The Thermotolerant Yeast Kluyveromyces marxianus Is a Useful Organism for Structural and Biochemical Studies of Autophagy*

Background: Autophagosome formation is mediated by multiple autophagy-related (Atg) proteins. Results: Essential Atg proteins of K. marxianus, which have superior thermostability and solubility, are identified. Conclusion: K. marxianus can be used as a novel organism to study autophagy. Significance: K. marxianus proteins are broadly applicable as tools for in vitro studies, not only in autophagy field but also in other fields. Autophagy is a conserved degradation process in which autophagosomes are generated by cooperative actions of multiple autophagy-related (Atg) proteins. Previous studies using the model yeast Saccharomyces cerevisiae have provided various insights into the molecular basis of autophagy; however, because of the modest stability of several Atg proteins, structural and biochemical studies have been limited to a subset of Atg proteins, preventing us from understanding how multiple Atg proteins function cooperatively in autophagosome formation. With the goal of expanding the scope of autophagy research, we sought to identify a novel organism with stable Atg proteins that would be advantageous for in vitro analyses. Thus, we focused on a newly isolated thermotolerant yeast strain, Kluyveromyces marxianus DMKU3-1042, to utilize as a novel system elucidating autophagy. We developed experimental methods to monitor autophagy in K. marxianus cells, identified the complete set of K. marxianus Atg homologs, and confirmed that each Atg homolog is engaged in autophagosome formation. Biochemical and bioinformatic analyses revealed that recombinant K. marxianus Atg proteins have superior thermostability and solubility as compared with S. cerevisiae Atg proteins, probably due to the shorter primary sequences of KmAtg proteins. Furthermore, bioinformatic analyses showed that more than half of K. marxianus open reading frames are relatively short in length. These features make K. marxianus proteins broadly applicable as tools for structural and biochemical studies, not only in the autophagy field but also in other fields.

Macroautophagy (hereafter referred to as autophagy), a fundamental cellular process conserved from yeast to mammals, mediates bulk degradation of cytoplasmic proteins and organelles in response to starvation (1)(2)(3)(4). Autophagy has attracted considerable interest in the fields of biological and medical sciences because it plays important roles in a variety of cellular events, including metabolic adaptation, stress response, quality control, development, tumor suppression, and renovation of cellular components (5)(6)(7). Morphologically, autophagy involves de novo formation of a double-membrane structure, called an autophagosome, that sequesters cytoplasmic materials. After the sequestration, the autophagosome fuses with lytic compartments (vacuoles in yeast and plants and lysosomes in mammals), leading to degradation of its contents (3,4).
Previous studies using the yeast Saccharomyces cerevisiae identified nearly 40 autophagy-related (Atg) 5 proteins involved in various types of autophagy (3,4,8). Among these, 18 Atg proteins (Atg1-Atg10, Atg12-Atg14, Atg16-Atg18, Atg29, and Atg31), defined as core Atg proteins (1), are crucial for the process of autophagosome formation. These Atg proteins are functionally and hierarchically classified into six subgroups as follows: the Atg1 complex (Atg1, Atg13, Atg17, Atg29, and Atg31); a vesicular membrane protein required for the early step of autophagosome formation (Atg9); the autophagy-specific PtdIns 3-kinase complex (Atg6 and Atg14; also includes Vps15 and Vps34); the phosphatidylinositol 3-phosphate effector complex (Atg2 and Atg18); and two ubiquitin-like conjugation systems (Atg3, Atg4, Atg5, Atg7, Atg8, Atg10, Atg12, and Atg16). Consequently, most of the key findings concerning the molecular basis of autophagosome formation have been obtained in the model yeast S. cerevisiae. In particular, the members of the two ubiquitin-like conjugation systems have been well characterized, because several lines of structurebased analyses (9 -15) and in vitro reconstitution studies (16 -18) have provided critical insights into their molecular functions. However, these structural and biochemical studies have been limited to a subset of the Atg proteins, because the rest of Atg proteins are difficult to prepare as recombinant proteins and cannot be purified efficiently from yeast cells. Therefore, the detailed functions of these Atg proteins remain to be elucidated.
One of the major problems in preparation of recombinant Atg proteins, most of which have been derived thus far from S. cerevisiae and mammals, is the modest stability of these proteins. We predicted that recombinant proteins derived from thermotolerant organisms would exhibit superior stability against high temperature and chemical reagents relative to their counterparts from S. cerevisiae. As an illustration of this principle, a heat-resistant TaqDNA polymerase derived from the thermophilic bacterium Thermus aquaticus is widely used in polymerase chain reaction techniques (19), and recently, Amlacher et al. (20)succeeded in reconstituting the structural modules of nuclear pore complexes using proteins from the thermophilic fungus Chaetomium thermophilum. Hence, we attempted to utilize the thermotolerant yeast strain Kluyveromyces marxianus DMKU3-1042, which can grow at temperatures above 49°C (21), because its Atg homologs are predicted to be thermostable and useful for structural and biochemical studies. In this study, we first identified the complete set of core Atg proteins of K. marxianus and then investigated their thermostability and solubility by biochemical analyses. Complementation assays showed that the K. marxianus Atg homologs can functionally replace their counterparts in S. cerevisiae cells. We propose that K. marxianus could be useful as a new model organism for further elucidation of the molecular details of autophagy.

