Research Article: The Lentiviral Integrase Binding Protein LEDGF/p75 and HIV-1 Replication

Date Published: March 28, 2008

Publisher: Public Library of Science

Author(s): Alan Engelman, Peter Cherepanov, B. Brett Finlay.


Retroviral replication proceeds through a stable proviral DNA intermediate, and numerous host cell factors have been implicated in its formation. In particular, recent results have highlighted an important role for the integrase-interactor lens epithelium-derived growth factor (LEDGF)/p75 in lentiviral integration. Cells engineered to over-express fragments of LEDGF/p75 containing its integrase-binding domain but lacking determinants essential for chromatin association are refractory to HIV-1 infection. Furthermore, both the levels of HIV-1 integration and the genomic distribution of the resultant proviruses are significantly perturbed in cells devoid of endogenous LEDGF/p75 protein. A strong bias towards integration along transcription units is a characteristic feature of lentiviruses. In the absence of LEDGF/p75, HIV-1 in large part loses that preference, displaying concomitant integration surges in the vicinities of CpG islands and gene promoter regions, elements naturally targeted by other types of retroviruses. Together, these findings highlight that LEDGF/p75 is an important albeit not strictly essential cofactor of lentiviral DNA integration, and solidify a role for chromatin-associated LEDGF/p75 as a receptor for lentiviral preintegration complexes. By now one of the best characterized virus–host interactions, the integrase-LEDGF/p75 interface opens a range of opportunities for lentiviral vector targeting for gene therapy applications as well as for the development of novel classes of antiretroviral drugs.

Partial Text

A key step in the retroviral lifecycle is the formation of the provirus, the integrated form of the viral cDNA that is produced during reverse transcription. Retroviral integration is promoted by the viral integrase (IN) enzyme, which enters the cell as a component of the virion particle. IN catalyzes two spatially and temporally distinct reactions within the context of the preintegration complex (PIC), a large structure derived from the virus core [1],[2]. During the initial reaction, which is called 3′ processing and happens soon after the cDNA is made, IN hydrolyzes a dinucleotide from each end of HIV-1 DNA [2],[3] (Figure 1). The second reaction, DNA strand transfer, takes place at the site of integration in the cell nucleus. IN uses the recessed 3′-OH groups created during 3′ processing to cut opposing strands of chromosomal DNA in a staggered fashion, concomitantly connecting the viral DNA 3′ ends to the generated 5′ overhangs [4]. The resultant DNA recombination intermediate harbors single-strand discontinuities that must be repaired to complete provirus formation (Figure 1). See [5] for a thorough overview of the mechanism of HIV-1 integration as well as the host cell factors that are implicated in the final DNA repair step.

LEDGF/p75, a member of the hepatoma-derived growth factor (HDGF) related protein (HRP) family, was initially implicated in lentiviral biology through its association with ectopically expressed HIV-1 IN in 293T cells [19]. Significantly, purified recombinant LEDGF/p75 protein stimulated HIV-1 IN catalytic function in vitro [19]. The interaction was independently discovered by analyzing proteins associated with HIV-1 IN in HeLa cells [5] and in a yeast two-hybrid screen for HIV-1 IN interactors [20].

HIV-1 IN is composed of three functional domains: the N-terminal domain (NTD), the catalytic core domain (CCD), and the C-terminal domain (CTD) (Figure 2B). Initial mapping experiments using fluorescent fusions expressed in live cells revealed that the CCD is minimally required for the interaction with LEDGF/p75, and highlighted a role for the NTD as an affinity enhancer [28]. A number of single amino acid substitutions within the CCD, including V165A [5], R166A [51], and Q168A [20], were soon thereafter shown to impair the IN–LEDGF/p75 interaction. Each of these changes rendered HIV-1 replication defective [20], [52]–[54], suggesting that the IN–LEDGF/p75 interaction might be essential for HIV-1 replication [20],[51],[55]. However, many mutations in IN exert so-called “class II” pleiotropic effects, whereby poorly understood aspects of IN biology extending beyond its innate catalytic function contribute to the overall replication defect [55]–[57]. Recent results indicate that PICs formed in the absence of LEDGF/p75 protein in vivo are fully competent to integrate the endogenous cDNA made during reverse transcription into exogenous target DNA in vitro [18]. Based on this, one would predict that IN mutant viruses defective for growth solely due to the inability to interact with LEDGF/p75 would yield PICs fully competent for integration in vitro. The replication defect caused by the Q168A mutation was suggested to result from the lack of cofactor binding [20], though a follow-up study indicated HIV-1Q168A behaved as a class II IN mutant virus [55]. PICs derived from class II mutants fail to support in vitro integration activity [58], indicating that PIC analyses would help to shed light on the specificity of the HIV-1Q168A replication defect.

