Vol. 63, No. 10 Truncated gag Products Encoded by Gv-1-Responsive Endogenous Retrovirus Locit PAUL F. POLICASTRO,1t MERETE FREDHOLM,"2 AND MICHAEL C. WILSON'* Department of Molecular Biology, Research Institute of Scripps Clinic, 10666 North Torrey Pines Road, La Jolla, California 92037,1 and Department of Animal Genetics, The Royal Veterinary and Agricultural University, Bulowsvej 13, 1870 Frederiksberg C, Denmark2 Received 7 March 1989/Accepted 16 June 1989 The conversion of endogenous or exogenous murine retroviruses to a leukemogenic phenotype involves recombination with retroviral sequences present in host genomic DNA. In the 129 Gix+ inbred strain, these endogenous sequences are replication defective but still express retroviral proteins under the apparent transcriptional control of the Gv-1 regulatory locus. To study the protein-coding potential of Gv-1-regulated endogenous retroviral loci, we used oligonucleotide probes directed to env deletion breakpoints identified in previously characterized cDNA clones. Four endogenous retroviral loci were isolated from a library of 129 Gi.+ genomic DNA with these probes. Three loci cloned with the env deletion probe del env-1 had virtually identical proviral inserts by restriction analysis. A unique locus was identified and cloned with the del env-2 probe, which must therefore represent a Gv-1-responsive element. Restriction enzyme and nucleotide sequence analyses indicated that the del env-1 and del env-2 loci represented members of the polytropic and modified polytropic classes of endogenous retrovirus, respectively. Despite this divergence, members of both classes contained identical deletions of 19 nucleotides within p309g9 and of 1,474 nucleotides from plO9ag into the reverse transcriptase-coding region of pol, suggesting that a recombination event had occurred between these proviral sequences prior to insertion within the genome. The del env-1 and del env-2 loci retained coding capacity for truncated gag polyproteins, confirmed by in vitro translation and immunoprecipitation of the protein products. Nucleotide sequence comparison of the untranslated leader (L) regions of the del env-l and del env-2 loci to a replication-competent ecotropic virus indicated regions that might be important to dispersion of these endogenous retroviral elements throughout the host genome. Replication-defective type C retrovirus sequences present in the mouse genome recombine with infectious ecotropic viruses to produce leukemogenic, polytropic recombinant viruses (9, 15, 17, 22, 25). Significantly, these recombinant viruses have exchanged sequence largely including the gp7O coding region of the env gene, which is thought to define retroviral tropism through either attachment and/or uptake of virus particles. The ecotropic viruses infect only mouse cells, while the recombinant polytropic viruses infect cells of both mouse and other species, and xenotropic viruses infect only non-mouse cells. The 30 to 50 endogenous retrovirus loci in any given inbred mouse strain consist of only a few or no ecotropic loci (21) and a variable number (5 to 20 of each) of loci bearing xenotropic (39, 42), polytropic, or modified polytropic env sequences (54). The modified polytropic class contains 18 amino acid changes and a nine-codon deletion within the env coding region relative to the polytropic class, without losing the capacity to contribute functional env sequence to infectious virus (53). The gp7O-coding 5' env sequences from a polytropic or modified polytropic locus are thought to contribute one obligate phenotype in the leuke- mogenic conversion process; the plSE-coding 3' env and adjoining U3 long terminal repeat (LTR) sequences from a xenotropic locus contribute the second (9, 10, 15, 22, 24, 42). Of the many polytropic and modified polytropic loci present * Corresponding author. t Publication 5789-MB from the Research Institute of Scripps Clinic. t Present address: Laboratory of Microbial Structure and Func- tion, National Institute of Allergy and Infectious Diseases, Rocky Mountain Laboratories, Hamilton, MT 59840. in the genome of an inbred strain, not all will be capable of providing the specific env sequences necessary for success- ful phenotype conversion of a replicating ecotropic virus. In addition to point mutations and short deletions leading to a shift in or loss of the translational reading frame (53), loci and transcripts have been characterized that are deleted for the majority of the env coding region (25, 28, 32). Recombi- nations involving the env region are easily detected antigen- ically by changes in the gp7O envelope protein in the leuke- mogenic conversion process, but endogenous loci deleted for env sequences may act as recombination partners in other regions of the retroviral genome. A Moloney murine leukemia virus (MoMuLV) mutant, normally NB-tropic but expressing a truncated 45-kilodalton (kDa) gag product and containing a 1-kilobase (kb) deletion, was rescued from the JLS-V11-NP cell line as an N-tropic, large XC plaque recombinant, presumably by recombination within gag with the endogenous BALB/c XC-negative, N-tropic retrovirus (5). MoMuLV mutants deleted within either the 3' end of the LTR (12) or the reverse transcriptase-coding region of pol (45) have been isolated as revertants from NIH 3T3 cells bearing new sequences that derive from recombination with one of several endogenous retroviral loci. The frequency of biologically significant recombination between infectious virus and endogenous loci may depend on RNA template exchange during reverse transcription (41) rather than on DNA-mediated recombination involving accessible endoge- nous loci in host chromatin. Since working models of leuke- mogenic conversion place env and U3 LTR recombinational events in the spleen and thymus (15, 23, 28), it is possible that endogenous loci involved in the recombination process exhibit tissue-specific expression. 4136 JOURNAL OF VIROLOGY, OCt. 1989, p. 4136-4147 0022-538X/89/104136-12$02.00/0 Copyright C) 1989, American Society for Microbiology on December 24, 2018 by guest http://jvi.asm.org/ Downloaded from
Transcript
Vol. 63, No. 10
Truncated gag Products Encoded by Gv-1-Responsive EndogenousRetrovirus Locit
PAUL F. POLICASTRO,1t MERETE FREDHOLM,"2 AND MICHAEL C. WILSON'*Department of Molecular Biology, Research Institute of Scripps Clinic, 10666 North Torrey Pines Road, La Jolla,
California 92037,1 and Department ofAnimal Genetics, The Royal Veterinary and Agricultural University,Bulowsvej 13, 1870 Frederiksberg C, Denmark2
Received 7 March 1989/Accepted 16 June 1989
The conversion of endogenous or exogenous murine retroviruses to a leukemogenic phenotype involvesrecombination with retroviral sequences present in host genomic DNA. In the 129 Gix+ inbred strain, theseendogenous sequences are replication defective but still express retroviral proteins under the apparenttranscriptional control of the Gv-1 regulatory locus. To study the protein-coding potential of Gv-1-regulatedendogenous retroviral loci, we used oligonucleotide probes directed to env deletion breakpoints identified inpreviously characterized cDNA clones. Four endogenous retroviral loci were isolated from a library of 129 Gi.+genomic DNA with these probes. Three loci cloned with the env deletion probe del env-1 had virtually identicalproviral inserts by restriction analysis. A unique locus was identified and cloned with the del env-2 probe, whichmust therefore represent a Gv-1-responsive element. Restriction enzyme and nucleotide sequence analysesindicated that the del env-1 and del env-2 loci represented members of the polytropic and modified polytropicclasses of endogenous retrovirus, respectively. Despite this divergence, members of both classes containedidentical deletions of 19 nucleotides within p309g9 and of 1,474 nucleotides from plO9ag into the reversetranscriptase-coding region of pol, suggesting that a recombination event had occurred between these proviralsequences prior to insertion within the genome. The del env-1 and del env-2 loci retained coding capacity fortruncated gag polyproteins, confirmed by in vitro translation and immunoprecipitation of the protein products.Nucleotide sequence comparison of the untranslated leader (L) regions of the del env-l and del env-2 loci to areplication-competent ecotropic virus indicated regions that might be important to dispersion of theseendogenous retroviral elements throughout the host genome.
Replication-defective type C retrovirus sequences presentin the mouse genome recombine with infectious ecotropicviruses to produce leukemogenic, polytropic recombinantviruses (9, 15, 17, 22, 25). Significantly, these recombinantviruses have exchanged sequence largely including the gp7Ocoding region of the env gene, which is thought to defineretroviral tropism through either attachment and/or uptakeof virus particles. The ecotropic viruses infect only mousecells, while the recombinant polytropic viruses infect cells ofboth mouse and other species, and xenotropic viruses infectonly non-mouse cells. The 30 to 50 endogenous retrovirusloci in any given inbred mouse strain consist of only a few orno ecotropic loci (21) and a variable number (5 to 20 of each)of loci bearing xenotropic (39, 42), polytropic, or modifiedpolytropic env sequences (54). The modified polytropic classcontains 18 amino acid changes and a nine-codon deletionwithin the env coding region relative to the polytropic class,without losing the capacity to contribute functional env
sequence to infectious virus (53). The gp7O-coding 5' env
sequences from a polytropic or modified polytropic locus arethought to contribute one obligate phenotype in the leuke-mogenic conversion process; the plSE-coding 3' env andadjoining U3 long terminal repeat (LTR) sequences from axenotropic locus contribute the second (9, 10, 15, 22, 24, 42).Of the many polytropic and modified polytropic loci present
* Corresponding author.t Publication 5789-MB from the Research Institute of Scripps
Clinic.t Present address: Laboratory of Microbial Structure and Func-
tion, National Institute of Allergy and Infectious Diseases, RockyMountain Laboratories, Hamilton, MT 59840.
