Formation of the (cid:1) 350-kDa Apg12-Apg5 (cid:1) Apg16 Multimeric Complex, Mediated by Apg16 Oligomerization, Is Essential for Autophagy in Yeast*

Autophagy, responsible for the delivery of cytoplasmic components to the lysosome/vacuole for degradation, is the major degradative pathway in eukaryotic cells. This process requires a ubiquitin-like protein conjugation system, in which Apg12 is covalently bound to Apg5. In the yeast Saccharomyces cerevisiae , the Apg12-Apg5 conjugate further interacts with a small coiled-coil protein, Apg16. The Apg12-Apg5 and Apg16 are localized in the cytosol and pre-autophagosomal structures and play an essential role in autophagosome formation. Here we show that the Apg12-Apg5 conjugate and Apg16 form a (cid:1) 350-kDa complex in the cytosol. Because Apg16 was suggested to form a homo-oligomer, we generated an in vivo system that allowed us to control the oligomerization state of Apg16. With this system, we demonstrated that formation of the (cid:1) 350-kDa pGEX-3X-Apg16, and pGEX-2T-Apg5 plasmids were transformed into Escherichia coli XL1 Blue MRF (cid:4) , and transformants were grown in LB medium containing 50 (cid:2) g/ml ampicillin to 0.5 A 600 . Recombinant protein expression was in- duced with 1 m M isopropyl- (cid:3) - D -thiogalactopyranoside during an additional 3-h culture at 37 °C. The recombinant proteins in the inclusion bodies were separated by SDS-PAGE and simultaneously stained with Coomassie Brilliant Blue G-250. The protein bands were cut off and eluted with elution buffer (100 m M Tris, pH 6.8, 0.05% SDS). The eluted proteins were used to immunize rabbits. Antiserum against Apg16 was subsequently affinity purified by passing serum over a column of His 6 tagged Apg16 recombinant protein coupled to CNBr-activated Sepha- rose 4B (Amersham Biosciences). For expression of His 6 -tagged Apg16, the pDEST17-Apg16 plasmid was transformed into BL21-SI competent cells (Invitrogen), and transformants were grown up to 0.5 A 600 at 30 °C in LB without NaCl medium (1% Bactotryptone, 0.5% yeast extract) containing 50 (cid:2) g/ml ampicillin, and expression of His 6 -Apg16 protein was induced for 2 h with 0.3 M NaCl. His 6 -tagged Apg16 was purified from inclusion bodies as described above. The purified His-Apg16 protein was covalently coupled to CNBr-activated Sepharose and incu- bated with antiserum overnight at 4 °C. After washing unbound anti-sera extensively, bound antibodies were eluted with 0.1 M glycine, pH 2.8. Eluates containing the affinity-purified at (cid:3) 80 °C. Anti-Apg5 antibody was purified likewise using GST-Apg5-conjugated

In eukaryotic cells, the majority of intracellular bulk degradation occurs in the lysosome/vacuole, an acidic compartment that contains various hydrolytic enzymes. Autophagy is the main pathway by which the cell delivers cytoplasmic components to the vacuole for degradation (1)(2)(3). In this process, cytoplasmic constituents, including organelles, are sequestered nonselectively by double-membrane structures termed autophagosomes, which subsequently fuse with the vacuole. The released inner membrane and sequestered components in the vacuolar lumen are degraded for reuse. Autophagy is a cellular survival response to starvation and plays an important role in developmental processes and cell differentiation.
