Artificial 64-Residue HIV-1 Enhancer-Binding Peptide Is a Potent Inhibitor of Viral Replication in HIV-1-Infected Cells

An artificial HIV-1 enhancer-binding peptide was extended by nine consecutive arginine residues at the C-terminus and by the nuclear localization signal of SV40 large T antigen at the N-terminus. The resulting synthetic 64-residue peptide was found to bind to the two enhancers of the HIV-1 long terminal repeat, cross the plasma membrane and the nuclear envelope of human cells, and suppress the HIV-1 enhancer-controlled expression of a green fluorescent protein reporter gene. Moreover, HIV-1 replication is inhibited by this peptide in HIV-1-infected CEM-GFP cells as revealed by HIV-1 p24 ELISA and real-time RT-PCR of HIV-1 RNA. Rapid uptake of this intracellular stable and inhibitory peptide into the cells implies that this peptide may have the potential to attenuate HIV-1 replication in vivo.


Introduction
Treatment of HIV-1 infection has made considerable progress. Nevertheless, all current therapeutic approaches have been hampered by viral resistance mutations, undesired side effects, and high treatment cost [1].
Therefore, additional concepts for inhibiting HIV-1 replication, particle assembly, and escape have been developed.
One approach to interfere with HIV-1 infection at an early stage is to block viral entry. Enfuvirtide, a synthetic peptide already in clinical use corresponding to the conserved extramembranous sequence 643 to 678 of the HIV-1 envelope glycoprotein gp41, was found to prevent the conformational change of gp41 necessary for HIV-1 fusion with CD4 + cells [2,3]. Other approaches include the construction of a highly colonizing Escherichia coli (E. coli) strain secreting HIV-1 gp41-hemolysin to block HIV fusion with target cells [4], the use of an interchain disulfide bondstabilized trimeric 45-residue fusion peptide [5,6] to inhibit the fusion of a large number of HIV isolates with target cells, and the inhibition of HIV-1 infection of CD4 + T cells by microbial HSP70 [7].
Similarly, the human monoclonal 5H/I1-BMV-D5 single-chain variable region fragment (scFv) antibody selected from phage-displayed libraries has been shown to bind to the N-terminal heptad repeat region of gp41 and thus inhibited the formation of fusion intermediates in vitro and the replication of diverse clinical isolates of HIV-1 [8]. In macaques, protection from vaginal challenge with simian human immunodeficiency virus was achieved by vaginally delivered inhibitors of virus-cell fusion; moreover, two small molecule inhibitors binding to viral gp120 and cellular CCR5, respectively, prevent receptor binding by gp120, 2 Advances in Virology and a 53-residue peptide containing the C-terminal heptad repeat region of gp41 inhibited the gp41-mediated virus-cell fusion [9].
However, HIV-1 particles interacting only weakly with fusion inhibitors would still be able to enter target cells [10]. This requires other strategies to repress the viral replication after the virus has invaded the target cells. These include Highly Active AntiRetroviral Therapy (HAART) treatment [1], the intracellular expression of an antisense gene targeting the HIV-1 envelope [11] and the sequence-specific targeting of the three Sp1 sites of the 5 -long terminal repeat (LTR) of HIV-1 by engineered zinc-fingers [12]. In the present work, we have designed a 64-residue protein that crossed plasma and nuclear membranes, recognized the two highly conserved NF-κB binding sites (enhancers) of the 5 -LTR of HIV-1, and acted as a strong inhibitor of HIV-1 replication in infected cells.
This synthetic 64-residue peptide (R64, Figure 1) contained, from N-to C-terminus, the nuclear localization signal of SV40 large T antigen, the recognition helix of bacteriophage 434 repressor (residues 28 to 36) flanked at both ends by two copies of the positively charged repressor sequence 37 to 44 and nine arginine residues (Arg 9 ) for plasma membrane transduction [13]. The recognition helix of 434 repressor was used because the nucleotide sequences of the two enhancers in the 5 -LTR region of HIV-1 were identical to repressor-binding sequences of the 434 operators, and shorter versions of R64 lacking Arg 9 [14] or both the Arg 9 and the NLS sequences (R42, [14,15]) had been shown to bind to the enhancer region of synthetic HIV-1 LTR DNA, thereby displacing the p50 subunit of NF-κB from its two binding sites in LTR and inhibiting HIV-1 enhancer-driven transcription [14,15].
Despite their potential immunogenicity, therapeutic peptides may play an important role as shown by their application in the treatment of prostate cancer in mice [16], in the prevention of HIV-1 infection [2,3], and the induction of Wilms' tumor-gene-(WT1)-specific cytotoxic T lymphocytes with resultant cancer regression in human [17]. Since covalently linked oligoarginine and related sequences efficiently mediate membrane transduction of polypeptides [18,19], R64 and similar peptides could be very useful in a combinatorial therapy against HIV-1 infection by targeting the NF-κB-binding sites of the HIV-1 enhancers [20].

