Identification of 14-3‑3 Proteins as Binding Partners of TRP Channels

Abstract

Transient receptor potential (TRP) channels are regulated by a diverse network of intracellular partners that govern their trafficking, stability, and functional expression at the plasma membrane. Here, we present a comprehensive and integrative characterization of 14-3-3 proteins as conserved binding partners of TRP channels. Leveraging the extensive structural repertoire of 14-3-3 complexes resolved to date, we combined large-scale sequence and structural analyses with molecular docking, coevolutionary inference, machine learning-based predictions, atomistic simulations, and targeted experimental validation to elucidate the molecular principles underlying TRP-14-3-3 recognition. Integration of these approaches into a unified consensus scoring framework revealed recurrent, solvent-exposed cytoplasmic motifs across the TRP channel family with a high propensity for 14-3-3 binding. Focusing on the TRPM4-14-3-3γ interaction, we identified an N-terminal cytoplasmic region of the channel as the primary 14-3-3 binding hotspot. Structural modeling and molecular dynamics simulations revealed a stable electrostatically driven interface, which was experimentally validated by fluorescence anisotropy assays. Moreover, biochemical and functional analyses demonstrated that TRPM4 interacts not only with 14-3-3γ but also with 14-3-3η, leading to a reduced channel-mediated sodium influx. Together, these findings establish 14-3-3 proteins as general and evolutionarily conserved regulators of TRP channels and provide a broadly applicable framework for identifying transient protein–protein interactions relevant to TRP channel dysregulation in disease.

Introduction

Protein–protein interactions (PPIs) have been identified as critical modulators of ion channel physiology and subcellular trafficking, influencing their localization, stability, and functional gating. , Given that proteins rarely act as isolated entities to carry out their biological functions, a crucial step toward elucidating the functional relationships within protein–protein complexes is the characterization of their physical interaction interfaces. Depending on their persistence, these interactions can be classified as transient or permanent, and are typically governed by specific sequence motifs or structural features that facilitate binding.

Through MS-based proteomic approaches complemented by computational analyses, , our group has identified a substantial set of interactor proteins that regulate the biophysical properties, trafficking, and membrane stability of transient receptor potential (TRP) channels. − Among these, the 14-3-3 protein familyin its seven isoforms (β, ε, η, γ, σ, τ, and ζ)is of particular interest for this study. 14-3-3 proteins constitute an evolutionarily conserved family of small regulatory molecules involved in a wide range of intracellular signaling pathways through the formation of protein–protein complexes. Structural analyses show that 14-3-3 proteins form symmetric cup-shaped dimers composed of two protomers rich in α-helices. Each protomer contains nine antiparallel α-helices and a single phosphopeptide-binding site. Genes encoding 14-3-3 proteins participate in many essential cellular processes, including cell-cycle control, apoptosis, signal transduction, metabolism, and the intracellular trafficking of membrane proteins.

The association of 14-3-3 proteins with ion channels was demonstrated in several systems. Using yeast two-hybrid (Y2H) screening, 14-3-3ε was shown to interact with HERG K⁺ channels, stabilizing the PKA-phosphorylated state of the channel and promoting its opening at negative potentials. Similarly, 14-3-3η was identified as a binding partner of the voltage-gated sodium channel Nav1.5, with functional consequences for the cardiac action potential. Interactions between 14 and 3-3 (ζ/ε and γ) and two-pore-domain potassium (K2P) channels have also been reported, where 14-3-3 proteins regulate surface expression and promote efficient trafficking of these channels to the plasma membrane. , Based on these findings, we hypothesized that 14-3-3 proteins may physically associate with members of the TRP channel family and modulate their functional properties and trafficking mechanisms, mirroring their regulatory roles in other ion channels.

The TRP family of ion channels comprises receptor-activated, nonselective cation channels that regulate multiple Ca²⁺ influx pathways and contribute to the permeation of other physiologically relevant ions. This superfamily includes 28 mammalian channels organized into seven subfamilies: TRPC (canonical), TRPV (vanilloid), TRPM (melastatin), TRPA (ankyrin), TRPP (polycystin), TRPML (mucolipin), and TRPN (NOMPC). Despite their functional diversity, all TRP channels share a conserved architecture consisting of four symmetrically arranged protomers, each containing six transmembrane helices that together form a central cation-selective pore. Their intracellular N- and C-terminal domains include subfamily-specific structural and regulatory elements that shape channel function and provide key sites for interactions with external proteins. TRP channels are widely expressed across diverse cell types and are broadly distributed in both excitable and nonexcitable human tissues. Dysregulation of their activity has been implicated in numerous pathological conditions, making TRP channels attractive therapeutic targets for neurological and cardiovascular disorders, pain, respiratory diseases, and cancer. ,

Molecular-level interactions between TRP channels and 14-3-3 proteins have been previously reported. Mutational analyses of TRPM7 identified key serine residues that regulate the channel stability and intracellular trafficking. Phosphorylation of these residues enables TRPM7 to interact with the signaling protein 14-3-3η, underscoring a regulatory role for 14-3-3 in channel trafficking. Likewise, the 14-3-3ζ isoform directly interacts with TRPM8 and is essential for LCK-dependent regulation of channel assembly, with LCK and 14-3-3ζ acting together to modulate TRPM8 multimerization and channel function. In addition, cell-surface expression of the TRPM4 channel has been shown to be regulated by 14-3-3 proteins, with the S88 residue in the intracellular region of the channel identified as a key determinant for its physical association with 14-3-3γ.

Building on this foundation, this study leveraged the extensive structural diversity of 14-3-3 proteins resolved to date, including complexes with a variety of ions, ligands, peptides, and protein partners. Using these structural resources, we performed integrated sequence and structure-based analyses that combined coevolutionary signals with machine learning-based predictions to derive a consensus framework for identifying molecular determinants of 14-3-3 binding within TRP channels. We then characterized the previously reported TRPM4-14-3-3γ interaction at atomic resolution using protein–protein docking and molecular dynamics simulations, followed by experimental validation of complex formation. This integrative approach led to the identification of a previously unrecognized N-terminal binding region in TRPM4 and revealed an additional 14-3-3 isoform associated with the channel. Collectively, our findings provide a comprehensive definition of 14-3-3 recognition motifs and interaction patterns across the TRP channel family, establishing 14-3-3 proteins as conserved and functionally relevant interaction partners for these channels.

