Protein Kinase a Regulates Gene-Specific Translational Adaptation in Differentiating Yeast

Downloaded from rnajournal.cshlp.org on October 3, 2021 - Published by Cold Spring Harbor Laboratory Press

Protein kinase A regulates gene-specific translational adaptation in differentiating yeast

PAVANAPURESAN P. VAIDYANATHAN, BORIS ZINSHTEYN, MARY K. THOMPSON, and WENDY V. GILBERT1 Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA

ABSTRACT Cellular differentiation is driven by coordinately regulated changes in gene expression. Recent discoveries suggest that translation contributes as much as transcription to regulating protein abundance, but the role of translational regulation in cellular differentiation is largely unexplored. Here we investigate translational reprogramming in yeast during cellular adaptation to the absence of glucose, a stimulus that induces invasive filamentous differentiation. Using ribosome footprint profiling and RNA sequencing to assay gene-specific translation activity genome-wide, we show that prolonged glucose withdrawal is accompanied by gene-specific changes in translational efficiency that significantly affect expression of the majority of genes. Notably, transcripts from a small minority (<5%) of genes make up the majority of translating mRNA in both rapidly dividing and starved differentiating cells, and the identities of these highly translated messages are almost nonoverlapping between conditions. Furthermore, these two groups of messages are subject to condition-dependent translational privilege. Thus the “housekeeping” process of translation does not stay constant during cellular differentiation but is highly adapted to different growth conditions. By comparing glucose starvation to growth-attenuating stresses that do not induce invasive filamentation, we distinguish a glucose-specific translational response mediated through signaling by protein kinase A (PKA). Together, these findings reveal a high degree of growth-state specialization of the translatome and identify PKA as an important regulator of gene-specific translation activity. Keywords: adaptation to stress; glucose starvation; PKA signaling; translatome specialization; yeast invasive growth

INTRODUCTION ies have identified hundreds of genes that are specifically required for invasive filamentous growth but are dispensable Free-living microbes must adapt to the availability of specific for rapid growth in the presence of glucose (Chin et al. nutrients in their environments. Remarkably, the yeast Sac- 2012; Ryan et al. 2012), highlighting the profound differences charomyces cerevisiae can adjust its growth and division rate between these two growth states. over a 10-fold range in response to nutrient levels (Brauer In addition to controlling filamentous differentiation, glu- et al. 2008). In addition to regulating growth rate, depletion cose availability regulates sweeping changes in gene expres- of specific nutrients induces adaptive cellular differentiation sion. The Ras/PKA pathway mediates most of the glucose- programs including sporulation, filamentation, and quies- induced transcriptional changes, with additional regulation cence. In haploid yeast, glucose starvation induces invasive by the AMP-activated protein kinase (AMPK) Snf1, the growth, a developmental program in which yeast cells form Hap2/3/4/5 transcription complex, and the Rgt network filamentous networks that invade the growth medium. This (Zaman et al. 2009; Broach 2012). The mRNA levels of is thought to function as a foraging response allowing the ∼40% of yeast genes change ≥2-fold in response to glucose yeast colony to “search” for glucose (Broach 2012; Cullen addition or withdrawal (Brauer et al. 2005; Zaman et al. and Sprague 2012). Invasive filamentous differentiation 2009; Arribere et al. 2011). Thus, the availability of glucose involves changes in cell polarity, actin cytoskeletal organiza- substantially affects the pool of cellular mRNA. tion, cell surface properties, and cell-cycle progression. Two The acute response to glucose withdrawal also profoundly conserved signaling pathways control glucose-regulated fila- affects cellular translation activity. Within 1 min, polysomes mentation–the Ras/PKA pathway and a mitogen-activated collapse and protein synthesis rates plummet (Ashe et al. protein kinase (MAPK) pathway (for review, see Cullen and Sprague 2012). Recent genome-scale gene deletion stud- © 2014 Vaidyanathan et al. This article is distributed exclusively by the RNA Society for the first 12 months after the full-issue publication date 1Corresponding author (see http://rnajournal.cshlp.org/site/misc/terms.xhtml). After 12 months, it E-mail [email protected] is available under a Creative Commons License (Attribution-NonCommer- Article published online ahead of print. Article and publication date are at cial 4.0 International), as described at http://creativecommons.org/licenses/ http://www.rnajournal.org/cgi/doi/10.1261/rna.044552.114. by-nc/4.0/.

912 RNA 20:912–922; Published by Cold Spring Harbor Laboratory Press for the RNA Society Downloaded from rnajournal.cshlp.org on October 3, 2021 - Published by Cold Spring Harbor Laboratory Press

Translatome specialization during differentiation

2000). We, and others, have recently shown that some genes media base lacking glucose. Immediately following glucose escape translational repression during the early stages of glu- withdrawal, polysomes collapsed (Fig. 1A), as previously de- cose withdrawal (Arribere et al. 2011; Castelli et al. 2011), scribed (Ashe et al. 2000). Polysomes gradually recovered to a demonstrating that the translation activity of individual genes reduced level by 3 h, at which point cells had resumed growth can be differentially regulated in glucose-starved cells. It is at a reduced rate (Fig. 1A; data not shown). Ribo-seq and not known how long-term adaptation to the absence of glu- RNA-seq were highly reproducible (R2 = 0.94–0.98) between cose affects translation globally or for individual genes. replicates (Fig. 1B) and revealed profound changes in the Here we have investigated translational regulation in transcriptome and translatome of starved cells. mRNA abun- starved differentiating yeast using ribosome footprint profil- dance was significantly affected for 32% of genes, and ribo- ing to determine gene-specific translation activity genome- some footprints were significantly affected for 45% of genes wide (Ingolia et al. 2009). We show that cellular adaptation (P < 0.05, based on analysis of replicates using counting sta- to the absence of glucose is accompanied by widespread tistics) (Fig. 1C). Gene expression changes were correlated and functionally coherent translational changes, which could with changes previously observed 2 h after glucose withdraw- not be detected by bulk polysome profiling approaches (Arri- al using microarrays to total mRNA and polysomal mRNA bere et al. 2011; Castelli et al. 2011). Specifically, transcripts (R2 = 0.64 and 0.65 for RNA-seq and Ribo-seq, respectively), from growth-promoting genes, including ribosomal proteins although Ribo-seq revealed larger fold changes than those (RPs) and ribosome biogenesis (RiBi) factors, are very effi- previously detected by bulk polysome profiling (Arribere ciently translated in rapidly dividing cells and preferentially et al. 2011). translationally repressed under differentiation-promoting Combining Ribo-seq with measurement of total tran- starvation conditions. In contrast, transcripts encoding mito- script abundance by RNA-seq can reveal changes in the chondrial proteins (mito mRNAs) are specifically translationally activated in starved cells. As a class, genes required A B RNA-seq Ribo-seq 40S 60S 80S Polysomes for invasive growth are significantly trans- 1.0 +glu 4 4 (RPKM) lationally up-regulated in response to 0.5 (RPKM) 10 glucose withdrawal, consistent with a pos- 2 10 2 0.0 R2 = 0.97 R2 = 0.98 itive role for translational regulation in -2 2 4 -2 2 4 1.0 cellular differentiation. Using a candidate -2 -2 +glu rep 2 log +glu rep 2 log -glu 10’ +glu rep1 log10 (RPKM) +glu rep 1 log10 (RPKM) gene approach, we identify the conserved A254 0.5 cAMP-dependent protein kinase (PKA) 0.0 4 4 (RPKM) (RPKM) as a major regulator of these gene-specific 2 2 10 1.0 10 glucose-responsive translation changes. -glu 180’ R2 = 0.95 R2 = 0.94 -2 2 4 Together, our results identify mecha- 0.5 -2 2 4

-2 -2 -glu rep2 log nisms for pervasive translational changes -glu rep 2 log 0.0 -glu rep 1 log10 (RPKM) -glu rep 1 log10 (RPKM) in response to glucose availability and re- Top Bottom veal a surprising degree of growth state- Ribo-seq RNA-seq C D 8 dependent specialization of the process 15 15 of translation. 4 10 10 TE RPKM RPKM 2 2 2 0

