The S. pombe CDK5 ortholog Pef1 regulates sexual differentiation through control of the TORC1 pathway and autophagy Free

In the fission yeast Schizosaccharomyces pombe, sexual differentiation is an essential process in order to transmit genetic information to future generations. This process is accompanied by alterations in both morphology and genetic dynamics, which are required to adapt to environmental changes (Davey, 1998; Egel et al., 1990; Engebrecht, 2003; Merlini et al., 2013). Nitrogen starvation leads to alterations in gene expression that lead to conjugation of haploid cells of the opposite mating types or cause cells to enter the G0 quiescent phase (Mata and Bähler, 2006; Marguerat et al., 2012; Sajiki et al., 2009; Shimanuki et al., 2007). In response to nitrogen starvation, the transcription factor Ste11 is activated and translocates from the cytoplasm to the nucleus, facilitating the expression of genes that encode mating pheromones and pheromone receptors, as well as a positive feedback to further transcribe Ste11 (Anandhakumar et al., 2013; Mata and Bähler, 2006; Qin et al., 2003; Sugimoto et al., 1991). Activated Ste11 also upregulates the expression of the mei2 gene, which encodes an RNA-binding protein that is a key regulator of meiosis and mediates the progression of successive sexual differentiation programs including meiotic division and sporulation (Qin et al., 2003; Sugimoto et al., 1991). Therefore, upregulation of Ste11 is a key event in the initiation of sexual differentiation in S. pombe.

Target of rapamycin complex 1 (TORC1) is a strong suppressor of the initiation of sexual differentiation in nutrient-rich conditions through both the inhibition of the nuclear accumulation of Ste11 and the promotion of the degradation of Mei2 by the ubiquitin-proteasome system (Otsubo et al., 2014, 2017; Valbuena and Moreno, 2010). TORC1, which is a kinase complex, is highly conserved between yeast and humans. Misregulation of TORC1 leads to various disorders, including cancer, type-2 diabetes and neurological disorders (Saxton and Sabatini, 2017).

In S. pombe, the TORC1 consists of the catalytic subunit, Tor2 [an ortholog of the mechanistic TOR (mTOR) in mammals], Mip1 (Raptor) and Wat1 (mLst8) (Matsuo et al., 2007; Takahara and Maeda, 2012), and its kinase activity is partially inhibited by the rapamycin–FKBP12 complex (Nakashima et al., 2010, 2012; Takahara and Maeda, 2012). Similar to the regulation of TORC1 in mammals, TORC1 in S. pombe localizes on the surface of vacuole (yeast lysosome), and is activated through interactions with Rhb1 (in mammals, Rheb GTPase), an essential activator of TORC1 in response to nitrogen sources (Murai et al., 2009; Urano et al., 2005, 2007; Uritani et al., 2006; Valbuena et al., 2012). The TSC complex, which is comprised of Tsc1 and Tsc2, negatively regulates TORC1, as it acts as a GTPase-activating protein (GAP) for Rhb1 (Chia et al., 2017; Ma et al., 2013; Matsumoto et al., 2002; van Slegtenhorst et al., 2004). Rag GTPases and Ragulator are essential for anchoring TORC1 to the lysosome and vacuole membrane in mammals and budding yeast, respectively. However, a recent study suggested that in fission yeast, the Rag GTPases Gtr1 and Gtr2 and Ragulator components Lam1–Lam4 are dispensable for the vacuolar localization of TORC1 and attenuate TORC1 activity to maintain the normal growth of cells (Chia et al., 2017). In addition to these TORC1 regulatory systems, recent reports suggest that tRNA precursors regulate TORC1 activity in response to nutrients to prevent the initiation of sexual differentiation (Otsubo et al., 2018). Thus, fission yeast is an excellent model organism because various unique TORC1 regulatory mechanisms have been identified. However, there is little information about the role of TORC1 in the regulation of sexual differentiation.

Under nutrient-rich conditions, TORC1 promotes protein synthesis via direct phosphorylation of Thr-415 of the ribosomal protein S6 kinase Psk1 (Nakashima et al., 2012), which is the equivalent of mammalian p70S6K (also known as RPS6KB1) phosphorylation at Thr-389. TORC1 in S. pombe also suppresses the initiation of autophagy through phosphorylation of Atg13 (Otsubo et al., 2017, 2018), similar to what is found in budding yeast and mammals. Autophagy acts as a bulk degradation system, but also plays an important role in driving the sexual differentiation process, as degraded proteins or organelles can be utilized as nutrient sources for promoting sexual differentiation (Mukaiyama et al., 2010). In addition, recycled nutrients reactivate TORC1 to terminate autophagy under starvation conditions (Shin and Huh, 2011; Yu et al., 2010). Thus, expanding knowledge about the regulatory mechanisms of balance between TORC1 activity and autophagy can provide essential insight into the mechanisms of sexual differentiation in fission yeast.

Cyclin-dependent kinases (CDKs) are Ser/Thr kinases that play a central role in cell cycle progression, and their activity is controlled by cyclins, other kinases and phosphatases (Malumbres and Barbacid, 2005). In S. pombe, pombe pho eighty-five 1 (Pef1) is a member of the CDK5 subfamily, which includes budding yeast PHO85, and mammalian CDK5 and CDK14–CDK18 (Cao et al., 2014). Unlike canonical cell cycle-related CDKs (e.g. CDK1 and CDK2), the CDK5 subfamily members are involved in the regulation of various cellular functions in post-mitotic cells, such as spermatogenesis, cell migration and autophagy (Cao et al., 2014; Dhavan and Tsai, 2001; Matsuda et al., 2017; Mikolcevic et al., 2012; Yang et al., 2010). Although reports suggest that Pef1 controls cell cycle and chronological life span and interacts with three cyclins, Clg1, Pas1 and Psl1 (Chen et al., 2013; Tanaka and Okayama, 2000), the downstream signaling pathway and activation mechanisms of Pef1 remain unclear. To expand the understanding of Pef1 function, we examined the role of pef1 deletion on cellular function. We found that Pef1 is involved in the regulation of mating efficiency through upregulation of the Ste11-Mei2 pathway. pef1 deletion reduced TORC1 activity during vegetative growth conditions but promoted autophagy and TORC1 reactivation during nitrogen starvation. Double deletion of clg1 and pas1 promoted autophagy and TORC1 reactivation as was seen in pef1-deficient cells, indicating that Clg1 and Pas1 are activators of Pef1 during the initiation of sexual differentiation. Our results provide new insight into the Pef1-mediated sexual differentiation machinery.

