eLife digest

elife

eLife

2050-084X

eLife Sciences Publications, Ltd

81056

10.7554/eLife.81056

Research Article

Evolutionary Biology

Microbiology and Infectious Disease

Selection on plastic adherence leads to hyper-multicellular strains and incidental virulence in the budding yeast

Ekdahl

Luke I

1†Salcedo

Juliana A

1†Dungan

Matthew M

1‡Mason

Despina V

1Myagmarsuren

Dulguun

1Murphy

Helen A

https://orcid.org/0000-0002-4363-4543

hamurphy@wm.edu11

https://ror.org/03hsf0573

Department of Biology, College of William and Mary

Williamsburg

United States

Harper

Diane M

Reviewing Editor

https://ror.org/00jmfr291

University of Michigan

United StatesLandry

Christian R

Senior Editor

https://ror.org/04sjchr03

Université Laval

Canada

†

These authors contributed equally to this work

‡

School of Medicine, Vanderbilt University, Nashville, United States

02112023

2023

e81056

1406202201112023

This manuscript was published as a preprint at .

04062022

2023

Ekdahl, Salcedo et al

http://creativecommons.org/licenses/by/4.0/

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Many disease-causing microbes are not obligate pathogens; rather, they are environmental microbes taking advantage of an ecological opportunity. The existence of microbes whose life cycle does not require a host and are not normally pathogenic, yet are well-suited to host exploitation, is an evolutionary puzzle. One hypothesis posits that selection in the environment may favor traits that incidentally lead to pathogenicity and virulence, or serve as pre-adaptations for survival in a host. An example of such a trait is surface adherence. To experimentally test the idea of ‘accidental virulence’, replicate populations of Saccharomyces cerevisiae were evolved to attach to a plastic bead for hundreds of generations. Along with plastic adherence, two multicellular phenotypes— biofilm formation and flor formation— increased; another phenotype, pseudohyphal growth, responded to the nutrient limitation. Thus, experimental selection led to the evolution of highly-adherent, hyper-multicellular strains. Wax moth larvae injected with evolved hyper-multicellular strains were significantly more likely to die than those injected with evolved non-multicellular strains. Hence, selection on plastic adherence incidentally led to the evolution of enhanced multicellularity and increased virulence. Our results support the idea that selection for a trait beneficial in the open environment can inadvertently generate opportunistic, ‘accidental’ pathogens.

eLife digest

Yeast are microscopic fungi that are found on many plants, in the soil and in other environments around the world. But, when given the chance, some yeasts are also good at infecting human and other animals and causing disease.

It has been proposed that some opportunistic microbes may have dual-use traits that evolved for one purpose in their natural environment but also incidentally allow them to infect animals. For example, a toxin that helps the opportunistic microbe compete against neighboring microbes may also weaken an animal. Or the ability of many individual microbe cells to clump together into structures known as biofilms on solid surfaces, or floating mats called flors on liquids, helps them to survive in harsh environments, whether in the soil or in the body of an animal.

To investigate this possibility, Ekdahl, Salcedo et al. examined whether artificially selecting yeast with a specific trait – the ability to stick to plastic beads – in the absence of any host animals would inadvertently also select for yeast that were good at causing disease. This trait was selected because it has not been previously linked to opportunistic yeast infections.

The team grew the yeast for 400 generations in tubes that each contained a plastic bead. At every generation, only yeast that stuck to the plastic bead were transferred to a fresh tube to grow the next generation. The experiments found that, not only did the ability of the yeast to stick to the plastic increase over time, but the yeast also evolved the ability to form biofilms and flors. Furthermore, the sticky yeast killed an insect host known as wax moth larvae more quickly than non-sticky yeast.

Together, these findings demonstrate that when microbes evolve in an environment that is devoid of any host animals, selection can inadvertently favor dual-use traits that also help the yeast to infect animals. Opportunistic yeast infections are of increasing concern in human patients, particularly those with weakened immune systems. Understanding which yeast traits are dual-use will help guide future efforts in combatting yeast and other opportunistic microbes.

yeastexperimental evolutionvirulenceadherencebiofilmFLO11wax moth

Research organismS. cerevisiae

http://dx.doi.org/10.13039/100000002

National Institutes of Health

R15-GM122032

Murphy

Helen A

http://dx.doi.org/10.13039/100000001

National Science Foundation

DEB-1839555

Murphy

Helen A

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Author impact statement

Yeast that were evolved to adhere to plastic surfaces for a few hundred generations became hyper-adherent and more virulent.

Introduction

The study of infectious disease often focuses on pathogenic microbes that either specialize in exploiting animal hosts or act as commensals that switch to pathogenesis when the delicate balance between host and microbe is perturbed. These microbes are presumed to have co-evolved complex adaptations that allow survival and reproduction in and on hosts. However, there exists a broad range of microbial organisms that live in the open environment (i.e. soil, vegetation, aquatic habitats) that are capable of causing disease when the opportunity presents itself (Brown et al., 2012). Such microbes also have adaptations that allow host exploitation, but the origin of these adaptations is unclear, as growth and survival in a host is not a required part of the lifecycle (Casadevall and Pirofski, 2007). For example, the soil-associated bacteria Pseudomonas aeruginosa (Selezska et al., 2012) and Burkholderia cepacia (Mahenthiralingam et al., 2008) can infect the lungs of cystic fibrosis patients.

The existence of virulence traits in environmentally-derived opportunistic pathogens may be due to selection favoring the traits for other uses in the non-host environment, which challenges the idea that co-evolution is a requirement of microbial pathogenesis and virulence. This hypothesis, first proposed and explored in the bacterial literature, is known as ‘coincidental selection’ (Levin and Svanborg Edén, 1990). Much of the research testing the coincidental selection hypothesis has focused on the biotic environment as a selective pressure, specifically the role of predatory bacteriophages and protists (reviewed in [Sun et al., 2018; Davies et al., 2016; Erken et al., 2013]). The relationship between virulence, bacteria, and their predators is complex (Brüssow, 2007). For example, a positive relationship between amoeba predation and virulence was shown in Escherichia coli (Adiba et al., 2010), while selection by amoeba predation led to a decrease virulence in P. aeruginosa (Leong et al., 2022). In the same bacterial species, an evolution experiment with both protist and phage predators showed that phage could mitigate the decrease in virulence associated with protist predation (Friman and Buckling, 2014). Despite the complexity, it is clear that selection from the biotic environment can strongly influence bacterial virulence.

A hypothesis similar to coincidental selection, known as ‘accidental virulence’, has been proposed (Casadevall and Pirofski, 2007) and explored in parallel in the eukaryotic microbial literature, also with a focus on the biotic environment (reviewed in Siscar-Lewin et al., 2022; Casadevall et al., 2019). For example, in the fungus Cryptococcus neoformans, traits that protect from predatory amoeba also play a role in human infection (Steenbergen et al., 2001), and selection by amoeba can increase the prevalence of such traits (Fu et al., 2021; Steenbergen et al., 2003). Similarly, co-culturing Paracoccidioides fungi with amoeba can lead to increased virulence (Albuquerque et al., 2019).

In both the bacterial and eukaryotic literature, there has been less focus on the role of selection imposed by the abiotic environment, although temperature is increasingly a consideration, as thermotolerance and halotolerance may make colonization in and on humans more likely (Casadevall, 2020; Garcia-Solache and Casadevall, 2010). For example, the emerging opportunistic yeast Candida auris, which is found in warm and salty coastal wetlands, can cause severe systemic infection (Arora et al., 2021). There is also experimental evidence that increased temperature can select for virulence in P. aeruginosa (Friman et al., 2011).

The dual-use virulence traits that can be under ‘coincidental selection’ and lead to ‘accidental virulence’ are numerous and range from toxin production, such as the production of gliotoxin in the soil-dwelling filamentous fungus Aspergillus fumigatus (Hillmann et al., 2015; Gupta et al., 2021), to protective structures, such as capsule formation in C. neoformans (Casadevall et al., 2003). Another type of trait, which is the focus of the research presented here, is adherence. Adherence is important for many microbial behaviors required for survival (e.g., biofilm formation) (West et al., 2007), but can also play a role in pathogenicity and virulence (Douglas, 2003; Hall-Stoodley et al., 2004). The ability to adhere to and invade tissues, as well as form communities resistant to anti-microbials, can be key to successful pathogenicity. In the soil-associated yeast, Blastomyces dermatitidis (Baumgardner and Laundre, 2001), which can cause lung infections, the deletion of a single adhesin gene abolishes pathogenesis (Klein, 2000).

The function of microbial traits in both the open environment and the host is crucial evidence for the coincidental selection-accidental virulence hypothesis. However, experimental evolution allows the idea to be tested directly; thus far, most evolution experiments have focused on co-culturing opportunistic microbes with predators (e.g., [Leong et al., 2022; Friman and Buckling, 2014; Fu et al., 2021; Steenbergen et al., 2003; Albuquerque et al., 2019; Friman et al., 2011; Mikonranta et al., 2012; Hosseinidoust et al., 2013]). In the research presented here, rather than altering the biotic or abiotic environment and determining the effect on virulence, we apply direct selection to a specific trait hypothesized to be dual-use in the biomedical model yeast Saccharomyces cerevisiae.

Aside from serving as a model for genetics and cell biology, and being found in a myriad of ecological niches around the globe (Peter et al., 2018), S. cerevisiae is also an opportunistic pathogen capable of infecting immunocompromised individuals, with reports of infections increasing (Aucott et al., 1990; Muñoz et al., 2005; Llopis et al., 2014; Enache-Angoulvant and Hennequin, 2005; Hennequin et al., 2000; Pérez-Torrado and Querol, 2015). As such, it has been a model for fungal pathogenesis-related traits (Clemons et al., 1994; Fraser et al., 2012; McCusker et al., 1994a; Phadke et al., 2018). Some environmentally derived strains are capable of adhering to surfaces and expressing associated aggregative phenotypes (Verstrepen and Klis, 2006). These range from biofilms on solid and semi-solid agar, to pseudohyphal growth and agar invasion, to floating mats on liquid surfaces (flors). Of these multicellular phenotypes, only invasive and pseudohyphal growth have been linked to pathogenicity in S. cerevisiae (McCusker et al., 1994a; Phadke et al., 2018; Palecek et al., 2002), although biofilm formation has been linked to pathogenicity in other fungi (Fanning and Mitchell, 2012). Not all strains are capable of expressing these phenotypes. And while some strains express multiple multicellular traits (Casalone et al., 2005; Zara et al., 2009), there does not appear to be a correlation among the numerous adherence phenotypes (Hope and Dunham, 2014). Despite overlap in conserved signaling and regulatory networks governing the traits, the ability of a strain to express multicellularity in one form does not necessarily suggest the ability to express it in another (Cullen and Sprague, 2012). This is not entirely surprising, since different environmental conditions likely favor different multicellular phenotypes.

Here, yeast populations were artificially selected for adherence ability in one context, in order to determine whether it led to an increase in virulence in another. Specifically, yeast were evolved to adhere to a plastic bead (Poltak and Cooper, 2011), then tested against wax moth larvae to estimate virulence. In replicate populations of two genetic backgrounds, the yeast increased in their ability to express multicellularity in numerous forms. Not all multicellular phenotypes responded in the same way, with pseudohyphal growth appearing to evolve independently from plastic adherence, biofilm formation, and flor formation. This phenotypic evolution demonstrates the complexity of the interacting genetic networks underlying yeast multicellularity. Along with these correlated effects of selection, the yeast also became more virulent. Our results experimentally demonstrate that selection on dual-use traits outside of a host environment can inadvertently favor pre-adaptations for virulence.

Results

The evolution experiment was conducted in two genetic backgrounds. While most S. cerevisiae strains can be pathogenic against wax moth larvae if administered with a high enough inoculum (Phadke et al., 2018), we chose two strains isolated from clinical settings, and therefore, had a known tendency toward human pathogenicity as well. These strains were highly heterozygous, with tens of thousands of SNPs in each genome (Magwene et al., 2011); thus, the strains contained standing genetic variation on which selection could act. The first strain, YJM311, was isolated from the bile tube of a patient in San Francisco in 1981 (McCusker et al., 1994b); its recombinant offspring vary in at least one form of filamentous growth (Lenhart et al., 2019). The second strain, YJM128, was isolated from the lung of a patient in Missouri in the 1980s (Tawfik et al., 1989). Both strains were engineered to constitutively express mCherry, then sporulated, digested, and germinated. Each pool of recombinant offspring was used to inoculate replicate populations.

