1Laboratoire de PhysiologieCellulaire Vgtale,DRDC, CEA-Grenoble,Grenoble, France
2Laboratoire de Chimiedes protines, DRDC,CEA-Grenoble, France
3Laboratoire de BionergtiqueCellulaire et Pathologique,DRDC, CEA-Grenoble, France
4UMR Physiologie Molculairedes Semences,Universit dAngers/INH/INRA,Angers, France
A survey of the plant mitochondrial proteomein relation to development
To expand the functional analysis of plant mitochondria, we have undertaken the build-ing of the proteome of pea mitochondria purified from leaves (green and etiolated),roots and seeds. In the first stage, we focused our proteomic exploration on the solubleprotein complement of the green leaf mitochondria. We used traditional two-dimen-sional polyacrylamide gel electrophoresis, in combination with size exclusion chroma-tography as a third dimension, to identify the major proteins and further resolve theirmacromolecular complexity. The two-dimensional map of soluble proteins of greenleaf mitochondria revealed 433 spots (with Coomassie blue staining) and around 73%of the proteins (in mass) were identified using three different approaches: Edmandegradation, matrix-assisted laser desorption/ionization mass spectrometry andelectrospray ionization tandem mass spectrometry. Quite a lot of the polypeptideswere present in multiforms which indicated the presence of isoforms or the occurrenceof post-translational modifications. Among these proteins, we uncovered an abundantfamily that was identified as aldehyde dehydrogenases, representing approximately7.5% of the soluble proteins. The comparative analysis of soluble mitochondrial pro-teomes led to the identification of a number of proteins which were specifically presentin root or in seed mitochondria, thus revealing the impact of tissue differentiation at themitochondrial level.
Keywords: Aldehyde dehydrogenase / Mass spectrometry / Metabolism / Plant mitochondria /Two-dimensional gel electrophoresis PRO 0210
Mitochondria are semiautonomous organelles whose uni-versally recognised function is to provide cellular ATP bythe process of oxidative phosphorylation. Recent devel-opments in cell biology, genetics and medical sciencehave placed mitochondria in the foreground as keyplayers in fields as diverse as evolution, cell death and agrowing number of severe pathologies . Comparedto their animal counterparts, plant mitochondria carryother primordial biosynthetic functions related to theautotrophic status of plants and their inability to avoidstress factors. Prominent features of plant mitochondriainclude the presence of additional NAD(P)H dehydro-genases and an alternative oxidase in the inner mem-
brane [5, 6] that confer an increased flexibility in energytransfer. In photosynthetic tissues, mitochondria are ob-ligatory partners of chloroplasts with respect to CO2light-dependent assimilation, due to their implication inthe photorespiratory pathway . Their contributioninvolves the oxidation of glycine at very high ratesbecause of a tremendous accumulation of the glycinecleavage system (or glycine decarboxylase complex) inthe matrix space .
More recently, it has been shown that plant mitochondriaare also implicated in the synthesis of essential cofactorssuch as biotin , folate , lipoate [11, 12] and vitaminC . These examples emphasise the functional com-plexity of plant mitochondria which cover a great varietyof biochemical reactions, from housekeeping to bio-genesis, in a coordinated fashion with the cellular en-vironment. However, in contrast to the vast and growingnumber of mitochondrial functions, relatively few mito-chondrial proteins have been characterised. The recentdevelopment of proteomics  opens the path to-ward a deeper exploration of mitochondrial functionsthrough their protein complement. Previous work dealingwith organelle proteome analysis has proven the useful-ness of the approach, for instance in the case of humanmitochondria , macrophage phagosome , thy-lakoid proteins of chloroplast  and chloroplast enve-lope membranes .
Correspondence: Dr. Jacques Bourguignon, Laboratoire dePhysiologie Cellulaire Vgtale, UMR 5019, CEA / CNRS / Uni-versit Joseph Fourier, Dpartement de Biologie Molculaire etStructurale, CEA-Grenoble, 17 rue des Martyrs, 38054 Grenoblecedex 9, FranceE-mail: firstname.lastname@example.orgFax: +33-4-38-78-5091
Abbreviations: ALDH, aldehyde dehydrogenase; FDH, formatedehydrogenase; GCS, glycine cleavage system; MDH, malatedehydrogenase; MnSOD, manganese superoxide dismutase;RNP, RNA binding protein; SCADH, short-chain alcohol dehydro-genase; SHMT, serine hydroxymethyltransferase; SVC, succi-nyl-CoA ligase
880 Proteomics 2002, 2, 880898
WILEY-VCH Verlag GmbH, 69451 Weinheim, 2002 1615-9853/02/0707880 $17.50+.50/0
Proteomics 2002, 2, 880898 Plant mitochondria proteomics and development 881
In the present work, we have undertaken a comprehen-sive functional analysis of plant mitochondria by buildingthe proteome of mitochondria isolated from differenttissues and organs from pea. In the first stage, wefocusedon the soluble protein complementwhosemacro-molecular complexity was further resolved by combiningsize exclusion chromatography with the traditional two-dimensional polyacrylamide gel electrophoresis separa-tion. We report here the 2-D map of soluble proteins ofmitochondria isolated frompea leaves (green or etiolated),roots and seeds together with the identification of themajor proteins, results which are discussed in a physio-logical context.
2 Materials and methods
2.1 Isolation of pea mitochondria andpreparation of a soluble mitochondrialprotein extract (matrix extract)
Mitochondria were isolated and purified from leaves of12-day old pea plants as described by Douce et al. .The procedure is composed of differential centrifugationfollowed by two steps of purification on Percoll/polyvinyl-pyrrolidone gradient. In the case of the purification ofmitochondria from etiolated leaves and roots, the pro-cedure used for the purification of the potato tuber mito-chondria was employed  with two steps of purificationon Percoll gradient. The mitochondria were isolated from16 h imbibed pea seeds according to the proceduredescribed by Benamar, Tallon and Macheral (manuscriptin preparation). These purified mitochondria were sub-mitted to an osmotic shock by dilution (1:20) in a buffercontaining 40mM MOPS (pH 7.4) and 1mM DTT and thesoluble proteins were released from the mitochondria bythree freeze-thaw cycles. After removal of mitochondrialmembrane by ultracentrifugation at 38000 rpm for 1 h(Beckman SW-40 rotor), the supernatant containing themitochondrial soluble proteins was concentrated on a3 kDa Macrosep (Pall, Portsmouth, UK) to a concentrationof 4050mg/mL.
2.2 Fractionation of the soluble mitochondrialproteins
The soluble protein extract (200 mg) prepared from pealeaf mitochondria was applied to a Sephacryl S-300 col-umn (Amersham Biosciences, Uppsala, Sweden) equili-brated with a buffer containing 50 mM Tris (pH 7.4) andDTT 1 mM. The proteins were then eluted, as function oftheir molecular mass, with the same buffer at 4C with aflow rate of 0.3 mL/min (fraction size, 3 mL), the glycine
cleavage system components and the serine hydroxy-methyltransferase being used as standards to calibratethe column.
2.3 Sample preparation for 2-D gels andIPG focusing
Soluble mitochondrial proteins (500 g) were diluted in asolution containing at the final concentration: 7 M urea,2.0 M thiourea, 4% w/v CHAPS, 20 mM DTT and 0.4%w/v carrier ampholytes (Pharmalytes 310; AmershamBiosciences) . Proteins were separated by 2-DE, es-sentially as previously described on home-made pH 48gradients .
2.4 Equilibration, SDS dimension and staining
After equilibration as previously described , the stripswere sealed on top of the 1.5 mm thick 2-D gel (Bio-Radvertical system; Hercules, CA, USA) with the help of1% low melting agarose in 0.2% SDS, 0.15 M N,N-bis(hydroxyethyl)N,N,N-tris(hydroxymethyl)aminomethane(Bis-Tris), 0.1 M HCl buffer supplemented with bromo-phenol blue as a tracking dye. SDS-PAGE was carriedout at constant power (15 W per gel) with cooling at10C, until the tracking dye reached the bottom of thegel. Gels were then incubated in a solution of 10% aceticacid and 50% methanol and stained with the same solu-tion containing 2.5 g/L Coomassie Blue (R-250).
2.5 Protein identification by N-terminalsequencing
When N-terminal sequencing was performed, the 2-Dgels were blotted onto PVDF membranes (ProBlott mem-branes; Applied Biosystems, Foster City, CA, USA) usingthe Bio-Rad TransBlot Cell according to the manufac-turers instructions. After transfer, at 90 V for 3 h, mem-branes were washed twice for 5 min with deionized waterand then stained for 15 min with the coloration solutioncontaining: 0.1% w/v Coomassie Blue R-250, 10% v/vacetic acid and 40% v/v ethanol. The membranes weredestained with a solution of 30% v/v ethanol. Proteinspots blotted on a PVDF membrane and stained withCoomassie blue were excised and sequenced on an ABI492 Procise sequencer (Applied Biosystems).
2.6 Protein digestion
The protein spots of interest were excised from Coomas-sie blue stained 2-D gels and washed with 25 mM ammo-nium hydrogenocarbonate (NH4HCO3) (pH 8.0) for 2 h,
882 J. Bardel et al. Proteomics 2002, 2, 880898
and then with 50% v/v acetonitrile, 25 mM NH4HCO3for another 2 h. This procedure was repeated until the gelpieces were destained and complete dehydration wasperformed with a vacuum centrifuge (Eppendorff Speed-Vac). Three to 5 L of a solution containing 0.1 to 0.5 g oftrypsin (porcine sequencing grade modified, Promega),25 mM NH4HCO3 (pH 8.0) and 10% v/v acetonitrile wasadded to the tube and the digestion proceeded at 37Cfor 3 to 4 h.
