Proc. Nati. Acad. Sci. USAVol. 84, pp. 4811-4815, July 1987Biochemistry

Engineered defective interfering RNAs of Sindbis virus expressbacterial chloramphenicol acetyltransferase in avian cells

(alphaviruses/RNA vectors/transfection)

ROBIN LEVIS, HENRY HUANG, AND SONDRA SCHLESINGERDepartment of Microbiology and Immunology, Washington University School of Medicine, 660 South Euclid Avenue, Box 8093, St. Louis, MO 63110

Communicated by Phillips W. Robbins, March 23, 1987

ABSTRACT We are investigating the feasibility of usingthe positive-strand RNA virus Sindbis virus and its defectiveinterfering (DI) particles as vectors for introducing foreigngenes into cells. In previous work we showed by deletionmapping of a cloned cDNA derived from one of the DI RNAsthat only nucleotides at the 3' and 5' termini of the RNA areessential for the DI RNA to be amplified after it is transfectedinto cells in the presence of helper virus. As a first step indeveloping a vector we replaced 75% ofthe internal nucleotidesof this DI cDNA with foreign sequences including the bacterialchloramphenicol acetyltransferase (CAT; EC 2.3.1.28) gene.DI RNAs transcribed from this cDNA were replicated andpackaged by helper Sindbis virus and became a major viralRNA species in infected cells by the third passage aftertransfection. They were also translated to produce enzymati-cally active CAT. CAT activity was measured at passage 3 butcould also be detected in transfected cells. DI RNAs containingthe CAT gene were translated in vivo and in vitro to produce twopolypeptides immunoprecipitable by anti-CAT antibodies. Onepolypeptide was identical in size to the authentic CAT poly-peptide; the other was the size expected for a protein initiatedat an upstream, viral-specific AUG in frame with the CATAUG. These studies establish that DI genomes of Sindbis viruscan tolerate the insertion and direct the expression of at leastone foreign gene.

A role for viruses as vectors for introducing foreign genes intocells can be traced back to the use of bacteriophage astransducing agents. Not until the development of recombinantDNA technology, however, have viruses been used to intro-duce a wide variety of genes into bacterial, animal, and plantcells (1, 2). The viruses now being used as vectors contain DNAgenomes, or, as with retroviruses, contain anRNAgenome thatis replicated through a DNA intermediate. But it should also bepossible for RNA viruses to act as expression vectors. ThecDNAs of several different virion RNAs have been cloned sothat the appropriate deletions and insertions can be made intothese genomes. The modified cDNAs can then be transcribedinto RNA using a bacteriophage or bacterial DNA-dependentRNA polymerase (3, 4). The feasibility of using an RNA viralgenome as a vector was demonstrated by French et al. (5), whoshowed that the bacterial chloramphenicol acetyltransferase(CAT; EC 2.3.1.28) gene could be inserted into the genome ofthe plant RNA virus, brome mosaic virus, and that it wasexpressed when the appropriate RNA transcripts were inocu-lated into plant protoplasts.We have been studying the RNA enveloped virus, Sindbis

virus, as a potential vector for introducing genes as RNA intoanimal cells. This virus has a wide host range and thereforemight be useful as a vector in many different cell types. Our firstapproach to developing this virus as a vector has been to insert

