(a) Schematic overview of the conformational diversity of available mammalian POT structures. Intermediate positions between states indicate partial gate opening. (b) Alphafold-predicted inwards-facing (IF) HsPepT2 structure (top view), highlighting potential inter-bundle extracellular gate interactions. (c) Outwards-facing (OF) Cryo-EM structure of apo RnPepT2 (7NQK, bottom view) (Parker et al., 2021), highlighting potential inter-bundle intracellular gate interations. (d) Ala-Phe substrate binding pose, representative cluster frame of 1 µs MD simulation from 7NQK structure with added ligand, for setup details see Methods and Materials. Purple dotted lines represent salt-bridge contacts, orange dotted lines other polar contacts. (e) ExxER motif salt-bridge cluster, representative cluster frame of 1 µs MD simulation from 7NQK structure.

(a) Illustration of the CVs used to quantify extra- and intracellular gate opening, consisting of inter-bundle centre-of-mass distances between the helical tips (top 11 residues) and bases (bottom 11 residues). (b)+(c) Triplicate 1 µs-MD simulations starting from OCC, showing the effects of different protonation and substrate binding states, projected onto the (b) Tip-CV and (c) Base-CV respectively. Violin plots are trajectory histograms, arrows link the CV values of the first and last frames.

(a) 2D-PMFs from REUS starting with MEMENTO paths, in different protonation states of candidate extracellular gating residues. (b) Projection of the 2D-PMFs in part a onto PC 1 using Boltzman reweighting. Shaded areas indicate convergence errors as the range of PMF values for a given CV value obtained with the first 40 %, the last 40 % and 100 % of sampling included (after alignment to the 100 % curve). H87 and D342 form an additive extracellular gate, where H87 protonation changes the relative OCC–OF state energies as well as the transition barrier, while D342 protonation only contributes in the transition region. Note that the individual PMFs are only determined by our REUS approach up to additive constants, and are shown aligned here at the OF state for convenience of comparison.

(a) 2D-PMFs from REUS starting with MEMENTO paths, in different protonation states of candidate intracellular gate-controlling residues. (b) Projection of the 2D-PMFs in part a onto PC 1 using Boltzman reweighting. Shaded areas indicate convergence errors as the range of PMF values for a given CV value obtained with the first 40 %, the last 40 % and 100 % of sampling included (after alignment to the 100 % curve). E53 and E622 protonation have additive and approximately equal effects on driving the OCC→IF transition. Note that the individual PMFs are only determined by our REUS approach up to additive constants, and are shown aligned here at the IF state for convenience of comparison.

Results of ABFE calculations, showing that the affinity of Ala-Phe substrate does not depend much on the conformational state (OF vs IF), but is signicantly decreased on E622 protonation.

(a) 2D-PMF for the OCC↔OF transition from REUS starting with Ala-Phe-bound PepT2 MEMENTO paths. The OCC state has an increased basin width in PC 2 (compare for figure 3a), and a transition path shifted in PC 2. (b) Projection of the PMF from panel a onto PC 1, showing how in holo PepT2, the OCC state is stabilised by ≈ 1 kcal mol−1. Shaded areas indicate convergence errors as the range of PMF values for a given CV value obtained with the first 40 %, the last 40 % and 100 % of sampling included (after alignment to the 100 % curve). Note that the individual PMFs are only determined by our REUS approach up to additive constants, and are shown aligned here at the OF state for convenience of comparison. (c) 2D-PMF for the OCC↔IF transition from REUS starting with Ala-Phe-bound PepT2 MEMENTO paths. The structure of the IF plateau is not significantly affected, but OCC is more flexibile in PC 1. (d) Projection of the PMF from panel c onto PC 1, showing how in holo PepT2, the OCC state has a broader basin, corresponding to intracellular-gate flexibility. Convergence error and alignment of PMFs is shown as in panel b.

(a) E53 and E56 pKa values from constant-pH MD simulations, in the apo and holo as well as the OF and OCC states, estimated as mean ± standard deviation from triplicate runs (using the full simulation data for fitting the titration curves). The presence of substrate raises the E56 pKa in either conformational state, while some effect on the E53 pKa may also exist in the OF state. (b) Illustration of a themodynamic cycle of E56 protonation and Ala-Phe binding, with edges filled in via CpHMD (converted into kcal mol−1 at pH 7) for the top and bottom transitions, and ABFE for the left and right edges. Notably, ABFE displays a response of Ala-Phe affinity to E56 that is consistent with the CpHMD results, and the cycle closes very well. The error in the cycle closure residual is estimated as a square root of the sum of squared standard deviations of the individual edges.

(a) Cell-based transport assays for PepT2 WT (transfected with 0.5 µg, n=12, and 0.8 µg, n=46, of DNA per well), empty plasmid vector (n=12) and PepT2 H87A, D342A, S321A (n=24 each) and I135L (n=12) mutants, all transfected with 0.8 µg of DNA. Diagram shows transport as fluorescence in post-assay lysate divided by total protein concentration, normalised to the WT (0.8 µg) mean. Bars are mean values plus minus standard deviation, and swarm plots samples corresponding to individual wells. Single asterisk indicates p < 10−3, double asterisks p < 10−4 significance levels for difference compared to (weaker transporting, 0.5 µg-transfected) WT, as evaluated using a two-tailed t-test. (b) Western-blot showing expression levels of WT and mutant GFP-labelled PepT2, with an anti-GFP primary antibody. All mutants express at levels between the WT transfected with 0.5 µg and 0.8 µg plasmid DNA. Cleaved GFP is also visible at low molecular weight, at levels comparable for WT and mutants.

Schematic overview of the PepT2 alternating-access transport cycle proposed in this work. Protons located at a question mark indicate a proton-transfer step with an as-of-yet unknown mechanism regarding intermediate residues.

Residue numbers used in the defintion of the tip-CV and base-CV.

Overview of all 1D-PMF sampling.

Overview of all 2D-PMF sampling.