|Home | About | Journals | Submit | Contact Us | Français|
Type III secretion systems (T3SSs) are widely distributed virulence determinants of Gram-negative bacteria. They translocate bacterial proteins into host cells to manipulate them during infection. The Shigella T3SS consists of a cytoplasmic bulb, a transmembrane region, and a hollow needle protruding from the bacterial surface. The distal tip of mature, quiescent needles is composed of IpaD, which is topped by IpaB. Physical contact with host cells initiates secretion and leads to assembly of a pore, formed by IpaB and IpaC, in the host cell membrane, through which other virulence effector proteins may be translocated. IpaB is required for regulation of secretion and may be the host cell sensor. However, its mode of needle association is unknown. Here, we show that deletion of 3 or 9 residues at the C terminus of IpaB leads to fast constitutive secretion of late effectors, as observed in a ΔipaB strain. Like the ΔipaB mutant, mutants with C-terminal mutations also display hyperadhesion. However, unlike the ΔipaB mutant, they are still invasive and able to lyse the internalization vacuole with nearly wild-type efficiency. Finally, the mutant proteins show decreased association with needles and increased recruitment of IpaC. Taken together, these data support the notion that the state of the tip complex regulates secretion. We propose a model where the quiescent needle tip has an “off” conformation that turns “on” upon host cell contact. Our mutants may adopt a partially “on” conformation that activates secretion and is capable of recruiting some IpaC to insert pores into host cell membranes and allow invasion.
The Gram-negative bacterium Shigella flexneri is the causative agent of human bacillary dysentery (or shigellosis), a disease characterized by invasion of, massive inflammation in, and destruction of the colonic mucosa (19). After ingestion, S. flexneri invades the gut lining (10, 53, 54), where it is phagocytosed by resident macrophages. S. flexneri kills macrophages by inducing apoptosis (21, 22, 59) and then forces its uptake into epithelial cells (1, 12). It becomes engulfed in an entry vacuole, which it also lyses to multiply within the cell cytoplasm (52). S. flexneri spreads to neighboring cells by forming a “comet tail,” nucleating host cell actin filaments using an asymmetrically distributed outer membrane protein (5, 29). Upon reaching the next epithelial cell, Shigella lyses its two-membrane vacuole, and the spread of infection continues (30, 46, 49, 50, 53, 54, 60). The S. flexneri genes required for invasion are clustered on a 31-kb fragment of a large virulence plasmid. This region encodes a type III secretion system (T3SS) (mxi and spa operons) and several effector proteins (ipa and ipg operons ).
Type III secretion systems (T3SSs) are important virulence factors in over 25 species of Gram-negative bacteria pathogenic for plants, animals, and humans. They are used to inject proteins into the plasma membrane or cytoplasm of host cells to manipulate them during infection. The overall T3SS architecture is conserved among different species (14); an extracellular structure, usually a straight, hollow “needle” constructed from only one type of protein, is connected to a transmembrane region consisting of two pairs of rings, a periplasmic channel, and a cytoplasmic “bulb.” The needle's distal end is terminated by a distinct structure, the tip complex (8, 38, 39). Physical contact with host cells initiates secretion of translocator proteins and leads to the formation of a pore in the plasma membrane. The direct connection of the pore to the needle may allow one-step injection of effector proteins into host cells (14, 38, 39).
For pore formation three proteins, known as the translocators, are required. In S. flexneri, these proteins are IpaBCD (invasion plasmid antigens [6, 11, 36]). IpaD is a hydrophilic protein that is constitutively present at the needle tip (17, 56). It mediates binding of IpaB to needles (23, 56), possibly after detection of bile salts in the gut lumen (40). IpaD is also required for IpaC binding to needles (56) and for insertion of IpaB and IpaC into the host cell membrane upon host cell contact (6). IpaB is a largely hydrophobic protein that is also present atop mature needles, although the amounts of this protein are smaller than those of IpaD (56). It later forms part of the pore in the host cell membrane together with IpaC (6). IpaC is also a hydrophobic protein, which is associated substoichiometrically with quiescent wild-type needles compared to IpaB and IpaD (56). Although IpaC is required for pore formation, it is not essential for membrane insertion of IpaB (6). IpaB may therefore insert first, but without IpaC no functional pore is formed (6).
We proposed that the finalized needle tip is composed of four molecules of IpaD and one molecule of IpaB (8, 23, 56). Determination of the atomic structure of IpaD allowed construction of an IpaD pentamer based on the crystal contacts in one of the crystal forms (23). This model proposes a helical arrangement with parameters similar to those of the MxiH needle (13). Due to the helical rise in the needle and pentamer, the interface between the fifth and first IpaD molecules is different from all other interfaces in the pentamer, and thus the final position may be filled not by IpaD but by IpaB (23).
