|
Advertisement | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
J. Biol. Chem., Vol. 279, Issue 45, 46851-46857, November 5, 2004
High Resolution Studies of the Afa/Dr Adhesin DraE and Its Interaction with Chloramphenicol*![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
From the
Received for publication, August 13, 2004
Pathogenic Escherichia coli expressing Afa/Dr adhesins are able to cause both urinary tract and diarrheal infections. The Afa/Dr adhesins confer adherence to epithelial cells via interactions with the human complement regulating protein, decay accelerating factor (DAF or CD55). Two of the Afa/Dr adhesions, AfaE-III and DraE, differ from each other by only three residues but are reported to have several different properties. One such difference is disruption of the interaction between DraE and CD55 by chloramphenicol, whereas binding of AfaE-III to CD55 is unaffected. Here we present a crystal structure of a strand-swapped trimer of wild type DraE. We also present a crystal structure of this trimer in complex with chloramphenicol, as well as NMR data supporting the binding position of chloramphenicol within the crystal. The crystal structure reveals the precise atomic basis for the sensitivity of DraE-CD55 binding to chloramphenicol and demonstrates that in contrast to other chloramphenicol-protein complexes, drug binding is mediated via recognition of the chlorine "tail" rather than via intercalation of the benzene rings into a hydrophobic pocket.
Diffusely adherent Escherichia coli has recently been accepted as a diarrheagenic strain of E. coli that causes watery diarrhea in young children (13). Diffusely adherent E. coli shares a characteristic pattern of infection of epithelial cells in which the bacteria form a diffuse pattern around the host cells and is accompanied by an elongation of the microvilli (4). Diffusely adherent E. coli is also characterized by the presence of an afa operon (5) that expresses the Afa/Dr adhesion complex on the surface of the bacteria. Although originally thought to be afimbrial, the Afa/Dr adhesins have been shown to form extended fimbrial structures on the surface of bacteria (59). These structures are made up of the E protein, which acts as both the adhesin and the major structural subunit (10, 11), and the D protein, which caps the structure and acts as the invasin (12). They mediate location of bacteria to the gut or urinary tract via adhesion to the mammalian complement regulatory protein, the decay accelerating factor (DAF1 or CD55). DAF is a 70-kDa protein found on the apical surface of host epithelial cells and attaches to the cell surface via a glycophosphotidylinositol anchor (1315). The Afa/Dr adhesins have also been shown to bind to members of the carcinoembryonic antigen family, although the significance of this is yet to be determined (16, 17).
The Afa/Dr adhesins comprise AfaE-I to AfaE-VIII, F1845, DraE, and entropathogenic E. coli Afa (9). Although these adhesins differ in sequence, they all bind to the third CCP domain of DAF with the exception of AfaE-VII and AfaE-VIII (18). Two of the Afa/Dr adhesins, AfaE-III and DraE, differ in sequence by only three amino acids but possess significantly different properties. For example, DraE has been reported to bind to the 7s domain of type IV collagen (19), but AfaE-III does not. The binding of DraE to both DAF and type IV collagen is blocked by the presence of chloramphenicol (2022), whereas the binding of AfaE-III to DAF is unaffected.
Recent structures (11) for an engineered form of the AfaE-III adhesin (AfaE-dsc) have allowed an atomic understanding of its assembly into fimbrae and have also allowed mapping of the sites of CD55 interaction onto the adhesin. Dr family fimbrial assembly is seen to proceed in the same way as bacterial pili assembly (11). Briefly, adhesin subunits are found to consist of an immunoglobulin-like fold that misses a central antiparallel This paper presents x-ray structures for native DraE and AfaE-III in isolation and also of a DraE-chloramphenicol complex and seeks to understand the basis for the differential receptor and inhibitor binding exhibited by the two proteins. These structures reveal a novel mode of chloramphenicol binding that provides potential for the design of a new class of anti-bacterial agents.
