Molecular communication of the membrane insertase YidC with translocase SecYEG affects client proteins

The membrane insertase YidC inserts newly synthesized proteins by its hydrophobic slide consisting of the two transmembrane (TM) segments TM3 and TM5. Mutations in this part of the protein affect the insertion of the client proteins. We show here that a quintuple mutation, termed YidC-5S, inhibits the insertion of the subunit a of the FoF1 ATP synthase but has no effect on the insertion of the Sec-independent M13 procoat protein and the C-tail protein SciP. Further investigations show that the interaction of YidC-5S with SecY is inhibited. The purified and fluorescently labeled YidC-5S did not approach SecYEG when both were co-reconstituted in proteoliposomes in contrast to the co-reconstituted YidC wild type. These results suggest that TM3 and TM5 are involved in the formation of a common YidC-SecYEG complex that is required for the insertion of Sec/YidC-dependent client proteins.


Results
A YidC mutant defective for Sec-dependent membrane insertion. Since the hydrophobic slide of YidC constitutes the major substrate contact site, we had systematically mutated the involved residues located in the transmembrane segments TM3, TM4 and TM5 of YidC (Fig. 1a), 18 . A quintuple serine mutant, termed 5S with mutations at the residues 430, 435, 468, 505 and 509 was tested in the E. coli strain MK6 that has an arabinose inducible promoter upstream of the chromosomal yidC gene 18 . Depletion of the chromosomally encoded YidC was performed by plating the culture on glucose medium that resulted in cell death (Fig. 1b, lane 3). Whereas the induction of the plasmid-controlled cysteine-less wild-type C 0 (lane 5) with IPTG allowed colony formation, the 5S mutant showed a growth defect and no colony growth was evident (lane 6).
The 5S mutant was then tested for the insertion of various substrates. Surprisingly, the insertion of the YidC-dependent M13 procoat was not affected and the cleavage to coat was as efficient as was observed with the wild-type YidC (Fig. 2a). M13 procoat was pulse-labelled for 1 min with 35 S-methionine and analysed by polyacrylamide gel electrophoresis (PAGE) after immunoprecipitation with an antiserum to coat. Only when the empty plasmid was present in the depleted cells, used as a control (lane 6; Fig. S1), did we observe an accumulation of the uncleaved procoat. Similarly, the YidC-5S mutant facilitated the translocation of the C-terminus of the C-tail protein SciP, a C-tail protein. SciP that has an additional cysteine residue at the translocated C-terminus 19 was modified by 4-acetoamido-4′maleimidylstilbene-2-2′disodium sulfonate (AMS) added to the periplasm (Fig. 2b). Previously, it was shown that SciP uses YidC for its membrane insertion but is independent of the Sec translocase 19,20 .
In contrast, the Sec/YidC-dependent subunit a of the FoF1 ATP synthase was affected for membrane insertion in the YidC depleted cells when the 5S mutant was expressed (Fig. 2c). Previously it has been shown that the membrane insertion of subunit a with a C-terminal extension of the periplasmic region (P2) of leader peptidase requires YidC and SecYEG 21 . The plasmid-encoded protein was expressed in JS7131 cells and pulse-labelled for 60 s. Then the cells were converted to spheroplasts and Proteinase K was added to the periplasmic space of the spheroplasts to probe for the translocation of subunit a (Fo-a-P2). When wild-type YidC was expressed, Foa-P2 was digested and a protease-resistant fragment was generated (indicated with an asterisk in Fig. 2c, lane 4), whereas in the YidC-5S expressing cells Fo-a-P2 was mostly resistant to the proteinase indicating that the membrane insertion