Structure-Function Analysis of the C-Terminal Domain of LcrV from Yersinia pestis

Abstract

LcrV, a multifunctional protein, acts as a positive regulator of effector protein secretion for the type III secretion system (T3SS) in Yersinia pestis by interaction with the negative regulator LcrG. In this study, LcrV was analyzed to identify regions required for LcrG interaction. Random-linker insertion mutagenesis, deletion analysis, and site-directed mutagenesis of hydrophobic amino acids between residues 290 and 311 allowed the isolation of an LcrV mutant (LcrV L291R F308R) defective for LcrG interaction. The new residues identified in LcrG interaction lie in helix 12 of LcrV; residues in helix 7 of LcrV are known to be involved in LcrG interaction. Helix 7 and helix 12 of LcrV interact to form an intramolecular coiled coil; these new results suggest that the intramolecular coiled coil in LcrV is required for LcrG interaction and activation of the T3SS.

Yersinia pestis, the etiological agent of plague, harbors a 70-kbp plasmid called pCD1 that is essential for Y. pestis virulence (23). pCD1 encodes a type III secretion system (T3SS) that is comprised of a structural type III secretion (T3S) apparatus (termed the Ysc) and the virulence-associated effector proteins, collectively called Yops (1). Y. pestis utilizes the Ysc to translocate Yops into the cytoplasm of host cells, where they disrupt cellular signaling cascades involved in phagocytosis and activation of the early innate immune response (2). Expression of the T3SS and Yops is activated at mammalian body temperatures, i.e., 37°C, while the secretion of Yops is activated by environmental cues (29). In vitro, secretion of Yops is activated by the depletion of Ca²⁺ from the growth media at 37°C (16). In vivo, the secretion of Yops is triggered by eukaryotic cell contact (7, 25). The process of toxin translocation is polarized; Yops are translocated directly into the cytoplasm of host cells and are not found in substantial amounts in the extracellular environment.

Under in vitro growth conditions lacking Ca²⁺, maximal Yop synthesis and secretion are attained, and the bacteria undergo a cessation of growth (termed growth restriction) (21). Conversely, addition of Ca²⁺ to the growth medium blocks secretion of Yops and prevents the bacteria from entering growth restriction. Blockage of Yop secretion prior to eukaryotic cell contact or in the presence of Ca²⁺ is mediated by several negative regulators, including YopN (8, 31), TyeA (10), SycN (3), YscB (11), and LcrG (14, 18, 26). Deletions of any of the negative regulators result in constitutive secretion of Yops, growth restriction, and maximal Yop expression under secretion-nonpermissive conditions.

In addition to the negative regulators, LcrV is a T3S protein that positively regulates Yop secretion. Positive regulation of Yop secretion by LcrV is attained by binding to the negative regulator LcrG (14, 18). Mutations in LcrG or LcrV that disrupt the interaction of LcrV and LcrG result in a constitutive blockage of Yop secretion (12, 14). The description of an LcrG mutant that lacked LcrV interaction led to the refinement of the LcrG titration model, which suggests how the interaction between LcrG and LcrV may control Yop secretion (14, 18). The interaction between LcrG and LcrV is thought to occur through hydrophobic coiled-coil domains (12, 13). Lawton et al. showed that LcrV interacts with LcrG through a predicted internal alpha helix located between amino acids 148 and 169 of LcrV (12). The resolved crystal structure of LcrV confirms the helical nature of the region between amino acids 148 and 169; however, this internal helix was found to mediate an intramolecular coiled-coil interaction with a helix at the C terminus of LcrV (5).

To determine if other regions in LcrV are involved in LcrG interaction, an LcrV insertional-mutant library was created and screened for mutants that failed to interact with LcrG in a yeast two-hybrid system. Our screen revealed three independent insertions of five amino acids at the C terminus of LcrV that dramatically reduced their ability to interact with LcrG. To further characterize the role of the C terminus of LcrV in mediating LcrV's interaction with LcrG, two C-terminal-truncation mutants of LcrV were created and studied for their ability to bind LcrG and to activate Yop secretion. Our results demonstrate that the C-terminal region between amino acids 290 and 311 of LcrV is required for interaction of LcrV with LcrG and the subsequent activation of Yop secretion. Using radical site-directed mutagenesis of hydrophobic amino acids in the C-terminal region of LcrV, we were able to identify residues that are crucial for binding LcrG and activating Yop secretion. In addition, we confirmed that the interaction between LcrG and LcrV occurs through hydrophobic forces.

