The aim of this study was investigate a possible interaction of the HER2- and CXCR4-receptors and their expression levels under treatment with their respective inhibitors in order to determine an impact of CXCR4-expression in HER2-positive esophageal carcinoma.
In vitro proliferation and migration of esophageal OE19 cells
To investigate the effects of the CXCR4- and HER2-receptors on cell proliferation in vitro OE19 cells were treated with AMD3100 and trastuzumab, respectively. While it was expected that trastuzumab treatment leads to a cell growth reduction, the effect of treatment with AMD3100 had not been defined. Cell migration of the CXCR4-positive cells however should be influenced by the chemoattractant SDF-1α.
The in vitro
cell proliferation assays reproducibly showed a significant reduction of cell proliferation under HER2- and CXCR4-receptor inhibition after treatment with trastuzumab (p
0.005) as well as with AMD3100 (p
0.02) (). As expected, treatment with trastuzumab led to a stronger reduction of proliferation than treatment with AMD3100. In the migration assay a dose-dependent effect on cell migration could be observed after treatment with SDF-1α (). Chemotaxis of CXCR4-positive OE19 cells could thus be induced by SDF-1α (). Results were reproduced with several concentrations (data not shown).
Figure 1 A Effect of trastuzumab and ADM3100 treatment on proliferation (%) of OE19 cells compared to control in the lactate-dehydrogenase assay. Receptor inhibition leads to reduced proliferation of cells. It shows a significant reduction of cell proliferation (more ...)
In vivo proliferation of primary tumor in the orthotopic model
To further investigate the behaviour of local and metastatic esophageal tumor growth in vivo
OE19 cells were implanted orthotopically into NMRI/nu mice. All mice developed primary tumors at the implantation site as shown by MRI two weeks after implantation. The bodyweight of the mice ranged from 18.5–26.3 g at the beginning and 22.5–32.7 g at the termination of the experiment. The means of bodyweights of each therapeutic group at the beginning and termination of the experiment are summarized in . No significant variation was observed. Mice were randomized two weeks after implantation into therapeutic groups. We have previously shown that tumor sizes two weeks after implantation were comparable between groups 
. At the time of termination of the experiment, an MRI scan was performed immediately before dissecting the animals. All animals reached the end point of the study without severe weight loss or other signs of tumor disease. Tumor weights were recorded and gave values between 0.01–3.9 g. Tumor volumetry was performed and confirmed the tumor weight results (). While weight values within the control, trastuzumab-treated group, and trastuzumab/AMD3100-treated group were more homogenous, values varied more strongly within the AMD3100-treated group. Tumor weights in the control group were significantly higher than in the trastuzumab-treated (p<0.0001) and trastuzumab/AMD3100-treated (p<0.0001) groups. Tumor weights in the AMD3100-treated group were significantly higher than in the trastuzumab-treated (p
0.04) and trastuzumab/AMD3100-treated (p
0.02) groups (). Although the effect of AMD3100 on the primary tumor weight was not as significant as the effect of trastuzumab, a potent effect was achieved by AMD3100 treatment alone compared to the untreated group. The tumor weights at time of autopsy correlated significantly with the volume measured by MRI (correlation coefficient: 0.837, p<0.01). Representative examples of magnetic resonance images for tumor evaluation with and without treatment are shown in .
Mean Body Weight, Tumor Weight and Volume of Mice.
Figure 2 A Significant differences in tumor weights between control and trastuzumab-treated groups (p<0.00), control and combination trastuzumab/AMD3100-treated groups (p<0.00), trastuzumab and AMD3100-treated groups (p=0.04), (more ...)
Metastastic potential of esophageal tumor cells in vivo
At the termination of the in vivo experiment, potentially metastatic tissues (lung, liver and lymph nodes) were sampled and histologically examined to evaluate the metastatic spread of the esophageal tumor for each treatment group. Additionally, micrometastases in liver and lung were detected by mRNA expression of the human gapdh gene by real-time PCR analysis from total RNA.
In the control group extensive metastastic spread was observed to lung, liver and lymph nodes. 20% of animals had aggressive metastatic spread to all compartments including lung, liver and multiple lymph nodes. In contrast, the AMD3100-treated group showed fewer metastases and only one animal expressed highly aggressive metastatic spread to three compartments (lung, liver, and lymph nodes). While metastatic spread in the combination-therapy group was not significantly lower than in the trastuzumab-treated group, metastasis presented solitarily.
When examining overall metastasic spread in H.E. and immunohistological staining there was significantly less metastatic spread found in the trastuzumab-treated group compared to the control group (p
0.002). In particular, a significant reduction of lung metastases could be observed compared to the control group after AMD3100, trastuzumab and combined treatment as well as to the liver after trastuzumab and combined treatment (). Micrometastic values are presented as delta ct-values of the control and treated groups.
Disseminated tumor cells in bone marrow
To determine the presence of disseminated tumor cells in the bone marrow, disseminated tumor cells (DTC) were isolated from bone marrow of mice by density gradient and visualized by chromagen immunostaining. The finding of cytokeratin-positive DTC in the bone marrow indicated the extent of tumor disease ().
