Formation of a Unique Cluster of G-Quadruplex Structures in the HIV-1 nef Coding Region: Implications for Antiviral Activity

G-quadruplexes are tetraplex structures of nucleic acids that can form in G-rich sequences. Their presence and functional role have been established in telomeres, oncogene promoters and coding regions of the human chromosome. In particular, they have been proposed to be directly involved in gene regulation at the level of transcription. Because the HIV-1 Nef protein is a fundamental factor for efficient viral replication, infectivity and pathogenesis in vitro and in vivo, we investigated G-quadruplex formation in the HIV-1 nef gene to assess the potential for viral inhibition through G-quadruplex stabilization. A comprehensive computational analysis of the nef coding region of available strains showed the presence of three conserved sequences that were uniquely clustered. Biophysical testing proved that G-quadruplex conformations were efficiently stabilized or induced by G-quadruplex ligands in all three sequences. Upon incubation with a G-quadruplex ligand, Nef expression was reduced in a reporter gene assay and Nef-dependent enhancement of HIV-1 infectivity was significantly repressed in an antiviral assay. These data constitute the first evidence of the possibility to regulate HIV-1 gene expression and infectivity through G-quadruplex targeting and therefore open a new avenue for viral treatment.


Introduction
G-quadruplexes (G-4s) are inter-and intramolecular fourstranded structures formed by G-rich DNA and RNA. G-4s are based on the formation of G-quartets, which are stabilized by Hoogsteen-type hydrogen bonds between guanines and by the interaction with cations located between the tetrads. G-quartets stack on top of each other to give rise to G-4s. G-rich oligonucleotides can be very polymorphic, and the adopted structures are dependent on several factors, including the base sequence, strand concentration, loop connectivities, and cations present. Such tetraplex DNAs have been an emerging topic in nucleic acids research because of the recent indication of their involvement in a series of key biological functions.
Initially G-4s have been shown to form in vitro [1] and in vivo [2] in the G-rich sequence of telomeres: stabilization of the G-4 folded structure has been proposed as an effective approach to inhibit telomerase activity in tumour cells [3]. Subsequently, it has been shown that G-rich sequences are tightly clustered immediately upstream and downstream of the transcription start site in proliferation-associated genes, therefore suggesting a role of tetraplexes as cis-acting regulatory elements in gene expression [4], in particular in genes important in cell signalling, which have representatives from the six hallmarks of cancer [5].
G-rich sequences capable of forming G-4 were also found within coding regions of the human genome, including minisatellites [6], immunoglobin heavy chain switch regions [7] and rDNA [8], and were shown to be the target of binding proteins [9]. The folding of DNA G-4 most likely occurs during single-stranded DNA formation, i.e. during transcription and replication [10]. Recent work has shown that G-4 can also form within coding regions: on the leading strand G-4s arise during replication and promote genetic instability in human and yeast [11,12]; on the lagging strand G-4-forming sequences were found to generate G loops during transcription both in vitro and in Escherichia coli [13,14]. Further, association between G-4single nucleotide polymorphisms and expression of the corresponding gene in individuals has been proposed [15]. Recently, cell-cycle dependent G-4 formation in living cells and their stabilization by G-4 ligands has been demonstrated [16].
The importance of G-4-forming sequences as regulatory systems has been so far demonstrated only in eukaryotic cells, however the presence of G-4 sequences has been recently pinpointed also in prokaryotic cells [23][24][25]. Similarly, it is likely that other organisms, such as viruses, have evolved analogous regulatory mechanisms. Nevertheless, research in this area has been so far very elusive.
The human immunodeficiency virus (HIV) is the etiological agent of the acquired-immuno-deficiency syndrome (AIDS). HIV establishes a persistent infection in human hosts, with the depletion of CD4+ lymphocytes, the major target cells of viral infection in vivo, eventually resulting in defective cellular immunity, and thus leading to full-blown AIDS [26]. HIV is a member of the genus Lentivirus, part of the family of Retroviridae. Two types of HIV have been characterized: HIV-1 and HIV-2. The strains of HIV-1 can be classified into four groups: the "major" group M, the "outlier" group O and two new groups, N and P. More than 90% of HIV-1 infections belong to HIV-1 group M [27]: within this group, at least nine genetically distinct subtypes or clades (A, B, C, D, F, G, H, J and K) of HIV-1 have been reported [28]. The viral genome consists of two copies of positive single-stranded RNA that codes for nine genes (gag, pol, env, tat, rev, vif, vpr, vpu, nef) [29].
