Bacterial DNA polymerases participate in oligonucleotide recombination

Abstract

Synthetic single-strand oligonucleotides (oligos) with homology to genomic DNA have proved to be highly effective for constructing designed mutations in targeted genomes, a process referred to as recombineering. The cellular functions important for this type of homologous recombination have yet to be determined. Towards this end, we have identified Escherichia coli functions that process the recombining oligo and affect bacteriophage λ Red-mediated oligo recombination. To determine the nature of oligo processing during recombination, each oligo contained multiple nucleotide changes: a single base change allowing recombinant selection, and silent changes serving as genetic markers to determine the extent of oligo processing during the recombination. Such oligos were often not incorporated into the host chromosome intact; many were partially degraded in the process of recombination. The position and number of these silent nucleotide changes within the oligo strongly affect both oligo processing and recombination frequency. Exonucleases, especially those associated with DNA Polymerases I and III, affect inheritance of the silent nucleotide changes in the oligos. We demonstrate for the first time that the major DNA polymerases (Pol I and Pol III) and DNA ligase are directly involved with oligo recombination.

Fig. 1.

Oligo processing at the replication fork. A. A replication fork with template (green) and newly synthesized (blue) DNA is shown. Pol III (yellow circles) extends from the 3′ ends of the Okazaki fragment and the leading-strand. B. A lagging-strand oligo (purple) is shown bound by Beta (grey ovals) to the template at a single-strand gap. Pol I, which degrades the RNA primer of the Okazaki fragment, can also degrade the 5′ end of the lagging-strand oligo after replacing Pol III. DNA polymerization by Pol I readies the recombinant oligo for DNA ligation and incorporation into the genome. C. A leading-strand oligo (purple) is shown bound by Beta (grey ovals) to the leading-strand template at the single-strand gap ahead of the newly synthesized leading-strand. An unknown exonuclease(s) (orange circle) processes the 3′ end of the oligo. Pol I and/or an unknown exonuclease (red circle) is involved in the degradation of the 5′ end of the leading-strand oligo.

Fig. 2.

Experimental outline for generating and analysing GalK⁺ recombinants. The experiments were performed as outlined. Although at least 48 independent recombinants were analysed, on occasion an individual colony yielded poor sequence and thus was not added to the data.

Fig. 3.

Recombination with multiply marked, lagging-strand oligos. A summary of oligos used to examine the effect of marker density and spacing is shown. Oligo 144 is 70 nt, oligo XT524 is 117 nt, and all others are 75 nt in length. Experiments were performed in the MMR mutant host, HME68. In (A) features of each oligo are shown with a diagram (to scale except XT524), indicating the number and placement of mismatches. The ‘+’ denotes the selected base and the ‘|’ denotes the unselected markers. The number above markers indicates the distance in bases from the 5′ end. Recombination efficiency is normalized to 10⁸ viable cells; efficiencies given here are a representative experiment from at least 3 independent experiments where variability was less than fivefold. 95% confidence limits can be found in Table S4. In (B), the percentage of recombinants that lost markers from either the 5′ or 3′ end are enumerated for each oligo. If a recombinant lost markers from both ends it is not recorded in either column, but such events happen and are shown (5′ & 3′) in Figs 4 and S1. Panel C indicates the Figure that contains the data for each of the oligos. NA: not applicable, either because no unselected markers are present on that end, or because there were no recombinants (XT21). For oligo XT21, 16 rare Gal⁺ colonies were recovered, 13 of which did not inherit the selected ‘T’ marker from the oligo, i.e. they were spontaneous revertants. The three isolates that have a TA’T’ codon could either be derived from the oligo (true recombinants) or be revertants (Fig. S1B and legend). These are unlikely to be true recombinants because no other oligo markers were inherited, and they occur at the same frequency as other spontaneous revertants.

Fig. 4.

