The role of WWP1-Gag interaction and Gag ubiquitination in assembly and release of human T-cell leukemia virus type 1

The PPPY motif in the matrix (MA) domain of human T-cell leukemia virus type 1 (HTLV-1) Gag associates with WWP1, a member of the HECT domain containing family of E3 ubiquitin ligases. Mutation of the PPPY motif arrests particle assembly at an early stage and abolishes ubiquitination of MA. Similar effects are seen when Gag is expressed in the presence of a truncated form of WWP1 that lacks the catalytically active HECT domain (C2WW). To understand the role of ubiquitination in budding, we mutated the four lysines in MA to arginines and identified lysine 74 as the unique site of ubiquitination. Virus-like particles produced by the K74R mutant did not contain ubiquitinated MA and showed a fourfold reduction in the release of infectious particles. Furthermore, the K74R mutation rendered assembly hypersensitive to C2WW inhibition; K74R Gag budding was inhibited at significantly lower levels of expression of C2WW compared with wild-type Gag. This finding indicates that the interaction between Gag and WWP1 is required for functions other than Gag ubiquitination. Additionally, we show that the PPPY⁻ mutant Gag exerts a strong dominant-negative effect on the budding of wild-type Gag, further supporting the importance of recruitment of WWP1 to achieve particle assembly.     

Budding of equine infectious anemia virus is insensitive to proteasome inhibitors

The only retrovirus protein required for the budding of virus-like particles is the Gag protein; however, recent studies of Rous sarcoma virus (RSV) and human immunodeficiency virus have suggested that modification of Gag with ubiquitin (Ub) is also required. As a consequence, the release of these viruses is reduced in the presence of proteasome inhibitors, which indirectly reduce the levels of free Ub within the cell. Here we show that the budding of equine infectious anemia virus (EIAV) from infected equine cells is largely unaffected by these drugs, although use of one inhibitor (MG-132) resulted in a dramatic block to proteolytic processing of Gag. This lack of sensitivity was also observed in transiently transfected avian cells under conditions that greatly reduce RSV budding. Moreover, insensitivity was observed when the EIAV Gag protein was expressed in the absence of all the other virus products, indicating that they are not required for this phenotype. An activity that enables EIAV to tolerate exposure to proteasome inhibitors was mapped to the C-terminal p9 sequence, as demonstrated by the ability of an RSV Gag-p9 chimera to bud in the presence of the drugs. Intriguingly, the p9 sequence contains a short sequence motif that is similar to a surface-exposed helix of Ub, suggesting that EIAV Gag may have captured a function that allows it to bypass the need for ubiquitination. Thus, the mechanism of EIAV budding may not be substantially different from that of other retroviruses, even though it behaves differently in the presence of proteasome inhibitors.     

Repositioning basic residues in the M domain of the Rous sarcoma virus gag protein

The first 86 residues of the Rous sarcoma virus (RSV) Gag protein form a membrane-binding (M) domain that directs Gag to the plasma membrane during budding. Unlike other retroviral Gag proteins, RSV Gag is not myristylated; however, the RSV M domain does contain 11 basic residues that could potentially interact with acidic phospholipids in the plasma membrane. To investigate this possibility, we analyzed mutants in which basic residues in the M domain were replaced with asparagines or glutamines. The data show that neutralizing as few as two basic residues in the M domain blocked particle release and prevented Gag from localizing to the plasma membrane. Though not as severe, single neutralizations also diminished budding and, when expressed in the context of proviral clones, reduced the ability of RSV to spread in cell cultures. To further explore the role of basic residues in particle production, we added lysines to new positions in the M domain. Using this approach, we found that the budding efficiency of RSV Gag can be improved by adding pairs of lysines and that the basic residues in the M domain can be repositioned without affecting particle release. These data provide the first gain-of-function evidence for the importance of basic residues in a retroviral M domain and support a model in which RSV Gag binds to the plasma membrane via electrostatic interactions.     

Functional chimeras of the Rous sarcoma virus and human immunodeficiency virus gag proteins.