Methods

Analysis of Available 14-3-3 Protein Structures

To obtain all Protein Data Bank (PDB) structures reporting complexes involving 14-3-3 proteins, we used the corresponding UniProt sequences and implemented a Python script based on the RCSB RESTful API v1.35.0. The rcsb_gene_name.value module was used to obtain the list of PDB entries associated with the seven 14-3-3 isoforms. The resulting structures were grouped according to the type of partner molecule bound to 14-3-3, namely, ion, ligand, peptide, or protein. Additionally, relevant metadata, including protein name, sequence length, species, and other structural annotations, was extracted to enable systematic characterization of the reported complexes. The structures were subsequently analyzed in PyMOL v3 (Schrödinger, LLC) to identify residues forming the interaction interface of each 14-3-3 protein using distance cutoffs of 5–10 Å. The corresponding residue sequences were exported in FASTA format and used to generate multiple sequence alignments (MSA) with MAFFT, aiming to detect conserved motifs or binding patterns. Sequence alignments were examined in Jalview, and residue conservation was quantified using WebLogo. To complement these analyses and detect structural features of both 14-3-3 proteins and their partners, we additionally employed the 3D-PP server, using a spacing threshold of 0.8 Å, a radius of 3 Å, an RMSD cutoff of 2 Å, and a minimum coverage of 80%. These analyses allowed us to define sequence motifs and structural characteristics that determine partner–protein binding to 14-3-3 proteins.

Identification of 14-3-3 Protein Binding Sites in TRP Channels

To identify potential 14-3-3 binding sites within TRP channels, we first generated sequence patterns and profiles using the PRATT and HMMsearch servers, employing the previously obtained multiple sequence alignments and structural profiles as input. The resulting motifs were then searched across the TRP channel family, focusing on those members with available sequence information in the UniProt database and corresponding three-dimensional structural data. In parallel, blind docking simulations between all 14-3-3 isoforms and representative structures of each TRP subfamily were performed using the LightDock software, restricting the search space to the intracellular regions of the channels. Docking calculations were carried out using 400 swarms, 200 glowworms, 100 steps, and activation of the membrane module, as described in Roel-Touris et al. Representative TRP channels included TRPA1 (PDB ID: 6PQP), TRPC3 (PDB ID: 7DXB), TRPC5 (PDB ID: 7D4P), TRPC6 (PDB ID: 7DXG), TRPM2 (PDB ID: 6MIX), TRPM4 (PDB ID: 5WP6), TRPM8 (PDB ID: 8BDC), TRPV3 (PDB ID: 6MHV), TRPV4 (PDB ID: 7AA5), and TRPV6 (PDB ID: 6E2F). The 14-3-3 isoforms analyzed corresponded to 14-3-3β (UniProt: P31946; PDB ID: 2C23), 14-3-3ε (UniProt: P62258; PDB ID: 2BR9), 14-3-3η (UniProt: Q04917; PDB ID: 2C74), 14-3-3γ (UniProt: P61981; PDB ID: 2B05), 14-3-3σ (UniProt: P31947; PDB ID: 6QHL), 14-3-3τ (UniProt: P27348; PDB ID: 2BTP), and 14-3-3ζ (UniProt: P63104; PDB ID: 7D8H). The docked structures were subsequently evaluated using the Rosetta InterfaceAnalyzer module, and contact-frequency analyses between TRP channels and 14-3-3 proteins were performed using a custom TCL script to identify key binding patterns.

Coevolutionary and Machine Learning-Based Contact Predictions

Protein–protein coevolution analyses were performed using paired multiple sequence alignments constructed for both the TRP channel family and the 14-3-3 protein family, ensuring that each sequence pair corresponded to the same species. Sequences for each protein family were obtained from the InterPro database, including TRPA (ID: IPR052076), TRPC (ID: IPR002153), TRPM (ID: IPR050927), TRPV (ID: IPR024862), and 14-3-3 (ID: IPR000308). The reference human sequences for each TRP representative and the corresponding 14-3-3 isoforms were obtained from UniProt and incorporated into their respective family data sets. Separate MSAs were then constructed using HMMER. The final MSAs were used as input for the coevolutionary analyses, performed with a unified coevolutionary framework implemented as a Google Colab notebook originally developed by Ovchinnikov et al. The resulting coevolution matrix was filtered to retain only interprotein residue contact pairs, excluding those located within the transmembrane regions of the TRP channels. To further strengthen this analysis, we predicted potential 14-3-3 binding sites in the representative TRP channels under study using the machine learning-based web server 14-3-3Pred. This tool integrates an artificial neural network, a position-specific scoring matrix (PSSM), and a support vector machine (SVM) to evaluate and score phosphopeptides with the potential to bind 14-3-3 proteins.

Calculation of a Consensus Score to Determine 14-3-3 Binding Sites

To define the residues within TRP ion channels most strongly predicted to interact with 14-3-3 proteins, we implemented a consensus strategy that integrated all previously described approaches: 14-3-3 binding pattern analysis, contacts obtained from blind docking, coevolutionary predictions, and machine learning-based binding site predictions (Figure A). The outputs from each method were weighted within a unified consensus scoring function to prioritize the regions of interest. This scoring function reflects, for each residue across the TRP channels analyzed, its overall propensity to participate in a 14-3-3 binding interface. To combine the heterogeneous outputs from each method, the values were partitioned into three groups, corresponding to high, intermediate, and low confidence levels. Values in these groups were assigned scores of 3, 2, and 1, respectively. The boundaries separating these groups were computed using the following thresholds:

$$\mathrm{lower}\mathrm{threshold}=\min (x)+\frac{\max (x)-\min (x)}{3}$$

$$\mathrm{u}\mathrm{p}\mathrm{p}\mathrm{e}\mathrm{r}\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{h}\mathrm{o}\mathrm{l}\mathrm{d}=\min (x)+\frac{2\ast (\max (x)-\min (x))}{3}$$

where x represents the output from any of the four predictive approaches and min (x) and max (x) correspond to the minimum and maximum values in that data set. Each value was then categorized according to

$$\mathrm{s}\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{e}(x)=\{\begin{array}{lll}3& & x>\mathrm{u}\mathrm{p}\mathrm{p}\mathrm{e}\mathrm{r}\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{h}\mathrm{o}\mathrm{l}\mathrm{d}\\ 2& \mathrm{l}\mathrm{o}\mathrm{w}\mathrm{e}\mathrm{r}\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{h}\mathrm{o}\mathrm{l}\mathrm{d}& <x\le \mathrm{u}\mathrm{p}\mathrm{p}\mathrm{e}\mathrm{r}\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{h}\mathrm{o}\mathrm{l}\mathrm{d}\\ 1& & x\le \mathrm{l}\mathrm{o}\mathrm{w}\mathrm{e}\mathrm{r}\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{h}\mathrm{o}\mathrm{l}\mathrm{d}\end{array}$$

