Phosphate starvation provides strong acquired resistance for H₂O₂ in C. glabrata but not in S. cerevisiae

Based on the evidence linking phosphate to oxidative stress and virulence in C. glabrata and C. albicans, we asked whether phosphate starvation could provide acquired resistance for severe H₂O₂ stress in C. glabrata and whether the behavior is different in the low pathogenic-potential relative, S. cerevisiae. To answer this question, wild type cells from both species were subjected to 45 minutes of phosphate starvation, followed by a severe H₂O₂ challenge (Fig 1A). We designed the experiment to be able to compare the ASR effect between species despite differences in their basal resistance level to H₂O₂, which affects the magnitude of ASR [32]. Through titration, we found that 100 mM and 10mM H₂O₂ resulted in a similar ~2.5% survival in C. glabrata and S. cerevisiae, respectively (S1A Fig, P = 1, Materials and Methods). Using these concentrations as the secondary stress, we found that phosphate starvation greatly enhanced the survival of C. glabrata after the H₂O₂ stress (Fig 1B) but had no detectable effect in S. cerevisiae (Fig 1C). A quantitative colony forming unit (CFU) assay confirmed this (Fig 1D, raw P = 0.0039 and 0.37 in C. glabrata and S. cerevisiae, respectively). To quantify the magnitude of ASR, we defined ASR-score as the fold increase in survival due to the primary stress treatment, i.e., the ratio between the survival rates with (r’) and without (r) the primary stress. Using this definition, we found a 45-minute phosphate starvation led to an ASR-score of 3.4 in C. glabrata (95% CI: [2.66, 4.26]), compared with 1.3 in S. cerevisiae (95% CI: [0.90, 1.63]). To comprehensively test the relationship between phosphate starvation and ASR for oxidative stress, we altered both the severity of the secondary stress and the length of the primary stress and asked how this affected ASR in both species. We found that more severe secondary stress (higher [H₂O₂]) led to stronger ASR in C. glabrata but not in S. cerevisiae at the concentrations tested (S1B Fig). When we increased the primary phosphate starvation treatment to 90 and 135 minutes, we observed a stronger ASR in C. glabrata (ASR-scores = 111 and 170, P < 0.05 after Bonferroni correction, S1C Fig). A similar trend was observed in S. cerevisiae (ASR-score = 11 and 22) but didn’t reach statistical significance (Bonferroni-corrected P = 0.7 and 0.5).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Phosphate starvation induces strong acquired resistance for H₂O₂ in C. glabrata and not in S. cerevisiae.

(A) Experimental design for quantifying Acquired Stress Resistance (ASR): cells were first treated with phosphate starvation (-Pi, or P) or mock-treated (M), then exposed to a secondary H₂O₂ (O) stress or a mock treatment (M). The resulting treatment regimens were referred to as PO, MO, PM and MM. After each treatment regimen, C. glabrata (B) and S. cerevisiae (C) cells were spotted on rich solid media and imaged after 24–48 hrs. Different H₂O₂ concentrations were used for the two species to achieve a similar survival rate without the primary stress. (D) Survival rates (%) after the secondary H₂O₂ challenge were quantified using Colony Forming Units either with (r’) or without (r) phosphate starvation as a primary stress. The biological replicates (n = 8) were shown as dots and their means as bars. Basal survival rates (r) were not different between species (P = 0.38), while there was a significant increase in survival due to the primary stress in C. glabrata but not in S. cerevisiae (P-values = 0.0039 and 0.37, respectively).

https://doi.org/10.1371/journal.ppat.1011748.g001

We also wondered whether the ASR for H₂O₂ is systematically different between the two species or whether phosphate starvation represents a distinct case of divergence. To answer this question, we tested two other stresses, heat shock and glucose starvation, and found that both led to ASR at a similar level between the two species (S2 Fig). Therefore, we conclude that C. glabrata and S. cerevisiae diverged particularly in the phosphate starvation-induced ASR for H₂O₂, where C. glabrata showed a strong ASR while the same primary stress elicited a much weaker effect in S. cerevisiae.

Lastly, we asked if phosphate starvation provides protection for other oxidative stress agents by replacing the secondary stress with either tert-butyl hydroperoxide (tBOOH), an alkyl hydroperoxide, or menadione (menadione sodium bisulfite, MSB), a superoxide agent. We found that phosphate starvation does not provide ASR for tBOOH, while a moderate ASR effect was observed for menadione compared to H₂O₂ (S3 Fig). We conclude that the existence and magnitude of phosphate starvation-induced ASR depends on the nature of the ROS. This result supports the prediction that ASR is specific to the primary and secondary stress combinations rather than being a general cross-stress protection mechanism [7].

Oxidative stress response genes were induced under phosphate starvation in C. glabrata

One potential mechanism for the observed ASR in C. glabrata is that phosphate starvation led to the induction of oxidative stress response (OSR) genes during the primary stress (Fig 2A). To test this, we curated a set of H₂O₂-response genes in the well-studied S. cerevisiae, which include antioxidants, proteases, chaperones and other genes to scavenge ROS and mitigate ROS-induced damages, as well as the TFs regulating them [33,34]. Using this reference set (S1 Table), we identified their orthologs in C. glabrata and asked if they were induced during phosphate starvation. To do so, we profiled the transcriptional response at 1 hour of phosphate starvation in C. glabrata and compared it with the response to the same treatment in S. cerevisiae from [35]. Of the 29 genes tested, 15 were significantly induced by at least two-fold in C. glabrata (FDR < 0.05), compared with three in S. cerevisiae (Fig 2B–2E). These include antioxidant genes such as CTA1, PRX1, SOD1 and SOD2. Induction of these OSR genes could lead to enhanced ROS scavenging and protection against oxidative damages in C. glabrata, consistent with its stronger ASR effect [36–39].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Phosphate starvation strongly induces many oxidative stress related genes in C. glabrata but not in S. cerevisiae.

