Alcohol oxidase and dihydroxyacetone synthase, the abundant peroxisomal proteins of methylotrophic yeasts, assemble in different cellular compartments

Peroxisomal protein import has been the subject of fruitful research in the past several years, and as a result many steps in this process have been identified (Hettema et al., 1999EF16; Kunau, 1998EF19; Subramani et al., 2000EF25). One of two peroxisomal targeting signals, PTS1 and PTS2, allows proteins destined to peroxisomes to bind to cytosolic receptors (Pex5p or Pex7p for PTS1 and PTS2, respectively). The substrate-receptor pair then binds to a docking complex (containing at least Pex13p, Pex14p and Pex17p) on the peroxisomal membrane. Several other membrane-associated proteins function downstream in a way that is ill-defined at present to promote translocation of the substrate. The medical significance of peroxisomal protein import is demonstrated by the peroxisomal deficiency diseases Zellweger syndrome and neonatal adrenoleukodystrophy, in which there is a global failure of peroxisomal matrix protein import (Fujiki, 1997EF9).

While most organelles that undergo protein import accept only unfolded substrates, several lines of evidence indicate that peroxisomes import folded proteins and oligomers: (1) monomers of oligomeric peroxisomal proteins and carriers that lack a PTS can be imported if coexpressed with partners containing a PTS (Elgersma et al., 1996EF7; Glover et al., 1994EF11; McNew and Goodman, 1994EF22); (2) microinjected protein aggregates and even colloidal gold particles that contain a PTS can be imported (Walton et al., 1995EF30); (3) aminopterin does not inhibit the import of dihydrofolate reductase into peroxisomes (Hausler et al., 1996EF15); (4) mistargeting of alanine:glyoxylate aminotransferase to mitochondria in the disease human primary hyperoxaluria type I requires mutations that both generate a mitochondrial targeting signal and disrupt dimerization (Leiper et al., 1996EF21); (5) in vitro import of protein dimers has been demonstrated (Brickner et al., 1997EF2); and (6) the PTS receptors Pex5p and Pex7p can be found within the peroxisomal matrix suggesting that they accompany substrate into this compartment and then recycle back to the cytosol (Dodt and Gould, 1996EF5; Szilard et al., 1995EF26; Zhang and Lazarow, 1995EF32).

Recent evidence has shown that peroxisomes are not unique in their ability to import folded proteins. The cvt pathway for the import of cytosolic precursors into the yeast vacuole involves the engulfment of cytosolic oligomers by membrane and fusion to the vacuole (Klionsky and Ohsumi, 1999). Thylakoid membranes of plants can export proteins from the stroma through three different pathways, one of them (the ΔpH pathway) accepts folded precursors as substrates (Creighton et al., 1995). A similar mechanism of export, the twin-arginine translocation pathway, is found in E. coli (Hunds et al., 1998).

While many experiments have clearly shown that peroxisomes can import oligomers, questions remain about the physiological importance of this mechanism. If a protein can be imported into peroxisomes as an oligomer, does it normally utilize that pathway? Is oligomeric import the exclusive pathway for peroxisomes? Since classical chaperones have not been found within the organelle, do peroxisomes have the capacity to assemble monomers into active oligomers, and do they ever do so?

To answer these questions we used a pulse-chase approach in the yeast Candida boidinii to ascertain the relationship between peroxisomal import and oligomerization. Methylotrophic yeasts such as C. boidinii, when cultured on methanol, contain large peroxisomes that consist almost exclusively of two oligomeric proteins, alcohol oxidase (AO), an octamer, and dihydroxyacetone synthase (DHAS), a dimer (Goodman, 1985; Veenhuis and Harder, 1987). We now report that AO is imported into peroxisomes over several minutes as a monomer, consistent with a previous finding that AO lacking a PTS could not be imported with subunits containing a PTS (Waterham et al., 1997). Octamerization of AO is demonstrated to occur in the peroxisomal matrix. By contrast, DHAS quickly dimerizes in the cytosol prior to its import.

