Loop L1 governs the DNA-binding specificity and order for RecA-catalyzed reactions in homologous recombination and DNA repair

In all organisms, RecA-family recombinases catalyze homologous joint formation in homologous genetic recombination, which is essential for genome stability and diversification. In homologous joint formation, ATP-bound RecA/Rad51-recombinases first bind single-stranded DNA at its primary site and then interact with double-stranded DNA at another site. The underlying reason and the regulatory mechanism for this conserved binding order remain unknown. A comparison of the loop L1 structures in a DNA-free RecA crystal that we originally determined and in the reported DNA-bound active RecA crystals suggested that the aspartate at position 161 in loop L1 in DNA-free RecA prevented double-stranded, but not single-stranded, DNA-binding to the primary site. This was confirmed by the effects of the Ala-replacement of Asp-161 (D161A), analyzed directly by gel-mobility shift assays and indirectly by DNA-dependent ATPase activity and SOS repressor cleavage. When RecA/Rad51-recombinases interact with double-stranded DNA before single-stranded DNA, homologous joint-formation is suppressed, likely by forming a dead-end product. We found that the D161A-replacement reduced this suppression, probably by allowing double-stranded DNA to bind preferentially and reversibly to the primary site. Thus, Asp-161 in the flexible loop L1 of wild-type RecA determines the preference for single-stranded DNA-binding to the primary site and regulates the DNA-binding order in RecA-catalyzed recombinase reactions.

free RecA crystals is considered to reflect an inactive state. The structure of the monomer and its packing arrangements are essentially identical to those of the previously reported 74 Å filaments (46, r.m.s.d. of 0.57-0.73 Å for Cαs in a monomer) except for loops L1 and L2, which are described in the main text. Our inactive RecA crystal provided an electron density map for most of both loops L1 and L2, unlike previously reported DNA-free RecA crystals (Fig. 1A). In the DNA-free RecA crystal structures reported to date, the electron densities for loops L1 and L2 are highly ambiguous, and the structures of these loops have been provided for four RecA structures: the free RecAs from E. coli (this study), and the RecAs from Mycobacterium smegmatis (msRecA) in the free (PDB ID: 2OFO), ADP-bound (PDB ID: 2OEP), and dATP-bound (PDB ID: 1UBG) forms (36)(37)(38).
The configuration of the protomers in our putative inactive filament distinctly differs from that in the activated RecA-DNA complex forming extended filaments, in which the helical pitches range between 92.4 -95.3 Å (22, Supplementary Fig. S1A, right panel). The core and the C-terminal domains in each protomer of this inactive RecA and those of the activated RecA-DNA complexes (22) superimposed well (within an rmsd. of 1.03 Å for Cαs of residues 45-156, 165-194 and 210-328), but the orientations of the N-terminal domains and the conformations of loops L1 and L2 significantly differed. The N-terminal domain (residues 1-33), containing an α-helix and a β-strand, is reportedly essential for RecA polymerization (50,51). In spite of the large difference in the configurations between the central core domains, the interprotomer interactions via the N-terminal domain were conserved in both the active and inactive filaments (Supplementary Fig. S1B). The flexible linker region modulates the location of the N-terminal domain, to retain the interprotomer contacts responding to the movement of the central core domains. The interaction surface area, excluding the N-terminal domain, was 684.0 Å 2 in this putative inactive filament, which is less than half of that in the active filament, ~1,460Å 2 . In the active filament, the hydrogen bonds formed between the γphosphate of ATP and K248 and K250 of the adjacent molecule are assumed to be critical to stabilize the active RecA-RecA interface (22) (Supplementary Fig. S1C). In the inactive free RecA filament, the putative catalytic residue Glu96 is shielded from the solvent by loop L2 of the adjacent protomer ( Supplementary Fig. S1D).

Preparation of DNA for biochemical analysis
The 90-nucleotide single-stranded DNA (10 μM in polynucleotide, 900 μM in nucleotides) was labeled with 33 P at the 5' terminus by an incubation in reaction buffer (10 μL) with [γ-33 P]ATP and 1U/μL T4 polynucleotide kinase at 37°C for 30 min, as recommended by the manufacturer (Megalabel DNA 5' end labeling kit, Takara Bio Inc.). The preparation was heated at 90°C for 2 min to inactivate the kinase, and the unreacted ATP was removed by two rounds of spin column separation. The labeled single-stranded DNA was diluted 80-fold in TE buffer (10 mM Tris-HCl, pH 7.5, and 0.1 mM EDTA).
The amounts of the labeled single-stranded DNA indicated are the calculated values, assuming 100% recovery during the labeling process.
Negatively supercoiled closed circular double-stranded (form I) pBluescript SK(-) DNA and M13 circular single-stranded DNA were prepared by the procedures described previously (25,26). pGsat4 form I DNA was described previously (52). ΦX174 circular single-stranded DNA was purchased from New England BioLabs, Inc. (Ipswich, MA, USA). It is noted that the form I DNA prepared by procedures including alkaline-treatment often traps non-canonical structures, and thus should not be used in the assay for homologous joint-formation.

DNA constructs for mutant recA expression
The site-directed mutagenesis of recA was performed as described previously (15), with a slight modification.