(A) Proposed transcriptional basis for the acquired stress resistance in C. glabrata: phosphate starvation induces both canonical phosphate homeostasis genes and also oxidative stress response genes, providing protection for the secondary H₂O₂ challenge. (B-E) Comparison of transcriptional induction of genes known to be involved in OSR in S. cerevisiae. Log2 fold changes after 1 hour of phosphate starvation were shown as the mean (bar) of 3 biological replicates (dots) for genes encoding antioxidants (B), protease components (C), molecular chaperones (D) and TFs involved in the OSR (E). The dotted lines indicate a 2 fold induction. Gene names were based on S. cerevisiae except for CTA1, which had two paralogs in S. cerevisiae and only one in C. glabrata. S. cerevisiae CTT1 is involved in OSR and its fold change is shown in (B). Asterisks above a gene name indicate that the gene was significantly induced in that species at an FDR of 0.05.

https://doi.org/10.1371/journal.ppat.1011748.g002

CTA1 is required for the ASR for H₂O₂ but not during phosphate starvation in C. glabrata

To test the hypothesis that the induction of OSR genes contributed to the ASR, we focused on CTA1, which has a well-established role in the resistance to H₂O₂ in C. glabrata [40]. We first used qRT-PCR to confirm that CTA1 is ~3 fold induced after 45 minutes of phosphate starvation (Fig 3A). We also confirmed that Cta1 protein levels increased during phosphate starvation by endogenously tagging Cta1 with GFP (Fig 3B). Notably, this strain had the same H₂O₂ resistance as the wild type (S4A Fig). To determine if CTA1 is required for the phosphate starvation-induced ASR in C. glabrata, we created a cta1Δ strain and compared its ASR phenotype to the wild type using comparable strengths of H₂O₂. We found that loss of cta1 largely abolished the phosphate starvation-induced ASR for H₂O₂ (Fig 3C). Quantitative CFU assays revealed a residual ASR effect in cta1Δ (ASR-score = 1.5, 95% CI [1.2, 1.8]), which was significantly lower than that in the wild type strain (ASR-score = 9.8, 95% CI [5.5, 15], Mann-Whitney’s U test comparing the two genotypes, P = 0.001) (Fig 3D). The ASR defect in cta1Δ was rescued by putting CTA1 back into its endogenous locus (S4B and S4C Fig). To determine if CTA1 induction is sufficient to confer ASR, we replaced the endogenous promoter of CTA1 with an inducible MET3 promoter. We confirmed that MET3pr-CTA1 was induced to a comparable level in the induction media as the endogenous CTA1 under phosphate starvation at 45 min (S5A Fig). Using this strain, we found that pre-inducing CTA1 enhanced the survival of the cells during the secondary stress as did phosphate starvation (S5B and S5C Fig). We therefore conclude that CTA1 induction is both necessary and sufficient for the ASR. It’s worth noting this doesn’t rule out the contribution of other OSR genes induced under phosphate starvation (Fig 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. C. glabrata CTA1 is induced during phosphate starvation and is required for the acquired resistance to severe H₂O₂ stress.

(A) CTA1 mRNA levels were assayed using qRT-PCR following 45 minutes of either 1.5 mM H₂O₂, phosphate starvation or mock treatment. Fold changes over the mock treated samples were shown using ACT1 as a reference. Two isogenic lab strains of C. glabrata were each assayed in triplicates. Bars represent the mean of each strain. An unpaired Student’s t-test comparing the ΔΔC_T values between both treatments with the mock found both to be significantly elevated (P < 0.01, strain was included as a covariate and found to be not significant). (B) Cta1 protein levels were monitored in strains carrying a genomic CTA1-GFP fusion using flow cytometry for 4 hours, during which cells were exposed to either mild H₂O₂ stress, phosphate starvation or mock treatment. The y-axis values are the median fluorescence intensities (a.u. = arbitrary unit). The dots show the mean of >3 biological replicates; the error bars show 95% confidence intervals based on bootstrapping and the lines are LOESS curves fitted to the data. (C) ASR spotting assays for wild type and cta1Δ C. glabrata strains. 2.5 mM H₂O₂ was used as the secondary stress for cta1Δ, which resulted in a similar basal survival rate as 100 mM H₂O₂ for the wild type. (D) ASR effects were quantified by comparing the survival rates after H2O2 treatments either with (solid circles) or without (open circles) the primary stress using a colony forming unit (CFU) assay. The difference in basal survival rates was not statistically significant between the wild type and the cta1Δ strain (P = 0.39). Phosphate starvation significantly increased the survival of both strains during the secondary challenge (raw P = 0.016 in both), but the ASR effect size was much smaller in cta1Δ (ASR-score = 9.8 and 1.5 in wild type and cta1Δ, respectively; Mann-Whitney U test P = 0.001 between the two).

https://doi.org/10.1371/journal.ppat.1011748.g003

One question that remained was whether the induction of CTA1 and other OSR genes served strictly to enhance the survival during the secondary stress, or if they were part of the primary stress response, with an unintended benefit for the secondary stress. To distinguish between these two hypotheses, we asked if Cta1, or the main TF for the canonical OSR, Yap1, were required for survival under phosphate starvation. We found that deletion of neither gene affected growth relative to the wild type during phosphate limiting conditions, while a strain deficient in the PHO response showed severe growth defects (S6A Fig). We conclude that the induction of CTA1 primarily serves the secondary H₂O₂ stress. Lastly, we found that while CTA1 is crucial for the phosphate starvation-induced ASR for H₂O₂, it is less important with mild H₂O₂ used as the primary stress (S6B Fig). This result supports the previous finding that the genetic basis for ASR is highly variable and dependent on the primary stress [11].

Msn4 and Skn7 contribute to CTA1 induction during phosphate starvation in C. glabrata

We reasoned that TFs involved in CTA1 induction during phosphate starvation must be key to the ASR. Four TFs were implicated in CTA1’s induction in C. glabrata under H₂O₂ stress, i.e., Yap1, Skn7, Msn2 and Msn4 [40]. To determine which one(s) contributes to CTA1 induction under phosphate starvation, we deleted each TF and measured Cta1-GFP induction. We found that yap1Δ and msn2Δ didn’t affect Cta1 induction, while msn4Δ and skn7Δ both exhibited decreased Cta1 induction during phosphate starvation (Fig 4A). A double mutant msn4Δ skn7Δ exhibited a stronger reduction in Cta1-GFP than either TFΔ alone, suggesting that both Msn4 and Skn7 were required for CTA1 induction (Fig 4A). However, since Cta1-GFP was still induced in the msn4Δ skn7Δ strain, additional TFs must be involved.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. TFs CgMsn4 and CgSkn7 jointly contribute to CTA1 induction during phosphate starvation in C. glabrata.

