Proteomics of the Chloroplast: Experimentation and Prediction Klaas Jan Van Wijk

trends in plant science Reviews Proteomics of the chloroplast: experimentation and prediction Klaas Jan van Wijk

New technologies, in combination with increasing amounts of plant genome sequence data, have opened up incredible experimental possibilities to identify the total set of chloroplast proteins (the chloroplast proteome) as well as their expression levels and post-translational modifications in a global manner. This is summarized under the term ‘proteomics’ and typically involves two-dimensional electrophoresis or chromatography, mass spectrometry and bioinformatics. Complemented with nucleotide-based global techniques, proteomics is expected to provide many new insights into chloroplast biogenesis, adaptation and function.

hloroplasts are chlorophyll-containing plastids and origi- bound to become another important tool for plant biology. Several nate from proplastids, which are generally maternally plant proteomics studies have been published in recent years10. Cinherited via the embryo. Although the study of the chloro- However, in these studies, protein identification was achieved plast is a classic field in plant biology, there is no good overview through Edman sequencing, which necessarily limited the identi- of the total set of chloroplast proteins (the chloroplast proteome). fication of proteins in terms of cost, speed and sensitivity. Mass Improvements in two-dimensional electrophoresis (2-DE) and spectrometry will allow identification at a much higher speed and mass spectrometry have, in combination with increasing amounts with 100–1000 times less protein. Two plant proteomics studies of sequence data from Arabidopsis, rice, maize and other plant using mass spectrometry have been published recently, one con- species, opened up fantastic experimental possibilities enabling cerning anoxia tolerance in maize root tips11 and the other on pea the chloroplast proteins as well as their expression levels and thylakoid proteins12. An explosion of plant proteomics initiatives post-translational modifications to be identified rapidly. This can be expected in the coming years. is summarized under the term ‘proteomics’. Proteomics typically involves biochemical purification techniques such as 2-DE, Compartmenting the chloroplast proteome chromatography or affinity purification, mass spectrometry and From a biochemical point of view, the chloroplast can be divided bioinformatics1,2. Complemented with other functional genomics into several compartments, with each compartment having its own techniques such as cDNA or oligonucleotide microarrrays (Box 1) specific subset of proteins, or subproteome. To characterize the and reverse genetics, a better understanding of chloroplast bio- chloroplast proteome fully, either experimentally or by prediction, genesis, adaptation to the environment, signal transduction and it is useful (and probably essential) to subdivide the chloroplast metabolic pathways can be obtained. proteome into such subproteomes13. Only then can we devise opti- Here, we discuss in detail such an experimental approach to the mal experimental strategies to identify and characterize most pro- characterization of the chloroplast proteome. A complete charac- teins, including those that are hydrophobic5, of low abundance or terization includes not only the identification of proteins but also transiently expressed14. For each of the chloroplast compartments, studies of their expression levels, post-translational modifications, we briefly review current knowledge of each corresponding sub- protein–protein interactions and apparent discrepancies between proteome and discuss possible experimental and theoretical strat- the identified proteins and their predicted protein sequence from egies for further characterization. nucleotide sequencing data. We also comment on possibilities and limitations for the theoretical prediction of the chloroplast pro- Stromules, the chloroplast envelope and vesicles teome based on targeting or presequence information. Starting from the cytosolic side of the chloroplast, the first compartment is the chloroplast outer and inner envelope. The enve- Experimental characterization of the chloroplast proteome by lope is the site of transport of metabolites, proteins and 2-DE and mass spectrometry messengers between plastids and the cytosol15,16. The inner enve- The improvement of 2-DE through the development of immobi- lope membrane is also a site for the biosynthesis of several prod- lized pH gradients3 and optimization of solubilization tech- ucts (e.g. lipids and pigments15,17,18) and has also been implicated niques4,5 now allows the reproducible separation of more than in DNA replication and transcription of the chloroplast genome19. 2000 proteins on a single 2-DE gel. Such gel-separated proteins Several protein complexes involved in the translocation of can be identified by mass spectrometry if genomic information is nucleus-encoded chloroplast proteins have been characterized in available1,2,6 (Box 2). In addition, mass spectrometry is a powerful great detail20,21. At least 100 protein bands can be resolved on one- tool for analyzing isoforms, secondary modifications of proteins dimensional electrophoresis (1-DE) silver- or Coomassie-stained (e.g. glycosylation and phosphorylation) and proteolysis using gels of purified inner and outer membranes. However, 1-DE gels low amounts (picomoles to attomoles) of proteins7–9. Thus, a pro- do not have sufficient resolution to obtain a global overview of the teomics approach allows us to bridge the gap between genomic envelope proteome or to study post-translational modifications or sequence information and the actual protein population in a cell. changes in protein expression. Proteomics is already an important tool in medical research and No successful systematic analysis of the envelope proteome has the analysis of yeast and prokaryotes. With the rapid progress of been carried out to date, which is partly related to the hydrophobic the sequencing of the Arabidopsis genome and ongoing EST and nature of this subproteome. The 2-DE of envelope membrane genomic sequencing of many agricultural crops, proteomics is proteins was reported to be unsuccessful in the recovery of

