INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The assembly and budding of Rous sarcoma virus (RSV), an avian retrovirus, are driven by its Gag polyprotein (Fig. 1) (for a review, see reference 28). Gag is synthesized on free ribosomes in the cytoplasm and then is transported to the plasma membrane. The membrane-binding (M) domain located at its N terminus is responsible for the specific targeting of the protein to the site of budding (19, 30), and approximately 1,500 Gag molecules interact to create the emerging particle (31). Although it is unclear when the interactions among Gag proteins first begin, the functions most important for the tight packing of the molecules are those of the interaction (I) domains located near the C terminus. These are thought to promote assembly by binding to RNA, and in their absence, only particles of low density are produced. These particles are also large and heterogeneous in size; however, I domains themselves are not sufficient for controlling particle dimensions. The primary size determinant of RSV is the capsid (CA)-spacer peptide (SP) sequence located in the central region of Gag. Deletions throughout this sequence result in the release of particles of normal density but large and heterogeneous size (13). The interactions mediated by the CA-SP sequence must be rather weak because they are insufficient for producing particles of normal size and density in the absence of I domains. Moreover, the M domain, I domains, and size determinant are incapable of mediating particle release ("pinching off"). This virus-cell separation step is thought to require host machinery recruited to the site of budding by the late (L) domain. During or shortly after particle release, the viral protease (PR) cleaves the Gag molecules into the mature products (matrix [MA], p2a, p2b, p10, CA, SP, nucleocapsid [NC], and PR), but this proteolytic activity is not required for budding.

View larger version (40K): [in this window] [in a new window]   FIG. 1.   RSV Gag chimeras. The wild-type RSV Gag protein (top, open boxes) was linked to the CA proteins of SV40 (cross-hatched boxes) and STMV (black boxes). Sites cleaved by the retroviral PR during release of the mature Gag products (MA, p2a, p2b, p10, CA, SP, NC, and PR) are indicated, and the associated numbers refer to amino acid residues, counting from the N terminus. The locations of assembly domains required for budding (M, L, and I) are indicated above the Gag protein, along with the size determinant (the CA-SP sequence). In the Gag-VP1 and Gag-STMV chimeras, the large M domain was replaced by MSrc, which is myristylated (squiggly line). Relevant restriction endonuclease sites used in the cloning procedures are indicated at their positions relative to the DNA. Sites in parentheses were destroyed as a result of the cloning. The letters at the ends of three of the chimeras indicate foreign amino acid residues introduced by the cloning methods.

The assembly of RSV particles does not absolutely require membranes, cells, or the functions of the M and L domains. Elegant in vitro studies have shown that the CA-SP-NC subsequence of Gag (which contains the size determinant and the I domains) can assemble into hollow tubes if RNA is present (6). When residues from the adjacent p10 sequence are also included, spherical particles of the proper size and morphology are obtained (7). These observations led us to consider the idea that the functions of the M and L domains are "add-ons" to the basic building block of the retrovirion. If this is the case, then it may be possible to replace the assembly functions of CA-SP-NC with capsid proteins from nonenveloped viruses. That is, it may be possible to insert capsid proteins from nonenveloped viruses into the RSV budding pathway by attaching the M and L domains to their amino termini. Our results, obtained from studies of the capsid proteins of simian virus 40 (SV40) and satellite tobacco mosaic virus (STMV), demonstrate that this is indeed the case. Moreover, the foreign capsid proteins were not mere passengers in the RSV budding pathway but provided interactions sufficient to constrain the size of the chimeric particles.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Previously described plasmid constructs and cells. The RSV gag gene was obtained from pATV-8, an infectious molecular clone of the RSV Prague C genome (25). The SV40 vp1 gene was acquired from the early SV40 deletion mutant plasmid dl1055 (37). The molecular clone of STMV, contained in a pBS plasmid, was a generous gift from J. A. Dodds (15, 16). All of the gag alleles were expressed in simian (COS-1) cells using a previously described SV40-based mammalian expression vector (35). A few of the gag alleles used in this study have been described elsewhere: pSV.Myr0 (wild type) (36), pSV.Myr1 (36), pSV.Myr1.3 h (36), pSV.Myr1.RNot (34), pSV.Myr1.MA6 (19), pSV.Myr1NC (39), pSV.Myr1.R3J (34), and Myr1() (4). In some cases, the activity of the retroviral PR was eliminated by substituting the aspartic acid in the active site with either isoleucine (D37I) (9, 26, 35) or serine (D37S) (9), changes which have no effect on particle release, density, or size. All plasmids were propagated in Escherichia coli DH-1 cells and selected using medium containing 100 µg of ampicillin per ml. COS-1 cells were grown in Dulbecco's modified Eagle medium (GIBCO BRL) supplemented with 3% fetal bovine serum and 7% bovine calf serum (HyClone, Inc.).

