Many transmembrane receptors that are destined for lysosomal degradation are directed through the multivesicular body (MVB) sorting pathway [1,2]. Internalized receptors are sorted into endosomal membrane invaginations, which then pinch off to form intralumenal vesicles within the MVB. This process is initiated by the ubiquitin-binding protein, Hrs, which recruits the endosomal sorting complex required for transport (ESCRT)-I to endosomal membranes by directly interacting with the ESCRT-I component, TSG101 (Tumor Susceptibility Gene 101) [3]. Following ESCRT-I recruitment, ESCRTs-II and III are sequentially localized to the endosomal membrane [1]. These complexes bind ubiquitylated receptors and are required for receptor sorting into the lumen of the MVB. The AAA type ATPase Vps4 then facilitates the disassembly of the ESCRT complexes prior to membrane fission, thereby ensuring that these complexes are available for further rounds of protein sorting [4].

In many respects, vesicle formation in the MVB is topologically identical to viral budding at the plasma membrane: both processes involve budding away from the cytosol. Most enveloped viruses have evolved strategies to gain access to cellular ESCRTs in order to mediate virion egress from the infected cell [5]. For example, HIV-1 recruits ESCRT complexes to sites of viral assembly through direct interactions between the Gag polyprotein and two cellular ESCRT proteins: TSG101 and AIP1/Alix [6-9]. Depletion of TSG101 or introduction of dominant negative mutants of AIP1/Alix arrests HIV-1 budding at a late stage and blocks viral particle release [7,10]. Likewise, depletion of TSG101 and other ESCRT components inhibits lysosomal downregulation of ligand-activated growth factor receptors, such as the EGF Receptor (EGFR) [11-13]. Given the fact that HIV-1 budding and EGFR downregulation both require ESCRT function, it is logical to question whether there is competition for cellular ESCRT components when both processes occur in the same cell at the same time.

We have previously shown that expression of HIV-1 Gag decreases the rate of EGF induced EGFR degradation [14]. This effect is dependent on the presence of an intact TSG101-binding sequence within the Gag polyprotein. As a result, activated EGFR accumulates in late endosomal compartments and Gag expressing cells exhibit higher levels of activated MAP Kinase. These findings indicate that HIV-1 Gag impinges upon the normal function of cellular ESCRT complexes during EGFR downregulation. In order to determine whether downregulation of other receptors is sensitive to HIV-1 Gag expression, we have now investigated the kinetics of lysosomal downregulation of CD4 and CXCR4, in the presence and absence of Gag. CD4 and CXCR4 function as the receptor and co-receptor respectively for the entry of HIV-1 X4 variants into target cells [15]. Regulation of the cell surface levels of these two proteins is critically important for HIV-1 pathogenesis.

CXCR4 is a seven transmembrane G-Protein Coupled Receptor (GPCR) expressed on the cell surface of various leukocytes such as neutrophils, monocytes and lymphocytes [16,17]. The ligand for CXCR4 is the chemokine stromal cell-derived factor (SDF-1), which regulates the movement of leukocytes during their development, homeostasis and inflammatory response [18]. Upon SDF-1 binding, CXCR4 is rapidly phosphorylated by GPCR kinase (GRK) and internalized via clathrin-coated pits [19]. SDF-1-bound CXCR4 is also monoubiquitylated by the Nedd4 like E3 Ub ligase AIP4 [20]; this promotes sorting of CXCR4 into the internal vesicles of the MVB prior to lysosomal degradation. While Hrs and Vps4 have been implicated in the lysosomal degradation of monoubiquitylated CXCR4 [20], no study has determined whether the ESCRT complexes play a role in this process. Hrs and Vps4 have been shown to function in ESCRT-dependent [3] as well as ESCRT-independent [21] pathways of receptor sorting. Identifying which of the two Hrs-dependent pathways is functional in the lysosomal downregulation of CXCR4 is important since this may also have implications for the trafficking of this receptor in HIV-1 infected cells.

