ERdd system in yeast
The in vivo behavior of the yeast-adapted ERdd tag was tested by expressing a GFP-ERdd fusion protein from a centromeric vector over a range of estradiol concentrations and assessing GFP fluorescence (Fig. 1B). In the absence of estradiol, specific GFP fluorescence was reduced by 95% compared to the fluorescence with 10 µM estradiol. At this highest tested estradiol concentration, specific GFP fluorescence of the GFP-ERdd samples reached 70% of that of untagged GFP. This indicated an exploitable dynamic window to leverage the ERdd tag to couple organism survival to estradiol supplementation by fusion to suitable essential proteins.
Essential GFP-ERdd library creation and high-throughput screening for estradiol response
We performed a large-scale screen of essential genes for the desired growth response when fused to the ERdd tag – uncompromised fitness in the presence of 1 µM estradiol (permissive condition) and a severe fitness impediment in its absence (restrictive condition). To this end, the yeast GFP collection was leveraged to facilitate ERdd tag integration. The yeast GFP collection contains 4,159 strains, in which ORFs are individually tagged with a C-terminal GFP and a downstream HIS3MX6 selectable marker gene. This constant region was used as a landing pad for C-terminal addition of the ERdd tag, allowing to use a single donor construct for editing of all strains of interest (Fig. 1C). The edit was mediated by a CRISPR/Cas9 system with an integration cassette also swapping the HIS3MX6 marker for a LEU2 marker gene, facilitating efficient editing and selection of correct edits. Out of 1103 S. cerevisiae ORFs deemed essential1, 822 are physically represented as strains in the yeast GFP collection. Out of these, 775 strains (94%) were converted into GFP-ERdd fusion strains (Supplementary Data S1). The others were either missing (24) in the library distribution we had, did not grow (4), or no Leu+/His- colonies could be obtained (19) upon transformation of the CRISPR/Cas components.
From the constructed essential GFP-ERdd library, estradiol dependent strains were identified by a bipartite screening (Supplementary Data S2). The primary screening for estradiol response was performed by pinning the generated essential GFP-ERdd library on YPD without (restrictive) and with 1 µM estradiol (permissive) and analyzing the colony size ratios (Fig. 2A). For most of the essential genes, ERdd-dependent degradation without estradiol appeared to not be substantial enough to cause a growth-inhibition phenotype, but about 5% of the screened strains showed a clear dependence on estradiol for proper growth (Fig. 2B). The 46 strains with the highest ratio of colony size between permissive and restrictive conditions were chosen for a subsequent screening in liquid culture in plate reader assays. For almost all of these strains differential growth in response to estradiol was observed in the liquid culture screening as well (Fig. 2C,D). Strains were ranked by the mean area under the difference curve between growth under permissive and restrictive conditions.
Most of the screened essential gene-GFP-ERdd fusions did not show a clear estradiol-dependent phenotype. To investigate whether this was due to a lack of ERdd-mediated essential protein degradation, the GFP-ERdd strains of the highly expressed essential genes PGK1, YEF3, TPI1 and GPM1 were assayed for growth and GFP fluorescence when cultured with and without estradiol (Figure S1). Relative fluorescence intensities suggested that substantial proportions of the respective fusion proteins were being degraded upon estradiol withdrawal (47–82% at 24h). However, evidently the remaining essential protein amounts were sufficient to support growth, as the strains showed either no or only minor growth inhibition in the absence of estradiol.
Estradiol response of single and double ERdd tagged strains
The genes corresponding to the six highest ranked strains in the secondary screening, SEC18, SES1, SPC110, SEC14, RRP46 and DIS3, were directly fused to ERdd in a BY4742 background without using a marker gene. The resulting six strains all showed a strong dependency on estradiol for growth (Fig. 3A, Figure S2). A genetic safeguard should not compromise the cell fitness under permissive conditions, as to not create an evolutionary incentive to inactivate the safeguard. Three of the strains, SPC110-ERdd, RRP46-ERdd and DIS3-ERdd, exhibited a growth indistinguishable from the parental strain in permissive medium (YPD with 1 µM estradiol) (Fig. 3B) and were used for further work. Of these, SPC110-ERdd displayed particularly favorable behavior, with a severe growth inhibition without estradiol and full restoration of growth already at 100 nM estradiol (Fig. 5A).
Subsequently, strains with two ERdd-tagged genes were created for combinations of SPC110, RRP46 and DIS3. All dual ERdd strains exhibit strict dependence on estradiol. Growth of SPC110-ERdd/RRP46-ERdd was fully restored with 100 nM estradiol, whereas the other two strains required a higher estradiol concentration, reaching wild-type-like growth at 1 µM estradiol (Fig. 4, Figure S3).
Escape rates and analysis of escapees
The estradiol regulated strains not exhibiting any growth impairment under permissive conditions, SPC110-ERdd, RRP46-ERdd, DIS3-ERdd and derived double ERdd-tagged strains, were assessed for the stringency of containment they offer. This was done by assaying the frequency of escape colonies occurring on restrictive medium. To this end, eight independent cultures of the assayed strain were plated on YPD without estradiol (restrictive environment) and at a suitable dilution on YPD with estradiol (permissive environment) to determine the number of colony-forming units.
