Phosphorylation of the transit sequence of chloroplast precursor proteins.

A protein kinase was located in the cytosol of pea mesophyll cells. The protein kinase phosphorylates, in an ATP-dependent manner, chloroplast-destined precursor proteins but not precursor proteins, which are located to plant mitochondria or plant peroxisomes. The phosphorylation occurs on either serine or threonine residues, depending on the precursor protein used. We demonstrate the specific phosphorylation of the precursor forms of the chloroplast stroma proteins ferredoxin (preFd), small subunit of ribulose-bisphosphate-carboxylase (preSSU), the thylakoid localized light-harvesting chlorophyll a/b-binding protein (preLHCP), and the thylakoid lumen-localized proteins of the oxygen-evolving complex of 23 kDa (preOE23) and 33 kDa (preOE33). In the case of thylakoid lumen proteins which possess bipartite transit sequences, the phosphorylation occurs within the stroma-targeting domain. By using single amino acid substitution within the presequences of preSSU, preOE23, and preOE33, we were able to tentatively identify a consensus motif for the precursor protein protein kinase. This motif is (P/G)X(n)(R/K)X(n)(S/T)X(n) (S*/T*), were n = 0-3 amino acids spacer and S*/T* represents the phosphate acceptor. The precursor protein protein kinase is present only in plant extracts, e.g. wheat germ and pea, but not in a reticulocyte lysate. Protein import experiments into chloroplasts revealed that phosphorylated preSSU binds to the organelles, but dephosphorylation seems required to complete the translocation process and to obtain complete import. These results suggest that a precursor protein protein phosphatase is involved in chloroplast import and represents a so far unidentified component of the import machinery. In contrast to sucrose synthase, a cytosolic marker protein, the precursor protein protein kinase seems to adhere partially to the chloroplast surface. A phosphorylation-dephosphorylation cycle of chloroplast-destined precursor proteins might represent one step, which could lead to a specific sorting and productive translocation in plant cells.

Chloroplasts and mitochondria contain their own genome; however, the vast majority of their proteins is encoded in the nucleus and synthesized in the cytosol. In general these proteins carry NH 2 -terminal targeting domains which direct them to the proper organelle in a posttranslational event. Whereas some knowledge about the components and mechanisms which are involved in the recognition and translocation of the precur-sors on the organellar surface has accumulated (Hirsch et al., 1994;Schnell et al., 1994;Gray and Row, 1995;Soll, 1995), almost nothing is known about the events which take place in the cytosol, after the precursor emerges from the ribosome and before binding to the chloroplast. A loosely folded precursor, however, seems to be a prerequisite for membrane translocation. In all import systems aggregation of highly hydrophobic membrane proteins and premature folding has to be prevented. To attain or maintain such an import-competent, soluble conformation, some (Waegemann et al., 1990) but not all proteins need the help of cytosolic factors in vitro (Pilon et al., 1992). A 70-kDa heat shock protein (hsc70) 1 was found to be involved in the transport of proteins to different destinations, e.g. mitochondria, endoplasmic reticulum, and chloroplasts (Deshaies et al., 1988;Zimmermann et al., 1988;Waegemann et al., 1990). It was also shown that precursor proteins readily interact with hsc70 during translation (Beckmann et al., 1990). Whereas this interaction seems a rather general phenomenon (Ellis and van der Vies, 1991), two cytosolic factors have been described which act as specific chaperones for mitochondrial precursors. These factors, called presequence binding factor and mitochondrial import stimulation factor, have been purified from rabbit reticulocyte lysate and rat liver cytosol, respectively, and were shown to recognize mitochondrial presequences and stimulate their import into the organelle (Murakami et al., 1992;Hachiya et al., 1993Hachiya et al., , 1995. Such cytosolic factors which interact specifically with the organellar targeting domain could be even more important in a plant cell, because the plastids represent an additional compartment to which the cytosolic synthesized precursors have to be properly routed. Whereas the mitochondrial presequences share an overall similar framework, i.e. positively charged residues, amphiphilic ␣-helical structure in plants, fungi, and mammals (von Heijne, 1986;Hartl et al., 1989), the plastid-directing transit sequences are much more heterogenous in length and secondary structure. The only common features of the chloroplast transit domains are an uncharged NH 2 -terminal region, a characteristic stromal processing site, and a particularly high content of serine and threonine residues (Karlin-Neumann and Tobin, 1986;von Heijne and Nishikawa, 1991;von Heijne et al., 1989). In order to understand the series of events which ultimately yield a specific and productive routing of precursor proteins to plastids, we have started an investigation on cytosolic components which could be involved in these processes.
Here we report the specific phosphorylation of several chloroplast precursor proteins within the stroma-targeting portion of the transit sequences, but not of their mature forms. Phosphorylation seems to inhibit import of precursor proteins into chloroplasts. Mitochondrial or peroxisomal precursors are not recognized by the protein kinase. The protein kinase, which belongs to the serine/threonine kinase family, is a plant-specific enzyme. It was shown to be located in the cytoplasm, but it also adheres to the chloroplast outer envelope. /SSU,preLHCP/LHCP,preOE23/iOE23,preOE33,preFd,and preF 1 ␤-Several plasmids for the overexpression of proteins in Escherichia coli were used. Pet 11c expression vector containing the cDNAs for preSSU and SSU, respectively, both from tobacco, were provided by Dr. R. Klein, University of Kentucky (Klein and Salvucci, 1992). cDNAs for preLHCP and LHCP from pea, both in pDS 12 expression vectors are described in Paulsen et al. (1990). Dr. K. Cline, University of Florida, provided the cDNA clones for preOE23, iOE23, and preOE33 from pea, all in pet 3c plasmids (Cline et al., 1993). The coding sequence for pea preFd was cloned in expression plasmid pet 21d by PCR using full-length preFd cDNA in BscII as template. The sequence of the used forward primer which contains an in frame NcoI site was 5Ј-CCCCCCCATGGCAACCACACCAGC-3Ј. The sequence of the reverse primer containing an in frame XhoI site was 5Ј-GAAAAAT-TAAGCCTCGAGATC-3Ј. The PCR product and vector pet 21d were digested with NcoI and XhoI and ligated. The preFd/pet 21d clone obtained was controlled by sequencing. The gene encoding the mitochondrial preF 1 ␤ from tobacco (Boutry and Chua, 1985) was obtained from Dr. Boutry, University of Louvaine, as a part-length clone (amino acid 1-513) in pEMBL 12. After digestion with NcoI which contains the initiation ATG and SalI, the insert was ligated into an appropriately digested pet 21d vector.

