A YidC-like Protein in the Archaeal Plasma Membrane

SUMMARY

Cells possess specialized machinery to direct the insertion of membrane proteins into the lipid bilayer. In bacteria, the essential protein YidC inserts certain proteins into the plasma membrane, and eukaryotic orthologs are present in the mitochondrial inner membrane and the chloroplast thylakoid membrane. The existence of homologous insertases in archaea has been proposed based on phylogenetic analysis. However, limited sequence identity, distinct architecture, and the absence of experimental data have made this assignment ambiguous. Here we describe the 3.5-Å crystal structure of an archaeal DUF106 protein from Methanocaldococcus jannaschii (Mj0480), revealing a lipid-exposed hydrophilic surface presented by a conserved YidC-like fold. Functional analysis reveals selective binding of Mj0480 to ribosomes displaying a stalled YidC substrate, and a direct interaction between the buried hydrophilic surface of Mj0480 and the nascent chain. These data provide direct experimental evidence that the archaeal DUF106 proteins are YidC/Oxa1/Alb3-like insertases of the archaeal plasma membrane.

Graphical abstract

INTRODUCTION

Membrane protein biogenesis is an essential process common to all cells. Most membrane proteins are co-translationally inserted into the eukaryotic ER or the prokaryotic plasma membrane following delivery of the ribosome-nascent chain complex (RNC) to the Sec61/SecYEG translocon by the signal recognition particle (Shao and Hegde, 2011; Akopian et al., 2013). Many of the components and core mechanistic features of this pathway are conserved in the three domains of life.

Alternative pathways exist for membrane proteins that do not access the Sec pathway but these appear less conserved. For example, tail-anchored (TA) membrane proteins are post-translationally inserted into the ER membrane following delivery to the Get1/2 insertion machinery by Get3 (Hegde and Keenan, 2011; Chartron et al., 2012; Denic, 2012). This pathway is conserved in eukaryotes but whether it operates in archaea or bacteria remains unclear (Borgese and Righi, 2010; Sherrill et al., 2011; Suloway et al., 2012).

Another pathway requires the YidC/Oxa1/Alb3 family of proteins, which are found in the bacterial plasma membrane, the inner mitochondrial membrane, and the chloroplast thylakoid membrane (Saller et al., 2012; Dalbey et al., 2014). These proteins function in different contexts as insertases, chaperones, and assembly factors. For example, bacterial YidC and chloroplast Alb3 can function as chaperones by associating with the Sec translocon during co-translational insertion of various substrates (Scotti et al., 2000; Urbanus et al., 2001; Klostermann et al., 2002; Nagamori et al., 2004). However, YidC, Alb3, and mitochondrial Oxa1 also function as Sec-independent insertases to drive the insertion of some topologically simple substrates (i.e., those lacking large or highly charged translocated regions) into the membrane. For example, the F₀c subunit of the Escherichia coli F₀F₁ ATPase is co-translationally inserted into the bacterial plasma membrane following delivery of the RNC to YidC (van Bloois et al., 2004; van der Laan et al., 2004). Together, the YidC/Oxa1/Alb3 proteins define a functionally diverse family that is broadly distributed in bacterial and eukaryotic genomes.

Intriguingly, proteins with low sequence similarity to the YidC/Oxa1/Alb3 family have been identified in archaea (Luirink et al., 2001; Yen et al., 2001; Zhang et al., 2009; Makarova et al., 2015). These proteins are annotated as Domain of Unknown Function 106 (DUF106), reflecting their highly diverged sequences and the absence of any experimental insight into their function. Thus, whether archaea possess bona fide members of the YidC/Oxa1/Alb3 family remains unknown.

To gain insight into the evolutionary distribution and mechanism of YidC/Oxa1/Alb3-type insertases, we carried out structural and functional studies of an archaeal DUF106 protein from Methanocaldococcus jannaschii (Mj0480). We determined the 3.5-Å resolution crystal structure of Mj0480, revealing an overall fold and unusual structural features that define it as a member of the YidC/Oxa1/Alb3 family. We also show that Mj0480, like YidC, binds selectively to stalled RNCs and can be crosslinked to the nascent chain via a conserved, lipid-exposed hydrophilic surface. These data provide direct experimental evidence that the archaeal DUF106 proteins function as YidC-like insertases within the archaeal plasma membrane.

RESULTS

Phylogenetic and Topological Analysis of Archaeal DUF106 Proteins

We constructed a phylogenetic tree based on a multiple sequence alignment of a representative set of YidC/Oxa1/Alb3 and archaeal DUF106 sequences. Three separate clades are observed, corresponding to mitochondrial Oxa1, bacterial YidC (including chloroplast Alb3 family members), and the archaeal DUF106 proteins (Figure 1A). Consistent with previous analyses, the YidC/Oxa1/Alb3 homologs are more closely related to each other than to the highly diverged archaeal DUF106 sequences (Luirink et al., 2001; Zhang et al., 2009).

Figure 1. Phylogenetic and Topological Analysis of the Archaeal DUF106 Family.