2.7 MALDI analysis
Mass spectra of the tryptic digests were acquired on aBiflex (Bruker-Daltonic, Bremen, Germany) MALDI-TOFmass spectrometer equipped with a gridless delayedextraction. The instrument was operated in the reflectormode. 0.5 L of each digest solution (in 25 mMNH4HCO3/10% acetonitrile) was deposited directly ontothe sample probe on a dry thin layer of matrix. The matrixwas prepared by mixing four volumes of saturated solu-tion of -cyano-4-hydroxy-trans-cinnamic acid (CCA) inacetone with three volumes of solution of nitrocellulose(10 mg/mL) in acetone/isopropanol (1:1). Deposits werewashed with 5 L of 0.1% v/v TFA before analysis. Amass list of peptides was obtained for each proteindigest. This peptide mass fingerprint was then sub-mitted to an appropriate software to identify the proteins(MS-Fit, available online at http://prospector.ucsf.edu/ucsfhtml3.4/msfit. htm, or ProFound, available online athttp://220.127.116.11/prowl-cgi/ProFound.exe). When aprotein could not be identified from its tryptic peptidemass map, the tryptic digest was extracted twice with a50% v/v acetonitrile solution containing 25 mM NH4HCO3.The digest solution and the extracts were then pooled,dried in a vacuum centrifuge, and desalted with ZipTipC18 (Millipore, Bedford, MA, USA) before the electrospraytandem mass spectrometry analysis.
Aquadrupole time-of-flight (Q-TOF) instrument (Micromass,Manchester, UK) was usedwith a Z-spray ion-source work-ing in the nanospray mode. Approximately 3 to 5 L of thedesalted sample was introduced into a needle (mediumsample needle; MDS Protana, Odense, Denmark) to runMS and MS/MS experiments. The capillary voltage was setto an average of 1000 V, and the sample cone to 50 V.Glufibrinopeptide was used to calibrate the instrument inthe MS/MS mode. MS/MS spectra were transformed usingMaxEnt3 (MassLynx; Micromass). Amino acid sequences,sequence tags,orpeptide ion fragments thatcouldbedeter-mined were used to screen the protein and expression tag(EST) databases with dedicated software: Pepfrag (http://
prowl1.rockefeller.edu/prowl/pepfrag.html), Scan (http://dna.stanford.edu/scan), BLAST (http://ncbi.nlm. nih.gov/blast/blast.cgi) or FASTA3 (http://www.ebi.ac.uk/fasta3/)for searching homologies.
2.9 Bioinformatic analysis
When indicated, the prediction of mitochondrial targetingfor protein sequences issued from database homologiessearch was performed with four different programs avail-able on the internet: MitoProt (http://www.mips.biochem.mpg.de/cgi-bin/proj/medgen/mitofilter) , Predotar V0.5(http://www.inra.fr/Internet/Produits/Predotar/), PSORT(http://psort.nibb.ac.jp) , and TargetP V1:01 (http://www.cbs.dtu.dk/services/TargetP/) .
3.1 Exploration of the proteome of pea leafmitochondria toward a 3-D analysis:gel filtration as the first dimension
Highly purified mitochondria were isolated from pealeaves according to Douce et al.  using a two-roundisopycnic purification on Percoll gradients. The solubleproteins were released by a combination of osmoticshock and freeze-thaw cycles and separated from mem-brane proteins by high speed centrifugation. The solublemitochondrial proteome was analysed by high resolution2-D PAGE that combines IEF on immobilized pH gradi-ents in the first dimension with SDS-PAGE as the seconddimension. Initial experiments done on wide range IPGstrips (pH 310) revealed that almost all the proteinsexhibited pIs between 4 and 8 (results not shown). Sucha distribution led us to prepare strips in this range toincrease the resolution during the first dimension. The2-DE map of the soluble proteins (500 g) from pea leafmitochondria is presented in Fig. 1. When the gel wasstained with Coomassie blue, 433 spots were detectedwith the PDQuest 2-D gel analysis software (Bio-Rad).The spots were quantified and the values obtained rang-ed from 972 to 1143997 arbitrary detection units for atotal integrity density of 17975406. This means that a pro-tein which represents 0.005% of the initial input can bevisualised. Since the information concerning pea proteinsin databases was limited (286 proteins in SWISS-PROTand 485 in TrEMBL, May 2002), our strategy was to useN-terminal Edman degradation as the primary tool foridentification of protein spots. After electroblotting onPVDF membranes, the major visible spots were analysedyielding a set of peptide sequence tags between 8 and 24residues that are listed in Table 1.
Proteomics 2002, 2, 880898 Plant mitochondria proteomics and development 883
Table 1. Pea proteins identified from the 2-D gel analyses of the soluble proteins of green leaf mitochondria (pI 4.0 to 8.0)
Name or description SpotNo.
Edman* or ESI-MS/MS MALDI pIobs
sequence % of identity/% of similarity
Proteins involved in the photorespiratory pathway
Glycine cleavage system P protein(GCS P)
ISVEALKPSDTF* 100 () 6.4 104
Glycine cleavage system H protein(GCS H)
386, 10, 17
SNVLDGLKYA* 100 () 4.2 14
Glycine cleavage system T protein(GCS T)
ATESELKKTVL* 100 383838
Dihydrolipoamide dehydrogenase(DLDH, also called E3) or glycinecleavage system L protein (GCSL)
54 6.1 57
47 6.3 57
37 6.5 57
Serine hydroxymethyl transferase(SHMT)
Proteins involved in the tricarboxylic cycle, carbon metabolism and bioenergetics
Malate dehydrogenase (MDH) 3416,4408 (8)
ATEPVPEAKVAILGA* 93/93 () 6.16.6
Succinyl-CoA Ligase ( subunit) (SUC ) 1515 CAA05024(Arath)
LNIHEYQGAELMSKYGVN* 94/100 () 5.4 41
Citrate synthase (CISY) 5508 PO20115(Arath)
LQMQSSTD* 88/100 () 6.8 45
Pyruvate dehydrogenase E1 subunit(PDH E1)
() 6.6 47
Pyruvate dehydrogenase E1 subunit(PDH E1)
() 6.3 41
Pyruvate dehydrogenase E1 subunit,PDH E1
2405 (9) P52904(Pisum)
XSXAKQMTVXDAL* 83/83 () 5.5 37
Pyruvate dehydrogenase E2 subunit,PDH E2 (dihydrolipoamide acetyl-transferase)
() 5.8 52
NAD-aldehyde dehydrogenase(homologue to the T cytoplasm malesterility restorer factor 2)(ALDH 14)
3638 (8)4625 (11)4615 (3)4616
Aldehyde dehydrogenase (ALDH 5) 5608 P93344(Nicotiana)
(+) 6.7 56
884 J. Bardel et al. Proteomics 2002, 2, 880898
Table 1. Continued
Name or description SpotNo.
Edman* or ESI-MS/MSMALDI MALDI pIobs
sequence % of identity/% of similarity
Aldehyde dehydrogenase (ALDH 6 & 7)(FIS1)
Aldehyde dehydrogenase (ALDH) 5609 Q9SU63(Arath)
() 6.8 54
Methyl-malonate semialdehydedehydrogenase (MMSDH1)
YGNGASLFTTSEAR 71/79 6.8 54
Aldehyde dehydrogenase (putativemethyl-malonate semialdehydedehydrogenase) (MMSDH2)
() 7.1 54
Short-chain alcohol dehydrogenase-likeprotein (SCADH)
AEAALYLASDESK 100 () 6.8 30
Formate dehydrogenase (FDH) 4516,5503 (5)
ATP synthase, subunit (ATP) 1608 (2),1626
ATP synthase, subunit (ATP) 1 Q39852(Glycine)
AKGSELPALKGDEML* 80/80 () 5.8 28
Proteins involved in protein folding and stabilisation
Heat shock protein 70 kDa (HSP70) 1621, 1622,1623 (7)
SRPAGNDVIGIDLGT* 100 () 5.25.4
Heat shock protein 60 kDa ( HSP60) 1617, 2629,2625 (4)
90/100 () 5.35.5
Proteins involved against stress
Manganese superoxyde dismutase(MnSOD)
96/96 () 6.16.6
Putative ascorbate peroxidase 2301 (5),2302 (20)
GTVSALASDPDQLK* 88/100 () 5.55.6
Nucleoside diphosphate kinase, NDPK 8021 (6) AAF08537(Pisum)
AELERTFIAIKPDGV* 100 () 7.8 16
Aspartyl-tRNA synthetase fragment(Asp-tRNA)
KGTLVPAKKENGT* 100 () 6.2 35
Thiosulfate sulfur transferase(Rhodanese) (TST)
XTQSAHTNEPVVS* 83/83 () 5.2 35
RNA-binding protein (RNP) 313 (2) Q40436(Arath)
GSAKLFVGG(I)SYXTD* 67/93 () 4.6 28
Translational elongation factor Tu(EF-Tu)
VDAVDDPELLELVEMELR 100 () 6.1 42
Proteomics 2002, 2, 880898 Plant mitochondria proteomics and development 885
Table 1. Continued
Name or description SpotNo.