Purification of RecA-wt and mutant recAs
The mutant recAs and RecA-wt were expressed from a derivative of the multicopy vector pKK223-3, which carried the mutant recA or recA + under the control of the tac promoter in a ΔrecA derivative of E.
coli. All RecAs were purified by the procedures described previously (25,26), with the following modifications. After ammonium sulfate precipitation, the proteins were fractionated by TOYOPEARL butyl-650M (TOSOH) column chromatography with a linear gradient of ammonium sulfate (1 M to 0 M), followed by hydroxyapatite (CHT-1, BioRad) column chromatography.
It is noted that we used fresh preparations of the mutant recAs and RecA-wt purified from freshly prepared cell-lysates obtained from newly transformed E. coli cells.
We experienced some experimental difficulties when we used the wild type RecA purified from old 60% glycerol stock of partially purified RecA in a freezer, which reproducibly showed very slow homologous joint-formation without the following dissociation phase (Supplementary Fig. S9A). We found that the RecA preparation (called RecA*) contained a higher amounts of mono-and di-oxidized forms, as revealed by mass-spectrometry performed (Supplementary Fig. S9C).

Fab fragment of the ARM193 antibody
The ARM193 antibody (approximately 4 mg/ml, purchased from Medical & Biological Laboratories Co.,Ltd (MBL)) was treated with papain, as described (44), and the Fab product was purified by DEAE-5PW (TOSOH) chromatography followed by Mono-S (Pharmacia) chromatography. The purified Fab fragment was tested for the ability to prevent RecA-filament formation, by a gel filtration assay on TSK-GEL G4000SW resin (TOSOH) (See ref. 44).

Crystallization
Prior to the preparation of crystallization droplets, RecA was mixed with the Fab fragments of the ARM193 antibody (ARM193Fab) at a 1:1 molar ratio at 277 K, to produce a protein solution including 1.9 mg/ml RecA protein and 2.5 mg/ml ARM193Fab, in 10 mM MES (pH 6.5) buffer. The crystallization droplets were prepared by mixing 10 μL of the protein solution and 10 μL of reservoir solution, which contained 24 % PEG400, 10 % glycerol, 10 mM MgCl2 and 0.1 M MES, pH 6.5. The crystals were grown by the hanging drop vapor diffusion method at 293 K. The ARM193Fab promotes the dissociation of spontaneously formed RecA filaments into monomers (44), and we expected that the addition of the ARM193Fab would improve the crystallization of RecA. However, the Fab was not included in the crystal that we obtained (see Supplementary Fig. S1).
The RecA crystal was briefly soaked in a cryo-protectant solution, containing 34 % PEG400, 18 % glycerol, 10 mM MgCl2 and 0.1 M MES (pH 6.5), and then flash cooled in a stream of N2 gas at 100 K.
X-ray diffraction data were collected at the RIKEN beamline II (BL44B2) at SPring-8 in Harima, Japan (53). The diffraction images were recorded on a CCD detector (MAR165), and were processed using the HKL2000 software package (54) up to 2.8 Å resolution. The initial phase was derived by molecular replacement with the program MOLREP in CCP4 (55), using the atomic coordinates of the E. coli RecA protein (PDB ID: 2REB) previously reported by Steitz's group (21) as a probe. The molecular structure was constructed and modified using TURBO FRODO (56), and refined using CNS (57).

Nano-electrospray mass spectrometry
Nano-electrospray mass spectra were acquired on a Waters (Manchester, UK) SYNAPT G2 quadrupole ion mobility time-of-flight mass spectrometer (see refs. 58,59). A 4 μL aliquot of 16 μM RecA in 50 mM ammonium acetate was deposited in a nanospray tip with an 8-μm internal diameter (HUMANIX, Hiroshima, Japan) and placed in a nanospray ion source. A capillary voltage of 0.8-0.9 kV and a sample cone voltage of 20-30 V were typically applied. Each mass spectrum was acquired in 1 s, and more than 100 mass spectra were accumulated and smoothed by the Savitzky-Golay method. All data analysis and processing were performed with MassLynx v4.1 (Waters).     The 90-nucleotide single-stranded [ 32 P]DNA (0.3 μM; these experiments only) and a mutant or wild type RecA (2.0 μM) were first incubated for 10 min, and the reaction was started by the addition of homologous (pBluescript SK(-)) closed circular double-stranded DNA (18 μM). After incubations for the indicated times, aliquots were withdrawn and the reaction was terminated. Unlike the standard conditions, the proteins were removed from the samples by a treatment with Proteinase K at 37°C, and then the reaction products were analyzed by agarose-gel electrophoresis. 32 P-signals were analyzed quantitatively. Mutant recAs used in these experiments are indicated within the panel.

Supplementary
Each point indicates the average obtained from two to three independent experiments. In the standard order of addition, the 90-nucleotide single-stranded [ 33 P]DNA (final 0.05 μM) and the indicated amounts of RecA-wt or recA-D161A were incubated for 5 min or more (preincubation), and the reaction was then started by the addition of homologous (pBluescript SK(-)) form I (negatively supercoiled closed circular double-stranded) DNA (18 μM). After incubations for the indicated times, aliquots were withdrawn and the amounts of homologous joints formed were measured, as described in Fig. 6.
Black symbols, standard order of addition of protein, single-stranded DNA and double-stranded DNA. Red symbols, double-stranded DNA was preincubated with RecA, and the reaction was then started by the addition of single-stranded [ 33 P]DNA. The protein amounts are indicated within the panels.   and Methods) showed a slow homologous joint-formation without the following dissociation of homologous joint, even when excess protein was present.
A. Abnormal homologous joint formation by RecA* under the standard conditions.