A yeast MAPK cascade regulates pexophagy but not other autophagy pathways

Autophagy maintains cellular homeostasis by the recycling of cytoplasmic constituents including damaged or redundant organelles. Two modes of autophagy of nonselective cargoes are recognized: macroautophagy and microautophagy. During macroautophagy (referred to as autophagy), which can be activated by starvation or depletion of amino acids, components of the cytosol are nonselectively engulfed by a double membrane (called the isolation membrane) into vesicles called autophagosomes. Fusion of the outer autophagosomal membrane with the vacuole (yeast lysosome) membrane delivers autophagic vesicles into the vacuole where hydrolases degrade and recycle the proteins and lipids, allowing cells to survive starvation (Yorimitsu and Klionsky, 2005). In microautophagy, the lysosome itself invaginates, sequesters cargo, and degrades and recycles the materials (Nakatogawa et al., 2009).

Recently, other autophagy-related pathways that degrade cargo selectively have been recognized. Such selective forms of autophagy include the cytosol to vacuole transport (Cvt) protein targeting pathway, peroxisome turnover by microautophagy (micropexophagy) or macroautophagy (macropexophagy), degradation of mitochondria (mitophagy), the ER (ER-phagy), or parts of the nucleus (micronucleophagy) and even ribosome degradation (ribophagy; Kraft et al., 2008; Farré et al., 2009). These forms of autophagy use many of the common core components of the autophagic machinery, superimposed on which are a variety of selectivity factors.

Signaling events in general autophagy have been studied to an extent. Nonselective or bulk autophagy is negatively controlled by the target of rapamycin (Tor), Sch9 protein kinase, and Ras/cAMP-dependent PKA pathways (Budovskaya et al., 2004; Yorimitsu et al., 2007). These kinases maintain proteins involved in autophagy induction, namely Atg1p and Atg13p, in their inactive, hyperphosphorylated states. Additionally, in mammalian cells, the MAPKs, extracellular signal–regulated kinase, and p38 play important roles in autophagosome maturation (Corcelle et al., 2007). In contrast, hardly any information exists regarding the signaling events that occur during selective autophagy. Although glucose sensing and transport affect pexophagy (Nazarko et al., 2008a,b), studies on pexophagy in Pichia pastoris have uncovered only one protein and an unknown kinase in pexophagy. The α subunit of phosphofructokinase is required for micropexophagy but not macropexophagy (Yuan et al., 1997), and the phosphorylation of the peroxisome receptor, Atg30p, by an unknown kinase is necessary for both macropexophagy and micropexophagy (Farré et al., 2008). However, the signaling cascade involved is unknown. Knowledge of these signaling pathways affecting selective autophagy is particularly important because during these processes, the nonselective autophagy machinery is adapted to degrade only specific cargoes (Nazarko et al., 2009). We addressed this question by studying the signaling pathways in Saccharomyces cerevisiae that trigger pexophagy (Evers et al., 1991; Chiang et al., 1996; Hutchins et al., 1999).

We screened the collection of viable haploid S. cerevisiae kinase/phosphatase deletion strains (Saleem et al., 2008) for defects in pexophagy. The assay tested the ability of strains to degrade the peroxisomal matrix protein, thiolase, fused to GFP (Pot1p-GFP). When wild-type cells were grown in oleate medium for 14–20 h, an increase in Pot1p-GFP expression was seen with the concomitant appearance of GFP-labeled peroxisomes (Fig. S1, a and b). Upon transfer to glucose medium, pexophagy was monitored by the appearance of free, protease-resistant GFP in the vacuole and a decrease in the Pot1p-GFP levels as detected by immunoblotting (Fig. 1 a). Within 6 h of shift to glucose medium, GFP appeared in the vacuole in wild-type cells (Fig. 1 b). Not surprisingly, the screen identified ATG1, a gene involved in all autophagy-related pathways including pexophagy (Yorimitsu and Klionsky, 2005; Nakatogawa et al., 2009). The atg1Δ cells were deficient in pexophagy (Fig. 1, a–c).

