Characterization of the Inhibition Mechanism of HIV-1 Nucleocapsid Protein Chaperone Activities by Methylated Oligoribonucleotides

Abstract

Since currently available therapies against HIV/AIDS still show important drawbacks, the development of novel anti-HIV treatments is a key issue. We recently characterized methylated oligoribonucleotides (mONs) that extensively inhibit HIV-1 replication in primary T cells at nanomolar concentrations. The mONs were shown to target both HIV-1 reverse transcriptase (RT) and the nucleocapsid protein (NC), which is an essential partner of RT during viral DNA synthesis. To further understand the mechanism of such mONs, we studied by isothermal titration calorimetry and fluorescence-based techniques their NC binding properties and ability to inhibit the nucleic acid chaperone properties of NC. Notably, we investigated the ability of mONs to inhibit the NC-induced destabilization of the HIV-1 cTAR (complementary DNA sequence to TAR [transactivation response element]) stem-loop and the NC-promoted cTAR annealing to its complementary sequence, required at the early stage of HIV-1 viral DNA synthesis. Moreover, we compared the activity of the mONs to that of a number of modified and nonmodified oligonucleotides. Results show that the mONs inhibit NC by a competitive mechanism whereby the mONs tightly bind the NC peptide, mainly through nonelectrostatic interactions with the hydrophobic platform at the top of the NC zinc fingers. Taken together, these results favor the notion that the mONs impair the process of the RT-directed viral DNA synthesis by sequestering NC molecules, thus preventing the chaperoning of viral DNA synthesis by NC. These findings contribute to the understanding of the molecular basis for NC inhibition by mONs, which could be used for the rational design of antiretroviral compounds targeting HIV-1 NC protein.

INTRODUCTION

Due to the emergence of strains resistant to the currently available drugs targeting the HIV-1 enzymes reverse transcriptase (RT), protease and integrase (Stanford HIV Drug Resistance Database, http://hivdb.stanford.edu), the development of novel anti-HIV agents and virucides is a major challenge. A promising target for anti-HIV agents is the nucleocapsid protein (NC), since it is highly conserved and essential during the early and late phases of HIV-1 replication (herein, the acronym “NC” without indication of the residues refers to the nucleocapsid protein in general, while in the description of particular experiments, the form of NC is specified). First, as a domain of the Gag structural polyprotein precursor, NC selects the genomic RNA and promotes its dimerization and packaging into newly formed viral particles (12, 37). Second, NC acts as a nucleic acid chaperone during reverse transcription by promoting the annealing of the cellular primer tRNA to the primer binding site (PBS) and the two obligatory DNA strand transfers necessary for the synthesis of a complete double-stranded viral DNA by RT (35, 41; reviewed in reference 13). In the first cDNA strand transfer, NC promotes the annealing of the cTAR (complementary DNA sequence to TAR [transactivation response element]) stem-loop from the strong-stop cDNA to the TAR sequence located at the 3′ terminus of the genomic RNA. This promotion results from a mechanistic switch from the poorly efficient loop-loop pathway that predominates in the absence of NC to a highly efficient zipping pathway through the stem termini (42). NC also chaperones the second strand transfer by promoting the annealing of the (−) and (+) DNA copies at the level of the PBS. This activation results from an NC-directed switch of (+)/(−)PBS annealing toward a loop-loop kissing pathway, as a consequence of the ability of NC to “freeze” PBS conformations competent for annealing via the loops (25). This NC activity is strictly dependent on the integrity of the hydrophobic platform at the top of the zinc fingers and is thought to play an important role in the specificity and fidelity of the second strand transfer. NC-promoted nucleic acid chaperoning involves several steps: (i) NC binding to its target sequences, (ii) destabilization of secondary and tertiary structures of the nucleic acids, and (iii) promotion of the annealing of the destabilized complementary sequences (for reviews, see references 13, 24, and 36).

A range of NC-targeting molecules with various mechanisms of action has been developed (see reviews in references 16 and 26). A strategy was to design molecules which bind NC with high affinity, such as GT- or GU-rich oligonucleotides (ONs)(21, 22, 43). Along this line, we recently designed small methylated single-stranded oligoribonucleotides (mONs) rich in G's and U's, which were found to inhibit the NC chaperone activity in vitro (28). Interestingly, such mONs impeded HIV-1 replication in TCD4⁺ cells at low nanomolar concentrations by severely impairing viral cDNA synthesis. After serial passaging of HIV-1 in the presence of such mONs, resistant viruses that contained mutations in NC and RT emerged (28), suggesting that these two viral proteins are the primary targets of mONs in infected cells.

In an attempt to understand the molecular determinants and the mechanism of the antiviral activity of such mONs, we studied the in vitro activity of these mONs and a series of related molecules, such as DNA and nonmethylated RNA analogues and random mONs lacking GU motifs. Their binding to NC and their effects on the molecular reactions underlying the first strand transfer reaction were quantitatively monitored by fluorescence spectroscopy and isothermal titration calorimetry (ITC). We found that the GU- and GT-rich oligonucleotides bound two NC molecules with an apparent binding affinity of 10⁶ to 10⁷ M⁻¹. The binding is enthalpy driven and mainly relies on nonelectrostatic interactions with the hydrophobic platform at the top of the NC zinc fingers. Thus, the antiviral activity of mONs is thought to rely on their strong affinity for NC and their ability to reduce the number of NC molecules bound to its viral nucleic acid targets, preventing NC molecules from chaperoning reverse transcription.