By genetic screens in the yeast Saccharomyces cerevisiae, the APG and AUT genes involved in autophagy were isolated (4,5). We found previously that four of the Apg proteins constitute a novel protein conjugation system, the Apg12 system (6). In this ubiquitin-like system, the carboxyl-terminal glycine of Apg12 (molecular mass is 21 kDa) is bound covalently to Lys 149 at the center of Apg5 (33 kDa). This conjugating reaction is catalyzed by Apg7 and Apg10. Apg7 is the Apg12-activating enzyme (7)(8)(9), and Apg10 functions as an Apg12-conjugating enzyme (10). Human homologs of Apg12 and Apg5 have been identified and undergo a similar covalent linkage (11), indicating that this conjugation system is conserved in mammalian cells. Analysis of an apg5 null mutant and a temperature-sensitive mutant suggested that Apg5 is required for formation or completion of sequestering vesicles in yeast (12). Recent morphological observation revealed that in yeast, most Apg5 exists diffusely in the cytoplasm, whereas only a small portion localizes to the pre-autophagosomal structures (13). Similarly, in a study of Apg5 in mouse embryonic stem cells, it was demonstrated that the Apg12-Apg5 conjugate is targeted from the cytoplasm to the autophagic isolation membranes during autophagosome formation (14). Immediately before or after completion of autophagosome formation, the Apg12-Apg5 conjugate dissociates from the isolation membrane. The Apg12-Apg5 conjugate is required for elongation of the isolation membrane. The Apg12-Apg5 conjugate is also required for association of Aut7/LC3 with the pre-autophagosomal membranes in both yeast and mammalian cells (13)(14)(15). But, although the Apg12-Apg5 conjugate is considered to be of vital importance for autophagosome formation, its molecular function is still poorly understood.
Apg16 was originally obtained by a two-hybrid screen using Apg12 as bait and was found to interact with the Apg12-Apg5 conjugate (16). It was then determined that Apg16 interacts directly with Apg5 but not with Apg12. Apg16 is a 150-amino acid protein (17 kDa) that contains a carboxyl-terminal coiledcoil motif (residues 58 -123) and associates with Apg5 at its amino-terminal region. It was suggested that Apg16 forms an oligomer through the coiled-coil region and functions as a linker to form the Apg12-Apg5⅐Apg16 multimeric complex (16). Because Apg16 is the only molecule thus far identified to interact with the Apg12-Apg5 conjugate, further characterization of Apg16 would provide valuable insights into the molecular role of the Apg12-Apg5 conjugate in autophagosome formation. In this study, we characterize the Apg12-Apg5⅐Apg16 protein complex and show, by use of a regulated oligomerization system, that formation of the ϳ350-kDa complex is required for functioning of the Apg12-Apg5 conjugate in the autophagic pathway.
Plasmid Construction-To create the glutathione S-transferase (GST) 1 -Apg12, GST-Apg16, and GST-Apg5 fusion constructs, the open reading frames of Apg12 and Apg16 were cloned into the blunted EcoRI and SmaI sites of pGEX-3X (Amersham Biosciences) to yield pGEX-3X-Apg12 and pGEX-3X-Apg16, respectively. The Apg5 open reading frame was cloned into the BamHI site of pGEX-2T to generate pGEX-2T-Apg5. For expression of His 6 -tagged Apg16, the Apg16 open reading frame was first cloned into the EcoRI site of pENT3C from the Gateway cloning system (Invitrogen). A His 6 -Apg16 expression plasmid (pD-EST17-Apg16) was then generated according to the manufacturer's instructions. To make the Apg16 -2FKBP F36V fusion constructs, we first created BamHI sites within the carboxyl-terminal coiled-coil region of Apg16 (after the amino acid residue 65 and/or 118) by mutagenizing pApg16, a pRS314 vector containing the APG16 gene (16), using the QuikChange TM site-directed mutagenesis kit (Stratagene). The fragment containing two tandem copies of FKBP F36V was PCR amplified from the plasmid pC4-Fv2E (provided by ARIAD Pharmaceuticals, Inc.) with primers that were designed to have in-frame BamHI sites in their 5Ј-ends: Fv2E5N, 5Ј-ATCGGGATCCAGGCGTCCAAGTCGAAACCA-3Ј; and FKBP3, 5Ј-GCATGGATCCGCGTAGTCTGGTACGTCGTACG-G-3Ј. The resulting 2FKBP F36V fragment was digested with BamHI and inserted into the mutated pApg16 plasmids, either into one of the two BamHI sites or between them, to generate pApg16 65 -2FKBP F36V , pApg16 118 -2FKBP F36V , and pApg16 ⌬65-118 -2FKBP F36V .
