The molecular basis of phosphite and hypophosphite recognition by ABC-transporters

Abstract

Inorganic phosphate is the major bioavailable form of the essential nutrient phosphorus. However, the concentration of phosphate in most natural habitats is low enough to limit microbial growth. Under phosphate-depleted conditions some bacteria utilise phosphite and hypophosphite as alternative sources of phosphorus, but the molecular basis of reduced phosphorus acquisition from the environment is not fully understood. Here, we present crystal structures and ligand binding affinities of periplasmic binding proteins from bacterial phosphite and hypophosphite ATP-binding cassette transporters. We reveal that phosphite and hypophosphite specificity results from a combination of steric selection and the presence of a P-H…π interaction between the ligand and a conserved aromatic residue in the ligand-binding pocket. The characterisation of high affinity and specific transporters has implications for the marine phosphorus redox cycle, and might aid the use of phosphite as an alternative phosphorus source in biotechnological, industrial and agricultural applications.

Fig. 1

The crystal structures of Te_PtxB and Pm_PhnD in complex with phosphite and HtxB in complex with hypophosphite. a–c A cartoon representation of the overall fold of Te_PtxB (grey, a), Pm_PhnD (blue, b) and HtxB (green, c), with the ligand drawn as spheres and coloured in atom colours (N = blue, P = orange, O = red, S = yellow, H = white). The termini are labelled. See also domain topology diagram in Supplementary Fig. 6. d–f A detailed view of the ligand-binding pocket, highlighting the interactions made between the ligand and the protein for Te_PtxB (d), Pm_PhnD (e) and HtxB (f). The protein backbones are drawn as a cartoon, with the bound ligand and surrounding sidechains drawn as sticks, and coloured as in parts (a–c). The single buried water molecule (HOH) is drawn as a red sphere. Hydrogen bonds are drawn as dashed lines and the P-H…π interaction is drawn as an orange dashed line. g A superposition of Te_PtxB and Pm_PhnD, viewed from above the phosphite binding site, shows how exchanging Y100 in Te_PtxB for a phenylalanine (F90) in Pm_PhnD requires D205 to be recruited to complete the network of hydrogen bonds around the phosphite ligand. The phosphite is drawn as partially transparent spheres and the buried water molecule is indicated as a red sphere. h A comparison of the binding pockets of Te_PtxB and HtxB showing how the interactions around two of the P-oxygen binding sites are largely spatially conserved, but the third site is modified in HtxB (H160 to F158) to essentially block binding of a third oxygen moiety. The hydrogen atoms of the ligand are highlighted in green for HtxB and in white for Te_PtxB, with most of the main chain atoms omitted for clarity

Fig. 2

A detailed comparison of the spatial arrangement of the binding pocket and network of interactions around the ligand. The interior surface (partially transparent, atom colours) of the binding pocket and surrounding residues (sticks) in a Te_PtxB; b Pm_PhnD; c HtxB and d E. coli PhnD (PDB:3P7I). In each case, the capping residues are highlighted with a dotted surface that represents the van der Waals radii of the atoms. The external surface is also shown in each case, demonstrating how the binding pocket in reduced phosphorus binding PBPs (a–c) is buried from the external solvent and in each case is much smaller than that of the phosphonate binding PhnD from E. coli (d). The protein backbone is drawn as a cartoon, with sidechains drawn in sticks and hydrogen bonds between the capping residues indicated by black dashes. Each figure (a–d) is accompanied by a schematic (e-h) that displays the hydrogen-bonding network between the protein and the oxygen atoms of the ligand. The 2-aminoethyl group of 2AEPn is abbreviated to 2AE in h

Fig. 3

A comparison of phosphite and methylphosphonate binding in Te_PtxB and Pm_PhnD. a Superposition of the complexes of Te_PtxB (grey) with phosphite (white hydrogen) and methylphosphonate (MPn; yellow methyl) showing that the residues around the binding pocket are in consistent positions in each structure apart from the capping tyrosine which moves by ~0.5 Å in the MPn complex (Y208, yellow) in response to the larger van der Waals radii of the methyl group of the ligand. b Superposition of the phosphite (white hydrogen) and MPn (cyan methyl) complexes with Pm_PhnD (blue) showing a similar movement of the capping tyrosine (Y206, cyan)

Fig. 4

The conformational change between the open and closed structures of Ps_PtxB. a, b A comparison of the open (purple) and closed (beige) structure of Ps_PtxB, superimposed on lobe 2, showing the 60° rotation of one domain relative to the other around an axis that lies between the two domains (red arrow). c Superimposing the open and closed structures on lobe 2 (lobe 1 is shown in grey) shows that the spatial arrangement of Y49 and Y203 is conserved in both the open and closed states, whilst Y94 and D16, which both sit on flexible loop regions, undergo large conformational movements (blue arrows) of 7 and 4.5 Å (cα- cα), respectively. The phosphite is shown as spheres

Fig. 5

Protein phylogeny and sequence alignments of reduced phosphorus compound PBPs. a Unrooted phylogenetic tree of reduced P compound transporter PBP amino-acid sequences. Phosphite binding PtxB (grey) and PhnD (purple) homologues cluster separately from each other and from C-P lyase-linked phosphonate binding PhnDs (yellow), HtxB homologues (green) and a putative ‘hybrid’ class of transporters (red). Proteins characterised in this study are marked with an asterisk and the scale bar represents the number of substitutions per site. b Sequence alignment of reduced phosphorus compound binding PBPs, coloured as in a. 2-Aminoethlyphosphonate-specific cap residues are highlighted yellow, phosphite-specific cap residues are highlighted grey (PtxB) or purple (PhnD) and hypophosphite-specific cap residues are highlighted green. Conserved residues that form H-bonds with ligand oxygen atoms are highlighted in blue. The position in the alignment is shown above the sequence and residue numbering for each protein at row ends. For residue numbering, the red (or orange, see below) value corresponds to the residue numbering of the structures, as the predicted N-terminal signal peptides were not present in recombinant proteins, and the black number in parentheses is the residue number of the unprocessed proteins. The second residue (following the initiator Met1) in the recombinant proteins used in this study is shown in red and bold in the sequence; in the case of proteins not directly studied here, the equivalent residues/residue numbers are shown in orange. For full details of the proteins included in this figure see Supplementary Table 3

Fig. 6

The volume of the binding pocket in Pm_PtxB is ~25% bigger than that of Te_PtxB and Pm_PhnD. Superposition of the Pm_PtxB/phosphite structure (pink) with those of Te_PtxB (grey) and Pm_PhnD (blue) (a) shows how a sequence difference (black arrow) in the binding pocket increases the volume of the cavity in Pm_PtxB. In Te_PtxB (b) and Pm_PhnD (c) Leu20 and Ile11 (respectively) provide a hydrophobic surface for the ligand to pack against within the binding pocket. In Pm_PtxB (d) Ile20 adopts an alternative rotamer making the hydrophobic region of the pocket slightly larger, as indicated by the asterisk (*)