A single zinc finger optimizes the DNA interactions of the nucleocapsid protein of the yeast retrotransposon Ty3.

Reverse transcription in retroviruses and retrotransposons requires nucleic acid chaperones, which drive the rearrangement of nucleic acid conformation. The nucleic acid chaperone properties of the human immunodeficiency virus type-1 (HIV-1) nucleocapsid (NC) protein have been extensively studied, and nucleic acid aggregation, duplex destabilization and rapid binding kinetics have been identified as major components of its activity. However, the properties of other nucleic acid chaperone proteins, such as retrotransposon Ty3 NC, a likely ancestor of HIV-1 NC, are not well understood. In addition, it is unclear whether a single zinc finger is sufficient to optimize the properties characteristic of HIV-1 NC. We used single-molecule DNA stretching as a method for detailed characterization of Ty3 NC chaperone activity. We found that wild type Ty3 NC aggregates single- and double-stranded DNA, weakly stabilizes dsDNA, and exhibits rapid binding kinetics. Single-molecule studies in the presence of Ty3 NC mutants show that the N-terminal basic residues and the unique zinc finger at the C-terminus are required for optimum chaperone activity in this system. While the single zinc finger is capable of optimizing Ty3 NC's DNA interaction kinetics, two zinc fingers may be necessary in order to facilitate the DNA destabilization exhibited by HIV-1 NC.

INTRODUCTION

Retrotransposons are mobile genetic elements, closely related to retroviruses, which use their own proteins in conjunction with cellular machinery to replicate independently of the genome. The yeast retrotransposon Ty3, and retroviruses such as human immunodeficiency virus type-1 (HIV-1), encode reverse transcriptase (RT), a DNA polymerase that converts the single-stranded RNA (ssRNA) genome of positive polarity into double-stranded DNA (dsDNA). During this reverse transcription process, two obligatory DNA strand transfers occur to generate the long terminal repeats (LTR) that control genomic DNA integration and its expression (1). In the early stages of replication, a tRNA primer anneals to the primer binding site (PBS) close to the 5′-end of the ssRNA, where RT synthesizes the complementary DNA (cDNA) strand, referred to as minus strand strong stop DNA (-sssDNA). While RT synthesizes -ssDNA, its RNaseH domain degrades the 5′-end of the RNA template. Further DNA polymerization requires -sssDNA transfer, where the newly synthesized single-stranded cDNA anneals to the 3′-untranslated terminal region (UTR) of the RNA template (1–3). However, thermodynamically stable structures within the cDNA and the 3′-UTR RNA make duplex formation improbable. The nucleocapsid (NC) protein acts as a nucleic acid chaperone that destabilizes these secondary structures and facilitates DNA–RNA hybrid formation (4–8). NC also participates in the second DNA strand transfer, where the plus DNA strand anneals to the structured PBS region of the minus DNA strand (9–12). These transfer events are required for RT to transcribe the remaining genomic RNA into dsDNA flanked by the LTR, which the homologous enzyme integrase inserts into the cellular genome.

Replication of retroelements requires the nucleic acid chaperone activity of NC proteins, which tend to be small cationic proteins with minimal structure other than one or two CCHC-motif zinc fingers (1–3,13–17). HIV-1 NC is 55 amino acids in length, with two zinc fingers and a basic N-terminal tail (1–3). It directs annealing of the tRNAiLys primer to the PBS of the genomic RNA (5,18,19). The viral RNA has 5′ and 3′ repeat regions that include stable trans-activation response element (TAR) hairpins, and HIV-1 NC facilitates rearrangement of these nucleic acids during the minus strand transfer step of reverse transcription (10–12). NC is also an essential nucleic acid chaperone involved in genomic RNA dimerization and virus assembly (20–24). The chaperone activity of HIV-1 NC involves nucleic acid aggregation, duplex destabilization and rapid binding kinetics (25–30). Nucleic acid aggregation, an effect due to protein-induced interactions that make the DNA molecule attracted to itself, is associated with the cationic domain of HIV-1 NC. Duplex destabilization involves aromatic residues in the zinc fingers, which stack with ssDNA bases (31,32). Thus optimal chaperone activity of HIV-1 NC is highly sensitive to the zinc-finger architecture, and relies on a delicate balance between destabilizing secondary structures and promoting complementary strand annealing.

