• C-X-C chemokine;
  • chemokine receptor;
  • esophageal cancer;
  • temozolomide


  1. Top of page
  2. Abstract


C-X-C ligand (CXCL) chemokines exert major roles in the biologic aggressiveness of esophageal cancer. In the current study, the authors investigated temozolomide (TMZ)-induced effects on activity of the CXCL chemokine network in human esophageal cancer cells. To the authors' knowledge, TMZ has not been investigated previously in experimental or clinical esophageal cancers.


A complete mapping of CXCL chemokines and their receptor messenger RNA was performed in 2 established human esophageal cancer cell lines (OE21 and OE33) and in 4 surgical samples from patients with esophageal carcinoma. The analyses pointed out the potential importance of CXCL2, and monitoring CXCL2 with quantitative videomicroscopy indicated that its biologic activity was silenced in OE21 esophageal cancer cells. TMZ-mediated antitumor activity was determined in vivo in an OE21 metastatic nude mice xenograft model.


The messenger RNA levels of CXC chemokines and their receptors were similar in both cell lines and in the 4 surgical specimens. CXCL2 depletion by small interfering RNA (siRNA) displayed marked effects on the proliferation of transfected OE21 cells. Chronic in vitro TMZ treatment of OE21 and OE33 cells markedly decreased CXCL2 and CXCL3 secretion. In vivo, TMZ induced significant delays in OE21 xenograft tumor development and improved the survival of OE21 xenograft-bearing mice, whereas cisplatin did not. Analyses performed on tissue samples from in vivo experiments revealed that TMZ also impaired tumor angiogenesis.


The current study emphasized the role of proangiogenic chemokines in esophageal cancer biology and indicated the possibility of using TMZ as a clinically compatible drug to impair the actions of the CXCL chemokine network in esophageal cancers. Cancer 2011. © 2010 American Cancer Society.

Despite decades of effort to improve outcomes, esophageal cancer remains a highly fatal disease. Annually, approximately 460,000 patients are diagnosed with esophageal cancer worldwide, and >380,000 patients die from this malignancy, which makes esophageal cancer the eighth most common cancer and sixth on the list of cancers that cause mortality.1 The overall 5-year survival rate for patients with esophageal cancer remains low because of late diagnosis, metastasis, and resistance of the tumor to radiotherapy and chemotherapy.2 Surgical resection as the treatment of choice remains a matter of debate, especially with modern strategies that incorporate neoadjuvant chemoradiation or perioperative chemotherapy for the treatment of locoregional disease. In fact, >50% of patients still have inoperable disease at presentation.1 A recently updated Cochrane review, which compared preoperative chemotherapy versus surgery alone and included 11 randomized trials that involved 2019 patients, ended with inconclusive evidence of a statistically significant benefit of standard chemotherapy.3 It is important to note that most of the standard agents (platines, taxanes, 5-fluorouracil) that have been used, in fact, are proapoptotic drugs and that esophageal cancers are naturally resistant to proapoptotic stimuli.3-6 Thus, novel therapeutic approaches are needed to combat esophageal cancers that display natural resistance to proapoptotic stimuli.

Temozolomide (TMZ) significantly improves outcomes in patients with glioblastoma, a malignancy that represents another example of an apoptosis-resistant cancer that is associated with a dismal prognosis.7, 8 Whole genomic analyses of TMZ-resistant glioma cells recently revealed, among other findings, increased expression levels of CXC chemokines.9

CXC chemokine ligands (CXCL) are a unique cytokine family that, based on structure/function and receptor binding/activation studies, exhibit either angiogenic or angiostatic biologic activity.10, 11 Many cancers have a complex chemokine network that affects the transcription of target genes involved in cell invasion, motility, interactions with the extracellular matrix, and the survival of cancer cells.11, 12 Numerous studies have demonstrated that CXCL7 is overexpressed in tumor cells in general (for a review, see Waugh and Wilson13) and in esophageal cancer in particular.14-17 The expression of CXCL12 and its receptor, CXCR4, reportedly play important roles in both esophageal squamous cell carcinoma and adenocarcinoma in relation to lymph node metastasis.18-21 Wang et al22 reported that CXCL1-CXCR2 and CXCL2-CXCR2 signaling contributes significantly to esophageal cancer cell proliferation and that this autocrine signaling pathway may be involved in esophageal tumorigenesis. Together, these data prompted us to seek a reliable experimental model for human esophageal cancer and to investigate 1) the importance of CXCLs in the biology of esophageal cancers, 2) TMZ-induced effects on CXCL expression, and 3) TMZ-induced antitumor effects on experimental esophageal cancers.


