|Home | About | Journals | Submit | Contact Us | Français|
Members of the YidC/Oxa1/Alb3 protein family mediate membrane protein insertion, and this process is initiated by the assembly of YidC·ribosome nascent chain complexes at the inner leaflet of the lipid bilayer. The positively charged C terminus of Escherichia coli YidC plays a significant role in ribosome binding but is not the sole determinant because deletion does not completely abrogate ribosome binding. The positively charged cytosolic loops C1 and C2 of YidC may provide additional docking sites. We performed systematic sequential deletions within these cytosolic domains and studied their effect on the YidC insertase activity and interaction with translation-stalled (programmed) ribosome. Deletions within loop C1 strongly affected the activity of YidC in vivo but did not influence ribosome binding or substrate insertion, whereas loop C2 appeared to be involved in ribosome binding. Combining the latter deletion with the removal of the C terminus of YidC abolished YidC-mediated insertion. We propose that these two regions play an crucial role in the formation and stabilization of an active YidC·ribosome nascent chain complex, allowing for co-translational membrane insertion, whereas loop C1 may be involved in the downstream chaperone activity of YidC or in other protein-protein interactions.
In eubacteria, membrane proteins constitute ~30% of the total proteins synthesized in the cell (1). They participate in signal transduction, respiration, energy generation, as well as the transport of a variety of ions, solutes, and macromolecules. The majority of the inner membrane proteins are destined for and inserted into the membrane via the SecYEG translocon-mediated pathway (2, 3). However, a specific subset of membrane proteins requires YidC insertase for integration into the membrane. Escherichia coli YidC is an abundant membrane protein, with ~2500 copies per cell (1) and it is involved in the insertion, folding and/or assembly of membrane proteins into the cytoplasmic membrane independently or in concert with the SecYEG translocon (4). E. coli YidC is essential for cell viability (5) and has been shown to function as an insertase in the membrane insertion of the filamentous phage Pf3 coat and M13 pro-coat proteins (5, 6) and the endogenous substrates F0c (7), MscL (8, 9), and TssL (10). In cooperation with the Sec translocase, YidC assists in the membrane insertion of CyoA (11, 12), NuoK (13), and F0a and F0b subunits of F1F0 ATPase (14), and the translocation of the periplasmic loop 1 and loop 2 of TatC (15). It also acts as a chaperone in the folding of lactose permease LacY and MalF (16, 17).
All members of the YidC/Oxa1/Alb3 protein family share a very conserved hydrophobic core region consisting of five transmembrane segments (TMS)2 connected by hydrophilic loops (18), but the E. coli YidC possesses an extra N-terminal TMS1 linked by a large periplasmic domain P1 to the TMS2 (19). Recently, Kumazaki et al. (20) reported the crystal structures of YidC2 from Bacillus halodurans and YidC of E. coli, both solved in the lipid environment. The high resolution structures revealed that the global arrangements of the conserved core regions in E. coli YidC and B. halodurans YidC2 are in high agreement with the folding of the five TMSs into a positively charged groove, whereas the hairpin-like C1 domain that consists of two antiparallel helices is more flexible in the E. coli YidC.
Like the SecYEG translocon, the YidC protein family facilitates co-translational substrate insertion involving ribosome binding to initiate the insertion process. The C-terminal regions of the mitochondrial Oxa1 and Streptococcus mutants YidC1 and YidC2 have been shown to be crucial for the contact with ribosomes (21,–23), and deletions within these domains compromised the protein insertion function. In E. coli, the C terminus of YidC is positively charged, but it is substantially shorter (13 residues; total charge +5) in comparison with Oxa1 (86 residues; +14), YidC1 (33 residues; +9), and YidC2 (61 residues; +14) (19, 22, 23). Recently, we have investigated the interaction of YidC with translation-stalled ribosomes by means of the fluorescence correlation spectroscopy (FCS) technique under physiological condition using nanodisc-embedded YidC to mimic the native lipids environment (24). In contrast to earlier models (25), we found that monomeric E. coli YidC was sufficient for ribosome binding, and thus it formed a minimal functional unit. The analysis highlighted the role of the C terminus of YidC in the ribosome binding but also suggested it to be not the sole determinant, so alternative ribosomal contacting sites were proposed within the positively charged C1 and C2 domains of YidC (25). The chimeric E. coli YidC with an extended C-terminal tail from Rhodopirellula baltica YidC (YidC-Rb) exhibited enhanced binding of translating ribosome forming primary interaction sites on the ribosomal rRNA helix 59 and the ribosomal protein L24, as shown by the cryo-electron microscopy structure of the YidC-Rb·RNC complex (26). Lately, Wickles et al. (27) built a structural model of E. coli YidC via the intramolecular co-variation analysis, which appeared in agreement with the experimentally solved structure. The model was applied to interpret the interaction of E. coli YidC with the RNC-F0c visualized in cryo-electron microscopy. The residues Tyr-370 and Tyr-377 in the C1 loop and Asp-488 in the C2 loop of YidC were suggested to be directly engaged in the ribosome binding at His-59 and the protein Leu-23, respectively (27). Substitution of the residues compromised the vegetative growth of cells; however, the role of cytoplasmic loops in ribosome binding and insertion activity of YidC has not been studied.
