Structure and mechanism of a tripartite ATP-independent periplasmic TRAP transporter

Abstract

In bacteria and archaea, tripartite ATP-independent periplasmic (TRAP) transporters uptake essential nutrients. TRAP transporters receive their substrates via a secreted soluble substrate-binding protein. How a sodium ion-driven secondary active transporter is strictly coupled to a substrate-binding protein is poorly understood. Here we report the cryo-EM structure of the sialic acid TRAP transporter SiaQM from Photobacterium profundum at 2.97 Å resolution. SiaM comprises a "transport" domain and a "scaffold" domain, with the transport domain consisting of helical hairpins as seen in the sodium ion-coupled elevator transporter VcINDY. The SiaQ protein forms intimate contacts with SiaM to extend the size of the scaffold domain, suggesting that TRAP transporters may operate as monomers, rather than the typically observed oligomers for elevator-type transporters. We identify the Na⁺ and sialic acid binding sites in SiaM and demonstrate a strict dependence on the substrate-binding protein SiaP for uptake. We report the SiaP crystal structure that, together with docking studies, suggest the molecular basis for how sialic acid is delivered to the SiaQM transporter complex. We thus propose a model for substrate transport by TRAP proteins, which we describe herein as an 'elevator-with-an-operator' mechanism.

Fig. 1. Structure of Photobacterium profundum SiaQM.

a Cryo-EM density for the SiaQM-${\mathrm{Mb}}_{\mathrm{Nb}07}^{\mathrm{HopQ}}$ complex in amphipol. The megabody is bound at the periplasmic face of SiaQM and makes contacts with both the Q- and M-subunits. Inset shows density and position of the amphipol belt. Cryo-EM density displayed at 7.5σ, as calculated by ChimeraX. b Cryo-EM density for the SiaQM-${\mathrm{Mb}}_{\mathrm{Nb}07}^{\mathrm{HopQ}}$ complex in MSP1D1-E. coli phospholipid nanodiscs (also displayed at 7.5σ). Density for lipids is shown in pink, while the contour of the disc is shown in grey. More detailed views are displayed in Supplementary Figs. 2 and 6. c Cartoon depicting the topology of SiaQM. The overall topology of SiaQM can be arranged into a transport domain and a rigid scaffold domain. The inverted topology of the M-subunit is seen in this cartoon, where Scaffold I, Arm I and Transport I form one repeat, with the remainder forming the second repeat. Hairpin helices HP_in and HP_out are indicated. d Structural organisation of SiaQM. Left: SiaQM structure, showing the Q-subunit (green), and the M-subunit coloured by domain (scaffold domains: purple and blue; transport domains: orange and gold; arm helices: pink). Right: Views from the cytoplasm and periplasm, showing the scaffold domains (green, blue and purple) bracing the transport domains (orange and gold), which are cradled between the two arm helices (pink). e Surface representation of the interface between SiaQ and SiaM. The surfaces of the contact residues at the interface are coloured by electrostatic potential, as calculated in ChimeraX. Highlighted are conserved residues that form salt bridges between the subunits, R32:E237 and R47:D50. Consurf analyses show that R32 and E237 are fully conserved across the sequences sampled, highlighting the structural and functional significance of this pair.

Fig. 2. Proteoliposome transport assays of SiaPQM.

