| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
|
Plant Physiology 135:723-734 (2004) © 2004 American Society of Plant Biologists Analysis of Curated and Predicted Plastid Subproteomes of Arabidopsis. Subcellular Compartmentalization Leads to Distinctive Proteome Properties1,[w]Computational Biology Service Unit, Cornell Theory Center (Q.S.) and Department of Plant Biology (K.J.v.W.), Cornell University, Ithaca, New York; and Stockholm Bioinformatics Center, Stockholm University, SE–10691 Stockholm, Sweden (O.E.)
Carefully curated proteomes of the inner envelope membrane, the thylakoid membrane, and the thylakoid lumen of chloroplasts from Arabidopsis were assembled based on published, well-documented localizations. These curated proteomes were evaluated for distribution of physical-chemical parameters, with the goal of extracting parameters for improved subcellular prediction and subsequent identification of additional (low abundant) components of each membrane system. The assembly of rigorously curated subcellular proteomes is in itself also important as a parts list for plant and systems biology. Transmembrane and subcellular prediction strategies were evaluated using the curated data sets. The three curated proteomes differ strongly in average isoelectric point and protein size, as well as transmembrane distribution. Removal of the cleavable, N-terminal transit peptide sequences greatly affected isoelectric point and size distribution. Unexpectedly, the Cys content was much lower for the thylakoid proteomes than for the inner envelope. This likely relates to the role of the thylakoid membrane in light-driven electron transport and helps to avoid unwanted oxidation-reduction reactions. A rule of thumb for discriminating between the predicted integral inner envelope membrane and integral thylakoid membrane proteins is suggested. Using a combination of predictors and experimentally derived parameters, four plastid subproteomes were predicted from the fully annotated Arabidopsis genome. These predicted subproteomes were analyzed for their properties and compared to the curated proteomes. The sensitivity and accuracy of the prediction strategies are discussed. Data can be extracted from the new plastid proteome database (http://ppdb.tc.cornell.edu).
Plastids are essential organelles of prokaryotic origin that are present in nearly every plant cell. Plastids are built up out of several compartments: (1) the outer and inner envelope membranes surrounding the plastids, providing a barrier for proteins and small molecules; (2) the soluble stroma within plastids, expected to contain thousands of different proteins; and (3) in the case of chloroplasts, the thylakoid membrane, an internal membrane system, holding the photosynthetic electron transport chain, as well as an unknown number of other proteins. The thylakoid membrane system forms large vesicles and its intrathylakoid space is called the thylakoid lumen, which contains an additional set of proteins.
To understand plastid function, biogenesis, and biosynthetic pathways, it is critical to characterize the plastid proteome: This includes protein expression levels, protein-protein interactions, and subplastid localization. A first step in the characterization of the plastid proteome is the careful assembly of all experimentally determined plastid proteins and their suborganellar localization. In addition to experimental data sets, prediction tools can and have been developed to predict the plastid proteome and suborganellar proteomes, with varying degrees of success (Abdallah et al., 2000
Numerous individual plastid proteins, as well as larger sets of plastid proteins, can be collected from the literature (for review, see van Wijk, 2004
Most proteins localized in plastids are synthesized as precursor proteins in the cytosol with a cleavable chloroplast transit peptide (cTP). Features of these cTPs are discussed by von Heijne (1989)
One particularly important question in plastid biogenesis is how nuclear-encoded integral membrane proteins (more than 500 are predicted) are targeted to the inner envelope and thylakoid membrane. With the exception of those proteins that carry a lumenal transit peptide (lTP) for targeting of the N terminus to the lumen (Mori and Cline, 2001
In this study, we carefully collected published experimental sets of integral thylakoid membrane proteins and integral inner envelope proteins. We analyzed these subproteomes, as well as the experimental thylakoid lumenal proteome, for their properties, with the objective of extracting potential predictors of subplastid localization and evaluating the number of putative chloroplast membrane proteins and their TMDs. Subsequently, we used the existing predictors, TargetP, SignalP, and TMHMM, as well as a newly developed predictor for the lumenal proteome, LumenP (Westerlund et al., 2003
Assembly of the Curated Experimental Integral Inner Envelope and Thylakoid Proteomes The nonredundant accession numbers for known thylakoid and inner envelope proteins of Arabidopsis were carefully collected from the literature and public databases. All proteins were then evaluated for being an integral membrane protein; for the most part, this was based on experimental data (e.g. not extractable with urea or salts, x-ray structure, or topology mapping by proteolysis, etc.) and, in some cases, based on prediction of TMDs by TMHMM or the consensus prediction reported in the Aramemnon database (http://aramemnon.botanik.uni-koeln.de/). If available, the number of experimentally determined TMDs for each protein was recorded. Importantly, only those proteins were listed for which it was very clear that they are integral either to the thylakoid membrane or to the inner envelope membrane. In total, 65 (75 when including alternative gene models in The Arabidopsis Information Resource [TAIR]; see also below) nuclear-encoded thylakoid membrane and 24 (27 when including different gene models) nuclear-encoded inner envelope membrane proteins were obtained (see Supplemental Table I at www.plantphysiol.org).
