JOURNAL OF VIROLOGY, Nov. 1993, P. 6365-63780022-538X/93/1 16365-14$02.00/0Copyright © 1993, American Society for Microbiology

Alternative Splicing of Human Immunodeficiency Virus Type 1mRNA Modulates Viral Protein Expression,

Replication, and InfectivityDAMIAN F. J. PURCELL* AND MALCOLM A. MARTIN

Laboratory of Moleclular Microbiology, National Institute ofAllergy andInfectious Diseases, Bethesda, Maryland 20892

Received 27 April 1993/Accepted 21 July 1993

Multiple RNA splicing sites exist within human immunodeficiency virus type 1 (HIV-1) genomic RNA, andthese sites enable the synthesis of many mRNAs for each of several viral proteins. We evaluated the biologicalsignificance of the alternatively spliced mRNA species during productive HIV-1 infections of peripheral bloodlymphocytes and human T-cell lines to determine the potential role of alternative RNA splicing in the regulationof HIV-1 replication and infection. First, we used a semiquantitative polymerase chain reaction of cDNAs thatwere radiolabeled for gel analysis to determine the relative abundance of the diverse array of alternativelyspliced HIV-1 mRNAs. The predominant rev, tat, vpr, and env RNAs contained a minimum of noncodingsequence, but the predominant nefmRNAs were incompletely spliced and invariably included noncoding exons.

Second, the effect of altered RNA processing was measured following mutagenesis of the major 5' splice donorand several cryptic, constitutive, and competing 3' splice acceptor motifs of HIV-1NL4-3. Mutations that ablatedconstitutive splice sites led to the activation of new cryptic sites; some of these preserved biological function.Mutations that ablated competing splice acceptor sites caused marked alterations in the pool of virus-derivedmRNAs and, in some instances, in virus infectivity and/or the profile of virus proteins. The redundant RNAsplicing signals in the HIV-1 genome and alternatively spliced mRNAs provides a mechanism for regulating therelative proportions of HIV-1 proteins and, in some cases, viral infectivity.

Eucaryotic cells control their metabolic activities by regulat-ing gene promoter activity and the processing of RNA andprotein. In the same way that the study of viral promoters hasserved as a paradigm for the promoters of their host mamma-lian cells, the investigation of viral RNA processing in mam-malian cells has provided an insight into various mechanismsused to regulate the steady-state level of a specific mRNA. Thecomplex nature of the processing of human immunodeficiencyvirus type 1 (HIV-1) RNAs provides an important model forhuman RNA processing pathways. All retroviruses requireRNA splicing to remove upstream gag and pol coding se-

quences from the env mRNA. In addition, HIV-1 generates a

distinctly complex pattern of spliced RNA to encode theessential regulatory proteins, Tat and Rev, as well as severalother proteins (Vif, Vpr, and Nef) needed for successfulreplication in vivo (3, 15, 18, 26, 32, 35, 36, 43, 47, 52). TheHIV-1 Rev protein binds viral RNA species that contain theRev-responsive element (RRE), located in the env gene,thereby promoting the export, and possibly the stability andtranslation, of partially spliced and unspliced RNAs from thenucleus into the cytoplasm for its translation and/or packaginginto progeny virions (2, 6, 7, 9, 12-14, 19, 20, 22, 23, 28, 33, 36,38). The Rev-RRE system alleviates the paradoxical require-ment for both spliced and unspliced HIV-1 RNA for successfulvirus replication. Rev protein also regulates the temporalchange from multiply spliced HIV-1 RNAs to partially splicedor unspliced RNAs during productive virus infection (27, 29).The splicing of HIV-1 RNA is extremely complex because of

the presence of both constitutive and alternatively used 5'RNA splice donor (SD) and 3' splice acceptor (SA) motifs.Numerous weak SA motifs, located toward the center of the

* Corresponding author.

genomic RNA, are competing points of ligation for splicing,and their alternate selection usually determines the proteinencoded by the mature RNA (3, 15, 18, 35, 43, 47). However,some of these mRNAs are multicistronic, encoding more thanone protein (15, 48, 49). Increased diversity of spliced mRNAsfor several HIV-1 proteins results from the alternative casset-ting of two noncoding exons into a proportion of transcripts(15, 43, 47). In addition, the use of several cryptic SA and SDsites may lead to the synthesis of novel chimeric viral proteins(5, 15, 46, 47). The varied use of these diverse splicing signalsresults in the synthesis of several sets of structurally differentRNAs that serve as alternative templates for the translation ofthe same protein, including the viral envelope, regulatory, andaccessory proteins. Because this complex pattern of RNAexpression is maintained among many HIV-1 isolates of di-verse origins (42, 43, 50), it is likely that this complexity iscritical for the successful completion of the HIV-1 infectiouscycle and not simply an inherent redundancy in viral RNAprocessing.The major advantage of examining regulated splicing of a

self-replicating entity such as HIV-1 is that such investigationscan use whole cells, rather than in vitro splicing extracts, andthe biological consequences can be readily correlated withRNA and protein expression as well as with virus infectivity.While mixed groups of mRNAs encode most HIV-1 proteins,it is unclear whether different cell types use alternative splicingto regulate HIV-1 RNA expression. The principal cell typesthat are infected by HIV-1, CD4+ lymphocytes and monocyte-derived macrophages, are known to alternatively splice RNAfrom some cellular genes, depending on the maturation or

activation status of the cell (e.g., CD45 [54, 55]) or on the celltype (e.g., CD46 [44]). Given the wide sequence diversity ofHIV-l strains (37), it is likely that sequence differences will

6365

Vol. 67, No. 11

on August 9, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

6366 PURCELL AND MARTIN

affect the splicing motifs of different virus isolates in view ofwhat is known in other systems (17, 34).We exploited the self-replicating capacity of HIV-1 to

examine the role of alternative RNA splicing in the regulationof virus replication and infectivity and to evaluate the relativeimportance of the different RNAs encoding HIV-1 proteins.First, we used a semiquantitative polymerase chain reaction(PCR) protocol that preserved the relative proportions of theHIV-1 RNA species in the 1.8- or 4.0-kb class of poly(A)+RNA to evaluate the steady-state levels of viral mRNAs duringproductive viral infection. Second, we introduced mutationsinto several SD and SA motifs of the HIV-1 proviral clonepNL4-3 to assess their effects on the composition and relativeabundance of alternatively spliced mRNAs during virus repli-cation and infection. HIV-1 splice site mutants permitted anexamination of the biological significance of the large redun-dant pool of spliced mRNAs and the potential role of alterna-tive RNA splicing in the regulation of HIV-1 during tissueculture infections.

MATERIALS AND METHODS

Construction of proviral mutants. The HIV-1 proviral mo-lecular clone pNL4-3 was constructed from the NY5 and LAV(LAI) HIV-1 isolates (1). A PCR-based mutagenesis protocolthat used a mutagenic oligonucleotide and a second primerpositioned near a convenient restriction endonuclease site wasused to generate a PCR product containing the mutation. Thisproduct was gel purified and used as a megaprimer with a thirdoligonucleotide positioned near a second convenient restric-tion site to generate DNA containing the mutated splice siteand the two restriction sites (31). These products were clonedinto the pCR1000 vector (Invitrogen, San Diego, Calif.),screened by restriction mapping, and then cloned back into theHIV-1 provirus by using the selected restriction sites. Theoligonucleotides used, with mutated nucleotides underlined,were as follows: for SD1, Odp.008 (5'-TGGCGTACTCTGCAGTCGCCGCC-3') with Odp.002 (5'-CTCTGGTAACTAGAGATCCCTCAG-3') and then Odp.007 (5'-CTCATCTGGCCTGGTGCAATAAGG-3'); for SA4b, Odp.023 (5'-AGGAGATGCTCAAGGC7lITYlGTCATG-3'), and for SA5, Odp.025(5'-GTCTCCGCTTlCTTCGAGCCATAGG-3'), each withOdp.021 (5'-GAATTGGGTGTCGACATAGCAG-3') andthen Odp.030 (5'-TTGFLT7AYTATTATlTTCCAAATTGTTCTC-3'); for SA6, Odp.028 (5'-GTGTTAGYTTATCTTGCACTGATTTGAAG-3') with Odp.030 and then Odp.021;and for SA7a, Odp.033 (5'-CTATAGTGAATTCAGTTAGGCAGGGAT-3'), for SA7a+7b, Odp.035 (5'-CTATAGTGAATTCAGTlTlTCGCAGGGATATT-3'), and for SA7,7a,7b, Odp.037 (5'-CTATAGTGAATTCAGFT[TCGCAGGGATATTCACCATTATCGT'Tl7CGTACCCACCTCCCCTATAGTG AA T A GA GTTAG G CAGGGATAT-3'), each withOdp.032 (5'-CCGCAGATCGTCCCAGATAAG-3') and thenOdp.031 (5'-AGTAGAGCAAAATGGAATGCCAC-3').Splice site mutant proviruses were sequenced to verify thepresence of the desired change as well as additional changesthat might have been introduced during the PCR procedure.Some mutations (ASA4b Tat G->S, ASA5 Tat R->S, ASA6Env K--S, ASA7a Env R->S, ASA7b Env R->S, and ASA7Env Q->R) changed the codon at that splice site. Otherchanges were as follows: pNLASD1, 756 A-4T (a silent changein the packaging signal); pNLASA4b, 6002 C-*T (Tat A-V,Rev L->F); pNLASA5, 6143 A--G (Env E->G); pNLASA6,6695 T->C (Env F->L); and pNLASA7a+7b, 7361 T->C (EnvF--L), 8069 C->A (Env L--M), 8107 G->T (Env W->C), and8321 T->C (Env S--P). Several other proviral splice site

mutants were sought; the resultant plasmids proved to beunstable, however, precluding their functional analysis.

Cell culture, transfections, and infections. HeLa cells, main-tained in Dulbecco's modified Eagle's medium supplementedwith 10% fetal calf serum (FCS), were obtained from theAmerican Type Culture Collection. CEM (12D7) cells, main-tained in RPMI 1640 medium with 10% FCS, were obtainedfrom Microbiological Associates (Gaithersburg, Md.), as wereperipheral blood mononuclear cells (PBMC), which werestimulated with phytohemagglutinin (0.25 ,ug/ml; WellcomeDiagnostics, Dartford, United Kingdom) for 96 h and grown inRPMI 1640 with 10% FCS and 10% interleukin-2 (PharmaciaDiagnostics, Fairfield, N.J.). HeLa cells (5 x 105) in T25 flaskswere cotransfected by the calcium phosphate coprecipitationtechnique (57) with 20 ,ug of proviral plasmid DNA and 0.5 ,ugof a human growth hormone reporter plasmid, pXGH5 (10);transfection efficiency was determined by a radioimmunoassayfor human growth hormone (Nichols Institute, San JuanCapistrano, Calif.). Virus production was monitored with anassay for reverse transcriptase (RT) activity, using [32P]TTPincorporation with an oligo(dT) * poly(A) template (59). Cells(2 x 105) were infected with 105 cpm of RT activity of anHIV-1 inoculum (approximately equivalent to a multiplicity ofinfection of 0.002) in 1 ml of RPMI 1640 for 2 h at 37°C beforeaddition of 4 ml of RPMI 1640 containing 10% FCS. The cellswere fed with RPMI 1640 containing 10% FCS at 2-dayintervals, and aliquots of the medium were assayed for RTactivity.

