Multiple Gag Domains Contribute to Selective Recruitment of Murine Leukemia Virus (MLV) Env to MLV Virions

Abstract

Retroviruses, like all enveloped viruses, must incorporate viral glycoproteins to form infectious particles. Interactions between the glycoprotein cytoplasmic tail and the matrix domain of Gag are thought to direct recruitment of glycoproteins to native virions for many retroviruses. However, retroviruses can also incorporate glycoproteins from other viruses to form infectious virions known as pseudotyped particles. The glycoprotein murine leukemia virus (MLV) Env can readily form pseudotyped particles with many retroviruses, suggesting a generic mechanism for recruitment. Here, we sought to identify which components of Gag, particularly the matrix domain, contribute to recruitment of MLV Env into retroviral particles. Unexpectedly, we discovered that the matrix domain of HIV-1 Gag is dispensable for generic recruitment, since it could be replaced with a nonviral membrane-binding domain without blocking active incorporation of MLV Env into HIV virions. However, MLV Env preferentially assembles with MLV virions. When MLV and HIV particles are produced from the same cell, MLV Env is packaged almost exclusively by MLV particles, thus preventing incorporation into HIV particles. Surprisingly, the matrix domain of MLV Gag is not required for this selectivity, since MLV Gag containing the matrix domain from HIV is still able to outcompete HIV particles for MLV Env. Although MLV Gag is sufficient for selective incorporation to occur, no single Gag domain dictates the selectivity. Our findings indicate that Env recruitment is more complex than previously believed and that Gag assembly/budding sites have fundamental properties that affect glycoprotein incorporation.

INTRODUCTION

To form an infectious particle, retroviruses must assemble several viral and host cell components. In general, retroviral assembly is an exclusive process where viral components only assemble with other viral components from the same or very closely related viruses, e.g., HIV-1 and -2 (1–6). A well-known exception to this exclusivity is the incorporation of viral glycoproteins. The ability of enveloped viruses to generate infectious particles with foreign viral glycoproteins has been termed pseudotyping and has been reported for a variety of viruses and glycoproteins, although first observed by Granoff and Hirst in 1954 (7, 8). Comparing pseudotyping and native glycoprotein incorporation can provide insight into the respective generic and selective mechanisms used for viral glycoprotein acquisition by many viruses. Are there mechanisms shared between diverse viruses for glycoprotein incorporation? Do some viruses have specific means to recruit their own glycoprotein during assembly/budding? We already have some insight into these questions.

The incorporation of native retroviral glycoproteins (Envs) likely involves specific interactions between Env and the structural Gag protein. For example, the N-terminal membrane-binding matrix domain (MA) of HIV-1 Gag has been reported to interact with the cytoplasmic tail domain (CTD) of HIV-1 Env, and this interaction is believed to be required for Env incorporation into virions (9–20). The HIV-1 Env CTD and MA have also been observed to coevolve in human infection, further underscoring the importance of their interaction (21). For MLV, the R-peptide, a 16-amino-acid peptide cleaved from the C terminus of the Env transmembrane subunit (TM) by the viral protease (PR), has been shown to physically associate with viral cores in cells (22). More recent work with MLV indicates that an interaction between MA and the Env CTD is required for MLV polarized budding (23).

Glycoprotein incorporation during pseudotyping is unlikely to occur through direct interactions. Although some glycoproteins, such as vesicular stomatitis virus G protein (VSV-G) and murine leukemia virus (MLV) Env, readily form pseudotyped particles with diverse viruses, other glycoproteins seem to function only with native and closely related viruses (24, 25). HIV-1 Env falls into this latter category (12). Our lab has demonstrated that MLV Env and VSV-G are enriched at HIV-1 (referred to hereafter as HIV) and Rous sarcoma virus budding sites, suggesting a conserved generic mechanism used by many viruses for glycoprotein incorporation (26). Although the CTDs of many glycoproteins are not needed for incorporating into virions, the physical aspects of glycoproteins that are needed for active incorporation remain a mystery. Likewise, the domains and functions of Gag needed for active incorporation are also not fully characterized.