A recent study examining the nucleic acid chaperone activity of NC proteins from several retroviruses established that the nucleic acid interaction characteristics that are important for nucleic acid chaperone activity varied significantly for different viruses (33,34). From these studies, it is clear that HIV-1 NC has optimal chaperone activity compared to other NC proteins. Although nucleic acid chaperones from HIV-1 and other retroviruses have been extensively studied, less is known about chaperone activity in LTR retrotransposons (4,35,36). Since Ty3 NC is a likely ancestor of HIV-1 NC, here we examine its nucleic acid chaperone properties and compare them to those of HIV-1 NC. In addition to allowing comparison of Ty3 NC to the paradigm of nucleic acid chaperone proteins, understanding these differences may also be useful for the development of drugs that target the NC protein from the rapidly evolving HIV-1 retrovirus (37–39). Ty3 NC is a 57-residue protein with one zinc finger and a basic N-terminal tail (40). Ty3 NC is monomeric in solution, and it is required for efficient annealing of the tRNAiMet primer on the bipartite Ty3 PBS and RNA dimerization (41). The UTR regions of the Ty3 genome also form secondary structures, and minus strand transfer requires Ty3 NC to overcome the energetic barrier in forming the DNA–RNA duplex necessary for further reverse transcription.

There are distinct structural differences between Ty3 NC, which has only one zinc finger with two aromatic residues, and HIV-1 NC, which has two zinc fingers with one aromatic residue each. The cationic tail of Ty3 NC, which has minimal structure, also contains an aromatic residue (tyrosine). Ty3 NC, with 20 basic residues (pI ≈ 11.5), has a slightly higher overall charge than HIV-1 NC (pI ≈ 10), with 15 basic residues. Although minimal information is available about the 3D structure of Ty3 NC, it is extremely likely to have high charge density, consistent with retroviral NC proteins of known structure such as HIV-1 NC and Moloney murine leukemia virus (MuLV) NC (42). Furthermore, the length and structural complexity of the UTR regions seem to be related to the chaperone activity of a number of retroviral NC proteins, and the secondary structures in the Ty3 UTR regions are not as thermodynamically stable as the HIV-1 TAR hairpins (25).

DNA stretching and other biophysical methods have been used to show that the properties of HIV-1 NC are specifically tuned to optimize its interactions with both dsDNA and ssDNA. Specifically, both zinc fingers are required to maintain their native order and structure, and seemingly minor mutations result in loss of rapid kinetics of the DNA–protein interaction (26). Other retroviral NC proteins exhibited less rapid kinetics in DNA stretching experiments, particularly MuLV and human T-cell lymphotropic virus type 1 (HTLV-1) NC (33,34). Notably, MuLV NC is the only NC protein studied that contained a single zinc finger, and it exhibited relatively slow kinetics. Thus, previous results suggest that two zinc fingers are generally required to optimize the kinetics of NC–DNA interactions (26,27,33,34). To determine the capability of a single zinc finger to facilitate NC–DNA interactions and to understand the connection between Ty3 NC's structure and function as a nucleic acid chaperone, we probed the thermodynamics and kinetics of wild type and mutant NC–DNA interactions with single DNA molecules. We found that, despite the presence of only a single zinc finger, wild-type Ty3 NC aggregates nucleic acids, promotes complementary strand annealing, and exhibits rapid kinetics resembling that of HIV-1 NC. Four mutants of Ty3 NC demonstrate that these chaperone properties require the zinc finger as well as the presence of its cationic N-terminal tail. Thus, Ty3 NC's single zinc finger is necessary and sufficient to facilitate rapid DNA–protein interactions, but a second zinc finger may be required for more effective nucleic acid destabilization by NC proteins, such as that observed for HIV-1 NC and TAR hairpins.