To date, FDA-approved antiretroviral drugs belong to several classes, based on their mechanism of action: 1) nucleoside and 2) non-nucleoside RT inhibitors; 3) protease, 4) fusion and 5) integrase inhibitors and 6) CCR5 antagonists [27]. Combinations of the existing drugs are very effective in slowing down progression to AIDS; however, the high mutation rate of HIV gives rise to resistance which ultimately impairs antiretroviral therapy. Therefore, there is an urgent need for new anti-HIV drugs with an innovative mechanism of action, possibly against highly conserved viral sites.
One attractive target for the anti-HIV therapy is the Nef protein.
Nef is a small myristoylated protein expressed early in the HIV-1 life cycle: it is a fundamental factor for efficient viral replication and pathogenesis in vivo; it also facilitates virus replication and enhances viral infectivity in vitro [30][31][32][33]. An essential role for Nef in vivo has been demonstrated in a subset of long-term non-progressors, HIV-infected individuals that do not progress to AIDS. Viral isolates from some of these individuals exhibit either a deletion in the nef gene or defective nef alleles [34]. In addition, rhesus macaques infected with an engineered strain of SIV that lacked the functional Nef protein also did not attain high viral loads and did not progress to clinical disease [35]. Nef alters host cell processes by several mechanisms: for example, to promote escape from the immune system and infectivity, it downregulates CD4 and MHC I expression on the cell surface, to enhance viral replication and infectivity it activates CD4+ CTL and downregulates/interacts with several cellular factors [36]. The Nef coding region, a 621 bp-long sequence located at the 3'-end of the viral genome, partially overlaps with the 3'-long terminal region (LTR).
The presence of G-4 forming sequences in the nef gene has never been reported. At the viral DNA level, G-4 forming sequences have only been described in a single-stranded portion (central DNA flap) of the reverse-transcribed preintegration HIV-1 genome, which specifically interacts with the viral nucleocapsid protein and protects the pre-integrated genome from nuclease degradation [37]. In the HIV-1 RNA genome, a G-4 structure has been proposed to promote dimerization of the two viral genome copies [38].
We present here the first comprehensive analysis of putative G-4 forming sequences in the HIV-1 nef coding region. We showed that three contiguous putative G-4 regions are present and that at least two are extremely conserved among most circulating HIV-1 strains. We provide evidence of their G-4 folding and stabilization in the presence of cations and G-4 binding compounds by spectroscopic and electrophoretic techniques. Finally, we demonstrate that G-4 ligands impaired Nef expression and significantly suppressed Nef-dependent enhancement of HIV-1 infectivity.

G-4 analysis of the HIV-1 genome
The HIV-1 Nef coding region (strain HXB2/LAI, NC_001802) was analyzed by QGRS Mapper (http:// bioinformatics.ramapo.edu/QGRS/index.php) for prediction of G-4 forming sequences in both coding and non-coding strands [39]. The putative G-4s were identified by the motif G x N y1 G x N y2 G x N y3 G x , where x was the number of guanine (G) tetrads and y n the length of loops connecting the G tetrads. The following restrictions have been applied: i) the number of tetrads had to be ≥ 2; ii) maximum length of QGRS was set to 30 bases; iii) at most one of the loops was allowed to be of zero length; iv) loop size from 0 to 15. The found QGRS were ranked based on the G-score, which is the likelihood to form a stable G-4, according to the following principles: a) shorter loops are more common than longer loops; b) G-4s tend to have loops roughly equal in size; c) the greater the number of G tetrads, the more stable the G-4.
Three G-4 putative sequences (Nef 8528 and Nef 8624 in coding strand, Nef 8547 in non-coding strand) were evaluated for their consensus sequence by aligning 3224 Nef sequences (HIV-1, M Group) from the HIV database (http:// www.hiv.lanl.gov/content/sequence/NEWALIGN/align.html) using Jalview (http://www.jalview.org/). Aligned sequences were searched for patterns expressed as Perl compatible regular expressions using GNU grep command line tool.