Experimental Procedures
Yeast Strains, Media, Plasmids, and Other Materials-Yeast strains used in this study are listed in supplemental Table 1. For cultivation of S. cerevisiae and K. marxianus cells, standard protocols of S. cerevisiae studies were used (22). Yeast cells were cultured at 30°C in nutrient-rich medium YPD (1% yeast extract, 2% bacto-peptone, 2% glucose) or SD/CA (0.17% yeast nitrogen base without amino acids and ammonium sulfate, 0.5% ammonium sulfate, 0.5% casamino acids, 2% glucose) supplemented with 20 g/ml adenine, 20 g/ml uracil, and/or 20 g/ml tryptophan. To induce autophagy, yeast cells were transferred to nitrogen starvation medium SD(ϪN) (0.17% yeast nitrogen base without amino acids and ammonium sulfate, 2% glucose) or treated with 0.2 g/ml rapamycin (Sigma). Gene deletions of S. cerevisiae cells were performed by using pFA6a-kanMX6, pFA6a-hphNT1, and pFA6a-natNT2 plasmids as reported previously (23). Gene deletions of K. marxianus cells were performed by using pFA6a-kanMX6, pFA6a-hphNT1, and pFA6a-natNT2 plasmids as reported previously (24). The K. marxianus cells expressing GFP-KmAtg8 were constructed; a DNA fragment, including the KmATG8 promoter, the KmATG8 gene, and the KmATG8 terminator (from 1000-bp upstream region of the initiation codon to 250-bp downstream region of the termination codon of the KmATG8 gene), was amplified from genomic DNA of K. marxianus and cloned into pFA6a-kanMX6 (23). A BamHI site was introduced into the resultant plasmid just downstream of the first codon of the KmATG8 gene by QuikChange site-directed mutagenesis (Stratagene), and a DNA fragment encoding GFP was inserted into the BamHI site, yielding pFA6a-GFP-KmAtg8-kanMX6. A DNA fragment, including the KmATG8 promoter, the GFP gene, the KmATG8 gene, and the KmATG8 terminator, was amplified from pFA6a-GFP-KmAtg8-kanMX6 and integrated into the K. marxianus chromosome as reported previously (24). The plasmids for integration of GFP-KmAtg8 FG and GFP-KmAtg8 FA were constructed by QuikChange site-directed mutagenesis (Stratagene). The plasmids for expression of KmAtg proteins in S. cerevisiae cells under control of their own promoter were constructed as follows: a DNA fragment, including the KmATG promoter, the KmATG gene, and the KmATG terminator (from 1000-bp upstream region of the initiation codon to 250-bp downstream region of the termination codon of the KmATG gene), was amplified from genomic DNA of K. marxianus and cloned into pRS316 (25) by using an In-Fusion cloning kit (Clontech). The plasmids for expression of KmAtg proteins in S. cerevisiae cells under control of ScTDH3 (GPD) promoter were constructed as follows: a DNA fragment including the KmATG gene was amplified from genomic DNA of K. marxianus and cloned into pRS316-GPDpro-PGKterm by using an In-Fusion cloning kit (Clontech).
Electron Microscopy-Cells were sandwiched between copper grids and rapidly frozen in liquid propane (Ϫ175°C) using Leica EM CPC (Leica), followed by substitution fixation in 2% osmium tetroxide dissolved in acetone containing 3% distilled water. Specimens were embedded in Quetol-651, sectioned, and observed with a transmission electron microscope (H-7500, Hitachi).
Preparation of Recombinant Proteins-ScAtg3, ScAtg7, and ScAtg8 were prepared as described previously (17). ScAtg10 was prepared as described previously (28). KmAtg7 and KmAtg10 were prepared as described previously (29). KmAtg3 and KmAtg8 were prepared as described previously (30). For NMR spectrometry of ScAtg10 and KmAtg10, 15 N-labeled proteins were prepared by growing Escherichia coli cells in M9 medium using 15 NH 4 Cl as the sole nitrogen source.
Differential Scanning Fluorimetry-Recombinant proteins (20 M) in 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM DTT supplemented with 50,000th volume of SYPRO Orange (Invitrogen) were heated from 25 to 95°C with a heating rate of 1°C/min. The fluorescence intensities were measured using an Mx3005P Real Time quantitative PCR system (Agilent Technologies) with excitation at 490 nm and emission at 530 nm. The midpoint temperature of the unfolding transition (T m ) was determined using GraphPad Prism software (GraphPad Software) from curve fitting to a Boltzmann equation (31).
NMR Spectrometry-NMR experiments were carried out at 298 K on a Varian UNITY INOVA 600 spectrometer. The sample solution of the 15 N-labeled ScAtg10 in 20 mM Tris-HCl, pH 7.5, 10 mM DTT, and the 15 N-labeled KmAtg10 in 20 mM phosphate buffer, pH 6.8, 100 mM NaCl were prepared for measurements of 1 H-15 N HSQC spectra. The NMR spectra were processed by NMRpipe (32), and data analysis was conducted using the Sparky program (33).

Bioinformatic Analyses of the Comprehensive Genome
Sequence of the Novel Thermotolerant Yeast Strain K. marxianus DMKU3-1042-One prospective approach to further understanding autophagy is utilization of novel organisms with advantageous features for investigating the molecular functions of Atg proteins. Here, we focused on a newly isolated thermotolerant yeast strain, K. marxianus (Km) DMKU3-1042, which can grow at temperatures up to 49°C (21) and whose genome sequence was determined recently (34). By bioinformatic approaches, we compared all of the open reading frames (ORFs) of the thermotolerant yeast K. marxianus and the model yeast S. cerevisiae. At first, we analyzed overall sequence alignments between K. marxianus and S. cerevisiae proteins (3,355 ORF pairs from 4,564 K. marxianus ORFs and 5,882 S. cerevisiae ORFs) using the BLAST and ClustalW web servers (Fig. 1A). These analyses showed that housekeeping proteins such as ribosomal proteins, mitochondrial proteins, and proteins involved in metabolism and nutrient utilization are nearly of the same length in S. cerevisiae and K. marxianus ( Fig. 1A and supplemental Table 2), potentially due to the fundamental importance of these proteins in cell proliferation. By contrast, several subgroups of K. marxianus proteins, annotated as involved in protein degradation, chromosome segregation, and morphogenesis, had significantly shortened primary sequences ( Fig. 1A and supplemental Table 2), potentially reflecting genetic diversity caused by adaptation to high temperature conditions, implying that these shortened K. marxianus proteins could have higher thermostability as compared with the S. cerevisiae counterparts.
