A genome-wide screen identifies overexpression of YLR053c as a rescue of cln3Δ bck2Δ lethality

We constructed a cln3Δ bck2Δ whi5::GAL1-WHI5 strain that is viable in glucose but not galactose medium due to conditional WHI5 expression [12,13]. We used synthetic genetic array (SGA) methodology to cross this query strain to an array of 5,280 strains that each of contained a GAL1-GST-ORF 2-μm high copy plasmid [40–42] and assessed the growth of the resulting array on galactose medium (Fig 1A and 1B). Three replicates of the screen identified 12 genes that reproducibly rescued the cln3Δ bck2Δ lethal Start arrest (S1 Table). These genes included the G1 cyclins CLN1 and CLN3 but not other known Start activators such as CLN2, BCK2, SWI4, and SWI6, possibly due to the high level of WHI5 expression used in our screen, and/or overexpression toxicity of particular genes. We recovered 10 genes not known to be Start regulators. Notably, the short hypothetical ORF YLR053c restored cln3Δ bck2Δ growth to the same extent as CLN3, as validated by direct transformation of the query strain with GAL1-YLR053c and GAL1-CLN3 constructs (Fig 1C). Other prospective high copy rescue genes, such as YEA4 (Fig 1C), could not be confirmed by direct transformation.

Download:

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

Fig 1. A genome-wide screen identifies YLR053c/NRS1 as a dosage suppressor of cln3Δ bck2Δ lethality.

(A) Schematic of SGA genetic screen for dosage suppressors of cln3Δ bck2Δ lethality. (B) Representative screen plate scored for growth on galactose medium. Red box, GAL1-YLR053c. (C) Comparison of growth at 30°C for the indicated strains streaked onto either glucose or galactose medium. YEA4 was a candidate hit that did not validate. (D) Chromosomal region around YLR053c/NRS1 on Chr. XII and translated 108 amino acid (12.7 kDa) protein sequence. (E) Ylr053c/Nrs1 protein sequence in S. cerevisiae (top) aligned with sequences of other yeast species. Conservation of a KKXNPFYVPSXVVREMV motif at the carboxyl terminus is indicated by a red bar. NRS1, Nitrogen-Responsive Start regulator 1; SGA, synthetic genetic array.

https://doi.org/10.1371/journal.pbio.3001548.g001

YLR053c encodes a 108 residue microprotein that is only poorly characterized. Microproteins are often encoded by newly evolved proto-genes that are thought to form a genetic reservoir that fuels adaptive evolution [43]. To gain insight into YLR053c locus evolution, we aligned the Ylr053c protein sequence from S. cerevisiae with predicted orthologs from other yeast species (Fig 1D and 1E). To estimate the extent to which YLR053c locus evolved across these species, we used the standard dN/dS metric that measures the ratio of single DNA site substitution rates at nonsynonymous codons (dN) versus synonymous codons (dS). The YLR053c locus appears to have evolved relatively rapidly within the Saccharomyces sensu stricto group, with a dN/dS ratio in the 98th, 79th, and 90th percentile in Saccharomyces mikatae, Saccharomyces bayanus, and Saccharomyces castellii, respectively, as compared with S. cerevisiae (see S1A Fig). In comparison, the evolution of other core Start regulators was closer to the genome median, with dN/dS ratios in the 74th, 77th, and 58th percentile for WHI5, 69th, 75th, and 31st percentile for SWI4, and 82nd, 51st, and 43rd percentile for SWI6. A 17 amino acid sequence at the Ylr053c carboxyl-terminal region was conserved even in more distant yeasts such as Kluyveromyces waltii (Fig 1E). Given its genetic role at Start and expression pattern (see below), we renamed the YLR053c gene NRS1 for Nitrogen-Responsive Start regulator 1.

Nrs1 is induced by rapamycin and nitrogen limitation

To understand the function of Nrs1, we first sought to characterize its endogenous expression at the protein level. We performed scanning Number and Brightness (sN&B) confocal microscopy to localize and quantify an Nrs1-GFPmut3 fusion protein at the subcellular scale in live wild-type cells under a range of conditions [22,44]. GFPmut3, a monomeric fast-folding GFP mutant [45], will be referred to here as GFP for brevity. Nrs1-GFP was not detected in cells grown on either SC + 2% glucose or SC + 2% raffinose medium (Fig 2A), consistent with previous analysis of YLR053/NRS1 mRNA levels [46]. In contrast, Nrs1-GFP was readily detected in the nucleus of cells grown overnight to mid log-phase in nitrogen-limited (YNB + 0.4% proline + 2% glucose, abbreviated YNB+Pro) medium (Fig 2A), also in accord with published genome-wide transcriptome analyses under various nutrient conditions [47] and the presence of binding sites for the Gln3 transcription factor in the NRS1 promoter [48,49]. Induction of Nrs1 upon the switch to nitrogen-limited YNB+Pro medium required a long incubation time, as Nrs1-GFP signal above autofluorescence background could not be detected after 7 hours in YNB+Pro but was clearly visible after 22 hours growth to mid-log phase in YNB+Pro (S1B Fig). Inhibition of TORC1 with 100 nM rapamycin induced Nrs1-GFP expression and nuclear localization within 1 hour in rich SC + 2% glucose medium (Fig 2A), as also evident by immunoblot analysis of an endogenously tagged Nrs1^13MYC strain treated with rapamycin (S1C Fig). These expression patterns were confirmed with an independent fusion of Nrs1 to wild-type GFP (S1D Fig). Hyperosmotic stress, oxidative stress, and DNA damage did not induce Nrs1 expression in SC + 2% glucose medium (S1E and S1F Fig). Immunoblot analysis of a Nrs1^13MYC strain in different carbon and nitrogen sources, as well as in rapamycin-treated and saturated cultures, confirmed the specific expression of Nrs1 only under conditions of nitrogen limitation or TORC1 inhibition (Fig 2B). A slow-migrating form of Nrs1 was induced by nitrogen limitation (Fig 2B) and also appeared first upon rapamycin treatment before conversion to the fast-migrating form (S1C Fig). To investigate the nature of this presumptive posttranslational modification (PTM) on Nrs1, we treated Nrs1^MYC immunoprecipitates with lambda phosphatase but found that this did not affect the slower-migrating species (Fig 2C). We also did not detect Nrs1-derived phosphopeptides by mass spectrometry (see below; S4 Table). We have not been able to determine the nature of this modification on Nrs1.

Download:

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

Fig 2. Nitrogen limitation and rapamycin treatment induce cell cycle–regulated Nrs1 expression.

