Chloroplast YidC homolog Albino3 can functionally complement the bacterial YidC depletion strain and promote membrane insertion of both bacterial and chloroplast thylakoid proteins.

A new component of the bacterial translocation machinery, YidC, has been identified that specializes in the integration of membrane proteins. YidC is homologous to the mitochondrial Oxa1p and the chloroplast Alb3, which functions in a novel pathway for the insertion of membrane proteins from the mitochondrial matrix and chloroplast stroma, respectively. We find that Alb3 can functionally complement the Escherichia coli YidC depletion strain and promote the membrane insertion of the M13 procoat and leader peptidase that were previously shown to depend on the bacterial YidC for membrane translocation. In addition, the chloroplast Alb3 that is expressed in bacteria is essential for the insertion of chloroplast cpSecE protein into the bacterial inner membrane. Surprisingly, Alb3 is not required for the insertion of cpSecE into the thylakoid membrane. These results underscore the importance of Oxa1p homologs for membrane protein insertion in bacteria and demonstrate that the requirement for Oxa1p homologs is different in the bacterial and thylakoid membrane systems.

Two pathways have been discovered in bacteria to insert membrane proteins into the inner membrane. The Sec-dependent pathway is used for most membrane proteins, while the Sec-independent pathway is used only for a few tested membrane proteins (for review see Refs. [1][2][3]. The cast of characters for the Sec-dependent pathway is SecA, Y, E, and G (4). SecY and SecE polypeptides are believed to form the core of the protein-conducting membrane channel (5,6), and SecA is the translocation ATPase (7) that catalyzes translocation of long periplasmic loops (8 -10). Known Sec-dependent membrane proteins are leader peptidase (Lep), 1 MalF, FtsQ, and AcrB (11)(12)(13)(14)(15). These proteins are targeted to the membrane by the signal recognition particle pathway involving Ffh and 4.5 S RNA, and FtsY (16 -18). Recently, a new component called YidC has been found to be associated with the Sec machinery. YidC appears to function in membrane protein insertion based on its interaction with the transmembrane region of FtsQ (14), leader peptidase (19,20), and mannitol permease (21) during their integration. Until very recently, the Sec-independent pathway was believed to function without the aid of a protein machinery (22,23). However, recent studies indicate that YidC is required for the membrane insertion of the Sec-independent procoat and pf3 coat protein (20,24,25).
In mitochondria and chloroplast, YidC homologs exist that have been shown to mediate membrane protein insertion (for review see Ref. 26). The mitochondrial homolog, Oxa1p, is required for the membrane insertion of a subset of inner membrane proteins (27)(28)(29). These Oxa1p-dependent proteins include proteins that are both matrix-encoded and nuclear-encoded that are imported from the cytosol into the matrix and then inserted into the inner membrane. The chloroplast homolog, Albino3 (Alb3), is required for integration of a subset of the light-harvesting chlorophyll-binding proteins (LHCP) into the thylakoid membrane (30 -32).
In this paper, we test whether the chloroplast Alb3 can functionally complement a YidC depletion strain. We find that Alb3 can complement the growth defect of a YidC depletion strain and that Alb3 promotes the insertion of Lep and M13 procoat protein as well as the chloroplast SecE (cpSecE) protein into the bacterial inner membrane. Surprisingly Alb3 is not required for the membrane insertion of the cpSecE into the thylakoid membrane. These studies emphasize the importance of Oxa1p homologs for insertion of proteins into the E. coli inner membrane and reinforce the recent results that membrane insertion into chloroplast thylakoid membrane can be independent of Alb3 (31,32).

EXPERIMENTAL PROCEDURES
Strains and Plasmids-E. coli strain JS7131 (MC1060, ⌬yidC, attB::R6Kori, ParaBAD Ϫ yidC ϩ , (Spec r )) was from our laboratory stocks. pMS119, which contains the tac promoter and lacI q , was used to express procoat and leader peptidase (Lep). pQE80 and pEH1 vectors, which contain the IPTG-inducible lacUV5 promoter, were used to express the light harvesting chlorophyll-binding protein (LHCP) and cpSecE, respectively. The cDNA sequence of Alb3 was amplified using the Arabi-dopsis thaliana cDNA library (from Arabidopsis Biological Resource Center at The Ohio State University) and cloned into pACYC184 (from New England Biolabs). The upstream region of the yidC gene including the promoter and ribosome-binding site was then cloned into the vector upstream from the alb3 start codon, yielding plasmid pACYC-Alb3. pACYC-mAlb3 was derived from pACYC-Alb3 where the predicted stroma-targeting sequence (1-55 amino acids) was replaced with a single methionine. XhoI and HpaI sites were introduced into the yidC and alb3 genes by site-directed mutagenesis, and then the two XhoI-HpaI fragments corresponding to 1-57 amino acids of YidC or Alb3 were swapped. The resulting plasmid, pACYC-H1Alb3, contains a yidC and alb3 fusion, which includes the first 57 amino acids of YidC, a linker valine residue, and 59 -462 amino acids of Alb3. The three-gene constructs, Alb3, mAlb3, and H1Alb3, were also cloned into pMS119 under the tac promoter to overexpress the plasmid-encoded Albino3 proteins. The first 57 amino acids of YidC and H1Alb3 were replaced by 1-57 amino acids from the precursor to the maltose-binding protein (preMBP), yielding two constructs: MBP-YidC and MBP-Alb3.
