The evolutionarily related beta-barrel polypeptide transporters from Pisum sativum and Nostoc PCC7120 contain two distinct functional domains.

Several beta-barrel-type channels are involved in the translocation or assembly of outer membrane proteins of bacteria or endosymbiotically derived organelles. Here we analyzed the functional units of the beta-barrel polypeptide transporter Toc75 (translocon in outer envelope of chloroplasts) of the outer envelope of chloroplasts and of a protein, alr2269, from Nostoc PCC7120 with homology to Toc75, both proteins having a similar domain organization. We demonstrated that the N-terminal region functions as a recognition and complex assembly unit, whereas the C terminus forms the beta-barrel-type pore. The pore region is, in turn, modulated by the N terminus of the proteins. The protein from Nostoc PCC7120, which shares a common ancestor with Toc75, is able to recognize precursor proteins destined for chloroplasts. In contrast, the recognition of peripheral translocon subunits by Toc75 is a novel feature acquired through evolution.

␤-barrel-type channels are involved in the translocation of polypeptides (1), the assembly of proteins in the outer membrane of endosymbiotic organelles (2)(3)(4), or in the assembly of proteins in the outer membrane of bacteria (5,6). These proteins belong to one class, which can be termed polypeptide-transporting ␤-barrel channels (2,4,7). Four proteins are in the focus of recent investigation, namely the bacterial outer membrane proteins Omp85 and ShlB, the mitochondrial outer membrane protein Tob55/Sam50, and the chloroplast outer envelope protein Toc75.
ShlB is an outer membrane protein involved in the secretion of hemolysins or adhesins in various Gram-negative pathogens (8,9). Omp85 is an essential component for outer membrane biogenesis in Neisseria meningitidis that might have two functions: the assembly of outer membrane proteins (5) and the translocation of lipids (10). Recently, it was discussed that the effect on lipid transfer by Omp85 depletion might be indirect and explained by an assembly defect of the required outer membrane protein, suggesting a function of Omp85 in outer membrane protein assembly only (11). As for ShlB, a ␤-barrel transmembrane structure was suggested for Omp85 (5). Recently, a new polypeptide-transporting protein was identified in the outer membrane of mitochondria and termed Sam50 (3), Tob55 (2), or mitochondrial Omp85 homologue (4). This protein facilitates the assembly of proteins into the outer membrane of mitochondria. Tob55/Sam50 is found in a larger complex with Mas37 (3,12) and Tob38/Sam35 (13)(14)(15).
The fourth investigated ␤-barrel-type polypeptide transporter is the 75-kDa subunit of the translocon of the outer envelope of chloroplasts, Toc75. Toc75 forms a complex with Toc34, Toc64, and Toc159 (16). In contrast to the other identified polypeptide transporters, such as Omp85, the translocation of proteins through Toc75 requires the action of assisting proteins, such as Toc159 (17), but still Toc75 seems to contain a preprotein-binding site as determined by electrophysiological measurements (1). Topological modeling of Toc75 from Pisum sativum (18) or Toc75-V from Arabidopsis thaliana (19) suggests a ␤-barrel-type structure. Previously, it was proposed that Toc75 might have evolved from the ShlB (20,21) or from the Omp85 class (5). This relationship to prokaryotic proteins is in line with the theory that chloroplasts have evolved from the ancestor of cyanobacteria (22,23).
Here we identified the Nostoc PCC7120 homologue of Toc75/ Omp85. To understand the evolutionary relationship between Toc75 and alr 2269, we investigated the specific properties of the two proteins. We here present experimental evidence that the N-terminal domain of ␤-barrel-type polypeptide transporters is involved in the recognition of substrates and complex assembly, whereas the C-terminal domain assembles the poreforming ␤-barrel.

