The Stable Interaction Between Signal Peptidase LepB of Escherichia coli and Nuclease Bacteriocins Promotes Toxin Entry into the Cytoplasm*

LepB is a key membrane component of the cellular secretion machinery, which releases secreted proteins into the periplasm by cleaving the inner membrane-bound leader. We showed that LepB is also an essential component of the machinery hijacked by the tRNase colicin D for its import. Here we demonstrate that this non-catalytic activity of LepB is to promote the association of the central domain of colicin D with the inner membrane before the FtsH-dependent proteolytic processing and translocation of the toxic tRNase domain into the cytoplasm. The novel structural role of LepB results in a stable interaction with colicin D, with a stoichiometry of 1:1 and a nanomolar Kd determined in vitro. LepB provides a chaperone-like function for the penetration of several nuclease-type bacteriocins into target cells. The colicin-LepB interaction is shown to require only a short peptide sequence within the central domain of these bacteriocins and to involve residues present in the short C-terminal Box E of LepB. Genomic screening identified the conserved LepB binding motif in colicin-like ORFs from 13 additional bacterial species. These findings establish a new paradigm for the functional adaptability of an essential inner-membrane enzyme.

The type I signal peptidase LepB (36 kDa) is an essential inner membrane enzyme of Escherichia coli and an integral part of the protein export machinery. LepB is a specialized serine protease that uses a Ser-90/Lys-145 catalytic dyad to perform the highly specialized task of removing the short N-terminal leader signal peptide of 15-30 amino acids from the N terminus of many secreted and membrane proteins (1,2). LepB spans the inner membrane twice with a large C-terminal domain protruding into the periplasmic space (3).
Colicins are antibacterial toxins produced and released under stress conditions that selectively kill non-colicin producing strains. Colicin D is a toxic tRNase known to kill target cells by cleaving the anticodon loop of all four isoaccepting tRNA Arg (4). A specific spontaneous mutation N274K of LepB was shown to result in the loss of sensitivity of E. coli target cells to colicin D (5). Nuclease colicins (DNase or RNase) must be translocated across both the outer and inner membranes to reach their targets in the cytoplasm (6,7). In this scenario LepB is specifically required for a late import step of colicin D (i.e. after the translocation step across the outer membrane). Moreover, the catalytic activity of LepB is not required for colicin D import and toxicity (8,9).
Colicin producers are protected against both endogenous and exogenous toxin molecules by the constitutive expression of an immunity (Imm) protein, which forms a neutral complex with nuclease colicins (6,10,11). Most colicins (e.g. colicins E) have structurally identifiable N-terminal, central, and C-terminal domains. During early steps of import, the first two domains are required for receptor binding and translocation across the outer membrane, respectively, whereas the C-terminal domain is needed for killing activity (12). The colicin D protein has an unusual tripartite organization. The N-terminal domain is required for both the binding of the colicin to the high affinity, iron-siderophore receptor, FepA, and for its subsequent translocation across the outer membrane (13,14). The passage of colicin D through the outer membrane is only partially understood, but it is dependent on the proton motive force of the cytoplasmic membrane, transduced by the TonB/ExbB-D import pathway (7,15). The 277-residue central domain (CD) 3 of colicin D is essential for cell killing. It plays dual roles in TonB-mediated import and formation of the colicin D immunity complex (16).
We demonstrated that, in the case of nuclease-type colicins, an endoproteolytic processing step is essential to transfer the cytotoxic C-terminal domain across the inner membrane into the cytoplasm (9,17). The inner-membrane protease FtsH (18) had previously been shown to be necessary for the toxicity of nuclease colicins (19), and we showed this requirement was at the level of colicin translocation across the inner membrane and that it is presumably responsible for processing of RNase or DNase colicins (9,20). This hijacked function of FtsH during colicin import is coherent with its usual biological role of degrading misassembled or damaged membrane proteins (21,22). In the case of colicin D, the productive FtsH-dependent processing may be mediated by LepB. This unexpected func-tion of LepB was suggested to involve in vitro an interaction with the central domain of colicin D (9).
In the present work we demonstrate in vitro that LepB forms a specific and highly stable complex with colicin D, and this interaction is essential for its import. We show that processing and translocation steps are dependent upon the colicin interaction with LepB. We report also that other RNase cytotoxins from two different species of Klebsiella need the novel noncatalytic activity of LepB to kill target cells. Genomic comparisons suggest that this LepB role is presumably required in the case of 10 other bacteriocins.

Bacterial Strains and Plasmids
E. coli K12 DH5␣ and XL1-blue were used as wild-type host strains for cloning and mutagenesis, respectively. All other strains and plasmids used or constructed are listed in Table 1.

Genetic Constructs and Mutagenesis
Construction of N-or C-terminally truncated colicin D derivatives-The 17 truncated colicin D derivatives (pColD -CD, -A, -B, -Del; Table 1) were created by PCR amplifications employing plasmid pColDI or pColDI(H611A) (carrying an inactivated catalytic domain) as template. The upstream oligonucleotide contained an NdeI restriction site including the ATG start codon in-frame with nucleotides of the cda (colicin D) coding sequence. The downstream oligonucleotide is complementary to cda and ended by an in-frame stop codon followed by a BamHI restriction site. A His 6 tag was added to the 5Ј-or 3Ј-primer used in each amplification. PCR products after digestion were cloned into plasmid pET11c (Table 1) and transformed into DH5␣ strain.

