Large-scale robot-assisted genome shuffling yields industrial Saccharomyces cerevisiae yeasts with increased ethanol tolerance

Background During the final phases of bioethanol fermentation, yeast cells face high ethanol concentrations. This stress results in slower or arrested fermentations and limits ethanol production. Novel Saccharomyces cerevisiae strains with superior ethanol tolerance may therefore allow increased yield and efficiency. Genome shuffling has emerged as a powerful approach to rapidly enhance complex traits including ethanol tolerance, yet previous efforts have mostly relied on a mutagenized pool of a single strain, which can potentially limit the effectiveness. Here, we explore novel robot-assisted strategies that allow to shuffle the genomes of multiple parental yeasts on an unprecedented scale. Results Screening of 318 different yeasts for ethanol accumulation, sporulation efficiency, and genetic relatedness yielded eight heterothallic strains that served as parents for genome shuffling. In a first approach, the parental strains were subjected to multiple consecutive rounds of random genome shuffling with different selection methods, yielding several hybrids that showed increased ethanol tolerance. Interestingly, on average, hybrids from the first generation (F1) showed higher ethanol production than hybrids from the third generation (F3). In a second approach, we applied several successive rounds of robot-assisted targeted genome shuffling, yielding more than 3,000 targeted crosses. Hybrids selected for ethanol tolerance showed increased ethanol tolerance and production as compared to unselected hybrids, and F1 hybrids were on average superior to F3 hybrids. In total, 135 individual F1 and F3 hybrids were tested in small-scale very high gravity fermentations. Eight hybrids demonstrated superior fermentation performance over the commercial biofuel strain Ethanol Red, showing a 2 to 7% increase in maximal ethanol accumulation. In an 8-l pilot-scale test, the best-performing hybrid fermented medium containing 32% (w/v) glucose to dryness, yielding 18.7% (v/v) ethanol with a productivity of 0.90 g ethanol/l/h and a yield of 0.45 g ethanol/g glucose. Conclusions We report the use of several different large-scale genome shuffling strategies to obtain novel hybrids with increased ethanol tolerance and fermentation capacity. Several of the novel hybrids show best-parent heterosis and outperform the commonly used bioethanol strain Ethanol Red, making them interesting candidate strains for industrial production. Electronic supplementary material The online version of this article (doi:10.1186/s13068-015-0216-0) contains supplementary material, which is available to authorized users.

: Saccharomyces yeasts vary in sporulation efficiency and spore viability. (A) The sporulation of each strain was induced on minimal sporulation medium and the sporulation efficiency, expressed as the percentage of tetrads, was determined by light microscopy. (B) After dissecting four full tetrads, the spore viability was expressed as the fraction of spores capable of forming a colony on solid medium. Strains were divided into five bins based on their spore viability. (C) The mating-type of germinated spores was determined, and each strain was classified as yielding haploid spores, diploid and haploid spores or diploid spores. S1: The ethanol production of industrial strains in static and stirred VHG fermentations. Thirteen strains, the eight parental strains as well as five additional high-ethanol producing strains, were tested for their fermentation capacity under both static and stirred conditions in 150 ml YP+35% (w/v) glucose. Under stirred conditions, each strain was tested in biological triplicates, except for P7 (Ethanol Red) which was tested in each fermentation batch (10 batches in total), and strain P4 which was tested twice in triplicates. Static fermentations were carried out once.     S4: Fermentation performance of F3 populations in VHG medium after re-inoculation. F3 populations obtained after random genome shuffling with different selection methods (growth, survival, no selection) were inoculated into YP+35% (w/v) glucose and re-inoculated two times (left; first reinoculation, right; second re-inoculation). The cumulative weight loss is a measure of CO2 production during the fermentation. Each line represents the average of two replicates (Ethanol Red), six replicates for growth-selected and survival-selected populations or twelve replicates for unselected populations. Error bars represent standard deviations. Then, the digested tetrad suspensions were divided over 96-well plats to generate masterplates that were subsequently used in a robot-assisted set-up to allow for mating. In this figure two examples of these masterplates are shown. The parental strains were crossed in all pairwise combinations by placing digested tetrad suspensions in close proximity on YPD agar using the robot. After mating, outcrossed hybrids, possessing resistance against both antibiotics, were recovered by replica-plating on YPD+hyg/kan. These hybrid pools were subcultured in liquid medium and subjected to a robotassisted ethanol tolerance screen on solid YPD supplemented with different ethanol levels as described in the Methods. After quantification, the best-performing populations were plated for single colonies on YPD supplemented with ethanol, and the fastest-growing colonies were re-tested for ethanol tolerance; the best-performing isolates were then used for the next round of shuffling, according to the same procedures (transformation of plasmids, sporulation, crossing). The crossing schemes are outlined in Figure 4. Hybrids for unselected genome shuffling schemes were isolated directly after the mating step, without carrying out any test for ethanol tolerance (not shown).  Overview of the contribution of each parental genome to targeted F1 hybrids selected for VHG fermentations. The percentage of the 15 F1 hybrids possessing a certain background is indicated, the "equal" bar indicates the percentage that each strain would contribute in case parental strains would contribute equally to these F1 hybrids.