Interestingly, cells deleted for the MAPK SLT2 gene (Torres et al., 1991), also known as MPK1 (Lee et al., 1993), were also defective in pexophagy. Unlike the wild-type but similar to atg1Δ cells, the slt2Δ cells showed neither a decrease in the levels of Pot1p-GFP nor the appearance of free GFP after 24 h of transfer to SD-N (pexophagy) medium (Fig. 1 a). Similarly, the slt2Δ cells were also defective in pexophagy of another peroxisomal matrix protein, Fox2, fused to the RFP (Fox2p–monomeric RFP; Fig. S1 c). Previous experiments showed that glucose-induced pexophagy is independent of the presence or absence of nitrogen (Yuan et al., 1997; Farré et al., 2008). However, the slt2Δ cells were proficient in peroxisome biogenesis as judged by the peroxisome morphology relative to wild-type cells, Pot1p expression, and growth in oleate medium (Fig. S1, a, b, and d). During pexophagy, a pexophagosome sequesters peroxisomes. To test whether pexophagosomes are formed in the slt2Δ cells, we used a protease protection assay wherein the formation of pexophagosomes would protect a peroxisomal membrane protein (Inp2p) from protease digestion. Postnuclear membrane fractions of cells incubated for 2 h in pexophagy conditions were treated with or without proteases (proteinase K and trypsin) in presence or absence of detergent. The atg24Δ cells, which form pexophagosomes but do not fuse to the vacuole, showed protease protection of Inp2p, as did wild-type cells (Fig. 1 d). In contrast, atg5Δ cells that do not form pexophagosomes showed protease sensitivity of Inp2. The slt2Δ cells displayed protease protection of Inp2p, suggesting normal pexophagosome formation in these cells (Fig. 1 d). All of the tested strains exhibited protease protection of the peroxisomal matrix marker, Pot1p, as expected. Among the six MAPKs present in S. cerevisiae, only slt2Δ cells showed a pexophagy defect (Fig. 1 e).

We next asked whether general autophagy or the Cvt pathways were perturbed in slt2Δ cells. Processing of the precursor Ape1p to the mature form by the constitutive Cvt pathway or by general autophagy, monitored during growth and starvation conditions, respectively (Klionsky et al., 2007), was unaffected in slt2Δ cells as compared with wild-type cells (Fig. 2 a). Accumulation of the cleaved GFP tag in the vacuole, resulting from the vacuolar delivery of GFP-Atg8p, is a measure of autophagy (Yorimitsu et al., 2007). Unlike atg1Δ cells, processing of GFP-Atg8p by general autophagy was unaltered in slt2Δ cells (Fig. 2 b). Furthermore, the slt2Δ cells were not defective for other selective autophagy pathways such as mitophagy (Fig. S2 a) and ribophagy (Fig. S2 b), indicating that general autophagy and several selective autophagy pathways other than pexophagy are normal in the slt2Δ cells.

Slt2p is a MAPK regulated by the PKC1 cell wall integrity (CWI) kinase signaling cascade (Levin, 2005). This pathway comprises of the MAPKKKK, Pkc1p, acting upstream of the MAPKKK, Bck1p, the functionally redundant MAPKKs, Mkk1p and Mkk2p, and the MAPK, Slt2p (Levin, 2005). We tested whether the upstream components of this pathway were also involved in pexophagy. The diploid PKC1/pkc1Δ cells (haploid-null strain is inviable) were completely blocked in degradation of Pot1p (Fig. 3 a). The bck1Δ cells degraded peroxisomes less efficiently and had an intermediate effect on pexophagy. Although independent deletions of MKK1 or MKK2 were normal for pexophagy (unpublished data), the double-deletion strain mkk1Δ mkk2Δ resembled the bck1Δ mutant in that it was partially blocked (delayed) in pexophagy. Because the bck1 and mkk1 mkk2 mutants failed to activate Slt2p as judged by the absence of phospho-Slt2p (unpublished data), the incomplete pexophagy block may be a consequence of aberrant cross talk between the PKC1 pathway and other MAPKs (McClean et al., 2007; García et al., 2009). For example, in S. cerevisiae, stimuli such as heat stress that activate the Slt2p pathway also activate the Hog1p pathway (Winkler et al., 2002). In view of this observation, it is plausible that blocking the transmission of a signal via the normal pathway (loss of Bck1 or of Mkk1 Mkk2) might aberrantly activate some other MAPK cascade.

