Characterization of the Ruler Protein Interaction Interface on the Substrate Specificity Switch Protein in the Yersinia Type III Secretion System

Abstract

Many pathogenic Gram-negative bacteria use the type III secretion system (T3SS) to deliver effector proteins into eukaryotic host cells. In Yersinia, the switch to secretion of effector proteins is induced first after intimate contact between the bacterium and its eukaryotic target cell has been established, and the T3SS proteins YscP and YscU play a central role in this process. Here we identify the molecular details of the YscP binding site on YscU by means of nuclear magnetic resonance (NMR) spectroscopy. The binding interface is centered on the C-terminal domain of YscU. Disrupting the YscU-YscP interaction by introducing point mutations at the interaction interface significantly reduced the secretion of effector proteins and HeLa cell cytotoxicity. Interestingly, the binding of YscP to the slowly self-cleaving YscU variant P264A conferred significant protection against autoproteolysis. The YscP-mediated inhibition of YscU autoproteolysis suggests that the cleavage event may act as a timing switch in the regulation of early versus late T3SS substrates. We also show that YscU_C binds to the inner rod protein YscI with a dissociation constant (K_d) of 3.8 μm and with 1:1 stoichiometry. The significant similarity among different members of the YscU, YscP, and YscI families suggests that the protein-protein interactions discussed in this study are also relevant for other T3SS-containing Gram-negative bacteria.

Introduction

Yersinia spp. share a common virulence plasmid that encodes a type III secretion system (T3SS)³ supporting the secretion of the Yersinia outer protein (Yop) effectors from the bacterium into eukaryotic host cells (1). After translocation into the host cells, Yop effectors counteract immune defense mechanisms such as phagocytosis or apoptosis and thereby promote the survival and propagation of the extracellular bacteria (2). Many pathogenic and symbiotic Gram-negative bacteria also utilize the T3SS to deliver proteins during the interactions with their hosts (3, 4). The T3SS (also referred to in the literature as the non-flagellar T3SS) evolved from the flagellum, and both systems present a similar architecture (5, 6) featuring a basal body assembly consisting of a multiringed complex that spans the bacterial envelope (5). However, the basal body of the flagellum is connected to an extracellular hook that forms a link with the flagellar filament. In contrast, the T3SS basal body is attached to a needle (in animal pathogens) or a pilus (in plant pathogens) that protrudes from the bacterial surface (5, 7, 8). The basal body and the needle/pilus together form the so-called needle complex that is the hallmark of the T3SS.

It is accepted that the needle complex is assembled in a stepwise manner, and in Yersinia spp., it has been shown that the inner membrane platform (formed by YscR, -S, -T, -U, and -V) is assembled independently of the outer membrane ring-forming proteins (YscC, -J, and -D) (9). After association of the inner and outer rings (mediated by YscJ), the secretion of the so-called early substrates such as the needle subunit YscF begins, which results in elongation of the needle complex. Activation of the T3SS impairs the secretion of the early substrates but triggers the secretion of the Yop effectors (late substrates) (10, 11). This modification of the secretion pattern was first described by Macnab and co-workers (12–14) in the flagellum and is called the substrate specificity switch. It has been shown that YscP plays a critical role in regulating the needle length (15–18). An YscP-null mutant or insertions within the YscP sequence triggered the formation of long needles. in contrast, shorter needles were produced when deletions were introduced in the YscP sequence (19). Furthermore, a minimal needle length is required to support Yop secretion (17). Together, these results suggest that the needle length is tightly regulated and that YscP functions as a molecular ruler (16). Although the ruler model has attracted significant interest, alternative models of needle length control such as the measuring cup model (15) and the molecular clock model (20) have also been proposed, and there is currently no consensus regarding the true nature of the needle length control mechanism. A systematic deletion analysis of YscP in Yersinia enterocolitica led to the identification of domains with specific functions (11, 19), including two distinct secretion signals (residues 1–35 and residues 97–137) and a substrate specificity switch domain located between residues 385 and 500 (11). Homologues of YscP with similar functions are found in the flagella and T3SSs of diverse bacterial species (21, 22).

YscU in Yersinia spp. and FlhB in the flagellum have also been linked to the substrate specificity switch (23). YscU is anchored in the inner membrane via four transmembrane helices and possesses a large C-terminal cytoplasmic domain named YscU_C (Fig. 1) (24, 25). YscU_C is characterized by a conserved NPTH motif that undergoes an autoproteolytic process (between Asp²⁶³ and Pro²⁶⁴) to generate a C-terminal peptide named YscU_CC that form a stable complex with the N-terminal part of the cytoplasmic domain named YscU_CN. Previous results obtained in our laboratories suggest that the positively charged residues within the linker between YscU_C and the membrane domain interact with the membrane lipids to associate YscU_C with the inner membrane (26). Accordingly, it has been shown that mutations at Asp²⁶³ or Pro²⁶⁴ that block YscU autoproteolysis interfere with both needle formation and Yop secretion (10, 27). Also, we have shown that the dissociation of YscU_C is required for Yop effector secretion (28). It was subsequently found that YscU_C contains a C-terminal secretion signal domain (29). The deletion of the last 15 residues of YscU triggered an increase in YscF secretion without affecting Yop secretion (29). Thus, like YscP, YscU is involved in the substrate specificity switch. Similar results were reported for FliK and FlhB, which are the flagellar homologues of YscP and YscU, respectively (30, 31). Interestingly, the phenotype observed in a yscP-null mutant was partially suppressed by single amino acid substitutions within YscU (YscU^A268F, YscU^Y287G, and YscU^V292T) (10). It was subsequently shown that these suppressor mutants are less stable and exhibit faster dissociation kinetics than the wild-type protein, which enabled Yop secretion to occur in the absence of YscP (10, 28). Direct interactions have been observed between FlhB and FliK (Salmonella enterica serovar Typhimurium) and between the homologous proteins in Pseudomonas aeruginosa, PscU and PscP (residues 31–36), supporting the proposed functional link between YscU and YscP homologues (10, 21, 22).

FIGURE 1.

Schematic representation of the YscU domain architecture. YscU features an N-terminal inner membrane-interacting domain (green) consisting of four transmembrane α helices, which were predicted using the TMHMM server. The cytoplasmic domain of YscU, denoted YscU_C, has been crystallized with a construct comprising residues 240–339 (Protein Data Bank code 2JLI) and is shown in a schematic representation (25). The NPTH motif is indicated in red, and the polypeptides resulting from autocleavage are colored orange (YscU_CN) and gray (YscU_CC), respectively. The linker sequence (residues 211–239) connecting the N-terminal domain with YscU_C has been shown to interact with the inner membrane (25–27) and is modeled as an α helix.

