The C-terminal regions of YidC from Rhodopirellula baltica and Oceanicaulis alexandrii bind to ribosomes and partially substitute for SRP receptor function in Escherichia coli


  • The authors declare no conflict of interest.


The marine Gram-negative bacteria Rhodopirellula baltica and Oceanicaulis alexandrii have, in contrast to Escherichia coli, membrane insertases with extended positively charged C-terminal regions similar to the YidC homologues in mitochondria and Gram-positive bacteria. We have found that chimeric forms of E. coli YidC fused to the C-terminal YidC regions from the marine bacteria mediate binding of YidC to ribosomes and therefore may have a functional role for targeting a nascent protein to the membrane. Here, we show in E. coli that an extended C-terminal region of YidC can compensate for a loss of SRP-receptor function in vivo. Furthermore, the enhanced affinity of the ribosome to the chimeric YidC allows the isolation of a ribosome nascent chain complex together with the C-terminally elongated YidC chimera. This complex was visualized at 8.6 Å by cryo-electron microscopy and shows a close contact of the ribosome and a YidC monomer.


Membrane proteins comprise about one third of the bacterial proteome, many of them being involved in essential physiological processes such as the respiratory chain function, uptake of nutrients or sensory mechanisms. To functionally assemble the wide variety of membrane-localized proteins and protein complexes, bacterial cells have evolved several membrane insertion and translocation machineries. For the insertion of inner membrane proteins in eubacteria, the membrane translocase SecYEG and the membrane insertase YidC as well as the SRP (signal recognition particle) targeting system are of central importance. SecYEG, YidC and the two SRP components Ffh and FtsY all are essential proteins. Insertion of integral membrane proteins into the lipid bilayer takes place co-translationally at membrane-bound ribosomes (Luirink and Sinning, 2004; Egea et al., 2005; Halic and Beckmann, 2005). Not fully clear, however, is how SRP recognizes the specific features of nascent chains of hydrophobic proteins and how ribosome binding to the membrane is mediated. Although it is known that the SRP receptor FtsY couples ribosome nascent chains (RNC) to the lipid bilayer surface, their transfer to the Sec translocase or YidC insertase has not yet been elucidated. Interestingly, the membrane binding of RNC can be supported by a SRP-FtsY fusion complex (Braig et al., 2011) suggesting that a membrane tethered SRP is functional. Nevertheless, SRP is involved during early translation of membrane proteins on cytosolic ribosomes presumably to prevent aggregation and to mediate the targeting of the ribosome-nascent-chain complex to the membrane surface via its receptor, FtsY. SRP and its receptor are also key players in protein targeting of secreted and integral membrane proteins to the eukaryotic endoplasmic reticulum (Doudna and Batey, 2004; Grudnik et al., 2009).

In eubacteria, the YidC insertase is a crucial component in the insertion of membrane proteins into the cytoplasmic membrane. It can either work independently inserting a subset of specific substrate proteins or it functions together with the Sec translocon (Dalbey et al., 2011). YidC has homologues in eukaryotic organelles, namely in chloroplasts and mitochondria. These homologues have highly conserved regions when compared with Escherichia coli YidC; however, there are a few major differences (Funes et al., 2011). The organellar proteins Alb3 of chloroplasts and Oxa-1 of mitochondria have only 5 transmembrane (TM) segments compared with 6 in the YidC proteins of most eubacteria and both Alb3 and Oxa-1 have a long C-terminal tail domain with pronounced clusters of positively charged amino acid residues. In contrast, the C-terminal tails of YidC in most Gram-negative bacteria are short and no essential function could yet be ascribed to this domain (Jiang et al., 2003). However, genomic analyses and sequence comparisons show that YidC homologues in many marine and extremophilic eubacteria have C-terminal extensions comparable to the eukaryotic organellar proteins (Kiefer and Kuhn, 2007). Recently, a phylogenomic analysis of the YidC/Oxa/Alb protein family revealed evidence for two lineages of YidC/Oxa/Alb evolution (Zhang et al., 2009). These observations may reflect the different classes of YidC proteins with and without the C-terminal extension, respectively.

In this work we present evidence for a possible (ancestral) function of this extended tail. It is known that in chloroplasts the C-terminal tail of Alb3 mediates binding to the organellar cpSRP54 (Falk and Sinning, 2010). In mitochondria, the Oxa-1 C-terminal extension is essential for targeting the ribosome to the inner membrane (Preuss et al., 2005). To study the possible roles of the YidC C-terminally extensions in bacteria, we analysed the function of the YidC tail regions of two marine prokaryotes in membrane targeting. The marine planctomycete Rhodopirellula baltica was first isolated from the Baltic Sea (Schlesner et al., 2004) as a unique bacterium that can anaerobically oxidize ammonium in specific cellular compartments, the anammoxosomes, and its 7.15 Mbp genome is one of the largest found in bacteria (Glöckner et al., 2003). Oceanicaulis alexandrii is a stalked, aerobic α-proteobacterium, isolated originally as a symbiont from the dinoflagellate Alexandrium tamarense (Strömpl et al., 2003).

To elucidate the function of the C-terminal tails of the marine YidC proteins, we fused these regions to the E. coli YidC protein and analysed them for their ability to bind to the ribosome and target the proteins to the inner membrane. We found a strong and specific binding of the chimeric YidC proteins with extended C-tails to isolated ribosomes in contrast to wild-type E. coli YidC, which did not show any binding to ribosomes in our assays. In addition, the chimera of E. coli YidC with the planctomycete YidC tail accounted for a partial recovery of a ftsY-defect in an E. coli depletion strain. We conclude that the C-terminal tail region in eubacterial YidC homologues has a similar function as proposed for the organellar homologue Oxa-1, namely to mediate membrane targeting of nascent membrane proteins.