CXCR4 and HER2 expression profile of OE19 cells in vivo
Firstly, OE19 cells were examined for their expression of CXCR4 and HER2. Not only the HER2-overexpression but also the amplification of its gene is of clinical relevance, thus Her2-amplification status was verified. OE19 cells showed a strong expression of CXCR4- and HER2-receptors in immunostaining () as well as an amplification of the Her2-gene in FISH (). Semiquantitive mRNA analysis showed expression of CXCR4 and Her2 compared with MDA-MB-231 and SKBr-3 cell lines ().
Figure 3 A CXCR4-expression of OE19 cells determined by fluorescence immunostaining (IgG1-control) B Confirmation of Her2-amplification determined by fluorescence in situ hybridization (red: Her2-gene loci, green reference CENT-17-loci) C
CXCR4 and HER-2 mRNA-expression (more ...)
Correlation of CXCR4- and HER2-expression in the orthotopic in vivo model
Secondly, primary tumor and metastatic tissues from the orthotopic model were examined for CXCR4- and HER2-expression. CXCR4- and HER2-expression was observed in all tumor bearing-tissues, including primary tumor, liver, lung and lymph node metastases (). HER2-expression correlated significantly with CXCR4-expression (correlation efficient 0.490, p<0.01).
Higher intensity of HER2-expression in metastases compared to primary tumor
To further evaluate the relevance of HER2- and CXCR4-correlation, a point-by-point diagram was designed (), in which each metastatic case was marked, indicating both the intensity of expression of the metastasis (y-axis) and the intensity of expression of its respective primary tumor (x-axis). According to the treatment group different symbols were used. The first diagram displays the HER2-intensity, the second the CXCR4-intensity.
Interestingly, a higher expression of HER2 and CXCR4 could be seen in metastases of all therapeutic groups compared to their respective primary tumors. The intensity of HER2- expression (score 1–3) of primary tumors and their respective metastases were applied in the first diagram in . The graph showed that the intensity of the HER2-positivity by immunostaining varies between tissues of treatment groups. While the HER2-positivity of primary tumor in the control group was limited to lighter intensity (score 1), metastases expressed HER2 at all intensities. Under AMD3100 treatment, however, HER2 was only expressed at stronger intensities (scores 2+3) in the primary tumor. Metastases in the AMD3100-treated group express HER2 almost exclusively at the highest intensity (score 3).
When applying the intensity scores of CXCR4-positivity (scores 1–3) of the primary tumors and their respective metastases to the second diagram in , the scores in the control group varied between lighter and medium intensities (scores 1+2). Although the intensity of CXCR4-expression levels of the primary tumors in the AMD3100 treated group reached higher expression levels (score 3), the intensity of metastatic CXCR4-expression did only extent to medium levels (score 2).
In both diagrams, the trastuzumab-treated group could not be evaluated in this way as there were no metastases. For the combined therapy group, there were only two metastatic cases making evaluation statistically not reliable.
Upregulation of HER2-expression under AMD3100 treatment
Hypothesising that the expression of CXCR4 and/or HER2, their respective pathways and their interaction are involved in mediating tumor progression and metastatic homing, each therapeutic group was evaluated separately as to the intensity of HER2- and CXCR4-expression levels. Using the same diagram () the intensity of HER2- and CXCR4-expression was compared between the treatment groups. In comparison to the control group, the HER2-intensities of metastases and primary tumors of the AMD3100-treated group are represented by higher scores throughout.
When statistically comparing the HER2 expression of primary tumor and metastases between therapeutic groups, a significantly higher HER2-expression was observed in the trastuzumab-treated group (p
0.003) and the AMD3100-treated group (p
0.003) compared to the control group. Upon examination of CXCR4-expression, a significantly higher expression was observed in the trastuzumab-treated group (p
0.003) and a higher expression in the AMD3100-treated group (p
0.065) compared to the control group. The combined therapy group (trastuzumab/AMD3100) neither showed significant HER2-expression or CXCR4-expression differences compared to the control group.
Validation of CXCR4- and HER2-coexpression in human esophageal carcinoma
To further validate the correlation of HER2 and CXCR4 that was found in the in vivo
studies, primary tumor tissues of 202 patients were examined for both HER2- and CXCR4-expression by immunostaining. Importantly, we have already previously shown that CXCR4-expression in a similar patient collective (136 patients) was associated with poor clinical outcome in esophageal cancer patients 
. All patients were operated in curative intention. M1 patients with distant lymph node metastases were included. Patient characteristics are summarized in .
Receptor staining was classified into low, medium and high expression. Of the HER2-positive tumor samples, only 14 showed high HER2-expression. In the primary tumor tissue of these patients (n
14), a significantly positive correlation (p
0.036) could be observed between high HER2 (57.14%) and high CXCR4 (42.86%) expression ().
HER2- and CXCR4-receptor expression.