INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The retrovirus Gag polyprotein directs the assembly and budding of virus particles from the plasma membrane of infected cells. Extensive functional mapping of several Gag proteins has led to the identification of three common assembly domains that are required for this activity. The N-terminal membrane-binding (M) domain directs Gag molecules from the cytoplasm to the inner leaflet of the plasma membrane (44, 55, 63) where aggregates of Gag proteins are formed through protein-protein interactions primarily involving the I (interaction) domains (4, 12, 22, 57). These interactions lead to the emergence of spherical particles that pinch off the membrane during the final step in the budding process, which is mediated by the L (late) assembly domain (27, 45, 58). Virus maturation occurs as the Gag polyprotein is cleaved into the major virus structural proteins by the viral protease (PR) to yield the matrix (MA), capsid (CA), and nucleocapsid (NC) proteins as well as several small peptides, and in the case of Rous sarcoma virus (RSV), PR itself. Reorganization of the virion architecture occurs following cleavage of the Gag protein. MA forms a spherical shell under the lipid envelope. At the center of the virion condenses an electron-dense core, a complex of the CA and NC proteins, viral RNA, and the replicative enzymes reverse transcriptase (RT) and integrase.

The Gag polyprotein also coordinates the packaging of Pol, Env, and the genomic RNA. The NC region of Gag is the primary trans-acting factor required for incorporation of the genome into virions (5, 54). Within NC reside two Cys-His boxes and an adjacent basic residue-rich region that interact directly with the viral RNA during encapsidation. Sequences outside of NC have also been implicated to play minor roles in RNA packaging (50, 54). With regard to RSV, sequences in MA are not required for specific genomic RNA incorporation into particles (50) although MA does have weak, nonspecific nucleic acid-binding properties (52). Packaging signals in the RNA itself consist of a cis-acting element known as  within the 5' untranslated region. In RSV, RNA encapsidation signals have been identified within a 270-nucleotide (nt) region between the primer binding site and the splice donor in gag (1) and within a 115-nt region spanning the direct repeats at the 3' untranslated end of the genome (2, 51).

The genomic RNA of all retroviruses exists within the virion as a noncovalently linked dimer of identical 35S RNA molecules. Sequences spanning nt 208 to 274 and 400 to 600 have been implicated in the dimerization. The dimer linkage structure appears to overlap the  packaging signal, a region required for in vivo incorporation of viral RNA (5, 6, 21, 31, 42). Dimerization likely occurs very early in the assembly pathway and is independent of PR activation (24, 25, 53). The NC protein has been shown to promote stability and maturation of dimers, although it is not required for dimerization per se (5). MA has not been previously implicated in genomic RNA dimerization.

Several lines of evidence indicate that MA possesses functions during replication in addition to membrane targeting and binding, although what these functions might be remains uncertain. Genetic analyses of RSV, murine leukemia virus and human immunodeficiency virus (HIV) have revealed many MA mutants that have little or no infectivity, even though particle assembly and budding proceed normally (11, 13, 16, 32, 36, 45, 48, 62). Characterizations of numerous HIV type 1 mutants have suggested a role for MA in accommodating the long cytoplasmic tail of its Env proteins (23, 37) as well as a poorly understood postfusion function (32). In contrast to HIV, it is unlikely that RSV MA has a role in glycoprotein packaging, since Env proteins lacking cytoplasmic tails or having foreign intracytoplasmic domains are packaged efficiently (47, 61). Nonetheless, the identification of numerous assembly-competent, noninfectious MA mutants for several different viruses compelled us to examine our RSV MA mutants more thoroughly.