The predicted IpaB structure supports this model. It may contain two long coiled-coil-forming regions, the first from residue 130 to residue 170 and the second at the extreme C terminus (residues 530 to 580). As the length and position of these coiled-coil-forming regions are similar to those of IpaD, the overall folds of these proteins might be similar, although IpaB would have a much larger C-terminal globular domain than IpaD (23). IpaD binds to the needle protein MxiH via its last 5 to 10 amino acids (17, 56); therefore, IpaB might also do so. These models for the needle tip and IpaB position the predicted C-terminal globular domain of IpaB at the very tip of the needle, making it the point of contact with the host cell. IpaB's C-terminal domain contains two predicted transmembrane helices (amino acids 313 to 346 and 400 to 423) within a larger hydrophobic domain (amino acids 310 to 430 [4, 18]). The hydrophobic domain also contains the IpaC binding region (amino acids 367 to 458), which could enable interaction of IpaC and IpaB atop needles and in the host cell membrane (41).
Therefore, IpaB and IpaD might “plug” the needle until it is activated (8, 34). The quiescent needle tip would be mostly in a closed conformation, similar to that of the modeled IpaD pentamer, which has no hole through its center (23). Activation of the tip complex by contact with the host cell would lead to membrane insertion of IpaB, which could trigger or stabilize a conformational change leading to opening of the pentamer, activation of secretion, subsequent recruitment of IpaC, and assembly of the pore (16, 56). An open conformation of the tip complex has not been seen yet in Shigella. However, the Yersinia homologue of IpaD, LcrV, may form an open structure at the needle tip (8, 15, 39).
The host cell contact signal must be transduced to the cytoplasmic side of the T3SS to trigger its activity. The needle probably plays an active role in transmitting the contact signal from the tip to the base of the T3SS. Indeed, mutagenizing the needle is sufficient to affect the ability of the T3SS to sense host cells (24, 37, 55) and regulate secretion (24, 55). The signal might be transmitted via structural changes in the needle subunits (7, 8).
The Shigella T3SS has three different functional states (56). Slow, low-level Ipa protein secretion prior to host cell contact is termed “leakage” (28). “Induction” describes the burst of Ipa protein secretion when a host cell is sensed (34) or upon addition of Congo red (CR), a small amphipatic dye molecule which acts as an artificial inducer of secretion (3). This event occurs within minutes. “Constitutive secretion” of Ipa proteins represents unregulated leakage and hence overall higher-level secretion than leakage and involves not only the Ipa proteins but also the “late effectors” that are involved in the intracellular stage of the bacterium's infectious cycle (47). It occurs in some needle mutants, where it is detectable only after hours (24). In ΔipaB and ΔipaD strains, constitutive secretion is much faster and is detectable in minutes. Therefore, this secretion is called “fast constitutive secretion” (56).
As deletion of ipaD or ipaB leads to fast constitutive secretion and unresponsiveness to Congo red, we proposed that IpaB is the host cell sensor and that the conformational change required for membrane insertion of IpaD-presented IpaB may be the signaling event triggering T3SS “activation” (56). This has been difficult to test because intrabacterial levels of IpaB are also used to regulate transcription of late effector genes, the products of which are usually secreted after initial activation of the T3SS. This initial activation leads to recruitment of IpaC to the needle tip (16, 56) and release of MxiC, which in turn leads to OspD1 secretion, both through the T3SS (9, 44). OspD1 acts as a repressor of the transcriptional activator MxiE, which regulates expression of late effector genes (31, 32, 44). MxiE also requires a coactivator, IpgC, which is the intrabacterial chaperone of IpaB and IpaC (35). Thus, once secretion is activated, IpgC is released from its substrates (31), MxiC and then OspD1 are secreted, and the IpgC-MxiE complex activates late effector genes.
To test both our model for IpaB assembly in the needle tip complex and the role of IpaB in regulating secretion at this location, we prepared mutants with short C-terminal IpaB truncations and examined their functional properties. We found that even small deletions at the IpaB C terminus affect the regulation of secretion, as well as bacterial adherence to and invasion of epithelial cells. These effects may be due to a conformational change in the mutants with IpaB C-terminal deletions at the needle tips, which leads to their weaker binding and partial IpaC recruitment at this location.
Table Table11 lists the strains used in this study. Bacteria were phenotypically selected on Congo red agar plates (33) and grown in Trypticase soy broth (Becton Dickinson) with the antibiotics indicated below at the following final concentrations: ampicillin, 100 μg/ml; kanamycin, 50 μg/ml; and trimethoprim, 5 μg/ml.
Full-length ipaB was amplified by PCR from pWR100 using primers ipa-F and ipa-R and cloned into pUC19 using the HindIII and PstI sites in the polylinkers, yielding pDR1. To generate the C-terminal deletions, the same strategy was used, but with different primers (Table (Table2);2); a combination of ipa-F and ipaB_min3Cterm_del_R1 was used to clone the gene for the 3-amino-acid deletion as pDR2, and ipa-F and ipaB_min9Cterm_del_R1 were used to clone the 9-amino-acid deletion as pDR4.
