Virus Evolution Reveals an Exclusive Role for LEDGF/p75 in Chromosomal Tethering of HIV

Retroviruses by definition insert their viral genome into the host cell chromosome. Although the key player of retroviral integration is viral integrase, a role for cellular cofactors has been proposed. Lentiviral integrases use the cellular protein LEDGF/p75 to tether the preintegration complex to the chromosome, although the existence of alternative host proteins substituting for the function of LEDGF/p75 in integration has been proposed. Truncation mutants of LEDGF/p75 lacking the chromosome attachment site strongly inhibit HIV replication by competition for the interaction with integrase. In an attempt to select HIV strains that can overcome the inhibition, we now have used T-cell lines that stably express a C-terminal fragment of LEDGF/p75. Despite resistance development, the affinity of integrase for LEDGF/p75 is reduced and replication kinetics in human primary T cells is impaired. Detection of the integrase mutations A128T and E170G at key positions in the LEDGF/p75–integrase interface provides in vivo evidence for previously reported crystallographic data. Moreover, the complementary inhibition by LEDGF/p75 knockdown and mutagenesis at the integrase–LEDGF/p75 interface points to the incapability of HIV to circumvent LEDGF/p75 function during proviral integration. Altogether, the data provide a striking example of the power of viral molecular evolution. The results underline the importance of the LEDGF/p75 HIV-1 interplay as target for innovative antiviral therapy. Moreover, the role of LEDGF/p75 in targeting integration will stimulate research on strategies to direct gene therapy vectors into safe landing sites.


Introduction
Although a dynamic interplay between the virus and host factors occurs at all steps of the HIV replication cycle, the identity and role of these cellular cofactors are poorly understood. LEDGF/p75 is a cellular cofactor of HIV-1 integrase (IN), which catalyses integration of the HIV-1 DNA into the host cell genome [1]. LEDGF/p75 was found to interact with HIV-1 IN following coimmunoprecipitation of nuclear extracts of cells stably expressing IN from a synthetic gene [2]. The tight binding between LEDGF/p75 and IN was corroborated independently [3,4] and found to be lentivirus specific [5][6][7]. The protein of 530 amino acids contains a PWWP domain at the N terminus that may be involved in chromatin binding [8,9]. The interaction of LEDGF/p75 with IN is mediated by a minimal IN-binding domain (IBD) [10,11], in the C-terminal region of the protein (amino acids 347 to 429). LEDGF/p75 appears to function as a tethering factor, linking IN to the chromatin [12,13]. A recent report suggested a modest contribution of LEDGF/p75 to the targeting of HIV-1 integration, implying a role for alternative cofactors [14].
Several studies have addressed the question whether LEDGF/p75 is an important cofactor for viral replication in vivo. The Q168 residue in HIV-1 IN was found to be critical for the interaction with LEDGF/p75 [4]. A virus clone containing this single point mutation was defective for replication due to a block at the integration step [4]. Two studies failed to observe a reduction in HIV-1 replication after partial knockdown of LEDGF/p75 [6,15]. Vandekerckhove et al. [16], however, proved that more potent, transient or stable knockdown of LEDGF/p75 results in a significant inhibition of HIV-1 replication in different cell lines. These data have been independently confirmed [17,18]. The most potent inhibition of HIV-1 replication was observed upon stable overexpression of C-terminal fragments of LEDGF/p75 fused to the enhanced green fluorescent protein (eGFP). These fragments compete with LEDGF/p75 for binding to IN. The absence of chromatin interaction domains renders these fragments dysfunctional for tethering integration [19].
In this study, we show that HIV can develop resistance to overexpressed truncation variants of its own cellular cofactor, although at the expense of a reduced affinity to LEDGF/ p75 and highly reduced replication kinetics. To our knowledge, such resistance phenotype to a cellular cofactor has not been described before. Detection of mutations at key positions in the LEDGF/p75-IN interface provides critical evidence of the importance of LEDGF/p75 for viral replication.