(A) sN&B images of untagged and NRS1-GFP strains. Cells were grown and imaged in SC + 2% glucose with or without 100 nM rapamycin, SC + 2% raffinose or nitrogen-limited medium (YNB +0.4% proline + 2% glucose + His, Leu, Met, Ura; labeled YNB+Pro), as indicated. Scale bar is 10 μm. The same intensity scale was used for all conditions. (B) Abundance of Nrs1^13MYC in various nutrient conditions as determined by immunoblot of anti-MYC immune complexes. raf, 2% raffinose; gly, 2% glycerol; eth, 2% ethanol; C-, no carbon source; sat, saturated culture; rap, 100 nM rapamycin; pro, 0.4% proline; N-, no nitrogen source; glu, 2% glucose. IgG indicates antibody light chain. A raw image of the original immunoblot is provided in S1 Raw Images. (C) Nrs1^MYC immunoprecipitates from a rapamycin treatment time course were either mock-treated (M, negative control), treated with lambda phosphatase (P), or lambda phosphatase + phosphatase inhibitors (P+I) prior to detection by anti-MYC immunoblot. In vitro phosphorylated recombinant Sic1 was used as a control to demonstrate activity of the phosphatase. (D) Absolute Nrs1 concentration in single cells grown and imaged in nitrogen-limited medium (YNB+Pro) as a function of cell size, as determined by sN&B. Blue and orange dots represent individual cells from 2 different experiments. The typical size range of cells at the G1/S transition (800 to 1,000 pixels, corresponding to 27 to 38 fL) is indicated. Cell-averaged total Nrs1 concentration (top) and nuclear concentration where Nrs1 nuclear localization was evident (bottom) are shown. Infrequent small cells with high Nrs1 levels had high autofluorescence and no nuclear localization of the signal. (E) Example of high-content confocal image of Nrs1-GFP cells grown to log phase in nitrogen-limited medium. Arrows indicate example cells with strong (green), dim (blue), or no nuclear Nrs1 (red) or inviable cells (white). The histogram summarizes Nrs1 signals in small unbudded (708 cells), large unbudded and small-to-medium budded (599 cells) and medium-to-large budded cells (186 cells) from 18 confocal images. Inviable cells with high autofluorescence, out-of-focus cells for which the budding pattern could not be ascertained and regions in which illumination was not homogeneous were not scored. All numerical values underlying panels D and E may be found in S1 Data. NRS1, Nitrogen-Responsive Start regulator 1; sN&B, scanning Number and Brightness; YNB+Pro, YNB + 0.4% proline + 2% glucose.

https://doi.org/10.1371/journal.pbio.3001548.g002

Nrs1 abundance peaks at the G1/S transition

We quantified Nrs1-GFP levels in asynchronous populations of live cells grown in nitrogen-limited YNB medium with our custom sN&B analysis software [22]. Nrs1-GFP concentrations, averaged over entire single cells, were between 5 and 20 nM. Higher levels were observed in cells close to the typical critical size at the end of G1 phase, where Nrs1-GFP was predominantly nuclear. In these cases, Nrs1 nuclear concentration was 60 to 80 nM (Fig 2D). These Nrs1 nuclear levels at Start were comparable to Swi4 (50 to 100 nM), Whi5 (100 to 120 nM), and Swi6 (130 to 170 nM) levels determined previously by the same method [22]. We further confirmed Nrs1-GFP expression and nuclear localization principally in large unbudded and small-to-medium budded cells in high-content images acquired by standard confocal microscopy (Fig 2E). The cell cycle–regulated expression pattern of Nrs1 protein suggested that it plays a role at Start, consistent with suppression of the cln3Δ bck2Δ arrest by NRS1 overexpression.

NRS1 genetically interacts with SBF and other Start regulators

To investigate the mechanism by which Nrs1 promotes Start, we examined genetic interactions of NRS1 with the main known Start regulators. Overexpression of NRS1 from the GAL1 promoter, either integrated at the NRS1 locus or from a 2-μm high copy plasmid [50], caused a pronounced small size phenotype, in agreement with its putative role as a Start activator (Fig 3A and 3B). To interrogate the genetic requirements for this size phenotype, we transformed deletion mutants of known Start regulators with the GAL1-NRS1 high copy plasmid and examined cell size epistasis. NRS1 overexpression almost entirely rescued the large size phenotype of a cln3Δ mutant and partially rescued the large size of a bck2Δ mutant (Fig 3B). A whi5Δ mutant was epistatic to NRS1 overexpression, whereas a nrs1Δ deletion did not exacerbate the larger size caused by WHI5 overexpression and only modestly increased the size of a whi5Δ mutant (Fig 3C). Notably, the small size caused by NRS1 overexpression was abrogated in swi6Δ and swi4Δ mutant strains (Fig 3D). This requirement for full SBF function was further demonstrated with a temperature-sensitive swi4-ts strain in which Swi4 binding to Swi6 is altered [51], grown at the semipermissive temperature of 30°C (Fig 3E).

Download:

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

Fig 3. NRS1 is genetically upstream of SBF.

Cell size distributions were determined for the indicated genetic combinations. (A) NRS1 overexpression alone. (B) NRS1 overexpression with either cln3Δ or bck2Δ mutations. (C) NRS1 overexpression and nrs1Δ mutation with either WHI5 overexpression or whi5Δ mutation. (D) NRS1 overexpression with either swi4Δ or swi6Δ mutations. (E) NRS1 overexpression with a swi4-ts mutation. Strains were transformed with either GAL1-NRS1, GAL1-WHI5, or empty vector control high copy plasmids as indicated. Strains bearing galactose-regulated constructs and associated controls were induced for 6 hours in SC + 2% galactose before size determination. WT plots shown in panels B, C, and D (left, middle) represent the same measurement. All numerical values underlying this figure may be found in S2 Data. NRS1, Nitrogen-Responsive Start regulator 1; SBF, SCB-binding factor; WT, wild-type.

https://doi.org/10.1371/journal.pbio.3001548.g003

In contrast to its overexpression, deletion of NRS1 did not detectably affect overall growth rate or cell size in a wild-type S288C (BY4741) laboratory strain, regardless of whether cells were grown in either nitrogen-replete or nitrogen-limited medium (S2A–S2D Fig). Moreover, in competitive growth experiments, nrs1Δ and wild-type cells had indistinguishable fitness (S1 Text and S2E Fig). Because the small size phenotype caused by NRS1 overexpression depended on SWI4 and SWI6, we hypothesized that MBF might partially compensate for loss of NRS1 function. We therefore generated an mbp1Δ nrs1Δ double mutant strain and evaluated growth in nitrogen-rich SC and nitrogen-poor YNB+Pro media. While the single mutant strains had no growth defect in either condition, the mbp1Δ nrs1Δ double mutant had a pronounced growth defect that was specific to nitrogen-poor conditions (Fig 4A). We furthermore considered the possibility that laboratory strains may have lost some requirements for NRS1 through the course of decades-long propagation in artificially rich nutrient conditions. To test this idea, we deleted NRS1 in the prototrophic wild yeast Saccharomyces boulardii, which shares >99% of its genome with S. cerevisiae including NRS1 [52]. An S. boulardii strain lacking NRS1 grew slower than wild-type cells in nitrogen-poor medium specifically (i.e., YNB+Pro+0.1% glucose medium not supplemented with amino acids, Fig 4B). Hence, under conditions of nitrogen limitation, NRS1 promotes growth in genetically crippled contexts in laboratory strains and is required for optimal growth of a wild variant of S. cerevisiae. Taken together, these genetic interactions suggested that Nrs1 promotes the G1/S transition by acting upstream of the Whi5-inhibited form of SBF in a manner that is independent of Cln3 and Bck2.

