1. Biochemistry and Chemical Biology
  2. Structural Biology and Molecular Biophysics
Download icon

The C-terminal tail of the bacterial translocation ATPase SecA modulates its activity

Research Article
  • Cited 0
  • Views 942
  • Annotations
Cite this article as: eLife 2019;8:e48385 doi: 10.7554/eLife.48385

Abstract

In bacteria, the translocation of proteins across the cytoplasmic membrane by the Sec machinery requires the ATPase SecA. SecA binds ribosomes and recognises nascent substrate proteins, but the molecular mechanism of nascent substrate recognition is unknown. We investigated the role of the C-terminal tail (CTT) of SecA in nascent polypeptide recognition. The CTT consists of a flexible linker (FLD) and a small metal-binding domain (MBD). Phylogenetic analysis and ribosome binding experiments indicated that the MBD interacts with 70S ribosomes. Disruption of the MBD only or the entire CTT had opposing effects on ribosome binding, substrate-protein binding, ATPase activity and in vivo function, suggesting that the CTT influences the conformation of SecA. Site-specific crosslinking indicated that F399 in SecA contacts ribosomal protein uL29, and binding to nascent chains disrupts this interaction. Structural studies provided insight into the CTT-mediated conformational changes in SecA. Our results suggest a mechanism for nascent substrate protein recognition.

https://doi.org/10.7554/eLife.48385.001

Introduction

In Escherichia coli, approximately a quarter of all newly synthesised proteins are transported across the cytoplasmic membrane by the Sec machinery (Cranford-Smith and Huber, 2018). Of these, the majority (~65%) require the activity of SecA for translocation across the membrane. SecA is an evolutionarily conserved and essential ATPase that is required for protein translocation in bacteria (Cranford-Smith and Huber, 2018). The catalytic core of SecA (amino acids ~1–832 in E. coli) contains five domains (Figure 1—figure supplement 1): nucleotide binding domain-1 (NBD1; amino acids 9–220 and 378–411), the polypeptide crosslinking domain (PPXD; 221–377), nucleotide binding domain-2 (NBD2; 412–620), the α-helical scaffold domain (HSD; 621–672 and 756–832) and the α-helical wing domain (HWD; 673–755). Binding and hydrolysis of ATP at the NBD1/NBD2 interface cause conformational changes in the HSD and HWD, which are required for the translocation of substrate proteins (Cranford-Smith and Huber, 2018; Collinson et al., 2015). In addition, the PPXD undergoes a large conformational change, swinging from a position near the HWD (the ‘closed’ conformation) to a position near NBD2 (the ‘open’ conformation) (Zimmer et al., 2008; Zimmer and Rapoport, 2009; Chen et al., 2015). SecA binds to substrate protein in the groove formed between the PPXD and the two NBDs, and the PPXD serves as a ‘clamp’ that locks unfolded substrate proteins into this groove when it is in the open conformation (Zimmer and Rapoport, 2009).

SecA can recognise its substrate proteins cotranslationally. SecA binds to the ribosome (Huber et al., 2011), and ribosome binding facilitates the recognition of nascent substrate proteins (Huber et al., 2017; Huber et al., 2011). The binding site for SecA on the ribosome includes ribosomal protein uL23, which is located adjacent to the opening of the polypeptide exit channel (Huber et al., 2011). Structural and biochemical studies indicate that ribosome binding is mediated by two regions in the catalytic core: the N-terminal α-helix (Singh et al., 2014) and the N-terminal portion of the HSD (Huber et al., 2011) (Figure 1—figure supplement 1). The structure of the SecA-ribosome complex was recently determined at medium resolution (~11 Å) by cryo-electron microscopy (Singh et al., 2014). However, the molecular mechanism governing the recognition of nascent substrate proteins is unknown.

In addition to the catalytic core, most SecA proteins contain a relatively long C-terminal tail (CTT; also known as the C-terminal linker [Hunt et al., 2002]), whose function is not well understood (Figure 1A). In E. coli, the CTT (833-901) contains of a small metal binding domain (MBD; 878–901) and a structurally flexible linker domain (FLD; 833–877). Although it is not required for protein translocation (Or et al., 2002; Or et al., 2005; Fekkes et al., 1997), E. coli strains producing a C-terminally truncated SecA protein display modest translocation defects (Or et al., 2005; Grabowicz et al., 2013). The MBD coordinates a metal ion (thought to be Zn2+) via a conserved CXCX8C(H/C) motif (Dempsey et al., 2004; Fekkes et al., 1999). In E. coli, the MBD is required for interaction of SecA with SecB (Kimsey et al., 1995; Breukink et al., 1995; Fekkes et al., 1997), a molecular chaperone that is required for the secretion of a subset of Sec substrate proteins (Randall and Hardy, 2002). Although not all SecA proteins contain an MBD, the MBD is conserved in many species that do not have a SecB homologue (Dempsey et al., 2004).

Figure 1 with 2 supplements see all
Phylogenetic analysis of the CTT and binding of E. coli CTT to the ribosome.

(A) Schematic diagram of the primary structure of SecA, SecAΔMBD and SecAΔCTT. Structures are oriented with the N-termini to the left, and the amino acid positions of the N- and C-termini are indicated. Residues of the catalytic core and the CTT are indicated below. Catalytic core, black; FLD, yellow; MBD, red. (B) Phylogenetic tree of the SecA proteins of 156 representative species from 155 different bacterial families. Species names are given as the five-letter organism mnemonic in UniProtKB (The UniProt Consortium, 2017). Taxonimical classes are colour-coded according to the legend. Leaves representing SecA proteins with an MBD are coloured black. Those with CTTs lacking a MBD are coloured red, and those that lack a CTT entirely are coloured yellow. Species that also contain a SecB protein are indicated with a star (*). (C) Logo of the consensus sequence of the MBD generated from the 117 species containing the MBD in the phylogenetic analysis. Positions of the metal-coordinating amino acids are indicated above. Amino acids that contact SecB in the structure of the MBD-SecB complex (Zhou and Xu, 2003) (1OZB) are indicated by arrowheads below. (D) Binding reactions containing 1 μM ribosomes, 10 μM SUMO-CTT and 10 μM AMS-modified SUMO-CTT (AMS-SUMO-CTT) were equilibrated at room temperature and layered on a 30% sucrose cushion. Ribosomes were then sedimented through the cushion by ultracentrifugation. Samples were resolved on SDS-PAGE and probed by western blotting against the Strep tag using HRP-coupled Streptactin. (E) 10 μM SUMO-CTT containing an N-terminal Strep(II)-tag was incubated with 1 μM purified ribosomes and treated with 5 mM or 25 mM EDC, as indicated. Samples were resolved by SDS-PAGE and analysed by western blotting by simultaneously probing against SecA (red) and ribosomal protein uL23 (green). The positions of SUMO-CTT, L23 and crosslinking adducts between them (*) are indicated at left.

https://doi.org/10.7554/eLife.48385.002

In this study, we investigated the role of the CTT in the recognition of nascent substrate proteins by SecA. Phylogenetic and sequence analysis of the CTT suggested that it could be involved in binding of SecA to ribosomes, which we confirmed using ribosome cosedimentation and chemical crosslinking approaches. Strikingly, disruption of the MBD alone or the entire CTT had opposing effects on multiple activities of SecA, suggesting that the CTT affects conformation of the catalytic core. Mass spectrometry, x-ray crystallography, and small-angle x-ray scattering experiments indicated that the FLD is bound in the substrate binding groove and affects the conformation of the PPXD. Finally, site specific chemical crosslinking suggested that binding of the MBD to the ribosome allows full-length SecA to interact with nascent substrate proteins. Taken together, our results provide insight into the molecular mechanism underlying nascent substrate recognition by SecA.

Results

Evolutionary distribution of the MBD of SecA

To investigate the evolutionary distribution of the CTT, we analysed the sequences of 156 SecA proteins from bacterial species in 155 phylogenetic families using ClustalOmega (McWilliam et al., 2013). The phylogenetic tree produced by this analysis generally placed SecA proteins from more closely related species (e.g. those in the same phylogenetic class) into similar groups (Figure 1B; Supplementary files 1 and 2). The majority of SecA proteins (143) contained a CTT (Figure 1B, red and black branches). Of these, 117 contained an MBD (Figure 1B, black branches). A small minority (13) lacked the CTT entirely (Figure 1B, yellow branches). Of the 69 SecA proteins from species that contained a SecB homologue (Figure 1B, starred species), only two lacked an MBD. The strong co-conservation of the MBD and SecB suggests that there is strong selective pressure to maintain the MBD in species possessing SecB, consistent with previous studies indicating that the MBD is required for binding of SecA to SecB (Fekkes, Fekkes et al., 1997; Fekkes et al., 1999; Zhou and Xu, 2003; Randall et al., 2004). However, a significant number of species that lack SecB (52) also contain a SecA protein with an MBD. Furthermore, many of the residues implicated in SecB binding were strongly conserved in these MBDs (Figure 1C, arrowheads) (Zhou and Xu, 2003). These results suggested that the MBD has an evolutionarily conserved function in addition to its role in binding to SecB.

Binding of the CTT to the ribosome

Many of the most highly conserved residues in the MBD consensus sequence (including in species that lack SecB) are positively charged and surface exposed (Figure 1C), which suggested that the MBD could also bind to the negatively charged surface of the ribosome. A fusion between the small ubiquitin-like modifier (SUMO) from Saccharomyces cerevisiae and the CTT of E. coli SecA (SUMO-CTT) co-sedimented with ribosomes through a sucrose cushion during ultracentrifugation, indicating that the CTT binds to ribosomes (Figure 1D, lanes 2 and 4). A shorter protein fusion containing only the MBD (SUMO-MBD) also cosedimented with ribosomes (Figure 1—figure supplement 2), indicating that the MBD is responsible for this ribosome binding activity, and modification of the metal-coordinating cysteines with AMS disrupted the ability of SUMO-CTT to cosediment with ribosomes (Figure 1D, lanes 3 and 5). Incubation of SUMO-CTT with ribosomes in the presence of 5 mM and 25 mM EDC (a non-specific crosslinking agent) resulted in the appearance of several crosslinking products. These products cross-reacted with antibodies against uL23 and SecA (Figure 1E), suggesting that the CTT binds in the vicinity of the opening of the polypeptide exit tunnel similar to full-length SecA.

Effect of C-terminal truncations on the affinity of SecA for ribosomes

We next determined the affinity of C-terminal truncation variants of SecA for the ribosome using fluorescence anisotropy (Huber et al., 2011) (Figure 2A and Table 1). The equilibrium dissociation constant (KD) of the complex between full-length SecA and ribosomes was ~640 nM, similar to previously published figures (Figure 2A and Table 1) (Huber et al., 2011). Truncation of the C-terminal 69 amino acids of SecA (SecAΔCTT) caused a modest, but reproducible, increase in the KD of the SecA-ribosome complex (920 nM) (Figure 2A and Table 1). However, truncation of the C-terminal 21 amino acids (SecAΔMBD) significantly increased the affinity of SecA for the ribosome (KD = 160 nM) (Figure 2A and Table 1). These differences in affinity were sufficient to affect the amount of SecA that cosedimented with ribosomes during ultracentrifugation (Figure 2B, lanes 4–6).

Figure 2 with 3 supplements see all
Effect of C-terminal truncations on SecA function in vitro and in vivo.

(A) 900 nM Ru(bpy)2(dcbpy)-labelled SecA (Wild type; circles), SecAΔMBD (ΔMBD; triangles) or SecAΔCTT (ΔCTT; squares) was incubated in the presence of increasing concentrations of purified 70S ribosomes. Because error bars corresponding to one standard deviation obscured the symbols, they were omitted from the graph. The equilibrium dissociation constant (KD) of the complex was determined by fitting the increase in fluorescence anisotropy from the Ru(bpy)2(dcbpy) (lines; Table 1). (B) 0.5 μM SecA, SecAΔMBD or SecAΔCTT was incubated in the absence (lanes 1–3) of ribosomes, in the presence of 0.5 μM vacant 70S ribosomes (lanes 4–9) or in the presence of 0.5 μM RNCs containing nascent SecM peptide (lanes 10–12). Where indicated, binding reactions were incubated in the presence of 100 mM (lanes 1–6) or 250 mM (lanes 7–12) potassium acetate (KOAc). Binding reactions were layered on a 30% sucrose cushion and ribosomes were sedimented through the sucrose cushion by ultracentrifugation. Ribosomal pellets were resolved by SDS-PAGE and stained by Coomassie. (C) 600 nM IAANS-VipB peptide was incubated with increasing concentrations of SecA (Wild type; circles), SecAΔMBD (ΔMBD; triangles) or SecAΔCTT (ΔCTT; squares). Confidence intervals represent one standard deviation. The KD for the SecA-peptide complex was determined by fitting the increase in IAANS fluorescence upon binding to SecA (lines; Table 1). (D) Growth of strains producing SecA (DRH1119; bottom left), SecAΔMBD (DRH1120; bottom right) and SecAΔCTT (DRH1121; top) on LB plates containing 100 μM IPTG.

https://doi.org/10.7554/eLife.48385.007
Table 1
Biochemical properties of wild-type and mutant SecA proteins.
https://doi.org/10.7554/eLife.48385.011
SecA variantKD Ribosomes*KD VipBBasal ATPase activityTM§
Wild type640 ± 33 nM0.9 μM0.053 ± 0.02 s−140.7 ± 0.09°C
SecAΔMBD160 ± 35 nM1.7 μM<0.001 s−142.0 ± 0.08°C
SecAΔCTT920 ± 38 nM5.9 μM0.91 ± 0.02 s−140.0 ± 0.1°C
SecAC885A/C887AND>10 μM<0.001 s−1ND
SecABpa852ND>10 μM<0.001 s−1ND
  1. *Equilibrium dissociation constant of the complex between SecA and non-translating 70S ribosomes as determined by fluorescence anisotropy. Confidence intervals are the standard error of the fit.

