| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
|
First published online October 21, 2005; 10.1105/tpc.105.035519 © 2005 American Society of Plant Biologists
Functional Differentiation of Bundle Sheath and Mesophyll Maize Chloroplasts Determined by Comparative Proteomics
|
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMETHODSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMETHODSREFERENCES |
|---|
; Nelson and Langdale, 1992). In this Kranz anatomy, each vein is surrounded by a ring of bundle sheath (BS) cells, followed by one or more concentric files of mesophyll (M) cells. BS cells have thick cell walls and contain centrifugally arranged chloroplasts with large starch granules and unstacked thylakoid membranes, whereas the M cells contain randomly arranged chloroplasts with stacked thylakoids and little or no starch (Edwards and Walker, 1983). Each cell type accumulates a distinct set of C4 photosynthetic enzymes (reviewed in Langdale, 1998; Sheen, 1999).
Numerous studies have focused on the analysis of BS and M cell differentiation and the relationship to formation of the Kranz anatomy (Edwards et al., 2004). Transcript accumulation studies in BS and M cells using in situ hybridization have been successful in determining spatial patterns and developmental gradients for several specific genes in maize (Furumoto et al., 2000) and in sorghum (Sorghum bicolor; Wyrich et al., 1998). Signals required for plastid-specific gene expression of C4 enzymes involve promoter sequences and other DNA regulatory elements (Sheen, 1999). Collectively, these and other studies show that M cells in maize leaves develop in a C3 pattern by default and in a C4 pattern through the influence of closely neighboring veins. Specialization of BS cells seems to be a combination of their procambial lineage and their vein adjacent position (Smith et al., 1996; Jankovsky et al., 2001). The golden 2 nuclear transcription factor seems to play an important role to promote BS chloroplast development and plastid-specific protein accumulation (Cribb et al., 2001). Several M/BS differentiation mutants were analyzed, such as bundle sheath defective1 (bsd1) (Langdale and Kidner, 1994; Hall et al., 1998) and bsd2 (Roth et al., 1996; Brutnell et al., 1999).
The fully differentiated BS and M chloroplasts each accumulate a distinct set of photosynthetic enzymes and proteins that enables them to cooperate in carbon fixation (Hatch and Osmond, 1976; Edwards and Walker, 1983). For instance, ribulose-1,5-bis-phosphate carboxylase/oxygenase (Rubisco) (RBCL and RBCS) accumulates exclusively in BS chloroplasts, whereas pyruvate phosphate dikinase (PPDK) and photosystem II accumulate in M chloroplasts. Differential protein accumulation of several other chloroplast proteins has been determined using chloroplast fractionation techniques, in combination with protein gel blot analysis, in situ immunolocalization, and green fluorescent protein fusions (reviewed in Edwards et al., 2001). In addition, several BS- and M-specific enzymatic activities were determined using nonaqueous fractionation techniques and pulse labeling techniques (Slack et al., 1969). Collectively, these biochemical studies showed that M cells carry out linear photosynthetic electron transport that produces the ATP and NADPH needed for autotrophic growth, to import pyruvate from the BS cells to generate phosphoenolpyruvate, and to convert the C4 acid oxaloacetate into the C4 acid malate. In complementary fashion, BS cells import and decarboxylate malate into pyruvate, using the released NADPH and CO2 to generate reduced carbohydrates in the Calvin cycle and convert the surplus of carbohydrates into starch. Additional NADPH is imported from the M cells via the triosphosphate shuttle. Differentiation of many other chloroplast functions (reviewed in Neuhaus and Emes, 2000), such as synthesis of fatty acids, nucleotides, hormones, and amino acids, is largely unknown in C4 plants.
The high mass accuracy, sensitivity, and high-throughput possibilities of mass spectrometry (MS), together with the availability of genomes, now allow the rapid identification of large sets of proteins (Aebersold and Mann, 2003; Ferguson and Smith, 2003; Lin et al., 2003). Proteomes of different organelles, membranes, and tissues of Arabidopsis thaliana and Medicago truncatula have been analyzed to various degrees, providing a good starting point for many functional studies (e.g., Watson et al., 2003; Friso et al., 2004; Heazlewood et al., 2004; Nuhse et al., 2004; Peltier et al., 2004a; Shimaoka et al., 2004). The large EST assembly of maize and recent efforts in sequencing of the maize genome (Gomez et al., 2002; Palmer et al., 2003) now facilitate meaningful studies on the maize proteome (Porubleva et al., 2001; Hochholdinger et al., 2004; Lonosky et al., 2004).
