The yeast Cbk1 kinase regulates mRNA localization via the mRNA-binding protein Ssd1

LATS/NDR kinases are an evolutionarily conserved subfamily of AGC (protein kinase A/G/C) kinases that are implicated in cellular transformation and growth control (Hergovich et al., 2006, 2008). Mammalian LATS and NDR kinases function as tumor suppressors and are implicated in a variety of cancers, including B cell lymphomas, prostate, pancreatic, and ovarian cancers. Despite their importance, the mechanisms by which LATS/NDR kinases influence cancer progression are poorly understood (Hergovich and Hemmings, 2009).

Data regarding fungal LATS/NDR kinase orthologues offer insight to the molecular mechanisms of mammalian LATS/NDR function. In yeast, two distinct LATS/NDR kinase orthologues control several aspects of cell growth and proliferation. One class of yeast LATS/NDR kinases, Saccharomyces cerevisiae Dbf2/Dbf20 and Schizosaccharomyces pombe Sid2, are key components of a conserved signaling network known as the S. cerevisiae mitotic exit network (MEN) and S. pombe septum initiation network (SIN) that coordinate cytokinesis and mitotic exit (Johnston et al., 1990; Komarnitsky et al., 1998; Hou et al., 2000; Luca et al., 2001; Balasubramanian et al., 2004). The other yeast LATS/NDR orthologues, S. cerevisiae Cbk1 and S. pombe Orb6, belong to a MEN/SIN-related signaling network that regulates polarized morphogenesis and maintenance of cell integrity (Verde et al., 1998; Racki et al., 2000; Bidlingmaier et al., 2001; Hou et al., 2003; Nelson et al., 2003).

S. cerevisiae Cbk1 is a late-acting kinase in the RAM signaling network, which is comprised of two kinases and four associated proteins (Jorgensen et al., 2002; Nelson et al., 2003). Each RAM protein is essential for viability and localizes to sites of polarized growth (Racki et al., 2000; Bidlingmaier et al., 2001; Du and Novick, 2002; Nelson et al., 2003). Current data suggest that Cbk1 and RAM regulate several functions, including polarized growth, mating efficiency, cell separation, and maintenance of cell integrity (Weiss et al., 2002; Nelson et al., 2003; Kurischko et al., 2005, 2008; Bourens et al., 2009; Jansen et al., 2009; Panozzo et al., 2010). Recent data suggest that Cbk1 also influences polarized secretion and Golgi function (Kurischko et al., 2008). Cbk1 controls cell separation via the daughter cell–specific transcription factor Ace2, which induces the expression of septum degradation enzymes at the end of mitosis (Colman-Lerner et al., 2001; Weiss et al., 2002; Bourens et al., 2008; Mazanka et al., 2008; Sbia et al., 2008). The mechanism of Cbk1 function in other processes is less understood; however, it is increasingly apparent that Cbk1 has a role in cell wall biogenesis. In support, the lethality of cbk1Δ and other RAM mutants is suppressed by moderate overexpression of certain cell wall biosynthesis proteins, including the mannoproteins Sim1, Srl1, and Ccw12 (Kurischko et al., 2005, 2008).

Perhaps the most important function for Cbk1, with respect to cell viability, is to regulate the mRNA-binding protein Ssd1. In support, CBK1 and all RAM genes are dispensable for viability in cells harboring ssd1Δ or the polymorphic ssd1-d mutation (Du and Novick, 2002; Jorgensen et al., 2002; Nelson et al., 2003; Kurischko et al., 2005). Moreover, Cbk1 phosphorylates Ssd1 in vitro and associates with Ssd1 via two-hybrid assays (Racki et al., 2000; Jansen et al., 2009). These data support the model that Cbk1 negatively regulates Ssd1 and suggest that the failure to inhibit Ssd1 is detrimental to cell integrity.

The molecular functions of Ssd1 are not well understood; however, Ssd1 is genetically linked to multiple cellular functions, including maintenance of cell integrity, mRNA processing, stress signaling, and cellular aging (Evans and Stark, 1997; Uesono et al., 1997; Moriya and Isono, 1999; Ibeas et al., 2001; Jorgensen et al., 2002; Kaeberlein and Guarente, 2002; Wheeler et al., 2003; Kaeberlein et al., 2004; Wojda et al., 2007; Jansen et al., 2009; Mir et al., 2009). Recent data also link Ssd1 to ribosome assembly/function (Li et al., 2009). Insight to Ssd1 functions comes from the identification of its associated mRNAs. Ssd1 was shown to specifically associate with a subset of mRNAs, including several that encode cell wall mannoproteins and cbk1 dosage suppressors (Hogan et al., 2008; Jansen et al., 2009). Intriguingly, some Ssd1-associated mRNAs, including SRL1, were shown to localize asymmetrically to the bud tip during polarized growth (Shepard et al., 2003; Hasegawa et al., 2008). Thus, Ssd1 may modulate the expression and/or mRNA localization of cell integrity proteins.

