Probing dynamics of HIV-1 nucleocapsid protein/target hexanucleotide complexes by 2-aminopurine

Abstract

The nucleocapsid protein (NC) plays an important role in HIV-1, mainly through interactions with the genomic RNA and its DNA copies. Though the structures of several complexes of NC with oligonucleotides (ODNs) are known, detailed information on the ODN dynamics in the complexes is missing. To address this, we investigated the steady state and time-resolved fluorescence properties of 2-aminopurine (2Ap), a fluorescent adenine analog introduced at positions 2 and 5 of AACGCC and AATGCC sequences. In the absence of NC, 2Ap fluorescence was strongly quenched in the flexible ODNs, mainly through picosecond to nanosecond dynamic quenching by its neighboring bases. NC strongly restricted the ODN flexibility and 2Ap local mobility, impeding the collisions of 2Ap with its neighbors and thus, reducing its dynamic quenching. Phe¹⁶→Ala and Trp³⁷→Leu mutations largely decreased the ability of NC to affect the local dynamics of 2Ap at positions 2 and 5, respectively, while a fingerless NC was totally ineffective. The restriction of 2Ap local mobility was thus associated with the NC hydrophobic platform at the top of the folded fingers. Since this platform supports the NC chaperone properties, the restriction of the local mobility of the bases is likely a mechanistic component of these properties.

INTRODUCTION

Nucleocapsid protein (NC), either as a domain of the Gag polyprotein precursor or as the mature protein, is essential for several important steps of the virus life cycle (1). For instance, selection of viral genomic RNA for packaging into virions is mediated by NC binding to the ψ encapsidation sequence within the untranslated region of the HIV-1 genome (2). Furthermore, NC chaperones the annealing of the primer tRNA to the primer binding site (PBS), and the two obligatory strand transfers necessary for the synthesis of a complete proviral DNA by reverse transcriptase (3,4). Mature NC is a basic protein of 55 amino acids (Figure 1) containing two highly conserved CX₂CX₄HX₄C zinc fingers (5). Mutations of the zinc coordinating residues and the two aromatic residues of the zinc fingers lead to completely noninfectious viruses (6–9).

Figure 1.

Structures of the native and mutated HIV-1 NC peptides used in this study: NC(1-55), NC(11-55), A¹⁶NC(11-55), L³⁷NC(11-55) and (SSHS)₂NC(11-55).

Most NC functions rely on interactions with nucleic acid targets. The structure of NC complexed to the SL2 and SL3 stem-loops as well as to the ACGCC sequence of the ψ encapsidation sequence have been solved by NMR (10–12). In all these complexes, the hydrophobic platform formed by the Val¹³, Phe¹⁶, Thr²⁴, Ala²⁵, Trp³⁷ and Met⁴⁶ residues at the surface of the two folded zinc fingers has been shown to be critical for binding. A particularly important role was notably demonstrated for Trp³⁷ that stacks with G residues in all these complexes.

Though 3D structures of NC complexes with several oligonucleotides (ODNs) are available, the dynamics of the ODNs in the complexes is still not well characterized. Dynamics of either single- and double-stranded nucleic acids plays an important role in their interaction with various ligands (13,14). For instance, in the HIV-1 RNA TAR stem-loop, the magnitude of the adaptive structure change due to ligand binding at a given site correlates with the degree of spontaneous internal motions at this site (15). It is likely that the local dynamics of ODNs are important for NC binding and activities as well.

Site-specific information on ODN dynamics can be provided by environmentally sensitive fluorescent probes, such as 2-aminopurine (2Ap), a fluorescent analog of adenine that can be selectively excited at 315–320 nm, where Trp does not absorb (16). The quantum yield of free 2Ap at pH 7.0 is 0.68 (17) but within ODNs it is significantly quenched due to interactions with its neighbor bases (18–22). Free 2Ap in water exhibits a single lifetime of about 10–11 ns, while multiple shorter lifetimes appear when 2Ap is incorporated into ODNs, indicating a highly heterogeneous environment of the probe. As a consequence, time-resolved fluorescence of 2Ap has been largely used to site-specifically characterize the interactions of ODNs with various ligands, including complementary nucleic acids strands and proteins (23–27).

To site-specifically investigate the changes in the dynamics of ACGCC during NC binding and correlate them with the known 3D structure of the NC/ACGCC complex (11), 2Ap was introduced either downstream or upstream to the CG motif. Moreover, since the environment of 2Ap at the 3′ or 5′ terminus of an ODN differs significantly from that at an internal position (18,28), complicating the data interpretation, an extra adenine was added at the 5′-terminus of ACGCC. This extra adenine was shown to not affect the binding of the ODN to NC (29). Moreover, substitutions of A2 and C5 with 2Ap within ¹AACGCC⁶ were also expected to not affect NC binding, since the corresponding residues of ACGCC only marginally interact with NC (11). Furthermore, since NC binds with strong affinity to TG motifs (29,30), we studied the interaction of NC with 2Ap-labeled AATGCC sequences. Finally, the native NC was substituted with several NC mutants (Figure 1) to investigate the role of the N-terminal basic domain and the folded finger domain as well as the hydrophobic platform on the top of the fingers on the dynamics of the labeled ODNs.

MATERIALS AND METHODS

Materials

NC(1-55) (further referred as NC), NC(11-55), (SSHS)₂NC(11-55), L³⁷NC(11-55) and A¹⁶NC(11-55) (Figure 1) were synthesized on a Applied Biosystems A433 peptide synthesizer as described (31). 2Ap-substituted DNA ODNs were synthesized and HPLC-purified by IBA GmbH (Germany). Absorbance spectra were recorded with a Cary 400 UV-visible spectrophotometer (Varian). ODN concentrations were calculated using the extinction coefficients (ε₂₆₀, M⁻¹ × cm⁻¹) of 58 050, 44 190, 51 390, 59 220, 45 360 and 52 560 for AACGCC, AApCGCC, AACGApC, AATGCC, AApTGCC and AATGApC, respectively. All experiments were performed in 50 mM HEPES, pH 7.5, at 20°C. ‘Fluorescence microscopy’ grade glycerol (Merck, Germany) was used. All other chemicals were purchased from Sigma, USA.