Circular dichroism and UV spectroscopy
DNA oligonucleotides used for spectral analysis were Nef8528 5'-GAGGAGGAGGTGGGT-3', Nef8547 5'-GGTCTTAAAGGTACCTGAGGTCTGACTGG-3', Nef8624 5'-GGGGGGACTGGAAGGG-3 ' and their complementary sequences, Nef8528-Compl ACCCACCTCCTCCTC, Nef8547-Compl CCAGTCAGACCTCAGGTACCTTTAAGACC, Nef8624-Compl CCCTTCCAGTCCCCCC. All oligonucleotides were diluted from stock to the final concentration (4 µM) in lithium cacodylate buffer (10 mM, pH 7.4) and, where appropriate, KCl 100 mM. All samples were annealed by heating at 95 °C for 5 min, gradually cooled to room temperature and measured after 24 h. Compounds at 16 µM final concentration were added after DNA annealing. CD spectra were recorded on a Jasco-810 spectropolarimeter (Jasco, Easton, MD, USA) equipped with a Peltier temperature controller using a quartz cell of 5-mm optical path length and an instrument scanning speed of 100 nm/min with a response time of 4s over a wavelength range of 230-600 nm. The reported spectrum of each sample represents the average of 2 scans at 20°C and is baseline-corrected for signal contributions due to the buffer. Observed ellipticities were converted to mean residue ellipticity (θ) = deg × cm 2 × dmol −1 (mol. ellip.). For the determination of T m , spectra were recorded over a temperature range of 20-95°C , with temperature increase of 5 °C/min. T m values were calculated as the first derivative of the melting profiles. For UV thermal unfolding, DNA oligonucleotides were diluted to final concentration of 4 µM or 40 µM. UV spectra were recorded on Lamba25 UV/Vis spectrometer (PerkinElmer) equipped with a Peltier temperature controller using a quartz cell of 10-mm optical path length and measuring absorbance at 295 nm. UV spectra were recorded over a temperature range of 20-95°C, with temperature increase of 1°C/min. The autozero function was applied on the corresponding buffer sample at 20°C.
DNA templates were diluted from stock to the final concentration (50 µM) in lithium cacodylate buffer (10 mM, pH 7.4) with 100 mM KCl and then let fold by heating at 95 °C for 3 min, gradually cooled to room temperature, and incubated at 4°C overnight. DNA templates were further diluted to a concentration of 1 µM and mixed with DNA primer (200 nM), 1X PCR reaction buffer (Applied Biosystems, Carlsbad, California, USA), and 0.1 mM dNTPs. Where appropriate, TMPyP4 was added. AmpliTaq Gold DNA polymerase (1U/ reaction, Applied Biosystem, Carlsbad, California, USA) was then added. Samples were subjected to 30 cycles: 95°C 30 sec, 60°C 30 sec, 72°C 30 sec. Reactions were stopped by ethanol precipitation. Marker lanes were generated on the labelled double stranded PCR product using the Maxam & Gilbert protocol. Briefly, ethanol precipitated PCR products were treated with formic acid (12 µl) for 5 min at 20°C. Reactions were stopped by ethanol precipitation. Samples were treated with piperidine 1 M for 30 min at 90°C; reactions were stopped on ice for 5 min. Samples were concentrated in SpeedVac. Markers corresponded to the C-rich complementary strand. The primer extension products and markers were separated on a 12% polyacrylamide denaturing gel and visualized by phosphorimaging (Typhoon FLA 9000, GE Healthcare, Milan, Italy).

Flow cytometry analysis
For FACS analysis, 8 × 10 4 of HEK 293T cells were seeded in a 12-well plate in 1 ml of DMEM/10% FBS medium and incubated for 24 hours. Cells were next transfected with pNef-HA EGFP-N1 by TransIT-293 Transfection Reagent (Mirus, Madison, WI, USA). After 4 hours, cells were treated with TMPyP4 or TMPyP2 (10 µM) and incubated overnight. After trypsinization, cells were washed with PBS and resuspended in 500 µl of PBS. To evaluate GFP expression, a total of 30000 events were acquired for each sample with an LRS 2 instrument using FACS DIVA Software (BD Bioscience, San Jose, CA, USA) and analyzed with FlowJo (Tree Star, OR, USA).