Cloning, Expression, and Purification of AfaE-dsc, native DraE, Native AfaE-III, and CD55Native AfaE-III and DraE (strain O75) were expressed separately as N-terminal hexahistidine fusions in E. coli strain M15 (Qiagen) as described previously (11). Briefly, the adhesins were purified from the supernatant using nickel affinity chromatography followed by size exclusion chromatography in an S75-Sepharose column (Amersham Biosciences) to separate the trimeric forms from small amounts of monomeric, dimeric, and aggregated forms of the adhesins. CD55 was purified and refolded from E. coli also as described previously (31). AfaE-dsc, DraE-dsc, and the mutants of the two were expressed from pRSETA plasmid in BL21(DE3) cells. They were purified via an N-terminal hexahistidine tag using nickel-nitrilotriacetic acid superflow resin (Qiagen) as described previously for AfaE-dsc (11, 30).
Chloramphenicol Sensitivity Assayed Using Surface Plasmon ResonanceNative AfaE-III and DraE were covalently coupled to the carboxylated dextran matrix on the surface on an activated CM5 sensor chip using the primary amine coupling kit (BIAcore AB). After activation according to the standard protocol, 0.5 mg/ml of DraE or AfaE-III in 10 mM sodium acetate (pH 4.5) was injected. Differing levels were immobilized ( Crystallization and Structure DeterminationCrystals of AfaE-III and DraE were grown at 21 °C using sitting drop vapor diffusion. AfaE-III was concentrated to 3.5 mg/ml, whereas DraE was concentrated to 1.5 mg/ml. AfaE-III-Se-Met crystals (I432) grew against a reservoir solution containing 2 M ammonium sulfate, 2% polyethylene glycol 400, and 0.1 M sodium HEPES (pH 6.87.0). Native AfaE-III crystals (P3121) grew against a reservoir of 0.2 M magnesium sulfate, 2021% polyethylene glycol 4000, and 0.1 M Tris-HCl (pH 7.8). DraE crystals (P212121) were obtained against a reservoir solution of 2 M ammonium sulfate, 0.1 M Tris-HCl (pH 7). Co-crystals of the DraE-Cm complex (P3) grew against a reservoir of 1.7 M ammonium sulfate, 100 µM Cm (as reported in biological assays (22)) or 2.8 mM Cm and 0.1 M NaHEPES (pH 7). All crystals were flash frozen in nitrogen gas, and data were collected at 100 K with crystals in a fiber loop. Data were collected in-house, at the European Synchrotron Radiation Facility (BM-14), Grenoble, France, and at the Synchrotron Radiation Source (stations 9.6 and 14.2), Daresbury, United Kingdom. The AfaE-III-Se-Met structure (Protein Data Bank accession number 1USZ [PDB] ) was solved at 3.1 Å by single-wavelength anomalous dispersion with the suite of programs autoSHARP (32) and built and refined (as all the other structures in this paper) with the program Xtalview (33) and Buster-TNT (41). The Se-Met model was used to solve the AfaE-III native crystal structure by molecular replacement with the program Molrep (34) (3.3 Å resolution, Protein Data Bank accession number 1UT2 [PDB] ). The native DraE structure (1.7 Å resolution, Protein Data Bank accession number 1UT1 [PDB] ) was also solved with Molrep using Protein Data Bank accession number 1USZ [PDB] as a search model. Finally, the DraE-Cm complex (1.9 Å resolution, Protein Data Bank accession number 1USQ [PDB] ) was again solved by molecular replacement with Molrep using the coordinates of the native DraE as the search model. For the DraE-Cm complex, rebuilding and refinement proceeded with the simple addition of water and remodeling of protein side chains until no difference density could be identified in the Fo Fc map other than that associated with the Cm, at which point the model for the Cm molecule was built. Chemical Shift Mapping15N-labeled NMR samples of AfaE-dsc, DraE-dsc, and mutants were prepared to a final concentration of 1 mM and a volume of 500 µl in 50 mM acetate buffer (pH 5.0), 100 mM sodium chloride, and 10% D2O. Samples were quantified using UV absorbance at 280 nm wavelength. Chloramphenicol was prepared up to a concentration of 7 mM in the same buffer as the protein and solubilized by mild heating. Chloramphenicol was added to each sample in increments described in the text. All experiments were carried out on a 500 MHz four-channel Bruker DRX500 spectrometer equipped with a z-shielded gradient triple resonance cryoprobe. Coordinates and Figure PreparationCoordinates and x-ray crystallographic data for the structures have been deposited at the Protein Data bank under the accession codes 1USQ [PDB] , 1USZ [PDB] , 1UT1 [PDB] and 1UT2 [PDB] . Figures were prepared using program AESOP.2
