Effects of 3′ Untranslated Region Mutations on Plus-Strand Priming during Moloney Murine Leukemia Virus Replication

Abstract

A conserved purine-rich motif located near the 3′ end of retroviral genomes is involved in the initiation of plus-strand DNA synthesis. We mutated sequences both within and flanking the Moloney murine leukemia virus polypurine tract (PPT) and determined the effects of these alterations on viral DNA synthesis and replication. Our results demonstrated that both changes in highly conserved PPT positions and a mutation that left only the cleavage-proximal half of the PPT intact led to delayed replication and reduced the colony-forming titer of replication defective retroviral vectors. A mutation that altered the cleavage proximal half of the PPT and certain 3′ untranslated region mutations upstream of the PPT were incompatible with or severely impaired viral replication. To distinguish defects in plus-strand priming from other replication defects and to assess the relative use of mutant and wild-type PPTs, we examined plus-strand priming from an ectopic, secondary PPT inserted in U3. The results demonstrated that the analyzed mutations within the PPT primarily affected plus-strand priming whereas mutations upstream of the PPT appeared to affect both plus-strand priming and other stages of viral replication.

Retroviral reverse transcription generates a double-stranded DNA copy of the single-stranded viral RNA genome. The primer used to initiate plus-strand DNA synthesis is a nucleolytic product of the viral genomic RNA. During minus-strand synthesis, the RNase H activity of reverse transcriptase (RT) degrades much of the genomic RNA as it forms an RNA-DNA duplex. However, the polypurine tract (PPT) fragment persists. The PPT’s protection from degradation and subsequent selection as the plus-strand primer require high degrees of molecular specificity (6).

All retroviral PPT regions are purine rich, but their composition differs from virus to virus. In Fig. 1, we aligned sequences of retrovirus and retroelement PPT regions. As in previous reports (34), −1 is defined as the nucleotide 5′ of the primer cleavage site. Among sequences in this compilation, the −4 position is 93% conserved and the −2 position is 86% conserved. Similar conservation for these positions is apparent in other PPT compilations (34), but PPT conservation is less pronounced in some retroviruses than others. For example, in spleen necrosis virus, the PPT itself differs significantly from the consensus (43), and a conserved T stretch whose presence upstream of many retroviral PPTs has previously been noted (31) is absent from caprine arthritis-encephalitis virus (37). Based on PPT length and a prominent oligoribonucleotide that primes plus-strand synthesis in vitro, we will consider the Moloney murine leukemia virus (M-MuLV) PPT to span from −1 through −13 (35).

FIG. 1.

Retroviral and retroelement PPT region sequences. Sequences were aligned by plus-strand primer cleavage site as indicated by the vertical line. The M-MuLV PPT is shaded; consensus bases greater than 75% conserved are shown in bold. Sequences are from M-MuLV (44), avian leukosis virus (ALV) (1), HIV (33), feline leukemia virus (FeLV) (9), human adult T-cell leukemia virus (HTLV) (42), mouse mammary tumor virus (MMTV) (26), Mason-Pfizer monkey virus (MPMV) (45), Rous sarcoma virus (RSV) (40), SIV (16), simian retrovirus type 1 (SRV) (32), mouse virus-like retrotransposon BVL-1 (VL30) (18), mouse intracisternal A-Particle (IAP) (19), caprine arthritis-encephalitis virus (CAEV) (37), and spleen necrosis virus (SNV) (43).

Early evidence for the role of the PPT region in plus-strand priming comes from the work of Sorge and Hughes (46), who showed that at least 9 and no more than 29 nucleotides upstream of the avian sarcoma virus (ASV) long terminal repeat (LTR) are required in cis for viral replication. Most subsequent plus-strand priming studies have been performed with model substrates or in permeabilized virions rather than during viral replication. Rattray and Champoux made point mutations in the M-MuLV PPT and found that sequences downstream of the PPT have no detectable effect on priming specificity in model reactions, but that mutations at −1, −2, −4, and −7 cause additional cleavage sites (34). Powell and Levin have shown that only the six G residues at the PPT 3′ end (that is, −1 through −6) are necessary for human immunodeficiency virus type 1 (HIV) plus-strand priming in model reactions (31). In those studies, plus-strand priming was the same with the PPT in two different sequence contexts on short primer templates, thus suggesting that sequences around the PPT do not affect plus-strand priming.

Here, we developed a system to examine the extent to which sequences within the PPT that alter priming in model reactions contribute to this process during virus replication. We also examined the roles of sequences upstream of the PPT which contact RT during plus-strand primer generation. Targeted alterations in both highly conserved regions within the M-MuLV PPT and in upstream regions were made, and the effects of these mutations on viral replication and on plus-strand priming were examined. Our results confirmed the importance of sequences within the PPT and also demonstrated that sequences upstream of the PPT were important for plus-strand priming. These 3′ untranslated region (3′ UTR) sequences upstream of the PPT appeared to function in other stages of viral replication as well.

MATERIALS AND METHODS

Plasmid construction.

3′ UTR mutations were introduced into a M-MuLV provirus plasmid and into a plasmid encoding a puromycin resistance (Puro^r)-conferring replication-defective retroviral vector. The proviral plasmids were derivatives of pNR163-4 or pMLV-neo. pNR163-4 is a derivative of a “tipless” M-MuLV plasmid (22) which encodes intact M-MuLV RNA. pNR163-4 contains a silent mutation that introduces an MfeI site in env at M-MuLV position 7746. pMLV-neo contains the neomycin resistance Neo^r gene in the backbone of a provirus plasmid. An EcoRI-to-XbaI fragment containing the Neo^r gene plus the PGK promoter and polyadenylation signals from a pPNT derivative (53) was subcloned into pNCA (8). The drug resistance gene is transcribed opposite from the M-MuLV LTR and is not incorporated into viral transcripts (Fig. 2A). Retroviral vectors were derivatives of pAM86-5 (22) (Fig. 2B). 3′ UTR mutations were introduced by using PCR-mediated site-directed mutagenesis and other standard recombinant DNA techniques. Sequences of all 3′ UTR regions from the end of env through early in U3 were confirmed by dideoxy sequencing using a Sequenase II kit (U.S. Biochemical).