MATERIALS AND METHODS

Since the full-length nucleocapsid protein NC(1-55) has a strong tendency to aggregate nucleic acids (40), the truncated form of this protein lacking the N-terminal domain [NC(11-55)] (Fig. 1A) and possessing much lower nucleic acid-aggregating properties was used here. In fact, NC(11-55) was shown to be an appropriate model to study the chaperone activities of the full-length NC protein (5, 8). NC(11-55) and NC(1-55) were synthesized by solid-phase chemistry, as previously described (15). The peptides were purified on reverse-phase high-pressure liquid chromatography (HPLC) columns, lyophilized, and stored in their zinc-bound form. The peptide concentrations were determined from their absorbance at 280 nm.

Fig 1.

Structure of the molecules used in this study. (A) HIV-1 nucleocapsid peptide NC(11-55). (B) Eosin label attached to mON-11. (C) cTAR and dTAR secondary structures. (D) Dyes attached to cTAR for monitoring the stem-loop destabilization and annealing reactions.

Labeled and unlabeled DNA sequences were synthesized at a 0.2-μmol scale by IBA GmbH Nucleic Acids Product Supply (Göttingen, Germany). The following single-stranded ONs were used: mON-9, all 2′-methylated RNA nonanucleotide 2′O-me-(GGUUUUUGU); ON-GT, nonmodified DNA undecanucleotide GGTTTTTGTGT; ON-GT-Acr, DNA undecanucleotide GGTTTTTGTGT with 9-aminoacridine conjugated to the 3′-phosphate through a 6-carbon spacer arm; ON-GU, nonmodified RNA undecanucleotide GGUUUUUGUGU; mON-11, all 2′-methylated RNA undecanucleotide 2′O-me(GGUUUUUGUGU); mON-11-Eos, mON-11 with eosin conjugated to the 3′-phosphate (Fig. 1B) (33); and mON-Rnd1 and mON-Rnd2, all 2′-methylated RNA random undecanucleotides 2′O-me(CUUAAGACCAU) and 2′O-me(UCUCACAACUC). To monitor the NC chaperone activities, we used cTAR (Fig. 1C) labeled with a pair of chromophores (Fig. 1D), namely, 5- and 6-carboxytetramethylrhodamine (TMR) or 6-carboxyrhodamine (Rh6G) and 5- and 6-carboxyfluorescein (Fl) or 4-(4′-dimethylaminophenylazo)benzoic acid (DABCYL). TMR and Rh6G were linked to the 5′ terminus via an amino linker with a 6-carbon spacer arm, while the 3′ terminus was labeled with Fl or DABCYL by using a special solid support with the dye already attached. Oligonucleotides were purified by polyacrylamide gel electrophoresis (PAGE), and their purity was checked by electrospray ionization-mass spectrometry. Concentrations of oligonucleotides and peptides were measured by UV absorption on a Cary 400 UV-visible spectrophotometer. The following extinction coefficients (260 nm) were used: 523,440 M⁻¹ × cm⁻¹ for cTAR and the DNA equivalent of the TAR sequence (dTAR) and 96,210 M⁻¹ × cm⁻¹ for ON-GT. All experiments were performed at 20°C, except for the temperature series of ITC titrations.

Time-resolved fluorescence measurements of the Acr-labeled ON were performed with a time-correlated, single-photon counting technique using the stable excitation pulses provided by a pulse-picked frequency-tripled Ti-sapphire laser (Tsunami, Spectra Physics) pumped by a Millenia X laser (Spectra Physics). The excitation pulses were at 429 nm with a repetition rate of 4 MHz. The emission was collected at 540 nm through an 8-nm band-pass monochromator (Jobin-Yvon H10). The single-photon events were detected with a microchannel plate Hamamatsu R3809U photomultiplier coupled to a Philips 6954 pulse preamplifier and recorded on a multichannel analyzer (Ortec 7100) calibrated at 25.5 ps/channel. The instrumental response function was recorded with a polished aluminum reflector, and its full width at half-maximum was 40 ps. Fluorescence intensity decays obtained at the magic angle were deconvolved with the instrument response function and analyzed as a sum of exponentials: I(t) = Σα_i exp(−t/τ_i), where I(t) is the fluorescence intensity collected at the magic angle at time t and α_i is the amplitude of the fluorescence lifetime τ_i such that Σα_i = 1. Time-resolved fluorescence anisotropy decays were recorded using a polarizer at horizontal or vertical positions and analyzed by the following set of equations:

$$I_{||}(t)=I(t)[1+2r(t)]/3$$

$$I_{\perp }(t):I(t)[1-r(t)]/3$$ (1)

$$r(t)=\frac{I_{||}(t)-{GI}_{\perp }(t)}{I_{||}(t)+{2GI}_{\perp }(t)}=r_0\mathrm{\Sigma }{\beta }_i\exp (-t/{\phi }_i)$$

where r₀ is the initial anisotropy; β_i is the amplitude of the rotational correlation time φ_i such that Σβ_i = 1; I_|| and I_⊥ are the intensities collected at emission polarizations parallel and perpendicular, respectively, to the polarization axis of the excitation beam; and G is the geometry factor at the emission wavelength, determined in independent experiments. The r₀ value was determined for ON-GT-Acr in the presence of 77% glycerol (vol/vol) from the extrapolation of the anisotropy decay curves to zero time and was found to be from 0.31 to 0.32. The minimal theoretical values of the rotational correlation times were calculated from the molecular masses (M) of the molecules and their complexes, assuming a spherical shape, by the following equation:

$$\phi =\frac{\eta M(v+h)}{RT}$$ 2

where η is the viscosity (assumed to be 0.01 P), M is the molar mass (assumed to be 3,753 and 5,205 g/mol for ON-GT-Acr and NC, respectively), T is the temperature (maintained at 293 K), υ is the specific volume of the particle (assumed to be 0.83 ml/g [34]), h is the hydration degree (assumed to be 0.2 ml/g for proteins), and R is the molar gas constant (8.31 J mol⁻¹ × K⁻¹).