Edman* or ESI-MS/MS MALDI pIobs
sequence % of identity/% of similarity
Homologs to proteins from Arabidopsis thaliana and Oryza sativa with unknown functions
Hypothetical 22.7 kDa proteinHyp 22.7K)
3004 (1) Q9LEV3(Arath)
() 5.9 17
F16N3 sequence hypothetical 28.1 kDaprotein
SEDVSHVPEIKDAE* 71/100 () 4.9 16
Unknown proteins 2310 (13) AAD24844(Arath)
(S)TKSKAIEHIVLFKV* 60/93 () 5.6 29
Hypothetical protein F24A 6.130(F24A6)
14 6.6 57
EST AU070430 2118 Q9SDD6(Oryza)
() 5.7 19
Unidentified proteins1304 (1) CTSAS(S)E(S)(SK)(K)(T)X(E)(E)(A)* 4.9 291732 (5) ESGVAESXA* 5.0 972307 (16),2308 (21)
3116 AKVATGTDILS(A)(A)SNF* 6.2 247324 DQIGDEIKPXN* 7.7 302 XTKVSNDPXTH* 5.7 124 (S)(S)(SP)PKISATISVGDK* 4.8 167 XXLYDKNVDDHS* 5.1 149 V(PLH)(SL)(TV)(AQ)(DE)(KE)(TY)(NQ)(AK)V(T)A* 4.1 1411 IETIAYEESRAHPDK(P)YT* 6.8 3314 (A)QSXKTXVLVTGXG* 5.4 3118 V(LH)(LS)(TP)(AP)(ED)(EK)
Ribulose 1, 5 bisphosphate carboxylase/oxygenase large subunit (RBCL)
81 34 6.8 52
Proteomic analysis of the soluble mitochondrial proteins. Pea leaf mitochondria were purified on Percoll gradients and thesoluble proteins were then extracted and separated by 2-DE as described in Sections 2.12.4. Proteins were identified byN-terminal Edman degradation, MALDI-TOF MS or ESI-MS/MS analyses followed by mining protein databases. Thesequences indicated with an asterisk were obtained by Edman degradation, the others by ESI-MS/MS. The percentageof identity was calculated between the amino acids present in the Edman or ESI-MS/MS tags and the sequence found inthe databases. For the calculation of the percentage of similarity, Awas considered similar to S, P, Tand G; I to L, M and V; Kto R and H; Y to F and W; D to E; and N to Q. In the case of MALDI-TOF MS analysis, coverage (cov.) corresponds to thepercentage of the full-length sequence covered by the matching peptides; (), MALDI-TOF MS analysis was done but theidentification of the protein by this method alone was not possible; this method was used to characterise the protein iso-forms; (), MALDI-TOF MS analysis not done. The spots corresponding to the identified proteins are also indicated in Figs.1 and 2. The spots were numbered using the PDQuest software (Bio-Rad) and with , and letters corresponding to the2-D gel image nomenclature presented in Fig. 2. Some of these spots (only labelled with Greek letters) are visible only in Fig.2. MM obs, molecular mass observed on the 2-D gel; pI obs, isoelectric point observed on the 2-D gel.Organism abbreviations : Arath, Arabidopsis thaliana; Cucurbita, Cucurbita sp.; Datisca, Datisca glomerata; Glycine, Gly-cine max; Linum, Linum usitatissimum; Medicago, Medicago sativa; Nicotiana, Nicotiana tabacum; Oryza, Oryza sativa;Pisum, Pisum sativum; Saccha, Saccharomyces cerevisiae; Solanum, Solanum tuberosum; Triticum, Triticum aestivum;Vigna, Vigna unguiculata; Zea, Zea mays.
886 J. Bardel et al. Proteomics 2002, 2, 880898
Figure 1. The soluble protein2-D gel map of pea leaf mito-chondria. Soluble proteins wereprepared from Percoll gradientpurified mitochondria and thenseparated by high-resolution2-DE. The proteins (500 g)were separated according totheir isoelectric point in the firstdimension and by SDS-PAGEin the second dimension. Thegel was calibrated for molec-ular mass (in kDa) and pI (in pHunits) by internal standards. Themajor spots were analysed byN-terminal Edman degradation,MALDI-TOF MS or ESI-MS/MS(see also Table 1). The imageused here is a representativegel stained with R-250 Coo-massie Brilliant Blue.
As it is difficult to get information from minor spots, wethought that chromatography on gel filtration might be anideal tool to enrich the minor proteins. The soluble proteinextract was fractionated by Sephacryl S-300 gel filtrationand analysed by SDS-PAGE (Fig. 2). Individual fractionswere pooled to build three subfractions (, and ) cor-responding respectively to high (above 250 kDa), medium(between 40 and 250 kDa) and low molecular mass pro-teins (less than 40 kDa). These subfractions were thenresolved by 2-D PAGE, electroblotted to PVDF mem-branes and stained with Coomassie blue. The results pre-sented in Fig. 2 clearly demonstrate the efficacy of gelfiltration for the enrichment of minor proteins. Besides,gel filtration provides additional information concerningquaternary structure of proteins. For instance, most ofthe polypeptides which appear in the blot are expectedto be components of oligomeric proteins. Using the , , subfractions, 36 additional spots shown in Fig. 2 weresubjected to protein sequencing, and resulting peptide
tags have been merged in Table 1. These peptides arementioned as 1 to 11, 1 to 8 or 1 to 21.
MALDI-TOF peptide mapping (see Section 2.7) was alsoconducted to analyse spots from Coomassie blue stainedgels. The MALDI-TOF procedure was generally used tocharacterise isoform enzymes or post-translational mod-ifications. ESI-MS/MS analyses were carried out whenthe proteins could not be identified by peptide massfingerprinting or Edman degradation. Several peptidesequence tags were obtained and eight further proteinswere unambiguously identified by the tandem mass spec-trometry technique (Table 1).
3.2 Identification of the major soluble proteinsof pea leaf mitochondria
The peptide sequence tags generated by Edman deg-radation or by MS/MS were analysed for similarity bymining nonredundant protein databases (SWISS-PROT,
Proteomics 2002, 2, 880898 Plant mitochondria proteomics and development 887
Figure 2. Soluble mitochondrialproteins separated by gel filtra-tion chromatography and thenanalysed by 1-D and 2-D PAGE.The soluble mitochondrial pro-teins were fractionated on aSephacryl S-300 column accord-ing to their molecular mass (seeSection 2.2) and 10 L of eachtube were analysed on 1-DSDS-PAGE stained with R-250Coomassie brilliant blue (upperpart of the figure). An aliquotof each tube (100 L of tubesnos. 1 to 23 called fraction ;100 L of tubes nos. 24 to 48called fraction and 100 L oftubes nos. 49 to 60 called frac-tion ) were pooled, concen-trated by ultrafiltration on micro-sep 3 K (Pall Filtron) and 500 gof these three fractions wereanalysed by 2-DE, blotted onProblottmembrane and stainedwith Coomassie blue (lower partof the figure). The major spotswere identified by Edman deg-radation and annotated on thethree different maps called , and , and also on Fig. 1 (seeTable 1).
TrEMBL, NCBI, TIGR) using FASTA and BLAST programs.The spots corresponding to the major soluble proteinsof pea leaf mitochondria were identified (Table 1) andare presented in a 2-D map (Figs. 1 and 2). This 2-D mapreveals that the major soluble proteins of pea leaf mito-chondria are the four components: GCS P, GCS T, GSC L(or DLDH, Fig. 1b), GCS H (Fig. 1c) of the glycine cleavagesystem (GCS) and the serine hydroxymethyltransferase(SHMT, Fig. 1b). These proteins were unambiguouslyidentified because they were cloned and characterised inpea (for a review see ). The GCS components togetherwith the SHMT were estimated, using the PDQuest soft-ware, to represent around 37% of the total soluble pro-teins, which confirms our previous observations .These components and especially the SHMT are spreadas different spots of similar molecular mass but that differin their pI values, indicating the presence of isoforms orpost-translational modifications (Fig. 1b and 2). In thecase of the H protein, the three spots identified in the gel (Fig. 2) correspond certainly to the normal H protein, theapoform (H protein without its lipoic acid cofactor) and theH protein loaded with methylamine, since these differentforms have been already characterised by MS . Themolecular chaperones, HSP60 and HSP70 (Fig. 1a), arevery abundant proteins in the mitochondrial space since
they represent around 3.4% and 1.8% of the total solubleproteins respectively. Like their chloroplastic counterparts, they are present in multiforms. The 2-D map in Fig. 1shows several other abundant spots corresponding tohousekeeping mitochondrial proteins involved in the tri-carboxylic acid cycle: malate dehydrogenase (MDH, Fig.1d) which represents 3.5% of the soluble proteins, succi-nyl-CoA ligase (SUC, Fig. 1a), citrate synthase (CISY,Fig. 1b), pyruvate dehydrogenase (PDH E1 and , Fig.1b; PDH E1, Fig. 1c; DLDH, Fig. 1b). We have also iden-tified a subunit of the F1-ATP synthase (ATP ) which isprobably released from the F0F1 coupling factor duringthe preparation of the soluble mitochondrial proteins(Fig. 1a and Table 1). The Mn-dependant superoxide dis-mutase (MnSOD, Fig. 1d), involved in the protectionagainst oxidative stress, is also well represented (around3%) in pea mitochondria by at least two isoforms. We alsodetected a putative ascorbate peroxidase (Table 1).
