This Article Abstract Full Text (PDF) Supplemental Data All Versions of this Article: 19/10/3170    most recent tpc.107.050989v1 Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via CrossRef Citing Articles via ISI Web of Science (4) Citing Articles via Google Scholar Google Scholar Articles by Reumann, S. Articles by Jahn, O. Search for Related Content PubMed PubMed Citation Articles by Reumann, S. Articles by Jahn, O. Agricola Articles by Reumann, S. Articles by Jahn, O.

Proteome Analysis of Arabidopsis Leaf Peroxisomes Reveals Novel Targeting Peptides, Metabolic Pathways, and Defense Mechanisms^[W]

^a Department of Plant Biochemistry, Georg-August-University of Goettingen, Albrecht-von-Haller-Institute for Plant Sciences, D-37077 Goettingen, Germany
^b Max-Planck-Institute of Molecular Plant Physiology, Metabolic Networks, D-14424 Potsdam, Germany
^c University of Potsdam, Institute of Biochemistry and Biology, 14469 Potsdam, Germany
^d Max-Planck-Institute of Experimental Medicine, Proteomics Group, D-37075 Goettingen, Germany
^e Deutsche Forschungsgemeinschaft Research Center for Molecular Physiology of the Brain, D-37073 Goettingen, Germany

² Address correspondence to sreuman{at}gwdg.de.

	ABSTRACT

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
We have established a protocol for the isolation of highly purified peroxisomes from mature Arabidopsis thaliana leaves and analyzed the proteome by complementary gel-based and gel-free approaches. Seventy-eight nonredundant proteins were identified, of which 42 novel proteins had previously not been associated with plant peroxisomes. Seventeen novel proteins carried predicted peroxisomal targeting signals (PTS) type 1 or type 2; 11 proteins contained PTS-related peptides. Peroxisome targeting was supported for many novel proteins by in silico analyses and confirmed for 11 representative full-length fusion proteins by fluorescence microscopy. The targeting function of predicted and unpredicted signals was investigated and SSL>, SSI>, and ASL> were established as novel functional PTS1 peptides. In contrast with the generally accepted confinement of PTS2 peptides to the N-terminal domain, the bifunctional transthyretin-like protein was demonstrated to carry internally a functional PTS2. The novel enzymes include numerous enoyl-CoA hydratases, short-chain dehydrogenases, and several enzymes involved in NADP and glutathione metabolism. Seven proteins, including ß-glucosidases and myrosinases, support the currently emerging evidence for an important role of leaf peroxisomes in defense against pathogens and herbivores. The data provide new insights into the biology of plant peroxisomes and improve the prediction accuracy of peroxisome-targeted proteins from genome sequences.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
Peroxisomes are ubiquitous cell organelles that compartmentalize primarily oxidative metabolic reactions. A characteristic of peroxisomes is their metabolic flexibility because their enzymatic constituents vary depending on the organism, the type of tissue, and the environmental conditions (for review, see Beevers, 1979; Hayashi and Nishimura, 2006). Plant peroxisomes differentiate into tissue- and developmental-specific variants, such as leaf peroxisomes in mature leaves and glyoxysomes in germinating seeds, and typically participate in metabolic pathways and networks that are spread over multiple subcellular compartments. For instance, the recycling of 2-phosphoglycolate during photorespiration requires orchestration of 15 chloroplastic, peroxisomal, and mitochondrial enzymes and numerous transport proteins to transfer the intermediates between cell compartments (for review, see Reumann, 2002; Reumann and Weber, 2006). Similarly, fatty acids are released from lipid bodies during seed germination and transferred to glyoxysomes, where the carbon chain is transformed into sucrose in coordination with mitochondrial and cytosolic enzymes. Compartmentalization of hydrogen peroxide–producing oxidases in peroxisomes offers the advantage that H₂O₂ and other reactive oxygen species (ROS) can immediately be detoxified at the site of production by catalase (CAT) and auxiliary antioxidative enzymes (for review, see del Rio et al., 2006). In addition to these basic metabolic functions, a role for peroxisomes in the biosynthesis of the plant hormone auxin and the signaling molecule jasmonic acid (JA) and in sulfur and nitrogen metabolism has been established (Zolman et al., 2001; Strassner et al., 2002; Nowak et al., 2004; Koo et al., 2006). Moreover, evidence is emerging from recent studies that peroxisomes not only have important functions in metabolism but also play an essential role in specific defense mechanisms conferring resistance against pathogen attack (Taler et al., 2004; Koh et al., 2005; Lipka et al., 2005).

Thanks to rapid advances in genome sequencing, the potential of computational approaches is increasingly being realized to predict peroxisomal proteins and explore the biology of plant peroxisomes. In silico analyses take advantage of conserved targeting sequences in peroxisomal proteins (i.e., the C- terminal tripeptide constituting the peroxisomal targeting signal type 1 [PTS1] of the prototype SKL> [> designates the extreme C terminus] or the cleavable PTS2 nonapeptide located in the N-terminal domain [prototype RLx₅HL]) to predict targeting of unknown proteins to peroxisomes (Emanuelsson et al., 2003; Kamada et al., 2003; Neuberger et al., 2003a, 2003b; Reumann, 2004; Hawkins and Boden, 2006). Except for few proteins (e.g., plant CAT and Arabidopsis thaliana sarcosine oxidase; Kamigaki et al., 2003; Goyer et al., 2004), all known matrix proteins from plant peroxisomes carry either a C-terminal PTS1 or an N-terminal PTS2 (Reumann, 2004). Plant PTS motifs have been defined by experimental and bioinformatics approaches and yielded partially overlapping results but differed in specificity and peptide identity. A relatively large number (24 to 120) of PTS1 peptides have been suggested to target proteins to peroxisomes in higher plants (Hayashi et al., 1997; Mullen et al., 1997; Kragler et al., 1998), but experimental limitations required data extrapolation from amino acid substitutions to all possible residue combinations within the three-residue motif and thereby reduced motif specificity. By in silico analysis of plant EST databases (Reumann, 2004), nine major and 11 minor PTS1 peptides were defined according to stringent criteria and thus considered as rather specific, though certainly yet incomplete. A search of the Arabidopsis genome for proteins carrying these peptides revealed 280 putative PTS1/2 proteins listed in the database AraPerox (Reumann et al., 2004; www.araperox.uni-goettingen.de). Despite considerable progress, prediction by PTSs is still limited by the low number of PTS-containing proteins that have been characterized at the molecular level, at least in one plant species (18 PTS1 and 14 PTS2 proteins; Reumann, 2004). However, the predictive algorithms for PTS1 proteins in particular can steadily be improved, as 12 additional proteins with putative PTS1s have been shown to be targeted to peroxisomes in the past 3 years and, at least in some cases, have been demonstrated to follow the PTS1 targeting pathway. The recently cloned cDNAs encode, for instance, auxiliary enzymes of fatty acid ß-oxidation, ROS metabolism, and JA activation, and one of two small heat shock proteins (sHsps); the second sHsp is the only additional PTS2 protein (Edqvist et al., 2004; Tilton et al., 2004; Goepfert et al., 2005, 2006; Leterrier et al., 2005; Lisenbee et al., 2005; Schneider et al., 2005; Turner et al., 2005; Koo et al., 2006; Ma et al., 2006; Helm et al., 2007; Zolman et al., 2007).

