The trans-Golgi network-associated human ubiquitin-protein ligase POSH is essential for HIV type 1 production

Abstract

HIV type 1 (HIV-1) was shown to assemble either at the plasma membrane or in the membrane of late endosomes. Now, we report an essential role for human ubiquitin ligase POSH (Plenty of SH3s; hPOSH), a trans-Golgi network-associated protein, in the targeting of HIV-1 to the plasma membrane. Small inhibitory RNA-mediated silencing of hPOSH ablates virus secretion and Gag plasma membrane localization. Reintroduction of native, but not a RING finger mutant, hPOSH restores virus release and Gag plasma membrane localization in hPOSH-depleted cells. Furthermore, expression of the RING finger mutant hPOSH inhibits virus release and induces accumulation of intracellular Gag in normal cells. Together, our results identify a previously undescribed step in HIV biogenesis and suggest a direct function for hPOSH-mediated ubiquitination in protein sorting at the trans-Golgi network. Consequently, hPOSH may be a useful host target for therapeutic intervention.

Release of the enveloped HIV type 1 (HIV-1) in most infected cells occurs by means of budding and subsequent pinching off of nascent virus particles from the plasma membrane. In HIV-1-infected macrophages, virus release proceeds through budding of nascent particles into the lumen of late endosomes and subsequent release of mature viruses by exocytosis (1, 2).

Virus egress is driven by the Gag viral polyprotein at the inner leaflet of the plasma membrane. Gag expression is sufficient to induce virus-like particle budding and release and, in the context of a full virus, to cause virus assembly and release (3). However, the mechanism that targets Gag to the plasma membrane is unknown.

Live-cell imaging of virus assembly detected Gag in late endosomes in a variety of cell types, including T cells, where HIV-1 egress occurs at the plasma membrane, suggesting that late endosomal membranes are viral assembly sites (4). In contrast, in another study, only mutant Gag was directed to late endosomes in HeLa and T cells (5). Consequently, it is unknown whether association of HIV-1 Gag with late endosomes is an obligatory step in subsequent movement to the cell membrane. Furthermore, analyses of murine leukemia virus- and Mason–Pfizer monkey virus-infected cells show association of Gag with late endosomal membranes and subsequent movement to the plasma membrane, the latter requiring a functional vesicular transport system (6, 7). It is therefore possible that retroviruses use different pathways to target Gag proteins to the plasma membrane.

Protein ubiquitination is a major regulator of intracellular protein transport. Ubiquitination regulates endocytosis and sorting of proteins into lysosomes (8) and recently was shown to serve as a signal for the targeting of proteins from the trans-Golgi network (TGN) to the vacuolar or to late endosomal/lysosomal compartments in yeast and higher eukaryotes, respectively (9–13).

A critical role for the TGN in HIV-1 biogenesis is indicated by recent findings on the incorporation of envelope glycoproteins (Env) into infectious HIV-1 particles. This process requires newly synthesized Env arriving at the plasma membrane to be endocytosed and subsequently routed back to the plasma membrane via the TGN. The retrograde transport and concomitant production of infectious particles is mediated by an interaction between the cytoplasmic tail of the gp41, the transmembrane domain of Env, and TIP47, a cytosolic protein that also is required for the recycling of cation-dependent mannose 6-phosphate receptors from late endosomes to the TGN (14, 15). Similarly, in Mason–Pfizer monkey virus-infected cells, recycled, rather than newly synthesized, Env incorporates into released viruses, yet Mason–Pfizer monkey virus-Gag does not colocalize with TGN markers (16). Therefore, it is possible that different retroviruses employ membranes of distinct intracellular organelles as assembly areas.

The TGN is a post-Golgi compartment that regulates protein trafficking pathways. For example, newly synthesized proteins traveling to the plasma membrane, to endosomes, or to specific membranes in polarized cells are sorted to their destination compartment at the TGN. Sorting at the TGN is facilitated by segregation of cargo into biochemically and functionally distinct membrane subdomains that subsequently pinch off from the TGN membrane and travel to the destination organelle. A constant recycling of membrane vesicles exiting and fusing with the TGN membrane allows maintenance of constant membrane volume and retrieval of essential sorting factors (17–19).