(A) Phylogenetic tree of representative eukaryotic and bacterial YidC/Alb3/Oxa1 insertases (gray) and members of the archaeal DUF106 family of unknown function (light blue). Branch lengths for the three main clades are indicated; note the high divergence of the archaeal proteins. The archaeal DUF106 protein from M. jannaschii (Mj0480) is indicated with an asterisk. (B) Members of YidC/Oxa1/Alb3 family are predicted to share a core topology comprising five TMs and a cytosolic-facing coiled-coil region between TM1 and TM2; these sequences are most divergent at their N and C termini (dashed lines). Depending on the algorithm, members of the archaeal DUF106 family are predicted to contain either three or four TMs, with a cytosolic-facing coiled coil between TM1 and TM2.

The archaeal DUF106 proteins share a low level of sequence identity (~10%–15%) with members of the YidC/Oxa1/Alb3 family. Moreover, they are generally smaller than their eukaryotic and bacterial homologs, which show large variations in length, particularly at their N and C termini (Figure 1B). Nevertheless, the predicted membrane topology and secondary structure reveal common features. The YidC/Oxa1/Alb3 proteins are thought to share a conserved core comprising five transmembrane (TM) helices (Saaf et al., 1998; Luirink et al., 2001; Kumazaki et al., 2014a, 2014b) and a cytosolic-facing coiled-coil region between the first two TMs. Likewise, COILS sequence analysis (Lupas et al., 1991) of the archaeal DUF106 proteins predicts a cytosolic-facing coiled-coil region between TM1 and TM2. However, in contrast to YidC/Oxa1/Alb3, the conserved transmembrane core of the shorter archaeal proteins is only predicted to contain three or four TMs (Figure 1B).

The Structure of Mj0480

To gain insight into the function of the archaeal DUF106 family, we expressed an archaeal homolog from M. jannaschii (Mj0480) in E. coli. Although we could purify milligram quantities of n-dodecyl-β-D-maltopyranoside (DDM)-solubilized Mj0480, it formed oligomers in a concentration-dependent manner (Figure S1B) and failed to crystallize. To facilitate crystallization, we generated high-affinity synthetic antibody fragments (sABs) using a novel strategy to screen a phage library against biotinylated nanodiscs reconstituted with Mj0480 (Dominik and Kossiakoff, 2015). These sABs were screened for the ability to bind DDM-solubilized Mj0480 and promote crystallization. One such sAB (M1), which appears to form 1:1 and 2:2 complexes with Mj0480 in DDM (Figures S1C and S1D), gave rise to crystals that diffracted anisotropically to ~3.5 A ° ; the structure of this complex was solved by molecular replacement using an antibody fragment as a search model (Table 1; Figure S2).

Table 1.

Crystallographic Data Collection and Refinement Statistics

	Mj0480 Complex
Data Collectiona
Space group	P21212
Cell dimensions
a, b, c (Å)	180.3, 251.4, 96.1
α, β, γ (°)	90, 90, 90
Wavelength (Å)	0.9791
Resolution (Å)	76.4–3.50 (3.63–3.50)b
Unique reflections	55,459
Completeness (%)	99 (99)
Multiplicity	6.8 (6.0)
Mean I/σI	8.9 (1.2)
Rmerge (%)	22.0 (210)
Rmeas (%)	23.8 (230)
CC1/2 (%)	99.7 (50.0)
Refinementc
Resolution range (Å)	76.4–3.50
Unique reflections	48,191
Rwork/Rfree (%)	25.0/30.0
No. of non-H protein atoms	17,849
Average B (Å2)	110
RMSD bond lengths (Å)	0.003
RMSD bond angles (Å)	0.68
Ramachandran outliers (%)	0.48

CC_1/2, half-dataset correlation; RMSD, root-mean-square deviation.

^aData collection statistics are prior to anisotropic truncation.

^bValues in parentheses refer to the high-resolution shell.

^cRefinement statistics are for the final model after refinement against anisotropically truncated and B-factor sharpened data.

The asymmetric unit comprises four Mj0480-sAB complexes packed in an antiparallel arrangement (Figure 2A) that gives rise to alternating layers of sAB and Mj0480 molecules in the crystal (Figure S2B). Two of the Mj0480 molecules are in direct contact and form an antiparallel dimer in which the C-terminal 50 residues from one monomer are swapped with those of the second monomer (Figure 2B).

Figure 2. A Domain-Swapped Dimer in the Mj0480-sAB Complex Crystal.

(A) The asymmetric unit comprises four sAB molecules sandwiched around four Mj0480 molecules; in this arrangement, the four Mj0480 molecules form two antiparallel, domain-swapped dimers (cyan and magenta; blue and green).

(B) Close-up showing the antiparallel arrangement of Mj0480 subunits (cyan, magenta) in the domain-swapped dimer. The swap is initiated at residue 149 (indicated by the asterisk), located in the loop following TM2, and extends through EH2 and TM3 to the C terminus.

(C) Comparison of the crystallographic open and modeled closed monomers, color-ramped from the N terminus (blue) to the C terminus (red). The closed monomer model is generated by connecting residue 148 of one monomer to residue 149 of the other.

(D) Model of the closed Mj0480 monomer (cyan) bound to a single sAB molecule (gray). The sAB-Mj0480 contacts are mediated by residues within the H2 and H3 loops (orange). See also Figure S2.

