Cbk1 regulation of the RNA binding protein Ssd1 integrates cell fate with translational control

Summary

Spatial control of gene expression, at the level of both transcription and translation, is critical for cellular differentiation [1-4]. In budding yeast, the conserved Ndr/warts kinase Cbk1 localizes to the new daughter cell where it acts as a cell fate determinant. Cbk1 both induces a daughter-specific transcriptional program and promotes morphogenesis in a less well-defined role [5-8]. Cbk1 is essential in cells expressing functional Ssd1, an RNA binding protein of unknown function [9-11]. We show that Cbk1 inhibits Ssd1 in vivo. Loss of this regulation dramatically slows bud expansion, leading to highly aberrant cell wall organization at the site of cell growth. Ssd1 associates with specific mRNAs, a significant number of which encode cell wall remodeling proteins. Translation of these messages is rapidly and specifically suppressed when Cbk1 is inhibited; this suppression requires Ssd1. Transcription of several of these Ssd1-associated mRNAs is also regulated by Cbk1, indicating that the kinase controls both the transcription and translation of daughter-specific mRNAs. This work suggests a novel system by which cells coordinate localized expression of genes involved in processes critical for cell growth and division.

Results and discussion

Cbk1 negatively regulates Ssd1

The budding yeast Ndr/warts family protein kinase Cbk1 controls spatially-regulated gene expression and morphogenesis [5-8]. The kinase is functionally restricted to the daughter cell and localizes to regions of cell growth and cytokinesis [7, 8]. Cbk1 drives an asymmetric cell fate decision by blocking export of the transcription factor Ace2 from the daughter cell nucleus, thereby activating a transcriptional program specific to the new daughter [5, 6, 12]. In addition to its role in transcriptional asymmetry, Cbk1 promotes cell growth through a separate mechanism that remains obscure [7, 8]. Furthermore, the kinase is essential for viability in cells that express a functional form of Ssd1, an RNA-binding protein that has been implicated in numerous processes [9-11, 13-18]. Despite extensive study, Ssd1's in vivo function remains unknown. We hypothesize that Ssd1 is involved in post-transcriptional control of gene expression, and that Cbk1 regulation of Ssd1 may represent a novel mechanism by which the kinase controls localized gene expression and morphogenesis.

Loss of Cbk1 function in cells expressing SSD1 ultimately causes cells to lyse [9, 19]. Overexpression of SSD1 is also deleterious [20]. Together, this suggests that Cbk1 inhibits Ssd1 and that unrestrained Ssd1 activity is lethal. Ssd1 physically interacts with Cbk1 [8, 21], suggesting direct regulation by the kinase. Indeed, Ssd1 contains eight predicted Cbk1 phosphorylation sites, with histidine at −5 and arginine or lysine at −2 or −3 relative to the putative phosphoacceptor residue (Fig. 1 A) [12]. Phosphoproteomic analyses have shown that at least six of these sites are phosphorylated in vivo [22]. Intriguingly, Ssd1 orthologs from distantly related fungi also contain similarly distributed putative Cbk1 phosphorylation sites (Fig. 1 A). In contrast, other proteins that contain an RNase II domain but are unrelated to Ssd1 show no enrichment for Cbk1 consensus motifs. We found that Cbk1 affinity purified from yeast cells robustly phosphorylated Ssd1 in vitro, and mutation of all the predicted phosphoacceptor residues abolished this phosphorylation (Fig. 1 B, S1).

Figure 1. Cbk1 phosphorylates and negatively regulates Ssd1.

