If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Correspondence Address correspondence to: Shane P. Duggan, Rm 5.408 Life Sciences Institute, Health Sciences Mall, University of British Columbia, Vancouver, BC Canada V6T 1Z3.
Department of Medicine, Division of Gastroenterology, University of British Columbia, Vancouver, British Columbia, CanadaLife Science Institute, University of British Columbia, Vancouver, British Columbia, CanadaCentre for Complement and Inflammation Research, Imperial College London, London, United Kingdom
Department of Clinical Medicine, Trinity Translational Medicine Institute, Trinity College Dublin, Dublin, IrelandDepartment of Gastroenterology, St James’ Hospital, Dublin, Ireland
Department of Medicine, Division of Gastroenterology, University of British Columbia, Vancouver, British Columbia, CanadaLife Science Institute, University of British Columbia, Vancouver, British Columbia, CanadaDepartment of Clinical Medicine, Trinity Translational Medicine Institute, Trinity College Dublin, Dublin, Ireland
Effective therapeutic approaches are urgently required to tackle the alarmingly poor survival outcomes in esophageal adenocarcinoma (EAC) patients. EAC originates from within the intestinal-type metaplasia, Barrett’s esophagus, a condition arising on a background of gastroesophageal reflux disease and associated inflammation.
Methods
This study used a druggable genome small interfering RNA (siRNA) screening library of 6022 siRNAs in conjunction with bioinformatics platforms, genomic studies of EAC tissues, somatic variation data of EAC from The Cancer Genome Atlas data of EAC, and pathologic and functional studies to define novel EAC-associated, and targetable, immune factors.
Results
By using a druggable genome library we defined genes that sustain EAC cell growth, which included an unexpected immunologic signature. Integrating Cancer Genome Atlas data with druggable siRNA targets showed a striking concordance and an EAC-specific gene amplification event associated with 7 druggable targets co-encoded at Chr6p21.1. Over-representation of immune pathway–associated genes supporting EAC cell growth included leukemia inhibitory factor, complement component 1, q subcomponent A chain (C1QA), and triggering receptor expressed on myeloid cells 2 (TREM2), which were validated further as targets sharing downstream signaling pathways through genomic and pathologic studies. Finally, targeting the triggering receptor expressed on myeloid cells 2-, C1q-, and leukemia inhibitory factor–activated signaling pathways (TYROBP–spleen tyrosine kinase and JAK-STAT3) with spleen tyrosine kinase and Janus-activated kinase inhibitor fostamatinib R788 triggered EAC cell death, growth arrest, and reduced tumor burden in NOD scid gamma mice.
Conclusions
These data highlight a subset of genes co-identified through siRNA targeting and genomic studies of expression and somatic variation, specifically highlighting the contribution that immune-related factors play in support of EAC development and suggesting their suitability as targets in the treatment of EAC.
This study defines immune-associated factors supportive of esophageal adenocarcinoma cell growth through small interfering RNA screening and pathologic and functional studies. Fostamatinib R788, targeting Janus kinase/signal transducer-and-activator of transcription and TYRO protein tyrosine kinase binding protein/spleen tyrosine kinase/AKT serine-threonine kinase pathways, induced esophageal adenocarcinoma cell death and reduced tumor burden in xenografted mice.
The rates of esophageal adenocarcinoma (EAC) have increased steadily over the past number of decades at an average of 2.8% per year in the most recent North American data.
Comparatively, the incidence rates of more prevalent cancers such as lung, prostate, colon, and breast cancer have seen marginal decreases in recent years.
Current curative approaches for EAC typically involve surgical esophagectomy or endoscopic resection in early T1 cancers, or chemotherapy plus esophagectomy in which limited invasion has occurred. The survival rate for EAC is 18% at 5 years compared with 90% and 91% for breast and colon, respectively.
Therefore, preventative and targeted therapies for EAC are urgently needed to increase survival rates and reduce tumor burden.
Targeted therapeutics for EAC have been hindered by the heterogeneous nature of patient-specific somatic mutations and copy number variations, the differing spectrum of mutations between patients and a poorly defined map of the molecular drivers of this cancer type. The most prevalent mutations or copy number variations have been reported in presently nonactionable tumor-suppressor genes such as TP53, ELMO1, CDKN2A, SMAD4, and ARID1A, rather than driver oncogenes,
The biomarker TP53 divides patients with neoadjuvantly treated esophageal cancer into 2 subgroups with markedly different outcomes. A p53 Research Group study.
Cancer Genome Atlas Research Network; Analysis Working Group: Asan University; BC Cancer Agency; Brigham and Women’s Hospital; Broad Institute; Brown University; Case Western Reserve University; Dana-Farber Cancer Institute; Duke University; Greater Poland Cancer Centre; Harvard Medical School; Institute for Systems Biology; KU Leuven; Mayo Clinic; Memorial Sloan Kettering Cancer Center; National Cancer Institute; Nationwide Children’s Hospital; Stanford University; University of Alabama; University of Michigan; University of North Carolina; University of Pittsburgh; University of Rochester; University of Southern California; University of Texas MD Anderson Cancer Center; University of Washington; Van Andel Research Institute; Vanderbilt University; Washington University; Genome Sequencing Center: Broad Institute; Washington University in St. Louis; Genome Characterization Centers: BC Cancer Agency; Broad Institute; Harvard Medical School; Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins University; University of North Carolina; University of Southern California Epigenome Center; University of Texas MD Anderson Cancer Center; Van Andel Research Institute; Genome Data Analysis Centers: Broad Institute; Brown University:; Harvard Medical School; Institute for Systems Biology; Memorial Sloan Kettering Cancer Center; University of California Santa Cruz; University of Texas MD Anderson Cancer Center; Biospecimen Core Resource: International Genomics Consortium; Research Institute at Nationwide Children’s Hospital; Tissue Source Sites: Analytic Biologic Services; Asan Medical Center; Asterand Bioscience; Barretos Cancer Hospital; BioreclamationIVT; Botkin Municipal Clinic; Chonnam National University Medical School; Christiana Care Health System; Cureline; Duke University; Emory University; Erasmus University; Indiana University School of Medicine; Institute of Oncology of Moldova; International Genomics Consortium; Invidumed; Israelitisches Krankenhaus Hamburg; Keimyung University School of Medicine; Memorial Sloan Kettering Cancer Center; National Cancer Center Goyang; Ontario Tumour Bank; Peter MacCallum Cancer Centre; Pusan National University Medical School; Ribeirão Preto Medical School; St. Joseph’s Hospital &Medical Center; St. Petersburg Academic University; Tayside Tissue Bank; University of Dundee; University of Kansas Medical Center; University of Michigan; University of North Carolina at Chapel Hill; University of Pittsburgh School of Medicine; University of Texas MD Anderson Cancer Center; Disease Working Group: Duke University; Memorial Sloan Kettering Cancer Center; National Cancer Institute; University of Texas MD Anderson Cancer Center; Yonsei University College of Medicine; Data Coordination Center: CSRA Inc.; Project Team: National Institutes of Health Integrated genomic characterization of oesophageal carcinoma.
Common amplifications have been reported in KRAS (21%), ERBB2/HER2 (19%), and EGFR (16%), and deletions in SMAD4 and CDKN2A. Targeted therapies based on these amplifications and mutations have largely been ineffective, except in the case of HER-2–positive gastric and junctional tumors, in which trastuzumab improved outcomes in metastatic patients.
Oesophageal Cancer Clinical and Molecular Stratification (OCCAMS) Consortium Mutational signatures in esophageal adenocarcinoma define etiologically distinct subgroups with therapeutic relevance.
Therefore, further cellular targets independent of, or harmonizing with, the mutational status of EAC are warranted to expand the repertoire of potential targeted therapeutic approaches.
The development of EAC is believed to progress from within the metaplastic lesion, Barrett’s esophagus (BE), supported by chronic inflammatory gastroesophageal reflux disease (GERD) in which chronic reflux of gastric and bile acids into the esophagus induces epithelial inflammation, ulceration, and epithelial erosion.
However, cytokine-mediated structural alterations within the esophageal epithelium may also arise without the need for reflux-mediated epithelial erosion.
Our laboratory and others have shown that esophageal epithelial cells release proinflammatory cytokines such as interleukin (IL)6, IL8, and IL1β in response to acid and bile acids that may become overexpressed in EAC partially as a result of reduced tribbles pseudokinase 3 (TRIB3) expression.
An integrative genomic approach in oesophageal cells identifies TRB3 as a bile acid responsive gene, downregulated in Barrett's oesophagus, which regulates NF-kappaB activation and cytokine levels.
Ursodeoxycholic acid inhibits interleukin 1 beta [corrected] and deoxycholic acid-induced activation of NF-kappaB and AP-1 in human colon cancer cells.
These cytokines result in greater numbers of immune cells infiltrating the epithelium where they initially are associated with basal cell and papillary hyperplasia without loss of surface cells.
However, the mechanistic links between this inflammatory milieu, the metaplastic process, and EAC cell growth or oncogenesis is poorly detailed and has yet to result in actionable targets for therapeutic intervention in patients presenting with EAC. Thus, our understanding of inflammatory or immune-mediated promotion of EAC is incomplete and requires further exploration to both treat and prevent this deadly disease. Herein, using functional genomic screens we show a distinct targetable EAC-specific immune-pathway–associated signature driving sustained oncogenic growth.