Bioinformatic analyses also showed that almost half of the K. marxianus proteins (48.7%, 1,635 ORFs) are shorter than their S. cerevisiae counterparts (Fig. 1B and supplemental Table  2), and the rest are nearly the same length (24.7%, 828 ORFs) or longer (26.6%, 892 ORFs). The total numbers of the comparable amino acid residues were 1,777,544 from K. marxianus ORFs and 1,799,483 from S. cerevisiae ORFs (supplemental Table 2), indicating that K. marxianus proteins are on average 1.22% shorter than S. cerevisiae proteins.
Next, we analyzed the predicted unfoldability of K. marxianus and S. cerevisiae proteins using the FoldIndex web server (35). FoldIndex scores calculated from the amino acid sequences of K. marxianus and S. cerevisiae (3,355 ORF pairs) revealed that K. marxianus proteins are relatively less disordered than their S. cerevisiae counterparts ( Fig. 1C and supplemental Table 3); the average scores of K. marxianus and S. cerevisiae ORFs were 0.0969 and 0.0934, respectively ( Fig.  1C, p ϭ 7.14E-8). In particular, the K. marxianus proteins that are shorter than S. cerevisiae counterparts are significantly less disordered (Fig. 1D, left, p ϭ 5.08E-14). Additionally, K. marxianus proteins that are nearly the same length as their S. cerevisiae counterparts are also relatively less disordered (Fig. 1D, middle, p ϭ 1.26E-6). Taken together, we found that almost half of the K. marxianus proteins are shorter in length and have a more ordered secondary structure than their S. cerevisiae counterparts, which might contribute to the superior thermotolerance of K. marxianus.
Identification of Atg Homologs in K. marxianus-Homology searches of the K. marxianus genome sequence identified a complete set of core KmAtg proteins by their sequence similarity to those of S. cerevisiae (Sc). As shown in Fig 1A). These features of KmAtg proteins strongly raised the possibility that KmAtg proteins has superior thermostability as compared with the ScAtg counterparts. In addition to KmAtg proteins, we identified several vacuolar enzymes frequently used in the autophagy field (Pep4, Prb1, Pho8, and Ape1) ( Fig.  2A). Therefore, we first tried to develop experimental methods to assess autophagy in K. marxianus cells.
Probes for Monitoring Autophagy in K. marxianus Cells-One of the easiest ways to evaluate the progression of autophagy is to monitor intravacuolar accumulation of autophagy-related structures called autophagic bodies (36). Thus, we performed several morphological analyses, including electron microscopy and fluorescence microscopy, to examine autophagy in K. marxianus cells. To observe accumulation of autophagic bodies, we deleted the KmPEP4 gene from K. marxianus cells; ScPep4 is a putative master enzyme required for activation of the vacuolar hydrolases responsible for degradation of autophagic bodies (36), and KmPep4 is highly conserved in K. marxianus as a single gene product ( Fig. 2A, 73.8% iden-tical to ScPep4). Electron microscopy revealed that in Kmpep4⌬ cells treated with the autophagy-inducing drug rapamycin, a large number of autophagic bodies accumulated in the vacuolar lumen (Fig. 2B, middle panel). This observation indicates that upon rapamycin treatment, autophagy was efficiently induced in K. marxianus cells. Next, we deleted the KmATG8 gene to determine whether KmAtg8 contributes to autophagosome formation, because Atg8 and its mammalian homolog LC3 have been widely used as an autophagosomal marker (37,38); furthermore, among the KmAtg homologs, KmAtg8 has the highest identity to its S. cerevisiae counterpart ( Fig. 2A, 85.3% identical to ScAtg8). As expected, electron microscopy revealed that no autophagic bodies formed in the absence of KmAtg8 (Fig.  2B, right panel), confirming that KmAtg8 is involved in autophagosome formation in K. marxianus. We next developed an alkaline phosphatase (ALP) assay, a method for quantitatively assessing autophagic activity that was originally established in S. cerevisiae (39). The vacuolar alkaline phosphatase KmPho8 is also conserved as a single gene product in K. marxianus ( Fig. 2A, 58.1% identical to ScPho8). A truncated form of KmPho8 that lacks its signal sequence, KmPho8⌬N44, is expressed in the cytoplasm, and its transport into the vacuole as an autophagosomal content and subsequent activation by vacuolar hydrolases can be quantitatively measured. In wild-type cells, ALP activity increased significantly upon nutrient starvation (Fig. 2C), and this increase was strictly dependent on the presence of the ALP probe KmPho8⌬N44 and the vacuolar hydrolases KmPep4 and KmPrb1 (Fig. 2C, left panel). By using FIGURE 1. Bioinformatic analyses of the comprehensive genome sequences of K. marxianus and S. cerevisiae. A, comparable ORFs aligned using the BLAST server were classified into three subgroups according to their relative lengths (see also supplemental Table 2). The comparable ORFs (total 3,355 ORF pairs) were annotated according to their functions. Asterisks, significantly fewer than in total ORFs (p Ͻ 0.005); triple asterisks, significantly more than in total ORFs (p Ͻ 0.005). B, almost half of all K. marxianus proteins are shorter than their S. cerevisiae counterparts. Km Ͻ Sc, the K. marxianus proteins are Ͼ0.5% shorter than the S. cerevisiae counterparts (1,635 ORFs in blue); Km ϭ Sc, the K. marxianus proteins are nearly the same length as the S. cerevisiae counterparts (828 ORFs in red); Km Ͼ Sc, the K. marxianus proteins are Ͼ0.5% longer than the S. cerevisiae counterparts (892 ORFs in green). Identities of the comparable ORFs (3,355 ORF pairs) from A were calculated using the LALIGN server. Asterisk, mature form of Atg8. C, unfoldability scores of all comparable K. marxianus and S. cerevisiae ORFs (3,355 ORF pairs) were assessed using the FoldIndex server (see also supplemental Table 3). Mean FoldIndex scores were analyzed statistically. D, mean FoldIndex scores of three subgroups in B were analyzed as in C. DECEMBER 4, 2015 • VOLUME 290 • NUMBER 49 FIGURE 2. Identification of Atg homologs and probes for monitoring autophagy in K. marxianus. A, Atg homologs in K. marxianus. Most KmAtg proteins are shorter than their ScAtg counterparts (red). Asterisk, mature form of Atg8. B, KmAtg8 is required for autophagosome formation in K. marxianus. K. marxianus wild-type (WT), Kmpep4⌬, and Kmpep4⌬ Kmatg8⌬ cells were grown at 30°C, treated with rapamycin for 4 h, and subjected to electron microscopy. C, KmAtg8 is essential for K. marxianus autophagy. K. marxianus wild-type (WT), Kmpep4⌬ Kmprb1⌬, and Kmatg8⌬ cells expressing an N-terminally truncated variant of KmPho8 (KmPho8⌬N44) were grown at 30°C and then shifted to nitrogen-starvation medium. Kmpho8⌬ cells were used as a control not expressing the probe KmPho8⌬N44 (without KmPho8⌬N44). After starvation for 5 h, the cells were harvested, and ALP activities were measured. a.u., arbitrary unit. D, GFP-KmAtg8 is transported into the vacuole. Wild-type K. marxianus cells expressing GFP-KmAtg8 (green) were treated with FM4-64 (magenta). The cells were grown at 30°C (Nut.) and then treated with rapamycin for 2 h (Rap. 2 h) or incubated in nitrogen-starvation medium for 2 h (Stv. 2 h). The cells were observed by fluorescence microscopy. E, C-terminal regions of ScAtg8 and KmAtg8. Atg8 is synthesized as a precursor form with an additional segment at its C terminus (blue) and cleaved by Atg4 to be converted into the mature form. KmAtg8 FG is a truncated variant that mimics a mature form. KmAtg8 FA contains an Ala substitution at residue 116 (red). F, KmAtg4 is responsible for the maturation of KmAtg8. GFP-KmAtg8, GFP-KmAtg8 FG , and GFP-KmAtg8 FA (green) were expressed in K. marxianus wild-type and Kmatg4⌬ cells. The cells were treated with FM4-64 (magenta), treated with rapamycin for 2 h, and then observed by fluorescence microscopy. G, GFP-processing assay in K. marxianus. The K. marxianus cells used in F were treated with rapamycin for 1 h, and then total lysates were prepared. Samples were subjected to immunoblot analysis using anti-GFP antibody and streptavidin-HRP as a loading control (biotinylated). proc.GFP indicates a processed form of the GFP moiety. this assay, we measured autophagic activity in Kmatg8⌬ cells. In the absence of KmAtg8, no autophagic activity was detected (Fig. 2C, right panel), suggesting that KmAtg8 is essential for K. marxianus autophagy. From these results, we concluded that the ALP assay is applicable for K. marxianus studies.

Utilization of K. marxianus for Autophagy Research
Atg8 (LC3 in mammals) is known to be localized on autophagosomal membranes (37,38). Hence, we constructed K. marxianus cells expressing GFP-KmAtg8 under the control of the KmATG8's own promoter. Fluorescence microscopy revealed that after rapamycin treatment (Fig. 2D, middle panel) or nutrient starvation (Fig. 2D, right panel), a significant portion of GFP-KmAtg8 was transported into the vacuole. These observations clearly indicate that GFP-KmAtg8 can be used as a probe to assess the progression of autophagy in K. marxianus cells.
In S. cerevisiae cells, the ubiquitin-like protein ScAtg8 is synthesized as a precursor form with an additional Arg residue at its C terminus; the Arg residue is subsequently cleaved by the cysteine protease ScAtg4 (40,41), resulting in conversion of ScAtg8 into a 116-residue mature form in which the C-terminal Gly residue is exposed (Fig. 2E). KmAtg8 is also predicted to be synthesized as a precursor form with an additional 8-residue sequence, which is cleaved by KmAtg4 to generate the 116residue mature form (Fig. 2E). To confirm the involvement of KmAtg4 in the cleavage of KmAtg8, we constructed a truncated form of KmAtg8 in which the C-terminal Gly residue (KmAtg8 FG ) is exposed, which is predicted to be functional as the mature form (Fig. 2E). In the absence of KmAtg4, wild-type GFP-KmAtg8 was not transported into the vacuolar lumen but instead dispersed in the cytoplasm (Fig. 2F, lower left panel). By contrast, GFP-KmAtg8 FG was transported to some extent into the vacuolar lumen even in the absence of KmAtg4 (Fig. 2F, lower middle panel), suggesting that KmAtg8 FG bypassed the requirement for cleavage by KmAtg4. We also confirmed that the GFP-KmAtg8 FA mutant (Fig. 2E), which contains an Ala substitution at residue 116, preventing its conjugation with a lipid phosphatidylethanolamine (PE) (16), was not transported into the vacuolar lumen even in wild-type cells (Fig. 2F, right upper panels). From these results, we conclude that KmAtg4 is responsible for the maturation of KmAtg8.