We also assembled extended experimental thylakoid and inner envelope integral membrane sets, adding additional integral membrane proteins for each membrane system, using data from Friso et al. (2004)
A number of parameters (length, pI, grand average of hydrophobicity [GRAVY; Kyte and Doolittle, 1982
Protein Size Very significant differences in protein size were observed between the experimentally identified lumenal, the integral thylakoid membrane, and the integral inner envelope proteomes (Table I). The envelope proteins were, on average, twice as large as the lumenal proteins and 75% bigger than the integral thylakoid proteins. The plastid-encoded integral thylakoid membrane proteins were, on average, 252 amino acids, but with a median value of only 140 amino acids (Table I).
Number of TMDs and Hydrophobicity The distribution of these predicted TMDs is shown in Figure 1, A to C. Figure 1A shows the distribution of predicted TMDs by TMHMM for the chloroplast- and nuclear-encoded thylakoid membrane proteins. Clearly, the thylakoid proteome is dominated by proteins predicted to have zero to three TMDs and a small group of proteins with nine or more TMDs (Fig. 1A). These TMD predictions were compared with the experimental TMD determinations reported in the literature. In the case of the chloroplast-encoded thylakoid membrane proteins, TMHMM slightly overpredicted the TMDs, with a total of 77 predicted for 72 known TMDs.
In the case of the nuclear-encoded thylakoid membrane proteins, a very significant percentage was not predicted to have any TMD. It turned out that TMHMM has a specific problem with predicting the 29 nuclear-encoded chlorophyll- and carotenoid- binding thylakoid membrane proteins (light-harvesting complexes [LHCs], Elips, PsbS, and Seps or Lils, Hlips, Scps, and Ohps; Jansson, 1999  ; Adamska, 2001). This LHC superfamily has an average of 2.82 TMDs, but TMHMM predicted only an average of 0.31. In contrast, Aramemnon listed quite an accurate prediction (Fig. 1), with an average of 2.10 TMDs. The poor prediction by TMHMM is most likely related to the presence of (conserved) positively and negatively charged residues within the TMDs used for helix-helix interactions (Adamska, 2001). Aramemnon and TMHMM both predicted very similarly that the curated envelope proteome falls into two classes of proteins with either one to four TMDs or nine or more TMDs (Fig. 1C). Surprisingly, approximately 50% of the curated lumenal proteome was predicted to be a membrane protein by Aramemnon. In contrast, only one out of the 53 confirmed lumenal proteins was a predicted membrane protein by TMHMM after removal of predicted cTPs. This overprediction was caused by the very long bipartite targeting sequence of lumenal proteins. The lTP, which is typically 40 to 50 amino acids down-stream of the N terminus, is identified as a TMD by many TMD predictors; a typical example, for instance, is the well-known lumenal electron transporter plastocyanin (At1g20340). A different way to characterize the integral membrane proteomes is by hydrophobicity, as calculated by the GRAVY index. The processed integral thylakoid proteome was, on average, slightly less hydrophobic than the processed inner envelope membrane proteome (Table I). The chloroplast-encoded thylakoid membrane proteome was significantly more hydrophobic than both nuclear-encoded proteomes (Table I). Cross-correlation of the GRAVY index with the number of predicted TMDs per amino acid listed in Aramemnon showed a linear correlation for the nuclear-encoded envelope and thylakoid membrane proteomes (Fig. 2, A and B). Since the chloroplast-encoded proteins are not in the Aramemnon database, we plotted for this group only the cross-correlation of the GRAVY index with the number of TMDs predicted by TMHMM (Fig. 2C); a good linear correlation was observed, as is evident from the regression analysis (Fig. 2C). This shows that the chloroplast-encoded thylakoid proteins are more hydrophobic because they have more TMDs per amino acid length.
