|Home | About | Journals | Submit | Contact Us | Français|
Human immunodeficiency virus type 1 (HIV-1) Vpr participates in nuclear targeting of the viral preintegration complex in nondividing cells and induces G2 cell cycle arrest in proliferating cells, which creates an intracellular milieu favorable for viral replication. Vpr also activates the transcription of several promoters and enhancers by a poorly understood mechanism. Vpr enhances glucocorticoid receptor (GR) signaling and may mediate the effects of steroids on HIV replication. More specifically, recombinant Vpr can potentiate virion production from U937 cells, downregulate NF-κB induction, and enhance programmed cell death, all effects also mediated by glucocorticoids. Vpr has been proposed to act as a GR coactivator, although other studies suggest that these enhancing effects are merely a consequence of G2 cell cycle arrest. We now demonstrate that Vpr functions as a GR coactivator and that this activity is independent of cell cycle arrest. In addition, we show that the Vpr-induced coactivation requires an intact glucocorticoid response element, that it is dependent on the presence of hormone and the corresponding receptor, and that it is mediated by the two highly conserved leucine-rich domains within Vpr that resemble the GR coactivator signature motif.
The genomes of human immunodeficiency virus types 1 and 2 (HIV-1 and HIV-2) encode several accessory proteins that are highly conserved in vivo and likely function to accelerate virus production. The vpr gene of HIV-1 encodes a 96-amino-acid, 14-kDa viral protein R (Vpr) that is expressed in infected cells and packaged into the virion through its interaction with the p6 region of the Gag precursor (6, 40, 62). Structural analysis of Vpr by nuclear magnetic resonance (48) and circular dichroism (35) predicts two α-helices, one located at the amino terminus between amino acids 17 and 34, and the other between amino acids 53 and 78. These helices likely play a role in Vpr dimerization (64) and perhaps binding to other proteins (63), likely through a leucine zipper-like mechanism (58). The carboxy-terminal region of Vpr corresponds to a less-well-characterized basic amino acid stretch between amino acids 79 and 96.
While Vpr is not required for viral replication in transformed cell lines or even cultured peripheral blood mononuclear cells (PBMCs) (39, 46), the open reading frame is maintained in vivo (17, 61), reflecting its importance in the HIV replicative life cycle. Vpr facilitates virus production in macrophages, perhaps by cooperating with integrase and matrix proteins to promote nuclear import of the viral preintegration complex (8, 15, 16, 21, 42, 56). In this regard, Vpr contains at least two distinct import signals that facilitate nuclear entry by a novel mechanism that may involve direct docking to the nuclear pore complex (25).
Vpr also induces G2 cell cycle arrest in proliferating human cells (4, 17, 19, 26, 43). In fact, each virion contains sufficient quantities of packaged Vpr to arrest infected T cells in the G2 phase of the cell cycle (24, 41). Such arrest during the initial phase of infection may create an intracellular environment conducive to improved transcription of the long terminal repeat (LTR) (17). However, at a later point in HIV replication, the G2 arrest induced by Vpr may lead to increased apoptosis even in cells infected with replication-defective forms of HIV (41, 51, 52, 59). The G2 arrest property of HIV-1 Vpr has been dissociated from the nuclear import properties by mutagenesis experiments (14, 36, 53). In the related HIV-2 and simian immunodeficiency virus (SIV) lentiviruses, the nuclear import and G2 cell cycle arrest properties are segregated between Vpx and Vpr2/VprSIV (13, 55).
In addition to increasing HIV LTR transcription, Vpr also upregulates the activity of several heterologous promoters and enhancers (7). It is possible that this effect is a byproduct of G2 cell cycle arrest. Increased transcription is observed in arrested cells following expression of Vpr by transfection or by infection (18, 53). However, it has been suggested that Vpr may increase transcription by physically interacting with various host factors, such as Sp1 (57), TFIIB (1, 2, 28), or an activated glucocorticoid receptor (GR) (28, 44). If Vpr binds DNA or other transcription factors directly in vivo, it likely enhances transcription through improved recruitment of p300/CBP (11).
Receptors for steroid hormones, thyroid hormones, and retinoic acid all belong to a superfamily of ligand-dependent transcription factors (for a review, see reference 54). These factors have sequence homology that includes an amino-terminal transcriptional activation domain (AD or AF-1), a central DNA-binding motif, and a carboxy-terminal, ligand-binding domain containing a second activation domain termed AF-2. Mutations in AF-2 can abolish signal transduction without affecting ligand binding or dimerization. Recently, steroid receptor coactivators (SRCs) have been identified as important members of the transcriptional activation complex that mediates nuclear hormone signaling. The recruitment of gene transcription machinery by the activated nuclear hormone receptor is dependent on this new class of coactivator proteins. The SRCs bind the hydrophobic cleft within the AF-2 domain through physical interaction of their signature LXXLL motifs, which are required for their function (9, 12, 20, 54).