Because HIV-1 production requires the exploitation of the host protein sorting and trafficking pathways and in light of reports that ubiquitination is involved in the late steps of the HIV-1 life cycle (20–22), we postulated that an E3 ubiquitin ligase activity is involved in HIV biogenesis. In this report, we describe the identification of a TGN-associated E3 ubiquitin ligase, human POSH (hPOSH), a homologue of murine POSH (23, 24), as a critical factor for the sorting of HIV Gag to the plasma membrane and for HIV-1 biogenesis.

Materials and Methods

RNA interference (RNAi) sequences, cloning and mutagenesis procedures, generation of anti-hPOSH antibodies, generation of constitutively expressing HeLa and Jurkat cell lines, and assay for virus-like particle release are described in Supporting Materials and Methods, which is published as supporting information on the PNAS web site.

Single Cycle Infectivity Assay. HeLa SS6 cells were initially transfected with RNAi and split the next day. Then, 48 h after the initial transfection, cells were cotransfected with another portion of RNAi, with a plasmid encoding vesicular stomatitis virus G protein (VSV-G) and HIV1_NL4–3 env-, in which the Nef coding sequences is replaced with an EGFP reporter (29, 30). Next, 36 h after the cotransfection, virus-containing supernatants were clarified by centrifugation (1,000 × g for 5 min), passed through a 0.45-μM-poresize filter (Schleicher & Schuell), and serially diluted (in triplicates). The diluted virus stock then was used to infect target Jurkat cells. Three days after infection, the percentage of infected cells was determined by FACS analysis of EGFP-expressing cells.

In Vitro Ubiquitination Assays. Purified recombinant E1 (100 ng), UbcH5c (E2) (250 ng), and maltose-binding protein–hPOSH–His tag (400 ng) were incubated in a final volume of 20 μl containing 50 mM Hepes·NaOH (pH 7.5), 1 mM DTT, 2 mM ATP, 5 mM MgCl₂, and 2.5 μg of ubiquitin. After incubation for 30 min at 37°C, hPOSH was isolated by metal affinity chromatography on Ni-nitrilotriacetic acid resin (Qiagen, Valencia, CA) according to the manufacturer's instructions. The isolated hPOSH was subsequently resolved on a 7.5% SDS gel and subjected to Western blot analysis with antiubiquitin antibodies (Covance Research Products, Denver, PA).

When testing hPOSH-V5 from HeLa SS6 cells, hPOSH was isolated by immunoprecipitation with anti-V5 antibodies coupled to Sepharose beads (Invitrogen). Beads were washed twice with solubilization buffer and twice with 50 mM Hepes·NaOH (pH 7.5). The hPOSH-V5 subsequently was incubated for 30 min at 37°C in an ubiquitination reaction mixture as described above except that a mixture of ubiquitin (2.5 μg) and biotinlyated ubiquitin (0.3 μg) (Boston Biochem, Cambridge, MA) was used. Ubiquitinated proteins were separated on 7.5% SDS/PAGE and detected by immunoblot analysis with streptavidin–horseradish peroxidase (Dako-Cytomation, Glostrup, Denmark).

Results

hPOSH, a Homologue of Murine POSH, Is Essential for Production of Infectious HIV-1. In an effort to identify E3s regulating virus budding, HeLa cell cultures were transfected with RNAi targeting various candidate E3 ligases. In addition to hPOSH, we tested the silencing effect of the Nedd4 family members previously implicated in retrovirus budding (31). After significant reduction of mRNA expression (Fig. 7, which is published as supporting information on the PNAS web site), cells were transfected with pNLenv1, which encodes an Env-deficient subviral Gag-Pol expression system (26), and the steady-state levels of virus-like particles (VLP) released into the culture medium were determined by Western blot analysis. Of all of the tested E3s, only the hPOSH RNAi significantly inhibited VLP release (Fig. 1A; compare lane 1 with lanes 3–7).

Fig. 1.

Effect of hPOSH silencing on HIV-1 release. (A) Effect of E3 silencing on HIV release. RNAi-treated cells were cotransfected with HIV-1_NL4–3 env- and VSV-G expression plasmid. Cellular and VLP detergent extracts of cells transfected with either lamin A/C RNAi (control) or RNAi targeting the indicated E3 ligases were resolved by SDS/PAGE and subjected to immunoblot analysis with anti-CA as described in Materials and Methods. (B) Effect of hPOSH inhibition on the production of infectious virus. Cells were transfected with lamin A/C (control) or hPOSH RNAi. A single cycle infectivity assay was subsequently carried out as described in Materials and Methods. The percentage of infected cells was determined by FACS analysis and was plotted against the corresponding dilution of the initial viral stock. The results are the mean value of triplicate incubations. (C) SEM of HeLa SS6 cells. Cells initially were transfected with either hPOSH scrambled RNAi (a and c) or hPOSH RNAi (b and d) and then either mock transfected (c and d) or transfected with pNLenv1 p6^ATAA (a and b). Subsequently, cells were subjected to SEM as described in Supporting Materials and Methods.

