- Email: [email protected]
RecA-DNA filament topology: The overlooked alternative of an unconventional syn–syn duplex intermediate
d n a r e p a i r 6 ( 2 0 0 7 ) 669–675
available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/dnarepair
Brief report
RecA-DNA filament topology: The overlooked alternative of an unconventional syn–syn duplex intermediate Richard Egel ∗ Institute of Molecular Biology and Physiology, University of Copenhagen, Ole Maaløe Vej 5, DK-2100 Copenhagen Ø, Denmark
a r t i c l e
i n f o
a b s t r a c t
Article history:
The helical filaments of RecA protein mediate strand exchange for homologous recombi-
Received 3 November 2006
nation, but the paths of the interacting DNAs have yet to be determined. Although this
Received in revised form
interaction is commonly limited to three strands, it is reasoned here that the intrinsic
17 December 2006
symmetry relationships of quadruplex topology are superior in explaining a range of obser-
Accepted 21 December 2006
vations. In particular, this topology suggests the potential of post-exchange base pairing in
Published on line 20 February 2007
the unorthodox configuration of syn–syn glycosidic bonds between the nucleotide bases and the pentose rings in the sugar–phosphate backbone, which would transiently be stabilized
Keywords:
by the external scaffolding of the RecA protein filament.
Recombinase
© 2007 Elsevier B.V. All rights reserved.
Homologous recombination DNA strand exchange Heteroduplex
1.
Introduction
Next to the semiconservative mode of replication, the capability of recombining with another molecule of corresponding sequence is a fundamental property of DNA. RecA-like proteins provide the recombinase function for homologous recombination to drive the initial strand exchange reaction in all organisms [1,2]. These proteins belong to the RecA–AAA+ superfamily of ATPases, which self-assemble in the active state with ATP and convert chemical energy into conformational change. RecA of Escherichia coli is typical for these recombinases, and most structural work has been devoted to this particular protein. RecA-type recombinases administer the proverbial search for a needle in the haystack: in trying to locate the perfect match for a given DNA sequence of typically 100 nucleotides
∗
Tel.: +45 4589 3746. E-mail address: [email protected]. 1568-7864/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.dnarep.2006.12.013
or more, they have to shift through potentially all the different sequence tracks of similar length throughout a complex genome. This means that literally thousands of unproductive encounters have to be assessed rapidly and reversibly, before a track of sufficient sequence homology allows the strand exchange reaction to be completed irreversibly. In vitro, RecA subunits with ATP assemble readily as helical filaments on single-stranded (ss) DNA. In this activated filament, a duplex DNA of any sequence can be incorporated as well. This interaction is transient for nonhomologous sequences, but in the case of a perfect match, strands are being exchanged according to the following formula: IN + COM::OUT → IN::COM + OUT Therein, the invading strand (IN) from the RecA filament displaces the outgoing strand (OUT) from the homologous
670
d n a r e p a i r 6 ( 2 0 0 7 ) 669–675
duplex, and the complementary strand (COM) to either one changes base pairing from one to the other. The substrate and product molecules are commonly referred to as homoduplex and heteroduplex, respectively. In contrast to ssDNA, a single duplex associates with RecA rather slowly, as limited by inefficient nucleation. However, if dsDNA is mechanically stretched beforehand, it rapidly forms a RecA filament as well [3]. Also, once nucleated by 4–5 ATP–RecA monomers, additional subunits assemble more readily on duplex DNA as well [4,5]. So far, the molecular coordinates have been worked out mainly for the RecA protein residues, but the exact paths of the participating strands of DNA in the central void of the filament are not yet known. Early models were subject to multiple degrees of freedom, which subsequently got constrained by new results. In fact, none of the models existing in the current literature appears to be fully compatible with all the relevant experimental data. Hence, some re-evaluation from first principles may be appropriate. Basically, the DNA in the helical RecA filament is right-handed as usual, but it is stretched by 50% and partly unwound by 40% with respect to B-form DNA [6,7]. Each RecA subunit covers three nucleotides, and about six subunits complete a turn of the helical filament. As each subunit only touches its nearest neighbors on either side, a furrow is left open between the turns of the ribbon, through which the central void can be entered from outside. Characteristically, one of the furrow rims is tiled quite smoothly by the protein subunits, while the other rim shows a pronged appearance, as formed by the protruding C-terminal domains of RecA. The stretching imposed upon the DNA effectively unstacks some of the base pairs, which is important for the activated intermediate to transiently disengage from base pairing in the original duplex. The RecA subunits are assumed to contact the sugar–phosphate backbone, rather than the bases [8], and they appear to bind a duplex from the minor groove [9,10]. Well before the era of RecA filaments, an imaginative and far-sighted structure was proposed for the possible strand exchange between two fully intertwined duplexes of DNA, as demonstrated by a space-filling model for the presumable preexchange complex [11]. A similar structure was also proposed for the post-exchange complex [12]. However, the remarkable quadruplex symmetry of these structures fell into disregard when it became clear that the RecA-driven exchange reaction primarily operates with three strands of DNA, not four [13]. Instead, subsequent model building concentrated on the assessment of various triple-helix arrangements, as stabilized by supplementary hydrogen bonds [14,15]. Notably, all current models of strand exchange are based on the lateral extrusion of bases from the COM strand. Such a movement within the plane of the original base pair cannot be performed without considerable buckling of the backbone, although part of this strain can be absorbed by changing the puckering of the pleated pentose rings [16]. In the triplex models the invading strand could enter the homologous duplex either from its major groove (preferred originally) or the minor one. This ambiguity seemed to be resolved by the finding that the outgoing strand left the duplex from the major groove [17]. Hence, it was indirectly concluded that the invading strand thus had to enter from the minor groove, which is not readily compat-
ible with the presumed binding of RecA protein to the minor groove as well. This is because a protein domain that binds to DNA across the minor groove is physically blocking that side of the binding site from interacting with another DNA, so that potential strand exchange could only be initiated from the major groove. Furthermore, according to recent analyses of strand exchange kinetics by fluorescense energy transfer, the relevance of hydrogen-bonded triple helixes as recombination intermediates has been questioned altogether, preferring the direct exchange of base pairing instead [18]. Moreover, a very informative experiment revived the possibility that four strands of DNA could simultaneously interact within a given RecA filament, provisionally referred to as recombination in trans [19]. When a RecA filament was preassembled on single-stranded DNA of a given sequence and supplemented with a nonhomologous duplex DNA, this duplex was activated for exchange with a homologous fourth strand, which was added last to the established filament. However, it very much depends on the topological arrangement whether the fourth strand actually approaches the duplex from the other side (the opposite groove “in trans”), with regard to the first single strand in the filament. According to earlier suggestions, the strongest DNA binding site, which is presumed to accommodate the invading single strand, has intuitively been placed deepest in the central furrow of the RecA filament [1,15]. Both strands of the homologous duplex could then straddle the furrow from rim to rim. In such an arrangement, the extra fourth strand would indeed approach the duplex from the opposite groove, as seen from the first single strand. Careful cross-linking studies, however, have located the single strand in the primary filament close to the smooth rim of the furrow [20]. Hence, the deepest site in the furrow is probably accommodating the complementary strand for the strand exchange reaction. The alternative model given below accounts for this condition, and both single strands mentioned above would approach the duplex from the same side.
