Virtual Lab Hunts New Anti-Sickling Drug Candidates for Sickle Cell Disease

Sickle cell disease remains one of the most common inherited blood disorders in the world, and despite decades of research, most patients still rely on treatments that manage symptoms rather than address the root cause. A new computational study from researchers at Guru Ghasidas Vishwavidyalaya in Bilaspur, India, published in Discover Chemistry, offers a glimpse of how modern in silico methods could change that picture. By combining bioisosteric molecular design, pharmacokinetic prediction, molecular docking, molecular dynamics simulation and quantum chemical calculations, the team identified two hybrid molecules that appear to bind sickle hemoglobin more strongly than currently approved drugs, while also showing encouraging safety profiles in predictive models.

The biological target of the study is the central event in sickle cell pathology: the polymerization of hemoglobin S. The disease stems from a single point mutation in the beta-globin gene, in which glutamic acid at the sixth position is replaced by valine. When oxygen levels fall, this seemingly small change causes hemoglobin molecules to clump into rigid fibers, deforming red blood cells into their characteristic sickle shape. These rigid cells clog capillaries, a process known as vaso-occlusion, triggering painful crises, hemolysis and progressive organ damage. Repeated cycles of polymerization and depolymerization also generate oxidative stress, depleting nitric oxide in the vasculature and fueling inflammation that involves platelets, neutrophils and the endothelium.

Current therapies reflect the limits of treating a downstream process. Hydroxyurea raises fetal hemoglobin levels and reduces sickling, voxelotor binds hemoglobin reversibly to keep it in its oxygenated state, and crizanlizumab blocks platelet-mediated adhesion. L-glutamine offers antioxidant support, while regular transfusions manage severe cases. The only curative option, hematopoietic stem cell transplantation from a matched sibling donor, is limited by donor scarcity, cost and serious risks including graft-versus-host disease. Gene editing therapies are emerging but remain inaccessible to most patients globally. This therapeutic gap motivated the Indian team to search for small molecules that could directly inhibit hemoglobin S polymerization while simultaneously counteracting oxidative stress.

The researchers chose two pharmacophores with complementary biological credentials. The first is caffeic acid, a phenolic compound found in coffee, wine, tea and propolis, which has documented antioxidant, anti-inflammatory and anti-carcinogenic properties. Previous experimental work showed that caffeic acid increases the delay time for hemoglobin polymerization, a direct anti-sickling effect, and significantly lowers the concentration of ferryl hemoglobin formed during peroxide reactions. The second is phthalimide, whose derivatives, including the thalidomide family, have been shown to increase gamma-globin expression and fetal hemoglobin production while lowering tumor necrosis factor alpha, giving them both anti-sickling and anti-inflammatory potential.

Using an AI-assisted online tool called Mol-Opt, the team generated bioisosteric analogues of both scaffolds, systematically swapping functional groups to improve potency, pharmacokinetics and safety while preserving the molecular features responsible for activity. This yielded 373 candidate analogues across the two series. Each was then screened through a battery of computational filters. The ADMET Lab 3.0 platform predicted absorption, distribution, metabolism, excretion and toxicity parameters, while the Osiris Property Explorer provided drug-likeness and drug-score assessments. Encouragingly, every analogue satisfied Lipinski’s rule of five, the Pfizer rules and, with few exceptions, the Golden Triangle criteria, suggesting the designed molecules occupy drug-like chemical space with reasonable prospects for oral absorption.

The pharmacokinetic and toxicity screens narrowed the field considerably. Among the caffeic acid derivatives, compounds such as C-168, C-2, C-29, C-195 and C-27 stood out for combining high drug scores with low predicted toxicity. Analogue C-2 proved especially notable, showing the lowest hemotoxicity and genotoxicity values in its series. In the phthalimide series, P-60 and P-41 displayed excellent ADME balance with minimal tissue accumulation and very low skin sensitization, although several otherwise promising analogues, including P-155, P-171, P-202 and P-248, were rejected because of alarmingly high predicted genotoxicity, illustrating how computational toxicity screening can eliminate liabilities before any laboratory synthesis takes place.