To elucidate the role of Cbk1 kinase on Ssd1 localization and function, we analyzed the effects of phospho-deficient and phosphomimetic Ssd1 mutants in vivo. Here, we provide evidence that Cbk1-mediated phosphorylations promote the cortical localization and function of Ssd1–mRNA complexes while repressing the growth inhibitory functions of Ssd1 at mRNA processing bodies (P-bodies) and stress granules. Further, our data indicate that Cbk1 inhibition and cellular stress promote Ssd1–mRNA interactions with P-bodies and stress granules, which sequester mRNAs from translation machinery, thus supporting the model that hypo-phosphorylated Ssd1 mediates the translational repression of its associated mRNAs. These data reveal a new role for LATS/NDR tumor suppressor kinases in mRNA regulation.

Current data suggest that Cbk1 kinase and Ssd1 are functionally linked to cell growth and maintenance of cell wall integrity (Jorgensen et al., 2002; Kurischko et al., 2005, 2008; Jansen et al., 2009); however, the molecular function of Ssd1 is poorly understood. The majority of Cbk1 kinase localizes to the bud cortex during bud growth and to the bud neck during cytokinesis, consistent with a role in polarized growth (Racki et al., 2000; Colman-Lerner et al., 2001; Weiss et al., 2002; Nelson et al., 2003). In contrast, Ssd1 was reported to localize uniformly throughout the cytoplasm (Uesono et al., 1997; Jansen et al., 2009).

Given the cooperative roles of Cbk1 and Ssd1 in polarized growth and cell wall biogenesis, we postulated that some Ssd1 must localize similarly to Cbk1. Thus, we reinvestigated the subcellular distribution of Ssd1 using cells expressing integrated Ssd1-GFP under the control of its physiological promoter. We observed a complex pattern of Ssd1-GFP localization in logarithmically growing cells, indicating that Ssd1 is spatially regulated. Most Ssd1 localized diffusely to the cytoplasm, as previously reported (Uesono et al., 1997; Jansen et al., 2009). In addition, Ssd1 concentrated to the bud cortex in some (13–40%, depending on the culture) small budded cells (Fig. 1). Ssd1 also localized to the bud neck in 4–9% of large budded cells (Fig. 1). These data indicate that a fraction of Ssd1 localizes similarly to Cbk1, which is consistent with a Cbk1-dependent function for Ssd1 during polarized growth.

To determine if Cbk1 kinase influences Ssd1 localization we monitored Ssd1-GFP in cells expressing an analogue-sensitive cbk1-as allele, which encodes a derivative of Cbk1 that is specifically inhibited by the drug 1NA-PP1 (Weiss et al., 2002). When cbk1-as cells were treated with 1NA-PP1 for 45 min, some Ssd1-GFP redistributed from the predominantly diffuse cytoplasmic localization to randomly distributed cytoplasmic puncta (Fig. 2 A). We rarely (<1% cells) observed Ssd1-GFP at the bud cortex or bud neck after Cbk1 inhibition, suggesting that the establishment or maintenance of Ssd1 at sites of polarized growth is Cbk1 dependent. Ssd1 localization was not affected in mock-treated cbk1-as cells or in 1NA-PP1–treated wild type (unpublished data). These data suggest that Ssd1 localization is modulated by Cbk1-dependent phosphorylation and support the hypothesis that Cbk1 kinase promotes the cortical localization of Ssd1 and/or represses the association of Ssd1 with cytoplasmic puncta.

We also monitored Ssd1 localization after its modest overexpression. We introduced low copy plasmids of SSD1-GFP under the control of the constitutive glyceraldehyde-3-phosphate dehydrogenase (GPD) promoter and monitored Ssd1 localization in asynchronous ssd1Δ cells. Modest Ssd1-GFP overexpression did not cause any obvious growth or cell integrity defects. However, up to 60% of the cells containing GPD-driven Ssd1-GFP plasmids contained prominent cytoplasmic foci that were usually excluded from buds (Fig. 2 B). Considering that most Cbk1 kinase localizes to buds during polarized growth (Racki et al., 2000; Colman-Lerner et al., 2001; Weiss et al., 2002; Nelson et al., 2003), these data are consistent with the model that moderate Ssd1 overexpression overwhelms regulatory Cbk1 phosphorylations in mother cells where Cbk1 concentrations are low, but not in buds where Cbk1 concentrations are higher.

The Ssd1 cytoplasmic puncta caused by modest Ssd1 overexpression or by Cbk1 inhibition were reminiscent of cytoplasmic mRNA P-bodies and stress granules, which sequester untranslated mRNAs and contribute to translational repression during cellular stress (Parker and Sheth, 2007; Anderson and Kedersha, 2008, 2009; Buchan et al., 2008). In support, Ssd1 was reported to partly colocalize with a P-body protein during stationary phase (Jansen et al., 2009). To determine if Ssd1 associates with P-bodies or stress granules during logarithmic growth, we monitored Ssd1 localization in cells expressing GPD-driven Ssd1-GFP and RFP-tagged P-body and stress granule proteins. We observed that ∼60% of the Ssd1-GFP cytoplasmic puncta colocalized with the core P-body protein Edc3 in logarithmically growing cells (Fig. 2 C). These data support the model that Ssd1 at least transiently associates with mRNA-sequestering protein complexes that mediate translational repression.

