Structural Dynamics of Neutral Amino Acid Transporter SLC6A19 in Simple and Complex Lipid Bilayers

ABSTRACT

B0AT1 (SLC6A19) is a major sodium‐coupled neutral amino acid transporter that relies on angiotensin converting enzyme 2 (ACE2) or collectrin for membrane trafficking. Despite its significant role in disorders associated with amino acid metabolism, there is a deficit of comprehensive structure‐function understanding of B0AT1 in lipid environment. Herein, we have employed molecular dynamics (MD) simulations to explore the architectural characteristics of B0AT1 in two distinct environments: a simplified POPC bilayer and a complex lipid system replicating the native membrane composition. Notably, our B0AT1 analysis in terms of structural stability and regions of maximum flexibility shows consistency in both the systems with enhanced structural features in the case of complex lipid system. Our findings suggest that diacylglycerol phospholipids significantly alter the pore radius, hydrophobic index, and surface charge distribution of B0AT1, thereby affecting the flexibility of transmembrane helices TM7, TM12, and loop connecting TM7‐TM8, crucial for ACE2‐B0AT1 interaction. Pro41, Ser190, Arg214, Arg240, Ser413, Pro414, Cys463, and Val582 are among the most prominent lipid binding residues that might influence B0AT1 functionality. We also perceive notable lipid mediated deviation in the degree of tilt and loss of helicity in TM1 and TM6 which might affect the substrate binding sites S1 and S2 in B0AT1. Considerably, destabilization in the structure of B0AT1 in lipid environment was evident upon mutation in TM domain, associated with Hartnup disorder through various structure‐based protein stability tools. Our two‐tiered approach allowed us to validate the use of POPC as a baseline for initial analyses of SLC transporters. Altogether, our all‐atoms MD study provides a platform for future investigations into the structure‐function mechanism of B0AT1 in realistic lipid mimetic bilayers and offers a framework for developing new therapeutic agents targeting this transporter.

1. Introduction

The disposition of drugs from cells is made through drug transporters, which are broadly categorized into influx and efflux transporters. Influx transporters, mostly the solute carrier (SLC) transporters, facilitate drug uptake into cells, while the efflux transporters, largely the ABC transporters, are accountable for moving drugs out of cells. Members of SLC transporter family are diverse and vital group of membrane proteins responsible for the transport of a wide range of solutes, including ions, sugars, amino acids, nucleotides, and organic molecules, across biological membranes [1]. With over 450 members classified into 65 families, the SLC transporters are integral to various physiological processes, such as nutrient absorption, waste removal, and the maintenance of cellular homeostasis [2]. SLC transporters are primarily either facilitative or secondary‐active. More precisely, they either use an electrochemical gradient to assist in the movements of substrates across membranes or rely on ion gradients created by ATP‐dependent pumps to transport substrates against their concentration gradient [3]. Nonetheless, there is still much to uncover concerning their dynamic behavior, such fine‐tuning of their transport pathways, and the process of ligand recognition. Understanding the classes and transportation mechanisms of these SLC transporters is crucial to understand their pivotal role in cellular functions, as they ensure the regulated movement of essential compounds across cell membranes, impacting metabolism, growth, and overall health [4].

Within the human proteome, the SLC6 transporter family stands out as one of the most significant protein groups, showing associations with conditions such as orthostatic intolerance and attention deficit hyperactivity disorder (ADHD) [5]. In the nervous system, SLC6 transporters plays a critical role in concluding synaptic transmission for numerous amino acids and neurotransmitters derived from amino acids. Moreover, they are also responsible for supplying necessary nutrients and osmolytes to both neurons and glial cells [6]. The SLC6 family encompasses various transporters responsible for the uptake of neutral and cationic amino acids [7]. In the epithelial cells of the intestines and renal tubules, SLC6A19 is crucial for the uptake and reuptake of neutral amino acids such as Glutamine (Gln), Phenylalanine (Phe), and some osmolytes. SLC6A19, otherwise referred as B0AT1 is also present, to a lesser extent, in other tissues including the skin, pancreas, prostate, stomach, and liver, where it governs the transport of neutral amino acids [8].

Considering the significant physiological functions of SLC transporters, it is reasonable that a defect in a single transporter can lead to a severe medical condition. Mutations in the SLC6A19 gene have been linked to a rare genetic condition known as Hartnup disease. This disorder is characterized by the impaired absorption of neutral amino acids in the small intestine and kidneys, resulting in symptoms such as skin rashes, neurological issues, and potentially psychiatric disorders [9, 10, 11].

The SARS‐CoV‐2 receptor, angiotensin‐converting enzyme‐2 (ACE2), is most abundantly expressed in the small intestine, where it is co‐expressed at the apical membrane, serving as the necessary partner for the expression and functioning of the B0AT1 [12]. Recent studies have shown that ACE2‐B0AT1 complex exists in the cell either as a dimer of heterodimers. Structural alignment of RBD‐ACE2‐B0AT1 ternary complex with the S protein of SARS‐CoV‐2 shows that two S protein trimers appear to be concurrently attaching themselves to an ACE2 homodimer. Differences were observed in the examination of the interaction interfaces of SARS‐CoV and SARS‐CoV‐2 RBD with ACE2. Some of these variations may enhance the interactions between SARS‐CoV‐2‐RBD and ACE2, while others are likely to decrease the affinity in comparison to SARS‐CoV‐RBD and ACE2 [12]. These discoveries offer valuable insights into the molecular mechanisms underlying coronavirus recognition and infection. Therefore, it is vital to provide a detailed account of the structural factors that dictate how ligands bind within the SLC family.

Over the past two‐decades, significant efforts have been made to provide new insights into B0AT1 structure and function at high‐resolution. Though the recently resolved cryo‐EM structures of B0AT1 (apo and substrate‐bound conformations) has unearthed a wealth of data on protein function but lacks an in‐depth explanation of how lipid modulation affects the protein behavior, underlining its effect on B0AT1 structure, function/behavior, and dynamics. Since lipids play integral roles in the biology of membrane proteins like B0AT1, studies extrapolating the molecular interactions between lipids, detergents, and SLC transporters is essential to understand the mechanisms used by synaptic membrane lipids to modulate SLC6 family transporters. In this scenario, molecular dynamics (MD) simulations have armored the stochastic nature of these membrane protein systems mostly driven by atomic fluctuations and thermal noise, however, the lack of such comprehensive study put forth a major gap in our understanding of the precise molecular mechanisms employed by B0AT1.

