Diatom Fucoxanthin Chlorophyll a/c-binding Protein (FCP) and Land Plant Light-harvesting Proteins Use a Similar Pathway for Thylakoid Membrane Insertion*

The light-harvesting proteins in plastids of different lineages including algae and land plants represent a superfamily of chlorophyll-binding proteins that seem to be phylogenetically related, although some of the light-harvesting complex (LHC) proteins bind different carotenoids. LHCs can be divided into chlorophyll a/b-binding proteins found in green algae, euglenoids, and higher plants and into chlorophyll a/c-binding proteins of various algal taxa. LHC proteins from diatoms are named fucoxanthin-chlorophyll a/c-binding proteins (FCP). In contrast to chlorophyll a/b-binding proteins, there is no information so far about the way FCPs integrate into thylakoid membranes. The diatom FCP preproteins have a bipartite presequence that is necessary to enable transport into the four membrane-bound diatom plastids, but similar to chlorophyll a/b-binding proteins there is apparently no presequence present for targeting to the thylakoid membrane. By establishing an in vitro import assay for diatom thylakoids, we demonstrated that thylakoid integration of diatom FCP depends on the presence of stromal factors and GTP. This indicates that a pathway involving signal recognition particles (SRP) is involved in membrane integration just as shown for LHCs in higher plants. We also demonstrate integration of diatom FCP into thylakoids of higher plants and vice versa SRP-dependent targeting of LHCs from pea and Arabidopsis into diatom thylakoids. The similar SRP-dependent modes of thylakoid integration of land plant LHCs and FCPs support recent analyses indicating a common origin of chlorophyll a/b- and a/c-binding proteins.

not present in modern cyanobacteria (1,2). They form a superfamily of thylakoid membrane-intrinsic chlorophyll-carotenoid proteins. The LHCs include the chlorophyll a/b-binding proteins from green algae and land plants as well as the chlorophyll a/c-binding proteins from chromophytic algae and dinoflagellates that may bind different carotenoids like fucoxanthin (FCPs in diatoms, phaeophytes, and others) and peridinin (intrinsic PCPs, many dinoflagellates) (2). Red algae only have chlorophyll a and use phycobilisomes as the major photosystem II antenna. The chlorophyll a-binding complexes isolated from the thylakoids of these organisms are also related to the LHCs of other eukaryotes (3,4). To avoid confusion we will use the term "LHC protein" for the light harvesting complexes in general, and "CAB protein" will refer to the chlorophyll a/b-binding proteins of green algae and higher plants.
The major PSII-associated chlorophyll a/b-binding proteins (LHCII) from higher plants have three ␣-helical, membranespanning regions as demonstrated by crystal structure analysis by Kü hlbrandt and co-workers (5). The first and the third helices are held together by ion pairs to ensure a compact complex to bring the carotenoids and the chlorophylls in close contact. Some CABs form trimeric homocomplexes (6). All genes encoding LHCs that are known so far reside in the nucleus. In plants, an N-terminal targeting signal ("transit peptide") allows the transport of the preproteins into the plastid stroma (7) and is exchangeable with other stroma-targeting sequences from plants (8). Targeting to the thylakoid membrane is accomplished by the mature protein itself. The third transmembrane domain is probably involved in targeting of CAB proteins to the thylakoids, whereas all three helices are needed for correct membrane insertion (9).
CAB proteins are inserted into the thylakoid membrane in a reaction requiring GTP and stromal factors (10). The integration process of the LHCs from land plants into thylakoid membranes involves binding to a chloroplast signal recognition particle (SRP), which consists of at least two proteins named cpSRP54 and cpSRP43, and to a further protein, FtsY (11)(12)(13). cpSRP54 and cpSRP43 are related to homologous proteins that are major factors of the cotranslational protein import pathway into the endoplasmic reticulum of eukaryotes (14). Overexpression and integration experiments in Escherichia coli show that integration does not depend on binding of chlorophylls and carotenoids to the LHC proteins (15).
Diatoms are representatives of the heterokont algae, which are assumed to have evolved by secondary endocytobiosis. This means that a eukaryotic host cell took up a photoautotrophic eukaryotic cell, probably an ancestor of modern red algae, followed by transformation of the endosymbiont into a plastid (16 -18). This process resulted in plastids having four bounding membranes. It has been demonstrated that protein import into these so-called complex plastids depends on a bipartite presequence and occurs in at least two steps (19).
