Lambda red recombineering in Escherichia coli occurs through a fully single-stranded intermediate

Abstract

The phage lambda-derived Red recombination system is a powerful tool for making targeted genetic changes in Escherichia coli, providing a simple and versatile method for generating insertion, deletion, and point mutations on chromosomal, plasmid, or BAC targets. However, despite the common use of this system, the detailed mechanism by which lambda Red mediates double-stranded DNA recombination remains uncertain. Current mechanisms posit a recombination intermediate in which both 5' ends of double-stranded DNA are recessed by λ exonuclease, leaving behind 3' overhangs. Here, we propose an alternative in which lambda exonuclease entirely degrades one strand, while leaving the other strand intact as single-stranded DNA. This single-stranded intermediate then recombines via beta recombinase-catalyzed annealing at the replication fork. We support this by showing that single-stranded gene insertion cassettes are recombinogenic and that these cassettes preferentially target the lagging strand during DNA replication. Furthermore, a double-stranded DNA cassette containing multiple internal mismatches shows strand-specific mutations cosegregating roughly 80% of the time. These observations are more consistent with our model than with previously proposed models. Finally, by using phosphorothioate linkages to protect the lagging-targeting strand of a double-stranded DNA cassette, we illustrate how our new mechanistic knowledge can be used to enhance lambda Red recombination frequency. The mechanistic insights revealed by this work may facilitate further improvements to the versatility of lambda Red recombination.

Figure 1.—

Previously proposed lambda Red-mediated dsDNA recombination mechanisms. Heterologous dsDNA is shown in green; Exo is an orange oval, and Beta is a yellow oval. In both mechanisms the recombination intermediate is proposed to be a dsDNA core flanked on either side by 3′ ssDNA overhangs. (A) The Court mechanism posits that (1) Beta facilitates annealing of one 3′ overhang to the lagging strand of the replication fork. (2) This replication fork then stalls and backtracks so that the leading strand can template switch onto the synthetic dsDNA. The heterologous dsDNA blocks further replication from this fork. (3) Once the second replication fork reaches the stalled fork, the other 3′ end of the integration cassette is annealed to the lagging strand in the same manner as prior. Finally, the crossover junctions must be resolved by unspecified E. coli enzymes (Court et al. 2002). (B) The Poteete mechanism suggests that (1) Beta facilitates 3′ overhang annealing to the lagging strand of the replication fork and (2) positions the invading strand to serve as the new template for leading-strand synthesis. This structure is resolved by an unspecified host endonuclease (red triangle), and (3) the synthetic dsDNA becomes template for both lagging and leading-strand synthesis. A second template switch must then occur at the other end of the synthetic dsDNA (Poteete 2008). The figure was adapted from the references cited.

Figure 2.—

Lambda Red mediated dsDNA recombination proceeds via a ssDNA intermediate. Instead of a recombination intermediate involving dsDNA flanked by 3′-ssDNA overhangs, we propose that one strand of linear dsDNA is entirely degraded by Exo (orange oval). Beta (yellow oval) then facilitates annealing to the lagging strand of the replication fork in place of an Okazaki fragment. The heterologous region does not anneal to the genomic sequence. This mechanism could account for gene replacement (as shown) or for insertions in which no genomic DNA is removed.

Figure 3.—

Strand bias in lambda Red ssDNA insertion recombination. Recombination frequencies were assessed for several leading-targeting and lagging-targeting complementary ssDNA pairs. Lagging-targeting strands were found to be more recombinogenic than leading-targeting strands. An asterisk indicates P < 0.05.

Figure 4.—

Strand-specific mismatch alleles were used to identify the strand of origin for each recombined mutation. The mismatched lacZ∷kanR cassette contained two consecutive mismatches at two loci in both flanking homology regions. Strand 1 was the lagging-targeting strand and strand 2 was the leading-targeting strand. If lambda Red dsDNA recombination proceeds via a ssDNA intermediate (left), (a) one Exo (orange oval) binds to a dsDNA end, (b) Exo fully degrades one strand while helping to load Beta (yellow oval) onto the remaining strand, and (c) this strand provides all of the genetic information during recombination. This figure shows the case in which the lagging-targeting strand is recombined (coding-strand genotypes: L1, AA; L2, AA; L3, TT; L4, TT), but the leading-targeting strand is also predicted to be observed (coding-strand genotypes: L1, CC; L2, CC; L3, GG; L4, GG). If the lambda Red recombination intermediate is a heterologous dsDNA core flanked by 3′-ssDNA overhangs (right), (a) one Exo binds to each dsDNA end, (b) Exo recesses both strands while helping to load Beta onto both 3′ overhangs, and (c) both strands provide genetic information for each recombination. Since Exo always degrades 5′ → 3′, the expected coding-strand genotypes for the Court and Poteete mechanisms would be L1, CC; L2, CC; L3, TT; L4, TT.

Figure 5.—

Testing the effect of strand protection on recombination frequency. Four lacZ∷kanR cassettes were tested to determine whether protecting one strand has a greater effect on recombination frequency than protecting the other strand. In each case, protection was accomplished through the placement of four phosphorothioate linkages on the 5′ end of a strand. Inset: Analysis of variance for lagging-targeting (Lag) phosphorothioation and leading-targeting (Lead) phosphorothioation. An asterisk (*) denotes phosphorothioation. Lagging-targeting phosphorothioation was found to significantly enhance recombination frequency, whereas leading-targeting phosphorothioation did not affect recombination frequency.