AraPerox. A database of putative Arabidopsis proteins from plant peroxisomes

Genome Analysis

AraPerox. A Database of Putative Arabidopsis Proteinsfrom Plant Peroxisomes1[w]

Sigrun Reumann*, Changle Ma2, Steffen Lemke2,3, and Lavanya Babujee

Georg-August-University of Goettingen, Albrecht-von-Haller-Institute for Plant Sciences,Department for Plant Biochemistry, D–37077 Goettingen, Germany (S.R., C.M., L.B.); andMolecular Biology Graduate Program, Georg-August-University of Goettingen, GZMB,D–37077 Goettingen, Germany (S.L.)

To identify unknown proteins from plant peroxisomes, the Arabidopsis genome was screened for proteins with putative majoror minor peroxisome targeting signals type 1 or 2 (PTS1 or PTS2), as defined previously (Reumann S [2004] Plant Physiol 135:783–800). About 220 and 60 proteins were identified that carry a putative PTS1 or PTS2, respectively. To further supportpostulated targeting to peroxisomes, several prediction programs were applied and the putative targeting domains analyzedfor properties conserved in peroxisomal proteins and for PTS conservation in homologous plant expressed sequence tags. Themajority of proteins with a major PTS and medium to high overall probability of peroxisomal targeting represent novelnonhypothetical proteins and include several enzymes involved in b-oxidation of unsaturated fatty acids and branched aminoacids, and 2-hydroxy acid oxidases with a predicted function in fatty acid a-oxidation, as well as NADP-dependentdehydrogenases and reductases. In addition, large protein families with many putative peroxisomal isoforms wererecognized, including acyl-activating enzymes, GDSL lipases, and small thioesterases. Several proteins are homologous toprokaryotic enzymes of a novel aerobic hybrid degradation pathway for aromatic compounds and proposed to be involved inperoxisomal biosynthesis of plant hormones like jasmonic acid, auxin, and salicylic acid. Putative regulatory proteins of plantperoxisomes include protein kinases, small heat shock proteins, and proteases. The information on subcellular targetingprediction, homology, and in silico expression analysis for these Arabidopsis proteins has been compiled in the publicdatabase AraPerox to accelerate discovery and experimental investigation of novel metabolic and regulatory pathways ofplant peroxisomes.

Peroxisomes constitute ubiquitous eukaryotic cellorganelles in which a large variety of oxidative met-abolic reactions are compartmentalized. Plant perox-isomes deserve increasing attention because the size ofthe peroxisomal proteome from plants seems to bemuch larger than that of yeast, trypanosomes, insects,and mammals (Emanuelsson et al., 2003) and becauseplants possess an exceptionally large number of met-abolically specialized microbodies. Apart from theirwell-known function in photorespiration and lipidmobilization plant peroxisomes also play a significantrole in nitrogen metabolism (Verma, 2002), degrada-tion of branched amino acids (Zolman et al., 2001), andbiosynthesis of plant hormones including jasmonicacid (JA) and auxin (Stintzi and Browse, 2000; Zolman

et al., 2001; Feussner and Wasternack, 2002), as well asin the production of the compatible osmo-solute Glybetaine (Nakamura et al., 1997).

Flavin-containing oxidases like glycolate oxidase,acyl-CoA oxidase, urate oxidase, and amine oxidase,which transfer electrons directly to molecular oxygenand generate the highly toxic by-product hydrogenperoxide (H2O2), play a central role in most peroxi-somal pathways. The high concentration of catalase(CAT) of up to 10% to 25% of the organelle’s proteinguarantees rapid detoxification of H2O2 at the site ofproduction and prevents, under normal conditions,leakage of the membrane-permeable molecule out ofperoxisomes and oxidation of extraperoxisomal pro-teins. Apart from a strict compartmentalization ofintraperoxisomal H2O2, peroxisomes are discussed toplay an important role in detoxification of reactiveoxygen species (ROS) generated outside of peroxi-somes and to be a source of ROS acting as signalmolecules (Willekens et al., 1997; Corpas et al., 2001;Titorenko and Rachubinski, 2004). The significance ofperoxisomes in tolerance of oxidative stress is furthersupported by a pronounced proliferation of plantperoxisomes upon application of oxidative stress(Lopez-Huertas et al., 2000). Further antioxidativeenzymes, such as superoxide dismutase, membrane-bound ascorbate peroxidase, monodehydroascorbatereductase, and glutathione peroxidase have been

1 This work was supported by the Deutsche Forschungsgemein-schaft (Re 1304/2–1) and by a Dorothea-Erxleben stipend from thegovernment of Lower Saxony (to S.R.).

2 These authors contributed equally to the paper.3 Present address: Department of Organismal Biology and Anat-

omy, University of Chicago, Cummings Life Science Centre, Room915, 920 E. 58th Street, Chicago, IL 60637–1508.

* Corresponding author; e-mail [email protected]; fax 49–551–395749.

[w]The online version of this article contains Web-only data.www.plantphysiol.org/cgi/doi/10.1104/pp.104.043695.

Plant Physiology, September 2004, Vol. 136, pp. 2587–2608, www.plantphysiol.org � 2004 American Society of Plant Biologists 2587

partly cloned or localized by biochemical means toplant peroxisomes and are thought to play an auxil-iary role in detoxification of ROS (Bunkelmann andTrelease, 1996; Kliebenstein et al., 1998; del Rio et al.,2002). Peroxisomal Ser-glyoxylate aminotransferasefrom Cucumis melo has only recently been shown toconfer enzymatic resistance against downy mildewcaused by the pathogen Pseudoperonospora cubensispossibly via enhanced production of photorespiratoryH2O2 (Taler et al., 2004). Apart from classical enzymesthat catalyze metabolic reactions, our knowledge onperoxisomal matrix proteins is rather limited due todifficulties in identifying low-abundance and induc-ible proteins by biochemical approaches. Evidence forthe existence of regulatory proteins in peroxisomes,such as heat shock proteins, kinases, and phospha-tases, is just emerging.

Most known peroxisomal matrix proteins are tar-geted to the peroxisomal matrix by the peroxisomaltargeting signal type 1 (PTS1), the C-terminal so-calledSKL-motif, whereas other matrix proteins carry a PTS2,which is a conserved nonapeptide of the prototypeRLx5HL that is embedded in the N-terminal domain(Gould et al., 1989; Swinkels et al., 1991; see alsoreferences in Reumann, 2004). Plant-specific PTS motifshave been deduced from experimental targeting stud-ies in vivo and provided valuable information. Theytend, however, to yield partially contradictory results,to deduce motifs of low specificity, and to neglect theproven role of accessory elements located in closeproximity to the PTS and displaying an auxiliarytargeting function (Hayashi et al., 1997; Mullen et al.,1997a; Kragler et al., 1998; Kato et al., 1998; Flynn et al.,1998). From an in silico study, in which protein andexpressed sequence tag (EST) databases were searchedfor a maximum number of sequences that are homol-ogous to PTS1- and PTS2-targeted plant peroxisomalproteins, it has been concluded that only a smallnumber of nine major PTS1 and two major PTS2peptides are widespread in plant proteins of differentorthologous groups and represent strong indicators forperoxisomal targeting (Reumann, 2004).

To identify novel proteins located in plant perox-isomes, we screened the Arabidopsis genome (TheArabidopsis Genome Initiative, 2000) for genes encod-ing proteins with putative PTSs. As deduced from theresults of various targeting prediction programs, ex-amination of the targeting domain for propertiesconserved in plant peroxisomal proteins, and analysisof PTS conservation in homologous ESTs, a largenumber of novel proteins are localized in peroxisomeswith high targeting probability. The presence of someof these enzymes in plant peroxisomes has beendemonstrated earlier by biochemical means and cannow be supported by genomic evidence. Other pro-teins have not been predicted in plant peroxisomesyet, and their postulated localization in plant perox-isomes yields surprising insights into the complexityof catabolic and biosynthetic pathways of plant per-oxisomes.

RESULTS

Extraction of Arabidopsis Proteins Carryinga Putative PTS

Novel proteins of a particular cell compartment canpossibly be detected by screening eukaryotic genomesfor genes encoding proteins that carry the respectivetargeting signals. In contrast to targeting signals ofperoxisomal membrane proteins, which have only beenidentified in a few cases and for which consensussequences remain to be deduced (Jones et al., 2001), thetargeting motifs of soluble matrix proteins have beenspecified by experimental and in silico studies. Ninemajor PTS1 tripeptides have been defined previouslyand are considered to indicate peroxisomal targetingwith high probability ([SA][RK][LM].without AKM.plus SRI. and PRL.; Reumann, 2004). In our initialscreen for PTS1-targeted proteins, these major PTS1peptides were supplemented by 11 minor PTS1 tripep-tides, which comprised 5 canonical PTS1 tripeptides(SKI., PRM., PKL., C[RK]L.) of the relatively re-strictive Hayashi motif ([SAPC][KR][LMI].; Hayashiet al., 1997) and 6 minor noncanonical PTS1 tripeptides(SRV., [SA]NL., SML., SNM., SSM.; Reumann,2004). The tripeptides SHL. and AHL. were addedas well because they have been shown to representefficient targeting peptides in yeast, to direct a passen-ger protein to plant peroxisomes, or to interact withtobacco Pex5 in the yeast-two-hybrid system (Elgersmaet al., 1996; Mullen et al., 1997a; Kragler et al., 1998;Lametschwandtner et al., 1998; Fig. 1A). This strategyallowed the detection of a maximum number of trueperoxisomal proteins while reducing the number offalse positives to a minimum, and was combined witha subsequent detailed analysis of targeting predictionfor identification and exclusion of nonperoxisomalproteins.

About 220 different open reading frames (ORFs)were extracted out of the Arabidopsis genome (alter-native splice or translation variants not counted) thatcarried one of these 22 PTS1 tripeptides. About 130proteins (59%) of these carried a major PTS1. Thelargest number of proteins contained the PTS1 SKL.(43; 0.15% of all predicted Arabidopsis proteins) orSRL. (31 proteins), followed by those with S[RK]I.,A[RK]L., or PKL. (each 9–16 proteins), and thenS[RK]M., and PRL. (each 6 proteins; Fig. 1A). About90 proteins carried a minor PTS1 or the tripeptides[SA]HL., of which a considerable number of 50proteins contained a noncanonical PTS1 tripeptide(SRV., SNL., and SSM., each 10–13 proteins;SML. and ANL., each 6–7 proteins; Fig. 1A). Ifrandom distribution of amino acids is assumed, onaverage 3.6 of the 28,800 predicted Arabidopsis pro-teins are expected to contain one of 8,000 possibleC-terminal tripeptides. Most of the PTS1 tripeptidesthat are present in a large number of Arabidopsis pro-teins well above average have recently been defined asmajor tripeptides and are considered strong indicators

Reumann et al.

2588 Plant Physiol. Vol. 136, 2004

for peroxisomal targeting. This rough correlation mostlikely reflects the late evolutionary stage of optimizingthe efficiency of peroxisomal targeting peptides andprovides further support for a specific targeting func-tion of these peptides. Vice versa, most canonicaltripeptides of the Hayashi motif (e.g. AKM., PKI.,CR[MI]., CKM.) that are present in a low number ofArabidopsis proteins (#2 proteins; data not shown),were also rare in plant homologs of PTS1-targeted per-oxisomal proteins and are not considered to target pro-teins efficiently to plant peroxisomes (Reumann, 2004).

In a similar way, about 60 proteins with one of twomajor (R[LI]x5HL) or nine minor PTS2 nonapeptides(R[QTMAV]x5HL, RLx5H[IF], and R[IA]x5HI) locatedin an N-terminal domain of defined size were extractedout of the Arabidopsis genome. The two major PTS2nonapeptides, RIx5HL and RLx5HL, were most fre-quent in Arabidopsis proteins (17 and 14 proteins,respectively), whereas a lower number carried one ofthe nine minor PTS2 peptides (1–5 proteins; Fig. 1B).Considering the number of 160,000 different tetrapep-tides (i.e. nonapeptides with four conserved residues at

Figure 1. Arabidopsis proteins with putative major or minor PTSs. The Arabidopsis genome was screened for genes encodingunknown proteins with a putative major or minor PTS1 or SHL. or AHL. (A) or with a putative PTS2 (B), following a previousdefinition (Reumann, 2004). For detection of PTS1-targeted proteins, nine major ([SA][RK][LM].without AKM. plus SRI. andPRL.), five minor canonical tripeptides (SKI., PRM., PKL., C[RK]L.) of the Hayashi motif ([SAPC][RK][LMI].; Hayashi et al.,1997), and six minor noncanonical PTS1 tripeptides (SRV., [SA]NL., SML., SNM., SSM.) were applied. They weresupplemented by SHL. and AHL., which have been shown to direct a passenger protein to plant peroxisomes and/or to interactwith tobacco Pex5 in the yeast-two-hybrid system (Mullen et al., 1997a; Kragler et al., 1998). For detection of PTS2-targetedproteins, two major (R[LI]x5HL) and nine minor PTS2 nonapeptides (R[QTMAV]x5HL, RLx5H[IF], R[IA]x5HI) were applied. Intotal, 282 different proteins were detected (additional splice and translation variants excluded), of which 131 and 80 proteinscarried a major and minor PTS1, respectively, and 11 proteins [SA]HL.; 31 and 29 proteins carried a major and minor PTS2,respectively. Two proteins (At5g27600 and At1g06460) contained both a PTS1 as well as a PTS2, and one hypothetical protein(At5g40900) carried two putative PTS2 nonapeptides in the N-terminal domain. The dashed lines indicate the number ofArabidopsis proteins expected to carry a particular PTS if random distribution of amino acids is assumed (3.6 proteins for aC-terminal tripeptide and 5.1 proteins for a nonapeptide with four conserved and five variable residues being located inthe targeting domain at position 2–38). The number of Arabidopsis proteins is compared with the frequency at which specificPTS peptides occurred in plant homologs of PTS-targeted proteins (Reumann, 2004).

Database of Arabidopsis Peroxisomal Proteins

Plant Physiol. Vol. 136, 2004 2589

positions 1, 2, 8, and 9) and the fact that these tetrapep-tides can occur at 29 different positions if the firstresidue of the PTS2 is located between position 2 and30, each tetrapeptide is expected to be present onaverage in 5.1 Arabidopsis proteins if random distri-bution is assumed. Thus, also the two major but none ofthe minor PTS2 nonapeptides are present in a numberof Arabidopsis proteins two to three times aboveaverage.