Isolation of HIV-1 mRNA, preparation of cDNA, and semi-quantitative PCR for spliced HIV-1 mRNA. Total cell RNAwas harvested by extraction with RNAzol (TelTest Inc.,Friendswood, Tex.) 16 or 36 h after transfection of HeLa cellsor immediately prior to the peak of RT production followinginfection of approximately 5 x 106 infected human PBMC;poly(A)+ RNA was selected by the Micro-Fast track oli-go(dT)-cellulose method (Invitrogen). For first-strand cDNAsynthesis, either poly(A)+ RNA or in vitro-transcribed RNA(see below) was denatured in 15 mM MeHgOH in the pres-ence of 130 mM f-mercaptoethanol and 0.5 ,ug of randomhexamers at 94°C for 2 min before reverse transcription (twotimes) for 1 h with murine leukemia virus RT, using a cDNAcycle kit (Invitrogen). Semiquantitative PCR analysis of cDNAfrom the 1.8-kb class of RNA was performed with oligonucle-otide primers Odp.045 (5'-CTGAGCCTGGGAGCTCTCTGGC-3'; positions 477 to 499) and Odp.032 (5'-CCGCAGATCGTCCCAGATAAG-3'; positions 8477 to 8498); for the4.0-kb class of RNA, primers Odp.045 and Odp.070 (5'-ACTATTGCTATTATTATTGCTACTAC-3'; positions 6094to 6115) were used. Twenty cycles of PCR were performedwith 1 U of Amplitaq (Perkin Elmer Cetus, Norwalk, Conn.),2 ,lI of first-strand cDNA, 0.2 mM each dATP, dCTP, dGTP,and dTTP, and 1 ,uM each primer in 10 mM Tris-HCI (pH8.3)-50 mM KCl-1.5 mM MgCl2-O.001% (wt/vol) gelatin byfirst incubating the mixtures for 5 min at 94°C and thensubjecting them to thermocycling (94°C for 1.5 min, 55°C for 1min, 72°C for 2.5 min) before finally incubating them at 72°Cfor 7 min. If the DNA concentration in a reaction mixture wasless than 10 ng/,l, as assessed by agarose gel electrophoresis,aliquots were sequentially reamplified in steps of five cyclesafter dilution 1:4 with fresh reaction mix to maintain linearamplification. Amplification products (100 ng) were radiola-beled by performing a single round of PCR as before but withaddition of 10 ,uCi of [32P]dCTP and subsequently analyzed byelectrophoresis on a 6% polyacrylamide-urea gel at sufficientcurrent to maintain a temperature of 65°C; bands were visu-alized by autoradiography or quantified with a Fujix BAS 2000

J . VlIROL.

on August 9, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

RNA SPLICING MUTANTS OF HIV-1 6367

Bio-image analyzer. An MspI digest of pBR322 was endlabeled and used as size markers. Controls used in PCRexperiments included cDNA from mock-transfected and -in-fected cells and poly(A)+ or in vitro-transcribed RNA (below)that was not incubated with RT during the cDNA synthesisreaction.

Analysis of PCR products and cloning of HIV-1 cDNAs.Bands were excised from the PCR gel, eluted in 0.5 Mammonium acetate-10mM magnesium acetate-I mM EDTA-0.1% sodium dodecyl sulfate (SDS) for 16 h at 4°C, and thenprecipitated with ethanol before rcamplification with the sameprimers. After it was confirmed that the PCR product migratedas a single band, residual primers were removed with a MagicPCR Prep column (Promega, Madison, Wis.), and the productwas directly sequenced by using end-labeled nested oligonu-cleotide primers in the dsDNA Cycle Sequencing System (LifeTechnologies, Inc., Gaithersburg, Md.). Purified PCR bandswere also cloned into the pCRII vector (Invitrogen) so that thesense strand was downstream of the T7 promoter sequence,and the identities of clones were confirmed by sequencing.HIV-l cDNA clones were linearized with SpeI, extracted withphenol and chloroform (50%, vol/vol), and precipitated withethanol, and RNA was transcribed in vitro by using T7 RNApolymerase (Promega). The DNA template was removed bydigestion (twice) with RNase-free DNase (Promega), and theRNA produced was directly quantified by spectrophotometryor density scanning of bands following agarose gel electro-phoresis. Selected in vitro-transcribed RNAs were diluted andthen mixed in known proportion as controls for subsequentcDNA synthesis and PCR amplification.

Analysis of HIV-1 proteins. Lysates were prepared fromtransfected HeLa cells (5 x 107 cells per ml) in 0.5% NonidetP-40 in 10 mM Tris-HCI (pH 7.4)-150 mM NaCI-I mMEDTA-1 mM phenylmethylsulfonyl fluoride. Different amountsof the cell lysate, standardized on the basis of human growthhormone activity, were separated on SDS-5 to 20% polyacryl-amide gradient gels, electroblotted onto Immobilon-P mem-branes (Millipore, Bedford, Mass.), and blocked with 5%powdered skim milk in phosphate-buffered saline. HIV-1proteins were detected by using serum (1:1,000) from anHIV-1-seropositive individual, or rabbit serum to a Nef N-ter-minal peptide (1:100) (21) or to purified NL4-3 gpl60/120(1:100) (58), and visualized with '251-labeled protein A byautoradiography or phosphoimaging analysis. Alternatively,cells were biosynthetically labeled with [35S]methionine-cys-teine (Tran 35S-label; ICN, Irvine, Calif.) as described previ-ously (59) and immunoprecipitated with rabbit serum to Revproduced in Escherichia coli (1:100) (gift from D. M. D'Agosti-no and G. N. Pavlakis). Levels of Rev functional activity weremeasured as described previously (24) by cotransfecting 0.5 [Jgof the pDM128 Rev reporter plasmid with 20 jig of proviralplasmid and 0.5 p.g of pXGH5 human growth hormonereporter plasmid and then assaying for chloramphenicol acetyl-transferase (CAT) activity on cell extracts that were standard-ized for growth hormone expression.

RESULTS

The selection of different competing SA sites in primaryHIV-1 transcripts leads to the alternative exon composition offully processed mRNA (Fig. 1). For example, functional Tatprotein, a transactivator of HIV-I transcription (8, 35, 40, 51),is expressed from the two types of mRNA using SA4 (Fig. IA):RNAs containing exon 4E, which is continuous from SA4 tothe poly(A) addition site (one-exon tat), or transcripts joiningexons 4 and 7 (two-exon tat) (Fig. IC) (45). Two other

competing SA sites, SA4a and SA4b (Fig. IA), mapping fewerthan 200 bases downstream from SA4, give rise to exons 4a and4b, which are spliced to exon 7 to generate mRNA for Rev(Fig. IC). Another competing splice site, SA5 (Fig. lA), is usedfor the expression of exons 5 and SE (Fig. IC); RNA speciesthat contain exon 5E encode envelope gp16O, and thosesplicing exon 5 to exon 7 encode Nef protein (15, 47). Thetranslation initiation sites of mRNAs using SA4a, SA4b, andSA5 have poor ribosome binding capacities and therefore havethe potential to be multicistronic: mRNAs containing exon

SE, 4aE, or 4bE encode Vpu and gpl6O envelope proteins,whereas transcripts joining exon 4a or 4b with exon 7 encodeboth Rev and Nef (15, 47, 48, 53). This increases the codingpotential of several HIV-1 mRNAs and the complexity ofHIV-1 mRNA pool. Further complexity results from theinterchangeable incorporation of two noncoding exons, exons2 and 3 (Fig. IC), into the spliced RNA species utilizing any ofthe downstream SA motifs.PCR protocols using primers that promote amplification of

limited subgroups of RNA have been used in conjunction withselective hybridization probes to map SA6, SD5, SA7a, andSA7b (Fig. lA) within env, thus generating frameshifting exons

6, 7a, and 7b (Fig. IC). These exons may generate novelchimeric Tat-Env-Rev (Tev or Tnv), Rev-X-Tat, and Tat-Envfusion proteins following transfection-infection by some deriv-atives of the HIV-I1 Al strain (5, 15, 46, 47). These are merelycryptic sites in the HIV-1NL4-3 genome (see below), and theirrole during HIV-1 replication is not clear (16).

Semiquantitative PCR analysis of spliced HIV-1 mRNA.Since Northern (RNA) blot analysis of HIV-1 RNA does not

distinguish the full array of alternatively spliced RNAs encod-ing the same viral protein, we used a sensitive semiquantitativePCR assay and urea-acrylamide gel analysis that discriminatedbetween RNA species differing in size by a single nucleotide.Two such assays were carried out. The first selectively detectedthe smaller multiply spliced 1.8-kb RNAs that use SD4 andSA7, SA7a, or SA7b (nef, rev', tat, or vpr) by using the Odp.045and Odp.032 primers (Fig. lD); the second specifically ana-

lyzed the larger 4.0-kb RNA species, which contain exons thatextend beyond SD4 into the env reading frame (vpu/env,one-exon tat, vpr, and viJf), by using the Odp.045 and Odp.070primers (Fig. IE). Representative gels, depicting the PCRproducts that resulted from random hexamer priming of themultiple species of HIV-1 RNA present in infected PBMCprior to the peak of RT production, are shown in Fig. 1D andE. Each of these cDNA bands was excised from the gel,reamplified, and directly sequenced. Several representativecDNA clones of each band were also introduced into thepCRII vector, which contains a T7 promoter adjacent to thecloning site, and also sequenced. To ascertain whether therelative quantity of the PCR-amplified cDNA bands visualizedon the gel faithfully represented the relative levels of thevarious RNA species isolated from infected cells, plus-senseHIV-1 RNA was synthesized from eight different cDNA clonesin vitro by using T7 RNA polymerase, directly quantified, andthen mixed in known proportions. Included among theseRNAs was one, designated cryptic, from a cDNA clonecontaining exons 1, 2x, 5, and 7, where exon 2x reads throughSD2 to a cryptic SD at position 5059 in vif. The RNA mixtureswere then used as templates for cDNA synthesis and PCRamplification, and the -P-labeled PCR products generatedwere analyzed by phosphoimaging (Fig. 2). We found that theproportion of radioactivity measured for each cDNA bandclosely matched the proportion of RNA added to the mixture

prior to cDNA synthesis (Fig. 2A). Thus, each cDNA reverse

VOL. 67, 1993

on August 9, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

6368 PURCELL AND MARTIN

A. Splice junctions- Splice

fL HJ donors

- SI): I

B. Open reading frames

C'D C7O'

Ln i" in 4'm G, C;i)

Spike CIDO1%IIt-I~Oil LnII ,1101>acptnjrs tiy t

-S: 2 3 4-ih: 6:

2 3 4

I'c L'Iev ' -'

7 -L , -LtE=LIn.-L

NT'P

PU'

CO CO

CS} Cc 0

..i.JCiD

(D C

CD G

_I c

iAAA,^.i.71) -87

..

tat21tcv.3

n,rvXr

C. Exons 02 *3 -5 06M41)M 4a- 4-4

-

- 3a

M.