Materials and Methods
Cell Lines and Cell Culture
FLO-1 (11012001), SK-GT-4 (11012007), and OE33 (96070808) cell lines were obtained from The European Collection of Authenticated Cell Cultures (ECACC). CP-D (CP-18821) cell lines were obtained from the American Type Culture Collection (ATCC). Validation of findings was performed in multiple esophageal EAC cell lines (SKGT4, FLO-1, and OE-33), which have been obtained from commercial sources and tested by STR profiling. CP-D cells were cultured in bronchial epithelial base/growth media (Lonza, Mississauga, Canada) with bullet kit additives and 5% fetal bovine serum (Gibco, Invitrogen, Carlsbad, CA). SKGT4, OE-33, and FLO-1 cells were maintained in RPMI1640 without L-glutamine media with 10% fetal bovine serum (Gibco, Invitrogen) and added penicillin/streptomycin/L-glutamine (Gibco). Cells were fed at 48 hours, and subcultured at 72 hours, before overconfluence, using trypsin-EDTA (Gibco) in Hank's balanced salt solution for detachment. Treatments with recombinant leukemia inhibitory factor (LIF) (30 ng/mL, ab57665; Abcam), IL6 (30 ng/mL; R&D Systems, Minneapolis, MN), native C1q (75 μg/mL; Sigma, Oakville, Ontario), and IL6 blocking antibody (AB206NA) experiments were performed in either 6- or 96-well cell culture plates for protein and viability experiments, respectively.
Clinical Tissue Collection, RNA Extraction, and Immunohistochemistry
Sample size for messenger RNA (mRNA) expression analysis of immune-related factors in BE (N = 17) and EAC (N = 23) patients and controls (N = 17) was determined through previous use of this clinical cohort in published reports.
The characterization of an intestine-like genomic signature maintained during Barrett's-associated adenocarcinogenesis reveals an NR5A2-mediated promotion of cancer cell survival.
Written, oral, and informed consent was provided by all patients who contributed tissue biopsy specimens for gene expression aspects of this study. Ethical approval for this study was granted by the Research Ethics Committee of Tallaght Hospital and St. James’ Hospital (Dublin, Ireland) and performed in accordance with the associated guidelines. The sample size for immune-histopathologic staining was determined by tissue availability and from previous use of BE cohorts (n = 28) and was scored by 2 independent pathologists. The retrospective cohort was processed with a Leica BOND-MAX automated immunohistochemistry processing unit (Leica, Concord, Ontario, Canada) using primary antibodies to C1QA (ab76425; Abcam, Cambridge, United Kingdom) and triggering receptor expressed on myeloid cells 2 (TREM2) (HPA012571; Sigma) at a concentration of 1:50 and 1:100, respectively. Institutional ethics approval was obtained for the use of EAC and control tissue from the University of British Columbia research ethics board (H17-01285). Slides were scanned with an Aperio slide scanner (Leica) and digitally magnified images were produced in ImageScope (Leica).
Small Interfering RNA–Mediated Silencing
Small interfering RNA (siRNA) smart pools, siGenome-SMARTpool (Thermo Fischer) for initial screening, or siON-TARGETplus (Thermo Fischer) in all other validation experiments were delivered to cells using Dharmafect transfection reagent (DF1-4; (Thermo Fischer)) using a reverse transfection protocol. Transfection reagent:siRNA complexes were added, as per the manufacturer’s instructions, to seeded but unattached cells in culture media (BEGM or RPMI) without fetal calf serum (FCS) or penicillin/streptomycin/L-glutamine, and allowed to incubate for 24 hours before replacement of the media with media containing FCS for 72 hours before analysis. Experiments performed in 6-, 24-, and 96-well cell culture plates were seeded with 2 × 105, 5 × 104, and 3 × 103 cells, respectively, in reverse transfection protocols. This method used siRNA (25 nmol/L) combined with 0.2 μL Dharmafect transfection reagent (Thermo Fischer).
High-Throughput siRNA Library Screening
The objectives of this study were to define therapeutically targetable regulators of esophageal cancer growth and survival. The initial siRNA library screen was performed in CP-D cells (n = 8 with replicate plate discarded, plates 2–4 used for 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide [MTT] assays). The human druggable siRNA library consisting of 6022 siRNA pools (4 individual gene targeting siRNAs at 20 μmol/L) formatted in 76 library plates (96-well v bottom) was maintained and re-aliquoted into a v bottom 96-well daughter library plate (Thermo Fischer) containing siBuffer (Thermo Fischer) to a concentration of 5 μmol/L using a plate replicator within a biosafety cabinet (Matrix/Thermo Fischer). Each daughter plate provides enough siRNA for 8 assay plates at 25 nmol/L. Negative (nontargeting siRNA pool) and positive controls (siRNA pool targeting GATA6), 25 nmol/L each, were added to columns 1 and 2 of 96-well v bottom plates. Reverse siRNA transfection plates then were prepared from these daughter plates through the distribution of transfection cocktail (1:100 μL DF4 in culture media) throughout all 76 transfection plates using electronic multichannel pipette delivery. Each transfection plate was individually mixed by swirling and tapping after completion and then was incubated for 20 minutes before distribution into assay plates sequentially (n = 8 replicate plates, 20 μL per well). An automated multichannel dispenser (Wellmate, Thermo Fischer, Waltham, MA) was next used to distribute 80 μL culture media per well containing 5 × 103 cells to create the assay plates. Cell culture media was replaced at 24 hours with full media containing FCS and incubated for a further 72 hours before the addition of 10 μL MTT reagent at 93 hours and dimethyl sulfoxide containing detergent reagent at 96 hours after a 3-hour incubation at 37°C. Plates were allowed to incubate at room temperature in the dark overnight before measuring absorbance at 570 nm. Screening results were analyzed in R statistical software using CellHTS2
(https://bioconductor.org) platform for statistical analysis of high-throughput RNA interference screens. The median of negative controls was used for within-plate normalization and furthermore to perform interplate normalization of replicate plates. The screening Z-factor was calculated as follows:
Screening results were ranked by robust Z score as calculated in RNAither with a cut-off value of -1.645 and a cell viability score less than 50%. Outliers were only removed with clear justification, such as an infection in a replicate well. Bioinformatic interpretation of siRNA screening results linked significant regulators of CP-D cell growth with immune-related pathways using n = 3 informatic platforms. Gene ontology–enrichment analysis was performed using Fatigo in Babelomics 5.0 (http://babelomics.bioinfo.cipf.es/) with identification list vs background reference druggable genome reference gene list and ClueGO gene ontology networking with Cytoscape plugin (http://apps.cytoscape.org/apps/cluego).
Protein Expression Analysis
Samples for protein expression analysis by Western blot were collected from 6-well plates at 72 hours for validation of siRNA-mediated silencing effects by scraping in ice-cold phosphate-buffered saline (PBS). The cells were then pelleted by centrifugation before resuspension in M-PER lysis solution (Pierce, Thermo Fischer) and incubation on ice for 10 minutes as per the manufacturer’s instructions. Protein concentration was determined by micro–bicinchoninic acid assay at 562 nm (Pierce, Thermo Fischer) to allow equal loading of protein in sodium dodecyl sulfate–polyacrylamide (SDS-PAGE) gel electrophoresis. Proteins separated by SDS-PAGE were transferred to a polyvinylidene difluoride membrane in semidry conditions in a power blotter (Pierce, Thermo Fischer) as previously described.
An integrative genomic approach in oesophageal cells identifies TRB3 as a bile acid responsive gene, downregulated in Barrett's oesophagus, which regulates NF-kappaB activation and cytokine levels.
The characterization of an intestine-like genomic signature maintained during Barrett's-associated adenocarcinogenesis reveals an NR5A2-mediated promotion of cancer cell survival.
Membranes were collected in PBS followed by incubation in milk protein–based blocking solution for 30 minutes. Primary antibodies to LIF, C1QA (ab108325; Abcam), TREM2 (Sigma), TYROBP or DAP12 (ab124834; Abcam), spleen tyrosine kinase (SYK) (D3Z1E; Cell Signaling), and p-SYK (Cell Signaling) were diluted in blocking solution at 1:1000, 1:500, 1:250, 1:300, 1:300, and 1:200, respectively, and incubated with the appropriate membranes for 24 hours at 4°C followed by washing with PBS-Tween and further room temperature incubation with appropriate secondary horseradish-peroxidase–conjugated antibody at a 1:2000 dilution for 2 hours. Chemiluminescent signals were detected using both radiograph film (TREM2, DAP12, SYK, p-SYK, C1QA and Actin) and cooled LCD camera lightbox (LIF, pSTAT3, and Actin) (Bio-Rad).
Gene Expression Microarray Re-analysis and Bioinformatics
The characterization of an intestine-like genomic signature maintained during Barrett's-associated adenocarcinogenesis reveals an NR5A2-mediated promotion of cancer cell survival.