  1. Top of page
  2. Abstract

Cell Lines, Tissue Samples, Media, and Compounds

Established cell lines

Two established human esophageal cancer cell lines were used in this study: the squamous cancer cell line OE21 (European Collection of Cell Cultures [ECACC] code 96062201) and the adenocarcinoma cell line OE33 (ECACC code 9607808), both obtained from Sigma-Aldrich (Bornem, Belgium). The cells were maintained in culture as detailed elsewhere.23

Esophageal cancer tissue samples

Four archived tissue samples from patients with esophageal squamous cell cancer were kindly provided by Professor Isabelle Salmon (Head of the Department of Anatomopathology at the Erasmus University Hospital, Free University of Brussels, Belgium).


TMZ was obtained from Schering Plough (Brussels, Belgium). Cisplatin was provided by Aldrich (Bornem, Belgium).

In Vitro Overall Cell Growth Determination

Overall cell growth was assessed using the 3-(4,5-dimethylthiazol-2yl)-diphenyltetrazolium bromide colorimetric assay (Sigma-Aldrich), as detailed previously.23 Each assessment was carried out 6 times.

Reverse Transcriptase-Polymerase Chain Reaction (RT-PCR) and Quantitative RT-PCR

Total RNA extraction, and standard and quantitative RT-PCR were performed as detailed elsewhere.24 The primers that were used, along with the experimental conditions for each primer, were provided by Invitrogen (Carlsbad, Calif) and were selected using HYBSIMULATOR software (Advanced Gene Computing Technology, Irvine, Calif). All primers that were used are listed in Table 1. The primers that were used for β-actin detection were as follows: forward, 5′-AAATCGTGCGTGACATTAAGG-3′; reverse, 5′-CTAAGTCATAGTCCGCCTAG-3′.

Table 1. Technical Details for Detecting C-X-C Ligand Chemokine (CXCL) and CXCL Receptor Expression by Reverse Transcription-Polymerase Chain Reaction Analysis
Targets (Synonym Names)NCBI Sequence No.Forward Primer SequencesReverse Primer SequencesAnnealing Temperature, °Ca
  • NCBI, National Center for Biotechnology Information; CXCL, C-X-C ligand chemokine; a, adenine; g, guanine; c, cytosine; t, thymine; CXCR, CXCL receptor; KSHV, Kaposi sarcoma-associated herpesvirus; Duffy, the Duffy antigen/chemokine receptor.

  • a

    The thermal profiles were as follows: predenaturation for 10 minutes at 95°C, denaturation for 30 seconds at 95°C, annealing for 45 seconds at the temperatures indicated in this table, elongation for 45 seconds and 1 minute at 72°C, and a final elongation for 10 minutes at 72°C. The polymerase chain reaction products were obtained after 35 thermal cycles.


Enzyme-Linked Immunosorbent Assays

The CXC chemokines growth-related oncogenic β (GROβ) and GROγ from enzyme-linked immunosorbent assay (ELISA) construction kits (Antigenix America, Huntington Station, NY) were used in accordance with the manufacturer's instructions for the quantification of human CXCL2 and CXCL3, respectively. The plates were coated with 1 μg/mL of capture antibody, and the tracer antibody was used at 0.25 μg/mL. Cell culture supernatants were collected after different treatments and times (as indicated in the figure legends), and multiple aliquots were taken and stored at −20°C until they were analyzed. The results from the ELISAs are represented as absorbance values. The concentrations of CXCLs are expressed as picograms per milliliter (pg/mL) and all samples were normalized to the cell count in the sample.

Flow Cytometry Analysis for Acridine Orange Measurements

TMZ anticancer effects can be induced by sustained autophagic processes,7 which can be assessed by quantifying acidic vesicular organelles (revealed as red fluorescence) after acridine orange (Sigma-Aldrich) staining.24 In acridine orange-stained cells, the cytoplasm and nucleus fluoresce green, and the acidic compartments fluoresce red. The intensity of the red fluorescence is proportional to the degree of acidity and the volume of acidic vesicular organelles, including autophagic vacuoles. TMZ-induced effects at the level of acidic vesicular organelles in OE21 and OE33 cancer cells were measured by flow cytometry, as detailed elsewhere.24 Each sample was evaluated in triplicate experiments.

