We have shown that the gatekeeper valine
We have shown that the gatekeeper valine to methionine substitutions in FGFR1 and FGFR3 resulted in reduction of sensitivity to selective inhibitors PD17307 and AZD4547; two other tested inhibitors, AP24534 and TKI258, retained their efficacy (Figs. 1, 2 and 4, Table 1). While the multi-kinase, type II inhibitor AP24534 (ponatinib) has been specifically developed to target the gatekeeper mutation in the BCL-ABL fusion protein occurring in response to imatinib treatment of chronic myeloid leukaemia (O\'Hare et al., 2009; Choi et al., 2010a), TKI258 (dovitinib) was originally developed as a novel class of RTK inhibitors targeting FLT3, KIT, VGFR and some PDGFR members, in addition to FGFR1-3 (Renhowe et al., 2009). TKI258 has been and continues to be tested in clinical trials including malignancies linked to FGFR genetic alterations (breast cancer, renal cell carcinoma and p97 cancer) and its other targets (such as KIT in gastrointestinal stromal tumours/GIST) (Andre et al., 2013; Konecny et al., 2013; Mazzola et al., 2014; Kang et al., 2013). Our findings that TKI258 can inhibit FGFR kinases incorporating gatekeeper substitutions (Fig. 1, 2 and 4) suggest that it could be also considered as a second-line treatment following failure of imatinib in GIST patients because one of the frequent causes of such failure is the intrinsic resistance mutation in the related KIT target, affecting the gatekeeper residue (Tamborini et al., 2006). Our data, including previously unavailable high-resolution crystal structure of TKI258 bound to a protein kinase, show that TKI258 is a type I inhibitor and binds to native FGFR1 without causing any structural alterations (Fig. 2B and C). Similar considerations apply to binding to FGFR3. When compared to type I FGFR1–3 selective inhibitors and type II inhibitor AP24534, TKI258 occupies a smaller region confined within the adenosine-binding pocket (Supplemental Fig. S6). Selective inhibitors PD173074, AZD4547 and NVP-BGJ398 (the latter two under clinical trials) share the 3,5-dimetoxy-phenil moiety and as shown for PD173074 and NVP-BGJ398 have closely overlapping binding pockets (Supplemental Fig. S6). Interestingly, FGFR1–4 are among only 15 out of 490 human protein kinases that have valine as the gatekeeper and the selective inhibitors are likely to involve an interaction with this uncommon residue (for example, V561 in FGFR1 interacts with PD173074 and NVP-BGJ398, Supplemental Fig. S6). Thus, it is possible that a shared single point mutation such as gatekeeper replacement would confer some degree of resistance to all these inhibitors. Despite a weaker binding to a phosphorylated, active form, more pronounced for the gatekeeper variant, TKI258 retains its inhibitory activity towards the gatekeeper replacement in FGFR1 and FGFR3 in vitro and in cells (Figs. 1, 2 and 4). Furthermore, the availability of the high-resolution crystal structure with TKI258 (Fig. 2B and C) can guide further modifications to increase efficacy and selectivity for FGFR while retaining activity towards the gatekeeper mutation. Therefore, TKI258 or its derivatives may provide therapeutic opportunity not only for patients with genetic alterations in FGFR but also for those with acquired resistance to several selective FGFR-inhibitors. Recently, a modification of the FGFR-specific inhibitor PD173074, has been reported that covalently targets a cysteine residue in the P-loop (Tan et al., 2014; Huang et al., 2015). The resulting compounds, in particular FIIN2 and FIIN3, proved effective towards mutations of the gatekeeper residue in all FGFR receptors (Tan et al., 2014). Compared to TKI258 the FIIN compounds occupy a space in the binding pocket that is closer to the side-chain residues of the gatekeeper replacements; however, due to its irreversible binding capability and additional stabilization FIIN retains efficacy towards these gatekeeper variants (Supplemental Fig. S7). Further development of FIIN compounds provides an alternative route towards effective second-line treatments.