From each ancestral strain, 10 replicate populations were evolved via serial transfer for 350–400 mitotic generations, half punctuated with sexual cycles every 40 generations, and two without beads as controls. YJM311 was evolved for 8 sexual cycles, while YJM128 was evolved for 9. Populations were grown in limiting medium in glass tubes in the presence of a plastic bead (Figure 1A–B). After growth, beads were washed, suspended in water, and sonicated to detach cells. The cell suspension was transferred to the next tube for growth (Figure 1—figure supplement 1). In YJM128, the sexual control failed to propagate after the first cycle. Therefore, a full complement of controls (three asexual and three sexual) were subsequently initiated and evolved in the same manner.

Figure 1.Evolution of bead adherence.

(A) Image of a 7 mm experimental bead; close-up image with attached cells expressing mCherry, scale bar represents 50 µm. (B) Whole population-bead adherence of replicate populations over the experimental cycles, as estimated at the end of the experiment from cryopreserved stocks. All replicates of one background were grown and measured in one assay to test their adherence ability (including control populations that were not evolved in the presence of a bead). Y-axis plots the number of cells adhering to a plastic bead on a log scale, as estimated by hemocytometer counts (± S.E.M.). Along with the ancestral timepoint, for each population at each timepoint, cells from eight replicate beads were counted in YJM311 (670 beads in total); for YJM128, four replicate beads were counted (542 beads in total).

Figure 1—source data 1.Counts of cells that adhered to plastic beads grown with replicate populations at different timepoints from YJM311/HMY7-derived populations.

Figure 1—source data 2.Counts of cells that adhered to plastic beads grown with replicate populations at different timepoints from YJM128/HMY355-derived populations.

Figure 1—figure supplement 1.Schematic of the Experimental Cycle.

Solid arrows represent steps in serial transfer; dashed lines represent punctuated sexual treatments. Populations were evolved on a 12-day cycle. In each cycle, populations were grown in 10 ml of minimal medium in a glass tube containing a sterile 7 mm polystyrene bead for 48 hr with rotation. The bead was removed, washed, suspended in water, and gently sonicated to detach cells from the bead. The cell suspension was used to inoculate the next 10 ml tube. After four serial transfers, asexual populations were refrigerated and sexual populations were sporulated for 48 hr. Asci were digested, germinated and mated. Refrigerated cultures and mated spores were used to begin the next 12-day cycle.

The number of cells attaching to the bead increased over time in the experimental populations (Figure 1; Supplementary file 1c). The data were analyzed with a linear mixed-effect model (LMM); coefficients for the treatment x cycle interaction estimate the effect a given treatment had over time on the number of cells adhering to a bead, as measured by hemocytometer counts. In both genetic backgrounds, the interaction coefficients were positive and sexual populations showed a larger effect than the asexual (YJM311, asexual*cycle = 0.054 (confidence interval ± 0.048), sexual*cycle = 0.184 (±0.048); YJM128, asexual*cycle = 0.47 (±0.242), sexual*cycle = 0.598 (±0.254)). Hence, as has been demonstrated in other evolution experiments in microorganisms (Zeyl and Bell, 1997; Goddard et al., 2005; McDonald et al., 2016; Kosheleva and Desai, 2018; Kaltz and Bell, 2002; Lachapelle and Bell, 2012), sexual populations showed increased adaptation in comparison to asexual populations.

In order to interrogate other effects of adherence selection, at the end of the experiment, ten individual clones were isolated from each population from four timepoints (for YJM311, cycles 2, 4, 6, 8, and for YJM128, cycles 1, 3, 6, 9). For each genetic background, over 400 clones, along with 20 ancestral recombinant offspring, were arrayed in a 96-well plate format for analysis of multicellular phenotypes.

Plastic adherence ability

The panel of clones was first assayed for plastic adherence ability (Figure 2A with Figure 2—figure supplements 1 and 2). Plastic adherence was measured with a microplate reader that detected the fluorescence signal of cells remaining attached to a well in which culture was grown to saturation and gently rinsed. As expected, plastic adherence increased over time in the clones from experimental populations (YJM311: control*cycle = 0.008 (±0.056), asexual*cycle = 0.040 (±0.040), sexual*cycle = 0.100 (±0.040); YJM128: control*cycle = 0.016 (±0.076), asexual*cycle = 0.012 (±0.102); sexual*cycle = 0.187 (±0.114); Supplementary file 1d). Fluorescent signal could have evolved over the course of the experiment; indeed, from the beginning, YJM128 produced a brighter fluorescent signal than YJM311, suggesting the existence of genetic variants that could influence fluorescence expression. Despite the potential for noise in the measurement, the signal of increased adherence throughout the experiment was apparent in both genetic backgrounds. These clonal data support the results of the whole-population adherence measurement (Figure 1), in which cells attaching to a plastic bead were counted manually with a hemocytometer.

Figure 2.Evolution of multicellular phenotypes.

Ten clones were isolated from each population at four timepoints and assayed in triplicate (except for flor formation, which had a single replicate). In all panels, large points represent the average of a treatment (asexual, sexual, control) ± 2 s.e.m.; smaller points represent the average of a replicate population ± s.e.m. Data at cycle 0 represent the average of 20 ancestral segregants. Representative images demonstrate the variation found in the phenotypes. (A) Plastic adherence was estimated by measuring the fluorescent signal of cells that adhered to the bottom of of a black, clear-bottom 96-well plate. (B) CCM was scored after growth on solid, glucose-limiting medium using the scale on the right, with 1 representing no biofilm and 5 the most structured colonies. (C) Flor formation was scored after growth in minimal medium using the scale on the right, with 0 representing no floating cells and 4 representing a full mat. (D) PSH was scored after growth on solid nitrogen-limiting medium. Images were processed to determine the percentage of growth pixels that were pseudohyphal compared to the central colony. The trajectory of replicate populations from each ancestral background can be found in Figure 2—figure supplements 1 and 2.

Figure 2—source data 1.Plastic adherence, biofilm, flor, and PSH measurements for ancestor and evolved clones from the YJM311/HMY7-derived replicate populations.

Figure 2—source data 2.Plastic adherence, biofilm, flor, and PSH measurements for ancestor and evolved clones from the YJM/128HMY355-derived replicate populations.

Figure 2—figure supplement 1.Evolved phenotypes by replicate population for the YJM311 background.

Phenotypes were assessed as in Figure 2. Larger points represent replicate population averages, with each replicate represented by a different shape/color combination (as in Figure 4); smaller points represent measurements from individual clones. Fluorescence, CCM, and PSH are average values of three replicate measurements per clone; Flor is based on one replicate.

Figure 2—figure supplement 2.Evolved phenotypes by replicate population for the YJM128 background.

We next measured the ability of the clonal panel to express three other seemingly different multicellular phenotypes.

Biofilm colony formation

The first multicellular phenotype was the ability to form complex colony morphology (CCM) on solid agar, which is indicative of the ability of a strain to form a differentiated biofilm colony, also known as a ‘fluffy colony’ (Kuthan et al., 2003; St’ovíček et al., 2010; Št’ovíček et al., 2014; Váchová et al., 2011; Maršíková et al., 2017). This phenotype is correlated with another multicellular phenotype (Hope and Dunham, 2014), mat formation, which is a biofilm that forms on semi-solid agar (Reynolds and Fink, 2001); we therefore only assayed CCM. Morphology was scored after growth on solid, glucose-limiting medium using a scale from 1 to 5, with 1 representing no biofilm and 5 representing the most structured colonies (Hope and Dunham, 2014; Granek and Magwene, 2010; Figure 2B with Figure 2—figure supplements 1 and 2).

In both genetic backgrounds, the selected populations increased in their ability to exhibit CCM compared to the ancestor, while the control populations either maintained or decreased their expression (YJM311: control*cycle = −0.024 (±0.056), asexual*cycle = 0.119 (±0.039), sexual*cycle = 0.136 (±0.039); YJM128: control*cycle = 0.007 (±0.020), asexual*cycle = 0.048 (±0.026); sexual*cycle = 0.052 (±0.030); Supplementary file 1e ). YJM311 evolved to exhibit stronger CCM than YJM128, despite the latter evolving for one more cycle.

Flor formation

The second multicellular phenotype was the ability to form a flor (or velum), which is a floating mat containing cells attached to one another in an extracellular matrix (Zara et al., 2009). Flors form at the liquid-air interface in static conditions and are most commonly found during sherry and wine making processes (Legras et al., 2016). Flor formation was scored after growth in minimal medium, with 0 representing no floating cells and 4 representing a full mat. The ability to form flors increased in both genetic backgrounds (Figure 2C with supplements 1 and 2), despite the cultures being grown with agitation. In YJM311, the ancestral clones showed no ability to generate flors, yet, its evolved populations did (control*cycle = 0.007 (±0.048), asexual*cycle = 0.075 (±0.034), sexual*cycle = 0.163 (±0.034); Supplementary file 1f). In YJM128, ancestral clones showed limited ability to form flors. The experimental populations either remained or increased in flor-forming ability, while control populations decreased in theirs (control*cycle = −0.109 (±0.032), asexual*cycle = 0.031 (±0.043); sexual*cycle = −0.003 (±0.048); Supplementary file 1f).

Pseudohyphal growth

The final phenotype, pseudohyphal growth (PSH), is a form of filamentous growth thought to represent a foraging strategy. It is characterized by substrate invasion and incomplete separation of mother-daughter cells growing in an elongated, unipolar budding pattern (Gimeno et al., 1992). This phenotype is sometimes correlated with invasive growth, so only PSH was assayed. Filamentous and invasive growth have been associated with pathogenicity and virulence in S. cerevisiae (McCusker et al., 1994a; Phadke et al., 2018; Palecek et al., 2002), as well as in other fungal pathogens of humans and plants (Lengeler et al., 2000). PSH was scored on solid nitrogen-limiting medium; images were processed to determine the percentage of growth that was pseudohyphal compared to the central colony. Unlike the previously assayed phenotypes, the two genetic backgrounds did not evolve similarly with respect to PSH.

In YJM311, the experimental populations did not increase in their PSH ability compared to the ancestor (Figure 2D with supplements 1 and 2); rather, all treatments showed some loss. Throughout the cycles, a moderate level of PSH was maintained in some of the experimental populations, with one sexual replicate doubling its PSH index, but it was lost in the controls and the other experimental populations (control*cycle = −0.61 (±0.60), asexual*cycle = −0.65 (±0.42), sexual*cycle = −0.97 (±0.43); Supplementary file 1g).

In YJM128, both the experimental and control populations increased in their PSH ability (control*cycle = 1.29 (±0.39), asexual*cycle = 0.61 (±0.51), sexual*cycle = 0.91 (±0.55); Supplementary file 1g ). This was true for the control population that was evolved in concert with all experimental populations, as well as the 6 control populations that were initiated subsequently.

The results from both genetic backgrounds suggest that the evolution of PSH was a response to selection in nutrient limiting conditions and not a response to selection for adherence, as the controls and experimental populations behaved similarly. Of all the adherence and multicellular phenotypes investigated, most of which appeared to increase throughout the experiment, PSH appeared to be independent from the others.

Overall, our phenotyping data show that selection on the ability to adhere to a plastic surface generated a correlated response in multiple multicellular phenotypes, and nutrient limiting conditions favored a further multicellular phenotype in one of the backgrounds.

Hyper-multicellularity

To understand the phenotypic landscape of the evolved populations and to determine whether the different forms of multicellularity evolved in concert in individual clones, the clonal phenotype data were combined in a principal components analysis (PCA) (Figure 3 with supplements 1 and 2). In YJM311, the loadings of the first two components, which explain 78% of the variation, show that evolved clones with the most extreme values of plastic adherence and flor formation do not tend to also excel at PSH. There were clones, however, that evolved to excel in all of the phenotypes, while not obtaining the most extreme values of the individual traits. In YJM128, the first two loadings explain 70% of the variation, and again, PSH appeared separated from the other multicellular phenotypes. Individual correlations between traits bear out this interpretation (Figure 3—figure supplements 3 and 4). When grouped by experimental treatments, clones from control, asexual, and sexual populations tended to occupy their own, somewhat overlapping, phenotypic space.

Figure 3.Evolved multicellularity.

(Top panel) Principal components analysis of clones from ancestral and evolved populations. The loadings of PC1 for YJM311 were –0.616*Flor - 0.573*PA - 0.493*CCM +0.221*PSH; for PC2, they were 0.851*PSH +0.506*CCM - 0.137*Flor. In YJM128, the loadings of PC1 were 0.601*CCM +0.535*PA +0.451*Flor +0.386*PSH; for PC2, they were 0.852*PSH - 0.226*PA - 0.472*Flor. PCA with population and cycle information can be found in Figure 3—figure supplements 1 and 2, while individual correlations can be found in Figure 3—figure supplements 3 and 4. (Bottom panel) Principal components analysis with highlighted points representing strains chosen for virulence assays: blue circles represent low multicellularity clones; orange squares represent hyper-multicellular clones; gray triangles represent the rest of the clonal panel. In YJM311, the non-multicellular clones were chosen from ancestral and control populations, while in YJM128, they were chosen from ancestral and early experimental populations.

