The Cationic Amphipathic α-Helix of HIV-1 Viral Protein R (Vpr) Binds to Nucleic Acids, Permeabilizes Membranes, and Efficiently Transfects Cells*

Viral protein R (Vpr) is a small protein of 96 amino acids that is conserved among the lentiviruses human immunodeficiency virus type 1 (HIV-1), HIV-2, and simian immunodeficiency virus. We recently sought to determine whether the karyophilic properties of Vpr, as well as its ability to bind nucleic acids, could be used to deliver DNA into cells. We have found that the C-terminal domain of Vpr-(52–96) is able to efficiently transfect various cell lines. Here, we show that the shortest active sequence for gene transfer corresponds to the domain that adopts a α-helix conformation. DNA binding studies and permeabilization assays performed on cells demonstrated that the peptides that are efficient in transfection condense plasmid DNA and are membranolytic. Electron microscopy studies and transfection experiments performed in the presence of inhibitors of the endocytic processes indicated that the major entry pathway of Vpr-DNA complexes is through endocytosis. Taken together, the results show that the cationic C-terminal α-helix of Vpr has DNA-condensing as well as membrane-destabilizing capabilities, both properties that are indispensable for efficient DNA transfection.

Unlike other retroviruses, HIV-1 is able to replicate in nondividing cells. How the preintegration complex is imported in the nucleus remains unclear. Four viral components of the preintegration complex reportedly have karyophilic properties, namely the integrase, the matrix protein, Vpr, and the central DNA flap (12)(13)(14)(15)(16). Active transport of proteins into the nucleus requires specific peptide signals referred to as nuclear localization signals. Although Vpr does not contain a canonical nuclear localization signal, it localizes to the nucleus, it can interact with host proteins related to nuclear transport such as importin-␣ and nucleoporins, and it promotes nuclear entry of viral nucleic acids in nondividing macrophages (17)(18)(19). Finally, it was reported that Vpr can disrupt the nuclear envelope, thereby providing a possible entry route for the preintegration complex (20).
The goal of nonviral gene transfer is to mimic the successful viral mechanisms for overcoming cellular barriers while minimizing the problems associated with the use of biological vectors. Most nonviral vectors are able to complex DNA and facilitate its entry into the cell as well as its escape from the endosome. Yet, nuclear transport remains the major bottleneck in successful gene transfer with synthetic DNA carriers (25,26). Considering the karyophilic properties of Vpr as well as its ability to bind nucleic acids (27,28), we have recently explored the possibility of using Vpr as a DNA transfection agent. We have found that the C-terminal fragment-(52-96) of Vpr, but not the whole protein, is able to deliver DNA efficiently into different cell lines (29). It was the first example of a peptide derived from a natural protein displaying such a high transfection activity in the absence of auxiliary agents. In the present work, we have determined the shortest Vpr sequence with gene transfer activity and studied the different steps of Vpr-mediated transfection from DNA compaction to endosomal escape of complexes.
DNA Retardation Assay-DNA binding was studied by means of agarose gel retardation assays. One g of DNA and increasing amounts of peptide were each diluted in 25 l of 150 mM NaCl and mixed. After 20 min, samples were electrophoresed through a 1% agarose gel using Tris borate-EDTA buffer, and DNA was visualized after ethidium bromide staining.
Transfection Experiments-Cells were plated 1 or 2 days before transfection to obtain a confluency of 60 -80% at the time of the experiment. For experiments performed in 24-or 48-well plates, 4 or 2 g, respectively, of plasmid DNA and the desired amount of peptide, pLys, DOTAP, or PEI were diluted into 100 or 50 l, respectively, of 150 mM NaCl and gently mixed. After 20 min of incubation, the mixture was diluted with serum-free medium to a final volume of 1 or 0.4 ml, respectively; 0.5 or 0.2 ml, respectively, of the transfection mixture was then put on each well of the duplicate for 3 h. The transfection medium was then replaced with DMEM, 10% fetal calf serum, and transgene expression was evaluated 24 -48 h after the beginning of the transfection. Each experiment was carried out several times; within a series, experiments were done in duplicates.