Candida boidinii was cultured in mineral medium (Sahm and Wagner, 1972) supplemented with 1% MeOH and 0.5 M MgSO₄. Addition of 0.5 M MgSO₄ to the culture medium was found to yield more reproducibility in labeling and expression of peroxisomal proteins during metabolic labeling. All of the following steps to generate and activate spheroplasts were performed at 30°C with pre-warmed reagents and incubations in a rotary shaker. 100 ml of cell culture at 1 OD₆₀₀/ml was harvested by centrifugation at 6,750 g for 10 minutes, resuspended in 30 ml H₂O, and repelleted. Cells were resuspended in 10 ml 0.1 M Tris-SO₄ (pH 9.3) with 10 mM DTT, and incubated for 10 minutes. Cells were harvested and washed in 30 ml 0.5 M MgSO₄. The pellet was resuspended in 10 ml 20 mM KHPi pH 7.5 with 0.5 M MgSO₄ and 1% MeOH, and then 15 μg of 100T Zymolyase was added. Spheroplasting was allowed to occur for approximately 45-60 minutes until conversion to spheroplasts was complete, as monitored microscopically by the ability of added H₂O to lyse a large majority of the cells. 20 ml 0.5 M MgSO₄ was then added, the cells were pelleted at 5,820 g for 10 minutes and then resupended in 10 ml mineral medium with MeOH and MgSO₄ (identical to culture medium) and incubated at 30°C for 1 hour to activate prior to labeling.

Activated spheroplasts were radiolabeled for 5 minutes with 30 μCi ³⁵S-methionine (New England Nuclear Life Science Products, Boston, MA) per 500 μl spheroplasts. Unlabeled methionine/cysteine, a solution of 250 mM each, was then added to a final concentration of 10 mM each for the chase, and cells were allowed to incubate for the times indicated. At each time point of chase, 200 μl of cell suspension was quickly pipetted into 1 ml of pre-chilled 1 M sorbitol on ice. (For the temperature shift experiments, the 15 degree shift was performed in a water bath using small volumes of cell suspension (less than 1 ml) and in thin plastic tubes to ensure rapid temperature equilibration.) Cells were harvested by centrifugation at 5,000 g spin for 5 minutes at 4°C in a microfuge. They were then resuspended in 200 μl 1 M SMA (1 M sorbitol, 5 mM MES (2-[N-morpholino]ethanesulfonic acid) pH 5.5, 0.2 mM AEBSF (4-(2-aminoethyl)benzenesulfonyl fluoride), then lysed by the addition of 800 μl 0.25 M (referring to sorbitol concentration) SMA. The lysate was vortexed three times for 10 seconds with a 5 minute incubation on ice in between each vortexing. Separation of cytosol (SUP) and organellar pellet (PEL) was performed by centrifugation at 20,000 g (in a microfuge) for 15 minutes at 4°C. For experiments in which the PEL was further fractionated, a duplicate PEL fraction was resuspended in 500 μl 30 mM Tris-HCl pH 9.0 with 5 mM EDTA to lyse peroxisomes contained therein. Separation of membrane (MEM) and matrix (MAT) fractions was then performed by a single centrifugation at 100,000 g for 15 minutes at 4°C in a Beckman TL-100 rotor. Immunoprecipitation of AO and DHAS was performed on each fraction using antibodies to the corresponding proteins (see Reagents and other methods). Radiolabel was visualized by fluorography and the dried gel was exposed to BioMax single emulsion film (Eastman Kodak, Rochester, NY).

The oligomeric state of AO contained in SUP and MAT fractions was determined by velocity sedimentation in sucrose gradients and immunoprecipitation of gradient fractions as previously reported (Goodman et al., 1984EF13). A similar procedure was performed to establish the oligomeric state of DHAS except gradients were centrifuged for 12 hours instead of 3.5 hours. Bovine serum albumin (BSA), mature DHAS and AO were used as molecular mass standards in the gradients and visualized by Coomassie Blue staining of SDS-PAGE gels.

Polyclonal antibodies for AO (Goodman et al., 1984EF13) and DHAS (Goodman, 1985EF12), and a monoclonal antibody against Pmp47 (Goodman et al., 1986EF14) used in immunoblots, have been previously described. Immunoblots were performed with Enhanced Chemiluminescence reagents as directed by the manufacturer (Amersham Life Science, Arlingon Heights, IL). SDS-PAGE (Laemmli, 1970EF20) was performed with 9% gels and a separating gel of pH 9.2. Formaldehyde dehydrogenase was assayed enzymatically (Egli et al., 1980EF6).

Analysis of peroxisomal targeting intermediates required a fractionation procedure to quickly, cleanly and reproducibly separate cytosol from peroxisomal matrix and membrane in high yield following pulse-chase. Such a method is outlined in Fig. 1A. Following pulse-chase of activated spheroplasts, cells were quickly lysed and subjected to a single centrifugation to separate cytosol (and light membranes, the SUP fraction) from peroxisomes (and other organelles, the PEL fraction). Fig. 1B demonstrates that this step results in the release of 94% of a cytosolic marker, formaldehyde dehydrogenase, a measure of the high efficiency of spheroplast lysis. Integrity of peroxisomes can be assessed in Fig. 1C, illustrating a protein stain of the fractions after SDS-PAGE. Alcohol oxidase and dihydroxyacetone synthase are the most abundant proteins in the cell, an indication of the massive proliferation of peroxisomes when cells are incubated in methanol medium (Goodman et al., 1984EF13; Sahm et al., 1975EF24). Very few of these matrix proteins are seen in the SUP fraction compared with the PEL fraction. Densitometry of these bands reveals at least 80% of these markers fractionate in the PEL (AO is generally above 90%), indicating that almost all peroxisomes are intact. Fig. 1D, illustrating an immunoblot against the integral peroxisomal membrane protein Pmp47 (Goodman et al., 1986EF14), shows that all of the peroxisomal membrane fractionates with the PEL fractions.

