Fibroblast growth factor-2–mediated FGFR/Erk signaling supports maintenance of cancer stem-like cells in esophageal squamous cell carcinoma

FGF-2 has a pivotal role for regulation of CSCs. The present study show for the first time that the Ras/Erk pathway is the primary downstream effector of FGFR signaling with regard to regulation of CSCs in ESCC.

Abstract

In esophageal squamous cell carcinoma (ESCC), a subset of cells defined by high expression of CD44 and low expression of CD24 has been reported to possess characteristics of cancer stem-like cells (CSCs). Novel therapies directly targeting CSCs have the potential to improve prognosis of ESCC patients. Although fibroblast growth factor-2 (FGF-2) expression correlates with recurrence and poor survival in ESCC patients, the role of FGF-2 in regulation of ESCC CSCs has yet to be elucidated. We report that FGF-2 is significantly upregulated in CSCs and significantly increases CSC content in ESCC cell lines by inducing epithelial–mesenchymal transition (EMT). Conversely, the FGFR inhibitor, AZD4547, sharply diminishes CSCs via induction of mesenchymal–epithelial transition. Further experiments revealed that MAPK/Erk kinase (Mek)/extracellular signal-regulated kinases (Erk) pathway is crucial for FGF-2–mediated CSC regulation. Pharmacological inhibition of FGF receptor (FGFR)–mediated signaling via AZD4547 did not affect CSCs in Ras mutated cells, implying that Mek/Erk pathway, downstream of FGFR signaling, might be an important regulator of CSCs. Indeed, the Mek inhibitor, trametinib, efficiently suppressed ESCC CSCs even in the context of Ras mutation. Consistent with these findings in vitro, xenotransplantation studies demonstrated that inhibition of FGF-2–mediated FGFR/Erk signaling significantly delayed tumor growth. Taken together, these findings indicate that FGF-2 is an essential factor regulating CSCs via Mek/Erk signaling in ESCC. Additionally, inhibition of FGFR and/or Mek signaling represents a potential novel therapeutic option for targeting CSCs in ESCC.

Introduction

Esophageal squamous cell carcinoma (ESCC) is the eighth most common cancer and the sixth leading cause of cancer death worldwide (1). Surgical resection, chemoradiotherapy and endoscopic treatments are standard curative therapies used for patients with early stage ESCC (2–5). However, in patients presenting with advanced stage ESCC, the effects of conventional therapy remain limited and patient prognosis is still not satisfactory. Although drugs targeting tyrosine kinases, such as epidermal growth factor receptor (EGFR) and human EGF receptor 2 (HER2), have been developed and widely used for various types of cancers including lung, breast, stomach and colon cancer as well as leukemia (6), molecular targeted drugs have not yet to be approved for ESCC therapy.

Cancer stem-like cells (CSCs) have been identified in as subsets of tumor cells in various types of cancers where these cells play essential roles in tumor initiation, maintenance, growth and progression. CSCs possess self-renewal, differentiation and migratory capabilities which may facilitate drug resistance, metastasis, recurrence and poor prognosis (7, 8). In ESCC, various markers have been used to identity CSCs, including CD44, CD24, CD90, CD271 and Aldehyde dehydrogenase 1 (ALDH1) (18, 19). Among these markers, we and other researchers have previously demonstrated that CD44^high/CD24^low cells possess a mesenchymal phenotype coupled with high tumorigenicity and resistance to anticancer drugs in vitro and in vivo, supporting CD44^high/CD24^low cells as a CSC population (9–11, 17). We have reported that epithelial–mesenchymal transition (EMT) is critical to enrich and maintain CSCs in ESCC. Furthermore, EGFR inhibitors suppress EMT-mediated CSC expansion in an organotypic 3D culture system of ESCC (10). EGFR inhibition, however, has no effect upon preexisting CSCs. Thus, a novel therapy directly targeting CSCs has great potential to improve prognosis of advanced stage ESCC patients.

Fibroblast growth factor-2 (FGF-2) also known as basic FGF binds to FGF receptors (FGFR) 1-4 to mediate downstream signaling through pathways such as Ras/Erk and phosphoinositide-3 kinase (PI3K)/Akt signaling and regulate cellular proliferation, differentiation and migration (12). FGF-2 also has critical roles in tumor progression and malignancy (13). In ESCC, FGF-2 has been identified as a significant factor related to recurrence and poor survival rate (14). Presently, the detailed roles of FGF-2 in CSCs and ESCC have yet to be fully elucidated.

