Genomic signatures of innovation and selection in the extremotolerant yeast Kluyveromyces marxianus

Abstract

Extremophiles can be the product of millions of years of evolutionary engineering and refinement. The underlying mechanisms can be quite distinct from the ones operating at earlier stages of trait innovation. In this work, we have developed the compost yeast Kluyveromyces marxianus, which diverged from its closest relative >20 million years ago, as a model for interspecies comparative biology and genomics. We applied a battery of growth assays to species of the Kluyveromyces genus and found that K. marxianus outperformed its relatives in a battery of heat and chemical stress conditions. We then generated and analyzed genomes from across the genus, to find derived genetic features associated with, and potentially causal for, K. marxianus traits. We found robust expansions in gene families in the K. marxianus genome, most notably among genes annotated as transmembrane transporters and in metabolism. In molecular-evolution tests, we identified adaptive protein variants at hundreds of genes, among which plasma membrane transporters were over-represented. Together, these signals enable a model for the molecular mechanisms and evolutionary pressures underlying K. marxianus traits, including gains in transporter function mediating stress resistance, and metabolic variants contributing to its capacity for rapid growth in challenging conditions. Such oligogenic architectures may be the rule rather than the exception in phenotypes that have evolved over long timescales.

Figure 1.. K. marxianus shows a unique heat and chemical stress resistance phenotype.

A, Phylogenetic tree of the Kluyveromyces genus with marine and terrestrial species delineated (Am-In et al. 2008; Varela, Puricelli, Ortiz-Merino, et al. 2019). Branch lengths are not to scale. B, Each panel reports the results of a growth profiling experiment across the genus in the indicated condition. In a given panel, each trace reports a growth timecourse for the indicated species, with the solid line displaying the mean across technical replicates and the faint outline showing the standard error. OD, optical density.

Figure 2.. The unique heat tolerance of K. marxianus only manifests during its growing stage.

A, Experimental workflow created with BioRender.com. From left, inoculum from a starting liquid culture is introduced into fresh liquid medium for regrowth at 28°C and then sampled at nutrient exhaustion (Stationary) or in log phase (Exponential). The liquid sample is subjected to a heat shock of 47°C for 30 minutes and then spotted as serial dilutions onto solid medium. Viable cells after heat shock grow into colonies after incubation at 28°C. B-C, Each row reports viability after heat shock of a sample of the indicated species in the indicated liquid growth phase. See Figure S3 for unshocked controls.

Figure 3.. Overall regulatory volatility by K. lactis relative to K. marxianus and stress-evoked repression of glycolysis in both species.

A, Shown are results of principal component (PC) analysis of K. lactis and K. marxianus transcriptomes across treatments. Each point reports the values of the top two PCs for one replicate transcriptome of the indicated species subjected to the indicated stress treatment or a 28°C unstressed control. In a given axis label, the value in parentheses reports the proportion of variance across the set of transcriptomes explained by the indicated PC. B,C, Each panel reports the effect of stress on expression of genes of the indicated Gene Ontology term in K. marxianus and K. lactis. In a given panel, each cell reports, for the indicated gene, the log₂ fold-change between expression levels in the indicated stress treatment and an unstressed control.

Figure 4.. Gene family expansions in K. marxianus.

A, The color of each cell reports the change in copy number, in the indicated genome of a Kluyveromyces or related species relative to its parent node in the phylogeny, of a gene family (orthogroup, OG) with significant expansion or contraction in the K. marxianus genome as inferred with CAFE 5. Results for all analyzed orthogroups are reported in Table S4. B, Each panel reports the positions of genes in the K. marxianus genome in one indicated gene family from A with annotations as major facilitator superfamily (MFS) transporters. In a given panel, each row represents a chromosome in K. marxianus, and each arrow reports a gene of the respective family and its transcriptional direction (not to scale).

Figure 5.. Phylogenetic analysis reveals accelerated amino acid evolution at hundreds of genes in K. marxianus.

A, Each bar shows the number of genes with significant evidence for positive selection on the indicated lineage (adjusted p > 0.05) in a phylogenetic PAML branch-site test. B, The set of bars of each color reports the distribution, across the genes from A, of the number of sites per gene inferred by PAML to be targets of positive selection on the indicated lineage. C, Amino acid alignment of an example region of a gene, KLLA0B02607g (S. cerevisiae ortholog FAT3/YKL187C), with evidence for positive selection from the PAML test in A and from the McDonald-Kreitman test of D. Highlighted columns show inferred targets of positive selection from the PAML test in B. Highlighting indicates the amino acid’s BLOSUM 62 matrix score with respect to the consensus amino acid. D, Each row reports a gene with evidence for adaptive protein variation between K. marxianus and K. lactis according to the McDonald-Kreitman test (adjusted p < 0.05). NI, neutrality index. Genes highlighted in green also showed significant evidence for positive selection on the branch leading to K. marxianus in phylogenetic tests (PAML branch-site p < 0.05; Table S5). Not shown here are genes with p-value < 0.05 as a product of exceptionally high or low amino acid polymorphism; see Methods and Table S5. E, The x-axis reports mean NI across a group of genes. The blue trace reports results from a resampling test for enrichment of low NI across the genes of the transmembrane transport Gene Ontology term: the y-axis reports the frequency, among random sets of genes, of the mean NI given on the x, and the true mean NI of genes of the transmembrane transport term is indicated by the red dashed line, corresponding to corrected empirical p = 0.0132 (Table S5).