Next, we compared production of infectious HIV-1 in the presence of normal or reduced levels of hPOSH during a single cycle of infection (27). Cells were pretreated either with hPOSH-specific or control RNAi and then transfected with an env-deleted HIV-1_NL4–3, DNA, and a plasmid encoding VSV-G. Because progeny HIV-1 viruses do not encode an envelope glycoprotein in their genome but, rather, incorporate envelope proteins (VSV-G) encoded by the helper plasmid, they allow a single round of infection. Because in this system infection cannot spread beyond one cycle, the number of infected cells upon incubation of progeny virus with naïve target cells is directly proportional to infectious virus titer. Thus, to determine virus titer in the RNAi-treated cells, an infectivity assay was performed by using serially diluted aliquots of the respective virus stocks. Determination of infected cells (Fig. 1B) indicated that undiluted stock from the control and hPOSH RNAi-treated cells gave infectivity values of 100% and 1%, respectively. A 1% infection rate was retained after a 1,000-fold dilution of the medium from control RNAi-treated cells, whereas a 100-fold dilution of the hPOSH-inhibited cells abolished infection. Thus, hPOSH silencing decreased the production of infectious virus by at least 2 orders of magnitude.

hPOSH Regulates Virus Production Independently of the Viral L-Domain and Upstream of Virus Budding at the Cell Membrane. Mutations in the HIV-1 late-domain (p6^Gag) PTAP motif confer a phenotype wherein HIV-1 particles remain tethered to the plasma membrane, fail to mature, and are characterized biochemically by the accumulation of a capsid (CA) processing intermediate p25 beside other Gag processing intermediates (32). Western blot analysis of Gag steady-state levels in cells expressing the pNLenv1 revealed that hPOSH silencing, although strongly inhibiting VLP release, did not cause accumulation of p25-CA in contrast to cells expressing a p6^Gag mutant version of pNLenv1 (p6^ATAAP) (Fig. 1A; compare lanes 2 and 5). This result indicated that the function of hPOSH in HIV-1 production was independent of the viral L-domain.

The apparent lack of efficient Gag processing defects upon inhibition of hPOSH expression prompted us to examine the effect of hPOSH silencing on budding of p6^ATAAP mutant virus by SEM. In control RNAi-treated cells, numerous cell surface-tethered virus particles were observed, consistent with inhibition of virus release (Fig. 1Ca). Pretreatment with hPOSH RNAi ablated accumulation of particles at the cell surface, confirming that hPOSH functioned independently of the virus L-domain and upstream of virus budding at the cell membrane (Fig. 1C; compare a and b). Notably, silencing of hPOSH expression did not have an adverse effect on the morphology of naïve cells (Fig. 1C; compare c and d).

Cloning of hPOSH. The complete coding sequence of hPOSH was deduced from two overlapping DNA sequences (GenBank accession nos. AK021429 and AB040927) and by comparison with the sequence of the mouse homologue (23).

The hPOSH cDNA encodes a predicted protein of 888 aa. Comparison of the human and murine POSH amino acid sequences reveals an overall identity of 90% (Fig. 8, which is published as supporting information on the PNAS web site). The protein encodes an amino-terminal RING-finger domain, four SH3 domains, and a region implicated in Rac binding (23) between the second and third SH3 domains (Fig. 8A). Immunoprecipitation of endogenous or exogenous hPOSH (see Figs. 8B and 4C, respectively) from cell extracts followed by Western blot analysis identified a protein of 110 kDa (as determined by SDS/PAGE), an apparent molecular weight that was higher than that calculated based on the hPOSH amino acid sequence (93 kDa). The level of the 110-kDa protein was significantly reduced after treatment with an hPOSH-specific RNAi (Fig. 9A, which is published as supporting information on the PNAS web site), confirming that it was hPOSH.

Fig. 4.

