Fig 1.
Current models for RecBCD action and Chi cleavage in the initiation of recombination and schematic depiction of the site of DSBR used in this work.
A. Current alternative hypotheses for the initiation of recombination by RecBCD [2,3]. RecBCD (pink figure) loads at the site of a DSB and translocates along the duplex DNA (i). During translocation, RecBCD either degrades both strands up to the recognition of a correctly oriented Chi site (with a preference for cleaving the 3’ terminal strand) or unwinds the duplex DNA without degrading it. Once a correctly oriented Chi site is recognised, the complex undergoes a conformational change and either up-regulates 5’ to 3’ cleavage while inhibiting 3’ to 5’ cleavage (ii) or nicks the 3’ terminal DNA strand and continues unwinding (iv). Both of these scenarios “Chi modulated DNA degradation” and “nick at Chi” lead to the formation ssDNA with a 3’ terminus, which is a substrate for the loading and polymerisation of RecA. These alternative hypotheses for the initiation of recombination lead to the formation of different structures of joint molecules (iii and v) and therefore to different biochemical steps following strand invasion and D-loop formation by RecA coated DNA. B. Map of the E. coli chromosomal depicting the two replichores and the site of DSBR used in this work. The chromosome of E. coli is drawn as a black line and the directions of replication of the left and right replichores are indicated by green and red arrows respectively. The regions of DSB induction in lacZ and of insertion of an ectopic terB site in ykgM-terB are shown in more detail. The palindrome and Chi arrays are shown by a black triangle and three coloured circles, respectively.
Fig 2.
MFA profiles of ΔrecG mutants and Rec+ strains of E. coli as a consequence of attempted DSBR at the lacZ locus.
The ratio of the normalized DNA copy number (or “relative enrichment”) of uniquely mapped sequence reads from exponentially growing cultures of the strains of interest are plotted along the y-axis against replichore-formatted genomic coordinates along the x-axis. The average relative enrichment of DNA in a ΔrecG mutant to a Rec+ strain is shown in the absence (A) or in the presence of an induced break at the palindrome (C), between Rec+ strains and ΔrecG mutants in the presence and in the absence of an induced break at the palindrome (C-D). The relative positions of the replication termination sites (terB, terC and terA), dif site and the location of palindrome are shown for each plot. The data are the averages of the two biological replicates shown individually in supporting Information S1 Fig and S2 Fig. Strains used were DL4184 (Rec+ Pal+), DL4201 (Rec+ Pal-), DL4311 (ΔrecG Pal+), and DL4312 (ΔrecG Pal-).
Fig 3.
Relative RecA ChIP-seq reads in the lacZ DSBR region (A-D) and terminus region (E-H) of the chromosome.
The raw data are shown in grey and smoothed data are shown in red. The smoothed data were plotted using a moving average filter with a 4 kb window. Red and green circles indicate Chi sites. Red Chi sites interact with RecBCD enzymes moving from right to left and green Chi sites interact with RecBCD enzymes moving from left to right. The closest Chi sites on either side of the DSB in lacZ were triple Chi arrays at 1.5kb from the palindrome, which have been used previously [9]. The positions of the rrnH operon, the palindrome at which a DSB is induced, termination sites (terA, terB and terC) and the site of resolution of chromosome dimers (dif) are all indicated. The direction of replication is indicated by green and red arrows for the left and the right replichore, respectively. A. and E. Rec+ Pal+; B. and F. Rec+ Pal-; C. and G. ΔrecG Pal+; D. and H. ΔrecG Pal-. Strains used were DL4184 (Rec+ Pal+), DL4201 (Rec+ Pal-), DL4311 (ΔrecG Pal+), and DL4312 (ΔrecG Pal-).
Fig 4.
The priA300 mutation suppresses the loss of DNA around a DSB in the absence of RecG and RuvAB.
A. NdeI digestion map of the region surrounding the palindrome locus. NdeI cutting sites and the distance between them are marked with black vertical arrows and numbers (in kb), respectively. The palindrome is indicated by a black triangle, Chi arrays by three coloured circles, the lacZ probe by a blue line and the lacZ.distal probe by a red line. B. Southern blot of a 1% agarose gel probed with a lacZ fragment (top) and a cysN control fragment (bottom). Strains used were DL4184 (Rec+ Pal+), DL4260 (ΔruvAB ΔrecG Pal+), DL5610 (ΔruvAB ΔrecG priA300 Pal+), DL4201 (Rec+ Pal-), DL4313 (ΔruvAB ΔrecG Pal-) and DL5611 (ΔruvAB ΔrecG priA300 Pal-). C. Quantification of the total amount of DNA at and around the break site. These values were first normalised to the values for the cysN control fragment. Then these ratios for the Pal+ strains were normalised to their Pal- controls. Finally, these ratios were normalised to the Rec+ ratio that was set to the value of 1. Error bars represent the standard error of the mean where n = 3. D. Schematic representation of the migration patterns of different species of branched DNA when separated on a two-dimensional native-native agarose gel. E. Two-dimensional native-native agarose gel electrophoresis. The DNA was detected using the lacZ.distal probe. Strains used were DL4243 (ΔruvAB Pal+), DL4257 (ΔruvAB Pal-), DL4260 (ΔrecG ΔruvAB Pal+), DL4313 (ΔrecG ΔruvAB Pal-), DL5610 (ΔrecG ΔruvAB priA300 Pal+) and DL5611 (ΔrecG ΔruvAB priA300 Pal-). F. Quantification of the DNA in the Y-arc and the X-spike normalised against the total branched DNA. Error bars represent the standard error of the mean where n = 3.
