Abstract
Purpose: The majority of gastric cancer patients who achieve an initial response to trastuzumab-based regimens develop resistance within 1 year of treatment. This study was aimed at identifying the molecular mechanisms responsible for resistance.
Experimental Design: A HER2+-trastuzumab sensitive NCI-N87 gastric cancer orthotopic nude mouse model was treated with trastuzumab until resistance emerged. Differentially expressed transcripts between trastuzumab-resistant and sensitive gastric cancer cell lines were annotated for functional interrelatedness by Ingenuity Pathway Analysis software. Immunohistochemical analyses were performed in pretreatment versus posttreatment biopsies from gastric cancer patients receiving trastuzumab-based treatments. All statistical tests were two-sided.
Results: Four NCI-N87 trastuzumab-resistant (N87-TR) cell lines were established. Microarray analysis showed HER2 downregulation, induction of epithelial-to-mesenchymal transition, and indicated fibroblast growth factor receptor 3 (FGFR3) as one of the top upregulated genes in N87-TR cell lines. In vitro, N87-TR cell lines demonstrated a higher sensitivity than did trastuzumab-sensitive parental cells to the FGFR3 inhibitor dovitinib, which reduced expression of pAKT, ZEB1, and cell migration. Oral dovitinib significantly (P = 0.0006) reduced tumor burden and prolonged mice survival duration in N87-TR mouse models. A higher expression of FGFR3, phosphorylated AKT, and ZEB1 were observed in biopsies from patients progressing under trastuzumab-based therapies if compared with matched pretreatment biopsies.
Conclusions: This study identified the FGFR3/AKT axis as an escape pathway responsible for trastuzumab resistance in gastric cancer, thus indicating the inhibition of FGFR3 as a potential strategy to modulate this resistance. Clin Cancer Res; 22(24); 6164–75. ©2016 AACR.
Translational Relevance
The majority of gastric cancer patients who achieve an initial response to trastuzumab-based regimens develop resistance within 1 year of treatment. The molecular mechanisms involved in trastuzumab resistance of gastric cancer remain uncharacterized. In this study, we propose an in vivo model in which trastuzumab therapy induces the selection of gastric cancers overexpressing FGFR3, which activates the PI3K/AKT/mTOR signaling pathway, sustaining, in turn, tumor growth and a more aggressive EMT phenotype. We confirmed our finding by demonstrating for the first time the overexpression of FGFR3 in paired pretreatment and postprogression bioptic samples from patients affected by advanced gastric cancer that relapsed upon trastuzumab therapy. Of translational relevance, we showed the therapeutic efficacy in vivo of the FGFR3 inhibitor dovitinib in models of trastuzumab-resistant gastric cancer. Our study provides the preclinical rationale to investigate the inhibition of FGFR3 as second-line treatment strategy in gastric cancer patients refractory to first-line trastuzumab-containing therapies.
Introduction
Gastric cancer is the fourth most commonly diagnosed cancer and the third leading cause of cancer-related death worldwide (1, 2). Trastuzumab, a recombinant humanized mAb directed against the human epidermal growth factor receptor 2 (HER2), is the only targeted agent to be approved for the first-line treatment of patients with HER2-overexpressing metastatic gastric or gastroesophageal junction adenocarcinoma (3). However, after an initial period of clinical benefit, patients almost inevitably progress, as the tumors become refractory to trastuzumab (4).
Several mechanisms involved in acquired resistance to trastuzumab have been described in breast cancer (5), including overexpression of Cyclin E (6), cross-talk between HER2 and other tyrosine kinase receptors (7, 8), or activation of downstream intracellular signal transducers such as SRC (9).
In contrast, the molecular mechanisms involved in resistance to trastuzumab in gastric cancer remain largely uncharacterized. Recently, integrated SNP array–based copy number and whole-exome sequencing analyses of data from HER2-amplified gastroesophageal adenocarcinoma revealed that more than half of the cases had additional oncogenic alterations at diagnosis that could potentially hamper the antitumor effect of anti-HER2 agents, including amplifications of cell cycle–related genes CCNE1 and CDK6, PI3K pathway activation by PI3KCA mutations, and gene amplification of other tyrosine kinase receptors such as EGFR and MET (10).
Solid cancers are molecularly heterogeneous, but among the large number of genetic alterations present at diagnosis only few of them represent relevant driver gene mutations that directly or indirectly confer a selective growth advantage (11). Although we recognize the potential contributions of additional oncogenic alterations present at diagnosis to tumor progression during therapy with trastuzumab, a number of genes that contain few or no mutations could be instead overexpressed, downregulated, or epigenetically altered, thus playing an equally important role in the development of drug resistance (12).
Therefore, in this study we aimed at directly identifying transcriptional mechanisms responsible for the resistance of gastric cancer to trastuzumab, which may represent new therapeutic targets in the search for ways to reverse the invariable escape of this disease from anti-HER2 therapies.