Box 1. Functional genomics tools to understand Box 2. Mass spectrometry for the analysis of proteins cellular processes and peptides

Transcriptomics Mass spectrometry is the preferred method in the study of protein Definition • The systematic analysis of accumulated identification. Mass spectrometers with ‘soft’ ionization techniques transcripts in the cell or tissue allow the rapid identification of proteins provided that genomic or Type of information • The accumulation of transcripts, indicating cDNA sequence is available. Protein identification and characteriz- the level of gene expression ation generally involves two types of mass spectrometers: Strong points • Extremely sensitive • Matrix-assisted laser desorption–ionization time-of-flight • Many different transcripts can be monitored (MALDI-TOF) mass spectrometers – accurately measure the simultaneously masses of a protein (mixture) or of proteolytic digests of gel Weak points • mRNA levels often do not correspond to separated or otherwise purified proteins. A selected protein spot protein accumulation is digested with a site-specific protease such as trypsin, resulting • Cannot study post-translational modifications in a set of peptides. The masses of the peptides are then mea- • Location of mRNA does not provide sured by MALDI-TOF MS, resulting in a list of peptide information about the location of the gene masses. For each entry in the nucleotide and protein databases, product the masses of the predicted tryptic peptides are calculated and compared (within the experimental mass accuracy) with the Proteomics list of measured peptide masses using web-based search Definition • The systematic analysis of the proteins engines (e.g. Protein Prospector, Mascot, Profound). The cor- (the proteome) of a cell, tissue, organelle rect protein will have many ‘matching’ peptides. Proteins can be or membrane identified even when they consist of a mixture of two or three Type of information • The identification and expression level proteins. This method relies on the mass accuracy (5–15 ppm) of the proteome and sensitivity (femtomole range) of the latest generation of • Post-translational modifications and protein– MALDI-TOF MS instruments. protein interactions (protein complexes) • Tandem mass spectrometer – often coupled to liquid chro- Strong points • Relatively fast and sensitive (femtomole matography, using electrospray ionization with a collision cell for range) identification of proteins induced fragmentation (the technique is thus abbreviated to • Protein–protein interactions can be studied ESI-MS/MS). When a protein cannot be positively identified • Monitors the protein directly, rather than by MALDI-TOF MS, peptide sequence tags are obtained by monitoring the mRNA ESI-MS/MS. The individual peptides are ‘screened’ in the first Weak points • With current technology, the number of section of the tandem mass spectrometer and selected peptides proteins that can be simultaneously followed are subsequently further fragmented along the protein backbone is not as high as with transcriptomics by collision with argon or nitrogen molecules. This is termed • Protein separation steps are still time- collision-induced dissociation. These peptide fragments are consuming then separated by a second analyzer to provide amino acid • Monitoring the expression levels of sequence information. The peptide sequence tags and mea- membrane proteins is technically demanding sured ion masses are used to search for proteins in the database using specialized software. These two types of instruments together allow the rapid identification of proteins, the accurate determination of partial amino acid sequences envelope proteins22, a phenomenon that has been reported for and the elucidation of post-translational protein modifications. membrane proteins in general5. However, it is at least possible to resolve smaller membrane proteins (1– 4 transmembrane domains) on 2-DE gels if soluble proteins as well as lipids are removed23. A similar strategy, using extraction with organic solvents, but with lipid transfer from the inner envelope to the thylakoids15,17,28. The 1-DE, has been used to identify several hydrophobic chloroplast concept of active vesicle formation within the chloroplast was fur- envelope proteins24. In this way, the most hydrophobic proteins ther supported in other studies29,30, and such vesicles might con- from chloroplast envelope preparations could be enriched while tain their own subproteome. excluding more hydrophilic proteins, thereby reducing the com- A low-density chloroplast membrane fraction with an acyl lipid plexity of the protein mixture. Using this method, a few integral composition similar to inner envelopes and thylakoid membranes envelope membrane proteins were identified, and ~5–10% of the has been isolated from the green alga Chlamydomonas rein- total envelope protein content were estimated to be hydrophobic hardtii31. Several chloroplast mRNA binding proteins were found proteins, representing at least 15–20 different proteins24. to be strongly enriched, suggesting that these membranes could be Intriguing tube-like structures protruding out of the chloroplast a site of chloroplast gene expression31. To elucidate the role of envelope were detected in chloroplasts and chlorophyll-free pro- these intriguing membranes in biogenesis, a more systematic plastids and plastids, using transgenic plants in which green fluor- analysis of their proteins is urgently needed. escent protein (GFP) was targeted into the chloroplast25,26. This confirmed less detailed observations reported between 1960 and Chloroplast stroma the 1980s (Ref. 27). These dynamic structures were named stro- The chloroplast stroma is a compartment with a high protein con- mules (for stroma filled tubules) and sometimes interconnected tent and many well known enzymes involved in carbon assimi- different plastids with the surrounding nuclei and mitochondria. lation, as well as many biosynthetic pathways. Although best Analysis of the proteome of these stromules could provide an known for their role in photosynthesis, chloroplasts synthesize insight into their specific function. many essential compounds, such as plant hormones, fatty acids