Here, we aimed to determine the regions of YidC involved in ribosome binding and further investigate the contribution of the YidC·ribosome physical interaction to the insertion process. Herein, we made sequential deletions within the C1 and C2 loops of E. coli YidC and studied their activity in vitro by means of biochemical and biophysical assays and also checked functional properties of these YidC variants in vivo. Truncations within loop C2 reduced ribosome binding to YidC at the membrane interface and combining deletions within C2 and at the C terminus of YidC also inactivated YidC in membrane insertion of F0c and MscL, suggesting that these two sites build a platform for ribosome docking. On the other hand, modifications within the loop C1 severely affected in vivo functioning of YidC but did not interfere with ribosome binding or substrate insertion, suggesting that the C1 loop is involved in chaperone/foldase activity of YidC or protein-protein interactions at the membrane interface or the insertion of yet unknown YidC substrates that are essential for cell viability.
The YidC depletion strain E. coli FTL10 (28) was a kind gift of Frank Sargent (University of East Anglia, Norwich, UK). E. coli SF100 (29) was used to express the YidC variants. Plasmid pKA107 was used as the template, in which the endogenous cysteine residue at position 423 of YidC was substituted for a serine, whereas a solvent-exposed cysteine was introduced at position 269, which did not affect the functionality (24). Primers were designed annealing to the flanking regions of the selected deletions in the cytosolic loops C1 and C2 of E. coli YidC (Table 1). PCR products containing truncated yidC genes were treated with DpnI at 37 °C for 2 h and then subject to gel purification. After phosphorylation and self-circularization, the resulting plasmids were multiplied in DH5α cells and verified by DNA sequencing. Plasmids containing the genes encoding for YidC mutated proteins were subject to Eco88I (AvaI) and HindIII double digestion, and the obtained DNA fragments with truncated yidC genes were recloned into fresh pTrc99a, yielding plasmids pKA120–126 (Table 1). As for the mutated yidC gene containing deletions in the C2 domain and at the whole C terminus, PCR was carried out using plasmid pKA125 as the template. The resulting yidC double truncated gene was subject to XbaI and HindIII double digestion and cloned into XbaI/HindIII double-digested pTrc99a vector, yielding the construct pYP00.
To study the functionality of E. coli YidC variants, E. coli FTL10 cells were transformed with plasmids harboring genes encoding for mutated YidC proteins and grown under aerobic or anaerobic conditions. For aerobic growth, cells were grown overnight in the LB medium supplemented with 0.2% arabinose (w/v) and ampicillin (100 μg/ml). To deplete YidC, overnight cultures were diluted into fresh LB medium supplemented with 0.2% glucose (w/v) and ampicillin (100 μg/ml) and grown at 37 °C for 3 h. After adjusting to the same A600, 10-fold dilution series of each culture were made and 3 μl of each dilution was spotted on LB plates in the presence of 0.2% arabinose (w/v) or 0.2% glucose (w/v) and 50 μm isopropyl 1-thio-β-d-galactopyranoside (IPTG). The plates were incubated at 37 °C at least 10 h, and the growth phenotype of the diluted cultures displayed on LB plates supplemented with glucose and IPTG represents the deletion effect on YidC function. For the anaerobic growth, cells were grown overnight at 37 °C in the basal anaerobic medium containing K2HPO4 (0.7%), KH2PO4 (0.2%), NH4Cl (0.1%), MgSO4·7H2O (0.02%), CaCl2·2H2O (0.002%), SL9 trace element solutions (0.1% v/v) supplemented with the carbon source glycerol (0.5% v/v) and the electron acceptor sodium fumarate (10 mm), and 0.2% arabinose (w/v) was supplied initially to induce the expression of endogenous yidC gene. To deplete the endogenously expressed YidC, overnight cultures were collected and washed twice with basal anaerobic medium and were diluted 2-fold into the same medium supplemented with glycerol and sodium fumarate. The procedure was repeated until the growth cessation. The cells were subsequently serially diluted, and 3 μl of the cell suspension were spotted on anaerobic medium plates containing 0.2% arabinose (w/v) or 50 μm IPTG and 0.5% glucose, respectively. The anaerobic growth was carried out in sealed bottles under O2-free environment.
