The research team, publishing an entirely in silico investigation, constructed a series of acridine-based molecules in which electron-donating and electron-withdrawing substituents were systematically varied on the terminal phenyl ring of the 9-anilinoacridine scaffold. The acridine nucleus is a planar, tricyclic aromatic system long prized in medicinal chemistry for its capacity to engage in π–π stacking, hydrophobic contacts and DNA intercalation, and several acridine derivatives have previously shown anticancer activity through topoisomerase inhibition, disruption of DNA replication and apoptosis induction. What remained poorly explored, the authors argue, was how novel substitutions on this scaffold would shape predicted interactions with ERα.

To answer that question computationally, the researchers anchored their workflow to a high-resolution X-ray crystal structure of the human ERα ligand-binding domain, PDB ID 2IOG, determined at 1.60 Å and co-crystallized with the ligand IOG in an antagonist-associated receptor conformation. Chain A, spanning residues 309 to 554, was prepared in Schrödinger’s Protein Preparation Wizard: crystallographic waters beyond 5 Å of the active site were removed, bond orders and protonation states were assigned at physiological pH, missing loops were rebuilt with Prime, and the structure was restrained-minimized with the OPLS4 force field to a heavy-atom RMSD cutoff of 0.30 Å. Three C-terminal residues and part of the Arg335 side chain were unresolved in the deposited structure and were repaired during preparation.

All 24 derivatives — designated II(a) through II(x) — were sketched in ChemDraw and converted to three-dimensional conformers with LigPrep, generating relevant protonation and tautomeric states at pH 7.0 ± 0.2. Docking was performed with Glide in extra-precision (XP) mode, using a receptor grid defined by the co-crystallized ligand, which was removed before docking. The protocol was validated by redocking IOG into the binding site: the superposition of experimental and redocked poses yielded a heavy-atom RMSD of just 0.2086 Å, supporting the reliability of the setup. Crucially, the authors stress that docking scores were used only for comparative prioritization and were not interpreted as experimental binding affinities.

The docking results singled out two derivatives. Compound II(x), a phenylhydrazone-substituted analogue, achieved the most favourable Glide XP Gscore of −11.85 kcal/mol and the best docking score of the series at −9.80 kcal/mol, while compound II(e), bearing a 2-nitrophenyl substituent, followed closely with a Gscore of −11.78 kcal/mol and a docking score of −9.32 kcal/mol. Interaction analysis revealed that II(x) formed a hydrogen bond with ASP351 and significant π–π stacking with TRP383, supplemented by hydrophobic contacts with LEU346, LEU349, MET343, MET421, LEU525 and TYR526. Derivatives II(n), II(w), II(i) and II(j) also scored favourably, whereas II(o), II(l) and II(s) were the weakest performers, with Gscores between −5.28 and −5.93 kcal/mol.

Because docking scores alone can mislead, the team ran Prime MM-GBSA calculations on the minimized docked complexes using the VSGB 2.0 solvation model. Here the ranking shifted in an instructive way: compound II(f), an ortho-aminophenyl imine derivative with only a moderate Gscore of −8.71 kcal/mol, produced the most favourable calculated binding free energy of the entire library at −99.68 kcal/mol, ahead of II(q) at −88.69 and II(c) at −79.68 kcal/mol. Energy decomposition attributed II(f)’s advantage to a strong van der Waals contribution of −63.14 kcal/mol, a lipophilic term of −40.56 kcal/mol, Coulombic energy of −40.76 kcal/mol and a hydrogen-bonding contribution of −3.43 kcal/mol. By contrast, II(e) and II(x) yielded markedly less favourable MM-GBSA estimates of −40.91 and −33.50 kcal/mol respectively — a divergence the authors interpret as evidence that docking scores and post-docking energetics capture different aspects of ligand–receptor recognition.

The pharmacokinetic picture, predicted with the QikProp module, was mixed. Molecular weights ranged from 325.41 to 492.49 g/mol, and predicted octanol/water partition coefficients (QPlogPo/w) spanned 3.46 to 7.30, indicating moderate-to-high lipophilicity across the series. Compound II(f) showed a molecular weight of 402.49 g/mol, a dipole moment of 3.00 D, hydrogen-bond donor/acceptor values of 2.5/3.5, a QPlogPo/w of 5.84 and one predicted Rule-of-Five violation. Most derivatives showed zero or one violation, but several displayed high lipophilicity and poor predicted aqueous solubility, with QPlogS values ranging from −4.67 to −8.48. Blood–brain barrier penetration was generally limited, and ProTox 3.0 predictions placed the compounds in toxicity classes 3 to 5, with predicted LD50 values of 160 to 2690 mg/kg; II(v) and II(w) showed the lowest predicted LD50 values, while hepatotoxicity and respiratory-toxicity alerts appeared for several derivatives.

The three top candidates — II(e) and II(x) for their docking scores, II(f) for its MM-GBSA energetics — were then subjected to 100 ns molecular dynamics simulations in Desmond, solvated in TIP3P water with 0.15 M salt and neutralizing counter ions. All four systems, including the co-crystallized ligand reference, showed an initial conformational adjustment followed by stable trajectories: protein RMSD settled around 1.8–2.2 Å for II(e), 2.0–2.4 Å for II(f), 2.5–2.9 Å for II(x) and 2.4–2.8 Å for the reference ligand in the latter half of the runs. Root-mean-square fluctuation analysis showed that the binding-site residues remained comparatively rigid, with the largest fluctuations confined to terminal and loop regions. Compound II(f) underwent a pronounced transient ligand rearrangement between roughly 20 and 45 ns, with ligand RMSD peaking near 4 Å before settling, yet it maintained recurrent contacts with Leu346, Ala350, Thr383, Leu387, Phe404 and Leu536, alongside intermittent hydrogen-bonding and water-mediated interactions, including recurring contacts with Glu353 and Asp351 across trajectory snapshots.

The authors are careful — unusually so for this genre — about what the results do and do not establish. The simulations comprised a single 100 ns trajectory per complex, so trajectory reproducibility remains untested. Docking was performed against a single antagonist-associated receptor conformation and therefore does not encompass the full conformational ensemble of ERα. No compound was synthesized, characterized or biologically tested: there are no NMR, mass spectrometric or purity data, no cytotoxicity assays, no receptor-binding or transactivation experiments, and no evidence of ERα antagonism or downstream signalling effects on the PI3K/AKT or MAPK pathways, which are discussed only as hypothetical consequences requiring validation. The authors also flag chemical-stability concerns, noting that II(f) and II(k) contain ortho-amino and ortho-hydroxy groups adjacent to imine functionalities that could permit hydrolysis, tautomerization or cyclization under experimental conditions.

Within those limits, the study delivers a concrete, hypothesis-generating shortlist. Compound II(f) emerges as the lead computational candidate on the strength of its favourable MM-GBSA energetics and persistent, dynamically rearranging contacts within the ERα pocket, flanked by II(e) and II(x). The next steps the authors propose are unambiguous: synthesize and fully characterize the prioritized structures, assess their solution-phase stability, then test ERα binding and functional activity, antiproliferative effects, selectivity and pharmacokinetic and toxicity profiles experimentally. Until that work is done, the computational evidence stands as a prioritization framework — a structural argument that these acridine derivatives can plausibly occupy the ERα ligand-binding cavity, awaiting the laboratory verdict on whether that occupancy translates into biology.

Subject of Research: Structure-based computational design and evaluation of 9-anilinoacridine derivatives targeting estrogen receptor-alpha in breast cancer

Article Title: Structure-based design and molecular modelling of 9-anilinoacridine derivatives targeting estrogen receptor-α in breast cancer

Article References: Rajagopal, K., Shanmugam, K., Raman, K., Uddandrao, V. V. S., Thakran, A., Dhevikarunaa, G. M., & Luseena, P. (2026). Structure-based design and molecular modelling of 9-anilinoacridine derivatives targeting estrogen receptor-α in breast cancer. Results in Chemistry, 31, Article 103975. https://doi.org/10.1016/j.rechem.2026.103975

Image Credits: AI Generated

DOI: 10.1016/j.rechem.2026.103975

Keywords: breast cancer, estrogen receptor-alpha, 9-anilinoacridine, molecular docking, MM-GBSA, molecular dynamics simulation, ADMET prediction, drug design, virtual screening, endocrine resistance, computational chemistry, Schrödinger Glide

The choice of PTP1B as the target is grounded in a growing body of evidence linking this enzyme to the neuroinflammatory cascade that drives Alzheimer’s pathology. PTP1B, also known as PTPN1, is best understood as a negative regulator of insulin and leptin signaling—it dephosphorylates insulin receptor substrates and dampens the JAK2 pathway. In the brain, however, its role extends into far more dangerous territory. In activated microglia, the brain’s resident immune cells, elevated PTP1B expression stimulates Src kinase activation by dephosphorylating its inhibitory tyrosine residue, which potentiates NF-κB signaling and amplifies the release of pro-inflammatory mediators including TNF-α, IL-1β, IL-6, iNOS, and COX-2. Pharmacological inhibition of PTP1B has been shown to diminish microglial activation and neuroinflammatory responses in both cell-based studies and animal models, suggesting that the enzyme actively contributes to disease progression through inflammatory mechanisms.

What makes PTP1B particularly attractive as a multi-target strategy is that it sits at the intersection of several intertwined pathological processes. Beyond its role in neuroinflammation, PTP1B directly regulates leptin receptors in hippocampal neurons and dephosphorylates tropomyosin receptor kinase B (TrkB), thereby downregulating brain-derived neurotrophic factor (BDNF) signaling—a pathway critical for synaptic plasticity and neuronal survival. It also acts as a negative regulator of IRS-1 and receptor tyrosine kinase signaling, contributing to the insulin resistance that is increasingly recognized as a risk factor for Alzheimer’s disease. Systemic metabolic conditions such as obesity and type-2 diabetes, both well-established risk factors for dementia, are linked with chronic neuroinflammation, mitochondrial impairment, and insulin resistance—all parameters that PTP1B inhibition could theoretically normalize. By contrast, the researchers argue, single-pathway approaches like BACE1 inhibitors, which failed in clinical trials due to psychiatric effects and hepatic toxicity, or GSK-3β inhibition, which carries a narrow therapeutic window, have struggled to translate into clinical success.