Here, we investigated which Gag components contribute to MLV Env enrichment at budding sites and recruitment into retroviral virions. Our findings indicate that MA is not required for MLV Env enrichment and incorporation during pseudotyping. Even though MLV Env forms pseudotyped particles with both foreign and native viruses, it is highly selective toward MLV when both MLV and HIV virions are produced from the same cells. Surprisingly, MLV MA is not necessary for this selectivity.

MATERIALS AND METHODS

Cells and plasmids.

293FT (Invitrogen) and 293T mCAT-1 were propagated in Dulbecco modified Eagle medium supplemented with 10% fetal bovine serum and 1% glutamine with periodic G418 selection. The HIV-CMV-GFP vector is a minimal pNL4-3 lacking Vpr, Vif, Vpu, and Nef that expresses green fluorescent protein (GFP) (27). Lyn-HIV and mMA-HIV were derived from the wild-type vector by replacing the MA domain with the Lyn membrane-binding domain (MGCIKSKRKD) or MLV MA, respectively, by standard cloning techniques while preserving the protease cleavage site (Fig. 1). The HIV SYNGP (28) coding sequence with a nonfunctional PR cloned into an internal ribosome entry site GFP expression vector was a gift from Paul Bieniasz. Wild-type MLV GagPol and ecotropic MLV Env vectors have been previously described (29). hMA-MLV and hMAΔp12-MLV were constructed from the wild-type vector by replacing either MA or MA and p12 with a codon-optimized HIV MA from pSYNGP (28). The ΔPol and leucine zipper variants of the MLV vectors were created by standard cloning techniques to remove the Pol coding sequence and replacing the coding sequence of NC with the coding sequence for the leucine zipper domain of CREB1 with a flexible linker at the N terminus (GSGSGRECRRKKKEYVKCLENRVAVLENQNKTLIEELKALKDLYCHKSD) without preserving the cleavage site (30). For MLV reporters, pQXCIP (Clontech) had GFP or tdTomato inserted by standard cloning procedures (31). The wild-type MLV Env had the coding sequence for the R peptide removed by standard cloning to create an R-peptide-deleted MLV Env. VSV-G expression vector was obtained from the NIH AIDS Research and Reference Reagent Program (32). Late domain defective Gag constructs were used for scanning electron microscopy (SEM) experiments. A vector expressing HIV Gag with the late domain (PTAP) changed to alanines was previously described (26). The Lyn membrane-binding domain was cloned into the previous vector, replacing HIV MA. For wild-type MLV, the GagPol vector had the primary late domain (PPPY) replaced with alanines. Also for SEM, a yellow fluorescent protein (YFP)-tagged MLV Env vector was used (29).

Fig 1.

Schematic of HIV and MLV Gag chimeras. HIV is black, MLV is white, Lyn is light gray, and the leucine zipper domain is dark gray. Most of the chimeras were engineered to maintain the complete protease cleavage site. The amino acid sequences at the chimeric junctions are shown with the upstream sequence in boldface and a slash indicating the viral protease cleavage site.

Infectivity assay.

293FT cells were transfected with the indicated viral and glycoprotein vectors using polyethyleneimine (33). Two days after transfection, supernatants were collected and frozen at −80, thawed, and clarified by centrifugation at 2,000 relative centrifugal force (RCF), and an aliquot was used to inoculate target cells. Two days after transduction, the target cells (293T mCAT-1) were harvested as a single cell suspension and fixed with 4% paraformaldehyde. A flow cytometer (Accuri) was used to analyze cells for GFP or tdTomato fluorescence, thus indicating HIV and/or MLV transductions, and to produce dot plots. Titers of the supernatants were then calculated based upon the cells infected and dilution of supernatant.

