Recurrent background mutations in WHI2 impair proteostasis and degradation of misfolded cytosolic proteins in Saccharomyces cerevisiae

Proteostasis promotes viability at both the cellular and organism levels by maintaining a functional proteome. This requires an intricate protein quality control (PQC) network that mediates protein folding by molecular chaperones and removes terminally misfolded proteins via the ubiquitin proteasome system and autophagy. How changes within the PQC network can perturb proteostasis and shift the balance between protein folding and proteolysis remain poorly understood. However, given that proteostasis is altered in a number of conditions such as cancer and ageing, it is critical that we identify the factors that mediate PQC and understand the interplay between members of the proteostatic network. In this study, we investigated the degradation of a thermally unstable cytosolic model substrate and identified a surprisingly high number of strains in the yeast knockout collection that displayed impaired turnover of the misfolded substrate. We found that this phenotype was caused by frequent background mutations in the general stress response gene WHI2. We linked this proteostatic defect to the lack of activity of the stress response transcription factor Msn2, potentially under conditions where the TOR pathway is active. Our results underscore how changes to the elaborate PQC network can perturb proteostasis and impair degradation of misfolded cytosolic proteins.

. Flow cytometry based screen for E3 ligases targeting Guk1-7-GFP for degradation. (a) Wild type cells expressing ectopic Guk1-GFP or Guk1-7-GFP were incubated with CHX for 4 hours at 25°C or 37°C and samples were collected at the indicated time points. Results represent the mean and standard deviation of three independent experiments. (b) Seventy non-essential E3 ligase deletion strains expressing Guk1-7-GFP were incubated with CHX at 25°C or 37°C for 2 hours and then analyzed by flow cytometry. Red line demarks strains with a relative loss of fluorescence (Rel. LoF) value of 0.75 or lower. (c) The top 20 strains were selected for further validation by flow cytometry with experiments performed as in (b). Data points in red correspond to strains displaying Guk1-7-GFP stabilization levels higher than that of ubr1∆. (d) Wild type or asi1∆ cells expressing Guk1-7-GFP were incubated with CHX at 25°C or 37°C for two hours prior to flow cytometry analysis. Results represent the mean and standard deviation of three independent experiments. (e) Wild type and asi1∆ cells co-expressing Guk1-7-GFP and an empty vector (EV) or ASI1 were treated as in (d). ns, *, ** and *** denote: not significant, p < 0.05, 0.01, and 0.005, respectively. ubiquitin ligase responsible for the degradation of the Guk1-7-GFP model substrate, we screened a collection of 70 non-essential E3 ligase deletion strains that were individually transformed with a plasmid encoding the Guk1-7-GFP fusion under the GPD promoter. Cultures were grown at 25°C and then divided and incubated in the presence of cycloheximide (CHX) for two hours at either 25°C or 37°C before performing flow cytometry analysis. For each deletion strain, the relative difference in median GFP fluorescence intensity between samples incubated at 25°C or 37°C was normalized to that of the wild type strain in order to calculate the relative loss of fluorescence (Rel. LoF) of Guk1-7-GFP (Fig. 1b). The collection was screened twice, and strains that had a relative loss of Guk1-7-GFP fluorescence value of 0.75 or lower, in at least one of the two rounds, were selected for further validation. A total of 20 strains met this criterion and were further analysed by flow cytometry in three independent experiments (Fig. 1c). In agreement with our previous findings, deleting UBR1 led to a 25% lower average loss of Guk1-7-GFP fluorescence compared to that of wild type cells 8 . Surprisingly, we identified twelve E3 ligase deletion strains with a greater impairment of Guk1-7-GFP degradation than that observed in ubr1∆ cells. Of these, seven strains had an average relative loss of Guk1-7-GFP fluorescence value of 0.5 or lower. These results indicate that an unusually high number of strains from our yeast knockout collection have a reduced capacity to eliminate misfolded cytosolic proteins.
A secondary mutation in WHI2 is linked to impaired proteostasis. We confirmed the results of our screen by performing CHX chase experiments with cells lacking ASI1, a member of the nuclear inner membrane Asi ubiquitin ligase complex (Fig. 1d) 12 . To determine whether the impaired turnover of the model substrate was caused by the absence of Asi1, we co-expressed Guk1-7-GFP with ASI1 under its endogenous promoter or with a control empty vector (EV) in asi1∆ cells. Addition of wild type ASI1 did not re-establish normal model substrate degradation levels in these cells (Fig. 1e). We obtained similar results with the six remaining E3 ligase deletions that displayed stabilization values comparable with ASI1 (DAS1, FAP1, HRT3, HUL5, UFD2, and UFD4) ( Supplementary Figs 1 and 2). This data implied that the decrease in Guk1-7-GFP degradation conferred by these strains could not be attributed to the absence of the assessed E3 ubiquitin ligases, but rather to that of another factor, such as a background mutation, or potentially an epigenetic factor. Similarly, we previously isolated several hits in a genome wide screen for factors involved in degradative PQC (e.g., YER071C and YJL141C) that we could not confirm after addback experiments expressing the deleted gene from a plasmid, indicating that this phenomenon was not limited to E3 ligase mutant cells 8 .