Steady-state fluorescence spectroscopy

Fluorescence spectra were recorded on FluoroMax3 and FluoroLog spectrofluorimeters (Jobin Yvon) equipped with thermostated cell compartments. Spectra were corrected for screening effects and buffer fluorescence. Quantum yield was calculated using free 2Ap as a reference [0.68 (17)]. 2Ap was excited at 315 nm. To determine the affinity of NC for the 2Ap-labeled ODNs, fixed amounts of NC were titrated with ODNs by monitoring the intrinsic Trp fluorescence of NC at 320-nm emission wavelength (to exclude 2Ap fluorescence). Affinity constants were determined from direct fitting of the experimental signal to the rewritten Scatchard equation:

1

where I and I_t are the intensities at a given and a saturating ODN concentration, respectively, I₀ is the intensity in the absence of ODN, N_t is the total ODN concentration, P_t is the total concentration of peptide, K_app is the apparent affinity constant, n is the number of binding sites. Alternatively, the K_app values were also obtained by titrating fixed amounts of 2Ap-labeled ODNs with the peptides and monitoring the 2Ap fluorescence at 370 nm. The data were fitted with a slightly modified version of Equation (1). The parameters were recovered from nonlinear fits of Equation (1) to experimental datasets by the Microcal Origin™ 6.0 program.

Time-resolved fluorescence measurements

Time-resolved fluorescence measurements were performed with the time-correlated, single-photon counting technique using the excitation pulses at 315 nm provided by a pulse-picked frequency tripled Ti-sapphire laser (Tsunami, Spectra Physics) pumped by a Millenia X laser (Spectra Physics) (32). The emission was collected through a polarizer set at magic angle and an 8-nm band-pass monochromator (Jobin-Yvon H10) at 370 nm. The instrumental response function was recorded with a polished aluminum reflector, and its full width at half-maximum was 40 ps. The mean lifetime <τ> was calculated from the individual fluorescence lifetimes, τ_i, and their relative amplitudes, α_i, according to <τ> = Σα_iτ_i. The population, α₀, of dark species of 2Ap within the ODNs was calculated by: α₀ = 1 − τ_free/(τ_sample × R_m), where τ_free is the lifetime of free 2Ap, τ_sample is the measured lifetime of 2Ap within the ODN (either free or bound to NC) and R_m is the ratio of the corresponding steady-state fluorescence intensities. The remaining amplitudes, α_ic, were recalculated from the measured amplitudes, α_i, according to: α_ic = α_i × (1 − α₀).

For time-resolved anisotropy measurements, the fluorescence decay curves were recorded at vertical and horizontal positions of the polarizer and analyzed by the following equations:

2

where β_i are the amplitudes of the rotational correlation times φ_i; 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 initial anisotropy value (0.33) was determined independently for 2Ap in 77% glycerol (v/v) from the extrapolation of the anisotropy decay curves to zero time. The theoretical values of the rotational correlation times were calculated from the molecular masses (M) of the molecules and their complexes, assuming spherical shapes, by:

3

where η is the viscosity (assumed to be 0.01 P), T is the temperature (maintained at 293 K), ν is the specific volume of the particle [assumed to be 0.83 ml/g, (33)], h is the hydration degree (assumed to be 0.2 ml/g for proteins) and R is the molar gas constant.

The cone semiangle (θ₀) providing an estimate for the local motions of 2Ap modeled as diffusion within a cone (34,35) was calculated as:

4

where S is the generalized order parameter: S = (β′₂/(β′₁ + β′₂))^1/2, β′₁ and β′₂ are the amplitudes of the rotational correlation times, recalculated such that Σ β′_i = r₀.

Time-resolved intensity and anisotropy data were treated with a nonlinear least-square analysis (39) using a homemade software (kindly provided by G. Krishnamoorthy). In all cases, the χ² values were close to 1, and the weighted residuals as well as the autocorrelation of the residuals were distributed randomly around zero, indicating an optimal fit.

RESULTS

2Ap fluorescence within free hexanucleotides

Fluorescence spectra of the AACGCC and AATGCC sequences labeled by 2Ap at position 2 or 5 showed a maximum emission wavelength at around 370 nm, similarly to free 2Ap (Figure 2). In contrast, the quantum yield of 2Ap within ODNs is much lower than that of the free fluorophore (Figure 2), in agreement with the literature (36). The quenching in respect with free 2Ap is more pronounced for 2Ap in the position 5 (38-fold intensity decrease) than in the position 2 (about 19-fold decrease) (Table 1), probably due to the flanking G4 residue, since guanine is the best quencher of 2Ap fluorescence among the bases (37). Moreover, the identical quantum yield of AApCGCC and AApTGCC suggests that 2Ap fluorescence does not depend on the nature of the pyrimidine base in position 3.

Figure 2.

Effect of NC and glycerol on the fluorescence spectra of AApCGCC and AACGApC. The fluorescence spectra of AApCGCC (dashed lines) and AACGApC (dash-dotted lines) sequences were recorded in the absence of NC either in buffer (red), in 77% glycerol (green) or in the presence of a 7.5-fold molar excess of NC (blue). Solid line is free 2Ap. Excitation at 315 nm, excitation and emission slits: 2 and 5 nm, respectively.

Table 1.

Steady-state and time-resolved fluorescence parameters for 2Ap-substituted hexanucleotides^a