CD4 is a transmembrane glycoprotein that is expressed on the surface of a subset of T cells as well as monocytes and macrophages. It normally functions as a signal transducer during T cell activation. Antigens, mitogens and PKC modulators such as PMA induce internalization of CD4, thereby regulating its cell surface expression [22,23]. PMA induces phosphorylation of CD4, resulting in its rapid internalization from the cell surface and degradation in lysosomes [24,25]. Similarly, the HIV-1 Nef protein induces endocytosis and lysosomal degradation of CD4 [26]. Both PMA and Nef-induced CD4 internalization are dependent on the clathrin adaptor protein AP-2 [27]. The sorting of CD4 from early endosomes into late endosomes/MVBs during Nef-induced downregulation has been shown to be dependent on the interaction of Nef with the β subunit of the coatomer protein-(COP)-1 complex and ARF1 [28,29]. How this sorting step occurs during PMA-induced CD4 downregulation and whether ESCRT complexes are involved is not known.

Given the fundamental importance of both CXCR4 and CD4 for both normal cell physiology and HIV-1 biology, we have examined the role of ESCRT I in downregulation of these two cellular proteins. SDF-1-induced downregulation of CXCR4 and PMA-induced downregulation of CD4 were monitored in cells depleted of endogenous TSG101 using siRNA directed against TSG101. We also monitored CXCR4 and CD4 downregulation in cells expressing HIV-1 Gag. Our findings indicate that SDF-1-induced CXCR4 downregulation is defective when ESCRT-I function is impaired, while PMA-induced CD4 downregulation is not. Consequently, the expression of HIV-1 Gag attenuated CXCR4 downregulation, while having little to no effect on CD4 downregulation. These results have important implications for the endocytic trafficking of CD4 and CXCR4 in normal as well as HIV-1 infected cells.

Figure 1 SDF-1 induced CXCR4 downregulation is TSG101-dependent. (A) COS-1 cells co-transfected with HA-tagged CXCR4 and either GFP, TSG101-GFP, or siRNA directed against TSG101, as indicated, were incubated with an anti-HA antibody for 1 hour on ice. Cells were (more ...)

In order to determine whether SDF-1 induced HA-CXCR4 downregulation is dependent on the ESCRT-I complex, cells were depleted of the critical ESCRT-I component, TSG101. Addition of siRNA directed against TSG101 resulted in 80% knockdown of endogenous TSG101 levels (Fig. 1B). SDF-1 induced HA-CXCR4 degradation was significantly attenuated in TSG101-deficient cells (Fig. 1A), as indicated by the retention of receptors in punctate structures even after 3 hours of incubation with SDF-1. An alternative method to interfere with TSG101 function was also implemented. Overexpression of full length TSG101 has been shown to inhibit ESCRT-I function and block EGF-induced EGFR downregulation [14,32]. COS-1 cells overexpressing TSG101 also exhibited attenuated HA-CXCR4 degradation (Fig. 1A). These data indicate that HA-CXCR4, like EGFR, is dependent on TSG101 function for SDF-1 mediated degradation.

To test this hypothesis, COS-1 cells were co-transfected with plasmids encoding HA-CXCR4 and Gag-GFP (GFP-tagged HIV-1 Gag). Cells expressing wild type HIV-1 Gag-GFP exhibited attenuated HA-CXCR4 degradation (Fig 2A). This effect of Gag was dependent on its TSG101-interacting PTAP sequence, located within the C-terminal p6 region of the Gag polyprotein. Cells expressing a Gag PTAP mutant efficiently degraded HA-CXCR4 (Fig. 2A).

Figure 2 HIV-1 Gag attenuates CXCR4 degradation in COS-1 cells. (A) COS-1 cells co-expressing HA-tagged CXCR4 (red) and either wild type (top panels) or LTAL, a PTAP mutant form (bottom panels) of Gag-GFP (green) were treated as described in Figure 1A. Only cells (more ...)

HA-CXCR4 degradation efficiencies were quantitated in cells expressing various GFP-tagged constructs. HA-CXCR4 degradation was decreased 3–6 fold in cells expressing TSG101-GFP or Gag-GFP, compared to cells expressing GFP (Fig 2B). A similar effect was noted in cells depleted of TSG101. In contrast, CXCR4 degradation in cells expressing the late domain mutant, LTAL Gag-GFP was nearly equivalent to that of control cells (Fig 2B). These results suggest that expression of wild type HIV-1 Gag interferes with the function of endogenous TSG101 and/or ESCRT-I machinery, resulting in increased accumulation of internalized, undegraded HA-CXCR4, following SDF-1 treatment.

