In Silico Discovery of Novel CDK1 Inhibitors from Linum Species for Targeted Cancer Therapy

The interaction between vicenin-2 isomer 2 (binding energy: −8.9413 kcal/mol) and 6GU7 revealed six potential hydrogen bonds: five hydrogen-donor interactions involving the amino acid residues Asp128, Asp146, Asp86, Ser84, and Asp128, and one hydrogen-acceptor interaction with Lys89. The corresponding interaction energies were −0.8, −2.6, −0.8, −1.8, −1.2, and −2.4 kcal/mol, with hydrogen bond distances of 3.18, 2.76, 3.24, 2.8, 3.34, and 2.81 Å, respectively (Figure 3).

Figure 3.

a) 2D and b) 3D interactions of vicenin-2 isomer 2 in the binding site of 6GU7. (in 3D: Ligand in purple, residues pocket in orange, hydrogen bonds in green, pi interactions in red)

This extensive hydrogen-bonding network aligns with computational studies of high-affinity CDK1 inhibitors, where interaction with Asp86 and Asp146 in the hinge region is consistently reported as critical for sub-micromolar activity ⁽³³⁾. The binding energy scores falls within the range reported for known flavonoid-based CDK1 inhibitors (−8.5 to − 10.2 kcal/mol) in similar docking studies using MOE. ⁽³⁴⁾

The highest number of hydrogen bond interactions between ligands and the 6GU7 complex was observed with 8,3ʹ,4ʹ-trihydroxyflavone-7-O-6(6ʺ-O-p-coumaroyl)-β-d-glucopyranoside (Figure 4), which achieved the second-lowest binding energy among all ligands tested (−8.8836 kcal/mol). This compound formed six hydrogen-donor interactions with the residues Asp128 (two bonds), Ile10, Asp86, and Asp146, as well as two hydrogen-acceptor interactions with Lys89, and one additional interaction with Thr14. The interaction energies were −2.0, −4.1, −2.1, −1.0, −3.9, −5.5, −3.8, −1.0, and −1.4 kcal/mol, with corresponding bond distances of 2.81, 2.69, 2.88, 3.16, 2.65, 2.74, 2.84, 3.30, and 4.26 Å.

Figure 4.

a) 2D and b) 3D interactions of 8,3ʹ,4ʹ-trihydroxyflavone-7-O-6(6ʺ-O-p-coumaroyl)-β-d-glucopyranoside in the binding site of 6GU7. 3D: Ligand in purple, residues pocket in orange, hydrogen bonds in green, pi interactions in red.

Moreover, most compounds identified from the EtOAc extracts of Linum species (LN and LT) established at least four potential hydrogen bonds with the 6GU7 complex. The detailed interaction data for beast 12 compounds are provided in the supplementary Information (Table S1).

The interaction between 6,4ʹ-dimethoxy-scutellarein-7-neohesperidoside and the 5LQF receptor A (binding energy: −9.8332 kcal/mol) revealed seven hydrogen bond interactions. These included two H-donor bonds with amino acids Ile10 and Glu12, and three H-acceptor bonds with Gln132, LEU83, and Glu12. Additionally, two π-H interactions were observed with Val18 and Ala145. The corresponding interaction distances were 2.83, 2.79, 2.92, 3.43, 2.97, 4.28, and 4.22 Å, with respective interaction energies of −3.0, −1.7, − 1.7, −1.4, −2.4, −0.6, and −0.8 kcal/mol (Figure 5). This binding affinity (−9.83 kcal/mol) is particularly notable, as it exceeds the threshold of −9.0 kcal/mol that Sankket et al ⁽³⁴⁾ correlated with sub-micromolar CDK1 inhibition in their benchmark study of 15 flavonoids as anticancer.

Figure 5.

a) 2D and b) 3D interactions of 6,4ʹ-dimethoxy in the binding sites of 5LQF receptor A. 3D: Ligand in purple, residues pocket in orange, hydrogen bonds in green, pi interactions in red.