of Fo-a-P2 was affected. We noticed that the expression of Fo-a-P2 leads to a retardation of the processing of proOmpA (Fig. S2) which is possibly due to a stalling effect of the non-cleared substrate from the translocase. Such a stalling has been previously observed for Sec/YidC-dependent substrates when the YidC function was compromised 8,21 . Taken together, the 5S mutant that showed a growth defect in the complementation assay was functional for the insertion of the Sec-independent proteins M13 procoat and SciP but was defective in the Sec-dependent subunit a insertion. Possibly, the 5S mutations inhibit the YidC-SecYEG contacts that are required for the interaction of YidC with SecY which is required for the membrane insertion of subunit a.  whether YidC or YidC-5S can interact with SecYEG, a single cysteine residue was introduced into the proteins to allow site-specific fluorescent labelling. In YidC and YidC-5S, residue W23 was replaced by a cysteine (Fig. 3a). This residue is located at the periplasmic face of TM1, is well exposed and the substitution to cysteine does not affect its function 22 . In SecY, M142 in the periplasmic loop P2 was used to introduce a single cysteine that is well-exposed in the periplasmic space 23 . The purified YidC protein was labelled with Atto520 and SecYEG with Atto647N (Figs. S3 and S4) with an efficiency of 18% and 24%, respectively. For the binding experiments, we used 0.55 µM YidC in 0.03% dodecyl maltoside and titrated increasing amounts of SecYEG. The donor fluorescence of YidC in the sample was excited at 516 nm. When the fluorescence of acceptor was recorded at 669 nm, energy transfer was observed and increased with the increasing acceptor concentration (Fig. 3b). A maximum of the FRET signal was reached when SecYEG attained a molar 1:1 ratio with YidC. Simultaneously, the donor fluorescence decreased which was recorded at 538 nm (Fig. 3c). The binding affinity between YidC and SecYEG   . For a control, the cleavage was followed under non-depleted conditions (+Ara, odd lanes). (b) SciP with a cysteine residue at 218 in the C-tail was co-expressed with YidC-5S (lanes 1, 2), YidC (lanes 3, 4) or where the empty plasmid was present (lanes 5, 6) and pulse-labelled for 2 min. The cells were non-treated (lanes 1, 3, 5) or treated with AMS to modify the cysteine and shift the protein by 0.5 kDa when the C-terminal tail was translocated to the periplasm (lane 2, 4, 6). (c) Membrane insertion of ATP synthase subunit a that was extended with the P2 domain of leader peptidase at the C-terminus was monitored in the YidC-depleted JS7131 cells. The cells expressing YidC-5S (lanes 1, 2) or YidC (lanes 3, 4) with subunit a-P2 were pulse-labelled with 35 S-methionine for 1 min, converted to spheroplasts and analysed by protease mapping. Proteinase K was added (lanes 2, 4) or not (lanes 1, 3). When the protein was digested with Proteinase K, a protease-resistant fragment was generated (see asterisk). For a control, the cellular protection of cytosolic proOmpA and the digestion of OmpA by the protease were documented (Fig. S2). In all cases, the total cell protein was immunoprecipitated and analyzed by SDS-PAGE and phosphorimaging. The percentage of the cleavage of M13 procoat, of the modification of SciP with AMS and the protease accessibility of Fo-a-P2 were quantified and the values are indicated. www.nature.com/scientificreports/ in detergent was determined with a K d of 27 (± 10) nM (Fig. S5). It is important to note that in these ensemble FRET experiments only a portion of the donor fluorescence energy is transferred to the acceptor and a variety of distances may exist between the dyes and the signal results from an average of these distances.