RESULTS 5 5 -glu log -4 -glu 3hr log -glu 3hr log

0 0 -8 Global translational reprogramming 0 5 10 15 0 5 10 15 -8 -4 0 4 8 in response to glucose starvation +glu log2 RPKM +glu log2 RPKM +glu log2 TE significantly increased in -glu (1701) significantly increased in -glu (1020) significantly increased in -glu (1391) To examine translational changes follow- significantly decreased in -glu (839) significantly decreased in -glu (829) significantly decreased in -glu (1533) ing glucose withdrawal, we performed FIGURE 1. Glucose withdrawal caused widespread translational changes. (A) Polysome profiles ribosome footprint profiling (Ribo- of cells grown in rich media with or without 2% glucose for the indicated times. (B) Ribo-seq seq), which consists of isolating and se- (right) and RNA-seq measurements (left) were reproducible. (C) Ribo-seq and RNA-seq libraries quencing ribosome-protected mRNA were prepared from cells starved for 3 h. Average ribosome densities (reads per kilobase per million reads) for 4576 genes that passed a minimum read cutoff were calculated from two biological fragments and gives a precise measure replicates. Genes that were significantly (P < 0.05) increased (red) or decreased (blue) were iden- of gene-specific translation activity ge- tified by analysis of replicates using counting statistics as described (Larsson et al. 2011). (D) nome-wide (Ingolia et al. 2009). Wild- Translational efficiencies (TEs) in glucose-fed cells vary widely from those in glucose-starved type Sigma 1278b cells were grown to cells. Average TEs for 5447 genes that passed a minimum read cutoff were calculated from two biological replicates. Genes that were significantly (FDR < 1%) increased (red) or decreased mid-log phase in rich media containing (blue) were identified by analysis of propagated RNA-seq and Ribo-seq variances (Materials 2% glucose and then transferred to rich and Methods).

www.rnajournal.org 913 Downloaded from rnajournal.cshlp.org on October 3, 2021 - Published by Cold Spring Harbor Laboratory Press

Vaidyanathan et al. translational efficiency (TE) for individual genes. Adaptation A Ribo-seq RNA-seq to glucose withdrawal was accompanied by pervasive transla- 100 100 Gini Coefficient: 80 Gini Coefficient: 80 tional regulation: 54% of all genes showed significant differ- +glu: 0.85 +glu: 0.79 -glu: ences in TE between glucose-grown and glucose-starved cells 60 -glu: 0.79 60 0.63 (FDR < 1%, based on propagation of RNA-seq and Ribo-seq 40 40 % of RPKMs % of variances) (Fig. 1D). Of particular note were 1391 genes with 20 20 significantly increased translation activity in starved cells, 0 0 0204060801000 20 40 60 80 100 which were not identified by bulk polysome profiling (Arri- % of genes % of genes bere et al. 2011). This difference could be explained by the B C fact that Ribo-seq can distinguish between mRNAs with dif- +glu -glu +glu 1 +glu 2 -glu 1 -glu 2 ferent numbers of ribosomes per mRNA instead of counting Top 50% Top 50% Ribo-seq 99 12 172 Ribo-seq 73104 38146 16 all polysomal mRNAs equally. RPKMs RPKMs Both microarray and sequencing-based methods can measure fold changes for individual genes between conditions, Top 50% Top 50% but only RNA-seq and Ribo-seq allow accurate quantita- RNA-seq 69 74 422 RNA-seq 5 138 22 43 453 70 RPKMs RPKMs tion of the contributions of individual genes to the total D E and translating mRNA pools. We exploited this characteristic 1.0 1.0 All genes (4350) All genes (4350) of sequencing-based gene expression data to determine how +glu Top 50% +glu Top 50% 0.8 0.8 different are the translatomes of two cellular growth states— Ribo-seq (111) Ribo-seq (111) rapidly dividing and starved differentiating. The translated 0.6 0.6 mRNA pools were strikingly specialized in both growth con- 0.4 0.4 Fraction of genes Fraction of genes ditions, as most ribosome-protected fragments came from 0.2 0.2 -28 -17 a small subset of genes (Gini coefficients of 0.85 and 0.79) p-value: 3.3e p-value: 2.1e 0.0 0.0 −4 −2 0 2 4 −4 −2 0 2 4 (Fig. 2A). After correction for differences in ORF length, a log2 TE (+glu) log2 ∆TE (-glu/+glu) mere 111 genes were sufficient to account for 50% of trans- F G 1.0 1.0 lated mRNA in rapidly dividing cells, and 184 genes com- All genes (4350) All genes (4350) -glu Top 50% -glu Top 50% 0.8 0.8 prised the top 50% in glucose-starved cells. Remarkably, Ribo-seq (152) Ribo-seq (152) there were only 12 genes in common between the two condi- 0.6 0.6 tions (Fig. 2B; Supplemental Table 1). The identities of these 0.4 0.4 Fraction of genes genes were reproducible between replicates (Fig. 2C). Based Fraction of genes 0.2 0.2 -72 -10 on these results, we infer that the majority of translation ini- p-value: 4.0e p-value: 2.9e 0.0 0.0 tiation events occurred on different mRNAs in glucose- −4 −2 0 2 4 −4 −2 0 2 4 starved compared to rapidly dividing cells. This conclusion log2 TE (-glu) log2 ∆TE (-glu/+glu) has profound implications for the roles of general translation FIGURE 2. Gene-specific translation efficiencies are adapted to growth factors (Discussion). The short gene lists comprising the top conditions. (A) Gini coefficient plots of yeast genes versus Ribo-seq (left) 50% of translating mRNAs were dominated by two function- and RNA-seq (right) reads per kilobase per million mapped reads al categories, one per condition. In glucose, cytoplasmic ribo- (RPKMs) represent the heavily skewed gene expression profiles in different growth conditions. The dotted horizontal lines represent the top somal proteins accounted for 80/111 genes (72%, P = 2.7 × 50% of RPKMs from each library. (B) The translated mRNA pool dif- −112 10 ). In cells adapting to glucose withdrawal, 119/184 of fered substantially depending on growth conditions (±glucose). Venn the most highly translated genes encode proteins that localize diagrams represent the genes that make up the top 50% of RPKMs to mitochondria (65% by GO component analysis, P = 2.7 × from each library. (C) The identities of the most highly expressed and −49 translated mRNAs were reproducible between biological replicates. 10 ). Thus, the majority of the translational machinery was Venn diagrams represent the genes that make up the top 50% of dedicated to synthesizing a different small set of proteins, RPKMs from each library. (D) Cumulative distribution (CDF) plot which depended on the growth state of the cells. for translational efficiency (Ribo-seq RPKM divided by RNA-seq Next we asked whether condition-dependent differences RPKM) in +glu of the top 50% of genes by Ribo-seq RPKMs in +glu. (E) CDF plot for changes in TE upon glucose withdrawal of the top in translational efficiency (TE) were consistent with growth 50% of genes by Ribo-seq RPKMs in +glu. (F) CDF plot for TE in state optimization of cellular translation activity. Elongation −glu of the top 50% of genes by Ribo-seq RPKMs in −glu. (G) CDF rates were assumed to be constant between genes, as suggest- plot for changes in TE upon glucose withdrawal of the top 50% of genes by Ribo-seq RPKMs in −glu. Individual replicates were used to identify ed by recent analysis (Ingolia et al. 2011). Gene-specific TEs the top 50% genes. TEs are the average of two biological replicates. were computed by normalizing ribosome footprint density to Significance was determined by the Kolmogorov-Smirnov test. total mRNA abundance. The most highly translated mRNAs in glucose-grown cells showed elevated TE compared to all − genes in the presence of glucose (P = 3.3 × 10 28) (Fig. 2D), duction in the average number of ribosomes per transcript − and this group was significantly more translationally re- (P = 2.1 × 10 17) (Fig. 2E). On the other hand, the most pressed following glucose withdrawal, i.e., had a greater re- highly translated mRNAs from glucose-starved cells had