As summarized for both PTSs, 282 Arabidopsisproteins were identified that contained either a major(162 proteins, 57%), a minor PTS (109 proteins, 39%),or [SA]HL. (11 proteins), of which most carrieda putative PTS1 (222, 79%). Of the proteins witha major PTS about 24 proteins (15%; Table I) can beclassified as known proteins from plant peroxisomeseither because their genes have been directly clonedfrom Arabidopsis and protein targeting to peroxi-somes has been demonstrated experimentally, orbecause they most likely represent the ortholog ofperoxisomal proteins from different plant species. Themajority of the extracted proteins, however, representunknown nonhypothetical proteins, plant homologs ofwhich have not yet been cloned and the function ofwhich remains to be studied (about 100 proteins witha major PTS; Table I). The unknown nonhypothetical

proteins with a major PTS comprise on the one handpartly annotated proteins that share some sequencesimilarity with mammalian or prokaryotic proteins (81proteins; Table I), and on the other hand ‘‘expressedproteins’’ without any detectable homology to func-tionally analyzed proteins in the database (22 pro-teins). For many unknown proteins, correspondingESTs are restricted to collections of plants subjected toabiotic stress conditions like dehydration, supportingthe idea that analysis of the entire proteome of plantperoxisomes is hardly possible by an experimentalproteome approach and needs to be supplemented bysuch a bioinformatics strategy. Of the extracted pro-teins with a major PTS about 35 ORFs (22%) still lackevidence for expression and encode hypothetical pro-teins (Table I). The only known peroxisomal matrixproteins not identified by this search are the threeisoforms of CAT with an unusual PTS1 and xanthinedehydrogenase/oxidase with an unknown PTS(Mullen et al., 1997b; Kamigaki et al., 2003).

Validation of Predicted Peroxisomal Localizationby Bioinformatics Tools

The PTS1 reveals a lower hierarchy as comparedto N-terminal signals (Neuberger et al., 2003a), and

Table I. Homology and targeting analysis of Arabidopsis proteins with a major PTS

Putative peroxisomal matrix proteins were extracted from the Arabidopsis genome based on the presence of one of nine major PTS1 tripeptides([SA][RK][LM]. without AKM. plus SRI. and PRL.) or one of two major PTS2 nonapeptides (R[LI]x5HL) according to a previous definition(Reumann, 2004). These proteins were classified based on homology and in silico expression. Known peroxisomal proteins (24 in total, 15%) compriseorthologs, the genes of which have been cloned from Arabidopsis or a different plant species and for which targeting of one ortholog to plantperoxisomes has been demonstrated experimentally. Unknown nonhypothetical proteins (103, 64%) mostly shared sequence similarity witheukaryotic or prokaryotic proteins, and their gene expression is supported by ESTs. In contrast to 22 of these unknown proteins annotated expressedproteins, the remaining 81 unknown nonhypothetical proteins (79%) were partly annotated and shared some sequence similarity with known proteinsin the public databases. Expression of hypothetical proteins (35 proteins, 22% of all proteins with major PTSs) was not yet supported by ESTs, and someof these genes may represent pseudogenes. Targeting prediction was considered significant if two independent programs concluded subcellulartargeting to the same subcellular compartment. High overall probability for targeting to peroxisomes was defined if targeting to peroxisomes waspredicted with high probability at least by one program and if no nonperoxisomal targeting signal was predicted with high accuracy by twoindependent programs. Medium and low overall targeting probability was deduced if one or two of these criteria were not fulfilled, respectively. Dueto a lack of prediction software for PTS2-targeted proteins, estimation of targeting probability was performed similarly for PTS2 proteins but restrictedto the categories of medium and low probability. Further support for peroxisome targeting is provided by analysis of the targeting domain for conservedproperties of PTS-targeted plant peroxisomal proteins and, for specific proteins, by analysis of PTS conservation in homologous EST (see AraPerox,Supplemental Figs. 1 and 4) but not included in the definition of overall targeting probability. n.d., Not defined.

Major PTS ProteinsNumber of Proteins Overall Probability of Targeting to Peroxisomes

Total High Medium Low

% % % %Total Total 162 100 53 39 8

Known peroxisome proteins 24 15 58 42 0Unknown nonhypothetical proteins 103 64 49 40 12Hypothetical proteins 35 22 63 34 3

PTS1 Total 131 100 66 30 5Known peroxisome proteins 16 12 88 12 0Unknown nonhypothetical proteins 90 69 56 38 7Hypothetical proteins 25 19 88 12 0

PTS2 Total 31 100 n.d. 77 23Known peroxisome proteins 8 26 n.d. 100 0Unknown nonhypothetical proteins 13 42 n.d. 54 46Hypothetical proteins 10 32 n.d. 90 10

Reumann et al.

2590 Plant Physiol. Vol. 136, 2004

PTS1-like C-terminal tripeptides are found in nonper-oxisomal proteins in which the tripeptide is not prop-erly exposed on the surface of the folded protein (seealso ‘‘Discussion’’). Thus, even though the Arabidopsisproteins with a major PTS are expected to include manytrue peroxisomal proteins, a moderate number of falsepositives need to be anticipated within this group aswell. The proteins with a minor PTS are localized inperoxisomes by definition with lower probability andare expected to include a considerable number ofnonperoxisomal proteins because peroxisomal target-ing of minor PTS is suspected to depend on targetingenhancing elements located next to the PTS (Reumann,2004). Discrimination between true positives and non-peroxisomal proteins by bioinformatics tools is a chal-lenging follow-up task of this genome screen. Somefalse positives are obvious and can be identified by theirannotation. For example, the cytoplasmic 20S protea-some subunits PAE1 and PAE2 (PAE1, At1g53850;ARL.; PAE2, At3g14290; SRL.), the putative plastiddivision protein FtsZ2-2 (At3g52750, PRL.), which isa paralog of plastidic FtsZ2-1, a closely related Arabi-dopsis homolog of the Glc-6-P translocator (At5g54800,AKL.), and several putative transcription factors witha minor PTS are probably nonperoxisomal proteins(data not shown).

To identify more subtle nonperoxisomal proteinsand provide further support for peroxisomal targetingof true proteins of peroxisomes, various subcellularprediction programs were applied. Predicted targetingto peroxisomes by a PTS1 can be supported by a fewsubcellular prediction programs, including TargetP,PSORT, and PSORTII. Only recently, new efforts havebeen undertaken to optimize the prediction of PTS1-targeted proteins (PeroxiP, Emanuelsson et al., 2003;PTS1 predictor, Neuberger et al., 2003a, 2003b). Plant-specific properties of the PTS, however, have still notbeen elaborated because of the predominance of CATisoforms with atypical PTSs (Neuberger et al., 2003a,2003b). Thus, negative prediction results for plantPTS1 proteins need to be handled with care. Algo-rithms for predicting PTS2-targeted peroxisomal pro-teins lack in all programs to date.

By contrast, plastidic transit peptides, mitochon-drial presequences, and ER signal peptides can some-times be predicted with high accuracy. Dual subcellulartargeting due to alternative splicing, an alternativeuse of two translational start codons or dependingon environmental conditions has been reported forsome peroxisomal proteins, such as plant isoforms ofHSP70 and Asp aminotransferase (Wimmer et al.,1997; Gebhardt et al., 1998). These proteins, however,are expected to represent an exception from thegeneral rule of an exclusive localization of peroxi-somal proteins in microbodies. The presence of non-peroxisomal targeting signals is therefore consideredto decrease the probability that the same protein islocalized in peroxisomes and necessitates a more de-tailed targeting analysis. Targeting prediction wasconsidered significant if two independent programs

predicted subcellular targeting to the same subcellularcompartment with high probability. High overallprobability for targeting to peroxisomes was definedif targeting to peroxisomes was predicted with highprobability at least by one program and if no non-peroxisomal targeting signal was predicted with highaccuracy by two independent programs. Medium andlow overall targeting probability was deduced if oneor two of these criteria were not fulfilled, respectively.Due to a lack of prediction software for PTS2-targetedproteins, estimation of targeting probability was per-formed similarly but restricted to the categories ofmedium and low probability. Regarding the 90 un-known nonhypothetical proteins with a major PTS1(Table I), most are targeted to plant peroxisomes withhigh (56%) or medium probability (38%). By contrast,a considerable percentage of the unknown nonhypo-thetical proteins with a major PTS2 are targeted toplant peroxisomes with low probability (46%; Table I).

Accessory elements that are located in close prox-imity of PTS peptides have been reported to play anauxiliary role in targeting proteins to peroxisomes(Mullen et al., 1997a, 1997b; Lametschwandtner et al.,1998). The targeting domain of both PTS1- and PTS2-targeted plant peroxisomal proteins was found to becharacterized by a high content of basic and Pro resi-dues and by a pronounced increase in pI (Reumann,2004). Especially for proteins with a minor PTS pep-tide and in case of an ambiguous prediction of sub-cellular localization, the presence of these conservedstructural elements can provide further support forperoxisomal targeting.

According to our experience, the most solid supportfor targeting of unknown proteins to peroxisomes,however, can be provided by the identification ofhomologous genes and ESTs that encode the samecompartment-specific ortholog and carry a PTS aswell, embedded in a moderately conserved targetingdomain. Prerequisites for a successful detection of PTSconservation in homologous genes/ESTs supportingperoxisomal targeting, however, are: (1) the absence ofparalogous nonperoxisomal Arabidopsis proteins ofhigh sequence similarity shared with the Arabidopsisprotein of interest; (2) expression of the gene of interestin a common plant organ, from which many plant ESTcollections have been generated; and (3) a certaindegree of sequence variation within the targetingdomain. Especially for proteins with a low-abundancePTS1, such as CKL., SML., SNL., or ANL., manyof which have previously only been detected in oneorthologous group (i.e. sulfite oxidase), or for proteinswith a putative PTS2, detection of a majority ofhomologous ESTs with major PTS peptides stronglysupports protein targeting to peroxisomes (Supple-mental Fig. 1, A–C, available at www.plantphysiol.org). On the other hand, if several presumablyorthologous ESTs from different plant species do notcarry a PTS1-like C terminal tripeptide, targeting ofthese proteins to peroxisomes is questionable (Supple-mental Fig. 1D). These Arabidopsis proteins with

Database of Arabidopsis Peroxisomal Proteins

Plant Physiol. Vol. 136, 2004 2591

a putative PTS, as well as the results of the analysis ofoverall targeting prediction, homology, and in silicoexpression by Digital Northern (Mekhedov et al.,2000), have been compiled in the database AraPerox(www.araperox.uni-goettingen.de; Fig. 2).

Proteins with overall low probability of targeting toperoxisomes were excluded from further analysespresented in this study. Focusing on proteins that aretargeted to plant peroxisomes with medium or highprobability, many currently unknown proteins andeven protein families with numerous putative perox-isomal isoforms were identified. These proteins in-clude large protein families of acyl-CoA activatingenzymes and thioesterase- and GDSL-lipase relatedproteins, many enzymes probably involved in b- anda-oxidation of unsaturated and branched-chain fattyacids, NADP-dependent dehydrogenases and reduc-tases, and enzymes involved in N metabolism as wellas regulatory proteins. The focus of the subsequentpresentation will be on enzymes, the peroxisomallocalization of which is strongly supported by pre-vious biochemical data, PTS conservation in homolo-gous ESTs, and/or by their homology to peroxisomalproteins from yeast and mammals.

A Large Family of Peroxisomal Isoforms ofAcyl-Activating Enzymes

Arabidopsis contains a large superfamily of 63mostly unknown acyl-activating enzymes (AAEs, orAMP-binding proteins; Fulda et al., 1997) that activatetheir acid substrates using ATP via an enzyme-boundadenylate intermediate, and include mainly acyl-CoAsynthetases and 4-coumarate-CoA ligases. As noticedearlier, a surprisingly large number of 17 enzymes ofthis family carry a putative PTS (Fulda et al., 2002;Shockey et al., 2002, 2003; Staswick et al., 2002;Supplemental Fig. 2). For long-chain acyl-CoA syn-

thetases 6 and 7, targeting to peroxisomes by a PTS1/2and an essential role in lipid mobilization has beendemonstrated by Suc-dependent germination of a dou-ble T-DNA knockout (k.o.) mutant (Fulda et al., 2002,2004). Clade V and the plant-specific clade VI inparticular contain a large number of mostly unknownenzymes that probably activate specific fatty acids inperoxisomes for subsequent metabolization (clade V,seven PTS1 proteins out of eight; clade VI, six PTS1proteins out of 14; Shockey et al., 2003; SupplementalFig. 2). Gene arrangement next to each other suggestssubstrate specification by recent gene duplication. Theenzymes of clade V are closely related to 4-coumarate-CoA ligases of the neighboring clade IV, which pro-duce CoA thioesters of a variety of hydroxy- andmethoxy-substituted cinnamic acids, which are pre-cursors of several phenylpropanoid-derived com-pounds, including anthocyanins, flavonoids, lignin,and coumarins (see references in Shockey et al., 2002,2003). The putative peroxisomal isoforms have beenproposed to catalyze CoA activation of benzoic acidderivatives or of very long-chain fatty acids (Shockeyet al., 2003). Combined homology data of this studysupport involvement of some AAEs in peroxisomalmetabolism of aromatic compounds, possibly includ-ing some plant hormones (see Fig. 7).

Enzymes Involved in b-Oxidation of Unsaturated

Fatty Acids

In addition to the basic b-oxidation machinery (Fig.3), degradation of mono- and polyunsaturated fattyacids, which are predominantly in cis configuration,requires auxiliary enzymes for elimination or isomer-ization of double bonds. Degradation of unsaturatedfatty acids with double bonds extending from even-numbered carbon atoms (e.g. linoleic acid) yieldsan intermediate, which is degraded either by an

Figure 2. Pictogram of the database AraPerox.About 280 putative matrix proteins from Arabidop-sis with a putative major or minor PTS have beencompiled in thedatabaseAraPerox (www.araperox.uni-goettingen.de). Summarized data on homol-ogy, targeting prediction, and expression analysisby Digital Northern are currently provided for allproteins with a major PTS. Known and putativeArabidopsis homologs of yeast andmammalianPEXproteins involved in peroxisome biogenesis havebeen added as well as membrane-associated en-zymes and integral membrane proteins involved inmetabolite transport.

Reumann et al.

2592 Plant Physiol. Vol. 136, 2004

NADP-dependent or an alternative pathway(Gerhardt, 1993; Graham and Eastmond, 2002; Hooks,2002; Fig. 3). In the NADP-dependent pathway,2,4-dienoyl-CoA (2-trans,4-cis-dienoyl-CoA) is re-duced to 3-trans-enoyl-CoA by an NADP-dependent2,4-dienoyl-CoA reductase (DECR) and further isomer-ized by the D3, D2-enoyl-CoA isomerase activity ofmultifunctional protein (MFP) to 2-trans-enoyl-CoA(Preisig-Muller et al., 1994), the common substrate ofMFP. Even though 2,4-DECR has been detected bio-chemically in plant peroxisomes, a predominant role ofthe NADP-independent pathway was concluded (seereferences in Gerhardt, 1993). The NADP-dependentpathway has been supported in yeast and mammals bycloning of NADP-dependent 2,4-DECR and a mono-functional peroxisomal D3, D2-enoyl-CoA isomerase(PECI) as well as by phenotypical analysis of a yeastk.o. mutant (van Roermund et al., 1998; Geisbrecht et al.,

1999; De Nys et al., 2001). An Arabidopsis protein of thefamily of weakly conserved short-chain dehydrog-enases/reductases (At3g12800, 298 residues) carriesthe major PTS1 SKL. and is the closest Arabidopsishomolog of mammalian and yeast 2,4-DECR (about40% identity over 250 residues, E value5 10239; De Nyset al., 2001; Gurvitz et al., 2001; Supplemental Fig. 3).Because targeting of this Arabidopsis protein to perox-isomes is supported by PTS1 conservation (Supple-mental Fig. 4A), this protein most likely represents theplant peroxisomal ortholog of NADP-dependent 2,4-DECR and supports the existence of the NADP-dependent pathway for degradation of unsaturatedfatty acids in plants (Fig. 3). According to our analysisand in contrast to a previous suggestion (Kamada et al.,2003), two other Arabidopsis proteins of the family ofshort-chain dehydrogenases/reductases (At4g05530,254 residues, SRL.; At3g59710, 302 residues, SKL.;

Figure 3. Alternative pathways of peroxisomal b-oxidation of unsaturated fatty acids in plants. Two Arabidopsis proteins(At3g12800, SKL.; At5g43280, AKL.) that are homologous to mammalian and yeast DECR and DCI, respectively, and carrya conserved PTS (see Supplemental Fig. 4, A and B), support the existence of an alternative NADP-dependent pathway fordegradation of unsaturated fatty acids with double bonds extending from even-numbered (left, A) and odd-numbered C-atoms(right, B) in plant peroxisomes. Except for NADPH of 2,4-DECR, other cofactors and small metabolites like CoASH, water, O2,H2O2, FAD, and NADH are omitted due to space limitations. The identified Arabidopsis homologs of 2,4-DECR and D3,5,D2,4-DCI are shaded gray. For possible monofunctional peroxisomal isoforms of enoyl-CoA hydratase, D3,D2-enoyl-CoA isomerase,and acyl-CoA dehydrogenase, see Figures 4, 7, and Supplemental Figure 4. ACX, Acyl-CoA oxidase.