Clones for in vitro transcription and translation have been described previously: pLHCP (33), iOE33 (34), and t23-PCm (35). PCR-based cloning was used to create mature cpSecE with a C-terminal Met. The C-terminal Met residue was added in order to visualize a proteaseprotected fragment of integrated cpSecE in thylakoid membranes. Using the plasmid pEH1cpSecE (36) as template, the mature cpSecE coding sequence was amplified by PCR using the forward primer 5Ј-GCGAATTCACCATGGCGACGAGTAATCTG-3Ј and the reverse primer 5Ј-GGTGTTCAAGACTTCTTCAGCATGTGAC-3Ј, which includes the ATG codon before the stop codon. An EcoRI site was included in the forward primer allowing the PCR products to be ligated into pGEM-4Z restricted with EcoRI and HincII. To place pElip2 in a direction suitable for transcription with SP6 RNA polymerase, PCR-based cloning of pElip2 was performed. To accomplish this, we used the parent plasmid (37) as a template and the forward primer, 5Ј-CACGAT-GGCCACAGCGTCGTTTAACATGCAGTCAG-3Ј, which added the codons for three additional residues, methionine, alanine, and threonine, at the N terminus. The M13/pUC reverse sequencing primer (New England Biolabs) was used as the reverse primer. The PCR fragment was restricted with HindIII and ligated into SmaI-and HindIII-restricted pGEM-4Z. The fidelity of the PCR was checked by DNA sequencing.
YidC Depletion-Unless otherwise indicated, JS7131 was grown overnight in LB medium with 0.2% arabinose. After washing with fresh LB twice, the overnight culture was 1:50 back-diluted into 1 ml of LB with either 0.2% glucose to deplete YidC or 0.2% arabinose to express YidC. For JS7131 bearing the pACYC-H1Alb3, the overnight culture was cultivated in the absence of arabinose and directly 1:50 backdiluted into 1 ml of LB with 0.2% glucose to suppress the expression of YidC. The cultures were grown at 37°C for about 3 h until a significant growth defect was observed between the glucose and the arabinose culture. Then the cells were transferred to M9 medium containing either 0.2% glucose or 0.2% arabinose, and the cultures were grown for an additional 30 min. At this point, the cells were ready for radiolabeling, protease-mapping, and signal peptide processing analysis (see below).
Protease Mapping Assay-To test the Alb3 requirement for membrane protein insertion, the membrane protein of interest was induced for 5-10 min by the addition of IPTG (1 mM, final concentration) to JS7131 cells prepared as above and labeled with trans-[ 35 S]methionine for 20 s. The sample was chilled on ice immediately, and the cells were collected by centrifugation. The cell pellet was subsequently resuspended in 0.5 ml of ice-cold sphero-buffer (40% sucrose, 33 mM Tris-HCl, pH 8.0) and treated with lysozyme (10 g/ml final concentration) for 15 min. Then the sample was split, and proteinase K was added to one aliquot. After 1 h of incubation on ice, 0.5 ml of chilled 20% trichloroacetic acid was added to each aliquot. The trichloroacetic acid-precipitated proteins were then analyzed by immunoprecipitation and SDS-PAGE (38) and by phosphorimaging.
Signal Peptide Processing Assay-Signal peptide cleavage was also used to monitor membrane insertion of proteins. Procoat synthesis was induced for 5 min in the JS7131 strain bearing pMS119-procoat, which was grown as described above, by the addition of 1 mM IPTG and subsequently labeled with trans-[ 35 S]methionine for 20 s. The radiolabeled sample was directly added to ice-cold trichloroacetic acid. The trichloroacetic acid-precipitated proteins were then subjected to immunoprecipitation and analyzed by urea-SDS-PAGE and phosphorimaging.
Protein Separation and Detection of E. coli Proteins-Samples were analyzed by SDS-polyacrylamide gel electrophoresis using a 15% poly-acrylamide gel except for procoat samples, which were analyzed using a 22% Urea-SDS-polyacrylamide gel. Radiolabeled proteins were visualized either by phosphorimaging or by fluorography. YidC and H1Alb3 proteins were detected by immunoblotting assay using the ECL Western blotting detection kit (Amersham Biosciences).