2. Base swivelling to a RecA-supported syn–syn duplex—a reasonable alternative to the traditional model? As to the four-fold symmetry of intertwining duplexes [11], a certain aspect is particularly suitable to facilitate the reversible exchange of base pairing in a stretched and unstacked recombinational intermediate. This concerns the glycosidic bonds between the nucleotide bases and the pentose rings in the sugar–phosphate backbones. Notably, all these bonds are diagonally oriented in equivalent positions. Hence, if the interacting strands in a cross-section of the RecA filament were externally constrained to three corners of a square, the complementary strand at the central position could readily assess the possibility of base pairing to either side by swivelling its base residues by 180◦ around the glycosidic bonds (Fig. 1). In such a configuration, the extra trans-strand could take the empty seat at the fourth corner, which should be readily accessible between the furrow rims. Incidentally, the flipping of bases by 180◦ would alternate between the sterical “anti” and
d n a r e p a i r 6 ( 2 0 0 7 ) 669–675
671
Fig. 1 – RecA filament in cross-section, binding three of four possible strands. (A) The three strands normally participating in the RecA-mediated exchange reaction are assigned specific binding positions at each RecA subunit. The symmetry relationship among the intertwined strands is drawn according to the quadruplex model of McGavin [11]. The invading strand (IN) binds at the upper edge, consistent with experimental data (site 1 in Fig. 6 of ref. [20]). The complementary strand (COM) should be adjacent at the central position (site 2 in Fig. 6 of ref. [20]). The outgoing strand (OUT) is drawn close to the other edge, probably with the weakest specific binding. The fourth corner of the square can potentially be entered by an extra strand (EX), as indicated by recombination “in trans” [19]. A:T pairing is shown in this example, since such base pairs are assumed to open before G:C pairs in the initial phase of homology probing [18]. In this topology, with 5 → 3 polarity, the IN and OUT strands (diamonds) are pointing away from the viewer (to the right in the longitudinal view of Fig. 2). The COM and EX strands (circles) have opposite polarity, pointing towards the viewer. The COM::OUT duplex is stretched from B-form DNA, straddled by protein binding across the minor groove. Accordingly, all bases are attached by “anti” glycosidic bonds. The stretch-induced unstacking of bases is assumed to facilitate full rotation around the glycosidic bonds, half turns leading to the “syn” configuration. (B) If the coplanar bases of IN and COM strands are flipped simultaneously (turning arrows in A), alternative base pairing is facilitated to form the IN::COM heteroduplex. A similar flip-over in the OUT strand could lead to OUT::EX base pairing in the case of trans-recombination. Notably, the IN::COM heteroduplex is still attached to RecA by minor-groove geometry, but the polarity of strands has been inverted with respect to the ordinary B-form. This is equivalent to saying, that in a mixed contiguous duplex, the minor and major grooves are switching sides at the transition point between B-form and non-B configuration (see Fig. 3) [Note that the ∼90◦ angle between the anchoring glycosidic bonds of each base pair in a cross-section defines the minor groove of the duplex; the major groove corresponds to the complementary angle of ∼270◦ .].
“syn” configurations relative to the pentose rings.1 The attractive implications of such a flip-flop movement should not be dismissed too lightly. As yet, however, the consideration of syn-isomers in RecA filaments has deliberately been excluded from modelling attempts [16,17,21], since common B-form DNA is composed of anti bonds exclusively. A three-dimensional representation of quadruple strand positions in the RecA filament is shown in Fig. 2. The protein ribbon is drawn according to ref. [22]. Important details of this fitting are the established polarity of the IN strand [23] and its lateral positioning at the smooth rim of the furrow [20]. Notably, the protruding C-terminal domain of RecA at the opposite rim has independent binding affinity to dsDNA [24], which may be relevant for threading the incoming duplex into
1 In anti, the angles of the glycosidic bond towards the nearest protons (H1 in the sugar ring, and H6 or H8 in a pyrimidine or purine moiety, respectively) zigzag away from each other. In syn, however, they point to the same side, which could be stabilized by nonpolar contacts to RecA.
the filament and/or accommodating the OUT strand in this model. The extra path of a fourth strand is not directly relevant for homology recognition in the common three-strand interaction, but it should be accessible for the extra strand provided in the trans-recombination assay [19]. Also, it could be relevant for the final position of the OUT strand in the postexchange complex, where it has been observed to be shifted out of register by three bases [21,25]. Its radial distance, too, appears to be increased at that stage [21]. How realistic is it to assume that the stretch-induced unstacking is indeed sufficient to allow the anti–syn isomerization to occur? To provide the necessary space for turning the bases, the purines in particular, it may in fact be helpful that the unstacking is considered to be discontinuous, or gapped, separating residual stacks of two or three nucleotides per RecA subunit [15,26], thus accumulating up to three increments of stretching around a single base pair. To develop this notion a step further yet, this may be part of a concerted, reiterative action in assessing the degree of sequence similarity for the productive strand exchange reaction. If the initial wave of anti–syn isomerization only affects every third base and there
672
d n a r e p a i r 6 ( 2 0 0 7 ) 669–675
Fig. 2 – Accommodation of four strands in the RecA filament. The intertwining-duplex model is schematically inscribed into the RecA filament structure [22], consistent with the cross-section in Fig. 1. The EX strand position (centered around the furrow) is empty in the common three-strand interaction, but it can be occupied during “trans” recombination. After successful strand exchange for an extended stretch of homology, the OUT strand is released from its binding sites and shifted to a new position—possibly resembling the EX strand in this diagram. The arrows at the right are indicating antiparallel 5 → 3 polarity of the individual strands. The directional polarity of the IN strand (largest arrow) has been verified experimentally [23].
is no perfect match at several positions of this kind, this would indicate early on that this encounter has not hit a homologous target—leading to disengagement by default. However, if several third-positions in a row allow alternative base pairing in syn, this would be a prospective token for suitable homology and the entire filament may slightly shift towards a second wave of anti–syn isomerization—again at every third position, yet one base removed from the initial wave. The second wave, too, could be aborted by default if base pairing in syn does not arise at most of these positions. On the other hand, if all three waves of shifted anti–syn isomerization result in alternative base pairing in syn, this could be the final signal to proceed with strand exchange for a much longer stretch. Such a reiterative scheme of homology assessment is not incompatible with another interesting notion that the initial wave(s) of base flipping might affect A::T pairs preferentially, before the more stable G::C pairs are being flipped as well in the final acceleration of the strand exchange reaction [27]. The experimental observation of a lateral displacement out of phase for the OUT strand by three bases [21,25] could be a direct consequence of skipped isomerization in several rounds. Alternatively, the radial and lateral displacement of the OUT strand relative to the post-exchange duplex [21] could be due to its continued binding to the C-terminal rim of RecA [24], while the heteroduplex may be shifted towards the helical axis [16,21]. While symmetry relationships are of utmost importance for reversibility in aborting nonhomologous encounters, directional asymmetry must eventually prevail in driving successful exchange events of homologous strands to completion. Hence, it is most critical for the model proposed above that the COM strand be positioned at the corner of a right angle relative to the two identical strands it switches partner with. The IN strand, on which the RecA filament is first assembled, starts out as a single strand so that its base residues are less constrained by residual stacking forces than the base pairs in the target duplex. In fact, it would be advantageous for the efficiency of this model if the IN-strand bases were config-
ured in syn preferentially. The strong interactions with RecA at its binding site could lead to such a preferential bias. On similar grounds, a slight preference for bases being in syn at the secondary binding site for the COM strand could contribute to a cumulative bias towards strand exchange in the case of substantial homology over a long stretch of heteroduplex formation. It is of interest to note that two flexible loops, which remained unlocalized in the earliest RecA crystal structure [28], appear to be specifically involved in the molding of both the primary and secondary DNA binding sites [20,21,29]. It is not inconceivable that certain amino acids of these loops may contribute to the gapped unstacking of base pairs in the homoduplex and the postulated bias towards the syn configuration of one base at a time. When a RecA–ssDNA complex, comprising as few as 3–5 nucleotides, was subjected to NMR analysis,2 the bases were resolved at remarkably well defined positions [30]. This shows that even a short single strand of DNA is rigidly constrained by its binding to RecA. The ordered structure was ascribed to a novel type of repetitive stacking between the aromatic rings of bases and the 2 -methylene moiety of the adjacent 2 -deoxyribose ring. Although the calculated structures were shown in the all-anti configuration exclusively, it is not clear from that paper whether this is a necessary conclusion from the NMR spectra directly. In a subsequent paper [16], the potential possibility of a similar stacking in syn was merely dismissed as being unlikely because it does not match with ordinary B-form DNA. On principle, the 2 -methylene moiety of a deoxyribose could stack equally well with either side of the adjacent aromatic base. Upon the completion of homologous strand exchange and facilitated by ATP hydrolysis, the filament disintegrates by releasing free subunits of RecA [31]. This should deprive the postulated heteroduplex syn–syn structure of its scaffolding
2
NMR: nuclear magnetic resonance.