Flow Cytometry Analysis for Apoptosis Measurements

Flow cytometry analyses of apoptotic cell death were carried out by using the terminal deoxynucleotidyl transferase deoxyuridine triphosphate nick-end labeling (TUNEL) technique based on a previously detailed experimental protocol.25 Briefly, OE21 cells were treated with TMZ once at a concentration of 100 μM and also 5 times at the same concentration. Control cells were left untreated. Apoptosis was then assessed by using the flow cytometry-based TUNEL assay.25 Both nonadherent cells and adherent cells were harvested, pooled, and fixed overnight (for 15 hours) with 1% paraformaldehyde (for 1 hour) and 70% ethanol. The next day, the cells were processed using the APO-bromodeoxyuridine (BrdU) kit (Sigma-Aldrich) according to the manufacturer's instructions. Immediately after the staining reaction, analysis was performed on a Cell Lab QUANTA flow cytometer (Beckman Coulter, Analis, Suarlee, Belgium) equipped with a 488-nm argon laser. Apoptosis was quantified by following the increase in fluorescein isothiocyanate-deoxyuridine triphosphate labeling in TMZ-treated cells compared with that in control (untreated) cells.

Immunofluorescence Analyses

Cells were cultured on coverslips and fixed with 3.7% formaldehyde in phosphate-buffered saline (PBS) for 20 minutes at 4°C. Fixed cells were permeabilized for 20 minutes by the addition of 0.2% (volume/volume) Triton X-100 and 10% (weight/volume) bovine serum albumin. Cells were washed twice with PBS and blocked in PBS that contained 0.1% bovine serum albumin for 1 hour at room temperature. Cells were then stained for 1 hour at room temperature with a polyclonal antibody against GROβ (CXCL2; 1:75 dilution) (Santa Cruz Biotechnology, Heidelberg, Germany). Antigens were detected using antigoat secondary antibody conjugated to Alexa Fluor 594 (Invitrogen). Coverslips were mounted on microscope slides with 10 μL Moviol agent (Calbiochem, VWR, Heverlee, Belgium). Images were captured using a ×40 microscope immersion objective (Zeiss Observer.Z1; Zeiss, Oberkochen, Germany) and a software-controlled Axiocam HRm Zeiss camera. Images were converted to stacks and navigated using AxioVison software (release 4.6; Carl Zeiss MicroImaging GmbH, Hamburg, Germany).

Transient Knock-Down of C-X-C Ligand Chemokine 2 (CXCL2) Expression by Means of an Anti-CXCL2 Small Interfering RNA in Human OE21 Esophageal Cancer Cells

The sequence of the CXCL2 siRNA (Eurogentec, Seraing, Belgium) that was used in the current work was 5′-gcaucgcccaugguuaaga-3′ for the sense sequence and 5′-ucuuaaccaugggcgaugc-3′ for the antisense sequence. An siRNA negative control (Eurogentec) was used as a process control. The antisense and sense strands of the siRNA were annealed by the manufacturer in 50 mM Tris, pH 7.5 to 8.0, and 100 nM NaCl in diethylpyrocarbonate-treated water. The final concentration of the siRNA duplex was 100 μM.

Cells either were left untreated (wild-type [wt]) or were transfected with 3-tetradecylamino-N-tert-butyl-N-tetradecylpropionamidine (DiC14-amidine) lipoplexes with or without siRNA (negative control: scrambled siRNA [scr] or CXCL2-specific siRNA). DiC14-amidine synthesis and liposome preparations were described previously.26 To form lipoplexes, DiC14-amidine/siRNA at a DNA concentration of 0.4 μM (usually 200 μL) was added to an equal volume of liposomes at 20 μg diC14-amidine/mL in RPMI medium (Invitrogen) while gently shaking the tube. The liposome/DNA mixture was allowed to stand for 20 minutes at room temperature before use. Under these conditions, the lipoplex has a cationic lipid/siRNA ratio of 7.5:1 (weight:weight), and the charge ratio was calculated as 4.54 positive charges for 1 negative charge. An additional lipid/siRNA ratio of 15:1 also was tested, and the charge ratio was calculated as 9.08 positive charges for 1 negative charge. The liposome/DNA mixture was diluted in RPMI buffer (up to a concentration of 0.032 μM in siRNA) and added to OE21 cells for 2 hours. On Day 2, each group of cells was pooled and replated for subsequent experiments. On Days 3, 5, 7 and 9, esophageal cancer cells were scraped into cold PBS buffer (for RNA extraction), fixed on glass coverslips for immunofluorescence analyses, or lysed directly in Cellytic reagent (Sigma-Aldrich). The efficiency of the anti-CXCL2 siRNA was evaluated by immunofluorescence analysis.