To understand the mechanism behind B0AT1 structural and conformational dynamics in the presence of lipids, herein, we have employed the state‐of‐the‐art all‐atoms MD simulations on B0AT1 in lipid mimetic bilayers to shed light on how lipid modulates B0AT1 structure. First, we used an oversimplified POPC system as a baseline for investigating the B0AT1‐lipid interactions and conformational dynamics. Later, another system, yet realistic lipid system was used for examining the physiologically relevant features of B0AT1 structure and lipid interactions. In conjunction with the structure‐based protein stability analysis of SLC6A19 (with both experimental and MD simulated structures) performed using the pathogenic mutants associated with Hartnup disease, we were able to shed light on the possible effect of pathogenic mutations on B0AT1 structure and function. Overall, our comprehensive study illustrates the importance of classical physics‐based simulations using structure‐based balanced force‐field to explore SLC6 transporter function and the role of the lipid membrane in substrate transport and regulation.

2. Materials and Methods

2.1. Membrane‐Protein System Assembly

The full‐length cryo‐EM structure of ACE2‐B0AT1 in both its closed (PDB ID: 6M18) [13] and open conformations (PDB ID: 6M1D) [13] were retrieved from RCSB Protein Data Bank (PDB). Atomic coordinates of B0AT1 monomer was saved from each complex to study the difference in their structural dynamics and conformations. The protein preparation wizard of Schrodinger Suite Maestro Release 2021 was used to prepare the structures of B0AT1 by optimizing the hydrogen‐bond network through predicting protonation states (using PROPKA3.1) at pH 7.0. All‐atom MD simulations for each B0AT1 conformation was initially performed in a simple lipid bilayer composed of 1‐palmitoy‐2‐oleoyl phosphatidylcholine (POPC) moieties followed by a complex lipid bilayer comprised of POPC:POPE: Cholesterol in a ratio of 2:2:1 for 6M18. The rotational and translational positions of B0AT1 in each lipid‐like environment were constructed using the PPM server with optimization of specified physio‐chemical and mechanical properties. To build a more realistic model, all three B0AT1‐lipid membrane systems were built in CHARMM‐GUI membrane builder program [14] using the structure‐based balanced CHARMM36m force field [15]. Explicit water molecules from TIP3P water model [16] was used to solvate all the systems with the addition of 0.15 M of NaCl counter ions through the auto‐ionizer to prevent the occurrence of long‐range electrostatic effects and electro‐neutralize the systems.

2.2. Lipid‐B0AT1 System Parameterization and Simulation

To elucidate the collective behavior of B0AT1 in a simplified POPC as well as complex realistic environment, GROMACS v2021.4 [17] was used to perform all‐atoms MD simulations, where the SHAKE algorithm [18] was used to propagate both the systems at an interval of 2 femtoseconds timesteps. The steepest descent algorithm [19] was used to perform a standard energy minimization procedure at 50 000 steps followed by a two‐step canonical NVT ensemble and four‐step isothermal‐isobaric NPT equilibration process. Berendsen barostat was used to apply periodic boundary conditions to each system. The particle‐mesh Ewald summation was used to calculate all the long‐range electrostatics while all the hydrogen bonds were constrained using the LINCs algorithm [20]. A production MD run of 300 ns each was run for B0AT1‐POPC‐TIP3 and 1 μs run was performed for B0AT1(6M18)‐realistic lipid systems at 303.15 K using a Nose‐Hoover thermostat [21, 22] and semi‐isotropic pressure coupling with Parrinello‐Rahmann barostat [23] at atmospheric pressure. The protocol for MD simulations has been adapted from our previous studies [24, 25, 26]. All the MD simulated data was processed and visualized using the Visual molecular dynamics (VMD) [27], PyMOL [28], XmGRACE, BIOVIA Discovery Studio Visualizer (BIOVIA DSV) and other GROMACS in‐built tools. Techniques used for analyzing the resultant trajectories are based on a few of our recent studies on membrane proteins [29, 30].

2.3. Lipid‐B0AT1 Simulation Analyses

2.3.1. Validation of Structural Stability, Elasticity, and Dynamics

Analyses of the MD simulated trajectories include calculating the stability and dynamics of both B0AT1 and lipid bilayer in all 3000 frames includes the calculation of root mean square deviation (RMSD) of the backbone atoms, radius of gyration (Rg), root mean square fluctuations (RMSF) of the Cα‐atoms, and solvent accessible surface area (SASA) using gmx rms, gmx gyrate, gmx rmsf, and gmx sasa. The correlated motions of each atom and the dynamic properties of B0AT1 in both open and closed conformations was calculated using gmx covar and gmx anaeig utility toolkits.

2.3.2. Conformational Heterogeneity and Free Energy Landscapes

To map the energy landscapes for both MD simulated B0AT1 systems and observe the transition of energy profile in each system, we analyzed the last 100 ns MD trajectories for both the systems using gmx sham module of GROMACS. The free energy landscapes (FELs) for each system was conducted based on their Rg or PC1 (principal component1) and RMSD or PC2 (principal component2) profiles to imitate the conformational variability in each B0AT1 system. Furthermore, to have a comparative illustration of B0AT1 in both systems we also extracted the representative snapshots possessing the least energy using get_timestamp.py script.

2.3.3. Dynamics of Lipid Bilayer and Their Interaction With B0AT1

In an effort to investigate the dynamic interactions between POPC and B0AT1, the influence of B0AT1 in POPC bilayer, and set POPC system as an ideal system to collect fundamental properties of B0AT1, we computed diverse membrane properties. We initially calculated the deuterium order parameters of POPC acyl chains (viz. sn1 and sn2) using gmx order to measure the orientation mobility of C–H bonds and then membrane density for POPC headgroups, tail groups, water, and phosphate groups using gmx density toolkit of GROMACS. To quantify and extract physical properties of POPC bilayer like area per lipid and membrane thickness with its local topology, we used a stand‐alone python‐based software, that is, FATSLiM [31].