Chimera of the membrane insertase YidC with altered C-terminal regions

Sequence comparison of YidC homologues show that the C-terminal regions in some marine bacteria is extended by numerous positively charged amino acid residues similar to the mitochondrial Oxa-1 and YidC2 found in Gram-positive bacteria. In contrast to the Gram-positive genomes, the marine bacteria only have a single yidC gene. The YidC homologue of the planctomycete R. baltica (Rb-YidC) has a size of 90 kD and a hydrophilic C-terminal region of 80 amino acid residues with a predicted isoelectric point (pI) of 10.45. The α-proteobacterium O. alexandrii YidC homologue (Oa-YidC) is 83 kD in size and has a positively charged C-terminal tail of 99 residues with an estimated pI of 9.85. In comparison, E. coli YidC (Ec-YidC) is 61 kD in size and its C-terminal region is only 16 residues long whereas Oxa-1 of Saccharomyces cerevisiae has a C-tail of 89 residues (Fig. 1A). To investigate the role of the extended C-terminal tails, chimeric forms of E. coli YidC were constructed where the C-terminal region of YidC was replaced by the C-terminal tail of R. baltica (YidC-Rb), O. alexandrii (YidC-Oa) and S. cerevisiae (YidC-Oxa), respectively (Fig. 1B).

Figure 1.

Diversity of the C-terminal regions of YidC homologues.

A. C-terminal tails of the E. coli YidC and the YidC homologues of marine bacteria. The sequence after TM segment 6 is shown for E. coli, R. baltica, O. alexandrii and, for comparison, the corresponding sequence of S. cerevisiae. The length and the number of positively charged amino acid residues are indicated. Also, the respective sequence of the chimeric proteins of Ec-YidC with the C-terminal tails of R. baltica (YidC-Rb) of O. alexandrii (YidC-Oa) and of S. cerevisiae (YidC-Oxa) is shown.

B. Schematic representation of the YidC proteins. The core domain of each YidC consists of 6 transmembrane segments and a large P1 domain.

To analyse whether the homologues and the chimera were functional, they were studied in the YidC depletion strain MK6. In this strain, the yidC promoter is replaced by an araC-araBAD promoter on the chromosome. Growth in the presence of glucose for more than 2 h results in the depletion of YidC and finally to cell death, whereas the presence of arabinose allowed normal growth (Fig. S1A and B). When the plate contained 0.4% glucose and 1 mM IPTG (Fig. S1C) the plasmid-derived expression of Ec-YidC, Oa-YidC, YidC-Rb, YidC-Oa and YidC-Oxa promoted growth. This clearly shows that the YidC homologue of O. alexandrii (Oa-YidC) and also the chimeric proteins YidC-Rb, YidC-Oa and YidC-Oxa are fully functional and can complement the wild-type YidC in E. coli, whereas the YidC homologue from R. baltica and S. cerevisiae Oxa-1 cannot. Therefore, only the three chimeras and Oa-YidC homologue were used for our further studies.

Ribosome binding of YidC is improved by the extended C-terminal tails

Since ribosome binding of the C-terminal region of the mitochondrial Oxa-1 protein has been documented (Jia et al., 2003), we were interested to see whether the extended C-tails of YidC homologues from Gram-negative marine bacteria have a similar function. Ribosomes were isolated from E. coli cells using a special chromatography system (Maguire et al., 2008) and incubated with the purified YidC proteins on ice. The ribosome-bound YidC was separated by centrifugation and the pellet and supernatant fractions were analysed by SDS-PAGE (Fig. 2). Whereas YidC-Rb and YidC-Oa readily bound to the 70S ribosomes and consequently were found in the pellet fraction (Fig. 2 and C), YidC from E. coli was found in the supernatant (Fig. 2A). Likewise, Oa-YidC was found associated with the ribosomes in the pellet fraction (Fig. 2D). When no ribosomes were present, virtually all YidC was found in the supernatant. The binding of YidC proteins with the extended C-terminal tails to the ribosome was salt sensitive, indicating that binding is mediated by ionic interactions (Fig. S2). From these data, we conclude that the charged C-terminal extensions are crucial for efficient binding of the ribosome to YidC. We suggest that the C-terminal tails of the marine YidC homologues can function as ribosome binding domains.

Figure 2.

Ribosome binding of YidC by the extended C-tails. The purified YidC proteins were incubated with E. coli ribosomes and separated by centrifugation. Pellet (P) and supernatant (S) were collected and analysed by SDS-PAGE, stained with Coomassie blue (upper panels) or analysed by Western blot using an anti-His antibody (lower panels). Ec-YidC (A), YidC-Rb (B), YidC-Oa (C), and Oa-YidC (D) were analysed. For control, the distribution of the YidC protein was monitored in absence of ribosomes. S1: ribosomal protein S1.