Although diatoms and other heterokont algae represent an essential part of aquatic ecosystems, contributing significantly to the total oxygen and biomass production, little is known about the photosystems of these organisms. Recent genetic analyses of Fcp genes show that similar to genes encoding CAB proteins Fcp genes belong to a multigene family. Several different Fcp genes have already been found in diatoms, and it is likely that there are even more present (20,21).
So far there are no reports about functional aspects of FCPs with respect to integration into thylakoid membranes and cooperation with the diatom photosystems. Therefore, the characterization of integration characteristics of FCP proteins might reveal the history and the relationship of FCPs with other carotenoid-binding light-harvesting proteins. Here we present the first data on thylakoid membrane integration of a member of the FCP family using a respective protein from the diatom Odontella sinensis. We have established an in vitro thylakoid import assay from purified diatom plastids enabling the study of FCP integration. Our results suggest that FCPs are integrated mainly by an SRP-dependent integration mechanism as found for LHCs. Complementary integration experiments were performed using LHC and FCP proteins as well as thylakoid membranes from land plant and diatom plastids.

EXPERIMENTAL PROCEDURES
Constructs Made from Fcp-The Fcp gene was derived from a cDNA library of the diatom O. sinensis inserted as EcoRI and XhoI fragments into the vector ZAPII (22). For a more effective expression the Fcp gene was cloned downstream of the SP6 promoter of the pSP73 vector (Promega) using the BamHI and KpnI sites. The Lhc gene (psAB80) from pea in pSP64 was kindly provided by Prof. Kenneth Cline (University of Florida) (23). The Lhca1 gene (Lhca1 is the gene encoding chlorophyll a/b-binding protein of LHCI, type 1) from Arabidopsis thaliana in pGEM4 was a gift from Dr. Bernhard Grimm (IPK Gatersleben, Germany).
Deletions of coding regions from Fcp were made by polymerase chain reaction using Pfu polymerase (Stratagene) according to the manufacturer's instructions. All modifications were verified by double-strand sequencing using the T7 sequencing kit (Amersham Pharmacia Biotech). The deletion of the signal sequence resulting in the intermediate FCP form, iFCP (FCP⌬1-15 F16M), was achieved by amplifying the Fcp gene including the vector and the Fcp gene but excluding the coding region for the signal sequence (primers 5ЈGCATCGATCATGGCCCCG-GCTCAGTCC3Ј and 5ЈCATGATCGATGCAGAGCCGGCGAGGAG3Ј). For intramolecular ligation of the polymerase chain reaction product a ClaI restriction site was introduced by the primers (underlined). For constructing the mature form of the FCP, mFCP (FCP⌬1-29), the cloned Fcp gene was cut with the restriction enzymes StuI and SmaI. After intramolecular ligation, the remaining gene was transferred into BamHI and KpnI sites of SP72 (Promega). All modified Fcp constructs were transcribed with SP6 RNA polymerase. Within the primer for the intermediate Fcp form (iFCP) a new start codon was included (bold letters). For expression of the mFCP translation product, the original ATG codon at position 31 was utilized.
Preparation and Subfractionation of Pea Chloroplasts and Import Experiments-Chloroplasts from pea were isolated (19) from 8-to 12day-old seedlings. Posttranslational import reactions, washing, and subfractionation of the plastids were performed as described. For protease treatment thermolysin (Fluka) was added to a final concentration of 0.2 mg/ml together with 2 mM CaCl 2 . Thylakoids were isolated according to Ref. 24.
Preparation of Chloroplasts and Thylakoids from Diatoms-The diatoms O. sinensis and Coscinodiscus granii were cultured as described (25). Plastids were isolated by gentle breakage of the cells in a Yeda press and subsequent separation of broken and intact plastids by centrifugation on a 40% Percoll cushion. Chlorophyll concentrations were determined as described (25). Isolated plastids were routinely checked for integrity by measuring oxygen evolution under illumination (25).
