Both the stroma and thylakoid lumen of tobacco ... - Wiley Online Library

Plant Biotechnology Journal (2008) 6, pp. 46–61

doi: 10.1111/j.1467-7652.2007.00298.x

Both the stroma and thylakoid lumen of tobacco chloroplasts are competent for the formation of disulphide bonds in recombinant proteins Julia Original Disulphide Bally Articles et bonds al.Publishing in proteins Blackwell Oxford, Plant PBI © 1467-7644 XXX 2007 Biotechnology UK Blackwell Publishing Journal Ltd Ltd and chloroplasts

Julia Bally1,2, Eric Paget1, Michel Droux2, Claudette Job2, Dominique Job2 and Manuel Dubald1,* 1

Bayer BioScience, F-69263 Lyon cedex 09, France

2

Centre National de la Recherche Scientifique-Bayer CropScience Joint Laboratory, Unité Mixte de Recherche 5240, F-69263 Lyon cedex 09, France

Received 30 May 2007; revised 7 August 2007; accepted 9 August 2007. *Correspondence (fax 33 4 72 85 20 34; e-mail [email protected])

Summary Plant chloroplasts are promising vehicles for recombinant protein production, but the process of protein folding in these organelles is not well understood in comparison with that in prokaryotic systems, such as Escherichia coli. This is particularly true for disulphide bond formation which is crucial for the biological activity of many therapeutic proteins. We have investigated the capacity of tobacco (Nicotiana tabacum) chloroplasts to efficiently form disulphide bonds in proteins by expressing in this plant cell organelle a well-known bacterial enzyme, alkaline phosphatase, whose activity and stability strictly depend on the correct formation of two intramolecular disulphide bonds. Plastid transformants have been generated that express either the mature enzyme, localized in the stroma, or the full-length coding region, including its signal peptide. The latter has the potential to direct the recombinant alkaline phosphatase into the lumen of thylakoids, giving access to this even less well-characterized organellar compartment. We show that the chloroplast stroma supports the formation of an active enzyme, unlike a normal bacterial cytosol. Sorting of alkaline phosphatase to the thylakoid lumen occurs in the plastid transformants translating the full-length coding region, and leads to larger amounts and more active enzyme. These

Keywords: alkaline phosphatase, chloroplast, disulphide bond, secretion, stroma, thylakoid lumen.

results are compared with those obtained in bacteria. The implications of these findings on protein folding properties and competency of chloroplasts for disulphide bond formation are discussed.

Introduction Plants offer a suitable alternative to microbial or animal expression for the production of recombinant proteins of industrial or pharmaceutical interest. They present a number of advantages compared with traditional systems, such as anticipated lower production costs, rapid scalability, absence of human pathogens and ability to fold and assemble complex proteins accurately (Ma et al., 2003). A number of plants have been successfully transformed to produce complex structures with native conformation, such as therapeutic antibodies or complex antigens (Warzecha and Mason, 2003; Arntzen et al., 2005). Moreover, some plant species have the potential to be suitable directly for oral immunization – the edible vaccine concept (Mason et al., 2002). However, there are

46

some limitations, in particular an often low yield of the recombinant protein when expressed via nuclear transformation. As an alternative to nuclear expression, chloroplast genetic engineering has emerged as an effective tool for the expression of recombinant proteins in plants (Daniell et al., 2004; Maliga, 2004; Dubald et al., 2007). The main reason is that this system potentially allows a very high expression level of recombinant proteins, which may represent up to 46% of total soluble proteins (De Cosa et al., 2001). Additional features of interest include the maternal inheritance of the plastidial genome in most species, reducing drastically transgene dissemination via the pollen (Zhang et al., 2003), and the targeted integration of introduced DNA by homologous recombination into a defined region of the plastid genome. Therefore, the location of transgenes is predictable, gene © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd

Disulphide bonds in proteins and chloroplasts 47

expression is uniform in the selected transgenic lines that are clonal in essence, and there is no segregation of the character in the progeny. The development of this technology has revolutionized fundamental plastid research and enabled many biotechnological applications, reflecting the commercial interest in using plastids as bioreactors (Daniell et al., 2004). Plastids are eukaryotic cell organelles that, according to endosymbiotic theory, derive from cyanobacterial ancestors (Margulis, 1975; Gray, 2004). Therefore, they still exhibit many prokaryotic features, such as gene organization in operons, and most prokaryotic mechanisms of gene expression. Indeed, the two genetic systems are so similar that, in general, plastid genes are fully functional in bacteria and vice versa (Brixley et al., 1997; De Cosa et al., 2001). Owing to this similarity, and knowing the failure of normal bacterial cytosol to support the formation of disulphide bonds in proteins (Kadokura et al., 2003), the question can be raised as to whether the chloroplast stroma is able to accumulate such recombinant proteins, which represent an important class of therapeutic proteins. Over recent years, a limited number of such proteins have been successfully expressed in transgenic plastids, such as human growth hormone (Staub et al., 2000), cholera toxin B (Daniell et al., 2001), human serum albumin (FernandezSan Millan et al., 2003), a single-chain antibody (Mayfield et al., 2003) and human interferon-α (Daniell et al., 2004). All of these recombinant proteins, except human serum albumin, were shown to be biologically active, suggesting that the stroma allows the correct formation of disulphide bonds in proteins. In marked contrast, in bacteria, this process occurs outside the cytoplasm, in the periplasm, owing to appropriate redox conditions and the presence of proteinfolding catalysts such as oxidoreductases and isomerases (Kadokura et al., 2003). In plants, even less is known about the distinct compartment of chloroplasts – the lumen of thylakoids – which accumulates a very restricted set of proteins (Peltier et al., 2000). These proteins are sorted to the lumen by secretion pathways present in the thylakoid membranes, such as SEC or TAT for soluble proteins, which are similar to those functioning in bacterial protein export (Settles and Martienssen, 1998). The question of whether this particular compartment is more or less appropriate than the stroma for accumulating recombinant disulphide bond-containing proteins has not yet been addressed. In this context, it should be noted that proteomic studies suggest that prime functions of the lumenal proteome include assistance in the folding and proteolysis of thylakoid proteins, as well as protection against oxidative stress (Peltier et al., 2002). This prompted us to evaluate the ability of chloroplast stroma and lumen to support the formation of disulphide bonds in proteins. To this

end, we tested whether a well-characterized bacterial enzyme, alkaline phosphatase (PhoA), could be expressed in a biologically active form in tobacco (Nicotiana tabacum) chloroplasts. This enzyme, normally localized in the periplasmic space of Escherichia coli, is a dimer consisting of two identical subunits encoded by the phoA gene, and each of the subunits contains two intramolecular disulphide bonds that are required for stability and catalytic activity of the enzyme (Sone et al., 1997). Hence, this protein has been widely used as a model to investigate disulphide bond formation in various biological backgrounds (Derman et al., 1993; Sone et al., 1997; DeLisa et al., 2003), although not yet in plant chloroplasts. In this work, tobacco plastid transformants were generated with two constructs, allowing the expression of either a full-length coding region of phoA, including its SEC-type signal peptide (this long form of the enzyme is referred to as PhoA-L), or a shorter mature protein deleted of its signal peptide but engineered to contain an additional methionine at the NH2 terminus (this short form of the enzyme is referred to as PhoA-S). It was hypothesized that the SEC-type signal peptide in PhoA-L might be able to direct the PhoA protein into the lumen of thylakoids, and this was shown to be the case. A comparison of the activity of PhoA produced in chloroplasts with these two genetic constructs is presented, shedding light on the folding properties of the stroma and lumen compared with the results obtained in bacterial systems.

Results Design of transformation vectors The full-length (PhoA-L) and N-terminally truncated (PhoA-S) coding regions of E. coli PhoA were inserted into a tobacco chloroplast expression vector, giving constructs pCLT516 and pCLT515, respectively (Figure 1). These vectors allow the targeted integration by homologous recombination of the genes of interest into the large single-copy region of the tobacco plastid genome, between the rbcL and accD genes (Figure 1). The selection cassette (aadA gene) encodes an aminoglycoside 3′-adenylyltransferase that provides resistance to spectinomycin (Svab and Maliga, 1993). PhoA-L and PhoA-S expression is controlled by tobacco plastid regulatory elements: the psbA light-inducible promoter including the 5′-untranslated region (5′UTR) and the 3′-end of the rbcL gene (3′rbcL) (Figure 1).