Microscopic and radiolabeling studies have revealed that vesi- and lipids, amino acids, vitamins (B1, K1 and E), purine and cles can form at the inner chloroplast envelope and are involved in pyrimidine nucleotides, and secondary metabolites such as

October 2000, Vol. 5, No. 10 421 trends in plant science Reviews

alkaloids or isoprenoids. Moreover, chloroplasts are also required for nitrogen and sulfur assimilation. In addition to these enzymes, (a) Isolation of chloroplasts the stroma also contains the transcriptional and translation Purification of chloroplast subproteomes machinery. No systematic analysis of the stromal proteome has been presented in the literature. Clearly, a large-scale protein analysis during chloroplast devel- pI opment, environmental changes and of (chloroplast) mutants 3.0 10.0 (b) could reveal connections between the many biosynthetic pathways and provide insight into chloroplast functions and signaling. However, it will be a challenge to resolve the proteins of lower MW abundance, given the presence of several abundant proteins. Removal of these dominant proteins (or protein complexes) by affinity purification or other separation techniques will probably be crucial to get a good overview of the stromal proteome.

Nucleoids and plastoglobules (c) Pick spots The chloroplast genome is organized in so-called nucleoids, in gel digestion which are associated with the inner envelope. Nucleoids contain extraction of peptides at least 15–20 proteins, which are poorly characterized19,32. Identi- fication of these proteins might give us an insight into gene activation, genome organization and, possibly, DNA replication (d) Identification by mass spectrometry during chloroplast division. Other interesting structures in the MALDI-TOF ESI-MS/MS chloroplast include the plastoglobules, which are lipid-containing particles in the stroma that are thought to serve as lipid reservoirs for thylakoid membranes. To date, two proteins (PG1 and fib- Peptide masses Sequence tags rillins or PAP) have been identified33,34.

Thylakoid membrane system (e) Bioinformatics The thylakoid membrane system contains four abundant multi- subunit protein complexes (Photosystems I and II, the ATP syn- thase, and the cytochrome b6f complex). Together, they contain (f) Verification of identified proteins ~70 proteins and carry out the photosynthetic reactions. The thy- Ð analysis of targeting signals lakoid membrane might contain many other proteins that are Ð comparison experimentally determined and involved in the biogenesis and regulation of these complexes35,36. Ð predicted MW and pI

