The human immunodeficiency virus type 1 Rev protein: ménage à trois during the early phase of the lentiviral replication cycle

Introduction

The retrovirus replication cycle starts with the entry of the virus into the host cell, followed by reverse transcription of the genomic viral RNA into viral cDNA. After translocation of the double-stranded DNA into the nucleus and integration into the host-cell genome, the early phase of the virus replication cycle is completed. Transcription of the integrated proviral DNA marks the start of the late phase, which also includes the subsequent splicing and trafficking events of the viral RNAs as well as translation, assembly and budding (Freed, 2001). In addition to the gag, pol and env genes common to all members of the family Retroviridae, lentiviruses also encode a number of accessory and regulatory proteins that regulate different steps of the lentiviral replication cycle in a mostly pleiotropic manner (reviewed by Malim & Emerman, 2008; Nekhai & Jeang, 2006).

Known functions of Rev in the late phase of the replication cycle

One of the additional proteins, the viral Rev protein, initially identified as a regulator of expression of virion proteins, was found to be required for expression of the viral structural proteins Gag, Pol and Env from the integrated proviral DNA. By binding to the Rev-responsive element (RRE), an RNA structure present on the unspliced RNA encoding Gag and GagPol and on singly spliced RNAs encoding Env, Rev tethers these transcripts to the cellular CRM-1-mediated nuclear-export pathway, leading to enhanced cytoplasmic levels of these RNAs and increased expression of the encoded proteins. However, this nuclear-export function does not seem to be the only effect of Rev during the late phase of the virus replication cycle. Consistent with an enhanced polysomal association of the RRE-containing RNAs, Rev was also found to stimulate protein expression levels to a greater extent than the cytoplasmic mRNA levels, suggesting an independent stimulatory effect of Rev on translation (recently reviewed by Grewe & Überla, 2010; Groom et al., 2009). More recently, we also observed that Rev enhances encapsidation of the genomic RNA into virions to a much larger extent than it affects cytoplasmic RNA levels (Blissenbach et al., 2010; Brandt et al., 2007). Nuclear RNA export, the corresponding modulation of the ratio of differentially spliced transcripts in nucleus and cytoplasm, the increase in translational efficiency, and stimulation of encapsidation are all exerted during the late phase of the virus replication cycle and, until now, no Rev mutants have been reported to be differentially impaired for one or several of these functions. Thus, it is conceivable that all of these functions are linked mechanistically to Rev's modulation of the ribonucleoprotein complex formed in the nucleus on RRE-containing RNAs (Grewe & Überla, 2010).

Novel function of Rev in the early phase of the replication cycle

In addition to its function during the late phase of the replication cycle, Rev was recently also found to inhibit the integration of the viral genome and might thus play a role in prevention of cellular superinfection. Exposure of susceptible cell lines to human immunodeficiency virus type 1 (HIV-1) at a high m.o.i. leads to a large number of reverse-transcribed viral cDNA copies per cell, but only one or two integration events are observed (Butler et al., 2001). Levin et al. (2009a) showed that the integration frequency in newly infected target cells is regulated by Rev. In their experiments, target cells were infected with either wild-type HIV-1 or a rev-deficient HIV-1 produced in Rev-expressing cells. Although similar numbers of viral DNA copies per target cell were observed for both viruses, infection with the rev-deficient virus resulted in an at least fivefold higher number of integration events (Levin et al., 2009a, 2010b). Consistent with an inhibitory effect of Rev on integration, the integration frequency of both viruses was reduced in target cells stably expressing Rev (Levin et al., 2009a). The inhibitory effect of Rev on the second virus was seen 4–8 h after infection with the first virus, coinciding with detection of Rev expression (Levin et al., 2010b). Although it has been shown that Rev can be expressed from integration-deficient HIV-1 mutants, it remains unclear whether Rev is expressed from unintegrated, as proposed by Levin et al., (2009a), or rapidly integrated proviral DNA after co-infection with integration-competent HIV-1. Interestingly, the Rev M10 mutant, which is a transdominant inhibitor of Rev's late-phase functions, conferred an integration frequency comparable to that of wild-type Rev (Levin et al., 2009a). Thus, the inhibitory effect of Rev during the early phase of the replication cycle can be separated genetically from the late-phase functions of Rev. However, it must be noted that a Rev mutant that is inactive in the early phase but active in the late phase is still to be discovered.