The observation of domain swapping in protein crystals is not unusual (Liu and Eisenberg, 2002) but it complicates assignment of the physiologically relevant species. Because the Mj0480 monomers are inverted (Figure 2B), it is unlikely that the domain-swapped dimer exists in the archaeal membrane. More likely, the C-terminal swap arises from the high protein concentration used for crystallization and the destabilizing effect of detergent. We therefore constructed an unswapped (closed) model of monomeric Mj0480 by connecting residue 148 of monomer A to residue 149 of monomer B (Figures 2B and 2C).

This Mj0480 monomer spans the membrane with a longest dimension of ~50 Å ; viewed from the cytosol, its overall length and width are ~25 and 36 Å. The N terminus of Mj0480 faces the extracellular space and is followed by an amphipathic helix (EH1) that lies nearly parallel to the plane of the membrane (Figure 3A). This helix ends in a sharp turn and is followed by the first of three bona fide TMs. The cytosolic-facing end of TM1 connects to TM2 via the predicted coiled-coil motif that is disordered in the crystal, presumably reflecting its flexibility. TM2 spans the membrane and packs against TM1 in an antiparallel orientation. This is followed by a short extracellular loop region and a second amphipathic helix (EH2) that also lies nearly parallel to the extracellular face of the membrane. Finally, an extracellular loop connects to TM3, which packs against TM1 and TM2 in the core of the protein and extends toward the disordered C terminus, which ends in the cytosol.

Figure 3. Overall Structure of the Monomeric Mj0480 Model.

(A) Cartoon representation viewed from the plane of the membrane and color-ramped from the N- (blue) to the C- (red) terminus. The structure comprises three transmembrane helices (TM1, TM2, TM3) and two amphipathic helices (EH1, EH2) that lie nearly parallel to the plane of the membrane on the extracellular side of the membrane. The predicted coiled-coil region connecting TM2-TM3 is disordered in the crystal (dotted lines).

(B) Crystallographic B factors are indicated using a three-color gradient from blue (60 Ų) to red (160 Ų).

(C) Electrostatic surface potential colored from negative (red) to positive (blue) reveals an exposed hydrophilic surface within the lipid bilayer.

(D) Sequence conservation of 33 archaeal DUF106-containing proteins mapped to the molecular surface of Mj0480 from most (pink) to least (cyan) conserved.

The lipid-exposed surface along one face of TM1–3 shows strong sequence conservation.

The cytosolic-facing half of Mj0480, including TM1–3, is loosely packed and likely dynamic, as shown by its relatively high B factors (Figure 3B). A striking feature of this region is the presence of a concave hydrophilic surface, presented by TM1–3, that is exposed to both the lipid bilayer and the cytosol (Figure 3C). Importantly, this surface is not accessible to the extracellular side of the membrane by virtue of the tightly packed hydrophobic core comprising residues from each of the five helical elements and the extracellular loops. Moreover, the hydrophilic character of the surface is conserved (Figures 3D and S3). Thus, a key structural feature of the archaeal DUF106 family is the presence of a conserved, lipid- and cytosolexposed hydrophilic surface presented by three transmembrane helices.

Comparison with Bacterial YidC

Despite only ~14% overall sequence identity, the Mj0480 structure shares striking architectural similarities with the recently determined crystal structures of Bacillus halodurans YidC2 (BhYidC) (Kumazaki et al., 2014a) and E. coli YidC (Kumazaki et al., 2014b). The Mj0480 structure superimposes on BhYidC with a root-mean-square deviation of 3.9 Å (over 105 core residues) and a Dali similarity Z score of 6.0 (Figure 4A) (Holm and Park, 2000). This analysis defines a core structural motif that includes EH1, TM1, TM2, and TM3 in Mj0480, and EH1, TM1, TM2, and TM5 in BhYidC (Figure 4A).

Figure 4. Archaeal Mj0480 and Bacterial YidC Share Key Structural Features.

(A) Structure-based alignment of Mj0480 (light blue) and BhYidC (gray; 3WO6) showing views from the plane of the membrane (left) and from the cytosol (right); the proteins superimpose with a root-mean-square deviation of 3.9 A ° over 105 equivalent residues (out of 141 visible).

(B) Structure-based topology cartoons for the YidC/Oxa1/Alb3 (left) and archaeal DUF106 (right) protein families.

(C) Close-up of the region highlighted in (A) (dashed box) showing details of the BhYidC hydrophilic groove; TM4 has been removed for clarity. Despite its location within the lipid bilayer, the groove is lined with polar and charged residues; R72, Q187, and W244 are functionally important for membrane protein insertion (Kumazaki et al., 2014a). The cytosolic-facing vestibule and cytosolic coiled coil (only partially shown for clarity) are lined with methionine side chains.

(D) Close-up of the corresponding region in Mj0480. The conserved, lipid-exposed surface in Mj0480 is lined with polar and charged residues (e.g., N42, T45, W107, Y180, S184, R192); the cytosolic-facing vestibule and the cytosolic coiled coil (disordered) are also enriched in methionine residues. See also Figure S3.

In YidC, this core motif forms part of an unusual element, a hydrophilic, lipid- and cytosol-exposed groove lined with the side chains of polar and positively charged residues (e.g., R72, Q187, and W244 in BhYidC) (Figure 4C) that facilitate the insertion of membrane protein substrates (Kumazaki et al., 2014a, 2014b). Although it is constructed from three TMs (TM1–3) instead of five as in YidC (TM1–5), the Mj0480 groove is also lined with polar (e.g., S38, Y180, S184, S188) and positively charged (e.g., K46, R192) side chains (Figures 4C and S3B). Thus, the presence of an unusual lipid- and cytosol-exposed hydrophilic groove is also conserved between bacterial YidC and Mj0480.

