Role of EscU auto-cleavage in promoting type III effector translocation into host cells by enteropathogenic Escherichia coli

Uncleaved forms of EscU support low levels of translocator and effector protein secretion into culture supernatants

Based on previous protein crystallography studies [26], we generated three recombinant plasmids that encode auto-cleaved or uncleaved histidine tagged forms of EscU (39 kDa) (see Materials and Methods). EscU-HIS (pJLT21), EscU(N262A)-HIS (pJLT22) and EscU(P263A)-HIS (pJLT23) were created for initial characterization studies. Unlike Shigella species where Congo Red is used to 'induce' in vitro type III secretion [33], culturing EPEC in DMEM 'induces' type III secretion [34, 35]. After culturing for 6 hours in DMEM, whole cell lysates and culture supernatants were collected from ΔescU strains harbouring pJLT21, pJLT22 and pJLT23. EscU auto-cleavage at the NPTH catalytic site is predicted to produce an 89 amino acid C-terminal product of 10.3 kDa. Immunoblotting whole cell lysates indicated that EscU-HIS was auto-cleaved due to the detection of an approximately 10 kDa species with anti-HIS antibodies (Figure 1A). A longer immunoblot exposure did not reveal any uncleaved EscU (39 kDa) suggesting complete auto-cleavage. In contrast, ΔescU whole cell lysates containing EscU(N262A) or EscU(P263A) produced a 39 kDa species detected by anti-HIS antibodies, a molecular weight consistent with uncleaved (intact) EscU.

To further characterize these strains, the respective culture supernatant fractions were evaluated. Under these growth conditions, four predominant protein species are routinely detected in secretion fractions and have been identified using protein micro-sequencing [36]. These include EspA (predicted molecular mass of 20.5 kDa, filamentous translocon protein [37], EspB (predicted molecular mass of 33 kDa, YopD orthologue), EspD (predicted molecular mass of 39.5 kDa, YopB orthologue) and EspC (predicted molecular mass 140 kDa, secreted by the type V secretion pathway). In contrast, low amounts of Tir and other type III effectors are secreted under these conditions but can be detected using immunoblotting approaches. As expected, ΔescU expressing EscU-HIS restored EspA, EspB and Tir protein secretion back to wild type EPEC levels (Figure 1B). ΔescU expressing either EscU(N262A) or EscU(P263A) had visibly lower amounts of protein species in their respective secretory profiles, however, a notable ~30kDa protein species was detected by Coomassie staining and could represent low levels of either EspB or EspD (predicted molecular masses of 33 and 39.6 kDa respectively). Immunoblotting with anti-EspA, anti-EspB and anti-Tir antibodies demonstrated reduced levels of EspA (~20%), EspB (~20%) and Tir (~70%) from ΔescU bacteria expressing either EscU(N262A) or EscU(P263A) relative to EscU (as determined by densitometric analyses). Immunoblotting the whole cell lysates of these strains demonstrated equal steady state amounts of EspA, EspB and Tir were present, ruling out the possibility of intracellular protein expression differences. Immunoblotting the same whole cell lysate samples with anti-EscC and anti-EscJ antibodies revealed equal amounts of the type III secretion apparatus ring forming proteins EscC and EscJ. This latter result indicates that EscC and EscJ protein expression levels are not altered by the cleavage status of EscU.

An ΔescNΔescU double mutant was generated to investigate if non-specific leakage from bacterial cells was occurring (perhaps due to overexpression of EscU or multi-copy effects). In the absence of EscN, the ATPase of the EPEC T3SS, type III secretion does not occur [38]. EspA, EspB and Tir were not observed in the secreted sample from the ΔescNΔescU double mutant by Coomassie staining (Figure 1C). Immunoblotting using antibodies against EspA, EspB and Tir did not detect these proteins in the ΔescNΔescU secretion fraction. Genetic complementation of ΔescNΔescU with plasmids expressing wild type EscN and EscU restored the secretion of EspA, EspB and Tir to wild type levels indicating that this double mutant strain could be rescued with multicopy plasmids expressing the appropriate proteins. Complementation of ΔescNΔescU with plasmids pJLT21, pJLT22 and pJLT23 (in the absence of pEscN) did not result in EspA, EspB and Tir secretion as assayed by Coomassie staining and immunoblotting (Figure 1C). Based on these data, the small amount of EspA, EspB and Tir in culture supernatants for ΔescU/pJLT22 and ΔescU/pJLT23 (Figure 1B and 1C) was due to EscU(N262A) or EscU(P263A) expression, and was EscN dependant. Importantly, plasmid mediated genetic complementation does not introduce leakage artefacts to the experimental system.