(A) Alignment of Ssd1 with orthologs. Ssd1 contains a type II RNase domain, but lacks catalytic residues. Orange indicates highly conserved regions. Predicted Cbk1 phosphorylation sites are marked with orange lines for sites found in highly conserved regions and with black lines for less well conserved regions. (B) Cbk1 phosphorylates Ssd1 in vitro. The N terminus of Ssd1 (residues 1 – 339) and Ssd1-8A were purified as GST-fusion proteins from bacteria. Cbk1-HA was purified from yeast cells (ELY140) and incubated with eluted Ssd1 protein and γ-³²P-ATP to assay in vitro phosphorylation. Phosphorylation was normalized first to Ssd1 protein levels and subsequently to the wildtype signal at five minutes. α-NT5 is a polyclonal antibody against a GST fusion of the first 100 amino acids of Cbk1; it therefore detects both the GST-Ssd1 fusion and Cbk1-HA. (C) Overexpression of Ssd1 is deleterious to cells. The hypomorphic cbk1-as2 allele exacerbates this effect, as does mutation of the eight Cbk1 consensus sites to alanine. Mutation of the Cbk1 consensus sites to aspartic acid rescues the lethality. CBK1 ssd1Δ (ELY853) and cbk1-as2 ssd1Δ (ELY624) cells carrying SSD1, ssd1-8A, or ssd1-8D on the inducible plasmid pEGKT were grown to stationary phase in selective media containing glucose. The effect of Ssd1 overexpression was assessed by plating serial dilutions on selective media containing either galactose (induced, left) or glucose (repressed, right).

If phosphorylation by Cbk1 inhibits Ssd1, then substitution of alanine at Cbk1 phosphorylation sites should create a hyperactive, lethal form of Ssd1. Conversely, substitution of acidic amino acids at these sites should mimic constitutive phosphorylation and render Cbk1 dispensable. We constructed ssd1 alleles under the control of a galactose-inducible promoter in which all eight predicted phosphoacceptor sites are substituted with alanine or aspartic acid (ssd1-8A and ssd1-8D, respectively). Overexpression of ssd1-8A was dominantly lethal (Fig. 1 C, data not shown). We were unable to construct a strain carrying this allele under the control of the endogenous SSD1 promoter. Mutation of these eight amino acids to alanine was not lethal when overexpressed as part of the truncated ssd1-d allele found in W303 cells (Fig. S2 A). Conversely, expression of the ssd1-8D allele was not deleterious in wildtype cells (Fig. 1 C).

We further hypothesized that if Cbk1 inhibits Ssd1, cells carrying a hypomorphic cbk1 allele should be more sensitive to Ssd1 overexpression. We tested this hypothesis by overexpressing SSD1 with a galactose-inducible promoter in cells carrying the analog sensitive cbk1-as2 allele, which can be specifically inhibited by the cell-permeable compound 1NA-PP1. This allele is also hypomorphic with partially compromised catalytic activity even in the absence of inhibitor [6]. We found that wild type Ssd1 overexpression was indeed dramatically more deleterious in cbk1-as2 cells in the absence of inhibitor (Fig. 1 C). In contrast, overexpression of the ssd1-8D allele in cbk1-as2 cells was not lethal, indicating that negative charges at the Cbk1 phosphorylation sites in this region render the kinase dispensable (Fig. 1 C). Taken together, these results indicate that Cbk1 inhibits Ssd1 in vivo by directly phosphorylating the protein. The dominant lethality of ssd1-8A complicates its use for phenotypic analysis. We therefore used 1NA-PP1 inhibition of cbk1-as2 in SSD1 cells to rapidly and conditionally hyperactivate Ssd1, allowing further analysis of its in vivo function.

SSD1 hyperactivation causes defective local cell wall expansion

To better understand the essential function of Cbk1 in SSD1 cells, we further characterized the phenotype that results when cbk1-as2 is inhibited with 1NA-PP1. In G1-synchronized SSD1 cells, inhibition of Cbk1 dramatically slowed bud expansion (Fig. 2 A). This did not occur in ssd1Δ cells (data not shown). This effect is similar to observations reported by Kurischko et al. [23], but we were unable to detect defects in cell polarity or post-golgi vesicle delivery (Y Lin, JM Jansen, and EL Weiss, submitted). Rather, we find a pronounced defect in cell wall organization in these cells. Bud growth requires localized expansion of the cell wall, a process that is a dynamic balance of synthesis and destruction of an extracellular network of glucan polymers and protein. Defects in the expression or function of hydrolytic enzymes required to remodel the extracellular lattice create an imbalance in the cell wall, leading to resistance to enzymatic digestion [24]. We found that inhibiting cbk1-as2 in SSD1 cells resulted in formation of an aberrant cell wall that was extremely resistant to digestion: after 2 hours of 1NA-PP1 treatment, cbk1-as2 SSD1 cells were highly refractory to lysis by treatment with cell wall digesting enzymes (Fig 2 B; Fig S3). In contrast, inhibition of cbk1-as2 in ssd1Δ cells caused only a slight increase in resistance to digestion (Figure 2 B).