To shed light into the modulatory and structural roles of lipids (POPC, POPE, and cholesterol in our case) on B0AT1 transporter, we directed our interest toward the identification and characterization of specific B0AT1‐lipid interactions and B0AT1 binding sites on POPC\POPE\cholesterol bilayer. Thus to assist our analysis, we used PyLipID [32], a python‐based package to detect binding sites, lipid residence time, and representative lipid bound poses. A dual cut‐off of 0.35–0.55 nm w.r.t protein surfaces was defined to determine the lipid contacts.

With a focus on scrutinizing the critical roles of B0AT1 to balance ions and uptake of amino acids and cations, we also calculated the dynamics of water transport in B0AT1 and calculated its pore radius profile using Channel Annotation Package (CHAP) using B0AT1‐POPC system [33].

2.3.4. Dynamics of Transmembrane Helices

To calculate the effect of POPC bilayer on B0AT1 transmembrane (TM) helices, we used the gmx helixorient for the calculation of the tilt angles and gmx helix to calculate the percentage of helix retention over the period of 300 ns timescale.

2.4. Computing Protein Stability (ΔΔG_MUT) of Pathogenic Mutants Associated With Hartnup Disease

A total of 14 reported pathogenic mutants of B0AT1 associated with the Hartnup disease were fetched from the UniProtKB database (https://www.uniprot.org/uniprotkb/Q695T7/variant-viewer). To understand the deleterious effects of these mutants and disease pathology, we utilized multiple protein stability (ΔΔG) tools and observed the structural and functional impacts of these mutants on the MD simulated B0AT1 open and closed conformations along with the experimental structures in the presence of POPC. We used DDMut [34], mCSM [35], HotMusic [36], PoPMusic [37], FoldX [38], and DynaMut2 [39] protein stability tools, keeping their default parameters intact to calculate the binding free energy of both the conformations of B0AT1.

3. Results and Discussion

3.1. Overall Topology and Architecture of B0AT1

B0AT1 comprises of 634 amino acids with 12 transmembrane (TM) helices [11] and is characterized by the presence of an extended loop connecting TM helices 7 and 8 (Figure 1A,B). This loop is critical for creating contacts with the single transmembrane proteins collectrin or ACE2 to enable their expression on the cellular surface of kidney [40] and intestine [41] respectively. Similar to most neurotransmitters like LeuT, TM1, and TM6 in B0AT1 are also broken into two segments. The primary orthosteric binding site, S1 for neurotransmitter inhibitors is harbored in the TM domain near this broken regions of TM1 and TM6, while some also bind to the allosteric site, which is present 10–12 Å away from S1 on the extracellular vestibule (Figure S1) [42, 43]. Most importantly, site S1 possess more affinity for serotonin uptake inhibitors than site S2.

Figure 1.

Structure of neutral amino acid transporter B0AT1 (SLC6A19) in lipid bilayer. (A) Experimental cryo‐EM structure of monomeric B0AT1 forming open conformation. (B) Structure of B0AT1 involved in forming ACE2 closed conformation. (C) Superposition of both closed and open conformations of B0AT1. (D) Conservation of residues across B0AT1. (Red zone highlights conserved residues, while white zones demonstrate residues possessing the ability to mutate) (E) Structure of B0AT1 showing extracellular and cytosolic regions and position of the transmembrane domain. (F) Orientation of B0AT1 protomer in lipid bilayer. (G) Superposition of MD simulated structure obtained from clustering with the experimental structures of B0AT1.

Both the cryo‐EM structures of B0AT1 in association with ACE2 differ by dimerization of the peptidase domain of ACE2, while closed state of ACE2 was evident only in the presence of SARS‐CoV‐2 RBD [13]. Superposition of B0AT1 monomers from both the B0AT1 structures show that they differ by an RMSD of 0.95 Å (Figure 1C), which suggests that even though little, B0AT1 undergoes conformational changes upon binding to ACE2‐RBD complex. Figure 1D highlights the regions of B0AT1 possessing highly conserved (red in color) and nonconserved (white in color) residues, indicating toward the fact that almost half of the residues in the TM domain are not conserved throughout the evolutionary timescale.

Although considerable research has been devoted to exploit the role of B0AT1 in numerous vital metabolic processes, little is known about its structural dynamics, conformational states and its association with its functional mechanism. Moreover, even with the advancements in cryo‐EM and X‐ray crystallography, specific dynamics and spatiotemporal lipid–protein interactions are difficult to resolve with the help of computational methods [44]. Therefore, we endeavored to extrapolate the above‐mentioned attributes of B0AT1 through classical all‐atoms MD simulations by mimicking a realistic membrane environment (Figure 1E,F). Superposition of both experimental and MD simulated structures of B0AT1 (Figure 1G) displayed a significant amount of deviation mostly in the loop regions in both the experimental and MD simulated structures in the presence of POPC, with an RMSD of 2.24 and 1.86 Å in 6M1D and 6M18 systems, respectively.

3.2. Properties of Diacylglycerol Phospholipid Bilayer

Our MD simulation of B0AT1 in a lipid mimetic system aids in providing a label‐free atomistic insight into the B0AT1‐POPC system. Studies have shown that the available homogenous and symmetric model bilayers are usually supposed to provide an approximate estimation for the properties of cellular membranes in the MD simulated ensembles like acyl chain unsaturation, chemistry of lipid headgroups and presence or absence of sterols. In general, POPC have been chosen as a common constituent for membrane proteins in mammalian cell membranes that is used for computational approaches [29, 30, 44]. Hence, we have also used a simplistic model of POPC bilayer for our study to explore the baseline protein–lipid interactions and their structural and conformational dynamics.

Recently, MD studies focusing on protein–lipid interactions have exhibited a direct impact of lipid bilayer on membrane proteins through any specific molecular contacts and strong binding or non‐direct modulatory effects due to global membrane properties like curvature, thickness, and fluidity [45]. For instance, coupling between the hydrophobic domain of a protein and the hydrocarbon region of a lipid bilayer can alter energetics of different conformations of a protein along with the localization of a protein within the membrane. Therefore, we also devoted our interest on acquiring crucial insights on specific membrane properties like their global structure, elasticity, and dynamic properties. Detailed analyses on specific B0AT1‐POPC interactions have been discussed below.