genes into a defective interfering (DI) genome. DI genomes aredeleted forms ofthe virion genome characterized by their abilityto interfere specifically with the replication of homologous orclosely related viruses (6). DI genomes of Sindbis virus containdeletions as well as rearrangements and repeats of the originalvirus sequence (7). The most prevalent, naturally occurring DIgenomes of Sindbis virus range in size from 2 to 2.5 kilobases(kb), corresponding to about one-sixth to one-fifth the size ofthevirion genome. We demonstrated by deletion mapping of acloned cDNA derived from one of the DI RNA genomes thatonly nucleotides at the 3' and 5' termini ofthe RNA are essentialfor the DI RNA to be amplified (4). The deletion analysis wascarried out by cloning a cDNA copy of a complete DI genomedirectly downstream from the promoter for the SP6 bacterio-phage DNA-dependent RNA polymerase. This cDNA wastranscribed into RNA, which was then transfected into chickenembryo fibroblasts in the presence ofhelper Sindbis virus. Afterone or two passages the DI RNA became the major viral RNAspecies in infected cells. Deletions extending over the entire DIgenome were made and only those in the 19-nucleotide regionat the 3' terminus and in the 162-nucleotide region at the 5'terminus destroyed the biological activity of the DI RNAtranscript. These results suggested that it might be possible tosubstitute foreign sequences for the dispensible internal se-quences. We have replaced 1689 internal nucleotides (75%) ofthe DI genome with foreign sequences, including the bacterialCAT gene. DI RNAs containing these foreign sequences notonly were amplified but also expressed CAT.

METHODSPlasmids, Transcription, and Transfection. The clone con-

taining the DI cDNA (KDI-25) has been described (4). ThepSV2cat plasmid (8) was a gift from John Majors (WashingtonUniversity, St. Louis). The insertion sites are indicated inFig. 1. The procedures for transcription and transfectionwere identical to those published (4, 9, 10). Transcripts to beused for in vitro translation were capped by including 1 mMm7G(5')ppp(5')G in the transcription reaction (11); theribonucleotide triphosphates were present at a concentrationof 0.5 mM. After a 60-min incubation at 41°C, 30 .tg ofDNaseI per ml was added for 15 min at 37°C to degrade the DNAtemplate.

S1 Nuclease Digestion. RNA and DNA samples to behybridized were coprecipitated in ethanol, resuspended in 12,ul of hybridization buffer (80% formamide/0.4 M NaCl/0.05M Pipes, pH 6.8/1 mM EDTA), heated at 75°C for 10 min, andthen allowed to reanneal at 54°C for at least 3 hr. Afterhybridization, 350 ,ul of S1 nuclease buffer (200 units of S1nuclease per ml/0.3 M NaCl/0.03 M NaOAc, pH 4.6/2 mMZnSO4/20 ,ug of DNA carrier per ml) was added and thesamples were digested for 30 min at 20°C. Samples were thenextracted with phenol and precipitated in ethanol (12).

Abbreviations: CAT, chloramphenicol acetyltransferase; DI, defec-tive interfering.

4811

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 8,

202

0

Proc. Natl. Acad. Sci. USA 84 (1987)

Protected fragments were denatured and analyzed by agarosegel electrophoresis (13).CAT Assay. The method described by Gorman et al. (8)

was followed, except for using 0.1 uCi (final concentration,17 ,uM) of [14C]chloramphenicol (60 mCi/mmol; 1 Ci 37GBq) in a final volume of 100 Fl. When the extracts wereprepared from cells infected with passage 2 virus and DIparticles, the reactions wdre carried out for 30 min. Whenextracts were prepared from cells after transfection, theincubation time was extended to 2 hr.

Immunoprecipitations. Monoclonal antibodies directedagainst CAT were a gift from C. Gorman (Genentech, Inc.,South San Francisco). Antibodies directed against theSindbis nonstructural protein, nsP1, were a gift from W. ReefHardy (California Institute of Technology, Pasadena, CA).Samples from cell extracts or from in vitro translationreactions in RIPA (1% Triton X-100/1% sodium deoxycho-late/0.15 M TrisHCl, pH 7.2) buffer were incubated with theantiserum for 2 hr at room temperature and were then treatedwith protein A-Sepharose CL-4B (Sigma) as described (14).Proteins were eluted from the resin in 1.5% NaDodSO4 andwere analyzed by polyacrylamide gel electrophoresis (15).

RESULTSReplacement of Sindbis Sequences with Foreign Sequences.