Computer-Assisted Phase-Contrast Videomicroscopy

The effects of anti-CXCL2 siRNA (vs control siRNA) on cell viability and cell growth were characterized in vitro by using computer-assisted, phase-contrast videomicroscopy, as described elsewhere.27, 28 Software that was developed in our laboratory was used to quantify the percentage of the area that was filled by the cells over the period of the experiment.28 Cells were monitored from Day 3 to Day 9 post-transfection. Photomicrographs were taken twice daily (magnification, ×10).

Quantitative Immunohistochemistry

The quantitative immunohistochemical assays have been described previously.29 For the current study, immunohistochemical labeling was performed using anti-inhibitor of differentiation 1 (anti-ID1) and anti-ID2 antibodies (dilution, 1:50) from Santa Cruz Biotechnology followed by staining with Meyer hematoxylin and Luxol Fast Blue. The ID-specific signals were evaluated quantitatively using a computer-assisted KS 400 imaging system (Carl Zeiss Vision, Hallbergmoos, Germany).

Animal Models

For the in vivo experimental models, 106 OE21 and OE33 cells were grafted subcutaneously into immunocompromized mice (6-week-old female nu/nu mice weighing 21-23 g; Janvier, Le Genest-St.-Isle, France). The potential therapeutic effects from chemotherapeutic agents were evaluated in the OE21 subcutaneous model. In the first assay, 5 mg/kg cisplatin was given as an intraperitoneal injection once weekly for 5 weeks. In the second experiment, 80 mg/kg TMZ was administered orally 3 times weekly for 6 weeks.

Tumor size was measured twice weekly with calipers and is expressed as the area (in mm2) by multiplying the 2 greatest perpendicular dimensions. Each mouse was killed for ethical reasons when it had lost 20% of its weight compared with the maximal weight gained by the mouse during the experiment. The tumors were removed, fixed in buffered formalin for 5 days, embedded in paraffin, and cut into 5-μm-thick sections. Histologic slides were stained with hematoxylin and eosin for morphologic analyses and vessel counts, as described previously.30 Immunohistochemical analyses also were carried out as detailed above. All in vivo experiments described in this study were performed in accordance with Authorization LA1230509 from the Animal Ethics Committee of the Federal Department of Health, Nutritional Safety, and the Environment (Brussels, Belgium).

Statistical Analysis

All statistical analyses were performed using Statistica software (StatSoft, Tulsa, Okla) with the Mann-Whitney and Gehan-Wilcoxon tests.


  1. Top of page
  2. Abstract

Establishment of an Experimental Model of Esophageal Cancer

In vitro, the OE21 model (a squamous cell carcinoma) displayed higher proliferation rates and resistance to confluence in vitro than the OE33 model (an adenocarcinoma) (Fig. 1B). In vivo, OE21 xenografts also grew more rapidly than those formed by OE33 cells (Fig. 1C). Histopathologic analyses revealed that OE21 xenografts displayed highly invasive features toward surrounding murine host tissues (Fig. 1D) and metastasized to the lungs and liver (data not shown), whereas OE33 xenografts did not (Fig. 1D). Therefore, the OE21 xenograft model was selected for subsequent in vivo analyses.

thumbnail image

Figure 1. These charts and photomicrographs illustrate how reliable experimental models of esophageal cancer were set up. (A) These photomicrographs show the morphologic characteristics of the 2 human esophageal cancer cell lines (OE21 and OE33) that were used in the current study. (B) This chart illustrates the global growth characteristics displayed by OE21 and OE33 cells, as assessed by means of a colorimetric 3-(4,5-dimethylthiazol-2yl)-diphenyltetrazolium bromide assay over 12 days and using different starting cell concentrations, as indicated on each chart. (C) Tumor growth characteristics are charted from the in vivo subcutaneous OE21 and OE21 xenograft models in nude mice. Ten mice were grafted with each cell line. Tumor sizes are reported as mean ± standard error of the mean (SEM) values. (D) These are typical hematoxylin and eosin-stained examples of the histopathology and morphologic characteristics of tumors from 2 in vivo xenograft models (original magnification, ×40).