Download:

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

Fig 4. NRS1 is required for optimal growth in poor nitrogen medium.

(A) Growth curves for WT (BY4741), nrs1Δ, mbp1Δ, and mbp1Δ nrs1Δ laboratory strains grown in nitrogen-limited YNB+Pro+0.1% glucose medium and nitrogen-rich SC+0.1% glucose medium. Curves represent the average of 3 different clones for each mutant strain. (B) Growth curves for WT and nrs1Δ S. boulardii strains were grown in either nitrogen-poor YNB+Pro+0.1% glucose minimal medium (i.e., not supplemented with any other amino acids) or nitrogen-rich SC+0.1% glucose medium. Three different isolates for the nrs1Δ S. boulardii strain were analyzed. Growth curves are the average of at least 12 independent colonies for each strain. To eliminate potential nutrient lag effects, cultures were rediluted into fresh medium, and their growth was monitored again. All numerical values are provided in S3 Data. NRS1, Nitrogen-Responsive Start regulator 1; WT, wild-type; YNB+Pro, YNB + 0.4% proline + 2% glucose.

https://doi.org/10.1371/journal.pbio.3001548.g004

Nrs1 interacts with SBF in vivo and in vitro

We next sought to identify Nrs1 protein interactors under conditions in which Nrs1 was endogenously expressed at physiological levels. We performed immunoaffinity purification on lysates from an endogenously tagged NRS1^13MYC strain and a control untagged strain grown in nitrogen-limited YNB+Pro medium. Analysis of the samples by mass spectrometry (see Methods and S1 Text) identified multiple peptides (S4 Table) from which the corresponding proteins were identified (>2 unique peptides, false discovery rate [FDR] <1%, S5 Table). Proteins specific to the Nrs1^13MYC sample were identified by subtracting proteins present in the untagged control and filtering hits against the CRAPome database of nonspecific interactions [53]. This workflow yielded 7 proteins that were specific to the Nrs1^13MYC sample (S3 Table and S3A Fig). Of these, Nrs1 was represented by 7 peptides (48% coverage), Swi4 by 5 peptides, and Swi6 by 5 peptides; the remaining 4 candidates were represented by only 2 or 3 peptides and thus of lower confidence. This unbiased analysis demonstrated that endogenous Nsr1 expressed under physiological conditions interacted with SBF. To confirm this result, we performed co-immunoprecipitation experiments on extracts of rapamycin-treated cells expressing either SWI4^3FLAG or SWI6^3FLAG alleles in combination with either NRS1^13MYC or WHI5^13MYC alleles, each expressed from their respective endogenous promoter. Swi4^3FLAG and Swi6^3FLAG were each immunoprecipitated with anti-FLAG resin and then blotted with antibodies against either FLAG or MYC epitope tags. Nrs1^13MYC and Whi5^13MYC were detected in immunoprecipitates of endogenous Swi4^3FLAG and Swi6^3FLAG from cells grown the presence of rapamycin but not in control immunoprecipitates from strains that lacked the FLAG-tagged alleles (Figs 5A and S3B). A similar experiment carried out with polyclonal anti-Swi4 and anti-Swi6 antibodies specifically detected endogenous SBF in Nrs1^13MYC immunoprecipitates from cells grown in the presence but not the absence of rapamycin (S3C Fig). Collectively, these results suggested that Nrs1 directly interacted with SBF under physiological conditions.

To determine if the interaction between Nrs1 and SBF was direct, we carried out in vitro binding assays with purified recombinant proteins. We titrated recombinant ^FLAGSwi4-Swi6 and ^FLAGMbp1-Swi6 complexes produced in baculovirus-infected insect cells [12] against recombinant ^GSTNrs1 produced in E. coli and immobilized on GSH-Sepharose resin. Swi4 and Swi6 were both captured with ^GSTNrs1 across the titration series, whereas control GSH-Sepharose resin did not bind SBF (Fig 5B). In contrast, ^GSTNrs1 did not detectably interact with the ^FLAGMbp1-Swi6 complex under the conditions tested, suggesting that specificity for Nrs1 is determined by the DNA-binding subunit and not the common Swi6 subunit of SBF/MBF. To test whether Nrs1 was able to interact with Whi5-bound SBF, we titrated purified recombinant ^GSTNrs1 into a preformed recombinant ^GSTWhi5-^FLAGSwi4-Swi6 complex. We observed that Nrs1 bound effectively even in the presence of Whi5 (Fig 5C). These results demonstrated that endogenous Nrs1 interacts directly and specifically with the Swi4-Swi6 complex, that no additional factors are needed for this interaction to occur, and that Nrs1 can bind SBF even in the presence of Whi5.

Download:

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

Fig 5. Nrs1 binds to SBF in vivo and in vitro.

(A) Swi4^3FLAG or Swi6^3FLAG complexes were immunoprecipitated from the indicated strains grown in the presence of 200 nM rapamycin for 2 hours and interacting proteins assessed by immunoblot with the antibodies against the indicated tags or proteins. Co-immunoprecipitation of Whi5^13MYC with Swi4^3FLAG and Swi6^3FLAG served as a positive control. Mr markers (M) are indicated for each blot. Pgk1 served as a loading control. Note the lower Mr form of Whi5 visible in the input blot is obscured by the IgG heavy chain signal in the immunoprecipitations. Data is representative of at least 5 independent experiments (see S3B Fig for a replicate experiment and S3C Fig for a reciprocal Nrs1^13MYC and Whi5^13MYC co-immunoprecipitation experiment performed with polyclonal anti-Swi4 and anti-Swi6 antibodies). (B) Recombinant ^GSTNrs1 immobilized on GSH-Sepharose resin was incubated with increasing concentrations of soluble purified SBF (^FLAGSwi4-Swi6) or MBF (^FLAGMbp1-Swi6). Bound proteins were analyzed by immunoblot with anti-Swi6 and anti-FLAG antibodies as indicated. GSH-Sepharose resin alone served as a negative control. (C) Recombinant ^FLAGSwi4-Swi6-^GSTWhi5 complexes containing 0.1 ug of ^GSTWhi5 were incubated with increasing amounts of recombinant His₆-tagged Nrs1 (0.1 ug, 0.2 ug, and 0. 5ug, equivalent to 4, 8, and 20 Nrs1:Whi5 molar ratio), immunoprecipitated using anti-FLAG beads to capture SBF–Whi5 complexes and probed with anti-GST antibody to detect Whi5 and anti-HIS₆ antibody to detect Nrs1. Co-immunoprecipitation with an irrelevant recombinant protein complex (Cdc4-^FLAGSkp1) on anti-FLAG beads served as a negative control for interaction specificity. Ponceau S stain was used to demonstrate equivalent input protein complexes in each lane. Raw images of all original immunoblots are provided in the S1 Raw Images. MBF, MCB-binding factor; NRS1, Nitrogen-Responsive Start regulator 1; SBF, SCB-binding factor.

https://doi.org/10.1371/journal.pbio.3001548.g005

Nrs1 binds to SBF-regulated promoters in vivo

We next assessed the presence of Nrs1^13MYC at the SBF-regulated promoters of CLN2 and PCL1 by chromatin immunoprecipitation (ChIP). For both genes, we specifically detected SCB-containing promoter sequences by PCR in cross-linked anti-MYC immune complexes purified from cells that expressed Nrs1^13MYC from the endogenous NRS1 locus (Fig 6A). The enrichment of PCL1 and CLN2 sequences in Nrs1^13MYC immunoprecipitates only in rapamycin-treated cells further demonstrated that Nrs1 specifically binds SBF-regulated G1/S promoter DNA (Fig 6A).

