An Endoplasmic Reticulum Retrieval Signal Partitions Human Foamy Virus Maturation to Intracytoplasmic Membranes

Abstract

Among all retroviruses, foamy viruses (FVs) are unique in that they regularly mature at intracytoplasmic membranes. The envelope glycoprotein of FV encodes an endoplasmic reticulum (ER) retrieval signal, the dilysine motif (KKXX), that functions to localize the human FV (HFV) glycoprotein to the ER. This study analyzed the function of the dilysine motif in the context of infectious molecular clones of HFV that encoded mutations in the dilysine motif. Electron microscopy (EM) demonstrated virion budding both intracytoplasmically and at the plasma membrane for the wild-type and mutant viruses. Additionally, mutant viruses retained their infectivity, but viruses lacking the dilysine signal budded at the plasma membrane to a greater extent than did wild-type viruses. Interestingly, this relative increase in budding across the plasma membrane did not increase the overall release of viral particles into cell culture media as measured by protein levels in viral pellets or infectious virus titers. We conclude that the dilysine motif of HFV imposes a partial restriction on the site of viral maturation but is not necessary for viral infectivity.

All retroviruses encode for type 1 transmembrane glycoproteins translated at the surface of the rough endoplasmic reticulum (ER) as a polyprotein precursor. For the majority of retroviruses, the newly synthesized glycoproteins are transported from the ER through the Golgi complex and to the plasma membrane. Glycoprotein folding, carbohydrate modification, and cleavage into surface (SU) and transmembrane (TM) subunits occur during transit to the cell surface. These glycoprotein modifications are necessary for viral replication and infectivity (for a review, see reference 15).

Foamy viruses (FVs), or spumaviruses, exhibit several unusual characteristics compared to all the other known members of the retrovirus family (33, 34). Furthermore, recent studies suggested that FVs utilize assembly strategies much more reminiscent of hepadnaviruses. In addition to similarities seen for the internal structural proteins and the SU proteins (7, 12, 20a, 37, 41, 44, 45), both FVs and hepadnaviruses bud predominantly intracellularly, both encode for an ER retrieval signal in their glycoproteins, and both require SU glycoproteins for viral release from the cell (1, 4, 8, 12, 13, 18, 29). For other retroviruses, intracytoplasmic budding is an unusual event; Gag alone is able to direct particle budding in the absence of Env (10, 38, 42); and a putative ER retrieval signal has not been identified for any other retroviral glycoprotein.

Since FVs were known to bud into the ER, we hypothesized that their glycoproteins must contain an ER sorting signal (13). The dilysine motif (KKXX) was one such signal known to function for ER retrieval of type 1 transmembrane glycoproteins (17). Nevertheless, it was surprising that a dilysine motif was encoded by the TM glycoprotein of all sequenced FVs (13, 41, 43). The dilysine motif consists of lysines at position −3 and either position −4 or −5 with respect to the carboxy terminus of the glycoprotein (17). Glycoproteins containing the dilysine signal resist bulk transport to the plasma membrane and are retrieved from post-ER compartments to the ER by cytosolic coat proteins (5, 20).

We previously reported that recombinant human FV (HFV) glycoproteins were sorted to the ER, a function that was mediated by the dilysine motif (12). In the present study, we attempted to define the role of the dilysine motif in the replication of FVs. By constructing infectious clones of HFV containing mutations in the dilysine motif, we found that mutant viruses matured at the plasma membrane to a greater extent than wild-type virus.

MATERIALS AND METHODS

Cells, transfections, and virus stock preparation.

Canine thymocytes (CF2Th cells) (kindly provided by T. Folks), baby hamster kidney (BHK-21) cells, and FV-activated β-gal (FAB) cells (kindly provided by M. Linial; see below for description) were maintained in Dulbecco’s modified Eagle medium (DMEM) supplemented with 5 to 10% fetal bovine serum. Transfections were performed by the LipofectAmine method (Gibco Life Technologies, Grand Island, N.Y.) according to the manufacturer’s instructions. Seven days after transfection of BHK-21 cells, cells in medium were scraped from the culture dishes. The cells and the media underwent three cycles of freezing and thawing in order to release the virus. The viral titers were then determined by the FAB assay as described below.

Constructs of HFV DNA.

