Biological context

Enteroaggregative Escherichia coli (EAEC) are a clinically important cause of diarrhea in both developing and industrialized countries. Infection can lead to watery diarrhea that is often persistent and inflammatory (Okeke and Nataro 2001). AAF-mediated attachment to host epithelium is a required first step in the infection process. The four known AAF alleles, AAF/I through AAF/IV share little homology and are antigenically distinct, yet all four have been implicated in mediating the same “stacked brick” pattern of HEp-2 cell adherence that defines the enteroaggregative pathotype. Studies of AAF/II mutants suggest that they have important roles in pathogenesis including induction of IL-8 release, disruption of tight junctions, and binding to extracellular matrix proteins (Farfan et al. 2008; Harrington et al. 2005). The major structural subunit (or pilin) of AAF/II is the protein AafA (GenBank: AAB82330.1), which is also the principal adhesin of EAEC strains expressing this allele. AafA has been shown to bind fibronectin, laminin, and type IV collagen in vitro but not BSA or type I collagen. Evidence suggests fibronectin is the primary receptor for EAEC adherence to cultured colonic epithelial cells (Farfan et al. 2008).

AAF are distantly related to the Afa/Dr adhesins, which are expressed by a subset of uropathogenic E. coli and diffusely adherent E. coli (Boisen et al. 2008). AafA exhibits 44.0% similarity and 14.8% identity to AfaE, the major pilin from AFA-III diffusely adherent E. coli that assembles into a flexible fimbrial structure. Despite the low identity between AafA and AfaE-III, significant sequence conservation is found among the chaperone and usher accessory proteins, as well as in the overall structure of the gene cluster. Due to theses similarities, the two systems are generally expected to possess similar macrostructure (Anderson et al. 2004a). Under this model, AAF are comprised of three subunit types: an usher, AafC that anchors the fibril to the outer membrane; a major subunit, AafA that polymerizes to form the fibril; and a minor subunit or tip antigen, AafB, found at its distal end. As with chaperone-usher systems generally, each AafA subunit should form an immunoglobulin fold that lacks the C-terminal or G-strand. Instead, that strand is found at the N-terminus of the mature subunit where it cannot fold into the G-strand cleft but instead occupies that cleft on an adjacent subunit, forming the basis for polymerization and fiber formation.

To make monomeric AafA for solution study, we have chosen an approach that has been successful for AfaE-III and other chaperone-usher systems (Anderson et al. 2004a; Choudhury et al. 1999). Specifically, we moved the N-terminal donor strand of AafA to the C-terminus along with and following an artificial turn and positioning linker that allowed it to complete the fold of the same subunit. The resulting monomeric “donor-strand-complemented” AafA, or AafA-dsc yields monodisperse multinuclear NMR spectra with broad chemical shift dispersion as expected for a well-folded protein (Fig. 1). Moreover, AafA-dsc retains the principal adhesive properties of the intact fimbriae, binding fibronectin, laminin, and type IV collagen but not BSA or type I collagen, and blocks bacterial binding either to plates coated with those proteins or to cultured colonic epithelial cells, demonstrating its relevance to in vitro studies of EAEC adherence (Farfan et al. 2008). Structural studies of AafA-dsc are underway, in support of our overall goal to develop anti-adhesive interventions that prevent or alleviate EAEC infection. We report here the 1H, 15N, and 13C resonance assignments for AafA-dsc as a first major step toward determining its structure.

Fig. 1
figure 1

Assigned 2D 1H-15N HSQC spectrum of AafA-dsc recorded with 4 transients at 500 MHz using 15N, 13C-labeled AafA-dsc at 308 K. Assigned residues that are too weak to be visible at this level are indicated with arrowheads

Full size image

Methods and experiments

Protein expression and purification

The design of AafA-dsc was based on a successful approach used for the related AfaE-III (Anderson et al. 2004b). The first 10 N-terminal amino acids of the native AafA (ktatstitvv) were removed and inserted at the C-terminus, downstream of a turn (dnkq) and six additional residues from the uncleaved AafA N-terminus (natava), added to position the donor strand and allow autonomous folding of the subunit. A synthetic aafA-dsc gene was inserted into plasmid pQE-30 as an N-terminal hexahistidine-tagged fusion which was then expressed in E. coli strain M15[pREP4] (Qiagen, Gaithersburg, MD). Cultures were grown in either lysogeny broth or M9 minimal medium containing either 15NH4Cl or both that and 13C-glucose (Cambridge Isotope Laboratories, Andover, MA). Media were supplemented with 50 μg·mL-1 carbenicillin or 50–100 μg·mL-1 ampicillin. Cultures were induced with 1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) at 37°C for 4–6 h, after which cells were collected by centrifugation at 6,000×g for 20 min, 4°C. Decanted pellets were frozen at −20°C. Cell pellets from 1 L media were resuspended in 20 mL denaturing buffer (50 mM sodium phosphate, pH 8.0, 300 mM NaCl, 8 M urea) and lysed by French Press (SLM Instruments, Urbana, IL). The fusion protein was purified by histidine affinity chromatography using NTA-agarose resin loaded with Ni+2 (Qiagen) under denaturing conditions. The purified protein was dialysed into phosphate buffer containing 1 M urea (50 mM sodium phosphate, pH 8.0, 300 mM NaCl, 1 M urea) and then into 50 mM acetate buffer with 50 mM NaCl at pH 5.0. Refolded protein was concentrated to 0.5 mM for the NMR experiments.

