Direct interaction of YidC with the Sec-independent Pf3 coat protein during its membrane protein insertion.

YidC is a newly defined translocase component that mediates the insertion of proteins into the membrane bilayer. How YidC functions in the insertion process is not known. In this study, we report that the Sec-independent Pf3 coat protein requires the YidC protein specifically for the membrane translocation step. Using photocrosslinking techniques and ribosome-bound Pf3 coat derivatives with an extended carboxyl-terminal region, we found that the transmembrane region of the Pf3 coat protein physically interacts with YidC and the bacterial signal recognition particle Ffh component. We also find that in the insertion pathway, Pf3 coat interacts strongly with YidC only after its transmembrane segment is fully exposed outside the ribosome tunnel. Interaction between Pf3 coat and YidC occurs even in the absence of the proton motive force and with a Pf3 coat mutant that is defective in transmembrane insertion. Our study demonstrates that YidC can directly interact with a Sec-independent membrane protein, and the role of YidC is at the stage of folding the Pf3 protein into a transmembrane configuration.

Most bacterial membrane proteins insert into the membrane utilizing the Sec translocase. The Sec translocase is composed of the integral membrane core subunits SecY and SecE, which function as protein transporters, and the peripheral subunit SecA, an ATPase that functions as a molecular motor to push the protein chain through the SecYE translocase by hydrolyzing ATP (1)(2)(3). The other components of the Sec translocase, such as SecG, SecD, SecF, and YajC, are not absolutely required for protein translocation. Prior to the membrane translocation step, the proteins need to be targeted to the membrane. In bacteria, this is achieved by two major pathways, one involving SecB (reviewed in Ref. 4) and the other involving the signal recognition particle (SRP) 1 (reviewed in Ref. 5). In the SecB pathway, SecB, a small molecular chaperone, targets exported proteins post-translationally to the membrane-bound SecA subunit of the Sec translocase (6). In the SRP pathway, SRP targets membrane proteins co-translationally to the membrane (7). In bacteria, SRP consists of a 48-KDa protein component, designated Ffh, and a 4.5 S RNA component. SRP is thought to bind to FtsY in its targeting cycle. After targeting by SRP, SecA can also be required for translocation of the periplasmic domains of some membrane proteins (8).
YidC, a homologue of the mitochondrial Oxa1 (9) and chloroplast Alb3 (10), is a newly identified translocase component in bacteria (11,12). YidC is absolutely essential for the membrane insertion of the Sec-independent M13 coat protein (13,14). It also stimulates the membrane translocation of the carboxyl-terminal domain of the Sec-dependent proteins, leader peptidase (Lep) and FtsQ (13,15,16), as well as promoting the insertion of the amino-terminal domain of Pf3-tagged leader peptidase (Pf3-Lep). For Sec-dependent membrane proteins, YidC has been shown to cooperate with Sec translocase and to play a possible role in the lateral movement of transmembrane segments out of the Sec translocase complex (15)(16)(17)(18)(19). Therefore, a current model is that YidC functions in association with the Sec translocase or works independently to insert proteins into membranes (18,20). Whether YidC works with or without the Sec translocase depends on the membrane protein being inserted.
Pf3 coat, a 44-amino acid membrane protein, is the major coat protein from Pseudomonas aeruginosa phage Pf3. When Pf3 coat is expressed in Escherichia coli, it inserts into the inner membrane with a single transmembrane segment adopting an N out /C in topology. Pf3 coat has been used as a model protein to study Sec-independent membrane insertion (21). It was widely believed that Pf3 coat inserts directly into the bilayer of the bacterial inner membrane without the assistance of proteinaceous factors (22,23). However, because recent studies show that the Sec-independent M13 procoat protein requires YidC for membrane insertion (13,14), Pf3 coat protein might first contact YidC and require its function for translocation across the membrane.
