Identification of Vpr mutants deficient for hCG1-binding
Previous studies have established that the localization of HIV-1 Vpr to the NE is related to its ability to interact with components of the NPC [23,25,26], including the nucleoporin hCG1 . In order to identify single-point mutations that altered the Vpr binding to hCG1, we generated a library of random Vpr mutants and used the yeast two-hybrid system to screen for hCG1-binding deficient Vpr mutants. Only mutants which retained the capacity to interact with UNG2 and HHR23A, two other known relevant host partners of Vpr [30,31] but failed to bind hCG1 were selected. Two Vpr mutants (clones 11 and 35) that still interacted with UNG2 and HHR23A were isolated (Fig. 1A and data not shown, respectively), as evidenced by growth of yeast-transformed cells on medium without histidine (-His) and β-gal activity. In contrast, these mutants did not bind to hCG1, since no growth on -His medium and β-gal activity was observed. Used as controls, the VprR90K mutant, which is known to abolish Vpr-induced G2-arrest , still bound both to hCG1 and UNG2, while the W54R mutant, which is deficient for binding to UNG2 , still interacted with hCG1 (Fig. 1A, lower panel). These results show that this yeast two-hybrid strategy is a powerful system to isolate specific hCG1-binding deficient Vpr mutants.
Figure 1 Identification of Vpr mutants deficient for binding to the nucleoporin hCG1. A) Screening for Vpr mutants defective for the interaction with hCG1. The L40 yeast reporter strain expressing the wt or mutated (clones 11 and 35, and Vpr-R90K and -W54R single-point mutants) HIV-1 Vpr fused either to LexABD (upper panels) or to the Gal4 DNA binding domain (Gal4BD) (lower panels), in combination with each of the Gal4AD-hybrids indicated on the top was analyzed for histidine auxotrophy and β-Gal activity. Double transformants were patched on selective medium with histidine (+His) and then replica-plated on medium without histidine (-His) and on Whatman filters for β-Gal assay. Growth in the absence of histidine and expression of β-galactosidase indicated an interaction between hybrid proteins. B) Amino acid substitutions found in the hCG1-binding deficient Vpr mutants (clones 11 and 35). Mutants were derived by error prone PCR-mediated mutagenesis from the primary sequence of the VprLai strain that is shown at the top. C) Isolation of single-point Vpr mutants defective for the interaction with hCG1. Single-point mutants derived from Vpr clones 11 and 35 fused to LexABD were expressed in L40 strain in combination with each of the Gal4AD-hybrids indicated on the top. Double transformants were assessed as described in A).
DNA Sequencing of clone 11 revealed 3 substitutions within the VprLai primary sequence (Leu23Phe, Leu67Gln and Arg73Gly), while clone 35 contained a single substitution (Lys27Met) (Fig. 1B). Each substitution from clone 11 was introduced in the Vpr sequence and the 3 single-point mutants were analyzed again for binding to hCG1 and UNG2. As shown in Fig. 1C, the L23F and K27M substitutions were sufficient to abrogate hCG1 binding without significant alteration of binding to UNG2. In contrast, the L67Q and R73G Vpr mutants still interacted with both hCG1 and UNG2. These results reveal that the L23F and K27M Vpr variants are specifically altered for the binding to hCG1.
As deduced from the 3D structure organization of Vpr resolved by NMR (see on Fig. 2A), the Leu23 and Lys27 residues are located in the first N-terminal α-helix H1 (residues 17–33) of Vpr which has amphipathic properties. Leu23 and Lys27 are separated by 3 residues and are thus located on the same face of the first α-helix (Fig. 2D). The connection between these two residues is favored by the formation of a hydrogen-bonding network through the O19/NH23, O23/NH27 and O27/NH31 atoms maintaining the structure of the α-helix. Moreover, the Corey, Pauling, and Koltun (CPK) representation, indicates that the Leu23 and Lys27 residues are located at the bottom of a pocket that is easily accessible to the solvent (Fig. 2B) and could constitute a binding site for hCG1. In addition, the Leu23 residue is hydrophobic and is surrounded by rather hydrophobic residues (Leu20, Trp54, Gly51 and Tyr47) that border one edge of the pocket (Fig. 2C), whereas the Lys27 residue is hydrophilic, positively charged and bordered by hydrophilic residues (Gln44, His40, Asn28 and Glu24) that constitute the second edge of the pocket. The potential structural modifications induced by substitution of Leu23 and Lys27 in Phe and Met, respectively, have been calculated by homology with the wild type Vpr protein using the Swiss-Model program [33-35]. The analysis indicated that the structure of the first α-helix (residues 17–33) is conserved as well as the hydrogen-bonding network allowing the stabilization of the 3 helices of HIV-1 Vpr. This supports the notion that the global 3D structure of the protein is not modified in these two Vpr mutants, as suggested from the yeast two-hybrid analysis.