The fluorescence quantum yields of 9-aminoacridine and ON-GT-Acr were calculated by the following equation:

$${\mathrm{\varphi }}_x={\mathrm{\varphi }}_r\left(\frac{S_x}{S_r}\right)\left(\frac{A_r}{A_x}\right)$$ (3)

where S_x and S_r are the integrated surface areas; A_x and A_r are the absorbances of the sample and the reference at the excitation wavelength, respectively; and ϕ_r is the quantum yield of the reference. Quinine sulfate (ϕ = 0.546 in 0.01 M H₂SO₄) and 9-aminoacridine (ϕ = 0.875 in water) were used as references with known quantum yield. Steady-state fluorescence emission spectra were recorded on either a Fluorolog or a FluoroMax spectrofluorometer (Jobin Yvon) equipped with a thermostated cell compartment.

NC/oligonucleotide binding experiments were performed by monitoring the intrinsic fluorescence emission (at 350 nm with excitation at 295 nm) of the NC Trp37 residue in titration experiments where a fixed amount of NC was reacted with increasing ON concentrations in 25 mM Tris-HCl, pH 7.5, 0.2 mM MgCl₂ and various concentrations of NaCl (30 to 500 mM). The fluorescence intensity was corrected for dilution, buffer fluorescence, and screening effects due to the ON absorbance. The fluorescence data were fitted to equation 4 as follows to recover the equilibrium association constant K_a:

$$\begin{gathered} I=I_0-\frac{(I_0-I_t)}{L_t}\times \\ \frac{[1+(L_t+n{N}_t)K_a]-\sqrt{{[1+(L_t+n{N}_t)K_a]}^2-4L_{t}n{N}_t{K}_a^2}}{2K_a} \end{gathered}$$ (4)

where L_t and N_t designate the total concentration of peptide and ON, respectively; I_t represents the fluorescence at the plateau when all the peptide is bound, whereas I₀ and I correspond to the fluorescence intensities of the peptide in the absence and in the presence of a given concentration of ON, respectively; and n represents the number of NC binding sites per ON molecule, determined independently in anisotropy decay measurements. The parameters were recovered from nonlinear fits of equation 4 to the experimental data sets by using the Origin software (Microcal). Alternatively, the NC/ON binding experiments were monitored by isothermal titration calorimetry (ITC), using a VPITC microcalorimeter (Microcal) at temperatures ranging from 10 to 42°C, in 50 mM HEPES, pH 7.4, 100 mM NaCl. Typically, experiments were performed by injecting 6-μl aliquots of a concentrated solution of NC(11-55) (145 μM) to the ON solution (2.9 μM) contained in the 1.4-ml heat-sensitive reaction cell, with 4-min intervals. The heat flow (μcal × s⁻¹) resulting from the reaction between the two partners was continuously recorded. The quantity of heat for each peptide injection was integrated over time. At each temperature used, the experimentally determined quantities of heat were corrected for the heat of dilution measured by performing a control experiment in which NC(11-55) was titrated into the buffer alone. Instrument control, data acquisition, and analysis were done with the VPViewer and Microcal Origin software provided by the manufacturer.

NC-promoted destabilization of stem-loop sequences was quantified according to a protocol previously developed in our laboratory (6). Briefly, we took advantage of the distance-dependent quenching of the Rh6G fluorescence by DABCYL linked to the 5′ and 3′ termini of cTAR to estimate the degree of fraying of the terminal bases of the cTAR stem. The experiments were performed in 25 mM Tris-HCl buffer, pH 7.5, 0.2 mM MgCl₂ containing 30 or 100 mM NaCl. For the ON conjugated with eosin whose fluorescence parameters significantly overlap with those of Rh6G, a cTAR species doubly labeled with AlexaFluor 350 (Invitrogen) and DABCYL was used instead of Rh6G-cTAR-DABCYL.

The cTAR-dTAR annealing kinetics were monitored as described before (23, 24). NC(11-55) was mixed with 5 nM TMR-5′-cTAR-3′-Fl and 50 nM dTAR, at a ratio of 11 NC molecules per cTAR or dTAR molecule. Then, the selected ON was added at 440 nM to both mixtures, and the mixtures were incubated for 5 min. The cTAR/dTAR annealing reaction was triggered by mixing the cTAR/NC(11-55)/ON mixture with the dTAR/NC(11-55)/ON mixture in 25 mM Tris-HCl, 0.2 mM MgCl₂, 30 mM NaCl, pH 7.5 buffer. The kinetic parameters of the cTAR/dTAR annealing reaction were recovered by fitting the experimental progress curves, obtained by monitoring the Fl fluorescence (excitation at 480 nm, emission at 520 nm) to the following equation:

$$I(t)=(I_f-I_{0NC})\times (e^{-k_{\mathrm{o}\mathrm{b}\mathrm{s}}\times (t-t_0)})$$ (5)

where I_0NC, I_f, and I are the initial fluorescence of the doubly labeled cTAR in the presence of NC(11-55), the fluorescence intensity of the final cTAR/dTAR complex, and the fluorescence intensity at a given time t, respectively; k_obs is the effective kinetic rate constant; and t₀ is the dead time. To compare the efficiencies of the ONs to alter the NC(11-55)-induced promotion of cTAR/dTAR annealing, the data were expressed as k_obs/k_obs0, where k_obs and k_obs0 are the effective association rate constants in the presence and in the absence of inhibitor, respectively.