The C-terminal extension of YidC-Rb can mediate the membrane targeting of MscL in the absence of a functional SRP system

The discovery of YidC isoforms with a C-terminal extension in mitochondria (Oxa-1) and Gram-positive bacteria (YidC2) suggested that they play a role in facilitating the cotranslational membrane targeting activity independent of the SRP pathway (Funes et al., 2009). This might also hold true for the C-terminal extensions of the marine YidC homologues of Gram-negative bacteria. The mechanosensitive channel protein MscL is targeted to the membrane of E. coli by SRP and is then inserted by YidC into the inner membrane (Facey et al., 2007). To investigate whether an extended C-terminal region of YidC can compensate for a loss of SRP function, the FtsY depletion strain IY26 and the Ffh depletion strain MC-ffh were transformed with pGZ119HE-N-his encoding the YidC proteins and a second plasmid pSF147 encoding the MscL protein. The cells were grown in glucose media to deplete the SRP receptor protein FtsY or the Ffh protein component of SRP, respectively. Membrane insertion of MscL was analysed by chemical modification with AMS, a membrane-impermeable sulfhydryl-reagent that shifts the protein mobility on SDS-polyacrylamide gels. Addition of AMS during the pulse chase experiments leads to a derivatization of the translocated cysteine-residue at position 68 of MscL and results in a shift of 0.5 kD of MscL, indicating its membrane insertion (Facey et al., 2007).

We tested whether the expression of Oa-YidC, YidC-Rb, YidC-Oa or YidC-Oxa can compensate for a loss of Ffh in the membrane targeting of MscL. Therefore, MC-ffh cells coexpressing the MscL-68C single cysteine mutant and the respective YidC variant were pulse-labelled with [35S]-Met, chased with non-radioactive methionine and then incubated with AMS. After treatment with AMS, proteins were acid-precipitated and immunoprecipitated. Derivatized and underivatized proteins were separated by SDS-PAGE and examined by phosphorimaging. In the absence of Ffh, AMS derivatization of MscL is reduced by about 40% (Fig. 3A and C). This corroborates that the insertion process of MscL is affected by Ffh depletion (Facey et al., 2007). Coexpression of the YidC proteins did not significantly restore membrane insertion of MscL under Ffh depletion conditions (Fig. 3A), in contrast to the cells where Ffh was coexpressed from a plasmid (lane 14, Fig. 3C).

Figure 3.

MscL is inserted into the membrane in the absence of the SRP receptor FtsY when YidC-Rb or YidC-Oxa is coexpressed. The membrane insertion of MscL was monitored by the AMS reactivity of a cysteine residue (C68) in the periplasmic loop of MscL.

A. Strain MC-ffh expressing the MscL cysteine mutant was grown in M9 minimal medium containing arabinose (cFfh+, lane 1). For depletion of Ffh (cFfh−, lane 2), the cells were grown in the presence of glucose. IPTG (1 mM) was added for 10 min to induce expression. Cells were incubated in the presence of AMS, pulse-labelled with [35S] Met for 2 min and chased with non-radioactive Met for 10 min. After quenching with 20 mM DTT, the radiolabelled samples were acid-precipitated and immuno-precipitated with an anti-His antibody and then subjected to SDS-PAGE and phosphorimaging. The arrowhead D denotes the AMS-derivatized protein. Parallel experiments were performed with cells coexpressing Ec-YidC (lanes 3 and 4), Oa-YidC (lanes 5 and 6), YidC-Rb (lanes 7 and 8), YidC-Oa (lanes 9 and 10) or YidC-Oxa (lanes 11 and 12). As a control, the derivatization of MscL was monitored with MC-ffh cells coexpressing plasmid-encoded Ffh (pFfh) in the absence of chromosomal Ffh (cFfh) (lane 14).

B. The FtsY depletion strain IY26 bearing the MscL cysteine mutant was grown under either FtsY depletion (lane 2) or FtsY expression conditions (lane 1). Samples were prepared and processed as described for MC-ffh. Parallel experiments were performed with cells coexpressing Ec-YidC (lanes 3 and 4), Oa-YidC (lanes 5 and 6), YidC-Rb (lanes 7 and 8), YidC-Oa (lanes 9 and 10) and YidC-Oxa (lanes 11 and 12). The lower panels show immunoblots of the Ffh or FtsY level under arabinose (+) and glucose (−) conditions, respectively.

C. The AMS-shift assays were quantified with ImageJ. In each experiment the amount of derivatized MscL in the presence of Ffh or FtsY was set to 100% (control). The relative amount of derivatized MscL in the absence of Ffh (red columns) or FtsY (blue columns) was calculated after coexpression of the indicated YidC protein from plasmid, respectively. Cells without coexpression are shown in lane (−). For several repeat experiments the standard deviations (bars) were calculated.

Next, we asked whether the C-terminal extensions of the YidC proteins could replace the function of the SRP receptor protein FtsY. To assess the ability of YidC-Rb, YidC-Oa or YidC-Oxa to mediate membrane targeting and insertion of MscL in the absence of FtsY, the same experiment, as described above for Ffh, was performed in the FtsY depletion strain IY26. When FtsY was present, about 50% of the MscL protein was shifted during the pulse time indicating that it was correctly inserted into the membrane (Fig. 3B, lane 1). However, when the cells were grown in the absence of arabinose to deplete FtsY, a lower proportion of MscL was derivatized by AMS (lane 2), showing that the targeting of MscL to the membrane and the insertion process is inhibited, somewhat stronger than after Ffh depletion. To investigate whether coexpression of the YidC proteins can compensate for the loss of FtsY and promote membrane insertion of MscL, a parallel experiment was performed with IY26 cells that express MscL together with Ec-YidC, Oa-YidC, YidC-Rb, YidC-Oa or YidC-Oxa, respectively (Fig. 3B and C). In the absence of FtsY, coexpression of the YidC-Rb and YidC-Oxa protein with the C-terminal extensions allowed substantial AMS derivatization of MscL (lanes 8 and 12). In contrast, the coexpression of YidC-Oa or Oa-YidC showed no significant improvement of the membrane insertion of MscL in the absence of FtsY (lanes 6 and 10). To exclude that the overexpression of a functional YidC protein causes an enhanced MscL insertion under FtsY depletion conditions, we coexpressed wild type E.coli YidC (Ec-YidC) and observed no improved MscL derivatization (Fig. 3C). Taken together, these data indicate that the membrane targeting and the subsequent insertion of MscL in the absence of FtsY are mediated by the C-terminal domain of the marine YidC homologue of R. baltica.