Synthesis and
Purification of the Peptides. The 20 to 64 residues long peptides were synthesized by the solid-phase method [22,23] on an Applied Biosystems 430A peptide synthesizer using Fmoc chemistry and following the instructions of the manufacturer. After deprotection and cleavage from the resin, they were purified on an Agilent 1100 HPLC system equipped with an autosampler, a diode array detector and an Interchim Uptisphere C18 column (250 × 4.6 mm). The peptides were eluted with a solvent gradient from 0.1% trifluoroacetic acid (TFA) to 0.075% TFA in acetonitrile. The purity of the peptides was demonstrated by ion spray mass spectrometry on a MALDI-TOF Bruker spectrometer and their concentration determined by quantitative amino acid analysis on a Biochrom 20 amino acid analyzer.

Band Shift Assays of R64-HIV LTR Interactions.
Assays followed standard procedures [21,25]. Binding reactions, performed in triplicate with identical results, contained R64 (Figure 1(b)), a synthetic 5 -end 32 P-labeled 70-base pair (bp) oligodeoxyribonucleotide (Figure 1(e)) enclosing the two NF-κB-binding sequences (enhancers, underlined residues), and a synthetic 51-bp competitor DNA (Figure 1(f)) in 20 μL phosphate buffer, pH 7.4 [26]. After incubation for 30 min at 37 • C, samples were analyzed by 10% polyacrylamide gel electrophoresis. After drying, the radioactivity in the gel was analyzed by phosphoimaging ( Figure 2). Figure 3) comprised positions -118 to -67 including two enhancer regions (Figure 3, vertical bars). The standard footprint procedure [27] was modified as follows. R64 was added to the labelled truncated HIV-1 LTR DNA and the mixture submitted to gel electrophoresis. The bands of the R64-DNA complex and of free DNA were excised, and sterically available N-7 positions of the guanine bases were methylated in the gel slices using dimethyl-sulfate [28]. After elution from the gel slices the DNA was cleaved at the sites of methylation by 10% piperidine [24], and the fragments formed were resolved by gel electrophoresis in a buffer containing 90 mM Tris-borate, pH 8.3, 25 mM EDTA. Dried gels were analyzed by phospho-imaging ( Figure 3). In parallel, sequencing of the labeled DNA by the method of Maxam and Gilbert [24] allowed localization of the binding sites of R64. The footprint assay was performed twice; the patterns obtained were in very good agreement.