Synthesis of Radiolabeled Precursors-Precursor proteins were synthesized from capped RNA using a wheat germ system in the presence of [ 35 S]methionine as described (39). Translation products were diluted 3-fold and adjusted to import buffer (IB; 50 mM Hepes/KOH, pH 8.0, 0.33 M sorbitol) containing 30 mM unlabeled methionine.
Protein Transport Assays Using Isolated Thylakoids-Standard transport assays included 50 l of 2ϫ thylakoids (1 mg/ml chlorophyll) in IB/10 mM MgCl 2 (IBM), 50 l of 2ϫ S.E. (equivalent to that obtained from 2ϫ chloroplasts), 5 mM ATP, 1 mM GTP, and 25 l of diluted radiolabeled precursor protein. IB was added to make the final volume 150 l. After incubation at 25°C for 30 min, the thylakoid membranes were pelleted (3200 ϫ g for 8 min) and resuspended in 500 l of IB. The amount of inserted protein was assayed by adding 25 l of thermolysin (2 mg/ml in IB/10 mM CaCl 2 ) and incubating 40 min on ice. Thermolysin treatment was terminated by addition of 100 l of IB/50 mM EDTA. Samples containing in vitro translated cpSecE were acetone-precipitated (41) and resuspended in 45 l of solubilization buffer. Otherwise, pelleted membranes were resuspended in 20 l of 10 mM EDTA and 25 l of 2ϫ solubilization buffer. After incubation at 70°C for 15 min, proteins contained in 10 l of each sample were separated by SDS-PAGE except for samples from cpSecE integration assays, which were separated by Tricine SDS-PAGE (42). Gels were dried and used to quantitate the level of protein transport from phosphorimages (Molecular Dynamics Typhoon).
Assay for Antibody Inhibition of Protein Import into Thylakoids-Antibodies against Alb3, cpSecY, and Tha4 were tested for their ability to inhibit insertion into thylakoids as described in (30). Overexpressed SecE and Alb3 are from Arabidopsis. After incubation with isolated thylakoids for 1 h, the unbound antibody was removed by washing the thylakoids in IBM, and standard transport assays were conducted, except that S.E. was included at twice the standard concentration.
Energetics of Thylakoid Import-Standard transport assays lacking additional ATP were used to test energetic requirements of precursor transport. IB was substituted for S.E. as indicated in the legend to Fig.  6 to examine the requirement for soluble factors (35). The non-hydrolyzable ATP and GTP analogs, AMP-PNP and GMP-PNP, were inspected for their effect on insertion by addition to 5 mM final concentration. The requirement for a ⌬pH was tested by including the proton ionophore nigericin (Fluka, 1 M final concentration in ethanol) as described (43). Control reactions contained ethanol without nigericin.
Import into Protease-treated Thylakoids-Thylakoids (0.18 g/ml chlorophyll) were treated with thermolysin (180 g/ml ϩ 0.9 mM CaCl 2 ) for 40 min on ice followed by the addition of EDTA as above to stop the reaction. Thylakoids were repurified over 7.5% Percoll cushions in IB/10 mM EDTA and prepared for import assays by washing two times in IB/10 mM EDTA and once in IB. The membranes were then resuspended in IBM containing 1 mM dithiothreitol and used in standard transport assays. Mock protease-treated thylakoids were used in control assays.

Alb3 Complements the Growth Defect of a YidC Depletion
Strain-To test whether the chloroplast Alb3 is a functional homolog of the E. coli YidC, we investigated whether Alb3 could complement the YidC depletion strain, JS7131. The YidC depletion strain, which is arabinose-dependent for growth, has an arabinose-inducible yidC gene permanently integrated at the attB locus of MC1060 and contains an in-frame deletion at the normal yidC locus. Complementation was assayed by testing whether Alb3 can restore growth of JS7131 on LB agar plates containing glucose.
The A. thaliana Alb3 is predicted to contain only five transmembrane segments (44) (see Fig. 1A) while E. coli YidC has six transmembrane segments with an N-terminal signal anchor domain (45). Alb3 lacks most of the N-terminal-translo-cated domain including the first transmembrane segment found in YidC. In our complementation studies, we made three specific Alb3 constructs. The first construct, Alb3, is the fulllength Alb3 (amino acids 1-462), including the stroma-targeting peptide. The second construct, mAlb3, lacked the stromatargeting peptide (amino acids 1-55). The third, H1Alb3, consisted of the first 57 amino acids of YidC joined to residues 59 -462 of Alb3. The YidC portion contains transmembrane segment one that functions as an uncleaved signal sequence. These pACYC184 constructs were introduced into the YidC depletion strain JS7131. JS-H1Alb3 grew under YidC-depleted conditions (0.2% glucose), while JS-Alb3 and JS-mAlb3 did not (Fig. 1C), showing that H1Alb3 complements the growth defect of YidC-depleted bacteria. No complementation was observed for the wild-type Alb3 either with or without the chloroplast stroma-targeting signal.