Figure 3—source data 1.Survival data for wax moth larvae injected with strains from YJM311/HMY7-derived populations.

Figure 3—source data 2.Survival data for wax moth larvae injected with strains from YJM128/HMY355-derived populations.

Figure 3—figure supplement 1.PCA of YJM311 Evolved Populations.

Each point represents a single clone; ancestor- green, control- gray through black, asexual- light blue through dark blue, sexual- light red through dark red. Colors become darker as cycle number increases. In the first panel, shapes as in Figure 4 to highlight different replicate populations.

Figure 3—figure supplement 2.PCA of YJM128 evolved populations.

Figure 3—figure supplement 3.Correlations among phenotypes for individual clones from YJM311 populations.

All correlations were significant with p<0.001. The adjusted R² for the relationships are as follows: PA vs. CCM: 0.167, PA vs. Flor: 0.306, PA vs. PSH: 0.020, CCM vs. Flor: 0.220, PSH vs Flor: 0.105, CCM vs. PSH: 0.015. Despite the statistical significance, most of these correlations explain little of the variance in the data; the presence of one multicellular phenotype does not necessarily predict the presence of another.

Figure 3—figure supplement 4.Correlations among phenotypes for individual clones from YJM128 populations.

All correlations were significant with p<0.001. The adjusted R² for the relationships are as follows: PA vs. CCM: 0.235, PA vs. Flor: 0.071, PA vs. PSH: 0.0294, CCM vs. Flor: 0.121, PSH vs Flor: 0.0162, CCM vs. PSH: 0.102. Despite the statistical significance, most of these correlations explain little of the variance in the data; the presence of one multicellular phenotype does not necessarily predict the presence of another.

In both backgrounds, as the populations evolved, there were individual clones that increased in all abilities, and became ‘hyper-multicellular’. Thus, a simple process of selection for plastic adherence led to correlated effects in multiple multicellular traits. These correlated effects were apparent both at the population-level, with mean phenotypes increasing in populations over the generations, but also at the individual-level with the evolution of hyper-multicellularity.

FLO11 length variation

One possible explanation for the increase in multiple forms of multicellularity is a change in a genetic element common to all four phenotypes. A genome-wide investigation into the genetic basis of three multicellular phenotypes (biofilm formation, PSH, and invasive growth) in a lab strain of S. cerevisiae found that each phenotype appeared to have its own set of hundreds of genes underlying its expression, but also some overlap in select transcription factors and signaling pathways (Ryan et al., 2012). Notably, the one element that all of the traits had in common, as do other aggregative phenotypes, is the requirement of the cell adhesin, Flo11p (Reynolds and Fink, 2001; Zara et al., 2005; Lo and Dranginis, 1998), which allows yeast cells to adhere to surfaces and other cells (Dranginis et al., 2007).

Flo11p is a cell surface protein with three domains: a C-terminal that facilitates attachment to the cell wall, an exposed N-terminal immunoglobulin-like domain that mediates cell adhesion (Kraushaar et al., 2015), and a low-complexity, serine-threonine rich B-domain of variable length that extends the adhesion domain away from the cell (Dranginis et al., 2007). The tandem repeats in the B-domain have been shown to be unstable (Fidalgo et al., 2006; Fidalgo et al., 2008) and to vary in length naturally (Zara et al., 2009; Oppler et al., 2019; David-Vaizant and Alexandre, 2018; Verstrepen et al., 2005). Differences in the length of this repetitive region have been shown to affect the strength of multicellular phenotypes in some genetic backgrounds (Zara et al., 2009; Fidalgo et al., 2008).

To determine whether FLO11 length changed throughout the experiment, amplicons of the gene were analyzed with electrophoresis in a subset of clones from the final timepoint (Figure 4). In the YJM311 populations, five out of eight experimental populations ended with an approximate 1000 bp length increase in some or all clones, while none of the control clones showed an increase in length. It is unknown whether the change in length was due to independent de novo mutations or selection favoring an existing allele. The similar allelic length in multiple replicate populations favors the latter explanation. It is possible that during the generation of the starting recombinant pool, there was a mutation that was not detected in the subset of ancestral clones later chosen for analysis. In this genetic background, FLO11 length is not correlated with the strength of plastic adherence, nor with the other three multicellular phenotypes (Figure 4—figure supplement 1).

Figure 4.<italic>FLO11</italic> length evolution.

In 8 ancestral clones and 5 clones per replicate population at the final cycle, the full gene was amplified and run through a BioAnalyzer to determine its length. Amplicons of this length have an accuracy of ±100 bp. x-axis: Anc refers to ancestor, A to asexual populations, and S to sexual populations, while a-d denote replicates; C refers to control populations, and a and s refer to the asexual and sexual controls, respectively. (A) In YJM311, it appears there were two major length alleles, with the possibility of derived variants with smaller changes in length. (B) In YJM128, it appears there were also two alleles, separated by ~500 bp. Clones from the final timepoint show variation in length. Correlations between length and different adherence phenotypes can be found in Figure 4—figure supplements 1 and 2.

Figure 4—figure supplement 1.<italic>FLO11</italic> Length by Phenotype for the YJM311 Background.

FLO11 length measurements, point shapes and point colors as in Figure 4. Each panel plots the allele length for a clone with one of its phenotypic measurements.

Figure 4—figure supplement 2.<italic>FLO11</italic> length by phenotype for the YJM128 background.

FLO11 length measurements, point shapes and point colors as in Figure 3. Each panel plots the allele length for a clone with one of its phenotypic measurements.

In YJM128, the ancestral pool likely had two alleles separated by a few hundred basepairs. The ending asexual populations appeared to have these alleles, with one much longer allele in a clone in one replicate. The ending sexual populations contained the ancestral alleles, as well as other variants both longer and shorter. Clones from one replicate could not be amplified (Sa), suggesting the possibility of a mutation in the region where the primers anneal. Again, FLO11 length was not correlated with the strength of plastic adherence, nor with the other three multicellular phenotypes (Figure 4—figure supplement 2).

Thus, while FLO11 length evolved during the experiment, it does not appear to be the cause of the correlated response to selection on adherence. However, this does not rule out the possibility that FLO11 plays a role. It is possible that expression of the gene, through its complex regulatory network (Rupp et al., 1999; Octavio et al., 2009; Bumgarner et al., 2009), is related to the phenotypic response to adherence selection.

Virulence

To test the coincidental selection-accidental virulence hypothesis, we sought to determine if the evolved changes had an effect on virulence, and were particularly interested in the unexpected evolution of hyper-multicellular clones. Virulence was measured using larvae of the greater wax moth, Galleria mellonella, an invertebrate model used to study microbial pathogenesis and virulence (Pereira et al., 2018), including in S. cerevisiae (Phadke et al., 2018). Using the phenotyping data and the PCA results as a guide, for each genetic background, we identified six hyper-multicellular clones and six non-multicellular clones (Figure 5, Supplementary file 1a, b). We attempted to identify evolved clones that excelled in all measured aggregative traits; therefore, not all were from the final time point, but all were from late in the experiment. In choosing the non-multicellular strains, clones were taken from a variety of time points, including the ancestor, control, and experimental populations at different time points. This allowed us to verify that it was the evolved hyper-multicellular phenotype and not just long-term growth in the evolution medium.

Figure 5.Virulence of Evolved Populations.

G. mellonella survival curves for strains highlighted in Figure 3; each strain was injected into 200 larvae for YJM311-derived clones or 180 larvae for YJM128-derived clones. Points represent Kaplan-Meier estimates. (A) Survival curves with confidence limits for non- and hyper-multicellular treatments; black triangles represents the control treatment injected with sterile water. (B) Survival for individual strains along with associated CCM images. The low multicellular curves are labeled for ease of identification. Strains used in survival analyses can be found in Supplementary file 1a and b.

Strains were grown in the medium in which they were evolved, then washed, adjusted for density, and injected into larvae. Larval survival and pupation were monitored for the next 7 days. Different batches of larvae can be variable in their response to microbial insult; therefore, to ensure reproducibility of results, the experiment was repeated multiple times with numerous batches of larvae, with each batch being challenged by all strains from a genetic background. For the clones derived from YM311, of the 2400 larvae injected with yeast, 1809 did not survive through day 7. Larvae injected with a hyper-multicellular strain were 1.28 times more likely to die than those injected with a non-multicellular strain (mixed effects Cox model: coefficient = 0.249, coeff. s.e. = 0.068, p<0.001) (Figure 5). For the clones derived from YM128, of the 2160 larvae injected with yeast, 1113 did not survive through day 7. Larvae injected with a hyper-multicellular strain were 1.29 times more likely to die than those injected with a non-multicellular strain (coefficient = 0.251, coeff. s.e. = 0.117, p=0.032) (Figure 5). When considering the survival curves of individual strains, non-multicellular strains were less virulent regardless of whether they were from ancestral, control or experimental populations.

Thus, the evolutionary changes brought about by selection for adherence to a plastic bead, led to the incidental evolution of increased virulence.

Discussion

Our results demonstrate that selection on one yeast trait can generate a correlated response in other traits— a common feature of organismal evolution (Price and Langen, 1992)— but here, the correlated traits may have included those associated with virulence. In this experiment, favoring the ability to adhere to plastic, a surface that is common in industrial, medical, and domestic settings (Geyer et al., 2017), led to a suite of aggregative phenotypes and increased virulence.

The coincidental selection-accidental virulence hypothesis proposes that selection for survival in harsh conditions may lead to traits that predispose microbes to virulence. However, harsh environmental conditions can also favor traits that favor the collective, or multicellular phenotypes (Tong et al., 2022), so it is perhaps not surprising that other forms of multicellularity increased throughout this experiment. Furthermore, in its long evolutionary history, S. cerevisiae has evolved the genetic capability to express multiple different multicellular phenotypes, most of which are induced in nutrient limiting conditions; the experiment presented here was performed in such conditions (glucose limited medium). Other experiments with yeast growing in nutrient limiting conditions have resulted in the unintended evolution of single aggregative behaviors (Hope et al., 2017). In this light, the correlated phenotypic response in this experiment is not entirely unexpected. However, the evolution of multiple multicellular phenotypes in two independent genetic backgrounds was not anticipated. Furthermore, nutrient limiting conditions may have affected PSH in our experiment, but it was not the main driver of the other multicellular phenotypes; rather, it was adherence selection that led to the evolution of hyper-multicellularity.

Previous research on a panel of environmental isolates found no correlation between the phenotypes assayed here (Hope and Dunham, 2014), and in a tractable lab strain capable of aggregative behaviors, each phenotype was associated with its own set of genes (Ryan et al., 2012). However, there is overlap in the requirement of FLO11 and its regulators. Despite the phenotypes being induced by different nutrient signals, there are numerous conserved signaling pathways contributing to filamentous, multicellular growth of all forms (e.g. cAMP-PKA, TOR, filamentous MAPK, Rim101) (Cullen and Sprague, 2012; Granek et al., 2011). It is well known that genetic background and genetic architecture can have strong effects on the expression and correlation of traits (Gasch et al., 2016). In the case of the filamentous phenotypes assayed here, a genetic background that contains variants in the main signaling pathways may lead to a correlation of the phenotypes, while variants expressed later in the development of the phenotype, that are specific to a single trait, may not lead to such a correlation. The effect of the different types of genetic variants suggests that some strains and genetic backgrounds are more likely to evolve virulence from selection in the open environment.

The strains used in this experiment each contain ~50,000 heterozygous sites and differ from each other by ~25,000 SNPs. It is possible that they contained genetic variation in canonical signaling pathways, allowing for the evolution of hyper-multicellular strains. Interestingly, in both backgrounds, pseudohyphal growth appeared to evolve independently of the other phenotypes. Future research will investigate the sorting of the genetic variation, as well as the new mutations, that led to the observed phenotypic evolution in these populations.

Understanding the processes that lead to the emergence of opportunistic fungal pathogens is of increasing importance. In 2022, the World Health Organization issued its first-ever report prioritizing 19 fungal pathogens for research and public health awareness (World Health Organization, 2022); of these, 11 are known to live in the environment (i.e. soil, wood, etc.), including three in the highest priority group (Cryptococcus neoformans, Candida auris, Aspergillus fumigatus). In our experiment, it is unclear which trait was associated with increased virulence: plastic adherence, a different multicellular trait, or general hyper-multicellularity. Regardless of the specific trait causing increased virulence, the experiment demonstrated that selection for a dual-use trait in an environment that is entirely devoid of host organisms can still inadvertently lead to virulence and pathogenicity. The experiment also demonstrated the role that sex can have in increasing rates of adaptation in fungi.