This includes processes such as biosynthesis and ligation of cofac- tors, and the insertion, folding or degradation of proteins. Based on preliminary information and postulated functions, at least 100 proteins (many of low abundance) are expected to be present. Fig. 1. Possible strategy for identifying chloroplast proteins using Recently, we have initiated a systematic analysis of thylakoid two-dimensional gel electrophoresis, mass spectrometry and tar- 12 geting analysis. (a) Purification of intact chloroplasts followed by proteins from pea (Fig. 1). We constructed high-resolution 2-DE fractionation of different chloroplast compartments, such as thyl- maps of lumenal and peripheral proteins of the thylakoid mem- akoids, stroma and envelope membranes. Further separation and brane system purified from intact pea chloroplasts. After correc- delipidation is required for the identification of membrane proteins. tion for possible isoforms and post-translational modifications, at (b) After isolation, chloroplast subproteomes are separated accord- least 200–230 different lumenal and peripheral proteins were cal- ing to their isoelectric point (pI) and then according to their mol- culated to be present. 61 proteins were identified by mass spec- ecular weight (MW), resulting in a two-dimensional gel. The spots trometry and Edman sequencing12, of which 33 had a clear are then visualized by Coomassie blue or silver staining. (c) function or functional domain, whereas no function could be Individual protein spots are selected, excised from the gel and assigned for ten proteins. For the other 18 proteins, no corre- digested with a site-specific protease, resulting in a set of peptides. sponding ESTs or full-length genes could be identified at the time (d) Extracted peptides from each gel spot are measured by matrix- assisted laser desorption–ionization time-of-flight (MALDI-TOF) of publication, in spite of experimental determination of a signifi- mass spectrometry. For further identification and characterization, cant amount of amino acid sequence. Ongoing genome sequenc- selected peptides are analyzed by electrospray tandem mass spec- ing is expected to identify these genes, and they can then be trometry (ESI-MS/MS), resulting in the generation of amino acid analyzed for functional domains. sequence tags. (e) The set of peptide masses measured by MALDI- Although pea is a good model system for biochemical work, TOF mass spectrometry are compared with the masses of the pre- there is one important drawback to using it for proteomics studies: dicted peptides for each entry in the sequence databases, possibly only a limited amount of DNA and protein sequences are avail- identifying the protein. The sequence tags obtained by ESI-MS/MS able for pea, and protein identification must therefore be carried are used for further confirmation of the identified protein or to ana- out based on the homology of pea proteins to well sequenced plant lyze post-translational modifications. (f) The identified proteins are species, such as Arabidopsis thaliana. Homology-based identifi- further analyzed for positive identification by analyzing targeting 12 information, such as chloroplast transit peptide or lumenal transit cation with mass spectrometry data is possible but generally peptide. In addition, the experimentally determined molecular requires a larger amount of experimentally determined protein mass and isoelectric point are compared with the predicted values, sequence tags or peptide mass fingerprints. Therefore, the use of after removal of predicted (cleavable) transit peptides. Arabidopsis, rice or other plant species that will be completely sequenced is highly favorable.