In this study, we show that in a strain in which YidC is depleted, the membrane insertion of wild-type Pf3 coat is severely affected. Using a photocrosslinking method and Pf3 coat mutants in which the carboxyl-terminal regions are extended, we found that YidC, but not SecY, interacts with Pf3 coat protein after its transmembrane segment emerges from the ribosome. We have also shown that YidC can interact with Pf3 coat when it is partitioned into but not translocated across the membrane, suggesting that YidC plays a specific role for membrane translocation.

MATERIALS AND METHODS
Strains and Plasmids-The E. coli YidC depletion strain, JS7131, was from our laboratory collection (13). For in vivo studies, the genes of Pf3 coat, its mutants, and Pf3-P2 (Pf3 coat extended with Lep P2 soluble domain at its carboxyl terminus) were under the control of the tac promoter in vector pMS119. pMS119 also carries the lacIq gene necessary for studies in JS7131. For in vitro studies, the truncated Pf3-P2 mRNA was transcribed from the T7 promoter in the pT7-7. pET610, harboring secYEG genes under the control of the Trc promoter, was a generous gift from Arnold Driessen's laboratory.
In Vivo Protease Mapping-JS7131 strain bearing wild-type Pf3 coat or its mutants, or Pf3-P2 in pMS119, was grown in LB medium with 0.2% arabinose overnight. The overnight culture was washed with LB medium twice and back-diluted 1 to 50 into fresh LB containing either 0.2% glucose or 0.2% arabinose. The cultures were grown for 2.5 h and shifted to M9 minimal medium supplemented with arabinose (to express YidC) or glucose (to deplete YidC) and grown for an additional 30 min. Ten min prior to labeling, 1 mM isopropyl-1-thio-␤-D-galactopyranoside was added. The cells were labeled with trans-[ 35 S]methionine for 20 s and chased for various times. In vivo protease mapping was performed to determine whether the amino-terminal domain of Pf3 coat protein translocates across the membrane as described (24). Briefly, radiolabeled cells were resuspended in 33 mM Tris HCl, pH 8.0, and a 40% sucrose solution, treated with 5 g/ml lysozyme, 1 mM EDTA (final concentrations), and incubated on ice for 15 min to convert the cells into spheroplasts. The spheroplasts were then treated with 0.5 mg/ml proteinase K for 1 h on ice. The reaction was quenched with phenylmethylsulfonyl fluoride (0.33 mg/ml, final concentration) and then precipitated with 10% trichloroacetic acid and subjected to immunoprecipitation or directly analyzed by SDS-PAGE/autoradiography. Pf3 coat can be analyzed directly without immunoprecipitation using a 22% SDSpolyacrylamide gel containing urea (25), because there is no background of radiolabeled proteins in the region of the gel of less than 6 KDa. Pf3-P2 can be analyzed by immunoprecipitation with anti-Lep serum.
Sodium Carbonate Extraction-The cells expressing Pf3 coat were labeled with trans-[ 35 S]methionine and converted into spheroplasts as described under "In Vivo Protease Mapping." 0.2 M sodium carbonate (pH 11.5) was added to the spheroplast suspensions. The suspensions were then vigorously vortexed and incubated on ice for 30 min. After centrifugation for 1 min in an Eppendorf centrifuge (14,000 rpm) to remove non-lysed cells and cell debris, the sample was subjected to ultracentrifugation (130,000 ϫ g) for 1 h at 4°C. The supernatant and membrane pellets were carefully separated, subjected to 10% trichloroacetic acid precipitation, and analyzed by SDS-PAGE and phosphorimaging.