Construction of preSSU
With the exception of preLHCP/LHCP plasmids which were transformed into the E. coli strain JM101 all plasmids were introduced in BL21De3 E. coli cells for expression. Expression and isolation of the overproduced proteins was done as described by Paulsen et al. (1990), with the exception that preF 1 ␤ was expressed at 30°C instead of 37°C. PreF 1 ␤ and preFd were expressed with a C-terminal His-Tag.
Site-directed Mutagenesis of preSSU, preOE23, and preOE33-Exchange of serine 31 and serine 34 of tobacco preSSU for alanine yielding cpreSSU-M31-S/A and cpreSSU-M34-S/A, respectively, or the exchange of both amino acids together for alanine resulting in cpreSSU-M31/34-S/A was done by PCR using preSSU cDNA in pet 11c as template. The primers for mutagenesis contain an AflII site, which is an endogenous restriction site in the preSSU cDNA near the mutated region. The exchange of serine for alanine was achieved by a single base exchange (T into G) for each serine. The used forward primer sequences were 5Ј-CACTGGCCTTAAGGCAGCTGCCTCATTCC-3Ј, 5Ј-CACTGG-CCTTAAGTCAGCTGCCGCATTCC-3Ј, and 5Ј-CACTGGCCTTAAGGC-AGCTGCCGCATTCC-3Ј for cpreSSU-31-S/A, cpreSSU-M34-S/A, and cpreSSU-M31/34-S/A, respectively. The universal reverse primer for all mutants was 5Ј-GTTAATTGGTGGCCACACCTGCATGCATTGC-AC-3Ј, which resembles a NsiI restriction site in the preSSU cDNA. The PCR products were digested with AflII and NsiI and ligated into the wild type preSSU cDNA which had been cut by the same restriction enzymes. The clones obtained were sequenced and transformed into BL21De3 cells for expression.
Exchange of serine 22 for alanine in preOE23 yielding cpreOE23-M22-S/A was done by recombinant PCR (Innis et al., 1990). For the two primary PCR runs cpreOE23/pet 3c was used as template. For the first amplification T7-promotor-primer (ϭ outside primer) was used as forward primer. The sequence for the reverse primer containing the point mutation (A into C) was 5Ј-CTTGGCGTTGAGCTAAGGTTCTAG-3Ј (ϭ inside primer). For the second amplification, the outside primer sequence was 5Ј-GATCAGATCTGATATCATC-3Ј, resembling a region in the multiple cloning site of pet 3c. As an inside primer resembling the complementary sequence of the inside primer of the first primary PCR, including the respective point mutation (T into G), we used 5Ј-CTA-GAACCTTAGCTCAACGCCAAG-3Ј. The two primary PCR products were purified by gel electrophoresis and polyethyleneglycol precipitation. The secondary PCR which was performed by mixing the two primary PCR products, which overlap in the inside primer sequence, with the above described outside primers produce full-length cpreOE23 containing the mutation serine 22 into alanine. This PCR product and pet 3c were digested with NdeI and EcoRV and ligated. The clone cpreOE23-M22-S/A was controlled by sequencing and transformed into BL21De3 cells to produce preOE23-M22-S/A. Exchange of threonine 21 for alanine in preOE33 was also done by recombinant PCR as decribed above. The two outside primers were identical with those of the preOE23 mutation. As inside primers we used 5Ј-GCGTAGCAACGCATTGCAGC-3Ј and 5Ј-GCTGCAATGCGTT-GCTACGC-3Ј, respectively, containing the mutations (A into G) and (T into C). The template for the two primary amplifications was cpreOE33/ pet 3c. The PCR product of the secondary amplification was digested with NdeI and SacI and ligated into appropriate digested pet 24c vector. The obtained clone was controlled by sequencing and transformed into BL21De3 cells to produce preOE33-M21-T/A.
PreSSU-M54-R/D was also created by recombinant PCR. As outside primers we used T7-promotor primer and T7-terminator primer. We used the two complementary inside primers 5Ј-CGGCGGAGATGTG-CAATGCATGC-3Ј and 5Ј-GCATGCATTGCACATCTCCGCCG-3Ј containing the mutation AGA into GAT and TCT into ATC, respectively. The template DNA for both primary amplifications was tobacco cpreSSU/pet 11c. The PCR product of the secondary amplification was digested with NcoI and NsiI and ligated into appropriate digested wild type cpreSSU/pet 21d. The obtained clone was controlled by sequencing and was transformed in E. coli BL21De3. The protein (preSSU-M54-R/D) was expressed with a C-terminal His-Tag.
Preparation of Cytoplasm-enriched Fraction from Pea Mesophyll Cells-Protoplasts were prepared from 10 -12-day-old pea seedlings as described by Hracky and Soll (1986). The membrane-free cytosolic fraction was frozen in small aliquots in liquid nitrogen and stored at Ϫ80°C.
Preparation of Chloroplasts and "Chloroplast Supernatant"-Intact chloroplasts were isolated from 10 -12-day-old peas (Waegemann and Soll, 1991). For the chloroplast supernatant, the isolated chloroplasts were kept in isotonic solution (Waegemann and Soll, 1991) on ice for 30 min in the dark. After reisolation of the organelles by centrifugation at 1500 ϫ g for 5 min, the supernatant was removed and centrifuged again at 10,000 ϫ g for 5 min. The resulting supernatant was used as the chloroplast supernatant and was prepared fresh for each experiment.
Phosphorylation Assay-A standard phosphorylation (50 l) was performed in 20 mM Tris/HCl, pH 7.5, 5 mM MgCl 2 , 0.5 M MnCl 2 with 2.5 M ATP plus 2-5 Ci of [␥-32 P]ATP (3000 Ci/mmol). For kinase activity, 10 -20 g of a cytoplasm-enriched protein fraction or 10 g of wheat germ extract protein were added. The assays contained 1-3 g of purified overexpressed proteins as substrates which were dissolved in 8 M urea prior to the experiment. The final urea concentration in the phosphorylation assay was 160 mM. The reaction was incubated for 5 min at room temperature and terminated with SDS sample buffer. The phosphorylation products were analyzed by SDS-PAGE (Laemmli, 1970) followed by autoradiography.