The ubiquitin ligase activity of hPOSH is required for HIV release. (A) Analysis of Gag-EGFP localization in HeLa cells. H187 cells were transfected with Gag-EGFP expression plasmid. (Upper) Then, 6 h after transfection, cells were transferred to 20°C for 2 h and then shifted back to 37°C for 5 h. (Lower) A parallel culture was maintained at 20°C for 7 h. Cells then were fixed and processed for confocal microscopy as described in Materials and Methods and in the Fig. 4 legend. (B) HIV-1 release at 37°C and 20°C. H187 cells were transfected with pNLenv1. Then, 12 h after transfection, the culture medium was replaced with fresh medium, and cells were incubated for 2 h at 20°C, after which the medium was replaced, and cells were further incubated for 5 h at 37°C. Control reactions were incubated in parallel at 20°C for 7 h. Subsequently, VLP were harvested, and aliquots were tested for reverse-transcriptase activity. The measured reverse-transcriptase activity was normalized to the relative Gag expression determined in parallel by quantitative anti-CA Western blot analysis using Cy3-conjugated secondary antibodies. (C) Cells were cotransfected with pNLenv-1 and expression plasmids for the indicated hPOSH proteins. Then, 12 h after transfection, cells were transferred to fresh medium at 20°C (2 h) and then to a fresh medium at 37°C (5 h). (Upper) Cells and VLP were processed, and reverse-transcriptase activity was determined as described in B.(Insets) Anti-V5 Western blot analysis of exogenous V5-hPOSH. (Lower) Quantitative analysis of Gag expression. Western blot analysis was performed with primary anti-CA antibody and a secondary Cy3-labeled anti-rabbit.

hPOSH Is an Ubiquitin-Protein E3 Ligase. The presence of a RING finger domain in hPOSH suggested that it might be an ubiquitin-protein ligase (E3) (33). The following three enzymes carry out covalent attachment of ubiquitin to target proteins: E1, the ubiquitin-activating enzyme; E2, an ubiquitin-conjugating enzyme; and an E3. The E3 serves the following two roles: it specifically recognizes ubiquitination substrates and simultaneously recruits an E2. Ligation of ubiquitin is initiated by the formation of an isopeptide bond between the carboxyl terminus of ubiquitin and an ε-amino group of a lysine residue on the target protein. Additional ubiquitin molecules can be further ligated to the initial ubiquitin molecule to form a polyubiquitinated protein (34). In the absence of an external substrate, E3s can catalyze self-ubiquitination; that is, transfer activated ubiquitin to a lysine side chain in the E3 polypeptide itself. Similar to transubiquitination, self-ubiquitination also depends on the action of E1 and an E2 (35).

When a bacterially expressed His-tagged hPOSH was incubated in vitro with E1, E2, ubiquitin, and ATP and then isolated by metal affinity chromatography, high molecular hPOSH-ubiquitin adducts were detected by antiubiquitin Western blot analysis only in the presence of a complete ubiquitin conjugation system. (Fig. 2A). We also tested the ubiquitination activity of a RING finger mutant hPOSH_V14A. The V14A mutation was inserted in a predicted E2–E3 interaction motif, Cys-x-(Val/Ile/Leu)-Cys (36–38). Indeed, when V5-epitope-tagged hPOSH and hPOSH_V14A were isolated from HeLa cells and subsequently incubated with E1, E2, ubiquitin-biotin, and ATP, high molecular weight ubiquitin chains were formed with wild-type (WT), but not mutant, hPOSH (Fig. 2B). The polyubiquitin chains migrating at a molecular weight lower than that of native hPOSH were identified by Western blot analysis with anti-V5 tag antibodies as truncated forms of hPOSH (data not shown), whereas the lower molecular weight bands in lanes 1 and 3 are most likely nonspecific, ubiquitinated proteins adhering to the beads. Together, these results confirm that hPOSH is an ubiquitin protein ligase.

Fig. 2.

hPOSH is an E3 ligase. (A) Self-ubiquitination of maltose-binding protein–POSH–his tag fusion protein was assayed as described in Materials and Methods. One of the ubiquitin-conjugating enzymes was omitted in each of the control reactions. Samples were resolved by SDS/PAGE and subjected to Western blot analysis with antiubiquitin antibodies. (B) Ubiquitination of exogenous hPOSH. Cells were transfected with empty vector, V5-tagged hPOSH, or hPOSH_V14A-encoding vectors. The tagged POSH then was immunoprecipitated with an anti-V5 antibody and incubated with E1, Ubch5c, ATP, and biotinylated ubiquitin as described in Materials and Methods. The reactions were stopped by the addition of SDS sample buffer, resolved by SDS/PAGE, transferred onto nitrocellulose, and blotted with streptavidin–horseradish peroxidase to detect ubiquitinated proteins.