d n a r e p a i r 6 ( 2 0 0 7 ) 669–675
Fig. 3 – Reestablishment of B-form DNA after dissolution of the RecA filament. The groove-inverted non-B arrangement of the presumptive intermediate should be inherently unstable when the external scaffolding of the protein barrel disappears. This figure focuses on the stacking discontinuity at the boundary between non-B and B-form DNA. For clarity, the ladder-like diagram at the left is presented without helical twist; a more realistic drawing is added to the right. As the syn configuration is unfavorable for compressed base stacking, especially for purines, the transition zone from non-B to B-form is likely to rearrange spontaneously by the processive extension of the B-form stack. This restacking may well occur non-enzymatically, resembling a domino effect of successive tumbling towards the stable B-form.
support. For all we know, base pairing in syn is unfavorable for DNA hybridization in free solution. Hence, after RecA dispersion, a limited stretch of less stable, syn-paired DNA would lie adjacent to the ordinary B-form (Fig. 3). At this structural discontinuity, the gap in relational base stacking should tend to equilibrate spontaneously towards the more compact and stable B-form, probably facilitated by Brownian movement alone and without the aid of particular helicase action to this effect. The ancillary stacking between the 2 -methylene moiety of deoxyribose and the adjacent base [30] may actually be helpful in pushing the bases of the non-B heteroduplex to the other side. In the aqueous environment of the unstacked intermediate state, it would not even be necessary to preserve the Watson–Crick hydrogen bonds of the tumbling base pairs during the shift, but the bases could flip-flop individually to fall into place in the stable B-form.
3.
Concluding remarks
3.1.
Why propose another model at this time?
Despite innumerous efforts to discern the molecular mechanism of RecA-mediated strand exchange, many basic and critical parameters of how this fundamental process is coordinated are still unknown. In contrast with more simple chemical reactions, the successful strand exchange for homol-
673
ogous recombination requires the unambiguous recognition of a matching pattern that extends over enormous distances at the atomic scale. It is indeed remarkable that a single protein, RecA, by assembling in a repetitive filament, can accomplish such a complicated task. To fully appreciate the way it works, of course, it is necessary to know the topological arrangement of the DNA strands to be compared and recombined. Such knowledge is still rather sketchy, and the triple-helix interactions assumed by current models are not yet satisfactory in detail. For one thing, they imply a discomforting degree of crowding at the minor groove of the target duplex, where both the RecA contacts and the invading ssDNA have to compete for access to the duplex DNA; for another, the complementary strand, which is the most important subject to the exchange reaction, has to wiggle about in a distorted path of its backbone, which is supposed to interact with RecA preferentially. Moreover, these distortions have to zigzag back and forth repeatedly, if the switching of base pairs does not proceed contiguously, but initially only occurs at A::T pairs—leaving all the original G::C pairs in place for a considerable time. On the other hand, the repetitive nature of the external RecA scaffold likely results in a stereotype pattern of equivalent binding sites to the backbones of the interacting strands of DNA. The quasi-quadruplex symmetry proposed here as an alternative model (with an empty seat at the fourth corner outside the filament) takes full advantage of such stereotype binding patterns in three intertwining threads. Also, it allows all the bases to switch partners without the need for compensatory lateral movements of the sugar–phosphate backbones. Thereby, the repeated symmetry of equivalent alternative positions around the helical filament could perfectly serve as an “activated intermediate” for the reversible base pair exchange reaction in the tentative probing for potential sequence similarity along a sizable stretch of duplex DNA. For the time being, this model is not stringently demanded by any particular experimental evidence, but neither is any one of the various triple-helix models proposed so far. Yet, as none of the traditional models of all-anti triplex interactions has convincingly suggested how repetitive RecA binding across the minor groove of the homoduplex should give way to the incoming single strand from the same side, the quasiquadruplex model with initial access to the duplex from the major groove and transitory base pairing in the syn–syn configuration deserves to be taken seriously as a respectable competitor to other current models. Only structural evidence from further experimentation can eventually disprove one or the other.
3.2. Which particular aspects should be more testable than others? The successful application of NMR difference spectra3 to the ordered adsorption of relatively short oligonucleotides to a much larger protein moiety [30] has shown that significant positional cues can be inferred, as to the structural fit
3
More precisely, the method in question is termed “transferred nuclear Overhauser effect (TRNOE) spectroscopy”.
674
d n a r e p a i r 6 ( 2 0 0 7 ) 669–675
imposed upon the ligand by the protein scaffold. Conceivably, this method could also be used to estimate the preponderance of syn or anti orientation of the base residues in the RecA-adsorbed oligonucleotides, by measuring the distance between H1 and H6Py or H8Pu of particular nucleotide residues. These protons should only approach each other in the syn configuration. To start with, a re-analysis of the primary ssDNA, as studied before [30], could give a useful hint. More critically, however, the corresponding orientation of the complementary strand should be assessed directly in the annealed heteroduplex. To this effect, an oligo comprising 3–5 nucleotides could be subjected to NMR analysis, before and after the stepwise addition of a preformed RecA–ssDNA filament of minimal size, say 4–5 RecA subunits [4,5] assembled on oligos comprising some 12–15 bases, with an internal sequence site complementary to the secondary oligo to be tested from the solution. It should be rewarding to subject the hypothetical models to such a focused test. Another procedure to assess the potential involvement of syn-configured intermediates could be based on rapid kinetic fluorescence probing of the strand exchange reaction [18]. If base pairing in syn indeed is stabilized by nonpolar contacts of RecA to protons H1 in the sugar ring, and H6 or H8 in the pyrimidine or purine moiety,1 these contacts may be disrupted by modified base analogs, such as 6-X thymidine and/or 8Y adenosin—X and Y being appropriately selected functional side groups. While such groups should have little effect on the planar extrusion of bases in the traditional triple-helix models, they are expected to impede the flip-flop rotation and/or the specific RecA contacts in the alternative model proposed above, and the exchange kinetics should be substantially perturbed by such substitutions.