Cytotoxicity assays
Evaluation of the cytotoxicity of the compounds in HEK 293T cells was performed using a 3-(4,5-dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide (MTT) assay. HEK 293T cells were seeded in 96-well plate (1,5 x 10 4 /well) and grown overnight. Test compounds were added to one series of triplicate wells and incubated at 37°C. After 48 h, MTT (Sigma-Aldrich, Milan, Italy) solution (5mg/ml in PBS) was added and incubated at 37°C. After 4h, a solubilizing solution (SDS 10%, HCl 10mM) was added to dissolve the formazan crystals and incubated overnight at 37°C. The blue-purple formazan corresponding to metabolically active cells was then measured spectrophotometrically. The absorbances at 620nm were read in a computer-controlled spectrophotometer with a 96-well plate reader (Sunrise, Tecan) using the program Magellan 4.0. The percentage of cell viability was calculated by using the median absorbance value of three wells compared to the control's wells. CC 50 was the concentration of the test compounds that caused death of 50% cell population.
TZM-bl cells [40,41] were seeded in 96-well plates (2 x 10 4 ) and grown overnight to permit adherence. Cells were then incubated with compounds in DMSO carrier solvent and incubated at 37°C. After 48 h, cytotoxicity was assessed using the Cell Titer Blue reagent (Promega) and the manufacturer's protocol.

Antiviral assays
Viral stocks were prepared by transfection of HEK 293T cells (ATCC) with wild-type and Nef-defective (ΔNef) proviral genomes (NL4-3 strain) and amplified in the T cell line, MT2 (NIH AIDS Research and Reference Reagent Program) as previously described [42][43][44]. HIV-1 infectivity was measured using the TZM-bl reporter cell line [40,41] (NIH AIDS Research and Reference Reagent Program). TZM-bl cells support HIV-1 replication in a Nef-dependent manner and contain a luciferase reporter under the control of the HIV-1 promoter [41,45]. Cells were seeded in 96-well plates (2 x 10 4 ) and grown overnight to permit adherence prior to treatment and viral infection. TMPyP4 was solubilized in DMSO and preincubated separately with both the cell culture medium (100 µL) and wild-type HIV-1 or ΔNef HIV-1 (100 µL) for 4 h prior to infection in a combined final volume of 200 µL. After 48 h at 37°C, the cells were washed with PBS and lysed in luciferase lysis buffer (Promega) by rocking for 15 min. Lysates (40 µL) were transferred to white 96-well plates, and 50 µL luciferase reagent (Promega) was injected into each well. Readings were recorded with a delay time of 2 s and an integration period of 10 s.

Computational analysis of the HIV-1 Nef coding region for the presence of putative G-4 forming regions
The HIV-1 Nef coding region (strain HXB2/LAI, NC_001802) was analysed with QGRS Mapper which is an online algorithms-based software program for recognition and mapping of putative Quadruplex forming G-Rich Sequences (QGRS) [39]. The found QGRS were ranked based on the Gscore, which is the likelihood to form a stable G-4.
Since the viral genome is retrotranscribed into doublestranded DNA and inserted into the human genome, analysis for QGRS was performed also on the reverse strand of the nef coding gene.
Four putative QGRS were located in the nef gene (Table 1): two on the forward (positions 8528 and 8624, where +1 is the first base of the HIV-1 genome as reported in GenBank, NC_001802) and two on the reverse strand (positions 8547 and 8727). In particular, the three that displayed the highest Gscore (≥ 20) were also adjacent to one another: the first on the forward strand (Nef8528) and the second on the reverse strand (Nef8547) were separated by just 4 nucleotides (nts); the second from the third sequence on the forward strand (Nef8624) by 48 nts (Figure 1A-B). If these three sequences were able to fold in G-4, this region of the HIV-1 proviral genome could constitute an important cluster of non-canonical DNA structures with possible effects on polymerase processing (and therefore impact on replication and transcription events) ( Figure 1C-D). It is interesting to note that all selected G-4 putative sequences code for amino acids of the Nef core, which is the most conserved region of the protein and is essential for interaction with cellular proteins to mediate key viral functions [46][47][48] (Figure 1D).
To establish the importance of the identified sequences from a virus standpoint, the degree of conservation in terms of sequence and G-4 formation among HIV-1 strains was assessed.