Histidine-tagged YidC proteins or SecYEG were expressed into E. coli SF100, and the inner membrane vesicles were prepared as described previously (24, 30). Target proteins were purified by Ni+-NTA affinitychromatography. Firstly, inner membrane vesicles were solubilized in 50 mm Tris-HCl, pH 8.0, 1% n-dodecyl β-d-maltoside (DDM), 100 mm KCl, and 10% glycerol at 4 °C for 30 min. As for YidC variants, 200 μm tris (2-carboxyethyl) phosphine was added to inhibit disulfide bonds formation. Then solubilized proteins were mixed with Ni+-NTA-agarose beads for 1 h at 4 °C on a rolling bank, washed with 3 bed volumes of buffer A (50 mm Tris-HCl pH 8.0, 0.1% DDM, 100 mm KCl, 10% glycerol) supplemented with 50 mm imidazole, and finally eluted with 300 mm imidazole in buffer A. For optional labeling, Ni+-NTA-bound YidC variants were incubated with the labeling buffer B (100 mm K2HPO4/KH2PO4, pH 7.0, 0.1% DDM, 100 mm KCl, 10% glycerol, and 200 μm tris (2-carboxyethyl) phosphine) supplemented with 200 μm Alexa Fluor 488 C5-maleimide (Life Technologies/Invitrogen) for 2 h at 4 °C on a rolling bank, followed by extensive washing with 10 mm imidazole and finally eluted by 400 mm imidazole in buffer B. Protein concentration and labeling efficiency were estimated spectrophotometrically. The extinction coefficient values for YidC at 280 nm and Alexa Fluor 488 at 500 nm were 96,000 and 72,000 cm−1 m−1, respectively.
To ensure a native membrane environment, the fluorescently labeled YidC variants were reconstituted into the lipid bilayers supported by the major scaffold protein (MSP) to form nanodiscs according to the previous method with minor modifications (24, 31) Synthetic membranes composed of DOPG:DOPE:DOPC (dioleoylphosphatidylglycerol: dioleoylphosphatidylethanolamine:dioleoylphosphatidylcholine) at a molar ratio of 3:3:4 were used to generate stable nanodiscs (32). To obtain nanodiscs containing single YidC molecule (YidCmono-Nd), detergent-destabilized lipids were mixed with MSP1D1 and YidC at the molar ration of 250:10:1. The YidC-containing nanodiscs were spontaneously assembled upon removal of the detergent by Bio-Beads SM2 sorbent (Bio-Rad), yielding nanodiscs with a diameter of ~10 nm (33). Size exclusion chromatography was then employed to separate the empty nanodiscs and YidCmono-Nd. YidC-enriched factions were used for FCS measurement. For preparing empty nanodiscs, 2% of fluorophore-conjugated DOPE-Atto 488 (Atto Tech) was added to the lipid formulation, and lipids and MSP were mixed at the molar ratio 25:1.
Generation of SecM-stalled ribosomes carrying the F0c nascent polypeptide chain was performed as described (24, 34). The stalled polypeptide chain is composed of N-terminal 3× Strep tag followed by a sequence encoding for the 1–44 residues of F0c containing the first TMS and flanking polar regions, with SecM arresting sequence fused to the C terminus. RNCs were expressed in vivo in E. coli and isolated by a StrepTactin affinity chromatography (IBA) as previously described (34). TnaC-stalled RNCs were prepared in vivo according to the published protocol (35). The N-terminal polyhistidine tag at the F0c nascent chain was removed upon specific proteolysis by the 3C protease.