Although kinases closely related to CDK5 (for instance mammalian CDK14–CDK18 or budding yeast PHO85) are involved in the regulation of various signal transduction pathways in post-mitotic cells (Cao et al., 2014; Dhavan and Tsai, 2001; Matsuda et al., 2017; Mikolcevic et al., 2012; Yang et al., 2010), the biological role and signaling pathways regulated by S. pombe Pef1 remain unclear. To examine the physiological functions of Pef1, we constructed a pef1-deficient strain and introduced LEU2-based pREP273 vectors encoding 3HA-tagged forms of Pef1 mutants into leu1-32 pef1Δ cells (Nakashima et al., 2014). Western blot analysis confirmed that endogenous Pef1 was not present in the pef1Δ strain and that there were similar levels of expression in Pef1-3HA mutants in pef1 mutant-introduced pef1Δ cells (Fig. S1A).

In order to determine mating efficacy, strains were grown on sporulation agar with supplements (SPAS), which contains small amounts of amino acids for efficient mating and sporulation, or Edinburgh minimal medium (EMM) without NH₄Cl (EMM-NH₄Cl) agar, a nitrogen-free medium. As seen in Fig. 1A, there were no differences in the mating efficiency in control cells between SPAS and EMM-NH₄Cl agar. Interestingly, the mating ability of pef1Δ cells was significantly decreased on the EMM-NH₄Cl agar as compared to the mating ability on the SPAS. This low mating efficiency in pef1Δ cells was restored by the expression of wild-type Pef1 (Pef1 WT), but not Pef1 K32R, which is a mutant that lacks kinase activity. Pef1 has two predicted phosphorylation sites, Thr-13 and Tyr-14, which correspond to Thr-14 and Tyr-15 of Cdc2 (Birot, 2016). Thus, we assessed the effect of phosphorylation of Pef1 on mating efficiency. The mating efficiency of these mutant (Pef1 T13A and Y14F)-expressing cells was unaltered as compared with Pef1 WT expressing cells on both the SPAS and EMM-NH₄Cl agar. These results suggest that Pef1 is required for mating processes in the absence of nitrogen sources. Thus, we suspected that nitrogen starvation evokes the phenotypes seen upon pef1 deletion. However, a potential phosphorylation of Pef1 at Thr-13 and Thr-14 does not seem to affect mating efficiency during sexual differentiation.

To determine the cause of low mating efficiency in pef1Δ cells, we examined the Ste11-Mei2 pathway, since it is known to play a critical role in switching sexual differentiation from mitosis. Ste11 is the upstream transcription factor of Mei2, and its expression levels are significantly increased upon nitrogen starvation (Anandhakumar et al., 2013; Mata and Bähler, 2006; Qin et al., 2003; Sugimoto et al., 1991). Quantitative (q)PCR analysis showed that the mRNA levels of ste11 and mei2 were increased after 4 h of starvation in control cells (pef1⁺), whereas the expression of these genes in the pef1Δ strain was silenced at all times in starved cells (Fig. 1B,C). Additionally, western blot analysis also revealed that the expression of Ste11 were increased by nitrogen starvation in a time-dependent manner, but this effect was strongly suppressed by pef1 deletion (Fig. 1D). In addition, the percentage of cells with nuclear accumulation of Ste11–GFP was decreased in pef1Δ cells at 6 h after nitrogen starvation as compared with control cells (Fig. 1E). These data indicate that Pef1 regulates the initiation of sexual differentiation by activating the Ste11-Mei2 pathway.

Next, we constructed a pef1⁺-GFP strain to obtain information about the subcellular localization of Pef1. Before the fluorescence microscopy analysis, the cells were stained with FM4-64 to visualize the vacuole. We observed that Pef1 localized in the cytoplasm and nucleus, and as well as having a punctate expression adjacent to vacuoles (indicated as a white arrowhead) (Fig. 2A; Fig. S1B). These puncta near vacuoles were abolished after treatment with rapamycin (Fig. 2B). In addition, the vacuoles in pef1Δ cells were smaller than those of control cells (Fig. 2C; Fig. S1C). The vacuole is an essential organelle for autophagic degradation of unnecessary cellular components; thus, we hypothesized that pef1 is involved in regulation of autophagy. To examine pre-autophagosomal structure (PAS) formation, we created a pef1Δ GFP-atg8⁺ double mutant, as Atg8 is the ortholog to proteins of the mammalian LC3 family, which are required for autophagosome maturation (Mukaiyama et al., 2010). Consistent with previous findings, PAS formation was unaffected by rapamycin treatment in control cells (Mukaiyama et al., 2010; Takahara and Maeda, 2012), whereas pef1 deletion increased PAS signals in response to rapamycin treatment (Fig. 2D). Western blot analysis confirmed that the amounts of truncated GFP, which is the product of autophagically degraded GFP–Atg8, were increased by rapamycin treatment in pef1Δ cells; however, there were no differences in autophagic activity in control cells (Fig. 3A). Additionally, the autophagic activity in pef1Δ cells during vegetative growth and the early phases of starvation (at 0.5–1 h without NH₄Cl) was slightly higher than that of control cells (Fig. 3B,C). However, the amount of degraded GFP–Atg8 at the late phase of nitrogen starvation was not significantly different between pef1Δ cells and in the control cells (Fig. 3B,C), suggesting that pef1 suppresses autophagy at steady state and the early phase of nitrogen starvation.

