The N-terminal amphipathic region of the Escherichia coli type III secretion system protein EspD is required for membrane insertion and function

Authors

  • Dayal Dasanayake,

    1. Institute of Parasitology and Centre for Host-Parasite Interactions, Macdonald Campus of McGill University, 21, 111 Lakeshore Road, Ste-Anne-de-Bellevue, Québec H9X 3V9, Canada.
    Search for more papers by this author
  • Manon Richaud,

    1. Institute of Parasitology and Centre for Host-Parasite Interactions, Macdonald Campus of McGill University, 21, 111 Lakeshore Road, Ste-Anne-de-Bellevue, Québec H9X 3V9, Canada.
    Search for more papers by this author
  • Normand Cyr,

    1. Institute of Parasitology and Centre for Host-Parasite Interactions, Macdonald Campus of McGill University, 21, 111 Lakeshore Road, Ste-Anne-de-Bellevue, Québec H9X 3V9, Canada.
    Search for more papers by this author
  • Celia Caballero-Franco,

    1. Institute of Parasitology and Centre for Host-Parasite Interactions, Macdonald Campus of McGill University, 21, 111 Lakeshore Road, Ste-Anne-de-Bellevue, Québec H9X 3V9, Canada.
    Search for more papers by this author
    • Present address: Department of Dermatology, Massachusetts General Hospital, 149 13th Street, Charlestown, MA 02129, USA.

  • Sabrina Pittroff,

    1. Institute of Parasitology and Centre for Host-Parasite Interactions, Macdonald Campus of McGill University, 21, 111 Lakeshore Road, Ste-Anne-de-Bellevue, Québec H9X 3V9, Canada.
    Search for more papers by this author
  • Ron M. Finn,

    1. Department of Biochemistry, University of Victoria, Petch Bldg., Room 258, Victoria, BC V8W 3P6, Canada.
    Search for more papers by this author
  • Juan Ausió,

    1. Department of Biochemistry, University of Victoria, Petch Bldg., Room 258, Victoria, BC V8W 3P6, Canada.
    Search for more papers by this author
  • Wensheng Luo,

    1. Division of Infectious Diseases, Department of Medicine, University of Maryland School of Medicine, 20 Penn Street, Baltimore, Maryland 21201, USA.
    Search for more papers by this author
  • Michael S. Donnenberg,

    1. Division of Infectious Diseases, Department of Medicine, University of Maryland School of Medicine, 20 Penn Street, Baltimore, Maryland 21201, USA.
    Search for more papers by this author
  • Armando Jardim

    Corresponding author
    1. Institute of Parasitology and Centre for Host-Parasite Interactions, Macdonald Campus of McGill University, 21, 111 Lakeshore Road, Ste-Anne-de-Bellevue, Québec H9X 3V9, Canada.
      E-mail armando.jardim@mcgill.ca; Tel. (+1) 514 398 7727; Fax (+1) 514 398 7857.
    Search for more papers by this author

E-mail armando.jardim@mcgill.ca; Tel. (+1) 514 398 7727; Fax (+1) 514 398 7857.

Summary

Enterohemorrhagic Escherichia coli is a causative agent of gastrointestinal and diarrheal diseases. These pathogenic E. coli express a syringe-like protein machine, known as the type III secretion system (T3SS), used for the injection of virulence factors into the cytosol of the host epithelial cell. Breaching the epithelial plasma membrane requires formation of a translocation pore that contains the secreted protein EspD. Here we demonstrate that the N-terminal segment of EspD, encompassing residues 1–171, contains two amphipathic domains spanning residues 24–41 and 66–83, with the latter of these helices being critical for EspD function. Fluorescence and circular dichroism analysis revealed that, in solution, His6-EspD1–171 adopts a native disordered structure; however, on binding anionic small unilamellar vesicles composed of phosphatidylserine, His6-EspD1–171 undergoes a pH depended conformational change that increases the α-helix content of this protein approximately sevenfold. This change coincides with insertion of the region circumscribing Trp47 into the hydrophobic core of the lipid bilayer. On the HeLa cell plasma membrane, His6-EspD1–171 forms a homodimer that is postulated to promote EspD–EspD oligomerization and pore formation. Complementation of ΔespD null mutant bacteria with an espDΔ66–83 gene showed that this protein was secreted but non-functional.

Introduction

Escherichia coli (E. coli) is a Gram-negative bacterium present in the intestinal microbiota of healthy humans (Marteau et al., 2001; Guarner and Malagelada, 2003). Certain strains of E. coli, however, have acquired a sophisticated mechanism for adhesion and alteration of the epithelial cells lining the human intestinal tract. Enterohemorrhagic (EHEC) and enteropathogenic (EPEC) E. coli are pathogens responsible for gastrointestinal infections (Russo and Johnson, 2003; Kaper et al., 2004). Association of EPEC or EHEC cells with host epithelial cells induces expression of a set of proteins required for the assembly of a structural organelle known as the type III secretion system (T3SS) (Garmendia et al., 2005). The T3SS facilitates translocation of proteins directly from the bacteria into the host epithelial cells. Effector proteins injected into gut enterocytes alter a myriad of eukaryotic cell functions. However, the most striking phenotype associated with pathogenesis is the remodelling of the host cell actin cytoskeleton leading to deformation of microvilli into dysfunctional pedestal-like structures, a process known as attaching and effacing (A/E) (Knutton et al., 1987; Frankel et al., 1998; Nougayrede et al., 2003; Shaw et al., 2005; Campellone et al., 2007).

Proteins responsible for the A/E lesion are encoded on a pathogenicity island of ∼ 35 kb referred to as the locus of enterocyte effacement (McDaniel et al., 1995; Pallen et al., 2005). In EPEC, the locus of enterocyte effacement has been estimated to contain 41 open reading frames that encode for effector molecules such as EspF, EspG, EspH and Map; regulatory proteins Ler and GrlA/GrlR (Jepson et al., 2003; Tu et al., 2003; Matsuzawa et al., 2005; Shifflett et al., 2005; Mellies et al., 2007); a bacterial outer membrane adhesin known as intimin; its receptor Tir (Shaw et al., 2002); and the secretion system that consists of the Esc, Sep, EspA, EspB and EspD proteins (Jarvis et al., 1995; Lai et al., 1997; Wolff et al., 1998; Goosney et al., 2000; Ide et al., 2001; Mota and Cornelis, 2005).

EspD is actively secreted through the T3SS and is essential for EPEC and EHEC attachment and effacement of gastroepithelial cells (Lai et al., 1997; Kresse et al., 1999; Daniell et al., 2001). Expression of EspD in the absence of epithelial cells results in the accumulation of this protein as aggregates in culture media (Daniell et al., 2001). To penetrate the host cell plasma membrane, it is postulated that EspD is secreted into the extracellular milieu and subsequently inserts into specialized domains on the host cell membrane to form a translocation pore through which effector proteins are transferred into the cytosolic compartment of host cells (Ide et al., 2001; Hayward et al., 2005). In EHEC EDL933 (O157:H7), espD encodes a 374-amino-acid protein (MW 39 kDa) predicted to contain two coiled-coil motifs and two transmembrane domains (Fig. 1A) (Wachter et al., 1999). Mutational analysis has confirmed that the coiled-coil domain situated proximal to the C-terminus (residues 330–347) is important for EspD self-association (Daniell et al., 2001). Despite the importance of EspD for pathogenesis, a detailed structure-function analysis of this protein has yet to be performed to elucidate the molecular mechanisms associated with the insertion of this protein into the host cell membrane. To biochemically dissect this process, the N-terminal region of EspD (His6-EspD1–171) was overexpressed and the interaction of this fragment with the host cell membrane investigated. Moreover, we show that His6-EspD1–171 in solution, or bound to the plasma membrane of HeLa cells, forms a dimeric structure. Finally deletion of the amphipathic region 66–83 from EspD abrogates activity.