FIG. 2.

Structures of proviral clones and retroviral vectors. (A) Proviral clones. Single PPT mutations were introduced into pNR163-4, and tandem PPT mutations were introduced into pMLV-neo. Direction of Neo^r transcription is indicated with an arrow. att_wt, wild-type att site. (B) pAM86-5, the retroviral vector into which PPT mutations were introduced. The direction of Puro^r transcription is indicated with an arrow. (C) Structures and predicted products of tandem-PPT constructs. The inserted secondary PPT, designated PPT2, and mutated att (att_mut) at the 5′ end of U3 are indicated. The first line indicates the structure of the transfected proviral constructs, the second line represents the structure of encapsidated RNAs, and the final lines represent the two kinds of reverse transcription products templated by tandem-PPT RNAs. Note that PPT regions are shown as disproportionately large to present detail important to this study. Primer extension products used to analyze PPT use (Fig. 8 and Table 3) are represented with gray dashed arrows.

Tandem-PPT constructs contained the following insertion in the U3 NheI site: AGCGGCCGCATAAAATAAAAGATTTTATTTAGTCTCCAGAAAAAGG GGGGAACAACAAAA. This PPT2 insertion is a duplication of sequences upstream of and including the PPT (underlined) followed by a mutated att (attachment) sequence. All mutant tandem-PPT plasmids retained this wild-type PPT sequence in PPT2 except −21/−28PPT2, which contained mutant upstream sequences in PPT2 and wild-type sequences upstream of PPT1.

Cells and viruses.

NIH 3T3 cells, D17 cells, Rat2 cells, and derivatives were grown in Dulbecco’s modified Eagle’s medium supplemented with 10% calf serum (Gibco). ΦNXA retroviral packaging cells (29) and derived cell lines were grown in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum (HyClone).

Stably transfected vector-producing cell pools were established by transfecting ΦNXA cells with vector plasmids, using Lipofectamine (Gibco) according to the manufacturer’s instructions. Puro^r cells were selected in 1 μg of puromycin (Sigma) per ml. More than 100 transfectants were pooled to generate each producer pool.

Viral supernatants (5 ml) were harvested from 10-cm-diameter plates of 90% confluent producer cells at 12-h intervals, filtered through 0.45-μm-pore-size filters (Fisher), and stored at −70°C prior to use. Virion proteins were quantified by determining RT DNA polymerase activity levels (52). Briefly, 5 μl of virus-containing culture medium was added to 25 μl of a mixture containing 60 mM Tris (pH 8.3), 24 mM dithiothreitol, 0.7 mM MnCl₂, 75 mM NaCl, oligo(dT) (6 μg/ml), poly(rA) (12 μg/ml), [α-³²P]dTTP (20 nCi/μl), 12 μM dTTP, and 0.06% Nonidet P-40 and then incubated for 60 min at 37°C. Duplicate 5-μl aliquots were spotted on DEAE-paper and washed with 0.3 M NaCl–30 mM sodium citrate (pH 7). Incorporated dTTP retained on washed DEAE-paper was quantified in a scintillation counter or with Image-quaNT software and a PhosphorImager (Molecular Dynamics). Serial dilutions in fresh culture medium of wild-type virus were assayed under the same conditions to establish a standard curve (Table 1).

TABLE 1.

Quantitative RT assay standard curve

Dilutiona	[α-32P]dTMP incorporated (cpm)
	Sample 1b	Sample 2b	Avg
1	1,133,341	985,115	1,059,228
0.5	578,791	542,558	560,675
0.25	266,503	277,304	271,904
0.125	146,212	140,005	143,109
0.0625	75,516	70,240	72,878
0.03125	41,951	40,014	40,983

^aSerial dilutions of wild-type M-MuLV. 

^bLowest reported values were >5-fold above background. 

Viral RNA levels were quantified by slot blotting as follows. Virus was pelleted from 7 to 9 ml of culture medium for 25 min at 3 × 10⁵ g at 4°C in a model M120EX microultracentrifuge (Sorvall). Virions were disrupted by incubation in 200 μl of a mixture containing 100 mM NaCl, 50 mM Tris (pH 7.5), 10 mM EDTA, 1% sodium dodecyl sulfate, proteinase K (100 μg/ml), and yeast tRNA (50 μg/ml) at 37°C for 30 min. Samples were extracted with an equal-volume mixture of phenol plus chloroform and precipitated with ethanol, and pellets were suspended in 400 μl of diethylpyrocarbonate-treated H₂O. Duplicate 200-μl samples or dilutions within the linear range of the assay were applied to a nylon membrane (Hybond N; Amersham) by using a BioDot slot blotter (Bio-Rad) and hybridized with a ³²P-labeled probe generated with a random primer kit (Boehringer Mannheim Biochemicals) under standard hybridization conditions (38). The probe was synthesized from a Puro^r-encoding fragment of pAM86-5. Quantification was performed with a PhosphorImager as described above. Serial dilutions of wild-type viral RNA were blotted, hybridized, and quantified to establish a standard curve (Table 2); 1× was defined as the first RNA dilution above the linear range of the assay.

TABLE 2.