For MD simulations, deuterium order parameter of any membrane is also often estimated to assess the accuracy of the force field used in the simulation [46]. As can be seen from Figure 2, for the hydrophobic acyl chains of POPC simulations, nearly all atoms produce identical results for analysis of sn‐1 and sn‐2 chains in both systems. Preferably, the presence of B0AT1 might have affected the acyl chains of POPC at position C15. An overall comparison of the ordering results of B0AT1‐POPC systems, leads us to the conclusion that the POPC order parameters from our study agree with the experimentally determined S_cd, apart from some deviation at carbon 15.

Figure 2.

Attributes of diacylglycerol phospholipids (deuterium order, density, area per lipid, and membrane thickness) with incorporated B0AT1 (SLC6A19) over a time scale of 300 ns.

Density profile of POPC bilayer was used to estimate their atomic or molecular distributions along a specific axis and observe any structural changes in the bilayer. It also provides us information regarding the changes in the lateral diffusion coefficient of lipid molecules. The density profile plot in Figure 2 indicates that in both the B0AT1 systems in POPC bilayer had a suitable distribution in both the upper and lower leaflets, suggesting that the structure of POPC membrane remains unaffected even in the presence of B0AT1.

Area per lipid and bilayer thickness are critical constraints for determining the orientation and conformation of a membrane protein. As can be seen from Figure 2, no significant changes in the area per lipid profile can be evident for both systems. However, we can certainly observe deviation in the thickness profile of POPC bilayer throughout the MD run. A nonsignificant yet distinct amount of deviation in the POPC membrane thickness led us to hypothesize that B0AT1 can exert critical effects on the lipid bilayer. However, the effect of B0AT1 in diverse lipid environment, other than POPC is yet to be performed.

3.3. Structural Dynamics and Elasticity of B0AT1 in POPC System

Typically, B0AT1 exhibits a structure comprising 12 membrane‐spanning domains (TM) with N‐ and C‐termini located within the cell. In the case of eukaryotic variants, both terminal ends are notably longer and have been demonstrated to play crucial roles in intricate regulatory functions, including protein transport, ion balance, and functional processes [47]. Lack of this apical neutral amino acid transporter has been consistently found to counteract elevated amino acids in phenylketonuria and urea cycle disorders in mouse models [48, 49]. While, a few other studies have shown that lack of B0AT1, through a series of mechanisms like elevated levels of FGF21, GLP‐1 [50], and reduced liver triglycerides [51] can lead to improved glucose tolerance, thereby protecting the kidney from any injuries. Another recent investigation on small molecule inhibition of B0AT1 has shown a possibility of the development of new therapeutic approaches for diseases like phenylketonuria and toxic accumulation of amino acids resulting in inborn errors of metabolism [52]. Therefore, a thorough structural and functional analysis of these transporters would open doors to the creation of novel compounds aimed at these critical drug targets [53].

Moving in the same direction, we evaluated the backbone RMSD, separately for both, the whole B0AT1 transporter and its transmembrane domain embedded in POPC systems to assess their stability. As can be seen from Figure 3A (left panel) a stable trend for 6M1D (RMSD = 0.31 ± 0.04 nm) was evident throughout the 300 ns timescale while a spike in the RMSD profile for 6M18 (RMSD = 0.29 ± 0.06 nm) was observed during the initial 100 ns, which attains a stable trend afterward. On the contrary, we observe a stable trend in the backbone RMSD profile for the TM domain of both systems, suggesting that the deviation in the RMSD profile of whole B0AT1 might have occurred due to the loops and short helices present either in the cytosol or extracellular region. Moreover, the probability distribution plot in Figure 3A (right panel) demonstrates a notable emergence of two peaks in the case of 6M18, which indicates toward the likelihood of the presence of multiple conformational states, while a single sharp peak can be evident in the case of 6M1D‐POPC system. The results of Rg and SASA calculations are in concordance with that of RMSD analysis (Table S1 and Figure S2).

Figure 3.

Dynamics stability and essential dynamics of B0AT1 transporter attached to both open and closed ACE2 conformations in lipid bilayers. (A) Dynamics stability of backbone RMSD of the whole SLC6A19 transporter along with its TM domain over 300 ns MD. (B) Residual fluctuations in the Cα‐atoms of SLC6A19 transporter measured using the snapshots from last 200 ns. (C) RMSF of the Cα‐atoms of the TM domains of SLC6A19 (TM1‐TM12). (D) Thermal fluctuation or computer B‐factor SLC6A19 transporter in both open and close conformation (shown in putty representation), highlighting the regions of maximum mobility. (E) Superposition of 100 structures from PC1 of both the B0AT1 systems. (F) The porcupine plots of B0AT1 (6M1D and 6M18) obtained using PC1 of both systems.

Conformational plasticity in B0AT1 and regions in TM domain aiding to maximum fluctuation in each system was estimated from the RMSF analysis of the Cα‐atoms (Figure 3B). The average RMSF value for both 6M1D‐POPC and 6M18‐POPC systems was calculated to be 0.12 ± 0.09 nm and 0.11 ± 0.08 nm, respectively. As can be evident from Figure 3B, the secondary structural elements prone to maximum mobility in B0AT1 can be perceived to be both the N and C‐terminal ends, loop regions, and TM12. To have a clear depiction of regions of maximum plasticity in the TM domain, we also plotted the RMSF profile of each TM helix as shown in Figure 3C. Evidently, among all the TM helices, TM7 and TM12 (toward the end of both the helices) seem to be the most flexible regions in both the POPC systems. On the other hand, we also observe a significant level of fluctuation in TM2, TM10, and TM11 of 6M1D‐POPC system in comparison with 6M18 system.

B‐factor (thermal fluctuation factor) putty representation of B0AT1 and its essential dynamics further supports the RMSF analyses (Figure 3D). Accessible motions of B0AT1 sampled conformations from both the B0AT1‐POPC MD trajectories and information related to atomic displacements were discerned through PCA. Such information was extracted by diagonalizing the covariance or correlation matrix and assessing their standard deviations of the atomic displacements of the main‐chain atoms. The trace values for B0AT1 in 6M1D‐POPC and 6M18‐POPC systems can be estimated as 57.01 and 49.79 nm², respectively. Lower trace value in 6M18‐POPC system suggests that B0AT1 occupies less conformational space when associated with closed state of ACE2. Furthermore, porcupine plots of the protein from both the systems locate the region and show the direction of motion in B0AT1 (Figure 3E,F). While similar pattern of movements is evident in both the systems, we also observe large outward movements in the loops connecting TM7‐TM8, TM3‐TM4 and helix in the N‐terminal end in 6M1D, and outward movements in loops connecting TM7‐TM8 was evident in 6M18 PC1 porcupine plot (Figure 3F). On the contrary, similar patterns of inward movements in loops connecting TM7‐TM8 and TM3‐TM4 were evident in PC2 porcupine plots with outward movements of helix in the N‐terminal end in both the systems (Figure S3). These fluctuations in loop connecting TM7‐TM8 can affect the binding of ACE2 and B0AT1. Hence, it can be concluded that the essential dynamics of B0AT1 coincides with our RMSF analyses.