In this report, we present our unexpected finding that addition of the Src membrane-binding sequence to the N terminus of RSV MA results in particles with altered genomic RNA dimerization. This finding raises the intriguing possibility that the MA sequence influences RNA dimer formation and/or stability.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Viruses and cells. The infectious RSV genome (pRC.V8) used in these studies was derived from pBH.RCAN.HiSV (pRCAN) (18) and bears the RSV Prague C gag gene from pATV8. This proviral vector carries a simian virus 40 early promoter-driven hygromycin resistance gene near the 3' end of the viral genome, within the boundary of the 3' long terminal repeat (15). The JD.Myr2 virus, which produces a myristylated Gag protein due to a E2G substitution, has been described previously (20). Avian cells used for transfection and infection experiments were QT6 cells, a chemically transformed quail fibroblast line (41). The maintenance of these cells was previously described (15). The simian virus 40-based pSV.Myr1 and pSV.Myr2 vectors used for expression of the Gag polyprotein in the COS-1 simian cell line have been described elsewhere (59).

Antisera. A polyclonal rabbit serum against RSV (56) was used for immunoprecipitations and immunoblotting. The anti-Env serum used recognizes the subgroup A envelope glycoprotein (gp85) and was kindly provided by Eric Hunter (University of Alabama, Birmingham).

Oligonucleotide-directed mutagenesis of the gag gene. Mutagenesis was performed by the method of Kunkel et al. (33), using the previously described recombinant bacteriophage MGAG as the template unless otherwise noted (59). The oligonucleotides used to make the indicated gag alleles were as follows: for myr1, 5'-GGATCAAGCATGGAATCCAGCAAAAGC; for myr1e, 5'-CCCGG TG GATCAAGCATG G GATCATCAAAATC TAAACC TAAG GATGAAGCCGTCATAAAGGTGAT; and for myr1e, 5'-GGATCAAGCATGGAATCATCAAAATCTAAAC. M13 DNAs bearing the mutations were identified by dideoxy sequencing. For each mutation, replicative-form DNAs from two independent clones were digested with SstI (nt 255) and HpaI (nt 2731), and the gag fragments were transferred into pRCAN for expression in avian cells. After digestion of the M13 DNA with SstI and Bg1II (nt 1630), each gag allele was cloned into pSV.GagSX for expression in mammalian cells (45).

Transfection, labeling, immunoprecipitation, and production of cell lines. COS-1 cells were transfected by the DEAE-dextran-chloroquine method, metabolically labeled with L-[³⁵S]methionine or [³H]myristic acid, and immunoprecipitated with anti-RSV serum as described previously (56, 60). QT6 cells were transfected by calcium phosphate precipitation and labeled with [³⁵S]methionine, and RSV proteins were immunoprecipitated as described elsewhere (45). Radiolabeled proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), detected by fluorography, and quantitated by laser scanning densitometry.

Virus infectivity assays. The ability of mutant and wild-type viruses to stably infect avian cells was measured in two ways. First, QT6 cells transfected with wild-type or mutant genomes were tested for the continued release of RT activity into the medium upon prolonged passage of the cells in culture. Only infectious genomes spread throughout the culture, integrate, and exhibit persistent virus release. For this, aliquots of culture supernatant were tested for RT activity periodically as described previously (15).

Protein and RNA composition of mutant viruses. Virus particles harvested from culture supernatants of QT6 cell lines were concentrated by ultracentrifugation, normalized according to RT activity, and subjected to immunoblotting with anti-RSV and anti-Env sera as described previously (15). Exogenous RT activity was determined as described previously (15). Viral RNA content of similarly produced particles was determined by Northern slot blot analysis using a ³²P-labeled riboprobe consisting of the gag sequence (15).