Taken together, the low number of well-characterized plant peroxisomal matrix proteins leads to four principal constraints limiting in silico predictions of peroxisomal PTS1/2 proteins: (1) insufficient coverage of functional targeting peptides, (2) limited knowledge of the identity and function of auxiliary targeting elements surrounding the short PTS peptides, (3) possible dominance of N-terminal nonperoxisomal targeting signals over C-terminal PTS1s, and (4) unpredictable masking of PTS surface exposure by polypeptide conformation, homo- and hetero-oligomerization, or posttranslational regulatory mechanisms. Thus, the currently available prediction tools suffer from both insufficient sensitivity (i.e., an incomplete detection of true PTS1/2 proteins) and insufficient specificity (i.e., the identification of a significant number of false positives). Notably, many proteins of peroxisomes cannot currently be predicted from genome sequences at all, including peripheral and integral membrane proteins, subunits that are imported into the matrix in a piggybacked fashion, and matrix proteins with internal PTS1-like peptides (e.g., yeast acyl-CoA oxidase and plant CAT; Klein et al., 2002; Kamigaki et al., 2003).

In the postgenomic era of plant peroxisomal research, a major challenge is the identification and functional characterization of plant-specific proteins. Many of the recently cloned cDNAs of plant peroxisomal proteins have been identified by computational homology analysis using the protein sequence of peroxisomal enzymes from mammals or fungi as query (Edqvist et al., 2004; Tilton et al., 2004; Goyer et al., 2004; Goepfert et al., 2005, 2006). Even though proven to be successful, this approach fails by definition to unravel plant-specific proteins. Instead, systematic large-scale experimental studies of plant peroxisomes are required. Technical advances in two-dimensional gel electrophoresis (2-DE), liquid chromatography (LC), and mass spectrometry (MS) laid the foundation for the progress in organellar proteome studies (for review, see Taylor et al., 2003; Peck, 2005; Glinski and Weckwerth, 2006) and for cell-specific profiling (Wienkoop et al., 2004). The major bottleneck of organellar proteomics is the purity of the cell compartment of interest. With respect to their fragile nature, the isolation of intact peroxisomes in high purity is difficult irrespective of the source organism but is even more challenging in plants, mainly due to the predominance of plastids as another contaminating cell compartment and a myriad of secondary metabolites with organelle-destabilizing effects in vitro. For these reasons, the first proteome studies of peroxisomes have only recently been reported (for review, see Saleem et al., 2006). As the availability of complete genome information or large EST collections is a prerequisite for straightforward protein identification and because peroxisome isolation protocols are hardly interchangeable between plant species, such proteome studies were hampered by the fact that the isolation of plant peroxisomes in high purity was restricted to few plant species (e.g., Pisum sativum and Spinacia oleracea), which lacked genome or EST sequence information. Complete genome analysis of Arabidopsis (Arabidopsis Genome Initiative, 2000) rendered this tiny weed the species of choice for proteome studies of plant peroxisomes, provided that the organelles could be isolated in sufficient purity and quantity. Initial proteome analyses of leaf peroxisomes and glyoxysomes from greening and etiolated Arabidopsis cotyledons have been reported, but probably due to insufficient organelle purity and/or technical limitations in protein identification, only few known and novel PTS-containing proteins have been identified (Fukao et al., 2002, 2003). Leaf peroxisomes are considered the variant of choice for a comprehensive long-term project aimed at the identification and functional analysis of all plant-specific proteins from peroxisomes because mature leaves are (1) physiologically the most significant plant tissue (site of photosynthesis, photorespiration, numerous biosynthetic pathways, and nitrogen and sulfur assimilation); (2) the most distinctive tissue of plants compared with yeast and mammals; and (3) second only to roots, the major plant organ that interacts with the environment (i.e., that perceives and immediately reacts to stress, both abiotic [high light, extreme temperature, drought, and hail] and biotic [pathogen and herbivore attack]).

With the objective to verify and ultimately improve the prediction accuracy of plant peroxisomal proteins from genome sequences, we applied the unbiased large-scale approach of organellar proteomics to leaf peroxisomes. A purification method was developed for leaf peroxisomes from mature Arabidopsis leaves in high purity by Percoll and sucrose density gradient centrifugation. Protein separation by 2-DE in combination with peptide mass fingerprinting (PMF) and peptide fragment fingerprinting supplemented by a gel-free shotgun approach led to the identification of 36 known and 42 novel peroxisomal proteins. Peroxisome targeting of the novel PTS-containing proteins was corroborated by the conservation of PTSs in orthologs of higher plants as revealed by analysis of protein and EST databases. By fluorescence microscopy, peroxisome targeting of 11 representative proteins carrying predicted PTS or PTS-related peptides was verified for fusion proteins with enhanced yellow fluorescent protein (EYFP) in onion epidermal cells. Thereby, three novel PTS1 tripeptides (SSL>, SSI>, and ASL>) and an internal PTS2 were established, providing valuable input for further development of prediction tools. The novel plant peroxisomal proteins indicate that leaf peroxisomes play an important role in catabolism of complex metabolites, in biosynthetic pathways, and in defense mechanisms against pathogens and herbivores.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
Isolation of Peroxisomes from Mature Arabidopsis Leaves in High Purity
To identify plant-specific peroxisomal proteins by a large-scale proteomics approach, a new isolation method for leaf peroxisomes from Arabidopsis was established with the objectives of (1) increasing the stability of peroxisomes in aqueous solution, (2) reducing the physical interaction of peroxisomes with mitochondria and chloroplasts, a pronounced phenomenon in Brassicaceae, and (3) increasing the specificity of peroxisome enrichment by a combination of Percoll and sucrose density gradient centrifugation. To this end, 3- to 4-week-old plants were harvested at the end of the dark period and gently ground in a buffer of high osmolarity. After separation of intact chloroplasts, but notably without peroxisome sedimentation by differential centrifugation, the supernatant was laid directly on top of a discontinuous Percoll density gradient (Figure 1A ). Cosedimentation of peroxisomes along with mitochondria and thylakoid membranes by differential centrifugation prior to isopycnic organelle separation increased irreversibly interorganellar adhesion and peroxisome contamination. Whereas chloroplasts, thylakoid membranes, and mitochondria were largely retained in the 15 and 38% (v/v) Percoll fraction near the top of the gradient, intact leaf peroxisomes passed the Percoll layer and were recovered at the bottom, visible as a whitish diffuse organelle sediment (Figures 1B and 1C; see Supplemental Figure 1 online). Peroxisome fractions of several Percoll gradients were combined and washed in Tricine/EDTA buffer containing 36% (w/w) sucrose. Approximately 7% of total activity of the leaf peroxisomal marker enzyme hydroxypyruvate reductase (HPR) of the crude extract was recovered in the first fraction of leaf peroxisomes (LP-P1) with a yield of 90 nkat HPR per 60 g fresh weight (Table 1 ). The peroxisome fraction was carefully homogenized and laid on top of a discontinuous sucrose density gradient (Figure 1D; see Supplemental Figure 1 online). After ultracentrifugation, intact leaf peroxisomes were enriched close to the bottom of the gradient, visible as a sharp white band, at the typical density of plant peroxisomes of 50 to 53% (w/w) sucrose (Figures 1E and 1F; see Supplemental Figure 1 online). Even though only half of the peroxisomes could be recovered, the contamination by chloroplasts and mitochondria was further reduced by a factor of three (Table 1). On average, 115 ± 29 µg protein were obtained with a specific HPR activity of 293 ± 76 nkat/mg protein (Table 1). The purity of leaf peroxisomes was high, as determined by the activities of marker enzymes for leaf peroxisomes (HPR), mitochondria (fumarase), and chloroplasts (NADP-dependent glyceraldehyde-3-phosphate dehydrogenase [GAPDH]) and the content of chlorophyll (thylakoids); contaminating chloroplasts and mitochondria were estimated to comprise only 0.1 and 1.7% of the total, respectively (Table 1).

