|Home | About | Journals | Submit | Contact Us | Français|
The ‘Nucleobase-Cation-Symport-1’, NCS1, transporters are essential components of salvage pathways for nucleobases and related metabolites. Here, we report the 2.85 Å resolution structure of the NCS1 benzyl-hydantoin transporter, Mhp1, from Microbacterium liquefaciens. Mhp1 contains 12 transmembrane helices, ten of which are arranged in two inverted repeats of 5 helices. The structures of the outward-facing open and substrate-bound occluded conformations were solved showing how the outward-facing cavity closes upon binding of substrate. Comparisons with the leucine (LeuTAa) and the galactose (vSGLT) transporters reveal that the outward- and inward-facing cavities are symmetrically arranged on opposite sides of the membrane. The reciprocal opening and closing of these cavities is synchronised by the inverted repeat helices 3 and 8, providing the structural basis of the ‘alternating access’ model for membrane transport.
Many membrane transporters are classified into three major groups. One group, the ‘primary’ active transporters, uses the energy released from light, redox reactions, or ATP hydrolysis to translocate substrates across the membrane. Another group, the ‘secondary’ active transporters, uses the free energy stored in an ion gradient for substrate transport. A third group carries out facilitated diffusion without energy input. The kinetics and thermodynamics of all types of transporters can, in principle, be explained by the “alternating access” model of molecular transport (1, 2). According to this model, a substrate-binding site located towards the centre of the protein in the membrane has alternating access to either side of the membrane as a result of reciprocal opening and closing of cavities connecting the binding site to either side of the membrane. This model is well studied and established for various transporters using kinetic and biochemical methods (3, 4). For the P-type ATPases and the ABC transporters, the mechanism has also been studied based on the structures of these proteins in various conformational states (5, 6). Secondary transporters are biochemically well-characterised, particularly lactose permease (7-9) and other members of the Major Facilitator Superfamily (MFS) transporters (10, 11), but here the structural basis of the “alternating access” mechanism is less well understood.
Here we show how structural studies of the secondary active membrane transporter, Mhp1, from Microbacterium liquefaciens provide insight into the mechanism of alternating access. Mhp1 mediates the uptake of indolyl methyl- and benzyl-hydantoins into M. liquefaciens, as part of a metabolic salvage pathway for their conversion to amino acids (12). Mhp1 is a member of the so-called nucleobase-cation-symport, ‘NCS1’, family 2.A.39 (13, 14) of transport proteins, which has at least 800 known homologues in eubacteria, archaea, fungi and plants, according to the UNIPROT database. Known substrates for the other NCS1 sub-family transporters include allantoin, uracil, cytosine (including the antifungals, 5-fluorocytosine and 5-fluorouracil), purines, thiamine, pyridoxal-based compounds and nicotinamide riboside (http://www.membranetransport.org/) (15, 16). The X-ray structure of the Mhp1 protein described in this paper revealed similarities of this cation-coupled transporter to the leucine transport protein LeuTAa (17-19), a member of the neurotransmitter-sodium-symporter family, ‘NSS’ 2.A.22, (13, 16, 20) and to the galactose transport protein vSGLT (21), a member of the solute-sodium-symporter family, ‘SSS’ 2.A.21. (13, 16, 22). Despite this structural similarity, the amino acid sequence of Mhp1 exhibits only an insignificant 15% identity to LeuTAa and 16% to vSGLT, as calculated by the LALIGN algorithm (http://www.ch.embnet.org/software/LALIGN_form.html) (23).
The two X-ray structures of Mhp1, in the outward-facing open conformation and in the outward-facing occluded conformation with benzyl-hydantoin, demonstrate–the conformational change consequent upon the binding of substrate from the outside of the membrane. Comparison of these Mhp1 structures with those of LeuTAa and vSGLT suggests a possible further conformational change, which allows the release of the substrate into the inside of the membrane. The inward- and outward- facing cavities of the transporters are symmetrically arranged on the opposite sides of the membrane using two inverted repeated segments related by an internal pseudo-two-fold axis parallel to the membrane. This comparative study suggestes how the reciprocal opening and closing of inward-facing and outward-facing cavities could be synchronised by helices 3 and 8 that connect the two cavities on the opposite sides of the membrane.
