FGFR2 Abrogation Intercepts Pancreatic Ductal Adenocarcinoma Development

Abstract

Activating KRAS mutations are a key feature of pancreatic ductal adenocarcinoma (PDAC) and drive tumor initiation and progression. However, mutant KRAS by itself is weakly oncogenic. Defining the pathways that cooperate with mutant KRAS to induce tumorigenesis could identify prevention and treatment strategies. Analyzing organoids and murine and human pancreatic specimens, we found that the receptor tyrosine kinase FGFR2 was progressively upregulated in mutant KRAS-driven metaplasia, precancerous lesions, and classical PDAC. In genetic mouse models, FGFR2 inactivation impeded mutant KRAS-driven transformation of acinar cells by reducing proliferation and MAPK pathway activation. FGFR2 abrogation significantly delayed tumor formation and extended the survival of these mice. Furthermore, FGFR2 collaborated with EGFR, and dual blockade of these receptor signaling pathways significantly reduced mutant KRAS-induced precancerous lesion formation. Together, these data have uncovered a pivotal role for FGFR2 in the early phases of pancreatic tumorigenesis, paving the way for future therapeutic applications of FGFR2 inhibitors for pancreatic cancer interception. Significance: FGFR2 inhibition reduces mutant KRAS signaling, which can impair mutant KRAS-expressing pancreatic cancer precursor lesions that are prevalent in the average healthy adult and delay pancreatic ductal adenocarcinoma progression.

Figure 1 –. Analysis of matched organoid cultures reveals up-regulation of Fgfr2 in pre-invasive compared to invasive KPC cells.

A. PCR-based genotyping of neoplastic-enriched cells from KPC tumors for assessing LOH of the wild-type Trp53 allele. Tumor ‘T34’ organoids and a 2D line derived from the same T organoids are analyzed as controls. B. Immunohistochemical staining for YFP (brown staining) and p53 (blue staining) in a KPCY tumor section. C. PCR-based genotyping for assessing the Trp53 status in neoplastic-enriched cells from KPC tumors after isolation (p0) and upon passaging as organoids (p1 to p4). D. PCR-based genotyping for assessing the Trp53 status in neoplastic-enriched cells from a KPC tumor after isolation (p0) and upon passaging as organoids (p1 to p4) in medium with or without Nutlin. +, positive PCR control E. T^m/LOH organoids are more aggressive in vivo than T^m/+ organoids. The Trp53 status of the organoid lines transplanted was assessed by PCR-based genotyping. Table of primary tumor occurrences. Hematoxylin and eosin (H&E) staining of tumors or pancreata (if no tumor was observed). Scale bars: 100 μm F. PCA of RNA-seq data of T organoid pairs (n = 3). Samples are colored based on their group: T^m/+ in pink and T^m/LOH organoids in blue. G. GSEA signature ‘KEGG p53 SIGNALING PATHWAY’ is repressed in T^m/LOH compared to T^m/+ organoids. NES, normalized enrichment score; FDR, false discovery rate. H. Fgfrs expression as TPM in T^m/+ and T^m/LOH organoids as determined by RNA-seq.

Figure 2 –. FGFR2 is progressively up-regulated in mutant KRAS-driven pancreatic metaplasia, pre-cancerous lesions and human Classical PDA.

A. Left, representative IF for the acinar marker AMY (white), the ductal marker Dolichos Biflorus Agglutinin (DBA) (green), FGFR2 (red) and 4′,6-diamidino-2-phenylindole (DAPI, blue) conducted on healthy pancreata from B6J mice (n = 5) and ADM and mPanINs from KC mice (n = 6). Scale bars, 25 μm. Right, quantification of staining plotted as mean ± SD. 3 images per mouse were quantified. Unpaired Student’s t test. B. Left, representative IF for CK19 (white), FGFR2 (red) and DAPI (blue) conducted on mPanINs from KC mice (n = 7). Scale bars, 25 μm. Right, quantification of staining plotted as mean ± SD. 3 images per mouse were quantified. C. Left, representative IF for CK19 (white), p53 (green), FGFR2 (red) and DAPI (blue) conducted on tumors from KPC mice (n = 6). Scale bars, 25 μm. Right, quantification of staining plotted as mean ± SD. 3 images per mouse were quantified. Unpaired Student’s t test. D. Representative IHC for FGFR2 conducted on human IPMN and PanIN1/2 lesions. E. Left, representative IF for FGFR2 (red), GATA6 (white), S100A2 (green) and DAPI (blue) conducted on human PDAs. Scale bars, 25 μm. Samples that present % GATA6⁺ S100A2⁻ cancerous cells > 60 are classified as ‘Classical’, samples with % GATA6⁻ S100A2⁺ cancerous cells > 30 are classified as ‘Basal-like’ and samples with % GATA6⁺ S100A2⁺ cancerous cells > 5 are classified as ‘IC’. Remaining samples that don’t meet these criteria are classified as ‘other’. Right, quantification of staining plotted as mean ± SD. Unpaired Student’s t test. F. Schematic representation of FGFR2 expression during pancreatic cancer progression in the KRAS^G12D-driven mouse model and in human samples.