Transfections in the presence of dimethylamiloride (final concentration 62.5-250 M) and cytochalasin B (final concentration 5-40 M) were performed as described above except that the drug was added to the cells in serum-free medium prior to the addition of complexes (10 and 30 min before transfection for dimethylamiloride and cytochalasin B, respectively). For transfections with methyl-␤-cyclodextrin (M␤CD) or cholesterol-charged methyl-␤-cyclodextrin (M␤CD-Chol; at a final concentration of 5-10 mM), cells were incubated for 1 h with the drug in serum-free medium before transfection. The transfection experiments involving chloroquine (Sigma) at a final concentration of 100 M were done as described above except that the drug was added after dilution of the complexes with DMEM, just prior to the addition of the transfection medium to the cells.
Transgene Expression-The luciferase assay was performed as described previously (29). Luciferase background was subtracted from each value, and the transfection efficiency, expressed as light units/10 s/well (with 1 light unit ϭ 10 counts), is the mean of duplicates. When drugs (dimethylamiloride, methyl-␤-cyclodextrin, and cholesterolcharged methyl-␤-cyclodextrin) were present during transfection, the results were normalized by a Bradford protein quantification assay. These results were then expressed as light units/10 s/mg of protein. The LacZ activity was measured by chemiluminescence as recommended by the manufacturer (Tropix).
Ethidium Bromide Exclusion Assay-One g of DNA was complexed with increasing amounts of peptide in a final volume of 50 l. Fifty l of a 150 mM NaCl solution containing ethidium bromide (8 g/ml) was then added to the complexes. The fluorescence resulting from ethidium bromide intercalation in DNA was measured with a 96-well fluorimeter (Spectramax, Gemini; excitation 485 nm, emission 590 nm). Results were expressed as the percentage of the maximum fluorescence signal when ethidium bromide was bound to DNA in the absence of competition.
Erythrocyte Lysis Assay-After centrifugation of 10 ml of fresh human blood for 10 min at 1000 ϫ g, the plasma and the white layer of leukocytes were removed. The erythrocytes were washed five times with 11 mM sodium citrate in Hepes-buffered saline, pH 7.3. The solution was then divided into two aliquots, which were washed three times and resuspended in assay buffer with the appropriate pH (200 mM sodium citrate, pH 5, or 11 mM sodium citrate in Hepes-buffered saline, pH 7.3) at a concentration of 10 8 cells/ml. A 75-L aliquot of erythrocytes was added in each well of a 96-well cone-type microtiter plate containing 75 l of a serial dilution of the compound to be tested in assay buffer. The plate was then gently shaken for 1 h at 37°C. Controls (100 and 0% lysis) were obtained by incubating erythrocytes with 4 l of Triton X-100 or assay buffer. After removal of the unlysed erythrocytes by centrifugation for 5 min at 1000 ϫ g, 75 l of the supernatant was transferred to a new microtiter plate (flat-bottom), and hemoglobin absorption was determined at 450 nm (background correction at 750 nm). The lysis percentage is given by the following formula: 100 ϫ [(OD 450 -OD 750 ) product Ϫ (OD 450 -OD 750 ) buffer ]/[(OD 450 -OD 750 ) Triton X-100 Ϫ (OD 450 -OD 750 ) buffer ].
Cell Permeabilization-HepG2 cells, plated in 24-well plates, were incubated for 1 h at room temperature to block endocytosis. The test compound, diluted in 250 l of PBS containing 5 g of ethidium bromide, was then added to the cells. After a 30-min incubation at room temperature in the dark under gentle shaking, cells were washed once with PBS, harvested with 1 mM EDTA/PBS, and analyzed by flow cytometry (FACScalibur, BD Biosciences).