A combined consensus score was then calculated for each residue i by summing the categorical scores received from all four predictive approaches:

$$\mathrm{c}\mathrm{o}\mathrm{m}\mathrm{b}\mathrm{i}\mathrm{n}\mathrm{e}\mathrm{d}\mathrm{s}\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{e}(i)={\mathrm{s}\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{e}}_{\mathrm{b}\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{i}\mathrm{n}\mathrm{g}\mathrm{p}\mathrm{a}\mathrm{t}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{n}\mathrm{s}}(i)+{\mathrm{s}\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{e}}_{\mathrm{d}\mathrm{o}\mathrm{c}\mathrm{k}\mathrm{i}\mathrm{n}\mathrm{g}\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{a}\mathrm{c}\mathrm{t}\mathrm{s}}(i)+{\mathrm{s}\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{e}}_{\mathrm{c}\mathrm{o}\mathrm{e}\mathrm{v}\mathrm{o}\mathrm{l}\mathrm{u}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}(i)+{\mathrm{s}\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{e}}_{14-3-3\mathrm{P}\mathrm{r}\mathrm{e}\mathrm{d}}(i)$$

To smooth the data and better identify local interaction hotspots, we applied a sliding-window procedure. In this approach, the combined scores of residues within a defined window were summed and assigned to the central residue, enabling detection of contiguous regions with elevated predicted 14-3-3 binding propensity:

$$\mathrm{sliding}\mathrm{sum}(i)=\left(\sum _{j=i}^{i+(w-1)}{\mathrm{combined}\mathrm{score}}_{(j)}\right)-\mathrm{baseline}$$

$$\mathrm{central}\mathrm{residue}(i)={\mathrm{residue}}_i+\frac{w}{2}$$

where i is the residue index, w is the window size, and baseline represents the minimum possible score for a window, which is subtracted from the sliding sum to standardize the resulting values. For this study, a window size of 27 residues was used, consistent with fragment lengths typically employed in Rosetta-based protein–peptide interaction predictions.

1.

Integrated consensus analysis of 14-3-3 binding sites in TRP channels. (A) Schematic overview of the consensus scoring strategy used to integrate multiple independent approaches, including 14-3-3 binding motif analysis, blind docking contacts, coevolutionary predictions, and machine learning-based binding site predictions. Individual scores from each method were classified into high, intermediate, and low confidence levels and combined into a unified consensus score, followed by smoothing using a moving-sum approach to identify continuous interaction regions. (B) Resulting consensus scores were mapped onto representative surface-rendered structures of TRPA1, TRPC3, TRPM4, and TRPV3. Shaded boxes denote the transmembrane region. Residues are colored according to their consensus score (color scale), with higher values indicating stronger support for 14-3-3 interaction.

Structural Modeling of the TRPM4-14-3-3γ Complex

To investigate the structural basis of TRP channel recognition by 14-3-3 proteins, we modeled the TRPM4-14-3-3γ complex, a system previously characterized experimentally. Protein–protein docking was performed using HADDOCK, employing the available structures of TRPM4 (PDB ID: 5WP6) and the 14-3-3γ homodimer (PDB ID: 2B05) as input. Missing TRPM4 segments with potential roles in partner recognition were reconstructed de novo using the Rosetta KIC protocol. Residue S88, proposed to be critical for 14-3-3 binding, was specifically modeled with Rosetta FloppyTail in its phosphorylated state to assess its solvent accessibility and interaction potential. For the docking calculations, TRPM4 residues identified through the consensus scoring analysis as putative 14-3-3 contact sites were designated as active residues, while for 14-3-3γ, the canonical binding groove was defined as the active region. A total of 1000 docked conformations were generated, and their interfacial energies were evaluated using the Rosetta InterfaceAnalyzer module. Structural clustering was subsequently performed with BitQT using a root-mean-square deviation (RMSD) cutoff of 2 Å. The most favorable cluster was selected based on the lowest interface binding energy and the shortest center-of-mass distance to the channel region predicted to interact with 14-3-3 proteins. The representative conformation from this cluster was analyzed using the Ensemble Cluster tool in UCSF Chimera.

Molecular Dynamics Simulations of the TRPM4-14-3-3γ Complex

The optimal docked conformation of the TRPM4-14-3-3γ complex was embedded into a pre-equilibrated palmitoyl-oleyl-phosphatidylcholine (POPC) bilayer, solvated with explicit TIP3P water molecules, neutralized, and ionized to a final NaCl concentration of 0.15 M. The system was first minimized for 100,000 steps, followed by a 27 ns equilibration phase at 310 K under isothermal–isobaric (NPT) conditions. During equilibration, soft harmonic restraints (progressively reduced from 20 to 0 kcal mol^–1 Å^–2) were applied to the proteins, and the Colvars module was used to enforce preservation of the protein–protein interface. Three independent unrestrained production replicas were then performed, yielding a total of 0.75 μs of simulation time. Temperature was maintained using a Langevin thermostat (damping coefficient: 1 ps^–1), and the pressure was controlled at 101.325 kPa with the Langevin piston method. Electrostatic interactions were treated using particle-mesh Ewald (PME) with an 8–9 Å real-space cutoff. Hydrogen mass repartitioning was applied to enable the use of a 4 fs time step, employing the Verlet-I/r-RESPA multiple time-step integrator. All simulations were performed with NAMD v3 using the CHARMM force field. Structural stability was assessed by computing the root-mean-square deviation (RMSD) across the simulation trajectories. Protein–protein contact frequencies were calculated as the fraction of trajectory frames in which any interprotein atomic pair was separated by ≤ 5 Å. All structural analyses were conducted using VMD 1.9.3.