MATERIALS AND METHODS
Construct Generation, Expression, and in Vitro Translation-Translation or expression of Toc34⌬TM, pSSU 1 , mSSU, pOE33, or pOE23 was described previously (24,25). psToc75 cDNA (1) was used as the template for the generation of psA, psB, psC, and psD ( Fig. 1) by PCR. The alr2269 cDNA was amplified from genomic Nostoc PCC7120 DNA. Constructs were generated by PCR, cloned into pTrcHis2 TOPO® TA (Invitrogen), and controlled by sequencing. BL21 cells (Novagen) transformed with plasmids were incubated in LB medium at 37°C, and expression was induced by the addition of 1 mM isopropyl 1-thio-␤-Dgalactopyranoside. Cells were harvested 2 h after induction and lysed using a French press (at 40 megapascals) in 50 mM NaP i , pH 8.0, 100 mM NaCl, and 2 mM ␤-mercaptoethanol (buffer A). Full-length proteins and C-terminal constructs (Fig. 1D) were accumulated in inclusion bodies. After centrifugation for 15 min at 25,000 ϫ g at 4°C, the pellet was resuspended in 50 mM NaP i , pH 8, 150 mM NaCl, 10 mM ␤-mer-captoethanol, 20 mM imidazole, and 5 M urea. The solution was centrifuged at 20,000 revolutions/min for 15 min at 4°C, the supernatant was subjected to nickel-agarose (Qiagen, Hilden, Germany), and the protein was purified according to the manual. Soluble fragments (Figs. 1D and 2A) were loaded onto nickel-agarose after centrifugation of lysed cells for 15 min at 25.000 ϫ g at 4°C. After purification, proteins were dialyzed against 50 mM NaP i , pH8, 100 mM NaCl. Their concentration was determined by Lowry analysis. The rapid translation system RTS 100 Escherichia coli HY Kit (Roche Applied Science) was used for in vitro translation. Antibodies against alr2269 were produced as previously described (1).
Cell Fractionation of Nostoc PCC7120 -Nostoc PCC7120 was grown photoautotrophically at 30°C in BG-11 medium (26) under constant illumination at 70 mol of photons m Ϫ2 s Ϫ1 with aeration by air containing 1% CO 2 . Cells were harvested at OD 750 ϭ 1.0 by centrifugation at 4,000 ϫ g for 10 min and then washed and lysed by French press at 40 megapascals. Broken cells were centrifuged at 48,000 ϫ g for 45 min at 4°C. The pellet was resuspended in 55% sucrose solution containing 20 mM HEPES, pH 8.0, and 0.2 mM phenylmethylsulfonyl fluoride. Floating sucrose density gradients were prepared by overlaying the cell/sucrose mixture with 40, 30, and 10% sucrose solutions and centrifuged at 130,000 ϫ g for 16 h at 4°C. Plasma membrane, thylakoid membrane, and outer membrane were collected, diluted in 20 mM HEPES, pH 8.0, and 1 mM phenylmethylsulfonyl fluoride and collected by centrifugation at 380,000 ϫ g for 1 h at 4°C and resuspended and stored at Ϫ20°C.
Binding Analysis-The peptides A1, B2, and E2 were synthesized at the Department of Peptide and Protein Chemistry, Charite (Berlin, Germany). Binding experiments using nickel-agarose or Toyopearl-AF-Tresyl 650 M (TosoHaas Corporation, Tokyo, Japan) were performed as described previously (24,25). Here, for one experiment similar molar amounts of preproteins (Fig. 2), synthetic peptides (Fig. 2), or constructs of alr2269 or Toc75 (Figs. 3 and 4) were coupled to the material. The coupled amount was controlled by adding defined amounts for coupling and analyzing the concentration of proteins or peptides in the flowthrough not associated with the matrix. For cross-linking analysis, wheat germ-translated 35 S-labeled pSSU was incubated with psToc75 or alr2269 constructs at 1 M in 50 mM NaP i , pH 8.0, and 100 mM NaCl in 50 l of bis(sulfosuccinimidyl)suberate (1 mM final, BS 3 ; Perbio, Bonn, Germany) was added and quenched after 45 min of incubation at 4°C by the addition of glycine (100 mM final). The binding efficiency was determined using AIDA software, expressed to the percent of loaded material and normalized to the indicated binding reaction.
Liposome Swelling Assay and Transport-specific Fractionation-Lipids for reconstitution were supplied by Nutfield Nurseries (Surrey, UK). Liposomes were prepared and proteins were reconstituted as described previously (17). The insertion was controlled by extraction. Swelling analysis was performed as previously established (27). The indicated amounts of NaCl or sucrose were added into liposome-containing solution, and the optical density at 500 nm (Fig. 5, B and C) or the integral of the optical density between 400 -700 nm (D and E) was determined. Transport-specific fractionation was previously described (28). For visualization, 0.1 mol % of rhodamine-phosphatidylethanolamine (Avanti Inc., Alabaster) was added to the lipid mixture.