Localization of MscL-GFP in the absence of Ffh or FtsY

To visualize the cellular localization of MscL, the protein was expressed as a MscL-GFP fusion (Maier et al., 2008). We analysed the localization of MscL-GFP in the FtsY depletion strain IY26 and in MC-ffh cells. The cells were grown in arabinose to allow normal membrane targeting or in glucose to deplete FtsY or Ffh, respectively. The YidC proteins were coexpressed and the cells were then inspected for the localization of MscL-GFP by fluorescence microscopy. In the presence of Ffh and FtsY, the fluorescence was evenly distributed at the membrane surface (Figs 4A and 5A, respectively). However, when the cells were grown in the presence of glucose to deplete Ffh, the MscL-GFP aggregated at the cell poles (Fig. 4B). Under FtsY-depleted conditions, the MscL-GFP protein was found in patches mostly at the cell poles (Fig. 5B). This is similar to the phenotype when YidC is depleted (Fig. 4C) in MK6 cells. Remarkably, coexpression of YidC-Rb in the absence of Ffh (Fig. 4D) restored the mislocalization of MscL-GFP and showed a distribution of the fluorescent signal similar to the coexpression of wild type Ffh encoded on a plasmid (Fig. 4F). The absence of FtsY in IY26 cells coexpressing YidC-Rb (Fig. 5C) allowed a cellular distribution of MscL-GFP similar to wild type conditions (Fig. 5A). These results, combined with the AMS derivatization assay indicate that YidC-Rb allows an improved membrane targeting of MscL in the absence of Ffh or FtsY whereas the correct insertion of MscL still requires Ffh.

Figure 4.

Localization of MscL-GFP in the absence of Ffh. Localization of MscL-GFP was studied in vivo by fluorescence microscopy. MC-ffh cells bearing the MscL-GFP fusion plasmid were grown in LB medium either in the presence of arabinose (A) or in the presence of glucose to deplete Ffh (B, D, E and F). Cells were induced with 1 mM IPTG for 2 h at 30°C and inspected by fluorescence microscopy. The coexpression of YidC-Rb under Ffh depletion conditions (D) showed a distribution of the fluorescent signal similar to the plasmid-encoded coexpression of wild type Ffh (F), whereas the coexpression of Ec-YidC (E) showed more MscL-GFP patches, but also a different cell localization compared with the single MscL-plasmid expressing Ffh depleted cells (B).

C. Localization of MscL-GFP in YidC depleted MK6 cells.

Figure 5.

YidC-Rb restores the mislocalization of MscL-GFP in the absence of FtsY. Localization of MscL-GFP in E. coli strain IY26 in the presence (A) or absence (B–D) of FtsY in vivo by fluorescence microscopy as described for MC-ffh. Mislocalization of MscL-GFP in the absence of FtsY (B) is prevented by coexpressing YidC-Rb (C), but not by Ec-YidC (D).

Nascent ribosome-linked chains of MscL interact with YidC-Rb

Since our results show increased ribosome affinity of YidC derivatives with an extended hydrophilic C-terminus, we performed a structural analysis of YidC-ribosome complexes. Using cryo-EM we determined the 3D structure of a complex consisting of a YidC-Rb bound to a translating ribosome carrying the MscL protein with its first two TM segments as a nascent chain stalled by an introduced TnaC sequence (Fig. 6). A reconstruction of the complex could be refined to 8.6 Å resolution after semi-supervised classification using competitive projection matching (Fig. S3). The TnaC-stalled ribosome has a strong density for tRNA located in the P-site indicating a high percentage of programmed RNCs in the final dataset. The path of the nascent chain can be traced from the peptidyl transferase centre (PTC) through the ribosomal tunnel into extra density at the tunnel exit representing YidC-Rb (Fig. 7). The position of this extra density agreed well with density observed in an earlier study of ribosome-bound E. coli YidC (Kohler et al., 2009). Moreover, we observed a similar interaction pattern between our C-terminally extended YidC-Rb and the ribosome. Ribosomal rRNA helix H59 and the ribosomal protein L24 appear to be the main contact sites with both types of YidC. Thus, the presence of the C-terminal extension increases the affinity of YidC to the ribosomes without changing its overall interaction mode. An additional contact to the ribosomal protein L29 was observed for the bound YidC-Rb protein. Notably, the presence and size of the detergent micelle surrounding the solubilized YidC-Rb was difficult to estimate at this resolution. Also, the large periplasmic domain of YidC-Rb is not resolved in the cryo-EM structures probably due to its dynamic motions relative to the membrane embedded domain. The size of the density indicates the presence of only one copy of YidC-Rb under our conditions (Fig. 6). Even when assuming the presence of a minimal micelle, a YidC homodimer, as had been suggested in a previous study (Kohler et al., 2009) could not be accommodated. This suggests that a single copy of YidC is stably bound to the ribosome and may be sufficient to act as a ribosome-bound insertase for MscL as also proposed by Kedrov et al. (2013).