pI The distribution of pI is strongly bimodal for the processed lumenal proteome (Table I; Fig. 3A). Removal of the cleavable cTPs and lTPs has a dramatic influence on the pI distribution, with median pI values of 8.53 for the unprocessed lumenal proteome and 5.31 for the processed proteome (Table I; Fig. 3A). In an earlier study of the thylakoid lumen proteome, we showed that these predicted pI values of processed proteins matched well (typically within 0.5 pH unit deviation) to the experimental pI values determined from two-dimensional gels (Peltier et al., 2002 ). This emphasizes that in silico proteome analysis of physical-chemical parameters should be done after removal of cleavable signal/transit peptides. Also, the processed integral thylakoid membrane proteome is characterized by acidic median (5.22) and average (6.0) pI values (Table I; Fig. 3A). In contrast, the processed (and unprocessed) inner envelope proteome is, on average, basic, with a median pI value of 8.7 and an average pI value of 8.1 (Table I; Fig. 3A). The two membrane proteomes showed a bi- or trimodal distribution of pI values (Fig. 3A). Manual inspection of the proteins at high and low pI did not suggest particular functions for the outliers in both populations. The low pI of the thylakoid membrane and lumenal proteome can be explained by a high content of the abundant acidic residues Asp (D) (pKr = 3.9) and Glu (E) (pKr = 4.07), when compared to the envelope proteome (Supplemental Fig. 1).
Cys Content Cys residues play an important role in redox reactions, as ligands and in stabilization of proteins and protein complexes by formation of disulfide bonds (Giles et al., 2003 ). Surprisingly, the Cys content is much lower for thylakoid proteins in the lumen and membrane, as compared to proteins in the inner envelope or elsewhere in the chloroplast (Table I). Importantly, the plastid-encoded integral thylakoid membrane proteome also has a low relative Cys content, with a median of 0.003 and an average of 0.005, expressed as the abundance ratio between Cys over all amino acid residues. This is about 3 times less than the Cys content in the envelope membrane proteome. The chloroplast-encoded stromal proteome has a much higher average Cys content of 0.015, clearly suggesting that the reduced Cys content in the thylakoid is related to its particular function. When expressed into the number of Cys residues per protein, 35% to 40% of the thylakoid lumen and membrane proteins have no Cys at all, whereas nearly all experimentally identified inner envelope membrane proteins have one or more Cys residues, with a peak at four Cys residues per protein (Fig. 3B). This significance will be discussed below.
cTP Characteristics
Functional Domains
The TargetP sensitivity (defined as the fraction of plastid-predicted proteins out of all true plastid proteins in our data sets) was 96%, 86%, and 78%, respectively, for the thylakoid lumen and membrane proteomes of the thylakoid and inner envelope (Supplemental Table II). These sensitivities are in the same range as the 85% reported originally (Emanuelsson et al., 2000 The experimental sets were then extended to encompass all known thylakoid proteins (lumenal, peripheral, and integral membrane), as well as proteins copurified with thylakoid and envelope proteins (stromal), totaling 283 proteins (Supplemental Table I). All are confirmed chloroplast proteins and they were initially identified by various biochemical and genetic approaches, thus avoiding experimental bias. TargetP predicted 89% of the 283 proteins correctly (Supplemental Table II) and 90% correctly after removal of 43 (15%) proteins with homology to a protein in the TargetP training set. A larger, unpublished, experimental data set with well over 400 stromal proteins from chloroplasts and non-green plastids confirmed this approximately 90% sensitivity (J.B. Peltier, Y. Cai, G. Friso, L. Giacomelli, V. Zabrouskov, Q. Sun, and K. van Wijk, unpublished data). The high sensitivity of TargetP is important for the prediction of plastid subproteomes, as detailed below.
We predicted the plastid proteome and subplastid localization by screening the latest annotation of the Arabidopsis genome (4.0 of ATH1.pep) using two prediction schemes (Fig. 4, A and B). The predictor, TargetP, was either combined with LumenP (Westerlund et al., 2003
It was predicted that 4,255 proteins have a cTP (14.9% of the total Arabidopsis proteome; Table II). Of those, 520 have one or more TMDs and are located in the inner envelope or thylakoid membrane, 291 (LumenP) or 285 (SignalP) are predicted to be soluble in the lumen, and an additional 57 integral membrane proteins were predicted to have an lTP. The remaining 3,387 proteins are assigned stromal proteins; however, it is important to note that these can be permanently or transiently associated with the thylakoid and/or inner envelope membrane via protein-protein, electrostatic, or hydrophobic interactions, as well as lipid anchors, but this is currently not possible to predict.