Selection of HIV-1 Strains Resistant to Inhibition by the Overexpressed C-Terminal Fragment of LEDGF/p75
Overexpression of LEDGF/p75 truncation mutants containing the IBD severely inhibits viral replication [19]. Figure  S1A represents a hypothetical scheme of HIV-1 integration in wild-type (WT) MT-4 cells and MT-4 cells overexpressing the LEDGF/p75 truncation mutant eGFP-D325 (MT-4 eGFP-D325). The fusion protein eGFP-D325 consists of the Cterminal fragment of LEDGF/p75 (amino acids 325 to 530) fused to the C terminus of eGFP. In WT MT-4 cells, the viral IN associates with LEDGF/p75 and is tethered to the chromatin, facilitating the integration of the viral cDNA in the host genome. Upon overexpression of eGFP-D325, the eGFP-D325 fusion competes with endogenous LEDGF/p75 for interaction with IN and thereby inhibits viral integration. Note that it is possible that recruitment of LEDGF/p75 versus D325 does not take place in the nucleus but may already occur in the cytoplasm [6]. Figure S1B displays replication of WT virus in MT-4 eGFP-D325 and MT-4 eGFP-D325 D366A cells. The control cells overexpress the D366A mutant of eGFP-D325 known to be defective for interaction with HIV-1 IN [20]. While HIV-1 replication is readily detected in MT-4 eGFP-D325 D366A mutant cell lines, HIV-1 replication is inhibited more than 20-fold 4 d postinfection in the MT-4 eGFP-D325 cells. The virus breaking through at 9 d postinfection is not resistant to the inhibitory effect of eGFP-D325 (unpublished data); limited HIV replication is allowed in cells where the eGFP-D325 fusion protein is competing for LEDGF/p75. Moreover, no mutations were detected in the IN gene of this virus (unpublished data). Therefore, we started to select resistant strains by repeated passaging of HIV-1 (NL4. 3) in MT-4 eGFP-D325 cells. As a control, virus was passaged in parallel in MT-4 eGFP-D325 D366A cells. By reducing the volume of the inoculum (i.e., decreasing the multiplicity of infection [MOI]), selective pressure was progressively increased during the experiment. Figure 1 shows the progressive increase in mutation frequency in the IN coding region while increasing selective pressure (decreasing MOI). Selection up to 21 passages (#21) in MT-4 eGFP-D325 cells resulted in a virus population with 100% of the strains containing both the A128T and the E170G mutation in the IN coding region. No mutations were identified in the IN coding region of the control virus passaged 21 times on the MT-4 eGFP-D325 D366A cells. To rule out cell type dependence, serial passaging of virus was also done in HeLaP4 eGFP-IBD Tet/off and HeLa P4 eGFP-D325 Tet/off cells (unpublished data) [19]. These selection experiments identified the mutations A128T, W131G, K156N, and G163E that mapped to the same regions in the IN gene. Since MT-4 cells better resemble natural host cells of HIV-1 infection, subsequent experiments focused only on the specific IN mutations found by selection in MT-4 eGFP-D325 cells. After 21 passages (#21), two mutations, V38A and S113N, were also found in the gp120 encoding regions of the selected virus population; however, they were present in only 25% and 50% of the virus population, respectively. Moreover, these mutations were also detected in the gp120 gene of the virus population passaged on MT-4 eGFP-D325 D366A cells, although in a lower percentage of the virus population. These mutations likely represent adaptation of the virus to cell culture and infection conditions.

Author Summary
Viruses have a limited genome and therefore exploit the cellular machinery of infected host cells to complete the replication cycle. Today there is a growing interest to unravel these virus-host interactions. Lentiviral integrases use the cellular protein LEDGF/p75 to tether the preintegration complex to the chromosome, although the existence of alternative host proteins substituting for the function of LEDGF/p75 has been proposed. Here, we used virus evolution to investigate the role and importance of LEDGF/p75 in HIV integration and replication. We selected a virus that is resistant against overexpression of the integrase binding domain of LEDGF/ p75. This virus displays a reduced affinity for its own cofactor but still depends on the cofactor for replication. We demonstrate the unique and essential role of LEDGF/p75 in HIV replication and support the LEDGF/p75 HIV-1 interplay as target for innovative antiviral therapy. Moreover, elucidation of the role of LEDGF/p75 in targeting integration will stimulate research on strategies to direct gene therapy vectors into safe landing sites.

Both A128T and E170G IN Mutations Are Necessary and Together Sufficient for Viral Rescue
To evaluate the importance of the described mutations for the observed resistant phenotype, these mutations were cloned in the HIV-1 NL4.3 molecular clone (pNL4.3) as single or double mutants (pNL4.3 A128T, pNL4.3 E170G, and pNL4.3 A128T/E170G, respectively). Susceptibility of the molecular clones to eGFP-D325 overexpression was determined by studying multiple round HIV-1 replications in MT-4 eGFP-D325 ( Figure 3A) and in MT-4 eGFP-D325 D366A cells ( Figure 3B). The experiments were carried out at an MOI of 0.5, and the different viral strains were normalized on p24 content. HIV-1 replication was clearly detected in MT-4 eGFP-D325 D366A cells for all strains used, although all mutated strains showed a reduced replication capacity compared with WT NL4.3. Whereas the double mutant NL4.3 A128T/E170G rescued HIV-1 replication in MT-4 eGFP-D325 cells, the single mutants NL4.3 A128T and NL4.3 E170G could only partially restore viral replication. Apparently, both IN mutations are necessary and together sufficient for the strong rescue phenotype of the selected HIV-1 strain in MT-4 eGFP-D325 cells.

Resistance to eGFP-D325 Overexpression Is Accompanied by Reduced Viral Replication Kinetics and Reduced Enzymatic Activity of IN
The mutations detected in the IN gene have not been described before. Therefore, we studied the impact of the IN To investigate whether the selected IN mutations affect the viral replication kinetics, the mutant molecular clones were examined for their ability to replicate in peripheral blood lymphocytes (PBLs) ( Figure 4B). All mutants showed substantially reduced replication fitness in comparison with WT NL4.3 ( Figure 4B). Replication kinetics of the double mutant NL4.3 A128T/E170G and the single mutant NL4.3 E170G were decreased 4-fold relative to the replication of WT NL4.3 as evaluated by p24 measurements. The NL4.3 A128T strain showed a 2-fold difference in replication capacity.