We, replaced 1689 nucleotides (nucleotides 242-1930) of theDI genome with 1492 nucleotides from pSV2cat. The codingregion for the CAT gene consists of only 657 nucleotides, butto determine if foreign sequences could replace the majorityof the viral sequences, we also included bacterial and simianvirus 40 sequences from the pSV2cat plasmid. Two cloneswere made (Fig. 1). In both clones the most 5' ATG is theAUG in the virion RNA that initiates translation of the non-structural genes (17). In the clone designated CT25, this ATG is

followed by two adjacent stop codons and three in-frame ATGsupstream from the ATG initiating translation of the CAT gene(Fig. 2). In the second clone, referred to as CTS253, the HindIIIsite at nucleotide 162 of CT25 was filled in so that the two stopcodons are out of frame with the most 5' initiating ATG. Inaddition, a polylinker was inserted at the Bal I site at base 241to permit substitution of other sequences downstream fromthe 5' viral sequences. With the inclusion of this polylinkerthe most 5' ATG is in frame with the ATG of the CAT gene,whereas the three intervening ATGs are out offrame (Fig. 2).Downstream from the termination codon ofthe CAT gene are800 nucleotides from pSV2cat followed by 328 nucleotides ofthe 3' end of KDI-25 including a poly(A) tail.RNA transcribed in vitro from either the CT25 or the

CTS253 plasmid was transfected into cells in the presence ofhelper Sindbis virus. Passaging ofthe progeny virus led to theaccumulation of DI RNAs identical in size to the inputtranscripts (Fig. 3). We established that these RNAs retainedthe complete foreign insert by S1 nuclease analysis. Theoriginal transcripts and the RNAs labeled in vivo during theformation of passage 3 were hybridized to either CT25 orCTS253 plasmid DNA and then digested With S1 nuclease.RNAs ofthe correct size were protected from digestion whenhybridization to the homologous DNA was carried out (Fig.3B). Hybridization to the heterologous DNA led to theprotection of a smaller fragment of the size expected if onlythose sequences downstream from the CAT AUG wereprotected (see Figs. 1 and 2). In addition to protection ofRNAs of the predicted size, lower molecular weight specieswere also present. These smaller molecules may be derivedfrom DI RNAs that had evolved dtiring passagiflg.The CT25 and CTS253 clones contain nucleotides 162-241

of the original DI cDNA. This region includes a 51-nucleotidestretch that is highly conserved in the genomes of alphavi-

SP6

.N"N11

N1-

N..N

N\

x

/

I/_KDI-25

rphI HI BllMph Ifind III

ATG CATI ~ ~ .PIubvz 0

Ball/Hind HI

40 ie ATG

3

Smn I

.N-"N

'NN

PsI

Hinc ll/Sna

Ahe IIIBa /Kpn

I,,, -,,,H?

FIG. 1. Diagrams showing Sindbis DI cDNAs and the insertion of the CAT gene. The construction of KDI-25 has been described (4). TheKDI-25 restriction enzyme sites into which the fragment from pSV2cat was inserted are indicated. The derivation ofCTS253 from CT25 requiredtwo modifications. The cohesive ends generated by digestion with HindIII were blunt-ended by treatment with DNA polymerase I large fragment,and a short polylinker was added downstream from the Bal I site at base 245. The polylinker has the following restriction enzyme sites: KpnI, Cla I, HindIII, and Xba I. The modified regions in CTS253 were sequenced by the dideoxynucleotide chain-termination method to verify thechanges made (16). Ori, origin; Amp R, ampicillin resistance.