422 October 2000, Vol. 5, No. 10 trends in plant science Reviews

The construction of high resolution 2-DE maps of Arabidopsis higher sensitivity) 10% of the proteins encoded by the assigned thylakoids has therefore been initiated and is well under way, with genes on chromosomes II and IV were estimated to be located more than 1000 protein spots analyzed by matrix-assisted laser in mitochondria and 14% to be located in the chloroplast. desorption–ionization time-of-flight mass spectrometry (J.B. Cleavage site prediction for identified chloroplast proteins was Peltier et al., unpublished). Analysis of the integral thylakoid pro- 40% correct when allowing an error of Ϯ2 amino acids. This teins is also in progress. These 2-DE maps can be used as refer- prediction can only be improved after generating more experi- ence maps to study post-translational modifications or the mental data. expression of thylakoid proteins under varying environmental Thylakoid proteins destined for the lumen have an additional conditions and during chloroplast development, or to evaluate the transit peptide that shows strong similarities to bacterial signal effect of specific gene deletions. In addition, these newly identi- peptides21,41. In the aforementioned thylakoid proteomics study, fied proteins will provide a useful database to more rationally the lumenal transit peptides of a set of 26 non-redundant proteins design strategies aimed at understanding thylakoid biogenesis. (16 known and 10 newly discovered) were compared by aligning the sequences according to their experimentally determined Prediction of the chloroplast proteome through analysis of cleavage sites. As expected, the presequence and cleavage site had targeting signals similar features to signal peptides in bacteria. However, the lume- Attempts have been made to predict the size of the chloroplast nal transit peptides also showed a strong presence of prolines at proteome based on location prediction programs and relationships the end of the hydrophobic domain, a nearly complete conser- to cyanobacterial ancestors, and estimates vary from 1950 to 2500 vation (25 out of 26) of the alanine at the Ϫ1 position and an abun- (Refs 37,38). Several programs are available on the Internet to dance of glutamic acid at the ϩ2 and ϩ4 positions. The prediction predict the cellular location of a protein. For chloroplasts, two program SignalP, which was originally developed to predict programs [PSORT and ChloroP (Box 3)] have been available for cleavable signal peptides of secretory proteins in bacteria and several years. Both programs were used on a set of identified eukaryotes, also predicted lumenal transit peptides and their nuclear encoded thylakoid proteins to test the prediction for cleavage sites, but with mixed success12. However, if SignalP chloroplast localization and the expected cleavage site by stromal could be adapted to lumenal signal peptides by using the specific processing peptidase(s)12. Ninety four percent of this set was cor- features of lumenal proteins as mentioned above, its prediction of rectly predicted to be located in the chloroplast by ChloroP, lumenal proteins might be significantly improved. whereas PSORT predicted chloroplast localization for 52%. Recently, a new chloroplast and mitochondrial predictor has Unfortunately, ChloroP also predicts a significant number of false been developed (Box 3). This program is still in development but positives, which are mostly mitochondrial proteins: out of a test it is reported to be good at discriminating between proteins local- set of 715 Arabidopsis entries in the SWISS-PROT database, 11% ized in the chloroplasts and mitochondria, and at recognizing pro- were falsely predicted as chloroplast proteins39. teins targeted to both organelles (I. Small, pers. commun.). However, the program TargetP has recently become available What does this mean for the use of these programs to identify (Box 3). This is basically a retrained ChloroP integrated with a (or confirm) chloroplast proteins and their location within the new mitochondrial predictor and a retrained version of SignalP chloroplast? When the complete annotated genome of Arabid- (Box 3). TargetP is reported to be about three times better than opsis or other higher plants becomes available, it is expected that ChloroP at discriminating between chloroplast and mitochondrial many chloroplast-localized proteins will be tentatively identified signal sequences and should therefore result in fewer false posi- based on presequence information. In particular, the lumenal pro- tives for chloroplast prediction40. Using N-terminal sequence teins should be predicted with high confidence and sensitivity by information, it discriminates between proteins destined for the initial screening with TargetP set at a low confidence level (or mitochondrion, the chloroplast, the secretory pathway and ‘other’ high sensitivity) followed by analysis with SignalP in Gram-nega- locations with a reported success rate of 85% on redundancy- tive mode. Recognition of outer envelope proteins based on pri- reduced test sets of mitochondrial, chloroplast, secretory, nuclear mary sequence information is difficult or impossible because and cytosolic plant proteins. However, these test sets excluded most lack obvious chloroplast targeting signals42. Several tRNA other organelles such as vacuoles and peroxisomes. synthetases have been reported to be targeted to both the mito- In a TargetP analysis of the complete Arabidopsis chromo- chondria and chloroplasts43, but they are not easily recognized by somes II and IV, the number of estimated chloroplast proteins prediction as chloroplast proteins. Several more dual targeted pro- was 171 (2.2% of 7798 assigned genes; Ref. 40) if using only teins are likely to be present. the highest level of confidence (corresponding to no false posi- Other chloroplast proteins that might escape positive prediction tives on the test sets). Allowing lower levels of confidence (or are those targeted not via information in the N terminus but by alternative mechanisms. This could include targeting via the C terminus, as observed for plant vacuoles44 and yeast mitochon- dria45, or targeting via the so-called ‘hitch-hiker’ mechanism that Box 3. Programs to predict the cellular location of has been observed in bacteria46, or targeting as polyproteins47. a protein Predicting whether chloroplast membrane proteins are destined for the thylakoid membrane or the (inner) envelope is currently ChloroP, TargetP and SignalP not possible because there are no obvious targeting signals. Thus, http://www.cbs.dtu.dk/services/ the most confident location prediction within the chloroplast is currently for the lumenal proteins. PSORT It is important to realize that, for these programs to work cor- http://psort.nibb.ac.jp/ rectly, the assignment of the start methionine needs to be correct. Several cases were observed in which incorrect assignment Chloroplast and mitochondrial predictor 12 http://www.inra.fr/Internet/Produits/Predotar/ resulted in a negative chloroplast (incorrect) prediction . In addition, chloroplast location prediction programs cannot determine actual protein accumulation. Finally, as a word of caution, the