Western blotting.

Virus was pelleted by centrifugation at >16,000 RCF for at least 1 h from supernatants obtained as for infectivity, with ritonavir (7 μg/ml, obtained through the NIH AIDS Research and Reference Reagent Program) added at the time of transfection where indicated. Viral pellets were resuspended and denatured by heating in sample buffer (50 mM Tris, 2% sodium dodecyl sulfate [SDS], 10% glycerol, 2% β-mercaptoethanol). Samples were then resolved by SDS-PAGE, transferred to 0.25-μm-pore-size polyvinylidene fluoride membranes, and immunoblotted with primary and secondary antibodies. The following two reagents were obtained through the NIH AIDS Research and Reference Reagent Program: HIV-1 p24 hybridoma (183-H12-5C) from Bruce Chesebro (34) and antiserum to HIV-1 p17 from Paul Spearman (35). Antibodies against MLV CA (R187, ATCC) and p15E (42/114) were produced from hybridomas (36, 37). Horseradish peroxidase-linked anti-mouse (A5278) and anti-rat (A5785) antibodies were obtained from Sigma. IRdye800 conjugated anti-rabbit was obtained from Rockland, Inc. (catalog no. 611-732-127). After washing, membranes were imaged with optimized brightness and contrast by either a LI-COR Odyssey or Fuji Film LAS-3000. Further adjustments of brightness and contrast to individual panels were made during figure assembly in Microsoft PowerPoint.

SEM.

Correlative SEM was performed as previously described (26). Briefly, 293T mCAT-1 cells were plated onto glass coverslips with a thin-layer gold coat in a grid pattern and transfected with the indicated late domain mutant Gag and Env constructs using PEI. Fluorescence microscopy was used to map transfected cells prior to fixation and labeling against the Env. Anti-GFP 20 antibody (G6539; Sigma) was used to label YFP-tagged MLV Env. Gold-labeled anti-mouse secondary was obtained from Jackson Laboratories (catalog no. 115-205-146). After critical point drying and carbon evaporation coating, samples were imaged with a Hitachi S4700 FE SEM at the University of Missouri Electron Microscopy Core Facility. Image brightness and contrast were adjusted in Microsoft PowerPoint for clear resolution of topology and gold particles.

RESULTS

Requirements of MA for generic recruitment of MLV Env.

The observation that some glycoproteins readily form pseudotyped particles with diverse viruses suggests that incorporation of these glycoproteins is passive or governed by a conserved generic mechanism. In the case of HIV pseudotyped with MLV Env, we have demonstrated that MLV Env is enriched at budding virions relative to the cell surface, suggesting that an active mechanism leads to incorporation (26). If a direct interaction mediates this pseudotyping, it would most likely be between the MLV Env CTD and HIV MA, since these are the only domains of the two proteins thought to be in close proximity in the virion. To determine whether HIV MA is required for active MLV Env recruitment and incorporation into virions, we replaced MA in HIV Gag with the membrane-binding domain from Lyn, a Tyr-kinase from the Src family (Fig. 1). This chimera's compatibility with MLV Env was determined by assaying infectivity, Env incorporation into virions and Env enrichment at budding sites (Fig. 2). Each assay was also performed with HIV and MLV for comparison. All three viruses produced infectious particles when pseudotyped with MLV Env. Lyn-HIV is less infectious than HIV but comparable to what has been reported with similar constructs (Fig. 2A) (38, 39). MLV is also less infectious than HIV; however, because the two viral expression systems are different, the infectivities cannot be directly compared. We wanted to determine whether the drop in infectivity for Lyn-HIV was due to a failure in MLV Env incorporation. Western analysis of MLV Env TM revealed that MLV Env is incorporated into virions of all three viruses, indicating that HIV MA is not required for MLV Env incorporation into HIV viral particles and the drop in infectivity for Lyn-HIV has another cause (Fig. 2B). The ratio of uncleaved (p15E) to cleaved (p12E) MLV Env TM is greater with HIV and Lyn-HIV than with MLV. This observation is consistent with a report indicating that HIV PR is less efficient than MLV PR at cleaving the R peptide from MLV Env TM (40). We also observed MLV Env enrichment on budding virions with all three viruses by scanning electron microscopy (SEM), recapitulating the results from the Western blotting (Fig. 2C). These data indicate that HIV MA is not required for the active recruitment and incorporation of MLV Env into HIV particles.