Initially, the presence of the exact sequences identified in the HIV-1 HXB2/LAI strain was analysed in 3224 nef sequences of the HIV-1 M group reported in the HIV database (http:// www.hiv.lanl.gov/content/sequence/NEWALIGN/align.html). Among these, 1538 sequences belonged to clade B, 612 to clade C, 486 to clade A, and 588 to other clades. As shown in Table 2, Nef8528, Nef8547 and Nef8624 were fairly well conserved in the M group, especially in the B subtype, where conservation was higher with respect to the other clades of the M group. Next, the possibility of G-4 formation was statistically analysed by maintaining the number and size of G repeats, while varying loop regions. Two cases were considered: in instance i) loops could diverge in base composition while maintaining a constant length; in instance ii) both loop composition and length were allowed to vary. As shown in Table 2, Nef8547 and Nef8624 reached a degree of conservation higher than 95% in both cases across all considered HIV subtypes. Nef8528 was conserved to a significant extent in clades A and B (up to 66.3%); its presence was negligible only in clade C. Consensus sequences and base conservation in each position are reported in Table 3.
Overall these data indicate that the G-4 pattern, therefore the possibility of G-4 folding, in the selected G-rich sequences in the Nef coding region is extremely conserved among circulating HIV-1 strains, at least for Nef8547 and Nef8624.
G-rich sequences in the nef region are able to fold in G-4 The actual folding of the selected nef sequences in a G-4 conformation was initially assessed by circular dichroism (CD) spectroscopy. In the case of Nef8528, which presents five GG-repeats, we used the minimal sequence that could fold into G-4 (see Materials and Methods). Since monovalent cations, in particular K + , are known to stabilize G-4s [49], the three nef sequences were incubated in the presence of increasing concentrations of K + or Na + . Both cations increased the CD signal, with K + exhibiting a remarkably higher effect than Na + . CD analysis is typically used to obtain information on the G-4 topology: in particular, a parallel-type G-4 conformation is manifested by a maximum at 260 nm and a minimum at 240 nm, while an antiparallel topology exhibits two maxima at 290 nm and 240 nm and a minimum at 260 nm [50]. Nef8528 and Nef8624 displayed a clear parallel-type conformation upon addition of K + and Na + , with a maximum at 260 nm and a minimum at 240 nm ( Figure 2). Conversely, Nef8547 presented a maximum at around 275 nm and a negative peak at 240 nm, with low K + /Na + dependence. Therefore, the 7-nt-long loop Nef8547 apparently did not naturally fold into a canonical G-4 conformation: to note, however, that similar CD spectra have  [73]. The protein moiety coded by the G-4 rich nucleotide region is highlighted by the rectangular yellow shape, indicating involvement of the conserved N-terminal Nef core region. Note that the first three nts of the Nef8528 sequence exactly code for the first amino acid of the protein core region. doi: 10.1371/journal.pone.0073121.g001 been reported for G-4 forming oligonucleotides with long loops (i.e. at least two loops ≥ 5 nts) [51].
Stability of the G-4 structures was next evaluated by CD thermal unfolding and melting temperatures (T m ) were calculated as the first derivative of the melting profiles ( Figure 3 and Table 4). In all cases the CD signal decreased with increasing temperature. For Nef8547 and Nef8624 a single transition between 25°C and 95°C was appreciable leading to discrete T m values (35°C and 59°C, respectively, Table 4). For Nef8528 two transitions were present: a first structural variation at 39°C and a second at T m > 50°C, where a clear inflection point was not observed. This behaviour could also be evinced by spectra overlapping, where two isosbestic points (asterisks in Figure 3) were detected, indicating the presence of at least three spectroscopically distinct species: the initial G-4 structure, a second folded form, likely a more flexible G-4 conformation, and the unfolded random coiled structure. In addition, T m values measured by UV thermal unfolding at oligonucleotide concentration of 40 µM were very similar to those obtained at 4 µM (Table 4), indicating a prevalent intramolecular G-4 folding.