To verify that the inactivation of TORC1 is responsible for the increase in autophagic activity in pef1Δ cells, we increased Tor2 expression in pef1Δ cells by using a pAL-KS-tor2⁺ vector. PAS signals were increased by treatment with rapamycin in pef1Δ cells, but these signals were suppressed by expression of Tor2 (Fig. S2A,B). The amount of truncated GFP was also partially decreased by the expression of Tor2 in pef1Δ cells (Fig. S2C), indicating that Pef1 might also regulate autophagy by partially suppressing TORC1 activity. Moreover, dephosphorylation and expression of Atg13 appeared to be increased upon pef1 deletion (Fig. 3A–C). To investigate phosphorylation levels of Atg13, we constructed an Atg13–3HA-expressing strain. As shown in Fig. 3D, Atg13–3HA was detected as sharp bands (labeled 1–3) between 130 kDa and 100 kDa. Although rapamycin stimulation did not detectably lead to a dephosphorylation of Atg13 in pef1⁺ cells, the dephosphorylation of Atg13 was promoted in pef1Δ cells as compared to rapamycin-stimulated pef1⁺ cells, indicating that Pef1 suppresses the phosphorylation of Atg13. However, Tor2 expression was unable to restore the expression of Atg13 (Fig. S2C). This result suggests that Pef1 controls the expression of Atg13 via a TORC1-independent pathway. On the other hand, dephosphorylation of Atg13 was immediately seen upon the addition of ammonium chloride in pef1Δ cells (Fig. S2D), indicating that the ability to uptake ammonium chloride in pef1Δ cells is intact. Collectively, we found that Pef1 is localized near the vacuole, controls vacuole size and suppresses autophagy through TORC1-dependent and -independent pathways. These findings imply that Pef1 regulates the progression of the sexual differentiation through control of autophagy.

We found that rapamycin sensitivity against autophagy is increased in pef1Δ cells (Figs 2D, 3A,D). Similar to our findings, previous comprehensive growth analysis showed that pef1Δ cells exhibit a slow growth in the presence of rapamycin (Doi et al., 2015). Therefore, we aimed to determine whether rapamycin affects the growth of pef1Δ cells. As shown in Fig. 4A, the growth of cells was decreased by pef1 deletion as compared to control cells (empty vector-expressing pef1⁺ strain). In addition, the growth of control cells was unaltered in the presence of rapamycin, but was suppressed by canavanine, a toxic amino acid. By contrast, the growth of pef1Δ cells was slightly inhibited by rapamycin, but not by canavanine. This result indicates that the deletion of pef1 increases rapamycin sensitivity, and resistance to canavanine. Expression of Pef1 (WT, T13A and Y14F) rescued these growth defects back to normal levels, with no differences in growth between these Pef1 mutants. This indicates that the phosphorylation of Thr-13 and Tyr-14 on Pef1 does not affect cell growth. The expression of Pef1 K32R rescued the growth defects of pef1Δ cells, but these cells still exhibited rapamycin sensitivity and canavanine resistance. Therefore, we hypothesized that inactivation of Pef1 reduces TORC1 activity and amino acid uptake, and this evidence suggests that the growth defects are due to the loss of Pef1 itself.

To assess whether Pef1 is involved in the regulation of TORC1 activity, we performed western blot analysis to measure phosphorylation levels of Sck1, a substrate of TORC1 (Nakashima et al., 2012). There was less Sck1 phosphorylation in pef1Δ cells during the vegetative growth stage (at the point of rapamycin addition, 0 min) as compared with the levels of control cells (Fig. 4B). In the phos-tag gel, the phosphorylation of Sck1 was decreased after treatment with rapamycin for 30 min in control cells, but was rapidly decreased in pef1Δ cells after only 15 min. This suggests that Pef1 positively regulates TORC1 activity. We also assessed phosphorylation of Psk1, but we were unable to distinguish the differences in Psk1 phosphorylation between control cells and pef1Δ cells during vegetative growth due to strong phosphorylation signals (Fig. 4C, –NH₄Cl 0 h). However, phosphorylation levels of Psk1 at 0.5 to 2 h after nitrogen starvation were below the detection limit in control cells, but were detected at low levels in pef1Δ cells (Fig. 4C; Fig. S3A). Phosphorylation of Psk1 in pef1Δ cells was ablated by treatment with rapamycin (Fig. 4C), indicating that the low level of phosphorylation of Psk1 was due to activation of TORC1.

To investigate the cause of low TORC1 activity during nitrogen starvation in pef1Δ cells, we examined whether autophagy was correlated with low TORC1 activity, as autophagy promotes TORC1 reactivation (Shin and Huh, 2011; Yu et al., 2010). Therefore, we constructed mutants with deletion of both pef1 and atg17, an essential gene for autophagy. Western blot analysis confirmed that the TORC1 reactivation was suppressed by the deletion of atg17 (Fig. S3B). As shown in Fig. 4D, the low Psk1 phosphorylation in pef1Δ cells during starvation was abolished in pef1Δ atg17Δ cells. However, the growth of pef1Δ atg17Δ cells was unaltered as compared with that of pef1Δ cells (Fig. 4E). These results indicate that the low TORC1 activity during nitrogen starvation in pef1Δ cells is due to excessive autophagy; however, the growth defects of pef1Δ cells are not caused by overactivation of autophagy.

Additionally, we examined the intracellular localization of the mammalian raptor ortholog, Mip1. Although pef1 deletion reduced the size of the vacuole, fluorescence from EGFP-tagged Mip1 was observed at the vacuole membrane in both control and pef1Δ cells (Fig. 4F); thus, it appears that Pef1 does not control TORC1 localization. Taken together, these results suggest that Pef1 positively regulates TORC1 activity under nutrient-rich conditions and that Pef1 suppresses the reactivation of TORC1 by blocking excessive autophagy activation during nitrogen starvation.