For thylakoid preparations the plastids were collected by a short centrifugation step (3,000 ϫ g, 1 min) and incubated subsequently for 2 min in HM medium (50 mM HEPES, pH 8, 5 mM MgCl 2 ) to break the plastids osmotically (at a concentration of 1 mg of chlorophyll/ml). Just before recovering the thylakoids by centrifugation at 15,000 ϫ g for 1 min, 1 volume of H2SM buffer (50 mM HEPES, pH 8, 680 mM sorbitol, 5 mM MgCl 2 ) was added. The resulting supernatant (stromal fraction) was kept on ice. The pelleted thylakoids were washed two or three times in HSM buffer (50 mM HEPES, pH 8, 340 mM sorbitol, 5 mM MgCl 2 ) to remove remaining stromal contaminations (15,000 ϫ g, 1 min) and were thoroughly resuspended in the stroma fraction or in HSM buffer at a concentration of 1 mg of chlorophyll/ml. EDTA-washed thylakoids were obtained by osmotic breakage of the plastids with HME buffer (50 mM HEPES, pH 8, 5 mM MgCl 2 , EDTA 10 mM) and washing of the thylakoids with HSME (HSM plus 10 mM EDTA). Protease-treated thylakoids were obtained by incubating osmotically broken thylakoids in HSMC buffer (50 mM HEPES, pH 8, 340 mM sorbitol, 5 mM MgCl 2 , 2 mM CaCl 2 ) including the protease thermolysin at a concentration of 0.2 mg/ml for 15 or 30 min. The protease was inactivated by adding 2.5 mM EGTA and was removed by washing in a larger volume of HSME 2 (HSME including 2.5 mM EGTA).
In Vitro Translation, Import Reactions, and Protease Treatment-Genes encoding LHC and FCP proteins were transcribed and translated in a coupled transcription/translation system (TNT System, Promega) as described in the manufacturer's introductions using [ 35 S]methionine (Amersham Pharmacia Biotech).
The conditions for the thylakoid integration experiments with pea or diatom thylakoids were similar except for the different osmotic conditions. For import experiments 45 l of thylakoids (0.5-1 mg of chlorophyll/ml) and 10 l of translation reaction were used. To distinguish between Sec-or SRP-dependent integration, thylakoids were resuspended in the stromal fraction. Inhibitors for different import pathways were added as described. ATP and GTP were supplied at concentrations of 8 mM. The integration reactions were started by adding the translated protein and incubating the assays at 25 (pea thylakoids) or 16 -18°C (diatom thylakoids) for 25-30 min. Some samples were illuminated (150 mol of photons/m 2 s). The reactions were stopped by placing the samples on ice.
All samples were divided into three parts. The first part was used directly for electrophoretic analysis and was taken as control (100%). The second and third parts were centrifuged, and the thylakoids were resuspended in HSM or HSMC, respectively, depending on optional further protease treatment. The conditions for protease treatment were identical as for the preparation of protease-treated thylakoids. All nonintegrated proteins were digested by thermolysin during an incubation on ice for 30 min. After deactivation of the protease with 2.5 mM EGTA, the thylakoids were collected and immediately prepared for the electrophoresis.
The third part of the import assay was washed in different solutions to verify the integration of LHC/FCP proteins. According to Breyton et al. (26), the thylakoids were incubated with five volumes of 2 M KSCN, 2 M NaCl, 6.8 M urea, or 85 mM Na 2 CO 3 /DTT, respectively, for 10 min at room temperature, were vortexed, and were subjected to two freeze/ thaw cycles including additional vortexing. The washed thylakoids were recovered by a 10-min centrifugation (22,000 ϫ g). To test the stability of FCP integration, this washing procedure was done three times, whereas in standard experiments all samples were washed once with 2 M KSCN. Afterwards, the thylakoids were washed in 100 l of HSM to remove the potassium ions before preparing the samples for electrophoresis.
To check the membrane specificity of FCP integration, 3 l of translation mixture and 10 l HSM were added to 4 l of microsomes. After incubation at 18°C for 30 min, the samples were washed with 100 l of 2 M KSCN as described above. Before recovering the membranes the samples were diluted to 3.5 ml with washing buffer. The microsomes were washed by centrifugation at 115,000 ϫ g for 45 min.
Electrophoresis, Fluorography, and Quantification-Samples were denatured by adding sample buffer and incubation for 3 min at 90°C and analyzed on 15% SDS-PAGE (27). For visualization of the radioactively labeled proteins, the gels were fixed in 30% ethanol, 10% acetic acid and soaked in Amplify (Amersham Pharmacia Biotech) before drying. Fluorographic signals were detected utilizing Kodak X-Omat x-ray film (Eastman Kodak Co.). Quantifications were performed on a PhosphorImager (fluoro-imager BAS-1800, Fuji). For calculation of integration efficiency, the protein bands in the control sample were set to 100%.