Download:

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

Fig 6. Nrs1 binds to SBF-regulated promoter DNA and activates SBF-driven transcription.

(A) Anti-MYC chromatin immunoprecipitates from WT untagged control and NRS1^13MYC strains either untreated or treated with 200 ng/mL rapamycin for 2 hours were probed for the presence of CLN2 and PCL1 promoter DNA sequences by real-time quantitative PCR. Bars indicate the mean fold-enrichment across 2 replicates, and error bars show the standard error on the mean. (B) Top panel: RICS vertical correlations from a NRS1-GFP strain grown in nitrogen-limited medium (YNB+Pro) as a function of the pixel shift for total and cytosolic Nrs1-GFP pools. RICS correlation for free GFP was used as a control for unconstrained diffusion. Data points show correlation averages over N FOV (N = 25 for nuclear Nrs1, N = 7 for cytosolic Nrs1, N = 12 for free GFP). Each FOV contained 1 to 5 cells. Error bars represent the standard error on the mean. Bottom panel: Scatter plots of fitted diffusion coefficients from individual FOVs of the same strains. Data from Swi6-GFP FOVs served as a control for constrained diffusion. All numerical values underlying panels A and B may be found in S5 Data. (C) Heatmaps of gene expression profiles upon galactose induction of GAL1-NRS1 and WT control strains. Genes were selected based on a gene rank score across biological triplicates and values shown are upper quartile-normalized log₂-read counts averaged across triplicates for each time point. Genes that were strongly up-regulated (gene score > 0.5) and down-regulated (gene score < −0.5) are shown. Genes were clustered for GAL1-NRS1 and WT heatmaps separately according to the default settings of the R Heatmap function. Positions of NRS1 (*), CLN2 (**), and PCL1 (***) are indicated. SBF (red), MBF (blue), and SBF/MBF (purple) target genes are indicated on the left side. Raw data underlying this panel may be found on the Gene Expression Omnibus database under the accession number GSE179366. FOV, field of view; MBF, MCB-binding factor; NRS1, Nitrogen-Responsive Start regulator 1; RICS, raster image correlation spectroscopy; SBF, SCB-binding factor; WT, wild-type; YNB+Pro, YNB + 0.4% proline + 2% glucose.

https://doi.org/10.1371/journal.pbio.3001548.g006

We predicted that Nrs1 binding to chromatin should reduce Nrs1 mobility in the nucleus. To test this hypothesis, we assessed the molecular dynamics of Nrs1-GFP in cells grown in nitrogen-limited medium using raster image correlation spectroscopy (RICS). The RICS method exploits the hidden time structure of imaging scans, in which adjacent pixels are imaged a few microseconds apart, to quantify the diffusional properties of fluorescent molecules [54–56]. Intensity correlations between pixels shifted along the direction of scanning (horizontal correlations), and along the orthogonal direction (vertical correlations), are averaged over multiple scans. Horizontal and vertical correlations that decay with increasing pixel shift are characteristic of the dynamical properties on 10 to 100 μs and 5 to 50 ms timescales, respectively. We found that the vertical RICS correlations of nuclear Nrs1-GFP signal decreased on a slower timescale than either free nuclear GFP or cytosolic Nrs1-GFP signal (Fig 6B). Fitted Nrs1-GFP diffusion coefficients in cells in which Nrs1 was predominantly nuclear were lower compared to cells in which it was mostly cytosolic and also lower than the diffusion coefficient of free nuclear GFP, a proxy for free diffusion in the nucleus. The Nrs1 diffusion coefficient was similar to that of Swi6-GFP which is largely DNA bound (Fig 6B). This result indicated that Nrs1 associates with a slow-moving nuclear component, most likely chromatin.

The apparent size epistasis between whi5Δ and GAL1-NRS1 prompted us to ask whether Nrs1 might reduce the association Whi5 with SBF and/or chromatin. However, ChIP analysis of Whi5^HA at the CLN2 and PCL1 promoters revealed that Whi5^HA–promoter interactions were not reduced in the presence of overexpressed NRS1 (S4A Fig). In addition, GAL1-NRS1 cells had wild-type levels of Whi5 as determined by sN&B quantification (S4B Fig), and GAL1-NRS1 did not alter Whi5^HA association with Swi4 or Swi6 in co-immunoprecipitation experiments (S4C Fig). Nrs1 and Whi5 also did not compete for SBF binding in in vitro binding assays with recombinant proteins (S4D Fig). Together, these results obtained using different in vivo, in vitro, biochemical, and imaging-based approaches suggested that Nrs1 binds SBF at G1/S promoters in vivo, but that the binding of Nrs1 does not alter Whi5 interactions with SBF or promoter DNA. These results led us to consider the possibility that Nrs1 might directly activate transcription.

Ectopic expression of NRS1 activates the G1/S regulon

To assess the potential role of Nrs1 in transcription, we used RNA sequencing (RNA-seq) to determine the genome-wide transcriptional response to ectopic NRS1 overexpression. Three biological replicates of nrs1::GAL1-NRS1 and wild-type cells were grown to log-phase in SC+2% raffinose medium, induced with galactose for 6 hours, and transcriptional profiles analyzed by RNA-seq. Gene scores were defined based on their expression fold-change before and after 6-hour induction, comparing for each gene the fold-change across multiple replicates to control for intersample expression variability and the effects of galactose (see Methods; all scores provided in S6 Table). Up-regulated genes in the GAL1-NRS1 samples included NRS1 itself as expected (rank 1) and many of the 139 SBF/MBF target genes that comprise the G1/S regulon as defined by integration of multiple microarray-based analyses of G1/S regulated genes [10]. These G1/S genes included CLN2 (rank 12) and PCL1 (rank 82), the promoters of which were bound by Nrs1 as shown above. Nrs1 also strongly induced the expression of HO (rank 10), RNR1 (rank 15), and HCM1 (rank 20), which are markers of the G1/S transition. Visualization of the data in a heatmap format illustrated the overall enrichment of SBF/MBF target genes (Fig 6C). Among the most strongly up-regulated genes (score >0.5), SBF targets were enriched 12-fold compared to random expectation (24 of 86 genes, p = 3*10⁻²⁰), MBF targets were enriched 7-fold (17 of 101 genes, p = 9*10⁻¹¹), but out of the 17 strongly up-regulated MBF targets, 16 were also SBF targets so that MBF-only targets were not enriched (1/53 genes, p = 0.65). In the wild-type control samples, no significant enrichment/depletion compared to random expectations was observed. We note that in addition to recognition of its cognate SCB elements, SBF can also activate genes that only contain MCB elements [57] such that the apparent specificity of Nrs1 for SBF is not undermined by the up-regulation of MBF genes in these experiments.