The Gag protein encoded by Rous sarcoma virus (RSV) is the only viral product required for the process of budding whereby virus particles are formed at the plasma membrane. Deletion analysis of this Gag molecule has revealed several regions (assembly domains) that are important for budding. One of these domains is located at the amino terminus and is needed for membrane binding. Another is located within the carboxy-terminal third of the protein. Though there is little sequence homology among the Gag proteins of unrelated retroviruses, it seemed possible that their assembly domains might be functionally conserved, and to explore this idea, numerous Gag chimeras were made. The results indicate that the first 10 amino acids of the human immunodeficiency virus (HIV) Gag protein can suppress the block to budding caused by deletions in the RSV MA sequence, much as described previously for the first 10 residues from the Src oncoprotein (J.W. Wills, R.C. Craven, R. A. Weldon, Jr., T. D. Nelle, and C.R. Erdie, J. Virol. 65:3804-3812, 1991). In addition, the carboxy-terminal half of the HIV Gag protein was fused to a truncated RSV Gag molecule, mutant Bg-Bs, which is unable to direct core assembly. This chimera was able to produce particles at a rate identical to that of RSV and of a density similar to that of authentic virions. Deletion analysis of the carboxy-terminal chimera revealed two small regions within the HIV NC protein that were sufficient for endowing mutant Bg-Bs with these properties. Chimeras lacking both regions produced particles of a low density, suggesting that these sequences may be involved in the tight packing of Gag molecules during assembly. In a related set of experiments, replacement of the RSV protease with that of HIV resulted in premature processing within the RSV sequence and a block to budding. Particle assembly was restored when the HIV PR activity was inactivated by mutagenesis. Collectively, the data presented here illustrate the functional similarities of Gag proteins from unrelated retroviruses.     

Retroviruses have differing requirements for proteasome function in the budding process

Proteasome inhibitors reduce the budding of human immunodeficiency virus types 1 (HIV-1) and 2, simian immunodeficiency virus, and Rous sarcoma virus. To investigate this effect further, we examined the budding of other retroviruses from proteasome inhibitor-treated cells. The viruses tested differed in their Gag organization, late (L) domain usage, or assembly site from those previously examined. We found that proteasome inhibition decreased the budding of murine leukemia virus (plasma membrane assembly, PPPY L domain) and Mason-Pfizer monkey virus (cytoplasmic assembly, PPPY L domain), similar to the reduction observed for HIV-1. Thus, proteasome inhibitors can affect the budding of a virus that assembles within the cytoplasm. However, the budding of mouse mammary tumor virus (MMTV; cytoplasmic assembly, unknown L domain) was unaffected by proteasome inhibitors, similar to the proteasome-independent budding previously observed for equine infectious anemia virus (plasma membrane assembly, YPDL L domain). Examination of MMTV particles detected Gag-ubiquitin conjugates, demonstrating that an interaction with the ubiquitination system occurs during assembly, as previously found for other retroviruses. For all of the cell lines tested, the inhibitor treatment effectively inactivated proteasomes, as measured by the accumulation of polyubiquitinated proteins. The ubiquitination system was also inhibited, as evidenced by the loss of monoubiquitinated histones from treated cells. These results and those from other viruses show that proteasome inhibitors reduce the budding of viruses that utilize either a PPPY- or PTAP-based L domain and that this effect does not depend on the assembly site or the presence of monoubiquitinated Gag in the virion.     

In vivo interference of Rous sarcoma virus budding by cis expression of a WW domain

For all enveloped viruses, the actual mechanism by which nascent virus particles separate or "pinch off" from the cell surface is largely unknown. In the case of retroviruses, the Gag protein drives the budding process, and the virus release step is directed by the late (L) assembly domain within Gag. A PPPPY motif within the L domain of Rous sarcoma virus (RSV) was previously characterized as being critical for the release of virions and shown to interact in vitro with the WW domain of Yes-associated protein (Yap). To determine whether WW domain-L domain interactions can occur in vivo, we attempted to interfere with the host cell machinery normally recruited to the site of budding by inserting this WW domain in different locations within Gag. At a C-terminal location, the WW(Yap) domain had no effect on budding, suggesting that the intervening I domains (which provide the major region of Gag-Gag interaction) prevent its access to the L domain. When positioned on the other side of the I domains closer to the L domain, the WW(Yap) domain resulted in a dramatic interference of particle release, and confocal microscopy revealed a block to budding on the plasma membrane. Budding was restored by attachment of the heterologous L domain of human immunodeficiency virus type 1 Gag, which does not bind WW(Yap). These findings suggest that cis expression of WW domains can interfere with RSV particle release in vivo via specific, high-affinity interactions at the site of assembly on the plasma membrane, thus preventing host factor accessibility to the L domain and subsequent virus-cell separation. In addition, they suggest that L domain-specific host factors function after Gag proteins begin to interact.     