Figure 2 Impact of the Vpr-L23F and -K27M substitutions on the three-dimensional structure of Vpr. A) 3D structure of HIV-1 Vpr , showing the three α-helices (residues 17–33, 38–50 and 54–77) represented in light blue, yellow and purple, respectively. The L23, K27, A30 and F34 residues are colored in red. The unstructured N- and C-terminal domains are represented in dark blue. B) CPK representation of Vpr. Residues are colored according to their hydrophobicity, except for L23 and K27 which are colored in yellow. The yellow box is enlarged in C), and this region shows a pocket that is organized around the L23 and K27 residues within the first α-helix and may represent a site for hCG1 binding. D) Helical-wheel diagram of the first α-helix of Vpr extending from a.a. D17 to F34. Residues L23, K27, A30 and F34 which have been mutated in the present study are indicated. Hydrophilic residues are in blue, whereas hydrophobic residues are in red.
Intracellular distribution of the Vpr mutants
Since HIV-1 Vpr localizes predominantly in the nucleus but also concentrates at the NE as a nuclear rim staining (Fig. 3A, middle panel) where it co-localizes with the nucleoporin hCG1 (left panel) , the cellular distribution of the two hCG1-binding deficient Vpr mutants was first analyzed. In contrast to the wt Vpr-GFP fusion, both Vpr-L23F and -K27M equally distributed between the cytoplasm and the nucleus (Fig. 3B), but they were excluded from the nucleolus. When expressed as HA-tagged proteins, these Vpr mutants similarly co-distributed in the cytoplasm and the nucleus, whereas wt HA-Vpr was concentrated into the nucleus and at the NE (data not shown). These data support that mutations of Vpr which alter its binding to hCG1 also impair its accumulation at the NE.
Figure 3 Subcellular distribution of the Vpr mutants. A) Colocalization of Vpr and hCG1 at the NE. HeLa cells co-expressing Vpr-GFP (middle row) and Myc-hCG1 (left row) fusion proteins were permeabilized with digitonin, fixed, and subsequently stained with an anti-Myc monoclonal antibody. B and C) Localization of wt and mutated Vpr-GFP fusions. HeLa cells expressing either GFP, wt Vpr-GFP, or the indicated Vpr-GFP mutants were fixed and directly examined. Cells were analyzed by epifluorescence microscopy, and images were acquired using a CCD camera. Scale bar, 10 μm.
In order to explore whether substitutions in the first α-helix had a general impact on the localization of Vpr, the cellular distribution of two other Vpr mutants (Vpr-A30L and -F34I) was also analyzed (Fig. 3C). In contrast with published observations , we found that Vpr-A30L was distributed between the nucleus and the cytoplasm and failed to concentrated at the NE. As previously reported , Vpr-F34I displayed a nucleocytoplasmic distribution. In contrast, other Vpr mutants with substitutions in the third α-helix or in the C-terminal flexible basic region of the protein, such as Vpr-W54R, -R80A and -R90K, were concentrated at the NE as efficiently as the wt Vpr-GFP fusion (Fig. 3C). Altogether, these results indicate that the first α-helix of Vpr contains the major determinants required for the nuclear localization of the protein.