a Transport by SiaPQM is dependent on an inwardly directed Na⁺ gradient and net transport is electrogenic as activity is enhanced when an inside negative membrane potential (ΔΨ, −117.1 mV) is imposed. Curves show external [³H]-Neu5Ac uptake into proteoliposomes reconstituted with SiaQM, in the presence of SiaP. Green circles, an inward sodium ion gradient (ΔμNa⁺) is present, with a membrane potential generated by valinomycin before measurement. Orange circle, ethanol was added instead of valinomycin as a control. Purple triangle, no ΔμNa⁺ was present, but a ΔΨ was imposed. Pink square, no ΔμNa⁺ was present and ethanol was added instead of valinomycin as a control. The plot presents the mean ± s.e.m. from five independent experiments (n = 5). b The dependence of Neu5Ac uptake into proteoliposomes based on SiaP. The conditions are the same as (a). SiaP is required for transport but the rate is not linear when the SiaP concentration is increased to 1.0 µM. All data are reported as means ± s.e.m. from four independent experiments (n = 4). c Dependence of Neu5Ac uptake into proteoliposomes based on external Na⁺ (green circle), with SiaP at 0.5 μM. Transport was measured in the presence of varying concentrations of external Na⁺-gluconate and fitted with the Hill equation, giving a Hill coefficient of 2.7 (95% CI = 1.9–4.0). The plot presents the mean ± s.e.m. from five independent experiments (n = 5). d Uptake of Na⁺, monitored by measuring the fluorescence emission of Sodium Green™, versus the uptake of [³H]-Neu5Ac into proteoliposomes. The mean Na⁺ uptake was 216 ± 41 nmol, while the mean sialic acid uptake was 82 ± 17 nmol (two-sided T-test P = <0.0001, determined using GraphPad Prism), giving a ratio of 2.6, consistent with the Hill coefficient in (c). The plot presents the mean ± s.d. from six independent experiments (n = 6) in each. e SiaQM transport activity is sensitive to the lipid environment. Transport was measured using proteoliposomes reconstituted with phosphatidylcholine from egg-yolk or E. coli total lipid extract. As a ΔΨ control, ethanol (grey bars) was used instead of valinomycin (orange bars). The bar presents the mean ± s.e.m. from four independent experiments (n = 4) in each. f Surface cutaway of SiaQM. The structure is in an inward-facing conformation, which is the substrate release state, with the presence of a large solvent-accessible cavity on the cytoplasmic face of the complex. Source data are provided as a Source data file.

Fig. 3. Predicted full TRAP complex.

a Interacting regions of SiaQM and SiaP, as determined using the algorithms RaptorX, Gremlin (Supplementary Tables 4 and 5) and AlphaFold. Surface representation of the SiaQM and SiaP structures with patches coloured and lettered on each surface to indicate the binding mode. The proposed binding surface on SiaQM largely involves the surface of the scaffold (green, purple and blue). TM3a aligns well with α2 of the P-subunit (B blue and D purple). The loop between TM7 and TM8 aligns well with α5 of the P-subunit (C blue). The two periplasmic loops of the Q-subunit are also predicted to interact with the P-subunit (E and F, green). The orange and pink patches represent the Transport I and Arm II components as coloured in Fig. 1b. b Mutagenesis of surface residues on both SiaP and SiaQM that are predicted to be important for function, as well as residues in the SiaM substrate and Na⁺ binding sites. The data are normalised to the wildtype SiaPQM transport rate (grey). Mutants are coloured according to the regions shown in Fig. 1a. Data are reported as means ± s.e.m. (error bars) from four independent experiments. Source data are provided as a Source data file. Symbols mark residues pairs depicted in (d). c Residue conservation mapped onto the SiaP and SiaQM structures. Cartoon representation of SiaP and SiaQM, coloured according to Consurf score. d Complex of the SiaQM cryo-EM structure and SiaP crystal structure based on the binding mode predicted by AlphaFold^, (the final model can be found in Supplementary Data 1). An interaction hotspot (inset, left and right) shows a number of titratable residues at the interface. e Modelling of the TRAP complex in the inward-facing (left) and outward-facing (right) conformations (cutaway, viewed from the scaffold domain, looking towards the substrate-binding site, white circle). Left, the experimental structures determined here are aligned based on AlphaFold predicted complexes. Right, an outward-facing model was generated by homology modelling the SiaM transport domain using LaINDY (PDB ID: 6wu4) as the template (the final model can be found in Supplementary Data 2). This model was then aligned to the scaffold domain of model I. Together, these models demonstrate how an elevator motion of the transporter could expose the substrate binding site of SiaM to SiaP.

Fig. 4. The TRAP ‘elevator-with-an-operator’ mechanism.

(I) The P-subunit (maroon) binds the substrate (cyan) with high affinity and undergoes a conformational change from the open to closed state. (II) The closed P-subunit then docks to the QM subunits (orange, purple and green). (III) We propose that docking induces a conformational change in the transporter to a state where Na⁺ ions (green) and the substrate can bind with greater affinity. This change is coupled to the allosteric modulation of the P-subunit to the open conformation, releasing the substrate to the transporter. (IV) The open state P-subunit, which presumably has lower affinity for the transporter, diffuses away, allowing the transporter to move to an inward-facing state (V), with the substrate and coupling ions then released into the cytoplasm. We note it is possible the conformational change induced in the transporter (II) may be either a local gating rearrangement, or a global elevator-type motion. Regardless, we suggest that the P-subunit is the ‘operator’ of the elevator, as transport without the P-subunit is negligible, as seen in Fig. 2b.