The C terminus of LcrV is required for LcrV's interaction with LcrG.

To identify regions in LcrV required for interaction with LcrG, random five-amino-acid insertions were created in LcrV using transposon-mediated linker insertion mutagenesis (mutation generation system; Finnzymes, Espoo, Finland) targeting lcrV on plasmid pJM15 (14). This LcrV library was screened for mutants that failed to interact with LcrG in a yeast two-hybrid system (13, 14). Screening of the mutant LcrV library for reduced LcrG interaction revealed three independent linker insertions near the C terminus of LcrV that dramatically reduced LcrG interaction in Saccharomyces cerevisiae (Table 1). Sequencing these LcrV mutants revealed in-frame insertions resulting in the addition of five amino acids after residues 277 (pTMV74; pJM15, SMRPQ insertion), 289 (pTMV84; pJM15, QLLRP insertion), and 292 (pTMV12; pJM15, DCGRT insertion) of LcrV. These results suggest that LcrV's C terminus is critical for LcrG interaction.

TABLE 1.

Two-hybrid analysis of LcrV mutants with LcrG

pJM17 derivative	Filter lift reaction	β-Gal activitya
pJM17 aloneb	White	7 ± 1
pJM15; wild-type LcrV	Blue	159 ± 13
pTMV12; LcrV insertion after aa 292	White	16 ± 7
pTMV74; LcrV insertion after aa 277	White	14 ± 4
pTMV84; LcrV insertion after aa 289	White	12 ± 4
pMH14; LcrV290	White	14 ± 5
pMH131; LcrV311	Blue	158 ± 4
pMH129; LcrV L291R F308R	White	21 ± 5
pMH130; LcrV L291A F308A	Blue	142 ± 4

^aLiquid β-gal activity (Miller units); data are means ± standard deviations.

^bThe value is derived from yeast expressing LcrV and an empty vector control.

To refine the region in LcrV's C terminus necessary for binding LcrG, two C-terminal truncation mutants of LcrV were fused by PCR to the GAL4 binding domain (Gal4-BD) in plasmid pJM17 (14) to remove the last 35 and 15 amino acids of LcrV to create LcrV290 (pMH14) and LcrV311 (pMH131). LcrV290 and LcrV311 were studied for their ability to interact with LcrG in yeast along with LcrG fused to the GAL4 activation domain (GAL4-AD) on a compatible vector (pACTG) (14). Liquid β-galactosidase (β-Gal) assays were used as a quantitative measurement for the LcrG-LcrV interaction. The β-Gal assays demonstrated that LcrV290 had dramatically reduced levels of β-Gal activity compared to the full-length LcrV (Table 1) and that LcrV311 had β-Gal activity similar to that of full-length LcrV, indicating that LcrV311 still binds to LcrG in yeast (Table 1). These results demonstrated that the region between amino acids 290 and 311 of LcrV was required for LcrG-LcrV interaction.

Phenotypes of LcrV290 and LcrV311.