Remarkable by its abundance is a protein family that weidentified as aldehyde dehydrogenase (ALDH, Fig. 1b).Indeed, pea leaf mitochondria contain at least nine differ-ent spots with different pI values and molecular massvalues, representing around 7.5% of the soluble mito-chondrial proteins. The major forms of ALDH are likely
888 J. Bardel et al. Proteomics 2002, 2, 880898
oligomeric since they have been identified in the subfrac-tion , corresponding to high molecular mass proteins(3, 8 and 11; Fig. 2). From the MALDI peptide map-pings, we were able to identify several distinct ALDH pro-teins (Table 1). A first group comprises four isoforms orpost-translationally modified proteins (called ALDH 14in Fig. 1b; three of these proteins correspond to 3, 8,11 in Fig. 2) that have a very high degree of similaritywith the maize RF2 nuclear restorer protein which isrequired to restore male fertility in the Texas cytoplasmmaize . Very recently, Liu et al.  demonstrated thatRF2 is a mitochondrial aldehyde dehydrogenase. A sec-ond ALDH (called ALDH 5 in Fig. 1) which exhibits a muchhigher expression in etiolated than in green leaves, sharesa high homology with the ALDH cloned from rice (calledALDH2A by Nakazono et al., ) and tobacco (calledTobALDH2A by Op den Camp and Kuhlmeier ), andalso to a lesser extent with maize RF2 (see Table 1). Wealso identified another group composed of two spots(ALDH 6 and 7 in Fig. 1) which can be distinguished fromthe other ALDH on the basis of higher molecular massand lower pI. These ALDH are similar to the flax fis1 geneproduct which was induced during susceptible infectionby flax rust (Melampsora lini) . The role of this alde-hyde dehydrogenase in the infection process is not clear.The last group of ALDH includes two proteins which wereclassified as methylmalonate semialdehyde dehydro-genase (MMSDH 1 and 2), a mitochondrial CoA depen-dent aldehyde dehydrogenase may be involved in the cat-abolism of valine, -alanine and thymine [34, 35]. One ofthese spots (corresponding to MMSDH1) also containsanother ALDH (Table 1).
Among the other major spots, we identified a formatedehydrogenase (FDH) (Fig. 1b), a short-chain alcoholdehydrogenase-like protein (SCADH) (Fig. 1d), a thiosul-fate sulphur transferase (TST) (Fig. 1c) and a nucleosidediphosphate kinase (NDPK) (Fig. 1d). NDPK was easilyidentified because it has been recently purified andcloned from pea [36, 37]. This enzyme which could belocated in the mitochondrial intermembrane space plays a major role in regulating the nucleoside triphos-phate balance and could be also involved in cell signal-ling. We have also identified in the 2-D map a translationalelongation factor (EF-Tu) (Fig. 1b; also Table 1) which hasbeen recently cloned and characterised from maize .In addition to EF-Tu, an aspartyl tRNA synthetase frag-ment (Asp-tRNA) (Fig. 1d) and a RNA binding protein(RNP) (Fig. 1c) were also identified.
Using the three different approaches (Edman degra-dation, MALDI-TOF MS and ESI-MS/MS), seventy-threespots were analysed and 31 proteins were identified.Five proteins correspond to Arabidopsis thaliana ESTs orputative proteins with unknown function and 12 proteins
could not be identified from their N-terminal sequences(Table 1). Based on mass distribution, the sum of the pro-teins characterised by this proteomic approach repre-sents around 73% of the total soluble proteins of pealeaf mitochondria.
3.3 2-D map of soluble proteins of peamitochondria isolated from green andetiolated leaves, roots and seeds
The soluble mitochondrial proteome was also studied in adevelopmental perspective in order to analyse the impactof tissue differentiation on the mitochondrial proteome.Mitochondria were isolated and purified from differentpea organs: green and etiolated leaves, roots and seeds.The mitochondria isolation procedure is described in theSection 2.1 and is composed of differential centrifugationcombined with two steps of purification on appropriatePercoll gradients (depending of the nature of the tissue).Soluble protein extracts prepared from these purifiedmitochondria were then analysed by 2-D PAGE. The fourgels corresponding to the soluble mitochondrial proteinsprepared from green and etiolated leaves, roots andseeds are presented in Fig. 3. Using the PDQuest soft-ware, we have identified on the one hand, the spots ofthe mitochondrial soluble fraction present in all the organsand on the other hand, the proteins present specifically inthe studied organ. The major spots that appear in mito-chondria of all studied organs were ascribed to house-keeping proteins such as the tricarboxylic acid cycle pro-teins (MDH, SUC . . .), HSP60, HSP70, MnSOD etc. Themajor mitochondrial proteins specific to etiolated leaf,root or seed are indicated by arrows in Fig. 3, the greenleaf 2-D map being the reference gel.
3.4 Proteomic changes associated withmetabolic evolution
The most striking feature that arises from the comparisonof the 2-D gel map of soluble proteins from green leafmitochondria and from etiolated leaf mitochondria (Fig. 3)is the very low abundance of proteins of the glycine cleav-age system (GCS P, GCS T, GCS H and GCS L) and serinehydroxymethyltransferase (SHMT) when plants weregrown in darkness. This dramatic accumulation of glycinedecarboxylase proteins in mitochondria from mesophyllcells of C3 plant leaves is very likely attributed to a light-dependent transcriptional control of the genes encodingthese proteins [7, 39]. Another mitochondrial protein thatseems to be induced by light was identified as a short-chain alcohol dehydrogenase-like protein (SCADH). This
Proteomics 2002, 2, 880898 Plant mitochondria proteomics and development 889
Figure 3. Comparison of thesoluble mitochondrial proteinspatterns of green leaf, etiolatedleaf (EL), roots (R) and seeds (S)from pea. Soluble proteins wereprepared from Percoll gradientpurified mitochondria, sep-arated by high-resolution 2-DEand the gels were stained withCoomassie Brilliant Blue. Themajor spots, specifically presentin each organ, were analysed byN-terminal Edman degradation,MALDI-TOF MS or ESI-MS/MS(see also Table 2).
protein with those of the GCS and SHMT make up themajor proteins of the soluble fraction induced by light. Incontrast, some spots are more abundant in the mitochon-dria isolated from etiolated leaves than those isolated
from green leaves. Seven of these proteins were analysedby Edman degradation, MALDI-TOF MS or ESI-MS/MS(Table 2, Fig. 3). Three correspond to ATP synthase sub-units, one to the E1 pyruvate dehydrogenase subunit,
Table 2. Pea proteins identified from 2-D gel analyses of the soluble proteins of etiolated leaf, seed and root mitochondria(pI 4.0 to 8.0)
Name or description SpotNo.
Edman* or ESI-MS/MS MALDI pIobs
sequence % of identity/% of similarity
ATP synthase, subunit (ATP) EL1 Q41534(Triticum)
13 5.3 55
ATP synthase, subunit (ATP) EL2 Q96252(Pisum)
LTVNFVLPYSSXXAAK 88/88 () 4.9 23
890 J. Bardel et al. Proteomics 2002, 2, 880898
Table 2. Continued
Name or description SpotNo.
Edman* or ESI-MS/MS MALDI pIobs
sequence % of identity/% of similarity
Putative ATP synthase subunit (pATP) EL3 Q39852(Glycine)
() 5.7 27
Pyruvate dehydrogenase E1 subunit(PDH E1)
() 5.5 38
Hypothetical 19.6 kDa protein(Hyp 19.6K)
() 5.1 17
Unidentified protein EL9 VTGYTEQFVFYFKLFYTVDNPTK
() 7.6 29
Chaperonin 60 (CPN60) EL5 P08926(Pisum)
27 4.9 60
Formate dehydrogenase (FDH) R1R2R3
Glutamate dehydrogenase probable(GDH)
() 6.5 44
Pyruvate dehydrogenase E1 subunit(PDH E1)
() 6.5 43
Cysteine synthase (CYS) R6 P32260(Spinacea)
() 6.4 37
Cysteine synthase (CYS) R7 Q9SXS7(Arath)
() 6.1 39
Arginase R9 P46637(Arath)
() 5.9 36
Unidentified protein R10,R11
Thiosulfate sulfurtransferase(rhodanese) (TST)
() 5.2 38
Thiosulfate sulfurtransferase (TST) S2 Q9ZPK0(Datisca)
() 5.2 38
Maturation protein PPM32 (MP) or LateEmbryogenesis Abundant protein
() 5.0 37
Proteomics 2002, 2, 880898 Plant mitochondria proteomics and development 891
Table 2. Continued
Name or description SpotNo.