Nrs1 confers transcriptional activity that can rescue G1/S-transcription deficient mutants

We next asked whether Nrs1 might itself function directly as a transcriptional activator. To test this hypothesis, we constitutively expressed a Nrs1-Gal4 DNA-binding domain fusion protein (GAL4^DBD-NRS1) in a reporter strain bearing the HIS3 gene under control of the GAL1 promoter. We used GAL4^DBD alone and GAL4^DBD fused to an irrelevant human gene (UBE2G2) that potently transactivates as negative and positive controls, respectively. Full-length NRS1, but not a truncated version that encoded only the conserved carboxyl terminus, was able to activate transcription of the HIS3 reporter and thereby allow cell growth in medium lacking histidine (Figs 7A and S5A).

Download:

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

Fig 7. Nrs1 has an inherent transcription activation function that can partially rescue G1/S-transcription deficient mutants.

(A) Transactivation of a HIS3 reporter by fusion of Nrs1 to the Gal4 DNA-binding domain (GAL4^DBD-NRS1) but not by fusion of the Nrs1 carboxyl terminus (GAL4^DBD-NRS1^Cter). GAL4^DBD-UBE2G2 and GAL4^DBD constructs served as positive and negative controls, respectively. Growth curves were determined in -His -Trp medium. All numerical values underlying this panel may be found in S6 Data. (B) Serial 5-fold dilutions of NRS1 and GAL1-NRS1 strains in WT, swi4-ts, mbp1Δ, and mbp1Δswi4-ts backgrounds were spotted onto SC + 2% glucose, SC + 2% raffinose and SC + 2% galactose, and grown for 5 days at 30°C. C1 to C4 indicate 4 independent clones of the mbp1Δ swi4-ts GAL1-NRS1 strain. (C) Serial 5-fold dilutions of WT or swi6Δ strains transformed with GAL1-WHI5, GAL1-WHI5^12A, GAL1-SWI6^SA4, GAL1-NRS1, or empty control plasmids or combinations thereof were spotted on rich medium containing 2% glucose or 2% galactose and grown for 2 days at 30°C. NRS1, Nitrogen-Responsive Start regulator 1; WT, wild-type.

https://doi.org/10.1371/journal.pbio.3001548.g007

Given the inherent transactivation capacity of Nrs1, we next investigated whether NRS1 overexpression could suppress the growth defects caused by loss of other G1/S activators. We first tested a cln1Δ cln2Δ cln3Δ MET-CLN2 strain, in which G1 cyclin activity is restricted to methionine-free media, with or without a <pGAL1-NRS1> plasmid. Repression of CLN2 by methionine caused a growth arrest, whether or not NRS1 was expressed, although ectopic expression of NRS1 did modestly reduce cell size and slightly increased cell proliferation (S5B Fig). Hence, NRS1 cannot compensate for the total lack of G1-cyclin activity. We then tested whether NRS1 overexpression could suppress the growth defects of mutants in which G1/S transcription was impaired, but in which Cln-Cdc28 activity was preserved. We first tested a mbp1Δ swi4-ts double mutant [58]. The growth of a GAL1-NRS1 mbp1Δ swi4-ts strain was improved with respect to the mbp1Δ swi4-ts control at a semipermissive temperature of 30° in galactose medium (Fig 7B; see S5C Fig for plate images with a 2-day extension of the incubation period), but not in other media or at the fully permissive temperature (S5B Fig). As a further test, we examined NRS1-mediated suppression in strains that expressed WHI5^12A and SWI6^SA4 alleles in which all Cdc28 consensus sites are mutated, a combination that abrogates the Cln-Cdc28 mediated relief of SBF inhibition by Whi5 [12,14]. Overexpression of NRS1 also partially rescued the growth defect of this strain (Fig 7C). These results suggested that Nrs1 provides an alternative mechanism of Start activation by augmenting G1/S transcription in the presence of Whi5-mediated inhibition.

Nrs1 can bypass Whi5 inhibition of SBF

To test whether physiological levels of Nrs1 targeted to SBF are sufficient to promote Start activation, we constructed a strain expressing a Whi5-Nrs1-GFP fusion protein from the endogenous WHI5 promoter. Experiments with this strain were carried out in SC complete medium to avoid other possible nitrogen-dependent parallel inputs to Start and to focus on Nrs1 function. The chimeric protein was expressed and had the expected molecular mass (S6A Fig) and retained the localization pattern of endogenous Whi5, with about 30% to 35% of pre-Start cells showing a prominent nuclear signal (Figs 8A and S6B). Cells expressing Whi5-Nrs1-GFP were almost as small as whi5Δ cells (Fig 8B). This phenotype was not due to the presence of the GFP tag since size distributions of strains expressing untagged versus GFP-tagged Whi5 were indistinguishable (S6C Fig). Moreover, cells expressing the chimeric protein grew at the same rate as wild-type and whi5Δ strains (S6D Fig). Strikingly, cells expressing Whi5-Nrs1-GFP were as small as cells that expressed GAL1-NRS1 in galactose medium (Fig 8C).

Download:

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

Fig 8. Tethered Nrs1 bypasses the Whi5-dependent lethal arrest of a cln3Δ bck2Δ strain.

(A) sN&B microscopy images of WHI5-GFP and WHI5-NRS1-GFP cells grown in SC + 2% glucose. Scale bars are 5 μm. The same intensity scale was used on both images. Fractions of pre-Start G1 cells were computed based on the assessment of nuclear GFP signal in WHI5-GFP cells (N = 245) and WHI5-NRS1-GFP (N = 95) cells. Absolute concentrations of Whi5-GFP and Whi5-Nrs1-GFP are shown in the plot. (B) Cell size distributions of WT, GAL1-NRS1, whi5Δ, and WHI5-NRS1-GFP strains in SC + 2% glucose. (C) Cell size distributions of WT, GAL1-NRS1, whi5Δ, and WHI5-NRS1-GFP strains in SC + 2% galactose. All numerical values underlying panels A–C may be found in S7 Data. (D) Genotype of 10 tetrads from a cln3Δ WHI5-NRS1-GFP X bck2Δ cross (top) and a cln3Δ WHI5-GFP X bck2Δ cross (bottom). For each tetrad, spore clone growth was assessed on SD-LEU (indicates cln3::LEU2), SC+NAT (indicates bck2::NAT^R), and SD-HIS (indicates whi5::WHI5-NRS1-GFP-HIS3 or whi5::WHI5-GFP-HIS3). Colored boxes indicate viable cln3Δ bck2Δ spore clones, all of which also contained the WHI5-NRS1-GFP construct. (F) Simplified schematic for Nrs1-dependent activation of Start. Red lines indicate nitrogen-limited conditions. Not all components of the Start machinery are shown. See text for details. NRS1, Nitrogen-Responsive Start regulator 1; sN&B, scanning Number and Brightness; WT, wild-type.