The 10 kDa EscU auto-cleavage product is membrane associated

The observation that uncleaved forms of EscU support very low levels of type III translocon and effector protein secretion was unexpected since EscU auto-cleavage has been suggested to provide a binding interface for protein substrate recognition at the base of the T3SS [26]. We therefore set out to evaluate the cleavage state of our EscU variants within sub-cellular fractions enriched for T3SS needle complexes.

To assess EscU auto-cleavage and to detect post-auto-cleavage products, we generated double tagged recombinant EscU forms. A hemagglutinin (HA) tag was fused to the N terminus and a FLAG tag was fused to the C-terminus of EscU. Using this strategy, wild type EscU auto-cleavage is predicted to produce a 29 kDa transmembrane polypeptide that can be recognized by anti-HA antibodies and a 10 kDa cytoplasmic polypeptide (amino acids 263-345) that can be recognized by anti-FLAG antibodies. ΔescU/pJLT24 (expressing HA-EscU-FLAG) demonstrated a wild type EPEC secretion pattern indicating that the presence of HA and FLAG tags did not inhibit EscU function (data not shown). A sub-cellular fractionation procedure to produce a membrane fraction enriched for T3SS needle complexes [39] was then used to evaluate the double tagged protein constructs in the escU null mutant. The membrane preparation derived from ΔescU/pJLT24 was probed with anti-HA antibodies and anti-FLAG antibodies which detected 29 and 10 kDa polypeptide species respectively (Figure 2). These polypeptide species are in complete agreement with the predicted sizes of EscU auto-cleavage products. No full-length EscU (39 kDa) was detected in the ΔescU/pJLT24 membrane fraction, suggesting complete auto-cleavage had occurred under these conditions. EscU(N262A) was detected exclusively at 39 kDa with anti-HA antibodies. Interestingly, EscU(P263A) appeared as a 39 kDa polypeptide along with a 29 kDa and 10 kDa polypeptides detected by anti-HA antibodies and anti-FLAG antibodies respectively. These data demonstrate that the EscU 29 and10 kDa auto-cleavage products localized to membrane fractions enriched for T3SS needle complexes and are in agreement with the crystal structure soluble domain interactions previously reported [26]. In addition, plasmid encoded EscU(P263A) is auto-cleaved in EPEC albeit at reduced levels compared to normal EscU.

The formation of functional T3SS needle complexes is believed to be a multistep process. For EPEC, T3SS needle complexes are less well characterized than those of Salmonella and Shigella species. Purified EPEC T3SS needle complex preparations often lack certain protein components that are highly conserved in all systems and hence expected to be part of a 'complete' T3SS needle complex. For example EscF, the putative needle protein has not been detected in highly purified EPEC needle preparations [20]. Antibodies to EscJ and EscN [39] were used to probe membrane fractions to assess the expression levels of these proteins. No change in the amount of cell envelope associated EscJ or EscN was observed in ΔescU bacteria expressing any of the EscU variants (Figure 2A and 2B). These data indicate that EscU auto-cleavage is not essential for EscN and EscJ localization to the cell envelope. In certain Yersinia and Shigella mutant genetic backgrounds and under defined growth conditions, the T3SS needle proteins (YscF or MxiH) are secreted in higher amounts [30, 31, 40] Since no one has detected polymerized EscF (needle protein) in purified EPEC needle complexes or secreted protein fractions, we are currently unable to assess whether EscU auto-cleavage is required for EscF polymerization or EscF protein secretion (see discussion).