CT25

4812 Biochemistry: Levis et A

rC%103

Dow

nloa

ded

by g

uest

on

Dec

embe

r 8,

202

0

Proc. Nati. Acad. Sci. USA 84 (1987) 4813

CTS253 0 GGATATAGTGGTGAGTATCCCCGCCTGTCACGCGGGAGACCGGGGTTCG

49 GTTCCCCGACGGGGAGCCAAACAGCCGACCAATTGCACTACCATCACA.

97 ATG.GAG.AAG.CCA.GTA.GTA.AAC.GTA.GAC.GTA.GAC.CCC.

133 CAG.AGT.CCG.TTT.GTC.GTG.CAA.CTG.CAA.AAA.GCT TC.AGC.

169 CGC.AAT.TTG.AGG.TAG. A. CAC.AGC.AGG.TCA.CTC.CAA.TTC.CCG.CAA.TTT.GAG.GTA.GTA.GCA.CAG.CAG.GTC.ACT.

205 ATG.ACC.ATG.CTA.ATG.CCA.GAG.CAT.TTT.CGC.ATC.TGG.CCA.AAT.GAC.CAT.GCT.AAT.GCC.AGA.GCA.TTT.TCG.CAT.

241 AGC.TTG.GCG.AGA.TTT.TCA.GGA.GCT.AAG.GAA.GCT.AAA.CTG.GGT.ACC.ATC.GAT.AAG.CTT.CTA.GAG.ATC.AGC.TTG.

277 ATG.GAG.---.---.---.---.---.---.----------------GCG.AGA.TTT.TCA.GGA.GCT.AAG.GAA.GCT.AAA.ATG.GAG.

ruses (19). We had shown that this region is not essential foramplification of KDI-25 RNA (4) but it does increase theefficiency of amplification (20). When we inserted the CATgene into a DI cDNA lacking the region from nucleotide 163to 241, no DI RNAs were detected after transcription,transfection, and passaging (data not shown). This resultsupports the hypothesis that the 51-nucleotide region playssome role in DI RNA amplification.CAT Activity in Cells Infected with CT25 or CTS253 DI

Particles. Extracts prepared from cells that had been infectedwith passage 2 Sindbis virus and either CT25 or CTS253 DIparticles were assayed for CAT activity (Fig. 4). Bothextracts converted chloramphenicol to the acetylated forms.The activity in extracts prepared from cells infected withCTS253 DI particles ranged from 0.7 to 1.5 nmol of['4C]chloramphenicol converted to the acetylated forms in 30min per 5 x 105 cells; the activity in extracts from cells

A B

SIN CT25 CTS253 RNA u3 T 3 T 3

49S --

26S- - - -

_- -_

1 2 3 4 5

DNA

FIG. 2. 5' Terminal sequencesof CT25 and CTS253 cDNAs. Thesequences of CT25 and CTS253are identical up to the modifiedHindIll site at nucleotide 162. Themost 5' ATG (AUG) initiates thetranslation ofthe Sindbis nonstruc-tural proteins; the codons in framewith that ATG are indicated. All ofthe ATGs are underlined and thestop codons are doubly under-lined. The most 3' ATG is theinitiation codon for the translationof CAT.

infected with CT25 DI particles was less by a factor of 2-10(see also the following section).CAT Activity in Cells Transfected with CT25 or CTS253

RNA. After transfection two or three passages are requiredbefore the DI RNA is clearly detected in infected cells underour usual labeling conditions (4). The high sensitivity of theCAT assay made it feasible to test those cells transfected withDI RNAs directly for CAT activity. Activity, detected 12 hrafter transfection with CTS253 and 15-18 hr after transfec-tion with CT25, depended on the presence of helper virus andthe CAT-containing DI RNA (Fig. 5). Extracts from cellsinfected with Sindbis virus alone had no CAT activity. Cellstransfected with CTS253 converted up to 50 pmol of[14C]chloramphenicol to acetylated forms in 30 min per 5 x 10cells. This activity cannot be compared directly to that obtainedin assays of extracts from passage 3 because only a smallfraction of the cells is transfected with DI RNA. (We estimate

Passage 3 RNA Passage 3 RNA

on o

(N 0r V3 tn-o c.