Materials and Methods
Cell lines, reagents, and in vitro studies
Human gastric cancer cell line NCI-N87 and human breast cancer cell lines MCF-7 and BT474 were obtained from ATCC. Human gastric cancer cell line YCC-2 was obtained from KCLB. Cell lines were authenticated by standard short tandem repeat (STR) DNA typing methodology before being purchased from cell bank. Cells were daily checked by morphology and routinely tested to be mycoplasma free by PCR assay. Generation of GFP+/luciferase+ NCI-N87 cell line (13), in vitro cell proliferation assay, wound-healing assay (14), IHC, protein extraction, and Western blotting (15) were performed as previously described, and for further details see Supplementary Materials and Methods.
Gene expression microarray and pathway analysis
RNA isolation and quantitative RT-PCR assay were performed as detailed in Supplementary Materials and Methods. Differences in gene expression between parental and resistant cells were examined using Illumina Human 48k-gene-chips (Illumina) as indicated in Supplementary Materials and Methods. Gene expression microarray data have been deposited in the GEO database (accession no. GSE77346). Differentially expressed transcripts were tested for network and functional interrelatedness using the IPA software program (Ingenuity Systems).
Establishment of gastric cancer cell lines in vivo resistant to trastuzumab and in vivo studies
Six- to 8-week-old female BALB/c athymic (nu/nu) mice were purchased from Harlan Laboratories. All mice were housed and treated in accordance with the guidelines of the Italian Ministry of Health Animal Care and Use Committee, and maintained in specific pathogen-free conditions at CIRSAL Animal Care Facility. The orthotopic implantation of gastric cancer cells was performed in six mice as described previously (16). Briefly, GFP+/luciferase+ NCI-N87 cell line was harvested from subconfluent culture by exposure to trypsin. Trypsinization was stopped with medium containing 10% FBS, and the cells were washed once with PBS. To produce orthotopic gastric tumors, 1 × 106 cells resuspended in 50 μL of PBS/Matrigel solution (1:1) were injected into the gastric wall of nude mice anesthetized with a 1.5% isoflurane–air mixture. To prevent such leakage, a cotton swab was held over the injection site for 1 minute. One layer of the abdominal wound was closed with wound clips (Auto-clip; Clay Adams). The mice tolerated the surgical procedure well, and no anesthesia-related deaths occurred. Tumor growth was monitored by bioluminescent imaging performed using a cryogenically cooled IVIS 100 imaging system coupled with a data-acquisition computer running the Living Image software program (Xenogen). When the resulting tumors became detectable, the mice were given 20 mg/kg of trastuzumab i.p. twice a week until the tumors suddenly recurred during continuous therapy. Treatment resistance developed in 4 of 6 mice. At evidence of advanced bulky disease, mice were euthanized using carbon dioxide inhalation. Four trastuzumab-resistant cell lines were established from excised tumors via repeated green fluorescent protein flow cytometric sorting with FACSAria II sorter (Becton Dickinson).
The subcutaneous heterotopic implantation of gastric cancer cells was performed as described previously (17). Tumor bearing mice were randomly assigned (n = 10 per group) to receive 20 mg/kg of trastuzumab i.p. twice a week for 4 weeks, or 40 mg/kg of dovitinib oral gavage daily for 4 weeks, or respective vehicles as a control. Tumor size was measured with a caliper by the modified ellipsoid formula (π/6) x AB2, where A is the longest and B is the shortest perpendicular axis of an assumed ellipsoid corresponding to tumor mass, as reported previously (18). All mice were weighed weekly and observed for tumor growth. When at least 6 of the 10 mice in a treatment group presented with bulky disease, the median survival duration for that group was considered to have been reached. At the median survival duration of the control group, the tumor growth in mice in all groups was evaluated. The mice were euthanized using carbon dioxide inhalation when evidence of advanced bulky disease developed or at cut-off volume of 2 cm3, which was considered the day of death for the purpose of survival evaluation.
Patients
Three patients were considered for analyses. For each patient, pre- and postresistance tumor samples were compared by immunohistochemical analyses. Informed consent was obtained from all patients. See Supplementary Materials and Methods for clinical history and details.
Statistical analysis
The results of in vitro proliferation were analyzed for statistical significance of differences by nonlinear regression analysis and are expressed as means and 95% confidence intervals (CI) for at least three independent experiments performed in quadruplicate. Statistical significance of differences in tumor growth was determined by the Mann–Whitney test; differences in survival duration were determined using a log-rank test. All statistical tests were two-sided, and a P value less than 0.05 indicated statistical significance. All statistical analyses were performed using GraphPad Prism software version 4.0c for Macintosh (GraphPad Software).