G2-arrest activity and cell death induction of the Vpr mutants
Since a functional link was reported between the targeting at the NE and the Vpr-induced cell cycle arrest [36,37], the G2-arrest activity of the Vpr-L23F and Vpr-K27M mutants was first assessed in T lymphocytes. HPB-ALL T lymphoid cells were transfected with wt or mutated HA-tagged Vpr expression vector together with a GFP expression vector (see Fig. 4C), and the DNA content was analyzed 48 h later by flow cytometry on GFP-positive cells after staining with propidium iodide. The results of four independent experiments are recapitulated on Fig. 4A. The Vpr-L23F mutant was affected but retained about 50% of the activity measured for the wt protein, while the Vpr-K27M mutant was more severely affected leading to a residual G2-arrest activity. Consistent with previous observations, the Vpr-F34I mutant was partially altered for the G2-arrest activity , while the Vpr-A30L mutant was completely defective [20,36] (Fig. 4A). As controls, the Vpr-R80A and -R90K variants, which still accumulated at the NE (Fig. 3C), were unable to induce a G2-arrest (Fig. 4A and Refs. [31,37]). The pro-apoptotic activity of the wt Vpr protein and the mutants was also assayed, 72 h after transfection, by flow cytometry analysis of the cell surface exposure of phosphatidylserine (PS) after staining with phycoerythrin-labeled Annexin V (Fig. 4B). Interestingly, the Vpr-induced pro-apoptotic activity of all the Vpr mutants, including Vpr-L23F and -K27M, strictly paralleled the results obtained in the cell cycle experiments (compare Fig. 4A and 4B), suggesting that induction of G2-arrest and apoptosis by HIV-1 Vpr are functionally related. As evidenced on Fig. 4C, the reduction in G2-arrest and cell death induction observed with the Vpr mutants could not be explained by important differences in their expression levels, since all mutants were correctly expressed in HPB-ALL T lymphoid cells.
Figure 4 G2-arrest and pro-apoptotic activities of the Vpr mutants. HPB-ALL T cells were transfected with the HA-tagged Vpr (wt or mutated) expression vectors in combination with the GFP expression vector. A) G2-arrest activity. The DNA content was analyzed 48 h after transfection by flow cytometry on GFP-positive cells after staining with propidium iodide. Results are expressed as the percentage of the G2M/G1 ratio relative to that of the wt HA-Vpr. Values are the means of four independent experiments. Error bars represent 1 standard deviation from the mean. B) Pro-apoptotic activity. Cell surface PS exposure was analyzed 72 h after transfection by flow cytometry on GFP-positive cells after staining with phycoerythrin-labelled Annexin V. Results are expressed as the percentage of GFP-positive cells displaying surface PS exposure relative to that measured with wt HA-Vpr. Values are the means of four independent experiments. Error bars represent 1 standard deviation from the mean. C) Expression of wt and mutated HA-tagged Vpr proteins. Lysates from HPB-ALL transfected cells were analyzed by western-blotting using anti-GFP (upper panels) and anti-HA antibodies (lower panels).
Altogether, these observations indicate that accumulation of Vpr at the NE is required but is not sufficient for its action on the cell cycle progression and the subsequent cell death. They also confirm that these two Vpr functions are functionally related.
Intracellular localization of Vpr mutants in primary human monocyte-derived macrophages
In order to confirm that Vpr also accumulated at the nuclear envelope in target cells relevant for HIV-1 replication, the distribution of both wt and mutated Vpr proteins was then analyzed in primary macrophages derived from monocytes (MDMs) isolated from buffy coats of healthy donors. As previously shown in HeLa cells (see Fig. 3), the wt Vpr-GFP fusion localized in the nucleus of MDMs but also concentrated at the NE as a punctuate staining likely corresponding to NPC structures (Fig. 5). A similar punctuate staining at the NE was observed in a myeloid cell line, such as THP-1 cells, expressing the Vpr-GFP fusion (not shown). Again, both Vpr-L23F and -K27M mutants failed to concentrate at the NE and predominantly localized in the cytoplasm as a diffuse staining. These data confirm that Vpr mutants deficient for hCG1-binding also fail to accumulate at the NE in primary macrophages.
Figure 5 Subcellular localization of wild type Vpr and Vpr mutants in human monocyte-derived-macrophages. MDMs expressing either GFP, wt Vpr-GFP, or the indicated Vpr-GFP mutants were fixed and analyzed by wide-field microscopy. Z stacks of fluorescent images were acquired using a piezo with a 0.2 μm increment and one medial section is shown (left panels). Phase contrast images of the same cells were acquired to identify the nucleus (right panels). Scale bar, 5 μm.