Download figure to PowerPoint

Determination of C-X-C Ligand Chemokine (CXCL) and C-X-C Chemokine Receptor (CXCR) Genomic Expression in Esophageal Cell Lines and Clinical Samples

The patterns of CXCL and CXCR messenger RNA (mRNA) expression were determined (using the standard PCR techniques detailed in Table 1) in OE21 and OE33 human esophageal cancer cell lines and in 4 tissue samples from patients with esophageal squamous cell cancer (CS1-CS4). The results are summarized in Table 2 and include 1 example of how we quantified the patterns illustrated in Figure 2A. The mRNA expression patterns were classified as “−” (absence of mRNA expression), “+” (expression), and “++” (marked expression) for each of the 15 human CXCLs (CXCL15 is specific to rodents) and their 9 CXC receptors.10, 11

thumbnail image

Figure 2. The expression of C-X-C ligand chemokine (CXCL) and CXCL receptor (CXCR) messenger RNA (mRNA) is illustrated in experimental models of esophageal cancer. (A) This experimental example of CXCL/CXCR detection was obtained using a standard reverse transcriptase-polymerase chain reaction (RT-PCR) technique. Complete results are presented in Table 2. CS1 through CS4 are 4 distinct surgical human esophageal squamous carcinoma samples. (B) Quantitative RT-PCR was used to determine CXCL and CXCR mRNA expression levels in human esophageal OE21 and OE33 cancer cell lines. Three samples were analyzed per sample, and data are reported as mean ± standard error of the mean (SEM) values. cDNA indicates complementary DNA; KSHV, Kaposi sarcoma-associated herpesvirus; Duffy, the Duffy antigen/chemokine receptor. (C) This is a schematic representation of CXCLs and CXCRs that were expressed at the mRNA level in OE21 and OE33 cancer cells. The cartoon was drawn from the data reported in A and B and in Table 2. CXCLs are represented as ellipses around the cell, and CXCRs are represented by wavy lines, which represent multipass membrane receptor symbols. pb indicates pair of bases.

Download figure to PowerPoint

Table 2. Determining C-X-C Ligand Chemokine (CXCL) and CXCL Receptor Expression Levels in 4 Clinical Samples From Patients With Esophageal Cancers (CS1-CS4) and in 2 Established Human Esophageal Cancer Cell Lines (OE21 and OE33) by Reverse Transcriptase-Polymerase Chain Reaction Analysisa
  Expression Level
  • CXCL indicates C-X-C ligand chemokine; CXCR, CXCL receptor; +, weakly positive, ++, positive; −, negative; KSHV, Kaposi sarcoma-associated herpesvirus; Duffy, the Duffy antigen/chemokine receptor.

  • a

    For 1 experimental example, see Figure 2A.

CXCL1 +++++++++
CXCL2 ++++++++++++
CXCL3 ++
CXCL4 +++
CXCL5 +++++++
CXCL6 ++++
CXCL7 ++++++
CXCL8 ++++++++++++
CXCL9 +++++++++
CXCL11 +++++++
CXCL12 ++++++++
CXCL13 ++++++++
CXCL14 ++++++
CXCL16 +++++++++++

These analyses revealed a similar expression pattern in the cell lines and the surgical specimens. Chemokines and receptors for which RNA expression was determined by using standard PCR methods at least in 1 of the 2 established cell lines were validated using quantitative RT-PCR (Fig. 2B). These investigations revealed the marked overexpression of proangiogenic chemokines, particularly CXCL1, CXCL2, and CXCL8 (Fig. 2B). Figure 2C provides a schematic illustration of the types of CXCL/CXCL receptor networks that were identified in the OE21 and OE33 models.

Characterization of a Transient Knock-Down (Small Interfering RNA) of C-X-C Ligand Chemokine 2 Expression on the In Vitro Biologic Behavior of OE21 Cells

Wang et al31 highlighted the central role of GRO/CXCR2 signaling in maintaining the continued proliferation of esophageal cancer cells. In line with those findings, our objective was to further investigate the involvement/role of CXCL2 in OE21 esophageal cancer cells using an siRNA approach and monitoring with computer-assisted videomicroscopy. Figure 3A indicates that untransfected cells (wt) and cells that were transfected with either empty cationic lipids or scrambled siRNA did not display any morphologic changes (Fig. 3A, brightfield photomicrographs), and the levels of CXCL2 remained unaltered. The 2 different siRNA:cationic liposome ratios that were used yielded similar results. Cells that were transfected with CXCL2 siRNA (in both ratios) displayed marked transient down-regulation of CXCL2 between Days 5 and 9 post-transfection (Fig. 3A).