In the present study, we report that FGF-2–mediated Mek/Erk signaling is an essential factor regulating CSCs in ESCC. Additionally, pharmacological inhibition of FGFR or Mek signaling induces mesenchymal–epithelial transition (MET) to efficiently eliminate preexisting CSCs. These findings highlight FGFR and Mek inhibitors as potential novel therapeutic options to target CSCs in ESCC.

Materials and methods

Cell lines and reagents

Esophageal cell carcinoma cell lines were obtained from Cell Resource Center for Biomedical Research, Tohoku University. The cells were authenticated by short-tandem repeat-PCR method. TE8 and HCE4 cells were grown in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 100 units/ml of penicillin G and 100 µg/ml of streptomycin. T-TeRAS cells, transformed esophageal epithelial cells with hTERT, SV40 Large T antigen and mutant Ha-Ras^V12 (27, 28) were cultured in Keratinocyte-SFM (Thermo Fisher Scientific, Waltham, MA) and DMEM (10% FBS) at a ratio of 1:1. Recombinant human FGF-2 was purchased from Miltenyi Biotec (Bergisch Gladbach, Germany). AZD4547 and trametinib were purchased from Cayman Chemical (Ann Arbor, MI). GDC0941 was purchased from Santa Cruz biotechnology (Santa Cruz, CA).

Real-time RT-PCR

RNA was extracted by RNeasy mini kit (Qiagen, Valencia, CA) and cDNA was synthesized using Prime Script^TM RT Master Mix (TaKaRa, Shiga, Japan) according to the manufacturer’s instruction. Real-time RT-PCR was done with SYBR® Green (TaKaRa) and Step One real-time PCR system (Applied Biosystems, Foster City, CA) with following primers; 5′-TTGCCGACAGGATGCAGAA-3′ and 5′-GCCGATCCACACGGAGTACT-3′ for β-Actin, 5′-TAAGGACACCCCAAATT CCA-3′ and 5′-ACTGCAATGCAAACTGCAAG-3′ for CD44, 5′-TCCCTGCTACC AGAGACCAA-3′ and 5′-ACAGCTCCATTGCCACTGTT-3′ for CD44s, 5′-AGC AACTGAGACAGCAACCA-3′ and 5′-CCGTGGTGTGGTTGAAATGG-3′ for CD 44v, 5′-GAATGACAACAAGCCCGAAT-3′ and 5′-ACCTCCATCACAGAGGTT CC-3′ for CDH1, 5′-AGGCGAGGAGAGCAGGATTT-3′ and 5′-AGTGGGTA TCAACCAGAGGGA-3′ for vimentin, 5′-GCTGTACTGCAAAAACGGGG-3′ and 5′-TAGCTTGATGTGAGGGTCGC-3′ for FGF-2. Primers for CD24 (QuantiTect primer QT00216811) were purchased from Qiagen (Hilden, Germany). β-Actin was used as an internal control.

Western blotting

Western blotting was done as described previously (26) using the following primary antibodies at the indicated titers: anti–β-Actin (1:2000), anti-Akt (1:1000), anti–phospho-Akt (Ser473, 1:1000), anti-Erk (1:1000) and anti–phospho-Erk (Thr202/Tyr204, 1:1000). These antibodies were purchased from Cell Signaling Technology (Danvers, MA).

Enzyme-linked immunosorbent assay (ELISA)

Sorted CD44^High/CD24^Low cells and CD44^High/CD24^High cells (TE8 and HCE4) were cultured in DMEM (10% FBS) for 48 h. Then, each conditioned medium was collected and centrifuged at 2000 rpm for 5 min. The supernatant was used for detection of FGF-2. FGF-2 concentration was measured using FGF basic (FGF-2) Human ELISA Kit (ab99979, Abcam, Cambridge, MA), following the manufacturer’s instruction.

Soft-agar colony formation assay

Soft-agar colony formation assays were done as described previously (27). In brief, 1 × 10³ cells were suspended in 0.67% agarose containing media and overlaid on top of a 1% agarose containing the medium per well (24 wells plate). 500 µl of medium with indicated reagents was added twice a week in each well and grown for 2 weeks. Colonies over 100 µm were counted. Experiments were done in hexaduplicate.

