Figure 1.
Cartoon representing the structure of the selected transformations around the pneumococcal cps locus.
The kanamycin resistance cassette (orange) on the donor DNA molecule is flanked by pbp genes (yellow), both of which are β lactam-sensitive alleles. The equivalent locus in the recipient chromosome is occupied by the much longer type 23F capsule biosynthesis cluster (represented by a series green boxes to indicate the relatively large number of genes comprising this cluster), which is flanked by pbp alleles that confer β lactam resistance on the host cell. The selected transformations involve in the exchange of DNA from the donor into the recipient, with boundaries in the regions shaded blue.
Figure 2.
Distribution of recombination events across the genome.
(A) Cartoon describing the structure of RSSs as discussed in the text. (B) A detailed view of the RSSs spanning and surrounding the cps locus. The annotation of the region is shown at the top, with the cps locus and flanking pbp genes marked. The red line denotes the extent of the ‘primary locus’ (see text). Underneath, in the panel indicated by the dashed boundary, the RSSs affecting this locus are indicated on the rows by black and grey blocks, as displayed in panel (A). There is a row for each of the 84 transformants, segregated according to the amount of DNA with which they were transformed. (C) Wider view of recombination across the genome. A simplified annotation of the 2,182,009 bp S. pneumoniae 23F-R genome is displayed across the top, with the site of the selected recombination (the cps locus) labelled along with other major chromosomal loci. The RSSs are displayed as indicated in panel (A), with the strains in the same order as in panel (B).
Figure 3.
Structure of recombinations within the primary locus.
(A) Annotation of two sections immediately flanking the cps locus, within the primary locus. The position of the cps locus is marked by a horizontal black bar, with a vertical grey shaded band underneath; this locus, across which all recombinations span, is not shown to scale relative to the flanking regions. The extent of an IS element insertion in S. pneumoniae 23F-R, not present in S. pneumoniae TIGR4Δcps, is marked by a second grey vertical band within the 5′ flank. A scale bar, labelled with the genome coordinates and with tick marks every 1 kb, is displayed beneath the annotation. (B) A plot of sequence identity showing the level of similarity between the donor and recipient throughout the locus. The value on the graph represents the proportion of the aligned flanking 100 bp that is identical in both strains for each base in the displayed region. (C) Distribution of recombination sizes. The selected RSSs spanning the cps locus from each of the sequenced 84 isolates are displayed as explained in Figure 2A. At the cps locus itself, all 84 isolates have the donor allele. On the left side, in (i), the RSSs are ordered by their 5′ boundary, while on the right side, in (ii), they are independently ordered by their 3′ boundary. The positions of these boundaries can be modelled as an exponential decay, which is represented by the red line and displayed rate parameter for each flank.
Figure 4.
Distribution of recombination sizes.
(A) Histogram showing the distribution of RSS sizes as L50R lengths. The vertical green line shows the length of the cps locus in the recipient's genome (21,373 bp). The recombinations longer than this are the selected events importing the kanamycin resistance marker. The smaller, unselected recombinations are modelled as an exponential distribution (red line), with the calculated rate parameter displayed. (B) Histogram showing the distribution of donor molecule lengths participating in primary recombinations, as estimated using the L50D lengths of the NCR boundaries. The vertical green line shows the length of the kanamycin resistance locus in the donor DNA (1,354 bp). (C) Histogram showing the distribution of secondary RSS sizes as L50R lengths. These are modelled as an exponential distribution, indicated by the red line and displayed rate parameter. (D) Histogram showing the distribution of secondary NCR L50D lengths, thereby estimating the sizes of donor molecules participating in secondary recombinations. These are also modelled as an exponential distribution, indicated by the red line and annotated rate parameter.
Figure 5.
Distribution of secondary recombinations with respect to sequence divergence between the donor and recipient strains.
The red line represents the proportion of non-overlapping windows of the recipient sequence, containing at least one marker SNP, that have a mean SNP sequence identity of, or greater than, the threshold marked on the x axis. The blue line represents the same statistic for the subset of windows that overlap with at least one detected secondary RSS. The two lines are very similar, suggesting that recombinations are not concentrated in a subset of windows that have a relatively high level of sequence identity between the donor and recipient. A hypergeometric test for enrichment of recombinations in windows with high levels of sequence similarity was performed at 0.1% intervals; the calculated p values are displayed as a black dashed line relative to the y axis on the right side, confirming the lack of a significant relationship.
Table 1.
The distribution of insertions, deletions and small interspersed repeats relative to RSSs.
Figure 6.
Frequencies of different base substitutions within secondary recombinations.
(A) Scatterplot showing the incidence of different substitutions outside the primary locus relative to their frequency in secondary recombinations. Points are coloured according to the efficiency with which they are transferred through transformation when mismatch repair is active (see key). This shows that SNPs repaired effectively by mismatch repair are the most common, and are represented in secondary recombinations at the same level as those SNPs repaired less effectively. (B) Scatterplot showing the frequency of different substitutions when imported at the extreme ends of RSSs (x axis) relative to their frequency as the nearest flanking marker SNP, which is not converted to the donor allele (y axis); this compares the types of substitution found at the two polymorphic sites that define BRs. This fails to provide any evidence that localised mismatch repair processes may affect the positioning of transformation event boundaries or be the cause of the mosaic structure of recombinations, as SNPs repaired more efficiently by mismatch repair are approximately equally represented within, and surrounding, secondary RSSs.
Figure 7.
Distribution of recombination events across the genome when testing the effect of MMR, displayed as described in Figure 2C.
The transformants are grouped according to their genotype, either wild type or mismatch repair defective. Both were transformed with 500 ng mL−1 donor DNA.
Figure 8.
Histogram comparing frequencies of detected recombination sizes in a collection of PMEN1 isolates and in vitro transformants.
The algorithm used to detect sequence imports in a sample of PMEN1 genomes was individually applied to each transformant sequence from the first experiment described in this work. In both graphs, the blue bars show the output of the algorithm applied to sequences from this experiment, whereas the red bars represent the data from the PMEN1 strains. (A) Comparison of the ten serotype switching recombinations characterised in PMEN1 with the transformation events spanning the cps locus selected in vitro. (B) Comparison of the unselected secondary transformation events characterised in vitro with the 615 homologous recombinations (i.e. excluding events overlapping prophage or conjugative elements) identified in the PMEN1 isolates.