The homologue of YscI in S. enterica serovar Typhimurium, PrgJ, has been proposed to form the inner rod, a structure that connects the needle to the basal body (20, 32). A study on Yersinia pseudotuberculosis showed that both YscP and YscU regulate YscI export (33). Indeed, YscI secretion by the T3SS is increased in a yscP-null mutant, whereas mutations in YscU_C reduced the secretion level of YscI in a ΔyscP strain. Interestingly, it was also recently shown that in Y. enterocolitica YscI is required for the export of YscP-PhoA hybrid into the periplasm (34). Moreover, an interaction between EscI and EscU (homologues of YscI and YscU, respectively) in Escherichia coli has been reported (35).

Here we report data regarding the interactions among YscU and the two proteins YscP and YscI. Using NMR spectroscopy, we showed that YscP binds to the C-terminal helix of YscU. The A335N point mutation within the YscP binding interface of YscU_C disrupted YscU-YscP interaction without affecting the YscU_C structure. The interaction was found to be functionally significant, and a yscU-null mutant expressing the YscU^A335N variant did not secrete Yops into the culture supernatant and did not induce cytotoxicity in HeLa cells. Furthermore, unlike a ΔyscP mutant or YscU mutants in the NPTH motif, YscU^A335N did not promote secretion of the needle subunit YscF. These results suggest that the YscU-YscP interaction is critical for secretion of both early and late substrates. In addition, we found that binding of YscP to the slowly autocleaving mutant YscU_C^P264A reduced the rate constant of the autocleavage event in vitro. This finding suggests that the binding of YscP to YscU_C can regulate the autocatalytic event, which in turn would affect the ability of YscU_C to dissociate and the subsequent secretion of its C-terminal polypeptide (YscU_CC). Finally, data from isothermal titration calorimetry and pulldown assays show that YscI interacts with YscU_C but not with either of the polypeptides resulting from autoproteolysis and dissociation (YscU_CN and YscU_CC).

Results

YscU Binds to the Disordered Segment of YscP

Previous genetic analysis showed that yscP deletion abolishes Yop secretion and that the wild-type phenotype can be partially rescued by the A268F, Y287G, and V292T suppressor mutations within YscU (10). Based on these results, it was proposed that an interaction between YscP and YscU was critical for Yop secretion. Here we characterize the interaction between the cytoplasmic domain of YscU (YscU_C) and YscP using NMR spectroscopy. A direct interaction between YscP and YscU_C homologues in P. aeruginosa (denoted PscP and PscU_C, respectively) has been observed previously by NMR (22); it was shown that a conserved N-terminal segment of PscP was responsible for its interaction with PscU_C. The domain architecture of PscP has been described as a “ball-and-chain” topology where the “ball”, i.e. the C-terminal domain, is folded and believed to function as a substrate switch, whereas the N-terminal “chain” polypeptide is disordered and flexible in solution (22). A similar architecture has been determined for the YscP homologue FliK from S. enterica serovar Typhimurium (21). A multiple sequence and structural alignment of YscP with FliK, PscP, and other homologues (supplemental Fig. 1) suggests that YscP has a structural topology resembling those described for FliK and PscP.

To investigate the domain architecture of YscP, we used three constructs (Fig. 2A) that were designed based on the results of a limited proteolysis experiment of YscP (supplemental Fig. 2) and a multiple sequence alignment (supplemental Fig. 1). These YscP variants were subjected to structural analysis using NMR and circular dichroism (CD) spectroscopy. First, size exclusion chromatography was used to determine the oligomeric state of YscP, YscP(342–442), and YscP(203–455) in solution. Interestingly, although the full-length YscP migrated as a 50-kDa monomer, YscP(342–442) and YscP(203–455) eluted with the column void volume, suggesting that both truncated variants aggregated in solution (supplemental Fig. 3). A CD spectrum of full-length YscP (Fig. 2B) indicated that the protein contained both unstructured and structured regions, and this was confirmed by the observation of bands at 205 and 220 nm in their CD spectra. The structured region of YscP was assigned to the C terminus of the protein by analyzing the CD spectra of the two truncated variants (YscP(203–455) and YscP(342–442)). To confirm that YscP(342–442) contained a structured domain, we acquired a ¹H-¹⁵N HSQC spectrum of a sample of the protein that was isotopically enriched with ¹⁵NH₄Cl. This NMR spectrum conclusively demonstrated that YscP(342–442) is a folded domain because it exhibits the large chemical shift dispersion typical of well folded proteins (supplemental Fig. 4). Although folded, the domain displayed tendencies to aggregate. The NMR spectrum was acquired at a concentration of 50 μm, and any increase in concentration resulted in precipitation of the protein. These results together demonstrate that YscP has a ball-and-chain domain architecture like its homologues. Next, we investigated the interaction between YscP and YscU_C by observing perturbations in the ¹H-¹⁵N HSQC spectrum of ¹⁵N-isotopically enriched YscU_C upon addition of the different YscP constructs. In general, interactions between partners in such cases induce changes in the chemical shifts or the line width of the peaks in the spectrum depending on the molecular masses of the interacting partners. The addition of YscP(342–442) did not trigger any changes in the NMR spectrum of YscU_C (Fig. 2C), suggesting that this YscP segment does not harbor the YscU_C interaction interface. In contrast, we observed extensive line broadening of the YscU_C resonances when YscP(203–455) was added (Fig. 2D), indicating that YscU_C becomes integrated into a high molecular weight complex following the addition of YscP(203–455). Based on this analysis, we propose that the YscU_C binding segment on YscP is localized between residues Glu²⁰³ and Arg³⁴¹.

FIGURE 2.

YscP domain architecture and its YscU_C-interacting segment. A, schematic representation of YscP and its truncated variants that were used in this study. B, far UV-CD spectra of YscP (black), YscP(342–442) (red), and YscP(203–455) (blue). All proteins were used at a concentration of 10 μm in 5 mm sodium phosphate, 30 mm NaCl, 1 mm DTT (pH 7.4) at 20 °C. mdeg, millidegrees. C, overlay of two-dimensional ¹H-¹⁵N HSQC spectra of free ¹⁵N-labeled YscU_C (blue) and ¹⁵N-labeled YscU_C mixed with unlabeled YscP(342–442) (red). D, overlay of two-dimensional ¹H-¹⁵N HSQC spectra of free ¹⁵N-labeled YscU_C alone (blue) and mixed with unlabeled YscP(203–455) (red).