In order to better understand the interaction of Pef1 and TORC1, we further explored these signaling pathways. Under eutrophic conditions, TORC1 generally promotes cell growth through interactions with its regulators, such as GATOR1, Rag and Ragulator. In S. pombe, it has been proposed that Gtr2 (mammalian ortholog RagC and RagD), Lam4 (mammalian ortholog LAMTOR4) and Npr2 (mammalian ortholog NPRL2) are required for maintaining adequate TORC1 activity and normal growth (Chia et al., 2017). Therefore, the disruption of these genes should lead to an imbalance in TORC1 activity, resulting in growth defects. Indeed, as shown in Fig. 5A, we confirmed that gtr2Δ, lam4Δ and npr2Δ cells were unable to grow on yeast extract with supplements (YES) but were able to grow on YES with rapamycin or on EMM. Similarly, pef1Δ gtr2Δ, pef1Δ lam4Δ and pef1Δ npr2Δ cells were not viable on YES but exhibited slow growth on YES with rapamycin and on EMM. The mating ability of gtr2Δ, lam4Δ and npr2Δ cells was significantly decreased on EMM-NH₄Cl agar, and were decreased to moderate levels on SPAS (Fig. 5B). However, the mating ability of pef1Δ gtr2Δ, pef1Δ lam4Δ, and pef1Δ npr2Δ cells was markedly low with both SPAS and EMM-NH₄Cl agar. To validate the effects of these double mutations on TORC1 activity, we examined the phosphorylation levels of Psk1. As shown in Fig. 5C–E, after nitrogen starvation, the Psk1 phosphorylation levels were low in gtr2Δ, lam4Δ and npr2Δ cells, as previously reported (Chia et al., 2017). However, phosphorylation levels were significantly increased in the each double mutant with pef1 and gtr2, lam4 or npr2 deletion (Fig. 5C–E; Fig. S3C).

We constructed GFP-atg8⁺ gtr2Δ, GFP-atg8⁺ lam4Δ, and GFP-atg8⁺ npr2Δ mutants to evaluate the levels of autophagy in these strains. During the early phases of nitrogen starvation in gtr2Δ, lam4Δ and npr2Δ cells, the amount of truncated GFP and dephosphorylated Atg13 were increased as compared with the levels in control cells (Fig. 5F–H). Rapamycin treatment did not affect truncated Atg8–GFP, Atg13 phosphorylation and PAS formation in these cells (Fig. 5I,J). Interestingly, high levels of truncated GFP and dephosphorylated Atg13 were detected in pef1Δ gtr2Δ, pef1Δ lam4Δ, and pef1Δ npr2Δ double mutants despite the nutrient rich conditions (Fig. 5F–I). In addition, the amount of truncated GFP and percentage of cells showing PAS formation in these double mutants were substantially increased by rapamycin treatment (Fig. 5I,J; Fig. S4A), indicating that these double mutants are still sensitive to rapamycin treatment. Based on these findings, we hypothesize that the double deletion of pef1 and gtr2, lam4 or npr2 causes excessive activation of autophagy, leading to abnormal TORC1 activation and thus blocking the initiation of sexual differentiation. The growth defects in pef1Δ gtr2Δ, pef1Δ lam4Δ, and pef1Δ npr2Δ cells are likely caused by the Pef1 deficit and TORC1 activation. Our results suggest that the GATOR1, Rag and Ragulator complex promotes the activation of TORC1 during the early phases of nitrogen starvation, or blocks excessive activation of autophagy through TORC1-independent pathways.

In addition to the GATOR1, Rag and Ragulator complex, the TSC/Rheb pathway is also conserved in S. pombe (Chia et al., 2017; Ma et al., 2013; Matsumoto et al., 2002; Murai et al., 2009; Urano et al., 2005, 2007; Uritani et al., 2006; van Slegtenhorst et al., 2004). Notably, previous studies have reported that pas1, which encodes the predicted activating subunit of Pef1, genetically interacts with tsc2, and is involved in the regulation of amino acid uptake and the G1 arrest of the cell cycle (van Slegtenhorst et al., 2005). Therefore, we hypothesized that pef1 genetically interacts with tsc1 or tsc2. Using a spot assay to assess resistance to canavanine, we found that pef1Δ, tsc1Δ and tsc2Δ cells all displayed resistance (Fig. 6A). Additionally, the pef1Δ tsc1Δ and pef1Δ tsc2Δ double mutants exhibited slow growth, rapamycin sensitivity and canavanine resistance to a similar extent to that found in pef1Δ cells. Mating efficiencies of tsc2Δ cells and pef1Δ tsc2Δ cells, as well as pef1Δ cells, were also significantly decreased on EMM-NH₄Cl agar as compared with SPAS (Fig. 6B). These findings suggest that the Pef1 and TSC complex are components of the same signaling pathway, as there were no differences between the pef1Δ and double mutants under these different conditions.

We next examined the effect on TORC1 activity in tsc1Δ and tsc2Δ cells. As shown in Fig. 6C,D, neither tsc1Δ nor tsc2Δ cells had decreased Psk1 phosphorylation under starvation conditions. Additionally, pef1Δ tsc1Δ and pef1Δ tsc2Δ cells were unable to attenuate Psk1 phosphorylation during starvation, with phosphorylation levels of the same intensity as tsc1Δ and tsc2Δ cells, respectively. These results suggest that tsc2 has a stronger genetic interaction with TORC1 than Pef1. In contrast, the amount of degraded GFP–Atg8 induced by nitrogen starvation was no different between control, tsc1Δ cells, and tsc2Δ cells, but that rapidly increased in pef1Δ tsc1Δ cells and pef1Δ tsc2Δ cells to the same level as pef1Δ cells (Fig. 6E,F). Rapamycin treatment increased levels of autophagy in both pef1Δ tsc1Δ cells and pef1Δ tsc2Δ cells to the same level as pef1Δ cells, but levels of autophagy remained unaltered in tsc1Δ cells and tsc2Δ cells (Fig. 6G,H; Fig. S4B). Therefore, pef1 might have a stronger genetic interaction with regards to autophagy regulation than tsc1 and tsc2. Therefore, Pef1 may positively regulate TORC1 via Tsc1 and Tsc2 but may regulate autophagy through a TSC-TORC1-independent pathway.

