Homology recognition without double-stranded DNA-strand separation in D-loop formation by RecA

Abstract RecA protein and RecA/Rad51 orthologues are required for homologous recombination and DNA repair in all living creatures. RecA/Rad51 catalyzes formation of the D-loop, an obligatory recombination intermediate, through an ATP-dependent reaction consisting of two phases: homology recognition between double-stranded (ds)DNA and single-stranded (ss)DNA to form a hybrid-duplex core of 6–8 base pairs and subsequent hybrid-duplex/D-loop processing. How dsDNA recognizes homologous ssDNA is controversial. The aromatic residue at the tip of the β-hairpin loop (L2) was shown to stabilize dsDNA-strand separation. We tested a model in which dsDNA strands were separated by the aromatic residue before homology recognition and found that the aromatic residue was not essential to homology recognition, but was required for D-loop processing. Contrary to the model, we found that the double helix was not unwound even a single turn during search for sequence homology, but rather was unwound only after the homologous sequence was recognized. These results suggest that dsDNA recognizes its homologous ssDNA before strand separation. The search for homologous sequence with homologous ssDNA without dsDNA-strand separation does not generate stress within the dsDNA; this would be an advantage for dsDNA to express homology-dependent functions in vivo and also in vitro.

unwind the double helix, the cc-dsDNA formed (Form IV) by the ligation of nicked circular dsDNA or topoisomerase I-treatment has a linking number (Lk0) that is identical to Tw0.Tw0 is Tw of B-form DNA (1 turn/10 base pair), and thus, LK0 = Tw0 = N/10, where N is total number of base pairs of cc-dsDNA examined.Because of fluctuations in the twisting number (not an integer) caused by thermal rotational movement around the axis of the double helix, the linking numbers of a cc-dsDNA population are distributed around a central (≅ average) number.Lk0 is represented as the central most prominent band of the set of 3 or 4 bands of Form IV in the electrophoretic profile, under the present experimental conditions (Sample 1 in Fig. 4. B -D, Sample 3 in Fig. 5A, Sample 4 in Fig. 6).If dsDNA is unwound before ligation by n turns, the cc-dsDNA formed by the ligation has a unique negative linking difference, ΔLk where ΔLk = Lk -Lk0.After ligation, the unwinding reagents are removed, the cc-dsDNA acquires negative (right-handed) supercoil (or negative (left handed) solenoid, which is not considered in this study), of which number is termed, writhing number (Wr).Among the three topological parameters of cc-dsDNA, there is a simple relationship: Lk = Wr + Tw (see refs. 65,66).
Experimental conditions, such as the presence of Mg 2+ and non-specific interactions with proteins, cause a small change in Tw.

Additional information about two-dimensional electrophoresis
In agarose-gel electrophoresis, the migration rate of cc-dsDNA depends on the absolute number of supercoils (Wr) and is independent of handedness: negative (right-handed) or positive (left-handed).
The signals (bands) of the cc-dsDNA molecules migrating near to the Form II signal (Form IV in the absence of ethidium bromide; Fig. 4A panel 1) and also near the topoisomer signals that migrates at maximum rates (the 1 st dimensional run of Form I and Form X in Fig. 4A panels 1, 2 and 4, and Fig. 5, A, B, D and E, and Fig. 6; topoisomers of which ΔLk is less than -12 in Figs.4B and 4C; Form IV in Fig. 5, B and C) are poorly separated and tend to overlap (82).Thus, the small degree of unwinding is not measurable within these regions.
Ethidium bromide intercalates between base pairs of dsDNA and unwinds dsDNA; it thus reduces the number of double-helix turns.If the dsDNA is cc-dsDNA, this reduction in the number of doublehelix turns (Tw) results in an increase by the equal number of Wr in the relationship denoted by the equation described above, Lk = Wr + Tw (see refs. 65,66).As the ethidium-bromide concentration increases, the extent of dsDNA unwinding increases (represented by a decrease in Tw), and Form IV acquires a positive Wr to migrate at a higher rate than Form II in gel electrophoresis.At an appropriate concentration (e.g. 15 nM), the signals of Form IV are well separated (Fig. 4A Panel 2; in the first dimensional electrophoresis in Fig. 4, B -D; Fig. 5, A, D -E, Fig. 6), and a small change in the linking number is sensitively detected as a stepwise shift of the band set (in the first dimensional electrophoresis in Fig. 4, B -D).To broadly count ΔLk, the reduction in the linking number from that of Form IV, we analyzed cc-dsDNA samples using two-dimensional electrophoresis, as shown in Fig. 4A Panel 4. In two-dimensional electrophoresis in the presence of ethidium bromide at 9-15 nM and 110-150 nM in the first and second dimensions, respectively, cc-dsDNA molecules, with various ΔLk, form an arch of discrete signals, as shown in Fig. 4A panel 4.

Determination of ΔLk of each DNA topoisomer
ΔLk for each DNA topoisomer was determined by counting the number of bands starting from the central most prominent band of Form IV (Reference band shown in Sample 1 in Fig. 4, B-D) to the band of the topoisomer in electrophoretic profiles.Upward steps of bands to the top followed by towards the lower left steps indicates a decrease in ΔLk (-1, -2, -3 and so on; see Fig. 4A, panel 4).
Sometimes, when their Wr is close to 0, a pair of topoisomers with adjacent ΔLk numbers is indistinguishable, because they migrate at almost the same rate as Form II in the first dimension.The overlap of signals was identified by comparison of electrophoretic profiles in the presence of different ethidium bromide conditions in the first dimension (Fig. 4, B -D) and by the brightness of the signal.
For example, the bands of which ΔLk is -5 and -6 overlap partly in Fig. 4B, but are separated in Fig. 4C and Fig. 4D.The bands of which ΔLk is -4 and -5 overlap completely in Fig. 4C, but are separated in Fig. 4D.The ΔLk of the band of which Wr is 0, is -5 in Fig. 4B and overlapped -4 and -5 in Fig. 4C, and -6 in Fig. 4D.
The turn number of the double helix-unwinding just before the ligation (or during topoisomerase Itreatment) by a specified interaction is determined as described in the main text, but more precisely determined by ΔLkave -ΔLkc, where ΔLkave is the average ΔLk of a set of bands of cc-dsDNA obtained under the specified conditions and ΔLkc is the average of ΔLk of the set of bands of cc-dsDNA obtained from the negative control.In this study, ΔLkc that was resulted from incubation with RecA and