We next examined the levels of GFP-KmAtg8 in the vacuole by immunoblot analysis. In wild-type cells, GFP-KmAtg8 was efficiently transported into the vacuole and processed to yield an ϳ28-kDa GFP fragment (Fig. 2G, lane 1), because the GFP moiety is resistant to vacuolar proteases (42). By contrast, no GFP fragment was detected in Kmatg4⌬ cells (Fig. 2G, lane 4), indicating that GFP-KmAtg8 was not transported into the vacuole in the absence of KmAtg4. In addition, GFP-KmAtg8 FG yielded the GFP fragment efficiently in wild-type cells and at somewhat lower levels in Kmatg4⌬ cells (Fig. 2G, lanes 2 and 5), whereas GFP-KmAtg8 FA yielded no GFP fragment even in wild-type cells (Fig. 2G, lanes 3 and 6). These results suggest that KmAtg8 and KmAtg4 function in a similar manner to their S. cerevisiae counterparts. Furthermore, the levels of the processed GFP fragment detected in the immunoblot analysis (Fig.  2G) were well correlated with the levels of intra-vacuolar GFP-KmAtg8 observed by fluorescence microscopy (Fig. 2F). These observations indicate that GFP-KmAtg8 can be used as a probe to monitor K. marxianus autophagy in both fluorescence microscopy and immunoblot analysis.
KmAtg Proteins Are Involved in Autophagosome Formation in K. marxianus Cells-Using GFP-KmAtg8 as a probe, we examined the involvement of other KmAtg proteins in autophagy. For this purpose, we constructed GFP-KmAtg8 strains lacking each KmATG gene. Fluorescence microscopy revealed that in the absence of core KmAtg homologs, the vacuolar transports of GFP-KmAtg8 were not observed at all (Fig.  3A), indicating that these KmAtg proteins are certainly involved in autophagy in K. marxianus. Their phenotypes could be classified into two groups. In cells depleted of KmAtg1, KmAtg2, KmAtg6, KmAtg9, KmAtg13, KmAtg14, or KmAtg18, GFP-KmAtg8 formed a punctate structure in proximity to the vacuolar membrane, which appeared to be the preautophagosomal structure (PAS) observed in S. cerevisiae cells (43). In contrast, the bright PAS puncta were barely detectable in cells depleted of KmAtg3, KmAtg4, KmAtg5, KmAtg7, KmAtg10, KmAtg12, or KmAtg16, all of which are members of the two ubiquitin-like conjugation systems (Fig. 3, A and B, green bars). These observations were consistent with those in S. cerevisiae, in which PAS assembly of GFP-ScAtg8 strictly requires the ubiquitin-like conjugation systems (44). We also observed that in addition to the bright PAS puncta, small GFP-KmAtg8-positive dots were observed in K. marxianus cells defective of the two ubiquitin-like conjugation systems (Fig. 3B, gray bars), which may imply additional functions of KmAtg8.
Fluorescence microscopy also revealed that KmAtg11, KmAtg17, KmAtg29, and KmAtg31 were not essential for the vacuolar transport of GFP-KmAtg8 (Fig. 3C, upper panels). Immunoblot analysis also showed that the GFP fragment derived from GFP-KmAtg8 was detected (although at somewhat reduced levels) in the absence of KmAtg11, KmAtg17, KmAtg29, or KmAtg31 (Fig. 3D, lanes 10 -13), indicating that GFP-KmAtg8 was at least partially transported into the vacuole in these cells. Previous studies using S. cerevisiae have shown that ScAtg17, ScAtg29, and ScAtg31 form a ternary complex that organizes the PAS scaffold responsible for starvation-induced autophagy (45) and that ScAtg11 plays a similar role to the ScAtg17-ScAtg29-ScAtg31 complex under nutrient-rich conditions (42). Therefore, we constructed the double-deletion mutants Kmatg11⌬ Kmatg17⌬, Kmatg11⌬ Kmatg29⌬, and Kmatg11⌬ Kmatg31⌬. In all of the double mutants, GFP-KmAtg8 was not transported into the vacuole but was largely dispersed in the cytoplasm (Fig. 3C, lower panels). These results suggest that KmAtg11 has a redundant function with KmAtg17, KmAtg29, and KmAtg31 in K. marxianus cells. Previous S. cerevisiae studies showed that ScAtg11 is involved in the biosynthesis of the vacuolar aminopeptidase Ape1 under nutrient-rich conditions (46,47). Ape1 is synthesized in the cytoplasm as a precursor form (prApe1) and subsequently transported into the vacuole via selective autophagy, called the cytoplasm-to-vacuole targeting pathway, to be processed into the mature form (mApe1). In K. marxianus cells, prKmApe1 was barely converted to the mature form in the absence of KmAtg11 under nutrient-rich conditions (Fig. 3D, lane 2), suggesting that KmAtg11 functions in the cytoplasm-to-vacuole targeting pathway in K. marxianus cells. Intriguingly, as judged DECEMBER 4, 2015 • VOLUME 290 • NUMBER 49 JOURNAL OF BIOLOGICAL CHEMISTRY 29511 by a reduction in the level of the processed GFP fragment (Fig.  3D, lane 10), KmAtg11 is required not only for selective autophagy under nutrient-rich conditions but also for starvation-induced bulk autophagy (Fig. 3D, lane 10) to nearly the same extent as KmAtg17, KmAtg29, and KmAtg31 (Fig. 3D,  lanes 11-13). These results suggest that although ScAtg11 is specifically involved in selective autophagy in S. cerevisiae, KmAtg11 plays an important role in bulk autophagy even under starvation conditions in K. marxianus. Taken together, these data indicate that as well as the core KmAtg proteins, KmAtg11, KmAtg17, KmAtg29, and KmAtg31 are also involved in bulk autophagy in K. marxianus.