in the genome of an inbred strain, not all will be capable ofproviding the specific env sequences necessary for success-ful phenotype conversion of a replicating ecotropic virus. Inaddition to point mutations and short deletions leading to ashift in or loss of the translational reading frame (53), lociand transcripts have been characterized that are deleted forthe majority of the env coding region (25, 28, 32). Recombi-nations involving the env region are easily detected antigen-ically by changes in the gp7O envelope protein in the leuke-mogenic conversion process, but endogenous loci deletedfor env sequences may act as recombination partners inother regions of the retroviral genome. A Moloney murineleukemia virus (MoMuLV) mutant, normally NB-tropic butexpressing a truncated 45-kilodalton (kDa) gag product andcontaining a 1-kilobase (kb) deletion, was rescued from theJLS-V11-NP cell line as an N-tropic, large XC plaquerecombinant, presumably by recombination within gag withthe endogenous BALB/c XC-negative, N-tropic retrovirus(5). MoMuLV mutants deleted within either the 3' end of theLTR (12) or the reverse transcriptase-coding region of pol(45) have been isolated as revertants from NIH 3T3 cellsbearing new sequences that derive from recombination withone of several endogenous retroviral loci. The frequency ofbiologically significant recombination between infectiousvirus and endogenous loci may depend on RNA templateexchange during reverse transcription (41) rather than onDNA-mediated recombination involving accessible endoge-nous loci in host chromatin. Since working models of leuke-mogenic conversion place env and U3 LTR recombinationalevents in the spleen and thymus (15, 23, 28), it is possiblethat endogenous loci involved in the recombination processexhibit tissue-specific expression.
4136
JOURNAL OF VIROLOGY, OCt. 1989, p. 4136-41470022-538X/89/104136-12$02.00/0Copyright C) 1989, American Society for Microbiology
We are interested in the regulation of transcription ofthese endogenous retrovirus sequences and have used the129 Gix+ and Gi,J strains to study two control loci, Gv-J andGv-2, that affect endogenous retrovirus expression (50-52).Alternative alleles at these loci behave in a semidominantand dominant fashion, respectively, to regulate endogenousretrovirus transcript (30, 31) and protein (29) levels. Thecongenic 129 Gix- strain is homozygous for the alternative ballele at the Gv-1 locus and shows 10- to 20-fold lower levelsof endogenous retrovirus expression than the Gix+ strain. Tostudy the sequences controlled by Gv-J, we characterizedcDNA clones representing transcripts from active endoge-nous retroviral loci in the 129 Gix+ mouse (32). Although themurine strain 129 genome lacks replication-competentecotropic proviral loci (21), some of the cDNA isolatesrepresent intact polytropic and modified polytropic envgenes that could serve in other mouse strains as donors inthe ecotropic-to-polytropic conversion process. OthercDNA isolates represent loci deleted for >1 kb of envsequence but are regulated coordinately with those contain-ing full-length env genes. These env deletion breakpoints canthus be used as convenient markers for specific loci that areactively transcribed and regulated by the Gv-1 locus, asopposed to the large number of loci that conserve envsequence but are not necessarily transcriptionally active.
In the present report, oligonucleotide probes directed totwo env deletion breakpoints were used to isolate provirus-like structural loci from the 129 Gix+ genome. Analysis ofthese loci by in vitro transcription and translation and bynucleotide sequencing reveals significant deletions of gagand pol domains that result in truncated gag protein prod-ucts. Three of the isolates contain nearly identical endoge-nous retrovirus sequence, inserted at different loci in thehost genome. In conjunction with the loss of the normal envgene splice acceptor sequence, the gag region encodes theonly possible protein products for this family of env-deletedloci. These loci should prove useful in identifying features ofhost chromatin and endogenous retrovirus structure thatresult in Gv-J-regulated transcription.
MATERIALS AND METHODS
Oligonucleotide synthesis and labeling. Two nonadecameroligonucleotides corresponding to the antisense strand ofpol-env sequence breakpoints identified (32) in cDNA iso-lates S13 (5-GAATACAGGGTCCCTTTCCA; del env-J) andE2 (5'-TACAAACTGGACCTTTCAT; del env-2) were syn-thesized by the phosphoramide method on an Applied Bio-systems 380-A DNA synthesizer. (The nucleotide sequencedata reported in this paper for del env-J and del env-2 willappear in the EMBL, GenBank, and DDBJ NucleotideSequence Databases under accession numbers M26005-DL1and M26006-DL2, respectively.) Crude oligomers were elec-trophoresed on a 20% acrylamide gel in 44.5 mM Tris-borate-1 mM EDTA, pH 8.3 (0.5x TBE), eluted in 4 ml of0.5 M ammonium acetate-10 mM magnesium acetate, con-centrated with n-butanol to 100 ,ul and precipitated with 1 mlof absolute ethanol. Stocks (5 pmol) were used for end-labeling with [_y-32P]ATP (7,000 Ci/mmol; Amersham Corp.)by using T4 polynucleotide kinase (New England BioLabs)as described before (35). Labeled oligonucleotides were useddirectly for library screening and colony hybridization orisolated on 20% acrylamide-8.3 M urea gels for Northern(RNA blot) and Southern hybridization.
Mice, nucleic acid isolation, and Southern and Northernblot hybridization. All mice were obtained at 6 to 8 weeks of
age from the mouse breeding colony at Scripps Clinic. DNAwas isolated from thymus and liver as described before (7).Genomic DNA was digested with a twofold excess ofrestriction enzymes under conditions recommended by themanufacturer and analyzed by electrophoresis on 0.7%agarose-2x TAE gels at 1 to 1.5 V/cm, 10 ,g of DNA perlane (1 x TAE is 40 mM Tris acetate [pH 8.1], 2 mM EDTA).Recombinant DNA digests were separated on 0.9% agarose-0.5x TBE gels. Transfer of DNA to nitrocellulose wasperformed as described before (47). RNA extraction fromtissues, polyadenylated RNA isolation by oligo(dT)-cellu-lose chromatography, electrophoresis in 0.9% agarose-2.2M formaldehyde gels, and transfer to nitrocellulose wereperformed as described before (30). Prehybridization andhybridization with oligonucleotide probes were performed at370C with 6x TESS-50 mM sodium phosphate-Sx Denhardtsolution (13)-100 p.g of denatured salmon sperm DNA perml-2 mM sodium PP,-0.1% sodium dodecyl sulfate (SDS)(20x TESS is 3 M NaCl, 100 mM TES [N-tris(hydroxymeth-yl)methyl-2-aminoethanesulfonic acid, pH 7.4], 100 mMEDTA). Labeled oligonucleotides were added to a finalconcentration of 1.8 pmol/ml at a specific activity of 0.5 x107 to 1 x 107 cpm/pmol. After >24 h of hybridization,washes were performed with 6x SSC (1x SSC is 0.15 MNaCl, 0.015 M sodium citrate) or with 3 M tetramethylammonium chloride (59) at temperatures defined in individ-ual experiments below. Prehybridization and hybridizationof nick-translated (43) MoMLV pol probe KpnI fragment,(nucleotides [nt] 2858 to 5577) (46) to new or regenerated (55)blots was performed at 420C in 50% formamide-5 x PIPESS,Sx Denhardt solution-100 ,ul of denatured salmon spermDNA-2 mM sodium PP1-0.5% SDS {20x PIPESS is 3 MNaCl, 100 mM PIPES [piperazine-N,N'-bis(2-ethanesulfonicacid), pH 8.1], 100 mM EDTA}. Washes were performedafter >36 h of hybridization.
Library construction and screening and phage DNA isola-tion. High-molecular weight DNA from 129 Gix+ mouse liverwas digested to completion with EcoRI and sedimented on alinear 10 to 40% sucrose density gradient (34), and DNA of7 to 20 kb in length was pooled after fractions were sized byelectrophoresis on 0.9% agarose gels. EcoRI-BamHI-di-gested EMBL4 DNA (Promega Biotec) was prepared asrecommended by the supplier and ligated to the sized mousegenomic DNA EcoRI fragments. Ligated DNA was addeddirectly to bacteriophage lambda DNA packaging mix (Gi-gapack; Stratagene), phage were plated on Escherichia coliNM539, and plaques were transferred to nitrocellulose filtersfor probe screening (58). Oligonucleotide probes del env-]and del env-2 were hybridized to library filters under theconditions described above, except that the oligomer con-centration was 0.72 pmol/ml. Plaques which cross-hybrid-ized to an endogenous cDNA pol-env nick-translated probe(PstI-EcoRI fragment of El) (32), under conditions de-scribed for Southern and Northern blot analysis, wereplaque purified with the del env probes. DNA was extractedfrom phage isolated by DE52 chromatography of brothlysates (19). EcoRI inserts were subcloned in pUC19 forrestriction enzyme mapping and hybridization after prelimi-nary characterization in the EMBL4 vector. Subclones wereprobed with fragments of the del env-2 cDNA E2 (32)representing 3' pol-plSE (HindIII-PstI) and U3 LTR (PstI-KpnI) sequences.