Plasmid pRK2mxiH (24) is a pKK223-3-based mxiH overexpression plasmid with both Ampr and Tpr genes. As the ipaB mutant plasmids also contain Ampr genes, we inactivated the β-lactamase gene in pRK2mxiH by mutating E26 into a stop codon by QuikChange mutagenesis (Stratagene) with primers pRK2_AmpS_R2 and pRK2_AmpS_F2. The correct plasmid was selected using ampicillin sensitivity, and the mxiH gene was sequenced. The resulting plasmid was designated pDR5.
S. flexneri strains were grown overnight in Trypticase soy broth with aeration. Bacteria were collected by centrifugation at 2,500 × g for 10 min at 4°C. The supernatants (“stationary phase”) were denatured in Laemmli sample buffer, subjected to SDS gel electrophoresis, and visualized by silver staining.
Bacteria collected during exponential growth (optical density at 600 nm [OD600] of ~1) were resuspended at an OD600 of 5 in phosphate-buffered saline (PBS), and Congo red was added to a final concentration of 200 μg/ml (to measure CR induction) or not added (for fast constitutive secretion). After incubation at 37°C for 15 min, the samples were centrifuged at 2,500 × g for 10 min at 4°C, and the supernatants were denatured, separated by SDS-PAGE, and silver stained as described above. The mouse monoclonal antibodies used for Ipa protein detection have been described previously and were generous gifts from Armelle Phalipon (anti-IpaB antibody, designated H16, and anti-IpaC antibodies, designated K24 and J22) and Kirsten Niebuhr (anti-IpgD antibody). The rabbit polyclonal sera against VirA and IpaD have also been described previously and were kind gifts from Brigitte Demers and Claude Parsot, respectively.
S. flexneri invasion of HeLa cells was assessed with a gentamicin protection assay. HeLa cells cultured in Dulbecco modified Eagle medium (DMEM) with a high concentration of glucose (Sigma) were seeded into 24-well plates at a density of 3 × 105 cells/well and grown overnight at 37°C in a humidified incubator under 5% CO2 (tissue culture incubator). Bacteria from overnight precultures were diluted and grown to exponential phase in Trypticase soy broth (OD600, ~1). They were collected by centrifugation at 4,500 rpm for 10 min at 10°C and resuspended in PBS to a concentration of 4.5 × 109 bacteria/ml. HeLa cells were washed twice with PBS, and 1 ml DMEM (Sigma) with 20 mM HEPES (pH 7.4) (Gibco) was added. Ten microliters of the prepared bacteria was transferred to the HeLa cells in a 24-well plate (multiplicity of infection [MOI] of 100). The plates were centrifuged for 10 min at 900 × g and room temperature and incubated for 30 min at 37°C. Free bacteria were removed by aspiration, and the cells were washed four times with PBS. Then 1 ml DMEM with 100 μg/ml gentamicin (Gibco) was added. After incubation for 1.75 h at 37°C in a tissue culture incubator, the gentamicin solution was aspirated, 1 ml of 0.1% (vol/vol) Triton X-100 (Sigma) in PBS was added, and the preparation was incubated for 10 min at room temperature to detach and lyse the HeLa cells. Serial dilutions of the cell lysate were plated on Luria-Bertani agar plates, and the colonies were counted the next day.
For the adhesion assay, the invasion assay was modified as follows. Infected HeLa cells were centrifuged for 10 min at 900 × g and room temperature and incubated for 30 min at 25°C (at which the T3SS is inactive ) instead of 37°C. The coverslips were then washed three times in PBS and fixed in 4% paraformaldehyde in PBS for 15 min. After cells were washed with H2O, they were stained with Giemsa stain (Sigma) for 5 min, washed twice with H2O, air dried, and mounted with Mowiol. Micrographs used to count cell-associated bacteria were taken with a Leica DMIRB inverted microscope at a magnification of ×100.
To assess vacuolar escape, the gentamicin protection assay was modified as follows. After incubation for 30 min at 37°C, free bacteria were removed by aspiration, and the cells were washed four times with PBS. Cells were treated with either 100 μg/ml gentamicin or 100 μg/ml gentamicin and 50 μg/ml chloroquine (Sigma), a weak base that kills bacteria localized in vacuoles (58). Then the cells were incubated for 1 h at 37°C. The gentamicin solution was then aspirated, the HeLa cells were lysed, and the bacteria were counted as described above. To confirm the effectiveness of the chloroquine treatment, RAW 264 cells were infected with both wild-type bacteria and mxiD mutant bacteria lacking a functional T3SS. The mxiD mutant bacteria, in contrast to the wild-type bacteria, could not survive the chloroquine treatment.