View larger version (25K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 1. Isolation of Peroxisomes from Mature Arabidopsis Leaves. The supernatant of the chloroplast sedimentation was loaded on top of a discontinuous Percoll gradient underlaid with fractions of increasing sucrose concentration (A). The Percoll concentrations refer to the concentrations prior to gradient centrifugation. After centrifugation, the Percoll gradient was fractionated from the top to the bottom into 21 2-mL fractions and analyzed for organellar marker enzyme activities, namely HPR (leaf peroxisomes) (B), fumarase (mitochondria), and NADP-dependent GAPDH (chloroplasts) (C). After washing, the peroxisome fractions of eight Percoll gradients were combined and homogenized, and the resuspension was laid on top of a discontinuous sucrose density gradient followed by ultracentrifugation. After fractionation from the top to the bottom into 14 1-mL fractions, the concentrations of protein and sucrose were determined (D). Marker enzyme activities were analyzed as before ([E] and [F]). Note that the scale between the first and second density gradients is increased by a factor of 5 for HPR and decreased for NADP-GAPDH (factor 5) and fumarase (factor 2).

View larger version (71K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 2. Two-Dimensional Protein Map of Isolated Leaf Peroxisomes from Arabidopsis. Leaf peroxisomal proteins (200 µg) of a specific HPR activity of 400 nkat HPR/mg protein were first separated according to pI by isoelectric focusing on commercial IPG strips (18 cm, nonlinear, pH 3.0 to 10) followed by SDS-PAGE on 10 to 15% (w/v) acrylamide gradient gels (A). For the sake of clarity, the annotations of the protein spots within the dashed frames are shown magnified in (B) and (C). Proteins were visualized by colloidal Coomassie blue staining. Stained proteins were excised manually and subjected to an automated platform for protein identification (Jahn et al., 2006). Two artificial Coomassie spots caused by dye crystals were eliminated from the figure. Abbreviations not listed in Supplemental Table 1 online or Table 2 are as follows: ACNA, (cytosolic) aconitase; alb., BSA; ATPase ß, ATP synthase subunit ß; efTU, elongation factor TU; GDC-P, Gly decarboxylase P protein; PSBP, oxygen-evolving enhancer protein 2; RubisCO LSU and SSU, ribulose-1,5-bis-phosphate carboxylase/oxygenase large and small subunits; SHMT, Ser-hydroxymethyltransferase; VDAC, voltage-dependent anion-selective channel.

Although two-dimensional gel-based proteomics approaches offer the advantage that protein integrity and information on protein abundance, isoform identity, and posttranslational modifications are retained, limitations with respect to the identification of small, basic, and hydrophobic proteins are reported. To complement the gel-based approach and improve coverage of the peroxisomal proteome, a gel-free technique was added that employs one-dimensional LC separation at the level of proteolytic peptides, commonly referred to as the shotgun approach (Link et al., 1999). This technique is very sensitive and requires only small amounts of total protein, as demonstrated recently by cell-specific protein profiling in pooled trichome and epidermis cells from Arabidopsis (Wienkoop et al., 2004). The proteins identified by this method largely overlapped with those of the gel-based approach; however, additionally, eight known and six novel proteins were detected only by the shotgun strategy. These proteins generally had either a basic isoelectric point (pI > 8.4) and/or were small proteins (<150 residues), underscoring the complementary nature of the two approaches applied. In total, we identified 78 nonredundant proteins, comprising 36 known and 42 novel proteins (Table 2 ; see Supplemental Table 1 online). Filtered mass spectra for protein identifications are stored in ProMEX, a mass spectral reference library for plant proteomics, that can be searched with peptide product ion spectra of unknown samples (Hummel et al., 2007).

View larger version (93K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 4. Verification of Peroxisome Targeting of Novel Peroxisomal Proteins Carrying Predicted Major or Minor PTS1s. The cDNAs of six novel peroxisomal proteins identified in isolated leaf peroxisomes from Arabidopsis and one pathway-associated protein (6PGL) were fused in frame to the 3' end of EYFP and transiently expressed in onion epidermal cells upon cell bombardment. Fluorescence microscopy was used to determine whether fusion proteins were targeted to punctate subcellular structures that coincided with peroxisomes labeled with a peroxisome-targeted version of CFP, MDH-CFP (Fulda et al., 2002). In addition, we analyzed the targeting of EYFP extended C-terminally by the putative 10–amino acid residue PTDs of the respective proteins. 6PGL was not yet identified in isolated leaf peroxisomes but analyzed in context of the detection of another peroxisomal isoform of the OPPP, 6PGDH. The following proteins were found to be targeted to peroxisomes by functional PTS1s (see text for full names): IDH (At1g54340, SRL>; A1, A2, 95% overlapping dots; [B]), SDRb (At3g12800, SKL>; [C1] and [C2]; 74% overlapping dots; [D]), HBCDH (At3g15290, PRL>; [E1] and [E2]; quantification difficult; [F]), 6PGL (At5g24400, SKL>; [G1] and [G2]; 95% overlapping dots; [H]), ECHIb (At4g14430, PKL>; [I1] and [I2]; 97% overlapping dots; [J]), and GSTT1 (At5g41210, SKI>; [K1] and [K2]; 95% overlapping dots [L]). The full-length fusion protein of 6PGDH (At3g02360) remained cytosolic (M), although it was shown to contain a functional PTS1 (SKI>; [N1] and [N2]; 91% overlapping dots).