Database of Arabidopsis Peroxisomal Proteins

Plant Physiol. Vol. 136, 2004 2593

E values . 10213), are more distantly related to yeastand mammalian 2,4-DECR even though they are tar-geted to plant peroxisomes with high probability aswell, and are expected to differ in substrate specificity(Supplemental Fig. 3).

Degradation of unsaturated fatty acids with doublebonds extending from odd-numbered carbon atoms(e.g. oleic acid) is thought to proceed to the intermedi-ate 3-cis-enoyl-CoA and relies on the above-mentionedD3, D2-enoyl-CoA isomerase activity of MFP for isom-erization to 2-trans-enoyl-CoA (Gerhardt, 1993;Preisig-Muller et al., 1994). In mammals and yeast,an alternative NADP-dependent pathway is used forthese fatty acids. In this pathway, the intermediate3-trans,5-cis-dienoyl-CoA is isomerized by D3,5-D2,4-dienoyl-CoA isomerase (DCI) to 2,4-dienoyl-CoA(2-trans,4-trans-dienoyl-CoA), where both NADP-dependent degradation pathways of unsaturated fattyacids merge (see above, Henke et al., 1998). Rathersurprisingly, an Arabidopsis protein that belongsto the family of enoyl-CoA hydratases/isomerases(At5g43280, 278 residues) shares about 40% iden-tity over the entire length with human D3,5, D2,4-DCI(Filppula et al., 1998) and is the only enzyme of about 10Arabidopsis enoyl-CoA hydratases with a putative PTSthat clusters together in clade II with mammalian D3,5,D2,4-DCI (Fig. 4). The Arabidopsis protein carries themajor PTS1 AKL., which is conserved or changed toSKL. in all detectable homologous ESTs (Supplemen-tal Fig. 4B) and thus most likely represents the plantperoxisomal ortholog of mammalian D3,5, D2,4-DCI.Successful complementation of the yeast mutant withthe Arabidopsis gene is required to conclusively provethe existence of this novel degradation pathway ofunsaturated fatty acids in plant peroxisomes. Apartfrom the isoforms of MFP and CHY1 (clades I and IV,respectively; Richmond and Bleecker, 1999; Zolmanet al., 2001), four other unknown Arabidopsis proteinswith a putative PTS belong to different clades of thesuperfamily of enoyl-CoA hydratases/isomerases (e.g.clade V, At1g60550, RLx5HL; clade VI, At4g16210,SKL.; Fig. 4), two of which interestingly cluster withPECI from fungi and mammals (clade III, At1g65520,SKL.; At4g14430, PKL.; see also Hooks, 2002; Fig. 4).

Provision of Reducing Equivalents for the PeroxisomalNADPH Pool

The enzyme 2,4-DECR requires stoichiometricamounts of intraperoxisomal NADPH for b-oxidationof unsaturated fatty acids with double bonds at evenand possibly also at uneven position (Fig. 3). Becausethe membrane permeability of plant and yeast perox-isomes seems to be restricted to diffusion of smallcarboxylates and to prevent passage of NAD(P) H (vanRoermund et al., 1995; Reumann et al., 1997, 1998),reduced equivalents need to pass the membraneby a malate/Asp-oxaloacetate shuttle (Mettler andBeevers, 1980; Reumann et al., 1994). In yeast andmammals, peroxisomal isoforms of NADP-dependent

isocitrate dehydrogenase (NADP-IDH) provideNADPH for degradation of unsaturated fatty acids(Henke et al., 1998; van Roermund et al., 1998). Plantperoxisomal isoforms of NADP-IDH have been char-acterized biochemically (Corpas et al., 1999; del Rioet al., 2002) but not yet at the molecular level. TheArabidopsis family of NADP-IDH comprises threeclosely related genes (Supplemental Fig. 5), one iso-form of which carries the C-terminal tripeptide SRL.(At1g54340, 416 residues). This enzyme shares about85% and 75% identity over the entire length with theputative cytosolic (At1g65930, 410 residues) and plas-tidic isoforms (At5g14590, 485 residues), respectively,the latter of which is extended at the N terminus bya transit peptide about 70 residues (Supplemental Fig.5). Expressed sequence tags that share highest se-quence similarity with the PTS1-containing NADP-IDH and also carry a PTS1, are found in many plantspecies and support targeting of the isoform to plantperoxisomes (Supplemental Fig. 4C).

The predicted localization of NADP-IDH and 2,4-DECR in plant peroxisomes indicates that NADPH isan important cofactor of peroxisomal metabolism nextto NADH (Fig. 5). Peroxisomal 2-oxophytodienoicacid (OPDA) reductase isoform 3 (OPR3) is a secondNADPH-dependent reductase of plant peroxisomes(Fig. 5). Besides NADP-IDH, the oxidative pentose-phosphate pathway (OPPP) has been discussed as analternative mechanism for intraperoxisomal genera-tion of NADPH, because all three enzymes of theoxidative part have been characterized biochemicallyin leaf peroxisomes from Pisum sativum (Corpas et al.,1998; Fig. 5). Two proteins paralogous to enzymes ofthe OPPP with a putative PTS1 can be detected in theArabidopsis genome. The second enzyme of the OPPP,6-phosphogluconolactonase (6PGL), has not beencloned from plants yet but the Arabidopsis genomeencodes five putative 6PGL, which share about 40%identity (over 250 residues) with mammalian 6PGLand yeast homologs (Collard et al., 1999). One enzyme(At5g24400, 325 residues), which surpasses the lengthof the others by a predicted transit peptide of about 70residues, carries additionally the putative PTS1 SKL.,suggesting dual targeting to plastids and peroxisomes(Supplemental Fig. 5). Indeed, several homologousESTs carry a putative PTS1 as well (Supplemental Fig.4D). Experimental support is required to verify thehypothesis of the existence of peroxisomal isoformsand their translation from the second Met, which isconserved in all homologs with a predicted transitpeptide (data not shown).

Of three Arabidopsis homologs of 6-phospho-gluconate dehydrogenase (6PGDH), one isoform(At3g02360, 486 residues) carries the minor PTS1SKI. and represents a peroxisomal candidate foroxidation of 6-P-gluconate to ribulose-5-P, NADPH,and CO2 (Fig. 5; Supplemental Fig. 5). The enzymeshares about 85% sequence identity with other full-length PTS1-containing homologs, many of which areannotated as cytosolic proteins (e.g. Medicago, SRI.;

Reumann et al.

2594 Plant Physiol. Vol. 136, 2004

Figure 4. Phylogenetic analysis of Arabidopsis enoyl-CoA hydratases/isomerases with putative PTSs. Five unknown Arabidopsisproteins with a putative PTS1 (At5g43280, AKL.; At4g14430, PKL.; At1g65520, SKL.; At1g60550, RLx5HL; and At4g16210,SKL.) that belong to the superfamily of enoyl-CoA hydratases/isomerases (ECH), were aligned with MFP and CHY1 homologsfrom Arabidopsis and Oryza sativa as well as prokaryotic and mammalian homologs of the family of enoyl-CoA hydratases/isomerases, and a phylogenetic tree was calculated using ClustalX (1,000 bootstraps). An unrooted phylogenetic tree wasgraphically constructed by Treeview. For eukaryotic proteins, recognizable PTS are indicated, and the five unknown Arabidopsisenoyl-CoA hydratases/isomerases are labeled by a gray box. Clade I of enoyl-CoA hydratases/isomerases comprises MFPhomologs from plants (At_MFP2, At3g06860, SRL.; At_AIM1, At4g29010, SKL.; and Cs_MFP, Cucumis sativus, Q39659,PRM.), bifunctional proteins (BFP) frommammals (Hs_BFP,Homo sapiens, NP_001957, SKL.; and Rn_BFP, Rattus norvegicus,NP_598290, SKL.), and a prokaryotic homolog (Po_FadB, Pseudomonas oleovorans, AAK83058). Clade II of enoyl-CoAhydratases/isomerases comprises D3,5,D2,4-DCI from mammals (Hs_DCI, AAH11792, SKL.; Rn_DCI, NP_072116, SKL.; andMm_DCI, Mus musculus, NP_058052, SKL.), two putative DCI (Ce_putDCI, Cenorhabditis elegans, NP_494954, SKL.; andNc_putECH, Neurospora crassa, XP_326933, SKL.), and two plant homologs (At5g43280, AKL.; and Os_putECH, Oryzasativa, NP_914858, SKL.). Clade III of enoyl-CoA hydratases/isomerases comprises PECI frommammals (Hs_PECI, NP_006108,SKL.; Mm_PECI, NP_035998, PKL.; and Rn_PECI, XP_214464, PKL.) and fungi (Sc_PECI, Saccharomyces cerevisiae,NP_013386) as well as three Arabidopsis homologs (At4g14430, PKL.; At1g65520, SKL.; and At4g14440). Clade IV of enoyl-CoA hydratases/isomerases comprises hydroxyisobutyryl-CoA hydrolases (HIBCH), such as the Arabidopsis protein At_CHY1(At5g65940, AKL.), two closely related homologs of CHY1 (At_putHIBCH_1, At2g30660, AKL.; and At_putHIBCH_2,At2g30650, AKL.), two enoyl-CoA hydratases from other plant species (Am_putECH, Avicennia maritima, AAF01467, AKL.;and Pa_putECH, Prunus armeniaca, AAB88874, AKL.), three Arabidopsis homologs without recognizable PTS (At1g06550,At4g31810, and At3g60510), and mammalian homologs from human (Hs_HIBCH_1, NP_055177, and RTx5HL) and mouse(Mm_ HIBCH, NP_666220; and RAx5HL). Clade V of enoyl-CoA hydratases/isomerases comprises dihydroxynaphthoic acidsynthases (DHNS) from prokaryotes (Hsom_DHNS, Haemophilus somnus, ZP_00132565; Pm_MenB, Pasteurella multocida,NP_246033; and Ec_DHNS, E. coli, AAA23682) and two plant homologs (At1g60550, RLx5HL; and Os_putECH, NP_917386,RLx5HL). Clade VI of enoyl-CoA hydratases/isomerases comprises two unknown enoyl-CoA hydratases from plants (At4g16210,SKL.; and Ca_putECH, Cicer arietinum, CAB61740, SKL.). Due to space limitations, bootstrap values are omitted in the centerof the tree.

Plant Physiol. Vol. 136, 2004 2595

Spinacia, SKI.; Oryza, AKM.; Supplemental Fig. 4E),and putative PTS1 tripeptides are detected in severalhomologous ESTs as well (Supplemental Fig. 4E). TheArabidopsis protein shares about 75% identity overthe entire length with the two other Arabidopsisisoforms (At5g41670, At1g64190), none of which car-ries rather surprisingly an N-terminal extension re-sembling a transit peptide (Supplemental Fig. 5). Bycontrast, a possible candidate for a peroxisomalhomolog of Glc-6-P dehydrogenase has not been de-tected yet in this study and may either be targetedto peroxisomes by an alternative mechanism or carryan unusual PTS. Alternatively, 6-P-gluconolacton mayenter the peroxisomal matrix directly from the cyto-plasmic pool.

Catabolism of Branched Amino Acids

Catabolism of branched amino acids poses a con-siderable challenge because 3-methyl and 2-oxo or2-hydroxyl groups represent a barrier for the coreenzymes of fatty acid b-oxidation. Therefore, thepostulated catabolic mechanism is a complex multi-step peroxisomal pathway that seems to be initiated inmitochondria (Gerbling and Gerhardt, 1989; Gerhardt,1993; Zolman et al., 2001; Graham and Eastmond,2002). Many biochemically characterized enzymes ofthese pathways remain to be cloned and localized ata subcellular level.

Regarding Leu catabolism, the 2-oxo acid has beenproposed to undergo first oxidative decarboxylationand to be further degraded via a free methyl-branchedfatty acid to 2-methyl-propanoyl-CoA, which can re-enter the core complex of b-oxidation (Gerhardt, 1993).In mammals, an alternative catabolic pathway for Leucatabolism has first been described in mitochondria,in which basically 3-hydroxy-3-methyl-glutaryl-CoA(HMG-CoA) is cleaved by a lyase to acetyl-CoA andacetoacetate. Only recently, a putative PTS1 has beenrecognized at the C terminus of mitochondrial HMG-CoA lyase, and targeting of the unprocessed form tomammalian peroxisomes has been demonstrated(Tuinstra et al., 2002). A single Arabidopsis gene ishomologous to mammalian HMG-CoA lyase andapparently encodes by use of two alternative trans-lation start Mets and possibly by alternative splicing ofthe 6th exon either a longer protein with a predictedmitochondrial presequence (At2g26800, 468 residues)or a shorter protein (433 residues) that is suspected tobe targeted to peroxisomes by the putative PTS1 SKI..Homologous C-terminal ESTs from a large number ofplant species, most of which also carry a putativePTS1, support targeting of the shorter isoform toperoxisomes and its role in an alternative pathway ofperoxisomal Leu catabolism (Supplemental Fig. 4F).Involvement of HMG-CoA lyase in peroxisomal ca-tabolism of aromatic compounds is also possible,whereas a role in isoprenoid metabolism like inmammals is not expected due to a plastidic localiza-tion of this pathway in plants.