https://doi.org/10.1371/journal.pbio.3001548.g008

As a reduction of Whi5 dosage in the Whi5-Nrs1-GFP fusion context could in principle explain the small size of this strain, we used sN&B to compare the nuclear concentrations of Whi5 in the wild-type and the Nrs1-fusion contexts. Whi5 levels in wild-type cells were 110 to 130 nM, consistent with those determined previously [22]. The Whi5-Nrs1-GFP chimera was present at a slightly lower concentration, 80 to 100nM (Fig 8A), close to endogenous Nrs1 nuclear levels in nitrogen-limited media (60 to 80 nM, see Fig 2D). This slight decrease in Whi5 abundance was unlikely to explain the small size of cells expressing Whi5-Nrs1-GFP since cell size is not strongly sensitive to WHI5 gene dosage [25,59]. We confirmed that hemizygous WHI5/whi5Δ diploid cells were only marginally smaller than wild-type diploid cells (S6E Fig) and, moreover, WHI5 overexpression only causes a 20% to 30% increase in mode size (Fig 3C), in agreement with previous results [25,59]. Consistently, our previously published Start model [22] also predicted that down-regulating Whi5 levels from 120nM to 85nM should not affect the critical size at Start (S6F Fig).

Finally, we predicted that the Whi5-Nrs1 fusion should be sufficient to rescue the lethal Start arrest of a cln3Δ bck2Δ strain. We crossed a cln3Δ strain bearing the integrated WHI5-NRS1-GFP allele to a bck2Δ strain and analyzed growth of dissected tetrads on selective media to identify spore genotypes (Fig 8D). We recovered many viable cln3Δ bck2Δ WHI5-NRS1-GFP triple mutant spore clones but did not recover any viable cln3Δ bck2Δ WHI5 double mutant clones (Fig 8D). As expected, no viable cln3Δ bck2Δ spore clones were obtained from a control cross of a cln3Δ whi5::WHI5-GFP and a bck2Δ strain (Fig 8D). As a further control to ensure that the Nrs1 fusion did not merely inactivate Whi5 in a nonspecific manner, we fused the above transcriptionally inactive carboxyl-terminal fragment of Nrs1 to Whi5 and showed that it did not reduce cell size (S7A Fig) nor rescue the cln3Δ bck2Δ lethality (S7B Fig). These results supported a model in which Nrs1 bypasses Whi5 inhibition by directly conferring transactivation activity on the Whi5-inhibited SBF complex.

Yeast strains construction and culture

All S. cerevisiae strains were isogenic with the S288C (BY4741) auxotrophic background (S2 Table). Standard molecular genetics methods were employed for genomic integration of carboxyl-terminal tagging cassettes [91] and verified by PCR or sequencing. Standard media were used for yeast growth: rich (XY: 2% peptone, 1% yeast extract, 0.01% adenine, 0.02% tryptophan); synthetic complete (SC: 0.17% YNB, 0.2% amino acids, 0.5% ammonium sulfate); or nitrogen-limited (0.17% YNB, 0.4% proline, supplemented with histidine, leucine, methionine and uracil to complement auxotrophies as needed). Prototrophic S. boulardii strains were grown in SC or nitrogen-limited (0.17% YNB, 0.4% proline) without amino acid supplements. To delete NRS1 in S. boulardii, the strain MYA-796 was transformed with pGZ110-Cas9-amdSYM-nrs1_sgRNA and plated on acetamide-containing medium. Deletion alleles were confirmed by DNA sequencing of the coding region and the pGZ110 plasmid was counter-selected on fluoroacetamide medium [92]. Carbon sources were added to 2% w/v as indicated. Unless otherwise specified, cells were grown to saturation in SC + 2% glucose and diluted 1 in 5,000 into fresh medium for 16 to 18 hours growth prior to experiments. These conditions ensured homeostatic growth in log-phase at the time of experiment at a culture density of 2 to 5 10⁶ cells/mL. For strains bearing constructs expressed under the inducible GAL1 promoter, pre-growth was in SC + 2% raffinose. Cell size distributions were acquired using a Beckman Coulter Z2 counter (Beckman Coulter, Brea, CA, USA). Growth curves were acquired at 30°C using a Tecan Sunrise shaker-reader (Tecan Group Ltd., Männedorf, Switzerland).

SGA screen

A cln3Δ bck2Δ whi5::GAL1-WHI5 strain was mated to an array of 5,280 GAL1-ORF fusion strains [41]. Following haploid selection, strains were scored for growth on selective medium containing 2% galactose. Three replicates of this screen were performed. For the third replicate, expression of the GAL4 transcription factor was increased in case it was limiting for expression of WHI5, the ORFs, or GAL genes required for growth in the presence of galactose at the sole carbon source. For this purpose, GAL4 was placed under the control of the ADH1 promoter in a cln3::LEU2 bck2::NAT^R whi5::Kan^R-pGAL1-WHI5 can1d mfa1::MFA1pr-spHIS5+ GAL4p::HphRpADH1-GAL4 query strain. Screen hits were validated by transformation of the GAL1-GST-ORF construct directly into the cln3Δ bck2Δ whi5::GAL1-WHI5 query strain.

Chromatin immunoprecipitation

Cells were grown in XY with 2% raffinose, induced with 2% galactose for 6 hours to an OD₆₀₀ ≤0.5 and fixed with 1% formaldehyde. Whole-cell extracts from 50 mL of culture were prepared by glass bead lysis, sonicated to shear chromatin DNA into fragments, and incubated with the appropriate antibody coupled to magnetic beads (Dynabeads PanMouse IgG). Immunoprecipitated DNA was washed, de-crosslinked, purified, and analyzed by quantitative real-time PCR. Reactions with appropriate oligonucleotides were set-up with SYBR Green PCR Master Mix (Applied Biosystems, Waltham, Massachusetts, USA) and carried out on an ABI 7500 Fast Real-Time PCR System. Enrichment at the CLN2 or PCL1 locus was determined after normalization against values obtained from input samples using SYP1 as the reference gene. ^HAWhi5 and Nrs1^13MYC ChIP experiments were performed in biological duplicates. The bar height on Fig 6A and S4A Fig represent the mean over duplicates, and the error bars represent the standard error on the mean.