Flow cytometry and fluorescence activated cell sorting (FACS)

FACSCantoII (BD Biosciences, San Jose, CA) was used for flow cytometry. Cells were suspended in Hank’s balanced salt solution containing 1% BSA (Sigma-Aldrich, Saint Louis, MI) and stained with PE/Cy7–anti-CD24 (Bio Legend, San Diego, CA) and APC–anti-CD44 at 1:40 (BD Biosciences, Franklin Lakes, NJ) on ice for 30 min. Dead cells were eliminated by 7-AAD (BD Biosciences) according to the manufacturer’s instruction. To purify CD44^High/CD24 ^low CSCs and CD44^High/CD24^High non-CSCs, cells were sorted by FACSAriaIII (BD Bioscience). ALDH activity was measured by using ALDEFLUOR® (Stem Cell Technologies, Vancouver, Canada) according to the manufacturer’s instruction.

Drug sensitivity assay (WST-8 assay)

5 × 10³ cells were seeded per well in a 96 wells plate. 24 h after seeding, drugs were added at following concentrations (5-Fluorouracil (5-FU): 5 µM, AZD4547: 2 µM, trametinib: 2.5 µM). After 72 h, WST-8 assay was carried out using Cell Counting Kit-8 (Dojindo, Kumamoto, Japan) according to the manufacturer’s instruction. Absorbance (A₄₅₀–A₆₃₀) was measured by Infinite® F200-HPHN plate reader (TECAN, Männedorf, Switzerland) and survival rate was determined.

Xenograft experiment in vivo

6-week-old female athymic BALB/C slc nu/nu mice were obtained from Japan SLC (Hamamatsu, Japan). Mice were treated in accordance with the institutional animal welfare guidelines of Hokkaido University. Cells were resuspended in Matrigel/DMEM (1:1) at 4 × 10⁶ cells/100 µl and injected subcutaneously into the four sites on the dorsal flanks of mice. Two weeks after cell injection, each drug solution in DMSO/PBS (1:1) was injected intraperitoneally at a volume of 200 µl/mouse (AZD4547: 5 mg/kg, trametinib: 2 mg/kg) every other day. Each drug was administrated from day 18 to 28. Mice were killed at day 30 (TE8) or 33 (T-TeRAS). Tumor size was measured weekly using digital calipers and tumor volume was calculated using the formula; Volume = ab²/2 (a = major axis, b = minor axis).

Immunohistochemistry (IHC)

Hematoxylin and Eosin (H&E) staining and IHC were performed as described previously (20). Deparaffinized sections were heat treated with antigen retrieval solution (Target Retrieval Solution, pH 9.0; Dako, Santa Clara, CA) at 95°C for 20 min using the Dako PT Link system. Then sections were incubated with anti–phosphorylated-FGFR1 antibody (phospho Y653+Y654, ab111124, Abcam) (1:50 overnight at 4°C), anti–phosphorylated-Erk (Phospho-p44/42MAPK, #4370, Cell Signaling Technologies) (1:200 for 60 min at room temperature), anti-vimentin (ACR048A, Biocare Medical, Pacheko, CA) (1:300 for 30 min at room temperature) and anti–E-cadherin (M3612, DAKO) (1:100 for 60 min at room temperature). Detection was performed using a standard polymer method according to the manufacturer’s instructions (EnVision Flex system, Dako).

Statistical analysis

In vitro experiments, results are presented as mean ± SD and were analyzed by 2-tailed Student’s t-test. For In vivo experiments, tumor volumes are presented as mean ± SEM and were analyzed by Tukey’s multiple comparison test. P < 0.05 was considered statistically significant for all studies. All analyses were performed using GraphPad Prism version 6 (GraphPad Software, San Diego, California).

Results

FGF-2 expression upregulated in CD44^High/CD24^Low ESCC CSCs

To examine expression level of FGF-2 in ESCC CSCs, we selected TE8 and HCE4 ESCC cell lines with mesenchymal characteristics and high expression of CD44. We sorted cells with high expression of both CD44 and CD24 (CD44^High/CD24^High) as well as those with high expression of CD44 and low expression of CD24 (CD44^High/CD24^Low). We have previously reported that CD44^High/CD24^Low cells have properties of CSCs in ESCC (10, 36). Several passages after cell sorting, expression levels of CD24 and CD44 cells were confirmed by FACS (Figure 1A). As we reported previously, CD44^High/CD24^Low CSCs have mesenchymal phenotype and CDH1 mRNA expression was significantly suppressed in CSCs (Figure 1B) and expression of E-cadherin was sharply decreased in CSCs confirmed by WB (Figure 1C). FGF-2 mRNA expression was significantly upregulated in CD44^High/CD24^Low cells isolated from TE8 and HCE4 cell lines (Figure 1D). We further determined concentration of FGF-2 by ELISA. Consistent with the result of real-time RT-PCR, FGF-2 was significantly more secreted in CD44^High/CD24^Low CSCs (Figure 1E).

