Metabolic remodeling and de novo mutations transcend cryptic variation as drivers of adaptation in yeast

Many organisms live in predictable environments with periodic variations in growth conditions. Adaptation to these conditions can lead to loss of nonessential functions, which could be maladaptive in new environments. Alternatively, living in a predictable environment can allow populations to accumulate cryptic genetic variation that may have no fitness benefit in that condition, but can facilitate adaptation to new environments. However, how these processes together shape the fitness of populations growing in predictable environments remains unclear. Through laboratory evolution experiments in yeast, we show that populations grown in a nutrient-rich environment for 1000 generations generally have reduced fitness and lower adaptability to novel stressful environments. These populations showed metabolic remodeling and increased lipid accumulation in rich medium which seemed to provide osmotic protection in salt stress. Subsequent adaptation to stressors was primarily driven by de novo mutations, with very little contribution from the mutations accumulated prior to the exposure. Thus, our work suggests that without exposure to new environments, populations might lose their ability to respond effectively to these environments.  Further, our findings highlight a major role of exaptation and de novo mutations in adaptation to new environments but do not reveal a significant contribution of cryptic variation in this process.

Methods
Cells were washed with PBS and resuspended in PBS. Data were collected using a BD LSR Fortessa Cell Analyzer machine in appropriate channels (FITC or PE-Texas Red or both). The R package ‘flowCore’ was used to read FCS files and to convert them to text format. The data points with FSC-A values more than 2000 and SSC-A values more than were considered yeast cells. The points that showed lower values than these thresholds represented cell debris or noise and hence, were discarded.  In addition, a filtering for cell aggregates was performed, since aggregates can generate stronger signals than single cells. To filter out cell aggregates, the ratio of FSC-A and FSC-H values was calculated, and data points showed two clusters. The data points with FSC-A/FSC-H values lower than 1.9 were considered as single cells and thus, were included in the quantification of the fluorescence signal.