Initial RNAi-based studies were central to establish important links between endogenous LEDGF/p75 protein and lentiviral IN expression levels and subcellular localization, yet they failed to reveal an important role for the cell factor in HIV-1 replication. In some RNAi-based studies, despite achieving what appeared to be very efficient reductions in cellular protein [30],[61], specific HIV-1 replication defects were not observed despite rigorous effort to identify them. At the time it was suggested that intracellular LEDGF/p75 levels might significantly exceed those required to effect normal lentiviral DNA integration [61], a hypothesis supported by subsequent RNAi-based work [16],[62],[63] and a gene knockout study [18]. Llano and colleagues [16] performed an elegant study whereby the expression of short-hairpin RNA (shRNA) was linked to that of green fluorescence protein (GFP) within the same lentiviral-based vector. Sorting the brightest green cells therefore ensured for selection of potent LEDGF/p75 knockdowns. Selected cells were moreover fractionated to analyze levels of chromatin-bound protein. In this way, HIV-1 infectivity levels were correlated to residual levels of chromatin-associated LEDGF/p75. In the absence of detectable protein, infection was reduced to 3.5% of that observed in the presence of normal LEDGF/p75 levels. Similarly significantly reduced levels of HIV-1 infection were observed in mouse embryo fibroblasts (MEFs) derived from LEDGF knockout as compared to littermate control animals [18]. The block in both cases was at integration: reverse transcription and the formation of two long terminal repeat (LTR)–containing DNA circles, a surrogate marker for PIC nuclear import, were normal, whereas integration was severely reduced [16],[18]. Although these results would seem to exclude a role for LEDGF/p75 in the nuclear import of the PIC, the baffling ability of lentiviruses to infect non-dividing cell types with high efficiency calls for further scrutiny of lentiviral PIC nuclear import in LEDGF-depleted cells under conditions of growth arrest. It is important to stress that even in a genetic knockout model completely devoid of LEDGF/p75 protein, HIV-1 integration was not ablated: LEDGF-null MEFs supported ∼10% of the level of HIV-1 integration achieved in control cells [18]. Hence, although important for lentiviral integration, LEDGF/p75 is clearly not strictly essential. The infectivity of Moloney murine leukemia virus (Mo-MLV), a γ-retrovirus, importantly did not depend on the presence of LEDGF/p75 in target cells [16],[18], providing biological relevance to the observations of lentiviral specificity in the IN–LEDGF/p75 interaction.

Establishment of the stably integrated provirus is a hallmark of retroviral replication, fundamental to the persistence of infection. However, mammalian genomic DNA is a vast target, a significant proportion of which is not transcriptionally active. What’s more, when integrated, the transcriptional activity of viral cDNA becomes sensitive to the local chromosomal environment [64]. Hence, it is not surprising that retroviruses do not leave integration entirely to chance, having evolved mechanisms for selecting suitable target loci. Indeed, the observed distributions of integrated proviruses along host chromosomal DNA are not random, and biases at the level of local DNA sequences as well as on the genomic scale have been described (reviewed in [13] and [50]). These biases appear to be genus-specific, and although the differences are sometimes subtle, three retroviral genera appear to stand out most distinctly. Lentiviruses, including HIV (both type 1 and type 2) [65],[66], simian immunodeficiency virus (SIV) [67],[68], FIV [69], and equine infectious anemia virus (EIAV) [70], are strongly biased towards integration into transcription units (TUs), with a preference for highly expressed genes. The γ-retrovirus Mo-MLV, in contrast to lentiviruses, prefers to integrate in the vicinity of transcription start sites and CpG islands [67],[71], while a spumaretrovirus, simian foamy virus (SFV), is biased against integrating into genes, yet nonetheless displays significant preferences for gene start sites and CpG islands [72],[73].

The question of how important LEDGF/p75 is to HIV-1 replication has caused fierce debates and remained controversial until recently. On one hand, viral replication defects caused by mutations in and around the LEDGF/p75-binding interface of HIV-1 IN indicated that the cofactor might play an essential role [20],[51],[55]. On the other hand, LEDGF/p75 depletion via RNAi significantly reduced but fell short of abrogating HIV-1 integration [16],[62],[63]. Since LEDGF knockout cells supported residual levels of HIV-1 integration, we must conclude that the cofactor is not essential for integration [18]. We concede that it is possible that some additional mechanisms rescue HIV-1 integration in the mouse knockout system, or that murine HRP2 takes over the role of LEDGF/p75. Nevertheless, these results are in agreement with a large body of experimental evidence demonstrating that the isolated HIV-1 IN protein can perform its catalytic functions [7],[10] (reviewed in [5] and [13]). Furthermore, PICs assembled in LEDGF-null cells were fully functional in vitro, indicating that the cofactor is not essential for the assembly or intrinsic activity of the HIV-1 complex. Though the endogenous cellular protein readily co-immunoprecipitated lentiviral PICs [30], the equivalent activities of PICs isolated from normal and LEDGF-null cells strongly suggest that it is the chromatin-bound pool of the protein that is functionally relevant (Figure 4).

LEDGF/p75 is an important host factor commandeering the integration of HIV-1 and likely other lentiviruses to active TUs [14],[18]. Although LEDGF/p75 was required for efficient HIV-1 integration and replication [62],[63], it is not essential, since stringently knocked-down human SupT cells and knockout MEFs supported residual provirus formation (approximately 10% of that seen in the presence of endogenous LEDGF/p75 levels) [16],[18]. Therefore, it appears that HIV-1 does not entirely rely on the host factor, and furthermore, it can be assumed that there is a background of LEDGF/p75-independent integration under normal infection conditions. Conceivably, integration into transcriptionally repressed or gene-poor regions contributes to the establishment of latent viral reservoirs and hence to the persistence of clinical infection [86]. It remains to be determined if the residual levels of integration in LEDGF-null cells depend upon HRP2, a close kin of LEDGF/p75.