Although these results provide insight to the pathways downstream from Pef1, little is known about the signaling pathways upstream of Pef1 that are involved in sexual differentiation. Previous studies have suggested that Pef1 physically interacts with the three cyclins Clg1, Pas1 and Psl1 (Chen et al., 2013; Tanaka and Okayama, 2000). However, basic information on the expression pattern, the intracellular localization and the stoichiometry of interactions between Pef1 and these cyclins remain unclear. To confirm interactions between Pef1 and these cyclins, we performed immunoprecipitation. As shown in Fig. 7A, a band for Psl1–GFP near 50 kDa and subtle band for Clg1–GFP near 75 kDa were detected in the Myc-eluted solutions from cells expressing Pef1–13myc, but the Pas1–GFP signal was below the detection limit. Additionally, a Pef1–13myc band near 63 kDa was clearly detected in the Clg1–GFP and Psl1–GFP immunoprecipitants, but was undetected in the GFP immunoprecipitate of Pas1–GFP. Based on these findings, we hypothesize that the interaction of Pef1 and Psl1 is stronger than that of Pef1 and Clg1. We also observed that Pef1–GFP and Psl1–RFP were colocalized in the nucleus, and Clg1–GFP and Pas1–RFP appeared to be colocalized with Pef1–GFP in the cytoplasm (Fig. 7B); however, Pas1–RFP signals were very weak.

We next examined the effects of nitrogen starvation on the interactions between Pef1 and cyclins. To this end, we used the pef1⁺-3HA pas1⁺-13myc strain because Pas1–GFP was undetectable. As shown in Fig. 7C–F, Pef1–13myc and Psl1–GFP were consistently expressed after nitrogen starvation, whereas expression of Clg1–GFP and Pas1–13myc were decreased after nitrogen starvation. Interaction intensities between Pef1 with both Clg1 and Pas1 were also decreased by nitrogen starvation (Fig. 7G,H), but Psl1 stably interacted with Pef1 after nitrogen starvation (Fig. 7I). Taken together, these results suggest that Pef1 interacts with Clg1 and Pas1 in the cytoplasm, and with Psl1 in the nucleus. Nitrogen starvation decreased the intensity of Pef1–Clg1 and Pef1–Pas1 interaction complexes, but not that of the Pef1–Psl1 complex. Therefore we hypothesize that Clg1 and Pas1 play a role in the nitrogen starvation response.

To validate the genetic interaction between pef1 and these cyclins, we constructed a series of cyclin-deleted mutants. In Fig. 8A, a spot assay showed that single deletion of clg1, pas1 and psl1 does not affect the growth, while double mutants with pef1 and cyclin deletion have a decreased growth response on YES to the same level as that of pef1Δ cells. However, the growth of pef1Δ clg1Δ cells on EMM was lower than that of pef1Δ cells, suggesting that clg1 deletion may affect growth through a pef1-independent pathway in the case of EMM condition. In the double or triple mutants with deletion of cyclins, clg1Δ psl1Δ and pas1Δ psl1Δ cells normally grew; however, clg1Δ pasl1Δ cells grew at the same rate as pef1Δ cells. These results indicate that the growth defect is caused by loss of interactions between Pef1, Clg1 and Pas1.

We also examined the mating efficiency of cyclin-deficient mutants (Fig. 8B). The mating ability of psl1Δ cells on EMM-NH₄Cl agar was significantly decreased when compared with the mating ability on SPAS; however, pas1Δ cells had no differences in mating efficiency between SPAS and EMM-NH₄Cl agar. Additionally, the mating efficiency of clg1Δ cells was significantly lower on both SPAS and EMM-NH₄Cl agar than that of wild-type cells but was restored by the double deletion clg1Δ pas1Δ. These results suggest that the mating process is promoted by Clg1 and Psl1 and suppressed by Pas1.

In order to assess whether these cyclins genetically interact with TORC1 signaling, we assessed Psk1 phosphorylation with single deletion of clg1, pas1 or psl1. In these mutants, the Psk1 phosphorylation levels after nitrogen starvation were similar to those of wild-type cells (Fig. S5A–C). However, the Psk1 phosphorylation levels in clg1Δ pas1Δ cells and clg1Δ pas1Δ psl1Δ cells remained high after nitrogen starvation (Fig. 8C, Fig. S5G), but other double mutants (clg1Δ psl1Δ and pas1Δ psl1Δ) had no differences in the levels of phosphorylation as compared to wild-type cells (Fig. S5E,F). Although clg1Δ pas1Δ cells maintained low Psk1 phosphorylation levels during nitrogen starvation, its phosphorylation levels were higher than those of pef1Δ cells. This suggests that Clg1 and Pas1 may also control TORC1 signaling through a Pef1-dependent and -independent pathway.

We also measured levels of autophagy by assessing truncated GFP during nitrogen starvation, and observed no differences between single mutants (clg1Δ, pas1Δ, psl1Δ) and wild-type cells (Fig. S5H). However, clg1Δ pas1Δ cells had high levels of autophagy during nitrogen starvation, comparable to the levels in pef1Δ cells (Fig. 8D; Fig. S6A,B) and after rapamycin stimulation (Fig. 8E; Fig. S7A,B). Vacuole size in clg1Δ pas1Δ cells was also reduced, similar to what is seen in pef1Δ cells (Fig. S8). These results suggest that Clg1 and Pas1 regulate both autophagy and TORC1. Taken together, these analyses indicate that Clg1 and Pas1 are activators of Pef1 in the regulation of TORC1 and autophagy. In contrast, the Pef1–Psl1 complex likely controls the initiation of the mating process through a distinct pathway separate from TORC1 and autophagy, such as a meiosis-related signaling pathway.