Regarding Val catabolism, oxidative decarboxyl-ation of the branched-chain 2-oxo acid is thoughtto yield 2-methyl-propanoyl-CoA, which can bedegraded by the basic b-oxidation enzymes to3-hydroxy-2-methyl-propanoyl-CoA. Instead of sub-sequent oxidation by MFP, 3-hydroxy-2-methyl-propanoate is released from the CoA ester by CHY1,a 3-hydroxyisobutyryl-CoA hydrolase (At5g65940, 378residues, AKL.), which belongs to the superfamily ofenoyl-CoA hydratases/isomerases (Fig. 4). The en-zyme has been identified in a screen of an Arabidopsispopulation mutagenized with ethyl methane sulfonatefor plants that are resistant to the naturally occurringauxin indole-3-butyric acid but sensitive to indole-3-acetic acid (IAA; Zolman et al., 2001). Two closelyrelated homologs of CHY1 (At2g30650, 410 residues;At2g30660, 376 residues, both AKL.; Zolman et al.,2001; Fig. 4) are candidate enzymes for hydrolysis ofslightly different substrates, for instance 3-hydroxy-propionyl-CoA, an intermediate of propionate catabo-lism (Gerhardt 1993), but are still hypothetical proteins.In the subsequent step of Val catabolism, 3-hydroxy-2-methyl-propionate (3-hydroxyisobutyrate) is oxidizedto 3-oxo-2-methyl-propionate by a yet unknown alco-hol dehydrogenase. Instead of an Arabidopsis proteinsimilar to 3-hydroxybutyryl-CoA dehydrogenases(At3g15290, 294 residues, PRL.), which is expectedto oxidize 3-hydroxy acids esterified to CoA (see Fig.7), the two other short-chain reductases that are more

Figure 5. Peroxisomal reactions involving the cofactor NADPH. Ac-cording to this study, NADP-IDH and two enzymes of the OPPP,namely 6-PGL and 6PGDH, are targeted to plant peroxisomes withhigh probability (see Supplemental Fig. 4, C–E) and can provideNADPH for reductive matrix enzymes. Two NADPH-dependentreductases of plant peroxisomes carry a PTS1, i.e. OPR3 of JAbiosynthesis, which converts OPDA to 3-oxo-2-(cis-2#-pentenyl)-cyclo-pentane-1-octanoic acid (OPC:8:0) and DECR of the NADP-dependentdegradation pathway of unsaturated fatty acids (see Fig. 3). Oxidation ofNADPH in the course of the ascorbate-glutathione cycle has also beenreported (Jimenez et al., 1997).

Reumann et al.

2596 Plant Physiol. Vol. 136, 2004

distantly related to 2,4-DECR and carry a putative PTSas well (Supplemental Fig. 3), may catalyze this re-action. A candidate for peroxisomal acyl-CoA dehy-drogenase that mediates a subsequent step of Valcatabolism (Zolman et al., 2001) and which has notbeen cloned from any organism yet, is the Arabidopsisprotein At3g06810 (SKL., 824 residues; data notshown). The enzyme is targeted to peroxisomes withhigh probability, contains the characteristic signatureof acyl-CoA dehydrogenases and is related to severalunknown homologs from mammals of the same size,all of which carry well-known putative PTS1 as well(e.g. ARM., AKL., SRL.).

A Novel Family of Small Peroxisomal Thioesterases

Acyl-CoA thioesterases catalyze the hydrolysis offatty acyl-CoAs to free fatty acids and CoASH. Acyl-CoA thioesterases have been implicated in mainte-nance of adequate levels of free CoA within theperoxisomal matrix for continuous fatty acid b-oxida-tion by releasing CoA that is linked to non- or poorly-metabolizable acyl-CoA intermediates. Since thesubstrate specificity of enzymes involved in metabo-lism of complex fatty acids, such as ricinoleic acid andpropionate, switches from CoA esters to free fattyacids (Gerhardt, 1993), several acyl-CoA thioesterasesor hydrolases, such as the above-mentioned hydrox-yisobutyryl-CoA hydrolase (CHY1; Zolman et al.,2001), may be required. Moreover, as b-oxidation offatty acids includes not only catabolism but also bio-synthesis of fatty acid derivatives, such as the impor-tant plant hormones JA, auxin, and possibly salicylicacid (SA; Stintzi and Browse, 2000; Zolman et al., 2001;Feussner and Wasternack, 2002), the precursor acyl-CoAs need to leave the b-oxidation complex at a cer-tain point and the hormones must be released either bycytoplasmic or, more likely, by peroxisomal acyl-CoAthioesterases (see Fig. 7).

Five distinct families of acyl-CoA thioesterases havebeen defined for prokaryotes and eukaryotes, wherethey are found in different cell compartments. Twomammalian acyl-CoA thioesterases have been local-ized to peroxisomes and revealed a broad substratespecificity (Jones and Gould, 2000; Hunt et al., 2002). Acorresponding plant homolog (ACH2, At1g01710, 427residues, SKL.; homologous to At4g00520, 283 resi-dues, AKL.), has been cloned only recently (Tiltonet al., 2004). According to biochemical activity andexpression data, the enzyme is unlikely to be linked tofatty acid oxidation as has been suggested for itseukaryotic homologs (Tilton et al., 2004). Our studyrevealed the existence of a large yet unknown Arabi-dopsis family of six small thioesterase-related proteinsof only about 150 residues in size (At1g04290,At1g48320, At2g29590, At3g16175, At3g61200, andAt5g48950). All six proteins carry a putative PTS1,such as SKL., AKL., and SNL. (Fig. 6), and theirpostulated function in plant peroxisomes is puzzlingat first glance. The plant enzymes cluster into two

different clades and are homologous to unknownsmall thioesterases from yeast, nematodes, insects,and mammals that strikingly do not carry a PTS1or PTS2 (Fig. 6). The three small thioesterases fromArabidopsis of clade I are homologous to prokaryoticsmall thioesterases, such as PaaI of Escherichia coli andAzoarcus evansii and FcbC of Arthrobacter sp., whichbelong to a novel catabolic pathway for the aromaticcompounds phenylacetic acid and 4-Cl-benzoate, re-spectively (Benning et al., 1998; Ferrandez et al., 1998;Thoden et al., 2003; Fig. 6). This novel prokaryoticpathway has been termed aerobic hybrid pathway,because both free and CoA esterified fatty acids arepathway intermediates and substrates of the partici-pating enzymes. Predicted targeting to peroxisomes ofthe homologs from Arabidopsis is supported by PTS1conservation in homologs from Oryza (Fig. 6). Thelarge number of small thioesterases in Arabidopsisand their presumable role in peroxisomal metabolismcontrast the situation in fungi, nematodes, insects, andmammals (Fig. 6) and argue in favor of a pronouncedplant-specific function of these enzymes, for instancein plant hormone biosynthesis (Fig. 7). Apart from thisprotein family, two other Arabidopsis proteins witha putative PTS1 belong to the family of esterases,lipases, and thioesterase and may also be involved inperoxisomal fatty acid metabolism (At3g62860, 348residues, SSM.; At5g11910, 297 residues, SRI.).

Peroxisomal Enzymes Involved in Biosyntheticb-Oxidation of Plant Hormones

The peroxisomal enzyme OPR3 reduces OPDA to theintermediate 3-oxo-2-(cis-2#-pentenyl)-cyclopentane-1-octanoic acid, the octyl chain of which is shortenedin three cycles of peroxisomal b-oxidation to an acetylgroup to yield JA (Figs. 5 and 7). Similarly, indole-3-butyric acid is converted by one cycle of peroxisomalb-oxidation to IAA (Zolman et al., 2001). SA, ahormone involved in thermotolerance and pathogenresistance, is synthesized from trans-cinnamic acidby way of a side branch of the phenylpropanoidpathway. After one cycle of b-oxidation, which isassumed to take place in peroxisomes, benzoicacid is further hydroxylated in meta position to yieldSA (Durner et al., 1997; Crozier et al., 2000; Fig. 7).The requirement for specific enzymes in prokaryotesstrongly suggests that specific peroxisomalhomologs of the enzymes of the basic b-oxidationmachinery are necessary for biosynthesis of thesearomatic or cyclic plant hormones. First, dependingon whether these hormone precursors enter the per-oxisomal matrix as free fatty acids or as CoA estersand depending on the substrate specificity of previousperoxisomal enzymes of the biosynthetic pathway(e.g. peroxisomal OPR3 acting on nonesterified sub-strates), the participation of peroxisomal AAEs needsto be postulated as an entrance point for peroxisomalb-oxidation. Second, considering the bulky size of thearomatic or cyclopentanone ring of IAA, SA, and JA

Database of Arabidopsis Peroxisomal Proteins

Plant Physiol. Vol. 136, 2004 2597

and the fact that the substrate specificity of otherenzymes, for instance that of the four peroxisomalisoforms of acyl-CoA oxidase, depends solely on thelength of straight-chain acyl-CoA esters (Hayashi et al.,1999; Hooks et al., 1999; Eastmond et al., 2000; Fromanet al., 2000), it is well possible that specific isoforms ofenoyl-CoA hydratases including monofunctional D3,D2-enoyl-CoA isomerases (Fig. 4) and 3-hydroxyacyldehydrogenases, catalyze the chain-shortening reac-tions of hormone precursors instead of MFP. Third andas outlined above, after thiolytic cleavage of the lastC2 unit, the plant hormones are probably releasedthe from CoA esters by peroxisomal acyl-CoA thio-esterases.

The gene encoding the small thioesterase FcbC ofArthrobacter sp. involved in degradation of 4-Cl-benzo-ate is located in an operon that comprises additionallythe acyl-activating enzyme FcbA, which is a 4-Cl-benzoate CoA ligase, and an enoyl-CoA hydratase(4-chlorobenzoyl-CoA dehalogenase, FcbB; Schmitz

et al., 1992; Fig. 7). The three homologous genes ofPseudomonas sp. strain CBS-3 are arranged likewise(Benning et al., 1998). The closest Arabidopsis homo-logs of FcbA are the 4-coumarate-CoA ligase-likeproteins with a putative PTS1 of clade V of AAEs(Supplemental Fig. 2; about 25% identity over theentire length) and other PTS1-containing AAEs. Theenzyme 4-chlorobenzoyl-CoA dehalogenase (FcbB)from Arthrobacter sp. is a monofunctional enoyl-CoAhydratase, several Arabidopsis homologs of whichcarry a putative PTS (Fig. 4).

The operon of E. coli involved in phenylacetatedegradation comprises 14 ORFs (Fig. 7), of whichPaaK, PaaF/G, and PaaI are homologous to FcbA,FcbB, and FcbC from Arthrobacter sp., respectively(Schmitz et al., 1992; Ferrandez et al., 1998). In addi-tion, PaaZ encoded in the same operon is homologousto betaine aldehyde dehydrogenases (BADH), one oftwo Arabidopsis isoforms is probably peroxisomal(At3g48170, SKL.; see below). The closest Arabidopsis

Figure 6. Phylogenetic analysis of an unknown Arabidopsis family of small thioesterases. Six putative small thioesterases(put_sT) from Arabidopsis of 155 to 188 amino acids in size, all of which carry a putative PTS1, were aligned with homologsidentified by BLAST analysis. The enzymes of clade I of plant small thioesterases share about 20% identity with smallthioesterases from prokaryotes involved in the degradation of aromatic compounds (4-Cl-benzoate, FcbCof Arthrobacter sp.; phenylacetate, PaaI of E. coli and Azoarcus evansii). For homologs carrying a putative PTS, the sequenceof the PTS peptide is indicated. Homologous unknown small thioesterases were also detected in yeast, nematodes, insects, andmammals but lack a putative PTS, indicating a nonperoxisomal localization in all organisms except for higher plants.Abbreviations: Ae_ PaaI, A. evansii, AAG28967; Ag_hom1, Anopheles gambiae, XP_308560; Asp_FcbC, Arthrobacter sp.,C48956, Thoden et al., 2003; At, Arabidopsis; At_put_sT1 to At_put_sT6, At1g48320, At5g48950, At3g61200, At1g04290,At2g29590, and At3g16175; Ce_hom1 to Ce_hom3, C. elegans, NP_495115, NP_872068, and NP_498872; Dm_hom1 andDm_hom2, Drosophila melanogaster, NP_647732 and NP_647730; Ec_ PaaI, E. coli, P76084, Ferrandez et al., 1998; Hs_sT,Homo sapiens, HTO12, PNAS-27, and NP_060943; Mm_sT,Mus musculus, NP_080066; Nc_hom1 and Nc_hom2,Neurosporacrassa, XP_325377 and XP_326708; Os_hom1 to Os_hom5, Oryza sativa, BAC79655, CAD40812, BAB90363, NP_913491,and NP_913490; Rn_sT, Rattus norvegicus, XP_214475; Sp_hom1, Schizosaccharomyces pombe, NP_596564.

Reumann et al.

2598 Plant Physiol. Vol. 136, 2004

Figure 7. Model of b-oxidation of aromatic and cyclic plant hormones in plant peroxisomes. The final steps of the biosynthesis ofthe aromatic or cyclic plant hormones JA, auxin, and SA are thought to take place in peroxisomes and involve one to three cycles ofb-oxidation for shortening of the side chain, but the enzymes involved in precursor activation, b-oxidation, and hormone releaseare currently unknown. Prokaryotes possess specific enzymes for degradation of aromatic compounds that participate in a novelaerobic hybrid pathway and are homologous to Arabidopsis proteins with a putative PTS, which accordingly are proposed to beinvolved in peroxisomal reactions of aromatic and cyclic compounds. The operon of Arthrobacter sp. involved in degradation of4-Cl-benzoate comprises three enzymes, namelyanAAE (FcbA; about 25% identical over the entire lengthwith 4-coumarate-CoAligase-like proteins of clade V and other PTS1-containing AAEs from Arabidopsis; see also Supplemental Fig. 2), an enoyl-CoAhydratase (4-chlorobenzoyl-CoA dehalogenase, FcbB; about 30% identical over 200 residues with Arabidopsis enoyl-CoA hy-dratases, e.g. At5g43280, AKL., At4g16210, SKL.; see Fig. 4), and a small thioesterase (FcbC; about 20% identicalwith the smallthioesterases from Arabidopsis of clade I; Fig. 6). The operon of E. coli involved in degradation of phenylacetate (Ferrandez et al.,1998) comprises 14 ORFs, three of which are homologous to FcbA (PaaK), FcbB (PaaF/G), and FcbC (PaaI) of Arthrobacter sp. Inaddition, an aldehyde dehydrogenase (PaaZ) is homologous to Arabidopsis BADH (At3g48170, SKL.; 20% identical over 400residues). A hydroxybutyryl-CoA dehydrogenase (PaaH) is homologous to a monofunctional hydroxybutyryl-CoA dehydrogenasefromArabidopsis (At3g15290, PRL.; about 40%identical over the entire length; seeSupplemental Fig. 4G). Theorder inwhich theenzymes of E. coli are involved in degradation of phenylacetate has not yet been analyzed in all detail and has been drawnaccording to available information.

Database of Arabidopsis Peroxisomal Proteins

Plant Physiol. Vol. 136, 2004 2599

homolog of PaaH is a monofunctional hydroxybu-tyryl-CoA dehydrogenase (At3g15290, 294 residues,PRL., about 40% identical over the entire length) withhigh targeting probability to peroxisomes due to PTS1conservation in homologous ESTs (Supplemental Fig.4G). Distinct isoforms of in total six acyl-CoA oxidasesand three thiolases in Arabidopsis may also be specificfor aromatic compounds. Taken together, these ho-mology data suggest that several Arabidopsis proteinswith a putative PTS are involved in peroxisomalbiosynthetic and possibly also catabolic reactions ofaromatic and cyclic plant hormones and pave the payfor straight-forward experimental strategies to testthese hypotheses.