Expression and Purification of TRPM4-N and 14-3-3 Proteins

The genetic material encoding human 14-3-3γ/η and the MHR1/2 domain of TRPM4 (TRPM4-N region) was codon-optimized for expression in Escherichia coli, synthesized by GenScript (Central, Hong Kong), and cloned onto a pET-28a­(+)-TEV vector. The TRPM4-N gene was synthesized with two exposed cysteine residues replaced by serine (C242S and C385S), and subsequent site-directed mutagenesis was performed to generate two solvent-exposed Ser-to-Cys variants for fluorescent labeling, namely, S27C and S148C. The S27C mutation is located near the N-terminus within the MHR1 domain, whereas S148C lies in the spatial vicinity of the previously reported S88 residue, whose phosphorylation has been described to enable 14-3-3 binding. The resulting plasmids were chemically transformed onto E. coli BL21 (DE3) cells, which were grown in LB medium at 37 °C until an OD₆₀₀ of 0.6–0.7 was reached. Protein expression was induced by adding isopropyl β-d-1-thiogalactopyranoside (IPTG), and cultures were incubated overnight at 20 °C. Cells were harvested by centrifugation and lysed by sonication in a buffer containing 50 mM Na₂HPO₄ (pH 8.0), 200 mM NaCl, and 20 mM imidazole. The lysate was clarified by high-speed centrifugation and subsequently used for immobilized metal affinity chromatography (IMAC). His-tagged proteins were purified using a Ni-NTA column (HisTrap, Cytiva) equilibrated in the same buffer and eluted with a buffer containing 500 mM imidazole. Protein purity was assessed by SDS-PAGE.

Fluorescent Labeling

The purified TRPM4-N S27C and S148C variants were further purified by size-exclusion chromatography (SEC) using an FPLC system equipped with a Superdex 200 Increase 10/300 GL column (GE Healthcare Bio-Sciences, USA), equilibrated with a buffer containing 50 mM Na₂HPO₄ (pH 8.0) and 150 mM NaCl. Each protein was subsequently concentrated to >60 μM, supplemented with 400 μM tris­(2-carboxyethyl)­phosphine (TCEP; Thermo Fisher Scientific), and incubated at room temperature for 30 min. Fluorescent labeling was performed by incubating the reduced protein with a 5-fold molar excess of BODIPY FL dye (Thermo Fisher Scientific) overnight at 4 °C. The reaction was quenched by diluting the sample in the same buffer supplemented with 5 mM 2-mercaptoethanol (2-ME), and excess dye was removed by a second SEC run under identical conditions, with the buffer additionally containing 2 mM 2-ME.

Fluorescence Measurements

Fluorescence anisotropy binding assays were performed at 37 °C using a Spark Multimode Microplate Reader (TECAN, Switzerland) operated in the fluorescence polarization mode. Parallel and perpendicular emission intensities were collected and converted to anisotropy values. Binding curves were measured in duplicate using a constant concentration (10 nM) of fluorescently labeled TRPM4-N mutants and increasing concentrations of unlabeled 14-3-3γ/η, ranging from 0 to 15 μM. Experiments were conducted in buffer containing 50 mM Na₂HPO₄ (pH 8.0), 150 mM NaCl, and 2 mM 2-mercaptoethanol (2-ME). Dissociation constants (K_D) were obtained by fitting the anisotropy data to a single-site binding model, assuming hyperbolic behavior. For sodium signal recordings, HEK293 cells were transiently transfected with FLAG-TRPM4 and HA-14-3-3η. 48 h after transfection, cells were incubated with 10 μM SBFI-AM (Thermo Fisher Scientific, Waltham, MA, USA; Cat#S1263) for 30 min at room temperature and then washed twice with modified Ringer’s medium (pH 7.4) containing 140 mM NaCl, 2.5 mM KCl, 10 mM glucose, 2 mM CaCl₂, 1 mM MgCl₂, and 10 mM HEPES. After 2 min of baseline recording in Ringer’s solution, cells were stimulated with 3 μM Necrocide-1 (MedChemExpress, Princeton, NJ, USA; Cat#HY-14307) for 13 min. Cells were excited at 340 ± 13 nm and 380 ± 13 nm, and fluorescence was recorded at 520 ± 13 nm every 5 s using a Synergy H1 microplate reader with a monochromator illumination system (BioTek, Winooski, VT, USA).

Pull-down Assays

GST and GST-14-3-3η fusion proteins were expressed in E. coli DH5α. Bacterial cultures were grown overnight at 37 °C in LB broth under constant shaking. Protein expression was induced by adding IPTG to a final concentration of 0.5 mM for 4 h at 37 °C. Bacterial cells were then harvested by centrifugation and resuspended in lysis buffer (50 mM Tris-HCl (pH 7.4), 50 mM NaCl, 5 mM MgCl₂, 1% (v/v) Triton X-100, 1 mM DTT, 1 mM PMSF, and 1× protease inhibitor cocktail (PIC)). Cells were lysed by sonication, and the lysate was centrifuged at 11,000 × g for 15 min at 4 °C. The resulting supernatant was subjected to overnight affinity purification at 4 °C using glutathione Sepharose 4B beads (GE Healthcare, Cat# 17-0756-01). The column was washed five times with wash buffer before further processing. For pull-down assays, immobilized GST fusion proteins were incubated with lysates from HEK293 cells transfected with FLAG-hTRPM4. Cells were lysed in a buffer containing 50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1 mM EDTA, 1 mM NaVO₄, 5 mM NaF, 1% v/v Triton X-100, 1 mM PMSF, and 1× PIC. Lysates were centrifuged at 11,000 × g for 10 min at 4 °C, and the resulting supernatant was incubated overnight at 4 °C with GST or GST-14-3-3η-containing beads. After incubation, beads were washed seven times with washing buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1 mM EDTA, 1 mM NaVO₄, 5 mM NaF, and 1% v/v Triton X-100). Bound proteins were eluted in reducing sample buffer (RSB; 60 mM Tris-HCl, pH 6.8, 25% (v/v) glycerol, 2% (w/v) SDS L5750, 14.4 mM 2-mercaptoethanol, and 0.1% (w/v) bromophenol blue), resolved by SDS-PAGE, and analyzed accordingly.