Electron Microscopy-Peptides were mixed with 0.02 g/l DNA in a final volume of 50 l of NaCl, 150 mM. Five l of the mixture was deposited onto an electron microscope grid covered with a thin carbon film previously activated by a glow discharge in the presence of pentylamine. The grids were then stained with 2% aqueous uranyl acetate, drained, and blotted. The observations were done with the annular dark-field mode in a Zeiss 902 EM, filtering out inelastically scattered electrons for enhanced contrast and resolution. For intracellular trafficking studies, 20 g of plasmid was mixed with peptide in a 150 mM NaCl solution in a final volume of 1 ml. After 20 min, serum-free medium was added, and the solution was pipetted onto the cells plated 1 day earlier in a 15-cm dish. After 3 h, the transfection medium was replaced with DMEM containing 10% serum. At different times, cells were fixed with medium containing 10% fetal calf serum and 2% glutaraldehyde, harvested, and centrifuged. The pellet was resuspended in Sörensen's buffer (67 mM phosphate buffer, pH 7.4) and then postfixed in 2% osmium tetroxide, dehydrated with ethanol and propylene oxide, and embedded in Epon. Ultrathin sections were prepared with an LKB Ultrotome. Sections of cells were colored with uranyl acetate and lead citrate and observed with a LEO 902 microscope.

RESULTS AND DISCUSSION
Definition of the Shortest Vpr-derived Peptide with Transfection Activity-To find the shortest sequence allowing efficient gene transfer, different subfragments of Vpr-(52-96) were synthesized, and their transfection activity was evaluated on two cell types. Increasing amounts of peptides were complexed to a luciferase expression plasmid (CMV-Luc) and incubated with the cells for 3 h. Luciferase activity was measured 30 h later. The cationic polymer 25-kDa PEI (30), one of the most efficient transfection reagents, was included as a positive control. On human HEK-293 cells, Vpr-(55-91), Vpr-(55-86), and Vpr-(55-82) allowed for gene transfer levels comparable with those obtained with PEI and about 10 times higher than those of Vpr-(52-96) (not shown). On human HepG2 cells, Vpr-(55-91) was the most active fragment, resulting in luciferase levels 1 log above those obtained with PEI. Vpr-(52-96), Vpr-(55-86), and Vpr-(55-82) were at least as active as PEI (Fig. 1). When five amino acid residues were removed from the N terminus of Vpr-(55-91) (Vpr-(60 -91)), the transfection activity was significantly reduced, indicating that this stretch is indispensable. On the other hand, the efficiency of Vpr-(52-75), which is seven residues shorter on the C-terminal end than Vpr-(55-82), was very low. Thus, among the different subfragments, the shortest active sequence for gene delivery is Vpr-(55-82) (Fig. 1). Inter-estingly, this sequence corresponds to the C-terminal domain, which adopts a ␣-helix conformation in Vpr-(1-96) (23). The Leu 60 and Leu 67 side chains are located on the hydrophobic side of the helix, and it was shown that they are involved in Vpr dimerization through a leucine zipper-type mechanism. Although the replacement of these two leucines by alanine residues eliminates Vpr dimerization (21), it preserved the helical structure of the peptide and had no effect on the transfection activity of Vpr-(52-96) (data not shown).
DNA Compaction-As shown in Fig. 1, DNA binding is a necessary but not sufficient condition for transfection activity of the Vpr-derived peptides. We wondered whether different structures of the DNA complexes could be related to differences in transfection efficiency between subfragments. We therefore evaluated the relative affinity of three peptides (Vpr-(55-91), -(60 -91), and -(55-86)) for plasmid DNA by performing an ethidium bromide exclusion experiment. This was achieved by preparing complexes of DNA at different charge ratios and adding an excess of ethidium bromide prior to spectroscopic analysis. DNA accessibility can be evaluated with this assay because a large increase in fluorescence is observed when the phenanthridium moiety of ethidium bromide intercalates DNA. Fig. 2A shows that maximal DNA condensation was reached for all the compounds tested, including poly-L-lysine, at a ϩ/Ϫ charge ratio between 1 and 2. Moreover, no differences were observed between transfecting (Vpr-(55-91) and Vpr-(55-86)) and poorly transfecting (Vpr-(60 -91)) peptides.
The structure of the DNA complexes was further characterized by electron microscopy. Fig. 2B, panel a, shows that DNA complexes obtained with the optimal amount of Vpr-(55-91) for gene transfer are large aggregates, likely to contain high copy numbers of plasmid DNA. Aggregates formed with subfragment-(60 -91) and -(55-86) were similar (Fig. 2B, panels b  and e), although sometimes unbound DNA fibers that extended outward from the condensed region were observed (panels c and d). In contrast, a large part of the DNA remained uncondensed in complexes generated with Vpr-(52-70), a fragment that contains only one positively charged amino acid (Fig. 2B,  panel f).