For other receptor D, the interaction with Vicenin-2 isomer 2 binding energy: −9.6997 kcal/mol (Figure 6), exhibited two hydrogen bonds and tow π-H interactions. These included one H-donor and one H-acceptor bonds with Leu83, as well as two π-H interactions involving Val18 and Ala145. The corresponding distances were 3.09, 3.40, 4.33, and 4.19 Å, with associated interaction energies of −0.7, −1.4, −0.7, and −0.8 kcal/mol, respectively (Figure 4b). Furthermore, the other compounds interactions with the 5LQF/A complex were described in supplementary Tables S2 and S3).

Figure 6.

a) 2D and b) 3D interactions of vicenin-2 isomer 2 in the binding sites of 5LQF receptor D. 3D: Ligand in purple, residues pocket in orange, hydrogen bonds in green, pi interactions in red.

The results of this study confirm that several compounds present in the Linum EtOAc extracts exhibit a strong affinity for CDK1 complexes, particularly for the CDK1/cyclin B1/CKS2 (5LQF) complex, suggesting their potential role as CDK1 inhibitors. Comparative analysis between the two protein complexes studied revealed that 5LQF is the most stable with the ligands tested. This may be attributed to the synergistic stabilizing effect of cyclin B1 and CKS2, which maintain the CDK1 conformation, thereby enhancing its interaction with inhibitors ⁽³⁵⁾.

One of the most promising compounds, 8,3ʹ,4ʹ-trihydroxyflavone- 7-O-6(6ʺ-O-p-coumaroyl)-β-D-glucopyranoside, was identified exclusively in the EtOAc extract of LN. This compound displayed the lowest binding energies at both 5LQF sites A and D, which may explain the stronger cell cycle blockade observed for the LN EtOAc extract compared to the LT EtOAc extract, as previously reported ⁽¹⁰⁾. Moreover, the major compounds identified in the EtOAc extracts of LN and LT, notably vicenin-2 and its isomers, showed high affinity for both CDK1 complexes (5LQF and 6GU7). The affinity ranges and interaction patterns of flavonoid derivatives in CDK1 docking studies were similar, with substantial hydrogen bonding and hydrophobic interactions impacting inhibitory potential. Retamal & Caballero 2016⁽³⁶⁾ found that flavonoid scaffolds docking simulations reveal different binding orientations and activity patterns due to hydrogen bond donors and hydrophobic interactions. Moreover, many biochemical tests shown that nitrogen-containing flavonoid analogues inhibit CDK1/Cyclin B, highlighting the importance of flavonoid scaffolds for CDK1 suppression. ⁽³⁷⁾

To confirm that the investigated compounds act as competitive ATP inhibitors, we analyzed their binding modes in comparison with the co-crystallized reference inhibitors NU6102 (in 5LQF) and AZD5438 (in 6GU7). Docking simulations revealed that all top-ranked compounds bind directly within the ATP-binding pocket and exhibit a high degree of pose similarity with these known competitive inhibitors. This spatial overlap with established ATP-competitive inhibitors provides strong computational evidence for a competitive mechanism, consistent with validation protocols used in kinase inhibitor discovery pipelines ⁽³⁸⁾

The binding pose superpositions in Figure 7 demonstrate that Vicenin-2 isomer 2 not only occupies the ATP-binding pocket but also forms a more robust hydrogen-bonding network than the reference compounds. Vicenin-2 forms six hydrogen bonds (described above) including the critical hinge interaction, compared to four for AZD5438 (Figure 7a) and only one for Dinaciclib (Figure 7b). This enhanced interaction profile within the same site provides compelling evidence that Vicenin-2 acts as a competitive inhibitor.

Figure 7.

Overlaid 2D interactions diagrams in 6GU7 receptor.

a) Vicenin-2 isomer 2 (green) and the reference inhibitor AZD5438 (red).

b) Vicenin-2 isomer 2 (green) and Dinaciclib (red).

Also 6,40-dimethoxy scutellarein-7-neohesperidiside demonstrate a superior interaction profile compared to NU6102. In the ATP-binding pocket of 5LQF receptor A (Figures 8a), our compound forms seven hydrogen bonds (described above), one more than the reference ligand. In receptor site D (Figures 8b), both compounds perform equally well, each forming seven hydrogen bonds, confirming the high affinity of our compound for the target sites. The findings align with earlier research indicating that dinaciclib is a powerful ATP-competitive inhibitor of CDK1 Parry et al., 2010 ⁽³⁹⁾and that NU6102 suppresses CDK1/2 in cancer cells under treatment, the increased affinity of NU6102 for 5LQF may result from the stabilizing effects of cyclin B1 and CKS2, establishing a dependable criterion for assessing natural compounds as prospective CDK1 inhibitors.