To evaluate the physiological significance of the LcrV truncations, the ability of LcrV290 and LcrV311 to activate Yop secretion in Y. pestis was analyzed. A nonpolar deletion in lcrV (ΔlcrV5; KIM8-3002.N1) was generated in Y. pestis KIM8.3002 (17) by using allele replacement as previously described (15, 17). The suicide plasmid pMHΔlcrV5 was constructed to remove the DNA sequence encoding residues 8 to 268 of LcrV to give the lcrV5 allele. pMHΔlcrV5 was made by cloning two PCR fragments approximately 1 kb upstream and downstream of lcrV into SalI- and BamHI-linearized pLD55 (15). The upstream and downstream DNA fragments were obtained using PCR with Deep Vent DNA polymerase (New England Biolabs, Beverly, MA). Primers used to amplify the upstream fragment were Delta-LcrV US SalI (5′-ACG CGT CGA CGA TAT CTG CTC GAA CAG A-3′) and Delta-LcrV US KpnI (5′-ATG GTA CCC TTA GGG TTT TGT TCG TAG GCT CTAATC-3′). Primers used to amplify the downstream fragment were Delta-LcrV DS KpnI (5′-CGG GTA CCT AAC CAC CAC CTG CTC GGA TAA GTC CAG GCC GC-3′) and Delta LcrV-DS BamHI (5′-CGC GGA TCC CAC TGA GGC TAT GGC GCT GAG CC-3′). The fragments were digested with SalI and KpnI or with BamHI and KpnI, respectively, and ligated simultaneously into pLD55. Replacement of lcrV with ΔlcrV5 resulted in an anticipated calcium-independent phenotype with no growth restriction (data not shown) and a decrease in Yop expression, demonstrated by decreased YopE expression (Fig. 1A), and a lack of Yop secretion (Fig. 1B), consistent with previous studies (22, 24, 27). As expected, transcomplementation of the ΔlcrV5 strain with LcrV restored Yop regulation (Fig. 1A) and secretion (Fig. 1B). LcrV290 and LcrV311 were cloned by PCR behind araBADp on plasmid pBAD18 (15), resulting in plasmids pMH13 and pMH122. Plasmids encoding LcrV290 and LcrV311 were separately electroporated (23) into the ΔlcrV5 strain to study the ability of the LcrV truncations to activate Yop secretion. Y. pestis strains were grown in the chemically defined medium TMH (28) in the presence or absence of Ca²⁺ for 4 h at 37°C and then harvested by centrifugation to separate secreted proteins from cellular proteins (17). The whole-cell fraction was examined by immunoblotting with anti-YopE and anti-LcrV to study the induction of lcr-regulated genes, such as yopE, and the stable expression of the LcrV truncations. Transcomplementation of ΔlcrV5 with LcrV290 did not restore Yop secretion in the absence of Ca²⁺ (Fig. 1B) and resulted in a Ca²⁺-independent phenotype with respect to growth (data not shown) and LCR induction (Fig. 1A). The failure of LcrV290 to induce Yop secretion is consistent with the expected phenotype of an LcrV incapable of LcrG interaction.

FIG. 1.

The C terminus of LcrV is required for LcrV's ability to activate Yop secretion. Equivalent amounts of whole-cell and cell-free culture supernatants samples were separated by SDS-PAGE in a 12.5% polyacrylamide gel and analyzed by silver staining for secreted Yops and immunoblotting for cellular proteins. (A) Immunoblotting of cellular proteins. Samples of the cellular fractions were analyzed by immunoblotting with anti-LcrV and anti-YopE. Cellular fractions of Y. pestis KIM8-3002 containing pBAD18 (vector; lanes 1 and 2) and Y. pestis KIM8-3002.N1 (ΔlcrV5) containing plasmids pBAD18 (empty vector; lanes 3 and 4), pAraV18 (LcrV; lanes 5 to 6), pMH13 (LcrV290; lanes 7 to 8), and pMH122 (LcrV311; lanes 9 to 10) were analyzed. (B) Cell-free fractions were separated by SDS-PAGE and silver stained to visualize secreted Yops. Cell-free fractions of Y. pestis KIM8-3002 containing pBAD18 (vector; lanes 1 and 2) and Y. pestis KIM8-3002.N1 (ΔlcrV5) containing plasmids pBAD18 (empty vector; lanes 3 and 4), pAraV18 (LcrV; lanes 5 to 6), pMH13 (LcrV290; lanes 7 to 8), and pMH122 (LcrV311; lanes 9 to 10) were analyzed.

In contrast, introduction of LcrV311 into the ΔlcrV5 strain restored Yop secretion in the absence of Ca²⁺; however, the amounts of Yops secreted into the medium were reduced compared to the levels secreted in the presence of full-length LcrV (Fig. 1B, lane 10), suggesting that residues past 311 may directly or indirectly influence interaction with LcrG. Intracellular levels of YopE were elevated in the culture grown without Ca²⁺, indicating the ability of LcrV311 to induce the low Ca²⁺ response (Fig. 1A, lane 10). Neither LcrV290 nor LcrV311 was found to be secreted, as determined by silver staining (Fig. 1B) or immunoblotting (data not shown), consistent with previous observations that C-terminal additions and deletions in LcrV abolish its ability to be secreted (4, 6, 24). The inability of LcrV311 to fully activate Yop secretion could be the result of the inability of LcrV311 to be secreted itself or the result of a requirement for the last 15 amino acids of LcrV for the full activation of Yop secretion despite their indispensability for binding LcrG. Taken together, these results with LcrV290 and LcrV311 demonstrate that the region between amino acids 290 and 311 of LcrV is involved in LcrG-LcrV interaction and the activation of Yop secretion in the absence of Ca²⁺.