    Equilibrium dissociation constant of the complex between SecA and IAANS-labelled VipB peptide as determined by change in fluorescence.

  2. Rate of ATP hydrolysis by SecA in the absence of substrate protein and SecYEG

    §Denaturation midpoint temperature as determined by the change in circular dichroism at 222 nm.

  3. not determined

Effect of truncations on affinity of SecA for nascent polypeptides

To investigate whether the truncations affected the affinity of SecA for nascent chains, we examined binding of SecAΔMBD and SecAΔCTT to ribosome nascent chain complexes (RNCs) containing arrested nascent SecM. SecM is a model nascent SecA substrate protein (Huber et al., 2017; Huber et al., 2011). Similar to full-length SecA (Huber et al., 2011), binding of SecAΔMBD and SecAΔCTT to non-translating ribosomes was sensitive to high concentrations of salt in the binding buffer (Figure 2B, lanes 7–9). Binding of SecA and SecAΔCTT to ribosomes in the presence of 250 mM potassium acetate was stabilised by the presence of arrested nascent SecM (Figure 2B, lanes 10 and 12). However, nascent SecM did not stabilise binding of SecAΔMBD to RNCs under the same conditions (Figure 2B, lane 11). These results suggested that SecAΔMBD is defective for binding to nascent substrate protein.

Site-specific crosslinking of SecA to ribosomes

To investigate binding of SecA to the ribosome in more detail, we incorporated p-benzoyl-L-phenylalanine (Bpa) into SecA at positions 56, 260, 299, 399, 406, 625, 647, 665, 685, 695, 748 and 796 using nonsense suppression (Figure 3A and B) (Singh et al., 2014; Huber et al., 2011; Chin et al., 2002). The side chains of the amino acids at these positions are located on the surface of SecA that binds to the ribosome (Figure 3A). Bpa contains a photoactivatable side chain that forms covalent crosslinks to nearby molecules containing C-H bonds. In the presence of purified 70S ribosomes, SecA containing Bpa at positions 399 (SecABpa399) and 406 (SecABpa406) produced additional high molecular weight bands in SDS-PAGE, which were recognised by α-SecA antiserum (Figure 3C). Analysis of the high-molecular weight band produced by SecABpa399 by mass-spectrometry (LC-MS/MS) indicated that it was an adduct between SecA and ribosomal protein uL29. uL29 is located adjacent to uL23 on the ribosomal surface, and both F399 and K406 appear to contact uL29 in the structure of the SecA-ribosome complex determined by cryo-electron microscopy (Figure 3A) (Singh et al., 2014). SecA containing Bpa at position 299 also produced a crosslinking adduct that migrated with a larger apparent molecular weight than the SecA-uL29 adduct (Figure 3C). However, the identity of the crosslinking protein is unknown. SecAΔMBDBpa399 also produced a high molecular weight crosslinking adduct in the presence of ribosomes, and LC-MS/MS confirmed the presence of both SecA and uL29 in the band (Figure 3D), indicating SecAΔMBD binds to ribosomes at the same site as full-length SecA.

Figure 3 with 2 supplements see all
Site-specific crosslinking of SecA to purified ribosomes and ribosome-nascent chain complexes.

(A and B) Sites of incorporation of Bpa in the structure of E. coli SecA. (A) Fit of the high resolution structure of SecA (PDB code 2VDA [Gelis et al., 2007]) and the 70S ribosome (PDB code 4V4Q [Schuwirth et al., 2005]) to the cryoEM structure of the SecA ribosome complex (EMD-2565 [Singh et al., 2014]). (B) View of SecA from the ribosome-interaction surface. Amino acid positions where Bpa was incorporated are represented in space fill (yellow). Positions that crosslink to ribosomal proteins are coloured red. The locations of the N-terminal α-helix of SecA and of ribosomal proteins uL23 (dark blue), uL29 (purple) and uL24 (cyan) are indicated. Structural models were rendered using Chimera v. 1.12 (Pettersen et al., 2004). (C) Bpa-mediated photocrosslinking of SecA variants to vacant 70S ribosomes. 1 μM purified ribosomes were incubated with 1 μM SecA containing BpA at the indicated position and exposed to light at 365 nm (above) or incubated in the dark. Crosslinking adducts consistent with the molecular weight of a covalent crosslink to ribosomal proteins are indicated with red arrowheads. The positions of full-length SecA and uncleaved SUMO-SecA protein are indicated to the right. (D) 1 μM SecABpa399 or SecAΔMBDBpa399 was incubated with 1 μM non-translating 70S ribosomes or 1 μM arrested RNCs containing nascent SecM (SecM-RNCs) and exposed to light at 365 nm. The positions of full-length SecA and the SecA-uL29 crosslinking adduct are indicated. In (C and D), samples were resolved using SDS-PAGE and probed by western blotting using anti-SecA antiserum. LC-MS/MS analysis of the high-molecular weight bands produced by SecABpa399 and SecAΔMBDBpa399 in the presence of vacant 70S ribosomes indicated that they contained both SecA and ribosomal protein uL29.

https://doi.org/10.7554/eLife.48385.012

Crosslinking of SecA and SecAΔMBD to RNCs

To investigate the effect of a nascent chain on binding of SecA to the ribosome, we incubated SecABpa399 with RNCs containing arrested nascent SecM or maltose binding protein (MBP) (Figure 3D and Figure 3—figure supplement 1). The presence of a nascent chain long enough to interact with SecA inhibited crosslinking of SecABpa399 to uL29 (Huber et al., 2017), but the presence of an arrested nascent chain that is too short to interact with SecA did not significantly affect crosslinking to uL29 (Figure 3—figure supplement 2). These results suggest that binding to nascent polypeptide causes a conformational change in SecA, which affects its interaction with the ribosome. In contrast, the presence of nascent substrate protein did not affect crosslinking of SecAΔMBD Bpa399 to uL29 (Figure 3D), consistent with the inability of SecAΔMBD to bind to nascent chains.

Effect of truncations on affinity for free polypeptides

We next examined the affinity of SecA, SecAΔMBD and SecAΔCTT for free polypeptide. To this end, we determined the affinity of SecA for a short peptide, VipB, which was labelled with an environmentally sensitive fluorophore (IAANS-VipB; Pietrosiuk et al., 2011) that produces an increase in fluorescence upon binding to SecA. The affinities of SecA and SecAΔCTT for IAANS-VipB (KD = 0.9 μM and 1.7 μM, respectively) were consistent with the previously reported affinity of SecA for unfolded substrate protein (Gouridis et al., 2009) (Figure 2C and Table 1). However, the affinity of SecAΔMBD for IAANS-VipB was significantly lower (KD = 5.9 μM). Furthermore, alanine substitutions in two of the metal-coordinating cysteines (SecAC885A/C887A) greatly reduced the affinity of SecA for IAANS-VipB (Table 1), suggesting that disrupting the structure of the MBD was sufficient to cause this decrease in affinity.

Effect of truncations on the ATPase activity of SecA

To investigate the effect of the truncations on the ATPase activity of SecA, we determined the basal ATPase rates of SecA, SecAΔMBD, SecAΔCTT and SecAC885A/C887A. The ATP turnover rate for full-length SecA was 0.05 s−1 (Table 1), consistent with previously reported figures (Huber et al., 2011). Deletion of the entire CTT caused a > 10 fold increase in the basal ATPase activity compared to the full-length protein (0.9 s−1) (Table 1), suggesting that the FLD inhibits the ATPase activity of SecA. SecAΔMBD and SecAC885A/C887A did not hydrolyse ATP at a detectable rate, suggesting that the MBD is required to relieve the FLD-mediated autoinhibition.

Effect of truncations on the folding of SecA

We next investigated the effect of the C-terminal truncations on the secondary structure content and thermal stability of SecA using circular dichroism (CD) spectroscopy. The CD spectra of the three proteins indicated that they were fully folded (Figure 2—figure supplement 1). However, the denaturation midpoint temperature (TM) of SecAΔMBD (42°C) was ~1.5°C higher relative to that of the full-length protein and ~2°C higher than that of SecAΔCTT (Figure 2—figure supplement 2), suggesting that SecAΔMBD was more stably folded than SecA or SecAΔCTT.

SecA truncation variants have differing abilities to complement the growth defect of a ΔsecA mutation

To investigate the effect of the C-terminal truncations on the function of SecA in vivo, we constructed strains in which the sole copy of the secA gene produced SecA, SecAΔMBD or SecAΔCTT under control of an IPTG-inducible promoter. Because SecA is required for viability, growth of these strains was dependent on the activity of the SecA variant in vivo. All three alleles complemented the viability defect caused by the ΔsecA mutation (Figure 2D) in an IPTG-dependent fashion and produced similar amounts of SecA (Figure 2—figure supplement 3), indicating that the truncated proteins were functional in vivo. SecAΔCTT and SecAΔMBD were not viable when incubated at room temperature, consistent with the cold-sensitive growth defect of secB mutant strains (Shimizu et al., 1997). However, cells producing SecAΔMBD grew poorly even at the permissive temperature (Figure 2D), consistent with the idea that truncation of the MBD alone inhibited the activity of SecA.

Autocrosslinking of the FLD in the substrate binding groove of SecA

In order to affect such a range of activities of SecA, we reasoned that the CTT likely interacts with the catalytic core. To investigate this possibility, we incorporated Bpa into the CTT at positions 852, 893 and 898. In order to distinguish between early termination products and full-length SecA, we fused a short polypeptide tag to the C-terminus of SecA, which causes SecA to be biotinylated in vivo (SecA-biotin) (Tagwerker et al., 2006; Huber et al., 2011). In addition, we fused hexahistidine-tagged SUMO to the N-terminus of SecA. Ni-affinity purified protein containing Bpa at position 852 (SecABpa852-biotin) migrated more rapidly than the other proteins in SDS-PAGE (Figure 4A), and purified SecABpa852-biotin interacted with streptavidin indicating it contains the C-terminal biotin (Figure 4B). In addition, purification of SecABpa852-biotin from cell lysates by the C-terminal biotin tag yielded proteins that migrated with molecular weights consistent with full-length SecA-biotin and the faster-migrating species (Figure 4—figure supplement 1). These results were consistent with the notion that the faster migrating SecABpa852-biotin species was the result of an internal crosslink within the protein and not early termination at position 852. The chemical basis for the high efficiency of crosslinking is unknown, but several possible explanations are treated in the Discussion section. Purified SecABpa852-biotin had a very low affinity for substrate protein and no detectable ATPase activity (Table 1), suggesting that SecABpa852-biotin occupies a conformation similar to that of SecAΔMBD.

Figure 4 with 1 supplement see all
Auto-crosslinking of the CTT to the catalytic core of SecA.

(A and B) 1 μM SUMO-tagged SecA-biotin containing Bpa at position 852, 893 or 898 in the CTT was incubated in the absence (-) or presence (+) of UV light at 365 nm. The protein samples were resolved using SDS-PAGE and visualised by (A) Coomassie staining or (B) western blotting against the C-terminal biotin tag. The positions of full-length SUMO-SecA is indicated. (C) Mass spectra of wild-type SecA-biotin (above, blue) and SecABpa852-biotin (below, red) in the region of 2450–2750 Da region. Wild-type SecA-biotin and SecABpa852-biotin were exposed to light at 365 nm and subsequently digested with trypsin. The masses of the tryptic fragments were determined using MALDI-TOF. (D) Structure of SecA (2VDA [Gelis et al., 2007]). The main body of the catalytic core is coloured blue, the PPXD is coloured cyan and the tryptic peptide that crosslinks to position 852 (amino acids 361–382) is highlighted in orange. The structural model was rendered using Chimera v. 1.12 (Pettersen et al., 2004).

https://doi.org/10.7554/eLife.48385.015

To investigate the site of the internal crosslink, we determined the molecular weights of the tryptic peptides of SecABpa852-biotin using mass spectrometry (MALDI-TOF). Tryptic fragments with masses greater than 860 Da were resolvable in the mass spectrum of both full-length SecA-biotin and SecABpa852-biotin (Supplementary file 1). Only one peptide in this size range, corresponding to amino acids 168–188, was absent from both spectra. As expected, the tryptic peptide containing position 852 (851-877) was absent from the mass spectrum of SecABpa852-biotin but not SecA-biotin. The only peptide absent from SecABpa852-biotin spectrum but present in wild-type SecA-biotin was the peptide corresponding to amino acids 361–382 (Figure 4C and Supplementary file 1). These results suggested that position 852 likely crosslinked to the region of SecA containing amino acids 361–382. Despite repeated attempts, the crosslinking adduct between peptides 851–877 and 361–382 could not be detected. However, this crosslinking adduct would be very large and would likely consist of a heterogeneous mixture of crosslinked peptides in different conformations. Both of these possibilities could have complicated detection of the adduct by mass spectrometry. Amino acids 361–382 are located in one of the two strands linking the PPXD to NBD1 in the groove where SecA binds to substrate protein (Figure 4D) (Cranford-Smith and Huber, 2018). Crosslinking of Bpa at 852 to this peptide would be consistent with previous work suggesting that the FLD binds in the substrate-binding groove (Hunt et al., 2002; Gelis et al., 2007).

