Wild yeast isolation by middle school students reveals features of North American oak populations of Saccharomyces cerevisiae and Kluyveromyces lactis

Abstract

Features of the natural life cycle of the budding yeast Saccharomyces cerevisiae were crucial to its domestication as a laboratory experimental model, especially the ability to maintain stable haploid clones and cross them at will to combine alleles via meiosis. Stable haploidy results from mutations in HO, which encodes an endonuclease required for haploid-specific mating-type switching. Previous studies found an unexpected diversity of HO alleles among natural isolates within a small geographic area. We developed a hands-on field and laboratory activity for middle school students in Denver, Colorado, USA to isolate wild yeast from oak bark, identify species via DNA sequencing, and sequence HO from S. cerevisiae isolates. We find limited HO diversity in North American oak isolates, pointing to efficient, continuous dispersal across the continent. By contrast, we isolated the “dairy yeast”, Kluyveromyces lactis, from a tree <10 m away and found that it represents a new population distinct from an oak population in an adjacent state, pointing to high genetic diversity. The outreach activity partnered middle school, high school, and university students in making scientific discoveries and can be adapted to other locations and natural yeast habitats. Indeed, a pilot sampling activity in southeast Texas yielded S. cerevisiae oak isolates with a new allele of HO and, from a nearby prickly pear cactus, a heat-tolerant isolate of Saccharomyces paradoxus.

INTRODUCTION

Despite an overall excess of workers seeking science, technology, engineering, and mathematics (STEM) careers, women and ethnic and racial minorities remain underrepresented in the academic sector of STEM fields in the United States (Fry et al., 2021). Middle school (grades 6–8 in the United States, typically ages 11–14) represents an important educational period in which interest in science as a career is either cultivated or lost (Tai et al., 2006). Recent research has demonstrated that informal, extracurricular experiential STEM-related activities can increase interest in future STEM careers among underrepresented populations (Maiorca et al., 2021). We sought to develop and test such an activity focused on biology and derived from our research experience with the budding yeast Saccharomyces cerevisiae.

Fermentation by S. cerevisiae to produce ethanol, carbon dioxide, and flavorful compounds has been exploited by humans for millennia. In the last century S. cerevisiae has also moved to the forefront of experimental molecular biology (Botstein and Fink, 2011), due in large part to efficient genetic manipulation coupled with rapid cell proliferation. The last decade has seen a surge in studies of the natural history of S. cerevisiae in the hopes of developing a better appreciation of how this experimental workhorse normally lives in the wild (Liti, 2015). Early investigations into the S. cerevisiae life cycle funded in part by beer industries in Denmark and the United States (Winge, 1935; Lindegren, 1945) revealed that while diploid yeast typically proliferate via budding (i.e., vegetatively), when starved for nitrogen most natural isolates are able to undergo meiosis accompanied by sporulation to produce haploid spores encased in specialized, stress-resistant cell walls. Without nutrients, a spore remains dormant and for weeks or longer retains the ability to germinate upon restoration of nutrients and either mate with an adjacent haploid cell or proliferate by budding (Brengues et al., 2002; Maire et al., 2020). Mating occurs only between haploid cells of compatible mating types, a with α. Most natural S. cerevisiae isolates are diploid (Peter et al., 2018), and in most cases a haploid spore of either haploid mating type is able to switch mating types (Fischer et al., 2021). Subsequent mating with a daughter cell of opposite mating type provides an efficient route for an isolated spore to return to diploidy.

A major advantage for experimental genetic analysis in S. cerevisiae is the ability to stably propagate haploid strains and mate them with other haploids under controlled conditions to produce diploid strains with desired genotypes. A key advance in this regard was the isolation of strains unable to switch mating types (“heterothallic”) (Lindegren and Lindegren, 1943) due to an apparent loss-of-function allele of the HO gene (Roberts and Winge, 1949). In strains capable of mating-type switching (“homothallic”), HO encodes an endonuclease that makes a single cut in the genome to initiate recombination-mediated exchange of alleles at the MAT locus (Strathern et al., 1982; Kostriken et al., 1983). Apparent loss-of-function ho alleles have been found frequently in other natural isolates (Katz Ezov et al., 2010; Fischer et al., 2021) and in strains used in wine-making (Mortimer, 2000). Since HO is not expressed in diploid cells (Jensen et al., 1983), any strain that is incapable of, or rarely undergoes, meiosis and/or sporulation would rarely or never express HO. Since HO is only expressed after a haploid cell buds once (Nasmyth, 1983), HO also remains repressed if spores mate immediately upon germination, which is common in some natural isolates (McClure et al., 2018). In either scenario the gene could accumulate mutations due to genetic drift in the absence of purifying selection.

Unexpectedly, however, the same alleles are found in isolates from geographically distant locations, inconsistent with drift in the absence of purifying selection. For example, the same ho allele originally found near Merced, California was also found in Mount Carmel, Israel, >10,000 km away (Katz Ezov et al., 2010). Ho evolved from a domesticated “selfish” genetic element, a homing endonuclease (Coughlan et al., 2020). These observations suggest that the common ancestors of isolates with shared ho alleles were widely dispersed across the planet, although we cannot exclude the possibility that some residual, ancestral Ho function apart from mating-type switching drove convergent evolution to arrive independently at the same alleles.

For “domesticated” strains associated with human activity, such dispersal could have been the work of humans, as previously suggested for the Merced strain (Katz Ezov et al., 2010), but many isolates with shared alleles come from primeval forests or other areas with little human activity (Wang et al., 2012) (see Figure 1). While the presence of primeval forest lineages in the bark of planted chestnut trees has been taken as evidence of human involvement in migration (Wang et al., 2012), the geographical distribution of ho alleles presents an unsolved puzzle. We note that in the age of recreational genetic ancestry testing, the general public is increasingly familiar with the idea of using gene sequences to infer relatedness between geographically distant individuals, yet many people fail to recognize practical applications, such as how genetic testing can prevent disease (Krakow et al., 2017). Thus the fundamental concepts underlying this puzzle are relatively straightforward to explain, and present an opportunity to increase genetic literacy with possible implications for future improvements in public health (Abrams et al., 2015; Little et al., 2022).

Figure 1. Geographic distribution of shared ho alleles among S. cerevisiae isolates from nature.

(A) Predicted structure of the Ho protein from the S288C genetic background (heterothallic) showing the locations of the regions used for the HO-sequence-based isolate grouping in (B): four single amino acid positions and one 36-residue region corresponding to residues 524–559 that is deleted in some isolates (“Δ36”). As delineated by Haber and Wolfe (Haber and Wolfe, 2005) the protein splicing domain is gray. The 36 residues deleted in the “Δ36” deletion are in blue. The structure is oriented so that the parts of the protein that bind DNA are at the bottom. (B) Location of origin for 41 strains of S. cerevisiae from the 1,011 genomes collection annotated in that study (Peter et al., 2018) as coming from “nature”. Strains were grouped according to HO gene sequences corresponding to the regions highlighted in (A). Key indicates genotypes by single-letter amino acid code for the five positions of interest in the order 189, 223, 405, 475 and, if present, the deletion starting at position 524. Numbers on location markers correspond to identifiers in Supplementary Table S1 (see workbook “for MapCustomizer”). Map positions are approximate and some location pins were slightly repositioned to reveal other pins otherwise obscured underneath or to show distant locations on this partial world map. Denver, the site of our isolation activities, is marked with a star.

