Binding of RecA Protein to Z-form DNA Studied with Circular and Linear Dichroism Spectroscopy*

Binding of RecA to poly(dG-m6dC) and poly(dG-dC) under B- and Z-form conditions was studied using circular dichroism (CD) and linear dichroism (LD). LD revealed a quantitative binding of RecA to Mg2+-in-duced Z-form poly(dG-mSdC) with a stoichiometry of 3.1 base pairs/RecA monomer, which is slightly larger than the 2.7 base pairs observed for the B-form. The LD spectra indicate a preferentially perpendicular orientation of DNA bases and a rather parallel orientation of the tryptophan residues relative to the fiber axis in both complexes. The association rate of RecA to Z-form DNA was found to be slower than to B-form. CD meas- urements showed that the polynucleotide conformation is retained upon RecA binding, and CD and LD confirm that RecA binds to both forms of DNA. The retrans- formed into both and in polynucleotides when the order of addition of RecA, ATPyS, MgC12 the kinetics of RecA binding to double-stranded in contrast to the binding of RecA to single-stranded DNA (16). For LD measurements, the solution mixed in a test tube immediately transferred to the Couette cell by aid of a syringe, enabling LD to be monitored 1-2 min after the mixing. In all cases, the mixing time referred to as time 0, and the binding rates are characterized by half-times (7112) referring to 50% of the total signal change. During the measurements, the shear was occa- sionally removed to verify base-line stability.

the complex with single-stranded DNA in absence of ATP-yS in which the DNA is slightly condensed (4, 7 ) and the DNA bases apparently unoriented (10-12).
It was recently reported that RecA binds with higher affinity to UV-irradiated DNA (21) and Z-form DNA (22). The latter result is particularly interesting in relation to recombinase activity: in the course of strand exchange between two DNAs, it may be speculated that one DNA strand is in a lefthanded configuration to form a double-helix with the complementary strand of another DNA molecule (23). Electron microscopy studies give some support to this hypothesis (24).
In this work we study the binding of RecA to Z-DNA using spectroscopic measurements (CD and LD) which provide information about hydrodynamic properties and structure but also about the conformation of the DNA in the complexes. Kinetic studies are also undertaken and the results compared with the binding kinetics with B-form DNA under corresponding conditions.

MATERIALS AND METHODS
RecA was purified as described elsewhere (18,25) except that a larger phosphocellulose column was used in order to eliminate impurities that may aggregate the RecA and RecA.DNA complexes. Under our experimental conditions, no significant turbidity was observed in CD above 310 nm. However, at NaCl concentrations below 30 mM a significant turbidity was noticed, especially in the presence of MnC12. Protein concentrations were determined spectroscopically using eZm = 2.17 X lo4 M" cm" (26). Poly(dG-dC) (lot OK 617910) and poly(dG-m5dC) (lot OC 617938) were obtained from Pharmacia LKB Biotechnology Inc. and used without further purification. Their concentrations were determined using e254 = 8.4 X lo3 M" cm" (in bases) and cZ5$ = 7.0 X lo3 M" cm", respectively.
CD and LD were measured on a Jasco 5-500 spectropolarimeter. LD measurements were carried out as described earlier (10)(11)(12)27) using an inner rotor Couette cell (28) at a shear gradient of 600 S-I. CD and LD measurements were conducted in a buffer containing 20 mM Tris-HCI, 40 mM NaCl, and 1 mM dithiothreitol, pH 6.9, at 20 "c, supplemented with indicated concentrations of MgC12 or MnC12. The concentration of ATPyS (Boehringer Mannheim) was usually 20 pM. Higher concentration of ATPyS could not be used because of distortion of CD spectra due to large absorption. However, 20 p M of ATPrS is expected to be more than sufficient to saturate 8-10 pM RecA that has been used here, because the dissociation constant of ATPyS. RecA is about 10" M (29, 30) and is expected to be even smaller in the presence of DNA. It was further verified by LD that an increase of the ATPyS from 20 to 200 PM did not significantly affect the association kinetics or the LD spectrum.
The DNA-RecA binding reaction was generally initiated by addition of ATPyS to an equilibrated DNA/RecA mixture, although it was found that the order of addition of RecA, DNA, ATPyS, and MgC12 does not affect the kinetics of RecA binding to double-stranded DNA: which is in contrast to the binding of RecA to single-stranded DNA (16). For LD measurements, the solution was mixed in a test tube and immediately transferred to the Couette cell by aid of a syringe, enabling LD to be monitored 1-2 min after the mixing. In all cases, the mixing time is referred to as time 0, and the binding rates are characterized by half-times (7112) referring to 50% of the total signal change. During the measurements, the shear was occasionally removed to verify base-line stability.