Marker loss pattern in wild type and polymerase mutant cells. Marker loss is compared for the lagging- and leading-strand oligo recombinants in ‘wild type’, HME68 (see A and B respectively). Similar experiments are shown for the Pol I mutant derivative of HME68, XTL70, which is defective for the 5′ → 3′ exonuclease activity (C and D), the Pol III mutant, XTL76 (E and F), and the ligase mutant, XTL47 (G and H). The recombination efficiency of the illustrated experiment is shown at the top of each panel. Normal variation in recombination frequency is less than fourfold. In each panel, the ‘Position’ row indicates the distance of the markers from the left end of the oligo as diagrammed, with the ‘+’ denoting the selected marker at position 39, which is the G of the TAG amber codon in the ‘Host’ sequence. The ‘Oligo’ row shows the sequence changes present on the oligo at the indicated position. The sequences of Gal⁺ recombinants are shown below the ‘Oligo’ row; grey shaded spaces indicate those bases that remained unchanged from the host and show where markers were lost from the oligo, white spaces indicate those markers that were inherited from the oligo. For each panel, recombinants are grouped according to their pattern of marker loss. The uppermost group of recombinants showed no marker loss. Groups with 3′, 5′, or 5′ & 3′ marker loss are indicated. The final group, ‘Other’, comprises recombinants with internal markers lost. In (A), (C), (E) and (G), the lagging-strand oligo XT13 (Table S2 and Fig. 3) was used for recombination. In (B), (D), (F) and (H), the leading-strand oligo XT14 (Table S2) was used. The oligos XT13 and XT14 are designed to create the same changes in the final product and have complementary sequences; their 5′ → 3′ orientation is indicated by the arrow above each panel. In all panels, the bases shown for the ‘Host’ are from the galK_am coding sequence. For simplicity of presentation, the oligo bases in (B), (D), (F) and (H) are shown as identical to those in (A), (C), (E) and (G) but in reality the leading-strand oligo (XT14) for (B), (D), (F) and (H) contains markers that are the complement of the sequence shown.

Fig. 4.

Marker loss pattern in wild type and polymerase mutant cells. Marker loss is compared for the lagging- and leading-strand oligo recombinants in ‘wild type’, HME68 (see A and B respectively). Similar experiments are shown for the Pol I mutant derivative of HME68, XTL70, which is defective for the 5′ → 3′ exonuclease activity (C and D), the Pol III mutant, XTL76 (E and F), and the ligase mutant, XTL47 (G and H). The recombination efficiency of the illustrated experiment is shown at the top of each panel. Normal variation in recombination frequency is less than fourfold. In each panel, the ‘Position’ row indicates the distance of the markers from the left end of the oligo as diagrammed, with the ‘+’ denoting the selected marker at position 39, which is the G of the TAG amber codon in the ‘Host’ sequence. The ‘Oligo’ row shows the sequence changes present on the oligo at the indicated position. The sequences of Gal⁺ recombinants are shown below the ‘Oligo’ row; grey shaded spaces indicate those bases that remained unchanged from the host and show where markers were lost from the oligo, white spaces indicate those markers that were inherited from the oligo. For each panel, recombinants are grouped according to their pattern of marker loss. The uppermost group of recombinants showed no marker loss. Groups with 3′, 5′, or 5′ & 3′ marker loss are indicated. The final group, ‘Other’, comprises recombinants with internal markers lost. In (A), (C), (E) and (G), the lagging-strand oligo XT13 (Table S2 and Fig. 3) was used for recombination. In (B), (D), (F) and (H), the leading-strand oligo XT14 (Table S2) was used. The oligos XT13 and XT14 are designed to create the same changes in the final product and have complementary sequences; their 5′ → 3′ orientation is indicated by the arrow above each panel. In all panels, the bases shown for the ‘Host’ are from the galK_am coding sequence. For simplicity of presentation, the oligo bases in (B), (D), (F) and (H) are shown as identical to those in (A), (C), (E) and (G) but in reality the leading-strand oligo (XT14) for (B), (D), (F) and (H) contains markers that are the complement of the sequence shown.

Fig. 5.

Marker loss pattern of lagging-strand oligo XT36 and leading-strand oligo XT418. Loss of markers at each position along the oligos was analysed. Each oligo has one marker every 6 nucleotides, and the x-axis indicates those positions relative to the 5′ end of each oligo. The per cent marker loss at each position is plotted on the y-axis. Note that 100% marker loss is at the origin. The lagging-strand oligo XT36 (Fig. S1A) is indicated by a solid line with closed circles, and the leading-strand oligo XT418 (Fig. S2A) is indicated by a dashed line with open circles.

Fig. 6.

Structure of the polA gene and mutants. This figure is approximately to scale. A. The polA gene encodes the three functional domains of the Pol I protein, which are indicated by the shaded boxes: (1) a 5′ → 3′ exonuclease that mediates nick translation during lagging-strand synthesis; (2) a 3′ → 5′ exonuclease that mediates proofreading; (3) a DNA polymerase activity. B. The resA1 amber mutation (TAG) is located between the two exonuclease domains. C. The sequence that encodes the 5′ → 3′ exonuclease domain of Pol I is replaced by the cat open reading frame to create the polA (5′ → 3′ exo)<>cat mutation (XTL70). D. The sequence that encodes the 5′ → 3′ exonuclease domain is replaced by the cat cassette with a stop codon followed by ATG, a translation restart (SD/AUG) for expression of the downstream sequences.