Kluyveromyces lactis is a less familiar yeast species for which there are also interesting questions about the geographical distribution of important genes. Recent studies provided evidence that ancestral strains were unable to ferment lactose, and lactose utilization arose multiple times independently due to strong selection imposed by dairy farmers and the acquisition of a cluster of genes (encoding the galactosidase Lac4 and the lactose permease Lac12) from a closely related species, Kluyveromyces marxianus (Friedrich et al., 2023). K. lactis isolates not associated with human activity lack the gene cluster and utilize lactose less efficiently (Friedrich et al., 2023). Genotypic and phenotypic characterization of new K. lactis isolates from non-industrial sources may provide additional insights into the geographical distribution of genes associated with lactose utilization. K. lactis thus provides a simple illustration of the concept that acquisition of a handful of genes can transform a “wild” organism into an industrially valuable resource, an idea that is broadly accessible to people with basic science knowledge.

We sought to develop a hands-on science outreach activity for middle-school students that addresses the question of how geographically disparate S. cerevisiae strains share common ho alleles. To broaden the impact of this activity, we further sought to involve university students pursuing Early Education degrees in both carrying out the hands-on activity and developing and evaluating assessments. As a template for others to adapt for their own purposes, here we describe the hands-on activities and present the biological insights derived from our initial efforts with middle-school and university students in the Denver, Colorado Metropolitan region. Pedagogical goals and assessments will be described in detail elsewhere. In addition to basic yeast species identification and HO sequence determination for S. cerevisiae isolates, we report genotypic and phenotypic characterization of both a natural isolate of K. lactis isolated during the outreach activity, and of a heat-tolerant isolate of Saccharomyces paradoxus, providing additional research topics readily accessible for outreach activities at the middle-school level.

MATERIALS AND METHODS

Oak bark isolation and culture

Sani-cloth Bleach germicidal disposable wipes (PDI # P54072) were used to sterilize the gripping surfaces of six-inch slip-joint pliers (Stanley # 84–097) just prior to bark isolation. The bark fragments were dropped into sterile 50-mL conical tubes. For bark collection in Texas, fragments were dropped into zip lock bags. In the lab, either 10–25 mL of sterile YPD+pen/strep (2% peptone, 1% yeast extract, 2% glucose, 100 units/mL penicillin, 100 μg/mL streptomycin; for the 2018 pilot) or 25 mL of Sniegowski’s enrichment medium (Sniegowski et al., 2002) was added to each tube, to submerge the bark. The penicillin and streptomycin were added to the YPD from a 100X stock (HyClone). The resulting cultures were maintained at room temperature (2018 pilot) or 30°C (2019 activity with CCU participation) without constant agitation; they were agitated occasionally, during periodic inspection for turbidity. Three-μL aliquots were removed and placed between a glass slide and #1 thickness 18 mm × 18 mm coverslip for examination by light microscopy with objectives of various magnifications (40X, 60X, or 63X). Images in the figures here were taken with an EVOSfl all-in-one epifluorescence microscope (Thermo-Fisher Scientific) with a 60X oil objective.

For YPD+pen/strep cultures that contained yeast-form cells, ~500 μL were used to inoculate 25 mL of Sniegowski’s enrichment medium (per L: 3 g yeast extract, 3 g malt extract, 5 g peptone, 10 g sucrose, 76 mL ethanol, 400 μg chloramphenicol, 1 mL 1M hydrochloric acid (Sniegowski et al., 2002)) in new 50-mL conical tubes. These cultures were also maintained at room temperature without constant agitation and inspected with a microscope in the same way. For Sniegowski’s cultures with yeast-form cells, aliquots were either diluted in sterile water and spread on solid YPD medium (2% agar) using sterile glass beads, or 3 μL of undiluted culture was spotted on a YPD plate and then the cells were streaked across the rest of the surface of the plate using a sterile toothpick. These plates were incubated at 30°C until colonies were visible. Cells from these colonies were scraped with a pipette tip and resuspended in 3 μL of water on a microscope slide before applying a glass coverslip and imaging.

Colonies composed of cells resembling S. cerevisiae were used for sporulation tests by scraping cells from the colony with a sterile toothpick and spreading on a sporulation plate (1% potassium acetate, 2% agar) in an area ~1 cm². Following incubation at 30°C for 1–2 days, cells from the sporulation plate were scraped off with a pipette tip and imaged. For those that harbored asci resembling S. cerevisiae (see Figure 2B), more cells were scraped from the sporulation plate and resuspended in 50 μL of zymolyase (Zymo Research #E1004) diluted 1:50 in water to digest ascus walls. Following ~10 min of digestion, 3 μL was spread in a strip on the surface of a YPD plate, and a tetrad dissection microscope was used to isolate individual spores. The YPD plate was incubated at 30°C until colonies formed. To test for homothallism, cells from individual colonies were spread on another sporulation plate, and the subsequent appearance of asci was taken as evidence of the presence of diploid cells, indicative of mating-type switching.

Figure 2. Tree map and schematic illustration of the outreach activity.

(A) Map of Cheesman Park with target trees from 2018 pilot activity labeled and annotated with species of yeast isolated, if any. (B) Circled numbers indicate the following steps in the wild yeast isolation activity. (1) Oak bark samples are incubated in liquid rich medium (YPD) with anti-bacterial antibiotics (penicillin and streptomycin, “pen/strep”). (2) After microscopic observation of YPD+pen/strep cultures, aliquots of those cultures that contained yeast-form cells were used to inoculate a more selective enrichment medium containing a different anti-bacterial antibiotic (Sniegowski’s with chloramphenicol). The micrograph at left shows an example image from a YPD+pen/strep culture from tree 26227, from which S. cerevisiae was ultimately isolated. (3) Microscopic examination of the selective cultures was used to identify those containing yeast-form cells, aliquots of which were plated to YPD agar plates to allow colonies to form. (4) Cells from individual colonies were examined with a microscope to identify those with yeast-form cells, portions of which were transferred to solid sporulation-inducing medium (1% potassium acetate). (5) Strains that generated asci resembling S. cerevisiae (the micrograph shows a representative isolate from tree 26227) were selected for inoculation from the YPD plate into liquid YPD medium. (6) Extraction of genomic DNA was followed by (7) PCR amplification of ITS2 sequences and subsequent Sanger sequencing. (8) Isolates identified as S. cerevisiae by BLAST analysis were further analyzed by PCR and (9) sequencing of the HO gene. The chromatograph shows actual HO sequence data from an S. cerevisiae isolate from tree 26204. (10) S. cerevisiae isolates were tested for homothallism by dissecting tetrad asci and allowing spores to germinate and form colonies on YPD. Cells from the resulting colonies were transferred to sporulation-inducing medium and examined for the presence of asci, indicative of mating-type switching after spore germination and mating to return to diploidy.