	Quantum yield	Rm (Ifree2Ap/I)	α0	τ1 (ns)	α1	τ2 (ns)	α2	τ3 (ns)	α3	τ4 (ns)	α4	<τ> (ns)
Free 2Ap	0.680b									10.2	1.00	10.2
AApCGCC	0.036	19.0	0.41	0.19	0.20	0.60	0.27	1.94	0.08	4.99	0.04	0.91
AApCGCC, glycerol 77%	0.289	2.35	0.11	0.39	0.21			3.78	0.30	8.17	0.38	4.86
AApCGCC + NC	0.090	7.57	0.36	0.17	0.15	0.69	0.24	2.01	0.11	6.58	0.14	2.09
AApCGCC + NC(11-55)	0.070	9.74	0.33	0.14	0.17	0.81	0.33	3.05	0.11	6.91	0.06	1.57
AApCGCC + (SSHS)2NC(11-55)	0.039	17.6	0.41	0.14	0.22	0.64	0.24	2.28	0.09	5.62	0.04	0.99
AApCGCC + A16NC(11-55)	0.047	14.6	0.33	0.11	0.29	0.59	0.23	2.28	0.10	6.55	0.05	1.05
AACGApC	0.018	37.4	0.79	0.08	0.10	0.55	0.04	2.20	0.04	4.93	0.03	1.27
AACGApC, glycerol 77%	0.176	3.86	0.38	0.24	0.24			2.60	0.10	8.18	0.28	4.25
AACGApC + NC	0.134	5.06	0.21	0.12	0.19	0.75	0.19	3.77	0.32	7.85	0.09	2.56
AACGApC + NC(11-55))	0.105	6.49	0.33	0.10	0.19	0.91	0.18	3.52	0.18	6.36	0.12	2.35
AACGApC + (SSHS)2NC(11-55)	0.020	34.0	0.80	0.08	0.09	0.72	0.04	2.79	0.05	6.01	0.02	1.49
AACGApC + A16NC(11-55)	0.108	6.28	0.25	0.15	0.24	0.85	0.17	3.03	0.20	5.98	0.14	2.15
AApTGCC	0.036	18.9	0.51	0.14	0.22	0.85	0.18	2.16	0.04	5.26	0.05	1.11
AApTGCC, glycerol 77%	0.287	2.37	0.0	0.12	0.27			2.03	0.23	7.60	0.50	4.30
AApTGCC + NC	0.147	4.62	0.20	0.20	0.17	1.04	0.30	3.49	0.15	7.54	0.18	2.77
AApTGCC + NC(11-55)	0.085	7.97	0.28	0.44	0.41	1.48	0.14	3.66	0.08	6.55	0.09	1.77
AApTGCC + (SSHS)2NC(11-55)	0.034	19.8	0.54	0.08	0.18	0.66	0.17	2.47	0.07	6.13	0.04	1.12
AApTGCC + A16NC(11-55)	0.045	15.1	0.49	0.10	0.19	0.65	0.20	2.55	0.08	6.56	0.04	1.30
AApTGCC + L37NC(11-55)	0.107	6.39	0.16	0.12	0.29	0.75	0.28	2.98	0.16	7.84	0.11	1.90
AATGApC	0.018	38.4	0.80	0.10	0.10	1.10	0.03	2.28	0.03	4.75	0.04	1.33
AATGApC, glycerol 77%	0.227	3.00	0.16	0.12	0.29			1.55	0.15	7.83	0.40	4.06
AATGApC + NC	0.125	5.45	0.25	0.12	0.22	1.02	0.18	3.81	0.30	6.58	0.05	2.50
AATGApC + NC(11-55)	0.126	5.39	0.22	0.19	0.28	1.25	0.13	3.95	0.28	6.58	0.09	2.43
AATGApC + (SSHS)2NC(11-55)	0.018	37.7	0.81	0.09	0.09	0.80	0.04	2.83	0.04	5.97	0.02	1.39
AATGApC + A16NC(11-55)	0.106	6.40	0.27	0.19	0.26	1.09	0.16	3.59	0.22	6.86	0.09	2.17
AATGApC + L37NC(11-55)	0.073	9.28	0.43	0.09	0.24	0.72	0.10	2.70	0.12	6.28	0.11	1.95

^aExperiments were performed with 1–8 μM 2Ap-labeled ODN in 50 mM HEPES, pH 7.5. The concentrations of NC derivatives were at least seven times higher. The amplitude values are corrected for the dark species, as described in Materials and Methods section. Standard deviations are lower than 15% for the quantum yield, 20% for lifetimes (except for τ₁, where they are below 45%), 20% for amplitudes and 15% for the mean lifetimes, respectively.

^bData from Ref. (17).

Since steady-state fluorescence is unable to report on the dynamics and heterogeneity of the fluorophore microenvironment, time-resolved fluorescence measurements were performed. Four components ranging from ≈100 ps to 5 ns were required for good fitting (χ² < 1.4) of the intensity decay curves (Table 1), demonstrating a large conformational heterogeneity of 2Ap in the studied ODNs. Multiple lifetimes are frequently observed in 2Ap-labeled ODNs and are usually associated with partially stacked structures with the shortest component reflecting decay from stacked conformations and the 8–10 ns components assigned to conformations where the 2Ap is extrahelical (37,38). In the studied ODNs, the longest lifetime (4–5 ns) was substantially lower than the lifetime of free 2Ap in water but comparable to that of 2Ap-labeled trinucleotides (19) and single-stranded ODNs (28). By analogy to the latter sequences, nanosecond conformational fluctuations of the 2Ap-labeled hexanucleotides in their excited state can drive 2Ap from an unstacked to a stacked conformation, leading to a fluorescence quenching and thus, a lifetime shortening.

The mean lifetime of 2Ap in the four ODNs (0.91–1.33 ns) was found to be 7.7- to 11.2-fold shorter than that of the free probe (10.2 ns) (Table 1). However, this difference in lifetimes was far less than the 19- to 38-fold difference in the corresponding quantum yields, suggesting the presence of a significant subpopulation of ‘dark’ species (α₀), i.e. species with a lifetime shorter than the detection limit of our equipment (≈30 ps) (Table 1). Generally, such apparently ‘null’ lifetime may appear as a result of either static quenching or very fast dynamic quenching. To discriminate between ‘true’ static and fast dynamic quenching, we studied the effect of viscosity on 2Ap fluorescence since only dynamic quenching is affected by viscosity. We observed that in 77% glycerol (v/v), both the quantum yield and the mean lifetime strongly increased (Table 1). Moreover, the number of components needed for good fitting of the decay curves dropped to 3, suggesting a less heterogeneous 2Ap microenvironment. All the observed lifetimes were substantially longer than in the absence of glycerol. The longest component became close to the lifetime of free 2Ap and its amplitude grew to 28–50%, indicating that a large subpopulation of 2Ap does not interact with its neighbors during the excited state lifetime in this case. Moreover, α₀ dramatically decreased at high viscosity (Table 1), demonstrating that the ‘null’ lifetime mostly represents ultra-fast dynamic quenching. The corresponding ultra-short lifetime that cannot be measured with our device likely corresponds to the 10-ps lifetime measured with a streak camera for 2Ap within single-stranded DNAs (39). Interestingly, α₀ values for ODNs labeled at position 5 were much higher than for ODNs labeled at position 2, probably due to the flanking G4 residue which causes a more efficient quenching than other bases (21,24,39).