Pef1 was originally identified as a kinase that contains a PSTAIRE motif, which is a highly conserved amino acid sequence of the CDK family (Tournier et al., 1997). There have been few studies that aimed to identify the function, activation mechanisms and downstream pathways of Pef1. In this study, we revealed that Pef1 regulates the initiation of sexual differentiation through controlling TORC1 activity, autophagy and the Ste11-Mei2 pathway (Fig. 8F). The deletion of pef1 decreased TORC1 activity toward Sck1 and increased autophagy activity in the nutrient-rich condition. In addition, the TORC1 reactivation induced by autophagy was promoted in pef1Δ cells during the early phase of nitrogen starvation. Thus, we predict that pef1 regulates the initiation of sexual differentiation through control of the balance between the TORC1 signaling pathway and autophagy.

To obtain a clue for the mechanism of the Pef1-TORC1 signaling pathway, we examined the genetic interaction between pef1 and TORC1 regulators. We found that the double deletion of pef1 and components of the GATOR1, Rag and Ragulator pathway causes excessive autophagy and TORC1 reactivation after nitrogen starvation. In mammals, nucleotide-bound Rags (RagA/B-GTP or RagC/D-GDP) tether TORC1 to the lysosomal membrane, where TORC1 is subsequently activated by the TSC/Rheb signaling pathway (Saxton and Sabatini, 2017). Although the deletion of GATOR1, Rag or Ragulator components in S. pombe did not alter the localization of Mip1, cells lacking these components were not viable in YES (Chia et al., 2017). Chia et al. also observed severe growth defects on YES when the wild-type Gtr1 (the counterpart of mammalian RagA/B) is artificially mutated so that it is locked in GTP-bound form (Gtr1 Q61L). Thus, it is hypothesized that GATOR1, Rag and the Ragulator complex suppresses TORC1 activity to maintain growth homeostasis in S. pombe. In contrast, we noticed that deletion of gtr2, lam4 or npr2 promoted autophagy during the early phases of nitrogen starvation, suggesting that these components possibly enhance TORC1 activity or suppress autophagy via a TORC1-independent manner. Therefore, simultaneous evaluation of the several TORC1 downstream pathways or cellular responses is necessary to reveal the true function of the GATOR1, Rag and Ragulator complex pathway in S. pombe.

The TSC complex is comprised of two proteins, hamartin and tuberin, which are products of the tuberous sclerosis genes tsc1 and tsc2. This complex serves as the GAP for the small G protein Rheb (Huang and Manning, 2008; Saxton and Sabatini, 2017). The mammalian TSC complex plays a critical role as the signal integrator of TORC1 regulation, and is important for sensing nutrition, growth factors and stress. TSC1 and TSC2 are the targets of various kinases, such as Akt, ERK and AMPK (Huang and Manning, 2008). CDK1–cyclin B also phosphorylates Thr-417, Ser-584, and Thr-1047 of TSC1 during the G2/M phase of the cell cycle in HEK293 cells, and phosphorylation of these sites leads to the inhibition of TSC complex activity (Astrinidis et al., 2003). Additionally, quantitative proteomic analysis indicates that Ser-862 of Tsc1 in S. pombe, which contains the CDK phosphorylation motif Ser-Thr-Pro, is phosphorylated during mitosis (Swaffer et al., 2018). This suggests that CDKs, including Pef1, in S. pombe regulate the activity of the TSC complex through direct phosphorylation of Tsc1. Alternative reports suggest that the deletion of tsc2 increases the expression of cyclin Pas1 (van Slegtenhorst et al., 2005). Therefore, the Pef1–Pas1 complex may create a positive or negative feedback loop via Tsc1 and Tsc2 to control TORC1 activity.

We also found that Pef1 is involved in the regulation of expression of Ste11 during nitrogen starvation (Fig. 1C,E,F). Ste11 plays an important role for upregulating various gene sets related to conjugation and initiation of meiosis, such as mei2 (Mata and Bähler, 2006; Sugimoto et al., 1991). Ste11 expression and activity are controlled by multiple pathways including mating pheromones, nutrition and stress (Anandhakumar et al., 2013; Otsubo and Yamamoto, 2012). For instance, the cAMP/PKA pathway suppresses the transcription of ste11 mRNA via direct inhibition of the zinc-finger transcription factor Rst2 (Higuchi et al., 2002; Kunitomo et al., 2000). The expression of Ste11 was increased in response to nitrogen starvation, but was suppressed by pef1 deletion. pef1 deletion also promoted the reactivation of TORC1. Therefore, we hypothesized that Pef1 upregulates expression of Ste11 through suppression of the TORC1 pathway; however, the Ste11-Mei2 pathway was strongly silenced in pef1Δ cells. This indicates that cAMP/PKA or another Ste11 regulatory pathway may also be impaired by pef1 gene disruption. Interestingly, mammalian PKA directly phosphorylates CDK16 and CDK18 (Graeser et al., 2002; Matsuda et al., 2014), which leads to alterations of their activities. Thus, Pef1 controls Ste11 through not only TORC1, but also via cAMP/PKA or other kinases, such as Pat1, Spk1 or Cdc2 (Anandhakumar et al., 2013; Otsubo and Yamamoto, 2012).

