Chemoradiation induces epithelial-to-mesenchymal transition in esophageal adenocarcinoma

Abstract

Multimodality treatment has advanced the outcome of esophageal adenocarcinoma (EAC), but overall survival remains poor. Therapeutic pressure activates effective resistance mechanisms and we characterized these mechanisms in response to the currently used neoadjuvant treatment against EAC: carboplatin, paclitaxel and radiotherapy. We developed an in vitro approximation of this regimen and applied it to primary patient-derived cultures. We observed a heterogeneous epithelial-to-mesenchymal (EMT) response to the high therapeutic pressure exerted by chemoradiation. We found EMT to be initiated by the autocrine production and response to transforming growth factor beta (TGF-β) of EAC cells. Inhibition of TGF-β ligands effectively abolished chemoradiation-induced EMT. Assessment of TGF-β serum levels in EAC patients revealed that high levels after neoadjuvant treatment predicted the presence of fluorodeoxyglucose uptake in lymph nodes on the post-chemoradiation positron emission tomography-scan. Our study shows that chemoradiation contributes to resistant metastatic disease in EAC patients by inducing EMT via autocrine TGF-β production. Monitoring TGF-β serum levels during treatment could identify those patients at risk of developing metastatic disease, and who would likely benefit from TGF-β targeting therapy.

Keywords: TGF-β; biomarker; chemoradiation; epithelial-to-mesenchymal transition; esophageal adenocarcinoma.

Figure 1

Chemoradiation therapy induces a mesenchymal morphology in a subset of EAC tumors. (a) 031M cells were either left untreated or subjected to a 14 days chemoradiation scheme and morphology was assessed by phase‐contrast microscopy and immunofluorescent staining using the indicated epithelial (CDH1) and mesenchymal (VIM) markers. Nuclei were stained using DAPI. Scale bar: 200 μm. (b) As for (a), using the 007B culture. [Color figure can be viewed at wileyonlinelibrary.com]

Figure 2

Chemoradiation therapy readily increases markers for EMT in EAC cells. (a) OE19, 031M and 007B cells were either left untreated or subjected to a 14 days chemoradiation scheme, and cell surface epithelial (CDH1, as CDH1 was not detected by flow cytometry in 031M cells EPCAM was used) and mesenchymal (VIM) markers were assessed using flow cytometry. (b) OE19, 031M and 007B cells in the same conditions as for (a), and gene expression of the indicated mesenchymal and cancer stem cell markers was determined using quantitative RT‐PCR. The bar graphs show means ± SD, n = 2. *p < 0.05, **p < 0.01, ***p < 0.001. [Color figure can be viewed at wileyonlinelibrary.com]

Figure 3

EAC cells produce TGF‐β in response to chemoradiation therapy and thereby induce EMT. (a) OE19, 031M and 007B cells were subjected to a 14 days chemoradiation scheme and the supernatant was used to measure the amount of free TGF‐β using ELISA. (b) OE19 and 31M cells were subjected to a 14 days chemoradiation scheme with or without the addition of fresolimumab, a TGF‐β neutralizing antibody, to the last 7 days of therapy and the supernatant was used to stimulate 293T cells. Supernatant from naive cancer cells was used as control. Different conditions were on the same membrane with equal exposure time, dashed lines indicate that membrane was cropped. (c) OE19, 031M and 007B cells were either left untreated, stimulated with recombinant TGF‐β for 14 days or subjected to a 14 days chemoradiation regimen and morphology was assessed by phase‐contrast microscopy. Scale bar: 200 μm. (d) Immunofluorescent staining of the indicated markers (CDH1 and VIM) on 031M cells, either subjected to 14 days of chemoradiation, recombinant TGF‐β or untreated. DAPI was used for nuclear staining. Scale bar: 200 μm. The bar graphs show means ± SD, n = 2. [Color figure can be viewed at wileyonlinelibrary.com]

Figure 4

Inhibition of TGF‐β signaling during chemoradiation can reverse EMT in EAC cells. (a) OE19 and 031M cells were subjected to chemoradiation for 14 days, with or without the addition of the TGF‐β neutralizing antibody (fresolimumab) in the final 7 days. Morphology was assessed by phase‐contrast microscopy. Arrows indicate type and time of treatment. Scale bar: 200 μm. (b) Morphological analyses of 031M cells in the following conditions; no treatment, 14 days chemoradiation, 7 days chemoradiation followed by 7 days observation, 7 days chemoradiation followed by 7 days chemoradiation with fresolimumab. Arrows indicate type and time of treatment. The white blocks in the arrows indicate time points at which images were taken. Scale bar: 200 μm. (c) Supernatant of 031M cells which were either left untreated or received 7 days of chemoradiation followed by 7 days of observation was used to measure the amount of free TGF‐β using ELISA. The bar graphs show means ± SD, n = 2. [Color figure can be viewed at wileyonlinelibrary.com]

Figure 5

TGF‐β inhibition reduces the migratory capacity of EAC cells exposed to chemoradiation. (a) OE19 and 031M cells were left untreated or subjected to recombinant TGF‐β for 14 days, chemoradiation for 14 days, with or without the addition of the TGF‐β neutralizing antibody (fresolimumab) in the final 7 days. Gene expression of the indicated mesenchymal markers was measured using quantitative RT‐PCR. The Bar graphs show means ± SD, n = 3. *p < 0.05, **p < 0.01 and ***p < 0.001. (b) Staining of VIM using flow cytometry on OE19 and 031M cells in the same condition as (a). The dashed histograms represent the isotype control. (c) Transwell migration assays on OE19 and 031M cells in the same conditions as (a). Data shown in the graphs are corrected for no‐attractant controls (medium without FCS), shown are technical duplicates of two individual experiments. p‐Values were determined by two‐way ANOVA and Tukey's multiple comparisons correction. The p‐values shown are compared to the control condition. The one‐phase exponential curves were plotted, including the SD. [Color figure can be viewed at wileyonlinelibrary.com]

Figure 6

High TGF‐β serum levels after chemoradiation correlate with FDG‐avid metastatic disease. (a) PFS status of EAC patients before start of therapy using a TGF‐β serum level cutoff at 22.75 pg/mL as determined using the cutoff finder (n = 63). (b) TGF‐β serum levels of EAC patients with various (neo)adjuvant treatments measured in serum samples at the time of metastatic or nonmetastatic disease (n = 175). (c) ΔTGF‐β serum level (post CR—baseline) was determined after chemoradiation in patients with or without FDG uptake in structures other than the primary tumor on post chemoradiation PET scan (n = 20). Statistical significance was evaluated using Mann‐Whitney U test. The dot plots show mean ± SD. (d) Post‐chemoradiation total body 18F‐FDG‐PET scan and fused PET‐CT image of a patient with no FDG uptake in structures other than the primary tumor with ΔTGF‐β serum level of −44. (e) Patient with FDG uptake in a new lymph node, which did not show up on the pretreatment scan, with ΔTGF‐β serum level of −4. Arrow indicates metastatic disease with pathological FDG uptake in upper jugular lymph node at level II. [Color figure can be viewed at wileyonlinelibrary.com]