Reconstitution of YidC and SecYEG into proteoliposomes. The purified and fluorescently labelled
proteins were separately reconstituted with 1,2 dioleoyl-sn-glycero-3-phosphocholine (DOPC) into proteoliposomes. For the reconstitution a molar protein-lipid ratio of 1:1000 was chosen. Equal amounts of the recon- www.nature.com/scientificreports/ stituted proteoliposomes were mixed and the fluorescence was recorded for the YidC-proteoliposomes, the SecYEG-proteoliposomes and the combined proteoliposomes (Fig. 4). The signal of the combined proteoliposomes (purple line) was compared to the sum of the two separate proteoliposomes (black dotted line). The match of the two curves shows that no FRET was detectable. In contrast, when YidC and SecYEG were coreconstituted (red line) the donor signal (at 538 nm) decreased and the acceptor signal increased (at 669 nm) when compared to the signal of the sum of the two separate proteoliposomes (black dotted line). A clear increase of the acceptor signal concomitant with a decrease of the donor signal was observed when SecYEG and YidC were present in the same liposome (a). This was not the case when SecYEG and YidC were present in separate liposomes (b). The Atto520 labelled YidC-5S mutant was then co-reconstituted with SecYEG and analysed in the same way (green line; Fig. 4c). Only small differences to the control were detectable (black dotted line). These results show that the YidC-5S mutant is impaired for SecYEG interaction when present in the same proteoliposome. In a titration experiment with detergent-solubilized YidC-5S and SecYEG only little, non-specific binding was observed (Fig. S6).
Disulphide cross-linking points to the SecY and YidC interacting residues. Single cysteine mutants of SecY and YidC, respectively, were used to pinpoint the residues that are involved in a direct molecular interaction. The respective plasmids were co-transformed into E. coli MK6 cells, induced with 1 mM IPTG for 20 min, pulse-labelled with 35 S-methionine for 3 min, treated with 200 µM DTNB and immunoprecipitated with an antiserum to YidC (Fig. 5). When a cysteine residue was placed at position 123 in the TM3 segment of SecY and at position 432 in the TM3 segment of YidC a cross-linked band was detected (lane 2) that did not appear in the presence of DTT (lane 1). Also, when a cysteine-less SecY variant was co-expressed with YidC 432C, no cross-linked band was detectable (lane 4) although the cellular amount of SecY was not affected (Fig. S7).
Likewise, single cysteine mutants showed that TM8 of SecY interacts with TM5 of YidC. When a cysteine residue was placed at position 328 of SecY (TM8) and 499 of YidC (TM5) a cross-linked band was detected (lane 7) that disappeared when the cells were treated with DTT (lane 8). Taken together, the disulphide cross-linking data show that the lateral gate of SecY, consisting of TM3 and TM8 directly interacts with the hydrophobic slide of YidC, namely the TM3 and TM5.

Discussion
For membrane insertion, a close interaction between the translocase SecYEG and the membrane insertase YidC is required for most of the membrane proteins that use the Sec/YidC dependent pathway. Previously, the existence of the "holo-translocon" has been demonstrated under overproducing conditions 2 . A low-resolution structure was presented that suggests that the lateral gate of SecY and the TM3 of YidC are in contact 23 . We present here further evidence that the lateral gate of the Sec translocase (TM3 and TM8) directly interacts with YidC (TM3 and TM5, respectively), by disulphide cross-links in vivo (Fig. 5). The close interactions between the lateral gate and the hydrophobic slide could generate an extended cavity between the pore of SecY and the hydrophilic groove of YidC in the complex (Fig. 6). This cavity could then be used to support the insertion of large membrane proteins allowing the substrate proteins to enter the YidC/SecYEG complex between the hydrophobic slide and the lateral gate. Indeed, a recent cryo-electron microscopic structure of a partially inserted proOmpA protein showed the involvement of TM3, TM7 and TM8 of the SecY lateral gate in the early insertion events 7 . www.nature.com/scientificreports/ To further explore the interaction and the binding affinity between SecYEG and YidC we used fluorescence resonance energy transfer (FRET) in vitro. The two labelled and purified proteins interact in detergent with a high affinity, which, however, might not reflect the in vivo situation. To investigate the interaction in the membrane both proteins were reconstituted first in separate liposomes (Fig. 4b). In this case, no FRET between SecYEG and YidC was observed. However, co-reconstitution of SecYEG and YidC led to an efficient FRET signal (Fig. 4a) that corroborates an interaction of the two components. In cells, the interaction between SecYEG and YidC might be different and influenced by additional factors. Previously, it was shown that SecF is directly involved in contacting YidC residues 24,25 . Also, multiple positions between the Sec translocase and YidC were identified as weakly interacting residues 17 .