3.4. Electrostatic Surface Potential Maps of B0AT1

Conformational changes in biomolecules like proteins and nucleic acids coupled with other biological macromolecules and small molecules are critical for the functioning of any biological processes including signaling [54], catalysis [55, 56], and assembly of macromolecular complexes [57]. These conformational changes can be easily distinguished because of the kinetic barriers present in each conformational state of the biomolecule. Moreover, the interconversion of conformers in the same states takes place more rapidly than the ones in separate state. The ensemble nature of such states set the basis for their interconversion. Use of high‐resolution structures in combination with MD simulations can be used as a starting point for determining such intricate details of different conformational states of membrane proteins like B0AT1 [58, 59]. In this study, we analyzed the FELs of B0AT1 associated with different conformations of ACE2, that is, both open and closed conformations embedded in POPC lipid bilayer, using gmx sham utility toolkit of GROMACS. As can be evident from Figure 4A, we can clearly observe the presence of multiple conformational states of B0AT1 in both the systems, with more in 6M18‐POPC system, which exhibits multiple peaks with least binding energies. To have a clear understanding of conformational changes that B0AT1 undergoes, we also extracted the snapshots possessing least binding energies.

Figure 4.

Conformational states and electrostatic surface potential map of B0AT1 embedded in POPC bilayer. (A) Free energy landscapes of B0AT1 in both the systems (6MID and 6M18) of B0AT1 with a snapshot from the low energy state. (B) Distribution of electrostatic surface potential charges of B0AT1 in both the MD simulated systems.

Electrostatic surface potential map in Figure 4B highlights the distribution of surface charges across B0AT1. Similar pattern of surface charges can be seen distributed over B0AT1 in both systems, with positive charges overlaying the TM domain and negative charges covering the extracellular and cytosolic regions of B0AT1 (Figure 4B: left panel). However, the channel transporting neutral amino acids in B0AT1 can be perceived to be surfaced mostly with negatively charged residues, with slight positive patches near the S1 site in the case of 6M1D and toward the cytosolic side in 6M18 (Figure 4B: right panel). We anticipate that these changes in the surface charges might have occurred due to the presence of POPC molecules.

3.5. Interaction of POPC Molecules With B0AT1

Yan et al. have shown that B0AT1 regulates the association of ACE2 and viral RBD [12]. However, the mechanism underneath this function of B0AT1 remains largely unknown. Existing preclinical data reveals that B0AT1 has been validated as a promising drug target for the treatment of numerous orphan diseases like alkaptonuria, urea cycle disorders and disorders of branched‐chain amino acid metabolism, which are characterized by the over‐abundance of neutral amino acids [42, 60]. Therefore, a better understanding of the structural characteristics and dynamics of B0AT1 in the presence of lipid can aid in strategizing the development of potential therapeutics against such diseases.

As mentioned above, proteins and lipids are not only essential for preserving cellular membranes but also influences many different cellular functions like allosteric modulation of many membrane proteins such as GPCRs and activation of ion channels [61, 62]. Many of such biological processes takes place on a few nanoseconds and nanometer length scale. To study the dynamics of such processes in membrane, MD simulations acts as a unique window that can provide us information regarding the mechanistic understanding of lipid and protein interactions. In the current research work, we utilized PyLipID to automatically determine specific POPC binding sites, characterize and quantify B0AT1‐POPC interactions. Initially, we determined the duration of POPC contacts which was calculated from the period of time when the molecule moves closer than the lower distance cutoff (0.35 nm) until it moves beyond the upper distance cutoff (0.55 nm). As can be seen from Figure 5A, residues in TM8 can be observed to exhibit the maximum POPC duration of 0.3 µs. In addition to POPC contact duration, we also calculated the number of lipid counts using the lower distance cutoff to assess the number of lipids surrounding B0AT1 (Figure 5B). Nondiffusive behavior of B0AT1, due to their interaction with POPC molecules were further evaluated through observing their dynamic behavior through their residence time. Residence time of any lipid–protein interaction is calculated from a survival time coefficient function and describes the relaxation of bound lipids (Figure 5C). The clustered bound poses of POPC molecules has been illustrated in Figure 5D. Residues exhibiting longest interaction with POPC was estimated to be Pro41 (TM1), Ser190 (loop connecting TM3 and TM4), Arg214 (TM5), Arg240 (loop connecting TM5 and TM6), Ser413 (TM8), Pro414 (TM8), Cys463 (TM9), and Val582 (TM12) (Figure 5E).

Figure 5.

Illustration of POPC mediated specific contacts with B0AT1 inferred through MD trajectories using a dual cut‐off in PyLipID. (A) Duration of POPC contacts with MD simulated B0AT1 systems. (B) Total number of POPC involved in interaction with B0AT1 systems. (C) Residence time of B0AT1 specific residues forming close contacts with POPC. (D) Clustered bound poses of POPC molecules within B0AT1. (E) Location of amino acid residue of B0AT1 exhibiting strong POPC interactions.

3.6. POPC Modulated Alteration of TM Helices

Structural analysis of members of SLC6 family have revealed that TM1‐TM5 attains a pseudo two‐fold symmetry with TM6‐TM10, aligned parallelly across the membrane. TM11 and TM12 are thought to regulate the process of B0AT1 dimerization [63]. The substrate binding site, S1 formed in the center of B0AT1 are mediated by two critical gating residues, that is, the intracellular gating residue Phe48 and extracellular gating residue Phe277. Comparison between the apo state of B0AT1 (6M18) and B0AT1 bound with Gln, Met, and Leu (6M17) reveals the presence of structural variation in the extracellular region of the substrate bound structures, that is, in the loop connecting TM7‐TM8 [12]. Our analysis on the structural dynamics of B0AT1 also suggests that loop connecting TM7‐TM8 is more prone to fluctuations, which is in accordance with previous studies [12].

