Discussion
In this study, we applied comparative phylogenetic methods for exploring the evolution of fermentative capacity of yeasts, using compiled traits. To address this problem, we applied the lasso-OU algorithm, which is designed to detect adaptive shifts in phenotypic values provided a phylogeny and a dataset. According to this procedure, the three proxies of fermentative capacity that we considered (i.e., glycerol production, ethanol yield and respiratory quotient, collectively “physiological traits” hereafter) are consistent with a single evolutionary episode; a whole-genomic duplication (WGD) that occurred in the evolution of yeasts ca. 150 million years ago (Dashko et al., 2014). Thus, our results partially support this idea but it indicates that the differentiation between lineages occurred perhaps later, according to the combination of calibrated phylogenies and traits (i.e., phenograms, Fig 4) 75 million years ago. Our results are then different of what was obtained originally by the authors (Hagman et al., 2003). This is probably due to the phylogenetic comparison, which permits to account for the evolutionary distances among species when comparing trait values.
Gene duplications represent a typical way for increasing phenotypic capacities (Zhang, 2003). For the Saccharomycotina clade, recent evidence suggests that the mechanism of genomic duplication was interspecies hybridization, an episode that provided stability to the recently formed allopolyploid (Marcet-Houben & Gabaldon, 2015). In fact, it is accepted that the yeast WGD likely involved mating between two different ancestral species followed by a doubling of the genome to restore fertility. Then, the duplicated genes were retained either through neofunctionalization or sub functionalization in many genomes, increasing performance under nutrient competitive conditions (Scannell et al., 2007, Chen et al., 2008). In fact, compared with other genes, paralogs that were generated after the WGD in yeasts have long-lasting regulatory effects (Thompson et al., 2013, Chen et al., 2008). In addition, genome content doubling has been recurrently observed in laboratory evolution assays using haploid lines (Fisher et al., 2018, Gallone et al., 2016, Voordeckers et al., 2015). For example, it was demonstrated that WGD in haploids provides an immediate fitness gain at the expense of slowing subsequent adaptation in autodiploids, however this positive effect can be condition dependent (Fisher et al., 2018, Chen et al., 2008). In this context, in wine fermentation, the selective environment of several domesticated yeasts, the greater dosage of genes permits a rapid consumption of nutrients and a competitive displacement of other microorganisms (Querol et al., 2003, Gutierrez et al., 2016). Hence, results suggesting that yeasts’ phenotypic diversification in ethanol yield, ethanol production, glycerol production, and CO2 production was modulated by the WGD is interesting (Piskur, 2001, Conant & Wolfe, 2007). The WGD should have facilitated the specialization on the fermentative niche through gene duplication and retention, including post-transcriptional regulation, finally producing lineages with a selective increase of useful genes for fermentation and eliminating others by purifying selection (Wolfe & Shields, 1997). Furthermore, paralog duplicated genes tend to have a wider gene expression variation pattern than singleton genes, likely explained by cis -effects as a key adaptation for the organism to respond and adapt to fluctuating environment (Dong et al., 2011).
One of the most important adaptive features of post WGD species is the capacity to consume glucose rapidly, then depleting media from nutrients, and hampering respiration in other non-fermentative cells (Gutierrez et al., 2016, Hagman & Piskur, 2015). Glucose uptake rate and metabolism directly impacts CO2 production levels, which is determined by glucose hexose transporters (encoded byHXT genes) (Luyten et al., 2002). The HXT genes have been extensively amplified in fungal lineages that have independently evolved aerobic fermentation (such as S. cerevisiae and C. glabrata ), while a reduction of the number of HXT genes has been reported in aerobic respiratory species (such as K. lactis ) (Lin & Li, 2011), in agreement with our results. Interestingly, there is a cost: since the fermentation process allows cells to rapidly convert sugars to ethanol, this goes at the expense of decreasing biomass production (Dashko et al., 2014). However, we did not detect such costs on growth rate measurements (here measured as dry mass growth rate), which appeared undifferentiated across lineages.
In Crabtree positive species, pyruvate is preferentially converted into acetaldehyde and subsequently, ethanol. In these species glycerol is synthesized by the reduction of dihydroxyacetone phosphate followed by dephosphorylation catalysed by glycerol-3-phosphate dehydrogenase (GPD1 ) and glycerol-3-phosphatase (GPP1 ) (Albertyn et al., 1994). These two enzymes have duplicated genes, GPD2 andGPP2 , originated from gene retention and adaptive sub-functionalization after the WGD (Wolfe & Shields, 1997). Moreover, functional divergence of ADH1 and ADH2 , the latest only present in Crabtree positive yeasts, allowed increasing ethanol production and converting it to acetyl CoA for subsequent utilization in the TCA cycle (Thomson et al., 2005, Zhou et al., 2017). Then, the enhanced glycerol production we also observed in fermentative yeasts (Fig 3d) represents a secondary adaptation for osmotolerance, as a mean to compensate for the increased external osmotic pressure of the fermentative environment.
Unicellular and multicellular organisms share essential aspects of their design and function, because of the methods for characterizing them many conceptual issues developed in one realm, maybe do not apply to the other (see a critical discussion in Goddard & Grieg, 2015). Here we considered the application of comparative phylogenetic methods (a family of methods developed for multicellular organisms) to characterize phenotypic evolution in unicellular organisms. We found that the analysis produced informative results, suggesting that (above the reasonable doubt) the WGD has visible effects on the phenotypic diversification of fermentative yeasts, more than other genomic rearrangements that were not identified by this analysis. Although literature is scant regarding comparative analyses in microorganisms, a handful of authors have tested adaptive hypotheses considering phylogenetic relationships (Gubry-Rangin et al., 2015, Nakov et al., 2014, Starmer et al., 2003, Ernst et al., 2003, Ravot et al., 1996). For instance, Ravot et al. (1996) inferred adaptive patterns of hyperthermophilic bacteria, based on the production of L-alanine in some clades. Also in bacteria, Ernst et al. (2003) analyzed (putative) adaptive radiations of picocyanobacteria supposedly associate with the presence of major accessory pigments as key innovations. Working with fermentative yeasts, Starmer et al. (2003) concluded (qualitatively) convergent adaptive features for the cactus-yeast community. In a comprehensive analysis, Gubry-Rangin et al. (2015) associated the high rates of diversification observed in terrestrial Thaumarchaeota (Archaea) to acidic adaptation of their ancestor. Although these authors did not exactly apply trait-based comparative analyses, they were the firsts to link evolutionary diversification to environmental adaptation in a prokaryotic phylum. Here we show that laboratory experiments combined with a comparative approach, could give important results for testing a given evolutionary hypothesis in microorganisms. We encourage authors to explore this possibility for testing evolutionary hypotheses in other lineages.
Acknowledgements . This study was funded by Fondo Basal CAPES 0002-2014 to
Francisco Bozinovic. J.F.Q.-G. (N° 21160901) and J.J.S-I (Nº 21160530) thanks a CONICYT-fellowship. Doctorado Nacional Chile /2016. R.F.N. and F.C. are funded by MIISSB Iniciativa Científica Milenio-MINECON. Roberto Nespolo also thanks a FONDECYT grant number 1180917. We also thank Cletus Kurtzman for kindly providing us with advice for the reconstruction of the phylogenies.
Table 1. Traits, units and meaning of the measured variables. All species were grown at similar conditions (batch cultivation) of media and temperature (25ºC), and traits are presented in standardized units to biomass. Variables were measured at the moment of maximum growth rate. Extended and detailed methods, as well as the descriptive statistics of all the variables are provided in the original reference (Hagman et al., 2013).