Radical site-directed mutagenesis of hydrophobic amino acids in α-12 of LcrV disrupts the ability to activate Yop secretion.

The crystal structure of LcrV shows that the region between amino acids 290 and 311 lies in a long C-terminal helix, helix 12 (α-12) (Fig. 2A) (5). α-12 is engaged by another internal helix, α-7, in an intramolecular coiled coil (Fig. 2A) (5). Interestingly, residues in α-7 of LcrV are involved in the LcrG-LcrV interaction (12). Sequence alignment of LcrV with PcrV from Pseudomonas aeruginosa revealed an amino acid identity of 39% (57% similarity) along the full length of both proteins (data not shown). The highest level of amino acid conservation is found at the C terminus of both proteins. The C-terminal region between residues 290 and 311 of LcrV is 77% identical (94% similar) to PcrV (Fig. 2B). Previous work demonstrated that the most highly conserved regions between LcrV and PcrV, which correspond to α-7 and α-12, are critical for LcrV-mediated secretion activation (24). Although interaction with LcrG was not analyzed in those studies, the results obtained are consistent with a lack of LcrG interaction (24). Analysis of the amino acid composition of residues 290 to 311 revealed the arrangement of hydrophobic residues in a heptad repeat characteristic of coiled-coil interactions (Fig. 2C). To gain insight into the importance of the hydrophobic amino acids in α-12 for the secretion activation role of LcrV, radical site-directed mutagenesis was used (9, 14). Hydrophobic residues between residues 290 and 311 were mutated to arginine using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) following the manufacturer's instructions. The resulting radical LcrV mutants were introduced into the ΔlcrV5 strain to analyze their expression and the ability of the radical LcrV mutants to activate Yop secretion. The radical LcrV mutants and their Yop secretion phenotypes are summarized in Table 2, and a representative of the secretion profile of each group is shown in Fig. 3B. All of the single point mutants in LcrV were stably expressed at levels similar to that of wild-type LcrV, and all the single LcrV point mutants retained wild-type secretion control (Fig. 3B, Table 2, and data not shown). In the absence of Ca²⁺, all of the single point mutants generated in LcrV showed wild-type Yop secretion except the LcrV F308R strain, which showed slightly reduced levels of Yop secretion (Table 2; Fig. 3B, lanes 9 and 10).

FIG. 2.

Analysis of the C-terminal domain of LcrV. (A) The region between amino acids 290 and 311 of LcrV lies in a C-terminal helix that is involved in an intramolecular coiled-coil interaction with α-7. The positions of α-12 (dark and light gray) and α-7 (dark gray) are indicated. The region between amino acids 291 and 311 is shown in light gray. (B) Pairwise alignments of the C-terminal region between residues 290 and 311 of LcrV with the C terminus of PcrV reveals 77% identity and 94% amino acid similarity (30). (C) Heptad repeat arrangement of the hydrophobic amino acids in the region between residues 290 and 311. a and d, the first and fourth positions of the heptad repeat, which are characterized by the presence of hydrophobic amino acids.

TABLE 2.

Effects of radical LcrV mutations on the secretion of Yops^a

LcrV	Secretion in medium:
	Without Ca2+	With Ca2+
None	Blocked	Blocked
Wild type	Activated	Blocked
L291R	Activated	Blocked
I294R	Activated	Blocked
F298R	Activated	Blocked
A301R	Activated	Blocked
I302R	Activated	Blocked
A304R	Activated	Blocked
L305R	Activated	Blocked
F308Rb	Reduced	Blocked
I309R	Activated	Blocked
L291R I294R	Activated	Blocked
L291R F298R	Activated	Blocked
L291R A301R	Activated	Blocked
L291R I302R	Activated	Blocked
I294R F298R	Activated	Blocked
A301R L305R	Activated	Blocked
I302R L305R	Activated	Blocked
L291R F308Rc	Blocked	Blocked
I294R F308Rb	Reduced	Blocked
A301R F308Rb	Reduced	Blocked
A301R F308Rb	Reduced	Blocked
I302R F308Rb	Reduced	Blocked
L305R F308Rb	Reduced	Blocked

^aSecretion phenotype of Yops produced by the ΔlcrV5 strain expressing various LcrV radical mutants. ΔlcrV5 strains expressing various LcrV constructs were grown in TMH for 4 h at 37°C in the presence or absence, of Ca²⁺ and the cell-free fractions were analyzed by SDS-PAGE followed by silver staining.