hPOSH Is Associated with the TGN. Next, we determined by confocal microscopy the intracellular localization of hPOSH. To this end, HeLa H153 and H187 cells, stably expressing hPOSH-specific or control RNAi, respectively, were stained with hPOSH-specific antibody and antibodies directed to various cellular compartments. The results demonstrate that hPOSH colocalized with TGN46, an integral protein of the TGN membrane (39), but not with markers for early endosomes [Early Endosome Antigen 1 (EEA1)], late endosomes (CD63), lysosomes (Lamp1) (Fig. 3A Upper), the nuclear membrane (nucleoporin; data not shown) or the endoplasmic reticulum (calnexin; data not shown). Anti-hPOSH staining of the TGN was specific because no staining was observed in H153 cells (Fig. 3A Lower) and ectopically expressed hPOSH also localized at the TGN (see Fig. 5). The lack of putative membrane-spanning domains within the hPOSH amino acid sequence and the temperature dependence of hPOSH association with low-density membranes upon cell fractionation (data not shown) suggest that hPOSH is peripherally associated with the cytoplasmic face of the TGN membrane. The hPOSH localization was resistant to treatment with brefeldin A (data not shown). Because brefeldin A targets the Arf G-protein exchange factors (40), insensitivity to brefeldin A indicates that the binding of hPOSH to TGN membranes is Arf-independent. Importantly, hPOSH also localizes at the TGN in primary human lymphocytes and macrophages, the principle human target cells of HIV-1 (Fig. 3B).

Fig. 3.

Intracellular localization of hPOSH. (A) hPOSH localizes at the TGN. H153 and H187 cells stably expressing RNAi targeting the hPOSH coding sequence or a scrambled sequence, respectively, were incubated with rabbit anti-hPOSH (hP^285–430) and mouse monoclonal antibodies directed to Lamp1, EEA1, or CD63 and sheep anti-human TGN46. Subsequently, cells were immunostained for confocal microscopy with Alexa Fluor 488-conjugated goat anti-rabbit (hPOSH detection), Cy3-conjugated donkey anti-sheep (TGN46 detection), and Alexa Fluor 546-conjugated goat anti-mouse secondary antibodies for detection of Lamp1, EEA1, and CD63. (Bars: 20 μM.) (B) Primary human lymphocytes and macrophages (10⁵ cells) were subjected to confocal microscopy for detection of TGN and hPOSH as described above. (Bars: 10 μM.) (C) Inhibition of VLP secretion from Jurkat cells. J3′UTR1–3 and J187 cells stably expressing RNAi targeting the hPOSH 3′ UTR or a scrambled sequence, respectively, were transfected with pNLenv1. Subsequently, VLP were quantified through measurement of reverse-transcriptase activity secreted into the culture medium as described in Supporting Materials and Methods.

Fig. 5.

POSH regulates transport of Gag to the plasma membrane. The intracellular distribution of Gag-EGFP was determined in H187 (A) and H153 (B) by confocal microscopy. Cells initially were transfected with empty plasmid (mock) or a plasmid encoding either V5-tagged hPOSH or hPOSH_V14A and subsequently with Gag-EGFP-encoding plasmid. Then, 6 h later, cells were fixed and immunostained with anti-V5 and anti-TGN46 and with Alexa Fluor 633-conjugated goat anti-mouse and Cy3-conjugated donkey anti-sheep antibodies, respectively. (Bars: 20 μM.)

hPOSH Regulates HIV Release in Human Lymphocytes. To further investigate the physiological function of hPOSH, we tested the effect of hPOSH silencing on HIV-1 VLP secretion from human lymphocytes. To this end, several Jurkat clones constitutively expressing hPOSH RNAi, derived from the 3′ UTR of the POSH mRNA, were generated. The cell lines were transfected with pNLenv1, and VLP secretion subsequently was quantified. The results indicated reduction of VLP secretion in all hPOSH-depleted Jurkat cell-lines (Fig. 3C), supporting a physiologically relevant regulation of HIV-1 secretion by hPOSH.