Supplemental Text
This supplemental text describes details of the different genome shuffling strategies. More information can be found in the main text.

Genome shuffling with random mating to generate pool of F1 hybrids
The mass mating procedure was started with 6.25x10 6 spores of each strain (i.e. a total of 5x10 7 spores). These spores were mixed in rich medium and allowed to germinate, mate and proliferate for 16h. After 16h, the cell density had increased to ~2x10 9 cells. Since we used diploid heterozygous parental strains, we assume each spore is genetically unique and, because the heterothallic nature of the strains avoided haplo-selfing of germinated spores, each hybrid formed is likely unique as well.
Assuming that all the spores first mated to form 2.5x10 7 diploids, and subsequently all the resulting diploids started proliferating, after the 16h incubation the population had undergone ~6.3 doublings; i.e. each unique diploid was present in ~80 copies at the end. The whole population was divided into aliquots which were frozen down at -80 °C; each aliquot harboring ~1.1*10 7 cells (0.58% of the entire population). Hence, we can expect that each F1 aliquot contains 0.0058x80=0.47 copies of each unique hybrid.

Random genome shuffling with growth selection
In order to select cells for the next round of shuffling during random mating strategy with selection for growth in the presence of ethanol, we plated dilution series on YPD plates supplemented with a range of ethanol concentrations. For the first round of selection, for each biological replicate, we started from a single F1 aliquot. We grew these cells in 50 ml YPD overnight, typically for n~7. After this overnight growth, typically 1.5-10% of the culture was subcultured in YPD+5%(v/v) ethanol and allowed to grow for 8h, and if we assume each initial hybrid was unique and grew with the same growth rate, we would obtain ~135 copies of each unique hybrid before plating. By plating serial dilutions, only a fraction of unique hybrids were present on a given plate. We always harvested a 10 -2 or 10 -3 dilution for the next round of shuffling; which implies that we subjected ~0.2 -0.02% of all unique hybrids to selection. After incubating the plates, cell growth was inspected, and we always took at least 150 unique hybrids (see Methods) to the next round of shuffling. After washing off and mixing the biomass, cells were frozen down, and the start for sporulation was carried out as described in Methods. After sporulation and mass mating similar population size was frozen down and the same assumptions were made as before.

Random genome shuffling with survival selection
In order to select cells for the next round of shuffling during random mating strategies with selection for survival of very high ethanol levels, we incubated cell populations for 16h in YPD supplemented with high ethanol concentrations. For the first round of selection, for each biological replicate, we started from five F1 aliquots. We pre-grew these cells in 50 ml YPD to an OD600~5 (~1.0*10 8 cells/ml), corresponding to n~7. If we assume at the start every unique hybrid was present twice, the number of each unique hybrid (assuming equal growth rates in YPD) would have increased to ~181. Then, per assay, 1.0*10 9 cells (~20% of the pre-culture) were harvested, i.e. ~36 copies of each unique hybrid were tested in each assay. Although there was 4% (w/v) glucose in this medium, the OD600 after 16h did not change significantly, indicating that at these high ethanol levels cell proliferation was inhibited. As a control, a portion of the culture was incubated in YPD without added ethanol, always undergoing 2-3 population doublings. After 16h incubation, we spun down the culture immediately, washed 1x with PBS, and resuspended the cells in a total volume of 1 ml PBS. 90% of this biomass was spread over nine petri plates containing SC+3%(v/v)+2%(w/v) agar, the remaining biomass was used for serial dilutions that were also plated. After incubating the plates, we always took at least 150 unique hybrids to the next round of shuffling (see Methods). After washing off and mixing the biomass, aliquots were created and frozen down, and the start for sporulation was carried out as described in Methods. After sporulation and mass mating the same assumptions were made as described above.