SCRaMbLE generates designed combinatorial stochastic diversity in synthetic chromosomes

Abstract

Synthetic chromosome rearrangement and modification by loxP-mediated evolution (SCRaMbLE) generates combinatorial genomic diversity through rearrangements at designed recombinase sites. We applied SCRaMbLE to yeast synthetic chromosome arm synIXR (43 recombinase sites) and then used a computational pipeline to infer or unscramble the sequence of recombinations that created the observed genomes. Deep sequencing of 64 synIXR SCRaMbLE strains revealed 156 deletions, 89 inversions, 94 duplications, and 55 additional complex rearrangements; several duplications are consistent with a double rolling circle mechanism. Every SCRaMbLE strain was unique, validating the capability of SCRaMbLE to explore a diverse space of genomes. Rearrangements occurred exclusively at designed loxPsym sites, with no significant evidence for ectopic rearrangements or mutations involving synthetic regions, the 99% nonsynthetic nuclear genome, or the mitochondrial genome. Deletion frequencies identified genes required for viability or fast growth. Replacement of 3' UTR by non-UTR sequence had surprisingly little effect on fitness. SCRaMbLE generates genome diversity in designated regions, reveals fitness constraints, and should scale to simultaneous evolution of multiple synthetic chromosomes.

Figure 1.

The circular synIXR synthetic chromosome comprises 43 segments, numbered consecutively as shown, with loxPsym sites (orange bars) serving as junctions between adjacent segments. Arrows indicate genes, with colors denoting essential genes (red), auxotrophic markers (purple), other nonessential genes (light blue), and the chloramphenicol resistance marker (green). The centromere, in segment 2, is shown as a black circle. The left and right ends of each segment are denoted “L” and “R,” and loxPsym sites are denoted by the unique half-sites of the segments they join. For example, the loxPsym site at the junction between segments 1 and 2 involves half-sites 1R and 2L.

Figure 2.

Rearrangements were observed in synIXR SCRaMbLE strains. Each SCRaMbLE strain is represented as a sequence of arrows. The color of each arrow indicates the segment number in the parental chromosome, and the direction of the arrow represents the orientation (SCRaMbLEgram visualization). A red border denotes a segment containing an essential gene. The names of slow growth strains are indicated in red text.

Figure 3.

Dot-plots illustrate the pattern of the rearrangements observed for six SCRaMbLE strains. In each case, the order of segments in the SCRaMbLE strain (y-axis) is compared with the parental order (x-axis), with orientation either the same (black dot) or inverted (blue dot). (A) Simple deletions in JS629 create gaps. (B) Inversions in JS738 appear as blue stripes perpendicular to the main diagonal. (C) Tandem duplications in JS621 create parallel stripes. (D) Inverted duplications in JS614 create alternating perpendicular and parallel stripes. (E,F) Increasingly complicated patterns in JS735 and JS710 arise from recombination events within previous events, particularly when duplications are already present.

Figure 4.

The fate of each segment in each strain is indicated as deleted (gray), inverted (blue), tandem duplication (orange), inverted duplication (red), complex (purple), or unaffected by any SCRaMbLE event and remaining at copy number 1 (white).

Figure 5.

(A) The distribution of copy number for each segment is shown across strains. (B) Each segment was involved in at least one duplication event.

Figure 6.

(A) The number of SCRaMbLE events of each type is depicted for each strain. While the number of events per strain is 6.2 ± 4.9, strains with duplications experience far more events overall. (B) Each event can lead to one or more novel junctions. The junctions observed in each strain are classified as parental, arising from specific types of events, or arising from multiple events (e.g., a deletion followed by an inversion) or have origins too complex to classify. (C) Novel junctions can create genome structures not observed in nature, including convergent coding domains lacking proper UTRs.

Figure 7.

(A) The number of events per strain has a long tail, with strains observed having 17 and 18 distinct recombination events. (B) Events are more likely at short distances but continue to be observed for long separations between recombination sites. There is a preponderance of deletions over inversions at the short lengths.

Figure 8.

In several strains, recombinations involving auxotropic markers have generated convergeons (convergent CDS regions joined by a loxPsym site) that nevertheless support the prototrophic phenotype.