Structural analysis of the SecA truncation variants

We next determined the crystal structure of SecAΔMBD at 3.5 Å resolution (6GOX; Figure 5A and Supplementary file 2). SecAΔMBD crystallised as a symmetric dimer in a head-to-tail configuration (Figure 5A). This structure was similar to that reported for the E. coli SecA homodimer in complex with ATP (Papanikolau et al., 2007) (PDB file 2FSG), except that (i) the PPXD is better resolved in 6GOX and (ii) the 6GOX dimer is symmetric and the 2FSG dimer is not. Consistent with previous studies, the structure of the PPXD was less well defined relative to the other domains of the catalytic core, consistent with the idea that the PPXD is structurally mobile (Zimmer and Rapoport, 2009; Gold et al., 2013). However, because the FLD was not resolved, its effect on the structure of SecA could not be determined.

Figure 5 with 1 supplement see all
SAXS analysis of SecA truncation variants.

(A) X-ray crystal structure of SecAΔMBD at 3.5 Å solved by molecular replacement. The main body of the catalytic core in the asymmetric unit (Protomer 1) is coloured orange with the PPXD highlighted in cyan. The crystallographic mate (Protomer 2) interacts with promoter one using an interface similar to that found in 2FSG (Papanikolau et al., 2007), suggesting that this is the dimer interface of the purified protein in solution. The position of the most C-terminal residue that could be resolved (proline 834) is noted with an asterisk in the right panel. (B–E) Overlay of 10 independent structural models of SecA (B, C), SecAΔMBD (D) and SecAΔCTT (E) generated from fitting to the SAXS data using CORAL. The main body of the catalytic core is coloured grey, and the flexible residues are not displayed. (B, D, E) To facilitate visualization of the asymmetry in the in the dimeric models, both protomeric partners of the dimer were overlaid and the PPXD was coloured (blue/magenta) according to the protomer. The MBD is not displayed in panel B. (C) To facilitate visualization of the position of the MBD in the full-length protein, both protomeric partners of the dimer were overlaid and the MBD of the dimer pair that was located nearest to position 596 of the depicted protomer (orange) was displayed. In panel C, the PPXDs of two protomers, which occupy the same space as the MBDs, are not displayed. (F) Plot of the position of the PPXD in partners of the SecA dimer predicted by structural modelling. The distance between the α-carbon of amino acid 314, which is located near the centroid of the PPXD, and amino acid 596 in NBD2 was determined for each protomer and plotted against the distance in the second protomer. SecA, black circles (FL); SecAΔMBD, orange triangles (ΔMBD); SecAΔCTT, blue squares (ΔCTT). The grey diagonal line indicates the position of the distances if the dimers were symmetric. χ2 values to the diagonal were calculated and used to determine p-values to test whether the asymmetry in the dimer was statistically significant.

https://doi.org/10.7554/eLife.48385.017

To investigate how the CTT affects the conformation of SecA in solution, we investigated the structures of SecA, SecAΔCTT and SecAΔMBD using small-angle x-ray scattering (SAXS) (Supplementary file 3). The SAXS spectra for all three proteins were similar in the low-q region, indicating that the overall shapes of the three proteins were similar, and the radii of gyration suggested that they were dimeric, consistent with previous studies (Woodbury et al., 2002) (Figure 5—figure supplement 1). However, the spectra of the proteins diverged in the mid-q region (Figure 5—figure supplement 1, arrow), indicating that there were differences in domain organisation. SecA has been crystallised in several distinct dimer configurations. The physiological configuration of the dimer and its relevance is an issue of on-going dispute (see discussion in Cranford-Smith and Huber, 2018). However, fitting of structural models of the E. coli SecA dimer based on PDB files 2FSG (Papanikolau et al., 2007), 2IBM (Zimmer et al., 2006), 1M6N (Hunt et al., 2002), 1NL3 (Sharma et al., 2003), 2IPC (Vassylyev et al., 2006) and 6GOX indicated that under the experimental conditions, the conformation of the dimer was similar to that found in 2FSG (χ2 = 3.66) and 6GOX (χ2 = 5.25) (Supplementary file 4).

To gain insight into the structural differences between the three proteins, we modelled the SAXS data by structural fitting. Because initial fitting runs indicated that the CTT was in close proximity to the catalytic core in models of full-length SecA and SecAΔMBD, we fixed the position of the FLD in subsequent fits so that it was consistent with the Bpa crosslinking results. The resulting models suggested that the PPXD was positioned considerably closer to NBD2 (i.e. more ‘open’) in SecA and SecAΔMBD than in SecAΔCTT (p=2.0×10−5 and 1.1 × 10−7, respectively) (Figure 5F). In models of SecAΔMBD and SecAΔCTT, the PPXDs in the two protomers of the dimers were positioned asymmetrically (p=1.8×10−8 and 0.0085, respectively) (Figure 5B–D,F). Finally, in models of full-length SecA, the MBD was positioned between NBD2 and the C-terminal portion of the HSD (amino acids 756–832) in both protomers of the dimer (Figure 5E). Localisation of the MBD to this region would position it directly adjacent to the ribosome-binding surface on the catalytic core (Huber et al., 2011; Singh et al., 2014).

Discussion

Our results indicate that the CTT controls the conformation of SecA and regulates its activity. Disruption of the MBD alone (i) increases the affinity of SecA for the ribosome, (ii) decreases the affinity of SecA for substrate protein, (iii) inhibits the ATPase activity of SecA, (iv) increases the thermal stability of SecA, (v) prevents SecA from undergoing a conformational change upon binding to nascent substrate protein and (vi) causes a defect in SecA function in vivo. However, disruption of both the MBD and the FLD results in a protein that behaves very similarly to full-length SecA, indicating that the FLD mediates these effects. Chemical crosslinking and structural modelling of the SAXS data for wild-type SecA suggest that the FLD interacts with the catalytic core (potentially binding in the substrate protein binding groove) and causes a conformational change in the PPXD. Gold et al. (2013) have suggested that opening of the PPXD when SecA is bound to substrate protein (i.e. enclosing the substrate protein within the binding groove by the PPXD clamp) activates the ATPase activity of SecA. Our work suggests that enclosure of the FLD by the PPXD has the opposite effect—that is, autoinhibition of SecA.

We propose that the MBD is the key for unlocking this autoinhibited conformation in the full-length protein. Previous work suggests that interaction of the MBD with SecB increases the affinity of SecA for polypeptides (Gelis et al., 2007). Our results raise the possibility that binding of the MBD to the ribosome activates SecA in a similar fashion. The absence of the MBD does not cause a strong defect in binding of SecA to ribosomes (indeed, SecAΔMBD has a higher affinity for ribosomes than full-length SecA and SecAΔCTT), and our results suggest that binding of the catalytic core to the ribosome would place the MBD in an ideal position to bind to the ribosomal surface. Thus, although the affinity of the MBD in isolation for the ribosome is relatively low, binding of the catalytic core could trigger binding of the MBD to the ribosome in the context of the full-length protein.

Taken together, our results allow us to propose a mechanism for the recognition of nascent substrate proteins by SecA (Figure 6): (i) interaction of the MBD with the ribosomal surface upon binding of the catalytic core of SecA to the ribosome destabilises the interaction between the FLD and the catalytic core; (ii) destabilisation of the FLD allows SecA to sample nascent polypeptides; (iii) the stable interaction of SecA with nascent substrate protein displaces the FLD from the substrate binding groove; and (iv) binding of SecA to nascent substrate protein causes a conformational change in SecA that leads to release from the ribosome.

Diagram of the proposed mechanism for recognition of nascent substrate proteins by SecA.

(a) In solution, SecA occupies an autoinhibited conformation with the FLD bound stably in the substrate protein binding site and the PPXD in the open conformation. (b) Binding of both the catalytic core and the MBD to the ribosomal surface causes the PPXD to shift to the open conformation, which destabilises binding of the FLD and allows SecA to sample nascent polypeptides. (c) Binding to the nascent substrate protein displaces the FLD from the substrate protein binding site and the PPXD returns to the open conformation, stabilising this interaction. Binding to nascent substrate releases SecA from the ribosomal surface.

https://doi.org/10.7554/eLife.48385.019

The physiological role of CTT-mediated autoinhibition is not yet known. One possibility is that autoinhibition prevents the spurious interaction of SecA with non-substrate proteins by only allowing it to interact with polypeptides in the presence of its ligands (e.g. translating ribosomes, SecB and potentially phospholipids/SecYEG) (Breukink et al., 1995; Gelis et al., 2007; Fekkes et al., 1999). Indeed, overproduction of substrate polypeptides causes a translocation defect in vivo (Wagner et al., 2007; Müller et al., 1989; Oliver and Beckwith, 1982), indicating that SecA can be overwhelmed by interactions with too many substrate proteins. In addition, the spurious translocation of cytoplasmic proteins can be toxic (van Stelten et al., 2009; Emr et al., 1978). Alternatively, the structure of the MBD could regulate the activity of SecA in response to physiological stress. For example, research by the Huber group suggests that the physiological ligand of the MBD is iron (BIORXIV/2019/613315). It is possible that the structure of the MBD is regulated in response to iron limitation or the redox state of the bound metal. If so, the partial activity of SecAΔMBD in vivo suggests that the CTT modulates the activity of SecA rather than inhibiting it completely.

Our results suggest that SecABpa852-biotin produces auto-crosslinks very efficiently. At least three factors could contribute to the high efficiency of auto-crosslinking in SecABpa852-biotin: (i) the amount of time the benzophenone group of the Bpa is in contact with the target molecule, (ii) the chemical reactivity of Bpa toward the target molecule, and (iii) the amount of time the benzophenone group stays in the activated state. First, the results of this study and others (Gelis et al., 2007) is consistent with the idea that the FLD is stably bound in the substrate protein-binding groove of SecA, which should result in a long-lived contact between position 852 and the amino acids lining the substrate binding groove. Second, although benzophenone can, in theory, react with any C-H bond, in practice it reacts with different efficiencies toward different amino acid side chains (Wittelsberger et al., 2006; Lancia et al., 2014). Finally, the chemical environment surrounding a benzophenone group (e.g. hydrophobicity, pH, etc.) can influence its photo-reactive properties (Barsotti et al., 2015; Barsotti et al., 2017). Because Bpa is typically incorporated at surface-exposed positions in order to capture protein-ligand interactions, these environmental effects are normally negligible. However, the hydrophobic environment surrounding the side chain of position 852 when bound in the substrate binding groove could have a significant effect on its reactivity.

The basic features of the catalytic core of SecA are highly conserved amongst bacteria, but different bacterial species contain a diverse array of loops and extensions. For example, our phylogenetic analysis indicated that many species contain alternative CTTs with structures that are significantly different from that of E. coli. These differences could allow SecA to be regulated in response to interaction with a different subset of interaction partners. Nonetheless, many of these alternative CTTs are highly positively charged (e.g. those of many Actinobacteria), suggesting that they may retain the interaction with the ribosome. Some phylogenetic groups (e.g. the Cyanobacteria) lack a CTT entirely. However, most of these species contain large loops in between the conserved elements of the catalytic core of the protein. Indeed, E. coli SecA also contains a ‘variable’ subdomain in NBD2 (amino acids ~ 519–547), which has been proposed to regulate its activity (Das et al., 2012). It is possible that the large loops in between the conserved features of the catalytic core could function analogously to the CTT in E. coli SecA.