View larger version (59K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 5. Subcellular Targeting Analysis of Novel Peroxisomal Proteins Carrying Unknown PTSs. The generation of EYFP fusion proteins and the fluorescence microscopic analyses were performed as described in Figure 4. Subcellular targeting analysis of ECH2 (At1g76150, SSL>; [A1] to [B2]; [A], 71% overlapping dots; [B], 95% overlapping dots), ATF1 (At1g21770, SSI>; [C1] to [D2]; [C], 60% overlapping dots; [D], 59% overlapping dots), and EH3 (At4g02340, ASL>; [E1] to [F2]; [E], 80% overlapping dots; [F], 73% overlapping dots) established SSL>, SSI>, and ASL> as novel functional PTS1 peptides in plants. Although analysis of full-length TLP324 (At5g58220) carrying a putative internal PTS2 (RLx5HL) at least indicated targeting to punctate subcellular structures (G), we could not provide conclusive evidence for peroxisomal localization of EYFP-TLP324 by double labeling owing to a strong cytosolic background fluorescence. When the essential His of the predicted internal PTS2 was mutated to Asp (RLx5HL to RLx5DL) to generate EYFP-TLPPTS2, the punctate fluorescence pattern disappeared (H), enabling us to identify the full-length version of TLP324, but not the two shortened splice variants lacking the predicted PTS2 (TLP311 and TLP286; see Figure 3) as a peroxisomal protein that is targeted to the organelle by an internal PTS2.

Even though fatty acid ß-oxidation is primarily performed by glyoxysomes of germinating seeds and is thought to play only a minor role in leaves, we detected not only a complete set of the core ß-oxidation enzymes but also several auxiliary enzymes in leaf peroxisomes. The core ß-oxidation enzymes included long-chain acyl-CoA synthetase isoform 7 (LACS7; At5g27600; SKL>); acyl-CoA oxidase isoform 4 (ACX4; At3g51840; SRL>); both Arabidopsis isoforms of multifunctional protein (i.e., multifunctional protein isoform 2 [MFP2; At3g06860; SRL>]) and abnormal fluorescence meristem 1 (AIM1; At4g29010; SKL>); and all three thiolase isoforms. Peroxisome Defective 1 (PED1/KAT2; At2g33150; RQx₅HL) and peroxisomal keto-thiolase isoform 2 (PKT2/KAT5; At5g48880; RQx₅HL) predominated on two-dimensional gels, whereas PKT1/KAT1 (At1g04710; RQx₅HL) was identified only by the shotgun approach.

The auxiliary enzymes of fatty acid ß-oxidation included ^3,5-^2,4- dienoyl-CoA isomerase (DCI; At5g43280; AKL>; Goepfert et al., 2005), enoyl-CoA hydratase isoform 2 (ECH2; At1g76150; Goepfert et al., 2006), sterol carrier protein isoform 2 (SCP-2; At5g42890; SKL>; Edqvist et al., 2004), and acyl-CoA thioesterase (ACH2; At1g01710; SKL>; Tilton et al., 2004). With respect to hormone biosynthesis, not only OPR3 but also 3-oxo-2-(2'-pentenyl)-cyclopentane-1-octanoic acid (OPC-8:0) CoA ligase 1 (OPCL1; At1g20510; SKL>), which activates OPC-8:0 (Koo et al., 2006), was detected. OPCL1 was represented by a low-abundance protein spot restricted to peroxisomes of highest purity, consistent with its low expression level under standard conditions and its induction by wounding. Two further acyl-CoA acitivating enzymes (AAEs) included AAE7/ACN1 (At3g16910; SRL>; Turner et al., 2005) and 4-coumarate-CoA ligase-like protein 1 (4CLP1; At4g05160; SKM>; Schneider et al., 2005). In agreement with the heat inducibility of one of the two sHsps from plant peroxisomes (HSP15.7-Px; Ma et al., 2006), only the constitutively expressed isoform ACD31.2-Px (At1g06460; RLx₅HF) was identified. As peroxisomal membranes were not further enriched prior to proteome analysis, the number of integral membrane proteins identified was low (APX3, PEX11d, and PEX14), which is consistent with the low content of membrane proteins in peroxisomes in general (Fujiki et al., 1982; S. Reumann, unpublished data).

In summary, the 36 known plant peroxisomal proteins identified in isolated leaf peroxisomes included all soluble enzymes involved in photorespiration, ROS metabolism, and JA biosynthesis. The detection of many core and auxiliary enzymes involved in fatty acid ß-oxidation and sulfur and nitrogen metabolism (see Supplemental Table 1 online) indicated that (1) the dynamic range of protein identification was sufficiently high to cover low-abundance enzymes that are predominantly expressed in peroxisome variants other than leaf peroxisomes and that (2) ß-oxidation–related enzymes may play a yet underestimated role in leaf peroxisomal metabolism.

Several matrix enzymes, although encoded by single Arabidopsis genes, often appeared on the two-dimensional gels as multiple protein spots arranged in a characteristic horizontal pattern. This pattern was most obvious for highly abundant matrix enzymes, namely GGT1, Ser-glyoxylate aminotransferase, HPR, and pMDH2, but also observed for minor enzymes, such as ascorbate peroxidase (APX3) and OPCL1 (Figure 2; see Supplemental Figures 2A to 2C online). The horizontal spot series of CAT3 differed slightly in that the differences in pI between spots were smaller (see Supplemental Figure 2D online). Although horizontal spot trains are often seen as artifacts of 2-DE, the uniform and regular patterns observed here might be indicative of the presence of charge-modifying posttranslational modifications, such as phosphorylation and acetylation.

Novel Plant Peroxisomal Proteins
Among the 42 novel proteins, a surprisingly large number had annotated functions related to fatty acid ß-oxidation (Table 2). One protein is a yet uncharacterized isoform of the extended superfamily of acyl-activating enzymes, namely AAE5 (At5g16370; SRM>), belonging to clade VI of largely unknown PTS1-carrying AAEs, as does AAE7/ACN1 (Shockey et al., 2003; Reumann et al., 2004; Turner et al., 2005). Regarding the superfamily of enoyl-CoA hydratases/isomerases (ECHIs; Reumann et al., 2004), with MFP and AIM1 being the most prominent enzymes, we now identified monofunctional members of plant peroxisomes. The first ECHI, referred to as isoform a (ECHIa; At4g16210; SKL>), belongs to clade VI of putative plant enoyl-CoA hydratases. Two further monofunctional ECHIs identified, namely, ECHIb (At4g14430; PKL>) and ECHIc (At1g65520; SKL>), have been grouped in clade III and are the closest Arabidopsis homologs of peroxisomal monofunctional ³-²-enoyl-CoA isomerase (PECI) from fungi and mammals (Geisbrecht et al., 1999; Reumann et al., 2004). Most intriguing of this enzyme family is ECHId, annotated as putative naphthoate synthase (NS/ECHId; At1g60550; RLx₅HL) because the enzyme is plant specific among eukaryotes and represents the Arabidopsis homolog of cyanobacterial MenB (clade V), an enoyl-CoA hydratase involved in phylloquinone biosynthesis (Johnson et al., 2001; Gross et al., 2006, see also references therein). This provides experimental evidence for peroxisome targeting of the single Arabidopsis homolog of MenB, further supported by conservation of the predicted PTS2 in homologous ESTs (see Supplemental Figure 3O online), and indicates that phylloquinone biosynthesis is partly compartmentalized in plant peroxisomes.