Antibodies-Anti-Apg12, anti-Apg16, and anti-Apg5 antibodies were prepared as follows. The pGEX-3X-Apg12, pGEX-3X-Apg16, and pGEX-2T-Apg5 plasmids were transformed into Escherichia coli XL1 Blue MRFЈ, and transformants were grown in LB medium containing 50 g/ml ampicillin to 0.5 A 600 . Recombinant protein expression was induced with 1 mM isopropyl-␤-D-thiogalactopyranoside during an additional 3-h culture at 37°C. The recombinant proteins in the inclusion bodies were separated by SDS-PAGE and simultaneously stained with Coomassie Brilliant Blue G-250. The protein bands were cut off and eluted with elution buffer (100 mM Tris, pH 6.8, 0.05% SDS). The eluted proteins were used to immunize rabbits. Antiserum against Apg16 was subsequently affinity purified by passing serum over a column of His 6tagged Apg16 recombinant protein coupled to CNBr-activated Sepharose 4B (Amersham Biosciences). For expression of His 6 -tagged Apg16, the pDEST17-Apg16 plasmid was transformed into BL21-SI competent cells (Invitrogen), and transformants were grown up to 0.5 A 600 at 30°C in LB without NaCl medium (1% Bactotryptone, 0.5% yeast extract) containing 50 g/ml ampicillin, and expression of His 6 -Apg16 protein was induced for 2 h with 0.3 M NaCl. His 6 -tagged Apg16 was purified from inclusion bodies as described above. The purified His-Apg16 protein was covalently coupled to CNBr-activated Sepharose and incubated with antiserum overnight at 4°C. After washing unbound antisera extensively, bound antibodies were eluted with 0.1 M glycine, pH 2.8. Eluates containing the affinity-purified antibodies were neutralized and stored at Ϫ80°C. Anti-Apg5 antibody was purified likewise using GST-Apg5-conjugated formyl-cellulofine. Antiserum to aminopeptidase I (API) was provided by Dr. Daniel J. Klionsky (University of Michigan).
Western Blotting of Total Cell Lysate-Total cell lysates were prepared as follows. Cells were grown to 1 A 600 unit/ml and, if necessary, starved for 3 h in SD (ϪN) medium. 10 A 600 units of cells were harvested by centrifugation and resuspended directly in 100 l of a solution containing 0.2 M NaOH and 0.5% 2-mercaptoethanol. Cells were incubated for 15 min on ice, after which 1 ml of cold acetone was added, and the cells were incubated further for 30 min at Ϫ20°C. After centrifugation at 15,000 ϫ g for 5 min, pellets were resuspended in 200 l of SDS-sample buffer and boiled for 5 min. Lysates were subjected to SDS-PAGE and immunoblotting.
Differential Centrifugation-Yeast cells were grown to 1 A 600 unit/ ml, and 50 A 600 units of cells were collected by centrifugation. Spheroplasts were generated and lysed by Dounce homogenization in lysis buffer (20 mM PIPES, pH 6.8, 5 mM MgCl 2 , 100 mM sorbitol, and protease inhibitors) with or without 150 mM NaCl. After a preclearing step at 100 ϫ g for 5 min, the lysates were subjected to low speed centrifugation at 13,000 ϫ g for 20 min to generate a pellet (P13) fraction. The resulting supernatant was centrifuged further at 100,000 ϫ g for 1 h to generate pellet (P100) and supernatant (S100) fractions.
Gel Filtration-Cells (50 A 600 units) were collected by centrifugation and converted to spheroplasts. Cell lysates were prepared with lysis buffer (20 mM PIPES, pH 6.8, 150 mM NaCl, 5 mM MgCl 2 , 100 mM sorbitol, and protease inhibitors) and centrifuged at 100,000 ϫ g for 1 h. The resulting supernatant was separated by size exclusion chromatography on a Superdex 200 column (Amersham Biosciences). The column was equilibrated with 50 mM Tris, pH 6.8, 150 mM NaCl, 1 mM dithiothreitol. Cell lysate (about 0.4 mg of protein in 200 l) was applied to and eluted from the column at a flow rate of 0.5 ml/min, and 0.8-ml fractions were collected. The column was calibrated with both high and low molecular mass gel filtration protein standards (Amersham Biosciences) containing ferritin (440 kDa), catalase (232 kDa), aldolase (158 kDa), albumin (67 kDa), ovalbumin (43 kDa), and chymotrypsinogen A (25 kDa).