RESULTS
B + 2-transition of Poly(dG-m6dC)-Methylated poly(dG-dC), referred to as poly(dG-m5dC), undergoes B + Z-transition at much lower Mg2+ concentrations (31) than unmodified poly(dG-dC) (32). Under our conditions (40 mM NaCl), poly(dG-m6dC) was transformed to Z-form by the presence of about 1.3 mM MgClz (Figs. 1 and 2). The rate of the B + Z transition was found to be very slow, especially at low MgC12 concentrations ( Fig. 1). In 0.5 mM MgC12, no significant amount of DNA in Z-conformation could be detected even after 3 h of incubation (Figs. 1 and 2b). Therefore, it was possible to study the binding of RecA to both Band Z-forms of the same DNA by changing the MgClz concentration. The use of the same DNA is crucial for a strict quantitative comparison since the binding of RecA to DNA can be anticipated to depend strongly on the base composition and chemical modification of the bases, as inferred from the case of single-stranded DNA (13, 15). The interaction of RecA with single-stranded DNA appears to be weaker with shorter DNA (3). Also, with double-stranded DNA the interaction appears to depend on the length of the DNA (22) and on chemical modification by UV irradiation (21). In contrast, the MgC12 concentration appears not to be crucial: changing the concentration from 0.5 to 5 mM does not significantly alter the binding kinetics or LD signal of RecA-double-stranded DNA complexes .2 Since the rate of the B + 2-transition of poly(dG-m5dC) is very slow in 1.3 mM MgC12 (Fig. l), the binding to both Band Z-forms could be studied at this MgC12 concentration. Binding t o B-form was monitored immediately after the addition of MgClZ to the polynucleotide, i.e. before the transition to Z-form reached a significant level. Under the same conditions, binding to the Z-form could be studied after incubation with MgCL for times sufficient to complete the transition to the Z-form. Thus, an unambiguous comparison between the binding to Zand B-forms of poly(dG-m5dC) was possible.

Conservation of Polynucleotide Conformation upon RecA
Binding-Poly(dG-m'dC) was incubated for at least 4 h in 1.3 or 2.5 mM MgC12 to ensure complete transformation to Zform. The CD spectra were essentially the same at these MgC12 concentrations (Fig. 2a) and were only slightly modified by addition of ATPyS-RecA, even after 10 h of incubation (Fig. 2a). The relatively modest changes in CD, especially when one takes into account the contribution from RecA. ATPyS (Fig. 2a), indicate a completely retained Z nucleotide conformation in the complexes with RecA.ATPyS. The binding to the B-form of the polynucleotide was studied in 0. 5   MgC12. The spectrum of ATPyS. RecA ( . . . . ) is also shown.
added immediately after the MgC12, before the slow transformation to Z-form had occurred). The CD spectra after 10 h of incubation are shown in Fig. 26. There is no significant difference between the spectra at these two MgC12 concentrations, but they are clearly different from those of the RecA complex with the polynucleotide in the Z-conformation. During the incubation, the positive band around 280 nm increases and the negative band centered at 250 nm decreases but the overalI changes of the CD spectrum are moderate, indicating that the polynucleotide remains in the right-handed B-conformation.
The amount of RecA used in these experiments was 1 monomer/3 base pairs of polynucleotide, which corresponds to the binding stoichiometry for the complex with natural double-stranded DNA (3,4). A doubling of the RecA concentration modified the CD signals by less than 10% (not shown).
Binding Stoichiometries-The binding of RecA to doublestranded calf thymus DNA increases the amplitude of the LD signal a t 260 nm by a factor of 18 and modifies the shape of the LD spectrum (12). Such a large spectral change strongly facilitates accurate quantitative determination of stoichiometry as compared to the use of the small changes in CD spectrum.
Various amounts of polynucleotide were mixed with 0.8 PM RecA, incubated for 10 h a t 20 "C, and LD spectra were recorded. For the Z-form of the polynucleotide the presence of RecA strongly enhanced the negative LD around 260 nm, and a small positive signal above 290 nm appeared. Also, with the B-form polynucleotide the amplitude of the LD signal around 260 nm increased drastically upon addition of RecA, but in this case the signal at 290 nm remained negative ( Fig.