Selection and screening of transformants From 20 bombarded leaves for each vector, seven and four spectinomycin-resistant events were selected for pCLT515

© 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd, Plant Biotechnology Journal, 6, 46–61

48 Julia Bally et al.

Figure 1 Recombinant plastome genetic structures. Wild-type tobacco plastome (a) and recombinant tobacco plastomes (b) after recombination with transforming vectors pCLT516 and pCLT515 and transgene integration. aadA, spectinomycin resistance gene; accD, acetyl-CoA carboxylase subunit; Prrn, tobacco 16SrDNA promoter; PsbA, tobacco psbA promoter; rbcL, Rubisco large subunit. PhoA-L and PhoA-S are the full-length and signal peptideless coding regions of Escherichia coli alkaline phosphatase, respectively. LHRR and RHRR are the left and right plastid recombination regions present in the transforming vectors, respectively. 1, 2 and 3, primers used for polymerase chain reaction analyses. The expected lengths of amplified fragments are indicated in base pairs and locate the rbcL and phoA probes used for Southern analyses.

(PhoA-S) and pCLT516 (PhoA-L), respectively. Integration of the foreign genes into the chloroplast genome was first checked by polymerase chain reaction (PCR) using a forward primer aligning on the phoA sequence and a reverse primer aligning on the accD sequence (see Figure 1). The expected amplification fragments of 957 bp and 906 bp were obtained for PhoA-L and PhoA-S transformants, respectively (Figure 2a–c). Thus, all these events have integrated the phoA transgene into the plastid genome, except for three events selected for pCLT515 (lanes 5–7 in Figure 2a). These three events most probably correspond to mutations providing resistance to spectinomycin, which are known to occur in tobacco (Svab et al., 1990). After regeneration and rooting, the transgenic

lines were transferred to the glasshouse. All lines were phenotypically normal and fertile. The T1 and T2 generations were propagated further for one transformant per construct. Progenies sown in vitro showed uniform resistance to spectinomycin, as expected for a plastid maternally inherited trait. To address the question of the homoplasmy of the transformants, PCR analysis was performed on the T0, T1 and T2 generations (Figure 2d). A forward primer aligning on the rbcL sequence and a reverse primer aligning on the accD sequence were used, so that an amplification product of 2306 bp is expected for the wild-type DNA only. Figure 2d clearly shows that this band is present in wild-type DNA, but absent from all transformants, testifying their homoplasmy.

© 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd, Plant Biotechnology Journal, 6, 46– 61


Figure 2 Confirmation of the integration of DNA constructs encoding alkaline phosphatase (PhoA) into the chloroplast genome by polymerase chain reaction (PCR) analyses. (a, b, c) Insertion of PhoA-S and PhoA-L into the chloroplast genome. Total DNA was extracted from wild-type and spectinomycin-resistant tobacco plants, and PCR was carried out using the primer pair No. 3 and No. 2, as described in ‘Experimental procedures’ and Figure 1. The 957-bp (PhoA-L) and 906-bp (PhoA-S) bands confirmed the correct integration of PhoA-S (lanes 1–4) and PhoA-L (lanes 8–11). Lanes 5–7 correspond to spectinomycin-resistant plants not having the PhoA-S construct integrated into the chloroplast genome. C, positive control corresponding to pCLT515 (a, b) and pCLT516 (c) vectors. (d) Homoplasmic PhoA-S and PhoA-L lines (T2 generation) identified with primers No. 1 and No. 2 as described in ‘Experimental procedures’ and Figure 1. Wild-type plants generated a 2306-bp product (WT lanes), whereas this band was absent for the transplastomic lines (lanes 1–4 for PhoA-S and lanes 5–8 for PhoA-L). WT, wild-type; M, 1-kb DNA ladder.

This finding was further verified by Southern blot analysis (Figure S1, see ‘Supplementary material’).

Analysis of PhoA expression in transgenic chloroplasts PhoA accumulation in leaves of transformed tobacco lines was examined in the T1 generation by proteomics and Western blotting after two-dimensional gel separation of proteins under denaturing conditions. PhoA expression was detected in both cases as multiple spots of roughly the same molecular weight, at around 45 kDa, which corresponds to the size of the PhoA monomer, but differing in isoelectric point (Figure 3). There was a substantial level of expression (Figure 3c–f ), in particular for the full-length PhoA-L protein (Figure 3e,f), as the corresponding spots are clearly visible by either silver nitrate (Figure 3c,e) or Coomassie blue (data not shown) staining of total separated proteins. All of these spots were isolated from preparative two-dimensional gels, six for

PhoA-S (spots 1–6) and 12 for PhoA-L (spots 7–18), and further characterized by liquid chromatography/tandem mass spectrometry (LC/MS-MS) analyses and Edman degradation. Mass spectrometry experiments identified all of these polypeptides as E. coli PhoA (Table S1, see ‘Supplementary material’). The N-terminal sequence of recombinant PhoA-L encoded by the full-length construct pCLT516 was established by Edman degradation in order to determine whether the bacterial signal peptide was cleaved in transgenic chloroplasts. The N-terminal sequences of four spots (7, 8, 9 and 10 in Figure 3e) were obtained (Table S1). In all cases, the N-terminal sequences were identical and started with residues FTPVTKAR, i.e. 15 amino acids after the translation start site (Figure 3g; Table S1). This specific cleavage of the bacterial signal peptide occurs, in chloroplasts, six amino acids before the normal processing site in E. coli (Figure 3g). This finding indicates that the bacterial signal peptide is recognized and cleaved by the chloroplast equivalent of the bacterial signal peptidase, known



Figure 3 Proteomic analysis of alkaline phosphatase (PhoA) accumulation in transgenic tobacco chloroplasts. Leaf protein extract (100 μg) from wildtype tobacco (a, b) and from plastid transformants pCLT515 expressing PhoA-S (c, d) and pCLT516 expressing PhoA-L (e, f) were separated by twodimensional gel electrophoresis, and the proteins were stained with silver nitrate (a, c, e) or analysed by Western blotting using anti-PhoA antibodies (b, d, f). The predicted N-terminal sequence of PhoA-L encoded by pCLT516 is shown, together with the N-terminal sequence experimentally determined from transgenic chloroplasts expressing PhoA-L (Table S1) and the normal signal peptide processing site in Escherichia coli (g). The labelled spots were identified by liquid chromatography/tandem mass spectrometry (LC/MS-MS) analysis as recombinant E. coli PhoA (Table S1).



to exist in the thylakoid membrane (Settles and Martienssen, 1998), but the processing sites differ in bacteria and chloroplasts. As expected, the N-terminal sequence of recombinant PhoA-S, encoded by pCLT515, was determined to be MRTPEM (Table S1), identical for two different spots (1 and 2 in Figure 3c). Therefore, this rules out the possibility that the occurrence of multiple spots, ascribable to recombinant PhoA in the two-dimensional gels, originates only from heterogeneity in the N-terminal part of the recombinant protein. The oxidation of methionine was noted in many internal tryptic fragments of PhoA-S and PhoA-L (Table S1), but these modifications, which can occur either in vivo or during sample processing, have no effect on the charge of this residue. Most importantly, protein oxidation (carbonylation) has been shown to result in protein stuttering, a phenomenon that can

Western blots again showed the presence of several spots differing in isoelectric point. This pattern suggests that the modifications affecting recombinant PhoA in chloroplasts (see Figure 3) are not specific to the plant organelles, although the phenomenon of protein stuttering was much less pronounced in recombinant bacteria than in recombinant plastids (cf. Figures 3 and 4). In marked contrast with the results obtained with the pCLT516 vector, the expression level of recombinant PhoA was clearly very low in the BL21 strain propagating the pCLT515 vector, which expresses the recombinant PhoA-S in the cytosol (Figure 4e,f). As reported previously (Bessette et al., 1999), the oxidizing Origami B strain allows the production of significant amounts of PhoA-S in the cytosol, this protein being detected as a major spot by silver staining (Figure 4g,h). Mass spectrometric data corre-

be detected on two-dimensional gels as satellite spots with molecular weights similar to the authentic protein, but separated from it in the isoelectric focusing dimension (Ballesteros et al., 2001). However, this phenomenon was not the explanation for the observed PhoA protein trains, as these spots were not revealed by staining with the specific reagent 2,4-dinitrophenylhydrazine (Ballesteros et al., 2001; Job et al., 2005; data not shown). Therefore, the observed protein trains result from either heterogeneity at the C-terminus of the polypeptide, or other post-translational modifications, not detected in the present LC/MS-MS analyses, which have an impact on the charge of the polypeptide (Gianazza, 1995).

sponding to the identification of recombinant PhoA in BL21 and Origami B bacterial strains transformed with the pCLT515 or pCLT516 vector are presented in Table S2.