^bMutation resulting in an LcrV mutant that showed reduced Yop secretion.

^cMutation resulting in an LcrV mutant that showed no Yop secretion.

FIG. 3.

Effects of the radical LcrV mutants on the secretion of Yops. Cultures were grown for 4 h at 37°C in TMH with or without Ca²⁺, and samples were fractionated into whole-cell and cell-free culture supernatants and analyzed by immunoblotting for cellular proteins and by silver staining for secreted Yops. (A) Western blot analysis of cellular proteins. Samples of the cellular fraction were analyzed by immunoblotting against anti-LcrV and anti-YopE. (B) Cell-free fractions were separated by SDS-PAGE and silver stained to visualize secreted Yops.

Since no single point mutation had a dramatic effect on LcrV's role in activating Yop secretion, double mutants of LcrV were created by introducing a second mutation. A list of the double point mutants in LcrV and their Yop secretion phenotypes is presented in Table 2. All of the double point mutants created were stably expressed and showed no Yop secretion in the presence of Ca²⁺ (Fig. 3, Table 2, and data not shown). The secretion phenotypes of the double point mutants fell into three classes. The first class of double mutants showed wild-type activation and regulation of Yop secretion. The positions of these mutations and their secretion phenotypes are summarized in Table 2. The secretion profile of a representative from this category, LcrV L291R/A301R, is shown in Fig. 3B (lanes 11 to 12). The second class of mutants consists of LcrV mutants that showed reduced Yop secretion in the absence of Ca²⁺. All mutants in the second class had an F308 mutation in addition to a second hydrophobic amino acid changed to arginine. This class contained five mutants summarized in Table 2, and the secretion profiles of two representatives (LcrV I294R F308R and LcrV L305R F308R) are shown in Fig. 3B (lanes 15 to 18). The third class of mutants included only one LcrV construct that showed no Yop secretion in the absence of Ca²⁺ (Fig. 3B, lanes 13 to 14). This LcrV construct had both L291 and F308 replaced with arginines (LcrV L291R F308R). To study the effect of these changes on the interaction between LcrV and LcrG, LcrV L291R F308R was introduced into an LcrV/GAL4-BD chimera and analyzed for interaction with an LcrG/GAL4-AD chimera in a yeast two-hybrid system. β-Gal activity of LcrV L291R 308R was dramatically reduced compared to that of the wild type, demonstrating that the combined changes of L291R and F308R in LcrV significantly reduced the LcrG-LcrV interaction (Table 1). Based on the data presented in this study, implicating α-12 involvement in LcrG interaction, and the results of Lawton et al. (12), involving α-7 in LcrG interaction, we propose that the intramolecular coiled coil formed by α-7 and α-12 of LcrV is required for the interaction of LcrV with LcrG. Unfortunately, the crystal structure of LcrG or an LcrG-LcrV complex is not available; thus, exactly how α-7 and α-12 of LcrV mediate LcrV's interaction with LcrG remains unclear, although coiled-coil interactions are believed to be a common mechanism of T3SS protein-protein interaction (20).

The interaction between LcrV and LcrG is hydrophobic.

Previous studies showed that the LcrG-LcrV interaction is mediated by hydrophobic forces (12, 13). This study identified the hydrophobic amino acids L291 and F308 of LcrV as crucial residues in mediating the LcrG-LcrV interaction. To confirm that the LcrG-LcrV interaction is hydrophobic, L291 and F308 were replaced with the moderately hydrophobic amino acid alanine. Introduction of LcrV L291A F308A into the ΔlcrV5 strain restored Yop regulation and secretion similar to those of wild-type LcrV (Fig. 3B, lanes 19 and 20), demonstrating that the hydrophobic properties of L291 and F308 are involved in LcrV's ability to activate Yop secretion. LcrV L291A F308A was introduced into an LcrV/GAL4-BD chimera and analyzed for interaction with an LcrG/GAL4-AD chimera in the yeast two-hybrid system to test LcrG-LcrV interaction. β-Gal activity of the LcrV L291A F308A was similar to that of wild-type LcrV, indicating that these alanine mutations did not affect the LcrG-LcrV interaction (Table 1). These analyses with LcrV L291A F308A confirm that the LcrG-LcrV interaction is mediated by hydrophobic amino acids, consistent with previous reports (12, 14).