Although cataloguing proteomes is useful, in the context of BS and M chloroplast differentiation, it is most important to compare protein accumulation in a quantitative manner. Initially, strategies based on two-dimensional gel electrophoresis (2-DE) were the main tool for comparative protein analysis. Examples of comparative 2-DE–based analysis in plants are protein expression profiling during seed filling and germination in Arabidopsis, M. trunculata, soybean (Glycine max), and pea (Pisum sativum; Gallardo et al., 2001, 2003; Schiltz et al., 2004; Hajduch et al., 2005). Very recently, several non-gel-based techniques have been developed that are based on differential labeling of proteomes with stable isotopes, followed by quantification in the mass spectrometer (Goshe and Smith, 2003; Ong et al., 2003). These non-gel-based methods allow better quantification and in principle much higher throughput analysis. In particular, the cleavable isotope coded affinity tag technique (cICAT) has been successfully employed for comparative proteome analysis (Hansen et al., 2003; Li et al., 2003; Tao and Aebersold, 2003; Conrads et al., 2004). However, these novel stable isotope techniques have not yet been applied in the analysis of plant proteomes.
We present here a systematic overview and a quantitative comparison of the soluble stromal proteomes of BS and M chloroplasts. The availability of the EST assembly and partial genome sequence allowed identification of specific members of protein families in many cases. This is important since different homologs within a protein family might have different functions and expression patterns, as will be shown in this study. The soluble, stromal proteomes of purified BS and M chloroplasts from young maize leaves were compared by a combination of 2-DE and image analysis, followed by identification of proteins by MS. To improve dynamic resolution, accuracy, and significance, we also used two non-gel-based techniques: a differential labeling with cICAT and a comparative analysis of unlabeled BS and M stromal proteomes by liquid chromatography electrospray mass spectrometry (LC-ESI-MS). In total, 
400 accessions in the EST unigene assembly were identified, providing an excellent overview of the stromal chloroplast proteomes in maize. Differential expression was determined for 125 chloroplast proteins, providing numerous novel insights in BS and M functional differentiation. These data are available via our Plastid Proteome Database (PPDB) at http://ppdb.tc.cornell.edu/ and linked to homologous plastid proteins in Arabidopsis and rice (Oryza sativa).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMETHODSREFERENCES |
|---|
2 h after the onset of the light period. We used a mechanical procedure for purification of M and BS chloroplasts, following suggestions from A. Barkan and R. Bassi and protocols adapted from Edwards and Black (1971) and Kannangara et al. (1977). This procedure is based on the much higher resistance of the BS cell wall to mechanical grinding as compared with the M cell wall. After BS and M chloroplast purification, both chloroplast types were broken by lysis and the membrane fractions removed by ultracentrifugation. The purity of M and BS chloroplast fractions was determined from stained SDS-PAGE gels using the abundant marker proteins Rubisco (RbcL at 54 kD) and PPDK (at 97 kD) for BS and M chloroplasts, respectively (Figure 1A). The one-dimensional gel electrophoresis (1-DE) analysis of BS and M purified stroma also clearly visualized additional BS/M-specific proteins, such as fructose biphosphate aldolase (FBA) and RBCS. This showed that our preparations were of high quality. For the analysis in this article, nine biologically independent preparations (pairs of BS and M) were used with cross-contamination levels between 5 and 15% but without any significant nonchloroplast contaminations. The cross-contamination of the BS chloroplasts by the M fraction was lower than we calculated, since it was shown that 10% of a light-independent PPDK form can accumulate in BS cells (Aoyagi and Nakamoto, 1985; Sheen and Bogorad, 1987a, 1987b). We deliberately did not use other types of BS/M preparations involving long enzymatic digestions of cell walls, likely resulting in unwanted induction of stress proteins.
|
5% BS stroma (Figure 1B). A corresponding stained protein profile (Ponceau Red on membrane) of these protein samples is shown as control for protein loading and to demonstrate the abundance of RBCL and PPDK in the maize leaf proteome (Figure 1C).
Overview and Outline of the Comparative Proteome Analysis of BS and M Chloroplast Stroma
Quantitative comparison of BS and M chloroplast proteomes is challenging because (1) obtaining separated BS and M intact chloroplasts is not trivial, (2) the maize genome has not been fully sequenced, (3) the maize genome is replete with chromosomal duplications and repetitive DNA, and (4) the stromal chloroplast proteomes are dominated by a small set of proteins involved in primary carbon metabolism. Therefore, we used three complementary proteomics techniques to overcome the genome/proteome complexity, to maximize the accuracy of protein quantifications, and to maximize the dynamic resolution. In total, nine independent BS/M chloroplast preparations were analyzed. The three techniques involved comparative 2-DE gels (n = 5) and two non-gel-based comparative proteomics techniques using either cICAT (n = 2) or parallel LC-MS runs of unlabeled M and BS stromal proteomes (n = 2) (assigned in this study as parallel LC-MS), as outlined in Figure 2. Quantification of proteins was done by extensive image analysis in case of the 2-DE gel analysis and from integrated peak areas (extracted single ion chromatograms) in case of the non-gel-based methods.
|
Comparative BS and M Stromal Proteome Analysis by 2-DE
Representative 2-DE gels of BS and M stroma are shown in Figure 3A. Quantitative data are indicated for a few selected spots as an example of how quantification data were processed (Figure 3B). A complete listing for all quantified spots with bar diagrams for each spot can be found in Supplemental Table 2 online. Corresponding interactive gels with protein spot identities that were quantified are available through PPDB.
|
; Peltier et al., 2004a) (see Supplemental Table 2 online).
