Sequestration of the PKC ortholog Pck2 in stress granules as a feedback mechanism of MAPK signaling in fission yeast

The protein kinase C (PKC) family is a member of the AGC kinase superfamily of serine/threonine protein kinases that play essential roles in cell growth and cell death (Newton, 2010) by transmitting extracellular signals to various downstream signaling pathways, including MAPK. In mammalian systems, the PKC–Raf–MEK–ERK signaling pathway modulates gene expression and cell proliferation in various cell types, including endothelial cells, T-cells and various cancer cell settings (Pintus et al., 2003; Hoyer et al., 2005; Li et al., 2012). Despite the importance of PKC signaling deterioration in various diseases, including cancers and neurodegenerative disorders such as Alzheimer's disease, how PKC regulates these pathophysiological roles remains obscure (Lucke-Wold et al., 2015; Callender and Newton, 2017; Isakov, 2018).

A similar PKC-dependent regulation of the MAPK signal transduction pathway has been described in both budding and fission yeasts. The yeast Saccharomyces cerevisiae produces only a single Pkc1, whereas the fission yeast Schizosaccharomyces pombe has two PKC homologs, Pck1 and Pck2 (Arellano et al., 1999; Fuchs and Mylonakis, 2009). In fission yeast, Pck2 is a primary upstream activator of the cell integrity MAPK signaling pathway, composed of a three-tiered Mkh1–Pek1–Pmk1 kinase cascade (Toda et al., 1996; Ma et al., 2006). Rho small GTPases, as well as PDK1 homologs, participate in the activation of PKC in both yeasts by mechanisms involving protein–protein interaction via the HR1 Rho-binding domains (Sayers et al., 2000) and phosphorylation at the activation loop (Madrid et al., 2015), respectively. Thus, conservation between the fission yeast Pcks and the mammalian PKN1/2 [also known as protein kinase C-related protein kinase (PRK)1/2] (Mukai, 2003) may exist at structural and regulatory levels (Sayers, et al., 2000).

Importantly, the biological activity of PKC enzymes is tightly regulated by their subcellular localization (Newton and Jhonson, 1998; Gould and Newton, 2008), as exemplified by PKC translocation to the plasma membrane in response to various stimuli. Recently, a research group demonstrated that PKC alpha (also known as PRKCA) in mammals localizes to stress granules (SGs) upon heat stress (HS) (Kobayashi et al., 2012), but the function and mechanism of PKC SG localization remain unclear. SGs are non-membranous cytoplasmic foci, composed of non-translating messenger ribonucleoproteins that rapidly aggregate in cells exposed to a variety of environmental stresses, including oxidative, hyperosmotic or heat stresses (Guil et al., 2006; Anderson and Kedersha, 2006, 2009). Representative components of SGs include poly(A)-binding protein (PABP; also known as PABPC1), translation factors such as eIF3 (also known as EIF3A), eIF4E and eIF4G (also known as EIF4G1), as well as specific RNA-binding proteins (RBPs), such as TIA-1 and TIAR (also known as TIAL1) (Panas et al., 2016). Notably, many signaling molecules, such as mTOR, RACK1 and calcineurin, are recruited to SGs, thereby constituting signaling hubs that can rewire signal transduction (Thedieck et al., 2013; Wippich et al., 2013; Arimoto et al., 2008; Higa et al., 2015). Despite recent progress in understanding the repertoires of signaling molecules contained in SGs, little is known about the mechanism or the physiological significance of PKC SG recruitment.

We have previously established molecular genetic screens for negative regulators of PKC-mediated Pmk1 MAPK signaling in fission yeast (Sugiura et al., 1998; Ma et al., 2006). The rationale for this screen relies on the finding that hyperactivation of Pmk1 MAPK signaling is toxic under some conditions, including Pck2 overexpression. We then isolated multi-copy suppressors of the MAPK hyperactivation-dependent cytotoxicity induced by Pck2 overexpression. We identified the dual-specificity phosphatase Pmp1; PP2C phosphatases Ptc1 and Ptc3, which inactivate Pmk1 MAPK; the SH3-adaptor protein Skb5, which spatially regulates Mkh1 MAPKKK (Sugiura et al., 1998; Takada et al., 2007; Kanda et al., 2016); and Ded1. Ded1 is similar to mammalian DDX3 (also known as DDX3X) and belongs to the highly conserved DEAD-box ATPases that are representative components of SGs (Grallert et al., 2000; Hilliker et al., 2011). Given the isolation of the SG-resident protein Ded1 as a negative regulator of PKC/MAPK signaling, together with the finding that PKC in mammals translocates to SGs, we hypothesized that SGs might be functionally relevant to repress Pck2-mediated MAPK signaling activation.

