Conformational flexibility driving charge-selective substrate translocation across a bacterial transporter

Bacterial membrane porins facilitate the translocation of small molecules while restricting large molecules, and this mechanism remains elusive at the molecular level. Here, we investigate the selective uptake of large cyclic sugars across an unusual passive membrane transporter, CymA, comprising a charged zone and a constricting N terminus segment. Using a combination of electrical recordings, protein mutagenesis and molecular dynamics simulations, we establish substrate translocation across CymA governed by the electrostatic pore properties and conformational dynamics of the constriction segment. Notably, we show that the variation in pH of the environment resulted in reversible modulation of the substrate binding site in the pore, thereby regulating charge-selective transport of cationic, anionic and neutral cyclic sugars. The quantitative kinetics of cyclic sugar translocation across CymA obtained in electrical recordings at different pHs are comparable with molecular dynamics simulations that revealed the transport pathway, energetics and favorable affinity sites in the pore for substrate binding. We further define the molecular basis of cyclic sugar translocation and establish that the constriction segment is flexible and can reside inside or outside the pore, regulating substrate translocation distinct from the ligand-gated transport mechanism. Our study provides novel insights into energy-independent large molecular membrane transport for targeted drug design strategies.


Molecular Dynamics Simulations
The missing residues in native CymA protein (residues 10 to 21) were modelled in CHARMM-GUI as described in the Methods section in the main text 1 .The protein was then protonated at pH 4.5 and 8 using the PDB2PQR webserver that uses PropKa to calculate the pKa values of the titratable amino acid residues (see Table S1) 2,3 .The charge of the native CymA protein at pH 4.5 and 8.0 was determined to be +14 and -6, respectively.The CymA protein was inserted in a DPhPC lipid bilayer containing 115 lipid molecules, each on the upper and the lower leaflets, resulting in 230 lipid molecules.The membrane-protein system was built in a tetragonal simulation box with box vectors: 10.2, 10.2 and 10.9 nm in the x, y and z-directions, respectively.The simulation system was solvated using TIP3P-charmm water molecules and neutralized by adding K + and Cl -ions to obtain 0.15 M KCl concentration.The simulation system with CymA, DPhPC lipid bilayer (PHPC in CHARMM-GUI), water and ions was prepared using the Membrane-Builder utility in the CHARMM-GUI webserver 1 .A threedimensional structure of the neutral α-cyclodextrin was obtained from the PubChem database (PubChem id: 444913).The six hydroxymethyl groups (-CH2OH) in neutral-α-cyclodextrin (αCD) were modified to methylammonium cations (-CH2-NH3 + ) to obtain a cationic-αcyclodextrin (am6αCD) molecule with +6 charge using the PYMOL software package.The geometry-optimized structure of the αCD molecule was then used as a cationic α-cyclodextrin.
The geometry optimization was performed using the Avogadro software package.The hydroxymethyl groups in neutral-α-cyclodextrin were modified to methyl sulfonate anions (-CH2-SO3 -) for obtaining an anionic α-cyclodextrin (s6αCD) molecule with -6 charge.To prepare the truncated CymA, we removed the first 15 residues of the native CymA and followed a similar protocol as described for native CymA for protonating ionizable amino acid residues at pH 4.5 and 8.0.The protein, lipids and ions were parameterized using CHARMM36 forcefield while the CD-based ligands were modelled using CHARMM general force field (CGENFF forcefield) [4][5][6][7] .Tables S5 and S6 provide the details for each system simulated in the present investigation.
The CD was inserted in the membrane protein system towards the extracellular side aligned with the z-axis.After an initial energy minimization using the steepest descents algorithm, the membrane-protein system was then equilibrated for 10 ns using a canonical (NVT) ensemble, followed by 20 ns equilibration using an isothermal and isobaric ensemble (NPT) with position restraints on protein, lipid and ligand heavy atoms.The Verlet cutoff scheme was employed with a minimum cutoff of 1.2 nm for the Lenard Jones interaction and short-range electrostatic interactions throughout the simulation 8 .Long-range electrostatic interactions were treated by the Particle Mesh Ewald (PME) summation method 9 .All bonds connected to hydrogen atoms were constrained using the LINCS algorithm 10 .The bonds and the angles of TIP3P water molecules were constrained using the SETTLE algorithm 11 .A time-step of 2 fs was used for all MD simulations.
During equilibration, the temperature and pressure were kept constant at 303.15 K and 1 bar using the Berendsen thermostat and barostat with time constants of 1 ps and 5 ps, respectively 12 .However, the Nose Hoover thermostat and Parinello-Rahman barostat were used for all production runs 13,14 .The equilibrated membrane-protein system was then subjected to steered MD simulation, where the CD was steered through the CymA from the extracellular to the periplasmic region by pulling the CD very slowly across the pore.The cyclodextrin was pulled through the pore slowly along the z-direction using a pull force with a force constant of 100 kJ mol -1 nm -2 and a pulling rate of 0.1 Å/ns.The steered MD simulations were run sufficiently long for the CD to translocate and move out of the pore.The snapshots from the steered MD simulations were used as initial configurations for the umbrella sampling simulations.We have run 121 umbrellas for reaction-coordinate values ranging from -3.00 nm to +3.00.The negative and positive values of the reaction coordinate refer to the CD translocation from the extracellular to the periplasmic side.Each umbrella was run for 10 ns total simulation time.
The harmonic restraint force constant was set to 1000 kJ mol -1 nm -2 for umbrella sampling simulations, ensuring good overlap between adjacent umbrella histograms.The free energy profile as a function of the reaction coordinate was obtained by reweighting the probability distributions of the reaction coordinate using the WHAM (Weighted Histogram Analysis Method) tool implemented in the GROMACS MD simulation software package 15 .All MD simulations were run using the GROMACS-2018 software simulation package.Periodic boundary conditions were employed in x, y, and z-directions.All snapshots were obtained using the molecular visualization packages PyMol and VMD 16 .
The details of the simulation systems are shown in Table S5 and Table S6, respectively.The simulation systems involving translocating the am6αCD at pH 8 and 4.5 are designated nat-am6αCDpH8 and nat-am6αCD4.5,respectively.The simulation system involving translocating s6αCD at pH 8 and 4.5 across the truncated CymA is designated trunc-s6αCDpH8 and trunc-s6αCDpH4.5,respectively.The simulation systems with am6αCD translocating through truncated CymA at pH 8 and 4.5 are referred to as trunc-am6αCDpH8 and trunc-am6αCDpH4.(MFEP) obtained using a post-string approach described by Morita et al 17