Figure 6.

Cryo-EM reconstruction of an RNC-YidC-Rb complex.

A. The 30S and the 50S ribosomal subunits are shown in yellow and grey, respectively; tRNA is shown in green and additional density accounting for YidC-Rb in red.

B. Cut through the density to view the ribosomal tunnel and fragmented nascent chain density in green.

C and E. Close up side view with the fitted molecular model of a 70S ribosome. L24 and L29 are shown in blue. YidC-Rb shows defined contacts to the ribosomal proteins L24, L29 and the helix H59.

D. Close up bottom view of the interaction area with a transparent density for YidC-Rb and the nascent MscL chain (green) surrounded by the YidC-Rb density.

Figure 7.

3D structure of the isolated density for the nascent chain – YidC-Rb complex at 8.6 Å.

A. Bottom view of the isolated YidC-Rb density. Density blot of YidC-Rb shows entering of the nascent MscL chain (green) in the centre of the YidC-Rb density (red). The asterisks mark the position of the ribosomal protein L29.

B. Side view of the translating ribosome carrying the MscL protein with its first 2 TM segments as a nascent chain stalled with an introduced TnaC sequence.

The ribosomal protein L29 binds to the C-terminal tail of YidC-Rb

To investigate whether the observed contact between YidC-Rb and L29 is caused by the extended C-tail of the protein, Streptavidin-tagged versions of L24 and L29 were purified and their interaction with YidC-Rb was analysed by pull-down experiments (Fig. 8). The purified ribosomal proteins were loaded on agarose matrix-immobilized YidC-RbNHis and YidCNHis, respectively. L24 and L29 co-eluted from the resin with the His-tagged YidC-Rb protein (lane 4), demonstrating that both ribosomal proteins physically interact with YidC-Rb. In contrast, L29 was unable to bind and therefore did not co-elute from the resin together with the His-tagged wild type E. coli YidC (lane 8), whereas L24 co-eluted in similar amounts as for YidC-Rb. These results show that the enhanced affinity of YidC-Rb to the ribosome is at least partially caused by the interaction of the C-terminal domain of the R. baltica YidC homologue and the ribosomal protein L29.

Figure 8.

Interaction of the ribosomal protein L24 and L29 with YidC-RbNHis and Ec-YidCNHis. Pull Down assays of L24 (upper panels) and L29 (lower panels) were performed with YidC-Rb (lane 1–4) and Ec-YidC (lane 5–8). About 120 μg of the respective YidC proteins was attached to NiNTA resin. Purified L24Strep and L29Strep (∼ 500 μg) proteins were loaded onto the immobilized YidC proteins. Lanes 1 and 5 show the total amount of added L24 and L29, respectively. Unbound protein was collected by gravity flow (lane 2 and 6). After three washes (lanes 3 and 7 show the last washing steps), bound proteins were eluted with 500 mM imidazole by centrifugation (lane 4 and 8). The proteins were acid-precipitated and analysed by SDS-PAGE and Coomassie staining.


This study shows that the extended C-terminal tail of YidC found in the marine Gram-negative bacterium R. baltica can partially substitute for the SRP receptor protein FtsY, in contrast to the native E. coli YidC. This is in accordance with recent data that show a functional overlap of the SRP-machinery and the elongated C-terminal tails of YidC/Oxa/Alb family members in co-translational protein insertion. The YidC homologue Oxa-1 of Saccharomyces cerevisiae binds to the large subunit of the mitochondrial ribosome close to the protein exit tunnel (Jia et al., 2003; Gruschke et al., 2010) and contributes to efficient co-translational membrane insertion of Cox2 and cytochrome bc1 (Szyrach et al., 2003). Since mitochondria lack the SRP system for membrane targeting, a direct affinity of the ribosome to Oxa-1 might be required for co-translational membrane insertion. In addition, a C-terminally truncated Oxa-1 mutant was restored by the coexpression of the bacterial Ffh protein in mitochondria (Funes et al., 2013). Consistent with this idea, it has been shown for the Gram-positive bacterium Streptococcus mutans, that yidC2 can only be deleted when SRP is present (Hasona et al., 2005). Since S. mutans has two YidC paralogues, YidC1 with a short C-terminal tail and YidC2 with an extended tail, it was speculated that the extended tail of YidC2 functions for ribosome binding and promotes co-translational membrane insertion (Funes et al., 2009). Here we present evidence that the functional overlap of the SRP system and the elongated YidC tails also exists in Gram-negative bacteria. We tested two C-terminal extensions from marine Gram-negative YidC homologues for their ability to function as a ribosome binding domain. The results verify that Ec-YidC binds very poorly to ribosomes in contrast to YidC with an extended C-tail (Kedrov et al., 2013). Both chimeras with the hydrophilic C-tails of the marine YidC homologues fused to the E. coli YidC moiety (YidC-Rb and YidC-Oa of R. baltica and O. alexandrii, respectively) showed efficient binding to ribosomes (Fig. 2).