Members of the YidC family contain a dynamic coiled-coil motif near the cytosolic entrance of the hydrophilic groove (Figure 4A). This motif is essential for YidC function in vivo (Chen et al., 2014; Kumazaki et al., 2014a). The corresponding region in Mj0480 is predicted to form a coiled coil but is disordered in the crystal (Figure 4A). Sequence analysis reveals that the cytosolic entrance (including the coiled-coil motif and cytosolic ends of the TMs) contains an unusual abundance of methionine side chains, a conserved feature of the archaeal DUF106 and bacterial YidC families (Figure S3). By analogy to other methionine-rich TM-binding proteins including SRP54 (Bernstein et al., 1989; Janda et al., 2010; Hainzl et al., 2011) and Get3 (Mateja et al., 2015), we propose that this region facilitates interaction with different hydrophobic substrates. Thus, despite their low sequence identity, Mj0480 and the bacterial YidCs share an overall fold and key structural features.

Functional Analysis of Mj0480

To directly test whether the archaeal DUF106 proteins function as YidC-like insertases, we attempted genetic complementation of YidC depletion in E. coil and of an Oxa1 knockout in S. cerevisiae mitochondria. Despite testing mesophilic and thermophilic archaeal DUF106 homologs, no complementation was observed. The inability of distantly related archaeal proteins to complement in these heterologous systems might reflect gross differences in function, or more subtle differences in lipid requirements, substrate specificity, or interaction partners.

As an alternative to complementation, we took advantage of the observation that during co-translational insertion of Sec-independent substrates, YidC and Oxa1 interact with the RNC (Jia et al., 2003; van Bloois et al., 2004; van der Laan et al., 2004). In E. coli, this is accomplished by selective binding of YidC to RNCs containing a substrate nascent chain (Kedrov et al., 2013), while mitochondrial Oxa1 recruits ribosomes even in the absence of a nascent chain (Jia et al., 2003).

To test this with archaeal Mj0480, we first isolated total 70S ribosomes and affinity-purified RNCs from E. coli. In the absence of a known archaeal substrate, we chose to monitor binding to RNCs containing the F₀c subunit of the F₀F₁ ATPase, a known substrate of both YidC and Oxa1 (van der Laan et al., 2004; van Bloois et al., 2005; Jia et al., 2007). To generate stable RNCs displaying only the first TM of F₀c, we used a previously characterized construct containing the stall sequence of SecM fused to the N-terminal region of F₀c (Figure S4) (Schaffitzel and Ban, 2007; Kedrov et al., 2013). Total E. coli 70S ribosomes or affinity-purified stalled RNCs were incubated with detergent-solubilized Mj0480, sedimented by ultracentrifugation, and analyzed for binding by immunostaining (Figure 5A). We observed binding only to stalled RNCs, suggesting that like E. coli YidC, Mj0480 binds selectively to ribosomes containing an exposed hydrophobic nascent chain.

Figure 5. Mj0480 Binds Selectively to RNCs Displaying a YidC Substrate and Can Be Photocrosslinked to Substrate via the Hydrophilic Groove.

(A) Total E. coli 70S ribosomes (100 nM) or affinity-purified stalled RNCs (100 nM) (see also Figure S4) were incubated with recombinant Mj0480 (1 µM) and subjected to ultracentrifugation. Supernatant (S) and pellet (P) (1×, 5×, and 10× volume equivalents of the supernatant) samples were analyzed by SDS-PAGE and immunoblotting (anti-His antibody).

(B) Stalled RNCs or total 70S ribosomes (100 nM) were incubated with proteoliposomes containing either wild-type (WT) or the indicated BpF-containing mutant (~100 nM) and subjected to UV crosslinking and analyzed by SDSPAGE and immunoblotting (anti-Strep antibody). Uncrosslinked nascent chain (NC) and its adducts with Mj0480 are indicated.

(C) Photocrosslinking data mapped to the structure of Mj0480. Positions that crosslink to the stalled nascent chain are highlighted in orange; non-crosslinked sites are shown in yellow. See also Figure S5.

To test whether direct interactions occur between the stalled substrate and Mj0480, we carried out a site-specific photocrosslinking analysis. For this, we designed a series of Mj0480 variants containing p-benzoyl-L-phenylalanine (BpF) substitutions at eight different sites. Seven of these are located within the bilayer, including four that line the hydrophilic groove, while the eighth site is located in a solvent-accessible position on the extracellular side of Mj0480. After purification, BpF-containing Mj0480 variants were reconstituted into proteoliposomes, incubated with stalled RNCs and subjected to UV crosslinking (Figure 5B; Figure S5).

All four bilayer positions lining the hydrophilic groove (I41, W107, F114, F185) could be crosslinked to the stalled F₀c substrate, as expected for a membrane-embedded docking site (Figure 5C). In contrast, only one (W179) of the three bilayer positions distal to the hydrophilic groove (F17, I32, W179) could be crosslinked to substrate. The solvent-exposed extracellular site (Y140) could also be crosslinked, possibly via the soluble N-terminal region of the stalled F₀c substrate that would be accessible after translocation; however, because some Mj0480 might reconstitute into liposomes in the opposite orientation, the Y140 crosslink is not diagnostic for substrate insertion. Together, these data demonstrate that the stalled nascent chain makes direct contacts with the lipid-exposed hydrophilic groove of Mj0480, similar to what has been shown previously for E. coli and B. halodurans YidC (Yu et al., 2008; Klenner and Kuhn, 2012; Kumazaki et al., 2014a).