ERT assays. Endogenous RT (ERT) reactions were performed as previously described (7, 8), with modifications as described below. Equivalent amounts of virus particles (normalized by RT content) were incubated with melittin (100 µg/ml, final concentration) at 41°C for 10 min prior to addition of ERT buffer (0.1 M Tris-HCl [pH 8.3], 25 mM NaCl, 3 mM magnesium acetate, 30 mM dithiothreitol, 1 mM dCTP, 1 mM dATP, 1 mM dGTP [final concentrations]). The reaction mixtures were divided equally into two tubes. To one set, nonradioactive was added to 1 mM (final concentration), and the reaction mixture was incubated at 41°C for 3 h. The reaction was stopped by addition of an equal volume of buffer containing 1% SDS, 20 mM EDTA, and 100 µg of yeast tRNA. The reaction products were phenol-chloroform extracted and ethanol precipitated. These samples were subjected to PCR analysis for the DNA products of reverse transcription, using the primers and conditions specified below. To the second set of reactions, 10 µCi of [³²P]TTP was added, and the remainder of the reaction was performed as described above. After ethanol precipitation, the radiolabeled ERT products were denatured with 0.05 M NaOH at 37°C for 30 min and electrophoresed in 1% agarose in Tris-borate-EDTA at 100 V for 3 h, and the gel was dried and subjected to autoradiography.

Analysis of products of reverse transcription by PCR. To detect viral DNA products, virus particles were obtained from QT6 cells lines stably expressing viral genomes and used to infect QT6 cells for 18 h. Low-molecular-weight DNA was extracted from duplicate 60-mm-diameter plates of infected cells according to the Hirt protocol (29). DNA was analyzed by using the following pairs of oligonucleotide primers designed to detect the indicated products of reverse transcription: minus-strand strong-stop DNA, primer 5 (R sequence; 5'-CTTCATGCAGGTGCTCGTAGTCG) and primer 3 (U5 sequence; 5'-GCCATTTTACCATTCACCACA); first strand switch, primer 8 (U3 sequence; 5-GGATTGGACGAACCACTGAA) and primer 3; and second-strand switch, primer 5 and primer 7 (5' untranslated region); 5'-CAACGACTCTCTGAGTTCTC). Prior to the reaction, one of the primers was labeled with [-³³P]ATP by using T4 polynucleotide kinase. PCR conditions were 97°C for 5 min and then 25 cycles of 94°C, 56°C, and 72°C for 30 s each. The PCR products were separated by electrophoresis in nondenaturing 6 or 8% polyacrylamide gels and detected by autoradiography.

Electrophoretic analysis of RNA dimerization. Virus particles produced from the stable lines or after transient transfection were collected by ultracentrifugation and resuspended on ice in either phosphate-buffered saline (containing 150 mM NaCl) or TNE buffer (10 mM Tris [pH 7.4], 100 mM NaCl, 1 mM EDTA). Virus particles were lysed in 50 mM Tris (pH 7.5)-10 mM EDTA-1% SDS-100 mM NaCl, with 50 µg of yeast tRNA as a carrier, according to the method of Fu and Rein (25). Equivalent amounts of viral RNA were electrophoresed through a 1% agarose gel in Tris-borate-EDTA (100 V for 3 h or 20 V for 17 h), denatured within the gel, and blotted to a nylon membrane (25). The blots were hybridized by using a gag riboprobe (as described above), washed under high-stringency conditions, and subjected to autoradiography.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

The initial objective of the experiments described below was to ascertain whether the M domain of a cellular protein could replace all of the functions of the RSV MA protein. We chose the myristylated N-terminal peptide of the Src oncoprotein, which is a potent signal for targeting and binding of proteins to the inner leaflet of the plasma membrane (10, 46, 49). When the first 10 residues of Gag are replaced with the first 10 residues of Src, the resulting chimera, named Myr1 (Fig. 1), retains the ability to direct particle assembly and budding from mammalian cells (Fig. 2A, lanes 1) (60). Indeed, the entire M domain of RSV Gag can be replaced with this sequence from Src without affecting budding (59). However, because the nucleotide substitution in myr1 removes the splice donor site (located at codon 7 of gag) which is needed for synthesis of env mRNA, this allele could not be used to study infectivity. For that reason we decided to add the coding sequence for the Src membrane-binding domain as a 5' extension of gag, thus preserving the splice site and creating a Gag chimera named Myr1E (Fig. 1). The Myr1E protein was found to produce particles at the normal rate when tested in mammalian cells (Fig. 2A, lanes 4 and 5), and like the Myr1 protein, it was labeled with [³H]myristic acid because of the presence of the sequence from Src (Fig. 2B).