MATERIALS AND METHODS

Single-molecule DNA stretching experiments were performed as previously described (35,43). Dual beam optical tweezers were used to capture biotin-labeled bacteriophage λ DNA between two streptavidin-coated polystyrene beads (Bangs Labs). Surrounding DNA molecules were rinsed out of solution with buffer (50 mM Na+, 10 mM HEPES, pH 7.5), and the captured DNA molecule was then stretched and released at a pulling rate of 100 nm/s to obtain DNA-only force-extension curves (Figure 1a, black), which are typically characterized by models of polymer elasticity. The worm-like chain (WLC) model describes dsDNA in terms of length b at a given force F: where Pds is the persistence length, Bds is the contour length and Sds is the stretch modulus. The freely jointed chain (FJC) model describes ssDNA in terms of length bss at a given force F:

Figure 1.

(a) Typical force-extension (solid) and release (dashed) curves of λ-DNA (black) obtained with optical tweezers. The WLC model (blue line) describes dsDNA. Near the dsDNA contour length, the molecule undergoes a force-induced melting transition, from dsDNA to ssDNA. The FJC model describes ssDNA (red line). Minimal hysteresis is evident in these solution conditions (50 mM Na+, 10 mM HEPES, pH 7.5). (b) Quantification of the hysteresis area ratio for a typical DNA extension and release curve. Force–extension (solid) and release (dashed) curve of DNA in the presence of Ty3 NC Δ2-NCp9 dd are shown in green. The WLC and FJC models are shown in blue and red, respectively. A linear combination of these two models is shown in black, indicating the fraction of ssDNA exposed to solution upon DNA extension. Relative hysteresis is the ratio of A1, the area between the stretch (solid green) and release (dashed green) curves, and A2, the area between the stretch (solid green) and melted DNA fraction (black) curves. (c) Equilibrium dissociation constant Kd determined from change in average melting force ΔFm as a function of protein concentration c fit to a simple DNA binding isotherm [Equations (3) and (5)]. Data points for mutant Δ2-NCp9 dd are shown with standard error bars, with a fit (blue line) that yields Kd = 20 (± 1) nM and saturated melting force ΔFmsat = 16 (± 0.5) pN. Kd was estimated for mutants Δ1-NCp9 and NCp9 dd with this method, but could not be obtained for mutant Δ2-NCp9, which did not affect DNA melting force.

Figure 1a shows the WLC (blue) and FJC (red) polymer models with typical parameter values [Bds = 0.34 nm/bp, Pds = 48 nm and Sds = 1200 pN in Equation (1), Bss = 0.55 nm/bp, Pss = 0.75 nm and Sss = 720 pN in Equation (2)] After obtaining force-extension curves of a single DNA molecule, the buffer was replaced with protein dilutions. Ty3 NC and its mutants were introduced into solution with the DNA held under slight tension in an effort to reduce formation of extremely stable protein–DNA aggregates during solution exchange. The first force-extension curve after protein exchange was therefore stretched from zero extension to the original starting extension after DNA release (example in Figure 1b, solid green curve, starting force of 10 pN). Subsequent force-extension curves begin and end at zero extension. At least three stretch–release cycles were performed for each measurement, and subsequent force-extension curves were the same within error in the case of wild type Ty3 NC. The 50 mM Na+ concentration was used for comparison with earlier studies of other NC proteins. Both the salt and protein concentration in vivo are higher than those used here, but the increased binding in low salt tends to compensate for the lower protein concentration.

Melting forces were determined by averaging along the length of the force-induced melting plateau in the presence of each protein. Hysteresis was quantified by calculating A1, the area between the force-extension (Figure 1b, solid green) and release (Figure 1b, dashed green) curves of DNA in the presence of wild-type Ty3 NC and its mutants. A linear combination of the WLC and FJC models (Figure 1b, blue and red, respectively) indicates the fraction of ssDNA generated upon DNA extension (Figure 1b, black), as described in (44). The total amount of hysteresis possible is A2, the area between the extension curve (Figure 1b, solid green) and this linear combination (Figure 1b, black). Relative hysteresis is the ratio of A1 and A2, ranging from 0.1 (±0.03) in the absence of protein, to a theoretical maximum of 1, which would indicate that all the ssDNA generated is bound by protein upon DNA release and does not dissociate on the time scale of the release (∼1 min). Averages and uncertainties (standard error) for all values reported were calculated using at least three measurements.