Virus Resistant to eGFP-D325 Overexpression Displays Reduced 2-LTR Circle Formation and Integration
To determine the step at which the replication cycle of eGFP-D325 resistant virus is affected, real-time quantitative PCR analysis was performed to quantify the formation of the different DNA species in the cell. To ensure single-round infection, 293T cell lines were infected with vesicular stomatitis virus G protein (VSV-G) pseudotyped virus. DNA was extracted at several time points postinfection. As shown in Figure S2A, none of the mutants affected the accumulation of viral DNA at early time points. On the other hand, the number of 2-LTR circles decreased 4-fold for the NL4.3 A128T, NL4.3 E170G, and NL4.3 A128T/E170G virus compared with WT NL4.3 ( Figure S2B). The 2-LTR circles are formed in the nucleus and are an indirect indication for nuclear import of the preintegration complex (PIC). Finally, the integration events in 293T cells were measured by realtime Q-Alu PCR. A 3-to 4-fold reduction in the number of proviruses was revealed for all HIV mutants ( Figure S2C).
To analyze the rescue of viral replication by real-time quantitative PCR, 293T eGFP-D325 cells and control 293T eGFP-D325 D366A cells were infected with WT VSV NL4.3 or VSV NL4.3 A128T/E170G. As shown in Figure S2D, the accumulation of viral DNA at early time points was not affected upon expression of eGFP-D325. As was previously stated for WT virus, the number of 2-LTR circles increased more than 10-fold in the 293T eGFP-D325 cells [19] ( Figure  S2E). However, this increase was significantly lower when the cells were infected with NL4.3 A128T/E170G virus. Integration could not be measured by real-time Q-Alu PCR, which makes use of an LTR primer that anneals to the lentiviral vector sequences used to make these cell lines. Therefore, infected cells were passaged at least four times (six cell divisions) to dilute nonintegrated DNA, ensuring that only integrated copies of viral DNA remained. The absence of 2-LTR circles at 48 h postinfection ( Figure S2E) is an experimental proof for the loss of nonintegrated DNA. Measuring total HIV-1 DNA after four passages pointed to a rescue in the number of proviruses in the 293T eGFP-D325 for the NL4.3 A128T/E170G virus in comparison with WT virus ( Figure S2F).

Confocal Imaging Suggests a Decreased Affinity of mRFP-A128T/E170G-IN s for eGFP-D325 and LEDGF/p75
Apparently, we selected a virus that can overcome a replication block by a fragment of an essential cofactor, by altering the interaction between IN and the fragment. Experiments were designed to directly demonstrate that the IN mutations affect the interaction with (the C-terminal fragment of) LEDGF/p75. First, confocal imaging was used to study the cellular localization of the different mutant INs ( Figure S3A). As expected, due to its interaction with endogenous LEDGF/p75, a transiently expressed monomeric red fluorescent protein-WT-IN s (mRFP-IN s ) fusion protein localized to the nucleus of WT HeLaP4 cells ( Figure S3A) [21][22][23][24][25][26]. Identical results were obtained for mRFP-A128T-IN s . However, mRFP-E170G-IN s and mRFP-A128T/E170G-IN s were more dispersed throughout the cell, although some residual nuclear localization could be observed. This suggests a reduced affinity of the mRFP-A128T/E170G-IN s for LEDGF/ p75. Indeed, upon overexpression of eGFP-LEDGF/p75, a remarkable relocalization of the mutant A128T/E170G IN to the nucleus was observed ( Figure S3C), which suggests that high levels of LEDGF/p75 can overcome the loss in affinity. As  To further characterize these interactions in vivo, we applied fluorescence cross-correlation spectroscopy (FCCS) in HeLa-P4 derived cell lines. This method allows the quantification of the interaction of two fluorescently labeled proteins in cellulo by measuring simultaneous diffusion. Because the diffusion of the complex mRFP-IN s and eGFP-LEDGF/p75 is strongly reduced in the nucleus-most probably due to interactions with chromatin-we could not quantify interactions in the nucleus. Therefore, the K150A NLS mutation was introduced in LEDGF/p75 to overcome the nuclear localization [27], and measurements were done in the cytoplasm of the cell. HeLaP4 eGFP-D325 or HeLaP4 cells overexpressing eGFP-LEDGF/p75(K150A) were transfected with plasmids encoding WT or double mutant mRFP-IN s . As The cross-correlation signal for the interaction of A128T/E170G mRFP-IN s with eGFP-D325 was more affected than that for interaction with full-length LEDGF/p75, suggesting that the mutations in viral IN differentiate between D325 and full-length LEDGF/p75.