dl.
For both nucleotide conformations the LD signal at 260 nm initially increased linearly with the polynucleotide/RecA ratio ( Fig. 4). In both cases, addition of larger amounts of polynucleotide changed the slope to a value corresponding to that of the free polynucleotide. The cross-over of the extrapolated initial and final slopes ( Fig. 4) corresponds to a stoichiometry of 1 RecA/2.7 base pairs of polynucleotides in B-form and 3.1 base pairs for the Z-form. These values are not significantly different from the 3 base pairs/RecA observed for natural double-stranded DNA (12). Note that the LD signal at 260 nm of free Z-form polynucleotide is about twice that of the free B-form ( Fig. 3), as has been observed previously (33), and that the LD increases upon RecA binding by a factor of 8.5 for the Z-form but about a factor of 40 for the B-form. The LD signal of the RecA complex with Z-form is thus about one-half of that with B-form.
Association Kinetics-In order to compare the binding rates of RecA to Z-and B-form of poly(dG-m'dC), the changes in CD and LD signals were normalized, assuming complete binding after 10 h of incubation. The rates determined with LD and CD were in good agreement with each other for both polynucleotide conformations (Fig. 5 ) , although the accuracy is believed to be better in the LD study because of the larger spectral changes. The binding rate was significantly slower to the Z-form of the polynucleotide: the half-time was found to be about 20 min for the binding to the B-form and about 40 min to the Z-form. For neither conformation did the MgClz concentration significantly influence the binding rate (Fig. 5).

Differences in Stability between Complexes with Z-and Bform DNA-
The Z-form of poly(dG-m'dC) in 2.5 mM MgC12 could be transformed back to B-form by addition of NaCl (Fig. 6). In 140 mM NaC1, complete retransformation of uncomplexed poly(dG-m6dC) into B-form took about 30 min (results not shown). In less than 100 mM NaC1, no significant change in CD spectrum was observed even after 15 min of incubation and only a minor increase in CD was observed in 120 mM NaCl (Fig. 6). The RecA complex with Z-form of poly(dG-m'dC) in 2.5 mM MgClz could also be retransformed to B-form in 140 mM NaCl (Fig. 6) within 30 min, and a small degree of retransformation was observed even in 120 mM NaCl (Fig. 6). Thus, the binding of RecA does not stabilize the Z-conformation of the polynucleotide. The CD spectrum of the fully retransformed polynucleotide in 200 mM NaCl was similar to that of the RecA complex formed directly with B-form of poly(dG-m'dC), although the intensity around 280 nm was slightly weaker. This suggests that most of the RecA remained bound to the polynucleotide after the transformation. The presence of 200 mM NaCl had no effect per se on the structure of the RecA complex with the B-form of poly(dG-m5dC) as judged from CD (Fig. 6).
Somewhat surprisingly, an increase of MgClZ concentration to 2.5 mM did not transform the polynucleotide in the RecA complex from B-form into Z-conformation even after 2 h of incubation ( Fig. 1). Such an increase in MgClz concentration completely transforms the free polynucleotide to Z-form within 2 h (Fig. 1). Even an increase to 10 mM MgClz did not cause any significant transformation as judged from CD and LD (not shown). These results indicate that the RecA complex with B-form of poly(dG-m'dC) is kinetically stable.