To quantify the effect of wild-type Ty3 NC on the shape of the force-induced melting plateau, transition width δF was determined from the slope at the midpoint of melting transition, as described in (27,45). The change in transition width is ΔF = δF – δF0, where δF0 is the transition width in the absence of protein. Change in transition width as a function of protein concentration c may be described by a simple DNA binding isotherm (46): where Kd is the equilibrium dissociation constant and the fractional DNA binding θ is given by the change in transition width: with δF defined as the transition width at saturated protein concentration. Standard error for measured data points was calculated from at least three measurements, and fits to Equations (3) and (4) were performed using χ2 analysis.

Ty3 NC mutants have negligible effect on the transition width, so Kd was obtained from their effect on the average melting force Fm, as reviewed in (43). Change in melting force is ΔFm = Fm – Fm, where Fm = 61.0 (0.5) pN, the melting force in the absence of protein. ΔFm as a function of protein concentration c may be described by Equation (3), where the fractional binding θ is given by the change in melting force: with F defined as the average melting force at saturated protein concentration. Data points with standard error were fit with Equations (3) and (5) to quantify Kd for mutant Δ2-NCp9 dd (Figure 1c). This method was used to estimate Kd for mutants Δ1-NCp9 and NCp9 dd, but mutant Δ2-NCp9 did not affect melting force over two orders of magnitude change in protein concentration.

Ty3 NC protein of 57 amino acids and its mutants were synthesized on the solid phase using fmoc (9-fluorenylmethoxycarbonyl)- and opfp (pentafluorophenyl ester)-protected amino acids and purified to homogeneity by HPLC (47).

RESULTS

Wild-type Ty3 NC binds DNA with rapid kinetics and slightly stabilizes dsDNA

Optical tweezers were used to stretch and release a single molecule of bacteriophage λ DNA in the absence of protein and in the presence of a nearly saturated concentration of wild type Ty3 NC (Figure 2a). While bacteriophage λ DNA is not the sequence acted on by Ty3 NC, it represents a random sequence, allowing us to measure the ability of the protein to remodel general nucleic acid structures. In the absence of protein, B-form dsDNA uncoils with a gradual increase in force at low extensions. As the molecule approaches its contour length of 0.34 nm/bp, the elasticity of the backbone causes a sharp increase in force. At ∼60 pN, the DNA molecule lengthens more than 1.5 times with minimal increase in force as it undergoes a cooperative force-induced melting transition from dsDNA to ssDNA (48–50). The force increases rapidly again at the end of this cooperative phase transition, at 0.6 nm/bp. As the DNA is released back to low extensions, the force-extension curve is almost completely reversible, exhibiting only minor hysteresis, which is the difference between the extension and release curves. The area between these curves characterizes the amount of hysteresis.

Figure 2.

(a and b) Typical force-extension (solid) and release (dashed) curves of λ-DNA in the presence of wild type Ty3 NC. (a) DNA only (black) and 5 nM wild type Ty3 NC (green). (b) 5 nM (green) and 25 nM (blue) wild type Ty3 NC. (c) Change in the transition width ΔF of DNA force-induced melting as a function of wild type Ty3 NC concentration. ΔF = δF – δF, where δF is the melting transition width in the presence of protein and δF0 = 3.6 (±0.3) pN, the melting transition width of DNA only. Standard error determined from at least three measurements was used to compute error bars for ΔF. A χ2 fit (blue line) to a simple DNA binding isotherm [Equations (3) and (4)] yields K = 3.5 (±0.5) nM and ΔF = 6.2 (±0.4) pN. Protein concentrations significantly above saturation (80–150 nM) were also included in the χ2 fit (data not shown).

In the presence of 5 nM wild type Ty3 NC, the force increases at low extensions, below the λ DNA contour length (Figure 2a). The transition at the end of the melting plateau also shifts to lower extensions, decreasing by ∼0.025 nm/bp. These effects are due to nucleic acid aggregation, in which protein-induced interactions make the DNA molecule attracted to itself. A 5-fold increase in protein concentration leads to small increases in these two effects, reflecting additional nucleic acid aggregation (Figure 2b).