Construction of Gag-VP1 and Gag-STMV chimeras. Myr1.VP1 (Fig. 1) was constructed by PCR using plasmid dl1055 as the template for the vp1 gene. Prior to PCR, dl1055 was digested with BamHI and ligated at a concentration of 10 µg/ml to remove the bacterial plasmid sequence. For PCR, the following upstream and downstream primers were used (the underlined sequence in each oligonucleotide corresponds to the particular restriction endonuclease site [in parentheses] used for cloning): 5'-TCTAAAAGCGGCCGCAGATGGCCCCAACAAAAAGA-3' (NotI) and 5'-AAAGCATAGATCTGACTGCATTCTAGTTGTGGTTT-3' (BglII).

Transfection of cells. COS-1 cells in 35- or 60-mm-diameter plates were transfected by the DEAE-dextran-chloroquine method as described previously (35). Before transfection, the plasmid DNAs were digested with XbaI and ligated at a concentration of 25 µg/ml. This step removes the bacterial plasmid sequence and joins the 3' end of the gag gene with the SV40 late polyadenylation signal for high-level expression (35). Typically, 1 µg of DNA was applied to each monolayer for 35-mm-diameter plates; 2 µg was used for 60-mm-diameter plates.

Metabolic labeling and immunoprecipitation. Cells were analyzed 48 h after transfection. To measure particle release, our standard method is to starve cells for 0.5 h in methionine-free, serum-free Dulbecco's medium and then metabolically label them with L-[³⁵S]methionine (50 µCi, >1,000 Ci/mmol) for 2.5 h as previously described (2, 9, 34-36). For sucrose gradient analysis, cells were starved for 0.5 to 1.0 h and then labeled in 0.6 ml (for rate-zonal gradients) or 1.0 ml (for isopycnic gradients) of medium for 2.5 to 8 h. In some instances, the cells were treated during the starvation and labeling period with the calcium ionophore A23187 (Sigma) at a concentration of 5 µM. The cells and growth medium from each labeled culture were mixed with lysis buffer containing protease inhibitors. The Gag proteins in each fraction were immunoprecipitated with a rabbit antibody against RSV (reactive with MA, CA, NC, and PR). Rabbit antiserum to SV40 (kindly provided by M. J. Tevethia [29]) or rabbit antiserum to VP1 (kindly provided by R. L. Garcea [17]) was also used in some experiments. The proteins were then electrophoresed in a sodium dodecyl sulfate (SDS)-12% polyacrylamide gel and detected by fluorography.

Sucrose gradient analysis. After the labeling period, the medium from each plate was collected and transferred to a microcentrifuge tube, and cellular debris was removed by centrifugation at 15,000 × g for 1 min. For the rate-zonal and isopycnic gradients, labeled particles of wild-type density and size were mixed with the samples to provide an internal control. To discriminate the protein species in this mixture of particles during subsequent gel analyses, the control either retained an active PR (i.e., Myr0) or had an inactive PR (i.e., Myr1.D37S or Myr1.3 h), as appropriate. The mixture was then layered onto 10 to 30% (for rate-zonal analysis) or 10 to 50% (for isopycnic analysis) sucrose gradients and centrifuged at 83,500 × g (26,000 rpm) and 4°C for 30 min (for rate-zonal gradients) or 16 h (for isopycnic gradients) in a Beckman SW41Ti rotor. Fractions (0.6 ml) were collected through the bottom of each tube and subsequently immunoprecipitated and processed for SDS-polyacrylamide gel electrophoresis analyses as described above. The resulting films were quantitated by laser densitometry. All gradient runs were repeated at least once to ensure consistent results.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