In this study, we explored the role of SGs in PKC/MAPK signaling activation in fission yeast. Heat stress at 45°C [high-heat stress (HHS)] induced both MAPK hyperactivation and a dynamic change in Pck2 intracellular localization from the plasma membrane, particularly enriched at the growing cell tips (hereafter referred to as ‘cell tips’), to the SGs. We show that Pck2 needs its kinase activity for its dispersal from the cell tips and subsequent recruitment into SGs. Notably, Pmk1 MAPK activity is a key determinant of Pck2 translocation into SGs, as MAPK signaling disruption limits Pck2 translocation into SGs. Consequently, Pck2, when localized at SGs, failed to reactivate MAPK signaling after HHS recovery. Intriguingly, HS at 45°C (i.e. HHS), but not at 43°C, preferentially partitioned Pck2 into SGs, thereby reducing Pck2 kinase activity in the soluble fraction. Taken together, these findings indicate a role for SGs as signaling hubs to intercept PKC/MAPK signaling activation upon HHS by spatial PKC sequestration. Failure to coordinate this ‘feedback circuit’ might lead to vulnerability to various PKC-regulated diseases, including cancer.

As a first step to investigate the functional connection between SGs and PKC-mediated MAPK signaling, we analyzed the effect of HS, a physiological SG-inducing stimulus, on the subcellular localization of Pck2. For this purpose, the endogenous Pck2 protein tagged with GFP (Pck2-GFP) was visualized, and the effect of HS was analyzed. Under the unstressed condition, fluorescence of the endogenous Pck2-GFP was mainly localized to the cell tips, as previously reported (Fig. 1A, 0 min, arrowheads) (Kanda et al., 2016). Upon HS (45°C), Pck2-GFP underwent dynamic changes in its intracellular localization from the cell tips to the dot-like structures (Fig. 1A, 20 min, 60 min, arrows). Quantification of the number of cells harboring the cell tip-localized Pck2 showed that HS induced Pck2 dispersal from the cell tips within 5 min after HS exposure (Fig. 1B, top, 5 min). By contrast, the number of Pck2 dots increased during HS (Fig. 1B, bottom). In mammalian cells and budding and fission yeasts, Pabp is a representative SG marker, and is routinely used to study SG dynamics in live cells (Kedersha and Anderson, 2007; Nilsson and Sunnerhagen, 2011; Brambilla et al., 2017). We, therefore, co-expressed Pabp tagged with tdTomato expressed from the endogenous loci to examine its colocalization with Pck2. The results showed that the Pck2-GFP dots colocalized with the Pabp-tdTomato dots upon HS (Fig. 1A), suggesting that Pck2 translocated into SGs upon HS. The number of Pabp dots also increased during HS, showing that Pck2 and Pabp displayed similar kinetics in terms of their translocation to dot-like structures upon HS. Quantification showed that approximately half of the Pck2 dots colocalized with Pabp dots (Fig. 1B, bottom). We also observed the colocalization of fluorescence of Pck2-GFP and tdTomato-fused Nrd1, which also serves as a representative SG-resident protein (Satoh et al., 2012), co-expressed from their endogenous loci upon HS, and, similar to Pabp, approximately half of the Pck2 dots colocalized with Nrd1 dots, further supporting the idea that Pck2 was recruited to SGs upon HS (Fig. 1C,D).

Cycloheximide (CHX) is known to be a potent inhibitor of SG formation (Kedersha et al., 2000; Mazroui et al., 2002; Mollet et al., 2008). We, therefore, used CHX to prevent SG formation and examined the impact of the inhibition of SG formation on Pck2-GFP localization (Fig. 1E). CHX treatment markedly inhibited the formation of SGs as judged by the decreased number of Pabp dots (Fig. 1F, bottom). The formation of Pck2-GFP dots under HS was also significantly inhibited by CHX treatment (Fig. 1F, top), suggesting that the formation of Pck2-GFP dots upon HS is dependent on SG formation.

What is the regulatory mechanism to stimulate Pck2 SG translocation? To investigate how Pck2 is recruited into SGs, we tested whether the kinase activity of Pck2 was required for its localization upon HS. For this, we created cells expressing the genomic version of the kinase-negative mutant Pck2 protein tagged with GFP (Pck2 K712W-GFP) under its native promoter (Fig. 2A). To gain a precise understanding of the impact of Pck2 kinase activity on the subcellular distribution of Pck2, we also created cells expressing the genomic version of the constitutively active mutant allele of Pck2, encoding Ala instead of Arg389 (Pck2 R389A) at the pseudo-substrate domain (Fig. 2A) (Watanabe et al., 1994).

To directly assess the in vitro kinase activity of these mutant Pck2 proteins compared with the wild-type (WT) Pck2, we used the PepTag system (Sukumaran and Prasadarao, 2002; Shao and Bayraktutan, 2013). The phosphorylated peptide (phosphorylated bands) migrates towards the positive electrode, whereas the non-phosphorylated peptide (non-phosphorylated bands) migrates towards the negative electrode. We first validated whether fission yeast Pck2 kinase activity can be measured using the PepTag system, and GST-Pck2, but not GST protein, expressed and purified from fission yeast was shown to induce the mobility shift of the phosphorylated peptide, which also appeared with the addition of mammalian PKC as a positive control (Fig. S1).