ml

- 7_ _

7aI 2 J7b

- S~~E4hE4aE4cE4E3E2E

D. PCR for 1.8 kb class of RNAP('GRGelI RNA name Exon content lBand size

- Vpr 2 1 /2 J 3a / 7 1 101 [itVprI 11/3a/7 1051 nt

Tat 4

Tat 3

I'aat 2_doo.. It

Tat I

Rev 12Rev 11

Rev 11)

Rev 8Revr 7Ref 6Nef 4ReReesr 4Nef 3Rev 3Rev 2Rev IN-ef 2

Nsef I

2/ / 4 / -7

113/ 7I1,4/7 4/

1 / 3 / 4t 7

/2/3/4i /7

/2 / 3/4 / 7

3 4c 713 4a 7

3 /4b 7

/ /4a 7I 2i4b 7/1 /'7

/ /744il 7/4 / 7/ 5/

788 rit

73>8 Tlt714 rIt664 iit629 nI

61 ntri605 lit589 Tit5 -7 9ri i

93 rin555 nit

537n nt5 15 ior

487 nt480 rit464 jit

-t -- Vif 2

_Vpr 4Vpr3Tat 8Tat 7Tfat 6TI'at 5Env 16Env 15

- / Env 14Ern 1 3

/ {nv 12Env 11Env 1(I

_lp - .Env 8W-f HlEiv 7

lEnv 6. Envr h

En\ 4

Env I

/ 7 396 Iot

E. PCR for 4.0 kb class of RNAPCUR CGel RNA name Exon content Band size

I / 2E

I 1/2 / 3EI /3EI /2/3/4E

3 /4E/2/4E/4E/2/3/4cE

I /2/3 4aEI / 2 /3 / 4bEI / 3/SE/3!4cE

I /3!4aEI /3/4bEI 1214cR/3/5E

I /2/4aEI 12/4bEI /2/5E/4cE

I /4aFI /4bEI /5E

1474 nt

1047 nt997 ot734 nt684 nt660 nt610 nt575 nt557 nt551 nt535 nt525 nt507 nt501 nt501 nt485 nt483 nt477 nt461 nt451 nt433 nt426 nt410 nt

J. VIROL.

I

on August 9, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

RNA SPLICING MUTANTS OF HIV-1 6369

transcribed from the in vitro RNA was amplified with equalefficiency with use of the Odp.045-Odp.032 primer pair.PCR analysis of the 1.8-kb class ofmRNA was performed on

samples from HeLa cells transfected with HIV-1 proviralclones or from human PBMC infected with cell-free HIV-1(Fig. 2B). RNA directed by the pNL4-3 cloned provirus DNAin transfected HeLa cells or by HIV-lNL4-3 in infected humanT lymphocytes yielded the same pattern of bands in numerousindependent experiments irrespective of whether RNA was

synthesized following transfection or infection. A similar band-ing pattern was also observed with RNA prepared from PBMCinfected with HIV-1LAI (Fig. 2B, lane 10). The results pre-

sented in Fig. 2 demonstrate that HIV-1 RNA is splicedequivalently in transfected HeLa cells and infected PBMC andthat the splicing pattern is similar for closely related strains ofHIV-1 such as NL4-3 and LAI.These control PCR amplifications also illustrate two other

important points: (i) for each RNA mixture, the relativeproportion of the different amplified RNA species, as deter-mined experimentally, closely approximated the original pro-

portion added, and (ii) only PCR products deriving from inputRNA were detected on the gel, indicating that only RNAsspliced by genuine processing pathways gave bands in thisanalysis. Abnormal cDNA species, potentially arising by tem-plate jumping during reverse transcription of RNA, did notappear as bands on the final autoradiogram. Thus, the PCRprotocol used here accurately reflects the relative quantity ofone HIV-1 RNA species compared with the others amplified inthe same reaction.

Identification of a new SA site for rev and vpu/env RNAs. OurPCR analyses of viral RNA synthesized in transfected HeLacells and infected PBMC indicated the presence of a previouslyunreported competing SA site, designated SA4c, that was usedto generate three novel rev mRNAs, rev3, rev6, and rev9 (Fig.1D), and three novel vpulenv mRNAs, env4, env9, and envl6(Fig. 1E). The SA4c site is 18 nucleotides (nt) upstream fromSA4a and is conserved among most, but not all, HIV-1 isolates(Fig. 3). The SA4c splice site exists in HIV-lHXB2, HIV-1LAI,HIV-lJR-CsF, and HIV-lBA-L, the strains used in earlier studiesmapping the splicing motifs of HIV-1, although its use was notnoted (3, 15, 18, 35, 43, 47). Since no new translation initiationsites are introduced into mRNAs using SA4c, transcriptscontaining exons 4c and 4cE would still encode typical HIV-1Rev or Env protein. However, structural changes introducedinto these mRNAs may affect their translatability. An ATGcodon exists in the HIV-1NL43 between SA4c and SA4a but ispresent in a context unfavorable for efficient translation (30).

Relative abundance of the alternatively spliced HIV-1mRNAs. Since PCR amplification of HIV-1 cDNA preservedthe relative proportion of the various alternatively splicedforms of HIV-1 RNA, we used phosphoimaging analysis to

directly determine the relative abundance of each of theseRNAs in a spreading HIV-1NL4-3 infection of PBMC (Fig. 2and 4). Within the 1.8-kb class of HIV-1 mRNAs (Fig. 4A), nef,rev, tat, and vpr species existed in a ratio of 56:34:9:1; within the4.0-kb class (Fig. 4B), env, tat, vpr, and vif species existed in a

ratio of 92:5:2:1. Of the nef mRNAs, nef2, which includesnoncoding exon 5 flanked by SA5 and SD4, was the predom-inant type, comprising 49% of all nef RNA and 28% of all1.8-kb RNAs. nefRNAs containing noncoding exon 3 (nef4) or

2 (nef3) or both 2 and 3 (nef5) were present in decreasingamounts. Nefl RNA, which lacks a noncoding exon, was theleast abundant nefmRNA. It is unclear whether the nontrans-lated RNA from exon 5 contributes to nef function, as the nefgene has not been evaluated as an RNA element, only as

protein.In contrast, the predominant spliced mRNAs for the other

HIV-1 proteins lacked noncoding exons. Only 20% of rev

mRNA includes a noncoding exon, and the use of noncodingexon 3 (rev7, rev8, and rev9) or both exons 2 and 3 (revl0,revll, and revl2) predominates over use of exon 2 alone (rev4,revS, and rev6). In addition, revl and rev2 mRNAs, utilizingSA4a and SA4b, respectively, occur approximately fivefoldmore frequently than rev3 mRNA, which uses the previouslyunreported SA4c. Both the one- and two-coding-exon forms oftat mRNA (1.8- and 4.0-kb class RNAs, respectively) infre-quently use noncoding exons 2 and 3 (tatl and tat5). However,when a tat mRNA contained a noncoding exon, exon 2 was

predominantly used (tat2 and tat6). The relative proportion ofthe one- and two-exon forms of tat or vpr mRNA could not bedetermined here, since the assays for the 1.8- and 4.0-kbmRNAs are merely semiquantitative among the species repre-sented in each PCR assay.

Eighty percent of env mRNAs use SA5 (exon SE); however,12% of env RNAs (env2, env3, and env4) utilize the upstreamSA4a, SA4b, or SA4c (exon 4aE, 4bE, or 4cE) when noncodingexons 2 and 3 were excluded. Noncoding exons 2 and 3 were

used only at low (<5%) frequency in env mRNA and mostly inconjunction with SA5 (envS, env8, and envl3). This studyclearly identifies mRNA species containing both noncodingexons 2 and 3 in the same transcript. This finding contrasts withprevious analyses of RNA directed by the HIV-lHXB2 strainthat identified these as mutually exclusive exons (15, 18, 47)but is in agreement with a recent report that characterized viral

FIG. 1. Identification by RT-PCR of alternatively spliced HIV-1 mRNAs containing a variety of exons that arise from the existence ofnumerous splice junctions encoded in the HIV-1 genome. (A) Map showing the locations of the SD and SA sites in the pNL4-3 proviral molecularclone of HIV-1, with each position shown in nucleotides from the start of the 5' long terminal repeat (LTR). The SD and SA sites are numberedas done by Schwartz et al. (47). (B) Organization of the HIV-1 genome. Open boxes show locations of the open reading frames that encode theHIV proteins. (C) The different HIV-1 exons generated from the use of different combinations of SA and SD motifs during RNA splicing areshown as bars and numbered as done by Muesing et al. (35). Exons represented by gray bars were not found in any of the RNA species examinedin these studies. Exons 1, 2, 3, and 5 do not encode HIV-1 protein, and those numbered with an E read through SD4 into the env gene. (D) Thevarious HIV-1 mRNA species falling into the 1.8-kb size range in Northern blot analyses were distinguished by acrylamide gel analysis ofPCR-amplified cDNA, using primers Odp.045 (positions 477 to 499) and Odp.032 (positions 8477 to 8498). A representative gel from PCR analysisof RNA from pNL4-3-infected PBMC is shown. The distinct HIV-1 RNA species were identified by direct PCR sequencing of excised bands andby cloning, and the identity, exon content, and size of the PCR product are shown. The mRNA species have been named according to the principalprotein that they encode and by their size, with the smallest RNA as 1. Faint bands were visible on longer exposures of the autoradiogram. revmRNAs are bicistronic and also encode Nef (15, 48, 49). (E) Representative acrylamide gel from PCR analysis of HIV-1 mRNA species fallinginto the 4.0-kb size range from pNL4-3-infected PBMC, using primers Odp.045 and Odp.070 (positions 6094 to 6115). The identity of each RNAspecies, determined by direct sequence analysis of excised bands, is shown with the exon content and size of the PCR product. RNA species notmatched to a band on the gel were difficult to detect except on very long gel exposures and were not quantified above the background level inphosphoimaging analysis. All env mRNAs are multicistronic and also encode Vpu and Nef (48, 49).

VOL. 67, 1993

on August 9, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

6370 PURCELL AND MARTIN

B.

100 100 100 100102 106 102 1000.4 5 10 500.2 2 8 58100 100 100 10095 98 97 95

50 10 5 0.458 8 0.2 0.1100 100 100 10071 82 84 790.4 5 10 500.1 2 10 58100 100 100 100117 123 113 115

Tat2

Tat1

cyptic

Nef4

Nef 3

Rev2

Nef2

4~

RNA mixtures r , *g

ABCD4tt4*~t q9¼~A B C D + ?q #hVpr2Vpr 1

Tat4

Tat3

P-'- - Tat2

Tati

Rev 12

-nquie Rev 1O,11.... ''Nef5RevB,9

Net3

Rev3_,n,. NRev2_ _ = = ~~~~~~~~~RevI-~~ ~~~ ~~~~~~~Nef2

Nef 1 50 10 5 0.471 7 5 0.8 Nef 1 *r

N

A..- Netl

1 234567In vitro RNA HeLa

89 10PBMC

FIG. 2. The relative proportions of alternatively spliced HIV-1 mRNAs were accurately determined by RT-PCR assay. The semiquantitativecapacity of the RT-PCR assay for the 1.8-kb class of spliced HIV-1 mRNA was evaluated by mixing predetermined concentrations of RNAtranscribed in vitro from HIV-1 cDNA clones with T7 RNA polymerase (A) and then performing reverse transcription, PCR amplification, labelingwith [32P]dCTP, and gel analysis (B). Proportions of each RNA added into four mixtures, A through D, are shown in panel A with the proportionof cDNA experimentally amplified by PCR (panel B, lanes 1 to 4) and determined by phosphoimaging analysis (shown in italics in panel A). (B)Lane 5 shows a pool of RNA mixtures A through D that were treated in parallel but without the addition of RT during cDNA synthesis. RNAfrom HeLa cells transfected with reporter human growth hormone only (lane 6) and with pNL4-3 (lane 7) or with RNA from PBMC (lane 8) andPBMC infected with HIV-lNL,43 (lane 9) or HIV-lLAI (lane 10) was examined in the same set of reactions as the control. The identity of eachamplified species is shown on the right (see Fig. 1D).