Colicins D Constructs Mutated in the Central Domain, LepB
Constructs Mutated in the Conserved Box E-Using as template the plasmids pColDI, point mutations within the N-terminal part of the colicin D central domain (between positions 413-454) were generated by QuikChange site-directed mutagenesis method (Stratagene) to give the series of pColDI-mut (Table 1). Similarly, LepB mutants within the C-terminal conserved Box E (Gly-272-Arg-282) were generated by using as template the plasmid pLEPB1-wt (pLEPB1-mut; Table 1). Two oligonucleotide primers containing the desired mutation were used, each complementary to opposite strands of the plasmid, to generate the mutant derivatives. The mutated DNAs were subsequently transformed into XL1-Blue super-competent cells, prepared in the presence of 10 mM RbCl.
Construction of Chimeric Colicin/Klebicin Variants-Chimeric colicin D(A317)/klebicin D(E321) variant was created by the three-step PCR amplifications method, modified from Mora et al. (16). Employing plasmid pColD1 (Table 1) as template, the first PCR used an upstream 113-mer oligonucleotide (containing a modified tac promoter and the cda ribosomal binding site, gctgaaatgagctgttgacaattaatcatcggctcgtataatgtgtggaattgtgagcggataacaatttcacacaaagaggtgttttt, followed by 30 nucleotides of the cda (colicin D) coding sequence starting from the fMet codon and a downstream oligonucleotide, complementary to cda and starting at A317, located upstream of the central domain. The second PCR amplification used an upstream oligonucleotide that is in part complementary to the downstream oligonucleotide used in the first PCR and that is in-frame with the codon E321 of kda (located at the start of the central domain of klebicin D) together with a downstream oligonucleotide, which is complementary to the stop codon of kdi (the immunity protein of klebicin D), using the plasmid pKlebD as template (Table 1), carrying the klebicin D operon. A denaturation-renaturation step of the mixed first two PCR products led to an annealing between complementary strands of the common amplified central part. This annealed DNA mix was employed as template for the third PCR amplification using the upstream oligo (first PCR) and the downstream oligo (second PCR), and this led to the amplification of a linear double-stranded chimeric colicin D/klebicin D DNA. The third PCR product was used directly as a template for in vitro protein synthesis. The same three-step PCR amplification technique was used to create hybrid colicin D(A317)/klebicin C(E223). Here the second PCR amplification used an upstream oligonucleotide that is in part complementary to the downstream oligonucleotide used in the first PCR (as described above), and it is in-frame with the codon E223 of kca (located upstream of the central domain of klebicin C) together with a downstream oligonucleotide complementary to the stop codon of kci (the immunity protein of klebicin C) by employing the plasmid pM-C04-KlebC2 (Table 1) as template. Wild-type and hybrid constructs were expressed in vitro in a coupled transcriptiontranslation, Zubay-S30 system from E. coli, optimized for linear DNA templates (Promega) at 37°C for 60 min.

Wild-type and Recombinant Protein Purification
Full-size wild-type colicin D in complex with the immunity protein carrying a C-terminal His 6 tag was produced as previously described (9) from BL21(DE3) cells carrying the plasmid pColDI. The complex was denaturated in 9 M urea, and then the His-tagged ImmD was retained by two passages over a Ni-NTA-agarose column (Protino, Macherey-Nagel), whereas the wild-type colicin D freed of ImmD present in the flowthrough was concentrated after dialysis by ultrafiltration (Amicon Ultra-4 10K NMWL; Millipore).
The 17 truncated, His-tagged colicin D derivatives (Table 1), including the whole or part of the central domain, and mutated colicin D-Imm(His 6 ) (pColDI-mut; Table 1) complexes were expressed and eluted by Ni-NTA affinity chromatography at 200 mM imidazole. Truncated colicin Derivatives that were insoluble (for instance the CD) were purified from inclusion bodies. Crude cell extracts were centrifuged, and the pellet was resuspended in 20 mM Tris, pH 7.5, 0.2 M NaCl containing 200 g of DNase, antiprotease mixture tablet, and lysozyme 1 mg/ml, incubated 30 min at 4°C, and then sonicated (4 ϫ 1 min). After centrifugation, the pellet was resuspended for 30 min in 50 mM Tris, pH 7.5, 0.2 M NaCl, and 9 M urea at 20°C. After centrifugation, the supernatant was loaded onto Ni-NTA column, eluted as described above, dialyzed against 20 mM Tris, pH 7.5 (renaturation), and concentrated by ultrafiltration.
His-tagged LepB, wild type or mutated, was produced from BL21(DE3) cells containing plasmid pLEPB1-wt, pLEPB(N274K), or pLEPB1-mut, respectively (Table 1). At A 600 ϭ 0.5 the expression of the LepB was induced by isopropyl 1-thio-␤-D-galactopyranoside (1 mM), and incubation was continued for 3 h at 37°C. LepB proteins were purified by Ni-NTA affinity chromatography (adapted from Paetzel et al. (23)). After centrifugation the cells were resuspended, incubated for 30 min at 4°C in loading buffer (25 mM Tris, pH 7.6, 0.1 M NaCl, 1% Triton X-100, 10 mM imidazole containing 200 g of DNase and antiprotease mixture tablet), and then broken by sonication. The supernatant was loaded onto a 1-ml Ni-NTA-agarose column and washed with 25 mM Tris, pH 7.6, 0.1 M NaCl, and 0.25% reduced Triton X-100 (a non-ionic and non-UV absorbing surfactant). The His-tagged solubilized LepB was eluted with 200 mM imidazole and then dialyzed against the same buffer. The D273A, R275A, and S281A mutant LepB could not be analyzed due to either their poor expression or the difficulty in extracting them from the detergent-solubilized membrane fraction.