We next sought to determine whether the observed phenotype was caused by a single background mutation. Therefore, we performed tetrad analysis on the haploid spores obtained from backcrossing the MATa asi1∆ strain to wild type MATalpha BY4742 cells. No discernable difference in growth rate was seen across the dissected spores. The selection marker KanMX, which confers resistance to geneticin in asi1∆ cells, and the impaired degradation phenotype did not appear to be linked. However, each segregated to the expected 2:2 ratio as is shown by a representative tetrad set (Fig. 2a). We obtained similar data with das1∆ cells (Supplementary Fig. 3). Data from CHX chase experiments confirmed that Guk1-7-GFP levels were significantly higher in spore c (ASI1) compared to those in the parental wild type and spore d cells after a two hour (p = 0.001 and p = 0.002) and four hour (p = 0.025 and p = 0.028) incubation at 37°C (Fig. 2b). These experiments indicate that the stabilization phenotype in the asi1∆ strain was likely due to background mutations at a single locus.
Next, we performed a complementation test to determine whether the secondary mutations responsible for the stability phenotype observed in the seven E3 ligase deletion strains are in the same locus, or different loci. Heterozygous diploids were produced by mating either a wild type strain (BY4741) or each of the seven E3 ligase deletions to two haploid strains derived from the asi1∆ backcross: one that presumably contained the secondary mutation (spore a in Fig. 2a), and the other that did not (spore b). Turnover of Guk1-7-GFP was normal in heterozygous diploids produced from mating spore a and BY4741 and diploids produced from crossing spore b with any of the E3 deletion mutants, or the wild type BY4741 (Fig. 2c). Conversely, all heterozygous diploids derived from mating E3 ligase deletions with spore a (harbouring the secondary mutation) demonstrated increased Guk1-7-GFP stability. These results indicate that these background mutations belong to the same complementation group, and suggest that the secondary mutations present in each strain are in the same gene.
To identify the locus containing the secondary mutation, we performed whole-genome sequencing on four haploid spores and their parental wild type and asi1∆ strains. Secondary mutations in the genes GLO4 and WHI2, which are approximately 3,000 bp apart on chromosome fifteen, co-segregated with the strains harbouring the increased Guk1-7-GFP stability phenotype. Interestingly, secondary mutations in the general stress response gene WHI2 have been identified previously in yeast knockout collections and genome evolution studies [13][14][15][16] . We identified a single nucleotide deletion in the coding sequence of WHI2. This mutation, hereinafter referred to as whi2-sc1, produces a frameshift introducing a premature stop codon and likely results in a loss of WHI2 function (Fig. 2d). By contrast, the coding sequence of GLO4, a mitochondrial glyoxalase, contained a single missense mutation.

Proteostasis is impaired owing to secondary mutations in WHI2.
To determine whether the mutation in WHI2 caused the observed stabilization, we expressed the WHI2 wild type open reading frame (ORF) from a plasmid in cells derived from the backcross. Whereas the addition of an empty vector did not rescue the phenotype, addition of WHI2 re-established normal Guk1-7-GFP degradation levels (Fig. 3a). Likewise, degradation of Guk1-7-GFP was impaired in the null whi2∆ strain from the yeast knockout collection, and degradation could be restored by the expression of WHI2 (Fig. 3b). Intriguingly, we found that glo4∆ cells had a similar reduction in Guk1-7-GFP degradation. Subsequent sequencing of the WHI2 locus in glo4∆ cells identified two point mutations that produce a premature stop codon, and expression of WHI2 in glo4∆ cells re-established normal turnover of Guk1-7-GFP ( Supplementary Fig. 3). These results suggest that the effect observed in glo4∆ cells is Scientific RepoRts | 7: 4183 | DOI:10.1038/s41598-017-04525-8 attributed to a loss of WHI2 function, not of GLO4, and that loss of WHI2 function is sufficient to strongly impair degradation of a misfolded cytosolic model substrate.
