Characterization of Eag chaperones and their associated effectors has thus far been limited to the EagT6-Tse6 and EagR1-RhsA chaperone-effector pairs from P. aeruginosa and Serratia marcescens, respectively (Cianfanelli et al., 2016a; Whitney et al., 2015). In both cases, the chaperone gene is found upstream of genes encoding its cognate effector and an immunity protein that protects the toxin-producing bacterium from self-intoxication (Figure 1A). We previously showed that EagT6 interacts with the N-terminal TMDs of Tse6, an observation that led us to hypothesize a general role for Eag chaperones in ‘solubilizing’ hydrophobic TMDs of effectors in the aqueous environment of the cytoplasm so they can be loaded into the T6SS apparatus (Figure 1B; Quentin et al., 2018). However, evidence supporting this general role is lacking because homology-based searches for additional Eag chaperones can yield difficult to interpret results due to a scarcity of conserved residues and homology of this protein family to the phage protein DcrB (Samsonov et al., 2002), which is widely distributed in both T6SS-positive and T6SS-negative organisms. Similarly, the identification of N-terminal TMD-containing PAAR effectors that might require Eag chaperones is also challenging because many PAAR domain-containing effectors lack TMDs (Shneider et al., 2013), and aside from being comprised of hydrophobic residues, the TMDs themselves are poorly conserved.

Figure 1 with 1 supplement see all

Download asset Open asset

In an attempt to overcome the challenges associated with identifying Eag-interacting T6SS effectors, we used jackhmmer to generate a sequence alignment hidden Markov model (HMM) for the N-terminal 60 residues of Tse6 using an iterative search procedure that queried the UniProtKB database (Johnson et al., 2010). We reasoned that if there exists a molecular signature present in effector proteins indicative of Eag chaperone association, it would be located within this region of Tse6 homologous proteins because it contains a known chaperone-binding site. Remarkably, the HMM we obtained revealed a nearly invariant AARxxDxxxH motif, which in Tse6 is found in the first 15 residues of the protein and is immediately N-terminal to its first TMD (Figure 1C). In total, our query identified over 2054 proteins containing this motif (Supplementary file 1 and Figure 1—figure supplement 1A). Among these candidate effectors, our search identified the recently characterized toxins Tre1, Tas1, and DddA as well as many toxins of unknown function indicating that our approach may have identified T6SS effectors with novel biochemical activities (Ting et al., 2018; Ahmad et al., 2019; Mok et al., 2020). Interestingly, prior to any knowledge of PAAR domains or Eag chaperones being involved in T6SS function, Zhang and colleagues noted the existence of this N-terminal motif in PAAR-containing proteins through an informatics analysis of bacterial nucleic acid degrading toxins (Zhang et al., 2011). Here, they refer to the motif and its adjacent TMD region as ‘prePAARTM’ because these sequence elements co-occur with one another and because they are both found N-terminal to PAAR domains. We have chosen to refer to the motif as ‘prePAAR’ because, as described below, our data indicate it has a function that is distinct from the TMD regions.

Examination of our putative effector sequences revealed that prePAAR is substantially enriched in bacterial genera with characterized T6SSs including Pseudomonas, Burkholderia, Salmonella, Shigella, Escherichia, Enterobacter, Yersinia, and Serratia. Interestingly, no prePAAR motifs were identified in Vibrio despite an abundance of species within this genus possessing highly active bacteria-targeting T6SSs. We next obtained all 56,324 available genomes from NCBI for the abovementioned genera and found that 26,327 genomes encode at least one prePAAR motif. After removing all redundant sequences, 6101 unique prePAAR-containing proteins present across 5584 genomes were used for further analyses (Supplementary file 2, List C). In these genomes, we determined that approximately 90% encode a single prePAAR motif, although instances where prePAAR is present up to six times within a single genome were also identified (Figure 1D). To determine if these unique proteins are probable TMD-containing T6SS effectors that require Eag chaperones for secretion, we next examined each prePAAR-containing protein and its associated genome for the following three criteria: (1) the existence of an Eag chaperone encoded in the same genome, (2) the presence of a downstream PAAR domain, and (3) predicted TMDs in the first 300 amino acids of the protein (Krogh et al., 2001; Käll et al., 2007). The location restriction in our TMD search was used in order to exclude C-terminal toxin domains that possess TMDs, which differ from N-terminal translocation TMDs in that they may not require chaperones for secretion (Mariano et al., 2019). We searched each genome for Eag proteins using an HMM for DUF1795 and found that 99.5% (5554/5584) of prePAAR-containing genomes also possessed at least one eag gene (Jones et al., 2014). In approximately 14% of the 5554 genomes analyzed, the number of prePAAR motifs matched the number of Eag homologues. In the remainder of cases, the number of Eag homologous proteins exceeded the number of prePAAR motifs, with a weighted average of 2.5 paralogues per genome. As is the case with eagT6-tse6 and eagR1-rhsA, ~90% of the identified prePAAR-containing effector genes appear directly beside an eag gene, whereas the remaining ~10% are found in isolation suggesting that their putative chaperone is encoded elsewhere in the genome. We removed pre-PAAR-containing protein fragments (proteins less than 100 amino acids in length) and further reduced redundancy by clustering sequences with 95% identity. Remarkably, in all but two of the remaining 1166 prePAAR-containing proteins, we identified a PAAR domain, indicating a probable functional relationship between prePAAR and PAAR. The two prePAAR-containing proteins lacking a PAAR domain were either adjacent to a gene encoding a PAAR domain-containing protein or directly beside T6SS structural genes. Finally, we searched 1166 prePAAR-containing proteins for TMDs and found that all protein sequences contained predicted TMDs with 86% having one region of TMDs and 14% having two regions of TMDs. In sum, our prePAAR-based search procedure identified thousands of candidate effector proteins possessing properties consistent with the requirement for an Eag chaperone for T6SS-dependent export.

To further analyze our collection of prePAAR-containing effectors, we built a phylogenetic tree from 1166 non-redundant effector sequences that represent the diversity present in our collection of sequences (Figure 1E). Interestingly, two distinct sizes of proteins emerged from this analysis: large prePAAR effectors that are on average 1196 amino acids in length and small prePAAR effectors comprised of an average of 443 amino acids (Figure 1E and Figure 1—figure supplement 1B). As noted previously, all effectors contained predicted TMDs; however, large effectors almost exclusively contained a single region of TMDs N-terminal to their PAAR domain, whereas most small effectors contained TMD regions N- and C-terminal to their PAAR domain. To distinguish between these two domain architectures, we hereafter refer to large, single TMD region-containing prePAAR effectors as class I and small, two TMD region-containing prePAAR effectors as class II. Notably, class I effectors also contain numerous YD repeat sequences, which are a hallmark of rearrangement hotspot (Rhs) proteins that function to encapsulate secreted toxins (Figure 1F; Busby et al., 2013). We also found a small subset of these effectors are encoded by two separate ORFs, the first encoding prePAAR-TMD-PAAR and the second encoding a protein containing Rhs repeats and a C-terminal toxin domain (Figure 1—figure supplement 1C). Conversely, class II effectors are distinguished by a GxxxxGxxLxGxxxD motif in addition to their second TMD region.