X-ray Structures for Oligomeric Forms of AfaE-III and DraEWhereas earlier structural work (11) used a form of AfaE-III engineered to be monomeric (AfaE-dsc), heterologous expression of native AfaE-III and DraE sequences in E. coli lacking the Dr fimbrial assembly machinery leads to the production of oligomeric forms of the adhesins. These oligomers have been extensively studied (e.g. Refs. 12, 35, and 36) and have been seen to possess the biological functions of the assembled adhesins (e.g. ability to bind host cell receptors, inhibition profiles, antigenic properties) with the exception that they do not assemble to form fimbrae. Therefore we sought to determine the structures of these oligomeric forms to facilitate comparisons with the structure of AfaE-dsc and interpretation of existing biological data. Size exclusion chromatography (11) has previously shown that the majority of the soluble oligomers are likely to be trimers and once purified, trimeric oligomers of AfaE-III and DraE readily produced crystals suitable for structure determination. The structure of an AfaE-III trimer was solved at a resolution of 3.3 Å using multiple anomalous dispersion methods and a selenomethionine-labeled form of AfaE-III (Table I, Fig. 1a, and "Experimental Procedures"). The structure of DraE was then solved using this model as a starting model for phasing by molecular replacement. In all three differently packed crystal forms have been obtained (two AfaE-III forms and a single DraE form), with the exception of the naturally differing side chains derived from the sequence differences, AfaE-III and DraE possess identical structures (Fig. 1b); the structures superimpose with a root mean square deviation of 0.6 Å over all backbone atoms. Moreover, structural comparison reveals a close superimposition between the average solution structure of AfaE-dsc (11) and the crystal structures of AfaE-III and DraE monomers, i.e. a root mean square deviation of 2.2 Å over the 135 equivalent residues (Fig. 1b).
Despite possessing an identical fold to monomeric AfaE-dsc, the trimer (Fig. 1a) consists of intermolecular disulfide bonded monomers in which the N-terminal donor strand is folded back to form an intramolecular, anti-parallel -strand G. This is facilitated by rearrangement of only five residues in the A1-A2 linker that allows -strand A1 to slot into an identical position in a neighboring subunit (Fig. 1, a and c). We believe this to be an artifactual domain swapping that occurs as a result of cytoplasmic expression in the absence of the chaperone and illustrates the potential problems that can occur in unassisted assembly of fimbrae/pilins. It also represents one of only a few observations describing cyclic domain-swapped trimers (38, 39) and reflects the inherent conformational malleability of these domains. Non-reducing SDS-PAGE (Fig. 1d) confirms this conclusion as analysis of adhesive organelles purified from the surface of Dr+ bacteria could not detect the existence of a disulfide bonded polymer. Despite this misfolding, the atomic structures of the trimeric adhesins differ from the correctly folded form at only five residues (those in the A1-A2 linker) because for the majority of the residues it is simply which subunit they originate from that differs rather than their location within the protein fold being changed. Therefore it is not surprising that earlier biological analyses using oligomeric adhesins found them to be biologically active (12, 35, 36).