In vitro transcription and translation and analysis of proteinproducts. To generate capped RNA transcripts, KpnI frag-ments extending from the 5' LTR R region to the 3' LTR andencompassing the entire endogenous retroviral genomic loci
3.1, 3.2, 11.1, and 15.3 were subcloned into the pSP18expression vector (Bethesda Research Laboratories). Simi-larly, KpnI fragments of MCF 247 and MX27 (generouslyprovided by Christine Holland and Jonathan Stoye, respec-tively) extending from the conserved KpnI site in the 5' LTRto the KpnI site within the 5' region of pol were alsosubcloned in pSP18. The plasmids were linearized witheither HindIll or EcoRI distal to the SP6 RNA polymeraseinitiation site, depending on the orientation of the insert inthe plasmid vector. Transcription reactions were performedwith SP6 as described before (16) in the presence of 500 liM7mGpppGm (Pharmacia), 100 mM GTP, CTP, ATP, andUTP, and S FLCi of [cx-32P]GTP (3,000 Ci/mmol; New En-gland Nuclear Corp.). The transcripts were extracted withphenol-chloroform, passed over G-50 Sephadex, and ethanolprecipitated before use in a translation assay consisting of 18[L1 of rabbit reticulocyte lysate (Amersham) and 20 ,uCi of[35S]methionine (1,127 Ci/mmol; New England Nuclear)programmed by 200 ng of RNA transcript. The translationreactions were stopped by adding equal volumes of 0.1 MNaCl, 20 mM Tris (pH 7.0) and 1% Triton X-100, and 2.5mM phenylmethylsulfonyl fluoride. To assay the translationproduct directly, 3 ,ul of the reaction mixture containing stopsolution was diluted with 45 [LI of sample buffer (2.5% SDS,5% P-mercaptoethanol, 20% glycerol, 0.125 M Tris [pH6.8]), boiled for 10 min, and applied to 10% SDS-polyacryl-amide gels (27). For immunoprecipitation of the in vitro-synthesized protein products, 3 to 5 ,ul of goat antiserumraised against Rauscher MuLV p15, p12, and p30 or normalgoat serum (provided by John Elder) per sample was preab-sorbed to protein A-Sepharose (Pharmacia) in PORT buffer(150 mM NaCl, 10 mM Tris [pH 8.5], 100 pg of ovalbuminper ml, 0.1% Nonidet P-40, and 0.5% Tween-20) for 2 h at4°C and washed in PORT buffer before incubation with thereaction mixture at 4°C overnight. The protein A-Sepharose-antibody complex was washed with PT buffer (150 mMNaCl, 10 mM Tris [pH 8.5]) before elution by boiling insample buffer. The translation products after electrophoresiswere visualized by impregnating the gel with En3Hance(New England Nuclear) and autoradiography.
Nucleotide sequence analysis. The dideoxy nucleotidechain termination method (44) with [35S]deoxyadenosine5'-(oa-thio)-triphosphate used for detection was applied topUC19 subclones of the EMBL4 inserts prepared by alkalinedenaturation (3) or by further subcloning into single-strandedbacteriophage M13mplO and mpll (35); standard M13 prim-ers and custom primers defined by internal sequences,synthesized as described above for the env deletion probes,were used to initiate reactions on the templates. Reactionproducts were resolved on 6% acrylamide buffer gradientgels (6).
RESULTS
env-deleted endogenous retroviral transcripts and proviralloci. We have previously described the structure and se-quences of cDNAs encoding the env and U3 LTR regions ofendogenous retroviruses as isolated from the spleen, liver,and epididymus of 129 Gi,+ inbred mice (31, 32). A numberof independent cDNA isolates displayed extensive deletionsof env that we have categorized as either del env-J or delenv-2 sequences. The S13 cDNA contains the del env-Ideletion, extending from nt 5676 at the 3' end of pol to nt7673 in the env p15E-coding region, while the E2 cDNArepresents the del env-2 deletion, extending from nt 5796 atthe 3' end of pol to nt 7696 in the env pl5E sequence
A [3__
FIG. 1. Oligonucleotide probes to endogenous retrovirus envdeletions del env-J and del env-2 identify specific transcripts fromtissues of 129 Gix+ mice. Polyadenylated RNA from spleen (S), liver(L), thymus (T) and epididymus (E) in duplicate lanes of a 0.9%agarose-2.2 M formaldehyde gel (see text), transferred to nitrocel-lulose and hybridized to 32P-labeled DNA probes. AX, Hindlll-digested bacteriophage lambda c1857 molecular weight markers. (A)Autoradiograph of first set of lanes hybridized to end-labeled 19-meroligonucleotide del env-i, washed at 56°C in 6x SSC, and exposedfor 56 h on preflashed XRP film (Eastman Kodak). (B) Autoradio-graph of second set of lanes hybridized to end-labeled 19-meroligonucleotide del env-2, washed at 42°C in 6x SSC, and exposedfor 24 h as in panel A. (C) Autoradiograph of second set of lanesstripped of initial probe (see text), prehybridized to nick-translatedMoMuLV pol region probe, washed at 42°C in 0.2x SSC, andexposed for 5 days on preflashed XAR film (Eastman Kodak).
(corresponding to AKV virus nucleotide sequence nomen-clature [20]). Each of these deletions was apparently medi-ated by a mechanism that involves recognition of directheptamer nucleotide repeats present in full-length endoge-nous retrovirus sequences at the deletion breakpoints. Wedesigned oligonucleotide probes to the novel sequence jux-taposition created by these deletionc, each with a length (19nt) that would not permit competitive hybridization to eitherend of the sites in nondeleted retroviral sequences. In thisway the oligonucleotide probes discriminate between locibearing these deletions and the numerous endogenous retro-virus loci present in the 129 Gix mouse genome. Theselected loci would thus include those particular endogenousretrovirus sequences, responsive to Gv-J control, fromwhich transcripts arose. We first tested the specificity of theoligonucleotides on endogenous retrovirus transcripts from129 Gix+ tissues and compared the hybridizing species withthe RNAs identified with an MoMuLV pol region probe (Fig.1). The del env-J probe identified a 5.0-kb transcript inthymus and a 4.0-kb transcript in liver (Fig. 1A), while delenv-2 identified a 5.2-kb spleen transcript (Fig. 1B). Thesetranscripts were described previously, by size and tissue oforigin, with the use of a full-length MoMuLV clone (30) anda 129 Gix+ endogenous retroviralpol-env cDNA clone (31) asprobes; levels of these transcripts were shown to be reduced10- to 20-fold in 129 Gij, mouse tissues, by Northern blothybridization analysis and by Si nuclease protection assay.In addition to these major del env-l and del env-2-hybrid-izing species, transcripts of similar size were present inlesser abundance in each tissue tested. In contrast, rehybrid-ization of the del env-2 blot to the MoMuLV pol region probeshowed that retrovirus-related RNA species were present insizes from 8.0 to 3.5 kb in these three tissues (Fig. 1C). The
FIG. 2. Genomic DNA blot analysis with oligonucleotide probesspecific to env-deleted endogenous retrovirus loci. EcoRl-digestedhigh-molecular-weight DNA separated on 0.7% agarose gels was
transferred to nitrocellulose and hybridized to 32P-labeled DNAprobes. (A) Autoradiograph of C57BL/6 (C57, lane 1) and 129 Gix+(129, lane 2) thymus DNA hybridized to nick-translated MoMuLV(Mo) pol fragment, washed at 42°C in 0.1x SSC-0.1% SDS, andexposed for 20 h on preflashed AR film (Eastman Kodak). 4A,
HindIll-digested bacteriophage lambda c1857 molecular weightmarkers. (B) Autoradiograph of 129 Gix+ (+, lanes 4 and 5) and 129Gix- (-, lane 6) liver DNA hybridized to end-labeled 19-meroligonucleotide del env-1 washed at 54°C in 3 M tetramethylammo-nium chloride-50 mM Tris (pH 8.0)-2 mM EDTA-0.1% SDS andexposed for 6 days on preflashed XRP film. (C) Autoradiograph of129 Gix, (+, lane 7) and 129 Gix- (-, lane 8) thymus and 129 Gix-(-, lane 9) liver DNA hybridized to end-labeled oligonucleotideprobe del env-2, washed, and exposed as in panel B. Arrowsindicate major hybridizing EcoRI fragments present in genomicDNAs.
del env-2 probe also detected a 1.7-kb RNA in spleen that didnot cross-hybridize to the MoMuLV pol probe (Fig. 1B),suggesting that it represents a spliced transcript that containsprimarily the plSE-coding portions of the deleted env gene.
Transcripts of this size and sequence specificity have beenreported previously (7, 23). There was a >1-kb discrepancybetween the size of the major del env-] and del env-2transcripts (5.2 to 5.0 kb) and the expected size (6.5 to 6.4kb) of an unspliced endogenous retrovirus RNA containing adeletion of 1.9 to 2.0 kb. This anomaly is examined further inthe genomic DNA analysis presented below.