RecA Binding to Z-form Poly(dG-dC) Induced by Mnz+-
We also carried out experiments with unmodified poly(dG-dC) using Mn2+ as a B + Z inducer (34). CD spectra of poly(dG-dC) in the presence of 2.5 mM MnClZ indicated a Btype conformation of the polynucleotide (Fig. 7b), although the positive band around 280 nm was smaller than in the absence of MnClZ (Fig. 7c). Heating the poly(dG-dC) to 60 "C inverted the initially B-form CD spectrum to a Z-form spec-   (-. . -) preheating, and in complex with RecA, with ( ---) and without (-) preheating. CD of free RecA with ATPrS is also shown ( . . . . ). trum within 5 min (not shown). After progressive cooling to ensure double-helical structure, the CD around 290 nm remained negative, indicating a retained Z-type conformation (Fig. 7a), although the negative band around 290 nm was smaller than for the Z-form induced by MgClZ, as observed previously (34). The polynucleotide remained in the Z-conformation for a t least 48 h when kept at 20 "C. Thus, the RecA binding to B-and Z-form poly(dG-dC) could be compared under identical experimental conditions. Addition of 10 p~ RecA and ATPyS to the Z-form of poly(dG-dC) slowly gave a slightly modified spectrum. The change of the CD at 290 nm was larger and occurred somewhat faster than for the RecA complex with the Z-form of poly(dG-m5dC) (Fig. 8). After 10 h of incubation the positive CD band around 270 nm was shifted slightly toward longer wavelengths but the small negative band around 300 nm remained (Fig.  7a), indicating that the polynucleotide still was in a Z-like conformation. Addition of RecA to the B-form of the polynucleotide increased the magnitude of the positive CD band around 280 nm, but the overall features of the CD spectrum remained unchanged indicating a retained B-type conformation. The increase of the positive band was larger (Fig. 7) and occurred slightly faster (Fig. 8) than for the complex with Bform poly(dG-m5dC) (Fig. 5).
With both the B-and Z-conformations of poly(dG-dC), the RecA association rates were faster than for the binding to methylated poly(dG-dC) observed above (compare Figs. 5 and 8). The half-times of the reactions were less than 15 min for both conformations. The difference in binding rates between the two forms was much smaller than with poly(dG-m5dC), although the binding to the Z-form appeared to be slightly slower (Fig. 8).
LD measurements were carried out to quantify the binding and to follow more accurately the association kinetics. The LD signal of free Z-form of poly(dG-dC) was twice that of the B-form (Fig. 9), which is the same as with poly(dG-m5dC) (Fig. 3), indicating a proper double-helical structure (33, 35, 36). For the B-form of poly(dG-dC) the LD at 260 nm increased about 20 times upon addition of RecA (Fig. 9), and the time dependence was similar to that observed in the CD measurements (Fig. 8). LD measurements of various amounts of RecA added to 25 PM of poly(dG-dC) in B-form revealed a stoichiometry of 2.8 base pairs/RecA (Fig. 10). The magnitude of the LD at 260 nm initially increased linearly with increasing RecA concentration and reached eventually a plateau. By contrast, the LD of the Z-form decreased upon addition of RecA and virtually vanished (Fig. 9). The determination of binding stoichiometry and binding rates from LD was therefore not possible, because the LD signal vanished before complete saturation occurred. The reason for the vanishing LD is at present unclear, but it seems likely that the complex aggregates and therefore does not orient in the flow field. Mn2+-induced Z-DNA has been reported to be prone to aggre-  To examine whether heating the poly(dG-dC) has any effects on the complex formation, we compared the binding of RecA to the polynucleotide before and after heating to 60 "C in the presence of 2.5 mM MgClz instead of MnC12. In absence of Mn2+, poly(dG-dC) remains in the B-form even after heating (Fig. 7c). In both cases a slow increase of the CD at 280 nm was observed upon addition of RecA (Fig. 8).
The changes in CD spectrum (Fig. 7c) were smaller than in the presence of MnC12 (Fig. 7b) and the binding rates much slower (Fig. 8): the half-time was more than 60 min with both the heated and non-heated polynucleotide, and the LD a t 260 nm increased about 20 times in magnitude upon RecA binding (not shown). These results indicate that heating the polynucleotides has no significant effect on RecA binding, whereas the nature of the metal ion clearly has.
Addition of NaCl to the free Z-form of poly(dG-dC) transforms it back to the B-form, as judged from the progressive decrease of the CD at 270 nm (results not shown). Complete transformation back to B-form was obtained in about 140 mM NaC1, and the CD spectrum was similar to that of the B-form of poly(dG-dC) in absence of MnC12. The Z-form of poly(dG-dC) complexed with RecA could also be transformed back to the B-form by addition of NaCl (not shown). This shows that the binding of RecA does not stabilize the MnClz-induced Zconformation of poly(dG-dC) either. After the retransformation the majority of the RecA remained bound to DNA as judged from CD.