The formation of large aggregates with cationic amphipathic peptides may be explained as follows (31): positive charged residues in the peptides interact electrostatically with the negative charge of phosphate in DNA, whereas the opposite hydro-phobic side creates interactions between peptide-DNA complexes, resulting in aggregation.
Taken together, the results obtained by electron microscopy and the ethidium bromide exclusion assay are in good agreement. They show that DNA compaction, even when complexes have similar structures (as judged by electron microscopy for Vpr-(55-91) and -(60 -91)), is not sufficient for efficient transfection.
Entry Pathway into the Cell-We have previously shown that Vpr-(52-96) delivers high numbers of DNA molecules into the cells (29). However, the mechanism by which the delivery is done is unknown. To evaluate this process, we used electron microscopy. At early times (i.e. after 2 h; Fig. 3) the Vpr-(52-96)-DNA complexes appeared as electron-dense particles at the cell surface. Then, as the duration of incubation increased, the complexes were taken up into the cell by an endocytic process. Once in the cytoplasm, the DNA particles were found exclusively within large vesicles and were no longer visible after 24 h (not shown). To investigate the uptake mechanism of DNA complexes, cells were treated before or during transfection with chemical agents that interfere with the endocytic processes. HepG2 cells were transfected with Vpr-(55-91) in presence of cytochalasin B, which inhibits phagocytosis and pinocytosis but not receptor-mediated endocytosis (32). The results show that the luciferase levels were slightly increased in the presence of the drug (data not shown). Similar observations have been reported by others using DNA formulations containing either a cationic lipid or a lipid/peptide mixture (33, 34).
Because we observed very large vesicles containing DNA complexes (Fig. 3), we further examined the uptake mechanism by using dimethylamiloride, a compound that inhibits macropinocytosis (35). In these experiments, performed with HEK-293 cells, we used a human recombinant adenovirus (Ad-LacZ) as negative control because macropinocytosis is not essential for viral uptake (36). We observed a 4-fold decrease in reporter gene expression in the presence of 250 M dimethylamiloride for Vpr-(55-91)-and PEI-mediated transfection, whereas the adenoviral transduction efficiency was slightly increased (data not shown). Thus, macropinocytosis is not an essential entry pathway for Vpr-(55-91)-DNA complexes. Finally, transfection was performed following cholesterol depletion of the plasma membrane with M␤CD (37). Cholesterol depletion results in the inhibition of clathrin-mediated endocytosis, although it also affects the structure and function of invaginated caveolae, including caveolae-dependent endocytosis (38). When HepG2 cells, which lack caveolae (39), were depleted of cholesterol prior to transfection by a 1-h treatment with M␤CD, the luciferase levels were decreased about 100-fold for Vpr-(55-91)-and PEI-mediated transfection (Fig. 4). The efficiency of the cationic lipid DOTAP was scarcely altered (Fig. 4), whereas that of the cationic lipid/DOPE formulation, LipofectAMINE, was strongly reduced (not shown). In fact, a decrease in DOTAP activity was detected only with higher concentrations of M␤CD. A similar significant reduction of the transfection efficiency after M␤CD treatment was recently reported with the lipidic formulation SAINT-2/DOPE (40).
The results show that the presence of cholesterol in the membranes is essential for efficient Vpr (but also PEI and lipid)-mediated transfection, suggesting that the major entry pathway is through clathrin-mediated endocytosis.
Hemolytic Activity of Vpr Fragments-The results described above do not exclude the possibility that DNA may also enter the cytoplasm directly. Indeed, peptides from the C-terminal region of Vpr including the conserved HFRIGCRHSRIG motif (Fig. 1) can cause permeabilization of yeast cells (41). Recent results showing that Vpr can gain access to intracellular compartments independently from the infection process support the idea that Vpr is membrane-active (24).