Replication in primary macrophages of the hCG1-binding deficient Vpr HIV-1 mutants
Finally, the relationship between the Vpr docking at the NE and HIV-1 replication in non-dividing cells was explored by analyzing the impact of the hCG1-binding deficient Vpr-L23F and -K27M mutations on viral replication in primary macrophages. The requirement of Vpr for early stages of the virus life cycle, including nuclear transport of the viral DNA (for review, see Ref. ), has been associated with its packaging into virions and the resultant presence in the cytoplasm of newly infected cells. Using a transient Vpr packaging assay in which HA-tagged Vpr is expressed in trans in virus producing cells , we therefore analyzed whether the two Vpr mutants were incorporated into virions. As evidenced in Fig. 6A, both Vpr-L23F and -K27M were efficiently packaged into purified virions, but a slight difference in the level of incorporation was repeatedly observed.
Figure 6 Impact of the Vpr mutations on HIV-1 replication in monocyte-derived macrophages. A) Packaging assay of the wt and mutated HA-tagged HIV-1 vpr into virus like particles. 293T cells were transfected with an HIV-1-based packaging vector lacking the vpr gene in combination with vectors for expression of the wt or mutated HA-tagged Vpr protein. 48 h later, proteins from cell and virion lysates were separated by SDS-PAGE and analyzed by Western blotting with anti-HA and anti-CAp24 antibodies. B and C) The L23F or K27M mutations were introduced into the vpr gene of the HIV-1YU-2 molecular clone. In B) Lysates from transfected 293T cells and virions isolated from cell supernatants were subjected to SDS-PAGE followed by Western blotting, using a rabbit polyclonal anti-Vpr and a mouse anti-CAp24 (provided from the NIH AIDS Research and Reference Reagent Program). In C) Replication of wild type and mutated HIV-1 in monocyte-derived macrophages. The wild type HIV-1YU-2 (WT, open diamonds) and the vpr-defective (ΔVpr, open squares), Vpr-L23F (black circles) and -K27M (black triangles) mutant viruses were produced by transfection of 293T cells with proviral DNAs. Monocyte-derived macrophages from four healthy donors were infected in triplicates with 0.5 ng of CAp24. Virus production was then monitored by measuring the p24 antigen by ELISA 10, 14 and 17 days after infection. Results are expressed as the level of p24 in the supernatants of infected cells. Values are the means of four experiments and error bars represent 1 standard deviation from the mean.
The L23F and K27M mutations were thus introduced into the vpr gene of the macrophage-tropic HIV-1YU-2 molecular clone. As a negative control, we used the isogenic vpr-defective mutant HIV-1YU-2ΔVpr, which contains two stop codons in frame without altering the vif open reading frame. As shown in Fig. 6B, both mutated VprYU-2 proteins were efficiently incorporated into purified HIV-1YU-2 virus particles, even if the slight difference in the level of incorporation of the two Vpr mutants evidenced in panel A was still apparent. We first verified that the Vpr mutant viruses did not show replication defects in cells permissive for vpr-defective virus replication in vitro. HeLa-CD4 cells or primary lymphocytes were first infected with equivalent inocula of wt or mutant viruses. Similar replication kinetics were observed for wt HIV-1YU-2, HIV-1ΔVpr, and the Vpr-L23F and -K27M mutant viruses (data not shown). We then infected monocyte-derived macrophages (MDMs) from 4 healthy seronegative donors with the same viral inocula. Consistent with previous reports [5-8], the vpr-defective virus showed a marked replication defect in MDMs from all donors (Fig. 5B). The Vpr-L23F and -K27M mutant viruses exhibited differential replication abilities according to the donor. Compared to the wt virus, a significant decrease of replication levels of the mutant viruses was observed in MDMs from three out of four donors (Fig. 5B, donors 1, 2 and 4). Conversely, the levels of replication of the Vpr-L23F and -K27M mutant viruses were similar to that of the wt virus in MDMs from another donor (Fig. 5B, donor 3). Although we cannot exclude that the replication defect observed in donors 1, 2 and 4 may be related to the differential virion incorporation evidenced in Fig. 5A, one can note that the levels of incorporation were sufficient, at least in donor 3, for efficient replication of the Vpr-L23F and -K27M mutant viruses. While these results confirm that the absence of Vpr expression consistently affects HIV-1 replication in primary macrophages, the Vpr-L23F and -K27M mutations lead to a replication defect in macrophages from most of the donors.