Anti-EphA2 Antibodies with Distinct In Vitro Properties Have Equal In Vivo Efficacy in Pancreatic Cancer

Abstract

The EphA2 receptor tyrosine kinase is overexpressed in a variety of human epithelial cancers and is a determinant of malignant cellular behavior in pancreatic adenocarcinoma cells. Moreover, it is expressed in tumor endothelium and its activation promotes angiogenesis. To better clarify the therapeutic potential of monoclonal antibodies (mAbs) directed to the EphA2 receptor, we generated a large number of mAbs by differential screening of phage-Ab libraries by oligonucleotide microarray technology and implemented a strategy for the rapid identification of antibodies with the desired properties. We selected two high-affinity and highly specific EphA2 monoclonal antibodies with different in vitro properties on the human pancreatic tumor cell line MiaPaCa2. One is a potent EphA2-agonistic antibody, IgG25, that promotes receptor endocytosis and subsequent degradation, and the second is a ligand antagonist, IgG28, that blocks the binding to ephrin A1 and is cross-reactive with the mouse EphA2 receptor. We measured the effect of antibody treatment on the growth of MiaPaCa2 cells orthotopically transplanted in nude mice. Both IgG25 and IgG28 had strong antitumor and antimetastatic efficacy. In vivo treatment with IgG25 determined the reduction of the EphA2 protein levels in the tumor and the phosphorylation of FAK on Tyr576 while administration of IgG28 caused a decrease in tumor vascularization as measured by immunohistochemical analysis of CD31 in tumor sections. These data show that in a pancreatic cancer model comparable therapeutic efficacy is obtained either by promoting receptor degradation or by blocking receptor activation.

Figure 1

In vitro characterization of IgG25 and IgG28. (a) FACS-based whole cell binding assay was performed with IgG25 and IgG28 on human MiaPaCa2 and (b) on mouse MC38-CEA cells to determine the apparent Kd. Mean Fluorescence Intensities (MFIs) obtained with IgG25 and IgG28 over the logarithm of their molar concentration (LogM) are reported. (c) Binding competition experiments of IgG25 and IgG28 with ephrinA1 on MiaPaCa2 cells, a control isotypic IgG1 antibody (Ctrl IgG) was used as negative control. IgG25 (filled circles), IgG28 (empty circles), Ctrl IgG (filled triangles). (d) EphA2 immunoprecipitation from lysates of cells treated with Ctrl IgG, IgG25, IgG28 and ephA1-Fc, followed by Western Blot with antiphosphotyrosine antibody. After stripping, the same filter was probed with EphA2 antibody as loading control. (e) FACS-based whole cell binding with IgG25 and IgG28 on mouse N2A cells transiently transfected with expression vectors coding for members of Eph A receptor family (EphA1, EphA2, EphA3, EphA4, EphA5, and EphA7).

Figure 2

IgG25- and IgG28-mediated EphA2 internalization and degradation in MiaPaCa2 cells. (a) EphA2 internalization in response to IgG25, IgG28, and ephrinA1-Fc treatment. mAbs and ephrinA1-Fc labeled in red while cell nuclei in blue. Images were acquired at 20× magnification. Localization was revealed 1 hour after incubation on cells either at 4°C or at 37°C. (b) Time course Western blot analysis of EphA2 degradation after treatment with control IgG, IgG25, IgG28, and ephrinA1-Fc; anti-β-actin was used as loading control. (c) Densitometric analysis of the level of EphA2 expression measured by Western Blot in cells treated with control IgG (asterisks), IgG25 (squares), IgG28 (triangles), and ephrinA1-Fc (circles). Data are expressed as percentage of EphA2 expression over time.

Figure 3

EphA2 downstream signaling in MiaPaCa2 cells treated with IgG25 and IgG28. Cell lysates of MiaPaCa2 were assayed by Western Blotting with antiphospho-Akt, antiphospho-ERK, and anti-phospho FAK (Tyr 576). Antibodies directed to total Akt, total FAK, and actin were used as loading controls.

Figure 4

In vivo activity of IgG25 and IgG28 in MiaPaCa2 orthotopic xenografts. (a) Pharmacokinetic analysis of IgG25 (empty circle) and IgG28 (full circle) following a single dose at 2 mg/kg. The straight line represents the trough level of the antibody before the next administration. (b) Tumor weight at day 35 following biweekly administration of control IgG, IgG25, and IgG28. Data, expressed in grams, represent the average of tumor weight from ten different animals (N = 10). Error bars indicate the standard error. (c) Impact of IgG treatment on metastatization: the metastatic incidence is given by the number of mice with metastases in each group. The total number of metastases counted for each group is also reported. Metastatic lesions were normally distributed in liver, spleen, stomach, diaphragm, and, in few cases, kidney tissues. Involvement of local lymph nodes was not detected. Asterisks indicate statistically significant differences with respect to the control group (Student's t-test; P<.05)

Figure 5

Effect of EphA2 antibodies on angiogenesis. (a) Representative CD31 immunostaining of paraffin embedded tumor section (magnification bar = 20 m). (b) Quantitative analysis of CD31 staining in tumors treated with control IgG, IgG25, and IgG28. The average data obtained from the analyses of two tumors selected from each group are reported as the percentage of CD31⁺ area in each entire section. In all panels, asterisks indicate statistically significant differences with respect to the control group (Student's t-test; P<.05).

Figure 6

EphA2 protein expression in tumors from treated mice. (a) Western Blot analysis with anti-EphA2 antibodies of  lysates from five tumors treated with control IgG, with IgG25, or six tumors treated with IgG28. (b) Densitometric analysis of the ratio between EphA2 levels revealed by anti-EphA2 antibody in nonsaturating conditions and actin expression measured with antiactin antibody. Data on y axis are expressed as arbitrary units.

Figure 7

IgG25 and IgG28 modulate in vivo EphA2 downstream signaling: Western Blot analyses of tumor lysates with antiphospho FAK (Tyr 576) and anti-phospho Akt. Antibodies to total Akt, total FAK, and actin were used as loading controls. P values were calculated with respect to average densitometric value of group treated with control IgG.