Analysis of PhoA expression in recombinant bacteria Expression cassettes that are dependent on plastid promoters, including the psbA promoter, are, in general, functional in other prokaryotic systems (Brixley et al., 1997). This prompted us to use the plant transformation vectors directly to analyse and compare PhoA expression in recombinant bacteria. For this purpose, two different E. coli strains were used. The first, BL21, is widely employed for recombinant protein production. The second is a double null mutant trxB/gor derivative of BL21 (referred to as Origami B) which has a more oxidizing cytosol than the wild-type strain, a feature allowing the production of disulphide bond-containing proteins within cells (Prinz et al., 1997; Bessette et al., 1999). Total protein extracts were prepared from the recombinant strains transformed with pCLT515 (PhoA-S) or pCLT516 (PhoA-L) and analysed by two-dimensional polyacrylamide gel electrophoresis (PAGE) and Western blotting as above (Figure 4; Table S2, see ‘Supplementary material’). PhoA was detected in abundance for both strains transformed with pCLT516 which targets the enzyme to the periplasm (Figure 4i,k–l). The

Recombinant PhoA activity in chloroplasts and bacteria The activity of PhoA expressed in BL21 bacteria (wild-type and trxB/gor mutant) and chloroplasts was examined in vivo. Recombinant bacteria or leaf discs from transgenic tobacco lines were incubated in the presence of the PhoA substrate 5bromo-4-chloro-3-indolyl phosphate (BCIP), which leads to the formation of a blue–purple reaction product (Figure 5a,b). With regard to PhoA-S expression in bacteria transformed with pCLT515, enzyme activity above wild-type level (no staining) was only observed with the transformed Origami B trxB/gor mutant strain (Figure 5a). Periplasmic PhoA-L expressed in bacteria transformed with pCLT516 was highly active in both strains (Figure 5a). In tobacco, penetration of the substrate into the tissues does not appear to be optimal, as staining mainly occurs at the periphery of the leaf discs (Figure 5b); however, it can be seen clearly that both constructs lead to active PhoA, although staining is less pronounced with pCLT515 (Figure 5b). PhoA activity was also assayed in leaf extracts from transgenic tobacco lines (T0 and T1 material) and recombinant bacteria (BL21 and its Origami B derivative). The specific activity of the enzyme was expressed relative to enzyme measurements carried out with a commercial PhoA preparation used as standard. The results again showed that the activity of cytosolic PhoA (PhoA-S encoded by pCLT515) was low in the bacterial strain BL21 (Figure 5c), most probably as a result of the reducing conditions of the bacterial cytosol and the absence of folding catalysts (Derman and Beckwith, 1991). By contrast, a more active PhoA-S enzyme was formed in both the



Figure 4 Analysis of alkaline phosphatase (PhoA) accumulation in recombinant bacteria. Bacterial protein extract (100 μg) from wild-type BL21 strain (c, d), from bacteria transformed with pCLT515 expressing PhoA-S (e, f for BL21 strain; g, h for BL21 trxB/gor strain) and from bacteria transformed with pCLT516 expressing PhoA-L (i, j for BL21 strain; k, l for BL21 trxB/gor strain) were separated by two-dimensional gel electrophoresis, and the proteins were stained with silver nitrate (a, c, e, g, i, k) or analysed by Western blotting using anti-PhoA antibodies (b, d, f, h, j, l). The relevant regions of the silver-stained gels and corresponding Western blots shown in (a) and (b) are reproduced for all samples analysed. The labelled spots were identified by liquid chromatography/tandem mass spectrometry (LC/MS-MS) analysis as recombinant Escherichia coli PhoA (Table S2).

cytosol of Origami B bacteria and the chloroplasts of transplastomic plants (Figure 5c). However, such an activity was only partial, as it was substantially lower than that measured from plants and bacteria transformed with pCLT516 and expressing the full-length enzyme PhoA-L (Figure 5c).

Localization of recombinant PhoA in chloroplast compartments There are two non-exclusive explanations for the striking difference in the recombinant PhoA expression levels observed



Figure 5 Activity of alkaline phosphatase (PhoA) in Escherichia coli and tobacco. (a, b) In vivo activity of recombinant PhoA in E. coli strains (a) and transgenic tobacco (b). The experiment shown in (b) is representative of assays repeated four times on 30 leaf discs from transformant and wild-type lines. (c) The relative specific activity from transgenic tobacco and recombinant bacteria was expressed as percentage values vs. the data obtained with standard PhoA, and the standard error of the mean is indicated.

in tobacco plants transformed with the pCLT515 and pCLT516 constructs (Figure 3). One possibility is that, for the recombinant PhoA obtained with the pCLT516 vector, the residual six amino acids from the bacterial signal peptide (see Figure 3g) contribute to the observed enhanced enzyme activity. Alternatively, the bacterial signal peptide might direct PhoA into the lumen of thylakoids, which may provide a more appropriate environment than the stroma for correct enzyme folding. To distinguish between these two possibilities,

we determined the precise localization of the full-length encoded PhoA-L within chloroplasts. To this end, immunocytolocalization studies on leaf sections were carried out initially, but the resolution of the gold labelling with the commercially available anti-PhoA antibodies was not sufficiently specific to decipher whether the recombinant protein was or was not present within the lumen of thylakoids. We therefore proceeded to the isolation of intact chloroplasts and their fractionation into stromal and lumenal extracts for



Figure 6 Localization of alkaline phosphatase (PhoA) within tobacco chloroplast compartments. (a, b, c) Western blot characterization of subcellular fractions carried out with 20 μg of protein extracts from plants expressing PhoA-S (lanes 1, 3 and 5) or PhoA-L (lanes 2, 4 and 6). (a) Characterization of chloroplast subcellular fractions using antibodies directed against a stromal (KARI) and a lumenal (TL29) protein. (b) Same experiment as in (a) using a less saturated Western blot. From densitometric analyses of lanes 1, 2, 5 and 6 in (b), the relative proportions of KARI and TL29 are 12% and 88%, respectively, for the chloroplast fractions and 4% and 96%, respectively, for the lumenal fractions. (c) Immunodetection of PhoA in subcellular chloroplast fractions. (d) PhoA specific activity for each subcellular fraction expressed as percentage values vs. the data obtained with standard PhoA.

both PhoA-L and PhoA-S, as described in ‘Experimental procedures’. Isolated fractions were first characterized by Western blotting using specific polyclonal antibodies directed against two soluble proteins that have a known localization within chloroplasts (Figure 6a). Ketol-acid reductoisomerase (KARI) is a stromal homodimeric enzyme involved in the biosynthesis of branched-chain amino acids (Dumas et al., 2001), whose 57-kDa mature monomer is encoded by a single gene (At3g58610) in Arabidopsis thaliana. The TL29 protein is an ascorbate regulator of approximately 29 kDa, also encoded by a single gene in Arabidopsis thaliana (At4g09010), and has been shown to be localized within the lumen of thylakoids (Kieselbach et al., 2000). As expected, both KARI and TL29 were detected in the whole chloroplast extracts (Figure 6a). The stromal extracts were virtually free of thylakoids, as the signal corresponding to TL29 was barely visible

(Figure 6a). The lumenal fractions showed a strong signal for TL29, as expected, but also significant contamination by the stroma, as judged by the use of KARI antibodies (Figure 6a). A less saturated Western blot was produced under the same experimental conditions (Figure 6b). By densitometric analysis of the lane in Figure 6b, it was estimated that the contamination of the lumen by the stroma was in the range 30%– 40%, indicating a two-fold enrichment of lumenal proteins in the purified lumenal fractions. It should be noted that contamination of thylakoid lumenal fractions by proteins from major cell compartments, in particular the stroma, is frequently observed, even in detailed proteomic studies on plastid compartments (Peltier et al., 2000). The same fractions were tested with antibodies directed against PhoA (Figure 6c). In agreement with the data shown in Figure 3, the recombinant enzyme was clearly detected in



whole chloroplast extracts of tobacco lines transformed with either pCLT515 or pCLT516 (Figure 6c). In the stromal fractions, both PhoA-S (pCLT515) and PhoA-L (pCLT516) were revealed by the anti-PhoA antibodies (Figure 6c). The results indicate that a substantial amount of PhoA-S is located in the stroma (Figure 6c). No difference in molecular weight was apparent between PhoA-S and PhoA-L in these stromal extracts (Figure 6c), suggesting, in agreement with the N-terminal sequence determinations (Figure 3g; Table S1) that some processing has occurred at the N-terminus of the full-length encoded PhoA-L, presumably by the thylakoid secretion machinery (Settles and Martienssen, 1998). With regard to the lumenal fractions, it was not surprising to detect PhoA-S, as these fractions exhibit some level of contamination by the stroma (Figure 6a,b). However, the lumenal localization of PhoA-L was supported by the fact that the amounts of PhoA-L and PhoA-S were similar in the stromal fractions (Figure 6c, cf. lanes 3 and 4), whereas PhoA-L was much more abundant than PhoA-S in whole chloroplasts (Figure 6c, cf. lanes 1 and 2). Therefore, it appears that PhoA-L is present in both the stroma and within the lumen of thylakoids. PhoA activity measurements were also expressed in terms of enzyme specific activity (Figure 6d). The measurements for PhoA-S were in accordance with the expected localization of the recombinant protein in the stroma. In the case of the catalytic activity ascribable to PhoA-L, the data were in good agreement with the Western blots, showing that PhoA-L was present in both the stroma and thylakoid lumen (cf. Figure 6c,d). Finally, PhoA activity was assayed in one-dimensional gels carried out in native conditions after separation of protein extracts corresponding to transformed tobacco plants expressing PhoA-S or PhoA-L. The data showed that both active PhoA-S and PhoA-L were detected in transgenic plant extracts, and exhibited an electrophoretic mobility similar to that of standard PhoA from E. coli (Figure S2, see ‘Supplementary material’). This finding demonstrates that the chloroplast recombinant enzymes are correctly assembled as active dimers in the transformed chloroplasts.