Construction of pSVL-KKS and pSVL-SSS has been previously described (12). These two env constructs encode for HFV envelope glycoproteins containing mutations to the ER retrieval signal under the control of a simian virus 40 promoter. All mutations were confirmed by dideoxynucleotide sequencing of the mutagenesis plasmid pALTER and within pSVL (36). To construct mutant full-length HFV clones, a 1,032-bp SpeI restriction fragment was isolated from pHSRV1-Mod (a wild-type HFV DNA clone that was provided by A. Rethwilm) and ligated into the SpeI restriction site of pSVL-KKS or pSVL-SSS. A 3,489-bp EcoRI restriction fragment from either pSVL-KKS or pSVL-SSS was then exchanged with the 3,489-bp EcoRI restriction fragment of pHSRV1-Mod to produce the mutant HFV DNA plasmids, pMod-KKS or pMod-SSS. Construction of pMod-RRR infectious clone was performed as follows: the PCR primers 5′-ACTGCCCAAGGAATATTTGGAACTGCCTT and 5′-ACGGGATCCGAATTCCAGAGGTGGAGGCTACTGATTCCTCCTTCTCGTAGG contained XcmI or BamHI restriction sites, respectively, with the latter primer predicted to encode three arginines (RRR) in place of the three lysines (KKK) of the ER retrieval signal. A pSP72 plasmid containing the wild-type env of HFV (11) was PCR amplified by using the above primers. The products were digested with XcmI and BamHI and cloned back into the same sites of plasmid pSP72. An EcoRI restriction fragment of this plasmid was then exchanged with pHSRV1-Mod as above to produce pMod-RRR. Virus stocks prepared after transfections with the plasmids were designated wild-type HFV, HFV-KKS, HFV-SSS, and HFV-RRR.

RIPA and Western blot assays.

The radioimmunoprecipitation assay (RIPA) was performed as previously described (11, 12, 26–28). Briefly, CF2Th or BHK-21 cells were labeled with [³⁵S]cysteine and [³⁵S]methionine 3.5 days after infection (multiplicity of infection [MOI] of 0.5 or 1). After radiolabeling and chase, the cells were lysed, and the medium was filtered through a 0.45-μm-pore-size filter and centrifuged through a 20% sucrose cushion. The cell lysates or pelleted material were immunoprecipitated with chimpanzee serum and protein A-agarose at 4°C overnight. Adult chimpanzees have a high prevalence of infection with FVs (14, 30), and all chimpanzee plasma samples used in these experiments were tested by RIPA to be certain that they reacted with the envelope and Gag proteins of HFV (11). The agarose beads were then washed and boiled in sodium dodecyl sulfate (SDS) for 5 min, and the immunoprecipitated proteins were characterized by SDS-polyacrylamide gel electrophoresis (PAGE) (19) on a 10% gel and autoradiography. Western blot assays were performed by harvesting cell lysate or pelleted virus as described above. After separation by SDS-PAGE on a 10 or 12% gel, the proteins were transferred onto a nitrocellulose membrane. Western blotting was performed using the Amersham ECL (enhanced chemiluminescence) detection system (Amersham Life Science, Arlington Heights, Ill.). Chimpanzee plasma and anti-human immunoglobulin G conjugated to horseradish peroxidase were used for antibody reagents.

Reverse transcriptase (RT) assay.

BHK-21 cells were infected at day 0 (MOI of 0.5) with wild-type HFV, HFV-SSS, or HFV-RRR. At the indicated times, 100 μl of supernatant was transferred from each of the infected cell cultures onto a 96-well plate; 75 μl of reaction cocktail {1 M Tris HCl (pH 8.0), 1 M dithiothreitol, 1 M Mn²⁺, 1 M KCl, 10% Triton X-100, 100 mM EGTA (pH 8.0), 1 μg of poly(dA-dT) per μl, 5 μCi [³⁵S]TTP} was added to the wells, and the solution was incubated at 37°C for 90 min. The reaction was stopped by adding 50 μl of 200 mM sodium pyrophosphate. Samples were adsorbed onto NA45 filter paper and air dried. After the filter was washed twice in 0.5 M NaPO₄, counts were measured with the AMBIS radioanalytic detection system (AMBIS Systems, Inc., San Diego, Calif.).

Infectivity assay.