Utilization of K. marxianus for Autophagy Research
Autophagic Activity Is Enhanced under High Temperature Conditions in K. marxianus but Reduced in S. cerevisiae-We next investigated whether high temperature stress induces autophagy in K. marxianus and S. cerevisiae cells. Under nutrient-rich conditions, neither the cytoplasm-to-vacuole targeting pathway nor autophagy was induced by heat stress at 37 or 42°C in both K. marxianus and S. cerevisiae cells. 6 By combination of rapamycin treatment and heat stress, GFP-KmAtg8 efficiently yielded GFP fragments (Fig. 3E, lanes 2 and 3), indicating that autophagic activity is enhanced by heat stress in K. marxianus. By contrast, the vacuolar transport of GFP-ScAtg8 was reduced by heat stress in S. cerevisiae (Fig. 3E, lanes  8 and 9), and GFP-ScAtg8 highly accumulated at the perivacuolar sites under high temperature conditions (Fig. 3F). These observations suggest that ScAtg proteins were somewhat inactivated by heat stress, which also represents clear advance of K. marxianus in elucidation of mechanisms of autophagy induction against the high temperature stress.
KmAtg Proteins Complement the Functions of ScAtg Proteins in S. cerevisiae-We next investigated to what extent KmAtg proteins suppress autophagic defects caused by the depletion of ScAtg proteins in S. cerevisiae. To this end, we examined the maturation of prApe1 in rapamycin-treated cells, allowing highly sensitive assessment of autophagosome formation. In the absence of core ScAtg proteins, prApe1 was not transported into the vacuole and therefore not converted to mApe1 (Fig. 4A, except for Scatg13⌬ cells because of their partial phenotype). Upon exogenous expression of KmAtg proteins under the control of their own promoters (ϳ1,000-bp upstream region of each KmATG gene), prApe1 was modestly converted to mApe1 in most cases (Fig. 4A, lanes 3, 5, 7 , 9, 11, 13, 15, 18, 20, and 24), suggesting that these KmAtg proteins can at least partly complement deletion mutants of their ScAtg counterparts. In the cases of KmAtg5, KmAtg7, KmAtg10, KmAtg12, and KmAtg16, mApe1 was not detected at all (Fig. 4A, lanes 22, 26,   28, 30, and 32), probably due to insufficient expression from the KmATG promoters in S. cerevisiae cells. Therefore, we expressed these KmAtg proteins under control of the GPD promoter (Fig. 4B). Under these expression conditions, KmAtg7, KmAtg10, and KmAtg12 suppressed defects in maturation of prApe1 (Fig. 4B, lanes 7, 10, and 13), suggesting that these proteins were functional in S. cerevisiae. By contrast, as judged by the maturation of prApe1, KmAtg5 and KmAtg16 failed to complement the functions of their S. cerevisiae counterparts. However, Scatg5⌬ cells expressing KmAtg5 yielded an extra band detected by anti-ScAtg12 antibodies (Fig. 4B, lane 4). This extra band migrated faster than that of the ScAtg12-ScAtg5 conjugate, one of the key products of the ubiquitin-like conjugation reaction (48). Based on the molecular weights of ScAtg5 (294 residues, 33.6 kDa) and KmAtg5 (271 residues, 31.2 kDa), it is conceivable that this extra band corresponds to the ScAtg12-KmAtg5 conjugate, the putative product of heterogeneous conjugation. Because ScAtg5 interacts directly with ScAtg16 in S. cerevisiae, we predicted that the ScAtg12-KmAtg5 conjugate would become functional in the presence of cognate KmAtg16 (49,50). As expected, co-expression of KmAtg5 and KmAtg16 led to efficient conversion of prApe1 into mApe1 (Fig. 4C, lanes 4 and 8), indicating that the proper interaction between KmAtg5 and KmAtg16 is required for their functions. Based on the results of these complementation assays, which demonstrated the functions of KmAtg proteins, we conclude that the fundamental molecular mechanisms underlying autophagosome formation are conserved between S. cerevisiae and K. marxianus.
KmAtg Proteins Are Thermostable Relative to ScAtg Proteins-As compared with the ScAtg proteins, most of the KmAtg proteins are relatively short in length ( Fig. 2A); sequence alignments between ScAtg and KmAtg proteins revealed that most ScAtg proteins contain several insertions (supplemental Figs. S1-S4). Thus, we assessed the thermostability of the KmAtg proteins. To date, four recombinant Atg proteins derived from both S. cerevisiae and K. marxianus cells (Atg3, Atg7, Atg8, and Atg10) have been prepared efficiently. Hence, we assessed the thermostability of these recombinant proteins by differential scanning fluorimetry (DSF). As compared with the ScAtg homologs, KmAtg3, KmAtg7, and KmAtg10 had relatively high T m values (i.e. the midpoint temperature at which they unfold). These data indicate that KmAtg3, KmAtg7, and KmAtg10 have superior thermostability relative to their S. cerevisiae counterparts (Fig. 5A).
