Molecular Dynamics and Quantum Chemical Insights into the Molecular Basis of Differential Toxicity of the Mycotoxin Deoxynivalenol (DON) and Its Microbial Derivatives DOM‑1 and 3-epi-DON

Abstract

Deoxynivalenol (DON) is a trichothecene mycotoxin from Fusarium species that contaminates cereal grains, posing health risks. DON binds to site A of the 60S ribosomal subunit in eukaryotes, triggering a ribotoxic stress response. Despite structural similarity, its metabolites, deepoxy-deoxynivalenol (DOM-1) and 3-epi-deoxynivalenol (3-epi-DON), are nontoxic, but the molecular basis of this difference is unclear. This study employed molecular docking, molecular dynamics (MD) simulations, and quantum chemical calculations based on Symmetry-Adapted Perturbation Theory (SAPT) to analyze the interactions of DON and its metabolites with the ribosomal site. Our results reveal that DON adopts a unique binding conformation, enabling strong and stable interactions with an Mg²⁺ ion and nucleotide U2873, maintaining its position within the ribosomal pocket throughout 500 ns of MD simulations. In contrast, DOM-1 and 3-epi-DON fail to sustain these interactions due to the absence of the epoxide group or altered hydroxyl orientation, leading to their dissociation. Additional MD simulations of the complex with another potent toxic trichothecene, verrucarin A showed stable binding at the same site, emphasizing the importance of these molecular interactions for toxicity. Further, quantum chemical analyses highlighted the energetic contributions of electrostatic and induction forces in stabilizing DON within the binding pocket. These data are in line with experimental studies in HCT116 human colon cancer cells confirming the lack of cytotoxicity of DOM-1 and 3-epi-DON and demonstrating differential ribotoxicity of DON and its metabolites. These results highlight a clear structure–activity relationship, where modifications at key positions markedly affect trichothecene binding and toxicity. Together, these findings advance our understanding of trichothecene toxicity and support the development of detoxification strategies and novel therapeutics targeting this ribosomal site.

Introduction

Mycotoxins are secondary metabolites produced by various filamentous fungi that often contaminate agricultural commodities and food products. , They are the most prevalent natural dietary toxins, contaminating up to 70% of global crop production. The risks associated with exposure to some of these compounds are well-characterized, and regulations and recommendations have been established for their maximum levels in food. Deoxynivalenol (DON) is a potent mycotoxin produced by Fusarium species that frequently contaminates cereal grains, including corn, wheat, oats, and rice, posing significant health risks to both humans and animals. − It belongs to the trichothecene family of mycotoxins, known for their potent inhibitory effects on protein synthesis. Acute DON intoxication has been shown to cause vomiting, nausea, and diarrhea, while chronic exposure has been linked to food refusal, anorexia, reduced body weight gain, and altered immune response. − DON binds to the A site of the 60S subunit of the eukaryotic 80S ribosome, initiating its ribotoxic effects. , However, its microbial metabolites, such as deepoxy-deoxynivalenol (DOM-1) and 3-epi-deepoxy-DON (3-epi-DON), are nontoxic despite sharing a similar structural framework with DON. , For instance, the studies in human intestinal cells, in intestinal explants, and in piglets have demonstrated that the metabolites of DON, specifically DOM-1 and 3-epi-DON, do not induce toxic effects, as evidenced by the absence of histological alterations and inflammatory responses in treated animals. , Despite their structural similarities, the molecular mechanisms that contribute to the higher toxicity of DON compared to its less toxic metabolites, DOM-1 and 3-epi-DON, remain largely unclear. Previous studies have primarily focused on experimental evaluations of DON and its metabolites, assessing their intestinal toxicity, gene expression changes, and ribosomal interactions using cellular models. − These studies demonstrated that while DON disrupts intestinal function and activates mitogen-activated protein kinases (MAPKs), its metabolites (DOM-1 and 3-epi-DON) do not induce toxicity or MAPKs activation. However, in comparison to DOM-1, 3-epi-DON still remains less explored, with limited information available on its molecular interactions and toxicological behavior. Additionally, previous structural analyses were limited to docking studies, which, while informative, do not provide comprehensive insights into the dynamics and stability of these interactions over time. , One inherent limitation of molecular docking is that the overall structural backbone is constrained to remain rigid (with only a few rotatable bonds permitted), thereby neglecting the dynamic processes that occur as the ligand approaches the binding pocket. − Additionally, the use of low-resolution structures (3.1–3.3 Å) in these previous studies may further contribute to inaccuracies in modeling toxin positioning within the catalytic site. Structural, dynamic, and energetic differences are essential for understanding the molecular mechanisms underlying the toxicity of DON and the reduced toxicity of its metabolites. A deeper understanding of the molecular basis of differential toxicity requires an analysis of ligand stability, interaction patterns, and structure–activity relationships, which are essential for toxicity prediction, drug design, and therapeutic interventions. Such insights are crucial for the development of detoxification strategies aimed at mitigating the harmful effects of toxic trichothecenes.