View larger version (20K): [in this window] [in a new window]   FIG. 1.   Schematic representation of wild-type and mutant MA proteins. The N-terminal region of the RSV Gag protein is shown at the top with the MA, p2, and p10 cleavage products indicated; the numbers below refer to amino acid residues. The M domain, as indicated, consists of the first 85 residues of MA. Expanded below are the N-terminal amino acid sequences of wild-type (WT) and mutant proteins. Residues identical to WT are indicated by dots. Boxes are drawn around residues derived from the Src membrane-binding domain, with the squiggle representing myristic acid modification. Myr2 has an E2G substitution, resulting in myristylation and removal of initiator methionine. In Myr1, the first 10 gag codons were replaced with the sequence encoding the myristylated Src N terminus. Myr1 has the wild-type glutamate substituted at position 2 of Myr1, resulting in the prevention of myristylation and the retention of methionine. Myr1E has nine codons specifying the Src M domain inserted after the gag AUG so that the Src sequence is extended from the Gag N terminus. In Myr1E, myristylation is abolished by a G2E substitution. The ability of each virus to bud and to infect avian cells is indicated by a plus sign (equivalent to wild-type level), minus sign (1% of wild-type level), or NT (not tested).

View larger version (62K): [in this window] [in a new window]   FIG. 2.   Effects of MA alterations on particle assembly and myristylation. (A) Analysis of budding in COS-1 cells. Transfected COS-1 cells were radiolabeled with [35S]methionine for 2.5 h, lysates and media were separated, and viral proteins were detected by immunoprecipitation with anti-RSV serum. Positions of the bands corresponding to Gag precursor, CA, MA, and PR proteins are indicated at the left. untr'f, untransfected. (B) Myristate labeling. Duplicate plates of COS-1 cells were transfected, labeled for 70 min with [35S]methionine or [3H]myristic acid, and immunoprecipitated as for panel A. (C) Budding in avian cells. QT6 cells were transfected with wild-type (WT) and mutant proviral plasmids and analyzed as for panel A. The position of the Gag-Pol protein is indicated at the left, and positions of molecular weight standards are shown at the right.

Placement of the Src peptide as an N-terminal extension potentially preserves the normal M domain of Gag. Evidence in support of this was obtained by eliminating the myristic acid addition sites of Myr1 and Myr1E with a G2E substitution, which inactivates membrane binding (17, 30). This resulted in the creation of Myr1- and Myr1E- (Fig. 1), and as expected, these proteins could not be labeled with [³H]myristic acid (Fig. 2B). The Myr1- protein was unable to direct budding because both M domains were destroyed: the initial insertion of the Src sequence was at the expense of the first 10 amino acids of Gag, which are necessary for activity of the RSV M domain (44); and the Src membrane-binding signal was disabled by the G2E mutation (Fig. 2A, lanes 2 and 3) (3). Therefore, the Myr1 protein was dependent on having a functional Src membrane-binding domain, whereas Myr1E- retained the ability to direct budding even though the inactive form of the Src peptide was present (Fig. 2A, lanes 6 and 7), indicating that the M domain of Gag remains intact. Thus, Myr1E appears to contain two, independent M domains.

To determine whether the budding activities of the Src extension mutants would be the same in avian cells, the myr1e and myr1e alleles of gag were cloned into the pRCAN proviral expression vector (18). Levels of particle production were normal upon transfection of these mutants into QT6 cells (Fig. 2C, lanes 2 and 3) and turkey embryo fibroblasts (data not shown). The rate of budding was indistinguishable for the Myr1E mutant compared with wild-type virus, as shown by pulse-chase analysis following transfection (data not shown).

