Fig 1.
EGFR downstream signaling and EGFR dimerization.
(a) A profile of the activation and transduction of EGFR downstream signals. (b) Binding-site residues (of WT EGFR) for two TKIs (Gefitinib and Erlotinib). (c) A sketch of the binding between an EGFR kinase mutant and a TKI, with primary binding-site residues listed. (d) The allosteric mechanism for kinase dimerization of EGFR and its partner. (e) The kinase dimerization for a TKI-blocked EGFR and its potential partner. (f) The average expressions of ErbB receptors in breast carcinoma, NSCLC and Colon carcinoma. (g) The proliferation of H460/TKI-R cells when treated with AG1024 or continuous Erlotinib. (h) The effects of depletion of c-Met, EGFR, ErbB-2 and ErbB-3 on cell proliferation of EBC-1 and H1993 cell lines.
Fig 2.
The three-dimensional templates in our structural analysis.
(a) and (b), The structures of gefitinib (PDB: 2ITY), with its atoms and essential pharmacophore (quinazoline ring) shown. (c) and (d), The structures of erlotinib (PDB: 1M17), with its atoms and essential pharmacophore (quinazoline ring) shown. (e) Complex 2ITY (in PDB), composed of an active WT EGFR kinase (N-lobe + C-lobe) and an inhibitor (gefitinib); major gefitinib-interacting residues are labeled as red balls. (f) to (i), The structures of the kinase domains of ErbB-2 (PDB: 3PP0), ErbB-3 (PDB: 4RIW), IGF-1R (PDB: 1K3A) and c-Met (PDB: 1R1W), respectively. (j) The asymmetric dimer that is formed by two WT EGFR kinases.
Fig 3.
Structural comparison of the TKI-binding caves of several EGFR WT or mutant proteins, with Gefitinib exhibited.
(a) to (c), Structural comparison among TKI-binding caves of WT EGFR, mutant L858R, and mutant L858R_T790M; major differences at the gatekeeper position (790) are labeled with green circles. (d) to (f), Binding-site residues of WT EGFR, mutant L858R, and mutant L858R_T790M, with residues colored as follows, LEU (red), ALA (green), GLY (blue), LYS (yellow), GLU (magenta), MET (cyan), THR (dark red), GLN (dark green), and PRO (dark blue). (g) to (i), Respective TKI-binding caves of mutants L858R, delE746_A750 and delL747_P753inS, with major structural differences at the gatekeeper position labeled. (j) to (l), Colored binding-site residues of mutants L858R, delE746_A750 and delL747_P753inS.
Fig 4.
Backbone RMSD curves of several EGFR mutant-partner systems in the 2-ns production MD simulations.
(a) to (c), The RMSD curves for the WT—ErbB-2, WT—IGF-1R and WT—c-Met systems, respectively. (d) to (f), The curves for the mutant L858R-related systems. (g) to (i), The scenarios where the mutant delE746_A750 is involved. (j) to (l), The curves for the systems concerning mutant delL747_P753insS.
Fig 5.
Binding free energy and its components of each mutant-inhibitor or mutant-partner system.
(a) to (c), The energies for the systems that each involves a mutant and a dimerization partner (ErbB-2, IGF-1R or c-Met). (d) and (e), The energies for those systems concerning a mutant and an inhibitor (gefitinib or erlotinib). (f) The total binding free energies extracted from parts a to e, with the mutation types shown. (g) The statistical results of the mutation types among our NSCLC patients.
Fig 6.
Total binding free energies of the mutant-inhibitor or mutant-partner systems, involving several specific mutants.
(a), The total binding free energies between several important EGFR kinases (WT, L858R, L858R_T890M) and a dimerization partner (ErbB-2, IGF-1R, c-Met) or an inhibitor (gefitinib, erlotinib). (b) to (f), Ranked binding free energies for the systems regarding all of our mutants (37 types + WT + L858R_T790M) in an ascending order.
Fig 7.
The difference between the total binding free energies of the mutant-inhibitor system and the mutant-partner system for each mutant.
(a) and (b), The binding free energy differences between the systems involving ErbB-2 and the two inhibitors (gefitinib and erlotinib), with the values ranked in the descending order. (c) and (d), The similar binding free energy differences between the systems concerning IGF-1R and the two inhibitors. (e) and (f), The binding free energy differences between the systems concerning c-Met and the two inhibitors.
Fig 8.
A regression analysis on different feature sets and the median PFS values for patient groups.
(a) The median PFS values (in months) for all the patient groups, with an outlier circled. Each blue point represents a patient group, where the included patients share the same EGFR mutation type and inhibitor. (b) to (j), The regression results based on features sets S1 to S9, after an outlier-removal from part a. The blue curves indicate the predicted PFS values for our patient groups.
Table 1.
RAE and RRSE values for the regressions.
Fig 9.
A regression tree generated based on feature set S3.
S3 = (MI1, MI2, MI3, MI4, MG1, MG2, MG3, MG4), which includes the interaction patterns, characterized by binding free energy components, in a mutant-IGF-1R or mutant-inhibitor system for each patient group. Patients in the same group share the same EGFR mutation type and inhibitor.
Fig 10.
Computational results for ErbB-3-partner systems.
(a) to (f), Respective backbone RMSD curves of ErbB-3—WT, ErbB-3—L858R, ErbB-3—L858R_T790M, ErbB-3—c-Met, ErbB-3—ErbB-2 and ErbB-3—IGF-1R systems, in the 2-ns production MD simulations. (g) Binding free energies and their components of ErbB-3-partner systems. (h) Binding free energies and their components of ErbB-3—EGFR mutant (averaged), ErbB-3—c-Met, ErbB-3—ErbB-2 and ErbB-3—IGF-1R systems. (i) Total binding free energy comparison among ErbB-3-partner systems. Binding free energies of ErbB-3—c-Met, ErbB-3—ErbB-2 and ErbB-3—IGF-1R systems are labeled with lines, and those of systems involving WT EGFR, L858R, L858R_T790, delE746_A750 and delL747_P753insS are marked with solid spheres. (j) Ranked total binding free energies of representative ErbB-3-partner systems.