The expression of genes whose siRNA-mediated silencing resulted in greater than 50% cell death and a Z score level <-1.645 were examined in the gene expression microarray (GEM) data sets of BE and EAC tissues. These GEM studies were analyzed at a Mann–Whitney significance level of P < .001. Results were hierarchically clustered by genes and samples using Euclidean distance and viewed through the data set of Kim et al
Cellular Assays of Viability, Apoptosis, and Gene Expression
Measurements of cell viability were performed after transfection of cells with the siRNA and/or treatment with recombinant or native proteins using the MTT assay (LONZA, Mississauga, Canada). Briefly, when recombinant (LIF and IL6) or native proteins (C1q and anti-IL6) were used, at the listed concentrations, these were added at 72 hours after transfection with siRNA followed by end point assays of viability (N = 3) or gene expression (N = 3) at 96 hours. In viability assays using siRNA-mediated gene inhibition, 10 μL of MTT reagent (30-1010K; ATCC) was added to each well 93 hours after transfection using a multichannel repetman (Thermo Scientific), and then incubated for 3 hours at 37°C. Detergent reagent (100 μL, 30-1010K; ATCC) was then added using a multichannel repetman (Rainin, Mississauga, Canada). Plates then were covered with tin foil to protect them from light and left at room temperature overnight. Cell viability was calculated by reading the absorbance at 570 nm using a spectrophotometer (Tecan Spark 10 M, Männedorf, Switzerland). In experiments using fostamatinib R406, cells were treated with R406 or vehicle dimethyl sulfoxide at the listed concentration for 24 hours followed be measurement of cell viability. Cellular growth arrest and apoptosis after R406 treatment was determined through flow-cytometric cell-cycle analysis using fluorescein isothiocyanate–labeled bromodeoxyuridine incorporation and a 7-AAD staining kit (BD Biosciences, San Jose, CA), as per the manufacturer’s instructions, and analyzed using a BD FACSVerse (BD Biosciences, San Jose, CA). Cells undergoing growth arrest were defined by a decreased percentage of cells in the S phase (R4) and a gain of those in G0/G1 (R3). Cells undergoing apoptosis were defined by a decreased percentage of cells in the S phase and a gain in both G0/G1 (R3) and sub-G1 populations (R6). Controls including deoxycholate treatment (200 μmol/L) and cells in culture media lacking FCS (resting-FCS) were used to define cell populations in apoptosis and growth arrest, respectively.
An integrative genomic approach in oesophageal cells identifies TRB3 as a bile acid responsive gene, downregulated in Barrett's oesophagus, which regulates NF-kappaB activation and cytokine levels.
The characterization of an intestine-like genomic signature maintained during Barrett's-associated adenocarcinogenesis reveals an NR5A2-mediated promotion of cancer cell survival.
gene expression analysis was performed by real-time relative reverse-transcription polymerase chain reaction using a 7900HT thermocycler with predesigned real-time primers and probes (Applied Biosystems, Foster City, CA). Relative fold changes in mRNA expression levels were calculated using the delta delta Ct method and glyceraldehyde-3-phosphate dehydrogenase as the denominator control gene. Mean values from control resting, nontargeting siRNA–transfected cells or esophageal squamous tissues were used as a base level from which to calculate altered expression. Nonparametric Mann–Whitney testing appropriate for real-time reverse-transcription polymerase chain reaction data was used to examine the significance of the observed changes.
Murine Subcutaneous Xenografts
Animal studies were conducted in accordance with the Imperial College Animal Welfare and Ethical Review Body of the UK Home Office for Laboratory Animal Care regulations following the Animal Research: Reporting In Vivo Experiments guidelines.
Six-week-old female NOD/SCIDgamma mice were obtained from Harlan Laboratories (Cambridge, UK) and maintained in individually vented cages at 5 animals per cage and water and food were provided ad libitum, as per institutional guidelines. At t0, 1 × 106 OE-33 and FLO-1 cells were injected subcutaneously into the left and right flanks, respectively, of n = 15 NOD/SCIDγ mice and allowed to implant for 5 days before the start of daily intraperitoneal treatment with fostamatinib R788. Cages of n = 5 animals were divided into 40 mg/kg R788, 80 mg/kg R788, or vehicle (carboxymethylcellulose [CMC])-only treatment groups, as previously reported by Suljagic et al,
The Syk inhibitor fostamatinib disodium (R788) inhibits tumor growth in the Emu-TCL1 transgenic mouse model of CLL by blocking antigen-dependent B-cell receptor signaling.
and tumors were measured by calipers every 3 days until maximal humane end points were reached, in this case when tumors reached 1500 mm3 in any direction, which occurred by 33 days. All authors had access to the study data and reviewed and approved the final manuscript
Statistical Analysis
The Student t test was used for statistical analysis of biologically replicated (N = 3) functional experiments. Nonparametric Mann–Whitney testing was used for analysis of significance in real-time gene expression experiments performed with both biological (N = 3–23) and technical replicates (N = 3).
An integrative genomic approach in oesophageal cells identifies TRB3 as a bile acid responsive gene, downregulated in Barrett's oesophagus, which regulates NF-kappaB activation and cytokine levels.
The characterization of an intestine-like genomic signature maintained during Barrett's-associated adenocarcinogenesis reveals an NR5A2-mediated promotion of cancer cell survival.
Statistical interpretation of siRNA screening results was achieved through Z score ranking of normalized median values. One-way analysis of variance and linear regression was used in significance testing of murine xenograft studies performed with GraphPad Prism software (La Jolla, CA).
Results
Druggable Genome Screening of EAC Cell Growth After siRNA-Mediated Silencing
Previous publications from our laboratory showed that silencing of the GATA6 transcription factor resulted in significant EAC cell growth arrest and apoptosis.
The characterization of an intestine-like genomic signature maintained during Barrett's-associated adenocarcinogenesis reveals an NR5A2-mediated promotion of cancer cell survival.
Positive and negative controls were tested for reproducibility in mock screening protocols. siRNA pools targeting GATA6 were used as a positive control in siRNA library screening of cell viability using the high-grade dysplastic (HGD) cell line CP-D, which reflects many of the pathways of activation in the majority of EAC cell lines. A full-scale, high-throughput siRNA library screen of CP-D cell line viability was implemented using a druggable genome siRNA library consisting of 6022 individual gene-targeting siRNA pools after optimization of screening platform and controls. The performance of this screen was assessed through analysis of the reproducibility between positive (siGATA6; n = 3 technical and n = 3 biological per genome library plate [n = 76]) and negative controls (nontargeting siRNA pools), both in the same siGenome sequence format as the druggable genome library. A clear separation of positive and negative controls was achieved resulting in a screening metric Z-factor of 0.67, indicative of high-quality screening performance. The observed viability data derived from the siRNA screen was distributed normally, permitting the use of standard Z score and robust Z score cut-off values of Z < -1.645, resulting in 299 and 252 gene transcript–targeting siRNA pools, respectively, that significantly affected EAC cell viability (Figure 1A–D). The mRNA transcripts targeted by these siRNAs included nucleic acid binding proteins, signaling molecules, hydrolases, receptors, and cytoskeletal proteins (Figure 1D). As might be expected, key nuclear and nucleic acid binding proteins such as RAD51, SYMPK, and RAN, with obvious oncogenic roles, achieved high Z score rankings (Figure 1D) supportive of the approach and the quality of the siRNA screen performed. Gene transcripts (n = 30) were selected from within this list at different levels of significance (top, mid, bottom third) for verification of expression (mRNA) in EAC cells and validation of the effects of their silencing in EAC cells using alternate siRNA formats (siON-target-plus) than used in the original siRNA library screen (siGenome) (Figure 1E). Good concordance with the original screening data was achieved while using this alternative siRNA format (Figure 1E).
Figure 1Regulators of cell proliferation and survival in EAC cells as defined by siRNA druggable genome library screening. The druggable genome siRNA library (siGenome build) was used in conjunction with a high-throughput end point MTT viability assay of using the CP-D cell line as per the Materials and Methods section. The most significant siRNAs/genes/wells resulting in alterations to EAC cell viability were defined through robust Z score ranking (Z < -1.645) which was normally distributed. (A) Nonranked Z scores of siRNA pools from siRNA screening ordered according to plate number. (B) Z score ranked individual siRNA pools after normalization and removal of substandard wells. (C) Normal distribution of Z sore rankings. (D) The Z score ranked list of siRNA targeted genes (Z < -1.645) with the strength of the heatmap color corresponding to level of statistical significance. (E) A selection of gene targeting siRNAs were validated (N = 3 biological replicates) in CP-D cells in separate independent experiments using an alternate siRNA format (siOn-target plus build), showing good concordance with the original screening data and a significant decrease in the expression of the associated genes (Mann–Whitney P < .0001 for all genes tested) by comparison with nontargeting controls (nontargeting siRNA pool [siNT]). Pos ctrl, siRNA targeting GATA6; Neg ctrl, nontargeting siRNA; Rel. exp., relative expression of associated mRNA by real-time reverse-transcription polymerase chain reaction. **P < .01, ***P < .001, in Student's t test.
Somatic Gene Amplifications Associate With Druggable Genome siRNA-Targeted Genes
The relationship between regulators of EAC cell growth as defined in siRNA screening data and somatic variation observed in EAC and esophageal squamous cell carcinomas (ESCCs), as recently defined by The Cancer Genome Atlas (TCGA) Research Network, was next examined.