DISCUSSION

Our structural and biochemical analysis provides direct experimental evidence that the archaeal DUF106 proteins are members of a larger YidC/Oxa1/Alb3 superfamily. We determined the crystal structure of Mj0480 and found that it shares a fold with bacterial YidC (Kumazaki et al., 2014a, 2014b), including a core TM1-cytosolic coiled-coil-TM2-XTM3 motif (Figure 4). This scaffold presents a conserved hydrophilic surface that is exposed to the membrane bilayer and the cytosol. Importantly, we found that this surface directly contacts a stalled membrane protein substrate (Figure 5B), similar to what has been observed previously with bacterial YidC (Yu et al., 2008; Klenner and Kuhn, 2012; Kumazaki et al., 2014a).

A conserved function of many bacterial YidC and mitochondrial Oxa1 homologs is the co-translational, Sec-independent insertion of certain topologically simple proteins into the membrane. This requires an interaction between YidC/Oxa1 and the ribosome. Oxa1 and some bacterial YidC homologs contain an extended, positively charged C terminus that allows ribosome binding in the absence of an associated nascent chain (Jia et al., 2003; Szyrach et al., 2003; Seitl et al., 2014). In contrast, E. coli YidC, which lacks a C-terminal extension, binds selectively to ribosomes displaying hydrophobic nascent chains (Kedrov et al., 2013). We showed that Mj0480, which also lacks a C-terminal extension, binds to stalled RNC-F₀c but not 70S ribosomes. Taken together, these structural and functional similarities establish the universality of the YidC/Oxa1/Alb3 family in all three domains of life.

Our data are consistent with a role for Mj0480 in co-translational Sec-independent insertion but this remains to be demonstrated directly. Doing so will necessitate identification of archaeal substrates that require a DUF106 protein for insertion in vivo and development of a reconstituted in vitro insertion assay that faithfully mimics the unique lipid composition of the archaeal plasma membrane.

It is intriguing that because Mj0480 lacks two of the five TMs present in bacterial YidC (i.e., TM3 and TM4), its hydrophilic groove is smaller and more exposed to the lipid bilayer. This may reflect differences in substrate specificity or in the mechanism of action. For example, the smaller hydrophilic groove might require Mj0480 to oligomerize or to coordinate with the Sec machinery to facilitate insertion and/or folding of certain substrates, in a manner analogous to the Sec-dependent chaperone activity of bacterial YidC (Scotti et al., 2000; Urbanus et al., 2001).

Finally, we note that the core TM1-cytosolic coiled-coil-TM2-X-TM3 topology observed here is predicted to be present in the eukaryotic integral membrane protein Get1/WRB, a subunit of a complex required for TA protein insertion into the ER membrane (Schuldiner et al., 2008; Mariappan et al., 2011; Vilardi et al., 2011; Wang et al., 2011). TA proteins, like Sec-independent YidC/Oxa1/Alb3 substrates, lack large translocated regions. It will be of interest to determine if Get1/WRB belong to a superfamily of functionally diverse proteins including YidC/Oxa1/Alb3 and the archaeal DUF106 proteins, whose main similarity is mediating insertion of topologically simple membrane proteins.

EXPERIMENTAL PROCEDURES

Phylogenetic Analysis

Archaeal DUF106 sequences were obtained using Mj0480 as a query in NCBI Blast with an E-value cutoff of 10⁻³⁰ and these were aligned with representative bacterial YidC, eukaryotic Oxa1 and eukaryotic Alb3 sequences using T-Coffee (Notredame et al., 2000) with the default parameters. Gaps were trimmed automatically using the Software TrimAl (Capella-Gutierrez et al., 2009) with a cutoff of 0.55. The resulting alignment was tested in Prottest (Darriba et al., 2011) to find the best evolutionary model, which for these data was an LG substitution matrix, empirical amino acid frequencies, fixed gamma shape parameter (1.491) with four substitution rate categories. The phylogenetic tree was then assembled in PhyML (Guindon et al., 2010) using these parameters.

Mj0480 Production

The gene encoding full-length M. jannaschii Mj0480 was PCR amplified from dried genomic DNA (ATCC #43067D-5), subcloned into a pET28a derivative (Novagen) encoding a C-terminal 6xHis tag, and verified by DNA sequencing. After transformation into E. coli BL21(DE3), a single colony was used to inoculate a terrific broth (TB) starter culture supplemented with 50 µg/ml kanamycin. After shaking for ~4 hr at 37 °C and 250 rpm, 1 ml of starter culture was used to inoculate 2 × 0.5 l of homemade TB autoinduction medium (Studier, 2005) supplemented with 50 µg/ml kanamycin. After ~16 hr at 37°C and 250 rpm, cells were harvested by centrifugation and resuspended in 100 ml buffer A (50 µM HEPES [pH 7.5], 200 mM NaCl) supplemented with 1 mM PMSF and 2 mM DNase I. Cells were lysed by passing three times through a high-pressure microfluidizer (Avastin), and the insoluble fraction was harvested by centrifugation for 40 min at 4°C/18,500 × g and stored at −80°C.