Analysis of Phosphorylated Amino Acid-For the determination of the phosphoacceptor amino acid, 10 phosphorylation assays with the respective protein substrate were performed and separated by SDS-PAGE. The gel was stained with Coomassie Blue, and the bands containing the phosphorylated proteins were excised from the gel. After equilibration (3 ϫ 10 min) in 125 mM Tris/HCl, pH 6.8, 0.1% SDS, 10% glycerol, 1 mm EDTA, 45 mM ␤-mercaptoethanol, the respective protein was electroeluted from the gel slices according to the manufacturer's instructions (Biometra, Göttingen, Germany). The eluted protein was precipitated with methanol and chloroform (Wessel and Flü gge, 1984) and subsequently hydrolyzed in 6 N HCl for 2 h at 110°C. The hydrochloric acid was removed by evaporation, and the sample was dissolved in H 2 O and analyzed by high voltage electrophoresis on Silica Gel 60 chromatography plates (Schleicher & Schü ll) according to Hunter and Sefton (1980). The electrophoresis was performed in the presence of 20 -40 g of phosphoserine and phosphothreonine as standards. After electrophoresis at 500 V for 4 h, the plate was dried, sprayed with ninhydrin reagent to mark the phosphoamino acid standards, and exposed on x-ray film.
Sucrose Synthase Assay-Sucrose synthase activity was measured by a coupled enzymatic reaction resulting in the conversion of NAD into NADH (Geigenberger and Stitt, 1993). The first step was performed in a final volume of 100 l in 20 mM Hepes/KOH, pH 7.0, 5 mM dithiothreitol, 100 mM sucrose, and 4 mM UDP with 70 g of protein of the chloroplast supernatant or cytoplasm. After 30 min at 25°C, the reaction was terminated by incubating for 4 min at 95°C. The produced UDP-glucose was measured in a second step which was performed in a final volume of 600 l with 50 l of the first enzyme reaction in 200 mM glycine, pH 8.7, 0.1 mM NAD, and 25 g of UDP-glucose dehydrogenase. NADH production was determined at 334 nm.
Preparation of Wheat Germ Extract-Unroasted wheat-germ was ground in liquid nitrogen and subsequently thawed in 40 mM Hepes/ KOH, pH 7.6, containing 100 mM potassium acetate, 1 mM magnesium acetate, and 4 mM dithiothreitol. After centrifugation of the homogenate at 10000 ϫ g for 10 min, the supernatant was spun again as above. The resulting supernatant was frozen in small aliquots in liquid nitrogen and stored at Ϫ80°C. Prior to use an aliquot was quickly thawed and centrifuged at 250,000 ϫ g for 15 min to remove membranes, ribosomes, and insoluble material. Partial Purification of the Precursor Protein Protein Kinase Activity-Precursor protein protein kinase activity was enriched from wheat germ extract. Immediately before each chromatography step the homogenate was spun at 250,000 ϫ g for 20 min to remove membranes and insoluble material. The supernatant (50 -60 mg of protein) was applied to a 18-ml Q-Sepharose FF column (Pharmacia Biotech Inc.) equilibrated in TEMP buffer (20 mM Tris/HCl, pH 6.5, 1 mM EDTA, 10 mM ␤-mercaptoethanol, 0.5 mM phenylmethylsulfonyl fluoride). After washing the column with 6 volumes of TEMP buffer, bound proteins were eluted with a linear NaCl gradient from 0 to 500 mM in TEMP buffer. The kinase activity eluted at 240 -440 mM NaCl. The active fractions of four Q-Sepharose purifications were pooled, adjusted to pH 7.5, concentrated in an ultrafiltration cell, and dialyzed overnight against 10 mM Tris/HCl, pH 7.5, 0.5 mM EDTA, 10 mM ␤-mercaptoethanol. The Q-Sepharose pool (50 -60 mg of total protein) was then adjusted to 1 M NaCl, centrifuged at 15,000 ϫ g for 10 min, and loaded subsequently onto two 50-ml phenyl-Sepharose FF columns (low substitution, Pharmacia) equilibrated in 20 mM Tris/HCl, pH 7.5, 10 mM ␤-mercaptoethanol, 1 M NaCl. After washing the bound proteins were eluted with 20 mM Tris/HCl, pH 7.5, 10 mM ␤-mercaptoethanol. The active fractions of both columns were pooled, concentrated (7-9 mg of total protein), and finally applied to a 120-ml Superdex 75 Hiload column (Pharmacia) equilibrated in 20 mM Tris/HCl, pH 7.5, 150 mM NaCl. Protein kinase activity was predominantly found in fractions 14/15 (0.15-0.25 mg of total protein). After concentrating to 100 l of Superdex-fraction 14/15 was stored in 50% glycerol at Ϫ20°C.
Protease Treatment of Chloroplasts-Intact isolated pea chloroplasts were incubated with thermolysin (100 g of thermolysin/mg of chlorophyll) in 330 mM sorbitol, 50 mM Hepes/KOH, pH 7.6, 3 mM MgCl 2 , and 0.5 mM CaCl 2 for 20 min on ice. The final chlorophyll concentration was 1 mg of chlorophyll/ml. The treatment was terminated by the addition of a 2-fold molar excess of ␣-macroglobulin for 15 min on ice. After reisolation of the chloroplasts by centrifugation (1500 ϫ g; 1 min). The organelles were resuspended in 330 mM sorbitol, 50 mM Hepes/KOH, pH 7.6.
Analytical Methods-Protein was determined by the method of Bradford (1976) using bovine serum albumin as a standard. SDS-PAGE was performed according to the method of Laemmli (1970). Silver staining of the gels was done as described in Blum et al. (1987).