Fluorescence anisotropy decay curves (Figure 3) of the 2Ap-labeled ODNs were fitted adequately with a two-exponential model (Table 2). Since the slower component φ₂ (0.68–0.83 ns) is slightly shorter than the 1 ns theoretical correlation time for the tumbling of a sphere with the same molecular mass than the ODN (33), the φ₂ component likely represents a combination of both tumbling and segmental motions, in line with the high flexibility expected for single-stranded ODNs. Moreover, the faster component φ₁ (0.14–0.28 ns) can be attributed to the local motion of 2Ap in the ODNs (37). The relative amplitude (β₁) of this faster component allowed to estimate the angular range of the local motion of 2Ap modeled as the diffusion in a cone semi-angle (θ₀) [Equation (4)] (34,37). In further keeping with the high flexibility of the ODNs, rather large θ₀ values were obtained (40–43°) indicating an important rotational freedom for 2Ap (Table 2). Noticeably, since the most quenched conformations (with lifetimes below 0.3 ns) contribute very little (<10%) to the total fluorescence intensity, the calculated diffusion angle likely describes mainly the local motion of the less quenched conformations.

Figure 3.

Experimental anisotropy decay curves of the 2Ap-labeled ODNs and their complexes. AApCGCC alone (blue), AApCGCC/NC (green) and AACGApC/NC (red). The corresponding fitted curves, using the parameters in Table 2 are in black.

Table 2.

Fluorescence anisotropy decay parameters for 2Ap-substituted hexanucleotides^a

	φ1 (ns)	β1	φ2 (ns)	β2	S	θ0 (degree)
2Ap	0.08	1.00			0	90.0
AApCGCC	0.28	0.59	0.83	0.41	0.633	42.9
AApCGCC + NC	0.30	0.38	4.31	0.62	0.787	31.8
AApCGCC + NC(11-55)	0.34	0.43	3.10	0.57	0.755	34.2
AApCGCC + (SSHS)2NC(11-55)	0.24	0.61	2.21	0.39	0.625	43.4
AApCGCC + A16NC(11-55)	0.36	0.44	3.11	0.56	0.755	34.2
AACGApC	0.18	0.57	0.76	0.43	0.656	41.7
AACGApC + NC	0.46	0.42	4.08	0.58	0.762	33.9
AACGApC + NC(11-55)	0.35	0.41	3.17	0.59	0.768	32.9
AACGApC + (SSHS)2NC(11-55)	0.30	0.61	2.21	0.39	0.625	43.8
AACGApC + A16NC(11-55)	0.43	0.41	3.52	0.59	0.768	33.2
AApTGCC	0.23	0.56	0.74	0.44	0.663	40.8
AApTGCC + NC	0.26	0.39	4.13	0.61	0.781	32.2
AApTGCC + NC(11-55)	0.33	0.40	3.04	0.60	0.775	32.7
AApTGCC + (SSHS)2NC(11-55)	0.32	0.67	1.99	0.33	0.575	47.1
AApTGCC + A16NC(11-55)	0.38	0.37	3.00	0.63	0.794	31.2
AApTGCC + L37NC(11-55)	0.42	0.36	3.86	0.64	0.802	30.6
AATGApC	0.14	0.54	0.68	0.46	0.678	39.7
AATGApC + NC	0.50	0.45	4.33	0.55	0.742	34.2
AATGApC + NC(11-55)	0.52	0.41	3.45	0.59	0.768	33.2
AATGApC + (SSHS)2NC(11-55)	0.31	0.69	2.27	0.31	0.554	48.3
AATGApC + A16NC(11-55)	0.44	0.41	3.23	0.59	0.768	33.2
AATGApC + L37NC(11-55)	0.28	0.44	3.03	0.56	0.751	34.0

^aβ₁ and β₂, the amplitudes of the components; S, generalized order parameter; θ₀, cone semi-angle for 2Ap local motion (calculated as described in Materials and Methods section). Standard deviations are <10% for the amplitudes, <30% for φ₁ and <20% for φ₂.

Effects of NC on the dynamics of the 2Ap-labeled hexanucleotides

Our objective was to characterize site-selectively the interaction of NC with the AACGCC and AATGCC sequences, by using 2Ap-labeled ODNs. However, this approach is only valid if the substitution of a natural base by 2Ap does not modify the interaction of NC with the ODNs. To check this, we compared the binding of the 2Ap-labeled sequences with the corresponding unlabeled sequences. The binding parameters were determined by monitoring the Trp emission at 320 nm (where 2Ap does not fluoresce) in the presence of increasing ODN concentrations in 50 mM HEPES, pH 7.5 (Figure 4). Due to the limited sensitivity of Trp fluorescence that does not permit to work at very low protein concentrations, binding experiments were performed with the NC(11-55) peptide that binds in the same way than the native NC, but with a lower affinity (29,40). K_app values of 4(±1) × 10⁶ M⁻¹, 9(±4) × 10⁶ M⁻¹ and 9(±3) × 10⁶ M⁻¹ were found for AApCGCC, AACGApC and AACGCC, respectively. For the two last sequences, similar affinities were obtained by monitoring the changes in the 2Ap fluorescence when adding increasing peptide concentrations (data not shown). Accordingly, no change in affinity accompanied the C→2Ap substitution at position 5. This result may be rationalized from the NMR data of the ACGCC/NC complex showing that NC binds the corresponding C residue only through a single H-bond with its N4 atom (11). It is thus likely that the N7 atom of 2Ap forms the same H-bond with NC since it is located virtually at the same place than the N4 atom in cytosine relatively to the sugar-phosphate axis (11). For AApCGCC, a limited affinity decrease was observed in respect with the native ODN. Since NMR showed no direct interaction between the corresponding A residue with NC, 2Ap may indirectly decrease the binding constant, by decreasing for instance the interaction of the neighbor C residue with NC. Since AATGCC was shown to bind NC much stronger than AACGCC (29), we used a buffer containing 0.1 M NaCl to get reliable binding constants with this ODN. No decrease of NC affinity for AATGApC appeared (K_app = 5(±3) × 10⁶ M⁻¹) as compared to AATGCC (K_app = 4(±1) × 10⁶ M⁻¹) whereas the affinity for AApTGCC (K_app = 8(±2) × 10⁶ M⁻¹) was slightly higher. As a consequence, 2Ap does not perturb significantly the interaction of the studied ODNs with NC and thus can be reliably used to probe the NC/ODN complexes.

