J. Mol. Biol. (1981) 151, 557-564

LETTEM TO THE EDITOR

Elongation of Duplex DNA by recA Protein

recA protein, which is essential for the recombination process in Escbrichia wli, was incubated in the presence of 5’-y-thiotriphosphate with circular plasmid pBR,G containing small single-stranded gaps. Stable complexes were formed which appear in the electron microscope as fibres with a diameter about five times that of naked DNA. Complex formation appears to be a co-operative process whereby the average rise per base-pair with respect to the fibre axis increases from 3.39 f0.08 .h to 520f0~18 A. The elongation of DNA by about 50”/0 is compatible with an unwinding of the double helix and an intercalating mode of binding of recA and/or 5’.y-thiotriphosphate to DNA.

For the recombination process in Escherichia coli, the reca gene product is essential (Clark, 1973; Kobayashi & Ikeda, 1978). The gene has recently been cloned and sequenced (Sancar et al., 1980). Its product, a protein of 37,842 M,, has been purified (Weinstock et al., 1979; Ogawa et al., 1979; Roberts et al., 1979; Shibata et al., 1979a) and used for crystallographic studies (McKay et al., 1980). The structure of the recombining DNA strands, however, is still an open question (for reviews, see DasGupta et al., 1980; Radding, 1978). In one model the pairing of duplex DNA molecules is thought to occur before any strand breakage, whereby four-stranded DNA structures would be an intermediate. In another model it has been postulated that a break in one strand of a donor duplex molecule would be followed by partial strand separation and the subsequent pairing of the broken, displaced single strand with another double-stranded molecule. The activity of recA protein has been called a strand transferase activity (DasGupta et al., 1980): since it mediates in vitro the formation of joint molecules between duplex DNA and single strands or double strands with single-stranded gaps. One has to assume that recA protein changes the conformation of duplex DNA in such a way that the access to the bases is facilitated. Indeed, it has been shown that single-stranded DNA4 stimulates recA protein to unwind duplex DNA to some extent (Cunningham et al.. 1979: Shibata et al., 19793).

West et al. (1980) have observed that recA protein can form stable complexes with double-stranded DNA containing small single-stranded gaps, if ATP is replaced by adenosine 5’-y-thiotriphosphat, called ATP[S]. We are trying to characterize such stable recA-DNA complexes by electron microscopy. In this letter we present evidence that in the presence of ATP[S] recA binds co-operatively to gapped double-stranded DNA, and we show that the average rise per base-pair is increased from about 3.4 A to about 5.2 A with respect to the fibre axis. This elongation is compatible with an unwinding of the double helix and with an intercalating mode of binding of recA to DNA.

557 0022-2836/81/270557~8 $02.00/O 0 1981 Academic Press Inc. (London) Ltd.

*558 A. STASIAK. E. DI CAPUA ASD TH. KOLLER

When we incubated in the presence of ATPIS] (f or reaction conditions see the legend to Fig. 2) a large excess of recA protein (Fig. 1) with gapped plasmid pBR,G (protein to DNA (w/w) ratio above 40 : I) and prepared the reaction product within one to two hours for electron microscopy by the prot,ein-free magnesium mounting method (Arcidiacono et al., 1980), we observed exclusively filaments that, were clearly much thicker than control DNA. In agreement with West et al. (1980) who mounted such complexes by the cytochrome c spreading method (Kleinschmidt & Zahn, 1959), many of the filaments were aggregated. These aggregates were not analysed further. but there was also a fair proportion of apparently single molecules

FIG. 1. Electrophoresis of recA protein in a 15yb (w/v) polyacrylamide gel. M, molecular weight markers x 103. recA protein was produced in E. coli K12 (&A, recA), strain KMIlOQ (McEntee, 1977) carrying the plasmid pDR1453 recA+ (Sancar & Rupp, 1979). The bacteria were a kind gift from Dr Howard-Flanders. The protein was purified according to Weinstock et al. (1979) and West et al. (1980). The protein concentration of the stock solution was estimated by the method of Lowry (Lowry el al.. 19.51). The apparent molecular weight, as estimated on the 15”/b polyacrylamide gel containing sodium dodecyl sulphate (Thoma et al., 1979). agrees with published data (Sancar & Rupp, 1979) for mcA protein

Form I DNA was incubated as described in the legend to Fig. 2 with recA protein at 37°C’ for 1 h at, a protein to DNA ratio (w/w) of 45 : 1. After extraction with phenol the DNA was analysed by electrophoresis in a l”/b (w/v) agarose gel. No conversion of form I DNA into nicked, circular or linear DNA could be detected, indicating little or no endonuclease activity of our recA sample.