(A) Cta1-GFP levels over 4 hours of phosphate starvation for wild type (WT) and TF deletion mutants. The dots represent the mean of > 3 biological replicates and the error bars indicate the 95% confidence interval based on 1000 bootstrap replicates. The lines are LOESS fit to the data. (B) A schematic showing the predicted binding sites (BS) for Skn7 and Msn2/Msn4 in C. glabrata (Materials and Methods). Based on the prediction, two types of cis-mutants were created: the internal deletions (4OL, 8OL) each removed 200 bp containing either the predicted Msn2/Msn4 BS or both Skn7 BS; four point mutation alleles each targeted either the single Msn2/Msn4 BS or one of the Skn7 BS predicted to have a higher affinity. (C, D) Comparing the Cta1-GFP induction driven by each of the mutant promoters in (B) compared with the WT and the corresponding TFΔ mutant.

https://doi.org/10.1371/journal.ppat.1011748.g004

We also measured Cta1 induction during H₂O₂ treatment and confirmed that both Yap1 and Skn7 were required, while msn4Δ resulted in a slight decrease in induction and msn2Δ had no measurable effect under the H₂O₂ concentration tested (S7 Fig). This showed that the induction of the same effector gene, CTA1, depended on different TF combinations under different stresses, a result consistent with previous findings in S. cerevisiae [10].

To determine if Msn4 and Skn7 directly regulate CTA1 under phosphate starvation, we first predicted their binding sites in the promoter of CTA1 (Fig 4B). This allowed us to construct a series of promoter mutants, including two internal deletions (200 bp) removing the predicted Msn4 or Skn7 binding sites, as well as point mutations predicted to abrogate Msn4 and Skn7 binding (Fig 4B). For both TFs, we found that all promoter mutants exhibited reduced Cta1-GFP levels similar to the TF deletion strain, with the 200 bp deletion mutants showing the most severe phenotypes (Fig 4C and 4D). These results strongly suggest that Msn4 and Skn7 directly regulate CTA1 during phosphate starvation. In summary, we found that Msn4 and Skn7, along with unidentified TFs, jointly regulate CTA1 induction in C. glabrata under phosphate starvation, while Yap1 and Skn7 were the main TFs responsible for its induction under H₂O₂ stress.

Msn4 translocates into the nucleus upon phosphate starvation in C. glabrata

The activity of Msn4 and its paralog Msn2 are partially regulated via nuclear translocation in both C. glabrata and S. cerevisiae [40–43]. We therefore asked if Msn4 orthologs respond differently to phosphate starvation in C. glabrata vs S. cerevisiae. Using GFP-tagged Msn4 and fluorescent microscopy, we found that phosphate starvation led to 47% of the C. glabrata cells to have nuclear localized Msn4 (Msn4^nuc), compared with < 4% in S. cerevisiae (Fig 5, Bonferroni corrected P < 0.01). This divergence is condition-specific: glucose starvation led to a similar level of Msn4^nuc in both species (Fig 5, 76% in C. glabrata vs 73% in S. cerevisiae, raw P = 0.7). No significant difference was observed between species under the no-stress condition (raw P = 1). These results further support CgMsn4 as one of the TFs mediating the phosphate starvation induced ASR in C. glabrata and that divergence in how Msn4 responds to phosphate starvation is a major contributor to the species divergence in ASR. In contrast, we found that the paralog Msn2 translocated into the nucleus in a comparable fraction of cells in the two species under phosphate starvation: 52% in C. glabrata and 54% in S. cerevisiae (S8 Fig, P = 0.8). Combined with the lack of effect of msn2Δ on Cta1 induction, we conclude that Msn2 is not involved in the ASR and doesn’t contribute to its divergence.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. C. glabrata Msn4 (CgMsn4) translocates into the nucleus upon phosphate starvation but not its ortholog in S. cerevisiae (ScMsn4).

Cellular localization of CgMsn4 (A) and ScMsn4 (B) under phosphate starvation (0 mM Pi, -Pi), glucose starvation (0.02% glucose, -Glu), and no stress conditions. All treatments were for 45 minutes. From left to right: i. CgMsn4-yeGFP and ScMsn4-mCitrine; ii. nucleus staining with DAPI in C. glabrata or labeled with Nh6a-iRFP in S. cerevisiae; iii. merge; iv. bright field. The percent of cells with nuclear-localized Msn4 and the total number of cells examined by microscopy were shown on the right. Scale bars are 5 μm in length.

https://doi.org/10.1371/journal.ppat.1011748.g005

The Greatwall kinase, Rim15, is an important signaling component in the ASR network