914 RNA, Vol. 20, No. 6 Downloaded from rnajournal.cshlp.org on October 3, 2021 - Published by Cold Spring Harbor Laboratory Press

Translatome specialization during differentiation moderate TEs in the presence of glucose (data not shown) lowing glucose withdrawal. Thus, widespread translational and very high TEs following glucose withdrawal (P = 4.0 × regulation leads to enhanced expression of specific genes un- − 10 72) (Fig. 2F) due to a significant increase in TE (P = der the cellular conditions where the gene products are re- − 2.9 × 10 10) (Fig. 2G). Fewer genes were required to account quired for optimal growth. for the top 50% of Ribo-seq reads compared to RNA-seq The translational and transcriptional changes caused by reads (Fig. 2B), which may be partly explained by translation- glucose starvation were largely coherent—ribosomal protein al privileging of highly expressed genes. Together, these re- genes were expressed at lower levels and also translated less sults show that glucose starvation induced pronounced efficiently, while mitochondrial protein genes were both gene-specific translational changes. These TE changes are transcriptionally up-regulated and translationally privileged consistent with a retooling of the translation machinery to in starved cells. It was recently argued, based on mathemati- better translate a dramatically altered mRNA substrate pool. cal principles, that using RNA-seq values to normalize Ribo- seq data may not be sufficient to eliminate the contribution of transcriptional changes to TE changes quantified by taking Translational control of starvation adaptation genes log2 ratios of Ribo-seq RPKMs (reads per kilobase per million Next we examined whether translational regulation dispro- mapped reads) to RNA-seq RPKMs. An alternative approach portionately impacted genes with important functions in cel- was proposed, which analyzes partial variance and uses a var- lular adaptation to starvation. Global gene deletion analysis iance shrinkage method for improving statistical inference, in has identified the genes required for starvation-induced in- a minimum of three biological replicates to assess changes in vasive growth (430) and adhesion to agar (329) (Chin et al. gene-specific translational efficiency (Larsson et al. 2011). 2012; Ryan et al. 2012). These sets of genes were preferentially We therefore performed a third independent experiment translationally up-regulated in glucose-starved cells compared and analyzed gene-specific translation changes caused by glu- − − to all genes (P = 2.1 × 10 4, 1.9 × 10 8). Globally, glucose cose withdrawal using the ANOTA R package. This alterna- withdrawal resulted in significant (FDR < 1%) changes in tive analysis revealed similarly significant translational TE for 54% of yeast genes and ≥2-fold changes in TE for regulation of RP and mito mRNAs in glucose-starved cells − − 29% of yeast genes (Figs. 1D, 3A). Notably, of the 870 genes (P = 5.3 × 10 20 and 1.9 × 10 12, respectively), thus ruling that were substantially translationally activated (▵TE ≥ out the trivial explanation of inadequate normalization of ri- +2-fold) in glucose-starved cells, mRNAs encoding mito- bosome footprints to total RNA levels as an explanation for chondrial proteins (mito mRNAs) formed the dominant cat- these TE changes. egory (Fig. 3B). Moreover, as a group, nuclear-encoded Next, we considered the possibility that increased and de- mito mRNAs were significantly better translated compared creased competition by mRNAs for limiting translation initi- − to all genes in glucose-starved cells (P =3×10 33). Mitochon- ation factors could be responsible for corresponding changes drial respiratory function is critical for yeast growth and filamentation in the A B Increased translational efficiency absence of glucose (Broach 2012), and 350 Replicate error 300 σ = 1.39 fold nuclear-encoded mitochondrial protein mitochondrion translation 61 250 genes are substantially overrepresented 200 mitochondrial organization 115 among the genes reported to be required 150 biological process 229 for invasive growth (159/430, P = 2.6 × 100 −29 Number of genes respiratory chain complex assembly 18 10 ) and adhesion (152/329, P = 50 −38 6.1 × 10 ). Thus, translational regula- 0 sexual sporulation 34 −4 −3 −2 −1 0 1 2 3 4 tion enhances synthesis of proteins that log2 ∆TE (-glu/+glu) 0 5 10 15 20 25 -log10 (p-value) are required for the adaptive cellular re- C Decreased translational efficiency sponse to glucose starvation. ribosome biogenesis 116 On the other hand, genes that are im- ribonucleoprotein complex biogenesis 121 ncRNA metabolic process 110 portant for rapid growth in the presence cellular component biogenesis 131 of glucose were selectively translationally rRNA processing 79 repressed during glucose starvation. carboxylic acid metabolic process 104 ≥ small molecule biosynthetic process 100 Genes with TE reduced 2-fold fell into translation 93 a few main GO categories: ribosome bio- 0 5 10 15 20 25 -log10 (p-value) genesis (16% of translationally repressed genes), cytoplasmic translation (13%), FIGURE 3. RP and mito mRNAs differentially translated upon glucose withdrawal. (A) metabolic processes (69%), and glucose Distribution of changes in TE in glucose-starved cells. (B) GO category enrichment of genes ≥ catabolism (3%). Furthermore, most whose TE increases 2-fold (FDR < 1%) in glucose starvation. The number of genes in each category is indicated to the right of the graph. (C) GO category enrichment of genes whose TE de- genes within these functional categories creases ≥2-fold (FDR < 1%) in glucose starvation. The number of genes in each category is showed significant reductions in TE fol- indicated to the right of the graph.

www.rnajournal.org 915 Downloaded from rnajournal.cshlp.org on October 3, 2021 - Published by Cold Spring Harbor Laboratory Press

Vaidyanathan et al. in mRNA abundance and translational efficiency for RP and A 1.0 B 1.0 All All mito mRNAs following glucose starvation. According to this RP RP model, the translational efficiency of RP mRNAs decreases 0.8 0.8 as their relative abundance decreases, while transcriptionally 0.6 0.6 up-regulated mRNAs such as the mito mRNAs become more 0.4 0.4 competitive under conditions where they comprise a much Fraction of genes larger portion of the total mRNA pool. To test this competi- Fraction of genes 0.2 p-value: 0.2 p-value: 2.4e-95 5.7e-16 tion hypothesis, we compared changes in total mRNA abun- 0.0 0.0 −6 −4 −2 0 2 4 6 −3 −2 −1 0 1 2 3 dance and TE for RP mRNAs in other conditions where the log2 ∆RNA-seq (-glu/+glu) log2 ∆TE (-glu/+glu) RP genes are transcriptionally down-regulated. Yeast Ribo- 1.0 1.0 All All seq and RNA-seq data were available from four relevant stress RP RP 0.8 0.8 conditions: glucose starvation (this study), amino acid starvation (Ingolia et al. 2009), nutrient deprivation during meiosis 0.6 0.6 (Brar et al. 2012), and oxidative stress induced by hydrogen 0.4 0.4 peroxide (H2O2) (Gerashchenko et al. 2012). Treatment Fraction of genes Fraction of genes 0.2 p-value: 0.2 p-value: -04 with H2O2 and induction of meiosis caused reductions in 1.5e-86 8.4e 0.0 0.0 RP mRNA levels that were well correlated with and compara- −6 −4 −2 0 2 4 6 −3 −2 −1 0 1 2 3 log ∆RNA-seq (H2O2/YPD) log ∆TE (H2O2/YPD) ble in magnitude to glucose starvation, yet H2O2 treatment 2 2 1.0 1.0 did not lead to translational repression of RP mRNAs (Fig. All All RP RP 4A–C; data not shown). Amino acid starvation caused 0.8 0.8 more moderate reductions in RP mRNA levels than either 0.6 0.6 glucose starvation or H2O2 treatment and did not lead to consistent reductions in TE (Fig. 4A,C). Thus, decreased 0.4 0.4 Fraction of genes mRNA abundance is not sufficient to cause decreased trans- Fraction of genes 0.2 p-value: 0.2 p-value: -08 lational efficiency of RP mRNAs. 1.3e-61 6.1e 0.0 0.0 Strong transcriptional up-regulation of mito mRNAs was −6 −4 −2 0 2 4 6 −3 −2 −1 0 1 2 3 specific to glucose starvation. Thus, to determine whether in- log2 ∆RNA-seq (-AA/YPD) log2 ∆TE (-AA/YPD) C creased mRNA abundance is generally associated with in- 8 6 creased TE, we examined a group of genes that are induced 4 in response to a variety of stresses as part of the environmen- 4 2 tal stress response (ESR) (Gasch et al. 2000). The ESR consists of regulated changes in ∼900 genes—down-regulation of the 0 0 −2 transcription of growth promoting genes and induction of −4 ∆RNA-seq (H2O2/YPD) ∆RNA-seq log2 ∆TE (H2O2/YPD) log2 stress-responsive genes—which are believed to allow cells 2 −4

