Skip to main content
Proceedings of the National Academy of Sciences of the United States of America logoLink to Proceedings of the National Academy of Sciences of the United States of America
. 2006 Aug 7;103(33):12353–12358. doi: 10.1073/pnas.0603433103

Competitive recruitment of the periplasmic translocation portal TolB by a natively disordered domain of colicin E9

Steven R Loftus *,, Daniel Walker *,, Maria J Maté *,†,, Daniel A Bonsor *, Richard James §, Geoffrey R Moore , Colin Kleanthous *,
PMCID: PMC1567883  PMID: 16894158

Abstract

The natively disordered N-terminal 83-aa translocation (T) domain of E group nuclease colicins recruits OmpF to a colicin-receptor complex in the outer membrane (OM) as well as TolB in the periplasm of Escherichia coli, the latter triggering translocation of the toxin across the OM. We have identified the 16-residue TolB binding epitope in the natively disordered T-domain of the nuclease colicin E9 (ColE9) and solved the crystal structure of the complex. ColE9 folds into a distorted hairpin within a canyon of the six-bladed β-propeller of TolB, using two tryptophans to bolt the toxin to the canyon floor and numerous intramolecular hydrogen bonds to stabilize the bound conformation. This mode of binding enables colicin side chains to hydrogen-bond TolB residues in and around the channel that runs through the β-propeller and that constitutes the binding site of peptidoglycan-associated lipoprotein (Pal). Pal is a globular binding partner of TolB, and their association is known to be important for OM integrity. The structure is therefore consistent with translocation models wherein the colicin disrupts the TolB–Pal complex causing local instability of the OM as a prelude to toxin import. Intriguingly, Ca2+ ions, which bind within the β-propeller channel and switch the surface electrostatics from negative to positive, are needed for the negatively charged T-domain to bind TolB with an affinity equivalent to that of Pal and competitively displace it. Our study demonstrates that natively disordered proteins can compete with globular proteins for binding to folded scaffolds but that this can require cofactors such as metal ions to offset unfavorable interactions.

Keywords: native disorder, thermodynamics, toxin


It is estimated that ≈4% of eubacterial and >30% of eukaryotic proteins contain stretches of >30 aa that are natively (intrinsically) disordered (1), many with functions yet to be defined. Native disorder in proteins occurs in regions of low sequence complexity, typically with a high glycine content and a prevalence of charged or polar residues and few bulky hydrophobic amino acids (2, 3). Their composition precludes the formation of a hydrophobic core or a stable three-dimensional fold, the protein instead remaining partially or completely unstructured under physiological conditions. The extent of native disorder in a protein can vary. For example, the entire interaction domain of the activator for thyroid hormone retinoid receptor is natively disordered (4), whereas native disorder is restricted to the N-terminal region of the eukaryotic translation initiation factor eIF4E (5).

Natively disordered states are important in many biological and disease processes; including, transcription, translation, cell cycle regulation, endocytosis, intracellular signaling, host–pathogen interactions, and cancer. For the most part, the role of natively disordered proteins is to bind other macromolecules (nucleic acids, membranes, and proteins) often expediting the assembly of larger molecular complexes. As studies of natively disordered proteins have accumulated, so the biological advantages inherent to such states have begun to emerge (2, 6). (i) Natively unfolded proteins undergo disorder–order transitions on binding to folded scaffolds that have characteristic thermodynamic signatures, where the significant entropic penalties associated with binding-induced folding are counterbalanced by a favorable enthalpy resulting in weak but specific complexation. (ii) Natively unfolded proteins can contain many linear binding epitopes allowing a relatively small domain to engage in multiple associations that if recapitulated by a folded protein would require a much larger polypeptide chain (7). (iii) “Fly-casting” facilitates recruitment of binding partners as natively unfolded proteins have a larger Stokes radius than a similarly sized globular protein (8). Through the study of the cellular uptake mechanism of colicins, we identify a functional paradigm for a natively disordered state that we term “competitive recruitment,” in which the protein folds on binding its target in the process competing with a prefolded globular protein.

Colicins are plasmid-encoded, SOS-induced protein antibiotics that specifically target Escherichia coli cells (9), with most Gram-negative microorganisms producing such bacteriocins. Typically killing cells by depolarizing the inner membrane (IM) or degrading nucleic acid in the cytoplasm, colicins first bind an outer membrane (OM) receptor before translocating into the cell via Tol or Ton proteins within the periplasm (10). We have been investigating the 60-kDa nuclease colicins E2–E9 that bind to the vitamin B12 receptor BtuB in the OM and translocate to the cytoplasm via the Tol system (11). E colicins share a high degree of sequence conservation (>90%) in the regions involved in receptor binding and translocation (≈460 aa), organized as a central coiled-coil receptor binding domain and an N-terminal translocation (T) domain, with the cytotoxic RNase or DNase located at the C terminus (12, 13). The N-terminal 83 aa of an E colicin T-domain are natively disordered as shown by three observations. First, this region is predicted to be disordered by predictive algorithms such as PONDR (14), Foldindex (15), and DISOPRED2 (1). Second, the crystal structure of ColE3 failed to resolve electron density for this region (13). Third, heteronuclear NMR studies of the homologous ColE9 have demonstrated this region to be unstructured and highly flexible (16, 17).

We showed recently that the natively disordered T-domain of E colicins recruits the porin OmpF in the OM to a preformed, high-affinity colicin–BtuB complex (18). It is thought that the T-domain is then lowered into the periplasm through the lumen of OmpF from where it must locate the 45-kDa translocation portal TolB to trigger cell entry (1921). In the present study, we set out to determine the molecular basis for TolB recruitment by ColE9 in the periplasm.