The Slt2p MAPK components were also analyzed for defects in the general autophagy and Cvt pathways. As with slt2Δ cells, maturation of precursor Ape1p was unperturbed in PKC1/pkc1Δ bck1Δ, and mkk1Δ mkk2Δ cells (Fig. 3 b). These mutants also processed GFP-Atg8p during autophagy conditions (Fig. 3 c). These results imply a significant involvement of the PKC1 pathway in pexophagy but not in general autophagy or Cvt pathways.

Signaling through the PKC1 pathway results in Mkk1p and/or Mkk2p activating Slt2p by phosphorylation on threonine 190 (T190) and tyrosine 192 (Y192) residues (Lee et al., 1993). When cells were grown on oleate medium, Slt2p was mostly in the unphosphorylated state but became phosphorylated at T190 and Y192 residues within 15 min after pexophagy induction in the wild-type cells (Fig. S3 a). When a Slt2p signaling mutant (TAYF; T190 and Y192 residues mutated to alanine and phenylalanine, respectively) was introduced in slt2Δ cells, this strain (slt2Δ + SLT2 TAYF) was defective in pexophagy (Fig. 4). In contrast, as expected, introduction of the SLT2 gene into slt2Δ cells (slt2Δ + SLT2) restored pexophagy. This indicates that signaling through the Mkk1p-Mkk2p phosphorylation sites (T190 and Y192) on Slt2p is necessary for pexophagy. Additionally, moderate overexpression of Slt2p accelerated pexophagy, whereas stronger overexpression inhibited pexophagy, even in wild-type cells (unpublished data). Our working hypothesis is that elevated levels of Slt2p might result in hyperactivation of Slt2p, which might either lead to increased downstream signaling or have a dominant-negative effect on pexophagy. These data also support the notion that the signals for pexophagy are most likely relayed in a linear fashion through the Slt2p MAPK cascade.

The activated Slt2p kinase phosphorylates downstream substrates. We investigated the requirement of the catalytic activity of Slt2p for pexophagy. Interestingly, the kinase-dead mutant (slt2Δ + SLT2 K54R) of Slt2p (Martín et al., 1993) was also unable to fully complement the pexophagy defect in slt2Δ cells (Fig. 4). Moreover, as observed by others (Martín et al., 1993), this Slt2p mutant was hyperphosphorylated at the T190 and Y192 sites, perhaps because of the inability to repress upstream signaling by feedback inhibition through its catalytic mechanism (Martín et al., 1993). Overall, these results demonstrate that during pexophagy, Slt2p probably phosphorylates hitherto unidentified substrates, and this downstream signaling via Slt2p is important for pexophagy.

How is the PKC1 pathway activated during pexophagy? In the case of the CWI pathway, cell surface sensors that are known to activate this MAPK cascade are the four WSC family members Wsc1p/Hcs77p/Slg1p, Wsc2p, Wsc3p, and Wsc4p (Verna et al., 1997; Levin, 2005), as well as Mid2p and its homologue Mtl1p (Ketela et al., 1999; Rajavel et al., 1999). The wsc mutants demonstrate weakened cell wall phenotypes, whereas Mid2p is a cell wall mechanosensor. Mtl1p serves a partially redundant function with Mid2p for CWI signaling at elevated temperatures. Pexophagy was completely blocked in the mid2Δ cells, similar to that in atg1Δ cells (Fig. 5 a). Although wsc2Δ, wsc4Δ, and mtl1Δ cells were proficient in pexophagy, wsc1Δ and wsc3Δ were partially delayed in pexophagy. Because Mid2p directly activates the Slt2p MAPK pathway and is not known to act independently of the Slt2p pathway, these results imply that Mid2p acts as an important sensor for pexophagy. Moreover, as several of these sensors play redundant roles, it is probable that Wsc1p and Wsc3p also contribute toward transducing this signal.