We next sought to confirm that WHI2 was also mutated in the other E3 ligase deletion strains in which Guk1-7-GFP degradation was impaired. Mutations in WHI2 sensitize cells to exposure to acetic acid, which lends itself to a convenient assay for Whi2 function 13 . The whi2∆ and all seven E3 ligase deletion strains were sensitive to acetic acid treatment ( Supplementary Fig. 3). Expressing WHI2 from a plasmid under its endogenous promoter restored cell viability in all strains, confirming data from the complementation test, suggesting that all strains contain secondary mutations in the same locus (Fig. 2c). We proceeded to sequence the WHI2 gene, including approximately one hundred base pairs upstream and downstream of the start and stop codons, in all twenty of the Mutations in WHI2 segregate with the Guk1-7-GFP stability phenotype. (a) Analysis of a representative tetrad derived from the mating of MATa asi1∆ and MATalpha BY4742 cells. Growth was assessed on the indicated plates (top) and the relative loss of fluorescence (Rel. LoF) of Guk1-7-GFP was determined following incubation with CHX at 25°C or 37°C for 2 hours (bottom). Results represent three independent experiments and p values were calculated with a one-way ANOVA and post-hoc Tukey HSD to assess significance, **denotes p < 0.01. (b) Spores c and d, produced from the asi1∆ backcross shown in (a), expressing Guk1-7-GFP were incubated with CHX at 25°C or 37°C for four hours. Samples were analysed by flow cytometry at the indicated time points. The results represent the mean and standard deviation of three independent experiments. P values were calculated with a two-tailed unpaired Student's t-test (*, **, and ns denote p < 0.05, 0.01, and not significant, respectively). (c) Complementation test in which the indicated diploids expressing Guk1-7-GFP were incubated with CHX for two hours at 25°C or 37°C and analysed by flow cytometry. MATa E3 ligase deletion strains or WT BY4741 were mated with MATalpha asi1∆ cells from spore a (red) and b (black). (d) Whole genome sequencing of BY4741, asi1∆, and the four asi1∆ backcross spores (shown in a) revealed a single base pair deletion in the coding sequence of WHI2 that co-segregates with the Guk1-7-GFP stability phenotype.
top E3 ligase deletion strains from our screen. Whereas ubr1∆ and six other strains had no apparent mutations, we identified unique WHI2 mutations in a total of thirteen strains (Fig. 3c). Of the WHI2 mutations identified, eleven are predicted to produce truncated proteins resulting from the introduction of a premature stop codon. Four of the eleven also contain additional C-terminal extensions (ranging from 4 to 31 amino acids in length) Figure 3. Absence of WHI2 leads to Guk1-7-GFP stability. (a) Relative loss of fluorescence (Rel. LoF) of Guk1-7-GFP in cells from spores c and d that carried a control empty vector (EV) or WHI2 were incubated with CHX at 25°C or 37°C for two hours and then analysed by flow cytometry. Results represent three independent experiments and p values were calculated with a one-way ANOVA and post-hoc Tukey HSD to assess significance (ns and ** denote not significant and p < 0.01, respectively). (b) Wild type and whi2∆ cells expressing Guk1-7-GFP along with an empty vector (EV) control or WHI2 were treated and analysed as in (a). (c) Positions of the WHI2 mutations in the indicated strains together with the predicted protein lengths depicted in red. Black boxes denote the C-terminal mismatch extensions. (d) Relative turnover of the indicated mutated proteins in WHI2 and whi2-sc1 cells that were treated and analyzed as in (a). Results represent three independent experiments. P values were calculated with a two-tailed unpaired Student's t-test (** and *** denote p < 0.01 and 0.005, respectively).
as the result of frameshift mutations. The mutations identified are relatively evenly dispersed along the length of the protein with the exception of a mutation free region, seventy amino acids in length, which was found approximately three quarters of the way into the protein. The presence of WHI2 mutations is unlikely to be restricted to E3 ligase mutant cells. Indeed, three other studies have reported the presence of secondary mutations in the WHI2 locus of strains from the yeast deletion collection 14,16,17 . Mutations in WHI2 have been speculated to provide a favourable fitness advantage under certain environments or in combination with other compensatory mutations 15,18 . However, we did not observe differences in the growth of haploid cells following tetrad analysis.
We next determined whether the degradation of other misfolded cytosolic proteins was impaired due to a loss of WHI2 function. We assessed the levels of three other GFP fusion mutants: Gus1-3, Pro3-1, and Ugp1-3 that are encoded by thermosensitive alleles of glutamyl-tRNA synthetase, delta-1-pyrroline-5-carboxylate reductase, and UDP-glucose pyrophosphorylase, respectively. We previously demonstrated that these mutant cytosolic proteins are degraded upon shifting cells to a higher temperature 8,19 . As for the Guk1-7 mutant, degradation of all three alleles was impaired in whi2-sc1 cells, although Pro3-1 to a lesser extent than the others (Fig. 3d). Altogether, these results indicate that loss of WHI2 function has a varied impact on the degradation of misfolded cytosolic proteins and that some quality control substrates are more dependent on the presence of Whi2.