As a first step toward validating our informatics approach for identifying Eag chaperone-effector pairs, we assessed the ability of several newly identified Eag chaperones to interact with the prePAAR-containing effector encoded in the same genome. We previously demonstrated that the class II effector Tse6 interacts with EagT6 and we similarly found that when expressed in E. coli, Eag chaperones from Enterobacter cloacae, Salmonella Typhimurium, Shigella flexneri, and Serratia proteamaculans co-purified with their predicted cognate effector (Figure 1G and Figure 1—figure supplement 1C). Collectively, these findings indicate that prePAAR proteins constitute two classes of TMD-containing T6SS effectors and that representative members from both classes interact with Eag chaperones.

We next sought to examine the specificity of Eag chaperones towards prePAAR effectors in a biologically relevant context. To accomplish this, we inspected our list of prePAAR effectors and found that the soil bacterium Pseudomonas protegens Pf-5 possesses both a class I and class II effector, encoded by the previously described effector genes rhsA and tne2, respectively (Tang et al., 2018). Furthermore, the genome of this bacterium encodes two putative Eag chaperones, PFL_6095 and PFL_6099, which have 25% sequence identity between them (Figure 2A). PFL_6095 is found upstream of rhsA and is likely co-transcribed with this effector, whereas PFL_6099 is not found next to either effector gene. To examine the relationship between these genes, we generated strains bearing single deletions in each effector and chaperone gene and conducted intraspecific growth competition assays against P. protegens recipient strains lacking the rhsA-rhsI or tne2-tni2 effector-immunity pairs. We noted that protein secretion by the T6SS of P. protegens is substantially inhibited by the threonine phosphorylation pathway, so we additionally inactivated the threonine phosphatase encoding gene pppA in recipients to induce a ‘tit-for-tat’ counterattack by wild-type donor cells (Figure 2—figure supplement 1A–B; Mougous et al., 2007; Basler et al., 2013). Consistent with the effector-immunity paradigm for bacteria-targeting T6SSs, wild-type P. protegens readily outcompeted ΔrhsA ΔrhsI ΔpppA and Δtne2 Δtni2 ΔpppA strains in a rhsA- and tne2-dependent manner, respectively (Figure 2B). Additionally, we found that a strain lacking PFL_6095 no longer exhibited a co-culture fitness advantage versus a ΔrhsA ΔrhsI ΔpppA recipient but could still outcompete tne2 sensitive recipients to the same extent as the wild-type strain. Conversely, a ΔPFL_6099 strain outcompeted ΔrhsA ΔrhsI ΔpppA but not Δtne2 Δtni2 ΔpppA recipients. Together, these data indicate that the delivery of RhsA and Tne2 into susceptible target cells requires effector-specific eag genes.

Figure 2 with 1 supplement see all

Download asset Open asset

To test the ability of PFL_6095 and PFL_6099 to act as RhsA- and Tne2-specific chaperones, respectively, we co-expressed each chaperone-effector pair in E. coli and examined intracellular effector levels by western blot. Consistent with functioning to promote cognate effector stability, accumulation of RhsA only occurred in the presence of PFL_6095, whereas Tne2 accumulated in cells containing PFL_6099 (Figure 2C). We next examined the stability-enhancing properties of PFL_6095 and PFL_6099 when expressed at native levels in P. protegens. Due to challenges associated with detecting RhsA and Tne2 in unconcentrated cell lysates, we constructed chromosomally encoded N-terminal decahistidine-tagged (his₁₀) fusions of RhsA and Tne2 to facilitate the enrichment of these proteins from P. protegens and confirmed that these fusions did not compromise the ability of these effectors to intoxicate recipients (Figure 3—figure supplement 1A–B). Following affinity purification, RhsA and Tne2 levels were assessed using RhsA and vesicular stomatitis virus glycoprotein epitope (VSV-G) antibodies, respectively. In line with our data in E. coli, we were unable to detect RhsA in the absence of PFL_6095 whereas Tne2 was absent in a strain lacking PFL_6099 (Figure 2D). Collectively, these data suggest that Eag chaperones exhibit a high degree of specificity for their cognate effectors. Based on our characterization of these genes, we propose to rename PFL_6095 and PFL_6099 to eagR1 and eagT2, respectively, to reflect their newfound role as chaperones for the prePAAR-containing effectors RhsA and Tne2.

Previous biochemical studies on the class II prePAAR effector Tse6 and its cognate chaperone EagT6 demonstrated that the two TMD regions of this effector each require an EagT6 chaperone for stability (Quentin et al., 2018). These findings suggest that there may exist a physical limitation to the number of TMDs that a single EagT6 chaperone can stabilize. Our finding that class I prePAAR effectors contain only one TMD region suggests that these proteins may only possess one Eag interaction site (Figure 3A). To test this, we constructed a RhsA variant lacking its N-terminal region (RhsA_∆NT) and co-expressed this truncated protein with EagR1 in E. coli. Consistent with our hypothesis, affinity purification of RhsA_∆NT showed that this truncated variant does not co-purify with EagR1 (Figure 3B). Additionally, expression of the 74 residue N-terminal fragment of RhsA in isolation was sufficient for EagR1 binding (Figure 3—figure supplement 1C). Our data also demonstrate that in contrast to wild-type RhsA, RhsA_∆NT is stable in the absence of EagR1 when expressed in E. coli indicating that the N-terminus imparts instability on the protein in the absence of its cognate chaperone. In P. protegens, we could readily detect rhsA_∆NT in a strain lacking eagR1, corroborating our findings in E. coli (Figure 3C). However, despite the enhanced stability of chaperone ‘blind’ RhsA_∆NT, a P. protegens strain expressing this truncation was no longer able to outcompete RhsA-sensitive recipient cells demonstrating an essential role for the chaperone-bound N-terminus during interbacterial competition (Figure 3D).

Figure 3 with 1 supplement see all

Download asset Open asset

After ruling out the possibility that truncating the N-terminus of RhsA affects the growth-inhibitory activity of its C-terminal toxin domain (Figure 3—figure supplement 1D), we next examined the ability of RhsA_∆NT to interact with its cognate secreted structural component of the T6SS apparatus. T6SS effectors encoded downstream of vgrG genes typically rely on the encoded VgrG protein for delivery into target cells (Whitney et al., 2014). Consistent with this pattern, PFL_6094 encodes a predicted VgrG protein, herein named VgrG1, which we confirmed is required for RhsA-mediated growth inhibition of susceptible target cells (Figure 3—figure supplement 1E). Furthermore, using a P. protegens strain expressing His₁₀-tagged RhsA and FLAG-tagged VgrG1 from their native loci, we found that these proteins physically interact to form a complex (Figure 3E). To test if the absence of the chaperone-bound N-terminus affects the formation of this complex, we used our E. coli co-expression system to purify RhsA-EagR1-VgrG1 complexes. These experiments show that RhsA_∆NT is not able to interact with VgrG1, even though this truncated protein possesses its PAAR domain, which in T6SS effectors lacking prePAAR and TMDs in their N-terminus, is sufficient for VgrG interaction (Figure 3—figure supplement 1F; Bondage et al., 2016).