The contact hemolysis assay was performed as described by Blocker et al. (6), with slight modifications. Bacteria from overnight precultures were diluted and grown to exponential phase in Trypticase soy broth (OD600 of ~1). They were collected by centrifugation at 2,200 × g for 10 min at 4°C, washed in Tris-saline (150 mM NaCl, 30 mM Tris; pH 7.4), and resuspended at a concentration of 1010 bacteria/ml. Sheep red blood cells (RBCs) (TCS Biosciences) were washed three times in Tris-saline with centrifugation at 2,000 × g for 7 min at 4°C and resuspended at a concentration of 5 × 108 cells/ml. Then 100 μl of bacteria was mixed with 100 μl of RBCs in round-bottom 96-well plates, centrifuged at 1,500 × g for 10 min at 10°C, and incubated at 37°C for 1 h. The cells were resuspended, and samples were centrifuged again. Then 100 μl of supernatant was transferred to a fresh plate, where its optical density at 405 nm was measured. Instead of bacteria, 0.5% Triton in Tris-saline was used to determine total hemolysis, and Tris-saline alone was used to determine the baseline.
Isolation of RBC membranes was performed as described previously (1).
HeLa cells cultured in DMEM with a high concentration of glucose (Sigma) were seeded into six-well plates at a density of 4.2 × 105 cells/well and grown overnight at 37°C in a humidified incubator under 5% CO2. Bacteria from overnight precultures were diluted and grown to exponential phase in Trypticase soy broth (OD600 of ~1). They were collected by centrifugation at 4,500 rpm for 10 min at 10°C and resuspended in DMEM-HEPES (DMEM [Sigma] with 20 mM HEPES [pH 7.4; Gibco]) to obtain a concentration of 1.4 × 109 bacteria/ml. HeLa cells were washed twice with PBS, and 3 ml DMEM-HEPES was added. Then 50 μl of the prepared bacteria was transferred to the HeLa cells in a six-well plate (MOI of 100). For every strain 12 wells were infected. The plates were centrifuged for 10 min at 900 × g and room temperature and incubated for 30 min at 37°C. Free bacteria were removed by aspiration, and the cells were washed three times with cold PBS. They were then gently scraped off the culture wells using 200 μl PBS per well, collected by centrifugation at 1,000 rpm for 5 min at 4°C, resuspended in 500 μl homogenization buffer (3 mM imidazole [pH 7.4], 250 mM sucrose, 0.5 mM EDTA) with protease inhibitors (Roche Complete mini EDTA free), centrifuged again at 2,500 rpm for 10 min at 4°C, and mechanically disrupted in 150 μl homogenization buffer with protease inhibitors by vigorous passage (ca. 12 to 30 times) through a 22-gauge needle using a 1-ml syringe. Lysis of the cell plasma membrane and release of intact but swollen nuclei were monitored by phase-contrast microscopy until approximately 70% of the nuclei were free but only a few nuclei had begun aggregating. Centrifugation at 3,000 rpm for 15 min at 4°C was used to pellet the bacteria, unbroken HeLa cells, and nuclei. The postnuclear supernatant was collected, diluted 1:2 in 3 mM imidazole (pH 7.4) with 63% (wt/wt) sucrose containing protease inhibitors, and loaded at the bottom of a 1-ml TLS-55 Beckman centrifuge tube. This mixture was topped with 200 μl of 3 mM imidazole (pH 7.4) with 40% (wt/wt) sucrose containing protease inhibitors and then overlaid with 200 μl homogenization buffer containing protease inhibitors. The floatation gradients were ultracentrifuged for 16 h at 15,000 × g and 4°C. The top 350 μl was collected and diluted by adding 600 μl cold PBS with protease inhibitors, and the membranes were pelleted by centrifugation at 166,000 × g for 30 min at 4°C in a Beckman benchtop ultracentrifuge. The supernatant was removed, and the pellet was collected in 10 μl PBS with protease inhibitors and processed for Western blotting.
Needles were purified from strains containing the pDR5 plasmid as described previously (56). The pDR5 plasmid never led to overexpression of needles that was as strong as the overexpression obtained with pRK2mxiH. As this occurred especially in the strains with two exogenous plasmids (i.e., the complemented strain with pDR1 and pDR5, the ipaBΔ3 strain with pDR2 and pDR5, and the ipaBΔ9 strain with pDR3 and pDR5), it might have been due to partial incompatibility between pUC19 and pRK2. Therefore, for the wild-type and ΔipaB mutant strains 450- or 600-ml cultures were grown. However, 1.8-liter cultures were grown for the complemented, ipaBΔ3, and ipaBΔ9 strains because of the low number of long needles observed with the electron microscope.
We generated plasmid constructs expressing 3- and 9-amino-acid C-terminal truncations of IpaB, as described in Materials and Methods, and used them to complement a ΔipaB strain. In the resulting strains, which were designated the ipaBΔ3 and ipaBΔ9 strains, IpaB is produced (data not shown) and secreted at approximately the same levels as it is in wild-type or the ΔipaB mutant carrying a construct encoding a wild-type version of IpaB (Fig. (Fig.1D,1D, top panel).