Figure 1.

Bioinformatic analysis of EspD. A. The schematic illustrates the structural features of EspD. Predicted coiled-coil segments are denoted by hatched boxes corresponding to residues 137–171 (COIL I) and 334–370 (COIL II). Two putative transmembrane domains situated between residues 175–200 and 228–250 are indicated by the black boxes. B. The plot shows the hydrophobic moment (grey line) and hydrophobic index (dashed line) for the EspD sequence calculated by the Heliquest algorithm. C. Sequences corresponding to the amphipathic helices (residues 24–41 and 66–83) were plotted on an Edmundson helical wheel. The hydrophobic sector of the helix is denoted by the black spheres. D. The plot shows the intrinsic disordered probability of the EspD determined by the PONDR algorithm. Regions of sequence with a probability index above 0.5 are postulated to adopt a disordered or molten globular structure tertiary structure. E. A pair wise alignment of the EspD from EPEC strain E2348/69 and EHEC EDL 933 was performed using the clustal×2 program.

Results

Bioinformatic analysis of EspD

In silico analysis of E. coli 0157:H7 EDL933 EspD revealed two putative transmembrane domains (Pallen et al., 1997; Wachter et al., 1999; Daniell et al., 2001) and two potential coiled-coil motifs situated between residues 138–167 (COIL I) and 330–350 (COIL II). The COILS algorithm (Lupas et al., 1991) predicted the COIL I region of EspD with 100% probability (Fig. 1A). Hydrophobic moment analysis of EspD using the program HeliQuest and AmphipaSeek (Sapay et al., 2006) revealed three amphipathic segments located between residues 24–42 (amphipathic region I), 66–83 (amphipathic region II) and 226–275 (amphipathic region III) (Fig. 1B). Analysis of amphipathic regions I and II on an Edmundson helical wheel revealed a hydrophobic sector circumscribed by two acidic residues (Fig. 1C), a structural feature of amphitropic proteins with membrane binding activity (Cornell and Taneva, 2006). Interestingly, amphipathic region III overlapped with the predicted second transmembrane domain of EspD (Fig. 1B).

Examination of the EspD sequence using PONDR, an algorithm that predicts the propensity of a protein to adopt an intrinsic or native disordered structure (Peng et al., 2005), revealed that EspD, and in particular the region spanning residues 1–171, favours an unfolded conformation (Fig. 1D), a characteristic of intrinsically disordered proteins (Peng et al., 2005). The portion of EspD with the highest predicted degree of order correlates with residues 140–290, a region that comprises the transmembrane helices (Fig. 1D).

The EPEC EspD, which shares 86% sequence similarity with EHEC EspD, has a comparable structural architecture (Fig. 1E).

EspD N-terminus forms dimers

Initial attempts to express full length EspD in the E. coli K12 strain ER2566 for structure-function analysis were complicated by the low levels of expression and formation of inclusion bodies. However, since bioinformatic analysis suggested that the N-terminal region of EspD contained putative membrane binding activity our studies focus on this fragment of the protein. To functionally characterize the N-terminal region of EspD, a fragment encompassing residues 1–171 (His6-EspD1–171) was expressed in E. coli ER2566 and purified by metal affinity chromatography on Ni2+-NTA. Rabbit antisera raised against His6-EspD1–171 selectively recognized the full length ∼ 39 kDa protein present in the culture supernatant of EPEC strain E2348/69 but showed no immunoreactivity with the culture supernatant from E. coli UMD870, a ΔespD mutant that does not express EspD (Lai et al., 1997) (data not shown) and confirmed the specificity of the EspD antisera.

Size exclusion chromatography (SEC) analysis of His6-EspD1–171 showed that this protein eluted with an apparent size of ∼ 50 kDa and suggested that this protein formed a stable dimer (Fig. 2A). Interestingly, a ∼ 20 and a ∼ 50 kDa species were detected in Western blots of samples resolved by SDS-PAGE without boiling (Fig. 2B, lane 2). To dismiss the possibility that His6-EspD1–171 exhibited anomalous migration on SEC or SDS-PAGE, cross-linking experiments were performed with glutaraldehyde. In the absence of glutaraldehyde, His6-EspD1–171 samples boiled prior to SDS-PAGE migrated as 20 kDa monomeric species (Fig. 2C, lane 1). However, treatment of His6-EspD1–171 with increasing concentrations of glutaraldehyde resulted in the appearance of a stable ∼ 50 kDa dimer species after boiling samples prior to SDS-PAGE. A low level population of higher molecular weight species was also detected (∼ 90 kDa) (Fig. 2C).

Figure 2.

Analysis of the His6-EspD1–171 structure. A. Quaternary structures of His6-EspD1–171 (thick black line), His6-EspD1–171Δ66–83 (thick grey line) and His6-EspD1–107 (thin black line) were analysed by size exclusion chromatography on a Superose 12 column using a wavelength of 230 nm to monitor protein elution. The Superose 12 column was calibrated with a standard mixture containing bovine IgG (160 kDa), BSA (66 kDa), ovalbumin (45 kDa) and equine myoglobin (17 kDa). B. Purified His6-EspD1–171 was re-suspended in sample buffer and resolved by SDS-PAGE with (lane 1) and without (lane 2) boiling prior to electrophoresis. C. Aliquots of His6-EspD1–171, His6-EspD1–171Δ66–83 and His6-EspD1–107 (10 µg 30 µl−1) untreated (−) or treated (+) 0.8 mM glutaraldehyde for 30 min then boiled in SDS-PAGE sample buffer and the oligomerization state EspD analysed by Western blot.

The COILs algorithm (Lupas et al., 1991) predicted that residues 137–171 corresponded to a coiled-coil motif (Fig. 1A). To assess the capacity of this region to a form stable coiled-coil interaction, a fragment encompassing residues 108–171 (EspD108–171) was generated by mild acid cleavage of the Asp107-Pro108 peptide bond. Sedimentation equilibrium analysis of EspD108–171 showed that this peptide exhibited a MW of 8.7 kDa that closely corresponded to the theoretical mass of monomer (data not shown). Glutaraldehyde cross-linking also failed to trap an EspD108–171 dimer species. Collectively, analytical centrifugation and cross-linking experiments suggest that the EspD108–171 was monomeric and that in solution no stable coiled-coil interaction was apparent.

To further delineate the dimerization domain, His6-EspD1–107 and His6-EspD1–171Δ66–83 were expressed in E. coli and the quaternary structure examined. SEC analysis of His6-EspD1–171Δ66–83 showed that like His6-EspD1–171 this protein eluted as a dimer with mass of ∼ 48 kDa (Fig. 2A, grey line). Cross-linking experiments with His6-EspD1–171Δ66–83 also showed the presence of dimeric species and suggested that amphipathic helix II (residues 66–83) was not essential for dimerization (Fig. 2C). Similar analysis of His6-EspD1–107 showed that this protein eluted as single peak with a mass of ∼ 28 kDa, indicative of this fragment forming dimers (Fig. 2A). This was further validated by glutaraldehyde cross-linking (Fig. 2C).

Interaction of His6-EspD1–171 with small unilamellar vesicles

To assess if the amphipathic regions in His6-EspD1–171 exhibited membrane binding activity, density flotation centrifugation experiments were performed using small unilamellar vesicles (SUVs). In this technique, proteins binding to SUVs will float to the top of the sucrose gradient, whereas unbound proteins remain at the bottom of the gradient. These experiments showed that comparable levels of His6-EspD1–171 bound to DOPC, DOPC:DOPE, DOPS, DOPC:DOPS, DOPC:SM:Chol and DOPC:DOPE:Chol SUVs (Fig. 3A), suggesting that the phospholipid composition was not a major factor regulating the initial binding to these model membranes. In the absence of SUVs, no His6-EspD1–171 flotation was detected. Flotation experiments performed in the presence of 500 mM NaCl did not alter the binding of His6-EspD1–171 to SUVs; indicating that the interaction of His6-EspD1–171 with these lipid bilayers was stabilized by hydrophobic, rather than electrostatic interactions (Fig. 3B).