In agreement with West et al. (1980) we observed recA-DNA complexes with gapped DNA, but not with covalently closed, circular (form I) or with nicked, circular DNA. This behaviour of recA protein was used to follow the gapping produced by exonuclease III : the enzyme reaction was stopped as soon as -90% of the DNA molecules formed complexes with recA. After a few weeks our recA protein lost the selectivity of reacting only with gapped DNA and, after reaction with nicked, circular (form II) DNA. thick filaments considered to represent recA-DNA complexes were also found. Since the morphology of the complexes was always the same whether freshly purified or several weeks old recA was used. we disregarded this loss of selectivity.

LETTERS TO THE EDITOR 559

which will be described below. Since we did not observe such thick filaments when either recA protein or ATP[S] was omitted, we considered these thick filaments as recA-DNA complexes. In some experiments complexes were fixed with 91% (v/v) glutaraldehyde for 15 minutes at 37°C before specimen preparation. Since their appearance was indistinguishable from unfixed samples, most of the experiments were performed without fixation. Once the complexes were formed, they were very stable and could be kept at 4°C for several days.

We have noticed (Arcidiacono et al., 1980) that using the magnesium method for electron microscope specimen preparation DNA molecules are lost if the grids are washed with distilled water. For this reason we kept the washing step as short as possible. Under these conditions, as will be described below, the contour length of plasmid pBR,G corresponded to the B-conformation of DNA. However, the drawback of the method was that a large proportion of the area of the specimens was still covered with residues of salt and buffer. The resulting rough background mainly influenced the visibility of the DNA not complexed with recA protein. For t’his reason it was not possible to perform an exact quantitative analysis of the number of complexed and uncomplexed DNA molecules as a function of the recA to DNA ratio used during complex formation.

recA protein and gapped, circular DNA were incubated at increasing protein to DNA (w/w) ratios. Below a ratio of 20 : 1 most of the molecules appeared as naked DNA, indistinguishable from control specimens without recA protein. At ratios between 20 to 40 : 1 the proportion of naked DNA decreased and the complexed molecules appeared as thick filaments either along the entire length (Fig. 2(a)) or along a distinct DNA segment (Fig. 2(b)). At ratios above 40 : 1 (experiments up to 80 : 1 were performed) no more naked DNA circles were observed. About 250/6 of the unaggregated molecules were clearly circular and fully covered with recA protein. The majority of the apparently single complexes, however, were rod-like, where many of the rods had a tail (arrow in Fig. 2(b)). These rod-like molecules were interpreted as partially recA-covered, folded, circular molecules where the naked part of the DNA appeared to be superhelically twisted. The partially covered molecules (the relaxed and those with the twisted tails) showed in all cases only one thick and one thin region, the thick regions being uninterrupted by naked DSA stretches. This observation, as well as the fact that at low protein to DNA ratios fully covered molecules were found beside unreacted DNA (Fig. 2(a)), suggest that the binding of recA to DNA is favoured in the immediate vicinity of already bound protein. This is conpatible with the co-operative binding mechanism proposed by West et al. (1980). It is known (West et al., 1980) that the initiation of recA binding to double-stranded DNA is only possible in single-stranded regions like gaps or tails. Our results do not allow us to decide whether during the co-operative binding process additional recA-binding is only possible in single-stranded regions (and thus displacing already bound recA), or whether recA can bind directly to double strands in the immediate vicinity of DNA-bound recA.

We mentioned above that the rod-like complexes mostly showed twisted appearing tails (Fig. 2(b)). When the reaction product was kept at 4°C for several days, these partially recA-covered molecules progressively disappeared in favour of fully covered circular complexes. If we assume that the binding of recA somehow

560 A. STAHIAK, E. DI C‘APU.4 8513 TH. KOLLER

FIG. 2. Complexes of recA protein with gapped circular plasmid pBRP. The protein to DNA ratio was 20 : 1 (a) and 40 : 1 (b). Note the difference in thickness between the mcA-covered strands and the naked duplex DNA.