The simufilam scaffold itself carries an unusual history. Originally developed to restore the native conformation of filamin A, a protein that becomes abnormally connected to α7 nicotinic acetylcholine receptors in Alzheimer’s disease, simufilam progressed through Phase I and Phase II trials with favorable safety and pharmacokinetic profiles, demonstrating restoration of CSF and blood biomarkers. But its Phase III evaluation was ultimately discontinued due to insufficient efficacy. The researchers are careful to note that this failure does not necessarily invalidate the underlying scaffold—the lack of clinical benefit may reflect bioavailability limitations, off-target effects, or trial-related challenges rather than a fundamental flaw in the molecule. Given simufilam’s demonstrated brain penetration, compact spirocyclic geometry, and chemically tractable 1,3,8-triazaspiro[4.5]decan-4-one core, the team hypothesized that rational structural modification could redirect its neuroprotective activity toward PTP1B allosteric inhibition.

The design strategy was methodical. The team generated 52 simufilam derivatives by introducing diverse substituents at two positions—the R1 site on the piperidine nitrogen and the R2 site on the imidazolidinone core—while preserving the parent scaffold. The R1 modifications were engineered to reduce amine basicity, tune lipophilicity to maintain log D between 2.0 and 3.5, and sterically protect the nitrogen against cytochrome P450-mediated N-dealkylation, addressing the well-known problem that highly basic tertiary amines restrict passive blood-brain barrier permeability. The R2 derivatization was designed to probe an adjacent, under-utilized binding pocket, optimizing van der Waals contacts and hydrogen bonding capabilities to enhance selectivity over off-target receptors. The N-benzyl group at position 1 was maintained as a lipophilic anchor to preserve baseline π-π hydrophobic interactions within the binding pocket.

Before any docking was attempted, the researchers employed network pharmacology to confirm that PTP1B was the right target. They predicted compound targets using SwissTargetPrediction, retrieved Alzheimer’s disease-related targets from DisGeNET, GeneCards, and GeDiPNet, and constructed a protein-protein interaction network of 128 common targets using the STRING database and Cytoscape software. The resulting network contained 128 nodes and 1448 edges, with an average node degree of 22.6. When ranked by degree, betweenness, and closeness centrality, PTP1B emerged as the top-ranked target with a degree of 92, betweenness of 1969.41, and closeness of 0.40—outperforming other key nodes including SRC, BCL2, AKT1, and CASP3. Gene ontology enrichment analysis across biological process, molecular function, and cellular component categories, along with Reactome pathway mapping, further confirmed that the 128 active targets converge on receptor tyrosine kinase signaling, PI3K/Akt regulation, and mitogen-activated protein kinase pathways—all central to Alzheimer’s pathogenesis.

Molecular docking was performed against the non-competitive allosteric site of PTP1B using the crystal structure PDB ID 1T49, which captures the enzyme in complex with a known allosteric inhibitor. The docking protocol was validated by re-docking the co-crystallized ligand, yielding an RMSD of 0.8 Å from the experimental pose and reproducing all key native non-covalent interactions as confirmed by the Protein-Ligand Interaction Profiler webserver. Among the 52 derivatives, A20 and A26 emerged as the top-scoring leads with predicted binding energies of −10.8 and −9.5 kcal/mol, respectively. The significantly lower energy of A20 indicates a more stable protein-ligand complex. Critically, both compounds engaged residues across the catalytic and adjacent non-competitive allosteric sites, satisfying a multi-tiered consensus strategy designed to distinguish true active leads from non-specific binders.

The mechanistic insight underlying A20’s inhibitory activity is elegant. PTP1B catalysis requires dynamic motion of the flexible WPD loop, which must transition from an open state to a closed conformation to correctly position catalytic Asp181 relative to the active-site P-loop for substrate cleavage. A20 occupies a hydrophobic C-terminal non-competitive allosteric site approximately 20 Å from the active site, delineated by helices α3, α6, and α7. The o-chlorophenyl hydrazine tail at R1 establishes hydrophobic contacts and halogen interactions with Leu192, Ala189, Phe280, and Leu281, while the spiro-hydantoin core anchors the compound within the surrounding helical framework. Occupancy of this site prevents the C-terminal α7 helix from adopting the packed, ordered conformation required to facilitate WPD loop closure. By locking the WPD loop in its open, inactive state, A20 decouples substrate binding from enzymatic cleavage, reducing catalytic turnover. Because this allosteric site is distinct from the highly conserved, positively charged active site, targeting this region provides a structural basis for high isoform selectivity and improved cellular permeability.

Molecular dynamics simulations in GROMACS over 1000 nanoseconds confirmed the stability of the PTP1B-A20 complex, with consistent RMSD, RMSF, radius of gyration, hydrogen bond count, and solvent accessible surface area values throughout the trajectory. More than three hydrogen bonds remained stable over the full simulation, further supporting binding persistence. Principal component analysis revealed that A20 restricts conformational flexibility of the protein, yielding a more compact ensemble compared to the unbound state, which traverses a broader conformational landscape. MM-PBSA calculations over 1000 snapshots from the MD trajectory displayed stable binding free energies for A20, with favorable van der Waals and electrostatic contributions. Density functional theory calculations at the B3LYP/6-311+G level characterized A20’s electronic properties, revealing a dipole moment of 4.39 Debye—more polar than A26 at 3.74 Debye but less polar than the co-crystal reference at 11.54 Debye. Frontier molecular orbital analysis showed HOMO-LUMO gaps of approximately 5.4 to 5.9 eV across the key interacting residues Tyr46, Asp193, and Glu276, consistent with the electronic reactivity patterns observed in docking.

Natural bond orbital analysis further revealed that A20 is pre-organized into a compact, hydrogen bond-stabilized orientation through intramolecular O/N···H contacts at distances of 1.732 to 1.772 Å, with dominant donor-acceptor interactions involving n(O/N) to σ(N-H/C-H)* hyperconjugation and π-π* delocalization across carbonyl and aromatic fragments. These electronic features energetically favor a pre-organized conformation for site binding. ADMET profiling indicated that A20 satisfies molecular weight criteria at 370 Da, hydrogen bond donor criteria at 4 donors, and lipophilicity criteria with QPlogPo/w of 1.74, though it exhibits a single Lipinski violation from an elevated hydrogen bond acceptor count of 12.2. The researchers acknowledge that all predicted binding affinities, interaction patterns, and pharmacokinetic properties are based on theoretical models and may not fully reflect biological behavior under physiological conditions. Future development will require systematic lead optimization, including bio-isosteric modifications to improve metabolic half-life, formulation into PLGA nanoparticles to enhance aqueous solubility and CNS delivery, and in vitro PTP1B inhibition assays to confirm the computational findings. A20 is synthetically accessible through a concise two-step synthesis and targets the non-conserved α3/α6/α7 allosteric sites, providing a scaffold to guide selectivity over the homologous TCPTP enzyme in future experimental studies.

Subject of Research: Computational drug redesign of simufilam derivatives targeting PTP1B for Alzheimer's disease neuroprotection

Article Title: Network pharmacology–driven identification of PTP1β and rational design of simufilam-based allosteric inhibitors for neuroprotection in Alzheimer’s disease

Article References: Malakar, V., Raman, R., Justin, A., SP, D., Roy, D., Malakar, C. C., Purohit, D., & Mali, P. C. (2026). Network pharmacology–driven identification of PTP1β and rational design of simufilam-based allosteric inhibitors for neuroprotection in Alzheimer’s disease. 3 Biotech, 16(11), Article 467. https://doi.org/10.1007/s13205-026-05086-2

Image Credits: AI Generated

DOI: 10.1007/s13205-026-05086-2

Keywords: Alzheimer's disease, PTP1B inhibition, simufilam derivatives, network pharmacology, molecular docking, molecular dynamics simulation, MM-PBSA, density functional theory, allosteric inhibitors, neuroinflammation, drug design, computational chemistry

The three derivatives, named AS1, AS4AE, and SA4AE, share a common 3,5-dimethoxybenzene scaffold but differ in the substituents attached at specific positions of the benzene ring. AS1 carries a para-hydroxyl group with an allyl ester, AS4AE features two allyl ester groups, and SA4AE bears a para-allyloxy group with a free carboxylic acid. The research, published in Discover Chemistry, represents the first study to combine density functional theory electronic characterization with molecular docking, molecular dynamics simulations, and Monte Carlo corrosion modelling to evaluate O-allylated syringic acid derivatives across both biological and anticorrosion domains within a single computational framework.

To understand the electronic behaviour of these molecules, the team performed geometry optimization using the B3LYP method with the 6-311G basis set, a well-established level of density functional theory. Frontier molecular orbital energies, global reactivity descriptors, and condensed Fukui indices were then extracted to characterize how electrons are distributed across each molecule and where chemical reactions are most likely to occur. The results revealed a fascinating hierarchy of reactivity. SA4AE emerged as the most reactive derivative with the smallest energy gap of 6.35 electronvolts, while AS1 showed the greatest tendency to donate electrons with the least negative HOMO energy of minus 7.07 electronvolts. The methoxy atoms and ring carbons were identified as the principal nucleophilic sites, the regions where electron density is most available for transfer to metal surfaces or biological receptors.

The biological target chosen for this investigation was the adenosine A1 receptor, a seven-transmembrane G-protein-coupled receptor expressed in the central nervous system, the cardiovascular system, and the kidney. Activation of this receptor regulates heart rate and atrioventricular conduction, contributes to neuroprotection and cardioprotection, and plays a role in sodium retention and reduced glomerular filtration rate. A1 receptor inhibitors are considered vital in the treatment of acute heart failure and asthma, and they may offer defensive benefits against kidney injury and dementia. The researchers retrieved the three-dimensional crystal structure of the receptor from the Protein Data Bank and prepared it for molecular docking simulations using AutoDock Vina.