We then compared the kinase activity of the cells expressing WT Pck2 (Pck2 WT-GFP), kinase-negative Pck2 (Pck2 K712W-GFP) or constitutively active Pck2 (Pck2 R389A-GFP) from the endogenous promoter. These Pck2-GFP proteins were expressed at similar levels, as shown by the analysis of whole-cell lysates blotted with anti-GFP antibodies (Fig. 2B,C). The kinase-negative Pck2 protein immunoprecipitated from the lysates of the mutant strain exhibited markedly reduced kinase activity against substrates compared with Pck2 WT (Fig. 2B). By contrast, the kinase activity of the constitutively active Pck2 R389A was significantly higher than that of the Pck2 WT kinase (Fig. 2C).

We then visualized the subcellular localization of the cells harboring the genomic version of Pck2 K712W-GFP and Pck2 R389A-GFP upon HS. Pck2 K712W-GFP localized at the cell tips in unstressed cells (Fig. 2D, 0 min). Notably, upon HS, the number of dots expressing the Pck2 K712W-GFP mutant protein was markedly lower (∼30% of that of Pck2 WT-GFP), and a significantly higher number (∼40%) of cells expressing Pck2 K712W maintained cell tip localization even after 60 min (Fig. 2D,E). Thus, Pck2 kinase activity is required for the dispersal of Pck2 from the cell tips and subsequent SG localization.

To evaluate the constitutively active nature of Pck2 R389A, we incubated the cells at a lower temperature of 44°C, because HS at 45°C (HHS) did not discriminate between Pck2 WT-GFP and Pck2 R389A-GFP in terms of their subcellular distribution at the SGs (Fig. S2). Upon HS at 44°C, the numbers of Pck2 dots and Pabp dots were ∼50% (three to six/cell) of those at 45°C (ten to 12/cell) (Fig. 2F; Fig. S3B). The number of Pck2 R389A-GFP dots upon HS (44°C) was almost 2.5-fold (eight/cell) of that of Pck2 WT-GFP (three/cell) upon HS for 60 min (Fig. 2F,G). Notably, cells harboring the Pck2 WT-GFP recovered their cell tip localization 60 min after the HS (44°C), whereas Pck2 R389A dispersed from the cell tips but failed to relocalize to the cell tips; instead, it largely translocated to the SGs (Fig. 2F,G, 60 min). Thus, in contrast to the kinase-negative Pck2, Pck2 R389A displayed higher SG localization and sustained dispersal from the cell tips, further supporting the possibility that Pck2 translocates into SGs in a kinase-dependent manner.

It should be noted that the cells expressing Pck2 K712W-GFP or Pck2 R389A-GFP were able to form SGs upon HS as efficiently as the cells harboring Pck2 WT, as judged by the number of Pabp dots during HS, indicating that Pck2 kinase activity does not affect the SG assembly process (Fig. S3A,B).

To investigate whether the active Pck2 preferentially locates to SGs, we undertook experiments to visualize both active and inactive Pck2 fused to a distinct fluorescent tag. For this, we visualized the endogenous kinase-dead (inactive) Pck2 K712W-GFP and investigated the effect of exogenous expression of the constitutively active version of Pck2 R389A fused to mCherry. In addition, the endogenously expressed active Pck2 R389A-GFP was visualized and the effect of exogenous expression of inactive mCherry-Pck2 K712W was investigated. The results showed that inactive Pck2 K712W-GFP was observed to persist in the plasma membrane and translocated to SGs to a lesser extent compared with the exogenously expressed mCherry-Pck2 R389A (Fig. S4A). By contrast, the endogenously expressed active Pck2 R389A potently translocated to SGs, whereas exogenously expressed inactive Pck2 barely translocated to SGs (Fig. S4B). The above data support the possibility that active Pck2 preferentially localizes to SGs. However, as these experiments were performed with exogenous expression of the active or inactive mutant Pck2, whether this reflects the physiological settings needs further study.

It has been reported that HS activates various signaling pathways, including Pmk1 MAPK. To investigate the relevance of the HS-induced Pck2 subcellular distributions to PKC-mediated signaling, we assessed the downstream Pmk1 MAPK activation during HS by evaluating the phosphorylation levels of Pmk1 tagged with GST under the endogenous promoter. HS elicited acute MAPK activation within 5 min, and a gradual increase was observed upon 20 min and 60 min HS (Fig. 3A). In pck2 deletion cells, the basal Pmk1 activity (0 min) was markedly low, and the acute MAPK activation 5 min after HS was almost abolished. Notably, the phosphorylation levels of Pmk1 in pck2 deletion cells were almost comparable with those in WT cells upon HS for 20 min and 60 min (Fig. 3A), indicating that Pck2 is not required for Pmk1 activation at a later stage of HS responses. We then evaluated the functional involvement of Pck1 in Pmk1 activation at a later stage of HS. The results showed that Pck1 deletion increased Pmk1 phosphorylation at an early stage, and no significant change in Pmk1 activation was observed at a later stage, of HS compared with the WT cells, thus indicating that Pck1 is not involved in Pmk1 activation at the late stage of HS (Fig. 3B). Although Pck1 and Pck2 share an essential function in cell viability (Toda et al., 1993), regarding Pmk1 activation upon HS, the functional redundancy was not detected.