Since the two oligonucleotides detected discrete subsetsof endogenous retrovirus transcripts, we used them to probe129 Gi1, and 129 Gix- genomic DNAs to determine thenumber of loci bearing these env region deletions. EcoRI-digested liver DNA from several 129 Gix+ and 129 Gix- micewas hybridized on Southern blots to the del env-l (Fig. 2B)and del env-2 (Fig. 2C) probes. In contrast to the selectivityof the oligonucleotide probes, hybridization of 129 Gix+ andC57BL/6 EcoRI-digested DNA with the MoMuLV pol probeidentified more than 20 different fragment sizes, many ofwhich probably represent more than one genomic locus (Fig.2A). Full-length endogenous retrovirus genomes normallycontain only one EcoRI site in the env region (9, 22, 25).Proviral inserts deleted for env will therefore migrate intacton single unique EcoRI fragments defined by restriction sitesin flanking host DNA. The oligonucleotide hybridizations
showed that del env-J transcripts could come from any offour different loci, on 13.5-, 10.0-, 9.0-, and 8.8-kb EcoRIfragments (Fig. 2B); a fifth hybridizing fragment, at 8 kb, wasnot retained at slightly higher wash stringency (data notshown). However, del env-2 transcripts appeared to haveonly one source, present on an 8.9-kb EcoRI fragment (Fig.2C). Identical hybridization patterns by the 129 Gix+ andcongenic 129 Gi,- DNAs (Fig. 2B and C) demonstrated thatspecific proviral loci responsible for Gv-1-regulated tran-scripts were retained in the 129 Gi,- congenic strain, whichcarries the Gv-lb allele of the nonpermissive C57BL/6 strain(51).
env-deleted loci contain a common gag-pol deletion. Toisolate the loci identified by the del env oligonucleotides, the7- to 20-kb size-fractionated EcoRI fragments of 129 Gix+DNA were ligated into the EMBL4 bacteriophage vector,and 4.2 x 105 plaques of the resulting library were screenedwith pooled del env-l and del env-2 oligonucleotide probes.Phage DNAs cross-hybridizing with the 129 Gix+ endoge-nous pol-env region cDNA clone were mapped with restric-tion enzymes and analyzed on Southern blots for hybridiza-tion with separate del env-] and del env-2 oligonucleotidesand with U3 LTR and pol-pl5E probes from the del env-2cDNA clone E2 (see Materials and Methods). Three delenv-J loci, 3.1, 3.2, and 11.1, were present on the 13.5-,10.0-, and 9.0-kb fragments, respectively, identified in ge-nomic Southern analysis (Fig. 3A). The 3.1 locus wasrepresented by five independent isolates, 3.2 by one isolate,and 11.1 by three isolates. None of the del env-] clonesrecovered represented the 8.8-kb genomic EcoRI fragmentidentified by the oligonucleotide probe (Fig. 2B), although afragment of this size would be compatible with the sizerequirement of the EMBL4 phage vector. The single delenv-2 locus, 15.3, was represented by three distinct isolatesthat each gave the same 8.9-kb EcoRI insert and internalrestriction map (Fig. 3A). The three del env-J loci wereessentially identical by restriction enzyme analysis withinthe proviral domains, yet flanking host DNA showed nosimilarities by the same criteria. By comparison with pub-lished endogenous type C murine retrovirus maps (25, 53)(Fig. 3B), the del env-J and del env-2 loci lacked approxi-mately 1.4 kb of sequence in addition to the env deletions.The distance between conserved U3 PstI sites was 4.8 and5.0 kb for the del env-] and del env-2 clones, respectively,but sizes of 6.3 and 6.4 kb are to be expected if theproviruses lack only the env sequences. Although the tran-script size from these loci appeared to be approximately 5 kbfor transcripts arising in spleen and thymus, a 4-kb del env-]RNA was found in liver (Fig. 1). This may be transcribedfrom the uncloned del env-] locus on the 8.8-kb EcoRIgenomic DNA fragment or from any one of the del env-1 lociexpressed preferentially in the liver and containing noveltranscript-processing signals. To resolve the discrepancybetween expected and actual proviral insert size, EcoRIinserts bearing the three del env-J loci and one del env-2locus were subcloned in pUC19 for further mapping andsequence analysis. Deletion of a gag-po junction region wassuggested by the conservation of 5' gag XbaI and Sacl sitesand a 3' pol XbaI site in subclones of all four isolates (Fig. 3)and by loss of Sacl, KpnI, and BamHI sites expected in the5' end of the full-length endogenous pol region (25, 53).To test the gag-pol junction region for the presence of
deletions, del env-I clone 3.1 and del env-2 clone 15.3 weresequenced from a conserved Sall site in the pol domain. Thenucleotide sequence revealed identical deletions of gag-polsequence between nt 2188 and 3675, relative to AKV virus
FIG. 3. (A) Restriction enzyme maps of cloned 129 Gix+ mouse genomic EcoRI fragments containing endogenous retrovirus inserts withdel env-J or del env-2 deletions of env. The three EcoRI fragments hybridizing to the del env-J oligonucleotide probe, 3.1, 3.2, and 11.1, arebracketed on the left; the single EcoRI fragment hybridizing to the del env-2 oligonucleotide probe, 15.3, is shown below the del env-] clones.(B) Polytropic consensus restriction enzyme map derived from references 25 and 53. Boundaries of gag, pol, and env genes are defined withrespect to restriction enzyme sites in an ideal intact provirus. LTR regions are indicated in boxes. All maps are oriented with 5' LTRs at leftand aligned relative to the positions of these LTRs in the cloned EcoRI fragments. Vertical lines indicate positions of restriction enzyme sites.Maps show positions of all sites for enzymes designated by single-letter code: E, EcoRI; H, Hindlll; X, XbaI; P, PstI; K, KpnI; S, Sacl; andB, BamHI. Positions and sizes of del pol, del env-I, and del env-2 deletions are shown below each restriction map. *, Sall site conserved inpol region of each clone, used to create subclones for nucleotide sequence determination of del pol deletion breakpoint of each clone (seetext). Host DNA flanking retroviral inserts was not mapped for Sall sites. (H), HindlIl site on polytropic consensus map is present only inmodified polytropic class endogenous retrovirus.
genome coordinates (20), in both endogenous isolates (datanot shown). This 1,474-nt deletion, named del pol, extendedfrom the 3' end of gag plO into the reverse transcriptase-coding region of pol (Fig. 3). As with the del env-] and delenv-2 deletions (32) and retroviral and cellular deletionsobserved by others (14; summarized in reference 28), the delpol junction retained one copy of a 7-bp direct repeat foundat the deletion breakpoints in the full-length AKV virussequence.Other distinguishing features of the 15.3 restriction map in
comparison to those of 3.1, 3.2, and 11.1 were the 3' polHindlIl site and the 5' env BamHI site in the del env-2 clone15.3 and the relative size of the U3 LTR regions. Thepresence of the 15.3 HindIll site indicated that this insertwas derived from a full-length provirus of the modifiedpolytropic class of endogenous retrovirus, while the absenceof the pol HindIll sites from the del env-J clones 3.1, 3.2,and 11.1 distinguished them as polytropic-related sequences(53, 54). The 15.3 BamHI site within env is conserved inendogenous, nonxenotropic retroviruses as a group and was
presumably also present in the nondeleted precursors to thedel env-J loci. Double digestion of the genomic isolates withPstI and KpnI showed that the U3 LTR regions of del env-Jclones were -50 nt shorter than the del env-2 U3 regions(550- versus 600-nt PstI-KpnI fragments; data not shown).The U3 size differences are presumably due to deletions inthe single-copy enhancer homolog in del env-J clones rela-tive to the del env-2 clone, similar to those reported previ-ously in cDNA sequence analysis (31, 33). Therefore, theseisolates share identical gag-pol deletions yet are significantlydivergent in LTR and env sequences, suggesting indepen-dent but equivalent deletion events in the two, or an ongoinggenetic exchange as defective proviruses.