The docking results were striking. AS4AE demonstrated the strongest receptor binding among the three derivatives with a docking score of minus 7.3 kilocalories per mole, matching that of adenosine itself under identical computational conditions. This suggests comparable thermodynamic favourability at the A1 receptor active site. The strong interaction was attributed to multiple hydrogen-bond interactions with the amino acid residues TYR12 and ASN254, at distances of 2.99 and 2.50 angstroms respectively, along with a carbon-hydrogen bond interaction with GLU172 at 3.13 angstroms. In comparison, SA4AE and AS1 formed fewer interactions and recorded lower binding affinities of minus 6.1 and minus 6.7 kilocalories per mole respectively.

To validate the stability of these receptor-ligand interactions over time, the team ran 100-nanosecond molecular dynamics simulations using the OpenMM engine within a cloud-based Google Colaboratory environment. The root mean square deviation values for both the adenosine and AS4AE complexes stabilized between 1.5 and 2.1 angstroms from 5 to 55 nanoseconds, indicating that the AS4AE derivative remains firmly anchored in the binding pocket for extended periods. While adenosine showed slightly greater overall stability after a transient destabilization event around 56 nanoseconds, the AS4AE complex maintained a stable interaction profile throughout the simulation, supported by a double hydrophobic interaction with the flexible residue LEU250.

Drug-likeness and pharmacokinetic profiling added another layer of confidence to the biomedical potential of these compounds. All three derivatives satisfied Lipinski’s rules with zero violations, showing molecular weights between 238.24 and 306.31, moderate lipophilicity values between 1.82 and 2.85, and low topological polar surface areas between 64.99 and 71.06 square angstroms. The boiled egg plot revealed strong human intestinal absorption and blood-brain barrier permeability for all three compounds, and none were predicted to be P-glycoprotein substrates. ADMET analysis confirmed the absence of hepatotoxicity, mutagenicity, and metabolic interference through CYP3A4 pathways, painting a clean toxicity profile across the series.

On the industrial side, the corrosion inhibition potential of these derivatives was assessed against the Fe(110) surface in an acidic environment using Monte Carlo adsorption and quench molecular dynamics simulations with the COMPASS II force field. The Fe(110) model surface was constructed by cleaving the iron bulk crystal along the 110 Miller plane and expanded into a 10 by 10 supercell with a 30-angstrom vacuum layer. The simulated acidic environment comprised 280 water molecules, 10 hydroxonium ions, and 10 chloride ions loaded into the solvent layer above the iron surface. AS4AE proved the strongest corrosion inhibitor with an adsorption energy of minus 5520 kilocalories per mole, forming a parallel orientation to the iron plane that displaces the corrosive medium and prevents access by corrosive agents.

The radial distribution function analysis revealed that all three inhibitors exhibit interaction peaks below 3.5 angstroms, suggesting that their binding to the metal surface occurs through chemisorption rather than simple physisorption. This means the molecules form genuine chemical bonds with the iron surface, creating a protective molecular film that blocks corrosive species from reaching reactive metal sites. The binding energy increased in the order of AS1, SA4AE, and AS4AE, with AS4AE achieving the highest binding energy at 70 picoseconds during the 500-picosecond simulation period, confirming its superior capacity to anchor itself to the metal surface in acidic conditions.

The dual promise of these O-allylated syringic acid derivatives is remarkable. The same structural features that favour biological receptor-ligand interaction, namely aromatic rings with heteroatom functionalities and pi-conjugation, also govern strong adsorption on metal surfaces. AS4AE stands out as the lead candidate for both pharmaceutical and anticorrosion applications, matching adenosine’s binding affinity at the A1 receptor while simultaneously outperforming the other derivatives in surface adsorption parameters. The researchers acknowledge that these results have not been validated experimentally and that future work should include experimental in vitro and in vivo analyses, as well as electrochemical validation of the anticorrosion efficiency on mild steel in acidic and alkaline solutions. Nevertheless, this computational study establishes clear structure-activity relationships that could guide the design of next-generation multifunctional molecules derived from one of nature’s most accessible phenolic acids.

Subject of Research: Chemistry

Article Title: Corrosion and biological inhibition activity studies of O-allylated syringic acid derivatives via computational approaches

Article References: Odozi, N. W., Ojo, N. D., Bello, K. A., & Ibrahim, D. A. (2026). Corrosion and biological inhibition activity studies of O-allylated syringic acid derivatives via computational approaches. Discover Chemistry, 3(1), 582. https://doi.org/10.1007/s44371-026-01043-3

Image Credits: AI Generated

DOI: 10.1007/s44371-026-01043-3

Keywords: syringic acid, corrosion inhibition, density functional theory, molecular docking, molecular dynamics simulation, O-allylation, adenosine A1 receptor, drug-likeness, Monte Carlo simulation, Fe(110) surface, Corrosion, biological

The NMDA receptor is a heterotetramer, typically assembled from two GluN1 and two GluN2A subunits, and it demands two co-agonists before it will open: glycine binds GluN1, glutamate binds GluN2A, and the magnesium block must be relieved. Once activated, the receptor undergoes a cascade of conformational changes that dilate a central pore, allowing calcium and sodium to pour into the cell. Because mutations in this receptor are linked to Alzheimer’s disease, depression, stroke, epilepsy, and schizophrenia, understanding the open state is not merely an academic exercise; it is a prerequisite for rational drug design. Yet the open state is transient, heterogeneous, and notoriously difficult to capture experimentally, which is precisely why the Buffalo team, led by Wenjun Zheng with first author Xing Liu, turned to simulation.

The researchers built a truncated model of the GluN1/GluN2A receptor containing only the ligand-binding domains and the transmembrane domain, the two layers that carry the essential channel-gating machinery. They started from a crystal structure of an allosterically inhibited GluN1/GluN2B receptor, generated a homology model for the rat GluN1 and GluN2A sequences using SWISS-MODEL, embedded the truncated receptor in a POPC lipid bilayer with CHARMM-GUI, and then drove it toward an active-like conformation with targeted molecular dynamics, using an active-state ligand-binding domain structure derived from a cryo-EM map as the target. Once there, they held the ligand-binding domains in their active conformation with positional restraints and let the transmembrane domain move freely across seven independent 800-nanosecond trajectories at 30 degrees Celsius.

The resulting ensemble revealed something striking: the channel pore is not a static tunnel but a violently fluctuating one. Analysis with the HOLE program showed the minimal pore radius swinging between 0 and 2.8 angstroms across the simulation frames. Using established criteria for ion-channel conductance, the team estimated that 55 percent of frames were water permeable, 35 percent partially ion-conducting, and less than 1 percent fully open. The authors interpret these rapid pore-radius fluctuations as a dynamic feature of the transmembrane domain in the active-like state, distinct from the much slower transitions between fully open and closed functional states. In other words, even when the receptor is nominally active, its gate breathes.

Zooming in on individual residues, the team calculated root mean square fluctuations across the ensemble and found pronounced flexibility in the S1-M1 linker, the M1-M2 loop, the M2-M3 loop, the M3-S2 linker, the S2-M4 linker, and the C-terminus of the M4 helix. Crucially, the two subunit types behaved differently. GluN1 showed higher flexibility in the S1-M1, M3-S2, and S2-M4 linkers, while GluN2A was more mobile in M1 and the M2-M3 loop. At the two proposed gates, the upper gate at the bundle crossing of the M3 helices and the lower gate at the apex of the M2 pore loops, the fluctuations were moderate but subtly asymmetric between subunits, hinting that GluN1 and GluN2A play distinct roles in controlling channel opening and closing.

Principal component analysis of the combined trajectories identified ten dominant modes of collective motion that together account for 83 percent of the structural variation. Three of these modes correlated significantly with pore radius, and all three showed subunit-dependent amplitudes, with GluN2A moving more vigorously near the lower gate while the two subunits moved similarly near the upper gate. The analysis also highlighted the M1-M2 loop, a region so flexible that it is often missing from experimentally solved structures, as a hub of collective motion. These findings reinforce a growing consensus that the NMDA receptor’s active state is not a single conformation but an ensemble of rapidly interconverting sub-states.

To connect motion to mechanism, the researchers catalogued every hydrogen bond and van der Waals contact between residue pairs across all 5,600 simulation frames and computed Pearson correlations between each interaction and the minimal pore radius. Interactions with correlations above 0.3 or below negative 0.3 were flagged as potentially favoring opening or closing, respectively. The analysis yielded five hydrogen-bonded pairs that appear to stabilize the open pore, including a cross-subunit contact between GluN1 M4 and the GluN2A M2-M3 loop, and fifteen pairs that appear to favor closure, many clustered in the flexible linkers that transmit force from the ligand-binding domains down to the pore. In total, 35 key residues emerged from the hydrogen-bond analysis, 33 of them evolutionarily conserved and 11 known sites of disease mutations recorded in ClinVar and HGMD.

The van der Waals analysis painted an even denser picture, identifying 144 residues whose nonpolar contacts correlate with pore dynamics, 125 of them conserved and 49 linked to disease. Because so many transmembrane residues are conserved and therefore plausibly functional, the correlation-based approach produced a long list of candidates that was too diffuse to guide experiments efficiently. To sharpen the predictions, the team turned to machine learning. They trained a random forest regressor to predict the minimal pore radius from pairwise interaction energies and hydrogen-bond occupancies, then used SHAP values, an explainable-AI technique borrowed from cooperative game theory, to rank the importance of every interaction. The result was a much sparser list: just 32 residues in the top 10 percent of importance, 29 of them conserved and 19 carrying disease mutations, an enrichment from under 25 percent to roughly 60 percent compared with the correlation analysis.