In Vitro Translation and Photocrosslinking-The E. coli S100 in vitro translation system was made according to Gold and Schweiger (26). Site-specific photocrosslinking was performed based on the procedure as described (27). Amber codons were introduced into the Pf3 coat gene at various positions to allow the incorporation of the photoactivatable amino acid analogue (Tmd)Phe (L-4Ј-(3-(trifluoromethyl)-3H-diazirin-3yl)phenylalanine) (28). Using T4 RNA ligase (Roche Molecular Biochemicals), (Tmd)Phe-pdCpA was covalently joined to the E. coli suppressor tRNA Asn (29) made using T7 Megashortscript in vitro transcription kit from Ambion. PCR was used to generate the truncated gene fragments from which the truncated mRNA transcripts were made using the T7 Megashortscript in vitro transcription kit. The inverted membrane vesicles were prepared according to the procedure (30). For photocrosslinking studies, the truncated mRNA and (Tmd)Phe-charged tRNA Asn were added to the E. coli S100 in vitro translation system to produce the [ 35 S]methionine-labeled Pf3 coat nascent protein. Inner membrane vesicles were added to the translation reactions 4 min after the reactions were initiated at 37°C, and the reaction was continued for 40 min at 37°C. The reactions were then irradiated with 360 nm of ultraviolet light for 30 min at 4°C to perform photocrosslinking. The reactions were then subjected to trichloroacetic acid precipitation or immunoprecipitation. The SecY positive control was synthesized using pET610 (containing the secY gene) with the T7 S30 circular DNA in vitro translation system (Promega). In this system, we took advantage of endogenous E. coli RNA polymerase instead of T7 RNA polymerase. For SecY immunoprecipitation, the samples were not heated, to avoid aggregation.

RESULTS
YidC Is Required for Wild-type Pf3 Coat Membrane Insertion-We have demonstrated that M13 Procoat, a Sec-inde-pendent protein, requires YidC for membrane insertion (13). Therefore, we were interested in whether the single membrane-spanning Pf3 coat may also require YidC for membrane insertion. Previously, Pf3 coat was shown to insert into the E. coli inner membrane without the aid of the Sec translocase (31). E. coli JS7131 cells (13) were grown in the presence of arabinose to express YidC or glucose to deplete YidC. Cells expressing Pf3 coat protein were labeled with trans-[ 35 S]methionine for 20 s and chased for various times. Protease mapping was performed to monitor the membrane insertion of Pf3 coat. Because the methionine at position 1 is the only methionine in the Pf3 coat protein, only Pf3 coat with an intact amino terminus can be detected by autoradiography (see Fig. 1a). When YidC was present (Fig. 1b, YidC ϩ ), Pf3 coat was inserted normally across the membrane with an N out /C in topology and was digested to a nonradiolabeled fragment by proteinase K added to the periplasmic side. In YidC-deficient cells (Fig. 1b, YidC Ϫ ), Pf3 coat was protected by the membrane from proteinase K digestion, indicating that it does not insert across the membrane even after a 2-min chase period. This demonstrates that Pf3 coat protein requires YidC for transmembrane insertion.
YidC Physically Interacts with the Transmembrane Region of Pf3 Coat-Because Pf3 coat requires YidC for membrane insertion in vivo, we analyzed whether YidC physically interacts with Pf3 coat during insertion and whether other proteins interact with Pf3 coat. Site-specific photocrosslinking (28) was applied to study Pf3 coat membrane biogenesis in vitro. With this technique, a truncated mRNA is generated that lacks the FIG. 1. YidC is required for membrane insertion of Pf3 coat and Pf3-P2 proteins. a, topologies of Pf3 coat and Pf3-P2 in the plasma membrane with the initiation Met (*) in the periplasm. b, proteinase K mapping of the Pf3 coat protein in the YidC depletion strain, JS7131. Cells were grown in the presence of arabinose (YidC ϩ ) or glucose (YidC Ϫ ), pulse-labeled for 20 s, and chased for the indicated times. The cells were then converted to spheroplasts and treated with or without proteinase K on ice for 1 h. After the reaction was quenched with phenylmethylsulfonyl fluoride, the samples were trichloroacetic acid-precipitated, acetone-washed, and analyzed by SDS-PAGE and phosphorimaging. c, proteinase K mapping of Pf3-P2 in the JS7131 strain. JS7131 expressing Pf3-P2 were grown in the presence of YidC (YidC ϩ ) or under YidC depletion conditions (YidC Ϫ ) and analyzed as described for panel b, except there was no chase. Pf3-P2, which is translocated across the membrane, is cleaved by proteinase K to generate a resistant fragment (Digested Pf3 P2).