Given Msn4’s role in the phosphate starvation-induced ASR in C. glabrata (Figs 4 and 5), we used it as a bait to identify upstream signaling components in the ASR network. In S. cerevisiae, a key regulator of Msn2/Msn4 activity is the Greatwall kinase, Rim15 [44,45]. Rim15 itself is regulated by multiple nutrient and stress-sensing kinases, e.g., TORC1, PKA and Pho80/85, making it a hub for integrating nutrient and stress signals [46,47]. We hypothesize that Rim15 regulates the nuclear localization of Msn4 during the primary stress in C. glabrata (Fig 6A). To test this, we compared the nuclear localization of CgMsn4-GFP in the wild type and rim15Δ backgrounds. We found that ~68% of the wild type cells had nuclear-localized CgMsn4 during phosphate starvation, compared with only 22% in the rim15Δ background (Fig 6B, P < 0.001). By contrast, both strain backgrounds showed a low fraction of cells with CgMsn4^nuc under high phosphate conditions (Fig 6B, P = 0.2). Next, we asked if Rim15 is required for CTA1 induction. We found that rim15Δ reduced Cta1-GFP induction to a similar level as msn4Δ, supporting its role as a regulator of the ASR (Fig 6C). msn4Δ rim15Δ had a more severe effect than either deletion alone, suggesting additional players besides the Rim15-Msn4 pathway (Fig 6C). Lastly, to assess the role of Rim15 in the phosphate starvation-induced ASR, we performed quantitative CFU assays to compare the wild type and rim15Δ strains at a comparable strength of H₂O₂ (Fig 6D). While there is a significant ASR effect in both strains (both raw P = 0.008), the ASR effect is weaker in rim15Δ cells (mean ASR-score = 3.45 for rim15Δ, 95% CI = [2.54, 4.33]; mean ASR-score = 6.29 for the WT, 95% CI = [4.58, 8.53], Mann-Whitney U test between genotypes, two-sided P = 0.04). This ASR defect was rescued by putting RIM15 back to its endogenous locus (S9 Fig). Taken together, these results strongly suggest that Rim15 is a crucial signaling component for the phosphate starvation-induced ASR in C. glabrata.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. The Greatwall kinase homolog Rim15 mediates phosphate starvation-induced ASR in C. glabrata.

(A) Proposed model for phosphate starvation-induced ASR in C. glabrata: phosphate starvation (-Pi) activates Rim15, which goes on to regulate Msn4’s activity and contributes to ASR. Dashed arrows show proposed connections; solid arrows are regulations supported by previous results; red question marks indicate specific downstream effects to be tested. (B) CgMsn4 nuclear localization in WT and rim15Δ strains in -Pi or rich media. Left: CgMsn4-GFP, % of cells with CgMsn4^nuc labeled on the lower left; right: bright field (BF) showing cells. Fisher’s Exact Tests (FET) were performed to compare the two strains under each condition. Raw, two-sided P-values were reported on the side. (C) Cta1-GFP induction during -Pi. Dots are the means of >3 biological replicates and the error bars the 95% CI based on 1000 bootstraps. (D) ASR assay for the wild type (WT) and rim15Δ. Shown are survival rates of the two strains at an equivalent H₂O₂ strength, either with (-Pi) or without (Mock) a primary stress. Each dot is an independent biological replicate (n = 7) and the red bar represents the mean. Basal survival rates were not different between the two strains (P = 0.5). The ASR effect is significant in both (raw, one-sided P = 0.008 for both), but is significantly lower in rim15Δ (mean ASR-score = 6.29 in WT vs 3.45 in rim15Δ, Mann-Whitney U test, two-sided P = 0.04).

https://doi.org/10.1371/journal.ppat.1011748.g006

Target-Of-Rapamycin Complex 1 (TORC1) is strongly inhibited by phosphate starvation in C. glabrata but not in S. cerevisiae

In S. cerevisiae, Msn2 and Msn4’s nuclear translocation is regulated by both the Target-of-Rapamycin Complex 1 (TORC1) and Protein Kinase A (PKA) pathways [41,45,48]. In S. cerevisiae, TORC1 responds slowly and weakly to phosphate limitation [49]. In the distantly related yeast pathogen C. albicans, TORC1 was shown to respond rapidly to phosphate availability [28]. We therefore hypothesized that TORC1 is more strongly and quickly inhibited by phosphate starvation in C. glabrata than in S. cerevisiae, contributing to the Msn4-mediated stress response and ASR. To test this, we used the phosphorylation of ribosomal protein S6 (Rps6) as a conserved readout of TORC1 activity [28,50,51]. We found that 60 minutes of phosphate starvation did not affect the total level of Rps6 but markedly reduced the abundance of phosphorylated Rps6 (P-Rps6) in C. glabrata (Bonferroni-corrected P = 0.06, Fig 7A and 7B). The same treatment, however, did not affect the proportion of P-Rps6 in S. cerevisiae (P = 1). By contrast, nitrogen starvation rapidly inhibited phosphorylation of Rps6 in both species (P = 0.06 and 0.07). These results confirmed that inhibition of TORC1 by nitrogen starvation is conserved between the two species while only in C. glabrata is TORC1 strongly inhibited by phosphate starvation at 60 minutes, consistent with their distinct ASR phenotypes (Fig 1). To determine which of the two species represents the ancestral state, we performed the same P-Rps6 assay in two outgroup species, K. lactis and L. walti, which diverged from S. cerevisiae and C. glabrata approximately 120 million years ago [15]. Both species exhibited intermediate levels of TORC1 inhibition by phosphate starvation at 60 minutes, suggesting that phosphate signaling to TORC1 is ancestral, but the signaling strength has evolved (S10 Fig). By contrast, in all four species TORC1 was rapidly inhibited by nitrogen starvation, suggesting that nitrogen is an evolutionarily conserved input for TORC1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. TORC1 is strongly inhibited by phosphate starvation in C. glabrata, likely contributing to the ASR via its proximal kinase, Sch9.

(A) A representative Western Blot for phosphorylated Rps6 (P-Rps6) and total Rps6 in both species under rich media, nitrogen starvation (-N) or phosphate starvation (-Pi) conditions. Red arrows point to the loss vs presence of the P-Rps6 band under -Pi in C. glabrata and S. cerevisiae, respectively. (B) Quantification of %P-Rps6 over total Rps6 (n = 3). *Bonferroni corrected P-values from Student’s t-tests were shown. (C) Inhibiting TORC1 by rapamycin induces Cta1-GFP in a dose-dependent manner. The dots, error bars and lines have the same meaning as before. (D) ASR for H₂O₂ with rapamycin as the primary treatment. Plotted are survival rates with or without rapamycin treatment (n = 12 for C. glabrata, n = 8 for S. cerevisiae). 60 mM and 6 mM of H₂O₂ were used as the secondary stress for the two species, resulting in a similar basal survival rate (P = 0.6). A Wilcoxon signed-rank test was used to compare the paired experiments with or without the primary treatment for each species. The raw, one-sided P-values were shown on the top. (E) Same plot as C, comparing Cta1-GFP induction in a phosphomimetic mutant of Sch9, a key proximal effector of TORC1, and a matching wild type Sch9 strain. (F) Same as D but with phosphate starvation (-Pi) as the primary stress, comparing the Sch9-3E mutant (n = 4) and the matching Sch9-wt control (n = 4). 100 mM and 40 mM of H₂O₂ were used as the secondary stress for the two genotypes, resulting in a similar basal survival rate (P = 0.7).