-c

U. 0 UCY J tJ-um u UJt.* t

'p.~l

1 2 3 4 5 6 7 8 9 10 11 12

CT25 CTS253

FIG. 3. Analysis of Sindbis viral and DI RNAs by agarose gel electrophoresis. Lanes 2 and 4 (labeled T), RNA transcripts from CT25 andCTS253. One microgram of these RNAs was used to transfect cells to prepare the first passage of virus and DI particles. The second passagewas used to infect cells to analyze viral and DI RNAs. Cells were labeled with [3H]uridine for 7 hr at 37°C and RNA was isolated by our standardprocedures (18). Lanes 1, 3, and 5 (labeled 3), RNAs isolated from cells during the formation ofpassage 3. SIN refers to cells infected with Sindbisvirus alone. (B) Protection of 3H-labeled DI RNAs from S1 nuclease digestion. RNA transcripts or RNA isolated from cells during the formationof passage 3 was hybridized with CT25 or CTS253 DNA. After digestion with S1 nuclease, RNAs were analyzed by agarose gel electrophoresis.Lanes 1-4, RNAs after hybridization with CT25 DNA; lane 5, CT25 RNA only; lane 6, CT25 RNA and CT25 DNA without hybridization. Lanes7-10, RNAs after hybridization with CTS253 DNA; lane 11, CTS253 RNA only; lane 12, CTS253 RNA and CTS253 DNA without hybridization.

Biochemistry: Levis et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 8,

202

0

Proc. Natl. Acad. Sci. USA 84 (1987)

12 hr 1 hr 18 hr

M.

* FIG. 4. CAT activity in cells* infected with passage 2 virus. Ex-,. tracts prepared from 2 x 106 cells

9 hr after infection were assayedfor CAT activity. Lane 1, cells

* * * infected with Sindbis virus alone;lane 2, cells infected with Sindbisvirus and CT25 DI particles; lane3, cells infected with Sindbis virus

1 2 3 and CTS253 DI particles.

that only 103 to 104 cells are transfected successfully.) Bypassage 3 most of the cells are infected with DI particles.

In five independent experiments CAT activity was at least10-fold higher in cells transfected with CTS253 DI RNA than incells transfected with CT25 DI RNA. The reason for thisdifference and for those seen at passage 3 is not clear. Ourfinding that the two DI RNAs amplify to essentially the sameextent by passage 3 (Fig. 3A) suggested that CTS253 RNA mightbe translated more efficiently than CT25 RNA. Our in vitrotranslation data, however, did not show any significant differ-ence in the translation efficiency of the two RNAs (see Fig. 6).

Translation of DI RNAs in Vivo and in Vitro. Cells infectedwith passage 3 virus populations containing either CT25 orCTS253 DI particles synthesized the viral proteins and twoadditional polypeptides (Fig. 6A). The larger molecularweight polypeptide in each sample was the size predicted fora fusion polypeptide that was initiated at the AUG upstreamfrom and in frame with the AUG of the CAT gene. (In CT25there are three upstream, in-frame AUGs that could poten-tially be used to initiate translation.) The smaller polypeptidewas the size of the authentic CAT protein (see Fig. 2). Thereare several ways to explain the presence oftwo polypeptides:(i) the DI RNAs might have evolved to form two species, eachof which was initiated at a different AUG; (it) the smallermolecular weight polypeptide arose by proteolytic process-ing of the larger; and (iii) the two polypeptides were trans-lated from the same RNA but by initiation at different AUGs.To distinguish among these possibilities, we carried out in