termination codon and therefore the nascent protein chains are not released from the ribosome. Ribosome-nascent chains trapped in the translocation process can then be subjected to cross-linking to identify interacting proteins. The length of wild-type Pf3 coat (44 amino acids) is too short for the photocross-linking technique, because 35-40 amino acids are estimated to reside within the ribosomal tunnel upon arrest of translation. To make Pf3 coat suitable for photocrosslinking, we have lengthened the protein with a sequence of the leader peptidase (Lep) soluble P2 domain (the hybrid is called Pf3 P2, Fig. 1a). The hybrid protein shows the same YidC dependence as Pf3 coat (Fig. 1c). Pf3-P2 was expressed in the JS7131 strain and subjected to proteinase K mapping. In the cells grown with arabinose to express YidC (Fig. 1c, YidC ϩ ), Pf3-P2 was digested to a smaller fragment, whereas in the cells grown with glucose to deplete YidC (Fig. 1c, YidC Ϫ ), Pf3-P2 was fully protected, indicating that Pf3-P2 is also YidC-dependent.
Ribosome-attached Pf3-P2 nascent proteins were site-specifically modified with the photoactivatable cross-linking group (Tmd)Phe (L-4Ј-(3-(trifluoromethyl)-3H-diazirin-3-yl)phenylalanine). Amber suppressor tRNA charged with (Tmd)Phe enabled incorporation of the photoprobe into the Pf3 coat transmembrane segment at positions Thr-20, Ile-27, and Leu-34 (Fig. 2a). A PCR method was employed to generate the truncated Pf3-P2 gene fragment for each amber mutant coding for a 91-amino acid protein. After synthesis, the Pf3 coat portion of 44 amino acids should be fully exposed from the ribosome, whereas the Lep P2 domain remains in ribosomal tunnel.
Each truncated mutant protein was synthesized in vitro in the presence of inverted membrane vesicles. During the synthesis, each reaction was exposed to UV radiation. Pf3 coat was cross-linked to YidC and also to Ffh when the (Tmd)Phe photoprobe was located within the transmembrane segment ( Fig.  2b) at positions Thr-20, Ile-27, and Leu-34. UV irradiation resulted in two major cross-linking products with apparent molecular weights of 60 and 70 KDa. Immunoprecipitation with antiserum to YidC or with antiserum to Ffh (Fig. 2b) confirmed that the 70-KDa cross-linking product corresponds to the YidC-Pf3 coat adduct and the 60-KDa product is the Ffh-Pf3 coat adduct. The molecular masses of the cross-linked products (Pf3-YidC and Pf3-Ffh adducts) are consistent with the combined sizes of the individual components (the 91-amino acid Pf3 coat protein is ϳ10 KDa; YidC, 60 KDa; and Ffh, 48 KDa). We also introduced the photoprobe into the Pf3 coat amino tail (at position Leu-12) and observed almost no crosslinking between Pf3 coat and YidC (data not shown). These studies are consistent with YidC promoting membrane insertion by physically interacting with the Pf3 coat hydrophobic segment and not with the hydrophilic region.
To test whether the interaction of the Pf3 coat to YidC occurs only during membrane translocation, photocrosslinking using the Pf3 coat L34 amber mutant (the total nascent chain length is 91 amino acids) was carried out after treating the translocation intermediates with puromycin/high salt (potassium acetate) (Fig. 2c). Puromycin/high salt treatment releases the truncated nascent protein from the ribosome so that no translocation intermediates will be formed (15). When the puromycin/high salt treatment was carried out for 5 min prior to UV irradiation (Fig. 2c, Puro ϩ), the photocrosslinking between YidC and Pf3 coat was substantially reduced compared with the nontreated sample (Fig. 2c, Puro Ϫ). This suggests that when Pf3-P2 is released from the ribosome, it cannot form translocation intermediates with YidC, which therefore indicates that the Pf3 coat-YidC interaction occurs only during Pf3 coat membrane insertion. Taken together, the photocrosslinking data clearly demonstrate that YidC as well as Ffh can interact physically with Pf3 coat during its membrane insertion.