FCS experiments were performed on the inverted confocal microscopy LSM710 in combination with the Confocor 3 module (Zeiss GmbH) and on a home-built set-up described previously (36). Prior to the experiment, the known diffusion coefficient of Alexa Fluor 488 in water was used to calibrate the confocal volume (D = 400 μm2/s) (37). To analyze the interactions between fluorescently labeled YidC and RNC-F0c, YidCmono-Nd was diluted to a final concentration of ~50 nm in buffer (50 mm HEPES-KOH, pH 7.2, 100 mm KCl, 5% (v/v) glycerol, and 5 mm MgCl2), and a 5-fold excess of RNCs was added. The reaction was incubated at room temperature for 5–10 min before subject to the FCS measurement. The fluctuation of fluorescence arising from the fluorescent YidC-Nd complexes was monitored and autocorrelated over the time to measure average diffusion time. Autocorrelation curves were fitted using the two-component model as described in the previous study (24, 34). The individual diffusion times for the two-component analysis were determined in separate experiments. The diffusion of YidC-Nd alone was taken from measurements in the absence of RNC-F0C, and the YidC-Nd·RNC-F0C complex was approximated by measurements with labeled RNC-F0C alone. During fitting, the two diffusion times were fixed, and only the amplitudes were varied. The fluorescence lifetime data available within our FCS data indicate that binding of RNC-F0C does not alter the fluorescence intensity. Hence, the amplitude of the two components represents their relative populations. The background fluorescence was analyzed using nonlabeled YidC-Nd and RNCs at stated concentrations. The intensity of the background fluorescence did not exceed 5% of the fluorescently labeled YidC-Nd.
In vitro synthesis of the insertion substrates was performed as described before (30). The in vitro transcription and translation was carried out at 37 °C for 20 min in the presence of EasyTag EXPRESS35S protein labeling mix (PerkinElmer Life Sciences) and empty liposomes or proteoliposomes containing YidC and/or SecYEG. A small fraction of the reaction was taken as a synthesis control.
To study the role of the cytoplasmic loops in the YidC function, we introduced sequential deletions within YidC as illustrated schematically in Fig. 1. Five deletions of 10 amino acids each were introduced within loop C1 starting at position 374. For loop C2 two deletions, 5 residues each, were introduced starting at position 485. To probe the activity of designed YidC variants in vivo, pTrc99a plasmids bearing the genes encoding for the YidC mutants were transformed into E. coli strain FTL10 (28), in which the endogenous chromosomal yidC gene was deleted, whereas a copy of the yidC gene was placed under the control of the arabinose-inducible promoter and integrated into the chromosome. Therefore, growth of the FTL10 is strictly arabinose-dependent. In the presence of arabinose, the transcription of the yidC gene is initiated, thereby exhibiting normal growth phenotype, whereas glucose supplementation inhibits the chromosomal yidC expression. The activity of designed YidC mutants can be examined by observing whether they rescue the growth defect of strain FTL10 when grown with glucose. Cells harboring the YidC variants displayed a robust growth phenotype under arabinose conditions (Fig. 2). However, under glucose and IPTG conditions, different growth phenotypes were observed. As expected, cells containing the empty vector pTrc99a were unable to grow without arabinose, confirming that depletion of native YidC led to the loss of cell viability (4), whereas IPTG-inducible wild-type YidC could compensate for the growth loss (Fig. 2, YidC). Remarkably, except for YidCΔ1 (Δ374–383, the numbers represent the position of the amino acid residues in YidC), which failed in growth complementation, all YidC mutants were capable of restoring the growth deficiency of the YidC depletion strain, although to varying extents (Fig. 2A, YidCΔ1–YidCΔ7). The deletions within the C1 loop had detrimental (YidCΔ1) or moderate (YidCΔ2-Δ5) effects on cell survival, suggesting an important role of the dynamic helical hairpin region for YidC activity (40, 41). Interestingly, a deletion at the tip of the hairpin had the least effect on cell growth (YidCΔ3), because it is unlikely to distort the hairpin structure. Both deletions within loop C2, YidCΔ6 (residues 485–489) and YidCΔ7 (residues 489–493), could fully rescue the growth defect of the YidC defective strain, confirming that the C2 domain of YidC is not crucial for its activity (42). This was also the case for the deletion of the C-terminal end of YidC (YidCΔC) (Δ536–547) in agreement with previous data (24).