Novel Paralogs of Glycolate Oxidase with a PostulatedRole in a-Oxidation

Glycolate oxidase (GOX) is the well-known firstperoxisomal enzyme of the glycolate pathway andcatalyzes the oxidation of glycolate to glyoxylatecoupled to the reduction of O2 to H2O2 via the cofactorflavinmononucleotide (Volokita and Somerville, 1987).The genome of Arabidopsis contains an unusuallylarge gene family of five GOX homologs (Reumann,2002). Two proteins (GOX1, At3g14420; GOX2,At3g14415) are closely related to each other and toSpinacia oleracea GOX, and their role in photorespira-tion is supported by a high number of ESTs derivedfrom photosynthetic tissue (Fig. 8A). The third GOXparalog (GOX3, At4g18360, AKL.) seems to have thesame substrate specificity but is specifically expressedin nongreen tissue (Kamada et al., 2003; Fig. 8A). Twomore distantly related Arabidopsis paralogs (HAOX1,At3g14130; HAOX2, At3g14150, about 60% identicalover the entire length) carry the minor PTS1 SML.and are, according to their digital expression profile,predicted to play a role in nonphotorespiratory me-tabolism (Fig. 8A). Prediction of different substratespecificity is based on the presence of two differingactive site residues as compared to GOX (Y-24/F andW-108/M; Lindqvist and Braenden, 1989; Fig. 8B).Interestingly, three peroxisomal GOX homologs havebeen identified recently in mammals (hydroxy acidoxidase 1-3 [HAOX1-3]; Kohler et al., 1999; Joneset al., 2000; Williams et al., 2000). In contrast to oneglycolate-oxidizing isoform (HsHAOX1), HsHAOX2and HsHAOX3 are specific for long-chain and me-dium-chain 2-hydroxy acids, respectively, and revealexchanges of active site residues as compared tophotorespiratory GOX that resemble those of thenovel Arabidopsis paralogs (Fig. 8B). It may bepredicted that the two novel GOX paralogs fromArabidopsis have a function closer to that ofHsHAOX2/3 and metabolize a substrate of longerchain length than glycolate. Because HsHAOX2/3are discussed to be involved in a-oxidation of fattyacids, a process that is only marginally understoodin both mammals and plants, the substrate specificityis expected to yield important insights into the

Figure 8. In silico study of an Arabidopsis family of GOX-relatedproteins by in silico expression analysis (A) and analysis of sequenceconservation of active site residues (B). ESTs corresponding to each ofthe five GOX homologs from Arabidopsis were identified by BLASTanalysis using TBLASTN and analyzed semiquantitatively according totissue-specificity, application of stress conditions, and the total numberof detected ESTs. The high number of ESTs for AtGOX1 (129) andAtGOX2 (37) contrasts the expression level of the other three homologs(AtGOX3, 2 ESTs each from collections of roots and stressed plants, 6ESTs in total; AtHAOX1, 1 ESTeach from collections of leaves, flowers,developing seeds, and stressed plants, 4 ESTs in total; and AtHAOX2, 1EST each from collections of leaves and flowers, 3 ESTs in total). Foranalysis of conservation of active site residues, homologs of plant GOXwere retrieved from the nonredundant database and aligned usingClustalX. The given numbers of the amino acids of the active sites referto GOX from S. oleracea (Lindqvist and Branden, 1989). The residuesinvolved in FMN binding were conserved in all homologs (data notshown). Abbreviations: At_GOX1 to GOX3, At3g14420, At3g14415,and At4g18360, respectively; At_HAOX1 and HAOX2, At3g14130 andAt3g14150, respectively; Csp, Cucurbita spec., T10242; So, S. oler-acea, P05414; Hs_HAOX1 to HAOX3, NP_060015, NP_057611, andAAF40201, respectively.

Reumann et al.

2600 Plant Physiol. Vol. 136, 2004

function of plant peroxisomes in a-oxidation of3-methyl- and 2-hydroxy acids.

Other Enzymes Related to Fatty Acid Catabolism

Lipolytic enzymes constitute a large family of dif-ferent classes, including not only true lipases, whichdisplay maximal activity toward water-insoluble long-chain triacylglycerides, but also carboxylesterases,which hydrolyze small ester-containing molecules atleast partly soluble in water, and various types ofphospholipases (Upton and Buckley, 1995; Arpignyand Jaeger, 1999). Class II carboxylesterases fromprokaryotes contain the so-called GDSL lipases thatcarry the active site residue Ser of the catalytic triad (S,D, and H) embedded in a slightly different motif(GDS(L) instead of GxSxG) and localized very close tothe N terminus (Arpigny and Jaeger, 1999). Prokary-otic GDSL lipases are membrane-bound or secretedesterases of in total about 600 residues in size com-prising a C-terminal autotransporting b-barrel domain(Riedel et al., 2003). Only limited information is avail-able on eukaryotic GDSL lipases to date.

Within an enormous family of GDSL-like enzymesin Arabidopsis, eight proteins carry a putative PTS1and share about 20% identity with the catalytic do-main of prokaryotic GDSL lipases, such as the crys-tallized esterase from Streptomyces scabies (Wei et al.,1995; Fig. 9). Most Arabidopsis GSDL-like lipases lackthe C-terminal autotransporter domain, suggestingthat they are soluble enzymes. At least six of these Ara-bidopsis proteins are expressed (At1g33811, ANL.;At2g38180, ARL.; At3g04290, SKI.; At4g18970,ARL.; At4g28780, SRI.; At5g08460, SRL.). TheGDSL motif of the Arabidopsis proteins contains anadditional Val residue at pos. 5 (GDSLV) like themammalian enzyme in contrast to prokaryotic en-zymes (GDSLS; Arpigny and Jaeger, 1999; Wilhelmet al., 1999; Fig. 9). The two other active site residues,Asp and His, are conserved as well (FxDx2HPT, Fig. 9;Wilhelm et al., 1999). If these isoforms of GDSL lipasesare indeed targeted to plant peroxisomes, they aresuspected to be involved in releasing fatty acids fromcomplex conjugates, for instance Glc and amino acidconjugates of plant hormones.

Peroxisomal Enzymes Involved in N Metabolism

With the peroxisomal enzymes uricase and xanthinedehydrogenase/oxidase, plant peroxisomes are in-volved in N metabolism and may compartmentalizefurther biosynthetic or catabolic pathways of N-con-taining metabolites. Aldehyde oxidase 2 (At3g43600,1321 residues, SNL.), which is distantly related toxanthine oxidase/dehydrogenase and contains thesame co-factor binding sites (Sekimoto et al., 1998;Seo et al., 2000), is one example and may be targeted toperoxisomes by the minor PTS1 SNL.. Polyaminesare crucial components for growth and cell prolifera-tion of all organisms and stimulate many reactions bytheir polycationic nature. Polyamines comprise theprimary amines putrescine and cadaverine as well asspermidine and spermine, of which the latter carryadditional secondary amino groups (Walden et al.,1997; Pandey et al., 2000; Perez-Amador et al., 2002).During catabolism of polyamines in plants, the pri-mary amines are subjected to oxidative deaminationby copper-containing diamine oxidases producingH2O2, ammonia, and aminoaldehydes. The secondaryamines present in spermine and spermidine can alsobe catabolized by a second class of enzymes, namelyflavin-containing polyamine oxidases, which formH2O2, 1,3-diaminopropan and aminoaldehydes. Thefirst polyamine oxidase has only recently been clonedfrom mammals, and postulated targeting to peroxi-somes by the putative PTS1 PRL. remains to bedemonstrated (Vujcic et al., 2003; Wu et al., 2003).Preferred substrates were N1-acetylated polyamines,indicating a physiological role of the enzyme in poly-amine back-conversion of spermine and spermidine tospermidine and putrescine, respectively (Vujcic et al.,2003; Janne et al., 2004). Plant peroxisomes may alsoplay a significant role in catabolism and/or back-conversion of polyamines as well, since three (poly)amine oxidases from Arabidopsis carry a putativePTS1 as noticed earlier (At1g65840, 497 residues,SRM.; At3g59050, 488 residues, SRM.; At2g43020,490 residues, SRL.; Kamada et al., 2003).

Even though the enzyme BADH, which catalyzesthe second and final biosynthetic step of the compat-ible osmosolute Gly betaine, is a chloroplastic enzymein Gly betaine-accumulating plant species, BADH

Figure 9. Sequence alignment of putative GDSL lipases from plant peroxisomes. Seven Arabidopsis GDSL-like lipases witha putative PTS1 were aligned. The numbers at the top correspond to the residues of the first protein (At1g33811). Due to spacelimitation, only the N-terminal GDSL domain, an internal domain with a putative active site residue (D174 for At1g33811), andthe C-terminal end of the Arabidopsis proteins with two putative active site residues (D336 and H339 for At1g33811) and thePTS1 are shown. The GDSL motif and conserved residues that correspond to active site residues in prokaryotic homologs areindicated at the bottom. The putative PTS1 tripeptides are underlined.

Database of Arabidopsis Peroxisomal Proteins

Plant Physiol. Vol. 136, 2004 2601

homologs of Oryza sativa and Hordeum vulgare areperoxisomal enzymes and also seem to confer salttolerance (Nakamura et al., 1997). Arabidopsis con-tains two closely related BADH isoforms, one of whichterminates with the PTS1 SKL. (At3g48170, 503residues). An analysis of ESTs suggests coexistenceof a peroxisomal and a nonperoxisomal (chloroplastic)isoform in dicots and monocots (S. Lemke and S.Reumann, unpublished data). Because the presumedhigh substrate specificity of chloroplastic BADH hasbeen questioned recently (Vojtechova et al., 1997;Sebela et al., 2000), the more general annotation as anaminoaldehyde dehydrogenase seems to be moreappropriate (Livingstone et al., 2003). The kineticparameter and subcellular localization of both BADHparalogs require reinvestigation to elucidate theirphysiological function, with special consideration ofthe production of aminoaldehyde derivatives in thecourse of polyamine catabolism.

Regulatory Proteins of Peroxisomes

Our knowledge on peroxisomal matrix proteins witha regulatory function is rather limited due to difficultiesin identifying low-abundance and inducible proteinsby biochemical approaches. Evidence for the existenceof regulatory proteins, such as heat shock proteins,kinases, and phosphatases, in peroxisomes is justemerging. Regarding heat shock proteins expressed inresponse to increased temperature and other forms ofabiotic stresses, a chloroplastic isoform of an HSP70homolog from Citrullus vulgaris is also targeted toperoxisomes by alternative use of two successive trans-lation start codons (Wimmer et al., 1997). Small heatshock proteins (sHSPs) are a ubiquitous superfamily ofHSPs characterized by the relatively small mass of thepolypeptide chain (16–42 kD) and the presence ofa conserved a-crystallin domain (Kim et al., 1998;Haslbeck, 2002). Plants house an exceptionally largefamily of sHSPs (Scharf et al., 2001). Experimentalevidence for the presence of sHSPs in peroxisomeshas not been provided yet for any organism. Twomembers of this family, however, carry a putativePTS1 (At5g37670, 137 residues, SKL.; At1g06460, 285residues, PKL. and RLx5HF) and may be targeted toplant peroxisomes. The second sHSP is intriguingbecause of the presence of a putative PTS1 (PKL.)as well as a putative PTS2 (RLx5HF), a feature thathas so far only been reported for LACS7 (Fulda et al.,2002).

Our largest gap of knowledge is probably that of theregulation of peroxisomal enzymes by posttransla-tional modifications and protein turnover. Some per-oxisomal enzymes, in particular CAT, have beensuggested to be modified posttranslationally by pro-teolytic cleavage (Kleff et al., 1997). Fragmentarysignaling cascades involving protein kinases andphosphatases have not been unraveled in any case.A protein kinase with yet unknown target was recentlyfound in glyoxysomes in a proteome study of Arabi-

dopsis (At3g17420, AKI.; Fukao et al., 2003). Acalcium-dependent protein kinase is anchored in theperoxisomal membrane by an acyl residue (Dammannet al., 2003). Calmodulin was found in a peroxisomalfraction and regulates CAT in the presence of Ca21

(Yang and Poovaiah, 2002), and a light-responsivenucleoside diphosphate kinase is reported to interactwith CAT (Fukamatsu et al., 2003). Our screen of theArabidopsis genome revealed seven putative proteinkinases that are suspected to mediate signal trans-duction across the peroxisomal membrane and to alterthe activity of peroxisomal enzymes by reversiblephosphorylation.

Turnover of peroxisomal enzymes may be regulatedby a peroxisomal homolog of the mitochondrial Lonprotease (At5g47040, 888 residues, SKL.), a homologof which has also recently been localized in mamma-lian peroxisomes (Kikuchi et al., 2004). A glutathionetransferase (At5g41210, 245 residues, SKI.), whichmay be a homolog of peroxisomal glutathione S-trans-ferase from mammals (Morel et al., 2004), or twoprotease-related proteins (At1g28320, 709 residues,SKL.; At2g41790, 970 residues, PKL.) may be furtherregulators of peroxisomal protein degradation.A Ca21-binding protein (At3g07490, 153 residues,SNL.) could be involved in signal transduction ofperoxisomal CDPK1 and CAT-activation by calmodu-lin, but the targeting function of the putative PTS1 cancurrently not be supported PTS1 conservation inhomologous ESTs (Supplemental Fig. 1D). One homo-log of a purple acid phosphatase also carries a putativePTS1 (At2g01880, 328 residues, AHL.). By contrast,considering the large number of more than 550 F-boxproteins in Arabidopsis (Kuroda et al., 2002) and thecurrent lack of evidence for expression of most of thegenes encoding about 8 F-box proteins with a putativePTS, most of these proteins are not expected to playa major role in plant peroxisomal metabolism.

DISCUSSION

In Silico Analysis of Protein Targeting to Peroxisomes

Well-defined conserved targeting signals that directnuclear-encoded proteins to a particular cell compart-ment offer the possibility to screen eukaryotic ge-nomes for unknown proteins that carry these signalsand are probably localized in the organelle of interest.The number of false positives relies primarily oncorrect gene prediction and on the specificity of theapplied targeting motifs. In the first years after pub-lication of the Arabidopsis genome (The ArabidopsisGenome Initiative, 2000), incorrect gene predictionaffected about 30% to 40% of our extracted proteins,for instance a ‘‘forever young protein’’ now predictedwith an internal SKL tripeptide (At4g27760; Kamadaet al., 2003). Gene prediction, however, has improvedconsiderably since and been supported by large-scalefull-length cDNA projects (Yamada et al., 2003) withthe result that only a minor portion of our extracted

Reumann et al.

2602 Plant Physiol. Vol. 136, 2004

proteins are estimated to be subjected to incorrect PTSprediction at present. However, a considerable num-ber of additional splice variants and alternative startMets are still being detected and require a regularupdate of genome screens.