B0AT1 has also been through clinical validations as a target for treating orphan diseases like alkaptonuria, urea cycle disorders, and ailments associated with branched‐chain amino acid metabolism that are characterized by the overabundance of neutral amino acids. This investigation by Wobst et al. also proofs the bioavailability of B0AT1 inhibitors and their ability to reach plasma concentration that is sufficient enough for inhibition of renal absorption [60]. Another recent study by Xu et al. focusing on exploring the molecular basis of inhibiting B0AT1 brings in light the potential of Trp56 on TM1 to form hydrogen bonds with the carbonyl group of the inhibitors JX225 and JX98. Furthermore, strong hydrophobic interactions between the inhibitors (JX225 and JX98) and residues of TM1, TM3, loop connecting TM5‐TM6 and TM7‐TM8, TM7, and TM8 [64].

To understand the dynamics of each TM helix of B0AT1, we calculated their degree of tilt and percentage of helix retention throughout the MD run. To begin with the observation, we calculated the degree of tilts for each TM helix in B0AT1‐POPC in both systems. As can be seen in Figure 6A and Table S2, maximum degree of tilt was observed in TM1, TM2, and TM6 in both the systems, while TM9, TM10, and TM12 also exhibited a tilt ranging from 13° to 15° in case of 6M1D, which might have resulted due to presence of lipid molecules. Tilt observed in TM1 and TM6 might affect the binding of the substrate at the S1 site [64]. Superposition of the top two cluster representatives of both the systems highlighting conformational differences and the snapshots extracted from the MD trajectories at 50 ns time interval have been demonstrated in Figure S4.

Figure 6.

Tilt and helical properties of each TM helix of B0AT1 MD systems. (A) Tilt of each TM helix of B0AT1 in 6M18 and 6M1D systems. (B) Percentage of helix retention in each TM helix of B0AT1 in each MD simulated system (Black: 6M1D and Red: 6M18) (C) Snapshots displaying the crucial helices of B0AT1 systems inferred at an interval of 50 ns.

Reduction in the percentage of helical content can affect some of the specific functions of any membrane protein. As can be observed from Figure 6B and Table S3, we observe a maximum loss of helicity in TM6 (~40%–50%) in both the systems followed by TM9 and TM1. Illustration of structural snapshots, extracted at every 50 ns time interval highlights the loss of helicity toward the N‐terminal end of TM11 in case of 6M1D at around 150 ns time interval. On the contrary, we observe a loss of helicity in TM1a in the case of 6M18 around 50, 150, 200, and 250 ns time period, which is retained back at the end of 300 ns.

3.7. Comparison of B0AT1 Dynamics in Simple and Complex Lipid Bilayer

In the past three decades, computational modeling and MD simulations have largely contributed to the understanding the behavior of membrane proteins in complex lipid systems with the development of atomistic forcefields in combination with gigantic computational resources [65]. This evolution in the computational biology field has not only boosted the investigation of complex systems with larger size but also made it possible to capture cellular processes ranging from membrane permeation of water, small molecules [66], pore formation [67], and many more. Therefore, to achieve a broader range of biophysical information about the structure‐function relationship in B0AT1 in a realistic environment, we performed all‐atoms MD simulation for 1 µs timescale. The results of our long‐range MD simulation in the presence of zwitterionic lipids, that is, phosphatidylcholine (PC), phosphatidyl ethanolamine (PE), and cholesterol that provide a stabilizing effect to the membrane proteins like B0AT1 have been summarized in Figure 7. As mentioned earlier, achieving insights into the stability of any membrane protein is crucial for understanding the molecular basis of their functions. The backbone RMSD profile of B0AT1 with an average value of 0.28 ± 0.07 nm, as shown in Figure 7A exhibits the pattern of atomic scale deviation throughout the MD run. We observe two different patterns of stabilization in backbone RMSD profile, where the molecule initially exhibits stabilization from ~100 to 300 ns with a sudden spike of atomic deviations and later attains a stable state from ~400 to 1000 ns timescale. The presence of multiple states of B0AT1 stabilization indicates toward the presence of multiple conformational spaces and enhanced stability in its native environment, which is well supported by the gyradius and SASA profile as illustrated in Figure S5 and Table S4, which is contrast to the result obtained from POPC environment. Later, our Cα‐RMSF profile (average RMSF of ~0.14 ± 0.15 nm) and elasticity assessments through PCA aided in identifying regions prone to maximum fluctuations in B0AT1 in its native environment. Notably, our overall plasticity analyses of all the systems in our study highlighted on the consistency in the region of maximum fluctuation irrespective of the lipid environment. However, the presence of complex lipid environment must be leading to rigorous motions in the C‐terminal ends, loop regions and TM12 as compared with the simplified POPC system. The superposition of the top two clusters, cluster 1 and cluster 2 against the experimental B0AT1 (6M18) has been demonstrated in Figure 7D, which clearly exhibits a deviation in the TM7. The difference in atomic deviations of the top two clusters, that is, cluster 1 and cluster 2 as compared with the experimental structure was found to be 1.50 and 1.350 Å, respectively. Furthermore, the RMSF analysis of each TM helix, as demonstrated in Figure 7E also coherently supports the results of POPC system with a slight increase in the values.

Figure 7.

Variation in B0AT1 structural dynamics observed in its native environment. (A) Time evolution of the backbone of B0AT1 observed through 1 µs timescale in its native environment. (B) Cα‐RMSF profile of 6M18 observed in its native environment. (C) PC1 and PC2 porcupine plots, exhibiting direction of motion in B0AT1 in its native environment. (D) unified view of structural variability in the experimental B0AT1 against the top two clusters (6M18) from 1 µs MD run. (E) Dynamic variability of B0AT1 TM helices observed through RMSF analysis over 1 µs MD simulation in a complex lipid environment. (F) Insights into lipid‐B0AT1 interactions observed through PyLiPID using trajectories of 1 µs MD run.

The FEL analysis further corroborates overlapping results obtained from both POPC and the complex lipid system (Figure S6). It can be concluded from the presence of consistent low‐energy states and similar energy barriers between the B0AT1 systems that, irrespective of lipid composition the protein maintains its structural integrity and dynamics.