The HS-induced kinetics of MAPK activation and Pck2 subcellular localization suggest that Pck2 dissociates from the cell tips within 5 min upon HS, which might facilitate the acute transmission of Pck2 signaling to downstream MAPK activation. Thus, this accounts for the dependence of Pmk1 activation on Pck2 at an early stage of HS (Fig. 3A, 5 min after HS). Consistently, the kinase-negative Pck2 K712W mutation abolished MAPK activity 5 min after HS. By contrast, at a later stage of HS responses, the Pck2 mutant cells exhibited a similar MAPK activation pattern to that of the pck2 deletion cells (Fig. 3C). Also, the cells expressing the Pck2 R389A mutant displayed higher phosphorylation levels of Pmk1 MAPK with or without HS (Fig. 3D), consistent with the constitutively active nature of Pck2 R389A kinase in vivo.

The vic phenotype (exhibiting growth in the presence of MgCl₂ plus the immunosuppressant FK506) is established as an indication of the downregulation of Pmk1 MAPK signaling based on the counteractive interaction between calcineurin and Pmk1 MAPK (Sio et al., 2005). Thus, this phenotype has been used as a highly sensitive biological readout for Pmk1 signaling activity in vivo (Ma et al., 2006; Doi et al., 2015; Sánchez-Mir et al., 2014). The Pck2 K712W mutant cells, similarly to pmk1 deletion cells, exhibited the vic phenotype as indicated by their growth in medium containing 0.12 M MgCl₂ and FK506, wherein the WT cells failed to grow (Ma et al., 2006) (Fig. 3E). By contrast, cells harboring the constitutively active Pck2 R389A were unable to grow in the presence of 0.06 M MgCl₂ and FK506, whereas the WT cells grew well, showing the vic-negative phenotype, which is an in vivo indication of MAPK activation (Fig. 3E). Thus, mutations that affect Pck2 kinase activity in vitro influence downstream Pmk1 signaling in vivo by transmitting the acute responses upon HS.

The data presented above provide evidence that Pck2 kinase activity has a stimulatory role for both Pmk1 MAPK signaling activation and Pck2 translocation into SGs upon HS. We then wanted to delineate the importance of Pck2 kinase activity or Pmk1 signaling activation in Pck2 SG translocation. To test whether Pmk1 MAPK activity plays a role in HS-induced Pck2 distribution, we expressed and imaged WT Pck2-GFP in pmk1 deletion cells treated with HS (Fig. 4A,B). In unstressed cells, Pck2 localization in pmk1 deletion cells was indistinguishable from that in WT cells, and Pck2 dispersal from the cell tips 5 min after HS was not substantially inhibited in pmk1 deletion cells (Fig. 4A,B), indicating that Pmk1 deletion does not affect Pck2 dissociation from the cell tips. Strikingly, in pmk1 deletion cells, Pck2 relocalized at the cell tips within 60 min upon HS, in contrast to the cytoplasmic (5 min) or the SG (20–60 min) Pck2 localization profile in WT cells (Fig. 4A). Conversely, Pck2 SG translocation was impaired in pmk1 deletion cells (∼50% of that in WT cells) (Fig. 4A,B). Pmk1 deletion does not affect SG assembly, as evidenced by the number of the Pabp dots (Fig. S3C) and Pck2 kinase activity (Fig. 4C), thus excluding the possibility that the observed alterations of Pck2 distribution can be induced indirectly via Pck2 or SGs. Thus, Pmk1 MAPK is required for Pck2 SG localization, but not for Pck2 dispersal from the cell tips, and Pmk1 signaling inhibition induces Pck2 relocalization to the cell tips.

Notably, the constitutively active Pck2 R389A-GFP, which exhibited a higher kinase activity and an enhanced SG localization ability in WT cells (Fig. 2C,F), exhibited markedly decreased localization at SGs in the Pmk1 deletion background compared with that of Pck2 R389A-GFP in WT cells (Fig. 4D,E, bottom). In pmk1 deletion cells, the number of Pck2 WT dots and that of the constitutively active Pck2 R389A-GFP dots were almost the same during HS (Fig. 4B,E, bottom; Table S3). Thus, Pmk1 deletion impaired the SG localization of active Pck2 R389A. By contrast, Pck2 R389A-GFP showed increased cell tip localization in pmk1 deletion cells compared with WT cells (Fig. 4B,E, top). These results strongly suggest that Pck2 kinase activity is required, but not sufficient, for the translocation of Pck2 to SGs, and that downstream MAPK activation is crucial for determining Pck2 translocation into SGs or relocalization at the cell tips. These data indicating the existence of a delicate balance of Pck2 localization between SGs and the cell tips suggest that Pmk1 signaling inhibition may induce Pck2 cell tip relocalization by reducing SG localization.