Immunoblotting

Following SDS-PAGE, proteins were transferred to nitrocellulose membranes (GE Healthcare Life Sciences), which were then blocked for 1 h with BLOTTO [4% w/v nonfat dry milk/0.1% v/v Tween-20 in Tris-buffered saline (TBS: 50 mM Tris-HCl, 150 mM NaCl, adjusted to pH 7.5)], followed by overnight incubation with anti-TRPM4 (Alomone Laboratories, ACC-044, Lot ACC044AN0502) and 2 h incubation with anti-GST (NeuroMab, N100/13) primary antibodies, diluted 1:1,000 at 4 °C. After 3 washes with BLOTTO, the membranes were incubated with the appropriate HRP-conjugated secondary antibody using a dilution 1:10,000 for 1 h at room temperature. After 3 washes for 10 min each with 0.1% v/v Tween-20/TBS, immunoblots were visualized using Pierce ECL Western Blotting Substrate (Thermo Fisher Scientific). The images were acquired with a Mini HD9 imager (Uvitec Ltd.) and quantified by measuring the optical density of the bands with ImageJ2/FIJI v.2.9.0 software.

Results

Structural Determinants of 14-3-3 Binding Specificity

Because 14-3-3 proteins interact with a wide variety of partners, including ions, ligands, peptides, and proteins, we performed an extensive search to retrieve all reported human 14-3-3 complex structures, enabling the identification of structural patterns associated with each type of binding partner. To initiate this analysis, genes belonging to the 14-3-3 family were identified through a UniProt search. A total of 12 genes were found, encoding the seven canonical 14-3-3 isoforms along with additional related proteins. A multiple sequence alignment of the isoforms revealed a high degree of conservation, with pairwise identities ranging from approximately 60 to 87% (Figure S1). Using these sequences as a reference, 488 3D structures containing 14-3-3 proteins were retrieved from the PDB. Of these, 463 corresponded to experimentally validated structures from Homo sapiens. The σ isoform showed the highest structural representation, accounting for 348 structures (Supplementary Table 1).

Among the human structures analyzed, 8 involved binding to metal ions, 265 were associated with small molecules, 154 featured peptide interactions, and only 7 captured complexes with full-length proteins. For each case, 14-3-3 residues located at the interaction interface were identified using distance cutoffs of 5 and 10 Å, enabling the determination of the principal contact sites mediating 14-3-3 interactions with its binding partners. Based on these data, contact-frequency profiles were projected onto the structure of the 14-3-3σ isoform (PDB ID: 1YWT) to identify the most recurrent consensus binding regions. For protein partners and peptide fragments, interactions predominantly occurred within the central amphipathic groove. In contrast, interactions with ligands and metal ions tended to localize to the outer surface of the 14-3-3 protein (Figure S2).

Complementing these analyses, structural binding patterns were examined by using the 3D-PP server. An initial set of 1128 patterns was obtained and subsequently filtered according to the criteria recommended by the developers, yielding 737 high-confidence structural motifs. Of these, 680 corresponded to protein and peptide binding patterns, whereas 53 were associated with ligands and 4 were associated with metal ions. These results were then compared to those derived from the contact-frequency maps, considering only protein and peptide interactions. The analyses showed strong concordance with previously identified interaction regions. Nevertheless, 3D-PP revealed additional putative interaction sites localized near helices H1 and H2 on the frontal face of the 14-3-3 structure.

Motifs and Sequence Patterns of 14-3-3 Binding Identified in TRP Channels

From the structural analysis of 14-3-3-bound complexes, sequence patterns corresponding to interface residues were extracted. A total of 652 motifs were initially identified, displaying residue conservation between 20% and 70%. To focus on robust and biologically meaningful patterns, motifs were filtered based on length (4-9 residues) and conservation (≥20%). This yielded a final set of 20 high-confidence patterns, comprising 10 motifs derived from protein partners and 10 from peptide sequences interacting with 14-3-3 (Table S2). Motifs identified in peptides were dominated by serine-containing sequences, while those from proteins showed more heterogeneous residue compositions, reflecting the increased structural complexity of full-length interactors.

Given the recent reports linking 14-3-3 proteins to TRP channel regulation, − we examined the presence of 14-3-3-associated motifs within TRP channel sequences obtained from the UniProt database. Among 206 TRP entries with resolved 3D structures, 31 motif-containing regions were identified across 8 organisms, corresponding to 18 distinct TRP family members. To evaluate whether these sites could indeed engage 14-3-3 proteins, we performed blind docking simulations using representative members of each TRP subfamily against all 14-3-3 isoforms. This unbiased strategy enabled us to assess the structural compatibility and the overall plausibility of TRP-14-3-3 interactions. An interaction was considered plausible if the predicted 14-3-3 contact region occurred within 5 Å of the motif-identified binding sites on the channel. Notably, four sequence motifs emerged as recurrent and potentially critical for 14-3-3 binding within the TRP family, as they were identified in representatives of the TRPA, TRPC, TRPM, and TRPV subfamilies: R-x(0–3)-S, S-x(1–3)-R, S-x­(3,5)-S, and L-x­(1,2)-K. Structural inspection revealed that these motifs are consistently located within cytoplasmic regions of the channels and comprise solvent-accessible residues, supporting their compatibility with 14-3-3 protein binding.

Interaction Sites Determined by Coevolutionary and Machine Learning-Based Analysis

Evolutionary coupling analyses were performed for representative members of the four TRP channel families studied (TRPA, TRPC, TRPM, and TRPV) to identify and corroborate residues potentially involved in interactions with 14-3-3 proteins. In all cases, the 14-3-3ζ isoform was selected as the interaction partner as it exhibits the highest minimum sequence identity relative to all other human 14-3-3 isoforms. Across all TRP representatives, evolutionary couplings were predominantly localized to cytoplasmic regions of the channels (Figure S3). Several TRP channels displayed clusters of coevolving residues concentrated in N- and C-terminal intracellular segments, suggesting these regions as preferential interfaces for 14-3-3 binding. Additionally, putative 14-3-3 binding sites were predicted using the 14-3-3Pred web server, which integrates three independent machine learning-based classifiers: an artificial neural network (ANN), a position-specific scoring matrix (PSSM)-based model, and a support vector machine (SVM). This analysis identified serine (S) and threonine (T) residues with high propensity for phosphorylation-dependent 14-3-3 binding in ten analyzed TRP channel members (Table S3). The convergence of predictions across multiple classifiers supports these residues as high-confidence candidates for mediating TRP-14-3-3 interactions and underscores the importance of integrating diverse methodological approaches to ensure reliable predictions.