Whether or not increased virulence caused by adherence selection is a general result in fungal microbes remains to be seen. In bacteria, the results are mixed. In Burkholderia cenocepacia, plastic bead selection led to an increase in biofilm phenotypes and mutations previously associated with chronic infections (Poltak and Cooper, 2011; Traverse et al., 2013). Yet, bead selection in P. aeruginosa led to a decrease in biofilm-related phenotypes. It also led to an increase in antibiotic resistance, thus mimicking changes seen in chronic infections (Azimi et al., 2020).

As humans continue to generate novel ecological niches at an unprecedented rate by encroaching on more habitats, using plastics unreservedly (Iroegbu et al., 2021), and especially, as the global climate changes (IPCC Core Writing Team, 2023), the potential for unintended selection grows (Casadevall, 2020). Clinically relevant strains of Escherichia coli can use microplastics in the environment as a reservoir and can even become more virulent after recovery from the ‘plastisphere’ (Ormsby et al., 2023), and it has been shown that plastics in the environment may harbor other pathogenic taxa (Wu et al., 2020). Warmer water temperatures and climate disruptions have been linked to the incidence of illness caused by the marine bacterium Vibrio vulnificus (Martinez-Urtaza et al., 2010), and an increase in global temperature has been hypothesized to be related to the simultaneous emergence of C. auris infections on multiple continents (Jackson et al., 2019; Megha et al., 2020). More generally, the narrowing of the gap between mammalian body temperatures and the ambient environment may create opportunities for fungi to exploit new host niches (Garcia-Solache and Casadevall, 2010). Thus, as new selective pressures act on populations with existing abundant genetic variation, there is the opportunity to coincidentally select a new generation of accidental pathogens.

Materials and methods

Key resources table

Reagent type (species) or resource	Designation	Source or reference	Identifiers	Additional information
Gene (Saccharomyces cerevisiae)	FLO11	Saccharomyces Genome Database	YIR019C
Strain, strain background (Saccharomyces cerevisiae)	YJM311	Gift from Paul Magwene at Duke
Strain, strain background (Saccharomyces cerevisiae)	YJM128	Gift from Paul Magwene at Duke
Genetic reagent (Saccharomyces cerevisiae)	HMY7	This paper		YJM311 homozygous for PGK1 tagged with mCherry-KanMX
Genetic reagent (Saccharomyces cerevisiae)	HMY355	This paper		YJM128 homozygous for PGK1 tagged with mCherry-HygMX
Biological sample (Galleria mellonella)	Galleria mellonella larvae	Vanderhorst Wholesale Inc.		https://www.waxworms.net/
Recombinant DNA reagent	pBS34	Addgene	RRID: Addgene_83796	Source of mCherry for tagging original strains
Sequence-based reagent	FLO11-for	This paper	PCR primers	GCCTCAAAA ATCCATATA CGCACACTA TG
Sequence-based reagent	FLO11-rev	This paper	PCR primers	TTAGAATAC AACTGGAAG AGCGAGTAG
Commercial assay or kit	MasterPure Yeast DNA Purification Kit	Lucigen	Cat #: MPY80200	DNA extraction kit
Commercial assay or kit	Agilent DNA 7500 kit	Agilent	Cat #: 5067–1506	PCR product length analysis kit
Other	7 mm polystyrene beads	American Educational Products	Product #: 3276	Plastic beads used in the evolution experiment (Figure 1A)

Strains

To generate strains appropriate for downstream phenotyping assays, the original diploid isolates were engineered to express a fluorescence protein by fusing mCherry to the C-terminal region of the highly expressed PGK1 gene, generating HMY7 (YJM311 PGK1-mCherry-KanMX) (Deschaine et al., 2018) and HMY355 (YJM128 PGK1-mCherry-HygMX). After being subject to selection for 8–9 cycles, clones with different multicellular phenotypes were isolated from each replicate population. Original isolates were generously provided by Paul Magwene (Duke University).

Media

Experimental populations were grown in Evolution Medium (EM; 0.17% yeast nitrogen base without ammonium sulfate and without amino acids, 0.1% glutamic acid, 0.1% dextrose) supplemented with G418 (200 μg/ml) or Hygromycin B (300 μg/ml). Cells were sporulated on solid medium (1% potassium acetate, 2% agar) and digested using an overnight zymolyase- β-glucuronidase procedure (Goddard et al., 2005; Granek et al., 2013). Phenotypes were assayed on YPD (1% yeast extract, 2% peptone, 2% dextrose, 2% agar), low dextrose (LD) YPD (0.1% dextrose), 2 X SLAD (0.34% yeast nitrogen base without ammonium sulfate and without amino acids, 2% dextrose, 50 μmol ammonium sulfate, 2% agar), or in liquid SD (0.17% yeast nitrogen base without amino acids and with ammonium sulfate, 2% dextrose).

Experimental evolution

HMY7 and HMY355 were grown in 10 ml YPD, sporulated, digested, grown to saturation in 10 ml EM, and used to inoculate 10 replicate populations: 4 sexual, 4 asexual, and 1 control of each reproductive type.

Experimental and control populations derived from YJM311 were evolved for 8 12-day cycles, for a total of ~350 generations; populations from YJM128 were evolved for 9 cycles, for a total of ~400 generations. In each cycle, experimental populations were grown in 10 ml of EM in a glass tube containing a sterile 7 mm polystyrene bead (American Education Products), population size ~2 x 10⁸. After 48 hr at 30 °C in a rotator drum, the bead was removed with sterile disposable forceps, washed twice, suspended in 500 µl of sterile H₂O in a microcentrifuge tube, and gently sonicated (UP200St with VialTweeter, Heischler Ultrasound Technology) to detach cells from the bead. The cell suspension was used to inoculate the next 10 ml EM tube. The number of cells on the bead varied over the experiment. Control populations were also grown in 10 ml of EM in a glass tube, but without the presence of a bead. Instead, 10 μl of culture were used to inoculate the next tube, which was approximately the same number of cells as that being transferred from bead adherence in the experimental populations at the start of the experiment. After 4 serial transfers, asexual populations were refrigerated and sexual populations were sporulated for 48 hr. Asci were digested overnight, and the spores resuspended in 1 ml of EM to allow germination and mating (population size ~10⁵ spores). Finally, the refrigerated cultures and the mated spores were used to begin the next 12-day cycle.

Population phenotyping

To estimate adherence evolution, all populations from all cycles were assayed using the same batch of medium. 10 ml EM cultures were inoculated with cryopreserved glycerol stocks and grown for 48 hr. From these, two replicate test tubes were inoculated with two beads in each, for a total of four beads per population per time point. The cultures were grown and the beads processed as in the experimental cycle; cell counts were made using a hemocytometer with the sonicated cell suspension. This entire process was repeated a second time, for a total of 8 beads per population per cycle for YJM311 populations.

Clonal phenotyping

Twenty clones were isolated from the ancestral population and 10 clones were isolated from each replicate population at four cycle timepoints: 2, 4, 6, 8, for YJM311, and 1, 3, 6, 9, for YJM128. The clonal strains were arrayed in a 96-well format and cryopreserved. To assay social phenotypes, saturated YPD cultures were resuspended and pinned to different media using a 96-pin multi-blot replicator (V&P Scientific no. VP408FP6), wrapped in parafilm, and incubated at 30 °C.

Plastic adherence

Clones were grown in 200 μl EM for 48 hr in three replicate black, clear-bottom, non-treated 96-well plates. Optical density was measured, then culture was removed, and plates were gently washed with water three times and dried upside down for 1 hr. Fluorescence readings were taken with a Spectramax M2e (Molecular Devices) and used as a proxy for the number of cells that remained attached to the wells. To account for differences in growth, each fluorescence reading was divided by the optical density of the well.

Flor formation

Clones were grown in 200 μl SD for 5 days and imaged on an Olympus SZX16 dissecting scope. Flor formation was scored using the scale in Figure 2.

Complex colony morphology (CCM)

Clones were pinned to 3 replicate LD omni trays, incubated for 7 days, and imaged on an EPSON Expression 11000 XL scanner. Colonies were scored for complexity using the scale in Figure 2.

Pseudohyphal growth

Clones were pinned to three replicate 2 X SLAD omni trays, incubated for 8 days, and scanned. Images were processed using a custom script that determined the percentage of colony pixels comprising the pseudohyphae (Lenhart et al., 2019).

Data analysis

Bead cell count data from experimental populations were log-transformed and analyzed using a mixed effects linear model in R (R Development Core Team, 2020) with the lme4 package (Bates et al., 2015). Replicate population within treatments (control, asexual, sexual) was considered a random effect. Because all populations were begun from a single ancestral pool, the intercept was set as the mean value of the ancestor and not allowed to vary among treatments. Therefore, the only fixed effect was the interaction between cycle and treatment, which tested the differences among the slopes of the three treatments. The analysis was performed on the number of cells counted on a hemocytometer, which ranged from zero to a few hundred cells in the later cycles. Thus, the coefficients represent the effect on the number of cells per cycle on these counts. Clonal data were analyzed similarly, with the untransformed average score of a phenotype as the independent variable. Thus, the coefficients represent the effect of the treatment over time on the measurement of the phenotype. Finally, the average phenotyping data for each clone were combined for a principal components analysis in R using the princomp function. Figures were produced using ggplot2 (Wickham, 2016).

Clones for virulence assay

The virulence assay was first conducted with clones from the YJM311 background. Clones were selected to reflect changes that occurred over the course of the entire experiment, specifically the evolution of hyper-multicellularity in experimental populations. Therefore, the low-multicellularity clones were chosen from the ancestral segregants, as well as early and ending control populations. The high-multicellularity clones were chosen from the ending experimental populations. Because CCM was correlated with plastic adherence and flor formation, it was used as an initial screen to find clones of interest (either smooth or very complex colonies); data on the other phenotypes were then investigated. Once a subset of clones was chosen, they were re-assayed to verify the phenotypes and used in virulence assays. In the second background, YJM128, clones were chosen in an attempt to isolate the effect of hyper-multicellularity, thus, other than ancestral segregants, all clones were chosen from the experimental populations using the same process. Since the high-multicellularity clones should have exhibited some level of all of the phenotypes (when possible) and PSH was evolving independently, the chosen clones do not stand out as the most extreme on the PCA plot.

Virulence assay

10 ml EM cultures of evolved and ancestral strains Supplementary file 1a were grown for 48 hr, washed and resuspended in sterile water to a concentration of 10⁹ cells/ml based on hemocytometer counts. 4 μl of culture or control water was injected into the final posterior proleg of Galleria mellonella larvae (Vanderhorst Wholesale Inc, https://www.waxworms.net) weighing on average 180 (±20) mg using a Hamilton PB600-1 Repeating Dispenser with a 27-gauge needle. Each strain was injected into 20 larvae on the same day using the same shipment of G. mellonella; 20 control larvae were injected at the start of the assay and at the end. The same assay was repeated the next day with the same shipment of larvae, for a total of 40 larvae/strain/shipment. Multiple shipments were used for the virulence measurements, for a total of 200 larvae per strain for YJM311-derived strains and 180 larvae per strain for YJM128-derived strains. After injection, larvae were incubated at 30 °C and survival was monitored for 7 days; larvae that turned black and no longer responded to tactile stimulation were considered dead and removed from the population, as were larvae beginning to pupate.

Data were analyzed with a mixed effects Cox model using the coxme package (Therneau, 2020) in R (R Development Core Team, 2020). Death was recorded as the day larvae were removed from the population; larvae were censored if removed for pupation. The model included treatment (high vs. low multicellular) as a fixed effect, and strain and larval batch as random effects.

FLO11 length

Of the clones assayed for multicellular phenotypes, 8 ancestral clones and 5 clones from the final time point of each replicate population were chosen for length analysis. Genomic DNA was extracted using the MasterPure Yeast DNA Purification Kit (Lucigen). FLO11 was amplified with Phusion polymerase (New England BioLabs) and primers targeting the entire gene (forward: GCC TCA AAA ATC CAT ATA CGC ACA CTA TG, reverse: TTA GAA TAC AAC TGG AAG AGC GAG TAG). Cycle conditions followed manufacturers recommendations and included a melting temperature of 58 °C and 3 min extension time. Gene length was estimated by running PCR amplicons through the Agilent 2100 BioAnalyzer using the Agilent DNA 7500 kit (as in ref [Oppler et al., 2019]).

Additional information

Competing interests

No competing interests declared

Author contributions

Data curation, Investigation, Methodology, Writing – review and editing, Conducted phenotyping assays; conducted all virulence assays

Data curation, Investigation, Methodology, Evolved all the populations and conducted phenotypic assays

Investigation, Conducted phenotyping assays

Methodology, Developed the evolution protocol

Investigation, Conducted FLO11 length analysis

Conceptualization, Data curation, Formal analysis, Funding acquisition, Writing - original draft, Project administration, Writing – review and editing

Additional files

MDAR checklist

Supplementary file 1.This file contains tables that list strains used for injections and full results from linear models.