Since our in vitro studies clearly showed a YidC-Rb-ribosome interaction, we wanted to test a possible role of the C-terminal YidC extensions in co-translational protein targeting and insertion in vivo. The C-terminal extension could act as an adaptor for a specialized function during the co-translational insertion of the YidC pathway. We observed that the insertion of the YidC substrate protein MscL in E. coli depends on the R. baltica tail of YidC when FtsY is depleted as shown by the AMS gel shift assay (Fig. 3B). Similarly, coexpression of YidC-Oxa resulted in an enhanced membrane insertion efficiency. Expression of the SRP-dependent protein MscL as a GFP fusion protein showed that it accumulated as patches mostly close to the cell poles when the receptor protein FtsY was depleted (Fig. 5B). Coexpression of YidC-Rb prevented the formation of these MscL-GFP aggregates and the fluorescence was found at the membrane (Fig. 5C) whereas coexpression of Ec-YidC (Fig. 5D) had no effect on the localization of MscL-GFP, indicating that the C-terminal tail of YidC-Rb can restore the membrane targeting of MscL in the absence of FtsY. It should be kept in mind that Ffh is still present and possibly required under these conditions.

When Ffh was depleted, the MscL-GFP clearly aggregated at the cell poles (Fig. 4B). Coexpression of YidC-Rb only partially restored membrane targeting and the MscL-GFP protein was found in patches but still localized near the cell poles. Membrane insertion of MscL assayed by the AMS gel shift assay was only marginally stimulated in the Ffh-depleted cells coexpressing YidC-Rb or YidC-Oxa (Fig. 3A, lanes 8 and 12). In conclusion, the extended C-tails of YidC can replace mainly the FtsY receptor function and not the Ffh function. We assume that the presence of Ffh might still be required to keep the newly synthesized MscL protein in an insertion-competent form. Without a functional SRP-system the hydrophobic non-inserted MscL protein most likely aggregates in the cytoplasm and is then prone to degradation, which occurs at the cell poles.

The improved binding of YidC-Rb to ribosomes enabled us to reconstitute the RNC-YidC-Rb complex in vitro at physiological pH. Using cryo-EM we could reconstruct the complex which shows YidC-Rb binding at the ribosomal tunnel exit (Fig. 6). Although our overall resolution was in the subnanometer range, alpha-helical secondary structure was barely resolved in the YidC density, most probably due to some flexibility within this region (Fig. S3). However, we could resolve the contact sites of YidC-Rb to the ribosome: The rRNA helix H59 shows the strongest contact followed by two ribosomal proteins L24 and L29. These contacts and an additional contact to L23 had been proposed from a lower resolution structure of the E. coli YidC-RNC with the FoC protein (Kohler et al., 2009). For YidC-Rb we observed a relatively weak connecting density with L23. The additional contact to L29 was confirmed by pull down assays (Fig. 8) with YidC-Rb and Ec-YidC, where L29 only co-eluted with YidC-Rb. Conclusively, the C-terminal tail of YidC-Rb faces towards the L29 moiety of the ribosome.

Recent studies on substrate-insertase interactions (Klenner and Kuhn, 2012; Neugebauer et al., 2012) have shown that multiple TM regions of YidC are involved in direct substrate contacts. A three-dimensional model of the hydrophobic YidC protein binding platform illustrates that the helices of the substrate protein are surrounded by the transmembrane regions of a YidC monomer (Klenner and Kuhn, 2012). A monomeric YidC binding the substrate protein MscL is consistent with our cryo-EM structure (Figs 6 and 7).

On the basis of the current structural and biochemical data, we propose for the FtsY depleted cells a sequential targeting and insertion mechanism of MscL involving the SRP, the nascent ribosome and the C-terminally extended domain of YidC (Fig. 9). The specific interacting target of SRP on the ribosome seems to be flexible and dependent on the conformational state of SRP. First, L23 offers a transient binding site for the N domain of Ffh while the M domain of the SRP binds the nascent chain. Upon recognition of an uncleavable signal sequence, the SRP binds with high affinity to the RNC, is repositioned and the contact between SRP and L29 is lost (Estrozi et al., 2011). We propose that, in this state, the C-terminal YidC-Rb tail can bind to the ribosome and the nascent chain is directly and co-translationally inserted in the absence of FtsY. Further studies are required to explicitly analyse at which stage SRP is displaced from the ribosome and when the transfer of the RNC to the YidC insertase occurs.

Figure 9.

Cartoon Model of co-translational targeting and insertion via a C-terminally extended YidC in the absence of the SRP-receptor FtsY. SRP binds the RNC complex and prevents aggregation and misfolding of the nascent chain. The complex is positioned on the YidC insertase by its C-terminal extension due to the interaction with L29. This results in a co-translational membrane insertion of MscL, similar to the SRP-FtsY pathway.

Another important question is why some marine bacteria have C-terminal extensions on their YidC homologues. R. baltica and O. alexandrii both have Ffh and FtsY genes on their chromosome, which most likely operate similar to E. coli. If YidC has a higher affinity to ribosomes, more ribosomes should be firmly bound to the membrane surface. This could be an advantage for cells that synthesize a large number of YidC-dependent membrane proteins. Another explanation is that the C-terminal extended tail of YidC in marine bacteria may be an ancestral remnant of a primordial translocation system operating without a SRP receptor. Possibly, the phylogenetically ancient function of the SRP was mainly chaperone-like rather than having a targeting function. The targeting function may first have been acquired by an ancestral YidC protein with an extended C-terminal tail mediating membrane binding of the substrates. Later on in evolution the SRP receptor was more efficient in membrane targeting and also allowed interaction with the Sec translocase. At this evolutionary stage, the C-terminal extension of the YidC proteins were dispensable. Marine prokaryotic organisms are, in many cases, unique in their genomic organization (Serres et al., 2009). Particularly in the planctomycete group, several gene duplications and protein motifs are found that only have counterparts in the archaeal or eukaryotic phyla (Studholme et al., 2004). Thus, marine bacterial groups may be a key to understand how complex cellular processes like protein translocation and membrane targeting may have evolved and spread over the whole organismic world.