Molecular dynamics (MD) simulations offer a powerful approach to study molecular recognition processes, enabling the exploration of conformational dynamics and interaction mechanisms in biomolecular systems. − This approach has been implemented to provide detailed insights into the stability, dynamics, and molecular interactions of ribosomal complexes at the atomic level. − In this study, we investigated the trichothecene mycotoxin deoxynivalenol (DON) and its metabolites, DOM-1 and 3-epi-DON, within the eukaryotic ribosome. The experimentally resolved DON–ribosome complex (PDB ID: 4U53) served as the structural reference, while docking was used to generate the corresponding complexes for DOM-1 and 3-epi-DON. This was followed by molecular dynamics (MD) simulations to investigate their stability and interaction profiles. This approach enabled a detailed examination of how structural modifications in DON influence binding behavior and overall complex stability. In MD simulations, DON remained stably bound, maintained by persistent interactions with an Mg²⁺ ion and predicted key residues. In contrast, the metabolites exhibit significantly altered interaction patterns due to minor structural modifications, leading to weaker interactions with the positively charged Mg²⁺ ion and the key residues; this ultimately reduces their stability within the binding pocket. From these studies, Mg²⁺ has emerged as a key factor in stabilizing ligand binding within the ribosomal A site, in addition to the contributions from key residues. These findings emphasize the importance of Mg²⁺ not only for understanding mycotoxin toxicity but also for guiding structure-based drug design efforts to mitigate ribotoxicity. In addition to DON and its metabolites, we also evaluated verrucarin A, a more toxic trichothecene compared to DON, which exhibited stable binding through conserved Mg²⁺ and RNA contacts, underscoring the importance of these interactions in toxicity. Residue-level interaction energy analysis employing MD simulation and Symmetry-Adapted Perturbation Theory (SAPT) − based on quantum chemical calculations revealed that electrostatic, polarization, and coordination forces were key contributors to DON’s stability, while these interactions were markedly reduced in its metabolites. Experimental assays in human colon cancer HCT116 cells supported these findings, ranking the compounds in toxicity as DON ≫ DOM-1 ≈3-epi-DON, with verrucarin A showing even greater toxicity. This further underscores the importance of additional key molecular interactions observed in the rRNA (rRNA) and verrucarin A complex. This study provides a detailed analysis of the structural and dynamic differences between DON and its metabolites, within the binding site, explaining their varying toxicities. These findings offer valuable insights into the molecular mechanisms driving the toxicity of trichothecenes. Furthermore, this research paves the way for future investigations focused on developing strategies to reduce mycotoxin contamination. Specifically, the insights gained could guide the design of small molecules aimed at targeting this same binding site to counteract or modify the toxic effects of these compounds, thus offering potential avenues for therapeutic or preventive approaches to mitigate the harmful impact of mycotoxins.

Results

Comparative Cytotoxicity of DON and Its Two Metabolites DOM-1 and 3-epi-DON

We first confirmed the lack of toxicity of DOM-1 and 3-epi-DON when compared to DON and verrucarin A. In this aim, the human colon cancer HCT116 cell line was used, as the intestinal toxicity of DON is well-established. ,, HCT116 cells were treated for 72 h with different concentrations of each compound or with the DMSO vehicle, and viability was measured by the CellTiter-Glo assay (Figure S1). DON and verrucarin A induced a dose-dependent loss of cell viability, with verrucarin A showing a massive cytotoxic activity compared to DON. Indeed, the half maximal inhibitory concentration (IC₅₀) of DON is 1.11 μM, which is 1643-fold higher than the IC₅₀ of verrucarin A (0.675 nM). As expected, the DON metabolites DOM-1 and 3-epi- DON do not affect cell viability, even at the highest concentrations of 40 μM and 80 μM, respectively.