Nitrogen starvation induces G1 arrest of the cell cycle, but the relationship between TORC1 signaling and the cell cycle machinery remained unclear. Recent studies have investigated the great wall kinase (Gwl)/endosulfin/PP2A/B55 pathways. Gwl directly phosphorylates endosulfin, and phosphorylated endosulfin associates with the phosphatase PP2A to inhibit PP2A activity, thus, resulting in activation of CDK1–cyclin B (Pérez-Hidalgo and Moreno, 2017). S. pombe contains three Gwl ortholog genes, ppk18, cek1 and ppk31 (Aono et al., 2019). The Ppk18-Igo1 (S. pombe endosulfin) pathway is also negatively regulated by the TORC1-Sck2 pathway, resulting in the activation of PP2A, which promotes the cell cycle (Chica et al., 2016; Pérez-Hidalgo and Moreno, 2017). Interestingly, previous reports have suggested that Pef1 physically interacts with the Gwl orthologs Cek1 and Ppk18 to control chronological life span (Chen et al., 2013); however, this relationship and the molecular mechanisms are poorly understood. Notably, Cek1, Ppk18 and Ppk31 contain many CDK phosphorylation motifs, Ser-Thr-Pro, in their amino acid sequences (Cek1, 20 sites; Ppk18, 21 sites; Ppk31, 4 sites). As the direct connection between CDK and Gwl is conserved among budding yeast, Xenopus and humans, it is possible that Pef1 controls the TORC1 pathway through direct phosphorylation of Gwl.

CDK activity typically oscillates during the cell cycle via switching of their regulatory subunits, phosphorylation or binding to CDK inhibitors (Malumbres and Barbacid, 2005). Mammalian CDK5 and budding yeast PHO85 also cooperate with their interacting partner to exert their physiological functions in response to environmental changes, such as neural differentiation or various stressors (Dhavan and Tsai, 2001; Huang et al., 2007). In budding yeast, ten cyclins have been identified as PHO85 partners, known as PHO85 cyclins (PCLs). Each PCL–PHO85 complex targets different substrates, resulting in the induction of multi-spectral cell responses including growth, alterations in cell morphology or the sensing of stress (Huang et al., 2007). Our data suggests that Pef1 also changes its interacting partner in response to nutrient conditions, as Pas1 and Clg1 dissociated from Pef1 during nitrogen starvation. Additionally, the double deletion of pas1 and clg1 caused similar growth, autophagy and TORC1 activity phenotypes to those of pef1Δ cells. Thus, we predict that the Pef1–Clg1 and Pef1–Pas1 complexes control TORC1 and autophagy. In contrast, psl1Δ cells showed no mating ability on EMM-NH₄Cl agar similar to pef1Δ cells, and Psl1 stably interacted with Pef1 in the nucleus despite nitrogen starvation conditions. This implies that the Pef1-–Psl1 complex controls nuclear-specific functions, such as meiosis progression, transcription or DNA repair. Alternately, it is expected that each S. pombe Pef1–cyclin complex targets different substrates in response to altered environmental conditions, similar to PCL–PHO85 complexes in S. cerevisiae. The deletion of pas1 may alter the activity of Cdc2, as previous studies suggest that Pas1 preferentially interacts with Cdc2 over Pef1 (Tanaka and Okayama, 2000). Therefore, we hypothesize that the growth defects in pef1Δ cells may be due to activation of Cdc2, as pef1 deletion may lead to increases in the levels of free Pas1. If so, further studies investigating how the expression ratio of Pef1:Cdc2 influences the cell cycle control would be of interest.

This study provides insight into the activation mechanisms and downstream signaling pathways of Pef1. As Pef1 is a member of the conserved CDK5 subfamily, elucidating its relationship with TORC1 could contribute a better understanding and these pathways and potential drug targets for a variety of human disorders including cancer, metabolic syndrome and neurodegenerative diseases. In order to develop a better understanding of the physiological functions of Pef1, future studies should incorporate proteomic and genomic techniques to identify substrates, activators or inhibitors of Pef1.

S. pombe strains used in this study are listed in Table S1. The cells were precultured on yeast extract with supplements (YES) or Edinburgh minimal medium (EMM), which contained 20 mg/ml glucose as a carbon source and 500 µg/ml ammonium chloride as a nitrogen source (Moreno et al., 1991). Before experiments, the cells were grown in EMM under shaking at 90 rpm at 25–30°C until they reached an optical density at 595 nm (OD₅₉₅) of 0.6–0.8. EMM-NH₄Cl, a nitrogen-free medium, was used for the starvation experiments. For spot assays, 6 μl of five-times serially diluted cultures were spotted on each medium and incubated for 2–4 days at 30°C. Rapamycin (Calbiochem) and canavanine (Sigma) were used at a final concentration of 200 nM and 60 μg/ml, respectively. For mating and sporulation, cells were cultured on sporulation agar with supplements (SPAS), which contained 45 mg/l adenine, leucine, uracil, histidine and lysine as amino acid sources, or EMM-NH₄Cl agar at 25°C for 2 days (Sabatinos and Forsburg, 2010). General and molecular genetic techniques followed standard protocols (Moreno et al., 1991; Okazaki et al., 1990; Sabatinos and Forsburg, 2010).

Direct chromosomal integration methods were used to integrate 3HA-hphMX, 13myc-hphMX, GFP-hphMX, or RFP-kanMX cassettes before terminal codons of the pef1, clg1, pas1 and psl1 genes. Gene disruptions were performed by replacing individual open reading frames (ORFs) with kanMX, hphMX or natMX cassettes (Bähler et al., 1998; Sato et al., 2005). To construct pef1⁺ expression plasmids, the cDNA encoding full-length pef1⁺ was cloned by PCR using a cDNA library (NBRP) and specific primers with restriction site sequences of NotI and SmaI at their 5′ end (lowercase), respectively (forward primer, 5′-gcactatgcggccgcATGAACTACCAAAGGCTT-3′; reverse primer, 5′-gcacttcccgggCTATGCGGTTAAAAACCAAGC-3′). The PCR product was subcloned into pREP273-3HA vector (Nakashima et al., 2014), which was generated by the replacement of multi-cloning sites together with the nmt41 promoter (Basi et al., 1993) and 3HA fragment from pSLF273 (NBRP). Point mutations of Pef1 were introduced by site-directed mutagenesis. To construct a tor2⁺ expressing vector, a PCR product, which consisted of tor2⁺ ORF and 1000 bp of its 5′ and 3′ flanking regions, was inserted into pAL-KS-vector using the infusion cloning method.