Fig 2.

MLV Env is recruited to and incorporated into viral particles lacking MA. (A) 293FT cells were transfected with plasmids expressing viral components for the indicated virus lacking Env and with a plasmid expressing MLV Env. Equal volumes of supernatant clarified of cell debris were used to transduce 293T mCAT-1 cells. GFP was expressed from the viral genomes and analyzed by flow cytometry (x axis). The y axis indicates forward scatter (FS). The percentages of GFP-positive cells are indicated in upper right of each plot. (B) Virus was purified from the remaining supernatant in panel A, and a Western blot performed against HIV CA, MLV CA, and MLV Env (p15E/p12E). (C) 293T mCAT-1 cells were transfected with plasmids expressing the indicated Gag constructs with late-domain mutations and YFP-tagged MLV Env. YFP was immunogold labeled for backscatter detection. Scanning electron micrographs were obtained for budding virions and labeled Env. Scale bars, 500 nm. All data shown are representative of or the average of data from at least three independent experiments.

Selective recruitment of MLV Env into MLV virions.

Although both HIV and MLV can produce infectious particles with MLV Env, we previously showed that when MLV and HIV virions are produced from the same cells and compete for a limiting amount of MLV Env, there is a disproportionate drop in infectious HIV particle production (41). In contrast, a competition for limiting VSV-G or an MLV Env lacking its CTD results in a near equivalent drop in infectious particle production for both viruses. Since the previous study used a different HIV expression system, we repeated this competition assay with our current HIV proviral expression construct. When assembling HIV and MLV virions compete for MLV Env, MLV infectivity remains robust at the apparent expense of HIV infectivity (Fig. 3A), supporting a model in which MLV virions have a selective mechanism to incorporate MLV Env that HIV virions lack. The dramatic reduction in HIV infectivity is not observed with VSV-G, indicating that the reduction seen with MLV Env is not caused by the limitation of general cellular resources. These results are consistent with MLV virions having a selective mechanism for incorporating MLV Env, although other explanations are possible.

Fig 3.

MLV outcompetes HIV for limited MLV Env. (A) Transfections and transductions were performed to produce HIV and MLV (alone or in combination) with limiting MLV Env or VSV-G. GFP was expressed from HIV genomes and tdTomato from MLV genomes. Fluorescence from GFP (x axis) and tdTomato (y axis) in transduced cells was assayed by flow cytometry. Representative dot plots are shown. The percentage of GFP positive cells in each plot is indicated in the lower right and of tdTomato-positive cells in the top left. (B) Infectivity was obtained from viruses alone or viruses in competition (comp) as in panel A with increasing amounts of glycoproteins. Upper graphs show the titer (infectious units [IU] per ml, y axis) for HIV and MLV at different glycoprotein concentrations. Lower graphs show the ratio of infectivity from competitions to infectivity without competition. The standard deviation is indicated by the error bars; the absence of lower-bound bars indicates a below-zero extension. (C) Western analysis of purified virions from the supernatant of cells transfected with plasmids for the indicated viral constructs and MLV Env, such that MLV Env would be expressed in a limiting manner. Western blotting was performed with antibodies against HIV CA, MLV CA, and MLV Env (p15E/p12E). Antibodies against the two CAs were used simultaneously. All data shown are representative of or the average of data from at least three independent experiments.