While wild-type bacteria secrete only low levels of Ipa proteins in stationary phase (“leakage”), the ΔipaB strain is a fast constitutive secreter (45, 56). The ipaBΔ3 and ipaBΔ9 strains also showed fast constitutive secretion (Fig. 1A and B). As observed for the ΔipaB strain, this involves not only a slight increase in secretion of an early effector, such as IpgD, but also massive stimulation of release of a late effector, such as VirA (Fig. 1B and D). Both proteins can be found in supernatants of ΔipaB, ipaBΔ3, and ipaBΔ9 bacteria growing in exponential phase (Fig. (Fig.1D,1D, lower panels) but not in the supernatants of wild-type bacteria or ΔipaB complemented bacteria, although they are stored at the same levels in all strains (data not shown) (56).
Another difference between wild-type and ΔipaB strains is their sensitivity to Congo red. Secretion can be induced in wild-type bacteria by addition of Congo red, whereas the ΔipaB strain is unable to respond to Congo red addition. We found that both the ipaBΔ3 and ipaBΔ9 strains are also largely uninducible by Congo red (Fig. (Fig.1C).1C). In conclusion, at present the ipaBΔ3 and ipaBΔ9 mutants are phenotypically indistinguishable from the ΔipaB strain in all type III secretion assays performed.
All previously described S. flexneri mutants displaying constitutive type III secretion and insensitivity to Congo red are also unable to invade HeLa cells. To analyze the invasiveness of the ipaBΔ3 and ipaBΔ9 mutant strains, gentamicin protection assays were performed, as described in Materials and Methods. In this assay the ΔipaB mutant is completely noninvasive (36). However, the ipaBΔ3 mutant seemed to be almost 3-fold more invasive than wild-type Shigella, while the ipaBΔ9 mutant was approximately 50% less invasive than the wild type but not significantly different from the ΔipaB complemented strain (Fig. (Fig.2A2A).
The ΔipaB strain was previously reported to display increased adhesion to eukaryotic cells through an unknown mechanism (20). Therefore, adhesion of the ipaBΔ3 and ipaBΔ9 mutants to HeLa cells was also checked by incubating the bacteria with host cells at 25°C, a temperature at which the T3SS is inactive (6) and hence cannot mediate invasion. While the ΔipaB mutant is approximately 4-fold more adhesive than the wild type, we found that the ipaBΔ3 and ipaBΔ9 mutants were ~5- and 3-fold more adhesive than the ΔipaB complemented strain, respectively (Fig. (Fig.2B).2B). These values did not change when we adjusted the MOIs to reflect the different adhesion capacities of the strains (not shown).
Taking the results of both sets of experiments into consideration and expressing the level of invasion as a percentage based on the number of intracellular bacteria relative to the number of adherent bacteria, we found that the ipaBΔ3 and ipaBΔ9 mutants had approximately 50% and 23% of the invasion capacity of the complemented strain, respectively. We concluded that the ipaBΔ3 and ipaBΔ9 mutants likely have mild and more substantial invasion defects, respectively, which are partially compensated for by their increased ability to bind to eukaryotic cells.
Insertion of translocon components into eukaryotic membranes is presumed to be a prerequisite for invasion of epithelial cells by Shigella. Therefore, we wanted to confirm that the ipaBΔ3 and ipaBΔ9 mutant strains could still form pores by inserting themselves and IpaC into membranes. A simple way to test this is a “contact hemolysis” assay (6). Sheep red blood cells were incubated with bacteria, and the release of hemoglobin was measured optically. The hemolytic activity of wild-type bacteria was defined as 100%. The ΔipaB mutant and both the ipaBΔ3 and ipaBΔ9 mutants were completely unable to lyse RBCs (Fig. (Fig.3A).3A). To test whether this was due to an inability of the mutated IpaBs to form functional pores along with IpaC or to an inability of the mutated IpaBs to insert themselves and/or allow IpaC insertion into membranes altogether, RBC membranes were prepared after contact with S. flexneri, as described previously (6). Wild-type bacteria insert IpaB and IpaC into the host cell membrane. In contrast, in the membrane fractions of the samples incubated with the ipaBΔ3 and ipaBΔ9 strains, no IpaB or IpaC was detected (Fig. (Fig.3B3B).
The results of the hemolysis assay apparently contradicted the ability of the mutants to invade HeLa cells. Thus, we tested whether the ipaBΔ3 and ipaBΔ9 strains are capable of associating IpaB and IpaC with HeLa membranes using a method equivalent to the method that we used for detection of protein association with red blood cell membranes. Eukaryotic membranes can be efficiently separated from bacteria by sucrose gradient floatation; bacterial proteins, such as the periplasmic protein DsbA (thiol:disulfide interchange protein), could not be detected in our preparations. Titrating the DsbA detection level showed that we would have detected 0.001% of the bacterial inoculum (not shown). We found that IpaB and IpaC were present in membranes isolated from HeLa cells infected with the ipaBΔ3 and ipaBΔ9 strains but not in membranes of cells infected with the ΔipaB mutant (Fig. (Fig.4).4). Taken together, the membrane association results suggest that insertion of the translocon components into RBC and insertion of the translocon components into epithelial cell membranes are not entirely equivalent.