RNA slot blot standard curve

Dilutiona	[α-32P]dCMP retained (cpm)
	Sample 1b	Sample 2b	Avg
1	15,553	12,633	14,093
0.5	13,230	12,813	13,022
0.25	6,909	6,829	6,869
0.125	2,507	2,457	2,482
0.0625	1,361	1,104	1,233

^aSerial dilutions of RNA isolated from culture media of cells chronically infected with wild-type M-MuLV. 

^bLowest reported values were ∼2-fold above background. 

Puro^r CFU titers were obtained by infecting 3T3 and D17 cells and counting Puro^r colonies. Infections were performed in the presence of 8 μg hexadimethrine bromide (Polybrene; Sigma) per ml for 2 h at 37°C. Two days after infection, Puro^r cells were selected in 6 (3T3 cells) or 2 (D17 cells) μg of puromycin per ml. CFU titers were normalized to viral RNA levels determined by RNA slot blotting and to virion protein levels determined by RT assay. Wild-type vector CFU per unit of RNA was assigned a value of 1. Titers, RNA levels, and virion levels were determined in at least three independent experiments using at least two different viral stocks for each mutant.

Replication efficiency assay.

Replication efficiency was determined as follows. Ten percent confluent 35-mm-diameter plates of 3T3 cells were transfected as described above with pNR163-4 or pMLV-Neo^r derivatives. After 2 days, all cells were passaged to 6-cm-diameter plates. Culture medium was sampled, and cells were passaged 1:10 every 3 days thereafter. Virus spread was assayed by RT activity (15). Each mutant was tested at least twice. Note that this experimental protocol prevented direct measurements of transfection efficiencies. However, all experiments included wild-type and mutant controls, and the consistency between wild-type and mutant values in each experimental repetition suggests that transfection efficiencies were similar for all samples within each experiment.

Virus spread following infection of fresh cells was monitored as follows. Equal amounts of each virus, quantified by RT assay as described above, were used to infect 10% confluent 3T3 cells. Culture medium was sampled once cells became confluent. Cells were subsequently passaged 1:5, and the medium was sampled every 2 days. Virus spread was detected by RT activity.

Preparation and analysis of viral DNA.

Low-molecular-weight DNA enriched in unintegrated viral DNA was extracted from Rat2 cells 24 h postinfection (17). Viral DNA was analyzed by PCR of the PPT region, using primers specific for env and U3, followed by restriction analysis. For sequencing, ClaI-to-NheI restriction fragments of these PCR products were introduced into pUC19, and individual subclones were sequenced.

PPT use for the tandem PPT constructs was analyzed by reiterative primer extension. Briefly, antisense U5 (CCCCGCTGACGGGTAGTC; complementary to positions 11 to 28 from the downstream edge of U5) and U3 (CCATCTGTTCCTGACCTTGATCTGAACTTC; complementary to positions 88 to 117 from the upstream edge of U3) primers were 5′ ³²P end labeled with polynucleotide kinase and [α-³²P]ATP (Amersham). Reaction mixtures of 50 μl contained 50 mM KCl, 10 mM Tris-HCl, 2mM MgCl₂, 200 μM deoxynucleoside triphosphates, 4 pmol of labeled primer, and 2 U of Taq polymerase. For each reaction, 1/10 of the low-molecular-weight DNA harvested from a 10-cm-diameter culture plate was used as the template. Primer extension products were generated by 30 cycles of denaturation (60 s at 94°C), annealing (60 s at 45°C), and extension (45 s at 72°C). Products were extracted with an equal-volume mixture of phenol plus chloroform, collected by ethanol precipitation, and suspended in 10% formamide. Products were denatured by heating at 80°C for 2 min prior to loading on a 6% polyacrylamide–8 M urea gel. Dried gels were exposed to autoradiography film. Products were quantified with a PhosphorImager as described above. Relative PPT use values obtained in separate experiments differed by about 10%.

Sequence alignments were performed with Lasergene MegAlign (DNASTAR, Inc.). BLAST searches (www.ncbi.nlm.nih.gov/BLAST/) were performed to identify the origin of sequences found in −21/−28 revertants.

RESULTS

Retroviral vectors and proviral clones with targeted alterations to the M-MuLV PPT region.

Targeted alterations were introduced into the 3′ UTRs of infectious proviral clones and replication-defective retroviral vectors (Fig. 2A and B). These mutations allowed us to examine effects of PPT region alterations on both a single round and multiple rounds of viral replication. The 3′ UTR mutations were also introduced into proviral clones containing secondary PPTs that served as competing plus-strand priming sites in cis (Fig. 2C).

Mutations (summarized in Fig. 3) included point mutations in highly conserved PPT positions and larger substitutions in sequences within and near the PPT. Our rationale in making upstream substitutions was that during cleavage of the plus-strand primer, RT is oriented with its RNase H active site at the 3′ end of the PPT and its DNA polymerase domain extending in the 5′ direction (20). Bound in this orientation, RT’s DNA polymerase domain contacts about 25 nucleotides upstream of the point of cleavage (54).

FIG. 3.

3′ UTR mutations. The wild-type M-MuLV sequence is on the first line. For mutants, bold indicates differences from the wild-type sequence and dashes indicate deletions. The PPT is shaded gray. The vertical line marks the PPT/U3 boundary and the normal site of plus-strand primer cleavage. Footprint of RT positioned to cleave the plus-strand primer (54) is underlined.