Figure 3.

Interaction of His6-EspD1–171 with SUVs. SUVs were mixed with 30 µg of His6-EspD1–171 in PBS (A) or PBS supplemented with 500 mM NaCl (B) prior to flotation centrifugation analysis. The distribution of His6-EspD1–171 in the sucrose gradient was analysed by silver stained SDS-PAGE. Lipids used for the preparation of SUVs included; dioleoyl phosphatidylcholine (DOPC), dioleoyl phosphatidylethanolamine (DOPE), dioleoyl phosphatidylserine (DOPS), sphingomyelin (SM), cholesterol (Chol).

Secondary structure analysis of His6-EspD1–171 showed that the amphipathic regions I and II were predicted to favour an α-helix configuration (Fig. 4A). To identify the membrane binding segments His6-EspD1–171 and His6-EspD1–171Δ66–83 were subjected to chemical and proteolytic degradation and the SUV binding activity was assessed. Flotation experiments performed with the purified partial acid hydrolysis fragments His6-EspD1–107 and EspD108–171 (Fig. 4B) showed that only His6-EspD1–107 bound to DOPC:DOPE:Chol SUVs (Fig. 4B). The capacity of the His6-EspD1–107 fragment to bind lipid bilayers was validated using the His6-EspD1–107 protein expressed in E. coli (data not shown). Treatment of His6-EspD1–171 with cyanogen bromide (CNBr), a reagent that cleaves at the C-terminus of methionine residues, generated a mixture of peptides ranging in size from 2.6 to 7.4 kDa (Fig. 4A). These peptides were all devoid of lipid bilayer binding activity. Collectively these data suggest that cleavage of His6-EspD1–171 at Met41 located between amphipathic helices I and II disrupted the membrane binding domain (Fig. 4A and C).

Figure 4.

Mapping the His6-EspD1–171 membrane binding domain. A. The secondary structure of His6-EspD1–171 was predicted using the GOR algorithm. Cylinders represent regions favouring α-helical structure and serrated boxes correspond to regions predicted to adopt a random or extended strand structure. Purified His6-EspD1–171 was cleaved using mild acid hydrolysis, cyanogen bromide or digested trypsin to generate an array of overlapping cleavage products. B–D. Products from the mild acid hydrolysis (B), cyanogen bromide (C) or tryptic digest (D) were subjected to flotation centrifugation analysis using DOPC:DOPE:Chol SUVs. E. Purified His6-EspD1–171Δ66–83 was incubated with and without DOPC:DOPE:Chol SUVs then subjected to sucrose density flotation and the distribution of this protein in the gradient assessed by Coomassie blue dye stained SDS-PAGE.

Similarly, digestion of native His6-EspD1–171 with trypsin generated ∼ 11 and 12 kDa peptides. Flotation experiments showed that both these peptides retained membrane binding activity (Fig. 4D). These peptides presumably correspond to N-terminal fragments with two missed tryptic cleavages, since complete degradation of His6-EspD1–171 would result in peptides of 0.1–4.0 kDa (Fig. 4A). Treatment of His6-EspD1–171 with clostripain, a protease that cleaves selectively at the C-terminus of arginine residues, generated an ∼ 11 kDa peptide that bound DOPC:DOPE:Chol SUVs (data not shown). These data are consistent with the results obtained with His6-EspD1–107.

The membrane binding activity of amphipathic helices I and II was further assessed by generating an internal deletion mutant lacking amphipathic helix II (His6-EspD1–171Δ66–83). Flotation experiments showed that His6-EspD1–171Δ66–83, like His6-EspD1–171, bound to DOPC:DOPE:Chol SUVs. In the absence of SUVs no His6-EspD1–171Δ66–83 was detected at the top of the gradient (Fig. 4E). Disruption of amphipathic helix I by substituting residues 32–46 with a 33-amino-acid sequence that is not predicted to form an amphipathic helix did not ablate fluorescence actin staining activity of EspD (Luo and Donnenberg, 2011). These data suggest that amphipathic helix I or II alone is sufficient for His6-EspD1–171 to bind lipid bilayers.

Structural analysis of His6-EspD1–171

To further investigate the interaction of His6-EspD1–171 with lipid bilayers we used the intrinsic fluorescence of the single tryptophan residue (Trp47) to monitor the structural changes associated with the binding of His6-EspD1–171 to SDS micelles or SUVs. In solution at pH 7.5, His6-EspD1–171 Trp47 exhibited an emission λmax at 357 nm at an excitation wavelength of 290 nm (Fig. 5A). Addition of SDS to His6-EspD1–171 resulted in a concentration dependent decrease in fluorescence intensity and a concomitant shift in the emission λmax to 344 nm in the presence of 8.0 mM SDS. The shift in the λmax to a lower wavelength is consistent with the relocation of Trp47 into the non-polar environment of the SDS micelle (Yau et al., 1998). Fluorescence quenching studies with acrylamide resulted in Stern-Volmer constants (Ksv) of 7.5 M−1 and 11.6 M−1 for His6-EspD1–171 in solution and in the presence of SDS micelles respectively. This decrease in the Ksv is consistent with the movement of Trp47 into the hydrophobic core of the SDS micelle, which protects Trp47 from dynamic quenching by acrylamide (data not shown).

Figure 5.

Fluorescence analysis of His6-EspD1–171. A. His6-EspD1–171 was titrated with increasing concentrations of SDS and the shift in the intrinsic fluorescence of Trp47 monitored at excitation wavelength of 290 nm. B and C. (B) The conformational changes in His6-EspD1–171 induced on binding to DOPS SUVs at pH 7.5 and pH 5.5 were assessed by monitoring changes in emission λmax of Trp47 and (C) by fluorescence quenching induced by acrylamide in the presence or absence of DOPS SUVs. Modified Stern-Volmer plots were generated by plotting the change in fluorescence at 350 nm as a function of the acrylamide concentration. D. The emission spectra for His6-EspD1–171 and His6-EspD1–171Δ66–83 were recorded in the absence or presence of DOPS SUVs at pH 5.5. E. The affect of pH on the intrinsic fluorescence was determined using 20 µM His6-EspD1–171 diluted into a universal buffer adjusted to pH 2.0–10.0. F. Fluorescence energy transfer from Trp47 to ANS was examined in the presence of 8.0 or 40 µM ANS using an excitation wavelength of 290 nm.

In contrast to SDS micelles, the binding of His6-EspD1–171 to DOPS, DOPC:SM:Chol and DOPC:DOPS:SM:Chol SUVs at pH 7.5 resulted in modest decreases in the fluorescence intensity without altering the emission λmax. These data indicated that His6-EspD1–171, bound to these SUVs, did not insert Trp47 into the hydrophobic core of the lipid bilayer (Fig. 5B). At pH 5.5, the binding of His6-EspD1–171 to DOPS SUVs was accompanied by a ∼ 8 nm blue shift in the emission λmax to 349 nm. This indicates that under acidic conditions DOPS facilitated a conformational change that promoted the insertion of Trp47 into the lipid bilayer (Fig. 5B). This is supported by acrylamide quenching experiments showing that at pH 5.5 His6-EspD1–171 exhibited a modified KSV value of 7.9 M−1 in solution and 17.5 M−1 when His6-EspD1–171 bound to DOPS SUVs at pH 5.5 (Fig. 5C). No change in the intrinsic fluorescence was observed with DOPC:SM:Chol or DOPC:DOPS:SM:Chol SUVs at an acidic pH (Fig. 5D). His6-EspD1–171Δ66–83 at pH 5.5 exhibited an emission λmax at 356 nm; however, no shift in the λmax was observed in the presence of DOPS SUVs at pH 5.5. These data suggest that deletion of amphipathic helix II impaired the insertion of the region circumscribing Trp47 into the hydrophobic core of the lipid bilayer (Fig. 5D). This contention was further validated by acrylamide quenching experiments, which showed that His6-EspD1–171Δ66–83 had comparable Ksv constants of 8.4 and 9.0 M−1, respectively, in the presence or absence of DOPS at pH 5.5 (Fig. 5C).