Several compounds have now been reported to stabilize the G-4 structure. Three G-4 ligands with different central cores (i.e. porphyrin for TMPyP4, acridine for BRACO-19, and perylene for PIPER) were incubated in the presence of each of the three nef G-rich oligonucleotides to check for the compound ability to induce/stabilize the G-4 topology. CD thermal unfolding analysis was employed to check stabilization of the G-4 conformation imposed by the G-4 ligands. CD spectra of all three oligonucleotides in the presence of G-4 ligands were characteristic of G-4 conformations. In particular, Nef8528 and Nef8624 maintained the initial parallel-like topology with TMPyP4 and PIPER; in the presence of BRACO-19 an additional positive peak appeared at 290 nm, which is characteristic of a G-quadruplex topology and might depict a shifting towards a hybrid topology. In the case of Nef8547, PIPER induced an antiparallel-like spectrum, while TMPyP4 and BRACO-19 stabilized hybrid-type G-4 structures ( Figure 4A), indicating that G-4 ligands are able to drive Nef8547 folding into a G-4 conformation. To note that in most cases an induced CD spectrum was observed in the UV/Vis absorption region of the ligands, further confirming oligonucleotide/compound interaction. G-4 ligands highly stabilized the G-4 conformations of Nef8528 and Nef8547 (Table 4). Nef8624 in general was less efficiently stabilized, likely due to the higher innate stability of this latter oligonucleotide. In cases where two transitions were observed, T m values for each transition were reported (Table 4). These data altogether confirm the ability of the nef sequences to fold Table 2. Statistical analysis of the conservation grade of the G-4 nef sequences or their G-4 patterns.  Table 3. Consensus sequences and percentages of base conservation at each position.   in G-4: G-4 ligands can bind, induce and stabilize their G-4 conformations.

G-4 name G-4 sequence or pattern Conservation grade (%)
Since the G-rich nef sequences in the proviral genome are normally embedded in a DNA double-helix, TMPyP4 was incubated with each nef oligonucleotide in the presence of its complementary counterpart in order to assess the ability of a G-4 ligand to promote G-4 folding from a double-stranded (ds) DNA. CD spectra of the ds oligonucleotides in the absence or in the presence of K + were very similar and characteristic of a B-DNA [52]. However, when TMPyP4 was added, the CD spectra clearly shifted, presenting two maxima at 290 and 260 nm, which are indicative of a G-4 conformation ( Figure 4B). Since TMPyP4 absorbance below 300 nm is extremely low [53], the observed molar ellipticity variation must be due to changes in the absorbance of the nucleic acid. These data show that nef sequences are normally present in a B-DNA conformation within the double-helix, and that a G-4 binder is able to induce their folding in G-4.
G-4 folding in the selected sequences was additionally proved by the Taq polymerase stop assay. The three G-4 nef oligonucleotides were designed in order to contain additional flanking bases at both the 5'-and 3'-ends (see Materials and Methods Section); in particular, an additional sequence at their 3'-end was used as primer annealing region. A 4-T linker region was added to separate the 3'-end of the primer and the first G of the G-4 portion. An additional oligonucleotide lacking the possibility to fold in G-4 was designed and used as negative control. Primer annealing and G-4 folding were obtained by incubating the template G-4 forming oligonucleotides and the primer in K + buffer at 95°C and slowly cooling down to room temperature. Taq polymerase was incubated with the different template/primer combinations in the presence of increasing amount of TMPyP4. As shown in Figure  5A, in the presence of both Nef8528 and Nef8624 templates, increasing drug concentrations induced arrest of the DNA polymerase processing at the T linker region, just before the G-4 folded region (lanes 10-12 and 17-18, Figure 5A), while no effect was detected in the negative control (lanes 2-6, Figure  5A). In the case of the Nef8624 template, at low amounts of TMPyP4, a polymerase pausing site was observed corresponding to the second G-tract of the Nef8624 oligonucleotide (* symbol, lanes 15-16, Figure 5A). In addition the polymerase was partially inhibited also in the absence of TMPyP4 (¤ symbol, lane 14, Figure 5A). We ascribed this behaviour to the fact that the G-4 conformation of Nef8624 was inherently very stable in the presence of K + (T m 59.8°C, see Table 3) and thus could affect polymerase activity even without G-4 ligands, similarly to other reported G-4 structures [54]. In the case of the Nef8547 template, several stop sites were observed in the G-4 forming template, while the polymerase was not inhibited in the control (compare lanes 2-5 and 7-10, Figure 5B). In particular, stop sites were clustered at G bases at low TMPyP4 concentrations (* symbols, lane 8, Figure 5B) and at the T-linker at higher G-ligand amounts (lane 10, Figure  5B).
Overall these data indicate that a G-4 binder can induce and stabilize the G-4 conformations at the nef DNA level.