HIV-1 LTR-Controlled InVitro Transcription in the
Absence and Presence of R64. OVEC-LTR, the plasmid used to test the activity of R64 in cell-free in vitro transcription, contained the two NF-κB and the three specificity protein 1 (Sp1) binding sites (positions -110 to -45) of HIV-1 LTR and a β-globin reporter gene [14,29]. As a control, plasmid Sp1-OVEC contained only Sp1 binding sites and lacked the first 19 base pairs of the β-globin gene [29]. Equal amounts of both plasmids were used (100 ng each) and the protocol described previously was followed [14]. RNA extraction, purification [30], and nuclease S1 digestion in the presence of a labeled 93-residue oligodeoxyribonucleotide [29,31] (corresponding to positions −18 to +75 of the coding strand of the β-globin reporter gene of OVEC-LTR) were followed by denaturing gel electrophoresis of the digests and phospho-imaging of the gel (Figure 4). The in vitro transcription experiments were performed in triplicate with nearly identical results.    Both extracts were analyzed by standard electrophoresis [32] and Western blotting [33] and R64 was detected colorimetrically after addition of polyclonal rabbit anti-R42 antibodies, incubation with goat anti-rabbit IgG-conjugated alkaline phosphatase, and addition of nitrotetrazolium blue chloride and the p-toluidine salt of the 5-bromo-4-chloro-3indolyl phosphate substrate (Sigma) [13,34] ( Figure 5). The experiment to study the intracellular localization of R64 was performed twice with almost identical results.
2.6. Incubation of R64 (Figure 1(b)) and R42 (Figure 1 . Transcription experiments containing the two plasmids, 100 ng each, and increasing amounts of R64 were performed as described in [14]. The total volume per experiment was 25 μL. Transcripts were hybridized with a labeled synthetic 93-mer DNA, corresponding to positions -18 to +75 of the β-globin reporter gene and then treated with nuclease S1. The resulting digests were submitted to gel electrophoresis in Tris-borate (89 mM, pH 7.5), EDTA (2 mM), urea (7 M    domain (PTD) at the N-terminus of the peptide chain. C20 (Figure 1(d)), a 20-residue control peptide, comprised the NLS and PTD sequences of R64, linked by two glycine residues.
HIV-1 stocks were prepared from cultures of human H9 lymphoblastoid cells (NIH AIDS Research and Reference Reagent Program) maintained in complete medium (RPMI-1640 supplemented with 2 mM L-glutamine, 24 mM sodium bicarbonate, and 10% heat-inactivated fetal calf serum (Life Technologies)) and chronically infected with HIV-1 strain III B (HIV-1 IIIB ). HIV-1-containing supernatant was filtered through a 0.2 μm filter, aliquoted, and stored at −70 • C until used.
Human CEM-GFP cells were in an exponential growth phase before being pelleted in 50 mL conical tubes (Falcon) to give ∼ 9.6 × 10 6 cells per tube.
Stocks of HIV-1 IIIB obtained from the supernatant of human H9 T cells were diluted 1 : 5 with complete medium, a concentration previously determined to give a continuous rate of HIV replication (data not shown). The CEM-GFP cell pellets were incubated with 1-ml aliquots of diluted HIV-1 and gently resuspended. After incubation at 37 • C for 2 h, excess virus was removed and the cells were thoroughly washed with RPMI-1640 medium.

7
At different time points, the HIV-1 p24 core protein in the cell-free culture supernatants of the wells was determined by quantitative ELISA as described previously [10]. Briefly, duplicate supernatants (25 μL each) of the cultures of the HIV-1-infected, peptide-treated CEM-GFP cells were diluted 1 : 6 with buffer and then added to microplate wells precoated with a mouse monoclonal anti-p24 antibody. This was followed by addition of a biotinylated human polyclonal anti-p24 antibody and incubation with streptavidin-horseradish peroxidase conjugate. Additional binding sites for streptavidin-horseradish peroxidase, created by treating the wells with biotinyl-tyramide [40], were reacted with a dilute solution of streptavidin-horseradish peroxidase as recommended by the manufacturer. Finally, the ophenylene diamine substrate was added and the resulting yellow color measured spectrophotometrically at 490 nm. p24 was quantified by comparing the absorbances of the samples with those of a standard curve obtained from p24 concentrations between 0.17 pg/mL and 10 ng/mL.
To determine the HIV-1 RNA levels in the culture supernatants, quantitative real-time RT-PCRs specific for the gag region of HIV-1 were performed as described in [41] using 1 μL of cell-free culture supernatant derived from the HIV-1-infected, peptide-treated CEM-GFP cells, or 1 μL of an RNA standard dilution series obtained from the supernatant of HIV-1 IIIB -infected human H9 lymphoblastoid cells and calibrated using the Amplicor (Roche) quantitative RT-PCR assay.
After eight days of incubation of the HIV-1-infected cells with peptides, possible peptide-derived cytotoxic effects were assessed using the MTT assay [38] with the following modifications. The infected CEM-GFP cells grown in the presence or absence of peptide were resuspended, duplicate 100 μL aliquots transferred to a 96-well ELISA plate (Nunc), and 10 μL of a solution of 5 mg MTT per mL PBS added to each well. After a four-hour incubation, 100-μL solubilization solution (10% SDS, 0.5% Triton X-100, 0.01 M HCl) was added per well and the plate further incubated at 37 • C overnight. Finally the absorption was measured in an ELISA plate reader (TecScan) at 570 nm with background subtraction at 630 nm.