To investigate the effect of distinct HS conditions on SG formation, we treated cells with modest HS (43°C) and severe HS (45°C; referred to as HHS). Notably, HS at 43°C barely induced Pck2 dots, in contrast to HS at 45°C, which induced many Pck2 dots (∼12/cell) colocalizing with Pabp (Fig. 5A–C). Importantly, HS at 43°C induced a small but significant number of Pabp dots, which appear to lack visible Pck2 dots (Fig. 5A–C). These data strongly suggested that only at 45°C (HHS), not modest HS at 43°C [low-heat stress (LHS)], led to accumulation of Pck2.

Regarding the Pck2 cell tip localization, LHS also induced Pck2 dispersal from the cell tips within 5 min, similar to HHS. However, Pck2 fully relocalized to the cell tips when cells were cultured at 43°C for 60 min, whereas Pck2 exposed to HHS remained in the SGs (Fig. 5A,B), indicating that Pck2 is more rapidly relocalized to the cell tips during recovery after HS at 43°C. Thus, SGs induced by HHS, but not by HS at 43°C, sequester Pck2, thereby impairing Pck2 localization at the cell tips, indicating the differential effect of HS (43°C and 45°C) on SG assembly and Pck2 localization.

To investigate the impact of the distinct HS conditions on the subcellular distribution and Pck2 kinase activity, we performed fractionation experiments and partitioned the cells into a ‘Soluble fraction (Sup.)’ and a ‘Condensed fraction (Pellet)’, following the scheme shown in Fig. S6A (using methods modified from Nilsson and Sunnerhagen, 2011; Wheelera et al., 2017). As shown in Fig. S6B and C, upon HHS for 20 min, Pck2, but not tubulin, partly partitioned into a Pabp-positive SG-containing fraction (Pellet) as suggested by the disappearance/decrease of Pck2/Pabp sedimentation from the pellet fraction by pre-treating the cells with CHX.

We then investigated the effect of distinct HS conditions on the subcellular distribution of Pck2 protein. The amount of Pck2-GFP immunoprecipitated from the soluble fraction by anti-GFP antibodies showed that HHS markedly reduced the abundance of Pck2 protein present in the soluble fraction (Fig. 5D). Quantification showed that, upon HHS for 60 min, the soluble fraction contained less than 5% of the Pck2 protein compared with that before HS (0 min) (Fig. 5F). By contrast, cells treated with HS at 43°C maintained more than 60% of Pck2 protein in the soluble fraction (Fig. 5F). In contrast to the soluble fraction, the pellet fraction treated with HHS accumulated a greater amount of Pck2 protein than that treated with HS (43°C) (Fig. 5F). Similarly to Pck2, HHS strongly induced Pabp partitioning from the soluble fraction into the pellet fraction, although HS (43°C) also induced significant partitioning of Pabp into the pellet fraction (Fig. 5G). These results are consistent with the observation with fluorescent microscopy that SGs induced by HHS specifically accumulated Pck2-GFP.

We then compared the in vitro Pck2 kinase activity immunopurified from the soluble fraction (Sup.) of the cells exposed to HS at 43°C and 45°C (Fig. 5D,E). The net Pck2 activity contained in the soluble fraction was markedly decreased when cells were treated with HHS, to ∼20% of that without HS (0 min) (Fig. 5E, red). By contrast, cells treated with HS (43°C) for 60 min almost entirely retained their Pck2 kinase activity compared with that in unstressed cells (Fig. 5E, black). These results suggest that SGs induced by HHS partitioned Pck2 into a Pabp-positive pellet fraction, thereby reducing the active Pck2 protein in the soluble fraction.

To investigate the impact of SGs on Pck2 localization upon HS, we expressed Pck2-GFP in several mutant backgrounds known to be defective in SG formation in S. pombe. These include nrd1 deletion cells, which are reported to display impaired SG formation upon HS (Satoh et al., 2012), and the phosphorylation mutant of the translation initiation factor eIF2α (eIF2α S52A) (Nilsson and Sunnerhagen, 2011), as in mammalian cells, the key event leading to the formation of SGs is the stress-induced phosphorylation of the translation initiation factor eIF2α. In these mutant backgrounds, the endogenous Pabp was visualized, and the effects of moderate HS (43°C) and HHS (45°C) on the localization of Pabp were examined. In nrd1 deletion cells, the number of Pabp-GFP dots was significantly decreased upon moderate HS (43°C) for 20 min and 60 min compared with that of the WT cells (Fig. 6A). However, the number of Pabp dots upon HHS (45°C) was not significantly affected by nrd1 deletion (Fig. 6B). Moreover, Pck2-GFP dots were similarly observed in both nrd1 deletion and WT cells during HHS (45°C) (Fig. 6C).