RESULTS

GU- and GT-rich ONs bind two molecules of NC with micromolar affinities.

As a first key step in further understanding the molecular mechanism of the antiviral activity of the GU-rich mONs, we characterized their binding parameters to NC, identified as a potential target of these mONs. The stoichiometry of binding (number of NC molecules bound to each ON) was determined by time-resolved fluorescence anisotropy, which is a reliable and nonperturbing method for the determination of the stoichiometry of molecular complexes in solution (34). The binding stoichiometry was determined using a DNA analogue conjugated with the 9-amino-acridine fluorescent dye (GGTTTTTGTGT-Acr) (Table 1). Anisotropy decay measurements provide rotational correlation time values which are proportional to the effective hydrodynamic radius and thus to the molecular weight (MW) of labeled molecules, both in their free form and in their complexes. The experimental decay curves were fitted satisfactorily to equation 1, assuming two rotational correlation times (φ) (χ² ≈ 1.5). The longest rotational correlation time was attributed to the tumbling of the free molecule or its complex. For ON-GT-Acr saturated with NC(11-55), the φ₂ value was 7.8 ns, well above the minimal theoretical rotational correlation time, φ_th= 6.0 ns (equation 2), expected for a spherical complex of two NC(11-55) molecules bound to one ON and below the corresponding value calculated for a 3:1 complex, φ_th = 8.2 ns. Therefore, the obtained value suggested that two NC(11-55) molecules are bound per ON and that the complex exhibits a nonspherical shape, which leads to a slow correlation time compared to a sphere of the same mass.

Table 1.

Time-resolved fluorescence and anisotropy decay parameters of ON-GT-Acr in the absence and the presence of NC(11-55)^a

ON-GT-Acr	τ1 (ns)	α1	τ2 (ns)	α2	τ3 (ns)	α3	Mean τ	φ1 (ns)	β1	φ2 (ns)	β2
−	0.20	0.67	2.45	0.21	8.6	0.12	1.77	0.23	0.17	1.7	0.83
+NC(11-55)	0.77	0.37	5.1	0.43	14.3	0.20	5.41	0.21	0.47	7.8	0.53

^a0.8 μM ON-GT-Acr (GGTTTTTGTGT-Acr), 16 μM NC(11-55), 0.05 M HEPES containing 0.1 M NaCl, pH 7.5. See Materials and Methods for further details. Results of 2 or 3 independent experiments are shown. Standard deviations for the fluorescence lifetimes (τ_i), the lifetime amplitudes (α_i), the rotational correlation times (φ_i), and their amplitudes (β_i) are less than 20%.

The binding affinities were first determined from titrations of NC with the ONs by monitoring the intrinsic fluorescence of Trp37. In the presence of an excess of the various ONs, the Trp fluorescence intensity dropped to approximately 5% of its value in the absence of ONs (Fig. 2A), as was previously observed with other nucleic acid sequences (e.g., references 5, 20, and 43). This strong fluorescence decrease is consistent with a stacking of the Trp37 residue (9) with the ON bases in the complex. Moreover, in line with the anisotropy decay data, the titration data sets were satisfactorily fitted to equation 4, assuming that two NC molecules are bound per ON, as shown in Fig. 2A with mON-11 taken as a representative example. In the presence of 30 mM NaCl, the apparent equilibrium binding constants (for a 2:1 stoichiometry) were about 1.3 × 10⁷ M⁻¹ to 2.8 × 10⁷ M⁻¹ for the GU- or GT-rich ONs and 4 × 10⁶ M⁻¹ for the random sequence (Table 2). Interestingly, only marginal changes in the affinity were observed when the 11-nucleotide (nt) GU-rich sequence was either 2′-O-methylated or coupled at its 3′ end to an eosin dye or shortened at its 3′ end by 2 nt. Moreover, an increase of the ionic strength only moderately reduced the affinities of the GU- or GT-rich ONs (Table 2; Fig. 2B), indicating a major contribution of nonelectrostatic interactions in the stability of the NC/ON complexes. In contrast, the affinity of the random sequence dramatically dropped upon increasing the NaCl concentration from 30 to 100 mM (Fig. 2B; Table 2), suggesting a strong contribution of electrostatic interactions in the complex stability in this case. The full-length NC [NC(1-55)] showed an affinity only twice as high for mON-11 [(2.9 ± 0.7) × 10⁷ M⁻¹] as compared to the truncated molecule NC(11-55), indicating that the binding of the NC is mainly supported by its zinc finger domain.

Fig 2.

NC binding to ONs, as monitored by the intrinsic fluorescence of the NC Trp37 residue. (A) Titration of NC(11-55) with mON-11. Excitation was at 295 nm, and emission at 350 nm. The indicated fluorescence intensities are corrected for the buffer fluorescence, dilution, and inner filter effect due to absorbance of the nucleic acids at 295 nm. Buffer: 25 mM Tris-HCl, pH 7.5, 0.2 mM MgCl₂, 30 mM NaCl. Solid line, fit assuming a 2:1 NC:ON stoichiometry; dashed line, fit assuming a 1:1 stoichiometry. (B) Effect of NaCl concentration on the NC/ON affinity constant. Symbols: squares, ON-GU; triangles, mON-Rnd1.

