An engineered bivalent neuregulin protects against doxorubicin-induced cardiotoxicity with reduced proneoplastic potential

Abstract

Background: Doxorubicin (DOXO) is an effective anthracycline chemotherapeutic, but its use is limited by cumulative dose-dependent cardiotoxicity. Neuregulin-1β is an ErbB receptor family ligand that is effective against DOXO-induced cardiomyopathy in experimental models but is also proneoplastic. We previously showed that an engineered bivalent neuregulin-1β (NN) has reduced proneoplastic potential in comparison with the epidermal growth factor-like domain of neuregulin-1β (NRG), an effect mediated by receptor biasing toward ErbB3 homotypic interactions uncommonly formed by native neuregulin-1β. Here, we hypothesized that a newly formulated, covalent NN would be cardioprotective with reduced proneoplastic effects in comparison with NRG.

Methods and results: NN was expressed as a maltose-binding protein fusion in Escherichia coli. As established previously, NN stimulated antineoplastic or cytostatic signaling and phenotype in cancer cells, whereas NRG stimulated proneoplastic signaling and phenotype. In neonatal rat cardiomyocytes, NN and NRG induced similar downstream signaling. NN, like NRG, attenuated the double-stranded DNA breaks associated with DOXO exposure in neonatal rat cardiomyocytes and human cardiomyocytes derived from induced pluripotent stem cells. NN treatment significantly attenuated DOXO-induced decrease in fractional shortening as measured by blinded echocardiography in mice in a chronic cardiomyopathy model (57.7±0.6% versus 50.9±2.6%, P=0.004), whereas native NRG had no significant effect (49.4±3.7% versus 50.9±2.6%, P=0.813).

Conclusions: NN is a cardioprotective agent that promotes cardiomyocyte survival and improves cardiac function in DOXO-induced cardiotoxicity. Given the reduced proneoplastic potential of NN versus NRG, NN has translational potential for cardioprotection in patients with cancer receiving anthracyclines.

Keywords: anthracyclines; protein engineering; translational cancer chemotherapy protocols.

Figure 1

Design of covalently linked bivalent neuregulin-1β (NN) and validation of its reduced pro-neoplastic potential compared to neuregulin-1β (NRG). (A) NN was produced via covalent linkage of two NRG domains separated by a flexible, protease-resistant spacer. Purity was indicated by Coomassie-stained gel. (B) Schematic of expected differential ErbB receptor complexation induced by NRG compared with NN in doxorubicin-sensitive breast cancer cells expressing ErbB2 and ErbB3. NRG is predicted to predominantly promote ErbB2/3 heterotypic interactions whereas NN is predicted to induce increased ErbB3 homotypic interactions. (C) Immunoblot analysis of ErbB2, ErbB3, Akt, and ERK1/2 phosphorylation in lysates from doxorubicin-sensitive human mammary ductal carcinoma cell line MDA-MB-175VII stimulated for 15min by NRG or NN. GAPDH expression was measured as a loading control. Data are representative of 3 independent experiments. (D) Cell number was assessed by measurement of DNA content via fluorescence (CyQUANT assay, ex = 485nm, em = 530nm) following stimulation of MDA-MB-175VII cells by control media (indicated by dashed line) or the indicated doses of NRG (left of slash) or NN (right of slash) (n=6). P values were calculated using a Wilcoxon rank-sum test. RFU = relative fluorescence units.

Figure 2

Bivalent neuregulin-1β (NN) stimulates ErbB receptor phosphorylation on cardiomyocytes. (A) Schematic of expected differential ErbB receptor complexation induced by neuregulin-1β (NRG) compared with NN on cardiomyocytes. NRG is predicted to predominantly promote formation of ErbB2/4 heterotypic interactions whereas NN is predicted to induce increased ErbB4 homotypic interactions. (B) Phospho-receptor tyrosine kinase array analysis of neonatal rat cardiomyocytes stimulated for 15min by 10nM NRG or NN or serum-free media (Vehicle). Data are representative of 3 independent experiments. Legend indicates relative locations of positive control (+), negative control (-), phospho-ErbB2 (pErbB2) and phospho-ErbB4 (pErbB4) spots on array membrane.

Figure 3

Bivalent neuregulin-1β (NN) induces intracellular signaling similar to NRG and phosphorylates known mediators of cardioprotection. (A) Expression data from phospho-kinase array of neonatal rat cardiomyocytes (NRCM) stimulated for 15min by 50nM NRG or 25nM NN, Vehicle = serum-free media (n=3). (B) Pathway analysis of data from (A) demonstrates similar protein phosphorylation patterns induced by NN and NRG. Data were thresholded to y-axis=0.20 and displayed in order of relative expression in Vehicle group from low to high as indicated. (C) Expression data from phospho-kinase array of neonatal rat cardiomyocytes (NRCM) stimulated for 24h by 50nM NRG or 25nM NN concurrently exposed to 1μM doxorubicin (DOXO), Vehicle = serum-free media, (n=3). (D) Pathway analysis of data from (C) demonstrates differential protein phosphorylation stimulated by DOXO compared to Vehicle. NN + DOXO and NRG + DOXO induced similar patterns that indicate significant regulation of numerous factors from their DOXO-induced state towards the Vehicle state, indicative of cardioprotective signaling. Data were thresholded to y-axis=0.75 and displayed in order of relative expression in Vehicle group from low to high as indicated.