DISCUSSION
We have investigated the interaction of RecA with Z-form of DNA in terms of binding stoichiometry and association kinetics using polarized light spectroscopy. These results are compared with the RecA binding to the same DNA in B-form under else identical or similar conditions. In addition to stoichiometric and reaction-mechanistic information, the spectroscopic measurements have also provided structural information about the RecA. DNA complexes.

Conformation of Polynucleotide i n RecA . D N A Complexes-
Circular dichroism measurements reveal that the binding of RecA does not significantly alter the conformation of the polynucleotide ( Figs. 1 and 7): Z-form DNA remains in the Zform and B-form DNA remains in the B-form after the binding of RecA. In no case was it observed that binding of RecA induces conformational changes of the polynucleotides, and it is concluded that the conformation of the polynucleotide in solution is preserved upon RecA binding.
Differences i n Global Structure of Complexes with Z -and Bform DNA-LD spectra of RecA complexes with B-and Zform DNA exhibit similar features: a negative band centered at 260 nm and a small positive band above 290 nm partly overlapping the negative DNA LD. The LD reflects the orientation of the prominent in-plane transition moments of the light-absorbing DNA base and tryptophan chromophores relative to a preferred orientation axis of the complex (12). This indicates, in both types of complexes, a perpendicular orientation of the DNA bases and a more parallel orientation of the aromatic planes of the tryptophan residues in RecA relative to the fiber axis of the complex. The binding stoichiometry is similar in all complexes: about 3 base pairs/RecA monomer, possibly slightly larger for the complex with Z-DNA. The organization of the DNA and the proteins appears thus to be similar in the two types of complexes, i.e. very probably a helical arrangement of RecA also around Z-form DNA as indicated for B-DNA (6)(7)(8)(9).
The amplitude of the LD for the RecA complex with Zform of poly(dG-m5dC) is, somewhat surprisingly, considerably smaller than with the B-form (Fig. 3). For the complex with the Z-form of poly(dG-dC) induced by Mn2+ the LD signal virtually vanishes (Fig. 9). This could indicate that the RecA complex with Z-form DNA is more flexible or more condensed (or aggregated) than the complex with B-form DNA, despite the fact that free Z-DNA is stiffer than free B-DNA (33). The difference may also be due to less perpendicular DNA base orientation in the Z-form complex.