The CA-SP sequence controls the size of RSV particles in cooperation with the I domains in the NC sequence (13). This prompted us to test the possibility that foreign capsid proteins can substitute for CA-SP-NC. We initially chose the major SV40 capsid protein, VP1, for several reasons. First, it contains a quantity of amino acids similar to that of CA-SP-NC (338 for CA-SP-NC versus 361 for VP1). Second, when expressed in the absence of all the other SV40 gene products, it self-assembles into icosahedral capsids within the nucleus of the cell, and when calcium ionophores are present, it also assembles in the cytoplasm (17), where retroviral budding takes place. Third, the particles produced by VP1 are similar in size to the spherical particles produced in vitro by MA-PR, a slightly longer form of CA-SP-NC that also contains the N-terminal p2 and p10 peptides of RSV Gag (40 nm for VP1 versus 50 nm for MA-PR) (7).

We began by constructing two Gag-VP1 chimeras (Fig. 1). Myr1.VP1 has VP1 in place of the last few residues of p10 and most of CA but retains the last 25% of CA as well as SP, NC, and PR. In Myr1.VP1t, the VP1 sequence is linked to the same N-terminal portion of Gag, but the C-terminal Gag sequence is absent. Both chimeras utilize M^Src, which is functionally equivalent to the M domain of RSV Gag (36), and both contain the L domain of RSV, which is required for the virus-cell separation step late in budding. However, Myr1.VP1t lacks both of the I domains required for the production of particles of normal density (2, 5).

Release of Gag-VP1 chimeras into the medium. The Gag-VP1 chimeras were expressed in COS-1 cells by an expression system that has been used in many previous studies of RSV Gag. Expression of the wild-type Gag protein (referred to as Myr0; Fig. 1) results in the release of virus-like particles into the growth medium (Fig. 2A, left panels), and these have been shown to be identical to authentic RSV (and the Src chimera Myr1) in terms of their rate of release, core morphology, size, density, and proteolytic maturation (2, 3, 10, 32, 35, 36). When the viral PR is inactivated by a single amino acid substitution (Myr0.D37I), budding continues at the normal rate, but Gag cleavage products are not released into the medium (Fig. 2A, right panels).

View larger version (74K): [in this window] [in a new window]   FIG. 2.   Release of Gag-VP1 chimeras within virus-like particles. COS-1 cells were transfected with the indicated DNAs; 48 h later, the cells were labeled with [35S]methionine for 2.5 h. (A) Three clones of Myr1.VP1 (with PR) and three clones of Myr1.VP1t (without PR) were analyzed for their ability to be released from transfected cells. Viral proteins in detergent lysates of the cells and growth medium were collected by immunoprecipitation with anti-RSV serum, separated by electrophoresis through an SDS-12% polyacrylamide gel, and visualized by fluorography. The arrowhead on the left indicates the predicted position of the p10-VP1-CA fusion protein released by the action of PR. The position of the uncleaved Myr1.VP1t chimera is indicated by the arrowhead on the right. Positions of molecular weight standards (in thousands) are indicated in the middle, and the positions of the full-length Gag protein (Pr76) and its detectable cleavage products are indicated on the left. (B) Trypsin resistance of the Gag-VP1 chimeras. COS-1 cells transfected with the indicated DNAs were labeled as described above. The growth medium was removed and divided into six equal portions that were treated for 30 min with trypsin, detergent (Triton X-100 [TX-100]), or trypsin inhibitor, as indicated above the lanes. Subsequently, trypsin inhibitor was added to each of the samples that did not receive any previously, and the surviving Gag proteins were immunoprecipitated with anti-RSV serum and analyzed by SDS-polyacrylamide gel electrophoresis. Only the portions of the gel containing the CA and p23 bands (Myr0 and SPG) and the VP1 fusion (Myr1.VP1 and Myr1.VP1t) are shown.