log r = 0.72 r = 0.55 to adapt to suboptimal environments. Stress response genes −8 −6 −8 −4 0 4 8 −6 −4 −2 0 2 4 6 were induced in all four stresses (Fig. 5A). However, the log2 ∆RNA-seq (-glu/+glu) log2 ∆TE (-glu/+glu) presence and magnitudes of TE changes varied significantly between individual stresses and did not simply parallel FIGURE 4. Translational regulation of RP genes is specific to individual stresses. (A,B) Translational and transcriptional changes of RP genes can changes in transcriptional activity (Fig. 5B). Taken together, be unlinked. CDF plots for changes in RNA-seq (A) and TE (B) of ribo- these results show that changing the relative abundance of an somal protein genes (RP genes) upon glucose withdrawal, oxidative mRNA does not lead to changes in its TE. Thus, translational stress (H2O2) (Gerashchenko et al. 2012), and amino acid starvation − regulatory mechanisms beyond simple competition for lim- ( AA) (Ingolia et al. 2009). (C) Comparison of changes in RNA-seq (left) and TE (right) between oxidative stress and glucose starvation. iting factors are at work in glucose-starved cells. TEs were computed only for genes with sufficient reads in both RNA- seq and Ribo-seq libraries from +glu and −glu samples. RNA-seq and TE changes are the averages of two biological replicates. Significance Translation changes are due to altered efficiency was determined by the Kolmogorov-Smirnov test. of initiation and mRNA sequestration Reduced TE can arise by two distinct mechanisms that affect ribosome recruitment: an increase in the size of the nontrans- mechanism underlying reduced TE of RP mRNAs, we per- lating pool of mRNA (e.g., by mRNA sequestration), or a formed polysome profiling followed by qRT-PCR for repre- decrease in the average number of ribosomes associated sentative RP genes. Relative mRNA abundance per fraction with each transcript due to reduced efficiency of initiation. from polysome profiles before or after glucose withdrawal These mechanisms are indistinguishable by Ribo-seq, as ei- was quantified by normalization to a dope-in control firefly ther leads to a decrease in the number of ribosome footprints luciferase RNA (FLUC). The distribution of 18S rRNA after recovered per mRNA molecule (Fig. 6A). To determine the normalization was determined for reference (Fig. 6B). In

916 RNA, Vol. 20, No. 6 Downloaded from rnajournal.cshlp.org on October 3, 2021 - Published by Cold Spring Harbor Laboratory Press

Translatome specialization during differentiation

A B ribosomes. Although still ribosome-associated, these RP 1.0 1.0 All All mRNAs shifted toward smaller polysomes in glucose-starved ESR up ESR up 0.8 ESR down 0.8 ESR down cells (Fig. 6C,D). On the other hand, RPL5, RPL6B, and RPS2

0.6 0.6 showed a pronounced accumulation of mRNA in the nontranslating fractions relative to the polysome fractions (Fig. 0.4 0.4 6E,F). Thus the translational repression of RP mRNAs in cells Fraction of genes Fraction of genes p-value: p-value: 0.2 0.2 6.4e-46 1.1e-07 adapting to prolonged glucose starvation proceeds by two -156 2.9e 3.8e-35 — 0.0 0.0 mechanisms a decrease in the frequency of initiation and −8 −4 0 4 8 −4 −2 0 2 4 log2 ∆RNA-seq (-glu/+glu) log2 ∆TE (-glu/+glu) an increase in the size of the nontranslating pool. 1.0 1.0 In contrast, the increase in TE of mito mRNAs in glucose- All All ESR up ESR up depleted cells was almost entirely due to increased polysome 0.8 0.8 ESR down ESR down size, rather than an increased fraction of actively translating 0.6 0.6 mRNA. Polysome profiling by qPCR revealed that represen-

0.4 0.4 tative mito mRNAs had relatively large amounts of free

Fraction of genes p-value: mRNA (13%–26% of total mRNA) in glucose-grown cells, p-value: Fraction of genes 0.2 0.2 4.1e-116 2.4e-01 consistent with previous genome-wide assessments, which 1.8e-198 8.2e-03 0.0 0.0 observed low ribosome occupancy for mito mRNAs under −8 −4 0 4 8 −4 −2 0 2 4 log2 ∆RNA-seq (H2O2/YPD) log2 ∆TE (H2O2/YPD) these conditions (Arava et al. 2003; Arribere et al. 2011). However, in glucose-starved cells, this ribosome-free pool 1.0 1.0 All All changed very little, whereas the ribosome-associated mito ESR up ESR up 0.8 ESR down 0.8 ESR down mRNAs shifted from light to heavier polysomes (Fig. 7A–

0.6 0.6 D). This constancy of the ribosome-free pool explains why translational up-regulation of mito mRNAs following glu- 0.4 0.4 cose withdrawal was not detected previously (Arribere et al. Fraction of genes Fraction of genes p-value: p-value: 0.2 0.2 6.1e-66 8e-07 2011). Therefore, mito mRNAs changed TE primarily due 1.8e-150 3.2e-31 0.0 0.0 to altered translation initiation rather than altered partition- −8 −4 0 4 8 −4 −2 0 2 4 log2 ∆RNA-seq (-AA/YPD) log2 ∆TE (-AA/YPD) ing between translating and nontranslating pools.

1.0 1.0 All All ESR up ESR up Inhibition of PKA signaling partially recapitulated 0.8 ESR down 0.8 ESR down translational effects of glucose withdrawal 0.6 0.6 What could be responsible for widespread changes in gene- 0.4 0.4 specific translational efficiencies specifically in response to

Fraction of genes p-value: p-value: Fraction of genes 0.2 0.2 1.3e-77 4.4e-06 glucose availability? Signaling through the Ras/PKA pathway 4e-188 5.9e-130 is required for the majority of glucose-responsive tran- 0.0 0.0 −8 −4 0 4 8 −4 −2 0 2 4 scriptional changes, and Ras/PKA activation is sufficient to log2 ∆RNA-seq (Meiosis/Vegetative) log2 ∆TE (Meiosis/Vegetative) recapitulate ∼90% of glucose-induced changes in mRNA FIGURE 5. Translational control mechanisms are stress-specific. (A,B) abundance (Zaman et al. 2009). Additionally, glucose star- Translational and transcriptional changes of environmental stress re- vation has been shown to inhibit PKA activity (Broach sponse (ESR) genes can be unlinked. CDF plots for changes in RNA- 2012). We therefore investigated whether the Ras/PKA path- seq (A) and TE (B) of ESR genes transcriptionally induced (ESR up) or repressed (ESR down) in a variety of stresses including glucose star- way also regulates translation of glucose-responsive mRNAs. vation, amino acid starvation (−AA) (Ingolia et al. 2009), oxidative PKA kinase activity was monitored by following phosphory- stress (H2O2) (Gerashchenko et al. 2012), and nutrient stress during lation of Pat1, which is phosphorylated by PKA on Ser-456 the meiotic program (Meiosis) (Brar et al. 2012). TEs were computed and Ser-457 (Ramachandran et al. 2011). To separate the ef- only for genes with sufficient reads in both RNA-seq and Ribo-seq libraries from +glu and −glu samples. RNA-seq and TE changes are the fects of inhibiting PKA from the glucose starvation response, averages of two biological replicates. Significance was determined by we exploited an analog-sensitive strain (pka-as) in which all the Kolmogorov-Smirnov test. three genes encoding the catalytic subunit of PKA (TPK1, TPK2, and TPK3) were mutated to render them sensitive to the ATP-analog 1NM-PP1. This inhibitor inactivates kinases the presence of glucose, almost all RP mRNA was associated whose active sites have been altered to accommodate binding with ribosome-containing fractions (5–12). Two modes of of a bulky side group and has no effect on the growth of cells translational repression were observed 3 h after glucose with- lacking the sensitizing mutation(s) (Bishop et al. 2001). drawal. For RPL9A, RPL2B, and RPP0, there was a modest in- Treatment of pka-as cells with 1NM-PP1 inhibited PKA ac- crease in ribosome-free mRNA (fractions 1–4), but most tivity (Fig. 8A) and caused a small reduction in bulk transla- mRNA (74%–93%) remained associated with one or more tion activity within 15 min and a larger reduction after 3 h

www.rnajournal.org 917 Downloaded from rnajournal.cshlp.org on October 3, 2021 - Published by Cold Spring Harbor Laboratory Press

Vaidyanathan et al.