Immunoprecipitation and immunoblot analysis

Protein extracts were prepared in lysis buffer (10 mM HEPES-KOH pH 7.9, 50 mM KCl, 1.5 mM MgCl₂, 1 mm EDTA, 0.5 mM DTT, 50 mM NaF, 50 mM sodium pyrophosphate, 1 mM Na₃VO₄ and Roche protease inhibitor cocktail; Roche, Basel, Switzerland) by glass bead lysis. On S3C Fig, the lysis buffer contained 200mM NaCl, 50 mM HEPES pH 7.5, 1 mM EDTA, 1 mM DTT, 1.5 mM MgCl₂, protease inhibitors, 2 mM NaF, and 2 mM Na-pyrophosphate. Due to low expression levels in the 60 to 150 nM range (Fig 2D and [22]), in some cases, we concentrated Nrs1^13MYC by immunoprecipitation prior to immunoblot detection. Immunoprecipitations were carried out at 4°C for 2 hours with indicated antibodies in soluble form (9E10 anti-MYC from EMD Millipore, Burlington, Massachusetts, USA, 05–419; M2 anti-FLAG from Sigma, St Louis, Missouri, USA, F1804) then bound to protein G resin (Pierce, Waltham, Massachusetts, USA, 20398) for 1 hour at 4° C, followed by either 1 or 3 washes with wash buffer (10 mM Tris-Cl pH7.9, 0.1% Triton X-100, 0.5 mM DTT, 0.2 mM EDTA, 10% glycerol, 150 mM NaCl; for Fig 5A, Wash buffer: 10 mM Tris-HCl pH7.9, 0.5 mM DTT, 0.2 mM EDTA, 150 mM NaCl), and resuspension in SDS sample buffer. In Fig 5A, the immunoprecipitations were performed as above except that cell pellets were resuspended in 200 mM NaCl, 50 mM HEPES-KOH pH 7.9, 5mM EDTA, 1 mM DTT, MCE protease inhibitors, 2 mM NaF, 2 mM Na-pyrophosphate, 0.1% NP-40 prior to lysis and complexes captured directly onto 15 μL of M2 anti-FLAG resin (anti-FLAG M2 affinity gel Sigma, A2220). For S3C Fig, the complexes were captured directly onto 15 μL of anti-MYC resin (Pierce, 20168). Proteins were resolved by SDS-PAGE, transferred onto nitrocellulose membrane, and immunoblotted with anti-HA, anti-Swi4 (gift from Brenda Andrews), anti-Swi6 (gift from Kim Nasmyth), anti-GST, anti-MYC 9E10, or anti-FLAG M2 antibodies, followed by detection with the appropriate HRP-conjugated secondary antibody [12,93].

In vitro binding assays

Recombinant GST-Nrs1 fusion protein was affinity purified with GSH-Sepharose 4B (Amersham Biosciences, Little Chalfont, Buckinghamshire, UK) in 50 mM HEPES-NaOH, pH 7.5, 150 mM NaCl, 5 mM EDTA, 5 mM NaF, 0.1% NP-40, 10% glycerol, supplemented with 1 mM PMSF. Recombinant SBF or MBF complex was purified with anti-FLAG resin (M2, Sigma) from insect cells co-infected with ^FLAGSwi4-Swi6 or ^FLAGMbp1-Swi6 baculovirus constructs in buffer supplemented with complete protease inhibitor cocktail (Roche) and then eluted with excess FLAG peptide [12]. Binding reactions were incubated at 4°C for 1 hour with rotation. Washed samples were resolved on SDS-PAGE gel followed by immunoblotting with anti-FLAG (M2, Sigma) and anti-Swi6. SBF-GST-Nrs1 and SBF-HA-Whi5 complexes were preformed in solution and immobilized on glutathione or anti-HA resin, respectively, incubated with GST-Nrs1 or ^HAWhi5, washed and analyzed as described above.

Immunoprecipitation and mass spectrometry analysis

Cell pellets from untagged and Nrs1^13MYC strains from 100 mL of culture at OD₆₀₀ = 1 were lysed in standard lysis buffer (50 mM Tris-HCl pH8.0, 150 mM KCl, 100 mM NaF, 10% glycerol, 0.1% tween-20, 1 mM tungstate, 1 mM DTT, 10 mM AEBSF, 10 mM pepstatin A, 10 mM E-64) [56] supplemented with protease inhibitors using a N₂(l) freezer mill. Lysates (0.5 mL) were incubated for 1 hour with 5 μL of anti-MYC antibody (Gentex, Zeeland, MI, USA) followed by 1-hour incubation with an additional 50 μL of GammaBind plus Sepharose beads (GE Healthcare, Chicago, IL, USA) to capture protein complexes. After multiple washes, samples were separated on Bio-Rad (Hercules, CA, USA) precast gels, and the entire gel lane was cut for each sample (CAPCA core facility, https://capca.iric.ca; see Supporting information Methods). Gel samples were destained, alkylated, and digested with trypsin for 8 hours at 37°C and peptides extracted in 90% ACN. Peptides were separated on a home-made C18 column connected to Q-Exactive HF Biopharma with a 56-minute gradient of 0% to 30% acetonitrile in 0.2% formic acid. Each full MS spectrum of extracted peptides was acquired at a resolution of 120,000, followed by acquisition of 15 tandem MS (MS–MS) spectra on the most abundant multiply charged precursor ions by collision-induced dissociation (HCD). Data were processed using PEAKS X (Bioinformatics Solutions, Waterloo, Ontario, Canada) and the UniProt yeast database. Variable selected posttranslational modifications were carbamidomethyl (C), oxidation (M), deamidation (NQ), acetyl (N-ter), and phosphorylation (STY). Control untagged and Nrs1^13MYC data were analyzed with Scaffold 4.8.9 at an FDR of 1% for at least 2 peptides with a likelihood of at least 99%.

RNA-seq analysis

Wild-type and GAL1-NRS1 strains were inoculated for overnight growth in SC + 2% raffinose at 30°C, then diluted in SC + 2% raffinose to an OD600 of 0.07–0.1, incubated at 30°C until an OD600 of 0.15–0.2 was reached, at which point a sample was collected for RNA extraction prior to galactose induction. For NRS1 induction, the cultures were either concentrated by centrifugation, resuspended in 20 mL SC+2% raffinose and added to 200 mL SC+2% galactose (replicate 2,3) or grown to an OD600 of 0.15 to 0.2 in 200 mL SC + 2% raffinose followed direct addition of 2% galactose to the culture (replicate 1). Time point 0 was collected immediately, followed by time points up to 6 hours. Total RNA was prepared as described in [94]. Briefly, cells were pelleted at 3500 rpm, disrupted using glass beads, phenol/chloroform extracted, and the aqueous phase containing nucleic acids was ethanol-precipitated. Contaminating genomic DNA was removed by DNase treatment (Qiagen, Hilden, Germany). Total RNA was quantified using Qubit (Thermo Scientific, Waltham, MA, USA) and RNA quality assessed with a 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). Transcriptome libraries were generated using the Kapa RNA HyperPrep (Roche) using a Poly-A selection (Thermo Scientific). Sequencing was performed on the Illumina NextSeq 500 system, aiming for 10 million single-end reads per sample, 40 million for raffinose, and final (6 hours) induced time points. RNA-seq experiments were performed in biological triplicate.