For details, see Materials and Methods (24). The structure of Mhp1 without substrate was solved using multiple isomorphous replacement in combination with 2-fold cross-crystal averaging (25). The model was refined against data extending to a resolution of 2.85Å with an R-factor of 24.0% and a corresponding R-free of 28.1% (Table1, table S1 and fig. S1) (26). The structure of the complex of Mhp1 with benzyl-hydantoin was solved from crystals grown in the presence of the substrate. Despite the limited resolution (4Å), difference maps clearly showed that the substrate is present in the binding site and there to be a structural rearrangement of residues 355-370. To avoid over-refinement, we only re-modeled this region followed by minimization with strict harmonic restraint (24). This partially remodelled structure gave an R-factor of 34.2 % and a corresponding R-free of 39.2 % (24).
The structure is comprised of 12 membrane-spanning helices as was predicted (Fig. 1A) (12), although previous assignment of transmembrane helices (TMs) based on homologous yeast NCS-1 transporters needed a slight revision (fig. S2) (15). The 12 transmembrane helices (TMs) are arranged in two repeating units (TMs 1-5, residues 20 to 190 and TMs 6-10, residues 204 to 388) connected by a 29-residue loop and followed by an additional two transmembrane helices (Fig. 1, A and B). The similarity of the two repeating units is such that 65 out of a possible 170 Cα pairs can be superimposed with an rms deviation of 2.5Å (24), despite the absence of significant sequence homology between the two units. The two units show an opposite topology with respect to the membrane and are related to each other by a rotation of 168° around an axis in the centre of the membrane and parallel to its plane. An inverted topology repeat is commonly observed for many membrane transporters and channels (17, 21, 27-31). The two repeat units are completely intertwined giving a central 4-helix bundle consisting of the 2 broken TMs, 1 and 6 associated with TMs 2 and 7 (Fig. 1). This bundle is coated, on the side of TMs 1 and 6, by a layer formed by the other 6 helices; TMs 3 and 8, facing directly towards TMs 1 and 6, form a long anti-parallel unit threading through two V-shaped structures formed by TMs 4 and 5 and TMs 9 and 10, respectively (Fig. 1A). The substrate- and cation-binding sites and the outward-facing cavity connecting these sites to the outsides of the membrane are all located in the space between the central four-helix-bundle and the outer helix layer (Fig. 1 and Fig. 2).
The amino acid sequence for TMs 11 and 12 and the C-terminal extension is poorly conserved among the NCS1 family (fig. S2) and the structural role of this region is unclear.
Although from the respective sequences it was not obvious, the fold of the core 10 helices of Mhp1 (TMs 1–10) is reminiscent of those of LeuTAa (TMs 1-10) (17) and the recently solved structure of vSGLT (TMs 2-11) (21) (fig. S3). Overall 249 out of a possible 460 Cα pairs between LeuTAa and Mhp1 can be superimposed with an rms deviation of 2.4Å (24). Using the same algorithm to superpose Mhp1 and vSGLT, 172 Cα atom pairs have an rms deviation of 2.2Å (fig. S4).
In the electron density maps of the Mhp1 crystals with benzyl-hydantoin (Fig. 3, A and B), a V-shaped density was clearly observed at a position almost identical to that of the leucine in the LeuTAa structure and close to that of the galactose in vSGLT (fig. S5). This site is located at the breaks in the discontinuous TMs 1 and 6, and facing TMs 3 and 8 (Fig. 1). It is located at the foot of the outward-facing cavity, which is composed of the neighbouring surfaces of TMs 1, 3, 6, 8 and 10 and allows access of the substrate to the binding site (Fig. 1 and Fig. 2A). The structure of l-5-benzyl-hydantoin taken from the Cambridge CSD (accession code IWEYAK) was consistent with the shape of this density and the molecule was placed between Trp117 (TM3) and Trp 220 (TM6) without requiring any modifications of torsion angles (Fig. 3, A and B).
In the current model, the hydantoin moiety forms a pi-stacking interaction with the indole-ring of Trp 117 and is within hydrogen bonding distance of Asn 318 and Gln 121 (Fig. 3A). Trp 117 and Asn 318 are conserved amongst all the transporters in the family and Gln 121 only varies in the uridine transporter, FUI 1 (fig. S2). Another conserved residue, Asn 314, is within hydrogen bonding distance of Asn 318 such that it may hold the asparagine side chain in position to interact with the substrate. The benzyl group of the substrate is situated next to Trp 220 and Gln 42. The side chain of Trp 220 moves into the binding site with respect to its position in the substrate-free structure and forms a pi-stacking interaction with the benzyl moiety.