Cellular stresses, such as glucose depletion, hypertonic stress, and heat shock trigger P-body aggregation and stress granule formation (Parker and Sheth, 2007; Anderson and Kedersha, 2008, 2009; Buchan et al., 2008). To test if cellular stress influences Ssd1 localization, we monitored Ssd1-GFP localization after glucose depletion. Ssd1 localized prominently to cytoplasmic puncta within 10 min of glucose depletion (Fig. 3; Fig. S1). Most of the Ssd1 cytoplasmic puncta (∼73–90%) colocalized with the P-body proteins Edc3, Lsm1, and Dcp2 (Fig. 3; Fig. S1). Ssd1 also colocalized with the stress granule proteins Pab1 and Pub1 upon glucose depletion. We observed similar results in salt-stressed cells (unpublished data). Collectively, these data suggest that Cbk1 kinase inhibition and cellular stress promote Ssd1 recruitment to both P-bodies and stress granules, which in turn may lead to the translational repression of Ssd1-associated mRNAs.

To biochemically corroborate data from the colocalization experiments, we conducted coprecipitation experiments with cells expressing Ssd1-GFP and representative TAP-tagged P-body and stress granule proteins. We discovered that Ssd1 specifically coprecipitated with the P-body or P-body–associated proteins Dcp2, Not1, and Xrn1 (Fig. 4). Ssd1 also coprecipitated with Pat1, a topoisomerase II–associated deadenylation-dependent mRNA-decapping factor necessary for both P-body and stress granule assembly (Buchan et al., 2008). These data confirm our colocalization experiments and suggest that a key function for Ssd1 involves P-body and stress granule interactions.

Because P-bodies and stress granules are comprised of mRNA–protein complexes, Ssd1 recruitment or binding to P-bodies might occur via mRNA interactions. To determine if Ssd1 mRNA binding is a prerequisite for associating with P-bodies, we analyzed the localization of truncated derivatives of Ssd1, such as Ssd1^1-450, which lack the mRNA-binding domain, as defined by Uesono et al. (1994). Upon stress induction (hypertonic stress, heat shock, and glucose depletion), Ssd1^1-450 prominently localized to P-bodies within several minutes of stress induction (unpublished data). These data indicate that the N-terminal 450–amino acid region of Ssd1 is sufficient for P-body recruitment and that establishment and maintenance of Ssd1–P-body interactions are independent of mRNA binding.

Ssd1 contains multiple Cbk1/LATS kinase consensus sequences for phosphorylation in its N-terminal region, as defined by Hao et al. (2008) and Mazanka et al. (2008). Significantly, in vivo expression of an Ssd1 mutant protein lacking multiple Cbk1 phosphorylation sites was reported to be toxic to yeast (Jansen et al., 2009); however, the nature of its toxicity and the effect of Cbk1-dependent phosphorylations on Ssd1 localization were not established. To explore the physiological significance of Cbk1-mediated phosphorylations on Ssd1 function and localization, we analyzed the phenotypes of Ssd1 phosphorylation site mutants. We converted nine putative Cbk1 phosphorylation sites to nonphosphorylatable alanines to yield Ssd1-9A. We also converted putative Cbk1 phosphorylation sites to phosphomimetic aspartate residues to yield phosphomimetic Ssd1-9D. We introduced low copy plasmids encoding wild-type or mutant Ssd1 into wild-type, ssd1Δ, and cbk1Δ ssd1Δ yeast strains and analyzed cellular phenotypes.

We discovered that Ssd1-9A expression was dominantly toxic to wild-type and ssd1Δ cells at 22°C and lethal at 37°C (Fig. 5). The Ssd1-9A transformants of wild-type and ssd1Δ cells were very fragile and propagated poorly. Microscopic analysis revealed that Ssd1-9A expression caused severe cell morphology and lysis defects that were remarkably similar to the phenotypes of conditional cbk1 mutants (Fig. 6 A). Ssd1-9A expression was also lethal in cbk1Δ ssd1Δ double mutant cells, as is wild-type Ssd1 (unpublished data). These data support the model that Cbk1 negatively regulates Ssd1 and indicate that hypo-phosphorylated Ssd1 causes dominant and severe growth and cell integrity defects.

To test if the toxicity of Ssd1-9A expression is dependent on mRNA binding, we constructed a plasmid expressing Ssd1-9A protein that lacks the mRNA-binding domain (RBD 686–788), as defined by Uesono et al. (1997). Ssd1-9A-RBDΔ expression was not toxic to ssd1Δ or wild-type cells and did not cause any obvious growth defects (Fig. 6 C). Furthermore, Ssd1-RBDΔ is not toxic to cbk1Δ cells, in contrast to wild-type Ssd1, indicating that the RNA-binding domain is essential for Ssd1 function (unpublished data). Ssd1-9A-RBDΔ expression also disrupted polarized Srl1 mRNA localization, similarly to ssd1Δ (see Fig. S4 B). Thus, the Ssd1 mRNA-binding domain is essential for Ssd1 function, in agreement with Uesono et al. (1997).