The number of Pabp-tdTomato dots was significantly decreased in the eIF2α S52A mutant cells upon moderate HS (43°C) (Fig. S5A), but not upon HHS conditions (45°C) (Fig. S5B). In addition, dynamic translocation of Pck2-GFP from the cell tips to the SGs upon HHS (45°C) was similarly observed in the eIF2α S52A mutant cells and WT cells (Fig. S5C). These results indicate that Pck2 translocation to SGs upon HHS was independent of Nrd1 or the phosphorylation of eIF2α, and suggest the distinct regulatory mechanism and/or structural components of SGs achieved in response to HHS.

A recent study reported the importance of SGs on TORC1 signaling reactivation during recovery from HS (Takahara and Maeda, 2012). To investigate the significance of Pck2 sequestration into SGs, we examined whether SGs affect Pmk1 signaling during the recovery process after HS by utilizing the dynamics of SG formation and dissolution to examine whether SGs affect reactivation of Pmk1 signaling. As shown in Fig. 7A, cells were exposed to HS at 45°C (HHS) for each indicated time (0, 5, 20 and 60 min; 1st HS), then incubated at 27°C for 2 h to recover from HS (Recovery 2 h). Then, the cells were again exposed to HS for 5 min to re-stimulate Pmk1 MAPK signaling. As shown in Fig. 7A, when cells were incubated at 27°C for 2 h after the 1st HS (Recovery 2 h), Pck2 was rapidly relocalized to the cell tips, when the exposure to the 1st HS was short, and the number of Pck2 dots (upon 1st HS) was small. Pck2 did not recover its cell tip localization even after a 2 h incubation at 27°C when cells were exposed to HS for more than 20 min. A 5 min exposure to HS (1st HS) did not impair Pck2 relocalization to the cell tips after the 2 h recovery from HS, whereas a 20 min exposure impaired Pck2 cell tip relocalization, and a 60 min exposure to HS retained Pck2 in SGs (Fig. 7A).

The apparent differences in Pck2 localization after recovery from HS suggested a role for SGs in sequestration of Pck2, thereby delaying the Pck2 relocalization at cell tips. To investigate whether the subcellular localization of Pck2 before HS may impact the HS-mediated MAPK signaling activation, the MAPK phosphorylation levels among the cells before (Recovery 2 h) and after the exposure to the second HS for 5 min (2nd HS 5 min) were compared. The second HS induced Pck2 dispersal from the cell tips in cells that exhibited the cell tip Pck2 localization (Fig. 7A, 0 min and 5 min at 1st HS). This Pck2 dispersal from the cell tips resulted in acute Pmk1 activation (up to 4-fold), which is almost equivalent to that achieved by the first HS (Fig. 7B, bottom). By contrast, HS failed to activate Pmk1 MAPK when Pck2 was localized in the cytoplasm or the SGs just before the 2nd HS exposure (Fig. 6B). The above data suggest the importance of Pck2 cell tip localization in reactivation of MAPK signaling, and likewise the role of SGs in the sequestration of Pck2, thereby preventing MAPK reactivation during HS.

The same experiments were again performed in nrd1 deletion cells wherein the SG assembly upon modest HS was defective. However, Pck2 SG translocation and Pmk1 MAPK signaling activation upon first and second HS exposure were similarly observed in nrd1 deletion cells (Fig. 7C,D). These results further indicate the uniqueness of Pck2-positive SGs upon HHS.

In this study, we explored the role of SGs in MAPK signaling regulation and showed that SGs induced by HHS preferentially sequester Pck2/PKC in a MAPK signaling-dependent manner, thereby suggesting a role for SGs in a feedback circuit to intercept PKC/MAPK signaling activation (Fig. 8).

HS is a well-known physiological inducer of SG assembly, as well as a stimulant that activates many signaling pathways, including MAPK. However, it is poorly understood how HS-induced MAPK signaling activation is functionally connected to SG assembly. In this study, we have explored the mechanism of HS-induced Pck2 SG recruitment and its impact on MAPK signaling activation. We found that Pck2 kinase activity is required for Pck2 dispersal from the cell tips and subsequent translocation into SGs, as evidenced by the sustained cell tip localization of the kinase-negative Pck2 and the preferential recruitment of the constitutively active Pck2 mutant version into SGs upon HS. We also found that Pmk1 MAPK activation has a crucial role in Pck2 subcellular distribution, as evidenced by the decreased Pck2 translocation into SGs and increased Pck2 relocalization to the cell tips in Pmk1 deletion. Thus, Pmk1 MAPK activation promotes Pck2 SG translocation and inhibits Pck2 relocalization to the cell tips. Stimulation of the basal MAPK activity by deleting pmp1⁺, encoding a dual-specificity MAPK phosphatase for Pmk1, enhanced Pck2 SG translocation (Fig. S7), further corroborating the importance of MAPK activation in stimulation of Pck2 SG partitioning. Importantly, the constitutively active Pck2 R389A mutant with high kinase activity failed to localize at SGs and instead exhibited a high cell tip localization profile in the Pmk1 deletion background, indicating that MAPK activity dictates whether Pck2 localizes to SGs or the cell tips. Thus, we conclude that Pck2 kinase activity is a prerequisite for Pck2 SG localization and that Pmk1 MAPK signaling activation is a critical determinant for Pck2 SG translocation.