Results

Residues 32–47 in the Natively Unfolded T-Domain of ColE9 Constitute the TolB Binding Epitope.

A “TolB box” (35DGSGW39) has previously been described that is part of the natively unfolded T-domain where mutations (of non-glycine residues) abolish colicin toxicity (21, 22). More recently, the potential TolB binding site has been extended by mutagenesis and NMR experiments (17, 23), but it is still not clear whether the site is a contiguous or discontiguous epitope or whether regions outside the natively disordered domain are involved. Two complementary approaches were taken to delineate precisely the binding site in ColE9. The first involved photocrosslinking using the heterobifunctional crosslinker 4-(N-maleimido)benzophenone (MBP) (24). MBP was attached to unique cysteines that replaced residues in the disordered T-domain (Arg-7, Ser-15, Thr-25, Val-29, Ser-34, Gly-38, Ser-40, Glu-42, Trp-46, Gly-51, Trp-56, Asn-64, Ser-75, and Ser-82) and irradiated with UV light when mixed with TolB in vitro. MBP-labeled ColE9 mutants could be readily crosslinked to TolB, but surprisingly these were distributed on either side of the previously identified TolB box sequence, appearing instead to form a photocrosslink footprint that spanned residues 34–46 (see Fig. 5a, which is published as supporting information on the PNAS web site).

The second approach utilized deletion analysis, in combination with gel-filtration chromatography and isothermal titration calorimetry (ITC) as assays for TolB binding (see Materials and Methods for details). From these studies, we deduced that (i) the first 61 aa of the disordered T-domain are sufficient for stoichiometric TolB binding; (ii) of the four deletions that were generated (T31–61, T1–44, T1–49, and T1–54), only one (T1–44) did not bind TolB (data not shown); and (iii) two synthetic peptides, Tpep30–50 and Tpep32–47, aimed at narrowing down the binding site on ColE9 yet further, bound TolB with the same equilibrium dissociation constant as intact colicin (Kd ∼ 1 μM) and with very similar thermodynamic parameters. ITC data for Tpep32–47 binding TolB are shown in Fig. 5b and compared with wild-type ColE9 in Table 1. The data show that complex formation is enthalpically driven and has a negative entropy, which is typical of binding-induced disorder–order transitions of natively disordered proteins. Critically, this sequence approximates the region of the colicin where MBP photocrosslinking to TolB was lost. Hence, the TolB binding site on the enzymatic ColE9 is contained within a linear 16-residue epitope spanning residues 32–47.

Table 1.

Thermodynamic parameters for TolB binding ColE9 and Pal obtained by isothermal titration calorimetry at 20°C

Protein ΔH, kcal/mol TΔS, kcal/mol ΔG, kcal/mol Kd, μM n
ColE9 −15.7 (±0.1) −7.7 (±0.1) −8.0 (±0.1) 1.00 (±0.09) 0.99 (±0.05)
ColE9 Tpep32–47 −14.7 (±0.1) −6.6 (±0.1) −8.1 (±0.0) 0.92 (±0.03) 1.07 (±0.03)
Pal −7.1 (±0.0) 2.7 (±0.0) −9.8 (±0.0) 0.047 (±0.002) 1.06 (±0.04)

Data generated from the titration of ColE9, ColE9 Tpep32–47, or Pal into TolB. Errors shown are from duplicate measurements. Experiments were conducted in 50 mM potassium phosphate buffer (pH 7.5) containing 50 mM NaCl and 10 mM EDTA. For further details, see Supporting Materials and Methods.

Structure of the ColE9 Tpep32–47–TolB Complex.

The structure of E. coli TolB has been described (25, 26), but as yet no high-resolution structure for this or any other Tol or Ton protein bound to a bacteriocin has been reported. To understand how ColE9 binds TolB in the periplasm, the structure of the ColE9 Tpep32–47–TolB complex was solved to a resolution of 2.0 Å (for data refinement statistics, see Table 2 and Supporting Materials and Methods, which are published as supporting information on the PNAS web site). The overall structure of the complex is shown in Fig. 1a, with the electron density of the bound ColE9 Tpep32–47 peptide shown in Fig. 1b. TolB is composed of two domains, an N-terminal α/β domain (residues 34–165) and a C-terminal, six-bladed β-propeller domain (residues 166–431). ColE9 Tpep32–47 binds to the “top” side of the β-propeller domain, distal to the N-terminal domain, with 15 of the 16 residues (3246) contacting TolB. Peptide binding does not result in any major structural changes to TolB, the rms deviation for Cα atoms relative to the native structure of TolB is 0.64 Å, and the accessible surface area that is buried on complex formation is 1,408 Å2. As with most β-propellers, a narrow channel runs through the domain, which in the case of TolB contains two Ca2+ ions (Fig. 1a); calcium had been included in the crystallization conditions. Only three TolB residues are involved in binding the Ca2+ ions (the side chains of Asp-337 and Asp-339 and the carbonyl of Val-340), with water molecules occupying the remaining coordination sites. Although neither Ca2+ ion is in direct contact with ColE9 Tpep32–47, they nevertheless influence colicin binding, a point we return to below.

Fig. 1.

Fig. 1.