Figure 4.

Effect of 2Ap substitution on the binding of NC(11-55) to AACGCC. The concentration of NC(11-55) was 1 μM in 50 mM HEPES, pH 7.5. The concentration of AApCGCC (filled circles), AACGApC (filled triangles) and AACGCC (filled squares) is expressed in strands. Solid lines correspond to the fit of the experimental points with Equation (1) and the parameters are given in the text.

Binding of NC was found to strongly increase the quantum yield of 2Ap within all species, although to different extents: from 2.5-fold for AApCGCC to more than 7-fold for AACGApC (Figure 2 and Table 1). Meanwhile, the position of the emission maximum did not change (Figure 2). Thus, NC binding significantly reduces the quenching of 2Ap fluorescence by its neighbor bases but does not change the polarity of its surrounding. As in the absence of NC, the time-resolved intensity decay of 2Ap within the ODN/NC complexes needed four lifetimes for good fitting. In the complexes, the 2Ap fluorescence decayed slower than in the absence of NC, with an increase in the mean lifetime by a factor of about two (Table 1). Significant differences appear as a function of the labeled position and the ODN nature. For AApCGCC, the increase in the mean lifetime is fully consistent with the quantum yield increase, indicating that NC negligibly affects α₀ in this case. In fact, the NC-induced increase in both the quantum yield and mean lifetime of this ODN is essentially due to an increase of the α₄ and τ₄ values. Since the τ₄ value approaches the lifetime of free 2Ap, unstacked conformers are probably driven less rapidly than in the free ODN to quenched conformers by conformational fluctuations. For AACGApC and the 2Ap-labeled TG-containing sequences, the NC-induced increase in quantum yield was found to be much higher than the increase in the mean lifetime, suggesting that NC dramatically decreases the α₀ value and thus, the ultra-fast dynamic quenching of 2Ap in these ODNs. This decrease in α₀ is accompanied by a large increase of the α₂ to α₄ values as well as an increase of the τ₃ and τ₄ values. Consequently, NC shifts the equilibrium of the 2Ap sub-populations to the less quenched ones and decreases the quenching efficiency in the latter. Interestingly, the effect of NC appears quite homogeneous since NC increases the summed populations (α₃ + α₄) of the two less quenched 2Ap conformers from about 10% to 35–40% in all three ODNs.

The effect of NC was found to be quite similar on the time-resolved anisotropy decays of the various 2Ap-labeled ODNs (Figure 3 and Table 2). In all cases, NC induces a striking increase of the φ₂ component, in line with the formation of NC/ODN complexes. The measured φ₂ values are significantly higher than the theoretical 3.3 ns value calculated for the tumbling of a 1:1 complex with a spherical shape. Thus, the shape of the complexes likely deviates from a sphere and, in contrast to the free ODNs, the φ₂ component only describes the tumbling motion of the complex, with no contribution of segmental motions. As a consequence, the binding of NC reduces the degrees of freedom of the ODNs. This conclusion is further substantiated by the decrease of the amplitude associated with the local motion. This decrease corresponds to a reduction by 6° to 11° of the cone semi-angle in which 2Ap rotates. Moreover, NC was found to also increase the value of the φ₁ component for 2Ap at position 5, but not in position 2, indicating that the binding of NC directly altered the flexibility of the 2Ap base at position 5. As in the free ODNs, the conclusions on the local mobility of 2Ap in the complexes probably concern mainly the less quenched subpopulations that contribute strongly to the total fluorescence intensity.

Effects of NC variants on the local dynamics of the 2Ap-labeled hexanucleotides

To determine the contributions of the N-terminal basic domain, the folded zinc fingers and the conserved aromatic residues (Phe¹⁶ and Trp³⁷) to the effects of NC on the dynamics of the 2Ap-labeled ODNs, we analyzed the fluorescence properties of these ODNs complexed with various NC mutants (Figure 1). The contribution of the N-terminal domain was investigated with the NC(11-55) mutant which includes the zinc finger domain but lacks the N-terminal domain. The contribution of the folded fingers was analyzed with the (SSHS)₂NC(11-55) mutant where all cysteines are substituted for serines, which prevents the binding of zinc and therefore the folding of the peptide (8). The contributions of the conserved Phe¹⁶ residue in the proximal finger and the conserved Trp³⁷ residue in the distal finger were investigated by using the A¹⁶NC(11-55) mutant with Phe¹⁶ replaced by Ala and L³⁷NC(11-55) with Trp³⁷ replaced by Leu, respectively. Since both Phe¹⁶ and Trp³⁷ residues do not participate to zinc binding (41), their substitution is expected to not alter the zinc-driven folding of the A¹⁶NC(11-55) and L³⁷NC(11-55) mutants. In contrast, since both residues participate to the hydrophobic plateau at the top of the folded zinc fingers, substantial alterations in this plateau should occur, likely explaining the significant alterations of the binding of the corresponding NC mutants to their ODN targets (7,40,42).

Both the steady-state and time-resolved fluorescence parameters of the complexes of NC(11-55) with AApCGCC, AACGApC and AATGApC were similar to those of the corresponding complexes with the native NC, suggesting that the NC-induced changes in the dynamics of these 2Ap-labeled ODNs are mediated by the zinc finger domain. In contrast to the aforementioned ODNs, an ∼2-fold decrease in the quantum yield and the amplitudes associated with the τ₃ and τ₄ lifetimes was observed when NC was substituted with NC(11-55) in its complex with AApTGCC. Thus, the N-terminal basic domain of NC likely contributes to the NC-induced changes of the local dynamics of 2Ap in this ODN (Table 1). Noticeably, the φ₂ correlation time of all tested ODNs was substantially lower in their complexes with the truncated mutant than with the native NC. This may be rationalized by the smaller molecular weight of the truncated peptide and probably, the more spherical shape of its complexes with the ODNs (Table 2).