A crystal structure of the dimeric catalytic core domain (CCD) of HIV-1 IN complexed to the IBD of LEDGF/p75 was previously reported [28]. A128 is part of the a3 helix in the IN-CCD, which forms a hydrophobic patch that accommodates the side chains of the LEDGF/p75 residues I365, F406, and V408 [28]. E170 is part of the so-called a4/5 connector, a six-residue connector linking helices a4 and a5 of the second IN chain (residues 166 through 171). These two regions were reported to be the main structural features of IN recognized by LEDGF/p75. Biomolecular modeling was performed to predict the influence of the IN mutations on the interaction of IN and LEDGF/p75 ( Figure S4A and S4B). Substituting alanine for threonine at position 128, and performing a structure energy minimization, resulted in a slightly decreased binding affinity of IN for IBD, with an increase in binding energy from À68.02 to À61.03 kcal/mol. Apparently, the bulky side chain of threonine causes sterical hindrance, thereby forcing the I368 residue of the IBD to adapt a different conformation. This occurs without a drastic influence on the conformation of residues in the LEDGF/ p75-IN binding interface. Substituting glutamic acid for glycine at position 170 has a more drastic effect. This mutation results in a different conformation of the loop disrupting two interactions. First, the double hydrogen bond between the hydrogens of the loop backbone at residues E170 and H171 and D366, a critical interacting residue on LEDGF/ p75, is broken. Second, the electrostatic interaction of E170 (IN) with K364 (LEDGF/p75) is abolished. These effects cause a significant decrease in binding affinity, with an increase in predicted binding energy from À68.02 to À56.38 kcal/mol. When both IN mutations are present, the overall binding energy for the IBD-IN complex is increased from À68.02 to À53.10 kcal/mol.

Virus Resistant to eGFP-D325 Overexpression Is Dependent on LEDGF/p75 for Replication
How does HIV overcome the replication block induced by the C-terminal fragment of its cofactor? FCCS measurements suggest that by altering the affinity of IN for the cofactor, HIV-1 may be capable to discriminate between endogenous LEDGF/p75 and the inhibiting C-terminal fragment (eGFP-D325). Alternatively, HIV-1 that has become deficient for interaction with LEDGF/p75 may replicate independently from LEDGF/p75, by employing another cofactor. The latter explanation has been put forward recently after analysis of integration sites in LEDGF/p75 knockdown cells [14]. LEDGF/ p75 knockdown only resulted in a modest shift in integration site selection. To address this question, replication of the virus resistant to eGFP-D325 overexpression was followed in LEDGF/p75 knockdown cells. Stable polyclonal MT-4 LEDGF/ p75 knockdown cells were generated using lentiviral vectors encoding shRNA targeting LEDGF/p75 (MT-4 p75À). As a control, four mutations were introduced in the short hairpin against LEDGF/p75, and MT-4 cells were generated (MT-4 mut). Efficiency of knockdown was verified by Western blotting ( Figure 7A). Infection of these cell lines with WT NL4.3 at an MOI of 0.5 resulted in a 5-to 10-fold reduction in viral replication as measured by p24 antigen production in the supernatant ( Figure 7B). In the control cells encoding the mutant hairpin (MT-4 mut), no reduction of HIV-1 replication was observed. No breakthrough of the mutant NL4.3 A128T/E170G virus was detected up to 10 d postinfection in MT-4 p75À cells (MOI 0.5) ( Figure 7B). Moreover, viral replication of the NL4.3 A128T/E170G virus was inhibited 10-fold more than that of WT virus in MT-4 p75À cells. In the control MT-4 mut cells, both strains showed equivalent viral replication. The inhibitory effect on the WT virus in the MT-4 p75À cells could be overcome by using a higher MOI (MOI 2.5) ( Figure 7C). However, at the same MOI, NL4.3 A128T/ E170G replication remained 10-fold lower. The same experiment was performed in the LEDGF/p75 knockdown Jurkat cells p75(À) Cl 2 (Jurkat p75À) and p75(À) Cl 2 backcomplemented (BC) (Jurkat BC) cells that were originally reported by Llano et al. [6]. Llano and coworkers did not detect any reduction in viral replication in these knockdown cells compared with the BC control cells. In agreement, we could not detect any inhibition of replication of WT NL4.3 in these cells ( Figure 7D). However, replication levels of A128T/ E170G virus were 10-fold lower in the Jurkat p75À cells, whereas no significant reduction of HIV-1 replication was observed in the BC Jurkat cells for any of the strains used. We reason that replication of the mutant virus should not be inhibited in cells with highly reduced levels of LEDGF/p75, if this replication occurs independently of LEDGF/p75. To the contrary, our experiments suggest that the virus resistant to eGFP-D325 overexpression has a reduced affinity for LEDGF/ p75, making it more vulnerable to reduction of LEDGF/p75 levels. It is clear that LEDGF/p75 is still employed by the virus for viral replication in eGFP-D325 overexpressing cells. Our data are thus at odds with the existence of another cofactor that can substitute for LEDGF/p75.