Results
In vivo selection of four gastric cancer models with acquired resistance to trastuzumab
We initially measured the expression levels of several members of the EGFR protein family in YCC-2 and NCI-N87 gastric cancer, and in the BT-474 and MCF-7 breast cancer cell lines (19), used as HER2-positive and HER2-negative controls, respectively. NCI-N87 gastric cancer cells showed a high basal expression of HER2 comparable with that of the HER2-positive BT-474 breast cancer cells (Fig. 1A). Consistently, with the expression of HER2, the NCI-N87 and BT-474 cells demonstrated in vitro a significantly higher sensitivity to trastuzumab (NCI-N87 IC50 = 4.74 μg/mL; BT-474 IC50 = 1.05 μg/mL) than did YCC-2 and MCF-7 cells (IC50 > 400 μg/mL; Fig. 1B).
In vivo selection of gastric cancer models with acquired resistance to trastuzumab treatment. A, Western blot analysis for the expression of ErbB family receptors in breast cancer, MCF-7 and BT-474, and gastric cancer, YCC-2 and NCI-N87 cell lines. B, Percent relative growth of MCF-7 and BT-474 breast cancer, and YCC-2 and NCI-N87 gastric cancer cell lines, 72 hours after treatment with increasing concentrations of trastuzumab. Vehicle-treated cells were assigned a value of 100% and designated as control. Means and 95% CIs of three independent experiments performed in quadruplicate are shown. C, GFP+/luciferase+ NCI-N87 cells were orthotopically injected into the gastric wall of four nude mice. When the resulting tumors became detectable, the mice were given 20 mg/kg of trastuzumab i.p. twice a week. As expected, the tumors responded dramatically to the treatment. The mice received treatment until the tumors suddenly recurred during continuous therapy with trastuzumab. At evidence of advanced bulky disease, mice were euthanized. Four novel trastuzumab-resistant cell lines, N87-TR1, N87-TR2, N87-TR3, and N87-TR4, were established from excised tumors via repeated GFP flow cytometric sorting. The tumor growth was quantified weekly on the basis of bioluminescence emitted by the tumor cells as the sum of all detected photons within the region of the tumor per second using a cryogenically cooled IVIS 100 imaging system coupled with a data-acquisition computer running the Living Image software program (Xenogen). A digital grayscale image was acquired, followed by the acquisition and overlay of a pseudocolor image representing the spatial distribution of detected photons emerging from the active luciferase within the mouse (representative image). D, Light- and fluorescent microscopic phenotype of the trastuzumab-sensitive NCI-N87 cells and trastuzumab-resistant N87-TR1, N87-TR2, N87-TR3, and N87-TR4 gastric cancer cells. E, Percent relative growth of trastuzumab-sensitive NCI-N87 cells and trastuzumab-resistant N87-TR1, N87-TR2, N87-TR3, and N87-TR4 gastric cancer cells, 72 hours after treatment with increasing concentrations of trastuzumab. Vehicle-treated cells were assigned a value of 100% and designated as control. Means and 95% CIs of three independent experiments performed in quadruplicate are shown. F, Fifty athymic nude mice bearing heterotopic NCI-N87, N87-TR1, N87-TR2, N87-TR3, and N87-TR4 gastric tumors were randomly assigned to 10 groups (n = 5 per group) to receive 20 mg/kg of either trastuzumab or saline (control) i.p. twice a week. Mice were sacrificed by carbon dioxide inhalation when evidence of advanced bulky disease developed. The day of sacrifice was considered the day of death from disease for the purpose of survival evaluation. Differences among survival duration of mice in each group were determined by log-rank test. NCI-N87, control versus trastuzumab, median survival = 66.5 days vs. undefined days, P = 0.0027, HR = 0.03652, 95% CI = 0.004217–0.3163; N87-TR1, control versus trastuzumab, median survival = 55 days vs. 49 days, HR = 0.7026, 95% CI = 0.04055–12.17, P = 0.8; N87-TR2, control versus trastuzumab, median survival = 54 days versus 42.5 days, HR = 1.338, 95% CI = 0.3258–5.491, P = 0.686; N87-TR3, control versus trastuzumab, median survival = 49 days versus 40 days, HR = 3.621, 95% CI = 0.5385–19.75, P = 0.1983; N87-TR4, control versus trastuzumab, median survival = 54.5 days vs. 44 days, HR = 2.143, 95% CI = 0.2905–15.8, P = 0.4547.
To study the molecular mechanisms of acquired resistance to HER2-targeted agents in gastric cancer, we established and validated four novel trastuzumab-resistant cell lines, N87-TR1, N87-TR2, N87-TR3, and N87-TR4 (Fig. 1C). Trastuzumab-resistant cell lines had a more spindle-shaped morphology (Fig. 1D) and exhibited significantly higher migration rates compared with trastuzumab-sensitive NCI-N87 (P < 0.001; Supplementary Fig. S1).