Enhanced Stability of DON in the Ribosomal 60S Subunit Complex

In the X-ray structure of DON bound to the eukaryotic 60S subunit of the 80S ribosome from Saccharomyces cerevisiae, the DON binding pocket includes key nucleotide residues U2869, U2873, G2874, U2875, G2816, A2820, C2821, and G2403. Notably, this binding pocket is composed solely of 25S rRNA, a crucial part of the peptidyl transferase center (PTC) in the 60S large subunit of the 80S ribosome. Additionally, an Mg²⁺ ion located at a distance of 2.7 Å from DON establishes electrostatic interactions and coordination with the nearest hydroxyl group of the ligand (Figure ). These residues in the X-ray complex likely play a crucial role in forming interactions with DON. Specifically, hydrogen-bond (H-bond) interactions between the oxygen of the epoxy ring in DON and the hydroxyl moiety of the sugar in the nucleotide residue U2873 appear to play a significant role in ligand binding. Likewise, the oxygen in the sugar ring of U2875 also forms an H-bond with the ligand and thus enhances the ligand stability in the binding pocket. Furthermore, DON appears to establish a hydrogen bond with the side chains of A2820 and G2403. Other residues predominantly engage in weak hydrophobic interactions to further enhance the stability of the complex with DON. As the crystal structures of the eukaryotic 80S ribosome in complex with DOM-1 and 3-epi-DON remain unavailable, the docking complexes generated for metabolites of DON exhibit comparable interactions (Figure ) with GOLD PLP fitness scores of 61.6 for DOM-1, 61.1 for 3-epi-DON, and 66.9 for DON. However, in the complex with 3-epi-DON, interactions with the Mg²⁺ ion appear weaker due to the increased distance resulting from the hydroxyl group switch at position 3. Similarly, the docking complex with DOM-1 lacks a hydrogen bond with U2873, as it contains a methylene group instead of a polar epoxide ring. Despite these minor structural differences, the binding poses of DON and its metabolites remain similar, occupying the same binding pocket in the docking simulations. Hence, these minor structure variations and static complexes alone do not fully explain or provide the molecular basis for the substantial differences in toxicity levels between DON and its metabolites. Therefore, to obtain deeper insights, 500 ns molecular dynamics (MD) simulations were performed on these complexes. The MD simulations confirm the stability of the receptor–ligand complexes, with the RNA backbone exhibiting average RMSD and RMSF values of 1.1–1.7 Å and 0.5–0.7 Å, respectively (Table S1, Figures S2, S3A, S4, and S5). Compared to its metabolites, the MD simulations of the DON–ribosome complex revealed a distinct pattern of additional key interactions, resulting from a unique and preferred binding posepotentially explaining the molecular basis for its higher toxicity. In this binding pose, hydrogen bonding interactions between the hydroxyl group at position 3 in the trichothecene ring of DON and the negatively charged oxygen of the phosphate group (PO₄ ^–) in U2873 played a pivotal role in ligand stability (Figure ), forming consistent interactions with over 95% occupancy (Figures S6 and S7A). Likewise, the epoxy group in DON formed a hydrogen bond with the hydroxyl group of the backbone sugar of residue U2873 in the X-ray complex but only showed 10% occupancy in the MD simulation. Nevertheless, being a polar group, it plays an important role in DON’s stability within the hydrophilic region of the binding pocket and played a key role in retaining DON near the highly charged Mg²⁺ ion, thereby facilitating further interactions. The MD results showed that in addition to the hydroxyl group at position 3 of the trichothecene ring, the oxygen atom at position 1 (O1) shifted closer to the Mg²⁺ ion, forming strong electrostatic interactions and coordination, with the O1–Mg²⁺ distance stabilizing at approximately 2.0–2.2 Å (Figures , S7B, S8, and S9). Consequently, this interaction stabilizes the unique binding conformation of DON, further enhancing the overall stability of the complex. These interactions and the binding conformation of DON were not observed in the X-ray crystal structure and only emerged during the MD simulation when a stable and unique binding pose was attained within the binding pocket. Furthermore, the oxygen atom in the sugar ring of U2875 established a crucial hydrogen bond with the hydroxyl group at position 7 of DON, exhibiting an occupancy of around 42% (Figure S6). Likewise, other crucial residues, including G2874 and A2872, play a role in significant hydrogen-bond interactions via the backbone sugar and nucleotide side chains. Other than polar interactions, DON also exerts hydrophobic interactions through its methyl group with the aromatic rings of A2820 and C2821. Ligand–residue interactions energy calculations performed using the MD trajectory of the DON–ribosome complex further highlight the significance of these interactions (Table S2). The Mg²⁺ ion exhibited the highest energy contribution at −105.0 kcal/mol, followed by U2873 with −18.2 kcal/mol. Other residues, including A2872, G2874, and U2875, contribute less than −10 kcal/mol but play a significant role in retaining DON within the binding pocket. For residues A2820 and C2821, van der Waals interactions were more prominent than electrostatic interactions in contributing to ligand binding. Although MD simulations provide valuable insights into the system’s stability and binding characteristics, they are limited in their ability to fully capture the detailed nature of noncovalent interactions, such as dispersion forces and induction. Therefore, quantum chemical calculations were conducted alongside MD simulations to further investigate the energetic contributions of key residues including side chain and backbone sugar and phosphate (PO₄ ^–) group in ligand–receptor interactions using SAPT. First, molecular electrostatic potential (MEP) analyses were performed to examine the charge distribution of DON and its metabolites (DOM-1 and 3-epi-DON), which clearly revealed differences among the compounds (Figure S10). The negative potential associated with the epoxide group in DON disappears in DOM-1, while in 3-epi-DON, the hydroxyl group at position 3 generates a negative potential oriented opposite to that in DON. These variations reflect distinct electrostatic characteristics that may influence their receptor interaction profiles. For further analysis, the F/I-SAPT approach was employed to break down pairwise interaction energies between selected molecular fragments into electrostatic, exchange repulsion, polarization, and dispersion components, providing deeper insight into the nature and strength of these interactions. This methodology has been successfully applied in similar systems, including a proflavine–DNA intercalation complex, RNA complexes, and G-protein complexes. The method was applied to the most representative binding pose, selected through clustering the MD simulation trajectory of the complex with DON. The results provided a more accurate and comprehensive picture of the ligand–receptor interactions (Table ). The interaction energy analysis highlighted the dominant role of Mg²⁺ and the phosphate group (PO₄ ^–) of U2873 in ligand binding, with Mg²⁺ contributing −139.6 kcal/mol and the phosphate group of U2873 contributing −16.7 kcal/mol. The results show that the Mg²⁺ ion caused a polarization effect on the two closest oxygen atoms in the ligand and formed strong electrostatic interactions with it. The analysis further indicates strong coordination between DON and Mg²⁺, dominated by a large induction energy (−73.2 kcal/mol) from ligand to metal and negligible reverse induction or dispersion. The close proximity (2.0–2.2 Å) of the hydroxyl group at C3 and the O1 oxygen to Mg²⁺ suggests bidentate coordination, with lone pair donation from both oxygen atoms reinforcing the polarization-driven nature of the interactions. Similarly, the phosphate group of U2873 mainly interacted through electrostatics, with dispersion interactions playing a secondary role. The Mg²⁺ ion likely influences these interactions by altering the local electronic environment, which can induce enhanced dispersion interactions between the phosphate group and DON. In contrast, for residues A2820 and C2821, the primary interactions were dispersion forces, which occurred between the DON and the side chains of these nucleotides. The results from the quantum chemical calculations align with the findings from the MD simulations, where similar patterns of interaction were observed, confirming consistency between the calculated interaction energies and the dynamic behavior captured in the simulations.

1.

Upper panel shows the 2D structures of DON, 3-epi-DON, and DOM-1, with position numbering indicated on the trichothecene core of DON. The lower panel displays the X-ray complex of DON (A) and the docking models of 3-epi-DON (B) and DOM-1 (C) with the 60S subunit (25 rRNA) of 80S ribosome. The complexes highlight the ligands in distinct colors: DON in cyan, 3-epi-DON in blue, and DOM-1 in salmon. All three ligands exhibit comparable binding poses, with minor differences in hydrogen bonding due to the absence of the epoxide group in DOM-1 and the inverted hydroxyl orientation at position 3 in 3-epi-DON.

2.

Binding pose of DON at site A of the 60S ribosome subunit, obtained from MD simulation, reveals key residues and its unique binding conformation (not observed in the X-ray complex), where its oxygen at position 1 (O1) forms more stable interactions with the Mg²⁺ ion in the binding site. These enhanced interactions contribute toward the ligand stability and increased toxicity of DON compared to its metabolites, DOM-1 and 3-epi-DON. (A) Key residues interacting with DON are shown in stick representation, with their thickness and color indicating the relative strength of van der Waals and electrostatic interactions, respectively. The Mg²⁺ ion (colored magenta) forms electrostatic interactions and forms coordination with O1 and the hydroxyl group at position 3 of DON. The corresponding interaction energy values are provided in Table S2. (B) The hydrogen-bond network and interaction patterns with key residues and the Mg²⁺ ion in the binding site. The occupancy of hydrogen bonds is detailed in Figure S6. The nucleotide residue U2873 and Mg²⁺ ion play a crucial role in the polar interactions.