Electrophysiological Measurements-Mega-9 (80 mM final) was added to purified psC, alr2269 or anaC. L-␣-phosphatidylcholine (type IV-S, Sigma) was dissolved in 80 mM Mega-9, 10 mM MOPS/Tris (pH 7.0). Both samples were mixed (1 mg of protein/20 mg of lipid) and dialyzed against 2 liters of 10 mM KCl, 10 mM MOPS/Tris, pH 7.0, for 2 h at 25°C and subsequently overnight at 4°C. Planar lipid bilayers were produced using the painting technique (29). A solution of 75 mg/ml L-␣-phosphatidylcholine in n-decan was applied to a hole in a Teflon septum, separating the two 3-ml chambers. To form a stable bilayer in 20 mM KCl, the 10 mM MOPS/Tris, pH 7.0, solution level was raised and lowered several times. The solution of the cis chamber was then exchanged to 250 mM KCl, 10 mM CaCl 2 , 10 mM MOPS/Tris, pH 7.0. Proteoliposomes were added to the cis chamber below the bilayer to allow the flow of the liposomes across the bilayer. The solution in the cis chamber was stirred to promote fusion. After fusion, the electrolytes in both compartments were changed to the final composition. Silver/AgCl electrodes were connected to the chambers through 2 M KCl-agar bridges. The electrode of the trans compartment was connected directly to the head stage of a current amplifier (GeneClamp500B, Axon Instru-ments, Union City, CA). The amplified currents were recorded using the pCLAMP9 software (Axon Instruments).
Circular Dichroism Measurements-Circular dichroism spectra were recorded in 1-nm steps for 2 s of integration time, and a slit width of 2 nm in 10 mM HEPES/KOH, pH 7.6, 100 mM NaCl on a Jobin Yvon CD6 spectrometer (Division d'Instruments, SA) at 22°C using a cuvette with a 1-mm path length. 15 spectra were recorded and averaged. The rough data were further manipulated as described previously (30).

RESULTS
Previously, the protein slr1227 from Synechocystis PCC6803 was described as a protein related to Toc75 (20,21). More recently, it was proposed that the Nostoc gene complement is most closely related to that of the ancestor of plastids based on the analysis of Nostoc punctiforme (22). To understand the evolutionary development of the polypeptide-transporting ␤-barrel protein Toc75, we wanted to compare the properties of psToc75 from P. sativum and a protein of the genus Nostoc. We therefore analyzed the genome of the two species N. punctiforme and Nostoc PCC7120 for the presence of a homologue of the polypeptide transporters Omp85 from N. meningitidis (5) and psToc75 (1) (supplemental figure, panel A). We identified two proteins related to psToc75/Omp85, namely Npun02006512 and alr2269. Sequence alignment of these two proteins revealed a sequence identity of 71% (supplemental figure). Because the two proteins identified from species N. punctiforme and Nostoc PCC7120 share such high identity, it is reasonable to use the protein encoded by the Nostoc PCC7120 gene alr2269 for further analysis (Fig. 1A). The protein encoded by alr2269 shares 19.4% identity and 29.3% similarity with psToc75 and 15.8% identity and 25.9% similarity with nmOmp85. The similarity in regard to the proposed topologies of the proteins is very high (Fig. 1D) (5,19). In addition to alr2269, two further homologues to psToc75 and Omp85 can be found in the Nostoc PCC7120 proteome, namely alr4893 and alr0075 (31,32). The two latter proteins are shorter than Omp85 with predicted molecular mass values of 72 and 54 kDa, respectively. Further, these two gene products are not as similar to psToc75 and Omp85 as alr2269 (not shown). Finally, in mass spectrometric analysis of outer membranes, only alr2269 could be detected (33). Hence, we have focused in this study on the analysis of alr2269.