Figure 2. cbk1-as2 SSD1 cells fail to grow due to aberrant cell wall remodeling.

(A) Loss of Cbk1 function leads to a dramatic bud growth delay. cbk1-as2 SSD1 (ELY617) cells were synchronized in G1 by elutriation. Cells were treated with either 1NA-PP1 or DMSO and fixed to assess budding over time (n > 200). (B) Ssd1 hyperactivation renders cells resistant to lysis by zymolyase treatment. Wildtype (BY4741), ssd1Δ (ELY853), cbk1-as2 SSD1 (ELY853), and cbk1-as2 ssd1Δ (ELY624) cells were grown to midlog phase and treated with 1NA-PP1 for 2 hours. Cells were then killed with azide and fluoride and treated with zymolyase. The percentage of cells remaining at indicated timepoints was assessed using OD₆₀₀. Points indicated the average of three trials, with standard error indicated. (C) Ssd1 hyperactivation causes a bud specific defect in cell wall organization. Electron micrographs of wildtype (BY4741), ssd1Δ (ELY853), cbk1-as2 SSD1 (ELY853), and cbk1-as2 ssd1Δ (ELY624) cells 2 hours after treatment with 1NA-PP1. Fixed cells were treated with zymolyase for 4.5 hours to digest the wall. Arrowhead: undigestable cell wall remaining on bud. Quantification indicates the number of daughters with intact cell wall by EM (p<10^-20). Scale bar, 0.5 μm

To determine if Cbk1 control of Ssd1 is critical for wall organization in growing buds, we examined electron micrographs of cells exposed to extensive wall digestion following fixation. We found that the cell wall of the bud was impervious to digestion in cbk1-as2 SSD1 cells treated with 1NA-PP1 (Fig. 2 C, red arrow). In contrast, the cell wall was completely removed from wild type, ssd1Δ, and cbk1-as2 ssd1Δ buds treated with 1NA-PP1. Loss of Cbk1 function in SSD1 cells ultimately leads to lysis, suggestive of defective cell wall organization [19]. In agreement with this, we found that the lethal effect of Cbk1 inhibition was partially rescued by addition of an osmotic stabilizer (Fig S2 B). This rescue was not complete, however, and osmotic stabilization did not ameliorate the effect of ssd1-8A overexpression (Fig S2 C). Overall, these findings indicate that loss of Cbk1 regulation of Ssd1 causes lethal defects in cell wall organization, likely by making the cell wall too rigid to expand with the growing bud.

Ssd1 binds specific RNAs in vivo

Ssd1 has been identified as an RNA-binding protein [10, 11]. To determine if the protein associates with specific mRNAs in vivo we affinity purified Ssd1 and identified associated RNAs using whole genome microarrays. We analyzed hybridization data from three independent biological replicates using the significance analysis of microarrays (SAM) algorithm, and found numerous mRNAs significantly enriched in the purification of tagged Ssd1 relative to a mock purification from an untagged strain (Fig. 3 A and Dataset S1). At a stringent 1% local false discovery rate (FDR) we found 152 unique mRNAs significantly enriched at levels ranging from twofold to over 100-fold compared to background. Using quantitative real time PCR (qPCR), we confirmed interactions of Ssd1 with a subset of mRNAs across a range of enrichment values (Fig. 3 B). Inhibition of Cbk1 did not affect Ssd1 association with mRNA (Fig. 3 C), indicating that this is not the activity regulated by Cbk1 phosphorylation.

Figure 3. Ssd1 is an RNA binding protein in vivo.

(A) Ssd1-associated RNAs were purified from mid-log untagged (BY4741) and SSD1-TAP (ELY819) cells in three biologically independent experiments. Data was analyzed using SAM [38]. The number of genes at given local FDRs are indicated. RNAs in the “other” category include sequences that map to antisense transcripts and non-transcribed regions. Highly repetitive sequences, such as retrotransposons, have been omitted for clarity. Cell wall genes are the only significantly enriched class. (B) Ssd1-associated RNAs were purified as in (A). Transcripts representing a range of enrichments were detected using quantitative PCR. Data is depicted on a log plot and represents one experiment out of at least three biological replicates. (C) Cbk1 does not affect Ssd1's ability to bind RNA. Ssd1-associated RNAs were isolated as in (A) and detected as in (B) from untagged (BY4741), SSD1-TAP (ELY819), cbk1-as2 SSD1-TAP (ELY974), and ssd1-8D-TAP (ELY1136), all treated with 1NA-PP1 for 10 minutes. mRNA association is normalized to Ssd1-TAP pulldown. Data from one representative experiment is shown. No reproducible differences were detected between the strains.