Cell cultures were mixed with 7% trichloroacetic acid, put on ice for at least 10 min, and then centrifuged (3300 rpm for 5 min). Cell pellets were washed twice with cold acetone and dried. Dried cells were lysed in buffer A [50 mM Tris-HCl pH 7.5, 150 mM NaCl, 5 mM EDTA, 10% glycerol, 20 mM β-glycerophosphate, 0.1 mM Na₃VO₄, 10 mM p-nitrophenyl phosphate (p-NPP), 10 mM NaF] containing 0.2% NP-40 and glass beads, and disrupted by vortexing. Whole-cell extracts were mixed with 3× SDS sample buffer, and boiled for 5 min. For immunoprecipitation, exponentially grown cells were treated with 1 mM PMSF for a few minutes under shaking at 90 rpm. After centrifugation, cell pellets were washed with diluted water containing 1 mM PMSF, and frozen using liquid nitrogen. Cells were lysed in buffer B [50 mM Tris-HCl pH 7.5, 150 mM NaCl, 5 mM EDTA, 10% glycerol, 20 mM β-glycerophosphate, 0.1 mM Na₃VO₄, 10 mM p-NPP, 10 mM NaF, 1 mM PMSF, protease inhibitor cocktail and phosphatase inhibitor cocktail (Roche)] with glass beads, and then homogenized using FastPrep 24 (MP-biomedicals: 6.5 m/s, 40 s for six times with a 5 min interval on ice). Cell lysates were diluted with an equal amount of buffer B containing 0.4% NP-40 and rotated for 30 min at 4°C. After centrifugation, supernatants were incubated with anti-HA affinity matrix (Roche) or protein G–Sepharose (GE healthcare) and rabbit anti-GFP antibody (MBL; 1:400; cat. no 598) at 4°C for 4 h. Immunoprecipitants were washed three times with buffer A containing 0.1% NP-40. For Myc elution, Myc immunoprecipitants were incubated with 1 mg/ml Myc peptide solution for 15 min at 30°C two times. Cell extracts, immunoprecipitants or Myc eluates were boiled with 3× SDS sample buffer, separated by SDS-PAGE, and then subjected to western blotting. Anti-phospho-p70 S6 K (Thr-389) antibody was purchased from Cell Signaling Technology (1:2000; cat. no 9206S). Anti-α-tubulin (B5-1-2) antibody was obtained from Sigma (1:2000; cat. no T5168). Anti-myc antibody was from Wako (1:3000; cat. no 017-21871). Anti-HA (12CA5) (1:8000; cat. no 11583816001) and mouse anti-GFP antibodies were from Roche (1:1000; cat. no 11814460001). Rabbit anti-GFP antibody was from MBL (1:1000; cat. no 598). Anti-PSTAIRE antibody was from Santa Cruz Biotechnology (1:1000; cat. no sc-53). Anti-Atg13 antibody was kindly provided by Akira Yamashita (National Institute for Basic Biology, Japan). After incubation with HRP- conjugated secondary antibodies, proteins were detected using the ECL Plus Western Blotting Substrate (Thermo Scientific). Intensities of immunoblotting bands were measured by Image J software (NIH).

Cells were cultured in EMM or EMM-NH₄Cl. RNA was extracted by disrupting cells by lysis in buffer C (0.2 M Tris-HCl pH 7.5, 0.5 M NaCl, 10 mM EDTA and 1% SDS) with glass beads. RNA was purified by removing proteins using phenol/chloroform/isoamyl alcohol mixed at 25:24:1 (pH 5.2) (Nakarai Tesque) followed by ethanol precipitation. For qPCR, 0.5 μg of total RNA was reverse transcribed to cDNA using ReverTra Ace qPCR RT Master Mix (TOYOBO). qPCR was performed using LightCycler 480 (Roche) and THUNDERBIRD SYBR qPCR Mix (TOYOBO). Primer sequences are described below. qPCR primerd were: ste11, forward primer 5′-CTTCTTACCCAGCAATGAGT-3′, reverse primer 5′-CAATTCTTGAGAATGAGCGG-3′; mei2: forward primer 5′-TGGAATTGACACAAGAACCA-3′, reverse primer 5′-GAGTACCCACTCTAGCTTTG-3′.

To calculate mating efficiency, cells were treated with 70% ethanol, and then stained with DAPI. Non-mating cells, zygote, ascus and free-spore were distinguished, respectively, by using the fluorescence microscopy BZ-8000 (Kyence). Mating efficiency was calculated as ratio of the number of diploid cells to total cells. To visualize vacuoles, cells were incubated with FM4-64 (Invitrogen/Thermo Fisher) for 30 min at 30°C under shaking at 110 rpm. Cells were grown in filtrated EMM on a lectin-coated glass-bottomed dish (Greiner) followed by image analysis using fluorescent microscopy BZ-9000 (kyence). BZ-analyzer software (kyence) was employed for measuring the size of vacuole and fluorescence intensity.

The data are shown as the mean±s.e.m. of the indicated number of observations. To determine significant differences, Student's t-tests and one-way analysis of variance (ANOVA) were performed using Prism ver. 8. P<0.05 was considered statistically significant.

The authors thank Mrs. Shino Oguri for her administrative support and warm encouragement. The authors also appreciate Satoshi Kawamoto, Taiki Nagano, Moena Okahisa, and Yui Miyamoto for their technical support. The authors also appreciate Drs Masamitsu Sato, Fuyuhiko Tamanoi, Akira Yamashita, Masayuki Yamamoto, and the National BioResource Project (NBRP) of the MEXT, Japan for providing S. pombe strains, plasmids and antibody.