Depletion of YidC also impairs cell growth under anaerobic conditions cause by defects in the membrane biogenesis of the anaerobic respiratory complexes, such as NADH dehydrogenase and the F1F0 ATPase (43). To determine whether the designed YidC variants could fulfill the growth complementation function under anaerobic conditions, E. coli strain FTL10 harboring the genes encoding for YidC variants were grown in anaerobic medium supplemented with 0.5% w/v glycerol and 10 mm sodium fumarate serving as the nonfermentable carbon source and electron acceptor, respectively. As shown in Fig. 2B, under arabinose-inducible conditions, the chromosomal copy of yidC genes in all strains were fully expressed to maintain growth. Similar to the growth pattern under aerobic conditions, in the presence of IPTG and glucose all the YidC variants, excluding YidCΔ1, were complementation positive and could support cell growth comparable with the wild-type YidC, indicating a generic phenotype of the mutants.
To rule out the possibility that the lack of activity of the YidCΔ1 mutant resulted from poor expression or structural instability, the total membrane fraction containing overexpressed YidCΔ1 protein was isolated and subjected to immunoblotting analysis using antiserum against the polyhistidine tag, wild-type YidC serving as the loading control. Immunoblotting confirmed that the YidCΔ1 variant was expressed at the same level as the wild-type YidC protein (data not shown). Taken together, our data verified that the C2 domain and the C terminus are dispensable for YidC activity, whereas the C1 domain (region 374–383) of YidC is a functionally critical region.
As described above, the YidCΔ1 mutant could not restore the growth of the YidC depletion strain. A recent study has also suggested that deletions within the C1 loop dramatically reduced the insertion of YidC-only substrates of Pf3 coat and M13 procoat but had a weak effect on insertion of E. coli endogenous protein CyoA (41). Here, we set out to test whether truncated variants of YidC could still fulfill the role as insertases in the in vitro membrane insertion of an endogenous substrate. F0c is a small membrane-embedded subunit c of F1F0 ATPase that exclusively relies on YidC for the membrane insertion (Fig. 3A) (7, 45). To test the YidC insertion activity, the established in vitro co-transcription, translation, and insertion assay in the presence of [35S]methionine and YidC-containing proteoliposomes was employed. Purified YidC variants were reconstituted into liposomes composed of E. coli phospholipids. SDS-PAGE demonstrated that all variants could be reconstituted into the liposomes as efficient as the wild-type YidC (Fig. 3, B and D). Subsequently, the lipid bilayer insertion efficiency of newly synthesized F0c was assessed by the protein accessibility to proteinase K.
As expected, there was no insertion in the absence of proteoliposomes, and the synthesized F0c was fully digested byproteinase K (Fig. 3C, buffer lane). Despite the background insertion resulting from the empty liposomes, a significant protease-protected fraction of F0c was observed when wild-type YidC-containing proteoliposomes were present (Fig. 3C, lane WT). Remarkably, protease K-resistant F0c was also observed when YidCΔ1-containing proteoliposomes were introduced in the in vitro transcription/translation system (Fig. 3C, lane Δ1). Also, all other designed YidC mutants demonstrated insertase activity of the F0c substrate at a level similar to the wild-type YidC (Fig. 3E). Our data suggest that the region 374–383 of YidC is not essential for insertion activity despite its disabled function in cell growth complementation. In addition, no other specific region within the cytosolic domains of YidC was solely found to be responsible for the insertase activity.
To probe the interaction of YidC with programmed ribosomes, we employed the highly sensitive confocal microscopy-based technique of FCS (46). In FCS, fluorescence fluctuations arising from labeled molecules migrating through the confocal volume of a laser beam are recorded and autocorrelated over the recording time. For the case of translational diffusion, the average residence time of the molecule within the illuminated confocal volume can be measured by FCS, and it is inversely proportional to the diffusion coefficient D, which is determined by the hydrodynamic radius of the molecule. Thus, interactions of a small membrane protein, such as YidC, with a large macromolecular complex, such as the ribosome, can be unambiguously detected because of a decrease in the protein's diffusion coefficient.