The −1/−6 substitution changed all PPT bases between −1 and −6, including the highly conserved −2 and −4 positions. The −7/−17 mutation altered the upstream half of the PPT. The −14/−19 and −21/−28 mutations were in sequences upstream of the canonical PPT. Both mutations lie within the predicted footprint of RT positioned to cleave the plus-strand primer (Fig. 3) (54) and within the region implicated by Sorge and Hughes as being required for viral replication in ASV (46). The −21/−28 substitutions also disrupted an A · T-rich region which is conserved upstream of several PPTs (Fig. 1). The −14/−19 substitutions extended purine-rich sequences to upstream of the PPT. None of the PPT substitutions altered the spacing of viral elements.

3′ UTR alterations and virus replication.

Proviral clones were transfected into 3T3 cells, transfected cells were serially passaged, and virus spread was monitored. In separate experiments, viral spread was monitored after 3T3 cells were infected with uniform amounts of infectious mutant viruses. Results are summarized in Fig. 4A and C.

FIG. 4.

Replication of 3′ UTR mutants. The leftmost edges of the dark bars indicate time points at which virus was first detected by RT activity. Standard deviations and number of experimental repetitions for each mutant are presented at the right side (Std.Dev., standard deviation; #, number of experimental repetitions). (A) Time course of spread of single-PPT virus after transfection of 3T3 cells. Values for −21/−28 and −14/−19 are averages of values obtained from those experiments where virus spread was detected. Variability in time course of viral replication for these mutants reflects the reversion that was required for viral spread. (B) Time course of spread of tandem-PPT virus after transfection of 3T3 cells. (C) Time course of spread of single-PPT virus after infection of 3T3 cells with mutant or revertant virus. Also shown is the replication time course for three dilutions of wild-type virus. All infections of mutants were performed with an amount of virus equivalent to 1× wild-type virus as determined by RT activity. Revertant sequences are shown in Fig. 5B and C. (D) Time course of spread of tandem-PPT virus after infection of 3T3 cells with mutant or revertant virus. Results are from a single experiment. The −21/−28 tandem rev1 revertant sequence is shown in Fig. 5d.

These assays revealed that single-base substitutions in highly conserved PPT residues caused relatively modest delays in virus spread, similar to delays observed when the input of wild-type virus was reduced 10- to 100-fold (Fig. 4A and C). The −7/−17 substitution in the upstream half of the PPT was only slightly more delayed than it was in the point mutants. Restriction and sequence analysis of viral DNAs produced after several rounds of replication revealed that for each of these mutants, the original mutation was retained throughout passage. Viral spread was not detected within 60 days after transfection for the −1/−6 substitution in the cleavage-proximal half of the PPT.

Replication of the −14/−19 mutant was detectable only after a significant delay in two experiments or not at all in a third experiment. Analysis of viruses detected after several rounds of cell passage revealed that the PPT region was altered. The revertant detected after one −14/−19 mutant transfection experiment (−14/−19rev1) contained a single substitution within the originally altered sequence (Fig. 5B). This substitution was initially detected because it obliterated the DraI restriction site that tagged the −14/−19 mutation (Fig. 5A, lanes 7 to 10). This revertant replicated more efficiently than the −4C and −7/−17 mutants (Fig. 4C). Virus from another −14/−19 transfection experiment (−14/−19rev2) reverted by undergoing more significant sequence alterations upstream of the PPT (Fig. 5B). In both cases, the reverted genome was the only viral product detectable among replicating virus DNA. In the third −14/−19 experiment, virus spread remained undetectable throughout 90 days of transfected cell passage.

FIG. 5.

Analysis of PPT regions after mutant virus replication. (A) PCR amplification of the region between env and the beginning of U3 from unintegrated viral DNA harvested 24 h after infection of Rat2 cells, and analysis for PPT mutation-associated restriction sites. Arrows indicate cut PCR product (products which retained restriction site that tagged parental mutation) and uncut fragments. Mutant tested, source of PCR template, and restriction enzymes used are indicated at the top. Passage 1 refers to DNA templated by virus generated after transfection of 3T3 cells and subsequent viral spread within the transfected cells; passage 2 refers to DNA templated by virus generated after additional replication in fresh cells of virus used for passage 1. Lanes: 1, marker (positions indicated in base pairs at the left); 2, no template PCR; 3 and 4, −21/−28 plasmid; 5 to 8, −21/−28rev5 virus; 9 and 10, −14/−19 plasmid; 11 and 12, −14/−19rev1 virus. In lane 5, the larger band corresponds to the revertant sequence (shown in panel C) and the shorter band corresponds to the original mutant sequence. The implications of digestion patterns observed here—that the −21/−28 mutant replicated as a mixed population of revertant and parental mutant virus but that all replication products of the −14/−19 mutant were revertants—was confirmed by subcloning and sequencing individual replication products. Slight differences in migration patterns between digested and nondigested lanes were due to salt effects. (B) Sequences of −14/−19 revertants. Shown are PPT and upstream sequences before and after replication in 3T3 cells from two individual experiments. Revertants from the first and second experiments are designated rev1 and rev2, respectively. (C) Sequences of −21/−28 revertants. The first number after the suffix “rev” indicates which transfection experiment yielded the indicated revertant, and the number after the dash indicates each sequence class found within the population of products from each transfection experiment. Insertions in rev3-1 and rev3-2 apparently derive from a VL30-type retroelement called BVL-1 (18), and sequences in rev4-2 and rev5 are from recombination with an endogenous murine retrovirus (47, 48). (D) −21/−28 tandem-PPT revertants. PPTA is the PPT acquired by one revertant between PPT1 and PPT2. The −21/−28 tandem rev1, 3 sequence was obtained twice in two separate transfection experiments; the −21/−28 rev2 sequence was obtained in one experiment. Parental mutations are indicated by bold capital letters; dashes indicate sites of insertions or gaps in the sequence; underlined bold lowercase letters indicate sequences that differ from the parental mutant. PPT sequences are shaded gray, and cleavage sites are indicated by a vertical line.