https://doi.org/10.1371/journal.ppat.1011748.g007

Inhibition of TORC1 leads to Msn4 nuclear localization, CTA1 induction and confers mild ASR in C. glabrata

To determine if TORC1 inhibition can lead to ASR, we used rapamycin to specifically inhibit TORC1 and monitored Msn4^nuc, Cta1-GFP induction and the ASR effect. We found that TORC1-inhibition induces Msn4^nuc in both species, with a stronger effect in C. glabrata: 45 min of exposure to 500 ng/mL of rapamycin causes 71.8% (79/110) of C. glabrata cells and 14.5% (19/131) of S. cerevisiae cells to have Msn4^nuc. Both were significantly higher compared with the mock-treated group (1.8% in C. glabrata and 1.2% in S. cerevisiae, raw FET P < 0.001 in both). Next, we followed Cta1-GFP induction in C. glabrata exposed to rapamycin and found a dose-dependent response except for the highest dose, which resulted in a lower induction than the next dose (Fig 7C). Lastly, we found 125 ng/mL rapamycin treatment for 45 minutes modestly but significantly enhanced the survival of C. glabrata following a secondary H₂O₂ stress (Fig 7D, mean ASR-score = 1.79, 95% CI [1.35, 2.28], raw P = 0.016). The same rapamycin dose resulted in a similar increase in survival in S. cerevisiae (mean ASR-score = 1.81, 95% CI [1.21, 2.47], raw P = 0.063) (Fig 7D). We conclude that TORC1 inhibition has the potential to induce acquired resistance for H₂O₂ in both species. This implies that divergence of the ASR occurred through TORC1’s differential response to -Pi.

TORC1 proximal effector, Sch9, contributes to phosphate starvation-induced ASR, while a downstream output of the Tap42-PP2A branch is minimally affected

The AGC kinase Sch9 is directly phosphorylated and regulated by TORC1 [49]. In turn, Sch9 regulates multiple downstream processes, including repressing stress responses by inhibiting Rim15 via phosphorylation, which prevents it from activating Msn4 [52,53]. We hypothesized that Sch9 negatively regulates the ASR in C. glabrata. To test this, we created a TORC1 phosphomimetic mutant of CgSch9, predicted to be constitutively active with respect to TORC1 signaling [49] (S11 Fig). This mutant showed reduced Cta1-GFP induction during phosphate starvation (Fig 7E). Moreover, a quantitative CFU assay failed to detect ASR in this mutant but did in the matching complement strain (Fig 7F, raw P = 0.69 and 0.063, respectively), further supporting TORC1’s role in the ASR.

Since TORC1 coordinates multiple cellular processes, we asked if phosphate starvation had limited or broad effects on TORC1’s downstream branches [54]. To test this, we monitored the Nitrogen Catabolism Repression (NCR) regulon, which is controlled by a different branch, Tap42-PP2A, rather than by Sch9, and is differentially induced under nitrogen starvation vs glucose starvation and other stresses in S. cerevisiae [55]. Using an established DAL80pr-GFP reporter [56], we found that nitrogen starvation strongly induced NCR-sensitive genes, but phosphate starvation had minimal effects in both S. cerevisiae and C. glabrata (S12 Fig). This suggests that evolution can wire an input such as phosphate limitation to a specific TORC1 output, e.g., stress response, without strongly impacting other outputs.

Experimental reproducibility and data availability

Each experiment was repeated at least two but mostly three or more times with >2 biological replicates each. Data analysis and visualization were performed in R v4.2. Raw data, R markdown files, and output files are available at https://doi.org/10.5281/zenodo.10004973. RNA-seq data for C. glabrata before and after 1 hour of phosphate starvation is available through the GEO database (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE244380).

Yeast media and growth conditions

Yeast cells were grown in the Yeast extract-Peptone Dextrose (YPD) medium or the Synthetic Complete (SC), using Yeast Nitrogen Base without amino acids (Sigma Y0626) supplemented with 2% glucose and amino acid mix. Unless specified, all stress treatments were performed with mid-log phase cells. Briefly, cells were grown overnight in either YPD or SC media, diluted in the morning to OD₆₀₀ = 0.2 and grown to OD₆₀₀ ~ 1. To apply the treatments, cells were collected by centrifugation at 3,000g for 5 minutes, washed 2–3 times with water and released into the treatment media. Nitrogen starvation medium was made using the Yeast Nitrogen Base without amino acids and without ammonium sulfate (Sigma, Y1251). Glucose starvation medium had 0.02% or 0.05% glucose in the SC medium. Phosphate starvation medium was made using Yeast Nitrogen Base with ammonium sulfate, without phosphates, without sodium chloride (MP Biomedicals, 114027812) and supplemented to a final concentration of 2% glucose, 1.5 mg/ml potassium chloride, 0.1 mg/ml sodium chloride and amino acids, as described previously [25]. Phosphate concentration in the medium was measured using a Malachite Green Phosphate Assay kit (Sigma, MAK307). The -MC medium used to induce MET3pr-CTA1 was made the same way as the SC medium, except methionine and cysteine were omitted from the amino acid mix. To test the effect of inhibiting TORC1 on ASR, Rapamycin (RPI, R64500) was added into the SC medium at the indicated concentrations. To apply H₂O₂ stress, H₂O₂ stock solution (30% w/w, Sigma H1009) was diluted into SC medium to the indicated concentrations immediately before the treatment, except in the MET3pr-CTA1 experiment, where -MC medium was used instead of SC medium to allow CTA1 to be induced during the secondary stress. To test additional ROS for the ASR effect, tert-butyl hydroperoxide (tBOOH 70% w/w in H₂O, Sigma 458139) was diluted directly into the SC medium, while menadione sodium bisulfite (Sigma M5750) was first dissolved in ddH₂O to make a 1M stock, and then the stock was diluted into the SC medium at the indicated concentrations immediately before the treatment.