In line with these results and to further strengthen our model that the A128T/E170 virus is still capable of interacting with LEDGF/p75 in vivo, we speculated that poor replication of the double mutant virus in WT cells could be rescued by overexpression of LEDGF/p75. For this purpose, MT-4 cells stably overexpressing LEDGF/p75 and MT-4 cells stably overexpressing eGFP were generated. Overexpression was verified by Western blot analysis ( Figure S5A). These cell lines were used to compare replication of WT NL4.3 virus and A128T/E170G NL4.3 virus ( Figure S5B). In the control MT-4 eGFP cells, replication kinetics of the double mutant NL4.3 A128T/E170G was decreased 2-fold in comparison to the WT HIV-1 NL4.3. However, in MT-4 cells overexpressing LEDGF/ p75, replication of both virus strains was similar. This experiment again suggests that poor replication of the IN mutants in WT cells is due to reduced affinity for LEDGF/p75 and that the A128T/E170 IN mutant virus still exploits LEDGF/p75 in vivo.

Discussion
We have previously characterized LEDGF/p75 mutants containing the IBD but lacking the chromosome attachment site (eGFP-D325 and eGFP-IBD) that strongly impair HIV-1 replication by competing for the interaction with IN [19]. We have now selected HIV-1 strains resistant to such a truncation mutant of LEDGF/p75 (eGFP-D325) (Figure 2). Two mutations were detected in the IN coding region at key positions in the LEDGF/p75-IN interface (A128T and E170G) (Figure 1). This corroborates that the specificity of the observed inhibition in viral replication in MT-4 eGFP-D325 cells is unambiguously based on the direct interaction of the overexpressed D325 fragments with IN. The available crystallographic data on the IN-IBD interface [28] were recently questioned after an in vitro mutagenesis study that failed to observe a strict correlation between the interaction of IN with LEDGF/p75 and the effect on HIV-1 replication [29]. Detection of the IN mutations A128T and E170G at key positions after resistance selection provides in vivo confirmation for the previously reported structure. As biomolecular modelling suggests, mutating these residues compromises the local structure of the IBDÀIN CCD interface, affecting the interaction with LEDGF/p75 truncation mutants ( Figure S4). Our data provide a striking example of the power of viral molecular evolution. Not only is the approach apparently more relevant than in vitro mutagenesis, a virus appears to discriminate between a fragment of a critical cofactor and the full-length cofactor itself.
To overcome inhibition by D325, mutations in the IN-IBD interface are selected. These mutations cripple the virus considerably ( Figure 4). It has been demonstrated before that mutations associated with antiviral resistance to diketo acids reduce the enzymatic activity of IN [30,31]. The conservation of the IN gene compared with other viral genes [32] suggests a tight restriction in amino acid flexibility. Still, our results show that viral replication is more affected than enzymatic activity. The most plausible explanation for the observed phenotype is that by altering the affinity for the inhibiting Cterminal fragment, interaction with the critical cofactor LEDGF/p75 is also jeopardized. Indeed, poor replication of the double mutant virus in WT cells could be rescued by overexpression of LEDGF/p75 ( Figure S5B). Although HIV strains have been selected to grow independently of restriction factors [33,34], this is to our knowledge the first example of selection of a virus that is partially defective for binding to its cellular cofactor. Although humans carrying a 32-base pair deletion of the CCR5 gene are resistant to HIV-1 infection, no R5 virus has yet been isolated that overcomes this inhibitory effect; infection is only possible via T-tropic virus that uses the CXCR4 coreceptor [35].
The reduced replication kinetics of the mutated virus were further analyzed by quantitative PCR analysis of 293T cells infected with VSV-G pseudotyped A128T/E170G virus. We detected a 4-fold decrease in the number of 2-LTR circles and in integrated provirus ( Figure S2). In previous experiments in which integration was impaired due to partial depletion of LEDGF/p75 [16,17] or to the presence of competing Cterminal fragments [19], the amount of 2-LTR circles was either constant or increased. The reduction in 2-LTR circles might indicate that the mutations in IN affect nuclear import of the PIC or the formation of 2-LTR circles. We cannot exclude that reduced interaction with LEDGF/p75 is responsible for this phenotype. Our previous finding that LEDGF/ p75 drastically increases the affinity of lentiviral IN for DNA may imply a role of LEDGF/p75 in the formation of the complex between IN and the viral DNA. This is the first experimental evidence suggesting that IN-LEDGF/p75 interaction might affect steps other than chromosomal tethering/ targeting. The documented presence of LEDGF/p75 in the cytoplasmatic PIC is consistent with this concept [6].
Analysis of DNA formation by the mutant virus in eGFP-D325 overexpressing 293T cells revealed that integration of the resistant virus in the 293T eGFP-D325 cells was restored ( Figure S2F). Whereas the number of 2-LTR circles produced by WT virus increased more than 10-fold in the 293T eGFP-D325 cells ( Figure S2E) as a result of the defect in integration ( Figure S2F) [19], the number of 2-LTR circles returned to almost normal levels when the cells were infected with A128T/ E170G virus. These data confirm that the C-terminal fragments specifically block the integration and/or targeting step.