The time course of detectable virus spread suggested that the −21/−28 mutant was severely impaired (Fig. 4A). Spread was observed in only 7 of 10 repetitions of the transfection experiment. When spread was detected, viral DNA was initially tested for retention of the NotI restriction site used to tag the original mutation, and several individual subclones of viral DNA from each transfection experiment were subsequently sequenced.

The −21/−28 mutant appeared to be weakly infectious on its own. In two experiments, the original mutation was still detectable after serial virus passage, but in most experimental repetitions, virus populations were dominated by revertants after extensive passage. Unlike other mutants, the virus found replicating after transfection with −21/−28 proviral plasmids often appeared as mixed populations of the parental mutant with one or more revertants (for example, Fig. 5A and −21/−28 revertant sequences in Fig. 5C). The time course of revertant spread through 3T3 cells was variable, with different revertant populations spreading either at the same rate as the wild type or else with significant residual delays.

Analysis of virus which emerged from two transfection experiments (−21/−28rev4-2 and −21/−28rev5) revealed that whereas a significant portion of the preintegrative DNA contained the original mutant sequence, the original substitution had been replaced with nonviral sequences in other viral DNAs within the population. A BLAST search identified these nonviral sequences as being highly homologous to portions of several retroelements, including endogenous murine retrovirus MX27, the modified polytropic endogenous murine retrovirus MX33, and the murine retrovirus LTR insertion linked to the hairless mutation (47, 48) (Fig. 5C). The presence of endogenous retrovirus sequences suggests that in these viruses, the original mutation had been reverted by patch repair (41). Less of the original −21/−28 mutant and more of the revertant were detected within the viral population after subsequent infection with this mixed virus than after the original transfection (Fig. 5A, lanes 6 and 8).

Other classes of revertants arose during separate −21/−28 transfection experiments. Some (for example, −21/−28rev2) contained only point mutations within the parental mutant sequences and may or may not have acquired increased replication fitness. In one repetition of the transfection experiment, virus was found with duplicate M-MuLV PPTs flanking a 61-bp insert that appeared to be derived from a VL30-type retroelement called BVL-1 (−21/−28rev3) (18). This insertion brought sequences similar to the T-rich PPT upstream region consensus (Fig. 1) into position flanking the downstream PPT. Yet another transfection experiment yielded a series of revertants (−21/−28rev1) with different lengths of additional A residues inserted into the PPT without alterations to the original −21/−28 mutations. The origin of function of these A-tract insertions is unclear, but they resulted in altered spacing between the PPT and the −21/−28 mutations. Further analysis will be required to determine whether all of the structurally disparate reversions that we observed restored the same function(s) to the −21/−28 mutant.

It is striking to note that one revertant of the −14/−19 mutant (−14/−19rev2 [Fig. 5]) had an A-tract insertion that extended the PPT in a manner similar to the insertions seen in some −21/−28 revertants and that A-tract insertions have also been observed in experiments using a different retroviral system. In a study of the simian immunodeficiency virus (SIV) T-rich stretch, Ilyinskii and Desrosiers found similar insertion of T-rich sequences upstream of the PPT and elongation of the PPT A stretch among revertants of T-stretch mutants (21).

Effects of 3′ UTR alterations on a single round of vector DNA synthesis.

To examine effects of 3′ UTR alterations on a single round of replication, we built mutations into retroviral vectors and determined vector-templated integrant titers. Vector producer cells were generated by stably transfecting packaging cells with vector plasmids, and virus was harvested from packaging cell pools. Vector DNA production in infected 3T3 or D17 cells was monitored by comparing Puro^r titers per unit of virion RNA. As predicted, we found that our PPT alterations, because they were located upstream of the plus-strand priming site, were not present at the tips of preintegrative DNA (see Fig. 8). Because our 3′ UTR mutations did not affect the structure of preintegrative DNA and hence should not affect its ability to become integrated, we assumed that reductions in titer per unit of RNA reflected reductions in amounts of DNA synthesized.

FIG. 8.

Primer extension analysis of PPT use among preintegrative DNA products of tandem-PPT viruses performed with a U5 antisense primer. Marker sizes (left) and PPT product sizes (right) are indicated in base pairs. Arrows indicate products from PPT1 and PPT2 use. ΔPPT1 and ΔPPT2 indicate products where one enhancer repeat was deleted from the PPT1 and PPT2 products, respectively. For −21/−28rev1, which has three PPTs, PPT1 and PPTA products are indicated (Fig. 5D). For the wild-type, −2C, −4C, −7/−17, −14/−19 and −21/−28PPT2 constructs, PPT1 is 640 bp, ΔPPT1 is 565 bp, PPT2 is 555 bp, and ΔPPT2 is 480 bp. For −21/−28rev1, PPT1 is 725 bp, ΔPPT1 is 650 bp, PPTA is 640 bp, ΔPPTA is 565 bp, PPT2 is 555 bp, and ΔPPT2 is 480 bp. Origins of the faint bands in lanes 3 and 6, which migrate at the same mobility as the −21/−28 PPT1 band, were not determined, but these bands are presumed to be products of rare triple-PPT viruses. The longer band in lane 1 corresponds with priming from the single wild-type M-MuLV PPT, and the shorter bands are enhancer deletions of this product. Longer products which resulted from primer annealing to the downstream LTR are not visualized in this figure.