N87-TR1, N87-TR2, N87-TR3, and N87-TR4 cells demonstrated a significantly higher resistance in vitro to trastuzumab than did the parental NCI-N87 cells (NCI-N87 IC50 = 4.74 μg/mL vs. N87-TR1, N87-TR2, N87-TR3, or N87-TR4 IC50 > 400 μg/mL; Fig. 1E). As expected, NCI-N87 tumors were sensitive to drug treatment, whereas mice bearing trastuzumab-resistant tumors had survival rates comparable with untreated control mice (Fig. 1F).
Identification of relevant biologic processes and genes by using global transcript profiling
To gain insight into the molecular mechanisms underlying trastuzumab-resistant phenotype in gastric cancer cells, we compared gene expression profiles in sensitive and resistant cells by microarray analysis to identify groups of genes associated with a specific signaling pathway or biological process. Differentially modulated transcripts in trastuzumab-resistant cells were enriched for genes implicated in mTOR signaling pathway and regulation of the eukaryotic initiation factors (eIF) 2 and 4 (Fig. 2A). Top differentially regulated transcripts in trastuzumab-resistant gastric cancer cell lines compared with NCI-N87 control cell line are summarized in Supplementary Table S1.
Identification of relevant biological processes and genes by using global transcript profiling. A, Signaling pathways enriched among genes differentially expressed in trastuzumab-resistant gastric cancer cells versus their sensitive control cell line. The X-axis represents the −log(10) P value for enrichment, with the threshold drawn at P = 0.05. B, Western blot analysis for the expression of ErbB family receptors in N87-TR gastric cancer cells compared to trastuzumab-sensitive NCI-N87 gastric cancer and BT-474 breast cancer cells, and trastuzumab-resistant YCC2 gastric cancer and MCF-7 breast cancer cells. C, Validation of ZEB-1, E-cadherin (CDH1), and vimentin (VIM) genes differentially expressed in trastuzumab-resistant cells versus parental sensitive cells. qRT-PCR data are expressed as the fold change in RNA expression between the gene of interest and β-actin. Mean and 95% CIs are shown. D, Western blot analysis for the expression of EMT markers. E, Validation of fibroblast growth factor (FGF)9 and FGF receptor 3 (FGFR3) genes differentially expressed in trastuzumab-resistant cells versus parental sensitive cells. qRT-PCR data are expressed as the fold change in RNA expression between the gene of interest and β-actin. Mean and 95% CIs are shown. F, Western blot analysis for the expression of FGFR3, and the activation of AKT and ERK1/2, in N87-TR cells and their control cell line. G, Interaction network derived from genes upregulated in trastuzumab-resistant gastric cancer cell lines versus the respective sensitive control cell line. Each interaction is supported by at least one literature reference identified in the Ingenuity Pathway Knowledge Base, with solid lines representing direct interactions and dashed lines representing indirect interactions. FGFR3, fibroblast growth factor receptor 3; FGF9, fibroblast growth factor 9; PIP2, phosphatidylinositol 4,5-bisphosphate; PIP3, phosphatidylinositol (3,4,5)-trisphosphate; AKT, V-Akt murine thymoma viral oncogene homolog; PI3K, phosphatidylinositol-4,5-bisphosphate 3-kinase; PDK1, pyruvate dehydrogenase kinase isozyme 1; CDH1, cadherin 1; TSC1, tuberous sclerosis 1; TSC2, tuberous sclerosis 2; Rheb, RAS-homolog enriched in brain; mTORC1, mechanistic target of rapamycin complex 1; mTORC2, mechanistic target of rapamycin complex 2; Rho, rhodopsin; Rock, Rho-associated coiled-coil containing protein kinase; VIM, vimentin; Raptor, regulatory associated protein of MTOR complex 1; GBL, G protein beta subunit-like; PRAS40, proline-rich Akt substrate of 40 kDa; mTOR, mammalian target of rapamycin; Rictor, RPTOR independent companion of MTOR complex 2; Proctor, protein observed with rictor-1; SIN1, SAPK-interacting protein.
Downregulation of HER2, EGFR, and HER3 expression in trastuzumab-resistant lines was evident, which was confirmed at the protein level by Western blot analysis (Fig. 2B). Consistent with this finding, although trastuzumab blocked the activation of AKT in sensitive NCI-N87 cells, it was completely ineffective in trastuzumab-resistant lines (Supplementary Fig. S2).