Discussion To examine disulphide bond formation in chloroplasts, tobacco plastid transformants were generated that expressed bacterial PhoA, a classical reporter molecule for that purpose (Sone et al., 1997). In this work, tobacco lines were characterized that had either the full-length coding region, including the bacterial signal peptide (PhoA-L), or a shorter sequence encoding the mature enzyme only (PhoA-S), integrated in the

plastid genome. In both cases, recombinant protein expression was easily detected by one- and two-dimensional PAGE of protein extracts. No phenotypic consequence was apparent in the transgenic tobacco lines expressing PhoA, despite the wide variety of compounds that this enzyme is potentially able to hydrolyse, including ATP (Heppel et al., 1962). Under denaturing conditions, the PhoA monomer was detected at approximately the same molecular weight position for PhoA-L and PhoA-S; however, the N-terminal sequencing of PhoA-L showed that the entire signal peptide had not been removed, as six residues were still present. Other post-translational modifications that influence the isoelectric point of the recombinant protein occurred in both cases, as demonstrated by the detection of multiple spots after two-dimensional PAGE. A similar pattern was also observed in bacteria, although to a much lesser extent. At present, we do not know the origin of these variants. Many types of processing or chemical modifications are known that can influence the isoelectric point of a protein, such as phosphorylation, deamidation and carbonylation (Gianazza, 1995). Yet, for another bacterial enzyme over-expressed in tobacco chloroplasts (p-hydroxyphenylpyruvate dioxygenase), only one form of the recombinant protein was detected in twodimensional gels (Dufourmantel et al., 2007), suggesting that the presently observed post-translational modifications are protein specific. Under native conditions, active PhoA-S and PhoA-L were detected in transgenic plant extracts, and had an electrophoretic mobility similar to standard PhoA from E. coli (Figure S2). This indicates that these chloroplast recombinant enzymes were correctly assembled as dimers. The separation of isolated chloroplasts into stromal and thylakoid lumenal fractions showed that PhoA-L was present in both compartments. This means that the bacterial SECtype signal peptide, which, in bacteria, targets the enzyme to the extracellular periplasm, is functional for secretion through the thylakoid membrane. The reverse also holds, as plant thylakoid signal peptides are able to target proteins for secretion in bacteria (Henry et al., 1997). Bacterial PhoA is, to our knowledge, the first example of a recombinant protein expressed by plastid transformants that has been shown to be sorted to the lumen. There are two other examples in the literature of bacterial genes encoding secreted proteins which have been expressed in tobacco plastid transformants. In one case, a thermostable recombinant xylanase from Bacillus subtilis was shown to accumulate at high levels in transformed tobacco chloroplasts. This enzyme was detected by one-dimensional sodium dodecylsulphate (SDS)-PAGE as a single band with an apparent molecular weight suggesting that the signal peptide had been removed; however, no



localization within chloroplast compartments was reported (Leelavathi et al., 2003). More recently, a recombinant lipoprotein from Borrelia burgdorferi was characterized in transformed tobacco chloroplasts; in this case, it was shown that the signal peptide was not removed (Glenz et al., 2006). From our results, at least two possibilities can explain the presence of significant amounts of PhoA-L in the stroma of transformed tobacco chloroplasts: (i) this PhoA-L pool corresponds to N-terminally processed molecules that are not secreted; or (ii) PhoA-L is N-terminally processed during secretion, but translocated back to the stroma. Although the latter possibility seems rather complex, it has recently been invoked in the case of two proteins secreted by the TAT pathway (Di Cola and Robinson, 2005). In particular, this possibility would fit with the fact that stromal PhoA-L and PhoA-S migrate at the same position by SDS-PAGE. Our work on bacterial PhoA expression in tobacco chloroplasts reveals new features concerning the capacity of the stroma and thylakoid lumen to allow the formation of disulphide bonds in recombinant proteins. Thus, PhoA-S lacking a bacterial signal peptide is found to be expressed as an active enzyme form in the stroma. This means that, unlike a normal reducing bacterial cytosol, the redox conditions in this plant cell compartment allow the correct formation of disulphide bonds necessary for recombinant PhoA activity. The present data also confirm previous reports describing the successful expression in the stroma of plastid transformants of other disulphide bondcontaining proteins (Staub et al., 2000; Daniell et al., 2001, 2004; Fernandez-San Millan et al., 2003; Mayfield et al., 2003). This study demonstrates the capacity of the lumen thylakoid space to support disulphide bond formation in recombinant proteins. In bacteria, the signal peptide normally directs PhoA to the SEC secretion pathway (Robinson et al., 2001). The finding that, in tobacco chloroplasts, the signal peptide of PhoA-L is removed but not processed at the same position as in E. coli is indicative of differences between these two prokaryotic systems for PhoA secretion. It would be interesting to determine whether secretion of PhoA-L to the lumen occurs as expected through a bacterial-like SEC pathway. Alternatively, other pathways, such as the TAT pathway (Derman et al., 1993; Di Cola and Robinson, 2005), could translocate folded PhoA (Settles and Martienssen, 1998; Gutensohn et al., 2006). A crosstalk exists between these various secretion pathways, as maize secA mutants that are deficient in the thylakoid SEC pathway can still export mature-sized plastocyanin or cytochrome f across the thylakoid membrane (Robinson et al., 2001). By comparing the expression and activity patterns of recombinant PhoA between chloroplasts and bacteria, it

appears that the properties of the tobacco wild-type stroma are similar to those found in the cytosol compartment of thioredoxin-reductase null mutants in E. coli, in particular Origami B strains, which support the efficient formation of disulphide bonds (DeLisa et al., 2003; Jack et al., 2004). In addition, it is worth noting that a slow folding of PhoA can occur in the cytosol of wild-type E. coli when the growth of cells is suspended, as the mechanism by which proteins are normally kept reduced in the bacterial cytoplasm fails to function in these conditions (Derman and Beckwith, 1995). As the chloroplast stroma is known to be maintained at a rather low redox potential during photosynthesis (Baier and Dietz, 2005), it is possible that folding assistants participate actively in the correct assembly of recombinant PhoA. The presence in the stroma of enzymes such as eukaryotic protein disulphide isomerase may be important in this respect (Levitan et al., 2005; Alergand et al., 2006). It can also be envisaged that the plastid stromal thioredoxins are not able to recognize recombinant bacterial PhoA, and hence to reduce its disulphide bonds, possibly for steric reasons. In conclusion, despite an observed contamination of lumenal fractions by stromal fractions (Figure 6a,b), our work demonstrates that the lumen of thylakoids is even more appropriate than the stroma for expressing proteins that require disulphide bonds, as PhoA-L has a higher specific activity than PhoA-S when extracted from transgenic chloroplasts. In this respect, the properties of the thylakoid lumen appear to be similar to those existing in the bacterial periplasm. These favourable conditions are probably a result of a more oxidizing redox potential in the lumen (Buchanan and Luan, 2005) and/or the presence of specific protein folding assistants. In this context, it is interesting to note that some lumenal enzymes require the formation of disulphide bonds for full activity (Gopalan et al., 2004; Shapiguzov et al., 2006). By contrast, a number of stromal proteins, in particular those linked to photosynthesis, are known to be activated when their disulphide bonds are reduced by thioredoxins (Jacquot et al., 2002). The process and regulation of thiol oxidation in the thylakoid lumen represents a currently very active and exciting research topic (Buchanan and Luan, 2005; SchlarbRidley et al., 2006). It must be stressed that the present findings are based on the characterization of a single reporter enzyme only, and therefore need to be evaluated under other experimental set-ups (e.g. day/night regulation) and with other recombinant proteins containing variable numbers of disulphide bonds. For instance, a recent study on the expression in the plastid stroma of human epidermal growth factor (hEGF), which contains three disulphide bonds, showed recombinant protein accumulation to be dependent on



illumination conditions (Wirth et al., 2006). Redox conditions in the chloroplasts and non-photosynthetic plastids, crosstalk between the stroma and the lumen, fluctuations according to light or dark conditions, and the possible role of folding assistants are areas that have just started to be examined (Baier and Dietz, 2005; Buchanan and Luan, 2005). Experiments are in progress to answer some of these fundamental questions and to explore further the lumen compartment as a new and original location for recombinant protein targeting.