YscP Interaction Interface on YscU_C

Despite intensive studies on YscP and YscU homologues, there are no high resolution structural data on the interaction interface on YscU_C responsible for its interaction with YscP. We identified this interaction interface using NMR spectroscopy by titrating unlabeled YscP to ¹⁵N-labeled YscU_C. The stepwise addition of unlabeled YscP to ¹⁵N-labeled YscU_C caused localized perturbations in the chemical shifts of the amide proton and nitrogen resonances of several YscU_C residues in the ¹H-¹⁵N HSQC spectrum of the protein (Fig. 3A). The binding interface was identified by displaying the residues with significant compounded chemical shift perturbations on the YscU_C structure (Fig. 3B). As expected for a well defined interaction interface, the YscP-interacting residues on YscU_C cluster to form a contiguous surface; the solvent-exposed residues Leu²⁸⁰, Ala³²⁸, Glu³³², Ala³³⁵, and Glu³³⁶ seem to form the YscP interaction interface. Interestingly, the interaction interface is centered on the C-terminal α helix of the YscU C-terminal domain (YscU_CC) (Fig. 3C). The titration experiment also enabled us to estimate the binding affinity (dissociation constant (K_d)) of the YscP-YscU_C interaction by monitoring changes in the chemical shifts of the methyl resonances of YscU_C as the concentration of YscP increased (Fig. 3D). These data were then fitted to an analytical function based on an assumed two-state binding reaction. The best fitted binding isotherm gave a K_d value of 430 ± 293 μm. Although the estimated K_d value suggests that the YscU-YscP interaction is relatively weak under our experimental conditions, this interaction may be modulated in vivo by other factors such as membrane anchoring (26) and/or additional protein partners.

FIGURE 3.

NMR-based identification of the YscP binding surface on YscU_C. A, overlay of the two-dimensional ¹H-¹⁵N HSQC spectra of free ¹⁵N-enriched YscU_C (100 μm) before (blue) and after addition of YscP (400 μm) (red). The spectra were acquired in PBS buffer containing 1 mm TCEP at 37 °C. Solvent-exposed residues that exhibit significant chemical shift perturbations upon YscP addition are marked with arrows. B, compounded chemical shift perturbations (ppm) between free ¹⁵N-labeled YscU_C and ¹⁵N-labeled YscU_C in complex with unlabeled YscP were calculated using Equation 1 and plotted against the YscU_C primary sequence. The threshold value used to define a significant chemical shift perturbation is indicated by the red line, and residues exhibiting significant shifts are represented by green dots except for solvent-accessible amino acid residues with significant chemical shift perturbations (Leu²⁸⁰, Ala³²⁸, Glu³³², Ala³³⁵, and Glu³³⁶), which are represented by red dots. C, display of the YscP binding interface on YscU_C. The structure of YscU_C illustrates the positions of the residues with significant chemical shifts (green) and the solvent-exposed residues with significant chemical shift perturbations (indicated as red ball and sticks). D, determination of the K_d value for the YscP-YscU_C interaction obtained by using Equation 2 to fit the changes in the chemical shifts of the methyl groups of YscU_C induced by YscP addition. Plots for three peaks in the methyl region are shown in black, red, and green, respectively. The error bars represent S.D.

Mutations at the YscU/YscP Binding Interface Regulate Yop and YscF Secretion

Based on the results presented above, we hypothesized that the interaction of YscU with YscP is mediated by the residues Leu²⁸⁰, Ala³²⁸, Glu³³², Ala³³⁵, and Glu³³⁶. To test our hypothesis and determine the physiological relevance of the YscU-YscP interaction, we performed a scanning mutagenesis analysis targeting these residues. The point mutations L280N, A328N, E332A, A335N, and E336A were separately introduced into the yscU coding sequence and cloned into the pBADmycHis-A plasmid. The resulting plasmids were then transformed into the pIB75 (ΔyscU) strain, which carries an in-frame deletion of the yscU gene. The ability of the different YscU variants to complement Yop secretion was assayed by culturing the cells in Ca²⁺-depleted medium with a temperature shift from 26 to 37 °C. These growing conditions mimic the host cell contact required for activation of secretion via the T3SS and lead to a massive secretion of Yop into the culture supernatant. The empty plasmid and the plasmid encoding YscU^WT were used as negative and positive controls, respectively. As expected, Yop secretion was restored by the plasmid encoding YscU^WT, and no Yop was detected in the supernatant of the strain carrying the empty plasmid (Fig. 4A). Similar amounts of Yop were detected in the supernatants of the strains expressing YscU^WT, YscU^E332A, and YscU^E336A (Fig. 4A). Thus, mutating residue Glu³³² or Glu³³⁶ to alanine did not affect the function of YscU or the secretion of Yop via the T3SS. Given that the YscU-YscP interaction is required for Yop secretion, this result suggests that Glu³³² and Glu³³⁶ may not be important residues for the binding of YscP to YscU. On the other hand, the mutation A335N drastically reduced the secretion of Yop into the supernatant (Fig. 4A, upper panel). The strain expressing YscU^L280N exhibited an intermediate phenotype with a decrease of about 80% of the Yop secreted by the wild type (Fig. 4A, upper panel). Lower amounts of intracellular Yop were detected for the strains bearing the empty plasmid or the plasmid expressing YscU^A335N (Fig. 4A). The attenuation of Yop secretion represses the expression of the yop genes, reducing the intracellular amount of the corresponding proteins. The ability of these different variants to generate a cytotoxic phenotype in HeLa cells was also assayed. Unlike YscU^WT and YscU^E332A, no cytotoxicity was generated by YscU^A335N (Fig. 4B), and the appearance of the cytotoxicity phenotype was delayed for YscU^L280N (Fig. 4B). Thus, both Yop secretion and translocation are affected by the A335N and L280N mutations. Taken together, these results indicate that Ala³³⁵ and Leu²⁸⁰ are critical for the interaction of YscU with YscP and Yop secretion via the T3SS. These mutations also impaired the secretion of the needle subunit YscF (Fig. 4A, third and fourth panels from the top). YscF is an early substrate, whereas Yop is a late substrate (34). Previous studies showed that the substrate specificity switch from early to late substrates is regulated by both YscU and YscP (10). The absence of YscF in the supernatant of the strain expressing YscU^A335N suggests that this mutation affects the secretion of both early and late substrates.

FIGURE 4.