However, transcriptome analysis revealed that the expression of ZEB1, a key member of the transcriptional complexes essential for the development of the epithelial-to-mesenchymal transition (EMT) genetic program, and that of the mesenchymal marker vimentin (VIM) were strongly upregulated, and that the expression of the epithelial marker gene E-cadherin (CDH1) was strongly downregulated in trastuzumab-resistant gastric cancer cell lines compared with sensitive control cell line (Supplementary Table S1; Fig. 2C). Consistently, the N87-TR1, N87-TR2, N87-TR3, and N87-TR4 cell lines had considerably lower levels of E-cadherin and higher levels of ZEB1 and vimentin protein expression than did their trastuzumab-sensitive cell lines when cultured in vitro (Fig. 2D). These results indicate EMT as a plausible underlying molecular mechanism responsible for the phenotypic changes observed in trastuzumab-resistant cells.
Among the top differentially regulated genes, we also found a consistent overexpression of the membrane receptor gene FGFR3 and of the gene coding for its ligand FGF9 in the N87-TR1, N87-TR2, N87-TR3, and N87-TR4 trastuzumab-resistant gastric cancer cell lines compared with NCI-N87–sensitive control cell lines (Fig. 2E). We validated that resistant cell lines had significantly higher levels of FGFR3, and of total and phosphorylated AKT than did their trastuzumab-sensitive parental counterparts. However, in all trastuzumab-resistant lines, we found a profound suppression of basal ERK1/2 phosphorylation, the other main intracellular signaling pathway activated by tyrosine kinase receptors (Fig. 2F).
We corroborated the most relevant changes in protein levels in vivo. All trastuzumab-resistant tumors had undetectable or low levels of HER2 expression. NCI-N87 tumors had low expression levels of FGFR3 and of phosphorylated AKT, high expression of E-cadherin, and no expression of ZEB1 and vimentin. In contrast, all trastuzumab-resistant tumors exhibited a significantly stronger expression of FGFR3 and phosphorylated AKT, no expression of E-cadherin, but high expression levels of ZEB1 and vimentin (Fig. 3).
Immunohistochemical analysis of trastuzumab-resistant versus trastuzumab-sensitive gastric xenograft tumors. Serial paraffin-embedded tumor sections were stained with antibodies against HER2, FGFR3, ZEB1, pAKT, E-cadherin, and vimentin proteins (scale bar: 200 μm; 10× magnification).
On the basis of the overall results, we propose a mechanistic model for the resistance to trastuzumab in gastric cancer in which an autocrine loop established through the overexpression of FGFR3 and of its ligand FGF9 could be responsible for the activation of the PI3K/AKT/mTOR signaling pathway, sustaining, in turn, tumor growth and EMT irrespectively of the anti-HER2 treatment (Fig. 2G).
Effects of FGFR inhibition in trastuzumab-resistant gastric cancer models
To validate our proposed model, we tested the activity of two different FGFR3 inhibitors, dovitinib and AZD4547, in vitro. All four trastuzumab-resistant cell lines were significantly more sensitive than NCI-N87 parental cells (dashed line) to the same extent to the inhibition of FGFR3 (All P < 0.0001; Fig. 4A). Furthermore, we found that in a clinically relevant concentration range, dovitinib was able to induce a measurable reduction of the phosphorylation of AKT (Fig. 4B), suppressed the expression of ZEB1 (Fig. 4C), and significantly (P < 0.01) inhibited migration in all trastuzumab-resistant cell lines (Fig. 4D and E).
In vitro antitumor activity of fibroblast growth factor receptor 3 inhibitor dovitinib in trastuzumab-resistant N87-TR cell lines. A, Percent relative growth of N87-TR cells and their control cell line, 72 hours after treatment with increasing concentrations of dovitinib or AZD4547. Vehicle-treated cells were assigned a value of 100% and designated as control. Means and 95% CIs of three independent experiments performed in quadruplicate are shown. Dovitinib: NCI-N87 IC50 = 2.189E−006 M versus N87-TR1 IC50 = 4.122E−007 M, or versus N87-TR2 IC50 = 5.297E−007 M, N87-TR3 IC50 = 6.127E−007 M, N87-TR4 IC50 = 5.764E−007 M, all P < 0.0001; AZD4547: NCI-N87 IC50 = 1.697E−005 M versus N87-TR1 IC50 = 1.038E−005 M, or versus N87-TR2 IC50 = 9.465E−006 M, N87-TR3 IC50 = 9.161E−006 M, N87-TR4 IC50 = 6.875E−006 M, all P < 0.0001. B, Western blot analysis for the activation of AKT after 24 hours treatment with dovitinib. C, Western blot analysis for the expression of ZEB1 after 72 hours treatment with dovitinib. D, Photographs of the wound area were taken by using phase-contrast microscopy immediately and 36 hours after the incision in untreated and treated cells. E, Levels of cancer cell migration of N87-TR cells and their control cell line after dovitinib treatment. Results are presented as percentages of the total distances between the wound edges enclosed by cancer cells. The mean values and 95% CIs from three independent experiments done in quadruplicate are shown. Relative migration: NCI-N87, control, mean = 15.91, 95% CI = 12.75–19.06, versus dovitinib, mean = 8.39, 95% CI = 5.24–11.55, P < 0.01; N87-TR1, control, mean = 93.85, 95% CI = 86.05–101.65, versus dovitinib, mean = 3.51, 95% CI = 1.84–5.17, P < 0.001; N87-TR2, control, mean = 100, 95% CI = 99.12–100.88, versus dovitinib, mean = 4.41, 95% CI = 3.53–5.28, P < 0.001; N87-TR3, control, mean = 93.84, 95% CI = 85.42–102.25, versus dovitinib, mean = 4.06, 95% CI = 2.14–5.99, P < 0.001; N87-TR4, control, mean = 100, 95% CI = 99.12–100.88, versus dovitinib, mean = 6.33, 95% CI = 5.11–7.55, P < 0.001. **, P < 0.01; ***, P < 0.001 (unpaired Student t test).