Fig 5.
Model depicting the proposed action of RecG and PriA in the RecBCD recombination pathway.
A. Revised model for the RecBCD recombination pathway. (i) The RecBCD enzyme recognises and binds to a DNA double-strand end. (ii) RecBCD generates a substrate with a 3’ end adjacent to a Chi site (shown as an arrow pointing in the direction of recombination stimulation) and a 5’ overhang. Continued unwinding by the RecBCD enzyme coupled to RecA loading onto the strand ending 3’ close to Chi allows the invasion of a target duplex and the formation of a D-loop. (iii) RuvAB binds to the Holliday junction end of the D-loop and migrates the Holliday junction away from the DSB end enlarging the D-loop. When a preferred recognition site for RuvC is encountered, the Holliday junction is resolved by cleavage and ligation. (iv) The RecG protein binds to the replication fork with a 5’ extended strand generated from the other end of the D-loop and unwinds the 5’ end while reannealing the parental DNA strands. (v) The action of RecG hands off the junction to PriA that binds in the correct manner to initiate the loading of DnaB. (vi) DnaB is loaded onto the lagging-strand template. (vii) DNA replication proceeds in the correct direction to restore the DNA lost in the region of the DSB. B. Hand-off between RecG and PriA ensures the correct loading of DnaB. The region between the dotted vertical lines is enlarged to show the binding of PriA. (i) RecG binds to a replication fork with an extended 5’ strand and unwinds this end while re-winding the parental template strands. (ii) This unwound fork is now in the right conformation to be bound by PriA in the orientation to load DnaB correctly onto the lagging-strand template. (iii) The hand-off reaction from PriA to PriB, to DnaT to DnaC to DnaB ensures that the replisome is reassembled. (iv) The replisome is loaded correctly to ensure the restoration of the DNA lost during the resection of the break. The box shows the domains of PriA as determined by X-ray crystallography [38]. The N terminus of the protein encodes the 3’ end-binding domain (3’BD–red). This is followed by a winged-helix domain proposed to interact with the parental DNA duplex (WH–orange). This is followed by two helicase lobes (HL1 –blue and HL2 –green). This is followed by a cysteine-rich region proposed to act as a wedge during helicase action (CRR–purple). Finally the protein is completed by a C-terminal domain that loops back round to the 3’BD (CTD–yellow). The priA300 mutation is predicted to lie in the HL1 domain. C. Action of PriA in the absence of RecG. The region between the dotted vertical lines is enlarged to show the binding of PriA. (i) A replication fork substrate with an extended 5’ new end is available to bind PriA but is not specifically remodelled for this hand-off in the absence of RecG. (ii) and (v) Because the 3’ end is readily available, PriA remodels the fork to ensure that the 3’ end is bound by the 3’BD and the parental duplex is bound by the WH domain. In structure (ii), the PriA helicase domains (HL1 and HL2) bind correctly to the lagging-strand template and in structure (v) the helicase domains bind incorrectly to the new lagging-strand. The PriA(K230R) helicase (present in the priA300 mutant) retains only the ability to bind correctly. (iii) and (vi) DnaB is loaded via the hand-off mechanism from PriA to PriB to DnaT to DnaC to DnaB. In (iii) DnaB is loaded correctly to the lagging-strand template and in (vi) DnaB is loaded incorrectly to the new lagging-strand. (iv) and (vii) A replication fork is reassembled. In (iv) the replication fork is assembled in the correct orientation to restore the DNA lost in the early stages of recombination at the site of the DSB. In (vii) the replication fork is assembled in the incorrect orientation and replicates the DNA flanking the DSB site. The direction of translocation of DnaB is indicated by a tan arrow. D. Action of PriA in the absence of RecG and RuvAB. The region between the dotted lines is enlarged to show the binding of PriA. (i) A replication fork with an extended 5’ “new” end is available for binding by PriA. However, the Holliday junction associated with the fork is not resolved by RuvABC and the fork itself is not remodelled by RecG. (ii) PriA has difficulty to remodel the fork to allow binding in the 3’ end-binding mode because the presence of the Holliday junction interferes with the required movement of the arms of the fork. This results in a significant proportion of molecules being bound by PriA in its helicase mode where it unwinds the parental duplex arms. (iii) This causes over-winding of the parental arms of the fork and under-winding of the D-loop, resulting in its dissociation. The direction of translocation of PriA is indicated by a black arrow.
Fig 6.
Model proposed for the over-replication of the terminus region between terA and terB in the absence of RecG.
(i) A replication fork is shown having traversed the terminus region in the direction from terA to terB where it is arrested. (ii) In the absence of RecG, PriA binds incorrectly at terB and causes a replication fork to be assembled that moves in the reverse direction towards terA. Whether the arrested fork is originally broken and repair is attempted prior to the assembly of the backward-directed replication fork is unknown. Whether or not the fork is broken, a single DNA end is generated adjacent to terB. (iii) The backward directed replication fork is blocked at terA and recombination of the DNA end with the intact duplex is attempted. (iv) The same process of assembly of a backward directed replication fork is set up at terA, this time moving towards terB. This game of ping-pong between terA and terB continues indefinitely with a preference for attempted DSBR events close to the ter sites but also with more internal sites derived from the D-loops generated from attempted recombination events. Replication forks can either start the process by being blocked at terB (as shown in (i)) or at terA. The combination of all the events occurring in the population results in the accumulation of DNA observed in a ΔrecG mutant between terA and terB.