Materials and methods

Key resources table
Reagent type
(species) or
resource
DesignationSource or
reference
IdentifiersAdditional
information
Strain, strain background (Escherichia coli K-12)MC4100Casadaban, 1976F- araD139 DlacU169 rpsL150 thi rbsR
Strain, strain background (Escherichia coli K-12)DRH1119This paperMC4100 ΔsecA::KanR λatt::placUV5-secA
Strain, strain background (Escherichia coli K-12)DRH1120This paperMC4100 ΔsecA::KanR λatt::placUV5-secAΔMBD
Strain, strain background (Escherichia coli K-12)DRH1121This paperMC4100 ΔsecA::KanR λatt::placUV5-secAΔCTT
Strain, strain background (Escherichia coli K-12)DRH663This paperMC4100 ΔsecA::KanR + pDH663
Strain, strain background (Escherichia coli B)BL21(DE3)Lab stockEscherichia coli Genetic Stock Center (CGSC), Yale University, USA. CGSC#: 12504 http://cgsc2.biology.yale.edu/Strain.php?ID=139459MC4100 ΔsecA::KanR + pDH663
Strain, strain background (Escherichia coli B)DRH584This paperBL21(DE3) containing plasmid pDH584
Strain, strain background (Escherichia coli B)DRH1166This paperBL21(DE3) containing plasmid pDH1166 and pSup-Bpa-6TRN
Genetic reagent (phage λ)lambda InChBoyd et al., 2000
AntibodyRabbit anti-SecA antiserumHuber et al., 2011(1:20000)
AntibodySheep anti-uL23 antiserumotherGift from R. Brimacombe (1:2500)
AntibodyRabbit anti-thioredoxin-1 antiserumSigma-Aldrich (St. Louis, MO, USA)Catalogue number: T0803(1:10000)
AntibodyGoat IR700-labelled anti-rabbitRockland (Philadelphia, PA, USA)Catalogue number: 611-130-122Discontinued (1:5000)
AntibodyGoat IR800-labelled anti-sheepRockland (Philadelphia, PA, USA)Catalogue number: 613-445-002(1:10000)
AntibodyDonkey HRP-labelled anti-rabbitGE HealthcareCatalogue number: NA934V(1:10000)
Recombinant DNA reagentpCA528Andréasson et al., 2010pET24 expression vector containing gene encoding His-tagged SUMO protein
Recombinant DNA reagentpCA597Andréasson et al., 2010pET24 expression vector containing gene encoding Strep-tagged SUMO protein
Recombinant DNA reagentpDH543This paperpCA597 containing portion of secA gene corresponding to amino acids829–901
Recombinant DNA reagentpDH934This paperpCA597 containing portion of secA gene corresponding to amino acids 875–901
Recombinant DNA reagentpDH625Huber et al., 2011pCA528 containg the secA gene
Recombinant DNA reagentpDH584Huber et al., 2011pCA528 containg the secA-biotin gene
Recombinant DNA reagentpDH1166This paperpDH584 containing amber codon at position corresponding to amino acid 852 in SecA.
Recombinant DNA reagentpDSW204Weiss et al., 1999pTrc99a-derived plasmid containing partially disabled trc promoter
Recombinant DNA reagentpDH692Huber et al., 2011pDSW204-derived plasmid producing SecA under control of an IPTG-inducible promoter
Recombinant DNA reagentpDH939This paperpDSW204-derived plasmid producing SecAΔMBD under control of an IPTG-inducible promoter
Recombinant DNA reagentpDH663Huber et al., 2011pTrc99b-derived plasmid producing SecA under control of an IPTG-inducible promoter and containing a Spectinomycin resistance gene in place of the bla (ampicillin resistance) gene
Recombinant DNA reagentpDH787Huber et al., 2017pCA597 containing full-length secM gene
Recombinant DNA reagentpDH784Huber et al., 2017pCA597 containing secM with internal deletion between region encoding signal sequence and the translation arrest sequence
Recombinant DNA reagentpDH894Huber et al., 2017pCA528 containing gene encoding malE gene that is translationally fused to sequence encoding the SecM translation arrest sequence
Recombinant DNA reagentpSup-Bpa-6TRNChin et al., 2002Plasmid producing orthologous tRNA and tRNA synthetase required for incorporation of Bpa
Peptide, recombinant proteinVipB peptidePietrosiuk et al., 2011
Peptide, recombinant proteinHRP-coupled StreptactinIBA Life Sciences (Goettingen, Germany)Catalogue number: 2-1502-001
Chemical compound4-acetamido-4′-maleimidylstilbene-2,2′-disulfonic acid (AMS)Invitrogen (Carlsbad, California)Catalogue number: A485
Chemical compound, drugHydrophilic streptavidin magnetic beadsNew England Biolabs (Ipswich, Massachusetts)Catalogue number: S1421S
Chemical compound, drugBenzophenylalanine (Bpa)Bachem (Santa Cruz, CA, USA)H-p-Bz-Phe-OH Article number:4017646.0005
Chemical compound, drug1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC)ThermoScientific PierceCatalogue number: 22980
Chemical compound, drugRu(bpy)2(dcbpy)Sigma Aldrich (St Louis, MO, USA)Product 96632Discontinued
 Software, algorithmATSAS v2.8.3European Molecular Biology Laboratory (EMBL) Hamburghttps://www.embl-hamburg.de/biosaxs/software.html
Software, algorithmPyMol v1.8.0.5Schrödinger Scientifichttps://pymol.org/2/
Software, algorithmGROMACSSchrödinger ScientificPronk et al., 2013
Chemical compound, drugPEG/Ion Screen 2 #39Hampton Research (Aliso Viejo, CA, USA)Product HR2-126
Chemical compound, drugMorpheusMolecular Dimensions (Newmarket, Suffolk, UK)Product MD1-46
OtherSuperose 610/300 GL columnGE HealthcareProduct 17517201Discontinued

Chemicals and media

All chemicals were purchased from Sigma-Aldrich (St. Louis, MO, USA) unless otherwise indicated. Anti-thioredoxin-1 antiserum was purchased from Sigma-Aldrich. Rabbit anti-SecA antiserum was a laboratory stock. Sheep anti-uL23 antiserum was a kind gift from R. Brimacombe. IR700-labelled anti-rabbit and IR800-labelled anti-sheep antibodies were purchased from Rockland (Philadelphia, PA). Horseradish peroxidase (HRP)-labelled anti-rabbit antibody was purchased from GE Healthcare. HRP-coupled streptactin was purchased from IBA Lifesciences (Goettingen, Germany). Bpa was purchased from Bachem (Santa Cruz, CA). Strains were grown in lysogeny broth (LB) containing kanamycin (30 μg/ml) or ampicillin (200 μg/ml) as required and in the concentration of isopropyl-thiogalactoside (IPTG) indicated.

Strains and plasmids

Strains and plasmids were constructed using standard methods (Miller, 1992; Sambrook and Russell, 2001). For protein-expression plasmids, the DNA encoding full-length SecA or fragments of SecA were amplified by PCR and ligated into plasmid pCA528 (His6-SUMO) or pCA597 (Strep3-SUMO) using the BsaI and BamHI restriction sites (Andréasson et al., 2008). UAG stop codons were introduced into plasmids expressing His-SUMO-SecA or His-SUMO-SecA-biotin using QuikChange (Agilent). Plasmid pSup-Bpa-6TRN was a kind gift from P Schultz. Strains DRH1119, DRH1120 and DRH1121 were constructed by cloning secA genes producing full-length SecA, SecAΔMBD and SecAΔCTT into pDSW204 (Weiss et al., 1999) and then introducing them onto the chromosome strain DRH663 (MC4100 ΔsecA::KanR + pTrcSpc-SecA) (Huber et al., 2011) using lambda InCh (Boyd et al., 2000). The pTrcSpc-SecA plasmid was then cured from the strain by plating on LB containing 1 mM IPTG. All three strains required >10 μM IPTG for growth on LB.

Phylogenetic analysis

The sequences of SecA for the given UniProtKB entry names (The UniProt Consortium, 2017) were analysed using ClustalOmega (McWilliam et al., 2013). The unrooted phylogenetic tree was rendered using iTOL (Ciccarelli et al., 2006). The logo of the consensus MBD sequence was generated using WebLogo (https://weblogo.berkeley.edu/logo.cgi).

Ribosome and protein purification

Ribosomes and arrested RNCs were purified as previously described (Rutkowska et al., 2008; Huber et al., 2011). SecA was purified as described previously (Huber et al., 2017). BL21(DE3) (laboratory stock) or BL21(DE3) containing plasmid pSup-Bpa-6TRN (Chin et al., 2002) was transformed with the appropriate plasmid and grown in LB in the presence of kanamycin at 37°C to OD600 ~1, induced using 1 mM IPTG and shifted to 18°C overnight. Cells were then harvested by centrifugation and lysed by cell disruption in buffer 1 (50 mM K·HEPES, pH 7.5, 500 mM NaCl and 0.5 mM TCEP [tris(2-carboxyethyl)phosphine]) containing cOmplete EDTA-free protease inhibitor cocktail (Roche). Unlabelled His-tagged proteins were affinity purified by passing over a 5 ml Ni-NTA HiTrap column (GE Healthcare), washed with buffer containing 50 mM imidazole and eluted from the column in buffer containing 250 mM imidazole. The eluted protein was cleaved with the SUMO-protease Ulp1 and the SUMO moiety was removed by passing over a 5 ml Ni-NTA HiTrap column. The partially purified protein was then concentrated (Centricon) and purified by size exclusion chromatography using a sepharose S-200 column (GE Healthcare). Bpa-labelled proteins were purified as described by Huber et al. (2017). For Strep-tagged proteins, lysates from cells producing SUMO-CTT and SUMO-MBD were passed over streptactin-coupled sepharose beads (IBA Lifesciences), washed extensively with buffer 2 (20 mM K·HEPES, pH 7.5, 100 mM potassium acetate, 10 mM magnesium acetate) and eluted using buffer 2 containing 10 mM desthiobiotin. SUMO-CTT was modified with 4-acetamido-4′-maleimidylstilbene-2,2′-disulfonic acid (AMS) by incubating 75 μM SUMO-CTT with 500 μM AMS in buffer 2 for 30 min on ice. AMS labelling was terminated by the addition of 500 μM β-mercaptoethanol. Efficient labelling was confirmed by the increased mobility of the modified protein in SDS-PAGE. For purification of SUMO-SecABpa852-biotin by its C-terminal biotin tag, lysates of cells producing SUMO-SecA-biotin and SUMO-SecABpa852-biotin were incubated with 50 μl hydrophilic streptavidin magnetic beads and washed five times with 1 ml buffer 2. The bound protein was eluted from the beads by boiling in 50 μl 1X Laemmli buffer and resolved on a 12% BioRad Stain-free Ready gel.

Western blotting

Western blots were carried out as previously described (Sambrook and Russell, 2001). Protein samples were resolved using ‘Any kD’ SDS-PAGE gels (BioRad) and transferred to nitrocellulose membranes. Membranes were probed using the indicated primary and secondary antisera or with HRP-streptactin. For HRP-based detection, membranes were developed using ECL (GE Healthcare) and visualised using a BioRad Gel-Doc. For IR700- and IR800-based detection, membranes were visualised using a LI-COR Odyssey scanner.

Ribosome cosedimentation

Ribosome cosedimentation experiments were carried out as previously described (Huber et al., 2017). Binding reactions were incubated in buffer containing 10 mM HEPES potassium salt, pH 7.5, 100 mM potassium acetate, 10 mM magnesium acetate, 1 mM β-mercaptoethanol for >10 min. The reaction mixture was then layered on top of a 30% sucrose cushion made with the same buffer and centrifuged at >200,000 x g for 90 min. The supernatant was discarded. The concentration of ribosomes in the pellet fractions were normalised using the absorbance at 260 nm.

Fluorescence anisotropy

The KD of the SecA-ribosome complex by fluorescence anisotropy was determined as previously described (Huber et al., 2011). SecA, SecAΔMBD and SecAΔCTT were labelled with Ru(bpy)2(dcbpy) and the fluorescence anisotropy was measured on a Jasco FP-6500 fluorometer containing an ADP303 attachment using an excitation wavelength of 426 nm (slit width 5 nm) and an emission wavelength of 640 nm (slit width 10 nm).

CD spectroscopy

The CD spectra of 2 µM solutions of full-length SecA, SecAΔMBD, or SecAΔCTT in 10 mM potassium phosphate buffer (pH 7.5) were measured at temperatures that promote folding (10°C) and denaturation (85°C) in a 0.1 cm cuvette using a Jasco J750 CD spectrometer. For thermal titrations, the temperature was raised 0.5 K/min from 30°C to 50°C and circular dichroism was measured at 222 nm.

Peptide binding

600 nM VipB peptide labelled with IAANS (Pietrosiuk et al., 2011) was incubated with increasing concentrations of SecA or the indicated SecA variant. The increase in IAANS fluorescence upon binding of SecA was measured using a Jasco FP-6500 fluorometer or a BMG Labtech CLARIOStar.

ATPase assays

ATPase activities were determined by measuring the rate of NADH oxidation in a coupled reaction (Kiianitsa et al., 2003). 1 μM SecA, or the respective SecA variant, was added to a solution containing 250 mM NADH, 0.5 mM phosphoenolpyruvate, 2 mM ATP, 20/ml lactate dehydrogenase, 100 U/ml pyruvate kinase and incubated at 25°C, 50 mM K·HEPES, pH 7.5 and 500 mM NaCl. The decrease in absorbance at 340 nm from the oxidation of NADH to NAD+ was measured using an Anthos Zenyth 340rt (Biochrom) absorbance photometer equipped with ADAP software. The rate of ATP hydrolysis was determined from the rate of NADH oxidation by dividing the rate of decrease in the absorbance by the extinction coefficient for NADH (6220 M−1 cm−1).

Chemical crosslinking

Non-specific crosslinking of SUMO-CTT to the ribosome using 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) were carried out as described previously (Huber et al., 2011). Site specific crosslinking of SecA containing Bpa was carried out as described by Huber et al. (2017).

Mass spectrometry

Auto-crosslinked samples were digested with sequencing grade trypsin and the masses of the tryptic peptides were determined using MALDI-TOF mass spectrometry. The identity of ribosomal crosslinking adducts was determined by excising the protein band from a Coomassie-stained gel and analysing the tryptic peptide fragments using liquid chromatography-tandem mass spectrometry (LC-MS/MS) identification (The Advanced Mass Spectrometry Facility, School of Biosciences, University of Birmingham).