Further support for the functional interaction between the Sec translocase and the hydrophobic slide of YidC is that the YidC-5S mutant that is defective in SecYEG interaction is partially inhibited in membrane insertion www.nature.com/scientificreports/ of a YidC/Sec-dependent substrate (Fig. 2c). When five of the hydrophobic slide residues, namely M430, A435, P468, F505 and F509 were substituted with serines, the interaction to SecYEG was severely reduced (Fig. 4c). All these mutations are localized in the upper part of the hydrophobic slide close to the periplasmic leaflet of the membrane. The YidC-5S mutant is affected to insert subunit a of the ATP synthase into the membrane. This protein has been studied with a carboxyl-terminal extension of the leader peptidase P2 domain 21 . Whereas the translocation of the amino-terminal domain of subunit a required SRP, YidC and the membrane potential, the translocation of the central loop is dependent on SRP, the SecYEG translocase and YidC. To our surprise, the five serine mutations in the hydrophobic slide had no negative effect on the insertion of the Sec-independent substrates M13 procoat and SciP (Fig. 2). Only when additional contacting residues were changed to serines, namely F424, L427, F476 and F502 in the YidC-9S mutant, was a Sec-independent substrate affected 11 . Presumably, the initial step to insert a YidC-only substrate is promoted by the hydrophobic residues close to the cytoplasm of the slide 15 .
We also found that YidC-5S cannot complement the depletion of wild-type YidC and is defective for growth (Fig. 1b). In conclusion, the Sec-dependent function of YidC is essential similar to the Sec-independent function of YidC as had been shown earlier 26 . Since the insertion of CyoA as an essential component of the ubiquinol oxidoreductase and the subunit c of the ATP synthase require YidC for insertion it was concluded that the depletion of YidC affecting Sec-independent substrates dissipates the transmembrane potential. This might be the main reason to cause cellular death.
Taken together, we show here that the YidC-5S mutant is defective in interacting with SecYEG and the insertion of Sec/YidC dependent subunit a of the ATP synthase into the membrane is affected. This can also explain why the mutant is not capable of complementing the YidC depletion strain and prevents normal cell growth.

Methods
Bacterial strains and plasmids. YidC depletion was carried out using E. coli strains MK6 11 and JS7131 8 .
The latter has a yidC gene with the araC-araBAD promoter in the lambda attachment site and a deletion in the original yidC gene. In the MK6 strain, the promoter of the yidC gene had been exchanged with the araC-araBAD promoter cassette. Depletion of the chromosomally encoded YidC was achieved by growth of MK6 in the presence of 0.4% glucose. For complementation assays, MK6 was transformed with pGZ119EH 27 expressing the respective YidC proteins under the control of the tac promoter. Co-expression of M13 procoat was performed with pMS-M13 procoat 28 , SciP with pMS-SciP-C 29 and pMS-Su-a-P2 21 , derivatives of the pMS119EH vector 30 . For the purification of SecYEG and YidC E. coli SF100 and C43 were used, respectively 5,29 . Translocation assay with the wild-type M13 procoat. The wild-type M13 procoat expressed from a pMS119EH derivative was co-expressed with YidC-5S or YidC to test for their capability of inserting a substrate. Overnight cultures of MK6 harboring two plasmids, one encoding the YidC mutants or the YidC control (in pGZ119EH) and the other the wild-type M13 procoat (in pMS119EH), were grown in LB (200 µg/mL ampicillin, 25 µg/mL chloramphenicol, 0.2% arabinose, 0.4% glucose) overnight at 37 °C. The cells were washed twice with LB medium and then inoculated 1:100 in fresh LB medium (200 µg/mL ampicillin, 25 µg/mL chloramphenicol, 0.4% glucose) and grown under YidC depletion conditions to an OD 600 = 0.5 at 37 °C. 200 µL were taken, washed twice and resuspended in 200 µL M9 medium lacking methionine. After 45 min growth at 37 °C, 1 mM IPTG was added for 10 min. 20 µCi [ 35 S]-methionine was added for 1 min, followed by the addition of 10% TCA. The samples were then processed for immunoprecipitation and PAGE as previously described 29 .