This binding mode is consistent with the observation that Mhp1 has a higher affinity for 5-indolyl-methyl-hydantoin than for benzyl-hydantoin (12), since the indole group in the hydantoin would form an even more extensive packing interaction with Trp 220 and, in addition, the side chain of Gln 42 could form a hydrogen bond with the nitrogen atom of the indole rings. This residue could play a role in the substrate specificity of the NCS1 transporters as suggested by sequence analysis (15).
The observed residues in the substrate-binding site are consistent with the results of mutational studies on a NCS1 family member, Fcy2, that transports cytosine into Saccharomyces cerevisae (32-35). Although Fcy2 is a distantly related homologue of Mhp1, the residues involved in the substrate and cation binding can be aligned with Mhp1 unambiguously (fig. S2). Three of the genetically selected Fcy2 mutants, which show an altered Km of substrate uptake, were substitutions in the segment 371 I-A-N-N-I-P-N 377 of Fcy2, which corresponds to the residues 311 - 318 of Mhp1 (32-34). Site-directed mutagenesis studies on these residues emphasized the role in the substrate binding of Asn374 and Asn377, which are equivalent to Asn314 and Asn318 of Mhp1, respectively (35).
In the benzyl-hydantoin complex structure, some conformational differences from the substrate-free Mhp1 structure are evident (Fig. 3, C and D). The N-terminal part of TM10 (Residues 355-370) moves into the outward-facing cavity. This occludes the substrate binding site from the outside space of the membrane (Fig. 2B and Fig. 3C). This movement seems to be triggered by a repositioning of Trp 220 located on TM6, which is adjacent to TM10.
We, therefore, have two conformations of the protein. We refer to the substrate-free structure as outward-facing open and to the substrate bound structure as outward-facing occluded.
Occluding the substrate binding-site from the outside of the membrane is essential to prevent the leakage of any molecules across the membrane. In LeuTAa, it was proposed that this should be achieved by the interactions between TMs 1 and 8 and TMs 6 and 3 (17). The binding site is occluded by the side chains of selected residues that pack over the substrate in LeuTAa. The occluding mechanism of the outward-facing cavity for Mhp1, therefore, seems to be different from that for LeuTAa. It is noteworthy that in the closed vSGLT outward-facing cavity (21), TM11 (equivalent to TM10 of Mhp1) adopts a conformation similar to TM10 in the occluded form of Mhp1 (Fig. 3D).
The electron density map at 2.85 Å resolution clearly indicates a possible cation-binding site at the C-terminal end of TM1a interacting with TM8. The site includes the carbonyl-oxygen-atoms of Ala 38, and Ile 41 of TM1 and the carbonyl-oxygen-atom of Ala 309, and the hydroxyl-oxygen-atoms of the side chains of Ser 312 and Thr 313, respectively (Fig. 3E and fig. S6). Presumably the dipole moment at the C-terminus of TM1a contributes to the binding as seen for other transporters (17, 21, 27, 28, 30). Currently, a sodium ion is modelled at this position such that the substrate atoms form a square pyramidal arrangement with the bond distances between 2.2 and 2.7 Å, which is too short for a water molecule. An equivalent site was observed and assigned as a sodium binding site in the structures of LeuTAa and vSGLT (fig. S6). The connection between the cation-binding site and the substrate-binding site could be made by residues Asn318 on TM8 and Gln42 on TM1. A requirement for cation-binding to form a proper substrate-binding site provides a basis for the coupling of the cation and substrate translocation.
To confirm the sodium dependency of Mhp1, which was not evident in whole cell uptake experiments (12), the binding of benzyl-hydantoin and/or sodium was measured using fluorescence quenching (Fig. 3F) (36). The results clearly show that the affinity of benzyl-hydantoin to the protein (apparent Kd 0.88 ± 0.27 mM) is raised over ten-fold in the presence of 15 mM sodium (apparent Kd 0.054 ± 0.007 mM). We have also observed that the benzyl-hydantoin increases the affinity of sodium to Mhp1 (data not shown); in the absence of benzyl-hydantoin, the apparent Kd for sodium is 1.15 ± 0.28 mM, and in the presence of 2 mM benzyl-hydantoin, the value is 0.15mM ± 0.04 mM. These results clearly indicate that the binding of sodium and benzyl-hydantoin are tightly coupled in Mhp1.