Association Kinetics
The association of RecA to double-stranded DNA is very slow in all cases, and the rate depends on the conformation of the DNA and on the nature of the positively charged divalent ions present. Under similar conditions and with the same DNA, the association rate to Z-form DNA is always slower than to the B-form ( Fig. 5 and 8 ) , a finding that seems to be in conflict with the conclusions of Blaho and Wells (22). Their conclusions were mainly based on the comparison of the binding of RecA to unmodified poly(dG-dC) in B-form and to the modified polynucleotides poly(dG-m5dC) and poly(dG-BrdC) in Z-form. However, it cannot be excluded that the modification of the DNA bases per se might influence the binding rates of RecA (13,15,21). When we compared the binding of RecA to unmodified poly(dG-dC) with the binding to the modified polynucleotide poly(dG-m5dC) under identical conditions (2.5 mM MgC12,compare Figs. 5 and 8), we found the association to the modified polynucleotide to be significantly faster irrespective of the nucleotide conformation. Because the lengths of the polynucleotides were not necessarily the same this comparison is not strictly justifiable, but our results might explain the observations of Blaho and Wells (22), and we conclude that modification of the DNA bases influences the association rates. A preferential binding of RecA to modified double-stranded DNA has also been observed previously both i n vivo and i n vitro: RecA binds more tightly and quickly to DNA modified by UV irradiation (21) or by the carcinogen N-2-a~etylaminofluorene.~*~ We find the RecA binding to poly(dG-dC) in the presence of Mn2' to be much faster than in presence of M P , as also reported by Blaho and Wells (22), and to depend much less on the conformation of the polynucleotide. However, Blaho and Wells observed in a filter binding assay (22) a slightly larger retention of the RecA complex with Z-form DNA than of the complex with the B-form, in contrast to the slightly slower binding of RecA to the Z-form of DNA reported here. This discrepancy might be due to a difference in the efficiency of retention or binding cooperativity of the complexes with Z-and B-form DNA. The filter binding assay is known to be influenced by other factors than binding affinity (37, 38). Another possibility might be partial denaturation of poly(dG-dC) caused by the heating done to induce the Z-form. Blaho and Wells (22) heated the poly(dG-dC) in the absence of NaC1, whereas we induced the Z-form in the presence of 40 mM NaC1. The higher ionic strength should increase the melting temperature and prevent strand separation. Even in 40 mM NaCl heating to above 70 "C followed by progressive cooling induces a certain degree of non-double helical structure, as judged from a decrease of the LD signal (results not shown), to which the binding of RecA is faster. Finally, it should be recalled that our measurements, in contrast to those of Blaho and Wells (22), were carried out in the presence of ATPyS which does not hydrolyze. In fact, they observed considerable lags in ATP hydrolysis rates with Z-DNA compared to B-DNA.
From these results it is concluded that the binding rate of RecA to double-stranded DNA (in presence of ATPyS) depends both on conformation and chemical modification of the DNA, as well as on the nature of the metal cations. RecA binds faster to DNA in B-conformation than in Z-conformation ( Figs. 5 and 8 ) , and faster to methylated poly(dG-dC) than to unmodified poly(dG-dC) ( Figs. 5 and 8) and also faster in the presence of Mn2+ compared to Mg" (Fig. 8).

Stability of RecA . D N A Complexes
For both poly(dG-dC) and poly(dG-m5dC) our conclusions regarding the relative stability of the RecA complexes with the polynucleotides in their B-and Z-conformational states N. Koffel-Schwartz and R. Fuchs, personal communication. X. Veaute, I. Bertrand, M. Takahashi, and R. Fuchs, unpublished results. can be summarized by the following reaction scheme (~~1~ is given in parentheses for the Me-induced Z-form of poly(dG-m5dC)). The RecA-DNA binding is characterized by a very high stability constant (3, 4), and the binding reaction can therefore be considered as practically irreversible. This is important to recall when considering the NaC1-induced retransformation of the B-form from the Z-form of polynucleotides in complex with RecA. Although we known as yet very little about the structural details of these complexes, it is not difficult to imagine how the compactly stacked array of RecA proteins may provide a (kinetic) barrier to major conformational changes in the DNA double helix. Thus, it is not astonishing that B-form poly(dG-dC), when complexed with RecA, is prevented from undergoing a transition to Z-form. What is more surprising, however, is the observation that divalent metal-induced Z-form is retransformed to B-form by addition of NaC1, also when associated with RecA. Furthermore, this reaction occurs at practically the same Na' concentration and with the same kinetics as in the absence of RecA, demonstrating that the protein interaction has no noticeable stabilizing effect on the Z-conformation.

Implications on Models for Recombination Mechanism
It has been speculated that one DNA molecule is in a lefthanded (Z-type) configuration in the genetic recombination (23). Our results clearly demonstrate that in vitro the DNA conformation is retained upon RecA binding and that only Z + B, but not the B + Z, conformational transition may occur in the RecA-DNA complex. We therefore suggest that if the Z-conformation is indeed involved in the in vivo recombination, the DNA should be in the Z-conformation prior to the RecA association. Once the RecA. Z-DNA complex is formed, the RecA should not prevent transformation to the B-form when the physical constraints disappear, for example after the cleavage of DNA.