SA

ACCAATTGCTATTGTAAAAAGTGTTGCTTTCATTGCCA----C-----------------

-A----------------------------------T----C-----------------

-- --_-_

4-SA4a

PTTGTTTCATGACAAPACF---------CA------__------A-------G-------CA----- -

---------CA--------------CA ---- --

SA4b

CCTI AC

G----__"I____-

4-Rev SAstart 5

GCATCTCCTT}{GCAC_________ __ __ _

_________ __ __ _

_________ __ __ _

------------_-_-AT - -F-------------------C----CA-G----__ G-------------------C----------------- --- C------A-----__G-----A--F---------------------________________ _---A--------G-- __--- Si-A----G---C--------------------------- -----C--T-A-AC---__G--

-AT-G -----------------C------------- -G------A-A--G--__G---J_ _ _ _ __

FIG. 3. Location of SA4c, a new SA site for rev and vpu/env mRNAs. Shown is alignment of nucleotide sequences of several HIV-1 isolatessurrounding the new SA site for rev and vpu/env mRNAs, SA4c (adapted from Myers et al. [37]).

A.

In vitro RNA mixtures

A B C D

Tat2

Tatl

cryptic

Net4

Nef3

Rev2

Nef2

NL4-3LAIHXB2SF2MNNY5

ADABALSF162

ELI

MAL

GAAGA

J. VIROL.

4-SA4c

Z

on August 9, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

RNA SPLICING MUTANTS OF HIV-1 6371

A.forward primer

Odp.045:II

- splice 2,* acceptors

I M_IU k}-111illor"Giv. -splc.e .-4! N a_~- splice1 donors

RNA species

-Nef 1-=mL.- Nef 2mmm-- Nef 3mom- Nef 4miii- Nef 5

Rev 1Rev 2 i.Rev 3 'Rev 4 -,--Rev 5 -Rev 6 -Rev 7Rev 8Rev 9Rev 10 -,_i-Rev 11Rev 12 -

1.8kb class of RNAreverse primer

:Odp.032.I II

3a 4ca7

I In7.,1"O*22 3Vlli2 3

- Tat 1- Tat 2- Tat 3- Tat 4

:L I. L.!.,I... . 4

:: Relative %

;;;i5 -j~

. ...- Af

4911247

-,-,-- 32 -j.... 40

8

11-

_Li~ 44_ 3-3

.-.. 2 -:_.. 2_ -1

4828169

- Vpr 1- Vpr 2 -.n-

7228 -.

B. 4.0 kb class of RNAforward primer reverse primer

Odp.1145 Odp.070.'

2 3. 4 cab - spliceacceptors

1 , '2 . pl3 4 do svpr Irevi I to~~L Ll w~~~~~it 1 Lrz| splice

I| 2 |~~~~~~

3 4 donors

RNA species

o. Env1 -.-Env 2-Env 3

Env 4Env -Env 6

in-Env7 -. Env 8

Env 9- Env 10- Env 11- Env 12.-- Env 13.-

in- Env 14- Env 15- Env 16

- Tat 5- Tat 6- Tat 7- Tat 8

Relative %

805

_6

mo-_ 1_o 5

_ 1

_io 1__

_NO O

85960

- Vpr 3- Vpr 4

- Vif 2

982

p 100

FIG. 4. Structure and relative abundance of each alternatively spliced HIV-1 mRNA. (A) Splice site usage of the 1.8-kb class of HIV-1 mRNAamplified by PCR of randomly primed cDNA, using primers Odp.045 and Odp.032. (B) Splice site usage of the 4.0-kb class of HIV-1 mRNAamplified by PCR of randomly primed cDNA, using primers Odp.045 and Odp.070. The solid boxes raised above the line represent the regionsof retained RNA. Shown at the right of each panel is the relative proportion of each mRNA species quantified by phosphoimaging from pNL4-3virus infections of PBMC, using semiquantitative PCR analysis for the 1.8- and 4.0-kb classes of RNA. In general, values less than 1 were notmeasured at levels significantly above background and were recorded as zero; however, the existence of these cDNA species was evident on longgel exposures.

RNA in infected MT-2 cells (50). No evidence was found for avif RNA among the 1.8-kb species that spliced SD4 to SA7(vifl).

Splice site mutants of pNL4-3. During HIV-1 mRNA splic-ing, the cellular spliceosome cleaves at GT and AG dinucle-otides within the SD and SA motifs, respectively. Several ofthese highly conserved dinucleotide motifs present in thepNL4-3 proviral DNA clone of HIV-1 were replaced withdifferent dinucleotides to block RNA splicing, using a PCR-based strategy (Fig. 5). Seven site-directed provirus mutantswere generated by inactivating (i) the constitutively used majorsplice donor, pNLASD1; (ii) the competing splice acceptor forthe first coding exon of rev, pNLASA4b; (iii) the competitivelyselected major splice acceptor for env and vpu, pNLASA5; (iv)the cryptic splice acceptor within env purportedly used togenerate the Tev protein (5, 46, 47), pNLASA6; (v) SA7a, themost 5' of two conserved cryptic SA sequences mapping 34 and28 bases upstream from the second coding exon of tat and rev,designated pNLASA7a; (vi) both the SA7a and SA7b crypticsites, pNLASA7a+7b; or (vii) both of these in combinationwith SA7, the constitutive SA for the second coding exon of tatand rev, pNLASA7+7a+7b. Some of these mutations alsoaltered an amino acid codon(s) overlapping the splice site(Materials and Methods). The integrity of all mutant proviralclones was confirmed by nucleotide sequence analysis of thereconstructed regions.

Consensus motif

Mutant name

pNLASD1

pNLASA4b

pNLASA5

pNLASA6

pNLASA7a

pNLASA7a+7b

pNLASA7+7a+7b

CAG (1I AAGTA IG

3' Spilice Acceptor

TTTTTTTTTTTTN C,,2CCCCCCCCCCCC NTC G

CTGCAGAGTACGCCA'1

TGACAAAAGCCTTGAG41

CATCTCCTATGGClPCG4i

TGTGTTAGTTTAIC11T

TTCTATAGTGAATI(CA4,i "I

TTCTATAGTGAAT1PCAGTT1PCG'1 'I I

AAT1CAGTTTCGCAGGGATATTCACCATTATCGTTTCGPA

FIG. 5. Base substitutions introduced into SD and SA motifs of thepNL4-3 proviral clone of HIV-1. The consensus motifs of mammalianSD and SA sites are shown at the top, with the essential two

dinucleotides of the motif shown in outlined font. These dinucleotideswere changed in the pNL4-3 proviral clone to the bases shown inoutlined font so as to inactivate several SD and SA sites. Arrowsindicate the point of cleavage and ligation of RNA during splicing.

t -7-'W'st I

VOL. 67, 1993

5' Splice Donor

!t

on August 9, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

6372 PURCELL AND MARTIN

A. 9 eB.

gpl60/120o Z gp160/120p55

...,>.. ~~~~ ~ ~~~~~~~~~~~~.. p55

p24 ..IE...-.....

p17 _ _

36 hrs

p24

p17

72hrs1 2 3 4 5 6 7 8 9 0 1 2

FIG. 6. Some HIV-1 splicing mutants direct the synthesis of analtered profile of HIV-1 protein. Western blot analysis of HIV-1protein detected with patient serum from HeLa cell lysates wasperformed 48 h (A) and 72 h (B) after cotransfection of 0.5 pLg of agrowth hormone reporter plasmid with 20 pLg of wild-type pNL4-3(lane 1), with RNA splice mutant provirus pNLtASD1 (lane 2),pNLASA4b (lane 3), pNLASA5 (lane 4), pNLASA6 (lane 5),pNLASA7a (lane 6), pNLASA7a+7b (lane 7), or pNLASA7+7a+7b(lane 8), or with reporter plasmid alone (lanes 9 and 0). The volume ofcell lysate was standardized for transfection efficiency according to thedetermination of human growth hormone in the culture supernatants.

Protein synthesis by splicing mutants of HIV-1. In the firstgroup of experiments, the effects of the splice site mutations onviral protein production were assessed by Western blotting(immunoblotting) lysates from transfected HeLa cells, using anAIDS patient's serum. As shown in Fig. 6, each of the mutantsexcept pNLASA7+7a+7b directed the synthesis of the samecomplement of HIV-1 proteins as did wild-type pNL4-3, withthe following exceptions. Mutation of SD1 caused a markeddecrease in the quantity of viral proteins accumulating in HeLacells during the first 48 h despite transfection efficiency equiv-alent to that of the wild type, as measured by coexpression ofa human growth hormone expression plasmid (Fig. 6A, lanes 1and 2). By 72 h after transfection, however, HIV-1 proteinaccumulation directed by mutant plasmid pNLASD1 was com-parable to the wild-type level (Fig. 6B, lanes 1 and 2). Mutationof SA4b, used for the production of rev and env mRNAs, led toconsistent and substantial elevations of gpl60/120 levels com-pared with the wild type (Fig. 6A, lane 3). In contrast, themutation of SA5 markedly reduced, but did not eliminate,gpl60/120 production (Fig. 6A, lane 4). Mutation of SA6 (fortev mRNA) and the SA7a or SA7a+7b cryptic splice sites (forthe second coding exon of tat and rev) had little if any effect onHIV-1 protein synthesis (Fig. 6A, lanes 5 to 7), indicating thatthese three SA sites (and exons 6, 7a, and 7b) play nosignificant role for HIV-1 protein synthesis in transfectedHeLa cells. In contrast, mutant pNLzASA7+7a+7b, whichcontains a triple splice site mutation including SA7, thein-frame acceptor used constitutively for the second codingexon of tat and rev, directs the synthesis of a markedly alteredprotein profile (Fig. 6A, lane 8), similar to that previouslyreported for Rev-deficient HIV-1 mutants: minimal levels ofp55, p24, or pl7gag protein and no detectable gpl60/120 (12,14, 20, 33) accompanied by a novel and abundant 20-kDaprotein that reacted with Tat antiserum (data not shown). Noincreased accumulation of HIV-1 structural proteins was notedat later times (72 or 96 h after transfection) despite the

HeLa transfection ct

%99 99>D 9999>¢COrC5>rc>rtA2;c,~~~~~z,-P7SstvX=sS7?<

622bp I

527bp i

404bp

1 2 3 4 5 6 7 8 910FIG. 7. HIV-1 proviruses mutated at cryptic splice sites synthesize

a wild-type profile of RNA. Semiquantitative RT-PCR analysis of the1.8-kb class HIV-1 RNA from HeLa cells transfected with 0.5 pLg ofreporter plasmid alone (lane 2) or cotransfected with 20 ,utg of pNL4-3(lane 3) or HIV-1 RNA splicing mutant pNLASD1 (lane 4),pNLASA4b (lane 5), pNLASA5 (lane 6), pNLASA6 (lane 7),pNLASA7a (lane 8), pNLASA7a+7b (lane 9), or pNLASA7+7a+7b(lane 10). An MspI digest of pBR322 is shown on the left (lane 1) as asize marker.

expression of levels of total cellular HIV-1 RNA similar towild-type levels by Northern blot analysis (not shown). Theseresults suggest that mutant pNLASA7+7a+7b failed to ex-press functional Rev and resulted in reduced amounts ofRev-dependent cytoplasmic mRNAs encoding Gag and Envproteins.mRNA analysis from cryptic splice site mutants of HIV-1.