Figure 1.

FGF-2 is upregulated in CD44^High/CD24^Low CSCs. (A) TE8 and HCE4 cells were stained with anti-CD44 and anti-CD24. Cells expressing high CD44 and low CD24 (CD24L; CD44^High/CD24^Low) or high CD44 and high CD24 (CD24H; CD44^High/CD24^High) were sorted by FACS. Expression levels of CD24 and CD44 were analyzed by flow cytometry after two passages (HCE4) and four passages (TE8) following sorting. (B) Expression of CDH1 was determined by real-time RT-PCR in sorted TE8 and HCE4 cells. (*P < 0.05 versus CD24H). (C) Expression of E-cadherin was determined by WB in sorted TE8 and HCE4 cells. (D) Expression of FGF-2 was determined by real-time RT-PCR in sorted TE8 and HCE4 cells. (*P < 0.05 versus CD24H cells). (E) FGF-2 concentration was determined by ELISA. (*P < 0.05 versus CD24H).

FGF-2 increases CD44^High/CD24^Low ESCC CSCs

To investigate the role of FGF-2 in regulation of CSCs, we sorted CD44^High/CD24^High non-CSCs and treated the sorted non-CSCs with FGF-2, then analyzed CD24 and CD44 expression by FACS, qRT-PCR and WB. FGF-2 significantly increased CD44^High/CD24^Low CSCs (Figure 2A and B). qRT-PCR analysis further confirmed suppression of expression of CD24 with concurrent upregulation of CD44 in the FGF-2–treated cells (Figure 2C). It has been reported that a class switch from variant isoforms (CD44v) to the standard isoform (CD44s) occurs during EMT (40). FGF-2 significantly downregulated CD44v and upregulated CD44s (Figure 2C). FGF-2 also suppressed expression of CDH1 (E-cadherin), an epithelial marker, while also increasing expression of vimentin, a mesenchymal marker (Figure 2C). Although upregulation of vimentin was not confirmed by WB, E-cadherin was sharply suppressed by FGF-2 (Figure 2D). ALDH activity, a hallmark of CSCs, was also increased by FGF-2 (data not shown). Phosphorylation of Akt and Erk, two downstream effectors of FGFR signaling, was also confirmed in response to FGF stimulation (Figure 2E). Then, to address the role of FGF-2 in anchorage-independent growth, we carried out soft-agar colony formation assay. In response to FGF-2, displayed enhanced colony formation capability (Figure 2F). These results indicate that FGF-2 induces EMT and upregulation of ESCC CSCs.

Figure 2.

FGF-2 stimulation increases CD44^High/CD24^Low ESCC CSCs. (A–F) CD44^High/CD24^High cells were sorted from TE8 or HCE4 cells, and the sorted cells were treated with or without FGF-2 (50 ng/ml) for 48 h. (A and B) Expression levels of CD24 and CD44 were analyzed by flow cytometry. (C) Expression levels of CD44, CD24, CDH1 and vimentin were determined by real-time RT-PCR in the sorted TE8 cells (*P < 0.05 versus control cells). (D) Expression levels of E-cadherin and vimentin were determined by WB in the sorted TE8 cells. (E) TE8 cells were preincubated with serum-free medium overnight then treated with or without FGF-2 (50 ng/ml) for 15 min. Levels of phosphorylated and total Erk1/2, Akt were examined by Western blotting with β-Actin as a loading control. (F) Anchorage-independent growth was examined by soft-agar colony formation assay with or without FGF-2 (50 ng/ml) for 2 weeks in TE8 cells (*P < 0.05 versus control).