A Next we examined gene-specific trans- mRNA reduced sequestration lational responses to PKA inhibition initiation to determine whether inactivating PKA

2-fold decreased TE was sufficient to recapitulate the selective B translational effects of glucose withdraw-

40S 60S 80S Polysomes al. Treatment of rapidly dividing pka-as 0.4 80S Polysomes + glu + glu cells with 1NM-PP1 for 15 min caused 0.4 - glu 0.3 - glu a shift of RP mRNAs (RPL2B, RPS2) to- 0.2 A254 0.2 ward smaller polysomes, indicating re- 0.1 duced translation initiation despite the

0.0 abundance 18S rRNA 0.0 presence of glucose. RP mRNA transla- 1 0 1 2 1 2 3 4 5 6 7 8 9 10 1 12 1 2 3 4 5 6 7 8 9 1 1 1 Fraction Fraction tion was further inhibited after 3 h of C 0.3 0.3 0.3 + glu RPL9A RPL2B RPP0 - glu incubation with the PKA inhibitor (Fig. 80S Polysomes 80S Polysomes 80S Polysomes 8C,D). In contrast, inhibiting PKA for 0.2 0.2 0.2 15 min increased translation activity of

0.1 0.1 0.1 mito mRNAs (COX11, MRPL24)inthe

mRNA abundance mRNA presence of glucose (left panels in Fig. 0.0 0.0 0.0 0 1 2 1 1 2 3 4 5 6 7 8 9 1 1 1 1 2 3 4 5 6 7 8 9 10 1 12 1 2 3 4 5 6 7 8 9 10 11 12 8E,F), but unlike RP mRNAs, translation D changes of mito mRNAs were attenuated 1.0 1.0 RPP0 + glu 1.0 RPL9A RPL2B - glu after 3 h of PKA inhibition in the pres- 0.8 0.8 0.8 ence of glucose (Fig. 8E,F). Taken togeth- 0.6 0.6 0.6 less more less more less more polysomal polysomal 0.4 polysomal polysomal 0.4 polysomal polysomal 0.4 er, these results show that PKA activity 0.2 0.2 0.2 is necessary for highly efficient transla- 0.0 tion of RP mRNAs even when glucose

Cumulative RNA fraction Cumulative RNA 0.0 0.0 2 1 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 1 1 2 3 4 5 6 7 8 9 10 1 12 E is available. In contrast, glucose avail- 0.3 RPL5 0.3 0.3 + glu ability eventually trumps PKA inactivity RPL6B RPS2 - glu 80S Polysomes 80S Polysomes 80S Polysomes for translational repression of the mito 0.2 0.2 0.2 mRNAs, but it takes time for cells to

0.1 0.1 0.1 adapt to the atypical situation of having

mRNA abundance mRNA high glucose and low PKA activity. 0.0 0.0 0.0 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 Thus, the mechanisms for PKA-mediat- F + glu ed RP mRNA translational activation 1.0 1.0 1.0 RPL5 RPL6B RPS2 - glu 0.8 0.8 0.8 and mito mRNA translational repression 0.6 0.6 0.6 are distinct and not merely two sides of less more less more less more 0.4 polysomal polysomal 0.4 polysomal polysomal 0.4 polysomal polysomal the same coin (see Discussion). 0.2 0.2 0.2 0.0 0.0

Cumulative RNA fraction Cumulative RNA 0.0 1 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 1 12 1 2 3 4 5 6 7 8 9 10 11 12 DISCUSSION Fraction Fraction Fraction Glucose starvation induces filamentous FIGURE 6. Two modes of translational regulation of RP mRNAs in glucose-starved cells. (A) differentiation accompanied by perva- Model for reduced TE by mRNA sequestration (right) or reduced frequency of translation initiation (left). (B) Polysome profiles of cells grown in rich media with or without 2% glucose for 3 h sive transcriptional changes that affect (left panel). qPCR analysis of 18S rRNA in polysome gradient fractions (right panel). (C) >40% of the genome. Here we showed Translating RP mRNAs (RPL9A, RPL2B, RPP0) associated with fewer ribosomes after 3 h of glu- that glucose starvation also causes wide- cose starvation. Relative mRNA abundance per fraction from polysome profiles before or after spread translational reprogramming of glucose withdrawal was quantified by qPCR normalized to dope-in control (FLUC) RNA. Error bars indicate standard deviation for technical replicates (n = 3). This is a representative the transcriptome. Genes encoding mi- of two replicate experiments. (D) CDF plots of mRNA abundance from fractions in C.(E)RP tochondrial proteins, which were tran- mRNAs (RPL5, RPL6B, RPS2) accumulated in ribosome-free fractions after 3 h of glucose star- scriptionally up-regulated in response to vation. Relative mRNA abundance per fraction from polysome profiles before or after glucose glucose withdrawal, were also transla- withdrawal was quantified by qPCR normalized to dope-in control (FLUC) RNA. Error bars indicate standard deviation for technical replicates (n = 3). This is a representative of two replicate tionally activated in glucose-starved cells. experiments. (F) CDF plots of mRNA abundance from fractions in E. Conversely, genes encoding cytoplasmic ribosomal proteins were translated less efficiently in addition to being based on a reduction in polysomes (Fig. 8B). Thus, inhibition transcribed at lower levels in the absence of glucose. These of PKA was sufficient to decrease translation even in the pres- concerted changes in gene-specific TEs contributed to a pro- ence of glucose. found growth state-dependent specialization of translation.

918 RNA, Vol. 20, No. 6 Downloaded from rnajournal.cshlp.org on October 3, 2021 - Published by Cold Spring Harbor Laboratory Press

Translatome specialization during differentiation

A mRNAs. Because the identities of the 0.3 0.3 0.3 + glu RSM10 MRPL24 PET117 - glu ∼5% of genes that comprise the major- 80S Polysomes 80S Polysomes 80S Polysomes 0.2 0.2 0.2 ity of translation activity differ between cellular growth conditions, the factor re- 0.1 0.1 0.1 quirements for bulk translation could mRNA abundance mRNA 0.0 0.0 0.0 be strongly condition-dependent. It will 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 B be important for future work to assess 1.0 1.0 1.0 + glu RSM10 MRPL24 PET117 - glu the gene-specific effects of translation 0.8 0.8 0.8 factors and to examine their roles under 0.6 less more 0.6 less more 0.6 less more different growth conditions. polysomal polysomal polysomal polysomal polysomal polysomal 0.4 0.4 0.4 We have shown that the mecha- 0.2 0.2 0.2 nisms of translational repression of RP