The force-induced melting plateau is slightly sloped in the presence of wild type Ty3 NC. This increase in transition width reflects a loss of DNA melting cooperativity, which indicates that the DNA molecule is more likely to undergo conformational rearrangements, an effect which has been observed for multiple nucleic acid chaperone proteins (27,34,45,46). A simple DNA binding isotherm [Equations (3) and (4)] fit to the change in transition width ΔF as a function of protein concentration c yields an equilibrium dissociation constant Kd = 3.5 (±0.5) nM and saturated change in transition width ΔFsat = 6.2 (±0.4) pN (Figure 2c, blue line).

The lack of hysteresis indicates that Ty3 NC exhibits rapid kinetics, allowing it to dissociate quickly from ssDNA. Rapid kinetics, which is characteristic of nucleic acid chaperones, is correlated with a protein's ability to facilitate re-annealing. In contrast, proteins that preferentially bind ssDNA, such as T4 gp32, induce significant hysteresis, and proteins that preferentially bind dsDNA, such as HMG, stabilize the duplex and increase the melting force (43). The melting force in the presence of wild type Ty3 NC is slightly higher than that of DNA only (Table 1), which indicates net stabilization of the DNA duplex. In order to elucidate how a protein that stabilizes duplex DNA can act as a nucleic acid chaperone, four mutants of Ty3 NC were used to further investigate the DNA binding role of the zinc finger and N-terminal tail (Figure 3). The N-terminal tail was deleted to varying degrees in mutants Δ1-NCp9 and Δ2-NCp9, leaving the zinc finger intact. In mutant Δ2-NCp9 dd, however, the zinc finger has also been deleted. Mutant NCp9 dd has a largely intact N-terminal tail, but no zinc finger.

Table 1.

Melting force and hysteresis area in the presence of wild type Ty3 NC and its mutants, measured at protein concentrations c (force-extension curves shown in Figures 2, 4–7), near or above Kd. Equilibrium dissociation constants Kd reported for 50 nM Na+

Ty3 NC protein	c (nM)	Fm (pN)a	Hysteresis (ratio)b	Kd (nM)c
Wild type	5	68.0 (±1.0)	0.26 (±0.03)	3.5 (±0.5)
Δ1-NCp9	20	59.4 (±0.7)	0.68 (±0.1)	12 (±3)
Δ2-NCp9d	50	60.6 (±0.1)	0.49 (±0.1)
NCp9 dd	3	71.5 (±0.4)	0.55 (±0.1)	3 (±2)
Δ2-NCp9 dd	13	67.1 (±0.5)	0.67 (±0.1)	20 (±1)

All values were calculated with at least three measurements, reflecting the uncertainty reported (standard error).

aThe melting force is an average along the length of the force-induced melting plateau in the presence of each protein. The melting force of DNA only in 50 mM Na+ is 61.0 (±0.5) pN.

bHysteresis is an area ratio that reflects amount of protein still bound upon DNA release (Figure 1b). The ratio increases with the amount of ssDNA observed upon DNA release, up to a maximum value of 1. The minimal hysteresis in the absence of protein yields an area ratio of 0.1 (±0.03).

Kd was quantified from the change in transition width in the case of wild-type Ty3 NC (Figure 2c). Change in melting force was used to determine Kd for Ty3 NC mutants (Figure 1c), with the exception of Δ2-NCp9, which did not affect melting force upon DNA binding.

dTy3 NC mutant Δ2-NCp9 did not appreciably bind DNA on the first stretch–release cycle (Figure 5a) over two orders of magnitude in protein concentration (1–100 nM). At protein concentrations 50 nM and above, binding that altered hysteresis was achieved only upon the third stretch–release cycle (Figure 2b and c) and this effect did not change significantly at higher concentrations. Therefore data from the third force-extension curve is included in Table 1 to provide a comparison of melting force and hysteresis ratio for this mutant.

Figure 3.

