GZD824 overcomes FGFR1‐V561F/M mutant resistance in vitro and in vivo

Abstract Abnormallyactivated FGFR1 has been validated as a therapeutic target for differentcancers. Although a variety of FGFR inhibitors have shown benefit in manyclinical patients with FGFR1 aberration, FGFR1 mutant resistance such as V561Mmutation, has been reported. To date however, no FGFR inhibitors have beenapproved to treat patients with FGFR mutant resistance. Herein, we report that GZD824, athird generation ABL inhibitor (Phase II, China), overcomes FGFR1‐V561F/M mutant resistance in vitro and in vivo. GZD824potently suppresses FGFR1/2/3 with an IC50 value of 4.14 ± 0.96, 2.77 ± 0.082, and 8.10 ± 0.15 nmol/L. It effectively overcomes FGFR1‐V561F/M and other mutantresistance in Ba/F3 stable cells (IC50:8.1–55.0 nM), and effectively inhibits the growth of Ba/F3‐FGFR1‐V561F/M mutantxenograft tumors in vivo (TGI=73.4%, 49.8% at20mg/kg, p.o, q2d). GZD824may be considered to be an effective drug to treat patients with FGFR1 abnormalactivation or mutant resistance in clinical trials.

Herein, we report the activities and mechanisms by which GZD824 can overcome FGFR1-V561F/M mutant resistance in vitro and in vivo, and explore the potential applications of GZD824 in patients with FGFR aberrant activation and mutant resistance. We report that FGFR1-V561F mutation is resistant to erdafitinib, pemigatinib, BGJ398, and TAS120 but is sensitive to GZD824. GZD824 strongly suppresses FGFR1-3 and effectively overcomes FGFR1-V561F/M and other mutant resistance in vitro and in vivo, and may serve as a novel drug in clinical trials of cancer therapy.

| In vitro kinase assays
FGFR1 and the Z′-Lyte kinase assay kit were purchased from Invitrogen (Waltham, MA, USA) and the assays were performed according to the manufacturer's instructions as described previously. 27  FBS, 100 U/ml of penicillin, and 50 mg/ml of streptomycin in 5% CO 2 at 37℃. The cells in the logarithmic growth phase were used for the experiments.

| In vitro cell proliferation Assay
Cells were seeded in 96-well plate in complete medium (1000-6000 cells/well) and cultured for overnight. Then the cells were exposed to gradient concentrations (0.0005-10 μM) of compounds. Cell proliferation was analyzed by Cell Counting Kit 8 (CK04, Dojindo Laboratories, Japan) after co-incubation for a further 72 h. IC 50 values were calculated by fitting of concentration-response curves using GraphPad Prism 8.01 software. Each assay was repeated at least three times. Each IC 50 value is displayed as mean ±SD of at least three independent experiments.

| Cell cycle analysis
Cells were treated with indicated concentrations of GZD824 or DMSO for 24 h. About 6 × 10 5 cells were re-suspended in 150 μl of BD Cytofix/Cytoperm buffer solution I for 10 min at room temperature. The cell suspensions were added with BD Cytofix/Cytoperm buffer solution II and incubated with another 10 min at room temperature, and then incubated with 200 μl of propidium iodide buffer (0.1 mg/ml) in the dark for 10 min at 4℃. Assays were performed with a Guava easy-Cyte flow cytometer (Merck, USA).

| Apoptosis assay
Cells were treated with indicated concentrations of GZD824 or DMSO for 48 h at 37℃. After incubation, cells were collected and washed twice with pre-cold PBS. About 6 × 10 5 cells were resuspended in 100 μl of 1× BD binding buffer solution (#556454, BD), and then stained with 7-ADD (#559925, BD) and Annexin V-PE (#556422, BD) in the dark for 15 min. Finally, 400 μl of 1×BD binding buffer solution were added to stop dyeing. The cells were then measured on a Guava easyCyte flow cytometer (Merck, USA).