Mutant WHI2 reduces substrate ubiquitination. To determine how an absence of WHI2 results in increased Guk1-7-GFP stabilization, we first needed to clarify what aspect of PQC is altered in the mutants. We first compared the thermodynamic stability of ectopically expressed Guk1-7 in cellular lysates by a cellular thermal shift assay (CETSA) 20 . Briefly, aliquots of cell lysate were heated to different temperatures, cooled, and then centrifuged to separate soluble fractions from precipitated proteins. Our protein of interest, Guk1-7, could then be detected in the soluble fraction by western blotting to generate thermal melt curves. The presence of Guk1-7 in the soluble fraction decreased rapidly at temperatures above 42°C in extracts from both wild type and whi2-sc1 strains (Fig. 4a). While not marked, slightly more Guk1-7 remained in the soluble fraction of whi2-sc1 lysates compared to wild type at 46.5°C and 48.8°C (p = 0.04 and p = 0.032). We next asked whether an absence of Whi2 influences protein folding by assessing NP-40 solubility. There was no significant difference in Guk1-7-GFP solubility between wild type and whi2-sc1 cells for cultures grown at 25°C, or following a short twenty minute incubation at 37°C (p = 0.18 and p = 0.07) (Fig. 4b). We further examined Guk1-7-GFP by fluorescence microscopy. After a twenty minute incubation at 37°C in the presence of CHX, Guk1-7-GFP was sequestered into inclusions that appear adjacent to, or overlapping with, nuclear staining in both wild type and whi2-sc1 cells (Fig. 4c). Consistent with our flow cytometry data (Fig. 1d), following a two hour incubation at 37°C in the presence of CHX, Guk1-7-GFP accumulated in large inclusions in whi2-sc1 cells (63% of cells contained an average of 1.2 inclusions, n = 72) whereas less Guk1-7-GFP was visible in WHI2 cells, which contained smaller inclusions (48% of cells contained an average of 1.5 inclusions, n = 67) (Fig. 4d). Together, these data suggest that an absence of WHI2 leads to slower clearance of Guk1-7-GFP without increasing thermal stability or promoting substrate solubility.
It is possible that mutations in WHI2 might generally alter the ubiquitin proteasome system. However, using the known proteasome substrate Deg1-GFP, we found no significant difference in its degradation in whi2-sc1 cells compared to wild type cells (p = 0.31, 0.40, 0.70, 0.52 for 10, 20, 30, and 60 minute time points, respectively) (Fig. 4e). We next asked whether an absence of WHI2 could affect Guk1-7-GFP ubiquitination. Quantification of ubiquitin levels was performed in three independent experiments and measured following pulldown of Guk1-GFP and Guk1-7-GFP from cultures grown at 25°C. Normalizing the ubiquitin levels to the amount of GFP tagged substrate eluted revealed that the poly-ubiquitination signal of Guk1-7-GFP in whi2-sc1 cells was significantly reduced compared to wild type cells ( Fig. 4f; average 32% reduction ± 6%, p = 0.0008, n = 3). Together, these data suggest that mutated WHI2 could impair Guk1-7-GFP degradation, in part by decreasing substrate ubiquitination.
Reduced proteostasic capacity in WHI2 mutants is linked to Msn2. Exposure to stressors such as heat, oxidative or osmotic shock, and nutrient starvation results in the transcriptional activation of approximately 200 genes in yeast 21 . Activation of this general stress response is mediated by binding of the partially redundant zinc finger transcription factors Msn2 and Msn4 to stress response elements (STRE) in the promoters of stress-response genes 21,22 . Studying the general stress response, mediated by Msn2/4 signalling, has gained renewed importance in light of a recent study showing that the majority of genes induced by heat shock in yeast are activated through Msn2/4 activity and not by Hsf1, as was previously believed 21 . Msn2 has been shown to form a complex with the plasma membrane phosphatase Psr1 and Whi2 23 . Therefore, we wished to assess the stability of Guk1-7-GFP in msn2∆ cells to determine whether the mutant WHI2 phenotype was potentially the result of an altered general stress response. Notably, we observed that an absence of MSN2, but not MSN4, led to a significant increase in Guk1-7-GFP compared to wild type (p = 0.003) with levels similar to those seen in whi2∆ cells (Fig. 5a). We confirmed that WHI2 was not mutated in msn2∆ cells. We then examined Guk1-7-GFP stability in the double msn2∆whi2∆ mutant and observed no additive effect (Fig. 5b). These results indicate that the reduced turnover in whi2∆ cells and in msn2∆ cells was likely caused by the disruption of the same pathway Therefore, the decreased Guk1-7-GFP degradation in the absence of Whi2 is likely associated to a general impairment of stress response factors acting downstream of Msn2. To confirm that mutants of WHI2 affect the Msn2 transcription factor, we assessed whether protein levels of Msn2 responsive genes were altered in the absence of Whi2 function. Therefore, we endogenously tagged Hsp12, Hsp42, and Hsp104 with GFP in wild type and whi2-sc1 strains, as these three heat shock proteins have been shown to be targets of Msn2 and their expression is induced following heat shock 24 . Steady state protein levels of Hsp12 and Hsp42 were approximately 90% lower in whi2-sc1 cells compared to wild type when examined by western blot (Fig. 5c). Levels of