Strandedness discrimination in peptide-polynucleotide complexes.

Preferential binding to single- or double-stranded nucleic acids is important for the activity of many proteins that process RNA and DNA. We have investigated the mechanism of strandedness discrimination with peptides derived from the putative DNA-binding domain of the RecA protein, a bacterial recombinase that modulates its affinity for single-stranded DNA by means of ATP binding and hydrolysis. Contributions of electrostatic and non-electrostatic interactions to binding of these peptides with polynucleotides were evaluated by fluorescence spectroscopy as a function of salt concentration and peptide charge. Binding of these peptides to single- and double-stranded nucleic acids was dominated by non-electrostatic interactions. Small electrostatic contributions selectively enhanced peptide complexation with single-stranded nucleic acids. Similar results were observed in control experiments carried out with tripeptides containing charged and aromatic amino acid residues. It was possible to modify the strandedness preference of peptide-polynucleotide complexes by changing electrostatic contributions to the binding free energy. These observations suggest a mechanism whereby some proteins that interact with DNA or RNA might determine and regulate their relative affinity for single- and double-stranded nucleic acids.

Preferential binding to single-or double-stranded polynucleotides is characteristic of numerous proteins involved in DNA replication, transcription, DNA repair, and homologous recombination (1). Single-stranded binding proteins transiently stabilize single-stranded polynucleotides without NTP hydrolysis (2,3). Helicases and recombinases may switch strandedness specificity by NTP-dependent mechanisms (4,5). These proteins generally bind to nucleic acids cooperatively or as multimeric complexes. It is important to understand how these proteins recognize single-and double-stranded polynucleotides and how they change strandedness preference.
Protein sequence motifs that recognize single-stranded DNA generally include both positively charged residues that can interact electrostatically with phosphate groups in the polynucleotide backbone and aromatic amino acids that can associate with nucleic acid bases via non-electrostatic interactions (2,3,6,7). The L2 loop, a putative DNA-binding domain of the RecA protein, has charged and aromatic amino acid residues characteristic of single-stranded DNA-binding proteins; these general features are found in many recombinases (Fig. 1a) (8 -10). Genetic and biochemical experiments show that point muta-tions in this loop disrupt RecA protein function in vivo and in vitro (11)(12)(13)(14)(15) and modify the affinity of the RecA protein for single-stranded DNA (15). This loop is not resolved in the crystal structure of the RecA-ADP complex (16), which suggests that it may be flexible in the absence of DNA and that, at least to a certain extent, it might interact with DNA independently of the protein scaffolding. Hence, a peptide corresponding to the L2 loop may reproduce some of the DNA binding behavior of the L2 sequence in the protein and is an attractive model polynucleotide-binding domain to study strandedness specificity.
We have investigated two peptides with the essential features of the L2 loop (Fig. 1b). In peptide wtw, the wild-type sequence was modified (F203W) in order to follow binding by fluorescence spectroscopy. Peptide 659w has a further modification (E207Q), which increases the net charge. KWK and AcKWK were included as control peptides with the same net charge as the L2 peptides, but with different amino acid sequences. We measured the binding of these peptides to the single-stranded polynucleotides poly(U), poly(A), and poly(dA) and to double-stranded DNA and poly(A)⅐poly(U) at low salt concentrations. Electrostatic and non-electrostatic contributions to the stability of the resulting complexes were evaluated from the effect of peptide charge on binding and from the stability of the peptide-polynucleotide complexes in the presence of high salt concentrations.