Acknowledgements I gratefully appreciated constructive discussion and critical ´ comments from Chantal Prevost, Edward H. Egelman, and Dirk Lankenau, which have been invaluable in streamlining a queer hunch into a set of coherent assumptions.
references
[1] M.M. Cox, The bacterial RecA protein as a motor protein, Annu. Rev. Microbiol. 57 (2003) 551–577. [2] A. Shinohara, M. Shinohara, Roles of RecA homologues Rad51 and Dmc1 during meiotic recombination, Cytogenet. Genome Res. 107 (2004) 201–207. [3] J.F. Leger, J. Robert, L. Bourdieu, D. Chatenay, J.F. Marko, RecA binding to a single double-stranded DNA molecule: a possible role of DNA conformational fluctuations, Proc. Natl. Acad. Sci. U.S.A. 95 (1998) 12295–12299. [4] R. Galletto, I. Amitani, R.J. Baskin, S.C. Kowalczykowski, Direct observation of individual RecA filaments assembling on single DNA molecules, Nature 443 (2006) 875– 878. [5] C. Joo, S.A. McKinney, M. Nakamura, I. Rasnik, S. Myong, T. Ha, Real-time observation of RecA filament dynamics with single monomer resolution, Cell 126 (2006) 515–527. [6] E.H. Egelman, X. Yu, The location of DNA in RecA-DNA helical filaments, Science 245 (1989) 404–407.
[7] E.H. Egelman, Does a stretched DNA structure dictate the helical geometry of RecA-like filaments? J. Mol. Biol. 309 (2001) 539–542. [8] M.C. Leahy, C.M. Radding, Topography of the interaction of recA protein with single-stranded deoxyoligonucleotides, J. Biol. Chem. 261 (1986) 6954–6960. [9] E. Di Capua, B. Muller, The accessibility of DNA to dimethylsulfate in complexes with RecA protein, EMBO J. 6 (1987) 2493–2498. [10] E. Tuite, U. Sehlstedt, P. Hagmar, B. Norden, M. Takahashi, Effects of minor and major groove-binding drugs and intercalators on the DNA association of minor groove-binding proteins RecA and deoxyribonuclease I detected by flow linear dichroism, Eur. J. Biochem. 243 (1997) 482–492. [11] S. McGavin, Models of specifically paired like (homologous) nucleic acid structures, J. Mol. Biol. 55 (1971) 293–298. [12] J.H. Wilson, Nick-free formation of reciprocal heteroduplexes: a simple solution to the topological problem, Proc. Natl. Acad. Sci. U.S.A. 76 (1979) 3641–3645. [13] M.M. Cox, Alignment of 3 (but not 4) DNA strands within a RecA protein filament, J. Biol. Chem. 270 (1995) 26021–26024. [14] V.B. Zhurkin, G. Raghunathan, N.B. Ulyanov, R.D. Camerini-Otero, R.L. Jernigan, A parallel DNA triplex as a model for the intermediate in homologous recombination, J. Mol. Biol. 239 (1994) 181–200. ´ [15] G. Bertucat, R. Lavery, C. Prevost, A molecular model for RecA-promoted strand exchange via parallel triple-stranded helices, Biophys. J. 77 (1999) 1562–1576. [16] T. Nishinaka, A. Shinohara, Y. Ito, S. Yokoyama, T. Shibata, Base pair switching by interconversion of sugar puckers in DNA extended by proteins of RecA-family: a model for homology search in homologous genetic recombination, Proc. Natl. Acad. Sci. U.S.A. 95 (1998) 11071–11076. [17] X. Zhou, K. Adzuma, DNA strand exchange mediated by the Escherichia coli RecA protein initiates in the minor groove of double-stranded DNA, Biochemistry 36 (1997) 4650–4661. [18] A.M. Lee, J. Xiao, S.F. Singleton, Origins of sequence selectivity in homologous genetic recombination: insights from rapid kinetic probing of RecA-mediated DNA strand exchange, J. Mol. Biol. 360 (2006) 343–359. [19] A.V. Mazin, S.C. Kowalczykowski, A novel property of the RecA nucleoprotein filament: activation of double-stranded DNA for strand exchange in trans, Genes Dev. 13 (1999) 2005–2016. [20] Y. Wang, K. Adzuma, Differential proximity probing of two DNA binding sites in the Escherichia coli RecA protein using photo-cross-linking methods, Biochemistry 35 (1996) 3563–3571. [21] V.A. Malkov, I.G. Panyutin, R.D. Neumann, V.B. Zhurkin, R.D. Camerini-Otero, Radioprobing of a RecA-three-stranded DNA complex with iodine 125: evidence for recognition of homology in the major groove of the target duplex, J. Mol. Biol. 299 (2000) 629–640. [22] N. Haruta, X. Yu, S. Yang, E.H. Egelman, M.M. Cox, A DNA pairing-enhanced conformation of bacterial RecA proteins, J. Biol. Chem. 278 (2003) 52710–52723. [23] A. Stasiak, E.H. Egelman, P. Howard-Flanders, Structure of helical RecA-DNA complexes. III: The structural polarity of RecA filaments and functional polarity in the RecA-mediated strand exchange reaction, J. Mol. Biol. 202 (1988) 659–662. [24] H. Aihara, Y. Ito, H. Kurumizaka, T. Terada, S. Yokoyama, T. Shibata, An interaction between a specified surface of the C-terminal domain of RecA protein and double-stranded DNA for homologous pairing, J. Mol. Biol. 274 (1997) 213– 221. [25] J. Xiao, S.F. Singleton, Elucidating a key intermediate in homologous DNA strand exchange: structural
d n a r e p a i r 6 ( 2 0 0 7 ) 669–675
characterization of the RecA-triple-stranded DNA complex using fluorescence resonance energy transfer, J. Mol. Biol. 320 (2002) 529–558. ´ [26] C. Prevost, M. Takahashi, Geometry of the DNA strands within the RecA nucleofilament: role in homologous recombination, Q. Rev. Biophys. 36 (2003) 429–453. [27] E. Folta-Stogniew, S. O’Malley, R. Gupta, K.S. Anderson, C.M. Radding, Exchange of DNA base pairs that coincides with recognition of homology promoted by E. coli RecA protein, Mol. Cell 15 (2004) 846–847. [28] R.M. Story, I.T. Weber, T.A. Steitz, The structure of the E. coli recA protein monomer and polymer, Nature 355 (1992) 318–325.