1. Decomposition of the Pairwise Interaction Energies (kcal/mol) between RNA Residues in the Ribosomal Binding Site and DON, 3-epi-DON, and DOM, Obtained from SAPT Calculations .

ligand/residue	receptor fragment	electrostatics	exchange	induction L → R	induction R → L	dispersion	total
DON	all	–132.9	123.6	–55.3	–6.9	–64.0	–135.6
	Mg2+ (3831)	–100.7	34.3	–73.2	0.0	0.0	–139.6
	U2873 (PO4 –)	–21.3	15.0	–2.9	–2.4	–5.1	–16.7
	U2875 (sugar)	–6.2	3.7	–1.0	–0.6	–3.3	–7.4
	U2873 (side chain and sugar)	–4.4	8.3	0.6	–0.6	–8.7	–4.8
	A2820 (side chain)	–3.2	7.7	0.8	–0.4	–9.5	–4.6
	C2821 (side chain)	–1.3	3.0	0.4	–0.2	–3.5	–1.6
DOM-1	all	–84.7	97.2	–29.6	–4.3	–66.4	–87.8
	Mg2+ (3831)	–63.0	15.4	–39.1	0.0	0.0	–86.6
	U2873 (PO4 –)	–23.2	13.1	–6.4	–1.1	–4.8	–22.4
	C2821 (side chain)	–10.2	16.3	–1.9	–0.4	–12.0	–8.2
	A2820 (side chain)	–3.6	8.7	0.4	–0.3	–8.4	–3.2
	G2874 (OH)	–1.2	1.1	0.5	–0.1	–1.4	–1.1
	U2873 (side chain and sugar)	–6.1	17.0	0.0	–0.3	–11.3	–0.7
3-epi-DON	all	–34.0	69.9	–9.2	–4.8	–60.0	–38.1
	U2873 (PO4 –)	–12.4	5.2	–1.9	–1.1	–2.4	–12.5
	C2821 (side chain)	–4.2	9.5	–0.8	–0.4	–9.5	–5.4
	A2820 (side chain)	–4.5	11.8	0.6	–0.5	–11.9	–4.4
	G2403 (side chain)	–6.6	8.1	1.3	–0.6	–5.7	–3.5
	Mg2+ (3831)	4.9	0.0	–7.8	0.0	0.0	–2.9
	U2875 (sugar)	–2.0	2.5	–0.3	–0.1	–2.0	–1.9
	U2873 (side chain and sugar)	–7.1	21.2	–0.6	–1.5	–12.5	–0.5
	G2874(OH)	–0.6	1.5	0.5	0.0	–1.3	0.0
verrucarin A	all	–83.8	87.5	–34.8	–5.7	–72.0	–108.9
	G2403 (side chain)	–18.9	19.0	–8.6	–0.9	–9.1	–18.6
	G2874	–13.4	17.1	–7.6	–0.6	–13.9	–18.5
	Mg2+ (3828)	–6.8	0.000	–11.2	0.0	0.0	–18.1
	G2400 (side chain)	–13.6	9.6	–3.2	–1.1	–4.693	–13.0
	C2821 (side chain)	–8.2	10.8	–3.1	–0.4	–10.4	–11.3
	U2873 (side chain and sugar)	–8.2	11.1	–0.1	–1.1	–10.8	–9.1
	U2875 (side chain and sugar)	–6.5	10.1	1.1	–1.0	–10.1	–6.4
	A2820 (side chain and sugar	–0.5	2.0	0.1	0.0	–2.983	–1.3
G2403	Mg2+ (3828)	–106.4	27.4	–67.4	0.0	–0.2	–146.6

^aL → R/R → L: Ligand → Receptor/Receptor → Ligand.

^bSAPT calculations were performed on the verrucarin A–80S ribosomal complex using the G2403 residue and Mg²⁺ ion.

^cThe term side chain refers to the nucleobase of each nucleotide. Notably, the Mg²⁺ ion and nucleotide residue U2873 play a key role in stabilizing DON within the binding pocket.

Structural Basis for Detoxification: Ribosomal Binding of DON Metabolites DOM-1 and 3-epi-DON

Despite their minor structural differences from DON, in vivo studies indicate that its metabolites, DOM-1 and 3-epi-DON, do not induce ribotoxic effects. , As the experimental structures of the eukaryotic 80S ribosome in complex with DOM-1 and 3-epi-DON are not yet available, molecular docking suggests that their predicted binding poses mimic the binding pose of DON in the X-ray structure of the DON–ribosome complex, maintaining most of the key interactions (Figure ). In order to explore the stability and energetics of the metabolites in the binding pockets, the complexes were subjected to MD simulations and quantum chemical calculations using the F/I-SAPT approach. The MD simulation results in all three replicates show that none of the metabolites were stable in the 500 ns simulation run as indicated by the higher RMSD and RMSF values in Table S1 and Figure S3B. DOM-1 exits the binding pocket after approximately 200 ns, highlighting its weaker binding stability and poor steric fit. This is likely due to lack of key polar interactions caused by the absence of the epoxide group in DOM-1, which also leads to higher exchange repulsion observed in F/I-SAPT. This causes the loss of favorable interactions with U2873 as observed in the MD simulation (Figures S7 and S9) and leads to dissociation from the binding site. The F/I-SAPT energy decomposition analysis (Table ) of the interactions between ligands and the nucleotide U2873 reveals that the epoxide-containing ligand (DON) exhibits significantly stronger binding affinity and stability compared to the methylene-containing ligand (DOM-1). The key difference in interaction energies lies in the exchange repulsion, which is much higher for DOM-1 (17.0 kcal/mol) compared to DON (8.3 kcal/mol). Although DOM-1 contributes stronger dispersion interactions (−11.3 kcal/mol compared to −8.7 kcal/mol for DON), these are outweighed by the epoxide’s superior electrostatic stabilization of the binding conformation and hydrogen bonding potential, especially with the sugar’s hydroxyl group in U2873. Overall, the total interaction energy with U2873 for DON (−4.8 kcal/mol) is much more favorable than for DOM-1 (−0.6 kcal/mol), indicating that epoxide substitution improves ligand binding by enhancing polar interactions and reducing steric repulsion. This suggests that the epoxide’s oxygen atom provides directional electrostatic stabilization of the ligand through hydrogen bonding or dipole interactions with U2873, allowing it to fit better in the binding site. In DOM-1, the absence of the polar epoxide group prevents it from aligning properly in the hydrophilic region of the binding site. As a result, it fails to achieve a binding pose comparable to DON and cannot maintain key interactions with the Mg²⁺ ion, making DOM-1 more prone to exit the binding site in MD simulation. The interaction energy analysis highlights the crucial role of Mg²⁺ in DOM-1 binding, contributing −86.6 kcal/mol to its binding energy. However, this contribution is comparatively lower than that of DON, as DOM-1’s electrostatic interactions with Mg²⁺ involve only one oxygen atom of the hydroxyl group at position 3. In contrast, DON establishes stronger electrostatic interactions through an additional oxygen atom (O1) in its most stable binding configuration (Figures and S8). This limited involvement of a second oxygen or the inability to achieve a binding configuration similar to DON further decreases DOM-1’s overall binding stability. In the docking complex of 3-epi-DON, the epoxide ring appears to form key interactions, including hydrogen bonding with the sugar moiety of nucleotide U2873 (Figure ). However, the orientation of the hydroxyl (OH) group at position 3 in the trichothecene ring system is inverted from that in DON, leading to altered binding interactions. Furthermore, MD simulations of the docking complex revealed significant instability. The ligand failed to maintain the binding pose for effective interaction with Mg²⁺ (Figure S9B), leading to its rapid exit from the binding site at the onset of the simulation. As previously reported, position 3 playing a crucial role in the binding of trichothecenes, the analysis of the complex shows that reversed orientation of the hydroxyl group at position 3 in 3-epi-DON increases the distance from the Mg²⁺ ion. This weakens interactions of 3-epi-DON with both the Mg²⁺ ion and the phosphate group of U2873 (Figure S7A), destabilizing the complex. As a result, 3-epi-DON fails to achieve the stable binding configuration observed in DON. Interaction energy decomposition analysis also indicates that 3-epi-DON interacts less effectively with U2873 compared to DON, primarily due to its altered hydroxyl orientation. Notably, 3-epi-DON shows significantly higher exchange repulsion (21.1 kcal/mol) compared to DON (8.2 kcal/mol) and DOM-1 (17.0 kcal/mol), indicating less favorable spatial fit within the ribosomal binding site. This suggests that although the epoxide ring in 3-epi-DON interacts with U2873, the lack of effective key interactions with Mg²⁺ ion further weakens its binding affinity, as reflected in its lower total interaction energy with U2873 (−0.5 kcal/mol, versus −4.8 kcal/mol for DON). These results from MD simulations and F/I-SAPT are consistent with our experimental data on cell toxicity and ribotoxicity (Figures S1 and ).