FGFR inhibition decreases CD44^High/CD24^Low ESCC CSCs

Since FGF-2 was upregulated in CD44^High/CD24^Low CSCs and FGF-2 significantly increased CSCs in ESCC, we hypothesized that FGFR inhibition could be used as a therapeutic reagent to target CSCs in ESCC. To test this hypothesis, we examined the effect of the FGFR inhibitor AZD4547 upon CSCs in ESCC cell lines. We sorted CD44^High/CD24^Low CSCs from TE8 and HCE4 cell lines. AZD4547 depleted CD44^High/CD24^Low CSCs in TE8 and HCE4 cells (Figure 3A and B). Consistent with the results of flow cytometric analysis, AZD4547 treatment augmented expression of CD24 and diminished expression of CD44 in CD44^High/CD24^Low CSCs (Figure 3C). AZD4547 significantly suppressed CD44s and increased CD44v (Figure 3C). AZD4547 also enhanced CDH1 expression while depleting expression of vimentin (Figure 3C). Consistent with the results of real-time RT-PCR, E-cadherin was upregulated and vimentin was sharply suppressed by AZD4547 confirmed by WB (Figure 3D). Moreover, soft-agar colony formation was decreased by AZD4547 (Figure 3E). Effects of AZD4547 on downstream of FGFR signaling were evaluated by Western blot, revealing that while phosphorylation of Erk was significantly suppressed by AZD4547 both in TE8 and HCE4 cells, phosphorylation of Akt was suppressed only in TE8 cells (Figure 3F and G), indicating that Ras/Erk signaling might have more important roles in regulating ESCC CSCs as compared to PI3K/Akt signaling.

Figure 3.

Pharmacological FGFR inhibition decreases CD44^High/CD24^Low ESCC CSCs. (A–G) CD44^High/CD24^Low cells were sorted from TE8 or HCE4 cells, and the sorted cells were treated with or without AZD4547 (2 µM) for 48 h. (A and B) Expression levels of CD24 and CD44 were analyzed by flow cytometry. (C) Expression levels of CD44, CD24, CDH1 and vimentin were examined by real-time RT-PCR in the sorted TE8 cells (*P < 0.05 versus DMSO control). (D) Expression levels of E-cadherin and vimentin were determined by WB in the sorted TE8 cells. (E) Anchorage-independent growth was examined by soft-agar colony formation. TE8 cells were treated with AZD4547(2 µM) or DMSO control for 2 weeks (*P < 0.05 versus DMSO control). (F and G) TE8 and HCE4 cells were preincubated with serum-free medium overnight and then treated with AZD4547 (2 µM) or DMSO control for 30 min. Levels of phosphorylated and total Erk1/2, Akt were examined by Western blotting with β-Actin as a loading control (AZD; AZD4547).

Ras/Erk signaling is essential for maintaining CD44^High/CD24^Low ESCC CSCs

To elucidate the role of downstream effectors of FGFR signaling in regulation of ESCC CSCs, we utilized trametinib or GDC0941 to inhibit Mek and PI3K inhibitor, respectively (Figure 4A). We sorted CD44^High/CD24^Low CSCs from TE8 and HCE4 cells and treated the sorted cells with the inhibitors. As expected, trametinib suppressed Erk phosphorylation and GDC0941 suppressed Akt phosphorylation with no detectable cross-reactivity between these inhibitors (Figure 4B). Trametinib dramatically suppressed CD44^High/CD24^Low CSCs, whereas PI3K inhibitor GDC0941 has no effect on CD44^High/CD24^Low CSCs (Figure 4C and D). Trametinib also promoted expression of CDH1 and suppressed expression of vimentin (Figure 4E and F), whereas GDC0941 suppressed expression of CDH1 and increased expression of vimentin. These results suggest that Ras/Erk signaling but not PI3K/Akt signaling might be essential for FGF-2–mediated CSC maintenance.

Figure 4.

Mek inhibition but not PI3K inhibition depletes CD44^High/CD24^Low ESCC CSCs. (A) Schematic of FGF-2/FGFR pathway. Downstream of FGFR, Ras/Erk and PI3K/Akt pathways are activated in response to FGF-2 stimulation. AZD4547 blocks FGFR, trametinib blocks Mek and GDC0941 blocks PI3K. (B) TE8 cells were preincubated with serum-free medium overnight and then treated with trametinib (2.5 µM), GDC0941 (1 µM) or DMSO control for 30 min. Levels of phosphorylated and total Erk1/2 and Akt were examined by Western blotting with β-Actin as a loading control. (C–F) CD44^High/CD24^Low cells were sorted from TE8 or HCE4 cells, and the sorted cells were treated with trametinib (2.5 µM), GDC0941 (1 µM) or DMSO control for 48 h. (C and D) Expression levels of CD24 and CD44 were examined by flow cytometry. (E and F) The sorted TE8 cells (E) and HCE4 cells (F) were treated with trametinib (2.5 µM), GDC0941 (1 µM) or DMSO control for 48 h. mRNA expression of CDH1 and vimentin were examined by real-time RT-PCR (*P < 0.05 versus DMSO control) (Tra; trametinib, GDC; GDC0941).