(a) YJM311 Strains used in virulence experiments. (b) YJM128 Strains used in virulence experiments. (c) Results of the mixed-effect linear model for whole-population cell count data over experimental cycles. (d) Results of the mixed-effect linear model for clonal plastic adherence data. (e) Results of the mixed-effect linear model for clonal CCM data. (f) Results of the mixed-effect linear model for clonal flor data. (g) Results of the mixed-effect linear model for clonal PSH data.

Data availability

All data generated or analysed during this study are included in the manuscript and supporting file. Source data files have been provided for Figures 1, 2, and 5. Original and evolved strains are available upon request to the corresponding author.

Acknowledgements

We thank Paul Magwene for strains, and Joseph Heitman and Anna Averette for guidance with G mellonella assays. The research was funded by National Institutes of Health grant R15GM122032 and National Science Foundation grant DEB-1839555 to HAM, and William & Mary Charles Center Summer Fellowships to LIE, DM, DVM, and JAS.

References

Adiba

Nizak

van Baalen

Denamur

Depaulis

2010

From grazing resistance to pathogenesis: the coincidental evolution of virulence factors

PLOS ONE5

e11882

10.1371/journal.pone.0011882

20711443

Albuquerque

Nicola

Magnabosco

DAG

Derengowski

Crisóstomo

Xavier

LCG

de Frazão

Guilhelmelli

de Oliveira

Dias

Hurtado

Teixeira

Guimarães

Paes

Bagagli

Felipe

MSS

Casadevall

Silva-Pereira

2019

A hidden battle in the dirt: Soil amoebae interactions with Paracoccidioides spp

PLOS Neglected Tropical Diseases13

e0007742

10.1371/journal.pntd.0007742

31589617

Arora

Singh

Wang

Yadav

Pawar

Singh

Padmavati

Chowdhary

Cowen

2021

Environmental isolation of Candida auris from the coastal wetlands of Andaman Islands, India

mBio12

e03181

10.1128/mBio.03181-20

Aucott

Fayen

Grossnicklas

Morrissey

Lederman

Salata

1990

Invasive infection with Saccharomyces cerevisiae: report of three cases and review

Reviews of Infectious Diseases12406411

10.1093/clinids/12.3.406

2193348

Azimi

Roberts

AEL

Peng

Weitz

McNally

Brown

Diggle

2020

Allelic polymorphism shapes community function in evolving Pseudomonas aeruginosa populations

The ISME Journal1419291942

10.1038/s41396-020-0652-0

32341475

Bates

Mächler

Bolker

Walker

2015

Fitting Linear Mixed-Effects Models Using lme4

Journal of Statistical Software67

i01

10.18637/jss.v067.i01

Baumgardner

Laundre

2001

Studies on the molecular ecology of Blastomyces dermatitidis

Mycopathologia1525158

10.1023/a:1012438029997

11761145

Brown

Cornforth

Mideo

2012

Evolution of virulence in opportunistic pathogens: generalism, plasticity, and control

Trends in Microbiology20336342

10.1016/j.tim.2012.04.005

22564248

Brüssow

2007

Bacteria between protists and phages: from antipredation strategies to the evolution of pathogenicity

Molecular Microbiology65583589

10.1111/j.1365-2958.2007.05826.x

17608793

Bumgarner

Dowell

Grisafi

Gifford

Fink

2009

Toggle involving cis-interfering noncoding RNAs controls variegated gene expression in yeast

PNAS1061832118326

10.1073/pnas.0909641106

19805129

Casadevall

Steenbergen

Nosanchuk

2003

“Ready made” virulence and “dual use” virulence factors in pathogenic environmental fungi--the Cryptococcus neoformans paradigm

Current Opinion in Microbiology6332337

10.1016/s1369-5274(03)00082-1

12941400

Casadevall

Pirofski

2007

Accidental virulence, cryptic pathogenesis, martians, lost hosts, and the pathogenicity of environmental microbes

Eukaryotic Cell621692174

10.1128/EC.00308-07

Casadevall

Guimaraes

Albuquerque

2019

The ‘Amoeboid Predator-Fungal Animal Virulence’ Hypothesis

Journal of Fungi5

10.3390/jof5010010

Casadevall

2020

Climate change brings the specter of new infectious diseases

The Journal of Clinical Investigation130553555

10.1172/JCI135003

31904588

Casalone

Barberio

Cappellini

Polsinelli

2005

Characterization of Saccharomyces cerevisiae natural populations for pseudohyphal growth and colony morphology

Research in Microbiology156191200

10.1016/j.resmic.2004.09.008

15748984

Clemons

McCusker

Davis

Stevens

1994

Comparative pathogenesis of clinical and nonclinical isolates of Saccharomyces cerevisiae

The Journal of Infectious Diseases169859867

10.1093/infdis/169.4.859

8133102

Cullen

Sprague

2012

The regulation of filamentous growth in yeast

Genetics1902349

10.1534/genetics.111.127456

22219507

David-Vaizant

Alexandre

2018

Flor yeast diversity and dynamics in biologically aged wines

Frontiers in Microbiology9

2235

10.3389/fmicb.2018.02235

30319565

Davies

Winstanley

Fothergill

James

2016

The role of temperate bacteriophages in bacterial infection

FEMS Microbiology Letters363

fnw015

10.1093/femsle/fnw015

26825679

Deschaine

Heysel

Lenhart

Murphy

2018

Biofilm formation and toxin production provide a fitness advantage in mixed colonies of environmental yeast isolates

Ecology and Evolution855415550

10.1002/ece3.4082

29938072

Douglas

2003

Candida biofilms and their role in infection

Trends in Microbiology113036

10.1016/S0966-842X(02)00002-1

Dranginis

Rauceo

Coronado

Lipke

2007

A Biochemical Guide to Yeast Adhesins: Glycoproteins for Social and Antisocial Occasions

Microbiology and Molecular Biology Reviews71282294

10.1128/MMBR.00037-06

Enache-Angoulvant

Hennequin

2005

Invasive Saccharomyces infection: a comprehensive review

Clinical Infectious Diseases4115591568

10.1086/497832

16267727

Erken

Lutz

McDougald

2013

The rise of pathogens: predation as a factor driving the evolution of human pathogens in the environment

Microbial Ecology65860868

10.1007/s00248-013-0189-0

23354181

Fanning

Mitchell

2012

Fungal biofilms

PLOS Pathogens8

e1002585

10.1371/journal.ppat.1002585

22496639

Fidalgo

Barrales

Ibeas

Jimenez

2006

Adaptive evolution by mutations in the FLO11 gene

PNAS1031122811233

10.1073/pnas.0601713103

16844788

Fidalgo

Barrales

Jimenez

2008

Coding repeat instability in the FLO11 gene of Saccharomyces yeasts

Yeast25879889

10.1002/yea.1642

19160455

Fraser

Levy

Chavan

Shah

Perez

Zhou

Siegal

Sinha

2012

Polygenic cis -regulatory adaptation in the evolution of yeast pathogenicity

Genome Research2219301939

10.1101/gr.134080.111

Friman

Hiltunen

Jalasvuori

Lindstedt

Laanto

Örmälä

Laakso

Mappes

Bamford

JKH

2011

High temperature and bacteriophages can indirectly select for bacterial pathogenicity in environmental reservoirs

PLOS ONE6

e17651

10.1371/journal.pone.0017651

21423610

Friman

Buckling

2014

Phages can constrain protist predation-driven attenuation of Pseudomonas aeruginosa virulence in multienemy communities

The ISME Journal818201830

10.1038/ismej.2014.40

24671085

Liporagi-Lopes

dos Santos

Tenor

Perfect

Cuomo

Casadevall

Olszewski

Wormley

2021

Amoeba predation of cryptococcus neoformans results in pleiotropic changes to traits associated with virulence

mBio12

e00567

10.1128/mBio.00567-21

Garcia-Solache

Casadevall

2010

Global warming will bring new fungal diseases for mammals

mBio1

e00061-10

10.1128/mBio.00061-10

20689745

Gasch

Payseur

Pool

2016

The power of natural variation for model organism biology

Trends in Genetics32147154

10.1016/j.tig.2015.12.003

26777596

Geyer

Jambeck

Law

2017

Production, use, and fate of all plastics ever made

Science Advances3

e1700782

10.1126/sciadv.1700782

Gimeno

Ljungdahl

Styles

Fink

1992

Unipolar cell divisions in the yeast S. cerevisiae lead to filamentous growth: regulation by starvation and RAS

Cell6810771090

10.1016/0092-8674(92)90079-r

1547504

Goddard

Godfray

HCJ

Burt

2005

Sex increases the efficacy of natural selection in experimental yeast populations

Nature434636640

10.1038/nature03405

15800622

Granek

Magwene

2010

Environmental and genetic determinants of colony morphology in yeast

PLOS Genetics6

e1000823

10.1371/journal.pgen.1000823

20107600

Granek

Kayıkçı

Magwene

2011

Pleiotropic signaling pathways orchestrate yeast development

Current Opinion in Microbiology14676681

10.1016/j.mib.2011.09.004

21962291

Granek

Murray

Kayrkçi

Magwene

2013

The genetic architecture of biofilm formation in a clinical isolate of Saccharomyces cerevisiae

Genetics193587600

10.1534/genetics.112.142067

23172850

Gupta

Srivastava

Osmanoglu

Brakhage

Dandekar

2021

Aspergillus fumigatus versus genus aspergillus: conservation, adaptive evolution and specific virulence genes

Microorganisms9

2014

10.3390/microorganisms9102014

Hall-Stoodley

Costerton

Stoodley

2004

Bacterial biofilms: from the natural environment to infectious diseases

Nature Reviews. Microbiology295108

10.1038/nrmicro821

15040259

Hennequin

Kauffmann-Lacroix

Jobert

Viard

Ricour

Jacquemin

Berche

2000

Possible role of catheters in Saccharomyces boulardii Fungemia

European Journal of Clinical Microbiology & Infectious Diseases191620

10.1007/s100960050003

Hillmann

Novohradská

Mattern

Forberger

Heinekamp

Westermann

Winckler

Brakhage

2015

Virulence determinants of the human pathogenic fungus Aspergillus fumigatus protect against soil amoeba predation

Environmental Microbiology1728582869

10.1111/1462-2920.12808

25684622

Hope

Dunham

2014

Ploidy-Regulated Variation in Biofilm-Related Phenotypes in Natural Isolates of Saccharomyces cerevisiae

G3: Genes, Genomes, Genetics417731786

10.1534/g3.114.013250

Hope

Amorosi

Miller

Dang

Heil

Dunham

2017

Experimental Evolution Reveals Favored Adaptive Routes to Cell Aggregation in Yeast

Genetics20611531167

10.1534/genetics.116.198895

28450459

Hosseinidoust

van de Ven

TGM

Tufenkji

2013

Evolution of Pseudomonas aeruginosa Virulence as a result of phage predation

Applied and Environmental Microbiology7961106116

10.1128/AEM.01421-13

IPCC Core Writing Team

2023

Climate change 2023: synthesis report. A report of the intergovernmental panel on climate change

Contribution of Working Groups I, II and III to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change

Iroegbu

AOC

Ray

Mbarane

Bordado

Sardinha

2021

Plastic pollution: a perspective on matters arising: challenges and opportunities

ACS Omega61934319355

10.1021/acsomega.1c02760

34368521

Jackson

Chow

Forsberg

Litvintseva

Lockhart

Welsh

Vallabhaneni

Chiller

2019

On the origins of a species: what might explain the rise of candida auris?