4.

Comparative analysis of protein translation inhibition induced by DON, verrucarin A, DOM-1, and 3-epi-DON using puromycin incorporation assay. Verrucarin A and DON markedly inhibit protein synthesis in a dose-dependent manner. In contrast, DOM-1 and 3-epi-DON exhibit minimal or no detectable effects.

Molecular Insights into the Stabilization of the Verrucarin A–80S Ribosome Complex

Verrucarin A, a potent type D trichothecene with a C-4 to C-15 macrocyclic linkage (Figure ), was used as a reference to assess ligand interactions in MD simulations. It inhibits eukaryotic protein synthesis by binding rRNA. Using the yeast 80S ribosome structure complexed with verrucarin A (PDB ID: 4U50), a 500 ns MD simulation showed that verrucarin A remained stably bound within the peptidyl transferase center of 25S rRNA, with low RMSD and RMSF values (Table S1, Figures S2, S3, and S4), similar to DON, indicating strong binding. Notably, U2873 formed a hydrogen bond with the epoxide ring of verrucarin A, with an occupancy of 93.7% (Figures S6 and S9A), highlighting its critical role in ligand stabilization. Further analysis of per-residue interaction energy contributions identified key residues involved in ligand recognition and stabilization (Figure ). The residue U2873 exhibited one of the key interactions (Table S2) with verrucarin A, showing a significant binding energy contribution of −17.2 kcal/mol. This interaction was primarily driven by electrostatic forces (−10.4 kcal/mol) and supported by van der Waals interactions (−6.8 kcal/mol).

3.

Upper panel illustrates the chemical structure of verrucarin A. The lower panel shows the binding pose of verrucarin A at site A of the 60S ribosome subunit, obtained from MD simulation. (A) Key residues interacting with verrucarin A are shown in stick representation, with their thickness and color indicating the relative strength of van der Waals and electrostatic interactions, respectively. The corresponding interaction energy values are provided in Table S2. (B) The hydrogen-bond network and interaction patterns with key residues in the binding site. The Mg²⁺ ion (colored magenta) does not directly interact with the ligand. However, the Mg²⁺ ion forms electrostatic interactions with one of the key residues G2403 which directly interacts with verrucarin A. The occupancy of hydrogen bonds is detailed in Figure S6.

Additional residues with significant interaction energy contribution, including U2875 (−11.2 kcal/mol), G2874 (−13.4 kcal/mol), and G2403 (−12.8 kcal/mol), also played a crucial role, indicating strong stabilization effects. Likewise, residues A2872, C2821, A2820, and G2400 exhibited moderate interactions (Table S2). Structurally, verrucarin A shares a core scaffold with DON but features an additional macrocyclic ring containing oxygen atoms. This ring, through its oxygen atoms, forms hydrogen bonds with the nitrogen atoms of the purine ring in residue G2403 (Figures and S6), contributing significantly to the ligand’s stability in the binding pocket. Furthermore, this purine ring engages in hydrophobic interactions with the macrocyclic ring of verrucarin A, further enhancing ligand’s stability. These interactions are facilitated by van der Waals forces between the nonpolar regions of the macrocyclic ring and the aromatic side chain of the residue. Notably, in the MD simulation of the complex with DON (Figure ), residue G2403 exhibits higher repulsion. In contrast, in the simulation of the complex with verrucarin A, this residue establishes stable interactions through persistent hydrogen bonding (Figure S6). In addition, residue G2400 also contributes a hydrogen bond, further stabilizing the placement of verrucarin A within the binding pocket. These unique interactions, enabled by the macrocyclic ring, likely account for the enhanced potency of verrucarin A relative to DON. In addition to these key residues, the Mg²⁺ ion plays an important role in ligand binding by interacting through strong coordination with the important residue G2403 in the binding pocket (Figure B). Mg²⁺ forms a strong coordination interaction (Figure S9B) with the O6 atom of the key residue G2403, dominated by electrostatics and donation of guanine lone pairs into Mg²⁺ orbitals, as revealed by F/I-SAPT analysis (Table ). Through this interaction, Mg²⁺ stabilizes G2403 in a favorable orientation, enabling the residue to form a key hydrogen bond with verrucarin A, as indicated by per-residue interaction energies from MD simulations and F/I-SAPT (Tables and S2). This indirect interaction underscores the importance of the Mg²⁺ ion in the overall binding mechanism. Compared to DON, the macrocyclic ring in verrucarin A promotes additional interactions, enhancing its binding affinity and more effectively inhibiting protein translation, as observed experimentally. These features likely contribute to verrucarin A being a more potent compound at the ribosomal binding site. Experimentally, verrucarin A exhibited markedly higher cytotoxicity and ribotoxicity than DON (Figures S1 and ), with an IC₅₀ in the low nanomolar range and strong inhibition of protein synthesis at similarly low concentrations. These findings are consistent with the enhanced binding stability and stronger interactions observed in MD simulations. Overall, these studies provide deeper insight into the molecular mechanisms of verrucarin A binding within the 60S ribosomal subunit and offer a comparative perspective with other inhibitors of eukaryotic protein synthesis. Additionally, this study sheds light on key structural interactions that contribute to their inhibitory effects, paving the way for potential therapeutic interventions targeting ribosomal function.

