Paradoxical cancer cell proliferation after FGFR inhibition through decreased p21 signaling in FGFR1-amplified breast cancer cells

Abstract

Fibroblast growth factors (FGFs) control various cellular functions through fibroblast growth factor receptor (FGFR) activation, including proliferation, differentiation, migration, and survival. FGFR amplification in ER + breast cancer patients correlate with poor prognosis, and FGFR inhibitors are currently being tested in clinical trials. By comparing three-dimensional spheroid growth of ER + breast cancer cells with and without FGFR1 amplification, our research discovered that FGF2 treatment can paradoxically decrease proliferation in cells with FGFR1 amplification or overexpression. In contrast, FGF2 treatment in cells without FGFR1 amplification promotes classical FGFR proliferative signaling through the MAPK cascade. The growth inhibitory effect of FGF2 in FGFR1 amplified cells aligned with an increase in p21, a cell cycle inhibitor that hinders the G1 to S phase transition in the cell cycle. Additionally, FGF2 addition in FGFR1 amplified cells activated JAK-STAT signaling and promoted a stem cell-like state. FGF2-induced paradoxical effects were reversed by inhibiting p21 or the JAK-STAT pathway and with pan-FGFR inhibitors. Analysis of patient ER + breast tumor transcriptomes from the TCGA and METABRIC datasets demonstrated a strong positive association between expression of FGF2 and stemness signatures, which was further enhanced in tumors with high FGFR1 expression. Overall, our findings reveal a divergence in FGFR signaling, transitioning from a proliferative to stemness state driven by activation of JAK-STAT signaling and modulation of p21 levels. Activation of these divergent signaling pathways in FGFR amplified cancer cells and paradoxical growth effects highlight a challenge in the use of FGFR inhibitors in cancer treatment.

Keywords: FGF2; FGFR1; JAK-STAT; Stemness; p21.

Fig. 1

FGF2 induced paradoxical growth effects through p21 modulation in FGFR1 amplified and non-amplified ER + cells. A. Spheroid images of FGFR1 amplified cell lines (CAMA1 and MDA-MB-134) and non-amplified cell lines (MCF7 and T47D) show the effects of FGF2 simulation with various doses (5, 25, 125 ng/mL) for 14 days. Bar equals 1000 μm. B. Cell number estimation for images of spheroids in Panel A, with DMSO treatments serving as controls and set to a fold of one. Asterisks (*) indicate that the FGF2 treatment is significantly lower than the DMSO treatment (p < 0.05), while hashtags (#) indicate that the FGF2 treatment is significantly higher than the DMSO treatment (p < 0.05). C. Comparison of the levels of FGFR1 (full length and ICD) and p21 in ER + cell 3D cultures that are either amplified or non-amplified in FGFR1, with or without FGF2 treatment (25 ng/mL) for 72 h. The relative fold change in p21 for each cell line is determined by the ratio of p21 to β-actin in the FGF2 vs. control. D. Immunoblotting shows the expression of p21 in the 3D culture of ER + cells after exposure to various FGF2 dose treatments for 72 h. E. Immunoblotting shows p21 protein levels in 3D cell cultures treated with FGF2 (25ng/mL) plus p21 inhibitor UC2288 (2.0 and 5.0 µM) for 72 h. F. Spheroid images of all four cell lines show the effects of UC2288 (0.2 and 1.0µM) with or without FGF2 treatment (25 ng/mL) for 14 days. Bar equals 1000 μm. G. Relative fold change of ratio between cell number in + FGF2 vs. -FGF2 spheroids in each treatment for the images in Panel F, with the ratios in the DMSO treatment serving as controls and set to a fold of one