Structure of the ColE9 Tpep32–47–TolB complex at 2-Å resolution. (a) Overall topology of the complex showing the colicin binding site in the β-propeller domain of TolB distal to the N-terminal α/β-domain. Also visible are the two Ca2+ ions localized in the β-propeller channel. (b) Electron density of the ColE9 Tpep32–47 peptide within the complex. Structure parameters for the complex are listed in Table 2. (c) Magnified view of the TolB binding surface where the atoms of the peptide are shown. The figure emphasizes the cyclizing hydrogen bond between Gly-32 and Asn-44 and the constriction between the two bound tryptophans that forces the chain toward the β-propeller channel.

The colicin binds within a canyon in the β-propeller domain blocking the central channel (Fig. 1c). Searches of the Dali server (27) failed to identify structures similar to that of ColE9 Tpep32–47, which adopts a distorted hairpin conformation. The peptide is anchored to the TolB canyon floor by the indole side chains of the two tryptophans (Trp-39 and Trp-46) with additional hydrophobic contacts made by Pro-45 and Ala-33 (Fig. 1c). The gap separating the hydrophobic docking sites for the tryptophans can only accommodate three residues in extended conformation. As six amino acids separate the two tryptophans, the chain becomes compressed, the resulting turn pushing colicin residues toward the channel at the base of the canyon. Comparing the protein–protein interaction of the ColE9 Tpep32–47–TolB complex with other β-propeller complexes suggests that its architecture is novel (28). β-propellers usually bind other proteins along their external surfaces, whereas the central cavity is generally where enzyme prosthetic groups are located. ColE9 Tpep32–47 binding is a combination of these binding modes, contacting both the external surface and the central cavity of the β-propeller.

Another unusual feature of the complex is that relatively few direct hydrogen bonds stabilize the ColE9 Tpep32–47–TolB complex. Discounting solvating waters, the colicin forms 21 hydrogen bonds; only 5 are direct intermolecular hydrogen bonds, a further 4 involve a bridging water molecule, and 12 are intramolecular. The importance of intramolecular contacts is most apparent around the TolB box sequence (35DGSGW39) of the colicin T-domain (Fig. 2a). Only Trp-39 from this sequence motif is involved in hydrogen bonding to residues in TolB, Asp-308 and Gln-336, the latter mediated via a bridging water molecule to the backbone carbonyl of the tryptophan. Instead, the TolB box forms a network of intramolecular hydrogen bonds that holds the peptide in its distinctive conformation within the confines of the β-propeller canyon. Colicin E9 Asp-35, in particular, is critical in this role because it forms three intramolecular hydrogen bonds, two to Ser-37 and one to Ser-40. Similarly, Ser-37 is hydrogen-bonded to the main-chain of Ser-40, whereas Asn-43 from further along the peptide forms a single hydrogen bond with the amide nitrogen of Gly-38 (Fig. 2a). Finally, the side chain of Asn-44 hydrogen-bonds the backbone nitrogen of the N-terminal glycine, Gly-32, effectively cyclizing the peptide in the combining site (Fig. 1c).

Fig. 2.

Fig. 2.

Hydrogen bonding networks in the ColE9 Tpep32–47–TolB peptide complex. (a) Stereoview of ColE9 TolB box residue interactions. The majority of the hydrogen bonds are intramolecular, involving backbone atoms of the TolB box (32DGSGW39) and the side chains of Asp-35, Ser-37, Ser-40, and Asn-43 of the colicin. These hold the colicin in its hairpin conformation and position Trp-39 in a pocket within the TolB canyon. (b) Stereoview of colicin interactions in and around the TolB channel. Two types of interactions are evident: direct hydrogen bonds involving ColE9 (Ser-41 and Glu-42) and TolB residues (Ser-205, His-246, and Glu-293), and water-mediated interactions that help stabilize the conformation of T-domain residues around the channel.

A consequence of the turn between the two tryptophans in ColE9 Tpep32–47 is that Ser-40, Ser-41, and Glu-42 project into the channel at the base of the canyon and it is here that the majority of hydrogen bonds with TolB are formed (Fig. 2b). Ser-41 hydrogen bonds to TolB Glu-293, whereas Glu-42 forms hydrogen bonds with three TolB residues, Ser-205, Ala-249, and His-246. Glu-42 also contacts Thr-292 indirectly through a water-mediated hydrogen bond. A network of water-mediated interactions helps stabilize the orientation of colicin side chains within the channel; this includes TolB Ser-205 in the wall of the channel, the backbone atoms of ColE9 Ser-41 and Glu-42, and the side chain of Ser-40, which hydrogen-bonds the amide nitrogen of Glu-42.

The results of previous ColE9 T-domain mutagenesis data can now be rationalized in the context of the TolB-ColE9 Tpep32–47 complex structure (2123). Mutations fall into two categories, those that abolish binding and those that result in weak binding to TolB. The first group includes Asp-35, Ser-37, Trp-39, Glu-42, and Trp-46, all of which the structure shows play critical roles in stabilizing the bound conformation of the colicin, either through hydrophobic interactions or by forming multiple hydrogen bonds. Mutations at these sites also render the colicin completely inactive as a toxin. The second group includes Ser-40, Ser-41, Asn-43, and Asn-44, of which all but Ser-41 are involved in forming single hydrogen bonds and where mutations of some (Ser-41 and Asn-43) retain partial biological activity.

The Natively Disordered ColE9 T-Domain Binds at the Pal Site on TolB.