The existence of the coding RNA was demonstrated by reverse transcription-mediated PCR (Fig. 1B). Immunodecoration of isolated fractions of Nostoc PCC7120 shows that the protein encoded by alr2269 is localized in the outer membrane (Fig. 1C, lane 3), but not in the thylakoid membrane (lane 1) or plasma membrane (lane 2). Topological modeling of alr2269 ( Fig. 1D) revealed two dominant regions, an N-terminal mostly soluble part and a C-terminal portion mostly involved in transmembrane ␤-strand formation. The localization of the domains in regard to the two sides of the membrane cannot be defined, because further experimental data have to be accumulated, as has been done for Toc75 (18,19). To test the predicted domain architecture, we constructed mutants of psToc75 and alr2269 comprising specific sections of these proteins. The first mutant (referred to as A in Fig. 1D, aa 161-470 of alr2269 and aa 149 -440 of psToc75, empty box and box with bricks) was chosen in regard to the proposed polypeptide transport-associated domain (POTRA) (34). The domain does not include the previously predicted transit peptide of psToc75 ending at amino acid 131 (35). AnaB (aa 388 -470) or psB (aa 365-440) (Fig. 1D, box with bricks) covers the third loop region in the topological model. The third construct (referred to as C in Fig. 1D, aa 469 -833 of alr2269 and aa 439 -809 of psToc75, box with crossed lines and box with transversal lines) represents the entire postulated pore-forming region (5,19). The last mutant (referred to as D in Fig. 1D, aa 702-833 of alr2269 and aa 678 -809 of psToc75, box with transversal lines) reflects the region with high similarity among the polypeptide transporters containing two structural motifs and eight proposed ␤-strands (31). The generated constructs of the N-terminal region (Fig.  1D, constructs A and B) were expressed in E. coli and purified as soluble proteins to high homogeneity ( Fig. 2A). The constructs C and D (in Fig. 5A) were expressed as insoluble proteins and purified under denaturing conditions. To confirm the structural content of the N-terminal constructs CD spectroscopy was performed. All N-terminal constructs showed a defined spectrum in the far UV region (Fig. 2B) (not shown for psA and psB), accounting for a secondary structure content (30). Using a simple approach for the estimation of the secondary structure content (30,36), we observed that the A constructs contained both helical (ϳ50%) and ␤-sheet (ϳ20%) content, whereas for the B constructs only helical content could be determined (ϳ65%). From the obtained result, we concluded Fractions were separated by SDS-PAGE, transferred to nitrocellulose, and incubated with antibodies against OxaI of Synechocystis PCC6803 or alr2269. D, a topological model of alr2269 was generated as described previously (17). Black boxes show strands covered by anaD and anaC, gray boxes show additional strands covered by anaC, and white boxes show putative strands present in the construct anaA. A schematic representation of alr2269 (top) or psToc75 (bottom) constructs is shown (for description, see "Materials and Methods"). The same boxes are shown in subsequent figures to indicate the used constructs. Numbers on top/bottom indicate first/last amino acids of the constructs. that the soluble expressed N-terminal constructs can be used for in vitro interaction analysis in solution.
psToc75 Contains an N-terminal Preprotein-binding Domain-Previously, it was suggested that psToc75 interacts with the transit peptide of chloroplast preproteins (1). We now investigated whether the N-terminal region of psToc75 could mediate this interaction. When a matrix charged with the precursor of the subunit of the oxygen-evolving complex of 23 kDa (pOE23) was incubated with radioactive labeled psA or psB (Fig. 2C, lanes 3 and 6) or 2-fold molar excess of psA or psB (Fig. 2D, lanes 1-9), both domains of psToc75 were associated with the matrix charged with the preprotein (Fig. 2C, lane 3; 2D, lanes 3 and 9) but not with the empty matrix or a matrix charged with the mature form of the small subunit of Rubisco (Fig. 2C, lane 6; 2D, lanes 6 and 12). Interestingly, the results for the qualitative (radioactive labeled psToc75 constructs; Fig.  2C) or quantitative approach (expressed psToc75 constructs; Fig. 2D) are comparable. In addition, the two psToc75 constructs associated in a similar quantitative manner with the precursor of the 33-or 23-kDa subunit of the oxygen-evolving complex (pOE33/23) (Fig. 2D, lane 9). These results suggest that psToc75 contains a transit peptide-binding site in its Nterminal region. To confirm this notion, synthetic peptides representing the transit sequence of the model preprotein pSSU were coupled to an affinity matrix. Here, the E2 peptide represents the N-terminal portion of the transit sequence, the A1 peptide represents the C-terminal portion without phosphorylated serine, and the B2 peptide represents the phosphoryl-ated C-terminal portion of the transit sequence (25, Fig. 2E). Incubation of these affinity matrices with psB revealed a physical interaction between this domain and the N-terminal portion of the transit sequence (Fig. 2F, lane 4). In contrast, psB did not bind to the C-terminal portion of the transit peptide independent of its phosphorylation state (Fig. 2F, lanes 2 and  3). Hence, we concluded that psToc75 recognizes the transit sequence of preproteins.
The Evolutionary Conservation of the N-terminal Preproteinbinding Domain-alr2269 of Nostoc PCC7120 encodes the protein with high similarity to psToc75 or Omp85 (Fig. 1A). We therefore compared the interaction of the N-terminal region of the psToc75 protein and alr2269 with preproteins destined for chloroplasts. For that, psA, psB, anaA, and anaB (Fig. 1D) were expressed, purified under non-denaturing conditions as soluble proteins ( Fig. 2A, lanes 1-8), and coupled to Ni-NTA to equal molar amounts of the indicated protein domains. Utilizing these psA, psB, anaA, or anaB affinity matrices for binding studies, an interaction with pSSU (Fig. 3B, lanes 3 and 5) but not with mSSU was observed (Fig. 3A, lanes 4 and 7), because mSSU was already eluted during the wash step (lanes 3 and 6).