Reads were aligned to all Ensembl yeast transcripts using Bowtie 2.2.5 [95] with default parameters. Read counts were tabulated for each gene, only considering alignments with an edit distance no greater than 5. Genes with at least 500 reads in at least 1 sample were included in the analysis. Gene expression levels for each sample, expressed as log2 read counts, were normalized to the upper quartile gene expression level for the same sample, as recommended in Bullard and colleagues [96], and represented as heatmaps Fig 6C.

For each of the 3 GAL1-NRS1 biological replicate samples, the expression after 6 hours galactose induction was normalized separately to the expression in raffinose (prior to induction) and to the wild-type sample from the same replicate after 6 hours induction. The former controls for the sample-to-sample variation, while the latter controls for the effect of the switch to galactose-containing media. To ensure to the best possible extent that observed expression differences are due to NRS1 expression and not to one of the 2 abovementioned sources of variation, the direction and magnitude of the differential expression of each gene had to be reproduced relative to both controls. We thus kept as the final score for each replicate the smallest absolute log2-fold change produced by either of the 2 comparisons and set the score to zero if the signs of the 2 disagreed. The values for each of the 3 replicates were averaged to yield the final gene score (S6 Table). This procedure was designed to control for the effects of galactose medium (i.e., normalizing by 6 hours wild-type samples) and for intersample expression variability (i.e., normalization to expression levels prior to induction of each culture and averaging over replicates). Hence, high scores for GAL1-NRS1 (positive or negative) could be obtained only for genes up-/down-regulated both in galactose compared to raffinose and compared to wild-type cells in galactose. Wild-type 6 hours scores (S6 Table) were obtained in the same fashion except that gene expression in wild-type cells after 6 hours in galactose were normalized separately to wild-type expression in raffinose and to GAL1-NRS1 after 6-hour induction. This control analysis is symmetrical and thus equivalent to the analysis performed on GAL-NRS1 samples.

High-content imaging

High-content images were acquired on an OPERA high-throughput confocal microscope (PerkinElmer, Waltham, MA, USA) equipped with a 60× water objective. Pixel resolution was 220 nm at 2*2 binning. 200 μL of culture were directly transferred to a Greiner Screenstar glass-bottom 96-well imaging plate and imaged within 30 to 45 minutes. GFP was excited with a 488 nm laser (800 ms exposure for endogenously tagged proteins and 320 ms for overexpressed proteins) and detected using a 520/35 nm band-pass filter. For WHI5-GFP and WHI5-NRS1-GFP strains, pre-Start G1 phase cells were identified by nuclear localization of the GFP signal, as computed using a custom image-analysis script written in MATLAB (MathWorks, Natick, MA, USA). The script masks individual cells using threshold-based detection of the autofluorescence background and determines the fraction of individual cells that display a pre-Start nuclear localization of the GFP signal.

sN&B and RICS fluorescence fluctuation microscopy

Live cells in log-phase were imaged on an Alba sN&B system (ISS, Champaign Illinois, USA) comprised of an inverted confocal Nikon Eclipse microscope equipped with a 100× water objective, a Fianium Whitelase continuous white laser with 488 nm emission filters and single photon APD detectors, as described previously [22]. In brief, 1.5 mL of cell culture was pelleted for 2 minutes at 3,000 rpm in a microfuge, resuspended in approximately 100 μL of culture supernatant and 3 μL deposited on a preset 65 μL drop of SC + 2% glucose + 2% agar gel medium on a circular glass coverslip (#1, VWR) that was encircled by an adhesive silicon ring. After 4 minutes drying time, a ConcanavalinA-coated (Sigma, 2 mg/mL) coverslip was gently pressed on top of the agarose to seal the pad against the adhesive silicon ring. Sealed pads were clamped in an AttoFluor chamber (Molecular Probes, Eugene, OR, USA) and immediately imaged for no more than 1.5 hours. Unless otherwise specified, the culture growth medium was reused to make the imaging pads in order to prevent inadvertent nutrient up- or downshifts. For this purpose, 1 mL of cell culture was pelleted at 15,000 rpm for 1 minute, and 500 μL of the supernatant was mixed with 10 mg of agarose and warmed for 1 to 2 minutes at 98°C to melt the agar, followed by application to a coverslip. For sN&B experiments in the presence of rapamycin, growth medium containing 100 nM rapamycin was also used to prepare the pad. To mitigate possible degradation of rapamycin during the pad preparation, an additional 100 nM of rapamycin was added to the pre-warmed agar mix just before making the pad. The final rapamycin concentration was therefore 200 nM, i.e., about 200 ng/mL, similar to rapamycin treatment time courses in Figs 5B and S1C, and the fixed-duration treatment in Fig 5A. Cells were pretreated 1 hour before imaging and imaging lasted about 1.5 hours, yielding a treatment duration between 1 hour and 2.5 hours across the various fields of view (FOVs). There were no significant differences in Nrs1-GFP signals across different FOVs.

sN&B imaging was performed using 20 raster scans of the same 30 μm–wide FOVs of 256 pixels (pixel size 117 nm), using an excitation power of 1 to 2 μW at 488 nm wavelength and a 64 μs pixel dwell time. sN&B images shown in the figures are projections of the 20 raster scans. Protein concentrations were extracted from sN&B data using custom analysis software [22]. RICS imaging was performed in a similar fashion to sN&B imaging but different parameters were used to improve correlation curves (pixel size of 48.8 nm, 50 frames, 20 μs pixel dwell time) as described previously [56]. RICS vertical correlations for individual FOVs were fitted to single mode–free diffusion models using the SimFCS analysis software. FOVs of poor-quality fit were discarded from diffusion coefficients plots.

Bioinformatics

The sequences of orthologs of YLR053c were obtained from the Fungal Orthogroups website. For Saccharomyces paradoxus and Saccharomyces mikatae, where no ORF was predicted at the expected locus, the YLR053c sequence was searched against whole genome assemblies using TBLASTN 2.2.27 to predict orthologs. The synteny of the regions in K. waltii and S. cerevisae was confirmed based on Orthogroups orthology assignments of the upstream (47.17997 and IES3) and downstream (47.18006 and OSW2) genes and the fact that each of these assignments were supported by highly significant Blast e-values. Alignments were generated using ClustalO 1.2.0 and displayed using the MView online tool.

Mathematical model of Start

Whisker and box plots for model predictions of the critical size at Start were obtained using the mathematical model and the data processing procedure described previously [22]. For each plot, 50 individual cells bearing variable Whi5 concentrations randomly picked in the 100 to 140 nM (for the WHI5-GFP strain) and 65 to 105 nM (for the WHI5-NRS1-GFP strain) ranges were simulated and plots generated in Excel using standard statistics of the distributions of critical size values for each strain.