Comparative Ribotoxicity of DON and Its Derivatives

In order to correlate the results of MD simulations and quantum chemical calculations (SAPT) of the different compounds with their ribotoxicity, protein translation inhibition was evaluated through puromycin incorporation into the elongating peptides (Figure ). After 2 h of exposure, DON inhibits protein translation in a dose-dependent manner, with an IC₅₀ of 11.99 μM. As expected from the MD simulation data, verrucarin A exhibited a more potent ribotoxic effect, with an IC₅₀ of 4.59 nM. For DOM-1 and 3-epi-DON, no obvious protein translation defect could be quantified, despite a very small decrease of puromycin signal at 40 μM of DOM-1. The data indicate that the ribotoxic potential correlates with the stability of the ligands and their interaction patterns within the binding site of the 60S ribosomal subunit of the 80S eukaryotic ribosome, as predicted by MD simulations and quantum chemical calculations.

Discussion

The findings from this study provide crucial insights into the molecular mechanisms underlying the differential toxicity between DON and its metabolites, DOM-1 and 3-epi-DON, in relation to their binding interactions with the eukaryotic ribosome. The MD simulations and quantum chemical calculations collectively demonstrate that the ribotoxicity of DON is primarily driven by its strong and persistent interactions with the Mg²⁺ ion and key residues within the ribosomal A site. Importantly, the conserved Mg²⁺ ion emerges from this study as a key determinant of DON’s higher ribotoxicity compared to its metabolites, as its coordination strongly stabilizes the toxin within the ribosomal binding pocket. In previous studies, both diffuse and site-specific Mg²⁺ binding have been described as key contributors to RNA folding, structural stabilization, and ligand recognition. − For instance, in riboswitches, Mg²⁺ ions have been reported to directly mediate ligand binding and conformational transitions. In the ribosomal complex with DON, the Mg²⁺-mediated strong interactions appear to be critical for the ribosome’s catalytic activity. In contrast, the reduced toxicity of DOM-1 and 3-epi-DON is attributed to their inability to maintain these critical interactions, leading to their dissociation from the binding site over time. The MD simulations revealed that DON exhibits stable binding throughout the 500 ns simulation, facilitated by persistent hydrogen bonding and electrostatic interactions with U2873 and other key residues. The presence of the polar epoxide group in DON enhances the ligand’s stability within the hydrophilic region of the binding site, playing a crucial role in maintaining its position within the ribosomal binding pocket. This allows DON to adopt a distinct binding conformation that facilitates efficient interactions (through O1 and the hydroxy group at position 3 in the trichothecene core) with the Mg²⁺ ion. In metabolites of DON, the lack of the epoxide moiety or a switched orientation of hydroxyl group at position 3, fail to achieve the same binding conformation, limiting their interaction efficiency. Due to the absence of these key structural features, DOM-1 and 3-epi-DON fail to establish strong interactions with U2873 and the Mg²⁺ ion. Furthermore, quantum chemical calculations using the SAPT approach provided a detailed decomposition of the ligand–receptor interaction energies, highlighting the dominant role of electrostatic and induction forces in stabilizing DON within the ribosomal A site. Notably, the interaction energy contributions from the Mg²⁺ ion were significantly higher for DON compared to its metabolites, further supporting the hypothesis that the differential toxicity of these compounds is driven by their ability to form and sustain key molecular interactions. The significantly higher exchange repulsion in DOM-1 and 3-epi-DON suggests that these metabolites fail to orient the ligand into an efficient binding conformation. This suboptimal positioning increases steric hindrance within the binding pocket, further destabilizing their interactions and ultimately causing these metabolites to exit the binding pocket.

To further evaluate interaction patterns and ligand stability, the study included a reference complex with verrucarin A, a highly toxic trichothecene. While binding the same ribosomal site as DON, it differs structurally and exhibits greater toxicity. Unlike DON metabolites, this compound remained firmly bound within the binding pocket during MD simulation, as it forms strong interactions with the predicted key residues. Moreover, the conserved Mg²⁺ ion also appears to play a crucial role in the binding. The compound demonstrated stable binding interactions similar to DON, reinforcing the idea that ribotoxicity is closely linked to specific interaction patterns within the ribosomal A site. The strong and persistent interactions observed in DON and verrucarin A contrast with the transient binding observed in the nontoxic DON metabolites, emphasizing the critical role of maintaining stable molecular interactions for ribotoxic effects. This mechanistic insight aligns with our experimental findings from ribotoxicity and cell viability assays, where the metabolites caused no cellular stress compared to DON and verrucarin A. Notably, the magnitude of difference between DON and verrucarin A ribotoxicity was similar to the difference between their cytotoxic potential.

Conclusion

The interactions of the mycotoxins, DON, verrucarin A, and the nontoxic metabolites DOM-1 and 3-epi-DON with the site A of the 60S ribosomal subunit were investigated using molecular dynamics studies. The results reveal that the mycotoxins produced stable and strong interactions with the ribosomal site, unlike metabolites, which are less strongly anchored into the catalytic site.