BA

PE2/El/E2I[

CAPSID- 412 3 4] nsPl-CAT

I..M-- 111-4 -CAT

1 2 3 4 5 6

0 0 0 0

1 2 3 4 5 6 7 8 9

FIG. 5. CAT activity in cells transfected with DI RNA. Cells (106)were transfected with 1 jig of the indicated RNA transcripts in thepresence of helper Sindbis virus. Extracts prepared at 12, 15, and 18hr after transfection were assayed for CAT activity. Lanes 1, 3, 5,and 8, activity present in cells that had been transfected with CT25RNA. Lanes 2, 4, 6, and 9, activity in cells transfected with CTS253RNA. Lane 7, a lack of activity in cells infected with Sindbis virusbut not transfected with a CAT-containing DI RNA. Lanes 8 and 9,lack of activity in cells transfected with either of these DI RNAs inthe absence of added Sindbis virus.

vitro translation reactions with the RNA transcripts obtainedfrom the CT25 and CTS253 plasmids and analyzed theproducts by polyacrylamide gel electrophoresis (Fig. 6A).Both RNAs produced two polypeptides identical in size tothe in vivo translated products. This result ruled out thepossibility that the two polypeptides were translated fromtwo DI RNAs that had evolved during passaging. It alsosuggests that the smaller protein was not a product ofproteolytic degradation, although we cannot eliminate thepresence of a protease or autoprotease-like activity in vivoand in the in vitro extracts.We established that the polypeptides synthesized in vivo

and in vitro contained CAT polypeptide sequences by im-munoprecipitating the samples with monoclonal antibodydirected against CAT protein. For CT25 and CTS253, twopolypeptides from the in vivo and the in vitro translationswere specifically immunoprecipitated by anti-CAT antibod-ies and were not precipitated by a nonspecific antiserum (Fig.6B; in vitro data not shown).The larger molecular weight polypeptide translated from

C

_ -nsPl

-nsPl-CAT.. .

,1 2 3 ,,J 5 6,antI CAT ant G (VSV )

n12 3 4,anti nsPl

FIG. 6. Polyacrylamide gel analysis ofproteins translated in vivo and in vitro from CAT-containing DI RNAs and of polypeptides precipitatedwith antibodies directed against CAT or the Sindbis nonstructural protein, nsP1. (A) In vivo and in vitro translations. Cells infected with Sindbisvirus or with Sindbis virus and DI particles during the formation of passage 3 were labeled with [35S]methionine for 1 hr, 7 hr after infection.Lane 1, cells infected with Sindbis virus alone; lane 2, cells infected with Sindbis virus and CT25 DI particles; lane 3, cells infected with Sindbisvirus and CTS253 DI particles. The rabbit reticulocyte lysate used for the in vitro translation reactions was purchased from Bethesda ResearchLaboratories and the reactions were carried out according to the manufacturer's instructions. Lane 4, reticulocyte lysate alone; lane 5, lysateplus CT25 RNA; lane 6, lysate plus CTS253 RNA. (B) Immunoprecipitation with anti-CAT and anti-G(VSV) (anti-G protein of vesicularstomatitis virus) antibodies. Samples from the [35S]methionine-labeled protein extracts shown in A were incubated with anti-CAT antibodies(lanes 1-3) or antibodies directed against the G protein ofVSV (lanes 4-6). The latter antibodies were used to measure nonspecific precipitation.Immunoprecipitations of extracts from cells infected with Sindbis virus alone are shown in lanes 1 and 4, with Sindbis virus plus CT25 particlesin lanes 2 and 5 and Sindbis virus plus CTS253 particles in lanes 3 and 6. (C) Immunoprecipitation with anti-nsP1 antiserum. Lane 1, in vitroproducts translated from CTS253 RNA as a marker for the CAT-nsP1 fusion protein. Samples from the [35S]methionine-labeled protein extractsshown in A were immunoprecipitated with anti-nsP1 antiserum: lane 2, cells infected with Sindbis virus and CT25 particles; lane 3, cells infectedwith Sindbis virus and CTS253 particles; lane 4, cells infected with Sindbis virus alone.