Notably, twelve of those 32 machine-learning hits had not been flagged by the Pearson correlation analysis at all, suggesting that nonlinear relationships between residue contacts and pore geometry exist that simple linear statistics miss. The machine-learning model, however, came with an honest caveat: it learned a trajectory-specific relationship that failed to generalize across independently sampled trajectories, meaning the model captured local statistical structure rather than a universal physical law. The authors are candid about this limitation, as well as about the artifacts that truncating the amino-terminal domain and restraining the ligand-binding domains may introduce, and they call for larger-scale simulations of the full-length receptor as computational hardware improves.

The study’s most actionable contribution may be its bridge to medicine. By annotating predicted gating residues with evolutionary conservation grades and known pathogenic mutations, the work offers a concrete, testable hypothesis: that disease-associated variants in the S1-M1 linker, the M2-M3 loop, and the M4 helix disrupt channel gating by perturbing the transient network of polar and nonpolar interactions that stabilizes the open pore. A recent functional study independently showed that the S1-M1 linker critically controls channel opening, lending weight to this framework. The next step, the authors suggest, is to simulate specific disease mutations to determine whether they produce gain-of-function or loss-of-function phenotypes, turning a 5.6-microsecond computational experiment into a roadmap for understanding, and eventually treating, the neurological disorders that arise when this molecular gate malfunctions.

Subject of Research: Molecular dynamics simulation and machine learning analysis of the active-like state of the GluN1/GluN2A NMDA receptor channel

Article Title: Molecular dynamics simulation of the truncated NMDA receptor in a partially restrained active-like state

Article References: Liu, X., & Zheng, W. (2026). Molecular dynamics simulation of the truncated NMDA receptor in a partially restrained active-like state. Frontiers in Chemical Biology, 5, Article 1894245. https://doi.org/10.3389/fchbi.2026.1894245

Image Credits: AI Generated

DOI: 10.3389/fchbi.2026.1894245

Keywords: NMDA receptor, molecular dynamics, GluN1, GluN2A, ion channel gating, machine learning, random forest, SHAP, hydrogen bonds, van der Waals interactions, synaptic plasticity, disease mutations

Now, a team of researchers from Presidency University Kolkata and the CSIR Indian Institute of Chemical Biology has taken a purely computational route to visualize one of these partnerships in atomic detail. Writing in Frontiers in Bioinformatics, Prakash Nandi, Izaz Monir Kamal, Saikat Chakrabarti and Sugopa Sengupta report a full-length three-dimensional model of yeast Top I and use it to predict, residue by residue, how the enzyme might dock onto Rpc82, a key subunit of RNA polymerase III. Their integrative pipeline—combining template-based structure modeling, molecular docking, molecular dynamics simulations and binding free-energy calculations—paints a plausible picture of a Top I–Rpc82 complex and delivers a testable blueprint for experiments that have never been possible before, largely because the enzyme itself had never been fully drawn.

The obstacle the team faced was structural, not conceptual. Two independent high-throughput studies in budding yeast, both using affinity chromatography followed by mass spectrometry, had already flagged a physical association between Top I and Rpc82, the largest subunit of the RNA polymerase III-specific trimer that also includes Rpc34 and Rpc31. Rpc82 is no bystander: it anchors to the clamp domain of the polymerase cleft, contacts duplex DNA downstream of the transcription bubble, and, through a disordered insertion in its third winged-helix (WH3) domain, engages the TFIIB-related factor Brf1 during pre-initiation complex assembly. Yet while the crystal structure of Rpc82 was available, Top I remained a 769-residue enigma. Only its DNA-binding domain had been resolved crystallographically, and even AlphaFold assigned very low confidence to the enzyme’s N-terminal region—the very region known to mediate protein–protein interactions with partners such as the RNA polymerase II subunit Rpb1 and the sirtuin Sir2.

The team’s answer was to model Top I piece by piece. They divided the protein into five segments: the N-terminal domain (D1, residues 1–142), the DNA-binding domain (D2, residues 143–363), the core domain (D3, residues 364–589), a linker region (D4, residues 590–698), and the C-terminal catalytic domain (D5, residues 699–769). Because the D2 crystal structure already existed, only the remaining four segments required prediction, and the researchers ran each through multiple engines—RoseTTAFold, I-TASSER, TrRosetta, ab initio modeling and DeepFold—rather than trusting any single algorithm. The best-scoring model for each domain was then validated against rigorous metrics: ERRAT scores above 95 percent, Verify3D values exceeding 85 percent, and Ramachandran plots showing more than 91 percent of residues in favored stereochemical regions.

Assembling the validated pieces with MODELLER produced a full-length model that held up remarkably well under scrutiny. The finished structure scored an ERRAT value of 93.43 percent, a Verify3D score of 86.35 percent, and placed 94.3 percent of residues in favored Ramachandran regions with only 0.1 percent in disallowed zones. When the team superimposed their model onto the crystal structure of human topoisomerase I—its closest structural relative, sharing 51.69 percent sequence identity—the overall root-mean-square deviation was just 1.021 angstroms, comfortably within the threshold considered indicative of an accurate model. Superimposition onto the AlphaFold prediction yielded an even tighter 0.7-angstrom agreement for the D2–D5 regions, and crucially, the new model reproduced a single alpha-helix in the N-terminal domain (residues 105–125) that AlphaFold had faintly glimpsed but could not resolve. Benchmarking against dozens of high-resolution eukaryotic crystal structures of comparable length confirmed that every quality metric sat within, and often above, reference ranges.

With a trustworthy Top I in hand, the docking question could finally be asked. The researchers ran blind docking using three independent servers—ClusPro, HDOCK and HADDOCK—followed by targeted docking in which the residues of Rpc82 already committed to binding Rpc34 and Rpc160 were masked, so that only surfaces genuinely accessible inside the living polymerase holoenzyme would be considered. Both strategies converged on the same answer: Top I preferentially engages the WH2–WH3 region of Rpc82, a peripheral patch of the protein that remains spatially free even within the assembled polymerase. The alternative surface, spanning WH1, WH4 and the coiled-coil domain, was set aside because it is pre-occupied by core subunits during holoenzyme assembly and an interaction there would create steric clashes. From the Top I side, the predicted interface was built primarily from the D1 N-terminal domain and the D3 core domain, with minor contributions from D2. The most favorable pose carried a binding energy of −11.5 kilocalories per mole as calculated by PDBePISA, involving 39 intermolecular contacts.

Molecular dynamics then put the model through the equivalent of a 200-nanosecond stress test in water, run in triplicate with the GROMACS engine and the AMBER99SB force field. The simulated complex held together. Root-mean-square deviation values settled around 0.8–1.0 nanometers, the solvent-accessible surface area remained steady at roughly 780–800 square nanometers, and the radius of gyration hovered near 4.3 nanometers, signaling sustained compactness. The buried interface area grew from about 30 to roughly 40 square nanometers over the trajectory—an increase the authors interpret as progressive tightening of the association. Strikingly, the interface itself proved more rigid than the flanks: the D3-mediated contact zone fluctuated by only about 0.3 nanometers, while the D1 side retained more local flexibility, and the non-interacting WH4 and coiled-coil regions of Rpc82 wobbled far more than the binding surface. A rigid interface amid flexible surroundings is exactly what one expects of a genuine, evolutionarily tuned protein–protein contact.

The energetic accounting reinforced the story. MM-PBSA analysis across 400 trajectory snapshots showed the complex’s total energy stabilizing near −37,300 kilocalories per mole in the later simulation stages, while the binding free energy of the interacting patch drifted toward strongly negative values, reaching roughly −300 to −310 kilocalories per mole by the end of the run. The binding energy-to-surface-area ratio of the modeled complex also fell squarely within the range observed for well-characterized eukaryotic heterodimeric complexes resolved at 3 angstroms or better, a benchmarking exercise that helps separate biologically credible interfaces from computational artifacts. Certain residues emerged as persistent hotspots: on Top I, the acidic cluster D50, E51, E126, E127 and E128 in D1, together with Y385, K400, V401, L403, E424, E427, Q486, G488, H489, Q490, Y507 and P509 in D3; on Rpc82, residues including R145, D148, E149, Y185, R368, H369 and H448 maintained contact for extended stretches. The team also observed transient secondary contact patches forming and dissolving over time, suggesting the complex may undergo adaptive rearrangements to preserve its grip under cellular conditions.

What would such a partnership be for? The functional logic is compelling. RNA polymerase III drives extraordinarily high levels of transcription of short genes—transfer RNAs, 5S ribosomal RNA and small nuclear RNAs—and recycles rapidly through repeated initiation cycles, generating intense local torsional strain. In budding yeast, loss of Top I has been shown to increase R-loop accumulation at heavily transcribed polymerase III loci, impairing elongation and pre-tRNA synthesis. A physical tether between Rpc82 and Top I would position the topoisomerase precisely where supercoils pile up, allowing it to relax DNA on demand and keep tRNA production flowing. The interaction would also echo established precedents: bacterial RNA polymerase’s beta-prime subunit contacts bacterial topoisomerase I, and Top I recruitment through RNA polymerase II stimulates relaxation activity during active transcription in flies and humans. Human topoisomerase I has even been reported to promote polymerase III reinitiation and termination through the TFIIIC factor complex.

The authors are careful to frame their results as a hypothesis rather than proof—the entire edifice rests on computation, and no wet-laboratory experiment has yet confirmed the interface. But the model now offers concrete targets for that confirmation: mutations of the identified hotspot residues could disrupt or preserve the interaction in vivo, providing a clean experimental test. Because both proteins are evolutionarily conserved, a comparable Top I–polymerase III interface may exist in human cells, where polymerase III transcription is frequently dysregulated in cancer. If future work validates the contact, the interface could represent a regulatory node—and, in disease contexts, a potential druggable one—linking the enzymes that read the genome with the enzymes that keep it from twisting itself apart.