Characterization of Endogenous Apg12-Apg5 Conjugate and
Apg16 -We generated polyclonal antibodies against Apg12 and Apg16 to examine the state of the endogenous Apg12-Apg5 conjugate and Apg16. Upon immunoblotting, most endogenous Apg12 exists in the form of the conjugate with Apg5 in wildtype cells (Fig. 1A). This is true in apg mutants except for ⌬apg5, ⌬apg7, and ⌬apg10. In these three apg mutants, which are defective in Apg12 conjugation (6), Apg12 was present exclusively in the unconjugated form. Apg16 was detected at the expected molecular size with affinity-purified anti-Apg16 antibody (Fig. 1B). When Apg16 was overexpressed, an additional band (* in Fig. 1B) was observed as reported previously (16), although it has not yet been characterized. The amounts of unconjugated Apg12, Agp12-Apg5 conjugates, and Apg16 did not change during nitrogen starvation ( Fig. 1B and data not shown).
We next examined the subcellular distribution of the endogenous Apg12-Apg5 conjugate and Apg16 by differential centrifugation. Previous studies demonstrated that distribution of the hemagglutinin (HA) epitope-tagged Apg12-Apg5 conjugate depends on lysis buffer conditions (12,16). Thus, we tested different buffer conditions and confirmed that the distribution of the endogenous Apg12-Apg5 conjugate was indeed affected considerably by the salt concentration of the lysis buffer. Yeast spheroplasts were homogenized with lysis buffer with or without a physiological concentration of salts. After removal of cell debris, the lysates were centrifuged at 13,000 ϫ g for 20 min to generate a pellet (P13) fraction. The resulting supernatant was centrifuged again at 100,000 ϫ g for 1 h to further separate it into pellet (P100) and supernatant (S100) fractions. Each fraction was subjected to immunoblotting with anti-Apg12 and anti-Apg16 antibodies. In buffer containing 150 mM NaCl, the Apg12-Apg5 conjugate was found primarily in the S100 fraction, whereas in the absence of salt it was found mainly in the P100 fraction (Fig. 1C). Apg16 displayed a similar distribution pattern (Fig. 1C). Because the distribution we observed using the salt-containing buffer is more consistent with our morpho-logical observations that most Apg5 is distributed throughout the cytoplasm in both yeast and mammalian cells (13,14), we used buffer containing physiological salt concentrations in the following experiments.
Cytosolic Apg12-Apg5 Conjugate and Apg16 Form a ϳ350-kDa Complex-Our previous study suggested that Apg16 forms an oligomer and cross-links the Apg12-Apg5 conjugate (16). To characterize the Apg12-Apg5⅐Apg16 complex further, we performed gel filtration analysis. The S100 fraction was subjected to gel filtration analysis using a Superdex 200 column and subsequently immunoblotted with anti-Apg12 and anti-Apg16 antibodies. The Apg12-Apg5 conjugate eluted mainly in a single peak in fractions corresponding to ϳ350 kDa. Most Apg16 was also recovered in these fractions ( Fig. 2A). Coelution of Apg12-Apg5 and Apg16 indicates that most of the Apg12-Apg5 conjugate and Apg16 form a ϳ350-kDa protein complex. Monomeric Apg12-Apg5 conjugate and Apg16 were scarcely detected. The ϳ350-kDa complex was already formed under nutrient conditions, and the elution profile was not affected by nitrogen starvation (Fig. 2B). In contrast, in ⌬apg16 cells, the ϳ350-kDa complex was not observed (Fig. 3A), indicating that the formation of this complex depends on Apg16. The Apg12-Apg5 conjugate was found in ϳ60-kDa fractions, corresponding to the sum of the molecular masses of Apg12 (21 kDa) and Apg5 (33 kDa) (Fig. 3A). We expected that Apg16 may cross-link two Apg12-Apg5 conjugates (16), but the size of the resulting complex was significantly larger than what would be expected from two sets of Apg12, Apg5, and Apg16. One possible explanation is that there may be additional components. However, as far as we examined by immunoprecipitation analysis using anti-Apg12 antibody, we could not detect any protein other than Apg12, Apg5, and Apg16 (data not shown). Furthermore, our observation that the main Apg12-Apg5 conjugate peak was shifted to fractions much smaller than half of 350 kDa in ⌬apg16 cells (Fig. 3A) suggests that the ϳ350-kDa complex is not dimeric. Therefore, we believe that the ϳ350-kDa complex represents an Apg12-Apg5⅐Apg16 multimeric complex, most probably tetrameric, the formation of which is mediated by Apg16 oligomerization.