Finally, to integrate 14-3-3 binding pattern analyses, blind docking contacts, coevolutionary predictions, and machine learning-based binding site predictions, a consensus scoring function was developed, as described in the Methods section. The resulting consensus scores were mapped onto representative members of each TRP channel family, including TRPA1, TRPC3, TRPM4, and TRPV3 (Figure B). Regions with elevated consensus scores were predominantly localized to cytoplasmic domains, whereas transmembrane segments showed little to no signal, consistent with the intracellular binding preference of the 14-3-3 proteins. High-scoring residues formed discrete interaction patches rather than uniformly distributed surfaces. Notably, despite family-specific differences in the distribution and density of the scores, recurrent interaction hotspots were observed across multiple TRP subfamilies. Together, these results reinforce the existence of shared, evolutionarily conserved intracellular determinants that are likely to mediate TRP-14-3-3 interactions.

Structural Modeling and Dynamic Characterization of the TRPM4-14-3-3γ Interaction

To validate the association between 14 and 3-3 proteins and TRP channels, a structural model of TRPM4 interacting with the 14-3-3γ homodimer was generated, as experimental evidence has previously demonstrated this interaction. Initial docking focused on S88 of TRPM4; however, this residue is buried in the available channel structure and remained inaccessible despite multiple modeling attempts using Rosetta FloppyTail (Figure S4), a protocol specialized in modeling flexible and intrinsically disordered regions, indicating that it is unlikely to constitute a primary 14-3-3 binding site. In contrast, as shown in Figure A, the consensus scoring analysis for the TRPM4-14-3-3γ interaction, derived from the integrative framework described above, identified the region encompassing residues 341-366 as a critical determinant for 14-3-3 binding. Using this region as the input for guided docking simulations, eight statistically significant conformational clusters were identified. The most populated and lowest-energy cluster revealed an interaction interface mediated by residues L221, Q224, L225, R227, and D228 on 14-3-3γ and R353, T356, R357, and E359 on TRPM4 (Figure B). The stability and persistence of this interface were further assessed by MD simulations (Figure S5). Analysis of three independent replicas revealed a well-defined interaction surface characterized by persistent inter-residue contacts. Contact maps showed a concentrated cluster of interactions involving TRPM4 residues 340-400 and 14-3-3γ residues 215-235 (Figure C). Notably, electrostatic and hydrogen-bonding interactions were consistently maintained across all simulation replicas (e.g., residue pairs E389-R227, R353-D228, and R248-E213), suggesting a dominant role in complex stabilization. Additional contacts contributed by neighboring residues further reinforced the interface, indicating that TRPM4 engages 14-3-3γ through a compact and cooperative interaction network (Figure D).

2.

Structural modeling of the TRPM4-14-3-3γ interaction. (A) Consensus scoring profile for TRPM4 identifying putative 14-3-3γ binding regions within the cytoplasmic domain. The transmembrane (TM) region of the channel is highlighted in teal. (B) Structural model of the TRPM4-14-3-3γ complex generated by protein-protein docking guided by the consensus scoring predictions. TRPM4 is shown in teal and the 14-3-3γ dimer in light orange. The inset highlights a close-up view of the predicted interaction interface, with key residues on both TRPM4 and 14-3- 3γ contributing to complex formation. (C) Contact map of the primary TRPM4-14-3-3γ interaction site calculated from three independent MD simulation replicas. The color scale represents the average inter-residue distances. The inset shows a representative snapshot of the simulated TRPM4-14-3-3γ complex. (D) Contact network representation of the TRPM4-14-3-3γ interface. Residues from TRPM4 (teal) and 14-3-3γ (light brown) are arranged in a circular layout, with connecting lines representing persistent interprotein contacts observed during the simulations.

Experimental Validation of the Preferred 14-3-3 Binding Site on TRPM4

Given the discrepancy between the TRPM4-14-3-3γ interaction site predicted in this study and the site previously reported by Cho et al., we designed an in vitro experimental strategy to determine the region of TRPM4 preferentially recognized by 14-3-3 proteins. To this end, the region predicted to mediate 14-3-3 binding was preserved, while restricting expression to the N-terminal domain of the channel. Specifically, the Melastatin Homology Regions 1 and 2 (MHR1/2, residues 1-389) of human TRPM4 were recombinantly expressed in E. coli and purified. Two serine-to-cysteine substitutions were introduced (S27C and S148C) to enable site-specific labeling with the fluorescent probe Bodipy FL (Figure A). As described in Kalinin et al., the accessible volumes calculated for a fluorophore at specific labeling positions indicate that fluorescence anisotropywhich reports on the rotational freedom of the moleculeis sensitive to protein binding events occurring in close spatial proximity. Accordingly, the binding of purified 14-3-3 isoforms near a labeled position is expected to restrict fluorophore motion, producing a measurable change in anisotropy.

3.

Experimental validation of the TRPM4-14-3-3 interaction and identification of the preferred binding site. (A) Structural model of the TRPM4 N-terminal region (MHR1/2) in complex with the 14-3-3γ dimer. The two labeling positions explored for fluorescence anisotropy experiments (S27C and S148C) are highlighted, illustrating their spatial relationship to the predicted 14-3-3 binding interface. (B) Fluorescence anisotropy binding curves for Bodipy FL-labeled TRPM4-MHR1/2 variants incubated with increasing concentrations of 14-3-3γ and 14-3- 3η.