In sharp contrast to NC and NC(11-55), the unfolded (SSHS)₂NC(11-55) mutant induced only limited changes in the steady-state and time-resolved parameters of the 2Ap-labeled ODNs. For instance, neither the quantum yield nor the α₀, α₁ and τ₁ values were significantly affected by (SSHS)₂NC(11-55), indicating that the peptide does not affect 2Ap in the most quenched conformers. In contrast, slight increases of about 0.3–0.6 ns and 1 ns were observed for the τ₃ and τ₄ values, respectively, indicating that (SSHS)₂NC(11-55) interacts with 2Ap in the less quenched conformers. The limited changes in the time-resolved intensity parameters were not due to poor binding, since this peptide was found to bind even stronger than NC(11-55) to the various target ODNs (data not shown), in line with previous data on other ODNs (43). Moreover, (SSHS)₂NC(11-55) increased the φ₂ correlation time of all 2Ap-labeled ODNs from 0.68–0.83 ns to 2–2.2 ns, confirming that this peptide binds to the ODNs. The correlation time of the complex is substantially lower than the 2.8 ns correlation time expected for the tumbling of a spherical 1:1 complex, indicating that the φ₂ component describes a combination of tumbling and segmental motions. Thus, (SSHS)₂NC(11-55) decreases the overall flexibility of the ODNs to a lesser extent than NC and NC(11-55).

Substitution of the Phe¹⁶ residue by Ala in the A¹⁶NC(11-55) mutant was found to decrease the affinity for the various ODNs by about 20-fold (data not shown), in line with a significant contribution of Phe¹⁶ in the binding process (42). Nevertheless, at the high concentrations needed for the time-resolved experiments, we calculated that more than 80% of the 2Ap labeled ODNs were bound to A¹⁶NC(11-55). Substitution of NC(11-55) by A¹⁶NC(11-55) only marginally altered the amplitudes associated with the different fluorescence lifetimes as well as the time-resolved anisotropy parameters of the ODNs labeled by 2Ap at the 5 position (Tables 1 and 2). The only significant change was the decrease of the τ₃ to τ₄ values, suggesting that A¹⁶NC(11-55) reduces less efficiently than NC(11-55) the quenching of 2Ap fluorescence by its neighbor bases in these conformations. As a consequence, the Phe¹⁶→Ala mutation induces only subtle changes in the interaction of NC with 2Ap at position 5. In sharp contrast, strong differences between NC(11-55) and A¹⁶NC(11-55) were observed when the two peptides interact with ODNs labeled at position 2. Indeed, the quantum yield and time-resolved intensity values of both AApCGCC and AApTGCC derivatives were much lower in the presence of A¹⁶NC(11-55) than in the presence of NC(11-55) and approached the values of the free ODNs, indicating that substitution of Phe¹⁶ with Ala prevented modifications of the local dynamics of 2Ap at position 2. However, the values of the longer correlation time of AApCGCC and AApTGCC bound with A¹⁶NC(11-55) and NC(11-55) were indistinguishable, indicating that both peptides decrease the overall flexibility of the ODNs to the same extent. Moreover, the two peptides provide similar β₁ values, suggesting that they similarly restrict the amplitude of the 2Ap motion.

Substitution of Trp³⁷ by Leu dramatically decreased (by two orders of magnitude) the binding constants of NC to the various ODNs, in line with the critical role of Trp³⁷ in nucleic acid recognition (6,10,11,29,44–48). As a consequence, only the more affine TG-containing ODNs could be used to obtain a sufficient level of binding (> 80%) in the time-resolved experiments. NC(11-55) and L³⁷NC(11-55) were found to substantially differ in their ability to alter the time-resolved fluorescence intensity parameters of the AATGApC sequence (Table 1). Indeed, the L³⁷NC(11-55) mutant induced about two times lower increase than NC(11-55) in the quantum yield of AATGApC, due to its lower efficiency in shifting the populations of the more quenched species toward the less quenched ones. Nevertheless, the L³⁷NC(11-55) mutant significantly increased the τ₃ and τ₄ values of AATGApC as well as their associated amplitudes, indicating that the mutated peptide was able to interact with the less stacked conformers, though less efficiently than the unmodified NC(11-55) peptide. In contrast to AATGApC, only limited differences in the fluorescence parameters of AApTGCC were observed in its complexes with NC(11-55) and L³⁷NC(11-55) (Table 1), suggesting that the Trp³⁷→Leu mutation affects essentially the ability of NC to perturb the local kinetics of 2Ap at position 5. This conclusion was further substantiated by the different effects of L³⁷NC(11-55) and NC(11-55) on the value of the short rotational correlation time of 2Ap at position 5, but not at position 2 (Table 2). In contrast to NC(11-55), the L³⁷ mutant only moderately slowed down the local motion of 2Ap at position 5.

DISCUSSION

In this study, we investigated the steady state and time resolved fluorescence properties of 2Ap-substituted AACGCC and AATGCC sequences and their complexes with native and mutated NC proteins. In the absence of NC, the complex intensity decays with four resolved lifetimes and one nonmeasurable ‘null’ lifetime suggest that 2Ap adopts at least five different conformations in these ODNs (18,49). Probably, even more geometries are explored but cannot be experimentally observed due to the limited resolution of the time-resolved measurements. The very low quantum yield as well as the time-resolved fluorescence parameters (and notably the absence of a free 2Ap-like component) of the 2Ap-labeled hexanucleotides are reminiscent of those previously observed with 2Ap-labeled trinucleotides (19). In further analogy with the latter sequences, a dramatic fluorescence increase of the 2Ap-labeled hexanucleotides was observed in a viscous medium, due to a strong decrease of the population of the most quenched conformers and a corresponding increase of the population of the less quenched conformers (Table 1). The fluorescence quenching of 2Ap appears thus highly dynamic and results probably from a charge transfer mechanism occurring during its collisions with its neighbors (19,20). This mechanism is in line with the strong quenching observed when 2Ap is next to a guanine, since this last residue shows the lowest redox potential among the bases (21,22,39). These fluorescence properties of the labeled hexanucleotides largely rely on their highly flexible structure, which allows conformational fluctuations in the picosecond–nanosecond (ps–ns) range. The high flexibility of the ODNs is substantiated by the large amplitude associated with their short correlation time as well as by the low value of their long rotational correlation time, showing strong contributions of both local and segmental motions that prevent observation of the tumbling motion. In this respect, even if unstacked 2Ap conformations are excited, rapid conformational changes will bring 2Ap close to one of its neighbor, allowing charge transfer and thus, fluorescence quenching. Moreover, the ‘null’ lifetime likely corresponds to a conformation where 2Ap stacks with one of its neighbors, leading to an almost immediate quenching.