Experimental procedures Construction of transformation vectors The phoA sequence including the bacterial signal peptide targeting the periplasm (PhoA-L) was amplified by PCR from E. coli DH5α genomic DNA with the primers EP-PhoA C (5′-TTATTTCAGCCCCAGAGCGG-3′) and EP-PhoATP N (5′-TGAAACAAAGCACTATTGCACTGGC-3′). The fragment amplified with Pwo polymerase was first inserted into pCR4Blunt-Topo (Invitrogen, Carlsbad, CA, USA), and then cloned into a tobacco chloroplast transformation vector, giving plasmid pCLT516 (GENBANK DQ882177). This vector carries a selection cassette encoding spectinomycin resistance (aadA gene) and targets the integration of the transgenes between the rbcL and accD tobacco plastid genes. The phoA coding region is placed under the control of the tobacco psbA promoter and rbcL terminator (Figure 1). The phoA gene without the bacterial signal peptide (PhoA-S) was amplified by PCR with primers EP-PhoA C and EP-PhoA N (5′-GGGTCATGAGGACACCAGAAATGCCTGT-3′). The fragment amplified with Pwo polymerase was first inserted into pCR4Blunt-Topo, and subsequently cloned into the same tobacco chloroplast transformation vector, giving plasmid pCLT515 (GENBANK DQ882176), as depicted in Figure 1.

Bacterial strains Escherichia coli strains DH5α [F-Φ80lacZΔM15 (lacZYA-argF) U169 deoR recA1 end A1 hsdR17 (rk–,mk–) phoA supE44 THi-1 gyrA69 relA1λ] from Invitrogen, BL21 [F– ompT hsdSB (rB–mB–) gal dcm (DE3) pLysE (CamR)] from Invitrogen and the BL21 derivative strain Origami B [F– ompT hsdSB (rB–,mB–) gal dcm lacY1 gor522::Tn10(TcR) trxB::kan] from Novagen (Darmstadt, Germany) were used as hosts for plasmids pCLT515 and pCLT516. Cultures were grown in Luria–Bertani (LB) medium supplemented with spectinomycin (100 mg/mL).

Generation of tobacco plastid transformants Plastid transformation and plant selection were carried out essentially as described by Svab and Maliga (1993). Briefly, sterile tobacco plants (Nicotiana tabacum cv. PBD6) were grown in solid Murashige and Skoog (1962) medium with 30 g/L sucrose (MS). Transformation was carried out by bombarding 4–5-week-old leaves (length, 4– 6 cm) with gold particles coated with plasmid pCLT515 or pCLT516, using a particle influx generator gun (Finer et al., 1992). Following incubation at 24 °C in MS medium supplemented with hormones, 6-benzylaminopurine (2 mg/L) and 1-naphthalene acetic acid

(0.05 mg/L) for 2 days, bombarded leaves were cut into small pieces (~1 cm2) and subjected to selection on medium containing 500 mg/L of spectinomycin. Resistant shoots obtained after about 6 weeks were rooted and transplanted to the glasshouse.

PCR and Southern blot analyses of total plant DNA Total plant DNA extraction on T0, T1 and T2 generations was performed using the DNeasy Plant Mini Kit (Qiagen, Valencia, CA, USA). Putative transgenic plants were screened by PCR using the specific set of primers No. 3 (5′-GCATGCCGCCAATGTTGTTG-3′) and No. 2 (5′-ACTGCCCATTGATTATTTAGCCC-3′) (Figure 1b) to demonstrate the presence of the phoA gene in the transgenic lines, and to determine whether the integration of foreign phoA-L or phoA-S genes had occurred in the chloroplast genome at the targeted site by homologous recombination, as described in Figure 1. Analysis for homoplasmy used a pair of primers No. 1 (5′-ATGTCACCACAAACACAGACTAAAGC-3′) and No. 2 (5′-ACTGCCCATTGATTATTTAGCCC-3′) flanking the transgene insertion site in the tobacco chloroplast genome (Figure 1a,b). PCRs were performed in an MJ Research thermocycler (MJ Research, Ramsey, MN, USA) using ReadyMix Taq PCR Reaction Mix (Sigma, St Louis, MO, USA). After denaturation for 2 min at 95 °C, samples were subjected to 30 cycles of amplification using the following temperature sequence: 95 °C for 30 s, 54 °C for 1 min and 72 °C for 2 min 30 s. PCR products were separated on 1% agarose gels. For Southern blot analyses, 5 μg of total DNA was digested with HindIII, separated on a 8% agarose gel and transferred to a nylon membrane (Bio-Rad, Hercules, CA, USA). The probes used for Southern blot analyses (Figure 1) were amplified using the following primer pairs: (i) 5′-GTAGAGAGCCGTTTATGAATGTCTTCG-3′ and 5′AAGGATGTCCTAAAGTTCCTCCACC-3′ for the rbcL probe; and (ii) 5′-TTATTTCAGCCCCAGAGCGG-3′ and 5′-GGGTCATGAGGACACCAGAAATGCCTGT-3′ for the phoA probe. The PCR fragments were purified and radiolabelled with 32P by random priming using the MEGAPRIME Kit (Qiagen). The last and most stringent wash was performed with 0.1 × standard saline citrate (SSC) containing 1% SDS (w/v) at 65 °C. Autoradiograms were revealed after 2 h of exposure at –80 °C, using an intensification screen.

Western blot analyses Transformed and untransformed leaves were frozen in liquid nitrogen and ground to a fine powder using a mortar and pestle. Protein extraction buffer [50 mM Tris-HCl, 1 mM dithiothreitol (DTT), pH 8.0], supplemented with protease inhibitor cocktail tablets (Roche Diagnostics, Basel, Switzerland), was added to the powder, and the mixture was incubated on ice for 20 min. The mixture was centrifuged at 4 °C at 13 000 g for 5 min, and the final supernatant was recovered for analysis. Protein concentration was determined by the Bradford method (Bradford, 1976) using a Protein Assay Reagent Kit from Bio-Rad. Protein samples were combined with Laemmli loading buffer supplemented with 10% (v/v) β-mercaptoethanol, boiled and run on a 12% SDS-PAGE gel (Laemmli, 1970). Native PAGE (without SDS and disulphide reductants) was performed with total proteins from leaves extracted in 50 mM Tris-HCl (pH 8.0), supplemented with protease inhibitor cocktail tablets (Roche Diagnostics). Extracts were not boiled and no reducing agent was added before loading on the gel and Western blot analysis (see Figure S2).



Separated proteins were transferred to poly(vinylidene difluoride) (PVDF) membranes (Bio-Rad) using a liquid electroblotting apparatus (Mini-Protean 3 Cell, Bio-Rad). After 2 h of saturation at room temperature with TTBS buffer [10 × TBS from Bio-Rad, 0.1% (v/v) Tween 20, pH 7.5] containing Western blocking reagent (Roche Diagnostics), membranes were washed three times with TTBS buffer and incubated overnight at 4 °C with TTBS buffer containing monoclonal antibody raised in mouse against PhoA (mAb1012, Chemicon International, Temecula, CA, USA). After three washes in TTBS buffer, membranes were incubated for 2 h at room temperature with TTBS buffer containing secondary antibodies raised in goat against mouse immunoglobulin Gs (IgGs) conjugated to PhoA (Sigma). Detection was performed after three washes in TTBS buffer and one wash in TBS buffer, with Immun-Star™ AP substrate Pack (Bio-Rad) on Hyperfilm™ ECL (Amersham, Buckinghamshire, UK).

pended in water supplemented with BCIP (20 mg/mL) and incubated at room temperature. Tobacco leaf discs were incubated overnight at 37 °C in sterile water supplemented with BCIP (20 mg/mL). This experiment was repeated four times on 30 leaf discs for all transformant and wild-type lines. The determination of PhoA activity was performed and repeated independently eight times using a colorimetric assay based on paranitrophenyl phosphate (pNPP) hydrolysis. The assay was performed by mixing in each well of a 96-well plate 260 μL of 100 mM glycine, 1 mM MgCl2, 1 mM ZnCl2 and small volumes (< 10 μL) of proteins extracted under native conditions from recombinant bacteria or plants. The enzymatic reaction was performed at 37 °C and initiated by adding 30 μL of pNPP substrate solution (one tablet in 5 mL of sterile water; Sigma). The assay was read against blank at 405 nm in a Beckman Coulter AD340 spectrophotometer (Beckman Coulter, Fullerton, CA, USA). Activity was expressed using the following equation