The Shigella translocator proteins each have been implicated in lysis of host invasion vacuoles during the bacterium's infectious cycle, although whether this is a direct or indirect effect is unknown, as is the mechanism of vacuolar lysis itself. This process has been hard to dissect because the translocators are a prerequisite for host cell invasion. Hence, it was of interest to test the C-terminal IpaB deletion mutants, which retained the capacity to invade, for their abilities to lyse the epithelial cell entry vacuole. This was done using the weak base chloroquine, which kills bacteria trapped in acidifying eukaryotic organelles (58), in a modified invasion assay. As the strains had different invasion efficiencies, the invasion efficiency in the absence of chloroquine for each strain was defined as 100% and was compared to the number of bacteria recovered in the presence of gentamicin and chloroquine for the same strain to obtain a percentage, which reflected the number of bacteria localized in the cytoplasm. We found that for the wild-type strain, the ΔipaB complemented strain, and the ipaBΔ3 and ipaBΔ9 strains approximately 60 to 70% of the invading bacteria reached the host cell cytoplasm (Fig. (Fig.5).5). These values did not change when the MOIs were altered to adjust for the strains' different invasion efficiencies (not shown). This was confirmed by the presence of actin comet tails, which are formed only when bacteria access the host cytoplasm, for both deletion mutants (data not shown). Therefore, C-terminal truncations in IpaB do not affect the ability of Shigella to lyse the epithelial cell invasion vacuole.
We wanted to analyze the needle tip composition of IpaB mutant strains. In an otherwise wild-type background, IpaB and IpaC are found in purified long needles in an IpaD-dependent manner (56). Both IpaD and IpaB were also immunolocalized to distal needles tips under wild-type conditions, and IpaC was found at this location in a specific needle mutant that displayed highly increased IpaC association with needles (56). Therefore, we decided to examine the Ipa association with purified long needles in the IpaB C-terminal mutant strains. A plasmid for overexpression of the needle protein MxiH was transferred into these strains, as well as into the control strains. Needle preparations of these strains were separated by SDS-PAGE and normalized for needle protein content by silver staining (Fig. (Fig.6).6). We found that the ipaBΔ3 and ipaBΔ9 strains had the same amount of IpaD in needle preparations as the control strains but that the IpaB and IpaC levels were different in the different strains. The extreme C terminus of IpaB is required for high-affinity needle binding as there was a gradual decrease in the amount of IpaB in needle preparations of the ipaBΔ3 and ipaBΔ9 mutants. IpaC is present in wild-type needles only substoichiometrically compared to IpaB and IpaD and therefore cannot always be detected in such preparations (56). Interestingly, the ipaBΔ3 and ipaBΔ9 strains both recruited substantially higher levels of IpaC to needles (Fig. (Fig.6)6) although they stored levels of IpaC similar to the levels stored by the wild-type and ΔipaB complemented strains (not shown).
IpaB is an essential S. flexneri virulence factor that is present atop mature, quiescent T3SS needles and is required for regulated secretion, invasion, and intracellular survival. How it binds needle tips was not known previously. We showed that the extreme C terminus of IpaB is required for binding to needles. In addition, we found that deletion of the extreme IpaB C terminus leads to increased recruitment of IpaC to needles, fast constitutive secretion, and hyperadhesion. However, the ipaBΔ3 and ipaBΔ9 strains are partially invasive and are able to escape from their macropinocytic entry vacuoles with normal efficiency.
Even deletion of IpaB's last three residues (IpaBΔ3) affects its ability to associate with needles. The effect is even more pronounced after deletion of the last nine residues (IpaBΔ9). In our samples, we assumed that both IpaB and IpaC are bound to needle tips and not bound nonspecifically to the outside of needles since neither IpaB nor IpaC is needle associated in a ΔipaD strain, which also constitutively secretes IpaB and IpaC (56). Therefore, our present findings support our model of needle tip organization where IpaD and IpaB are assumed to have the same overall fold and to bind the distal end of needles with the same region, their extreme C termini. This model also proposes interactions between IpaD and IpaB. The C-terminal globular domain of IpaD is involved in the binding of IpaB to needle tips (23), but the required regions in IpaB have not been characterized yet. The latter interaction could account for residual binding of IpaBΔ3 and IpaBΔ9 to needles.
Deletion of the extreme C terminus of IpaB leads to increased recruitment of IpaC to needles. In wild-type needles IpaC is bound substoichiometrically compared to IpaB and IpaD, and this is probably why we did not detect IpaC in needles of the wild-type strain. Thus, there is substantial recruitment of IpaC to needles after deletion of the C terminus of IpaB; however, this recruitment seems to be quantitatively less than that observed for the mxiH(Q51A) and mxiH (P44A+Q51A) mutants (56). These needle mutants also have defects in regulating the T3SS; both are constitutive secreters, and the latter is also unresponsive to Congo red (24).