Stabilization of nef G-4 sequences by G-4 ligands affects gene expression
To check the effect of G-4 stabilizing ligands on gene expression, a GFP-based reporter gene assay was set up. The nef gene was cloned upstream of a GFP coding region in a plasmid optimized for protein expression in mammalian cells. In principle, G-4 folding in the nef sequence should impair the polymerase activity on the DNA template, therefore reducing expression of the fused Nef-GFP protein. Transfected cells were treated with TMPyP4 (10 µM) or a control compound, TMPyP2, a structural analogue of TMPyP4 which is not able to bind G-4s [55]; both compounds did not show toxicity on HEK293T cells up to 100 µM (data not shown). GFP fluorescence was quantified by flow cytometry-based analysis.
As shown in Figure 6A, mean of fluorescence consistently decreased in the presence of TMPyP4, while slightly increased upon treatment with the control TMPyP2 indicating a selective impairment of gene expression mediated by the interaction of the G-4 ligand with the nef sequence.

Effect of nef G-4 binding compounds on HIV-1 infectivity
Having observed an impairment of Nef protein expression levels, the effect of G-4 stabilizing ligands was next assessed in HIV-1 infected cells. As Nef has been previously shown to enhance HIV-1 infectivity [31,33,56,57], we evaluated the impact of TMPyP4 on infectivity using the TZM-bl reporter cell line [40,41]. TZM-bl cells support HIV-1 replication in a Nefdependent manner and contain a luciferase reporter driven by the HIV-1 LTR in response to infection with HIV-1 [41,42]. As shown in Figure 6B, TMPyP4 impaired the enhancement of Nef-mediated viral infectivity in a dose-dependent manner, while the negative control (TMPyP2) had no effect ( Figure 6C). In addition, a Nef-deleted virus (ΔNef) showed no significant effects by treatment with either TMPyP4 or TMPyP2 at the highest concentrations ( Figure 6B-C). Cytoxicity assays further confirmed the observed impairment of viral infectivity as minimal toxicity was observed over the range of concentrations tested (0.1-6 µM) ( Figure 6D). Taken together, these results provide support for a G-4 mediated, Nef-directed anti-retroviral mechanism of action.

Discussion
To our knowledge, this is the first report indicating a) the presence of conserved G-rich sequences that can fold in the G-4 conformation within the HIV-1 proviral genome, b) the important role played by G-4 stabilization in both polymerase and viral infectivity inhibition.
The Nef protein has a fundamental function in vivo and its lack of activity (observed in Nef mutant HIV strains) prevents progression to the clinical development of AIDS [34]; therefore its depletion could have critical antiviral effects.
Here we demonstrated that in the HIV-1 Nef coding region three G-rich/G-4 prone tracts were closely clustered and located on both the leading and lagging strands where G-4mediated induction of genetic instability and arrest of RNAdependent polymerase have been described [11,13]. In addition, in a supercoiled environment, i-motif conformations may arise in the C-rich complementary sequences independently of G-4 formation and pH [1,58]. We thus foresaw the possibility of creating a structured environment upon induction of non-canonical nucleic acid conformations, with the possibility of blocking enzymes involved in Nef protein expression. The identified G-4 forming sequences comprised at least four tracts of two consecutive Gs, with the possibility to form G-4s with two stacked tetrads. Although in principle two G-tetrads would confer less stability than more extended Gtetrads, the existence and biological role of G-4s with two stacked tetrads has been reported in several cases [59][60][61][62][63]. Interestingly, in certain instances, two-tetrads G-4s showed higher stability than three-tetrads G-4s [64][65][66][67].
Upon analysis of all HIV-1 strains available in databases, we demonstrated that the G-rich region in the nef coding gene, and in particular the G pattern necessary for G-4 folding, is conserved within the HIV-1 M group and in its subtypes largely distributed worldwide. This information suggests that G-4 structures may form at this level and may also be involved in important viral functions. If this was the case, we argued that stabilization induced by G-4 binders would exert an effect observable at the viral level. All three putative G-4 nef sequences were able to fold in a G-4 conformation in the presence of G-4 binders. In most cases G-4 ligands varied the initial G-4 topology and augmented the T m during temperature unfolding experiments, indicating an effective interaction between them and the G-4 nef oligonucleotides. Importantly, all G-4 nef sequences were able to form G-4 structures even in the presence of their complementary oligonucleotides. In the case of Nef8528 and Nef8547, they both existed in a fully double-stranded state and G-4 folding was induced upon incubation with the G-4 ligand. Conversely, Nef8624 was present in a partial G-4 topology even in the absence of the G-4 binder, as attested by both CD analysis and partial inhibition of the DNA polymerase activity prior to G-4 compound addition. This is in accordance with the higher T m measured for the G-4 structure of Nef8624.