Most importantly, we sought to understand the variation in lipid interaction profile of B0AT1 in its realistic environment as compared with the simple lipid (POPC) system. The calculated residence time and lipid contact duration have been demonstrated in Figure 7F, where the residence time for each of the lipids were the same, however, variation in the duration time of POPE were noticeable. Furthermore, from extensive analysis of the MD simulated B0AT1 in complex lipid bilayer, we came into conclusion that, residues Ser413, Pro414, Cys463, Trp581, Val585, and Leu468 shows the longest interaction and highest occupancy for POPC, which is similar to what we observe in our earlier analysis in POPC system. On the other hand, we also noted several new residues that exhibited stronger interactions with lipid components such as Arg240, Leu111, Val508, Thr243, Val232, Leu464, Ile239, Ile263, and Val532. Above residues show longer interaction and highest lipid occupancy with POPE, while Thr218, Ile492, Ser594, Thr596, Ile83, Leu199, Tyr600, Gly591, Trp195, Cys203, Met210, Val592, Leu595, Val587, and Val554 displayed consistent interaction with cholesterol.

3.8. Properties of Channel Pathway and Its Pore Facing Residues

B0AT1 has a broad selectivity for neutral amino acids with low apparent affinity for its substrate. The order of substrate affinity for B0AT1 (higher to lower), studied over these years is, methionine > branched‐amino acids > glutamine > asparagine > phenylalanine > cysteine > serine > glycine > tyrosine > threonine > histidine > proline > tryptophan [68]. A structural investigation on B0AT1 substrate recognition and its mechanism of transport reveals that Trp56 could also behave as a pseudo‐substrate symport‐effector. On the other hand, l‐tryptophan has been studied as a poor B0AT1 substrate with 90% lower V_max when compared with the most preferred substrate l‐methionine [69] that binds to site S1 of B0AT1 [70]. Therefore, it is likely that l‐tryptophan might behave as an allosteric inhibitor and displace the sidechain of Trp56, which can slow down the process of transportation [64].

Functional studies have shown that B0AT1 co‐transports 1Na⁺ for each amino acid, which is why a rise in the affinity for substrates can be observed in case of hyperpolarization. There are also existing studies which suggests the phenomenon of water transport in amino acid transporters with a coupling coefficient ranging from 50 to 500 water molecules per substrate molecule [71]. Cotransporters like B0AT1 allows the uphill secondary active transport of hydrophilic molecules to exploit the electrochemical gradient of a cation [72, 73, 74, 75]. It has also been observed that these cotransporters can act as water pathways which might occur either with the nonaqueous substrates or osmotic gradients generated by accumulation of intracellular substrates. However, a thorough analysis on such mechanism is yet to be done.

In our current atomistic MD simulation, we were able to observe a water pathway in B0AT1 through CHAP annotation. We computed the pore facing residues which might be involved in the co‐transportation of Na⁺ ion and amino acids (Figure 8). Residues facing the pore were identified to be Cys49, Leu52, Gly53, and Val55 in TM1, Leu234 (TM2), Phe279, and Leu281 in TM6, Val393, and Gly397 in the loop connecting TM7‐TM8, Leu484, and Gly490 in TM9. To further elucidate the molecular mechanism underlying opening and closing of these gates and exploit the pore hydration and conductive status of B0AT1, we calculated its hydrophobic index and pore radius. We observed the amount of water transport through B0AT1, associated with both open and closed conformations of ACE2. As can be seen clearly, a reduction in the amount of water flow can be observed in 6M18, with decreased pore radius and increased hydrophobic index of the pore. This variation in the properties of the pore might have occurred due to lipid modulated conformational changes in B0AT1 that are required to form the open state of ACE2 in association with SARS‐CoV‐2 RBD.

Figure 8.

Water accessibility within the MD simulated B0AT1, hydrophobic index, and pore radius profile estimated using CHAP annotation tool. Pore facing residues are highlighted in black.

3.9. B0AT1 Pathogenic Mutants Associated With Hartnup's Disease

Hartnup's disease is an autosomal recessive condition that can cause neurological, mental, and dermatological symptoms. Excessive neutral amino acid loss in the intestine and urine is an important feature of this illness, which is caused by inherited defects in an amino acid transporter situated in the apical membrane. Mutations in the cytoplasmic and transmembrane regions of SLC6A19, the recently cloned neutral amino acid transporter, were recently found in Hartnup's disease patients [76]. As mentioned before, SLC619 gene encodes B0AT1 protein that is highly expressed in the kidneys and small intestine. It helps in the transport of neutral amino acids; mutation in the protein is a primary candidate for Hartnup's disease [11]. This clinical presentation is due to deficits in neutral amino acids and their downstream metabolites. Sequencing studies revealed that there are 21 mutations including missense, splicing site, frameshift, and nonsense mutations affecting functionality of the protein and lead to the disease conditions [77].

Here, in this study, we analyzed 14 missense mutations that are reported to have pathogenic effect on protein functionality (Figure 9). These mutations in the protein might have the effect on its stability and integrity and our aim is to explore how these mutations affects the functionality of the protein in the presence and absence of lipid ensembles. Electrostatic surface potential map of mutated B0AT1 exhibits a change in the surface charge distribution of the pore, with a patch of positive and neutral charges developed toward the lower half of the protein (Figure 9B). Lipid interaction with the protein provide a native environment and also affects the helical properties of B0AT1 as shown previously in the study. We, therefore, analyzed the effect of mutations on protein stability, in both experimental (Figure 9C,D) and lipid ensembled‐MD‐simulated structures (Figure 9E,F) with open and closed conformations for better understanding of transportation process. Here, we utilized six varied state‐of‐the‐art structure‐based protein stability tools for the ∆∆G estimation upon missense mutations to minimize the biases and errors (Table S5–S8).

Figure 9.

Distribution and computing the effect of pathogenic mutations associated with Hartnup's disease in B0AT1. (A) The distribution of pathogenic mutants in different TM helices and loops of B0AT1. (B) Electrostatic surface potential map of mutated B0AT1, highlighting the channel of amino acid transport within the TM domain. (C) Assessment of protein stability (ΔΔG) of reported pathogenic variants in open state of B0AT1 protein (6M18). (D) Effect of reported mutants on stability of simulated structure 6M18. (E) Effect of pathogenic mutants on the stability of B0AT1 protein in close conformation (6M1D). (F) Effect of pathogenic mutants on the stability of MD simulated B0AT1 protein in close conformation. Negative values represent protein destabilization, while stability of any mutated protein is represented through positive values.