Results are summarized in Fig. 6. All mutants except the −7/−17 mutant consistently yielded lower colony counts per unit of viral RNA than the wild-type vector, suggesting that most of the mutations affected plus-strand priming. Note that the titers for some mutants (most notably the −7/−17 mutant) were highly variable, suggesting that some mutations may have been unstable. Data in Fig. 6 represent three assay repetitions; variability for this mutant was also evident in additional assays not shown. The −14/−19 and −21/−28 mutants had low titers. This finding was consistent with the inability of viruses harboring these mutations to undergo multiple rounds of replication without reverting but indicated that some reverse transcription was possible for these mutants. The −1/−6 mutant, replication of which was never detected, yielded a vector titer which was 1% of that of the wild type, comparable to the titer reported for a vector with no PPTs (3). Note that Fig. 6 presents titers as the number of drug-resistant colonies per unit of RNA; values for colonies per unit of virion were similar. Whereas RNA loss during purification was not measured, the good agreement in values for colonies per unit of RNA and per unit of virion obtained in experimental repetitions suggests that variations in RNA loss during purification were not great enough to seriously affect our results.

FIG. 6.

Single-PPT mutant vector titers. Colonies per unit of RNA ratios for wild-type PPT vectors were given a value of 1. Error bars reflect variation in values for four (wild type [wt]), three (−2C, −4C, −7/−17, −14/−19, and −21/−28 mutants), two (mutant att), and one (−1/−6 mutant) repetitions of each experiment. Methods for virus and RNA quantification are described in Materials and Methods.

Establishment of a tandem-PPT system for studying the effects of 3′ UTR alterations.

We also examined plus-strand priming from a nonessential site by using tandem-PPT viruses (Fig. 2C). Our intention with these tandem-PPT constructs was to assess effects on plus-strand priming separately from other replication effects. Bowman et al. have shown that if two PPTs are built into a single vector, each can serve to prime plus-strand synthesis some of the time (2). In our experiments, we cloned a secondary PPT and flanking sequence (PPT2) into the upstream edge of U3 (Fig. 2C). Previous studies have shown that insertions at this site have minimal consequences for viral replication (23, 49). For these experiments, we introduced alterations into the PPT1 region while maintaining wild-type sequences in PPT2.

In our constructs, PPT2 was adjacent to a mutated integration att site (Fig. 3). This was intended to ensure that if a deletion occurred between the tandem PPTs during reverse transcription, then the virus would be rendered noninfectious since the resulting DNA would lack functional integration signals on one end. Note that despite our intention of creating an unusable att site (5, 7), single-PPT virus containing our mutated att could replicate without reversion, albeit with a significant delay relative to the wild type (Fig. 4A). Vectors with this att mutation yielded a titer slightly higher than that obtained with integrase mutants (Fig. 6A) (27, 51).

PCR analysis of preintegrative DNAs generated after several rounds of replication showed that both PPTs were retained as intended in the vast majority of replication products (Fig. 7), suggesting that DNAs ending with the mutant att were integration defective and at a selective disadvantage for further replication. This maintenance of both PPTs ensured that the reverse transcription machinery was provided with a choice of two competing PPTs during nearly all rounds of viral replication.

FIG. 7.

PCR analysis of tandem-PPT viral DNA after passage through 3T3 cells. The region between envelope and U5 was amplified from unintegrated proviral DNA purified from Rat2 cells infected with tandem-PPT virus. The size of the mutant att PCR product corresponds to the expected product size if the tandem PPT had been deleted between PPT1 and PPT2. Arrows indicate one, two, and three PPT product sizes. Marker positions are indicated in base pairs at the left.

Effects of tandem PPT 3′ UTR mutations on viral replication.

Tandem-PPT proviral plasmids were transfected into 3T3 cells, and viral spread was monitored (Fig. 4B and D). Replication of −2C, −4C, and −7/−17 tandem-PPT mutants was detectable at time points similar to those for the corresponding single-PPT mutants (Fig. 4A). The −14/−19 mutant replicated more efficiently in the tandem-PPT context than in the single-PPT virus. After several rounds of replication, we detected tandem −14/−19 replication products that did not contain reversions, suggesting that sequences in the ectopic PPT might have compensated for the −14/−19 defect. Surprisingly, the −1/−6 tandem-PPT mutant reproducibly failed to replicate, even though the presence of PPT2 might have been predicted to allow replication as efficient as that of mutant att virus. The stage where replication of this mutant was blocked was not examined. PCR and sequence analysis of DNA synthesized by all tandem-PPT viruses that did replicate revealed that all tandem-PPT mutants except the −21/−28 mutant retained their original two PPTs without reversions even after prolonged passage (Fig. 7 and data not shown). The −21/−28 mutant yielded two env/U3 region PCR products, one which corresponded to the presence of the original two PPT and another which was longer (Fig. 7, lane 9).

Sequence analysis of the longer −21/−28 products revealed that it contained a third wild-type PPT (PPTA) between the two PPTs of the original construct (Fig. 5D). This revertant contained a −21/−28 PPT region plus wild-type att followed by a wild-type PPT plus wild-type att and then another wild-type PPT plus mutant att. This third PPT insertion, which presumably arose via intermolecular template switching during reverse transcription of copackaged tandem −21/−28 RNAs, was observed in all three repetitions of this experiment. However, as with the −21/−28 mutation in the single PPT, some unaltered virus with the original tandem −21/−28 sequence could be detected in addition to the altered sequence among preintegrative DNAs produced in one experiment. When the −21/−28 mutation was introduced into PPT2 instead of PPT1, the resultant −21/−28 PPT2 virus was capable of replicating with wild-type efficiency without reversion (Fig. 4B).

Assessing PPT use with the tandem-PPT system.