What is the mechanism underlying the resistant phenotype observed? Our in vitro data ( Figure 5) and in vivo confocal microscopy analysis ( Figure S3) both reveal experimentally that the mutations that reduce interaction with eGFP-D325 also decrease interaction with full-length LEDGF/p75. The available structural data support this result. The fact that A128T/E170G IN displays a reduced affinity for full-length LEDGF/p75 seems to suggest at first sight that HIV can replicate without LEDGF/p75 perhaps using an alternative cofactor. In fact, because of the subtle alterations in target site selection after LEDGF/p75 depletion, the existence of such a secondary targeting cofactor has been proposed [14]. However, replication of the resistant virus in LEDGF/p75 knockdown MT-4 cells was inhibited 10-fold more than that of WT virus ( Figure 7B). Moreover, replication of A128T/ E170G virus was also severely impaired in the Jurkat p75À cells, which support normal replication of WT virus [6] ( Figure 7D). It follows that LEDGF/p75 is still employed by the resistant virus that is capable to grow in eGFP-D325 overexpression cells. In fact, poor replication of the double mutant virus in WT cells is rescued by overexpression of LEDGF/p75 ( Figure S5B), corroborating that the A128T/E170 IN mutant virus still exploits LEDGF/p75 in vivo. Although we cannot formally exclude the remote possibility that both mutations also affect binding of the alternative cofactor, its existence has become unlikely. Reduction of the affinity of IN for LEDGF/p75 results in a clear phenotype even in cells with partial depletion of LEDGF/p75. Hence, the failure by Llano et al. [6] to detect a replication deficit in Jurkat p75À is due to an insufficient knockdown, as proposed [1,16]. Apparently, reduced interaction with LEDGF/p75 is preferred rather than ineffective targeting to the chromatin in the absence of LEDGF/p75.
Replication of A128T/E170G virus in MT-4 cells overexpressing LEDGF/p75 truncation mutants suggests that its affinity for eGFP-D325 is more reduced than for LEDGF/p75, taking into account that eGFP-D325 is in excess of LEDGF/ p75 (unpublished data). Differential impact of the A128T/ E170G mutation on interaction of recombinant IN with fulllength LEDGF/p75 or the C-terminal fragment could not be documented in vitro by pull-down ( Figure 5A) or Al-phaScreen technology (unpublished data). Probably, these assays do not reflect the in vivo situation. In vivo FCCS experiments, however, revealed that mutant IN has a different affinity for the cofactor and the inhibiting C-terminal fragment ( Figure 6). Domains in LEDGF/p75 other than the C-terminus may contribute to the interaction with IN. Alternatively, the fusion of D325 to eGFP may also aid A128T/E170G IN to preferentially bind to LEDGF/p75.
The importance of LEDGF/p75 for HIV replication has been demonstrated over time using mutagenesis [4], RNA interference [16,17], or overexpression of truncation mutants [19]. However, none of these experiments could exclude the existence of other cofactors capable of tethering the PIC to the chromosomes. Our data indicate that HIV is incapable to circumvent the LEDGF/p75 function during integration. Our approach may prove valuable in validating other cellular cofactors of virus replication. Demonstration of the biological relevance of the available LEDGF/p75-IN interface, the impaired replication kinetics upon resistance selection to an interaction inhibitor, and the exclusive role in HIV integration all support the attraction of LEDGF/p75 as an antiviral target.

Materials and Methods
Plasmids and viral vectors. The construction of the lentiviral vector transfer plasmids encoding shRNA targeting LEDGF/p75 was described previously (unpublished data). The construct pmRFP-IN s was used for overexpression of mRFP-IN s in HeLaP4 cells [19], and the construct peGFP-p75 was used to overexpress eGFP-tagged human LEDGF/p75 in HeLaP4 cells [12]. For bacterial expression of nontagged LEDGF/p75, C-terminal His 6 -tagged HIV-1 IN and Nterminal His 6 -tagged HIV-1 IN, the plasmids pCPnat75 [12], pKB-IN6H [12], and pRP1012 [36] were used, respectively. To generate the A128T or E170G single mutants and double mutants, site-directed mutagenesis was performed using the Kirsch and Joly method [37]. The presence of the expected mutations was confirmed by DNA sequencing of the entire IN coding region to ensure no other mutations were present in the molecular clone. To construct the lentiviral transfer plasmids pCombi_LEDGF_IRES_Puro, eGFP was removed from the plasmid pCombi_eGFP_IRES_Puro [19] by BamHI-SpeI digestion and replaced by BamHI-LEDGF-XbaI. This fragment was obtained by PCR amplification using the primers LEDGF-forw (59-GGC GGG ATC CAG ACA CCA TGA CTC GCG ATT TCA AAC C-39) and LEDGF-rev (59-CAG GTC TAG ACT AGT TAT CTA GTG TAG AAT CCT TC-39) using pEF1-LEDGF-Back [6] as a template. Subsequently, the resulting products were digested with BamHI and SpeI. Production of lentiviral vectors was performed as described previously by Geraerts et al. [38].
Virus strains. The HIV-1 molecular clone pNL4.3 [40] was obtained through the AIDS Research and Reference Reagent Program, Division of AIDS, National Institute of Allergy and Infectious Diseases, National Institutes of Health, contributed by Dr. Malcolm Martin (Bethesda, Maryland, United States). To generate the A128T, E170G, and A128T/E170G IN mutated viruses, site-directed mutagenesis was performed using the Kirsch and Joly method [37]. The presence of the expected mutations was confirmed by DNA sequencing of the entire IN coding region. Virus productions were performed as described by De Rijck et al. [19].