Secretion profiles and cytotoxicity of Y. pseudotuberculosis expressing various point mutants of YscU. A, secretion profiles of Y. pseudotuberculosis strain YPIII/pIB75 transformed with plasmids encoding different point mutants of YscU. Culture supernatants were precipitated with TCA, separated on a Tris/Tricine gel, and stained with Coomassie R-250 (upper panels). Cell pellets of the corresponding strains were analyzed with anti-Yop antibody (second panel from the top). The amounts of secreted YscF (third panel from the top) and intracellular YscF (bottom panel) were analyzed using an anti-YscF antibody. The position of YscF is indicated by the black arrows, and the black asterisks indicate an unspecific band. B, HeLa cells were infected with bacteria at a multiplicity of infection of 10 for 45 min. After fixation with formaldehyde, the cytoskeleton actin filaments were stained with green phalloidin-Alexa Fluor 488 (Invitrogen) and nucleic acids were stained with DAPI (blue). Uninfected HeLa cells (panel a) exhibited non-cytotoxic morphologies similar to cells infected with strains bearing the empty plasmid (panel b). Cytotoxicity was observed for 100% of HeLa cells infected with the strains expressing YscU^WT (panel c) and YscU^E332A (panel e). YscU^L280N induced delayed cytotoxicity (panel d). YscU^A335N generated no cytotoxicity (panel f).

The results of the complementation analysis presented above showed that the residue Ala³³⁵ is critical for the function of YscU. The A335N mutation could thus impair either the folding of YscU or its interaction with YscP. To determine whether the A335N mutation affects the structure of YscU, we acquired a ¹H-¹⁵N HSQC spectrum of YscU_C^A335N and compared it with that of the wild type. An overlay of the two spectra revealed no significant differences in chemical shifts, and the majority of their cross-peaks have similar intensities and positions (Fig. 5A). The only residues that showed significant chemical shift perturbations are in close proximity to the mutation site, so we can conclude that the mutation does not affect the overall structure of YscU_C (Fig. 5, B and C). We then investigated whether YscU_C^A335N retained the ability to interact with YscP by performing an NMR titration of unlabeled YscP against ¹⁵N-labeled YscU_C^A335N at 37 °C (Fig. 5D) as we had done previously with YscU_C^WT. A comparison of the spectra from the titrations of YscP against YscU_C^WT and YscU_C^A335N showed that the chemical shift perturbations of YscU_C^A335N were significantly smaller than for YscU_C^WT (Fig. 5D). Furthermore, a K_d value for the interaction between YscU_C^A335N and YscP could not be quantified because the chemical shift changes were too small (supplemental Fig. 5). These data clearly indicate that the YscU_C^A335N mutant retains the structure of the wild-type protein but has lost its YscP binding activity.

FIGURE 5.

Disruption of YscP binding by the A335N replacement in YscU_C. A, overlay of two-dimensional ¹H-¹⁵N HSQC spectra of ¹⁵N-labeled YscU_C^WT (blue) and YscU_C^A335N (red) at concentrations of 100 μm. The buffer consisted of PBS and 1 mm TCEP at 37 °C. Solvent-exposed residues that exhibit significant chemical shift perturbations upon mutation are indicated with arrows. B, compounded chemical shift perturbations (ppm) between ¹⁵N-labeled YscU_C^WT and ¹⁵N-labeled YscU_C^A335N calculated by Equation 1 and plotted against the YscU residue number. The threshold value used to define significant chemical shift changes is indicated by the red line. The residues with the largest chemical shift perturbations (Ser²³⁶, Leu²⁸², Glu³³⁶, Leu³³⁸, and Arg³³⁹) cluster around the mutation site and are shown in blue; position 335 is shown in red on the YscU_C crystal structure in C. D, overlay of two-dimensional ¹H-¹⁵N HSQC spectra of ¹⁵N-labeled YscU_C^A335N at a concentration of 100 μm alone before (blue) and after addition of unlabeled YscP (red) at a molar ratio of 1:4 in PBS.

YscP Protects the Slow Processing Mutant P264A from Autocleavage

It has been shown previously that the dissociation of YscU_C into the two polypeptides YscU_CC and YscU_CN is required for Yop secretion and that YscU_CC is subsequently secreted into the culture supernatant (27, 29). The secretion of YscU_CC through the T3SS is facilitated by a C-terminal secretion sequence, and the non-secreting phenotype observed for yscP-null mutants is partially rescued by the secondary suppressor mutants YscU^A268F, YscU^Y287G, and YscU^V292T (10, 33). These suppressor mutants display elevated rate constants for the dissociation of YscU_C into YscU_CN and YscU_CC relative to wild-type YscU_C (28). To determine whether YscP can affect the dissociation of YscU_C, we studied the dissociation kinetics of YscU_C in the absence and presence of YscP by NMR spectroscopy as described previously (28). The experiments revealed that YscP does not significantly change the dissociation kinetics of YscU_C (supplemental Fig. 6). It is important to note that wild-type YscU_C is fully cleaved in these experiments because of the protein's long handling time during purification. To determine whether YscP can modulate the rate of autoproteolysis, we studied the slowly autocleaving mutant P264A (27) whose autoproteolysis proceeds slowly enough to allow purification of partially uncleaved protein. The in vitro cleavage kinetics of this mutant into the two polypeptide fragments YscU_CN and YscU_CC can be followed by Western blotting with an anti-YscU_CC antibody (Fig. 6 and supplemental Fig. 7) at 37 °C at 0, 1, and 2 days after purification. Each sample generated two bands in the immunoblot due to the incomplete autoproteolysis of the YscU_C^P264A variant; the uppermost band corresponded to uncleaved YscU_C, and the lower band corresponded to the cleaved YscU_CC polypeptide (Fig. 6). The total protein content for each sample was calculated from the sum of the bands corresponding to YscU_C and YscU_CC. The relative abundance of the uncleaved full-length protein (i.e. YscU_C variant) was then calculated as the percentage of total protein concentration. Following this analysis, it is evident that YscP can reduce the rate of YscU_C^P264A autoproteolysis in a concentration-dependent manner (Fig. 6). To test whether the effect is specific, we turned to the double mutant YscU_C^P264A/A335N, which has both slow autoproteolysis and a weakened YscP binding affinity. The YscP dose dependence for the protection against autoproteolysis is significantly weaker for the double mutant in comparison with the YscU_C^P264A single mutant. This observation suggests that the effect on autoproteolysis mediated by YscP is specific and driven by the interaction of YscP with the identified binding surface of YscU_C. Taken together, the data show that YscP can protect the slowly cleaving YscU_C mutant P264A from autoproteolysis in vitro.

FIGURE 6.