Translocation assay of SciP with AMS. The pMS119EH plasmid-encoded SciP-C 19,29 was co-expressed with pGZ119EH encoding YidC-5S in MK6 cells under YidC depletion conditions. The cells were grown in LB (100 µg/mL ampicillin, 37 µg/mL chloramphenicol, 0.4% glucose) to an OD 600 = 0.5 at 37 °C. Then 1 mL culture was sedimented, washed and resuspended in M9 minimal medium lacking methionine and split into two cultures (+/− AMS). The cells were grown for another 45 min at 37 °C, induced with 1 mM IPTG for 10 min. 2.5 mM AMS (Molecular Probes) was added for 1 min followed by pulse labelling with 15 µCi 35 S-methionine for 2 min. Nonradioactive l-methionine was added for 10 min, and the AMS reaction was quenched by the addition of 10 mM DTT for further 10 min. The cells were acid-precipitated and immunoprecipitated (his tag antibody) as previously described 29 .
Membrane translocation of subunit a of ATP synthase. The ATP synthase subunit a-P2 was coexpressed with one of the YidC constructs to examine membrane insertion following the protocol of Spann et al. 29 . Overnight cultures of the YidC depletion strain JS7131 bearing two plasmids, one encoding the YidC defective mutant or the YidC blank control (pGZ119) and the other encoding subunit a-P2 (pMS119), were grown in LB (50 μg/mL ampicillin, 25 μg/mL chloramphenicol, 0.2% arabinose) overnight at 37 °C. After washing the cells twice with LB medium, the cells were inoculated 1:100 in fresh LB medium (50 µg/mL ampicillin, 25 µg/mL chloramphenicol, 0.2% glucose) and grown under YidC depletion conditions for 4 h. The cells were then washed with M9 medium and transferred into M9 lacking methionine, and grown for 30 min at 37 °C. Then, the culture was induced with 1 mM IPTG for 7 min, and then pulse-labelled with [ 35 S]-methionine for 1 min. Cells were collected by pelleting in a microfuge at 4 °C and resuspended in 40% sucrose and 33 mM Tris-HCl (pH 8.0) and converted into spheroplasts by treatment with lysozyme (5 µg/mL) and 1 mM EDTA (pH 8.0) on ice for 30 min. The spheroplasts were treated with or without proteinase K (0.8 mg/mL final concentration) for 1 h on ice. 5 mM PMSF (final concentration) was added to stop the protease reaction. Subunit a-P2 was immunoprecipitated with leader peptidase antibody and the samples were analyzed by SDS-PAGE and phosphorimaging. Purification of YidC and SecY. For purification, 2 L of LB medium containing 12.5 µg/mL chloramphenicol or 100 µg/mL ampicillin, respectively, was used to grow the E. coli C43 cells bearing pGZ119-YidC-5S or pMS119EH-YidC-wt or E. coli SF 100 cells bearing pTrc99a-SecYEG plasmid 31 . After reaching an OD 600 of 0.6 for YidC and 0.9 for SecYEG, respectively, protein expression was induced for 2 h with 0.5 mM IPTG and the cells were harvested by centrifugation (7000 g, 4 °C, 10 min). The cells were shock-frozen in buffer 1 (20 mM Tris-HCl pH 7.5 for YidC and pH 8.0 for SecYEG, 300 mM NaCl, 10% glycerol) supplemented with 1 mM PMSF and 1 mM DTT. To the thawed cells 1 mM MgCl 2 and DNase I were added. After cell disruption by a One-Shot cell disruption system (Constant Systems Ltd) and separating the cell debris by centrifugation (17, Supplementary Fig. S3). The labelled proteins were shock-frozen and stored protected from light at − 80 °C.