We next confirmed that JS7131 expressing H1Alb3 does not express YidC when grown in glucose. Immunoblot analysis using YidC antiserum revealed that there was no detectable YidC in JS7131 expressing H1Alb3 that was grown in glucose ( Fig. 2A). The top band shown in Fig. 2A is a nonspecific band that is precipitated by our YidC antiserum. As a control, we confirmed that JS7131 only synthesizes YidC in the presence of arabinose (left lane) but not in glucose (middle lane). Western blotting analysis using Alb3 antiserum was performed to detect the presence of Alb3. Fig. 2B shows that H1Alb3 is produced in glucose-grown JS7131 when expression is controlled by both the yidC promoter (lane 1) and the tac promoter (lane 4). There is some degradation of H1Alb3 (* depicts the degradation products) that can be seen when this protein is overproduced (lane 4), possibly because it cannot insert and assemble efficiently under overproducing conditions. The wild-type Alb3 (mAlb3) and Alb3 containing the targeting signal (Alb3) are not expressed in detectable amounts in JS7131 either with the yidC or tac promoter (Fig. 2B).
Because the H1Alb3 mutant was functional, but not Alb3, we asked if there was a functional requirement for the N-terminal 57 amino acids of YidC. One possibility was that the YidC region was required because it contained a hydrophobic domain that helped in the membrane translocation of the Alb3 or YidC N-terminal lumenal domain. Therefore, we tested whether the first 57 amino acids of the precursor to the maltose-binding FIG. 1. Alb3 complements the growth defect of the YidC depletion strain JS7131. A, the membrane topology of YidC and mature Alb3 predicted by ToppredII. B, the three Alb3 constructs used in the complementation studies. The full-length Alb3 consists of 462 amino acids that include the stroma-targeting sequence (the first 55 amino acids). mAlb3 is the mature part of Alb3. The first 55 amino acids that correspond to the stroma-targeting sequence were replaced by a methionine. H1Alb3 is a fusion between the YidC H1 region and the mature part of Alb3 that includes the first 57 amino acids of YidC, a linker valine residue, and amino acids 59 -462 of Alb3. The three constructs were cloned into pACYC184 (from New England BioLabs) downstream from the yidC gene promoter, resulting in three plasmids: pACYC-Alb3, pACYC-mAlb3, and pACYC-H1Alb3. C, the YidC depletion strain JS7131 bearing the three different Alb3 plasmids was streaked on 0.2% arabinose or 0.2% glucose LB plates and incubated at 37°C overnight. Only JS7131 containing the pACYC-H1Alb3 grew on both arabinose and glucose plates.

FIG. 2. YidC and H1Alb3 expression in JS7131 detected by Western analysis.
A, JS7131 was grown in LB with 0.2% arabinose overnight. After washing with fresh LB medium twice, the overnight culture was 1:50 back-diluted in LB with either 0.2% glucose to deplete YidC or 0.2% arabinose to express YidC. For JS7131 bearing pACYC-H1Alb3, the overnight culture was cultivated in the absence of arabinose and directly back-diluted 1:50 in LB medium with 0.2% glucose added to suppress the expression of YidC. After about 3 h of growth, cells were collected, and the cell lysate was analyzed by SDS-PAGE and subjected to Western analysis using antiserum against YidC. The upper background bands precipitated with our YidC antiserum demonstrate that an equal amount of cell lysate was loaded in each lane. B, an overnight culture of JS7131 harboring pACYC-Alb3, -mAlb3, or -H1Alb3 was grown in LB with 0.2% arabinose. The overnight culture was then back-diluted into LB medium with 0.2% glucose and grown for an additional 4 h. JS7131 harboring pMS119-Alb3, -mAlb3, or H1Alb3 was grown for 2 h, induced with 1 mM IPTG, and grown for an additional 2 h. The cell lysate was analyzed by SDS-PAGE and immunoblotted with antiserum against Alb3. The * indicates possible proteolytic fragments of H1Alb3.
protein containing a cleavable signal sequence could replace the first 57 amino acids within YidC or H1Alb3. Fig. 3A shows that MBP-YidC is functional as this construct complements the growth defect of the YidC depletion strain. Immunoblot analysis shows that MBP-YidC is expressed and slightly smaller than the wild-type YidC, due most likely to cleavage of the preMBP presequence by signal peptidase (Fig. 3B). This shows conclusively that the first 57 amino acids of YidC do not have a functional requirement. Notably, the construct MBP-Alb3 does not restore growth to the YidC depletion strain (Fig. 3A), most likely because the protein is not expressed as it is not detected on an immunoblot (data not shown).