Analysis of Gene Ontology (GO) terms shows that Ssd1-associated mRNAs are significantly enriched for messages encoding proteins that function in cell wall organization (Fig. 3 A; Table 1). Intriguingly, three of these wall-related mRNAs (CTS1, DSE2, and SUN4) are transcriptional targets of the Cbk1-regulated transcription factor Ace2 [7, 25]. Recently, Hogan et al. [10] reported a set of 59 Ssd1-associated mRNAs, similarly enriched for cell wall-related genes, as part of a larger study of mRNA binding proteins. We identified 14 of these messages, all of which are involved in cell wall-related processes. The non-overlapping set may be attributable to differences in experimental protocol or false negatives excluded in statistical analysis. Taken together with other recent findings that implicate Ssd1 in cell wall integrity [18, 26], our findings and the work of Hogan et al. suggest that Ssd1 plays an important role in the remodeling and organization of the wall, an essential process that is highly localized to sites of cell growth and cytokinesis.

Table 1.

BGL2*	cell wall
CTS1*
DSE2*
FLO1
HPF1
LRE1*
SCW10*
SCW4*
SIM1*
SRL1*
SUN4*
TOS1*
UTH1*
HSL1*	bud morphogenesis checkpoint
KCC4
MMR1*	bud mitochondria
FLC1	bud site; wall maintenance
GYL1	bud site; trafficking

^*also identified by Hogan et al.

Cbk1 inhibits Ssd1-mediated translational repression of bound mRNAs

The proteins encoded by Ssd1-associated mRNAs prominently include hydrolytic enzymes involved in normal wall morphogenesis. For example, Ssd1 associates with mRNAs encoding the chitinase Cts1 and seven different glucanases, including the closely related SUN family proteins Sim1, Sun4, and Uth1. Absence of these proteins might result in formation of a wall that is unusually rigid. Consistent with this, deletion of several of these genes results in elevated resistance to enzymatic digestion [27, 28].

The aberrant cell wall organization observed upon Cbk1 inhibition (Fig. 2) suggests that Ssd1 hyperactivation is lethal because it reduces the expression of associated mRNAs encoding proteins that remodel the wall at the growth site. We surmised that this might occur through accelerated mRNA decay or suppressed translation. Consistent with a role in mRNA metabolism, Ssd1 contains a conserved C-terminal RNase II domain related to the Dis3 family of nucleases, although Ssd1 and its closest orthologs lack catalytically essential residues [11]. Furthermore, we found that Ssd1-GFP localized strongly to cytoplasmic foci upon glucose starvation or entry into stationary phase (Fig. S4 A). A subset of these foci contained Dcp2, a hallmark of cytoplasmic P bodies that are sites of mRNA storage and degradation during periods of translational repression (Fig. S4 A; [29]). Inhibition of Cbk1 did not affect Ssd1 localization under these conditions (Fig. S4 B).

We assessed Ssd1's role in mRNA decay. Decay rates of selected mRNAs were considerably faster in SSD1 cells, but this effect was not specific to Ssd1-associated messages (Fig. S5 A, C). Cbk1 inhibition further stimulated Ssd1-mediated mRNA decay, and also accelerated mRNA decay in an Ssd1-independent manner (Fig. S5 B – D). Thus, Ssd1 promotes degradation of both bound and unbound mRNAs, suggesting a broad and nonspecific role in mRNA decay that may be secondary to a more direct effect on specifically associated messages.