Cumulative RNA fraction Cumulative RNA 0.0 0.0 0.0 0 1 2 0 1 2 1 1 2 3 4 5 6 7 8 9 1 1 1 1 2 3 4 5 6 7 8 9 1 1 1 1 2 3 4 5 6 7 8 9 10 1 12 mRNAs and translational activation of C + glu mito mRNAs are downstream of PKA 0.3 COX11 0.3 RSM19 0.3 RSM27 - glu 80S Polysomes 80S Polysomes 80S Polysomes signaling. Increased translation of mito 0.2 0.2 0.2 mRNAs in glucose starvation or following inhibition of PKA is not simply 0.1 0.1 0.1 an indirect consequence of reduced RP mRNA abundance mRNA 0.0 0.0 0.0 1 1 2 3 4 5 6 7 8 9 1 mRNA translation (e.g., by reduced com- 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 1 12 10 1 12 petition for limiting initiation factors), D + glu 1.0 COX11 1.0 RSM19 1.0 RSM27 - glu because the two translational responses 0.8 0.8 0.8 could be uncoupled. After 3 h of PKA in- 0.6 less more 0.6 less more 0.6 less more polysomal polysomal hibition with 1NM-PP1 in the presence 0.4 polysomal polysomal 0.4 polysomal polysomal 0.4 0.2 0.2 0.2 of glucose, RP mRNA translation was

Cumulative RNA fraction Cumulative RNA 0.0 0.0 0.0 still repressed, while mito mRNA transla- 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 Fraction Fraction Fraction tion had returned to the lower level typi- cal of these transcripts in glucose-grown FIGURE 7. Increased polysome size of mito mRNAs in glucose-starved cells. (A) Mitochondrial cells. Thus, separate PKA-dependent protein mRNAs (RSM10, MRPL24, PET117) associated with heavier polysomes after 3 h of mechanisms regulate the efficiency of glucose starvation. mRNA abundance per fraction was determined as in Figure 5. This is a representative of two replicate experiments. (B) CDF plots of mRNA abundance from fractions in translation initiation on RP and mito A.(C) Mitochondrial protein mRNAs (COX11, RSM19, RSM27) associated with heavier poly- mRNAs. Differential activity of PKA iso- somes after 3 h of glucose starvation. mRNA abundance per fraction was determined as in forms is a potential source of stress spe- Figure 5. This is a representative of two replicate experiments. (D) CDF plots of mRNA abun- cificity in the gene-specific effects of dance from fractions in C. PKA on translation. The catalytic subunit of PKA is encoded by three separate In fact, the majority of translating ribosomes were engaged genes in S. cerevisiae (TPK1, TPK2, and TPK3). In our exper- with mRNAs from <5% of genes in each condition, and iments, all three PKA catalytic subunits were inhibited with translatomes were substantially less diverse than transcrip- 1NM-PP1. Given that the Tpk isoforms have distinct sub- tomes, particularly in glucose-starved cells. strate preferences (Ptacek et al. 2005) and subcellular locali- Not only were translatomes highly specialized in both zations and show isoform-specific changes in localization growth conditions, the identities of the most highly translated upon glucose withdrawal (Tudisca et al. 2010), it will be im- messages were almost nonoverlapping between conditions. portant to determine which isoform(s) mediate translation- This observation has profound implications for studying al activation of RP mRNAs and translational repression of translation and for understanding the impact of transla- mito mRNAs. Additionally, other glucose-responsive signal- tion factors and RNA-binding proteins on gene expression. ing pathways may also contribute to gene-specific transla- According to our data, a hypothetical factor that was abso- tional control. Indeed, the minor differences in polysomal lutely required for translation of 95% of genes might cause distributions of RP mRNAs between glucose-starved and only a twofold change in bulk translation, depending on PKA-inhibited cells offer some support for this possibility. which 95% of genes it affected. Thus, the general importance It will be important for future studies to measure the relative of such a factor could easily be overlooked by relying on contributions of each of these signaling networks toward bulk assays. Conversely, a factor that causes bulk transla- starvation-induced translational repro-gramming. tion defects when mutated or depleted in a particular condi- Our comparative analysis of different stress conditions tion, and which has therefore been considered as a “general” suggests that only a small fraction of the genome is sub- translation factor, may actually play a specialized role in ject to similar translational regulation in all stresses, which promoting translation of relatively few highly translated contrasts with much greater similarity between stresses at

www.rnajournal.org 919 Downloaded from rnajournal.cshlp.org on October 3, 2021 - Published by Cold Spring Harbor Laboratory Press

Vaidyanathan et al.

the transcriptional level. Thus, control mechanisms that regulate gene translation in one condition may not necessarily A B 80S Polysomes be utilized in a different growth or stress condition. Because the functions of most Input IP: Pat1 0.4 -1NM-PP1 +1NM-PP1 15’ yeast translation factors and cytoplasmic α-PKA Pat1 254 +1NM-PP1 3hr (phospho) A Sub 0.2 RNA-binding proteins have been charac- α-myc Pat1 (total) terized primarily, if not exclusively, in Antibody - + - + - + 0.0 cells growing in glucose, the factors re- 1NM-PP1 -15’3hr - 15’ 3hr 1 2 3 4 5 6 7 8 9 10 11 12 Top Bottom sponsible for translational activation of C 0.3 RPL2B 0.3 RPL2B 1.0 RPL2B mito mRNAs and other starvation-in- 80S Polysomes 80S Polysomes 0.8 duced transcripts are likely to be relative- 0.2 0.2 0.6 -1NM-PP1 #1 -1NM-PP1 #2 ly uncharacterized. Intriguingly, more 0.4 +1NM-PP1 15’ 0.1 0.1 +1NM-PP1 3hr than 70 annotated yeast RNA binding 0.2 mRNA abundance mRNA mRNA abundance mRNA proteins changed expression ≥2-fold in

0.0 0.0 fraction Cumulative RNA 0.0 0 1 2 1 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 1 1 1 1 2 3 4 5 6 7 8 9 10 1 12 starved cells, suggesting that many as- D 0.3 RPS2 0.3 RPS2 1.0 RPS2 pects of post-transcriptional gene reg- 80S Polysomes 80S Polysomes 0.8 ulation may be altered to adapt to the 0.2 0.2 0.6 -1NM-PP1 #1 transformation of the mRNA pool by -1NM-PP1 #2 0.1 0.4 +1NM-PP1 15’ glucose starvation. It will be interesting 0.1 +1NM-PP1 3hr 0.2 mRNA abundance mRNA mRNA abundance mRNA to see whether cellular differentiation

0.0 0.0 fraction Cumulative RNA 0.0 1 2 3 4 5 6 7 8 9 1 1 10 11 12 1 2 3 4 5 6 7 8 9 10 1 12 1 2 3 4 5 6 7 8 9 10 1 12 in multicellular eukaryotes is associated E with a similarly high degree of transla- 0.3 COX11 0.3 COX11 1.0 COX11

80S Polysomes 80S Polysomes 0.8 tome specialization. If so, this could illu- 0.2 0.2 0.6 -1NM-PP1 #1 minate the basis for the surprisingly -1NM-PP1 #2 0.4 +1NM-PP1 15’ 0.1 0.1 tissue-specific effects of some translation +1NM-PP1 3hr 0.2

mRNA abundance mRNA factors and regulators. mRNA abundance mRNA

0.0 0.0 fraction Cumulative RNA 0.0 0 1 2 1 1 2 3 4 5 6 7 8 9 1 1 1 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 1 12 F 0.3 MRPL24 0.3 MRPL24 1.0 MRPL24 MATERIALS AND METHODS 80S Polysomes 80S Polysomes 0.8 0.2 0.2 0.6 -1NM-PP1 #1 -1NM-PP1 #2 Yeast strains and culture conditions 0.4 +1NM-PP1 15’ 0.1 0.1 +1NM-PP1 3hr 0.2 mRNA abundance mRNA