Even though PTSs with their conservation of three tofour residues suggest a high specificity of the deducedtargeting motifs, a clear determination of allowedresidue combinations and the definition of plant-specific PTS motifs remain a challenge. Prediction ofperoxisomal targeting is made difficult because eventhe presence of a well-known PTS (e.g. SKL., SRL.,or RLx5HL) is not a sufficient criterion for unambig-uous targeting of a ‘‘natural’’ protein to the peroxi-some matrix and contrasts experimental resultsobtained with artificial fusion proteins. Some proteinswith a putative PTS are not targeted to peroxisomesbecause the tiny PTS1 tripeptide in particular is notproperly exposed on the surface of the folded proteinor because peroxisomal targeting of low-abundance‘‘weaker’’ PTS depends on the presence of marginallydefined accessory elements that substantially enhancethe affinity to its receptor. Furthermore, the C-terminalPTS1 is probably of lower hierarchy as compared toN-terminal signals (Neuberger et al., 2003a) because ofthe temporary advantage of N-terminal signals thatcan already be recognized by cytosolic chaperones andreceptors when the 3#-end of the mRNA encoding theC-terminal PTS1 is still being translated.

The Choice of the Screening Peptides

Our screen of the Arabidopsis genome for peroxi-somal proteins is characterized by a restrictive appli-cation of targeting peptides, as we included onlypeptides that had previously been detected in a signif-icant number of plant homologs of PTS-targetedmatrix enzymes (Reumann, 2004). In contrast to a com-parable genome analysis (Kamada et al., 2003), 10peptides of the Hayashi motif (i.e. A[RK]I., [PC]RI.,[PC]K[MI].; Hayashi et al., 1997) and 10 additionaltripeptides with H at position 22 ([SA]H[MI].,[PC]H[LMI].) were excluded in our study due toa current lack of experimental and in silico support fortheir peroxisomal targeting properties and in light ofthe subsequent difficulties in sorting out false posi-tives by further bioinformatics analyses. Our in silicoanalysis strongly suggested that position-specific res-idues of functional PTS peptides cannot be combinedfreely and that two to three low-abundance residuesare unlikely to constitute functional PTS1 tripeptides(Reumann, 2004). For the same reasons, the large num-ber of additional PTS peptides included in more per-missive experimentally determined PTS motifs werealso omitted (e.g. PTS1, [SAPCGT][RKHNL][LMIY].;Mullen et al., 1997a; Kragler et al., 1998; PTS2,[RK]x6[HQ][ALF]; Flynn et al., 1998). A low numberof true positives, however, may indeed be hiddenamong these Arabidopsis proteins with putativePTS peptides other than those defined as major and

minor PTS. For instance, a homolog of monodehy-droascorbate reductase carries the putative PTS1 AKI.(At3g52880; Kamada et al., 2003) and has beencharacterized biochemically in plant peroxisomes(Bunkelmann and Trelease, 1996). The analysis ofPTS conservation in novel proteins in the course ofthis study and the detection of homologous ESTswith PTS peptides that had so far been restricted tosingle orthologous groups (e.g. SML.), unique ESTs(e.g. AKI., CRI., SSL.), or which had completelybeen absent (RIx5HV, SSI., SNN., AML., or FKL.;see Supplemental Figs. 1 and 4; Reumann, 2004),indicate that additional peptides represent low-abundance but functional PTS1 tripeptides. These pep-tides will be defined as minor PTS peptides and addedto our genome screen upon experimental localizationof one of these novel orthologs to peroxisomes. Ourlimitation dealing with the localization of the PTS2nonapeptide within the N-terminal 38 residues is alsomore restrictive (compared to Kamada et al., 2003) butsupported by an analysis of more than 150 plantsequences (R of position 1 of the PTS2 at most atposition 26; data not shown).

On the other hand, about 50 Arabidopsis proteinswith one of six noncanonical C-terminal tripeptides(e.g. SRV., S[NM]L., ANL.) and some proteins witha minor PTS2 are specific to our genome screen and areexpected to include true positives (Supplemental Fig.1). Despite the empirically chosen threshold of thedefinition of major and minor PTSs we judged a dis-tinction between major and minor peptides necessarybecause of their differing rates of false positives, andfocused in the subsequent in silico analyses primarilyon those proteins with a major PTS. Proteins witha minor PTS can provide valuable hints on novelmetabolic pathways and regulatory functions, butneed to be analyzed thoroughly by application offurther bioinformatics tools in each case.

Bioinformatics Validation of Peroxisomal Targeting

Despite our restrictive application of PTS peptides,some proteins have been extracted from the Arabi-dopsis genome that are obviously nonperoxisomalproteins, and a considerable number of false positivesneed to be anticipated among proteins with a minorPTS. According to our experience, distinction of trueand false positives is essential before drawing pre-mature conclusions with respect to the identification ofnovel proteins from plant peroxisomes and beforeproceeding with experimental studies (Kamada et al.,2003). Several bioinformatic strategies have been ap-plied in our study to increase the prediction accuracyof protein targeting to peroxisomes. These analy-ses comprised algorithm-based prediction of PTS1-targeted proteins as well as nonperoxisomal targetingsignals, analysis of the targeting domain for propertiesconserved in plant peroxisomal proteins, and analysisof PTS conservation in homologous ESTs. Even thoughsubcellular targeting prediction of different programs

Database of Arabidopsis Peroxisomal Proteins

Plant Physiol. Vol. 136, 2004 2603

is difficult to compare and to summarize semiquanti-tatively, we provided a rough definition of high to lowoverall targeting probability of unknown proteins toperoxisomes, which is to be regarded as a startingpoint for subsequent bioinformatics analyses andexperimental verification of the predicted subcellularlocalization. Our conclusion that about 90% of theunknown nonhypothetical Arabidopsis proteins witha major PTS are targeted to peroxisomes with mediumto high probability indicates a high rate of truepositives (Table I). The results of this study, accordingto which two further enzymes, namely 6PGL andHMG-CoA lyase, most likely carry targeting signalsfor both peroxisomes and a second cell compartment,demonstrate that the presence of nonperoxisomaltargeting signals does not preclude targeting to per-oxisomes per se. In our context, peroxisomal targetingprediction can best be strengthened by PTS conserva-tion in homologous plant ESTs (Supplemental Figs. 1and 4). This approach, however, relies on relativelyhigh expression levels in common organs, from whichmost cDNA libraries have been generated, and is, forinstance, not applicable for most low-abundance reg-ulatory proteins or those belonging to large genefamilies, such as F-box proteins, protein kinases, andheat shock proteins.

Novel Proteins from Plant Peroxisomes

Several unknown proteins with a putative PTS arelocalized in peroxisomes with high probability asindicated by overall targeting prediction, EST analysis,and functional evidence provided by homology anal-ysis. The enzymes 2,4-DECR, D3,5, D2,4-DCI, andNADP-IDH have been localized to yeast and mam-malian peroxisomes and are involved in b-oxidation offatty acids with double bonds at even- and possiblyalso odd-numbered carbon atoms. Three Arabidopsisproteins are probably orthologous, catalyze the samereaction as their counterparts in yeast and mammals,and strongly suggest that degradation of unsaturatedfatty acids by NADP-dependent pathways playsa more pronounced role in plant peroxisomal metab-olism than previously assumed. It will be interesting togain insights into the physiological conditions underwhich these alternative pathways are active and howthey are regulated. In addition, we have presentedindications that two isoforms of the OPPP, namely6PGL and 6PGDH, are targeted to peroxisomes andmay provide NADPH for JA biosynthesis or DECRactivity (Fig. 5). These molecular data support theprevious biochemical characterization of the enzymesin peroxisomes (Corpas et al., 1998). Because minoramino acid exchanges can alter the substrate specific-ity, experimental verification of the postulated sub-cellular localization and substrate specificity of allnovel proteins is mandatory.

Two novel paralogs of GOX have been detected thatare expected to catalyze a non-photorespiratory re-action and to be involved in a-oxidation of 3-methyl-

and 2-hydroxy fatty acids, such as the intermediate2-hydroxy-3-methyl-butanoate of Leu catabolism andD-2-hydroxy-8:0 of the degradation of ricinoleic acid(Gerhardt, 1993). Analysis of postulated targeting ofAtHAOX1/2 to peroxisomes by the rare PTS SML.and their substrate specificity may yield awaitedinsights at the molecular level into peroxisomala-oxidation of fatty acids in plants.

Complete degradation of complex fatty acids hasbeen postulated to be accomplished in a hybrid path-way, involving enzymes with specificity for free fattyacids as well as those converting CoA esters (Gerhardt,1993). Depending on their substrate specificity, severalAAEs and thioesterases may be involved. The numberof 15 largely unknown AAEs as well as 6 smallthioesterases all with putative PTSs supports the ideathat complex fatty acids can be oxidized in peroxi-somes by a sequential action of thioesterases, hydroxyacid oxidases, and AAEs. By contrast, our data accord-ing to which GDSL lipases play a role in peroxisomallipid metabolism raises currently more questions thanit provides answers. Much remains to be learned aboutunusual input and output metabolites of peroxisomalb-oxidation and their catabolic pathways.

Apart from metabolic enzymes and peroxisomebiogenesis proteins (PEX proteins), our knowledgeon the presence and function of further peroxisomalmatrix proteins is rather limited. Evidence for theexistence of heat shock proteins, kinases, and phos-phatases in peroxisomes, however, is supported by thedetection of homologs with a putative PTS in thisstudy. In addition of an HSP70 and a DnaJ homolog(HSP40; Wimmer et al., 1997; Diefenbach and Kindl,2000), two members of the large plant family of smallHSPs (Scharf et al., 2001) may be targeted to plantperoxisomes and help refolding of heat-labile matrixproteins. A few protein kinases have recently beenlocalized to plant peroxisomes (Yang and Poovaiah,2002; Dammann et al., 2003; Fukamatsu et al., 2003;Fukao et al. 2003). Our screen of the Arabidopsisgenome reveals seven additional putative proteinkinases that are suspected to mediate signal trans-duction across the peroxisomal membrane and to alterthe activity of peroxisomal enzymes by reversiblephosphorylation. In line with this result many proteinsfrom leaf peroxisomes seem to be posttranslationallymodified, of which some modifications most likelyrepresent reversible phosphorylations (L. Babujee andS. Reumann, unpublished data). Accumulating evi-dence also suggests that turnover of peroxisomalenzymes is highly regulated, for instance under tran-sition of glyoxysomes to leaf peroxisomes and viceversa during senescence. Whether a homolog ofthe Lon protease, a zinc protease, or a glutathioneS-transferase is active in peroxisomes and involved inorchestrated degradation of matrix proteins needs tobe elucidated. The considerable number of partlyannotated unknown proteins not mentioned in thisarticle as well as the 22 completely unknown Arabi-dopsis proteins with a major PTS lacking significant

Reumann et al.

2604 Plant Physiol. Vol. 136, 2004

sequence similarity to any known proteins (Table I) areexpected to bear further surprises.

The Potential of a Genomics Approach

Even though proteome approaches, in which thepowerful resolution of two-dimensional gel electro-phoresis is combined with high-sensitivity mass spec-trometry, allowed considerable advances in the area of‘‘organelle proteomics’’ and led to the identification ofmany unknown proteins (Peltier et al., 2000; Fukaoet al., 2002, 2003; Heazlewood et al., 2004), detection ofmost regulatory and inducible proteins of small,fragile, and low-abundance cell organelles is restrictedto an in silico genomics approach. Indeed, for many ofour Arabidopsis proteins with a putative PTS, corre-sponding ESTs are only detected in collections fromplants subjected to specific stress conditions. About15% of the extracted proteins are still hypothetical,some of which may be induced under currently un-known conditions or in highly distinct plant organs.

Plant homologs of peroxisomal proteins from fungiand mammals can often be detected by homologyanalysis using general protein databases. The mainadvantages of our genome screen and the compilationof all available information on putative peroxisomalmatrix proteins in a user-friendly html-based databaseare: (1) identification of plant-specific peroxisomalproteins; (2) recognition of distantly related homologs(e.g. small thioesterases); (3) detection of proteins thefunction of which has not yet been connected to theorganelle of interest (e.g. Lon protease, GDSL lipases);(4) coverage of the major portion of soluble matrixproteins from plant peroxisomes; and (5) assembly of,at first glance, apparently unrelated metabolic mosaicpieces to novel metabolic pathways. The latter isillustrated by the detection of Arabidopsis proteinswith a putative PTS that are homologous to prokary-otic enzymes of a novel prokaryotic pathway fordegradation of aromatic fatty acids (Benning et al.,1998; Ferrandez et al., 1998; Thoden et al., 2003). Closeplant homologs of the three enzymes involved indegradation of 4-Cl-benzoate, i.e. an acyl-activatingenzyme, a monofunctional enoyl-CoA hydratase, anda small thioesterase, belong to Arabidopsis proteinfamilies with a considerable number of members withhigh peroxisome targeting probability. If prokaryotesrequire specific isoforms for b-oxidation-related reac-tions of these aromatic compounds, specific isoformsneed to be postulated for higher plants as well. Towhich extend aromatic and cyclic compounds, such asaromatic amino acids, chinones, and the plant hor-mones auxin, SA, and JA are degraded in the perox-isomal compartment, is currently largely unknownand requires experimental investigation, consideringthe influence of their turnover rate on cytoplasmichormone concentrations and the induction of short-term signaling cascades. Peroxisomal AAEs and smallthioesterases, both specific for aromatic and cyclicsubstrates also need to be postulated for biosynthesis

of auxin, SA, and JA and are probably PTS-carryingmembers of the protein families presented here. It isalso possible that the two isoforms of MFP are notsuitable for conversion of aromatic and cyclic inter-mediates and that peroxisomal isoforms of monofunc-tional enoyl-CoA hydratase and 3-hydroxyacyl-CoAdehydrogenase catalyze these reactions. Analysis ofthe corresponding Arabidopsis T-DNA insertion k.o.mutants may yield interesting insights into the role ofperoxisomes in the regulation of plant hormone levels.

CONCLUSIONS

We have screened the Arabidopsis genome for novelproteins of plant peroxisomes carrying putative PTSs.These proteins have been compiled in the databaseAraPerox and have been supplemented by known andputative Arabidopsis PEX homologs of yeast andmammalian proteins involved in peroxisome bio-genesis (Charton and Lopez-Huertas, 2002) and bymembrane-associated enzymes as well as membraneproteins involved in metabolite transport. Detailedinformation on the prediction of subcellular localiza-tion is provided to sort out peroxisomal from non-peroxisomal proteins and to design straight-forwardstrategies for experimental validation of postulatedprotein targeting. In addition, quick access to analyt-ical in silico data on homology and gene expression aswell as literature on experimentally studied homologsis provided and expected to facilitate the identificationof new functions and complex biochemical pathwaysof plant peroxisomes.

MATERIALS AND METHODS

Identification of Arabidopsis Proteins with a Putative

PTS1 or PTS2

Genes encoding Arabidopsis proteins with putative PTS1 or PTS2 were

identified using the software PatternSearch at the Arabidopsis Information

Resource (TAIR) server (www.arabidopsis.org). Major and minor PTS1

tripeptides and PTS2 nonapeptides have been defined in a previous publica-

tion (Reumann, 2004). The localization of the PTS2 nonapeptide (R of position

1 of the nonapeptide) was restricted to position 2 to 30 of the N-terminal

domain based on the analysis of about 150 plant homologs of PTS-targeted

proteins (Reumann, 2004). ORF prediction was verified by homology analysis

against the nonredundant database at the National Center for Biotechnology

Information (NCBI; www.ncbi.nlm.nih.gov/). Homologous plant ESTs were

retrieved and analyzed as described (Reumann, 2004). Homologous se-

quences were aligned and bootstrap analysis performed using ClustalX.