thumbnail image

Figure 3. Transient C-X-C ligand cytokine 2 (CXCL2) expression was down-regulated by means of a small interfering RNA (siRNA) approach. (A) These are morphologic illustrations of the efficiency of using an anti-CXCL2 siRNA compared with a control siRNA (SCR) to decrease CXCL2 expression in OE21 cells 3 days, 5 days, 7 days, and 9 days after siRNA transfection into the cells. All CXCL2/GROβ images were captured with the same exposure time. Between 40 and 80 cells were analyzed for each experimental condition, and representative data are shown. CT indicates cationic lipid. (B) The global growth ratio was determined by using a quantitative videomicroscopy approach for 9 days (Day+3 [D+3] through D&plus9) after the transfection of a control (SCR; open bars) versus 2 anti-CXCL2 siRNA transfection conditions (siCXCL2 transfected at 2 ratios; gray bars and black bars) into OE21 cells. Data are illustrated as the mean ± standard error of the mean (SEM) values calculated in triplicate experiments; 5 images per sample per time point were analyzed. (C) Surface colonization was evaluated by using in-house software. Data are illustrated as the area of colonized (covered) surface expressed as the mean ± SEM values calculated in triplicate experiments; 5 images per sample per time point were analyzed.

Download figure to PowerPoint

The transient depletion of CXCL2 had effects on the proliferation of transfected cell, as demonstrated by a decrease of approximately 50% in the global growth rate (in terms of cell numbers) of OE21 cells, and the effects remained after CXCL2 re-expression (Day 9 post-transfection) (Fig. 3B). Furthermore, transient CXCL2 depletion had a marked effect on the proliferation of transfected cells, as demonstrated by a reduction in their colonization capacity (Fig. 3C).

Temozolomide Decreased C-X-C Ligand Chemokine (CXCL) and CXCL Messenger RNA Secretion by Esophageal Cancer Cells

Figure 4A,B illustrates that the repeated (chronic) in vitro treatment of OE21 cells with TMZ (for 5 consecutive days) markedly decreased CXCL2 and CXCL3 mRNA and protein secretion (as evaluated by ELISA assay), whereas a single (1 day) in vitro treatment with TMZ did not significantly decrease CXCL2 and CXCL3 secretion. Similar data were obtained for OE33 cancer cells (data not shown).

thumbnail image

Figure 4. The influence of temozolomide (TMZ) is characterized in the OE21 esophageal cancer cell line. (A) Secretion of the C-X-C ligand cytokines CXCL2 and CXCL3 was assayed by using enzyme-linked immunosorbent assays. OE21 cells were left untreated (control [CT]; hatched bars) or were treated either once (gray bars) or 5 times weekly (black bars; 8 hours of TMZ treatment each day with the replacement of fresh medium containing TMZ each day) with 100 μM TMZ. The amounts of CXCL secretion were measured 24 hours after the final TMZ treatment. Data are illustrated as the mean ± standard error of the mean (SEM) values calculated in triplicate experiments and are normalized with respect to the cell counts. (B) Quantitative reverse transcriptase-polymerase chain reaction was used to analyze CXCL2 and CXCL3 messenger RNA expression levels in the human esophageal OE21 cancer cell line 24 hours after the last treatment with 100 μM TMZ (once [TMZ 1X]or 5 times weekly [TMZ 5X]). Three samples were analyzed per specimen, and the data are reported as the mean ± SEM values. cDNA indicate complementary DNA. (C) This chart illustrates the flow cytometric (FCM)-related quantification of acidic vesicular organelles (revealed as red fluorescence) after acridine orange staining in OE21 cancer cells. The cytoplasm and nucleus fluoresce green in acridine orange-stained cells. All experiments were conducted in triplicate. (D) FCM-related quantification of apoptosis was determined by using the terminal deoxynucleotidyl transferase deoxyuridine triphosphate nick-end labeling (TUNEL) technique 3 days and 6 days after the last treatment with 100 μM of TMZ (once or 5 times weekly). Each experiments was conducted in triplicate.