Titration experiments showed that His6-EspD1–171 did not exhibit any significant conformational change in solution at pH 5.0–9.0. However, at pH 2.0–5.0 His6-EspD1–171 undergoes a structural change as shown by the marked blue shift in the λmax of Trp47 emission fluorescence (Fig. 5E).

8-anilinonaphthalene-1-sulphonate (ANS), an extrinsic fluorescence probe that binds hydrophobic surfaces, was used to examine if His6-EspD1–171 membrane binding domain was exposed and capable of binding ANS. In addition since the excitation spectra of ANS overlaps with the emission spectra of Trp47, we employed fluorescence resonance energy transfer to assess if this tryptophan residue was in close proximity to the membrane binding domain. Excitation of His6-EspD1–171 at 290 nm in the presence of 8 or 40 µM ANS resulted in a concentration dependent decrease in the His6-EspD1–171 fluorescence intensity at 353 nm and a concomitant increase in the ANS fluorescence at 479 nm due to a fluorescence resonance energy transfer event (Wu and Brand, 1994) between Trp47 and the ANS (Fig. 5F). In the absence of His6-EspD1–171, no ANS fluorescence was observed at an excitation of 290 nm.

The circular dichroism (CD) spectrum of His6-EspD1–171 at pH 5.5 or 7.5 was dominated by minima at 200 nm, a feature that is characteristic of proteins with a high content of random structure. Indeed, deconvolution of the His6-EspD1–171 secondary structure showed that this protein contained 3% α-helix, 31% β-sheet, 20% β-turn and 44% random coil, a composition that was in agreement with the intrinsic disorder structure predicted by PONDR (Figs 1C and 6A). These data suggest that the His6-EspD1–171 conformation was not significantly affected by a pH change (Fig. 6 and Table 1). Interestingly, binding of His6-EspD1–171 to SDS micelles at pH 5.5 or 7.5 induced dramatic structural changes that increased the α-helix content of this protein ∼ 7–12-fold (Table 1 and Fig. 6). Addition of His6-EspD1–171 to DOPS SUVs (phospholipid : protein ratio of 100:1) at pH 5.5 triggered similar conformational changes that increased the α-helical content of this protein from 2% to 15% (Table 1). Surprisingly, no significant alteration in the α-helical content was induced by the binding of His6-EspD1–171 to DOPS SUVs at pH 7.5 (Table 1 and Fig. 6B). However, a shift in the minima from 200 nm to 208 nm suggests that the insertion of His6-EspD1–171 into DOPS SUVs influence the tertiary fold of this protein causing it to adopt a more ordered structure (Uversky, 2002). This result correlates with intrinsic fluorescence studies showing that DOPS SUVs induce a notable blue shift in the Trp47 emission only under acidic conditions. DOPC:SM:Chol or DOPC:DOPS:SM:Chol SUVs, despite binding His6-EspD1–171, failed to induce any notable changes in the CD spectra at pH 7.5 or 5.5 (data not shown).

Figure 6.

Structural analysis of His6-EspD1–171. A and B. The secondary structure changes induced when His6-EspD1–171 binds to SDS micelles (A) or DOPS SUVs (B) at pH 5.5 or pH 7.5 was examined by CD. C. The equilibrium unfolding of His6-EspD1–171 induced by increasing concentrations of urea was analysed by intrinsic fluorescence or CD. Unfolding of His6-EspD1–171 was monitored by the red shift in the emission λmax of Trp47. Perturbation in the secondary and tertiary structure of His6-EspD1–171 was examined by in the molar ellipticity at 222 nm. D. Refolding of His6-EspD1–171 in pH 5.5 buffer alone or buffer containing DOPS or DOPC:DOPC:SM:Chol SUVs was monitored by intrinsic fluorescence 25°C. E. The affect of urea denaturation on His6-EspD1–171 quaternary structure was assessed by glutaraldehyde cross-linking in the presence of increasing concentrations of urea (0–3.0 M). The affect of urea on His6-EspD1–171 dimer species was assessed by Western blot analysis (inset). Dimer species were quantified from blots using the Image J software to measure area.

Table 1.  Secondary structure assignment for His6-EspD1–171.
 pH 5.5pH 7.5pH 5.5 SDSpH 7.5 SDSpH 5.5 DOPSpH 7.5 DOPS
α-Helix2%3%24%16%15%3%
β-Sheet45%31%19%28%37%35%
β-Turn19%20%19%17%19%22%
Random coil34%44%28%38%28%39%

We next examined if SUVs impacted the refolding of His6-EspD1–171. His6-EspD1–171 in 8.0 M urea exhibited a fluorescence emission λmax at 358 nm. Dilution of denatured His6-EspD1–171 into pH 5.5 buffer resulted in a shift in the λmax to 353 nm. This is consistent with the refolding of this protein and the relocation of Trp47 to a less solvent exposed environment. Dilution of His6-EspD1–171 in the presence of DOPS SUVs at pH 5.5 was accompanied by a marked hypsochromic shift in the λmax to 343 nm. This indicated that the refolding of this protein on SUVs involved the insertion of Trp47 into the lipid bilayer (Fig. 6C). A modest 2 nm blue shift in the λmax to 351 nm was observed when refolding was performed in the presence of DOPC:DOPS:SM:Chol SUVs (Fig. 6C), indicating that these vesicles did not induce the structural changes necessary for insertion of Trp47 into the hydrophobic core of these membranes. Surprisingly, refolding experiments performed at pH 7.5 resulted in an emission λmax of 353 nm in buffer and in the presence of DOPS, and DOPC:DOPS:SM:Chol SUVs (data not shown).

The thermodynamic stability of the His6-EspD1–171 structure was next assessed by equilibrium unfolding. The denaturation of His6-EspD1–171 exhibited biphasic kinetics with transition points occurring at 0.25 and 2.5 M urea (Fig. 6D). Comparable biphasic kinetics were observed for the guanidinium hydrochloride denaturation (data not shown). CD analysis of urea denatured His6-EspD1–171 also revealed a biphasic unfolding process with transition midpoints in the 222 nm ellipticity occurring at 0.35 and 2.8 M urea (Fig. 6D).

To assess the effect of denaturants on His6-EspD1–171 structure, we employed glutaraldehyde cross-linking to trap oligomeric species. Western blot analysis revealed that the most significant loss of the His6-EspD1–171 dimer occurred at urea concentration > 1.0 M (Fig. 6E). Collectively, the cross-linking, CD and intrinsic fluorescence experiments suggest that the biphasic unfolding kinetics observed for His6-EspD1–171 correspond to the dissociation of the dimer followed by the denaturation of the monomeric species.