Treeview was used for graphical presentation of phylogenetic analyses. Molec-

ular mass and pIs were either taken from the information file at the TAIR

server or, in cases of obvious mistakes, calculated using the Expasy server

(www.expasy.org/tools/pi_tool.html). Targeting, expression, and homology

analyses were updated in January and February 2004 and gene prediction,

annotation, and expression of hypothetical proteins in April 2004.

Targeting Analysis

For targeting prediction the following programs were applied: TargetP

(www.cbs.dtu.dk/services/TargetP/; Emanuelsson et al., 2000), PSORT,

PSORTII, iPSORT (www.psort.org/; Nakai and Kanehisa, 1991; Horton and

Nakai, 1997; Bannai et al., 2002), Predotar (www.inra.fr/predotar/), Mitoprot

(www.mips.biochem.mpg.de/cgi-bin/proj/medgen/mitofilter; Claros and

Database of Arabidopsis Peroxisomal Proteins

Plant Physiol. Vol. 136, 2004 2605

Vincens, 1996), and PTS1 predictor (http://mendel.imp.univie.ac.at/

mendeljsp/sat/pts1/PTS1predictor.jsp; Neuberger et al. 2003a, 2003b), and

DBSubLoc (www.bioinfo.tsinghua.edu.cn/guotao/predict.html; Guo et al.,

2004). Targeting prediction by PeroxiP (Emanuelsson et al., 2003) was not avail-

able yet during preparation of the manuscript. Targeting to a particular cell

compartment was considered reliable if high scores were given for the same

organelle by two independent programs (thresholds, 0.7 for PSORT, TargetP,

Predotar, and Mitoprot; 70% for PSORT II and DBSubLoc; 0.0 for PTS1

predictor).

Expression Analysis by Digital Northern

The nucleotide sequence of the protein of interest was retrieved using

ENTREZ (www.ncbi.nlm.nih.gov/Entrez/). The EST database was searched

for the presence of corresponding Arabidopsis ESTs using TBLASTN

(www.ncbi.nlm.nih.gov/BLAST/). Expressed sequence tags with a minimum

length of 150 bp and a sequence identity of $95% were considered matches to

the protein of interest. For analysis of GOX expression, $97% sequence

identity at the nucleotide level was chosen for At_GOX/2.

ACKNOWLEDGMENTS

We thank Katharina Pawlowski, Martin Fulda, Ivo Feussner, and Hans-

Walter Heldt for stimulating discussions and suggestions on the manuscript.

The support of our research project by Hans-Walter Heldt and Ivo Feussner is

appreciated. The help of Carolin Farke and Caroline Mayer in setting up the

database is acknowledged. If gene cloning and functional analysis of

peroxisomal proteins is based in major parts on information provided by

the database AraPerox, citation of this article is appreciated.

Received March 28, 2004; returned for revision June 14, 2004; accepted June 16,

2004.

LITERATURE CITED

Arabidopsis Genome Initiative (2000) Analysis of the genome sequence of

the flowering plant Arabidopsis thaliana. Nature 408: 796–815

Arpigny JL, Jaeger KE (1999) Bacterial lipolytic enzymes: classification and

properties. Biochem J 343: 177–183

Bannai H, Tamada Y, Maruyama O, Nakai K, Miyano S (2002) Extensive

feature detection of N-terminal protein sorting signals. Bioinformatics

18: 298–305

Benning MM, Wesenberg G, Liu R, Taylor KL, Dunaway-Mariano D,

Holden HM (1998) The three-dimensional structure of 4-hydroxyben-

zoyl-CoA thioesterase from Pseudomonas sp. Strain CBS-3. J Biol Chem

273: 33572–33579

Bunkelmann JR, Trelease RN (1996) Ascorbate peroxidase. A prominent

membrane protein in oilseed glyoxysomes. Plant Physiol 110: 589–598

Charton W, Lopez-Huertas E (2002) PEX genes in plants and other

organisms. In A Baker, IA Graham, eds, Plant Peroxisomes: Biochem-

istry, Cell Biology and Biotechnological Applications, Ed 1. Kluwer

Academic Publishers, Dordrecht, The Netherlands, pp 385–426

Claros MG, Vincens P (1996) Computational method to predict mitochon-

drially imported proteins and their targeting sequences. Eur J Biochem

241: 779–786

Collard F, Collet JF, Gerin I, Veiga-da-Cunha M, Van Schaftingen E (1999)

Identification of the cDNA encoding human 6-phosphogluconolacto-

nase, the enzyme catalyzing the second step of the pentose phosphate

pathway (1). FEBS Lett 459: 223–226

Corpas FJ, Barroso JB, del Rio LA (2001) Peroxisomes as a source of

reactive oxygen species and nitric oxide signal molecules in plant cells.

Trends Plant Sci 6: 145–150

Corpas FJ, Barroso JB, Sandalio LM, Distefano S, Palma JM, Lupianez JA,

Del Rio LA (1998) A dehydrogenase-mediated recycling system of

NADPH in plant peroxisomes. Biochem J 330: 777–784

Corpas FJ, Barroso JB, Sandalio LM, Palma JM, Lupianez JA, del Rio LA

(1999) Peroxisomal NADP-dependent isocitrate dehydrogenase. Char-

acterization and activity regulation during natural senescence. Plant

Physiol 121: 921–928

Crozier A, Kamiya Y, Bishop G, Yokota T (2000) Biosynthesis of hormones

and elicitor molecules. In BB Buchanan, W Gruissem, RL Jones, eds,

Biochemistry and Molecular Biology of Plants, Ed 1. American Society

of Plant Physiologists, Rockville, MD, pp 850–929

Dammann C, Ichida A, Hong B, Romanowsky SM, Hrabak EM, Harmon

AC, Pickard BG, Harper JF (2003) Subcellular targeting of nine calcium-

dependent protein kinase isoforms from Arabidopsis. Plant Physiol 132:

1840–1848

De Nys K, Meyhi E, Mannaerts GP, Fransen M, Van Veldhoven PP (2001)

Characterisation of human peroxisomal 2,4-dienoyl-CoA reductase.

Biochim Biophys Acta 1533: 66–72

del Rio LA, Corpas FJ, Sandalio LM, Palma JM, Gomez M, Barroso JB

(2002) Reactive oxygen species, antioxidant systems and nitric oxide in

peroxisomes. J Exp Bot 372: 1255–1272

Diefenbach J, Kindl H (2000) The membrane-bound DnaJ protein located

at the cytosolic site of glyoxysomes specifically binds the cytosolic

isoform 1 of Hsp70 but not other Hsp70 species. Eur J Biochem 267:

746–754

Durner J, Shah J, Klessig DF (1997) Salicylic acid and disease resistance in

plants. Trends Plant Sci 2: 266–274

Eastmond PJ, Hooks MA, Williams D, Lange P, Bechtold N, Sarrobert C,

Nussaume L, Graham IA (2000) Promoter trapping of a novel medium-

chain acyl-CoA oxidase, which is induced transcriptionally during

Arabidopsis seed germination. J Biol Chem 275: 34375–34381

Elgersma Y, Vos A, van den Berg M, van Roermund CW, van der Sluijs P,

Distel B, Tabak HF (1996) Analysis of the carboxyl-terminal peroxi-

somal targeting signal 1 in a homologous context in Saccharomyces

cerevisiae. J Biol Chem 271: 26375–26382

Emanuelsson O, Elofsson A, von Heijne G, Cristobal S (2003) In silico

prediction of the peroxisomal proteome in fungi, plants and animals.

J Mol Biol 330: 443–456

Emanuelsson O, Nielsen H, Brunak S, von Heijne G (2000) Predicting

subcellular localization of proteins based on their N-terminal amino

acid sequence. J Mol Biol 300: 1005–1016

Ferrandez A, Minambres B, Garcia B, Olivera ER, Luengo JM, Garcia JL,

Diaz E (1998) Catabolism of phenylacetic acid in Escherichia coli.

Characterization of a new aerobic hybrid pathway. J Biol Chem 273:

25974–25986

Feussner I, Wasternack C (2002) The lipoxygenase pathway. Annu Rev

Plant Biol 53: 275–297

Filppula SA, Yagi AI, Kilpelainen SH, Novikov D, FitzPatrick DR,

Vihinen M, Valle D, Hiltunen JK (1998) Delta3,5-delta2,4-dienoyl-

CoA isomerase from rat liver. Molecular characterization. J Biol Chem

273: 349–355

Flynn CR, Mullen RT, Trelease RN (1998) Mutational analyses of a type 2

peroxisomal targeting signal that is capable of directing oligomeric

protein import into tobacco BY-2 glyoxysomes. Plant J 16: 709–720

Froman BE, Edwards PC, Bursch AG, Dehesh K (2000) ACX3, a novel

medium-chain acyl-coenzyme A oxidase from Arabidopsis. Plant Physiol

123: 733–742

Fukamatsu Y, Yabe N, Hasunuma K (2003) Arabidopsis NDK1 is a compo-

nent of ROS signaling by interacting with three catalases. Plant Cell

Physiol 44: 982–989

Fukao Y, Hayashi M, Hara-Nishimura I, Nishimura M (2003) Novel

glyoxysomal protein kinase, GPK1, identified by proteomic analysis of

glyoxysomes in etiolated cotyledons of Arabidopsis thaliana. Plant Cell

Physiol 44: 1002–1012

Fukao Y, Hayashi M, Nishimura M (2002) Proteomic analysis of leaf

peroxisomal proteins in greening cotyledons of Arabidopsis thaliana.

Plant Cell Physiol 43: 689–696

Fulda M, Heinz E, Wolter FP (1997) Brassica napus cDNAs encoding fatty

acyl-CoA synthetase. Plant Mol Biol 33: 911–922

FuldaM, Schnurr J, Abbadi A, Heinz E, Browse J (2004) Peroxisomal Acyl-

CoA synthetase activity is essential for seedling development in

Arabidopsis thaliana. Plant Cell 16: 394–405

Fulda M, Shockey J, Werber M, Wolter FP, Heinz E (2002) Two long-chain

acyl-CoA synthetases from Arabidopsis thaliana involved in peroxisomal

fatty acid beta-oxidation. Plant J 32: 93–103

Gebhardt JS, Wadsworth GJ, Matthews BF (1998) Characterization of

a single soybean cDNA encoding cytosolic and glyoxysomal isozymes

of aspartate aminotransferase. Plant Mol Biol 37: 99–108

Geisbrecht BV, Liang X, Morrell JC, Schulz H, Gould SJ (1999) The mouse

gene PDCR encodes a peroxisomal delta(2), delta(4)-dienoyl-CoA re-

ductase. J Biol Chem 274: 25814–25820

Reumann et al.

2606 Plant Physiol. Vol. 136, 2004

Gerbling H, Gerhardt B (1989) Peroxisomal degradation of branched-chain

2-oxo acids. Plant Physiol 91: 1387–1392

Gerhardt B (1993) Catabolism of fatty acids (a- and b-oxidation). In TS

Moore Jr, ed, Lipid Metabolism in Plants. CRC Press, Boca Raton, FL, pp

527–565

Gould SJ, Keller GA, Hosken N, Wilkinson J, Subramani S (1989)

A conserved tripeptide sorts proteins to peroxisomes. J Cell Biol 108:

1657–1664

Graham IA, Eastmond PJ (2002) Pathways of straight and branched

chain fatty acid catabolism in higher plants. Prog Lipid Res 41: 156–181

Guo T, Hua S, Ji X, Sun Z (2004) DBSubLoc: Database of protein subcellular

localization. Nucleic Acids Res 32 (Database issue): D122–D124

Gurvitz A, Hamilton B, Ruis H, Hartig A (2001) Peroxisomal degradation

of trans-unsaturated fatty acids in the yeast Saccharomyces cerevisiae.

J Biol Chem 276: 895–903

HaslbeckM (2002) sHsps and their role in the chaperone network. Cell Mol

Life Sci 59: 1649–1657

Hayashi M, Aoki M, Kato A, Nishimura M (1997) Changes in targeting

efficiencies of proteins to plant microbodies caused by amino acid

substitutions in the carboxyl-terminal tripeptide. Plant Cell Physiol 38:

759–768

Hayashi H, De Bellis L, Ciurli A, Kondo M, Hayashi M, Nishimura M

(1999) A novel acyl-CoA oxidase that can oxidize short-chain acyl-CoA

in plant peroxisomes. J Biol Chem 274: 12715–12721

Heazlewood JL, Tonti-Filippini JS, Gout AM, Day DA, Whelan J, Millar

AH (2004) Experimental analysis of the Arabidopsis mitochondrial

proteome highlights signaling and regulatory components, provides

assessment of targeting prediction programs, and indicates plant-

specific mitochondrial proteins. Plant Cell 16: 241–256

Henke B, Girzalsky W, Berteaux-Lecellier V, Erdmann R (1998) IDP3

encodes a peroxisomal NADP-dependent isocitrate dehydrogenase

required for the beta-oxidation of unsaturated fatty acids. J Biol Chem

273: 3702–3711

Hooks MA (2002) Molecular biology, enzymology, and physiology of

b-oxidation. InA Baker, IA Graham,eds,Plant Peroxisomes:Biochemistry,

Cell Biology and Biotechnological Applications, Ed 1. Kluwer Academic

Publishers, Dordrecht, The Netherlands, pp 19–55

Hooks MA, Kellas F, Graham IA (1999) Long-chain acyl-CoA oxidases of

Arabidopsis. Plant J 20: 1–13

Horton P, Nakai K (1997) Better prediction of protein cellular localization

sites with the k nearest neighbors classifier. Proc Int Conf Intell Syst Mol

Biol 5: 147–152

Hunt MC, Solaas K, Kase BF, Alexson SE (2002) Characterization of an

acyl-CoA thioesterase that functions as a major regulator of peroxisomal

lipid metabolism. J Biol Chem 277: 1128–1138

Janne J, Alhonen L, Pietila M, Keinanen TA (2004) Genetic approaches to

the cellular functions of polyamines in mammals. Eur J Biochem 271:

877–894

Jimenez A, Hernandez JA, Del Rio LA, Sevilla F (1997) Evidence for the

presence of the ascorbate-glutathione cycle in mitochondria and perox-

isomes of pea leaves. Plant Physiol 114: 275–284

Jones JM, Gould SJ (2000) Identification of PTE2, a human peroxisomal

long-chain acyl-CoA thioesterase. Biochem Biophys Res Commun 275:

233–240

Jones JM, Morrell JC, Gould SJ (2000) Identification and characterization

of HAOX1, HAOX2, and HAOX3, three human peroxisomal 2-hydroxy

acid oxidases. J Biol Chem 275: 12590–12597

Jones JM, Morrel JC, Gould SJ (2001) Multiple distinct targeting signals in

integral peroxisomal membrane proteins. J Cell Biol 153: 1141–1150

Kamada T, Nito K, Hayashi H, Mano S, Hayashi M, Nishimura M (2003)

Functional differentiation of peroxisomes revealed by expression pro-

files of peroxisomal genes in Arabidopsis thaliana. Plant Cell Physiol 44:

1275–1289

Kamigaki A, Mano S, Terauchi K, Nishi Y, Tachibe-Kinoshita Y, Nito K,

Kondo M, Hayashi M, Nishimura M, Esaka M (2003) Identification of

peroxisomal targeting signal of pumpkin catalase and the binding

analysis with PTS1 receptor. Plant J 33: 161–175

Kato A, Takeda-Yoshikawa Y, Hayashi M, Kondo M, Hara-Nishimura I,

Nishimura M (1998) Glyoxysomal malate dehydrogenase in pumpkin:

cloning of a cDNA and functional analysis of its presequence. Plant Cell

Physiol 39: 186–195

Kikuchi M, Hatano N, Yokota S, Shimozawa N, Imanaka T, Taniguchi H

(2004) Proteomic analysis of rat liver peroxisome: presence of peroxi-

some-specific isozyme of Lon protease. J Biol Chem 279: 421–428

Kim KK, Kim R, Kim SH (1998) Crystal structure of a small heat-shock

protein. Nature 394: 595–599

Kleff S, Sander S, Mielke G, Eising R (1997) The predominant protein in

peroxisomal cores of sunflower cotyledons is a catalase that differs in

primary structure from the catalase in the peroxisomal matrix. Eur J

Biochem 245: 402–410

Kliebenstein DJ, Monde RA, Last RL (1998) Superoxide dismutase in

Arabidopsis: an eclectic enzyme family with disparate regulation and

protein localization. Plant Physiol 118: 637–650

Kohler SA, Menotti E, Kuhn LC (1999) Molecular cloning of mouse

glycolate oxidase. High evolutionary conservation and presence of an

iron-responsive element-like sequence in the mRNA. J Biol Chem 274:

2401–2407

Kragler F, Lametschwandtner G, Christmann J, Hartig A, Harada JJ (1998)

Identification and analysis of the plant peroxisomal targeting signal 1

receptor NtPEX5. Proc Natl Acad Sci USA 95: 13336–13341

Kuroda H, Takahashi N, Shimada H, Seki M, Shinozaki K, Matsui M

(2002) Classification and expression analysis of Arabidopsis F-box-

containing protein genes. Plant Cell Physiol 43: 1073–1085

Lametschwandtner G, Brocard C, Fransen M, Van Veldhoven P, Berger J,

Hartig A (1998) The difference in recognition of terminal tripeptides as

peroxisomal targeting signal 1 between yeast and human is due to

different affinities of their receptor Pex5p to the cognate signal and to

residues adjacent to it. J Biol Chem 273: 33635–33643

Lindqvist Y, Braenden CI (1989) The active site of spinach glycolate

oxidase. J Biol Chem 264: 3624–3628

Livingstone JR, Maruo T, Yoshida I, Tarui Y, Hirooka K, Yamamoto Y,

Tsutui N, Hirasawa E (2003) Purification and properties of betaine

aldehyde dehydrogenase from Avena sativa. J Plant Res 116: 133–140

Lopez-Huertas E, Charlton WL, Johnson B, Graham IA, Baker A (2000)

Stress induces peroxisome biogenesis genes. EMBO J 19: 6770–6777

Mekhedov S, de Ilarduya OM, Ohlrogge J (2000) Toward a functional

catalog of the plant genome. A survey of genes for lipid biosynthesis.

Plant Physiol 122: 389–402

Mettler IJ, Beevers H (1980) Oxidation of NADH in glyoxysomes by

a malate-aspartate shuttle. Plant Physiol 66: 555–560

Morel F, Rauch C, Petit E, Piton A, Theret N, Coles B, Guillouzo A (2004)

Gene and protein characterization of the human glutathione S-trans-

ferase kappa and evidence for a peroxisomal localization. J Biol Chem

279: 16246–16253

Mullen RT, Lee MS, Flynn CR, Trelease RN (1997a) Diverse amino acid

residues function within the type 1 peroxisomal targeting signal.

Implications for the role of accessory residues upstream of the type 1

peroxisomal targeting signal. Plant Physiol 115: 881–889

Mullen RT, Lee MS, Trelease RN (1997b) Identification of the peroxisomal

targeting signal for cottonseed catalase. Plant J 12: 313–322

Nakai K, Kanehisa M (1991) Expert system for predicting protein local-

ization sites in gram-negative bacteria. Proteins 11: 95–110

Nakamura T, Yokota S, Muramoto Y, Tsutsui K, Oguri Y, Fukui K, Takabe

T (1997) Expression of a betaine aldehyde dehydrogenase gene in rice,

a glycinebetaine nonaccumulator, and possible localization of its protein

in peroxisomes. Plant J 11: 1115–1120

Neuberger G, Maurer-Stroh S, Eisenhaber B, Hartig A, Eisenhaber F

(2003a) Motif refinement of the peroxisomal targeting signal 1 and

evaluation of taxon-specific differences. J Mol Biol 328: 567–579

Neuberger G, Maurer-Stroh S, Eisenhaber B, Hartig A, Eisenhaber F

(2003b) Prediction of peroxisomal targeting signal 1 containing proteins

from amino acid sequence. J Mol Biol 328: 581–592

Pandey S, Ranade SA, Nagar PK, Kumar N (2000) Role of polyamines and

ethylene as modulators of plant senescence. J Biosci 25: 291–299

Peltier JB, Friso G, Kalume DE, Roepstorff P, Nilsson F, Adamska I, van

Wijk KJ (2000) Proteomics of the chloroplast: systematic identification

and targeting analysis of lumenal and peripheral thylakoid proteins.

Plant Cell 12: 319–341

Perez-Amador MA, Leon J, Green PJ, Carbonell J (2002) Induction of the

arginine decarboxylase ADC2 gene provides evidence for the involve-

ment of polyamines in the wound response in Arabidopsis. Plant Physiol

130: 1454–1463

Preisig-Muller R, Guhnemann-Schafer K, Kindl H (1994) Domains of the

tetrafunctional protein acting in glyoxysomal fatty acid beta-oxidation.

Database of Arabidopsis Peroxisomal Proteins

Plant Physiol. Vol. 136, 2004 2607

Demonstration of epimerase and isomerase activities on a peptide

lacking hydratase activity. J Biol Chem 269: 20475–20481

Reumann S (2002) The photorespiratory pathway of leaf peroxisomes. In A

Baker, IA Graham, eds, Plant Peroxisomes: Biochemistry, Cell Biology

and Biotechnological Applications, Ed 1. Kluwer Academic Publishers,

Dordrecht, The Netherlands, pp 141–189

Reumann S (2004) Specification of the peroxisome targeting signals type 1

and type 2 of plant peroxisomes by bioinformatics analyses. Plant

Physiol 135: 783–800

Reumann S, Bettermann M, Benz R, Heldt HW (1997) Evidence for the

presence of a porin in the membrane of glyoxysomes of Ricinus

communis L. Plant Physiol 115: 891–899

Reumann S, Heupel R, Heldt HW (1994) Compartmentation studies on

spinach leaf peroxisomes. II. Evidence for the transfer of reductant from

the cytosol to the peroxisomal compartment via a malate shuttle. Planta

193: 167–173

Reumann S, Maier E, Heldt HW, Benz R (1998) Permeability properties

of the specific porin of spinach leaf peroxisomes. Eur J Biochem 251:

359–366

Richmond TA, Bleecker AB (1999) A defect in beta-oxidation causes

abnormal inflorescence development in Arabidopsis. Plant Cell 11:

1911–1924

Riedel K, Talker-Huiber D, Givskov M, Schwab H, Eberl L (2003)

Identification and characterization of a GDSL esterase gene located

proximal to the swr quorum-sensing system of Serratia liquefaciens MG1.

Appl Environ Microbiol 69: 3901–3910

Scharf KD, Siddique M, Vierling E (2001) The expanding family of

Arabidopsis thaliana small heat stress proteins and a new family of

proteins containing alpha-crystallin domains (Acd proteins). Cell Stress

Chaperones 6: 225–237

Schmitz A, Gartemann KH, Fiedler J, Grund E, Eichenlaub R (1992)

Cloning and sequence analysis of genes for dehalogenation of 4-chloro-

benzoate from Arthrobacter sp. strain SU. Appl Environ Microbiol 58:

4068–4071

Sebela M, Brauner F, Radova A, Jacobsen S, Havlis J, Galuszka P, Pec P

(2000) Characterisation of a homogeneous plant aminoaldehyde de-

hydrogenase. Biochim Biophys Acta 1480: 329–341

Sekimoto H, Seo M, Kawakami N, Komano T, Desloire S, Liotenberg S,

Marion-Poll A, Caboche M, Kamiya Y, Koshiba T (1998) Molecular

cloning and characterization of aldehyde oxidases in Arabidopsis thali-

ana. Plant Cell Physiol 39: 433–442

Seo M, Koiwai H, Akaba S, Komano T, Oritani T, Kamiya Y, Koshiba T

(2000) Abscisic aldehyde oxidase in leaves of Arabidopsis thaliana. Plant J

23: 481–488

Shockey JM, Fulda MS, Browse JA (2002) Arabidopsis contains nine long-

chain acyl-coenzyme A synthetase genes that participate in fatty acid

and glycerolipid metabolism. Plant Physiol 129: 1710–1722

Shockey JM, Fulda MS, Browse J (2003) Arabidopsis contains a large

superfamily of acyl-activating enzymes. Phylogenetic and biochemical

analysis reveals a new class of acyl-coenzyme a synthetases. Plant

Physiol 132: 1065–1076

Staswick PE, Tiryaki I, Rowe ML (2002) Jasmonate response locus JAR1

and several related Arabidopsis genes encode enzymes of the fire-

fly luciferase superfamily that show activity on jasmonic, salicylic,

and indole-3-acetic acids in an assay for adenylation. Plant Cell 14:

1405–1415

Stintzi A, Browse J (2000) The Arabidopsis male-sterile mutant, opr3, lacks

the 12-oxophytodienoic acid reductase required for jasmonate synthe-

sis. Proc Natl Acad Sci USA 97: 10625–10630

Swinkels BW, Gould SJ, Bodnar AG, Rachubinski RA, Subramani S

(1991) A novel, cleavable peroxisomal targeting signal at the amino-

terminus of the rat 3-ketoacyl-CoA thiolase. EMBO J 10: 3255–3262

Taler D, Galperin M, Benjamin I, Cohen Y, Kenigsbuch D (2004) Plant eR

genes that encode photorespiratory enzymes confer resistance against

disease. Plant Cell 16: 172–184

Thoden JB, Zhuang Z, Dunaway-Mariano D, Holden HM (2003) The

structure of 4-hydroxybenzoyl-CoA thioesterase from Arthrobacter sp.

strain SU. J Biol Chem 278: 43709–43716

Tilton GB, Shockey JM, Browse J (2004) Biochemical and molecular

characterization of ACH2, an Acyl-CoA thioesterase from Arabidopsis

thaliana. J Biol Chem 279: 7487–7494

Titorenko VI, Rachubinski RA (2004) The peroxisome: orchestrating

important developmental decisions from inside the cell. J Cell Biol

164: 641–645

Tuinstra RL, Burgner JW II, Miziorko HM (2002) Investigation of the

oligomeric status of the peroxisomal isoform of human 3-hydroxy-3-

methylglutaryl-CoA lyase. Arch Biochem Biophys 408: 286–294

Upton C, Buckley JT (1995) A new family of lipolytic enzymes? Trends

Biochem Sci 20: 1798–1799

van Roermund CW, Elgersma Y, Singh N, Wanders RJ, Tabak HF (1995)

The membrane of peroxisomes in Saccharomyces cerevisiae is imperme-

able to NAD(H) and acetyl-CoA under in vivo conditions. EMBO J 14:

3480–3486

van Roermund CW, Hettema EH, Kal AJ, van den Berg M, Tabak HF,

Wanders RJ (1998) Peroxisomal beta-oxidation of polyunsaturated fatty

acids in Saccharomyces cerevisiae: isocitrate dehydrogenase provides

NADPH for reduction of double bonds at even positions. EMBO J 17:

677–687

Verma DPS (2002) Peroxisome biogenesis in root nodules and assimilation

of symbiotically-reduced nitrogen in tropical legumes. In A Baker, IA

Graham, eds, Plant Peroxisomes: Biochemistry, Cell Biology and Bio-

technological Applications, Ed 1. Kluwer Academic Publishers, Dor-

drecht, The Netherlands, pp 191–220

Vojtechova M, Hanson AD, Munoz-Clares RA (1997) Betaine-aldehyde

dehydrogenase from amaranth leaves efficiently catalyzes the NAD-

dependent oxidation of dimethylsulfoniopropionaldehyde to dimethyl-

sulfoniopropionate. Arch Biochem Biophys 337: 81–88

Volokita M, Somerville CR (1987) The primary structure of spinach

glycolate oxidase deduced from the DNA sequence of a cDNA clone.

J Biol Chem 262: 15825–15828

Vujcic S, Liang P, Diegelman P, Kramer DL, Porter CW (2003) Genomic

identification and biochemical characterization of the mammalian poly-

amine oxidase involved in polyamine back-conversion. Biochem J 370:

19–28

Walden R, Cordeiro A, Tiburcio AF (1997) Polyamines: small molecules

triggering pathways in plant growth and development. Plant Physiol

113: 1009–1013

Wei Y, Schottel JL, Derewenda U, Swenson L, Patkar S, Derewenda ZS

(1995) A novel variant of the catalytic triad in the Streptomyces scabies

esterase. Nat Struct Biol 2: 218–223

Wilhelm S, Tommassen J, Jaeger KE (1999) A novel lipolytic enzyme

located in the outer membrane of Pseudomonas aeruginosa. J Bacteriol 181:

6977–6986

Willekens H, Chamnongpol S, Davey M, Schraudner M, Langebartels C,

VanMontaguM, Inze D, Van CampW (1997) Catalase is a sink for H2O2

and is indispensable for stress defence in C3 plants. EMBO J 16:

4806–4816

Williams E, Cregeen D, Rumsby G (2000) Identification and expression

of a cDNA for human glycolate oxidase. Biochim Biophys Acta 1493:

246–248

Wimmer B, Lottspeich F, van der Klei I, Veenhuis M, Gietl C (1997) The

glyoxysomal and plastid molecular chaperones (70-kDa heat shock

protein) of watermelon cotyledons are encoded by a single gene. Proc

Natl Acad Sci USA 94: 13624–13629

Wu T, Yankovskaya V, McIntire WS (2003) Cloning, sequencing, and

heterologous expression of the murine peroxisomal flavoprotein,

N1-acetylated polyamine oxidase. J Biol Chem 278: 20514–20525

Yamada K, Lim J, Dale JM, Chen H, Shinn P, Palm CJ, Southwick AM, Wu

HC, Kim C, Nguyen M, et al (2003) Empirical analysis of transcriptional

activity in the Arabidopsis genome. Science 302: 842–846

Yang T, Poovaiah BW (2002) Hydrogen peroxide homeostasis: activation

of plant catalase by calcium/calmodulin. Proc Natl Acad Sci USA 99:

4097–4102

Zolman BK, Monroe-Augustus M, Thompson B, Hawes JW, Krukenberg

KA, Matsuda SPT, Bartel B (2001) chy1, an Arabidopsis mutant with

impaired b-oxidation, is defective in a peroxisomal b-hydroxyiso-

butyryl-CoA hydrolase. J Biol Chem 276: 31037–31046

Reumann et al.

2608 Plant Physiol. Vol. 136, 2004