The Chloroplast Alb3 Promotes Membrane Insertion of Secindependent and Sec-dependent Membrane Proteins in E. coli-Because the E. coli YidC is essential for efficient membrane insertion of the Sec-independent M13 procoat protein and the Sec-dependent leader peptidase (Lep) (20), we first tested whether H1Alb3 is required for insertion of these two proteins when YidC is depleted. Cells expressing procoat were pulselabeled with trans-[ 35 S]methionine for 20 s and chased for 5 and 120 s. Signal peptide processing of procoat is completely normal when H1Alb3 is expressed (Fig. 4A). As a control, we confirmed that processing is completely blocked when JS7131 grown in glucose (YidC-depleted cells) in the absence of H1Alb3, but is efficient when grown in arabinose (ample YidC). In addition, H1Alb3 promotes membrane insertion of Lep (Fig.  4B). Cells expressing Lep were pulse-labeled with trans-[ 35 S]methionine for 20 s and analyzed for protease accessibility. Cells were converted to spheroplasts and then treated with or without protease. When JS7131 is expressing H1Alb3, Lep is completely accessible to protease, indicating that the large C-terminal domain of Lep is translocated across the membrane. In the same spheroplasts, there was no lysis as Band X (46), an unidentified cytoplasmic protein, was protected from protease digestion. The efficiency of Lep insertion was just as efficient as when YidC was present (arabinose-grown cells). Translocation of the C-terminal domain of Lep was impaired in YidC-deficient JS7131 cells (glucose-grown cells, Fig. 4B) as previously reported (20). In this study where Lep was examined, the overnight culture was grown in LB medium containing both arabinose and glucose. The glucose was added to the overnight culture as it suppresses the expression of YidC, allowing a more effective depletion of YidC in the subsequent growing of JS7131 cells in glucose medium.
Alb3 Promotes the Insertion of the Chloroplast cpSecE into the E. coli Inner Membrane-The complementation and insertion data above directly links the role of Oxa1p homologs in the translocation pathways in bacteria and in the plant chloroplast. To test this idea further, we tested whether H1Alb3 can function to insert a chloroplast thylakoid membrane protein into the E. coli inner membrane. Because the chloroplast SecE has been shown to functionally substitute for SecE in E. coli (36) we tested whether this protein would require Alb3 for insertion. In this study, we used cpSecE, which lacks the chloroplast stromatargeting signal. This protein spans the thylakoid membrane once with a short C-terminal tail in the lumenal space. H1Alb3 and cpSecE-containing cells were pulse-labeled for 20 s, converted to spheroplasts, and protease-mapped. As can be seen in Fig. 5A, cpSecE was efficiently degraded by protease, indicating it inserted across the membrane when H1Alb3 was present. Likewise, cpSecE was inserted, although to a lesser extent, when JS7131 was grown in the presence of YidC (arabinose). No detectable cpSecE was inserted into the membrane when YidC was absent (glucose).
Given that H1Alb3 promoted efficient insertion of the cpSecE, we tested whether the chloroplast LHCP, which requires Alb3 for membrane insertion (30), would insert into the E. coli inner membrane. Previously, it had been shown that overexpression of LHCP in E. coli lead to its accumulation in inclusion bodies (47). Therefore, we tested whether the accumulation may have arisen because insertion was blocked due to the protein not being able to utilize YidC. H1Alb3-containing cells may then allow efficient insertion. However, we observed FIG. 3. The YidC N-terminal domain (amino acids 1-57) does not play a functional role. A, JS7131 bearing the pACYC vector without insert or with pACYC vector encoding MBP-YidC or MBP-Alb3 was streaked out on a 0.2% arabinose or 0.2% glucose LB plate and incubated at 37°C overnight. Only JS7131 bearing the plasmid expressing MBP-YidC grew on both arabinose and glucose plates. B, JS7131 bearing the pACYC vector or the pACYC-MBP-YidC vector was grown, analyzed by SDS-PAGE, and subjected to Western analysis using YidC antiserum as described in Fig. 2B.   FIG. 4. H1Alb3 promotes the membrane insertion of procoat and leader peptidase. JS7131 or JS7131 bearing pACYC-H1Alb3 were grown as described in Fig 2A. These strains also contained the pMS119 vector with the procoat and leader peptidase introduced. These membrane proteins encoded by the pMS119 vector were induced by 1 mM IPTG and labeled with [ 35 S]Met for 20 s. A, signal peptide processing was used as a measure of membrane insertion of procoat. After labeling with trans-[ 35 S]Met for 20 s and chasing with non-radioactive methionine for the indicated times, the sample was directly added to ice-cold trichloroacetic acid. The trichloroacetic acid-precipitated proteins were subjected to urea-SDS-PAGE, and the radioactive protein bands were visualized by phosphorimaging. B, protease mapping assay was employed to monitor the translocation of the large C-terminal domain of Lep to the periplasmic space. After radiolabeling, the cells were converted to spheroplasts, and an aliquot was treated with proteinase K on ice for 1 h. The labeled proteins were trichloroacetic acid-precipitated, followed by immunoprecipitation with antiserum against Lep. The samples were then subjected to SDS-PAGE, and the radiolabeled proteins visualized by phosphorimaging. (See "Experimental Procedures.") no obvious membrane insertion either with H1Alb3 or YidC (Fig. 5B), showing that insertion requires other components that are lacking in E. coli. Likewise, no insertion was also found with PsbW (data not shown), a thylakoid protein shown to insert by a spontaneous pathway in chloroplast (31,48), suggesting that in E. coli spontaneous insertion is not possible for thylakoid membrane proteins.