To demonstrate that FGFR3 is a druggable target to overcome acquired resistance to trastuzumab in gastric cancer, 40 mice were injected with trastuzumab-sensitive NCI-N87 or trastuzumab-resistant N87-TR4 gastric cancer cells and randomly assigned to receive oral dovitinib or its vehicle as a control. Dovitinib was completely inactive in NCI-N87 tumor bearing mice (Fig. 5A). Conversely, N87-TR4 gastric cancer bearing mice treated with dovitinib experienced a significant reduction in tumor burden. Accordingly, only the N87-TR4 tumor bearing mice treated with dovitinib demonstrated a significantly longer median survival duration (P = 0.0006; Fig. 5B). In this regard, control N87-TR4 tumors from mice treated with oral vehicle showed a moderate to strong expression of phosphorylated AKT. Conversely, N87-TR4 tumors from mice treated with dovitinib demonstrated only very weak or no expression of phosphorylated AKT (Fig. 5C).
Antitumor activity of oral fibroblast growth factor receptor 3 inhibitor dovitinib in vivo in N87-TR4 and NCI-N87 gastric tumor xenografts. A and B, Left, 40 athymic nude mice bearing subcutaneous heterotopic NCI-N87 or N87-TR4 gastric tumors were randomly assigned to four groups (n = 10 per group) to receive daily oral dovitinib (40 mg/kg) or vehicle for 4 weeks. Mice were sacrificed by carbon dioxide inhalation when reached cut-off volume of 2 cm3. The day of sacrifice was considered the day of death from disease for the purpose of survival evaluation. Differences among survival duration of mice in each group were determined by log-rank test. Right, tumor growth at the median survival duration of the control group (65 and 40 days for NCI-N87 and N87-TR4, respectively). Error bars are 95% CIs. (NCI-N87 tumor volume: control, mean = 1,921 mm3, 95% CI = 1,836–2,007; dovitinib, mean = 1,813, 95% CI = 1,574–2,051; survival: control versus dovitinib, median survival = 65 days versus 62 days, HR = 0.7750, 95% CI = 0.2664–2.255, P = 0.64; N87-TR4 tumor volume: control, mean = 1,910, 95% CI = 1,459–2,362; dovitinib, mean = 1,086, 95% CI = 803–1,369; survival: control versus dovitinib, median survival = 40 versus 63 days, HR = 0.1413, 95% CI = 0.04611–0.4331, P = 0.0006) **, P < 0.01. C, Six mice bearing N87-TR4 gastric tumors were randomly allocated (n = 3 per group) to receive 40 mg/kg daily oral dovitinib or vehicle as control. Treatments were continued for 4 weeks. Tumors were excised 1 day after the end of treatments. Paraffin-embedded gastric tumor sections stained immunohistochemically with antibodies against pAKT.
Validation of the FGFR3/AKT axis as an escape pathway for trastuzumab resistance in gastric cancer patients
To corroborate the clinical relevance of our findings, we identified three patients affected by advanced gastric cancer who received trastuzumab-containing treatments and for which there were available both pretreatment and postprogression bioptic samples. The clinical history for each patient is depicted in Fig. 6. Immunohistochemical analyses were performed on both pretreatment and postprogression samples to determine the expression of FGFR3, activated AKT, and of the EMT marker ZEB1. We demonstrated that in all three patients, samples taken at trastuzumab progression exhibited significantly higher FGFR3 expression than their matched pretreatment samples. Consistently, we observed higher expression levels of activated AKT and ZEB1 in postprogression bioptic samples when compared with their respective pretreatment samples (Fig. 6).
Analysis of pretreatment and postprogression biopsies from gastric cancer patients and clinical history. Paraffin-embedded sections from patients treated with trastuzumab who underwent relapse were immunohistochemically stained with antibodies against FGFR3, p-AKT, and ZEB1 (scale bar, 200 μm; 10× magnification).