675
[29] C. Cazaux, J.S. Blanchet, D. Dupuis, G. Villani, M. Defais, N.P. Johnson, Investigation of the secondary DNA-binding site of the bacterial recombinase RecA, J. Biol. Chem. 273 (1998) 28799–28804. [30] T. Nishinaka, Y. Ito, S. Yokoyama, T. Shibata, An extended DNA structure through deoxyribose-base stacking induced by RecA protein, Proc. Natl. Acad. Sci. U.S.A. 94 (1997) 6623–6628. [31] J.M. Cox, O.V. Tsodikov, M.M. Cox, Organized unidirectional waves of ATP hydrolysis within a RecA filament, PLoS Biol. 3 (2005) e52.
available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/dnarepair
Brief report
RecA-DNA filament topology: The overlooked alternative of an unconventional syn–syn duplex intermediate Richard Egel ∗ Institute of Molecular Biology and Physiology, University of Copenhagen, Ole Maaløe Vej 5, DK-2100 Copenhagen Ø, Denmark
a r t i c l e
i n f o
a b s t r a c t
Article history:
The helical filaments of RecA protein mediate strand exchange for homologous recombi-
Received 3 November 2006
nation, but the paths of the interacting DNAs have yet to be determined. Although this
Received in revised form
interaction is commonly limited to three strands, it is reasoned here that the intrinsic
17 December 2006
symmetry relationships of quadruplex topology are superior in explaining a range of obser-
Accepted 21 December 2006
vations. In particular, this topology suggests the potential of post-exchange base pairing in
Published on line 20 February 2007
the unorthodox configuration of syn–syn glycosidic bonds between the nucleotide bases and the pentose rings in the sugar–phosphate backbone, which would transiently be stabilized
Keywords:
by the external scaffolding of the RecA protein filament.
Recombinase
© 2007 Elsevier B.V. All rights reserved.
Homologous recombination DNA strand exchange Heteroduplex
1.
Introduction
Next to the semiconservative mode of replication, the capability of recombining with another molecule of corresponding sequence is a fundamental property of DNA. RecA-like proteins provide the recombinase function for homologous recombination to drive the initial strand exchange reaction in all organisms [1,2]. These proteins belong to the RecA–AAA+ superfamily of ATPases, which self-assemble in the active state with ATP and convert chemical energy into conformational change. RecA of Escherichia coli is typical for these recombinases, and most structural work has been devoted to this particular protein. RecA-type recombinases administer the proverbial search for a needle in the haystack: in trying to locate the perfect match for a given DNA sequence of typically 100 nucleotides
∗
Tel.: +45 4589 3746. E-mail address: [email protected]. 1568-7864/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.dnarep.2006.12.013
or more, they have to shift through potentially all the different sequence tracks of similar length throughout a complex genome. This means that literally thousands of unproductive encounters have to be assessed rapidly and reversibly, before a track of sufficient sequence homology allows the strand exchange reaction to be completed irreversibly. In vitro, RecA subunits with ATP assemble readily as helical filaments on single-stranded (ss) DNA. In this activated filament, a duplex DNA of any sequence can be incorporated as well. This interaction is transient for nonhomologous sequences, but in the case of a perfect match, strands are being exchanged according to the following formula: IN + COM::OUT → IN::COM + OUT Therein, the invading strand (IN) from the RecA filament displaces the outgoing strand (OUT) from the homologous
670
d n a r e p a i r 6 ( 2 0 0 7 ) 669–675
duplex, and the complementary strand (COM) to either one changes base pairing from one to the other. The substrate and product molecules are commonly referred to as homoduplex and heteroduplex, respectively. In contrast to ssDNA, a single duplex associates with RecA rather slowly, as limited by inefficient nucleation. However, if dsDNA is mechanically stretched beforehand, it rapidly forms a RecA filament as well [3]. Also, once nucleated by 4–5 ATP–RecA monomers, additional subunits assemble more readily on duplex DNA as well [4,5]. So far, the molecular coordinates have been worked out mainly for the RecA protein residues, but the exact paths of the participating strands of DNA in the central void of the filament are not yet known. Early models were subject to multiple degrees of freedom, which subsequently got constrained by new results. In fact, none of the models existing in the current literature appears to be fully compatible with all the relevant experimental data. Hence, some re-evaluation from first principles may be appropriate. Basically, the DNA in the helical RecA filament is right-handed as usual, but it is stretched by 50% and partly unwound by 40% with respect to B-form DNA [6,7]. Each RecA subunit covers three nucleotides, and about six subunits complete a turn of the helical filament. As each subunit only touches its nearest neighbors on either side, a furrow is left open between the turns of the ribbon, through which the central void can be entered from outside. Characteristically, one of the furrow rims is tiled quite smoothly by the protein subunits, while the other rim shows a pronged appearance, as formed by the protruding C-terminal domains of RecA. The stretching imposed upon the DNA effectively unstacks some of the base pairs, which is important for the activated intermediate to transiently disengage from base pairing in the original duplex. The RecA subunits are assumed to contact the sugar–phosphate backbone, rather than the bases [8], and they appear to bind a duplex from the minor groove [9,10]. Well before the era of RecA filaments, an imaginative and far-sighted structure was proposed for the possible strand exchange between two fully intertwined duplexes of DNA, as demonstrated by a space-filling model for the presumable preexchange complex [11]. A similar structure was also proposed for the post-exchange complex [12]. However, the remarkable quadruplex symmetry of these structures fell into disregard when it became clear that the RecA-driven exchange reaction primarily operates with three strands of DNA, not four [13]. Instead, subsequent model building concentrated on the assessment of various triple-helix arrangements, as stabilized by supplementary hydrogen bonds [14,15]. Notably, all current models of strand exchange are based on the lateral extrusion of bases from the COM strand. Such a movement within the plane of the original base pair cannot be performed without considerable buckling of the backbone, although part of this strain can be absorbed by changing the puckering of the pleated pentose rings [16]. In the triplex models the invading strand could enter the homologous duplex either from its major groove (preferred originally) or the minor one. This ambiguity seemed to be resolved by the finding that the outgoing strand left the duplex from the major groove [17]. Hence, it was indirectly concluded that the invading strand thus had to enter from the minor groove, which is not readily compat-
ible with the presumed binding of RecA protein to the minor groove as well. This is because a protein domain that binds to DNA across the minor groove is physically blocking that side of the binding site from interacting with another DNA, so that potential strand exchange could only be initiated from the major groove. Furthermore, according to recent analyses of strand exchange kinetics by fluorescense energy transfer, the relevance of hydrogen-bonded triple helixes as recombination intermediates has been questioned altogether, preferring the direct exchange of base pairing instead [18]. Moreover, a very informative experiment revived the possibility that four strands of DNA could simultaneously interact within a given RecA filament, provisionally referred to as recombination in trans [19]. When a RecA filament was preassembled on single-stranded DNA of a given sequence and supplemented with a nonhomologous duplex DNA, this duplex was activated for exchange with a homologous fourth strand, which was added last to the established filament. However, it very much depends on the topological arrangement whether the fourth strand actually approaches the duplex from the other side (the opposite groove “in trans”), with regard to the first single strand in the filament. According to earlier suggestions, the strongest DNA binding site, which is presumed to accommodate the invading single strand, has intuitively been placed deepest in the central furrow of the RecA filament [1,15]. Both strands of the homologous duplex could then straddle the furrow from rim to rim. In such an arrangement, the extra fourth strand would indeed approach the duplex from the opposite groove, as seen from the first single strand. Careful cross-linking studies, however, have located the single strand in the primary filament close to the smooth rim of the furrow [20]. Hence, the deepest site in the furrow is probably accommodating the complementary strand for the strand exchange reaction. The alternative model given below accounts for this condition, and both single strands mentioned above would approach the duplex from the same side.