YscP binding attenuates the rate of YscU autocleavage. A, immunoblotting detection of the degree of autocleavage of YscUC^P264A and YscU_C^P264A/A335N using anti-YscU_CC as the primary antibody. Autocleavage was monitored over 2 days at 37 °C with varying concentrations of YscP. The YscU_C:YscP ratio was varied from 0 to 2 as indicated at the top of the figure. Each sample shows two bands in the immunoblot: the top band corresponds to uncleaved YscU_C, whereas the lower band corresponds to the YscU_CC polypeptide. B, the immunoblotting experiments were analyzed by quantifying the proportion of uncleaved YscU_C^P264A (blue) and YscU_C^P264A/A335N (orange) in the samples after 2-day incubation. Values shown in the graph are means with error bars representing ± S.D. (the experiment was repeated four times).

YscU Binds to the Inner Rod Protein YscI at Molar Ratio 1:1

The proposed T3SS structural model suggests that the protein YscI (Yersinia spp.), MxiI (Shigella spp.), or PrgJ (Salmonella spp.) polymerizes into an inner rod to form a conduit for protein secretion through the periplasm (20, 33, 36, 37). YscI is secreted via the T3SS, and it has been shown that YscI secretion is regulated by both YscP and YscU (33). It has also been shown that the inner rod protein EscI, a homologue of YscI in enteropathogenic E. coli interacts with the YscU homologue EscU (35). We therefore investigated whether YscU interacts with YscI and attempted to determine the binding interface (Fig. 7) by performing GST pulldown experiments in which YscU_C, YscU_CN, and YscU_CC were GST-tagged and used as bait to co-purify YscI-His₆. YscI-His₆ only bound to GST-YscU_C; no YscI remained associated to GST alone, GST-YscU_CN, or GST-YscU_CC (Fig. 7A). This interaction pattern suggests that the interaction of YscI with YscU_C is abolished by the dissociation of YscU_CC from YscU_CN and/or structural changes of YscU_CC and YscU_CN. To further characterize the binding of YscI to YscU, we analyzed this interaction using isothermal titration calorimetry (ITC) (Fig. 7B). The binding is characterized by a distinct exothermic binding process that can be robustly fitted to a one-site binding model (Fig. 7B) with a K_d of 3.8 μm and a 1:1 stoichiometry.

FIGURE 7.

The inner rod protein YscI interacts with YscU_C but not with YscU_CN or YscU_CC. A, GST pulldown assay using different variants of YscU and polyhistidine-tagged YscI. Top panel, verification of binding of GST, GST-YscU_C, GST-YscU_CN, and GST-YscU_CC to glutathione-Sepharose beads. This was accomplished by applying aliquots of the different proteins to glutathione-Sepharose beads, extensive washing to remove unbound material, and subsequent SDS-PAGE analysis of aliquots of beads bearing the bound proteins. The gels were stained with Coomassie R-250 for detection. Black asterisks indicate degradation products. Bottom panel, GST and GST-tagged YscU variants were incubated with purified YscI. After several washes, resin-associated proteins were eluted and separated on the gel for immunoblotting detection with an anti-His₆ antibody. The black arrow indicates the position of YscU_CC in the GST-YscU_C sample. B, raw data from an ITC experiment involving addition of YscU_C to YscI are shown in the upper panel; the lower panel shows the integrated heats obtained after subtracting the heat of dilution. The best fitting binding isotherm is shown as a solid line. The fitted parameters are K_d = 3.8 μm, ΔH⁰ = −17 kcal mol⁻¹, ΔS⁰ = −33.8 cal mol⁻¹ degrees⁻¹, and a stoichiometry “n” of 0.8 at 25 °C. C, far-UV CD spectrum of YscI. The experiment was conducted with 10 μm YscI in a 5 mm sodium phosphate buffer (pH 7.4) containing 30 mm NaCl and 1 mm DTT at 25 °C. mdeg, millidegrees.

Previous structural analyses of two YscI homologues, MxiI and PrgJ, using CD and NMR spectroscopy have shown that they are partially folded in solution (36). Although MxiI, PrgJ, and YscI have equivalent functions in the T3SS, their sequence similarity is relatively low (supplemental Fig. 8). The CD spectrum of YscI features distinct CD bands in the region between 205 and 220 nm, which is indicative of significant secondary structure content (Fig. 7C).

We attempted to characterize the binding interface of YscI on YscU_C by performing two-dimensional NMR experiments. A titration of unlabeled YscI against ¹⁵N-labeled YscU_C resulted in massive line broadening of the peaks corresponding to YscU_C residues due to aggregation of YscI in solution into complexes that are too large to permit the detection of YscI-bound YscU_C resonances. However, a two-dimensional ¹H-¹⁵N HSQC experiment using ¹⁵N-labeled YscU_C^A335N showed that the binding of YscI was not abolished by the mutation A335N in contrast to that of YscP (supplemental Fig. 9). The interactions of YscP and YscI with YscU thus appear to occur at different binding surfaces.

Discussion

Type III secretion systems are complex nanomachines spanning the cell wall of Gram-negative bacteria. They are constituted by around 25 proteins whose assembly requires that the individual components are added in a specific order (38). Furthermore, the T3SS substrates (Yops in Yersinia) that are destined for the eukaryotic cytosol are not exported before cell contact between the pathogen and the eukaryotic target cell has been established (39). Thus, there exists a secretion hierarchy allowing early T3SS substrates to be secreted before the secretion of the late substrates is allowed. This process was first described for the flagellar T3SS and was named “substrate specificity switching” by Macnab (12). In Yersinia spp., YscP and YscU have been shown to be essential components of the substrate specificity switching mechanism. A yscP mutant assembles elongated needles, but it is unable to secrete Yop effectors (19). Similar phenotypes have also been reported for the homologous proteins in Salmonella and Shigella T3SSs (31, 40). Interestingly mutations in the cytoplasmic domain of yscU can suppress a yscP mutant phenotype, leading to decreased export of the needle protein YscF and increased Yop secretion (10, 27). These genetic results argue strongly for a direct YscP and YscU interaction that is important to enable Yersinia to regulate the substrate switch. YscU and its homologues in other bacteria undergo autocatalytic cleavage at a conserved NPTH motif, generating a C-terminal 10-kDa peptide fragment called YscU_CC in Yersinia. YscU_CC remains bound to the membrane-anchored part of YscU after cleavage (27). Removal of Ca²⁺ from the growth medium results in Yop secretion as well as secretion of YscU_CC, indicating that YscU is dissociated under these conditions (28). Furthermore, we also showed previously that yscP suppressor mutants mapping to the yscU_CC part of the yscU gene destabilized the interaction between the two YscU polypeptides generated by autocleavage, indicating that YscU_CC dissociation is an essential step in substrate switching in Yersinia. It has therefore been proposed in line with these findings that the decisive step in substrate switching is the autocatalytic event and that YscP is a key regulator of this process.