To release matrix from the PEL fraction, the pH was raised to 9.0 (Goodman et al., 1984EF13), and the sample was subjected to high speed centrifugation (Fig. 1A). Virtually all the AO and DHAS bands remained in the supernatant (MAT, Fig. 1C), while Pmp47 was recovered only in the pellet (MEM, Fig. 1D). The data show that the matrix fraction contained virtually all the AO and DHAS from the pellet (Fig. 1C) and no peroxisomal membrane (Fig. 1D). In summary, there is little cross contamination between cytosol (in the SUP), matrix and peroxisomal membrane.

Using pulse-chase and this fractionation protocol, we followed the trafficking of newly synthesized AO and DHAS, and the results up to 15 minutes of chase are shown in Fig. 2A,B. Alcohol oxidase chases from the cytosol (SUP) to peroxisomes (PEL) as has been shown previously in a more detailed analysis (Goodman et al., 1984). A similar pattern was observed for DHAS, although the kinetics of association with peroxisomes was considerably faster (Goodman, 1985). The fractions of AO and DHAS found in the SUP could not be sedimented at 100,000 g, indicating that they are in the cytosol (data not shown). During this time course both proteins appeared in the peroxisomal matrix. We always saw a lag in appearance of both labeled proteins between the PEL and in the MAT (Fig. 2A). Very little AO or DHAS was seen in the MEM fraction at any time point. This loss suggested that the immunoprecipitation from the membrane fraction was ineffective. When the fractions were visualized directly without immunoprecipitation (Fig. 2C), there was clearly a significant amount of newly synthesized AO and DHAS associated with the membrane fraction at early time points. In fact, more AO precursor is observed in the membrane than the matrix at 0 and 1 minutes.

The oligomeric state of AO in the cytosol and matrix, as determined by velocity sedimentation, as a function of time after synthesis, is shown in Fig. 3. AO is a monomer in the cytosol at all time points examined, suggesting that oligomerization does not happen in this compartment. As AO appears in the matrix (1 minute chase), it is mostly monomer. Most of the monomer that we detect in the matrix is very sensitive to proteolysis during sample processing, much more so than monomers in the cytosol, and is clipped to a stable intermediate (indicated by asterisks in Fig. 3). We have been unable to fully prevent this degradation with protease inhibitors. At longer chase times, octamer accumulates in the matrix, while monomers still enter peroxisomes (15 minutes). In some experiments AO of an intermediate size is seen (15 minute point, open arrowhead, more apparent at lower temperatures), which may indicate dimeric or tetrameric forms. This experiment demonstrates that the peroxisomal matrix has the capacity to oligomerize AO.

Parallel fractions from the same experiment were centrifuged in the sucrose gradients for longer times to separate monomeric from dimeric DHAS (Fig. 4). In contrast to AO, DHAS clearly dimerizes in the cytosol prior to import. Even at 0 minutes of chase about half of DHAS is already a dimer; the ratio increases further with longer chase. Dimeric DHAS is seen in the matrix, even at the earliest time points. Considering the efficient dimerization of DHAS in the cytosol and the appearance of dimer in the matrix at the earliest time point, the simplest interpretation of our data is that the dimeric form of DHAS is the major if not the only species that crosses the peroxisomal membrane.

Having shown that AO is imported as a monomer and DHAS as a dimer, we next performed import experiments at lower temperatures to attempt to uncover mechanistic differences in the import of the two substrates. At 15°C, the association of both AO and DHAS with peroxisomes revealed a complex triphasic kinetic pattern with AO exhibiting slower kinetics in all three phases. The slower import kinetics of AO were not attributable to a decrease in the rate of octamerization in the matrix (data not shown). Although AO appears to cross the peroxisomal membrane more slowly than DHAS, our data do not allow us to conclude that mechanistic differences exist between the import of these two substrates.

In summary, we have been able to monitor simultaneously the import and assembly of endogenous peroxisomal proteins in C. boidinii. Our data show that AO enters peroxisomes as monomers and that the peroxisomal matrix has the capacity to assemble octamers from this precursor. By contrast, DHAS dimers assemble quickly in the cytosol before translocation into the matrix.