Discussion
In this study, we sought to identify the molecular mechanisms responsible for the resistance of gastric cancer to anti-HER2 treatment trastuzumab. We demonstrated that trastuzumab induces the selection of gastric cancer cells overexpressing FGFR3 and its specific ligand FGF9. This autocrine loop activates the PI3K/AKT/mTOR signaling pathway, sustaining, in turn, tumor growth and a more aggressive EMT phenotype. To our knowledge, this is the first study to provide evidences that targeting the kinase activity of FGFR3 could be a valid approach to modulate acquired resistance to trastuzumab in gastric cancer.
FGFR3 is a member of a family of four tyrosine kinase receptors that contribute to carcinogenesis by stimulating tumor proliferation, survival, and neoangiogenesis (20). Recent studies provide evidence for an increasing role or FGFR signaling as a key mediator of resistance to several anticancer therapies (21, 22), including the dual HER2 and EGFR inhibitor lapatinib (23). In our study, we demonstrated that FGFR3 is overexpressed in gastric cancer tumor models selected for resistance to trastuzumab. Most importantly, we confirmed our finding by demonstrating for the first time the overexpression of FGFR3 in paired pretreatment and postprogression bioptic samples from patients affected by advanced gastric cancer that relapsed upon therapies with trastuzumab.
FGF9 is one of the members of the heparin binding polypeptide ligands of the FGF family, which presents a unique affinity (Kd: 0.25 nmol/L) for FGFR3 (24). FGF9 was initially identified as an important secreted mediator for the epithelial-to-mesenchymal cell signaling during embryonic development (25). In this study, we measured a significantly higher expression of FGF9 in trastuzumab-resistant gastric cancer cell lines than in the sensitive parental line, thus supporting the existence of an autocrine loop between FGF9 and overexpressed FGFR3.
Either the interaction of FGFR3 with the specific ligand FGF9 or the ligand-independent dimerization induced by the overexpression of FGF receptors may lead to the transphosphorylation of their tyrosine kinase domains, activating, in turn, the MAPK and the PI3K/AKT/mTOR signaling pathways (20). The continued activation of the PI3K/AKT/mTOR signaling pathway is the most common alteration associated with the resistance to anti-HER2 therapies in breast cancer (26, 27). More recent clinical evidences suggested that activating mutations in the PI3KCA gene encoding the p110a catalytic subunit of the PI3K enzyme (28, 29), rather than the loss or inactivating mutations of phosphatase and tensin homolog on chromosome 10 (PTEN; ref. 30), might be important biomarkers to identify breast cancer patients resistant to anti-HER2 therapies. However, less is known about the role of FGFR pathway in anti-HER2 resistance in breast cancer. It has been demonstrated that FGFR2 is a pivotal molecule for the survival of lapatinib-resistant cells, suggesting that a switch of addiction from the HER2 to the FGFR2 pathway enable cancer cells to become resistant to HER2-targeted therapy (23). In our study, we demonstrated that the AKT signaling pathway is significantly activated in trastuzumab-resistant gastric cancer models, if compared with sensitive control. More significantly, we demonstrated for the first time the overactivation of the AKT signaling pathway in postprogression bioptic samples from gastric cancer patients receiving therapies with trastuzumab, if compared with their respective pretreatment samples. Most importantly, we demonstrated that the overactivation of the AKT signaling pathway in trastuzumab-resistant gastric cancer models is dependent on the overexpression of FGFR3 and can be modulated by the inhibition of this receptor.
EMT is a transdifferentiation process that converts tumor cells with an epithelial phenotype into highly motile mesenchymal cells (31). The PI3K/AKT/mTOR signaling pathway demonstrated an essential role in mediating EMT, thought the activation of both mTOR complex 1 (mTORC1; ref. 32), and mTORC2 (33, 34). In our study, we found an EMT phenotype and increased cell motility in trastuzumab-resistant gastric cancer models. We demonstrated that the induction of this phenotype parallels the overactivation of the AKT signaling pathway, and is sustained by the autocrine loop between overexpressed FGF9 and FGFR3, as it can be modulated by the FGFR3 inhibitor dovitinib. Of clinical relevance, we showed for the first time the overexpression of the EMT marker ZEB1 in postprogression bioptic samples from gastric cancer patients receiving therapies with trastuzumab, if compared with their respective pretreatment samples.
Dovitinib (TKI258, Novartis) is a multitargeted receptor tyrosine kinase inhibitor, which strongly binds to FGFR3 and inhibits its phosphorylation (NCI Drug Dictionary). This agent is currently explored as a monotherapy in the second- and third-line setting [NCT01719549], and in combination with docetaxel in the second-line setting (NCT01921673) for patients with gastric cancer. In our study, we showed that dovitinib significantly reduced the activation of AKT and the expression of ZEB1, inhibiting, in turn, the growth and the migratory properties of trastuzumab-resistant cells. Of translational relevance, we showed the therapeutic efficacy of treatment with dovitinib in an in vivo model of trastuzumab-resistant N87-TR4 gastric cancer xenografts.