EscU auto-cleavage is necessary for functional translocation of type III effector proteins into human cells

The role of EscU auto-cleavage in effector injection during EPEC infection has not been evaluated. We therefore set out to evaluate the role of EscU auto-cleavage during EPEC infection of human cells. We used C-terminal HIS tagged EscU forms for these experiments as we noted the complementation efficiency for these constructs were better than the dual HA and FLAG tagged constructs (data not shown). escU mutants expressing EscU-HIS variants were tested for their ability to translocate (inject) the effector Tir into human cells. While the EPEC bundle forming pilus (BFP) is known to mediate early and initial adherence to host cells during infection, T3SS mediated Tir translocation into host cells is required for intimate EPEC adherence (mediated by a Tir/Intimin interaction). In addition, Tir translocation results in F-actin 'pedestal' structures directly beneath adherent bacteria. As expected after a three-hour infection, it was found that the ΔescU infection had markedly fewer bacteria intimately associated to HeLa cells (compared to the wild type infection) and could not induce host cell F-actin rearrangement (Figure 3A). Infection with ΔescU/pJLT21 fully restored intimate adherence and F-actin pedestal structures to wild type levels, indicating that EscU is required for pedestal formation. The EscU(N262A) variant encoded by ΔescU/pJLT22 had similar defects in intimate adherence and pedestal formation as ΔescU (Figure 3A). In contrast, EscU(P263A) supported an apparent increase in bacterial intimate adherence and formed short actin pedestals (see inset). Notably all strains express BFP, suggesting that the intimate adherence differences are related to T3SS and EscU function. We further quantified the number of intimately adherent bacteria by microscopic counts. These analyses revealed a significant deficiency in intimate adherence for both the escU null mutant and escU expressing EscU(N262A) (Figure 3B).

To further support the Tir injection and actin pedestal observations, we employed a Tir-TEM-1 β-lactamase fusion protein (expressed in EPEC and ΔescU strains) to report on Tir translocation. This approach uses living cells loaded with a fluorescent substrate that can be cleaved by β-lactamase and has been used in EPEC/EHEC/Citrobacter to quantitatively monitor type III effector translocation [41–45]. Using this approach, a Tir-TEM-1 fusion protein was translocated by wild type EPEC but not ΔescU (Figure 3C). ΔescU/pJLT21 demonstrated translocation of Tir-TEM-1 near wild type levels while ΔescU/pJLT23 supported significantly less translocation albeit above ΔescU levels. ΔescU/pJLT22 was unable to support Tir-TEM1 translocation and appeared similar to ΔescU. These results demonstrate that EPEC strains with auto-cleaved forms of EscU supported the translocation of Tir-TEM-1 fusion proteins into infected HeLa cells whereas strains with uncleaved EscU or the absence of EscU did not.

In the absence of EscU auto-cleavage, novel Tir polypeptides are detected in culture supernatants

The HeLa cell infection experiments established a substantial role for EscU auto-cleavage in Tir and presumably other type III effector injection by EPEC. The in vitro secretion assay experiments shown in Figure 1 reveal predominant EPEC translocon protein secretion (EspABD) and very low levels of effector proteins. In contrast, EPEC sepD mutants are known to hypersecrete abundant levels of type III effector proteins under the same growth conditions, including Tir, NleA, NleH, NleG and EspZ among others [35, 39] (also see Figure 4A). We reasoned that the ΔsepD EPEC strain would be a suitable genetic background to gain some insight into the role of EscU auto-cleavage with respect to in vitro type III effector secretion. A ΔsepDΔescU double mutant was generated and grown under secretion inducing conditions followed by collection of the secreted protein fractions. The secreted protein fraction derived from ΔsepDΔescU was visibly lacking many protein species compared to that of ΔsepD (Figure 4A). Trans-complementation of ΔsepDΔescU with pJLT21 restored secretion back to that of ΔsepD with respect to protein amounts and profile. In contrast, the ΔsepDΔescU/pJLT22 did not restore a ΔsepD secretion profile. Interestingly, ΔsepDΔescU/pJLT23 secreted low amounts of a protein species with a similar apparent molecular mass as Tir (Figure 4A). Untagged cis-complemented sepD::escU(N262A) and sepD::escU(P263A) strains (expressing the respective escU allele from the chromosome) were generated by allelic exchange and were found to produce the same secretion profile as the respective plasmid complemented strains (Figure 4A). Immunoblotting with monoclonal anti-Tir antibodies revealed that Tir secretion occurred at variable levels when EscU or EscU variants were expressed although for EscU(N262A), a novel lower molecular weight polypeptide was detected with anti-Tir antibodies (Figure 4B). This novel polypeptide species was consistently absent from ΔsepDΔescU/pJLT21 or pJLT23 and the parent ΔsepD strain.