Table 2.

Binding constants of NC(11-55) to ONs, as determined from titrations monitored through the intrinsic fluorescence of Trp37^a

ON	Ka, M−1 × 10−6, 30 mM NaCl	Ka, M−1 × 10−6, 100 mM NaCl
GGTTTTTGTGT	28 ± 10	9 ± 2
2′O-me(GGUUUUUGUGU)Eos	6.2 ± 0.5	5 ± 1
2′O-me(GGUUUUUGUGU)	14 ± 1	7 ± 3
2′O-me(GGUUUUUGU)	17 ± 4	7.5 ± 0.2
GGUUUUUGUGU	13 ± 7	6 ± 3
2′O-me-(CUUAAGACCAU)	4.1 ± 0.8	<0.1

^aA model assuming two identical binding sites (equation 4) was used to fit the fluorescence titration curves. Mean values ± SD for the equilibrium association constant K_a were obtained from 2 or 3 independent titrations.

To more accurately determine the binding parameters and get further insight into the nature of the interactions, we investigated the interaction between mON-11 and NC(11-55) at 100 mM NaCl by ITC at various temperatures. A representative ITC data set is shown in Fig. 3. Each injection of NC(11-55) into the mON-11 solution was followed by a negative peak of the experimental signal (heat flow), indicating that the reaction is exothermic (Fig. 3A). The quantities of heat as a function of the peptide/ON molar ratio (Fig. 3B) were fitted to a binding model assuming two independent and nonequivalent NC binding sites on mON-11, which was found to be the simplest model to satisfactorily describe the calorimetric data. The values of the equilibrium association constants for the two binding sites, K_a1 and K_a2, were found to be 3.4 × 10⁶ M⁻¹ and 0.19 × 10⁶ M⁻¹, respectively, and thus differed by 1 order of magnitude. The values of the corresponding binding enthalpies, ΔH₁ and ΔH₂, were −7.2 and −10.9 kcal · mol⁻¹, respectively. These parameters were then used to calculate the entropy changes ΔS_1,2 and the free energy of binding ΔG_1,2 (Table 3). The ΔH₁ and ΔH₂ values were found to be almost temperature independent, suggesting that the molar heat capacities are close to zero (ΔC_p1,2 = 12 and 43 cal · mol⁻¹ · K⁻¹). For the higher-affinity site, both ΔH₁ and TΔS₁ values were respectively negative and positive and thus favorable for the interaction over the experimental range of temperatures. In contrast, negative TΔS₁ values corresponding to an unfavorable increase of order were observed for the binding of NC(11-55) to its low-affinity site. Interestingly, while the K_a1 value derived from the ITC data was in good agreement with the binding constants obtained from the fluorescence titration (Table 2), a significant deviation was observed for the K_a2 value. This deviation is likely due to the adsorption of the NC(11-55) peptide to the walls of the quartz cuvette (44), which leads to an incremental fluorescence decrease. As a consequence, deviations from the real binding curve are especially important in the last points of the fluorescence titration curves. Since these points strongly contribute to the determination of the lowest-affinity binding site, an overestimation of its binding constant might occur.

Fig 3.

Determination of NC(11-55)/mON-11 binding parameters by ITC. (A) Representative ITC titration of 2.9 μM mON-11 with NC(11-55) (0.145 mM in the syringe) at 20°C. The baseline is also shown. (B) Plot of the quantities of heat as a function of the NC(11-55)/ON molar ratio. The data are corrected for the readouts of the blank titration. The solid line represents the best fit of the binding isotherm to a model with two independent and different binding sites.

Table 3.

Temperature dependence of the thermodynamic parameters determined by ITC for the 2′O-me(GGUUUUUGUGU)/NC(11-55) interaction^a

Temperature (°C)	Ka1,2 (M−1 × 10−5)	ΔH1,2 (kcal · mol−1)	TΔS1,2b (kcal · mol−1)	ΔG1,2 (kcal · mol−1)
10	24 ± 9	−7.2 ± 0.4	1.1 ± 0.5	−8.7 ± 3
	1.7 ± 0.1	−11.0 ± 0.5	−4.2 ± 0.5	−6.8 ± 0.4
15	27 ± 9	−7.7 ± 0.3	0.8 ± 0.3	−8.5 ± 2
	1.5 ± 0.1	−10.1 ± 0.4	−3.3 ± 0.3	−6.8 ± 0.4
20	34 ± 9	−7.2 ± 0.2	1.5 ± 0.4	−8.7 ± 2
	1.9 ± 0.1	−10.9 ± 0.3	−3.9 ± 0.3	−7.0 ± 0.3
25	53 ± 8	−7.0 ± 0.1	2.1 ± 0.4	−9 ± 1
	1.7 ± 0.1	−10.1 ± 0.3	−3.0 ± 0.2	−7.1 ± 0.3
30	45 ± 10	−7.5 ± 0.3	1.7 ± 0.6	−9.2 ± 3
	2.2 ± 0.2	−9.2 ± 0.4	−1.8 ± 0.2	−7.4 ± 0.5
37	48 ± 10	−6.7 ± 0.4	2.8 ± 0.9	−9.5 ± 2
	2.7 ± 0.2	−10.6 ± 0.5	−2.9 ± 0.4	−7.7 ± 0.7
40	75 ± 20	−7.4 ± 0.5	2.5 ± 1.0	−10 ± 3
	2.4 ± 0.3	−8.2 ± 0.3	−0.5 ± 0.1	−7.7 ± 0.9
42	75 ± 20	−7.1 ± 0.5	2.8 ± 1	−10 ± 3
	2.7 ± 0.2	−10.0 ± 0.4	−2.6 ± 0.2	−7.5 ± 0.5

^aData are reported as mean values ± SD for 2 or 3 experiments. Experiments were performed in 50 mM HEPES, pH 7.4, 100 mM NaCl.