Two-dimensional gel electrophoresis

Units of PhoA activity / mL = ( ΔA405nm/min test − ΔA405nm/min blank) × (total volume reaction in mL) 18.5 × ( volume of extract in mL)

Proteins were analysed by two-dimensional gel electrophoresis as described previously (Görg et al., 1987; Job et al., 2005). Isoelectrofocusing was carried out with protein samples corresponding to about 100 μg of total proteins extracted from leaves of tobacco plastid transformants or from saturated overnight cultures of E. coli. Proteins from the various extracts were separated using gel strips forming an immobilized non-linear gradient from pH 3 to pH 10 (Immobiline Dry Strip pH 3–10 NL, 18 cm; Amersham Pharmacia Biotech, Freiburg, Germany). Strips were rehydrated for 14 h at 22 °C with the thiourea–urea lysis buffer, as described previously (Harder et al., 1999), containing 2% (v/v) Triton X-100, 20 mM DTT and the protein extracts. Isoelectric focusing was performed at 22 °C in a Multiphor II system (Amersham Pharmacia Biotech) for 1 h at 300 V and 7 h at 3500 V. The gel strips were then equilibrated for 2 × 20 min in 2 × 100 mL of equilibration solution containing 6 M urea, 30% (v/v) glycerol, 2.5% (w/v) SDS, 0.15 M bis-Tris and 0.1 M HCl (Görg et al., 1987; Harder et al., 1999). DTT (50 mM) was added to the first equilibration solution, and iodoacetamide (4%, w/v) was added to the second (Harder et al., 1999). Separation in the second dimension was carried out in polyacrylamide gels (10% w/v acrylamide, 0.33% w/v piperazidine diacrylamide, 0.18 M Trizma base, 0.166 M HCl, 0.07% w/v ammonium persulphate and 0.035% v/v Temed). Electrophoresis was performed at 10 °C in a buffer (pH 8.3) containing 25 mM Trizma base, 200 mM taurine and 0.1% (w/v) SDS for 1 h at 35 V and 14 h at 110 V. Ten gels (200 × 250 × 1.0 mm) were run in parallel (Isodalt system, Amersham Pharmacia Biotech). The twodimensional gels were stained with silver nitrate according to Blum et al. (1987) for densitometric analyses and Shevchenko et al. (1996) for mass spectrometric analyses. Two-dimensional gels were also stained with the GelCode blue stain from Pierce (Rockford, IL, USA). Stained gels were scanned with a UMAX Powerlook III scanner equipped with Magicscan version 4.5 from UMAX Data Systems (Amersham Biosciences, Buckinghamshire, UK).

PhoA assays For in vivo assays, bacterial strains expressing either PhoA-S or PhoAL were grown overnight at 37 °C in liquid LB medium supplemented with antibiotic (500 mg/L spectinomycin). For each strain, 1 mL of saturated culture was centrifuged, and the bacterial pellet was resus-

where 18.5 mM−1 cm−1 corresponds to the millimolar extinction coefficient of pNPP at 405 nm. The specific activity of PhoA in the different extracts was expressed as the ratio between the number of units of enzymatic activity per milligram of protein and the amount of phosphatase, normalized by the amount of total protein, detected in the extracts on onedimensional Western blots using the chemiluminescent reaction monitored with Quantity One software from Bio-Rad. For PhoA activity detection after native PAGE and transfer, the PVDF membrane was incubated at room temperature with nitroblue tetrazolium chloride (NBT) and BCIP (NBT/BCIP ready to use tablets, Roche Diagnostics) dissolved in water (Figure S2).

Protein localization For chloroplast fractionation into stromal and thylakoid preparations, the protocol was adapted from Kieselbach et al. (1998). Tobacco leaves (around 100 g) were blended twice at very low speed and at 4 °C in a Waring blender (HGBSSSS6 model) for 2–3 s in 330 mM sorbitol, 50 mM HEPES-KOH (pH 7.8), 10 mM KCl, 1 mM ethylenediaminetetraacetic acid (EDTA), 0.15% (w/v) bovine serum albumin, 4 mM sodium ascorbate and 7 mM cysteine. The resulting mixture was filtered through three layers of nylon mesh (20 μm), and the filtrate was centrifuged for 20 min at 1500 g at 4 °C. The pellets were resuspended in 25 mL of 330 mM sorbitol, 50 mM HEPES-KOH (pH 7.8) and 10 mM KCl, centrifuged for 20 min at 1500 g at 4 °C and resuspended in 25 mL of the same buffer. The resulting solution was layered over a two-step Percoll gradient (8 mL of 40%, 4 mL of 80%). The gradients were centrifuged at 2000 g for 10 min at 4 °C. The lower green band containing intact organelles was isolated and the chloroplasts were disrupted in 10 mM sodium pyrophosphate buffer (pH 7.8). The thylakoids were then collected by centrifugation in 0.6 M sucrose at 13 000 g for 10 min at 4 °C. The supernatant containing the stromal extract was separated from the pellet containing dark-green thylakoids. The latter was washed twice with each of the following buffers: (i) 10 mM sodium pyrophosphate (pH 7.8) to remove residual stromal proteins; (ii) 2 mM Tricine (pH 7.8), 300 mM sucrose to remove unidentified extrinsic thylakoid membrane proteins; (iii) 30 mM sodium phosphate (pH 7.8), 50 mM NaCl,



5 mM MgCl2 and 100 mM sucrose (fragmentation buffer). Intact thylakoids were disrupted by sonication on ice, three times for 10 s each. The sonicated thylakoids were then centrifuged for 5 min in an Eppendorf centrifuge at 4 °C, and the lumen extract was obtained by centrifugation of this supernatant at 20 000 g for 10 min. Isolated fractions were characterized on Western blots using polyclonal rabbit antibodies directed against a soluble lumenal protein (TL29) from Arabidopsis thaliana (Kieselbach et al., 2000), and a soluble stromal protein (KARI) from Spinacia oleracea (Dumas et al., 2001).

Protein identification by mass spectrometry and Edman sequencing Proteins characterized in this work were identified by matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) analysis performed by Dr Maya Belghazi (INRA, Nouzilly, France). The spots were excised from two-dimensional PAGE gels with sterile tips and placed in 1.5-mL sterile tubes. Each spot was rinsed and reduced with 10 mM DTT, alkylated with 55 mM iodoacetamide and incubated overnight at 37 °C with 12.5 ng/mL trypsin (sequencing grade; Roche Diagnostics) in 25 mM NH4HCO3. The tryptic fragments were extracted, dried, reconstituted with 2% (v/v) acetonitrile, 0.1% formic acid and sonicated for 10 min. Analysis of tryptic peptides by MS-MS was performed on a nanoelectrospray ionization quadrupole timeof-flight hybrid mass spectrometer (Q-TOF Ultima Global; Waters Micromass, Milford, MA, USA) coupled with a nano-high-performance liquid chromatograph (nano-HPLC) (Cap-LC; Waters). The samples were loaded and desalted on a C18 precolumn (LC-Packings PepMap C18, 5 mm, 100 Å, 300 μm × 5 mm) at a flow rate of 20 μL/min isocratically with 0.1% formic acid. The peptides were separated on a C18 column (Atlantis dC18, 3 μm, 75 μm × 150 mm, Nano Ease; Waters). After washing with solvent A (water–acetonitrile, 98/2 v/v; 0.1% formic acid), a linear gradient from 5% to 60% of solvent B (water–acetonitrile, 20/80 v/v; 0.1% formic acid) was developed over 80 min at a flow rate of 180 nL/min. The Q-TOF spectrometer was operated in the Data Dependent Analysis mode using a 1-s MS survey scan on three different precursor ions. The peptide masses and sequences obtained were either matched automatically to proteins in a non-redundant database (National Center for Biotechnology Information, NCBI) using the Mascot MS/MS Ions Search algorithm (http://www.matrixscience.com) or BLASTed manually against the current databases. For N-terminal identification of proteins, proteins were transferred from two-dimensional gels to PVDF Western blotting membranes (Boehringer, Mannheim, Germany), and the spots of interest were submitted to amino acid sequencing performed by Dr Jacques d’Alayer (Institut Pasteur, Paris, France) by automated Edman degradation of the peptides with a PE Applied Biosystems sequencer (Applied Biosystems, Foster City, CA, USA). All sequences determined in this work are available as supplementary material in the online version of the article.

Acknowledgements This work was supported by the French Ministry of Industry and Bayer BioScience (PhD thesis support to J.B.). We wish to thank Renaud Dumas (CEA, Grenoble, France) and Wolfgang Schröder (Umea University, Umea, Sweden) for

providing rabbit polyclonal sera directed against KARI and TL29, respectively.