Laboratory and reference yeast strains

For PFGE, we used S. cerevisiae strain FY2742 (S288C derivative MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0 RME1 TAO3 MKT1, where the alleles at RME1, TAO3 and MKT1 are from the SK1 strain background (Kloimwieder and Winston, 2011) and prototrophic wild-type diploid S. paradoxus strain N17 (Naumov et al., 1990). For lactose utilization, we used an S. cerevisiae isolate from tree 26204 (Table 1) that we named MMY0374 and K. lactis NRRL Y-1140 ATCC #8585 (Dujon et al., 2004; Zivanovic et al., 2005).

Table 1.

Oak trees sampled and yeast species isolated in the fall 2018 outreach activity.

Oak (species, number, GPS coordinates)	YPD+pen/strep yeast	Sniegowski’s yeast	Yeast-like colony	Sporulation	ITS species
Quercus alba, 26033, 39.732216, −104.968465	+	+	−	ND	ND
Quercus robur, 26203, 39.731964, −104.965107	ND	+	−	ND	ND
Quercus robur, 26204, 39.731964, −104.965107	+, +	+, +	+, +	+, ?	S. cerevisiae, Torulaspora delbrueckii
Quercus robur, 26205, 39.731992, −104.965094	ND, ND	+, +	−, +	ND, −	Pichia kudriavzevii
Quercus robur, 26206, 39.732035, −104.965094	+, ND	+, +	+, −	+, ND	ND, ND
Quercus robur, 26224, 39.732450, −104.965087	ND, +	+, +	+, +	+, +	ND, ND
Quercus robur, 26225, 39.732450, −104.965087	+, +	+, +	+, +	+, +	K. lactis, ND
Quercus robur, 26226, 39.732450, −104.965087	ND, ND	+, +	−, +	ND, +	ND, ND
Quercus robur, 26227, 32.732450, −104.965087	+, ND	+, +	+ (x3), +	+, +	S. cerevisiae (x2), unknown
Quercus alba, 26252, 39.732924, −104.964659	ND	+	+	?	Torulaspora delbrueckii
Quercus bicolor, 26321, 39.736155, −104.967111	ND	+	−	ND	ND
Quercus alba, 26561, 39.731806, −104.968078	ND	+	+	?	Torulaspora delbrueckii
Quercus rubra, 26775, 39.734859, −104.965500	ND	+	+ (x2)	?, ND	Debaryomyces hansenii, Candida oleophila

ND, not determined.

+, cells or colony with yeast-like appearance were detected

−, no cell or colony with yeast-like appearance was detected

?, ambiguous/unclear results

(x2), two distinct colonies or isolates were found.

DNA isolation, PCR, and single-gene sequencing

Genomic DNA was isolated from cells pelleted from 5-mL overnight YPD liquid cultures or scraped from YPD plates using a modified E. coli plasmid DNA isolation protocol/kit (Johnson et al., 2015). Standard Taq (New England Biolabs #M0273) or Phusion (New England Biolabs #M0530) or Q5 (New England Biolabs #M0491) DNA polymerase was used for PCR according to the manufacturer’s instructions, with 1 μL of genomic DNA as template. PCR for species identification (White, 1990) used primers ITS1 (TCCGTAGGTGAACCTGCGG), ITS4 (TCCTCCGCTTATTGATATGC), NL1 (GCATATCAATAAGCGGAGGAAAAG), and/or NL4 (TCCTCCGCTTATTGATATGC). PCR for HO sequencing used primers 5’ScereHOfw (TCCATATCCTCATAAGCAGC), ScHOmidre (CCTGCGAGTACTGGACCAAA), S.cereHOmidfw (CAGCTACTGTGACGACCAGG), HOmidfw2 (GGTCTGTGGTTAGGTGACAG), HOmidre2 (TCTCACCCTGGAAATCATCC), HOendrev (TACCACAACTCTTATGAGGC), and/or 3’HOre (TGGCGTATTTCTACTCCAGC). After inactivation/digestion of primers and nucleotides via treatment with Exonuclease I (Thermo Scientific Fermentas #EN0581) and alkaline phosphatase (Thermo Scientific Fermentas #EN0581), the products were sequenced via Sanger sequencing with the same primers. Contigs were assembled using the CAP3 Sequence Assembly Program (https://doua.prabi.fr/software/cap3) (Huang and Madan, 1999). For one amplicon from genomic DNA of MMY0358 using the primers 5’ScereHOfw and HOmidre2, the PCR product was purified using the DNA Clean & Concentrator-5 kit (Zymo Research #D4013) and direct long-read sequencing was performed by Quintara Biosciences, Inc.

Maximum temperature estimation

We viewed daily temperature data for Denver from the months of June, July, and August of 2012–2022 at https://www.weather.gov/bou/local_climate and averaged the values for the highest daily temperature recorded for each month. No data were available for July 2015. For College Station, Texas, we used online weather data at https://www.weather.gov/wrh/Climate?wfo=hgx and browsed the monthly highest max temperature from 2012–2019 at Easterwood Field, College Station.

Phylogenetic analysis

The K. lactis ACT1 phylogenetic tree was based on nucleotides 819–1904 where position 1 is the start codon. S. cerevisiae ACT1 sequences were aligned using the Simple Phylogeny tool at the EMBL-EBI server (www.ebi.ac.uk/Tools/phylogeny/simple_phylogeny/) from a Clustal Omega alignment performed at the same server. Input Parameters were: Tree Format: phylip, No distance correction, gaps were not excluded, and a Neighbour-joining Clustering Method was used.

Pulsed-field gel electrophoresis

Yeast strains were grown in 7 mL liquid YPD cultures at 30°C for 24 hours. Cells were pelleted and resuspended in low melting point agarose to form plugs. Plugs were treated with zymolyase to remove cell walls and proteinase K to degrade protein components. Plugs were loaded into a 1% gel made in tris boric acid EDTA buffer (45 mM Tris-borate 1 mM EDTA). The gel was then run in the same buffer in a BioRad CHEF DRII Mapper for 60 hours with initial switch and final switch times 50s and 150s, respectively. To visualize size-resolved chromosomes, the gel was stained with ethidium bromide.

Lactose utilization assay

5-mL YPD liquid cultures were inoculated with colonies of Y0329, Y0374, or NRRLY-1140 from YPD plates and rotated at 30°C overnight. 1 mL of each culture was pelleted in a microcentrifuge tube by a pulse of centrifugation at maximum speed. The medium was removed and the cells were resuspended in 1 mL sterile distilled deionized water, then pelleted again. After resuspending the cells again in 1 mL water, 100 μL of each was mixed with 15 mL of lactose (per L: 6.7 g yeast nitrogen base with amino acids and ammonium sulfate (Research Products International # Y20030, 100 g lactose) or glucose (per L: 6.7 g yeast nitrogen base with amino acids and ammonium sulfate, 20 g glucose) medium in a petri dish, then 150 μL of this suspension was transferred with a multichannel pipette to each of 16 wells of a flat-bottom 96-well plate. The plate was incubated for 22 hr 9 min at 30°C in a BioTek Cytation 3 plate reader and the absorbance at 595 nm was read every 5 minutes following orbital shaking. Eight wells away from the plate edges were used for analysis.