Alb3 Is Not Required for Insertion of cpSecE into the Thylakoid-We examined also the cpSecE membrane biogenesis pathway in chloroplast to determine whether Alb3 is required. Radiolabeled cpSecE, without the targeting signal and containing an added Met residue at the C terminus, was synthesized using an in vitro system (translation products) (Fig. 6A) and added to a chloroplast lysate. cpSecE is recovered with the pelleted thylakoid fraction (ϪProt) and remains quantitatively bound to the thylakoid even after extraction with alkali (ϪProt/ NaOH). As a control, we showed that the thylakoid-processed iOE33, which is a lumenal protein, is extracted by alkali (ϪProt/NaOH, lower panel). These results with cpSecE indicate that cpSecE has integrated into the membrane. The addition of protease produces a shorter resistant fragment that is protected from the protease by the membrane (ϩProt). The protease-protected fragment disappears when the integrity of the membranes is disrupted by the addition of detergent (Prot/ T100). To test the requirement for Alb3, we used anti-Alb3 antibody prepared against the stroma-facing thylakoid domain of Alb3. Previously, this antibody (␣-ALB3) was shown by Moore et al. to inhibit the membrane integration of LHCP while having no effect on Sec and Tat-dependent substrates (30). Fig.  6B shows that the addition of the anti-Alb3 antibody (␣-Alb3) has no effect on the membrane insertion of cpSecE while inhibiting the integration of LHCP. The addition of anti-cpSecY antibody (␣-cpSecY) also has no affect while strongly inhibiting the Sec-dependent substrate iOE33. Finally, the addition of anti-Tat antibody prepared against Tha4 (␣-Tha4) had no effect on cpSecE insertion while inhibiting the translocation of t23- FIG. 5. The chloroplast H1Alb3 is essential for the insertion of cpSecE into E. coli inner membrane. JS7131 cells bearing H1Alb3 and a plasmid encoding cpSecE (A), and LHCP (B) were grown as described in Fig. 2A. The expression of cpSecE and LHCP was induced by the addition of 1 mM IPTG to JS7131 containing pEH1cpSecE or pQE80LHCP respectively, and their membrane insertion was monitored by protease mapping assay as described under "Experimental Procedures." FIG. 6. Alb3-independent insertion of SecE into thylakoid membranes. A, mature chloroplast cpSecE inserts into pea thylakoids in the correct orientation. Transport assays were conducted with chloroplast lysates and radiolabeled proteins (see "Experimental Procedures") corresponding to mature cpSecE and iOE33 (indicated to the left of the phosphorimage). Recovered thylakoid membranes were then divided into separate aliquots that were either washed with IB (ϪProt), washed with 0.1 M NaOH (ϪProt/NaOH), treated with thermolysin (ϩProt), or treated with thermolysin in the presence of 1% Triton X-100 (ϩProt/T100). Translation products (TP) are shown for comparison to mature (m) OE33 and to a characteristic cpSecE degradation product (DP), which is observed following protease treatment of thylakoids. B, cpSecE integration is unaffected by antibodies that inhibit translocation by the thylakoid Sec, TAT, and Alb3 localization pathways. Thylakoids were incubated with 10 mM Hepes-KOH, pH 8.0 (No antibody), pre-immune sera (PI), or immune serum directed against GST-ALB3 (␣-ALB3), cpSecY (␣-cpSecY), or Tha4 (␣-Tha4). Following removal of unbound antibody, thylakoids were examined for the ability to transport radiolabeled precursors indicated to the left of each image (see "Experimental Procedures"). Postassay treatment with thermolysin created the characteristic degradation products (DP) of cpSecE and pLHCP and eliminated all but the correctly transported mature (m) OE33 and plastocyanin (PC). Numbers below each lane represent the relative efficiency of transport for each precursor compared with buffertreated thylakoids. The transport pathway used by each precursor is indicated in parentheses.
PCm, a Tat-dependent substrate (35). The data taken together suggest that cpSecE does not require SecY, Tat, or Alb3 for its insertion into the thylakoid membrane.
