Drug efflux proteins are widespread amongst microorganisms, including pathogens. They can contribute to both natural insensitivity to antibiotics and to emerging antibiotic resistance and so are potential targets for the development of new antibacterial drugs. The design of such drugs would be greatly facilitated by knowledge of the structures of these transport proteins, which are poorly understood, because of the difficulties of obtaining crystals of quality. We describe a structural genomics approach for the amplified expression, purification and characterisation of prokaryotic drug efflux proteins of the 'Major Facilitator Superfamily' (MFS) of transport proteins from Helicobacter pylori, Staphylococcus aureus, Escherichia coli, Enterococcus faecalis, Bacillus subtilis, Brucella melitensis, Campylobacter jejuni, Neisseria meningitides and Streptomyces coelicolor. The H. pylori putative drug resistance protein, HP1092, and the S. aureus QacA proteins are used as detailed examples. This strategy is an important step towards reproducible production of transport proteins for the screening of drug binding and for optimisation of crystallisation conditions to enable subsequent structure determination.

No abstract

No abstract

A series of 46 natural nucleosides and analogues (mainly adenosine-based) were tested as inhibitors of [U-(14)C]uridine uptake by the concentrative, H(+)-linked nucleoside transport proteins NupC and NupG from Escherichia coli. The two evolutionarily unrelated transporters showed similar but distinct patterns of inhibition, revealing differing selectivities for the different nucleosides and their analogues. Binding of nucleosides to NupG required the presence of hydroxyl groups at each of the C-3' and C-5' positions of ribose, while binding to NupC required only the C-3' hydroxyl substituent. The greater importance of the ribose moiety for binding to NupG is consistent with the evolutionary relationship between this protein and the oligosaccharide: H(+) symporter (OHS) subfamily of the major facilitator superfamily (MFS) of transporters. For both proteins the natural alpha-configuration at C-3' and the natural beta-configuration at C-1' was mandatory for ligand binding. N-7 in the imidazole ring of adenosine and the amino group at C-6 were found not to be important for binding and both transporters showed flexibility for substitution at C-6/N(6); one or both of N-1 and N-3 were important for adenosine analogue binding to NupC but significantly less so for binding to NupG. From the different effects of 8-bromoadenosine on the two transporters it appears that adenosine selectively binds to NupC in an anti- rather than a syn-conformation, whereas NupG is less prescriptive. The pattern of inhibition of NupC by differing nucleoside analogues confirmed the functional relationship of the bacterial transporter to members of the human concentrative nucleoside transporter (CNT) family and reaffirmed the use of the bacterial protein as an experimental model for these physiologically and clinically important mammalian proteins. The specificity data for NupG have been used to develop a homology model of the protein's binding site, based on the X-ray crystallographic structure of the disaccharide transporter LacY from E. coli. We have also developed an efficient general protocol for the synthesis of adenosine and three of its analogues, which is illustrated by the synthesis of [1'-(13)C]adenosine.

We have devised methods in which cross-polarization magic-angle spinning (CP-MAS) solid-state NMR is exploited to measure rigorous parameters for binding of (13)C-labeled substrates to membrane transport proteins. The methods were applied to two proteins from Escherichia coli: a nucleoside transporter, NupC, and a glucuronide transporter, GusB. A substantial signal for the binding of methyl [1-(13)C]-beta-d-glucuronide to GusB overexpressed in native membranes was achieved with a sample that contained as little as 20 nmol of GusB protein. The data were fitted to yield a K(D) value of 4.17 mM for the labeled ligand and 0.42 mM for an unlabeled ligand, p-nitrophenyl beta-d-glucuronide, which displaced the labeled compound. CP-MAS was also used to measure binding of [1'-(13)C]uridine to overexpressed NupC. The spectrum of NupC-enriched membranes containing [1'-(13)C]uridine exhibited a large peak from substrate bound to undefined sites other than the transport site, which obscured the signal from substrate bound to NupC. In a novel application of a cross-polarization/polarization-inversion (CPPI) NMR experiment, the signal from undefined binding was eliminated by use of appropriate inversion pulse lengths. By use of CPPI in a titration experiment, a K(D) value of 2.6 mM was determined for uridine bound to NupC. These approaches are broadly applicable to quantifying binding of substrates, inhibitors, drugs, and antibiotics to numerous membrane proteins.

NupG from Escherichia coli is the archetype of a family of nucleoside transporters found in several eubacterial groups and has distant homologues in eukaryotes, including man. To facilitate investigation of its molecular mechanism, we developed methods for expressing an oligohistidine-tagged form of NupG both at high levels (>20% of the inner membrane protein) in E. coli and in Xenopus laevis oocytes. In E. coli recombinant NupG transported purine (adenosine) and pyrimidine (uridine) nucleosides with apparent K(m) values of approximately 20-30 microM and transport was energized primarily by the membrane potential component of the proton motive force. Competition experiments in E. coli and measurements of uptake in oocytes confirmed that NupG was a broad-specificity transporter of purine and pyrimidine nucleosides. Importantly, using high-level expression in E. coli and magic-angle spinning cross-polarization solid-state nuclear magnetic resonance, we have for the first time been able directly to measure the binding of the permeant ([1'-(13)C]uridine) to the protein and to assess its relative mobility within the binding site, under non-energized conditions. Purification of over-expressed NupG to near homogeneity by metal chelate affinity chromatography, with retention of transport function in reconstitution assays, was also achieved. Fourier transform infrared and circular dichroism spectroscopy provided further evidence that the purified protein retained its 3D conformation and was predominantly alpha-helical in nature, consistent with a proposed structure containing 12 transmembrane helices. These findings open the way to elucidating the molecular mechanism of transport in this key family of membrane transporters.

No abstract

The binding of the transport inhibitor forskolin, synthetically labelled with (13)C, to the galactose-H(+) symport protein GalP, overexpressed in its native inner membranes from Escherichia coli, was studied using cross-polarization magic angle spinning (13)C NMR. (13)C-Labelled D-galactose and D-glucose were displaced from GalP with the singly labelled [7-OCO(13)CH(3)]forskolin and were not bound to any alternative site within the protein, demonstrating that any multiple sugar binding sites are not simultaneously accessible to these sugars and the inhibitor within GalP. The observation of singly (13)C-labelled forskolin was hampered by interference from natural abundance (13)C in the membranes and so the effectiveness of double-quantum filtration was assessed for the exclusive detection of (13)C spin pairs in sugar (D-[1,2-(13)C(2)]glucose) and inhibitor ([7-O(13)CO(13)CH(3)]forskolin) bound to the GalP protein. The solid state NMR methodology was not effective in creating double-quantum selection of ligand bound with membranes in the 'fluid' state (approx. 2 degrees C) but could be applied in a straightforward way to systems that were kept frozen. At -35 degrees C, double-quantum filtration detected unbound sugar that was incorporated into ice structure within the sample, and was not distinguished from protein-bound sugar. However, the method detected doubly labelled forskolin that is selectively bound only to the transport system under these conditions and provided very effective suppression of interference from natural abundance (13)C background. These results indicate that solid state NMR methods can be used to resolve selectively the interactions of more hydrophobic ligands in the binding sites of target proteins.