Ubiquitin Ligase Activity of hPOSH Regulates TGN-to-Plasma Membrane Sorting of Gag. Recent evidence indicates that Gag is targeted to the plasma membrane by means of late endosomes (1). Our findings indicate that silencing of hPOSH inhibits budding at the cell surface and that hPOSH associates with the TGN, a sorting compartment for proteins destined for the plasma membrane or the endosomal/lysosomal compartment. We therefore tested the possibility that hPOSH controls HIV-1 production at the TGN by regulating Gag plasma membrane localization.

To investigate whether Gag localizes to the TGN before its arrival at the cell membrane, we used a method that allows a reversible, temperature-dependent block of vesicular trafficking in post-Golgi compartments (41). This method was used previously to measure TGN-to-plasma membrane export (42) and Mason–Pfizer monkey virus Gag trafficking (16). H187 cells were transfected with a plasmid encoding a fusion between Gag and EGFP and shortly thereafter were shifted to 20°C for 2 h. One culture was subsequently shifted to 37°C to allow transport, whereas a parallel culture was kept at the restrictive 20°C temperature throughout the experiment. Subsequent visualization of Gag-EGFP by confocal microscopy demonstrated exclusive staining of the TGN at 20°C (Fig. 4A Lower). A fraction of Gag was detected at the cell periphery only after further incubation at 37°C (Fig. 4A Upper). In a similar experiment, pNLenv1-transfected cells were unable to release VLP at 20°C, yet further incubation at 37°C restored VLP release (Fig. 4B).

To investigate the requirement for hPOSH-mediated ubiquitination for Gag membrane localization, cells were cotransfected with a plasmid encoding either epitope-tagged hPOSH or hPOSH_V14A together with pNLenv-1 and then shifted to 20°C. VLP release was measured after additional incubation at 37°C. As expected by the requirement for hPOSH, VLP were efficiently released from H187 cells but not from H153 cells. WT hPOSH, but not the V14A RING mutant, stimulated VLP production in H153 to the levels observed in H187 cells. Expression of the RING mutant in H187 cells significantly inhibited VLP release, suggesting a dominant-negative effect (Fig. 4C). Nevertheless, expression of native hPOSH also had a mild inhibitory effect on VLP release in H187 cells. Because V14A expression level was higher then that of native hPOSH (Fig. 4C Insets), we could not completely exclude the possibility that overexpression of hPOSH per se exerted a general toxic effect or that it indirectly affected virus release through regulation of Gag expression. However, these possibilities appear unlikely because Pr55-Gag expression was unaffected by either native or mutant hPOSH expression (Fig. 4C Lower) and because mutant (but not WT) hPOSH expression inhibited Gag transport to the plasma membrane in H187 cells (see Fig. 5A). The weak activation of virus release in H153 cells by hPOSH_V14A may be due to either a residual activity or sequestration of a yet-unknown negative regulator.

To test directly whether hPOSH regulates Gag transport to the plasma membrane, we compared the intracellular distribution of newly synthesized Gag in H187 and H153 cells. Cells were transfected with Gag-EGFP expression plasmid and6hlaterwerefixed and permeabilized for indirect immunofluorescence. Visualization of Gag-EGFP by confocal microscopy indicated that in the H187 control cells, Gag associated with both the plasma membrane and the TGN (Fig. 5A Top). Upon expression of exogenous hPOSH, Gag disappeared from the TGN and was mainly present at the plasma membrane, indicating acceleration of Gag transport (Fig. 5A Middle). Nevertheless, ectopic expression of hPOSH did not stimulate virus release from H187 cells (Fig. 4B), suggesting that the final stages of particle budding and release at the cell surface became rate limiting. In agreement with the marked inhibition of VLP release in H187 cells (Fig. 4C), hPOSH_V14A expression redistributed Gag into intracellular, punctate loci that were largely nonoverlapping with the TGN (Fig. 5A Bottom). In hPOSH-depleted H153 cells, Gag showed punctate intracellular staining similar to that induced by hPOSH_V14A in H187 (Fig. 5; compare A Lower with B Upper). In agreement with the ability to restore virus release, expression of hPOSH in H153 cells also restored Gag membrane localization, whereas expression of hPOSH_V14A had no effect (Fig. 5B Middle and Bottom, respectively).

The blockage of Gag plasma membrane localization by hPOSH RNAi and the inability of mutant hPOSH_V14A to restore production of VLP in hPOSH-inhibited cells (Figs. 4 and 5) suggest that targeting of Gag from the TGN to the plasma membrane requires the ubiquitination activity of hPOSH.