Generating yeast strains and plasmids

Genetic transformation of both S. cerevisiae and C. glabrata was performed using the LiAc method as described in [69]. C. glabrata knockout strains were generated with auxotrophic markers HIS3 and URA3 and the nourseothricin drug marker NAT. ~200 bp of flanking homology sequences were added to either a deletion cassette or a knock-in construct using overlap PCR. Transformants were confirmed by colony PCR on both sides of the gene. To construct CTA1 promoter mutants, a URA3 cassette was used to replace the endogenous promoter, after which the mutant allele constructed by PCR mutagenesis was used as the repair template to swap out the URA3 and selecting with 5-fluoroorotic acid (5-FOA) (GoldBio, F-320). A similar approach was used to replace the endogenous CTA1 promoter with the promoter from the C. glabrata MET3 gene. The CgMET3 promoter was amplified from pCU-MET3 [70] (AddGene #45336). The Msn2 and Msn4-GFP expression plasmids were made by inserting the coding sequence of the genes in between the PDC1 promoter and GFP in the pCU-PDC1-GFP vector [71] (AddGene #45342). The ScDAL80-GFP reporter was constructed by amplifying the 1kb upstream region of ScDAL80 and placing it in front of a yeGFP in a pRS426 (ATCC #77107) backbone. Gibson cloning was performed according to the protocol of the Gibson Assembly kit (NEB, E5520). To construct the Sch9-3E-NAT mutant and the matching Sch9-WT-NAT control strains, we first used PCR-based mutagenesis to construct the Sch9-3E fragment (S11 Fig). This and the Sch9-WT fragments were joined with the NAT cassette via overlap PCR. Then, ~200bp of flanking sequences homologous to the endogenous SCH9 5’ and 3’UTR were added on both sides using overlap PCR. The end product was used as a linear template to transform an SCH9 wild type strain of C. glabrata. Positive transformants were selected on YPD+NAT plates and validated by genomic PCR and sequencing. Yeast strains, plasmids and qRT-PCR primers used in this study are listed in Tables 1, 2 and 3, respectively.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Yeast strains used in this study.

https://doi.org/10.1371/journal.ppat.1011748.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Yeast expression plasmids used in this study.

https://doi.org/10.1371/journal.ppat.1011748.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. qRT-PCR Primers used in this study.

https://doi.org/10.1371/journal.ppat.1011748.t003

Spotting and Colony Forming Units (CFU) assays

A semi-quantitative spotting assay or a quantitative Colony Forming Unit (CFU) assay was used to measure the survival rate after a treatment. For the spotting assay, the treated cells were 10x serially diluted in ddH₂O. 5 μl of each dilution was spotted onto an SC plate and allowed to dry before the plate was incubated at 30C for 16–36 hours. An image was taken at the end of the incubation using a homemade imaging hood fitted with an iPad Air. For the CFU assay, post-treatment cells were diluted 100 to 1,000-fold and plated on an SC plate. The dilution was chosen to produce roughly 10–500 colonies per plate. The plates were incubated for 48 hours and CFU was manually counted using a lab counter (Fisher Scientific).

Calibrating H₂O₂ concentrations

To apply an equivalent strength of H₂O₂ treatment to species and genotypes with different basal survival rates, we measured the cell survival rates after treatment with different concentrations of H₂O₂. Specifically, we exposed S. cerevisiae cells to a series of H₂O₂ concentrations from 0mM to 10mM, with 2mM steps. Similarly, we exposed C. glabrata cells H₂O₂ concentrations from 0mM to 100mM with 20mM steps. CFU assays were performed after 2 hours of treatment to estimate the survival rate. A Mann-Whitney U test was used to determine the significance of differences. Concentrations resulting in similar survival rates across species or genotypes were chosen as the condition for the secondary stress in the ASR experiment.

Acquired stress resistance (ASR) assay

During the primary stress phase, mid-log phase cells were collected, washed, and released into either the SC medium for a mock treatment or the primary stress media, e.g.,—Pi for phosphate starvation. After 45 minutes (or as indicated) of incubation with shaking at 30°C, cells were collected by centrifugation and resuspended at OD₆₀₀ = 0.2 in the secondary stress treatment media, then incubated the same as before for 2 hours. After the secondary stress, cells were diluted and directly used for the spotting or CFU assay. CFU multiplied by the dilution rate was then used to calculate r (MO/MM), r’ (PO/PM) and the ASR-score (r’/r). Three to eight biological replicates were performed.

RNA-seq profiling for C. glabrata under phosphate starvation

C. glabrata wild type cells were grown in SC overnight, diluted into fresh SC the next morning to OD₆₀₀ = 0.1, grown for another 2–3 hours until the culture reached mid-log phase (OD₆₀₀~0.6). At this point, two biological replicates of pre-starvation samples were collected using a cold methanol quenching method [35,72]. Briefly, 5 mL of culture was added directly into 7.5 mL of pre-chilled methanol (−50°C) and incubated in an ethanol-dry ice bath at that temperature for at least 20 min. The remaining culture was collected by filtration, washed with equal volume of pre-warmed no phosphate SC media and then released into the pre-warmed no phosphate SC. After 1 hour incubation with shaking at 30°C, three biological replicates of the starved samples were collected in the same way as above. When all samples were in the ethanol-dry ice bath for > 20 minutes, cells were collected by centrifugation and quickly washed with ice-cold water to remove the methanol and resuspended in RNAlater solution (Qiagen, 76104) for at least 2 hours. Cells were centrifuged to remove the RNAlater, flash-frozen in liquid nitrogen and stored at -80 C until RNA-extraction. For each sample, ~5x10⁷ cells were collected and total RNA was extracted using a MasterPure Yeast RNA purification kit (Biosearch Technologies, MPY03100) following the manufacturer’s protocol. RNA-seq libraries were prepared with the TruSeq RNA Library Preparation Kit v2 (Illumina) with the mRNA purification option. The resulting libraries were sequenced on an Illumina HiSeq 4000, which produced on average 10 million 50 bp single end reads for each sample. Raw and processed data are available at GEO (GSE244380).