Figure 8.

Overlaid 2D interactions diagrams:

a) 6,4ʹ-dimethoxy (green) and the reference inhibitor NU6102 (red) in 5LQF receptor A.

b) Vicenin-2 isomer 2 (green) and Dinaciclib (red) in 5LQF receptor D.

This affinity, which exceeds that of the reference inhibitors, suggests a promising inhibitory potential likely responsible for the observed antiproliferative effects. The docking results also revealed that several other compounds present in the EtOAc extracts of LN and LT, including 6,4ʹ-dimethoxyscutellarein- 7-neohesperidoside, luteolin-7-O-β-d-glucoside, rutin, violanthin, foliasalacioside B1, isovitexin 2ʺ-Oarabinoside, vitexin-2ʺ-rhamnoside, and malvidin 3-O-β-galactoside, exhibited strong binding affinities toward CDK1, with binding energies lower than those of the reference inhibitors tested. This aligns with computational studies showing that 37 flavonoids often outperform inhibitors in CDK1 docking due to their extended interaction networks ⁽³⁶⁾. This high affinity reinforces our previous findings: inhibition of CDK1 prevents activation of the cyclin B/CDK1 complex, a key regulator of the G2/M transition, thereby explaining the G2/M cell cycle arrest observed in PC3 cells following treatment with the EtOAc extracts from both plants. Similar correlations between CDK1 docking scores and G2/M arrest have been reported for other natural products. ⁽⁴⁰⁾

These findings suggest that the identified natural compounds could serve as promising candidates for the development of CDK1-targeted inhibitors in innovative cancer therapies.

Analysis of the molecular interactions revealed a complex network of bonds, mainly hydrogen bonds, but also hydrophobic and polar interactions. These multiple interactions confer notable stability to the formed complexes, an essential requirement for effective CDK1 inhibition.

Hydrogen bonds ensure precise anchoring of the ligands within the active site, promoting an optimal orientation for interaction, whereas hydrophobic interactions significantly contribute to the stabilization of the ligand-protein complex. This multimodal binding mechanism plays a key role in the molecular recognition and selective inhibition of enzymatic targets. The combination of hydrophobic filling, electrostatic steering, and optimized hydrogen bonding represents a comprehensive binding strategy observed in successful kinase inhibitor designs. ⁽⁴¹⁾

In line with the findings of Sahu et al. ⁽⁴²⁾, hydrophobic forces play a fundamental role in the final phase of the binding process, filling hydrophobic cavities within the active site, reducing void spaces, and reinforcing complex cohesion through van der Waals interactions. Electrostatic forces also contribute to the initial long-range molecular recognition, facilitating the orientation of ligands toward the catalytic pocket.

Another factor influencing complex stability lies in the specific geometry of the binding systems. As demonstrated by Rupniak et al. ⁽⁴³⁾, a temporarily stabilized ring system, dependent on the conformation of donor and acceptor groups, can reduce the energetic penalty associated with desolvation. Such a conformation may enable ligands like vicenin-2 to navigate the protein structure more effectively, thereby enhancing their capacity to establish long-lasting and specific interactions with CDK1.

Finally, the work of Thorson et al. ⁽⁴⁴⁾ emphasizes that even a single additional hydrogen bond can enhance the biological activity of a ligand by up to a hundredfold, underscoring its critical role in optimizing both affinity and selectivity.

This principle is reflected in our findings, where the stabilized conformations of the ligand–CDK1 complexes strengthen anchoring within the active site, supporting a competitive inhibition mechanism with strong therapeutic potential.

Overall, these observations confirm not only the structural stability of the formed complexes but also their pharmacological relevance as selective CDK1 inhibitors. They thus open up promising perspectives for the development of new naturally occurring antiproliferative agents targeting CDKs.