To better assess whether assembling MLV virions outcompete HIV virions for limited MLV Env, we repeated the competition with a fixed amount of the HIV and MLV constructs and increasing amounts of MLV Env or VSV-G expression vectors (Fig. 3B). The amounts of HIV and MLV used with each glycoprotein were adjusted such that both viruses had roughly equivalent titers at the maximum viral glycoprotein level. If MLV Envs were sequestered away from HIV virions by MLV virions, HIV infectivity would rapidly increase in the competition assay once MLV Env becomes nonlimiting, i.e., when MLV virions are saturated with MLV Env. If a different mechanism causes the competition, we would expect HIV infectivity to remain static or increase in conjunction with MLV infectivity. Individually, both HIV and MLV infectivity increase to a maximal range as amounts of either viral glycoprotein increase. When in competition, HIV infectivity is reduced by ∼100-fold at low concentrations of MLV Env, whereas MLV infectivity is unperturbed. Only at higher MLV Env concentrations, when MLV infectivity reached its maximum, did HIV infectivity recover to levels similar to HIV alone. This pattern indicates that MLV Env is selectively incorporated into MLV virions at the expense of HIV virions. The loss of infectivity at low MLV Env concentrations is in contrast to that seen with VSV-G, where the infectivities of both viruses increase in the competition at rates similar to when not in competition, indicating that there is little selective incorporation of VSV-G. We also observed similar recovery to a maximal infectivity in the competition regardless of the glycoprotein, indicating that the selectivity we observe at limiting glycoprotein concentrations is not due to the limitation of general cellular resources that would affect assembly and release of virions. The selectivity for MLV Env and the contrast with VSV-G is also apparent when we compared the ratio of infectivity of each virus when expressed in competition to when it is expressed by itself (Fig. 3B, lower graph). These data strongly support the hypothesis that MLV has a selective mechanism, which HIV lacks, to recruit MLV Env to assembling virions.

Next, we wanted to directly assay whether the competition is due to selective recruitment and incorporation of MLV Env. Normally, Western blot analysis would not be able to make such a distinction. However, cleavage of MLV Env's R peptide allows an exception. The R peptide is cleaved by PR during maturation of the retrovirus (42). Since PR of both HIV and MLV cleave off the R peptide, a protease defective virus will maintain the full-length MLV Env TM. Thus, a competition between wild-type and protease defective viruses allows us to discern which virions contain Env by resolving the cleaved (p12E) and uncleaved (p15E) MLV Env TM (Fig. 3C). To prevent HIV PR activity, we used a GagPol construct with an inhibitory mutation in PR. For MLV, we removed the Pol coding sequence, which includes PR, from the MLV GagPol expression construct. Competition assays with limiting amounts of MLV Env were performed as before, except instead of assaying for infectivity, virions were purified and analyzed by Western blotting for MLV Env TM. Individually, purified MLV virions contained cleaved Env (lane 3), while HIV virions lacking a functional protease contained uncleaved Env (lane 2). In competition, the vast majority of MLV Env is cleaved, indicating its presence predominantly in MLV virions (lane 5). With the reciprocal competition, the cleavage patterns are reversed. Although we had previously observed MLV Env TM to be incompletely processed when associated with HIV virions (Fig. 2B), here we observed that Env is cleaved efficiently in purified HIV virions (Fig. 3C, lane 1). This difference is likely due to the limited amount of MLV Env that is incorporated into virions here. MLV virions lacking PR had only uncleaved Env (lane 4). When HIV virions and MLV virions lacking PR are produced in competition, nearly all Env is uncleaved (lane 6), again indicating that Env is present almost exclusively in MLV virions and that Pol is not needed for this selectivity. That in both cases the cleavage pattern in the competition was that of MLV alone also reconfirms that the Gag proteins from the two viruses do not coassemble. These data are consistent with MLV having a selective mechanism to incorporate MLV Env into virions.

Selective MLV Env recruitment does not require MLV MA.