Truncation of gag ORF yields products of 32 and 34 kDa.One consequence of the 1,474-nt gag-pol deletion in the delenv-J and del env-2 clones is the truncation of the gag openreading frame (ORF) and possibly the production of a novelgag-pol fusion product. The sequence juxtaposition arisingfrom the deletion predicts an ORF extension of 10 codonsbeyond the breakpoint before a new termination site corre-
FIG. 4. del env-I and del env-2 gag regions encode truncated polypeptide products. SP6 RNA polymerase-directed transcripts offull-length or partial retrovirus inserts in the transcription vector pSP18 pGEM were capped with 7mGppp and translated in a rabbitreticulocyte lysate in the presence of [35S]methionine (see text). (A) Autoradiogram of portions of the total translation products from mixescontaining no added RNA (-), del env-J clone 3.1, 3.2, and 11.1 transcripts, or del env-2 clone 15.3 transcripts separated on a 12%polyacrylamide gel and detected as described in the text. (B) Autoradiogram of radiolabeled translation products from mixes containing MCF247, del env-] clone 3.1, or del env-2 clone 15.3 transcripts unreacted with antiserum (-) or reacted with antisera to Rauscher virus gagproteins p15 (a-p15), p12 (a-pl2), or p30 (a-p3O), recovered by protein A-Sepharose adsorption and analyzed as described for panel A. (C)Autoradiogram of portions of total translation products from mixes containing transcripts of del env-J clone 3.1, endogenous polytropic cloneMX27, or MCF 247 analyzed as in panel A. Protein molecular mass standards are as indicated (in kilodaltons).
sponding to nt 3701 to 3703 of the AKV virus sequence. Totest these defective loci for their gag region coding potential,KpnI fragments of 4.8 (del env-J) and 4.9 (del env-2) kb,extending from the 5' U5 LTR to the equivalent restrictionsite in the 3' LTR, were subcloned in a vector which allowsSP6 RNA polymerase transcription of inserted sequences.Transcripts from each reaction were capped with 7mGppp invitro and translated in a rabbit reticulocyte lysate in thepresence of radiolabeled methionine. Transcripts from 3.1,3.2, and 11.1 directed translation of a 34-kDa protein, andtranscripts from 15.3 directed translation of a 32-kDa protein(Fig. 4A). If these proteins originated at the equivalent ATGutilized by AKV virus, their apparent molecular mass onSDS-polyacrylamide gel electrophoresis suggests productsthat extend halfway through the p30 region of gag, sincethey are half the size expected if terminated in pol asdescribed above. The translation products from 3.1 and 15.3were immunoprecipitated with individual p15, p12, and p30Rauscher gag protein antisera to determine whether the 34-and 32-kDa products indeed represented the expected cod-ing region (Fig. 4B). These antisera showed that the 34- and32-kDa products contained p15 and p12 determinants (32-kDa protein reactivity with p12 antiserum was observedafter a longer exposure of the autoradiogram shown in Fig.4B). In contrast, only the del env-] locus 3.1 productconserved minor antigenic reactivity with the Rauscher p30antiserum. An MCF 247 gag p65 translation product dem-onstrated uniform recovery with each antiserum in parallel.Thus, the del env-J and del env-2 transcripts encode similar,but not identical, truncated gag protein products that areless than 280 residues long, whereas the gag-pol deletionthey share predicts products -520 amino acids in length.Virus-negative NIH Swiss mice express p12 and p30 liverantigens that cross-react with anti-AKR MuLV and anti-Rauscher MuLV sera but are distinct from the homologousviral antigens (48, 49). BALB/c and 129/J tissues contain p30
but not p12 reactivity to anti-AKR virus sera (36). Based onthe abundance of transcripts from the del env-J and del env-2loci in 129 Gix+ mouse tissues (Fig. 1) (30) and the ability ofSP6-derived transcripts of the cloned loci to direct transla-tion of products reactive with Rauscher gag protein antisera,we suspect that the del env-J and del env-2 loci are respon-sible for a portion of the gag antigen observed in 129 Gi,+tissues.We were interested in the possibility that even full-length
endogenous proviruses might contain similar cryptic gagregion termination signals. To test this, we compared the 3.1and MCF 247 cell-free products with the product fromMX27, a polytropic provirus clone from HRS/J mice that isapparently intact, by restriction enzyme mapping (53). Thiscomparison (Fig. 4C) showed that MX27 encoded a polypep-tide of 63 kDa, only slightly smaller than the product fromthe infectious MCF 247 clone. Although this suggests thatthe MX27 gag region does contain a premature terminationsignal or an in-frame deletion relative to MCF 247, it alsodemonstrates that gag products of 32 to 34 kDa are not acommon feature of all endogenous polytropic and modifiedpolytropic proviruses.
Nucleotide sequence confirms structure of gag polypeptides.To identify the ORFs resulting in the observed gag proteinproducts from the del env-J and del env-2 clones, wedetermined 5' gag nucleotide sequences for 3.1 and 15.3.Since the 5' leader (L) region of gag has been reported foronly one other endogenous murine polytropic provirus,AL10 (40), the L regions were included in the 3.1 and 15.3analyses. Figure 5 shows these data, beginning with the Rregion of the 5' LTR and extending 1,435 nt through theapparent ORFs of the del env-J and del env-2 gag regions.The previously reported partial AL10 sequence is includedfor comparison of L regions (see below). Analysis of the gagprotein-coding sequence showed that, relative to the AKVecotropic virus genomic sequence (20), 3.1 and 15.3 shared
AKV .G.....T.... A. .T......T.C. .... AT.....T.......... T.. A.... CT..0......G O.GGACTG....CTC. A.G.... TOO .. .C.GArg Ser Ser Asn Ser Pro Oly GIyLeu SerAspGlyAspGly Arg
1110 1140 1170 1200
3. 1 AGAAGGCACCTCAGTCCCCTTCAGTTTGCGGGAGAAACTCGACGCCACTTCAGATCACTCCGCAT15.3.........................................C.C ......... .~.....AKV. CA .T....T. .T. A. ..TO. .L-euLeuProTrpCysL.uAspCysG lyG IuGlO uThrLeuPoG InArgThrProProProProArgH isSerH isSerA IsAKV.......A.. ... ..T.. A.C.G ... ......A . C.C...AA...CC.0....A...T..G....C... C.....TTO.... ..T ... TTTT
ThrSer IleProAla Ile Lys Thr Lou1230 1280 1290 1320
19 bp deletionGlyGlyAspGlyGlnLeuGlnTyrTrpProPheS.rSerS.rAspLu-----------ThrLeuProPheLeuLysll.GInVaIAsns..
a Comparisons of del env-l clone 3.1 and del env-2 cloned 15.3 sequenceswith those for comparable stretches of the AKV ecotropic virus genome;nucleotide numbering is as shown in Fig. 5.
b R, Redundant region; U5, unique 3-flanking region in 5' LTR; L,untranslated leader region of gag; p15, p12, and p30, protein-coding regions ofgag.
I Divergence (Div) calculated as number of gaps introduced to createalignment plus number of nonidentical nucleotides divided by total nucleotidelength, expressed as percent. Conservation (Cons) calculated as number ofidentical nucleotides divided by total nucleotide length, expressed as percent.
identical 19-nt deletions in the 5' end of the p30-codingregion, between nt 1369 and nt 1389 of the aligned sequencesin Fig. 5 and corresponding to nucleotides 1341 through 1359ofAKV virus. This deletion occurred 20 codons into the p30coding region, and the frameshift resulted in a 10-codonextension of the 3.1 reading frame before termination. The15.3 clone contained a single-base-pair deletion upstreamwithin the p12-coding sequence at nt 1236 (Fig. 5); theresulting frameshift encoded 44 amino acids in an alternateORF that terminated at the 19-nt deletion within p30. Be-sides single-nucleotide substitutions relative to the AKVvirus sequence, there were also a 21-nt deletion between nt798 and nt 820 in the 15.3 sequence and a 3-nt deletion at nt1179 to 1183 in both the 3.1 and 15.3 sequences. These hada neutral effect on the gag ORF but served to reduce therelative size of the 15.3 product.These sequence analyses explain both the mass differ-
ences of the 3.1 and 15.3 gag products (Fig. 4A) and theRauschcer anti-gag protein antiserum reactivities of theseproducts (Fig. 4B). The 3.1 and 15.3 products should haverespective masses of approximately 29.8 and 27.2 kDa, bythe derived amino acid sequence shown in Fig. 5. Asrepresented by clone 3.1, the del env-J family conserved gagprotein-coding capacity relative to 15.3 by virtue of thesingle-nucleotide deletion of the del env-2 clone at nt 1236 inthe p12-coding region. Table 1 summarizes the 3.1 and 15.3polytropic and AKV virus ecotropic sequence comparisonsfor gag polypeptide-coding regions and for upstream un-translated regions shown in Fig. 5. On the nucleotide level,15.3 did not show greater gag region divergence from AKVvirus than did 3.1; both polytropic clones conserved greater
del env-1 loci 3.1 5 -TGGAGGCCCCAGCGAGAT- 3'
3.2 -TGGAGGT T CC ACCGA GAT -
11.1 - TGGAGGT TCCACCGA GAT -
del en-2 locus 15.3 - TGGAGGTC CCACCGA GAT -
tRNAgln 3'HoACCUCCAAGGUGGCUCUA- 5FIG. 6. PBSs of del env-J and del env-2 proviral loci do not
conserve tRNA primer base-pairing compatability. Sequences fromloci 3.1 and 15.3 are identical to those shown at nt 147 to 164 in Fig.5. Sequences from loci 3.2 and 11.1 were determined in parallelanalyses. The tRNAGln sequence is from reference 38 and repre-sents the 18 nt at the 3' end of a rat liver glutamine tRNA. Asterisks(*) underneath retroviral nucleotides indicate a base mismatch withthe complementary tRNA sequence at the bottom.