FAB cells contain the β-galactosidase gene under the control of the HFV long terminal repeat; only cells infected with HFV will express the transactivating protein (Bel-1 or Tas) necessary for expression of β-galactosidase. BHK-21 cells were infected (MOI of 0.5) at day 0 with wild-type HFV, HFV-KKS, or HFV-RRR; cell supernatants were collected at 2, 4, or 6 days after infection. The FAB assay was performed as previously described (46). At the indicated times, cells were incubated with dilutions of virus (resuspended in DMEM) for 2 h at 37°C; the virus was removed and replaced with DMEM containing 5% fetal bovine serum. Two days after infection, the cell monolayers were fixed (0.5% glutaraldehyde in phosphate-buffered saline [PBS]) for 5 min and washed three times with PBS. The cells were then incubated 4 to 16 h at 37°C with staining solution (4 mM potassium ferrocyanide, 4 mM potassium ferricyanide, 2 mM MgCl₂, 0.4 mg of 5-bromo-4-chloro-3-indolyl-β-d-galactopyranosidase [X-gal] per ml). After the staining solution was removed and the cells were washed with PBS, the number of blue cells per well was determined under a light microscope.

EM.

CF2Th cells were infected with wild-type HFV, HFV-KKS, or HFV-SSS. Four days after infection, the culture media were removed and the cells were washed twice with PBS. The cells were subsequently fixed (1% glutaraldehyde) for 1 h and stored in PBS solution for analysis by electron microscopy (EM).

Sequencing of viral RNA.

BHK-21 cells were infected with wild-type HFV, HFV-KKS, or HFV-SSS. Virus was harvested from infected cells after 7 days and used to infect fresh cells. After nine passages (9 weeks), total mRNA was isolated with Tri-Reagent (Molecular Research Center, Inc., Cincinnati, Ohio) and reverse transcribed with Superscript RT (Gibco BRL). The resulting DNA was then amplified by PCR using primers (5′-GCTGAGCTCCTTCGACTGG-3′ and 5′-CTCATTTCCTCTGGTGTGG-3′) that flank the nucleotides encoding for the ER retrieval signal. The products were gel purified and sequenced with a Perkin Elmer-Applied Biosystems model 377 automated sequencer.

RESULTS

The ER retrieval signal directed HFV budding to intracellular membranes.

We previously demonstrated that recombinant HFV glycoproteins containing dilysine motif mutations were expressed to a greater degree at the cell surface (3, 12). For several enveloped viruses, including human immunodeficiency virus type 1, (HIV-1), the membrane site of glycoprotein localization determined the site of viral budding (21, 31, 39). We hypothesized that infectious clones lacking a complete ER retrieval signal in the glycoprotein should bud preferentially from the plasma membrane. To investigate the role of the dilysine motif in viral budding, we engineered infectious molecular clones of HFV which encode dilysine motif mutations (Fig. 1). Because the lysine at position −3 is essential for efficient ER retrieval of the glycoprotein (12), we substituted an env gene that encoded a mutation of the lysine at −3 (KKS) into the infectious wild-type clone pHSRV1-Mod (Fig. 1). Additionally, since the three positively charged lysines at the carboxy terminus of the HFV envelope glycoprotein may be important for charge interactions with other viral proteins (9), we constructed additional mutant viral clones carrying replacements of all three lysines with either serines (SSS) or arginines (RRR). The resultant mutant plasmid clones were designated HFV-KKS, HFV-SSS, and HFV-RRR (Fig. 1).

FIG. 1.

Mutations to the ER retrieval signal. The dilysine motif located in the cytoplasmic tail of the TM glycoprotein of HFV is shown (boldface KKK) along with the clones containing partially (HFV-KKS) or fully (HFV-SSS and HFV-RRR) mutated ER retrieval signals. LTR, long terminal repeat; PR, protease; WT, wild type.

CF2Th cells were infected with wild-type HFV, HFV-KKS, or HFV-SSS. Four days postinfection, cells were fixed and studied by EM (Fig. 2). While viral budding at the plasma membrane was occasionally seen for wild-type virus (Fig. 2A and D), the majority of wild-type virus was seen inside the cell (Fig. 2D). More virus was budding from or present at the plasma membrane in cells infected with HFV-KKS or HFV-SSS than in cells infected with wild-type HFV (Fig. 2B, C, E, and F). By visual estimation, approximately three times as much mutant virus budded at the plasma membrane. A significant amount of viral budding continued to occur intracytoplasmically for both mutants (Fig. 2E and F), but less intracytoplasmic budding was seen compared with wild-type virus. Similar results were obtained with cells infected with HFV-RRR, and the same phenotype was noted in both BHK-21 and human embryonic lung (HEL) cells (data not shown). Taken together, our results indicated that the dilysine ER retrieval signal of the FV envelope glycoprotein played a significant role in partitioning the maturation of FV particles to intracytoplasmic membranes. However other factors or signals must also participate in the partitioning of viral maturation since intracytoplasmic budding continued in the absence of the dilysine motif ER retrieval signal.