| Western blot analysis (WB)
Cells were treated with different concentrations of compound or DMSO for indicated time. Then, cells were lysed with 1 × SDS sample lysis buffer (recommended by CST). After ultra-sonication and boiling, the cell lysates were loaded and then electrophoresed to 10%-12% SDS-PAGE gel, and then the separated proteins were transferred to PVDF membranes. After blocking with 5% BSA in TBST solution (0.5% Tween-20) at room temperature for 4 hrs, the membranes were incubated with the corresponding primary antibody (1:2000-1:200) at 4℃ overnight. After washing with TBST for 3-5 times, the HRP-conjugated secondary antibodies were incubated for 2 hrs. The protein signals were detected with Amersham Imager 600 system (GE, Boston, MA, USA) using an ECL western blotting detection kit (Thermo Fisher Scientific, Waltham, MA, USA).

| Induced fit docking studies
Ligands were built and prepared using LigPrep (Maestro Version 11.9, Schrodinger LLC, New York, NY, USA) with the default settings and the OPLS3e force field. The crystal structure of FGFR1 in complex with ponatinib (PDB code: 4V04) was extracted from the Protein Data Bank (http:// www.pdb.org). The protein was prepared using the Protein Preparation Wizard workflow in Maestro to assign bond orders and add hydrogens. Crystal water molecules and all heteroatom (het) residues except those in ponatinib were removed. Docking studies were performed using the induced fit docking (IFD) protocol of Schrödinger (Schrödinger LLC, New York, NY, USA) software. The centroid of the cocrystal structure ligand ponatinib was used as the center of the grid box. Residues within 5 Å of the ligand poses were refined. The default protocols were used to perform induced fit docking.

| Mouse xenograft tumor models
All animal studies were approved by the Institutional Animal Care and Use Committee. Male CB17-SCID mice were purchased from Vital River Laboratory Animal | 4877 Technology Inc. (Beijing, China). The right flanks of SCID mice were subcutaneously injected with 2 x 10 6 Ba/F3-FGFR1, Ba/F3-FGFR1-V561F, or Ba/F3-FGFR1-V561M cells, respectively. When the average tumor volume reached 100-200 mm 3 , mice bearing tumor were randomly divided into groups and treated with compounds or vehicle. All the compounds were suspended in 0.5% CMC-Na aqueous solution. GZD824 (10 or 20 mg / kg, q2d), BGJ398 (qd), or vehicle (q2d) were administrated by gavage for 12-20 days. The body weight and tumor volume of animals were measured once every 2 days. Tumor volumes were calculated as L×W×W/2, where L and W are the length and width of the tumor, respectively. At the experiment terminal, tumor tissues were dissected, paraffin embedded, and sectioned for hematoxylin and eosin (H&E), immunohistochemistry (IHC), and TUNEL analysis.

| Cellular thermal shift assay (CETSA)
Engagement between FGFR1 protein and GZD824 was assessed by CETSA. For each group, 2 × 10 7 cells were collected and resuspended in PBS after treatment with GZD824 (1 μM for H1581, 100 nM for Ba/F3-FGFR1-V561F) or DMSO 4 h, then the cell suspensions of each group were divided into eight equal parts. Pairs consisting of one experimental aliquot and one DMSO aliquot were heated from 38.0 to 57.8℃ for 3 min. Then, the samples were exposed to liquid nitrogen for 10 min, and then placed in a water bath at 37℃ for 10 min. After three freeze-thaw cycles, cell lysates were centrifuged at 12000 rpm at 4℃ for 10 min, and then boiled for 15 min. The soluble supernatant was used for WB analysis.

