Phenylalanine Transport through LAT1: Insights from Molecular Dynamics, Steered Molecular Dynamics, and Targeted Molecular Dynamics

Abstract

L-type amino acid transporter 1 (LAT1) is a heterodimeric membrane protein that primarily facilitates the transport of phenylalanine, along with other amino acids such as valine, leucine, isoleucine, tryptophan, and tyrosine across the cell membrane. It is predominantly expressed at the blood–brain barrier. While LAT1 has been extensively studied in the context of drug delivery, its specific transport pathway for phenylalanine and the related conformational dynamics remains largely unexplored. Investigation of the transport pathway of phenylalanine via LAT1 is also very important, as excessive amounts of phenylalanine can cause phenylketonuria (PKU). Inhibiting LAT1’s transport could reduce phenylalanine’s entry into the brain, aiding in the management of its levels and alleviating the adverse effects of PKU. To investigate the phenylalanine transport mechanism and the conformational dynamics of LAT1, microsecond-long molecular dynamics (MD) as well as steered molecular dynamics (SMD) and targeted molecular dynamics (TMD) simulations were performed. The results of this study identify essential structural motifs that enable the isomerization of LAT1 among outward-facing, inward-facing, and occluded states. A significant conclusion drawn from this research is the identification of seven critical transmembrane helices that play central roles along the transport pathway, namely, TM1, TM4, TM6, TM7, TM8, TM9, and TM12. Root-mean-square deviation (RMSD) analyses across the structural models further indicate that TM1 and TM6 are consistently engaged during both extracellular and intracellular phenylalanine transitions, whereas TM9 and TM12 contribute more specifically to the outward and inward opening events, respectively. The observed conformational changes offer fresh insights into the alternating-access mechanism of LAT1 during phenylalanine transport, contrasting with conclusions from structural models based on other substrates transport through LAT1. By elucidating this transport mechanism, the study will contribute to developing LAT1 inhibitors that can selectively regulate brain phenylalanine levels without disrupting the transport of other essential amino acids.

Introduction

L-type amino acid transporter 1 (LAT1) is a transmembrane protein that mediates the uptake of large neutral amino acids (LNAAs) across cell membranes and plays a central role in cellular nutrition and metabolic regulation. It is highly expressed at the blood–brain barrier (BBB) and also facilitates the transport of essential substrates, including thyroid hormones, L-DOPA, gabapentin, and various metabolites into the brain. LAT1 exhibits pronounced substrate specificity, with a preferential transport hierarchy of amino acids as follows: phenylalanine > tryptophan > leucine > isoleucine > methionine > histidine > tyrosine > valine. The amino acids transported by LAT1 possess bulky side chains that engage effectively with the hydrophobic binding pocket of LAT1 through extensive van der Waals and hydrophobic interactions, while remaining neutral at physiological pH. In addition, they retain free α-amino and α-carboxyl groups, which form conserved hydrogen-bonding interactions with residues S66, G67, F252, and G255, providing a common anchoring framework for substrate binding. In contrast, small amino acids (e.g., glycine and alanine) lack sufficient side-chain volume for stable hydrophobic contacts, whereas charged amino acids (e.g., arginine and aspartate) are energetically unfavorable within the hydrophobic pocket. − Aromatic substrates (phenylalanine and tryptophan) engage key gating residues such as F252 (TM6) through π–π interactions and optimally occupy multiple hydrophobic subpockets, resulting in stronger binding compared to purely aliphatic substrates. , Together, these structural and physicochemical factors govern LAT1 substrate selectivity and transport preference.

Targeting LAT1 offers a strategic approach to fight diseases such as phenylketonuria (PKU) and cancer. In phenylketonuria (PKU), phenylalanine accumulates to toxic levels in the brain due to impaired metabolism. This elevated phenylalanine disrupts normal neurological function and can lead to severe cognitive and developmental deficits if left untreated. Given the role of LAT1 in facilitating phenylalanine transport across the BBB, modulating its function presents a potential therapeutic strategy. By limiting phenylalanine influx into the brain, LAT1-targeted approaches may reduce neurotoxic accumulation and alleviate the neurological burden associated with PKU, offering a promising avenue for improving patient outcomes. In addition to the BBB, the LAT1 transporter is also expressed in the cerebral cortex, blood–retina barrier, testis, placenta, bone marrow, and various types of cancer cells. LAT1 is frequently overexpressed in various human cancers, including breast, prostate, lung, colorectal, head, neck, and gliomas, − reflecting the high demand for amino acids to support rapid growth and proliferation. Inhibitors of LAT1 function by limiting the availability of amino acids to cancer cells, thereby hindering protein synthesis and cellular proliferation through downregulation of mammalian target of rapamycin complex 1 (mTORC1) signaling. − Similarly, abnormal LAT1 expression, regulated by c-Myc, plays a significant role in colon cancer progression. The identification of LAT1 inhibitors has facilitated the development of novel compounds that act as substrate mimetics, effectively competing for binding at the transporter’s active site. ,− Furthermore, phenylalanine analogues, such as 3-([1,1′-biphenyl]-3-yl)-2-aminopropanoic acid, have been found to strongly inhibit LAT1.

Recent cryo-EM studies have resolved human LAT1–4F2hc (Figure A) in complex with several high-affinity ligands and inhibitors, providing key structural insights into inhibitor recognition. In these structures, potent LAT1 inhibitors (IC₅₀ ∼ 100–250 nM) such as HNX (2∼{S})-2-azanyl-7-(naphthalen-1-ylmethoxy)-3,4-dihydro-1∼{H}-naphthalene-2-carboxylic acid bind within the canonical substrate-binding pocket, stabilizing an outward-facing occluded conformation and preventing amino acid transport. Complementary cryo-EM structures of LAT1–4F2hc in the inward-open conformation with another potent LAT1 inhibitor, BCH (2 amino-2-norbornanecarboxylic acid) further illustrate how inhibitors occupy the binding site and highlight the essential role of LAT1 during the substrate transport. A recent cryo-EM study captured LAT1 in an outward-facing conformation with bound phenylalanine, revealing a widely open configuration of TM1b and TM6a, with Phe252 adopting an upward orientation that exposes the substrate-binding pocket to the extracellular environment; in this state, the phenyl ring of l-Phe is oriented toward Gly255. Consistent with these structural observations, our integrated molecular dynamics (MD), targeted MD (TMD), and steered MD (SMD) simulations demonstrate that the outward-facing conformation is primarily driven by the opening of TM1 and TM6, as discussed in detail in the following sections. No other experimentally resolved LAT1 structure with bound phenylalanine is currently available, limiting detailed characterization of phenylalanine-specific transport mechanisms.

1.

Structural model of phenylalanine-bound LAT1 and schematic illustration of LAT1’s role in disease and therapeutic targeting. (A) Structural model of phenylalanine (magenta) bound to LAT1 (orange), complexed with the transmembrane glycoprotein 4F2hc (tan), (B) Conceptual schematic illustrating LAT1’s contribution to cellular growth and its involved in autophagy, protein synthesis, and cell proliferation. Its critical role in tumor initiation and progression makes it a promising therapeutic target with several LAT1 inhibitors showing promise in suppressing tumor cell proliferation.