Size Exclusion Chromatography (SEC) of in Vitro Mixed LepB and Colicin D
Purified, full-size (FS), wild-type colicin D, and LepB(His 6 ) alone were analyzed individually by SEC. In addition, purified FS colicin D and a slight excess of His-tagged LepB were mixed in 25 mM Tris, pH 7.5, 0.1 M NaCl, and 0.25% reduced Triton X-100, incubated for 1 h at 25 or 37°C, and then analyzed by SEC. Purified proteins or protein mixes were run on a Superdex 200GL 10/300 column (GE Healthcare) monitored by an AKTA purifier FPLC system (GE Healthcare). 500-l fractions were collected for detection of the LepB-colicin D complex and comparison of the SEC profile of the mixed proteins with that of colicin D or LepB analyzed alone. Proteins in the elution fractions were separated on 15% SDS-PAGE.

Real-time Surface Plasmon Resonance Assays
The SPR biosensor real-time experiments were performed on a Biacore 2000 instrument (GE Healthcare) equilibrated at 25 or 37°C with running buffer (25 mM Tris, pH 7.5, 0.1 M NaCl, 0.3 mM EDTA, 0.05% reduced Triton X-100). The SEC purified, His-tagged LepB protein (encoded by plasmid pLEPB1-wt or pLEPB(N274K)) was captured on Ni-NTA sensor chip (GE Healthcare) surface at an immobilization level of 150 resonance units (pg.mm Ϫ2 ). Colicin D was then injected at different concentrations with a flow rate of 20 l min Ϫ1 for 750 s. The dissociation of the colicin D-LepB complexes was monitored for a 10-min flow of the running buffer without colicin D. The association and dissociation profiles were double-referenced, i.e. both the nonspecific background signals from a streptavidine-His 6 reference surface and those from blank experiments (using running buffer without protein) were subtracted. The experimental data of six independent injections per concentration were overlaid and analyzed using the BIAevaluation 4.1 software (GE Healthcare) considering a 1:1 Langmuir binding model. The experimental SPR signals were normalized against the fitted maximum binding signal (Rmax), and the figures thus show the proportion of LepB molecules that are complexed on the y axis ("LepB site occupancy %"). Between binding assays, the surface was regenerated by successive injections of EDTA 0.35 M (2 min), NaOH 50 mM (1 min) and SDS 0.1% (1 min) followed by washing with the running buffer.
To test the interaction of the N-terminally His-tagged CD of colicin D with the His-tagged LepB, the latter was covalently immobilized in an oriented fashion onto a nitrilotriacetic acidderivatized sensorchip according to Jomain et al. (24) with different surface densities. Immobilized recombinant VEGF (at resonance units ϭ 3000) receptor was used as a non-relevant control protein for the interaction. Colicin D CD was injected with a flow of 20 l min Ϫ1 for 240 s and followed then by two regeneration injections of Gly-HCl 10 mM, pH 2.5.

Detection of LepB Interaction with Colicin D Derivatives by Far Western Blotting
Purified truncated colicin D derivatives (2-5 g) were separated on a 15% SDS-PAGE, transferred to nitrocellulose membrane, and then stained by Amido Black. A denaturation-renaturation step according to Wu et al. (25) was performed to refold the proteins. After incubation for 1 h at 37°C with purified wild-type LepB (1 g/ml) and successive washing steps, the binding of LepB to colicins was analyzed by an anti-LepB antiserum and ECL. Data were collected in triplicate, and the LepBbinding rates averaged.

Killing Activity Test of Bacteriocins
The cytotoxicity of wild-type purified colicin D or colicin D molecules carrying point mutations in the central domain as well as hybrid colicin-klebicins synthesized in vitro (Zubay-S30) was analyzed in vivo by growth inhibition (halo) assay. Undiluted colicins (and 10-fold serial dilutions (5 l) in the case of purified colicins) were spotted onto a lawn of sensitive AD202 or mutant A38 (carrying the LepB(N274K) mutation) cells on LB plates and then incubated overnight. To increase the binding of in vitro synthesized hybrid colicin Derivatives to the outer-membrane receptor FepA, the killing activity test was performed in the presence of the iron chelator 2,2Ј-dipyridyl (200 M). Wild-type klebicin D was found to be non-toxic onto E. coli strains.