Binding of His6-EspD1–171 to HeLa and red blood cells

Addition of His6-EspD1–171 to HeLa and human red blood showed that this protein bound tightly to these cells. Subcellular fractionation of HeLa cells revealed that His6-EspD1–171 was quantitatively detected in the membrane fraction (Fig. 7A). Sequential extraction of membranes with high salt and alkaline sodium carbonate showed that the bulk of His6-EspD1–171 remained associated with the membrane pellet, suggestive of a non-polar interaction with the lipid bilayer. His6-EspD1–171 exhibited a biophysical behaviour similar to that of the epidermal growth factor receptor an integral membrane protein. Disruption of HeLa membranes with Triton X-100 resulted in of His6-EspD1–171 (Fig. 7A). Insertion of His6-EspD1–171 into human red blood cell membrane did not induce haemolysis. This is not surprising since the transmembrane domains that would insert through the plasma membrane in the formation of functional translocon pore are absent from His6-EspD1–171. Trypsin treatment of HeLa cells loaded with His6-EspD1–171 showed that this protein was rapidly degraded to a core doublet of 11 and 12 kDa within ∼ 15 min and confirmed that a significant portion of His6-EspD1–171 anchored to the plasma membrane remained susceptible to proteolysis (Fig. 7B). A similar protease resistant doublet was observed following trypsin digestion of His6-EspD1–171 in solution or on liposomes (Fig. 4B). Chemical cross-linking showed that His6-EspD1–171 bound to HeLa cells as a stable dimer (Fig. 7C). Interestingly, a minor tetrameric population (100 kDa) was also detected. Similar experiments performed with the deletion mutant His6-EspD1–171Δ66–83 showed that this protein also bound to HeLa cell membranes and was stabilized non-polar interactions (data not shown). These data suggest that the initial contact of EspD with the host cell membrane is mediate by several segments within the N-terminal region of protein.

Figure 7.

Interaction of His6-EspD1–171 with HeLa cells. A. HeLa cells were loaded with His6-EspD1–171 and a crude membrane (M) fraction was prepared. The membrane fraction was sequentially treated with 500 mM NaCl, 100 mM sodium carbonate pH 11.5, and 1% Triton X-100 and then separated into a supernatant (S) and pellet (P) fractions and the distribution of His6-EspD1–171 was analysed by Western blot. Blots were also probed to assess the distribution of the integral membrane marker protein epidermal growth factor receptor. B. Intact HeLa cells loaded with His6-EspD1–171 were treated with trypsin and the His6-EspD1–171 cleavage products examined by Western blot analysis. C. Intact HeLa cells were incubated with His6-EspD1–171 for 2 h at 37°C and the quaternary structure of protein bound to the plasma membrane was trapped by treating cells with 0.2–2.0 mM glutaraldehyde at 4°C for 20 min.

Effect of deleting amphipathic region II on EspD

The EPEC espD null mutant UMD870 transformed with pQE80-espD exhibited expression of EspD in the cell pellet and culture supernatant in the absence of IPTG. Addition of 25 µM IPTG to the media increased expression and secretion of EspD > 10-fold (Fig. 8A). In contrast, UMD870 transformed with pQE80-espDΔ66–83 expressed lower levels of EspDΔ66–83 in the absence of IPTG. However, the expression levels were dramatically increased in the presence of 25 µM IPTG (Fig. 8A), albeit the total amount of EspDΔ66–83 was diminished compared with EspD. Secretion of EspDΔ66–83 was markedly impaired as the levels of this mutant protein in the culture supernatant were ∼ 20-fold lower when compared with EspD (Fig. 8A). UMD870 cell transformed with the empty vector pQE80 showed no detectable EspD (data not shown).

Figure 8.

Complementation of UMD870 with EspD. A. E. coli UMD870 transformed with the pQE80 vectors encoding EHEC EspD or espDΔ66–83 were cultured in DMEM media in the presence or absence of 25 µM IPTG. Culture supernatants were harvested and the expression and secretion of the EspD/espDΔ66–83 assessed by Western blot analysis. B. For fluorescence actin staining experiments, HeLa cells were infected with UMD870 transformed with pQE80 encoding EHEC EspD (panels a and b) or espDΔ66–83 (panels c and d) and then stained with fluorescein isothiocyanate-phalloidin and examined by phase-contrast (panels a and c) or fluorescence microscopy (panels b and d).

To assess the importance of amphipathic region II for translocation pore formation and injection of effector proteins required for A/E a fluorescence actin staining assay was performed (Lai et al., 1997). Infection of HeLa cells with UMD870 transformed with pQE80-espD or pQE80-espDΔ66–83 showed that both bacterial cell lines attached to HeLa cells and formed microcolonies (Fig. 8B, panels a and c). However, staining of HeLa cells with FITC-phalloidin revealed a large accumulation of actin filaments under UMD870 secreting EspD. In contrast, UMD870 secreting EspDΔ66–83 showed no detectable actin accumulation under the microcolonies indicative of EspDΔ66–83 failing to form a functional translocation pore.

Discussion

EspD is a critical component of the T3SS required for attachment and pathogenesis of EPEC and EHEC. Numerous studies have demonstrated that EspD is secreted into the extracellular milieu and subsequently inserts into the host cell plasma membrane to form a translocation pore (Wainwright and Kaper, 1998; Kresse et al., 1999; Wachter et al., 1999; Daniell et al., 2001; Ide et al., 2001). To dissect the initial molecular events associated with the recruitment and insertion of EspD into the host cell plasma membrane, we initiated a structural characterization of the N-terminal region of EspD encompassing residues 1–171 (His6-EspD1–171). Bioinformatic analysis predicted that EspD adopts an intrinsically disordered structure. The propensity of the EspD N-terminus to favour a native unfolded state was substantiated by CD measurements showing that His6-EspD1–171 is dominated by a distinctive minimum at 200 nm, a spectral feature characteristic of proteins with a prominent random or molten globular structure (Uversky, 2002). The native unfolded structure of EspD (Fig. 1) is advantageous since threading of newly synthesized proteins via the T3SS needle apparatus dictates that these proteins need to be devoid of tertiary structure to pass through the narrow 2–3 nm conduit formed by EspA (Sekiya et al., 2001; Delahay and Frankel, 2002).

Surprisingly, despite the intrinsic disordered structure, His6-EspD1–171 formed a stable homodimeric structure in solution and on the plasma membrane of HeLa cells. These data, together with previous studies on the C-terminal coiled-coil (Daniell et al., 2001), suggested that the N- and C-terminal portions of EspD are important for stabilizing EspD–EspD interactions necessary for translocation pore assembly. However, sedimentation equilibrium centrifugation and cross-linking studies using a peptide spanning residues 108–171 (EspD108–171) failed to detect a dimeric species. The fragment encompassing residues 1–107, however, formed a stable dimer in solution. These data suggest that a segment within the first 107 residues was responsible for dimerization of His6-EspD1–171. The amphipathic domain spanning residues 66–83 is not required for oligomerization since deletion of this region did not disrupt dimerization. Moreover, it should be stress that prior dimerization of His6-EspD1–171 is not necessary for membrane binding since urea denatured protein diluted into liposome preparations showed that this protein had a comparable membrane binding activity (data not shown). This is important since the full length EspD emerging from the T3SS needle would be in a largely unfolded configuration and bind the membrane as a monomer and then subsequently oligomerize on the host plasma membrane. Our studies however do not exclude the possibility that amphipathic domain I (residues 24–41) or that a region more proximal to the N-terminus of EspD may be involved in forming an EspD–EspD interaction. Fine mapping studies are presently underway to define the protein–protein interactions that stabilize the His6-EspD1–171 homodimerization.