It was not surprising that the presence of the Src sequence was compatible with budding, since a variety of other peptide sequences have been added as extensions from RSV Gag without affecting budding or infectivity (39; T. Nelle and J. W. Wills, unpublished data). Therefore, we fully expected recombinant RSV genomes carrying the myr1e or myr1e allele of gag to be infectious in avian cells. However, this was not the case.

Infectivity of Myr1E. To analyze infectivity, RCAN vectors that express the myristylated and nonmyristylated forms of Myr1E were transfected into QT6 cells. Proviral plasmid DNAs containing myr2 (the E2G substitution that creates a myristylated form of Gag) or wild-type gag sequences were included as infectious controls. One of two sets of duplicate transfected plates was labeled with [³⁵S]methionine at 18 h posttransfection to see whether the transfections were successful. In each case, Gag proteins were readily detected within cells after a 1-h labeling period (Fig. 3A). To monitor the persistence of infection, cells from the second set of plates were passaged every 3 to 5 days to allow infectious viruses to spread throughout the cultures, and periodically the cells were tested for continued production of Gag proteins by radiolabeling and immunoprecipitation. Although abundant Myr1E particles had been released immediately after transfection, no evidence of infection was found after six passages (approximately 30 days), and the Myr1E virus had disappeared from the culture (Fig. 3A). Numerous transfection experiments of this type have been repeated by different investigators in our laboratory, using multiple clones of the myr1e allele expressed in the Prague A and RCAN genomes in both QT6 cells and turkey embryo fibroblasts, and persistent infection has never been observed. On the other hand, the Myr1E and Myr2 mutants spread in QT6 cultures at the same rate as the wild-type virus and readily establish infection (Fig. 3A). The precise magnitude of the defect in Myr1E virus has not been determined; however, our previous experience with these assays indicates that the defect must be on the order of 10⁵ or greater relative to the wild type.

View larger version (22K): [in this window] [in a new window]   FIG. 3.   Persistence of virus production in long-term cultures. (A) Duplicate plates of QT6 cells were transfected with proviral constructs expressing the indicated gag alleles. At 18 h posttransfection, the Gag precursor in cell lysates was detected after a 1-h labeling period by immunoprecipitation with RSV antiserum (transient expression). Persistent expression of Gag was measured similarly after passage 6. non-inf., noninfectious CA mutant L171I, included as a negative control; untr'f, untransfected. (B) In two separate experiments, infections of QT6 cells were initiated by transfection as for panel A; then the accumulation of RT activity in the culture medium was monitored over a period of 30 or 33 days posttransfection. The large difference between the two experiments in the scale on the RT activity axis was due to differences in specific activity of the isotope and in the particle concentration factor.

Biochemical analysis of Myr1E virus particles. To determine whether the replication defect of Myr1E was due to a failure to package required viral components, virus particles were examined for incorporation of Env glycoproteins, genomic RNA, and Gag and Gag-Pol proteins. With regard to Env, it was possible that extension of the Src sequence from the N terminus of Gag might interfer with its synthesis, transport, or incorporation into the viral envelope because the splice donor for env mRNA resides within the gag gene in RSV. Therefore, the first few amino acids of Gag are initially attached to the nascent Env protein. If the plasma membrane-binding signal of Src was dominant over the Env signal peptide, then targeting of Env to the endoplasmic reticulum would not occur and little or no glycoprotein would be produced and packaged in the Myr1E virions. This is not the case, however, because Myr1E particles produced by transfection of QT6 cells contained abundant Env (gp85) as do wild-type and Myr1E viruses (Fig. 4A).