In vivo crosslinking has established that TolB and Pal form a complex at the OM in the bacterial periplasm (29), where Pal is localized by virtue of its N-terminal lipoyl tether. Using a suppressor screen, Lazzaroni and coworkers (30) identified a number of residues in TolB that potentially comprise a Pal binding site. When mapped onto the structure of the ColE9 Tpep32–47–TolB complex, it becomes clear that there is significant overlap with the ColE9 T-domain binding site (Fig. 3), which includes residues within the TolB β-propeller channel that form hydrogen bonds with the colicin (for example, His-246 and Thr-292). However, there are no biochemical or structural data showing that the Pal and E colicin binding sites on TolB overlap, and thus we investigated this in more detail using mutagenesis. To evaluate the impact of any TolB mutations on the ability to bind Pal, the soluble 13-kDa periplasmic domain of E. coli Pal (residues 65–173) was overexpressed and purified (see Materials and Methods). Far-UV circular dichroism spectroscopy showed that the protein contained ≈40% α-helix and 15% β-sheet, which is in good agreement with the x-ray structure of E. coli Pal (PDB entry 1OAP). Initially, we used analytical gel-filtration chromatography to confirm the domain bound TolB in a 1:1 complex (data not shown). ITC established the Kd for the complex as 47 nM with binding being enthalpically and entropically favorable (Table 1), which contrasts with the energetics of the disordered ColE9 T-domain binding to TolB under the same conditions.

Fig. 3.

Fig. 3.

The ColE9 T-domain recruits TolB by binding at the Pal site. Depicted is a section through the β-propeller domain surface showing residues implicated in binding Pal (colored purple), identified by the second site suppressor screen of Ray et al. (30), in the context of bound ColE9 Tpep32–47 (ribbon).

A TolB Thr292Ile mutation was isolated in the study of Ray et al. (30) that had a large effect in the phenotypic screens that were used. We generated this mutation as well as an alanine mutant at this position. We also substituted His-246 for alanine. Mutant TolB proteins were assayed for ColE9 T-domain and Pal binding in vitro by ITC, and bacterial strains carrying these mutations were tested for their susceptibility toward ColE9 and SDS (see Fig. 6, which is published as supporting information on the PNAS web site). Growth in the presence of 2% SDS provides a convenient assay for OM integrity and hence TolB function in vivo. Our phenotypic assays made use of an E. coli K-12 ΔtolB strain that is unable to grow in the presence of SDS and is resistant to enzymatic colicins. The strain was transformed with an arabinose-inducible plasmid encoding wild-type and mutant TolBs targeted to the periplasm by the gIII signal sequence.

Both His246Ala and Thr292Ile TolB-expressing bacteria were completely resistant to the colicin and showed no toxin binding, whereas Thr292Ala TolB showed 100-fold reduced binding and diminished colicin toxicity relative to the wild-type control (Fig. 6). The TolB mutations had an even greater impact on Pal binding because none were able to bind Pal in vitro (Fig. 6), which is unlikely to be due to misfolding of the mutants because all of the purified proteins had far-UV CD spectra identical to wild-type TolB (data not shown). Moreover, unlike wild-type TolB-containing E. coli, the three mutant strains showed little growth in the presence of 2% SDS (Fig. 6) demonstrating that all had compromised TolB function in vivo. We conclude that the binding sites for ColE9 and Pal on TolB must overlap to a significant degree and that, as anticipated, disruption of the Tol–Pal interaction destabilizes the bacterial OM.

Ca2+ Ions Enhance the Association of the Natively Disordered T-Domain of ColE9 with the TolB β-Propeller.

The data in Fig. 3 show how TolB-dependent colicins associate at the Pal site, supporting a competitive binding mode for these bacteriocins. With the realization that the β-propeller channel of TolB binds two Ca2+ ions, we investigated whether calcium (or magnesium) could influence the binding of ColE9 T-domain (Tpep30–50) and Pal for TolB by ITC. Because of buffer/metal ion compatibility issues, we opted for Hepes buffer at 20°C, pH 7.5, 50 mM NaCl, and containing either 5 mM EDTA or 1 mM CaCl2: the measured Kds for the TolB–Pal complex were 27 and 90 nM, respectively, whereas those for TolB-ColE9 T-domain were 900 and 84 nM, respectively (data not shown). Ca2+ ions therefore weaken Pal binding by 3-fold but enhance ColE9 T-domain binding by >10-fold, with similar changes seen for Mg2+ ions. The ability of the colicin T-domain to compete directly with Pal for binding to TolB was shown by MBP-photocrosslinking experiments. When Pal was present in the incubations, T-domain could only be crosslinked to TolB in the presence of Ca2+ ions (Fig. 4a). We conclude that ColE9 T-domain can compete with Pal for binding to TolB but that this requires Ca2+ or Mg2+ ions to be bound within the β-propeller channel, which render their affinities for TolB essentially equivalent.

Fig. 4.

Fig. 4.

Ca2+ ions modulate the binding of the ColE9 T-domain to TolB. (a) MBP-photocrosslinking of ColE9 Trp56Cys T-domain is inhibited by Pal (2 μM) in the absence of Ca2+ ions. TolB and ColE9 (≈2 μM) were crosslinked as described in Materials and Methods. The addition of 1 mM Ca2+ ions abolished the ability of Pal to inhibit ColE9 T-domain crosslinking to TolB. The arrow indicates the migration position of the crosslinked adduct. The figure also highlights how the electrostatic potentials of the TolB β-propeller canyon are influenced by calcium ions. (b) The surface is negatively charged in the absence of Ca2+ ions weakening the affinity for ColE9 T-domain. (c) Ca2+ binding within the channel results in a more positively charged surface, which binds ColE9 Tpep32–47 >10-fold more tightly (see text for details). Bound ColE9 Tpep32–47 peptide (shown as yellow ribbon in b and c) has a net charge of −2 at neutral pH due to Asp-35 and Glu-42.