In Vivo Detection of Colicin D Processing
50-ml cultures of wild-type or mutant E. coli strains were grown in LB to A 600 ϭ 0.4 at 37°C in the presence of 0.2 mM 2,2Ј-dipyridyl, which increases the expression of the FepA receptor (26), then 1 mg of wild-type or mutant colicin D-ImmD complex was added to the cultures for 45 min. The cells were harvested, washed three times, resuspended in phosphate buffer, pH 7.4 (1ϫ PBS, Euromedex), and treated with proteinase K at 100 g/ml for 1 h at 37°C to hydrolyze residual colicin molecules bound to the cell surface. After four additional washings the colicin-and proteinase K-treated cells were resuspended in 2 ml of PBS buffer with a protease inhibitor complex (Complete EDTA-free; Roche Applied Science) and then disrupted by sonication. Proteins of the S100 cytoplasmic fraction were separated on 15% SDS-PAGE. The processed colicin D form (PF) was detected by Western blot analysis using an antiserum to the colicin D (9).

In Silico Identification of Conserved Amino Acid Motifs at and Near to the LepB Binding Sites (LBS) Located in the Central Domain of E. coli Colicin D
The plasmid-encoded colicin D (pColD CA23) protein originating from E. coli, the colicin D central domain (Glu-314 -Met-590), and a shorter peptide segment centered on the LBS were used independently as bait to detect homologous ORFs. The 3241 completely sequenced bacterial chromosomes provided by the NCBI repository were analyzed locally with a C# script using TBLASTN on the SYNTTAX genomic database (27

LepB and Colicin D Form a Stable Complex in Vitro-
We became interested in the hijacked function of LepB in colicin D import because of one intriguing possibility that the colicin-LepB interaction replaces or completes the electrostatic associations previously reported between several E-type, nuclease colicins, and the inner membrane (19). Although such a chargedependent interaction has been proposed as essential to initiate the late import steps, this was not found to be relevant for colicin D (9, 28).
As a first step in this study, we sought to ascertain the existence of a complex formed by LepB and colicin D. We attempted to reconstitute in vitro the complex from separately purified wild-type LepB(His 6 ) and FS colicin D. Both partners were purified, mixed together, and then incubated at 25 or 37°C for 1 h. The presence of an in vitro assembled complex was assayed by SEC analysis (Fig. 1A). Free FS colicin D ran as a single peak, with an elution volume of 14.2 ml, close to that for free LepB (14 ml). The apparent molecular mass of ϳ145 kDa indicated that in solution colicin D (75 kDa) has a dimeric structure, like RNase colicin E3 (29). The membrane protein LepB (36 kDa) in solution is bound to the micelle formed by the detergent Triton X-100 (ϳ90 kDa; Ref. 30). The apparent molecular weight of LepB (ϳ145 kDa) presumably indicates a monomeric structure. At both temperatures we obtained the same dichotomic elution profiles ( Fig. 1A) with the mixed proteins (i.e. peaks 1 and 2, with elution volumes of 11.9 ml and 14.2 ml, respectively). The molecular weight of the LepB-ColD heterodimer, including Triton binding in solution, could be expected to be ϳ220 kDa (145 ϩ 75 kDa). However, our SEC elution data were consistent with the presence of a LepB-ColD complex in peak 1, corresponding to an apparent molecular mass of ϳ430 kDa, implying that in vitro the complex may behave as a dimer of heterodimers. More complex was formed at 37°C, the area under the peak 1 was ϳ1.5-2-fold larger at 37°C than at 25°C, whereas peak 2, corresponding to isolated LepB and FS colicin D, was diminished by about a factor of 2. The analysis of SEC fractions of peak 1 by SDS-PAGE indicated that both partners assembled into a stable complex and supported the presence of LepB and colicin D in a 1:1 stoichiometry both at 25°C and at 37°C (Fig. 1B), as suggested by the SEC analysis.
We confirmed the existence of the LepB-ColD complex after the co-expression of His-tagged LepB and C-terminally truncated, non-toxic colicin D (Met-1-Met-590) that included both the N-terminal-and central domains but not the toxic tRNase domain. Colicin D (Met-1-Met-590) was co-purified with solubilized LepB(His 6 ) by Ni-NTA affinity chromatography, and the complex was found to be stable by SEC (Fig. 1,  C and D).
In Vitro Binding Affinity of Colicin D to LepB Peptidase-The formation of a stable complex between the wild-type partners shown above by SEC (Fig. 1) prompted us to determine by surface plasmon resonance (SPR) the kinetic constants character-izing the in vitro binding of FS colicin D to LepB. Representative real-time association-dissociation profiles corresponding to a large range of colicin D concentrations injected over wild-type LepB(His 6 ), immobilized by nickel chelation on the sensor chip surface, showed a stable and specific interaction between the partners at both 25 and 37°C (Fig. 2, A and B). The SPR response corresponding to the LepB site occupancy by colicin D is proportionally increased, with higher colicin D concentrations. Remarkably, the k off (2.6 Ϯ 0.4 ϫ 10 Ϫ4 s Ϫ1 ) was found to be identical at both temperatures, which indicated that the complex once formed is stable (t1 ⁄ 2 ϭ 45 min) and that its dissociation was not dependent on the temperature. In contrast, the  Table 1). The sonication of cells was followed by 1 h of incubation at 25°C and then solubilized LepB was co-purified by Ni-NTA affinity chromatography with colicin D (Met-1-Met-590) as shown by SDS-PAGE. Two elution fractions together were subsequently analyzed by SEC and gave two peaks (black curve). The LepB-ColD (Met-1-Met-590) complex, eluted as peak A, had an apparent molecular mass of ϳ430 kDa. Peak B was close to the single elution peak of the free monomeric LepB chromatogram (blue curve). The monomeric LepB with the micelle formed by the detergent Triton X-100 (90 kDa) had an apparent molecular mass of ϳ145 kDa, as also shown in A. D, the SEC elution fractions were analyzed by SDS-PAGE. The stoichiometry of the LepB-ColD complex appeared to support a 1:1 molar ratio of the two partners (boxed fractions; see also B). The free LepB was enriched in peak B (indicated by a horizontal arrow). DECEMBER  ; thus this interaction should presumably be sufficient in vivo to localize colicin D close to the inner-membrane surface once it is present in the periplasmic space. A higher binding affinity at 37°C than 25°C is consistent with the more extensive complex formation detected by SEC at 37°C (Fig. 1A).