Comparing culture growth rates at different temperatures

Yeast cells grown on a solid YPD plate were resuspended in liquid YPD in sterile glass culture tubes, from which 100 μL was transferred to 8 wells of a 96-well plate of PCR tubes that was cut in half. The plate was incubated in a thermal cycler using a program with gradient of temperatures over 8 wells from 30°C to 45°C. The lid temp was set to 50°C. After 21 hr incubation, 50 μL of each culture was transferred to the wells of a 96-well plate, after first reading the absorbance at 595 nm of the empty wells using a Cytation 3 plate reader (Bio-Tek). The absorbance of each culture was measured and the values of the empty wells were subtracted.

Yeast whole genome sequencing

DNA extraction and sequencing was performed by The Sequencing Center, Fort Collins, Colorado. Zymo Research Quick-DNA Fungal/Bacterial Microprep Kit (#D6007) was used to extract DNA from K. lactis strain MMY0329. Cells (10–20 mg wet weight) were resuspended in 200 μL phosphate buffered saline (PBS) and added to a ZR BashingBead^™ Lysis Tube (0.1 & 0.5 mm) with 750 μL Lysis Solution. The tube was processed in a Qiagen TissueLyser LT bead beater at 50 Hz for 5 min and centrifuged at 10,000 × g for 1 min. 400 μL supernatant was transferred to a Zymo-Spin^™ IV Spin Filter in a Collection Tube and centrifuged at 7,000 × g for 1 min. ß-mercaptoethanol was added to Genomic Lysis Buffer to a final concentration of 0.5% (v/v). 1.2 mL Genomic Lysis Buffer was added to the filtrate. 800 μL of the mixture was transferred to a Zymo-Spin^™ IC Column in a Collection Tube and centrifuged at 10,000 × g for 1 min. The flow-through was discarded and the previous step was repeated once. 200 uL DNA Pre-Wash Buffer was added to the Zymo-Spin^™ IC Column and centrifuged at 10,000 × g for 1 min. 500 μL g-DNA Wash Buffer was added to the Zymo-Spin^™ IC Column and centrifuged at 10,000 × g for 1 min. The Zymo-Spin^™ IC Column was transferred to a 2-mL microcentrifuge tube and 20 μL of DNA Elution Buffer was added directly to the column matrix. After 3 min. incubation, the column was centrifuged at 10,000 × g for 30 seconds to elute ultra-pure DNA.

The Illumina Nextera XT DNA Library Prep Kit was used to prepare extracted yeast DNA for sequencing. For tagmentation, 10 μL of Tagment DNA Buffer and 5 μL of extracted gDNA were added to each plate well in a PCR plate and mixed by pipetting. 5 μL of Amplicon Tagment Mix was added to each plate well, mixed by pipetting and the plate was centrifuged at 280 × g for 1 min. A predefined tagmentation program was run on an ABI GeneAmp PCR System 9700 thermal cycler. Immediately following its completion, 5 μL of Neutralize Tagment Buffer was added to each plate well and mixed by pipetting. The plate was centrifuged at 280 × g for 1 min followed by incubation at room temperature for 5 min. For amplification and indexing, Illumina Index 1 (i7) adapters and Index 2 (i5) adapters were uniquely assigned to each sample on an index adapter plate. 15 μL of NexteraPCR Master Mix (NPM) was added to each plate well and mixed by pipetting. The Index/NPM mixture was transferred to the PCR plate, mixed by pipetting, and centrifuged at 280 × g for 1 min prior to PCR. Library cleanup was performed using AMPure XP beads (Beckman). The PCR product was centrifuged at 280 × g for 1 min and 50 μL of product was transferred to a new PCR plate. 30 μL of AMPure XP beads was added to each plate well and the plate was shaken at 1800 rpm for 2 min. The plate was incubated at room temperature for 5 min and placed on a magnetic stand until the liquid was clear. All supernatant was removed and discarded and the beads were washed twice with 200 μL 80% ethanol. The plate was air-dried on a magnetic stand for 15 min. 52.5 μL of RSB was added to the beads and the plate was shaken at 1800 rpm for 2 min. The plate was incubated at room temperature for 2 min and placed on a magnetic stand until the liquid was clear. 50 μL of supernatant was transferred to a new PCR plate. To check library size distribution, undiluted library was analyzed on an Agilent Technologies 2200 TapeStation using Agilent Technologies High Sensitivity D1000 ScreenTape.

44 μL of Library Normalization Additives 1 (LNA1) per sample and 8 uL of Library Normalization Beads 1 (LNB1) per sample were added to a 15-mL conical tube. 45 μL of the LNA1/LNB1 mixture was added to each sample PCR plate well. The plate was shaken at 1800 rpm for 30 min and placed on a magnetic stand until the liquid was clear. The supernatant was removed and discarded. The beads were washed twice with 45 μL of Library Normalization Wash 1 and 30 μL of 0.1 N sodium hydroxide was added to each plate well. The plate was shaken at 1800 rpm for 5 min. After a 5-min incubation, the samples were resuspended and shaken at 1800 rpm for 5 min. The plate was placed on a magnetic stand until the liquid was clear. The supernatant was transferred to a new PCR plate and the plate was centrifuged at 1000 × g for 1 min. 5 μL of each library was transferred from the PCR plate to a 2-mL microcentrifuge tube. A fresh dilution of 0.1 N sodium hydroxide was prepared by combining 900 μL of laboratory-grade water with 100 μL of stock 1.0 N sodium hydroxide. The denaturation process involved combining 5 μL of 1 nM library with 5 μL of 0.1 N sodium hydroxide, vortexing and incubating at room temperature for 5 min. 5 μL of 200 mM Tris-HCl, pH 7.0, was added to the denatured library, vortexed and centrifuged at 280 × g for 1 min. The denatured library was diluted with 985 μL Hybridization Buffer, vortexed and centrifuged at 280 × g for 1 min. The resulting diluted library was transferred to a 2-mL microcentrifuge tube and further diluted with 320 μL Hybridization Buffer to achieve a final loading concentration of 1.8 pM. A 5% PhiX spike-in control was added to the final library. Yeast whole genome sequencing was performed on an Illumina MiniSeq short-read sequencer using a standard Illumina workflow and configured for 2 × 150 bp paired-end reads and MiniSeq flow cell.