We also analyzed their thermostability using an in vitro ubiquitin-like conjugation assay, a well established reconstitution system consisting of Atg7 (E1 enzyme), Atg3 (E2 enzyme), Atg8 (ubiquitin-like protein), and PE-containing liposomes (16).   1 h), and then total lysates were prepared. The samples were subjected to immunoblot analysis using anti-GFP antibody, anti-ScApe1 antibody (cross-reacted with KmApe1), and streptavidin-HRP as a loading control (biotinylated). proc.GFP, prKmApe1, and mKmApe1 indicate the processed form of the GFP moiety, the precursor form of KmApe1, and the mature form of KmApe1, respectively. CBB, Coomassie Brilliant Blue. E, GFP-processing assay under high temperature conditions. K. marxianus cells expressing GFP-KmAtg8 and S. cerevisiae cells expressing GFP-ScAtg8 were grown at 30°C, shifted to 37 or 42°C, and treated with rapamycin for 1 h. F, K. marxianus cells used in E were observed by fluorescence microscopy. DECEMBER 4, 2015 • VOLUME 290 • NUMBER 49

Utilization of K. marxianus for Autophagy Research
When incubated at 30°C, both ScAtg8 and KmAtg8 were efficiently conjugated to PE in a time-dependent manner (Fig. 5B). However, when ScAtg3 was preincubated at 60°C, ScAtg8 was not conjugated to PE (Fig. 5B, left panel), suggesting that ScAtg3 was inactivated by the heat treatment. By contrast, KmAtg3 retained its E2 enzyme activity to some extent even after heat treatment at 60°C (Fig. 5B, right panel). These results suggest that KmAtg3 is more thermostable than ScAtg3, consistent with the differences in their T m values obtained by DSF analysis (Fig. 5A).
KmAtg Proteins Are Highly Soluble Relative to ScAtg Proteins-To further investigate intrinsic features of KmAtg proteins, we prepared six pairs of ScAtg and KmAtg proteins (Atg3, Atg5, Atg7, Atg8, Atg10, and Atg12, all of which are components of the ubiquitin-like conjugation systems) by using the PUREfrex cell-free translation system (GeneFrontier), not containing other intracellular components such as heat shock chaperones. Both ScAtg and KmAtg proteins were efficiently produced; however, ScAtg5, ScAtg7, and ScAtg12 were poorly recovered after high speed centrifugation, suggesting that they aggregated under these conditions (Fig. 5C). In contrast to the ScAtg proteins, the KmAtg proteins except for KmAtg7 were efficiently recovered in the supernatant fraction (Fig. 5C), implying that the KmAtg proteins used in this assay were more soluble than the corresponding ScAtg proteins.
KmAtg10 Exhibits a High Resolution NMR Spectrum-Recently, we reported the solution structure of KmAtg10, determined by NMR spectrometry (29). During our attempt to determine the KmAtg10 structure, we also prepared recombinant ScAtg10; however, the NMR spectrum showed that ScAtg10 was somewhat aggregated and therefore not suitable for structural determination (Fig. 5D, left panel). By contrast, a high resolution NMR spectrum could be obtained from KmAtg10 (Fig. 5D, right panel). Sequence alignment between KmAtg10 and ScAtg10 revealed that ScAtg10 contains an extra 13-residue segment located inside the four-stranded ␤-sheet of the KmAtg10 structure (supplemental Fig. S2E). We assumed that the relatively short length of the KmAtg10 sequence might contribute to high resolution of its NMR spectrum (Fig. 5D, right panel), as well as its improved thermostability as assessed by DSF analysis (Fig. 5A), allowing the practical determination of its structure (29). In addition to KmAtg10, we succeeded in determining the structures of other KmAtg proteins, including KmAtg5 (29), the KmAtg18 homolog KmHsv2 (51), the KmAtg7-KmAtg10 complex (30), and the KmAtg1-KmAtg13 complex (52). Taken together with our recent progress in structural biology, these findings indicate that KmAtg proteins, which are more thermostable and soluble than their ScAtg counterparts, are suitable for biochemical and structural studies.

Discussion
In this study, with the goal of expanding the scope of autophagy research, we demonstrated that the newly isolated thermotolerant yeast strain K. marxianus DMKU3-1042 (21) represents a novel experimental system with thermostable Atg proteins. We first identified a complete set of KmAtg proteins essential for autophagosome formation in K. marxianus (Fig.  3), most of which can, at least in part, functionally substitute their counterpart ScAtg proteins in S. cerevisiae (Fig. 5). These findings showed that the basal molecular mechanisms underlying autophagosome formation are conserved between these two species. Sequence alignments and bioinformatic analyses showed that most KmAtg proteins are apparently shorter than DECEMBER 4, 2015 • VOLUME 290 • NUMBER 49 their S. cerevisiae counterparts ( Fig. 2A). Furthermore, Atg proteins are highly diverse between these two species as compared with other proteins (most KmAtg proteins have relatively low identity; located in the left side of the blue box indicating Km Ͻ Sc in Fig. 1B). Our observations suggest that K. marxianus will be useful for studies of autophagy. First, similar to S. cerevisiae cells, autophagy was efficiently induced in K. marxianus cells by nutrient starvation or rapamycin treatment. Second, because the fundamental molecular mechanisms underlying autophagosome formation are conserved between S. cerevisiae and K. marxianus, the novel insights obtained in K. marxianus studies, such as structural information, will be directly applicable to in vivo analysis using the well characterized model yeast S. cerevisiae, and vice versa. Third, K. marxianus cells grow rapidly with doubling times of 45-60 min at 37°C and reach much higher density than S. cerevisiae cells (53), allowing us to perform experiments rapidly and efficiently. Fourth, standard protocols for K. marxianus genetics, such as knock-in and knockout techniques, have been established (24,54,55). Fifth, autophagy is modestly induced in K. marxianus cells even at temperatures above 47°C. 6 Based on these features, K. marxianus is thought to be applicable for studies aimed at elucidating the molecular functions of Atg proteins by in vitro analyses, as well as the physiological roles of autophagy in vivo, including cellular quality control and stress response to high temperature. In fact, under the high temperature conditions at 37 or 42°C, while autophagic activity was reduced in S. cerevisiae cells, autophagic activity was significantly enhanced in K. marxianus cells (Fig. 3E).