Subject of Research: In silico prediction of the molecular interaction between yeast DNA topoisomerase I and the RNA polymerase III subunit Rpc82

Article Title: Exploring molecular interactions between yeast topoisomerase I and RNA polymerase III subunit, Rpc82 using an in-silico approach

Article References: Nandi, P., Kamal, I. M., Chakrabarti, S., & Sengupta, S. (2026). Exploring molecular interactions between yeast topoisomerase I and RNA polymerase III subunit, Rpc82 using an in-silico approach. Frontiers in Bioinformatics, 6, Article 1897635. https://doi.org/10.3389/fbinf.2026.1897635

Image Credits: AI Generated

DOI: 10.3389/fbinf.2026.1897635

Keywords: topoisomerase I, RNA polymerase III, Rpc82, Saccharomyces cerevisiae, protein-protein interaction, molecular docking, molecular dynamics simulation, binding free energy, DNA supercoiling, transcription, structural modeling, genome stability

The study began with a large-scale computational sweep through publicly available blood transcriptome data. Using the limma statistical framework, the team compared gene expression in severe COVID-19 patients against healthy controls, drawing on dataset GSE171110, which comprised 44 patient samples and 10 controls. A parallel analysis of the gout dataset GSE160170, with six patients and six controls, applied the same thresholds: an adjusted p-value below 0.05 and a log2 fold change of at least 1. The COVID-19 cohort yielded 3,520 differentially expressed genes, while the gout cohort produced 1,860. Intersecting the two lists revealed just 19 genes shared by both inflammatory disorders — a compact transcriptional module that the authors interpret as the core molecular signature driving the convergent inflammatory response.

Functional enrichment analysis of these 19 shared genes pointed strongly toward the immune system. Gene Ontology terms highlighted B cell proliferation, immunoglobulin biosynthesis, cytokine activity, and cytokine receptor binding. This B-cell signal is particularly intriguing. In severe COVID-19, aberrant B cell activation and autoantibody production have been recognized as key drivers of immunopathology and lung damage. Gout, by contrast, has traditionally been viewed as a disease of innate immunity, driven by neutrophils and macrophages responding to urate crystals. Yet emerging evidence implicates B cells and autoantibody responses in chronic tophaceous gout as well. The convergence on humoral immunity in both diseases suggests a shared mechanism that has been underappreciated until now. KEGG pathway analysis further linked the shared gene set to autoimmune conditions including rheumatoid arthritis, asthma, and inflammatory bowel disease, underscoring the systemic nature of the overlapping inflammatory program.

To distill the 19 shared genes down to actionable targets, the researchers constructed a protein-protein interaction network using the STRING database at a high-confidence threshold of 0.7. The resulting network contained 19 nodes and 41 edges, with CXCL8, JUN, and IL10 showing the highest connectivity. Applying the CytoNCA plugin in Cytoscape, the team evaluated five centrality metrics — degree, closeness, eigenvector, local average connectivity, and betweenness — and identified six genes that ranked highest across all measures: CXCL8, MPO, JUN, MMP1, IL4, and IL10. These hub genes were significantly upregulated in both disease cohorts compared with healthy controls, and correlation heatmaps revealed strong positive co-expression among them in each dataset, suggesting they operate as a synergistic regulatory unit rather than isolated players.

Enrichment analysis of the six hub genes reinforced their relevance to both pathologies. Gene Ontology terms included positive regulation of MHC class II biosynthetic process, myeloid leukocyte activation, and regulation of isotype switching — processes central to adaptive immunity and implicated in the dysregulated humoral responses of severe COVID-19. KEGG pathway analysis tied the hub genes to the IL-17 signaling pathway, T cell receptor signaling, and rheumatoid arthritis. The IL-17 axis, in particular, has been independently implicated in both gout pathogenesis and COVID-19-associated hyperinflammation, positioning these six genes as convergent nodes within a shared inflammatory network.

The pivotal moment came when the team intersected these six hub genes with the predicted protein targets of colchicine, the ancient anti-inflammatory drug long used to treat acute gout attacks and recently repurposed for COVID-19 with encouraging results in trials such as GRECCO-19. Using the SwissTargetPrediction platform, the researchers generated a list of 100 putative colchicine targets from its PubChem entry. Only one of the six hub genes appeared on that list: MMP1. This single overlap suggested that colchicine’s therapeutic benefit in both diseases might flow, at least in part, through direct engagement of this matrix metalloproteinase — a mechanism not previously described.

To test this hypothesis computationally, the team performed molecular docking between colchicine and the MMP1 protein structure obtained from the Protein Data Bank. The docking simulation, run with AutoDock Vina, revealed a stable binding conformation with a calculated binding energy of −6.4 kcal/mol, a value within the range typically considered favorable for drug-target interactions. Colchicine formed conventional hydrogen bonds with the ASN-180 and ALA-182 residues of MMP1, at distances of 2.3 and 2.6 angstroms respectively, and was further anchored by a network of hydrophobic contacts within the catalytic pocket. Because these hydrogen bonds sit within the enzyme’s catalytic domain, the authors propose that colchicine binding may directly interfere with MMP1’s proteolytic activity, thereby modulating extracellular matrix remodeling and inflammatory cell migration.

Docking alone cannot guarantee that a ligand remains bound under physiological conditions, so the team followed up with a 100-nanosecond all-atom molecular dynamics simulation using GROMACS 2023 and the AMBER99SB-ILDN force field. The root mean square deviation values equilibrated within the first 10 nanoseconds and maintained a stable plateau thereafter, indicating a steady conformational state. Root mean square fluctuation analysis showed consistently low local motion across most residues, and the radius of gyration declined gradually over the first 60 nanoseconds before stabilizing around 1.46 nanometers, reflecting progressive compaction of the complex. The number of intermolecular hydrogen bonds remained steady at one to two throughout the trajectory, and comparison of snapshots at 0 and 100 nanoseconds showed no substantial dissociation or deformation of colchicine within the binding pocket. Together, these results support a durable and specific interaction between the drug and its newly identified target.

Beyond colchicine, the researchers explored a dual screening strategy for additional therapeutic candidates targeting the six hub genes. Among the top 10 predicted small molecules were simvastatin, which has established anti-inflammatory properties and has been reported to attenuate NLRP3 inflammasome activation; ciprofloxacin, a fluoroquinolone with immunomodulatory effects; and nimesulide, a selective COX-2 inhibitor that could theoretically ease gout flares, though its hepatotoxicity profile warrants caution. The team also reviewed traditional Chinese medicine formulations reported to act on the hub genes, identifying six prescriptions including Ma Xing Shi Gan Decoction, Yi-Fei-Jian-Pi-Tang, and Taodan Granules. Yi-Fei-Jian-Pi-Tang showed the broadest targeting spectrum, modulating five of the six hub genes, while Taodan Granules targeted three. The authors caution that several computational hits, such as nickel sulfate and electrocorundum, are environmental agents rather than drugs, and that all predictions require rigorous experimental validation before clinical translation.

Perhaps the most striking finding came from single-cell RNA sequencing analysis of peripheral blood mononuclear cells, drawing on datasets GSE188172 and GSE211783 and encompassing more than 94,000 cells across severe COVID-19 patients, gout patients, and healthy controls. After quality control and batch correction using Harmony integration, the team annotated eight major immune cell subpopulations and found that MMP1 expression was significantly upregulated in COVID-19 patients, with the elevation localized predominantly to B cells and plasma cells rather than myeloid compartments. This B-cell-restricted pattern was further accentuated in ventilated COVID-19 patients compared with non-ventilated cases, implying a positive correlation between MMP1 expression in B cells and disease severity. In gout patients, by contrast, MMP1 showed a more distributed expression across monocytes and macrophages, consistent with the innate immune-driven pathology of crystal-induced inflammation. The authors suggest that while MMP1 serves as a shared molecular hub at the transcriptomic level, its cellular sources and functional roles may be disease-context-dependent, and that therapeutic targeting may ultimately require cell-type-specific strategies. Taken together, the findings position MMP1 as a promising candidate for patients facing the double burden of gout and severe COVID-19, while underscoring the substantial gap that remains between computational prediction and clinical application.

Subject of Research: Shared molecular pathways between gout and severe COVID-19, focusing on MMP1 as a common hub gene and drug target

Article Title: MMP1 serves as a common link and drug target between gout and severe COVID-19 infection

Article References: Ma, H., Geng, Y., Liu, M., Feng, T., Lv, Y., Wang, H., & Niu, Y. (2026). MMP1 serves as a common link and drug target between gout and severe COVID-19 infection. Results in Chemistry, 31, Article 103967. https://doi.org/10.1016/j.rechem.2026.103967

Image Credits: AI Generated

DOI: 10.1016/j.rechem.2026.103967

Keywords: MMP1, gout, severe COVID-19, colchicine, molecular docking, molecular dynamics simulation, single-cell RNA sequencing, NLRP3 inflammasome, hub genes, bioinformatics, inflammation, drug repurposing

The challenge stems from the nature of TCR recognition itself. Naturally occurring TCRs often bind tumor antigens with suboptimal affinity, and simply boosting binding strength can backfire, reducing specificity and raising the risk of off-target attacks on healthy tissue. Affinity enhancement strategies have historically produced receptors that cross-react with unrelated peptides, a potentially lethal side effect. Writing in Frontiers in Chemical Biology, Mario Frezzini and Daniele Narzi argue that the field needs more than raw computational power; it needs mechanistic understanding, a way of knowing not just which receptor variants bind well, but why they bind well and what molecular consequences each mutation carries.

Artificial intelligence has transformed the search for better TCRs in recent years. Deep learning models can explore vast stretches of TCR sequence space, predicting which amino acid sequences are compatible with a desired structural framework and generating libraries of candidate variants in silico. Structure-based deep learning approaches have shown genuine promise in proposing receptor variants with improved binding properties. But these tools, the authors contend, are largely data-driven and descriptive: they rely on static structural representations and scoring functions that fail to capture the dynamic, ever-fluctuating nature of protein-protein interactions. The result is an embarrassment of riches—thousands of candidates that are difficult to prioritize, and little insight into the physical determinants of recognition.

Molecular dynamics simulations sit at the opposite end of the spectrum. By explicitly modeling every atom of a TCR-peptide-MHC complex as it jiggles, flexes, and interacts in a simulated watery environment, MD reveals conformational plasticity, residue-specific energetic contributions, and the cooperative behavior of the binding interface. Decades of simulation work have shown that TCR recognition is anything but a rigid lock-and-key event; the complementarity-determining regions, the receptor’s antigen-grasping loops, fluctuate significantly, contributing to both adaptability and specificity. The catch is computational cost. Reliable MD characterization demands all-atom simulations under physiological conditions, roughly ten independent replicas per system, and sampling times of at least hundreds of nanoseconds per replica to reach statistical convergence. That makes MD suitable for analyzing only a few tens of closely related variants, not for screening hundreds of candidates.

Rather than viewing these limitations as weaknesses, Frezzini and Narzi propose exploiting them deliberately. In their framework, MD is positioned as a low-throughput, high-information filter—a step that examines a carefully chosen handful of candidates in exquisite detail. The pipeline begins with biologically grounded receptor selection: repertoire sequencing of tumor-infiltrating lymphocytes, peripheral blood, or antigen-specific T cell populations, combined with bioinformatic analysis of clonal expansion, persistence, and CDR3 sequence similarity, identifies candidate receptors with known or putative tumor reactivity. AI-driven design then explores the local sequence space around these validated scaffolds, generating variant libraries. Fast pre-screening tools—empirical scoring functions, rapid energetic estimates, or machine learning predictors—trim the candidate pool to a manageable size before the expensive physics-based refinement begins.

At the heart of the workflow, MD serves as an integrative mechanistic engine rather than a brute-force screener. Trajectory analyses quantify how flexibly the CDR loops move, which interaction patterns stabilize the complex over time, and, through residue-specific energy decomposition schemes, which individual amino acids contribute favorably to binding. Crucially, the simulations also map allosteric communication pathways—networks of dynamically coupled residues that connect the binding interface to distal regions of the receptor. Local perturbations can propagate through the structure and modulate the behavior of residues tens of angstroms away, meaning that mutations in one part of the receptor can reshape binding chemistry at another. This dual description of energetics and allostery gives the pipeline a mechanistic criterion for prioritizing variants that goes far beyond static sequence-based metrics.

The power of this approach is illustrated by the authors’ own recent computational work on a pair of closely related engineered TCRs, c796 and its parent c728. The two receptors differ by a single conservative mutation, E31D, in the CDR1 loop, yet experiments show the variant binds its target peptide-MHC complex roughly six-fold more tightly. Counterintuitively, detailed energetic analysis revealed that the mutated residue itself contributes nothing directly to the improved affinity. Instead, the single substitution triggers a redistribution of interaction energies across multiple residues, including positions in distal CDR loops that now stabilize the interface more favorably. Network analyses of simulation trajectories traced allosteric pathways linking the mutation site to these distant regions. In other words, binding affinity in TCR systems emerges from a cooperative web of interactions, and even a minimal sequence change can produce substantial functional consequences through dynamic, long-range mechanisms—precisely the kind of hidden logic that only MD-based mechanistic analysis can expose.

The framework’s final stage returns to the laboratory, where two complementary experimental technologies validate the computationally refined candidates. Surface plasmon resonance, the gold standard for biophysical characterization, directly measures the association and dissociation rates of TCR-peptide-MHC binding, providing quantitative affinity and kinetic parameters for a limited set of high-confidence variants. But SPR requires recombinant protein expression and purification, offers relatively low throughput, and cannot capture the conformational and energetic complexity that governs T cell activation. That gap is filled by cellular reporter assays such as RAPTR, which evaluate TCR activation in response to specific peptide-MHC complexes within a living cellular environment, capturing downstream signaling and functional output at higher throughput. Importantly, binding affinity alone does not necessarily correlate with functional efficacy, so these orthogonal layers of validation—molecular binding on one side, cellular function on the other—are essential for identifying receptors that are both strong and clinically useful.

What distinguishes the proposed strategy from previous pipelines is its iterative, bidirectional character. Energetic fingerprints and allosteric maps derived from MD are fed back into the AI design models, narrowing and sharpening the sequence space they explore in the next round. Experimental results, in turn, benchmark and improve the computational predictions. The authors describe this as a closed-loop, multi-scale design paradigm, building on recent demonstrations that lab-in-the-loop AI-guided optimization can work for therapeutic antibody engineering. Emerging generative tools such as AlphaFlow and BioEmu may eventually accelerate conformational sampling upstream, and the authors envision hybrid strategies in which AI-generated structural ensembles initialize diverse conformations for explicit-solvent MD refinement—though they caution that these newer methods have not yet been validated for multichain TCR-peptide-MHC complexes or mutation-induced energy changes.

The approach is not without caveats. MD-based mechanistic insight is only as reliable as the structural model underlying the simulations, and accurate models of TCR-peptide-MHC complexes remain challenging to obtain because of the intrinsic flexibility of the CDR loops and the complexity of protein-protein recognition. AI tools like AlphaFold have improved structural prediction enormously, but performance for receptor-ligand complexes is still limited without experimental structural data. The authors therefore stress that every mechanistic hypothesis generated in silico must be experimentally validated before guiding the next round of optimization. If the framework delivers on its promise, however, the payoff could be substantial: a scalable, adaptive design engine that unifies statistical learning with physical principles, capable of producing the high-affinity, exquisitely specific, and safe T cell receptors that next-generation cancer immunotherapy demands. The convergence of AI, molecular dynamics, and experimental validation may finally turn TCR engineering from an empirical art into a predictive science.

Subject of Research: Integrating artificial intelligence, molecular dynamics simulations, and experimental assays for mechanism-aware engineering of T cell receptors in cancer immunotherapy

Article Title: Bridging AI, molecular dynamics, and experiments for mechanism-aware TCR engineering in cancer immunotherapy

Article References: Frezzini, M., & Narzi, D. (2026). Bridging AI, molecular dynamics, and experiments for mechanism-aware TCR engineering in cancer immunotherapy. Frontiers in Chemical Biology, 5, Article 1909618. https://doi.org/10.3389/fchbi.2026.1909618

Image Credits: AI Generated

DOI: 10.3389/fchbi.2026.1909618

Keywords: T cell receptor engineering, cancer immunotherapy, artificial intelligence, molecular dynamics simulations, allosteric communication, peptide-MHC complex, surface plasmon resonance, RAPTR assay, protein design, adoptive T cell therapy, binding affinity, computational biology

Wettability, in essence, describes how fluids spread over and adhere to rock surfaces. It is quantified through the contact angle, the angle formed where a droplet meets a solid at a three-phase junction, governed thermodynamically by the Young-Laplace equation. For CO2-brine-rock systems, researchers classify wettability into five regimes based on the brine contact angle: strong water-wetness from 0 to 50 degrees, weak water-wetting from 50 to 70 degrees, neutral wetting between 70 and 110 degrees, mild CO2 or oleic wetting from 110 to 130 degrees, and strong CO2 or oil-wetting from 130 to 180 degrees. These angular thresholds are far from academic trivia. When a shale surface shifts from water-wet toward weakly water-wet, residual water becomes trapped and crude oil movement is restricted. Conversely, if a caprock turns CO2-wet, its sealing performance deteriorates and the risk of CO2 leakage rises sharply.

The review highlights how the field has moved beyond the limitations of conventional bulk contact-angle testing. Researchers now deploy integrated multi-scale schemes: nanoscale in-situ atomic force microscopy, microscale visualization using shale-based microfluidic chips, and macroscale joint measurements via nuclear magnetic resonance and mercury injection capillary pressure. Molecular dynamics simulations fill in the extreme conditions that laboratory experiments cannot reach, modeling fluid behavior in nanoslits under reservoir temperatures and pressures. On the bulk scale, NMR-based methods built on spontaneous imbibition trials yield a Normalized Wettability Index ranging from −1, indicating complete oil-wettability, to +1, indicating complete water-wettability, with values between −0.5 and 0.5 signifying the mixed-wettability states common in organic-rich shales.

Yet the measurement landscape remains fragmented. Contact angles measured at different locations, or with different techniques, can yield contradictory results, with deviations between macroscopic contact angle measurements, microscopic NMR, and nanoscale molecular simulation reaching 15 to 20 percent. The sessile drop method works only on ideal smooth surfaces; the captive bubble technique reduces brine diffusion errors but suffers from alignment issues; the Wilhelmy plate and tilted plate methods each carry their own constraints. Because real CO2 injection involves penetration into pore spaces rather than spreading across flat surfaces, discrepancies between pore-scale and core-scale data persist, and no unified correction system yet exists.

The controlling factors of wettability alteration form the mechanistic heart of the review. Temperature effects remain controversial: some experiments show contact angles dropping as temperature rises, consistent with increasing CO2-brine interfacial tension, while others report the opposite or negligible trends. Molecular dynamics studies generally agree that above the critical temperature of 304.41 K, water droplets flatten and their contact angles decrease progressively, but the smooth, single-mineral substrates assumed in most simulations diverge from real reservoir conditions. Pressure, by contrast, more consistently pushes shale toward CO2-wettability: rising pressure increases CO2 density and strengthens intermolecular adhesion between rock and CO2 molecules, while reducing CO2-brine interfacial tension. Under extreme supercritical conditions of 30 MPa and 343 K, the review notes, system wettability may even undergo complete reversal to a fully CO2-wet state.