Edman* or ESI-MS/MS MALDI pIobs
sequence % of identity/% of similarity
Maturation protein PPM32 (MP) or LateEmbryogenesis Abundant protein
S5, S6, S7 Q9SPJ6(Glycine)
TAETTGEVAGAATEALK 65/82 () 5.0 35
Heat shock protein, 22 kDa (HSP22) S8 P46254(Pisum)
NTNAMRQY* 100 () 5.8 22
Heat shock protein, 22 kDa (HSP22) S9 P46254(Pisum)
LNMTDLLTDNPVLSAASR 94/94 () 5.5 22
Heat shock protein, 22 kDa (HSP22) S10 P46254(Pisum)
DDLLLSDVFDPFSPPR 100 () 6.1 22
Formate dehydrogenase (FDH) S11 Q9S7E4(Arath)
() 6.8 47
Dormancy related protein orSeed maturation protein (SMP)
() 6.9 35
T26B15.8 protein (T26B) S19 O80889(Arath)
() 5.6 28
RNA-binding protein (RNP) S22 Q9FNR1(Arath)
() 4.8 31
Contamination by protein bodies
Convicillin precursor (CVCL) S20 P13915(Pisum)
NYDEGS* 100 () 5.55.8
Vicilin precursor (VCLC) S21 P13918(Pisum)
Provicilin precursor (type B) (VCLB) S24 P02854(Pisum)
LEAAFNTNYEEAEK 93/93 () 6.5 22
Provicilin precursor (type B) (VCLB) S25 P02854(Pisum)
NEQLQDLDLFVNSVDLK 82/94 () 7.4 36
Vicilin (VCL) S15S17 P02856(Pisum)
QLQDLDLFVNWSVLK 80/80 () 6.86.9
Albumin 2 (PA2) (Alb) S27 P08688(Pisum)
() 5.5 29
Albumin 2 (PA2) (Alb) S28 P08688(Pisum)
FFPFFEGTVFENGNNAAYR 83/86 () 5.3 29
The major spots corresponding to soluble mitochondrial proteins specifically present in etiolated leaves (EL), seeds (S) androots (R) were identified by Edman degradation, MALDI-TOF MS or ESI-MS/MS followed by mining protein databases. Thesequences indicated with an asterisk were obtained by Edman degradation, the others by ESI-MS/MS. The percentage ofidentity was calculated between the amino acids present in the Edman or ESI-MS/MS tags and the sequence found in thedatabases. For the calculation of the percentage of similarity, A was considered similar to S, P, Tand G; I to L, M and V; K toR and H; Y to F and W; D to E; and N to Q. In the case of MALDI-TOF MS analysis, coverage (cov.) corresponds to thepercentage of the full-length sequence covered by the matching peptides; (), MALDI-TOF MS analysis was done but theidentification of the protein by this method alone was not possible; this method was used to characterise the protein iso-forms; (), MALDI-TOFMS analysis not done. The spots corresponding to the identified proteins are also indicated in Fig. 3.MM obs, molecular mass observed on the 2-D gel; pI obs, isoelectric point observed on the 2-D gel.Organism abbreviations : Arath, Arabidopsis thaliana; Datisca, Datisca glomerata; Glycine, Glycine max; Pisum, Pisumsativum; Spinacea, Spinacea oleracea; Solanum, Solanum tuberosum; Triticum, Triticum aestivum.
892 J. Bardel et al. Proteomics 2002, 2, 880898
one is homologue to an hypothetical protein of A. thaliana,one is unidentified and another corresponds to a plastidialprotein contamination, the chaperonin 60 (CPN 60).
Figure 3 also highlights the proteins specifically ex-pressed in mitochondria isolated from roots. These pro-teins were analysed and identified by MS and N-terminalsequencing. The different sequencing data are presentedin Table 2. One of the major proteins induced specificallyin roots corresponds to formate dehydrogenase (FDH).The enzyme seems to be present in the mitochondria inmultiforms (isoforms or post-translationally modified)since four spots corresponding to this enzyme werefound on the root 2-D map, two also being present ingreen leaves (see Fig. 1). Cysteine synthase (CYS, 2spots), arginase, a putative glutamate dehydrogenase(GDH) and the E1 pyruvate dehydrogenase (PDH E1)subunit are other enzymes which seem to be induced oroverexpressed in root mitochondria.
When the 2-D map of the soluble proteins isolated frompea seeds was compared to the other maps and in par-ticular with the green leaf 2-D map, it clearly appearedthat a high number of spots were specifically present inseed mitochondria. Twenty-eight different spots were an-alysed by MS and N-terminal sequencing and the peptidesequence tags generated by these analysed were addedto Table 2. Unfortunately, some spots were attributed toseed reserve proteins (convicilin, vicilin, provicilin, albu-min), thus revealing contamination of the mitochondrialfraction by protein bodies. Among the other spots, wefound three abundant proteins that were unambiguouslymitochondrial, and whose abundance suggests an impor-tant physiological function in seed mitochondria. Weidentified two spots (S1 and S2, Fig. 3) as a thiosulfatesulphur transferase (rhodanese), an enzyme that wasrecently characterized in plants and which is expectedto carry several potential functions such as the formationof iron-sulphur centres, cyanide detoxification and assim-ilation of sulphur .
Another abundant protein was identified as formate de-hydrogenase, which was also abundant in root mito-chondria. An unexpected strongly expressed mitochon-drial protein that we found in seed is HSP22, which wasfound as several spots of similar molecular weight thatsuggest post-translational modifications. MitochondrialHSP22 is a heat shock protein [43, 44] that has also beenshown to be induced by oxidative stress . Besidesthese genuine mitochondrial proteins, we identified sev-eral proteins that have not been assigned previously tothe organelle (Table 2). Since we found evidence of con-tamination by protein bodies, we questioned the potentialcellular localisation of the proteins that emerged with thehighest scores from database homology searches. Four
different programs (MitoProt, PSORT, Predotar, TargetP,see Section 2.9) were systematically used for these anal-yses.
One of the major proteins (S3 and S4), appearing in thegel as several spots with an apparent molecular massof 36 kDa (Fig. 3) but also with a lower molecular mass(S5, S6, S7), gave its highest database hit with a seedmaturation protein PPM32 (Acc. Q9SPJ6) from Glycinemax that was annotated as an LEA III protein . SincePPM32 (18.8 kDa) was only half the molecular mass ofS3 (S4) protein, we performed a BLAST search on Arab-idopsis sequences using its accession as a query. Weretrieved two high scoring sequences (Acc. At5g44310,At4g21020) respectively annotated as LEA-like protein(37 kDa) and desiccation related protein (29.4 kDa). Thethree accessions (Q9SPJ6, At5g44310, At4g21020)derived from data mining with S3 sequence tags exhibitan overwhelming hydrophilicity and features characteris-tic of LEA protein of group III, such as repeatedsequences and enrichment in small (alanine, glycine) andcharged (lysine, glutamate, aspartate) residues that to-gether accounted for more than 50% of the molar frac-tion. The three accessions were submitted to target pep-tide predictions with the four programs described above.All three proteins were assigned a medium to high prob-ability of targeting to mitochondria, generally with threeout of four programs (results not shown). It is thereforereasonable to hypothesize that S3 (S4) is an LEA proteinlocalized within mitochondria.
The S22 protein (Fig. 3) was clearly identified as a glycine-rich RNA binding protein (RNP), the closest accessionsbeing two Arabidopsis sequences (Q9C909, Q9FNR1).The subcellular predictions performed on these proteinsindicated a strong probability of mitochondrial targetingwith 3/4 prediction programs (results not shown). In addi-tion, the N-terminal putative transit peptide sequencewas excluded from the eukaryotic RNA recognitiondomain (PROSITE PS50102) and the glycine-rich region(PROSITE PS50315) characteristic of glycine-rich RNAbinding proteins. It is therefore highly probable that S22is a mitochondrial member of this protein family. More-over, this protein has also been identified in pea leaf mito-chondria (Fig. 1). The S18 protein tags (Fig. 3, Table 2)were found similar to two almost identical Arabidopsisproteins (Acc. Q9CLL3, Q9FZ42) and to a soybean pro-tein (Acc. Q9LLQ6). These putative proteins are relatedto dormancy, seed maturation and are possible membersof the short-chain alcohol dehydrogenase family. How-ever, the mitochondrial localisation of these proteins israther doubtful since none of the programs predicted atargeting to mitochondria, suggesting that S18 is a prob-able contaminant. The peptide tags of S19 (Fig. 3, Table 2)
Proteomics 2002, 2, 880898 Plant mitochondria proteomics and development 893
gave a significant match with an Arabidopsis protein(Acc. O80889) that bears a thiolase signature (PROSITEPDOC00092) that suggests a possible role of S19 in fattyacid oxidation. While fatty acid oxidation is known tooccur in animal mitochondria and peroxisomes, in plants,the pathway takes place mainly in peroxisomes. However,a complete oxidation pathway has been discovered inhigher plant mitochondria  and both organelles con-tribute to fatty acid oxidation in pea . The predictionof the subcellular localization of O80889 supported a per-oxisomal or cytosolic localization, but without excluding amitochondrial targeting (TargetP, PSORT). In the light ofthe available data, it seems too early to make any conclu-sions about the mitochondrial localisation of S19 protein.
4.1 Main features of plant mitochondrialproteomes
Taking into consideration the central role of mitochondriain bioenergetics and metabolism of plant cells, we haveundertaken a detailed proteomic analysis of this orga-nelle. Since mitochondria are rich in membrane-asso-ciated proteins and solubility of membrane proteins couldbe a critical issue [20, 49], we decided to initiate the pro-teomic work by the characterization of soluble mitochon-drial proteins. The strategy which consists of separatingthe analysis of the soluble proteome from that of themembrane proteome has the advantage of determiningthe suborganelle localization of the proteins. Soluble pro-teins were prepared from pea leaf mitochondria purifiedon a Percoll gradient and submitted to high resolution2-DE. Among the 433 spots detected with Coomassieblue, 78 spots, which correspond to around 73% (inmass) of detectable proteins, were analysed either byN-terminal Edman sequencing, MALDI-TOF MS or ESI-MS/MS. The combination of gel filtration chromatographywith classical 2-DE significantly improved the resolutionof the proteome (Fig. 2). Besides, such a three-dimen-sional analysis provides valuable information about theoligomeric status of proteins, the composition of multi-enzymatic complexes and protein-protein interactions.