View larger version (32K): [in this window] [in a new window]   FIG. 4.   Biochemical properties of virus particles. (A) Envelope incorporation. Equivalent amounts of virions from culture supernatants were collected by ultracentrifugation from QT6 cells transfected with mutant or wild-type (WT) proviral plasmids. Levels of Env (SU) expression were detected by immunoblotting with anti-Env serum. untr'f, untransfected. (B) Gag processing. Virus particles from panel A were analyzed by immunoblotting with anti-RSV serum to detect processed Gag proteins CA and MA. (C) RNA genome incorporation. Virus particles were normalized for RT content, RNA was purified, serial fourfold dilutions were performed, and the RNA was applied to a slot blot apparatus, bound to nitrocellulose, and detected with a radiolabled gag riboprobe.

Postinfection levels of reverse transcription. Wild-type or Myr1E virus particles produced by cell lines stably expressing each proviral genome were used to infect monolayers of QT6 cells. Eighteen hours later, low-molecular-weight DNA was isolated from infected cells (29) and used as a template for PCR amplification with primers designed to detect the products of reverse transcription (Materials and Methods). As shown in Fig. 5A, the level of minus-strand, strong-stop DNA was markedly reduced in Myr1E-infected cells. To determine the magnitude of the defect, the low-molecular-weight DNA was diluted serially from 10¹ to 10³ and subjected to PCR analyses. The amount of strong-stop DNA isolated from Myr1E-infected cells was reproducibly 100- to 1,000-fold reduced compared with wild-type infection (Fig. 5B). DNA products corresponding to later steps of reverse transcription (first- and second-strand switches) were undetectable after Myr1E infection (data not shown). These findings led us to investigate whether there was an intrinsic defect in reverse transcription attributable to the mutant virus particles.

View larger version (38K): [in this window] [in a new window]   FIG. 5.   Detection of viral DNA after infection. (A) Virus particles produced from stably expressing cell lines were normalized, concentrated, and used to infect QT6 cells. Low-molecular-weight DNA was isolated and used as a template in PCRs with primers specific for minus-strand strong-stop DNA as described in Materials and Methods. In both panels, molecular size markers were run in the right lane (ladder). The PCR product amplified from minus-strand strong-stop DNA is predicted to be 86 bp in length with our primers. cl., clone. (B) Quantitative analysis of strong-stop DNA. Low-molecular-weight DNA isolated as described for panel A was serially diluted as indicated and amplified as described above. Virus particles used for infection were derived from independently derived cell lines (clones 1 and 2) stably expressing the myr1e genome.

ERT activity in Myr1E virus particles. To address the possibility that the Myr1E mutation had disrupted a structure within the virion needed for reverse transcription, ERT assays were performed. Virus particles from two independent cells lines stably expressing the myr1e genome or cells chronically infected with wild-type virus were permeabilized with melittin and incubated with deoxyribonucleotides. Viral DNA was isolated from the particles and analyzed by PCR using primers to detect the earliest products in the reverse transcription reaction, the minus-strand strong-stop DNA. As shown in Fig. 6A, approximately equivalent levels of strong-stop DNA were synthesized in virus particles from both clones of Myr1E and the wild-type virus. To allow more quantitative comparisons among samples, limiting serial dilutions of the viral DNAs from the ERT reactions were subjected to PCR with primers that amplify minus-strand strong-stop DNA and products of the first- and second-strand transfer reactions. In each step, similar amounts of viral DNA were detected for the mutants and wild type (Fig. 6B). When the same population of virus particles was analyzed for ERT activity using radioactively labeled nucleotides, a comparable result was obtained (data not shown).

View larger version (35K): [in this window] [in a new window]   FIG. 6.   ERT activity. Permeabilized visions were incubated with nucleotides, and viral DNA was isolated and subjected to PCR analysis as for Fig. 5. (A) Intravirion minus-strand strong-stop DNA for particles isolated from stable cell lines. Molecular size markers were run in the right lane (ladder). (B) Serial 10-fold dilutions of ERT products to amplify sequences corresponding to minus-strand strong-stop (86-bp), first-strand switch (117-bp), and second-strand switch (300-bp) DNA products.