View larger version (62K): [in this window] [in a new window]   FIG. 3.   Requirement of the M and L domains for the release of VP1 chimeras. Viral proteins from cell lysates and growth medium were analyzed as described in the legend to Fig. 2. The positions of the full-length Gag protein (Pr76) and its detectable cleavage products are indicated on the left, while the VP1-containing fusion proteins are indicated by the arrowheads on the right of each panel. (A) Analysis of M domain mutants. (B) Analysis of an L domain mutant. , antiserum.

Mechanism of release for Gag-VP1 chimeras. It has been reported that SV40 particles can be released from polarized cells by a mechanism that (like budding) does not involve cell lysis (8). Therefore, it was of interest to examine the pathway utilized by the Gag-VP1 chimeras. Several approaches were used, and all of the results argue strongly that the retroviral budding pathway is utilized by the chimeras.

(i) Trypsin resistance. If the chimeras travel the retroviral budding pathway to reach the growth medium, then they should be surrounded by a lipid bilayer, which would protect them from digestion with exogenously added PRs. Analyses of the chimeras revealed that both are released in a trypsin-resistant form (Fig. 2B), including the MA and PR species produced by Myr1.VP1 (data not shown). (The partial sensitivity of Myr1.VP1t in this particular experiment was not reproducible.). Trypsin resistance was lost when detergent was added, consistent with the removal of protective membranes. These characteristics are identical to those of wild-type Gag (Myr0; Fig. 2B) (14, 32, 35) but are unlike those of Gag proteins with a signal peptide (SPG; Fig. 2B) (14), which travel the secretory pathway and are released into the medium in a trypsin-sensitive, soluble form. These results suggest that the chimeras are membrane enclosed.

(ii) Necessity of the M and L domains for Gag-VP1 release. Retrovirus budding requires an M domain for plasma membrane targeting and an L domain for particle release. If the VP1 chimeras travel the budding pathway, then the removal of either domain should abolish their release into the medium. To inactivate M^Src, we used a previously described mutant, Myr1(), which fails to bud (Fig. 3A) because the site of myristylation has been destroyed (4). This change was introduced into both of the Gag-VP1 chimeras to create Myr1().VP1 and Myr1().VP1t (Fig. 1), and the resulting myristate-lacking forms were found to be incapable of particle release despite good levels of expression in the cells (Fig. 3A); hence, the M domain is required.

(iii) Subcellular localization. VP1 contains nuclear targeting information (12, 38) which enables it to be efficiently transported to the nucleus, where it is rapidly assembled into virus-like particles (17). If the M and L domains of Gag efficiently direct VP1 into the budding pathway, then the chimeras should not be found in the nucleus. Consistent with this notion, cell fractionation and immunofluorescence experiments using various sera (antisera to RSV, SV40, and VP1) failed to reveal the presence of the chimeras in the nucleus, although T antigen (which is constitutively expressed in COS-1 cells) was readily detectable in the nucleus in both assays (data not shown). Moreover, the chimeras remained cytoplasmic even when the M domain was inactivated by eliminating the site of myristylation (data not shown). It is possible that the nuclear targeting information of VP1, which resides near the N terminus (18), is masked by fusion to the p10 sequence. In any case, it is clear that the attached Gag sequences are dominant over the nuclear targeting signals of VP1, enabling this capsid protein to bud from the plasma membrane of the cell.

VP1 can substitute for retroviral I domains. The I domains within the RSV Gag protein (Fig. 1) provide the major regions of interaction and enable the tight packing of molecules needed for the efficient release of particles of normal density (2, 5, 33). When the I domains are absent, fewer particles are produced and those that are released have a lower density, as measured in isopycnic sucrose gradients (e.g., mutant NC; Fig. 4A). Deletions elsewhere in Gag have little or no effect on particle density (13). Because Myr1.VP1 contains the RSV I domains, we expected the particles produced by that chimera to have the same density as control particles, and this was found to be the case (Fig. 4B). However, it was unclear what to expect for Myr1.VP1t. This chimera buds with an efficiency that is reduced only twofold relative to that of RSV Gag (Fig. 2A; compare with Myr0.D37I), even though it lacks both of the RSV I domains and all of the CA-SP sequence (Fig. 1). Analysis of Myr1.VP1t particles in density gradients revealed that they have a density that is indistinguishable from that of particles produced by wild-type Gag (Fig. 4C), a result that strongly suggests that interactions provided by VP1 can substitute for those of the I domains. This hypothesis is further supported by a deletion mutant that lacks the C-terminal 109 amino acids of VP1 (Myr1.VP1ABt; Fig. 1) and that was found to have reduced particle density (Fig. 4D). To what extent the assembly of Myr1.VP1t (or Myr1.VP1) bears resemblance to that of authentic SV40 will require further investigation, but the following experiment shows that there is some similarity.