Thus, our prediction scheme suggests that 13.5% of the predicted plastid proteins are integral membrane proteins, compared to 22% predicted membrane proteins for the complete Arabidopsis genome (Table II). The consensus TMD prediction listed in the Aramemnon database currently reports that 24% of the 4,028 cTP proteins are predicted membrane proteins, compared to 25% of the Arabidopsis proteome (Table II; see "Discussion" for further details).
We compared the two prediction strategies (outlined in Fig. 4, A and B) for the lumenal proteome. The total number of predicted proteins was very similar for the two approaches, and both approaches gave very similar sensitivities (56% and 60%; Supplemental Table II; Fig. 4C). Twenty-six out of 51 confirmed lumenal proteins were correctly predicted by both approaches, and both approaches predicted some additional known lumenal proteins correctly, as well as 150 proteins with no confirmed lumenal location. However, both strategies also predicted an additional, nonoverlapping set of over 100 proteins. Since the rest of the lumenal proteome is unknown, it is hard to evaluate these two nonoverlapping sets.
We then compared the features of the experimental lumenal proteome with the predicted lumenal proteomes. The average and median lengths of proteins in the two predicted lumenal proteomes were 50% to 60% longer than the experimental lumenal proteome. This suggests that either the experimental set was biased toward smaller proteins or the predicted proteome contained a number of false positives. Comparison of the length of the cTP + lTP between experimental sets and predicted sets shows higher median and average values for the predicted sets (Fig. 4D). This suggests that the cutoff for maximum cTP + lTP length was too high. Strikingly, the cTP + lTP length for the predicted TAT proteins was very similar to the experimentally determined length, suggesting a more robust prediction for this subclass, as also demonstrated earlier (Peltier et al., 2002
The properties (length, GRAVY, pI, number of TMDs, and relative Cys content) of the predicted subproteomes, before and after removal of the predicted cTPs and lTPs, were analyzed and displayed using frequency distribution plots (data not shown). Mean and average values were calculated (Supplemental Table IV). The median and average lengths of the predicted chloroplast integral membrane proteome (inner envelope and thylakoid) after removal of the predicted cTP are 327 and 382 amino acids, respectively. This is in between the values for the experimental integral thylakoid proteome (213/244; Table I) and the integral inner envelope membrane proteome (398/427). The average and median pI values of the predicted total plastid membrane proteome are trimodal, with an average pI value of 7.39 (Supplemental Table IV).
Several proteins have been identified that are integral to the thylakoid membrane and have an N-terminal lumenal transit peptide. Known cases are CFo-II (At4g32260), psbW (At2g30570), PsbX (At2g06520), psbT1 (At3g21055), and psaF (At1g31330). Insertion of these proteins seems to occur without assistance of other proteins or energy requirements (Robinson et al., 2000
The TAIR database reports more than one gene model for 1,411 (5%) out of 27,170 genes of the annotated Arabidopsis genome (1,141 genes with two, 109 with three, and 17 with four or more gene models; each gene model might be biologically relevant). The 4,255 gene models predicted to have a cTP represented 4,013 genes (Table II). cTP prediction differentiated between the gene models for 36 out of those 4,013 genes. In the case of the curated subproteomes, TargetP prediction was only rarely affected by the different gene models presented in TAIR. Exceptions include carbonic anhydrase (At3g01500), for which three forms are present in TAIR (.1, .2, and .3). At3g01500.1 is not predicted to go to the plastid, whereas version .2 and .3 are correctly predicted to be plastid localized. The explanation is that At3g01500.1 is N-terminally truncated, thus lacking a proper cTP.
For a complete understanding of plant functions and biosynthetic and signaling pathways, it is important to determine and characterize the proteomes at different subcellular locations. This will also be important in long-term efforts to develop faithful, quantitative models for plant processes (for discussion, see Raikhel and Coruzzi, 2003 ).