Cancer Genome Atlas Research Network; Analysis Working Group: Asan University; BC Cancer Agency; Brigham and Women’s Hospital; Broad Institute; Brown University; Case Western Reserve University; Dana-Farber Cancer Institute; Duke University; Greater Poland Cancer Centre; Harvard Medical School; Institute for Systems Biology; KU Leuven; Mayo Clinic; Memorial Sloan Kettering Cancer Center; National Cancer Institute; Nationwide Children’s Hospital; Stanford University; University of Alabama; University of Michigan; University of North Carolina; University of Pittsburgh; University of Rochester; University of Southern California; University of Texas MD Anderson Cancer Center; University of Washington; Van Andel Research Institute; Vanderbilt University; Washington University; Genome Sequencing Center: Broad Institute; Washington University in St. Louis; Genome Characterization Centers: BC Cancer Agency; Broad Institute; Harvard Medical School; Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins University; University of North Carolina; University of Southern California Epigenome Center; University of Texas MD Anderson Cancer Center; Van Andel Research Institute; Genome Data Analysis Centers: Broad Institute; Brown University:; Harvard Medical School; Institute for Systems Biology; Memorial Sloan Kettering Cancer Center; University of California Santa Cruz; University of Texas MD Anderson Cancer Center; Biospecimen Core Resource: International Genomics Consortium; Research Institute at Nationwide Children’s Hospital; Tissue Source Sites: Analytic Biologic Services; Asan Medical Center; Asterand Bioscience; Barretos Cancer Hospital; BioreclamationIVT; Botkin Municipal Clinic; Chonnam National University Medical School; Christiana Care Health System; Cureline; Duke University; Emory University; Erasmus University; Indiana University School of Medicine; Institute of Oncology of Moldova; International Genomics Consortium; Invidumed; Israelitisches Krankenhaus Hamburg; Keimyung University School of Medicine; Memorial Sloan Kettering Cancer Center; National Cancer Center Goyang; Ontario Tumour Bank; Peter MacCallum Cancer Centre; Pusan National University Medical School; Ribeirão Preto Medical School; St. Joseph’s Hospital &Medical Center; St. Petersburg Academic University; Tayside Tissue Bank; University of Dundee; University of Kansas Medical Center; University of Michigan; University of North Carolina at Chapel Hill; University of Pittsburgh School of Medicine; University of Texas MD Anderson Cancer Center; Disease Working Group: Duke University; Memorial Sloan Kettering Cancer Center; National Cancer Institute; University of Texas MD Anderson Cancer Center; Yonsei University College of Medicine; Data Coordination Center: CSRA Inc.; Project Team: National Institutes of Health Integrated genomic characterization of oesophageal carcinoma.
Somatic variation (SV) at a frequency above 5% in all esophageal cancers was noted in 49 siRNA target genes from the druggable genome screen, at a cut-off value of Z < -1.645 in 2 independent clinical cohorts of esophageal cancers (EAC and ESCC) and gastric carcinomas (gastroesophageal junction and gastric carcinomas), respectively. Distinct clustering of gene-associated amplifications within specific patient samples could be defined through the oncoprint of both sequencing studies that, when curated, resulted in both cancer tissue- and patient-specific SV patterns (Figure 2). Amplifications associated with GTF2H4, GUCA1B, ABCC10, RUNX2, TREM2, POLH, and FRS3, whose silencing reduced the viability of EAC cells in screening data, were specific to EAC rather than ESCC (Figure 2A), significantly higher in gastroesophageal junctions than gastric carcinomas (Figure 2B), and were all co-encoded at chromosome 6p21.1. The remaining amplifications separated into 2 further clusters (TONSL, GPAA1, and TG; and KCNQ2, BIRC7, and TFAP2C) that were co-located at chromosome 8q24.3 and 20q13.33, respectively, but did not distinguish between EAC and ESCC cancer types (Figure 2). In total, 12 of 49 somatic events identified clustered at these 3 chromosomal locations. The chromosome location 6p21 is noted for the presence of the immune major histocompatibility complex locus recently associated with the development of BE,
Barrett's and Esophageal Adenocarcinoma Consortium (BEACON); Esophageal Adenocarcinoma GenEtics Consortium (EAGLE); Wellcome Trust Case Control Consortium 2 (WTCCC2) Genome-wide association studies in oesophageal adenocarcinoma and Barrett's oesophagus: a large-scale meta-analysis.
a precursor to EAC, in genome-wide association studies. Both 8q24.3 and 20q13.33 loci have not traditionally been associated with BE. Comparatively, SV in squamous lineage epithelial cell–associated transcription factors TP63/SOX2, which do not sustain cell growth according to screening data, was associated predominantly with ESCC, as previously described. Conversely, SV in ERBB2 and vascular endothelial growth factor A (VEGFA) was EAC-specific. Thus, results garnered directly from siRNA druggable genome library screening of HGD CP-D cells aligns with the genomic fingerprint observed in genomic studies of EAC and further supports the argument that these events may occur early in the EAC cancer sequence and are specific to EAC over ESCC. It further highlighted chromosomal changes occurring at noted immune-associated loci in a significant number of patients.
Figure 2Overlap between EAC-associated somatic variation and siRNA-targeted genes defined by druggable genome siRNA library screening of EAC cells. (A) TCGA data charting SV in EAC and ESCC viewed through a curated oncoprint of 13 siRNA-targeted genes with a Z score < -1.645 in screening data and a frequency of SV > 5%, selected for co-enrichment and clustering of the gene amplification events as determined by GISTIC through cBioportal (http://www.cbioportal.org/index.do). (B) Parallel analysis of TCGA data containing gastroesophageal junctional (GEJ) and GCs results in similar clustering of gene amplifications events in subsets of both of these tumors types. Amplification events of siRNA-targeted genes that altered EAC cell growth as per druggable genome screening data encoded at chromosome (C) 6p21.1, (D) 8q24.3, and (E) 20q13.33.
We subsequently used gene ontologies to identify the gene/protein/pathways predominantly enriched, as regulators of EAC cell growth, within the siRNA library screening data. Cytoscape and the associated plug-in ClueGO
were next used to explore gene ontology (GO) enrichments and biological networks within siRNA screening data with a robust Z score cut-off value of Z<-1.645 (Figure 1). This resulted in 202 GO terms defining the linked biological processes associated with these genes and accordingly were used further to derive the top-level relationships between the GOs based on the similarity of their associated genes. This process showed the significant presence of 6 immune-related GO clusters from a total of 10 network nodes (Figure 3). The most significant biological processes were those related to cytokine-mediated signaling pathways and cellular responses to molecules of bacterial origin, which together represented 52% of all total relationships (Figure 3). Further validation of these findings was achieved through similar analyses using the FatiGO module of the Bablelomics 5.0 platform for functional interpretation of genomics, resulting in broadly similar immune associations (data not shown). These approaches commonly emphasized a distinct and significant over-representation of immune system processes within the genes positive for regulation of CP-D cell growth. Secreted immune factors included C1QA, LIF, IL22, TGFB1, TGFB2, PROK1, THSB2, and CSF3. Membrane-bound immune factors included TREM2, TLR2, IL9R, ITGB3, ICAM1, and TNFRSF8. Cytoplasmic factors with immune relationships included NLRC4, TRAF2, and GAB2. Nuclear immune-related proteins included RELA, RUNX2, NR1H3, EGR1, CTCF, FOSL1, GATA1, MNDA, and BCL6. GERD-induced inflammation has long been associated with the development of esophageal metaplasia and EAC-associated factors.
Furthermore, recent evidence has proposed that low-grade epithelial inflammatory responses without inflammatory cell infiltrates, para-inflammation, may be induced by persistent tissue stress and contribute to cancer promotion in murine models.
Thus, the observed immune-pathway associations observed in CP-D cells may reflect the inflammatory origins of this disease.
Figure 3Gene ontology analysis of siRNA screening data. (A) ClueGO gene function clustering of gene ontologies, associated with siRNA gene targets, defines significant linkage to cytokine-mediated signaling pathway and cellular response to molecule of bacterial origin from the list of gene-targeting siRNAs defined in Figure 1. (B) GO networking of ClueGO results details the overlaps and interactions between the functional groupings. NF-κB, nuclear factor κB.
Overexpression of LIF, C1QA, and TREM2 in the BE-Associated Cancer Sequence
To examine the expression levels and thus targetability of the genes defined from siRNA library screening, we performed an integrative re-analysis of EAC gene expression microarray (GEM) studies, using methods previously described
The characterization of an intestine-like genomic signature maintained during Barrett's-associated adenocarcinogenesis reveals an NR5A2-mediated promotion of cancer cell survival.
(Figure 4A). This approach highlighted 62 siRNA-targeted genes that are putatively overexpressed in EAC tissues including immune-associated genes C1QA, TREM2, LIF, TGFB1, THBS2, HLA-G, HLA-DQB1, GDF15, IL32, IL15RA, C1S, ICAM1, and TRAF4. When this list is ranked by the strength of effects on EAC cell viability in siRNA screening data, 3 secreted factors were prominent within this list: LIF, C1QA, and triggering receptor expressed on myeloid cells (TREM2) (Figure 4B). Secreted factors may have future theranostic potential and thus were an important focus of this study. JAK-STAT signaling pathways, which are activated by IL6 and its family member LIF, have been shown to regulate C1q production in macrophages.