4814 Biochemistry: Levis et al.

qua

Dow

nloa

ded

by g

uest

on

Dec

embe

r 8,

202

0

Proc. Natl. Acad. Sci. USA 84 (1987) 4815

the CTS253 RNA transcript and also found in those cellsinfected with virus and CTS253 DI particles should contain 21amino acids from the viral nonstructural protein, nsPl, inaddition to the CAT sequences. That this is indeed the casewas shown by immunoprecipitation of the in vivo labeledextracts with antiserum directed against the nsP1 protein(Fig. 6C). The polypeptides precipitated by this antiserumwere the size expected for authentic nsP1 and for the fusionprotein between nsP1 and CAT. The 10-12 amino acids thatshould be at the amino terminus of the larger molecularweight protein translated by CT25 RNA are not in the samereading frame as nsP1, and the CT25 fusion protein was notimmunoprecipitated by this antiserum.

DISCUSSIONAn important conclusion from the work presented here is thatit is possible to replace at least 75% (1689 internal nucleo-tides) of a DI genome of Sindbis virus with foreign sequencesand still obtain RNA that can be replicated and encapsidatedby Sindbis-specific proteins. In many respects this result isnot surprising. The internal viral sequences in this DI RNAare rearrangements of viral sequences and therefore aredistinct from those present in the parental viral genome.Furthermore, deletions in this region do not destroy thebiological activity of the DI RNA. We did find, however, thatthose DI RNAs in which 900 or more internal nucleotides aredeleted are unstable and rapidly evolve to a size more typicalof endogenously generated DI RNAs (4). In contrast, DIRNAs of the correct size range, but containing foreignsequences, showed the same size stability as the RNAscontaining only viral sequences. Thus, although internalsequences are not specifically required for replication andpackaging, they appear to play a role in DI RNA selection andevolution by maintaining a critical size for the DI genome.A second conclusion from this study is that DI RNAs can

direct the expression of at least one foreign gene. This resultsupports the concept that Sindbis DI RNAs may be devel-oped as vectors for introducing RNAs into cells. For a DIRNA to be an efficient and useful vector it should bereplicated and translated efficiently. The CAT-containing DIRNAs described here are replicated and packaged well; theconcentrations of the DI RNA and Sindbis 26S mRNA ininfected cells by passage 3 were similar (Fig. 3A). The amountof CAT protein translated from the DI RNA, however, wassignificantly less than the amount of capsid and envelopeproteins produced by the 26S mRNA (Fig. 6A). It is possiblethat the CAT polypeptides are translated to high levels butare unstable in these infected cells. A more likely explanationis that these DI RNAs are translated inefficiently. DI RNAsare selected for their ability to compete with the parentalgenome in replication and packaging and if these DI RNAsare sequestered in replication complexes they would not beavailable for translation. Furthermore, the 5' termini of CT25RNA and CTS253 RNA consist of nucleotides 10-75 of a rattRNAAsP a sequence not normally found at the 5' termini ofmRNAs (21). This sequence might form a structure thatinterferes with the ability of ribosomes to bind to or traversethe mRNA. Since other viral sequences can also serve as the5' terminus of Sindbis DI RNAs (22), it may be possible tomodify the 5' terminal sequences to enhance translationwithout compromising replication.

Translation of the DI RNAs in vitro and in vivo produced twopolypeptides containing CAT sequences. The M, 24,000 poly-peptide translated from both RNAs was identical in size toauthentic CAT protein and the larger polypeptides were thesizes predicted if they were initiated at the upstream in-frameAUGs indicated in Fig. 2. A number of viral mRNAs initiatetranslation at more than one AUG (23). In these cases the most

5' terminal AUG is not in an optimal "context" with a purinein position -3 and a guanine residue in position +4 (24). Whenthe first AUG is in a suboptimal context, initiation can alsooccur at another, downstream AUG. In CT25 and CTS253 DIRNAs, the AUG upstream ofand in frame with the CAT AUGand the actual CAT AUG are both in an optimal context forinitiation. Although the most likely explanation for the presenceof the two CAT-specific polypeptides seems to be that they areinitiated separately, our data can also be explained by some typeof proteolytic processing.