Non-Gel-Based Quantification Using cICAT and Parallel Ion Chromatograms
The cICAT experiments identified 305 proteins (ZmGI) in BS and/or M chloroplasts (see Supplemental Table 1 online), of which 59 proteins could be quantified (see Supplemental Table 3 online). In total, 100 proteins (ZmGI) were identified with parallel LC-MS, with 69 proteins in BS stroma and 75 proteins in M stroma (see Supplemental Table 1 online). Forty-three of those proteins were identified in both plastid types and expression ratios could be quantified, while 19 proteins were only identified in BS and 31 only in M. All peptide sequences with BS:M ratios from both approaches can be found in Supplemental Table 3 online and are also available via PPDB, as extractable tables, as well as on the protein report page for each accession.
Analysis of the Collective Data Set: Identification, Quantification, and Function
The data from the three methods described above were merged. Collectively, 400 ZmGI accessions were identified from searches against ZmGI, and a Venn Diagram for identified proteins by the three methods is shown in Figure 4A. Clearly, the methods were complementary in their identification success rates, with the cICAT experiments identifying most accessions (Table 1). In total, 327 proteins were identified by searching against the maize genome sequences (AZM), and 277 by homology-based searches against the predicted rice proteome (OsGI) (Table 1). A complete listing of all identified ZmGI, AZM, and OsGI accessions can be found in Supplemental Table 1 online.
|
|
) as a basis. BLAST alignments of proteins identified in maize resulted in a set of Arabidopsis homologs, which we classified into 25 MapManBins covering a wide range of pathways and functions. This includes primary carbon metabolism and photosynthesis (Calvin cyle, oxidative phosphopentose pathway [OPPP], glycolysis, and starch metabolism) totaling 16%, amino acid metabolism (6%), lipid metabolism (3%), and nitrogen and sulfur assimilation (2%). Proteins involved in protein synthesis, folding, and degradation represented 24% of all identified proteins (Figure 5). BS- and M-localized enzymes of the C4 shuttle and the photosynthetic electron transport chain were also found. Several enzymes in tretrapyrrole synthesis and vitamin biosynthesis were identified. Of the identified proteins, 21% were classified in a miscellaneous category with a majority of unknown proteins. Thus, the analysis covered a wide range of plastid functions (Figure 5).
|
Collectively, we were able to assign differential BS:M chloroplast expression ratios for 125 ZmGI accessions (Figure 4B). The expression ratios of 20 proteins were quantified by three methods, and 29 proteins by two methods, while the remaining 76 proteins were quantified by only one method, in most cases 2-DE gels. The expression ratios for these 125 ZmGI accessions are summarized in Table 2 and plotted on a logarithmic scale, grouped by functional class (Figures 6 to 9). Rather than calculating one ratio for each quantified protein, we chose to report all of the values obtained from each method, providing an interval of calculated expression ratios. This gives a better sense of the data set and the biological significance than presenting a single average value for each protein. All detailed quantifications are also presented in Supplemental Table 4 online. Coherent expression patterns were observed for the large majority of identified proteins.
|
|
|
|
|
Compiled Expression Profiles and Functional Differentiation of BS and M Chloroplasts
In the remaining section of Results, we present the quantitative observations in terms of functions. We extensively compare the data with published literature, in an effort to pull together comparative BS/M maize (and sorgum) data resulting from diverse methodologies, such as transcript analysis, enzymatic measurements, and others. Importantly, new information of the BS/M distribution of less-studied metabolic pathways was determined, and strong quantitative differences in proteins involved in (regulation of) plastid gene expression and biogenesis are presented. The protein profiling data are summarized in Table 2 and corresponding Figures 6 to 9
. We recommend using the protein report pages in PPDB to obtain a more interactive integration of the results.