Preparation and Characterization of Peptides and Polynucleotides-
Peptides wtw, 659w, KWK, and AcKWK ( Fig. 1b) were synthesized with an Applied Biosystems Model 430A apparatus by standard t-butoxycarbonyl chemistry and purified by reversed-phase HPLC 1 on an Aquapore C8 column with a 0 -70% acetonitrile gradient in 0.1% trifluoroacetic acid. Peptide purity was evaluated by HPLC, and molecular weights were confirmed by fast atom bombardment mass spectroscopy. Peptide stock concentrations were determined from Trp absorption at 280 nm in water at 25°C after confirming that absorption of the peptides in water was the same as in 6 M guanidinium hydrochloride (17). Trp and polynucleotides were purchased from Sigma. Polynucleotides were exhaustively dialyzed against reaction buffer, and their concentrations were determined in 0.1 M Tris acetate, pH 7.4, at 25°C from published extinction coefficients (18,19). Solutions were prepared with filtered deionized water (Milli-Q, Millipore Corp.). Titrations were carried out in low salt buffer (10 mM Tris acetate, pH 7.4) or high salt buffer (1 M KCH 3 CO 2 , 10 mM Tris acetate, pH 7.4, for DNA, poly(A)⅐poly(U), and poly(U); 0.2 M KCH 3 CO 2 , 10 mM Tris acetate, pH 7.4, for poly(A) and poly(dA)). Concentrations of peptides are expressed in units of moles of peptide molecule, whereas concentration units for polynucleotides are moles of nucleic acid residues.
Fluorescence Spectra and Titration-Fluorescence spectra were measured with an SLM 500 spectrofluorometer (AMINCO). Samples were excited at 292 nm, and fluorescence was recorded as the integrated emission at 335-500 nm. Bandwidths for excitation and emission were 2 nm. UV absorbance and fluorescence of the peptide solutions varied linearly with peptide concentration, indicating the absence of association between peptide molecules in the concentration ranges used. * This work was supported in part by grants from the Association pour la Recherche sur le Cancer and the Ligue Nationale Contre le Cancer (to N. P. J.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
During titration, aliquots of polynucleotide stock solution were added to peptide solutions. The samples were agitated and allowed to equilibrate in a thermostated sample holder at 25°C. In some experiments, titrations were carried out in low salt buffer, and the electrostatic contributions to the resulting complex were subsequently investigated by titrating with 2 M KCH 3 CO 2 (or NaCH 3 CO 2 ), 10 mM Tris acetate, pH 7.4. In other experiments, non-electrostatic binding was measured by titrating the peptide with polynucleotide in high salt buffer. Data are reported as fluorescence quenching, Q obs ϭ (F 0 Ϫ F)/F 0 , where F 0 is the initial fluorescence in the absence of polynucleotide and F is the fluorescence during titration. Fluorescence intensity was corrected for dilution, and inner filter effects were corrected by the relation (20) F corr ϭ F obs antilog(A ex /2), where F corr and F obs are the corrected and observed fluorescence, respectively, and A ex is the absorbance at the excitation wavelength (292 nm), which was calculated from the concentrations of polynucleotide and peptide with the following extinction coefficients: ⑀ 292 ϭ 104 M Ϫ1 cm Ϫ1 (poly(A) and poly(dA)), (poly(A)⅐poly(U)), and ⑀ 292 ϭ 3200 M Ϫ1 cm Ϫ1 (peptides). Measurements during titration showed that the absorbance of polynucleotides in the reaction mixture obeyed Beer's law at 292 nm over the entire concentration range of the experiments. Experiments were carried in a 0.4 ϫ 1.0-cm quartz cuvette with the 0.4-cm path oriented in the direction of the excitation beam in order to minimize absorbance corrections. Dividing the calculated absorbance by 2 did not significantly alter the shape of the fluorescence titration curves, and the relative magnitudes of the binding constants in Table I were unchanged, indicating that our qualitative conclusions are independent of uncertainties in inner filter correction.