It has been suggested that glucocorticoids and recombinant Vpr influence the HIV life cycle in an analogous fashion (44). Specifically, the Vpr-mediated enhancement of virus production demonstrated in cultured macrophages was mimicked by substitution of glucocorticoids and, correspondingly, inhibited by the addition of the GR antagonist RU486 (mifepristone). Moreover, exogenous addition of recombinant Vpr to cells appears to stimulate glucocorticoid activity in cultured cells with respect to apoptosis, increased virion production from U937 cells, and downregulation of NF-κB (3). Treatment with RU486 inhibits all of these effects. The observation that SRCs contained the signature motif LXXLL and the presence within Vpr of an LQQLL element at amino acids 64 to 68 raised the possibility that Vpr functions as a GR coactivator. Kino and colleagues have described such an activity (28). However, others have directly disputed this finding and suggest that the G2 cell cycle arresting properties account for these coactivator-like effects of Vpr (14). In this regard, replacement of L64 with alanine in the LQQLL motif (AQQLL or L64A) both inhibits the transcriptional activating properties of Vpr and reduces the degree of G2 cell cycle arrest (28, 36).
In view of these uncertainties, we further explored the possibility that Vpr acts as a GR coactivator. Because it is known that multiple LXXLL interaction domains function synergistically in SRCs (9, 38, 47), we also examined the possible cooperation between the two leucine-rich motifs of Vpr with respect to G2 cell cycle arrest as well as GR coactivation.
The plasmid containing the glucocorticoid response element (GRE) derived from the tyrosine aminotransferase (TAT) 5′ regulatory region and the control plasmid lacking the TAT3 repeats have been described (33). The rat GR encoded by the 6RGR plasmid (32) was used in coimmunoprecipitation assays and in transfection of CV1-B cells, which lack an endogenous cortisol receptor. The GRIP1-encoding plasmid (23) was kindly provided by Michael Stallcup (University of Southern California). All Vpr constructs were inserted into the pCMV4 vector containing three hemagglutinin (HA) repeats at the amino terminus. Mutations were generated by oligonucleotide-directed PCR and confirmed by sequencing. Expression was verified by Western blot analysis. All constructs that induced G2 arrest as well as the Vpx construct expressed proteins at nearly equivalent levels (data not shown).
All transfections were performed using calcium phosphate for precipitation of DNA. Cells were plated at 300,000/well in six-well plates containing Dulbecco's modified Eagle's medium (Gibco-BRL, Gaithersburg, Md.) with 5% “stripped” fetal bovine serum (charcoal treated at 10 g/liter for 1 h to remove endogenous steroids), penicillin G at 100 U/ml, and streptomycin at 100 μg/ml. Transfections were performed 18 h after plating at a constant DNA concentration of 4 μg per well. Cells were washed 24 h after transfection. At 48 h, the medium was replaced with medium containing the indicated steroid and analyzed 3 h later. Transfected cells were assayed for luciferase activity after lysis with 1× lysis buffer (Promega), adding reagents A and B (Amersham) to 20 μl of lysate, and measuring light output on a Microbeta 1450 Trilux luminescence counter (Wallac Company).
293T cells were plated in 100-mm dishes, transfected, and cultured for 48 h. Cells were washed with phosphate-buffered saline (PBS) and lysed for 20 min, and the supernatant was incubated with monoclonal mouse antibody HA.11 immobilized on Sepharose Fast Flow beads (BabCO) for 1 h at 4°C and then washed three times. The beads were then boiled for 5 min to dissociate any bound proteins. Lysis and wash buffer consisted of 50 mM HEPES, 150 mM NaCl, 5 mM EDTA, and 0.5% NP-40 detergent. Western blotting of the proteins that had bound the immunoprecipitation beads or proteins from the lysates was performed using polyclonal antibodies (Affinity Bioreagents) specific for the GR.