A further energy decomposition analysis using the quantum mechanics FI/SAPT method provided insight into the contributing components to the interaction energy. The calculations emphasize the crucial role of key nucleotides, in particular U2873, as well as the importance of Mg²⁺ for understanding toxicity of DON and verrucarin A. The consistent correlation between interaction stability and known toxicity levels highlights the potential of this structure–function analysis as a predictive tool for assessing the ribotoxic potential of trichothecenes and related compounds. Such an approach could be used to screen novel or modified compounds for ribosome-binding capacity and likely toxicity, aiding in early risk assessment and ligand design. Moreover, this study highlights the potential for designing small molecules that could competitively inhibit DON binding by targeting the same ribosomal site, thereby offering a novel approach for therapeutic intervention against trichothecene-induced toxicity. Overall, this study provides a comprehensive molecular-level understanding of the differential toxicity of DON and its metabolites. The combination of MD simulations and quantum chemical calculations underscores the importance of key molecular interactions in dictating ligand stability and toxicity. Future research exploring potential detoxification strategies could leverage these interaction patterns to develop effective approaches for mitigating the effects of mycotoxins or other toxic compounds binding to the same site.

Methods

Most of the calculations performed in this study were carried out on the “Olympe” supercomputing center, at the University of Toulouse (France) (https://www.calmip.univ-toulouse.fr/les-services/hpc/caracteristiques-techniques-dolympe ) consisting of 374 computing nodes, each with two 18-core processors rated at 2.3 GHz.

Molecular Docking

The molecular docking studies were conducted using the GOLD (Genetic Optimization for Ligand Docking) software. Atomic coordinates from the crystal structure of the yeast 80S ribosome complexed with DON (PDB code 4U53) were obtained from the Protein Data Bank. , This complex consists of a mixture of RNA, peptide chains, Mg²⁺ ions, and the bound ligand, DON. Additionally, the Mg²⁺ ion (MG 3831), which resides in the binding pocket and plays a role in bridging interactions with the ligand and RNA residues G2816 and G2403, was retained as part of the receptor for docking simulation. To ensure accurate docking, hydrogen atoms, which were missing in the original crystal structure, were added, and the terminal C and N residues of the protein were capped with NMA (N-acetyl) and ACE (acetyl) groups, respectively. The ligands-DON, DOM-1, and 3-epi-DON were then prepared to ensure the correct protonation state at physiological pH 7.4, which is essential for accurate docking predictions. The docking procedure was first performed using DON to validate that the approach could correctly predict the binding pose, which was then compared to the X-ray crystal structure of the complex. Once the method’s accuracy was confirmed, the same docking protocol was applied to DOM-1 and 3-epi-DON, allowing for comparisons of their binding affinities and modes of interaction. The GOLD docking approach employs the ChemPLP scoring function alongside a genetic algorithm as the search strategy to optimize the binding orientation and energy of the ligand, enabling reliable predictions of ligand–receptor interactions.

Molecular Dynamics Simulation

The initial structures for DON (PDB code: 4U53) in complex with 80S ribosome were retrieved from the Protein Data Bank (PDB). Due to the large size of the overall complex, protein and RNA only residues within a 35 Å radius of the ligands were selected (Figure S11). Many other studies have successfully employed reduced models, in which only residues located within 20–35 Å of the catalytic site are retained. − Ligand parameters were generated using the General Amber Force Field (GAFF2) with AM1-BCC charges, employing the Antechamber tool available in the CHARMM-GUI , input generator module. The systems included K⁺ and Cl^– ions at a concentration of 0.15 M. All the input files for simulations employing AMBER , were prepared using CHARMM-GUI. All molecular dynamics simulations were conducted using Amber22 with the ff19SB force field for proteins, OL3 for RNA, and GLYCAM_06 for sugars. Water molecules were modeled using the TIP3P water model and the system was solvated in a 100 ųcubic simulation box with periodic boundary conditions. Mg²⁺, K⁺, and Cl^–ions were modeled using Amber’s default 12–6 Lennard-Jones (LJ) parameters, ensuring compatibility with the TIP3P water model. After solvation and addition of salt ions, the ribosome–DON complex comprised approximately 91,000 atoms: 9,460 from the macromolecules (RNA and amino acids), 130 K⁺, 77 Cl^–, and 88 Mg²⁺ ions, and about 27,100 water molecules. Initial energy minimization was carried out in two stages: 4000 steps using the steepest descent method followed by 6000 steps using conjugate gradient minimization. The system was gradually heated to 310 K in the NVT ensemble over 100 ps, employing a Langevin thermostat with a friction coefficient of 2.0 ps^–1 for equilibration and production phases. The equilibration process was conducted with pressure control (NPT ensemble) and used a 1 fs time step. It consisted of eight sequential steps where positional restraints on nucleotides, proteins, and ligands were gradually released from 5 to 0.25 kcal·mol^–1·Å^–2. The final equilibration phase involved a brief 1 ns simulation (NPT ensemble, 2 fs time step). During this, the constraints (10 kcal·mol^–1·Å^–2) were applied only to the backbone phosphorus and adjacent oxygen atoms of nucleotides, as well as to the α carbon and nitrogen atom of the backbone in protein residues beyond 25 Å from the ligand to maintain geometry during production. This initial step was discarded from subsequent analyses. Both equilibration and production simulations were performed under isotropic pressure control using a Berendsen barostat. A cutoff distance of 9 Å was applied for nonbonded interactions, including van der Waals and electrostatics. For each trajectory, frames were saved every 100 ps, generating a total of 5000 frames over 500 ns of simulation. Three independent replicates were generated for each system. Each production run began from the same equilibrated structure but used different random seeds for the Langevin thermostat, ensuring statistical independence without altering initial velocities or coordinates. All replicates were included in the trajectory analyses.

Trajectory Analysis

Trajectory analyses included root-mean-square deviation (RMSD), root-mean-square fluctuation (RMSF), and hydrogen-bond occupancy all of which were performed using CPPTRAJ from the AmberTools22 suite. Hydrogen-bond analysis was carried out using a donor–acceptor distance cutoff of 3.5 Å and a minimum angle of 130°. The interaction energies between the nucleotide residues and the ligand were computed using 2500 frames from the trajectory with the namdenergy.tcl script (v1.4) from Nanoscale Molecular Dynamics (NAMD). For analysis, residues within 10 Å of the ligand were selected to capture key binding interactions. The ligand–residue energies were taken as an average number from three replicates. The MD trajectory clustering was performed using the k-means algorithm implemented in CPPTRAJ, after removing water and ions. Clustering was based on the ligand RMSD, generating a representative structure for each cluster. The representative structure from the largest cluster was used for subsequent analysis and SAPT − calculations.