Afforded containment varied drastically between the three genes when individually tagged with ERdd. SPC110-ERdd had the lowest escape frequency of 7.0×10− 8, followed by DIS3-ERdd with 6.1×10− 7. RRP46-ERdd kept growing very slowly on restrictive plates, leading to the continued emergence of escaper colonies with prolonged incubation, making the quantification of an escape frequency infeasible. Combining SPC110-ERdd with an ERdd tag on either RRP46 or DIS3 resulted in stringently contained strains with escape frequencies below the detection limit of this assay, not yielding a single escapee colony. Subsequently, assays for low escape frequencies were carried out for these strains, plating about 5×109 colony forming units (CFUs) on restrictive plates, which were replica plated for single colony detection. Again, no escapee colonies were observed. Accordingly, escape frequencies of SPC110-ERdd/RRP46-ERdd and SPC110-ERdd/DIS3-ERdd were below the assay’s detection limit of 2×10− 10. In contrast, ERdd tagging both RRP46 and DIS3 yielded an escape frequency of 4.3×10− 7, not markedly improving containment over singly tagged DIS3-ERdd (6.1×10− 7). Interestingly, both RRP46 and DIS3 code for a subunit of the exosome complex essential for RNA metabolism.
Eight escapee strains from each of the three singly ERdd-tagged were isolated, and the ERdd-tagged gene was Sanger sequenced. For the sequenced SPC110-ERdd escapees, seven unique escape mutations were found, mapping to the C-terminal part of the Spc110 protein or the linker to the ERdd tag (Fig. 5). All of them result in a premature stop, leading the absence of the ERdd tag. Seven of the eight mutants were missing a C-terminal portion of the protein. The C-terminal region of Spc110p is known to be involved in binding to calmodulin, but not essential for viability (21). Interestingly, no mutations within the ERdd tag itself were found, indicating its resistance against escape by truncation. Based on this finding, we generated an ERdd fusion omitting the non-essential C-terminal domain, SPC110Δ845-ERdd. This indeed increased containment stringency without compromising growth under permissive conditions, resulting in a strain with an estradiol dependence that was similarly favorable as that of SPC110-ERdd; growth of SPC110Δ845-ERdd likewise was fully restored at 100 nM estradiol (Fig. 5C). The strain’s escape frequency was below the detection limit of the standard escape assay, not yielding escape colonies. In a low escape frequency assay an escape rate of 4.3×10− 9 was determined. All of the sampled SPC110Δ845-ERdd escape mutations mapped to the linker or ERdd tag, demonstrating that escape routes through truncating mutations in SPC110 itself have been closed off by removal of its C-terminal region.
The sampled RRP46-ERdd, DIS3-ERdd, and doubly tagged RRP46-ERdd/DIS3-ERdd escapees had no escape mutations in the respective fusion genes. To elucidate their escape mechanism, the whole genome of three clones of each were sequenced by Illumina sequencing (Supplementary Data S3). For each of the six escapees with an ERdd-tagged DIS3 gene, a mutation likely to reduce proteasome activity could be found. Three unique mutations (two non-sense, one frameshift) of RPN4, a transcription factor stimulating expression of proteasome genes were sampled. Further, a non-sense mutation of a chaperone for proteasome maturation (UMP1), and non-synonymous mutations of two proteins involved in proteasome regulation, RPN3 and RPT5, were found. Likewise, for one of the RRP46-ERdd escapees a non-synonymous mutation of a proteasome subunit was the likely cause for containment escape. For the other two RRP46-ERdd escapees no obvious suppressor mutations could be identified.
Competitive fitness of contained strains
The perfect containment system imparts no fitness defect under permissive conditions as to not provide an evolutionary incentive to inactivate the containment system under prolonged culture. The strains contained by single or dual ERdd fusion to SPC110, DIS3, or RRP46 and respective combinations showed growth that was indistinguishable from that of the parental BY4742 with 1 µM estradiol in growth assays. However, a much more stringent measure for fitness is the competition against the parental strain under repeated, lab-typical batch cultures, cycling through different physiological states. Competition cultures were carried out over ten such cultures cycles for a total of 100 generations in YPD with 1 µM estradiol (Fig. 6A), assessing the frequency of the parental and safeguarded strain at 0, 50 and 100 generations from 46 to 48 samples each. None of the above-mentioned single and dual ERdd strains appeared to be outcompeted by the WT parental strain. This apparent lack of evolutionary incentive for containment inactivation should result in a high stability of the containment system in prolonged culture under permissive conditions. This was directly assessed by cultivation of eight independent SPC110-ERdd cultures over 100 generations under permissive conditions followed by assaying the escape frequencies. The measured average escape frequency after 100 generations was 2.2×10− 8, not having increased following prolonged culture.
Proteome analysis of contained strains
To investigate effects of the introduction of ERdd tags on cell physiology under permissive conditions, cells grown with 1 µM estradiol were harvested in mid-log phase to analyze their proteomes by LC-MS. Normalized protein abundances of the safeguarded strains were compared to those of the parental strain. Between 2939 and 2943 protein abundances could be compared for each of the strains. For SPC110-ERdd no significant perturbances of the proteome were detected, whereas SPC110Δ845-ERdd (1 up-, 2 down-regulated), RRP46-ERdd (5 upregulated) and DIS3-ERdd (1 upregulated) showed apparent detectable changes of the proteome. Interestingly, in both RRP46-ERdd and DIS3-ERdd cytochrome b2 appeared to be upregulated. Both RRP46 and DIS3 are components of the RNA exosome complex, but the relationship to cytochrome b2, a mitochondrial protein involved in lactate utilization, is unclear.
[1] https://rdrr.io/bioc/SLGI/man/essglist.html