than 80% of the p15-coding sequence relative to AKV virusand equivalent loss of conservation (77 and 73%, respec-tively) through the p12- and p30-coding sequences. In thesame regions, 3.1 and 15.3 maintained greater than 90%sequence identity to each other. This relatedness is consis-tent with other data on ecotropic-polytropic divergence (54)and emphasizes the significance of the deletion at nt 1236 tocoding potential in the del env-2 clone 15.3.PBS and 3' leader region are not conserved. The R and U5
regions of the 3.1 and 15.3 5' LTRs showed relatively strong(-90%) conservation of sequence compared with AKV virus(Table 1), whereas the untranslated leader (L) sequenceswithin gag showed greater than 20% divergence comparedwith AKV virus. This divergence was seen first in thepresumed tRNA primer-binding site (PBS) for initiation ofminus-strand reverse transcription, nt 147 to 164 in Fig. 5, inwhich 3.1 showed 6 bp changes and 15.3 showed 5 bpchanges relative to AKV virus. In murine type C retrovi-ruses in general, the PBS conserves complementarity toproline tRNA (12). However, the PBS sequence of thepresent 15.3 clone, of a replication-competent revertant ofan MoMuLV deletion mutant recovered from NIH 3T3 cells(12), and of the AL10 endogenous polytropic clone reportedpreviously (40) and shown in Fig. 5 all contained a 17 of 18nt match to the 3' terminus of a rat glutamine tRNA. Clone3.1 diverged from complementarity to this tRNA primer at 3of 18 nt, in addition to its divergence from AKV virussequences, suggesting that it cannot bind to either primer.We examined the PBS sequences of del env-] clones 3.2 and11.1 to determine whether these clones share this divergentsequence with 3.1. The results of this comparison are shownin Fig. 6. The 3.2 and 11.1 isolates showed perfect comple-mentarity to the glutamine tRNA and were identical to thePBS of an endogenous retrovirus isolate from an RFM/Unstrain mouse (38). Divergence of the 3.1 PBS may indicatethat this locus is no longer capable of helper virus-dependentreplication, whereas the two other del env-J loci and the delenv-2 locus may maintain this capacity. Since we suspectedthat these PBS modifications in del env-J clone 3.1 were
FIG. 5. Nucleotide sequence analysis of 5' LTR R and U5 regions and apparent gag ORFs of del env-l clone 3.1 and del env-2 clone 15.3.The positive strand sequence for clone 3.1 is shown on the top line. Nucleotide sequence of equivalent regions for clone 15.3, AKV virus (19),and endogenous retrovirus clone AL10 (40) are indicated below the 3.1 sequence. The sequence of AL10 as reported is incomplete for theregion of comparison. Identical nucleotides relative to 3.1 are indicated by dots; nucleotide differences are indicated by the substituted base,and insertions or deletions are indicated by dashes. The R and U5 domains of the 5' LTR and the PBS and splice donor sequences of the gagL region are indicated above the 3.1 sequence by brackets. The apparent 3.1 gag polyprotein ORF is indicated above the 3.1 sequence bythe three-letter amino acid code; p15, p12, and p30 segments of the polyprotein are indicated above the ORF by brackets. Differences fromthis amino acid sequence encoded by 15.3, AL10, and AKV virus are indicated directly below their respective nucleotide sequences by theappropriate three-letter amino acid code. Termination codons in the 3.1 and 15.3 ORFs are indicated by three asterisks (***).
FIG. 7. Host proviral junctions demonstrate 4- to 5-nt duplica-tion of host DNA sequence. Nucleotide sequences were determinedby using primers directed to conserved sequences in the LTRs ofcloned endogenous retroviral cDNAs (32). Subclones of del env-J
loci 3.1, 3.2, and 11.1 and del env-2 locus 15.3 containing separate 5'and 3' LTRs were used as templates. Duplicated sequence inflanking mouse genomic DNA is underlined. Inverted terminalrepeat sequences at the 5' and 3' ends of proviruses are separated bydashes.
errors by helper virus reverse transcriptase, we examinedthe host-proviral junction sequences for each of the clonedloci to determine whether integration at the 3.1 locus waspossibly defective due to helper function. The sequences(Fig. 7) indicated that each insertion event resulted induplication of 4 to 5 bp of host sequence, as expected fromnormal integrase activity (56).Three additional functions encoded by the L region are the
splice donor sequence required for env mRNA processingand the dimerization and encapsidation sequences requiredfor genome incorporation into the virion (11). The splicedonor sequence identified in AKV virus (20), at nt 205 to 211(Fig. 5), was fully conserved in the 3.1 and 15.3 clones. Datafrom 129 Gix+ strain cDNA clones indicate that the spliceacceptor upstream of env is part of the 1,997-base-pair (bp)pol-env sequence deleted in del env-J isolates, while thisacceptor is intact and fully conserved upstream of the1,990-bp deletion in del env-2 clones (32). These transcript-processing signals may be recognized in vivo, as suggestedby the 1.7-kb del env-2-specific transcript present in spleensof 129 Gix+ mice (Fig. 2B).The splice donor sequence was contained in a part of the
L region that extends from the 3' end of the PBS at nt 165 tont 420 (Fig. 5), which is generally conserved (-87%, Table 1)between the endogenous isolates and AKV virus. Dimeriza-tion of full-length retroviral transcripts involves joining atthe 5' ends of each molecule, which may involve this 5'portion of L, including the PBS (11). Efficient encapsidationalso requires conservation of sequences within L, althoughadditional sites within gag may enhance recognition (1, 2, 4).These processes may allow packaging of transcripts from thedel env-J and del env-2 loci in the presence of helper virus ifL region sequences 5' of nt 420 are the major determinants.Sequence conservation was significantly altered from nt 421to nt 667, however, with an overall sequence identity of-60% between either of the endogenous isolates and AKVvirus (Table 1). Comparison of the previously publishedendogenous AL10 sequence with that of AKV virus in the 5'end of the L region revealed 85.5% conservation of thesequence from nt 165 to 420, approximately equal to thatshown by the 3.1 and 15.3 loci. However, in the 3' segmentof L, AL10 contained a 68-nt deletion and a 24-nt insertionwhich contributed to conservation of only 38.5% of thesequence between nt 421 and 667. Most of the conservedsequence in the latter comparison lay between nt 498 and596, with strongly conserved sequence from nt 535 to 579.This was also the most conserved stretch in comparisonsbetween AKV virus and the present del env-] and del env-2
clones, suggesting that this portion of the L region isinvolved in dimerization or encapsidation.
DISCUSSION
We anticipated that design of oligonucleotide probes fortwo cDNA-encoded endogenous retroviral env region dele-tions would result in isolation of two distinct, Gv-J-regulatedproviral loci from the 129 Gix+ host genome. In fact, the delenv-J probe identified a family of four loci, at least three ofwhich contained essentially the same endogenous polytropicproviral sequences. In the del env-J family, at least, conser-vation of the LTR promoter and L region sequences seemsto be adequate for helper virus-dependent replication andreinsertion in the host germ line. One recent report (18)suggests that defective retroviruses can in fact undergointracellular transposition at measurable frequencies. Themultiplicity of del env-J loci causes confusion, however,with regard to the source of Gv-J-regulated transcripts.Since different-sized del env-J RNAs were observed inthymus and liver, it seems likely that there are internaldifferences between these loci, in addition to the PBSalterations detected between the 3.1 locus and the 3.2 and11.1 loci. It also seems likely that tissue-specific factors areresponsible for differential expression from the loci generat-ing 5-kb and 4-kb transcripts; whether these factors repre-sent elements in cis, such as chromatin configuration at theproviral insertion, or in trans, such as diffusible DNA-binding proteins, is a matter for further investigation. In thecase of the sole del env-2 locus in the 129 genome, there canbe only one source for the 5.2-kb transcript observed in thespleen, and it therefore seems likely that this particularinsertion site and the U3 LTR sequence of 15.3 are bothpermissive to expression in a certain population of spleencells. It should be noted that the 3' U3 LTRs from both delenv-J and del env-2 cDNAs function as promoters in anembryonal carcinoma cell line (33). The transcription pat-terns seen in Fig. 1 only identify relative levels of in vivoexpression, since the del env-J deletion has been found inspleen and liver cDNAs and the del env-2 deletion has beenfound in epididymus and liver cDNAs (31, 32; unpublishedobservations).With the identification of 15.3 as the only possible source
of del env-2 transcripts, we were able to corroborate otherlines of evidence indicating that Gv-J regulation acts intrans. Although retroviral probes hybridized to RNA from129 Gix- tissues detect the same array of transcripts as seenin 129 Gix+ tissues, signal intensities are 10-fold lower or less(30). It is possible that a few endogenous loci responsible formost transcripts in 129 GixJ mice were deleted in the processof selecting the Gv-lb phenotype seen in the 129 Gi,jcongenic mice. However, the del env-2 probe recognized thesame 8.9-kb EcoRI fragment in 129 Gix- DNA as in 129 Gix+DNA, showing that a transcriptionally active, Gv-J-regu-lated endogenous retroviral locus was not excised or lostduring the derivation of the congenic strain carrying thealternative Gv-lb regulatory allele (51).The polytropic and modified polytropic endogenous vi-