Experimental proteomics using modern mass spectrometry has become a powerful tool, with continuous improvements in sensitivity, dynamic resolution, and quantification (Aebersold and Mann, 2003
Experimental proteome analysis of subcellular compartments can provide extensive protein sets to either train predictors or extract experimental parameters for design of filters, as we demonstrated earlier for the thylakoid lumen (Peltier et al., 2002 In this study, we focused on curation, analysis, and prediction of the chloroplast subproteomes encoded by the nuclear and plastid genomes of Arabidopsis. We carefully assembled proteins located in either the thylakoid membrane system or the plastid inner envelope membrane, with the goal of obtaining a better overview of the respective integral membrane proteomes and finding specific features for each set of proteins, possibly with predictive value for membrane localization. Significant sets of proteins could indeed be assembled and their analysis showed clear differences in the properties of each membrane proteome and associated functions. It is unlikely that these differentiating properties are due to an experimental bias, since these proteins and their corresponding genes were originally identified using various strategies, ranging from reverse and forward genetics screens, highly specific cross-link experiments, as well as more recent proteomics approaches involving gels or chromatography, followed by different types of mass spectrometry techniques.
The pI distribution was bimodal for both membrane systems (possibly trimodal for the envelope proteome), with, on average, a basic integral inner envelope membrane proteome and an acidic integral thylakoid membrane proteome. These pI distributions were strongly affected by removal of the predicted transit peptides. Currently, there is no good explanation for this pI distribution of thylakoid and envelope proteome, but it is likely related to the pH in the lumenal, stromal, and intra-envelope space. It is unclear if and how this connects to the positive-inside rule, which states that membrane proteins have, on average, a net positive charge on the loops facing the cis-side of the membrane (Gavel et al., 1991
Cys residues have a unique reactivity and they are involved in catalysis, redox activity, structural stabilization, and metal binding (Frand et al., 2000
Thylakoid proteins were, on average, smaller, more acidic, and, most significantly, contained less Cys residues when compared to the inner envelope proteome. The question is if these differences contain enough predictive power in order to discriminate integral inner envelope membrane proteins from integral thylakoid membrane proteins when applied to uncharacterized proteins. A three-dimensional plot in which pI, number of Cys residues, and protein length are combined shows that the expanded curated thylakoid and inner envelope membrane proteins (83 and 39 proteins, respectively; see Supplemental Table I) are generally well separated (Fig. 5A). Recently, published experimental envelope proteome studies (Ferro et al., 2002
The cTP cleavage site motifs and upstream and downstream amino acid regions for the different (curated and extended) subproteomes did not exhibit any striking differences and did not offer any predictive value for subplastid localization.
TMHMM was successful at predicting TMDs for both plastid- and nuclear-encoded thylakoid membrane proteins that were not part of the LHC superfamily (Jansson, 1999 It is interesting to note that, when using TMHMM, only 13% of the predicted plastid proteins have predicted TMDs whereas 22% of all annotated Arabidopsis genes are predicted to encode for membrane proteins. There could be two additive biological reasons for this relatively low percentage of membrane proteins in the plastid: (1) a plastid is not a complete organism and has only one internal membrane compartment (the thylakoid), whereas the complete cell has many types of organelles and corresponding membrane systems; and (2) the cTP set includes proteins in both non-green plastids without internal thylakoid membrane system and chloroplasts. The soluble, stromal proteome is expected to be quite different between non-green plastids and chloroplasts, whereas the integral membrane composition of the inner envelope might not be significantly different and an internal membrane system is absent in non-green plastids. The diversity of the plastid would thus add a lot of extra soluble proteins, but possibly few integral membrane proteins. The consensus prediction listed in the Aramemnon database suggests that the number of membrane proteins in the plastid is 24%, similar to the 25% prediction of membrane proteins in the total Arabidopsis proteome. Manual evaluation of this discrepancy between TMHMM and the accessions listed in Aramemnon for the plastid predicted proteome suggests that many of these discrepancies represent soluble proteins. This is likely resulting from assignment of a significant percentage of lumenal transit peptides and cTPs as TMDs by a number of predictors used in Aramemnon.
Two strategies were used to predict lTPs in both the predicted membrane and soluble proteins. Interestingly, the predicted membrane proteins with lTP have a low Cys content relative to the other predicted subproteomes, suggesting that a significant population was indeed located in the thylakoid. Otherwise, the prediction of the lumenal proteome still seems to be difficult, even when including the experimentally derived filters and considering the fairly large training set (>200 proteins) used for developing LumenP. As mentioned before, only the lumenal proteins with a TAT motif could be fairly well predicted.