C4 Carbon Shuttle, Carbon Fixation, OPPP, and Glycolysis
This category is composed of the most abundant stromal proteins identified on 2-DE gels, representing 34 and 65%, respectively, of the total spot volume in M and BS chloroplasts. Within the C4 carbon shuttle, we confirmed the known localization of MDH and PPDK in M chloroplasts and NADP-ME in BS. We quantified all 11 Calvin cycle enzymes, several of which overlap with the OPPP or glycolysis. Preferentially expressed in BS chloroplasts were the enzymes specific to the Calvin cycle (Rubisco [TC233714, TC234038, and TC234963], phosphoribulokinase [PRK; TC221089 and TC225896], and sedoheptulose-1,7-biphosphatase [TC239473]) and those shared between the Calvin cycle and the OPPP (ribulose 5-phosphate-3-epimerase [TC234954], ribulose-5-phosphate isomerase [TC221577], and transketolase [TC235000]). The BS preferential localization of Rubisco, ribulose-5-phosphate isomerase, PRK, FBA (TC219359 and TC219361), and ME (TC234846) do support the activity measurements of M and BS chloroplasts separated by nonaqueous fractionation (Slack et al., 1969). To our knowledge, the differential expression of ribulose 5-phosphate-3-epimerase and transketolase (involved in OPPP and Calvin cycle) has not been shown before. Rubisco activase is preferentially expressed in the BS chloroplasts, which was indirectly shown by in situ mRNA hybridization in maize (Ayala-Ochoa et al., 2004) and by differential cDNA screening in sorghum (Wyrich et al., 1998). Two homologs of triose phosphate isomerase (TC233907 and TC233912) involved in Calvin cycle and glycolysis were strongly expressed (between fivefold and ninefold) in the M chloroplasts. This settles the conflicting observations between transcript analysis in sorghum (Wyrich et al., 1998) and enzyme activity assays on total BS and M cellular extracts (Ku and Edwards, 1975).
The BS:M expression ratio of phosphoglycerate kinase (PGK) was different for two identified homologs (TC219624 and TC219625). The parallel LC quantifications were able to distinguish between the two isoforms. PGK-TC219625 was abundantly expressed in M; in each of the LC experiments, six peptides were quantified with twofold to sixfold higher accumulation in M chloroplasts. The expression of the other isoform (PGK-TC219624) was equally distributed (BS:M = 1.1) between M and BS. cICAT and 2-DE gel measurements did not distinguish between the two homologs since only shared peptides for the two PGK isoforms homologs were resolved. PGKs have a role in the reduction phase of the Calvin cycle as well as in glycolysis. The role of PGK-TC219625 is possibly to ensure high rates of triose phosphate reduction in the M chloroplast. We note that the (de)activation mechanisms for the two PGK isoforms are unknown but should be investigated; it is conceivable that the isoform more highly expressed in M chloroplasts has an activation mechanism specifically adapted to M localization. The activity of PGK was shown to be equally distributed between the M and BS chloroplasts prepared by nonaqueous chloroplast isolation (Slack et al., 1969) and by activity measurements on total cell extracts from M protoplasts and BS strands (Ku and Edwards, 1975). A subsequent article, however, suggested an increased M localization (Usuda and Edwards, 1980). These earlier conflicting reports are most likely due the activity and detection of the individual chloroplast PGK isoforms, with the background activity of the third cytoplasmic homolog.
The two non-gel-based methods consistently showed that GADPH-A (TC219897) was equally distributed between M and BS chloroplasts, while GADPH-B (TC234510) preferentially accumulated in the M chloroplast (2.5- to 4.8-fold up). The 2-DE analysis of the acidic GADPH-B also showed an increased accumulation in the M chloroplasts. Measurements of total GADPH activities showed equal distribution between M and BS chloroplasts (Slack et al., 1969; Ku and Edwards, 1975). In C3 plants, GAPDH-B can form a homotetramer in addition to formation of a heterotetramer with GAPDH-A1,2 (Baalmann et al., 1996). It is most likely that GAPDH oligomers in M maize chloroplasts differ in the GAPDH A/B protein ratio as compared with GAPDH complexes in BS chloroplasts. In C3 plants, GAPDH oligomers reversibly associate with PRK and a small CP12 linker protein to help activation of PRK (Baalmann et al., 1996; reviewed in Graciet et al., 2004). This likely regulates the carbon flow from the Calvin cycle to the oxidative pentose phosphate cycle (Tamoi et al., 2005). Interestingly, we quantified the maize CP12 (TC223290) homolog with a BS:M ratio of 4.2. The presence of CP12 in the BS fractions strongly suggests that formation of the GAPDH-CP12-PRK supercomplex takes place in the BS chloroplast (not in M chloroplast), in agreement with the role of the BS chloroplast in the Calvin cycle. The CP12 protein was not earlier detected in maize.
Starch Biosynthesis and Carbohydrate Metabolism
We found a preferential BS localization for several enzymes involved in starch synthesis in agreement with earlier observations (Spilatro and Preiss, 1987; Lunn and Furbank, 1997). The small subunit of ADP-glucose pyrophosphorylase (TC232071), the first committed step in starch synthesis, was 1.7-fold higher in BS. Isoforms of the large subunit ADP-glucose pyrophosphorylase glucose-1-phosphate adenyltransferase (TC222533 and TC242174) were also significantly higher in B chloroplasts, whereas aldose-1-epimerase (TC237704) and soluble starch synthase (TC236897) were 5- and 3.3-fold higher, respectively, in BS chloroplasts. It is important to point out, however, that we did observe starch accumulation in the M chloroplasts of older (20 d) maize leaves during chloroplast preparations (W. Majeran, unpublished data). This suggests that starch synthesis is mostly limited to BS chloroplasts when the leaf is a sink, but when leaf tissue becomes a source, M chloroplasts also have significant rates of starch synthesis.