Mechanisms of the regulatory role of Rev during the early phase

What could be the mechanism behind this early-phase effect of Rev? As viral cDNA synthesis itself is not affected, Rev could inhibit either steps after cDNA synthesis but before integration, or the integration process itself. Reduction of the stability and incorrect trafficking of the pre-integration complex by Rev have, to our knowledge, not been studied and, therefore, the possibility that these potential mechanisms contribute to Rev's function during the early phase of the replication cycle cannot be excluded.

Substantial evidence has been accumulated for a direct inhibitory effect of Rev on the integration process. Rev can interact with integrase (IN) as demonstrated by in vitro binding assays, intracellular bimolecular fluorescence complementation in yeast cells and co-immunoprecipitation from different transfected or infected human cells (Levin et al., 2009a, 2010a; Rosenbluh et al., 2007). This interaction can be blocked by specific Rev- and IN-derived peptides, also suggesting that these peptide sequences are part of the binding interface of Rev and IN, respectively (Hayouka et al., 2008; Levin et al., 2009a, b, 2010a; Rosenbluh et al., 2007). In vitro assays revealed inhibition of the enzymic activity of IN by association with Rev or the Rev-derived peptides (Hayouka et al., 2008; Levin et al., 2009b; Rosenbluh et al., 2007). This inhibitory effect of Rev could be blocked by the two IN-derived peptides, which disrupt the Rev–IN interaction by binding to Rev (Levin et al., 2009b). Even more important, in virus-infected cells the Rev-derived peptides induced a defect in integration, whereas the IN-derived peptides stimulated integration, similar to the situation after infection with a rev-deficient virus (Levin et al., 2009a, b; Rosenbluh et al., 2007).

However, IN is not the only target by which Rev could mediate its suppressive effect on the integration frequency. The LEDGF/p75 protein has been shown to be an important cellular co-factor of integration, presumably by tethering the pre-integration complex to the host chromatin via direct interaction with IN and/or inhibition of proteasomal degradation of IN (recently reviewed by Engelman & Cherepanov, 2008). Rev also binds to LEDGF/p75 in a number of different in vitro and infectivity assays, and the interaction of Rev with LEDGF/p75 can be blocked by LEDGF/p75- and Rev-derived peptides (Levin et al., 2010a). In vitro protein–protein binding experiments with different peptides as competitors revealed that Rev and IN compete for the same binding site on LEDGF/p75. Consistently, the same LEDGF/p75-derived peptides can prevent the formation of the Rev–LEDGF/p75 and IN–LEDGF/p75 complexes (Levin et al., 2010a). Whilst Rev and IN bind to the same site of LEDGF/p75, the interaction between Rev and IN is mediated by regions of the proteins differing from their LEDGF/p75-binding sites. In this ménage à trois, each of these three proteins can interact with the other two proteins. Hence, Rev was also shown to promote dissociation of the IN–LEDGF/p75 complex. A model in which Rev binds to IN and LEDGF/p75, thereby blocking IN activity and preventing tethering of the pre-integration complex to the host-cell chromosome, could well explain Rev's suppressive effect on the integration frequency. It is important to note that this function was also mediated by the M10 mutant of Rev, which is a transdominant inhibitor of the functions of Rev during the late phase of the replication cycle (Levin et al., 2010a).

Potential benefits of Rev-mediated inhibition of integration for HIV-1 replication

What selective advantage could the inhibition of an essential step in the lentiviral replication cycle by Rev provide to HIV-1? As Rev is presumably expressed at sufficiently high levels only after the encoding genome has integrated, prevention of superinfection of the infected cell with subsequent viruses seems to be a plausible function (Nethe et al., 2005). In particular, reinfection of an infected cell by viruses released from the same cell seems to contribute little to virus fitness. Downregulation of surface receptors is a common mechanism used by different viruses to ensure efficient release of virus progeny from an infected cell. Expression of neuraminidase by influenza viruses is a classic example (reviewed by Wagner et al., 2002). For HIV, the viral Env, Nef and Vpu proteins have also been shown to prevent superinfection by downmodulation of the CD4 receptor (reviewed by Levesque et al., 2004; Lindwasser et al., 2007). However, as Rev's inhibition of integration blocks superinfection at a post-entry step, at which the virions have already lost their infectivity, this mode of blocking superinfection should not promote efficient release of progeny viruses. Thus, the main function of Rev's suppressive effect on the integration frequency seems to be to maintain the viability of the infected cell by limiting the number of integration events in each cell, thereby preventing genotoxicity. Entry of larger numbers of virions into the same cell by formation of a viral synapse (Jolly & Sattentau, 2004; Jolly et al., 2004; McDonald et al., 2003) and high local concentrations of infectious virions in lymphoid tissues (Fox et al., 1991) could indeed lead to excessive integration. Gratton et al. (2000) obtained evidence for multiply infected cells in germinal centres, but the PCR analyses performed do not allow discrimination between integrated and non-integrated viral DNA. In cell-culture experiments, Levin et al. (2009a) observed a correlation between the integration frequency and cell death after infection. Infection of cells with a rev-deficient virus resulted in a high integration frequency, but no infectious HIV-1 was produced from the infected cells. Despite this unproductive infection, cell death was observed in up to 40 % of the cells (Levin et al., 2009a). Infection of cells with HIV-1 encoding the RevM10 mutant or infection with a pseudotyped HIV-1 env-deletion mutant expressing Rev led to a reduced integration frequency, coinciding with higher cell viability. However, the higher number of integration events in cells infected with the rev-negative mutant might not only increase genotoxicity. Due to the higher number of integrated viral DNA copies, the Rev-independent Tat and Nef proteins could be expressed at higher levels in cells infected with the rev-deficient virus. Thus, the possibility that the observed increase in cytotoxicity in the absence of Rev is due to higher expression levels of potentially cytotoxic, Rev-independent genes cannot be excluded. However, at present, the hypothesis that the higher number of integration events leads to genotoxic effects seems more plausible.

