Membrane Permeation of Psychedelic Tryptamines by Dynamic Simulations

Palmisano and colleagues situate this work within the recent revival of interest in classic psychedelics, which are structurally related to serotonin (5-HT) and act as agonists at the 5-HT2A receptor. Previous pharmacological and animal research links 5-HT2A agonism to rapid neurophysiological and neuroplastic changes that may underlie therapeutic effects in conditions such as depression and anxiety. However, there is mechanistic uncertainty regarding why closely related 5-HT2A agonists differ in their neuroplastic and subjective effects. Recent evidence suggests that activation of intracellular pools of 5-HT2A receptors — requiring a compound to cross the neuronal plasma membrane — may be a key determinant of these differences, and that lipophilicity correlates with neuroplasticity-promoting properties in earlier studies. This study aims to characterise, by molecular simulation, how structural modifications to tryptamine derivatives influence their ability to permeate a model lipid bilayer and thereby access intracellular receptor pools. The investigators selected 12 tryptamines (tryptamine, serotonin, 5-methoxy-tryptamine, 4-hydroxy-tryptamine and their N-methyl and N,N-dimethyl analogues) and used classical molecular dynamics (MD) with umbrella sampling (US) to compute potentials of mean force (PMFs), local diffusivity, and effective permeability coefficients (log Peff). The intent was to isolate the effects of N‑alkylation, indole-ring substitution and substitution position, and protonation state on membrane permeation, with an eye toward informing rational design of psychedelics with altered intracellular access.

The study used all-atom classical MD and umbrella sampling to model permeation of 12 tryptamine derivatives across a homogeneous POPC bilayer. Ligand geometries were optimised at the B3LYP/cc-pVDZ level and ligand partial charges were derived via the Merz–Singh–Kollman scheme at HF/6-31G*; ligand parameters otherwise came from GAFF. The bilayer was built with CHARMM-GUI and comprised 100 POPC lipids, solvated with about 5000 TIP3P water molecules and NaCl at 0.15 mol L-1, in a rectangular box of dimensions x = 58 Å, y = 58 Å, z = 86 Å. Lipid and water parameters were taken from Lipid17 and TIP3P; ion Lennard–Jones parameters followed Joung and Cheatham. Simulations were performed with the CUDA version of AMBER20. After energy minimisation and staged equilibration (including an NVT heating to 303.15 K and NPT equilibration with semi‑isotropic pressure scaling), a 200 ns production run was used to equilibrate the membrane before ligand insertion. Each ligand was inserted at the membrane centre (z = 0 Å), the system was re-equilibrated, and a restrained production run of 100 ns (harmonic restraint on COM distance) preceded a pulling simulation that moved the ligand from z = 0 Å to z = 36 Å at 1 Å ns-1 to generate initial windows for US. SHAKE constrained bonds to hydrogen, non-bonded cut-off was 10 Å and electrostatics used particle-mesh Ewald with 1 Å grid spacing. Umbrella sampling covered the z‑coordinate from the bilayer centre (z = 0 Å) to bulk water (z = 36 Å) using 45 windows at 0.8 Å spacing. Each window produced a 40 ns trajectory and the full 40 ns per window was analysed. PMFs were obtained by the weighted histogram analysis method. Local diffusivity D(z) and PMFs W(z) were combined to estimate effective resistivity and effective permeability (Peff); log Peff values were reported. To decompose interaction contributions at minima and maxima along PMFs, MM‑GBSA binding free energies were calculated using 500 frames per calculation and split into van der Waals, electrostatic, polar solvation and nonpolar terms. Simulations included both neutral species for all compounds and explicit simulations of protonated forms for TRY and N,N-TRY to assess the influence of ionisation.