X-ray Structure of a DraE-chloramphenicol ComplexThe high degree of structural similarity between AfaE-III and DraE does not facilitate an understanding of their subtly different biology; in particular, the precise structural basis of the differential susceptibility to inhibition of CD55-binding by Cm (22). Therefore we decided to determine the structure of a DraE-Cm complex. High resolution data for the DraE-Cm complex (1.9 Å, Table I) were collected from co-crystals grown in the presence of 2.8 mM Cm (see "Experimental Procedures"). These crystals were in space group P3 and were phased by molecular replacement with the DraE coordinates described above (see "Experimental Procedures"). Calculation of unbiased Fo Fc difference Fourier maps prior to construction of a model for the bound Cm (Fig. 2a) allowed unambiguous positioning of the drug, and the coordinates for the complex were subject to further rounds of refinement to yield the model described in Table I (Fig. 2). Other data sets collected from crystals that were grown in the absence of Cm and then soaked in Cm at a range of concentrations (down to 0.1 mM) prior to data collection showed Cm to be bound in identical fashion (data not shown). The DraE-Cm complex structure reveals that Cm binds in a surface pocket between the N-terminal portion of strand B and the C-terminal portion of strand E and lies within the CD55-binding site identified by chemical shift mapping (11). Cm is anchored at one end by van der Waals contacts of its CHCl2 moiety with the methylene groups of Gly-113 and Gly-42 and Pro-40 and Pro-43 residues and at the other end by stacking of the p-nitro group against the side chain of Ile111 (Fig. 2). The ring of Tyr-115 closes the bottom of this hydrophobic pocket with contacts to the chlorine atoms and additional contacts to the C-
This Cm binding site is seen to be in close proximity to two of the three sequence differences between AfaE-III and DraE (Fig. 3a) and may therefore provide an explanation for the reported lack of activity for Cm against AfaE-III binding of CD55. In the crystal structure of AfaE-III, the Thr88Met mutation puts the side chain of Met-88 into the hydrophobic pocket occupied by Cm in DraE (Fig. 3b), and the Ile111Thr mutation replaces the Ile, which packs against the p-nitro group and neighboring residues crucial for Cm recognition. These structural differences seem sufficient to produce the reported lack of Cm sensitivity for AfaE-III. Carnoy and Moseley (22) have demonstrated previously that the single Ile111Thr mutant completely abolishes Cm binding. Our data suggest that the Thr88Met mutation may also play an important role in the lack of Cm binding to AfaE-III.
To provide additional evidence for the observations seen in the crystal structure of the DraE-Cm complex, we performed an NMR chemical shift mapping experiment on monomeric DraE-dsc in the presence of increasing concentrations of Cm. A number of heteronuclear single quantum coherence peaks shift upon titration with Cm, which is consistent with a weak but specific interaction between Cm and DraE (Fig. 2c). Based on changes in 15N against 1H ppm, significant peak shifts were grouped into large, medium, and small. The largest chemical shift changes were seen in amide peaks corresponding to residues Tyr-115 and Ile-114, both of which are implicated by our crystal structure to be crucial for the DraE-Cm interaction. Furthermore, the medium and small peak shifts map to the region of the DraE-Cm interface (Fig. 2, d and e). These observations were further probed with surface plasmon resonance experiments to examine CD55 binding by the adhesins in the presence of Cm (Fig. 3c). These studies confirm earlier reports in that CD55 binding by DraE is seen to be Cm sensitive with 50% inhibition seen to occur with 2.8 mM Cm, whereas AfaE-III binding to CD55 is not significantly altered by Cm at this concentration. These results confirm that the subtle alterations seen between the AfaE-III and DraE are indeed sufficient to ablate Cm sensitivity. The structure of the DraE-Cm complex may provide the start point for the design of novel antimicrobial compounds because, in contrast to earlier structures for Cm-ligand complexes (37, 40) where binding of Cm involves burial of the benzene ring, the interactions seen here are focused on burial of the chlorines. This observation may allow design of novel therapeutics that maintain the geometry of the adhesin-Cm interaction without the nitro-benzene ring and hence avoid the toxic side effects associated with therapeutic use of Cm caused by intercalation into the human ribosome. Because the lack of sensitivity to Cm of the AfaE-III is seen to be dependent on covering of the Cm site by a single side chain we speculate that agents designed to bind with higher affinity would be able to displace this side chain and so achieve more wide ranging action against this whole class of bacterial adhesins.
* This work is supported by The Wellcome Trust research leave award (to S. M.), British Biotechnology Science Research Council Grant 43/B16601 (to S. M. L.), Arthritis Research Campaign Grant L0534 (to S. M. L.), Medical Research Council studentship (to D. P.), and an Engineering and Physical Sciences Research Council studentship (to K .L. A.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The atomic coordinates and structure factors (codes 1USQ
[PDB]
, 1USZ
[PDB]
, 1UT1
[PDB]
, and 1UT2
[PDB]
) have been deposited in the Protein Data Bank, Research Collaboratory for Structural Bioinformatics, Rutgers University, New Brunswick, NJ (http://www.rcsb.org/).
1 The abbreviations used are: DAF, decay accelerating factor; Cm, chloramphenicol.
2 M. Noble, personal communication.
We thank Martin Walsh at BM14, European Synchrotron Radiation Facility and the staff of stations 9.6 and 14.2 at the Synchrotron Radiation Source, Daresbury for assistance with data collection.
This article has been cited by other articles:
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Advertisement | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||