Mutation of L291 to L291R in LcrV in combination with F308R completely disrupted LcrV's ability to activate Yop secretion. In addition, LcrV L291R F308R had reduced interaction with LcrG in the yeast two-hybrid system demonstrating that LcrV L291R F308R is defective in LcrG interaction. In LcrV, L291 and F308 appear to interact with L164 and A120, respectively (Fig. 4). Changing L291 and F308 into the bulky arginine might push α-12 away from α-7, which would destabilize the intramolecular coiled coil in LcrV and disrupt the ability of LcrV to interact with LcrG. The disruption of the LcrG-LcrV interaction by changing the hydrophobic residues L291 and F308 to arginine suggests that the hydrophobic nature of L291 and F308 is crucial for LcrV's function in secretion activation via LcrG interaction. This contention that the hydrophobic nature of L291 and F308 is involved in LcrG interaction is supported by mutation of L291 and F308 to the moderately hydrophobic amino acid alanine. This mutation did not affect LcrV's function regarding secretion activation or binding to LcrG, demonstrating that the hydrophobic nature of L291 and F308 is important in the LcrG-LcrV interaction.

FIG. 4.

Positions of amino acid L291 and F308 on the crystal structure of LcrV. (A) Positions of residues A120, L164, L291, and F308. L291 is in α-12, and L291 appears to interact with L164 on α-7. F308 on α-12 appears to interact with A120 on α-5. (B) Multiple-sequence alignment of the LcrV homologs in Pseudomonas aeruginosa, Photorhabdus luminescens, and Aeromonas hydrophila. Sequence alignment was performed using T-Coffee, and amino acid sequences of the regions homologous to α-7 and α-12 of LcrV are shown (19). Positions of A120, L164, L291, and F308 are indicated by asterisks.

The intramolecular coiled coil is not required for LcrV dimerization.

LcrV has been reported to form dimers and higher-order oligomers (12). In this study and in other unpublished observations, sodium dodecyl sulfate (SDS)-stable LcrV dimers were detected using immunoblotting with anti-LcrV (Fig. 5). To study the role of the intramolecular coiled coil formed by α-7 and α-12 in the formation of LcrV dimers, an LcrVΔα-7 mutant was created. Plasmid pMH61, encoding LcrVΔα-7, was constructed by cloning two PCR fragments encoding residues 1 to 152 and residues 165 to 326 of LcrV into pBAD18. This mutant, LcrVΔα-7, is deleted for the residues comprising α-7 (residues 152 to 165), and was tested along with full-length LcrV (pAraV18), LcrV290 (pMH13), and LcrV311 (pMH122) for the ability to form LcrV dimers. The truncated protein LcrV290 had all of α-12 removed, while LcrV311 was missing a small part of α-12 (Fig. 2A). These LcrV deletion constructs were overexpressed in E. coli to rule out other T3S proteins that might bind to LcrV. Samples from E. coli expressing the LcrV deletions constructs were separated by SDS-polyacrylamide gel electrophoresis (PAGE) and immunoblotted with anti-LcrV. LcrVΔα-7, LcrV290, and LcrV311 were all able to form dimers (Fig. 5). These results suggest that neither α-7 nor α-12 is required for LcrV dimerization.

FIG. 5.

The intramolecular coiled-coil motif in LcrV is not required for LcrV's dimerization. E. coli (Novablue) cells carrying pBAD18 (empty vector; lane 1), pAraV18 (full-length LcrV; lane 2), pMH61 (LcrVΔα-7; lane 3), pMH13 (LcrVΔα-12, lane 4), and pMH122 (LcrV311, lane 5) were grown in LB plus arabinose for 3 h. Cells were harvested by centrifugation, and cellular fractions were precipitated with trichloroacetic acid overnight at 4°C. Samples were separated by SDS-PAGE in a 15% polyacrylamide gel and analyzed by immunoblotting with anti-LcrV. Positions of the LcrV dimers are indicated by arrowheads.