We conclude that analysis of curated plastid subproteomes from existing literature suggests a striking difference in Cys content between integral inner envelope membrane proteins and integral and lumenal thylakoid proteins, in addition to significant differences in protein length, pI, and TMD distribution. Analysis of these assembled thylakoid and inner envelope membrane proteomes did not reveal an obvious sorting signal for either membrane system. Despite partial success, suborganellar location prediction is still in its infancy. More large-scale experimental identifications of subproteomes from different subcellular locations are needed to improve subcellular localization scheme predictions.
Collection and Curation of Experimental Subproteomes The literature and public databases were carefully screened for plastid proteins for which the plastid sublocalization was determined. All proteins were then evaluated as being an integral membrane protein. This was mostly based on experimental data (e.g. not extractable with urea or salts; x-ray structure, or topology mapping by proteolysis, etc.) and, in some cases, based on prediction of TMDs by TMHMM or the consensus prediction reported in the Aramemnon database.
Proteins believed to possess a cTP were extracted using TargetP (accepting all reliability classes; Emanuelsson et al., 2000
Molecular weight, pI, GRAVY, and amino acid composition were calculated using the Emboss software suite (Rice et al., 2000
The positive training set for TargetP consisted of a set of 141 confirmed chloroplast proteins from different higher plant species (data set available at http://www.cbs.dtu.dk/services/TargetP/datasets/datasets.php). A BLAST search was carried out to find the Arabidopsis orthologues for each of these 141 proteins; E-values were between 9.10–9 and 0, with 90% below 10–40, indicating that homologs were found for all 141 proteins. Due to one-to-many or many-to-many homology relationships, the actual number of Arabidopsis homologs to these 141 proteins was 113 (see Supplemental Table I). Only 15 of these 113 proteins were part of the most conservative curated lumenal, thylakoid membrane, and inner envelop set (1 in the lumenal set, 13 in thylakoid membrane set, and 1 in the inner envelope).
The database engine for the PPDB (http://ppdb.tc.cornell.edu/) is an MS SQL Server. The Web interface for the PPDB is developed on ASP.NET platform using C# language. The functional domain prediction was based on PFAM analysis, with a cutoff E-value at 0.1 (Bateman et al., 2002
We thank members of the van Wijk lab for helpful discussions and corrections and Prof. Gunnar von Heijne for comments and support. Rainer Schwacke is gratefully acknowledged for sending an Aramemnon data set and for helpful discussions. Received February 13, 2004; returned for revision March 25, 2004; accepted April 14, 2004.
1 This work was supported by the National Science Foundation (MCB no. 0090942) and NYSTAR (grant to K.J.v.W.). All large-scale data collection at Cornell was conducted using the resources of the Cornell Theory Center, which receives funding from Cornell University, New York State, federal agencies, foundations, and corporate partners. 
2 Present address: Molecular Biophysics and Biochemistry Department, Yale University, New Haven, CT 06520.
[w] The online version of this article contains Web-only data. www.plantphysiol.org/cgi/doi/10.1104/pp.104.040717. * Corresponding author; email kv35{at}cornell.edu; fax 607–255–7979.