^bCalculated from ΔG = −RTlnK_a and ΔG = ΔH − TΔS.

Taken together, our data show that the GU-rich mONs bind two NC molecules with rather high affinities, confirming that NC is a potential target of such mONs. Interestingly, the affinities of the GU-rich mONs are much higher than those of random guanine-free mONs and rely mainly on nonelectrostatic interactions with the central zinc finger domain, which is critical for the NC functions (13, 17). As a consequence, the GU-rich mONs are expected to hinder or prevent the NC activities mediated through NC binding to its target nucleic acid sequences.

The GU- and GT-rich ONs inhibit NC-promoted cTAR destabilization by a competitive mechanism.

The relatively high affinities of the GU- and GT-rich ONs for NC suggested that they might efficiently compete with the nucleic acid sequences chaperoned by NC, such as for instance cTAR. To characterize the effect of the GU- and GT-rich ONs on the NC chaperone activities, we first monitored their effect on the NC-promoted destabilization of a doubly labeled cTAR (4, 8). When the stem of cTAR is closed, the Rh6G fluorophore at its 5′ end is efficiently quenched by the DABCYL quencher at its 3′ end, so that the fluorescence of the doubly labeled cTAR is very low. The addition of NC leads to a partial melting of the cTAR stem, causing an increase in the distance between the Rh6G and DABCYL dyes and thus a partial restoration of the Rh6G fluorescence. In line with this, we observed a fluorescence intensity increase of the doubly labeled cTAR by about 6 or 7 times upon the addition of a saturating NC(11-55) concentration (Fig. 4A). Upon subsequent addition of the GU- or GT-rich ONs to the preformed NC(11-55)/cTAR complex, a drastic decrease in Rh6G fluorescence was observed, indicating that the ONs reverse the effect of NC on cTAR. About 50% of the destabilizing effect of NC(11-55) was inhibited by an approximately 2-fold excess of the GU- or GT-rich ONs over cTAR, while almost full suppression of the effect of NC(11-55) was reached at a 16-fold excess of the ONs (Fig. 4B). All GU- and GT-rich ONs showed similar inhibition values on the NC-promoted cTAR destabilization, while the random sequences were much less efficient (Fig. 4B and D). The 50% inhibitory concentrations (IC₅₀s) for the ONs depended linearly on [cTAR], in line with a competitive inhibition mechanism (Fig. 4C). No direct stabilizing effect of the ONs on cTAR was observed in control experiments in the absence of NC(11-55) (data not shown), confirming that NC is the target of the ONs.

Fig 4.

Effect of the ONs on the NC(11-55)-promoted destabilization of the cTAR stem-loop. (A) Effect of ON-GT on the fluorescence spectrum of 25 nM Rh6G-5′-cTAR-3′-DABCYL. Excitation was at 520 nm. Dotted line, Rh6G-5′-cTAR-3′-DABCYL in the absence of NC(11-55) and ON-GT; dash-dot-dotted line, Rh6G-5′-cTAR-3′-DABCYL in the presence of 275 nM NC(11-55); solid lines, same in the presence of increasing ON-GT concentrations (indicated by the arrow). a.u., arbitrary units. (B) Dependence of the fluorescence intensity of Rh6G-5′-cTAR-3′-DABCYL in the presence of NC(11-55) on the ON concentration. Representative curves. 25 nM Rh6G-5′-cTAR-3′-DABCYL, 275 nM NC(11-55), 30 mM NaCl. Symbols: diamonds, mON-11; stars, mON-Rnd2. (C) Dependence on cTAR concentration of the IC₅₀ of ON-GT for the inhibition of the NC-induced cTAR destabilization. The IC₅₀s were determined in the presence of an 11-fold molar excess of NC(11-55) over cTAR. (D) IC₅₀s for the inhibition of the NC-induced cTAR destabilization by the various ONs. Experiments were performed with 25 nM cTAR and 275 nM NC(11-55). *, 50% inhibition of cTAR destabilization was not reached at the highest tested concentration of mON-Rnd2 (800 nM), so the bar in panel D represents a lower-bound estimate of its IC₅₀.

To investigate the possible formation in these conditions of a ternary complex containing cTAR, NC(11-55) and the ON, fluorescence resonance energy transfer (FRET) experiments were performed using carboxyfluorescein (Fl) attached to the 3′ end of cTAR as the donor and eosin attached to the 3′ end of mON-11-Eos as the acceptor. Due to the significant overlap between the Fl emission and eosin absorbance spectra, a close proximity between these dyes in a ternary complex should induce an Fl fluorescence decrease. The absence of any Fl intensity decrease in the presence of a 57 to 250 nM concentration of mON-11-Eos, which significantly inhibits the NC(11-55)-induced cTAR destabilization, suggested that no ternary complex is formed (data not shown). As a consequence, the ONs likely inhibit the NC chaperone activity by a fully competitive mechanism.

The GU- and GT-rich ONs inhibit NC-promoted cTAR/dTAR annealing.