Under otherwise wild-type conditions, IpaB and IpaC bind independent of each other to the needle tip (56), suggesting that they can also do this in the IpaB mutant strains with C-terminal mutations. This might explain the increase in the amount of IpaC when the amount of IpaB is actually decreased in the IpaBΔ9 needles compared with the IpaBΔ3 needles. The increased IpaC binding to needle tips in our ipaB mutant strains can perhaps be explained by the increasingly altered IpaB proteins inducing an increasing transition between different functional states of the needle tip. The model proposes an “on” conformation for the needle tip when a host cell is sensed and that IpaC is recruited upon transition from an “off” state to an “on” state. The minor deletions in IpaBΔ3 and IpaBΔ9 might therefore lock these proteins in increasingly “on” conformations, while they concomitantly decrease the affinity of the proteins for needles. In vivo, a change in the conformation could decrease the affinity for the needle tip and prime IpaB for insertion into the membrane. However, we would not expect wild-type IpaB to be lost in vivo since it would have to remain attached to needles to allow formation of a translocon pore within host membranes that is continuous with the needle.
Deletion of the last three amino acids in IpaB is enough to deregulate secretion. In fact, presently no differences have been observed between the secretion patterns of the ΔipaB, ipaBΔ3, and ipaBΔ9 strains; all of these strains are fast constitutive secreters (56).
It has been proposed that activation of secretion is ultimately sensed by the presence of free IpgC, the intrabacterial chaperone for IpaB and IpaC, and by free MxiE (32), which is bound by OspD1 prior to its own secretion (44). In the ΔipaB strain, the remaining translocators and OspD1 are constitutively secreted (9); therefore, free IpgC and MxiE are always present and act together to activate expression and secretion of late effectors (26, 31).
We show here that the ipaBΔ3 and ipaBΔ9 strains are also constitutive secretors of the translocators and early effectors. Therefore, they probably secrete OspD1 as well. Furthermore, they secrete late effectors. However, the binding of IpaBΔ3 and IpaΔ9 to IpgC cannot be affected as it is required for stable, nontoxic expression of IpaB possessing its hydrophobic regions (35). Moreover, the IpgC binding site in IpaB (amino acids 51 to 72) is far from the mutated C terminus of our mutants (27, 41). Yet, Parsot et al. reported that an ospD1 mutant showed significantly increased expression of late effector genes even before induction of secretion; this indicates that some IpgC molecules are not titrated by IpaB and IpaC even under these baseline conditions (44). In addition, since the ipaBΔ3 and ipaBΔ9 strains secrete IpaC and IpaB constitutively, this could alter the dynamics of how IpgC interacts with these proteins (although the levels of stored IpaB and IpaC appear not to be altered in these mutants [data not shown]). In any case, we suppose that there is enough free IpgC in these mutants to allow a substantial increase in late effector expression and secretion compared with the late effector expression and secretion in the wild-type strain.
In conclusion, in view of our data, which allowed us to separate the effect of IpaB's association at needle tips from its liberation of IpgC for the first time, we propose that the functional state of the needle tip is the initial activation signal. Our data for the composition of the needle tip in the ipaBΔ3 and ipaBΔ9 mutant strains support the model for the transition between an “off” or closed state and an “on” or open state of the needle tip. The lack of CR inducibility in the ipaBΔ3 and ipaBΔ9 strains, which may be sensed at the needle tip (56), also supports this hypothesis. The dominance of the IpaB and IpaD proteins at needle tips in generating the signal that activates secretion may also explain why a ΔipaC mutant is not a constitutive secreter (36), unlike the ΔipaB strain, even though the absence of IpaC must lead to some increase in the amount of free IpgC.
Both the ipaBΔ3 and ipaBΔ9 strains are invasive, although their needle tip composition differs from that of wild-type bacteria, and show deregulated secretion. In fact, the invasion by the ipaBΔ3 strain is 3-fold greater than that by the complemented strain. This result might be explained by the hyperadhesion phenotype of the C-terminal deletion mutants.
Adhesion of S. flexneri to host cells is not well understood. The presence of the virulence plasmid enhances adhesion to HeLa cells (36, 42), but the protein(s) responsible remains elusive. IpaB, IpaC, and IpaD are not required for efficient adhesion, unlike the findings for the highly homologous Salmonella pathogenicity island 1 secretion system of Salmonella enterica serovar Typhimurium (25). Indeed, ΔipaD and ΔipaB mutants even exhibit hyperadhesion (20, 36). However, hyperadhesion is not correlated with at least slow constitutive secretion since one slow constitutive secreter, mxiH(D73A), is hyperadhesive, while another slow constitutive secreter, mxiH(P44A+Q51A), is not hyperadhesive (not shown).