If Cbk1 phosphorylation prevents the toxicity of Ssd1, then the lysis phenotypes of cbk1 loss-of-function mutants and cells expressing Ssd1-9A must occur via similar mechanisms. If this hypothesis is correct, then some cbk1 dosage suppressors may abrogate or diminish the toxicity of Ssd1-9A expression. We previously established that high copy plasmids of the cell wall biogenesis proteins, Ccw12, Sim1, Srl1, and Uth1 suppress the lethality of cbk1 loss-of-function mutants (Kurischko et al., 2005 and Fig. S2). Thus, we introduced Ssd1-9A plasmids into cells containing cbk1 dosage suppressor plasmids (pCCW12, pSIM1, pSRL1, pUTH1, pZRG8) and assayed cell morphology and cell integrity by microscopy. We discovered that the cellular lysis and aberrant morphology phenotypes of Ssd1-9A cells were partially rescued by these cbk1 dosage suppressors at 22°C, 34°C (Fig. 6, A and B), and 37°C (not depicted). These data support the hypothesis that the toxicity of Ssd1-9A expression and the cell integrity defects of cbk1 loss-of-function mutants occur via the same or similar mechanisms.

Because Ssd1 localizes to P-bodies in response to Cbk1 inhibition, we hypothesized that hypo-phosphorylated Ssd1-9A constitutively localizes to P-bodies and stress granules. Thus, we analyzed the localization of GFP-tagged Ssd1-9A in ssd1Δ cells. Because constitutive Ssd1-9A expression (via GPD promoter) was toxic for cell growth, we conducted experiments with cells containing galactose-inducible Ssd1-9A plasmids. When cells were grown in repressing conditions (glucose-containing medium), Ssd1-9A–GFP was not detectable by fluorescence microscopy and did not cause any obvious phenotype (unpublished data). Upon 2 h galactose induction, Ssd1-9A localized prominently to cytoplasmic puncta in nearly all cells (Fig. 7 A). Most of the Ssd1-9A cytoplasmic puncta colocalized with P-body markers. We never observed phospho-deficient Ssd1-9A at bud tips or bud necks. These results suggest that the lethality of ssd1-9A and conditional cbk1 mutants correlates with enhanced or constitutive Ssd1 localization to P-bodies and stress granules. Furthermore, these data support the model that hypo-phosphorylated Ssd1 is toxic to cells via P-body–dependent translational repression of Ssd1-associated mRNAs. In agreement, Ssd1 is no longer toxic when it cannot bind mRNAs (Fig. 6 C and Fig. S4 B).

To determine the role of Cbk1 phosphorylation on Ssd1 function, we analyzed the phenotypes of cells expressing phosphomimetic Ssd1-9D. In contrast to phospho-deficient Ssd1-9A, phosphomimetic Ssd1-9D was not toxic to wild-type, ssd1Δ, or cbk1Δ ssd1Δ cells when expressed on low copy plasmids under the control of its endogenous promoter or the GPD promoter (Fig. 5 and unpublished data). Moreover, the morphology and viability of cells expressing Ssd1-9D plasmids were indistinguishable from corresponding control cells containing empty vectors. These data support the model that Cbk1 negatively regulates Ssd1 by phosphorylation, and thereby prevents the dominant toxic/lethal effects of hypo-phosphorylated Ssd1.

To determine if Cbk1 phosphorylation directly influences Ssd1 localization, we analyzed Ssd1-9D-GFP localization in asynchronous cells. Like wild-type Ssd1, most Ssd1-9D localized diffusely to the cytoplasm. Ssd1-9D also localized prominently to the bud cortexes and bud necks of some small and large budded cells, similar to wild-type Ssd1 (Fig. 7 B). Significantly, and in contrast to wild-type Ssd1 or Ssd1-9A, Ssd1-9D rarely localized to bright cytoplasmic foci (P-bodies) when expressed via the constitutive GPD promoter. These data support the model that Cbk1-dependent phosphorylations prevent Ssd1 from associating with P-body aggregates, but allow Ssd1 to localize to sites of polarized growth.

Ssd1 binds a subset of mRNAs, many of which encode cell wall biogenesis proteins (Hogan et al., 2008; Jansen et al., 2009). Intriguingly, several Ssd1-associated mRNAs are encoded by cbk1 dosage suppressors, such as SRL1, SIM1, CCW12, and UTH1 (Kurischko et al., 2005; see Fig. S2). Moreover, two of these mRNAs (SRL1 and UTH1) are known to polarize to the bud cortex in an actin- and myosin V–dependent fashion (Shepard et al., 2003). In light of the polarized localizations of wild-type Ssd1 and phosphomimetic Ssd1-9D in some budded cells, it seemed plausible that Ssd1 influences asymmetric mRNA localization during polarized growth.

To determine whether Ssd1 influences mRNA localization, we monitored SRL1 mRNA localization in vivo using a GFP-tagging strategy described previously (Haim et al., 2007; Haim-Vilmovsky and Gerst, 2009). We compared the pattern of SRL1 mRNA localization in asynchronously growing ssd1Δ cells containing Ssd1 plasmids or empty vector. In ∼50% of the cells expressing wild-type SSD1, SRL1 mRNA polarized to 1–3 prominent spots in the bud (Fig. 8; Shepard et al., 2003). We also observed that SRL1 mRNA localized to multiple faint spots randomly distributed throughout the cytoplasm in ∼40–50% of the cells. The overall distribution of SRL1 mRNA localization was similar in small, medium, and large budded cells, indicating that the different patterns of SRL1 mRNA localization were not obviously enhanced during cell cycle progression (unpublished data). We obtained similar results with wild-type cells (unpublished data).