To exclude the unspecific binding of YidC to ribosomes, the measurements were performed with Alexa Fluor 488-labeled YidC reconstituted into a lipid membrane provided by nanodiscs (24) (Fig. 4). Autocorrelation curves were fitted using the two-component model as described previously (24, 34). As suggested by the autocorrelation function of nanodisc-embedded YidC, diffusion coefficients for all YidC variants varied within a narrow range of 39 ± 2 cm2/s. To monitor the binding of YidC to ribosomes, RNC-F0c complexes (Fig. 4) were added in 5-fold excess to YidC-Nd, and the binding efficiency was analyzed based on the shift of the autocorrelation curves caused by the change in the size-dependent diffusion coefficient (Fig. 5A). Interactions of wild-type YidC-Nd with RNCs caused a prominent shift in the autocorrelation trace, whereas only weak binding accounting for ~20% of nanodiscs was observed for the C-terminal truncated YidC at the same ribosome:YidC ratio (Fig. 5A) in agreement with previously published data (24).
RNC binding was analyzed for all YidC variants, and their binding efficiencies were calculated. Deletions within the C1 domain of YidC had no influence on the ribosome binding efficiency (Fig. 5B), and the YidCΔ1 variant, which severely compromised YidC functionality in the cell viability assay (Fig. 2), exhibited full binding efficiency to RNCs (Fig. 5C). In contrast, deletion of the region 485–489 within the C2 domain (YidCΔ6) caused a strong reduction in YidC·ribosome complexes. Only 20% of YidCΔ6 were associated with ribosomes that was similar to YidCΔC variant. The deleted region included a conserved residue, Asp-488, which was previously suggested as a potential “hot spot” in ribosome binding (27). Another truncation within the C2 loop, YidCΔ7, had a weaker effect on RNC binding, but it required prolonged incubation up to 20 min to achieve an equilibrium between the free and RNC-bound states. No temporal variations in complex assembly were observed for other YidC variants. This suggests that the modification within the C2 loop altered the binding kinetics, either reducing the association rate or allowing rapid dissociation of the ribosome prior the nascent chain insertion, hence the slow overall binding. The nascent chain exposed contains the first transmembrane domain of F0 subunit c and a polar purification Strep tag, which needs to be translocated upon the TM insertion, thus posing a potential bottleneck for the insertion process. To test the hypothesis that the tag weakens the binding, we employed RNC-F0c with a cleavable purification tag of polar N-terminal amino acids that could be specifically cleaved by 3C protease. When the purification tag was removed from the F0c nascent chain, rapid YidCΔ6:RNC assembly at wild-type levels was observed, whereas binding of other YidC variants was not affected (Fig. 5C) as compared with the tagged RNC-F0c binding (Fig. 5B). With the YidCΔC, weak RNC-F0c binding is observed. To test whether this was mediated by YidC or by hydrophobic interactions between nanodiscs and the transmembrane domain of F0c, we analyzed interactions between RNC-F0c and empty nanodiscs containing fluorophore-conjugated lipids. Only 7% of empty nanodiscs were observed in complex with RNCs (Fig. 5, B and C), suggesting that the truncated variant YidCΔC demonstrated residual affinity to programmed ribosomes. Thus, we conclude that deletions within the C2 loop destabilize the YidC·ribosome complex.
To further investigate the assembly of YidC·RNC complexes, a double truncated YidC variant (YidCΔ8) was cloned and expressed in which both the region 485–489 of C2 and the C terminus were removed (Figs. 6A and and5,5, D and E). The nanodisc-reconstituted YidCΔ8 variant demonstrated a weakened interaction with RNC-F0c, although the interaction appeared to be higher than for the YidCΔC variant (Fig. 5C). YidCΔ8 was further subjected to activity tests via the growth complementation and F0c membrane insertion assays. Although no effect was observed for the individual deletions YidCΔ6 and YidCΔC (Fig. 1), the combination of these truncations within YidC only weakly suppressed the growth deficiency of the YidC depletion strain both under aerobic (Fig. 6B) and anaerobic conditions (data not shown). To analyze the membrane insertion of F0c, YidCΔ8-containing proteoliposomes were introduced into the in vitro transcription/translation reaction system, and insertion was tested. Remarkably, the double deletion in YidC had a detrimental impact on its insertion activity in vitro. Upon removal of the region 485–489 and the C terminus, YidC was not capable of stimulating the membrane insertion of F0c (Fig. 6C), whereas single truncated YidC variants were fully functional in the assay.