Finally, we showed that cells transfected with CAT-containing DI RNA produce enough CAT protein to detectenzymatic activity. In our previous assay two or threepassages were necessary before the DI RNA was easilyidentified. This required not only that the RNAs be replicatedbut also that they be packaged. Our ability to measure CATactivity after transfection will permit us to distinguish se-quences in the DI RNA that are required for replication fromthose that may be essential only for encapsidation.We thank Drs. Charles Rice and John Majors for their advice and

assistance and Dr. C. Gorman for sending us the anti-CAT antibod-ies. This work was supported by Grant A111377 from the NationalInstitute of Allergy and Infectious Diseases and by the Monsanto/Washington University Biomedical Research Contract. R.L. hadbeen a Stephen Morse Fellow in Microbiology and Immunology.1. Rigby, P. W. J. (1983) J. Gen. Virol. 64, 255-266.2. Gluzman, Y., ed. (1982) Eukaryotic Viral Vectors (Cold Spring

Harbor Laboratory, Cold Spring Harbor, NY).3. Bujarski, J. J., Ahlquist, P., Hall, T. C., Dreher, T. W. &

Kaesberg, P. (1986) EMBO J. 5, 1769-1774.4. Levis, R., Weiss, B. G., Tsiang, M., Huang, H. & Schlesinger,

S. (1986) Cell 44, 137-145.5. French, R., Janda, M. & Ahlquist, P. (1986) Science 231,

1294-1297.6. Perrault, J. (1981) Curr. Top. Microbiol. Immunol. 93,

151-207.7. Monroe, S. S. & Schlesinger, S. (1984) J. Virol. 49, 865-872.8. Gorman, C. M., Moffat, L. F. & Howard, B. H. (1982) Mol.

Cell. Biol. 2, 1044-1051.9. Melton, D. A., Krieg, P. A., Rebagliati, M. R., Maniatis, T.,

Zinn, K. & Green, M. R. (1984) Nucleic Acids Res. 12,7035-7056.

10. Al-Moslih, M. I. & Dubes, G. R. (1973) J. Gen. Virol. 18,189-193.

11. Konarska, M. M., Padgett, R. A. & Sharp, P. A. (1984) Cell38, 731-736.

12. Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982) MolecularCloning: A Laboratory Manual (Cold Spring Harbor Labora-tory, Cold Spring Harbor, NY).

13. Carmichael, G. G. & McMaster, G. K. (1980) Methods En-zymol. 65, 380-391.

14. Anderson, D. J. & Blobel, G. (1983) Methods Enzymol. 96,111-120.

15. Laemmli, U. K. (1970) Nature (London) 227, 680-685.16. Sanger, F., Nicklen, S. & Coulson, A. R. (1977) Proc. Natl.

Acad. Sci. USA 74, 5463-5467.17. Strauss, E. G., Rice, C. M. & Strauss, J. H. (1984) Virology

133, 92-110.18. Weiss, B., Rosenthal, R. & Schlesinger, S. (1980) J. Virol. 33,

463-474.19. Ou, J.-H., Strauss, E. G. & Strauss, J. H. (1983) J. Mol. Biol.

168, 1-15.20. Schlesinger, S., Levis, R., Weiss, B. G., Tsiang, M. & Huang,

H. (1987) in Positive Strand RNA Viruses, eds. Brinton, M. &Rueckert, R. (Liss, New York), pp. 241-250.

21. Monroe, S. S. & Schlesinger, S. (1983) Proc. Natl. Acad. Sci.USA 80, 3279-3283.

22. Tsiang, M., Monroe, S. S. & Schlesinger, S. (1985) J. Virol.54, 38-44.

23. Kozak, M. (1986) Cell 47, 481-483.24. Kozak, M. (1986) Cell 44, 283-292.

Biochemistry: Levis et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 8,

202

0