To determine if Ssd1 influences translation of its bound mRNAs, we fractionated polysomes by centrifugation on a sucrose gradient and quantified the associated mRNAs using Northern blots. Actively translated messages are associated with multiple ribosomes; therefore, co-migration with high molecular weight polysomes refects an mRNA's translational activity [30]. Since numerous mRNAs decrease in abundance upon Cbk1 inhibition in SSD1 cells, this approach provides a more reliable estimate of a message's translation efficiency than bulk protein levels. Overall polysome profiles were indistinguishable between SSD1 and ssd1Δ cells, as well as between SSD1 and ssd1-8D cells (Fig. S6 A – D). Furthermore, the distribution of representative transcripts were largely overlapping between these strains. This is consistent with the mild phenotype associated with loss of SSD1 function and SSD1 overexpression (Fig. 1), which suggests that the protein is normally tightly negatively regulated.

To determine if hyperactivation of Ssd1 selectively affects translation of bound mRNAs we inhibited the cbk1-as2 allele with 1NA-PP1 and examined polysome distribution of selected messages. Bulk polysome profiles were similar in inhibitor-treated cbk1-as2 SSD1 and cbk1-as2 ssd1Δ cells (Fig. 4 A), indicating that there was no global effect on translation. However, upon Cbk1 inhibition, Ssd1-associated mRNAs (UTH1, SUN4, CTS1, SIM1, TOS1) were dramatically depleted from polysome fractions in cbk1-as2 SSD1 cells. These messages remained polysome-associated in cbk1-as2 ssd1Δ cells (Fig. 4 B, S6 G). A similar effect was evident in comparison of cbk1-as2 SSD1 cells treated with 1NA-PP1 or DMSO for 10 minutes (Fig. S6 E, F). In marked contrast, mRNAs that did not bind Ssd1 (ACT1 and PGK1) remained polysome-associated when Cbk1 was inhibited (Fig. 4 B, Fig. S6 E, F). Thus, Cbk1 inhibits Ssd1's direct role in the translational repression of bound targets.

Figure 4. Cbk1 inhibits Ssd1-mediated translational repression of bound transcripts.

(A) Polysome profiles of RNA from cbk1-as2 SSD1 (ELY617) and cbk1-as2 ssd1Δ (ELY624) cells that were grown to mid-log phase and treated for 10 minutes with 1NA-PP1. RNA was separated on a 15-50% sucrose gradient and its distribution analyzed with a continuous flow cell. No gross differences in polysome profiles are evident. Grey boxes highlight polysomes. (B) Material from (A) was collected in 75 second intervals, resulting in 18 fractions, and assayed by Northern blotting with the indicated radiolabeled probes. Depletion of Ssd1-associated mRNAs (UTH1, SUN4, CTS1) from polysome fractions is evident upon cbk1-as2 inhibition in SSD1 cells (orange line) compared with ssd1Δ cells (blue line). Quantification from one representative experiment out of three independent biological replicates is shown. Grey box highlights polysomes. (C) Ssd1 affects Uth1 protein levels. Wildtype (BY4741), ssd1Δ (ELY853), cbk1-as2 SSD1 (ELY853), and cbk1-as2 ssd1Δ (ELY624) cells were grown to mid-log phase and treated with 1NA-PP1 or DMSO for 15 minutes. Lysates were generated with urea and proteins separated by SDS-PAGE.

It is unlikely that Ssd1-mediated translational repression is a secondary effect of mRNA degradation: Ssd1 accelerates ACT1 decay, but the polysome association of this transcript is unaffected. Ssd1's role in translational repression is specific to associated mRNAs, while acceleration of mRNA decay is not specific, suggesting that it may be indirect.

These results predict a decrease in the abundance of proteins encoded by Ssd1-associated mRNAs when Ssd1's activity increases. We examined levels of Uth1 in SSD1 and ssd1Δ strains with both normal and reduced Cbk1 function using an antibody against endogenous Uth1 protein [31]. Consistent with our hypothesis, the abundance of Uth1 was lower in cells expressing functional Ssd1 (Fig. 4 C). The steady state level of Uth1 was dramatically reduced in cbk1-as2 SSD1 cells, consistent with the hypomorphic nature of this allele. Addition of 1NA-PP1 for 15 minutes resulted in a further decrease in Uth1 abundance. This effect was eliminated by deletion of Ssd1. The level of Pgk1, a message which is not associated with Ssd1, was unaffected by either Ssd1 and Cbk1 loss of function (Fig. 4 C). While Uth1 protein levels could be modulated by either mRNA abundance or translation effects, when combined with polysome association data these findings further suggest that Ssd1 acts as a Cbk1-regulated inhibitor of translation.