We next examined whether accumulation of intracellular HA-CXCR4 caused alterations in SDF-1 mediated signaling. GPCRs are known to be rapidly desensitized after ligand binding and internalization. One would therefore predict that accumulation of intracellular, inactivated receptors would not alter signaling. To test this hypothesis, the time course of pERK formation, a downstream readout of SDF-1 mediated CXCR4 signaling, was monitored. As depicted in Figure 2C, cells expressing Gag-GFP exhibited identical kinetics and levels of pERK production when compared to cells expressing GFP. Thus, accumulation of intracellular HA-CXCR4 did not result in altered SDF-1 induced CXCR4 signaling in Gag expressing cells.

Previous studies have shown that T cells have large intracellular stores of CXCR4 that can be mobilized by treating the cells with PMA and ionomycin [17]. Indeed, in order to observe SDF-1 induced CXCR4 degradation in Jurkat cells, we needed to inhibit the synthesis of new receptors continuously with cycloheximide and incubate the cells with SDF-1, PMA and ionomycin (Fig. 3A). Efficient expression of HIV-1 Gag was achieved by transducing Jurkat cells with Gag-GFP-encoding lentiviruses. At a multiplicity of infection (MOI) of 10, over 90% of the cells expressed Gag-GFP (Fig. 3B). Incubation of Gag-GFP expressing Jurkat cells with SDF-1, PMA and ionomycin revealed that downregulation of endogenous CXCR4 was clearly attenuated by expression of wt Gag-GFP (Fig. 3C,D,E). In contrast, cells expressing the late domain mutant, LTAL Gag-GFP, exhibited CXCR4 degradation rates more similar to the LacZ control (Fig 3D,E). Notably, cell surface levels of CXCR4 at steady state were not altered in HIV-1 Gag expressing cells (Fig. 3F). Thus, experiments in Jurkat cells reveal that expression of HIV-1 Gag attenuates downregulation of endogenous CXCR4 in the presence of SDF-1, PMA and ionomycin. These data are similar to those reported in Figure 2, for SDF-I induced HA-CXCR4 degradation.

Figure 3 HIV-1 Gag attenuates SDF-induced CXCR4 downregulation in Jurkat T cells. (A) Jurkat T cells were pre-treated with cycloheximide, then incubated in the presence (filled squares) or absence (open squares) of SDF, PMA and ionomycin for the indicated times. (more ...)

Previous studies have quantitated CD4 degradation kinetics by monitoring levels of metabolically (pulse) labeled CD4 over time in untreated and PMA-treated cells [24,33,34]. Pulse labeled CD4 has been shown to proceed to the cell surface via the secretory pathway within 30–60 minutes after synthesis, internalize via endocytosis, and undergo degradation in lysosomes [34]. We were unable to immunoprecipitate endogenous CD4 from Jurkat T cells using a wide range of available anti-CD4 antibodies. We therefore examined PMA-induced downregulation of exogenously expressed CD4, which was readily radiolabeled and immunoprecipitated using a monoclonal anti-CD4 (Leu3a) antibody. Trafficking of exogenous CD4 has been shown to accurately represent that of endogenous CD4 [35], and we have previously shown that CD4 is trafficked to the cell surface in transfected COS-1 cells [36,37].

COS-1 cells expressing exogenous CD4 were metabolically labeled with ³⁵S-Met/Cys-translabel for 10 minutes, then chased in non-radioactive medium in the presence or absence of PMA. PMA induced a significant decrease in CD4 levels over a period of 6 hours (Fig. 4A). The experiment was then repeated in cells depleted of endogenous TSG101 using siRNA. At early time points, CD4 degradation was slightly attenuated in TSG101 depleted cells (Fig. 4B, 2 hour time point). However, by 6 hours, CD4 was degraded as efficiently in TSG101-depleted cells as in control cells. Under similar conditions of TSG101 knockdown, EGF induced EGFR degradation was dramatically inhibited (Fig. 4D).