To determine the relative abundance of products primed by PPT1 and by PPT2 during a single cycle of reverse transcription, we infected fresh Rat2 cells with tandem-PPT viruses and harvested preintegrative DNAs at a uniform time point postinfection.

Two species of preintegrative DNAs should exist within the populations templated by our tandem-PPT viruses: one with a left end that results from the use of PPT1 and a second that results from the use of PPT2 (Fig. 2C). By comparing the ratio of the two species among the DNA products of a virus whose PPTs were both wild type to the ratio of these species for a virus with one wild-type and one mutant PPT, we examined the effects of our PPT alterations on plus-strand initiation site selection.

We assumed that differences in product prevalence reflected differences in the DNA products’ production. This is because by their nature, PPT sequences specify viral DNA ends but do not become a part of those ends. Both the PPT1- and PPT2-primed DNA products of each tandem-PPT RNA had the same terminal sequences as the corresponding products from every other tandem-PPT RNA, and hence the rate of integration of these DNAs will not be affected by which PPT specified their ends. Note that mutant PPT products which are relatively abundant at one time point should remain relatively abundant at all time points, thus ensuring that assessing any single time point should accurately reveal trends among mutants in relative PPT use. However, the precise magnitude of differences among mutants would vary at different time points since some viral DNAs with intact integration att termini would become integrated and removed from the preintegrative pool over time.

The preintegrative DNAs were analyzed by primer extension assays with U5 antisense end-radiolabeled primers (Fig. 2C). The products were separated on polyacrylamide gels, and the ratios of PPT1- and PPT2-primed products were compared (Fig. 8 and Table 3). Note that due to frequent reverse transcription-associated deletion of a repeated enhancer element in U3, use of each PPT is represented on the gel in Fig. 8 as two bands: one resulting from the use of a given PPT and the second resulting from the use of that same PPT accompanied by enhancer deletion. The sum of the deleted and nondeleted products was used to compare PPT usage frequency (Table 3).

TABLE 3.

Relative PPT use^a

Virus	Relative mutant PPT use
Wild-type tandem	1.00
−2C	0.51
−4C	0.29
−7/−17	0.35
−14/−19	0.94
−21/−28rev1	0.13
−21/−28PPT2	0.24

^aQuantified by calculating the ratio of both bands that result from priming using the mutant PPT (mutant PPT plus Δmutant PPT) to total priming (the sum of PPT1, ΔPPT1, PPT2, and ΔPPT2). The mutant PPT is PPT1 for −2C, −4C, −7/−17, and −14/−19; the mutant PPT is PPT2 for −21/−28PPT2. −21/−28rev1 use was quantified by calculating the ratio of PPT1 plus ΔPPT1 to total priming (the sum of PPT1, ΔPPT1, PPTA, ΔPPTA, PPT2, and ΔPPT2). Wild-type PPT in PPT1 was assigned a value of 1, and other values were normalized to this number. Values are the averages of two experiments. 

Under these conditions, products primed by PPT1 were about two and a half times as prevalent as those from PPT2 among DNAs templated by virus with two wild-type PPTs (Fig. 8, lane 2). In contrast, the wild-type PPT2 was selected more frequently than PPT1 when the PPT1 site contained the −2C, −4C, or −7/−17 mutation (lanes 3 to 5). The −14/−19 mutation had little apparent effect on primer selection, and its calculated utilization was similar to that of the wild type in this assay (Fig. 8, lane 6; Table 3). For the tandem −21/−28 PPT mutant, the wild-type PPT was favored over the mutant site to an extent similar to that observed with the conserved PPT point mutations (Fig. 8, lane 8; Table 3). Similar selectivity against the −21/−28 mutation as a priming site was observed with the −21/−28 revertant, which retained the −21/−28 mutation in one PPT (Fig. 5D) but which also contained two wild-type PPTs—PPT2 and PPTA (Fig. 8, lane 7). Note that whereas mutant PPT use was calculated considering the presence of three PPTs for the −21/−28 mutant, low levels of three PPT products were also detected for the wild type, −2C mutant, and −14/−19 mutant by primer extension (Fig. 8, lanes 2, 3, and 6). In contrast, these three PPT products were not detected by PCR (Fig. 7, lanes 4, 5, and 8). This discrepancy may be due to an overrepresentation of short products by PCR and the ability of primer extension to detect rare products. If this three-PPT product is considered for the −14/−19 mutant, then its priming may be somewhat more impaired than calculated here.

The plus-strand initiation site was examined by using primer extension with a U3 antisense primer that allowed resolution of single base length differences. Although the conditions used would not have allowed detection of minor species, the major plus-strand initiation site for all mutants appeared to be the same as for the wild type (not shown).

DISCUSSION

We have examined the effects of altering specific highly conserved positions in the M-MuLV PPT and flanking region on the ability of the virus to replicate and to synthesize vector DNAs. Our results demonstrated that the conserved regions within the PPT contributed to optimal plus-strand priming. However, each of several conserved features (positions −2, −4, and −7/−17) was dispensable individually, thus suggesting a degree of redundancy among plus-strand priming determinants. Alterations to 3′ UTRs adjacent to and upstream of the PPT affected both plus-strand priming and other aspects of replication.

Mutations in the conserved −2 and −4 positions, as well as a mutation (−7/−17) which converted the cleavage site-distal half of the PPT to a series of pyrimidine residues, were compatible with replication. However, replication of these mutants was delayed relative to the wild type, and use of the mutant PPTs was highly disfavored when a wild-type PPT was available in cis.

The −1/−6 mutant, which contained substitutions in the cleavage site-proximal half of the PPT, did not replicate and yielded a titer 1% of that of a wild-type vector. Bowman et al. found that a vector containing a deletion of both the PPT and att site, which should not plus-strand prime, formed less than 2% of the Neo^r resistant titer of a vector containing these sequences (3). Thus the −1/−6 mutant vector titer suggests this PPT was nonfunctional.