Some of the NCS1 family transporters from yeasts and bacteria, including Mhp1, were thought to be proton- rather than sodium-dependent (12, 15, 16). Sodium dependence in whole cell transport assays may be obscured by the presence of a separate compensating sodium transport system in the membrane. In the case of Mhp1 the sensitivity of assays is reduced by the low solubility and the lipophilic nature of the substrate, preventing the testing of sodium dependence in proteoliposomes. It is also possible that Mhp1 and other NCS-1 transporters have a flexible cation selectivity, like the MelB sugar-cation symporter (37).
The two structures of Mhp1 reveal the conformational change of the transporter upon the binding of substrate from the outside of the membrane. The structure of vSGLT provides further insight into the transport process because it is in an inward-facing conformation closed with the substrate in its binding site (Fig. 2C and Fig. 4). We refer to this as an inward-facing, occluded conformation according to the assignment in (8). There, the cavity connecting the substrate-binding site to the inside of the membrane is observed although the site is still occluded (Fig. 2C, fig. S7 and fig. S8). In this vSGLT structure the outward-facing cavity, as also observed in LeuTAa and substrate-bound Mhp1, is completely closed (Fig. 2C). The observed inward-facing cavity in vSGLT is made of the neighbouring surfaces of TMs 1, 3, 5, 6 and 8, (Fig. 2C, fig. S7 and fig. S8), which are symmetrically related to the outward-facing cavity observed in Mhp1 on the opposite side of the membrane. Thus these structures are related by an approximate two-fold axis that is parallel to the membrane (fig. S7). In Fig. 4, we compare the conformations of the core 10 helices (TMs 1-10) of outward-facing Mhp1 and inward-facing vSGLT. These structures suggest a reciprocating oscillation between symmetrical states opening alternately to each side of the membrane.
In the following, we propose a possible mechanism of molecular transport across the membrane by Mhp1 based on the X-ray structures (Fig. 4).
The site of the cation uptake and release is also controlled by the conformational changes, because the ion-binding site is located between TMs 1 and 8 (Fig. 4D). The cation-binding sites of Mhp1 and LeuTAa are very similar (fig. S6) and are a part of the surface of the outward-facing cavity, whereas the one for vSGLT is significantly different and is a part of the inward-facing cavity occurring as a consequence of the conformational change of TMs 3 and 8. This, together with the strong coupling of substrate- and cation-binding, should form the basis of substrate-cation symport. The coordinated and reciprocating conformational changes observed on both sides of the membrane provide a structural basis for the widely-held view of an alternating access model deduced from kinetics data.
The structures of a Nucleobase-Cation-Symport-1 family transporter reveal the molecular basis of the alternating access model of membrane transport.
This research was funded primarily by the Biotechnology and Biological Sciences Research Council, BBSRC [grant numbers B17935, BB/C51725], with important contributions from Ajinomoto Inc., the European Membrane Protein consortium, EMeP [grant number LSHG-CT-2004-504601], the Membrane Protein Structure Initiative, MPSI [grant number BBS/B/14418], and the Wellcome Trust [grant number [062164/Z/00/Z]. We wish to thank the Membrane Protein Laboratory (MPL) at the Diamond Light Source Ltd for use of the MPL facilities. Jeff Abramson kindly provided us with the coordinates of vSGLT before their release from the Protein Data Bank. We are also grateful to Simon Phillips, Steve Baldwin, Phil Gilmartin, Sheena Radford, David Drew, Mika Jormakka, Kunihiko Watanabe and Momi Iwata for support or advice. JSPS provided personal funding to S.Y. and the Leverhulme Trust to P.J.F.H. S.W. was a recipient of an European Molecular Biology Organization EMBO, long-term fellowship. The coordinates and the structure factors for Mhp1 without substrate and the hydantoin complex have been deposited in the Protein Data (entries 2JLN and 2JLO, respectively).
Supporting Online Material
Materials and Methods
Figs. S1 to S8