The insoluble fraction was resuspended in buffer A supplemented with 1mM PMSF and 1% DDM (Anatrace) and incubated for ~1.5 hr at 4°C with gentle mixing. The detergent soluble supernatant was isolated by centrifugation for 40 min at 4°C/18,500 × g and batch-purified by cobalt-affinity chromatography. After washing the column with buffer A supplemented with 20mMimidazole and 0.04% DDM, protein was eluted in buffer A with 200 mM imidazole and 0.04% DDM. Typically, the protein was further purified by size exclusion chromatography (Superdex 200 10/300 GL, GE Healthcare) in 10 mM HEPES (pH 7.5), 200 mM NaCl, 0.04% DDM. Peak fractions were pooled, concentrated to 5 mg/ml in a 100-kDaMWCOAmicon Ultra centrifugal filter (Millipore), and large aggregates were removed by centrifugation at 15,000 × g for 10 min at 4°C before aliquots were flash frozen in liquid nitrogen and stored at −80°C. Protein concentrations were determined by A₂₈₀ using a calculated extinction coefficient.

For photocrosslinking experiments, site-directed Mj0480 amber mutants were generated by QuikChange mutagenesis (Stratagene) and verified by DNA sequencing. After co-transformation with pEVOL-pBpF (Chin et al., 2002) into E. coli BL21-Gold(DE3) (Agilent), single colonies were used to inoculate TB starter cultures supplemented with 50 µg/ml kanamycin and 50 µg/ml chloramphenicol. After ~16 hr at 37°C and 250 rpm, 20 ml of starter culture was used to inoculate 1 l of TB supplemented with 50 µg/ml kanamycin and 50 µg/ml chloramphenicol. After the cells reached ~0.6 A₆₀₀, 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG), 0.2% arabinose, and 1 mM p-benzoylphenylalanine (BpF) (Bachem) were added and the culture was grown for an additional 5 hr at 37°C and 250 rpm. Purification was as above.

Nanodisc Phage Display Selection

sABs were obtained essentially as described (Dominik and Kossiakoff, 2015). Briefly, biotinylated MSP1 E3D1 (Addgene) was incubated with purified Mj0480 and n-undecyl-β-D-maltopyranoside resolubilized chicken egg phosphatidylcholine (PC) lipids (Avanti), as described (Ritchie et al., 2009). Excess detergent was removed by overnight incubation at 4°C with Bio-Beads SM-2 Adsorbents (Bio-Rad). Mj0480-loaded nanodiscs were purified using Ni-NTA FastFlow Resin (Qiagen) and size exclusion chromatography (Superdex 200 10/300 GL, GE Healthcare) in buffer A. After validating immobilization and release from Streptavidin MagneSphere Paramagnetic Particles (Promega), biotinylated Mj0480 nanodiscs were used as a target in a five-round competitive phage display library selection protocol using Library E (kindly provided by S. Koide) (Miller et al., 2012). Initial validation was performed by single point direct phage ELISA against Mj0480 nanodiscs and empty nanodiscs as a control. Clones with A₄₅₀ higher than 0.2 (4× average background) were sequenced and unique clones were kept for further analysis.

Synthetic Antibody Fragment Production

sABs were subcloned into the expression vector RH2.2 (gift from S. Sidhu), transformed into E. coli BL21-Gold(DE3) (Agilent) and a single colony was used to inoculate 15 ml overnight starter cultures in 2xYT medium (Fisher) supplemented with 50 µg/ml ampicillin. 10 ml of starter cultures was used to inoculate a 1-l 2xYT culture supplemented with 50 µg/ml ampicillin; after growing to an OD₆₀₀ of ~0.8–1.0, cells were induced with 1 mM IPTG and grown for an additional 4 hr at 37°C and 220 rpm.

Cells were harvested by centrifugation and disrupted in lysis buffer containing 20mMsodium phosphate (pH 7.4), 500mMsodium chloride, 1mMPMSF, and 2 mM DNase, using a high-pressure microfluidizer (Avestin). Lysate was heated to 60°C for 30 min, cleared by centrifugation, and loaded onto a HiTrap MabSelect SuRe 5-ml column (GE Healthcare) equilibrated with buffer containing 20 mM sodium phosphate (pH 7.4) and 500 mM NaCl. After washing with 10 column volumes of equilibration buffer, protein was eluted with 0.1 M acetic acid. Peak fractions were loaded onto an ion exchange Resource S 1-ml column (GE Healthcare). After washing with 50 mM sodium acetate (pH 5.0), sABs were eluted with a linear 0%–100% gradient of buffer containing 50mMsodium acetate (pH 5.0) and 2Msodium chloride. Fractions containing pure sAB were pooled, neutralized with 50 mM HEPES (pH 7.5), dialyzed against buffer containing 50mMHEPES (pH 7.5) and 200mMsodium chloride, concentrated, and stored in aliquots at −80°C.