Structural Role of LepB for Bacteriocin Import
Under the same conditions, the injection of colicin D at different concentrations on immobilized, His-tagged LepB (N274K) protein (from the A38 mutant (Table 1), which is resistant to colicin D killing), showed no detectable interaction at either temperatures (Fig. 2C). It is striking that the absence of an interaction is not the consequence of a global structural unfolding in LepB due to the N274K mutation, as the catalytic function was shown to be unaffected in A38 strain (5). Thus, the absence of interaction may be attributable either directly to the replacement of the Asn-274 residue by a Lys or indirectly to local conformational changes induced by the point mutation. Additional SPR investigations confirmed previous Far Western blot analyses showing that the N-terminally His-tagged central domain of colicin D (CD, purified from pColD-CD; Table 1) specifically recognized the immobilized wild-type LepB, covalently coupled to the sensor chip. No interaction was detected with a non-relevant VEGF-receptor protein (Fig. 2D). Experiments conducted in parallel showed that the rate of interaction increased proportionally with the concentration of the CD of colicin D injected on the covalently immobilized LepB (Fig. 2E), as also observed with the full-size colicin D ( Fig. 2A). Further analysis of the binding affinity of CD to LepB could not be investigated by SPR, as the precise oligomerization state of the purified, isolated CD could not be determined unambiguously.  Table 1) injected at 25°C (380 g/ml) over different densities (resonance units (RU)ϭ 750 -6100) of covalently immobilized wild-type LepB(His 6 ) or VEGF receptor (resonance units ϭ 3000; used as a non-relevant protein) are shown. E, wild-type LepB(His 6 ) was covalently immobilized on the surface of sensor chip. Serial dilutions of purified colicin D CD were injected for 450 s at 25°C. The corresponding color-coded association profiles, deduced from the LepB-site occupancy by colicin molecules in real time, are shown.
However, the global structural integrity of the purified CD could be confirmed by circular dichroism spectroscopy (Fig. 3A).
LepB Recognizes Specifically a Small Peptide Motif Inside of the Colicin D CD-17 truncated, colicin D-derived constructs were obtained by PCR amplification and subsequent cloning into a pET11 vector, yielding pColD-A, -B, -CD, and pColD-Del(1-14) ( Table 1). The purified colicin derivatives, shown in Fig. 4A, were analyzed by Far Western blotting for their ability to bind to LepB. As expected, only colicin D derivatives carrying the CD were found to be the specific target of LepB binding (Fig.  4C). Neither the presence of the N-nor C-terminal domains affected the binding (Fig. 4, A and C; ColD compared with the derivatives A or B or CD). No interaction was detected with derivatives either starting at or downstream of position 444 (derivatives 2, 3, and 5) or those located before position 410 (derivatives 9 and 13) (Fig. 4C). In contrast, colicin D derivatives 12 or 6, starting at position 410 and ending at position 444, respectively (Fig. 4, A and C), exhibited an interaction with LepB. Accordingly, with all other truncated colicin derivatives straddling the region 410 -444 (constructs 1, 4, 7, and 8), binding with LepB was observed. The importance of the CD peptide motif 410 -444 in LepB recognition is highlighted by the observation that truncated derivatives starting (derivative 10) or ending (derivative 11) at position 422 showed no interaction with LepB. However, the derivative 14 containing residues 410 -437 exhibited a full interaction signal with LepB. This 28 amino acid-long peptide motif is hereafter referred to as the LepB binding site, LBS (Fig. 4A).
To confirm that the recognition of LBS by LepB was responsible for its stable interaction with colicin D, we tested whether extra copies of LBS-containing colicin D sequences could compete for binding LepB with colicin D derivatives. Thus, wildtype purified LepB preincubated with the truncated colicin derivative 6 or 14 (both containing the LBS sequence; Fig. 4A) was shown by Far Western blotting to have a strongly reduced ability (12-20% of free LepB) to bind the same and other truncated colicin derivatives separated on SDS-PAGE (Fig. 4D). In contrast, the intensity of the wild-type interaction signal observed with LepB was unchanged when it was preincubated with the truncated colicin D derivative 5, which does not contain the LBS sequence. Thus, the non-catalytic role of LepB required for import (9) is presumably based upon the site-specific interaction with the incoming colicin D.
The Structural Function of LepB Is Also Required for the Toxicity of Other RNase-type Cytotoxins-Klebicins D and C produced by K. oxytoca and K. pneumoniae, respectively, have a ribonuclease activity and a central domain like colicin D (16, 31) (Fig. 4B). We identified a conserved peptide motif within the LBS sequence present in all three bacteriocins (AG-GSD-VPGR; named LBS consensus motif) (Fig. 4A). This observation suggested that colicin D may not constitute the only substrate that requires a stable structural interaction with LepB. The N-terminal import domains (I) of klebicins C and D are not homologous to colicin D (Fig. 4B). Klebicin D shows no toxicity against wild-type E. coli, suggesting that it uses a Klebsiellaspecific import pathway. Klebicin C is toxic for E. coli, but its translocation across the outer membrane of E. coli does not require the FepA receptor, unlike colicin D, but remains dependent on the energy transducing TonB-system (Refs. 7 and 15 and data not shown). To investigate a possible LepB requirement in klebicin import into E. coli target cells, we sought to dissociate the klebicin-specific translocation step across the outer membrane from the putative LepB-dependent crossing of the cytoplasmic membrane. To this aim, we constructed hybrid colicin D-klebicin D (or -klebicin C) in which the intact N-terminal import domain of klebicin D or C was replaced with that of colicin D (Fig. 4B). Thus, translocation of hybrid derivatives into the periplasm became dependent on FepA and TonB, as for colicin D. As shown by a growth inhibition test (Fig. 5A), the in vitro expressed wild-type and hybrid bacteriocins exhibited a strong toxicity on wild-type E. coli. The same bacteriocins all completely lost their toxicity on the LepB(N274K) mutant strain (Fig. 5A). We also verified that an FtsH-deficient strain is fully insensitive to purified klebicin C and the hybrid colicins, like colicin D (data not shown). The in vivo toxicity test provided evidence that the LepB protein also contributes to the B, the spectra of mutated LepB, measured in the presence of 0.25% reduced (non-UV absorbing) Triton X-100, were found to be comparable in pattern and intensity to wild-type (WT) LepB, indicating that the point mutations did not alter the secondary structure of the proteins. DECEMBER 25, 2015 • VOLUME 290 • NUMBER 52