Selection of resistance. The resistance selection of HIV-1 (NL4.3) against overexpression of the C-terminal part of LEDGF/p75 was initiated at a high MOI (25) in MT-4 eGFP-D325 cells. Simultaneously, as a control, virus was also passaged in MT-4 eGFP-D325 D366A cells. Every 3 to 4 d, the MT-4 cell culture was monitored for the appearance of HIV-induced cytopathic effect (CPE). When CPE was observed, the cell-free culture supernatant was used to infect fresh, uninfected MT-4 eGFP-D325 or MT-4 eGFP-D325 D366A cells. To increase selective pressure, the volume of supernatant transferred was progressively decreased during the experiment from 1 ml to 50 ll for 2 3 10 5 cells cultured in 2 ml of medium.
PCR amplification and sequencing of the coding regions of IN. Proviral DNA extraction of infected MT-4 eGFP-D325 cells was performed using the QIAamp blood kit (Qiagen, http://www.qiagen. com). PCR amplification and sequencing of IN encoding sequences were done as described previously [30]. Mutations present at least 25% of the global virus population can be detected as a mixture with the WT amino acid by means of population sequencing.
HIV-1 infection. Infection of MT-4 and Jurkat cells was performed with 1 3 10 6 cells in 1 ml of medium at an MOI of 0.1, 0.5, or 2.5 for 3 h. Different virus strains were normalized on p24 levels. After 3 h, cells were washed twice with PBS and resuspended in 10 ml of MT-4 medium. HIV-1 replication was monitored by quantifying p24 antigen in the supernatant via ELISA (Alliance HIV-1 p24 ELISA kit; Perkin Elmer, http://www.perkinelmer.com).
To study replication kinetics in MT-4 cells, inoculants of various HIV-1 strains containing equal amounts of HIV-1 p24 antigen (10, 5, and 2.5 pg/ml) were added to MT-4 cells (50,000 cells/ml). From 3 d postinfection on, aliquots of cell-free supernatants were taken for p24 level determination.
Real-time quantitative PCR analysis. One day prior to transduction, 293T cells were seeded in 24-well plates at approximately 2 3 10 5 cells per well. Transductions with the VSV-G pseudotyped viruses were carried out at an MOI of 10. Prior to infection, VSV-G virus stocks were treated with RNase-free DNase I (0.2 U/ll) (Roche Applied Sciences). Virus was added to the cells in the presence of DMEM-1% FCS. After 3 h of incubation, medium was replaced by DMEM containing 10% FCS. In each 24-well plate, noninfected 293T cells were incubated in parallel. DNA extractions and quantification of late reverse transcripts, 2-LTR circles, and integrants were done as described earlier [42,43]. Analysis of DNA extracted from lentivirus tranduced cells was done as described earlier [19].
FCCS analysis. FCCS measurements were performed on a commercial LSM510/ConfoCor2 system (Carl Zeiss, Jena, Germany). The 488-nm line of the Ar þ -laser (AOTF 0.1%, approximately 25 lW) was used to excite eGFP, and the 543-nm line of the HeNe laser (AOTF 7%, approximately 70 lW) was used to excite mRFP1. The excitation light was reflected by a dichroic mirror (HFT 488/543) and focused through a type C-Apochromat 340/1.2W objective lens. The fluorescence emission light was split by a second dichroic mirror (NFT 570) into two separate beam paths and passed through a 505-to 530-nm bandpass filter for eGFP fluorescence and a 600-to 650-nm bandpass filter for mRFP1 fluorescence. Each confocal pinhole diameter was set to 70 lm. After preparation of the cells in an eight-well chambered coverglass (NUNC A/S, Roskilde, Denmark), laser scanning microscopy was used to search for cells suitable for FCCS. The laser beam was focused at a selected spot, and an FCCS measurement (10 3 20 s) was performed. Auto-and cross-correlation curves were evaluated by a Levenberg-Marquardt nonlinear leastsquares fitting to a two-component model, using the OriginPro 7.5 software (OriginLab, Northampton, Massachusetts, United States). The relative cross-correlation is calculated from the following formula: CC rel. ¼ G cross (0)/G green (0), with G cross (0) and G green (0) being the amplitudes of the cross and green autocorrelation curves, respectively [19].
Western blotting. Western blotting was performed as described previously [16].
Substrate and target DNA used in the enzymatic IN assay. The HPLC-purified deoxyoligonucleotides, INT1 (59-TGT GGA AAA TCT CTA GCA GT) and INT2 (59-ACT GCT AGA GAT TTT CCA CA), corresponding to the U5 end of the HIV-1 LTR, were purchased from Amersham Biosciences (http://www.amersham.com). The oligonucleotide INT1 was purified through a 20% denaturing polyacrylamide/ urea gel and was radioactively labeled at the 59-end using polynucleotide T4 kinase and [c-32 P]ATP (Amersham Biosciences). The DNA substrate and target for IN reactions were made by annealing INT1 and INT2. An equimolar mixture of the two oligonucleotides in the presence of 100 mM NaCl was heated shortly at 95 8C and allowed to cool slowly to room temperature.