Binding of NC induces a strong decrease of the ODN flexibility, with notably a disappearance of the segmental motions, as shown by the values of the long rotational correlation times (Table 2). This decrease in ODN flexibility is consistent with the well-resolved structure of ACGCC in its complex with NC(12-53) (11) as well as with the ‘freezing’ by NC of the nanosecond dynamics of a dsDNA (50). The effect of NC on ODN flexibility is in line with an ‘adaptive binding’ mechanism where a flexible nucleic acid sequence becomes more ordered upon binding of a protein (51,52). As a consequence of this decreased flexibility, collisions of 2Ap with its neighbors needed for fluorescence quenching are restricted, explaining the increased quantum yields and the shift toward the less quenched populations (associated with the longer lifetimes) in the presence of NC. In addition, NC also affects the local motion of 2Ap at positions 2 and 5 by decreasing its excursion angle, in full line with the NMR structure of the ACGCC/NC(12-53) complex showing that the local motion of the corresponding bases in ACGCC is constrained by the Asn¹² and Gln⁴⁵ residues, respectively (Figure 5). Though the effect of NC on the fluorescence properties of the 2Ap-labeled ODNs is qualitatively similar to that of glycerol, it differs by an important aspect. Indeed, glycerol induces similar increases of the 2Ap quantum yield in the four ODNs due to a mechanical decrease of the number of collisions of 2Ap with its neighbors by viscosity. In contrast, NC differently affects the dynamics of the bases at positions 2 and 5, since the quantum yield of 2Ap in AApCGCC increases by only 2.5-fold while it increases by about 7-fold in AACGApC. Moreover, NC does not affect the local motion rate of 2Ap at position 2 while it slows down the local motion of the probe at position 5. This slowing down may be due to an interaction of 2Ap at position 5 with the side chain of Gln⁴⁵ (Figure 5) since a H-bond was reported between this Gln⁴⁵ side chain and the corresponding C residue in the ACGCC/NC(12-53) complex (11). Alternatively, the effect on the local motion of 2Ap may also be an indirect consequence of the interaction of its neighbors with NC. In this respect, the strong fluorescence increase of 2Ap at position 5 (giving a fluorescence quantum yield that is only five times lower than that of the free ODN) is likely a consequence of the strong stacking of G4 with the Trp³⁷ residue that freezes the mobility of G4 and thus restricts its collisions with 2Ap (11,29). Moreover, this stacking also rotates the guanine residue in respect to the phosphate ribose axis, further preventing its collisions with 2Ap. In contrast, the stacking of C3 with Phe¹⁶ appears less stable and constrained since two orientations of this couple were found in the complex. Moreover, the A1 residue likely protrudes out of the complex and is thus quite mobile. Thus, both C3 and A1 residues are still able to collide with 2Ap in this complex, explaining its efficient quenching.

Figure 5.

Structure of the NC(12-53)/ACGCC complex. NC protein backbone, blue ribbon; Phe¹⁶ and Trp³⁷ residues, magenta. Zinc atoms are as gray spheres. The side chains of Asn¹² and Gln⁴⁵ are represented in CPK color code. The A (yellow) and C (green) residues of the ODN which have been substituted by 2Ap in our study are in bold. The G residue that interacts with Trp³⁷ is in pink.

The fluorescence signal of 2Ap can also be used to characterize the complex of AATGCC with NC. From the strong similarities of the steady state and time-resolved fluorescence parameters of AACGApC and AATGApC complexed with NC, it can be inferred that G4 stacks similarly with the Trp³⁷ residue in both complexes. More generally, since the 3′ half of the ODNs interacts mainly with the distal finger of NC, this interaction is likely similar in both complexes. In contrast, the stronger NC-induced increase of 2Ap fluorescence in AApTGCC as compared to AApCGCC suggests that the flexibility of the 5′ half of the ODN is more strongly decreased by NC in the former sequence. This observation may be related to a stronger interaction of T3 as compared to C3 with NC, due to a stable interaction of the methyl group of T3 with the aromatic protons of Phe¹⁶ as well as with the methyl groups of Val¹³ as Thr²⁴ (N. Morellet, personal communication). This stable interaction with T3 may strongly hinder its collisions with 2Ap. Alternatively, the binding of T3 to NC may favor a direct interaction of NC with 2Ap, preventing its collisions with neighbor bases. However, the absence of change in the rate of the 2Ap local motion does not favor this hypothesis. The more ‘frozen’ complex obtained with the TG-containing sequence is in line with its stronger binding constant as compared to the CG-containing sequence (29).

Comparison of the native NC with the truncated NC(11-55) derivative indicated that the finger domain plays a central role in restricting the ODN flexibility and local dynamics of 2Ap at positions 2 and 5. In AApTGCC, both the ODN flexibility and 2Ap local dynamics are further restricted by the terminal basic domain of NC (Table 1). The stronger dependence of the fluorescence parameters of AApTGCC as compared to AApCGCC on the N-terminal domain suggests that this domain interacts tighter with 2Ap at position 2 in the former ODN. This tight interaction is likely a consequence of a different binding mode of the proximal NC finger to AApTGCC that orientates differently the N-terminal domain.