JOURNAL OF BIOLOGICAL CHEMISTRY 30789
toxicity of these two other distinct nuclease cytotoxins, implying that LepB may have a wider, non-catalytic role in bacteriocin import. To investigate this hypothesis, we tried to identify by TBLASTN any ORFs carrying a central region similar to that of colicin D in entirely sequenced bacterial genomes and plasmids. Our inspection showed that, in addition to colicin D and klebicins C and D, the LBS consensus motif (G-GSD-VPGRD) was present in many ORFs encoded by chromosomal or plasmid-borne cda-like genes, present in 10 other species, all belonging to Gram-Gammaproteobacteria (Fig. 6). A less conserved LBS consensus was identified in the phylogenetically distant P. aeruginosa. The genomic analysis also identified in all colicin D-like ORFs a second highly conserved peptide motif ( 459 V-PVRG 465 in E. coli, 26 amino downstream of the LBS consensus) (Fig. 6). However, this conserved sequence was not required for the interaction of colicin D with LepB, as truncated peptides missing this sequence (i.e. constructs 6 and 14) still interacted with LepB (Fig. 4, A and C).
Function of LepB in Colicin D Import and Cell Killing-The loss of the interaction of mutated LepB(N274K) with colicin D showed above in vitro and the fact that this mutation specifically prevented both the colicin processing and the sensitivity of target cells to colicin D supported the structural role of LepB in colicin D cytotoxicity (9,20). We attempted to analyze whether (i) the recognition of LepB may be specifically dependent on conserved amino acids belonging to the LBS of colicin D and (ii) the direct interaction between colicin D LBS and LepB is an indispensable prerequisite for the FtsH-dependent colicin processing and subsequent cell killing. Point mutations (mainly to alanine) were introduced to replace amino acids in six positions of the colicin D LBS consensus (Fig. 5B). The Ala replacements had no or only a slight effect on colicin D toxicity, as shown by a growth inhibition test (Fig. 5B). For instance, colicin D mutated at positions S426A, P430A, or R432A exhibited a weak ϳ10-fold reduced toxicity. However, a charged aspartate introduced to replace Ala-422 or Gly-423 almost completely abolished the toxicity. As A422V showed no loss in toxicity, the extra negative charge is probably responsible for the loss in toxicity, at least in the case of A422D. A more graduated response was found by analyzing the interaction between the mutated colicin D proteins and wild-type LepB by Far Western blotting (Fig. 5C). Only with A422D or G423D was the interaction with LepB almost completely abolished (Ͻ10%), whereas S426A, P430A, and R432A exhibited an ϳ50% reduction in interaction. Cell killing by nuclease colicins is known to require only a few copies of the nuclease domains to reach the cytoplasm (7). Thus, loss of 50% of the binding observed between these three mutated colicin D and LepB remains compatible with an efficient, only 10-fold decreased cell killing (Fig. 5, B  and C). Finally, we tested the level of processing in vivo of several mutated colicin D and observed a wild-type level of the processed colicin D tRNase form (PF; Ref. 9) in the cytoplasm of target cells (AD202) treated by all of them, except for A422D and G423D (Fig. 5D). No PF of these two mutated colicins was detected in the cytoplasm. The processing defect thus appeared to be the direct consequence of the absence of their productive interaction with LepB at the inner membrane. All these results indicated that the structural role of LepB is a key player in the process of colicin D uptake.
Involvement of the Box E Residues of LepB in the Interaction with Colicin D-Four strictly conserved residues of the LepB Box E (positions 272-282) are critical for maintaining a functional enzyme (32) (Fig. 7A). The analysis of the solvent accessibility using the ESPript program (33) showed that most Box E residues are buried near the catalytic center, whereas only Asp-276, Asn-277, and Ala-279 have intermediate or full accessibility to water molecules. In addition to the N274K mutation of strain A38, we have constructed several mutants at (and in the vicinity of) Asn-274. The interactions of these mutated LepB proteins with colicin D constructs (derivatives A and 6; Fig. 4A) were detected by Far Western blotting. The level of interaction was diminished by a factor of ϳ2 in the case of N274A, whereas the N274K (and N274D, data not shown) mutants showed only a marginal interaction (Ͻ5%) with the central domain (derivative A) or the truncated colicin D derivative 6 (Fig. 7A). The mutation of partially exposed and non-catalytic Asn-277 to an Ala had only a weakly reduced effect (10%) on colicin D binding, whereas N277K showed a strongly reduced binding to the same colicin D derivatives (Fig. 7A). Even though Asn-274 is a strictly conserved residue in prokaryotes, its mutation either to alanine or to Lys had no effect on the enzymatic activity of LepB (5,32). The buried Gly-272 (an essential residue for signal peptidase activity), when mutated to Ala also abolished the interaction with colicin D constructs (Fig. 7A). In contrast, the water-accessible Ala-279 mutated to Gly exhibited only a moderately reduced level of interaction (85% that of wild type). The global conformational integrity of mutated LepB, either at Box E-residues essential for the catalytic activity (Gly-272; Ref. 32) or at residues found to be relevant only for the interaction of LepB with colicin D (Asn-274 and Asn-277; Fig. 3B) was checked by circular dichroism spectroscopy. The study of these mutations on the three-dimensional structure of LepB (34) showed that the replacement of the asparagine at 274 or 277 by the smaller residue alanine does not introduce any significant steric interference (Fig. 7B). The introduction of the bulkier charged lysine residue at either Asn-274 or Asn-277 as well as the N274D change in LepB potentially impairs the local conformation. Thus, the interaction with colicin D may be abolished, as possibly a consequence of electrostatic or steric hindrance affecting the surface-exposed antiparallel ␤ sheet 8 (amino acids 188 -196) or ␤ sheet 9 (amino acids 204 -209) (35) close to the colicin D LBS would be located (Fig. 7B).