The YidC-Pf3 Coat Interaction Is More Efficient As the Transmembrane Segment of Pf3 Coat Is Extended Further from the Ribosomal Tunnel-To study at which stage Pf3 coat interacts with YidC during membrane insertion, we used truncated mRNA to synthesize nascent Pf3 coat proteins of different lengths, i.e. 56-Pf3, 64-Pf3, 72-Pf3, 80-Pf3, and 91-Pf3, which have 56, 64, 72, 80 and 91 amino acids, respectively. In addition, the photoprobe (Tmd)Phe was incorporated by introducing an amber codon at position Thr-20 in 56-Pf3 and 64-Pf3 or at position Leu-34 in 72-Pf3, 80-Pf3, and 91-Pf3 (Fig. 3a). For nascent chains that remain bound to the ribosomes, about 35 amino acids will be trapped within the ribosomal tunnel. Therefore, as the nascent chains increase in length from 56 to 64 to 72 residues, the Pf3 coat proteins gradually expose their transmembrane regions (residues 19 -36) from the ribosomal tunnel, whereas in 80-Pf3 or 91-Pf3 the transmembrane segment is fully exposed. The series of truncated nascent proteins should therefore represent insertion intermediates (32) by which the Pf3 coat protein leaves the ribosome to contact and insert into the membrane. Photo-cross-linking was performed to investigate at which stage of the membrane biogenesis pathway YidC interacts with Pf3 coat. To do this, the photocrosslinking products of the nascent chains were analyzed either directly (Fig. 3b) or after immunoprecipitation with antiserum to YidC (Fig. 3c, left panel, IP, YC). The data show that the interaction between YidC and Pf3 coat is most intense when the transmembrane segment of Pf3 coat is fully exposed from the ribosome (80-Pf3 and 91-Pf3). This finding indicates that YidC interaction is more efficient when the carboxyl-terminal region of the Pf3 coat protein is exposed from ribosomal tunnel.
We also checked whether YidC in the photocrosslinking described above was cross-linked to SecY. The photocrosslinking products of the different Pf3 coat truncated proteins (56-Pf3, 64-Pf3, 72-Pf3, 80-Pf3, and 91-Pf3) were immunoprecipitated with antiserum to SecY (Fig. 3c, left panel, IP, SY). No SecY-Pf3 coat cross-linking products were detected among the series of the nascent proteins, which represent different stages of Pf3 coat membrane insertion (Fig. 3c, left panel, IP, SY). As a positive control, we showed that the antiserum to SecY immunoprecipitated 35 S-labeled SecY, which was synthesized in vitro in an E. coli S30 translation system (Fig. 3c, right panel). These data are consistent with Pf3 coat protein not contacting SecY. The fact that YidC mediates membrane insertion of Pf3 coat, a Sec-independent protein, implies that YidC can work independently of the Sec translocase.
YidC Interacts with Membrane-partitioned Pf3 Coat to Promote Its Transmembrane Configuration-Pf3 coat membrane insertion can be understood as a process of three steps; i.e. the first step is targeting of Pf3 coat protein to the membrane, the second step is partitioning of the Pf3 coat hydrophobic domain into the membrane lipid bilayer, and the final step is the formation of a transmembrane helix with concomitant membrane translocation of the amino-terminal region (23). In the third step, the proton motive force (pmf) is required for the electrophoretic transfer of the amino-terminal tail, which contains two negatively charged amino acid residues (33). Our data obtained by photocrosslinking suggests that YidC functions at the stage of membrane insertion, which might be the membrane partitioning step (the second step) or the orientation step (the third step), to form the transmembrane form of the protein. Therefore, we dissected the membrane insertion of Pf3 coat to determine at which stage YidC functions.