Throughout this work, we observed that only 8% of thespots submitted to N-terminus Edman degradation wereblocked. In contrast, it is estimated that 50% of the proteinin nature are blocked by post-translational modification ofN-terminal residues (acetylation, formylation) or cyclisationof an N-terminal glutamine residue to pyroglutamic acid. The low rate of blocking we observed is probably dueto the fact that themajority of the solublemitochondrial pro-teinsare nuclearencodedandprocessedupon import, thus
removing the potential N-terminal residue modification. Aninvaluable advantage of the results obtained via Edmandegradation of intact polypeptides lies in the determinationof the actual N-terminus of the mature proteins. In plantmitochondria, more than 95% of proteins are nuclearencoded and targeted to the organelle through transloca-tion machinery that recognizes a targeted N-terminal pep-tide. Since in most cases, this presequence is removedupon import by endoproteases, the N-terminus determina-tion of the mature protein provides essential informationabout the cleavagesite and thepresequenceprimary struc-ture when the nucleotide sequence of the precursor isknown. The improvement of targeting prediction programswill facilitate the annotation of genomic data that open thepath to functional genomics. Because of the diversity inmitochondrial presequences and cleavage sites, the pre-diction programs remain knowledge based, and in spite ofthe development of better algorithms and neural networks,a major obstacle toward their improvement lies in the scar-city of actual data aboutmature proteins . Theworkpre-sentedhere hasallowed thedeterminationof42differentN-terminus sequences of pea mitochondrial proteins, whichwill be helpful to prediction programs, and also crucialwhen, for example, overexpression of a recombinant pro-tein is considered.
Among the mitochondrial proteins analysed, a large pro-portion seems to be post-translationally modified. Animportant issue will be to determine the nature of themodifications and the functional significance of theirabundance in the soluble compartment of plant mito-chondria. With regard to this result, it is interesting tonote that almost half of the proteins identified in the analy-sis of Medicago truncatula root proteome were found asisoforms . Such high levels of post-translational mod-ifications will certainly add an additional dimension to thecharacterization of plant proteomes.
Besides the in-depth exploration of the soluble proteomeof pea leaf mitochondria, we have done a comparativeprofiling of the mitochondrial soluble proteomes isolatedfrom different tissues or organs: green leaves, etiolatedleaves, roots and seeds. The comparative study has ledto the analysis of 45 other spots either by N-terminalEdman sequencing, MALDI-TOF MS or ESI-MS/MS.From image analysis of the 2-D gels, it is obvious that theproteomic profiles share a high number of common pro-teins that logically correspond to housekeeping and basicfunctions of mitochondria (Krebs cycle, protein folding)(see Table 1) that do not request further discussion.Rather we will comment on the prominence of aldehydedehydrogenases present in all tissues before switchingto the prime modifications of proteomic profiles thatmust be related to specific physiological functions.
894 J. Bardel et al. Proteomics 2002, 2, 880898
4.2 Prominence of aldehyde dehydrogenase
A major result inferred from this study is that aldehydedehydrogenase (ALDH) is one of the major mitochondrialproteins strongly expressed in plant mitochondria with thehighest level in leaves and roots where at least nine ALDHisoforms represent a large proportion of the soluble pro-teins ( 7.5%), highlighting the importance and the com-plexity of the aldehyde dehydrogenase family in the plantmitochondrial matrix space. Aldehydes, which can begenerated from a high number of reactions in the cell, arehighly reactive molecules which interact with cellularnucleophiles such as proteins and nucleic acids. Amongthe most effective pathway for aldehyde detoxificationis the irreversible oxidation to their corresponding car-boxylic acids by ALDHs. A vast literature exists about thehuman acetaldehyde dehydrogenase which functions inthe detoxification pathway of dietary ethanol. The humanaldehyde dehydrogenase gene family contains at leasttwelve different members some of which have quitespecific, and others broad, substrate preferences .Class 1 and 3 ALDHs contain both constitutively and indu-cible cytosolic forms, whereas class 2 consists of consti-tutive mitochondrial enzymes . In yeast, mitochondrialALDH are apparently encoded by two genes, ALD4 andALD5, whose products differ mainly by their NAD andNADP cofactor preferences [54 and references therein].In plants, apart from betaine ALDH which is involved inthe biosynthesis of betaine implicated in salinity anddrought adaptation , aldehyde dehydrogenases havereceived little attentionuntil the last few years. ALDHmightbe involved in the detoxification of reactive aldehydesaccumulated within mitochondria (under stress condi-tions). For example, it has been shown recently that 4-hydroxy-2-nonenal, an aldehydeproduct of lipidperoxida-tion is a powerful inhibitor of the mitochondrial decarbox-ylating enzymes . Such a compound must berapidly eliminated to avoid dysfunctioning of cell respira-tion but also of photorespiration in concerned tissues.
AlthoughALDHwere postulated to play a role in the detox-ification of occasional aldehydes , their high amountin plant mitochondria argue in favour of a major role fortheses enzymes. Mitochondrial ALDHs could neutralizethe harmful effect of aldehydes produced by a misfunc-tioning of some enzymes of the tricarboxylic acid cycle(such as pyruvate dehydrogenase, PDH; -ketoglutaratedehydrogenase, KGDH), the glycine cleavage system(GCS) or the serine hydroxymethyltransferase (SHMT).Suchenzymes couldproduce, in the absence of free coen-zyme A or tetrahydrofolate (H4F), aldehydes such as acet-aldehyde (PDH), succinaldehyde (KGDH) or formaldehyde(GCS, SHMT). This idea is supported by the fact that theGCS, in absence of H4F, produces formaldehyde as a con-
sequence of the rapid destabilization of the methylamine-loaded lipoamidearmofHprotein [12, 59].Accordingly,wehave therefore detected spectrophotometrically the pre-sence of a formaldehyde dehydrogenase activity in thematrix of pea leaf mitochondria (which does not requirethe presence of glutathione; result not shown) and whichcould correspond to one of the ALDHs identified in thiswork. This formaldehyde dehydrogenase seems to be dif-ferent from the glutathione dependent formaldehydedehydrogenase also known as class 3 alcohol dehydro-genase . The formate produced could thenbe oxidizedby the mitochondrial formate dehydrogenase that weidentified in all the tissues.
ALDH might be also involved in primary metabolic path-ways (i.e. ethanolic fermentation). In Populus deltoides, ithas been shown that xylem ethanol was rapidly incor-porated into TCA intermediates via acetaldehyde andacetate . Tadege and Kuhlmeier  have shownthat, during tobacco pollen development, pyruvate decar-boxylase and alcohol dehydrogenase (ADH) are stronglyinduced. Op den Camp and Kuhlmeier  suggest thatALDH, also highly expressed in reproductive tissues,could play a role in detoxifying the cytosolic acetaldehydeproduced by an aerobic ethanolic fermentation whichoccurs during pollen development. They also proposedthat in tobacco pollen, cytosolic ALDH could function ina pyruvate dehydrogenase bypass similar to the bypassthat was proposed in the yeast Saccharomyces cerevi-siae . More recently, Boubekeur et al.  suggestedthat in yeast, cytosolic acetaldehyde resulting from pyru-vate decarboxylase activity could be transported and oxi-dized by ALDH in mitochondria, thus generating a mito-chondrial pyruvate dehydrogenase bypass. However,since acetaldehyde is an extremely reactive product, it isdifficult to believe that it should cross the mitochondrialmembranes without provoking damage. Indeed, we didnot measure any oxidation of acetaldehyde by intact plantmitochondria (results not shown), although the com-pound was readily oxidized by a soluble matrix extract( 30 nmol/min/mg of protein) with a very low apparentKm for acetaldehyde (510 M). Therefore, we suggestthat mitochondrial ALDH is involved in the detoxificationof aldehydes produced inside the mitochondria. A pos-sible source of mitochondrial acetaldehyde could beethanol, produced inside the cell or transported fromanother tissue (roots for example). Ethanol could beused as a powerful respiratory substrate with the helpof a mitochondrial alcohol dehydrogenase and aldehydedehydrogenase. Such a pathway would confer additionalflexibility to respiratory carbon fluxes in plants by allow-ing ethanol, a highly reduced byproduct of fermentation,to be efficiently oxidized in mitochondria (two moleculesof NADH produced by one molecule of ethanol). Alterna-
Proteomics 2002, 2, 880898 Plant mitochondria proteomics and development 895
tively, one can postulate that in conditions where pyru-vate decarboxylase is active, a potential ethanol bypasswould sustain a physiological carbon flux in which eth-anol could not be considered as a final product butrather as an intermediary metabolite. In support of theexistence of such a pathway, it should be pointed outthat an ADH has been identified in yeast and humanmitochondria [17, 65], and that we have detected ashort-chain ADH (SCADH, Fig. 1, Table 1) present inlarge amounts in pea leaf mitochondria and probablyinduced in light-grown tissue. Interestingly, when pyru-vate decarboxylase was overexpressed in tobacco acetaldehyde and ethanol accumulated only upon inhibi-tion of respiration (anoxia, inhibitors) which indicates thatthey could be actively metabolized in normal conditionsand in particular through the respiratory ethanol bypasswe propose. This is sustained by the fact that exogenousethanol provided to tobacco leaves under normoxia con-ditions is efficiently metabolized . This pathway couldhave a great significance in terms of cytoplasmic malesterility since a mitochondrial ALDH was identified asthe nuclear restorer factor (RF2) of T-cytoplasm maize.Indeed, the ethanol bypass might be essential to in-crease the capacity of respiratory carbon flux thatfuels the high energy demand of germinating pollen. Theoperation of the bypass can proceed independently ofthe TCA cycle and the pyruvate dehydrogenase, andthus generate acetate required for fatty acid and lipidbiosynthesis.