To determine whether MLV MA is necessary for the selectivity of MLV Env recruitment, we created a construct that has HIV MA replacing that of MLV (hMA-MLV), but maintains the MLV cleavage site. This construct is more informative than a Lyn replacement, as it maintains high infectivity and has a shared epitope with HIV. A competition assay was performed, as in Fig. 2, between HIV and hMA-MLV for MLV Env (Fig. 4A). As we saw with wild-type MLV, hMA-MLV infectivity is unperturbed in the competition while HIV infectivity is reduced, surprisingly indicating that MLV MA is not required for the selectivity. We also performed Western blotting on virions purified from the same supernatant used for infectivity to ensure that the selectivity could not be explained by differences in virion production. We probed against HIV MA, which was detected from both viruses, and against HIV and MLV CA. The observed level of HIV MA is similar between the two viruses alone, and the observed CA levels are also similar between each virus alone and the competition. The similarities among the viral components indicate that the selectivity toward hMA-MLV is not due to differences in virion production but is indeed due to the same selective incorporation of MLV Env that was observed with wild-type MLV. We also performed a competition assay with increasing amounts of glycoproteins, as in Fig. 3B, and similarly saw HIV infectivity suffer when in competition with hMA-MLV for MLV Env (Fig. 4B). Here hMA-MLV was less infectious with MLV Env than wild-type MLV indicating that MLV MA might play some role in a virion's compatibility with MLV Env. However, the reduction to HIV infectivity in the competition was just as striking as with wild-type MLV. So whatever function MLV MA plays, it is again shown to be unnecessary for the selective recruitment.

Fig 4.

MLV MA is not required for MLV virions to outcompete HIV virions for MLV Env. (A) Transfections, transductions, and Western analyses were performed with an MLV construct with MLV MA replaced with HIV MA. The infectivity of each virus was determined by flow cytometry, and the virus titer was calculated (infectious units [IU] per ml, y axis). Purified virus from the remaining supernatant was used to assay for relative levels of HIV MA, HIV CA, and MLV CA by Western blotting. Titers and blots are from the same experimental repeat. (B) Infectivity of HIV and hMA-MLV when in competition over when alone was determined with increasing amounts of glycoproteins as in Fig. 3B. All data shown are representative of or the average of data from at least three independent experiments.

Domains required for selective MLV Env recruitment.

Since MLV MA is dispensable for the selectivity for MLV virions in the competition, we sought to determine which other domains of MLV Gag contribute. We generated an hMA-MLV construct where we replaced MLV Gag's nucleocapsid domain (NC) and Pol with a leucine zipper, which has been shown to recapitulate NC's multimerization function required for particle formation and budding (hMA-MLV-Z) (43, 44). To determine whether p12 is required for the selectivity, we replaced MLV MA and p12 with HIV MA in backgrounds with or without Pol (hMAΔp12-MLV and hMAΔp12-MLVΔPol). Finally, we created a construct that both lacked p12 and had the leucine zipper in place of NC and Pol (hMAΔp12-MLV-Z). Although none of these constructs can produce infectious particles, the reduction to HIV infectivity with MLV Env and VSV-G in our competition assay was determined with each construct (Fig. 5A). All four constructs reduced the infectivity of HIV for both glycoproteins, although much more so with MLV Env than VSV-G, indicating that they maintain the selective mechanism for MLV Env incorporation. The exception to this trend was the construct only retaining the CA domain from MLV, which reduced HIV infectivity with both glycoproteins to the same extent, indicating that MLV CA alone is not sufficient for selective recruitment. To further assess whether these constructs were being released and incorporating MLV Env at the expense of HIV, we again assayed R-peptide cleavage with the constructs lacking Pol (Fig. 5B). Deletion of p12 impairs release and protease processing of virions (45). Even so, proteins from the constructs that lacked p12 were recovered from supernatants and associated with p15e. In agreement with the infectivity data, removal of either p12 or NC did not impair MLV's ability to incorporate MLV Env at the expense of HIV. However, when only MLV's CA remains, the MLV construct is no longer able to prevent HIV from incorporating MLV Env. These data indicate that no one domain of MLV Gag is responsible for the selectivity for MLV Env. Since p12 and NC are dispensable for the selectivity individually but not in concert, two explanations for the selectivity are likely. MLV CA may be responsible for the selectivity but requires the presence of either p12 or NC. Alternatively, CA may not contribute to the selectivity, while both p12 and NC independently participate in strong mechanisms for selective MLV Env recruitment.