Signaling by the Slt2p MAPK cascade is important for various cellular functions including CWI, pH sensing, osmoregulation, and the unfolded protein response (Levin, 2005; Serrano et al., 2006; Scrimale et al., 2009). Data presented in this study show a specific role of the Slt2p signaling pathway in pexophagy in several ways. First, other MAPKs do not affect pexophagy. Second, absence of a specific cell surface sensor, Mid2p, known to activate the Slt2p pathway, blocks pexophagy, which is analogous to the activation of the Slt2p MAPK cascade by single sensors for other cellular processes (Martín et al., 2000). Third, although the Slt2p-TAYF phosphomutant is blocked in pexophagy, artificially activating Slt2p by heat stress or Calcofluor white (not depicted) during peroxisome biogenesis conditions did not trigger pexophagy in either wild-type or atg1Δ cells (Fig. S3 b). Fourth, the presence of osmotic support (1M sorbitol), a treatment known to suppress Slt2p phosphorylation and the effects of cell wall damage (de Nobel et al., 2000), did not prevent pexophagy or the appearance of phospho-Slt2p during pexophagy in wild-type cells (Fig. S3 c). Thus, the activation of Slt2p is a necessary prerequisite for pexophagy; however, it is not sufficient to drive peroxisome degradation. There is indeed precedent for the activation of cellular processes by the integration of dual signaling pathways. In S. cerevisiae, the expression of the floccullin gene, FLO11, depends on the integration of signals from the PKA/cAMP pathway and the STE20 MAPK cascade (Vinod and Venkatesh, 2007).

In addition to the activation of Slt2p, pexophagy needs a second signal. Wild-type cells degrade peroxisomes when shifted from a fatty acid carbon source to a sugar such as glucose or galactose, as shown in this study. Because slt2Δ cells did not degrade peroxisomes in these conditions, we wondered whether the impairment was specific to sensing glucose or nitrogen starvation. We found that wild-type cells did degrade peroxisomes and Slt2p was also phosphorylated when switched from oleate to other sugars such as galactose, either in the presence or absence of nitrogen as a supplement (data not depicted in the presence of nitrogen; Fig. S3 d). In contrast to wild-type cells, slt2Δ cells were blocked in pexophagy irrespective of the aforementioned conditions, suggesting that pexophagy-specific Slt2p signaling is necessary for peroxisome degradation. This second signal, activated extracellularly by hexoses (glucose and galactose), has been shown to require the G protein–coupled receptor, Gpr1, and the G protein, Gpa2, which act via the PKA–cAMP pathway (Nazarko et al., 2008b).

Modeling of FLO11 expression pathway shows that a weak signal provided by the MAPK cascade, although sufficient to provide only a basal level of FLO11 expression, is amplified significantly by cross talk with the PKA pathway (Vinod and Venkatesh, 2007). Given that glucose-induced pexophagy also activates the PKA–cAMP pathway (Nazarko et al., 2008b), we propose that similar signal integration might be necessary between the PKA and Slt2p pathways to trigger pexophagy in S. cerevisiae.

The Slt2p pathway receives multiple/alternate inputs, resulting in differential activation of the pathway to bring about the desired functions. Although six sensors have been implicated in signaling to the Slt2p MAPK module, not all of them are involved in every Slt2p pathway–dependent process. The choice of a particular sensor in activating this MAPK cascade in response to differential stress conditions such as alkaline pH or heat stress is unclear. For example, overexpression of Mid2p partially complements the deletion of WSC1 (Ketela et al., 1999). Thus, the sensors, although specific, have overlapping functions. Glycosylated Mid2p acts as a mechanosensor and activates the PKC1 pathway via its association with Rom2p, a guanine nucleotide exchange factor for Rho1p (Philip and Levin, 2001). Because cell wall composition is influenced by changes in carbon source (Lesage and Bussey, 2006), and as induction of pexophagy involves adapting cells from oleate to glucose medium, it is probable that Mid2p might either directly or indirectly sense these cell wall modifications to activate Slt2p.