To examine the effect of the ONs on NC-promoted cTAR/dTAR annealing, we used a doubly labeled cTAR species, with Fl and TMR fluorophores attached to the 3′ and 5′ termini of cTAR stem, respectively (23). In the closed species, the Fl and TMR dyes are close and form a nonfluorescent heterodimer (7). In contrast, conversion of the cTAR stem-loop into the 55-bp dTAR/cTAR extended duplex greatly increases the distance between the chromophores, leading to a full recovery of the Fl emission peak at 520 nm (23) (Fig. 5A). Thus, the fluorescence of TMR-5′-cTAR-3′-Fl allows a sensitive real-time monitoring of the annealing kinetics in solution. Spontaneous annealing of dTAR to cTAR in the absence of NC is very slow, taking more than 36 h to be complete in the present conditions. In contrast, saturating NC concentrations [with 11 NC(11-55) molecules per cTAR or dTAR] strongly accelerate this process, by increasing the annealing rate by 3 orders of magnitude. Interestingly, the GU- and GT-rich ONs markedly slowed down the NC(11-55)-promoted cTAR/dTAR annealing reaction. In contrast, the guanine-free mON-Rnd2 induced only a limited decrease in the reaction rate (28) (Fig. 5B).

Fig 5.

Inhibition of NC(11-55) promoted cTAR/dTAR annealing by the ONs. (A) Fluorescence spectra of 5 nM TMR-5′-cTAR-3′-Fl in the absence (solid line) or in the presence (dashed line) of 55 nM NC(11-55) and after annealing with 50 nM dTAR (dash-dotted line). Excitation was at 480 nm. (B) Effect of mON-11 on NC-promoted cTAR/dTAR annealing reaction. The annealing reaction was recorded in the absence of NC(11-55) (dash-dotted line, spontaneous cTAR/dTAR annealing), in the presence of 605 nM NC(11-55) (solid line), or in the presence of both 605 nM NC(11-55) and 440 nM mON-11 (dashed line). Data are from the work of Grigorov et al. (28). a.u., arbitrary units. (C) Dependence of the residual annealing activity of NC(11-55) (calculated as described in Materials and Methods) on the ratio of the inhibitor to the sum of cTAR and dTAR concentrations. Symbols: closed diamonds, mON-11; open diamonds, mON-11-Eos; closed squares, ON-GT; 605 nM NC(11-55), 5 nM cTAR, and 50 nM dTAR. Mean values ± standard deviations (SD), n = 2 to 4. (D) Annealing activity of NC(11-55) (605 nM) on 5 nM cTAR and 50 nM dTAR in the presence of 440 nM the various ONs (8-fold molar excess over cTAR plus dTAR); values ± SD, n = 2 to 5.

By systematically investigating the inhibition of the NC(11-55) chaperone activity by these ONs, we found that the observed annealing rate was reduced by approximately 50% in the presence of a 2- to 3-fold molar excess of mON-11 or mON-11-Eos over cTAR and dTAR (Fig. 5C). ON-GT and the nonmethylated ON-11, as well as mON-9, inhibited the annealing about two times less efficiently than mON-11, suggesting that both the methylation of sugars and the 3′-terminal GU motif contribute to the inhibitory effect of mON-11 (Fig. 5D). In contrast to the annealing rate (k_obs), the final value of the fluorescence intensity after annealing completion was not affected by the ONs (Fig. 5B), indicating that all cTAR molecules will finally anneal with dTAR in the presence of the ONs, in line with a competitive inhibition mechanism. Indeed, the ONs likely sequester a fraction of the NC(11-55) molecules, thus reducing the effective peptide concentration bound to cTAR and dTAR. As a result, the annealing is slowed down but not prevented since cTAR-dTAR annealing is a nearly irreversible process due to the large free energy decrease accompanying the stem-loop → duplex transition. Therefore, though only a limited number of NC molecules can bind to cTAR and dTAR at a given time, these bound NC molecules can still promote the irreversible annealing of cTAR with dTAR, so that the duplexes can accumulate with time.

DISCUSSION

To get an insight into the inhibition mechanism of HIV-1 replication by modified GU-rich mONs (28), we quantitatively characterized their binding to NC and their inhibition of the NC nucleic acid chaperone activity.

We found that two NC molecules bind to the 11-nt mONs, in line with the previously reported occluded binding size of 5 or 6 nucleotides per NC molecule (2, 14, 21, 22, 43). The equilibrium binding constants of the GU- and GT-rich ONs at 100 mM NaCl (Table 2) were close to the reported affinities of NC to TG- or UG-containing hexanucleotides or dodecanucleotides (2, 3, 43), suggesting that the GU or GT motifs provide a major contribution to NC binding to such ONs. Interestingly, an increase of the ionic strength only moderately reduced the affinities of the GU- and GT-rich ONs but caused a dramatic drop for the random sequence (Table 2, Fig. 2B), indicating a major contribution of the nonelectrostatic interactions for the binding of NC to the GU- and GT-rich ONs. This strong contribution of nonelectrostatic interactions to the binding constant of these ONs is in agreement with the key role played by the GU and GT motifs in the interaction with NC (22, 43).