mRNA abundance mRNA Ribo-seq was performed on WT yeast strains

0.0 0.0 fraction Cumulative RNA 0.0 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 in the Sigma 1278b background (YWG025) Fraction Fraction Fraction (see Table 1). Yeast cultures were grown in YPAD (1% yeast extract, 2% peptone, 0.01% FIGURE 8. PKA activity promoted efficient translation of RP mRNAs and inhibited translation adenine hemisulfate, 2% dextrose) at 30°C of mito mRNAs. (A) PKA-dependent phosphorylation of Pat1 was lost in an analog-sensitive PKA in baffled flasks with vigorous shaking. Glu- strain (tpk1-as tpk2-as tpk3-as) following treatment with 1NM-PP1 for 15 min and 3 h. A myc- cose-starved cultures were prepared from tagged version of the PKA substrate Pat1 was immunoprecipitated, and the relative level of phos- – phorylation was determined by Western blotting with an anti-PKA substrate antibody (α-PKA YPAD cultures grown to OD600 = 1.0 1.1, Sub). Phosphorylated Pat1 in the IP lanes does not necessarily align with any of the bands in harvested by centrifugation for 5 min at the input lane because there are hundreds of PKA substrates in the cell (Ptacek et al. 2005) 12,000g, resuspended in prewarmed YPA me- that may be contributing to the signal in the input panel. 1NM-PP1 is stable in yeast over the dium lacking glucose and returned to shaking time scale of this experiment, thus the slight increase in reactivity to the anti-PKA sub antibody at 30°C for various times. The pka-as strain (a observed in the input lanes after 3 h +1NM-PP1 is unlikely to be due to reactivation of PKA. It gift from James Broach) was grown in syn- may be due to increased protein kinase C (PKC)-dependent phosphorylation when PKA is inhibited, because PKC recognizes a closely related sequence, (R/K) (R/K) X (S∗) (Hydrophobic) (R/ thetic SD medium with 2% dextrose to an K). (B) Inhibiting PKA reduced bulk translation activity in the presence of glucose. Polysome pro- OD600 = 0.4. PKA activity was inhibited files were analyzed after treatment with 1NM-PP1 for 15 min and 3 h. Representative traces from with 20 μM final concentration 1NM-PP1 three independent replicates are shown. (C,D) Inhibiting PKA for 15 min or for 3 h decreased (EMD Millipore) for various times. translation of RP mRNAs (RPL2B [C] and RPS2 [D]) in the presence of glucose. Polysome analysis of mRNAs (left and center) was performed as in Figure 5. Combined CDF plots of mRNA abundance from fractions including both time points of 1NM-PP1 inhibition are shown on the right.(E,F) Inhibiting PKA in the presence of glucose for 15 min increased translation of Ribo-seq and RNA-seq mito mRNAs (COX11 [left panel, E] and MRPL24 [left panel, F]). However, after 3 h of PKA in- Ribo-seq and RNA-seq were performed essen- hibition, translation of mito mRNAs was similar to uninhibited glucose-replete conditions (center panels, E and F). Relative mRNA abundance per fraction was determined as in Figure 5. tially as described (Ingolia et al. 2009; Brar Combined CDF plots of mRNA abundance from fractions including both time points of et al. 2012) with minor modifications. Yeast 1NM-PP1 inhibition are shown on the right. cultures were harvested by centrifugation.

920 RNA, Vol. 20, No. 6 Downloaded from rnajournal.cshlp.org on October 3, 2021 - Published by Cold Spring Harbor Laboratory Press

Translatome specialization during differentiation

RNA was purified from polysomal gradient samples by extraction TABLE 1. Yeast strains with 25:24:1 acidic phenol:chloroform: isoamyl alcohol and fol- Strain Genotype Source lowed by ethanol precipitation.

YWG025 MATa ura3 leu2 trp1 his3 Hiten Madhani YWG926 W303-1B tpk1M164G tpk2M147G James Broach Quantitative RNA analysis tpk3M165G YWG927 W303-1B tpk1M164G tpk2M147G This study mRNA abundance per polysome fraction was determined by qRT- tpk3M165G [2µ::HIS3 pCUP1- PCR. To control for differences in RNA recovery, RP and mito 6xmyc-Pat1] mRNA signals were normalized to synthetic poly(A)-tailed firefly YWG939 MATa ura3 leu2 trp1 his3 [2µ:: This study luciferase control mRNA or synthetic rluB mRNA from Escherichia HIS3 pCUP1-6xmyc-Pat1] coli. Reverse transcription was performed using SuperScript III (Invitrogen) or Avian Myeloblastosis Virus Reverse Trancriptase (AMV-RT; Promega) using a mix of oligo-dT and random hexamer Cells were lysed in PLB (20 mM HEPES-KOH, pH 7.4, 2 mM Mg primers. Quantitative PCR was performed with SYBR Fast reagents (Kapa Biosystems) according to the manufacturer’s instructions us- (OAc)2, 100 mM KOAC, 1% Triton-X 100, 0.1 mg/mL CHX, 3 mM DTT) by vortexing with glass beads. RNA-seq libraries were pre- ing a Roche Lightcycler 480. Gene-specific primer sequences are pared from total RNA purified from cell extract by the hot phenol available upon request. method. PKA activity assay Data analysis PKA activity was determined by monitoring the phosphorylation Sequencing reads were processed as described (Ingolia et al. 2009) state of PKA substrate Pat1 as described (Ramachandran et al. using Python scripts developed in-house. To determine differential 2011). Briefly, equal amounts of log-phase yeast cultures were har- ≥ expression of genes, raw sequencing read counts were analyzed using vested and resuspended in 5% TCA for 30 min to precipitate all the edgeR package (Robinson et al. 2010). Genes for which there cellular proteins. TCA-treated cells were washed first in 50 mM were at least 128 read counts across both replicate libraries in each Tris-Cl, pH 7.5, then in acetone; and dried in a speed vac. Cell ex- condition were included. The first eight codons from each gene tracts for immunoprecipitation were prepared in breakage buffer were excluded from the gene expression calculation in all libraries. (50 mM Tris-Cl, pH 7.5, 1 mM EDTA, 15 mM PNPP, 60 mM β The translation efficiency of a gene was computed as the ratio of -glycerophosphate, 5 mM DTT, 100 µg/mL aprotonin, 1 µg/mL the Ribo-seq read density on the gene to the RNA-seq read density. pepstatin, 10 µg/µL leupeptin, 1 mM PMSF, and one tablet of To be included in the TE calculation, we required that genes pass a Roche complete protease inhibitor cocktail [EDTA-free] per 50 minimum read cutoff of at least 128 counts over both replicates of mL buffer) by agitation with glass beads on a Fastprep-24 (MP Bio- the Ribo-seq and RNA-seq libraries. To determine significant TE medicals) at maximum speed for 3 × 45 sec or on a multivortexer changes, raw footprint and total counts for each gene (excluding for 15 min at 3000 rpm. Myc-tagged Pat1 was immunoprecipitated the first eight codons) were first processed by DESeq (Anders and from the clarified lysates with anti-myc monoclonal antibody Huber 2010). TEs were computed and variances determined by stan- (Sigma). Immunoprecipitates were collected on Protein A Mag dard error propagation (Lee and Forthofer 2006). The resulting Sepharose beads (GE Healthcare) and washed six times with wash- means and variances were used to compute an unpaired Z statistic, ing buffer (50 mM Tris-Cl, pH 7.5, 150 mM NaCl, 1% Igepal, 60 β which was used to compute a two-tailed P-value by integration of the mM -glycerophosphate, 0.1 mM sodium orthovanadate, 15 mM standard normal distribution. P-values were adjusted for multiple p-nitrophenylphosphate, 100 µg/mL aprotinin, 20 µg/mL leupeptin, testing by the method of Benjamini and Hochberg (Benjamini and 1 mM DTT, and one tablet of Roche complete protease inhibitor Hochberg 1995). An alternative metric for computing translation cocktail [EDTA-free] per 25 mL buffer). The level of PKA phos- changes using the ANOTA R package (Larsson et al. 2011) gave sim- phorylation was assessed by Western blotting with an anti-PKA substrate antibody (Cell Signaling) that recognizes phospho-Ser/Thr ilar results. Plots were made using GraphPad Prism version 5.0b for ∗ ∗ Mac OSX and the Matplotlib Python package (Hunter 2007). peptides of the sequence RRXS /T .