First, we investigated whether YidC is important for the partitioning of the hydrophobic domain of Pf3 coat into the membrane. We used sodium carbonate extraction, which distinguishes peripherally bound proteins from integral membrane proteins. Pf3 coat was pulse-labeled for 20 s in YidC-induced JS7131 cells (treated with arabinose) or YidC-depleted JS7131 cells (treated with glucose). After converting the cells into spheroplasts, the cells were extracted with sodium carbonate (pH 11.5) and then subjected to ultracentrifugation to separate the membrane fraction (pellet) and cytosolic fraction (supernatant). As a control, we confirmed by using proteinase K mapping that the insertion of Pf3 coat was essentially 100% blocked when YidC was depleted (data not shown). The carbonate extraction study showed that almost all the Pf3 coat was found in the membrane fraction when YidC was expressed by the addition of arabinose (Fig. 4a, YidC ϩ ). In the YidC-depleted cells, only ϳ40% of the Pf3 coat is in the membrane fractions, and ϳ60% of Pf3 coat was extracted into the supernatant (Fig. 4a, YidC Ϫ ). This indicates that without YidC, more than half of the Pf3 coat cannot stably partition into the membrane and is extracted. The 40% of Pf3 coat detected in the membrane was still protected from proteinase K digestion, indicating that this portion of Pf3 coat is not oriented correctly in the N out /C in topology. These data suggest that YidC plays a role to orient Pf3 coat in the transmembrane configuration. Therefore, when YidC is absent, the hydrophobic region of Pf3 coat does not span the membrane and can be more easily extracted by sodium carbonate.
The pmf is involved in membrane translocation of the negatively charged residues located within the amino-terminal region of Pf3 coat protein. What is the relationship between the function of YidC and the pmf? We applied photocrosslinking to investigate whether the interaction between Pf3 coat and YidC still occurs if the pmf is destroyed. The 91 amino acid-nascent Pf3 coat protein with amber mutation at L34 was used for photocrosslinking in the presence or absence of 140 M carbonyl cyanide m-chlorophenylhydrazone (CCCP), a protonophore that dissipates the pmf. As previously shown by Kiefer and Kuhn (23), CCCP can efficiently abolish the pmf under these in vitro conditions. Fig. 4b (left panel) shows that there is no difference in the YidC-Pf3 coat photocrosslinking products between the non-CCCP-treated and CCCP-treated photocrosslinking reactions. This result indicates that the pmf is not necessary for YidC to interact with Pf3 coat and that YidC does not function after the pmf-requiring step. Rather, YidC might act synergistically with the pmf to orient Pf3 coat in the membrane, whereby the pmf acts upon the negatively charged amino-terminal tail and YidC functions by interaction with the hydrophobic segment. In addition, because YidC is a membrane protein, the cross-linking of Pf3 coat to YidC corroborates that Pf3 coat can partition into the membrane in the absence of the pmf.
We next examined a Pf3 coat mutant, Pf3-4N, which cannot translocate its hydrophilic domain across the membrane, as judged by protease mapping experiments (23). This mutant lacks negative charges in the amino-terminal tail on which the pmf acts, and therefore the pmf cannot drive the neutral Nterminal tail across the membrane. Truncated nascent chains of 91 residues with the photoprobe at the Leu-34 position were synthesized to study possible interactions with YidC. The photocrosslinking products were immunoprecipitated with YidC antibody and analyzed by SDS-PAGE/autoradiography. Fig. 4b  (right panel) shows that strikingly Pf3-4N can interact with YidC with approximately the same efficiency as wild-type Pf3 coat. We believe that this Pf3 coat mutant, like the wild-type Pf3 coat that accumulates when the pmf is abolished, has an N in /C in topology. In both cases, the proteins are partitioned into the membrane and are able to interact with membrane-bound YidC (see Fig. 4c for proposed intermediates).