RESULTS

Disparate origins of S. cerevisiae isolates with shared alleles at HO

Others previously noted the diversity of alleles of HO among the genome sequences of 1,011 S. cerevisiae isolates and the fact that many are predicted to be non-functional for mating-type switching (Peter et al., 2018; Fischer et al., 2021). We examined the geographical origin of those strains with respect to the HO allele of each. We focused on five locations in the HO coding sequence, four single residues (189, 223, 405 and 475) where substitutions were found in a heterothallic fig isolate compared to the “wild-type” sequence from homothallic strains (Meiron et al., 1995), and one 36-residue region (residues 524–559) previously documented as being absent in a heterothallic isolate from Brazil (Argueso et al., 2009). The residue at position 189 is not conserved in related yeast species and lies in a linker between the protein splicing and endonuclease domains of Ho (Figure 1A). Ala, Thr or occasionally Ser is encoded here, with no apparent effect on Ho function. Residue 223 lies in one of two “endonuclease motifs” conserved between Ho and PI-SceI, a closely-related yeast endonuclease (Bakhrat et al., 2004), and is Gly in PI-SceI and all known homothallic strains as well as in strains from primeval Chinese forests (e.g. strain BAM) where S. cerevisiae is thought to have originally evolved (Wang et al., 2012; Peter et al., 2018). Ser in this position is sufficient to render an otherwise active allele inactive (Ekino et al., 1999). Residue 405 lies in the endonuclease domain but outside conserved regions (Haber and Wolfe, 2005). Leu is found here in many homothallic strains but alleles encoding both Ser in this position and Leu475 show some activity (Ekino et al., 1999). Residue 475 lies in one of two zinc finger domains predicted to directly contact DNA, and the H475L substitution was originally proposed to disrupt Ho function (Meiron et al., 1995) but a subsequent study found some mating-type switching activity of an allele encoding Leu475 (and Ala189 Gly223 and Ser405) (Ekino et al., 1999). The deleted region is interesting in two ways. First, the deletion rather precisely removes the other zinc finger, hence the deletion allele might bind a different DNA sequence, or none at all. Second, it is flanked by repeats of an 8-nucleotide sequence that likely act to promote homologous recombination that excises the intervening sequences (Argueso et al., 2009). Thus the deletion would be expected to occur spontaneously at a higher rate than a point mutation.

To look for shared alleles, we downloaded the HO coding sequences from the 1,011 genomes study (http://1002genomes.u-strasbg.fr/) and used Microsoft Excel to search within each for short (7–16-nucleotide) sequences surrounding each of the positions introduced above, then grouped isolates that shared the same sequences for all five positions. Since in the 1,011 genomes mixed sequencing results at a given position are indicated with the standard single-letter codes for multiple nucleotides (e.g., R for A or G), we detected heterozygosity in many cases and some groups shared heterozygosity at specific positions. We ignored the 124 isolates for which HO was deliberately deleted in the lab, and focused on the 41 isolates annotated as being of “natural” origin.

As shown in Figure 1B and Supplementary Table S1 for seven HO groups, isolates from the same group originated in geographically disparate locations. A phylogenetic tree made from the HO sequences recapitulated most of the major relationships established in the 1,011 genomes study using thousands of SNPs across the genome (Figure S1). Satisfyingly, the group heterozygous at residues 189 (Ala/Thr), 223 (Asp/Gly) and 475 (Leu/His) clusters with the 1,011 genomes “mixed origin” clade near the group with Ala189, Asp223, and Leu475 and the group with Thr189, Gly223, and His475 (Figure S1). Mixing between these groups could explain the origins of the two alleles in the heterozygous strains. Both strains with the 36-residue deletion (ALC and ALH, isolated from distinct locations in Ecuador) also encode a single-nucleotide deletion at codon 138 resulting in a predicted truncation after 148 residues (Figure S1B). The same frameshift is also found in three strains from Mount Carmel (BDC, BDF, and BDG) that lack the 36-residue deletion but otherwise differ by only three nucleotides from the Ecuadorian strains (Supplementary Table S1). If the frameshift eliminates Ho function, it is easy to imagine how the 36-residue deletion could subsequently arise and persist by avoiding purifying selection. How isolates from Ecuador and Israel so closely resemble each other is more mysterious.

A subsequent study of the 1,011 S. cerevisiae isolates assessed sporulation ability (De Chiara et al., 2022). There was no obvious correlation among natural isolates between any specific allele of HO and the ability to sporulate efficiently (Supplementary Table S1, see workbook “Natural isolates”). These observations do not support our speculation (see Introduction) that a failure to sporulate in the wild is the source of allelic variation in HO because it precludes the ability of purifying selection to maintain specific (functional) HO alleles.

Engaging middle-school students in isolation of wild yeast from oak bark

We sought out middle school students near our institutions to develop an activity engaging those students in the search for natural S. cerevisiae isolates and HO alleles. Morey Middle School in Denver, Colorado is an urban school serving sixth- to eighth-graders in the Denver Public School district. We contacted the sixth grade Science teacher at Morey at the beginning of the 2018/2019 and 2019/2020 school years to determine when and how our activity might best fit with her planned curriculum, and further designed our activity to meet Colorado middle school academic standards and learning expectations for Life Science (Supplementary Table S2). A ~40-minute didactic lecture presentation in the classroom, accompanied by image- and video-rich slides, introduced the specific topics. Altogether, the didactic material and the hands-on activity (see below) match six of twelve Colorado state Life Science learning standards (Supplementary Table S2). We chose dates for the lectures and hands-on activity in between the Science classes’ units on Microbiome, Metabolism and Traits and Heredity.

To develop a hands-on activity, we adapted existing protocols developed by others for yeast isolation from the bark of oak trees (Sniegowski et al., 2002). Though Saccharomyces yeasts are likely minor components (Kowallik et al., 2015), oak bark is a known habitat for many yeast species, including S. cerevisiae (Naumov et al., 1998). Cheesman Park is within walking distance (1 km) of Morey Middle School and is maintained by the City of Denver, which also maintains an interactive online inventory of tree species (TreeKeeper, denverco.treekeepersoftware.com). TreeKeeper and Google Maps (http://www.google.com/maps/) were used to prepare a map of Cheesman Park in which all target oak trees were identified by numbers (from TreeKeeper) and assigned global positioning satellite (GPS) locations using Google Maps. Trees were grouped by threes and assigned to groups of students.

In the 2018 pilot activity, an adult led groups of four to six students to oak trees of a variety of species, where they collected ~2-cm pieces of bark using kits of nitrile gloves, 50-mL conical tubes pre-filled with 10 mL liquid yeast peptone dextrose medium supplemented with the antibacterial drugs penicillin and streptomycin (“YPD+pen/strep”), pliers, sterilizing wipes, and a laminated tree map (Figure 2A). Bark (one piece per culture) was chosen from parts of the trees on or near their trunks, 0.5 to 1 m from the soil, as a compromise between ease of accessibility and distance from suspected sources of “contamination” (humans and domesticated animals). Bark cultures (two per tree, 15 trees) were incubated in the classroom for several days and then transferred to the McMurray lab. Samples from each culture were examined by transmitted light microscopy. Yeast-form (ovoid) cells were present in most YPD+pen/strep cultures, but non-yeast cells were predominant (Figure 2B). At this point, the students were asked to complete during class an online multiple-choice quiz (prepared in Google Forms; Figure 2B) distinguishing yeast-form cells from others in micrographs from selected cultures.