Discussion
The E. coli LepB, a type I signal peptidase, is a key membranebound enzyme of the protein secretion process that has been identified in eubacteria, archaea, and eukaryotes (36). LepB cleaves the signal peptide and thereby releases the mature protein into the periplasm (35,37). In addition to this canonical scenario, our current work has demonstrated a second function of LepB, which is of ensuring a stable association of the incoming colicin D with the inner membrane during its import. This step appeared to be necessary for the processing of colicin D by FtsH before the subsequent translocation of the toxic domain across the inner membrane into the cytoplasm (17). FtsH acts preferentially on unfolded proteins rather than unfolding the substrates themselves (21). Consequently, it is reasonable to assume that this novel LepB activity implies a structural destabilization of the colicin molecules so that the unstructured C-terminal tails can protrude into the catalytic center of FtsH.
Association of the positively charged and periplasmically exposed nuclease domains with the anionic phospholipids of the inner membrane has been demonstrated to be essential for the toxicity of several E-type nuclease colicins (19,38,39). Docking onto the inner membrane was proposed to drive the unfolding of the catalytic domain (28), thereby exposing it as a substrate for FtsH. The central domains of colicin D and klebicins C and D as well as the catalytic domains of colicin D and klebicin D do not exhibit an excess of positive charges, facilitating their electrostatically driven interaction with the inner membrane. Thus, in the case of these incoming bacteriocins, the interaction with LepB can be understood as a way of ensuring or at least improving their association with the cytoplasmic membrane.
In vitro experiments demonstrated the strong and site-specific interaction between LepB and the tRNase colicin D with a K d in the nanomolar range. The high affinity binding of colicin D with LepB may subsequently make the catalytic domain of colicin D available to the FtsH protease for its subsequent processing. Attachment of the incoming colicin to membrane components appears to be essential throughout the colicin import process. Both the association with an outer membrane receptor (40) and the interaction of colicin D and other bacteriocins with the inner membrane LepB, as demonstrated here, seem to necessitate a high affinity interaction to initiate the translocation into the periplasm and cytoplasm, respectively.
Colicin D hydrolysis by the outer-membrane protease OmpT in vitro, reported as requiring the simultaneous attachment of  (Zubay-S30 system), wild-type colicin D (ColD), klebicin C (KlebC), hybrid colicin D/klebicin D (ColD-KlebD), and colicin D/klebicin C (ColD-KlebC) by growth inhibition test is shown. Undiluted aliquots were spotted, as indicated by black points, onto a lawn of AD202 wild-type or A38 LepB(N274K) mutant E. coli strain. Dark halos indicate a clear zone of growth inhibition (toxicity), and no clearing reveals the resistance to wild-type or hybrid colicins. B, nine point mutations to alanine, aspartate, or valine introduced into or close to the LepB-binding site (blue-boxed peptide sequence located in the colicin D central domain) are indicated. The killing activity of these purified mutated colicins D was quantified by spotting aliquots of undiluted colicin (numbered 0) and serial 10-fold dilutions up to 10 Ϫ5 (numbered 1-5) directly onto a lawn of AD202 strain and analyzed as in A. Turbid zones indicate marginal toxicity. C, wild-type or mutated FS colicins (ϳ50 ng) were separated by 15% SDS-PAGE and detected by Amido Black staining (upper panel). Their