To gain insight into how EagR1 binding facilitates RhsA interaction with VgrG1, we next performed negative-stain electron microscopy (EM) to examine the configuration of each subunit within this complex. To facilitate the accurate identification of each component, we obtained class averages of purified VgrG1, RhsA_∆NT, RhsA-EagR1 complex, and RhsA-EagR1-VgrG1 complex (Figure 4—figure supplement 1A–H). As expected, isolated VgrG1 and RhsA_∆NT proteins appeared as characteristic spike- and barrel-shaped proteins, respectively (Spínola-Amilibia et al., 2016; Busby et al., 2013; Figure 3F and G). Intriguingly, images of RhsA-EagR1 complexes contained a sphere-shaped object that likely represents a subcomplex between EagR1 and the N-terminus of RhsA (Figure 3H). Lastly, the class-averages of RhsA-EagR1-VgrG1 complexes revealed a close association of EagR1 and RhsA with the tip of the VgrG spike, which is likely mediated by the PAAR domain of RhsA (Figure 3I). Interestingly, although both complexes exhibit significant rotational flexibility, the average distance between the subcomplex formed by EagR1 and the N-terminus of RhsA is substantially greater in the absence of VgrG1 (average distance: 2.68 nm, n = 27 classes versus 1.20 nm, n = 26 classes) (Figure 4—figure supplement 1F–H). When taken together with our biochemical experiments, these structural data indicate that EagR1 stabilizes the N-terminus of RhsA, which may also orient the effector such that it can interact with its cognate VgrG.

In addition to a TMD-containing region, the N-terminus of prePAAR effectors also harbors the prePAAR motif itself. However, the negative stain EM images of RhsA-EagR1-VgrG1 particles presented herein and our previously determined single-particle cryo-EM structure of a complex containing Tse6-EagT6-VgrG1 are of insufficient resolution to resolve the structures of chaperone-bound effector TMDs or the prePAAR motif (Quentin et al., 2018). Therefore, to better understand the molecular basis for chaperone-TMD interactions and to gain insight into prePAAR function, we initiated X-ray crystallographic studies on both class I and class II effector-chaperone complexes. Efforts to co-crystallize P. protegens EagR1 with the prePAAR and TMD-containing N-terminus of RhsA were unsuccessful. However, the EagR1 homologue SciW from Salmonella Typhimurium crystallized in isolation and in the presence of the N-terminus of the class I prePAAR effector Rhs1 (Rhs1_NT), allowing us to determine apo and effector-bound structures to resolutions of 1.75 Å and 1.90 Å, respectively (Figure 4 and Table 1). Similar to RhsA, we confirmed that a Rhs1_∆NT variant was unable to bind its cognate chaperone, SciW (Figure 3—figure supplement 1G). The structure of the EagT6 chaperone was previously solved as part of a structural genomics effort and we were additionally able to obtain a 2.60 Å co-crystal structure of this chaperone in complex with the N-terminal prePAAR and first TMD region of the class II effector Tse6 (Tse6_NT) (Figure 4 and Table 1).

Figure 4 with 1 supplement see all

Download asset Open asset

The overall structure of SciW reveals a domain-swapped dimeric architecture that is similar to the previously described apo structure of P. aeruginosa EagT6, although each chaperone differs in its electrostatic surface properties (Figure 5—figure supplement 1A–D; Whitney et al., 2015). A comparison of the chaperone structures in their apo and effector-bound states shows that upon effector binding, both chaperones ‘grip’ the prePAAR-TMD region of their cognate effector in a claw-like manner (Figure 4A–D). Although our biochemical data indicate that Eag chaperones exhibit a high degree of specificity for their associated effector, the internal surface of the claw-shaped dimer contains a number of conserved residues that make critical interactions with the TM helices in both complexes (Figure 5A–F). For example, I22 and I24 of EagT6 create a hydrophobic surface in the ‘palm’ of the claw, which is flanked on either side by symmetrical hydrophobic surfaces comprised of A62, L66, L98, F104, and I113 (Figure 5B–D). Furthermore, the conserved hydrophilic residues S37, S41, Q58, and Q102 also interact with the bound effectors by making bifurcated hydrogen bonds to amide or carbonyl groups in the peptide backbone of the TM helices (Figure 5E–F). These polar interactions between chaperone and effector TM helices are striking because they are reminiscent of polar interactions seen within the helical packing of alpha helical transmembrane proteins, which often use serine and glutamine residues to mediate inter-helical interactions via bifurcated hydrogen bonds between side chain and main chain atoms (Dawson et al., 2002; Dawson et al., 2003; Adamian and Liang, 2002). Additionally, EagT6 and SciW provide ‘knob-hole-like’ interactions, which also feature prominently in membrane protein packing (Curran and Engelman, 2003). Knob-hole interactions involve a large hydrophobic residue on one TM helix acting as a ‘knob’ to fill the hole provide by a small residue such as glycine or alanine on another TM-helix. TM holes are typically created by GxxxG/A motifs such as those found in G19-A24 (Rhs1) and G25-A30 (Tse6). In this case, the conserved Eag chaperone residue L66 provides a knob for the A24/30 hole (Figure 5E–F). Given that the Eag chaperone dimer creates a hydrophobic environment with complementary knob-hole interactions for its cognate effector TM helices, and interacts with TM helices via side chain to main chain hydrogen bonds, we conclude that Eag chaperones provide an environment that mimics transmembrane helical packing to stabilize prePAAR effector TMDs in the cytoplasm prior to effector export from the cell.

Figure 5 with 1 supplement see all

Download asset Open asset

We next compared the conformation of the bound prePAAR-TMD fragments between our effector-chaperone co-crystal structures. Interestingly, despite the abovementioned similarities between the SciW and EagT6 structures, the conformation of the N-terminal fragment of their bound prePAAR effector differs significantly. In the SciW complex, Rhs1_NT adopts an asymmetric binding mode whereby the effector fragment does not make equivalent molecular contacts with both chains of the two-fold symmetrical chaperone dimer (Figures 4A and 5F). The first TM helix (residues 19–33) binds to the hydrophobic cavity of one SciW protomer whereas the remaining hydrophobic region of Rhs1, which consists of two anti-parallel alpha-helices connected by a short 3₁₀ helix, occupies the remainder of the binding surface. Phenylalanine residues F20 and F43 likely play an important role in the asymmetric binding of Rhs1 to SciW because their hydrophobic side chains insert into equivalent hydrophobic pockets found in each SciW protomer (Figure 5E). By contrast, Tse6_NT exhibits a pseudosymmetric binding mode with EagT6 (Figures 4C and 5F). In this structure, two alpha-helices of Tse6 each occupy equivalent Eag binding pockets and run in the opposite direction to match the antiparallel arrangement of the EagT6 dimer. For example, A7 and A30 of Tse6 interact with equivalent sites in their respective chaperone protomers (Figures 4B and 5E). These two helices, which consist of prePAAR and a TM helix, flank a central TM helix whose C-terminus extends into the solvent, likely indicating the location of the downstream PAAR domain in the full-length effector.