One plausible rationale for Pmk1 activity-dependent Pck2 SG translocation is a role for SGs in a negative feedback mechanism to repress PKC/MAPK hyperactivation induced by HS. Consistently, Pmk1 deletion stimulated Pck2 relocalization to the cell tips, which is essential for stimulating Pck2-dependent MAPK activation (Kanda et al., 2016). We also presented evidence supporting the role of SGs in spatial restraint of MAPK signaling hyperactivation. Pck2 localized at the cell tips can activate Pmk1 MAPK, whereas Pck2 localized at the SGs failed to reactivate Pmk1 in response to HS (Fig. 7), showing the relevance of Pck2 distribution in HS-mediated MAPK signaling (re)activation. Thus, SGs provide a benefit as a means to avoid the detrimental effects of HS by allowing a time delay for the cell to recover the Pck2 cell tip localization by sequestering Pck2 (Fig. 8). Therefore, Pck2 recruitment into SGs resulted in repression of MAPK reactivation during recovery from HS. Various mechanisms of PKC downregulation have been reported, including protein degradation, membrane trafficking and PKC dephosphorylation, by affecting PKC protein levels and/or PKC subcellular localization from/to the plasma membrane (Newton, 2009; Madrid et al., 2017). Downregulation of PKC signaling by reducing the soluble PKC protein/activity by SGs (Fig. 7) could provide a novel way to spatiotemporally control PKC/MAPK hyperactivation, which can induce cell death under certain conditions (Gutiérrez-Venegas et al., 2009).

A similar strategy to repress MAPK signaling activation to sequester MAPK signaling regulator into SGs has been reported for RACK1, which is a signaling scaffold protein that binds to the stress-responsive MAPKKK MTK1 (Arimoto et al., 2008). SGs sequester RACK1 under HS, thereby inhibiting activation of MTK1-mediated SAPK signaling. In budding yeast, TORC1 signaling is also regulated by SGs via transient sequestration of TORC1 into SGs (Takahara and Maeda, 2012). Our study results not only present an additional example of the role of SGs as signaling hubs to sequester PKC to repress MAPK signaling, but also demonstrate the observation that SGs preferentially recruit active Pck2 in a MAPK activation-dependent manner. Based on the findings that SGs preferentially recruit Pck2 only when Pmk1 MAPK signaling is active, and that there seems to exist a threshold (45°C) for Pck2 translocation into SGs, SGs might serve as a safeguard to prevent MAPK signaling hyperactivation by sequestering active Pck2.

Our data indicate that only SGs induced by HS at 45°C (i.e. HHS) can recruit Pck2 and that deletion of nrd1, one of the key components of SGs, did not affect SG formation upon HHS (Figs. 5A and 6B). These data indicate the heterogeneity of SGs stimulated upon distinct stress conditions and highlight the uniqueness of the Pck2-positive SGs induced upon HHS. Currently, it remains unknown how the activation of MAPK by HHS triggers Pck2 SG translocation and the underlying molecular mechanisms that distinguish HHS from LHS. It may be intriguing to speculate that MAPK activation modulates (presumably by phosphorylation) a putative Pck2-binding partner protein, which would facilitate Pck2 recruitment into SGs. Previous reports identified yeast Ataxin2 (Pbp1) as a negative regulator of TORC1 signaling (Yang et al., 2019). Pbp1 was shown to bind TORC1, specifically during respiratory growth, and inhibits TORC1 signaling via a low-complexity region by a phase-separation mechanism (Yang et al., 2019). Our genetic screen identified ded1⁺ as a suppressor of MAPK signaling hyperactivation-induced cell death by Pck2 overexpression. This finding raises an intriguing possibility that Ded1, which is an SG-resident protein and an evolutionarily conserved regulator of phase separation, may suppress MAPK hyperactivation by promoting SG assembly/phase separation. Although a complete analysis and characterization of the molecular function of Ded1 in relevance to SGs and/or phase separation will form the basis of a future report, these results strongly suggest that the formation of Ded1-positive SGs appears to be relevant to repress Pck2-mediated MAPK signaling activation. Further confirmation of this view will rely on the identification of the proteins involved in SG assembly and/or Pck2 binding partners, which stimulate Pck2 SG translocation via physical interaction.

In summary, we have presented data that uncover a role for SGs as a platform to modulate the spatial distribution of Pck2, which contributes to control MAPK hyperactivation, as well as the presence of Pck2-positive granules specifically induced upon HHS. Given the importance of PKC in various physiological/pathological settings in higher organisms, our findings may shed light on the fundamental processes regulated by PKC and its control by SG assembly.