Fig 5.

Mapping of the non-MA-mediated selective recruitment of MLV Env. (A) Infectivity of HIV with MLV Env or VSV-G was determined when in competition with the indicated constructs. Shown is the ratio of HIV infectivity in each competition to HIV infectivity without competition (control). The P value from a Student t test for the difference between the reduction in infectivity for MLV Env and VSV-G are shown for each construct. (B) Western analysis was performed as before on purified virions from cells expressing MLV Env with HIV and/or the indicated MLV constructs lacking Pol. All data shown are representative of or the average of data from at least three independent experiments.

MLV MA hinders R-peptide cleavage by HIV PR.

MLV MA and Env have been reported to interact, which may explain the difference in infectivity that we observed between MLV and hMA-MLV with MLV Env (23). These observations raise the question of whether MLV MA independently contributes to selective recruitment of MLV Env even though it is not required. To address this question, we created an HIV construct with HIV MA replaced with MLV MA, mMA-HIV. This chimera is considerably less infectious than wild-type HIV with MLV Env (Fig. 6A). Western analysis revealed that while mMA-HIV incorporates MLV Env, it is perturbed in R-peptide cleavage (Fig. 6A). R-peptide cleavage mediates the fusogenicity of MLV Env, so the uncleaved TM is consistent with the low infectivity (46, 47). To see whether infectivity could be recovered by circumventing the need for R-peptide cleavage, we used a truncated MLV Env missing the R peptide. Infectivity of mMA-HIV increased with the truncated Env, while HIV infectivity was unaffected. The chimera's infectivity was still greatly reduced compared to that of HIV however, indicating that while R-peptide cleavage may account for some of the loss to infectivity, this chimera has other defects as well (Fig. 6A). This mutant's defect in HIV PR cleavage of the R peptide but not of Gag suggests that MLV MA does physically interact with the tail of MLV Env. Assaying for competition by infectivity with this chimera is not feasible due to its poor infectivity with wild-type MLV Env and because the truncated MLV Env displays little selectivity in competition (41). Determining MLV Env incorporation is still possible by analyzing R-peptide cleavage, which we performed using hMA-MLV and mMA-HIV in the presence of ritonavir, a specific inhibitor of HIV protease (Fig. 6B). We observed that most of MLV Env TM is cleaved in the competition, with little to no reduction from the amount observed with hMA-MLV alone. There was also uncleaved MLV Env visible in the competition, although greatly reduced compared to mMA-HIV alone. To rule out the possibility that differences in virion production levels are the cause of the observed incorporation pattern, we varied the amount of mMA-HIV used in the competitions. Although increasing the amount of mMA-HIV did slightly increase the amount of uncleaved MLV Env observed, both when expressed alone or in competition, the change is modest and there is little to no impact on the level of hMA-MLV-associated cleaved MLV Env in the competition. In all, these results suggest that MLV MA does interact with MLV Env but that its contribution, if any, to recruitment is minor.

Fig 6.

MLV MA interacts with MLV Env. (A) HIV or mMA-HIV were pseudotyped with MLV Env or MLV Env lacking the R peptide (ΔR) and analyzed for infectivity and by Western blotting of virions. Titers and blots are from the same experimental repeat. (B) Western analysis of purified virions from cells expressing hMA-MLV and various amounts (quantity of plasmid used in transfection is shown) mMA-HIV alone or in competition with limiting MLV Env. Ritonavir was added at transfection for all samples. All data shown are representative of or the average of data from at least three independent experiments.