We also show that YidC can be cross-linked to another Pf3 coat mutant, Pf3-RD, although its membrane insertion orientation (C out /N in ) is opposite to that of wild-type Pf3 coat (33) (Fig. 4b, right panel). Taken together, the photocrosslinking data demonstrate that the Pf3 coat-YidC interaction occurs independently of the charges flanking the transmembrane segment of Pf3 coat. These findings are also consistent with the idea that the Pf3 coat-YidC interaction takes place prior to or at the same time as the pmf-requiring step. DISCUSSION We have presented data showing for the first time that YidC interacts directly with a Sec-independent membrane protein, namely the Pf3 coat protein, and promotes its membrane insertion. Previously, we have shown that YidC promotes the membrane insertion of the Sec-independent M13 procoat protein (13). However, we could not rule out the possibility that YidC depletion was causing an indirect effect, thereby inhibiting membrane protein insertion. In this study, we have found, using photocrosslinking, that ribosome-bound Pf3 coat nascent chains are cross-linked to YidC when the photoprobe is located either in the center or toward the amino-or carboxyl-terminal ends of the transmembrane segment (Fig. 2). Moreover, we found that YidC binds to a nontranslocated membrane protein, as it is directly cross-linked to Pf3-4N, which cannot insert across the membrane (Fig. 4).
In our photocrosslinking studies, we extended the carboxyl terminus of Pf3 coat such that we could fully expose the hydrophobic domain (residues 19 -36) during synthesis with the protein still attached to the ribosome. This was necessary because Pf3 coat is too short when the ribosome is attached, as around 35ϳ40 amino acids residues are within the ribosome. The Pf3 coat with the extended carboxyl-terminal region, called Pf3-P2, was completely YidC-dependent for membrane insertion (Fig.  1). Efficient cross-linking to YidC was observed only when the carboxyl-terminal region of Pf3 coat had emerged from the ribosome. No cross-linking of Pf3 coat was observed to SecY. These data are consistent with Pf3 coat being inserted by means of a Sec-independent mechanism (31). Interestingly, our cross-linking results show some differences and similarities to those obtained with FtsQ, a Sec-dependent membrane protein, which has been investigated using the same photocrosslinking methodology (15). They are different because the hydrophobic domain of FtsQ first inserts in an environment around SecY and then moves toward YidC. The proposed function of YidC is to integrate the transmembrane regions into the membrane bilayer (19). Our studies with the Sec-independent Pf3 coat are similar to those of FtsQ in that both of these proteins contact YidC efficiently when the transmembrane segments are fully exposed from the ribosomal tunnel. This similarity suggests that YidC may function in a common way for Sec-dependent and Sec-independent proteins. The actual function of YidC might be that of a membrane chaperone to enable the hydrophobic segments of either Sec-dependent or Sec-independent proteins to properly integrate into the lipid bilayer in a transmembrane configuration.
Although the precise function of YidC is not known, it is required for the translocation of the hydrophilic domain of Pf3 coat across the membrane. We found that the Pf3-RD mutant, which inserts also with the inverted topology, interacts with YidC (Fig. 4). Importantly, the Pf3-4N mutant, which does not translocate its hydrophilic domain across the membrane, still FIG. 4. YidC mediates Pf3 coat membrane translocation. a, sodium carbonate extraction shows that YidC affects membrane partitioning under alkali conditions. JS7131 cells expressing Pf3 coat were grown in the presence of arabinose (YidC ϩ ) or glucose (YidC Ϫ ) and pulse-labeled for 20 s with [ 35 S]methionine. The samples were split into two aliquots, one for the sodium carbonate extraction study (upper panel) and the other for protease mapping (as a control; data not shown). The supernatant (S) and pellet (P) fractions were prepared as described under "Materials and Methods." b, left panel, Pf3 coat interacts with YidC in the absence of a pmf. Photocrosslinking was performed in the presence or absence of CCCP treatment with the 91amino acid Pf3 coat nascent chain containing the photoprobe at position L34. CCCP (140 M, final concentration) was added to the photocrosslinking reaction 4 min after translation was initiated. The Pf3-YidC adduct on the gel is indicated with an arrow. b, right panel, photocrosslinking of YidC with wild-type