FIG. 2.

The ER retrieval signal partitioned viral budding away from the plasma membrane. CF2Th cells were infected with either wild-type HFV (A and D), HFV-KKS (B and E), or HFV-SSS (C and F) and then fixed and sectioned for EM analysis.

The ER retrieval signal did not impede viral exit from the cell.

Because the dilysine mutant clones had shifted part of their viral maturation to the plasma membrane, we expected their yield of extracellular virus would be greater than for wild-type HFV. To address this issue, CF2Th cells were infected at an MOI of 1 with wild-type HFV, HFV-KKS, or HFV-SSS. At 3.5 days after infection, cells were metabolically labeled for 1 h and chased for 12 h, and the media and cells were harvested. To visualize virus-specific proteins present within extracellular virus particles, the media were filtered through a 0.45-μm-pore-size filter and centrifuged through a 20% sucrose cushion. After immunoprecipitation, the proteins were resolved by SDS-PAGE. The levels of pellet-associated viral proteins (SU, TM, and Gag) were observed to be similar for wild-type and mutant viruses (Fig. 3; compare lanes 6 to lanes 7 and 8). The amount of processed glycoprotein (gp80) was decreased for the cell lysate fraction of wild-type infected cells (compare lane 2 with lanes 3 and 4). A slight decrease in the amount of pelleted virus was noted for HFV-SSS; however, this difference was neither significant nor consistent among experiments. In contrast to the cell-associated fraction, we noted that mature glycoprotein subunits (as determined by gp80 or gp47) were present in the released viral particles of wild-type as well as ER retrieval mutant viruses (Fig. 3, lanes 6 to 8). To ascertain that these results are not limited to a particular cell type, the RIPA was repeated with BHK-21 cells. The media were collected 4 days after infection and pelleted through 20% sucrose cushion. The amounts of extracellular virus were not significantly different for wild-type HFV, HFV-KKS, and HFV-SSS (Fig. 4A, lanes 6 to 8). Similar to the experiment with CTh2 cells, decreased amounts of glycoprotein cleavage were seen in wild-type HFV-infected cells (Fig. 4A; compare lane 2 to lanes 3 and 4. Infection of HEL cells yielded similar results.

FIG. 3.

The ER retrieval signal did not inhibit production of extracellular virus. CF2Th cells were infected with wild-type HFV (lanes 2 and 6), HFV-KKS (lanes 3 and 7), or HFV-SSS (lanes 4 and 8); 3.5 days after infection, cells were radiolabeled for 1 h, chased for 12 h, and immunoprecipitated with chimpanzee plasma. Lanes 1 to 4, cell lysates; lanes 5 to 8, media; lanes 1 and 5, mock infection. Media were pelleted through a 20% sucrose cushion. Proteins were separated by SDS-PAGE (12% gel).

FIG. 4.

Similar levels of extracellular virus were released from a different cell line, as visualized by RIPA and Western blot analysis. BHK-21 cells were infected with wild-type (lanes 2 and 6), HFV-KKS (lanes 3 and 7), or HFV-SSS (lanes 4 and 8). Cells were labeled overnight for RIPA (A) or harvested for Western blot analysis (B). The medium was pelleted through a 20% sucrose cushion, and proteins were separated by SDS-PAGE (10% gel). Lanes 1 and 5, mock infection.

Labeling experiments may not be an accurate measure of the total amount of extracellular virus production, a consideration especially significant with respect to a mutation that may disrupt the normal replication kinetics of the virus. For example, a mutant virus might yield an increased level of total extracellular virus but have a longer time to viral release. In such a case, the RIPA would underestimate the actual amount of released virus. Western blot analysis, which is not dependent on radiolabeling, was therefore performed to evaluate the total amount of Gag present in the viral particles released from BHK-21 infected with wild-type HFV, HFV-KKS, or HFV-SSS. Four days after infection, media were filtered and centrifuged through a sucrose cushion as described above, and Western blotting was performed. The amounts of Gag (p74/70) present in released viral particles were again similar for mutant and wild-type viruses (Fig. 4B; compare lane 6 to lanes 7 and 8). The HFV surface glycoproteins were poorly recognized by the chimpanzee plasma in the Western blot, and only the precursor (gp130) was detected. Results for HEL cells were similar. As indicated by both radiolabeling and Western blot assays using three different cell types, the amounts of pellet-associated viral proteins for wild-type and mutant viruses were not different. Taken together, these results showed that although electron micrographs indicated the ER retrieval signal of wild-type HFV directed viral maturation away from the plasma membrane, this did not reduce the total level of extracellular virus as measured by levels of pellet-associated viral proteins.