Nitrogen Assimilation
Two nitrogen assimilation pathways can be distinguished in leaves. The primary N assimilation involves chloroplast nitrite reductase (NiR), Gln synthase 2 (GS2), ferredoxin-dependent Glu synthase (Fd-GOGAT), and Asp transaminase (AspT). Secondary N assimilation is performed by the enzyme couple Gln synthase 2 and Fd-GOGAT, which allows the reassimilation of ammonium produced as the last product of the photorespiration, in particular in C3 species (Weber and Flugge, 2002).
We were able to quantify the accumulation of four chloroplast enzymes (with one isoform each) implicated in N assimilation. Three enzymes were predominantly expressed in the M chloroplasts and are NiR (TC222347 and TC222348, 2.5- to 6-fold), Fd-GOGAT-1 (TC236410 or GLU1 in Arabidopsis, 5-fold or unique in M), and AspT (TC219944, 2- to 11-fold). GS (TC220868) synthase was identified only on 2-DE gels, and its BS:M ratio was close to one.
Several groups analyzed the localization of nitrogen assimilation in C4 plants and designated M cells as the compartment where primary N assimilation takes place (Rathnam and Edwards, 1975, 1976; Harel et al., 1977; Becker et al., 1993). However, conflicting conclusions were presented for the localization of GS and Fd-GOGAT involved in secondary N assimilation (Rathnam and Edwards, 1976; Harel et al., 1977; Becker et al., 1993, 2000). This might well have been the result of the lengthy preparations of M and BC cells using enzymatic digestion at 37C. A preferential localization of AspT activity in the M was previously described, but the authors associated it with the cytoplasmic fraction (Hatch and Mau, 1973). AspT catalyzes the fixation of ammonium on oxaloacetate to yield aspartate. It constitutes a metabolic link between N assimilation and amino acid synthesis. The increased accumulation of AspT in M chloroplasts is consistent with the fact that both the N assimilation (see above) and enzymes of amino acid synthesis pathways preferentially accumulate in M chloroplasts.
S-Assimilation
We identified one enzyme in the sulfur import pathway, the ATP-sulfurylase (TC235924, similar to Arabidopsis ATP SULFURYLASE2), which catalyzes the first step of sulfate import (sulfate fixation onto an adenyl-sulfate). Its accumulation was increased in the BS chloroplasts (twofold), which was consistent with previous localization of primary sulfur assimilation in the BS cells by activity measurements (Burgener et al., 1998). Cystein was proposed to be the transport factor for sulfur between M and BS cells (Burgener et al., 1998). A cystein synthase (TC235388 and TC235390) was slightly higher in the M chloroplast (1.6-fold), suggesting that secondary assimilation of H2S takes place in M chloroplasts.
Photorespiration
Chloroplast phosphoglycolate phosphatase (TC229733) accumulated preferentially in BS chloroplasts (BS:M = 2.6), in agreement with phosphoglycolate phosphatase activity measurements (Baldy and Cavalie, 1984). The significance and capacity of photorespiration in C4 photosynthesis in maize is, however, a matter of debate and is driven by the O2/CO2 partial pressures in BS cells. Clearly, it is lower in NADP-ME type C4 plants, such as maize, than in NAD-ME or phosphoenolpyruvate carboxykinase type C4 plants (Yoshimura et al., 2004).
Lipid Metabolism and Hormones
We quantified three proteins in lipid biosynthesis: acyl carrier protein 1 (ACP1; TC223112 and TC228772), ß-hydroxyacyl-ACP dehydratase (TC240700), and S-malonyltransferase (TC223360). We also found a lipid transfer protein 7a2b (LTP-7a2b; AI966827). ACP1 showed an equal distribution between M and BS chloroplasts. The ß-hydroxyacyl-ACP dehydratase was a low abundant spot in the 2-DE gel analysis, at the detection limit of the analysis. S-malonyltransferase was found to be preferentially expressed in the M chloroplasts (BS:M ratio 0.38), and the lipid transfer protein 7a2b showed a unique M localization.
We identified two types of enzymes, lipoxygenase 1 (LOX1) and LOX2 that are involved in the production of oxylipins, a diverse group of fatty acid derivatives that includes the hormone jasmonate and defensive metabolites (Feussner and Wasternack, 2002). LOX1 (AW157962) and LOX2 (TC234252, TC237970, and TC237971) all showed a high preference for M accumulation (fourfold up to ninefold). The high preferential accumulation of lipoxygenases in M chloroplasts has not been determined earlier and warrants further investigation.