One protein spot was identified as a monofunctional hydroxyacyl-CoA dehydrogenase, annotated as hydroxybutyryl-CoA dehydrogenase (HBCDH; At3g15290) and carrying the predicted major PTS tripeptide PRL> (Reumann et al., 2004). Of the experimentally uncharacterized Arabidopsis family of small thioesterases (STs) with six PTS1-possessing members (Reumann et al., 2004), the second isoform of clade I (ST2; At1g04290; SNL>) is now experimentally associated with plant peroxisomes. Five novel proteins belong to the superfamily of short (polypeptide)-chain dehydrogenases/reductases (SDRs) and include four monofunctional SDRs (SDRa, At4g05530, SRL>; SDRb, At3g12800, SKL>; SDRc, At3g01980, SYM>; and SDRd, At3g55290, SSL>) and a bifunctional short- and medium-chain dehydrogenase/reductase (BSMDR; At1g49670; SRL>) with an N-terminal SDR and a C-terminal medium-chain dehydrogenase/reductase domain of the subgroup of NADPH:quinone reductases (COG0604, Qor). The latter was identified specifically by the gel-free approach.

Further novel enzymes with predicted catalytic activities related to fatty acid ß-oxidation included epoxide hydrolase isoform 3 (EH3; At4g02340; ASL>) and one of two Arabidopsis homologs of acetoacetyl-CoA thiolase (ACAT2, EC 2.3.1.9, At5g48230, which is 80% identical with ACAT1, At5g47720; Carrie et al., 2007). GOX3 (At4g18360; AKL>) is a yet unknown isoform that shares 85% identity at the amino acid level with GOX1/2 but is predominantly expressed in nonphotosynthetic tissue and presumably plays a role outside of photorespiration (Kamada et al., 2003; Reumann et al., 2004). Enzymes associated with peroxisomal NADP metabolism included isoforms of NADP-dependent isocitrate dehydrogenase (IDH; At1g54340; SRL>) and 6-phosphogluconate dehydrogenase (6PGDH; At3g02360; SKI>), one of three enzymes of the oxidative pentose phosphate pathway (OPPP). Several novel plant peroxisomal enzymes are predicted to catalyze GSH-dependent reactions. These identified proteins included an isoform of glutathione reductase (GR; At3g24170; TNL>), isoform 1 of the extended Arabidopsis family of glutathione S-transferases (GSTT1, formerly GST10, At5g41210, SKI>, EC 2.5.1.18; Dixon et al., 2002), and the single Arabidopsis homolog of S-(hydroxymethyl)glutathione dehydrogenase (HMGDH; At5g43940; EC 1.1.1.284; Dolferus et al., 1997).

Seven identified proteins are functionally associated with plant defense mechanisms, which was rather unexpected. Four enzymes are ß-glucosidases of the superfamily of glycoside hydrolase family 1, many of which are reported to mediate plant defense against herbivores (Xu et al., 2004; Table 2). The ß-glucosidases included ß-glucosidase 1 (BGL1; At1g52400; Stotz et al., 2000), PYK10 (At3g09260), and the myrosinase homolog thioglucoside glucohydrolase 2 (TGG2; At5g25980; Barth and Jander, 2006). All three defense-related enzymes were apparently upregulated in leaf peroxisomes isolated from senescent plants (see Supplemental Figures 2E and 2F online); TGG1 (At5g26000; Barth and Jander, 2006) was specifically identified in the latter (data not shown). The myrosinase-associated protein EPITHIOSPECIFIER MODIFIER1 (ESM1; At3g14210), which alters glucosinolate hydrolysis and insect resistance (Zhang et al., 2006), was identified in highly purified leaf peroxisomes as well. Two other defense proteins identified comprised one of three Arabidopsis homologs of vertebrate macrophage migration inhibitory factor (MIF; At3g51660; SKL>) and ozone-induced protein 1 (OZI1; At4g00860), a plant-specific small polypeptide of only 80 residues that is induced by ozone and pathogen attack (Sharma and Davis, 1995).

Likewise unexpected was the identification of proteins with annotated functions related to nucleic acid metabolism. These six proteins included two Gly-rich RNA binding proteins (GRP7 and 8; At2g21660 and At4g39260; Staiger et al., 2003), two DEAD-box RNA helicases (RH11, At3g58510; RH37, At2g42520), an unknown RNA binding protein (RBP; At4g17520), and a nucleotide binding protein (NBP; At4g16566). Except for NBP carrying the PTS1-like C-terminal tripeptide SKV>, all proteins related to nucleic acid metabolism lack predicted PTSs.

Proteins with other annotated functions included two enzymes reported to be involved in Cys and Met biosynthesis. O-acetylserine sulfhydrylase (OASS A1; At4g14880) is a presumably cytosolic isoform lacking an obvious PTS. Isoform 1 of cobalamin (vitamin B12)-independent Met synthase (METE1; At5g17920; EC 2.1.1.14; Ravanel et al., 2004) carries the PTS1-related C-terminal tripeptide SAK>. One of two Arabidopsis homologs of betaine aldehyde dehydrogenase (BADH; At3g48170; SKL>) was identified as well, homologs of which had previously been localized to the peroxisome matrix in monocotyledons (Nakamura et al., 1997). Another protein spot (ATF1; At1g21770; SSI>) was identified as a closely related homolog of a recently crystallized minimal acetyl transferase (ATF2; At1g77540; SSI>; Tyler et al., 2006). The single Arabidopsis transthyretin-like protein (TLP; At5g58220) is homologous to mammalian transthyretin, a transport protein for vertebrate hormones (Eneqvist et al., 2003; Hennebry et al., 2006a, 2006b). Interestingly, the full-length protein contains a predicted PTS2 (RLx₅HL) outside of the N-terminal domain in a linker region that connects two enzymatic domains, whereas two alternative splice variants from Arabidopsis lack the putative targeting signal (Figure 3A ; see Supplemental Figure 3P online; Hennebry et al., 2006a; Ramazzina et al., 2006). By MS sequencing of three tryptic peptides covering major parts of the linker region, including the putative PTS2, the TLP homolog identified in Arabidopsis peroxisomes was demonstrated to represent the full-length protein, strengthening the idea that the internal PTS2 peptide targets full-length TLP to peroxisomes (Figure 3). Lastly, four novel proteins lacked annotated domains and sequence similarity with known proteins deposited in the protein databases; three of these unknown proteins (UPs) carried PTS-like tripeptides (UP1, At1g52410, SSL>; UP2, At3g15950, SLN>; and UP3, At2g31670, SSL>; Table 2).