The first subcomponent of complement, C1q, triggers the production of IL-8, IL-6, and monocyte chemoattractant peptide-1 by human umbilical vein endothelial cells.
In vitro modulation of C1q mRNA expression and secretion by interleukin-1, interleukin-6, and interferon-gamma in resident and stimulated murine peritoneal macrophages.
Both C1Q and TREM2 mediate signaling events through TYROBP(DAP12)-SYK pathways and are associated commonly with innate and chronic inflammatory conditions.
Higher levels of LIF, C1QA, and TREM2 mRNA were observed in BE and EAC biopsy specimens when compared with squamous control tissues using an independent clinical cohort (P < .0001, Mann–Whitney) (Figure 4C–E). There are currently no sufficiently accessible databases of EAC patient GEM data with associated survival outcomes. However, a significant association between LIF (P = 5.4E-4), C1QA (P = 3.7E-7), and TREM2 (P = 5.5E-6) GEM expression pattern and patient survival could be observed in lung adenocarcinoma (P = 6E4; P = .05), with no associations in squamous lung, ovarian, or breast carcinomas as viewed through the kmplot database. Similarly LIF and C1QA, but not TREM2, expression was associated with poor outcome in gastric adenocarcinoma (www.kmplot.com).
The expression of LIF, C1QA, and TREM2 as novel esophageal associations was next localized by immunohistochemistry in BE and EAC tissues (Figure 5). Normal squamous esophageal tissue showed negligible expression of LIF, C1QA, and TREM2, with some expression in the underlying lamina propria and local immune cells (Figure 5). By comparison, consistent and strong staining of LIF, C1QA, and TREM2 was observed in the glandular epithelial cells and surfaces of BE in addition to isolated lymphoid cells within the interstitial spaces (Figure 5). Similarly, strong staining of LIF, C1QA, and TREM2 was observed in EAC tissues (Figure 5), further validating the preceding findings at the mRNA level.
Figure 4LIF, C1QA, and TREM2 mRNA expression in the EAC cancer sequence. (A) Hierarchical gene and condition clustering of siRNA gene targets (Z-score < –1.645 in siRNA library screening data) and tissues (EAC and squamous control tissues [SQ]) respectively, in gene expression microarray data (P < .001 in moderated t test and FC > 1.5) as previously reported.
The characterization of an intestine-like genomic signature maintained during Barrett's-associated adenocarcinogenesis reveals an NR5A2-mediated promotion of cancer cell survival.
Blue and red dendrites of conditional clustering signify EAC and SQ tissues respectively. Heat map yellow and blue coloring denote low and high expression levels respectively. In hierarchical clustering GEM data yellow and blue coloring reflects low and high expression respectively. (B) Top 10 genes overexpressed in EAC tissues ranked by the strength of effect achieved in siRNA screening data (robust Z score). Real-time reverse-transcription polymerase chain reaction analysis of (C) LIF, (D) C1QA, and (E) TREM2 mRNA expression in normal esophageal (SQ), BE, and EAC tissues. ****P < .0001 in Mann–Whitney testing for differences in mRNA tissue expression. GAPDH, glyceraldehyde-3-phosphate dehydrogenase.
Figure 5Immunohistochemistry confirms expression of LIF, C1QA, and TREM2 in BE and EAC tissues. Immunohistochemical staining pattern of (A) LIF, (B) C1QA, and (C) TREM2 in normal squamous (SQ), BE, and EAC tissues showing glandular membranous and perinuclear expression in epithelial cells in addition to expression by infiltrating or resident immune cell types. Images are from digitally zoomed fields of view from scanned slides using ImageScope (Leica). Original magnification: top, 10×; bottom, 20×.
LIF Maintains EAC Cell Survival, Constitutive Levels of phospho-STAT3, IL6, and C1QA
Previous work from our laboratory has shown that LIF was specifically produced by EAC cells in response to bile acid exposure with no such induction observed in squamous cell lines.
An integrative genomic approach in oesophageal cells identifies TRB3 as a bile acid responsive gene, downregulated in Barrett's oesophagus, which regulates NF-kappaB activation and cytokine levels.
In more recent murine models of esophageal dysplasia, reduced levels of dysplastic lesions have been observed in Il-6 and Il-1b double-knockout mice, implicating the IL6-STAT3 pathway in supporting EAC development.
The involvement of LIF, an IL6 family member, in EAC oncogenesis has not been explored. Thus, we next studied the role of LIF signaling in EAC cell lines. Silencing of LIF in multiple EAC cell lines resulted in significantly reduced levels of target mRNA (Figure 6A), secreted LIF protein (Figure 6B), and reduced cell viability in multiple EAC cell lines (Figure 6C), as suggested from siRNA screening data. The lower proliferative rate of LIF-silenced EAC cells could be returned to normal levels by treatment with recombinant LIF protein (Figure 6C), which was associated with re-expression of LIF mRNA (Figure 6D), LIF protein (Figure 6E), and a consequent return of constitutive phospho-STAT3-Y705 levels in EAC cells (Figure 6F). Treatment of resting EAC cells with exogenous LIF marginally induced IL6 mRNA levels, with no alterations under treatment with recombinant IL6 protein (Figure 6G). Importantly, substantially reduced levels of IL6 mRNA (80%, t test P < .001) were observed in LIF-silenced EAC cells, which could be returned to basal levels when treated with either exogenous recombinant LIF or recombinant IL6 protein (Figure 6G). Thus, silencing LIF overproduction by EAC cells allows the STAT3 pathway to once again become responsive to exogenous IL6 or LIF. This renewed responsiveness to exogenous IL6 could be blocked through the use of an IL6-blocking antibody, resulting in lower levels of target IL6 mRNA transcript (Figure 6G). However, siRNA- and blocking antibody–mediated inhibition of IL6 action in LIF-silenced cells did not result in any significant effects on EAC cell growth or LIF levels (Figure 6H), placing LIF hierarchically above IL6 expression and STAT3 activation in EAC cell lines. Because LIF, C1q, and TREM2
are all secreted factors it is likely that their pathways may converge in the maintenance of continued EAC cell growth. During validation of LIF silencing, reduced levels of C1QA mRNA were observed in LIF-silenced EAC cells (Figure 6I), consistent with the reported STAT3 regulation of C1q production in macrophages.
In vitro modulation of C1q mRNA expression and secretion by interleukin-1, interleukin-6, and interferon-gamma in resident and stimulated murine peritoneal macrophages.
Similarly, reduced levels of LIF mRNA expression were observed in C1QA-silenced EAC cells, with no detectable changes in TREM2 levels under either conditions (Figure 6I). Thus, a partial interdependence between LIF and C1QA mRNA levels was uncovered, suggestive of overlap between these pathways.
Figure 6LIF hierarchically regulates STAT3 phosphorylation, IL6 expression, and cell proliferation in EAC cells. SiRNA-mediated silencing of LIF expression results in reduced LIF mRNA (A) expression, (B) secretion, and (C) a recombinant-recoverable reduction in cell proliferation in CP-D, SKGT4, and OE-33 EAC cell lines. Treatment of LIF silenced cells with recombinant LIF protein (rLIF) results in (D) recovery of LIF mRNA in CP-D, SKGT4, and OE33 EAC cells and (E) protein expression levels in SKGT4 EAC cells. (F) STAT3-Y705 phosphorylation levels in LIF-silenced SKGT4 EAC cells are rescued by treatment with recombinant LIF protein as determined by Western blot. (G) Silencing LIF expression in SKGT4 EAC cells results in decreased IL6 mRNA expression that is recovered upon exposure to rLIF protein. (H) Neither silencing IL6 expression by siRNA-mediated silencing nor treatment with anti-IL6 antibody affects LIF mRNA levels or SKGT4 EAC cell growth. (I) Silencing LIF expression results in decreased expression of C1QA mRNA but not TREM2 in SKGT4 EAC cells. ∗P < .05, **P < .01, ***P < .001, and ****P < .0001 in either Student t test (proliferation) or Mann–Whitney testing (mRNA expression). siNT, nontargeting siRNA pool; siLIF, LIF-targeted siRNA pool; rLIF, recombinant LIF protein; rIL6, recombinant IL6 protein.
Autocrine Non–Complement-Mediated Promotion of EAC Cell Growth by C1q
A growing body of evidence suggests nonhumoral immune roles for complement proteins such as C1q-mediated activation of Wnt signaling pathways in epithelial cells.