LAT1 is a member of the heterodimeric amino acid transporters (HATs) family, which consists of two subunits: a light chain (SLC7) responsible for amino acid transport, and a heavy chain (SLC3) that facilitates plasma membrane localization and stabilization of the light chain. The LAT1 protein comprises 12 putative transmembrane segments (TMs) arranged in a bilayer structure (Figure A and Supporting Information, Table S1). The inner layer consists of five TMs (TM1, TM3, TM6, TM8, and TM10), surrounded by an outer layer of seven TMs (TM2, TM4, TM5, TM7, TM9, TM11, and TM12). Notably, both the N- and C-terminal ends of LAT1 are localized intracellularly. In addition, the 4F2hc protein, also known as CD98, is a transmembrane glycoprotein that serves as a heavy chain subunit of the LAT1 complex (Figure B). It is essential for the proper folding, trafficking, and function of LAT1. While the 4F2hc-LAT1 complex is vital for cellular function and metabolism, particularly in amino acid transport to the brain and other tissues, recent studies indicate that LAT1 is the primary transport unit, with 4F2hc primarily acting as a molecular chaperone to ensure LAT1’s correct localization in the membrane.

As a member of the amino acid/polyamine/choline (APC) transporter superfamily, LAT1 exhibits conformational dynamics that are likely analogous to those of LeuT, a well-characterized neurotransmitter transporter within the same superfamily. Both transporters are believed to operate via an alternating-access mechanism, in which the outward-facing conformation permits substrate binding, followed by a transition to an inward-facing state that enables substrate release into the cytoplasm. This mechanistic model is supported by kinetic study of a substrate, gabapentin, that also transport through LAT1. The estimated Michaelis–Menten parameters in brain endothelial cells are K _m = 530 μM and V _max = 7039 pmol/million cells/min versus other processes; K _m = 923 mM and V _max = 3656 pmol/million cells/min. Although the transport rate is increased by about 10-fold with LAT1, it still operates on a millisecond time scale. The structural conservation of the LeuT fold with LAT1 further supports the plausibility of an alternating-access transport mechanism.

In this study, the structural dynamics and key functional domains of the LAT1 involved in phenylalanine transport were investigated using replicate long-time scale molecular dynamics (MD) simulations. Because the LAT1-mediated transport processes occur on the millisecond time scale and are beyond the reach of conventional MD simulations, we employed targeted molecular dynamics (TMD) and steered molecular dynamics (SMD) simulations to elucidate the transport mechanism and LAT1 conformational states during the phenylalanine translocation path. These integrated computational strategies facilitate the identification of key residues and structural domains involved in phenylalanine recognition, binding, and translocation through the LAT1. The resulting insights offer a valuable framework for the rational design of LAT1 inhibitors as potential therapeutic alternatives for the treatment of phenylketonuria (PKU).

Materials and Methods

Two crystal structures for LAT1 complexed with substrate are available in the PDB data bank, which represents HNX (2∼{S})-2-azanyl-7-(naphthalen-1-ylmethoxy)-3,4-dihydro-1∼{H}-naphthalene-2-carboxylic acid bound outward-occluded structure (PDB 7DSK) and BCH (2 amino-2-norbornanecarboxylic acid) bound inward-open structure (PDB 6IRT). At the time we started this study, no LAT1 structures with bound phenylalanine were available, so these models were used as the starting structures for molecular docking and MD simulations. To our surprise, structural comparison with the cryo-electron microscopy structure, representing the outward-facing conformation, revealed a very similar phenylalanine binding mode to that observed in our docked LAT1–Phe complex for the outward-occluded structure used for MD simulations (Supporting Information, Figures S1 and S2).

Phenylalanine Docking to LAT1

Molecular docking was performed using AutoDock Vina to predict the binding orientation of l-phenylalanine within 7DSK and 6IRT structures. First, the substrates were removed from these structures. For 7DSK, docking calculations were carried out using a cubic grid box of 25 × 25 × 25 ų, centered at coordinates (−0.6481, 0.2555, 0.4586), corresponding to the center of mass (COM) of the protein. Similarly, the docking calculations using 6IRT were performed using a cubic grid box of 25 × 25 × 25 ų, centered at the protein center of mass (COM) located at coordinates (−0.9321, 0.8297, 0.3184). Using each LAT1 model structure as a target, nine distinct binding poses of the phenylalanine substrate were generated. Binding affinities were evaluated for all poses, and the lowest-energy conformations were selected as the most favorable binding modes. The best docking scores were −5.9 kcal·mol^–1 for the outward-occluded structure and −6.0 kcal·mol^–1 for the inward-open structure (Supporting Information, Table S2). Structural comparisons with experimentally resolved ligand-bound LAT1 structures revealed that the docked phenylalanine adopts binding positions closely overlapping with the native ligands (Supporting Information, Figures S1 and S2). More importantly, a recent cryo-EM structure of LAT1 in an outward-facing conformation with bound phenylalanine reveals a binding pose that closely matches our docked phenylalanine-LAT1 complex in the outward-facing state (Supporting Information, Figure S1). The lowest energy docked complexes were subsequently used as the starting configurations for molecular dynamics simulations to assess binding stability and to investigate substrate translocation mechanisms.

MD Simulation

All simulation systems were constructed using the Membrane Builder module of CHARMM-GUI. A target cubic simulation box of approximately 110 × 110 × 110 ų was employed. The membrane of the outward-occluded protein consisted of a symmetric lipid bilayer comprising 404 lipid molecules (202 per leaflet), designed to reproduce the average lipid composition of human brain microvascular endothelial cell (BMEC) membranes. , The bilayer composition included cholesterol (CHL, 116 molecules; 28.7%), OSM (76 molecules; 18.8%), SAPE (60 molecules; 14.9%), SOPE (32 molecules; 7.9%), SLPC (32 molecules; 7.9%), SAPS (32 molecules; 7.9%), SAPC (32 molecules; 7.9%), POPC (16 molecules; 4.0%), and SAPI (8 molecules; 2.0%), with lipids evenly distributed between the two leaflets. The inward-open state membrane was also modeled using the identical lipid composition and proportional distribution of all components. All of the systems were built with the all-atom CHARMM36 force field incorporating CMAP corrections, the TIP3 water model, and the CHARMM additive force field for lipids (see Supporting Information, Figure S3, for a setup view of the system). The depth of the water solvent layer above and below the bilayer was 25 Å. Potassium and chloride ions were added to neutralize the system and achieve a physiological salt concentration of 0.15 M KCl. This resulted in the addition of 106 K⁺ and 69 Cl^– ions for the outward-occluded LAT1 system, and 113 K⁺ and 75 Cl^– ions for the inward-open LAT1 system. LAT1 was centered at Z = 0 with the membrane normal parallel to the Z axis. Periodic boundary conditions were applied using a cuboid cell with starting dimensions of 108.8 × 108.8 × 108.4 ų. The minimization and MD simulations were carried out utilizing the NAMD 2.14 software suite. , Both the systems were minimized for 1000 steps with a combination of steepest descent and conjugate gradient algorithm to relax atomic positions. Equilibration is carried out in multiple stages, gradually lifting restraints across the system, starting with amino acid side chains, followed by solvent molecules, ions, lipids, and finally the entire protein. As recommended by CHARMM-GUI, six equilibration phase input files were created using both NVT and NPT ensembles, and each of these equilibration steps was run for 1 ns to ensure a smooth transition to a fully equilibrated system. To further verify the stability of these systems, equilibration simulations were extended to 100 ns with 0.1 kcal·mol^–1·Å^–2 positional restraints, for both the outward-occluded (7DSK) and inward-open (6IRT) LAT1 structures (Supporting Information, Figure S4). The production molecular dynamics (MD) simulation for the outward-occluded structure was carried out as three independent replicas, with two replicas of 230 ns each and one extended replica of 370 ns, during which phenylalanine translocated from the transporter to the extracellular side. In contrast, the inward-open structure was simulated as three independent replicas of 1000 ns each. Both simulations were performed using a 2 fs integration time step. The pressure was maintained at 1 atm using a Nosé-Hoover Langevin piston control, and the temperature was kept at 303.15 K by coupling to a Langevin thermostat. Long-range electrostatic interactions were treated with the particle mesh Ewald algorithm. The nonbonded interactions were smoothly cutoff at 12 Å, and the nonbonded neighbor list was updated at every time step.