ITC experiments revealed that the binding of NC to the high-affinity site on the inhibitory ONs probably relies in large part on the interaction of the hydrophobic platform at the top of the NC fingers with the GU or GT motifs. Indeed, this interaction and, notably, the stacking between Trp 37 and guanine residues probably account for the entropically favorable release of water molecules suggested from the positive ΔS₁ value. This gain in entropy may also result, at least in part, from the release of counterions (i.e., the polyelectrolyte effect resulting from the interaction between the positively charged side chains of the peptide and the ON phosphates) (38). The slightly positive molar heat capacity is also consistent with the release of water molecules due to hydrophobic interactions (39) and the contribution of electrostatic interactions but is far less than for the interaction of the HIV-1 Tat protein with heparin sulfate, which is thought to be entirely electrostatically driven (46). NC binding to the low-affinity site (characterized by K_a2 and ΔH₂) is only enthalpy driven (TΔS <0). Here again, the small positive ΔC_p value suggests the participation of both electrostatic and nonionic interactions, while the negative entropy change ΔS₂ likely reflects a more ordered conformation resulting from the loss of freedom due to the ON/NC association. Thus, the 9-nt and 11-nt ONs bind two NC molecules with micromolar affinities, mainly by nonelectrostatic interactions between the GU and GT motifs and the NC hydrophobic platform. Since the affinities for the GU- and GT-rich ONs are similar to those for the NC-specific binding sites on the viral RNA and DNA, i.e., 1.7 × 10⁷ M⁻¹ at 30 mM NaCl for cTAR (5), 2 × 10⁶ M⁻¹ at 100 mM NaCl for (−)PBS stem-loop (10), and ≈10⁷ M⁻¹ at 25 mM NaCl for the SL2 and SL3 stem-loops of the genomic packaging signal (1, 14), the GU- and GT-rich ONs can compete with the viral sequences for NC binding in infected cells.

As a result of this binding competition, a reduction of the number of NC molecules bound to its target sequences during reverse transcription is expected. This may notably affect the TAR RNA and cTAR DNA sequences involved in the first cDNA strand transfer. Since both the nucleic acid destabilization and the annealing promotion components of NC chaperone activity depend on the level of nucleic acid coating by NC molecules (8, 42; reviewed in reference 13), these two components were found to be significantly altered in the presence of the inhibitory ONs which compete with the stem-loops to bind NC. Indeed, a 2- to 3-fold molar excess of mON-11 over cTAR inhibited by a factor of 2 the NCp7-induced destabilization of cTAR and the rate of NCp7-promoted cTAR/dTAR annealing. How these changes will affect the reverse transcription process could be tentatively answered based on a recent work on the NCp7-promoted (−)/(+)PBS annealing reaction (25). NCp7 was shown to induce via its hydrophobic platform a mechanistic switch from an annealing pathway through the PBS single-strand overhangs to a kissing loop pathway that is thought to play an important role to ensure the specificity and fidelity of the second strand transfer. A similar NC-induced mechanistic switch was observed for the TAR/cTAR annealing reaction, since increasing concentrations of NC progressively shifted the kissing loop reaction pathway that predominates in the absence of NC to a pathway through the stem termini that is the only one when the cTAR and TAR sequences are coated by NC molecules (42). This pathway through the stem termini relies on the ability of NC to destabilize the cTAR termini through its hydrophobic platform (5) and to screen the repulsion between the negatively charged ONs, through its positively charged basic residues (45). Therefore, sequestration of NC molecules by mONs will decrease the number of NC molecules bound to the TAR and cTAR sequences and thus prevent the specific annealing pathway, which likely ensures the specificity and fidelity of the first strand transfer. In addition, a reduction of the nucleic acid destabilization effect of NC will prevent it from blocking nonspecific self-primed cDNA synthesis (18, 27, 30–32, 35), which will further impact on the fidelity of reverse transcription and the synthesis of viral DNA.

One may note that the potencies of the ONs in inhibiting NC-directed cTAR/dTAR annealing did not correlate strictly with the affinities for NC(11-55) (Table 2 and Fig. 5D). For instance, while mON-11 and mON-9 show similar affinities for cTAR, the second showed significantly lower inhibition of the NC-promoted cTAR/dTAR annealing and lower antiviral activity (28). One possible explanation is that in the cellular context, the difference in affinities of NC for mON-11 and mON-9 is more pronounced than in solution, providing a better match between the binding and antiviral activities. Alternatively, the two mONs may differ for their interaction with RT, which was found to be a target for both mONs, since resistant viral clones contained mutations in both RT and NC sequences (28). However, using the most active mON-11 inhibitor added up to a 100 μM concentration, we did not observe any inhibition of the RNA- and DNA-dependent DNA polymerase activities of purified RT (reference 27 and data not shown). In contrast, the RNase H activity of purified RT was inhibited by mON-11, but at much higher concentrations (IC₅₀ = 20 μM) than for NC activities (data not shown), suggesting that RT may not be targeted directly by the mONs. In this respect, the sequestration of NC molecules by the mONs may not only inhibit the NC nucleic acid chaperone activity but also prevent the binding of NC with RT (11, 19, 29), which may be required for the fidelity and efficiency of reverse transcription.

In conclusion, we report here that by sequestering NC molecules, the GU-rich mONs alter the NC-chaperoning reactions and/or the NC/RT interaction during viral DNA synthesis by the RT enzyme. This, in turn, is thought to alter the contribution of NC to the DNA strand transfers and the specificity and fidelity of the reverse transcription process (13). This sequestration of NC molecules is most probably responsible for the inhibition of HIV replication in cells (28) by such mONs. Based on this proposed mechanism of the mONS, it would be useful in a next step to determine the 3-dimensional (3D) structure of the mON/NCp7 complexes in order to more precisely identify the binding determinants of the mONs. This would in turn allow the rational design of potential therapeutic molecules able to bind NCp7 molecules with high affinity and thus to inhibit the viral steps in which NCp7 molecules play a key role.