Journal of Fungi5

10.3390/jof5030058

Kaltz

Bell

2002

The ecology and genetics of fitness in chlamydomonas: Xii. Repeated sexual episodes increase rates of adaptation to novel environments

Evolution; International Journal of Organic Evolution5617431753

10.1111/j.0014-3820.2002.tb00188.x

12389719

Klein

2000

Molecular basis of pathogenicity in Blastomyces dermatitidis: the importance of adhesion

Current Opinion in Microbiology3339343

10.1016/S1369-5274(00)00100-4

Kosheleva

Desai

2018

Recombination alters the dynamics of adaptation on standing variation in laboratory yeast populations

Molecular Biology and Evolution35180201

10.1093/molbev/msx278

29069452

Kraushaar

Brückner

Veelders

Rhinow

Schreiner

Birke

Pagenstecher

Mösch

Essen

2015

Interactions by the Fungal Flo11 Adhesin Depend on a Fibronectin Type III-like Adhesin Domain Girdled by Aromatic Bands

Structure2310051017

10.1016/j.str.2015.03.021

25960408

Kuthan

Devaux

Janderová

Slaninová

Jacq

Palková

2003

Domestication of wild Saccharomyces cerevisiae is accompanied by changes in gene expression and colony morphology

Molecular Microbiology47745754

10.1046/j.1365-2958.2003.03332.x

12535073

Lachapelle

Bell

2012

Evolutionary rescue of sexual and asexual populations in a deteriorating environment

Evolution; International Journal of Organic Evolution6635083518

10.1111/j.1558-5646.2012.01697.x

23106714

Legras

J-L

Moreno-Garcia

Zara

Garcia-Martinez

Mauricio

Mannazzu

Coi

Bou Zeidan

Dequin

Moreno

Budroni

2016

Flor yeast: new perspectives beyond wine aging

Frontiers in Microbiology7

503

10.3389/fmicb.2016.00503

27148192

Lengeler

Davidson

D’souza

Harashima

Shen

W-C

Wang

Pan

Waugh

Heitman

2000

Signal transduction cascades regulating fungal development and virulence

Microbiology and Molecular Biology Reviews64746785

10.1128/MMBR.64.4.746-785.2000

Lenhart

Meeks

Murphy

2019

Variation in filamentous growth and response to quorum-sensing compounds in environmental isolates of Saccharomyces cerevisiae

G3: Genes, Genomes, Genetics915331544

10.1534/g3.119.400080

Leong

Poh

Williams

Lutz

Hoque

Poh

Yee

BYK

Chua

Givskov

Sanderson-Smith

Rice

McDougald

Nikel

2022

Adaptation to an amoeba host leads to Pseudomonas aeruginosa isolates with attenuated virulence

Applied and Environmental Microbiology88

e02322

10.1128/aem.02322-21

Levin

Svanborg Edén

1990

Selection and evolution of virulence in bacteria: an ecumenical excursion and modest suggestion

Parasitology100 SupplS103S115

10.1017/s0031182000073054

2235060

Llopis

Hernández-Haro

Monteoliva

Querol

Molina

Fernández-Espinar

2014

Pathogenic potential of Saccharomyces strains isolated from dietary supplements

PLOS ONE9

e98094

10.1371/journal.pone.0098094

24879417

Dranginis

1998

The Cell Surface Flocculin Flo11 Is Required for Pseudohyphae Formation and Invasion by Saccharomyces cerevisiae

Molecular Biology of the Cell9161171

10.1091/mbc.9.1.161

Magwene

Kayıkçı

Granek

Reininga

Scholl

Murray

2011

Outcrossing, mitotic recombination, and life-history trade-offs shape genome evolution in Saccharomyces cerevisiae

PNAS10819871992

10.1073/pnas.1012544108

21245305

Mahenthiralingam

Baldwin

Dowson

2008

Burkholderia cepacia complex bacteria: opportunistic pathogens with important natural biology

Journal of Applied Microbiology10415391551

10.1111/j.1365-2672.2007.03706.x

18217926

Maršíková

Wilkinson

Hlaváček

Gilfillan

Mizeranschi

Hughes

Begany

Rešetárová

Váchová

Palková

2017

Metabolic differentiation of surface and invasive cells of yeast colony biofilms revealed by gene expression profiling

BMC Genomics18

814

10.1186/s12864-017-4214-4

29061122

Martinez-Urtaza

Bowers

Trinanes

DePaola

2010

Climate anomalies and the increasing risk of Vibrio parahaemolyticus and Vibrio vulnificus illnesses

Food Research International4317801790

10.1016/j.foodres.2010.04.001

McCusker

Clemons

Stevens

Davis

1994a

Saccharomyces cerevisiae virulence phenotype as determined with CD-1 mice is associated with the ability to grow at 42 degrees C and form pseudohyphae

Infection and Immunity6254475455

10.1128/iai.62.12.5447-5455.1994

7960125

McCusker

Clemons

Stevens

Davis

1994b

Genetic characterization of pathogenic Saccharomyces cerevisiae isolates

Genetics13612611269

10.1093/genetics/136.4.1261

8013903

McDonald

Rice

Desai

2016

Sex speeds adaptation by altering the dynamics of molecular evolution

Nature531233236

10.1038/nature17143

26909573

Megha

Arunaloke

Bernhard

2020

On the origin of Candida auris: Ancestor

Environmental Stresses, and Antiseptics. mBio11

e02102

10.1128/mBio.02102-20

Mikonranta

Friman

Laakso

2012

Life history trade-offs and relaxed selection can decrease bacterial virulence in environmental reservoirs

PLOS ONE7

e43801

10.1371/journal.pone.0043801

22937098

Muñoz

Bouza

Cuenca-Estrella

Eiros

Pérez

Sánchez-Somolinos

Rincón

Hortal

Peláez

2005

Saccharomyces cerevisiae fungemia: an emerging infectious disease

Clinical Infectious Diseases4016251634

10.1086/429916

15889360

Octavio

Gedeon

Maheshri

Akhtar

2009

Epigenetic and Conventional Regulation Is Distributed among Activators of FLO11 Allowing Tuning of Population-Level Heterogeneity in Its Expression

PLOS Genetics5

e1000673

10.1371/journal.pgen.1000673

Oppler

Parrish

Murphy

2019

Variation at an adhesin locus suggests the possibility of self-discrimination in the yeast Saccharomyces cerevisiae

Proceedings of the Royal Society B-Biological Sciences 286

20191948

10.1098/rspb.2019.1948

Ormsby

White

Metcalf

Oliver

Quilliam

2023

Clinically important E. coli strains can persist, and retain their pathogenicity, on environmental plastic and fabric waste

Environmental Pollution326

121466

10.1016/j.envpol.2023.121466

36958655

Palecek

Parikh

Kron

2002

Sensing, signalling and integrating physical processes during Saccharomyces cerevisiae invasive and filamentous growth

Microbiology148893907

10.1099/00221287-148-4-893

11932437

Pereira

de Barros

Fugisaki

Rossoni

Ribeiro

de Menezes

Junqueira

Scorzoni

2018

Recent advances in the use of galleria mellonella model to study immune responses against human pathogens

Journal of Fungi4

128

10.3390/jof4040128

Pérez-Torrado

Querol

2015

Opportunistic strains of Saccharomyces cerevisiae: a potential risk sold in food products

Frontiers in Microbiology6

1522

10.3389/fmicb.2015.01522

26779173

Peter

De Chiara

Friedrich

Yue

J-X

Pflieger

Bergström

Sigwalt

Barre

Freel

Llored

Cruaud

Labadie

Aury

J-M

Istace

Lebrigand

Barbry

Engelen

Lemainque

Wincker

Liti

Schacherer

2018

Genome evolution across 1,011 Saccharomyces cerevisiae isolates

Nature556339344

10.1038/s41586-018-0030-5

29643504

Phadke

Maclean

Zhao

Mueller

Michelotti

Norman

Kumar

James

2018

Genome-wide screen for Saccharomyces cerevisiae genes contributing to opportunistic pathogenicity in an invertebrate model host

G3: Genes, Genomes, Genetics86378

10.1534/g3.117.300245

Poltak

Cooper

2011

Ecological succession in long-term experimentally evolved biofilms produces synergistic communities

The ISME Journal5369378

10.1038/ismej.2010.136

Price

Langen

1992

Evolution of correlated characters

Trends in Ecology & Evolution7307310

10.1016/0169-5347(92)90229-5

R Development Core Team

2020

R: A language and environment for statistical computing

Vienna, Austria

R Foundation for Statistical Computing

http://www.r-project.org

Reynolds

Fink

2001

Bakers’ yeast, a model for fungal biofilm formation

Science291878881

10.1126/science.291.5505.878

11157168

Rupp

Summers

Madhani

Fink

1999

MAP kinase and cAMP filamentation signaling pathways converge on the unusually large promoter of the yeast FLO11 gene

The EMBO Journal1812571269

10.1093/emboj/18.5.1257

Ryan

Shapiro

Kurat

Mayhew

Baryshnikova

Chin

Lin

Cox

Vizeacoumar

Cheung

Bahr

Tsui

Tebbji

Sellam

Istel

Schwarzmüller

Reynolds

Kuchler

Gifford

Whiteway

Giaever

Nislow

Costanzo

Gingras

Mitra

Andrews

Fink

Cowen

Boone

2012

Global gene deletion analysis exploring yeast filamentous growth

Science33713531356

10.1126/science.1224339

Selezska

Kazmierczak

Müsken

Garbe

Schobert

Häussler

Wiehlmann

Rohde

Sikorski

2012

Pseudomonas aeruginosa population structure revisited under environmental focus: impact of water quality and phage pressure

Environmental Microbiology1419521967

10.1111/j.1462-2920.2012.02719.x

22390474

Siscar-Lewin

Hube

Brunke

2022

Emergence and evolution of virulence in human pathogenic fungi

Trends in Microbiology30693704

10.1016/j.tim.2021.12.013

35058122

Steenbergen

Shuman

Casadevall

2001

Cryptococcus neoformans interactions with amoebae suggest an explanation for its virulence and intracellular pathogenic strategy in macrophages

PNAS981524515250

10.1073/pnas.261418798

Steenbergen

Nosanchuk

Malliaris

Casadevall

2003

Cryptococcus neoformans virulence is enhanced after growth in the genetically malleable host Dictyostelium discoideum

Infection and Immunity7148624872

10.1128/IAI.71.9.4862-4872.2003

12933827

St’ovíček

Váchová

Kuthan

Palková

2010

General factors important for the formation of structured biofilm-like yeast colonies

Fungal Genetics and Biology4710121022

10.1016/j.fgb.2010.08.005

20728557

Št’ovíček

Váchová

Begany

Wilkinson

Palková

2014

Global changes in gene expression associated with phenotypic switching of wild yeast

BMC Genomics15

136

10.1186/1471-2164-15-136

24533484

Sun

Noorian

McDougald

2018

Dual Role of Mechanisms Involved in Resistance to Predation by Protozoa and Virulence to Humans

Frontiers in Microbiology9

1017

10.3389/fmicb.2018.01017

29867902

Tawfik

Papasian

Dixon

Potter

1989

Saccharomyces cerevisiae pneumonia in a patient with acquired immune deficiency syndrome

Journal of Clinical Microbiology2716891691

10.1128/jcm.27.7.1689-1691.1989

2671026

Therneau

2020CoxmeMixed Effects Cox Models

Tong

Bozdag

Ratcliff

2022

Selective drivers of simple multicellularity

Current Opinion in Microbiology67

102141

10.1016/j.mib.2022.102141

35247708

Traverse

Mayo-Smith

Poltak

Cooper

2013

Tangled bank of experimentally evolved Burkholderia biofilms reflects selection during chronic infections

PNAS110E250E259

10.1073/pnas.1207025110

Váchová

Stovícek

Hlavácek

Chernyavskiy

Stĕpánek

Kubínová

Palková

2011

Flo11p, drug efflux pumps, and the extracellular matrix cooperate to form biofilm yeast colonies

The Journal of Cell Biology194679687

10.1083/jcb.201103129

21875945

Verstrepen

Jansen

Lewitter

Fink

2005

Intragenic tandem repeats generate functional variability

Nature Genetics37986990

10.1038/ng1618

16086015

Verstrepen

Klis

2006

Flocculation, adhesion and biofilm formation in yeasts

Molecular Microbiology60515

10.1111/j.1365-2958.2006.05072.x

16556216

West

Diggle

Buckling

Gardner

Griffin

2007

The social lives of microbes

Annual Review of Ecology, Evolution, and Systematics385377

10.1146/annurev.ecolsys.38.091206.095740

Wickham

2016ggplot2: Elegant Graphics for Data AnalysisCham

Springer-Verlag

10.1007/978-3-319-24277-4

World Health Organization

2022WHO Fungal Priority Pathogens List to Guide Research, Development and Public Health ActionWHO

Zhang

Zhao

Niu

2020

Colonization characteristics of bacterial communities on microplastics compared with ambient environments (water and sediment) in Haihe Estuary

The Science of the Total Environment708

134876

10.1016/j.scitotenv.2019.134876

31740062

Zara

Bakalinsky

Zara

Pirino

Demontis

Budroni

2005

FLO11-based model for air-liquid interfacial biofilm formation by Saccharomyces cerevisiae

Applied and Environmental Microbiology7129342939

10.1128/AEM.71.6.2934-2939.2005

15932987

Zara

Pinna

Marceddu

Budroni

2009

FLO11 gene length and transcriptional level affect biofilm-forming ability of wild flor strains of Saccharomyces cerevisiae

Microbiology15538383846

10.1099/mic.0.028738-0

19729408

Zeyl

Bell

1997

The advantage of sex in evolving yeast populations

Nature388465468

10.1038/41312

9242403

10.7554/eLife.81056.sa0

Editor's evaluation

Harper

Diane M

Reviewing Editor

https://ror.org/00jmfr291

University of Michigan

United States

Using experimental evolution and virulence assessment in a model system, this valuable study examines how yeast virulence can coincidentally evolve following selection for plastic adherence. The strength of evidence is solid. The work presents interesting experimental systems, and the findings will be of interest to investigators in the field of experimental evolution and evolution of fungal pathogens.