ruses appear to have diverged most recently of the fourrecognized classes of type C murine retroviruses, based oncomparison of env gene-coding potential (53). Besides theexpected similarities in restriction site maps between the delenv-J and del env-2 loci, we found that they shared identicalgag-pol deletions mediated by direct heptamer repeats at thedeletion breakpoints and identical gag p30 region 19-ntdeletions which interrupted the gag ORF. Since these vi-
ruses are members of the polytropic and modified polytropicclasses, one possibility is that these deletion events occurredindependently. If so, it is also possible that the del env-] loci,appearing to originate from a single, multiply-deleted provi-rus, could each have arisen independently from differentpolytropic viruses undergoing equivalent deletion eventsduring replication. Alternatively, the shared gag-pol deletioncould have arisen from RNA template exchange between atranscript of a polytropic gag-pol-deleted, 3.1-like locus anda transcript from a progenitor of the 15.3 polytropic locusbearing only the del env-J deletion, during minus-strandreverse transcription. If minus-strand DNA transcriptionterminated prematurely within the pol gene of the 15.3template and was completed on a 3.1 RNA template, plus-strand DNA synthesis would then generate a molecule withthe U3 LTR sequence of the original 15.3 polytropic locus(41). In this model, the diversification of the 3.1-like loci byretrovirus-mediated transposition and accumulation of addi-tional mutations by the integrated recombinant 15.3 locuswould then occur independently.The analysis of the coding potential of these loci was
facilitated by the use of in vitro transcription and translation,followed by immunological identification of products. Theretroviral gag gene is particularly amenable to this methodbecause transcript splicing is not required to allow recogni-tion of the initiator AUG and complete ORF translation bythe reticulocyte system. Also, the high degree of sequenceconservation in this region of the type C retroviral genomepermits the use of antibodies raised against Rauscher virus,encoding polypeptides of presumed ecotropic origin, todetermine conservation of antigenic determinants. Loss ofthe env gene splice acceptor site in the del env-J familyprevents the dominant mode of retroviral transcript process-ing, ensuring that the truncated 34-kDa molecule we de-scribe is the major protein product from these loci. In thecase of the del env-2 locus 15.3, splice donor and acceptorsites are conserved and apparently functional. This leads tosignificant levels of both the 5.2-kb unspliced transcriptdirecting translation of the 32-kDa protein and the 1.7-kbtranscript that probably contains only 200 nt of p1SE-codingsequence from the 3' end of the env ORF. Preliminaryevidence suggests that 129 Gix+ mice do contain gag prod-ucts similar to those translated in vitro. Immunoblots ofprotein extracted from 129 Gix+ tissues reacted with anti-Rauscher p12 serum to show the presence of a 30-kDaantigen in spleen and a 33-kDa antigen in thymus (M. C.Wilson, unpublished), consistent with the size and tissue-specific expression of the respective cell-free product sizes(Fig. 5) and tissue-specific expression (Fig. 1) of the delenv-2 and del env-J loci. We have not yet identified thecoding sequence changes which result in the apparent trun-cation of the MX27 gag product (Fig. 4) relative to the MCF247 product. The size difference of -4 kDa (63 versus 67kDa) suggests that the plO ORF is interrupted, but immuno-logical characterization will be necessary to determinewhether p15, p12, and p30 determinants are present, since aninsertion or deletion mutation could result in a long exten-sion in another ORF.There is little prior evidence of gag sequence variation and
its effect on infectious retrovirus functions, and the isolatescharacterized here may be useful in creating recombinantswith infectious ecotropic virus. It has been demonstratedthat an endogenous proviral sequence bearing a PBS com-patible with tRNAGIn can serve as a PBS donor in rescue ofa defective MoMuLV genome (12). The 3 of 18 nt basemismatch of the 3.1 locus PBS with tRNAGln does not rule
out the possibility that it is compatible with another uniden-tified primer. Recombinants containing the extensive 3' Lregion or more conservative p15 and p12 gag sequencedifferences reported here might have significantly alteredphenotypes as infectious virus. The expression of theseendogenous sequences and the presence of defective gaggene products do not seem to block replication by infectiousretrovirus (57). Conversely, the stretches of conserved se-quence in the L region noted above may be adequate toensure encapsidation of endogenous defective retroviraltranscripts in the presence of replication-competent ge-nomes.
ACKNOWLEDGMENTS
We thank J. H. Elder for valued discussions and for criticalcomments on the manuscript, as well as the Rauscher antisera usedin this study. We also thank Jonathan Stoye and John Coffin for theMX27 and Christine Holland for the MCF 247 proviral DNA clones;and we thank Michelle Dietrich for expert secretarial assistance.
This work was supported in part by Public Health Service grantCA-33730 to N;C.W. from the National Institutes of Health, post-doctoral fellowship CA07788 to P.F.P., and a Danish NationalResearch Council (SJVF) grant to M.F.
LITERATURE CITED1. Adam, M. A., and A. D. Miller. 1988. Identification of a signal in
a murine retrovirus that is sufficient for packaging of nonretro-viral RNA into virions. J. Virol. 62:3802-3806.
2. Armentano, D., S.-F. Yu, P. W. Kantoff, T. V. Ruden, W. F.Anderson, and E. Gilboa. 1987. Effect of internal viral sequenceson the utility of retroviral vectors. J. Virol. 61:1647-1650.
3. Bartlett, J. A., R. K. Gaillard, Jr., and W. K. Joklik. 1986.Sequencing of supercoiled plasmid DNA. Biotechniques 4:208-210.
4. Bender, M. A., T. D. Palmer, R. E. Gelinas, and A. D. Miller.1987. Evidence that the packaging signal of Moloney murineleukemia virus extends into the gag region. J. Virol. 61:1639-1646.
5. Besmer, P., H. Fan, M. Paskind, and D. Baltimore. 1979.Isolation and characterization of a mouse cell line containing adefective Moloney murine leukemia virus genome. J. Virol.29:1023-1034.
6. Biggin, M. D., T. J. Gibson, and G. F. Hong. 1983. Buffergradient gels and 35S label as an aid to rapid DNA sequencedetermination. Proc. Natl. Acad. Sci. USA 80:3963-3965.
7. Blin, N., and D. W. Stafford. 1976. A general method forisolation of high molecular weight DNA from eukaryotes.Nucleic Acids Res. 3:2303-2308.
8. Boccara, M., M. Souyri, C. Magarian, E. Stavnezer, and E.Fleissner. 1983. Evidence for a new form of retroviral env
transcript in leukemic and normal mouse lymphoid cells. J.Virol. 48:102-109.
9. Chattopadhyay, S. K., M. R. Lander, S. Gupta, E. Rands, andD. R. Lowy. 1981. Origin of mink cytopathic focus-forming(MCF) viruses: comparison with ecotropic and xenotropic mu-
rine leukemia virus genomes. Virology 113:465-483.10. Cloyd, M. W., and S. K. Chattopadhyay. 1986. A new class of
retrovirus present in many murine leukemia systems. Virology151:31-40.
11. Coffin, J. 1984. Structure of the retroviral genome, p. 261-268.In R. Weiss, N. Teich, H. Varmus, and J. Coffin (ed.), RNAtumor viruses, vol. 1. Cold Spring Harbor Laboratory, ColdSpring Harbor, N.Y.
12. Colicelli, J., and S. P. Goff. 1987. Isolation of a recombinantmurine leukemia virus utilizing a new primer tRNA. J. Virol.57:37-45.
13. Denhardt, D. T. 1966. A membrane-filter technique for thedetection of complementary DNA. Biochem. Biophys. Res.Commun. 23:641-646.
14. Dutta, A., L.-H. Wang, T. Hanafusa, and H. Hanafusa. 1985.Partial nucleotide sequence of Rous sarcoma virus-29 provides
evidence that the original Rous sarcoma virus was replicationdefective. J. Virol. 55:728-735.
15. Evans, L. H., and F. G. Malik. 1987. Class II polytropic murineleukemia viruses (MuLVs) of AKRIJ mice: possible role in thegeneration of class I oncogenic polytropic MuLVs. J. Virol.61:1882-1892.
16. Green, M. R., T. Maniatis, and D. A. Melton. 1983. HumanP-globin pre-mRNA synthesized in vitro is accurately spliced inXenopus oocyte nuclei. Cell 32:681-694.
17. Hartley, J. W., N. K. Wolford, L. J. Old, and W. P. Rowe. 1977.A new class of murine leukemia virus associated with develop-ment of spontaneous lymphomas. Proc. Natl. Acad. Sci. USA74:789-792.
18. Heidmann, T., 0. Heidmann, and J.-F. Nicolas. 1988. Anindicator gene to demonstrate intracellular transposition ofdefective retroviruses. Proc. Natl. Acad. Sci. USA 85:2219-2223.
19. Helms, C., M. Y. Graham, J. E. Dutchik, and M. V. Olson. 1985.A new method for purifying lambda DNA from phage lysates.DNA 4:39-49.
20. Herr, W. 1984. Nucleotide sequence of AKV murine leukemiavirus. J. Virol. 49:471-478.
21. Jenkins, N. A., N. G. Copeland, B. A. Taylor, and B. K. Lee.1982. Organization, distribution, and stability of endogenousecotropic murine leukemia virus DNA sequences in chromo-somes of Mus musculus. J. Virol. 43:26-36.