The ER retrieval signal was not required for HFV infectivity.

Although in the above experiments extracellular viral particles were detected in the media of cells infected with HFV clones containing dilysine motif mutations, it remained uncertain if these released viral particles represented infectious virus. The ER retrieval mutants HFV-SSS and HFV-RRR were used in this experiment to determine if the charged residues in the glycoprotein cytoplasmic domain play a role in Gag-Env interaction. If these charges are important for Gag-Env interactions in FVs, then their disruption should significantly decrease viral replication. To address this question, BHK-21 cells were infected at day 0 (MOI of 0.5), and the media were collected and filtered through a 0.45-μm-pore-size filter at 2, 4, or 6 days after infection. Serial dilutions of the virus-containing media were inoculated onto FAB cells. Because FAB cells encode β-galactosidase under the control of an HFV long terminal repeat promoter, infection with HFV results in expression of β-galactosidase, which degrades X-Gal to produce blue cells as detected by light microscopy. Two days after inoculation, the FAB cells were fixed and stained with X-Gal, and the blue cells were counted. The mutant viruses HFV-SSS and HFV-RRR were infectious, albeit at slightly (<1 log) lower titers than wild-type HFV (Fig. 5A). The replication kinetics of the mutant and wild-type viruses were generally similar, which demonstrated a peak of virus production at 4 days after infection and a slight decrease by day 6. However, there appeared to be a slight delay in replication for the mutants. At day 6, the titer of mutant HFV-RRR was still increasing slightly, while the mutant HFV-SSS titer had just reached a plateau (Fig. 5A). HEL cells exhibited similar results, indicating that these findings were not limited to a particular cell type (not shown). Assays of RT activity in the media of infected BHK-21 cells again demonstrated similar infectivity profiles for wild-type and mutant viruses, although the levels were increasing for all viruses at day 6 (Fig. 5B). Taken together, the results indicated that the ER retrieval signal, though required for optimal viral replication, was not absolutely necessary for HFV infectivity. Additionally, it did not appear that the positive charges found in the ER retrieval signal played a role in virus replication kinetics, as the mutants HFV-SSS and HFV-RRR (both lacking the ER retrieval signal) demonstrated similar replication kinetics.

FIG. 5.

Viral replication kinetics were similar for wild-type and mutant clones. BHK-21 cells were infected with wild-type (WT) or mutant viruses, and the media were collected at the indicated times. The media were inoculated onto FAB cells to measure viral infectivity by numbers of blue cells (A) or by quantification of RT activity in the media (B).

Because of the slight delay in mutant virus replication (Fig. 5A), we hypothesized that disruption of the ER retrieval signal had created a disadvantage for viral replication under cell culture conditions. Mutant viruses may therefore be under pressure to revert to wild type for more efficient growth. To identify possible revertants, BHK-21 cells were infected with HFV-KKS or HFV-SSS and passaged in cell culture for a 9-week period. Viral RNA was subsequently harvested, reverse transcribed to DNA, amplified by PCR, cloned, and sequenced. Neither mutant had generated any nucleotide changes in the region of the ER retrieval signal. Furthermore, even after 9 weeks of cell passage, RIPA of BHK-21 cells infected with the mutant viruses maintained glycoprotein cleavage and electrophoretic mobility characteristic of the mutant glycoprotein (gp80) (not shown). The same phenotype as seen with viruses that were minimally passaged prior to infection of BHK-21 cells was observed (Fig. 4A, lanes 2 to 4). This latter result did not rule out the possibility of second-site mutations; however, the mutant phenotype was still detected after 9 weeks of viral passage, making second-site mutations less likely to have occurred.