2. Base swivelling to a RecA-supported syn–syn duplex—a reasonable alternative to the traditional model? As to the four-fold symmetry of intertwining duplexes [11], a certain aspect is particularly suitable to facilitate the reversible exchange of base pairing in a stretched and unstacked recombinational intermediate. This concerns the glycosidic bonds between the nucleotide bases and the pentose rings in the sugar–phosphate backbones. Notably, all these bonds are diagonally oriented in equivalent positions. Hence, if the interacting strands in a cross-section of the RecA filament were externally constrained to three corners of a square, the complementary strand at the central position could readily assess the possibility of base pairing to either side by swivelling its base residues by 180◦ around the glycosidic bonds (Fig. 1). In such a configuration, the extra trans-strand could take the empty seat at the fourth corner, which should be readily accessible between the furrow rims. Incidentally, the flipping of bases by 180◦ would alternate between the sterical “anti” and
d n a r e p a i r 6 ( 2 0 0 7 ) 669–675
671
Fig. 1 – RecA filament in cross-section, binding three of four possible strands. (A) The three strands normally participating in the RecA-mediated exchange reaction are assigned specific binding positions at each RecA subunit. The symmetry relationship among the intertwined strands is drawn according to the quadruplex model of McGavin [11]. The invading strand (IN) binds at the upper edge, consistent with experimental data (site 1 in Fig. 6 of ref. [20]). The complementary strand (COM) should be adjacent at the central position (site 2 in Fig. 6 of ref. [20]). The outgoing strand (OUT) is drawn close to the other edge, probably with the weakest specific binding. The fourth corner of the square can potentially be entered by an extra strand (EX), as indicated by recombination “in trans” [19]. A:T pairing is shown in this example, since such base pairs are assumed to open before G:C pairs in the initial phase of homology probing [18]. In this topology, with 5 → 3 polarity, the IN and OUT strands (diamonds) are pointing away from the viewer (to the right in the longitudinal view of Fig. 2). The COM and EX strands (circles) have opposite polarity, pointing towards the viewer. The COM::OUT duplex is stretched from B-form DNA, straddled by protein binding across the minor groove. Accordingly, all bases are attached by “anti” glycosidic bonds. The stretch-induced unstacking of bases is assumed to facilitate full rotation around the glycosidic bonds, half turns leading to the “syn” configuration. (B) If the coplanar bases of IN and COM strands are flipped simultaneously (turning arrows in A), alternative base pairing is facilitated to form the IN::COM heteroduplex. A similar flip-over in the OUT strand could lead to OUT::EX base pairing in the case of trans-recombination. Notably, the IN::COM heteroduplex is still attached to RecA by minor-groove geometry, but the polarity of strands has been inverted with respect to the ordinary B-form. This is equivalent to saying, that in a mixed contiguous duplex, the minor and major grooves are switching sides at the transition point between B-form and non-B configuration (see Fig. 3) [Note that the ∼90◦ angle between the anchoring glycosidic bonds of each base pair in a cross-section defines the minor groove of the duplex; the major groove corresponds to the complementary angle of ∼270◦ .].
“syn” configurations relative to the pentose rings.1 The attractive implications of such a flip-flop movement should not be dismissed too lightly. As yet, however, the consideration of syn-isomers in RecA filaments has deliberately been excluded from modelling attempts [16,17,21], since common B-form DNA is composed of anti bonds exclusively. A three-dimensional representation of quadruple strand positions in the RecA filament is shown in Fig. 2. The protein ribbon is drawn according to ref. [22]. Important details of this fitting are the established polarity of the IN strand [23] and its lateral positioning at the smooth rim of the furrow [20]. Notably, the protruding C-terminal domain of RecA at the opposite rim has independent binding affinity to dsDNA [24], which may be relevant for threading the incoming duplex into
1 In anti, the angles of the glycosidic bond towards the nearest protons (H1 in the sugar ring, and H6 or H8 in a pyrimidine or purine moiety, respectively) zigzag away from each other. In syn, however, they point to the same side, which could be stabilized by nonpolar contacts to RecA.
the filament and/or accommodating the OUT strand in this model. The extra path of a fourth strand is not directly relevant for homology recognition in the common three-strand interaction, but it should be accessible for the extra strand provided in the trans-recombination assay [19]. Also, it could be relevant for the final position of the OUT strand in the postexchange complex, where it has been observed to be shifted out of register by three bases [21,25]. Its radial distance, too, appears to be increased at that stage [21]. How realistic is it to assume that the stretch-induced unstacking is indeed sufficient to allow the anti–syn isomerization to occur? To provide the necessary space for turning the bases, the purines in particular, it may in fact be helpful that the unstacking is considered to be discontinuous, or gapped, separating residual stacks of two or three nucleotides per RecA subunit [15,26], thus accumulating up to three increments of stretching around a single base pair. To develop this notion a step further yet, this may be part of a concerted, reiterative action in assessing the degree of sequence similarity for the productive strand exchange reaction. If the initial wave of anti–syn isomerization only affects every third base and there
672
d n a r e p a i r 6 ( 2 0 0 7 ) 669–675
Fig. 2 – Accommodation of four strands in the RecA filament. The intertwining-duplex model is schematically inscribed into the RecA filament structure [22], consistent with the cross-section in Fig. 1. The EX strand position (centered around the furrow) is empty in the common three-strand interaction, but it can be occupied during “trans” recombination. After successful strand exchange for an extended stretch of homology, the OUT strand is released from its binding sites and shifted to a new position—possibly resembling the EX strand in this diagram. The arrows at the right are indicating antiparallel 5 → 3 polarity of the individual strands. The directional polarity of the IN strand (largest arrow) has been verified experimentally [23].
is no perfect match at several positions of this kind, this would indicate early on that this encounter has not hit a homologous target—leading to disengagement by default. However, if several third-positions in a row allow alternative base pairing in syn, this would be a prospective token for suitable homology and the entire filament may slightly shift towards a second wave of anti–syn isomerization—again at every third position, yet one base removed from the initial wave. The second wave, too, could be aborted by default if base pairing in syn does not arise at most of these positions. On the other hand, if all three waves of shifted anti–syn isomerization result in alternative base pairing in syn, this could be the final signal to proceed with strand exchange for a much longer stretch. Such a reiterative scheme of homology assessment is not incompatible with another interesting notion that the initial wave(s) of base flipping might affect A::T pairs preferentially, before the more stable G::C pairs are being flipped as well in the final acceleration of the strand exchange reaction [27]. The experimental observation of a lateral displacement out of phase for the OUT strand by three bases [21,25] could be a direct consequence of skipped isomerization in several rounds. Alternatively, the radial and lateral displacement of the OUT strand relative to the post-exchange duplex [21] could be due to its continued binding to the C-terminal rim of RecA [24], while the heteroduplex may be shifted towards the helical axis [16,21]. While symmetry relationships are of utmost importance for reversibility in aborting nonhomologous encounters, directional asymmetry must eventually prevail in driving successful exchange events of homologous strands to completion. Hence, it is most critical for the model proposed above that the COM strand be positioned at the corner of a right angle relative to the two identical strands it switches partner with. The IN strand, on which the RecA filament is first assembled, starts out as a single strand so that its base residues are less constrained by residual stacking forces than the base pairs in the target duplex. In fact, it would be advantageous for the efficiency of this model if the IN-strand bases were config-
ured in syn preferentially. The strong interactions with RecA at its binding site could lead to such a preferential bias. On similar grounds, a slight preference for bases being in syn at the secondary binding site for the COM strand could contribute to a cumulative bias towards strand exchange in the case of substantial homology over a long stretch of heteroduplex formation. It is of interest to note that two flexible loops, which remained unlocalized in the earliest RecA crystal structure [28], appear to be specifically involved in the molding of both the primary and secondary DNA binding sites [20,21,29]. It is not inconceivable that certain amino acids of these loops may contribute to the gapped unstacking of base pairs in the homoduplex and the postulated bias towards the syn configuration of one base at a time. When a RecA–ssDNA complex, comprising as few as 3–5 nucleotides, was subjected to NMR analysis,2 the bases were resolved at remarkably well defined positions [30]. This shows that even a short single strand of DNA is rigidly constrained by its binding to RecA. The ordered structure was ascribed to a novel type of repetitive stacking between the aromatic rings of bases and the 2 -methylene moiety of the adjacent 2 -deoxyribose ring. Although the calculated structures were shown in the all-anti configuration exclusively, it is not clear from that paper whether this is a necessary conclusion from the NMR spectra directly. In a subsequent paper [16], the potential possibility of a similar stacking in syn was merely dismissed as being unlikely because it does not match with ordinary B-form DNA. On principle, the 2 -methylene moiety of a deoxyribose could stack equally well with either side of the adjacent aromatic base. Upon the completion of homologous strand exchange and facilitated by ATP hydrolysis, the filament disintegrates by releasing free subunits of RecA [31]. This should deprive the postulated heteroduplex syn–syn structure of its scaffolding
2
NMR: nuclear magnetic resonance.