Hsp104 appear to be less affected by a lack of Whi2 function. Flow cytometry experiments support the finding of lower steady state levels of Hsp12 (Fig. 5c) at 25°C in whi2-sc1 cells compared to wild type. Hsp42 and Hsp104 showed a similar trend, but their levels were much lower in both wild type and whi2-sc1 cells (Supplementary Fig. 4). In wild type cells, levels shift assay of Guk1-7 fused to a six histidine tag in lysates of WHI2 and whi2-sc1 cells grown at 25°C. One representative anti-His western blot is shown. The graph represents the means and standard deviations of Guk1-7 levels from three independent experiments. P values were calculated with a two-tailed unpaired Student's t-test (*denotes p < 0.05). (b) WHI2 and whi2-sc1 cells expressing Guk1-7-GFP were grown at 25°C or shifted to 37°C for 20 min. Total cell lysate (T), soluble (S), and pellet fractions (P) were immunoblotted with an anti-GFP antibody. One representative blot is shown and the ratio of soluble fraction to total cell lysate is noted and represents the mean and standard deviation of three independent experiments. Although, slightly lower levels were recovered in the soluble fractions of whi2-sc1 cells, the differences with wild type cells were not significant. (c) WHI2 or whi2-sc1 cells expressing Guk1-7-GFP were grown at 25°C and then incubated at 25°C or 37°C for 20 minutes with CHX before fixation and imaging. The scale bar represents 5 µm. (d) As in (c) except cells were incubated in the presence of CHX for 2 hours at 25°C or 37°C. (e) Turnover of a constitutive substrate of the UPS. WHI2 and whi2-sc1 cells expressing Deg1-GFP under the Cup1 promoter were incubated with CHX at 30°C and samples were collected at the indicated time points. Membranes were immunoblotted with an anti-GFP antibody and an anti-Pgk1 antibody as a loading control. The graph represents the mean and standard deviation of three independent experiments. (f) Guk1-GFP or Guk1-7-GFP were immunoprecipitated of GFP fluorescence increased for all three Hsp proteins following incubation at 37°C (but not in the presence of CHX), whereas GFP fluorescence levels were lower in whi2-sc1 cells (Fig. 5d and Supplementary Fig. 4). These data suggest that the steady state levels of the assessed Msn2 stress responsive genes are altered in whi2-sc1 cells, even in the absence of stress.
The activity of the Msn2/Msn4 transcription factors is regulated and fine tuned by multiple factors including the glucose-responsive cyclic AMP (cAMP)-protein kinase A 25 . Intriguingly, while performing WHI2 addback experiments we noticed a pronounced decrease in Guk1-7-GFP stability when adding a second plasmid bearing the auxotrophic marker leucine. To further investigate this observation, we transformed wild type and whi2-sc1 cells with a plasmid expressing Guk1-7-GFP that contained one of the following selection markers: histidine, uracil, or leucine. Consistent with our previous data, the loss of Guk1-7-GFP fluorescence was 53% lower in whi2-sc1 cells compared to wild type when the histidine marker was used (Fig. 6a). However, when whi2-sc1 cells containing the leucine selection marker were grown in synthetic media without additional leucine, Guk1-7-GFP degradation persisted and levels were approximately two fold lower. Uracil selection resulted in an intermediate phenotype. Leucine was previously shown to activate the TORC1 kinase complex that can also inhibit Msn2/4 [26][27][28][29] . TOR is a conserved serine threonine protein kinase which, as part of the TORC1 complex, promotes growth by linking protein synthesis to extrinsic signals such as nutrient levels and environmental stresses. Treating cells with the antifungal rapamycin mimics nutrient starvation and stress, thereby inhibiting TORC1 and allowing a number of stress and growth responsive transcription factors (such as Msn2) to translocate into the nucleus 30 . One possibility is that Whi2 is only required to maintain Msn2 active when TORC1 is stimulated in the presence of high levels of exogenous leucine. In support of this view, impaired Guk1-7-GFP degradation was restored in whi2-sc1 cells upon the addition of increasing amounts of leucine (Fig. 6b). Conversely, when TORC1 was inhibited in the presence of rapamycin, Guk1-7-GFP degradation in whi2-sc1 cells was comparable to that of wild type (Fig. 6c). These data would suggest that Whi2 is required for Msn2 function under conditions where TOR is active.

Discussion
Temperature sensitive alleles have proven to be fruitful model substrates when used to elucidate the existence and function of PQC pathways. In this study, we used the thermally unstable Guk1-7 allele as a model PQC substrate to identify the E3 ubiquitin ligase, or ligases, responsible for it's temperature dependent degradation. Screening a yeast deletion collection using a flow cytometry based approach resulted in the identification of a number of putative E3 ligase hits. Further validation, however, suggested that the phenotype observed was produced by an indirect effect. Subsequent complementation analysis and whole genome sequencing revealed secondary mutations in the general stress response factor WHI2 that account for the observed increase in Guk1-7-GFP stability. WHI2 promotes Guk1-7-GFP degradation via substrate ubiquitination with no effect on solubility or thermal stability.