October 2000, Vol. 5, No. 10 423 trends in plant science Reviews

currently available chloroplast predictors cannot identify chloro- techniques (such as affinity purification and gel filtration) fol- plast proteins of the unicellular alga C. reinhardtii, which is an lowed by mass spectrometry or yeast two-hybrid systems will be important model organism in the study of chloroplast biogenesis35,48. essential to unravel biochemical functions at a molecular level. Proteomics studies on chloroplast biogenesis mutants obtained by Post-translational modifications, isoforms and mRNA editing phenotypic screening, chlorophyll fluorescence screens55, other Characterization of the chloroplast proteome involves the identifi- phenotypic criteria or reverse genetics will help to pinpoint the cation not only of proteins but also of their post-translational mod- function of the gene products involved in chloroplast biogenesis ifications. Many post-translational modifications of chloroplast and function. proteins have been reported, including methylation (for RbcS)49, carbamylation (for RbcS, RbcL)50, glycosylation (for CF)51 and Acknowledgements palmitoylation (for D1)52. However, no global analysis of post- I gratefully acknowledge members of my laboratory and Olof translational modifications has been carried out to date. With Emanuelsson for numerous stimulating discussions and critically the recent developments of mass spectrometry and 2-DE, such reading the manuscript. My funding was provided by the Swedish analysis is now feasible and could lead to many interesting and National Research Council (NFR), the Swedish Agricultural unexpected observations. For instance, systematic mapping of Research Council (SJFR), the Swedish Strategic Funds (SSF) and phosphorylation of chloroplast proteins during light–dark transi- the Carl Trygger Foundation. tions is likely to reveal unexpected kinetics and substrates, possibly leading to more insight into redox-driven signal transduction References chains and (de)activation mechanisms in the chloroplast. 1 Roepstorff, P. (1997) Mass spectrometry in protein studies from genome to Proteomics approaches can also provide a global overview of function. Curr. Opin. Biotechnol. 8, 6–13 splicing, mRNA editing and the expression of isoforms. Several 2 Yates, J.R., III (2000) Mass spectrometry: from genomics to proteomics. cases of mRNA editing53 and splicing54 have been described for Trends Genet. 16, 5–8 chloroplast proteins, and multigene families have been reported 3 Görg, A. et al. (1988) The current state of two-dimensional electrophoresis in pea and spinach. These phenomena can, of course, also be with immobilized pH gradients. Electrophoresis 9, 531–546 detected from a careful analysis of the mRNA (or cDNA), but pro- 4 Herbert, B. (1999) Advances in protein solubilisation for two-dimensional teomics is clearly a strong alternative or complementary way to electrophoresis. Electrophoresis 20, 660–663 elucidate such events. 5 Santoni, V. et al. (2000) Membrane proteins and proteomics: un amour impossible? Electrophoresis 21, 1054–1070 Functional proteomics of the chloroplast 6 Dainese, P. et al. (1997) Probing protein function using a combination of gene Protein reference maps of subfractions of various organisms con- knockout and proteome analysis by mass spectrometry. Electrophoresis 18, structed by 2-DE are expected to become a central tool for organ- 432–442 izing and understanding proteomes. A few web sites with small 7 Kuster, B. and Mann, M. (1998) Identifying proteins and post-translational organized 2-DE databases are already available (e.g. yeast, modifications by mass spectrometry. Curr. Opin. Struct. Biol. 8, 393–400 E. coli, plant plasma membranes and chloroplasts; Box 4). Refer- 8 Burlingame, A.L. et al. (1998) Mass spectrometry. Anal. Chem. 70, 647R–716R ence 2-DE maps will be used to follow differential protein express- 9 Wilkins, M.R. et al. (1999) High-throughput mass spectrometric discovery of ion and post-translational modifications. A combination of protein post-translational modifications. J. Mol. Biol. 289, 645–657 proteomics with other large-scale techniques such as DNA 10 Thiellement, H. et al. (1999) Proteomics for genetic and physiological studies microarrays will be needed because they are expected to be highly in plants. Electrophoresis 20, 2013–2026 complementary (Box 1). DNA- and RNA-based technologies are 11 Chang, W.W. et al. (2000) Patterns of protein synthesis and tolerance of more sensitive tools, which can handle much larger numbers of anoxia in root tips of maize seedlings acclimated to a low-oxygen genes simultaneously. By contrast, proteomics determines the environment, and identification of proteins by mass spectrometry. Plant presence of the gene product directly (rather than at the mRNA Physiol. 122, 295–318 level) and can also monitor post-translational modifications, 12 Peltier, J.B. et al. (2000) Proteomics of the chloroplast. Systematic including proteolysis. identification and targeting analysis of lumenal and peripheral thylakoid Only after the experimental determination of further chloro- proteins. Plant Cell 12, 319–342 plast proteins and the cleavage site of their targeting signals can 13 Cordwell, S.J. et al. (2000) Subproteomics based upon protein cellular the prediction of chloroplast localization and cleavage sites be location and relative solubilities in conjunction with composite two- improved. To identify the functional role of newly discovered dimensional electrophoresis gels. Electrophoresis 21, 1094–1103 proteins, additional approaches (e.g. reverse genetics) will be 14 Corthals, G.L. et al. (2000) The dynamic range of protein expression: a needed in combination with proteomics. Systematic analysis of challenge for proteomic research. Electrophoresis 21, 1104–1115 protein–protein interactions using non-denaturing biochemical 15 Joyard, J. et al. (1998) The biochemical machinery of plastid envelope membranes. Plant Physiol. 118, 715–723 16 Soll, J. and Tien, R. (1998) Protein translocation into and across the chloroplastic envelope membranes. Plant Mol. Biol. 38, 191–207 Box 4. 2-DE databases 17 Marechal, E. et al. (1997) Lipid synthesis and metabolism in the plastid envelope. Physiol. Plant 100, 65–77 Yeast, E. coli and others 18 Bruce, B.D. (1998) The role of lipids in plastid protein transport. Plant Mol. http://www.expasy.ch/ch2d/ Biol. 38, 223–246 19 Sato, N. et al. (1999) Do plastid envelope membranes play a role in the Plant plasma membranes http://sphinx.rug.ac.be:8080/ppmdb/index.html expression of the plastid genome? Biochimie 81, 619–629 20 May, T. and Soll, J. (1999) Chloroplast precursor protein translocon. FEBS Chloroplasts Lett. 452, 52–56 http://www.biokemi.su.se/chloroplast/ 21 Keegstra, K. and Cline, K. (1999) Protein import and routing systems of chloroplasts. Plant Cell 11, 557–570