Nucleotide Metabolism
Adenylate kinase (TC236532 and TC236534) and inorganic pyrophosphatase (InPyr; TC218860) both accumulated preferentially in the M chloroplast, with BS:M ratios of 0.3 and 0.5, respectively. Adenylate kinase removes and recycles AMP produced in the reversible PPDK reaction, while InPyr is also indirectly needed to activate PPDK through dephosphorylation. Thus, the relatively high M expression is likely to be dictated by the large flux through PPDK. InPyr also plays a role (by removal of pyrophosphate) to drive the forward reaction of plastid-localized ADP-glucose pyrophosphorylase toward starch biosynthesis. The observed expression ratios are consistent with their M localization based on activity assays in maize (Slack et al., 1969).
Tetrapyrrole Synthesis
We quantified three enzymes in the tetrapyrrole synthesis pathway implicated in three steps of the chlorophyll/heme biosynthesis pathway: (1) formation of the precursor 5-aminolevulinic acid (ALA), (2) synthesis of protoporphyrin, and (3) incorporation of Mg2+ into the porphyrin ring (for reviews, see Beale, 1999; Eckhardt et al., 2004). The formation of ALA is the limiting step for the rest of the pathway. We identified Glu-1-semialdehyde 2,1-aminomutase (TC227768), which catalyzes the rearrangement of Glu-1-semialdehyde into ALA. This stromal enzyme accumulated preferentially in M (BS:M 0.40). The subsequent enzyme, 
-aminolevulinic acid dehydratase (TC240729), which catalyzes the asymmetric condensation of two ALA to form porphobilinogen, also accumulated at a somewhat higher level in M chloroplasts. We identified the Mg-chelatase subunit I (TC223168), similar to Mg-chelatase I-2 from Arabidopsis (At5g45930.1), several enzymatic steps further downstream in the chlorophyll biosynthesis pathway. This enzyme showed a strong preferential M accumulation (BS:M 0.19), likely reflecting the higher demand of chlorophylls due to photosystem II (PSII) accumulation.
Defense
Hydroxamic acids (Hx) possess antifungal and insecticidal properties and are abundantly accumulating in rye (Secale cereale), wheat (Triticum aestivum), and maize cell walls as a glucoside-linked precursor (Hx-glc) (Niemeyer, 1988). Upon tissue damage, the ß-D-glucosidases come in contact with Hx-glc and release the Hx that can be converted into a benzoxalolinone. IMBOA (2-O-ß-D-glucopyranosyl-4-hydroxyl-7-methoxy-1,4-benzoxazin-3-one) is the main Hx-glc in maize and wheat.
Each of the three quantification methods used in this study showed that ß-D-glucosidase (TC220472) accumulates at much higher levels in BS chloroplasts (BS:M ratio >6.7). This is in agreement with previous observations showing that ß-D-glucosidases were found in maize plastids and proplastids of the BS cells (Nikus and Jonsson, 1999; Nikus et al., 2001). Although the major role of ß-D-glucosidase was attributed to defense, it was also suggested to play a role in development and growth regulation, since transient expression of maize ß-D-glucosidase in tobacco (Nicotiana tabacum) protoplasts was able to release cytokinin from an inactive cytokinin-O-glucoside (Brzobohaty et al., 1993).
Proteins Involved in Redox Regulation
Ferredoxins are not only essential in linking thylakoid-localized photosynthetic electron transport to NADPH production via Fd-NADP+ reductases (FNR), but also provide redox equivalents to the different thioredoxins (Trxf, -m, -x, and -y) via Fd-Trx reductase. These different Trx regulate several plastid-localized pathways by changing the redox state of thiol groups (S-S<
2SH) (Buchanan and Balmer, 2005).
We identified and quantified both Fd1 (TC220059 and TC238105) and Fd2 (TC223586). Fd1 accumulated predominantly in the M chloroplast (twofold to fivefold), while Fd2 accumulated predominantly in the BS chloroplast (threefold to fourfold). This is in agreement with transcript analysis of Fd1 and Fd2 (Matsumura et al., 1999; Furumoto et al., 2000). Preferential accumulation of Fd2 in BS chloroplasts was shown by immunoblots, while Fd1 was found to be equally distributed over both chloroplast types (Kimata and Hase, 1989). It seems that Fd1 and Fd2 have evolved to interact with different partners. They possess similar redox potentials but have different Km values for interaction with FNR, suggesting different cellular functions. Indeed, expression of maize Fd1 in Fd-deficient strains of cyanobacteria (Plectonema boryanum) led to a high light sensitive phenotype, while the strain expressing Fd2 showed a nitrogen deficiency phenotype (Kimata-Ariga et al., 2000).
We identified and quantified BS and M expression ratios for three FNR identifiers: TC219223, TC219224, and TC219226. All three sequences were similar to Arabidopsis FNR-2. All of them (FNR: TC219223, TC219224, and TC219226) showed a preferential localization in M chloroplasts (from twofold up to sevenfold). This is consistent with previous M localization of an FNR protein by rocket electrophoresis (Broglie et al., 1984).