In ssd1Δ cells containing empty vector, SRL1 mRNA polarity was diminished by ∼10-fold compared with SSD1 cells (Fig. 8). Moreover, the percentage of ssd1Δ cells with multiple faint SRL1 mRNA puncta or no spots was ∼30% and severalfold greater, respectively, than corresponding SSD1 cells (Fig. 8). These data indicate that Ssd1 is critical for establishing or maintaining asymmetric SRL1 mRNA localization during polarized growth.

If Cbk1-mediated phosphorylations simply inactivate Ssd1, then constitutively phosphorylated Ssd1 should phenocopy ssd1 loss-of-function alleles, such as ssd1Δ. However, in light of the cortical localization of wild-type Ssd1 and Ssd1-9D, we thought it to be likely that Cbk1 stimulates some cortical Ssd1 functions and protein interactions. To test if Ssd1-9D can rescue the SRL1 mRNA localization defects in ssd1Δ cells, we monitored SRL1 mRNA localization in ssd1Δ cells expressing Ssd1-9D plasmids. Significantly, Ssd1-9D expression restored the diminished SRL1 mRNA polarity in ssd1Δ cells to a greater degree (∼1.4-fold) than wild-type Ssd1 (Fig. 8). Moreover, Ssd1-9D expression reduced the percentage of cells with dispersed SRL1 mRNA puncta by approximately two- to threefold, in comparison to corresponding ssd1Δ cells and SSD1-expressing cells. These data indicate that Cbk1-phosphorylated Ssd1 promotes and enhances the polarity of SRL1 mRNA in asynchronously growing cells.

Because Ssd1-9A constitutively localizes to P-bodies, we predicted that Ssd1-9A expression also brings its associated mRNAs to P-bodies. Thus, we monitored SRL1 mRNA localization in cells expressing galactose-inducible RFP-tagged Ssd1-9A. Significantly, within 2 h of Ssd1-9A induction, SRL1 mRNA dispersed to multiple cytoplasmic puncta, all of which colocalized with Ssd1-9A (Fig. 9). We observed the same punctate SRL1 mRNA localization in cells expressing HA-tagged Ssd1-9A (Fig. S4). In contrast, SRL1 mRNA did not disperse or redistribute to P-bodies when wild-type Ssd1 or Ssd1-9D expression was induced for the same duration (unpublished data). Corresponding immunoblots confirmed that Ssd1-9A and wild-type Ssd1 were expressed to similar levels in these experiments (unpublished data). These results support the model that in the absence of Cbk1 phosphorylations, hypo-phosphorylated Ssd1 sequesters its associated mRNAs in P-bodies and stress granules, where the mRNAs are translationally repressed. These data further suggest that enhanced Ssd1–mRNA interactions with P-bodies and stress granules are the primary cause of Ssd1 toxicity in cbk1 mutants.

Collectively, our data support the model that the essential function of Cbk1 is to modulate Ssd1 and mRNA metabolism during polarized growth (Fig. 10). For this model, Cbk1 phosphorylates Ssd1 to promote the polarized localization of mRNAs involved in bud growth and cell wall biogenesis. Thus, Cbk1-phosphorylated Ssd1 may function to promote protein synthesis at cortical sites of growth. In contrast, Cbk1 inhibition or cellular stress lead to elevated levels of dephosphorylated Ssd1 and thereby promote Ssd1 association with P-bodies and stress granules. We suggest that Ssd1 associations with P-bodies and stress granules lead to the translational repression of Ssd1-associated mRNAs. Thus, the cell lysis phenotypes of ssd1-9A and conditional cbk1 mutant cells are likely caused by the reduced expression of cell wall proteins that are encoded by Ssd1-associated mRNAs. In support, the polysome fraction of several Ssd1-associated mRNAs is dramatically decreased upon Cbk1 inhibition (Jansen et al., 2009). Furthermore, the lethality of ssd1-9A and cbk1 mutants is rescued by moderate overexpression of several Ssd1-associated mRNAs encoding cell wall proteins (Srl1, Ccw12, Sim1, and Uth1).

Several lines of evidence suggest that Cbk1-phosphorylated Ssd1 contributes to asymmetric mRNA localization. First, both phosphomimetic Ssd1-9D and wild-type Ssd1 localize to bud cortex and bud neck. Moreover, two Ssd1-associated mRNAs, SRL1 and UTH1, localize asymmetrically to the bud tip (Shepard et al., 2003). Both mRNAs encode cell wall biosynthesis proteins and suppress cbk1 and ssd1-9A mutants when moderately overexpressed (Kurischko et al., 2005; Fig. 6 and Fig. S2). Most notably, the polarized localization of SRL1 mRNA is dramatically reduced in ssd1Δ cells, and is restored and modestly enhanced by Ssd1-9D expression (Fig. 8). Based on the physical interactions between Ssd1 and SRL1 mRNA and the observation that Ssd1 transiently localizes to the bud cortex, we hypothesize that Ssd1 directly modulates the delivery, retention, and translation of at least some of its associated mRNAs to the sites of polarized growth. Thus, Cbk1 and Ssd1 may help ensure that its associated mRNAs are translated where they are needed.