Specific Phosphorylation of Chloroplast Precursor Proteins-
Posttranslational modification of proteins by phosphorylation is an important mechanism in the regulation of cellular processes. The introduction of a phosphate group into a given polypeptide chain is likely to alter its three-dimensional conformation. The import-competent conformation or the exposure of certain motifs for recognition by chaperones, presequence binding factors, or receptors could be influenced by a phosphorylation event within the presequence. The high content of serine and threonine residues in the chloroplast transit sequences suggests a phosphorylation of precursor proteins. To address this question, several chloroplast precursor proteins and their mature proteins were overexpressed in E. coli, recovered in a partially purified form from inclusion bodies (Fig. 1A), and subjected to in vitro phosphorylation (Fig. 1B). As examples of typical soluble stromal proteins, we used the precursor and the mature forms of the small subunit of ribulose-bisphosphate carboxylase/oxygenase (preSSU/SSU) and of ferredoxin (preFd/Fd). As a member of the thylakoid membrane the precursor and the mature form of the chlorophyll a/b-binding protein of the light-harvesting complex II (preLHCP/LHCP) were used. The transit sequence of preLHCP has typical stroma directing properties and is cleaved by the stromal proc-FIG. 1. Overexpression and phosphorylation of proteins from different plant cell compartments. A, isolated bacteria-expressed proteins were subjected to SDS-PAGE and stained with Coomassie Blue. Mature Fd was isolated from spinach leaves according to Yasunobo and Tanaka (1980). B, the purified proteins, which were recovered from inclusion bodies, were dissolved in 8 M urea and subjected to in vitro phosphorylation in the presence of cytoplasm (20 g of protein) and [␥-32 P]ATP. PreMDH and MDH were purified from E. coli as soluble proteins and were also adjusted to 8 M urea prior to the phosphorylation reaction. An autoradiogram is shown. essing peptidase in one step in the stroma. Two precursor proteins of the oxygen-evolving complex associated with photosystem II (preOE23⅐preOE33) were also analyzed. These proteins, which reside at the lumenal side of the thylakoid membranes, possess, in contrast to preLHCP, bipartite presequences consisting of an amino-terminal stroma-targeting domain and an additional thylakoid transfer domain. The stroma-targeting domains of preOE23 and preOE33 are cleaved off in the stroma to yield intermediate sized precursors which are further translocated across the thylakoids and processed finally in the thylakoid lumen. For the OE23 protein this intermediate precursor lacking the stroma-targeting domain, but still possessing the thylakoid transfer domain (iOE23), was also available. Beside these chloroplast proteins we also used overexpressed precursors from the other organelles in the plant cell to which precursor proteins are directed by NH 2 -terminal-cleavable presequences in a posttranslational manner. As a constituent of plant mitochondria, the precursor of the ␤-subunit of the ATPase (preF 1 ␤) from tobacco was expressed, and as a member of the peroxisomes, we used the precursor and the mature form of malate dehydrogenase (preMDH/MDH) (gift of Ch. Gietl, Mü nchen) (Gietl et al., 1994). Each of the above described proteins was subjected to an in vitro phosphorylation assay with [␥-32 P]ATP in the presence of cytoplasmic proteins isolated from pea mesophyll protoplasts (Fig. 1B). All chloroplast precursors examined,i.e. preSSU,preFd,preLHCP,preOE23, and preOE33 became phosphorylated (indicated by Ͼ in Fig. 1B). Their mature forms or the intermediate processed form of OE23 (iOE23), however, were not able to serve as a substrate for the cytosolic protein kinase activity, indicating that the phosphorylation occurred in the stromal targeting domain of the transit sequence. The cytoplasm used in the phosphorylation assays is likely to contain several protein kinases (Ranjeva and Boudet, 1987). We observed the phosphorylation of several endogenous proteins in the absence of added precursor proteins (Fig. 1B, lane 1). The pattern of endogenous phosphorylation varied with the different batches of cytoplasm used, while the precursor protein protein kinase activity seemed more stable. This could be due to the long isolation procedure of a cytosolic fraction from protoplasts (see "Materials and Methods") and to different lengths of storage. The endogenous phosphoproteins were similar to those described by Hracky and Soll (1986). To ascertain that this phosphorylation is specific for chloroplast precursor proteins and not a general event, other precursor proteins carrying NH 2 -terminal presequences but which are destined for different locations inside the plant cell were also subjected to in vitro phosphorylation. As shown in Fig. 1B the plant mitochondrial preF 1 ␤ could not be phosphorylated. The peroxysomal preMDH and also the mature MDH were phosphorylated only very weakly. We conclude that the radioactive label is probably incorporated into the mature part of MDH. Altogether these data demonstrate that a protein kinase is localized in cytoplasm enriched from pea mesophyll cells which is able to specifically phosphorylate precursor proteins destined for the chloroplast compartment. The phosphorylation seems to occur exclusively in the stroma-targeting domain of the presequence.
Determination of the Phosphorylation Sites-For most protein kinases the amino acids serine, threonine, or tyrosine serve as phosphoacceptor amino acids. A common feature of all chloroplast transit sequences is their high content (20 -30%) of serine and threonine. To determine the identity of the phosphoacceptor group the 32 P-labeled precursor proteins were separated by SDS-PAGE, excised from the gel, and electroeluted. The eluted proteins were precipitated by CHCl 3 /MeOH followed by acid hydrolysis at 110°C. The hydrolysate was ana-lyzed by high voltage electrophoresis on silica gel thin layer chromatography plates. We found that serine serves as the phosphate acceptor in preSSU ( Fig. 2A), preFd (Fig. 2B), and preOE23 ( Fig. 2C), whereas preOE33 was phosphorylated exclusively on a threonine residue (Fig. 2D). In the case of preL-HCP, both amino acids incorporated labeled phosphate in almost equal amounts (Fig. 2E). An endogenous phosphorylation site with threonine as phosphate acceptor is present very close to the NH 2 terminus of mature LHCP. This site is involved in the regulation of light harvesting in the thylakoids in vivo (Bennett, 1991). Thus this internal phosphorylation site might also be recognized in our in vitro phosphorylation experiments when the precursor form and cytosol are used in the phosphorylation assay. Taken together these results suggest that the functional precursor protein protein kinase belongs to the serine/threonine kinase family.
To determine the phosphorylation sites exactly and to find a putative consensus motif in the chloroplast transit peptides phosphorylated preSSU was treated with the protease trypsin. This resulted in one radioactive labeled fragment (not shown). Sequencing of this peptide revealed that this fragment represents indeed a part of the preSSU transit sequence from tobacco. The fragment comprises amino acid 31-39 of preSSU. Three serines are present in this fragment which could serve as potential phosphate acceptors (Fig. 3). Two of these serines were exchanged for alanine by in vitro mutagenesis. The mutations affected serine 31 and serine 34 which were exchanged either separately resulting in the clones cpreSSU-M31-S/A and cpreSSU-M34-S/A or simultaneously yielding cpreSSU-M31/34-S/A (Fig. 3). The three mutated cDNAs were subsequently expressed in E. coli (Fig. 4A, lanes 1-4, C) and subjected to in vitro phosphorylation as before (Fig. 4A, lanes 1-4, 32 P). The phosphorylation experiments of the mutated precursor proteins resulted in a reduced 32 P incorporation into preSSU-M31-S/A (Fig. 4A, lane 2, 32 P) and no incorporation into preSSU-M34-S/A and preSSU-M31/34-S/A (Fig. 4A, lanes  3 and 4, 32 P). From these data we conclude that serine 34 is the actual phosphoacceptor in tobacco preSSU, whereas serine 31 might function in the recognition or binding of the kinase.