Cell cycle analyses were performed by using green fluorescent protein (GFP) as a marker to distinguish transfected and untransfected cells. Experiments were performed with pEGFP expression vector (Clontech) cotransfected at a 1:8 ratio with an HA expression vector cloned in frame with Vpr or mutants of Vpr. 293T cells were prepared for cell cycle analysis by initial trypsinization followed by fixation for 30 min in 2% formaldehyde. The cells were washed with PBS and treated with RNase A (1 mg/ml) and propidium iodide (10 μg/ml) in PBS for 30 min. Cellular DNA content in the transfected (GFP+) and untransfected (GFP−) cells was assessed with a FACScan flow cytometer (Becton Dickinson).
The potential action of Vpr as an SRC was compared to coactivation by GR-interacting protein 1 (GRIP1), a well-characterized SRC (22, 23). Both GRIP1 and Vpr significantly enhanced luciferase activity driven by the GRE in the presence of increasing amounts of dexamethasone, with a constant amount of transfected GRIP1 or Vpr expression plasmid together with the endogenous GR expressed in 293T cells (Fig. (Fig.1).1). These proteins induced a similar shift in the coactivation curve, although Vpr exhibited roughly 50% of the activity of GRIP1. In several replicates, Vpr produced a three- to sixfold greater GR-mediated response than was obtained in cells treated with similar concentrations of dexamethasone and control vector. The weak induction of GRE obtained in the latter condition presumably reflects the action of an endogenous coactivator.
Subsequently, we examined the ability of Vpr to produce a dose-dependent coactivation effect and the capacity of RU486 to inhibit such a response (Fig. (Fig.2).2). Transfecting increasing amounts of Vpr in the presence of a constant amount of total DNA induced a dose-dependent increase in endogenous GR coactivation. Furthermore, this coactivation was specifically inhibited by the introduction of the RU486 antagonist when used at a concentration higher than that of dexamethasone. We controlled for nonspecific Vpr transactivation through induction of G2 cell cycle arrest (7, 14, 59) by including a Rous sarcoma virus (RSV) promoter-driven, β-galactosidase reporter in all cell transfections. This reporter is activated by G2 arrest and modestly induced by Vpr (7). Accordingly, differences occurring secondary to the G2 arrest or transfection efficiency could be accounted for by normalizing the GRE-driven luciferase activity to overall galactosidase activity. Together, these data suggest that Vpr exerts a dose-dependent effect on the GRE independent of changes related to G2 cell cycle arrest.
We further established the specificity of the Vpr-mediated GR coactivation by examining the dependence on a GRE. In the presence of three GRE repeats, HIV-1 Vpr caused an almost 3-fold induction over similarly hormone-treated cells transfected with a background vector and a 16-fold increase over non-hormone-treated cells (Fig. (Fig.3A).3A). The small amount of basal coactivation seen may have been due to residual hormones remaining in the charcoal-treated serum supplement or more likely represents oscillations of basal expression, as there was no coactivation above the vector control. When the hormone response element was absent, parallel experiments revealed that coactivation was abolished in the presence and absence of the agonist. We next examined whether the analogous proteins of HIV-2 were likewise able to coactivate the GR. Vpx has no identifiable leucine-rich domain, while Vpr from HIV-2 (Vpr2), which can induce G2 cell cycle arrest, contains a single domain homologous to HIV-1 Vpr that is centered on the pentamer LQRAL. Vpr2 displayed about 50% of the coactivator activity of HIV-1 Vpr, while Vpx had no coactivation potential despite being expressed at higher levels (data not shown). Together, these findings indicate that Vpr2 but not Vpx has a moderate GR coactivation function that is both hormone and GRE dependent.
As a final control, we tested whether the Vpr coactivation was receptor dependent (Fig. (Fig.3B).3B). The monkey kidney cell line CV-1B lacks endogenous cortisol receptors (33). Using these cells, the Vpr coactivator response did not occur unless GR was also provided during transfection. In the presence of GR, a 2.5- to 3-fold enhancement of GRE-luciferase activity was observed. Notably, Vpr is unable to induce G2 cell cycle arrest in CV1-B cells, again providing strong evidence that cell cycle alterations are not solely responsible for the induction of the GRE.