Evidence for a second function of the MA sequence in the Rous sarcoma virus Gag protein.

During retrovirus assembly, Gag proteins bind to the inner leaflet of the plasma membrane to initiate the budding process. The molecular basis of this protein-lipid interaction is poorly understood. For the human, immunodeficiency virus type 1 Gag protein, we recently reported that the membrane-binding domain resides within the N-terminal 31 amino acids and consists of two components: myristate and a cluster of basic residues, which together promote membrane binding in vitro and budding in vivo (W. Zhou, L. J. Parent, J. W. Wills, and M. D. Resh, J. Virol. 68:2556-2569, 1994). The positively charged residues associate electrostatically with acidic phospholipids to stabilize membrane binding, while myristate provides membrane-binding energy via hydrophobic interactions. Here we demonstrate that the human immunodeficiency virus type 1 Gag membrane-binding domain can fully replace the membrane-targeting function of the N-terminal 100 residues of the non-myristylated Rous sarcoma virus (RSV) Gag protein. To further explore the importance of myristate and basic residues in membrane binding, we developed a gain-of-function assay whereby budding was restored to defective mutants of RSV Gag. Detailed mutational analysis revealed that the position, number, and context of charged residues are crucial to budding. Myristate provides additional membrane-binding energy, which is critical when a Gag protein is near the threshold of stable membrane association. Finally, viruses with altered matrix (MA) proteins that are noninfectious, even though they produce particles with high efficiency, were identified. Thus, we present the first evidence that the RSV MA sequence plays two distinct roles, membrane binding during particle assembly and a second, as yet undefined function required for viral infectivity.     

Cumulative mutations of ubiquitin acceptor sites in human immunodeficiency virus type 1 gag cause a late budding defect

The p6 domain of human immunodeficiency virus type 1 (HIV-1) Gag has long been known to be monoubiquitinated. We have previously shown that the MA, CA, and NC domains are also monoubiquitinated at low levels (E. Gottwein and H. G. Krausslich, J. Virol. 79:9134-9144, 2005). While several lines of evidence support a role for ubiquitin in virus release, the relevance of Gag ubiquitination is unclear. To directly address the function of Gag ubiquitination, we constructed Gag variants in which lysine residues in the NC, SP2, and p6 domains were mutated to arginine either in individual domains or in combination. Using these mutants, we showed that in addition to MA, CA, NC, and p6, SP2 is also mono- or di-ubiquitinated at levels comparable to those of the other domains. Replacement of all lysine residues in only one of the domains had minor effects on virus release, while cumulative mutations in NC and SP2 or in NC and p6 resulted in an accumulation of late budding structures, as observed by electron microscopy analysis. Strikingly, replacement of all lysine residues downstream of CA led to a significant reduction in virus release kinetics and a fivefold accumulation of late viral budding structures compared to wild-type levels. These results indicate that ubiquitination of lysine residues in Gag in the vicinity of the viral late domain is important for HIV-1 budding, while no specific lysine residue may be needed and individual domains can functionally substitute. This is consistent with Gag ubiquitination being functionally involved in a transient protein interaction network at the virus budding site.     

Biochemical characterization of rous sarcoma virus MA protein interaction with membranes

The MA domain of retroviral Gag proteins mediates association with the host cell membrane during assembly. The biochemical nature of this interaction is not well understood. We have used an in vitro flotation assay to directly measure Rous sarcoma virus (RSV) MA-membrane interaction in the absence of host cell factors. The association of purified MA and MA-containing proteins with liposomes of defined composition was electrostatic in nature and depended upon the presence of a biologically relevant concentration of negatively charged lipids. A mutant MA protein known to be unable to promote Gag membrane association and budding in vivo failed to bind to liposomes. These results were supported by computational modeling. The intrinsic affinity of RSV MA for negatively charged membranes appears insufficient to promote efficient plasma membrane binding during assembly. However, an artificially dimerized form of MA bound to liposomes by at least an order of magnitude more tightly than monomeric MA. This result suggests that the clustering of MA domains, via Gag-Gag interactions during virus assembly, drives membrane association in vivo.     