Fig. 2

FGF2 caused paradoxical proliferation can be reversed using specific FGFR inhibitors. A. Spheroids images formed by FGFR1 amplified cells (CAMA1 and MDA-MB-134) and non-amplified cells (MCF7 and T47D) when incubated with FGFR inhibitors PD166866 (FGFR1i, 1.0µM), Alofanib (FGFR2i, 1.0µM), H3B-6527 (FGFR4i, 1.0µM), and AZD4547 (FGFR1-3i, 1.0µM) in the presence or absence of FGF2 for 14 days. Bar equals 1000 μm. B. Relative fold change of ratio between cell number in + FGF2 vs. -FGF2 spheroids for each treatment in Panel A. The ratios in the DMSO control group are serving as controls and set as a fold of one. C. Immunoblotting shows the levels of p21 and CCND1 expression, as well as the phosphorylation of Stat1/3 and Erk1/2 in ER + cell 3D cultures that were incubated with specific FGFR inhibitors (1.0µM for all inhibitors) and FGF2 (25ng/mL) for 72 h

Fig. 3

TAS-120 reversed FGF2 induced paradoxical growth effects in FGFR1 amplified and non-amplified ER + BC cells. (A) Spheroid images of FGFR1 amplified cells (CAMA1 and MDA-MB-134) and non-amplified cells (MCF7 and T47D) are shown after incubation with different doses (0.2 and 1.0µM) of TAS-120 and FGF2 (25 ng/mL) for 14 days. Bar equals 1000 μm. (B) Relative fold change of ratio between cell number in + FGF2 vs. -FGF2 spheroids in Panel A are depicted, with the DMSO control group serving as controls and set to a fold of one. C-D. Relative mRNA expression level of FGFR1 (C) and ratio of p21/CCND1 (D) in ER + cell 3D cultures incubated with different doses (0.2 and 1.0 µM) of TAS-120 plus FGF2 (25 ng/mL) for 72 h. The mRNA expression was normalized to RPLP0 and the control treatments, using DMSO, are set to a fold of one. E. Plots that depict the relationship between the expression levels of FGF2 (X-axis) and the CDKN1A/CCND1 ratio (Y-axis) in 601 ER + breast cancer patients from TCGA. The three linear fit curves and 95% C.I. (grey shaded area) demonstrate the interaction effects between FGF2 and FGFR1 expression, where FGFR1 expression levels are grouped into tertiles (high n = 206; low n = 199; med n = 196)

Fig. 4

FGF2 lead to opposite effects on cell cycle in FGFR1 amplified and non-amplified ER + cells. A-D. The proportion of cells in each phase of the cell cycle was evaluated through PI staining-based flow cytometry analysis in FGFR1 amplified cell lines CAMA1 and MDA-MB-134 (A and B) and non-amplified cell lines MCF7 and T47D (C and D). The cells were collected from 2D cultures after being exposed to FGF2 (25ng/ml) with or without TAS-120 (1.0 µM) for 24 hours. The symbol “#’ indicates that the G1 phase in FGF2-treated cells is significantly higher compared to DMSO-treated cells (p < 0.05). The symbol “*’ shows that the G1 phase in FGF2 plus TAS-120 treated cells is significantly lower compared to FGF2-treated cells (p < 0.05). E-G. TCGA ER + breast cancer patients data analysis for interactions in the generalized linear model between FGF2 (X-axes):FGFR1 expression and their impact on GSEA pathways related to M phase (C), G2/M checkpoints (F) and S phase (G) (Y-axes, n = 601). The three linear fit curves and 95% C.I. (grey shaded area) are shown to indicate the interaction effects between FGF2 and FGFR1 expression, where FGFR1 expression levels were grouped by tertiles (high n = 206; low n = 199; med n = 196)