Plasmid ~BIt~G2.17 containing 4961 base-pairs (F. Meyer C H. Weber, persona1 communication] was grown in E. oli strain HBlOl (a kind gift from Ch. Weissmann). It was extracted by a cleared lysate (Triton X100) and was purified by 2 extractions with phenol. sucrose gradient centrifugation and by banding in a CsCl density gradient, Relaxed circular (form II) DNA, called nicked DNA, was obtained by reacting for 30 min at 30°C the closed circular (form I) DNA (93 mg/ml) with DNAase 1 from Boehringer (2 x 10S4 mg/ml) in the presence of 0.5 pg ethidium bromide/ml. Gapped DNA was produced by treating the nicked DNA (65 pg/ml) for up to 20 min at 25°C with exonuclease III from Miles (1 ~1 stock/ml).

recA protein and DNA were reacted as described by West et al. (1930). The buffer contained 50 mM- Tris’HCl (pH 7.5), 91 mM-EDTA, 20 mM-NaCl. lOmM-MgCl,, 05m~-ATP]S]. 2 to 5pg DNA/ml and

LETTERS TO THE EDITOR 561

puts a lock on the gap(s) in a way hindering free rotation of the DNA strands, the twisted tails indicate a recA-mediated change of the winding angle of the DNA double helix. Indeed, it has been shown that recA, in the presence of single-stranded DNA (Cunningham et al., 1979), leads to an unwinding of duplex DNA. The rod- like molecules are, in most cases, only partially covered (especially after short reaction times). This is best explained by assuming that the constraint given to the molecule by the unwinding hinders and slows down the binding of additional recA molecules and, therefore, only after long reaction times the partially covered constrained molecules disappear in favour of fully covered relaxed circles.

We frequently observed that the complexes showed quite regular cross-striations (Fig. 2). The significance of this substructure will be described in a forthcoming paper.

An average contour length of 1.68 pm was measured for nicked, circular DNA (not reacted with exonuclease III) prepared in the absence of recA protein (ratio 0 : 1, Table 1). Similar lengths were obtained for DNA considered to be free of recA protein on specimens prepared with samples at a low protein to DNA ratio (2 : 1 and 6 : 1). Taking into account the known number of 4961 base-pairs for pBR,G, we compute an average rise per base-pair close to 3.4 A (Table 1). Such a distance is compatible with the B-conformation of duplex DNA (Wing et al., 1980). With regard to the recA-DNA complexes, it can be seen that the contour lengths measured are independent of the recA/DNA ratio (Table l), in agreement with the co-operative binding mechanism discussed above. They were also independent of the age of the complexes. Taking all the data of Table 1 into account, we calculate an average contour length of 2.58 + 099 pm and an average rise per base-pair with respect to the fibre axis of 5*20+0.18 A.

As described above, at low recA to DNA ratios, circular molecules partially covered with recA were found (Fig. 2(b)). The contour length of four of such molecules was measured. We assumed for the recA-covered part of the strand an average rise per residue of 5.2 b and for the uncovered part of the molecule one of 3.4 A. Adding the obtained numbers of base-pairs we computed nearly the expected 4961 base-pairs (5130, 4833, 5193 and 4684) for the entire molecule. The same result was obtained with six additional molecules measured on specimens that were dehydrated in ethanol instead of air-drying (not shown). Within the thin, uncovered part of such strands we never observed two strands or a branching of the strands. For this reason we assume that only one duplex DNA molecule is complexed with recA protein within these thick circular strands. The elongation of

4 to 2OO~g recA protein/ml were incubated at 37°C for 60 min. The formed complexes. as judged by electron microscopy, were stable at 4°C for several days.

Specimens of recA-DNA complexes were prepared by a modified magnesium method (Aricidiacono et al.. 1980): 1 ~1 of the reaction mixture (containing about 5 yg DNA/ml) was diluted in 4 1.11 of mounting buffer (30 m&f-triethanolamine . HCl, pH ‘7.5, 10 mM-magnesium acetate). The droplet was placed on a fresh sheet of Parafilm. After 10 min it was touched with a carbon-coated grid freshly treated for 30 8 in a plasma cleaner (Harrick, intensity position 10). After 15 8 the grid was touched very briefly (about 1 s) onto redistilled water and was immediately blotted dry on filter paper. Grids were rotary shadowed at an angle of 7” with carbon/platinum evaporated from an electron gun.

562 A. STASIAK. E. UI CAPUA ASI) TH. KOLLEK

TAHLE 1

C’ontvur length and average rise per base-pair with re.Ypect to j&e axis of fully protein-covered recA-DNA complexes

and of naked form I I DM

recA/DNA ratio

(w/w)

Contour lengt>h

(tLm)

Rise per base-pair with Number of molecules respect to tibre axis

(A) Naked C’OVfXetl

0 1 1 ,ci6 + @02 3.35 * 0.04 10 2:l 1.71_+@05 3,44+0.11 15 6:l 1.68 + 0.03 3.38 + (ho7 10

30 1 2.58 + 0.07 521 +013 II 36 : 1 2.56fO.10 516*02 10 45 : 1 2.58T@ll a~l9+@%3 13 60 : 1 2.59 + 0% 5.22f0.12 - 11

Overall average: Naked molecules Covered molecules

1f8+004 3.39 * om 35 2.58 + 0.09 5.20*0.18 45

The magnification of electron micrographs was estimated by measuring the circumferences of Latex spheres (0.31 pm diam.. Polaron Equipment Ltd, Watford, England).