Here we used solution state NMR spectroscopy to identify the YscP binding interaction interface on the cytoplasmic domain of YscU (YscU_C). The interaction surface is centered on the C-terminal α helix of YscU_C, and disruption of the surface by introducing the A335N point mutation dramatically reduced Yop secretion and cytotoxicity on HeLa cells. On the basis of biophysical analyses with circular dichroism and NMR spectroscopy, we show that YscP has a ball-and-chain architecture with an extended N-terminal flexible segment attached to a folded C-terminal domain. A similar architecture has been demonstrated previously for the orthologue PscP from P. aeruginosa (22) and for FliK from S. enterica serovar Typhimurium (21). These results suggest that the ball-and-chain architecture is strongly conserved among the YscP class of bacterial proteins. By performing a deletion analysis of YscP, we showed that the protein interacts with YscU via amino acid residues in its disordered N-terminal segment. This is analogous to the scenario observed in P. aeruginosa where N-terminal residues of PscP interact with the YscU orthologue PscU (22). It has been proposed that the C-terminal folded domain of PscP is located at the growing tip of the needle structure (22) and that the N-terminal flexible segment spans the entire T3SS to make contact with PscU located in the cytoplasm. This model is consistent with the fact that YscP has been identified as a secreted protein in Yersinia (41) and that it is also surface-located. In accordance with these results, YscP has been postulated to function as a “molecular ruler” that defines the correct length of the needle structure (19). An issue that remains to be answered is the nature of the mechanism that enables discrimination between early and late substrates. As discussed, we have shown previously that autoproteolysis of YscU enables secretion of the C-terminal YscU polypeptide (YscU_CC) and that its secretion parallels that of the Yops (28). Destabilized YscU mutants that displayed increased rate constants of dissociation also enable secretion of Yops at 30 °C, a temperature at which the wild-type strain effectively displays no Yop secretion. In addition, we previously identified a C-terminal secretion signal in YscU_CC (29). Taken together, these findings suggest that YscU_CC dissociation and subsequent secretion enabled by autoproteolysis play an important role in regulating Yop secretion. However, a contradictory hypothesis exists whereby the role of autoproteolysis is to shape the three-dimensional structure of YscU and thereby enable the protein to engage in downstream protein-protein interactions (42). This hypothesis is supported by the observation that uncleaved SpaS (the YscU orthologue in Salmonella) could not be detected inside bacteria. In addition, in an elegant experiment, the authors engineered a SpaS variant that was cleaved by an external protease; also with this experiment no regulatory function of SpaS cleavage could be verified. Taken together, these observations suggest that self-cleavage of SpaS has no regulatory function regarding the substrate specificity switch (42). However, it is possible that the time resolution in the experiments did not allow detection of uncleaved SpaS. Hence, in the context of wild-type SpaS, it is currently problematic to test regulatory aspects of self-cleavage. To test whether autoproteolysis could be regulated in Yersinia, we turned to the slowly cleaving YscU_C mutant P264A. The YscU_C^P264A mutant has a half-life for autoproteolysis that enables biochemical experiments aimed at probing the role of YscP for the rate of autoproteolysis and/or dissociation. To test for potential effects on YscU_C dissociation kinetics in response to YscP binding, we relied on our established NMR protocol for quantifying this event (28). We could not observe any statistically significant changes in the rate constant of wild-type YscU_C dissociation in the absence or presence of YscP. Using the YscU_C^P264A mutant and a protocol based on Western blotting with an antibody against YscU_CC, we showed that the presence of YscP significantly reduced the rate constant of autoproteolysis. This finding opens the possibility that YscU might be maintained in a state that prevents late substrate secretion when bound to YscP and that release of YscP enables YscU self-cleavage, YscU_CC dissociation, and activation of late substrate secretion. This inference is consistent with the model of Bergeron et al. (22) where PscP is anchored to PscU on the cytoplasmic side with the N-terminal flexible segment threaded through the needle and the C-terminal folded domain located at the tip of the needle.

It was recently shown by Lloyd and co-workers (33) that the formation of the inner rod by the YscI protein exhibits a critical role in the substrate specificity switch. They also showed that YscU and YscP indirectly control Yop secretion by regulating the export of YscI. These results are in perfect agreement with the view of Marlovits and co-workers (43) who showed that the inner rod was absent in YscP mutants, leading to the conclusion that the inner rod is essential for substrate specificity switching. In addition, it has also been shown that EscI interacts with EscU (35). In line with these observations, we found that YscI interacted with the cytosolic domain (YscU_C), but in contrast YscI was unable to interact with either of the two peptides resulting from autoproteolysis. Thus, we conclude that YscI interacts with YscU and that autoproteolysis per se does not affect YscI interaction with the two resulting YscU polypeptides as long as they are interconnected. In addition, we recently showed that induction of the switch to late Yop secretion also resulted in secretion of YscU_CC (28). This result shows that YscU_CC is dislocated from the remaining part of YscU after induction of the switch (28). One possibility based on these findings is that YscU anchors the inner rod to the T3SS and that induction of the switch results in disassociation of the inner rod followed by Yop secretion. In accordance with this hypothesis and the results presented herein, we further suggest that YscP/YscU binding blocks autoproteolysis, and thus YscU_CC dissociation followed by YscI release is inhibited. This mechanism requires that YscU_C harbor distinct YscI and YscP binding surfaces, and this feature was established in the current study. In this scenario, YscP could fit very well to serve as the enigmatic T3SS surface sensor because it has also been shown that a domain of YscP is surface-located prior to induction of the switch (41). Thus, YscP can serve as a link between the exterior of the T3SS and proteins regulating the substrate specificity switch, thus fulfilling the basal criteria for a surface sensor. More work is needed in this context, and we now focus our efforts to test this hypothesis.

Experimental Procedures

Bacterial Strains, Plasmids, and Growth Conditions

The bacterial strains and plasmids used in this study are listed in supplemental Table S1. Standard molecular biology methods were used to generate the different plasmid constructs used in the experiments. The PCR primers used in the different clonings are listed in supplemental Table S2. The sequences of all the constructs were verified by sequencing (Eurofins MWG Operon). E. coli strains were grown in Luria-Bertani broth (LB) or on Luria agar plates at 37 °C. Y. pseudotuberculosis strains were grown at 26 °C in LB or on Luria agar plates. Antibiotics were added to the medium for selection according to the resistance markers carried by the plasmid at the following concentrations: kanamycin, 50 μg/ml; carbenicillin, 100 μg/ml; and chloramphenicol, 35 μg/ml.