With the knowledge of this phosphorylation site we analyzed other transit sequences for homologies. A similar motif was found in preOE23. The respective serine 22 was exchanged for alanine resulting in cpreOE23-M22-S/A (Figs. 3 and 4B, lane  2, C). The phosphorylation experiment using the mutated protein clearly shows that preOE23-M22-S/A is not phosphoryl- FIG. 2. Phosphoamino acid analysis of the phosphorylated chloroplast precursor proteins. Precursor proteins were subjected to in vitro phosphorylation with wheat germ extract (10 g of protein) and separated by SDS-PAGE. After electroelution and precipitation the proteins were hydrolyzed in 6 N HCl and subsequently analyzed by high voltage electrophoresis on silica gel thin layer chromatography plates. The positions of the phosphoamino acid standards phosphoserine (P-Ser) and phosphothreonine (P-Thr) are indicated beside the respective autoradiogram. A,preSSU;B,preFd;C,preOE23;D,preOE33;E,preLHCP. ated any more indicating that serine 22 is indeed the phosphoacceptor amino acid in the pea preOE23 transit peptide (Fig. 4B, lane 2, 32 P). A similar sequence homology was found in preOE33 (Fig. 3). From the phosphoamino acid analysis we knew that this protein is labeled on a threonine residue. Therefore we exchanged threonine 21 for alanine resulting in cpreOE33-M21-T/A (Fig. 3). The overexpressed protein (Fig.  4C, lane 2, C) was subjected to the kinase assay (Fig. 4C, lane 2, 32 P). To our surprise the mutated protein was only slightly less phosphorylated than the wild-type precursor (Fig. 4C, compare lane 1, 32 P and 2, 32 P). This observation could be due to three reasons. First, the mutation did not affect the phosphoacceptor amino acid. Second, threonine 21 acts as phosphoacceptor group, but the kinase can also use another residue with a lower efficiency if threonine 21 is not available. And third, with the exchange of threonine 21 into alanine a new phosphoryla-tion site was created which could also be used by the kinase with reduced efficiency. In order to address this question we analyzed the phosphoacceptor amino acid in preOE33-M21-T/A by acid hydrolysis and thin layer chromatography. As shown in Fig. 4D the mutated preOE33-M21-T/A was now exclusively phosphorylated on a serine in contrast to the wild type (Fig.  2D). From these data we conclude that threonine 21 is the phosphoacceptor group in pea preOE33. We cannot decide now whether the kinase is able to switch to serine phosphorylation or whether a new phosphorylation motif was created.
Altogether these results suggest a consensus motif in the chloroplast transit peptides which could serve as a putative recognition site for the cytosolic precursor protein protein kinase. We postulate that the motif consists of a turn-promoting residue (P/G) followed by a basic amino acid (R/K), an hydroxylated group (S/T) and finally by the actual phosphoacceptor amino acid (S/T) (Fig. 3). The spacing between this residue seems to be variable (see Fig. 3 and "Discussion").
The Precursor Protein Protein Kinase Is Present in Different Plant Sources-In order to find a suitable source for the enrichment of the cytosolic protein kinase and for other studies, a soluble extract from wheat germ was tested for its capability in chloroplast precursor phosphorylation. The wheat germ extract showed specific phosphorylation of all chloroplast precursor proteins tested (Fig. 2), but not of their mutated forms (not shown) and was therefore used as a starting material for purification (see "Materials and Methods"). 2 Interestingly, rabbit reticulocyte lysate, which was also tested as a source, was not able to use chloroplast precursors as substrate for protein phosphorylation, suggesting that the kinase is a plant-specific enzyme (not shown). Irrespective of the source of the precursor protein protein kinase, i.e. wheat germ (Fig. 2), partially purified preparation (Fig. 5), or pea cytosol (Fig. 1), its specificity was established using various precursors and their mutated forms.
Significance of Precursor Phosphorylation for Protein Import-In a first approach import studies were conducted using the mutated preSSU-M34-S/A which can not longer serve as a substrate for the precursor protein protein kinase. These experiments revealed that the mutated protein is imported into chloroplasts as efficient as the wild type preSSU indicating that the phosphorylation of the transit sequence is not prerequisite for the translocation event in vitro ( Fig. 5B and not shown). Earlier results had, however, demonstrated that the phosphatase inhibitors NaF and NaMoO 4 inhibit protein import into plastids in a reversible manner (Flü gge and Hinz, 1986;Schindler et al. 1987) raising the possibility that dephosphorylation of the precursor protein during the import process might indeed be a regulatory step.
Different experimental approaches were used to test this idea (Fig. 5, A and B). First we wanted to know at what stage during import dephosphorylation of phosphorylated preSSU (P-preSSU) occurred, i.e. early during translocation at the envelope membranes or in the stroma before or during processing. To elucidate this a preSSU processing mutant was constructed and synthesized from the tobacco cDNA clone in analogy to a preSSU mutant described for pea (Archer and Keegstra, 1993). Due to an amino acid exchange (M54-R/D) close to the stromal processing site, this precursor is impaired in normal processing but yields an intermediate form (iSSU), which is about 2 kDa smaller than the precursor form. Import efficiency of the mutant preSSU is similar to wt-preSSU (Fig. 5A, lanes 5-8). The intermediate processing site in preSSU-M54-R/D is most likely NH 2 -terminal of the phosphorylation site in preSSU, which is 2 K. Waegemann, unpublished results. at amino acid position 34. We reasoned that, if removal of the phosphate from preSSU would occur via or after processing, iSSU-M54-R/D should still be phosphorylated. To test this we used [␥-35 S]ATP␥S to label both wt-preSSU and preSSU-M54-R/D. ATP␥S is accepted as a phosphate donor by protein kinases but thiophosphorylated proteins are not or much more slowly dephosphorylated by protein phosphatases (McGowan and Cohen, 1988;MacKintosh, 1993). This seems also true in plants as thiophosphorylated preSSU is not dephosphorylated by a soluble protein extract from pea chloroplasts over a 60-min period (not shown). 3 35 S-Labeled wt-preSSU is imported into intact chloroplasts and processed, and the mature form appears protease-protected inside the organelles (Fig. 5A, lanes 5 and 6). The mutant 35 S-labeled preSSU-M54-R/D is imported and processed with similar efficiency as the wt-preSSU, only processing yields an intermediate form (iSSU) (Fig. 5A, lanes 7 and 8). This intermediate form is also detected in the experiment using wt-preSSU but to a much lesser extent. In parallel assays nonradioactive preSSU and preSSU-M54-R/D, both labeled by phosphorylation with a partially purified kinase fraction from a wheat germ lysate and [␥-35 S]ATP␥S prior to import into chloroplasts, were used. The thiophosphorylated forms of preSSU and preSSU-M54-R/D accumulated at the chloroplast surface in comparison to preSSU or preSSU-M54-R/D phosphorylated by ATP, presumably because complete translocation was inhib-ited or slowed down by the thiophosphate group in the precursor proteins. No labeled iSSU or iSSU-M54-R/D was detectable (Fig. 5, lanes 1-4) inside chloroplasts, although the thiolabeled precursor cannot be dephosphorylated in the time course (5 min) of the import experiment (see above) (not shown). This indicates that the thiophosphorylated precursor had not yet reached the stroma for processing. In contrast, the chloroplastbound preSSU and preSSU-M54-R/D yield two new distinct thiophosphorylated translocation intermediates (P-TimA and P-TimB) upon protease treatment (Fig. 5A, lanes 2 and 4). In addition the thiophosphorylated precursor forms of both proteins seem to be more protease-protected than the controls (Fig.  5A, lanes 6 and 8). P-TimA and P-TimB are recovered in the envelope membrane fraction and not in the soluble chloroplast protein (not shown). These data indicate that dephosphorylation of preSSU might be necessary for complete translocation into the stroma, i.e. if dephosphorylation is hindered by introduction of a thiophosphate group, the import process comes to an early stop shortly after binding.