Fig. 5

FGFR1 overexpression upregulated p21 to inhibit cell cycle progression. A. Immunoblotting shows the levels of FGFR1, p21, and stat1/stat3 activation in 2D cultures of FGFR1 amplified cells (CAMA1 and MDA-MB-134) and non-amplified cells (MCF7 and T47D) after transfection with pCMV-XL6-FGFR1 plasmid or an empty pCMV-XL6 vector (as controls). B. Cell cycle phase proportions were determined by PI staining-based cell cycle FACS analysis in all four cell lines cells with FGFR1 and empty vector transfection in 2D culture. A ‘#’ symbol indicates G1 phase in FGFR1 transfected cells significantly higher than in empty vector transfected cells; C. 3D spheroid images are shown for four cell lines with FGFR1 and empty vector transfection after 14 days 3D culture. Bar equals 1000 µm. D. Relative fold change of cell number in FGFR1 transfected vs. vector control spheroids for images in (C). The cells transfected with empty vector are used as controls and set as fold one. E. Time course of cell number change over 14 days culture. for FGFR1 and vector control transfected cell lines. D and F, a ‘*’ symbol indicates the cells transfected with FGFR1 significantly lower than the cells transfected with empty vector (p < 0.05). F. cancer cell growth rate for cells transfected with FGFR1 and vector control. For D and F, a ‘*’ symbol indicates the cells transfected with FGFR1 significantly lower than the cells transfected with empty vector (p < 0.05)

Fig. 6

The JAK2 inhibitor AZD1480 reversed the paradoxical proliferation caused by FGF2. A. Spheroid images show the effects of the JAK1 inhibitor Solcitinib (1.0 µM) and the JAK2 inhibitor AZD1480 (0.2 µM) on FGFR1 amplified cells (CAMA1 and MDA-MB-134) and non-amplified cells (MCF7 and T47D) with and without FGF2 treatment (25 ng/mL) for 14 days. Bar equals 1000 μm. B. Relative fold change of ratio between cell number in + FGF2 vs. -FGF2 spheroids is calculated for each treatment. The ratios in DMSO treatment are used as controls and are set to a fold of one. C. Immunoblotting shows the effect of specific FGFR inhibitors (1.0 µM for all FGFR inhibitors) on p21 expression levels, JAK2 and Stat1/3 activation in ER + cell 3D cultures after 72 h of treatment with FGF2 (25 ng/mL)

Fig. 7

Amplified FGFR1 promotes FGF2 induced cancer stemness through JAK-STAT pathway. A-B. Relative percentage changes of CSL cells/live cells (A) and relative percentage change of total mammosphere area for the 1st round culture (B) are shown in four ER + cell 3D cultures incubated with FGF2 (25ng/mL) plus PD166866 (1.0µM), TAS-120 (1.0µM), AZD1480 (0.2µM), SGC-CBP30 (10µM) and UC2288 (1.0µM). Cells from spheroids were incubated with FGF2 plus inhibitors for 72 h before ALDEFLUOR/CD44 staining or counting and replating for 1st round of mammosphere culture for 7 days. The DMSO treatments are used as controls and set as fold one. C-E. TCGA ER + breast cancer patients data analysis for interactions in the generalized linear model between FGF2 (X-axis):FGFR1 expression and their impact on ssGSEA pathways related to stemness (C-D) and JAK-STAT (E) (Y-axis, n = 601). The three linear fit curves and 95% C.I. (grey shaded area) are shown to indicate the interaction effects between FGF2 and FGFR1 expression, where FGFR1 expression levels were grouped by tertiles (high n = 206; low n = 199; med n = 196)

Fig. 8

Brief figure summarizes the FGF2-FGFR1-p21 regulatory pathway in FGFR1 amplified ER + BC cells. FGF ligands can activate both the classical FGFR MAPK cascade to promote proliferation and a collateral JAK-STAT signaling to upregulate p21, which leads to a growth inhibitory effect and a stem cell-like state. In cancer cells with FGFR1 amplification, FGF ligands addition leads to enhanced upregulation of p21 and stem cell-like state. By using a range of inhibitors target different nodes of FGFR signaling, including FGFR1 (PD186866), pan-FGFR (TAS-120), JAK2 (AZD1480), STAT1 (Fludarabine), STAT3 (BAY2353), CBP/P300 (SGC-CBP30), ERK1/2 (Ulixertinib), and p21 (UC2288), these different states could be reversed