View larger version (20K): [in this window] [in a new window]   FIG. 4.   Particle density of the Gag-VP1 chimeras. COS-1 cells transfected with the indicated Gag derivatives were labeled with [35S]methionine for 8 h. The medium from each plate was collected and mixed with labeled control particles of normal density. The mixtures were then layered onto 10 to 50% isopycnic sucrose gradients and centrifuged for 16 h. Fractions were collected, immunoprecipitated with anti-RSV serum, electrophoresed in an SDS-12% polyacrylamide gel, and detected by fluorography. Arrows indicate the direction of sedimentation. (A) Myr1.NC, an RSV mutant that lacks I domains. (B) Myr1.VP1, containing I domains. (C) Myr1.VP1t, lacking I domains. (D) Myr1.VP1ABt, lacking I domains and a C-terminal segment of VP1.

Gag-VP1 chimeras respond to calcium ionophores. To ascertain whether VP1 can substitute for the size determinant of RSV, we used rate-zonal sedimentation analysis. Retroviral particles move quickly through the gradient as a well-defined band, and these were used as an internal control within each gradient (Fig. 5). As previously reported (13), mutants of RSV that have deletions within the CA-SP sequence make particles that are very large and heterogeneous (e.g., mutant Myr1.R3J; Fig. 5A, left panel). Similar heterogeneity was also observed with both of the Gag-VP1 chimeras (Fig. 5B and C, left panels). However, it is well documented that the assembly of VP1 requires calcium (11, 22, 23) and that particles are not found in the cytoplasm (where retroviral budding is initiated) unless ionophores are used to increase the levels of calcium in this compartment (17). Therefore, we repeated the sedimentation analysis using particles obtained from cells that were treated with the calcium ionophore A23187. Although the CA-SP mutant did not respond to A23187 (Fig. 5A, right panel), both Gag-VP1 chimeras did and, as a result, the particles became much more uniform in size (Fig. 5B and C, right panels). Moreover, the chimeric particles were found to sediment more slowly than the internal control particles; the peaks were shifted two (Myr1.VP1) and four (Myr1.VP1t) fractions higher in the gradient, indicative of a considerable change in particle size. The smaller size of Myr1.VP1t particles probably was due to the absence of the mass contributed by the long C-terminal Gag sequence and not to the presence of the short foreign peptide (Fig. 1) because an identical shift in sedimentation rate was seen when this peptide was removed (data not shown). In contrast, the particles from Myr1.VP1ABt did not appear to respond significantly to increased levels of calcium (Fig. 5D). The inability of this chimera to produce uniformly sized particles could be explained in part by its limited ability to produce particles of the proper density (Fig. 4D). Collectively, these results suggest that VP1 has some assembly capabilities when inserted into the budding pathway.

View larger version (34K): [in this window] [in a new window]   FIG. 5.   Calcium ionophores change the size of Gag-VP1 particles. COS-1 cells expressing the indicated Gag derivatives were labeled with [35S]methionine in the presence or absence of the calcium ionophore A23187. The medium from each plate was collected and mixed with labeled control particles of normal size. The mixtures were then layered onto 10 to 30% rate-zonal sucrose gradients and centrifuged for only 30 min. Fractions were collected, immunoprecipitated with anti-RSV serum, electrophoresed in an SDS-12% polyacrylamide gel, and detected by fluorography. Arrows indicate the direction of sedimentation. (A) Myr1.R3J.D37S, an RSV mutant with a defective size determinant. (B) Myr1.VP1, a chimera that lacks the RSV size determinant but has I domains. (C) Myr1.VP1t, a chimera that lacks the RSV size determinant and I domains. (D) Myr1.VP1ABt, a chimera that lacks I domains and a C-terminal segment of VP1.