A lack of interpretable electron density prevented modelling of Rhs1’s entire AARxxDxxxH prePAAR motif in our Rhs1_NT-SciW co-crystal structure. However, the DxxxH portion of this motif is part of a short 3₁₀ helix that orients the aspartate and histidine side chains such that they face outward into solvent (Figure 5—figure supplement 1E–G). By contrast, we were able to model the entire prePAAR motif of Tse6_NT and in this case, the motif forms an alpha helix that binds the hydrophobic pocket of an EagT6 protomer. In this structure, the two conserved alanine residues of prePAAR make contact with the EagT6 chaperone, whereas the arginine, aspartate and histidine residues are solvent exposed (Figures 4C and 5F). Remarkably, despite existing in different secondary structure elements, the D11 and H15 prePAAR residues of Tse6 are located in a similar 3D location as their D9 and H13 counterparts in Rhs1 (Figure 5G). It should be noted that the modelled conformation of Tse6_NT appears to be locked into place by crystal packing, suggesting that in solution Tse6’s prePAAR motif may exhibit significant conformational flexibility and can dissociate from EagT6 as is observed for the prePAAR motif of Rhs1 (Figure 5—figure supplement 1H–I). In support of this, we previously showed that addition of detergent disrupts the interaction between EagT6 and Tse6 suggesting that Eag chaperone-effector interactions are labile, likely because chaperone dissociation is required prior to effector delivery into target cells (Quentin et al., 2018). Intriguingly, docking our high-resolution EagT6-Tse6_NT crystal structure into our previously determined lower resolution Tse6-EagT6-VgrG1 cryo-EM map orients the D11 and H15 prePAAR residues of Tse6 in a position that suggests they interact with its PAAR domain (Figure 5H). In sum, our structural analyses of prePAAR shows that this region is likely dynamic, and its mode of interaction varies for class I and class II prePAAR effectors. However, both Eag chaperones bind the N-terminus of their cognate effector such that the conserved aspartate and histidine residues of prePAAR are positioned to potentially be involved in interactions with the downstream PAAR domain, and thus may play a role in effector-VgrG interactions.

To test if prePAAR influences PAAR function, we next conducted mutagenesis analysis on Tse6 because its PAAR-dependent interaction with its cognate VgrG protein, VgrG1a, can be monitored in vivo by western blot. During denaturing electrophoresis, Tse6 appears in two forms: (1) a high-molecular-weight species corresponding to Tse6-VgrG1a complex and (2) a low-molecular-weight species indicative of free Tse6 (Whitney et al., 2015). Deletion of vgrG1a only affects complex formation, whereas deletion of the eagT6 gene results in a substantial reduction in the levels of both species providing a means to differentiate residues involved in effector-chaperone versus effector-VgrG interactions (Quentin et al., 2018). Using this readout, we engineered P. aeruginosa strains expressing Tse6 D11A and H15A single amino acid substitutions and a D11A/H15A double substitution and examined the consequences of these prePAAR mutations on Tse6 interactions. In support of a role in promoting proper folding of PAAR, Tse6-VgrG1a complex formation was substantially reduced in a strain expressing the Tse6^D11A variant and abolished in a strain expressing Tse6^{D11A, H15A} (Figure 6A). We next examined the effect of these mutations on T6SS-dependent delivery of Tse6 into target cells by subjecting these P. aeruginosa strains to growth competition assays against Tse6-sensitive recipients. In agreement with our biochemical data, strains expressing Tse6 harboring a D11A mutation exhibited a substantial reduction in co-culture fitness consistent with an inability of these mutant proteins to form a complex with VgrG1a (Figure 6B).

Figure 6 with 2 supplements see all

Download asset Open asset

To better understand why Tse6’s PAAR domain requires prePAAR for function, we compared its sequence and predicted structure to the X-ray crystal structure of the ‘orphan’ PAAR domain c1882 from E. coli, which does not contain additional components such as TMDs or a toxin domain (Shneider et al., 2013). Interestingly, this analysis suggested that the PAAR domain of Tse6 lacks an N-terminal segment, which, based on the structure of c1882, is potentially important for the proper folding of this domain (Figure 6C). We next extended this structural analysis to include all PAAR domains of the prePAAR effectors that we experimentally confirmed bind Eag chaperones. In all cases, the N-terminal segment of each PAAR domain was missing (Figure 6—figure supplement 1A). We also noted that the prePAAR motif possesses significant sequence homology to the N-terminal segment of c1882, suggesting that even though this stretch of amino acids exists on the opposite side of the first TMD region of Tse6, it may comprise the missing segment of Tse6’s PAAR domain (Figure 6—figure supplement 1B). Lending further support to this hypothesis, when we artificially fused Tse6’s prePAAR motif (residues 1–16) with its PAAR domain (residues 77–163) and generated a structural model, we found that the first 16 residues of Tse6 fill the missing structural elements of Tse6’s PAAR domain (Figure 6—figure supplement 1C). Based on this information, we hypothesized that prePAAR is necessary for PAAR domain function and thus, facilitates VgrG binding. To explore this experimentally, we assessed the effect of prePAAR on VgrG binding in vivo and in vitro. We started by generating two Tse6 mutants expressed from their native locus in P. aeruginosa. The first mutant lacks prePAAR, TMD1, and TMD2 (Tse6^{∆prePAAR, ∆TMDs}), while the second contains prePAAR fused to PAAR and thus lacks only its TMDs (Tse6^∆TMDs). We next assessed the ability of these truncated forms of Tse6 to form a complex with VgrG1a in vivo, as described above. In these experiments, we found that prePAAR and PAAR are together both necessary and sufficient for the formation of Tse6-VgrG1a complexes in vivo (Figure 6D). Of note, the amount of the complex formed by the Tse6^∆TMDs mutant is less than the parent strain, which may be due to unstable structural elements that arose from suboptimal boundaries selected for truncating the effector. We next assessed the formation of this complex in vitro and found that co-incubation of Tse6, EagT6 and VgrG1a after overexpression in E. coli leads to the formation of SDS-resistant Tse6-VgrG1a complexes, whereas doing so with a strain expressing Tse6^{D11A, H15A} does not (Figure 6E and Figure 6—figure supplement 1F). Importantly, these mutations do not affect overall levels of Tse6 in cells or affect its ability to bind to EagT6, indicating that these mutations do not have a global destabilizing effect on Tse6 (Figure 6E). Together, these data suggest that the PAAR domains of prePAAR effectors exist as ‘split PAAR’ due to the presence of N-terminal TMDs.