To further verify the inhibitory effect of mutations on the YidC activity, membrane insertion of another YidC exclusive substrate MscL was explored. Similarly to F0c, MscL is a double-spanning membrane protein (Fig. 6D), but with both the N terminus and the long C terminus protruding into the cytoplasmic environment and requiring SRP for membrane targeting and insertion (9, 47). After treatment with proteinase K, the C-terminal region of membrane-inserted MscL was largely digested. The amount of the truncated N-terminal fragment MscL(ΔC) was visualized by phosphor imaging and quantified to assess the insertion efficiency (Fig. 6, D and E). The wild-type YidC and the YidCΔ1 variant bearing the N-terminal deletion within the C1 loop were equally functional with respect to MscL insertion, suggesting that the protein is capable of interacting with SRP. In contrast, YidCΔ8 was totally inactive in stimulating the membrane insertion of MscL, whereas full activity was observed for the individual deletions YidCΔ6 and YidCΔC (Fig. 6, D and E). The data suggest that loop C2 together with the C terminus are essential for stabilizing the functional interaction of YidC with translating ribosomes, thereby allowing for the membrane insertion, and these domains may cooperate in maintaining the YidC activity. It is likely that the double deletion within YidC perturbed its structure, which stimulated recruitment of ribosomes as observed in FCS experiments without forming a functional YidC·RNC complex.
In E. coli, YidC functions either independently or in cooperation with SecYEG translocon. In the SecYEG-mediated co-translational pathway, YidC is suggested to localize in the vicinity of the lateral gate of SecY to facilitate the release of the TMSs from the SecYEG channel into the lipids bilayers and to chaperone the folding and/or assembly (44, 48). To further examine whether YidCΔ8 is still functional in the SecYEG-YidC insertion pathway, the membrane insertion of NuoK was explored. NuoK is a small membrane subunit K of the NADH dehydrogenase (Fig. 7A), requiring both SecYEG and YidC for efficient membrane insertion (13). To study the in vitro membrane insertion of NuoK, YidC together with SecYEG were integrated into liposomes to obtain SecYEG-YidC-containing proteoliposomes (Fig. 7B). Subsequently, YidC-SecYEG proteoliposomes were introduced into the NuoK synthesis system, and the amount of proteinase-protected NuoK fragments (13) was analyzed and quantified as described (Fig. 7C). A decrease in the amount of membrane-inserted NuoK was observed for the YidCΔ8 mutant, down to ~70% of the wild-type YidC. With YidCΔ1 and YidCΔ6, full functionality was retained with respect to the membrane insertion of NuoK. This suggests that the cytosolic loop 2 and the C terminus are needed for the efficient insertion of NuoK that is YidC- and Sec-dependent.
Despite recent structural insights, the exact molecular mechanisms of YidC-mediated membrane protein biogenesis remain unknown. YidC has been described to participate in a range of cellular reactions, including ribosome and SRP binding, Sec translocon interactions, and insertion, folding, and oligomerization of its substrates (1). Hence, a comprehensive methodological approach is required to address particular aspects of the insertase functionality. Previous studies using the C-terminal deletion suggested that this domain is involved in ribosome binding (24), as has also been shown for the YidC homologues in Gram-positive and Gram-negative bacteria (24,–26). However, with the E. coli YidC, a substantial residual binding of translating ribosomes was observed, and the truncated YidC variant was fully active in insertion and could rescue growth of YidC-depleted cells. Thus, the C terminus of E. coli YidC was thought not to be the only determinant for communication with the ribosome, and the positively charged cytosolic loops were suggested as potential binding partners (24, 27). In this study, we applied both in vivo and in vitro analyses to investigate the roles played by the cytosolic domains of E. coli YidC in its functioning. We could demonstrate that the cytosolic loop C2 and the C-terminal end of YidC determine ribosome binding and the insertion activity, whereas the C1 loop is essential for cell viability and may be involved in downstream protein biogenesis.