The decisive test came from molecular docking against the crystal structure of deoxyhemoglobin S, available in the Protein Data Bank under the identifier 2HBS at 2.05 angstrom resolution. Using the Pyrex implementation of AutoDock Vina, the team docked the analogues into the hemoglobin active site and visualized the interactions with ICM Browser Pro. The phthalimide analogue P-2 achieved the study’s best binding affinity at minus 10.4 kilocalories per mole, followed by P-56 at minus 10.3. Among the caffeic acid derivatives, C-2 led with minus 9.0 kilocalories per mole, ahead of C-72 and C-68. For comparison, the approved drugs hydroxyurea, L-glutamine and voxelotor were docked under the same conditions, and the top analogues outperformed them, forming networks of hydrogen bonds and hydrophobic contacts with residues including LYS-66, ASN-102, ASP-99, VAL-98, LEU-28 and PHE-41.

Because a good docking score does not guarantee that a ligand remains bound in the dynamic environment of a real protein, the two lead compounds were subjected to 100-nanosecond molecular dynamics simulations using the Desmond package in the Schrödinger suite. The P-2 hemoglobin complex proved remarkably stable: after an initial equilibration period, the ligand’s root mean square deviation, fitted to the protein, settled within 1.2 to 2.2 angstroms and remained steady for the rest of the simulation. The C-2 complex showed somewhat greater fluctuation, with average protein RMSD values between 0.8 and 3.2 angstroms, but still maintained its binding pose within the active pocket. Contact analysis revealed that the ligands were held in place by a combination of hydrogen bonds, hydrophobic interactions and water bridges involving residues such as PHE-41, LYS-95, VAL-98, ASP-99 and GLU-43, providing a plausible structural basis for sustained anti-sickling activity.

To probe the electronic foundations of the two leads, the researchers performed density functional theory calculations using the ORCA 5.0.4 program at the B3LYP level with the def2-TZVP basis set. Compound C-2 displayed a HOMO energy of minus 8.63 electron volts localized over its catechol pi-system, identifying it as the principal electron-donating region, while its LUMO at minus 5.23 electron volts spread across the nicotinate ring and ester carbonyl groups. The resulting HOMO-LUMO gap of 3.40 electron volts indicates moderate electronic delocalization and efficient charge-transfer pathways, properties associated with reversible electron transfer and enhanced biological activity. Compound P-2 showed a smaller gap of 2.62 electron volts, with the HOMO distributed over its aromatic indole and benzoyl amide regions and the LUMO concentrated at the carbonyl nicotinate position, suggesting good conjugation and moderate chemical reactivity across the amide framework.

The authors conclude that the combined computational workflow, spanning bioisosteric design, ADMET prediction, docking, dynamics and quantum chemistry, has delivered two credible lead candidates, P-2 and C-2, that merit progression toward laboratory validation. Experimental studies to confirm their anti-sickling activity are reportedly underway. As with any purely computational study, the predictions remain hypotheses until synthesis, binding assays and animal or clinical testing are completed, and the high genotoxicity flags on several analogues underline the need for careful optimization. Nevertheless, the work demonstrates how a rational, multi-method in silico pipeline can compress the early stages of drug discovery, potentially cutting costs and timelines for a disease that has waited decades for a truly transformative small-molecule therapy.

Subject of Research: Computational screening of phthalimide and caffeic acid analogues as potential anti-sickling agents for sickle cell disease

Article Title: Computational screening of some newer potential anti-sickling compounds using docking, ADMET, DFT, and molecular dynamics approaches

Article References: Verma, D., Ahirwar, B., Vaishnav, Y., & Gahrewal, M. (2026). Computational screening of some newer potential anti-sickling compounds using docking, ADMET, DFT, and molecular dynamics approaches. Discover Chemistry, 3(1), Article 495. https://doi.org/10.1007/s44371-026-00942-9

Image Credits: AI Generated

DOI: 10.1007/s44371-026-00942-9

Keywords: sickle cell disease, hemoglobin S polymerization, molecular docking, ADMET, molecular dynamics simulation, DFT, bioisosteres, phthalimide, caffeic acid, drug discovery, computer-aided drug design, vaso-occlusion