Mutational analysis revealed that most of the mutations are displaying negative ∆∆G values, indicating destabilization upon substitutions. From the sequence analysis, it was found that D173N is the most frequent mutation associated with B0AT1 that can lead to complete inhibition of transportation in the protein [78]. However, the presented data reveals that there are slight negative ∆∆G values in experimental structure with both open and closed conformations while MD simulated structure with lipid bilayer showed a higher negative value signifying the alteration in protein stability and functionality in native state. SLC6A19 mutants (R57C) in oocytes showed significant reduction in leucine absorption, demonstrating that the mutation resulted in a poorly functioning protein [11]. R57 is located in TM1 and proposed to have a role in closing the pore at extracellular gate by interacting with D486. But substitution with neutral amino acid C57 causes disruption of interaction and leads to B0AT1 malfunction. Similarly, in our analyses, all the states in both experimental and MD simulated structure exhibited decreased stability of the protein. A69 is a part of highly conserved motif NGGGGAF in 2nd TM helix that undergoes conformational changes during the transportation cycle [79]. It is a surface exposed residue that could be involved in the interaction with accessory proteins. Experimental data suggests that A69T mutant, increased cell surface expression of this A69T‐B0AT1 protein upon contact with ACE2. One possible reason for lacking the transport function in ACE2/A69T‐B0AT1 heterodimers expressed on cell surface, suggesting that the associated ACE2 protein suppresses A69T‐B0AT1 activity [79]. Our study suggests that A69T mutation results in the destabilization of both the open and closed states in experimental and MD simulated structures, while destabilization is more pronounced in the lipid assembled structures. Similarly, B0AT1 G93R mutant, present in TM2 exhibits no transport activity when expressed alone or co‐expressed with intestinal protein ACE2 or renal paralogue collectrin [11, 79]. Our study also shows destabilization of the protein in all the tested structures. R240Q is located at the apex, protruding outside for the interaction with accessory proteins. As mentioned above in our POPC‐B0AT1 interaction analysis, R240 also significantly interacts with POPC molecules. R240Q mutation inhibits B0AT1 from interacting with collectrin in the kidney and ACE2 in gut. Thus, resulted in reduced surface expression in both the kidney and intestine, thereby explaining the onset of the disorder in individuals carrying this mutation [41]. Hence, we anticipate that mutation of R240Q might be regulated by lipid molecules. Similar to R57C, L242P mutant sited on 3rd extracellular loop, showed no transport activity or expression in X. laevis oocyte in the presence or absence of other accessory proteins such as ACE2 or collectrin. Although the ∆∆G values are in all the structures indicating destabilization, but the presence of lipids in the MD simulated structure enhances the destabilization as reflected in the ∆∆G values.

In our computational analysis, G284R mutation shows a high rate of destabilization for open and closed states in both experimental and lipid simulated structures. The effect is more pronounced in MD simulated open state structure. The above residue is present in the central region of TM6 in the protein that can be well associated with TM1 to complete the transport cycle [11, 79]. Glycine is a small amino acid residue which provides flexibility but substitution with charged amino acids with a side chain is unable to afford flexibility, hence hinder the transportation function. All the other mutants such as R328C, E405K, E501K, D517G, and P579L are also reported in the B0AT1 protein affects its functionality [77]. The presence of lipid ensembles in the simulated structures augment protein destabilization which is observed in their ∆∆G values. Overall computational estimation suggests that the reported mutations affect the protein stability and thus resulted in impaired protein function and interaction with accessory proteins. In the presence of lipid bilayer and in open state conformation, the effect is more appreciated.

4. Conclusion

Amino acid homeostasis can be considered as a crucial diagnostic factor against orphan diseases [80]. Our comparative structural investigation of B0AT1 in both simplified POPC and complex lipid systems provide the evidences on how the presence of lipids affects the structure and dynamics of the protein. While the complex lipid system undoubtedly provided us some additional and specific informations like the enhanced stability of B0AT1, and increased flexibility of TM7, TM12, and loop regions, specifically loop connecting TM7‐TM8. In the presence of a complex system, the overarching conclusions of our study remain consistent with the previous findings observed in POPC system.

Our extensive structural investigation on B0AT1 dynamics in both the POPC systems, however, provides a baseline for studying the fundamental behaviors of B0AT1 like deviation in the helical properties of TM1 and TM6 without the addition of complex lipids which might affect the dimensions of the substrate binding sites of B0AT1, that is, S1 and S2 sites. Moreover, the presence of multiple conformational states and a shift in the pore radius, hydrophobic index, and surface charge distribution of B0AT1 water channel were evident in the MD simulated system of B0AT1 associated with the open conformation of ACE2, suggesting the possibility that specific B0AT1‐lipid interactions are required for formation of ACE2 open state conformation. However, we also evident a clear deviation in the thickness and order parameter profile of POPC bilayer in B0AT1 structure associated with the open state of ACE2, which suggests that B0AT1 might also affect the structure of lipid molecules.

Finally, the mutational analysis of both experimental and lipid ensembled‐MD simulated structures of B0AT1 indicate destabilization of the protein upon substitutions to nonnative amino acids, especially in the MD simulated structures, in the presence of POPC molecules, signifying the role of lipids in the proper functionality of the protein which is well pronounced in the case of R240Q and G287R. However, the dynamic behavior of B0AT1 in lipid environment, other than POPC and the effect of mutants, is yet to be explored. Moreover, the limitations of our study also lie in the use of classical MD simulations in lipid bilayers, which we are attempting to recover through the use of advanced MD simulations and extending our investigation of B0AT1 in its native environment to achieve a broader range of biophysical information regarding its structure‐function mechanism and the effect of mutations in our forthcoming research work.

Author Contributions

Budheswar Dehury: conceptualization, supervision, data curation, formal analysis, methodology, software, visualization, writing–original draft. Sarbani Mishra: formal analysis, methodology, software, visualization, writing–original draft. Sunita Panda: formal analysis, methodology, software, writing–original draft. Sanghamitra Pati: project administration, writing–review and editing.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting information.

Supporting information.

Data Availability Statement

All research data presented in this study are included in the manuscript and supporting material.