Cbk1-phosphorylated Ssd1 may influence SRL1 mRNA localization directly or indirectly via promoting or stabilizing mRNA interactions with either mRNA localization machinery or with cortical proteins. In S. cerevisiae, 24 mRNAs, including SRL1 mRNA, localize to the bud tips via myosin V (Myo4)–mediated transport along actin cables (Shepard et al., 2003; Martin and Ephrussi, 2009). Polarity landmark proteins such as Sec3 were also shown to facilitate asymmetric mRNA and cortical ER interactions (Aronov et al., 2007). Although SSD1 and the mRNA polarity machinery are not essential for viability in yeast, the directed localization of mRNAs provides a mechanism for regulating gene expression with precise temporal and spatial control. The spatial distribution of mRNAs is more economical than the post-translational protein delivery with regard to localized protein expression because mRNAs can be translated many times (Martin and Ephrussi, 2009).

The association of Ssd1 with P-bodies and stress granules suggests a role for Ssd1 in translation repression. P-bodies are conserved RNA protein granules that serve as storehouses for nontranslated mRNAs and sometimes target mRNAs for degradation (Bruno and Wilkinson, 2006; Sheth and Parker, 2006; Buchan et al., 2008; Gallo et al., 2008; Nissan and Parker, 2008). They consist of translation repressors, mRNA decapping proteins, and a 5′–3′ exonuclease. P-bodies are particularly important for translational repression during cellular stress, when repression of many mRNAs is crucial to halt growth and enhance cell survival (Bruno and Wilkinson, 2006; Sheth and Parker, 2006; Buchan et al., 2008; Gallo et al., 2008; Nissan and Parker, 2008). P-bodies are critical for the formation of stress granules, which are comprised of aggregates of untranslating mRNAs, a subset of translation initiation factors, the 40S ribosome subunit, and poly(A)-binding proteins, such as Pab1 and Pub1 (Anderson and Kedersha, 2008, 2009; Buchan et al., 2008). Stress granules form in response to translation initiation defects and cellular stress and dynamically interact with P-bodies, suggesting an exchange of mRNPs. Both P-body– and stress granule–mediated mRNA sequestration provides a rapid and reversible mechanism for translational repression. Our data regarding Ssd1 and P-body/stress granule interactions support the model that in the absence of Cbk1-mediated phosphorylations, Ssd1-associated mRNAs are transitionally repressed via P-body and/or stress granule sequestration. In agreement, SRL1 mRNA colocalizes with Ssd1-9A (Fig. 9), which constitutively localizes to P-bodies. Moreover, Cbk1 kinase inhibition reduces the levels of several Ssd1-associated mRNAs in polysome fractions (Jansen et al., 2009).

Our results also support the model that some cellular stresses (glucose depletion, hypertonic stress, heat shock) promote the P-body and stress granule association of Ssd1 and subsequent translational repression of cell growth proteins. The properties of the phospho-deficient ssd1-9A mutant suggest the existence of a stress-induced phosphatase that promotes Ssd1–P-body/stress granule complexes by removing Cbk1 phosphorylations (Fig. 10). Alternatively or additionally, there may be mechanisms that inhibit Cbk1 activity or disrupt protein interactions in response to stress.

In principle, unphosphorylated Ssd1 could repress the expression of cell wall proteins by promoting mRNA degradation. Indeed, the mRNA-binding domain of Ssd1 shares some homology with RNase II (Uesono et al., 1997). However, there is no experimental evidence supporting the notion that Ssd1 promotes mRNA degradation in vitro or in vivo. Recombinant Ssd1 has no RNase activity in vitro (Uesono et al., 1997) and microarray and RT-PCR experiments reveal no changes in the steady-state levels of Ssd1-associated mRNAs (Li et al., 2009 and Fig. S3). Moreover, SRL1 mRNA is clearly detectable in ssd1Δ or ssd1-9A cells by microscopic and RT-PCR methodology (Fig. 9; Figs. S4 and S5). Intriguingly, recent data indicate that Ssd1 is required to stabilize a population of CLN2 mRNA after heat shock (Ohyama et al., 2010). Taken together, Ssd1 may be required to prevent the degradation of some mRNAs during heat shock and other cellular stresses.

Cbk1 may influence asymmetric protein expression via both Ssd1-dependent and Ssd1-independent mechanisms. Indeed, it is well established that Cbk1 regulates daughter cell–specific transcription of cell separation proteins via Ace2 transcription factor (Colman-Lerner et al., 2001; Weiss et al., 2002). Cbk1 phosphorylates Ace2 on its N-terminal nuclear export sequence (NES), thereby interfering with its Crm1-dependent nuclear export while promoting Ace2 transcriptional activity in the daughter cell nucleus (Jansen et al., 2006).

Intriguingly, the mRNAs of several Ace2-regulated genes, such as CTS1 and DSE2, were shown to bind Ssd1 (Hogan et al., 2008; Jansen et al., 2009; Fig. S3). Moreover, CTS1 mRNA was shown to be transcribed only in the daughter cell and to colocalize with the cortical ER (Bourens et al., 2008). Thus, it is likely that Cbk1-phosphorylated Ssd1 also mediates the spatial distribution and expression of these and other Ssd1-associated mRNAs, as we have demonstrated for SRL1 mRNA.

Cbk1 may also regulate the asymmetric expression of proteins via polarized secretion mediators. In support, Cbk1 influences secretion and Golgi-mediated glycosylation (Kurischko et al., 2008). Moreover, the polarized secretion machinery is critical for the asymmetric localization of several mRNAs (Aronov et al., 2007). Thus, Cbk1 may regulate polarized growth and cell wall biosynthesis via the integrated coordination of multiple processes, from daughter cell–specific transcription to polarized mRNA localization and secretion.

