Large-Scale Computational Screening Identifies First in Class Multitarget Inhibitor of EGFR Kinase and BRD4

Abstract

Inhibition of cancer-promoting kinases is an established therapeutic strategy for the treatment of many cancers, although resistance to kinase inhibitors is common. One way to overcome resistance is to target orthogonal cancer-promoting pathways. Bromo and Extra-Terminal (BET) domain proteins, which belong to the family of epigenetic readers, have recently emerged as promising therapeutic targets in multiple cancers. The development of multitarget drugs that inhibit kinase and BET proteins therefore may be a promising strategy to overcome tumor resistance and prolong therapeutic efficacy in the clinic. We developed a general computational screening approach to identify novel dual kinase/bromodomain inhibitors from millions of commercially available small molecules. Our method integrated machine learning using big datasets of kinase inhibitors and structure-based drug design. Here we describe the computational methodology, including validation and characterization of our models and their application and integration into a scalable virtual screening pipeline. We screened over 6 million commercially available compounds and selected 24 for testing in BRD4 and EGFR biochemical assays. We identified several novel BRD4 inhibitors, among them a first in class dual EGFR-BRD4 inhibitor. Our studies suggest that this computational screening approach may be broadly applicable for identifying dual kinase/BET inhibitors with potential for treating various cancers.

Figure 1. Transcriptional profiles of EGFR and BRD4 inhibitors are distinct.

The box plot shows correlation matrix variances for each comparison of signatures generated by target inhibitors in the MCF7 cell line at 10 μM for 24 hours. P-values obtained by Welch’s two sample t-test comparing the populations of pairwise correlations.

Figure 2. High-throughput computational screening pipeline to identify dual inhibitors of EGFR kinase and BRD4(1).

Figure 3. ROC curves of the BRD4 docking data fusion model.

The activity cutoff is (a) pIC₅₀ ≥ 5 and (b) pIC₅₀ ≥ 6. The ROC curve is true positive rate (sensitivity) over false positive rate (1 - specificity).

Figure 4. Docking scores versus reported BRD4 pActivities (pIC₅₀/pK_d/pK_i) and Pearson correlation results of representative BRD4 inhibitors by chemotypes.

Figure 5. Predicted probability of EGFR activity (EstPGood for pIC₅₀ ≥ 6) vs BRD4 aggregate/fusion docking score for 908 compounds that were filtered from commercial libraries using the EGFR kinase Laplacien-modified Naïve Bayes classifiers and physicochemical properties.

Circles represent purchased compounds tested in the BRD4(1) assay, categorized by their activity against BRD4 and EGFR. Only compounds that were confirmed active against BRD4 were further tested for activity against EGFR; any of the black circles therefore could also be EGFR actives.

Figure 6. Concentration response profile of 2870

(A) Average concentration-response profile of compound 2870 and known BRD4 inhibitor I-BET151 using the BRD4(1) biochemical alpha screen assay (n = 3). Emission data was normalized using DMSO and is reported as % response. Average IC₅₀ of 2870 was found to be 9.02 μM against BRD4(1). (B) The EGFR kinase radioisotope filter binding assay was performed at a substrate concentration of 10 μM against 2870 and Staurosporine (n = 3). Average IC₅₀ of 2870 was found to be 0.044 μM against EGFR kinase.

Figure 7. Confirmed active compounds identified by the computational screening pipeline

Activity from the alphascreen assay against BRD4(1) and closest topological similarity to known BRD4(1) binders is also shown. Five other compounds comprising the same sulfonamide scaffold as 8302 were also confirmed actives but are not displayed.

Figure 8. Molecular dynamics analysis Receptor ligand poses as obtained by docking simulations of 2870 in (A) BRD4(1) and (B) EGFR TKD and primary receptor ligand interactions (H-bonds, hydrophobic, ionic, water bridges) of 2870 for the duration of the MD simulations with (C) BRD4(1) and (D) EGFR TKD respectively.

The top panel in (C, D) shows the total number of specific contacts the protein makes with the ligand over the course of the trajectory. The bottom panel shows which residues interact with the ligand in each trajectory over time. Some residues make more than one specific contact with the ligand, which is represented by a darker shade of orange, according to the scale to the right of the plot.