This opens the possibility that "active" DNA complexes could enter the cell by membrane permeabilization. Alternatively, the permeabilization activity of Vpr could be required for endosomal escape. To determine whether the membranolytic activity is required for efficient gene transfer, we evaluated the capacity of different subfragments of Vpr to lyse freshly prepared human erythrocytes. Increasing amounts of either Vpr-(52-96), -(55-91), -(60 -91), or -(1-51) were incubated with erythrocytes at neutral pH. After 1 h, the amount of hemoglobin released was measured by spectrophotometry. Fig. 5A  shows that Vpr-(1-51) was completely inactive, whereas Cterminal subfragments were all hemolytic at various degrees ((55-91) Ͼ (52-96) Ͼ (60 -91)). This activity was maintained in the presence of DNA. Interestingly, it was strongly inhibited at acidic pH (Fig. 5B).
These results indicate that there is a correlation between hemolysis and the activity of Vpr derivatives as transfection agents. Such a correlation has been described for other peptides such as those derived from influenza HA2, which potentiate pLys-DNA complexes and destabilize membranes at acidic pH (42). In contrast to these peptides, Vpr subfragments act alone and are inactivated at acidic pH, suggesting a different mechanism for membrane permeabilization.
Cell Membrane Permeabilization Activity of Vpr Derivatives-The permeabilization activity of Vpr fragments was also evaluated on HepG2 cells. Cells were first incubated for 1 h at room temperature to reduce endocytic processes, and the peptide was then added together with ethidium bromide, a poorly membrane-permeant molecule that becomes strongly fluorescent upon binding to DNA. Positive control was obtained by incubating the cells with melittin, a highly permeabilizing peptide (43), whereas incubation with ethidium bromide alone was used as negative control. The results show that several peptides, including Vpr-(55-91), -(60 -91), -(52-70) (Fig. 6), and -(55-86) and -(55-82) (not shown), induced an increase of the cell fluorescence. Among the three peptides shown in Fig. 6, Vpr-(55-91) in the absence of DNA had the highest activity followed by Vpr-(60 -91) and Vpr-(52-70). In the presence of DNA, the permeabilization efficiency of the peptides was not altered, except for Vpr-(55-91) (Fig. 6). These results demonstrate that Vpr subfragments are able to permeabilize plasma membranes of mammalian cells in the presence of DNA.
Influence of Cholesterol Content on Permeabilization Activity-As described above, reduction of the membrane cholesterol content reduces the transfection efficiency of Vpr-(55-91). Besides inhibiting clathrin-mediated endocytosis, cholesterol depletion can modulate the membrane disruption activity of peptides (44). To evaluate the influence of cholesterol content on the membrane disruption activity of Vpr, we pretreated HepG2 cells with either M␤CD, to deplete plasma membrane of cholesterol, or with M␤CD-Chol complex, to enrich the membranes with the sterol. The M␤CD Ϯ Chol treatments did not significantly modify the permeabilization activity of Vpr-(55-91) (not shown). We then checked whether M␤CD-Chol treatment results in an enhanced transfection efficiency. HepG2 membranes were enriched with cholesterol before transfection. The results show that Vpr-(55-91)-mediated transfection was slightly increased under these conditions compared with control, whereas the efficiency of DOTAP and PEI was slightly reduced (not shown).
These results indicate that the membrane cholesterol content only moderately modifies the permeabilization activity of Vpr subfragments. Thus, the 2-log decrease, observed on transfection efficiency after M␤CD treatment (Fig. 4), is not due to the inhibition of the permeabilization activity.