Recent data suggest that the role of Cbk1 in mRNA metabolism extends beyond Ssd1. Notably, Cbk1 and Ssd1 are genetically linked to other mRNA-associated proteins, such as the PUF family RNA-binding protein Mpt5 and the spliceosome-associated Brr1 (Kaeberlein and Guarente, 2002; Bourens et al., 2009). MPT5 and BRR1 were identified as suppressors of the mating defect of cbk1 mutants and MPT5 deletion rescues the lethality of cbk1Δ mutations (Bourens et al., 2009). ssd1 mpt5 double mutants also display enhanced phenotypes, suggesting that they function in parallel mRNA processes (Kaeberlein and Guarente, 2002). Intriguingly, Mpt5 contains multiple consensus sites for Cbk1 phosphorylation (Mazanka et al., 2008; Bourens et al., 2009), suggesting that Cbk1 directly regulates cell growth and development via multiple mRNA-associated processes.

Collectively, our data suggest unanticipated functions for LATS/NDR kinases in regulating cell growth and cancer development via post-transcriptional mRNA-dependent mechanisms. Our experiments also provide a link between LATS/NDR kinases and stress signaling. Because alterations in stress responses are a major hallmark of transformation, it is possible that an important tumor suppressor function of LATS/NDR kinases is to repress cell growth during cellular stress, perhaps via similar mechanisms as Cbk1. Finally, data regarding Ssd1-dependent mechanisms for asymmetric mRNA localization may also reveal a possible mechanism for LATS/NDR kinase in mediating polarized morphogenesis. For example, mutations in one of the Drosophila and Caenorhabditis elegans Cbk1 orthologues cause cellular morphogenesis defects that are particularly evident in neurons (Hergovich et al., 2006; Jan and Jan, 2010). Because neuronal morphogenesis requires the polarized delivery of mRNAs to growth cones and dendritic spines, it is tempting to speculate that the neuronal morphogenesis defects of LATS/NDR mutants could be the direct consequence of defective mRNA localization mechanisms. Further work on this important class of kinases and their substrates will clarify their specific mechanisms in controlling cell morphogenesis and proliferation.

Standard yeast genetics and culture methods were used as described previously (Guthrie and Fink, 1991; Kurischko et al., 2005). The strains used in this paper are listed in Table I. The Ssd1-GFP strains were constructed by integration of PCR-based cassettes, as described previously (Longtine et al., 1998). All TAP-tagged yeast strains were obtained from Invitrogen.

SRL1 mRNA in vivo imaging was performed as described previously (Haim et al., 2007). In brief, at the 3′ UTR of SRL1 the cassette containing 12 MS2-CP binding sites, together with the S. pombe his5+ gene as a selective marker, was integrated. The selective marker was deleted and the strain was transformed with the MS2-CP-GFP–containing plasmid to visualize the tagged mRNA. Plasmids for mRNA tagging were provided by Dr. Jeffrey Gerst (Weizmann Institute of Science, Rehovot, Israel).

Where designated, cells were exposed to different stresses as described previously (Buchan et al., 2008; Nissan and Parker, 2008). For glucose depletion, the cells were spun down, washed with glucose-free medium, resuspended in the same medium, and incubated for 10–15 min. For hypertonic stress induction, logarithmically growing cells were pelleted, resuspended in medium containing 1 M NaCl, and incubated for 5–10 min.

The plasmids and oligonucleotides used in this paper are listed in Table II and Table III. SSD1 was subcloned from YEp13-SSD1 (a gift from C. Boone, University of Toronto, Ontario, Canada) into pUC19 for further manipulations. SSD1-9A and SSD1-9D mutations were synthesized by Geneart AG and correspond to the following amino acids substitutions: S to A at positions 42, 104, 126, 152, 164, 228, 295, and 319, and T to A at position 261 (SSD1-9A); S to D at positions 40, 42, 122, 126, 160, 164, 228, 259, and 319 (SSD1-9D). SSD1-9A and SSD1-9D plasmids were constructed by subcloning the N-terminal HpaI–SpeI fragment of mutated SSD1 into pUC19-SSD1. From there, the entire SSD1 (promoter region and ORF with 3′ UTR) was cloned into pRS316 and pRS415 to create FLE1079, FLE1080, FLE1081, FLE1083, and FLE1087 plasmids. Wild-type and mutant SSD1 were subcloned into Gateway-compatible GFP, dsRED (RFP), or HA plasmids, as described previously (Alberti et al., 2007). All Gateway-compatible vectors were provided by Dr. Aaron Gitler (University of Pennsylvania, Philadelphia, PA). The wild-type SSD1 ORF without the Stop codon was amplified by PCR using the corresponding oligos and cloned into the pDONR221 vector. The corresponding SSD1-9A and -9D plasmids were constructed by restriction enzyme–mediated cloning. Ssd1-RBDΔ and Ssd1-9A-RBDΔ plasmids were constructed by deleting the internal XbaI fragment of SSD1. UTH1 was isolated from the plasmid YGPM12a13 (Thermo Fisher Scientific) by excising the NaeI–BglII fragment and subcloning it into the NaeI–BamHI sites of pGP564, creating FLE1272.