Fig. S5 :
Fig. S5: Representative configuration of the hydrated CymA embedded in DPhPC lipid bilayer MD-simulation system with am6αCD.

Fig. S6 :
Fig. S6: Multiple Configurations of the am6αCD while translocating through the native CymA.

Fig. S7 :
Fig. S7: A one-dimensional free energy profile was obtained for am6αCD.translocation through native CymA at pH 4.5 and 8. Fig. S8: Interaction of am6αCD with truncated CymA at pH 4.5 and 8. Fig. S9: Interaction of s6αCD with native CymA under different pH conditions.Fig. S10: Interaction of s6αCD with truncated CymA under different pH conditions.

Fig. S16 A
Fig.S16A minimum free energy path (MFEP) obtained using a post-string approach.

Fig. S2
Fig. S2 Gating pattern of native CymA.a) Electrical recordings of single native CymA at ±50 mV in pH 4.5.The unitary conductance (n=25) and current amplitude histogram are shown.b) Electrical recordings of single native CymA at ±50 mV in pH 8.0.The unitary conductance (n=25) and current amplitude histogram are shown.c) Electrical recordings of single native CymA at ±100 mV in pH 4.5 and d) pH 8.0.All points current amplitude histogram is shown.e) I-V curve obtained from a single native CymA in pH 4.5 and 8.0.The current signals were digitally filtered at 7 kHz.

Fig. S3
Fig. S3 Gating pattern of truncated CymA.a) Electrical recordings of single truncated CymA at ±50 mV in pH 4.5.The unitary conductance (n=25) and current amplitude histogram are shown.b) Electrical recordings of single truncated CymA at ±50 mV in pH 8.0.The unitary conductance (n=25) and current amplitude histogram are shown.c) Electrical recordings of single truncated CymA at ±100 mV in pH 4.5 and current amplitude histogram are shown.d) Electrical recordings of single truncated CymA at ±100 mV in pH 8.0 and current amplitude histogram are shown.e) I-V curve obtained from a single truncated CymA in pH 4.5 and 8.0.The current signals were digitally filtered at 7 kHz.