424 October 2000, Vol. 5, No. 10 trends in plant science Reviews

22 Adessi, C. et al. (1997) Two-dimensional electrophoresis of membrane proteins: 42 Reumann, S. and Keegstra, K. (1999) The endosymbiotic origin of the protein a current challenge for immobilized pH gradients. Electrophoresis 18, 127–135 import machinery of chloroplastic envelope membranes. Trends Plant Sci. 4, 23 Molloy, M.P. et al. (2000) Proteomic analysis of the Escherichia coli outer 302–307 membrane. Eur. J. Biochem. 267, 2871–2881 43 Peeters, N.M. et al. (2000) Duplication and quadruplication of Arabidopsis 24 Seigneurin-Berny, D. et al. (1999) Differential extraction of hydrophobic thaliana cysteinyl- and asparaginyl-tRNA synthetase genes of organellar proteins from chloroplast envelope membranes: a subcellular-specific proteomic origin. J. Mol. Evol. 50, 413–423 approach to identify rare intrinsic membrane proteins. Plant J. 19, 217–228 44 Di Sansebastiano, G.P. et al. (1998) Specific accumulation of GFP in a non- 25 Kohler, R.H. and Hanson, M.R. (2000) Plastid tubules of higher plants are acidic vacuolar compartment via a C-terminal propeptide-mediated sorting tissue specific and developmentally regulated. J. Cell Sci. 113, 81–89 pathway. Plant J. 15, 449–457 26 Köhler, R.H. et al. (1997) Exchange of protein molecules through connections 45 Lee, C.M. et al. (1999) The DNA helicase, Hmi1p, is transported into between higher plant plastids. Science 276, 2039–2042 mitochondria by a C-terminal cleavable targeting signal. J. Biol. Chem. 274, 27 Wildman, S.G. et al. (1962) Chloroplast and mitochondria in living plant cells: 20937–20942 cinematographic studies. Science 138, 434–436 46 Rodrigue, A. et al. (1999) Co-translocation of a periplasmic enzyme complex 28 Morré, D.J. et al. (1991) Stromal low temperature compartment derived from by a hitchhiker mechanism through the bacterial tat pathway. J. Biol. Chem. inner membrane of the chloroplast envelope. Plant Physiol. 97, 1558–1564 274, 13223–13228 29 Hugueney, P. et al. (1995) Identification of a plastid protein involved in 47 Mant, A. and Robinson, C. (1998) An Arabidopsis cDNA encodes an apparent vesicle fusion and/or membrane protein translocation. Proc. Natl. Acad. Sci. polyprotein of two non-identical thylakoid membrane proteins that are U. S. A. 92, 5630–5634 associated with photosystem II and homologous to algal ycf32 open reading 30 Park, J.M. et al. (1998) A dynamin-like protein in Arabidopsis thaliana is frames. FEBS Lett. 423, 183–188 involved in biogenesis of thylakoid membranes. EMBO J. 17, 859–867 48 Hippler, M. et al. (1998) Chlamydomonas genetics, a tool for the study of 31 Zerges, W. and Rochaix, J.D. (1998) Low density membranes are associated bioenergetic pathways. Biochim. Biophys. Acta 1367, 1–62 with RNA-binding proteins and thylakoids in the chloroplast of 49 Grimm, R. et al. (1997) Postimport methylation of the small subunit of Chlamydomonas reinhardtii. J. Cell. Biol. 140, 101–110 ribulose-1,5-bisphosphate carboxylase