View larger version (34K): [in this window] [in a new window] [as a PowerPoint slide]   Figure 3. Domain Structure of Arabidopsis Transthyretin-Like Protein and Its Splice Variants. The single Arabidopsis TLP (At5g58220) is a bifunctional protein with an N-terminal decarboxylase (COG3195, OHCU decarboxylase) and a C-terminal hydrolase domain (COG2351, 5-HIU hydrolase) synthesized in three splice variants (A). The full-length variant (TLP324) carries a predicted PTS2 (RLRIIGGHL, position 182 to 190) in the linker region between the two functional domains. In addition to the peptide mass information obtained (shown in [A] as black bars with numbering of tryptic peptides), mass spectrometric sequencing by MALDI-TOF-MS unambiguously confirmed the presence of the tryptic peptides T17, T19, and T22 (shown in [A] also as sequences, PTS2 amino acids in bold face), thereby indicating the integrity of the linker region of full-length TLP324. Except for the tryptic dipeptide T18 (LR) that escaped detection due to its small size, the entire PTS2 nonapeptide was covered by the sequencing analysis, as can be deduced from the fragment ion mass spectrum of T19 (B). By contrast, both shorter splice variants of TLP (TLP311 and TLP286) lack the internal PTS2 and TLP286 additionally the bulky conserved His residue (His209 of TLP324, corresponding to His14 of TLP/PucM of Bacillus subtilis) that is implicated in determining the entrance diameter of the internal channel (Jung et al., 2006). Thus, the shorter transcriptional variants probably differ in subcellular localization and function from the peroxisomal isoform. See Supplemental Figure 3P online for details of the central amino acid sequence of the three Arabidopsis TLP splice variants and plant homologs. OHCU, 2-oxo-4-hydroxy-4-carboxy-5-ureidoimidazoline.

On average, 30 plant ESTs were retrieved for each novel protein. Except for one EST homologous to SDRa terminating with FRL>, the orthologs of nine novel proteins with predicted PTS1 (ECHIa/b/c, ST2, SDRa/c, BSMDR, GSTT1, and MIF) carried approximately four different C-terminal tripeptides each, all of which represented major or minor PTS1s (see Supplemental Figure 3 online). Five other novel proteins had previously been reported to carry conserved PTS1 (SDRb/DECR, HBCDH, IDH, 6PGL, and 6PGDH; Reumann et al., 2004). Likewise, the postulated peroxisome-targeting function of the PTS1-related tripeptide SSI> present in ATF1 was supported by the detection of several homologous ESTs possessing either minor noncanonical PTS1s (SSM> or SNM>) or closely related tripeptides, such as SHI> and SNI> (see Supplemental Figure 3N online). In addition, most plant homologs of ATF1 carried a basic residue at position –4, which is reported to enhance peroxisome targeting by weak PTS1 tripeptides (Mullen et al., 1997; Maynard et al., 2004). Regarding SDRc (SYM>), only PTS1-like peptides were present in homologous ESTs (e.g., SF[ML]>, AYM>; see Supplemental Figure 3H online). However, the high content of basic amino acid residues and Pro in front and the lack of acidic residues in the C-terminal tripeptides themselves were consistent with protein targeting by the PTS1 pathway. Although a dissection of plant homologs of ACAT2 into apparent orthologs of ACAT1 or ACAT2 was difficult, many EST sequences were found to carry C-terminal 10–amino acid extensions either terminating with predicted (SKL>) or novel PTS1s (SSL>; see below) or PTS1-like tripeptides (SAL> or SLL>; see Supplemental Figure 3J online).

Regarding novel proteins carrying predicted PTS2, the peroxisome targeting function of the predicted PTS2 in naphthoate synthase (NS/ECHId; At1g60550; RLx₅HL) was well supported by conservation of the nonapeptide as major or minor PTS2 (R[LIMV]x₅HL) in plant orthologs. Moreover, characteristic amino acid residues predicted to act as targeting-enhancing elements (Arg and Pro) were present in close proximity to the PTS2 (Reumann, 2004; see Supplemental Figure 3O online). Likewise, the predicted internal PTS2 (RLx₅HL) of TLP was conserved as putative major or minor PTS2s in all homologous plant ESTs spanning this region (e.g., R[LIMV]x₅HL and RIx₅HM) except for one slightly divergent nonapeptide (RVx₅HV) present in Medicago (see Supplemental Figure 3P online). Even though full-length TLP variants appear to predominate among plant ESTs, some plant species, such as Oryza and Triticum, were found to express alternative splice variants similar to Arabidopsis that exactly lack this nonapeptide (Figure 3B; see Supplemental Figure 3P online; Hennebry et al., 2006a).

Experimental Verification of Predicted PTSs
To experimentally verify protein targeting to peroxisomes and the peroxisome-targeting function of predicted PTSs, subcellular targeting of seven representative fusion proteins with EYFP was analyzed by fluorescence microscopy upon transient cDNA expression in onion epidermal cells. Four proteins carried predicted major PTS1 peptides (SRL>, SKL>, or PRL>), whereas three proteins possessed predicted minor PTS1s (SKI> or PKL>) of lower peroxisome-targeting probability (Reumann, 2004). For all proteins, subcellular targeting of two constructs was analyzed (i.e., EYFP fusions of the full-length proteins of interest and fusions attaching only the predicted PTS1 targeting domain [PTD], comprising the C-terminal 10–amino acid residues, including the putative PTS1 tripeptide, to EYFP).

NADP-dependent IDH (At1g54340; SRL>) was chosen as one of the proteins carrying predicted major PTS1 peptides. NADP-dependent IDH, fused at its N terminus with EYFP (EYFP-IDH), was indeed detected in punctate structures in onion epidermal cells (Figure 4A 1). In double transformants expressing simultaneously a peroxisome-targeted cyan fluorescent protein (CspMDH-CFP; Fulda et al., 2002; Ma et al., 2006), the yellow fluorescent subcellular structures were identified conclusively as peroxisomes (Figures 4A1 and 4A2). EYFP extended by the C-terminal 10–amino acid residues of IDH, including the predicted PTS1 SRL> (EYFP-PTD_IDH), was directed from the cytosol to punctate structures, thereby identifying SRL> as the PTS of peroxisomal IDH (Figure 4B). Two further proteins, namely, the full-length fusion proteins of the predicted DECR homolog SDRb (At3g12800; SKL>) and the monofunctional hydroxyacyl-CoA dehydrogenase HBCDH (At3g15290; PRL>), were targeted to peroxisomes as well (Figures 4C1, 4C2, 4E1, and 4E2). EYFP extended by the PTDs of these proteins showed a punctate fluorescence pattern, characterizing the C-terminal tripeptides SKL> and PRL> as the targeting signals of SDRb/DECR and HBCDH, respectively (Figures 4D and 4F).