Download figure to PowerPoint

Figure 4C shows that both chronic treatment and single treatment did induce increases in acidic vesicular organelle amounts in OE21 cells that were treated 5 times with TMZ. Figure 4D shows that both chronic and single treatments with 100 μM temozolomide induced apoptotic cell death in OE21 cancer cells 3 days after the last TMZ treatment.

Temozolomide Contributes In Vivo Therapeutic Benefits in OE21 Esophageal Cancer Xenografts

Cisplatin, a proapoptotic drug that is widely used to treat esophageal cancers, did not impair OE21 xenograft development or growth in immunocompromised mice (Fig. 5A). Accordingly, cisplatin was unable to contribute any significant therapeutic benefit in terms of overall survival in OE21 tumor-bearing mice (Fig. 5C). In contrast, we observed a significant delay in OE21 xenograft growth with TMZ using the same experimental model (P = .02; Mann-Whitney test at Day 63, the last day all mice were alive) (Fig. 5B). OE21 xenograft-bearing mice that received TMZ treatment also displayed a significant improvement in overall survival compared with untreated mice (P = .02; Gehan-Wilcoxon test) (Fig. 5D).

thumbnail image

Figure 5. Chemotherapy-induced antitumor effects in esophageal cancer were evaluated in an in vivo xenograft model. OE21 xenografts in nude mice were treated with (A,C) cisplatin (5 mg/kg given intraperitoneally [i.p.] once weekly for 5 weeks; the treatment was started 22 days after the tumor graft) or (B,D) TMZ (80 mg/kg administered orally [p.o.] 3 times weekly for 6 weeks; the treatment was started 21 days after the tumor graft). Each group included 10 mice. Respective treatment schedules and durations are indicated by black arrows above each graph. CT indicates control. (A,B) Tumor size was measured twice weekly with a caliper. The area (in mm2 [sq-mm]) was determined by multiplying the 2 greatest perpendicular dimensions and is expressed as the mean ± standard error of the mean (SEM). (C,D) Kaplan-Meier curves represent comparisons of the overall survival of treated mice versus untreated (control) mice.

Download figure to PowerPoint

It must be emphasized that, in cisplatin versus control experiments (Fig. 5A,C), the median survival of control mice was approximately 200 days; whereas, in TMZ-versus-control experiments (Fig. 5B,D), the median survival of control mice was <100 days. These differences translate into distinct levels of locoregional and/or metastatic processes (see Fig. 1) with the OE21 model from 1 experiment to another (data not shown).

Temozolomide Decreases Angiogenesis in Esophageal Cancer Xenografts

Taking into consideration the observed, in vitro, TMZ-induced down-regulation of proangiogenic CXCL secretion in esophageal cancer cells (Fig. 4A) and the TMZ-induced antiangiogenic effects demonstrated previously in gliomas,32 we investigated potential TMZ-induced antiangiogenic effects in OE21 esophageal cancer xenografts. Figure 6A illustrates the significantly decreased global tumor vascularization in tumors from TMZ-treated mice (represented by decreased numbers of blood vessels in treated samples) compared with control, untreated xenografts.

thumbnail image

Figure 6. Temozolomide (TMZ)-induced antiangiogenic effects are characterized in the OE21 esophageal cancer xenograft model. (A) Quantitative angiogenesis data for tumor samples derived from untreated control and TMZ-treated OE21 xenograft-bearing mice are expressed as percentages of counted blood vessels (the control [CT] level was set at 100%). Ten fields per tumor sample were analyzed. (B) This is a morphologic illustration of immunostaining with 2 inhibitor of untreated differentiation (ID) antibodies (ID1 and ID2) in tumor samples that were obtained from in vivo studies. (C) ID-dependent signals were evaluated quantitatively by using a computer-assisted KS 400 imaging system (Carl Zeiss Vision, Hallbergmoos, Germany). In each case, 15 fields corresponding to a surface area that ranged between 60,000 μm2 and 120,000 μm2 were scanned. The computer-assisted morphometric analyses of the parameters for immunohistochemical expression of each marker were quantitative, and the results represent the mean optical density, which corresponds to the staining intensity. SEM indicates standard error of the mean.

Download figure to PowerPoint

TMZ-induced antiangiogenic affects reportedly are mediated, at least in part, through the down-regulation of ID1 and ID2 factors in glioma.32, 33 Accordingly, we used quantitative immunohistochemistry to investigate the expression of ID1 and ID2 in tumor samples from untreated mice and from TMZ-treated OE21 xenograft-bearing mice. TMZ induced the significant down-regulation of ID1 protein expression (P < .05; Mann-Whitney test), whereas ID2 expression remained unaltered by TMZ treatment (Fig. 6B,C).


