ERBB3/HER2 signaling promotes resistance to EGFR blockade in head and neck and colorectal cancer models.

EGFR blocking antibodies are approved for the treatment of colorectal cancer and head and neck squamous cell carcinoma (HNSCC). Although ERBB3 signaling has been proposed to limit the effectiveness of EGFR inhibitors, the underlying molecular mechanisms are not fully understood. To gain insight into these mechanisms, we generated potent blocking antibodies against ERBB3 (REGN1400) and EGFR (REGN955). We show that EGFR and ERBB3 are coactivated in multiple HNSCC cell lines and that combined blockade of EGFR and ERBB3 inhibits growth of these cell lines more effectively than blockade of either receptor alone. Blockade of EGFR with REGN955 strongly inhibited activation of ERK in HNSCC cell lines, whereas blockade of ERBB3 with REGN1400 strongly inhibited activation of Akt; only the combination of the 2 antibodies blocked both of these essential downstream pathways. We used a HER2 blocking antibody to show that ERBB3 phosphorylation in HNSCC and colorectal cancer cells is strictly dependent on association with HER2, but not EGFR, and that neuregulin 1 activates ERBB3/HER2 signaling to reverse the effect of EGFR blockade on colorectal cancer cell growth. Finally, although REGN1400 and REGN955 as single agents slowed the growth of HNSCC and colorectal cancer xenografts, the combination of REGN1400 plus REGN955 caused significant tumor regression. Our results indicate that activation of the Akt survival pathway by ERBB3/HER2 limits the effectiveness of EGFR inhibition, suggesting that REGN1400, which is currently in a phase I clinical trial, could provide benefit when combined with EGFR blocking antibodies.

Monoclonal antibody classification based on epitope-binding using differential antigen disruption.

Currently, classifying a population of specific antigen-reactive monoclonal antibodies (mAbs) according to their epitope-binding properties has been limited to competition assays. Such assays are time consuming, labor intensive and restricted to the number of mAbs in the experiment. To overcome this problem, a differential antigen disruption-based antibody profiling procedure was developed. This procedure rapidly classifies specific antigen-reactive mAbs into epitope-related groups by measuring the binding signal of the antibodies to a set of structurally disrupted antigens and then clustering the antibodies according to the similarity of their binding profiles. The clustering results generated by differential antigen disruption showed a significant concordance with those generated by competition experiments. Therefore, differential antigen disruption method opens an opportunity to assess the entire population of antigen-reactive mAbs according to their epitope-binding properties. In doing so, a set of representative antibodies can be drawn to describe the epitope complexity for systematically exploring their functions.

Kinetic analysis of a high-affinity antibody/antigen interaction performed by multiple Biacore users.

To explore the reliability of Biacore-based assays, 22 study participants measured the binding of prostate-specific antigen (PSA) to a monoclonal antibody (mAb). Each participant was provided with the same reagents and a detailed experimental protocol. The mAb was immobilized on the sensor chip at three different densities and a two-step assay was used to determine the kinetic and affinity parameters of the PSA/mAb complex. First, PSA was tested over a concentration range of 2.5-600 nM to obtain k(a) information. Second, to define the k(d) of this stable antigen/antibody complex accurately, the highest PSA concentration was retested with the dissociation phase of each binding cycle monitored for 1h. All participants collected data that could be analyzed to obtain kinetic parameters for the interaction. The association and the extended-dissociation data derived from the three antibody surfaces were globally fit using a simple 1:1 interaction model. The average k(a) and k(d) for the PSA/mAb interaction as calculated from the 22 analyses were (4.1+/-0.6) x 10(4) M(-1) s(-1) and (4.5+/-0.6) x 10(-5) s(-1), respectively. Overall, the experimental standard errors in the rate constants were only approximately 14%. Based on the kinetic rate constants, the affinity (K(D)) of the PSA/mAb interaction was 1.1+/-0.2 nM.