We also performed gel filtration analysis on ⌬apg5 and ⌬apg12 cells. In ⌬apg5 cells, the ϳ350-kDa complex was not present, and the Apg12 and Apg16 peaks were detected in fractions of low molecular size (Fig. 3B). The eluting peak position of Apg16 (fractions 11 and 12) was larger than the expected molecular size of the Apg16 monomer. When we performed gel filtration analysis using ⌬apg5⌬apg16 cells expressing both Myc Apg16 and HA Apg16, the tagged Apg16 was recovered in the same fractions (data not shown). Coimmunoprecipitation analysis revealed that the amount of the Myc Apg16-HA Apg16 dimer was very little, and most Apg16 existed as a monomer in the absence of Apg5 (data not shown), suggesting that Apg16 in fractions 11 and 12 represents the monomeric form. Apg16 might not be a globular protein and behave aberrantly in size exclusion chromatography. Taken together, these results suggest that the Apg5-Apg16 interaction would be important for Apg16 to form an oligomer. In contrast, in ⌬apg12 cells, Apg5 and Apg16 were mainly coeluted in ϳ250-kDa fractions, which would represent the complex made up of four sets of Apg5 and Apg16 (Fig. 3C). Therefore, even in the absence of Apg12, Apg5 and Apg16 could form the tetrameric complex.
To confirm that formation of the ϳ350-kDa complex is mediated by Apg16, we reconstituted the complex formation using two different cell lysates: one was prepared from ⌬apg16 cells, in which the Apg12-Apg5 conjugate existed as a monomeric form (Fig. 4A), and the other was from ⌬apg5⌬apg12 cells overexpressing Apg16, in which most Apg16 was monomeric (Fig. 4B). These two cell lysates were mixed, incubated at 4°C overnight, and subjected to gel filtration analysis. In contrast to the premixed sample (Fig. 4A), the ϳ350-kDa complex appeared clearly in the incubated mixture (Fig. 4C). We could not detect the shift of Apg16 probably because it was below the detectable level by the antibody, judging from the band intensity of overproduced Apg16 (data not shown). We also observed a small but significant shift of the Apg12-Apg5 fractions after a 1-h incubation of the lysates (data not shown).
Apg16 Oligomerization Is Required for Formation of the ϳ350-kDa Apg12-Apg5⅐Apg16 Complex-Because the above experiments suggested that formation of the ϳ350-kDa complex is mediated by the Apg16 homo-oligomer, we next attempted to control the oligomerization state of Apg16 to determine whether the formation of this complex is required for autophagy. We employed a regulated oligomerization system, which allows drug-induced multimerization of proteins of interest (18,19). This system is based on FKBP and its small ligand FK506. A bivalent drug AP20187, which is created by chemically linking two FK506 derivatives, is able to cross-link fusion proteins containing the FKBP F36V domain, in which Phe 36 of wild-type

FIG. 1. Expression and subcellular distribution of endogenous
Apg12-Apg5 conjugate and Apg16. A, lysates from wild-type (SEY6210; WT) cells and each apg disruptant were immunoblotted using anti-Apg12 antibody. Positions of Apg12-Apg5 conjugate and unconjugated Apg12 are indicated. B, lysates from ⌬apg16, wild-type cells, wild-type cells starved for 3 h, and ⌬apg16-overexpressing Apg16 (by a 2 plasmid) were subjected to immunoblotting using anti-Apg16 antibody. Positions of Apg16 and its uncharacterized modified form (*) are indicated. C, spheroplasts generated from wild-type cells were homogenized in lysis buffer with or without 150 mM NaCl as described under "Experimental Procedures." After removing debris, total lysates were subjected to low speed centrifugation at 13,000 ϫ g for 20 min to generate a pellet fraction (P13). The resulting supernatant was separated into pellet (P100) and supernatant (S100) fractions by centrifugation at 100,000 ϫ g for 1 h. Each fraction was subjected to immunoblotting using anti-Apg12 and anti-Apg16 antibodies.