Figure 4 PMA-induced CD4 downregulation is TSG101 and Vps4 independent. (A) COS-1 cells transfected with CD4 were labeled with trans-35S-label for 10 minutes, then chased in non-radioactive medium in the presence or absence of 50 ng/ml PMA for the indicated times. (more ...)

In order to determine whether the AAA-type ATPase Vps4 plays a role in PMA-induced CD4 downregulation, a dominant negative form of Vps4 (Vps4E228Q-GFP) was co-expressed with CD4. At early time points, CD4 degradation was slightly attenuated, but by 6 h, Vps4E228Q-GFP expressing cells had degraded CD4 as efficiently as cells expressing GFP (Fig. 4C). In contrast Vps4E228Q-GFP strongly inhibited EGF-induced EGFR degradation (Fig. 4E). Taken together, these findings suggest that PMA-induced CD4 degradation may be partially dependent on TSG101 and Vps4 function and only during early times after exposure to PMA. With prolonged PMA treatment, CD4 degradation can proceed in the absence of functional TSG101 and Vps4.

Figure 5 HIV-1 Gag does not attenuate PMA-induced CD4 downregulation. (A) COS-1 cells co-expressing CD4 and either GFP or Gag-GFP were treated as described in Fig. 5A. Data from one representative experiment (out of three) is shown. (B) Data shown represents the (more ...)

A summary of all the measurements of PMA-induced CD4 degradation represented as the percent of undegraded CD4 remaining after 6 hours of PMA treatment is shown in Fig. 5B. PMA-induced CD4 degradation was not affected by either depletion of endogenous TSG101 or by overexpression of Vps4E228Q or HIV-1 Gag, suggesting that under these conditions, lysosomal degradation of CD4 can proceed in the absence of ESCRT-I and Vps4. Taken together, our studies clearly demonstrate that downregulation of CXCR4, but not CD4, is attenuated by HIV-1 Gag-mediated recruitment of ESCRT complexes. Thus the ability of HIV-1 Gag to impinge on the cellular endocytic pathway is selective, attenuating only ESCRT-dependent processes in this pathway.

In this study, we show that HIV-1 Gag, as well as TSG101, differentially affect the kinetics of downregulation of the HIV-1 co-receptors CXCR4 and CD4. SDF-1-induced CXCR4 downregulation was sharply reduced when TSG101 function is inhibited, while PMA-induced CD4 downregulation was relatively unaffected. Depleting TSG101 using siRNA directed specifically against TSG101 has been shown to result in a reduction in the cellular levels of the other two components of the ESCRT-I complex, Vps28 and hVps37 [11]. Consequently, TSG101-depleted cells have fewer functional ESCRT-I complexes. Thus, our observations imply that CXCR4, but not CD4, is dependent on the ESCRT-I complex for its lysosomal degradation.

Divergent receptors, such as the Epidermal Growth Factor Receptor (EGFR) and the Toll-like receptor, TLR4, have been shown to be dependent on Hrs, ESCRT-I and Vps4 for their transport from early endosomes to late endosomes to lumenal vesicles in the MVB [3,38]. However, this is not the only route into the MVB. For example, sorting and lysosomal degradation of the Delta Opioid Receptor (DOR), a GPCR, is Hrs and Vps4 dependent but does not require TSG101 (ESCRT-I) [21]. Another recent study showed that lysosomal sorting of the GPCR, PAR-1 (Protease-activated receptor-1) does not require Hrs or TSG101 [39]. Moreover, MVB sorting of the melanosome protein Pmel17 is completely independent of any of the members of the 'Vps' family of proteins [40]. It is also important to note that not all ESCRT components are essential for HIV-1 budding and release. For example, depletion of ESCRT-II components or of AIP1/Alix has minimal effects on HIV-1 budding [6,41]. These studies clearly indicate that protein sorting and MVB biogenesis are complex processes involving multiple points of entry and regulation.

Besides modulating the levels of receptors at the cell surface, the cellular endocytic pathway plays a major role in the attenuation of ligand-induced receptor-mediated signaling [42]. For EGFR and other Receptor Tyrosine Kinases, sequestration of the activated receptor into the internal vesicles of the MVB is required to shut down signaling. This process occurs 2–3 hours after ligand binding. Activated EGFR remains in early and late endosomes during this time and is capable of mediating continued intracellular signaling [43]. We previously showed that expression of HIV-1 Gag increases the amount of EGFR that remains in endosomes after EGF addition [14]. As a consequence, Gag expressing cells exhibit hyperactivated and prolonged MAP kinase signaling. Increased MAP kinase activation is advantageous to the virus as it has been shown to enhance HIV-1 replication and infectivity [44,45].