Construction of FCP Precursor
Proteins-Different constructs were made from the diatom FCP precursor protein by deleting regions of the Fcp gene coding for different parts of the presequence (Fig. 1). iFCP was constructed by deleting the signal peptide domain necessary for cotranslational transport across the chloroplast ER (the outermost of the four diatom plastid-bounding membranes) and therefore represents the putative intermediate preprotein possessing a transit peptide domain only. The position of the new start codon was chosen on the basis of prediction of possible signal peptidase cleavage sites according to the "Ϫ3,Ϫ1" rule of von Heijne (28) indicating a cleavage site at Ala 15 and Phe 16 . mFCP represents the mature form as found within the plastids (20).
Isolation of Functionally Intact Diatom Thylakoids-To analyze the integration mode of FCPs into the thylakoid membranes from diatoms, we established an in vitro import assay for diatom thylakoids. Functionally and morphologically intact plastids were isolated from the marine centric diatoms O. sinensis and C. granii and purified by centrifugation through a Percoll cushion as described (25), followed by osmotic rupture of the plastids. The integrity of the resulting thylakoids was analyzed by measuring electron transport capabilities (the isolated thylakoids were routinely checked and typically yielded ϳ80 mol of O 2 ⅐mg chlorophyll Ϫ1 ⅐min Ϫ1 in the presence of methyl viologen and ADP/P i ) and the stability of the lightinduced proton gradient using the fluorophore 9-aminoacridine (29). The ⌬pH-dependent fluorescence quench was measured as described (30). The thylakoid membranes were able to build up a stable proton gradient after illumination. This proton gradient was sensitive to low amounts of the uncoupler nigericin (data not shown). One important factor for the ability of the thylakoids to build up a stable pH gradient are the conditions used for osmotic breakage of the plastids. We found that osmotic rupture of the plastids is possible with low osmotic buffer for 1 min, but subsequently the thylakoids need to be replaced in a buffer containing at least 300 mM sorbitol to avoid swelling and increase of proton permeability.
Analysis of Membrane Integration of FCP Proteins-LHC proteins do not get proteolytically processed during or after membrane integration; therefore, integration cannot be visualized by a shift in molecular mass. Structural comparisons between LHCs and FCPs indicate that this might be similar in FCPs as there is obviously no cleavable presequence present for thylakoid targeting (21). In this study we used varying stringent washing procedures of the membranes, and we compared the results with protease protection assays. These procedures were established to distinguish between loosely membranebound protein and membrane-integrated protein. To exclude the possibility that parts of the presequence needed for plastid import might be involved in the integration reaction, we analyzed integration of mFCP as well as of iFCP. Both precursor proteins show similar integration characteristics.
According to Breyton et al. (26) we checked different procedures for removing peripheral proteins from thylakoid membranes at conditions where no FCP integration should occur (no addition of stromal extract and GTP). Fig. 2A demonstrates that urea and KSCN have the best potential to elute loosely bound FCP protein from diatom thylakoid membranes after incubation for 10 min followed by harsh vortexing and freeze/ thaw cycles. After addition of radiolabeled FCP protein to the thylakoids and subsequent incubation, they were treated as indicated. The amount of residual protein was determined by separating thylakoid membrane proteins by SDS-PAGE and subsequent quantitative analysis of the residual radioactivity of the FCP band using a PhosphorImager. Thylakoids that were washed with buffer only served as control (100%). First experiments were performed by incubation of thylakoids and radiolabeled FCP protein without addition of further substrates, allowing only minor integration of FCP. Subsequent washing of the membranes with 6.8 M urea, 2 M KSCN, and 2 M NaCl, respectively, resulted in a residual amount of radiolabeled FCP protein of 7-18% (mFCP) of thylakoid-associated protein ( Fig. 2A). In contrast washing with 85 mM Na 2 CO 3 /DTT was not suitable to extract loosely bound proteins (77% residual protein). To improve the KSCN washing procedure, we tested the integration stability of mFCP by additional washing steps including vortexing and freeze/thaw cycles (Fig. 2B). First washing of the membranes with 2 M KSCN resulted in a remaining radioactivity of 27%, whereas during the following washing steps only a marginal further extraction of the membrane protein occurred. Repeated washing steps with urea, however, resulted in a nearly complete removal of FCP probably due to membrane disintegration by the harsh conditions. None of the washing procedures resulted in a clearly defined amount of residual membrane-bound protein under varying conditions. Based on our results repeated washing steps with 2 M KSCN were the best choice to remove most of unspecifically bound proteins and to allow reproducible results for analyzing the effectiveness of FCP integration in the following experiments.