Crystallization and Data Collection

Purified Mj0480 was mixed with 1.2-fold molar excess of sAB M1 and incubated for 1 hr at 4°C (Figure S1). The Mj0480-sAB complex was separated from excess sAB by size exclusion chromatography (Superdex 200 10/300 GL, GE Healthcare) in 50 mM HEPES (pH 7.5), 200 mM NaCl, 0.04% DDM. Peak fractions were pooled, concentrated in a 100-kDa MWCO Amicon Ultra centrifugal filter (Millipore), centrifuged for 10 min at 15,000 × g at 4°C, and either used immediately for crystallization or stored in aliquots at −80°C.

Crystals of the Mj0480-sAB complex were grown at room temperature by hanging drop vapor diffusion. Equal volumes of a protein solution containing 10 mg/ml complex was mixed with a reservoir solution containing 0.08 M calcium acetate and 18% PEG 3350. Crystals were cryoprotected in 0.08 M calcium acetate, 18% PEG 3350, 20% ethylene glycol, 0.04% DDM, and flash frozen in liquid nitrogen.

Two datasets were collected from different regions of a single crystal at 100 K at APS beamline 24ID-E (λ = 0.9791 A °) on an ADSC Q315 detector. The data were processed, scaled, and merged using XDS/XSCALE (Kabsch, 2010) as implemented in Xia2 (Winter, 2010). The strongest dataset was severely anisotropic, with diffraction limits of 3.5 A ° along b* and c* but only 5.5 A ° along a*. The second dataset was lower resolution but more isotropic, with diffraction extending to 3.6 and 3.8 A ° along b* and c*, respectively, and 4.4 A ° along a*. After merging, the resolution limit of the combined data was 3.5A ° (half-dataset correlation is >50%). Data collection and processing statistics are listed in Table 1.

Structure Determination and Refinement

The Mj0480-sAB complex structure was determined to 3.5 A ° by molecular replacement with PHASER (McCoy et al., 2007), using an sAB (PDB: 3PGF, with its complementarity determining regions (CDR) omitted) (Rizk et al., 2011) as the search model. Initial maps showed four sABs packed closely together in the asymmetric unit, with their CDRs pointing outward toward weak helical density. Cycles of manual model building and refinement (torsional non-crystallographic symmetry restraints applied throughout) with COOT (Emsley et al., 2010) and PHENIX (Adams et al., 2010) allowed us to trace the backbone of Mj0480 (Figure S2). During the later stages of refinement, the data were elliptically truncated and B-factor sharpened (−30.3 Ų) using the UCLA diffraction anisotropy server (http://services.mbi.ucla.edu/anisoscale) (Strong et al., 2006).

The Mj0480 electron density was strongest in the extracellular-facing regions and this allowed us to assign the amino acid sequence. Residual density extending outward from the cytosolic-facing side of TM1 and TM2 likely corresponds to the coiled-coil region seen in YidC, but because the density is very weak, we did not build into it. There is also weak density present around the sAB H3 loop; this may correspond to an ordered DDM molecule. The final model contains four Mj0480 and four sAB monomers. Refinement and validation statistics are listed in Table 1.

Sequence conservation from an alignment of 33 archaeal DUF106 proteins sharing between 20%–91% sequence identity to Mj0480 was mapped to the surface of Mj0480 using the ConSurf server (Glaser et al., 2003) (http://consurf.tau.ac.il). Structure figures were generated with PyMOL (http://www.pymol.org).

Complementation Assays

The Mj0480 gene and two closely related mesophilic archaeal homologs (from Methanococcus aeolicus and Methanococcus maripaludis) were cloned into pQE81-L vector encoding a C-terminal 6xHis tag using Gibson assembly (Gibson et al., 2009); we also cloned chimeras in which the gene of interest was fused to the first 330 residues of E. coli YidC. Full-length YidC was used as a positive control and protein expression was verified by western blot using an anti-6xHis antibody.

After transforming the E. coli JS7131 YidC depletion strain (gift from R. Dalbey) (Samuelson et al., 2000) with the appropriate experimental construct, single colonies were picked and grown for 4–6 hr in lysogeny broth (LB) medium supplemented with 0.2% arabinose, 25 µg/ml spectinomycin, and 100 µg/ml ampicillin, taking care that cells never entered the stationary phase. Next, cells of OD₆₀₀ 0.2 were pelleted for 5 min at 5,000 × g at 4°C, washed with 1 ml of LB, resuspended to a final OD₆₀₀ of 0.1, and then 2 µl of 1:10 serial dilutions were spotted on LB plates supplemented with spectinomycin, ampicillin, and either 0.2% arabinose (positive control) or 0.2% glucose (negative control). The latter condition was necessary as the JS7131 strain sometimes spontaneously reverts to a wild-type phenotype. For the experimental condition, we tested IPTG concentrations between 20 and 500 µM.

To test Oxa1 complementation, we generated a yeast W303-1A Doxa1 knockout strain by homologous recombination. This strain was verified by PCR and exhibited an inability to grow on a nonfermentable carbon source as previously reported (Preuss et al., 2005). Complementation was tested by cloning either the Mj0480 gene or its M. aeolicus homolog into a Yeplac195 vector (gift from B. Glick) with a C-terminal 6xHis tag. Each gene was fused to the Oxa1 mitochondrial targeting sequence and optionally also the long Oxa1 C-terminal tail, which is important for its function (Preuss et al., 2005). A vector carrying a C-terminally 6xHis-tagged Oxa1 was used as a positive control and empty vector was used as a negative control.