Figure 3 -. FGFR2 is repressed by TGFβ signaling.

A. Heatmap of average z-score values of Fgfr2, Fgfr1, Snai1 and Fn1 expression as determined by RT-qPCR in T^m/+ organoids after growth in Complete organoid medium, in Minimal medium or Minimal medium supplemented with TGFβ (n = 3). B. Fgfr2 and Fgfr1 expression as determined by RT-qPCR in T6^m/+ and T23^m/+ organoids after growth in Minimal medium supplemented with different doses of TGFβ or vehicle (n = 3). Results show mean ± SD. Unpaired Student’s t test. C. Violin plots showing the expression of Fgfr2, Epcam, Fgfr1 and Zeb1 in the different populations of KPC pre-invasive and invasive cells (40). D. Representative RNA ISH of Fgfr1 (magenta), Vim (cyan) combined with IF for CK19 (white), FGFR2 (green), p53 (orange) and DAPI (blue) in tumors from KPC mice. Scale bar, 50 μm. E. Left, representative RNA ISH of Cdkn2a/ARF (cyan), Vim (yellow) combined with IF for YFP (white), CK19 (red), FGFR2 (green), p53 (orange) and DAPI (blue) conducted on tumors from KPCY mice (n = 5). Scale bar, 20 μm. Right, quantification of staining plotted as mean ± SD. 4 images per mouse were quantified. Unpaired Student’s t test. F. FGFR2 z-score values in SMAD4 wild-type and altered PDA samples (44). Results show mean ± SD. Unpaired Student’s t test.

Figure 4 -. FGFR2 is dispensable for pancreas recovery following injury.

A. Left, representative H&E staining conducted on pancreatic samples from C Fgfr2^+/+ and Fgfr2^f/f mice at different time points after saline or cerulein treatment (day −1, day 0). Scale bar, 50 μm. Right, quantification of the fraction of normal pancreas (n = 5-6). Results show mean ± SD. Unpaired Student’s t test. B. Left, representative IF for CK19 (red), amylase (white) and DAPI (blue) conducted on pancreatic samples from C Fgfr2^+/+ and Fgfr2^f/f mice at different time points after saline or cerulein treatment (day −1, day 0) (n = 5-6). Scale bars are 500 μm for main images and 50 μm for insets. Right, quantification of staining plotted as mean ± SD. Unpaired Student’s t test.

Figure 5 –. FGFR2 abrogation impedes mutant KRAS-induced transformation of acinar cells by reducing proliferation and MAPK pathway activation.