22. Khan, A. S. 1984. Nucleotide sequence analysis establishes therole of endogenous murine leukemia virus DNA segments information of recombinant mink cell focus-forming murine leu-kemia viruses. J. Virol. 50:864-871.
23. Khan, A. S., F. Laigret, and C. P. Rodi. 1987. Expression ofmink cell focus-forming murine leukemia virus-related tran-scripts in AKR mice. J. Virol. 61:876-882.
24. Khan, A. S., and M. A. Martin. 1983. Endogenous murineleukemia proviral long terminal repeats contain a unique 190-base-pair insert. Proc. Natl. Acad. Sci. USA 80:2699-2703.
25. Khan, A. S., W. P. Rowe, and M. A. Martin. 1982. Cloning ofendogenous murine leukemia virus-related sequences fromchromosomal DNA of BALB/c and AKR/J mice: identificationof an env progenitor of AKR-247 mink cell focus-formingproviral DNA. J. Virol. 44:625-636.
26. Krieg, A. M., A. F. Khan, and A. D. Steinberg. 1988. Multipleendogenous xenotropic and mink cell focus-forming murineleukemia virus-related transcripts are induced by polyclonalimmune activators. J. Virol. 62:3545-3550.
27. Laemmli, U. K. 1970. Cleavage of structural proteins during theassembly of the head of bacteriophage T4. Nature (London)227:680-685.
28. Laigret, F., R. Repaske, K. Boulukos, A. B. Rabson, and A. F.Khan. 1988. Potential progenitor sequences of mink cell focus-forming (MCF) murine leukemia viruses: ecotropic, xenotropic,and MCF-related viral RNAs are detected concurrently inthymus tissues of AKR mice. J. Virol. 62:376-386.
29. Lerner, R. A., C. B. Wilson, B. C. Del Villano, P. J. McConahey,and F. J. Dixon. 1976. Endogenous oncornaviral gene expres-sion in adult and fetal mice: quantitative, histologic, and phys-iologic studies of the major viral glycoprotein, gp7O. J. Exp.Med. 143:151-166.
30. Levy, D. E., R. A. Lerner, and M. C. Wilson. 1982. A geneticlocus regulates the expression of tissue-specific mRNAs frommultiple transcription units. Proc. Natl. Acad. Sci. USA 79:5823-5827.
31. Levy, D. E., R. A. Lerner, and M. C. Wilson. 1985. The Gv-1locus coordinately regulates the expression of multiple endoge-nous murine retroviruses. Cell 41:289-299.
32. Levy, D. E., R. A. Lerner, and M. C. Wilson. 1985. Normalexpression of polymorphic endogenous retroviral RNA contain-ing segments identical to mink cell focus-forming virus. J. Virol.56:691-700.
33. Levy, D. E., R. D. McKinnon, M. N. Brolaski, J. W. Gautsch,and M. C. Wilson. 1987. The 3' long terminal repeat of atranscribed yet defective endogenous retroviral sequence is acompetent promoter of transcription. J. Virol. 61:1261-1265.
34. Maniatis, T., E. F. Fritsch, and J. Sambrook. 1982. Molecularcloning: a laboratory manual, p. 284-285. Cold Spring HarborLaboratory, Cold Spring Harbor, N.Y.
35. Maxam, A. M., and W. Gilbert. 1980. Sequencing end-labeledDNA with base-specific chemical cleavages. Methods Enzymol.65:499-560.
36. McClintock, P. R., J. N. Ihle, and D. R. Joseph. 1977. Expres-sion of AKR murine leukemia virus gp-71-like and BALB/c-likeantigens in normal mouse tissues in the absence of overt virusexpression. J. Exp. Med. 146:422-434.
37. Messing, J., and J. Vieira. 1982. A new pair of M13 vectors forselecting either DNA strand of double-digest restriction frag-ments. Gene 19:269-276.
38. Nikbakht, K. N., L. R. Boone, P. L. Glover, F. E. Myer, andW. K. Yang. 1987. Characterization of a molecular clone ofRFM/Un mouse chromosomal DNA that contains a full-lengthendogenous murine leukemia virus-related proviral genome. J.Gen. Virol. 68:683-693.
39. O'Neill, R. R., A. S. Khan, M. D. Hoggan, J. W. Hartley, M. A.Martin, and R. Repaske. 1986. Specific hybridization probesdemonstrate fewer xenotropic than mink cell focus-formingmurine leukemia virus env-related sequences in DNAs frominbred laboratory mice. J. Virol. 58:359-366.
40. Ou, C.-Y., L. R. Boone, and W. K. Yang. 1983. A novelsequence segment and other nucleotide structural features in thelong terminal repeat of a BALB/c mouse genomic leukemiavirus-related DNA clone. Nucleic Acids Res. 11:5603-5620.
41. Panganiban, A. T., and D. Fiore. 1988. Ordered interstrand andintrastrand DNA transfer during reverse transcription. Science241:1064-1069.
42. Quint, W., W. Boelens, P. V. Wezenbeek, T. Cuypers, E. R.Maandag, G. Selten, and A. Berns. 1984. Generation of AKRmink cell focus-forming viruses: a conserved single-copy xeno-trope-like provirus provides recombinant long terminal repeatsequences. J. Virol. 50:432-438.
43. Rigby, P. W. J., M. Dieckmann, C. Rhodes, and P. Berg. 1977.Labeling deoxyribonucleotide acid to high specific activity invitro by nick translation with DNA polymerase I. J. Mol. Biol.113:237-251.
44. Sanger, F., S. Nicklen, and A. R. Coulson. 1977. DNA sequenc-ing with chain-terminating inhibitors. Proc. Natl. Acad. Sci.USA 74:5463-5467.
45. Schwartzberg, P., J. Colicelli, and S. P. Goff. 1985. Recombina-tion between a defective retrovirus and homologous sequencesin host DNA: reversion by patch repair. J. Virol. 53:719-726.
46. Shinnick, T. M., R. A. Lerner, and J. G. Sutcliffe. 1981.Nucleotide sequence of Moloney murine leukemia virus. Nature(London) 293:543-548.
47. Southern, E. M. 1975. Detection of specific sequences amongDNA fragments separated by gel electrophoresis. J. Mol. Biol.98:503-517.
48. Stephenson, J. R., S. A. Aaronson, P. Arnstein, R. J. Huebner,and S. R. Tronick. 1974. Demonstration of two immunologicallydistinct xenotropic type C RNA viruses of mouse cells. Virol-ogy 61:56-63.
49. Stephenson, J. R., S. R. Tronick, R. K. Reynolds, and S. A.Aaronson. 1974. Isolation and characterization of C-type virusgene products of virus-negative mouse cells. J. Exp. Med.139:427-438.
50. Stockert, E., E. A. Boyse, H. Sato, and K. Itakura. 1976.Heredity of the Gix thymocyte antigen associated with murineleukemia virus: segregation data simulating genetic linkage.Proc. Natl. Acad. Sci. USA 73:2077-2081.
51. Stockert, E., E. A. Boyse, Y. Obata, H. Ikeda, N. H. Sarkar, andH. A. Hoffman. 1975. New mutant and congenic mouse strainsexpressing the murine leukemia virus-associated thymocytesurface antigen Gix J. Exp. Med. 142:512-517.
52. Stockert, E., L. J. Old, and E. A. Boyse. 1971. A cell surfacealloantigen associated with murine leukemia virus: implicationsregarding chromosome integration of the viral genome. J. Exp.Med. 133:1334-1355.
53. Stoye, J. P., and J. M. Coffin. 1987. The four classes ofendogenous murine leukemia virus: structural relationships and
potential for recombination. J. Virol. 61:2659-2669.54. Stoye, J. P., and J. M. Coffin. 1988. Polymorphism of endoge-
nous proviruses revealed by using virus class-specific oligonu-cleotide probes. J. Virol. 62:168-175.
55. Thomas, P. S. 1980. Hybridization of denatured RNA and smallDNA fragments transferred to nitrocellulose. Proc. Natl. Acad.Sci. USA 77:5201-5205.
56. Varmus, H., and R. Swanstrom. 1984. Replication of retrovi-ruses, p. 369-512. In R. Weiss, N. Teich, H. Varmus, and J.Coffin (ed.), RNA tumor viruses, vol. 1. Cold Spring HarborLaboratory, Cold Spring Harbor, N.Y.
57. Villar, C. J., T. N. Fredrickson, and C. A. Kozak. 1988. Effect ofthe Gv-1 locus on Moloney ecotropic murine leukemia virus-induced disease in inbred and wild mice. Curr. Top. Microbiol.Immunol. 137:250-255.
58. Woo, S. L. C. 1979. A sensitive and rapid method for recombi-nant phage screening. Methods Enzymol. 68:389-395.
59. Wood, W. I., J. Gitschier, L. A. Lasky, and R. M. Lawn. 1985.Base composition-independent hybridization in tetramethylam-monium chloride: a method for oligonucleotide screening ofhighly complex gene libraries. Proc. Natl. Acad. Sci. USA82:1585-1588.