Fluorescence anisotropy measurements revealed distinct binding behaviors for the two TRPM4-MHR1/2 variants tested (Figure B). For the S148C construct, located near the previously proposed S88 site, only minimal changes in anisotropy were observed upon increasing concentrations of either 14-3-3γ or 14-3-3η, indicating little to no restriction of fluorophore motion. This suggests that the binding of 14-3-3 proteins does not occur in spatial proximity to the S88 region. In contrast, the S27C variant, positioned near the interaction site predicted in this study, exhibited a concentration-dependent increase in anisotropy for both 14-3-3 isoforms. This behavior is consistent with direct binding near the S27 region, resulting in reduced rotational freedom of the fluorophore. Together, these results support the conclusion that binding of 14-3-3 to TRPM4 occurs away from the previously proposed S88 site, favoring instead the N-terminal region identified by our consensus-guided structural analysis. Furthermore, the calculated dissociation constants (K _d) indicate moderate-affinity interactions within a physiologically relevant range, with values of 14 μM for 14-3-3γ and 8.5 μM for 14-3-3η, consistent with transient and specific protein–protein interactions.

14-3-3η Isoform Interacts with TRPM4 and Inhibits Sodium Influx

As demonstrated by the fluorescence anisotropy analyses, additional 14-3-3 isoforms beyond those previously reported are likely to interact with TRPM4. In line with this observation, mass spectrometry-based proteomics analyses performed by our group indicated a potential interaction between TRPM4 and 14-3-3η. To validate this interaction, we performed GST pull-down assays using lysates from HEK293 cells expressing FLAG-tagged human TRPM4. The TRPM4 channel was specifically detected in pull-downs performed with GST-14-3-3η, but not with GST alone, indicating a specific association between TRPM4 and the 14-3-3η isoform (Figure A). We next investigated whether 14-3-3η regulates the channel activity. Sodium signaling recordings were performed using the sodium-sensitive fluorescent probe SBFI-2AM in HEK293 cells expressing FLAG-TRPM4 and HA-14-3-3η. Notably, the results revealed a reduction in sodium influx triggered by the TRPM4 activator necrosulfonamide-1 (Nec-1) when HA-14-3-3η was coexpressed (Figure B–C). Based on the TRPM4-14-3-3η interaction interface, we designed cell-permeable peptides corresponding to residues 341-366 of TRPM4 fused to the HIV-1 TAT sequence (TAT-TRPM4), synthesized by GenScript, to disrupt the interaction between the two proteins. Sodium signaling recordings were then performed in HEK293 cells expressing FLAG-TRPM4 and HA-14-3-3η after preincubation with 10 μM TAT-TRPM4 for 24 h. Under these conditions, a partial rescue of sodium influx was observed in the presence of HA-14-3-3η (Figure C–D). In contrast, incubation with TAT-TRPM4 in the absence of HA-14-3-3η resulted in a reduction of sodium influx, suggesting that the peptides may also disrupt interactions with endogenous 14-3–3 isoforms.

4.

Biochemical and functional evidence for TRPM4 regulation by the 14-3-3η isoform. (A) Immunoblots from GST pull-down assays performed using lysates from HEK293 cells expressing FLAG-tagged TRPM4. The GST-14-3-3η fusion protein is detected at the expected molecular weight (∼54 kDa) using an anti-GST antibody. (B) Representative traces (mean ± SEM) of cytosolic sodium signals measured with SBFI-AM in HEK293 cells. (C) Area under the curve (AUC; mean ± SD) from five independent experiments in HEK293 cells transfected with empty vector, FLAG-TRPM4, and FLAG-TRPM4 + HA-14-3-3η. Statistical analysis was performed using one-way ANOVA followed by Dunnett’s multiple-comparisons test. Small dots represent individual wells; large circles represent the mean of each independent experiment. (D) Representative traces (mean ± SEM) of cytosolic sodium signals measured with SBFI-AM in HEK293 cells preincubated with TAT-TRPM4 peptides. (E) AUC (mean ± SD) from five independent experiments in HEK293 cells transfected with FLAG-TRPM4 or FLAG-TRPM4 + HA-14-3-3η and preincubated with TAT-TRPM4 peptides. The same statistical analysis and data representation described in panel B were used.

Discussion

Ion channels are integral membrane proteins that are synthesized in the endoplasmic reticulum, processed through the Golgi apparatus, and subsequently delivered to the plasma membrane via vesicular trafficking, where membrane fusion enables their insertion into the lipid bilayer. This multistep transport process is regulated by specific protein partners. Of particular relevance to our research, 14-3-3 proteins have emerged as key modulators of TRP channel trafficking and surface expression, suggesting a targeted role in ion channel trafficking dynamics. −

In this study, we present a comprehensive and integrative characterization of 14-3-3 proteins as binding partners of TRP channels. By combining structural analyses, coevolutionary and machine learning-based predictions, atomistic simulations, and targeted experimental validation, we define the molecular determinants that mediate TRP-14-3-3 recognition. Our systematic examination of all available human 14-3-3 complex structures revealed a clear segregation of interaction regions depending on the nature of the binding partner: peptides and full-length proteins predominantly engage the canonical amphipathic groove, whereas ligands and metal ions interact with peripheral surfaces of the 14-3-3 dimer. Notably, most of the analyzed complexes corresponded to the σ isoform, a variant of particular interest due to its established relevance in multiple cancer types. Furthermore, by mapping key sequence and structural determinants of 14-3-3 recognition across the TRP channel family, our work expands the current understanding of how 14-3-3 proteins associate with TRP channels and reveals conserved molecular features shared across multiple TRP subfamilies. Specifically, we identified several recurrent short linear motifs enriched in serine, threonine, and polar residues, located within cytoplasmic regions of TRP channels. This motif architecture aligns with the well-established preference of 14-3-3 proteins for phosphorylated serine or threonine residues, , suggesting that multiple TRP channels may be regulated through similar phospho-dependent mechanisms. At the same time, it is important to recognize that 14-3-3 proteins can also engage certain partners in a phosphorylation-independent manner through electrostatic and structural complementarity, particularly when negatively charged or acidic residues functionally mimic phospho-dependent interactions. Notably, the convergence of motif-based analysis with blind docking, evolutionary couplings, and machine learning predictions strongly supports the existence of shared, evolutionarily conserved 14-3-3 interaction determinants across multiple TRP subfamilies.