Titers for the −7/−17 mutant were variable but generally consistent with the model that the U3-proximal half of the PPT is sufficient for plus-strand priming (31). −7/−17 mutant use in tandem PPT virus was consistent with this notion. However, replication of the −7/−17 mutant was more impaired than its colony-forming titer would have predicted, which may indicate that this mutation affected more than one replication stage.

Our results suggest sequences outside the PPT contribute to plus-strand priming. As revealed in assays of tandem-PPT products, a mutant PPT region that contained the −21/−28 alterations was used significantly less often than a PPT containing the wild-type sequence. These findings may indicate that altering the PPT flanking sequences affected the conformation of the substrate or that contacts made by the DNA polymerase domain are important to PPT recognition, or both. M-MuLV was less tolerant of changes to these sequences than of changes to some highly conserved positions within the PPT, since replication of the −21/−28 mutant was severely delayed and sometimes detectable only after reversion had occurred.

There is experimental precedence for the concept that sequences or structures upstream of RNase H cleavage sites or their interactions with RT may affect retroviral RNA-DNA duplex cleavages. Footprinting analysis reveals that M-MuLV RT protects the regions from −27 to +6 of the template strand (numbered relative to the site of polymerization) and from −26 to −1 of the primer strand when RT is bound in the position to polymerize DNA (54). Additionally, some mutations in RT’s DNA polymerase domain have been reported to affect RNase H activity (12, 14, 24, 30, 36, 39, 50). Hence, contacts made by RT’s DNA polymerase domain as well as the RNase H domain may affect plus-strand priming.

It has been suggested that PPT recognition may involve a unique nucleic acid structure that contributes to the PPT’s resistance to RNase H cleavage (10, 31). Double-stranded DNA typically forms a B-form helix, and double-stranded RNA forms an A-form helix. DNA-RNA hybrids adopt an intermediate conformation called H form (11). H-form helices can be modeled to contact RT’s RNase H domain differently from A or B form helices, and these extra contacts may confer RNase H specificity for RNA-DNA hybrids (11). Circular dichroism studies suggest that polypurine-rich RNA-DNA hybrids have a structure distinct from that of random sequences (31) and that polypurine-rich sequences have wider major and minor grooves than random-sequence RNA-DNA hybrids (10). These structural distinctions might contribute to the RNase H resistance of PPT sequences. However, purine richness per se cannot be responsible for PPT recognition, since Sorge and Hughes have shown that a mere stretch of purines is not sufficient for replication (46) and Luo et al. have demonstrated species specificity in PPT use (25).

Because the −21/−28 replication is more severely impaired than would be expected due to its effects on plus-strand priming, some other step of viral replication may be affected by this mutation. In a study performed contemporaneously with this one, Ilyinskii and Desrosiers found that in SIV, mutations of this A · T-rich region, which they referred to as the U box, caused a block in reverse transcription at a stage between the first strong-stop template switch and plus-strand initiation (21). A similar role for the M-MuLV T-rich region upstream of the PPT might explain the results we find here.

Several of our mutants addressed the effects on priming of purine richness in the 3′ UTR. In contrast to the other mutations, the −14/−19 mutation, which elongates the PPT, did not appear to affect plus-strand priming significantly. This mutation had less effect on PPT selection in tandem-PPT virus than did the other mutations, and the tandem −14/−19 virus replicated more efficiently than the mutant single-PPT virus. These results suggest that the mutation principally affected replication steps other than plus-strand priming and that these defects were overcome to at least some degree by the presence of the ectopic PPT. Because this mutation was able to support plus-strand priming in the tandem-PPT context, the −14/−19 alterations seem unlikely to exert their negative effects on replication through interfering with priming by perturbing the local structure of the PPT region. However, the −14/−19 mutation likely affects some early step in the replication of the virus, since this mutation had a significant effect on CFU titer per unit of encapsidated RNA when incorporated into replication-defective vectors.

For M-MuLV and many other retroviruses, the PPT lies in an untranslated region downstream of the envelope gene. For M-MuLVs, this genome region plays no known roles in replication other than in plus-strand initiation. However, we found that some alterations to this region have defects more severe than would be predicted to occur due to effects on plus-strand priming. These findings suggest that the M-MuLV 3′ UTR may function in replication steps other than plus-strand priming, as has been reported for some other simple retroviruses.

One role that has previously been described for retroviral 3′ UTRs is in the regulation of RNA trafficking. For example, because the main difference between avian leukosis virus and ASV is the presence of the host-derived src gene in its 3′ UTR, it was previously believed that the 3′ UTR might be a functionally dispensable, genetically malleable region (13). However, it is now apparent that DR1 and DR2 in the 3′ UTR of Rous sarcoma virus function in nuclear export of unspliced RNA (28). These unspliced RNAs are used as genomes for new virions and as mRNA for the Gag and Pol proteins. The simian retrovirus and Mason-Pfizer monkey virus contain in this region a cis-acting sequence, named the constitutive transport element, which serves a similar RNA transport function (4, 55).

The mechanism by which M-MuLV unspliced RNAs exit the nucleus is unknown, and the uncharacterized functions of the M-MuLV 3′ UTR— for which we have found evidence—may include unspliced RNA export. However, evidence for impairment of early replication stages by some mutants examined here support the notion that the M-MuLV 3′ UTR may also participate in processes not previously known to rely on this region of retroviral genomes. We are currently exploring these putative replication roles of the M-MuLV 3′ UTR through further mutagenesis, revertant analysis, and complementation studies.