Abdallah F, Salamini F, Leister D (2000) A prediction of the size and evolutionary origin of the proteome of chloroplasts of Arabidopsis. Trends Plant Sci 5: 141–142[CrossRef][ISI][Medline] Adamska I (2001) The Elip family of stress proteins in the thylakoid membrane of pro- and ukaryota. In E-M Aro, B Anderson, eds, Regulation of Photosynthesis, Vol 11. Kluwer Academic Publishers, London, UK Aebersold R, Mann M (2003) Mass spectrometry-based proteomics. Nature 422: 198–207[CrossRef][Medline] Barber J (2002) Photosystem II: a multisubunit membrane protein that oxidises water. Curr Opin Struct Biol 12: 523–530[CrossRef][ISI][Medline]
Bateman A, Birney E, Cerruti L, Durbin R, Etwiller L, Eddy SR, Griffiths-Jones S, Howe KL, Marshall M, Sonnhammer EL (2002) The Pfam protein families database. Nucleic Acids Res 30: 276–280 Ben-Shem A, Frolow F, Nelson N (2003) Crystal structure of plant photosystem I. Nature 426: 630–635[CrossRef][Medline] Bjellqvist B, Basse B, Olsen E, Celis JE (1994) Reference points for comparisons of two-dimensional maps of proteins from different human cell types defined in a pH scale where isoelectric points correlate with polypeptide compositions. Electrophoresis 15: 529–539[CrossRef][ISI][Medline] Bruce BD (2000) Chloroplast transit peptides: structure, function and evolution. Trends Cell Biol 10: 440–447[CrossRef][ISI][Medline] Chou ML, Fitzpatrick LM, Tu SL, Budziszewski G, Potter-Lewis S, Akita M, Levin JZ, Keegstra K, Li HM (2003) Tic40, a membrane-anchored co-chaperone homolog in the chloroplast protein translocon. EMBO J 22: 2970–2980[CrossRef][ISI][Medline] Claros MG, Brunak S, vonHeijne G (1997) Prediction of N-terminal protein sorting signals. Curr Opin Struct Biol 7: 394–398[CrossRef][ISI][Medline]
Emanuelsson O (2002) Predicting protein subcellular localisation from amino acid sequence information. Brief Bioinform 3: 361–376 Emanuelsson O, Nielsen H, Brunak S, von Heijne G (2000) Predicting subcellular localization of proteins based on their N-terminal amino acid sequence. J Mol Biol 300: 1005–1016[CrossRef][ISI][Medline] Emanuelsson O, von Heijne G (2001) Prediction of organellar targeting signals. Biochim Biophys Acta 1541: 114–119[Medline]
Ferro M, Salvi D, Brugiere S, Miras S, Kowalski S, Louwagie M, Garin J, Joyard J, Rolland N (2003) Proteomics of the chloroplast envelope membranes from Arabidopsis thaliana. Mol Cell Proteomics 2: 325–345
Ferro M, Salvi D, Riviere-Rolland H, Vermat T, Seigneurin-Berny D, Grunwald D, Garin J, Joyard J, Rolland N (2002) Integral membrane proteins of the chloroplast envelope: identification and subcellular localization of new transporters. Proc Natl Acad Sci USA 99: 11487–11492 Frand AR, Cuozzo JW, Kaiser CA (2000) Pathways for protein disulphide bond formation. Trends Cell Biol 10: 203–210[CrossRef][ISI][Medline]
Friso G, Giacomelli L, Ytterberg AJ, Peltier JB, Rudella A, Sun Q, Wijk KJ (2004) In-depth analysis of the thylakoid membrane proteome of Arabidopsis thaliana chloroplasts: new proteins, new functions, and a plastid proteome database. Plant Cell 16: 478–499 Froehlich JE, Wilkerson CG, Ray WK, McAndrew RS, Osteryoung KW, Gage DA, Phinney BS (2003) Proteomic study of the Arabidopsis thaliana chloroplastic envelope membrane utilizing alternatives to traditional two-dimensional electrophoresis. J Proteome Res 2: 413–425[CrossRef][ISI][Medline] Gavel Y, Steppuhn J, Herrmann R, von Heijne G (1991) The ‘positive-inside rule’ applies to thylakoid membrane proteins. FEBS Lett 282: 41–46[CrossRef][ISI][Medline] Giles NM, Giles GI, Jacob C (2003) Multiple roles of cysteine in biocatalysis. Biochem Biophys Res Commun 300: 1–4[CrossRef][Medline] Gygi SP, Rist B, Gerber SA, Turecek F, Gelb MH, Aebersold R (1999) Quantitative analysis of complex protein mixtures using isotope-coded affinity tags. Nat Biotechnol 17: 994–999[CrossRef][ISI][Medline] Inaba T, Li M, Alvarez-Huerta M, Kessler F, Schnell DJ (2003) atTic110 functions as a scaffold for coordinating the stromal events of protein import into chloroplasts. J Biol Chem Jansson S (1999) A guide to the Lhc genes and their relatives in Arabidopsis. Trends Plant Sci 4: 236–240[CrossRef][ISI][Medline]