References Alergand, T., Peled-Zehavi, H., Katz, Y. and Danon, A. (2006) The chloroplast protein disulfide isomerase RB60 reacts with a regulatory disulfide of the RNA-binding protein RB47. Plant Cell Physiol. 47, 540–548. Arntzen, C., Plotkin, S. and Dodet, B. (2005) Plant-derived vaccines and antibodies: potential and limitations. Vaccine, 23, 1753–1756. Baier, M. and Dietz, K.J. (2005) Chloroplasts as source and target of cellular redox regulation: a discussion on chloroplast redox signals in the context of plant physiology. J. Exp. Bot. 56, 1449 –1462. Ballesteros, M., Fredriksson, A., Henriksson, J. and Nyström, T. (2001) Bacterial senescence: protein oxidation in non-proliferating cells is dictated by the accuracy of the ribosomes. EMBO J. 20, 5280 –5289. Bessette, P.H., Aslund, F., Beckwith, J. and Georgiou, G. (1999) Efficient folding of proteins with multiple disulfide bonds in the Escherichia coli cytoplasm. Proc. Natl. Acad. Sci. USA, 96, 13 703– 13 708. Blum, H., Beier, H. and Gross, H.J. (1987) Improved silver staining of plant proteins, RNA and DNA in polyacrylamide gels. Electrophoresis, 8, 93–99. Bradford, M.M. (1976) A rapid and sensitive method for the quantification of microgram quantities of protein utilizing the principle of protein–dye binding. Anal. Biochem. 72, 248–254. Brixley, P.J., Guda, C. and Daniell, H. (1997) The chloroplast psbA promoter is more efficient in Escherichia coli than the T7 promoter for hyperexpression of a foreign protein. Biotechnol. Lett. 19, 395–399. Buchanan, B.B. and Luan, S. (2005) Redox regulation in the chloroplast thylakoid lumen: a new frontier in photosynthesis research. J. Exp. Bot. 56, 1439–1447. Daniell, H., Lee, S.B., Panchal, T. and Wiebe, P.O. (2001) Expression of the native cholera toxin B subunit gene and assembly as functional oligomers in transgenic tobacco chloroplasts. J. Mol. Biol. 311, 1001–1009. Daniell, H., Cohill, P.R., Kumar, S. and Dufourmantel, N. (2004) Chloroplast genetic engineering. In: Molecular Biology of Plant Organelles: Chloroplasts and Mitochondria (Daniell, H. and Chase, C., eds), pp. 443– 490. Dordrecht: Springer. De Cosa, B., Moar, W., Lee, S.B., Miller, M. and Daniell, H. (2001) Overexpression of the Bt cry2Aa2 operon in chloroplasts leads to formation of insecticidal crystals. Nat. Biotechnol. 19, 71–74. DeLisa, M.P., Tullman, D. and Georgiou, G. (2003) Folding quality control in the export of proteins by the bacterial twin-arginine translocation pathway. Proc. Natl. Acad. Sci. USA, 100, 6115–6120. Derman, A.I. and Beckwith, J. (1991) Escherichia coli alkaline phosphatase fails to acquire disulfide bonds when retained in the cytoplasm. J. Bacteriol. 173, 7719–7722. Derman, A.I. and Beckwith, J. (1995) Escherichia coli alkaline phosphatase localized to the cytoplasm slowly acquires enzymatic activity in cells whose growth has been suspended: a caution for gene fusion studies. J. Bacteriol. 177, 3764–3770. Derman, A.I., Prinz, W.A., Belin, D. and Beckwith, J. (1993) Mutations that allow disulfide bond formation in the cytoplasm of Escherichia coli. Science, 262, 1744–1747. Di Cola, A. and Robinson, C. (2005) Large-scale translocation reversal within the thylakoid Tat system in vivo. J. Cell. Biol. 171, 281–289.



Dubald, M., Tissot, G., Dufourmantel, N. and Goutorbe, F. (2007) The engineering of recombinant plastids in higher plants. In: Recent Advances in Plant Biotechnology (Kumar, A., ed.), in press. New Delhi: IK International Publishers. Dufourmantel, N., Dubald, M., Matringe, M., Canard, H., Garçon, F., Job, C., Kay, E., Wisniewski, J.P., Ferullo, J.M., Pelissier, B., Sailland, A. and Tissot, G. (2007) Generation and characterization of soybean and marker-free tobacco plastid transformants overexpressing a bacterial 4-hydroxyphenylpyruvate dioxygenase which provides strong herbicide tolerance. J. Plant Biotechnol. 5, 118–133. Dumas, R., Biou, V., Halgand, F., Douce, R. and Duggleby, R.G. (2001) Enzymology, structure, and dynamics of acetohydroxy acid isomeroreductase. Acc. Chem. Res. 34, 399–408. Eibl, C., Zhurong, Z., Beck, A., Minkyun, K., Mullet, J. and Koop, H.U. (1999) In vivo analysis of plastid psbA, rbcL and rpl32 UTR elements by chloroplast transformation: tobacco plastid gene expression is controlled by modulation of transcript levels and translation efficiency. Plant J. 19, 333–345. Fernandez-San Millan, A., Mingo-Castel, A., Miller, M. and Daniell, H. (2003) A chloroplast transgenic approach to hyper-express purify human serum albumin, a protein highly susceptible to proteolytic degradation. J. Plant Biotechnol. 1, 71–79. Finer, J.J., Vain, P., Jones, M.W. and McMullen, M.D. (1992) Development of the particle inflow gun for DNA delivery to plant cells. Plant. Cell. Rep. 11, 323–328. Gianazza, E. (1995) Isoelectric focusing as a tool for the investigation of post-translational processing and chemical modifications of proteins. J. Chromatogr. A. 705, 67–87. Glenz, K., Bouchon, B., Stehle, T., Wallich, R., Simon, M.M. and Warzecha, H. (2006) Production of a recombinant bacterial lipoprotein in higher plant chloroplasts. Nat. Biotechnol. 24, 76–77. Gopalan, G., He, Z., Balmer, Y., Romano, P., Gupta, R., Heroux, A., Buchanan, B.B., Swaminathan, K. and Luan, S. (2004) Structural analysis uncovers a role for redox in regulating FKBP13, an immunophilin of the chloroplast thylakoid lumen. Proc. Natl. Acad. Sci. USA, 101, 13 945–13 950. Görg, A., Postel, W., Weser, J., Günther, S., Strahler, J.R., Hanash, S.M. and Somerlot, L. (1987) Elimination of point streaking on silver stained two-dimensional gels by addition of iodoacetamide to the equilibration buffer. Electrophoresis, 8, 122–124. Gray, M.W. (2004) The evolutionary origins of plant organelles. In: Molecular Biology of Plant Organelles: Chloroplasts and Mitochondria (Daniell, H. and Chase, C., eds), pp. 15–36. Dordrecht: Springer. Gutensohn, M., Fan, E., Frielingsdorf, S., Hanner, P., Hou, B., Hust, B., Klosgen, R.B. (2006) Toc, Tic, Tat et al.: structure and function of protein transport machineries in chloroplasts. J. Plant. Physiol. 163, 333–347. Harder, A., Wildgruber, R., Nawrocki, A., Fey, S.J., Larsen, P.M. and Görg, A. (1999) Comparison of yeast cell protein solubilization procedures for two-dimensional electrophoresis. Electrophoresis, 20, 826–829. Henry, R., Carrigan, M., McCaffrey, M., Ma, X. and Cline, K. (1997) Targeting determinants and proposed evolutionary basis for the Sec and the Delta pH protein transport systems in chloroplast thylakoid membranes. J. Cell. Biol. 136, 823–832. Heppel, L.A., Harkness, D.R. and Hilmoe, R.J. (1962) A study of the substrate specificity and other properties of the alkaline phosphatase of Escherichia coli. J. Biol. Chem. 237, 841–846. Jack, R.L., Buchanan, G., Dubini, A., Hatzixanthis, K., Palmer, T. and Sargent, F. (2004) Coordinating assembly and export of complex bacterial proteins. EMBO J. 23, 3962–3972.