interaction with purified wildtype LepB was detected in vitro by Far Western blotting (lower panel) with anti-LepB antiserum. The efficiency of interaction was quantified and expressed as a percentage of value measured with wild-type colicin D (wt, 100%). D, in vivo detection of the colicin D processed form (ColD PF, 12.4 kDa) from the S100 cytoplasmic fraction of an OmpT-deficient colicin D-sensitive strain treated by purified wild-type or mutated colicins D. Total S100 extracts were separated by 15% SDS-PAGE, and the ColD PF was detected by Western blotting. Mutations provoking a weak or an almost full defect on toxicity on the interaction with LepB and on the processing are indicated in green and red, respectively.
the LepB protein to the substrate (9), supports the structural role of LepB. LepB binding, ϳ200 residues upstream of the OmpT cleavage site, may introduce a local, structural unfolding into colicin D that thus exposes it to hydrolysis by OmpT. Thereby, LepB seems to function similarly to several chaperones or to other adaptor proteins that are reported as necessary for the degradation of various natural substrates by FtsH. For instance, such a chaperone scenario implying FtsH was reported in the case of DnaK/J-mediated degradation of 32 factor at lower temperatures or for HflK/C-modulated protein quality control of the inner membrane (18,41,42).
In addition to Asn-274, two other nearby residues, Asn-277 and Gly-272, also located in the conserved C-terminal Box E of LepB (32), were shown in vitro to be essential for the interaction with colicin D. In the case of the first two residues a dramatic loss was only observed when the asparagine was changed to a bulkier positively charged lysine or by a negatively charged aspartate (at N274D). In contrast, the mutation of Gly-272 to alanine is sufficient to abolish the interaction. Although essential, the contribution of these partly "buried" residues for the interaction appeared to be possibly indirect with regard of the study of their mutations on the solved crystal structure of LepB. The interaction site on colicin D was shown to be located to a 28-amino acid-long LBS motif within the central domain. We also showed that the C-terminal half of the LBS was found to be conserved in the central domains of two other RNases bacteriocins encoded by Klebsiella species, whose toxicity is also strictly dependent on LepB and FtsH. The LBS consensus motif was identified by genomic screening in nuclease-type bacteriocins that carry various import and/or catalytic domains in 13 different bacterial species, suggesting that LepB is a common participant in their import. This observation is supported by the fact that the klebicin C operon was previously reported to have been inserted into a lambdoid prophage, which would explain klebicin diversification as a consequence of bacteriophage-mediated horizontal transfer (31). Thus, the requirement of LepB for colicin import seems to be linked to the spread of a colicin D-type central domain during diversification of bacteriocins by exchanges and mixing of functional domains either by recombination (31,43,44) or mediated by bacteriophages (31). It should be noted that when we used only the single LBS sequence (Fig. 4A) for the in silico analysis, the presence of an additional (and longer in comparison with E. coli) consensus sequence (G-GSD-VP-RN-EAM; Fig. 6) was identified near the adenylation domain of the bacitracin synthetase, a 9088-amino acid-long protein complex encoded by the Gramϩ bacteria, P. mucilaginosus. The hypothetical role of LepB in the functioning of this key enzyme of the non-ribosomal synthesis of the antimicrobial polypeptide bacitracin (45) still remains elusive.