Computed permeation profiles and derived log Peff values showed clear structure–permeation relationships across the set of tryptamines. Overall, all neutral compounds migrated from bulk water (z = 36 Å) to the lipid headgroup region (≈ z = 25 Å) without an entry barrier, and each compound displayed a free‑energy minimum located roughly 10–14 Å from the bilayer centre depending on substitution. Dimethylated (N,N) compounds uniformly exhibited positive log Peff (indicating permeability). Most N‑methylated variants were also permeable except for N‑SER. Within the unmethylated group, 5‑methoxy‑tryptamine (OME) was the most permeable, followed by 4‑hydroxy (PSI), tryptamine (TRY) and serotonin (SER). N‑alkylation had a pronounced effect: methylation decreased polarity, shifted the free‑energy minimum progressively closer to the lipid tails (TRY minimum at z = 13.6 Å, N‑TRY at z = 12.0 Å, N,N‑TRY at z = 10.4 Å), increased van der Waals interactions with tails (ΔGvdW increased from primary to tertiary amine) and reduced desolvation penalties for substituted ligands. Those changes produced deeper minima for N,N species and lowered the barrier at the bilayer centre, explaining their higher permeability. Indole substitutions at position 5 also affected minima depth but not their position for N,N compounds. Among the tertiary amines, interaction strength ranked N,N‑OME > N,N‑TRY > N,N‑SER, with the methoxy substituent producing the deepest minima because strong van der Waals contacts outweighed solvation penalties. For unmethylated and N‑methyl species, indole substitution shifted both the position and depth of minima: SER derivatives tended to sit nearer the headgroups (z ≈ 15.2 Å) due to –OH interactions with polar heads, while OME derivatives located closer to tails (z ≈ 11.2 Å) because of hydrophobic contacts. Comparing positional isomers PSI (4‑OH) and SER (5‑OH) across alkylation states revealed nuanced effects. PSI compounds were generally less polar, exhibited stronger van der Waals interactions and a smaller desolvation penalty than SER counterparts, producing lower energy barriers and slightly deeper minima for PSI and N‑PSI relative to SER and N‑SER. However, in the double‑methylated pair the trend reversed: N,N‑SER showed a deeper minimum than N,N‑PSI, attributed to shifts in the balance of electrostatics, van der Waals and solvation contributions. Protonation strongly impeded permeation. Explicit PMFs for protonated TRY+ and N,N‑TRY+ showed deep minima near the polar headgroups dominated by electrostatic interactions, with phosphate groups contributing about 66% of ligand–headgroup electrostatics (TRY+: choline 20%, phosphate 66%, glycerol 14%; N,N‑TRY+: choline 24%, phosphate 66%, glycerol 10%). Both protonated species faced steep barriers to the bilayer centre of ≈14 kcal mol-1, yielding very low effective permeabilities and supporting the conclusion that neutral forms must cross the bilayer and then be reprotonated in the cytosol. The simulation-derived log Peff values correlated well with available experimental dendritogenesis efficacy data from a previous study and aligned with earlier milogP predictions, giving cross‑validation of the computational trends reported.

The investigators interpret their findings as supporting the hypothesis that membrane permeation of tryptamine derivatives is strongly modulated by N‑alkylation, indole substitution and protonation state, and that these physicochemical properties can help explain differential access to intracellular 5‑HT2A receptor pools proposed to underlie distinct neuroplastic and behavioural effects. They emphasise that neutral species are required for transbilayer crossing, which is consistent with the large energetic penalty experienced by protonated forms and with previous experimental observations that intracellular receptor activation is important for neuroplasticity. In positioning the results relative to earlier work, the authors note agreement with a prior study that reported a link between lipophilicity and neuroplastic effects, and with Vargas and colleagues' experimental observations on intracellular receptor activation. They also report concordance between their computed log Peff values and earlier milogP predictions, and show that specific modifications (dimethylation, 5‑methoxy substitution, and certain positional isomerisms) have predictable effects on permeability because they tune the balance between van der Waals interactions with lipid tails and desolvation/electrostatic penalties. Key limitations acknowledged include the use of a homogeneous POPC membrane model rather than a complex neuronal membrane containing diverse lipids and cholesterol, and the absence of an explicit concentration gradient or ion‑trapping description that could affect transport dynamics. The authors also note the relative scarcity of experimental permeability and activity data for some compounds (notably PSI derivatives), which limits exhaustive validation. They present their computational protocol as a state‑of‑the‑art approach that nonetheless would benefit from additional experimental comparisons and from extensions that include membrane compositional complexity and ion‑trapping effects.

Palmisano and colleagues conclude that MD combined with umbrella sampling can robustly characterise how small chemical modifications alter membrane permeation of tryptamine psychedelics. Their simulations show that all neutral tryptamines studied partition to energy minima located roughly 10–14 Å from the bilayer centre, that N,N‑dimethylation and 5‑methoxy substitution most strongly increase permeability by deepening minima and reducing central barriers, and that protonated (charged) species face steep central barriers (≈14 kcal mol-1) and are effectively impermeable. The authors stress that these mechanistic insights into how alkylation, indole substitution and positional isomerism modulate van der Waals, electrostatic and solvation contributions may inform the rational design of psychedelic compounds with altered intracellular access, while recognising limitations related to membrane model simplicity and limited experimental benchmarks.