Utilization of K. marxianus for Autophagy Research
We also observed some differences between S. cerevisiae and K. marxianus. Whereas ScAtg11 mostly functions in selective autophagy under nutrient-rich conditions in S. cerevisiae, KmAtg11 is required for both selective autophagy and starvation-induced autophagy in K. marxianus (Fig. 3C). One possible explanation for this is that during evolution, KmAtg11 might have acquired an additional function related to the basal mechanisms of autophagosome formation under starvation conditions. Alternatively, ScAtg11 might have become specific for selective autophagy in S. cerevisiae cells. We also found that although either ScAtg11 or ScAtg17 is crucial for PAS formation in S. cerevisiae cells (44), PAS assembly of GFP-KmAtg8 was observed even in the absence of the scaffold proteins in K. marxianus (Fig. 3B, green bars, Kmatg11⌬ Kmatg17⌬, Kmatg11⌬ Kmatg29⌬, and Kmatg11⌬ Kmatg31⌬), which was strictly dependent on two ubiquitin-like conjugation systems (Fig. 3B, green bars, Kmatg3⌬, Kmatg4⌬, Kmatg5⌬, Kmatg7⌬, Kmatg10⌬, and Kmatg16⌬). These observations suggest that in K. marxianus cells, GFP-KmAtg8 can assemble to form the PAS-like puncta, irrespective of scaffold proteins such as KmAtg11 and KmAtg17. In mammals, neither Atg11 nor Atg17 is conserved, in which FIP200 plays a key role in the initial step of autophagosome formation (56,57). The differences between S. cerevisiae and K. marxianus cells would help us to further understand the common mechanisms underlying the initial step of autophagosome formation, which involves divergent scaffold proteins, including Atg11, Atg17, and FIP200. Alternatively, KmAtg8 could have additional functions other than autophagosome formation in K. marxianus cells. We found that even in the absence of the two ubiquitin-like conjugation systems, GFP-KmAtg8 formed several small dots in proximity to the vacuolar membrane (Fig. 3C, gray bars). A previous study has reported that in Pichia pastoris Atg8 is involved in not only autophagy but also in vacuolar membrane dynamics in a lipidation-independent manner (58). KmAtg8 may also have a role in vacuolar morphogenesis or other biological events in a lipidation-independent manner.
From the standpoints of structural and biochemical studies, the most important property of K. marxianus is its thermotolerance. Our in vitro analyses revealed that several KmAtg proteins have higher thermostability and solubility than ScAtg proteins (Fig. 4), potentially due to their shorter primary sequences ( Fig. 2A). By using these thermostable recombinant proteins, we have actually succeeded in obtaining structural information about KmAtg1, KmAtg5, KmAtg7, KmAtg10, KmAtg13, and the KmAtg18 homolog KmHsv2 (29,30,51,52). Taken together, our findings clearly indicate that K. marxianus has the advantages of thermostability and solubility, making it an especially suitable model organism for structural and biochemical studies. Recently, thermotolerant and thermophilic novel organisms have begun to be used in structural biology. For example, Amlacher et al. (20) uncovered molecular details of nuclear pore complexes by using the thermophilic fungus C. thermophilum. In the autophagy field, Ragusa et al. (59) succeeded in determining the structure of the Atg17-Atg31-Atg29 ternary complex derived from the thermotolerant yeast Lachancea thermotolerans. The results of these studies are consistent with the interpretation that thermotolerant or thermophilic organisms are useful for structural and biochemical studies. Currently, structural data about several Atg proteins, such as Atg2, Atg9, Atg11, and Atg14, have not been obtained because of technical difficulty of preparing these proteins. We expect that utilization of K. marxianus will help us to obtain structural information about these Atg proteins and will facilitate the elucidation of the molecular mechanisms underlying autophagosome formation.
Our bioinformatic analyses showed that, in addition to the KmAtg proteins, almost half of K. marxianus proteins are shorter and more ordered than their S. cerevisiae counterparts FIGURE 5. KmAtg proteins are more thermostable and soluble than ScAtg proteins. A, KmAtg7, KmAtg3, and KmAtg10 are more thermostable than ScAtg7, ScAtg3, and ScAtg10, respectively. Recombinant Atg proteins (Atg7, Atg3, Atg8 FG , and Atg10) derived from K. marxianus (red) and S. cerevisiae (black) were subjected to DSF analysis. B, KmAtg3 is more thermostable than ScAtg3. Recombinant Atg proteins (0.22 M Atg7, 0.22 M Atg3, and 5 M Atg8 FG ) derived from S. cerevisiae (left panel) and K. marxianus (right panel) were subjected to in vitro PE-conjugation assay (350 mM PE-containing liposome). Before conjugation reaction, ScAtg3 and KmAtg3 were preincubated at 30, 55, or 60°C for 90 min, and the conjugation reaction was performed at 30°C for 10, 30, 60, and 90 min. The samples were subjected to urea-containing SDS-PAGE followed by Coomassie Brilliant Blue staining. The PE-conjugated form of Atg8 (Atg8-PE) was quantitated. Total amounts of Atg8 were defined as 100%. RfU, relative fluorescence unit. C, strep-tagged Atg proteins (Atg7, Atg3, Atg8 FG , Atg10, Atg12, and Atg5) derived from K. marxianus and S. cerevisiae were expressed at 37°C for 4 h by using the PUREfrex cell-free translation kit (GeneFrontier). After the translation, the total reaction mixtures (T) were centrifuged at 78,000 ϫ g for 30 min, and supernatants were prepared (S 78 ). The samples were subjected to immunoblot analysis using anti-Strep antibody. D, NMR spectra of ScAtg10 (left panel) and KmAtg10 (right panel). (Fig. 1). As general features of the K. marxianus proteome, the shortened primary sequences and ordered secondary structures may explain the superior thermotolerance of this organism. We hope that utilization of K. marxianus will expand the range of structural biology and biochemistry by providing stable recombinant proteins for a multitude of applications.