Plasmids encoding RFP-tagged P-body and stress granule proteins were obtained from Dr. Roy Parker (Howard Hughes Medical Institute, University of Arizona, Tucson, AZ) and Dr. Charles Cole (Dartmouth Medical School, Hanover, NH).

Co-immunoprecipitation and immunoblot analyses were conducted as described previously (Kurischko et al., 2005, 2008) using monoclonal anti-HA antibody (Roche), monoclonal anti-GFP antibody (Roche), or polyclonal anti-TAP antibody (Thermo Fisher Scientific). For coIP, cells were resuspended in lysis buffer (150 mM NaCl, 50 mM Hepes, 1% NP-40, 60 mM 2-glycerophosphate, 1 mM EDTA, 2 mM DTT, 10 mM NaF, and protease inhibitors) and lysed in a Mini-Beadbeater (Biospec Products). 2 mg of protein extract was precleared with protein G–Sepharose (Invitrogen), incubated with 2 μg anti-GFP antibody (Roche), and precipitated with ∼50 μl protein G–Sepharose for 1 h at 4°C. The protein-bound beads were washed with lysis buffer and resuspended in 50 μl SDS sample buffer and loaded on 7.5% SDS-PAGE (15 μl per lane). Immunoblots were probed with primary antibodies and secondary AP-conjugated anti–mouse and anti–rabbit antibodies (Promega) and processed for ECF, as per manufacturer’s protocol (Invitrogen). Immunoblots were digitized and, where designated, quantified with a STORM PhosphorImager (GE Healthcare). For whole-cell immunoblots, cells were TCA precipitated and processed for immunoblots, as described previously (Kurischko et al., 2008).

RT-PCR experiments with immunoprecipitated complexes were performed as described previously (Peritz et al., 2006). We used the MasterPure kit (Epicentre Biotechnologies) for yeast RNA purification. The cDNA was synthesized with the Transcriptor First Strand cDNA Synthesis kit (Roche). The oligos used to amplify fragments of SUN4, CTS1, DSE2, SCW4, SRL1, CCW12, and SIM1 are listed in Table III.

qRT-PCR was performed using the Lightcycler 1.5 system (Roche). The initial cDNA denaturation was for 10 min at 95°C and was followed by 50 cycles: denaturation at 95°C for 1 s; annealing at 55°C for 5 s; elongation at 72°C for 8 s. The oligos for ACT1, ADH1, CCW12, SIM1, and SRL1 are listed in Table III.

Routine fluorescence microscopy was performed with a fluorescence microscope (model DMR5; Leica) equipped with a 100x Plan Apochromat 1.46 NA oil objective and a 16-bit cooled EMCCD camera (ImagEM; Hamamatsu Photonics), as described previously (Kurischko et al., 2008). Most colocalization experiments were conducted on a spinning disk confocal system controlled by MetaMorph software (MDS Analytical Technologies). The spinning disk microscope was an inverted microscope (model DMI4000; Leica) equipped with a 100x HCX Plan Apochromat 1.46 NA oil objective, a spinning disk confocal system (CSU-10; Yokogawa) and a 16-bit cooled EMCCD camera (ImagEM; Hamamatsu Photonics). Laser excitation was provided by a 488-nm (Spectra Physics) and a 561-nm laser (Cobolt Jive) controlled through the LMM5 module (Spectral Applied Research). The emissions were collected at 503–552 nm for GFP and 583–650 nm for RFP. Z stacks were taken for a total thickness of 1.8–3.4 µm at a step size of 0.2 µm. Image capture and analysis was controlled via Volocity (PerkinElmer) or MetaMorph software.

Fig. S1 shows that physiologically expressed Ssd1 colocalizes with P-bodies in glucose-depleted cells. Fig. S2 shows that UTH1 is a dosage suppressor of cbk1-8 mutants. Fig. S3 shows that Ssd1 precipitates SRL1, SIM1, CCW12, SUN4, DSE2, SCW4, and CTS1 mRNAs and does not affect steady-state mRNA levels. Fig. S4 shows that Ssd1-9A but not Ssd1-9A-RBDΔ expression causes SRL1 mRNA to localize to P-bodies. Fig. S5 shows confirmation of SRL1 mRNA-tag expression in wild-type and ssd1Δ cells.

The authors thank Dr. Lingli Zhang (The School of Veterinary Medicine’s Core Imaging Facility at University of Pennsylvania) for help with confocal microscopy and Amy Praestgaard (University of Pennsylvania Center for Clinical Epidemiology and Biostatistics) for assistance with the statistical analysis of mRNA localization patterns. We are grateful to Drs. Aaron Gitler, Jeffrey Gerst, Charles Cole, and Roy Parker for providing various plasmids and reagents. We thank Dr. Christopher Herbert (CNRS, Gif-sur-Yvette, France), Dr. Malcolm Whiteway (National Research Council of Canada), Dr. Beverely Errede (University of North Carolina), Dr. Erfei Bi (University of Pennsylvania), Dr. Phong Tran (University of Pennsylvania), and Dr. Aaron Gitler (University of Pennsylvania) for many helpful discussions and Drs. Michael Atchison and Xuan Pan (University of Pennsylvania) for assistance with qPCR.

This work was supported by American Cancer Society grant RSG0508401 and National Institutes of Health grant R01-GM-060575 (to F.C. Luca).