In contrast, GPCR-mediated signaling is attenuated within minutes after ligand binding. CXCR4, like most GPCRs, is desensitized rapidly through the action of GPCR kinases (GRKs), which phosphorylate the receptor on several C terminal Ser/Thr residues [19]. Phosphorylation of CXCR4 promotes binding of β-arrestins, which sterically hinder the association of heterotrimeric G proteins with the receptor and thereby block signal transduction [46]. The increase in intracellular CXCR4 induced by expression of HIV-1 Gag did not result in a change in SDF-1 mediated CXCR4 signaling, as judged by MAP kinase activation (Fig 2C). This finding is consistent with known GPCR biology and confirms that internalized CXCR4 in Gag expressing cells is desensitized and does not signal.

β-arrestin binding to GPCRs also serves to recruit components of the endocytic machinery including clathrin and AP-2, thereby mediating the internalization of the receptor [47]. Following internalization, CXCR4 colocalizes with Hrs-positive endosomes [20]. While Hrs and Vps4 have been implicated in CXCR4 downregulation [20], no role for TSG101 or ESCRTs had been established in this process until now. Our data (Figs 1, 2) strongly suggest that SDF-1 induced CXCR4 downregulation is TSG101 and ESCRT-I dependent. Given that HIV-1 Gag competes with Hrs for TSG101 in vitro [48], and that overexpression of TSG101-binding regions of Hrs inhibits HIV-1 release [49], we hypothesized that expression of Gag would compete for TSG101 binding and function in vivo. Our observations that TSG101/ESCRT-I dependent downregulation of CXCR4 (Figs 2, 3) and EGFR [14] are attenuated in HIV-1 Gag expressing cells indicate that Gag functionally depletes the ESCRT complexes, thereby interrupting other ESCRT-dependent pathways in the cell.

Two other HIV-1 proteins, Nef and Env, have been shown to interact with or regulate CXCR4. Thus, an important question is whether levels of CXCR4 are altered in the context of an HIV-1 infected cell. Multiple studies have addressed this issue, primarily by quantitating the amount of cell surface CXCR4. A recent study reported that HIV-1 Nef induces downregulation of CXCR4 from the cell surface of infected cells [50]. The authors propose that Nef-mediated CXCR4 downregulation may protect against superinfection. Superinfection is detrimental to viral replication because the accumulation of unintegrated viral DNA results in the induction of cytopathic effects in the host cell [51]. However, multiple other studies have shown that HIV-1 Nef does not downregulate cell surface levels of CXCR4 [52-54] and that maximal protection from superinfection involves an unidentified mechanism that is independent of CXCR4 downregulation [54]. Similarly, we observed no change in cell surface levels of CXCR4 in HIV-1 Gag-expressing cells (Fig. 3F). In contrast, others have, in some cases, seen an upregulation in cell surface expression of CXCR4 in HIV-1 infected CD4+ T cells [55]. SDF-1 induced CXCR4 signaling could potentially be beneficial to viral replication since it results in the activation of transcription factors such as NFκB [16], which are known to increase HIV-1 LTR promoter activity [44]. It is also important to note that HIV-1 Env protein (shed from virus particles or infected cells) can bind to CXCR4 and thereby trigger apoptotic signals. However, CD4 and CXCR4 expression are both required for apoptotic signaling by Env in CD4 T cells [56]. Since CD4 is efficiently removed from the surface of productively infected cells (see below), only uninfected/bystander CD4 T cells express both CD4 and CXCR4 and are therefore susceptible to Env-induced apoptosis [56].

Thus, CXCR4 downregulation might not be essential for HIV-1 replication. We speculate that during the late stages of the viral life cycle when mostly structural proteins such as Gag are expressed, SDF-1 induced CXCR4 downregulation is attenuated resulting in the accumulation of densensitized CXCR4 within intracellular compartments. These receptors could, in the long run, contribute to maintaining or replenishing the cell surface levels of CXCR4 in HIV-1 infected cells.