The expression of C1QA was next examined in multiple EAC cell lines by Western blot (Figure 7A). Post screen verification experiments showed that inhibition of C1QA by siRNA-mediated silencing results in reduced cell growth of the CP-D cell line. This was confirmed further in a wider panel of additional EAC cell lines shown to express C1QA protein, with no effects observed in HEK293 cells that express limited levels of C1q protein (Figure 7B). Inhibition of traditional C1q complement function through exposure of EAC cells to C1INH did not result in any observable effect on EAC cell growth (Figure 7B) or expression levels (Figure 7C and D) by comparison with siRNA-mediated silencing of C1QA expression. Critically, the reduced levels of cell growth observed in C1QA-silenced EAC cells (42%; P < .001) could be returned to normal levels (98%, P = NS) after treatment with native C1q protein (Figure 7E), indicative of the involvement of a distinct non–complement-mediated signaling mechanism in response to C1q protein. Strikingly, treatment of resting and nontargeting control-transfected EAC cells with native C1q protein resulted in increased EAC cell proliferation (Figure 7E). These data are supportive of an autocrine C1q-driven cell survival/growth mechanism in EAC. During validation of C1QA mRNA levels in transfected cells it was also discovered that treatment of resting and C1QA-silenced EAC cells with native C1Q protein results in a feedback-like mechanism, leading to reduced levels of C1QA mRNA transcript (37%, P < .001; 11.5%, P < .001 resting and siC1QA, respectively) (Figure 7F). This further supports the presence C1q-mediated signaling pathways in EAC cells promoting cell growth and control of localized C1q production.
Figure 7C1q is produced by EAC cells and promotes cell proliferation in an autocrine manner. (A) Expression of C1QA protein in EAC cell lines by Western blot. (B) Silencing of C1QA results in reduced EAC cell viability with no effects on HEK-293 cells. (C) C1QA protein expression by Western blot after siRNA-mediated silencing or treatment with C1 inhibitor protein C1INH in SKGT4 EAC cells. (D) Densitometry of panel C. (E) Cell viability and (F) C1QA mRNA levels in SKGT4 EAC cells after either silencing of C1QA or treatment of siC1QA-silenced cells with native C1q to examine recovery of proliferative phenotypes. (G) Cell viability of multiple EAC cell lines after siRNA-mediated silencing of TREM2. (H) Treatment of TREM2-silenced SKGT4 EAC cells with rLIF or C1q native protein resulted in the rescue of the observed growth arrest. (I) Expression of TREM2 and C1q-associated ITAM docking protein TYROBP (DAP12), signaling intermediary SYK and p-SYK in multiple EAC cell lines. **P < .01, ***P < .001, and ****P < .0001 in either Student t test (proliferation) or Mann–Whitney testing (mRNA expression). siC1QA, siRNA pools targeting complement C1q subcomponent subunit A; siNT, nontargeting siRNA pool; siTREM2, siRNA pools targeting triggering receptor expressed on myeloid cells 2.
TREM2 and Its Associated Signaling Pathway Regulates EAC Cell Survival
TREM2 is a transmembrane and secreted factor, and similar to C1q, recently has been associated with Alzheimer’s disease, innate and humoral immune responses, and triggering constitutive production of cytokines through the TYROBP (DAP12) adaptor protein.
TREM2 mRNA expression directly correlates with that of C1QA (rs = 0.69), C1QB (rs = 0.71), C1QC (rs = 0.72), and the ITAM adapter protein TYROBP (rs = 0.73) expression in EAC tissues using TCGA data. Validation of TREM2-mediated support of EAC cell growth was next performed in a wider panel of EAC cell lines transfected with TREM-2 targeting siRNAs, confirming similar suppression of cell growth as observed in the original screen and post screen verification (Figure 7G). No significant effects on the expression of TREM2 were noted in LIF- or C1QA-silenced cells (Figure 6I). However, treatment of TREM2-silenced SKGT4 EAC cells with either recombinant LIF or native C1q proteins resulted in a statistically significant rescue of cell viability (Figure 7H). TREM2, in a similar fashion to C1q, may mediate cell signaling through immunoreceptor tyrosine-based motif proteins such as TYROBP, leading to downstream SYK pathway activation in a variety of cell types.
In Western blot experiments, TREM2, TYROBP (DAP12), SYK, and phospho-SYK expression was validated in a panel of EAC cell lines showing the presence of the associated signaling pathway in esophageal cell lines (Figure 7I).
Targeting the Immune Signature of EAC Cells With Fostamatinib
SYK inhibitors such as fostamatinib (R788 prodrug; R406 active format) have shown promise in treating allergic and autoimmune conditions, such as rheumatoid arthritis (RA), in addition to B-cell lineage malignancies.
both of which communicate signaling from LIF and IL6 through the STAT3 transcription factor. Because LIF, C1q, and TREM2 signaling were prominent in the regulation of EAC cell survival/growth, we examined the ability of fostamatanib R406 to target the JAK/STAT and SYK/AKT pathways and induce growth arrest or apoptosis in EAC cells. Under increasing doses of R406 a significant decrease in both STAT3 and AKT phosphorylation was observed in SKGT4 EAC cells as determined by the phospho/total protein ratio in enzyme-linked immunosorbent assays (Figure 8A) and via Western blot of phospho-STAT3 and phospho-AKT (Figure 8B). Furthermore, significant levels of apoptosis (18.7% ± 1.7 sub-G0/G1 or R6; P < .001), as defined by 7-AAD/bromodeoxyuridine staining, were observed in SKGT4 cells treated with 12.5 μmol/L R406 by comparison with vehicle control (3.8% ± 0.87 sub-G0/G1 or R6) in addition to increased growth arrest (31.5% ± 2.3 G0/G1; P < .001) at 24 hours (Figure 8C and D). These findings were replicated in EAC cell lines known to form robust tumors in nude mice, OE33 cells, and FLO-1 cells over a range of R406 concentrations and measuring cell proliferation and apoptosis at 24 hours (Figure 8E). To further show the efficacy of fostamatinib for therapeutic use, OE33 (left flank) and FLO1 (right flank) cells were allowed to engraft in NOD/SCIDγ mice for 5 days before segregation into treatment groups (n = 5) for daily injection of vehicle or fostamatinib R788 prodrug at 40-mg/kg and 80-mg/kg treatment. Significantly reduced tumor volumes were achieved for both cell lines (linear regression P < .0001) in both treatment groups, with no significant difference between 40-mg/kg and 80-mg/kg doses. Over the first 30 days after engraftment, treatment with 80 mg/kg R788 significantly suppressed tumor growth, however, some evidence of refractory growth pattern in response to the fostamatinib R788 inhibitor was observed at day 33 (Figure 8F).
Figure 8Targeting SYK and JAK with fostamatinib results in EAC cell growth arrest, apoptosis, and reduced xenografted-tumor burden. (A) Ratio of total STAT or AKT protein to phospho-STAT3 or phosphor-AKT protein after a 90-minute treatment of SKGT4 EAC cells with fostamatinib R406 as measured by enzyme-linked immunosorbent assay. (B) Western blot of phospho-STAT3, phospho-AKT after treatment with either dimethyl sulfoxide (DMSO) (-) or R406 10 nmol/L(+). (C and D) Fluorescence-activated cell sorter analysis of SKGT4 EAC cell-cycle phase after treatment of SKGT4 EAC cells with fostamatinib R406 (active drug), DMSO (vehicle), DCA (positive control), or low FCS media (R-FCS negative control). (E) Fluorescence-activated cell sorter analysis of OE33 and FLO1 EAC cell-cycle phase after treatment with a concentration curve of fostamatinib R406. (F) Xenografted tumor volume measurements (n = 5) in NOD/SCIDγ mice after engraftment with FLO1 (left flank) and OE33 (right flank) cell lines and daily treatment with either CMC (vehicle) or fostamatinib R788 (prodrug) at 40 or 80 mg/kg. R4, S-phase cell population; R3, G0/G1 phase cell population; R5, G2/M phase cell population; R6, sub-G1 early apoptosis cell population; CMC, carboxymethylcellulose; R, resting; R-FCS, resting cells in culture media without DMSO; R406, active metabolite of SYK inhibitor fostamatinib; R788, fostamatinib, a prodrug of the active metabolite and SYK inhibitor R406. ∗P < .05, **P < .01, ***P < .001, ****P < .0001 in either Student’s t test (fluorescence-activated cell sorter and enzyme-linked immunosorbent assay data) or linear regression (R406 xenograft study data in panel E). FITC, fluorescein isothiocyanate; Veh, vehicle.
Herein, we define a cohort of targetable factors whose siRNA-mediated silencing inhibited EAC cell growth. Critically, we also show that somatic variation observed in EAC directly associates with genes sustaining EAC cell growth as defined by siRNA library screening. The screening approach, although only completed in CP-D HGD cells, however, resulted in significant findings that could be verified through pathologic and functional studies using both alternate siRNA formats, recombinant proteins, and rescue experiments supporting the overall immune-signature associations described herein. EAC develops in a background of GERD-induced inflammation and it recently was postulated that inflammatory factors may be directly responsible for the induction of metaplasia development rather than a process of caustic erosion followed by metaplastic replacement.
The data suggest that the low-grade epithelial inflammation induced by a stressor in conjunction with tumor-suppressor loss (p53 mutation) results in a macrophage mimicry by tumor cells.
In addition, recent sequencing studies have defined glioblastoma subsets with distinct complement response signature derived from CD45-depleted tumors.
Our detection of an enriched immune signature underlying the promoters of EAC cell viability as determined by siRNA library screening, thus is supported by considerable evidence. Moreover, nuclear proteins guiding genome integrity and mitotic fidelity also were prominent regulators of cancer cell survival, as might be expected, from the siRNA screening data and included SYMPK, RAD51, and RAN, which all were overexpressed in GEM data of EAC. The presence of RAD51 within this list likely reflects the instability associated with the EAC cell genome and previously has been linked with alterations in copy number and heterozygosity in EAC cells.