Binding Constants-The formalism of McGee and von Hippel (21, 22) for noncooperative binding of ligand on an infinite lattice was used to determine apparent binding constants. Fluorescence quenching curves were calculated as a function of binding constant (K obs ; M Ϫ1 ), site size (n), and Q max . K obs is an apparent binding constant for the overall equilibrium, [peptide] ϩ [polynucleotide] % [peptide-polynucleotide], which may include electrostatic and non-electrostatic interactions. Q max is the quenching of peptide complexed to polynucleotide. We have fit the fluorescence quenching data assuming n ϭ 3 or 15, the number of amino acid residues in the peptide. If data are available for sufficient peptide binding (Ͼ80% of maximum quenching), then both Q max and K obs can be determined from the best fit of the fluorescence data. Alternatively, Q max can be determined independently of K obs and n by analyzing titrations of several concentrations of peptide with the binding density function method (23). Good agreement for Q max was observed by these two methods where comparison was possible. In some cases (such as titration in high salt buffer), we were unable to add sufficient polynucleotides to unambiguously calculate Q max from the titration curve or to perform binding density function analysis. We assumed in these experiments that Q max was the same in low and high salt buffers. We found Q max ϭ 1 within experimental error for all reactions studied, except titration of peptide KWK with poly(U) at pH 5.8, where Q max ϭ 0.85 Ϯ 0.05, in agreement with previously published results (19). Representative fits are shown in Figs. 2-4. We calculated the contributions of electrostatic interactions to binding (K obs el ) by K obs el ϭ K obs /K obs nel , where K obs is the binding constant in 10 mM Tris acetate, pH 7.4, at 25°C, and K obs nel is the non-electrostatic contribution to binding estimated by the titration of peptides in high salt buffer.

Electrostatic and Non-electrostatic Contributions to Stability of Peptide-Polynucleotide
Complexes-Since titration of peptides with polynucleotides quenched the peptide fluorescence (Figs. 2-4), quenching can be used to measure the binding of peptide to polynucleotide. Both electrostatic and non-electrostatic interactions contributed to the binding of peptide 659w to single-stranded poly(A) and poly(U) at 25°C, pH 7.4; this can be seen by the partial decrease in quenching of the peptidepolynucleotide solution with increasing KCH 3 CO 2 (Fig. 2,  squares). Complexes between peptide wtw and single-stranded nucleic acids were less sensitive to salt (Fig. 2, circles), indicating that they were stabilized to a lesser extent by electrostatic interactions. Non-electrostatic interactions stabilized the complexation of both wtw and 659w with double-stranded poly(A)⅐poly(U) and DNA in these reaction conditions without significant contributions from electrostatic binding (Fig. 3). The non-electrostatic nature of this binding is apparent from the incapacity of KCH 3 CO 2 to disrupt the peptide complexes with poly(A)⅐poly(U) or DNA (Fig. 3) and from the similar titration curves observed in low and high salt buffers (data not shown). Furthermore, the fluorescence curves observed upon titration of 659w or wtw with double-stranded nucleic acids in low salt solutions were insensitive to the different charges of the peptides, as would be expected for non-electrostatic binding.
To quantitate electrostatic and non-electrostatic contributions to binding, we determined the relative contributions of K obs el and K obs nel from titration curves measured at various salt concentrations (Table I). Electrostatic binding did not significantly contribute to the stability of any peptide complex with double-stranded poly(A)⅐poly(U) or DNA. Furthermore, binding of peptide wtw to single-stranded polynucleotides was also driven by non-electrostatic interactions. Electrostatic interactions contributed selectively to the stability of complexes between peptide 659 and single-stranded polynucleotides; the calculated electrostatic binding constants were 2 orders of magnitude less than the non-electrostatic binding constants.
Effect of Peptide Sequence and Reaction Conditions on Nonelectrostatic Interactions-It has been previously reported that short peptides of polylysine containing Trp residues bind to polynucleotides primarily by electrostatic interactions (18,19,(25)(26)(27). To investigate whether these differences with our observations were a consequence of particular peptide sequences or different reaction conditions, we compared the binding of KWK to poly(U) under our conditions at pH 7.4 with the same reaction at pH 5.8, which has been previously well studied (19).
Binding constants (Table II) were determined from the fluorescence quenching of 6 -16 M KWK with poly(U) in 20 mM KCH 3 CO 2 , pH 5.8. Fluorescence quenching of the resulting complexes diminished upon addition of KCH 3 CO 2 , and the linear part of the plot of log(K obs ) versus log[K ϩ ] had a slope of Ϫ1.50 Ϯ 0.22. The extrapolated binding constant at 1 M monovalent cation, log(K obs ) ϭ 0.9 Ϯ 0.4, is taken as a measure of the non-electrostatic contribution to binding (24). These results agree with previous reports (19) that electrostatic interactions stabilize KWK-poly(U) complexes at low salt concentrations and pH 6. However, peptide KWK bound to poly(U) at this pH in the presence of 1 M KCH 3 CO 2 (Table II). Furthermore, KWK bound to poly(U) in 10 mM Tris acetate, pH 7.4, by non-electrostatic interactions at low salt concentrations judging from log(K obs ) extrapolated to [K ϩ ] ϭ 1 M (Fig. 4 and Table II).