Intriguingly, in mammals, PKC signaling and its deterioration are well appreciated to be closely associated with various diseases, and abnormal aggregation of PKC subtypes was reported to underlie the pathogenesis of neurodegenerative diseases such as spinocerebellar ataxia type 14 (Verbeek et al., 2008). SGs are attracting attention as RNA-centric signaling hubs to recruit many signaling molecules, and SG assembly communicates a ‘state of emergency’ by intercepting and sequestering components of various signaling pathways. We, therefore, assume that the MAPK-dependent PKC recruitment system involving SGs may be a universal survival strategy that cells adapt to combat stress responses.

Schizosaccharomyces pombe strains used in this study are listed in Table S1. The media and genetic, molecular and biological methods were as previously described (Kanda et al., 2016). Pck2 K712W and R389A were generated by inverse PCR using PrimeSTAR Max DNA Polymerase (TaKaRa). The primers used are summarized in Table S2. Proteins were C-terminally tagged with GFP, tdTomato or GST expressed from the respective endogenous loci (Bähler et al., 1998). S. pombe strains are described in Table S1, and we configured their authentication and tested for contamination every time.

Unless otherwise stated, cells were cultivated at 27°C in Edinburgh minimal medium (EMM) (Sabatinos and Forsburg, 2010). Prior to stress treatment, the cells were grown to mid-log phase (optical density at a wavelength of 660 nm=0.5). HS was imposed by transferring the culture tubes to a water bath at 43°C or 45°C for the indicated time. After each stress treatment, the culture medium was chilled in ice water for 5 min. The cells were harvested by brief centrifugation at 4°C.

GST pull-down was carried out as previously described (Kanda et al., 2016).

The scheme for centrifugal separation is shown in Fig. S6A. Protein lysates were prepared in lysis buffer [1% Triton X-100, 30 mM Tris-HCl (pH 8.0), 2 mM EDTA]. Lysates were centrifuged at 800 g for 3 min, and then the supernatants were collected. Subsequently, the supernatants were centrifuged at 815 g or 3300 g for 15 min, and then the supernatants were collected (soluble fraction). The pellets after the second centrifugation were washed in lysis buffer and centrifuged at 815 g or 3300 g for 15 min again. After the removal of the supernatants, the pellets were collected (condensed fraction). GFP-tagged proteins were immunoprecipitated (IP) using anti-GFP antibody (Ma et al., 2006) with Pierce TM Protein A/G Magnetic Beads (Thermo Fisher Scientific, 88803).

Antibodies used are listed in Table S2. Membranes were developed with Chemi-Lumi One Super (Nacalai Tesque, Japan). Protein levels were quantified using relative intensities of all bands that were quantified using MULTI GAUGE Ver. 3.2 software (Fujifilm, Japan).

PKC activity was measured by the PepTag^® assay (Promega, Japan) as previously reported (Sukumaran and Prasadarao, 2002; Shao and Bayraktutan, 2013). In vitro kinase reaction was performed at 30°C for 30 min in a mixture of purified PKC (primarily of α, β and γ isoforms with lesser amounts of δ and ζ isoforms; Walton et al., 1987), GST- or GFP-tagged Pck2 mutants [purified by IP; described in the ‘Centrifugal separation and immunoprecipitation (IP)’ section] and 25 µl of the PepTag^® solution containing 5 µl 5×PKC reaction buffer, 1 µl P-L-S-R-T-L-S-V-A-A-K peptide, 1 µl peptide protection solution and 18 µl ultra-pure water. The reaction was terminated at 95°C for 10 min. Each sample was separated by electrophoresis using a 1.0% agarose gel at 100 V for 20 min. The phosphorylated peptide was quantified using ImageJ software (http://rsb.info.nih.gov/ij/). Protein kinase activity was analyzed for the level of total Pck2 by western blotting and the phosphorylated peptide (Figs. 2B,C and 4C). Given the varying nature of the Pck2 protein levels in the soluble fraction upon heat stress, the net kinase activity of Pck2 from the soluble fraction was analyzed by western blotting for the level of tubulin from the soluble fraction as a loading control and the phosphorylated peptide (Fig. 7D).

Microscopy and miscellaneous methods were carried out as previously described (Kanda et al., 2016).

The quantification of cell tip localization was performed for at least two individual datasets, which analyzed up to 100 cells. Quantification of stress granules foci was performed for at least two individual datasets, which summed up to 50 counted cells.

All results are expressed as mean±s.d. of several independent experiments. Data were analyzed using a one-way ANOVA, followed by Student's t-test (Figs. 1F, 2B,C,E,G, 3A–D, 4B,C,E, 5B,C,E, 6A,B,C, 7D; Figs. S2B, S3A–C, S5A–C, S7B) or one-way ANOVA, followed by a post hoc test using Dunnett's multiple comparison (Fig. 6B). P-values less than 0.05 were regarded as significant.

We thank Dr T. Toda and the Yeast Resource Center (YGRC/NBRP; http://yeast.lab.nig.ac.jp/nig) for providing strains and plasmids; and Professor William Figoni for critical reading of the manuscript. We are grateful to the members of the Laboratory of Molecular Pharmacogenomics for their support.

The peer review history is available online at https://jcs.biologists.org/lookup/doi/10.1242/jcs.250191.reviewer-comments.pdf