Quantum Chemical Calculation

SAPT calculations were performed on the most representative complexes of DON and verrucarin A obtained after clustering the trajectory using CPPTRAJ. For 3-epi-DON and DOM-1, which were unstable during the MD simulations, the docking complexesincluding all relevant RNA bases and the nearby Mg²⁺ ion at the binding sitewere used for the calculations. The residues at terminals were capped with hydroxy and methyl groups. The system up to 359 atoms was included in the SAPT calculations. To prepare the system, the hydrogen atoms were added using Biovia Discovery studio 2024 (Dassault Systèmes, 10 rue Marcel Dassault, CS 40501, F-78946 Vélizy-Villacoublay Cedex) and the structural optimization of the ligand–receptor complexes was performed in two stages using ORCA5.0. , Initially, hydrogen atoms were optimized using Grimme’s 3-corrected Hartree–Fock (HF-3c) method while keeping other atoms fixed. Subsequently, the entire complex was optimized, excluding the phosphorus atoms, the two adjacent oxygens, and the C1′ atom of the RNA backbone sugar to preserve the structural integrity. Grimme’s HF-3c method employs the minimal-basis set “MINIX”, which is designed to balance computational efficiency with reasonable accuracy, making it suitable for addressing systematic errors in small-basis Hartree–Fock calculations. Energy minimization is deemed converged, and the optimization process is halted when the maximum energy gradient and the root-mean-square (RMS) gradient fall below 0.0003 and 0.0001, respectively. The resulting coordinates (Figure S12) were used to perform the SAPT calculations. The side chains of RNA bases far from the binding site were removed and capped with hydrogen atoms and methyl group before performing the SAPT calculations. The calculations were conducted utilizing the jun-cc-pVDZ basis set at the density-fitted Hartree–Fock (DF-HF) level, with PSI4 software (version 1.9.1).

Cell Culture and Treatments

Human colon cancer cells HCT116 were maintained in McCoy’s 5A (modified) medium (Gibco, Life technologies) supplemented with 10% heat-inactivated calf serum and 0.5 mg/mL penicillin/streptomycin (P/S). Cells were grown at 37 °C in a humidified atmosphere containing 5% CO₂ and subcultured every 2–3 days. DON (prepared as a 5 mM solution in water) was purchased from Sigma-Aldrich. Deepoxy-deoxynivalenol (DOM-1, prepared as a 5 mM solution in DMSO) was purchased from Sigma-Aldrich and prepared as previously described. Verrucarin A (prepared as a 10 mM solution in DMSO) was purchased from Biorbyt. 3-epi-Deepoxy-DON (3-epi-DON, prepared as a 10 mM solution in DMSO) was produced as previously described.

Cell Viability Assay

HCT116 cells were seeded in triplicate at a density of 10,000 cells per well in 96-well plates. One day after seeding, cells were treated during 72 h. Viability was assessed using the CellTiter-Glo Luminescent Cell Viability Assay (Promega) according to the manufacturer’s instructions. IC₅₀ values were calculated with Quest Graph IC50 Calculator, AAT Bioquest (https://www.aatbio.com/tools/ic50-calculator, accessed on 28 May 2025).

Protein Synthesis Analysis

HCT116 cells were seeded at a density of 50,000 cells per well in 96-well plates. One day after seeding, cells were treated during 2 h, and puromycin was added for the last 30 min. The measurement of protein synthesis by puromycin labeling was performed as described previously. Puromycin was immunodetected by In-Cell-Western using an antipuromycin antibody (clone 12D10 diluted 1:5000; Millipore, Molsheim, France). GAPDH was used as a control and immunodetected with an anti-GAPDH antibody (ABS16 diluted 1:5000, Millipore). Secondary antibodies were diluted 1:5000 (IRDye 800CW; Rockland, and IRDye 680RD Licor). The puromycin signal was first normalized with the GAPDH signal and then with the fluorescence of puromycin-labeled control cells. IC₅₀ values were calculated with Quest Graph IC50 Calculator, AAT Bioquest (https://www.aatbio.com/tools/ic50-calculator, accessed on 15 October 2025).

AmberTools and Amber22 (https://ambermd.org) are available under license. ORCA 5.0 (https://www.faccts.de/orca/), VMD 1.9.4 (https://www.ks.uiuc.edu/Research/vmd/), NAMD (https://www.ks.uiuc.edu/Research/namd/), and PyMOL (https://github.com/schrodinger/pymol-open-source) are available free of charge for academic users. CHARMM-GUI (https://www.charmm-gui.org/) is an accessible Web-based platform to interactively build complex systems. Psi4 (https://github.com/psi4/psi4) is an open-source code. Biovia Discovery Studio v22.1 (https://www.3ds.com/fr/products/biovia/discovery-studio) is a product of Dassault Systèmes and is commercially available. Discovery Studio Visualizer, a molecular graphics environment (https://discover.3ds.com/discovery-studio-visualizer-download) is available free of charge.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.5c09756.

• Dose-dependent cytotoxicity data for DON, DOM-1, 3-epi-DON, and verrucarin A; RMSD and RMSF analyses of ligand–rRNA complexes; hydrogen-bond occupancy profiles; time evolution of key atomic distances; superimposition of MD-derived and crystallographic binding poses; molecular electrostatic potential maps; models of the rRNA A site used for MD simulations; quantum-mechanical models used in F/I-SAPT calculations; and interaction energy decompositions for the rRNA–ligand complexes (PDF)

• Coordinates, topology, and parameter files used for the equilibration and production MD simulations of DON, 3-epi-DON, DOM-1, and verrucarin A (ZIP)

• Input files, coordinates, and output files for the F/I-SAPT calculations performed on the complexes of the same set of compounds (ZIP)

#.

Sir William Dunn School of Pathology, University of Oxford, South Parks Road, Oxford, OX1 3RE, United Kingdom

This work was supported by the French National Research Agency (ANR) under the GenoMyc project (ANR-22-CE34-0022).

The authors declare no competing financial interest.