NMR spectroscopy

NMR samples were prepared in 90% H2O/10% D2O at pH 5.0. Spectra were recorded at 308 K on Bruker DRX500, DRX600, Avance II 800 MHz, and Avance II 950 MHz spectrometers equipped with triple resonance, z-axis pulsed-field gradient cryoprobes. The following experiments were performed (reviewed in Sattler et al. (1999) unless otherwise noted): CBCA(CO)NH, HNCACB, HNCO, HN(CA)CO, (H)CCH-TOCSY, H(C)CH-TOCSY, HBHA(CBCACO)NH (Grzesiek and Bax 1993), (H)CCCONH (Celda and Montelione 1993), 15N-edited NOESY-HSQC and 13C-edited HMQC-NOESY. Assignments were extended by resolving overlap using 4D 13C/15N-edited HMQC-NOESY-HSQC data collected at 600 MHz, and 4D 13C/13C-edited HMQC-NOESY-HMQC data collected at 800 MHz on a sample prepared in 100% D2O. Pulse programs were written in-house or downloaded from the pulse program library at the National Magnetic Resonance Facility at Madison (NMRFAM) or provided by the spectrometer manufacturer. The data were processed with NMRPipe (Delaglio et al. 1995) and analyzed with NMRView (Johnson and Blevins 1994) or Sparky (T. D. Goddard and D. G. Kneller, SPARKY 3, Univ. of California, San Francisco, unpublished source).

Extent of assignment and data deposition

Connectivities between residues were established using MARS in an iterative manner, adding or adjusting assignments manually (Jung and Zweckstetter 2004). Figure 1 shows the 1H-15N HSQC spectrum of AafA-dsc with assignments labeled. The backbone chemical shifts of two non-proline residues excluding the N terminal affinity tag were not observable in triple resonance spectra, namely Q52 and R110. Furthermore, eight residues appear very weakly on 15N-HSQC spectra, namely I5, D18, C35, A36, M40, A44, T132, and A133, indicated by arrowheads on the figure. Chemical exchange is suspected in regions of missing assignments or weak signal; experiments are currently underway to explore this possibility and the time-scales on which it may be occurring. Assignments were determined for 98% of each of backbone 13C, non-proline backbone 15N, and backbone amide 1H atoms. Sidechain assignments were determined for 95% of 1H atoms attached to carbon, and 83% of 13C atoms (96% excluding those not typically assigned, namely Arg Cζ, Asp Cγ, Asn Cγ, Glu Cδ, Gln Cδ, His Cγ, Phe Cγ, Trp Cγ, Cδ2, Cε2, and Tyr Cγ and Cζ). Chemical shift values have been deposited in the Biological Magnetic Resonance Bank database (http://www.bmrb.wisc.edu/) under accession number 16748.

Secondary structure

Secondary structure was predicted on the basis of chemical shift index (CSI) using TALOS and from sequential NOE correlations observed in 3D and 4D NOE data. The CSI shows that AafA-dsc folds to establish 12 β-strands, consistent with the solution structure of AfaE (Anderson et al. 2004b; Wishart and Sykes 1994). Although less than 20% homology is shared between AafA-dsc and AfaE-dsc, their secondary structure is similar and comprised of β-sheets with similar registries and NOE correlations. Inter-strand NOEs between the N-terminal donor strand and the native C-terminus provide good evidence that the structure is stably folded via donor strand complementation. Differences between the two secondary structures include the absence in AafA-dsc of a short 3–10 helix that is seen in AfaE-dsc, which may yet become evident in the final structure (Fig. 2).

Fig. 2
figure 2

Secondary structure predictions for AafA-dsc. Chemical shift deviations from random coil values are shown for 1Hα, 13Cα, 13Cβ, and backbone carbonyl 13C. Arrows indicate probable β-strands based on TALOS predictions and NOE correlations. Residues that were introduced between the native C-terminus and the displaced donor strand are underlined. Deviations larger than ranges shown are indicated with an asterisk (*). Unassigned nuclei are marked with a dagger (†)

Full size image