In orphan PAAR proteins, such as c1882, DxxxH motifs are necessary for Zn²⁺-coordination and are therefore necessary for proper folding of this domain (Shneider et al., 2013). In agreement with this precedent, the conserved histidine residue in the DxxxH portion of Tse6’s prePAAR motif is predicted to be in the same 3D position as the first zinc-coordinating histidine residue of c1882 (Figure 6—figure supplement 1D). To extend this comparison further, we conducted in silico analyses to examine the evolutionary relationship and potential Zn²⁺-binding residues in 564 orphan PAARs and 1765 prePAAR effectors. We found that orphan PAAR sequences are ancestral to split PAAR domains and that while orphan PAAR proteins typically contain a total of four histidine and/or cysteine Zn²⁺-coordinating residues, prePAAR effectors only contain three in their PAAR domain with the fourth likely being provided by the prePAAR motif (Figure 6—figure supplement 1E and Figure 6—figure supplement 2). In support of this prediction, we found that Tse6-VgrG1a complexes formed by the D11A or H15A variants were susceptible to heat treatment under denaturing conditions whereas the wild-type complex remained intact (Figure 6—figure supplement 1F–G). Collectively, our experimental data and informatics analyses indicate that unlike orphan PAAR proteins, which contain all the necessary molecular determinants for proper folding, prePAAR effectors may contain inherently unstable PAAR domains that require a prePAAR motif to ensure their proper folding thus enabling their interaction with their cognate VgrG protein (Figure 7).

Figure 7

Download asset Open asset

As demonstrated in Figure 2, EagR1 and EagT2 from P. protegens interact with and stabilize RhsA and Tne2, respectively, but not vice versa. The inability of these chaperones to interact with non-cognate effectors can be explained by the two starkly contrasting binding modes observed in our co-crystal structures. The class I prePAAR effector and RhsA homolog Rhs1 interacts with its cognate chaperone SciW in an asymmetric manner, whereas the class II effector and Tne2 homolog Tse6 adopts a pseudosymmetric binding mode whereby two separate alpha helices interact with each EagT6 chaperone protomer in a similar location. Our structural analyses suggest that Rhs1 residues F20 and F43 play a critical role in its asymmetric binding mode because the aromatic side chains of these amino acids insert into hydrophobic pockets present in SciW that are lacking in EagT6. The backs of these hydrophobic binding pockets in SciW are formed by residue G108, whereas in EagT6 the equivalent space is occupied by the side chain of F104. As such, the F20 and F43 side chains of Rhs1’s TMD are able to insert into the hydrophobic pockets of SciW but are unable to do so with EagT6 due to a steric clash that would occur with F104. Conversely, the equivalent residue to F20 in the TMD of Tse6 is A26. The small side chain of alanine lacks the volume needed to fill the hydrophobic binding pocket of SciW and thus this interaction would likely contribute less free energy of binding with SciW. We expect that these same structural features contribute to the effector specificity observed for EagR1 and EagT2 as these chaperones also possess an alanine and phenylalanine at the corresponding positions in their hydrophobic pockets. Furthermore, the equivalent positions to Rhs1 in the TMD of RhsA has large hydrophobic residues (I23/L46), whereas the TMD of Tne2 has an alanine residue (A26) that aligns with A26 of Tse6. Although our structural models and bioinformatic analyses of chaperone-effector pairs suggest this pattern of Eag chaperone specificity is widespread, additional biochemical experimentation will be needed to demonstrate the generalizability of these findings.

Another important difference observed in the class I and class II chaperone-effector interactions is the position of the prePAAR region in each chaperone-effector pair. The asymmetric chaperone-TMD interaction allows SciW to ‘shield’ the hydrophobic regions of Rhs1’s N-terminus from the aqueous milieu while also positioning its prePAAR motif in such a way that would allow it to interact with PAAR. By contrast, the pseudosymmetric binding mode of Tse6 to EagT6 appears to be much more dynamic as interpretable electron density for bound Tse6 was only observed when the effector fragment was held in place by interactions with an adjacent complex in the crystallographic asymmetric unit. Consequently, we speculate that even though the Tse6’s prePAAR motif appears less accessible than that of Rhs1, it is likely highly dynamic in solution and thus may adopt a markedly different conformation when in complex with PAAR.

Despite containing a primarily beta-sheet secondary structure, Eag chaperones interact with effector TMDs by mimicking the interactions that occur between the helices of alpha-helical membrane proteins, which, to our knowledge, is a unique mechanism for a chaperone-effector interaction. Upon binding their cognate effector, we hypothesize that Eag chaperones not only shield effector TMDs from solvent but also distort their structure to prevent potential hairpin formation and erroneous insertion into the inner membrane of the effector-producing cell. Because Eag-interacting TMDs have likely evolved to insert into bacterial membranes, a mechanism to prevent self-insertion is probably necessary prior to export. Recent work studying the secretion of TMD-containing effectors of the bacterial type III and type IV secretion systems found that shielding TMDs to prevent inner membrane insertion is a critical step for proper targeting to the secretion apparatus (Krampen et al., 2018). However, membrane protein effectors of these secretion systems have evolved to target eukaryotic, not bacterial, membranes and thus may not require stringent control of TMD conformation prior to export. Indeed, unlike the Eag chaperones presented here, a previously studied T3SS chaperone was shown not to distort the conformation of effector TMDs, whose conformation remained similar before and after membrane insertion (Nguyen et al., 2015).

Current evidence also suggests that Eag chaperones are not secreted by the T6SS (Cianfanelli et al., 2016a; Quentin et al., 2018). This leads to two important questions: (1) when do Eag chaperones dissociate from their cognate effector? (2) How do effector TMDs remain stable after their dissociation from the chaperone? Although no definitive answers exist for either of these questions, given that effector-chaperone interactions are maintained after effector-VgrG complex formation, chaperone dissociation presumably occurs immediately before or during a T6SS firing event. One way this could be accomplished is through chaperone interactions with components of the T6SS membrane and/or baseplate subcomplexes, which might induce chaperone-effector dissociation. The lumen of the T6SS apparatus may also serve to mitigate the susceptibility to degradation observed for prePAAR effectors in the absence of Eag chaperones because the inner chamber of the T6SS apparatus may shield effectors from the protein homeostasis machinery of the cell.