In vivo homodimerisation of HTLV-1 Gag and MA gives clues to the retroviral capsid and TM envelope protein arrangement

During retroviral particle formation, the capsid precursors (Gag) associate with the cell membrane via their matrix (MA) domain to form viral assembling particles. After budding, Gag and its proteolytically matured MA, form a shell in the released immature and mature particles, respectively. Although the arrangement of Gag domains in vitro and their radial organisation in retroviral particles have been extensively studied, little is known concerning Gag inter-subunit interactions in authentic retroviruses. We report that human T-cell leukemia virus type 1 Gag homodimerises in the cell via a disulphide bonding at cysteine 61 in the MA domain. Most Gags are homodimeric after budding and MAs are also dimeric in mature authentic virions. Molecular modelling of the MA domain indicates that non-covalent interactions at the MA dimer interface may also be important for Gag (and MA) dimerisation. In addition, all amino acids previously reported to be involved in MA-transmembrane (TM) interactions are located on the MA face opposite to the dimer interface. The model reveals that homodimerisation is compatible with a hexameric network of Gag and MA dimers that look like the hexameric networks observed for other retroviruses. These data, together with previous studies, lead us to propose a supra-molecular arrangement model in which the transmembrane glycoproteins of the virion envelope are anchored in a hexameric cage hole formed by the MA.     

Analysis of human immunodeficiency virus type 1 Gag ubiquitination

Ubiquitin is important for the release of human immunodeficiency virus type 1 (HIV-1) and several other retroviruses, but the functional significance of Gag ubiquitination is unknown. To address this problem, we decided to analyze Gag ubiquitination in detail. A low percentage of the HIV-1 p6 protein has previously been shown to be ubiquitinated, and published mutagenesis data suggested that Gag ubiquitination is largely lost upon mutation of the two lysine residues in p6. In this study, we show that Gag proteins lacking the p6 domain or the two lysine residues within p6 are ubiquitinated at levels comparable to those of the wild-type Gag protein. We detected monoubiquitinated forms of the matrix (MA), capsid (CA), and nucleocapsid (NC) proteins in mature virus preparations. Protease digestion of Gag polyproteins extracted from immature virions indicated that ubiquitinated MA, CA, and possibly NC are as abundant as ubiquitinated p6. The HIV-1 late-domain motifs PTAP and LRSLF were not required for Gag ubiquitination, and mutation of the PTAP motif even resulted in an increase in the amount of Gag-Ub conjugates detected. Finally, at steady state, ubiquitinated Gag proteins were not enriched in either membrane-associated or virus-derived Gag fractions. In summary, these results indicate that HIV-1 Gag can be monoubiquitinated in all domains and that ubiquitination of lysine residues outside p6 may thus contribute to viral release and/or infectivity.     

Positionally independent and exchangeable late budding functions of the Rous sarcoma virus and human immunodeficiency virus Gag proteins.

The Gag proteins of Rous sarcoma virus and human immunodeficiency virus (HIV) each contain a function involved in a late step in budding, defects in which result in the accumulation of these molecules at the plasma membrane. In the Rous sarcoma virus Gag protein (Pr76gag), this assembly domain is associated with a PPPY motif, which is located at an internal position between the MA and CA sequences. This motif is not contained anywhere within the HIV Gag protein (Pr55gag), and the MA sequence is linked directly to CA. Instead, a late assembly function of HIV has been associated with the p6 sequence situated at the C terminus of Gag. Here we demonstrate the remarkable finding that the late assembly domains from these two unrelated Gag proteins are exchangeable between retroviruses and can function in a positionally independent manner.     

Ubiquitination of human immunodeficiency virus type 1 Gag is highly dependent on Gag membrane association

Ubiquitin is important for the release of human immunodeficiency virus 1 (HIV-1) and several other retroviruses. All major domains of the HIV-1 Gag protein are monoubiquitinated, but the modifying machinery and the function of HIV-1 Gag ubiquitination remain unclear. Here, we show that the induction of a late budding arrest by mutation of the HIV-1 PTAP motif or by specific inhibition of selected ESCRT components leads to an increase of Gag-ubiquitin conjugates in cells, which coincides with an accumulation of detergent-insoluble, multimerized Gag at the plasma membrane. Membrane flotation experiments revealed that ubiquitinated Gag is highly enriched in membrane-bound fractions. Based on these findings, we propose that a blocking of virus release results in increased Gag ubiquitination as a consequence of its prolonged membrane association. Consistent with this, ubiquitination of a membrane-binding-defective (G2A)Gag mutant was dramatically reduced and the ubiquitination levels of truncated Gag proteins correlated with their abilities to bind to membranes. We therefore propose that membrane association and multimerization of HIV-1 Gag proteins, rather than a specific motif within Gag, trigger recognition by the cellular ubiquitination machinery.     