CesT membrane localization is altered in the absence of EscU auto-cleavage

In a previous report, we have demonstrated that the multicargo type III chaperone CesT mediates effector 'docking' at the inner membrane in an EscN-dependent manner [39]. CesT is also required for Tir stability in the EPEC cytoplasm [46, 47] and mediates efficient secretion of at least 9 type III effectors [39]. It has also been demonstrated that CesT contributes to effector translocation [42, 43]. The detection of a small novel Tir polypeptide in culture supernatants derived from bacteria expressing EscU(N262A) suggested that CesT-Tir interactions may be altered in the absence of auto-cleavage. We therefore set out to investigate CesT-Tir, CesT-EscU interactions in context of EscU auto-cleavage using bacteria that expressed HA-tagged EscU variants. Total cell lysates and membrane preparations were generated from the ΔescU mutant expressing either EscU, EscU(N262A) or EscU(P263A) followed by SDS-PAGE and immunoblotting analyses. Total CesT levels were unchanged in all the strains, indicating that EscU auto-cleavage does not influence CesT protein expression or stability (Figure 5). As reported previously [39], CesT was detected within the membrane fraction for wild type EPEC (Figure 5). Band intensity (chemiluminescent signals) was quantified using densitometry normalized to EscJ levels within the same membrane fraction. A reduced amount of membrane associated CesT was observed for ΔescU and ΔescU expressing either EscU(N262A) or EscU(P263A), as determined by densitometric analyses. The reduced amount was statistically significant for the escU null mutant compared to wild type EPEC, although this significance did not extend to the EscU variants. Next, the membrane fractions were subjected to sucrose gradient fractionation to assess CesT membrane localization patterns. EscJ and intimin are inner and outer membrane proteins respectively and hence served to identify inner and outer membrane enriched fractions. For ΔescU expressing HA-EscU-FLAG, a strong enrichment of CesT was found within inner membrane fractions. In contrast, HA-EscU(262)-FLAG and HA-EscU(263)-FLAG showed a more diffuse pattern of CesT membrane association, with a considerable amount of CesT protein localizing to less dense fractions within the gradient. These observations suggested that CesT function could be altered or less efficient in the absence of EscU auto-cleavage. We therefore carried out a co-immunoprecipitation assay, using anti-CesT antibodies, to assess CesT-effector interactions. Moreover, it has been shown that HpaB, a type III chaperone, interacts with HrcU [48] (EscU homologue) and hence we asked whether CesT interacts with EscU. Affinity purified anti-CesT antibodies co-immunoprecipitated equal amounts of Tir from all bacterial lysates (Figure 6). This was expected, since CesT is required for Tir stability [46, 47], and an earlier result that showed equal steady state levels of Tir in whole cell lysates expressing EscU variants (Figure 1). In contrast, both auto-cleaved and un-cleaved forms of EscU were not co-immunoprecipitated with anti-CesT antibodies.

Taken together, these data indicate that total CesT membrane levels were not statistically different for EscU variant expressing strains, although the nature of CesT association with the inner membrane was altered in the absence or with limited EscU auto-cleavage. CesT retained normal effector binding function in the absence of EscU auto-cleavage and EscU did not co-immunoprecipitate with CesT.

Bacterial Strains and Growth Media

Isolation of Genomic and Plasmid DNA

Genomic DNA was isolated from bacterial strains using the Purogene genomic DNA isolation kit (Gentra systems). Plasmids were isolated from bacterial strains using the QIAprep spin miniprep kit (Qiagen).