Experimental procedures

Bacterial strains and plasmids

YidC depletion was carried out using E. coli strain MK6 (Klenner et al., 2008). 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 (Lessl et al., 1992) expressing the respective YidC proteins under the control of the tac promoter. For overexpression and purification of the YidC proteins from E. coli, R. baltica and O. alexandrii N-terminal decahistidine variants cloned into pET16b (Novagen, Darmstadt, Germany) were used in E. coli C43 (DE3). For the AMS derivatization studies, the I68C MscL mutant (Facey et al., 2007) was expressed by induction with IPTG from the vector pMS119 (Samuelson et al., 2001). The YidC proteins, containing an N-terminal His10tag, were coexpressed from the vector pGZ119HE in strains MC-ffh and IY26, respectively. The FtsY depletion strain IY26 (BW25113-Kan-AraCP-ftsY) was obtained from E. Bibi. FtsY is under the control of the araBAD promoter and operator. The Ffh depletion strain MC-ffh (MC1061-Kan-AraCP-ffh) is from our collection and was derived from E. coli strain MC1061 (Casadaban and Cohen, 1980) by homologous recombination of the araC-araBAD operator region together with a kanamycin cassette as a selective marker into the chromosome upstream of the ffh gene (Datsenko and Wanner, 2000).

Construction of YidC chimera

Genomic DNA from R. baltica and O. alexandrii was used as a template to amplify the homologous yidC genes with flanking Nde restriction sites. Two chimeric proteins of the E. coli YidC were constructed with the C-terminal region of YidC replaced. We first introduced an Nsi restriction site at amino acid position 539 of the E. coli YidC using site-directed mutagenesis. The C-terminal extensions of both marine YidC homologues were amplified with flanking Nsi sites. The resulting PCR products were digested with Nsi and cloned into the E. coli yidC resulting in YidC-Rb (E. coli YidC 1–540 and R. baltica YidC 748–827), YidC-Oa (E. coli YidC 1-540 – O. alexandrii YidC 575-673) and YidC-Oxa (E. coli YidC 1-540 – S. cerevisiae Oxa-1 272-360).

Complementation of E. coli YidC

Escherichia coli MK6 cells bearing the respective pGZ119EH plasmids expressing YidC were grown in Luria broth (LB) medium with 0.2% arabinose and 0.4% glucose to an OD600 of 1.0. Cells were washed once with LB, serially diluted and spotted on LB plates containing 0.2% arabinose or 0.4% glucose, respectively. In the presence of 1 mM IPTG, the plasmid-encoded YidC proteins were expressed. The plates were incubated overnight at 37°C.

Recombinant protein expression

Conditions for overexpressing and purification of the YidC proteins from E. coli, O. alexandrii and the chimeric proteins YidC-Rb and YidC-Oa in E. coli C43 (DE3) were similar for all pET16b constructs. Cells were grown in 4 L LB containing 100 μg ml−1 ampicillin at 37°C and induced with 1 mM IPTG at an OD600 of 0.6. The induced culture was grown for another 2 h at 37°C. The cells were pelleted, resuspended and lysed using a French pressure cell (8000 Ibs in−2). The lysate was centrifuged for 10 min at 8000 g to spin down the unbroken cells and the membranes were pelleted by centrifugation for 1 h at 140 000 g. Membranes containing the respective YidC protein were extracted with 1% n-dodecyl β-D-maltoside (DDM; w/v) overnight at 4°C. The protein was purified by Ni chelating chromatography and isolated in elution buffer (50 mM NaOAc2 pH 5, 150 mM NaCl, 0.1% DDM, 400 mM imidazole, 1 mM DTT). After elution the main fractions were dialysed against elution buffer without imidazole.

In vitro ribosome-binding assay

The ribosomes were isolated from a cell extract of the E. coli strain MRE600 according to Maguire et al. (2008). The isolated ribosomes and the purified YidC proteins were incubated on ice for 1 h (100 nM ribosomes; 500 nM protein) in ribosome-binding buffer (20 mM Tris-HCl pH 7.4, 25 mM MgOAc2, 100 mM NaCl, 0.1% DDM, 1 mM DTT), with a total reaction volume of 500 μl and then centrifuged for 3 h at 200 000 g. The pellet was directly resuspended in SDS loading buffer, whereas the supernatant was precipitated with 10% trichloroacetic acid (TCA) overnight and then subjected to SDS-PAGE analysis and subsequent Coomassie staining or immunoblotting and detection with antibodies against the His10tag. For quantification, the total amount of added protein (input) was loaded on the gel.