Studies conducted in several different bacteria suggest that many T6SSs export multiple effectors during a single firing event (Cianfanelli et al., 2016a; Silverman et al., 2013; Hood et al., 2010). The precise subcellular location for effector delivery in recipient cells is not well understood; however, it is noteworthy that many effectors that interact with Hcp or C-terminal extensions of VgrG target periplasmic structures such as peptidoglycan or membranes (Flaugnatti et al., 2016; Silverman et al., 2013; Brooks et al., 2013; LaCourse et al., 2018). In contrast, all characterized prePAAR proteins act on cytoplasmic targets by mechanisms that include the hydrolysis of NAD⁺ and NADP⁺, ADP-ribosylation of FtsZ, pyrophosphorylation of ADP and ATP, and deamination of cytidine bases in double-stranded DNA (Whitney et al., 2015; Ting et al., 2018; Ahmad et al., 2019; Mok et al., 2020). This observation supports the proposal that the TMDs in prePAAR effectors function to promote toxin entry into the cytoplasm of target cells (Quentin et al., 2018). Two possibilities for how this occurs include a discrete toxin translocation event that takes place after the initial delivery of effectors into the target cell periplasm or that effectors are delivered directly into the target cell cytoplasm during a T6SS firing event. The large size of Rhs repeat-containing class I prePAAR effectors favors the latter model because it is unlikely that the 2–3 N-terminal TM helices found in these proteins could form a translocation pore for the C-terminal toxin domain. Instead, we propose that the TMDs of prePAAR effectors acts as molecular grease that coats the tip of the VgrG spike allowing it to effectively penetrate target cell membranes during a T6SS firing event. It should be noted that PAAR effectors with nuclease activity that lack N-terminal TMDs have been identified, suggesting that other cell entry mechanisms likely exist, and future work may address whether these proteins have important motifs or domains that permit an alternative translocation mechanism into recipient cells (Pissaridou et al., 2018; Jana et al., 2019).

Crystal structures of single domain PAAR proteins suggest that this domain folds independently and is highly modular (Shneider et al., 2013). Indeed in many instances, PAAR domains appear in isolation (orphan PAAR) and do not require additional binding partners to interact with VgrG (Wood et al., 2019). The initial characterization of PAAR domains established seven groups of PAAR proteins, with the most abundant being orphan PAARs (55% of 1353 PAAR proteins) while the remaining groups represent PAAR proteins with N- and/or C-terminal extensions (Shneider et al., 2013). Our data demonstrate that PAAR domains with N-terminal extensions possess prePAAR, which we predict may be required for the proper folding of the downstream PAAR domain. Based on our structural modelling and sequence alignments, the ability of prePAAR to assist with PAAR domain folding may in part be due to its participation in coordinating the zinc ion found near the tip of this cone-shaped protein. Our sequence analysis also suggests that while orphan PAARs contain four zinc-coordinating histidine and/or cysteine residues, the PAAR domain of prePAAR effectors contains only three, suggesting that the fourth ligand required for tetrahedrally coordinated Zn²⁺ is provided by prePAAR. In this way, the PAAR domain of prePAAR effectors is split into two components, which likely come together to form a structure that can interact with VgrG and undergo T6SS-mediated export. One consequence of this ‘split PAAR’ domain arrangement is that the TMDs are tethered to PAAR via their N- and C-terminus, which would restrict the mobility of the TMDs and ensure their positioning on the surface of PAAR. We speculate that the proper arrangement of prePAAR effector TMDs on the surface of PAAR is likely critical for the ability of the T6SS spike complex to effectively penetrate target cell membranes during a T6SS firing event. Future studies focused on capturing high-resolution structural snapshots of assembled prePAAR-TMD-PAAR complexes will be needed to further support this proposed mechanism.

In summary, our mechanistic dissection of prePAAR effectors and their cognate chaperones has revealed fundamental new insights into bacterial toxin export and membrane protein trafficking. The unique ability of T6SSs to potently target a wide range of bacteria in a contact-dependent manner may permit their use in different biomedical applications, such as the selective depletion of specific bacterial species in complex microbial communities (Ting et al., 2020). An in-depth understanding of the mechanisms that that underlie T6SS effector recruitment and delivery will be of critical importance for such future bioengineering efforts.

Pseudomonas strains used in this study were derived from P. aeruginosa PAO1 and P. protegens Pf-5 (Supplementary file 4). Both organisms were grown in LB medium (10 g L⁻¹ NaCl, 10 g L⁻¹ tryptone, and 5 g L⁻¹ yeast extract) at 37°C (P. aeruginosa) or 30°C (P. protegens). Solid media contained 1.5% or 3% agar. Media were supplemented with gentamicin (30 μg mL⁻¹) and IPTG (250 μM) as needed.

Escherichia coli strains XL-1 Blue, SM10 and BL21 (DE3) Gold or CodonPlus were used for plasmid maintenance and toxicity experiments, conjugative transfer and protein overexpression, respectively (Supplementary file 4). All E. coli strains were grown at 37°C in LB medium. Where appropriate, media were supplemented with 150 μg mL⁻¹ carbenicillin, 50 μg mL⁻¹ kanamycin, 200 μg mL⁻¹ trimethoprim, 15 μg mL⁻¹ gentamicin, 0.25–1.0 mM isopropyl β-D-1-thiogalactopyranoside (IPTG), 0.1% (w/v) rhamnose or 40 μg mL⁻¹ X-gal.

Primers were synthesized and purified by Integrated DNA Technologies (IDT). Phusion polymerase, restriction enzymes and T4 DNA ligase were obtained from New England Biolabs (NEB). Sanger sequencing was performed by Genewiz Incorporated.

Plasmids used for heterologous expression were pETDuet-1, pET29b, and pSCrhaB2-CV. Mutant constructs were made using splicing by overlap-extension PCR and standard restriction enzyme-based cloning procedures were subsequently used to ligate PCR products into the plasmid of interest.

In-frame chromosomal deletion mutants in P. aeruginosa and P. protegens were made using the pEXG2 plasmid as described previously (Hmelo et al., 2015). Briefly, 500–600 bp upstream and downstream of target gene were amplified by standard PCR and spliced together by overlap-extension PCR. The resulting DNA fragment was ligated into the pEXG2 allelic exchange vector using standard cloning procedures (Supplementary file 5). Deletion constructs were transformed into E. coli SM10 and subsequently introduced into P. aeruginosa or P. protegens via conjugal transfer. Merodiploids were directly plated on LB (lacking NaCl) containing 5% (w/v) sucrose for sacB-based counter-selection. Deletions were confirmed by colony PCR in strains that were resistant to sucrose, but sensitive to gentamicin. Chromosomal point mutations or tags were constructed similarly with the constructs harboring the mutation or tag cloned into pEXG2. Sucrose-resistant and gentamicin-sensitive colonies were confirmed to have the mutations of interest by Sanger sequencing of appropriate PCR amplicons.

We previously showed that a D1404A mutation was sufficient to attenuate, but not abolish, the toxicity of RhsA and allows for the cloning of this toxin in the absence of its immunity gene (Tang et al., 2018). Therefore, to assess the toxicity of the full-length effector and a truncated variant, we cloned RhsA^D1404A or RhsA_∆NT^D1404A into the rhamnose-inducible pSCrhaB2-CV vector. These plasmids were co-transformed with an IPTG-inducible pPSV39 vector containing or lacking EagR1, respectively (see Supplementary file 5). Stationary-phase overnight cultures containing these plasmids were serially diluted 10⁻⁶ in 10-fold increments and each dilution was spotted onto LB agar plates containing 0.1% (w/v) L-rhamnose, 250 µM IPTG, trimethoprim 250 µg mL⁻¹ and 15 µg mL⁻¹ gentamicin. Photographs were taken after overnight growth at 37°C.