In the case of pSSU, we could not detect any protein in the final wash fraction (not shown). To confirm that the observed interaction is truly mediated by the proteins investigated, pSSU or pOE33 and the N-terminal construct of either psToc75 (psA and psB) or alr2269 (anaA and anaB) were incubated in solution, followed by immunoprecipitation using antibodies specific against psToc75 (Fig. 3C, lanes 1 and 2) or alr2269 (lanes 3 and  4). In all cases, the complexes could be immunoprecipitated using either the psToc75 (Fig. 3D, upper panel, lanes 4 and 7) or alr2269 antibodies (lower panel, lanes 4 and 7). The same antibodies did not precipitate the preproteins without the addition of psToc75 or alr2269 constructs (not shown). In line with this, no precipitation of the preproteins by the corresponding preimmune serum could be obtained (not shown). The direct interaction was further confirmed by chemical cross-linking in solution. When pSSU was incubated with the N-terminal constructs, specific cross-links were observed (Fig. 3E, lanes  1-4, triangles). Cross-links to psA or anaA migrate at 60 -66 kDa, and cross-links to psB or anaB migrate at 40 -45 kDa. These cross-links are in the expected size range. The crosslinks are specific to the constructs added and to the presence of the transit peptide, because they are not observed while crosslinking the translation product itself (Fig. 3E, star) or using the mature form of SSU (Fig. 3E, lanes 6 -9). The pSSU-psA/psB cross-links are more pronounced than pSSU-anaA/anaB crosslinks (Fig. 3E, compare lanes 1 and 3; 2 and 4), suggesting that psToc75 reveals a higher affinity for the preproteins (Fig. 3, A  and B). To confirm that the interaction between alr2269 and the preprotein is indeed specific and targeted to the N-terminal region, the full-length protein or the construct representing the pore-forming domain (anaC) were reconstituted into liposomes. Subsequently, the liposomes not containing proteins (Fig. 3F,  lane 2), containing the full-length protein (Fig. 3F, lane 3), or the pore-forming region of alr2269 (Fig. 3F, lane 4) were incubated with 35 S-labeled pSSU. Only in the presence of alr2269 was an increased interaction of the proteoliposomes with the preprotein in comparison to the empty liposomes observed (Fig.  3F, compare lanes 2 and 3). In addition, the association of 35 S-labeled pSSU with alr2269 proteoliposomes (Fig. 3G, filled circles) but not with empty liposomes (Fig. 3G, triangles) or liposomes containing the pore-forming region (Fig. 3G, squares) was found to be time-dependent. This confirms the specificity of the interaction, because other outer membrane proteins not involved in the translocation do not stimulate the association of pSSU with lipid membranes (25). Hence, alr2269 contains an

N-terminal region specifically interacting with preproteins.
To compare the affinity of the two proteins, identical amounts of psB and anaB were coupled to an affinity matrix, as judged from the quantification of the protein not coupled to the matrix. Subsequently, the matrix was incubated with increasing amounts of pSSU, and the bound protein was quantified (Fig. 3H). Here the binding of pSSU was more efficient to psB than to anaB, suggesting a higher affinity of this construct for the preprotein. To confirm this observation, a matrix coated with pSSU was incubated at the same time with similar molar amounts of psA and anaB (not shown) or psB and anaA (Fig.  3I). This assay allowed a direct comparison of the binding affinity by subsequent immunodecoration. Here an increase of associated psToc75 constructs was observed by the addition of increasing amounts of the psA/anaB (not shown) or psB/anaA combinations (Fig. 3I, lower panel, lanes 2-4). In contrast, the association of the N-terminal constructs of alr2269 (Fig. 3I, lanes 2-4 for psB/anaA) (not shown for psA/anaB) did not increase in relation to the added concentration. Even more, at the highest concentration used (20 M), a lower amount of protein was associated than at 10 M (Fig. 3J, closed circles for anaA) (not shown for anaB). This can be explained by direct competition between the Toc75 and the alr2269 constructs for binding sites on the pSSU-coated matrix. This supports the notion that the specific interaction of the N-terminal domain of alr2269 to the preprotein is not as strong as the interaction of the homologous region in psToc75. In conclusion, the N-terminal domains of alr2269 and psToc75 are able to interact with preproteins but with different affinity.