Pulse chase and immuno-precipitation analysis

For the AMS derivatization studies, E.coli MC-ffh (Ffh depletion) and IY26 (FtsY depletion) were transformed with pMS-MscLI68C and pGZ-YidC-RbN-His, pGZ-YidC-OaN-His, pGZ-YidC-OxaN-His, pGZ-Oa-YidCN-His or pGZ-Ec-YidCN-His, respectively. The MscL mutant and the YidC proteins were coexpressed by induction with IPTG. To deplete cells of Ffh or FtsY, the cells of overnight cultures were washed twice with LB to remove the arabinose and back diluted 1/150 into fresh LB medium containing 0.4% glucose (w/v). Cells were grown for at least 3 h in glucose. The cells were transferred to M9 minimal medium for 60 min prior to pulse chase labelling. Cells coexpressing the MscL mutant and the YidC proteins were induced for 10 min with 1 mM IPTG. Cells were radiolabelled with [35S] Met (10 μCi ml−1 culture) for 2 min. After labelling, cells were incubated with 4-acetamido-4′-maleimidylstilbene-2,2′-disulfonic acid (AMS; 2.5 mM final concentration; Molecular Probes) and non-radioactive L-methionine was added (final concentration, 500 μg ml−1) for 10 min and then combined with 20 mM DTT for 10 min to quench the AMS reaction. After quenching, samples were acid-precipitated, resuspended in 10 mM Tris, 2% SDS and immuno-precipitated with anti-His antibodies (Sigma-Aldrich). The samples were then analysed in 14% SDS-polyacrylamide gels, examined by phosphorimaging and quantified with ImageJ (Schneider et al., 2012).

Fluorescence microscopy

For the localization studies a MscL-GFP fusion protein was expressed from a pMS119 derivative. The strains IY26 and MC-ffh were transformed with the MscL-GFP encoding plasmid and pGZ119HE derivatives encoding the YidC homologues or YidC chimera, respectively. Strains were grown overnight at 37°C in LB medium with 0.2% arabinose and 0.4% glucose, washed twice with LB and diluted 1/150 in fresh LB medium with arabinose (0.2%) to allow normal membrane insertion or in glucose (0.4%) to deplete FtsY or Ffh, respectively. Cells were grown at least 3 h to an OD600 of 0.6, and protein expression was then induced with 1 mM IPTG for 2 h at 30°C. Cells were collected by centrifugation, washed twice with LB medium and resuspended in 2 mM EDTA, 50 mM Tris-HCl pH 8.0 and incubated overnight at 4°C. The cell suspension (corresponding to 50 μl cells) was applied to a polylysine-coated cover glass (Sigma-Aldrich) and examined by a Zeiss AXIO Imager M1 fluorescence microscope. Emission was detected with a filter set specific for GFP.

Reconstitution of the RNC-YidC complex

Ribosome nascent chain complexes (RNCs) were purified as described (Seidelt et al., 2009). The translation system was programmed with an mRNA coding for an N-terminal His-tag followed by amino acid residues 1–115 of MscL and the TnaC stalling sequence at the C-terminus. This results in the first two transmembrane helices of MscL being fully emerged from the ribosomal exit tunnel. For cryo-EM analysis RNCs were reconstituted with purified YidC-Rb for 30 min at 37°C.

Electron microscopy and image processing

The reconstituted RNC-YidC complex was vitrified on 2 mm pre-coated Quantifoil R3/3 holey carbon supported grids using a Vitrobot Mark IV (FEI Company). For automated data collection on the Titan Krios TEM (FEI Company) under low dose conditions (∼ 20 e A−2) the magnification was set to nominal 75 000× with a defocus range between −1 μm and −3.5 μm was used. The microscope was operated at 200 KeV and a magnification of 148 721× at the plane of CCD using a 4k × 4k TemCam-F416 CMOS camera (TVIPS GmbH) resulting in an image pixel size of 1.12 A (object scale).

A total of 14 165 micrographs were collected of which 4488 were selected manually for further processing based on information content of the power spectra and particle density on the grid. The data processing was performed using SPIDER software package (Frank et al., 1996). The defocus of each micrograph was determined using the TF ED command in SPIDER and particles were automatically selected using SIGNATURE (Chen and Grigorieff, 2007). The complete data set of 140 266 particles was aligned to an empty 70S ribosome that was generated using the crystal structure of an E. coli ribosome (Schuwirth et al., 2005). Using semi-supervised classification (Penczek et al., 2006) it was possible to sort for subpopulations showing distinct ribosomal conformations and ligands (+/− E-site tRNA, +/− Ligand at tunnel exit). A final dataset of 51 903 particles resulted in a density map refined to 8.6 Å resolution according to Fourier Shell Correlation (FSC at 0.5 cut-off) showing P-site tRNA and high occupancy of YidC at the ribosomal exit site.

Pull-down assays

One hundred and twenty micrograms of purified YidCNHis and YidC-RbNHis protein was attached to 50 μl bed volume NiNTA resin, respectively. Immobilization of the YidC proteins was performed in 1 ml binding buffer containing 50 mM Tris pH 7, 150 mM NaCl, 1 mM DTT and 0.05% DDM (YidC) or 0.1% LDAO (YidC-Rb) for 2 h at 4°C. Unbound protein was collected by gravity flow. L24 and L29 were purified using N-terminal Strep-tag fusion proteins and eluted in buffer E (100 mM Tris pH 8, 150 mM NaCl, 1 mM EDTA, 2.5 mM desthiobiotin). The ribosomal proteins (∼ 500 μg) were loaded onto the immobilized His-YidC proteins. After 3 washing steps with 200 μl wash buffer (50 mM Tris pH 7, 100 mM NaCl, 1 mM DTT), the bound proteins were eluted with 500 mM imidazole by centrifugation.


We would like to thank Ross Dalbey and Sebastian Leptihn for critical reading of the manuscript. This work was supported by grants from the Deutsche Forschungsgemeinschaft Ku 749/6, an advanced investigator grant from ERC (to R.B.) and a stipend from the Landesgraduierten-förderung Baden-Württemberg.