In sharp contrast to NC(11-55), (SSHS)₂NC(11-55) minimally affects the fluorescence parameters of the four tested 2Ap-labeled ODNs. This does not result from a decreased binding level since this mutant was shown to bind to various ODNs with high affinity (40,46). The poor effect of (SSHS)₂NC(11-55) is likely a consequence of its inability to restrict the ODN flexibility and 2Ap local dynamics. As a result, the distribution of the 2Ap conformers and the quenching of 2Ap by its neighbors are similar to those in the free ODN. Thus, the folding of the finger domain appears critical for restricting the ODN flexibility and 2Ap local dynamics. This crucial role of folding is likely related to the hydrophobic platform (formed by the Val¹³, Phe¹⁶, Thr²⁴, Ala²⁵, Trp³⁷ and Met⁴⁶ residues) at the folded finger surface that plays a key role in ODN binding (10,11,40,53) and specific structural changes (10–12). In line with this hypothesis, only a limited stacking of Trp³⁷ with the ODN bases was observed in the complexes with (SSHS)₂NC(11-55), as shown by the limited decrease of Trp³⁷ fluorescence (data not shown). This explains that the G4 residue is not ‘frozen’ by stacking with Trp³⁷ and can thus efficiently quench 2Ap at the position 5 in the complex. Stacking interactions with Phe¹⁶ may be prevented as well in the complexes with (SSHS)₂NC(11-55) since the interaction of Phe¹⁶ with ACGCC has been shown to be largely altered with a NC mutant exhibiting a distorted proximal finger (11). The limited restriction of the mobility of the ODN bases in the complexes with (SSHS)₂NC(11-55) is in line with a binding of the unfolded peptide mainly to the phosphate backbone. This binding is likely mediated mainly through electrostatic interactions that compensate for the loss of hydrophobic interactions with the bases (40,46). Since unfolded NC peptides are unable to destabilize the secondary structure of cTAR (40) and alter the cooperativity of the helix–coil transition of λ DNA molecules (54), the restriction of the ODN flexibility and local dynamics of the bases is certainly essential for the destabilizing component of NC chaperone activity. In addition, since mutations that prevent the folding of the finger domain lead to completely noninfectious viruses (6–9), ‘freezing’ of the ODN dynamics may be a key feature of NC during the viral life cycle.

The relationship between the restriction of the ODN dynamics and the biological activities of NC was further substantiated with the A¹⁶NC(11-55) and L³⁷NC(11-55) mutants. Both mutations led to fully noninfectious viruses (6), likely by altering the nucleic acid binding and chaperone properties of NC (40,42,46,48). With A¹⁶NC(11-55), the local dynamics of 2Ap at position 5 and its close neighbors was restricted to the same level than with NC(11-55) while in contrast, the local dynamics of 2Ap at position 2 in the presence of A¹⁶NC(11-55) was comparable to that in the free ODNs. Accordingly, the distal finger of A¹⁶NC(11-55) likely interacts with the 3′ half of the ODN sequences in the same way than the unmodified NC(11-55). In contrast, the Phe¹⁶→Ala mutation deeply modified the interaction of the proximal finger with the 5′ half of the ODN sequences, probably due to the inability of Ala¹⁶ to interact with C3 and prevent its rapid collisions with 2Ap at position 2. Symmetric effects were observed with the L³⁷NC(11-55) mutant, which showed a strongly reduced ability to prevent quenching of 2Ap at position 5. This indicates that Leu³⁷ is probably unable to interact with the G4 base and thus cannot prevent the fast collisions of this residue with 2Ap at position 5. In addition, the change induced by L³⁷NC(11-55) in the rate of the 2Ap local motion further suggests that in contrast to the unmodified peptide, the Gln⁴⁵ residue of L³⁷NC(11-55) does probably not interact with 2Ap at position 5 in the complex. Nevertheless, in contrast to (SSHS)₂NC(11-55), both A¹⁶NC(11-55) and L³⁷NC(11-55) mutants restrict the overall ODN flexibility, as shown by the absence of any contribution from segmental motions in the long rotational correlation time of their complexes with the 2Ap-labeled ODNs (Table 2). Accordingly, the complexes with these two mutants are probably stabilized through hydrophobic interactions between the correctly folded fingers of the mutants and the ODN bases. Taken together, the data with the A¹⁶NC(11-55) and L³⁷NC(11-55) mutants unambiguously confirmed that the aromatic residues of the hydrophobic plateau at the top of the folded fingers play a key role in the nucleic acid recognition and the restriction of the ps–ns local dynamics of the nucleotide subdomains to which they bind.

In conclusion, 2Ap fluorescence allowed us to characterize sub-nanosecond and nanosecond dynamics of NC target hexanucleotides and their complexes with NC. The effects of NC binding on 2Ap fluorescence are explained by restriction of the ODN flexibility and 2Ap local mobility that impedes collisions of 2Ap with neighbor bases and thus, largely decreases its dynamic quenching. This is consistent with the ‘adaptive binding’ model of protein–nucleic acid interaction where the conformational freedom of single-stranded ODNs becomes restricted due to protein binding. The ‘freezing’ of the ODN flexibility seemed to be mainly supported by the folded zinc finger domain. The restriction of the local dynamics of the bases was more specifically attributed to the hydrophobic platform at the top of the folded fingers and notably, to the stacking of Trp³⁷ with the G base and the interaction of Phe¹⁶ with the preceding C base. Since this hydrophobic plateau also supports the NC chaperone properties and notably the ability of NC to destabilize the stem of various stem-loops (40), the restriction of the ps–ns ODN dynamics is likely a mechanistic component of these properties. In this respect, together with the increase in the distance and the rotation of the consecutive bases which interact with Phe¹⁶ and Trp³⁷ (10,11), the restriction of the local motion of these bases and their close neighbors may induce the disruption of base pairs in the double strand segments destabilized by NC (32,55–59). Thus, the NC-induced distortion of the ODN structure and local ‘freezing’ of the bases likely constitute the initial events responsible notably for the NC-chaperoned opening/closing of the cTAR DNA stem (56,60–62). Moreover, the NC-induced conformational changes and restriction of the local base motion probably favor the interaction of the loops of stem-loop sequences with their complementary sequences, explaining the loop–loop promoted annealing of (–)PBS/(+)PBS (63) and mini-TAR sequences (64) as well as the NC-promoted formation of PBS (65,66) and DIS (67) kissing loop homodimers. These hypotheses on the dependence of NC chaperone properties on the NC-promoted changes in the local dynamics of various 2Ap-labeled stem-loop sequences are currently investigated. Finally, since the NC-induced changes in the 2Ap fluorescence parameters were shown to depend on the position and the context of the probe, 2Ap can be used to site-specifically report on the binding of NC to more complex sequences.