We previously identified the E3 ubiquitin ligase Ubr1 from a genetic screen for factors involved in degradative protein PQC of Guk1-7 8 . Ubr1 activity alone, however, was not sufficient to account for the bulk of substrate degradation. In some cases, the nuclear E3 ligase San1 and cytosolic ligase Ubr1 have been shown to act in parallel to degrade cytosolic substrates 5,19 . We found no role for San1, and no additive effect with the double ubr1∆san1∆ mutant in the degradation of the Guk1-7 substrate. Therefore, in this study we wanted to identify the E3 ubiquitin ligases, in addition to Ubr1, that are responsible for the proteasomal degradation of Guk1-7. We identified a number of potential hits in our screen which included: ASI1, DAS1, FAP1, HRT3, HUL5, UFD2, and UFD4. ASI1 is a RING domain family member localized to the inner nuclear membrane and is part of the Asi complex that acts as a branch of the ERAD degradation pathway independent from HRD3 and DOA10 31,32 . Both DAS1 and HRT3 are putative F-box SCF ubiquitin ligases 33,34 . A homologue of the human transcription factor NF-X1, FAP1 also confers resistance to rapamycin by acting as a ligand for FKBP12 35 . HUL5 is a member of the HECT ubiquitin ligase family. Involved in cytoplasmic PQC of short-lived misfolded proteins, it is also necessary for the increased ubiquitination observed as part of the heat shock quality control response 36 . UFD2 is both an E3 and E4 enzyme with mutants being hypersensitive to protein misfolding stressors 37 . Finally, UFD4, like HUL5, is also a member of the HECT family of E3 ligases and physically interacts with Ubr1 to increase processivity of ubiquitin chain formation in the N-end rule pathway 38 . While an absence of these seven genes resulted in significant stabilization of our model substrate, we were surprised to have identified some of them in our screen. Guk1 is found in both the cytoplasmic and nuclear compartments of yeast cells as assessed by GFP tagging and fluorescence microscopy 8,39 . It was, therefore, not entirely unexpected for some hits to be nuclear proteins. However, based on the ascribed functions for some of the ligases, and the nature of our substrate, it was puzzling to have identified ASI1, DAS1, FAP1, and HRT3 in our screen.
Once our seven top hits failed to validate in addback experiments, we then went on to identify secondary mutations in the coding sequence of the general stress response gene WHI2 in thirteen of twenty E3 ligase deletion strains tested. A null WHI2 deletion phenocopied the mutants, therefore suggesting that these mutations lead to a loss of Whi2 function. Mutations in WHI2 have been reported in laboratory based evolution studies and are speculated to provide a favorable fitness advantage under certain environments or in combination with other compensatory mutations 15,18 . However, we did not observe striking differences when comparing growth of haploid cells on YPD following tetrad analysis. To our knowledge, ours is the first report of WHI2 mutations with GFP-Trap beads from lysates of WHI2 or whi2-sc1 cells grown at 25°C and expressing a control empty vector, Guk1-GFP, or Guk1-7-GFP. Samples were eluted and immunoblotted with anti-ubiquitin, anti-GFP, and anti-Pgk1 antibodies. Ubiquitination levels were normalized to levels of unconjugated Guk1-7-GFP that was immunoprecipitated and averaged from three independent experiments. Values denote averaged relative quantification levels plus or minus standard deviations. Uncropped images for all Western blots can be found in Supplementary Fig. 5. in E3 ubiquitin ligase deletion strains. Nevertheless, the presence of WHI2 mutations is unlikely restricted to E3 ligase mutant cells, as we have demonstrated here. Three other studies have reported the presence of secondary mutations in the WHI2 locus of strains from the yeast deletion collection 14,16,17 . These reports highlight the confounding effects of suppressor mutations on genetic studies performed using yeast deletion collections. In the most recent study, approximately 30% of all strains sequenced carried unique mutations in WHI2 and/or five other genes 16 . Most of the mutations identified were frameshift or nonsense mutations, suggesting a loss of function phenotype as we have shown in this report. Finally, it was found that serially passaging whi2∆ strains under conditions with a prolonged stationary phase resulted in an increased abundance of the deletion strains relative to a wild type control. This suggests that secondary mutations in these genes might be found at higher frequencies as the result of selecting for mutants that delay the onset of the stationary phase under laboratory conditions, such as those used to create the yeast deletion collection.
As part of the general stress response, Whi2 forms a complex with the plasma membrane phosphatase Psr1 and the zinc finger transcription factor Msn2 23 . Upon exposure to stress, Msn2 translocates into the nucleus where it can bind to stress response elements (STRE) in promoters of stress responsive genes. We found that like WHI2, an absence of MSN2 had a positive effect on Guk1-7-GFP stability. We also demonstrated that some Msn2 stress responsive genes have lower steady state levels in strains lacking Whi2 function. This suggests that the increase in Guk1-7-GFP stability observed in WHI2 mutants is mediated by factors downstream of Msn2. Alternatively, WHI2 may be responsible for Msn2 stability and an absence of, or mutations in, WHI2 could result in lower levels of Msn2 and therefore an altered general stress response in these strains. Studying the general stress response, mediated by Msn2/4 signalling, has gained renewed importance in light of a recent study examining the role of Hsf1 in the transcriptional response to heat shock. The authors of that study concluded that the majority of genes induced by heat shock in yeast are activated through Msn2/4 activity and not by Hsf1, as was previously believed 21 .