The Trx play an important role in redox regulation (activation and deactivation) of many biosynthetic and other activities in plastids (Balmer et al., 2004). The distribution of the different Trx between the M and BS chloroplasts has not been reported earlier, which is somewhat surprising given the central role of the Trx in regulating chloroplast function. Based on their closest homologs in Arabidopsis, we identified and quantified Trx-m2, Trx-m4, and Trx-f2. For Trx-m4 (TC221334) and Trx-m2 (TC228810), only common peptides were identified showing their preferential accumulation in the M chloroplast (2.6 to 4). In the 2-DE spot 348 (only present in M stroma), Trx-f2 (TC220464) was identified; however, its quantification could not be exclusively determined as it comigrated with Trx-m4 (TC221334). The non-gel-based methods quantified unique peptides for two other Trx-m4 isoforms (TC238760 and TC224103), which showed preferential accumulation in M (BS:M ratio 0.45) and BS (BS:M ratio 2.4), respectively (Table 2; see Supplemental Table 4 online).
Proteins Involved in Detoxification of Reactive Oxygen Species
Chloroplasts have an elaborate enzymatic system to detoxify different reactive oxygen species (ROS), such as superoxide, hydrogen peroxide, and lipid peroxides. M chloroplasts have high rates of linear photosynthetic electron transport and high production rates of molecular oxygen through the water-splitting activity of PSII, whereas BS chloroplasts only carry out cyclic electron flow. It is thus quite likely that there are substantial differences in the quantity and quality of ROS in these two chloroplast types. However, differential expression levels of the BS and M chloroplast ROS defense systems are unknown. We quantified the BS/M chloroplast expression of Cu-Zn superoxide dismutase (SOD), glutathione disulfide reductase (GR), and several peroxiredoxins (Prx). GR (TC227052) and Cu-Zn SOD (TC237182) each accumulated in M and BS chloroplasts. We observed an equal distribution of GR over both plastid types. Both gel-based and non-gel-based analysis suggested that the BS:M expression ratio of CuZn SOD (TC237182) was between 0.4 and 0.7.
Several studies tried to measure SOD activity in BS and M maize cells (but not in purified chloroplasts), with inconsistent results between the different studies (Foster and Edwards, 1980; Doulis et al., 1997). Possibly this is due to the presence of different SOD homologs and different isolation procedures. Comparisons between the total maize leaf extracts and rapid M sap extractions showed that glutathione (Burgener et al., 1998), dehydroascorbate reductase, and GR were localized in the M, while ascorbate peroxidase was found in the total leaf extracts (Doulis et al., 1997). As GR mRNA was distributed in both compartments, a translational regulation mechanism of differential expression was proposed (Pastori et al., 2000).
Prx are Trx-dependent peroxidases that have a capacity to scavenge peroxides. They constitute an antioxidant system that operates in parallel to the ascorbate-glutathione system (reviewed in Dietz, 2003). In Arabidopsis, four types of Prx are targeted to plastids (Prx IIE, PrxQ, 2-Cys PrxA, and 2-Cys PrxB). We quantified the expression of maize EST products 2-CysPrx-TC234345 and 2-CysPrx-TC234346, which both showed high degrees of similarity to Arabidopsis 2-Cys PrxB (At5g06290.1), and TC223042, which was closest to Prx IIE (At3g52960.1). Both 2-Cys Prx (TC234345 and TC234346) and Prx IIE (TC223042) accumulated preferentially in M chloroplasts (1.8- up to 7-fold). The 2-Cys Prx were also identified on 2-DE gels in a small train of spots (spots 362, 459, and 125), migrating in the region between 47 and 53 kD at about twice the expected molecular mass (of 26 kD). These spots represented likely 2-Cys Prx dimers, despite the denaturing conditions of the 2-DE gel. The basic oligomeric state of Prx is a dimer that under reducing conditions can assemble into larger 2-Cys Prx oligomers (Konig et al., 2002). To our knowledge, the distribution of the different Prx homologs over M and BS chloroplasts was not determined in published literature. The preferential accumulation of Prx proteins in the M chloroplasts is consistent with the high linear electron flow and production of active oxygen species, such as H2O2 (Doulis et al., 1997) in M chloroplasts. A wide range of electron donors has been shown to interact with Prx. Some Prx can accept electrons from both Trx and glutaredoxins (Rouhier et al., 2001), whereas others can only use Trx as electron donor. It is interesting to note that the preferential expression of Prx (TC223042, TC234346, and TC234345) in M is correlated with the expression in M chloroplasts of two isoforms of Trx (TC238760 and TC221334) and one glutaredoxin (TC225992, 3.5-fold).
Chloroplast Chaperones, Isomerases, and Proteases
The two major housekeeping chloroplast chaperone families (Cpn60/20 and Hsp70/GrpE) were relatively abundant on the 2-DE gels. The Cpn60/20 complex is composed of three types of proteins, namely Cpn60
and Cpn60ß (homologs to Escherichia coli GroEL) and Cpn20 (homolog to E. coli GroES). The identified Cpn60 homologs (TC220350, TC235184, and TC236031), Cpn60-ß (TC219523 and TC219522), and Cpn20 (TC233810 and TC236117) all showed an equal distribution between M and BS chloroplasts, with BS:M ratios between 1 and 1.4. This clearly shows that the Cpn60/20 chaperone system is not exclusively functioning in Rubisco assembly (since Rubisco is absent in M chloroplasts).