Comparing the transcriptional response to phosphate starvation between C. glabrata and S. cerevisiae

The RNA-seq raw reads were mapped to the C. glabrata genome downloaded from the Candida Genome Database (CGD, RRID:SCR_002036, version s02-m02-r09), using Bowtie v1.1.1 (RRID:SCR_005476) with the option ‘-m 1—best–strata’ [73]. The resulting SAM files were sorted using Samtools v1.2 (RRID:SCR_002105) [74] and the number of reads per transcript was counted using Bedtools2 (RRID:SCR_006646) [75], with the option ‘bedtools coverage -a BAM_file -b genome_annotation.bed -S -s sorted -g Chrom.length`. Gene features for C. glabrata were downloaded from CGD, version s02-m07-r04. The count matrix was filtered to remove lowly expressed genes (335 genes with <1 read per sample on average). Next, we used the trimmed mean of M-values (‘TMM’) method in the EdgeR (RRID:SCR_012802) package to calculate the normalization factors for scaling the raw library sizes [76], and applied voom transformation [77] to remove the mean-variance relationship on the log2 transformed count data. A log2 fold change was calculated for each gene by subtracting the mean of the pre-starvation samples from each of the 1-hr starvation sample values. For S. cerevisiae, the microarray data from [35] was downloaded from the GEO database (RRID:SCR_005012). Within the series GSE23580, three samples corresponding to the three biological replicates comparing the 1 hour phosphate starved sample to a mock-treated sample were extracted (GSM578408, 578424, 578440). The log2 ratios were already background-subtracted, normalized, and used directly for comparisons. To compare orthologous gene expression, we first curated a list of oxidative stress response genes based on the literature (S1 Table) [33,34]. To identify their homologs in C. glabrata, we downloaded the orthology and best-hit mapping between the two species from CGD. The best-hit mapping was based on less stringent criteria and only used when an orthology mapping was not available. The resulting table of gene expression log2 fold changes is available in S2 Table.

CTA1 induction by quantitative real-time PCR (qRT-PCR)

C. glabrata cells were mock-treated or treated with phosphate starvation or 1.5 mM H₂O₂ for 45 minutes. ~2-4x10⁷ cells were harvested by centrifugation, snap-frozen in liquid nitrogen and stored in -80°C until RNA extraction. Two biological replicates were collected per condition. RNA was extracted using a MasterPure Yeast RNA purification kit (Biosearch Technologies, MPY03100). cDNA was synthesized using SuperScript III First-Strand Synthesis System (Invitrogen, 18080051) and treated with RNase A (ThermoFisher, EN0531). We used the LightCycler 480 SYBR Green I Master (Roche, 04707516001) and qRT-PCR was performed on a Roche LC480 instrument. Three technical replicates were performed for each sample. The PCR program was 95°C for 5 minutes, followed by 45 cycles of 95°C for 10 seconds, 60°C for 20 seconds, and 72°C for 20 seconds. Data analysis was done using the instrument software, including automatic baseline subtraction and C_T value quantifications. ACT1 was used as a reference gene and its C_T values were subtracted from those of CTA1. We then subtracted the mean of the ΔC_T for the mock-treated samples from the technical replicates for the phosphate starvation and H₂O₂-treated ones to obtain ΔΔC_T, which were used for downstream analyses.

Cta1-GFP time course by flow cytometry

Cells were grown to mid-log phase in SC medium in 96 deep-well plates (VWR, 82051). After washing with ddH₂O twice and spinning down at 3,000g for 5 minutes, cells were resuspended in 0mM Pi SC medium, H₂O₂-containing SC medium or SC medium (7.5mM Pi), incubated at 30°C shaking at 200 rpm for 4 hrs. At the indicated time points, an aliquot of cell culture (40–50uL) was drawn and diluted into ddH₂O (360uL–400uL) for flow cytometry on an Attune NxT instrument fitted with an autosampler (Thermo Fisher). For each sample, 30,000 events were recorded; non-cell events and doublets were removed based on FSC-H and FSC-W. Median fluorescence intensity (MFI) was used for Cta1-GFP protein expression analysis.

Binding site prediction for CTA1 promoter

To predict the binding sites (BS) for Skn7 and Msn4 in the CgCTA1 promoter, we first retrieved the 1000 bp DNA upstream of the start codon for the CTA1 gene in the C. glabrata CBS138 genome (CGD, version s02-m02-r09). We then scanned this sequence with the expert curated motifs for Skn7 and Msn4 from the YeTFaSCo (yetfasco.ccbr.utoronto.ca) database. To perform the scanning, we first selected the two motifs for Skn7 and single expert curated motif for Msn4, “added them to the cart”, and used the “scan using cart” function. We adopted the default 75% minimum percent of maximum score cutoff, and set the background percent A/T to 0.3, which was based on the mean GC% in 1000 bp upstream sequences in the C. glabrata genome calculated in-house. The motifs and prediction results were available in S1 Text.

Fluorescent microscopy for Msn4 and Msn2 nuclear localization

C. glabrata cells were transformed with the indicated plasmids and grown on SC Ura- plates for selection. Transformed cells were then grown and maintained in SC Ura- liquid medium prior to the experiments. C. glabrata cells were grown to mid-log phase in SC Ura- medium and treated with SC, glucose starvation, or 0 mM Pi SC media for 30 mins. Subsequently, 4’,6-diamidino-2-phenylindole (DAPI, Sigma, D9542) was diluted in the respective medium and added to the cultures at a final concentration of 10 ug/ml. Cells were incubated with DAPI in the dark for 15 minutes before DAPI was removed and cells were washed twice with ddH₂O. Finally, cells were re-resuspended in the treatment medium. Fluorescent imaging was performed on a Leica SP8 confocal imaging system (Carver Center for Imaging, the University of Iowa). A 488 nm laser was used for yeGFP excitation and 495–585 nm for emission; a 405 nm laser was used for DAPI excitation and 430–550 nm for emission. In S. cerevisiae, Msn2-mCherry and Msn4-mCitrine cells were prepared as described above excluding the DAPI staining. For Msn4-mCitrine, 488 nm was used for excitation and 501–547 nm for emission; for mCherry, 552 nm was used for excitation and 570–649 nm for emission; for visualizing the nucleus with the Nhp6a-iRFP marker, a 638 nm laser was used for excitation and 648–790 nm for emission.