| Statistical analysis
Data are expressed as the mean ±SD of three independent experiments. Two-tailed Student's t-test was used for comparison between the two groups, and one-way ANOVA was used for comparison among the multiple groups. Differences with p < 0.01 and p < 0.05 were considered very significant or significant, and marked as ** and *.
A commercial binding affinity assay, KINOMEscan™ was used to evaluate the potential activities of GZD824 against FGFR mutant resistance. It was shown that GZD824 possesses high binding affinities with FGFR1/2/3/4, with K d values of 7.3 ± 0.3, 15.5 ± 3.5, 17.0 ± 1.4, and 19.0 ± 1.4 nM (Table 1), respectively, and also displays strong binding affinities to mutants FGFR2-N550K, FGFR3-G697C, FGFR3-K650E, and FGFR3-V555 M, with K d values of 17.0 ± 5.7, 39.5 ± 7.8, 11.0 ± 0.0, and 19.5 ± 4.9 nM, respectively. We also investigated the binding properties of GZD824 with FGFR1 by a CETSA. The results showed that in the presence of GZD824, the stability of FGFR1 or FGFR1-V561F proteins, but not of the control protein GAPDH, was significantly enhanced ( Figure 1C and 1D). These results show that GZD824 can inhibit the activation of FGFRs and overcome FGFR mutant resistance.
The results revealed that GZD824 exerted strongly antiproliferative activity against the cancer cells harboring FGFR aberrations in vitro, especially in FGFR1-FGFR3 cell lines, with IC 50 values of 4.0-91.0 nM (Table 2). However, the activity of GZD824 against A549 and MDA-MB-231 cancer cells was minor with IC 50 values of >1.0 μM, since no FGFR expression was observed ( Table 2). Under the experimental conditions, the IC 50 value of GZD824 is similar to that of BGJ398, but lower than that of ponatinib.

| GZD824 inhibits phosphorylation of FGFR1 and downstream proteins
To study the mechanism of action of GZD824, the effects of GZD824 on the FGFR1 signaling pathway were investigated in H1581 FGFR1 overexpression , A204 FGFR1 overexpression ,

F I G U R E 1 GZD824 is a pan-FGFR inhibitor. (A) Chemical structure of GZD824; (B) GZD824 inhibits FGFR1/2/3/4 kinase activities in vitro
by FRET-based Z′-Lyte assay; (C) GZD824 possess binding activities with FGFR1 in H1581 lung cancer cells by CETSA; (D) GZD824 possess binding activities with FGFR1-V561F in Ba/F3-FGFR1-V561F cells by CETSA. Cells were treated with DMSO or GZD824 (1000 nM for H1581, 100 nM for Ba/F3-FGFR1-V561F) for 3 h, respectively. Drug-treated cells were heated at gradient temperature from 38℃ to 57.8℃ for 3 min and then lysed, centrifuged to separate the soluble fractions from precipitates. The soluble fractions were analyzed by immunoblotting. Relative band intensities were plotted as a function of temperature followed a sigmoidal trend. The corresponding sigmoidal curve is on the right. GZD824 protected FGFR1 against thermal degradation, suggesting that the compound interacts with the receptor Ba/F3-FGFR1, Ba/F3-FGFR1-V561F, Ba/F3-FGFR1-V561M, and A549 FGFR1 lowexpression by western blotting analysis. As shown in Figure 2, treatment with GZD824 for 4 h revealed dose-dependent inhibition of FGFR1 phosphorylation, the downstream FRS2 and ERK1/2 proteins progressing from 3.7 nM to 300 nM in H1581, Ba/F3-FGFR1, and Ba/ F3-FGFR1-V561F cells. GZD824 displayed no obvious effect on the total level of FGFR1 protein expression in H1581, A204, and A549 cell lines (Figure 2A) but showed an obvious degradation of FGFR1 in all Ba/F3 stable cells at doses of 33.3 nM or higher, with an increased ubiquitination level of FGFR1 ( Figure S2B). Further research validated that the proteasome inhibitor MG132 can rescue FGFR1 degradation induced by GZD824, which supports that this is a proteasome-dependent pathway ( Figure S2A). However, its effects on FGFR1 downstream signal pathway in A549 cells were very weak as a negative control.
Under the same conditions, upon treatment with 300 nM of pemigatinib, TAS120 or BGJ398 in Ba/F3-FGFR1-V561F and Ba/F3-FGFR1-V561M, no effects on activation of FGFR1 and its signal pathway were observed ( Figure S3).