In addition to the washing steps the incorporation of the FCPs into thylakoid membranes was monitored by utilizing the protease thermolysin. It has been demonstrated that some of the CAB proteins from higher plants are protected against the proteases trypsin and thermolysin when integrated into the thylakoid membrane (24). In our experiments a limited incubation of thylakoids following FCP integration assays (0.2 mg/ml thermolysin, 2 mM CaCl 2 ) on ice for 20 min was sufficient to degrade peripheral and free FCP, while leaving integrated protein intact (data not shown). Therefore protease protection assays were used to verify the results obtained by washing with KSCN.
Integration of FCP into Diatom Thylakoids-Factors needed for FCP integration into diatom thylakoids were analyzed by incubation of mFCP and iFCP, respectively, from O. sinensis with diatom thylakoids under a variety of different experimental conditions. Both proteins indicated identical integration features as shown for mFCP (Fig. 3). Comparable results were also obtained with thylakoid preparations from both diatoms O. sinensis and C. granii. By using the washing procedure with 2 M KSCN as described above, we found a low amount of membrane-integrated FCPs when only thylakoids and radioactively labeled FCP proteins were incubated. A clear enhancement of FCP integration was obtained by addition of stromal extracts and GTP (8 mM) during incubation, after washing, or protease treatment. In this case an average of 2.5-3 times higher amounts of FCP remained in the thylakoid membranes. Variations of the concentration of GTP during incubation show a clear GTP dependence of FCP integration (Fig. 4). A GTP concentration of about 12 mM turned out to be sufficient for maximum FCP integration. Thylakoid washing and protease treatment resulted in comparable results (Fig. 3).
Different integration conditions were checked to exclude the

FIG. 1. N-terminal presequence domains of FCP precursors and derived constructs used in this study showing predicted signal peptides and transit peptides as indicated.
For construction of iFCP the signal peptide domain and for mFCP the complete presequence have been removed and substituted by new methionine residues to enable the respective new starts of translation. Dots indicate that only a partial sequence of the mature proteins is shown. possibility that other known protein integration pathways might also be involved in FCP integration. The Sec system is involved in the import reaction of the lumenal OEE33 protein (31) and depends on the presence of stromal factors and of ATP. Addition of stromal extract and/or ATP to the integration assay did not result in an obvious enhancement of FCP integration above the level obtained without additions, which would be expected if the Sec system would also be involved (Fig. 3). To make sure that the low level of FCP integration obtained with-out any additions was not due to the presence of ATP or stromal factors carried over with the thylakoid preparation, we added an inhibitor of the Sec pathway, sodium azide (5 mM), as well as the ATP-hydrolyzing enzyme apyrase. In both cases integration was comparable to experiments without any additions. In land plants lumenal proteins like OEE23 and OEE16 are transported into the thylakoids by a pH gradient (31). Apparently there is no additional ⌬pH-depending FCP integration because performing FCP integration in the light or in the dark as well as the addition of nigericin, which abolishes transmembrane proton gradients, did not show obvious effects.
The integration of the FCPs is not disturbed by the presence of the N-terminal presequence which is necessary for targeting of pre-FCP into the plastids in vivo. The full-length FCP precursor integrates into the thylakoid membrane as well as the intermediate iFCP or the mature form mFCP. This is consistent with investigations of the higher plant LHCs (32). Spontaneous Integration of FCP?-After repeated washing steps or protease treatment as described above, we always found a certain amount of FCP to be associated with the thylakoid membrane even in the absence of stromal factors or GTP. This might be due to factors carried over from the in vitro translation system (rabbit reticulocyte lysate) or insufficient removal of unspecifically bound proteins. However, a stronger dilution of the added radiolabeled FCP protein in the integration assay still resulted in membrane-associated FCP protein.