Discussion

In this work we have identified an association of newly synthesized Gag with the TGN. We have further demonstrated that RNAi-mediated silencing and inhibition of hPOSH-ubiquitination activity prevented Gag from localizing at the plasma membrane and inhibited virus secretion. The arrest of Gag at the TGN and the abrogation of VLP release upon inhibition of vesicular transport (Fig. 4 A and B) suggest that Gag transits at the TGN en route to the plasma membrane. The results showing restoration of VLP secretion and Gag membrane localization in hPOSH-depleted H153 cells by expression of native, but not RING mutant, hPOSH and, in contrast, displacement of Gag from the TGN and attenuation of VLP secretion upon expression of a hPOSH RING mutant in H187 cells (Figs. 4C and 5) indicate that targeting of Gag to the plasma membrane is regulated by hPOSH-mediated ubiquitination.

We were unable to identify differences in Gag ubiquitination in the presence or absence of hPOSH (data not shown). We therefore postulate that hPOSH affects Gag trafficking indirectly by regulating the sorting of Gag-containing cargo vesicles. Consequently, the earlier findings that ubiquitination is involved in virus budding and release (20–22) imply that a distinct E3 ligase is involved in the late stages of the HIV-1 life cycle.

It is becoming apparent that retroviral Gag transport to the plasma membrane is mediated by Gag/Env interaction at intracellular membranes and that incorporation of processed Env (gp120/gp41) into infectious viruses likely requires passage through the TGN (14, 16). In this study, we obtained inhibition of virus production through hPOSH silencing by using an env-deficient subviral DNA, suggesting that hPOSH directly affected Gag transport. Nevertheless, with reduced levels of hPOSH, VLP secretion could still be detected (Fig. 1A), whereas production of infectious, pseudotyped, HIV-1/VSV-G was inhibited by 2–3 orders of magnitude (Fig. 1B). The milder effect on virus release relative to virus infectivity could be explained if immature particles were released through an hPOSH-independent pathway or if hPOSH also regulated Env assembly/transport. Based on the results presented in this study, we propose two models for virus assembly. According to the first model, newly synthesized Gag and Env cluster at the TGN before movement to the plasma membrane (Fig. 6A). According to the second model, newly synthesized Gag is sorted by means of the TGN to late endosomes where it assembles with Env before transport to the cell membrane (Fig. 6B). According to both models, inhibition of hPOSH results in diminished delivery of Gag to the plasma membrane and inefficient Gag-Env assembly, followed by production of defective progeny viruses.

Fig. 6.

Proposed pathways for HIV-1 Gag trafficking to the cell membrane. (A) Newly synthesized Gag assembles with endocytosed Env at the TGN before transport to the cell surface. (B) Newly synthesized Gag is sorted by means of the TGN to late endosomes (LE) where it assembles with Env before transport to the cell membrane. Consistent with both mechanisms, the role of hPOSH (indicated by a star) is to facilitate egress of Gag cargo vesicles from the TGN.

Cotransport of Env and CA may be a common pathway for the assembly of other enveloped viruses (i.e., viruses that mature through budding at the host cell membrane). For example, immunofluorescence and EM studies of West Nile virus assembly indicate that viral envelope proteins and CAs are closely associated throughout the virus life cycle and are cotransported from a perinuclear region to the plasma membrane via microtubules (43).

The model for cotransport of Gag/Env to the plasma membrane is analogous to the cotransport mechanism of the nicotinic acetylcholine receptor and the membrane-associated acetylcholine receptor-escort protein Rapsyn to the neuro-muscular synapse (44). Intriguingly, Rapsyn and Gag share common functional and structural features. Both proteins self-cluster, associate with membranes by means of a myristoyl moiety, and require a second amino-terminal basic motif for traffic to the cell membrane (5, 25, 28, 45, 46).

In conclusion, hPOSH is a host ubiquitin ligase essential for HIV-1 production and thus constitutes a potential target for therapeutic intervention in HIV-1-infected patients. Moreover, considering the possibility that sorting of enveloped virus's structural proteins to the plasma membrane utilizes common pathways, hPOSH may constitute a broader-spectrum antiviral target. Further insight into the function and mechanism of hPOSH under normal physiological conditions and during HIV infection will be facilitated by identification of the subcellular compartment to which Gag is diverted when hPOSH is inhibited as well as by the identification of hPOSH-interacting proteins and ubiquitination substrates.