Deletion of the last three amino acids in IpaB (IpaBΔ3) is enough to confer a hyperadhesive phenotype to bacteria similar to that of the ΔipaB strain; the ΔipaB, ipaBΔ3, and ipaBΔ9 mutants are 2-, 5-, and 3-fold fold more adhesive than the ΔipaB complemented strain, respectively. The reason for the hyperadhesiveness of the ΔipaB, ipaBΔ3, and ipaBΔ9 strains is still not known, and it is unclear how to investigate it further.
On the other hand, it is likely that the ipaBΔ3 and ipaBΔ9 strains show increasing invasion defects. Indeed, when we expressed invasion as a percentage calculated by comparing the number of intracellular bacteria with the number of adherent bacteria, we found that the values for the ipaBΔ3 and ipaBΔ9 mutants were 50% and 23% of the value for the complemented strain, respectively. The fact that there is a partial correlation between increased adhesion and invasion strengthens the notion that, as in S. enterica serovar Typhimurium, there is a specific adhesion process prior to entry of Shigella into host cells, which deserves further investigation.
So far, Shigella's ability to invade epithelial cells has been tightly correlated with its ability to perform “contact hemolysis” by inserting the translocon components into red blood cell (RBC) membranes (6). However, to our surprise, like the ΔipaB strain, the ipaBΔ3 and ipaBΔ9 strains are completely unable to form pores in RBC membranes and release hemoglobin. Although it has been noted previously that RBC lysis does not mirror invasion (48, 57), to our knowledge no other mutant Shigella strain has been reported to be totally unable to insert translocators into RBC membranes and still be invasive.
To investigate this further, we purified membranes from HeLa cells after infection with different Shigella strains, and IpaB and IpaC were present in the membranes isolated from HeLa cells infected with the ipaBΔ3 and ipaBΔ9 strains. However, our assay cannot distinguish whether these proteins are inserted into the membranes or only associated with them because it is difficult to produce enough membrane material to test whether the association of the proteins resists a high salt level or high pH (6). The small amount of membranes that could be isolated may also explain why it was hard to detect IpaC reliably in the ΔipaB complemented strain sample (Fig. (Fig.4).4). Moreover, no pore formation within, or lysis of the plasma membrane by, wild-type Shigella could be detected in HeLa cells (A. Blocker, unpublished). This means that we cannot be sure that the increase in the amount of IpaB and possibly the increase in the amount of IpaC detected in membranes from cells infected with the ipaB C-terminal deletion mutants compared to the amounts detected in membranes from cells infected with the ΔipaB complemented strain was due to increased protein release or increased protein insertion by these mutants.
In summary, the fact that insertion of the translocon components into RBC and insertion of the translocon components into epithelial cell membranes are not entirely equivalent probably reflects biological differences between the two assays. The C-terminal IpaB mutants seem to be able to recognize something in human HeLa cell membranes that they cannot recognize in sheep RBC membranes. This could be due to either a difference between the two species or a difference in the membrane organization of the different cell types. Therefore, we continue to propose that in wild-type Shigella the conformation of the needle tip and possibly IpaB changes upon activation by a host cell. This conformation might prime the translocators for membrane insertion. We also propose that our C-terminal mutant proteins constitutively adopt an activated conformation(s) and are thus able to insert into HeLa membranes, although their affinity for the needle tip is decreased.
The C-terminal deletions in IpaB do not seem to impair lysis of the vacuole, as both the ipaBΔ3 and ipaBΔ9 strains could be found in the host cell cytoplasm with efficiencies similar to that of the wild type. The mechanism of vacuolar lysis by Shigella is also unclear, but it seems to involve factors encoded in the virulence plasmid, in particular in the “entry region” (52). As a hydrophobic protein that inserts into the host plasma membrane, IpaB is a prime candidate for subsequent involvement in vacuolar lysis. Yet, our data indicate that, if this protein is involved, its location and possibly conformation at the needle tip are not crucial for maximal efficiency of the process.
In summary, we have shown the functional importance of the extreme C terminus of IpaB; it is required for efficient needle tip binding, and its absence also affects regulation of secretion and adhesion to and possibly invasion of host cells. Our results support the existence of an “on” or open conformation and an “off” or closed conformation for the needle tip, where the transition between these two states acts as a signal for activation of secretion. Collecting more evidence for or against the existence of this structural transition and its involvement in regulation of secretion is an important task for the future.
Hiroaki Nishioka and Saroj Saurya are thanked for their help with development of the HeLa cell membrane floatation assay.
A.D.R. was supported by the Max Weber Program of the Elite Network of Bavaria, Germany. I.M.A. was supported by UK Medical Research Council project grant G0401595 and Wellcome Trust project grant 082398 to A.J.B. S.J. was funded by UK Medical Research Council project grant G0400389 to Susan M. Lea. A.J.B. was partially supported by the Guy G. F. Newton Senior Research Fellowship and by grant G0401595. A.K.J.V. was funded by EC Marie Curie postdoctoral fellowship MEIF-CT-2005-023694.
Editor: J. B. Bliska
Published ahead of print on 19 January 2010.