Flotation centrifugation experiments with SUVs verified that His6-EspD1–171 membrane binding to anionic and zwitterionic lipid bilayers was stabilized by hydrophobic, rather than electrostatic interactions. His6-EspD1–171 also bound tightly to HeLa cell membranes and suggested that the initial association with lipid bilayers was independent of the phospholipid composition. Our mapping studies indicate that the amphipathic helices spanning residues 24–41 and 66–83 are important for the membrane binding activity of EspD. Similar structural features have been identified in the related pore forming proteins SipB and IpaB from Salmonella and Shigella respectively (Hayward et al., 2000; McGhie et al., 2002; Hume et al., 2003). Analysis of residues 66–83 on an Edmundson wheel revealed the presence of a characteristic non-polar helical face that drives membrane insertion and stabilizes binding by making hydrophobic contacts with the lipid bilayer core (Cornell and Taneva, 2006). We postulated that the N-terminal amphipathic helices I and II are important for membrane anchoring of newly secreted EspD and for promoting insertion of the regions circumscribing these helices. Binding via this amphipathic helix would facilitate the recruitment of EspD to the host cell membrane prior to oligomerization or assembly of the translocation pore. The binding of His6-EspD1–171 alone to HeLa or human red blood cells was not sufficient to disrupt cellular integrity and suggests that His6-EspD1–171 only penetrated into the outer leaflet of the plasma membrane and confirmed that this segment of EspD alone is not sufficient for pore formation.

Complementation of the ΔespD mutant UMD870 showed that wild-type EspD and EspDΔ66–83 were expressed in these bacteria; albeit the levels of EspDΔ66–83 secreted into the culture media were ∼ 20-fold lower when compared with EspD; a phenotype that was similar to EPEC null mutants lacking the chaperones CesD or CesD2 (Wainwright and Kaper, 1998; Neves et al., 2003). In addition to being involved in membrane binding, amphipathic helix II may represent a binding site for chaperones CesD2 or CesD, which is necessary for maintaining EspD in a secretion competent conformation (Wainwright and Kaper, 1998; Neves et al., 2003). Moreover, the lower steady state intracellular levels of EspDΔ66–83, when compared with EspD, imply that the amphipathic helix II may be important to protect EspD from rapid degradation if this protein is not secreted into the culture supernatant.

Interestingly, the interfacial region of the His6-EspD1–171 amphipathic helix is demarcated by two glutamate residues (Fig. 1). Previous studies have demonstrated that acidic residues at this location impart a selectivity that allows peptides to preferentially bind anionic lipid bilayers containing phospholipids, such as phosphatidylserine (Dunne et al., 1996; Johnson et al., 2003). However, the exact role of these glutamate residues in EspD will need to be determined by additional site-directed mutagenesis studies. This selectivity is driven by a localized surface pH gradient that may be 1–2 pH units lower than the bulk solution (Eisenberg et al., 1979). This more acidic environment at DOPS SUV membrane surface likely alters the charge properties of His6-EspD1–171 increasing its hydrophobicity, which facilitates lipid bilayer insertion of the region circumscribing Trp47 and triggers a approximately sevenfold increase in the α-helix content of His6-EspD1–171 (Leenhouts et al., 1995; Sui, 2000; Johnson et al., 2003). This hypothesis is supported by the finding that a shift in pH from 7.5 to 5.5 in solution was not sufficient to induce a noteworthy structural change in His6-EspD1–171 (Fig. 5). Although the binding of His6-EspD1–171 to DOPS SUVs at neutral pH provoked only a modest secondary structure rearrangement (Table 1), there was a notable shift in the CD spectral band from 200 nm to 208 nm. This 8 nm shift, together with the appearance of a shoulder peak at 222 nm, indicates that the association with anionic, but not zwitterionic, membranes causes a conformational change in which His6-EspD1–171 adopts a more tightly packed configuration (Uversky, 2002). Alterations in the His6-EspD1–171 structure require a significant negative charge distribution in the lipid bilayer since incorporation of 20% DOPS into DOPC:DOPS:SM:Chol SUVs had no notable impact on the His6-EspD1–171 architecture. Phosphatidylserine is known to be an important constituent for the incorporation of the E. coli bacterial receptor protein Tir (Kenny et al., 1997; Race et al., 2006) and PopD, a Pseudomonas aeruginosa homologue of EspD (Faudry et al., 2006) into the host cell membrane. It is tempting to speculate that the structural rearrangements induced by the interaction of His6-EspD1–171, in particular residues 66–83, with the lipid bilayer are instrumental in driving the translocation of the two EspD transmembrane domains into the host cell plasma membrane. This is supported by the finding that deletion of amphipathic helix II did not abrogate membrane binding but disrupted the structural changes required for insertion of the region proximal to Trp47 into anionic lipid bilayers. Loss of this amphipathic helix in addition to reducing the level EspDΔ66–83 also impaired the ability of this mutant EspD to restore a functional T3SS system in UMD870 E. coli. When residues 62–76 from EPEC EspD (EspDΔ62–76) were replaced with residues specified by restriction endonuclease sites, the resulting protein could not be detected by immunoblotting. In contrast, EPEC EspD mutant variants in which residues 32–46 or 77–91 were replaced with the same residues (Deng et al., 2007) complement the ΔespD deletion in UMD870 as shown by the FAS test (data not shown). These data further support the contention that amphipathic region II (residues 66–83) is important for pore formation.

Microbial pathogens are known to exploit membrane microdomains such as lipid rafts, which are enriched for sphingomyelin and cholesterol, as invasion portals (Rosenberger et al., 2000; Lafont and van der Goot, 2005; Allen-Vercoe et al., 2006). Contact of EPEC or enterotoxigenic E. coli with HeLa or T84 cells has been shown to cause a rapid mobilization of phosphatidylethanolamine and phosphatidylserine to the cell surface (Barnett Foster et al., 1999; Crane et al., 1999). This surface display of phosphatidylethanolamine, an adherence ligand for bundle-forming pilus, has led to the notion that this lipid aids in E. coli attachment and microcolony formation on gastrointestinal epithelial cells (Barnett Foster et al., 1999; 2000; Tobe and Sasakawa, 2002; Allen-Vercoe et al., 2006). It is interesting to note that the accumulating evidence suggests that the externalization of phosphatidylserine is also dependent on lipid rafts (Gargalovic and Dory, 2003; Zech et al., 2009), structures required by EPEC and EHEC for attachment and pedestal formation (Goosney et al., 2000; Riff et al., 2005; Allen-Vercoe et al., 2006; Johnson et al., 2009). It is possible that distribution of phosphatidylserine in lipid rafts may not be homogeneous since this phospholipid can partition into non-raft like regions (Geisse et al., 2004; Zech et al., 2009). This partition may favour binding EspD and translocation pore formation. Finally, we demonstrated that co-incubation of large unilamellar vesicles with EPEC resulted in the insertion of EspD into vesicles of varying phospholipid compositions. However, injection of the effector protein Tir into the lumen of these vesicles was detected only when phosphatidylserine was included as a constituent of the lipid bilayer (Caballero-Franco and Jardim, in preparation).

Experimental procedures

Chemicals and reagents

Restriction endonucleases and DNA-modifying enzymes were obtained from Invitrogen (Burlington, ON) or New England Biolabs (Ipswich, MA). Cell culture reagents were purchased from Invitrogen. The pQE80 vector and Ni2+-NTA affinity beads were purchased from Qiagen (Mississauga, ON). Electrophoresis and chromatography products were obtained from Bio-Rad (Mississauga, ON). Sequencing grade porcine trypsin and clostripain were purchased from Promega (Madison, WI) and Worthington Enzymes (Lakewood, NJ). Phospholipids were procured from Avanti Polar Lipids (Alabaster, AB). Cholesterol and bovine sphingomyelin were obtained from SigmaAldrich (St Louis, MO). All other reagents were of the highest quality commercially available.