Yop Secretion Analysis and Immunoblotting Quantification

To induce Yop secretion, Y. pseudotuberculosis strains were first grown at 26 °C for 2 h in Ca²⁺-depleted LB medium (medium containing 5 mm EGTA and 20 mm MgCl₂) and then at 37 °C for 3 h. Cultures were started at an A₆₀₀ of 0.1. Samples were treated as described previously (28) and separated on a Tris-Tricine polyacrylamide gel. Proteins were either stained with Coomassie R-250 or transferred onto a PVDF membrane for immunoblotting. Anti-Yop, anti-His, and anti-YscF antibodies were diluted at 1:5,000. Horseradish peroxidase-conjugated anti-rabbit antibody was diluted at 1:10,000 (GE Healthcare). Proteins were detected with a chemiluminescence detection kit (GE Healthcare).

To monitor the effect of YscP binding on YscU autocleavage, YscP was mixed with 20 μm YscU_C variants at different molar ratios (0, 0.01, 0.1, 0.25, 0.5, 1, and 2) in PBS with 1 mm TCEP and incubated at 37 °C. Samples were taken after 0, 1, and 2 days and diluted to a concentration of 1 μm in the YscU_C variant before mixing with Laemmli buffer (Bio-Rad). The proteins in the samples were then separated by SDS-PAGE and transferred to a PVDF membrane (Bio-Rad) for immunoblotting. Anti-YscU_CC was diluted at 1:10,000. Goat anti-rabbit IgG (heavy + light)-HRP conjugate antibody was diluted at 1:20,000. Transferred proteins were detected with a chemiluminescence detection kit (Bio-Rad). Densitometric analysis of the bands was performed using the ImageJ software package based on the same immunoblotting samples (44).

HeLa Cell Cytotoxicity Assay

Y. pseudotuberculosis cultures were started at an A₆₀₀ of 0.1 in LB medium containing 1 mm CaCl₂. After 1 h of growth at 26 °C, cultures were shifted to 37 °C for 2 h. HeLa cells were infected for 45 min at a multiplicity of infection of 10. Cytotoxicity assays were performed as described by Rosqvist and co-workers (45) with some modifications. For immunostaining, samples were subsequently fixed with 4% paraformaldehyde and permeabilized with 0.5% Triton X-100 (Sigma-Aldrich). Unspecific binding was suppressed by treatment with 0.1 m glycine in PBS containing 1% bovine serum albumin. Alexa Fluor 488-phalloidin (Life Technologies) and DAPI were used to stain the actin cytoskeleton and nucleic acids of the cells, respectively. Samples were mounted in mounting medium (Dako) and examined with a fluorescence microscope (Nikon Eclipse C1 Plus).

Protein Expression and Purification

Proteins produced in the BL21(DE3)/pLysS strain were grown in Luria broth at 37 °C. When an A₆₀₀ of 0.6 was reached, protein production was induced by adding 1 mm isopropyl β-d-1-thiogalactopyranoside (IPTG) to the culture medium and allowing the bacteria to grow for a further 4 h. Bacteria were harvested by centrifugation for 30 min at 5,000 × g, and the pellets were frozen at −80 °C. For ¹⁵N labeling, bacteria were grown in M9 minimal medium containing ¹⁵NH₄Cl. After addition of 1 mm IPTG at an A₆₀₀ of 0.6, ¹⁵N-labeled proteins were isolated from bacteria grown overnight at 30 °C and harvested as described above.

GST-tagged variants of YscU_C and YscP were purified as described previously (28). Bacteria were resuspended in the lysis buffer (50 mm Tris, 100 mm NaCl, 2 mm DTT (pH 7.4)) and lysed by sonication. The lysates were then centrifuged for 30 min at 30,000 × g at 4 °C. The supernatant was loaded on a GSTrap column (GE Healthcare) pre-equilibrated with the lysis buffer, the column was washed with 10 column volumes of the same buffer, and then the GST-tagged protein was eluted with 50 mm Tris, 100 mm NaCl, 20 mm glutathione (pH 7.4) buffer. Glutathione was eliminated from the eluate using a centrifugal filter device, and the GST tag was cleaved by incubating the resulting samples with 10 units of PreScission protease (GE Healthcare)/mg of protein overnight at 4 °C. The free GST was eliminated by passing the sample through a GSTrap column with the purified proteins remaining in the column flow-through. The purification of YscP was finalized by loading untagged YscP on a HiPrep 26/60 Sephacryl S-100HR gel filtration column (GE Healthcare) after a buffer exchange to PBS containing 2 mm DTT (pH 7.4). YscU_C was loaded on a Sepharose SP cation exchange column (Sigma-Aldrich) after a buffer exchange to 10 mm Tris containing 25 mm NaCl (pH 7.4) and eluted using an NaCl gradient from 25 to 800 mm in 10 mm Tris (pH 7.4). Both YscP and YscU_C were concentrated in PBS containing 1 mm TCEP (pH 7.4) using centrifugal filter devices.

His₆-YscP(203–455) was produced and lysed as described above. The supernatant was then loaded on a HisTrap column (GE Healthcare). The protein was eluted with an imidazole gradient from 10 to 700 mm in PBS. After removal of imidazole using centrifugal filter units, His₆-YscP(203–445) was loaded on a 26/60 Sephacryl S-100HR gel filtration column.

His₆-maltose-binding protein (MBP)-YscP(342–442) with a double His₆-MBP tag was produced overnight at 25 °C by adding 1 mm IPTG when the A₆₀₀ of the culture reached 0.6. Inclusion bodies were solubilized in 50 mm Tris containing 100 mm NaCl and 8 m urea (pH 7.4). After centrifugation for 30 min at 30,000 × g at 4 °C, the supernatant was dialyzed against the same buffer containing 4 m urea for 2 h at 4 °C. The procedure was then repeated with buffers containing 2 m and then 1 m urea. The final dialysis was carried out overnight against 50 mm Tris containing 100 mm NaCl and 10 mm imidazole after which the refolded protein was loaded onto a HisTrap column and eluted with an imidazole gradient from 10 to 700 mm in the loading buffer. The imidazole concentration was then reduced to 10 mm using centrifugal filter devices, and the His₆-MBP tag was removed by overnight incubation at room temperature with 100 units of tobacco etch virus protease/mg of protein. 8 m urea was added to the resulting solution, and the uncleaved His₆-MBP-YscP(342–442) was removed by affinity chromatography on a HisTrap column with the YscP(342–442) remaining in the flow-through. The buffer was then exchanged to 6 m urea in PBS, and the protein was purified by size exclusion chromatography using a HiPrep 26/60 Sephacryl S-100HR column.