To obtain further experimental evidence for this notion, we conducted import experiments in the presence of NaF, a broad range protein phosphatase inhibitor (Fig. 5B). Unlabeled wt-preSSU was phosphorylated by [␥-32 P]ATP and a partially purified kinase fraction from wheat germ extracts (see "Material and Methods"). 32 P-Labeled preSSU bound to chloroplasts in the absence or presence of NaF. Protease treatment of organellar bound precursors yielded 32 P-labeled P-TimA and P-TimB. In the absence of NaF, 32 P-labeled translocation inter-3 J. Soll, unpublished results. 1 and 2) and preSSU-M54-R/D (lanes 3 and 4) were incubated with a partially purified kinase from wheat germ in the presence of 10 Ci of [␥-35 S]ATP␥S (specific activity, 3000 Ci/mmol) prior to the chloroplast import experiment. 35 S-Labeled preSSU (lanes 5 and 6) and preSSU-M54-R/D (lanes 7 and 8) were treated as above except that unlabeled ATP was used. Import experiments (Waegemann and Soll, 1991) were done in the presence of 3 mM ATP and chloroplasts equivalent to 10 g of chlorophyll for 5 min at 25°C. After completion of the import reaction, organelles were either not treated (odd numbers) or treated with the nonpenetrating protease thermolysin (even numbers). All other manipulations are indicated on top of the figure. B, PreSSU was labeled as above either with [␥-32 P]ATP (lanes 1-4) or with [␥-35 S]-ATP␥S (lanes 5-8), or 35 S-labeled preSSU was phosphorylated with unlabeled ATP (lanes 9 -12 as above). Lanes 13-16, 35 S-labeled preSSU-M34-S/A was incubated with unlabeled ATP and partially purified kinase from wheat germ extract. The pretreated precursors were added to intact purified pea chloroplasts, and import was assayed in the absence or presence of 50 mM NaF as mediates accumulate to about 10 -20% of the NaF level (compare Fig. 5B, lanes 2 and 4). When we used thiophosphorylated preSSU in a standard import assay in the presence of NaF, binding of preSSU to chloroplasts increased, compared to that in the the absence of NaF (Fig. 5B, lanes 5 and 7). Upon protease treatment P-TimA and P-TimB and protease protected precursor protein were detected in the absence (Fig. 5B,  lanes 5 and 6) of NaF. In the presence of NaF the amount of protease protected preSSU and P-TimA and P-TimB increased similar to the increased amount of binding (Fig. 5B, lanes 7 and  8). The detection of P-TimA and P-TimB in the absence (Fig. 5,  A and B) or presence (Fig. 5B, lanes 1-8) of NaF indicates that NaF per se does not deviate the precursor from its normal import to a bypass pathway. This is corroborated by earlier findings (Flü gge and Hinz, 1986;Schindler et al., 1987). In a parallel experiment we used 35 S-labeled wt-preSSU which had been incubated prior to the import with unlabeled ATP and partially purified kinase. In the absence of NaF, import occurred normally (Fig. 5B, lanes 9 and 10), while in the presence of NaF almost no import was detectable, and bound preSSU accumulated again (Fig. 5B, lane 11). Subsequent protease treatment resulted in the occurrence of P-TimA and P-TimB (Fig. 5B, lane 12, indicated by an arrowhead). The difference in labeling intensities of P-TimA and P-TimB in Fig. 5B, lanes 8 and 12, is due to (i) differences in the specific activity of the labeled educts, i.e. [␥-35 S]ATP␥S and 35 S-labeled preSSU and (ii) substoichiometric phosphorylation of preSSU in vitro, i.e. the import reactions presented in Fig. 5B, lanes 9 -12, contain two different precursor proteins, one which is phosphorylated and one which is not. The nonphosphorylated wt-preSSU subpopulation probably gives rise to the additional translocation intermediates (Tim1-4) seen in Fig. 5B, lane 12. The yield of Tim1-4 in these experiments varied, probably due to the different ratio of phosphorylated to nonphosphorylated preSSU (a typical result out of five repeats is shown in lanes 9 -12). Tim1-4 are most likely identical to deg 1-4, which have been described as translocation intermediates in the preSSU import pathway into chloroplasts for the reticulocyte lysate-synthe-sized precursor protein (Waegemann and Soll, 1991). To test this we used the nonphosphorylatable mutant preSSU-M34S/A subjected it to a mock phosphorylation treatment and assayed its import characteristics in the absence or presence of NaF. In the absence of NaF, preSSU-M34S/A imports normally and is processed to the mature form (Fig. 5B, lanes 13 and 14). In the presence of NaF, preSSU accumulates at the chloroplast surface, and almost no import occurs. After a protease treatment, translocation intermediates appear that are not related to phosphorylation, namely Tim1-4 (Fig. 5B, lanes 15 and 16). Translocation intermediate homologues to P-TimA and P-TimB are not generated to a significant amount from preSSU-M34S/A. From these data we conclude that a phosphorylated chloroplast precursor protein cannot be imported into the organelle, but that dephosphorylation is required sometime after binding but before translocation.