After transforming the knockout strain with the appropriate construct using the lithium acetate method (Gietz and Woods, 2002), a single colony was restreaked on fresh plates, then used to inoculate CSM-Ura cultures that were grown until they became cloudy (OD₆₀₀ <2). Next, cells of OD₆₀₀ 0.5 were centrifuged for 4 min at 2,600 × g at 4°C and resuspended in 1 ml of cold water. 2 µl of this and 5× 1:10 serial dilutions were spotted on yeast peptone agar plates supplemented with either 2% glucose or 3% glycerol; plates were incubated at room temperature, 30°C, and 37°C. Only strains transformed with the Oxa1 gene grew on 3% glycerol plates.

Ribosome Binding Assays

Total 70S E. coli ribosomes were purified according to a published protocol (Wu et al., 2013) with the exception that we used E. coli BL21(DE3) cells. Stalled RNCs were obtained using a construct containing a 3× Strep-tag II sequence fused to the first transmembrane segment of the YidC substrate F₀c and to the SecM stalling sequence, similar to a previously utilized construct (Kedrov et al., 2013). An artificially synthesized gene was Gibson assembled into pET21a and transformed into BL21(DE3) cells. Stalled RNC-F₀cs were expressed and purified according to the published protocol (Wu et al., 2013) (Figure S4). Both the total 70S ribosomes and the stalled RNCs were quantified by A₂₆₀, flash frozen, and stored at −80°C.

Binding assays were performed by incubating 100 nM ribosomes or RNCs with 1 µM Mj0480 in 1.5 ml of assay buffer containing 50 mM Tris (pH 7.5), 150 mM KCl, 5 mM MgCl₂, and 0.04% DDM. After 1 hr on ice, the samples were pelleted for 3 hr at 100,000 × g at 4°C in a TLA100.3 rotor. An aliquot of the supernatant was saved for analysis and the rest was discarded. Pellets were washed with 300 µl of assay buffer, resuspended in 50 µl of SDS sample buffer, and then adjusted to 1×, 5×, and 10× volume equivalents of the supernatant sample. Supernatant and pellet samples were analyzed by SDS-PAGE and western blotting with a Penta-His HRP conjugated antibody (Qiagen catalog #34460).

Site-Specific Photocrosslinking Assays

Chloroform stocks of egg PC and 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine lipids (Avanti) were mixed in a glass tube in a 4:1 mass ratio and supplemented with 10mMDTT. The lipid solution was dried in a glass tube under a stream of nitrogen followed by overnight lyophilization. Lipids were solubilized to a final concentration of 20 mg/ml in buffer B containing 50 mM HEPES (pH 7.4), 150 mM potassium acetate, and 5 mM magnesium acetate by end-over-end mixing for 1 hr at room temperature with intermittent vortexing. The resulting milky suspension was extruded 25 times through 200-nm polycarbonate membranes at 74°C using an Avanti mini-extruder. The final clear liposome solution was aliquoted, flash frozen in liquid nitrogen, and stored at −80°C.

Proteoliposomes were prepared by mixing gel-filtration purified wild-type Mj0480 or BpF mutants (final protein concentration of 0.3 µM) with 10 µl of liposomes and adjusting to a final volume of 110 µl in reconstitution buffer (50 mM HEPES [pH 7.4], 150 mM potassium acetate, 5 mM magnesium acetate, 1 mM DTT, and 0.1% deoxy big CHAP). After 15 min on ice, 50 mg of activated Bio-Beads SM-2 adsorbents (Bio-Rad) were added and the mixture was incubated overnight at 4°C with end-over-end mixing. After removing the Bio-Beads, the supernatant was centrifuged for 10 min at 15,000 × g at 4°C to remove large aggregates. Reconstitution efficiency was estimated by stain-free SDS-PAGE and Image J analysis.

Photocrosslinking analysis was carried out by incubating ~100 nM wildtype Mj0480 or BpF proteoliposomes with 100 nM of 70S ribosomes or stalled RNCs for 1 hr at 4°C in buffer B. As a loading control, an aliquot of each sample was removed prior to UV treatment and analyzed for Mj0480 by SDS-PAGE and immunoblotting with a Penta-His HRP conjugated antibody (Qiagen #34460) (Figure S5). Half of the remaining sample was set aside as the non-UV treated control. The other half was photocrosslinked via BpF for 15 min on ice in a 96-well polypropylene plate using a 365-nm longwave UV spot lamp placed ~4 cm from the sample. Subsequently, the UV and non- UV samples were handled identically. Samples were treated with RNase cocktail (Ambion) for 35 min at 37°C and then diluted into buffer B with 1% Fos-choline-12 (FC-12, Anatrace). The detergent-solubilized samples were bound to Ni-NTA resin (Qiagen), washed extensively, and eluted in SDS-PAGE loading buffer. The samples were analyzed by SDS-PAGE using a 26-well Criterion TGX precast 4%–20% Tris-glycine gels (Bio-Rad) and western blotting using an anti-Strep-tag II antibody (Abcam #ab76949) to monitor the stalled substrate.

Highlights.

• An archaeal DUF106 protein from M. jannaschii (Mj0480) adopts a YidC-like fold

• Mj0480 binds selectively to stalled ribosome-nascent chain complexes

• The nascent chain directly contacts a lipid-exposed hydrophilic surface of Mj0480

• This work establishes the universality of YidC proteins in all three domains of life