The energetics of cpSecE insertion was investigated by monitoring insertion into isolated thylakoids under a variety of conditions (Fig. 7). cpSecE, which inserts correctly, gives rise to degradation products (DP; upper panel) after treating the thylakoid-inserted cpSecE with protease. Unlike the SRP/Alb3-dependent LHCP and the Sec-dependent iOE33, insertion of cpSecE did not require the addition of a stroma extract (compare ϪS.E., and ϩS.E. lanes) and was unaffected by addition of GMP-PNP or AMP-PNP. GMP-PNP is an inhibitor of the SRP, and AMP-PNP almost certainly inhibits the ATPase activity of cpSecA. Although assays did not contain additional ATP, the amount of ATP in S.E. and the translation products was enough to support iOE33 transport. A slight reduction in cpSecE membrane insertion was observed when nigericin was added to collapse the pH gradient, which completely blocks the insertion of a Tat-dependent substrate t23-PCm and inhibits insertion of LHCP as well. The energetic studies here show that cpSecE inserts by an SRP-independent route, in contrast to the Alb3-dependent LHCP protein. The data also shows that cpSecE insertion is stimulated by the pH gradient.
As a second test to show that Alb3 is not involved in the membrane insertion of cpSecE into the thylakoid membrane, we tested whether cpSecE could insert into protease-treated thylakoids. Fig. 8 shows studies where mature cpSecE is added to thylakoids that were previously treated with or without thermolysin and then, after transport, were treated again with thermolysin to assay the amount of cpSecE inserted. There was only a slight reduction of cpSecE that inserts into proteasetreated thylakoid (lane P) in comparison to mock proteasetreated thylakoids (MP). Similar results were found for pElip2, which was previously shown to insert by a spontaneous mechanism (bottom panel) (37). These results are different from those observed for the Tat (⌬pH)-, Sec-, or SRP/Alb3-dependent substrates where pretreatment of thylakoids with protease (compare P with MP lanes) abolishes translocation into or across the membrane (P). Taken together, these data suggest that cpSecE integration into thylakoid membranes takes place by a spontaneous mechanism. DISCUSSION The Oxa1p/YidC/Alb3 family is a newly discovered group of proteins that participates in a novel transport pathway for insertion into the mitochondrial and bacterial inner membrane and the chloroplast thylakoid membrane (49). Homologs in this evolutionarily conserved pathway are found in eubacteria, archaebacteria, and in mitochondria and chloroplasts of eukaryotes (26). Almost nothing is known about the structure and function of these Oxa1p homologs.
In this report, we show that the Arabidopsis Alb3 can complement the growth defect of a YidC depletion strain, when the first 57 amino acids of YidC containing the first transmembrane segment are linked to mature Alb3 (H1Alb3). The mature Alb3 or the precursor form of Alb3 containing the stromatargeting sequence cannot substitute for YidC. Possible reasons are that they could not be expressed in E. coli since they were not detected in the Western study or that the albino3 proteins are very unstable or they fail to assemble across the membrane. We found that a functional YidC could be made when the first 57 amino acids of YidC were replaced with the maltose-binding protein preprotein region that contains a cleavable leader sequence. This indicates that there is not a functional requirement for the N-terminal YidC portion of the molecule for Albino3 or YidC itself. Most likely the YidC Nterminal portion is necessary for YidC and H1Alb3 function because this region contains an uncleaved signal that promotes translocation of the N-terminal domain.
The fact that H1Alb3 can complement the YidC depletion strain demonstrates that the chloroplast and eubacterial , and 95% ethanol for a control (Ethanol). All assays were posttreated with thermolysin to reveal characteristic degradation products (DP) of cpSecE and pLHCP and to eliminate all but the correctly transported mature (m) OE33 and plastocyanin (PC). Numbers beneath the lanes represent the percentage of transport relative to assays conducted with S.E. (ϩSE). All other notations are as in Fig. 6.   FIG. 8. Chloroplast cpSecE integrates into protease treated thylakoids. Transport assays were conducted with isolated thylakoids that were either mock protease-treated (MP) or protease-treated with thermolysin (P) (see "Experimental Procedures"). Following thylakoid incubation with radiolabeled proteins, all assays were posttreated with thermolysin to reveal characteristic degradation products (DP) of cpSecE and pLHCP and to eliminate all but the correctly transported mature (m) OE33 and plastocyanin (PC). All other notations are as in Fig. 6.