Extract preparation, polysome gradient fractionation, DATA DEPOSITION and RNA isolation Sequencing data were deposited in the NCBI Gene Expression Yeast polysomes were prepared as described (Arribere et al. 2011). In Omnibus (GEO) with accession number GSE51532. brief, cycloheximide was added to cell cultures to a final concentration of 0.1 mg/mL for 2 min before harvesting cells by centrifugation SUPPLEMENTAL MATERIAL (5 min at 4,000g). Cells were lysed by vortexing with glass beads in cold PLB. The crude extracts were clarified (20 min at 14,700g), Supplemental material is available for this article. and 20 OD260 units from the resulting supernatants were applied to linear 10%–50% sucrose gradients in PLB and spun for 2.75 h ACKNOWLEDGMENTS at 35,000 rpm in a Beckman SW41 rotor. Fractions were collected on a BioComp gradient master, and 20 ng of control firefly luciferase We thank H. Madhani, G. Fink, and J. Broach for yeast strains; RNA was added to each fraction before further processing. P. Herman for plasmids; J. Weissman for advice on ribosome

www.rnajournal.org 921 Downloaded from rnajournal.cshlp.org on October 3, 2021 - Published by Cold Spring Harbor Laboratory Press

Vaidyanathan et al.

profiling; MIT BioMicro Center for assistance with sequencing; and Chin BL, Ryan O, Lewitter F, Boone C, Fink GR. 2012. Genetic variation D. Bartel, C. Burge, Q. Justman, and members of the Gilbert labo- in Saccharomyces cerevisiae: circuit diversification in a signal trans- – ratory for helpful discussions. This work was supported by the NIH duction network. Genetics 192: 1523 1532. Cullen PJ, Sprague GF. 2012. The regulation of filamentous growth in R01 (GM094303). yeast. Genetics 190: 23–49. Gasch AP, Spellman PT, Kao CM, Carmel-Harel O, Eisen MB, Storz G, Received January 25, 2014; accepted March 14, 2014. Botstein D, Brown PO. 2000. Genomic expression programs in the response of yeast cells to environmental changes. Mol Biol Cell 11: 4241–4257. REFERENCES Gerashchenko MV, Lobanov AV, Gladyshev VN. 2012. Genome-wide ribosome profiling reveals complex translational regulation in re- Anders S, Huber W. 2010. Differential expression analysis for sequence sponse to oxidative stress. Proc Natl Acad Sci 109: 17394–17399. count data. Genome Biol 11: R106. Hunter JD. 2007. Matplotlib: a 2D graphics environment. Comput Sci Arava Y, Wang Y, Storey JD, Liu CL, Brown PO, Herschlag D. 2003. Eng 9: 90–95. Genome-wide analysis of mRNA translation profiles in Saccharomy- Ingolia NT, Ghaemmaghami S, Newman JRS, Weissman JS. 2009. ces cerevisiae. Proc Natl Acad Sci 100: 3889–3894. Genome-wide analysis in vivo of translation with nucleotide resolu- Arribere JA, Doudna JA, Gilbert WV. 2011. Reconsidering movement of tion using ribosome profiling. Science 324: 218–223. eukaryotic mRNAs between polysomes and P bodies. Mol Cell 44: Ingolia NT, Lareau LF, Weissman JS. 2011. Ribosome profiling of mouse 745–758. embryonic stem cells reveals the complexity and dynamics of mam- Ashe MP, De Long SK, Sachs AB. 2000. Glucose depletion rap- malian proteomes. Cell 147: 789–802. idly inhibits translation initiation in yeast. Mol Biol Cell 11: 833– Larsson O, Sonenberg N, Nadon R. 2011. anota: analysis of differential 848. translation in genome-wide studies. Bioinformatics 27: 1440–1441. Benjamini Y, Hochberg Y. 1995. Controlling the false discovery rate: a Lee ES, Forthofer RN. 2006. Analyzing complex survey data.InSeries: practical and powerful approach to multiple testing. J R Stat Soc Quantitative applications in the social sciences (ed. Liao TF), Vol. 71. Series B Stat Methodol 57: 289–300. SAGE Publications, Thousand Oaks, CA. Bishop AC, Buzko O, Shokat KM. 2001. Magic bullets for protein ki- Ptacek J, Devgan G, Michaud G, Zhu H, Zhu X, Fasolo J, Guo H, Jona G, nases. Trends Cell Biol 11: 167–172. Breitkreutz A, Sopko R, et al. 2005. Global analysis of protein phos- Brar GA, Yassour M, Friedman N, Regev A, Ingolia NT, Weissman JS. phorylation in yeast. Nature 438: 679–684. 2012. High-resolution view of the yeast meiotic program revealed Ramachandran V, Shah KH, Herman PK. 2011. The cAMP-dependent by ribosome profiling. Science 335: 552–557. protein kinase signaling pathway is a key regulator of P body foci for- Brauer MJ, Saldanha AJ, Dolinski K, Botstein D. 2005. Homeostatic ad- mation. Mol Cell 43: 973–981. justment and metabolic remodeling in glucose-limited yeast cul- Robinson MD, McCarthy DJ, Smyth GK. 2010. edgeR: a Bioconductor tures. Mol Biol Cell 16: 2503–2517. package for differential expression analysis of digital gene expression Brauer MJ, Huttenhower C, Airoldi EM, Rosenstein R, Matese JC, data. Bioinformatics 26: 139–140. Gresham D, Boer VM, Troyanskaya OG, Botstein D. 2008. Coordi- Ryan O, Shapiro RS, Kurat CF, Mayhew D, Baryshnikova A, Chin B, nation of growth rate, cell cycle, stress response, and metabolic activ- Lin Z-Y, Cox MJ, Vizeacoumar F, Cheung D, et al. 2012. Global ity in yeast. Mol Biol Cell 19: 352–367. gene deletion analysis exploring yeast filamentous growth. Science Broach JR. 2012. Nutritional control of growth and development in 337: 1353–1356. yeast. Genetics 192: 73–105. Tudisca V, Recouvreux V, Moreno S, Boy-Marcotte E, Jacquet M, Castelli LM, Lui J, Campbell SG, Rowe W, Zeef LAH, Holmes LEA, Portela P. 2010. Differential localization to cytoplasm, nucleus or Hoyle NP, Bone J, Selley JN, Sims PFG, et al. 2011. Glucose P-bodies of yeast PKA subunits under different growth conditions. depletion inhibits translation initiation via eIF4A loss and subse- Eur J Cell Biol 89: 339–348. quent 48S preinitiation complex accumulation, while the pentose Zaman S, Lippman SI, Schneper L, Slonim N, Broach JR. 2009. Glucose phosphate pathway is coordinately up-regulated. Mol Biol Cell 22: regulates transcription in yeast through a network of signaling path- 3379–3393. ways. Mol Syst Biol 5: 245.

922 RNA, Vol. 20, No. 6 Downloaded from rnajournal.cshlp.org on October 3, 2021 - Published by Cold Spring Harbor Laboratory Press

Protein kinase A regulates gene-specific translational adaptation in differentiating yeast

Pavanapuresan P. Vaidyanathan, Boris Zinshteyn, Mary K. Thompson, et al.

RNA 2014 20: 912-922 originally published online April 23, 2014 Access the most recent version at doi:10.1261/rna.044552.114

Supplemental http://rnajournal.cshlp.org/content/suppl/2014/04/03/rna.044552.114.DC1 Material

References This article cites 25 articles, 13 of which can be accessed free at: http://rnajournal.cshlp.org/content/20/6/912.full.html#ref-list-1

Creative This article is distributed exclusively by the RNA Society for the first 12 months after the Commons full-issue publication date (see http://rnajournal.cshlp.org/site/misc/terms.xhtml). After 12 License months, it is available under a Creative Commons License (Attribution-NonCommercial 4.0 International), as described at http://creativecommons.org/licenses/by-nc/4.0/. Email Alerting Receive free email alerts when new articles cite this article - sign up in the box at the Service top right corner of the article or click here.

To subscribe to RNA go to: http://rnajournal.cshlp.org/subscriptions