The Function of the N-terminal Domain in Complex Assembly-The
Toc complex contains four subunits, namely Toc75, Toc34, Toc64, and Toc159 (16). To analyze the role of the N-terminal domain of psToc75 in complex assembly, an affinity matrix coated with psA or psB was incubated with radioactive labeled Toc159, Toc64 (not shown), or the cytosolic domain of Toc34 (Fig. 4A). No direct interaction of Toc159 or Toc64 with any of the constructs could be observed (not shown). In contrast to Toc159 and Toc64, Toc34⌬TM binds to psA (Fig. 4A, , lane 3). This result was confirmed by immunoprecipitation of the complex using psToc75 antibodies (not shown). Furthermore, the missing interaction of Toc34⌬TM to psB (Fig. 4B) points to a receptor-binding site distinct from the precursor-binding site. Supporting this notion, Toc34 was not able to compete for the interaction of radioactive labeled pSSU with psA or anaA (Fig. 4C, lanes 2 and 5), even though an association was observed (not shown). In contrast, the addition of peptides (E2 ϩ A1) (Fig. 2E) representing the transit peptide of pSSU (Fig. 4C, lanes 3 and 6) drastically reduced the recognition of the preprotein.
Besides hetero-oligomerization, ␤-barrel proteins of the Omp85 class were found to assemble homo-oligomeric complexes (e.g. (37)). Therefore, we investigated the ability of the N-terminal region of alr2269 to facilitate homo-oligomerization. We demonstrated that the N-terminal domain of alr2269 interacts with itself, because the radioactive labeled protein was efficiently co-eluted with the coupled protein (Fig. 4D, lane  3, anaA). However, no significant interaction of anaA with anaB, psA, or psB could be observed (Fig. 4D, lanes 3 and 5). In line with this, only a weak interaction of anaB with any of the constructs was observed (Fig. 4E, gray bars, anaA). Summarizing, both proteins (psToc75 and alr2269) are able to recognize preproteins with a domain localized in the second half of the N terminus (Fig. 4F, bar with bricks). In addition, the more extreme N-terminal domain (Fig. 4F, open box) is involved in complex formation.
Characteristics of the C-terminal Region of Toc75 and alr2269 -After establishing that the N-terminal domain of psToc75 is involved in preprotein recognition and complex formation, we wanted to confirm the C-terminal location of the pore-forming region as postulated by the topological model (Fig. 1D). Therefore, psC, psD, anaC, or anaD was expressed, purified, and reconstituted into liposomes of outer envelope lipid composition. The reconstitution was confirmed by silver staining of the protein content of the proteoliposomes (Fig. 5A). Only in the case of psC was a second minor proteinaceous band observed. Carbonate extraction confirmed that this protein is not reconstituted (not shown). The insertion of the constructs into liposomes was taken as a first hint that the C-terminal region might indeed be involved in pore formation. To further analyze the pore formation, we went on to establish a liposome swelling assay using proteoliposomes containing psC or psD. First, the proteoliposomes were subjected to a hypertonic NaCl medium, and the time-dependent change of the optical density as a measure of the change of the ultrastructure of the lipo-somes was determined (Fig. 5B). The optical density of the empty liposomes increased in a time-dependent manner (Fig.  5B, solid line) due to liposome deformation upon osmotic pressure resulting from the addition of salt (38 -40). This shift could not be observed using liposomes containing psC (Fig. 5B, dashed line) or psD (solid gray line). We concluded that the reconstituted constructs were able to form pore-like structures and to exchange solutes. This behavior was not dependent on the salt concentration in the range up to 100 mM NaCl (Fig.  5C). Furthermore, reconstituting Toc34 (not shown) or the Nterminal region of psToc75, psA (Fig. 5C, white squares) did not alter the behavior of the empty liposomes, supporting the notion that the C-terminal constructs facilitate solute exchange and that the N-terminal region cannot form a pore by itself. Next, we analyzed whether the same behavior can be obtained  3 and 6). Bound protein was collected and visualized by phosphorimaging. 100% of the translation product is shown (TP). D, 35 S-labeled anaA was incubated with psA, psB (upper panel), anaA, or anaB (lower panel) coupled to Ni-NTA (empty). Flowthrough (FT, lanes 2, 4, and 6) and elution (E, lanes 3, 5, and 7) were collected and radioactivity-visualized. 100% of the translation product is shown (TP, lane 1). E, the interaction between the fragments of psToc75 or alr2269 and anaA (black bars) or anaB (gray bars) was quantified and normalized to the interaction efficiency of anaA dimerization. The average of at least three experiments is shown. F, a model of the interacting regions and the recognized targets for alr2269 and psToc75 is shown. for alr2269. Again, liposomes containing psC, psD, anaC, or anaD adapted to an osmotic pressure by the addition of 100 mM NaCl (Fig. 5D), as seen from the comparison to empty liposomes. Liposomes containing full-length alr2269 (Fig. 5D, circles) responded to the osmotic pressure in the same manner as liposomes containing the C-terminal constructs. To confirm our observation, sucrose was added to the liposomes (Fig. 5E). Empty liposomes were shrinking, which resulted in reduction of the optical density (not shown), as no deformation of the liposomes occurred (as had previously occurred by the addition of salt). When the C-terminal constructs of psToc75 and alr2269 were reconstituted into the liposomes, they were able to facilitate the compensation of the osmotic pressure (Fig. 5E), as determined by an increase of the ratio between the optical density of protein-loaded liposomes and empty liposomes. To confirm this result, empty liposomes and proteoliposomes were subjected to a transport-specific fractionation using an isoosmotic density gradient. Here, liposomes migrate into the gradient only when osmolyte exchange-facilitating channels are present (28). All liposomes containing the C-terminal constructs migrated into the gradient (Fig. 5F, fraction 2). Hence, the proteins were able to support the exchange of the osmolytes, suggesting that the C-terminal portion is involved in pore formation.