Jacquot, J.P., Rouhier, N. and Gelhaye, E. (2002) Redox control by dithiol-disulfide exchange in plants. I. The chloroplastic systems. Ann. N.Y. Acad. Sci. 973, 508–519. Job, C., Rajjou, L., Lovigny, Y., Belghazi, M. and Job, D. (2005) Patterns of protein oxidation in Arabidopsis seeds and during germination. Plant Physiol. 38, 790–802. Kadokura, H., Katzen, F. and Beckwith, J. (2003) Protein disulfide bond formation in prokaryotes. Annu. Rev. Biochem. 72, 111–135. Kieselbach, T., Hagman, A., Andersson, B. and Schroder, W.P. (1998) The thylakoid lumen of chloroplasts. Isolation and characterization. J. Biol. Chem. 273, 6710–6716. Kieselbach, T., Bystedt, M., Hynds, P., Robinson, C. and Schroder, W.P. (2000) A peroxidase homologue and novel plastocyanin located by proteomics to the Arabidopsis chloroplast thylakoid lumen. FEBS Lett. 480, 271–276. Laemmli, U.K. (1970) Cleavage of structural protein during the assembly of the head of the bacteriophage T4. Nature, 227, 680–685. Leelavathi, S., Gupta, N., Maiti, S., Ghosh, A. and Reddy, V.S. (2003) Overproduction of an alkali- and thermo-stable xylanase in tobacco chloroplasts and efficient recovery of the enzyme. Mol. Breed. 11, 59–67. Levitan, A., Trebitsh, T., Kiss, V., Pereg, Y., Dangoor, I. and Danon, A. (2005) Dual targeting of the protein disulfide isomerase RB60 to the chloroplast and the endoplasmic reticulum. Proc. Natl. Acad. Sci. USA, 102, 6225–6230. Ma, J.K.C., Drake, P.M.W. and Christou, P. (2003) The production of recombinant pharmaceutical proteins in plants. Nat. Rev. 4, 794–802. Maliga, P. (2004) Plastid transformation in higher plants. Annu. Rev. Plant. Biol. 55, 289–313. Margulis, L. (1975) Symbiotic theory of the origin of eukaryotic organelles; criteria for proof. Symp. Soc. Exp. Biol. 29, 21–38. Mason, H.S., Warzecha, H., Mor, T. and Arntzen, C.J. (2002) Edible plant vaccines: applications for prophylactic and therapeutic medicine. Trends Mol. Med. 8, 324–329. Mayfield, S.P., Franklin, S.E. and Lerner, R.A. (2003) Expression and assembly of a fully active antibody in algae. Proc. Natl. Acad. Sci. USA, 100, 438– 442. Murashige, T. and Skoog, F. (1962) A revised medium for rapid growth and bioassays with tobacco tissue culture. Physiol. Plant. 15, 473– 497. Peltier, J.B., Friso, G., Kalume, D.E., Roepstorff, P., Nilsson, F., Adamska, I. and van Wijk, K.J. (2000) Proteomics of the chloroplast: systematic identification and targeting analysis of lumenal and peripheral thylakoid proteins. Plant Cell, 12, 319–341. Peltier, J.B., Emanuelsson, O., Kalume, D.E., Ytterberg, J., Friso, G., Rudella, A., Liberles, D.A., Soderberg, L., Roepstorff, P., von Heijne, G. and van Wijk, K.J. (2002) Central functions of the lumenal and peripheral thylakoid proteome of Arabidopsis determined by experimentation and genome-wide prediction. Plant Cell, 14, 211–236. Prinz, W.A., Aslund, F., Holmgren, A. and Beckwith, J. (1997) The role of the thioredoxin and glutaredoxin pathways in reducing protein disulfide bonds in the Escherichia coli cytoplasm. J. Biol. Chem. 272, 15 661–15 667. Robinson, C., Thompson, S.J. and Woolhead, C. (2001) Multiple pathways used for the targeting of thylakoid proteins in chloroplasts. Traffic, 2, 245–251. Schlarb-Ridley, B.G., Nimmo, R.H., Purton, S., Howe, C.J. and Bendall, D.S. (2006) Cytochrome c6A is a funnel for thiol oxidation in the thylakoid lumen. FEBS Lett. 580, 2166–2169.



Settles, A.M. and Martienssen, R. (1998) Old and new pathways of protein export in chloroplasts and bacteria. Trends Cell. Biol. 8, 494–501. Shapiguzov, A., Edvardsson, A. and Vener, A.V. (2006) Profound redox sensitivity of peptidyl-prolyl isomerase activity in Arabidopsis thylakoid lumen. FEBS Lett. 580, 3671–3676. Shevchenko, A., Wilm, M., Vorm, O. and Mann, M. (1996) Mass spectrometric sequencing of proteins in silver-stained polyacrylamide gels. Anal. Chem. 68, 850–858. Sone, M., Kishigami, S., Yoshihisa, T. and Ito, K. (1997) Role of disulfide bonds in bacterial alkaline phosphatase. J. Biol. Chem. 272, 6174–6178. Staub, J.M., Garcia, B., Graves, J., Hajdukiewicz, P.T.J., Hunter, P., Nehra, N., Paradkar, V., Schlittler, M., Carroll, J.A., Spatola, L., Ward, D., Ye, G. and Russell, D.A. (2000) High-yield production of a human therapeutic protein in tobacco chloroplasts. Nat. Biotechnol. 18, 333–338. Svab, Z. and Maliga, P. (1993) High-frequency plastid transformation in tobacco by selection for a chimeric aadA gene. Proc. Natl. Acad. Sci. USA, 90, 913–917. Svab, Z., Hajdukiewicz, P. and Maliga, P. (1990) Stable transformation of plastids in higher plants. Proc. Natl. Acad. Sci. USA, 87, 8526–8530. Warzecha, H. and Mason, H.S. (2003) Benefits and risks of antibody and vaccine production in transgenic plants. J. Plant. Physiol. 160, 755–764. Wirth, S., Segretin, M.E., Mentaberry, A. and Bravo-Almonacid, F. (2006) Accumulation of hEGF and hEGF-fusion proteins in chloroplast-transformed tobacco plants is higher in the dark than in the light. J. Biotechnol. 125, 159–172. Zhang, Q., Liu, Y. and Sodmergen (2003) Examination of the cytoplasmic DNA in male reproductive cells to determine the potential for cytoplasmic inheritance in 295 angiosperm species. Plant Cell. Physiol. 44, 941–951.

Supplementary material The authors have provided the following supplementary material, which can be accessed alongside the article at http://www.blackwell-synergy.com. It contains two figures and two tables. Figure S1 Southern blot analyses. Total DNA extracted from transformants expressing the short (PhoA-S) (pCLT515, lanes 1–4) or long (PhoA-L) (pCLT516, lanes 5–8) form of alkaline phosphatase was analysed by Southern blotting with the rbcL (a) or phoA (b) probe. The analysis was conducted with total DNA from generation T0 in lanes 1 and 5, from one T1 progeny in lanes 2 and 6, and from two T2 progenies in lanes 3, 4, 7 and 8. WT, wild-type sample. The samples were digested by the restriction enzyme HindIII, separated by agarose gel electrophoresis, blotted on to a nylon membrane and hybridized with 32P-labelled probes covering the rbcL and phoA coding regions, as depicted in Figure 1. A unique DNA fragment at around 7 kb is detected in all the lanes using the rbcL probe, confirming the integration of the transgenes at the

predicted locus between the rbcL and accD plastome genes (arrow). The phoA probe reveals a major band at around 8 kb for the transgenic samples (arrow), testifying to the presence of the phoA gene in the plastid genome of these lines. A lower molecular weight band at around 3 kb is also visible in all lanes, except for wild-type tobacco, which is always in the same ratio with the predominant band at 8 kb. This unexpected fragment might correspond to recombinations between introduced and resident tobacco plastid sequences, as shown by Eibl et al., (1999). Such events have been detected in the first tobacco plastid transformants generated by Svab and Maliga (1993) using the aadA selectable marker gene placed under the control of tobacco plastid expression elements. Finally, no difference is observed in this pattern between T0, T1 and T2 generations, showing that the introduced transgenes are stably inherited. Figure S2 Characterization of chloroplast alkaline phosphatase (PhoA) after native polyacrylamide gel electrophoresis. (a) PhoA activity after protein transfer on to nitrocellulose membrane and incubation with 5-bromo-4-chloro-3-indolyl phosphate/nitroblue tetrazolium chloride (BCIP/NBT) substrate. (b) Corresponding Western blot analysis using antibodies directed against PhoA. Twenty micrograms of tobacco leaf proteins were loaded from wild-type extracts (wt) and independent extracts (1 and 2) from transgenic lines generated with vectors pCLT515 (PhoA-S) and pCLT516 (PhoA-L). Two micrograms of PhoA from Escherichia coli (standard) were loaded as control. MWM, molecular weight marker from Bio-Rad (dual colour). These data show that recombinant PhoA is expressed in transgenic tobacco as an active homodimer (the monomeric unfolded form is inactive; Sone et al., 1997). Table S1 Peptide sequences corresponding to recombinant alkaline phosphatase expressed in tobacco chloroplasts. Spot numbers correspond to Figure 3 in the main text Table S2 Peptide sequences corresponding to recombinant alkaline phosphatase expressed in bacterial strains BL21 and BL21 trxB/gor. Spot numbers correspond to Figure 4 in the main text This material is available as part of the online article from: http://www.blackwell-synergy.com/doi/abs/10.1111/ j.1467-7652.2007.00298.x (This link will take you to the article abstract). Please note: Blackwell Publishing are not responsible for the content or functionality of any supplementary materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.