Mek inhibition effectively suppresses CD44^High/CD24^Low CSCs in Ras mutated ESCC cells

To further assess the role of Ras/Erk pathway in CD44^High/CD24^Low ESCC CSC dynamics, we evaluated the influence of FGFR inhibition in T-TeRAS cells which have mutated Ras. AZD4547 had no effect upon CD44^High/CD24^Low CSCs in T-TeRAS cells (Figure 5A). Expression levels CD24 and CD44 as well as CDH1 and vimentin were also unaffected by AZD4547 treatment in T-TeRAS cells (Figure 5B). While AZD4547 efficiently suppressed Akt phosphorylation in T-TeRAS cells, Erk phosphorylation remained (Figure 5C). We continued by treating T-TeRAS cells with trametinib or GDC0941, which sharply suppressed phosphorylation of Erk or Akt, respectively (Figure 5D). Trametinib, but not GDC0941, also significantly decreased CD44^High/CD24^Low CSCs (Figure 5E). Consistent with the result of flow cytometric analysis, expression level of CD24 was increased and expression level of CD44 was suppressed by trametinib, but not by GDC0941 (Figure 5F). Trametinib also increased expression of CDH1 while concurrently diminishing expression of vimentin, suggesting loss of mesenchymal characteristics with concomitant acquisition of epithelial traits in T-TeRAS cells with Mek inhibition (Figure 5F). Expression of neither CDH1 nor vimentin was impacted by GDC0941 (Figure 5F). T-TeRAS sensitivity to 5FU, a chemotherapeutic agent used to treat ESCC patients, was significantly enhanced by trametinib with AZD4547 showing no additional effect (Figure 5G). These results identify Ras/Erk signaling as a critical factor facilitating FGF-2–mediated regulation of CD44^High/CD24^Low ESCC CSCs.

Figure 5.

CD44^High/CD24^Low ESCC CSCs are regulated by FGFR/Ras signaling. (A and B) T-TeRAS cells were treated with AZD4547 (2 µM) or DMSO control for 48 h. (A) Expression levels of CD24 and CD44 were analyzed by flow cytometry. (B) mRNA expression of CD44, CD24, CDH1 and VIM were examined by real-time RT-PCR. (AZD; AZD4547, n.s.; not statistically significant versus DMSO control). (C) T-TeRAS cells were preincubated with serum-free medium overnight then treated with AZD4547(2 µM) or DMSO control for 30 min. Phosphorylated and total Erk1/2, Akt were examined by Western blotting with β-Actin as a loading control (AZD; AZD4547). (D) T-TeRAS cells were treated with trametinib (2.5 µM), GDC0941(1 µM) or DMSO control for 48 h. Expression levels of CD24 and CD44 were detected by flow cytometry. (E) T-TeRAS cells were treated with trametinib (2.5 µM), GDC0941(1 µM) or DMSO control for 48 h. mRNA expression of CD24, CD44, CDH1 and VIM were examined by real-time RT-PCR (Tra; trametinib, GDC; GDC0941) (*P < 0.05 versus DMSO control, n.s.; not statistically significant versus DMSO control). (F) T-TeRAS cells were preincubated with serum-free medium overnight and then treated with AZD4547(2 µM), trametinib (2.5 µM), GDC0941(1 µM) or DMSO control for 30 min. Phosphorylated and total Erk1/2, Akt were examined by Western blotting with β-Actin as a loading control (Tra; trametinib, GDC; GDC0941). (G) T-TeRAS cells were treated with or without 5-FU, AZD4547 and trametinib for 72 h. Cell viability was measured by WST-8 assay. (*P < 0.05 versus 5FU (-) AZD (-) Tra (-), ^#P < 0.05 versus 5-FU (+) AZD (-) Tra (-)) (AZD; AZD4547, Tra; trametinib, GDC; GDC0941).