10.7554/eLife.81056.sa1

Decision letter

Harper

Diane M

Reviewing Editor

https://ror.org/00jmfr291

University of Michigan

United States

Smith

Daniel

ReviewerJohns HopkinsUnited States

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Selection on plastic adherence leads to hyper-multicellular strains and incidental virulence in the budding yeast" for consideration by eLife. Your article has been reviewed by 2 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Arturo Casadevall as the Senior Editor. The following individual involved in the review of your submission has agreed to reveal their identity: Daniel Smith (Reviewer #1).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

The reviewers have provided detailed comments and made recommendations for improving the paper. Furthermore Reviewers 1 and 2 had discussions among themselves and with me regarding what would constitute a successful revision and here is a distillation of that discussion, which is in addition to the Reviewer comments below.

– As noted by Reviewer it is very important to address the relationship to previous work on 'coincidental selection for virulence' and place your findings in that context.

– You must address the concerns about how the virulence assays were conducted. For example, It was not clear to Reviewer 1 why the paper lacked a simple 'ancestor' versus 'evolved with beads' versus 'evolved without beads' comparison. The concern here is that pooling across generations and stratifying by a correlated phenotype could introduce biases that confound the question of phenotype under observation.

– There was concern about the pooling of the data. Although the unpooled data was in the supplemental data this should be on the main figures. Pooling is a problem because the isolates come from different cycles and conditions, and the readers should be able to see which are the ancestral/control and late-cycle experimental isolates. Showing that enhanced virulence phenotype is present in 10-12 of the hyper multicellular isolates makes their data stronger than showing them pooled as one.

– The paper needs to do a better job in describing the selection of the 20 low hyper multicellular strains and the 20 high hyper multicellular strains and include the Supplemental Tables 1 and 2 in the main text. This could help the reader see which strains were used and also show that the low hyper multicellular strains are generally control or ancestral isolates. They might also be able to write which strain is which in the virulence graph keys.

Please know that I will be sending this paper back to the reviewers if a revised version is submitted to eLife. Hence, I encourage you to address all the issues raised by the review. Personally, I think this is a very interesting paper and I hope that it can be revised such that it would be acceptable to the journal. The comments/criticisms are extensive but if addressed will result in a much stronger paper.

Reviewer #1 (Recommendations for the authors):

Overall, I found this paper very interesting, and I think the experiments were thorough and well-designed, and I am very impressed with the volume of the G. mellonella infections. It is amazing how much the yeast changed after the 8/9 cycles of selection for adhesion to plastic. I think the paper does a good job describing the multicellular (and hyper-multicellular) phenotypes that arose from the adhesion to plastic. I also found it interesting that the PSH evolved independently of plastic adhesion and was likely due to nutrient deprivation in the media. Below are some suggested edits and changes you could make to the presentation of the data that could help effectively show the massive amount of work done.

Questions:

1. How many of the bead-adhering cells get removed during the sonication and washing process? Are there strongly adherent cells remaining? If so, would you then be selecting for medium-plastic rather than strong adherence?

2. Have you looked at the cell surface hydrophobicity of some of these strains? A MATH assay (Microbial Adhesion to Hydrocarbon) is relatively straightforward to try, although it might be fickle and highly variable if the baseline hydrophobicity is low. I would expect the plastic-selected ones to have higher hydrophobicity which is sometimes associated with enhanced virulence.

3. Line 114: you mention the possibility of fluorescence evolution over time. Although it seems unlikely (and you give good reasons why), if you wanted to verify could you measure the mean fluorescence intensity of an individual cell? (i.e. take an image under the microscope and measure the mean gray value of the individual cells using an image processing software like ImageJ/FIJI. A mean gray value of 0 is black/no fluorescence signal and 255 would be white/brightest fluorescent signal. It would be expected that the fluorescence of each cycle would be equal.)

4. Do you think the phenotypes are reversible? Genetics is discussed a lot, but there is a possibility that this is epigenetics-based.

Data Presentation:

1. In my opinion, there is a lot of great data visualization that is put in the supplementary figures that could be included in the main figures/table. Namely, I think including a version of Table S1 and S2, Figure S6/7, Figure S8/9 (as part of Figure 4), and Figure S10 (instead of the current Figure 3B) in the main manuscript would be helpful.

2. Figure 1 may benefit from including Figure S1 as the first panel. I think it might help the reader better understand the evolution schema. Figure S1 could also be labelled directly on the illustration to improve understanding. Figure 1 should also have a scale bar in the fluorescence microscope mage in Panel A.

3. The presentation of the linear mixed-effect model data on lines 95-101, 112-114, 137-139, 151-152, 172-177 is unclear and took a while to understand. It is not clear what the units are for these data, which would be important to know and better understand the results. Including the units and removing the formula and including that information in the methods it would help the clarity. Could you compare these counts in any other way such as averaging the values of the sexual and asexual groups instead at each cycle and do an ANOVA comparing to the control?

4. Connecting the dots of the average values over the cycles in Figure 2 would help illustrate the trends better.

5. Can you correlate the fungal virulence with the quantified phenotypes of the individual strains (Similar to Figure S6 and S7) using either % mortality, pathogenic potential (PP or PPT), or median survival time as measures? I think that maybe correlating virulence to a specific multicellular phenotype show if one, in particular, is responsible for the increased virulence (as mentioned in Line 328-332).

6. Similarly, have you looked at if the longer FLO11 genotype correlated with the virulence?

7. If space permits, for Figure 3, write "high multicellularity" and "low multicellularity" in the key instead of "high" and "low" for better clarity.

8. In Figure 4, it might be easier to understand the X axis if instead of (a-d) it would be 1-4, and/or there was a key in the figure itself rather than just in the figure legend. I am also not sure if "Cs" is a typo or if I cannot find what the lowercase "s" stands for. I also think the data in Figure S8/9 is really interesting and would be nice to include in the main Figure 4.

General suggestions:

1. In general, I think the introduction would benefit from including more background information and elaborating on some of the points that I mention below.

2. It might be helpful to add more examples of predominately saprophytic/environmental microbes that also happen to be pathogens (lines 10,11).

3. The sentence in Line 20 about C. auris in the environment seems to be missing the part of the sentence describing the environmental pressure (salinity? Temperature?) that theoretically drove its adaptation into being a good pathogen.

4. An additional sentence or two after Line 22 discussing how adhesion, biofilm formation, and hydrophobicity are involved in virulence would be helpful (i.e. helping microbes stick to tissue and invade), and additional examples of microbes where adhesion is important like Candida albicans.

5. While there are reports of S. cerevisiae infection, they are relatively rare and the fungus is better known as a very useful model for fungal biology in the lab. Elaborating on the environmental and research role of S. cerevisiae would be good in the introductory paragraph (beginning line 28).

6. Is there a specific reason why the strains are referred to as YJM311 and YJM128 in the results and discussion, but in the methods are referred to as HMY7 and HMY355 (the strain named after being engineered to be drug-resistant and express mCherry)?

7. Since the virulence data is in Figure 3, it should be brought up in the results before Figure 4 (FLO11 length), or the figures can be combined/rearranged accordingly.

8. I think you are underselling your results in your description lines 299-305 because, besides PSH, the control conditions did not show increased multicellular phenotypes in the same way that the selected cultures did.

9. Might help to put a standalone sentence about the dramatic increase in plastic/microplastic, hydrocarbon, and metal pollution in the environment following the sentence from 334-336.

10. Number and unit should be separated by a space, for example, 10ml should be 10 ml, 200µl should be 200 µl, etc. Mostly in the methods section text.

Reviewer #2 (Recommendations for the authors):

I managed to misread 'plastic evolution' several times, attempting to interpret it as 'evolution of plasticity' (a topic that does intersect with virulence, see eg ref [1]). To avoid others making a similar mistake, perhaps rephrase it as 'evolution on plastic' or similar?

The FLO gene length analysis left me wondering why not look at sequence or structure? Looking only at gene length seemed a bit of a blunt instrument to get at adhesion behavior, and the lack of association doesn't tell us much (to be fair, this point is made by the authors).

Turning to analysis choices, the authors focus on analyzing changes in time, which is entirely reasonable (e.g. 'does bead attachment increase/decrease thru evolutionary time'?). But this leaves open whether the evolutionary trajectories end up in significantly different places – so I would suggest analyzing differences across endpoints (eg for evolved with beads versus evolved without beads).

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Selection on plastic adherence leads to hyper-multicellular strains and incidental virulence in the budding yeast" for further consideration by eLife. Your revised article has been evaluated by a Senior Editor and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

Reviewer #1 (Recommendations for the authors):

Thank you to the authors for revising the manuscript and including as many of the edits and changes that were feasible, especially given the circumstances of life. I think the revised manuscript is in good shape, and I don't think any of the not performed suggested experiments were necessary for the manuscript or for the final outcome of the project. One strong suggestion is that the Figure 1 supplemental with the graphical abstract can be better labelled directly on the figure so there are labels and descriptions without having to read the figure legend.

Reviewer #2 (Recommendations for the authors):

In this revised manuscript, the authors have addressed several of my prior concerns, leading to clear improvements. Overall the paper is generally well written, and I appreciate the thoughtful and frank rebuttals. Yet the responses to issues 2-4 (especially 3) still leave me with ongoing concerns.

1. Integration with prior literature on 'co-incidental selection'.

The authors do a very nice job in connecting the literature threads on 'co-incidental selection' and 'accidental virulence'. This will help the field.

2. Experimental evolution controls.

The authors clarify that Figure 1 bead attachment data is from a separate phenotyping experiment under standard bead conditions, run after the experimental evolution was completed. This clarifies my primary concern. I suggest the authors add a sentence to spell this out in the figure legend, to avoid similar confusion. I also note that the specific design of the control experiment is not specified in the methods. Given there was no bead, what exactly was the passaging protocol? (I expect it was some standard volume transfer). How comparable is this design to the bead design, in terms of approx. number of cells transferred? Substantial differences in the passaging bottleneck could raise issues in the interpretations of comparisons to the control.

3. Sample pooling for virulence assays.

The authors provide an honest summary of their rationale for their experimental design, but little in the way of changes to the MS. As things stand, I still have significant concerns over the current MS. The specific rationale for individual strain selection into their 'low/high' groups is quite mysterious, which further reduces confidence. Line 344: 'using PCA results as a guide,..' – what does this mean? The selected strains are not close to extreme positions in the PCA (see PCA lower panels). At a minimum, the authors need to present clear and unambiguous criteria for strain selection. Without this, I am concerned that the contrasts don't even meet the authors goal of directly filtering on high versus low multicellularity phenotypes. Even with a more specific, unambiguous set of criteria for strain selection, my previous concerns remain. By selecting the virulence assay on phenotypes (seemingly blind to passage time or treatment), we weaken the ability to directly assess the impact of different experimental treatments on evolutionary outcomes.

4. FLO gene length typing is a blunt instrument.

The authors agree, they have sequence data but are holding it back for another paper. The current data provides very little insight into the molecular mechanisms underlying the phenotypic differences reported.

In summary, the MS is improved, and it provides useful data that will benefit the field. Yet my earlier concerns are not fully addressed. Addressing concerns 2-4 would help increase the rigor and repeatability of the research.

10.7554/eLife.81056.sa2

Author response

Essential revisions:

– As noted by Reviewer it is very important to address the relationship to previous work on 'coincidental selection for virulence' and place your findings in that context.

We wholeheartedly agree with this criticism. Our previous literature search and review focused exclusively on the fungal literature, and was missing this important context. We had no intention of ignoring previous work and apologize for the omission. To the best of our ability, given the space constraints, we have updated the Introduction to include this hypothesis and the subsequent work testing it. We also expanded this section to include more review of the fungal literature, as it is comparable to that which we highlight from the bacterial literature. This review of the literature has even generated new ideas for our own research, and we are grateful for the urging of the reviewers.

This is a reasonable criticism and to be honest, one we struggled with when we were choosing the strains to assay for virulence. As noted in the paper, the evolution of hyper-multicellular strains was unexpected and surprising, and we were therefore focused on the effect of this trait (or suite of traits) on virulence. Hyper-multicellularity was only observed in populations evolved with a bead, and not in the ancestral clones or those evolved without a bead.