Results and Discussion
We have shown earlier that a 42-residue peptide derived from the DNA-binding domain of bacteriophage 434 repressor bound specifically to the enhancer region of HIV-1 long terminal repeat and repressed the in vitro transcription of HIV-1 enhancer-containing plasmids [14]. In the present work we have demonstrated that this peptide was active in HIV-1-infected cells after it had been extended by a protein transduction domain to cross the plasma membrane and by a nuclear localization signal to enter the nucleus. The best results were obtained with peptide R64 (Figure 1(b)).

DNA Binding Specificity of R64 and R62 at the Two HIV-1 LTRNF-κB Binding
Sites. The binding specificity of R64 and R62 was determined by competitive band shift and footprint assays. Band shift assays in the presence of R64 (Figure 2) showed that, in the experimental conditions chosen, an approximately 100-fold molar excess of competitor DNA (Figure 1(f)) was required to displace the peptide from the enhancer-containing HIV-1 LTR target DNA (Figure 1(e)).
The result of the band shift assay performed in the presence of R62 was very similar (data not shown).
Footprint analysis followed a procedure [42] in which protein-DNA complexes and free DNA were first separated by band shift electrophoresis and then treated with a DNAcleaving reagent in the gel matrix. The resulting DNA fragments were eluted from the gel and resolved by gel electrophoresis. In the present work, the electrophoretically separated bands of an R64-HIV-1 LTR DNA complex and free HIV-1 LTR DNA were excised from the gel and the DNA in the bands was treated with dimethyl sulfate, eluted from the gel slices, and then incubated with piperidine [24,28]. The DNA fragments obtained were separated by electrophoresis on a sequencing gel (Figure 3). The results showed that the G bases of the two NF-κB binding sites (enhancers, marked by vertical bars) of HIV-1 LTR were largely protected from methylation in the presence of R64 (Figure 3, lanes 3 to 5) indicating that R64, like R42 [26], could compete with NF-κB for HIV-1 enhancer binding and thus inhibit enhancer-controlled transcription. This was confirmed by the following experiments.

In Vitro Transcription Suppression.
In HeLa cell nuclear extracts, R64 suppressed the transcription of a reporter gene from an HIV-1 enhancer-containing plasmid (OVEC-LTR) more strongly than the transcription of this gene from a control plasmid lacking the HIV-1 enhancer region (Sp1-OVEC) [14,29] (Figure 4, lanes 5 and 6). R62, largely identical with R64 except that an N-terminal bipartite nuclear localization signal [43] embraces the protein transduction domain (Figure 1(c)), was less specific in the suppression of HIV-1 enhancer-controlled transcription than R64 despite its lower net positive charge (+25 versus +28, data not shown).