Crystal structures of YidC proteins from B. halodurans and E. coli (20, 40), as well as co-evolution based modeling (27), demonstrate that the large cytosolic loop C1 is folded into a helical hairpin that protrudes from the membrane plane with a tilt determined by an N-terminal proline residue (position 371 within E. coli YidC). Point mutations Y370A and Y377A at the N-terminal end of the loop inactivate YidC in vivo (27), whereas mutations, but not deletions, within the hairpin did not interfere with YidC functionality. This domain forms a potential interface for ribosome binding. Previous studies provided conflicting views on its role. Deletion of the entire C1 domain (positions 371–416) inactivated YidC for growth complementation and membrane insertion of Pf3 and M13 procoat proteins (41), whereas in another study, the C1 region was found to be dispensable for YidC activity (42). We observed that the YidCΔ1 mutant bearing an N-terminal deletion of the helical hairpin was severely compromised in growth complementation both under aerobic and anaerobic conditions. Because YidCΔ1 was steadily expressed, the phenotype was not caused by the in vivo instabilities and suggested an essential role of the region 374–383 within the C1 domain of YidC in cell viability, in accordance with the recent data (27). However, neither ribosome binding nor membrane protein insertion was affected, as tested in vitro for the YidC-only and the SecYEG-associated pathways. Other deletions within the C1 loop had moderate effects on cell viability but also did not reveal defects in ribosome interactions or membrane insertion of F0c and MscL. Thus, there is a low correlation between the YidC insertase activity and in vivo functionality for this domain. We hypothesize that the C1 region of YidC is involved in the membrane insertion of yet unknown substrates that are essential for cell viability or that it plays an important role in chaperone activities such as correct folding of inserted membrane proteins, substrate oligomerization, or protein-protein interactions at the membrane interface.
The conformation of the C2 loop could not be resolved in crystal structures of YidC, suggesting that it is flexible. The loop is built of ~10 amino acids and contains one conserved charged residue Asp-488. A negatively charged residue is essential in this position for cell viability (27). When introducing short deletions within this loop, we observed fully functional YidC with respect to cell viability and substrate membrane insertion. Although these deletions did not affect the insertase activity of YidC in vitro, they inhibited or slowed down the assembly of YidC·RNC complexes. In contrast to the C-terminal deletion, ribosome interactions could be restored either upon prolonged incubation or by increasing the hydrophobicity of the nascent chain. Thus, the loop C2 ensures the stable docking of a translating ribosome to YidC, whereas modifications within this region lead to enhanced ribosome dissociation. Hence, the YidC·ribosome complexes may dissociate before the initiation of the co-translational membrane insertion, leading to termination of the membrane insertion events.
Remarkably, combining deletions within the C2 loop (region 485–489) and at the C terminus of YidC (YidCΔ8) abolished the functionality of YidC in the membrane insertion of the YidC-only substrates F0c and MscL. It is unlikely that this phenotype arises from the instability of the YidCΔ8 variant because a considerable amount of overexpressed YidCΔ8 could be detected, purified, and reconstituted into the liposomes and nanodiscs. Moreover, using NuoK as a model substrate, we found that YidCΔ8 was also less efficient in the SecYEG-YidC co-translational insertion pathway. The double deletion of the C2 and C terminus likely induces a conformational alteration of YidC that negatively affects insertion.
In conclusion, our findings provide insights into the associations between the insertion activity of YidC and ribosome binding. We demonstrated that the cytosolic loop C2 together with the C terminus of E. coli YidC are critical for the assembly of functional YidC·RNCs complexes at the membrane interface, whereas the loop C1 is involved in other vital functions of YidC.
We thank Stephan Wickles for valuable discussions and for providing the pEM36-F0c plasmid and Anne-Bart Seinen for helpful discussions and contributions on drawing the structure of the YidC variants.
*This work was supported by Chemical Sciences and Earth and Life Sciences that are both subsidized by the Netherlands Foundation for Scientific Research and by the Chinese Scholarship Council. This work was also supported by a European Research Council Advanced Grant (to R. B.), Deutsche Forschungsgemeinschaft Grant SFB 646 B11 (to D. C. L.), and funds from the Excellent Clusters Nanosystems Initative Munich, the Center for Integrated Protein Science Munich, the Ludwig-Maximilians-Universität München via the LMUInnovativ BioImaging Network, and the Center for NanoScience (to D. C. L.).
2The abbreviations used are: