Osimertinib confers potent binding affinity to EGFR kinase domain duplication

Dear editor, Epidermal growth factor receptor (EGFR) kinase domain duplication (KDD) is a rare form of EGFR alteration in nonsmall-cell lung cancer (NSCLC) with a prevalence of 0.12% (13/10,759) in East Asian patients.1 When with this aberration, in-tandem and in-frame duplications occurs in exons which encode the EGFR kinase domain, and EGFR signaling is activated by the forming of intramolecular asymmetric dimers.2 EGRF KDD, first reported in 2015, is now recognized as one of oncogenic drivers for lung cancer. KDD occurring in exon 18–25 is the most common form, as the unusual events in exon 17–25 and exon 14–26 were also reported.1 Limited data have shown that NSCLCs with EGRF-KDD displayed sensitivity to several EGRF tyrosine kinase inhibitors (TKIs), but the efficiency varied, with a total of 6 out of 11 cases achieving partial response as reported.1-7In vitro study demonstrated that erlotinib, afatinib and osimertinib preserved the antitumor response by inhibiting EGFR-KDD tyrosine phosphorylation in a dose-dependent manner.2 The reported progression-free survival (PFS) of EGFR-KDD to TKIs from literature was ~6 months.1-4 Interestingly, the duration of response to first-generation TKIs varied greatly. The best responses were achieved on gefitinib with a 6-year PFS.3 On the other hand, the effectiveness of third-generation EGFR-TKI, osimertinib, also appears to be encouraging from pilot study, in which durable response was observed after resistance to first and second-generation EGFR-TKIs.1 Thus, the underlying mechanism is worthy in-depth investigation. We computationally constructed the three-dimensional structure of EGFR-KDD based on the crystal structure of EGFR wild-type (WT; PDB ID:2GS6). Generally, the two kinase domains of EGFR-KDD linked by a loop adopt an asymmetric position as previously reported (Fig. 1a).8 Comparing with the crystal structure of EGFR-WT, the modeled structure of EGFR-KDD shows no significant difference in the activated kinase domain, especially in the adenosine triphosphate (ATP)-binding site including hinge region, helixC, P-loop and activation loop (Fig. 1b). However, when we analyzed the binding of TKIs in the context of EGFR-KDD model with molecular dynamic (MD) simulation, and compared to that of EGFR-WT, we observed significant structural deviation in the first-generation TKI (gefitinib) with a root-mean-square deviation (RMSD) of 4.895 å, followed by the second-generation drug afatinib (RMSD: 2.279 å). In contrast, the third-generation TKI osimertinib was found to occupy the ATP-binding site of EGFR-KDD most stably among all these drugs (RMSD: 1.347 å). We also calculated the binding energy for these small-molecule inhibitors in the context of either EGFR-KDD or EGFR-WT with molecular mechanic-Poisson–Boltzmann surface area (MM-PBSA) method.9 Consistent with structural observations, the most significant difference was observed in the binding energy of gefitinib, increased from −32.21 ± 5.07 kcal/mol (EGFR-WT) to −18.51 ± 5.27 kcal/mol (EGFR-KDD), suggesting the reduced binding affinity of gefitinib against EGFR-KDD. In contrast, comparable binding energy was found for osimertinib, −26.89 ± 4.51 kcal/mol for EGFR-WT and −28.49 ± 4.49 kcal/mol for EGFR-KDD, and small increase was found for afatinib, −37.01 ± 5.62 kcal/mol for EGFR-WT and −30.47 ± 4.71 kcal/mol for EGFR-KDD. It should also be noted that there is additional contribution of covalent bond in both afatinib and osimertinib, which would further facilitate the target binding to EGFR-KDD. Collectively, these computational results provide the first evidence to support that osimertinib, compared to the first generation TKI (e.g., gefitinib), is more favorably to bind with and thus inhibit EGFR-KDD. However, given the limitation of currently used computational method, it was hard to make comparisons between the second-generation TKI (e.g., afatinib) and the third-generation TKI (e.g., osimertinib), both of which serve as covalent binders toward EGFR-KDD and WT, and such interaction was not considered in the present study. Yours sincerely, Rui Jin Jie Li Zhou Jin Yuefei Lu Yang W. Shao Wen Li Guofang Zhao Yang Xia Data are available on request from the authors.