When the 1.8-kb class of RNA from HeLa cells transfectedwith HIV-1 proviral DNA clones containing splice site muta-tions was analyzed, it was clear that several expressed the sameRNA pattern as did wild-type pNL4-3 (Fig. 7). For example,mutants pNLASA6, pNLASA7a, and pNLASA7a+7b wereindistinguishable from the wild type (Fig. 7, lanes 7 to 9),indicating that the SA6, SA7a, and SA7b sites are not usedwith significant frequency by HIV-1NL43, even though it isvirtually identical to HIV-lHXB2, the strain in which thesesplice sites were originally described. Furthermore, no cDNAsusing the SA6, SA7a, or SA7b splice site were ever detectedamong the PCR-amplified bands that were directly sequencedfrom cells transfected or infected with derivatives of HIVLAI orHIV-lN"-3, and no clone generated from these PCR-ampli-fied cDNAs could be shown to utilize these splice sites. Thus,the SA6, SA7a, and SA7b cryptic sites are used extremelyrarely, if at all, and may be active only in the HIV-1LAIderivative HIV-lHxB2.We found that careful selection and testing of PCR primers

was required, in conjunction with the use of poly(A)+ RNAand nonspecific cDNA priming, to avoid the amplification ofaberrant cDNAs. Our semiquantitative PCR assays dependedon equal use of all the HIV-1 RNA species in either the 1.8- or4.0-kb mRNA classes as competing templates for amplifica-tion. Other protocols may selectively amplify transcripts usingSA6, SA7a, and SA7b (47): reports characterizing the se-

J. VIROL.

"I

n'.) ,cp

.+ V %,:.'1118 ll.v ..

..:

on August 9, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

RNA SPLICING MUTANTS OF HIV-1 6373

A. e B. Q

9~, 'o

Vpr3- -

Tat4Tat3.Tat2-

Tati- _ _ _

Nef5-Rev6-Nel4-

Revl,-Net2- F

Net t-__

Tat7-Tat6-Tat5- __

Envg-Env8- ..Env5-Env4-Env3-Env2- _Envi- -__

1 2 3 4 5 1 2 3 4 5FIG. 8. An altered profile of spliced RNA is synthesized by HIV-1

provirus mutated at constitutive or competing splice sites. Shown issemiquantitative RT-PCR analysis of the 1.8-kb class HIV-1 RNA (A)and the 4.4- to 5.5-kb class of HIV-1 RNA (B) from HeLa cellstransfected with 20 p.g of pNL4-3 (lane 1) or HIV-1 RNA splicingmutant pNLASD1 (lane 2), pNLA&SA4b (lane 3), pNLA&SA5 (lane 4),or pNLASA7+7a+7b (lane 5).

quence of cDNA clones around the SA7 splice site (15) or

using hybridization analyses of cDNAs (50) also failed to

detect usage of SA7a or SA7b.Analysis of mRNA from HIV-1 with mutated constitutive

and competitive splice sites. The cDNAs amplified from cellstransfected with mutant pNLASD1 had a pattern very similarto that of the wild type but with a slightly slower electro-phoretic mobility (compare lanes 3 and 4 in Fig. 7). Directsequence analysis of these PCR bands as well as severalindividual cDNA clones indicated that spliced RNA frompNLASD1 used a GT dinucleotide 4 nt downstream from themutated major SD (constitutive SD1) site as a cryptic SD sitein this HIV-1 proviral DNA. The HIV-1 proteins translatedfrom these slightly larger mRNAs were indistinguishable fromthe wild type (Fig. 6), but the rate of splicing of these viraltranscripts may be slower, perhaps explaining the delay inprotein expression previously observed. This cryptic SD (incor-rectly annotated as the major SD in the alignment of thisregion by Myers et al. [37]) is strongly conserved among allHIV-1 isolates. Because the cryptic SD1 site may be used onlywhen the genuine site is inactivated, it is likely that the strongconservation of this alternative splice site results from addi-tional selective pressures. Nevertheless, its existence greatlyreduces the possible loss of virus infectivity due to a sponta-neous mutation affecting the major SD that would otherwiseblock the production of functional spliced mRNAs.The cDNAs amplified from the remaining mutants differed

from the pNL4-3 wild-type pattern (Fig. 7 and 8). For theprovirus mutant inactivating the constitutive SA7 for thesecond coding exon of tat and rev, pNLASA7+7a+7b, every

cDNA band clearly differed in size compared with the wild typefollowing amplification of the 1.8-kb species of HIV-1 mRNA(Fig. 8A, lane 4); the overall pattern was similar to thewild-type pattern except that each band had a faster electro-

phoretic mobility. A direct sequence analysis of these PCRbands and sequencing of individual pNLASA7+7a+7b cDNAclones indicated that all used an AG dinucleotide situated 20nt downstream from SA7 as the alternative SA site; thisresulted in cDNAs that were 20 bases shorter than the wildtype. The activation of this cryptic downstream SA site,however, precluded the generation of mRNA capable ofencoding a functional Rev protein and resulted in an immu-noblot devoid of HIV-1 structural proteins (Fig. 6A, lane 8). Itshould be noted that all of the wild-type 1.8-kb spliced RNAswere transcribed as truncated species by mutant pNLASA7+7a+7b except for the nefl RNA, which was not detectedon the autoradiogram shown in Fig. 8A. The cDNA bandsobtained for pNLASA7+7a+7b after PCR for the 4.0-kbmRNA were identical to those of wild-type virus, indicatingthat RNA splicing to SA and SD sites upstream of SA7 was notaffected by the absence of functional two-exon Rev protein.This finding demonstrates that Rev is not required for splicesite selection in HIV-1 RNAs; however, protein expressionfrom RNAs containing an RRE clearly requires functional Revprotein.

Mutation of each of two competing SA sites, SA4b and SA5,caused a different usage frequency for neighboring SA sites butnot the activation of any cryptic sites. The first competing SAmutant, pNLASA4b, failed to generate bands for revl (Fig. 8A,lane 2), rev4, rev7, and revlO (evident after long gel exposures;not shown) following PCR with primers for the 1.8-kb mRNAspecies and for env2 (Fig. 8B, lane 2), env6, envlO, and envl4(evident after long gel exposures; not shown) after PCR withprimers for the 4.0-kb mRNA. The failure to detect thesebands was consistent with the absence of RNAs splicing to themutagenized SA4b site. All other cDNAs amplified frompNLASA4b were identical to wild-type cDNAs. The cDNApattern associated with the second competing SA mutant,pNLASA5, lacked several predominant nef species (nef2, nef3,nef4, and nef5) after PCR for the 1.8-kb mRNA (Fig. 8A, lane3), as well as the major env species (envl, env5, env8, andenvl3) after PCR for the 4.0-kb mRNA (Fig. 8B, lane 3),reflecting the absence of RNAs splicing to the mutagenizedSA5 site. In addition, there was a compensatory increase in theuse of SA4, SA4a, SA4b, and SA4c, resulting in increasedlevels of revl, rev2, rev3, tatl, tat2, tat3, tat4, tat5, tat6, tat7,env2, env3, and env4 mRNAs.

Translational consequences of altered RNA profile for com-petitive splice site mutants. Since the mRNAs using thecompeting SA4a, SA4b, SA4c, and SA5 splice sites are multi-cistronic, potentially coding several proteins from each mRNA(15, 48, 49), we examined the effect of altered proportions ofHIV-1 mRNAs on protein expression (Fig. 9). MutantpNLAiSA4b directed the synthesis of a significantly increasedlevel of envelope gpl60/120 and Nef but decreased amounts ofRev compared with wild-type provirus (Fig. 9A to C, lanes 3).Because Rev protein plays a prominent role in regulating thelevel of gpl60/120, we evaluated Rev functional activity bymeasuring the Rev-dependent rescue of CAT activity followingcotransfection of pNLASA4b with the pDM128 Rev reporterplasmid (Fig. 9E and F) (24). Again, Rev activity directed bypNLASA4b, as measured in this assay, was lower than thatdirected by wild-type pNL4-3. Thus, the observed elevation ofgpl60/120 was not due to any increase in Rev activity. Signif-icantly, the expression of both Rev-dependent gpl60/120 andRev-independent Nef proteins were increased in similar pro-portions. These proteins share SA5 as the predominant com-peting SA for their mRNAs. Thus, the increased synthesis ofgpl60/120 and Nef from the pNLASA4b mutant most likelyreflects the increased use of SA5 (and SA4c) and results in

VOL. 67, 1993

on August 9, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

6374 PURCELL AND MARTIN

A. '.

* gpl60/120

1 2 3 4

B.,15 y.eas* t,..,...

__ Nef

1 2 3 4

C.