Most other translocation pathways known from higher plant thylakoids have been ruled out by our experiments; however, it is possible that other so far unknown transporters might assist FCP integration. Therefore, we treated diatom thylakoids with thermolysin (0.1 mg/ml) prior to addition of FCP protein to degrade any protein translocators that might be actively involved in protein insertion. Additionally, we tried to extract electrostatically bound proteins from the thylakoids by pretreatment with 5 mM EDTA. In both cases the same amount of associated FCP protein was detected as in control assays (data not shown). This might suggest that a part of the FCP proteins might be able to integrate spontaneously as reported for the CF 0 II subunit of chloroplast ATPase from higher plants or ELIP proteins (24,33). To verify this we analyzed FCP integration into a completely different type of membrane, pancreatic ER microsomes. We incubated radioactively labeled iFCP protein with microsomal vesicles for 30 min. After washing the membranes repeatedly with 2 M KSCN including freeze/thaw cycles, we still found 13% of the radioactivity after recovering the membrane pellet (not shown).
Integration of FCP Protein into Higher Plant Thylakoid Membranes-To compare CAB and FCP integration in the same experimental system, we incubated LHC proteins from pea and from Arabidopsis and FCP from Odontella individually with thylakoids isolated from pea plastids. As shown in Fig. 5, the integration characteristics of LHC from pea into diatom thylakoid membranes under various conditions are similar to the respective experiments with FCP integration (Fig. 3). Only the addition of GTP and stromal proteins resulted in a high rate of FCP integration. Similar results were obtained for LHC from Arabidopsis. In reciprocal experiments we could show that integration of diatom FCP into thylakoid membranes from pea also depends on GTP and stromal factors (Fig. 6). DISCUSSION LHC proteins in all organisms analyzed so far have to cross at least two membranes before being inserted into the thylakoid membrane. In higher plants and in green and red algae, two plastid envelope membranes have to be traversed, whereas in heterokont algae FCPs have to be transported across four membranes to enter the stromal compartment. In both cases N-terminal presequences are utilized for correct targeting. Presequences of FCPs and other nucleus encoded plastid preproteins from diatoms have a bipartite structure. The individual properties of these two domains have been demonstrated in vitro (19) and in vivo. 2 In plastids of higher plants several pathways have been described for insertion of proteins into the thylakoid membrane or targeting to the thylakoid lumen (for review see Ref. 31). Lumenal proteins like plastocyanin or the subunit of the oxygen evolving system OEE33 were found to be transported by a system homologous to the bacterial Sec system. In contrast to the Sec pathway, a ⌬pH-dependent system allows the translocation of more tightly folded proteins (twin arginine translocase; see Ref. 34). For both systems an Nterminal targeting domain is required for transport, which is found between the transit peptide for chloroplast import and the mature protein. Some proteins that need to be inserted into the thylakoid membrane use different mechanisms. Insertion of CAB proteins is dependent on the presence of GTP and different stromal proteins, which are homologous to the SRP transport system at ER membranes. Initial reports demonstrated that integration into thylakoid membranes required ATP (35), but later experiments demonstrated that GTP promotes membrane insertion 10 times more effectively (10). Other proteins that are related to CAB proteins like early light-inducible proteins (ELIPs), PsbB, and PsbW are sug-  gested to integrate spontaneously into the membrane without the aid of proteins or energetic components (24). It is still unclear whether these different integration processes are due to structural differences between CAB and ELIP proteins. The mode of thylakoid membrane insertion of LHC proteins so far has been analyzed for CAB proteins of higher plants only. In this study we therefore have addressed the question whether FCP proteins follow the same insertion pathway as CAB proteins and whether this process might be a general feature for LHC proteins from all groups of photoautotrophic eukaryotes.
To analyze the conditions for integration of FCP from a diatom in a homologous system, we have established a procedure to isolate functional thylakoids from isolated diatom plastids. In contrast to thylakoids of higher plant plastids in diatoms, no grana structures are found except lamellae consisting of triple thylakoid stacks. The main problem obtaining functionally intact thylakoids from marine diatoms, indicated by the capability to build up stable proton gradients, was to find osmotic conditions to break up the plastids but to avoid osmotic rupture of the thylakoids. Previous thylakoid preparations from diatoms resulted in thylakoid membranes that showed electron transport activities, but no stable proton gradients have been demonstrated yet (36). Isolated plastids from marine diatoms need approximately twice the sorbitol concentration as used for chloroplasts from higher plants to avoid osmotic rupture and to keep plastidic functional integrity, which has been demonstrated by light-dependent oxygen evolution (25). Therefore, for thylakoid preparations we limited the osmotic rupture of the plastids in a low osmotic buffer to a very short time before increasing the osmolarity again to stabilize the thylakoids. Another advantage of the preparation of thylakoids from purified intact plastids is the possibility to obtain pure stromal extracts for supplementation during integration assays.