Whilst limiting cytotoxicity can clearly increase the efficiency of virus replication, it is less clear whether preventing superinfection without enhancing virus release confers an evolutionary advantage. On the one hand, inhibition of superinfection might have evolved as a mechanism by which HIV-1 variants compete with each other for dominance in the viral quasispecies population within each infected individual. On the other hand, computer simulations suggest that superinfection speeds up the development of sequence diversity by facilitating recombination between different viruses (Bocharov et al., 2005), thereby potentially favouring immune escape and other adaptation processes (reviewed by Blackard et al., 2002; Burke, 1997; Najera et al., 2002). Thus, blocking superinfection might actually provide an evolutionary disadvantage to the virus.

This raises the question of whether blocking of integration by Rev indeed prevents superinfection. Expression of the early HIV-1 tat, rev and nef genes after infection with integration-deficient HIV-1 has been detected repeatedly, but expression levels seem to be too low to promote late-gene expression and production of infectious particles (reviewed by Wu, 2004). More recently, however, it has been demonstrated that integration-deficient mutants can be rescued by co-infection with wild-type HIV-1 (Gelderblom et al., 2008). Co-infection with wild-type virus led to late-gene expression from the unintegrated HIV-1 mutant DNA and even transfer of genomic RNA of the IN-deficient mutant into target cells at an unexpectedly high frequency. Levin et al. (2009a, 2010b) demonstrated clearly that expression of Rev by the first infecting HIV-1 reduces subsequent integration events by a second HIV-1. However, in these experiments, replication of the second virus has not been analysed. If the only effect of Rev during the early phase of the replication cycle is to block integration to prevent genotoxicity, then the data of Gelderblom et al. (2008) would actually indicate that the second virus can replicate without integration and therefore contribute to the development of sequence diversity.

Current model for the role of Rev during the early phase of the HIV-1 replication cycle

In conclusion, we are left with two models for the role of Rev during the early phase of the HIV-1 replication cycle (Fig. 1⇓). In both models, the reverse-transcribed cDNA of the first HIV-1 virion is translocated to the nucleus. In the absence of Rev, IN interacts with chromatin-bound LEDGF/p75, leading to integration, transcription and assembly of new virions. Once Rev is expressed from either non-integrated or already-integrated HIV-1 DNA, the situation for the next HIV-1 entering the cell is changed. By an interaction of Rev with the IN of the incoming pre-integration complex of the second virus and by binding of Rev to LEDGF/p75, integration of the second HIV-1 DNA is inhibited (Fig. 1⇓, second virus). This probably prevents genotoxicity due to multiple integration events. The two models now differ in the inhibition of superinfection. In the first model (Fig. 1a⇓), the non-integrated DNA is not transcribed efficiently, leading to an abortive infection of the second virus. This would reduce genotoxicity by limiting the number of integration events, but impair superinfection-dependent sequence diversification of HIV-1. In the second model (Fig. 1b⇓), viral regulatory proteins expressed by the integrated genome of the first HIV-1 activate transcription and translation from the non-integrated genome of the second HIV-1. Genomic RNA from both viruses is encapsidated into budding particles, leading to release of infectious virus, transferring the genomes of both viruses. According to this model, Rev would reduce genotoxicity by limiting the number of integration events without impairing superinfection-dependent sequence diversification. However, whether Rev-mediated inhibition of integration indeed allows rescue of the second virus by the first awaits further elucidation.