Recombinant Plasmid Construction

To generate pJLT21, pJLT22 and pJLT23, DNA fragments corresponding to the relevant escU allele were amplified by PCR using primers JT8 and JT10 from isolated plasmid DNA of pETescU_HIS, pETescU(N262A)_HIS and pETescU(P262A)_HIS, respectively. The resulting 1.5 kb products were purified and treated with restriction enzymes and cloned into pACYC184 treated BamHI and SalI. To generate pJLT24, pJLT25 and pJLT26, DNA fragments corresponding to the relevant escU allele were amplified by PCR using primers HAEscU and EscURevBglII using plasmid DNA as template. The resulting PCR products were purified and sub-cloned into pFLAG-CTC vector using XhoI and BglII.

To generate pTir-bla, primers XH1 and XH2 were used to PCR amplify the tir open reading frame (without the stop codon) using EPEC genomic DNA as template. The resulting PCR product was treated with AseI and EcoRI and cloned into NdeI/EcoRI treated pCX341 (generously provided by I. Rosenshine) [43] to create pTir-bla. The resulting plasmid construct was electroporated into EPEC and transformants were selected using tetracycline. Expression of Tir-TEM1 was verified by immunoblotting using anti-TEM1 antibodies (QED Biosciences).

Construction of mutants in EPEC E2348/69

A chromosomal deletion of escU was generated using allelic exchange [39]. Chromosomal DNA regions flanking the escU open reading frame were amplified from EPEC genomic DNA by PCR using primer pairs JT1/JT2 and JT3/JT4. The resulting 0.9 kb and 1.2 kb products were treated with NheI and then combined in a 1:1 ratio followed by the addition of T4 DNA ligase. After an overnight incubation at 16°C, an aliquot of the ligation reaction was then added to a PCR with primers JT1 and JT4 which generated a 2.1 kb product. The product was digested with SacI and cloned into pRE112 using E. coli DH5αλpir as a cloning host. The resulting plasmid PΔescU was verified using primers JT1 and JT4 by sequencing. PΔescU was then transformed into the conjugative strain SM10λpir which was then mated with EPEC E2348/69. EPEC integrants harbouring PΔescU on the chromosome were selected by plating onto solid media supplemented with streptomycin and chloramphenicol. The resulting colonies were then plated onto sucrose media (1% [w/v] tryptone, 0.5% [w/v] yeast extract, 5% [w/v] sucrose and 1.5% [w/v] agar) and incubated overnight at 30°C. The resulting colonies were screened for sensitivity to chloramphenicol, followed by a PCR using primers JT1 and JT7 to verify deletion of the escU from the chromosome. Cis-complementation mutants were generated using the same allelic exchange approach using primers NT278 and NT279 for escU(N262A) and primers NT281 and NT282 for escU(P263A) genetic constructs. To generate the ΔescNΔescU and ΔsepDΔescU double mutants, SM10λpir/PΔescU was conjugated with ΔescN [65], ΔsepD [66] as described above. For genetic trans-complementation studies, the appropriate plasmids were transformed into electrocompetent strains followed by antibiotic selection.

In vitrosecretion assay

Secretion assays were performed as previously described [39] with some minor modifications. To aid in the precipitation of proteins from secreted protein fractions, bovine serum albumin (100 ng) was added as a carrier protein during the precipitation step.

Isolation of bacterial membrane fractions

Total crude bacterial membrane preparations were generated from EPEC strains using a previously established protocol that preserves peripherally associated proteins and enriches for T3SS needle complexes [39]. Fractionation of membrane preparations was achieved using sucrose density gradients as previously described [39].

Immunoprecipitation

Immunoprecipitations with EPEC cell lysates were performed as previously described [39]. Briefly, 500 ng of affinity purified polyclonal anti-CesT antibody was added to 50 μl of Protein A conjugated agarose beads (Invitrogen) followed by washing as directed by the manufacturer. The antibody-bead mixture was blocked in phosphate buffered saline (PBS, 137 mM NaCl, 2.7 mM KCl, 4.3 mM Na₂HPO₄, 1.47 mM KH₂PO₄) supplemented with 1% (w/v) bovine serum albumin and then added to lysate preparations and incubated overnight at 4°C on a rotator. The samples were gently pelleted and the agarose beads were washed 3 times with PBS. The agarose beads were then exposed to 100 mM glycine (pH 2.2) to elute bound proteins and neutralized with 1 M Tris (pH 8.8) and then prepared for SDS-PAGE.