Most importantly, the ability to induce G-4 folding with consequent inhibition of viral infectivity was shown in HIV-1 infected cells. Within the cellular context of viral infection, the G-rich sequences are embedded in a double stranded state and yet small amounts of G-4 ligand were sufficient to impair the Nef-mediated enhancement of HIV-1 infectivity. The significant effect observed on the wt HIV-1 virus in Nef sensitive cells [45] and the negligible effect on the ΔNef virus further indicate an efficient inhibition of Nef protein expression.
Recent research on G-4 has revealed strong stabilization of tetraplex structures at the RNA level [68,69]. It is thus possible that, besides the DNA, the nef mRNA is also stabilized by G-4 compounds, therefore enhancing inhibition of Nef expression in infected cells. The fact that Nef8624, the most stable G-4 sequence, was able to pause the polymerase independently of the presence of G-4 ligands, suggests that it may naturally function as a molecular switch to assist protein expression or other unanticipated viral functions.
Moreover, G-4 stabilization within the nef coding region of the viral genome may impair not only Nef expression through inhibition of transcription that directly generates mRNAs, but also overall transcription for production of new copies of the RNA genome to be assembled in the viral progeny.
An additional point to be considered is that available G-4 ligands were developed principally against the telomeric G-4, and were shown to kill tumour cells through telomere dysfunction and telomerase inhibition [70,71]. G-4 stabilizing compounds interact with the G-4 structure both at the aromatic G-quartet level, which is common to all G-4 conformations, and at the loops, which conversely are unique to each G-4 folding [72]. Here we have shown that the G-4 ligand TMPyP4 is effective towards the virus with minimal effect on cell viability. Therefore a therapeutic window can be envisaged that would consent the employment of known G-4 binders as anti-HIV compounds. In addition, rational design of new G-4 ligands directed towards the tested G-4 nef sequences would likely augment the selective effects towards the virus. Detailed structural characterization of the G-4 nef sequences will be the first step to allow molecular modelling of new G-4 binding leads.
The results presented here may open a new avenue in the development of antiviral compounds with an unprecedented mechanism of action.  Figure 6. Effects of the stabilization of the nef G-4s by TMPyP4 on gene expression and viral infectivity in Nef sensitive cells. A) Effect of TMPyP4 and TMPyP2 on Nef-GFP expression measured by flow-cytometry. HEK 293T cells were transfected with a Nef-GFP encoding plasmid and treated with TMPyP4 or TMPyP2 (10 µM) for 24 h. Results are shown as percent mean of fluorescence relative to the control cells incubated ± SD (n = 4). Statistical difference was observed for TMPyP4 (p<0.05), but not for TMPyP2. B) and C) TZM-bl cells were infected with wild-type (black bars) and ΔNef (grey bars) HIV NL4-3 in the presence of either the G-4 ligand, TMPyP4 (B), or the negative control compound, TMPyP2 (C). After 48 h, infectivity was assessed as relative luciferase activity in infected cells. Results are shown as percent infectivity relative to the control cells incubated with carrier solvent (DMSO) ± SEM (n = 3). In B), no statistical difference was observed across ΔNef infected cells, even at the highest concentration (p > 0.15). In C), no statistical difference was observed for the wild-type virus at 3 µM and 6 µM (p >0.345 and >0.325, respectively) relative to the untreated control. The negative control compound, TMPyP2, further had no impact on the ΔNef virus at any concentration tested (p>0.29 at 6 µM). D) TZM-bl cell viability in the presence of compounds was assessed via the Cell-Titer Blue assay (Promega). TZM-bl cells were incubated with the indicated concentrations of compounds for 48 h and cell viability was assessed via the Cell-Titer Blue assay relative to control cells incubated with carrier solvent. Assays were done in triplicate. doi: 10.1371/journal.pone.0073121.g006