Structure of (a) wild-type Ty3 NC, (b) Δ1-NCp9, (c) Δ2-NCp9, (d) NCp9 dd and (e) Δ2-NCp9 dd. Basic residues shown in blue and zinc coordinating residues shown in green.

Figure 5.

Typical force-extension (solid) and release (dashed) curves of (a) DNA only (black) and in the presence of (a–c) 50 nM Ty3 NC mutant Δ2-NCp9. First stretch–release curve shown in green (a and b), second stretch–release curve shown in blue (b and c), and third stretch-release curve shown in red (c).

Cationic tail deletions destroy the rapid kinetics and duplex stabilization of wild type Ty3 NC

The N-terminal tail of mutant Δ1-NCp9 is missing the first 16 residues, compromising the protein's DNA binding affinity by ∼5-fold (Table 1). Kd was estimated as 12 (±3) nM using a simple DNA binding isotherm to approximate the decrease in melting force as a function of protein concentration [Equations (3) and (5)]. In striking contrast to wild type Ty3 NC, force-extension curves in the presence of 20 nM Δ1-NCp9 exhibit significant hysteresis (Figure 4a). The hysteresis ratio for Δ1-NCp9 is 0.68 (±0.1), relative to 0.26 (±0.03) for wild type (Table 1). This indicates that the mutant does not dissociate from ssDNA on the timescale of the experiment, leading to the loss of rapid kinetics. Subsequent stretches reflect incomplete protein dissociation (Figure 4b). Δ1-NCp9 induces less aggregation than wild type Ty3 NC, and its effect on the force-induced melting transition width is nominal. Furthermore, the melting force is 59.4 (±0.7) pN in the presence of the mutant, which is significantly lower than the 68.0 (±1.0) pN melting force in the presence of wild type Ty3 NC (Table 1). This result reflects weak duplex destabilization by Δ1-NCp9.

Figure 4.

Typical force-extension (solid) and release (dashed) curves of (a) DNA only (black) and in the presence of (a and b) 20 nM Ty3 NC mutant Δ1-NCp9. First stretch–release curve shown in green (a and b), and second stretch–release curve shown in blue (b).

To identify the DNA binding activity of the zinc finger, we examined the Δ2-NCp9 mutant. Deletion of the N-terminal tail leaves a mutant that primarily consists of the zinc finger, and has DNA binding affinity more than an order of magnitude smaller than that of wild type. Although Δ2-NCp9 does not affect DNA melting force or transition width to allow for an estimate of Kd, protein concentrations smaller than 50 nM reflected no DNA binding. In the presence of 50 nM protein, the first force-extension curve lies almost exactly on top of the DNA only curve, showing a negligible effect (Figure 5a). The second stretch–release cycle exhibits some hysteresis, indicating slow kinetics (Figure 5b), which increases further in the third force-extension curve (Figure 5c). The Δ2-NCp9 mutant is the only Ty3 NC protein for which successive stretch-release cycles show additional protein binding. The third stretch-release cycle is required to observe an effect comparable to the other proteins presented in Table 1 because Δ2-NCp9 binds DNA with significantly lower affinity than wild type or the other mutants. Furthermore, these effects remain the same within error, even up to 100 nM protein concentration. Therefore the hysteresis ratio of 0.49 (±0.1) is reported in Table 1 for comparison. There is less hysteresis than that for Δ1-NCp9 (Table 1), which is consistent with the likelihood that Δ2-NCp9 has higher charge density.

DNA stretching curves in the presence of mutants dominated by the zinc finger imply that the zinc finger of Ty3 NC contributes to nucleic acid aggregation and duplex destabilization. Both mutants with a compromised N-terminal tail also lack the rapid kinetics of wild type Ty3 NC, demonstrating that the cationic tail is required to promote nucleic acid annealing.