in chloroplasts. FEBS Lett. 408, 32 Sato, N. et al. (1993) Detection and characterization of a plastid envelope 350–354 DNA-binding protein which may anchor plastid nucleoids. EMBO J. 12, 555–561 50 Smith, H.B. et al. (1988) Subtle alteration of the active site of ribulose 33 Pozueta-Romero, J. et al. (1997) A ubiquitous plant housekeeping gene, PAP, bisphosphate carboxylase/oxygenase by concerted site-directed encodes a major protein component of bell pepper chromoplasts. Plant mutagenesis and chemical modification. Biochem. Biophys. Res. Commun. Physiol. 115, 1185–1194 152, 579–584 34 Kessler, F. et al. (1999) Identification of proteins associated with 51 Maione, T.E. and Jagendorf, A.T. (1984) Partial deglycosylation of plastoglobules isolated from pea (Pisum sativum L.) chloroplasts. Planta 208, chloroplast coupling factor 1 (CF1) prevents the reconstitution of 107–113 photophosphorylation. Proc. Natl. Acad. Sci. U. S. A. 81, 3733–3736 35 Wollman, F.A. et al. (1999) The biogenesis and assembly of photosynthetic 52 Mattoo, A.K. and Edelman, M. (1987) Intramembrane translocation and proteins in thylakoid membranes. Biochim. Biophys. Acta 1411, 21–85 posttranslational palmitoylation of the chloroplast 32-kDa herbicide-binding 36 van Wijk, K.J. Proteins involved in biogeneis of the thylakoid membrane. In protein. Proc. Natl. Acad. Sci. U. S. A. 84, 1497–1501 Advances in Photosynthesis: Regulatory Aspects of Photosynthesis (Vol. 8) 53 Sugita, M. and Sugiura, M. (1996) Regulation of gene expression in (Andersson, B. and Aro, E-M., eds), Kluwer (in press) chloroplasts of higher plants. Plant Mol. Biol. 32, 315–326 37 Abdallah, F. et al. (2000) A prediction of the size and evolutionary origin of 54 Mano, S. et al. (1997) Stromal and thylakoid-bound ascorbate peroxidases are the proteome of chloroplasts of Arabidopsis. Trends Plant Sci. 5, 141–142 produced by alternative splicing in pumpkin. FEBS Lett. 413, 21–26 38 Martin, W. and Herrmann, R.G. (1998) Gene transfer from organelles to the 55 Meurer, J. et al. (1996) Isolation of high-chlorophyll-fluorescence nucleus: how much, what happens, and why? Plant Physiol. 118, 9–17 mutants of Arabidopsis thaliana and their characterisation by 39 Emanuelsson, O. et al. (1999) ChloroP, a neural network-based method for spectroscopy, immunoblotting and northern hybridisation. Planta predicting chloroplast transit peptides and their cleavage sites. Protein Sci. 8, 198, 385–396 978–984 40 Emanuelsson, O. et al. (2000) Predicting subcellular localization of proteins based on their N-terminal amino acid sequence. J. Mol. Biol. 300, 1005–1116 Klaas Jan van Wijk is at the Dept of Biochemistry, Arrhenius 41 Dalbey, R.E. and Robinson, C. (1999) Protein translocation into and across the Laboratories, Stockholm University, S-10691 Stockholm, Sweden bacterial plasma membrane and the plant thylakoid membrane. Trends (tel ϩ46 8 162420; fax ϩ46 8 153679; e-mail [email protected]). Biochem. Sci. 24, 17–22

October 2000, Vol. 5, No. 10 425