The platinum-shadowed specimens of recA-DNA complexes to be analysed quantitatively were coated with Alcian blue as described by Labhart & Keller (1980). They were then touched t,o 5~1 droplets of the commercial Latex stock solution diluted 10 x in redistilled water. Specimens were always mounted in the electron microscope with the Latex spheres facing downwards. The average diameter of the Latex spheres was tested by adsorbing them onto a carbon-grating replica grid from Fullam Standard deviations given were calculated according to a normal propagation law of errors (contour lengths of molecules, circumferences of Latex spheres). Length measurements were made as described by Keller el al. (1978).

the DNA upon reaction with recA and ATP[S] must therefore be real and cannot be due to aggregation of two or more DNA molecules.

In the nearest-neighbour exclusion model for the binding of intercalating reagents to duplex DNA (Cairns, 1962; Brothers, 1968: Bauer & Vinograd, 1970) the presence of an intercalator between t)wo base-pairs excludes the next nearest,- neighbour site. At saturation, propagation of this effect along the entire DN4 molecule would result in occupancy of 500/b of the possible intercalat,ion sit,es. Model-building studies and fibre X-ray diffraction studies with DNA saturated with platinum intercalation reagents indicate that, intercalators bound to DSA distribute at, I@2 A intervals along t,he helix axis (Bond et al.. 1975: Lippard et al.. 1976). This distance is in excellent agreement with the average rise per base-pair of 52&@12 ,k (or about 1@4 &2 b ase-pairs) for recA-DNA complexes. Indeed intercalation has been proposed as a possible mechanism for protein-nucleic acid recognition (Gabbay et al., 1973), and recA protein does have in its amino acid sequence potential intercalative residues (Sancar et al., 1980). In addition one has also to consider the possibility that ATP[S]. stabilized by recA protein, could intercalate intro the double helix. However, one has to realize that’. in contrast to

LETTERS TO THE EDITOR 563

the recA binding, the intercalation of ethidium bromide and other dyes occurs in a non-co-operative fashion (Bauer & Vinograd, 1970; Lippard, 1978). Therefore, full saturation of a DNA with intercalating dyes seems to be impossible (for a review see Lippard, 1978). The structure of unwound DNA upon reaction with intercalating residues may be essential for the initial stages in the recombination process. Unfortunately the detailed stereochemistries of the intercalative mechanisms are not yet known; suggestions have been made on the bases of theoretical model-building and with the crystal structures of dinucleotide duplex complexes (Alden & Arnott, 1975,1977; Sobell et al., 1977; Arnott et al., 1980: Shieh et al., 1980).

In conclusion we confirm by electron microscopy that recA protein binds co- operatively to gapped duplex DNA in the presence of ATP[ S]. Upon reaction DNA is elongated by about 500/b and thus appears to be unwound. This may suggest that recA and ATP[S] bind to duplex DNA according to the nearest-neighbour exclusion model for the binding of intercalating reagents. A structural analysis of recA-DNA complexes is in progress and will be described in a forthcoming paper.

We thank Drs H. J. Vollenweider, F. Thoma and F. <Jay for discussions and critical reading of the manuscript, and Mrs H. Mayer-Rosa for technical help. This work was supported by Schweizerischer Nationalfonds zur Fijrderung der wissenschaftlichen Forschung (to Th.K.).

Institut fur Zellbiologie ETH Zurich, Honggerberg 8093 Zurich, Switzerland

A. STASIAK~ E. DI CAPUA TH. KOLLER

Received 31 March 1981

REFERENCES

Alden, C. J. & Arnott, S. (1975). Nucl. Acids Res. 2, 1701-1717. Alden. C. J. & Arnott, S. (1977). Nucl. Acids Res. 4, 3855-3861. Arcidiacono, A., Stasiak, A. & Koller, Th. (1980). Proc. 7th Eur. Congr. Electr. Microsc.

(Brederoo, P. & de Priester, W., eds), vol. 2, pp. 516-523, Seventh European Congress on Electron Microscopy Foundation, Leiden.