This study, however, had some limitations. The orthotopic xenograft tumor models used are limited by the artificial microenvironment in an immunocompromised host and could not faithfully recapitulate the histopathologic features of the human disease, thus, impairing the representativeness of the results observed to human patients. Moreover, the four different TR cell lines developed and analyzed in this study derive from NCI-N87 cells, thus, we could not exclude that different mechanisms of resistance could have been emerged by starting the selection process from different HER2-positive gastric cancer cell models. In this regard, we corroborated the clinical relevance of our findings for human gastric cancer patients by detecting for the first time a significantly higher expression of FGFR3, activated AKT, and of the EMT marker ZEB1 in posttreatment biopsies when compared with their respective pretreatment biopsies from three gastric cancer patients receiving trastuzumab-based treatments. The limited number of patients' samples available and the potential heterogeneity in the primary tumor and among the different sites of metastasis analyzed could represent additional potential limitations of this study.
Two prior studies reported the development of trastuzumab-resistant NCI-N87 cells (35, 36). Eto and colleagues established a resistant cell line by culturing NCI-N87 cells in vitro in the presence of increasing concentration of trastuzumab from 5 to 100 μg/mL along 6 months (35). Similarly, Yang and colleagues selected in vitro another resistant cell line by growing NCI-N87 over a 1-year period in increasing concentration of trastuzumab up to 10 μg/mL. During the development of the resistance, NCI-N87 cells underwent EMT, as shown by the increase of ZEB1 expression and of their migratory properties. However, trastuzumab was still able to suppress the phosphorylation of AKT in this cell model, and the activation of an IL6/STAT3/Jagged-1/Notch positive feedback signaling loop was indicated as associated with the acquisition of trastuzumab resistance (36). In our study, we developed four different trastuzumab-resistant models through, not an in vitro, but an in vivo process of selection by using a clinically relevant model of orthotopic growth of tumors in the mouse stomach. We verified the resistance of these models to trastuzumab in vitro, up to 400 μg/mL, as well as in vivo. Although we confirmed the acquisition of an EMT phenotype and an overexpression of ZEB1 in the trastuzumab-resistant models when compared with sensitive cells, we did not measure any significant overexpression of IL6 or Jagged-1 in our models (data not shown).
In conclusion, we propose a model in which trastuzumab induces the selection of gastric cancer cells overexpressing FGFR3 and its ligand FGF9. This autocrine signaling loop sustains the activation of the PI3K/AKT/mTOR signaling pathway, and in turn, tumor growth and a more aggressive EMT phenotype. Most importantly, our study provides the preclinical rationale to investigate the inihibition of FGFR3 as second-line treatment strategy in gastric cancer patients refractory to first-line trastuzumab-containing therapies.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: G. Piro, C. Carbone, S. Giacopuzzi, D. Melisi
Development of methodology: G. Piro, I. Cataldo, A. Sbarbati, D. Melisi
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): G. Piro, C. Carbone, F.D. Nicolantonio, S. Giacopuzzi, G. Aprile, F. Simionato, F. Boschi, M. Zanotto, R. Santoro, G. de Manzoni, A. Scarpa, D. Melisi
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): G. Piro, C. Carbone, I. Cataldo, F.D. Nicolantonio, G. Aprile, V. Merz, A. Scarpa, D. Melisi
Writing, review, and/or revision of the manuscript: G. Piro, F.D. Nicolantonio, G. Aprile, V. Merz, A. Scarpa, G. Tortora, D. Melisi
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): F. Simionato, M. Mihaela Mina, G. de Manzoni, D. Melisi
Study supervision: G. de Manzoni, G. Tortora, D. Melisi
Grant Support
This work was supported in part by the Associazione Italiana per la Ricerca sul Cancro (AIRC) Start-Up no. 10129, and 5 per mille no. 10016 (to D. Melisi), and by the AIRC grants IG 11930, 5 per mille 12182, 12214, and PRIN no. 2009 × 23L78_005 (to G. Tortora). Partial support was also provided by grant Farmacogenomica' 5 per mille 2009 MIUR from Fondazione Piemontese per la Ricerca sul Cancro—ONLUS (to F.D. Nicolantonio) and Fondo per la Ricerca Locale (ex 60%), Università di Torino, 2014 (to F.D. Nicolantonio).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Acknowledgments
We thank Licia Montagna for the technical execution of immunohistochemistry. Part of the work was performed at the Laboratorio Universitario di Ricerca Medica (LURM) Research Center, University of Verona.
Footnotes
Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).
- Received January 29, 2016.
- Revision received May 26, 2016.
- Accepted June 1, 2016.
- ©2016 American Association for Cancer Research.
References
Volume 22, Issue 24