Figure 4

Bivalent neuregulin-1β (NN) has similar cardioprotective qualities to neuregulin-1β (NRG) in vitro. (A) Neonatal rat cardiomyocytes (NRCM) were exposed to 1μM doxorubicin (DOXO) and were stimulated with the indicated concentrations of NRG or NN in serum-free conditions for 24h. The control condition was NRCM in serum-free media (Vehicle). Cell viability was quantified via fluorescence (CyQUANT assay, ex = 485nm, em = 530nm) (n=8). (B) Presence of cleaved caspase-3/7 was assessed in NRCM after exposure to 1μM DOXO and stimulation with NRG or NN for 24h via fluorescence (SensoLyte assay, ex = 354nm, em = 442nm) (n=8). The control condition was NRCM in serum-free media (Vehicle). For A and B, P values were determined using a Wilcoxon rank-sum test with a Bonferroni correction following an initial Kruskal-Wallis test to determine statistical significance; data are representative of at least 2 independent experiments; RFU = relative fluorescence units. (C) Immunoblot analysis of γ-H2A.X in NRCM stimulated with NRG (100nM) or NN (50nM) concurrently exposed to 1μM DOXO for 24h. NRCM were incubated with treatments for 24h prior to DOXO addition (48h total treatment time). (D) Immunoblot analysis of γ-H2A.X in human induced pluripotent stem cell-derived cardiomyocytes (iPSC-CM) stimulated with NRG (100nM) or NN (50nM) concurrently exposed to 10μM DOXO for 24h. iPSC-CM were incubated with treatments for 24h prior to DOXO addition (48h total treatment time). For C and D, GAPDH expression was assessed as a protein loading control, data are representative of 3 independent experiments, and the control condition was NRCM in serum-free media (Vehicle).

Figure 5

Bivalent neuregulin-1β (NN) does not interfere with doxorubicin (DOXO) activity on anthracycline-sensitive cancer cells. (A) Mammary adenocarcinoma line SK-BR-3 cells were concurrently exposed to DOXO at the indicated concentrations and stimulated with 50nM NRG or 25nM NN. Cell viability relative to control (serum-free media; 100%) was quantified via fluorescence (CyQUANT assay, ex = 485nm, em = 530nm) (n=5, data are representative of 3 independent experiments). IC₅₀ values were calculated based on the best-fit line generated with GraphPad Prism software. (B) Identical experimental conditions were used with HeLa cells. (C) HeLa cells were exposed to serum-free media (Vehicle) with or without 1μM DOXO for 24h with concurrent stimulation by 50nM NRG or 25nM NN as indicated and immunoblot analysis of γ-H2A.X was performed on lysates. GAPDH expression was assessed as a protein loading control. Data are representative of 3 independent experiments.

Figure 6

Bivalent neuregulin-1β (NN) confers functional protection from doxorubicin (DOXO) toxicity in vivo. (A) Acute cardiomyopathy was induced with a single 20mg/kg intraperitoneal (ip) injection of DOXO. Mice were treated for 7d with Vehicle (0.2% BSA in PBS), neuregulin-1β (NRG), or NN at 100μg/kg ip per day, with 3 injections occurring prior to DOXO administration. Mice given the same NRG or NN treatments without DOXO injection were used as controls and are shown to the right of the vertical dashed line. Fractional shortening (FS) as assessed by blinded echocardiography at 7d is shown. The horizontal dotted line indicates the average FS of all animals at baseline echocardiography performed 2d prior to initial injection. P values were determined using a Wilcoxon rank-sum test with a Bonferroni correction (α=0.0125) following an initial Kruskal-Wallis test to determine statistical significance; no P values in this panel meet the criteria for statistical significance. (B) Immunoblot analysis of γ-H2A.X in cross-sectional heart lysates of study animals. GAPDH expression was assessed as a protein loading control. (C) Chronic DOXO-induced cardiomyopathy was induced with weekly serial 4mg/kg intraperitoneal (ip) injections of DOXO as shown. Beginning on the day of the initial DOXO injection for each 4-week series, mice were injected daily for 7d with the vehicle solution, 0.2% BSA in PBS (DOXO), neuregulin-1β (NRG), or NN at 50μg/kg ip per day. Mice that were not injected with DOXO (Vehicle) were used as controls. Fractional shortening (FS) as assessed by blinded echocardiography 2wks prior to initial DOXO injection and 6, 12, 22 and 24wks following initial DOXO injection, respectively, is shown. Only endpoint values (24wks) were analyzed; P values were determined using a Wilcoxon rank-sum test with a Bonferroni correction (α=0.0125) following an initial Kruskal-Wallis test to determine statistical significance. For this panel, P values less than 0.0125 are considered statistically significant.