The overall integration assay. The enzymatic integration reactions were carried out as described previously [44][45][46] with minor modifications [30]. Specific activities of different enzyme preparations were determined in this assay. The extent of overall integration was based on measuring the amounts of strand transfer products of the mutant INs in comparison with WT IN. These data were determined using the OptiQuant Acquisition and Analysis software (Perkin Elmer Corporate).
His 6 Tag IN Pull-down Assay. His 6 tag IN pull-down assay was adapted from Maertens et al. [12], with minor modifications. Binding of IN to LEDGF/p75, IBD-MBP, and D325-MBP was assayed in 25 mM Tris-HCl (pH 7.4), 0.1% Nonidet P-40, and 20 mM imidazole containing 150 mM NaCl, in the presence of 1 mM MgCl 2 (binding buffer). Then, 6 lg of recombinant C-terminal His 6 -tagged IN was incubated with 40 ll of Ni-NTA-agarose for 10 min at 4 8C. Next, 6 lg of LEDGF/p75 in 200 ll of binding buffer supplemented with 4 lg of BSA was added to the mixture. Following 4-h incubation at 4 8C, the agarose beads were recovered by centrifugation for 2 min at 13,000 rpm at 4 8C and washed several times with 300 ll of binding buffer. Bound proteins were eluted in 20 ll of binding buffer supplemented with 200 mM imidazole and 1% SDS and analyzed by 11% SDS-PAGE followed by staining with Coomassie Blue R-250 [12].
AlphaScreen. To confirm the loss of interaction of the mutant INs to the D325 fragment, AlphaScreen technology (Perkin Elmer) was performed. In a 384-well plate, 0, 30, 100, or 300 nM of IN was added to 0, 10, 30, or 100 nM MBP-D325 at room temperature in assay buffer (25 mM Tris/HCl [pH 7.3], 150 mM NaCl, 1 mM MgCl 2 , 0.01% Tween-20, 0.1% BSA). The mixture was incubated for 90 min at room temperature. For subsequent binding of the proteins to the beads, Nichelate-coated acceptor beads and streptavidin-coated donor beads premixed with biotinylated anti-MBP antibody (Vector Laboratories, http://www.vectorlabs.com) were added to a final concentration of 20 lg/ml. All dilutions were done in assay buffer and, if necessary, the final volume of the reaction was adjusted to 25 ll. In order to bind the anti-MBP antibody to the streptavidin-coated donor beads, the antibody was dialyzed overnight against assay buffer without BSA, added to the beads in a final concentration of 50 nM, and incubated for 1 h at room temperature. After addition of the beads to the binding assay, light exposure was omitted and the incubation was carried out for an additional 60 min before analyzing the interaction in the EnVision plate reader (Perkin Elmer).
Biomolecular modeling. In order to investigate the structural influences caused by the mutations on the LEDGF/p75 HIV-1 IN interface, a biomolecular modeling approach was used. Structural data of the binding of these two proteins are limited to the crystal structure 2B4J of the HIV-CCD dimer with two IBD fragments [28]. Unresolved loops in the crystal structure were first completed using the B chain of the CCD 1BIS structure [47]. Four different complexes were generated: the WT, the single mutated, and the double mutant structure. Both CCD IN chains of the dimer were mutated. After mutation of the residues and refinement of the mutated residue, a sphere with a radius of 9 Å was subsequently minimized using a CHARMM-based force field [48]. Crystallographic water molecules were incorporated, and an additional water box was used during the energy minimization. In order to analyze the influence of the mutations, the nonbonded interaction energy of the interface and the contribution of the electrostatic energy, Van Der Waals energy, and hydrogen bonding energy were computed for the LEDGF/p75 IN interface and the residues of the interface. Crystallographic water molecules bound in the interface were taken in account for these computations. The computations were performed using MOE (The Molecular Operating Environment, Chemical Computing Group, 1255 University Street, Suite 1600; Montreal, Quebec, Canada H3B 3X3) (modeling) and Brugel [49] (analysis).    (A) Binding of the two WT catalytic core domains with the IBD (PDB code 2B4J). A128 is part of the a3 helix in the IN-CCD, which forms a hydrophobic patch that accommodates the side chains of the LEDGF residues I365, F406, and V408 [28]. E170 is part of the so-called a4/5 connector, a six-residue connector linking helices a4 and a5 of the second IN chain (residues 166 through 171). (B) Representation of the model showing the effect of mutating A128 to T128 and E170 to G170. For A128T, the bulky side chain of threonine causes sterical hindrance, thereby forcing the I368 residue of the IBD to adapt a different conformation. This occurs without a drastic influence on the conformation of residues in the LEDGF/p75 IN binding interface. The E170G mutation has a more drastic effect. This mutation results in a different conformation of the loop disrupting two interactions. First, the double hydrogen bond between the hydrogens of the loop backbone at residues E170 and H171 and D366, a critical interacting residue on LEDGF/p75, is broken. Second, the electrostatic interaction of E170 (IN) with K364 (LEDGF/p75) is abolished. Found at doi:10.1371/journal.ppat.0030047.sg004 (6.5 MB TIF).