GCE-SCRaMbLE allows tight control of Cre recombinase activity
The activity regulation of Cre enzyme by GCE was previously proposed in success using engineered pyrrolysyl-tRNA synthetase/tRNACUA (PylRS/tRNACUA) pairs to encode photocaged amino acids20,21. However, the efficiency of these developed PylRS/tRNACUA pairs is low in yeast22. Thus, we introduced an engineered leucyl-tRNA synthetase/tRNA pair (LeuOmeRS/tRNACUA) to the translational machinery of yeast cells to achieve efficient incorporation of O-methyl-L-tyrosine (OMeY) into the Cre recombinase in response to UAG stop codon23. The Cre enzyme is driven by the daughter-cell-specific promoter of SCW11 to avoid on-going recombination events post-SCRaMbLE. In this way, the expression of full-length Cre recombinases depends on the presence of OMeY in cells (Fig. 1a). In previous study, the estrogen-binding domain (EBD) was fused to the C-terminal of Cre recombinase to allow the estradiol-inducible regulation of SCRaMbLE24. However, the Cre recombinase functions as a tetrameric synaptic complex25, and we suspect that the fusion of C-terminus EBD might affect its activity. To test our hypothesis, we measured the recombination kinetics of purified Cre and Cre-EBD recombinases using an intramolecular excision assay26. Our result showed that the Cre-EBD exhibited 2.2-fold reduced affinity and 3-fold slower turnover rates compared with Cre without EBD, highlighting the presence of EBD does impair Cre enzyme activity (Supplementary Fig. 1). Therefore, the EBD was eliminated from our proposed GCE-SCRaMbLE system.
We next attempted to select a permissive residue of Cre enzyme for site-specific incorporation of OMeY. The N-terminal region of Cre recombinase was found to tolerate an insertion mutation27. Thus, codons corresponding to selected N-terminal residues (L5, L11, L14 and A18) were individually replaced to the UAG stop codon via site-directed mutagenesis. To test the performance of GCE-SCRaMbLE in yeast cells, a previously developed reporter system containing a loxP-flanked terminator cassette between the ADH1 promoter and the CDS of GFP reporter protein16 was utilized to measure the recombination frequency (Fig. 1b). We transformed three plasmids encoding the LeuOmeRS/tRNACUA pair, GFP reporter and Cre recombinase variants (CreUAG5, CreUAG11, CreUAG14, and CreUAG18) into yeast cells, followed by culturing cells in the presence and absence of 1 mM OMeY. Our data showed GCE-SCRaMbLE system could effectively trigger terminator deletion via recombination between two loxP sites induced by OMeY, leading to the detection of fluorescence signal. In addition, we observed CreUAG5-mediated GCE-SCRaMbLE was most efficient as the yeast cells expressing CreUAG5 exhibited the highest fluorescence intensity compared with other Cre variants. Consistent with the previous report19, we found that the Cre-EBD system was indeed leaky: around 20% relative fluorescence signal was observed in the absence of β-estradiol. In contrast, our GCE-SCRaMbLE system showed minimal fluorescence signals at the background level in the absence of OMeY (Fig. 1b). Next, we investigated whether GCE-SCRaMbLE could be regulated by OMeY concentration. Using most active Cre recombinase variants (CreUAG5 and CreUAG14), we found the relative fluorescence intensity was positively correlated with OMeY concentration at 0, 0.5, 1, 2, 5 and 10 mM (Fig. 1c, d). Overall, we demonstrated GCE-SCRaMbLE exhibited low basal activity and allowed tight control of loxP-mediated recombination in a dose-dependent manner in our designed plasmid-based system.