Salinity and ion type add another layer of complexity. Higher salinity increases advancing and receding contact angles and drives shale from hydrophilic toward mixed and CO2-wet states, a mechanism explained by electrical double layer screening: dissolved ions adsorb onto oppositely charged rock surfaces, forming hydrophobic layers that reduce surface polarity. Divalent ions outperform monovalent ones in this regard. Experimental data show contact angles rising from 26 to 31 degrees for NaCl and from 29 to 43 degrees for MgCl2 as concentration increases from 0.1 to 1.0 molar, with the hydrophobic effect ordered MgCl2 greater than CaCl2 greater than NaCl. High-salinity environments enhance CO2 permeation tendencies, which the authors flag as unfavorable for both shale gas exploitation and geological storage.

Organic matter introduces perhaps the most consequential variable. Total organic carbon content generally correlates with increasing CO2-wettability: shales with high TOC tend to be CO2-wet, while low-TOC shales remain strongly water-wet and are better suited as caprocks. But the relationship is not universal. In the Silurian Longmaxi marine shale of the Sichuan Basin, contact angle varies parabolically with TOC, first decreasing and then increasing, a pattern attributed to how mobile organic components distribute within the matrix. In Permian and Triassic transitional and terrestrial shales, the correlation inverts, likely governed by organic maceral composition—vitrinite being more water-wet, liptinite more oil-wet. Even shales with TOC as high as 7.9 percent can remain hydrophilic overall when the mineral assemblage, which constitutes 80 to 90 percent of sample volume, dominates the wetting behavior.

Chemical reactions compound these effects over time. Dissolved CO2 forms carbonic acid, whose hydrogen ions attack feldspar, carbonate, and clay minerals, reshaping pore-throat structures and mineral compositions. X-ray diffraction studies after supercritical CO2 exposure show elevated quartz proportions alongside reduced carbonate and clay abundances. Mineral dissolution enlarges pores and reduces water-wettability, while secondary precipitation of newly formed minerals can block throats and, in the long run, enable the most stable form of carbon trapping: mineral trapping. Surface chemistry shifts in parallel, as acidic conditions catalyze the condensation of hydrophilic silanol groups into Si-O-Si linkages, reducing hydrogen bonding with water and raising contact angles.

The review is unflinching about the field’s gaps. Most experiments run for less than 168 hours, far too short to capture the slow extraction of organic matter and continuous mineral dissolution that unfold over years or decades in actual reservoirs, processes that could trigger a secondary wettability reversal. Laboratory cores measuring 25 by 50 millimeters cannot account for fracture distribution and lithofacies heterogeneity at kilometer-scale reservoirs, and no scale-conversion verification has yet been performed. Machine learning models trained on simple mineral systems—using algorithms from support vector machines to random forests and adaptive neuro-fuzzy inference systems—struggle to capture the complexity of real shale, though emerging approaches embed physical constraints such as interfacial tension thermodynamics directly into loss functions.

The practical stakes are considerable. The authors identify a favorable lithological assemblage for storage site selection: high-TOC shale in the lower interval, suited to adsorption trapping—which accounts for nearly 98 percent of trapped gas in shale—capped by low-TOC, water-wet shale above for structural trapping. Wettability regulation technologies, including anionic surfactants such as alpha-olefin sulfonates, nanoparticle treatments with SiO2 and Al2O3, and surfactant foams, have achieved 18 to 25 percent recovery improvements at laboratory scale, though engineering applicability remains unproven. Even microbial effects, mediated through biosurfactants and biofilm-induced changes to pore geometry, remain nearly unstudied in shale-CO2 systems. As the world pursues dual goals of shale hydrocarbon development and industrial carbon reduction, the review makes clear that mastering the invisible chemistry of wetting—at scales from single molecules to entire basins—is the prerequisite for storing CO2 safely beneath our feet.

Subject of Research: Wettability alteration of the CO2-water-shale system and its implications for geological carbon storage and enhanced shale hydrocarbon recovery

Article Title: Wettability alteration of CO2-water-shale system: state of the art and future directions

Article References: Luo, C., Yin, X., Zeng, B., Yuan, J., Huo, X., Lin, H., & Yin, N. (2026). Wettability alteration of CO2-water-shale system: state of the art and future directions. Environmental Earth Sciences, 85(16), Article 428. https://doi.org/10.1007/s12665-026-13163-1

Image Credits: AI Generated

DOI: 10.1007/s12665-026-13163-1

Keywords: wettability alteration, CO2 geological storage, shale reservoirs, contact angle, carbon sequestration, total organic carbon, molecular dynamics simulation, nuclear magnetic resonance, caprock integrity, brine salinity, supercritical CO2, enhanced oil recovery

The quantum limit, first articulated in theoretical work by Pendry and Landauer in the 1980s, describes the maximum heat current that a single perfect channel can carry. Each independent mode of vibration, or phonon, contributes a fixed quantum of thermal conductance, and no material, however pure, can exceed this ceiling for a given channel. Measuring such tiny heat flows, on the order of picowatts per kelvin, has historically required cryogenic temperatures and exquisitely engineered nanostructures. Earlier landmark experiments measured the quantum of thermal conductance in suspended silicon nitride membranes at low temperatures, and subsequent work probed ballistic phonons in carbon nanotubes and superlattices. But the fundamental limits of phonon-mediated heat transport at the atomic scale, at ambient conditions, remained unknown, largely because the experimental tools did not exist.

A team led by Longji Cui at the University of Colorado Boulder, working with collaborators at the Instituto de Ciencia de Materiales de Madrid, the Universidad Autónoma de Madrid, and other institutions, set out to close that gap. Their solution was a custom-fabricated scanning thermal microscope probe of unusual stiffness and sensitivity. The probe, dubbed Hs-SThM for high-stiffness scanning thermal microscopy, incorporates a niobium nitride sensing element defined by electron-beam lithography and reactive sputtering, metallized with chromium and gold for electrical isolation. Finite element simulations reveal directional stiffnesses reaching 2,000 newtons per meter laterally and an extraordinary 50,000 newtons per meter along the pressing axis, values that allow the tip to form and hold stable contacts with surfaces at the atomic scale without buckling or drifting.

The measurement strategy borrows a page from the field of molecular electronics. For decades, researchers have used mechanically controllable break junctions and scanning tunneling microscopes to form and rupture single-atom electrical contacts, capturing the discrete conductance signatures of individual atoms and molecules. The Colorado-led team adapted this concept to heat, creating what they call phononic break junctions. The probe is brought into contact with a surface and then slowly withdrawn at a controlled speed of roughly 0.11 nanometers per second. Thermal feedback, monitored through the probe’s temperature response, signals the transition from near-field radiative heat transfer across a vanishing gap to solid contact, marked by a sharp jump in measured conductance. As the tip retracts further, the contact narrows through a nanocontact to a single atomic bridge, and the thermal conductance is recorded throughout in real time.

The material under scrutiny was amorphous silicon dioxide, the glass that forms the insulating layers of virtually every integrated circuit ever made. In bulk form, this glass is a poor thermal conductor: its disordered network of silicon and oxygen atoms scatters vibrational energy into diffusive modes known as diffusons, alongside localized modes called locons and only a minority of truly propagating modes. Macroscopic thermal conductivity is correspondingly low. Yet when the researchers pulled their probe away from a silica surface and the contact thinned to just a few atoms, the measured thermal conductance did not collapse. Instead, it settled onto reproducible plateaus, with the finest junctions supporting a phononic thermal conductance of up to forty percent of the quantum limit at room temperature.

That figure is all the more striking when compared with crystalline metals. The atomically thin glass junctions conducted heat roughly an order of magnitude better, per phonon channel, than the phonon thermal conductance of crystalline gold atomic contacts, which the team measured in parallel as a reference system. Gold, despite its perfect lattice, is a poor phonon conductor because its heavy atoms vibrate at low frequencies. Silica, with its light oxygen atoms, supports much higher vibrational frequencies, and this difference proved decisive. Through molecular dynamics simulations performed with non-equilibrium methods and spectral heat current decomposition, the researchers found that the phononic thermal conductance of an atomic junction is dictated not by the degree of structural order but by the Debye frequency, the maximum vibrational frequency the material can support.

The simulations, run with the LAMMPS package using Tersoff-parameterized potentials for the silicon-oxygen system and embedded-atom-method potentials for gold, reproduced the experimental conductance traces in striking detail. Spectral decomposition of the heat current across a one-dimensional silica contact revealed that the energy distribution entering the junction through one plane exits essentially unperturbed at another, a hallmark of coherent transport. Most of the transmitted heat lies below the Debye frequency cutoff, and the vibrational spectrum is preserved across the junction rather than scrambled by the amorphous structure. In other words, at the atomic limit, the glass behaves not as a diffusive insulator but as a ballistic waveguide for coherent phonons, with structural disorder failing to impose the fundamental penalty that dominates its macroscopic behavior.

Statistical rigor was essential to rule out artifacts. Because each rupture event samples a different local atomic configuration, the team recorded thousands of traces and applied an unsupervised machine-learning pipeline combining principal component analysis with Gaussian mixture model clustering. The clustering, performed on jump-independent feature vectors without any manual labels, identified five distinct trace families, and multiple independently identified clusters showed a peak or shoulder near the dominant conductance maximum of approximately 1.2 times the single-mode quantum of thermal conductance. This repeated spectral weight across clusters indicates that the characteristic conductance is not produced by one anomalous trace subclass but reflects a statistically favored atomic-scale junction configuration in the amorphous network, with cluster-dependent variations arising from local bonding geometry and rupture pathway.

The implications extend well beyond fundamental physics. Heat dissipation is a first-order problem in nanoscale electronic and quantum devices, and the finding that intrinsic disorder does not fundamentally constrain atomic-scale heat flow opens a new design space. If the Debye frequency, rather than crystallinity, is the governing parameter, then materials can be selected and engineered for atomic-scale thermal links on the basis of their vibrational spectra alone. Light-atom glasses and other disordered solids, long dismissed as thermal dead ends, may serve as efficient phononic interconnects where crystalline channels are impractical. The break-junction platform itself offers an experimental testbed for studying wave-like coherent phonon transport at room temperature, complementing cryogenic quantum-coherence experiments on phononic crystals and superlattices.