| GZD824 induces G0/G1 phase arrest and apoptosis in Ba/F3-FGFR1-V561F cells
To further identify the efficacy of GZD824, the effects of GZD824 on cell cycle and apoptosis in Ba/F3-FGFR1 and Ba/F3-FGFR1-V561F cell lines were detected by flow cytometry analysis. The results displayed that GZD824 dose-dependently induces G0/G1 phase arrest and apoptosis in Ba/F3-FGFR1 and Ba/F3-FGFR1-V561F. Treatment with 100 nM of GZD824 led to 70.88% and 71.13% G0/ G1 phase arrest after 24 h, and 65.73% and 71.01% apoptosis after 48 h in Ba/F3-FGFR1 and Ba/F3-FGFR1-V561F cells, respectively ( Figure 3A and 3C). Under the same conditions, even treatment with 1000 nM of NVP-BGJ398, pemigatinib, and TAS120 showed a weak effect on cell cycle and apoptosis in Ba/F3-FGFR1-V561F cells ( Figure S5). WB analysis also showed that GZD824 dose-dependently decreases the protein levels of CDK2, cyclin D2, CDK4, cyclin E, and the activating cleavage of caspase-3 and caspase-9 in the above FGFR1 model cells ( Figure 3B and 3D). The antiproliferative activities of the compounds were evaluated using CCK-8 assay. The cells were treated with compound or 0.1% DMSO for 72 h. Each data is expressed as mean ±SD of at least three independent experiments.

| GZD824 as a type II FGFR1 inhibitor overcomes FGFR1-V561F/M mutant resistance
In order to explain the mechanism of sensitivity of GZD824 to the FGFR1-V561M/F mutation, we performed an induced fit docking. The docking results of GZD824 with FGFR1 (PDB code: 4V04) shown in Figure 4 revealed that GZD824 as a type II inhibitor binds to the DFG-out conformation of FGFR1-WT, FGFR1-V561M, and FGFR1-V561F. The 1H-pyrazolo [3,4-b] pyridine core of GZD824 occupies the adenine pocket of FGFR1 kinase and forms hydrogen bond interactions with the backbone of Ala564 in the hinge region of the FGFR. The methylphenyl ring occupies the hydrophobic pocket located behind the gatekeeper residue Val561. The amide moiety is involved in two hydrogen bond interactions with the backbone nitrogen atom of Asp641 from the DFG motif and the side chain of the strictly conserved glutamate of the aC-helix (Glu531), which is characteristic of type II inhibitors. The side chain of Phe642 is extended toward the solvent, creating an induced-fit hydrophobic pocket into which the trifluoromethylphenyl ring binds. The terminal piperazinyl nitrogen atom of GZD824 is protonated and the resulting tertiary ammonium ion forms a hydrogen bond with the carbonyl oxygen atom of residue Ile620 in the catalytic loop. As shown in Figure 4, GZD824 binds to FGFR1-V561M in a very similar fashion to that adopted by FGFR1-WT. Although the 2.8 Å increase in length that occurs upon replacing a valine residue with methionine, GZD824 could participate in favorable van der Waals interactions with the gatekeeper Met without causing steric clashes. In the GZD824-FGFR1 V561F docking complex, the side chain of Phe561 rotates to conserve the electrostatically favorable face-to-face π-π interaction with the methylphenyl ring of GZD824. Ponatinib binds to FGFR1 WT, FGFR1 V561M, and FGFR1 V561F in a very similar fashion to GZD824. BGJ398, pemigatinib, and TAS120 are not type II inhibitors and reasonable binding conformations are not found between these and FGFR1-V561M or FGFR1-V561F.