  1. Top of page
  2. Abstract

A better understanding of the molecular events involved in the development of esophageal cancer may offer opportunities to identify diagnostic markers, therapeutic targets, or prognostic indicators. In line with these possibilities, Wang et al22 used combinational DNA microarray analysis to identify genes potentially involved in the development of esophageal squamous cell carcinoma and reported that primary esophageal tumor tissues expressed elevated levels of the chemokines CXCL1 (GROα) and CXCL2 (GROβ) relative to their expression in adjacent normal tissues. The expression and role(s) of the proteins that belong to this family, with the exceptions of CXCL8 and CXCL12 and its receptor CXCR4, has not been studied extensively in esophageal cancer.14-21 The current study reveals similar mRNA expression patterns in the OE21 and OE33 cell lines and in the surgical specimens and the marked overexpression of proangiogenic chemokines, such as GRO CXCLs (CXCL1, CXCL2, and CXCL3) and CXCL8.

Previous reports have indicated significant improvements in the outcome of patients with glioma who received TMZ.7 Therefore, our objective was to assess the potential beneficial effects of TMZ in models of esophageal cancer, which is an apoptosis-resistant cancer, because, to our knowledge, TMZ has never been studied in this setting. We detected notable in vivo antitumor effects with TMZ (Fig. 5B,D). It is noteworthy that the delay in tumor mass development was sustained during TMZ treatment and progressed after the cessation of TMZ treatment (Fig. 5B). This effect may have been the result of the TMZ-induced down-regulation of CXCL2 and CXCL3 secretion that we observed with chronic, in vitro TMZ treatment, but not with a single TMZ treatment (Fig. 4A). The current results indicate that siRNA targeting of CXCL2 indeed markedly affected esophageal cancer cell proliferation (Fig. 3).

It appears that the antitumor effects of TMZ also may be the result of TMZ-induced proautophagic effects that overcome resistance to apoptosis.7 Indeed, when sustained, TMZ-induced proautophagic effects34, 35 lead to late apoptosis,36 even in cancer cells that resist various proapoptotic stimuli, such as glioma cells.7, 8

The antiangiogenic properties of TMZ may present a third possibility for the antitumor effects of TMZ in OE21 xenograft models, which are mediated at least in part through ID factor signaling.37 The absence of a TMZ-mediated effect on ID2 expression, although it had an inhibitory effect on ID1 expression (Fig. 6C), suggests that TMZ induces vascular endothelial growth factor (VEGF)-independent antiangiogenic effects, because ID2 functions as a master regulator of VEGF expression.37 The combination of TMZ and VEGF-targeted therapy presents a novel therapeutic strategy that should be investigated for its potential to further improve the outcome of patients with esophageal cancer. In line with this possibility, Noma et al38 reported that the essential role of fibroblasts in esophageal squamous cell carcinoma-induced angiogenesis was mediated through the release of VEGF. Furthermore, Ren et al16 reported that human esophageal carcinomas express and secrete large amounts of macrophage migration-inhibitory factor, which may serve as an autocrine factor in angiogenesis through its effects on VEGF and interleukin-8. Hepatocyte growth factor has been proposed as a useful biomarker for esophageal tumor progression, because it promotes cancer cell migration and VEGF and CXCL8 expression.17 Finally, it is noteworthy that, in their review of emerging drugs for esophageal cancer, Homs et al1 emphasized a growing interest in the effect of angiogenesis inhibitors on esophageal cancer.

Collectively, our results to date 1) emphasize the role of proangiogenic chemokines in esophageal cancer biology, 2) bring a novel proposal for therapy involving CXCL chemokine inhibition to improve clinical outcomes, and 3) demonstrate noteworthy therapeutic benefit with TMZ in experimental esophageal cancer models, such as effects on both cancer cell proliferation and CXCL-dependent but VEGF-independent angiogenesis. Although several therapeutic agents currently are under investigation (as monotherapy or multiagent therapy) in several clinical trials for esophageal cancers,1 to our knowledge, no trial has been designed to evaluate the effects of TMZ. The current results suggest that TMZ administration may be a potential new therapeutic strategy for esophageal cancer.


  1. Top of page
  2. Abstract

The current work was supported by grants from the Fonds Yvonne Boel (Brussels, Belgium).


  1. Top of page
  2. Abstract