d n a r e p a i r 6 ( 2 0 0 7 ) 669–675
Fig. 3 – Reestablishment of B-form DNA after dissolution of the RecA filament. The groove-inverted non-B arrangement of the presumptive intermediate should be inherently unstable when the external scaffolding of the protein barrel disappears. This figure focuses on the stacking discontinuity at the boundary between non-B and B-form DNA. For clarity, the ladder-like diagram at the left is presented without helical twist; a more realistic drawing is added to the right. As the syn configuration is unfavorable for compressed base stacking, especially for purines, the transition zone from non-B to B-form is likely to rearrange spontaneously by the processive extension of the B-form stack. This restacking may well occur non-enzymatically, resembling a domino effect of successive tumbling towards the stable B-form.
support. For all we know, base pairing in syn is unfavorable for DNA hybridization in free solution. Hence, after RecA dispersion, a limited stretch of less stable, syn-paired DNA would lie adjacent to the ordinary B-form (Fig. 3). At this structural discontinuity, the gap in relational base stacking should tend to equilibrate spontaneously towards the more compact and stable B-form, probably facilitated by Brownian movement alone and without the aid of particular helicase action to this effect. The ancillary stacking between the 2 -methylene moiety of deoxyribose and the adjacent base [30] may actually be helpful in pushing the bases of the non-B heteroduplex to the other side. In the aqueous environment of the unstacked intermediate state, it would not even be necessary to preserve the Watson–Crick hydrogen bonds of the tumbling base pairs during the shift, but the bases could flip-flop individually to fall into place in the stable B-form.
3.
Concluding remarks
3.1.
Why propose another model at this time?
Despite innumerous efforts to discern the molecular mechanism of RecA-mediated strand exchange, many basic and critical parameters of how this fundamental process is coordinated are still unknown. In contrast with more simple chemical reactions, the successful strand exchange for homol-
673
ogous recombination requires the unambiguous recognition of a matching pattern that extends over enormous distances at the atomic scale. It is indeed remarkable that a single protein, RecA, by assembling in a repetitive filament, can accomplish such a complicated task. To fully appreciate the way it works, of course, it is necessary to know the topological arrangement of the DNA strands to be compared and recombined. Such knowledge is still rather sketchy, and the triple-helix interactions assumed by current models are not yet satisfactory in detail. For one thing, they imply a discomforting degree of crowding at the minor groove of the target duplex, where both the RecA contacts and the invading ssDNA have to compete for access to the duplex DNA; for another, the complementary strand, which is the most important subject to the exchange reaction, has to wiggle about in a distorted path of its backbone, which is supposed to interact with RecA preferentially. Moreover, these distortions have to zigzag back and forth repeatedly, if the switching of base pairs does not proceed contiguously, but initially only occurs at A::T pairs—leaving all the original G::C pairs in place for a considerable time. On the other hand, the repetitive nature of the external RecA scaffold likely results in a stereotype pattern of equivalent binding sites to the backbones of the interacting strands of DNA. The quasi-quadruplex symmetry proposed here as an alternative model (with an empty seat at the fourth corner outside the filament) takes full advantage of such stereotype binding patterns in three intertwining threads. Also, it allows all the bases to switch partners without the need for compensatory lateral movements of the sugar–phosphate backbones. Thereby, the repeated symmetry of equivalent alternative positions around the helical filament could perfectly serve as an “activated intermediate” for the reversible base pair exchange reaction in the tentative probing for potential sequence similarity along a sizable stretch of duplex DNA. For the time being, this model is not stringently demanded by any particular experimental evidence, but neither is any one of the various triple-helix models proposed so far. Yet, as none of the traditional models of all-anti triplex interactions has convincingly suggested how repetitive RecA binding across the minor groove of the homoduplex should give way to the incoming single strand from the same side, the quasiquadruplex model with initial access to the duplex from the major groove and transitory base pairing in the syn–syn configuration deserves to be taken seriously as a respectable competitor to other current models. Only structural evidence from further experimentation can eventually disprove one or the other.
3.2. Which particular aspects should be more testable than others? The successful application of NMR difference spectra3 to the ordered adsorption of relatively short oligonucleotides to a much larger protein moiety [30] has shown that significant positional cues can be inferred, as to the structural fit
3
More precisely, the method in question is termed “transferred nuclear Overhauser effect (TRNOE) spectroscopy”.
674
d n a r e p a i r 6 ( 2 0 0 7 ) 669–675
imposed upon the ligand by the protein scaffold. Conceivably, this method could also be used to estimate the preponderance of syn or anti orientation of the base residues in the RecA-adsorbed oligonucleotides, by measuring the distance between H1 and H6Py or H8Pu of particular nucleotide residues. These protons should only approach each other in the syn configuration. To start with, a re-analysis of the primary ssDNA, as studied before [30], could give a useful hint. More critically, however, the corresponding orientation of the complementary strand should be assessed directly in the annealed heteroduplex. To this effect, an oligo comprising 3–5 nucleotides could be subjected to NMR analysis, before and after the stepwise addition of a preformed RecA–ssDNA filament of minimal size, say 4–5 RecA subunits [4,5] assembled on oligos comprising some 12–15 bases, with an internal sequence site complementary to the secondary oligo to be tested from the solution. It should be rewarding to subject the hypothetical models to such a focused test. Another procedure to assess the potential involvement of syn-configured intermediates could be based on rapid kinetic fluorescence probing of the strand exchange reaction [18]. If base pairing in syn indeed is stabilized by nonpolar contacts of RecA to protons H1 in the sugar ring, and H6 or H8 in the pyrimidine or purine moiety,1 these contacts may be disrupted by modified base analogs, such as 6-X thymidine and/or 8Y adenosin—X and Y being appropriately selected functional side groups. While such groups should have little effect on the planar extrusion of bases in the traditional triple-helix models, they are expected to impede the flip-flop rotation and/or the specific RecA contacts in the alternative model proposed above, and the exchange kinetics should be substantially perturbed by such substitutions.