DISCUSSION

Based on knowledge of certain other viruses, we surmised that the envelope glycoprotein of HFV would play a significant role in determining the site of HFV budding. On the other hand, redirecting the glycoprotein of HIV-1 to the ER did not change the budding site of viral particles (32, 35). In the present study, EM indicated that the ER retrieval signal of HFV played an important role in establishing the intracellular site of budding; however, other determinants diverting HFV budding to an intracellular site must also exist since budding across intracellular membranes continued to be observed for mutant viruses (Fig. 2E and F). Unexpectedly, metabolic labeling and Western blot analysis of viral proteins did not demonstrate an expected increase in the levels of HFV particles in the media of cells infected with ER retrieval signal mutant viruses. These results were confirmed in multiple experiments using different cell lines. Cotransfection of plasmids encoding for HFV Gag and either wild-type or dilysine motif mutant glycoproteins also indicated that wild-type and mutant particles were present in the media at similar amounts (3). Therefore the results obtained were not dependent on different efficiencies of viral infectivity since these coexpression experiments did not produce infectious virus.

It is not clear why changing the major site of HFV budding to the plasma membrane did not increase the amount of extracellular HFV. Although viral budding occurs at the ER membranes, transit through the cell via the Golgi apparatus and release at the plasma membrane may continue to take place. Despite the fact that wild-type HFV glycoprotein precursor gp130 was not efficiently cleaved when expressed alone (12), wild-type HFV particles collected from the media of infected cell cultures contained mostly the proteolytically processed subunits gp80 and gp47 (Fig. 3 and 4A, lane 6 in both). Thus, after wild-type HFV buds into the ER, in order for viral glycoprotein cleavage to occur, the virus must transit through the secretory pathway to access the cellular subtilisin-like proteases in the distal Golgi apparatus. Furthermore, virus released by cell lysis without transit through the Golgi apparatus would presumably result in particles containing the uncleaved precursor gp130, but such particles have not been described. Finally, FVs produced a noncytopathic infection in certain lymphocyte cell lines (24, 25, 47), also shown to be the site of persistent viral infection in vivo (40), suggesting that FVs have developed mechanisms for exiting the cell without causing cytopathology.

The reason why FVs but no other retroviruses have evolved a highly conserved ER retrieval signal in their envelope glycoprotein remains unclear. While it was required for efficient intracellular budding of virus, it was unnecessary for the production of extracellular infectious virus. Furthermore, the signal was not required for in vitro replication of HFV, and only minor effects on the replication kinetics with ER retrieval mutants were observed. Potentially, the presence of the three lysines of the dilysine motif could facilitate Gag-Env assembly through important charge interactions. In addition, the coordinated interaction between Gag and Env may require that these proteins be colocalized to the site of viral maturation. However, these possibilities are not supported by our work demonstrating that neutralizing the entire charge of the glycoprotein’s cytoplasmic domain (HFV-SSS) did not result in a significant change in infectious virus titer, nor did it affect Gag incorporation into viral particles. Overall, our findings suggest that the ER retrieval signal does not play a significant role for the replication of FV in immortalized cell types during laboratory cultures. This conclusion was further supported by the absence of detectable revertant mutants after 9 weeks of mutant virus passage in such cells.

Other possible roles that the ER retrieval signal plays in vitro have not been studied. A protein sorting signal somewhat analogous to the dilysine motif may be seen with other retroviruses, where a tyrosine (located on the cytoplasmic tail of the TM glycoprotein) was shown to be important for decreasing the level of SU glycoprotein when expressed in the absence of other viral particles. This effect was partially abrogated when HIV-1 Gag was coexpressed with Env (6), a finding similar to our present and earlier results which demonstrated that when the HFV glycoprotein is expressed alone, most of it remains in the ER (12). The present study showed that when the glycoprotein is expressed as part of a complete viral infection, a significant portion of the glycoprotein exits the ER, suggesting a role for other viral proteins in facilitating glycoprotein transit through the Golgi network and beyond. It is also interesting that the same tyrosine signal was shown to be important for basolateral targeting of lentiviral particles (2, 22), again analogous to the present work, where the ER retrieval signal was shown to be important for efficient intracytoplasmic targeting of FV particles.

By decreasing the amount of SU glycoprotein expression, FVs may minimize the level not only of syncytium formation but also of apoptosis, which has been demonstrated with FV infections of cells (16, 23). Another possibility is that by directing the major site of assembly intracytoplasmically, FVs temporarily escape immune detection, thereby giving the virus an important advantage early in the replication cycle. FVs have undoubtedly devised several mechanisms of immune system evasion, though data on this question are scant. The conservation of the dilysine motif in FVs isolated from several different mammalian species indicates that this signal must play an extremely important role in the replication of these viruses. This paper describes some of its in vitro effects, but clearly more will be learned of this highly conserved motif.