While performing WHI2 addback experiments, we observed that Guk1-7-GFP stability depended upon adequate levels of leucine being present in the media. Amino acids, especially leucine, have been shown to regulate the TOR signalling pathway 29 . TOR is a conserved serine threonine protein kinase, which as part of the TORC1 complex promotes growth by linking protein synthesis to extrinsic signals such as nutrient levels and Figure 5. Msn2 is linked to reduced proteostatic capacity in WHI2 mutants. (a) Wild type, whi2∆, msn2∆, and msn4∆ cells expressing Guk1-7-GFP were analysed by flow cytometry following a two hour incubation at 25°C or 37°C in the presence of CHX. P values were calculated with a one-way ANOVA and post-hoc Tukey HSD to assess significance (*, **, and ns denote p < 0.05, 0.01, and not significant, respectively). (b) Wild type, whi2∆, msn2∆, and msn2∆whi2∆ cells expressing Guk1-7-GFP were analysed by flow cytometry following a two hour incubation at 25°C or 37°C in the presence of CHX. P values were calculated with a one-way ANOVA and post-hoc Tukey HSD to assess significance (** and ns denote p < 0.01 and not significant, respectively). (c) WHI2 and whi2-sc1 strains with either Hsp12, Hsp42, or Hsp104 endogenously tagged with GFP were grown at 25°C. Membranes were immunoblotted with an anti-GFP antibody and an anti-Pgk1 antibody as a loading control. (d) WHI2 and whi2-sc1 strains with Hsp12 endogenously tagged with GFP were grown to log phase and then incubated at 25°C, 37°C, or 37°C in the presence of CHX for 2 hours prior to anlysis by flow cytometry. Uncropped images for all Western blots can be found in Supplementary Fig. 5. environmental stresses. Treating cells with the antifungal rapamycin mimics nutrient starvation and stress, thereby inhibiting TORC1 and allowing a number of stress and growth response transcription factors (such as Msn2) to translocate into the nucleus 30 . We propose a model in which Whi2 is required to maintain Msn2 functionality when TORC1 is active (e.g., in the presence of high concentrations of leucine in the media). Under these conditions, mutations in WHI2 would result in Msn2 phosphorylation and inhibition of the downstream induction of stress responsive genes, thereby preventing degradation of misfolded cytosolic proteins such as Guk1-7. Conversely, when TORC1 activity is decreased or inhibited (e.g., in LEU2 cells deprived of exogenous leucine or in the presence of rapamycin) Msn2 is activated in a WHI2 independent manner and Guk1-7 is degraded with similar dynamics as in wild type cells. Our findings of a link between TORC1 and the Whi2 general stress response in the degradation of our model substrate might help explain some puzzling results we obtained from a previous screen we conducted with Guk1-7-GFP 8 . In our previous screen we had a high false positive rate and a number of hits which, at the time, were unexpected. Two such hits were the kinases Yak1 and Rim15, which have been shown to directly phosphorylate Msn2 in vitro and to translocate from the cytosol into the nucleus following mTORC1 inhibition 40,41 . There is one striking point that remains unanswered: Msn2 is thought to be mostly active after a stress to induce expression of stress response genes, but in our conditions translation of newly expressed genes would be prevented by cycloheximide. Therefore, the impaired turnover of our model substrate is likely the result of an imbalance of the proteostatic network prior to the stress (i.e., Msn2 basal activity is also required in unstressed conditions). Indeed, protein steady state levels of several Msn2 targets (e.g. Hsp12) were lower in whi2-sc1 cells.

Materials and Methods
Yeast strains, media, and growth conditions. The Saccharomyces cerevisiae strains used in this study are listed in Table S1. Yeast strains were cultured in synthetic media with 2% dextrose (lacking the appropriate amino acids for plasmid selection) or YPD (1% yeast extract, 2% peptone, 2% dextrose) and grown at 25°C with shaking unless indicated otherwise. When not specified otherwise, cultures in log phase were obtained by diluting overnight saturated cultures grown at 25°C to an OD 600 = 0.2 and grown for 4-6 hours until log phase OD 600 = 0.8-1.0 was reached.