Fig. S4
Fig. S4 Interaction of am6αCD with native CymA under different pH conditions.a) Electrical recordings of single native CymA in the presence of am6αCD (10 µM, trans) at +50 mV and -75 mV in pH 4.5 b) Electrical recordings of single native CymA in the presence of am6αCD (10 µM, trans) at +50 mV and -75 mV in pH 8.0.Insets show the corresponding off and on dwell time histograms of CD blocking fitted with a monoexponential probability function.c) Scatter plots of current block amplitudes versus dwell time (off) of CD blocking with native CymA are shown.The current signals were digitally filtered at 7 kHz.

Fig. S5
Fig. S5 Representative configuration of the hydrated CymA embedded in DPhPC lipid bilayer MD-simulation system with am6αCD.The water layer on top and bottom of the lipid bilayer is shown as gray transparent slabs.The hydrophobic chains of the lipids are shown in cyan color.The phosphate groups of the lipids are shown as vdW spheres where the P-atom is shown in yellow and the O-atom in red colored spheres.The native CymA is shown in the green cartoon representation.The am6αCD is shown using vdW sphere representation where the C-atoms are shown as cyan spheres, H-atoms as white and O-atoms as red-colored spheres.

Fig. S6
Fig. S6 Multiple configurations of the am6αCD while translocating through the native CymA.The CD assumes different configurations from the extracellular side's entrance to the periplasmic side's exit.The native CymA pore is shown in a cartoon (gray) representation.In a) The CD was shown using a stick representation, and the C-atoms were colored differently, rendering different conformations of the CD at various stages of the translocation event.In b)CD configurations with space-filling sphere representations for visual comparison.

Fig. S7 A
Fig. S7 A one-dimensional free energy profile was obtained for am6αCD translocation through native CymA at pH 4.5 and 8.0.a) The error bars associated with the free energy calculation are calculated using a bootstrapping method implemented in gmx wham utility of GROMACS MD simulation software.The error bars are black (pH 4.5) and blue (pH 8.0) transparent filling on the respective free energy profiles.b) The umbrella histogram (121 windows) overlaps for native CymA umbrella sampling simulations at pH 8.0 for RC1 ranging from -3 to +3 nm.c) The free energy profiles for am6αCD translocation through native CymA at pH 8.0 obtained from 10 ns (blue) and 20 ns (black) simulation times for 121 umbrella windows spanning a range of -3.0 to +3.0 nm for the z-component of protein-ligand center-to-center distances or RC1.

Fig. S8
Fig. S8 Interaction of am6αCD with truncated CymA at pH 4.5 and 8.0.a) Electrical recordings of single truncated CymA in the presence of am6αCD (10 µM, trans) at +50 mV and -75 mV in pH 4.5 b) Electrical recordings of single truncated CymA in the presence of am6αCD (10 µM, trans) at +50 mV and -75 mV in pH 8.0.Insets show the corresponding off and on dwell time histograms of CD blocking fitted with a monoexponential probability function.c) Scatter plots of current block amplitudes versus dwell time (off) of CD blocking truncated CymA are shown.The current signals were digitally filtered at 7 kHz.

Fig. S9
Fig. S9 Interaction of s6αCD with native CymA under different pH conditions.a) Electrical recordings of single native CymA in the presence of s6αCD (10 µM, trans) at -50 mV and -75 mV in pH 8.0 b) Electrical recordings of single native CymA in the presence of s6αCD (100 µM, trans) at -50 mV and -75 mV at pH 8.0.The current signals were digitally filtered at 7 kHz.

Fig. S10
Fig. S10 Interaction of s6αCD with truncated CymA under different pH conditions.a) Electrical recordings of single truncated CymA in the presence of s6αCD (100 µM, trans) at -25 mV and -75 mV in pH 4.5 b) -125 mV and +50 mV in pH 4.5.c) Electrical recordings of single truncated CymA in the presence of s6αCD (100 µM, trans) at -25 mV and -75 mV in pH 8.0.c) Scatter plots of current block amplitudes versus dwell time (off) of CD blocking truncated CymA are shown.The current signals were digitally filtered at 7 kHz.