FKBP is replaced with valine. AP20187 has a 1,000-fold higher affinity for FKBP F36V than for wild-type FKBP. Thus, a fusion protein containing two copies of FKBP F36V is capable of forming oligomers in the presence of AP20187. Because Apg16 is suggested to form an oligomer through its carboxyl-terminal coiled-coil region (16), we modified Apg16 by inserting two tandem copies of FKBP F36V after amino acid 65 or 118 (described as Apg16 65 -2FKBPF 36V and Apg16 118 -2FKBP F36V , respectively) or by using it to replace nearly the entire coiled-coil region (Apg16 ⌬65-118 -2FKBP F36V ), where it would disrupt the coiled-coil region and thereby inhibit natural Apg16 oligomerization (Fig. 5A). We then tested whether each Apg16 -2FKBP F36V fusion protein oligomerized in an AP20187-dependent manner. ⌬apg16 cells expressing each Apg16 -2FKBP F36V fusion protein were treated with or without 0.1 M AP20187, and cell lysates were separated on a Superdex 200 column as described above. Among these constructs, only the Apg16 65 -2FKBP F36V fusion protein, which was inserted with 2FKBP F36V immediately after the beginning of the coiled-coil region, allowed for ligand-dependent oligomerization. Without   FIG. 2. Apg12-Apg5 and Apg16 form a ϳ350-kDa complex in the cytosol. The S100 fractions were prepared from wild-type cells that were growing (A) or starved for 3 h in SD (ϪN) medium (B) and separated by size exclusion chromatography on a Superdex 200 column. Each fraction was subjected to immunoblotting using anti-Apg12 and anti-Apg16 antibodies. Positions of molecular mass standards (in kDa) are shown. V, void fraction.
FIG. 4. The ϳ350-kDa complex is formed in vitro using ⌬apg16 and ⌬apg5⌬apg12 cell lysates. The S100 fractions from ⌬apg16 cells (A) and ⌬apg5⌬apg12 cells (YNM117) overexpressing Apg16 from a 2 plasmid (B) were prepared. Equal volumes of these two cell lysates were mixed, incubated at 4°C overnight, and subjected to gel filtration analysis using a Superdex 200 column. Each fraction was analyzed by immunoblotting using anti-Apg12 (A and C) and anti-Apg16 antibodies (B). AP20187, the Apg16 65 -2FKBP F36V was detected in fractions corresponding to ϳ200 kDa (Fig. 5Ba). In this peak, Apg16 65 -2FKBP F36V associated with the Apg12-Apg5 conjugate because Apg16 65 -2FKBP F36V was detected at ϳ55 kDa in ⌬apg5 cells, which was probably monomeric Apg16 65 -2FKBP F36V (Fig. 5Be). Because the sum of the molecular masses of Apg12, Apg5, and Apg16 65 -2FKBP F36V is about 100 kDa, our data suggest that dimeric Apg12-Apg5⅐Apg16 65 -2FKBP F36V could be formed even in the absence of AP20187. The remaining coiled-coil region of Apg16 65 -2FKBP F36V might function partially (see "Discussion"). After AP20187 treatment, the peak of Apg16 65 -2FKBP F36V was clearly shifted to fractions corresponding to ϳ400 kDa (Fig. 5Bc). This change of peak indicates that Apg16 oligomerizes upon AP20187 treatment. When the Apg16 118 -2FKBP F36V fusion protein, in which 2FKBP F36V was inserted at almost the end of the coiled-coil region, was expressed in ⌬apg16 cells, the ϳ400-kDa complex was present irrespective of the AP20187 treatment (data not shown). Expression of the Apg16 ⌬65-118 -2FKBP F36V was too low to be evaluated (data not shown).
Using this controlled Apg16 oligomerization system with the Apg16 65 -2FKBP F36V construct, we examined the complex state of the Apg12-Apg5 conjugate. In the absence of AP20187, the Apg12-Apg5 conjugate was eluted mainly at ϳ60 kDa, equivalent to the molecular mass of the conjugate (Fig. 5Bb). After treatment with AP20187, the Apg12-Apg5 conjugate was found to elute in ϳ400-kDa fractions (Fig. 5Bd), together with oligomerized Apg16 65 -2FKBP F36V . This peak corresponds to the wild-type ϳ350-kDa Apg12-Apg5⅐Apg16 complex; insertion of 2FKBP F36V (about 24 kDa) into Apg16 most likely accounts for the difference. These results suggest that formation of the ϳ350-kDa Apg12-Apg5⅐Apg16 complex requires Apg16 oligomerization and that we were effectively able to regulate the formation of the complex with this AP20187-dependent oligomerization system.