FIG. 5. Dephosphorylation of preSSU is required for complete translocation into pea chloroplasts. A, PreSSU (lanes
The Cytosolic Precursor Protein Protein Kinase Is Associated with the Chloroplast Outer Surface-Our initial results indicated that the chloroplast precursor specific protein kinase is a soluble enzyme in the cytoplasm (Fig. 6, lanes 1 and 2). Proteins exist, however, which have a dual localization, i.e. soluble and membrane-associated. We thus wanted to test whether a portion of the kinase is also bound to the chloroplast surface. For this purpose isolated intact pea chloroplasts were incubated with preSSU in the presence of [␥-32 P]ATP. This experiment shows that chloroplasts are indeed able to phosphorylate preSSU (Fig. 6, lanes 3-6). However, only a small portion of the labeled precursor could be isolated together with the organelles after centrifugation (Fig. 6, lanes 3 and 4), whereas the main portion of 32 P-labeled preSSU was recovered in the supernatant (Fig. 6, lanes 5 and 6). These results are consistent with the observation that the initial interaction of precursor proteins with the chloroplast surface is a reversible process Perry and Keegstra, 1994). Tight and irreversible binding of preSSU to the chloroplasts requires between 20 and 50 M ATP (Olsen et al., 1989), concentrations which are about 10-fold higher than those used for the phosphorylation of preSSU, which was 2.5 M. When the organelles were pre- FIG. 6. Localization of the protein kinase activity. A cytoplasm-enriched fraction from pea mesophyll cells (20 g of protein) was incubated in the absence (lane 1) or presence (lane 2) of overexpressed preSSU with [␥-32 P]ATP. A protein kinase assay was performed with isolated intact pea chloroplasts (equivalent to 15 g of chlorophyll) in the absence (lanes 3 and 5) or presence (lanes 4 and 6) of preSSU. After completion of phosphorylation, the organelles were reisolated by centrifugation. Either the pellet, containing the intact chloroplasts (P) or the supernatant (S) was subjected to SDS-PAGE and autoradiography. Intact chloroplasts were pretreated without (lanes 7 and 8) or with (lanes 9 and 10) thermolysin (Th) (10 g of Th/100 g of chlorophyll) as indicated in the figure. After termination of the digestion with a 2-fold molar excess of ␣-macroglobulin, the organelles were reisolated by centrifugation and subjected to in vitro phosphorylation in the presence (lanes 8 and 10) or absence (lanes 7 and 9) of preSSU. After completion of the reaction, the assays were divided into pellet and supernatant by centrifugation. The autoradiogram of resulting supernatants (S) is shown. In vitro phosphorylation with chloroplast supernatant, which was prepared as described under "Materials and Methods," was performed in the absence (lane 11) or in the presence (lane 12) of preSSU. The position of preSSU is indicated by an arrowhead in each autoradiogram. treated with thermolysin, a protease which cannot enter the organelle prior to the phosphorylation assay, the protein kinase activity was significantly diminished. This indicates that the protein kinase is exposed on the chloroplast surface (Fig. 6,  lanes 7-10), while internal chloroplast protein kinases were not affected, e.g. phosphorylation of phosphoglucomutase in the soluble fraction at about 60 kDa and phosphorylation of the thylakoid membrane localized LHCP at 24 -29 kDa (Bennett, 1991) remained largely unchanged. When isolated chloroplasts were left on ice in isotonic buffer and reisolated after 30 min by centrifugation, the resulting supernatant, referred to as chloroplast supernatant, contained most of the phosphorylating activity (Fig. 6, lanes 11 and 12). The respective chloroplast pellet had an residual activity of about 10% (not shown). These data suggest that a portion of the precursor protein protein kinase is in a loose association with the organellar surface.
To test the specificity of the association of this largely cytosolic enzyme with the chloroplasts, we analyzed the distribution of sucrose synthase, a cytosolic marker enzyme (Stitt and Steup, 1985), with that of the protein kinase activity in isolated cytoplasm and in chloroplasts. Quantification of the kinase activity and sucrose synthase activity in the chloroplast supernatant and in isolated cytoplasm shows that both fractions were able to phosphorylate preSSU, but sucrose synthase activity could be detected exclusively in the cytoplasm (Fig. 7). Therefore we conclude that the protein kinase is a cytosolic enzyme which might specifically interact and adhere to the chloroplast surface. The nature of this interaction remains to be elucidated.

DISCUSSION
Most chloroplast proteins are synthesized in the cytosol and have to be imported into the chloroplasts. The import is mediated by an NH 2 -terminal transit sequence which contains all of the information necessary for translocation in vitro. However, the transit sequences of the different chloroplast proteins are very heterogenous in length and also in secondary structure. Attempts to dissect the presequences into functional domains, which are important for certain steps in protein import, have been successfully conducted for some precursor proteins, i.e. preSSU (Reiss et al., 1989) and preFd (Pilon et al., 1995). But unfortunately these motifs do not fit to the presequences of all or most chloroplast precursors. On the way to find such a common motif, we addressed the question whether a specific phosphorylation in the transit sequence could be such a motif. Here we have reported on the phosphorylation of five different overexpressed chloroplast precursor proteins. Each of the proteins,i.e. preSSU,preFd,preLHCP,preOE23,and preOE33, was labeled in the transit sequence. Precursors containing a bipartite presequence, i.e. preOE23 and preOE33, are phosphorylated in the stromal targeting domain of the transit peptide as indicated by the fact that iOE23 could not serve as a substrate for the precursor protein protein kinase. Other precursor proteins with NH 2 -terminal cleavable presequences, i.e. mitochondrial preF 1 ␤ or peroxysomal preMDH, were not phosphorylated or like the MDH in the mature part of the protein. We are currently trying to overexpress more mitochondrial precursor proteins from plants to broaden the basis for this notion.