Recent integrative genomic analyses of esophageal and gastric carcinomas, performed as a component of the TCGA project, has characterized the associated somatic and epigenetic alterations to facilitate improved classification and selection of appropriate therapeutic modalities.
Cancer Genome Atlas Research Network; Analysis Working Group: Asan University; BC Cancer Agency; Brigham and Women’s Hospital; Broad Institute; Brown University; Case Western Reserve University; Dana-Farber Cancer Institute; Duke University; Greater Poland Cancer Centre; Harvard Medical School; Institute for Systems Biology; KU Leuven; Mayo Clinic; Memorial Sloan Kettering Cancer Center; National Cancer Institute; Nationwide Children’s Hospital; Stanford University; University of Alabama; University of Michigan; University of North Carolina; University of Pittsburgh; University of Rochester; University of Southern California; University of Texas MD Anderson Cancer Center; University of Washington; Van Andel Research Institute; Vanderbilt University; Washington University; Genome Sequencing Center: Broad Institute; Washington University in St. Louis; Genome Characterization Centers: BC Cancer Agency; Broad Institute; Harvard Medical School; Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins University; University of North Carolina; University of Southern California Epigenome Center; University of Texas MD Anderson Cancer Center; Van Andel Research Institute; Genome Data Analysis Centers: Broad Institute; Brown University:; Harvard Medical School; Institute for Systems Biology; Memorial Sloan Kettering Cancer Center; University of California Santa Cruz; University of Texas MD Anderson Cancer Center; Biospecimen Core Resource: International Genomics Consortium; Research Institute at Nationwide Children’s Hospital; Tissue Source Sites: Analytic Biologic Services; Asan Medical Center; Asterand Bioscience; Barretos Cancer Hospital; BioreclamationIVT; Botkin Municipal Clinic; Chonnam National University Medical School; Christiana Care Health System; Cureline; Duke University; Emory University; Erasmus University; Indiana University School of Medicine; Institute of Oncology of Moldova; International Genomics Consortium; Invidumed; Israelitisches Krankenhaus Hamburg; Keimyung University School of Medicine; Memorial Sloan Kettering Cancer Center; National Cancer Center Goyang; Ontario Tumour Bank; Peter MacCallum Cancer Centre; Pusan National University Medical School; Ribeirão Preto Medical School; St. Joseph’s Hospital &Medical Center; St. Petersburg Academic University; Tayside Tissue Bank; University of Dundee; University of Kansas Medical Center; University of Michigan; University of North Carolina at Chapel Hill; University of Pittsburgh School of Medicine; University of Texas MD Anderson Cancer Center; Disease Working Group: Duke University; Memorial Sloan Kettering Cancer Center; National Cancer Institute; University of Texas MD Anderson Cancer Center; Yonsei University College of Medicine; Data Coordination Center: CSRA Inc.; Project Team: National Institutes of Health Integrated genomic characterization of oesophageal carcinoma.
Carcinoma-specific amplifications defined include ERBB2, VEGFA, and GATA6 in EACs, and TP63, SOX2, and CCND1 in ESCCs. When genes supportive of EAC cell growth as defined through siRNA screening of CP-D cells were examined in TCGA data of esophageal tumors a significant intersection was uncovered composed of 49 genes with somatic frequencies greater than 5%. Intriguingly, 14% (7 of 49) of these genes showing gene amplifications, at similar frequencies to ERBB2 and VEGFA amplifications, were encoded at Chr6p21.1 and specifically associated with EAC rather than ESCC carcinomas. This further supports the hypothesis that the observed somatic variation is occurring in differentiation pathways specific to EAC. Chromosome 6p21 and the major histocompatibility locus in particular has been linked consistently with risk of both BE and EAC in genome-wide association studies and follow-up meta-analyses further supporting the inflammatory links to this cancer type.
Barrett's and Esophageal Adenocarcinoma Consortium (BEACON); Esophageal Adenocarcinoma GenEtics Consortium (EAGLE); Wellcome Trust Case Control Consortium 2 (WTCCC2) Genome-wide association studies in oesophageal adenocarcinoma and Barrett's oesophagus: a large-scale meta-analysis.
Cancer Genome Atlas Research Network; Analysis Working Group: Asan University; BC Cancer Agency; Brigham and Women’s Hospital; Broad Institute; Brown University; Case Western Reserve University; Dana-Farber Cancer Institute; Duke University; Greater Poland Cancer Centre; Harvard Medical School; Institute for Systems Biology; KU Leuven; Mayo Clinic; Memorial Sloan Kettering Cancer Center; National Cancer Institute; Nationwide Children’s Hospital; Stanford University; University of Alabama; University of Michigan; University of North Carolina; University of Pittsburgh; University of Rochester; University of Southern California; University of Texas MD Anderson Cancer Center; University of Washington; Van Andel Research Institute; Vanderbilt University; Washington University; Genome Sequencing Center: Broad Institute; Washington University in St. Louis; Genome Characterization Centers: BC Cancer Agency; Broad Institute; Harvard Medical School; Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins University; University of North Carolina; University of Southern California Epigenome Center; University of Texas MD Anderson Cancer Center; Van Andel Research Institute; Genome Data Analysis Centers: Broad Institute; Brown University:; Harvard Medical School; Institute for Systems Biology; Memorial Sloan Kettering Cancer Center; University of California Santa Cruz; University of Texas MD Anderson Cancer Center; Biospecimen Core Resource: International Genomics Consortium; Research Institute at Nationwide Children’s Hospital; Tissue Source Sites: Analytic Biologic Services; Asan Medical Center; Asterand Bioscience; Barretos Cancer Hospital; BioreclamationIVT; Botkin Municipal Clinic; Chonnam National University Medical School; Christiana Care Health System; Cureline; Duke University; Emory University; Erasmus University; Indiana University School of Medicine; Institute of Oncology of Moldova; International Genomics Consortium; Invidumed; Israelitisches Krankenhaus Hamburg; Keimyung University School of Medicine; Memorial Sloan Kettering Cancer Center; National Cancer Center Goyang; Ontario Tumour Bank; Peter MacCallum Cancer Centre; Pusan National University Medical School; Ribeirão Preto Medical School; St. Joseph’s Hospital &Medical Center; St. Petersburg Academic University; Tayside Tissue Bank; University of Dundee; University of Kansas Medical Center; University of Michigan; University of North Carolina at Chapel Hill; University of Pittsburgh School of Medicine; University of Texas MD Anderson Cancer Center; Disease Working Group: Duke University; Memorial Sloan Kettering Cancer Center; National Cancer Institute; University of Texas MD Anderson Cancer Center; Yonsei University College of Medicine; Data Coordination Center: CSRA Inc.; Project Team: National Institutes of Health Integrated genomic characterization of oesophageal carcinoma.
Peri-operative chemotherapy with or without bevacizumab in operable oesophagogastric adenocarcinoma (UK Medical Research Council ST03): primary analysis results of a multicentre, open-label, randomised phase 2-3 trial.
Thus, the significant association between genes sustaining EAC growth in siRNA screening data and their expression from Chr6p21.1 supports the hypothesis that alterations at this locus also may contribute to oncogenesis through enhanced proliferative rates in addition to that mediated by alterations to VEGFA levels and angiogenic pathways. Comparatively, no consistent links have been detailed between Chr8q24.3 and Chr20q13.33 and either BE, EAC, or ESCC occurrence genomic studies. No ESCC-specific signature was associated with the 49 gene amplifications highlighted in our study. However, a significant ESCC-specific signature could be defined when squamous markers TP63 and SOX2 were analyzed, although neither SOX2 nor TP63 sustain EAC cell growth according to screening data. Gene amplification events that were defined at Chr8q24.3 and Chr20q13.33 were not cancer-type–specific and occurred at similar frequencies in either EAC or ESCC. Thus, the integration of somatic variation data with druggable genome screening results highlights a significant overlap that supports the validity of the potential targets identified in this study.
Analysis of druggable genome screening results, using multiple GO software platforms,
provided strikingly similar interpretations of the gene-function relationships. These findings indicated the involvement of immune-related pathways in the regulation of EAC cell growth and survival. The development of BE, and by relationship EAC, previously has been associated with GERD-promoted inflammation. Recent reports in murine models have shown that viral-mediated overexpression of Il1b in the upper alimentary tract can lead to an Il6-dependent Barrett's-like metaplasia, with dysplasia.
In our study, neither silencing of IL6 expression nor anti-IL6 blocking affected EAC cell growth or survival in human cell lines. However, silencing of the IL6 family member, LIF, resulted in significantly reduced EAC cell proliferation, pSTAT3 and IL6 levels in SKGT4 EAC cells that could be rescued through treatment with exogenous recombinant LIF. Critically, silencing LIF expression also resulted in reduced C1QA levels, showing an interdependency between these pathways. STAT3 is a known SYK pathway substrate further supporting this finding. Because both C1q and TREM2 similarly use SYK-mediated signaling pathways in immune cell types, these proteins were selected for further analysis.
The C1QA gene encodes a subcomponent of the trimeric C1q complement protein normally associated with the innate surveillance system thought to largely occur from bone marrow–derived cells such as macrophages and dendritic cells.