Plasmids. Plasmids used in this study are listed in Table S3. Guk1-GFP (BPM453), Guk1-7-GFP (BPM458), and Guk1-7-His 6 (BPM717) expressed from the GPD1 promoter in pRS313 were generated in a previous study 8 ; Guk1-7-GFP was subcloned with ApaI and SacI sites in pRS315 to generate BPM609 and with XhoI and SacII Figure 6. Deficient proteostatic response under conditions where the TOR pathway is active. (a) Guk1-7-GFP was expressed from CEN/ARS plasmids with histidine, uracil, or leucine auxotrophic markers in wild type or whi2-sc1 cells. Cultures were incubated with CHX for two hours at 25°C or 37°C before being analysed by flow cytometry. (b) Wild type and whi2-sc1 cells were co-transformed with Guk1-7-GFP and pRS315 (LEU2). Cultures were grown in synthetic drop out media containing the indicated amounts of leucine and then incubated for two hours in the presence of CHX at 25°C or 37°C before flow cytometry analysis. (c) WHI2 and whi2-sc1 cells were handled as in (a) except for being treated with 200 nM rapamcyin for two hours at 25°C prior to the two hour incubation at 25°C or 37°C. sites in pRS316 to generate BPM781. To generate the E3 ligase addback plasmids (BPM748, ASI1; BPM749, DAS1; BPM750, FAP1; BPM751, HRT3; BPM752, HUL5; BPM753, UFD2; BPM754, UFD4), the open reading frames and approximately 500 bp of both endogenous 5′ and 3′ UTR was PCR amplified from genomic DNA (BY4741) and inserted into pRS316 using XhoI and XmaI sites for all but HUL5 where SacII and XhoI sites were used. The WHI2 (BPM863, BPM914) addback plasmids were generated as for the E3 ligases except ligated into pRS315 or pRS316 using SacI and XmaI sites, respectively. Flow cytometry. Cells in log phase were treated with 100 µg/mL cycloheximide and incubated at either 25°C or 37°C as indicated. GFP fluorescence was measured for 50,000 cells using a FACSCalibur (BD) flow cytometer. Median GFP fluorescence values were obtained using FlowJo software. For chase experiments, percentage remaining values were calculated by normalizing the median GFP fluorescence intensity values for each time point to the initial t = 0 measurement. To calculate the relative loss of fluorescence for single time-point measurements, the difference of GFP fluorescence values for samples incubated at 25°C or 37°C was normalized to that of the 25°C sample. To perform multiple strain comparisons, the relative loss of fluorescence values (as calculated above) for each strain was normalized to that of the wild type BY4741 strain.
Sequencing. Whole-genome sequencing and library preparation was performed at the NextGen Sequencing facility at the University of British Columbia's Biodiversity Research Centre. Yeast cells were grown overnight to saturation in YPD at 25°C and genomic DNA was extracted using standard protocols 42 . Barcoded libraries for each strain were created according to Illumina protocols (Illumina 2011, all rights reserved) and 100-bp paired end fragments were sequenced by pooling all six libraries and run on a single lane of an Illumina HiSeq2000. The short-read aligner BWA was used to map sequence reads to the yeast reference genome S288C version R64 (Saccharomyces Genome Database, SDG, http://www.yeastgenome.org) 43 . Single-nucleotide variants (SNVs) were identified using the SAMtools toolbox and then each SNV was annotated with a custom-made Perl script using gene data downloaded from SDG on January 21, 2014 44 . IGV viewer was used to visually inspect read alignments in the regions of candidate SNVs 45, 46 . WHI2 plate assay. Yeast cultures were grown overnight at 25°C in 5 mL YPD to OD 600 = 1-2 and then diluted to OD 600 = 0.2 in 5 mL YPD and left to grow for 2 hours at 25°C. 1 mL was kept as an untreated control and the remaining 4 mL of culture was treated with 200 mM acetic acid for 4 hours at 25°C. Treated and untreated cultures were serially diluted fivefold in 1X PBS and plated on solid media. Plates were incubated at 30°C for 2 days before being imaged with a Gel Doc XR + System (Bio-Rad). per 3 mg of lysate) for 2 hours at 4°C. Beads were washed three times in lysis buffer before samples were eluted with 3X SDS buffer. Equal volumes of samples were resolved by SDS-PAGE. Membranes were immunoblotted with mouse anti-GFP (Roche, 1:2,500), rabbit anti-Pgk1 (Acris Antibodies, 1:10,000), and mouse anti-ubiquitin (Millipore, 1:2,500) primary antibodies and secondary antibodies (Mandel Scientific, 1:10,000).
Fluorescence microscopy. Cultures grown to log phase were divided and then incubated at either 25°C or 37°C for 20 minutes and then fixed in 3.7% formaldehyde for 15 minutes at room temperature. Fixed cells were rinsed with 1X PPS (0.1 M potassium phosphate, pH 7.5 with 1 M sorbitol) and then permeabilized with 0.1% Triton X-100 in 1X PPS for 10 minutes. For nuclear staining, permeabilized cells were incubated for 10 minutes with 1 µg/mL Hoechst 33342 in 1X PPS. Slides were prepared by mounting cells in 1X PPS and covered using concanavalin A treated coverslips. Imaging was performed with a 63X oil immersion objective on an inverted Zeiss Axio Observer microscope. Global contrast was adjusted using Zeiss Zen Pro software and images were then cropped for presentation in ImageJ.
Statistical analysis. Data are presented as mean ± SD unless otherwise stated. Comparisons were made using the two-tailed Student's t-test and differences were considered significant at a p-value of <0.05. When indicated, multiple strains were compared with a one-way ANOVA and post-hoc Tukey HSD to assess significance.