Zinc-finger deletions destroy rapid kinetics and induce strong duplex stabilization

To identify the DNA binding properties of the N-terminal tail, we examined two mutants that lack the zinc finger entirely, NCp9 dd and Δ2-NCp9 dd. The zinc-finger deletion in mutant NCp9 dd leaves the cationic tail nearly intact, resulting in DNA binding affinity most similar to that of wild-type Ty3 NC, with Kd estimated as 3 (±2) nM (Table 1). The initial force-extension curve in the presence of 3 nM NCp9 dd exhibits a significantly higher melting force of 71.5 (±0.4) pN, which reflects strong dsDNA stabilization (Figure 6a, Table 1). The second stretch exactly follows the previous release curve, indicating that the mutant binds ssDNA irreversibly on the timescale of the experiment (Figure 6b). NCp9 dd also induces strong DNA aggregation, indicated by an increase in the DNA stretching force to 10–20 pN below the DNA contour length, an effect that dominates subsequent stretches.

Figure 6.

Typical force-extension (solid) and release (dashed) curves of (a) DNA only (black) and in the presence of (a and b) 3 nM Ty3 NC mutant NCp9 dd. First stretch–release curve shown in green (a and b), and second stretch–release curve shown in blue (b).

The Δ2-NCp9 dd mutant is Δ2-NCp9 without the zinc finger. Deletion of both the N-terminal tail and the zinc finger results in a small protein likely to have a high charge density, with DNA binding affinity an order of magnitude smaller than that of wild type Ty3 NC. A simple DNA binding isotherm [Equations (3) and (5)] fit to change in melting force ΔFm as a function of protein concentration c yields Kd = 20 (±1) nM with saturated change in melting force ΔFmsat = 16 (±0.5) pN. The force-extension curve in the presence of 13 nM Δ2-NCp9 dd exhibits duplex stabilization, DNA aggregation (10 pN force is required to stretch DNA below the dsDNA contour length), and large hysteresis (Figure 7a). In contrast with Δ2-NCp9 (Figure 6b), subsequent stretch–release cycles reflect incomplete protein dissociation (Figure 7b). DNA force-extension curves in the presence of mutants without the zinc finger indicate that the cationic tail of Ty3 NC contributes strongly to nucleic acid aggregation, and is largely responsible for dsDNA stabilization. Both mutants also induce strong hysteresis, and this inhibition of DNA annealing abolishes rapid kinetics, which requires both the zinc finger and the N-terminal tail of Ty3 NC.

Figure 7.

Typical force-extension (solid) and release (dashed) curves of (a) DNA only (black) and in the presence of (a and b) 13 nM Ty3 NC mutant Δ2-NCp9 dd. First stretch–release curve shown in green (a and b), and second stretch–release curve shown in blue (b).

DISCUSSION

Recent studies of nucleic acid chaperone proteins from retroviruses and retrotransposons have shown that aggregation, duplex destabilization and rapid kinetics are often characteristic of their chaperone activity. Single-molecule stretching experiments have also shown that an increase in the slope of the DNA melting plateau is positively correlated with nucleic acid chaperone activity (34,45,46). Force-extension curves in the presence of wild type Ty3 NC exhibit this increase in transition width, and demonstrate that aggregation of both single- and double-stranded nucleic acids is a key component of its chaperone activity. The increase in the force at extensions below the DNA contour length has also been observed with retroviral NCs such as HIV-1 NC and RSV NC (34). In addition, the decrease in ssDNA contour length has been reported for ORF1p, the nucleic acid chaperone from the LINE-1 retrotransposon (45,46). DNA stretching curves in the presence of Ty3 NC also exhibit very little hysteresis, indicating that the protein dissociates quickly from ssDNA, allowing the strands to anneal, and rapid kinetics is a key component of its nucleic acid chaperone activity. These results contrast significantly with those obtained in the presence of MuLV NC, the only other single zinc finger NC protein studied by DNA stretching (34). DNA stretching curves in the presence of MuLV NC exhibit significant hysteresis and much less aggregation. Although the previous results suggested that two zinc fingers may be required for rapid DNA interaction kinetics, this study shows that rapid kinetics can be observed even with a single zinc finger NC protein.