Protein expression

Escherichia coli EHEC 0157:H7 EDL933 (provided by Dr S. Gruenheid, McGill University) was propagated in Luria–Bertani (LB) media supplemented with 50 µg ml−1 of nalidixic acid. The open reading frame encoding residues 1–171 of EspD (His6-EspD1–171) was amplified by PCR using the primer pair 5′-ACGGATCCATGCTTAACGTAAATAACG-3′ and 5′-TCGAGCTCAAAG ACCTGGCCAACAATT-3′ (restriction sites are underlined) with 30 cycles of denaturation at 95°C for 30 s, annealing at 53°C for 10 s, extension at 68°C for 30 s and EHEC 0157:H7 EDL933 genomic DNA as the template. The PCR product (513 bp) was subcloned into the pQE80 vector using the BamHI and SacI restriction sites to generate the pQE80-His6-espD1–171 expression vector. The full length wild-type EHEC espD was amplified using the primer pair 5′-ACGGATCCATGCTTAACGTAAATAACG-3′ and 5′-TCGAGCTCTTAAATTCGGCCACT AACAATACG-3′ and the PCR product was cloned into the BamHI and SacI restriction sites of the pQE80 vector (pQE80-espD). The pQE80-espDΔ66–83 deletion mutant was generated using QuikChange protocol as described above and pQE80-espD as the template.

Deletion mutants were generated with the QuikChange PCR protocol using Pfx DNA polymerase, pQE80-His6-espD1–171 as the template and the primer pairs 5′-CTCAGGTGAATACCGTTGACTAACAGCAAATGATGATGATGG-3′ and 5′-CCATCATC ATCATTTGCTGTTAGTCAACGGTATTCACCTGAG-3′ to insert a stop codon at positions 322–324 (His6-espD1–107) for expression of an His6-EspD fragment encompassing residues 1–107. The internal deletion mutants lacking residues 66–83, His6-EspD1–171Δ66–83 and EspDΔ66–83 were generated using pQE80-His6-espD1–171 or pQE80-espD as the template and the primer pair 5′-GTCACTCATTAG TGACGCCCACAAGTCGCACTGAGGAGGC-3′ and 5′-GCCTCCTCAGTGCGACTTGTGG GCGTCACTAATGAGTGAC-3′ respectively to generate the expression constructs pQE80-His6-espD1–171Δ66–83.

Escherichia coli strain ER2566 cells (New England Biolabs) transformed with pQE80-His6-espD1–171 was grown to an OD600 of 0.6 in LB media containing 100 µg ml−1 ampicillin. Protein expression was induced with 0.2 mM IPTG for 4 h at 20°C. Bacterial cell pellets were suspended in 15 ml of lysis buffer A (50 mM sodium phosphate buffer pH 7.5, 500 mM NaCl) and lysed by French press (Thermo-Fisher Sci. Waltham, MA). Lysates were clarified by centrifugation and the supernatant was applied to a 2 ml Ni2+-NTA affinity column pre-equilibrated with lysis buffer A. The column was washed with 150 ml of buffer A containing 10 mM imidazole. His6-EspD1–171 was eluted with a 80–320 mM imidazole step gradient. Fractions containing His6-EspD1–171 were pooled and further purified by chromatography on a DEAE cellulose column (2 × 10 cm) equilibrated in 20 mM sodium phosphate. The DEAE void volume fraction containing His6-EspD1–171 was collected, dialysed against 20 mM sodium phosphate pH 7.5 and concentrated using an Amicon centrifugal filter device (2.0 mg ml−1), and stored at −80°C.

Analysis of espD mutations by fluorescence actin staining assay

The espD null mutant EPEC strain UMD870 (Lai et al., 1997) transformed with the pQE80-espD or pQE80-espDΔ66–83 were grown overnight in LB media containing 100 µg ml−1 ampicillin and 50 µg ml−1 kanamycin with shaking at 37°C. Cultures were diluted 1:50 in DMEM and incubated for 3 h without shaking at 37°C in a 5% CO2 environment. Cultures were treated for 1 h with 25 µM IPTG and the culture supernatant and cell pellets were harvested to determine EspD and EspDΔ66–83 expression and secretion. The fluorescence actin staining assay, in which filamentous actin in HeLa cells underlying sites of EPEC A/E activity are detected by fluorescence microscopy, was performed as described (Luo and Donnenberg, 2006) in media containing 25 µM IPTG.

Cleavage of His6-EspD1–171

Partial acid hydrolysis was performed by re-suspending His6-EspD1–171 (2.0 mg) in 300 µl of 44% formic acid and incubating at 42°C for 72 h. The digest mixtures were lyophilized and the peptide pellet dissolved in 8.0 M urea, 10 mM Tris HCl pH 7.5 and applied to a Ni2+-NTA spin column (200 µl packed beads). The C-terminal fragment encompassing residues 108–171 (EspD108–171) was eluted in the column void volume. The hexahistidine tagged N-terminal fragment (His6-EspD1–107) was eluted with 400 mM imidazole. His6-EspD1–107 and EspD108–171 fragments were lyophilized and desalted on a Sephadex G-10 column. Cyanogen bromide (CNBr) cleavage of His6-EspD1–171 was performed by re-suspending 2.0 mg of protein in 150 µl of 70% formic acid containing 2.0 mg of CNBr and incubating the mixture in the dark for 18 h at 20°C. CNBr digests were lyophilized and the peptides dissolved in 100 µl of 50 mM Tris HCl pH 7.5. Protease digests were performed by treating 500 µg of His6-EspD1–171 dissolved in 300 µl of Tris HCl pH 7.5 with 5 µg of porcine trypsin or clostripain for 60 min at 20°C. Proteolysis was terminated by addition of a protease inhibitor cocktail (Roche Applied-Sciences, Laval, QC).

Western blot analysis

EspD specific antiserum was generated by immunizing rabbits with His6-EspD1–171 (Pacific Immunology, Ramona, CA). Proteins were resolved by SDS-PAGE, transferred onto PVDF membrane, blocked with 3% skim milk powder in PBS containing 0.1% Tween-20 (PBST) and probed with anti-His6-EspD1–171 specific rabbit antisera (1:2000). Primary antibodies were detected with goat anti-rabbit HRP-conjugated secondary antibody. Blots were developed with the Perkin Elmer chemiluminescence reagent (Pierce).

Size exclusion chromatography

The quaternary structure of His6-EspD1–171, His6-EspD1–171Δ66–83 and His6-EspD1–107 were examined by SEC on a Beckman Coulter 32 Karat high performance liquid chromatography system equipped with Superose 12 HR10/30 (GE HealthCare) equilibrated with 50 mM sodium phosphate pH 7.5, 150 mM NaCl buffer. Purified protein (10–20 µg) was injected and the column was developed at a flow rate of 0.5 ml min−1. Fractions were collected for Western blot analysis. The SEC column was calibrated with a protein mixture containing thyroglobulin (670 kDa), bovine IgG (160 kDa), ovalbumin (45 kDa), equine myoglobin (17 kDa), ubiquitin (8.3 kDa) and vitamin 12 (1.3 kDa).

Sedimentation equilibrium centrifugation

Analytical ultracentrifugation data were obtained using a Beckman XL-I centrifuge equipped with an An-55 rotor at 20°C with rotor speeds of 20 000, 30 000 and 40 000 r.p.m. in two-sector cells with column heights of 1.2 cm using a peptide concentration of 0.85 mg ml−1 in PBS. Data was analysed using UltraScan 9.9 (ver. 1114) using a theoretical solution density for PBS and a theoretical partial specific volume based on the primary sequence of EspD108–171 (Demeler, 2005).