X-ray crystallography

SecAΔMBD was crystallised by mixing 2 μl of purified protein (100 μM) with 2 μl of a 9:1 mixture of PEG/Ion 2 Screen # 39 buffer (0.04M Citric acid, 0.06M BIS-TRIS propane, pH6.4, 20% Polyethylene glycol 3,350) (Hampton Research, Aliso Viejo, CA, USA) and Morpheus condition 47, box 2 (0.1M Amino acids, 0.1M Buffer system 3, pH 8.5, 50% v/v Precipitant Mix 3) (Molecular Dimensions, Newmarket, UK) in a 48-well MRC MAXI plate (Molecular Dimensions, Newmarket, UK). Crystals appeared within six days and were fully matured by two weeks. Crystals were analysed at Diamond light source, and the structure was solved by molecular replacement using PDB file 2FSG at 3.5 Å resolution (Figure 1—source data 1). The structure was deposited at RCSB under PDB file 6GOX.

SAXS measurements

Synchrotron radiation X-ray scattering data were collected on the ESRF BM29 BioSAXS beamline (Grenoble) (Figure 1—source data 2). An in-line Superose 6 10/300 GL column (GE Healthcare) was used to ensure that the protein was free from aggregates and that it occupied a single oligomeric state during data collection. The sample-to-detector distance was 3 m, covering a range of momentum transfer s = 0.03–0.494 Å−1 (s = (4π·sin θ)/ λ, where 2θ is the scattering angle, and λ = 0.992 Å is the X-ray wavelength). Data from the detector were normalised to the transmitted beam intensity, averaged, placed on absolute scale relative to water and the scattering of buffer solutions subtracted. All data manipulations were performed using PRIMUSqt and ATSAS (Petoukhov et al., 2012). The forward scattering I(0) and radius of gyration, Rg were determined by Guinier analysis. These parameters were also estimated from the full scattering curves using the indirect Fourier transform method implemented in the program GNOM, along with the distance distribution function p(r) and the maximum particle dimensions Dmax. Molecular masses of solutes were estimated from SAXS data by comparing the extrapolated forward scattering with that of a reference solution of bovine serum albumin. Computation of theoretical scattering intensities was performed using the program CRYSOL. SAXS data has been deposited at the SASBDB (www.sasbdb.org) with accession codes: SASDDY9 (full-length SecA), SASDDZ9 (SecAΔMBD) and SASDE22 (SecAΔCTT).

Molecular modelling of SAXS data

For modelling based on SAXS data, multiple fits were performed to verify the stability of the solution, and to establish the most typical 3D reconstructions using DAMAVER. Guinier analysis of the SAXS data indicated that the protein was dimeric under the conditions used for SAXS. Structural models of the E. coli SecA dimer were generated by aligning the structure of SecA in PDB file 2VDA to PDB files 2FSG, 2IBM, 2IPC, 1M6N, 1NL3 and 6GOX using PyMol v. 1.8.0.5 and refining using GROMACS (Pronk et al., 2013). Because the CTT is not resolved in the structures used for modelling, these models were fitted to the SAXS spectrum of SecAΔCTT using FoXS (Schneidman-Duhovny et al., 2016) (Supplementary file 1). The structures of full-length SecA, SecAΔMBD and SecAΔCTT were modelled by fitting the 2FSG dimer to the respective SAXS data by multi-step rigid body refinement using CORAL (Petoukhov et al., 2012). The positions of NBD1, NBD2, HSD and HWD were fixed in all models. The regions corresponding to residues 1–8, 220–231, 367–375 were defined as linkers and modelled as flexible. The PPXD was allowed rigid body movement in all three models. The FLD (residues 829–832 in SecAΔCTT and 829–880 in SecAΔMBD and full-length SecA) were modelled as flexible. For full length SecA, the MBD was modelled using PDB file 1S × 0 and allowed rigid-body movement. Because initial modelling indicated that the FLD was in close contact with the catalytic core, and because photocrosslinking indicated the FLD was bound in the substrate binding groove, residues 851–854 were modelled to form a small β-sheet with residues 222–225 and 373–375 and allowed rigid body movement. All ten independently generated fits for SecA and SecAΔMBD produced plausible structural models (χ2 = 0.97 ± 0.02 and 1.87 ± 0.09, respectively). Ten of 15 of the fits of SecAΔCTT produced plausible structural models (χ2 = 1.82 ± 0.18). In the remaining five fits, the position of the PPXD in one of the two protomers was inconsistent with previously published structures of SecA and occupied a non-realisitic conformation, suggesting increased mobility of the PPXD in SecAΔCTT.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
    The way is the goal: how SecA transports proteins across the cytoplasmic membrane in bacteria
    1. T Cranford-Smith
    2. D Huber
    (2018)
    FEMS Microbiology Letters, 365, 10.1093/femsle/fny093, 29790985.
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
    SecA cotranslationally interacts with nascent substrate proteins in vivo
    1. D Huber
    2. M Jamshad
    3. R Hanmer
    4. D Schibich
    5. K Döring
    6. I Marcomini
    7. G Kramer
    8. B Bukau
    (2017)
    Journal of Bacteriology, 199, 10.1128/JB.00622-16, 27795329.
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
    A Short Course in Bacterial Genetics
    1. JH Miller
    (1992)
    Cold Spring Harbor: Cold Spring Harbor Press.
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
    Molecular Cloning: A Laboratory Manual (Third Edition)
    1. J Sambrook
    2. DW Russell
    (2001)
    Cold Spring Harbor: Cold Spring Harbor Press.
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
    Localization of FtsI (PBP3) to the septal ring requires its membrane anchor, the Z ring, FtsA, FtsQ, and FtsL
    1. DS Weiss
    2. JC Chen
    3. JM Ghigo
    4. D Boyd
    5. J Beckwith
    (1999)
    Journal of Bacteriology 181:508–520.
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59

Decision letter

  1. Ramanujan S Hegde
    Reviewing Editor; MRC Laboratory of Molecular Biology, United Kingdom
  2. John Kuriyan
    Senior Editor; University of California, Berkeley, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

[Editors’ note: a previous version of this study was rejected after the first round of revisions, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]

Thank you for submitting your article "The C-terminal tail of the bacterial translocation ATPase SecA modulates its activity" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and John Kuriyan as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

There are two main issues to be addressed experimentally:

1) Experiments to determine whether and which divalent metal(s) impact function of the C-terminal domain. A complete analysis is probably beyond the scope of this paper, but initial experiments to examine the metal requirement were judged to be important.

2) Disentangling the role of the nascent chain versus a programmed ribosome by analysis of RNCs containing short nascent chains.

In addition, a number of clarifications, controls, and fuller explanations were requested (see the full reviews appended below). Each of these seem relatively straightforward. I suspect that for many of these, you will already have the answer from experiments not shown in the manuscript, although some might also require repeating experiments with some variations.

Finally, it is worth addressing the issue of the interplay between the SecA-nascent chain interaction studied here as it relates to known interactions of SecA with lipids and SecYEG. Experimental investigation of this issue is beyond the scope of this study, but discussing this issue is worthwhile.

Reviewer #1:

The manuscript submitted by the Huber lab relates to the regulation of the essential motor ATPase responsible for the bulk of bacterial protein secretion. The results build up to a hitherto unknown mechanism for its auto-inhibition and activation by nascent substrates at the ribosome exit site. The submission compiles high quality data, which have been sensibly interpreted, so the story is plausible. It is a story that should certainly be worth considering for those of us interested in SecA, and who have until now completely ignored the regulatory feature of the C-terminus. The paper will also be of interest to a much wider readership, and also for folk interested in the development of antibiotics. Thus, the work could be worthy of publication in eLife if the following points and suggestions can be addressed/considered.

To be addressed:

1) The metal binding site seems to have an important role here. Does it require metals to function? If so, which ones – Zn2+? Probably some/many of the experiments should be compared with and without divalent metals.

2) Figure 3A shows important evidence for interaction of flexible linker with the catalytic core by intra-molecular crosslinking of residue 852. The authors say the SecA-852 band is fully crosslinked, in contrast to 893 and 898, because of its higher apparent mobility. I would like the authors to redo this experiment, but with equal quantities of SecA, just to check that it is not a loading artefact. Also, why is the crosslink not induced by UV? Were the uncrosslinked samples kept in the dark at all times?

3) Concerning the X-ray structure. If the FLD influences the conformation of SecA, as stated, then it should be evident in the high-resolution structure. How does the new structure compare to others determined without the FLD? Did the full-length protein fail to crystallise? This figure could do with some improvement by colouring the different domains and highlighting the location of the C-terminus and FLD. The two monomers of the dimers could also be distinguished somehow.

Reviewer #2:

The study from Jamshad and colleagues investigates the function of the C-terminal domain of SecA, a key factor in protein export in bacteria. The authors show that in E. coli the C-terminal domain of SecA, as well as binding SecB, can also bind to the ribosome. Loss of the entire CTT leads to no obvious growth defect and actually enhances ribosome binding, suggesting that it likely has a regulatory function. Indeed through crosslinking, structural and functional analyses a model emerges where the metal binding domain of the CTT acts to control substrate loading in response to ribosome binding. Overall, the experiments are largely well performed and the study provides new insight into SecA function. Following attention to the points below I would be supportive of publication.

In Figure 1D the SUMU-CTT is in large excess over ribosomes, it would be important to see a 10% input to assess the efficiency of binding. It would also be informative to assess whether SecA lacking the CTT is able to inhibit binding.

In Figure 2B, the effect of the nascent chain is established by comparing empty 70S ribosomes with a stalled SecM RNC. This could reflect changes due to the nascent chain, but also the fact that the ribosomes are now programmed. This could be distinguished using a shorter nascent chain that has not yet emerged from the exit tunnel. Moreover, data from the Wintermeyer group also indicate that interaction of the nascent chain with the uL23 inside the tunnel can also trigger changes in binding of SRP, another exit site ligand, which also interacts via uL23.

In Figure 2D, there is no control to show that SecA is essential in the strain being used (e.g. control strain lacking exogenous SecA with and without IPTG). The writing on the back of the plate is also quite intrusive. It would be essential to show the expression levels of the proteins relative to endogenous levels of SecA. It could be that the mutants are unstable in vivo and this causes the phenotype.

The functional analysis of the mutants would be more complete if their ability to support protein translocation (either in vivo or in vitro) was evaluated.

In Figure 3, there is a loss of two peptides from the MS analysis. Was the crosslinked peptide also detectable? This would strengthen the identification of the two peptides as the crosslinking site.

Was the auto-crosslinked complex tested for structural analysis? This might have stabilised the CTD allowing it to be visualised.

Does the ∆MBD construct with the 852 suppressor also form the auto crosslink? This would strengthen the argument that the locked ∆MBD conformation is a state the wild-type protein also encounters and not a non-physiological dead-end state.

In addition to 852, crosslinkers were also incorporated in the MBD at positions 893 and 898, but it was not clear why these were not further analysed. In particular, it would seem logical to text if these give rise to crosslinking to rProteins as that might permit further pinpoint localisation of the MBD at the ribosome surface or relative to the rest of SecA.

If Figure 5C, what is higher UV-specific band seen just above the SUMO-SecA band with the 209 construct? Is this a crosslink to another ribosomal protein? If so, that might further define the positioning of SecA relative to the ribosome.

In Figure 5D, the crosslink to the ∆MBD seems smaller. Is it established that this is still a crosslink to uL29 and not a different rProtein? This would be important to establish and confirm that ∆MBD binds in the same position as the WT, but cannot reposition upon binding the nascent chain.

In the authors model, peptide binding leads to release of SecA and the substrate from the ribosome. Can this be tested by releasing the nascent chain in the RNC.SecA complex with puromycin?

Reviewer #3:

The authors investigated the functions of SecA's C-terminal tail. Sedimentation and crosslinking assays with the purified motor protein and purified ribosomes or nascent chain ribosome complexes revealed that the metal binding domain binds to the ribosomal surface. The binding event causes the PPXD to shift to the open conformation, thereby destabilizing the flexible linker domain, which – according to small-scale angle scattering experiments – resides in the substrate protein binding groove. In turn, the autoinhibition is released and SecA is able to sample nascent polypeptides.

The manuscript offers new insight into the recognition process between SecA and secretory polypeptides. The approach appears to be sound. Yet both membrane and SecYEG are missing in the emerging picture. That is, SecA's substantial affinity to lipid membranes and SecYEG is not mentioned. How do the newly determined KD values compare to those previously published for the interaction of SecA with lipid membranes and with the translocon? Does such affinity comparison favor SecA interactions with the nascent chain (i) in the cytoplasm or (ii) at the membrane surface?

[Editors’ note: what now follows is the editors’ decision letter after the first round of revisions.]

Thank you for submitting the revised manuscript entitled "The C-terminal tail of the bacterial translocation ATPase SecA modulates its activity" for consideration by eLife. Your revision has been carefully reviewed by a Reviewing Editor and a Senior Editor.

After carefully examining your responses to the editorial and reviewer comments, we find that the revision does not adequately address the concerns that were raised during review. Thus, the primary conclusions of this study are not supported with the level of rigor and depth expected for eLife. A detailed explanation of the concerns is provided below.