Steered MD Simulation

As phenylalanine did not translocate to the extracellular region even after replicating 1000 ns MD simulations, we subsequently employed steered MD simulations to promote its transport. In this study, constant velocity pulling steered molecular dynamics (SMD) simulations were performed, in which the selected SMD atom (α-carbon atom of phenylalanine substrate) is connected to a dummy atom via a virtual spring, and a time-dependent harmonic potential U (r, t) is incorporated into the system’s Hamiltonian. The force between the two atoms is measured using the following equation

$$\mathrm{F⃗}=-\nabla U$$

$$U={\frac{1}{2}k[\mathit{vt}-(\mathrm{r⃗}-{\mathrm{r⃗}}_0)·\mathrm{n⃗}]}^2$$

Here, the potential energy is represented by U, while k denotes the spring constant. The velocity of pulling is indicated by v, and t refers to time. r⃗ is the actual position of the SMD atom, r⃗₀ is the initial position of the SMD atom, and last n⃗ is the direction of pulling. This setup enables the SMD simulation to effectively pull the phenylalanine out of the LAT1 protein into the intracellular site. The SMD simulations were carried out using the NAMD 2.14 software suite. Multiple trial SMD simulations were performed to optimize the parameters required for capturing the unbinding of phenylalanine from LAT1 in the occluded state and its subsequent translocation toward the intracellular side. The final parameters were set with a spring constant of 0.3 kcal/mol·Å² and a pulling velocity of 0.000004 Å/ps. The SMD direction was defined as X = 0, Y = 0, and Z = −1.0. A total of six replicate SMD simulations were performed, each with a duration of 5 ns.

Targeted MD Simulation

In targeted molecular dynamics (TMD), a restraint energy term is incorporated into the total energy to investigate large-scale transitions between two known conformations of LAT1, the outward-open state (a structural model obtained from a 230 ns MD simulation of 7DSK LAT1 structure) and the inward-open state (a structural model obtained from an SMD simulation where Phe is outside in the intracellular side) of LAT1. The restraint energy term is proportional to the square of the difference between two root-mean-square deviation (RMSD) values: (a) the RMSD from the current coordinates to the final coordinates, and (b) the RMSD that would occur from the current coordinates to the final coordinates if a linear transition were to occur between the initial and target coordinates. The strength of the force pulling the system toward its target coordinates is expressed by k, which represents the spring constant. The force on each atom is given by the gradient of the potential

$${U_{\mathrm{TMD}}=\frac{1}{2}\frac{k}{N}[\mathrm{RMS}(t)-{\mathrm{RMS}}^{*}(t)]}^2$$

where RMS­(t) is the instantaneous best-fit RMS distance of the current coordinates from the target coordinates and RMS*­(t) evolves linearly from the initial RMSD at the first TMD step to the final RMSD at the last TMD step. The spring constant k is scaled down by the number N of targeted atoms. ,

To investigate the transport pathway, a total of six TMD simulations were performed. The simulations were run for 10 ns with a k of 100 kcal/mol/Ų, which determines the magnitude of the elastic potential that attracts each atom to its final coordinates. Simulations were performed at a temperature of 303.15 K, which was maintained using a Langevin thermostat. The pressure was set to 1 bar and controlled using a Nosé-Hoover Langevin piston. The integration time step was 2 fs, and simulation data, including atomic coordinates and velocities, were saved every 0.5 ps. The simulations were carried out using NAMD software, and the systems were prepared using the Charmm-Gui membrane builder. ,,,

In addition, VMD was used for visualization, while Discovery Studio facilitated the identification of potential intermolecular interactions between phenylalanine and residues across different conformational states of LAT1.

Principal Component Analysis (PCA) and Conformational Clustering Analysis

Principal component analysis (PCA) was performed using aligned backbone Cα coordinates extracted from the MD, SMD, and TMD trajectories to characterize the dominant collective motions associated with phenylalanine transport through LAT1. Prior to PCA, all trajectory frames were structurally aligned to the reference LAT1 structure (first frame) to remove global translational and rotational motion. The covariance matrix of backbone Cα positional fluctuations was diagonalized, and trajectories were projected onto the first two principal components (PC1 and PC2), which captured the dominant transport-associated motions.

To visualize conformational sampling, the PC1–PC2 projections were converted into relative density landscapes using Gaussian kernel density estimation (KDE) implemented through the scipy.stats.gaussian_kde module in Python. Density values were normalized relative to the maximum sampled density within each data set, and low-density regions (3% of the maximum density) were masked to improve visualization clarity. The resulting KDE-smoothed distributions were rendered as filled contour maps. Highly populated conformational regions identified from the PCA density landscapes were selected as representative transport intermediates. Representative structures were obtained from frames nearest to the local density maxima within each region. Additional low-density (LD) conformations surrounding the major PCA regions were also selected for comparative Boltz-2 binding affinity analysis.

To characterize the conformational states sampled during the SMD and TMD simulations, the displacement vector of the phenylalanine substrate relative to its position in the intracellular space was computed for each trajectory frame using MDAnalysis. The displacement was decomposed into Cartesian components (dX, dY, dZ), and the scalar Euclidean distance to the target position was calculated per frame over the simulation trajectory. K-means clustering (k = 3 for SMD and K = 6 for TMD) was applied to the four-dimensional feature space (dX, dY, dZ, distance) after standardization. Clusters were ordered by decreasing mean distance from the target position. The centroid of each cluster was identified as the frame closest to the cluster mean in feature space. The conformational changes were visualized in the dZ–dY subspace, with centroid positions connected sequentially to illustrate the substrate transition.

Binding Affinity Prediction Using Boltz-2

To evaluate the relative binding stability of phenylalanine (Phe) across different conformational states of LAT1, binding affinity predictions were performed using the Boltz-2 deep learning framework. Representative structures were extracted from high-density regions identified through principal component analysis (PCA) of steered molecular dynamics (SMD) and targeted molecular dynamics (TMD) simulations. For each cluster, a corresponding noncluster structure was selected from nearby conformations outside the defined PCA basins to enable direct comparison.

The amino acid sequence of LAT1 was provided, and the ligand was defined using its SMILES representation in the .yaml file. Each representative LAT1–Phe complex was also supplied as a template structure during prediction. The ligand binding site was specified by defining surrounding residues within a 6 Å radius of the Phe position in the template structure (as pocket contacts in the yaml file), ensuring consistent localization of the binding region across predictions. Boltz-2 affinity predictions were performed using the predict module, with multiple sequence alignment (MSA) generation enabled via the ColabFold MSA server and inclusion of learned interaction potentials.

For each input, Boltz-2 outputs two primary metrics: affinity_pred_value and affinity_probability_binary. The predicted binding affinity (affinity_pred_value, denoted as y) from Boltz-2, expressed as log₁₀ (IC₅₀ in μM), was converted to pIC₅₀ (in kcal/mol) using

$${\mathrm{pIC}}_{50}(\mathrm{kcal}/\mathrm{mol})=(6-y)\times 1.364$$ 1

The affinity_probability_binary corresponds to the predicted probability of the ligand being a binder. In this study, pIC₅₀ values were used to compare relative binding strengths across different conformational states, with higher pIC₅₀ indicating stronger predicted binding. All cluster and noncluster conformations were evaluated independently under identical computational settings.

A schematic overview of the complete computational workflow employed in this study is shown in Figure . The flowchart integrates the sequential steps of molecular docking, conventional molecular dynamics (MD), steered molecular dynamics (SMD), and targeted molecular dynamics (TMD) simulations applied to different LAT1 conformational states, providing a consolidated representation of substrate transport modeling.

2.

Computational workflow to investigate LAT1 conformational states and phenylalanine (Phe) transport through LAT1. Phenylalanine was docked into the outward-occluded (PDB ID: 7DSK) and inward-open (PDB ID: 6IRT) structures of LAT1, followed by (1) three independent replicate conventional molecular dynamics (MD) simulations for each system to obtain conformational states of LAT1 and phenylalanine translocation pathway, (2) six replicate steered molecular dynamics (SMD) simulations to promote phenylalanine translocation toward the intracellular space as phenylalanine did not exit in the intracellular space with conventional MD simulations, (3) six targeted molecular dynamics (TMD) simulations to confirm the complete transport pathway and to obtain important LAT1 conformational states.

Results and Discussion

Phenylalanine Transport from Outward-Occluded to Outward-Open State: Key Conformational States of LAT1 Revealed by Replicate MD Simulations

Three replica molecular dynamics (MD) simulations (each ≥ 230 ns) of the outward-occluded state of LAT1 reveal that the phenylalanine, which was initially located near the center of LAT1, Structure I (Figure C­(I)), migrates to the extracellular site, Structure IV (Figure C­(IV)) of LAT1. Structure I represents the outward-occluded conformation of LAT1 in complex with phenylalanine, docked on the LAT1 structure derived from the reference PDB ID 7DSK. A recent cryo-EM structure of LAT1 in an outward-facing conformation with bound phenylalanine closely agrees with the binding orientation observed in our docked outward-facing LAT1–Phe complex (Supporting Information, Figure S1). The MD simulation reveals the pathway by which phenylalanine translocates from the occluded site to the outward site, as well as the specific amino acids of LAT1 that interact with phenylalanine (Figure ). A video clip of the MD simulation, provided in the Supporting Information, illustrates the phenylalanine transport from the outward-occluded to the outward-open state of LAT1. The stability of the LAT1 structure during the MD simulation was a crucial aspect of our analysis, and to evaluate this, two key metrics, RMSD and root-mean-square fluctuation (RMSF), were employed. The RMSD of the protein backbone Cα atoms was performed to determine if the structure of LAT1 remains stable throughout the MD simulation. The RMSD plot, as shown in Figure B, indicates that the LAT1 structure deviates by approximately 2–3 Å from the initial geometry. Furthermore, the RMSF calculation provided valuable insights into the dynamic behavior of LAT1 residues, quantifying their average deviation from their initial positions over time. The RMSF analysis shows the fluctuations of the residues of LAT1 (Supporting Information, Figure S5). By analyzing the RMSF values, regions of the protein that exhibit high flexibility due to interaction with phenylalanine and those that remain relatively rigid during phenylalanine transport can be identified. Overall, increased residue fluctuations have been observed in residues 65–75, 155–170, 220–240, 300–330, 365–375, 385–400, 420–430, and 480–490.

3.

Phenylalanine transport from the outward-occluded to outward-open state of LAT1 revealed by replicate MD simulations. (A) PCA population landscape (PC1–PC2) showing four high-density conformational basins (Structures I–IV) identified during phenylalanine translocation. (B) RMSD of LAT1 backbone Cα atom obtained from the triplicate MD simulations. (C) Representative phenylalanine (pink) binding positions within each LAT1 conformational state (Structures I–IV). (D) Residues of LAT1 interacting with phenylalanine within each LAT1 conformation state.

From the MD simulation, it was possible to identify several conformational states of LAT1, represented with Structures I to IV, upon substrate transport from outward-occluded to outward-open conformation (Figure C). The population of these conformational states was obtained by using principal component analysis, which depicts the system’s energy as a function of its conformational coordinates (Figure A and Supporting Information, Figures S6 and S23). The red color indicates regions of increased population density, indicating a conformational state that LAT1 prefers after phenylalanine binding, whereas the blue color indicates regions of low population density. The PCA analysis, as shown in Figure A, revealed four conformationally stable structures in LAT1 during phenylalanine transport, the last and the final structure, Structure IV (Figure C­(IV)), representing an outward-open conformation where phenylalanine is outside the protein.

These structures were then examined to identify the residues that are interacting with the phenylalanine during its transport. Figure D depicts the interactions between phenylalanine and LAT1 residues in the LAT1 Structures I, II, III, and IV (Figure C). The frequency of phenylalanine interacting with residues of LAT1 has shown higher occupancies for Ser144 (88%), Phe252 (86%), Gly255 (82%), Ser66 (77%), Val148 (76%), Gly65(75%), Ile63(75%), and Ala253 (74%) (Supporting Information, Table S3). Phenylalanine interacts with Gly65, Ser144, Ser66, Phe252, and Gly255 in Structure I; Ile63, Gly65, Ser66, Val148, Phe252, and Ala253 in Structure II; Gly65, Ser66, Val148, and Phe252 in Structure III; and Tyr155, Lys158, Val313, and Pro328 in Structure IV. In Structure I, the carbonyl oxygen of phenylalanine forms a hydrogen bond with the hydroxyl group of Ser66 (Phe^O/Ser66^HO), while the flexible residues Gly255 and Gly65 adopt favorable conformations to accommodate the phenylalanine, and the phenylalanine also engages in hydrophobic interactions with Phe252. In a recent cryo-EM structure of LAT1 in an outward-facing state, bound phenylalanine is also coordinated by residues including Phe255, Ser66, and Gly255. In this configuration, the aromatic ring of l-Phe is positioned toward Gly255 and engages in van der Waals interactions with its Cβ atom, supporting earlier evidence that Gly255 is important for accommodating larger amino acid substrates. In Structure II, phenylalanine forms backbone-mediated hydrogen bonds with Ser66 (Phe^OH/Ser66^N), Phe252 (Phe^NH/Phe252^O), Gly255 (Phe^NH/Gly255^O), and participates in a nonpolar interaction with Val148. In Structure III, phenylalanine engages in hydrophobic interactions with Val148 and forms hydrogen bonds with Ser66 (Phe^O/Ser66^NH) and Phe252 (Phe^NH/Phe252^O). Finally, in Structure IV, phenylalanine engages in hydrophobic interactions with Val313 and Pro328, and a π–π stacking interaction with Tyr155 (Supporting Information, Figure S7).

Key Conformational States of LAT1 Occluded and Inward-Occluded States Revealed by Replicate Long-Time Scale MD simulations

As described in the Materials and Methods section, phenylalanine was docked into the inward-open LAT1 structure obtained from PDB ID: 6IRT. Following energy minimization and 100 ns equilibration of the fully solvated system, the substrate may have adopted a confined binding state that resembled an inward-occluded or occluded state, as three independent 1000 ns molecular dynamics simulations showed no phenylalanine release events. Four major conformational states were identified based on the PCA density landscape (Figure A and Supporting Information, Figures S8 and S24) in which phenylalanine release was hindered. Although the PCA density landscapes from the three MD simulations show slight variations along the principal components, they primarily represent occluded conformational states, as phenylalanine does not exit through LAT1 on either the intracellular or extracellular side in any of the simulations. The stability of LAT1 throughout the MD simulation was assessed by analyzing the RMSD of the protein. The RMSD plot indicates that the LAT1 structure deviates by approximately 2.5–3.5 Å for all backbone Cα atoms of LAT1 (Figure B). The RMSF analysis highlights LAT1 residue fluctuation during MD simulations, where increased residue fluctuations were observed in residues ranging from 110–120, 160–170, 220–240, 250–270, 355–370, 420–435, 450–460, and 475–490 (Supporting Information, Figure S9). This suggests that certain regions of the protein are more flexible and prone to conformational changes. PCA analysis reveals four distinct conformational clusters of LAT1 during phenylalanine translocation, as shown in Figure A, while the last structures (Figure C­(IV)) may correspond to the inward-occluded state. Figure D depicts the interactions between phenylalanine and LAT1 residues for the Structures I, II, III, and IV, as shown in Figure C. The frequency of phenylalanine interacting with residues of LAT1 has shown higher occupancies for Ser338 (85%), Tyr259 (83%), Gly337 (81%), Ala201(77%), Lys204(77%), Thr62 (76%), and Gly61 (62%) (Supporting Information, Table S4). In Structure I, phenylalanine interacts with Ser338, Ala201, Lys204, and Gly337, while in Structure II, it engages with Arg141, Gly256, Gly341, and Ser342; in Structure III, it interacts with Thr62, Ala201, and Lys204, whereas in Structure IV, it forms contacts with Gly61, Thr62, Ala201, Lys204, Gly337, and Ser338. In Structure IV, Phe forms hydrogen bonds with backbone amide groups of Gly61 (Phe^OH/Gly6^O) and Lys204 (Phe^OH/Lys204^N), and it also undergoes hydrophobic interaction with Ala201 and Gly337 (Figure S10).

4.

Key conformational states of LAT1 during inward-occluded phenylalanine binding revealed by replicate long-time scale MD simulations. (A) PCA population landscape (PC1–PC2) showing four high-density conformational basins (Structures I–IV) identified during the 1000 ns simulations, in which phenylalanine remains confined within LAT1. (B) RMSD of LAT1 obtained from the triplicate 1000 ns MD simulations. (C) Representative phenylalanine (pink) binding positions within each LAT1 conformational state (Structures I–IV). (D) Residues of LAT1 interacting with phenylalanine within each LAT1 conformation state.

Phenylalanine Transport from Inward-Occluded to Inward-Open State: Key Conformational States of LAT1 Revealed by Steered MD Simulations

Since phenylalanine remained trapped in the occluded or inward-occluded states during the MD simulations, we next performed six replicate 5 ns steered molecular dynamics (SMD) simulations using the final MD snapshot (Figure C­(IV)), in which phenylalanine is positioned near the intracellular exit of LAT1 yet remains occluded. We applied a time-dependent external force to guide phenylalanine downward toward the intracellular region of the membrane protein system. All six SMD simulations, within 1 ns (Table S5), showed phenylalanine translocating toward the extracellular space along the same pathway (Figure and Supporting Information, Figure S11). PCA analysis (Figure A and Supporting Information, Figures S11 and S25) of the SMD simulations revealed three distinct conformational states of LAT1 (Structures I–III in Figure D) during phenylalanine transport, with the final state, Structure III (Figure D­(III)), corresponding to the inward-open conformation. However, as these PCA projections were derived from biased nonequilibrium SMD trajectories, the resulting conformational density landscapes do not represent equilibrium free-energy surfaces. Nevertheless, the qualitative features of these landscapes may still be accurate and provide useful insight into the conformational space sampled during transport, which may minimally enable identification of major conformational states and transition intermediates along the transport pathway. To further characterize the conformational transitions observed during transport, we performed additional structural clustering analysis on the trajectories generated from the SMD simulations to identify representative conformational states with greater accuracy, although the phenylalanine moved rapidly under the applied pulling force toward the intracellular side, resulting in fast conformational transitions along the transport pathway. Still, we were able to extract similar conformations as previously identified from the PCA analysis (Supporting Information, Figure S21). The clusters were generated by computing the displacement vector of the phenylalanine substrate relative to its intracellular reference position, decomposing the trajectory into dX, dY, and dZ components, and applying k-means clustering (k = 3) in the displacement features space.

5.

Phenylalanine transport from the inward-occluded to inward-open state of LAT1 revealed by replicate steered MD (SMD) simulations. (A) PCA population landscape (PC1–PC2) showing three high-density conformational basins (Structures I–III) identified during SMD-driven phenylalanine translocation. (B) RMSD of LAT1 backbone Cα atom obtained from the six independent 5 ns SMD simulations. (C) Predicted binding affinity (pIC₅₀) of phenylalanine (Phe) to LAT1 across the three conformational states identified from PCA analysis was compared with three low-density (LD) basins around the conformation states. (D) Representative phenylalanine (pink) binding positions within each LAT1 conformational state (Structures I–III). (E) Residues of LAT1 interacting with phenylalanine within each LAT1 conformation state.

At the start of each simulation (t = 0 ns), the force required to initiate phenylalanine release was approximately 500 pN, reflecting strong interactions with Gly61, Thr62, Ala201, Lys204, Gly337, and Ser338 (Figure E­(I)). The force profile gradually declined as phenylalanine became less bound to the protein residues (Supporting Information, Figure S12). The RMSD analysis of all six SMD simulations shows that the LAT1 structure deviates by a maximum of 1.2 Å from its initial state, suggesting that pulling phenylalanine downward does not cause abrupt structural variations in LAT1 (Figure B). RMSF analysis of LAT1 indicates elevated flexibility in residues 110–120, 160–170, 230–240, 260–270, 360–370, 420–430, and 480–490, although the fluctuations do not exceed 1.3 Å (Supporting Information, Figure S13).

To evaluate the relative stability of the substrate along the forced transport pathway in SMD simulation, the predicted binding affinity (pIC₅₀) for the three high-density conformational structures (Structure I, II, and III) and three nearby low-density structures (LD I, II, and III) were compared with Boltz-2 (Figure C). Structures I, II, and III consistently exhibited higher pIC₅₀ values, indicating stronger binding affinity compared to the LD structures (Figure B). Structures I and II display the highest binding affinity, pIC₅₀ ≈ 6.08 and 6.37 kcal/mol, respectively, reflecting stable substrate engagement within intermediate regions of the transport pathway. Structure III, which is located near the intracellular exit gate, shows reduced binding affinity with pIC₅₀ ≈ 5.19 kcal/mol. Overall, all of the LD structures consistently show reduced binding with pIC₅₀ ≈ 4.49–5.05 kcal/mol (Figure B). The relatively small changes in Boltz-2-predicted affinity, ranging from 0.70 to 1.32 kcal/mol for conformations I, II, and III, are expected because the analyzed structures represent sequential transport intermediates rather than distinct binding modes. The results suggest that LAT1 maintains favorable phenylalanine interactions throughout the transport, with highly populated PCA conformations showing slightly stronger substrate stabilization

The Structures I–III obtained from SMD simulation were analyzed to investigate potential interactions between phenylalanine and LAT1 residues, as shown in Figure E (also in Supporting Information, Figure S14). The frequency of phenylalanine interacting with residues of LAT1 has shown higher occupancies for Tyr259 (59%), Thr62 (45%), Ala201 (35%), Gly337(30%), Gly61 (29%), Ser338 (25%), and Lys204 (21%) (Supporting Information, Table S6). In Structure I, phenylalanine interacted with Gly61, Thr62, Ala201, Lys204, Gly337, and Ser338. Structure II revealed interactions with Thr62, Lys204, Tyr259, Gly337, and Ser338, and Structure III showed interactions between the phenylalanine and Leu370 and Leu371, residues located at the intracellular exit of LAT1. In Structure I (corresponding to Structure IV in Figure D), phenylalanine undergoes a hydrophobic interaction with Ala201, Gly337, and Ser338, and is tightly surrounded by Gly61 and Gly337, indicating a stabilized binding environment (Supporting Information, Figure S14).

In Structure II, phenylalanine showed hydrogen bonds with Tyr259 (Phe^OH/Tyr259^O) and Lys204 (Phe^O/Lys204^NH), and is also surrounded by Ser338 and Gly337. In Structure III, phenylalanine exhibited hydrophobic interactions with Leu371 and Leu370. A video clip of the combined MD and SMD simulation showing phenylalanine transport toward the intracellular side of LAT1 is provided in the Supporting Information.

Phenylalanine Transport from Outward-Open to Inward-Open State: Key Conformational States of LAT1 Revealed by Targeted MD Simulations

Six independent targeted MD (TMD) simulations were employed to identify key conformational states of LAT1 during phenylalanine transport and to assess whether these conformational states resemble those observed in MD and SMD simulations. PCA analysis of the TMD simulations identified six LAT1 conformations during phenylalanine transport, ranging from an outward-open (Structure I) to an inward-open state (Structure VI), as illustrated in Figure . The plots for the PCA analysis for the remaining five simulations are shown in Supporting Information (Figures S15 and S26). These simulations required approximately 10 ns to transport phenylalanine from the extracellular to the intracellular space (Supporting Information, Table S7). All TMD simulations consistently demonstrate a similar pathway and conformational transitions of LAT1 during phenylalanine transport. Similar to the SMD analysis, the PCA projections derived from biased nonequilibrium TMD trajectories do not yield conformational density landscapes that represent equilibrium free-energy surfaces. Nevertheless, the qualitative features of these landscapes may still be accurate and provide useful insight into the conformational space sampled during transport, while minimally enabling identification of major conformational states and transition intermediates along the transport pathway. To further characterize the conformational landscape revealed during Phe transport, we performed an additional clustering analysis on the trajectories generated from the TMD simulations to identify representative conformational states more accurately. A comparison of the clustering results with the PCA analysis showed that the representative conformations previously identified from the PCA projections aligned closely with the dominant clustering regions obtained from the trajectory analysis, particularly for Structures I to VI (Supporting Information, Figure S22). This agreement between the two analytical approaches provides a more robust structural interpretation of the transport pathway and further supports the conformational transitions observed during the phenylalanine transport.

6.

Phenylalanine transport from the outward-open to inward-open state of LAT1 revealed by replicate targeted MD (TMD) simulations. (A) PCA population landscape (PC1–PC2) showing six high-density conformational basins (Structures I–VI) identified during TMD-driven phenylalanine translocation. (B) RMSD of LAT1 backbone Cα atom obtained from six independent 10 ns TMD simulations. (C) Predicted binding affinity (pIC₅₀) of phenylalanine (Phe) to LAT1 across the six conformational states identified from PCA analysis was compared with six low-density (LD) basins around the conformation states. (D) Representative phenylalanine (pink) binding positions within each LAT1 conformational state (Structures I–VI). (E) Residues of LAT1 interacting with phenylalanine within each LAT1 conformation state.

A video clip of the TMD simulation, provided in the Supporting Information, illustrates the transport process of phenylalanine through LAT1. RMSD analysis of LAT1 backbone Cα atoms for all six TMD simulations indicates that conformational transitions, including outward-open, occluded, and inward-open states, occur smoothly, with the final inward-open structure deviating by approximately 4 Å from the outward-open state (Figure B). Additionally, RMSF analysis highlights significant fluctuations in specific residues, particularly in the regions spanning 110–120, 160–170, 180–190, 220–240, 260–280, 315–325, 365–375, 380–390, 415–435, and 470–490, with some residues showing RMSF between 3 and 8 Å (Supporting Information, Figure S16).

To evaluate the relative stability of the substrate along the forced transport pathway in TMD simulation, the predicted pIC₅₀ for the six high-density conformational structures (Structures I, II, III, IV, V, and VI) and six nearby low-density structures (LD I, II, III, IV, V, and VI) were compared with Boltz-2 (Figure C). All of the structures, I to VI, consistently exhibited higher pIC₅₀ values (5.44–6.89 kcal/mol), indicating stronger substrate-binding affinity compared to the LD I to VI (4.37–5.18 kcal/mol) (Figure C). Binding affinity increases in intermediate states, with Structures III and IV exhibiting the highest pIC₅₀ values, 6.85 and 6.89 kcal/mol, respectively, indicating optimal stabilization of phenylalanine within the central binding pocket. Toward the intracellular side, Structure VI shows reduced binding affinity, pIC₅₀ = 5.44 kcal/mol, consistent with the progressive weakening of interactions during substrate release. Similar to the SMD results, the relatively small differences in the Boltz-2-predicted affinity observed in TMD, ranging from 0.51 to 2.42 kcal/mol for conformations I, II, III, IV, V, and VI, suggest that LAT1 maintains favorable interactions with phenylalanine throughout the transport process. Highly populated PCA-derived conformations or clusters exhibit slightly stronger substrate stabilization, supporting their role as favorable transport intermediates.