For YscI-His₆ purification, bacteria were resuspended and lysed in 50 mm Tris containing 100 mm NaCl (pH 7.4). The lysates were then centrifuged for 30 min at 30,000 × g at 4 °C. Because YscI-His formed inclusion bodies, the pellet was dissolved in a binding buffer of 50 mm Tris containing 100 mm NaCl and 7 m urea (pH 7.4) and then centrifuged. The resulting supernatant was loaded on a HisTrap column, and urea was eliminated by washing the column with a non-urea buffer of 50 mm Tris containing 100 mm NaCl (pH 7.4). The protein was then eluted with an imidazole gradient from 0 to 1 m in 50 mm Tris containing 100 mm NaCl (pH 7.4). Imidazole was eliminated by dialysis overnight at 4 °C into an appropriate buffer for NMR or isothermal titration calorimetry.

NMR Spectroscopy

¹⁵N-Labeled proteins were studied at 100 μm in PBS containing 1 mm TCEP and 10% D₂O. In the YscP titration experiments, the quantity of added YscP was chosen to achieve a final concentration of 50, 100, 200, 300, or 400 μm and a YscU_C:YscP molar ratio of 0.5, 1, 2, 3, or 4. ¹H-¹⁵N HSQC spectra were acquired at 37 °C using a Bruker Avance III HD spectrometer at 850 MHz equipped with a 5-mm TCI HCN triple resonance cryoprobe. The NMR spectra were processed with Topspin (Bruker Biospin) or NMRPipe and visualized using ANSIG for Windows (46, 47). Compounded chemical shift perturbations for each backbone ¹H-¹⁵N signal were computed using the following equation.

$$\Delta \delta \hspace{0.17em}\left(\text{ppm}\right)=\left|\Delta {\delta }_{\text{H}}\right|\hspace{0.17em}+0.2\times \left|\Delta {\delta }_{\text{N}}\right|$$ (Eq. 1)

where Δδ_N and Δδ_H are the chemical shift perturbations of the amide nitrogen and amide proton of the assigned residue in the free and complexed forms of the protein, respectively. The calculated chemical shift perturbations were plotted against residue numbers to localize the affected surfaces.

In the binding studies, the chemical shifts of the methyl resonances in ¹H one-dimensional NMR spectra of YscU_C were monitored during YscP addition. The YscP-YscU_C interaction occurs under the fast exchange regime on the NMR time scale, so binding constants (K_d) were estimated by fitting the experimental data to Equation 2 based on an assumed 1:1 binding reaction (P + L ↔ PL) where YscP is regarded as “ligand L,” YscU_C is regarded as “protein P,” and the YscP-YscU_C complex is regarded as “PL”. Δδ_max is the perturbation of the chemical shift when the binding site is saturated (48).

$$\Delta \delta \hspace{0.17em}\left(\text{ppm}\right)=\frac{\Delta {\delta }_{\max }}{2}\left(\left(1+x+\frac{K_d}{P}\right)-\sqrt{{\left(1+x+\frac{K_d}{P}\right)}^2-4x}\right)$$ (Eq. 2)

where P is concentration of protein and x is molar ratio of ligand:protein. The reported K_d is an average value based on the individual K_d values of the analyzed peaks.

Limited Trypsin Proteolysis

Limited trypsin proteolysis was utilized to determine the domain architecture of YscP. A 22 μm solution of the full-length YscP protein in PBS was incubated with sequencing grade trypsin (Promega) at ratios of 1:100, 1:1,000, and 1:10,000, and samples were taken after 1-, 5-, 10-, and 20-min incubation at 4 °C. The peptide content of these samples was separated on a 15% SDS-acrylamide gel and stained with Coomassie R-250. The bands corresponding to stable fragments were then excised and identified by ultraperformance liquid chromatography-mass spectroscopy/mass spectroscopy.

CD

Far-UV CD spectra were recorded in a 0.1-cm quartz cuvette on a Jasco J-810 spectropolarimeter equipped with a Peltier element for temperature control. The spectra were recorded using 10 μm protein solutions in 5 mm sodium phosphate, 30 mm NaCl, 1 mm DTT (pH 7.4) buffer. Spectra were acquired at 20 °C from 260 to 190 nm in continuous scanning mode with 0.5-nm steps, a bandwidth of 2 nm, and a scan speed of 50 nm/min. Five spectra per sample were accumulated and averaged to improve the signal to noise ratio. The appropriate solvent spectrum background was subtracted from each of these spectra.

ITC

ITC experiments were performed in PBS buffer (pH 7.4) at 25 °C using a MicroCal Auto iTC200 instrument (GE Healthcare). A 900 μm YscU_C solution in the syringe was titrated against 90 μm YscI in the cell. The experiment involved 20 injections of 2 μl of the titrant protein (YscU_C) with 150 s between injections. The resulting raw data were fitted to a 1:1 binding model using the Origin 7.0 software package (GE Healthcare).

Pulldown of GST-tagged YscU_C Variants with YscI

Proteins were produced as described above. GST-tagged YscU variants were resuspended in PBS buffer containing 1% Triton X-100, sonicated, and centrifuged for 20 min at 20,000 × g at 4 °C. The resulting supernatants were then incubated with a glutathione-Sepharose 4B slurry (GE Healthcare) for 1 h at 4 °C with gentle rocking. Bound proteins were washed with PBS containing 0.1% Triton X-100 and then used in an interaction assay. The YscI pellet was resuspended in PBS buffer containing 8 m urea, sonicated, and centrifuged for 20 min at 20,000 × g at 4 °C. The supernatant was then incubated with HIS-Select Nickel Affinity Gel (Sigma-Aldrich) for 1 h at 4 °C. Bound YscI was washed with PBS containing 0.1% Triton X-100 and eluted with 0.4 m imidazole in the same buffer. The GST-tagged YscU variants bound to the beads were incubated with 10 μm YscI for 3 h at 4 °C with gentle rocking. The beads were then washed with PBS containing 0.1% Triton X-100, and the proteins that remained bound to the beads were eluted using SDS-PAGE sample buffer and analyzed by immunoblotting.

Author Contributions

O. H., F. H. L., H. W.-W., and M. W.-W. designed experiments, analyzed data, and managed the project. O. H. and P. R. conducted experiments to characterize the domain structure of YscP. O. H. performed experiments to study the binding interface of YscP binding on YscU, YscP protection for the mutant P264A from autocleavage, and interaction of YscU and YscI by ITC. F. H. L. performed Yop and cytotoxicity assays for functional characterization of mutations at the YscU/YscP binding interface in vivo and GST binding assay for YscI to GST-tagged YscUC variants and discussed the project. T. E. performed secretion experiments. The manuscript was written by O. H., M. W.-W., F. H. L., and H. W.-W. All authors discussed the results and commented on the manuscript.