Virtual screening yields two drug candidates that disable c-Myc in colorectal cancer

Colorectal cancer remains one of the most common solid tumors worldwide, accounting for roughly ten percent of all newly diagnosed cancers and associated mortality. For patients whose disease has metastasized, the five-year survival rate still sits below twenty percent, a stark figure despite decades of progress in endoscopic resection, surgery, and chemoradiotherapy. That gap between treatment advances and patient outcomes has pushed researchers toward new molecular targets, and few targets are as tantalizing — or as frustrating — as c-Myc. The transcription factor is a central oncogenic driver, upregulated in about seventy percent of colorectal cancer patients, and experiments in mouse models have shown that knocking down c-Myc suppresses tumorigenesis. Yet c-Myc has long been labeled undruggable, and no c-Myc-targeted small-molecule compound has entered clinical trials for colorectal cancer treatment.

A research team has now reported a systematic attempt to change that. Their strategy centered on a specific weakness in c-Myc biology: the protein exerts its oncogenic activity primarily by pairing with a partner transcription factor called Max. The resulting c-Myc–Max heterodimer recognizes the canonical E-box DNA sequence, CACGTG, and modulates the expression of downstream target genes through either transcriptional activation or repression. Because this protein–protein interaction sits at the heart of c-Myc-driven tumorigenesis, disrupting the dimerization interface offers a way to block c-Myc without needing to grip the protein itself. Prior work supports the concept — compound B13, for example, has been shown to block the c-Myc/Max interaction and suppress primary colorectal tumor growth in xenograft mouse models — but the field still lacks potent, optimized inhibitors developed specifically for colorectal cancer.

The discovery workflow began in silico. Using the Surflex docking algorithm within the Sybyl-X2.1 software package, the team targeted the c-Myc/Max complex structure deposited in the Protein Data Bank under the identifier 1NKP. They selected a structurally well-characterized binding pocket composed of Leu917, Phe921, and Lys939 on c-Myc, together with Arg212, Arg215, Asp216, Ile218, Lys219, Phe222, and Arg239 on Max — the same interface strategy reported in earlier studies. Candidate molecules came from the ChemDiv database, a library of 1.6 million commercially available compounds with purity above ninety-five percent. Before any docking took place, the library was filtered through Lipinski’s Rule of Five, which imposes limits on molecular weight, hydrogen bond donors and acceptors, and lipophilicity to exclude non-drug-like molecules. That first round of screening retained about 1.06 million compounds.

The screening then narrowed the field in stages. A low-precision docking pass rapidly assessed the binding affinity of the 1.06 million survivors against the target pocket, and only the top one percent — roughly 10,600 compounds — advanced. The team then re-docked those candidates under high-precision, standard parameters and kept the top 500 ranked molecules. Finally, an integrated strategy combining docking score ranking with structural clustering analysis selected four compounds for commercial purchase: 6069–0336, C201–1652, D144–0281, and E959–1030, with docking scores of 8.263, 8.695, 8.874, and 9.036 in units of −log Kd, respectively. Nuclear magnetic resonance spectra confirmed the identity of the four purchased compounds, and the entire funnel — from 1.6 million molecules to four testable hits — illustrates how modern virtual screening can compress an otherwise impossible chemical search into a manageable bench experiment.

The four hits then faced living cells. The team tested them against SW480 and Caco-2 colorectal cancer cell lines, both well characterized and both known to depend on c-Myc signaling for proliferation and survival, using a CCK-8 viability assay across concentrations from one to twenty micromolar over forty-eight hours. Three of the four compounds — C201–1652, D144–0281, and E959–1030 — significantly reduced the viability of both cell lines in a dose-dependent manner, while 6069–0336 showed comparatively weak cytotoxicity. Quantitatively, E959–1030 was the most potent, with an IC50 of 4.81 micromolar in SW480 cells and 11.34 micromolar in Caco-2 cells, followed by C201–1652 at 9.51 micromolar in SW480 and D144–0281 at 16.98 and 12.94 micromolar in the two lines, respectively. All three active compounds showed minimal cytotoxicity toward HEK293T cells within the tested range, whose IC50 values exceeded twenty micromolar, yielding selectivity indices above 1.0. Benchmarking against the classic c-Myc inhibitor 10,058-F4, whose IC50 exceeded fifty micromolar in both cancer lines, the screened compounds displayed superior antiproliferative potency.

Apoptosis assays added a second layer of evidence. Defective apoptotic signaling is what allows cancer cells to evade cell cycle arrest and proliferate uncontrollably, so compounds that restore programmed cell death are considered reliable candidates in anti-tumor drug development. Flow cytometry using an annexin V-APC/7-AAD detection kit showed that C201–1652, D144–0281, and E959–1030 were significantly more potent than 6069–0336 at inducing apoptosis in SW480 cells, and all three effectively promoted apoptosis in Caco-2 cells as well. These effects occurred at low toxic doses. The authors caution, however, that HEK293T is an immortalized transformed cell line, so the observed selectivity reflects only preliminary in-vitro safety relative to that reference line and cannot be extrapolated to primary normal colon epithelial cells.