Fig. S11
Fig. S11 Interaction of neutral αCD with CymA pores at different pHs.a) Electrical recordings of single truncated CymA in the presence of αCD (100 µM, trans) at +50 mV in pH 4.5.Electrical recordings of single truncated CymA in the presence of αCD (10 µM, trans) at +50 mV in pH 6.0 and pH 8.0.Insets show the corresponding off and on dwell time histograms of CD blocking fitted with a monoexponential probability function.b) Electrical recordings of a single native and truncated CymA in the presence of neutral αCD (10 µM, trans) at +50 mV in pH 6.0.The current signals were digitally filtered at 7 kHz.

Fig. S12
Fig. S12 Interaction of am6αCD, s6αCD and αCD with truncated CymA.a) Electrical recordings of single truncated CymA in the presence of am6αCD (10 µM, trans) at +50 mV in pH 8.0 b) Electrical recordings of single truncated CymA in the presence of s6αCD (100 µM, trans) at -50 mV in pH 4.5.c) Electrical recordings of single truncated CymA in the presence of neutral αCD (10 µM, trans) at +50 mV in pH 6.0.The current signals were digitally filtered at 7 kHz.

Fig. S13
Fig. S13 Interaction of am6αCD and s6αCD with truncated CymA at pH 8 in orbit16.a) Schematic of CymA pore orientation in planar lipid bilayer system.b) Schematic of CymA pore orientation in Orbit 16 system.c) Electrical recordings of single truncated CymA in the presence of am6αCD (10 µM, cis) at -25 mV and -50 mV in pH 8.0.d) Electrical recordings of single truncated CymA in the presence of s6αCD (100 µM) at +25 mV and +50 mV in pH 8.0.The current signals were filtered at 10 kHz and sampled at 20 kHz.

Fig. S14
Fig. S14 Molecular model of substrate translocation through CymA.a) Crystal structure of native CymA highlighting the presence of resolved N terminus segment (residues 1-9).b) Modelled native CymA highlighting the entire N terminus segment in the pore.c) Translocation of am6αCD across native CymA and associated N terminus segment conformational dynamics.

Fig. S15
Fig. S15 CymA protein residues involved in polar interactions with am6αCD a) amino acid residues involved in polar interactions with the ligand where the Nterm loop and the ligand coexist inside the pore when the ligand crosses the pore's center.The configuration refers to the Fig. 6C snapshot in the main text.b) amino acid residues involved in polar interactions with the ligand where the Nterm loop is inside while the ligand has translocated almost outside the pore towards the periplasmic region.The configuration refers to the Fig. 6D snapshot.The protein is shown in cartoon representation (grey), the N-term loop region (residue index 1-15) is shown in cartoon representation (red) and the amino acid residues involved in polar interactions are shown as sticks (C-cyan, O-red, N-blue, H-white) and the am6αCD ligand is shown in sticks representation (C-yellow, O-red, H-white) and the steered MD trajectory is projected onto the 2D-FES as a function of RC1 and RC2, for am6αCD translocation through native CymA at pH 8.

Fig. S17
Fig. S17 Single-channel properties of fully truncated CymA.a) Electrical recordings of fully truncated CymA insertion at +50 mV and +150 mV in pH 8.0.b) Electrical recordings of fully truncated single CymA at +25 mV and +50 mV showing sub conductance states in pH 8.0.c) Electrical recordings of fully truncated single CymA showing unstable states at +25 mV and +50 mV in pH 8.0.The current signals were filtered at 2 kHz and sampled at 10 kHz.

Fig. S18
Fig. S18 Interaction of neomycin with truncated CymA.a) Electrical recordings of single truncated CymA in the presence of neomycin (10 µM, trans) at +25 mV.b) +50 mV in pH 8. Insets show the corresponding off and on dwell time histograms of neomycin blocking fitted with a monoexponential probability function.The current signals were digitally filtered at 7 kHz.

Table S5 .
Description of the different components of the simulation systems: system

Table S6 .
Description of simulation systems: Simulation system nomenclatures and their corresponding pHs, simulation box vectors in X, Y and Z-directions, and the total number of atoms present in simulation boxes.Natoms refers to the total number of atoms in the simulation systems.