Stationary-phase overnight cultures of E. coli (DE3) BL21 CodonPlus, P. aeruginosa ∆retS or P. protegens were inoculated into 2 mL or 50 mL LB at a ratio of 1:500, respectively. Cultures were grown at 37°C (E. coli and P. aerugionsa) or 30°C (P. protegens) to OD 0.6–0.8. Upon reaching the desired OD, all samples were centrifuged at 7600 x g for 3 min. The secreted fraction in P. aeruginosa or P. protegens samples was prepared by isolating the supernatant and treating it with TCA (final conc: 10% (v/v)) as described previously (Quentin et al., 2018). The cell pellet was resuspended in 60 µL PBS, treated with 4X laemmli SDS loading dye and subjected to boiling to denature and lyse cells. For experiments examining the stability of Tse6-VgrG1a complexes, P. aeruginosa cells were resuspended in 60 µL PBS and subjected to six freeze-thaw cycles prior to mixing with 2X laemmli SDS loading dye. For preparation of P. protegens and E. coli cell fractions containing his-tagged complexes, cells were resuspended in lysis buffer containing 50 mM Tris-HCl (pH 8.0), 250 mM NaCl, 10 mM imidazole and purified according to the protocol described below (see Protein expression and purification).

A tetracycline-resistant, lacZ-expression cassette was inserted into a neutral phage attachment site (attB) of recipient P. aeruginosa and P. protegens strains to differentiate these strains from unlabeled donors. P. protegens recipient strains also contain a ΔpppA mutation to stimulate T6SS effector secretion to induce a T6SS ‘counterattack’ from P. protegens donor strains (Basler et al., 2013).

For intraspecific competitions between P. aeruginosa or P. protegens donors against isogenic recipient that lack the indicated effector-immunity pairs, stationary-phase overnight cultures were mixed in a 1:1 (v/v) ratio.

Initial ratios of donors:recipients were counted by plating part of the competition mixtures on LB agar containing 40 μg mL⁻¹ X-gal. The remainder of each competition mixture was spotted (10 µL per spot) in triplicate on a 0.45 μm nitrocellulose membrane overlaid on a 3% LB agar plate and incubated face up at 37°C for 20–24 hr. Competitions were then harvested by resuspending cells in LB and counting colony forming units by plating on LB agar containing 40 μg mL⁻¹ X-gal. The final ratio of donor:recipient colony forming units were normalized to the initial ratios of donor and recipient strains.

All plasmids used for co-purification experiments (chaperone-effector pairs, tagged variants of P. protegens proteins and Tse6 prePAAR mutants), RhsA-RhsI-EagR1-VgrG complex for negative-stain EM, Hcp (PFL_6089) and RhsA_∆NT used for antibody development or the SciW, EagT6-Tse6_NT complex and the SciW-Rhs1_NT complex used for crystallization were expressed in E. coli BL21 (DE3) Gold or CodonPlus cells. Important differences in expression strategy used are indicated below.

Chaperone-effector pairs (e, effector; c, chaperone) from: Pseudomonas aeruginosa (e: PA0093, c: PA0094), Salmonella Typhimurium (e: SL1344_0286, c: SL1344_0285), Shigella flexneri (e: SF0266, c: SF3490), Enterobacter cloacae (e: ECL_01567, c: ECL_01566) and Serratia proteamaculans (e: Spro_3017, c: Spro_3016) were co-expressed using pET29b containing the predicted chaperone and pETDuet-1 harboring the full-length effector and its predicted immunity determinant. A similar co-expression strategy was employed for the RhsA_∆NT-RhsI complex, RhsA-RhsI-EagR1-VgrG1 complex, Tse6 and the Tse6 prePAAR variants, Tsi6 and EagT6 (see Supplementary file 5 for details). VgrG1a was expressed in isolation in pETDuet-1 and Hcp (PFL_6089) was expressed in pET29b. For P. protegens, all purified proteins were expressed from their native locus.

For the expression of chaperone-effector pairs and the Tse6 prePAAR mutants, a 1 mL overnight culture of expression strains was diluted in 50 mL of LB broth and grown at 37°C (E. coli) until OD 0.6–0.8. 40 mL overnight cultures were grown for all other of expression strains and were diluted into 2 L of LB broth and grown to OD₆₀₀0.6–0.8 in a shaking incubator at 37°C. For most samples, protein expression was induced by the addition of 1 mM IPTG and cells were further incubated for 4.5 hr at 37°C. Expression of large protein complexes (>150 kDa) in E. coli, such as the chaperone-effector pairs from Salmonella and Enterobacter, RhsA_∆NT-RhsI and RhsA-RhsI-EagR1-VgrG1 complexes were induced at 18°C and incubated at this temperature overnight. One millilitre overnight cultures of P. protegens strains expressing the desired tagged protein was diluted in 50 mL of LB broth and grown at 30°C (P. protegens) unitl OD 0.8. Cells were harvested by centrifugation at 9800 g for 10 min following incubation. For the RhsA-EagR1-VgrG1 complex and the experiments containing Tse6 prePAAR mutants, the pellets for cells expressing the cognate VgrG were combined with the pellets containing effectors, as described above. Pellets from 50 mL culture were resuspended in 3.5 mL lysis buffer (50 mM Tris-HCl pH 8.0, 300 mM NaCl, 10 mM imidazole), whereas those from 2 L of culture were resuspended in 25 mL of lysis buffer prior to rupture by sonication (6 × 30 s pulses, amplitude 30%). Cell lysates were cleared by centrifugation at 39,000 g for 60 min and the soluble fraction was loaded onto a gravity flow Ni-NTA column that had been equilibrated in lysis buffer. Ni-NTA-bound complexes were washed twice with 25 mL of lysis buffer followed by elution in 10 mL of lysis buffer containing 400 mM imidazole. The Ni-NTA purified complex was further purified by gel filtration using a HiLoad 16/600 Superdex 200 column equilibrated in 20 mM Tris-HCl pH 8.0 150 mM NaCl or phosphate buffered saline (for samples used for antibody development only).