Discussion

Colicin Recruitment of the Translocation Portal TolB Requires a Natively Unfolded N Terminus.

The 16-residue TolB binding site in ColE9 has been shown by NMR to be unstructured and flexible in solution but also to contain regions of reduced mobility relative to other parts of the disordered T-domain that take the form of local hydrophobic clusters centering on the tryptophan side chains of Trp-39 and Trp-46 (16, 17). Similar clustering is seen around bulky hydrophobic residues in urea-denatured states of proteins (31, 32). NMR has further demonstrated that association of the T-domain with TolB perturbs these hydrophobic clusters, with sequences flanking these regions remaining unaffected (16, 33). The present structure now shows how the two tryptophans act as the main attachment points of the colicin on the TolB surface. With the exception of the C-terminal glycine, every residue of the ColE9 T-domain peptide is involved in stabilizing the bound conformation, most by intra- or intermolecular hydrogen bonds (Fig. 2). Consequently, the underlying thermodynamics of binding reflect not just binding-induced folding within the TolB canyon but also disruption of the hydrophobic clusters in water. Although binding is weak, specific recognition of TolB by the natively disordered T-domain is emphasized by the fact that no crosslinking could be detected within the binding site, whereas adjacent sequences were readily crosslinked (Fig. 5a), indicating that MBP-labeling likely abolished binding to TolB.

Because the soluble periplasmic domain of Pal is a globular protein, the question arises as to why the colicin T-domain should be natively unfolded before binding TolB. This requirement for native disorder is undoubtedly a consequence of the colicin cellular uptake mechanism because the narrow lumen of an OmpF monomer is thought to be the entry point for the unfolded polypeptide into the periplasm before translocation across the OM (19). Thereafter, association with TolB would be expedited because OmpF is known to be closely associated with Tol proteins (34). The role of native disorder in colicin translocation is further illustrated by the pore-forming colicin N, which uses OmpF both as a receptor and translocator to the periplasm where the toxin binds TolA. The first 90 aa of ColN are natively disordered, of which 27 residues comprise the TolA binding site (35, 36). Native disorder is therefore a mechanistic prerequisite for Tol-dependent colicin translocation, allowing the toxin to penetrate the OM via trimeric porins and contact Tol proteins in the periplasm.

Colicins Disrupt Host Protein–Protein Interactions in the Periplasm.

Tol and Ton proteins are essential periplasmic systems in Gram-negative bacteria that have been subverted by group A (such as ColN and ColE2-E9) and group B (such as ColB and ColIa) colicins, respectively, for translocation across the OM. TolAQR are orthologues of TonB/ExbB/ExbD; both are IM systems coupled to the proton-motive force, and each (TolA and TonB) traverse the periplasm. The physiological role of the Ton complex is to energize the OM for active transport of ligands through nutrient receptors, such as FepA (37), whereas that of Tol is less well defined but involves maintenance of outer-envelope integrity. It is thought that group B colicins disrupt protein–protein interactions in the periplasm of the host because they possess a “TonB box” sequence at their N terminus equivalent to a sequence at the C terminus of OM nutrient receptors that signals to TonB its liganded status (37). The present work shows that such disruption extends to the group A colicin ColE9 where the disordered T-domain associates at the Pal site on TolB. The TonB box of group B colicins forms a continuous β-sheet with the C-terminal domain of TonB, most likely equivalent to that of a ligand-bound OM receptor (38, 39). In contrast, ColE9 adopts a hairpin conformation in the confines of the β-propeller domain of TolB. We conclude that both Ton- and Tol-dependent colicins interfere with host protein–protein interactions but in very different ways. The outcome is ostensibly the same, however, because each results in the disruption of a protein–protein complex near the OM allowing the toxin to tap into the proton-motive force across the IM via proteins that traverse the periplasm.

Tol-Dependent Colicin Translocation Across the OM.

TolB and Pal form a subcomplex that connects essential OM proteins such as Lpp and OmpA to the IM via the TolAQR complex (40). Disruption of any of these associations through mutation causes instability of the OM, with the cells becoming periplasmically leaky and hypersensitive to detergents and antibiotics (41). Such mutations also display characteristic tol phenotypes whereby they are tolerant of colicin intoxication (42), the same phenotype exhibited by the TolB mutations generated in this study (Fig. 6). Bénédetti and coworkers (43, 44) have shown that colicin T-domain sequences secreted directly into the periplasm (via the sec pathway) generate the same pleiotropic effects as tol mutations and that this results in decreased in vivo crosslinking between TolB and Pal. This finding suggested that Tol-dependent colicins might perturb host protein–protein interactions to cause local disruption of the OM and so expedite their entry into the periplasm (45). Our study provides the first structural and biophysical data in support of such a mechanism. The colicin thus bound to TolB might then be able to traverse the destabilized membrane by “hitching a ride” on the energized TolAQR complex, which is known to bind TolB at its N-terminal α/β domain (46).

Competitive Recruitment and the Role of Divalent Cations.

