An octanol hinge opens the door to water transport

Despite their prevalent use as a surrogate for partitioning of pharmacologically active solutes across lipid membranes, the mechanism of transport across water/octanol phase boundaries has remained unexplored. Using molecular dynamics, graph theoretical, cluster analysis, and Langevin dynamics, we reveal an elegant mechanism for the simplest solute, water. Self-assembled octanol at the interface reversibly binds water and swings like the hinge of a door to bring water into a semi-organized second interfacial layer (a “bilayer island”). This mechanism is distinct from well-known lipid flipping and water transport processes in protein-free membranes, highlighting important limitations in the water/octanol proxy. Interestingly, the collective and reversible behavior is well-described by a double well potential energy function, with the two stable states being the water bound to the hinge on either side of the interface. The function of the hinge for transport, coupled with the underlying double well energy landscape, is akin to a molecular switch or shuttle that functions under equilibrium and is driven by the differential free energies of solvation of H2O across the interface. This example successfully operates within the dynamic motion of instantaneous surface fluctuations, a feature that expands upon traditional approaches toward controlled solute transport that act to avoid or circumvent the dynamic nature of the interface.


S1 Literature Analysis
: Previous theoretical and computational publications that have focused upon the water/octanol biphasic systems. Note that the works of Wip↵ and coworkers [1,2,3,4] have examined metal-ligand complexe distributions across water/octanol interfaces as it pertains to liquid-liquid extraction. Computational protocols involved mixing/demixing (non-equilibrium) and potential of mean force were employed to understand the energetic favorability in the two bulk phases (where the distance coordinate is not the transport reaction coordinate). [5,6] adopt mass-transfer models to understand the extraction equilibria and kinetics solute complexes between the aqueous and octanol phases.

S2 Simulation Protocols/Models
Figure S1: Schematic representation of the octanol/water liquid-liquid interface simulation box. Bulk water and octanol molecules are depicted in gray and cyan lines respectively. Extracted water molecules into the octanol phase and the interfacial water layer are described with red oxygens, while two adjacent water layers that directly beneath the interfacial water layer with green and blue oxygens, respectively (only shown in the right half cell). The bilayer structure of octanol molecules in the vicinity of the interface (left half box) is highlighted with cyan bonds for alkyl carbon tail and orange sphere for oxygens in the hydroxyl head group.

S2.1 Convergence of Simulation Data
Equilibrium is ascertained through monitoring a number of properties, including the interfacial tension, concentration of water in the octanol phase, and number of transport events between the two phases (vide infra) The black curve in the left panel of Figure S4 shows the scaled water density profile along the z axis by its bulk density in the octanol-rich phase. It can be observed that there presents a significant region of enhanced water concentration around 17Å from the GDS toward the octanol phase, which is in good agreement with the previous simulation. The average densities of water and octanol more than 1 nm from the GDS in the octanol-rich phase yields an average mole fraction of water of 0.26, in good agreement with the experimental water solubility in octanol [103] (see Table 1). The left panel of Figure S4 presents the density profiles of octanol atom types in the right half the simulation box along the z axis (scaled by their bulk densities in the octanol phase). The oscillation in the hydroxyl Oatom density profile draws the further interest, with its first peak sitting exactly at the position of the Gibbs Dividing surface (GDS) (z = 0). The density has a minimum near 9Å from the GDS, which is in good agreement with previous simulations having a similar distances of 8 [79] or 9Å [32]. We further calculate the average octanol molecular end-to-end distance (hydroxyl oxygen to methyl carbon) to be 8.68Å, coinciding with the di↵erence between the first maximum and minimum of the hydroxyl oxygen density profile. The density oscillation propagates further along the z dimension of the simulation box, with the second peak sitting ⇠17Å from the GDS, a value comparable to twice the octanol molecular end-to-end length. The average end-to-end distance of octanol molecules in the second layer of the bilayer structure has a value is 8.54Å. It is worth to note that the sum of this obtained value and the average interfacial octanol molecular  end-to-end distance is confirmed to be equal to the di↵erence between the aboved-mentioned maxima of the hydroxyl oxygen density profile. The distribution of the molecular end-to-end distance of the octanol backbone Figure S4: Scaled density profiles from the octanol-water liquid-liquid interface system for water (black line), octanol hydroxy oxygens (red line), octanol methylene carbons (green line), and octanol methyl carbons (blue line) (left panel). Average orientation of water dipoles (black lines) and end-to-end atoms (hydrogen from OH groups and methyl carbon) for octanol (blue lines). The angle is with respect to the z axis, with zero representing no orientational preference for both profiles (right panel).

S3.3.2 Determining E a for Transport.
The stochastic flipping of individual octanol between Layer-1 () Layer-2 was first examined. For A, we presume that in order for an octanol to stochastically flip, it must not be hydrogen bonded to any other species, and thus A is estimated to be the rate at which octanol within a layer loses its hydrogen bonds to other octanol and H 2 O. Based upon the compositions of all transferring clusters of octanol and water (Layer-1 () Layer-2), the overall reaction for transport only of a single water using the hinge mechanism is: To be able to employ the Arrhenius equation using this chemical reaction is however not possible, as to determine the rate constant, the both the rate of transfer and the total concentration of all reactant and products is needed. Within the analyses performed only the successful transfer processes are observed, as such the total concentration is not known.
Instead, the Arrhenius equation is employed for the following reaction: This reaction encompasses both the formation of the H 2 O(oct) 1.63 reactive hinge structures from all water and octanols in the layer to and the transport of water in the hinge form. In this case, the concentrations of all reactant and product species are known.
Two estimates of the prefactor A have been employed: 1) using the rate of new hydrogen bond formation between water-octanol (in that a water-octanol cluster cannot form without the formation of a HB), or 2) the rate of new waters that adsorb to the surface and hydrogen bond with octanol in Layer-1 and Layer-2. The complete table of all values and associated calculated E a values with all statistical uncertainties is presented below, employing the following equation: Using the predicted E a values and the associated statistical errors, a series of Langevin Dynamic simulations were performed using a double well potential, as described in the Computational Methods section. The Table  below presents the fitted parameters of the double well analytical representation for the di↵erent respective E a values (and those including statistical uncertainties), along with the equilibrated ratio of particles in the left-hand (Layer-1) side and right-hand (Layer-2) side of the well potential. Table S8: Arrhenius equation fitting data for the molecular transport in the process of water transport between Layer-1 () Layer-2. Transport rates are in /10 ps and concentration is in number of molecules. Prefactor A was estimated based upon two di↵erent rates: (a) the rate of new hydrogen bond formation in Layer-1 or Layer-2, and (b) the rate of new water adsorption into Layer-1 or Layer-2.