We identified two highly homologous HSP70 homologs (TC220215 and TC235785). In the cICAT and 2-DE gel analysess, we were not able to distinguish the BS:M expression ratio for each of the HSP70 proteins individually. Collectively, their BS:M expression ratio ranged between 0.9 and 5.8, depending on the quantification method and protein spot. However, in the parallel LC-based quantifications, we were able to determine the BS:M expression ratios for each of the two homologs. This showed an equal distribution for HSP70-TC220215 and a strong (sixfold) preferential expression of HSP70-TC235785 in the M chloroplasts. It is noteworthy that one of the interacting partners of HSP70, the GrpE (TC222867) also showed a fivefold preferential M expression. This result is consistent with a previous sorghum cDNA library screening where HSP70 expression ranged from 1 up to 10 in M cells (Wyrich et al., 1998). This result suggests that besides the expression of chaperones in both plastid types as housekeeping proteins, HSP70-TC235785 seems to have a specific function in M chloroplasts.
Proteome analysis of chloroplasts in Arabidopsis showed accumulation of a relatively large number of protein peptidyl-prolyl cis-trans isomerases of different categories, many of which seem localized in the thylakoid lumen (Peltier et al., 2002; Schubert et al., 2002). Here, we identified and quantified BS:M expression ratios of three of these lumenal isomerases, apparently released into the chloroplast stroma. Tlp21-TC228225 and Tlp40-TC239826, homologs to the cyclophylins Tlp21 and Tlp40, both showed a BS:M expression ratio of 0.5 in the 2-DE gel experiments. FK506-TC247336, a homolog of Arabidopsis At5g45680 assigned FKBP13, was preferentially expressed in the BS chloroplasts (BS:M = 5). FKBP13 was found in yeast two-hybrid experiments as interacting with the cytochrome b6f complex Rieske subunit (Gupta et al., 2002).
Interestingly, we also quantified the expression ratio of ClpC-TC235372, a homolog of the HSP100 ClpC chaperone family. The BS:M expression ratio was 0.4. ClpC1 in Arabidopsis accumulates in the stroma likely delivering substrate to the Ser-type ClpP/R protease complex, while an envelope-associated form is involved in protein import. We also quantified two ClpP proteins (TC245457 and TC225197) with BS:M ratios ranging from 0.3 to 0.59. The alignment of TC245457 and TC225197 with the members of the ClpP/R family in Arabidopsis did not identify clear-cut homologs.
Plastid Protein Expression
Both BS and M chloroplasts contain the same plastid genome (with 104 genes), encoding for four rRNAs and 30 tRNAs and some 70 proteins involved in plastid gene expression (e.g., ribosomal proteins, tRNA synthetases, etc.) as well as photosynthesis (RBCL and thylakoid proteins) (Maier et al., 1995). Expression of these plastid genes likely involves hundreds of nuclear-encoded proteins, each imported into the chloroplast (reviewed in Barkan and Goldschmidt-Clermont, 2000).
We quantified BS/M expression for seven plastid ribosomal proteins. Ribosomal protein L1 (TC219835) was equally expressed in both compartments, while L12.1 (TC222916), S1 (TC221841), S5 (TC239133), and L10 (TC224551) predominantly or exclusively accumulated in M chloroplasts. This suggests that 70S ribosomes in M and BS chloroplasts are different in composition, in agreement with differential accumulation of chloroplast L2 in an earlier study (Zhao et al., 1999).
Homologs to Arabidopsis plastid-specific ribosomal proteins PSRP2 (TC238798) and polyprotein of Ef-Ts (PSRP7-EF-Ts fusion protein; TC226754) were preferentially expressed in the M chloroplasts, with BS:M ratios of 0.29 and 0.14, respectively. It was proposed that PSRP2 is part of a plastid translational regulatory module on the 30S ribosomal subunit structure for the possible mediation of nuclear factors on plastid translation (Yamaguchi et al., 2000). We identified TC226754 as a 125-kD protein on the 2-DE gels. It is likely a homolog of a polyprotein of Ef-Ts gene originally described in Chlamydomonas reinhardtii as a single-copy gene containing PSRP-7 and EF-Ts sequences (Beligni et al., 2004). Translation of this gene resulted in a 110-kD pro-protein, and the majority of this protein was likely posttranslationally processed into the 65-kD protein PSRP-7 and a 55-kD EF-Ts. Our MS data for this protein covered both the N- and the C-terminal parts containing the PSRP7 and EF-Ts sequences, respectively. We did not identify the 65- and 55-kD maturation products.
We found a