To enrich for yeast, aliquots of cultures with any visible growth (turbidity) were transferred to Sniegowski’s medium, which has a high ethanol content (Sniegowski et al., 2002). 21 Sniegowski’s cultures contained yeast-form cells (Table 1). Dilutions of each were plated to solid YPD medium, from which single colonies were examined for yeast-form cells. 15 resulting yeast strains were examined microscopically after transfer to sporulation medium; six formed asci resembling Saccharomyces (Table 1; see Figure 2). Species identification was performed using rDNA internal transcribed spacer (ITS2) sequences amplified by PCR from genomic DNA. Sequencing identified three S. cerevisiae isolates; seven other strains were a variety of non-Saccharomyces yeasts, including Kluyveromyces lactis (Table 1). Sequence quality from one strain was too poor for unambiguous species identification, and four were not sequenced (Table 1).

Notably, we did not isolate Saccharomyces paradoxus, a species closely related to S. cerevisiae and often found together in the same oak forests or even on the same tree (Sniegowski et al., 2002; Sampaio and Gonçalves, 2008; Liti et al., 2009; Zhang et al., 2010). The failure to isolate Saccharomyces paradoxus from Denver oaks was not unexpected, since prior research suggested that the geographical distribution of S. cerevisiae versus S. paradoxus on oak trees can be predicted by the maximum summer temperatures (Robinson et al., 2016). Using available local climate data we estimated the average maximum summer temperature for Denver from 2012–2022 as ~36.75°C, within the maximum temperature range considered optimum for wild S. cerevisiae (25–38°C) but too hot for S. paradoxus (18–31°C), at least in the oak habitat. These data also agree with recent S. paradoxus isolation results from China (He et al., 2022).

Homothallism was tested in three of the S. cerevisiae isolates by inducing sporulation, isolating individual spores on rich medium, and testing cells within the resulting colonies for the ability to sporulate, as an indicator of the presence of diploid cells. All three isolates sporulated with very high efficiency (see Figure 2). HO coding sequences were amplified by PCR and sequenced. All three sequences are of the group with Ala189 Gly223 Leu405 Leu475 and no deletion and are identical to the sequence found in homothallic strains isolated from oak trees in Missouri (“T7” strain) or Pennsylvania, USA (YPS163 strain) (Sniegowski et al., 2002; Fay and Benavides, 2005). We note that this HO allele must be functional despite encoding Leu405 and Leu475, since YPS163 (Fay and Benavides, 2005) and all our isolates are homothallic.

At this point, about a week after the bark gathering, a final classroom lecture introduced the students to the concept of species identification by sequencing a region of the genome that is sufficiently conserved between species to allow comparison but is sufficiently variable to distinguish between closely related species. The lecture included instructions on how to use the NCBI BLAST server to find the species with the best match. Then, each student was sent a link to access an anonymized ITS2 sequence from one of the oak bark isolates and asked to perform a BLAST alignment and enter in an online form their best guess as to the species identity. The students were also informed about the homothallism and HO sequencing results for the S. cerevisiae isolates. This activity concluded the pilot outreach.

The activity was repeated, with some changes, with a new class of middle school students in the 2019–2020 school year. Most important was the inclusion of first-year university students from Colorado Christian University (CCU). Nine Biology majors and 6 Early Education majors were present at Morey for the bark isolation, where they led groups of middle-school students in the activities. Since the Sniegowski’s medium was effective at enriching for yeast, the YPD+pen/strep step was skipped, and bark samples were cultured directly in Sniegowski’s enrichment medium. The cultures were monitored by CCU students in the Woodman lab, and they also tested for sporulation. Additionally, simple light microscopes owned by Morey Middle School were used with microscope slides and coverslips in a hands-on activity in which the students examined the bark cultures in the classroom. A digital eyepiece camera (Levenhuk M300 BASE) connected to a laptop computer was used to project live video from one microscope to a projection screen in the classroom, providing real-time instruction for the students to follow. Finally, the students completed custom assessments to gauge effectiveness of the activity. These assessments and related educational activities will be described elsewhere.

Geographical distribution of HO alleles shared with Denver oak isolates

One new S. cerevisiae isolate was obtained, found to be homothallic, and the HO sequence was identical to the three from the other Cheesman Park oak isolates. We examined the origins of all 20 isolates from the 1,011 genomes study that share this HO sequence. This group includes 12 of the 13 S. cerevisiae isolates in the “North American Oak” clade, and four other isolates from North American oaks that were not placed in that clade. One of the isolates is from the “Mosaic region 3” clade, while the remaining four are from human food (from India or Japan) or a North American insect (Supplementary Table S1). Notably, the other isolate from the 1,011 genomes “North American Oak” clade, strain ACD, has an HO sequence that differs at six nucleotides and encodes an Ho protein of the “Ala189 Ser223 Ser405 Leu475 and no deletion” group. Its HO allele is identical to only four others from the 1,011 genomes, one North American oak isolate (ADH), a human clinical isolate from the USA (ABB), and two wine isolates, one from Russia (ACQ) and one from Ivory Coast (ACK). All other isolates from ostensibly natural North American sources among the 1,011 genomes have different HO sequences (Supplementary Table S1). These observations suggest that there is little genetic variation among S. cerevisiae residing in North American oaks, and little mixing between North American oak populations and other natural populations on the same continent. Shared sequences with isolates from other continents presumably reflect either human-assisted transportation or an ancient diversity of alleles prior to global distribution. We note that the 2019 isolate came from the same English oak (Quercus robur) tree as did one of the S. cerevisiae isolates from the 2018 activity. A subsequent isolation activity in 2023 again isolated S. cerevisiae from some of the same English oaks, plus other nearby oaks of the same species, and also from two Northern Red oaks (Quercus rubra; unpublished results).

A K. lactis isolate from oak bark represents a new natural population

To ask if the lack of genetic diversity of our oak isolates extended to other yeast species, we further characterized the oak bark isolate of K. lactis to compare it with the distinct populations of the species that have been established in the literature. Pulsed-field gel electrophoresis (PFGE) of intact chromosomes revealed a small number of large chromosomes (Figure 3A), most similar to previously characterized isolates of a “new” North American genetic population (Lyutova et al., 2022). Notably, one member of the “new” North American population was isolated from an oak in Arizona (Naumova et al., 2004), a state adjacent to Colorado.

Figure 3. Molecular genetic analysis of a Denver oak isolate of K. lactis points to a new population.