Unlike SDF-induced CXCR4 downregulation, Gag expression had little to no effect on PMA-induced CD4 downregulation (Fig. 5). PMA is a phorbol ester that binds to and activates protein kinase C (PKC) [57]. PKC is normally activated upon binding of antigen to the T-cell receptor and its associated CD4 [58]. Activated PKC phosphorylates CD4 on its cytoplasmic tail and induces CD4 internalization and lysosomal degradation [25,59]. Several studies have shown that PMA treatment mimics the mechanism of antigen induced CD4 downregulation [23,60]. Surprisingly, little is known about how internalized CD4 gets sorted into the internal vesicles of the MVB prior to lysosomal degradation. In the present study, we show that PMA-induced CD4 downregulation can occur efficiently in the absence of functional ESCRT-I and Vps4 (Fig. 4) and that expression of HIV-1 Gag has no effect on this process (Fig. 5). These findings indicate that Gag affects only ESCRT-dependent processes. We therefore predict that lysosomal degradation of CD4 should not be impeded by Gag in an HIV-1 infected cell. Indeed, loss of cell surface CD4 is a hallmark of HIV-1 infection [51]. After virus entry, it is essential that HIV-1 efficiently downregulates CD4 for multiple reasons. CD4 downregulation is important to prevent superinfection of the infected cell [61]. In addition, cross-linking of CD4 in the absence of T cell receptor activation results in the generation of non-proliferative or apoptotic signals [62]. Viral transcription is also inhibited under these conditions [63]. Many studies have also reported that cells overexpressing CD4 exhibit a drastic inhibition of virion release [64,65]. Moreover, the presence of CD4 at the cell surface appears to significantly reduce the infectivity of released virions [66,67]. Exactly how CD4 exerts these effects is unclear, but these observations establish the critical need for HIV-1 to downregulate CD4 in infected cells. Three different viral proteins, Nef, Env and Vpu have evolved to ensure that cell surface CD4 is downregulated soon after entry (by the early protein Nef) and that transport of newly synthesized CD4 to the cell surface at late stages of infection is blocked (by late proteins Env and Vpu) [51,68]. Thus, by the time Gag proteins are expressed in an infected cell, most of the surface CD4 has already been downregulated by Nef.

Our observations indicate that expression of HIV-1 Gag functionally depletes cellular ESCRT complexes. As a consequence, Gag expression modulates ESCRT-dependent but not ESCRT-independent receptor sorting pathways in the host cell. These findings are likely to be highly relevant to HIV-1 pathogenesis as they shed light on the mechanisms used by HIV-1 proteins to dysregulate normal cell physiology and to potentiate viral replication.

For colocalization experiments, cells were incubated with monoclonal anti-EEA1 or monoclonal anti-CD63 antibody following the 3 hour SDF treatment. Cells were then stained with the Alexa Fluor^® 594-conjugated goat anti-rabbit antibody (for HA-CXCR4) and a Cy5-conjugated anti-mouse antibody (for EEA1 or CD63). Colocalization was measured on a pixel-by-pixel basis using MetaMorph software (Universal Imaging Corp., Downingtown, PA). EGF-induced EGFR downregulation experiments were performed as previously described [14].

ESCRT (endosomal sorting complex required for transport); multivesicular body (MVB); TSG101 (Tumor Susceptibility Gene 101); EGF Receptor (EGFR); SDF-1 (stromal cell-derived factor).

The author(s) declare that they have no competing interests.

RRV performed all experiments and wrote the majority of the manuscript. RRV and MDR jointly conceived the rationale for the study and designed the experiments. MDR wrote part of the manuscript, and edited the entire manuscript. Both authors read and approved the final manuscript.

We thank Dr. Tia Rai and Nathaniel Martinez for fruitful discussions and critical analysis of the data; Dr. Jan Hendrix and Diane Domingo for technical assistance with flow cytometry; Drs. Stanley Cohen, John Wills, Uta von Schwedler and Lionel Ivashkiv for kind gifts of reagents, and Raisa Louft-Nisenbaum for expert technical assistance. This work was supported by NIH Grant CA 72309.