Pf3 coat (WT) and mutant Pf3 coat proteins containing the photoprobe at position L34 of the 91residue nascent chain. The in vitro translation and photocrosslinking was performed as described in the legend for Fig. 2b. The YidC-Pf3 coat adducts and the Pf3 coat nascent proteins are indicated with an arrow. c, a schematic depicting the ribosome-bound Pf3 coat proteins interacting with YidC when the membrane potential is abolished by treating samples containing Pf3-P2 with CCCP (left panel). The right panel shows the Pf3-4N mutant interacting with YidC even though the Pf3 coat cannot insert in a transmembrane form with the amino-terminal region translocated across the membrane. P, M, and C, periplasm, membrane, and cytosol, respectively. The "X" indicates that ⌬⌿, the membrane electrical potential component of the pmf, has been abolished, and the symbol with a wavy arrow through the center of the X indicates that translocation of the amino terminus of the Pf3 coat proteins has been blocked. interacts with YidC (Fig. 4), indicating that YidC recognizes the nontransmembrane coat protein. In addition, YidC still interacts with nascent Pf3 coat proteins that are in the process of inserting in the membrane even when the pmf is abolished by the addition of CCCP. We propose that YidC and the pmf act synergistically to promote membrane insertion, whereby YidC associates with hydrophobic segments and the pmf acts on negatively charged residues. When YidC is depleted within the cell membrane, the hydrophobic region of wild-type Pf3 coat still partitions into the membrane, but some of it falls off of the membrane under alkaline conditions (Fig. 4a). This is in contrast to the M13 procoat protein, which still partitions into the membrane and does not fall off of the membrane in YidCdepleted cells (14). However, M13 procoat protein has an additional hydrophobic segment that most likely promotes the hydrophobic interaction of the protein with the membrane.
In addition to YidC interacting with Pf3 coat with an extended carboxyl-terminal region, we found that Ffh may contact the Pf3 coat protein in the membrane targeting and insertion pathway (Figs. 2 and 3). The Ffh-Pf3 coat interaction might be enhanced because Pf3-P2 is bound to the ribosome and has a better chance to interact productively with Ffh at the ribosome. The wild-type Pf3 coat protein does not require Ffh for membrane insertion. 2 We propose a new model for Pf3 coat membrane biogenesis (Fig. 5) based on the data in this study and on previously published work (23,33). After Pf3 coat is targeted to the surface of the membrane, its hydrophobic segment partitions into the membrane and moves into an environment near YidC, with the Pf3 coat polar amino-terminal tail region and positively charged carboxyl-terminal region located in the aqueous cytosol. Finally, the electrical potential (⌬), a component of the pmf, translocates the amino-terminal tail across the membrane by an electrophoretic mechanism, and at the same time YidC mediates the transmembrane segment insertion.
In conclusion, we demonstrate for the first time, using the photocrosslinking approach, that YidC acts directly on a Secindependent substrate to promote its membrane insertion. YidC functions at a stage of membrane insertion by associating with the membrane-bound hydrophobic region (Fig. 3) and then helping the inserting protein to orient into the transmembrane form (Fig. 4). The YidC-Pf3 coat interaction can occur even when the pmf is disrupted with CCCP and with a Pf3 coat protein mutant that cannot integrate across the membrane. Taken together, the data support the notion that the YidC-Pf3 coat interaction takes place during or before the pmf-requiring step of membrane protein insertion. Pf3 coat is synthesized and then targeted to the inner membrane by the interaction of the positively charged residues in the carboxyl-terminal region with the negatively charged membrane surface. After Pf3 coat membrane targeting, the Pf3 coat hydrophobic region partially integrates and then interacts with YidC. The pmf and YidC then cooperate to promote translocation. The membrane electrical potential component (⌬⌿) of the pmf is responsible for translocating the negatively charged amino-terminal region of Pf3 coat into the periplasm, and YidC mediates stable hydrophobic segment integration into a transmembrane configuration. The carboxyl terminus of Pf3 coat is retained in the cytosol by interaction of its positive charges with the negatively charged membrane surface. P, M, and C represent periplasm, membrane, and cytosol, respectively.