We examined several mutants to investigate the relationship between the structure of Ty3 NC and its nucleic acid chaperone function. Stretching curves in the presence of Δ1-NCp9, which has a partially deleted N-terminal tail, indicate that the zinc finger contributes to weak duplex destabilization, an effect that balances the strong duplex stabilization of the cationic tail. Although the distribution of charge depends upon the structure of Ty3 NC mutants bound to nucleic acids, which is unavailable, both mutants without the zinc finger must have high charge density. Both of these mutants, NCp9 dd and Δ2-NCp9 dd, significantly increase the melting force, to 71.5 (±0.4) pN and 67.1 (±0.5) pN, respectively (Table 1). The N-terminal tail mutants demonstrate that the cationic residues on the N-terminus stabilize dsDNA, while the zinc finger destabilizes dsDNA. When the cationic region is deleted, DNA destabilization by the zinc finger, which contains two aromatic residues that could potentially stack with ssDNA, dominates the overall thermodynamics of the DNA-protein interaction. When the zinc finger is deleted, the protein strongly stabilizes dsDNA.

While duplex stabilization such as that observed for wild type Ty3 NC is likely to be somewhat detrimental to nucleic acid chaperone function, this property is a consequence of the high charge density of the protein. A high charge density is required for efficient nucleic acid aggregation, which is a primary requirement for nucleic acid chaperone activity. Thus, wild type Ty3 NC balances aggregation by its cationic residues and duplex destabilization by its zinc finger, resulting in overall mild duplex stabilization in conjunction with strong aggregation. Although all four mutants aggregated both dsDNA and ssDNA, the effect was strongest with zinc finger deletion. Mutant Δ2-NCp9, which is primarily the zinc finger, showed the least DNA aggregation as well as the weakest DNA binding. Even at 100 nM protein concentration, nearly 4-fold higher than saturated binding of wild type Ty3 NC, multiple stretch–release cycles were required to see an effect. In contrast, the zinc-finger deletion of NCp9 dd is likely to result in higher charge density, preserving DNA binding affinity. These properties are consistent with the ability of NCp9 dd to facilitate tRNA primer annealing, promote genomic RNA dimerization, and initiate cDNA synthesis in vitro because these processes do not require significant duplex destabilization to proceed (47).

The rapid kinetics of wild type Ty3 NC is highly sensitive to protein charge and structure, and deletion of either the zinc finger or the cationic tail forms mutants that induce hysteresis, which demonstrates that rapid kinetics requires both the zinc finger and the N-terminal tail. In fact, only wild type Ty3 NC has rapid kinetics, quickly switching between binding single- and double-stranded nucleic acids. The weak duplex destabilization effect of the zinc finger and the strong duplex stabilization effect of the cationic tail work in concert to transiently destabilize secondary structures in the 5′- and 3′-UTR and facilitate annealing of the DNA:RNA duplex during reverse transcription (2,51). Thus, a single zinc finger is sufficient for a highly charged NC protein to exhibit rapid DNA–protein interaction kinetics. Although the high charge density of the protein strongly stabilizes dsDNA, the single zinc finger also destabilizes DNA, and the resulting balance of these two effects makes Ty3 NC a very strong nucleic acid aggregation protein with only weak duplex stabilization. If Ty3 required significant duplex destabilization to facilitate minus strand transfer, as is the case for HIV-1, a second zinc finger would likely be needed. In fact, HIV-1 NC probably evolved a second zinc finger for this purpose, in conjunction with the requirements of other critical HIV-1 replication processes, due to the high thermodynamic stability of its TAR hairpins. This suggests that nucleic acid chaperone activity of NC proteins is specifically tuned to secondary structures in the UTR regions of their genomic RNA. The nucleic acid chaperone properties are then optimized to maximize DNA aggregation while facilitating only the necessary secondary structure rearrangements required for reverse transcriptase to synthesize a complete functional viral DNA.

FUNDING

National Institutes of Health (R01GM72462); National Science Foundation (MCB-0744456); Institut National de la Santé et de la Recherche Médicale; Agence Nationale de Recherches sur le Sida et les Hépatitis Virales; National Science Foundation Integrative Graduate Education and Research Traineeship Program (DGE-0504331 to K.C.). Funding for open access charge: National Institutes of Health.

Conflict of interest statement. None declared.