Editorial assessment of the author reply to reviewers:

The role of a metal requirement for MBD function was not examined satisfactorily. The sole experiment to address this issue, Figure 1D, is poorly controlled and uninterpretable. First, the signals in lane 7 and lane 2 are essentially indistinguishable (lane 7 is fainter, but the band is spread over a larger area, so the overall difference is hard to appreciate). Without replicates and statistics, one cannot reasonably infer relevance from such a modest difference. Even if one were to take the difference at face value, the effect cannot be attributed to the MBD from the information provided; it could possibly be due to an effect of EDTA on some part of the ribosome. There are no controls to verify that metal was in fact removed from the CTT. The authors have other experimental approaches at their disposal. For example, the purified CTT could be stripped of metal (e.g., with high concentration EDTA), the stripped metal and EDTA removed by dialysis or desalting, the extent of removal verified, and the metal-free CTT (compared to protein in which the metal was added back) used in the assays. Alternatively (or in addition), one could use mutants in the metal binding domain. In short, the issue of a metal requirement was not addressed.

The request to clarify the supposed UV-independent and highly efficient crosslinking from residue 852 was also not addressed adequately. This is a crucial experiment because it is the main direct evidence for the CTT engaging the catalytic core of SecA. The problem with UV-independence is that one cannot be certain the crosslink actually formed, with the alternative interpretation being a migration artefact with no crosslinks. The manuscript partially allays this concern, but the failure to detect a peptide is not especially strong evidence on its own. Based on what is shown, it could even be the case that BpA was not incorporated into 852 (amber suppression efficiency is known to be position-sensitive). While the authors could argue that the biotin blot shows amber suppression, this blot is very over-loaded and there is no negative control of a comparably overloaded SecA lacking the biotin tag. For these reasons, the conclusion that increased migration represents extremely efficient UV-independent intramolecular crosslinking through some unknown mechanism is not sufficiently compelling evidence to support a critical conclusion of the paper. If the authors' claim that this crosslink forms during the purification is correct, they can simply treat intact bacteria without or with UV, then harvest directly into SDS and monitor by blotting for the biotin tag. This should show a clear UV-dependent size shift to a position where the purified protein migrates. Other approaches are also feasible, including positioning BpA at nearby residues where the environment would be different, and hence, avoid constitutive UV-independent crosslinking. As it stands, a key conclusion of the study remains in doubt.

The request to provide the reader some estimate of the efficiency of binding in Figure 1D (first point of reviewer 2) was either ignored or misunderstood. The reviewer was asking simply to run 10% of the input sample on the same gel as the pulldowns so one could assess whether the amount pulled down was more or less than 10% of what went into the reaction. If it is far less, one could run 1% of the input (or whatever is roughly in the suitable range). The point is to provide the reader with a rough idea of how much is being pulled down in the experiment because this impacts how believable the findings are.

Similarly, the request to verify that SecA lacking the CTT fails to compete for binding in Figure 1D was also not addressed (second point of reviewer 2). This is a rather standard control, and it most certainly would affect the conclusions were it to compete similarly to WT SecA.

The reasonable request to test SecA levels in the different mutant strains relative to endogenous levels of SecA was addressed in a rather convoluted manner and did not adequately address the concern. IPTG-induced toxicity is not a good surrogate for expression levels because the basis of this toxicity could be different for different mutants (i.e., for some, it may have to do with SecA function, for others, it might have to do with protein aggregation or inappropriate interactions). It should be a straightforward matter to directly test expression levels by blotting using widely available SecA antibodies. Similarly, the request to directly test protein translocation is both reasonable and straightforward. The concern is not that SecA has some other function, but rather that ∆MBD is having its effect via a dominant toxicity unrelated to its function (which is certainly plausible).

The aberrant migration of the crosslink to ∆MBD relative to the corresponding crosslink to full length SecA in Figure 5D was not addressed. While it is true that ∆MBD migrates faster than wild type, the key concern was that the shift upon crosslinking was far smaller for ∆MBD than for full length, suggesting the possibility that the crosslinking partners are different. In this region of the gel, this difference in size shift is pretty substantial. It is therefore not acceptable to simply assume both crosslinks are to uL29.

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for submitting your article "The C-terminal tail of the bacterial translocation ATPase SecA modulates its activity" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and John Kuriyan as the Senior Editor. The reviewers have opted to remain anonymous.

This revised article generally satisfies the reviewer comments. As you will see below, reviewer 2 feels you should be more cautious in some of your conclusions that are not fully convincing, particularly the intramolecular crosslink. In reading the article and evaluating the revisions myself, I have similar reservations.

Please adjust the text accordingly and submit a final version as per the instructions below.

Reviewer #1:

The manuscript is a new submission of a paper previously submitted to eLife, of which I was a reviewer.

Following review there were substantial (but not major) problems that needed addressing with the support of additional experiments and controls. It turned out that these could not be fully resolved within 2 months and the paper was returned to the authors.

I've now read the cover letter carefully and looked at the relevant sections of the new manuscript, and the JBC submission (available at bioRxiv) concerning the metal binding site. The new data addresses my initial query about the nature of the metal ligand, and consequences of binding.

It seems to me that all other concerns have also been addressed and would be happy to see the manuscript published.

Reviewer #2:

The authors have now addressed most of the major issues I had identified from the first round of reviews. They nicely show that a programmed short nascent chain does not block the crosslink to uL29 strengthening the authors model. They also show the 10% input for the binding assay in Figure 1, this reveals that the level of binding is relatively low, probably 1-2% of the input protein, i.e. binding to around 10-20% of ribosomes. They have also now attempted the suggested competition assay control (for original Figure 1D) but obtained inconclusive results with the ∆CTT construct. I would agree this compromises the interpretation of the competition experiment, hence its removal is sensible. It is puzzling why the assay was so variable considering that the WT and ∆CTT constructs apparently behaved well for the structural analysis.

Protein levels of the in vivo expression of the constructs is now shown and look similar, albeit without a loading control. I still also think it would have been informative to also look at the effects of the SecA mutants on translocation rather than growth alone.

The revised section of the text dealing with the internal crosslink now doesn't actually mention the auto-crosslink formation, save in the section- and figure headings. The section should explain the behaviour of the three constructs +/- UV and then introduce the uv-independent crosslink explanation. Also, as I mentioned previously, and was discussed by the other reviewers, the lack of identification of the crosslinked peptide tempers the definitive confirmation that the faster-migrating band is actually crosslinked. So this remains a slight weak point in the manuscript.

The altered mobility of the ∆MBD-uL29 crosslink is now addressed by the use of mass-spectrometry to confirm its identity.

Reviewer #3:

I have no additional comments to the revised manuscript.

https://doi.org/10.7554/eLife.48385.034

Author response

[Editors’ note: the author responses to the first round of peer review follow.]

There are two main issues to be addressed experimentally:

1) Experiments to determine whether and which divalent metal(s) impact function of the C-terminal domain. A complete analysis is probably beyond the scope of this paper, but initial experiments to examine the metal requirement were judged to be important.

2) Disentangling the role of the nascent chain versus a programmed ribosome by analysis of RNCs containing short nascent chains.

In addition, a number of clarifications, controls, and fuller explanations were requested (see the full reviews appended below). Each of these seem relatively straightforward. I suspect that for many of these, you will already have the answer from experiments not shown in the manuscript, although some might also require repeating experiments with some variations.

Finally, it is worth addressing the issue of the interplay between the SecA-nascent chain interaction studied here as it relates to known interactions of SecA with lipids and SecYEG. Experimental investigation of this issue is beyond the scope of this study, but discussing this issue is worthwhile.

We have revised the manuscript in response to the comments by the reviewers. We feel that these revisions have significantly improved the manuscript, and we thank the reviewers for their very insightful feedback. We also have included a significant amount of new experimental material, which addresses the concerns raised in the decision letter. Our revisions include:

1) Crosslinking experiments indicating that SecABpa399 crosslinks to uL29 on RNCs containing arrested nascent SecM that is too short to interact with SecA (Figure 5—figure supplement 1B). These results suggest that the loss of crosslinking to uL29 in Figures 5D and Figure 5—figure supplement 1A was the result of interaction of SecA with the nascent chain (i.e. was not due to the state of SecM-stalled ribosome).

2) Ribosome binding experiments indicating that the presence of a metal chelator partially inhibits binding of the CTT to the ribosome (Figure 1D). These results suggest that binding of the MBD to its metal cofactor is important for efficient binding to the ribosome.

We did not include experiments addressing the identity of the physiological metal ligand. The identity of the metal cofactor does not affect our conclusions, and the apparent identity of the metal cofactor (zinc) is well established. Furthermore, although we have accumulated a significant amount of evidence suggesting that the physiological metal ligand is actually iron, we feel that challenging the established knowledge merits its own publication. Indeed, we are in the process of preparing this work for publication. A portion of this work is already publicly available as an author pre-print at BioRxiv (doi: 10.1101/173039), and we have discussed the possible implications of this in the revised Discussion section.

The reviewers also asked that we discuss the interplay between the SecA-nascent chain interaction and the already established interaction of SecA with SecYEG and lipids, and we have expanded our discussion to accommodate this request.

Reviewer #1:

The manuscript submitted by the Huber lab relates to the regulation of the essential motor ATPase responsible for the bulk of bacterial protein secretion. The results build up to a hitherto unknown mechanism for its auto-inhibition and activation by nascent substrates at the ribosome exit site. The submission compiles high quality data, which have been sensibly interpreted, so the story is plausible. It is a story that should certainly be worth considering for those of us interested in SecA, and who have until now completely ignored the regulatory feature of the C-terminus. The paper will also be of interest to a much wider readership, and also for folk interested in the development of antibiotics. Thus, the work could be worthy of publication in eLife if the following points and suggestions can be addressed/considered.

To be addressed:

1) The metal binding site seems to have an important role here. Does it require metals to function? If so, which ones – Zn2+? Probably some/many of the experiments should be compared with and without divalent metals.

As noted in our Introduction, previous work suggests that the MBD binds to zinc. Unpublished research by our groups suggest that the physiological ligand of the MBD could be iron. However, we are not challenging the notion that MBD binds to zinc in the present work. Furthermore, both zinc and iron (as well as other metals) can promote the interaction of SecA with SecB. Determining the effect of any particular metal on the affinity of the MBD for the ribosome is extremely challenging. Ribosomes tend to contain large amounts of transition metal ions (including both zinc and iron). In addition, many cysteine-based iron-binding motifs will also bind to zinc with a similar or higher affinity. Because of the difficulty of these experiments and because the identity of the metal did not affect our conclusions, we did not address this topic.

2) Figure 3A shows important evidence for interaction of flexible linker with the catalytic core by intra-molecular crosslinking of residue 852. The authors say the SecA-852 band is fully crosslinked, in contrast to 893 and 898, because of its higher apparent mobility. I would like the authors to redo this experiment, but with equal quantities of SecA, just to check that it is not a loading artefact. Also, why is the crosslink not induced by UV ? Were the uncrosslinked samples kept in the dark at all times?

We have repeated this experiment numerous times with different levels of loading. We have noted the reproducibility of this experiment in the revised manuscript. We occasionally see that the protein migrates as a double band at early steps in purification but always converts to the aberrantly migrating form in Figure 3A and B following subsequent purification steps. We speculate that the reason for the increased efficiency of crosslinking is that the chemical environment surrounding position 852 activates the Bpa (or lowers the energy required for photoactivation of the Bpa). However, although the chemical basis for this increased efficiency is not known, it does not affect our conclusions.

3) Concerning the X-ray structure. If the FLD influences the conformation of SecA, as stated, then it should be evident in the high-resolution structure.

Not necessarily. If our model is correct, the FLD-induced autoinhibited state is only semi-stable. Thus, packing of the protein into crystals (a non-physiological condition) could easily favour a non-autoinhibited conformation. Indeed, we carried out the SAXS experiments because we were concerned about this possibility. We have made this line of thought more apparent in the revised text.

How does the new structure compare to others determined without the FLD?

As noted in the Results section, our structure is very similar to the previously published crystal structure of full-length SecA from E. coli (PDB file: 2FSG). The notable differences are that our structure is symmetrical and that the PPXD is better resolved. We have noted these differences more explicitly in the revised text.

Did the full-length protein fail to crystallise?

We did not attempt to crystallise the full-length protein as full-length SecA from multiple species has already been crystallised (and published) multiple times. As noted in the Introduction, the CTT is not resolved in any of these structures.

This figure could do with some improvement by colouring the different domains and highlighting the location of the C-terminus and FLD. The two monomers of the dimers could also be distinguished somehow.

The protomeric subunits of the apparent dimer (i.e. the “monomers”) are coloured differentially as noted in the figure legend. We have revised the figure and legend to make this point more clear. In addition, we have marked the location of the most C-terminal residue that is still resolvable with an asterisk in the figure.

As mentioned in the Results section, the FLD is not resolved in this structure. Its location is not marked because it is not known.

Reviewer #2:

The study from Jamshad and colleagues investigates the function of the C-terminal domain of SecA, a key factor in protein export in bacteria. The authors show that in E. coli the C-terminal domain of SecA, as well as binding SecB, can also bind to the ribosome. Loss of the entire CTT leads to no obvious growth defect and actually enhances ribosome binding, suggesting that it likely has a regulatory function. Indeed through crosslinking, structural and functional analyses a model emerges where the metal binding domain of the CTT acts to control substrate loading in response to ribosome binding. Overall, the experiments are largely well performed and the study provides new insight into SecA function. Following attention to the points below I would be supportive of publication.