The mechanistic experiments produced the study’s most striking finding. Western blot analysis revealed that D144–0281 and E959–1030 significantly downregulated c-Myc protein levels in both SW480 and Caco-2 cells, consistent with the obvious suppression produced by the positive control 10,058-F4. Yet quantitative PCR showed no significant change in c-Myc mRNA levels in either cell line, indicating that the compounds do not directly regulate c-Myc transcription but instead appear to affect the protein post-transcriptionally — a pattern the authors note mirrors the small-molecule inhibitors MYCMI-7 and MY05, which trigger c-Myc protein destabilization and degradation without obvious changes in transcript levels. Dual luciferase reporter assays, in which HEK293T cells were transfected with a c-Myc expression plasmid, an MYC response element-luciferase reporter, and a Renilla reference plasmid, then demonstrated that D144–0281 and E959–1030 suppressed c-Myc-driven transcriptional activity in a dose-dependent manner, while C201–1652 showed no obvious inhibitory effect. GST pull-down assays using recombinant GST-c-Myc and His-Max proteins further indicated that high doses of D144–0281 and E959–1030 interfered with the physical interaction between c-Myc and Max.

Molecular dynamics simulations then stress-tested the predicted binding. The docked conformations served as starting structures for 200-nanosecond simulations in Amber18, with systems solvated in TIP3P water, heated from zero to 300 kelvin over 100 picoseconds, and run at a two-femtosecond time step. Root mean square deviation analysis showed all systems fluctuating between two and five angstroms, with average values of 4.1 angstroms for the apo-protein, 4.6 angstroms for the c-Myc/Max–D144–0281 complex, and 2.9 angstroms for the c-Myc/Max–E959–1030 complex; both ligand-bound systems stabilized within ten nanoseconds. Root mean square fluctuation analysis revealed that significant movements occurred mainly at the protein termini, far from the binding site, while residues at the active center fluctuated below two angstroms. MM/GBSA calculations yielded binding free energies of −40.72 kilocalories per mole for D144–0281 and −32.39 kilocalories per mole for E959–1030, both indicating favorable and spontaneous interactions. Energy decomposition showed D144–0281 binding mainly through van der Waals contacts with the hydrophobic region, engaging Leu917, Lys918, Phe921, Lys939, and Leu943 on c-Myc plus Ile218, Lys219, Phe222, and Arg239 on Max, whereas E959–1030 relied predominantly on electrostatic interactions and formed a stable hydrogen bond with the side chain of Arg239 on Max.

Virtual ADME profiling with SwissADME painted a mixed but encouraging drug-likeness picture. D144–0281 has a molecular weight of 443.45 grams per mole and E959–1030 weighs 390.52 grams per mole, both within the accepted range for oral agents, and both compounds are predicted to show high gastrointestinal absorption, zero Lipinski violations, and a bioavailability score of 0.55. Both are also predicted to cross the blood-brain barrier — a property the authors flag as a potential liability rather than purely an advantage, since it raises the risk of off-target central nervous system exposure. Metabolic profiling identified distinct safety concerns: D144–0281 inhibits multiple cytochrome P450 isoforms, elevating the risk of drug-drug interactions, while E959–1030 inhibits only CYP2D6 and CYP3A4, carrying a relatively lower metabolic interaction risk.

The authors are candid about what the study does and does not prove. The docking was based on an apo protein crystal structure with no co-crystallized ligand, so conventional redocking validation was impossible, and the Surflex scoring function provides only relative binding affinity estimates rather than absolute binding free energies; the MD simulations support the stability of the predicted complexes, but the binding poses remain putative and require further experimental validation. The GST pull-down results demonstrate disruption of the c-Myc/Max interaction but do not constitute direct biophysical evidence of physical binding between the compounds and either protein — surface plasmon resonance assays will be needed to confirm direct target engagement. The work was restricted to in-vitro characterization without prior medicinal-chemistry optimization, so solubility, metabolic stability, and pharmacokinetic profiles remain unknown, and no xenograft assays or combination studies with clinical colorectal cancer chemotherapeutics were performed. Still, by occupying an identical binding pocket of the c-Myc/Max complex, interfering with the heterodimer, suppressing c-Myc protein post-transcriptionally, and blocking downstream transcriptional activity, D144–0281 and E959–1030 establish a basis for further preclinical evaluation of c-Myc/Max inhibitors for colorectal cancer.

Subject of Research: Discovery of c-Myc/Max complex inhibitors for colorectal cancer through virtual screening and in vitro validation

Article Title: Integrated virtual screening and in vitro evaluation identify potential c-Myc/max complex inhibitors for colorectal cancer

Article References: Miao, Z., Ge, G., Chen, A., Liu, X., Kuang, Y., & Cui, H. (2026). Integrated virtual screening and in vitro evaluation identify potential c-Myc/max complex inhibitors for colorectal cancer. Results in Chemistry, 31, Article 103982. https://doi.org/10.1016/j.rechem.2026.103982

Image Credits: AI Generated

DOI: 10.1016/j.rechem.2026.103982

Keywords: colorectal cancer, c-Myc, Max, virtual screening, molecular docking, protein-protein interaction, apoptosis, molecular dynamics, MM/GBSA, ADME, drug discovery, in vitro