Pharmacological inhibition of FGFR or Mek suppresses ESCC tumor growth in vivo

To further assess the therapeutic potential of FGFR or Mek inhibition, we next utilized a preclinical in vivo xenotransplantation model of ESCC. Mice bearing established TE8 xenograft tumors displayed suppression of tumor growth following treatment with either AZD4547 or trametinib (Figure 6A). Both AZD4547 and trametinib suppressed phosphorylation of FGFR and Erk, leading to sharp suppression of vimentin and enhanced expression of E-cadherin (Figure 6B). Mice bearing T-TeRAS tumors where Ras is mutated displayed limited attenuation of tumor growth upon treatment with AZD4547 (Figure 6C) despite marked suppression of FGFR phosphorylation in these tumors (Figure 6D). In contrast, trametinib-treated animals displayed remarkable suppression of tumor growth (Figure 6C) concomitant with decreased Erk phosphorylation (Figure 6D). Vimentin was downregulated in trametinib-treated animals. E-cadherin was slightly increased in trametinib-treated animals although expression levels of E-cadherin were very low in all three groups in mice bearing T-TeRAS tumors (Figure 6D). The preclinical in vivo studies indicate that the FGFR-Ras/Erk signaling axis may represent a novel molecular target for ESCC therapy.

Figure 6.

Pharmacological inhibition of FGFR/Erk signaling suppresses ESCC tumor growth in vivo. (A) TE8 tumor-bearing mice were treated with AZD4547 (5 mg/kg), trametinib (2 mg/kg) or DMSO control via intraperitoneal injection every other day. Tumor volume was measured at indicated time points (DMSO, n = 8; AZD4547, n = 7; trametinib, n = 7; *P < 0.05 versus DMSO). (B) H&E and immunohistochemistry staining for phosphorylated FGFR, phosphorylated Erk1/2, vimentin and E-cadherin were performed in TE8 xenograft tumors. (C) T-TeRAS tumor-bearing mice were treated with AZD4547 (5 mg/kg), trametinib (2 mg/kg) or DMSO control via intraperitoneal injection every other day. Tumor volume was measured at indicated time points (DMSO, n = 4; AZD4547, n = 6; trametinib, n = 6; *P < 0.05 versus DMSO control). (D) H&E and immunohistochemistry staining for phosphorylated FGFR, phosphorylated Erk1/2, vimentin and E-cadherin were performed in T-TeRAS xenograft tumors.

Discussion

In most cancers, FGF-2 promotes disease progression and high expression of FGF-2 or FGFRs is significantly associated with poor prognosis of advanced stage cancer patients (13). FGF-2 promotes cancer cell proliferation and makes tumor cells more resistant to anticancer drugs (30). FGF-2 also alters the tumor microenvironment, stimulating stromal and inflammatory cells while also promoting angiogenesis (39). Recently, Fessler E. et al. reported that FGF-2 in the tumor microenvironment has pivotal roles for regulating CSCs. In glioblastoma, FGF-2, secreted by tumor microvascular endothelial cells, reverts differentiated glioblastoma cells to undifferentiated CSCs (29). Thus, FGF-2 plays important roles in regulating CSCs in diverse types of cancers. Although the role of FGF-2 in regulation of CSCs in ESCC is not been fully understood, FGF-2 has tumor promoting effects in ESCC. Enhanced expression of FGFR3-IIIc promotes ESCC proliferation via autocrine FGF signaling (21), and FGF-2/FGFR1 signaling regulates survival and migration of tumor-associated macrophages and cancer cells (22). In several tumor types, cancer cells that have undergone EMT acquire stem cell-like phenotypes and ability to promote invasion, metastasis and drug resistance (34). We have also reported that EMT mediated by transforming growth factor (TGF)-β is a critical process for regulation of CD44^High/CD24^Low CSCs in ESCC (10, 25, 27). Considering that FGF-2 drives EMT in some cancers (23, 24), it’s possible that FGF-2 might have important roles in regulating CSCs and promoting poor prognosis in ESCC. In the present study, we have demonstrated that FGF-2 expression is significantly increased in CD44^High/CD24^Low ESCC CSCs and that exogenous FGF-2 increases CD44^High/CD24^Low CSCs with characteristics consistent with EMT (Figures 1 and 2). Furthermore, FGFR inhibitors suppress CSCs and facilitate upregulation of E-cadherin with concurrent downregulation of vimentin, suggesting MET (Figure 3). Thus, the present study highlights the importance of FGF-2 in regulation of CSCs in ESCC with potential implications for cell plasticity.