DISCUSSION

Research to date suggests that native Env recruitment by retroviruses involves interactions between MA and the CTD of Env. With HIV, mutations to MA can abolish Env recruitment and can be compensated for by deletion of Env's CTD (12–16, 20). HIV Env and Gag have also been shown to have interactions affecting assembly (10–12, 17–19). For instance, placement of HIV MA into visna virus can convey the ability to incorporate HIV Env (12). Evidence for interactions between MA and Env have also been found for SIV and MLV (17, 22, 23, 48, 49). However, some glycoproteins can be used by unrelated viruses to mediate infections, indicating a mechanism of incorporation other than native evolved interactions. Although it has been proposed that such incorporation is passive, we have previously demonstrated that with VSV-G and MLV Env, glycoproteins become enriched at sites of retroviral assembly and budding, suggesting an active recruitment process (26).

We had hypothesized that the observed active incorporation of MLV Env into HIV particles is due to a shared common mechanism mediated by MA for both MLV and HIV. However, our hypothesis was disproven when we observed that active incorporation of MLV Env does not require an MA domain. Infectivity of Lyn-HIV is impaired however, recapitulating the importance of MA in steps of the viral life cycle other than assembly (38, 39, 50). Thus, the mechanism that leads to active incorporation during pseudotyping remains unknown. Evidence that MLV Env's CTD is also dispensable for active incorporation further suggests that direct interactions between Env and Gag are not required (41). A shared trafficking destination and indirect interactions are the remaining explanations for active incorporation. Currently, our favored model is that Env has an affinity for the unique lipid environment created by a budding virus, possibly involving lipid rafts and tetraspanin-enriched domains that others have observed (51–54).

Even though MA is not required for generic recruitment of MLV Env into viral particles, it remained possible that selective recruitment of MLV Env to MLV particles is mediated by MLV MA. MLV Gag and MLV Env's CTD are known to interact, and we expected these interactions to be from MA and to contribute to a selective recruitment mechanism (22, 23, 41). To determine whether MLV Env was selectively recruited into MLV virions, we developed competition assays based upon infectivity and R-peptide cleavage. Selective recruitment was observed, but surprisingly it does not require MLV MA. Indeed, no single domain seemed to be individually required for the selective recruitment.

The exact mechanism for selective recruitment and the contributions of each domain still need to be elucidated. Perhaps the most obvious possible explanation for selectivity is a physical interaction, either direct or indirect. MA is the only domain close enough to the membrane that a direct physical interaction is likely (55). Our observation that MLV MA prevents cleavage of the R peptide when replacing HIV MA does suggest a physical interaction exists. However, this chimera is still outcompeted by hMA-MLV, indicating that MLV MA does not play a major role in selective MLV Env recruitment. Another possible mechanism for selective recruitment could be trafficking. MLV Gag may assemble at sites earlier in MLV Env's trafficking path than where HIV virions assemble. This model is supported by evidence suggesting that MLV virions can bud into internal vesicles (29, 56). Studies have also shown that MLV Gag and Env affect each other's trafficking, although this is at least in part dependent upon MA (23, 57–60). Finally, CA also determines the specific hexameric lattice and diameter of virions, which may better accommodate MLV Env's CTD in MLV virions either sterically or by effects on the lipid microenvironment (55, 61, 62). MLV CA alone, however, was not sufficient for the selective recruitment. So, if CA does contribute to recruitment, then p12 or NC must provide a redundant function, possibly a second interaction site for a cellular protein or simply structural stabilization for CA. The other possible explanation for our mapping results is that both MLV p12 and NC independently have mechanisms for the selective recruitment of MLV Env. Further studies will be needed to clarify which recruitment mechanisms are used and which domains are responsible for them.