Transport and processing of the Rous sarcoma virus Gag protein in the endoplasmic reticulum.

The Gag proteins of replication-competent retroviruses direct budding at the plasma membrane and are cleaved by the viral protease (PR) just before or very soon after particle release. In contrast, defective retroviruses that bud into the endoplasmic reticulum (ER) have been found, and morphologically these appear to contain uncleaved Gag proteins. From this, it has been proposed that activation of PR may depend upon a host factor found only at the plasma membrane. However, if Gag proteins were cleaved by PR before the particle could pinch off the ER membrane, then the only particles that would remain visible are those that packaged smaller-than-normal amounts of PR, and these would have an immature morphology. To distinguish between these two hypotheses, we made use of the Rous sarcoma virus (RSV) Gag protein, the PR of RSV IS included on each Gag molecule. To target Gag to the ER, a signal peptide was installed at its amino terminus in place of the plasma membrane-binding domain. An intervening, hydrophobic, transmembrane anchor was included to keep Gag extended into the cytoplasm. We found that PR-mediated processing occurred, although the cleavage products were rapidly degraded. When the anchor was removed, allowing the entire protein to be inserted into the lumen of the ER, Gag processing occurred with a high level of efficiency, and the cleavage products were quite stable. Thus, PR activation does not require targeting of Gag molecules to the plasma membrane. Unexpectedly, molecules lacking the transmembrane anchor were rapidly secreted from the cell in a nonmembrane-enclosed form and in a manner that was very sensitive to brefeldin A and monensin. In contrast, the wild-type RSV and Moloney murine leukemia virus Gag proteins were completely insensitive to these inhibitors, suggesting that the normal mechanism of transport to the plasma membrane does not require interactions with the secretory pathway.     

Role of Nedd4 and ubiquitination of Rous sarcoma virus Gag in budding of virus-like particles from cells

Rous sarcoma virus (RSV) budding requires an interaction of the L domain within the p2b region of Gag with cellular Nedd4-family E3 ubiquitin protein ligases. Members of our laboratories previously demonstrated that overexpression of a fragment of the chicken Nedd4-like protein (LDI-1 WW) inhibits Gag release in a dominant-negative manner (A. Kikonyogo, F. Bouamr, M. L. Vana, Y. Xiang, A. Aiyar, C. Carter, and J. Leis, Proc. Natl. Acad. Sci. USA 98:11199-11204, 2001). We have now identified the complete 3' end of LDI-1 and determined that it has a C-terminal ubiquitin ligase HECT domain, similar to other Nedd4 family members. While overexpression of the full-length LDI-1 clone (LDI-1 FL) had little effect on Gag budding, an LDI-1 FL mutant with a substitution in the HECT domain catalytic site blocked Gag release, similar to LDI-1 WW. The coexpression of Gag and hemagglutinin-tagged ubiquitin (HA-Ub) resulted in the detection of mono- and polyubiquitinated forms of Gag in cells and mostly monoubiquitinated Gag in virus-like particles (VLPs). When the Nedd4-binding site (L domain) was deleted, ubiquitinated Gag was not detected. Interestingly, the release of Gag with ubiquitin covalently linked to the C terminus (Gag-Ub) was still blocked by LDI-1 WW. To understand the mechanism of this inhibition, we examined cells expressing Gag and LDI-1 WW by electron microscopy. In the presence of LDI-1 WW, VLPs were found in electron-dense inclusion bodies in the cytoplasm of transfected cells. In contrast, when cells that coexpressed Gag-Ub and LDI-1 WW were examined, inclusion bodies were detected but did not contain VLPs. These results indicate that the ubiquitination of Gag is dependent upon Nedd4 binding to the L domain and suggest that Nedd4 has additional functions during RSV release besides the ubiquitination of Gag.     

The PPPY motif of human T-cell leukemia virus type 1 Gag protein is required early in the budding process

Domains required late in the virus budding process (L domains) have been identified in the Gag proteins of a number of retroviruses. Here we show that the human T-cell leukemia virus type 1 candidate L domain motif PPPY is indeed required for virus production. Strikingly, however, mutation of this motif arrested virus particles at an earlier stage in the budding process than was seen for mutation of the L domain motifs thus far described for retroviruses. In view of the exchangeability of such domains, we propose that the retrovirus budding process may involve a continuum from bud formation to membrane fission.     