A. Representative IHC for FGFR2 conducted on pancreatic samples from C and KC mice at different time points after saline or cerulein treatment (day −1, day 0). Scale bar, 50 μm. B. Left, representative bright field images (scale bar, 90 μm), H&E staining (scale bar, 20 μm) and IF for CK19 (green), amylase (magenta) and DAPI (blue) (scale bar, 10 μm) conducted on acinar clusters and associated terminal ductal epithelium from KC Fgfr2^+/+ and Fgfr2^f/f mice at different time points after culture in collagen. Right, quantification of staining plotted as mean ± SD (n = 5). Unpaired Student’s t test. C. Protein expression analysis in acinar clusters and associated terminal ductal epithelium from KC Fgfr2^+/+ and Fgfr2^f/f mice at different time points after culture in collagen, as determined by Western blotting. Loading control, COFILIN. D. GSEA signature ‘HALLMARK KRAS SIGNALING UP’ is increased in acinar-enriched explants from KC Fgfr2^+/+ compared to Fgfr2^f/f mice at day 2 of culture in collagen. NES, normalized enrichment score; FDR, false discovery rate. E. GSEA signature ‘HALLMARK G2M CHECKPOINT’ is increased in acinar-enriched explants from KC Fgfr2^+/+ compared to Fgfr2^f/f mice at day 5 of culture in collagen. NES, normalized enrichment score; FDR, false discovery rate. F, G. Left, representative IF for CLU (yellow), FGFR2 (red), phospho-ERK1/2 (F) or Ki67 (G) (green), and DAPI (blue) conducted on pancreatic samples from KC Fgfr2^+/+ mice at day 7 and 30 after cerulein treatment (day −1, day 0) (n = 5-8). Scale bars are 500 μm for main images and 20 μm for insets. Right, quantification of staining plotted as mean ± SD. Paired Student’s t test.

Figure 6 -. FGFR2 inactivation delays KRAS^G12D-driven pancreatic tumorigenesis.

A, B. Representative H&E staining of the pancreata and highest-grade lesions from 7- to 11-months-old age-matched KC Fgfr2^+/+ and Fgfr2^f/f mice (n = 7) (A) or from age-matched KC Fgfr2^+/+ and Fgfr2^f/f mice 30 days after cerulein treatment (day −1, day 0) (n = 8) (B). Scale bars are 4 mm for pancreas images, 100 μm for insets and 50 μm for highest grade lesions images. S, spleen; D, duodenum; P, pancreas C, D. Quantification of the fraction of diseased pancreas. Results show mean ± SD. Unpaired Student’s t test. E, F. Classification of highest-grade lesions. G. Representative H&E staining of the pancreata and highest-grade lesions from 4-months-old KPC Fgfr2^+/+ and Fgfr2^f/f mice (n = 4). Scale bars are 5 mm for pancreas images and 50 μm for insets. S, spleen; D, duodenum; P, pancreas H. Quantification of the fraction of diseased pancreas. Results show mean ± SD. Unpaired Student’s t test. I. Classification of highest-grade lesions. J. Kaplan–Meier survival curve of percent survival for KPC Fgfr2^+/+ (n = 25) and KPC Fgfr2^f/f (n = 24) mice. Notches represent enrolled mice that died from causes other than tumor (i.e., lymphoma, papilloma). Log-rank Mantel–Cox test. Table of median survival (days).

Figure 7 –. Dual FGFR2 and EGFR inhibition delays mutant KRAS-driven pancreatic tumorigenesis.

A. FGFR2 protein expression in T^m/+ organoids knocked-out for Fgfr2 or control Rosa26 treated with vehicle, gefitinib or erlotinib as determined by Western blotting. Loading control, ACTIN. B. Fold change of growth inhibition upon erlotinib treatment of T^m/+ organoids knocked-out for Fgfr2 or control Rosa26 (n = 5). Results show mean ± SD. Unpaired Student’s t test. C. Fold change of growth inhibition upon erlotinib treatment of T^m/+ Fgfr2^f/f organoids expressing Fgfr2 cDNA or control (n = 6). Results show mean ± SD. Unpaired Student’s t test. D. Protein expression analysis of T^m/+ organoids treated with vehicle, lirafugratinib, erlotinib or the combination as determined by Western blotting. Loading control, COFILIN. E. Relative growth (day 4 / day 1) of T^m/+ organoids treated with vehicle, erlotinib, lirafugratinib or the combination (n = 5). Results show mean ± SD. Unpaired Student’s t test. F. Quantification of the fraction of diseased pancreas in 2-months-old KC mice treated with cerulein for 2 days and the indicated inhibitors for 10 days (n = 6). Results show mean ± SD. Unpaired Student’s t test. G, H. Left, representative IF for CK19 (cyan), CLU (yellow), phospho-ERK1/2 (red) and DAPI (blue) (G) and for CK19 (red), amylase (white), Ki67 (green) and DAPI (blue) (H) conducted on pancreatic samples from KC mice treated with cerulein for 2 days and the indicated inhibitors for 10 days (n = 6). Scale bar, 2 mm. Right, quantification of staining plotted as mean ± SD. Unpaired Student’s t test.