GCE-SCRaMbLE generates genome-wide recombination in synthetic yeast chromosomes
Taking advantage of the high controllability of GCE-SCRaMbLE system, we sought to determine the effect of Cre recombinase abundance, the conformation of synthetic chromosome and genome ploidy on the SCRaMbLE outcome. A series of Sc2.0 yeast strains were used in this study: haploid yeasts harboring a single synthetic chromosome in both linear and circular form (synII, ring_synII) and multiple synthetic chromosomes (syn2369R: synII, synIII, synVI, and synIXR), heterozygous diploid yeast strains generated by mating the forementioned haploid synthetic yeast strains with wild-type haploid yeast (BY4741 or BY4742). We transformed Cre recombinase variants (CreUAG5 and CreUAG14) and LeuOmeRS/tRNACUA pair plasmids into these strains growing in the presence of OMeY to promote GCE-SCRaMbLE. In addition, four concentrations of OMeY at 1, 2, 5 and 10 mM were chosen for the induction using CreUAG14. In total, 23 test groups were designed to fully dissect the influence and corresponding contribution of each factor during the SCRaMbLE process (Fig. 2a, Supplementary Table. 1).
SCRaMbLE occurrence rate was found to be positively correlated with cell death rate presumably because recombination events could lead to the deletion of essential genes11. To determine the appropriate induction time, cultured cells of haploid strains induced by GCE-SCRaMbLE with 1 mM OMeY were plated at different time points for monitoring cell viability. We found cells exhibited the highest mortality after OMeY treatment for 24 hours for most haploid strains (Supplementary Fig. 2). Thus, to maximize viable DNA deletion by GCE-SCRaMbLE, the induction time at 24-hour was chosen for this study. To prevent early events that might dominate SCRaMbLE recombination outcomes, we performed 30 independent SCRaMbLE inductions in each group without selective pressures. For each induction, two colonies were randomly chosen post-recovery. To this end, 1380 (= 23×30×2) GCE-SCRaMbLEd yeast strains were selected for SCRaMbLEd genome reconstruction and rearrangement analysis. In parallel with GCE-SCRaMbLE induction, four yeast pools (haploid syn2369R and ring_synII strains expressing CreUAG5 and CreUAG14 recombinase variants respectively) in the absence of OMeY were also prepared for deep sequencing (~ 17,000×) to identify potential “leaky” events that might be overlooked by single colony selection.
We confirmed that no recombination was observed on synthetic chromosomes when designed yeast pools were cultured in the absence of OMeY, highlighting the extremely low basal activity of GCE-SCRaMbLE in its uninduced state. By whole-genome reconstruction of all 1380 GCE-SCRaMbLEd yeast strains, we observed SCRaMbLE events in 29.71% of the strain collection. In total, 1731 rearrangement events were identified in SCRaMbLEd cells, with an average number of SCRaMbLE events per strain at around 4. No intra-chromosomal events were observed even though the quadruple-synthetic strain was utilized. Among all recombination events that occurred in cells, three types of events including deletion, inversion and duplication, were observed at the frequencies of 44.2%, 40.5% and 15.3%, respectively (Fig. 2b). To estimate the efficacy of GCE-SCRaMbLE to generate rearrangements through random recombination between two loxPsym sites, we built a rearrangement landscape by mapping junctions involving distinct recombination events to each loxPsym site on different synthetic chromosomes. We observed the widespread occurrence of GCE-SCRaMbLE-mediated recombinations in all synthetic chromosomes (Fig. 2c). The synII chromosome showed higher number of recombination events than the other chromosomes, because all testing strains contain this chromosome. Taken together, our data demonstrated that GCE-SCRaMbLE was able to generate synthetic yeast derivatives with genome-wide rearrangements.
Recombination frequency correlates positively with Cre enzyme abundance regulated by GCE-SCRaMbLE
As GCE-SCRaMbLE controls Cre recombinase synthesis by incorporating OMeY into the growing peptide during the translation stage, the cellular abundance of Cre was expected to be positively correlated with the OMeY concentration. We performed quantitative western blot analysis to measure the cellular abundance of Cre variants (CreUAG5 and CreUAG14) with a gradient of OMeY concentrations. As expected, the cellular abundance of CreUAG14 was perfectly correlated with the OMeY concentration (Fig. 3a). Consistent with our finding that the efficiency of GCE-SCRaMbLE mediated by CreUAG5 was higher than that of CreUAG14 (Fig. 1b), we also observed a higher abundance of CreUAG5 than that of CreUAG14 in yeast cells grown in the medium supplemented with 1 mM OMeY (Fig. 3a). In addition, compared with CreUAG5, we observed no readthrough of the amber codon in the gene encoding CreUAG14 without OMeY, demonstrating GCE-SCRaMbLE using CreUAG14 has very low basal expression in the uninduced state.