The structures identified in Figure D were further analyzed to determine the LAT1 residues interacting with phenylalanine during its transport. The first and last structures of phenylalanine do not interact much with the residues of LAT1, as they are almost out of the protein. Figure E illustrates the interaction of phenylalanine with residues within the Structures II, III, IV, and V. The particular type of interaction is shown in Supporting Information, Figure S17. The frequency of phenylalanine interacting with residues of LAT1 has shown higher occupancies for Tyr259 (45%), Arg141 (35%), Ile140 (34%), Gly256 (33%), Gly255 (255%), Phe252 (24%), Thr345 (24%), Ser338 (20%), Val70(16%), and Val151(15%) (Supporting Information, Table S8). In Structure II, phenylalanine interacts with Gly67, Val70, and Val151, while in Structure III, Gly255, Gly256, and Ile140 interact with phenylalanine. Structure IV reveals interactions with Arg141, Ser338, and Thr345, whereas in Structure V, phenylalanine interacts with Thr194 and Tyr259. In Structural model II, phenylalanine engages in hydrophobic interactions with Val70 and Val151. In Structural model III, phenylalanine is stabilized through hydrophobic interaction with Ile140, Gly255, and Gly256, while in Structure IV, phenylalanine establishes hydrogen bonds with Thr345^O and Ser338^O and electrostatic interaction with Arg141. In Structure V, phenylalanine demonstrates hydrogen bond with Thr194 (Phe^OH/Thr194^O) and Pi-alkyl interaction with Tyr259. During the TMD, Phe350 exhibited minimal fluctuation compared to other residues, as indicated by its low RMSF (about 3 Å), making it a reliable reference point for tracking the time-dependent change in distance between the Cα atoms of the substrate phenylalanine and Phe350 as LAT1 transitions from the outward-open to the inward-open conformation. MD and SMD simulations also showed minimal RMSF for Phe350 (see Supporting Information, Figures S5, S9, and S13). Across all six TMD simulations, the Cα–Cα distance between phenylalanine and Phe350 began at an average of ∼42 Å in the outward-open state, when phenylalanine first enters LAT1, and gradually decreased through occluded intermediates to a minimum average of ∼14 Å. The distance then increased slightly to ∼19 Å, as phenylalanine exited via the inward-open state. The slower decrease observed after ∼4 ns reflects the prolonged residence of phenylalanine in occluded LAT1 conformations before its final release (Supporting Information, Figure S18).

Key Phenylalanine–LAT1 Interactions Conserved across MD, SMD, and TMD Simulations

A comparative analysis of LAT1 structures obtained from molecular dynamics (MD), steered molecular dynamics (SMD), and targeted molecular dynamics (TMD) simulations reveals conserved phenylalanine binding modes during transport across the transporter. Occupancy analysis of different residues interacting with the Phe over time reveals key residues that are involved with Phe binding (as shown in Supporting Information, Table S9 and Figure S19). In the outward-occluded state, phenylalanine consistently interacts with Gly255 in both MD (Structure I, Figure D) and TMD simulations (Structure III, Figure E) with occupancies of 82 and 30%, respectively. As phenylalanine progresses slightly toward the intracellular side within the occluded conformation, it interacts with Ser338 in both MD (Structures I and IV, Figure E) and TMD (Structure IV, Figure E) with occupancies of 85 and 20%, respectively, as well as with Arg141 in MD (Structure II, Figure E) and TMD (Structure IV, Figure E) with occupancies of 27 and 35%, respectively. When phenylalanine approaches the exit site while LAT1 remains largely closed, still in inward-occluded states, it interacts with Tyr259, as observed in SMD (Structures II, Figure E) and TMD (Structure V, Figure E) with occupancies of 59 and 45%, respectively. These findings highlight Gly255, Ser338, Arg141, and Tyr259 as key residues in facilitating the movement of phenylalanine across the membrane (as shown in the Supporting Information, Figure S19 and Table S9) and suggest that they may serve as strategic targets for the design of LAT1 inhibitors aimed at limiting phenylalanine transport.

LAT1 Conformational States from Extracellular and Intracellular Views

This study identifies key transmembrane helices (TMs) that undergo significant conformational changes during phenylalanine transport through LAT1. Specifically, TMs involved in phenylalanine entry from the extracellular side and exit toward the intracellular side were compared to the outward-occluded (PDB 7DSK) and inward-open crystal structures (PDB 6IRT) of LAT1 (Figure ). The outward-facing open conformation, obtained from a 230 ns molecular dynamics (MD) simulation and viewed extracellularly (red), was superimposed onto the outward-occluded crystal structure (PDB ID: 7DSK, blue), and the RMSD values for different transmembrane domains are calculated (See Supporting Information, Table S10). This comparison revealed that the transition to the outward-open state primarily involves outward displacement of TM1, TM6, and TM9, with calculated RMSD values of approximately 6.0, 4.5, and 4.0 Å, respectively (Figure A). A similar outward-facing conformation generated via targeted MD (TMD) simulation confirmed these findings, showing the highest shifts for TM1, TM6, and TM9 with RMSD values of 3.8, 2.3, and 2.8 Å, respectively (Figure B).

7.

Outward-facing, occluded, and inward-facing conformational states of LAT1, as viewed from the extracellular (left) and intracellular (right) perspectives. (A) Extracellular view of the outward-facing open model (red) obtained from conventional MD simulation, superimposed on the outward-occluded crystal structure (PDB 7DSK, blue). (B) Extracellular view of the outward-facing open model (red) obtained from TMD simulation, superimposed on the outward-occluded crystal structure (PDB 7DSK, blue), showing a similar change as seen in (A). Transmembrane helices TM1, TM6, and TM9 undergo significant displacement during these conformational changes. (C) Intracellular view of the inward-facing open model (red) obtained from SMD simulation, superimposed on the inward-open crystal structure (PDB 6IRT, blue). (D) Intracellular view of the inward-facing open model (red) obtained from TMD simulation, superimposed on the inward-open crystal structure (PDB 6IRT, blue), showing a very similar change as seen in (C). Major displacements are observed in TM1, TM6, and TM12 during these transitions.

On the intracellular side, the inward-facing open model obtained from steered MD (SMD) simulation was aligned with the inward-open crystal structure (PDB ID: 6IRT, blue). This comparison demonstrated that TM1, TM6, and TM12 undergo major rearrangements during intracellular opening, with RMSD values of 2.9, 1.9, and 3.2 Å, respectively (Figure C). The inward-facing model generated from TMD simulation also showed nearly identical displacements for these helices, especially for TM1, further validating the results (Figure D). Together, these findings indicate that TM1 and TM6 are consistently involved in both extracellular and intracellular transitions, while TM9 and TM12 contribute specifically to the outward and inward openings, respectively. The consistent results across MD, SMD, and TMD simulations highlight the critical role of these transmembrane helices in mediating conformational changes necessary for phenylalanine translocation through LAT1.

Key Transmembrane Domains Involved in Phenylalanine Transport across LAT1

This study provides new mechanistic insights into how LAT1 transports phenylalanine, highlighting a substrate-specific isomerization process in which the transporter undergoes conformational changes to accommodate and translocate phenylalanine. The findings also support the alternating-access model, whereby LAT1 shifts between inward- and outward-facing conformations to facilitate substrate exchange across the membrane. To quantify conformational dynamics during phenylalanine translocation, root-mean-square deviations (RMSD) were calculated for each transmembrane helix (TM) across six structural intermediates (Structures I–VI) derived from targeted molecular dynamics (TMD) simulations (see Supporting Information, Table S11 for detailed RMSD values). Seven key helices, TM1, TM4, TM6, TM7, TM8, TM9, and TM12, were identified as major contributors to substrate-induced rearrangements. A comparison of the RMSD and the Structure I and Structure II (Figure I,II) reveals significant movement in TM8 and TM9, with RMSD values of 1.4 and 1.3 Å, respectively. As phenylalanine progressed from Structure II to Structure III (Figure II,III), the largest displacements were observed in TM1 (2.9 Å), TM4 (1.4 Å), and TM7 (1.3 Å). During the transition from Structure III to IV (Figure III,IV), TM6 (mainly residues 259–261) opened toward the intracellular side, accompanied by a shift in TM7, with RMSD values of 1.8 and 1.3 Å, respectively. From Structure IV to V (Figure IV,V), TM1 underwent bending to accommodate phenylalanine, while TM8 also moved intracellularly, showing RMSDs of 2.7 and 1.1 Å, respectively. The transition from Structure V to VI (Figure V,VI) involved intracellular opening of TM1, TM6, and TM12, with RMSDs of 1.9, 1.6, and 1.0 Å, respectively. Finally, after phenylalanine exit, the system returned from Structure VI (Figure VI) to outward-facing Structure I (Figure I), where TM1, TM6, and TM9 reopened toward the extracellular space, with corresponding RMSDs of 6.4, 2.8, and 3.2 Å, respectively. These dynamic changes highlight a coordinated series of TM rearrangements that highlight phenylalanine translocation through LAT1, reinforcing the transporter’s role as a flexible and regulated molecular conduit.