Analysis of genomic RNA secondary structure. To determine whether the Myr1E genome was normal with respect to the formation of dimers and the stability of the dimer complex, RNA was isolated from equivalent amounts of particles produced by cells transiently transfected with mutant or wild-type proviral DNA (25). The RNA was electrophoresed under nondenaturing conditions prior to Northern analysis using a radioactively labeled gag riboprobe. Under these electrophoretic conditions, the majority of the wild-type genomic RNA migrates as a high-molecular-weight complex containing noncovalently bound dimers (Fig. 7A). Strikingly, viral RNA purified from Myr1E particles was almost entirely monomeric (Fig. 7A). Very little, if any, RNA was seen at the position expected for dimeric RNA. Virus particles isolated from cell lines stably expressing mutant or wild-type virus were also examined for genomic RNA content (Fig. 7B). Again, the wild-type viral RNA was dimeric, and genomic RNAs from both Myr1E cell lines appeared as monomers (Fig. 7B, right). In contrast, Myr1E particles contained RNA in dimeric form that migrated at the same position as wild-type dimers (Fig. 7B, left). This is the expected result given that Myr1E has normal infectivity, and dimerization is believed to be critical for infectivity. In some experiments there is dramatically less total RNA visible in the Myr1E lanes of nondenaturing Northern blots compared to wild-type virus even though slot blot analysis reveals only a slight difference, suggesting that the Myr1E monomeric RNA is more susceptible to degradation during extraction than is the wild-type RNA.

View larger version (44K): [in this window] [in a new window]   FIG. 7.   Nondenaturing Northern blot analysis of viral RNA. Genomic RNA was isolated from virus particles produced by transient transfection of proviral plasmids (A) or from two independently arising stably producing cell lines (clones [cl.] 1 and 2) (B). The positions of dimers (D) and monomers (M) are indicated by arrows.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In this report, we have characterized a mutant, called Myr1E, which efficiently produces virus particles that are defective in viral DNA synthesis following infection despite being normal in reverse transcription measured in an ERT assay. The lack of dimers in the RNA isolated from Myr1E particles raises interesting questions about the possible role of the MA sequence in viral RNA dimerization, the requirement for dimerization prior to RNA packaging, and the ability of RT to transcribe monomeric RNA.

From electron microscopy images, the point of RNA dimer linkage in the RSV genome has been described as an approximately 50-nt sequence positioned around nt 511 as reported by one group (35) or around nt 466 (42). Other dimer-promoting sequences have been identified at nt 208 to 270 and 400 to 600 (6, 21, 35). An autocomplementary sequence at nt 258 to 274 identified by in vitro studies appears to contribute to dimerization of avian leukosis virus RNA (21), although the importance for this sequence in vivo has not been ascertained. The insertion mutation that gives rise to Myr1E begins immediately upstream of the normal gag initiation site, which is at nt 380. Although the insertion lies between the regions known to be important for dimer formation in vivo, it does not directly overlap any of these sequences. In addition, there are two other reasons why it seems unlikely that the interference with dimerization resides entirely at the level of RNA structure. First, the myristate-lacking Myr1E makes dimeric RNA and is fully infectious yet differs from noninfectious Myr1E at only one nucleotide (codon 2, GGA to GAA). It is unlikely that such a minor change would so severely interfere with dimerization. Second, we have found other, unrelated RSV MA mutants that also have altered dimerization (to be presented elsewhere). However, it remains of interest to determine whether the single nucleotide difference between Myr1E and Myr1E could indeed have a cis effect on dimerization, because this region has not been previously implicated in contributing to the dimer linkage structure.