Beyond the identification of 14-3-3 proteins as binding partners of TRP channels, a central contribution of this work is the redefinition of the TRPM4-14-3-3γ interaction interface. Although previous studies proposed S88 as a critical determinant for 14-3-3 binding, our structural and conformational analyses suggest that S88 remains largely buried within the channel architecture, even when modeled in its phosphorylated state and subjected to extensive conformational sampling. The absence of solvent exposure or local rearrangements compatible with 14-3-3 engagement indicates that S88 is unlikely to function as a primary binding site, at least not in the resolved channel conformation currently available or under the structural and biochemical conditions examined in this study. Instead, our consensus scoring approach identified a distinct N-terminal region as the most probable interaction hotspot. Guided docking and molecular dynamics simulations revealed a stable interface mediated by complementary polar and charged residues on both TRPM4 and 14-3-3γ, with hydrogen bonding and electrostatic interactions dominating the binding mechanism. These features align with the transient yet specific nature characteristic of many 14-3-3 protein–partner interactions and are consistent with canonical 14-3-3 recognition principles, which typically involve hydrogen-bond networks and positively charged residues surrounding phosphoserine/threonine sites. ,,

Our computational predictions were experimentally validated through fluorescence anisotropy assays using site-specific labeling of the TRPM4-MHR1/2 region. Notably, the S148C variant, located near the previously proposed S88 site, showed negligible anisotropy changes upon addition of 14-3-3γ or 14-3-3η, indicating the absence of binding events in the spatial proximity of this region. However, it remains possible that this site constitutes a low-affinity interaction interface that is not readily detected under the experimental conditions employed in this study. In contrast, the S27C variant, positioned adjacent to the interface identified by our integrative analysis, displayed a clear concentration-dependent increase in anisotropy, consistent with a reduced fluorophore rotational freedom caused by 14-3-3 binding. These results provide strong evidence that 14-3-3 recognition does not occur near S88 but instead favors the N-terminal region predicted by our consensus model. Furthermore, the dissociation constants measured for 14-3-3γ and 14-3-3η (14 and 8.5 μM, respectively) fall within the low-micromolar range typical of transient regulatory protein–protein interactions, , supporting a model in which 14-3-3 proteins act as dynamic modulators rather than constitutive channel subunits. Such affinities are well suited to biological processes, such as channel folding, assembly, and trafficking, where reversible interactions are required.

Finally, our data demonstrate that TRPM4 interacts with additional 14-3-3 isoforms beyond 14-3-3γ, including 14-3-3η, thereby expanding the scope of previously reported associations. GST pull-down assays support a direct association between TRPM4 and 14-3-3η, while sodium imaging experiments reveal that this isoform inhibits TRPM4-mediated sodium influx. Disruption of the TRPM4-14-3-3η interface using TAT-TRPM4 peptides partially restored sodium influx, indicating a direct and functional regulatory role of this interaction. These results, together with the anisotropy measurements, challenge the notion that TRPM4 binding to 14-3-3 proteins is isoform-specific, as previously reported. Instead, our results indicate that TRPM4 associates with multiple 14-3-3 isoforms, suggesting a more flexible and context-dependent regulatory mechanism. Notably, the inhibitory effect observed for 14-3-3η contrasts with the previously reported role of 14-3-3γ, highlighting isoform-dependent functional outcomes. This divergence suggests that distinct 14-3-3 isoforms may fine-tune TRPM4 behavior by differentially modulating channel trafficking, stability, or gating. Further studies will be required to elucidate the molecular basis and physiological relevance of this inhibition.

Conclusions

Our findings position 14-3-3 proteins as general and evolutionarily conserved regulators of TRP channels, extending their well-established roles in other ion channel families. By integrating sequence and structural analyses with computational modeling and experimental validation, we establish a broadly applicable framework for identifying and characterizing transient protein–protein interactions in complex membrane systems. Our results show that TRP-14-3-3 interactions preferentially occur within solvent-exposed N-terminal regions of the channels, consistent with established 14-3-3 recognition mechanisms that may involve phosphorylated residues, as well as broader electrostatic and structural complementarity. Through combined computational and experimental approaches, we demonstrate that TRPM4 associates not only with 14-3-3γ but also with 14-3-3η, challenging the notion of strict isoform specificity and revealing a broader regulatory landscape. Given the high level of sequence conservation among 14-3-3 isoforms and the comparable binding affinities observed, additional isoforms are likely to contribute to TRP channel regulation. Together, these findings support a model in which 14-3-3 proteins act as dynamic modulators of TRP channel trafficking, localization, and function. Future studies should elucidate how post-translational modifications, cellular signaling pathways, and tissue-specific isoform expression patterns shape TRP-14-3-3 interactions in vivo. Moreover, systematic targeting of this protein–protein interface may enable the development of novel therapeutic strategies to modulate TRP channel function and surface expression in channelopathies and other TRP-associated diseases.

Input files for the docking and molecular dynamics used in the generation of the results, the multiple sequence alignments from which coevolution was calculated, and the Python scripts used for every part of the Methods section, including analyses of available 14-3-3 structures, identification of 14-3-3 binding sites in TRP channels, docking workflows, coevolutionary and machine learning-based contact predictions, consensus scoring procedures, and the TRPM4-14-3-3γ structural model, are available through Zenodo at 10.5281/zenodo.18272233.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jcim.6c00092.

• A multiple sequence alignment of human 14-3-3 isoforms, gene information, and the distribution of available PDB structures across the different 14-3-3 isoforms. Contact-frequency maps highlighting preferred interaction regions for distinct classes of 14-3-3 binding partners, conserved sequence motifs identified at 14-3-3 interaction interfaces, and coevolutionary coupling profiles between TRP channels and 14-3-3 proteins. Machine learning-based predictions of 14-3-3 binding residues in TRP channels, together with a conformational analysis evaluating the accessibility of residue S88 in TRPM4, and RMSD analyses of the equilibrated MD simulations (PDF)

A.V.-J. and P.G.-D. designed the computational studies. N.P.-V., M.G.-A., and N.S.-G. performed computational modeling, simulations, and data analysis. D.M., I.S., J.A., E.M.-B., E.M., and O.C. conducted the experimental assays. A.V.-J. and P.G.-D. wrote the manuscript. All authors contributed to data interpretation, manuscript revision, and scientific discussion.

The authors declare no competing financial interest.

Published as part of Journal of Chemical Information and Modeling special issue “Computational Chemistry in the Global South: The Latin American Perspective”.