Acknowledgements I gratefully appreciated constructive discussion and critical ´ comments from Chantal Prevost, Edward H. Egelman, and Dirk Lankenau, which have been invaluable in streamlining a queer hunch into a set of coherent assumptions.
references
[1] M.M. Cox, The bacterial RecA protein as a motor protein, Annu. Rev. Microbiol. 57 (2003) 551–577. [2] A. Shinohara, M. Shinohara, Roles of RecA homologues Rad51 and Dmc1 during meiotic recombination, Cytogenet. Genome Res. 107 (2004) 201–207. [3] J.F. Leger, J. Robert, L. Bourdieu, D. Chatenay, J.F. Marko, RecA binding to a single double-stranded DNA molecule: a possible role of DNA conformational fluctuations, Proc. Natl. Acad. Sci. U.S.A. 95 (1998) 12295–12299. [4] R. Galletto, I. Amitani, R.J. Baskin, S.C. Kowalczykowski, Direct observation of individual RecA filaments assembling on single DNA molecules, Nature 443 (2006) 875– 878. [5] C. Joo, S.A. McKinney, M. Nakamura, I. Rasnik, S. Myong, T. Ha, Real-time observation of RecA filament dynamics with single monomer resolution, Cell 126 (2006) 515–527. [6] E.H. Egelman, X. Yu, The location of DNA in RecA-DNA helical filaments, Science 245 (1989) 404–407.
[7] E.H. Egelman, Does a stretched DNA structure dictate the helical geometry of RecA-like filaments? J. Mol. Biol. 309 (2001) 539–542. [8] M.C. Leahy, C.M. Radding, Topography of the interaction of recA protein with single-stranded deoxyoligonucleotides, J. Biol. Chem. 261 (1986) 6954–6960. [9] E. Di Capua, B. Muller, The accessibility of DNA to dimethylsulfate in complexes with RecA protein, EMBO J. 6 (1987) 2493–2498. [10] E. Tuite, U. Sehlstedt, P. Hagmar, B. Norden, M. Takahashi, Effects of minor and major groove-binding drugs and intercalators on the DNA association of minor groove-binding proteins RecA and deoxyribonuclease I detected by flow linear dichroism, Eur. J. Biochem. 243 (1997) 482–492. [11] S. McGavin, Models of specifically paired like (homologous) nucleic acid structures, J. Mol. Biol. 55 (1971) 293–298. [12] J.H. Wilson, Nick-free formation of reciprocal heteroduplexes: a simple solution to the topological problem, Proc. Natl. Acad. Sci. U.S.A. 76 (1979) 3641–3645. [13] M.M. Cox, Alignment of 3 (but not 4) DNA strands within a RecA protein filament, J. Biol. Chem. 270 (1995) 26021–26024. [14] V.B. Zhurkin, G. Raghunathan, N.B. Ulyanov, R.D. Camerini-Otero, R.L. Jernigan, A parallel DNA triplex as a model for the intermediate in homologous recombination, J. Mol. Biol. 239 (1994) 181–200. ´ [15] G. Bertucat, R. Lavery, C. Prevost, A molecular model for RecA-promoted strand exchange via parallel triple-stranded helices, Biophys. J. 77 (1999) 1562–1576. [16] T. Nishinaka, A. Shinohara, Y. Ito, S. Yokoyama, T. Shibata, Base pair switching by interconversion of sugar puckers in DNA extended by proteins of RecA-family: a model for homology search in homologous genetic recombination, Proc. Natl. Acad. Sci. U.S.A. 95 (1998) 11071–11076. [17] X. Zhou, K. Adzuma, DNA strand exchange mediated by the Escherichia coli RecA protein initiates in the minor groove of double-stranded DNA, Biochemistry 36 (1997) 4650–4661. [18] A.M. Lee, J. Xiao, S.F. Singleton, Origins of sequence selectivity in homologous genetic recombination: insights from rapid kinetic probing of RecA-mediated DNA strand exchange, J. Mol. Biol. 360 (2006) 343–359. [19] A.V. Mazin, S.C. Kowalczykowski, A novel property of the RecA nucleoprotein filament: activation of double-stranded DNA for strand exchange in trans, Genes Dev. 13 (1999) 2005–2016. [20] Y. Wang, K. Adzuma, Differential proximity probing of two DNA binding sites in the Escherichia coli RecA protein using photo-cross-linking methods, Biochemistry 35 (1996) 3563–3571. [21] V.A. Malkov, I.G. Panyutin, R.D. Neumann, V.B. Zhurkin, R.D. Camerini-Otero, Radioprobing of a RecA-three-stranded DNA complex with iodine 125: evidence for recognition of homology in the major groove of the target duplex, J. Mol. Biol. 299 (2000) 629–640. [22] N. Haruta, X. Yu, S. Yang, E.H. Egelman, M.M. Cox, A DNA pairing-enhanced conformation of bacterial RecA proteins, J. Biol. Chem. 278 (2003) 52710–52723. [23] A. Stasiak, E.H. Egelman, P. Howard-Flanders, Structure of helical RecA-DNA complexes. III: The structural polarity of RecA filaments and functional polarity in the RecA-mediated strand exchange reaction, J. Mol. Biol. 202 (1988) 659–662. [24] H. Aihara, Y. Ito, H. Kurumizaka, T. Terada, S. Yokoyama, T. Shibata, An interaction between a specified surface of the C-terminal domain of RecA protein and double-stranded DNA for homologous pairing, J. Mol. Biol. 274 (1997) 213– 221. [25] J. Xiao, S.F. Singleton, Elucidating a key intermediate in homologous DNA strand exchange: structural
d n a r e p a i r 6 ( 2 0 0 7 ) 669–675
characterization of the RecA-triple-stranded DNA complex using fluorescence resonance energy transfer, J. Mol. Biol. 320 (2002) 529–558. ´ [26] C. Prevost, M. Takahashi, Geometry of the DNA strands within the RecA nucleofilament: role in homologous recombination, Q. Rev. Biophys. 36 (2003) 429–453. [27] E. Folta-Stogniew, S. O’Malley, R. Gupta, K.S. Anderson, C.M. Radding, Exchange of DNA base pairs that coincides with recognition of homology promoted by E. coli RecA protein, Mol. Cell 15 (2004) 846–847. [28] R.M. Story, I.T. Weber, T.A. Steitz, The structure of the E. coli recA protein monomer and polymer, Nature 355 (1992) 318–325.
675
[29] C. Cazaux, J.S. Blanchet, D. Dupuis, G. Villani, M. Defais, N.P. Johnson, Investigation of the secondary DNA-binding site of the bacterial recombinase RecA, J. Biol. Chem. 273 (1998) 28799–28804. [30] T. Nishinaka, Y. Ito, S. Yokoyama, T. Shibata, An extended DNA structure through deoxyribose-base stacking induced by RecA protein, Proc. Natl. Acad. Sci. U.S.A. 94 (1997) 6623–6628. [31] J.M. Cox, O.V. Tsodikov, M.M. Cox, Organized unidirectional waves of ATP hydrolysis within a RecA filament, PLoS Biol. 3 (2005) e52.