Analysis of the Intracellular Localization of R64.
R64 was shown to be present in both the cytosolic and nuclear extract of R64-incubated CHO-K1 cells using a double antibody assay following electrophoresis and Western blotting of the two extracts ( Figure 5). R64 was detected when its concentration in the incubation mixture (∼10 5 cells/3 mL medium) was above 1.3 μM. The minute increase in the amount of R64 in the nuclear extracts going from 1.7 μM to 2.3 μM R64 in the incubation mixtures ( Figure 5, lanes 5 and 6) indicated that the "receptor" for nuclear protein import [44] was almost saturated at 1.7 μM R64 in the incubation mixture and that excess intracellular R64 remained in the cytosol ( Figure 5, lanes 9 and 10). in the nuclear extract of R64-incubated CHO-K1 cells ( Figure 5), it was tested whether this peptide was active in intact CEM-GFP cells [35] which contain an HIV-1 LTRcontrolled GFP reporter gene whose expression is stimulated, via activation of NF-κB, by HIV-1, TNF-α, interleukin-1, lipopolysaccharide, and PMA [36,37,45]. The HIV-1 p24 ELISA was performed as described [10]. In Figure 7, the results obtained with R64, R62 (Figure 1(c)), R42 (Figure 1(a)), and C20 (Figure 1(d), comprising only the nuclear and the plasma membrane transduction signals of R64) are compared. Noticeable p24 production in the presence of 0-7 μM R64 (Figure 7(a)) started approximately six days after addition of the peptide except in the samples containing 3.5 and 7 μM R64, respectively, in which HIV-1 p24 levels remained close to zero. The production of p24 in the presence of 0-7 μM R62, R42, and C20 (Figures 7(b), 7(c) and 7(d)) started uniformly after approximately five days of incubation and then rose logarithmically without efficient inhibition by these peptides.
Real-time RT-PCR experiments [41] showed that the number of HIV-1 RNA copies per mL supernatant of infected cells dropped by nearly three orders of magnitude in the presence of 7 μM R64 (Figure 9(a)) but remained constantly high in the presence of 7 μM R62, R42 or C20 in the infection experiments (Figures 9(b), 9(c) and 9(d)). At maximal concentration of R64, HIV-1 replication was inhibited more than 300-fold. Figure 9 also indicates the degree of growth inhibition of the HIV-1-infected CEM-GFP cells in the presence of R64, R62, R42 and C20, respectively. The MTT assay used for this purpose measures metabolically active, viable cells. The maximum of 20% inhibition observed in the presence of R64 (Figure 9(a)) indicates that 80% of the cells are still metabolically active and viable. A loss of only 20% of the metabolic capacity of these cells cannot account for more than 300-fold inhibition of HIV-1 replication observed with R64.
R62, which differed from R64 only in sequence and arrangement of the nuclear localization and plasma membrane transduction signals, had no antiviral activity in HIV-1-infected CEM-GFP cells. From earlier work we knew that R42 was not able to pass the plasma membrane and therefore was inactive in these assays. C20 presumably reached the nucleus but its sequence did not have enhancer-binding specificity.
In Figures 8 and 10, by using the CalcuSyn 1.1 (Biosoft) software package [39], median-effect plots were derived from dose-effect curves obtained for R64 as antiretroviral compound using the median-effect equation: where D is the dose, D m is the dose required for a 50% effect, f a is the fraction affected by the dose, f u is the unaffected fraction and m is a coefficient of the sigmoidicity of the dose-effect curve. The median-effect dose (D m ) was used to describe the 50% effective dose (ED 50 ) and in this case, this ED 50 corresponds to IC 50 (50% inhibitory concentration) which is the drug level needed to block 50% of HIV's normal replication in vitro. R64 peptide titrations of HIV-1-infected CEM-GFP cells analyzed by p24 ELISA and one tube real-time RT-PCR, showed a 50% inhibitory R64 concentration between 0.588 and 0.493 μM, respectively.
In summary, based on previous work [14,15,18,26], we have designed a 64-residue peptide (R64, Figure 1(b)) containing a central HIV-1 enhancer recognition helix and terminal nuclear localization and plasma membrane transduction signals. This peptide bound to the two HIV-1 enhancers and thus competed with NF-κB, the major activator of HIV-1 replication [46,47]. After incubation of CHO-K1 cells with R64, the peptide could be localized to cytosol and nucleus of these cells. In CEM-GFP cells, it suppressed the HIV-1 enhancer-driven expression of the GFP gene after PMA stimulation and inhibited viral transcription and replication after HIV-1 infection in a concentrationdependent manner. Most likely, R64 displaced NF-κB from its two binding sites in HIV-1 LTR despite the presence of NF-κB binding sites on many cellular genes [48]. An artificial 62-residue peptide (R62, Figure 1  LTR enhancer region. However, it was active only in a cellfree in vitro transcription assay and did not inhibit HIV-1 replication in infected cells. Therapeutic peptides have important applications in human diseases such as prevention of HIV-1 infection [2,3], induction of tumor-specific cytotoxic T-lymphocytes [17], and as therapeutic vaccines [49]. R64 seemed to fulfil some of the requirements as a therapeutic peptide: the Western blots of the cytosolic and nuclear extracts of R64-incubated CHO-K1 cells ( Figure 5) showed no fragments, indicating stability of the peptide against intracellular proteolytic degradation; R64 reduced the viral replication by almost three orders of magnitude and was not cytotoxic at a concentration of 7 μM in the medium of HIV-1-infected CEM-GFP cells ( Figure 9); and uptake of R64 into cells appeared to be fast ( Figure 6), lowering the immunogenic potential of this peptide.