Finally, since the ALDH family encompasses enzymeswith various substrate specificity, we have undertakena biochemical approach to identify and study in moredetail the different ALDH activities present in plant mito-chondria. Among the group of spots we identified, an-other candidate could correspond to the succinic semi-aldehyde dehydrogenase that has been assigned a mito-chondrial localization and which has a similar molecularmass (453 kDa) .
4.3 Tissue specific variations
The analysis of green and etiolated leaf mitochondriasoluble proteomes reveals several differences, in agree-ment with the previous study that compared 2-D maps ofmitochondria isolated from green and etiolated pealeaves . The most remarkable observation concernsthe four proteins of GCS and SHMT which are known tobe highly induced in light-grown tissue because of theirengagement in the photorespiratory pathway . Never-theless, these proteins are still present in nonphotosyn-thetic cells which is in agreement with the work of Mouil-lon et al. . Indeed, serine and glycine catabolism aredirectly connected to C1 metabolism. Serine was shown
to be the main donor of C1 units through the cytosolicSHMT reaction and the glycine formed during this pro-cess is rapidly oxidized by the mitochondrial GCS-SHMTenzymatic system, which is therefore required in all planttissues. We found some polypeptides preferentially ex-pressed in etiolated leaf mitochondria: ATPase subunits,ALDH2A, PDH subunits, cysteine synthase and severalunidentified proteins. However, the interpretation of thesedifferential expressions will require further work on eachsystem. The root mitochondrial soluble proteome is char-acterized by the presence of several distinct proteins notfound (or present in low amount) in the leaf counterpart.A major protein is a formate dehydrogenase (FDH), whichis present as several isoforms or post-translationallymodified proteins. Formate dehydrogenase is known tobe highly expressed in nongreen tissues and induced byenvironmental stress in leaves . We demonstratedfor the first time that FDH was also a major protein of seedmitochondria, which were indeed able to oxidise formateat a low but significant rate (12 nmol of O2 per min andper mg protein). In contrast, FDH isoforms were barelydetectable in green and etiolated leaf mitochondria, indi-cating that their expression in pea is not linked to dark-ness but rather to a tissue-specific program. A detailedmolecular and biochemical work will be required toanalyse the different isoforms and the regulation of themechanism of post-translated modifications. It shouldbe pointed out that the metabolism of formate, in partic-ular its origin, in higher plants has not been clearly eluci-dated, although it is closely related to C1 carbon andserine metabolism . Nevertheless, it seems that apotential source of formate could derive from form-aldehyde dehydrogenase activity. Another interesting fea-ture of root mitochondria is the occurrence of severalenzymes involved in amino acid metabolism (cysteinesynthase, arginase, glutamate dehydrogenase). It is thuslikely that root mitochondria are actively engaged inamino acid metabolism.
The seed mitochondria soluble proteome appeareddramatically different from the other mitochondrial pro-teome arrayed in this study, although the mitochondrialhousekeeping enzymes were readily detected. We identi-fied one of the major protein as rhodanese, a thiosulfatesulfur transferase (TST). The existence of rhodanese inplants has been highly controversial and only demon-strated recently in Arabidopsis after EST mining and sub-sequent characterization of cDNAs and recombinant pro-teins . Although the exact function of thiosulfatesulfur transferase remains unclear, it has been postulatedto play a role in several mechanisms such as the forma-tion of iron-sulfur centers, sulfur assimilation and cyanidedetoxification . TST abundance in pea seed mito-chondria is rather intriguing and will require in-depth
896 J. Bardel et al. Proteomics 2002, 2, 880898
investigation to elucidate its actual function in relationto germination. It is interesting to note the existence ofseveral spots of TST that differ by their pIs, which couldbe an indication of post-translational modification likephosphorylation, which have been described in the caseof rhodanese .
One of the major proteins (S3), appearing in the gelas several spots with an apparent molecular mass of36 kDa, was tentatively identified as an LEA (late embryo-genesis abundant) protein. Further cloning work is in pro-gress to establish the primary structure and confirm themitochondrial localization of the protein. Interestingly,two proteins that accumulated in maize seedlings mito-chondria in response to cold stress, were detected byantibodies against dehydrins, which are LEA proteins ofgroup II . LEA proteins are hydrophilic proteins thataccumulate to high level during seed maturation or waterdeficit in vegetative tissues of plants. They have beengrouped in at least six distinct families on the basis ofsequence similarity [81, 82]. Because of their extremehydrophilic nature, LEA proteins have been proposed toplay a role in the preservation of cell components (mem-branes, proteins, cytosol) during dehydration. Recently, agroup III LEA protein from the pollen of Typha latifoliawas shown in vitro to act synergistically with sucrose inthe formation of the glassy state that confers long-termstability to the cytosol in the dry state . If confirmed,protein S3 would be the first member of the LEA proteinfamily to be characterized in plant mitochondria, and itsrole is likely to be related to the desiccation tolerancethat is a major trait of most seeds. Our attention wasdrawn by the relative abundance of HSP-22 in seed mito-chondrial proteome. HSP-22, which was discovered as aheat shock protein in pea leaf mitochondria  has alsobeen reported in tomato to appear under oxidative stressconditions . It is thus tempting to hypothesize thatHSP 22 expression is under a genetic control linked toseed development, as it has been demonstrated for sev-eral cytosolic HSPs . However, since the protein wasshown to be remarkably stable after its rapid expression, it should be established whether its presence in seedmitochondria is not simply a consequence of the hightemperatures that a seed is likely to endure on the motherplant during seed development and maturation.
5 Concluding remarks
As a whole, the proteomic study described in this workproved to be very informative since it brought to light aglobal view of the soluble protein composition of plantmitochondria in different tissues. The body of data pro-vides new insights into the functions, physiological roles
and biogenesis of the mitochondria and it is now obviousthat organ and tissue differentiation are intimately linkedwith functional differentiation of mitochondria, which canbe visualized through their proteome complement. How-ever, our results pinpointed some of the difficulties inher-ent to proteomic technology such as the analysis of minorproteins, which can at least be partially overcome by theuse of a third dimension such as the gel filtration step wehave used. Another drawback of subcellular proteomeanalysis is that contamination by other fractions is almostunavoidable. It is thus essential to confirm the genuinelocalization of new or original proteins by in silico predic-tions, and ultimately by in vivo transport studies per-formed with reporter fusion protein constructs.
In order to increase knowledge about the original plantmitochondrial functions, we are performing biochemicaland molecular studies of the main mitochondrial pro-teins highlighted during this work. In addition, a similarapproach will be engaged to study membrane proteinsand we are considering a more thorough exploration ofthe mitochondrial proteome to identify minor proteins.
Note added in proof
During the processing of this manuscript, two reportswere published which describe the analysis of the mito-chondrial proteome from heterotrophically cultured cellsof Arabidopsis thaliana (Kruft et al., Plant Physiol. 2001,127, 16941710; Millar et al. Plant Physiol. 2001, 127,17111727). While the majority of mitochondrial house-keeping enzymes were logically found in all studies, eachwork brought to light original proteins, thus demonstrat-ing the potential of the proteomic approach to deciphermitochondrial functions.
Thisworkwas supportedby the RgionRhne-Alpes, bythe Centre National de Recherche Scientifique (Phys-ique et Chimie du Vivant, PCV 98122), by Gnoplanteprogramme nouveaux outils (NO 19993663) and by theCommissariat lEnergie Atomique. We thank Dr. Jac-ques Joyard for useful discussions and for his support. Wethank also Dr Hughes Dunlop for critical reading of themanuscript.
Received November 16, 2001
 Green, D. R., Reed, J. C., Science 1998, 281, 13091312. Gray, M. W., Burger, G., Lang, B. F., Science 1999, 283, 1476
1481. Saraste, M., Science 1999, 283, 14881493. Wallace, D. C., Science 1999, 283, 14821488.
Proteomics 2002, 2, 880898 Plant mitochondria proteomics and development 897
 Moller, I. M., Rasmusson, A. G., Trends Plant Sci. 1998, 3,2127.
 Vanlerberghe, G. C., McIntosh, L., Annu. Rev. Plant Physiol.Plant Mol. Biol. 1997, 48, 703734.
 Douce, R., Bourguignon, J., Macherel, D., Neuburger, M.,Biochem. Soc. Trans. 1994, 22, 184188.
 Vauclare, P., Diallo, N., Bourguignon, J.,Macherel, D., Douce,R., Plant Physiol. 1996, 112, 15231530.
 Baldet, P., Alban, C., Douce, R., FEBS Lett. 1997, 419, 206210.
 Rbeill, F., Macherel, D., Mouillon, J. M., Douce, R., EMBOJ. 1997, 16, 947957.