Vpr as Helper for Polylysine-mediated Transfection-pLys-DNA complexes escape rather inefficiently from internal vesicles, but endosomolytic agents such as chloroquine or fusogenic peptides can be used as helper during pLys-mediated transfection (42,45). We reasoned that if Vpr subfragments are able to disrupt membranes, then they should be able to enhance pLysmediated transfection. To evaluate this, pLys-CMV-Luc complexes were pre-formed, and Vpr or Vpr complexed with salmon sperm carrier DNA was added. Fig. 7 shows that in the presence of Vpr-(55-91) or Vpr-(60 -91), the luciferase activity obtained with pLys was increased 23-and 37-fold, respectively, whereas the addition of an excess of pLys (ϮDNA) did not enhance transfection. The helper effect of the two Vpr deriva- tives was almost as important as that of chloroquine. When the peptides were complexed to DNA, the helper effect was still observed, although it was slightly less important. These results suggest that Vpr-(55-91) and -(60 -91), free or complexed to DNA, are able to enhance the escape of pLys-DNA complexes from endocytic vesicles. However, it cannot be excluded that other effects of Vpr, such as an increased cell entry due to aggregation of the complexes or a more efficient nuclear transport of DNA, are also implicated in this enhancement.
Taken together, these results show that the N-terminal domain of Vpr-(52-96) plays an important role during transfection, probably by enhancing the efficiency of endosomal escape of DNA.
The aim of this study was to clarify how a particular family of nonviral DNA carriers, namely peptides derived from Vpr of HIV-1, acts. In the ethidium bromide exclusion assay, maximal DNA condensation with different Vpr fragments was obtained at the same charge ratio as polylysine. However, the capacity to compact DNA is not sufficient to allow transfection. As demonstrated by different permeabilization assays, the Vpr-derived transfecting peptides are also able to destabilize membranes. As the major entry pathway is through endocytosis, our results suggest that the permeabilizing activity of Vpr peptides allows endosomal escape of DNA. Release of the DNA probably takes place before acidification of the endosome occurs, because the membranolytic activity is strongly reduced at acidic pH. The escape of DNA from endosomes remains, however, a rate-limiting step with complexes being trapped in endocytic vesicles even at 10 -14 h post-transfection (Fig. 3). Once released into the cytosol, DNA must be protected against enzymatic degradation and transported into the nucleus. We showed previously that when Vpr-(52-96)-DNA complexes are incubated with DNase I for 1 h at 37°C, the integrity of the plasmid is not preserved (29). Thus, DNA degradation may be another limiting step. As indicated by electron microscopy, the Vpr-DNA complexes tend to form multimolecular aggregates, which are too large to cross an intact nuclear membrane. Therefore, the ability of Vpr to transfect nondividing cells has to be investigated. However, it is possible that smaller particles are also generated. Given that reporter expression is due to a minority of plasmids entering the nucleus (29), DNA transfection may be mediated by small particles, whereas aggregates, representing the major population of the complexes, are not productive. Alternatively, the results obtained by de Noronha et al. (20) open the possibility that aggregates enter the nucleus by disrupting the nuclear envelope.
The smallest active fragment corresponds to the C-terminal domain, which adopts a ␣-helix conformation in Vpr- . This may suggest that this particular conformation is important for efficient gene transfer. In fact, it is interesting to note that other cationic amphipathic peptides with high transfection activities, such as KALA (46) or ppTG20 (47), are also characterized by their capacity to bind DNA, destabilize membranes, FIG. 7. Vpr fragments as the auxiliary agent for pLys-mediated gene transfer. The PEI-DNA (N/P ϭ 6.7) (10 g of Vpr-(55-91)/2 g of DNA (charge ratio (ϩ/Ϫ) ϭ 2.3), 20 g of Vpr-(60 -91)/2 g of DNA (ϩ/Ϫ ϭ 5), and 2 g of pLys/2 g of DNA (ϩ/Ϫ ϭ 1.7)) complexes were generated as described under "Experimental Procedures." The other formulations were prepared as follows: 2 g of pLys/2 g of CMV-Luc in 150 mM NaCl were diluted in serum-free DMEM containing 10 g of Vpr-(55-91) Ϯ 2 g of carrier DNA (salmon sperm DNA), 20 g of Vpr-(60 -91) Ϯ 2 g of carrier DNA, or 2 g of pLys Ϯ 2 g of carrier DNA. Transfection of HEK-293 cells was then carried out as described previously. The luciferase activity was determined 30 h later. The transfection efficiency was expressed as total light units/10 s/well, and the mean of duplicates is shown.
Because the sequence of a peptide can be modified easily, we can envision further improvement of the transfection efficiency of Vpr either by sequence alteration or by introducing a motif that provides cell specificity.