We next aimed to explore whether and how the expression level of the Cre recombinase affects SCRaMbLE recombination outcomes. The concentration of OMeY and cellular abundance of Cre variants are both contributing factors to the expression level of Cre recombinase. Here we investigated each factor and its influence in totally six subgroups of GCE-SCRaMbLEd strains, with different OMeY concentrations or Cre variants, in which 180 haploid strains per group were randomly selected for sequencing analysis. Similar to the results from the plasmid based GFP reporter system, our result showed a positive correlation between the concentration of OMeY and the occurrence rate of SCRaMbLE (Fig. 3b, c). The SCRaMbLE recombination frequency within a post-SCRaMbLE yeast population remained steady once the concentration of OMeY reached and exceeded 5 mM, suggesting the effective concentration of Cre enzyme became saturate. Strains with 8 and 11 recombination events were found when GCE-SCRaMbLE was induced by 5 and 10 mM OMeY, respectively. In contrast, the number of events was lower in the condition of 1 and 2 mM OmeY, at up to 3 and 4 events per strain, respectively (Fig. 3c). For the Cre variants, we found SCRaMbLE frequency and the number of SCRaMbLE events per cell were higher in yeast cells expressing CreUAG5 than that for CreUAG14 (Fig. 3d, e), which could be explained by cellular abundance of these Cre variants. We also tested whether the level of Cre enzyme played a role in the distributions of different recombination types and no evident correlation was revealed (Supplementary Fig. 3). Our result suggests that the regulation of SCRaMbLE recombination frequency could be achieved by fine-tuning the concentration of OMeY and utilizing proper Cre recombinase variants.
Genome ploidy is a key factor affecting deletion capability by SCRaMbLE and resultant proportion of distinct types of events
Deletions of chromosome fragments bearing essential genes via SCRaMbLE could lead to loss of viable cells in haploid strains, thus restricting the extent of deletable chromosome contents and potentially decreasing the genomic diversity post-SCRaMbLE. This issue could be addressed in the diploid background, allowing the synthetic chromosomes to undergo rearrangements while essential genes in native alleles remain unaffected. A recent study showed heterozygous diploid strains are more tolerant to genome rearrangements by SCRaMbLE compared to haploid strains13. To better understand the effect of genome ploidy on deletion capability via SCRaMbLE, we compared the loss of chromosome content on the target synthetic chromosome using 60 haploid strain and another corresponding 60 heterozygous diploid strains, subjected to the same GCE-SCRaMbLE induction. In general, the number of strains with deletion and the degree of chromosome loss was significantly higher in the diploid background (Fig. 4a and Supplementary Fig. 4). For diploid strains, we found 19 strains showing reduced contents in the ring_synII with the lowest chromosome retention rate at 13.95%. In contrast, deletion events on the circular synII were observed in only 13 haploid strains, with the chromosome retention rate higher than 98%. Similarly, we found 26 diploid cells have chromosome content loss (chromosome retention rate ranging from 29.37–99.96%) in linear synII after GCE-SCRaMbLE induction, while only 11 haploid cells with minimal loss of synII content less than 1% were observed.
To further characterize the genomic diversity between haploid and diploid SCRaMbLEd strains, we generated a rearrangement landscape by mapping recombination loci to their corresponding loxPsym sites. We observed two notable effects of ploidy on the SCRaMbLE outcomes. First, the SCRaMbLE rearrangement frequency on the target chromosome synII was significantly higher across the entire chromosome in diploid strains than in haploid counterparts (Fig. 4b). The SCRaMbLE frequency for different regions of synII ranged from 7.92–17.08% in diploid strains, whilst in haploid strain the average frequency of synII is as low as 1.11%. In addition, we noticed a significant difference in the proportion of distinct events between haploid and diploid cells (Fig. 4b). As the symmetry of loxPsym sites allows recombination in either orientation, deletions and inversions were observed with approximately equal frequency in diploid strains as expected. In contrast, the deletion event represented the smallest proportion of rearrangement types in haploid cells; and these deletion events were scattered sparsely through the synII, which could be explained by the distribution of essential genes on synII (Fig. 4b). Taken together, our findings highlight that ploidy has a significant impact on the type, diversity and proportion of SCRaMbLE-mediated rearrangement events.