The ϳ350-kDa Complex Is Required for Autophagy-Next, we used the oligomerization system to determine whether the formation of the ϳ350-kDa complex is necessary for autophagy. ⌬apg16 cells expressing Apg16 65 -2FKBP F36V were incubated with AP20187 overnight and then starved with nitrogen-free medium containing 1 mM phenylmethanesulfonyl fluoride and 0.1 M AP20187 for 8 h. Cells were then examined by light microscopy for accumulation of autophagic bodies in vacuoles (20). Typical accumulation of autophagic bodies was observed in AP20187-treated cells but was rarely detected in untreated cells (Fig. 6A).
Induction of autophagy was also confirmed by examining the maturation of a vacuolar enzyme, API. In S. cerevisiae, API is synthesized in a pro-form (prAPI) and delivered from cytoplasm to vacuole via the cytoplasm to vacuole targeting pathway, an autophagy-related pathway (21)(22)(23). During starvation, API is transported to the vacuole via the autophagic pathway. Delivery to the vacuole leads to maturation of API into its active form (mAPI). ⌬apg16 cells expressing Apg16 65 -2FKBP F36V were grown in medium with or without 0.1 M AP20187 and starved for 3 h. Lysates were prepared and subjected to immunoblotting using anti-API antibody. Although most API was processed to the mature form in wild-type cells (Fig. 6B, lane 3), mature API was scarcely detected in ⌬apg16 cells expressing Apg16 65 -2FKBP F36V (Fig. 6B, lane 1). As expected, maturation of API was restored after treatment with AP20187 (Fig. 6B, lane 2) in a concentration-dependent manner (Fig. 6C). Taken together, the results of these experiments suggest that formation of the ϳ350-kDa Apg12-Apg5⅐Apg16 complex is required for autophagy. DISCUSSION In the present study we have demonstrated that Apg12, Apg5, and Apg16 form a ϳ350-kDa multimeric complex that exists stably in the cytosol. Our approach, utilizing the regu- lated oligomerization system, demonstrated concretely that formation of the ϳ350-kDa complex, which depends on oligomerization of Apg16, is required for autophagy. Our study also demonstrated that this oligomerization system works well in yeast cells in vivo.
In our previous experiment using HA-tagged Apg12 ( HA Apg12), less than half of the Apg12 was found conjugated to Apg5 (6), whereas most of endogenous Apg12 was found attached to Apg5 (Fig. 1A). It is conceivable that plasmid-derived expression of HA Apg12 affects the conjugation efficiency. HA tagging may also affect the subcellular distribution determined from differential centrifugation experiments. As shown in Fig.  1C, the endogenous Apg12-Apg5 conjugate was recovered primarily in the cytosolic fraction, in contrast to previous reports showing that most HA-tagged Apg12-Apg5 was pelletable (16), even when lysis buffer containing salt was used (12). Combined with our recent morphological studies demonstrating that most Apg12-Apg5 is found in the cytoplasm (13,14), we concluded that this is the primary localization of endogenous Apg12-Apg5. Under the present conditions, a small portion of Apg12-Apg5 was still observed in the pellet fractions (Fig. 1C). How-ever, its amount was not changed during nitrogen starvation or in ⌬apg6 cells, in which the pre-autophagosomal structures were not formed (13). Therefore, it is dubious that the conjugate in the pellet fractions represents any real physiological localization.
Upon gel filtrate analysis, we found that almost all of the Apg12-Apg5 conjugate and Apg16 are included in the ϳ350-kDa complex, whereas the monomeric forms of each component were scarcely detected. This is not simply because each monomer is unstable; free Apg12-Apg5 conjugate and Apg16 were easily detected in ⌬apg16 and ⌬apg5 cells, respectively, neither of which contain the ϳ350-kDa complex (Fig. 3). In the previous report, we suggested that Myc Apg16 interacts efficiently with Apg12-HA Apg5 but inefficiently with unconjugated HA Apg5 (16). However, gel filtration analysis of endogenous proteins showed that Apg16 could interact quite well with unconjugated Apg5, though still somewhat weaker than with conjugated Apg5 (Fig. 3C). Therefore, the primary role of Apg12 conjugation would not be to strengthen the Apg5⅐Apg16 association. It is also unknown which takes place earlier in vivo, Apg12-Apg5 conjugation or Apg5⅐Apg16 interaction. Al- though both are possible, these two molecular interactions do not depend on each other, i.e. the Apg12-Apg5 conjugate is generated in ⌬apg16 cells (Fig. 1A), and Apg5 could form a complex with Apg16 in the absence of Apg12 (Fig. 3C).