We next sought to confirm that the LQQLL motif present in HIV-1 Vpr at residues 64 to 68 was responsible for mediating Vpr-induced glucocorticoid coactivation and to further distinguish the G2 arrest phenotype of Vpr from its coactivation properties. We also examined the possibility that a more amino-terminal LLEEL pentamer might be involved in coactivation. While this latter motif is “reversed” in orientation from the classical SRC signature motif, such determinants have been implicated in coactivation by known SRCs (Beatrice Darimont, personal communication). Furthermore, SRCs such as GRIP1 contain multiple LXXLL motifs that function in concert and may serve as a model for Vpr-mediated functions. Table Table11 summarizes the ability of transfected Vpr to induce G2 cell cycle arrest when these two leucine-rich α-helices were mutated individually or in tandem. While no leucine in the amino-terminal helix was involved in G2 arrest, the L64A (A64QQLL) mutation alone disrupted the ability of Vpr to delay cells at the G2 checkpoint. These findings are consistent with data from Mahalingam et al. (36) but disagree with the results described by Kino et al. (28), who found that this mutation abolished coactivation without affecting G2 arrest. The single L67A (LQQA67L) and L68A (LQQLA68) substitution mutations did not affect the arrest phenotype, but the composite LQQA67A68 mutation did. Vpr2 contains an analogous leucine-rich motif (LQRAL) that is similar to that of the LQQA67L mutant of HIV-1 Vpr, and the wild-type form is able to cause G2 arrest. Of special note, Vpr mutants that retained the ability to induce G2 arrest were consistently expressed at much higher levels than those that failed to induce arrest. In the analysis of coactivation, these different levels of Vpr expression must be carefully considered. We chose to compare Vpr mutants that retained G2-arresting properties.
The LQQA67L and LQQLA68 mutations of Vpr in the carboxy leucine-rich motif (H2) only partially reduced coactivation potential (approximately 50%) (Fig. (Fig.4A).4A). Therefore, we had further evidence prompting us to examine the potential contribution of the amino-terminal leucine-rich motif (H1) to the coactivation response. Mutation of all three leucines in the first helix (A22A23EEA26) did not compromise the ability of Vpr to induce G2 arrest but did reduce GR coactivation by approximately 50%. When the A22A23EEA26 and LQQA67L mutations were combined (A22A23EEA26/LQQA67L), the G2 arrest phenotype was maintained but coactivation was abolished (Fig. (Fig.4B).4B). Thus, both leucine-rich domains appear to cooperate to facilitate GR coactivation. A similar coactivation pattern is seen with GRIP1, in which multiple LXXLL motifs contribute to SRC function (9, 47).
Subsequently, we assessed the physical interaction of Vpr and GR to further address the role of each leucine-rich motif in coactivation. Wild-type Vpr and GRIP1 can bind to the GR under similar conditions, whereas the protein phosphatase 2A regulatory subunit Aα, a control protein, did not bind. Mutations that replaced key leucines in both motifs interfered with the ability of Vpr to bind to GR (Fig. (Fig.5).5). Of note, the addition of exogenous glucocorticoid was not required for coimmunoprecipitation of either Vpr or GRIP1 with GR. These results suggest either that sufficient endogenous hormone is present in serum-supplemented medium or that this binding occurs in the absence of hormone when these proteins are expressed at high levels.
We next addressed whether the Vpr proteins encoded by patient-derived viruses shared the GR coactivation properties observed with NL4-3-derived Vpr. Vpr was amplified from PBMCs isolated from two HIV-infected patients and inserted into pUC-19, whereupon multiple clones were sequenced (Fig. (Fig.6A).6A). Both leucine-rich motifs were found to be highly conserved among the quasispecies obtained from these two patients. One clone (2.28) had a single leucine mutation with a phenylalanine at position 68, but the coding sequence of this clone was interrupted by a stop codon (Z) at amino acid 9, thereby removing the pressure for sequence conservation. Another clone from the same patient (2.07) had a deletion in the first helix, but expression of this clone was not detectable by Western blot analysis. Four Vpr alleles (1.06, 1.23, 2.10, and 2.12) containing conserved leucine motifs were tested for the ability to coactivate endogenous GR in 293T cells by cotransfection with the GRE-luciferase reporter. They displayed nearly the same degree of coactivation as the NL4-3 Vpr clone (4- to 6-fold, magnified by the addition of dexamethasone to 40- to 60-fold) (Fig. (Fig.6B),6B), despite lower levels of protein expression (Fig. (Fig.6C).6C).
HIV-1 contains several accessory proteins that play various roles in the viral life cycle. Vpr is highly conserved in vivo and has been implicated in accelerating virus production by facilitating entry of the preintegration complex into the nuclei of nondividing cells and by causing G2 cell cycle arrest of infected T cells, leading to increased LTR transcription. Although transactivation by Vpr correlates with its ability to cause G2 arrest, we now show that Vpr-mediated coactivation of the GR is a distinct phenotype, confirming and extending the work of Kino and colleagues (28). In addition to the LQQLL motif present in the carboxy-terminal helix, we find a contribution of the amino-terminal LLEEL motif to the coactivation response. A search of the GenBank database revealed that both of the leucine-rich domains are highly conserved among various patient isolates. In fact, we have now sequenced over 100 intact Vpr quasispecies from 10 patients and found that both helices are completely conserved in all but one intact isolate.