Arnott, S., Bond, P. J. & Chandrasekaran, R. (1980). Nature (London), 28’7, 561-563. Bauer, W. R. & Vinograd, J. (1970). J. MOE. Biol. 4’7, 419-435. Bond, P. J.. Longridge, R., Jennette, K. W. & Lippard, S. J. (1975). Proc. Nut. Acad. Sci.,

U.S.A. 72, 4825-4829. Cairns, J. (1962). Cold Spring Harbor Symp. Quant. Biol. 2’7, 31 l-318. Clark, A. (1973). Annu. Rev. Genet. 7, 67-86. Crothers, D. M. (1968). Biopoolymers, 6, 575-584. Cunningham, R. P., Shibata, T., DasGupta, C. & Radding, C. M. (1979). Nature (London),

281, 191-195. DasGupta, C., Shibata, T., Cunningham, R. P. & Radding, C. M. (1980). Cell, 22, 4377446. Gabbay, E. J., Sanford, K., Baxter, C. A. & Kapicak, L. (1973). Biochemistry, 12,4021-4029. Kleinschmidt, A. K. & Zahn, R. K. (1959). Zeitschr. Naturf. 14, 779779.

t Present address: Polish Academy of Sciences, Institute of Biochemistry and Biophysics, Ul Rakowiecka 36. 02-532 Warszawa, Poland.

564 .I. STASIAK. E. DI CAPCA rlND TH. KOLLEK

Kobayashi, 1. & Ikeda, H. (1978). NoZ. Gen. &net. 155, 25-29. Keller, Th., Kiibler, O., Portmann, R. dt Sogo, J. M. (1978). J. Nol. Biol. 120, 121~.-131. Labhart, P. & Keller, Th. (1980). Proc. 7th Eur. Congr. Electron Microsr. (Brederoo, P. & de

Priester, W., eds), vol. 2, pp. 542-543, Published by Seventh European Congress on Electron Microscopy Foundation, Leiden.

Lippard, S. J. (1978). Ace. Chem. Res. 11: 211-217. Lippard, S. J., Bond. P. J., Wu. K. C. & Bauer, W. R. (1976). Science, 194, 726-728. Lowry, 0. H., Rosebrough, X. ,J., Farr, A. L. & Randall. R. I. (1951). ,I. Biol. Chem. 193,265~

275. McEntee, K. (1977). J. Bacterial. 132, 904-911. McKay. D. B., Steitz, T. A., Weber, I. T.. West, S. C. 8 Howard-Flanders, 1’. (1980). J. Biol.

Chem. 255, 6662. Ogawa, T., Wabiko, H., Tsurimoto, T., Horii, T., Masukata, H. & Ogawa, H. (1979). (loid

Spring Harbor Symp. Quant. Biol. 43, 909-915. Radding, C. M. (1978). Annu. Rev. Biochem. 47, 847-880. Roberts, J. W., Roberts, C. W.. Craig, XT. L. & Phizicky, E. M. (1979). Cold Spring Harbor

Symp. Quant. Biol. 43, 917-920. Sancar, A. & Rupp, W. D. (1979). Proc. LVat. Acad. Sci.. ljT.S.d. 76, 3144-3 148. Sancar, A., Stachelek, C., Konigsberg, W. & Rupp, D. W. (1980). Proc. Jat. Acad. 9ci..

1J.S.d. 77, 261 l-2615. Shibata, T., DasGupta, C., Cunningham, R. I’. & Raddinp. (‘. M. (1979a). Proc. IVat. A4cccd.

hi., U.S.A. 76, 1638-1642. Shibata, T., Cunningham, R. P., DasGupta. C. & Radding, (‘. M. (1979b). Proc. L&f. =Icad.

Sci., U.S.A. 76, 5lOC-5104. Shieh, H. S., Berman. H. M,, Dabrow, M. 8r Neidle, S. (1980). IVUCL. =1cids Res. 8, 85-97. Sobell, H. M., Tsai, C. C., ,Jain, S. C. &, Gilbert, S. G. (1977). J. Xol. Biol. 114, 333-336. Thoma, F., Keller, Th. & Klug, A. (1979). J. Cell Biol. 83, 403-427. Weinstock, G. M., McEntee. K. & Lehman, I. R’. (1979). Proc. *Vat. Acad. Ski.. V.S..4 76,

126-130. West, S. C., Cassuto, E.. Mursalim, J. & Howard-Flanders. 1’. (1980). Proc. .\:at. Acad. Ski..

U.S.A. 77, 2569-2573. Wing, R.. Drew, H., Takano, T., Broka, (‘., Tanaka, S., Itakura, K. & Dickerson, K. IX.

(1980). iTature (London). 287, 755-758.

Edited by S. Brenner