From our analysis, it is most likely that the ϳ350-kDa complex consists of four sets of Apg12-Apg5⅐Apg16 proteins. Using the regulated oligomerization system, we showed successfully that formation of the ϳ350-kDa complex is mediated by Apg16 homo-oligomer. Although Apg16 65 -2FKBP F36V and Apg12-Apg5 were clearly shifted to ϳ400 kDa by AP20187 treatment, it was intriguing that these proteins were detected at the fraction corresponding to ϳ200 kDa in the absence of AP20187. Because 2FKBP F36V was inserted immediately after the beginning of the coiled-coil region, the remaining coiled-coil region of Apg16 might mediate formation of dimeric Apg12-Apg5⅐Apg16 65 -2FKBP F36V complex. If that is the case, it was conceivable that our system regulates tetramer formation by cross-linking the dimeric Apg16 65 -2FKBP F36V . We demonstrated that the ϳ400-kDa Apg12-Apg5⅐Apg16 65 -2FKBP F36V complex formed by such a manner was functional, suggesting that the wild-type ϳ350-kDa complex might be also formed by dimer-dimer association. In general, tetramer formation mediated by dimer-dimer contact is possible as observed in the cases of EPS15 (24), F 1 -ATPase inhibitor protein IF1 (25), and Mnt repressor (26), although four-stranded coiled-coils have been reported (27). We also tried to test whether Apg16 inserted with a single copy of FKBP F36V domain at the same site was able to function. However, the resulting Apg16 65 -1FKBP F36V was still functional in the absence of AP20187, probably because of insufficient interference of the tetramer formation, and thus could not be evaluated (data not shown). The regulated oligomerization system used in this study is based on the interaction between FK506 and FKBP. Another natural ligand of FKBP is rapamycin, which is known to induce autophagy (28). The rapamycin⅐FKBP complex inhibits Tor kinase, which plays a role in nutrient sensing, thus inducing autophagy (28 -30). Therefore, we tested AP20187 to make sure it did not possess a rapamycin-like activity that could affect our results. We confirmed that in contract to rapamycin, AP20187 itself did not induce autophagy (data not shown). The possibility that AP20187 could induce autophagy through the Tor-mediated pathway was also rejected because AP20187 and nitrogen starvation had such strong additive effects on autophagy and API maturation (Fig. 6).
In our experiment, the Apg16 65 -2FKBP F36V fusion construct allowed for ligand-dependent recovery of the formation of the Apg12-Apg5⅐Apg16 complex and autophagic activity, suggesting that the Apg12-Apg5 conjugate would not function unless cross-linked by Apg16. On the other hand, ⌬apg16 cells expressing Apg16 118 -2FKBP F36V were defective in autophagy, even though the ϳ400-kDa complex was present (data not shown). This may indicate that the extreme carboxyl-terminal region of Apg16 plays a role in something other than oligomerization. We suggested recently that one role of Apg16 might be recruitment of the Apg12-Apg5 conjugate to the pre-autophagosomal membrane (13). Unfortunately, we could not conclude whether Apg16 oligomerization is essential for Apg12-Apg5 recruitment, because the signal intensity of GFP-Apg5 on preautophagosomes was very weak when using the oligomerization system. In any case, our data suggest that the ϳ350-kDa Apg12-Apg5⅐Apg16 complex is formed in the cytosol and then targets to the pre-autophagosomal membrane. Even though there is no apparent homolog of Apg16 in other species, it is likely that a functional counterpart exists to cross-link Apg12-Apg5 conjugates. Indeed, we observed that the mammalian Apg12-Apg5 conjugate is included in a large protein complex. 2 Considering that Apg12-Apg5 preferentially localizes to the outer side of the elongating autophagic isolation membrane in mammalian cells (14), the possibility that the Apg12-Apg5⅐Apg16 complex functions like coat proteins is an attractive one. Further studies aimed at investigating this ϳ350-kDa complex are likely to improve our understanding of the molecular mechanisms of autophagosome formation.