Koo AJ, Ohlrogge JB (2002) The predicted candidates of Arabidopsis plastid inner envelope membrane proteins and their expression profiles. Plant Physiol 130: 823–836 Krogh A, Larsson B, von Heijne G, Sonnhammer EL (2001) Predicting transmembrane protein topology with a hidden Markov model: application to complete genomes. J Mol Biol 305: 567–580[CrossRef][ISI][Medline] Kyte J, Doolittle RF (1982) A simple method for displaying the hydropathic character of a protein. J Mol Biol 157: 105–132[CrossRef][ISI][Medline] Liu Z, Yan H, Wang K, Kuang T, Zhang J, Gui L, An X, Chang W (2004) Crystal structure of spinach major light-harvesting complex at 2.72 A resolution. Nature 428: 287–292[CrossRef][Medline]
Moller S, Croning MD, Apweiler R (2001) Evaluation of methods for the prediction of membrane spanning regions. Bioinformatics 17: 646–653 Mori H, Cline K (2001) Post-translational protein translocation into thylakoids by the Sec and DeltapH-dependent pathways. Biochim Biophys Acta 1541: 80–90[Medline] Nakai K (2000) Protein sorting signals and prediction of subcellular localization. Adv Protein Chem 54: 277–344[ISI][Medline] Nielsen H, Engelbrecht J, Brunak S, von Heijne G (1997) A neural network method for identification of prokaryotic and eukaryotic signal peptides and prediction of their cleavage sites. Int J Neural Syst 8: 581–599[CrossRef][Medline] Nielsen H, Krogh A (1998) Prediction of signal peptides and signal anchors by a hidden Markov model. Proc Int Conf Intell Syst Mol Biol 6: 122–130[Medline]
Ott CM, Lingappa VR (2002) Integral membrane protein biosynthesis: why topology is hard to predict. J Cell Sci 115: 2003–2009
Peltier JB, Emanuelsson O, Kalume DE, Ytterberg J, Friso G, Rudella A, Liberles DA, Soderberg L, Roepstorff P, von Heijne G, et al (2002) Central functions of the lumenal and peripheral thylakoid proteome of Arabidopsis determined by experimentation and genome-wide prediction. Plant Cell 14: 211–236
Raikhel NV, Coruzzi GM (2003) Achieving the in silico plant. Systems biology and the future of plant biological research. Plant Physiol 132: 404–409 Rice P, Longden I, Bleasby A (2000) EMBOSS: the European Molecular Biology Open Software Suite. Trends Genet 16: 276–277[CrossRef][ISI][Medline] Robinson C, Thompson SJ, Woolhead C (2001) Multiple pathways used for the targeting of thylakoid proteins in chloroplasts. Traffic 2: 245–251[CrossRef][ISI][Medline]
Robinson C, Woolhead C, Edwards W (2000) Transport of proteins into and across the thylakoid membrane. J Exp Bot 51 Spec No: 369–374
Schleiff E, Eichacker LA, Eckart K, Becker T, Mirus O, Stahl T, Soll J (2003) Prediction of the plant beta-barrel proteome: a case study of the chloroplast outer envelope. Protein Sci 12: 748–759
Schneider T, Stephens RM (1990) Sequence logos: a new way to display consensus sequences. Nucleic Acids Res 18: 6097–6100
Schwartz R, Ting CS, King J (2001) Whole proteome pI values correlate with subcellular localizations of proteins for organisms within the three domains of life. Genome Res 11: 703–709 Sipos L, von Heijne G (1993) Predicting the topology of eukaryotic membrane proteins. Eur J Biochem 213: 1333–1340[ISI][Medline] Stroebel D, Choquet Y, Popot JL, Picot D (2003) An atypical haem in the cytochrome b(6)f complex. Nature 426: 413–418[CrossRef][Medline]
Tu CJ, Peterson EC, Henry R, Hoffman NE (2000) The L18 domain of light-harvesting chlorophyll proteins binds to cpSRP43. J Biol Chem 275: 13187–13190 van Wijk KJ (2004) Chloroplast proteomics. In D Leister, ed, Plant Functional Genomics. The Haworth Press, Inc., Book Division, Binghamton, NY VanBogelen RA, Schiller EE, Thomas JD, Neidhardt FC (1999) Diagnosis of cellular states of microbial organisms using proteomics. In Process Citation. Electrophoresis 20: 2149–2159[CrossRef][Medline] von Heijne G, Steppuhn J Hermann SG (1989) Domain structure of mitochondrial and chloroplast targeting peptides. Eur J Biochem 80: 535–545
Westerlund I, Von Heijne G, Emanuelsson O (2003) LumenP–a neural network predictor for protein localization in the thylakoid lumen. Protein Sci 12: 2360–2366 This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
RESULTS
-barrel proteins, proteins with predicted 