In Figure 1D the SUMU-CTT is in large excess over ribosomes, it would be important to see a 10% input to assess the efficiency of binding.

The purpose of the experiments depicted in Figure 1D and E was to demonstrate that the MBD binds the ribosome and that it most likely does so at a specific site. Under physiological conditions the MBD is always connected to the catalytic core. The affinity of this domain for the ribosome in the absence of the catalytic core is not entirely relevant and does not affect our conclusion that the MBD binds to the ribosome near the site where full-length SecA binds the ribosome.

For the purposes of review, we note that preliminary experiments suggest that the affinity of SUMO-CTT for the ribosome is relatively weak. Lowering the input concentration of SUMO-CTT to 10% decreases the amount of co-sedimenting SUMO below the detection limit for western blotting. Concentrating the protein to carry out the converse experiment (10-fold more SUMO-CTT) results in protein solubility issues.

It would also be informative to assess whether SecA lacking the CTT is able to inhibit binding.

This is a technically challenging experiment that could provide limited (albeit potentially interesting) insight into the structure of the SecA-ribosome complex. Moreover, the results would not affect our conclusions. Thus, it is clearly beyond the scope of this work.

In Figure 2B, the effect of the nascent chain is established by comparing empty 70S ribosomes with a stalled SecM RNC. This could reflect changes due to the nascent chain, but also the fact that the ribosomes are now programmed. This could be distinguished using a shorter nascent chain that has not yet emerged from the exit tunnel.

Because we have already published this experiment, we have highlighted this publication more explicitly in the revised manuscript. We have also addressed this concern in greater detail with new experimental results presented in Figure 5—figure supplement 1B.

Moreover, data from the Wintermeyer group also indicate that interaction of the nascent chain with the uL23 inside the tunnel can also trigger changes in binding of SRP, another exit site ligand, which also interacts via uL23.

We assume the reviewer refers to the publication by Bornemann et al. (Nat Struct Mol Biol. 15(5):494-9). This paper suggests that the SRP binds with high affinity to translating ribosomes containing a signal anchor that is still buried in the exit channel. The possibility of programmed ribosomes is a very intriguing, but there is very little evidence to support their existence. Because the ribosomal structure is thought to be relatively rigid, it is difficult to imagine how signals could transmitted from deep in the exit channel to the ribosomal surface. In addition, ribosome profiling experiments by the Bukau group (Nature. 536(7615):219-23) do not support the existence of the high-affinity SRP-ribosome complex proposed by Bornemann et al. in vivo. Finally, mechanistic models published in subsequent papers by the Wintermeyer group (e.g.Nucleic Acids Res. 45(20):11858-11866) appear to suggest that the SRP does not distinguish between ribosomes containing nascent signal anchor motifs buried in the exit channel and those that do not.

In Figure 2D, there is no control to show that SecA is essential in the strain being used (e.g. control strain lacking exogenous SecA with and without IPTG).

The IPTG-dependent growth of these strains was noted in the Materials and methods section. We have also noted the IPTG-dependent growth in the Results section of the revised manuscript to make this more clear. We did not depict this experiment since the results were an empty plate.

The writing on the back of the plate is also quite intrusive.

We apologise for the intrusiveness of the labels. Because we take digital photographs of our plates, these labels are present to record the identity of the strain in the image. We cannot write directly on the image without altering it. Thus, we have retained the writing.

It would be essential to show the expression levels of the proteins relative to endogenous levels of SecA. It could be that the mutants are unstable in vivo and this causes the phenotype.

IPTG-induced expression of SecA is toxic when produced from a plasmid, and SecAΔMBD (and SecAΔCTT) becomes toxic at a similar concentration of IPTG, suggesting it is just as stable as wild-type SecA. In addition, the growth of strains producing SecAΔMBD from the chromosome was similar at all IPTG concentrations tested (10 μM to 1 mM), indicating that the growth defect is not due to differences in the steady-state level of the proteins. These results were noted in the original manuscript, and we have highlighted the more explicitly in the revised manuscript.

The functional analysis of the mutants would be more complete if their ability to support protein translocation (either in vivo or in vitro) was evaluated.

We assume that the growth defect of the strains expression SecAΔMBD is due to a defect in protein translocation since the only known essential function of SecA is as a translocation ATPase. We agree it would be interesting to carry out a full analysis of different Sec substrate proteins to determine which are affected. However, this is clearly beyond the scope of the current work.

In Figure 3, there is a loss of two peptides from the MS analysis. Was the crosslinked peptide also detectable? This would strengthen the identification of the two peptides as the crosslinking site.

We spent months trying to identify a mass peak consistent with the crosslinked peptides to no avail. It is worth noting that the mass of the crosslinked peptide is very large – even for a crosslinked peptide. Because signal tends to decrease with the size of the peptide, our inability to detect this mass peak was disappointing but not surprising. Other factors – such as inhibition of trypsin as a result of conformation of the crosslinked protein or simply because the crosslinked peptide just doesn’t “like to fly” – could also have contributed to the low signal.

Was the auto-crosslinked complex tested for structural analysis? This might have stabilised the CTD allowing it to be visualised.

Yes, it is possible to purify the auto-crosslinked protein in limited amounts with a significant amount of effort, and preliminary experiments suggest it is possible to crystallise this protein. Although these preliminary results are promising, producing crystals of sufficient quality would require a substantial amount of effort. Moreover, it’s possible that the resulting structure may not provide significant insight into the structure of the autoinhibited state. Thus, this experiment is beyond the scope of the current work.

Does the ∆MBD construct with the 852 suppressor also form the auto crosslink? This would strengthen the argument that the locked ∆MBD conformation is a state the wild-type protein also encounters and not a non-physiological dead-end state.

Unfortunately, the amount of time, effort and resources required to construct the mutant, purify the Bpa-incorporated protein and analyse it by mass spectrometry make it impossible to complete in a timely fashion.

In addition to 852, crosslinkers were also incorporated in the MBD at positions 893 and 898, but it was not clear why these were not further analysed. In particular, it would seem logical to text if these give rise to crosslinking to rProteins as that might permit further pinpoint localisation of the MBD at the ribosome surface or relative to the rest of SecA.

We analysed all three of these SecABpa variants (i) for their ability to form the internal crosslinks and (ii) for their ability to crosslink to the ribosome. (Indeed, we initially purified these proteins in an attempt to determine the location of the CTT-binding site on the ribosomal surface.) We did not include these results because, with the exception of the internal crosslink in SecABpa852, none of the mutants produced internal or SecA-ribosome crosslink. The potential reasons for the negative result are myriad.

If Figure 5C, what is higher UV-specific band seen just above the SUMO-SecA band with the 209 construct? Is this a crosslink to another ribosomal protein? If so, that might further define the positioning of SecA relative to the ribosome.

We were intrigued by this band for the same reasons as the reviewer. If the previous structural models are correct, position 299 should be well positioned to interact with uL24. The size of the adduct is rather larger than expected for a crosslink to a ribosomal protein (which tend to be relatively small), but crosslinks do tend to make proteins run aberrantly. We spent several months attempting to determine the identity of the crosslinking partner but could not.

In Figure 5D, the crosslink to the ∆MBD seems smaller. Is it established that this is still a crosslink to uL29 and not a different rProtein? This would be important to establish and confirm that ∆MBD binds in the same position as the WT, but cannot reposition upon binding the nascent chain.

SecAΔMBD consistently migrates faster than full-length SecA in SDS-PAGE (see Figure 2B). Although the explanation suggested by the reviewer could be correct, we assume the difference in the running properties of this protein (and crosslinking adducts) is due to the fact that SecAΔMBD is 21 amino acids shorter than full-length SecA.

In the authors model, peptide binding leads to release of SecA and the substrate from the ribosome. Can this be tested by releasing the nascent chain in the RNC.SecA complex with puromycin?

This is an intriguing implication of our model but investigating it would clearly be beyond the scope of the present work. For example, investigating the rate of release would require kinetic measurements. As far as we understand, puromycin-induced release of arrested nascent polypeptides is very slow (and thereforelikely much slower than the binding kinetics of SecA for nascent chain). Fluorescence-based approaches may provide a more viable approach. However, these would take months or years to develop and optimise.

Reviewer #3:

The authors investigated the functions of SecA's C-terminal tail. Sedimentation and crosslinking assays with the purified motor protein and purified ribosomes or nascent chain ribosome complexes revealed that the metal binding domain binds to the ribosomal surface. The binding event causes the PPXD to shift to the open conformation, thereby destabilizing the flexible linker domain, which – according to small-scale angle scattering experiments – resides in the substrate protein binding groove. In turn, the autoinhibition is released and SecA is able to sample nascent polypeptides.

The manuscript offers new insight into the recognition process between SecA and secretory polypeptides. The approach appears to be sound. Yet both membrane and SecYEG are missing in the emerging picture. That is, SecA's substantial affinity to lipid membranes and SecYEG is not mentioned. How do the newly determined KD values compare to those previously published for the interaction of SecA with lipid membranes and with the translocon? Does such affinity comparison favor SecA interactions with the nascent chain (i) in the cytoplasm or (ii) at the membrane surface?

These are interesting questions, and we thank the reviewer for raising them. We have included additional discussion to address the interplay between these different forms of SecA.

[Editors' note: what now follows is the author responses to the editors’ decision letter after the first round of revisions.]

Editorial assessment of the author reply to reviewers:

The role of a metal requirement for MBD function was not examined satisfactorily. The sole experiment to address this issue, Figure 1D, is poorly controlled and uninterpretable. First, the signals in lane 7 and lane 2 are essentially indistinguishable (lane 7 is fainter, but the band is spread over a larger area, so the overall difference is hard to appreciate). Without replicates and statistics, one cannot reasonably infer relevance from such a modest difference. Even if one were to take the difference at face value, the effect cannot be attributed to the MBD from the information provided; it could possibly be due to an effect of EDTA on some part of the ribosome. There are no controls to verify that metal was in fact removed from the CTT. The authors have other experimental approaches at their disposal. For example, the purified CTT could be stripped of metal (e.g., with high concentration EDTA), the stripped metal and EDTA removed by dialysis or desalting, the extent of removal verified, and the metal-free CTT (compared to protein in which the metal was added back) used in the assays. Alternatively (or in addition), one could use mutants in the metal binding domain. In short, the issue of a metal requirement was not addressed.

To address this issue, we have included new experimental results in Figure 1D in which we modified the metal-coordinating cysteines in SUMO-CTT with AMS. Ribosome co-sedimentation experiments indicate that AMS modification almost completely abolishes binding to the ribosome. These results indicate that the MBD must be correctly folded in order to bind to ribosomes. We have referenced these results in subsection “Binding of the CTT to the ribosome.”.

In addition, the original editor’s summary requested more detail about “which divalent metal(s) impact function of the C-terminal domain”. At the time of revision, we were in the process of preparing a manuscript to address this issue, which was not yet ready of submission. The results described in this manuscript strongly suggest that the physiological ligand of the MBD is iron. This manuscript is currently being revised for submission to the Journal of Biological Chemistry and is available as an author preprint through BioRxiv (doi: 10.1101/613315). We reference these new results and speculate on their significance in paragraph four of the Discussion section in the revised manuscript. In the resubmission, purified SecA, SUMO-CTT and SUMO-MBD are bound to zinc. The identity of the metal could affect the strength of the interaction with the ribosome. However, it clearly does not disrupt ribosome binding entirely. Furthermore, the Zn-bound form of SecA binds with high affinity to SecB and is functional in vitro. Finally, maintaining SecA in the iron-bound form is technically challenging for a large range of reasons (e.g. contaminating zinc in purification buffers, rapid oxidation of iron under standard experimental conditions, etc.).

The request to clarify the supposed UV-independent and highly efficient crosslinking from residue 852 was also not addressed adequately. This is a crucial experiment because it is the main direct evidence for the CTT engaging the catalytic core of SecA. The problem with UV-independence is that one cannot be certain the crosslink actually formed, with the alternative interpretation being a migration artefact with no crosslinks. The manuscript partially allays this concern, but the failure to detect a peptide is not especially strong evidence on its own. Based on what is shown, it could even be the case that BpA was not incorporated into 852 (amber suppression efficiency is known to be position-sensitive). While the authors could argue that the biotin blot shows amber suppression, this blot is very over-loaded and there is no negative control of a comparably overloaded SecA lacking the biotin tag. For these reasons, the conclusion that increased migration represents extremely efficient UV-independent intramolecular crosslinking through some unknown mechanism is not sufficiently compelling evidence to support a critical conclusion of the paper. If the authors' claim that this crosslink forms during the purification is correct, they can simply treat intact bacteria without or with UV, then harvest directly into SDS and monitor by blotting for the biotin tag. This should show a clear UV-dependent size shift to a position where the purified protein migrates. Other approaches are also feasible, including positioning BpA at nearby residues where the environment would be different, and hence, avoid constitutive UV-independent crosslinking. As it stands, a key conclusion of the study remains in doubt.