Finally, we observed that there are differential kinetics of Slt2p phosphorylation during pexophagy (sustained phospho-Slt2p levels; Fig. S3 a) as compared with alkaline pH stress (Serrano et al., 2006) and hypotonic shock (transient activation of phospho-Slt2p; Davenport et al., 1995). This feature has been observed previously for the outputs of MAPK pathways.

What are the downstream targets of Slt2p? We show that although the kinase activity of Slt2p is important for pexophagy, Slt2p is not required for pexophagosome formation. Although several studies have identified downstream transcriptional targets of Slt2p, our hypothesis is that transcriptional changes alone are not likely to trigger pexophagy because the degradation of peroxisomes by macropexophagy is not affected by blocking cellular translation (cycloheximide treatment) in the yeast P. pastoris (Tuttle and Dunn, 1995; Sakai et al., 1998). Rather, it is likely that novel, nontranscriptional targets of Slt2p, such as protein components involved in pexophagosome movement to (Monastyrska et al., 2006), or fusion with, the vacuole, may be involved.

The PKC1 MAPK cascade is highly conserved from yeast to mammals. Recently, it was shown in mammalian cells that JNK1, a MAPK, stimulates starvation-induced autophagy by phosphorylating Bcl2, an autophagy inhibitor that keeps Beclin1, a component of the autophagy-specific PtdIns-3 kinase, in check (Wei et al., 2008). Interestingly, autophagy has also been shown to be modulated by mammalian PKC isoforms: PKC-δ, with contradicting roles (Ozpolat et al., 2007; Chen et al., 2008), and PKC-θ, in the autophagic response to ER stress (Sakaki et al., 2008). Previous data have also suggested the involvement of the MAPKs Erk1/2 as activators of autophagy in some mammalian cell types, but not in others, and a role for the MAPK p38 (and its yeast orthologue HOG1) as a negative regulator of autophagy (Meijer and Codogno, 2006; Prick et al., 2006). To our knowledge, the results reported in this paper show for the first time an involvement of the MAPK cascade in selective rather than nonselective autophagy pathways, highlighting a broad, evolutionarily conserved role for MAPK cascades in autophagy-related pathways.

S. cerevisiae strains were grown in rich medium (1% yeast extract, 2% peptone, and 2% glucose [YPD]) or defined synthetic medium (SD; 0.17% yeast nitrogen base [YNB], 0.5% ammonium sulfate, 2% glucose, and amino acids as required) at 30°C on a shaker set at 250 rpm. For peroxisome induction, cells were transferred to oleate medium (1% oleate, 5% Tween-40, 0.25% yeast extract, 0.5% peptone, and 5 mM phosphate buffer). Pexophagy was induced by transferring cells to SD-N medium (0.17% YNB without ammonium sulfate and 2% glucose) or SG-N (0.17% YNB without ammonium sulfate and 2% galactose). Yeast strains used in this study are listed in Table S1.

We used the S. cerevisiae viable deletion set of kinases and phosphatase mutants that has GFP genomically tagged to the peroxisomal matrix marker Pot1p (Saleem et al., 2008). Cells were grown in YPD medium and transferred to 4 ml (A₆₀₀ ∼0.5–1/ml) oleate medium for peroxisome induction. After overnight incubation, 2 ml of the culture was washed twice with SD-N and further incubated for 24 h in the same medium. Lysates were prepared using the rapid alkaline lysis method.

For GFP-Atg8p–processing assays, GFP-ATG8 under its endogenous promoter was subcloned (EcoRI–XhoI cloning) from pRS414GFPAut7 (provided by D.J. Klionsky, University of Michigan, Ann Arbor, MI; Reggiori et al., 2005) into the 2µ URA3 plasmid vector pRH1430 (provided by R. Hampton, University of California, San Diego, La Jolla, CA). The SLT2 plasmids p2188 pRS315[MPK1-3xHA], p2193 pRS315[mpk1(K54R)-3xHA], and p2190 pRS315[mpk1(T190A Y192F)-3xHA] (Kim et al., 2008) were provided by D.E. Levin (Boston University, Boston, MA). The 3× HA tags from these plasmids were removed by QuikChange mutagenesis (Agilent Technologies) with the primers 5′-GGATTAGATA­GAAAATATTTT­TAGGACAAAAA­ACTATAAGTAA­CCGG-3′ and 5′-CCGGTTACTT­ATAGTTTTTTG­TCCTAAAAATA­TTTTCTATCTA­ATCC-3′.