Thus, the expression and production of C1q by EAC epithelial cells, in our study, and its promotion of sustained EAC cell growth through non–complement-pathway–mediated autoregulatory signaling was unexpected. Silencing of the complement C1q binding protein (C1QBP or GC1QR), which normally binds to the globular heads of C1q protein, thus inhibiting C1 activation, has been observed to result in reduced lamellipodia formation and consequently reduced cell migration and invasion of cancer cells in a number of studies.
However, silencing C1QBP was not observed as a significant regulator of EAC cell growth in our screening data. A clear increase in metaplastic epithelial and luminal cell surface expression of C1QA was observed in tissue sections from BE patients that was mirrored at the mRNA level in both BE and EAC by comparison with squamous control tissues. Similarly, TREM2 immunostaining reflected that observed at the mRNA level. Dense epithelial cytoplasmic and cell surface expression of TREM2 was noted in goblet cell–positive glands of BE and, intriguingly, in putative neuroendocrine cells noted within glandular structures. A number of studies have linked inappropriate activation of both TREM2 and C1Q function independently with the development of Alzheimer’s disease through microgliosis and augmented complement-mediated synaptic pruning, respectively, supporting a shared functional relationship between these proteins.
The functions associated with TREM2 have to date remained rooted in immune and neuronal cell types and osteoclasts, with few studies on its role in carcinogenesis. However, recent reports have indicated that silencing TREM2 in glioma cells may result in cell death and thus provide a rational therapeutic target.
Unfortunately, no GEM database with associated survival metrics for EAC patients currently exists in a publicly accessible form. However, a significant preferential association between high LIF, C1QA, and TREM2 mRNA expression and poorer survival outcome was defined in such a database of lung and gastric adenocarcinoma patients in our analysis of kmplot data.
Treatment of TREM2-silenced EAC cells with exogenous native C1q protein partially rescued EAC cell viability, supportive of a shared signaling pathway. Components of the signaling pathways used by TREM2 and C1q in immune cells, immunoreceptor tyrosine-based activation motif (ITAM) docking protein TYROBP and SYK, were expressed in EAC cell lines. SYK expression also was confirmed in EAC cell lines and was responsive to fostamatinib treatment, resulting in decreased cell cycle S-phase percentages and increased EAC cell death.
The oral SYK inhibitor fostamatinib is currently under clinical and experimental trials for the treatment of a range of conditions such as immune thrombocytopenia, RA,
The JAK2/STAT3 pathway mediates cytokine induction of receptor activator of nuclear factor kappa-B ligand (RANKL) in synovial fibroblasts in RA and thus combined inhibition of both SYK and JAK by fostamatinib may be beneficial. Importantly, JAK2 is central in mediating responses to IL6 and LIF through IL6R and gp130 cell surface receptors.
Thus, the use of R788/fostamatinib for the treatment of EAC, as suggested by this report, may mediate therapeutic effects through JAK/STAT and SYK, key pathways in the inflammatory promotion of EAC and the regulation of EAC cell proliferation and survival as defined in this study.
In summation, these findings further show and provide molecular evidence for the involvement of inflammatory factors in the development of EAC. Furthermore, the data provide actionable targets for the development of EAC therapeutics that may support standard chemotherapy-based approaches and provides proof of concept for one such therapeutic target pathway. Although we have focused on the subset of genes associated with the SYK JAK-STAT pathway, further studies will address the additional immunologic pathways uncovered in this study and we anticipate that this approach will uncover further new therapeutic options.
Acknowledgments
The authors thank Anthony Davies, Rocio Seoane, and Lucy Baudino for technical support.
References
Thrift A.P.
Whiteman D.C.
The incidence of esophageal adenocarcinoma continues to rise: analysis of period and birth cohort effects on recent trends.
The biomarker TP53 divides patients with neoadjuvantly treated esophageal cancer into 2 subgroups with markedly different outcomes. A p53 Research Group study.
Cancer Genome Atlas Research Network; Analysis Working Group: Asan University; BC Cancer Agency; Brigham and Women’s Hospital; Broad Institute; Brown University; Case Western Reserve University; Dana-Farber Cancer Institute; Duke University; Greater Poland Cancer Centre; Harvard Medical School; Institute for Systems Biology; KU Leuven; Mayo Clinic; Memorial Sloan Kettering Cancer Center; National Cancer Institute; Nationwide Children’s Hospital; Stanford University; University of Alabama; University of Michigan; University of North Carolina; University of Pittsburgh; University of Rochester; University of Southern California; University of Texas MD Anderson Cancer Center; University of Washington; Van Andel Research Institute; Vanderbilt University; Washington University; Genome Sequencing Center: Broad Institute; Washington University in St. Louis; Genome Characterization Centers: BC Cancer Agency; Broad Institute; Harvard Medical School; Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins University; University of North Carolina; University of Southern California Epigenome Center; University of Texas MD Anderson Cancer Center; Van Andel Research Institute; Genome Data Analysis Centers: Broad Institute; Brown University:; Harvard Medical School; Institute for Systems Biology; Memorial Sloan Kettering Cancer Center; University of California Santa Cruz; University of Texas MD Anderson Cancer Center; Biospecimen Core Resource: International Genomics Consortium; Research Institute at Nationwide Children’s Hospital; Tissue Source Sites: Analytic Biologic Services; Asan Medical Center; Asterand Bioscience; Barretos Cancer Hospital; BioreclamationIVT; Botkin Municipal Clinic; Chonnam National University Medical School; Christiana Care Health System; Cureline; Duke University; Emory University; Erasmus University; Indiana University School of Medicine; Institute of Oncology of Moldova; International Genomics Consortium; Invidumed; Israelitisches Krankenhaus Hamburg; Keimyung University School of Medicine; Memorial Sloan Kettering Cancer Center; National Cancer Center Goyang; Ontario Tumour Bank; Peter MacCallum Cancer Centre; Pusan National University Medical School; Ribeirão Preto Medical School; St. Joseph’s Hospital &Medical Center; St. Petersburg Academic University; Tayside Tissue Bank; University of Dundee; University of Kansas Medical Center; University of Michigan; University of North Carolina at Chapel Hill; University of Pittsburgh School of Medicine; University of Texas MD Anderson Cancer Center; Disease Working Group: Duke University; Memorial Sloan Kettering Cancer Center; National Cancer Institute; University of Texas MD Anderson Cancer Center; Yonsei University College of Medicine; Data Coordination Center: CSRA Inc.; Project Team: National Institutes of Health
Integrated genomic characterization of oesophageal carcinoma.
An integrative genomic approach in oesophageal cells identifies TRB3 as a bile acid responsive gene, downregulated in Barrett's oesophagus, which regulates NF-kappaB activation and cytokine levels.
Ursodeoxycholic acid inhibits interleukin 1 beta [corrected] and deoxycholic acid-induced activation of NF-kappaB and AP-1 in human colon cancer cells.
The characterization of an intestine-like genomic signature maintained during Barrett's-associated adenocarcinogenesis reveals an NR5A2-mediated promotion of cancer cell survival.
The Syk inhibitor fostamatinib disodium (R788) inhibits tumor growth in the Emu-TCL1 transgenic mouse model of CLL by blocking antigen-dependent B-cell receptor signaling.
The first subcomponent of complement, C1q, triggers the production of IL-8, IL-6, and monocyte chemoattractant peptide-1 by human umbilical vein endothelial cells.
In vitro modulation of C1q mRNA expression and secretion by interleukin-1, interleukin-6, and interferon-gamma in resident and stimulated murine peritoneal macrophages.
Peri-operative chemotherapy with or without bevacizumab in operable oesophagogastric adenocarcinoma (UK Medical Research Council ST03): primary analysis results of a multicentre, open-label, randomised phase 2-3 trial.
Author contributions S. P. Duggan and C. Garry performed and designed the studies; F. M. Behan and S. Phipps contributed to technical components of screening implementation; A. Zaheer, M. Kirca, J. V. Reynolds, S. McGarrigle, H. Kudo, R. Goldin, S. E. Kalloger, and D. F. Schaeffer provided clinical and pathologic support and analyses; S. P. Duggan and J. Strid performed and evaluated animal model experiments; A. Long provided in vitro experimental expertise and interpretation; S. P. Duggan and D. Kelleher planned the study design and supervised the study; and all authors edited the manuscript.
Conflicts of interest The authors disclose no conflicts.
Funding Supported by the Health Research Board Ireland (HRB-TRA.2007.11 (S.D., C.G., S.P., D.K., A.L.) and the HRB PhD Scholars program (F.M.B.)), Imperial College London (H.K., R.G., J.S., D.K.), the University of British Columbia Start-up funds F15-04616, and a donation in memoriam by the Chan family (S.D., D.K.).
30019-5&id=fx1.jpg){kind=link}
30019-5&id=gr1.jpg){kind=link}
30019-5&id=gr2.jpg){kind=link}
30019-5&id=gr3.jpg){kind=link}
30019-5&id=gr4.jpg){kind=link}
30019-5&id=gr5.jpg){kind=link}
30019-5&id=gr6.jpg){kind=link}
30019-5&id=gr7.jpg){kind=link}
30019-5&id=gr8.jpg){kind=link}