At pH 5.8, the ␣-amino group of peptide KWK is protonated (pK ϭ 7.5) (26). Deprotonation or screening of the charge at the amine by 1 M salt might partially explain our observed nonelectrostatic interactions. To investigate this possibility, we compared the binding of peptides KWK and AcKWK to poly(U) at pH 5.8 (Table III). Acetylation of the ␣-amine decreased the apparent binding constant in low salt solutions as expected from loss of ϩ1 net positive charge. However, increasing salt concentrations only slightly diminished fluorescence quenching of the resulting complex. Hence, in contrast to peptide KWK, non-electrostatic interactions contributed to the stability of the AcKWK-poly(U) complex at low salt concentrations. The binding constants of AcKWK in low salt (Table III) and KWK in high salt (Table II) were identical, suggesting that AcKWK may be a good model for binding of KWK to poly(U) at high salt   b and d, respectively). Symbols and reaction conditions are the same as described for Fig. 2. concentrations, pH 5.8. Finally, interactions between Trp and poly(U) probably account for non-electrostatic binding; the equilibrium constant for binding of tryptophan to poly(U) under these conditions closely resembled that for the tripeptide AcKWK (Table III). Taken together, these results suggest that the charge on the ␣-amine of peptide KWK influences the mechanism by which this molecule binds to poly(U). If this moiety was protonated, (pH 5.8), interactions with the polynucleotide in low salt solution were almost entirely electrostatic; if the charge of this moiety was neutralized by acetylation or pH or if it was screened by high salt concentrations, non-electrostatic interactions became important. The influence of electrostatic contributions on the binding of KWK to single-and double-stranded polynucleotides ( Fig. 4 and Tables I and III) resembled that observed for peptide 659w (Figs. 2 and 3 and Table I). In all cases, electrostatic interactions significantly stabilized complexes of both peptides with single-stranded polynucleotides and had little or no detectable contribution to the stability of complexes with double-stranded nucleic acids.

DISCUSSION
The nucleic acid-binding domains of proteins that preferentially recognize single-stranded nucleic acids are believed to be characterized by positively charged amino acid residues that interact electrostatically with the polynucleotide phosphate backbone and aromatic amino acids that can stabilize the complex through non-electrostatic interactions (2, 3, 6, 7). We have evaluated electrostatic and non-electrostatic contributions to the complexation of peptides with single-and double-stranded polynucleotides. The peptides studied were derived from the L2 DNA-binding domain of the RecA protein, a bacterial recombinase that cycles, in an ATP-dependent reaction, between conformations with high and low affinity for single-stranded DNA. These were compared with model tripeptides (KWK and AcKWK) that have the same charge as the L2 peptides (Fig. 1).
Two types of experiments showed that complexes of all peptides with double-stranded polynucleotides were stabilized by non-electrostatic interactions (Figs. 3 and 4 and  Tables I and III). First, fluorescence quenching of peptides by poly(A)⅐poly(U) and DNA was insensitive to salt concentration. Second, altering the charge of the peptide did not affect binding to double-stranded nucleic acids; peptide charge was modified by either an amino acid substitution in the case of the L2 peptides or acetylation of the ␣-amino group of peptide KWK. Complexes of peptides with single-stranded polynucleotides were stabilized mainly by non-electrostatic interactions. Recent filter binding studies have also shown that peptides based on the amino acid sequence of the L2 loop of the RecA protein bound to single-stranded DNA at high salt concentrations and neutral pH (28) presumably by non-electrostatic interactions. However, we observed that the stability of single-stranded complexes was also partially sensitive to the charge of the peptide and to salt concentration (Table I). Hence, in our experiments, non-electrostatic interactions stabilized peptide complexes with both single-and double-stranded polynucleotides; electrostatic interactions, on the other hand, contributed selectively to the stability of single-stranded complexes.