Competition between natively disordered proteins for a common scaffold has been reported previously. The transcriptional-adapter zinc finger-1 (TAZ1) domain of the transcriptional regulator CBP is bound by disordered domains of HIF1a and CITED2, which fold around TAZ1 (reviewed in ref. 6). To our knowledge, there have been no previous reports of a natively disordered protein competing with a globular protein for binding to a folded scaffold. The degree of overlap awaits structural resolution of the TolB–Pal complex but is likely to be substantial given the dramatic in vitro and in vivo effects of the single TolB β-propeller mutations on both ColE9 and Pal binding. Competition experiments reinforce this view because the ColE9 T-domain can dislodge Pal from its complex with TolB but only in the presence of Ca2+ ions (Fig. 4a), which increase its binding affinity for TolB by >10-fold. Ca2+ ions switch the electrostatic surface potential of the β-propeller channel from negative to positive, favoring binding of the negatively charged ColE9 T-domain peptide (pI ∼ 3.7) (Fig. 4 b and c) while having minor impact on the binding of Pal, which has no net charge at neutral pH.

These observations raise the question of whether the effects of divalent cations are physiologically relevant. Because the function of TolB has yet to be established, this remains unclear, but we note that Ca2+ and Mg2+ ions are known to be concentrated in the periplasm relative to the cytoplasm and external environment (47, 48), implying that alkaline earth metal ions are bound to TolB under physiological conditions. We speculate that these metal sites are involved in native TolB function or stability, their presence having the secondary effect of expediting its competitive recruitment by the natively disordered T-domain of a translocating colicin.

Materials and Methods

Bacterial Strains and Plasmids.

Plasmid pRJ379 encoding TolB with a C-terminal His6-tag but lacking its 23-aa signal sequence is described in ref. 26. All other plasmids were constructed by standard methods (for further details, see Supporting Materials and Methods). E. coli strain JM83 was used as a general host strain for cloning, and BL21 (DE3) was used for protein expression. Cultures were grown in LB broth or on LB agar plates and supplemented with ampicillin at 100 μg·ml−1 or kanamycin at 50 μg·ml−1, when required.

Protein Expression and Purification.

Proteins were overexpressed from BL21 (DE3) and purified by nickel affinity chromatography and gel filtration. Synthetic peptides were obtained from Pepceuticals (Nottingham, United Kingdom) (for further details, see Supporting Materials and Methods).

Labeling and Crosslinking.

N-His6ColE9 (16 μM) in 50 mM Mops/3 M GnHCl (pH 7.0) was incubated for 30 min at room temperature in the presence of 50 μM DTT. MBP, from a stock solution of 10 mM in DMSO, was added to a final concentration of 120 μM, mixed by pipetting, and immediately quenched by the addition of 240 μM DTT. The labeled proteins were dialyzed overnight against 50 mM Mops/100 mM NaCl (pH 7.0). Crosslinking reactions were performed in 0.5-ml Eppendorf tubes. MBP-labeled N-His6ColE9 (1 μM) and TolB (4 μM) were mixed and placed directly on a transilluminator for 10 min, and the products were run on an SDS/10% PAGE gel and visualized by Coomassie staining.

Crystallization and Structure Determination.

A 1:1 TolB–ColE9 Tpep32–47 complex (10 mg·ml−1 in water) was prepared by adding equimolar amounts of both components and used for crystallization trials without further purification. Crystals of the TolB–ColE9 Tpep32–47 complex were obtained by the hanging-drop vapor-diffusion method. One microliter of TolB–Tpep32–47 complex was mixed with the same volume of 24% polyethylene glycol monomethyl ether 5000/80 mM CaCl2/100 mM Hepes (pH 7.5) and equilibrated against a reservoir of 500 μl containing this solution. Crystals were flash-cooled by using paratone as a cryoprotectant, and a data set was collected at 100 K on beamline ID23 at the European Synchrotron Radiation Facility (Grenoble, France). Crystals belonged to the space group P21 and contained four molecules of the complex (TolB/ColE9; A–D/E–H) in the asymmetric unit giving a Matthews coefficient of 2.1 for a 40.7% solvent content. The structure was solved by molecular replacement with the program MOLREP and the structure of TolB (PDB entry 1C5K) as the search probe. For details of the structure determination, see Supporting Materials and Methods. Molecular graphics images were produced by using the UCSF Chimera package (Resource for Biocomputing, Visualization, and Informatics, University of California, San Francisco, CA).

Supplementary Material

Supporting Information

Acknowledgments

We thank Andrew Leech and Berni Strongitharm (York Technology Facility) for expert assistance with many of the biophysical measurements conducted in this study and Nick Housden and Jennifer Potts (University of York) for comments on the manuscript. We acknowledge the European Synchrotron Radiation Facility for provision of synchrotron radiation facilities and thank Didier Nurizzo for assistance in using beamline ID23. This work was supported by the Wellcome Trust and the Biotechnology and Biological Sciences Research Council of the United Kingdom.

Glossary

Abbreviations

IM

inner membrane

ITC

isothermal titration calorimetry

MBP

4-(N-maleimido)benzophenone

OM

outer membrane.

Footnotes

Conflict of interest statement: No conflicts declared.

This paper was submitted directly (Track II) to the PNAS office.

Data deposition: The atomic coordinates and structure factors have been deposited in the Protein Data Bank, www.pdb.org (PDB ID code 2IVZ).