His6-EspD1–171 binding to HeLa and human red blood cells

HeLa cells were obtained from the American Type Culture Collection (Rockville, MD) and cultured in DMEM media containing 10% heat inactivated fetal bovine serum, 10 U ml−1 penicillin-streptomycin and 100 mM HEPES at 37°C and 5% CO2. HeLa cell cultures grown to 75% confluence were incubated for 3 h at 37°C with 30 µg of His6-EspD1–171 in 1.0 ml of serum free DMEM. Cells were washed with PBS, scraped and re-suspended in 1.0 ml of ice cold PBS containing protease inhibitor cocktail (Roche Applied Sciences, Laval, QC). Cells were lysed with a Dounce homogenizer and a crude membrane fraction was isolated by centrifugation at 14 000 r.p.m. for 30 min at 4°C. Membranes were extracted with 150 µl of PBS containing 500 mM NaCl, 100 mM sodium carbonate pH 11.5 or 1% Triton X-100 for 15 min at 0°C. The membranes were then separated into supernatant and pellet fractions by centrifugation at 45 000 r.p.m. for 30 min at 4°C. Alternatively, HeLa cells (3 × 107) loaded with His6-EspD1–171 were washed with cold PBS to remove unbound protein and then treated with porcine trypsin (25 µg ml−1 PBS) at 4°C. Digests were terminated with a protease inhibitor cocktail. HeLa cells were re-suspended in SDS-PAGE sample buffer and His6-EspD1–171 degradation was analysed by Western blot.

Chemical cross-linking

For cross-linking experiments, HeLa cells loaded with His6-EspD1–171 were treated with 0.2–2.0 mM glutaraldehyde at 4°C for 20 min and the reaction was quenched with 100 mM Tris HCl pH 8.0. His6-EspD1–171, His6-EspD1–171Δ66–83, His6-EspD1–107 and EspD108–171 (10 µg 30 µl−1) in PBS were incubated with 0.8 mM glutaraldehyde for 30 min at 25°C. The reaction was terminated with 4 µl of 1 M Tris-HCl pH 8.0. Cross-linked products were analysed by Western blot using anti-His6-EspD1–171 antisera.

Liposome flotation

Thin films were prepared with dioleoyl phosphatidylcholine (DOPC), dioleoyl phosphatidylserine (DOPS), dioleoyl phosphatidylethanolamine (DOPE), sphingomyelin (SM) and cholesterol (Chol) using the mixtures of DOPC, DOPC:DOPE (2:1), DOPS, DOPC:DOPS (2:1), DOPC:DOPE:Chol or DOPC:SM:Chol (1:1:1) by evaporating the chloroform under nitrogen gas stream and storing samples under vacuum for 24 h to remove residual solvent. Lipid films were rehydrated in 50 mM Tris HCl, 150 mM NaCl, pH 7.5 (TBS) (2 mg lipid ml−1) with vigorous mixing to generate multilamellar vesicles. They were then sonicated to produce SUVs. SUVs were annealed for 1 h at 37°C and multilamellar liposomes were removed by centrifugation at 14 000 r.p.m. for 20 min. SUVs (300 µg) were mixed with His6-EspD1–171 (30 µg) in 90 µl of PBS or PBS supplemented with 500 mM NaCl and then incubated for 45 min at 37°C. Reaction mixtures were made up to 45% sucrose then overlaid with 4 ml of 35% sucrose in PBS and 1 ml of PBS. Flotation was performed by centrifugation at 50 000 r.p.m. for 2 h at 16°C in a Beckman-Coulter SW55 rotor. The gradients were fractionated (750 µl) and proteins were precipitated with 15% trichloroacetic acid and analysed by Coomassie blue stained SDS-PAGE or Western blot.

Circular dichroism

A solution of His6-EspD1–171 (8.0 µM), dialysed against 10 mM sodium phosphate and 10 mM NaCl pH 7.5, was titrated with 0–8.0 mM SDS and CD spectra were recorded from 190 nm to 260 nm. Similarly, His6-EspD1–171 (8.0 µM) was mixed with DOPS, DOPC:SM:Chol or DOPC:DOPS:SM:Chol SUVs at a lipid : protein ratio of 0:1 to 100:1 and was incubated at 25°C for 10 min prior to CD analysis. CD spectra were recorded for SUVs alone to control for background scattering.

For urea denaturation experiments, His6-EspD1–171 (8.0 µM) was diluted in urea solutions (0–8.0 M) and incubated for 16 h at 4°C prior to recording CD spectra from 210 to 250 nm. Spectra were recorded at 23°C on a JASCO 810 instrument using a 1 mm cuvette at a scan rate of 100 nm min−1 and a bandwidth of 1 nm. Five spectra were collected and averaged. Secondary structure content of His6-EspD1–171 was analysed using the CDSSTR algorithm with the reference set SP175 at DichroWeb server (http://dichroweb.cryst.bbk.ac.uk).

Fluorescence spectroscopy

His6-EspD1–171 was diluted in 40 mM sodium phosphate pH 7.5 (1.6 µM) and the fluorescence spectrum was recorded following titration with SDS (0–25 mM). His6-EspD1–171 samples were titrated with phosphate buffer to control for changes associated with dilution. Spectra were recorded from 300 to 420 nm at a scan rate of 60 nm min−1 using an excitation wavelength of 290 nm and slit widths of 5 nm. Fluorescence quenching was examined by titrating His6-EspD1–171 with acrylamide (0–0.25 M) in the absence or presence of 8.0 mM SDS. The interaction of His6-EspD1–171 with lipid bilayers was examined by the addition of SUVs (DOPS, DOPC:DOPS:SM:Chol and DOPC:SM:Chol) to a final lipid : protein ratio of 75:1 and recording the fluorescence spectra from 300 to 420 nm at 30°C. Binding of His6-EspD1–171 to SDS micelles and SUVs was performed in 40 mM sodium phosphate pH 7.5 and 50 mM sodium acetate pH 5.5 buffers. Fluorescence spectra were recorded on a Varian Cary Eclipse spectrofluorometer. For pH titration experiments, a universal buffer containing 40 mM each of formic acid, acetic acid, sodium phosphate and Tris base was prepared and the pH was adjusted with HCl or NaOH. For chemical denaturation experiments, His6-EspD1–171 (20 µM) was diluted in 0–8.0 M urea or 0–6.0 M guanidinium hydrochloride (GuHCl) in PBS and the samples were incubated for 20 h at 4°C prior to spectroscopic analysis. Refolding experiments were performed by diluting a 3.5 mg ml−1 His6-EspD1–171 solution in 8.0 M urea to a final concentration of 0.1 M urea into 50 mM sodium acetate pH 5.5 or buffer containing 20 µg ml−1 of DOPS or DOPC:DOPS:SM:Chol SUVs. Samples were incubated at 25°C for 10 min prior to recording spectra. Fluorescence energy transfer experiments were performed by adding increasing concentrations of ANS to His6-EspD1–171 (20 µM). The spectra were recorded from 300 to 550 nm using an excitation wavelength of 290 nm.

Bioinformatic analysis

Transmembrane domains, coiled-coil motifs and secondary structure prediction were performed using the TMPRED, COILS and the GOR algorithms (http://www.expasy.org). The hydrophobic moment of EspD was calculated using Heliquest (http://heliquest.ipmc.cnrs.fr) and AmphipaSeeK (http://npsa-pbil.ibcp.fr) algorithms. The intrinsic disorder of EspD was predicted using the PONDR software package (Peng et al., 2005).

Acknowledgements

This work was supported by grants from the Natural Science and Engineering Research Council of Canada (to AJ) and Canadian Institutes of Health Research for partial salary support for DD and MR and a Regroupements Strategiques grant from the Fonds de recherche sur nature et les technologies Québec (FQRNT) and Public Health Service Award AI32074 (to MSD) from the National Institutes of Health. We thank Dr Dieter P. Reinhardt of McGill University for access to the spectropolarimeter. Access to PONDR® was provided by Molecular Kinetics (6201 La Pas Trail – Ste 160, Indianapolis, IN 46268; 317-280-8737; E-mail: main@molecularkinetics.com). VL-XT is copyright©1999 by the WSU Research Foundation, all rights reserved. PONDR® is copyright©2004 by Molecular Kinetics, all rights reserved.

Ancillary