Rous sarcoma virus gag has no specific requirement for phosphatidylinositol-(4,5)-bisphosphate for plasma membrane association in vivo or for liposome interaction in vitro

The MA domain of the retroviral Gag protein mediates interactions with the plasma membrane, which is the site of productive virus release. HIV-1 MA has a phosphatidylinositol-(4,5)-bisphosphate [PI(4,5)P₂] binding pocket; depletion of this phospholipid from the plasma membrane compromises Gag membrane association and virus budding. We used multiple methods to examine the possible role of PI(4,5)P₂ in Gag-membrane interaction of the alpharetrovirus Rous sarcoma virus (RSV). In contrast to HIV-1, which was tested in parallel, neither membrane localization of RSV Gag-GFP nor release of virus-like particles was affected by phosphatase-mediated depletion of PI(4,5)P₂ in transfected avian cells. In liposome flotation experiments, RSV Gag required acidic lipids for binding but showed no specificity for PI(4,5)P₂. Mono-, di-, and triphosphorylated phosphatidylinositol phosphate (PIP) species as well as high concentrations of phosphatidylserine (PS) supported similar levels of flotation. A mutation that increases the overall charge of RSV MA also enhanced Gag membrane binding. Contrary to previous reports, we found that high concentrations of PS, in the absence of PIPs, also strongly promoted HIV-1 Gag flotation. Taken together, we interpret these results to mean that RSV Gag membrane association is driven by electrostatic interactions and not by any specific association with PI(4,5)P₂.     

Late domain function identified in the vesicular stomatitis virus M protein by use of rhabdovirus-retrovirus chimeras

Little is known about the mechanisms used by enveloped viruses to separate themselves from the cell surface at the final step of budding. However, small sequences in the Gag proteins of several retroviruses (L domains) have been implicated in this process. A sequence has been identified in the M proteins of rhabdoviruses that closely resembles the PPPPY motif in the L domain of Rous sarcoma virus (RSV), an avian retrovirus. To evaluate whether the PPPY sequence in vesicular stomatitis virus (VSV) M protein has an activity analogous to that of the retroviral sequence, M-Gag chimeras were characterized. The N-terminal 74 amino acids of the VSV (Indiana) M protein, including the PPPY motif, was able to replace the L domain of RSV Gag and allow the assembly and release of virus-like particles. Alanine substitutions in the VSV PPPY motif severely compromised the budding activity of this hybrid protein but not that of another chimera which also contained the RSV PPPPY sequence. We conclude that this VSV sequence is functionally homologous to the RSV L domain in promoting virus particle release, making this the first example of such an activity in a virus other than a retrovirus. Both the RSV and VSV motifs have been shown to interact in vitro with certain cellular proteins that contain a WW interaction module, suggesting that the L domains are sites of interaction with unknown host machinery involved in virus release.     

The matrix protein of human immunodeficiency virus type 1 is required for incorporation of viral envelope protein into mature virions.

Accumulating evidence suggests that the matrix (MA) protein of retroviruses plays a key role in virus assembly by directing the intracellular transport and membrane association of the Gag polyprotein. In this report, we show that the MA protein of human immunodeficiency virus type 1 is also critical for the incorporation of viral Env proteins into mature virions. Several deletions introduced in the MA domain (p17) of human immunodeficiency virus type 1 Gag polyprotein did not greatly affect the synthesis and processing of the Gag polyprotein or the formation of virions. Analysis of the viral proteins revealed normal levels of Gag and Pol proteins in these mutant virions, but the Env proteins, gp120 and gp41, were hardly detectable in the mutant virions. Our data suggest that an interaction between the viral Env protein and the MA domain of the Gag polyprotein is required for the selective incorporation of Env proteins during virus assembly. Such an interaction appears to be very sensitive to conformational changes in the MA domain, as five small deletions in two separate regions of p17 equally inhibited viral Env protein incorporation. Mutant viruses were not infectious in T cells. When mutant and wild-type DNAs were cotransfected into T cells, the replication of wild-type virus was also hindered. These results suggest that the incorporation of viral Env protein is a critical step for replication of retroviruses and can be a target for the design of antiviral strategies.