A novel mechanism for localized control of gene expression

Cbk1 control of Ssd1 may constitute a heretofore-unappreciated signaling mechanism linking translation and stabilization of mRNAs with localized processes requiring their products. We previously showed that active Cbk1 is restricted to the daughter cell [12]. Ssd1 might therefore be preferentially inhibited there, resulting in daughter-specific translation of bound messages. Cbk1 concentrates at areas of active cell growth and wall remodeling, appearing at the site of the expanding bud until late in mitosis and then accumulating at the bud neck immediately prior to actomyosin ring contraction [6]. Cbk1 might inhibit Ssd1 near these sites, allowing localized expression of associated mRNAs. Such localized protein synthesis may help ensure that hydrolytic enzymes important for wall destruction are delivered to appropriate sites, where the machinery that inserts new material into the wall balances the remodeling activity of these enzymes to allow for expansion of the bud. During cell separation, Cbk1 at the bud neck may similarly promote localized translation of enzymes that would have deleterious effects if expressed throughout the cell. Alternatively, Cbk1 activated at sites of growth and cell separation might inhibit Ssd1 more broadly throughout the cell, linking translation of Ssd1-bound mRNAs throughout the cytoplasm to the status of a localized process at the cell periphery.

Ssd1's association with mRNAs encoding the cell separation enzymes Cts1, Dse2, and Sun4 is notably relevant to Cbk1's cell fate control function. Expression of these daughter-specific messages is driven by the transcription factor Ace2, which is activated by Cbk1 [5, 6]. Thus, the kinase may act in a coherent feed-forward loop, turning on transcription of certain genes in the daughter cell and inhibiting a protein that blocks their translation. These connections exemplify coordinated control of multiple processes that converge to reinforce gene expression required to generate asymmetric behavior of mother and daughter cells.

Local control of translation is important for differentiation from yeast to humans. For example, the Puf family protein Khd1 is implicated in the translational repression of ASH1, a daughter-specific gene required for mating type switching in budding yeast [1]. Yck1 relieves this repression by phosphorylating Khd1 [1]. In metazoans, Src performs a similar function, regulating the translation of β-actin through the RNA-binding protein Zbp1 [32]. Cbk1 is the downstream-most component of a signaling pathway called the RAM network (Regulation of Ace2 and Morphogenesis) that is conserved from yeast to metazoans [33], and Ssd1 is similar to the broadly distributed Dis3-related proteins. It will therefore be interesting to determine if related pathways in metazoans also participate in post-transcriptional control of gene expression. While the cell wall is obviously not present in all eukaryotes, localized growth remains necessary to drive cellular expansion and remodeling. Pathways related to the RAM network appear to be involved in such processes. For example, in D. melanogaster the essential Cbk1-related kinase tricornered is critical for normal morphogenesis of cellular extensions and dendrite organization [34-36]. Localized specification of the complement of translatable mRNAs might contribute to organization of these and similar structures.

Experimental Procedures

Yeast strains, plasmids, and cell growth assays

All yeast strains are derived from S288c, BY4741 (Open Biosystems, see Strain table, Table S1). All strain and plasmid construction was confirmed by DNA sequencing and Western blotting. For cell growth assays, stationary phase cells grown in synthetic medium with glucose (SD) lacking uracil were diluted to OD 0.2. Seven-fold serial dilutions were plated on SD lacking uracil and grown 2-3 days at 30°C.

Cell synchronization and microscopy

For bud growth and polarity assessment, 1 L of cells were grown to mid-log phase at 30°C and shifted to 20°C for 1 hour. Cells in G1 were isolated from this population using centrifugal elutriation and collected by filtration. Approximately 300 mL of elutriated cells were concentrated into 1 mL of fresh media containing either 25 μM 1NA-PP1 or an equivalent volume of DMSO, for a final OD of ∼10. Cells were grown at 24°C and aliquots taken at indicated times to assess budding.