If it is not the RNA sequence of myr1e that disrupts dimer formation, then the MA protein itself might influence dimerization in trans. The strikingly different phenotypes of Myr1E and Myr1E provide insight into how MA might be involved. The two mutant proteins differ only at the second residue (Glu-to-Gly), but this difference allows myristylation of the Myr1E protein. Because the myristylated Src sequence is known to be a strong membrane-binding signal, the Myr1E MA protein might be tightly attached to the viral membrane, making MA unavailable for involvement in RNA dimerization after budding and virus maturation. Further support for this hypothesis is provided by earlier reports that the RSV MA protein has nucleic acid-binding properties (52). Although it has not been shown to specifically bind viral RNA with high affinity, within the confines of a virus particle, the concentration of MA may be high enough to allow MA to play a role in dimerization. The tighter-than-normal membrane-binding hypothesis also suggests that MA needs to be in close proximity to the genomic RNA for it to have a direct effect on dimer formation; however, for RSV it is unknown whether any MA molecules actually associate with the viral core. Perhaps just a few MA molecules might be required to influence dimerization of the RNA because the area of dimer linkage is only a small region of the genome. Small amounts of MA protein could be very difficult to detect within the ribonucleoprotein complex in the interior of the virion.

Our findings in no way suggest a role for the RSV MA domain in viral RNA packaging (i.e., the specific selection of viral RNA for incorporation into virions). Earlier studies have shown that MA is not needed for RNA incorporation and that deleting the N-terminal half of MA results in a minimal reduction in the amount of genomic RNA incorporated into virus-like particles (less than a 25% decrease compared with Myr1) (50). However, dimerization of the RNA was not examined in that study.

One interesting interpretation of our data is that Myr1E packages a lower amount of RNA because it incorporates monomers rather than dimers. If so, then the processes of dimerization and RNA packaging would have to be separable functions. We hasten to point out that our data do not rule out the possibility that dimers are initially packaged into particles but are unstable due to failure of the dimers to mature into stable structures. This has been proposed as the explanation for rapidly harvested virus, NC mutants, and protease-deficient viruses that contain dimeric RNA with increased electrophoretic mobility and decreased thermostability (9, 19, 24-26, 40). However, in contrast to those studies, we did not detect any dimers for Myr1E under standard low-temperature RNA extraction conditions identical to those used by others (25).

If indeed only monomers are present in Myr1E particles, then the efficient ERT activity suggests that dimerization might not be required for reverse transcription. However, we cannot rule out that the RNA within Myr1E particles is dimeric under the gentle permeabilizing conditions of the ERT reaction, even though only monomers are seen following extraction of the viral RNA. Likewise, it is puzzling that reverse transcription is so severely impaired after infection. These findings suggests two main alternative conclusions: (i) the Myr1E viral RNA is able to be reverse transcribed in the protected milieu of the virion but not in the environment of the cell, perhaps due to instability of the dimeric structure; or (ii) the problem with dimer formation has nothing to do with the inability of Myr1E to perform reverse transcription after infection; rather, the Myr1E MA protein might be unable to participate in a different step early in infection. We hope to be able to distinguish between these possibilities by examining other noninfectious mutants involving the MA sequence for dimer formation and postentry RT activity.

The interpretations above have implied an active role for MA at a step after budding; however, it is important to point out an alternative and equally plausible model. It could be that the wild-type MA protein is not directly involved in dimer formation or stability or in the initiation of proviral synthesis after infection, but rather the altered and possibly misshapen Myr1E MA protein interferes with these functions, which are performed by other viral factors. A defect in genomic RNA dimerization in Myr1E does not necessarily mean that the normal MA protein provides this function in the wild-type virus.

On the other hand, we do find it compelling that there is remarkable similarity in the three-dimensional structures of all retroviral MA proteins, each having four major alpha helices separated by flexible loops, with the first two helices overlapping the second two (14, 28, 38, 39). The structures of these evolutionarily diverse MA proteins are conserved even though the mechanisms that they use for membrane binding are not (e.g., the requirements for myristate and the clustering of basic residues differ for RSV and HIV) (55, 63). This observation suggests that there are other shared and essential functions among MA proteins of different viruses. The conserved MA structure might have a role during another step in replication, perhaps one involved in RNA dimerization or efficient proviral DNA synthesis.