Regarding proteins carrying predicted minor PTS1 peptides, the fusion protein between EYFP and ECHIb (At4g14430; PKL>), one of two predicted homologs of mammalian PECI (Geisbrecht et al., 1999), was targeted to punctate structures in onion epidermal cells that coincided with CFP-labeled peroxisomes (Figures 4I1 and 4I2). EYFP, extended by the C-terminal 10–amino acid residues of ECHIb, including the predicted PTS1 PKL> (EYFP-PTD_ECHIb), was directed from the cytosol to punctate structures, thereby identifying the minor PTS1 PKL> as the PTS of peroxisomal ECHIb (Figure 4J). Likewise, the isoform GSTT1 (At5g41210) with the putative PTS1 SKI> was shown to be located in peroxisomes, as the yellow fluorescent spots coincided with cyan fluorescent peroxisomes upon expression of EYFP-GSTT1 (Figures 4K1 and 4K2). The transferase was imported into peroxisomes via the PTS1 pathway because attachment of the C-terminal 10–amino acid domain of GSTT1 to EYFP showed a punctate pattern of fluorescence (Figure 4L). The full-length protein of the NADPH-producing dehydrogenase 6PGDH (At3g02360; SKI>) fused to the C-terminal end of EYFP was the only full-length fusion protein that remained cytosolic, for yet unknown reasons (Figure 4M). The C-terminal 10–amino acid residues of 6PGDH, however, directed EYFP to punctate subcellular structures that moved along cytoplasmic strands and coincided with peroxisomes (Figures 4N1 and 4N2). Although not identified within the proteome analysis, subcellular targeting of 6P-gluconolactonase (6PGL), the precedent enzyme of 6PGDH in the peroxisomal OPPP (Corpas et al., 1998, 1999), was also investigated using one specific isoform (At5g24400) out of five Arabidopsis homologs. This enzyme carries both a predicted N-terminal transit peptide and a conserved PTS1 (SKL>; Reumann et al., 2004). Whereas 6PGL-EYFP was targeted to plastid-like structures consistent with the presence of a functional N-terminal transit peptide (see Supplemental Figure 4 online), the inversely arranged fusion protein, EYFP-6PGL, was targeted to peroxisomes (Figures 4G1 and 4G2), as was EYFP-PTD_6PGL with the PTS1 SKL> (Figure 4H). These data further strengthen the idea that this 6PGL isoform is targeted to both plastids and peroxisomes in vivo, probably regulated by alternative translation (Reumann et al., 2004).

Taken together, predicted peroxisome targeting of novel proteins identified in isolated leaf peroxisomes was experimentally confirmed for five of the six full-length candidate proteins and additionally for 6PGL as a metabolically related enzyme; all predicted PTS1s were characterized as the functional targeting signals of the respective proteins. Peroxisome targeting of full-length 6PGDH may have been disturbed by the N-terminal EYFP tag or depend on yet unknown auxiliary import factors missing in the standard expression system of onion epidermal cells.

Characterization of Novel Targeting Peptides
We next investigated whether representative novel proteins carrying PTS1-related peptides at the extreme C terminus or PTS2 peptides outside of the defined N-terminal domain are indeed imported into peroxisomes and whether these peptides function as targeting signals. The three PTS1-related tripeptides chosen for these experiments were SSL>, SSI>, and ASL>. The full-length fusion protein composed of EYFP and Arabidopsis enoyl-CoA hydratase ECH2 (At1g76150; SSL>) was targeted to peroxisomes (Figures 5A1 and 5A2). Extension of EYFP by the C-terminal 10–amino acid residues of ECH2 directed the reporter protein to punctate structures that coincided with peroxisomes (Figures 5B1 and 5B2); thus, the C-terminal PTS1-like tripeptide SSL> is the PTS1 of ECH2 and a novel functional PTS1 tripeptide in higher plants. Likewise, EYFP-ATF1 (At1g21770) terminating with SSI> was detected in peroxisomes (Figures 5C1 and 5C2). The putative PTS1 domain of this protein directed EYFP to punctate structures that coincided with peroxisomes, establishing SSI> as a second novel plant PTS1 (Figures 5D1 and 5D2). For epoxide hydrolase isoform 3 (EH3; At4g02340) carrying the C-terminal PTS1-like tripeptide ASL>, organelle targeting was difficult to demonstrate owing to a pronounced cytosolic background staining. Nevertheless, protein targeting to peroxisomes could be demonstrated by double labeling in a few cells (Figures 5E1 and 5E2). Attachment of the C-terminal 10–amino acid residues of EH3 to EYFP directed the reporter protein from the cytosol to punctate structures that coincided with peroxisomes (Figured 5F1 and 5F2). With these results, we conclusively demonstrated targeting of another three proteins to peroxisomes (ECH2, ATF1, and EH3) and established the PTS1-like peptides SSL>, SSI>, and ASL> as novel functional PTS1 peptides for higher plants.

By current definition, PTS2 nonapeptides are confined to the N-terminal domain comprising 40 amino acid residues. To determine whether the bifunctional TLP identified in Arabidopsis leaf peroxisomes (Figures 2 and 3, Table 2) was targeted to peroxisomes by the internally located putative PTS2 (RLx₅HL), the full-length cDNA of TLP was cloned downstream of EYFP and transiently expressed in onion epidermal cells. The fusion protein EYFP-TLP was detected mainly in the cytosol. However, despite the strong background fluorescence, targeting of EYFP-TLP to punctate structures suggestive of peroxisomes was observed in some cells (Figure 5G ). The low contrast between the yellow fluorescent punctate structures and the background, however, allowed conclusive organelle identification as peroxisomes by double labeling only in a single cell (data not shown). In an alternative approach, the His residue known to be essential for the targeting function of PTS2 (Kato et al., 1998; Reumann, 2004; Ma et al., 2006) was mutated to Asp (RLx₅HLRLx₅DL). Indeed, the mutated fusion protein EYFP-TLPPTS2 was no longer targeted to peroxisomes but exclusively detected in the cytosol and the nucleus, as was EYFP alone (Figure 5H). Thus, these data demonstrate targeting of a eukaryotic TLP homolog to peroxisomes and an internal location of a functional PTS2 in an endogenous eukaryotic protein.

Analysis of Artificial Copurification of Large Protein Complexes with Leaf Peroxisomes
The defense- and nucleic acid metabolism-related proteins identified in isolated leaf peroxisomes were represented by a relatively high number of proteins lacking known PTSs. Some of these defense-related proteins, such as BGL1, PYK10, and Brassica myrosinases, are reported to form polymers (Eriksson et al., 2002; Nagano et al., 2005; Lee et al., 2006). Likewise, nucleic acid metabolism-related proteins may be components of large protein complexes, such as ribosomes, exosomes, spliceosomes, or P-bodies (for review, see Eulalio et al., 2007). Even though unlikely, it was conceivable that these proteins were accidentally copurified with leaf peroxisomes through large complexes of similar sedimentation behavior in density gradient centrifugation. To distinguish between a physiologically relevant or artificial association of defense- and nucleic acid metabolism-related proteins with leaf peroxisomes, the isolation protocol was modified in such a way that intact leaf peroxisomes were lysed in hypoosmotic media by conditions that are not expected to reduce the sedimentation rate of protein complexes. To this end, Arabidopsis leaves were worked up in grinding and washing buffer lacking sucrose under otherwise identical conditions. Indeed, the fractions of the sucrose density gradient corresponding to those of intact leaf peroxisomes (fractions 11 and 12, Figure 1) contained substantially less total protein and largely lacked HPR activity, indicating that intact leaf peroxisomes were efficiently lysed and no longer recovered (data not shown). When such a fraction volume corresponding to 200 µg leaf peroxisomal protein in the presence of osmoticum was subjected to 2-DE, none of the defense-related proteins (BGL1, TGG2, PYK10, or ESM1) were detected by Coomassie staining (see Supplemental Figure 5 online). Similarly, the nucleic acid metabolism-related proteins GRP7 and 8 and RBP were not visible. Although two very faint spots may coincide with the RNA helicases RH11 and RH37 (see Supplemental Figure 5 online), these data as a whole strongly argue against an artificial copurification of high molecular weight complexes formed by the defense- or nucleic acid metabolism-related proteins with leaf peroxisomes and support a physiologically relevant association of most if not all of these PTS-deficient proteins with leaf peroxisomes.