*.- Rer)- 2 3

E. Mock pNL4-3

0 5 10.5 0.5 0.5

pNL.NSA4b

50.5 0.5

pNL.\SA5

5 1 Ltg provirus0.5 0.5 Ltg pDM128

* AcCMW. .....SA7

CM

0 14.0 13.7 3.1 0.4 26.2 11.6 % conversion

FIG. 9. Translational consequences of alternative usage of competing splice sites by mutant HIV-1 proviruses. Western blot analysis with rabbitserum to gpI60/120 (A) or Nef (B) and immunoprecipitation with rabbit serum to Rev from HeLa cell lysates prepared 48 h after cotransfectionwith 0.5 ,ug of human growth hormone reporter plasmid alone (lane 1) or with 20 ,ug of pNL4-3 (lane 2), pNLASA4b (lane 3), or pNLA&SA5 (lane4) (C). Lysate volumes were standardized according to human growth hormone determination. Functional Rev activity was measured with use ofthe pDM128 Rev-dependent reporter plasmid of Hope et al. (24) (D) by measuring the amount of CAT activity rescued due to Rev-RREinteraction after cotransfecting the indicated amounts of plasmids pDM128 and pNL4-3, pNLASA4b, pNLASA5 with 0.5 jig of human growthhormone reporter plasmid (E). The percentage of chloramphenicol (CM) converted to the acetylated forms (AcCM) is shown at the bottom. SV40,simian virus 40 promoter and enhancer; LTR, long terminal repeat.

increased synthesis of both proteins. This occurs despite theloss of mRNA species using SA4b which encode Rev. Sinceincreased use of SA4c does not prevent the reduction of Revexpression, if is likely that multicistronic mRNAs using SA4care less efficient for Rev expression than for Nef or gpl60/120expression.HeLa cells transfected with pNLASA5 express very low

levels of gpl60/120 and Nef proteins but elevated levels of Revcompared with wild-type pNL4-3 (Fig. 9A to C, lanes 4).Elevated Rev activity was also measured in the assay ofRev-dependent rescue of CAT activity, showing that the lowlevel of gpl60/120 expression did not result from any deficiencyin Rev function. The low-level expression of gpl60/120 mostlikely results from the inefficient translation of the multicis-tronic env2, env3, and env4 mRNAs, which are present atrelatively increased levels in cells transfected with pNLASA5compared with the wild-type pNL4-3 (Fig. 8B, lanes 1 and 3).This result indicates that the multicistronic env mRNAs aremarkedly less competent for gpl60/120 expression than is envl.pNLASA5 also fails to synthesize the major nef RNAs (nef2,through nef5; Fig. 4), although the levels of the nefl cDNAspecies, which results from splicing of SD1 directly to consti-tutive SA7, were equivalent in the pNLASA5- and wild-type-transfected cells (Fig. 8A). This low expression of Nef proteinindicates that the increased amount of bicistronic revinef andenvinef mRNAs encoded by mutant pNLASA5 fails to com-pensate for the loss of the predominant monocistronic nefmRNAs. The elevated levels of Rev protein observed demon-strate that the bicistronic revinef and envinef mRNAs encodeRev with significantly higher efficiency than Nef.

Infectivity of splicing mutants of HIV-1. The ability of theHIV-1 splice site mutants to generate progeny virions wasassessed by measuring the RT activity released into themedium following cotransfection of HeLa cells with proviral

and human growth hormone DNAs (Fig. 10A). All of thesplice site mutants except pNLASA7a+7b generated less par-ticle-associated RT than did the wild-type plasmid pNL4-3;pNLASA7+7a+7b and pNLASD1 produced only 2 and 15%of viral progeny, respectively, compared with the wild type.The infectivity of the splice site mutants was evaluated by

inoculating CEM (12D7) cells with equal amounts of virusharvested from transfected HeLa cell supernatants as deter-mined by RT assay (for wild-type pNL4-3, the multiplicity ofinfection was approximately 0.002) (Fig. lOB). Spreading in-fection was established by one of the two proviruses containinga mutated constitutive site. Mutant pNLASD1 was infectiousbut exhibited delayed replication kinetics compared with wild-type pNL4-3, reflecting the delayed kinetics of protein synthe-sis by this mutant (Fig. lOB). Lower efficiency of RNA splicingfrom the cryptic donor activated in pNLASD1 is very likelyresponsible for the delayed replication and infectivity kinetics.Mutant pNLASA7+7a+7b, which lacks the constitutive SA7site, could not infect CEM (12D7) cells, reflecting its inabilityto produce Rev and the HIV-1 structural proteins.

Mutations affecting competing SA sites in HIV-1 had differ-ent effects on viral infectivity. Mutant pNLASA4b, which failsto generate several HIV-1 mRNAs (revl, rev4, rev7, and revlOand env2, env6, envlO, and envl4) and consistently expressedelevated levels of gpl60/120, exhibited growth kinetics similarto the wild-type virus kinetics (Fig. lOB). In this case, alteredsplicing of viral mRNA had no effect on viral infectivity.Mutant pNLASA5, which lacks the processing site for themajor envl mRNA species, was not infectious despite beingable to synthesize low levels of gpl60/120 in transfected HeLacells by utilizing alternative SAs (Fig. 6A, lane 4; Fig. 9A).These contrasting results demonstrate that alternative splicingto competing SA sites can affect gpl60/120 production, leadingto synthesis of either fully infectious or defective virion.

D.

Rev dependent reporter plasmid pDM128

SD4

SV40 CAT RRE

* *. s ,* @ .

J. VIROL.

on August 9, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

RNA SPLICING MUTANTS OF HIV-1 6375

A.20000

- 15000

jk.E

a,6 10000

5000

0

Bo40000

'a

0.ie

30000 -

200000

10000

DISCUSSION

0

* pNL4-3 wildtype* pNLASDIO pNLASA4b0 pNLASA5O pNLASA613 pN1ASA7aa pNLASA7a+7bM pNLASA7+7a+7b* Mock

pNL4-3 wildtypepNLApNLA SA4b

pNLA SASpNLApNLA

pNLA SA7a+7b

pNLA SA7+7a+7b

Mock

0

--+-0---

20

Days post-infection

30

FIG. 10. Efficiency of virion production and infectivity of HIV-1splicing mutants. (A) Particle production by HIV-1 RNA splicingmutants was measured by using the accumulation of RT activity 40 hafter cotransfecting HeLa cells with 20 jig of mutant provirus plasmidsand 0.5 ,ug of human growth hormone reporter plasmid. The RTactivity was standardized for transfection efficiency by using the ratio ofhuman growth hormone compared with wild-type pNL4-3 virus. (B)Kinetics of infection of HIV-1 RNA splicing mutants following trans-fer of cell-free virion obtained from transfection supernatants (105 cpmfrom the RT assay) to CEM (12D7) cells.

The infectivity of mutants affecting the cryptic splice siteswas more subtle and could not be predicted from their abilityto generate virus particles in the transfection experimentsshown in Fig. 1OA. Mutant pNLASA7a was infectious, butpeak virus production was delayed 4 to 5 days compared withthe wild type. Mutant pNLASA7a+7b, which directed largeamounts of progeny virion production following transfection ofHeLa cells, was not infectious. Mutant pNLASA6 was also notinfectious, despite directing the synthesis of the wild-typecomplement and quantities of protein and spliced mRNAs(Fig. 6 and 7). These latter results could indicate that thecryptic SA6 and SA7b sites participate in some other aspect ofRNA processing (e.g., folding or branch point formation).Alternatively, amino acid changes introduced into the enve-

lope protein as a result of mutagenesis '(see Materials andMethods) may eliminate virus infectivity, possibly by affectinggpl60 processing as suggested by reduced amounts of gpl20observed in Fig. 6A. In a previous study, a similar mutation ofSA6 in the HIV-lHXB2 isolate resulted in the loss of infectivity,whereas SD5 mutants, also defective for Tev expression, had a

wild-type phenotype (16).

Relative proportions of alternatively spliced HIV-1 mRNAs.Various regulatory mechanisms control the expression andfunction of HIV-1 during a cycle of virus infection. Theregulation of RNA processing is one such prominent mecha-nism, and the balanced splicing of genomic length RNA into acomplex set of alternative RNA transcripts is required for thesynthesis of several viral proteins essential for replication.Several transcripts are capable of expressing each of theregulatory and accessory HIV-1 proteins, and most of thesetranscripts have the potential to encode two or more proteinswith different efficiencies. To evaluate the importance of thecomplex group of mRNAs synthesized during infection byHIV-1, we first rigorously determined the identities and rela-tive quantities of viral mRNAs resulting from transfection andinfection experiments. Our analysis shows that some RNAspecies are synthesized in preference to others. Generally, themost highly spliced forms of RNA that exclude noncodingexons are most common except in the case of nef, in which casethe inclusion of the 68-nt noncoding exon 5 is favored. Apreviously unrecognized SA site for rev- and env/vpu-encodingmRNAs (designated SA4c) was identified among a cluster ofcompeting SA sites in the tat coding sequence. This site wasselected at fivefold-lower frequency than the SA4a or SA4bsite for both rev and env mRNAs in PBMC infected withHIV-lNL-3 or HIV-lLAI. The SA4c site is used by many strainsof HIV-1 and is the predominant SA used for rev mRNA bysome HIV-1 strains (42). The addition of the new SA4c site tothe central competing SA sites (SA4, SA4a, SA4b, SA4c, andSA5) determined that 16 alternative mRNAs may encodegpl60/120. However, most of these exist at very low levels, andthe most common env mRNAs either used SA5 or excludedboth noncoding exons 2 and 3. The shortest possible envtranscript (envl) accounts for 80% of all env RNA directed byHIV-'NL4-3.Competing SA site usage determines gpl60/120 levels and

virus infectivity. Changing the balanced usage of competingsplice sites caused alterations in the proportions of both RNAand protein species and, in some cases, viral infectivity. HIV-1mutants containing changes affecting SA4b gave rise to anincreased proportion of the mRNA species using neighboringSA4a, SA4c, and SA5. This caused elevated expression ofenvelope gpl60/120 but no loss of virus infectivity. Increasedsynthesis of HIV-1 env mRNA using SA5, which more effi-ciently yields gpl60/120, is likely to explain the increase ingpl60/120. Mutation of the closely adjacent SA5, the major SAfor env mRNA, resulted in increased use of SA4a, SA4b, andSA4c but was accompanied by a marked reduction in both theexpression of envelope gp160/120 and virus infectivity. Thereduced levels of gpl60/120 were not associated with reducedRev activity, and Gag protein production was not altered.These results demonstrate that the level of expression ofenvelope proteins can be dramatically altered by forcingdifferent splicing patterns on HIV-lNLA-3 through the dispro-portionate usage of the seemingly redundant SA sites in thisregion of the viral genome.Two determinants could control the selection of the com-

peting SA sites if this alternative splicing mechanism were tooperate in vivo. First, the different sequence structure of thecompeting SA motifs (Fig. 3) or the branch point structure(s)in individual HIV-1 strains could alter the balance of the SAusage. The location of the branch point(s) for the competingHIV-1 SA sites is unknown. We have confirmed that HIV-1strains with different sequences have different splicing patternsin a survey of various HIV-1 isolates exhibiting variable

VOL. 67, 1993

on August 9, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

6376 PURCELL AND MARTIN

tropism (42). Second, the activation status or type of cellharboring an HIV-1 provirus could also affect the balance ofSA usage.Our analysis of splice site mutants suggest that an alteration

in the balanced selection of the competing SA sites may havea profound effect on the ability of some HIV-1-infected cells toproduce infectious virion. The potential deficit of gpl60/120protein resulting from underutilization of SA5 in a wild-typeHIV-1 isolate may reduce the infectivity of viral progeny byinterfering with adsorption-penetration step of the virus lifecycle. This mechanism may contribute to some of the nonin-fectious virus present in the plasma of infected individuals (4,11, 39, 41).

