Diagram of the variably phosphorothioated (VPT) cassette series. Homology arms to lacZ are shown in blue, and the inserted heterologous kanR gene is shown in gold; the lagging-targeting strand is represented as the top strand, and the leading-targeting strand is represented as the bottom strand. The red asterisk indicates the presence of 4 consecutive phosphorotiohate bonds. B) In vitro Lambda Exo digestion of non-phosphorothioated (VPT1, left), single end phosphorothioated (VPT2, center), and dually phosphorothioated (VPT4, right) dsDNA cassettes. Samples were digested with no enzyme (lanes 1, 5, & 9) and tenfold-increasing amounts of enzyme, from left to right. The bottom band is dsDNA, and the top band is the ssDNA product after Lambda Exo degrades one of the two strands. PAGE analysis confirms that dually phosphorothioated cassettes are protected from degradation. C–E) Recombination frequencies for the VPT cassette series in EcNR2 (C), nuc4⁻ (D), and EcNR2.xseA ⁻ (E). Insertion frequencies were measured as the number of kanamycin resistant recombinants over the total number of cfu (as plated on non-selective media). These data are presented as the mean with the error bars as standard deviation; n = 3 for EcNR2 and EcNR2.xseA ⁻, and n = 2 for nuc4⁻. Numerical data are shown in . F& G) Cassette insertion frequencies were measured in technical replicates for EcNR2, nuc4⁻, and EcNR2.xseA ⁻. Data in F) & G) are presented as the mean with the standard error of the mean. The ratio between VPT4 and VPT1 (F) indicates a strain’s ability to recombine cassettes with terminal PT bonds preventing the direct action of Lambda Exo. The ratio between VPT4 and VPT7 (G) indicates whether a strain is more able to process terminal PT bonds (VPT4) in comparison with internal PT bonds (VPT7).

A non-phosphorothioated lacZ::kanR dsDNA cassette with two 2 bp sets of mismatch mutations (placed 7&8 bp from each end of the cassette) was recombined into EcNR2 and EcNR2.xseA ⁻ cells. Resulting kanR recombinants were then genotyped in order to determine whether they inherited these distally-located mutations. The distribution of mutations in EcNR2 (two independent experiments, n = 180) and EcNR2.xseA ⁻ (two independent experiments, n = 177) was plotted as the frequency of clones inheriting 0 (empty black bars), 1 (hatched black bars), or 2 (filled black bars) sets of mutations. B) When the inheritance of each 2 bp mutation is considered individually, both show increased preservation in EcNR2.xseA ⁻, although this is only statistically significant for the 2 bp mutation on the 3′ end of the cassette (as defined with respect to the lagging-targeting strand). Data are presented as the mean with the standard deviation from the mean. C & D) A 90mer lagging strand-targeting oligonucleotide was designed to introduce 7 stop codons (placed at the +2, +23, +35, +49, +61, +73, +87 base pair positions with respect to the 5′ end) into the lacZ gene. This oligo was transformed into EcNR2, EcNR2.xseA ⁻, and nuc5⁻ cells. LacZ⁻ cells were identified and their lacZ genes sequenced in order to determine which mutations were conferred in a given clone. The average total number of stop codons introduced into a given clone is shown in C). Both EcNR2.xseA ⁻ (**p = 0.0004, n = 54) and nuc5⁻ (**p = 0.001, n = 46) exhibit similar, statistically significant increases in total mutation inheritance compared to EcNR2 controls (n = 45). A breakdown of each strain’s inheritance of each mutation is given in D). Both EcNR2.xseA ⁻ and nuc5⁻ show similarly increased inheritance of 3′-located mutations, but no increased inheritance of 5′-located mutations. No clones inherited the mutation encoded at position +2. Data in C) are presented as the mean with the standard error of the mean.

CoS-MAGE was carried out in three strains (EcNR2, EcNR2.xseA ⁻, and nuc5⁻), using sets of ten oligos encoding TAGTAA mutations, and a co-selection oligo designed to revert a mutated selectable marker located within 500 kb of the targeted loci. Set 1 (A) was co-selected with chloramphenicol acetyltransferase (cat, inserted at the mutS locus). In comparison with EcNR2 (n = 319), both EcNR2.xseA ⁻ (**p = 0.0001, n = 135) and nuc5⁻ (***p<0.0001, n = 257) show statistically significant increases in mean oligo conversion, a decreased proportion of clones exhibiting no allele conversions, and more clones with 5+ conversions. Set 2 (B) was co-selected with beta lactamase (bla, inserted with the λ prophage). Here, nuc5⁻ (n = 142) shows a statistically significant increase in recombineering performance compared to both EcNR2 (***p<0.0001, n = 268) and EcNR2.xseA ⁻ (***p<0.0001, n = 184). Set 3 (C) was co-selected with tolC. Here, nuc5⁻ (n = 139) shows a statistically significant increase in mean allele conversion compared to EcNR2 (*p = 0.002, n = 327). EcNR2.xseA ⁻ (n = 92) shows an intermediate phenotype between EcNR2 (p = 0.2) and nuc5- (p = 0.3). All oligos used in this experiment had two PT bonds on both ends. Data shown in the right panels are presented as the mean with the standard error of the mean. Statistical significance is denoted using a starred system where ns denotes a non-significant variation, * denotes p<0.003, ** denotes p<0.001, and *** denotes p<0.0001.

The effect of oligo phosphorothioation on the CoS-MAGE performance of the various nuclease knockout strains was studied. Variants of recombineering oligo Set 1 (described in ) with no PT bonds (0 PT) and four PT bonds on both ends (4 PT) were compared with the initial version of Set 1, which had two PT bonds on both ends (2 PT). Recombination was performed with cat co-selection as prior. For EcNR2 (A), n = 133 (0 PT), n = 319 (2 PT), n = 186 (4 PT). For EcNR2.xseA ⁻ (B), n = 94 (0 PT), n = 92 (2 PT), n = 86 (4 PT). For nuc5⁻ (C), n = 132 (0 PT), n = 257 (2 PT), n = 183 (4 PT). Data shown in the right panels are presented as the mean with the standard error of the mean. Statistical significance is denoted using a starred system where ns denotes a non-significant variation, * denotes p<0.003, ** denotes p<0.001, and *** denotes p<0.0001.