Gag-STMV chimeras. Having found that VP1 can substitute for the assembly functions contained within CA-SP-NC, we next constructed Gag-STMV chimeras to see whether our findings could be extended to a capsid protein from another nonenveloped virus. STMV encapsidates a single-stranded RNA genome and requires coinfection with tobacco mosaic virus for propagation (1, 15, 16). This plant virus is much smaller (only 17 nm) than SV40 and RSV, and its coat protein is half the size of CA-SP-NC (159 versus 338 amino acids). It does not require calcium for assembly, as would be predicted for a virus that replicates in the cytoplasm.

View larger version (60K): [in this window] [in a new window]   FIG. 6.   Release of Gag-STMV chimeras into the growth medium. COS-1 cells were transfected with the indicated DNAs and analyzed as described in the legend to Fig. 2. The positions of the full-length Gag protein (Pr76) and its detectable cleavage products are indicated on the left. The arrowheads on the right indicate the STMV-containing fusion proteins.

View larger version (34K): [in this window] [in a new window]   FIG. 7.   Analysis of particle density and size of the Gag-STMV chimeras. Particles released from COS-1 cells expressing the indicated Gag-STMV chimeras were mixed with control retroviral particles and analyzed for size (A and B) or density (C and D) in sucrose gradients. Arrows indicate the direction of sedimentation. (A and C) Myr1.STMV, lacking the RSV size determinant but containing I domains. (B and D) Myr1.STMV.t, lacking the RSV size determinant and I domains.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

One of the most fundamental ways to group viruses is by the presence or absence of a surrounding lipid bilayer. Enveloped viruses usually acquire this bilayer as their viral components bud through one of the membranes of the infected cell. Acquisition of the envelope is essential for infectivity, since the viral proteins required for attachment to host cells reside on the outer surface of this membrane. In contrast, viruses without envelopes are assembled in the cytoplasm or the nucleus of the cell in a manner that is largely independent of membranes. These viruses are generally released when the infected cell dies, and the functions needed for binding and entering another cell are intrinsic parts of the exposed surface of the capsid structure.

Although the assembly pathways of enveloped and nonenveloped viruses appear to be very different, the experiments described here suggest that their distinguishing features may be far simpler than expected. In particular, we have shown that the major capsid proteins from two very different icosahedral viruses can be inserted into the retroviral budding pathway merely by adding M and L domains to their N termini. This simple alteration shifts their assembly from the nucleus (for VP1) or cytoplasm (for STMV) to the plasma membrane, whereupon membrane-enclosed, virus-like particles are released at rates comparable to those of wild-type retroviruses. Far from being inert passengers, these foreign proteins appear to interact with each other to substitute for both the size (e.g., the CA-SP sequence) and density (e.g., the I domains) determinants of Gag.

Although particles of relatively uniform and smaller size are produced by the chimeras, rate-zonal gradient analysis does not provide information about their morphology. It is possible that enveloped, icosahedral capsids are not produced when the M and L domains are present and that the driving forces needed for Gag interactions are provided by various SV40 and STMV assembly intermediates that begin, but fail to complete, their normal folding pathways. Future electron microscopy studies may shed light on this notion. Nonetheless, there are three reasons for concluding that the foreign capsid proteins are active participants in defining chimeric particle size. First, deletions throughout the CA-SP sequence result in very heterogeneously sized particles (13), but this heterogeneity is suppressed by VP1 or the STMV coat protein. Second, for the Gag-VP1 chimeras, the particles respond to increased levels of calcium, moving from a very heterogeneous profile to a more uniform distribution. Third, for both chimeras, the particles not only are relatively homogeneous in size but also are smaller than the control particles. In contrast, CA-SP-NC mutants (e.g., Myr1.R3J; Fig. 5A) produce a wide range of particle sizes but not ones that are smaller than those of the wild type (13).

In summary, we have demonstrated that the structural proteins from two icosahedral viruses can be inserted into the budding pathway by placing an M domain and an L domain on their N termini. In addition, these capsid proteins function within this context to create particles of relatively uniform size, thereby replacing the CA-SP-NC sequence of Gag. The modularity and exchangeability of these viral proteins underscore the conservation of function among evolutionarily distinct viruses.