Targeting promiscuous heterodimerization overcomes innate resistance to ERBB2 dimerization inhibitors in breast cancer

Abstract

Background: The oncogenic receptor tyrosine kinase (RTK) ERBB2 is known to dimerize with other EGFR family members, particularly ERBB3, through which it potently activates PI3K signalling. Antibody-mediated inhibition of this ERBB2/ERBB3/PI3K axis has been a cornerstone of treatment for ERBB2-amplified breast cancer patients for two decades. However, the lack of response and the rapid onset of relapse in many patients now question the assumption that the ERBB2/ERBB3 heterodimer is the sole relevant effector target of these therapies.

Methods: Through a systematic protein-protein interaction screen, we have identified and validated alternative RTKs that interact with ERBB2. Using quantitative readouts of signalling pathway activation and cell proliferation, we have examined their influence upon the mechanism of trastuzumab- and pertuzumab-mediated inhibition of cell growth in ERBB2-amplified breast cancer cell lines and a patient-derived xenograft model.

Results: We now demonstrate that inactivation of ERBB3/PI3K by these therapeutic antibodies is insufficient to inhibit the growth of ERBB2-amplified breast cancer cells. Instead, we show extensive promiscuity between ERBB2 and an array of RTKs from outside of the EGFR family. Paradoxically, pertuzumab also acts as an artificial ligand to promote ERBB2 activation and ERK signalling, through allosteric activation by a subset of these non-canonical RTKs. However, this unexpected activation mechanism also increases the sensitivity of the receptor network to the ERBB2 kinase inhibitor lapatinib, which in combination with pertuzumab, displays a synergistic effect in single-agent resistant cell lines and PDX models.

Conclusions: The interaction of ERBB2 with a number of non-canonical RTKs activates a compensatory signalling response following treatment with pertuzumab, although a counter-intuitive combination of ERBB2 antibody therapy and a kinase inhibitor can overcome this innate therapeutic resistance.

Keywords: Breast cancer; ERBB2; Heterodimers; Pertuzumab; Receptor tyrosine kinases.

Fig. 1

Targeting ERBB2 with therapeutic monoclonal antibodies. a Western blotting showing the expression of ERBB2, pAkt^S473 and Akt in a panel of breast cancer cell lines. b Cell viability assays performed on the panel of lines with trastuzumab and pertuzumab at the concentrations indicated for 5 days (n = 6, mean ± SD). c Combinatorial cell viability assays with trastuzumab and pertuzumab at the concentrations indicated for 5 days (n = 6, mean). Raw data is presented in Additional file 2: Figure S1B. d Western blotting showing the phosphorylation of ERBB3^Y1289 and Akt^S473 following treatment with trastuzumab (100 nM), pertuzumab (100 nM) and the combination of both, at the time points indicated. Blots were re-probed with an actin antibody for quantification. A representative image is shown from three independent replicates (n = 3, mean ± SD). All drug treatments were significantly different from control. For simplicity, the indicated significance compares the combination to trastuzumab only (*p < 0.05, **p < 0.01)

Fig. 2

ERBB2 interaction screen. a Schematic of the BiFC assay used to visualise ERBB2 dimerization. b Confocal fluorescence microscopy of HEK-293 T cells following transfection with plasmids containing a full-length Venus control, ERBB2-V1, ERBB2-V2 or co-transfected with ERBB2-V1 and ERBB2-V2 (scale bar = 25 μm). c Dendrogram showing the sequence-based relationship between all 61 RTKs and their ERBB2 interaction status detected using the BiFC assay and high-content imaging. The dendrogram was adapted from Manning et al. 2002 (24). d Confocal fluorescence microscopy of HEK-293 T cells following transfection with plasmids containing ERBB2-V1 and the V2 tagged RTKs indicated (scale bar = 15 μm)

Fig. 3

Analysis of RTK activation. a An example phospho-RTK antibody array showing the location of relevant RTKs within the array. b Phospho-RTK arrays developed using lysates from ZR-75-30, BT474 and AU565 cells treated with either trastuzumab (100 nM, 4 h) or pertuzumab (100 nM, 4 h). c AU565 and BT474 cells were treated with pertuzumab (100 nM) for the time points indicated. Western blotting was performed with lysates from these cells with the total and phospho-RTK antibodies indicated. Data is presented as a heat map showing the relative changes in phosphorylation (n = 3, mean), raw data and statistical analysis is presented in Additional file 2: Figure S4