Reconstitution of YidC and SecYEG in proteoliposomes.
YidC and SecYEG were reconstituted into DOPC-liposomes by the extruder and Bio-Bead method. The proteins were dialyzed to remove the glycerol, mixed in a molar lipid-to-protein ratio of 1:1000 with liposomes and extruded 15 times. To remove the detergent, the proteoliposomes were twice rotated with activated Bio-Beads for 1 h at 4 °C. The proteoliposomes were collected by centrifugation (Beckman Airfuge, 20 psi, 10 min, 4 °C), resuspended in buffer 1 (see purification) without glycerol and immediately used for measurements.

Förster resonance energy transfer (FRET) measurements.
For measuring FRET, the donor dye (Atto520) was excited at 516 nm (Fluorolog, HORIBA Jobin Yvon GmbH). The fluorescence spectrum was recorded at 520-700 nm to detect the fluorescence maximum of the donor at 538 nm and the fluorescence maximum of the acceptor at 669 nm.

FRET data analysis.
To determine the K d value of the SecY-YidC interaction in detergent a titration was performed and the fluorescence spectrum of each titration point was recorded. Because of the increasing volume and the addition of acceptor, the concentration of the labelled proteins changed during titration. To correct this effect of dilution and concentration, respectively, buffer controls (buffer to YidC-Atto520 and SecY-Atto647N to buffer) were performed equivalent to the titration of the proteins. The corrected maxima of the signals were plotted against the concentration of the acceptor protein (SecY-Atto647N) and linearized according to Lineweaver-Burk to determine the K d value. For the measurements on the reconstituted samples, no buffer controls were needed. To determine the background signal, the proteins were reconstituted separately, in addition to the mixed samples. The sum of the signals of the two separately measured SecYEG and YidC proteoliposomes was used for normalizing the signal of the mixed samples to determine the increase or decrease of the fluorescence because of FRET.
In vivo disulphide cross-linking. For disulphide cross-linking, single cysteine mutants of SecY and YidC were prepared. The E. coli MK6 strain with the chromosomal YidC under the control of the araC-araBAD promoter was co-transformed with a pTrc99a plasmid carrying the mutated secYEG genes and a pGZ119EH plas- www.nature.com/scientificreports/ mid carrying the mutated yidC gene. Cultures were grown at 37 °C in LB medium supplemented with 100 µg/ mL ampicillin, 12.5 µg/mL chloramphenicol and 0.4% glucose. After reaching an OD 600 of 0.4 to 0.5 following growth for 3 h for depletion of chromosomal YidC, 400 µL of the cultures were washed twice with M9 minimal medium and incubated in 400 µL M9 minimal medium for 45 min at 37 °C. After induction with 1 mM IPTG for 20 min, the proteins were pulse-labelled with 35 S-methionine for 3 min and treated for 10 min with freshly prepared 200 µM DTNB for oxidation. The protein pellets after TCA precipitation were washed with acetone and boiled for 5 min at 95 °C in 100 µL 2% SDS/10 mM Tris-buffer (pH 8). 1 mL of cold TENTX-buffer (150 mM NaCl, 10 mM Tris, 1 mM EDTA, 2% Triton-X 100, pH 8) and 20 µL Staph A (10% in PBS) were added for reabsorption (45 min). For the immunoprecipitation, antiserum to YidC was added to the supernatant and incubated overnight. After precipitation with 30 µL Staph A and washing twice with TENTX-buffer and once with TENTX-buffer without Triton-X 100, the samples were divided in half. The samples were resuspended in SDS sample buffer with or without DTT and analyzed by SDS-PAGE and following phosphor imaging.

Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request including the raw data of Figs. 3, Fig. 4 and Fig. S6.