(A) Pulsed-field gel electrophoresis of intact chromosomes from S. cerevisiae (“S. cer”, strain FY2742), S. paradoxus (“S. par”, strain N17), and the K. lactis isolate from an oak in Cheesman Park, Denver (“MMY0329”). Numbers to the left of the gel indicate sizes in Mb of the S. cerevisiae chromosomes based on the reference genome (S288C Genome Annotation Release R64.3.1, April 2021). The arrow/cursor icon is irrelevant. (B) Transmitted light micrographs of cells of the Denver oak K. lactis isolate MMY0329 following cultivation on rich agar medium (“rich medium”) or 1% potassium acetate (“sporulation medium”). Scale bars are 5 μm. (C) Phylogenetic tree of K. lactis isolates from around the world, including the Denver isolate and, as a control, K. marxianus, based on ACT1 sequence. Color-coding is used to label strains according to their published clade assignment based on >100,000 SNPs genome-wide (Friedrich et al., 2023). “70–4” is presumed to be the closest relative of the ancestral K. lactis strains prior to the acquisition of the LAC4–LAC12 cluster (Friedrich et al., 2023). (D) Sequence alignment of selected nucleotides within ACT1 comparing the Denver isolate (MMY0329) to sequences from other K. lactis populations. Nucleotide “1” corresponds to the 75^th nucleotide (inclusive) past the start codon. “•” denotes identity to the first/top sequence. (E) Short-read whole-genome sequence data from the Denver K. lactis isolate for the region adjacent to the location of the LAC4–LAC12 insertion in the dairy reference strain NRRL Y-1140. Colored bars represent individual sequence reads. (E) As in (D) but for a region of the right arm of chromosome 3 showing the presence there of genes C19547, C19519, and C19492. (F) Culture growth, as assessed by absorbance at 595 nm, of cultures of the indicated strains in liquid medium containing either 2% glucose or 10% lactose as the carbon source. MMY0374 is an S. cerevisiae isolate from oak 26204 (Table 1). Measurements were made every 5 minutes for 8 replicate cultures. Error bars show the 95% confidence interval. Decrease in absorbance for MMY0329 near the end of the experiment corresponded to clumping of cells around the periphery of the culture well.

To better understand the relationship of our isolate with documented populations, we performed short-read whole-genome sequencing and assembled the genome using either of two reference strains. The assembled genome contained six chromosomes ranging in size from 1.6 to 2.2 Mbp, fully consistent with our PFGE results. K. lactis isolates from the wild are typically haploid (Friedrich et al., 2023), and the diploid phase is usually short-lived, with meiosis taking place shortly after mating-type switching and mating, all of which are induced by starvation/stress (Herman and Roman, 1966; Barsoum et al., 2011). While we did not directly test the ploidy of our isolate, we observed what appeared to be elongated mating cells in addition to asci in a culture transferred to sporulation medium (Figure 3B), which we interpret as evidence that our isolate is haploid and underwent mating type switching and mating prior to sporulation. Due to extra copies of mating type genes present at the silent mating type loci, short-read DNA sequencing cannot determine which allele is present at the mating type locus, and we do not know if the strain is MATa or MATα.

A recent study suggested that polymorphisms in the ACT1 gene allow powerful discrimination between genetic populations (Lyutova et al., 2022), so we constructed a phylogenetic tree based on a larger ACT1 region from short-read sequences of 41 “domesticated” (dairy) and wild isolates plus the reference strain NRRL Y-1140. This tree almost perfectly recapitulated the relationships described by analysis of >100,000 genome-wide SNPs (Friedrich et al., 2023) (Figure 3C), confirming the power of the ACT1 locus to capture genetic differences. Our isolate did not fit in any of the populations analyzed previously (Figure 4B). We also directly compared our isolate with eight isolates characterized by Sanger sequencing of a portion of the ACT1 gene (Lyutova et al., 2022). Our isolate had a distinct combination of SNPs, notably distinct from the “new” North American population (Figure 3D). No read from our sequencing mapped to the LAC4–LAC12 gene cluster of the reference strain (Figure 3E). The presence of three genes (C19547, C19519, and C19492) on the right arm of chromosome 3, rather than on the left arm of chromosome 4, further distinguishes our isolate from the K. lactis var drosophilarum strains that lack LAC4–LAC12 (Figure 3F). Consistent with the lack of LAC4–LAC12, our isolate did not utilize lactose well (Figure 3G). Indeed, our isolate proliferated nearly as quickly as S. cerevisiae in glucose-containing medium and nearly as slowly as S. cerevisiae in lactose-containing medium, whereas the dairy strain NRRL Y-1140 proliferated at about the same rate in the two sugars (Figure 3G). These data are consistent with a model in which our isolate is the founding member of a previously unstudied genetic population of wild K. lactis that is adapted to living on oaks.

Figure 4. Sampling the diversity within species and genus by varying the geography and ecology.

(A) Map of the continental United States showing the location (blue star) in the state of Texas where the oak and sampling was performed. (B) The unique combination of SNPs discovered in the S. cerevisiae HO coding sequence from the Texas oak isolates (MMY0351, MMY0355 and MMY0356) is shown by a sequence alignment of selected nucleotides. Nucleotide “1” corresponds to the first nucleotide of the start codon. “•” denotes identity to the first/top sequence, which is the allele found in all our Denver isolates. (C) A photograph of the prickly pear cactus sampled in the Texas oak forest, March 2019. Strain MMY0358 was isolated from the fruit. (D) The growth in liquid rich medium (YPD) at the indicated temperatures was assessed for the indicated strains by measuring the absorbance at 595 nm following ~21 hr of incubation. (E) The same strains as in (D) were streaked from frozen glycerol stocks to solid rich (YPD) medium, incubated overnight at 30°C, and then cells were scraped from the streaks and resuspended in water. 10-fold serial dilutions were spotted onto new YPD plates with a metal 48-prong replicator and the plates were incubated at the indicated temperature overnight before being photographed.

A new HO allele in S. cerevisiae isolated from Texas oaks

By venturing outside of Denver, we found that it was possible to isolate North American oak strains with previously undescribed HO sequences. Outside the context of the outreach activity, we used the same isolation method to obtain S. cerevisiae from seven post oak trees (Quercus stellata) in College Station, Texas (Figure 4A). All the isolates were homothallic. We sequenced HO from three of the Texas isolates and found that they were identical, and represented a new allele, encoding the same protein sequence as the other North American oaks, but carrying synonymous SNPs that make it distinct from all others reported here and elsewhere (Figure 4B). The closest available match, differing by only a single SNP, is represented 14 times among the 1,011 isolates, including two isolates from Pennsylvania oaks (CMP, CMQ) and one from an oak in Spain (BBR).

A heat-tolerant Texan isolate of Saccharomyces paradoxus

Interestingly, two of the 14 strains described above (YCX and GAT, which represent the same isolate, originally named UWOPS87–2421) came from a prickly pear in Hawai’i (Opuntia megacantha) (Cubillos et al., 2009; Liti et al., 2009). We noticed prickly pears of undefined species (likely Opuntia engelmannii) growing in the same Texas oak forest, and isolated from the fruit of one plant (Figure 4C) a strain that we initially assumed to be S. cerevisiae. After failing to amplify HO using the same primer pair that was successful with the oak isolates (Figure S2A), we used other primer pairs targeting smaller HO fragments and reduced the annealing temperature in the PCR. Two combinations yielded products that were near the expected sizes (Figure S2A), but sequencing revealed that the amplicons instead were 99% identical to portions of the MHP1 or FLO11 genes of a strain of S. paradoxus (Figure S2B,C). We found presumptive sites of primer annealing to the S. paradoxus MHP1 and FLO11 loci (Figure S2D), suggesting that the amplicons arose from spurious primer annealing in the absence of perfect matches to the template genomic DNA. Flo11 (also called Muc1 in S. cerevisiae) is a cell surface protein harboring many repeats of a short motif, encoded by tandemly repeated sequences that are prone to recombination-mediated variation in copy number (Verstrepen et al., 2005). One presumptive primer binding sequence fell within one such repeat, likely explaining the doublet of PCR products (Figure S2D). To directly test the hypothesis that the prickly pear isolate is S. paradoxus, we performed ITS sequencing. A 1424-bp ITS sequence was a perfect match with S. paradoxus (Figure S2E).