References

  • 1.Ward J. J., Sodhi J. S., McGuffin L. J., Buxton B. F., Jones D. T. J. Mol. Biol. 2004;337:635–645. doi: 10.1016/j.jmb.2004.02.002. [DOI] [PubMed] [Google Scholar]
  • 2.Dunker A. K., Brown C. J., Lawson J. D., Iakoucheva L. M., Obradovic Z. Biochemistry. 2002;41:6573–6582. doi: 10.1021/bi012159+. [DOI] [PubMed] [Google Scholar]
  • 3.Fink A. L. Curr. Opin. Struct. Biol. 2005;15:1–7. doi: 10.1016/j.sbi.2005.01.002. [DOI] [PubMed] [Google Scholar]
  • 4.Demarest S. J., Martinez-Yamout M., Chung J., Chen H., Xu W., Dyson H. J., Evans R. M., Wright P. E. Nature. 2002;415:549–553. doi: 10.1038/415549a. [DOI] [PubMed] [Google Scholar]
  • 5.Gross J. D., Moerke N. J., von der Haar T., Lugovskoy A. A., Sachs A. B., McCarthy J. E. G., Wagner G. Cell. 2003;115:739–750. doi: 10.1016/s0092-8674(03)00975-9. [DOI] [PubMed] [Google Scholar]
  • 6.Dyson H. J., Wright P. E. Nat. Rev. Mol. Cell Biol. 2005;6:197–208. doi: 10.1038/nrm1589. [DOI] [PubMed] [Google Scholar]
  • 7.Gunasekaran K., Tsai C.-J., Kumar S., Zanuy D., Nussinov R. Trends Biochem. Sci. 2003;28:81–85. doi: 10.1016/S0968-0004(03)00003-3. [DOI] [PubMed] [Google Scholar]
  • 8.Shoemaker B. A., Portman J. J., Wolynes P. G. Proc. Natl. Acad. Sci. USA. 2000;97:8868–8873. doi: 10.1073/pnas.160259697. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Brooks I. Crit. Rev. Microbiol. 1999;25:155–172. doi: 10.1080/10408419991299211. [DOI] [PubMed] [Google Scholar]
  • 10.Lazdunski C. J., Bouveret E., Rigal A., Journet L., Lloubès R., Bénédetti H. J. Bacteriol. 1998;180:4993–5002. doi: 10.1128/jb.180.19.4993-5002.1998. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.James R., Penfold C. N., Moore G. R., Kleanthous C. Biochimie. 2002;84:381–389. doi: 10.1016/s0300-9084(02)01450-5. [DOI] [PubMed] [Google Scholar]
  • 12.Kleanthous C., Kühlmann U. C., Pommer A. J., Ferguson N., Radford S. E., Moore G. R., James R., Hemmings A. M. Nat. Struct. Biol. 1999;6:243–252. doi: 10.1038/6683. [DOI] [PubMed] [Google Scholar]
  • 13.Soelaiman S., Jakes K., Wu N., Li C. M., Shoham M. Mol. Cell. 2001;8:1053–1062. doi: 10.1016/s1097-2765(01)00396-3. [DOI] [PubMed] [Google Scholar]
  • 14.Romero P., Obradovic Z., Kissinger C. R., Villfranca J. E., Dunker A. K. Proc. IEEE Int. Conf. Neural Networks; 1997. pp. 90–95. [Google Scholar]
  • 15.Uversky V. N., Gillespie J. R., Fink A. L. Proteins. 2000;41:415–427. doi: 10.1002/1097-0134(20001115)41:3<415::aid-prot130>3.0.co;2-7. [DOI] [PubMed] [Google Scholar]
  • 16.Collins E. S., Whittaker S. B., Tozawa K., MacDonald C., Boetzel R., Penfold C. N., Reilly A., Clayden N. J., Osborne M. J., Hemmings A. M., et al. J. Mol. Biol. 2002;318:787–804. doi: 10.1016/S0022-2836(02)00036-0. [DOI] [PubMed] [Google Scholar]
  • 17.Macdonald C. J., Tozawa K., Collins E. S., Penfold C. N., James R., Kleanthous C., Clayden N. J., Moore G. R. J. Biomol. NMR. 2004;30:81–96. doi: 10.1023/B:JNMR.0000042963.71790.19. [DOI] [PubMed] [Google Scholar]
  • 18.Housden N. G., Loftus S., Moore G. R., James R., Kleanthous C. Proc. Natl. Acad. Sci. USA. 2005;102:13849–13854. doi: 10.1073/pnas.0503567102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Kurisu G., Zakharov S. D., Zhalnina M. V., Bano S., Eroukova V. Y., Rokitskaya T. I., Antonenko Y. N., Wiener M. C., Cramer W. A. Nat. Struct. Biol. 2003;10:948–954. doi: 10.1038/nsb997. [DOI] [PubMed] [Google Scholar]
  • 20.Zakharov S. D., Eroukova V. Y., Rokitskaya T. I., Zhalnina M. V., Sharma O., Loll P. J., Zgurskaya H. I., Antonenko Y. N., Cramer W. A. Biophys. J. 2004;87:3901–3911. doi: 10.1529/biophysj.104.046151. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Bouveret E., Rigal A., Lazdunski C., Bénédetti H. Mol. Microbiol. 1998;27:143–157. doi: 10.1046/j.1365-2958.1998.00667.x. [DOI] [PubMed] [Google Scholar]
  • 22.Garinot-Schneider C., Penfold C. N., Moore G. R., Kleanthous C., James R. Microbiology. 1997;143:2931–2938. doi: 10.1099/00221287-143-9-2931. [DOI] [PubMed] [Google Scholar]
  • 23.Hands S. L., Holland L. E., Vankemmelbeke M., Fraser L., Macdonald C. J., Moore G. R., James R., Penfold C. N. J. Bacteriol. 2005;187:6733–6741. doi: 10.1128/JB.187.19.6733-6741.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Dorman G., Prestwich G. D. Biochemistry. 1994;33:5661–5673. doi: 10.1021/bi00185a001. [DOI] [PubMed] [Google Scholar]