| DISCUSSION
FGFR1-FGFR4 exhibit various abnormal activations in different tumors, including overexpression, mutation and fusion, drive malignant tumor proliferation, migration, and invasion. These observations suggest that FGFR1-FGFR4 may be potential targets of broad-spectrum targeted therapy for many cancers. The clinical treatment of FGFR2 and FGFR3 abnormal tumors has been successful, and two marketed drugs (erdafitinib and pemigatinib) have been approved to treat those two kinds of tumors, respectively. Although dozens of FGFR inhibitors have been developed, they have not been used successfully in the treatment of patients with abnormal FGFR1/4 activation and are still at the stage of clinical research.
FGFR1 can be considered a therapeutic target for many cancers from different tissues. 28 It has been reported that FGFR1 amplifications emerge in various of cancers, such as NSCLC (12%-17%), prostate cancer (16%), myxofibrosarcoma (20%), bladder cancer (9%), esophageal cancer (9%), colorectal cancer (6%), and in undifferentiated pleomorphic sarcomas (7%). 3,4 FGFR1 fusions such as BCR-FGFR1 29 and FGFR1OP2-FGFR1 30 have been found in acute myeloid leukemia and other cancers. Due to resistance mediated by target mutation or bypass signal activation, treatment of FGFR1 amplified cancers with FGFR inhibitors is currently unsatisfactory. Reported data indicate that cancer cells harboring FGFR1 amplifications can acquire resistance to FGFR inhibitors mediated by overexpression of NRAS, MET amplification, mutational inactivation of PETN, and activation of AKT. 31 Gatekeeper mutations such as FGFR1-V561M (FGFR2-V564F/I, FGFR3-V555M) are, however, the predominant resistant mechanism during treatment with FGFR-targeted therapies. 19 Although many FGFR inhibitors such as erdafitinib, pemigatinib, and TAS120 have been developed, it has been reported that only LY2874455 and ponatinib can overcome FGFR1-V561M mutant resistance.
The gatekeeper mutation could generate a steric clash, or eliminate critical hydrogen bonds required high-affinity binding preventing inhibitors binding. Crystal structure of FGFR1-V561M displayed that this gatekeeper mutation can stabilize the hydrophobic spine to promote catalytic activation, which was validated by FGFR1-V561M conferring a 38-fold increase in k cat relative to FGFR1-WT by a kinase kinetic assay. 32 FIN-2, FIN-3, and TAS-120, which have been designed to form a covalent bond with FGFR kinases to overcome FGFRs mutant resistance, were all successful in overcoming FGFR2-K659N, C491A, and other mutations but not FGFR2-V564F in cells or patients. In this paper, we first report the sensitives of FGFR1-V561F to FGFR inhibitors which similar to V561M are insensitive to pemigatinib, erdafitinib, BGJ398, and TAS120. We were surprised to find that GZD824 can effectively overcome V561F mutant resistance in cells and animals with an IC 50 lower than that of ponatinib. Our induced fit docking assays suggested that GZD824 possess a different binding model compared to those of type I inhibitors, which bind to FGFR1-V561F/M in a very similar fashion to FGFR1-WT, a type II FGFR1 inhibitor.
It has been reported that BGJ398 effectively inhibits the activation of ERK but not AKT in FGFR1-amplified lung cancer cell lines. 1 In this research, GZD824 was found to exert properties similar to those of BGJ398 in H1581 and A204 cells, which displays obvious inhibition of pFRS2 and pERK at a dose of 3.7 nM or higher, but with no effect on pAKT, even at a dose of 300 nM. In FGFR2 overexpressed SNU-16 and in NCI-H716 cells, GZD824, at levels from 3.7 to 300 nM can strongly inhibit pAKT. We suggest these differences may be due to the different backgrounds of the cell lines.
In summary, we report GZD824 as a pan-FGFR inhibitor which inhibits the signaling pathways of FGFR1 kinases and suppresses the proliferation of cancer cells harboring overexpression or mutant activation of FGFR1 in vitro and in vivo. GZD824 overcomes FGFR1-V561F, V561M, K656E, and K656N mutant resistance, and effectively inhibits the growth of Ba/F3-FGFR1-V561F/M mutant xenograft tumors in vivo. F I G U R E 4 Binding mode analysis of GZD824 to FGFR1 WT , FGFR1 V561M , and FGFR1 V561F . (A) Molecular docking of GZD824 into FGFR1 WT (generated from PDB code 4V04). (B) Molecular docking of GZD824 into FGFR1 V561M (generated from PDB code 4V04). (C) Molecular docking of GZD824 into FGFR1 V561F (generated from PDB code 4V04). (D) Overlay of the binding conformation of GZD824 from the FGFR1 WT , FGFR1 V561M , and FGFR1 V561F , respectively GZD824 may be considered to be an effective drug to treat patients with FGFR1 abnormal activation or mutant resistance in clinical trials.