Infection of HeLa cells

HeLa cells [American Type Culture Collection (ATCC)] were seeded onto sterile glass coverslips at a density of 1 × 10⁵ /ml, grown for 24 hrs and then infected with various EPEC strains at a multiplicity of infection of 50 for 3 hours. The infected HeLa cells were then prepared for microscopy as previously described [35]. Images were detected using a Zeiss Axiovert 200 inverted microscope and captured using a Hamamatsu ORCA-R2 digital camera. Microscopy based quantification of EPEC intimate adherence (binding index) was performed as previously described [67]. Briefly, GFP positive bacteria (which were identified by GFP fluorescence) that were associated with actin pedestals were quantified. At least 50 cells were examined per sample.

β-lactamase reporter assays

Type III effector-TEM1 fusion reporter assays for EPEC strains were performed as previously described [42] with minor modifications. Briefly, HeLa cells (seeded to confluence in 96 well, black, clear bottom plates [Costar 3603]) were infected with a MOI of approximately 50 for 2 hours using bacteria that had been pre-activated in DMEM +10% FBS for 2 hours at 37°C, 5% CO₂. After 1 hour of infection, IPTG was added to a final concentration of 0.5 mM. The infected cells were gently washed twice with DMEM and then loaded with CCF2/AM using a Toxblazer kit (Invitrogen). The 96 well plate was incubated for 90 min in the dark and then placed in a Victor X plate reader (Perkin Elmer) set to read fluorescence using an excitation filter for 405 nm and emission filters for 460 nm (blue signal)/530 nm (green signal). Blue/green signal ratios and statistical significance (two sided Student's t test) were calculated as previously described [42]. The presented data are mean values of the results from three experiments.

Protein electrophoresis and Immunoblotting

All protein samples were separated by SDS-PAGE as described [68]. Separated polypeptides were visualized by Coomassie G-250 blue staining or Sypro Ruby Red (Bio-Rad, according to manufacturers directions). Pre-stained Broad Range Protein Markers (New England Biolabs, cat. # 7708) were used for standards. For immunoblotting, separated polypeptides were transferred to Immobilon-P membrane (Millipore), blocked with skim milk (5% [w/v] in Tris buffered saline + Tween20 [TBS-T]) or BSA, and then incubated with specific antibodies at working concentrations; anti-EscJ, 1:500 [69]; anti-EscN, 1/500 [39]; anti-EspB, 1:200 [70]; anti-EspA [71]; anti-intimin 1:1000 (gift from J. Leong); anti-HA, 1:5000 (gift from R. Duncan); anti-FLAG, 1:5000 (Sigma); anti-DnaK, 1:5000 (Calbiochem), anti-Tir, 1:1000 [35]; anti-TEM1, 1:2000 (QED Biosciences); goat anti-mouse conjugated to horse radish peroxidise (HRP), 1:5000 (Rockland immunochemicals); goat anti-rabbit conjugated to HRP, 1:5000 (Rockland immunochemicals); goat anti-rat conjugated to HRP, 1:5000. Anti-CesT polyclonal antibodies were raised in New Zealand white rabbits against a synthetic peptide (LENEHMKIEEISSSDNK) corresponding to the C-terminal region of CesT (Pacific Immunology, CA, USA, [NIH Animal Welfare Assurance Number: A4182-01]). Final bleeds were affinity purified against the peptide by the supplier and used in immunoblots at a 1:10000 dilution. Immunoblots were developed using an enhanced chemiluminescence reagent (ECL, GE Healthcare) and data captured on a VersaDoc 5000 MP (Bio-Rad). Densitometry measurements to evaluate band intensity from chemiluminescent signals in immunoblotting experiments were performed using Quantity One software (Bio-Rad). Immunoblots were imaged simultaneously and within exposure times that were within an empirically determined linear range of signal detection.

Ethics statement

The grant proposal supporting this research was reviewed by the Dalhousie University Ethics Officer. Ethics approval was not required as the research does not involve human subjects, primary human cell lines/samples or animals.