Some of the liposomes containing the longer constructs (psC and anaC) behaved like empty liposomes (Fig. 5F, fraction 1). Interestingly, when liposomes containing full-length alr2269 were subjected to osmotic pressure changes by sucrose, no significant difference from empty liposomes could be observed in the swelling assay (Fig. 5E, circles). Subjecting these proteoliposomes to transport-specific fractionation revealed a similar behavior as empty liposomes (Fig. 5G, lanes 1 and 2). This suggests that a region not present in the C-terminal constructs has to be involved in the gating or sizing of the translocation. The different response of the full-length protein containing proteoliposomes obtained for NaCl and sucrose might be explained as follows. In contrast to the truncated protein, fulllength alr2269 might be tightly gated and does not facilitate the exchange of molecules as large as sucrose. To confirm this notion, electrophysiological experiments were performed. Using single channel analysis experiments, we could demonstrate that the channels alr2269 and anaC have the same reverse potential (Fig. 6A) of U rev ϳ 17 mV and the same main conductance of ϳ500 nS (Fig. 6B). This observation suggests a similar structure of the interior of the channel, supporting the notion that the C-terminal portion forms the pore. Interestingly, analyzing the open and closed traces of both molecules (Fig. 6C) revealed a fundamental difference of both channels. AnaC shows either a stable open or closed state (Fig. 6C, anaC). In contrast, the full-length protein shows several minor conductance levels and changes between the conductance states with high frequency. This indicates that the region not present in anaC is involved in the gating of the ␤-barrel polypeptide channel. This notion could be confirmed by analysis of the C-terminal portion of Toc75. Here we obtained the same stability of the opening and closing state for both psC and anaC (Fig. 6C), which is significantly different from that of fulllength psToc75 (1,41). For psC, a reverse potential of U rev ϳ 40 mV was observed (not shown), which is close to the previously reported 48 mV for the full-length Toc75 (41). Hence, the gating of the C-terminal localized channel by the N-terminal domain of the protein might be a general characteristic of the polypeptide-transporting proteins. DISCUSSION Recently, a relationship between ␤-barrel-type polypeptide transporters of the outer membrane of endosymbiotic-derived organelles and the bacterial outer membrane has been reported (2,4,31). In addition, evidence for the evolutionary conservation of some features of the proteins has been presented (37). To investigate the relationship between the proteins from prokaryotes and eukaryotes with a common ancestor, we have studied psToc75 (16) and its related protein from Nostoc PCC7120, alr2269. Previously, it was suggested that proteins of the ␤-barrel polypeptide-transporting channel family have two distinct domains (34,42), which is also reflected by the topological model of alr2269 (Fig. 1D). In line with this, we demonstrated that the N-terminal region of psToc75 acts as a specific receptor for proteins containing an N-terminal transit sequence (Fig. 6D). This observation supports the previously postulated binding site within psToc75 (1). Furthermore, the protein not involved in the translocation of preproteins across the outer envelope was able to specifically recognize the transit sequence containing proteins (Fig. 3). Hence, the interaction between the preproteins and alr2269 revealed a prerequisite of this feature in the common ancestor of both proteins. The interaction between the ␤-barrel polypeptide transporter and preprotein is directed toward the N-terminal region (Fig. 3F). This leads to the question of whether the pore-forming protein itself offered the first binding site for preproteins after the cyanobacterium was incorporated into the host cell. The more extreme N-terminal region of the two proteins was found to be involved in complex formation of the translocon (Fig. 4). Here, the mode of interaction differs between both proteins. The region of psToc75 containing a loop facing the cytosol is involved in hetero-oligomerization by recognizing the receptor Toc34 (Figs. 4, A and B, and 6D), whereas the N terminus of alr2269 is involved in homo-oligomerization (Fig. 4). Such homo-oligomerization was also obtained for other prokaryotic proteins (37). The divergence of the amino acid sequence of the extreme N terminus (Fig. 1A) supports a different mode of complex formation. In line with this, no clear homologues of receptors of the Toc machinery were identified in cyanobacteria (16).