For the first genetic background that we assayed (YJM311), for the low-multicellular treatment, we exclusively chose ancestors and clones from populations evolved without beads (cycle 2 and cycle 8). The results from this design (or strain choice) would allow us to say that the virulence increased in populations evolved with beads (and was not due to evolution in the medium alone). When it came time to choose strains/clones from the second background (YJM128), for the low-multicellular treatment, we again chose a few ancestors, but were also interested in finding low-multicellular clones from populations evolved with a bead. This would allow us to say that clones evolved in the same conditions, but with different resulting multicellular phenotypes, had different patterns of virulence. As can be seen by the clones we chose, most of those strains came from earlier in the experiment (cycles 1 and 3). For the same reason, we also chose hyper-multicellular strains from a timepoint earlier than the last one (cycle 6). We were really trying to get at whether it was the hyper-multicellularity or evolution with a bead, but were limited to what phenotypes evolved at different timepoints in the experiment.

Given the number of larvae required to assay the virulence of each clone, we could not reasonably assay more strains than we did. For better or worse, we made the choices described above. In our first genetic background, we could in principle make the “simple” comparison of ancestor vs. evolved with bead, and ancestor vs. evolved without bead, but the number of strains in the different categories would be unequal. Instead, we choose to focus on the phenotype, which was what we were interested in.

To be completely transparent and for ease of understanding, we have labeled all the low multicellular survival curves in the figure (Figure 5), which should help the readers navigate the results.

As noted above, we have updated the figures to include the unpooled data. We have clearly labeled the individual curves and included images of the clonal phenotypes.

We have done our best to comply with this request (see above).

We appreciate the encouragement and hope the revised version is acceptable.

Below we respond to the suggestions from each reviewer, focusing on the ones that were not included in the summary above.

Reviewer #1 (Recommendations for the authors):

Questions:

We haven’t investigated this systematically. We have occasionally looked at the beads under the microscope to verify that the cells have been removed. It is certainly possible that after long-term selection, there could be cells that adhere so strongly that sonication does not remove them and they would not be included in the inoculum for the next generation. At this point, we have no evidence for that occurring.

This is an interesting suggestion and one we may use for future research, as we are very interesting in the evolved multicellular changes. But in our opinion, this is out of the scope of the current manuscript, as it doesn’t directly address the idea of virulence evolution.

This is also an interesting suggestion, but since the plastic adherence assays show the expected patterns (increase over the cycles); thus, in our opinion, this extra assay is unlikely to contribute significantly to the manuscript.

4. Do you think the phenotypes are reversible? Genetics is discussed a lot, but there is a possibility that this is epigenetics-based.

While it is possible that there is an epigenetic component, we believe the majority of the phenotype is genetic. Given the abundance of evidence that there is a genetic basis for each of the traits assayed here (mapping studies, etc.) and the extensive amount of genetic variation in each clinical isolate (~40-50K heterozygous sites), it is reasonable to surmise that the sorting of this genetic variation is leading to the gradual evolution of increase in multicellularity throughout the experiment.

Data Presentation:

We thank the reviewer for the compliment and have taken the suggestion to update Figure 4 (now Figure 5) by including the data from Figure S10 and Table S1. For the other sets of supplementary figures, we intend to take advantage of the eLife format and link these supplementary figures to the main figures in the text. Specifically, when Figure 3 is viewed (the PCA), there will be links to Figures S6/7 to show the individual correlations. Similarly, when Figure 4 (FLO11 evolution) is viewed, there will hyperlinks to Figures S8/9. In this way, the supplementary figures will be viewed in the context in which they are discussed in the manuscript. This is also true for Figure 1 with a hyperlink to S1.

We thank the reviewer for pointing out the issue of interpreting the coefficients. We have updated the methods to clarify, and added language to the results when the linear model is first introduced.

4. Connecting the dots of the average values over the cycles in Figure 2 would help illustrate the trends better.

We have taken this suggestion and made this edit.

6. Similarly, have you looked at if the longer FLO11 genotype correlated with the virulence?

At the reviewer’s suggestion, we have analyzed the virulence data using a Cox survival analysis and extracted the hazard ratio for each strain. Not surprisingly, each of the multicellular phenotypes was correlated with this measure of virulence. The reason is that the two categories of strains were chosen to have either all or none of the multicellular phenotypes, and as noted in the manuscript, one category had a higher virulence than the other (i.e., high and low multicellularity). Thus, we cannot point to one particular phenotype as being responsible. FLO11 was not correlated with virulence.

7. If space permits, for Figure 3, write "high multicellularity" and "low multicellularity" in the key instead of "high" and "low" for better clarity.

The updated figure does not have the space, but the figure legend clarifies the groups.

We thank the reviewer for pointing this out; the legend has been updated to clarify the nomenclature. As noted above, Figures S8-9 will be hyperlinked to the figure in the main text.

General suggestions:

1. In general, I think the introduction would benefit from including more background information and elaborating on some of the points that I mention below.

2. It might be helpful to add more examples of predominately saprophytic/environmental microbes that also happen to be pathogens (lines 10,11).

7. Since the virulence data is in Figure 3, it should be brought up in the results before Figure 4 (FLO11 length), or the figures can be combined/rearranged accordingly.

9. Might help to put a standalone sentence about the dramatic increase in plastic/microplastic, hydrocarbon, and metal pollution in the environment following the sentence from 334-336.

10. Number and unit should be separated by a space, for example, 10ml should be 10 ml, 200µl should be 200 µl, etc. Mostly in the methods section text.

All of the above suggestions have been very useful and have been adopted, and the manuscript has been updated in the appropriate locations.

Reviewer #2 (Recommendations for the authors):

We have removed references to plastic evolution.

The sequence analysis of these populations will be presented in a different manuscript. We agree that a length analysis is a blunt tool, but length variation is highly variable/mutates quickly and has been shown to be associated with adherence levels—it has even been suggested as a tool for quick evolution (Verstrepen et al., 2005. Nature Genetics). Thus, we investigated the evolution of length variation.

[Editors’ note: what follows is the authors’ response to the second round of review.]

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

Reviewer #1 (Recommendations for the authors):

We have added text to the figure to supplement the legend. We believe it is clarifying and hope the reviewer agrees.

Reviewer #2 (Recommendations for the authors):

1. Integration with prior literature on 'co-incidental selection'.

The authors do a very nice job in connecting the literature threads on 'co-incidental selection' and 'accidental virulence'. This will help the field.

Thank you. As noted in our previous response, exploration of this literature has generated new research ideas for our lab, so we are grateful for being pointed to it.

2. Experimental evolution controls.

Based on this suggestion, we have added language that makes it explicit that the data from the experiment and the phenotyping are different (see Figure 1 legend).

As for the transfers, thank you for pointing this out. We should have included this detail in the manuscript, and have since added text to the methods (lines 420-423). “Control populations were also grown in 10 ml of EM in a glass tube, but without the presence of a bead. Instead, 10 ul of culture were used to inoculate the next tube, which was approximately the same number of cells being transferred from bead adherence in the experimental populations at the start of the experiment.”

However, to be clear, we cannot guarantee that the number of cells being transferred was comparable throughout the entirety of the experiment. As experimental populations became more adherent, more cells would have been transferred, and as control populations adapted to the medium, more would have been transferred. This is often the case in microbial evolution experiments when passage is by volume. At the start of the experiment, bottlenecks from passaging are consistent among treatments, but may diverge among replicates as evolution proceeds.

3. Sample pooling for virulence assays.

We appreciate and respect the reviewer’s concern about the clones we selected for the virulence assay. As noted in the previous response, we cannot now change the clones we selected, so we are left with only one option, which is to be as transparent as possible. To that end, we previously updated Figure 5 to include details about the clones that were chosen. We have now also added text to the Methods to provide details about our selection logic and criteria (lines 481-495). Unfortunately, there is nothing more we can do to address this concern. We leave the ultimate fate of the manuscript in the hands of the reviewers and editors.

For the sake of total transparency, below we provide details about our process from our lab notes. As best we could, we sought to find clones that had none of the phenotypes in the low strains and all of the multicellular phenotypes in the high strains, and also came from various specified timepoints. This ended up being more of an art than a science, which we describe below, in part because PSH evolved independently from the other phenotypes. As a reminder, we first assayed YJM311, so we chose high strains from the end of the experiment and low strains from the control populations and ancestors. When we next assayed YJM128, we were interested in isolating hyper-multicellularity as a trait, so chose low and high strains from experimental populations throughout (as well as ancestors).

The process was as follows:

1) In order to avoid sampling the same clonal background twice, we aimed to choose only one clone from a replicate population at a given timepoint.

2) Since we were ultimately interested in hyper-multicellularity, we focused our initial selection criteria more on the phenotypes that appeared to be a correlated response than on plastic adherence itself (evolved populations from later timepoints would not have survived the experiment if they were not adhering). We knew from the PCA results that CCM (complex colony morphology) strongly correlated with flor formation and plastic adherence, and PSH was evolving independently from the other phenotypes. We therefore used CCM for the first step in the screen, as colony morphology was a clear, visible phenotype with high replicability among replicate plates. Thus, for low strains, we sought smooth colonies and for high strains, we sought highly complex colony morphology (and when possible, seemingly different architectures).

3) Low Strains:

For the low strains, we aimed to find clones that had none of the multicellular phenotypes, or at the very least muted versions of them. For example, there were 20 ancestral clones from each genetic background; most of these had some form of PSH, CCM and plastic adherence. In YJM311, the ancestral clones had no flor formation, while in YJM128, they did. So for the ancestral clones, we chose the ones with the lowest scores on the combination of the three phenotypes that existed in our panel, otherwise, we would have had no ancestral clones.

For the other low strains, we began by screening the CCM plates for the smoothest colonies in the timepoints/treatments we were interested in. For YJM311, which was the first genetic background we conducted virulence assays for, we looked only for clones from control populations (early and late in the experiment). After finding the smoothest colonies from different replicates, we looked at the scores on PSH and flor formation. It was often not possible to have everything in the later timepoints: sometimes for the lowest CCM, there was slight PSH or flor formation (e.g., C7s-8-8). We therefore also looked for the lowest PSH scores, and then investigated the CCM and flor scores (e.g., C7a-8-6).

For YJM128, we attempted to get low strains from experimental populations. The procedure was similar: visibly scan the CCM plates for smoothness to begin the process, then investigate the other phenotypes. It turned out that low strains were not available later in the experiment.

4) High Strains

For the high strains, we aimed to get clones that had multiple, ideally all, of the multicellular phenotypes. We began by scanning the CCM plates for complex colony morphology (and when possible, different morphological architectures). We then investigated flor scores, plastic adherence and PSH of these clones. We were trying to balance finding clones with abilities in all phenotypes, not just the most extreme version of a single one.

For YJM 311, we were only interested in the last time point and were able to find clones from different replicate lines. In this genetic background, PSH decreased over the experiment, so we had to find clones with high CCM, PSH ability that was as least as strong as the average ancestral ability, and some flor formation.

In YJM 128, we were interested in obtaining clones from throughout the experiment; we followed a similar procedure. We began by screening CCM, then the other phenotypes. The one exception was (A8a-9-8), which had very high PSH and flor formation, but medium CCM.

After choosing the clones of interest, all strains were re-assayed to verify the scores they received in the high throughput assays.

To be clear, we stated that we used PCA results as a guide because these results gave us the insight of which traits were correlated, but it does not mean (nor did we mean to suggest) that we used the six clones with top scores on the first two PCs. We have updated the language in the manuscript. It is important to note that when looking at the PCA plot, PSH is antagonistic with the other multicellular phenotypes. Thus, the reasons we couldn’t just simply choose the most extreme points on the plot are: (1) these points do not necessarily exhibit all of the phenotypes of interest (because of the uncoupling of PSH and the others), (2) they may not have been from the correct treatment/timepoint, and (3) we were only selecting one clone from each replicate/timepoint and clones from a replicate/timepoint tended to cluster (see the supplemental figures for Figure 3).

We hope this explanation lends transparency to our process.

4. FLO gene length typing is a blunt instrument.

This manuscript is focused on the evolution of phenotypes and not molecular mechanisms. It is true that we have sequenced the replicate populations at multiple time points; however, there are ~50K SNPs being sorted in the sexual populations, and single clones fixing in the asexual populations (that differ by many SNPs, many of which likely have nothing to do with selection for plastic adherence). We are working through these messy data and have not uncovered an obvious candidate gene(s) responsible for the observed phenotypes. In fact, our QTL-mapping approach is yielding more fruitful genetic results for following up on potential alleles. We can assure the reviewer that if we had an exciting insight into the molecular mechanism, we would include it in the manuscript.

We have done our best to address concerns 2-4 and hope our responses are satisfactory.