Fig. 4

Receptor interactions with ERBB2. a Western blotting showing the co-immunoprecipitation of the indicated receptors with ERBB2 from BT474 cells and AU565 cells. Each line was treated with pertuzumab (100 nM) for the time period indicated, and immunoprecipitation was performed with either a mouse IgG control or ERBB2 monoclonal antibody. Blots are representative of three experiments. b Confocal fluorescence microscopy imaging of proximity mediated ligation assays showing the interaction between ERBB2 and the indicated RTKs in BT474 and AU565 cells (scale bar = 25 μm)

Fig. 5

Targeting pertuzumab-induced signalling. a AU565 and BT474 cells were treated with pertuzumab (100 nM) for the time points indicated. Western blotting was performed with lysates from these cells with the total and phospho-antibodies indicated. Data is presented as a heat map showing the relative changes in phosphorylation (n = 3, mean); raw data and statistical analysis are presented in Additional file 2: Figure S6. b AU565 cells expressing the ERK-KTR clover biosensor were treated with pertuzumab (100 nM). Live cell imaging was performed to measure the cytoplasmic to nuclear (C/N) ratio of the ERK-KTR biosensor at the time points indicated (mean ± SEM). c Western blotting showing the co-immunoprecipitation of GRB2 and SHC with ERBB2 from BT474 cells and AU565 cells. Each line was treated with pertuzumab (100 nM) for the time period indicated and immunoprecipitation was performed with either a mouse IgG control or ERBB2 monoclonal antibody. Blots are representative of three experiments. d AU565 cells were treated with pertuzumab (100 nM) or UO126 (10 μM) for 30 min. Western blotting was performed with the antibodies indicated. Blots are representative of three experiments. e Cell viability assay was performed using the AU565 cell line with pertuzumab at the concentrations indicated, in the presence or absence of UO126 (10 μM) for 5 days (n = 6, mean ± SD, *p < 0.05). f AU565 cells were treated with the combination of pertuzumab (100 nM), BMS-777607 (1 μM) or cabozantinib (1 μM) for 30 min. Western blotting was performed with the antibodies indicated. Blots are representative of three experiments. g Cell viability assay was performed using the AU565 cell line with pertuzumab at the concentrations indicated, in the presence or absence of BMS-777607 (1 μM) or cabozantinib (1 μM) for 5 days (n = 6, mean ± SD)

Fig. 6

Synergy between pertuzumab and lapatinib in vitro. a Cell viability assays performed on a panel of ERBB2+ breast cancer lines with lapatinib at the concentrations indicated for 5 days (n = 6, mean ± SD). b Western blotting showing the co-immunoprecipitation of GRB2 and SHC with ERBB2 from AU565 cells treated with pertuzumab (100 nM) or lapatinib (100 nM) as indicated, for 30 min. Immunoprecipitation was performed with either a mouse IgG control or ERBB2 monoclonal antibody. Blots are representative of three experiments. c BT474 cells and d AU565 cells were treated with pertuzumab (100 nM) or lapatinib (100 nM) as indicated, for 30 min. Western blotting was performed with the antibodies indicated. Blots are representative of three experiments. e BT474 cells and f AU565 cells were used to perform combinatorial cytotoxicity assays with lapatinib and pertuzumab at the concentrations indicated for 5 days (n = 6, mean). Raw data is presented in Additional file 2: Figure S6C, D. Synergy was calculated as a Combination Index (CI), using Compusyn (v1)

Fig. 7

Synergy between pertuzumab and lapatinib in vivo. a IHC staining for RTK expression as indicated in the HCI-012 PDX model. (Scale bar = 25 μm). b Proximity mediated ligation assays detected heterodimers between ERBB2 and other RTKs as indicated, shown in red. (Scale bar = 25 μm). c Tumour growth following treatment with either vehicle, pertuzumab (12 mg/kg loading dose, 6 mg/k6 maintenance dose, intraperitoneal injection, once weekly), lapatinib (50 mg/kg, oral gavage, 5 times a week) or the combination of pertuzumab and lapatinib (n = 6, mean ± SEM, *p < 0.05). d Growth rate for each treatment arm calculated from a linear regression analysis (n = 6, mean ± SEM, **p < 0.01). e IHC staining for Ki67 from day 14 tumours. Analysis was performed on three random fields of view from three tumours for each arm (n = 9, mean ± SD, **p < 0.01)