8.

Potential phenylalanine transport pathway through LAT1 identified by TMD simulations, highlighting transmembrane domains (1, 4, 6, 7, 8, 9, and 12) that shift to enable substrate movement. Comparable domain motions were confirmed by conventional MD and steered MD simulations.

Similar results were observed from MD and SMD simulations, where substantial movements of TM1, TM4, TM6, TM7, TM8, TM9, and TM12 were identified as phenylalanine translocated through LAT1. The corresponding RMSD values for all of the structures are given in the Supporting Information in Tables S12–S14. For example, during the transition from the extracellular-facing (Figure C­(IV)) to occluded state (Figure C­(I)), TM1, TM4, TM6, and TM9 exhibited significant movement, with RMSD values of 1.5, 1.4, 1.6, and 1.6 Å, respectively (see Supporting Information, Table S12, for RMSDs of all of the TMs). Furthermore, during the transition from the occluded state (Figure C­(I)) to the inward-open state (Figure D­(III)), which corresponds to Structures IV and VI in Figure D, TM1, TM6, TM8, and TM12 exhibited significant conformational changes, with RMSD values of 4.3, 3.0, 2.0, and 2.4 Å, respectively (see Supporting Information, Table S13 for RMSDs of all of the TMs), facilitating the transport of phenylalanine to the intracellular side. Collectively, these findings from TMD, MD, and SMD simulations underscore the coordinated, segmental rearrangement of key transmembrane helices during phenylalanine transport, reinforcing the dynamic nature of LAT1’s alternating-access mechanism.

Following the initial submission of our manuscript, a cryo-EM study reported LAT1 in an outward-facing conformation with bound phenylalanine, revealing a pronounced opening of TM1b and TM6a that facilitates substrate translocation. Consistent with these structural observations, our MD and TMD simulations capture a similar outward-facing state characterized by the coordinated opening of TM1 and TM6 in the presence of bound phenylalanine. In agreement with the cryo-EM structure, the phenyl ring of Phe in our simulations is oriented toward Gly255. As no additional experimentally resolved structures of LAT1 bound to phenylalanine are currently available, our integrated MD, SMD, and TMD simulations provide detailed mechanistic insights into phenylalanine translocation and the associated transmembrane helix rearrangements. Notably, our results further highlight the critical involvement of TM9 and TM12 during the passage of the substrate through the intracellular region. Collectively, these findings support a sequential transport mechanism in which TM1 and TM6 govern substrate binding, extracellular opening, and occluded state stabilization, while coordinated motions of TM9 and TM12 facilitate intracellular gating.

Conclusion

This study provides comprehensive mechanistic insights into the transport of phenylalanine through the L-type amino acid transporter 1 (LAT1), using a combination of replicate molecular dynamics (MD), targeted molecular dynamics (TMD), and steered molecular dynamics (SMD) simulations. By capturing the conformational transitions of LAT1 across outward-facing, occluded, and inward-facing states, we identified key transmembrane helices (TMs), notably TM1, TM4, TM6, TM7, TM8, TM9, and TM12, as central to substrate translocation. RMSD analyses across structural models mainly revealed the TM1 and TM6 to be consistently engaged during both extracellular and intracellular Phe transitions, whereas TM9 and TM12 contribute more specifically to the outward and inward opening events, respectively. The correspondence of these findings across MD, TMD, and SMD simulations affirms the dynamic nature of LAT1 and highlights these helices as potential allosteric or drug-targetable sites. The phenylalanine transport pathway was further characterized by assessing binding interactions, conformational rearrangements, and the sequential engagement of LAT1 domains. Simulations revealed that phenylalanine’s passage is guided by conformationally driven gating mechanisms, consistent with the alternating-access model. Importantly, the inability of phenylalanine to exit in the absence of specific conformational shifts illustrates the finely tuned nature of substrate release, governed by coordinated helix dynamics. In addition to elucidating transport mechanisms, this work lays the foundation for therapeutic development. Our simulations revealed key conserved interactions and critical residues essential for phenylalanine translocation, providing valuable insights that can be utilized in rational drug design. The identification of major TMs involved in this process, TM1, TM6, TM9, and TM12, positions these domains as strategic targets for inhibiting phenylalanine transport in pathological contexts such as phenylketonuria (PKU). Designing LAT1 inhibitors based on these mechanistic insights may facilitate the development of new therapeutic strategies for diseases associated with dysregulated amino acid transport, including cancer, metabolic disorders, and neurological conditions. In summary, this study advances the structural and functional understanding of LAT1, highlights the value of integrative computational approaches, and opens promising avenues for translational applications targeting amino acid transporters.

All data are available upon reasonable request. MD simulations were performed using NAMD 2.14 on the Expanse supercomputer at SDSC through ACCESS support. Ambertools23 was used for RMSD and RMSF analysis. Xmgrace (v5.1.25) was used to generate data plots. VMD (v1.9.4a57) was used to generate molecular graphics and is available at (https://www.ks.uiuc.edu/Research/vmd/). Relative binding affinity predictions were performed using Boltz-2, available at (https://github.com/jwohlwend/boltz).

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jpcb.6c00494.

• Combined MD and SMD simulation video for phenylalanine transport from occluded to towards the intracellular side of LAT1 (MOV)

• MD simulation video for phenylalanine transport from outward-occluded to outward-open state (MPG)

• Phe_docked_lat1_Inward (PDB)

• Phe_docked_lat1_outward (PDB)

• LAT1 transmembrane domain information, docking pose of Phe in LAT1 7DSK and 6IRT structures, RMSD of equilibration simulations for Phe, and protein systems, RMSD and RMSF analyses from replicate MD, SMD, and TMD simulations, PCA cluster plots for all replicate simulations, pulling force vs Time plot for all SMD simulation, comparative RMSD tables for TM domains across MD, SMD, and TMD simulations, binding affinity of Phe in different structures corresponding to different conformational states of LAT1 across SMD and TMD with Boltz-2, and visual representation of phenylalanine-LAT1 interactions at different conformational states during the MD, SMD, and TMD simulations (PDF)

• TMD simulation video for phenylalanine transport from outward-open to inward-open state (MPG)

S.M.I. designed the study, wrote and edited the manuscript with input from all authors. M.M.H. and K.S. performed the simulation studies, conducted analyses, and edited the manuscript. S.M.I. secured funding and supervised the work.

The authors declare no competing financial interest.