The researchers anticipate that their approach will enable the engineering of high-performance thermal phononic devices, including thermal diodes that pass heat preferentially in one direction and thermal transistors that modulate heat currents, the functional analogs of their electronic counterparts. Earlier work has demonstrated phonon-based routing and rectification in nanostructures such as a graphite thermal Tesla valve driven by hydrodynamic phonon transport, and coherent phonon control has been used to manage interactions between high-frequency phonons and single quantum systems. An atomic-scale platform that works at ambient temperature, with picowatt resolution and reproducible single-atom contacts, brings those device concepts closer to practical realization. The data and code underlying the study have been released openly, and the team suggests that the same thermal-feedback break-junction technique can be extended to other disordered materials, molecular junctions, and hybrid inorganic-organic contacts, charting a path toward a technology in which heat itself is guided, switched, and computed with at the scale of individual atoms.

Subject of Research: Coherent phonon heat transport through atomic-scale amorphous silicon dioxide junctions

Article Title: Thermal conductance of atomic-scale glass reaches 40% of quantum limit

Article References: Yelishala, S. C., Zhu, Y., Martinez, P. M., Mateos-Lopez, O., Xiong, W., Chooi, D., Kimura, S., Cuevas, J. C., Vilhena, J. G., & Cui, L. (2026). Thermal conductance of atomic-scale glass reaches 40% of quantum limit. Nature Nanotechnology. https://doi.org/10.1038/s41565-026-02308-9

Image Credits: AI Generated

DOI: 10.1038/s41565-026-02308-9

Keywords: phonon transport, thermal conductance, quantum limit, amorphous silicon dioxide, break junction, scanning thermal microscopy, ballistic transport, Debye frequency, nanoscale heat dissipation, thermal diode, molecular dynamics simulation, glass

A new computational study published in Discover Chemistry by Zara Javaid, Hafiza Salaha Mahrosh, Saad Serfraz and Amer Jamil of the University of Agriculture, Faisalabad, set out to close this gap by screening plant-derived molecules against two of the virus’s most strategically important proteins. The team assembled a library of 1,000 phytochemicals with documented antiviral, antioxidant or anticancer activity, drawn from the IMPPAT database, Dr. Duke’s Phytochemical and Ethnobotanical Database and published literature, and docked them against glycoprotein 42, the envelope protein that mediates viral entry into B cells, and Epstein–Barr nuclear antigen 1, the latency protein that maintains and partitions the viral episome during cell division.

The choice of targets reflects a deliberate dual strategy. Glycoprotein 42 is a surface-exposed glycoprotein that binds HLA class II molecules and works alongside the gH and gL glycoproteins to drive membrane fusion, making it an accessible point of attack during the earliest stages of infection. Peptide inhibitors directed at this protein have already shown nanomolar antiviral activity in prior work, validating its therapeutic potential. EBNA1, by contrast, is a 641-amino-acid nuclear protein expressed in every EBV-linked cancer and is essential for viral genome replication and persistence; no FDA-approved inhibitor of it exists. By targeting both an entry factor and a latency factor simultaneously, the researchers aimed to identify compounds with combined antiviral and antineoplastic potential, an approach that distinguishes their workflow from earlier studies focused on single targets or synthetic libraries.

The docking screen, performed with the AutoDock Vina engine inside PyRx, produced clear front-runners. Against glycoprotein 42, the steroidal saponin hecogenin achieved the strongest binding energy of −8.9 kcal/mol, forming a hydrogen bond with glutamine 92 and hydrophobic contacts with tyrosine 167, phenylalanine 129 and histidine 191. Other strong performers included baicalin, islandicin, swertisin, hortensin, pedalitin, epigallocatechin, pinoresinol, hesperetin and verbenalin. Against EBNA1, the plant sterol cyclobranol led with a score of −8.2 kcal/mol, anchoring itself in a hydrophobic pocket through contacts with phenylalanine 508, tyrosine 510 and leucine 539. Notably, acyclovir, used as a comparator control, scored only −4.7 kcal/mol against glycoprotein 42 and −4.9 kcal/mol against EBNA1, placing it last among the screened ligands and suggesting that the top phytochemicals bind both viral proteins more favorably than the standard clinical antiviral.

The authors took care to validate their docking protocol. The native co-crystallized ligand of EBNA1 was extracted and redocked under identical parameters, reproducing the experimental binding geometry with a heavy-atom root-mean-square deviation of 0.958 angstroms, well within the accepted threshold of 2 angstroms. Redocking validation was not possible for glycoprotein 42, which was solved in its unbound form without a co-crystallized ligand, a structural limitation the team acknowledged openly. Docking scores, they stressed, are estimates of binding compatibility rather than predictions of biological potency, and experimental confirmation remains essential.

From the top-ranked compounds, the researchers applied Lipinski’s rule of five as an initial filter for oral drug-likeness. Hesperetin, hecogenin, islandicin, hortensin, pedalitin, pinoresinol and verbenalin passed with no violations, while brassicasterol, swertisin, caftaric acid and epigallocatechin each violated a single criterion. Compounds with more than two violations, including cyclobranol, stigmasterol, baicalin and taraxerol, were excluded. The authors noted that natural products often carry greater structural complexity than synthetic drugs, so broader physicochemical descriptors were reserved for the later pharmacokinetic assessment rather than used as absolute exclusion criteria at this early stage.

The study then moved into network pharmacology, linking the surviving compounds to their predicted human gene targets through SwissTargetPrediction and the STITCH database, and cross-referencing these against EBV- and EBV-cancer-associated genes retrieved from GeneCards and DisGeNET. The intersection yielded 132 shared targets, which were mapped into a protein–protein interaction network using STRING and visualized in Cytoscape. The network contained 132 nodes and 827 edges, and degree-based analysis with the CytoHubba plugin identified ten hub genes: TP53, HSP90AA1, AKT1, SRC, BCL2, EGFR, TNF, ESR1, ERBB2 and PIK3CA. Gene Ontology and KEGG enrichment showed these genes clustered in apoptosis regulation, kinase signaling, the PI3K-Akt pathway and viral oncogenesis pathways, including EBV infection itself and Kaposi’s sarcoma-associated herpesvirus infection, consistent with the virus’s known strategy of suppressing host cell death and hijacking kinase cascades to maintain latency and drive proliferation.

When the eight drug-like phytochemicals were docked against the EBV-associated cancer proteins, hecogenin again dominated, achieving the best score against nine of the ten targets. Its strongest interaction was with the TP53 DNA-binding domain at −10.9 kcal/mol, stabilized by hydrogen bonds with glycine 112, phenylalanine 113 and serine 116, followed by −10.6 kcal/mol against PIK3CA and −10.4 kcal/mol against SRC. Pinoresinol, swertisin and hesperetin also performed well. To test whether the hecogenin–TP53 interaction could survive realistic physical conditions, the team ran a 200-nanosecond molecular dynamics simulation in GROMACS using the CHARMM36 force field on an NVIDIA H100 GPU. The protein backbone remained stable throughout, with RMSD values of 2.0 to 2.8 angstroms and a nearly constant radius of gyration, while the ligand’s internal geometry barely changed, with a ligand RMSF of only about 0.35 angstroms. The ligand did, however, wander considerably relative to its docked pose, and hydrogen bond analysis revealed mostly transient polar contacts, indicating that the complex is held together primarily by hydrophobic and van der Waals forces in a dynamic rather than rigid binding mode.

Predicted pharmacokinetics added further support. ADMET profiling with SwissADME and pkCSM indicated no AMES toxicity, hepatotoxicity or skin sensitization for the lead compounds, and both hecogenin and pinoresinol were predicted to cross the blood–brain barrier and show high gastrointestinal absorption, placing them in the favorable yolk region of the Boiled Egg plot. As P-glycoprotein substrates, both would be actively pumped out of the central nervous system, limiting CNS exposure but potentially improving peripheral safety. The authors were candid about the caveats: hecogenin’s broad affinity across structurally unrelated proteins raises the possibility of promiscuous, hydrophobicity-driven binding rather than genuine multitarget specificity, and although the compound triggered no PAINS alerts, orthogonal assays such as surface plasmon resonance and thermal shift experiments will be needed to separate true polypharmacology from assay interference.

Hecogenin itself is a steroidal saponin from Agave sisalana with a prior record of anti-inflammatory, antimicrobial and antitumor research, while pinoresinol is a lignan from sesame seeds and olive oil, and hesperetin and swertisin are flavonoids with documented anticancer and immunomodulatory properties. The study’s real contribution may lie as much in its framework as in its lead compound: an integrated pipeline that couples large-scale docking, network pharmacology, enrichment analysis, ADMET prediction and molecular dynamics into a reproducible workflow for discovering mechanism-based antiviral agents from natural sources. The researchers emphasize that all findings are computational and hypothesis-generating, and that in vitro and in vivo validation is required before hecogenin or its counterparts can be considered genuine therapeutic candidates against EBV infection and the cancers it leaves in its wake.

Subject of Research: Computational screening of phytochemicals as multitarget inhibitors of Epstein–Barr virus infection and EBV-associated malignancies

Article Title: Network pharmacology and molecular docking-based screening of phytochemicals as multitarget inhibitors of Epstein–Barr virus infection and associated malignancies

Article References: Javaid, Z., Mahrosh, H. S., Serfraz, S., & Jamil, A. (2026). Network pharmacology and molecular docking-based screening of phytochemicals as multitarget inhibitors of Epstein–Barr virus infection and associated malignancies. Discover Chemistry, 3(1), Article 577. https://doi.org/10.1007/s44371-026-01029-1

Image Credits: AI Generated

DOI: 10.1007/s44371-026-01029-1

Keywords: Epstein–Barr virus, network pharmacology, molecular docking, hecogenin, phytochemicals, EBNA1, glycoprotein 42, molecular dynamics simulation, ADMET, oncogenic viruses, drug discovery, TP53