Early fluorescence studies (26,29) reported that peptide KWK bound to single-and double-stranded polynucleotides at low salt concentrations primarily by electrostatic interactions. The observed preference of KWK for single-stranded polynucleotides (26,29) likely reflects the larger electrostatic contributions to the stability of these complexes, compared with doublestranded polynucleotides, and is consistent with our results. However, in those experiments, inner filter effects were evaluated from fluorescence quenching of the reactants in solutions with high salt concentrations where complexes were assumed to entirely dissociate. Correcting for inner filter effects by this procedure could subtract potential non-electrostatic interactions from titrations at low salt concentrations, which may in part explain the reported absence of non-electrostatic binding.
Oligolysine peptides (K) n -⑀-N-DNP-K (n ϭ 3-8) (24,30) and KWK n (n ϭ 1-8) (25) have been shown to bind to polynucleotides by an entirely electrostatic process. We also observed that non-electrostatic contributions did not significantly stabilize complexes between single-stranded poly(U) and KWK at pH 5.8 in low salt solution (Table II) as previously reported (19,25). However, deprotonation or acetylation of the ␣-amine of KWK or high salt concentrations increased the non-electrostatic contribution to this reaction ( Fig. 4 and Tables I-III). These observations suggest that in low salt solutions at pH 5.8, KWK binds to polynucleotides through the positively charged ␣amino group. Interactions between the ␣-amino group and the polynucleotide may explain why peptides with a total charge of z Ͻ ϩ4 bind to single-stranded polynucleotides with different thermodynamics than more highly charged oligolysine molecules (25). Taken together, these results show that peptides with an uncharged ␣-amine extremity and a small net positive charge bind to polynucleotides in low salt solutions qualitatively differently than highly charged oligolysine molecules. Their large non-electrostatic interactions compared with those of polylysine oligopeptides suggest different peptide-nucleic acid contacts in the two cases.
One important consequence of these results is that electrostatic contributions to the binding free energy can modulate the strandedness preference of peptide-nucleic acid complexes. For  example, reducing the net positive charge of the peptide by an amino acid substitution (659w 3 wtw) had no effect on peptide binding to double-stranded poly(A)⅐poly(U) at pH 7.4, but decreased binding to the component single-stranded polynucleotides poly(A) and poly(U) (Figs. 2 and 3 and Table I). Likewise, neutralization of the positive charge of the ␣-amino group (KWK 3 AcKWK) had little effect on binding to doublestranded DNA at pH 5.8, but decreased binding to singlestranded poly(U) (Table III). Similarly, at pH 7.4, electrostatic interactions stabilized complexes of peptide KWK with poly(A) or poly(U) more than with double-stranded poly(A)⅐poly(U) (Fig. 4 and Table I).
These results show that, at least for certain peptide-polynucleotide complexes, increasing electrostatic binding can selectively enhance the affinity of the peptide for single-stranded nucleic acids. It should be noted that this observation cannot be explained by the relative charge density of single-stranded nucleic acids, which is lower than that of duplex molecules (27). We speculate that non-electrostatic interactions might constrain the peptide molecule and that weak electrostatic contacts might be favored by the greater conformational flexibility of single-stranded polynucleotides.
If these results can be extrapolated to protein-nucleic acid complexes, then protein conformational changes that modify the net charge of the binding domain could regulate strandedness preference of the protein. We note, for example, that negatively as well as positively charged amino acids are present in homologous DNA-binding domains of recombinases from phage to man (Fig. 1a) (8 -10); negative charges in the DNA-binding domain might allow these proteins to modulate their affinity for single-stranded DNA by means of allosteric conformational changes that modify the relative electrostatic and non-electrostatic contributions to polynucleotide binding. Alteration of the relative affinity of the L2 loop of the RecA protein for single-and double-stranded DNA could have important consequences for the mechanism of action of the RecA-DNA nucleoprotein filament during homologous recombination.