Oxa1p homologs are truly functional homologs. Like YidC, the chloroplast H1Alb3 promotes the membrane insertion of the Sec-independent procoat protein and Sec-dependent Lep, suggesting that H1Alb3 may function in association with or independent of the Sec translocase as YidC does in E. coli. In addition, H1Alb3 can also promote the membrane insertion of the chloroplast thylakoid cpSecE into the E. coli inner membrane. Our studies demonstrate that Alb3 in chloroplast and YidC in bacteria function in similar and conserved ways in membrane assembly. These studies are in line with work over the last decade that has shown that, in general, transport across the bacterial inner membrane and thylakoid membrane is remarkably conserved (3,50). Like bacteria, the chloroplast has the SecA, Y and E homologs, and there is no reason to believe that Sec-dependent insertion is drastically different. In fact, the chloroplast thylakoid cpSecE can complement the E. coli SecE-depletion strain (36). Similarly, chloroplasts have a homologous Tat pathway found in bacteria that can export proteins with metal cofactors in a folded conformation (50).
We also examined the Alb3-dependent thylakoid protein LHCP for membrane insertion in E. coli. However, LHCP was not inserted even in cells expressing H1Alb3 (Fig. 5B). This defect in insertion may arise because E. coli lacks the novel SRP component, SRP 43, that is required for targeting in the chloroplast thylakoid system (51). Similarly, PsbW, which inserts into the thylakoid membrane by a spontaneous pathway in chloroplasts, does not insert into the inner membrane in E. coli even with H1Alb3 present (data not shown). Based on this result, it seems that E. coli cannot support spontaneous insertion of thylakoid membrane proteins.
Although YidC and H1Alb3 are essential for insertion of the cpSecE and bacterial SecE into the E. coli inner membrane, 2 surprisingly Alb3 is not required for insertion of the cpSecE into thylakoids. We found that antibody against Alb3 did not inhibit the integration of cpSecE into thylakoids, even though it blocked the insertion of the Alb3-dependent LHCP (Fig. 6B). In fact, cpSecE insertion may occur by a novel pathway since antibodies that inhibit transport by the thylakoid Sec or Tat pathways had no effect on cpSecE insertion. Neither stromal factors nor nucleotide triphosphates were needed for cpSecE membrane insertion, while only the depletion of the ⌬pH was found to reduce cpSecE insertion by ϳ45% (Fig. 7). Strikingly, cpSecE inserted with high efficiency into protease-treated thylakoids (Fig. 8), which blocked the insertion of the Alb3-dependent protein LHCP. These results bolster the observation of Robinson and co-workers where they find that to date most of the tested thylakoidal membrane proteins do not apparently require Alb3 and insert by a novel, possibly spontaneous mechanism (31,32). Of course, we cannot rule out that under normal in vivo conditions Alb3 is involved in cpSecE assembly. Moreover, there are other Oxa1p homologs in Arabidopsis (26), one of which is predicted with high probability to reside in the chloroplast.
Given these contrasting requirements for an Oxa1p homolog in the bacterial and chloroplast system for SecE, our hypothesis is that the requirement for Alb3 changed following the endosymbiotic event that gave rise to chloroplasts. Whereas cpSecE originally depended on Alb3 in the bacterial ancestor of chloroplasts, the dependence changed when much of the ancestral genome including the cpSecE gene moved to the nucleus of the plant eukaryotic cell. As a result, it became necessary for SecE to be imported into the chloroplast and then inserted posttranslationally rather than by a cotranslational mechanism. This change may also account for some differences in the targeting and translocation pathways: SRP43 is needed in the chloroplast thylakoidal targeting system, not in the bacterial system; chloroplasts lack the Sec translocase components SecD, SecF, and SecG, which are found in bacteria (52); and, as stated above, many chloroplast thylakoid membrane proteins utilize a novel "unassisted" membrane protein insertion pathway that requires no known translocation components (31). In this context, it is predicted that cpSecE insertion in cyanobacteria will be dependent on a homologue of Alb3/YidC. Another possible reason for the different requirement for Alb3 in the E. coli inner membrane and chloroplast thylakoids is the difference in lipid composition. The thylakoid membrane is unusual in that a high percentage of the lipids are glycoglycerolipids, whereas E. coli contains mainly phospholipids.
Taken together, the complementation studies reported here demonstrate that the chloroplast Alb3 functions in a similar way as the E. coli YidC. This is in line with other studies that have shown that the translocation machineries present in bacteria and chloroplasts such as the Sec translocase and the Tat translocase are remarkably similar (2,50,53). Oxa1p, Alb3, and YidC belong to the same protein family and were found to be involved in membrane protein assembly in mitochondria, chloroplast, and bacteria, respectively. The fact that Alb3 can functionally substitute for YidC further demonstrates that the Oxa1p/Alb3/YidC group represents a conserved translocation pathway in the three systems. Moreover, while YidC is required for membrane insertion of bacterial SecE in E. coli, 2 Alb3 is not required for the insertion of cpSecE into the thylakoid membranes. This may suggest that functional changes in the Oxa1p/Alb3/YidC pathway occurred during the evolutionary process or that the substrate itself has changed from a ribosome-bound nascent protein to a ribosome-released fulllength protein such that one system requires an Oxa1p homolog and the other does not.