Fig 2: Representative pictures of FGFR2IIIb and FGFR2IIIc expression in gastric cancer. (A) IHC score of FGFR2IIIb and FGFR2IIIC. FGFR2IIIb and FGFR2IIIc were mainly expressed at cell membrane (arrows and inset). The immunoreactivity of FGFR2IIIb and FGFR2IIIc were evaluated according to the intensity of membranous staining score 0–3, as follows: 0, negative; 1 +, weak at cytosol; 2 +, moderate at cell membrane; 3 +, strong at cell membrane. FGFR2IIIb and FGFR2IIIc expression were considered positive when scores were ≥ 2. Bar = 100 μm. (B) Representative cases stained with H&E, FGFR2IIIb, FGFR2IIIc, pan-FGFR2, and CEA of gastric cancer. Heterogeneous expressions of FGFR2IIIb isoform and FGFR2IIIc isoform were found in a tumor (Case 1 and Case 2). The pan-FGFR2 detected both FGFR2IIIb isoforms and FGFR2IIIc isoforms. FGFR2IIIb expression (arrows) and FGFR2IIIc expression (asterisks) were positive in the same tumor, but both were not positive in the same tumor cells. FGFR2IIIc dominant expression in a metastatic lymph node disease was found in Case 1. CEA determined cancer cells in the primary tumor and the lymph node. In contrast, Case 3 shows FGFR2IIIb expression, but not FGFR2IIIc expression. × 100. Bar = 200 μm.

Fig 3: Characterization of the FGFR2-SKI fusion identified in the KCC_P_4043 PDX model. a Integrative Genomic Viewer results for breakpoint regions of chromosome 10 (containing FGFR2) and chromosome 1 (SKI). b Junction break point sequence of FGFR2-SKI fusion. Pink, FGFR2 exon 17; blue, SKI exon 2. c Analysis of FGFR2 in the KCC_P_4043 using SNP arrays. Top three tracks are copy number raw logRatio data from the array; middle three tracks are copy number segmentation from the raw logRatio; bottom three tracks are smoothed copy number signal. The red box highlights FGFR2 on chromosome 10. d Schematic of the FGFR2-SKI fusion in KCC_P_4043. Domain structure and amino acid residues of FGFR2 and SKI are indicated. In FGFR2, IgI–IgIII: immunoglobulin 1–3. TM: transmembrane domain. In SKI, R-smad: corresponding binding domain. DHD: Dachshund homology domain. SAND: Sp100, AIRE1, NucP41/75 and DEAF1. NLS: nuclear localization sequence. The dotted line highlights the junction between FGFR2 and SKI. e Confirmation of FGFR2-SKI expression by Western blotting. Protein lysates from 9 PDX samples were immunoblotted with a FGFR2 N-term antibody. MFM-223 and MDA-MB-468 lysates were used as positive and negative controls, respectively

Fig 4: Crosstalk between FGFR2 and KIT in IM-resistant GISTs. (A) IM-resistant GISTs were treated with IM (1 µmol/L) for 48 h alone (middle panel) or in the presence of BGJ398 (1 µmol/L), a selective FGFR inhibitor (bottom panel), and subjected to the double immunofluorescence staining for FGFR2 and KIT. To outline the nucleus, the images were also merged with DAPI staining. (B) Quantitative analysis of relative fluorescence intensity of FGFR2 and KIT in GIST T-1R cells treated for 72 h with DMSO (control), IM (1 µmol/L) alone, or in the presence of BGJ398 (1 µmol/L). (C,E) Confocal images illustrating the co-localization pattern between FGFR2 (green) and KIT (red) in GIST T-1R cells. Merged image is shown in yellow in (E). Images were also merged with DAPI staining in (C) to outline the nucleus. (D) Pearson’s correlation analysis (Pearson’s R) for FGFR2 and c-KIT fluorescence pattern in IM-treated, control, and IM-naive cells is shown.

Fig 5: Signaling from FGFR2 S780A is prolonged compared to wild-type FGFR2. (a) U2OS-R2 wild-type or U2OS-R2 S780A cells were kept for two hours in serum-free media before addition of 100 ng/mL FGF1 and heparin (20 U/mL) for the indicated periods of time. Cycloheximide (10 µg/mL) was added at the beginning of the starvation period and kept throughout the experiment. After lysis, the cellular material was analyzed by immunoblotting using the indicated antibodies. A p in front of the name of the antibody indicates that it recognizes the phosphorylated form of the protein. Quantifications of five independent experiments are presented in the graph. The bands corresponding to phosphorylated receptor were normalized to γ-tubulin and within each experiment to the time point of 15 min. Error bars denote the standard deviation. The difference between FGFR2 wild-type and the S780A mutant was significant (p ≤ 0.001, 3-way ANOVA, Holm-Sidak test, n = 5). (b) U2OS-R2 wild-type (wt) or U2OS-R2 S780A (m) cells were kept for two hours in serum-free media before addition of indicated concentrations of FGF1 in the presence of heparin (20 U/mL) for 15 min. After lysis, the cellular material was analyzed by immunoblotting using the indicated antibodies. A p in front of the name of the antibody indicates that it recognizes the phosphorylated form of the protein. Quantifications of three independent experiments are presented in the graph. The bands corresponding to the phosphorylated receptor/ERK1/2/PLCγ were normalized to LNF (γ-tubulin). Error bars denote the standard deviation. The difference between FGFR2 wild-type and S780A mutant was significant (** p ≤ 0.01, *** p ≤ 0.001, 3-way ANOVA, Holm-Sidak test, n = 3). (c) Internalization of DL550-FGF1 in FGFR2 and FGFR2 S780A cells is reduced upon U0126 treatment. U2OS-R2 and U2OS-R2 S780A cells were incubated with 200 ng/mL DL550-FGF1 and heparin (50 U/mL) for 30 min. The cells were pretreated as indicated with U0126 (20 µM) for 30 min before addition of FGF1. The cells were then fixed, stained with Hoechst, and analyzed by confocal microscopy. The images were taken at fixed intensity settings, and brightness/contrast was adjusted in the same way for all images. Scale bar: 5 µM. Quantifications of four independent experiments were performed as described in materials and methods and are presented in the graph. In total, 269 U2OS-R2 cells, 262 U2OS-R2 cells treated with U0126, 247 U2OS-R2 S780A cells, and 213 U2OS-R2 S780A cells treated with U0126 were quantified. Outliers were removed according to the 1.5*IQR outlier rule. Error bars denote the SEM (n = 4). Due to a general variation in the intensity between experiments, the means of U0126 treated cells for each cell line in each experiment were normalized to the mean of the corresponding cell line in the same experiment (** p ≤ 0.05, * p ≤ 0.05, two-sided t test, n = 4). (d) In vitro phosphorylation of FGFR2 by active recombinant ERK1. Lysates from U2OS, U2OS-R2, and U2OS-R2 S780A cells were subjected to FGFR2 immunoprecipitation (IP). The immunoprecipitated materials were next incubated with [γ-32P]-labelled adenosine triphosphate and recombinant active ERK1 (rec. ERK1) in the presence of PD173074. After washing, the samples were subjected to SDS-PAGE and analyzed by autoradiography and immunoblotting (IB). One representative experiment is shown.