  • 25.Abergel C., Bouveret E., Claverie J. M., Brown K., Rigal A., Lazdunski C., Bénédetti H. Struct. Fold. Des. 1999;7:1291–1300. doi: 10.1016/s0969-2126(00)80062-3. [DOI] [PubMed] [Google Scholar]
  • 26.Carr S., Penfold C. N., Bamford V., James R., Hemmings A. M. Struct. Fold. Des. 2000;8:57–66. doi: 10.1016/s0969-2126(00)00079-4. [DOI] [PubMed] [Google Scholar]
  • 27.Holm L., Sander C. Trends Biochem. Sci. 1995;20:478–480. doi: 10.1016/s0968-0004(00)89105-7. [DOI] [PubMed] [Google Scholar]
  • 28.Jawad Z., Paoli M. Structure (London) 2002;10:447–454. doi: 10.1016/s0969-2126(02)00750-5. [DOI] [PubMed] [Google Scholar]
  • 29.Bouveret E., Derouiche R., Rigal A., Lloubès R., Lazdunski C., Bénédetti H. J. Biol. Chem. 1995;270:11071–11077. doi: 10.1074/jbc.270.19.11071. [DOI] [PubMed] [Google Scholar]
  • 30.Ray M.-C., Germon P., Vianney A., Portalier R., Lazzaroni J. C. J. Bacteriol. 2000;182:821–824. doi: 10.1128/jb.182.3.821-824.2000. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Neri D., Billeter M., Wutrich K. Science. 1992;257:1559–1563. doi: 10.1126/science.1523410. [DOI] [PubMed] [Google Scholar]
  • 32.Klein-Seetharaman J., Oikawa M., Grimshaw S. B., Wirmer J., Duchardt E., Ueda T., Imoto T., smith L. J., Dobson C. M., Schwalbe H. Science. 2002;295:1719–1722. doi: 10.1126/science.1067680. [DOI] [PubMed] [Google Scholar]
  • 33.Tozawa K., Macdonald C. J., Penfold C. N., James R., Kleanthous C., Clayden N. J., Moore G. R. Biochemistry. 2005;44:11496–11507. doi: 10.1021/bi0503596. [DOI] [PubMed] [Google Scholar]
  • 34.Rigal A., Bouveret E., Lloubès R., Lazdunski C., Bénédetti H. J. Bacteriol. 1997;179:7274–7279. doi: 10.1128/jb.179.23.7274-7279.1997. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Gokce I., Raggett E. M., Hong Q., Virden R., Cooper A., Lakey J. H. J. Mol. Biol. 2000;304:621–632. doi: 10.1006/jmbi.2000.4232. [DOI] [PubMed] [Google Scholar]
  • 36.Anderluh G., Gokce I., Lakey J. H. J. Biol. Chem. 2004;279:22002–22009. doi: 10.1074/jbc.M313603200. [DOI] [PubMed] [Google Scholar]
  • 37.Ferguson A. D., Deisenhofer J. Biochim. Biophys. Acta. 2002;1565:318–332. doi: 10.1016/s0005-2736(02)00578-3. [DOI] [PubMed] [Google Scholar]
  • 38.Peacock R. S., Weljie A. M., Howard S. P., Price F. D., Vogel H. J. J. Mol. Biol. 2005;345:1185–1197. doi: 10.1016/j.jmb.2004.11.026. [DOI] [PubMed] [Google Scholar]
  • 39.Shultis D. D., Purdy M. D., Banchs C. N., Wiener M. C. Science. 2006;312:1396–1399. doi: 10.1126/science.1127694. [DOI] [PubMed] [Google Scholar]
  • 40.Clavel T., Germon P., Vianney A., Portalier R., Lazzaroni J. C. Mol. Microbiol. 1998;29:359–367. doi: 10.1046/j.1365-2958.1998.00945.x. [DOI] [PubMed] [Google Scholar]
  • 41.Bernadac A., Gavioli M., Lazzaroni J. C., Raina S., Lloubès R. J. Bacteriol. 1998;180:4872–4878. doi: 10.1128/jb.180.18.4872-4878.1998. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Webster R. E. Mol. Microbiol. 1991;5:1005–1011. doi: 10.1111/j.1365-2958.1991.tb01873.x. [DOI] [PubMed] [Google Scholar]
  • 43.Bouveret E., Rigal A., Lazdunski C., Bénédetti H. Mol. Microbiol. 1997;23:909–920. doi: 10.1046/j.1365-2958.1997.2751640.x. [DOI] [PubMed] [Google Scholar]
  • 44.Bouveret E., Journet L., Walburger A., Cascales E., Bénédetti H., Lloubès R. Biochimie. 2002;84:413–421. doi: 10.1016/s0300-9084(02)01423-2. [DOI] [PubMed] [Google Scholar]
  • 45.Lazzaroni J. C., Dubuisson J. F., Vianney A. Biochimie. 2002;84:391–397. doi: 10.1016/s0300-9084(02)01419-0. [DOI] [PubMed] [Google Scholar]
  • 46.Walburger A., Lazdunski C., Corda Y. Mol. Microbiol. 2002;44:695–708. doi: 10.1046/j.1365-2958.2002.02895.x. [DOI] [PubMed] [Google Scholar]
  • 47.Jones H. E., Holland I. B., Campbell A. K. Cell Calcium. 2002;32:182–192. doi: 10.1016/s0143416002001537. [DOI] [PubMed] [Google Scholar]
  • 48.Nikaido H. Mol. Biol. Rev. 2003;67:593–656. doi: 10.1128/MMBR.67.4.593-656.2003. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supporting Information

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences

RESOURCES