We were surprised to find S. paradoxus here, where the maximum temperature (41°C during 2012–2019) significantly exceeds 38°C, the suggested maximum for this species (see above). When we compared liquid culture growth in rich medium at different temperatures, the prickly pear isolate of S. paraxodus (hereafter MMY0358) was similar to an S. cerevisiae strain isolated from a nearby oak tree (Figure 4D). By contrast, an S. paradoxus strain (N17) originally isolated from an oak in Russia (Naumov, 1987) had a lower maximum growth temperature (Figure 4D). Plating the same strains on solid rich medium at 37°C confirmed that high temperature tolerance correlated with the climate of origin, not the species (Figure 4E).

DISCUSSION

Here we describe new biological insights resulting from an activity that engaged children in middle school classrooms in a relatively simple set of hands-on procedures targeted to sixth grade learning objectives in Biology. To broaden the impact of the activity, we also involved university students and PhD students from two university campuses. Moreover, two of the co-authors, N.B. and V.S., designed and executed the grouping analysis of the HO sequences from the 1,011 genomes database while they were high school students, as part of the Anschutz Medical Campus CU Science Discovery STEM Research Experience. Thus this activity represents a true partnership spanning multiple ages and levels of education.

The original goal of our activity was to isolate wild S. cerevisiae in the hopes of identifying a variety of alleles of HO, as was achieved at Mount Carmel (Katz Ezov et al., 2010). Taken together with published sequences, however, our results point to limited diversity within North American oak populations. Thus the original puzzle of shared HO alleles spread all over the world led to a new puzzle of nearly invariable HO sequences across a continent. The Mount Carmel site (dubbed “Evolution Canyon”) features extreme variation in microclimate within a small area, and soil and leaves from many different plants were also sampled to obtain the S. cerevisiae isolates (Ezov et al., 2006). The limited diversity among North American oak isolates may reflect limited diversity of the ecological niches we sampled in Denver, rather than limited geographical diversity per se.

Indeed, by widening the geographical and ecological ranges of our sampling, we broadened the diversity of our isolates. In southeast Texas we discovered a previously unidentified HO allele in S. cerevisiae living on oaks, and, from a single plant of a different species surrounded by those same oaks, isolated a strain of S. paradoxus that has apparently adapted to survive the hot Texas summers. We note that previous studies identified a North American population of S. paradoxus, called SpB, with distinctly higher maximum growth temperatures than European and other North American populations (Leducq et al., 2014), and the S. paradoxus strains that best match the MHP1 and FLO11 sequences from the Texas isolate belong to the SpB population. Furthermore, a very recent study of open agave fermentations also found S. paradoxus in parts of Mexico that are similarly hot (Gallegos-Casillas et al., 2024). Prickly pears are widespread in the same parts of Mexico, and it is tempting to speculate that the S. paradoxus strain we found is part of a heat-tolerant North American population (likely SpB) covering a large geographical region. Given this precedent, it should be possible to explore greater Saccharomyces diversity in a middle-school outreach context by sampling other plant species, including unroasted coffee and cacao beans of defined origin (Ludlow et al., 2016), fruit, and/or flour. The latter could be combined with a sourdough starter activity (Landis et al., 2021). Simple phenotypic analysis (e.g. growth temperature) could also be performed in the classroom setting to generate additional useful insights.

We did not originally plan to analyze K. lactis, but recent advances in the study of K. lactis population genetics (Lyutova et al., 2022; Friedrich et al., 2023) inspired us to carry out genomic and phenotypic characterization of the single isolate we obtained. Unlike the case with S. cerevisiae, our analysis of the K. lactis isolate indicates more heterogeneity among North American oak populations for this species. Indeed, based on others’ findings and the hypothesis that in some scenarios HO might escape purifying selection (see Introduction), we had hoped to find allele diversity in the S. cerevisiae HO gene, but ultimately a K. lactis gene (ACT1) encoding a highly conserved protein (actin) shows much more variation.

We consider two explanations for this difference that are not mutually exclusive. First, the genes may be fundamentally different, with the allele at HO or a nearby gene conferring some functional advantage to S. cerevisiae in the oak habitat, whereas K. lactis ACT1 is a functionally neutral readout of relatedness. However, S. cerevisiae oak isolates cluster relatively closely together based on analysis of five (Fay and Benavides, 2005) or 12 (Legras et al., 2007) non-HO sequences or the entire genome (Liti et al., 2009; Peter et al., 2018). Furthermore, phylogenetic analysis of ACT1 sequence alone can recapitulate S. cerevisiae relatedness about as well as does HO sequence (Figure S3). There is no reason to suspect that K. lactis ACT1 is under different evolutionary constraints.

Second, the species may be fundamentally different in the mode and extent of their dispersal. K. lactis may be more like non-domesticated Saccharomyces species (e.g. oak populations of S. paradoxus), for which genetic differentiation increases rapidly with physical distance, pointing to limited dispersal (Koufopanou et al., 2006; Boynton and Greig, 2014; He et al., 2022). By contrast, North American oak populations of S. cerevisiae may experience continuous homogenization via rapid/effective dispersal. Indeed, none of the Colorado oaks we sampled is native, and each is <100 years old, hence dispersal must be efficient to populate such new trees. The similarity of North American S. cerevisiae oak isolates to each other but not to North American isolates from other natural sources implies that oak strains enjoy a privileged mode of dispersal compared to non-oak S. cerevisiae isolates. Considering the very recent isolation of S. cerevisiae from butterfly intestines (Bendixsen et al., 2021) and of multiple yeast species from the digestive tracts of migrating Taiwanese butterflies (Lin et al., 2022), and given the importance of Colorado oaks as butterfly roosts (Scott, 2020), we speculate that migrating insects like butterflies may be important dispersal vectors. It may be possible to include insect sampling in future outreach activities. There may also be strict genetic requirements for stable occupancy of the oak bark habitat, such that non-oak strains arrive just as frequently on oaks but are unable to persist there. Indeed, the fact that we isolated S. paradoxus from a cactus within meters of oak trees on which only S. cerevisiae was found is probably best explained by the population of S. paradoxus in this region of North America being adapted to non-oak niches.

DATA AVAILABILITY

Strains and plasmids are available upon request. The new S. cerevisiae HO sequence (BankIt2837613 Seq PP902360), S. paradoxus MHP1 (BankIt2842207 Seq PP935650) and FLO11 sequences (BankIt2842203 Seq PP935649), and the K. lactis genome sequence (project PRJNA1128167) have been deposited at NCBI.