For electron micrographs of the cell wall, cells were grown to mid-log phase and treated with 25 μM 1NA-PP1 for 2 hours at 30°C, harvested by centrifugation and washed twice with deionized H₂O. For detailed methods see Supplemental Materials. In brief, cells were fixed in gluteraldehyde, digested with zymolyase for 2.5 hours, stained with osmium tetroxide and uranyl acetate.

Cell wall integrity

Cells were grown to mid-log phase in synthetic media with glucose and treated with 25 μM 1NA-PP1 or DMSO for 2 hours at 30°C. 1 OD of cells was pelleted from each culture, washed in 1 mL zymolyase assay buffer (20 mM NaF, 20 mM NaN₃, 10 mM Tris-HCl), and resuspended in 1 mL zymolyase assay buffer containing 10 μL 10T zymolyase and 1 μL β-mercaptoethanol. The zymolyase-treated cell suspensions were incubated at 37°C, and OD₆₀₀ was monitored over three hours with a 96-well plate reader.

In vitro kinase assays

Protein purification and kinase assays were performed as previously described and in further detail in the Supplemental Methods section [37]. In brief, Cbk1-HA was purified from log phase cells grown in YPD using α-HA (a generous gift from Dr. R. Lamb) and protein G sepharose beads. The first 339 amino acids of Ssd1 and Ssd1-8A were cloned into the pGAT2 vector and expressed in ER2566 E. coli cells (NEB). The kinase was divided evenly between 1 μg of each GST- Ssd1 N terminal fragment fusion and GST alone. A sample was removed from the reaction at indicated timepoints and the resulting signal was normalized to protein levels and then to wild type Ssd1 signal.

Purification of Ssd1-associated RNAs

Ssd1-associated RNAs were purified and detected as described in the Supplemental Methods. Briefly, RNA was extracted from Ssd1-TAP purified from yeast. Amino-allyl labeled DNA was synthesized and hybridized to arrays using a mixture of anchored oligo-dT and random hexamers for reverse transcription. Arrays were scanned with an Agilent G2505B scanner (Agilent Technologies, Santa Clara, CA). The arrays contain 23,321 unique 70-mer probes covering the S. cerevisiae genome and consists of the commercially available 70-mer oligo set: Operon Yeast Set v1.1 which has been supplemented with a custom designed 70-mer oligo set representing genic and intergenic regions. RNA abundance was also measured using Affymetrix GeneChip Yeast Genome 2.0 Arrays (Affymetrix, Santa Clara, CA). Data from three independent biological replicates was analyzed using Significance Analysis of Microarrays [38]. Threshold was set at 1% local FDR. See supplement for full data table.

Detection by qPCR was performed as above, but the purification protocol was scaled-down 10-fold. Reverse transcription was primed only with oligo dT and messages were detected by incorporation of Sybr green (Invitrogen) into amplicons generated with the primers listed in the Supplemental Methods. Relative concentrations were obtained using standard curves generated from 5-fold serial dilutions of genomic DNA.

Polysome profiles and Northern blots

For a detailed description of polysome profiles see Supplemental Methods. In brief, cells were grown to mid-log phase and cycloheximide was added to a final concentration of 0.1 mg/mL. Lysates were prepared using glass bead lysis and cleared by centrifugation at 4000 rpm for 2 minutes. Approximately 90 A₂₅₄ units were loaded on an 15 – 50% sucrose gradient, and centrifuged in a Beckman SW41 rotor at 39000 rpm for 2.5 hours at 4°C. RNA was detected by reading A₂₅₄ using a continuous flow cell. Fractions of 0.75 mL were collected every 75 seconds and RNA was extracted with phenol chloroform/isoamyl alcohol (Sigma Aldrich) and then chloroform. RNA was ethanol precipitated and resuspended in 10 μL of sterile water. 5 μL was denatured in loading dye for 15 minutes at 55°C and loaded onto a formaldehyde-based denaturing agarose gel. Blots were transferred to nylon membrane and prehybridized in Church buffer at 65°C for 2 hours. Radiolabeled probes were hybridized to the blots at 65°C for 4 hours. Blots were washed twice for 10 minutes in 2×SSC with 0.1% SDS and exposed to phosphorimager screen overnight. Signal was quantified using ImageQuant (GE Healthcare) and each fraction was normalized to total signal. Blots were stripped with boiling 0.1% SDS in water twice for 1 hour at 65°C and then reprobed.