Total and P-Rps6 Western Blot

Mid-log phase cells were subjected to either mock or starvation treatments in parallel. Cell lysis buffer (0.1M NaOH, 0.05M EDTA, 2% SDS, 2% β-mercaptoethanol) and loading buffer (0.25M Tris-HCL pH6.8, 50% Glycerol, 0.05% bromophenol blue) were prepared and protein extraction was performed as described in [78]. Cell lysates were separated by SDS-PAGE at 200 volts for 40 minutes, transferred onto nitrocellulose membranes (Bio-Rad, 1662807), and stained with a total protein stain (Li-COR, 926–11015), which we used for normalization in place of a loading control, as recommended by Li-COR. The membranes were blocked with 3% BSA (RPI, A30075) at room temperature for 1 hr. Two blots for the same samples were probed with either anti-phospho (Ser/Thr) Akt substrate rabbit polyclonal antibody (RRID:AB_330302, Cell Signaling Technology # 9611) or Anti-RPS6 rabbit antibody (RRID:AB_945319, Abcam ab40820) to detect the phosphorylated Rps6 (P-Rps6) and total S6, respectively. After washing, the blots were incubated with a secondary antibody (RRID:AB_621843, LI-COR, 926–32211) and imaged on a Li-COR Odyssey Fc scanner (Carver Center for Genomics, the University of Iowa). The intensity of P-Rps6 and total Rps6 were quantified using the Li-COR Image Studio software, following the manufacturer’s instructions for the Revert 700 Total Protein Stain for Western Blot Normalization. Briefly, the band of interest for each lane was selected using same-sized rectangles. We then subtracted the background, calculated as the median of the top and bottom regions of a band. We did the same analysis for the total protein stain image, except we included the entire lane where bands are visible. We then normalized the Western blot signals in each sample to the corresponding total protein stain to obtain the intensity for P-Rps6 or total Rps6. Finally, we divided the two values to get the P-Rps6/total Rps6 ratio.

Assaying cell morphology by light microscopy

To determine the percent of unbudded cells during a phosphate starvation time course, wild type C. glabrata and S. cerevisiae cells were grown as described above to mid-log phase, collected by centrifugation, washed once with pre-warmed 0 mM Pi media and diluted to 5x10⁶ cells/mL in 0 mM Pi media. At 0, 20, 45, 75, 105 and 135 minutes into the time course, an aliquot of cells was taken from each sample, sonicated for 5s to break up any cell clumps, diluted 10-50x and counted on a hemocytometer (Marienfeld, 0640030) under a light microscope. The number of unbudded cells over the total number of cells were recorded by manually examining > 100 cells over four fields of view. The numbers from the four fields of view were summed for downstream analysis.

Statistical analyses and tests

For quantitative ASR assays using CFU, the data from each replicate experiment was used to calculate survival rates with or without the primary stress, expressed as r = CFU_MO / CFU_MM and r’ = CFU_PO / CFU_PM. The fold change in survival due to the primary stress, expressed as r’/r, aka ASR-score, was calculated for each experiment. We estimated the mean and 95% confidence interval for the ASR-score with 1000 bootstraps using the smean.cl.boot() function in the Hmisc package. To identify equivalent strengths of ROS for different species and strains, a Mann-Whitney U test was used to test for differences in the basal survival rates (r) and a two-sided P-value was reported. To assess the statistical significance of the ASR effect in a strain, we used the paired r and r’ estimates from each replicate and performed a Wilcoxon signed-rank test if n > = 4. When n < 4, a paired Student’s t-test was used since the former test has very low power at such a small sample size. The one-sided P-value, with the alternative hypothesis being r’ > r, was reported in the legend and text. To compare the ASR effect sizes between two strains, a Mann-Whitney U test (aka Wilcoxson rank-sum test) was performed for the paired data. A two-sided P-value was reported. When more than one comparison was made, a Bonferroni correction was applied. For transcriptomic comparisons, we performed a t-test for the triplicates of the log2 fold changes for 29 genes in both species. The alternative hypothesis was that the true log2 fold change was greater than 0 (induced under phosphate starvation). The resulting P-values were adjusted for multiple comparisons using the Benjamini-Hochberg procedure (FDR). Genes with an FDR < 0.05 and a log2 fold change > 2 were considered significantly induced. For CTA1 induction by qRT-PCR, the ΔΔC_T values from the three technical replicates for the phosphate starved, H₂O₂-treated and mock-treated samples were analyzed using a linear model, with the two stress conditions treated as dummy variables and the biological replicate included as a covariate (R: lm(ddCt ~ treatment + strain), where treatment is a factor with levels = {“Mock”, “-Pi”, “H₂O₂”}). Raw P-values for each level were reported. For Cta1-GFP basal and induced levels, a linear model was used to compare the different genotypes and the P-values were Bonferroni-corrected (multiplied by 5 mutants). For Msn4 nuclear localization, cells with nuclear-localized vs cytoplasmic Msn4 were counted under a fluorescent microscope (>100 cells in total per strain x condition). A Fisher’s Exact Test (FET) was performed for cells with or without nuclear-localized Msn4 and between species/genotypes. P-values were calculated using an exact method. When multiple comparisons were made, a Bonferroni correction was applied unless otherwise noted in the text or legend. For Western Blot analysis of P-Rps6 / total Rps6, the samples being compared were run on the same gel for each replicate. Thus, a paired Student’s t-test was performed, and a two-sided P-values after Bonferroni correction was used to assess the significance of the difference between the treatment conditions.