Biological significance of multicistronic HIV-1 mRNAs.Many of the alternative HIV-1 mRNAs are multicistronic,encoding more than one protein when tested in in vitrotranslation systems (15, 47). Some of the splice site mutantsdescribed above selectively eliminate several of these mRNAsfrom the HIV-1 RNA pool, allowing an evaluation of thesedepleted transcripts in the context of virus replication. Theinactivation of SA5 from provirus mutant pNLASA5 precludedthe synthesis of the major species of env mRNA (envl) as wellas the predominant nef mRNAs (nef2 through nef5). Duringproductive infection of PBMC with wild-type HIV-1, envl andnef2 through nef5 mRNAs comprise 80% of the total env and95% of the total nef transcripts, respectively. Intuitively, theelimination of these mRNAs could be compensated for by theincreased use of other neighboring SAs which could be used tosynthesize alternate multicistronic mRNAs with coding poten-tial for the deficient Env and Nef proteins. Our analysesshowed that multicistronic mRNAs generated only limitedamounts of Nef and gpl60/120 proteins which proved inade-quate for the production of infectious progeny virion. Aftertransfection of HeLa cells with mutant pNLL\SA5, minimalsynthesis of gpl60/120 (-2% of the wild-type level) and lowlevels of Nef (-12% of the wild-type level) were measured. Inour studies, bicistronic rev mRNAs encoded Nef protein with2- to 3-fold-lower efficiency than did monocistronic nefmRNAs (in agreement with results of Schwartz et al. [49]) andgpl60/120 was translated from envl mRNA approximately25-fold more efficiently than from other env mRNAs. Giventhe proportions of the different HIV-1 mRNAs measured inthis study, neither multicistronic env RNA containing the Rev,Vpu, and Env translation initiation sites (env2 through env5)nor the multicistronic rev RNA species, also capable of direct-ing the synthesis of Nef (15, 47), significantly contributes to thetotal cellular pool of gpl60/120 or Nef proteins during virusreplication. In contrast, a mutation affecting one of the SAs forrev mRNA (SA4b) resulted in rev mRNAs using SA4a andSA4c that fully compensated for altered SA usage and pro-duced wild-type levels of Rev functional activity.

Mutation of constitutive splice sites activates cryptic sites.Mutation of the major SD (SD1) in the pNL4-3 provirusslowed the kinetics of RNA and protein synthesis and thekinetics of a spreading virus infection. An alternative crypticmajor SD signal, four bases downstream, was activated. Be-cause the cryptic SDI site may be used only when the genuinesite is inactivated, it is likely that the strong conservation of thisalternative splice site results from additional selective pres-sures. Nevertheless, its existence greatly reduces the possibleloss of virus infectivity due to a spontaneous mutation affectingthe major SD that would otherwise block the production offunctional spliced mRNAs. The efficiency of RNA splicingfrom this cryptic donor ( 3AGA l GUACGCC+7) may belower than that from the genuine donor site ( 3CUG l GUGAGUA+7) and would therefore be responsible for the

exon ;5'SD intron

adenovirus GGGGU G A G U A C U

pNL4-3 GGU G A G U A C

pNLASD1 ACCA-3 -2 -1 +1 +2 +3 +4 +5 +6 +7 +8 +9

snRNA t t t t t t t I tinteractionsmappedto t t t t t t Tadenovirusmajor late SD t t t t t

major late SD

genuine SD1cryptic SD1

Ul snRNA

U5 snRNA

U6 snRNA

FIG. 11. HIV-1 SD1 has greater potential to interact with Ul, U5,and U6 snRNAs than does the cryptic SD1. Shown is alignment of theSD site of adenovirus major late pre-mRNA with the genuine andcryptic SD1 sites of pNL4-3 and pNLASD1; homologous nucleotidesare boxed. Locations of the adenovirus nucleotides identified byWassarman and Steitz (56) as strongly (4) or weakly interacting (t)with Ul, U5, and U6 snRNAs are shown at the bottom.

delayed replication and infection kinetics of mutantpNLASD1. Comparison of these two HIV-1 SD motifs with theadenovirus major late 5' SD (-3GGG I GUGAGUA+7) (Fig.11), used to map nucleotide interactions with U1, U5, and U6small nuclear RNAs (snRNAs), provides an explanation forthe possible lower splicing efficiency of the cryptic SD1 (56).The nucleotides at sites of strong interaction between theadenovirus SD and the Ul, US, and U6 snRNAs of thespliceosome are the same for wild-type HIV-1 SD1 but mis-matched at several positions for the cryptic HIV-1 SD1,potentially destabilizing the interaction of the cryptic SD1 withthe spliceosome (Fig. 11). The decreased affinity of the crypticHIV-1 SD with several important spliceosomal snRNAs mayexplain the delayed infection kinetics of mutant pNLASD1.

Mutation of SA7, in conjunction with SA7a and SA7b,resulted in the expression of aberrant RNA and protein speciesand the loss of virus infectivity as a consequence of theactivation of a downstream cryptic SA site. The use of thislatter site shifted the reading frame for the second coding exonof Rev, thereby causing a concomitant loss of Rev function.Despite the apparent lack of expression of functional Rev bymutant pNLASA7+7a+7b, all other HIV-1 SD and SA siteusage was equivalent to that of wild-type pNL4-3, indicatingthat the two-exon Rev protein is not required for the selectionof splice sites in HIV-1 RNAs with or without an RRE.However, assembly of the spliceosome may still be required forRev to activate RNA containing an RRE (6).

Significance of cryptic splice sites in HIV-1 replication. Wefound that three previously mapped splice acceptor sites, SA6,SA7a, and SA7b (47), were never used during HIV-1NL4-3replication in PBMC and HeLa cells and that mutation ofthese cryptic sites had no apparent effect on HIV-1 RNA orprotein synthesis. These sites were unlikely candidates ascommonly used ligation points for the synthesis of maturemRNA because HIV-1 isolates, other than derivatives ofHIV-lLAI, do not possess the SA6 splicing signal (37), and theuse of SA7a and SA7b sites would preclude the in-frame entryinto the second coding exon of rev and tat, leading to thesynthesis of novel Rev-Tat-Env chimeric translation productsand no functional Rev protein. The ability of mutantpNLASA7a to suc&essfully replicate and infect human cellsfurther demonstrates that this site is not essential in the HIV-1life cycle. Unexpectedly, mutants pNL/SA6 and pNLASA7a+7b were not infectious, suggesting that these sites might playsome other role in RNA processing or that the amino acidstructure around these mutagenized SAs is critical for infec-tivity. Mutational analysis of the tev splicing signals (SD5) in

J. VIROL.

on August 9, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

RNA SPLICING MUTANTS OF HIV-1 6377

the context of HIV-1HIB indicated that Tev was not requiredfor virus infections of CD4+ T-cell lines and activated PBMC.However, the tev splicing signal (SA6) in HIV-1HIB stabilizedneighboring suboptimal splice sites and maintained the bal-anced proportion of spliced and unspliced RNAs (16). The tevsplicing pattern may be limited to certain derivatives of HIV-1L.A and could have arisen as a result of inefficient splicingsignals throughout the primary RNA transcript in this isolate.

In contrast to the uncertain role of cryptic splice sites inHIV-1, changing the usage of the competing splice sites in thecenter of the HIV-1 genome caused alterations in the propor-tions of both HIV-1 RNA and protein species, and this affectedviral infectivity. Similarly, mutation of the major SD motifmarkedly reduced the synthesis of viral RNA and protein aswell as viral infectivity. Therefore, we have shown that alter-native splicing of mature HIV-1 mRNA is yet another poten-tial mechanism for the regulation of HIV-1 expression. Thus,the multiple redundant splice signals in the central region ofthe HIV-1 genome play a greater role than simply providingredundant strategies for mRNA synthesis in the event ofunwanted mutations at these sites.

ACKNOWLEDGMENTS

We thank A. Buckler-White for oligonucleotide synthesis and DNAsequencing, K. Peden for providing the pLAI provirus clone, T. G.Parslow for pDM128, G. Pavlakis for antiserum to Rev, and K. Strebelfor antisera to gpl60/120. For helpful discussions and critical review ofthe manuscript, we thank K.-T. Jeang, F. Maldarelli, E. Freed, G.Englund, R. Willey, K. Peden, and S. Venkatessan.D. F. J. Purcell was supported by a C. J. Martin fellowship from the

National Health and Medical Research Council of Australia.

REFERENCES1. Adachi, A., H. E. Gendelman, S. Koenig, T. Folks, R. Willey, A.

Rabson, and M. A. Martin. 1986. Production of acquired immu-nodeficiency syndrome-associated retrovirus in human and non-human cells transfected with an infectious molecular clone. J.Virol. 59:284-291.

2. Amigo, S., and I. S. Chen. 1991. Rev is necessary for translationbut not cytoplasmic accumulation of HIV-1 vif, vpr, and envlvpu 2RNAs. Genes Dev. 5:808-819.

3. Arrigo, S., S. Weitsman, J. A. Zack, and I. S. Chen. 1990.Characterization and expression of novel singly spliced RNAspecies of human immunodeficiency virus type 1. J. Virol. 64:4585-4588.

4. Bagnarelli, P., S. Menzo, A. Valenza, A. Manzin, M. Giacca, F.Ancarani, G. Scalise, P. E. Varaldo, and M. Clementi. 1992.Molecular profile of human immunodeficiency virus type 1 infec-tion in symptomless patients and in patients with AIDS. J. Virol.66:7328-7335.

5. Benko, D. M., S. Schwartz, G. N. Pavlakis, and B. K. Felber. 1990.A novel human immunodeficiency virus type 1 protein, tev, sharessequences with tat, env, and rev proteins. J. Virol. 64:2505-2518.

6. Chang, D. D., and P. A. Sharp. 1989. Regulation by HIV Revdepends upon recognition of splice sites. Cell 59:789-795.

7. D'Agostino, D. M., B. K. Felber, J. E. Harrison, and G. N.Pavlakis. 1992. The Rev protein of human immunodeficiency virustype 1 promotes polysomal association and translation of gag/poland vpu/env mRNAs. Mol. Cell Biol. 12:1375-1386.

8. Dayton, A., J. G. Sodroski, C. A. Rosen, W. C. Goh, and W. A.Haseltine. 1986. The trans-activator gene of the human T-celllymphotropic cirus type III is required for replication. Cell 44:941-947.

9. Dayton, E. T., D. M. Powell, and A. I. Dayton. 1989. Functionalanalysis of CAR, the target sequence for the Rev protein ofHIV-1. Science 246:1625-1629.

10. Denolo, F. M., D. D. Moore, and H. M. Goodman. 1981. Humangrowth hormone DNA sequence and mRNA structure: possiblealternative splicing. Nucleic Acids Res. 9:3719-3730.

11. Embretson, J., M. Zupanic, J. L. Ribas, A. Burke, P. Racz, K.

Tenner-Racz, and A. Haase. 1993. Massive covert infection ofhelper T lymphocytes and macrophages by HIV during theincubation period of AIDS. Nature (London) 362:359-362.

12. Emerman, M., R. Vazeux, and K. Peden. 1989. The rev geneproduct of the human immunodeficiency virus affects envelope-specific RNA localization. Cell 57:1155-1165.

13. Feinberg, M. B., R. F. Jarrett, A. Aldovini, R. C. Gallo, and F.Wong-Staal. 1986. HTLV-III expression and production involvecomplex regulation at the levels of splicing and translation of viralRNA. Cell 46:807-817.

14. Felber, B. K., M. Hadzopoulou-Cladaras, C. Cladaras, T. Cope-land, and G. N. Pavlakis. 1989. Rev protein of human immuno-deficiency virus type 1 affects the stability and transport of the viralmRNA. Proc. Natl. Acad. Sci. USA 86:1495-1499.