Although transforming growth factor-β (TGF-β) is one of the most well-known EMT-inducing stimuli, additional growth factors such as FGF, EGF, hepatocyte growth factor and insulin-like growth factor (IGF) have been implicated in EMT (37). Interestingly, cross talk between TGF-β and FGF signaling has been revealed to regulate EMT. Shirakihara et al. have shown that TGF-β regulates isoform switching of FGFRs from FGFR2b to FGFR1c, thereby causing the cells to become sensitive to FGF-2 (38). Cross talk between FGFR signaling and other signaling pathways has also been reported to contribute to CSC dynamics. In non-small cell lung cancer, FGFR1 promotes acquisition of a stem cell-like phenotype by upregulating expression of GLI2, a crucial transcriptional mediator of Hedgehog signaling (31). In triple negative breast cancer, Notch1 and FGFR signaling cooperatively regulate CSCs and sustain drug resistance against mechanistic target of rapamycin 1/2 inhibition (32). In hepatocellular carcinoma tumor initiating cells, the IGF and FGF signaling cascades are upregulated and contribute to drug resistance against sorafenib (33). In sum, these previous reports demonstrate the importance of FGFR signaling in CSC biology. We have previously reported that EGFR inhibitors, erlotinib and cetuximab, block EMT and suppress expansion of CSCs in ESCC (10). It is tempting to speculate that interaction between FGF, EGF and TGF-β signaling might cooperatively regulate EMT and expand CD44^High/CD24^Low CSCs in the context of ESCC; however, detailed mechanistic studies are required to investigate such a possibility.

Mitogen-activated protein kinases (MAPK) signaling and PI3K/Akt signaling are the primary signaling cascades activated downstream of FGFR signaling (15, 16), however, where downstream signaling pathway regulates CSCs in ESCC remained an open question. FGFR inhibitor via AZD4547 suppressed CSCs both in TE8 and HCE4 cells. Interestingly, although Erk phosphorylation was suppressed in both cell lines, Akt phosphorylation was unaltered in HCE4 cells (Figure 3G). These results implied that MAPK signaling might play a more essential role in regulating CSCs in ESCC. In line with this notion, Mek inhibitor but not PI3K inhibition suppressed ESCC CSCs (Figure 4C and D, Figure 5E). Furthermore, in T-TeRAS cells, where mutant Ha-Ras^V12 is overexpressed and Ras/Mek/Erk pathway is constitutively activated, FGFR inhibitor failed to suppress Erk phosphorylation and had no influence upon CD44^High/CD24^Low CSCs. In contrast, Mek inhibition via trametinib (Figure 5D and E) or U0126 (data not shown) sharply suppressed Erk phosphorylation and depleted CSCs in T-TeRAS cells. Thus, we show for the first time that FGF-2/FGFR/Ras/Erk signaling is the primary signaling cascade involved in regulation of CD44^High/CD24^Low CSCs in ESCC. Furthermore, we show that AZD4547 and trametinib are effective at promoting tumor regression in a preclinical in vivo ESCC xenograft model (Figure 6). Currently, trametinib has been approved for and has greatly improved the prognosis of patients with advanced BRAF mutated melanoma (35). Our findings indicate that trametinib may represent a novel therapeutic option for targeting CSCs in ESCC, a strategy that has the potential to benefit patients with advanced ESCC, a population for whom available therapies remain ineffective.

In conclusion, FGF-2 expression is significantly upregulated in CD44^High/CD24^Low ESCC CSCs and has a pivotal role for regulating CSC maintenance. Furthermore, the present study show for the first time that the Ras/Erk pathway is the primary downstream effector of FGFR signaling with regard to regulation of ESCC CSCs. Inhibitors against FGFR signaling and particularly Mek inhibitors may represent a novel therapeutic tool for targeting CSCs and improving outcomes in patients with advanced ESCC.

Abbreviations

CSC

cancer stem-like cell

EGFR

epidermal growth factor receptor

EMT

epithelial–mesenchymal transition

Erk

extracellular signal-regulated kinases

ESCC

esophageal squamous cell carcinoma

FACS

fluorescence activated cell sorting

FGF

fibroblast growth factor

FGFR

FGF receptor

MAPK

mitogen-activated protein kinases

Mek

MAPK/Erk kinase

PI3K

phosphoinositide-3 kinase

TGF-β

transforming growth factor-β