Dynamic Changes in Yeast Phosphatase Families Allow for Specialization in Phosphate and Thiamine Starvation

Abstract

Convergent evolution is often due to selective pressures generating a similar phenotype. We observe relatively recent duplications in a spectrum of Saccharomycetaceae yeast species resulting in multiple phosphatases that are regulated by different nutrient conditions - thiamine and phosphate starvation. This specialization is both transcriptional and at the level of phosphatase substrate specificity. In Candida glabrata, loss of the ancestral phosphatase family was compensated by the co-option of a different histidine phosphatase family with three paralogs. Using RNA-seq and functional assays, we identify one of these paralogs, CgPMU3, as a thiamine phosphatase. We further determine that the 81% identical paralog CgPMU2 does not encode thiamine phosphatase activity; however, both are capable of cleaving the phosphatase substrate, 1-napthyl-phosphate. We functionally demonstrate that members of this family evolved novel enzymatic functions for phosphate and thiamine starvation, and are regulated transcriptionally by either nutrient condition, and observe similar trends in other yeast species. This independent, parallel evolution involving two different families of histidine phosphatases suggests that there were likely similar selective pressures on multiple yeast species to recycle thiamine and phosphate. In this work, we focused on duplication and specialization, but there is also repeated loss of phosphatases, indicating that the expansion and contraction of the phosphatase family is dynamic in many Ascomycetes. The dynamic evolution of the phosphatase gene families is perhaps just one example of how gene duplication, co-option, and transcriptional and functional specialization together allow species to adapt to their environment with existing genetic resources.

Keywords: Candida glabrata; acid phosphatase; gene duplication and loss; parallel evolution; phosphate metabolism; thiamine metabolism.

Figure 1

RNA-seq of thiamine-starved C. glabrata. (A) Normalized abundance of transcripts plotted for wild-type C. glabrata and a Cgthi3Δ strain that were grown for 4 h in thiamine starvation relative to cells grown in thiamine replete medium (strains grown in singlicate for this experiment). Transcripts that increased in abundance during starvation are visualized as circles that are above the majority of the genes. The seven most highly induced genes are CgTHI3-dependent. (B) A list of the most highly induced genes during thiamine starvation and their fold induction relative to the thiamine replete sample. Wild-type cells were grown for 2 h and 4 h to investigate the timing of induction; however, only the 4 h time point is presented in part (A).

Figure 2

CgPMU3 is tightly regulated by thiamine whereas CgPMU2 is regulated by both thiamine and phosphate. Plasmids containing the full-length promoter of either CgPMU2 or CgPMU3 driving expression of YFP were transformed into C. glabrata wild-type, thi3Δ, and pho4Δ strains and assayed for fluorescence using flow cytometry. Error bars represent the standard deviation of three biological replicates. We performed a Student t-test to confirm that CgPMU2 expression declined in the Cgthi3Δ strain; however, there is still a significant increase in expression relative to the high phosphate conditions. We additionally confirmed no statistical decline in the Cgpho4Δ strain during thiamine starvation in the CgPMU3p-YFP. The ** indicates a p value less than 0.001 and NS indicates a p value higher than 0.05. (Left) Strains were grown in thiamine replete (High) and starvation (No) conditions. Both CgPMU2 and CgPMU3 promoters are induced by thiamine starvation and this induction is regulated by CgThi3. (Right) Strains were grown in phosphate replete (High) and starvation (No) conditions. Only the CgPMU2 promoter is induced by phosphate starvation and this induction is regulated by CgPho4.

Figure 3

Repeated, relatively recent duplications of PHO5 family in the Ascomycetes. (A) and (B) Cartoon example for the two evolutionary scenarios: (A) gene duplication precedes speciation, resulting in the orthologs (1A and 2A, 1B and 2B) to cluster first before they were joined by the duplication event; (B) speciation precedes gene duplication, resulting in the paralogs (1A and 1B, 2A and 2B) to cluster first before they were joined by the speciation event. Red and blue dots indicate gene duplication events and speciation events, respectively. (C) Maximum likelihood tree inferred based on protein sequence alignment of PHO5 homologs in four species of Ascomycete yeasts. Numeric values next to the internal nodes indicate bootstrap values in support of the phylogeny shown (1000 replicates run). Red dots indicate inferred gene duplication events at nodes with strong bootstrap support (> 980/1000). The inset shows the species phylogeny. Gene loss events are not shown in this tree but are in Figure S3.

Figure 4

Phosphate-repressible phosphatases are more tightly regulated than thiamine-repressible phosphatases. S. cerevisiae, C. glabrata, and S. pombe wild-type strains were grown in replete, thiamine starvation, and phosphate starvation conditions. qPCR on reverse-transcribed RNA from these strains determined the amount of transcript for various phosphatase genes. Transcript levels were normalized to transcript levels for ACT1, which does not change its expression in response to nutrient conditions, and fold induction was calculated relative to replete for both thiamine (Left) and phosphate (Right) starvation conditions. Error bars represent the standard deviation of the average of fold induction for three independently grown biological replicates. We compared with Student t-tests the two paralogs in each condition, and the p value below 0.05 is indicated with one * and 0.001 with **. If the p value was above 0.05 it was considered not significant (NS). (Left) Thiamine-repressible phosphatases are not tightly regulated, as the other paralogs show some induction. (Right) Phosphate repressible phosphatases are tightly regulated, as there is no induction of the other paralogs.

Figure 5

CgPMU3 encodes a functional thiamine pyrophosphatase (TPPase). C. glabrata wild-type, pmu2Δ, and pmu3Δ strains, containing an empty vector, were grown in thiamine/TPP starvation, thiamine replete, and TPP replete conditions. Plasmids containing CgPMU2 or CgPMU3 under the control of the ScPHO5 promoter were introduced into the deletion strains and grown in the same conditions. The ScPHO5 promoter expresses at a low level regardless of thiamine concentration in C. glabrata (Orkwis et al. 2010). Optical density at 600 nm was measured to determine growth. All strains grow to a high density when thiamine is supplied. Cgpmu3Δ shows a defect when TPP is the sole source of thiamine and only addition of CgPMU3 rescues this defect. Error bars represent the standard deviation of the average of three independently grown biological replicates. The ** indicate a p value below 0.001 when those samples are compared to wild-type with TPP, and all of the other TPP samples are not statistically significantly different from one another.

Figure 6

PHO5 and PMU1 protein family members have different specificities against substrates - some are better TPPases and others are more broad-range organic phosphatases. (Left) The CgPMU3 ORF was deleted and replaced with ORFs of phosphatases from various yeasts so that they are under the control of the CgPMU3 promoter and inducible in growth medium containing TPP as the sole thiamine source. The ability to hydrolyze TPP to thiamine was assayed by measuring growth using optical density at 600 nm. (Right) The phosphatase ORFs were cloned into plasmids under the control of the ScPHO5 promoter, which is highly expressed during phosphate starvation, and these plasmids were introduced into a Cgpmu2Δ, which has minimal PNPPase activity. Cells were assayed for PNPP hydrolysis, indicated by an increase in OD₄₀₀/OD₆₀₀. (Right, images) The strains containing the plasmids were grown on solid medium lacking phosphate and assayed for 1-napthyl-phosphate hydrolysis, which is indicated by a red color. Activity was assayed after 5 min and 45 min. For the bar graphs, error bars represent the standard deviation of the average of three to six replicates. We compared with Student t-tests the two paralogs in each condition, and a p value below 0.05 is indicated with one * and 0.001 with **. If the p value was above 0.05 it was considered not significant (NS).