15. Furtado, M. R., R. Balachandran, P. Gupta, and S. M. Wolinsky.1991. Analysis of alternatively spliced human immunodeficiencyvirus type-1 mRNA species, one of which encodes a novelTAT-ENV fusion protein. Virology 185:258-270.

16. Gottlinger, H. G., T. Dorfman, E. A. Cohen, and W. A. Haseltine.1992. The role of tnv protein and tnv RNA splicing signals inreplication of HIV-1 IIIB isolates. Virology 189:618-628.

17. Green, M. R. 1991. Biochemical mechanisms of constitutive andregulated pre-mRNA splicing. Annu. Rev. Cell Biol. 7:559-599.

18. Guatelli, J. C., T. R. Gingeras, and D. Richman. 1990. Alternativesplice acceptor utilization during human immunodeficiency virustype 1 infection of cultured cells. J. Virol. 64:4093-4098.

19. Hadzopoulou-Cladaras, M., B. K. Felber, C. Cladaras, A. Atha-nassopoulos, A. Tse, and G. N. Pavlakis. 1989. The rev (trs/art)protein of human immunodeficiency virus type 1 affects viralmRNA and protein expression via a cis-acting sequence in the envregion. J. Virol. 63:1265-1274.

20. Hammarskjold, M., J. Heimer, B. Hammarskjold, I. Sangwan, L.Albert, and D. Rekosh. 1989. Regulation of human immunodefi-ciency virus env expression by the rev gene product. J. Virol.63:1959-1966.

21. Hammes, S. R., E. P. Dixon, M. H. Malim, B. R. Cullen, and W. C.Green. 1989. Nef protein of human immunodeficiency virus type 1:evidence against its role as a transcriptional inhibitor. Proc. Natl.Acad. Sci. USA 86:9549-9553.

22. Heaphy, S., C. Dingwall, I. Ernberg, M. J. Gait, S. M. Green, J.Karn, A. D. Lowe, M. Singh, and M. A. Skinner. 1990. HIV-1regulator of virion expression (Rev) binds to an RNA stem-loopstructure located in the Rev-response element region. Cell 60:685-693.

23. Holland, S. M., N. Ahmad, R. K. Maitra, P. Wingfield, and S.Venkatesan. 1990. Human immunodeficiency virus Rev proteinrecognizes a target sequence in Rev-responsive element RNAwithin the context of RNA secondary structure. J. Virol. 64:5966-5975.

24. Hope, T. J., X. J. Huang, D. McDonald, and T. G. Parslow. 1990.Steroid-receptor fusion of the human immunodeficiency virus type1 Rev transactivator: mapping cryptic functions of the arginine-rich motif. Proc. Natl. Acad. Sci. USA 87:7787-7791.

25. Katz, R. A., and A. M. Skalka. 1990. Control of retroviral RNAsplicing through maintenance of suboptimal processing signals.Mol. Cell. Biol. 10:696-704.

26. Kestler, H. W., D. J. Ringler, K. Mori, D. L. Panicali, P. K. Sehgal,M. D. Daniel, and R. C. Desrosiers. 1991. Importance of the nefgene for maintenance of high virus loads and for development ofAIDS. Cell 65:651-662.

27. Kim, S., R. Byrn, J. Groopman, and D. Baltimore. 1989. Temporalaspects of DNA and RNA synthesis during human immunodefi-ciency virus infection: evidence for differential gene expression. J.Virol. 63:3708-3713.

28. Kjems, J., A. D. Frankel, and P. A. Sharp. 1991. Specific regulationof mRNA splicing in vitro by a peptide from HIV-1 Rev. Cell67:169-178.

29. Klotman, M. E., S. Kim, A. Buchbinder, A. DeRossi, D. Baltimore,and F. Wong-Staal. 1991. Kinetics of expression of multiply splicedRNA in early human immunodeficiency virus type 1 infection oflymphocytes and monocytes. Proc. Natl. Acad. Sci. USA 88:501 1-5015.

30. Kozak, M. 1991. An analysis of vertebrate mRNA sequences:

VOL. 67, 1993

on August 9, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

6378 PURCELL AND MARTIN

intimations of translational control. J. Cell Biol. 115:887-903.31. Landt, O., H. P. Grunert, and U. Hahn. 1990. A general method

for rapid site-directed mutagenesis using the polymerase chainreaction. Gene 96:125-128.

32. Lang, S. M., M. Weeger, C. Stahl-Hennig, C. Coulibaly, G.Hunsmann, J. Muller, H. Muller-Hermelink, D. Fuchs, H.Wachter, M. M. Daniel, R. C. Desrosiers, and B. Fleckenstein.1993. Importance of vpr for infection of rhesus monkeys withsimian immunodeficiency virus. J. Virol. 67:902-912.

33. Malim, M. H., J. Hauber, S. Le, J. V. Maizel, and B. R. Cullen.1989. The HIV-1 Rev trans-activator acts through a structuredtarget sequence to activate nuclear export of unspliced viralmRNA. Nature (London) 338:254-257.

34. McKeown, M. 1992. Alternative mRNA splicing. Annu. Rev. CellBiol. 8:133-155.

35. Muesing, M. A., D. H. Smith, C. D. Cabradilla, C. V. Benton, L. A.Lasky, and D. J. Capon. 1985. Nucleic acid structure and expres-sion of the human AIDS/lymphadenopathy virus. Nature (Lon-don) 313:450-458.

36. Muesing, M. A., D. H. Smith, and D. J. Capon. 1987. Regulationof mRNA accumulation by a human immunodeficiency virustrans-activator protein. Cell 48:691-701.

37. Myers, G., B. Korber, J. A. Berzofsky, R. F. Smith, and G. N.Pavlakis. 1992. Human retroviruses and AIDS 1992. Los AlamosNational Laboratory, Los Alamos, N.Mex.

38. Olsen, H. S., P. Nelbock, A. W. Cochrane, and C. A. Rosen. 1990.Secondary structure is the major determinant for interaction ofHIV Rev protein with RNA. Science 247:845-848.

39. Pantaleo, G., C. Graziosi, J. F. Demerest, L. Butini, M. Montroni,C. H. Fox, J. M. Orenstein, D. P. Kotler, and A. S. Fauci. 1993.HIV infection is active and progressive in lymphoid tissue duringthe clinically latent stage of disease. Nature (London) 362:355-358.

40. Peterlin, B. M., P. A. Luciw, P. J. Barr, and M. D. Walker. 1986.Elevated levels of mRNA can account for the trans-activation ofhuman immunodeficiency virus. Proc. Natl. Acad. Sci. USA 83:9734-9738.

41. Piatak, M., Jr., M. S. Saag, L. C. Yang, S. J. Clark, J. C. Kappes,K.-C. Luk, B. H. Hahn, G. M. Shaw, and J. D. Lifson. 1993. Highlevels of HIV-1 in plasma during all stages of infection determinedby competitive PCR. Science 259:1749-1754.

42. Purcell, D. F. J., G. Englund, and M. A. Martin. Unpublished data.43. Robert-Guroff, M., M. Popovic, S. Gartner, P. Markham, R. C.

Gallo, and M. S. Reitz. 1990. Structure and expression of tat-, rev-,and nef-specific transcripts of human immunodeficiency virus type1 in infected lymphocytes and macrophages. J. Virol. 64:3391-3398.

44. Russell, S. M., R. L. Sparrow, I. F. C. McKenzie, and D. F. J.Purcell. 1992. Tissue-specific and allelic expression of the comple-ment regulator CD46 is controlled by alternative splicing. Eur. J.Immunol. 22:1513-1518.

45. Sadaie, M. R., J. Rappaport, T. Benter, S. F. Josephs, R. Willis,

and F. Wong-Staal. 1988. Missense mutations in an infectioushuman immunodeficiency viral genome: functional mapping of tatand identification of the rev splice acceptor. Proc. Natl. Acad. Sci.USA 85:9224-9228.

46. Salfeld, J., H. Gottlinger, R. Sia, R. Park, J. Sodroski, and W.Haseltine. 1990. A tripartate HIV-1 tat-env-rev fusion protein.EMBO J. 9:965-970.

47. Schwartz, S., B. K. Felber, D. M. Benko, E. M. Fenyo, and G. N.Pavlakis. 1990. Cloning and functional analysis of multiply splicedmRNA species of human immunodeficiency virus type 1. J. Virol.64:2519-2529.

48. Schwartz, S., B. K. Felber, E. M. Fenyo, and G. N. Pavlakis. 1990.Env and Vpu proteins of human immunodeficiency virus type 1are produced from multiple bicistronic mRNAs. J. Virol. 64:5448-5456.

49. Schwartz, S., B. K. Felber, and G. N. Pavlakis. 1992. Mechanismof translation of monocistronic and multicistronic human immu-nodeficiency virus type I mRNAs. Mol. Cell Biol. 12:207-219.

50. Smith, J., A. Azad, and N. J. Deacon. 1992. Identification of twonovel human-immunodeficiency virus type 1 splice acceptor sitesin infected T-cell lines. J. Gen. Virol. 73:1825-1828.

51. Sodroski, J. G., R. Patarca, C. A. Rosen, F. Wong-Staal, and W. A.Haseltine. 1985. Location of the trans-activating region on thegenome of human T-cell lymphotropic virus type III. Science229:74-77.

52. Strebel, K., D. Daugherty, K. Clouse, D. Cohen, T. Folks, andM. A. Martin. 1987. The HIV "A" (sor) gene product is essentialfor virus infectivity. Nature (London) 328:728-730.

53. Strebel, K., T. Klimkait, and M. A. Martin. 1988. A novel gene ofHIV-1, vpu, and its 19-kilodalton product. Science 241:1221-1223.

54. Streuli, M., L. R. Hall, Y. Saga, S. F. Schlossman, and H. Saito.1987. Differential usage of three exons generates at least fivedifferent mRNAs encoding human leukocyte common antigens. J.Exp. Med. 166:1548-1566.

55. Streuli, M., and H. Saito. 1989. Regulation of tissue-specificalternative splicing: exon-specific cis-elements govern the splicingof leukocyte common antigen. EMBO J. 8:787-796.

56. Wassarman, D. A., and J. A. Steitz. 1992. Interactions of smallnuclear RNA's with precursor messenger RNA during in vitrosplicing. Science 257:1918-1925.

57. Wigler, M., A. Pellicer, S. Silverstein, and R. Axel. 1978. Biochem-ical transfer of single-copy eucaryotic genes using total cellularDNA as donor. Cell 14:725-731.

58. Willey, R. L., T. Klimkait, D. M. Frucht, J. S. Bonifacino, andM. A. Martin. 1991. Mutations within the human immunodefi-ciency virus type 1 gpl60 envelope glycoprotein alter its intracel-lular transport and processing. Virology 184:319-329.

59. Willey, R. L., D. H. Smith, L. A. Lasky, T. S. Theodore, P. L. Earl,B. Moss, D. J. Capon, and M. A. Martin. 1988. In vitro mutagen-esis identifies a region within the envelope gene of the humanimmunodeficiency virus that is critical for infectivity. J. Virol.62:139-147.

J. VIROL.

on August 9, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from