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
Only 50 plant peroxisomal matrix proteins have been established for Arabidopsis either experimentally or by homology to peroxisomal proteins from other plant species (see summary at www.araperox.uni-goettingen.de). The low number of known peroxisomal enzymes in general and of plant-specific proteins in particular severely restricts our current knowledge of the organelle's physiological functions and alternative protein targeting mechanisms. The two most promising large-scale approaches to ultimately characterize the proteome of plant peroxisomes in its entity are (1) bioinformatics analyses predicting PTS1 and PTS2 proteins from genome sequences in combination with subcellular targeting studies of candidate fluorescent fusion proteins, and (2) proteome analyses. As in silico predictions are currently limited by several parameters (see Introduction) and because highly pure peroxisomes can finally be isolated from Arabidopsis, proteome analysis is the method of choice. This study demonstrates that proteome data yield important information not only directly on the identity of novel proteins and the recognition of newfound metabolic pathways but also indirectly on protein targeting. Leaf peroxisomal proteome analysis enables us to (1) recognize novel targeting peptides, (2) extend the location of PTS2 domains, and (3) posit the existence of novel PTS1/2-independent targeting pathways.

Plant-specific peroxisome functions, such as photosynthesis-related pathways, hormone biosynthesis, and pathogen defense mechanisms, are expected to be most comprehensively depicted by proteome analyses of Arabidopsis peroxisomes from mature rosette leaves rather than heterotrophic suspension-cultured cells or germinating seeds, even though these tissues offer significant technical advantages. The challenges in isolating highly pure Arabidopsis leaf peroxisomes include not only obvious traits (e.g., low leaf fresh weight and predominance of chloroplasts) but also unexpected parameters, such as the extreme fragility of Arabidopsis leaf peroxisomes in aqueous extracts, probably owing to the high concentrations of secondary metabolites, and their pronounced physical association with chloroplasts and mitochondria in Brassicaceae. For in-depth proteome analysis, we optimized our previous method (Ma et al., 2006) in several aspects, such as plant growth and time of harvest, differential centrifugation, and by the use of both Percoll and sucrose density gradients in particular (see Results and Methods). Contaminating proteins of chloroplasts and mitochondria, the major contaminants of plant peroxisomes, were hardly detectable in the final fraction by sensitive biochemical methods. The higher apparent content of contaminants estimated from the spot representation on two-dimensional gels (5%) compared with the activities of compartment-specific marker enzymes (0.1% plastids and 1.7% mitochondria; Table 1) may reflect facilitated solubilization of the rather acidic nonperoxisomal proteins compared with the neutral or basic peroxisomal proteins.

The Proteome Map of Arabidopsis Leaf Peroxisomes
The success of organellar proteome studies is determined by the number of novel true positives that indeed reside in the compartment of interest. The gel-based and shotgun approaches applied as complementary protein identification strategies maximized proteome coverage. The 78 proteins identified represent approximately twice as many proteins as reported previously in peroxisomes from rat liver (Kikuchi et al., 2004) or plants (Fukao et al., 2002, 2003). The 33 known matrix proteins include 65% of all currently known matrix proteins from plant peroxisomes (51 Arabidopsis proteins; Table 3 ). As this ratio has been obtained from only one of several plant peroxisomal variants and from standard plants, the coverage is considered high. Comprehensive proteome analyses of other tissue-specific and developmental peroxisome variants (e.g., glyoxysomes and gerontosomes) and from plants subjected to abiotic or biotic stresses are expected to promote the detection of the remaining known and further predicted proteins. A comprehensive membrane proteome analysis will be made possible by the availability of larger organelle quantities and the selective enrichment of peroxisomal membranes.

The horizontal spot pattern of several matrix enzymes is suggestive of charge-modifying posttranslational modifications, such as phosphorylation and acetylation, both of which are currently being investigated in greater detail. Some peroxisomal protein kinases are reported and several are predicted, along with phosphatases, to be matrix targeted (Fukao et al., 2002, 2003; Dammann et al., 2003; Reumann et al., 2004). Acetylation of internal Lys residues also decreases the pI of a polypeptide and has recently been recognized as an important regulatory signal in many cellular processes (for reviews, see Polevoda and Sherman, 2002; Fuchs et al., 2006). ATF2 (At1g77540; SSI>), whose x-ray structure has recently been resolved, is a minimal acetyl transferase of unknown subcellular localization (Tyler et al., 2006). We have now established its closely related homolog, ATF1 (At1g21770; SSI>; 80% identical with ATF1) as a novel constituent of plant peroxisomes by multiple lines of evidence, including in vivo EYFP targeting studies (Figures 5C and 5D). Conservation of the novel PTS1 SSI> of ATF1 in ATF2 strongly suggests that the entire two-member protein family is peroxisomal in Arabidopsis. Protein acetylation may thus be an as yet unrecognized regulatory mechanism of plant peroxisomal enzymes. Proteins showing multiple horizontal spot patterns (see Supplemental Figure 2 online) are reasonable candidate substrates of ATF1.

Targeting Verification for Improving Targeting Prediction
By in silico analysis for conservation of the predicted PTS in homologous plant ESTs (Reumann, 2004), the peroxisome targeting function of the predicted PTS peptides was supported for 17 of 28 novel proteins (Table 2; see Supplemental Figure 3 online; Reumann et al., 2004). Exceptions included novel proteins, whose PTS1-like peptides were unaltered in homologous ESTs, and Brassicaceae-specific proteins. Notably, no evidence against peroxisomal targeting, for instance, acidic amino acid residues at any position of the tripeptide, was obtained for any of the proteins. By subcellular analysis of reporter protein fusions, peroxisome targeting of six out of seven representative novel proteins with predicted PTS1s was confirmed. Artificial localization data possibly caused by cDNA expression from the strong 35S cauliflower mosaic virus promoter were excluded by colocalization of the proteins of interest with a cyan fluorescent peroxisome marker, as applied in previous studies (Fulda et al., 2002; Goepfert et al., 2006; Ma et al., 2006