It turned out to be difficult to determine the exact amount of FCP integrated into the thylakoid membrane. Most washing procedures were strong enough to remove unspecifically bound FCP protein, but the procedures turned out to be very harsh resulting in a removal of most of the protein when employed repeatedly. However, using one single washing step with KSCN and standardization of the procedure resulted in reproducible results that were supported by the protease protection assays allowing the analysis of effectors on protein insertion. Our integration experiments revealed a very similar membrane integration behavior for FCPs and CAB proteins. Integration of FCP was depending on stromal factors and on the presence of GTP. Variations in the GTP concentration clearly show a dependence on GTP and resulted in a saturation of FCP integration at concentrations of 10 -12 mM GTP. This result is in agreement with integration experiments of CAB proteins into land plant thylakoids (10). A slight effect of ATP on insertion was observed repeatedly in the protease assays.
Despite harsh washing procedures and protease treatment after incubation, we always found a certain amount of FCP protein cosedimenting together with the thylakoid membranes even in the absence of GTP and stromal proteins. The same result was obtained even after pretreatment of the thylakoids with protease to destroy proteins possibly being involved in protein translocation. Even in heterologous membranes like ER vesicles from canine pancreas, a certain amount of FCP was recovered after incubation and subsequent washing/centrifugation steps. The membrane washing procedures with chaotropic salts we applied have been demonstrated to be effective in removing loosely bound proteins (26). It remains unclear whether they were effective enough to remove all of the unspecifically bound FCPs in our experiments. Protease treatments of the thylakoids after the integration reaction indicate that in fact a certain amount of FCPs might be protease-protected. On the other hand CAB integration into diatom thylakoids and vice versa also resulted in a certain amount of residual FCP protein in the absence of stimulating factors. This could mean that so far we are not able to distinguish between spontaneous FCP integration and unspecifically bound FCP protein. A spontaneous integration of FCPs would be in contrast to CAB proteins, which clearly need GTP and SRP proteins. LHC proteins were suggested to have evolved from duplication and fusion of monospanning high light-inducible proteins, an ELIP-related protein (37). The result that ELIPs can integrate spontaneously into the thylakoid membrane as well as assisted by the SRP complex (24) indicates that these proteins might be able to follow parallel insertion pathways. If this should turn out to be true for FCPs, both pathways might have existed for insertion of the ancestor of present day LHC proteins. Perhaps such spontaneous integration capabilities were lost during the evolution of CAB proteins. Analysis of the integration mode of CABs from more "primitive" green algae might give further insight into this question.
Phylogenetic analyses have shown that FCP proteins are more closely related to LHC proteins from red algae and to PCP proteins from dinoflagellates than to CAB proteins (1), which reflects the putative phylogenetic origin of heterokont plastids from red algal ancestors (16). Separation of red and green algae (the latter leading to the land plants) from a common ancestor is supposed to have occurred early in the evolution of autotrophic eukaryotes. We demonstrated that the diatom FCP can integrate into pea thylakoids as well as pea LHC into diatom thylakoids. This is very interesting because it shows that both protein types use the same apparently very conserved import machinery. This cross-functionality in such divergent organisms as diatoms and land plants indicates that the SRP-dependent integration of light-harvesting proteins may have been developed very early in the evolution of plastids and has essentially remained unchanged. This is especially surprising as cyanobacteria, which are thought to be related to the ancestors of plastids, apparently do not have light-harvesting proteins, but phycobilisomes instead. Although it is still possible that the cyanobacterial ancestor of plastids differed from modern cyanobacteria, this indicates that the light-harvesting proteins as well as the needed thylakoid-targeting system might have evolved shortly after primary endocytobiosis. The genetic history of the transfer of Fcp genes during secondary endocytobiosis from the endosymbiont genome to the nucleus of the host cell is also unclear. As FCPs seem to be generally nucleusencoded, it is likely that they might have been transferred to the nucleus after primary endocytobioses and therefore had to be transferred in a second step after secondary endocytobiosis, this time from the nucleus of the endosymbiont to the nucleus of the second host. An intron within the Fcp gene of the related brown algae Laminaria (38) between signal and transit peptide domain of the FCP precursor and phylogenetic analyses of light-harvesting proteins from cryptomonads and chloroarachniophytes in fact suggest a two-step gene transfer (39).