Circularization of synII leads to enhanced inter-chromosomal contacts and increased number of SCRaMbLE events
Chromosome in circular conformation could lead to many complex rearrangement events after SCRaMbLE15,17. Here we further dissect its influence on the SCRaMbLE outcomes using strains carrying linear or circular synII in haploid or diploid background subjected to GCE-SCRaMbLE induction under the same condition. As expected, we found the average number of recombination events per SCRaMbLEd strain was much higher in cells bearing circular synII than that in the linear counterpart regardless of genome ploidy. Specifically, the average numbers of recombination per SCRaMbLEd strain in haploid and diploid cells were, respectively, 1.12 and 2.67 for linear synII, versus 1.40 and 8.86 for circular synII.
The linear and circular synII share the almost identical sequence but have distinct chromosomal conformation. We next exploited Hi-C to investigate the difference in the 3D structure between the linear and circular synII. To allow the qualitative comparison, 2D contact maps (bin size at 10 kb) and 3D projections for both linear and circular synII were generated based on the ligation frequencies between DNA restriction fragments (Fig. 5a). In general, we found circularization of synII led to apparent changes in its overall structure. The circular synII exhibited increased intra-chromosomal contacts than the linear counterpart, especially between the two subtelomeric regions (30 kb length) at the chromosome ends, which are designated as regions of interest (ROI) in the contact maps and 3D representations (Fig. 5a). A previous study suggests 3D proximity of loxPsym sites on synthetic chromosomes is closely related to the likelihood of deletion event via SCRaMbLE14. We hypothesized that the increased intra-chromosomal contacts by circularization of synII chromosome would lead to increased occurrence of SCRaMbLE events in circular synII compared to linear synII. To test this, we performed ultra-deep sequencing (80,000×) to analyze genome rearrangements of two SCRaMbLEd cell populations that bear linear and circular synII respectively. Only heterozygous diploid strains subjected to GCE-SCRaMbLE (1 mM OMeY) were utilized in the analysis to increase rearrangement frequency. Indeed, we observed more SCRaMbLE events in the circular synII than that in the linear synII (10769 versus 7278). For quantitative analysis, we plotted the number of SCRaMbLE events against the normalized contact counts per bin between the regions involved in these recombination events. A significant positive correlation was revealed between the intra-chromosomal contact strength and the number of SCRaMbLE events (Fig. 5b). In addition, this positive correlation was stronger in circular synII compared that to linear synII. Overall, these findings are consistent with our hypothesis that the frequency of SCRaMbLE events is positively correlated to the intensity of intra-chromosomal contacts.
Since the extra intra-chromosomal contacts in circular synII compared with linear synII mainly derived from the contacts between ROI and the rest part of the chromosome, we next investigated whether these ROI-dependent interactions were responsible for the increased number of SCRaMbLE events in circular synII. We focused on the intra-chromosomal contacts between the ROI in the left arm and other chromosome regions in linear synII or circular synII by means of a bait chromosome capture approach28. The contact counts between the ROI and every 10 kb region in the right arm of synII appeared to be higher in the circular form than that in the linear counterpart, and the contact discrepancies increased from centromere to the chromosome end (Fig. 5c). In a similar manner, we also compared the number of recombination events specifically occurred between the ROI and the rest part of linear and circular synII. We found the number of ROI-involved SCRaMbLE events was much higher in circular synII than that in the linear synII, especially at the telomere adjacent region of the right arm (Fig. 5c). Taken together, our results showed increased intra-chromosomal contacts, mainly from the interactions between the ROI and the other chromosomal parts in synII, gave rise to more ROI-dependent rearrangement events.