Vpr likely multimerizes through interactions at the amino-terminal leucine-rich domain (64) and binds heterologous proteins through interactions at the carboxy-terminal leucine domain (63). Vpr is a relatively small protein, and mutations throughout its 96 amino acids interfere with its various phenotypes differentially (36). While each of the two leucine-rich domains may promote an interaction with a specific binding partner, our finding that the L68A (LQQLA68) mutant of Vpr retained G2-arresting properties except when it also had mutations in the first leucine-rich helix implies that the two helices may cooperate to promote arrest. While mutations of each α-helix interfered with binding of Vpr to GR (Fig. (Fig.5)5) and partially compromised the coactivation function, we cannot distinguish between the possibilities that the mutations interfere with direct binding to GR and that multimerization of Vpr is required for coactivation. In this regard, it has been shown that the binding domain of Vpr for the p6 region of the Gag precursor resides in the amino terminus but is affected by mutations in the carboxy terminus (60). It should be noted that Vpr binding to Gag also involves an LXXLF motif in p6 (29, 34).
Vpr can be detected in the serum of HIV-1-infected patients, and exogenously applied, recombinant Vpr has been shown to enhance virus production from various latently infected cell lines and PBMCs (30, 31). Extracellular Vpr is thought to have several potential effects in vivo, including enhancement of virion production. Even in replication-defective forms of the virus, there is sufficient Vpr in the virion to induce G2 arrest in the absence of de novo Vpr synthesis within the infected cell. This finding has led some to hypothesize that Vpr may play a role in immune suppression involving non-productively infected cells (41). Of note, a similar function has been ascribed to glucocorticoids. Since Vpr in the virion can induce G2 cell cycle arrest, which in turn may lead to programmed cell death, it may be that Vpr depletes cell populations by inducing apoptosis, an effect also known to be mediated by steroids.
Recombinant Vpr possesses glucocorticoid-like activity that can be inhibited by antibodies to Vpr or by the glucocorticoid antagonist RU486 (30, 31, 44). Specifically, recombinant Vpr can act as a glucocorticoid on cultured cells by inducing apoptosis, increasing virion production from U937 cells, and downregulating NF-κB through an upregulation of IκB synthesis. However, it is now becoming clear that glucocorticoids may not modulate IκB expression in order to exert their effects (5, 10, 27, 45). Nevertheless, the parallels between glucocorticoid and recombinant Vpr action are intriguing given the fact that Vpr can function as an SRC. This is particularly interesting with respect to increased virus production in HIV-infected cultured H9 T cells in the presence of Vpr and glucocorticoids (37, 49). In fact, the HIV-1 provirus contains an intact and functional GRE within the vif open reading frame (49).
These observations have led to the proposal that glucocorticoid antagonists be used as antiretroviral agents (3, 44). However, Soudenyns and Wainberg (50) have been quick to point out that the effects of glucocorticoids on HIV replication in vitro and in vivo are diverse and varied and should be further explored at the molecular level before any such agents are used clinically. We have been unable to show any effect of glucocorticoids or RU486 on HIV-1 replication in cultured H9 cells, PBMCs, or primary macrophages (data not shown). Likewise, isogenic viruses lacking only the Vpr open reading frame replicated with kinetics similar to those of the wild-type constructs despite the addition of exogenous dexamethasone or RU486. Since Vpr may also coactivate hormone receptors for estradiol and progesterone (28), perhaps another agonist or antagonist hormone ligand might have more profound effects on virus production. Alternatively, it is likely that the plethora of effects exerted by glucocorticoids in vivo on T-cell function and virus-host cell interactions as well as the effects of circulating Vpr cannot be accurately recapitulated with in vitro culture systems. While it is clear that Vpr can function as an SRC, further studies are necessary to determine whether this function contributes to the spread or production of HIV or to the pathogenesis of AIDS.
We thank Michael Stallcup for providing the GRIP1 plasmid and Beatrice Darimont for sharing unpublished results. We also thank J. Lo and M. Schambelan for providing clinical specimens.
This work was supported by the UCSF-GIVI Center for AIDS Research, grant NIH P30 MH59037, and SFGH GCRC grant M01 RR00083.