Yeast cells were cultured in oleate medium containing 5 µg/ml FM 4–64 in log-phase cultures (A₆₀₀ ∼0.3–0.4/ml) and transferred to SD-N. Images were captured at room temperature using a Plan Apochromat 100× 1.40 NA oil immersion objective on a motorized fluorescence microscope (Axioskop 2 MOT plus; Carl Zeiss, Inc.) coupled to a monochrome digital camera (AxioCam MRm; Carl Zeiss, Inc.) and processed using AxioVision software (version 4.5; Carl Zeiss, Inc.).

For preparation of extracts by alkaline lysis, cells were centrifuged, washed with sterile distilled water, and resuspended in 0.2 M NaOH and 0.01 vol β-mercaptoethanol for 5 min. After centrifugation at 15,000 g for 1 min at room temperature, the cell pellet was resuspended in SDS-PAGE sample loading buffer and boiled for 10 min. For the rest of the biochemical assay, samples were prepared by TCA extraction. Samples (A₆₀₀ 0.3 equivalent) were resolved in 12% SDS-PAGE followed by Western blotting with anti-Pot1p (rabbit; 1:5,000; Subramani Laboratory), anti-Ape1p (rabbit; 1:5,000, Klionsky Laboratory; Reggiori et al., 2005), anti-GFP (mouse; 1:2,000; Roche), anti–total Slt2p (1:200; MPK1; Santa Cruz Biotechnology, Inc.), or anti–phospho-Slt2p (1:1,000; phospho-p44/42 MAPK; Thr202/Tyr204; Cell Signaling Technology). Secondary antibodies were either anti–rabbit or anti–mouse polyclonal (both 1:10,000; Roche) followed by enhanced chemiluminescence (GE Healthcare).

Cells grown overnight in oleate medium were transferred to SD-N for 2 h. Cells were collected (1 gm), washed with sterile water, and resuspended in 4 ml zymolyase buffer (0.5 M KCl, 5 mM MOPS/KOH buffer, pH 7.2, and 10 mM Na₂SO₃) containing 0.5 mg zymolyase 100 T/g cells. After incubating for 30 min at 30°C, spheroplasts were harvested and homogenized in a dounce homogenizer (20 strokes) in the presence of homogenization buffer (5 mM MES/KOH, pH 5.5, and 1 M sorbitol; 3 ml/g cells). The postnuclear supernatant was obtained by collecting the supernatant of homogenates centrifuged at 1,000 g. Similarly, the 200,000 g pellet was obtained by centrifugation of the postnuclear supernatant at 200,000 g. The pellets (100 µg protein) were resuspended in 50 µl homogenization buffer with or without 200 µg proteinase K and 100 µg trypsin. The proteases were added in the presence or absence of 1% Triton X-100. After incubation on ice for 30 min, samples were processed by TCA precipitation for immunoblotting of peroxisomal membrane and matrix markers.

Figs. S1–S3 show analyses of the role of Slt2p in peroxisome biogenesis, mitophagy, and ribophagy, as well as the role of Slt2p in pexophagy under different physiological conditions. Table S1 lists the strains used in this study.

We would like to thank Richard Rachubinski, Barbara Knoblach (University of Alberta, Edmonton, Alberta, Canada), Javier Arroyo (Universidad Complutense de Madrid, Madrid, Spain), María Molina (Universidad Complutense de Madrid, Madrid, Spain), David E. Levin, Maho Niwa, Randolph Hampton, and Daniel Klionsky for sharing plasmids and strains and Gaurav Agrawal (University of California, San Diego, La Jolla, CA) for help and advice.

This work was supported by a National Institutes of Health grant (GM GM69373 to S. Subramani).