Phosphoamino acid analysis showed that the phosphorylation of the chloroplast transit sequences occurs on serine or threonine residues. For preSSU, preOE23, and preOE33, the phosphorylation site was determined exactly. Comparison of these sequences revealed a loose motif for a consensus sequence which is defined by a turn-promoting residue (P/G) followed by a spacer, a basic amino acid (R/K), a hydroxylated residue (S/T), a spacer, and the actual phosphoacceptor amino acid (S*/T*). The length of the spacer can vary from protein to protein. The kinase belongs to the serine/threonine protein kinase family and was shown to be a cytoplasmic enzyme which might associate loosely with chloroplasts. The precursor protein protein kinase was detected in phylogenetically different plants, the dicotyledonous pea versus the monocotyledonous wheat and in developmentally very different stages, i.e. pea leaf mesophylls cells and wheat germ embryos. To date it is not clear whether the chloroplast precursors are phosphorylated in the cytoplasm during translation or before binding to the organelle, or if they are phosphorylated at the chloroplast surface during initial phases of chloroplast-precursor interaction . Import studies using the nonphosphorylatable preSSU-M34-S/A revealed that the mutated proteins are imported into chloroplasts as efficient as the wt-preSSU, indicating that phosphorylation of the transit sequence is not prerequisite for the translocation in vitro. However, if phosphorylation of plastiddestined precursor proteins would be complete during or after translation in vivo, then dephosphorylation by a protein phosphatase could be a regulatory step in the import process, i.e. a phosphorylated precursor would not be translocated into chloroplasts. This is not without precedent in posttranslational protein translocation. It has been reported that the nuclear import of several proteins, e.g. lamin B 2 , SV40 T-antigen, and yeast transcription factor SW15 (Hennekes et al., 1993;Jans et al., 1991;Moll et al., 1991;Rihs et al., 1991), is inhibited by phosphorylation of a serine or threonine residue near the nuclear localization signal. Only upon dephosphorylation are these proteins imported into the nucleus. To evaluate such a possibility, we used two approaches. First, thiophosphorylated precursor proteins were used for import. The thiophosphate group is not or only very slowly released from the phosphoprotein by all known protein phosphatases (McGowan and Cohen, 1988;MacKintosh, 1993). In addition we used preSSU-M54-R/D, a precursor form that imports normally but is aberrantly processed. The processing site of preSSU-M54-R/D is most likely NH 2 -terminal to the phosphorylation site; however, no thiophosphorylated imported iSSU-M54-R/D was detectable in standard import assays. The thiophosphorylated preSSU-M54-R/D bound normally to intact chloroplasts and became partially FIG. 7. Protein kinase activity is associated with chloroplasts. Protein kinase activity of cytoplasm or chloroplast supernatant (35 g of protein, respectively) was determined by performing an in vitro phosphorylation experiment in the presence of preSSU and [␥-32 P]ATP. Radioactive labeled preSSU was quantified in both reactions by laser densitometry of the autoradiograms. Sucrose synthase activity was measured in identical amounts (70 g of protein) of both fractions as described under "Materials and Methods." Maximal enzymatic activity of the protein kinase and sucrose synthase in the cytoplasm was set to 100%.
protease-protected, indicating that it had moved into the translocation machinery. The phosphorylated translocation intermediates P-TimA and P-TimB are different in size to translocation intermediates (deg 1-4) described before for wt-preSSU, which was synthesized in a reticulocyte lysate system (Waegemann and Soll, 1991). The translocation intermediates deg 1-4, which are identical to Tim1-4, thus seems to originate from the nonphosphorylated precursor form, e.g. after dephosphorylation (this study, Fig. 5B, lanes 12 and 16) (Waegemann and Soll, 1991). Tim1-4 are most prominent in import experiments in the presence of preSSU-M34-S/A, the nonphosphorylatable mutant, corroborating this idea.
Furthermore, P-TimA and P-TimB are detected from a wt-preSSU phosphorylated either by [␥-35 S]ATP␥S or by phosphorylation with [␥-32 P]ATP. The chloroplast-bound [ 32 P]preSSU yields P-TimA and P-TimB upon protease treatment, demonstrating that the thiophosphate group did not result in an artificial translocation intermediate. P-TimA and P-TimB are also detected from organellar bound 35 S-labeled preSSU phosphorylated with unlabeled ATP in the presence of NaF upon protease treatment. NaF is a protein phosphatase inhibitor, which had been shown before to reversibly inhibit preSSU import into chloroplasts (Flü gge and Hinz, 1986;Schindler et al., 1987). Since in vitro phosphorylation of precursor proteins is substoichiometric, P-TimA and P-TimB appear in combination with the nonphosphorylated Tims. Hence it is not yet possible to experimentally follow the dephosphorylation of preSSU and the concurrent appearance of SSU. It remains to be established that a cycle of precursor phosphorylation-dephosphorylation is a general and essential part of the import process in vivo. Other possibilities have also to be considered, e.g. this process could represent a regulatory phenomenon, which is activated only under certain developmental or environmental conditions to control precursor protein import into plastids.
The precursor protein protein phosphatase is most likely localized in the outer envelope membranes since P-TimA and P-TimB are membrane-bound early translocation intermediates. The precursor protein protein phosphatase would represent a new yet to be identified component of the chloroplast import machinery. Preliminary experiments to classify the protein phosphatase using specific inhibitors failed. Calyculin and microcystin, potent inhibitors of PP1 and PP2A (MacKintosh and MacKintosh, 1994), were without influence on preSSU import. NaMoO 4 (40 mM) had the same inhibitory effect as NaF, while vanadate (1 mM) (Hunter, 1995) was without influence. 3 To test for PP2B-or PP2C-specific activities is more complex in a crude system (Shenolikar, 1994) and was not tried in this study.
NaF seems to inhibit an additional step in the import pathway downstream to dephosphorylation of the precursor protein since preSSU-M34-S/A is not completely imported in the presence of NaF. Precursor import into chloroplasts requires nucleoside triphosphates at various steps in the translocation pathway, e.g. for binding  or for pulling the precursor across the membrane by the ATP-dependent action of hsc70 (Stuart et al., 1994). Furthermore, key components of the chloroplast outer envelope import machinery, i.e. OEP86 (Hirsch et al., 1994) and OEP34 (Seedorf et al., 1995;Schnell et al., 1994), are prominent phosphoproteins of this membrane. Their activity might be regulated by a phosphorylation-dephosphorylation cycle (Soll, 1995), which could also be influenced by NaF and result in the inhibition of complete preSSU-M34-S/A translocation into chloroplasts. One of these events or one yet to be identified process seems also be inhibited by NaF and impairs import.
Protein phosphatases do not merely reverse the effect, which is provoked by phosphorylation, but are highly regulatory enzymes themselves (Mumby and Walter, 1993;Hunter, 1995). Thus, the function(s) of a precursor protein protein phosphatase might be to regulate or modulate precursor uptake in dependence of the chloroplast import competency, e.g. energy or redox status. Another possibility might be the developmental regulation of import into chloroplasts during chloroplast differentiation and maturation. It has been reported that mature chloroplasts from fully expanded pea leaves loose their competence to import precursor proteins, while chloroplasts from young, rapidly expanding leaves showed highest import rates (Dahlin and Cline, 1991). Another possibility would be that the precursor protein protein kinase is not always in an activated mode in vivo, i.e. a nonphosphorylated precursor could short-circuit the protein phosphatase. It is clear, that much remains to be learned on the exact purpose of this regulatory circuit in protein import into chloroplasts.
In conclusion phosphorylation of chloroplast destined precursor proteins in the cytosol in combination with their dephosphorylation at the chloroplast envelopes might represent one step in a cascade of events which ultimately lead to specific sorting and translocation to plastids in plant cells.