Preparation of samples for crystallization sciW (SL1344_0285) was synthesized with codon optimization for E. coli and cloned into the vector pRSETA with the restriction sites NdeI/HindIII (Life Technologies). This construct includes an N-terminal 6-his tag and an HRV 3C protease cleavage site (MGSSHHHHHHSSDLEVLFQGPLS). SciW-Rhs1_NT and EagT6-Tse6_NT complexes were co-expressed using pETDUET-1. Note that the EagT6 construct has a C-terminal VSV-G tag (see Supplementary file 5). Cells were grown in LB broth to OD₆₀₀ 0.6 at 37°C at which point protein expression was induced by the addition of 1 mM IPTG. The temperature was reduced to 20°C and cultures were allowed to grow overnight. Cells were harvested by centrifugation and resuspended in lysis buffer followed by lysis with an Emulsiflex-C3 (Avestin). The lysate was cleared by centrifugation at 16,000 rpm for 30 min and the supernatant passed over a nickel NTA gravity column (Goldbio) followed by washing with 50 column volumes of chilled lysis including PMSF, DNase I, and MgCl₂. Proteins were eluted with five column volumes elution buffer then purified by gel-filtration using an SD75 16/60 Superdex gel filtration column equilibrated in gel-filtration buffer (GF) with an AKTA pure (GE Healthcare). For SciW, after affinity purification the protein was dialyzed in GF buffer O/N at 4°C and the His-tag removed during dialysis using HRV 3C protease. The digested SciW was passed over a nickel NTA gravity column and the flow through was collected. SciW was further purified using an SD75 16/60 Superdex gel filtration column equilibrated in GF buffer.

The buffers used were as follows: SciW lysis buffer (20 mM Tris pH 7.5, 500 mM NaCl, 20 mM imidazole); SciW elution buffer (20 mM Tris pH 7.5, 500 mM NaCl, 500 mM imidazole); SciW GF buffer (20 mM Tris pH 7.5, 250 mM NaCl, 1 mM 2-Mercaptoethanol); SciW-Rhs1_NT and EagT6-Tse6_NT complexes lysis buffer (20 mM Tris pH 8.0, 150 mM, 25 mM imidazole); elution buffer (20 mM Tris pH 8.0, 150 mM, 500 mM imidazole); and GF buffer (20 mM Tris pH 8.0, 150 mM NaCl, 1 mM 2-Mercaptoethanol).

SciW was concentrated to 7, 14, and 22 mg mL⁻¹ for initial screening using commercially available screens (Qiagen) by sitting-drop vapor diffusion using a Crystal Gryphon robot (Art Robbins Instruments). The crystallization conditions for SciW were 22 mg mL⁻¹with a 1:1 mixture of 0.1 M Tris HCL pH 8.5, 25% (v/v) PEG 550 MME at 4°C. EagT6-Tse6_NT complex was concentrated to 5, 10 and 20 mg mL⁻¹ and screened for crystallization conditions as per SciW. The final crystallization conditions were 20 mg mL⁻¹ with a 1:1 mixture of 0.2M Magnesium chloride, 0.1M Bis-Tris pH 5.5, and 25% (w/v) PEG 3350 at 4°C. SciW-Rhs1_NT complex was concentrated to 15, 20, and 25 mg mL⁻¹ and screened for crystallization as per SciW. The crystallization conditions were 25 mg mL⁻¹ protein with a 1:1 mixture of 0.2M Ammonium sulfate, 0.1M Bis-Tris pH 5.5, and 25% (w/v) PEG 3350 at 4°C.

Diffraction data from crystals of SciW and EagT6-Tse6_NT complex were collected in-house at 93K using a MicroMax-007 HF X-ray source and R-axis 4++ detector (Rigaku). Diffraction data from SciW-Rhs1_NT crystals were collected at the Canadian Light Source at the Canadian Macromolecular Crystallography Facility Beam line CMCF-ID (08ID-1). SciW crystals were prepared by cryo-protection in mother liquor plus 38% PEG 550 MME and flash freezing in liquid nitrogen. Crystals of EagT6-Tse6_NT and SciW-Rhs1_NT complexes were prepared in the same manner with increasing the concentration of PEG3350 to 35–38%. All diffraction data were processed using XDS (Kabsch, 2010). Phases for SciW were determined by the molecular replacement-single anomalous diffraction (MR-SAD) technique. A home-source data set was collected from SciW crystals soaked in cryo-protectant containing 350 mM NaI for one-minute before flash freezing. EagT6 (PDB: 1TU1) was used as a search model and phases were improved by SAD using the Phenix package (Adams et al., 2010). Phases for both the EagT6-Tse6_NT and SciW-Rhs1_NT complexes were obtained by molecular replacement using EagT6 (PDB: 1TU1) and SciW as search models, respectively, with the Phenix package. Initial models were built and refined using Coot, Refmac and the CCP4 suite of programs, Phenix, and TLS refinement (Emsley et al., 2010; Murshudov et al., 1997; Winn et al., 2011; Winn et al., 2001). Data statistics and PDB codes are listed in Table 1. The ligands identified included polyethylene glycol (PEG) for SciW and sulfate ions for the SciW-Rhs_NT complex. Additionally, the side-chain of residue C11 in SciW chain B was observed to be covalently bound to 2-Mercaptoethanol to form CME (S,S-(2-hydroxyethyl)thiocysteine). The coordinates and structure factors have been deposited in the Protein data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, NY (http://www.rcsb.org). Molecular graphics and analysis were performed using Pymol (Schrödinger, LLC) and UCSF Chimera (Pettersen et al., 2004).

Western blot analyses of protein samples were performed as described previously for rabbit anti-Tse1 (diluted 1:5,000; Genscript), rabbit anti-FLAG (diluted 1:5,000; Sigma), rabbit anti-VSV-G (diluted 1:5,000; Sigma), rabbit anti-Hcp1 (P. aeruginosa) (diluted 1:5,000, Genscript) and detected with anti-rabbit horseradish peroxidase-conjugated secondary antibodies (diluted 1:5,000; Sigma) (Ahmad et al., 2019). Rabbit anti-Hcp (P. protegens) was used at a 1:5000 dilution. Western blots were developed using chemiluminescent substrate (Clarity Max, Bio-Rad) and imaged with the ChemiDoc Imaging System (Bio-Rad).

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

The authors thank Jianhua Zhao for electron microscopy expertise, Sarah Trilesky and Matthew Walker for their assistance with cloning and protein purification and Peter Stogios, Seemay Chou, James Holton and Atanas Radkov for crystallography expertise. SA and KKT were supported by Ontario Graduate Scholarships and AGM holds a Cisco Research Chair in Bioinformatics. Part of the research described in this paper was performed using beamline 08ID-1 at the Canadian Light Source, a national research facility of the University of Saskatchewan, which is supported by the Canada Foundation for Innovation (CFI), the Natural Sciences and Engineering Research Council (NSERC), the National Research Council (NRC), the Canadian Institutes of Health Research (CIHR), the Government of Saskatchewan, and the University of Saskatchewan. This work was supported by the Max Planck Society (to SR) and by grants from CFI (34531 to AGM and 37841 to GP), NSERC (RGPIN-2017–05350 to JCW and RGPIN-2018–04968 to GP) and CIHR (PJT156129 to JCW and PJT156214 to AGM). Computer resources were supplied by the McMaster Service Lab and Repository computing cluster, funded in part by grants to AGM from CFI and Compute Canada (http://www.computecanada.ca).

1. Received: September 4, 2020
2. Accepted: December 14, 2020
3. Accepted Manuscript published: December 15, 2020 (version 1)
4. Accepted Manuscript updated: December 17, 2020 (version 2)
5. Version of Record published: December 30, 2020 (version 3)

© 2020, Ahmad 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.