 Wada, H., Shintani, D., Ohlrogge, J., Proc. Natl. Acad. Sci.USA 1997, 94, 15911596.
 Gueguen, V., Macherel, D., Jaquinod, M., Douce, R., Bour-guignon, J., J. Biol. Chem. 2000, 275, 50165025.
 Siendones, E., Gonzlez-Reyes, J. A., Santos-Ocaa, C.,Navas, P., Crdoba, F., Plant Physiol. 1999, 120, 907912.
 Abbott, A., Nature 1999, 402, 715720. Cellis, J. E., Gromov, P.,Curr. Opin. Biotech. 1999,10, 1621. Rossignol, M., Curr. Opin. Biotechnol. 2001, 12, 131134. Rabilloud, T., Kieffer, S., Procaccio, V., Louwagie, M. et al.,
Electrophoresis 1998, 19, 10061014. Garin, J., Diez, R., Kieffer, S., Dermine, J.-F. et al., J. Cell
Biol. 2001, 152, 165180. Peltier, J.-B., Friso, G., Kalume, D. E., Roepstorff, P. et al.,
Plant Cell 2000, 12, 319341. Seigneurin-Berny, D., Rolland, N., Garin, J., Joyard, J., Plant
J. 1999, 19, 217228. Douce, R., Bourguignon, J., Brouquisse, R., Neuburger, M.,
Methods Enzymol. 1987, 148, 403415. Rabilloud, T., Adessi, C., Giraudel, A., Lunardi, J., Electro-
phoresis 1997, 18, 307316. Rabilloud, T., Valette, C., Lawrence, J. J., Electrophoresis
1994, 15, 15521558. Claros, M. G., Vincens, P., Eur. J. Biochem. 1996, 241, 779
786. Nakai, K., Kanehisa, M., Genomics 1992, 14, 897911. Emanuelsson, O., Nielsen, H., Brunak, S., Gunnar von
Heijne, G., J. Mol. Biol. 2000, 300, 10051016. Douce, R., Bourguignon, J., Neuburger, M., Rbeill, F.,
Trends Plant Sci. 2001, 6, 167176. Oliver, D. J., Neuburger, M., Bourguignon, J., Douce, R.,
Plant Physiol. 1990, 94, 833839. Cui, X., Wise, R. P., Schnable, P. S., Science, 1996, 272,
13341336. Liu, F., Cui, X., Horner, H. T., Weiner, H., Schnable, P. S.,
Plant Cell 2001, 13, 10631078. Nakazono, M., Tsuji, H., Li, Y., Saisho, D. et al., Plant Physiol.
2000, 124, 587598. op den Camp, R. G., Kuhlemeier, C., Plant Mol. Biol. 1997,
35, 355365. Roberts, J. K., Pryor A., Plant J. 1995, 8, 18. Goodwin, G. W., Rougraff, P. M., Davis, E. J., Harris, R. A.,
J. Biol. Chem. 1989, 264, 1496514971. Yoshida, A., Rzhetsky, A., Hsu, L. C., Chang, C., Eur. J. Bio-
chem. 1998, 251, 549557. Struglics, A., Hakansson, G., Eur. J. Biochem. 1999, 262,
765773. Escobar Galvis, M. L., Hakansson, G., Alexciev, K., Knorpp,
C., Biochimie 1999, 81, 10891096. Choi, K. R., Roh, K. S., Kim, J. K., Sim, W. S., Gene 2000,
 Oliver, D. J., Annu. Rev. Plant Physiol. Plant Mol. Biol. 1994,45, 323338.
 Hatzfeld, Y., Saito, K., FEBS Lett. 2000, 470, 147150. Papenbrock, J., Schmidt, A., Eur. J. Biochem. 2000, 267,
145154. Papenbrock, J., Schmidt, A., Eur. J. Biochem. 2000, 267,
55715579. Lenne, C., Douce, R., Plant Physiol. 1994, 105, 12551261. Lenne, C., Block, M., Douce, R., Biochem. J. 1995, 311,
805813. Banzet, N., Richaud, C., Deveaux, Y., Kazmaier, M. et al.,
Plant J. 1998, 13, 519527. Chow, T. Y., Liu, S. M., Lin, T. Y., Hsing, Y. I. C., Plant Physiol.
1999, 121, 1054. Dieuade, M., Couee, I., Pradet, A., Raymond, P., Biochem. J.
1993, 296, 199207. Masterson, C., Wood, C., Biochem. Soc. Trans. 2000, 28,
757760. Adessi, C., Miege, C., Albrieux, C., Rabilloud, T., Electro-
phoresis 1997, 18, 127135. Lottspeich, F., in: Rabilloud, T. (Ed.), Proteome Research:
Two-dimensional Analysis and Identification Methods,Springer Verlag, Berlin, Heidelberg, Germany 2000,pp. 181196.
 Emanuelson, O., Nielsen, H., Brunak, S., von Heijne, G.,J. Mol. Biol. 2000, 300, 10051016.
 Mathesius, U., Keijzers, G., Natera, S. H. A., Weiman, J. J.et al., Proteomics 2001, 1, 14241440.
 Lindahl, R., Crit. Rev. Biochem. Mol. Biol. 1992, 27, 283335.
 Navarro-Avino, J. P., Prasad, R., Miralles, V. J., Benito, R. M.,Serrano, R., Yeast 1999, 15, 829842.
 Weretilnyk, E. A., Hanson, A. D., Proc. Natl. Acad. Sci. USA1990, 87, 27452749.
 Humphries, K. M., Yoo, Y., Szweda, L. I., Biochemistry 1998,37, 552557.
 Humphries, K. M., Szweda, L. I., Biochemistry 1998, 37,1583515841.
 Millar, A. H., Leaver, C. J., FEBS Lett. 2000, 481, 117112. Guilhaudis, L., Simorre, J. P., Blackledge, M., Marion, D.
et al., Biochemistry 2000, 39, 42594266. Shafqat, J., El-Ahmad, M., Danielsson, O., Martinez, M. C.
et al., Proc. Natl. Acad. Sci. USA 1996, 93, 55955599. MacDonald, R. C., Kimmerer, T. W., Plant Physiol. 1993, 102,
173179. Tadege, M., Kuhlemeier, C., Plant Mol. Biol. 1997, 35, 343
354. Pronk, J. T., Wenzel, T. J., Luttik, M. A., Klaassen, C. C. et al.,
Microbiology 1994, 140, 601610. Boubekeur, S., Bunoust, O., Camougrand, N., Castroviejo,
M. et al., J. Biol. Chem. 1999, 274, 2104421048. Lutstorf, U., Megnet, R., Arch. Biochem. Biophys. 1968,
126, 933944. Bucher, M., Brandle, R., Kuhlemeier, C., EMBO J. 1994, 13,
27552763. Busch, K. B., Fromm, H., Plant Physiol. 1999, 121, 589597. Remy, R., Ambard-Bretteville, F., Colas des Francs, C., Elec-
trophoresis 1987, 8, 528532. Bourguignon, J., Rbeill, F., Douce, R., in: Singh, B. K.
(Ed.), Plant Amino Acids, Marcel Dekker, New York 1999,pp. 111146.
 Mouillon, J. M., Aubert, S., Bourguignon, J., Gout, E. et al.,Plant J. 1999, 20, 197205.
898 J. Bardel et al. Proteomics 2002, 2, 880898
 Colas des Francs-Small, C., Ambard-Bretteville, F., Small,I. D., Remy, R., Plant Physiol. 1993, 102, 11711177.
 Hourton-Cabassa, C., Ambard-Bretteville, F., Moreau, F.,Davy de Virville, J. et al., Plant Physiol. 1998, 116, 627635.
 Hanson, A. D., Gage, D. A., Shachar-Hill, Y., Trends PlantSci. 2000, 5, 206213.
 Rbeill, F., Douce, R., in: Kruger, N. J., Hill, S. A. Ratcliffe, R.G., (Eds.), Regulation of Primary Metabolic Pathways inPlants, Kluwer Academic, Dordrecht 1999, pp. 5399.
 Igamberdiev, A., Bykova, N. V., Kleczkowski, L. A., PlantPhysiol. Biochem. 1999, 37, 503513.
 Ogata, K., Volini, M., J. Biol. Chem. 1990, 265, 80878093. Schmidt, A., Jager, K., Annu. Rev. Plant Physiol. Plant Mol.
Biol. 1992, 43, 325349. Westley, J., Adv. Enzymol. 1973, 39, 327368.
 Ogata, K., Dai, X., Volini, M., J. Biol. Chem. 1989, 264, 27182725.
 Borovskii, G. B., Stupnikova, I. V., Antipina, A. I., Downs,C. A., Voinikov, V. K., J. Plant Physiol. 2000, 156, 797800.
 Bray, A. E., in: Singh, A. (Ed.), Stress Induced Gene Ex-pression in Plants, Hardwood Academic, New York 1994,pp. 123.
 Delseny, M., Gaubier, P., Hull, G., Saez-Vasquez, J. et al., in:Singh, A. (Ed.), Stress Induced Gene Expression in Plants,Hardwood Academic, New York 1994, pp. 2559.
 Wolkers, F. W., McCready, S., Brandt, W. F., Lindsey, G. G.,Hoekstra, F. A., Biochem. Biophys. Acta 2001, 1544, 196206.
 DeRocher, A. E., Vierling, E., Plant J. 1994, 5, 93102.