The proposed topological dissection of the proteins (Fig. 1D) in two functional regions suggests a pore function of the C terminus. Investigations of ShlB (42) and topological modeling of psToc75 (19) or Omp85 (5) support such speculation. As presented herein, the C-terminal portion of psToc75 or alr2269 allowed the passage of salt or sucrose (Fig. 5). The electrophysiological properties, such as the reverse potential and the main conductance of psC (not shown) and Toc75 (41) or alr2269 and the C-terminal portion of the protein, were comparable (Fig. 6). This result underlined the conclusion that the C-terminal domain builds the pore and questions the two proposed membrane-inserted strands in the N-terminal domain (Fig. 1D). In contrast to the C-terminal portion, the full-length alr2269 was not able to transport sucrose (Fig. 5, E and G). Guided by this observation, we speculate that the N-terminal portion has a function in gating of the channel (Fig. 6D). This notion is supported by the obtained difference of the gating behavior of the full-length protein and the C-terminal domain. The Cterminal domain itself shows a low gating profile, whereas the full-length protein rapidly changes its conductance (Fig. 6C). The obtained channel characteristic of alr2269 confirms its relationship to the ␤-barrel-type polypeptide transporters. The permeability ratio (P K ϩ /P Cl Ϫ ) calculated by the Goldman-Hodgkin-Katz voltage equation is 2.2:1, revealing a cation selectivity (Fig. 6A), as previously established for psToc75 (1) or synToc75 (20). Further, a similar main conductance as found for other ␤-barrel-type transporters was obtained. psToc75 and synToc75 have a main conductance of 400 pS at similar experimental conditions (18), which is close to the 500 pS measured for alr2269. For filamentous hemagglutinin C, a main conductance of 1200 pS (at 1 M KCl) or 300 pS (at 200 mM KCl) was reported (43), which again reveals a similar main conductance as found for alr2269 when extrapolated to similar experimental conditions. A pore diameter of 1.7 nm can be estimated for alr2269 based on the main conductance. This conclusion is based on the assumption of a cylindrical conformation, considering that the conductivity within the pore is ϳ5-fold lower than in the bulk medium (43). The obtained pore dimension is in line with the proposed function as polypeptide transporter as the translocation of unfolded polypeptide chains (1) is possible. The diameter of alr2269 is smaller than the pore diameter of HMW1B (2.7 nm) obtained by swelling analysis (37). This difference in the pore diameter of the proteins might also reflect the differences in the techniques used. However, both channels have an interior dimension that would allow the translocation of polypeptides, and the findings for the alr2269 and psToc75 relationship might be extrapolated to the other proteins of this class.
Interestingly, the two constructs representing the extreme C-terminal region (psD and anaD) show channel activity as well (Fig. 5). Using the swelling assay in the presence of 300 mM NaCl, we obtained an adaptation for proteoliposomes containing psC, anaC, or alr2269 but not of the liposomes containing psD and anaD (not shown). This leads to the speculation that the two constructs form multimeric ensembles within the membrane, which are disrupted in the presence of high concentration of salt (Fig. 6D). Therefore, the transport activity of psD and anaD reflecting eight conserved ␤-strands (31) leads to the hypothesis that ␤-barrel proteins, similar to helical proteins (44), have evolved from a minimal structural unit. Such minimal structure might be represented by toxins containing two or four segments and forming a multimeric ␤-barrel channel (45) or by the smallest monomeric ␤-barrel membrane proteins containing eight ␤-strands (45). Therefore, psD and anaD might reflect the evolutionary base unit for the development of the ␤-barrel pore.