To address this issue, we present new evidence in Figure 4—figure supplement 1. In this figure, we purified SUMO-SecA-biotin and SUMO-SecABpa852-biotin via their C-termini directly from cell lysates using streptavidin-coated magnetic beads. The two most prominent bands in the C-terminally purified SUMO-SecABpa852-biotin migrated with molecular weights identical to full-length SecA-biotin and to the faster migrating species in Figure 4 (previously Figure 3), indicating that both species are full-length SUMO-SecABpa852-biotin. We have referenced these new results in subsection “Auto-crosslinking of the FLD in the substrate binding groove of SecA” of the revised text. Exposure of lysates to light did not significantly increase the amount of the lower (apparent) molecular weight species. However, this result is consistent with our model if (a) crosslinking is very efficient and (b) the different populations of SecA (autoinhibited and uninhibited) are stable. We also note that previous NMR studies (Gelis et al., 2007) and our SAXS results are consistent with the proposed site of crosslinking.

We have also included additional text to provide a clearer explanation for the high efficiency of auto-crosslinking of SUMO-SecABpa852-biotin (paragraph five in the Discussion section). To summarise: the efficiency of Bpa crosslinking is influenced by three factors: (i) the amount of time the benzophenone group of the Bpa is in contact with the target molecule, (ii) the chemical reactivity of Bpa toward the target molecule and (iii) the amount of time the benzophenone group stays in the activated state. First, the results of this study and others (Gelis et al., 2007) is consistent with the idea that the FLD is stably bound in the substrate protein-binding groove of SecA, which could result in a long-lived contact between position 852 and the binding pocket. Second, although benzophenone can, in theory, react with any C-H bond, in practice it reacts with different efficiencies toward different amino acid side chains (Lancia et al., 2014; Wittelsberger et al., 2006). Finally, environment (hydrophobicity, pH, etc.) influences the photo-reactive properties of many aromatic compounds. Indeed, the excitation and emission spectra of 4-hydroxy-benzophenone, which has the same photo-reactive group as Bpa, are strongly affected by hydrophobicity and pH (Barsotti et al., 2015; Barsotti et al., 2017). These environmental conditions are normally assumed to be negligible because Bpa is normally incorporated at surface-exposed positions because in order to capture protein-ligand interactions. However, because our results indicate that position 852 is buried in a hydrophobic groove, the environment surrounding the Bpa could affect the photo-reactive properties of the benzophenone in unexpected ways. Any (or all three) of these factors could have influenced the efficiency of auto-crosslinking in this experiment. We apologise for the unclear explanation for the high efficiency of crosslinking in our previous reply.

The request to provide the reader some estimate of the efficiency of binding in Figure 1D (first point of reviewer 2) was either ignored or misunderstood. The reviewer was asking simply to run 10% of the input sample on the same gel as the pulldowns so one could assess whether the amount pulled down was more or less than 10% of what went into the reaction. If it is far less, one could run 1% of the input (or whatever is roughly in the suitable range). The point is to provide the reader with a rough idea of how much is being pulled down in the experiment because this impacts how believable the findings are.

We apologise. We completely misunderstood what the reviewer was requesting. We have addressed this issue by including 10% loading controls in the new experimental results depicted in Figure 1D. These results suggests that <20% of the ribosomes in the binding reaction are bound by SUMO-CTT in binding reactions containing 1 μM ribosome and 10 μM SUMO-CTT. Although this result suggests that the affinity of the MBD alone for the ribosome is relatively low, it is consistent with the decrease small decrease in the affinity of SecAΔCTT for the ribosome and it is consistent with the idea that the MBD normally binds to the ribosome in conjunction with the catalytic core. We have discussed have elaborated on these issues more extensively in paragraph two of the Discussion section.

Similarly, the request to verify that SecA lacking the CTT fails to compete for binding in Figure 1D was also not addressed (second point of reviewer 2). This is a rather standard control, and it most certainly would affect the conclusions were it to compete similarly to WT SecA.

As requested, we have repeated the binding competition experiment, but the results of this experiment were variable. The reason for the limited reproducibility is unclear. For example, it is possible that there is some experiment-to-experiment variability in the oxidation state of the metal-coordinating cysteines or in the bound metal cofactor in the MBDs of in SUMO-CTT and full-length SecA. However, the results presented in Figure 1 (and supplements) indicate that: (i) the MBD is well conserved (if not universally); (ii) the pattern of conservation suggests a binding partner besides SecB; (iii) the MBD binds to the ribosome; and (iv) the MBD binds to a site near uL23. Because it is possible to make these conclusions without the binding competition experiments, we have removed these results from the resubmitted manuscript.

The reasonable request to test SecA levels in the different mutant strains relative to endogenous levels of SecA was addressed in a rather convoluted manner and did not adequately address the concern. IPTG-induced toxicity is not a good surrogate for expression levels because the basis of this toxicity could be different for different mutants (i.e., for some, it may have to do with SecA function, for others, it might have to do with protein aggregation or inappropriate interactions). It should be a straightforward matter to directly test expression levels by blotting using widely available SecA antibodies. Similarly, the request to directly test protein translocation is both reasonable and straightforward. The concern is not that SecA has some other function, but rather that ∆MBD is having its effect via a dominant toxicity unrelated to its function (which is certainly plausible).

To address this issue, we present new results in Figure 2—figure supplement 3, which indicate that SecA is produced at similar levels in the three strains depicted in Figure 2D. We have referenced these results in the revised manuscript.

The aberrant migration of the crosslink to ∆MBD relative to the corresponding crosslink to full length SecA in Figure 5D was not addressed. While it is true that ∆MBD migrates faster than wild type, the key concern was that the shift upon crosslinking was far smaller for ∆MBD than for full length, suggesting the possibility that the crosslinking partners are different. In this region of the gel, this difference in size shift is pretty substantial. It is therefore not acceptable to simply assume both crosslinks are to uL29.

To address this concern, we analysed the crosslinking adduct produced by SecA∆MBDBpa399 using LC-MS/MS. This analysis indicated that SecA∆MBDBpa399 crosslinks to uL29. We have included this data in subsection “Site-specific crosslinking of SecA to ribosomes” of the revised text and in the legend for Figure 3.

[Editors' note: the author responses to the re-review follow.]

This revised article generally satisfies the reviewer comments. As you will see below, reviewer 2 feels you should be more cautious in some of your conclusions that are not fully convincing, particularly the intramolecular crosslink. In reading the article and evaluating the revisions myself, I have similar reservations.

Please adjust the text accordingly and submit a final version as per the instructions below.

Reviewer #2:

The authors have now addressed most of the major issues I had identified from the first round of reviews. They nicely show that a programmed short nascent chain does not block the crosslink to uL29 strengthening the authors model. They also show the 10% input for the binding assay in Figure 1, this reveals that the level of binding is relatively low, probably 1-2% of the input protein, i.e. binding to around 10-20% of ribosomes. They have also now attempted the suggested competition assay control (for original Figure 1D) but obtained inconclusive results with the ∆CTT construct. I would agree this compromises the interpretation of the competition experiment, hence its removal is sensible. It is puzzling why the assay was so variable considering that the WT and ∆CTT constructs apparently behaved well for the structural analysis.

We agree that the lack of reproducibility is puzzling given how reproducible the experiment was in the past. SUMO-CTT has a very low affinity for the ribosome and because it is prone to misfolding due to oxidation or sequestration of the bound metal. Thus, some level of variability due to small differences in the fraction of active protein between batches is expected.

Protein levels of the in vivo expression of the constructs is now shown and look similar, albeit without a loading control. I still also think it would have been informative to also look at the effects of the SecA mutants on translocation rather than growth alone.

We thank the reviewer for noticing this. In addition to ensuring that we loaded samples with the same cell density, we also blotted against thioredoxin-1 as a loading control. However, we inadvertently forgot to include this loading control in the figure. We have revised Figure 2—figure supplement 3 to include the loading control.

The revised section of the text dealing with the internal crosslink now doesn't actually mention the auto-crosslink formation, save in the section and figure headings.

We did mention the auto-crosslink in this paragraph. However, our reference to it was not very clearly written. We apologise for the resulting confusion. We have now more explicitly mentioned this in the revised subsection (subsection “Auto-crosslinking of the FLD in the substrate binding groove of SecA”).

The section should explain the behaviour of the three constructs +/- UV and then introduce the uv-independent crosslink explanation.

The reason for the highly efficient intramolecular crosslinks is unknown. We speculate on several possibilities in the Discussion section. However, we agree that this could be confusing to the reader. We have therefore explicitly referred the reader to the Discussion where the possibilities are in depth.

Also, as I mentioned previously, and was discussed by the other reviewers, the lack of identification of the crosslinked peptide tempers the definitive confirmation that the faster-migrating band is actually crosslinked. So this remains a slight weak point in the manuscript.

We understand the reviewer’s (and the editor’s) concern. We have tempered our conclusions in the Results section (subsection “Auto-crosslinking of the FLD in the substrate binding groove of SecA”) and our interpretation in the Discussion (paragraph one and three) to address this weakness.

https://doi.org/10.7554/eLife.48385.035

Article and author information

Author details

  1. Mohammed Jamshad

    Institute for Microbiology and Infection, School of Biosciences, University of Birmingham, Birmingham, United Kingdom
    Contribution
    Supervision, Investigation, Methodology, Writing—review and editing
    Contributed equally with
    Timothy J Knowles
    Competing interests
    No competing interests declared
  2. Timothy J Knowles

    Institute for Microbiology and Infection, School of Biosciences, University of Birmingham, Birmingham, United Kingdom
    Contribution
    Resources, Data curation, Validation, Investigation, Methodology
    Contributed equally with
    Mohammed Jamshad
    Competing interests
    No competing interests declared
  3. Scott A White

    Institute for Microbiology and Infection, School of Biosciences, University of Birmingham, Birmingham, United Kingdom
    Contribution
    Data curation, Formal analysis, Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  4. Douglas G Ward

    Institute of Cancer and Genomic Sciences, University of Birmingham, Birmingham, United Kingdom
    Contribution
    Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  5. Fiyaz Mohammed

    Institute of Immunology and Immunotherapy, University of Birmingham, Birmingham, United Kingdom
    Contribution
    Resources, Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  6. Kazi Fahmida Rahman

    Institute for Microbiology and Infection, School of Biosciences, University of Birmingham, Birmingham, United Kingdom
    Contribution
    Investigation, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
  7. Max Wynne

    Institute for Microbiology and Infection, School of Biosciences, University of Birmingham, Birmingham, United Kingdom
    Contribution
    Investigation, Writing—review and editing
    Competing interests
    No competing interests declared
  8. Gareth W Hughes

    Institute for Microbiology and Infection, School of Biosciences, University of Birmingham, Birmingham, United Kingdom
    Contribution
    Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1228-6152
  9. Günter Kramer

    Center for Molecular Biology of Heidelberg University (ZMBH), German Cancer Research Center (DKFZ), ZMBH-DKFZ Alliance, Heidelberg, Germany
    Contribution
    Conceptualization, Resources, Funding acquisition, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7552-8393
  10. Bernd Bukau

    Center for Molecular Biology of Heidelberg University (ZMBH), German Cancer Research Center (DKFZ), ZMBH-DKFZ Alliance, Heidelberg, Germany
    Contribution
    Funding acquisition, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0521-7199
  11. Damon Huber

    Institute for Microbiology and Infection, School of Biosciences, University of Birmingham, Birmingham, United Kingdom
    Contribution
    Conceptualization, Supervision, Funding acquisition, Investigation, Methodology, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    d.huber@bham.ac.uk
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-7367-3244

Funding

Biotechnology and Biological Sciences Research Council (BB/L019434/1)

  • Mohammed Jamshad
  • Damon Huber

Biotechnology and Biological Sciences Research Council (BB/P009840/1)

  • Timothy J Knowles
  • Gareth W Hughes

Biotechnology and Biological Sciences Research Council (MIBTP)

  • Max Wynne

Deutsche Forschungsgemeinschaft (FOR 1805)

  • Günter Kramer
  • Bernd Bukau

Deutsche Forschungsgemeinschaft (SFB 638)

  • Günter Kramer
  • Bernd Bukau

Wellcome (099266/Z/12/Z)

  • Fiyaz Mohammed

Deutsche Forschungsgemeinschaft (KR3593/2-1)

  • Günter Kramer

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank A McNally and members of the Bukau, Mayer, Henderson, Lund, Grainger, Cole and Rossiter groups for helpful advice and discussions. We thank B Zachmann-Brand, Jingli Yu and the functional genomic facility (School of Biosciences, University of Birmingham) for technical assistance. We are grateful to the University of Birmingham Protein Expression Facility for use of their facilities.

Senior Editor

  1. John Kuriyan, University of California, Berkeley, United States

Reviewing Editor

  1. Ramanujan S Hegde, MRC Laboratory of Molecular Biology, United Kingdom

Publication history

  1. Received: May 10, 2019
  2. Accepted: June 26, 2019
  3. Accepted Manuscript published: June 27, 2019 (version 1)
  4. Version of Record published: July 10, 2019 (version 2)

Copyright

© 2019, Jamshad et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 942
    Page views
  • 153
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

  1. Further reading

Further reading

    1. Biochemistry and Chemical Biology
    2. Cell Biology
    Amal Alex et al.
    Tools and Resources
    1. Biochemistry and Chemical Biology
    2. Structural Biology and Molecular Biophysics
    Ching-Ju Tsai et al.
    Research Article Updated