To examine the physiological significance of oligomeric import and determine whether matrix is capable of forming oligomeric proteins, we undertook a pulse-chase approach to directly observe these processes. We chose AO and DHAS from C. boidinii because each of these proteins comprise over 10% of cell protein and are very easy to detect. Our pulse-chase results depended on technical improvements over previous work in C. boidinii (Goodman, 1985; Goodman et al., 1984). Cell growth and spheroplasting conditions were adjusted to yield efficient and reproducible radiolabeling and complete cell lysis while keeping peroxisomes intact. This procedure obviated the need to discard a low-speed pellet, which can contain most of the cellular organelle mass (data not shown), thus skewing the correct ratio of substrates in cytosol versus in organelles. The subsequent lysis of the peroxisomal membrane by raising the pH allowed a fast separation of cytosol, peroxisomal membrane and matrix. This procedure was then coupled with sucrose velocity sedimentation to determine the compartment in which assembly of oligomers occurred.

These observations are novel in two ways. First, this is the only direct demonstration to our knowledge that the peroxisomal matrix can support the formation of quaternary structure. Our results suggest that the import of microinjected AO octamers into peroxisomes does not reflect the physiological import mechanism for these substrates (Walton, 1996). Once AO monomers cross the membrane, an intermediate can be detected that migrates between monomer and octamer in sucrose velocity gradients. It is not clear whether this oligomeric form represents homomers or heteromers. Why does oligomerization of AO occur in the matrix and not the cytosol? There are three possibilities, which are not mutually exclusive: (1) conditions within the matrix are favorable for spontaneous assembly of AO octamers (an optimal pH or specific ion concentration, for example); (2) proteins within the matrix catalyze assembly into octamers; or (3) molecular chaperones within the cytosol prevent assembly. Chaperones may not be required for octamer formation since monomers that have been dissociated from holoenzyme by treatment with glycerol can spontaneously reassemble into active enzyme (Boteva et al., 1999). However, dissociation is irreversible once monomers are sufficiently denatured to release cofactor. Does the monomeric form that is imported into peroxisomes contain FAD? Riboflavin starvation in the yeast Hansenula polymorpha leads to inefficient import of AO, suggesting that binding to FAD normally precedes import (Evers et al., 1994). Nevertheless, data is also available to suggest involvement of chaperones in the assembly process. Cytosolic hsp70 is known to keep proteins unfolded and competent for import into mitochondria and endoplasmic reticulum (Deshaies et al., 1988) and could play a similar role in preventing AO assembly in the cytosol. There is also evidence for the involvement of matrix chaperones: mutants in H. polymorpha that fail to generate peroxisomes assemble AO in the cytosol, as if folding factors found in the matrix in wild-type cells catalyze this process in the cytosol of the mutants (Tan et al., 1995). We have attempted to form oligomers by incubating metabolically pulse-labeled monomers with concentrated matrix but have not yet been successful since we have been unable to control the action of a protease activity to which monomeric AO is particularly sensitive (data not shown). This activity is apparent in Figs 3, 5.

Second, this report is the first study in which the extent to which a protein oligomerizes prior to peroxisomal import has been determined. Our data show that DHAS extensively and perhaps completely dimerizes prior to import since very little if any monomeric DHAS is detected in the peroxisomal matrix. These data suggest that oligomeric import is the normal mechanism for translocation of this substrate, unlike AO. Since our results are based on kinetic analysis alone, we cannot rule out the possibility that DHAS binds to the membrane as a dimer, quickly dissociates into monomers, and reassembles into dimers on the matrix side of the membrane following translocation. This pathway appears very unlikely. The only known chaperones that have the ability to dissociate and unfold proteins and protein aggregates are those in the Hsp100 family, and studies in vivo and in vitro have shown that this process requires several minutes (Glover and Lindquist, 1998EF10); we see no such lag in our import reactions, we do not detect a buildup of DHAS in the membrane prior to import, and the amount of monomers at any time in the matrix is negligible. Certainly the simplest interpretation of our data is that the dimeric form itself is imported.

The experiments at low temperature were an attempt to kinetically dissect the import process of AO and DHAS. Import is slowed for both substrates but especially for AO. Further experiments in a more genetically manipulatable system may allow us to pinpoint the rate limiting step of the import process at a molecular level.

We thank Pamela A. Marshall and members of the Goodman lab for many insightful discussions. Additional thanks to the laboratory of Margaret Phillips. The authors gratefully acknowledge funding from NIH grant GM-31859 and Robert A. Welch Foundation grant I-1085 for this work.