Context-dependent role of angiopoietin-1 inhibition in the suppression of angiogenesis and tumor growth: implications for AMG 386, an angiopoietin-1/2-neutralizing peptibody

Abstract

AMG 386 is an investigational first-in-class peptide-Fc fusion protein (peptibody) that inhibits angiogenesis by preventing the interaction of angiopoietin-1 (Ang1) and Ang2 with their receptor, Tie2. Although the therapeutic value of blocking Ang2 has been shown in several models of tumorigenesis and angiogenesis, the potential benefit of Ang1 antagonism is less clear. To investigate the consequences of Ang1 neutralization, we have developed potent and selective peptibodies that inhibit the interaction between Ang1 and its receptor, Tie2. Although selective Ang1 antagonism has no independent effect in models of angiogenesis-associated diseases (cancer and diabetic retinopathy), it induces ovarian atrophy in normal juvenile rats and inhibits ovarian follicular angiogenesis in a hormone-induced ovulation model. Surprisingly, the activity of Ang1 inhibitors seems to be unmasked in some disease models when combined with Ang2 inhibitors, even in the context of concurrent vascular endothelial growth factor inhibition. Dual inhibition of Ang1 and Ang2 using AMG 386 or a combination of Ang1- and Ang2-selective peptibodies cooperatively suppresses tumor xenograft growth and ovarian follicular angiogenesis; however, Ang1 inhibition fails to augment the suppressive effect of Ang2 inhibition on tumor endothelial cell proliferation, corneal angiogenesis, and oxygen-induced retinal angiogenesis. In no case was Ang1 inhibition shown to (a) confer superior activity to Ang2 inhibition or dual Ang1/2 inhibition or (b) antagonize the efficacy of Ang2 inhibition. These results imply that Ang1 plays a context-dependent role in promoting postnatal angiogenesis and that dual Ang1/2 inhibition is superior to selective Ang2 inhibition for suppression of angiogenesis in some postnatal settings.

Figure 1

The effect of combined Ang1 and Ang2 inhibition on the growth of Colo205 tumor xenografts. A, mice implanted with Colo205 cells were treated with either AMG 386 (5.6 mg/kg twice per week), Fc control (5.2 mg/kg), L1-7(N) (2.0 mg/kg), or mL4-3 (3.2 mg/kg) daily; or the combination of L1-7(N) and mL4-3 (at the same dosing regimens used in the single-agent groups). Where necessary, Fc control protein was added to match the total amount of protein delivered in the combination group (5.2 mg/kg). Viable tumor burden (B) and total blood vessel area (C) were determined following 7 days of peptibody administration. Data are mean values ± SE. *, P < 0.05 versus L1-7(N), mL4-3, and Fc; §, P < 0.05 versus Fc; †, P < 0.01 versus Fc; ‡, P = 0.02 versus mL4-3. D, confocal microscopic images of tumor endothelial cells (CD31; green) and basement membrane (type IV collagen; red). After 26 days of treatment with Fc, mL4-3, or L1-7(N), tumor vessels were accompanied by basement membrane with varying abnormalities (arrowheads), but the vessels are continuous. By comparison, a discontinuous (regressing) vessel in a tumor treated with both mL4-3 and L1-7 (N) was observed (right). The region of vessel regression, which lacks endothelial cells, is spanned by an empty basement membrane sleeve (arrow). Scale bar, 15 µm.

Figure 2

The effect of combined Ang1 and Ang2 inhibition in the presence of VEGF pathway antagonism. A, mice implanted with Colo205 cells were treated with human IgG1 (0.4 mg/kg, i.p.), bevacizumab (0.4 mg/kg, i.p.), AMG 386 (0.11 mg/kg, s.c.), or the combination of bevacizumab and AMG 386 (at the same dose and schedule as the single-agent groups) twice per week. B, viable tumor burden was determined at the end of the experiment (n = 10). Bev, bevacizumab. C, Colo205 tumor-bearing mice were treated with IgG1 (0.4 mg/kg, i.p.; n = 10) or bevacizumab (0.4 mg/kg, i.p.) in combination with either AMG 386 (2.8 mg/kg, s.c.) or L1-7(N) (2.8 mg/kg, s.c.) twice per week (n = 25 per group). Fc or IgG1 control proteins were added to match the total amount of protein delivered in the combination group in both experiments. D, viable tumor burden (n = 23–25). Data are mean values ± SE. *, P < 0.0001 versus bevacizumab or AMG 386; §, P < 0.05 versus bevacizumab or AMG 386; †, P ≤ 0.05 versus IgG1; ‡, P < 0.05 versus bevacizumab + L1-7(N) at terminal measurement.

Figure 3

The effect of Ang1 and Ang2 antagonism on tumor endothelial cell proliferation, corneal angiogenesis, and retinal angiogenesis. A, BrdUrd incorporation in endothelial cells. Colo205 tumor-bearing mice were treated with Fc, AMG 386, L1-7(N), mL4-3, or L1-7(N) plus mL4-3. Bar, mean endothelial:total mouse cell BrdUrd ratios. Data are mean values ± SE. *, P < 0.05 versus Fc. The effect of inhibition of Ang1 and Ang2 on bFGF-induced (B) and VEGF-induced (C) corneal angiogenesis following treatment with Fc, L1-7(N), mL4-3, and L1-7(N) plus mL4-3. Data are mean values ± SE. †, P < 0.002 versus Fc + bFGF; ‡, P < 0.01 versus Fc + VEGF. D, inhibition of Ang2 prevents oxygen-induced neovascularization in the mouse retina. Pups were treated with Fc, L1-7(N), mL4-3, or L1-7(N) plus mL4-3. Data are mean values ± SE. §, P < 0.0001 versus Fc.

Figure 4

Ang1 and Ang2 inhibitors cooperatively suppress ovarian follicular angiogenesis. Fc, mL4-3, L1-7(N), or an mL4-3/L1-7(N) combination were administered to superovulated mice (two independent experiments). Data are mean values ± SE. *, P = 0.005 comparing mL4-3/L1-7(N) combination versus either single-agent alone; #, P < 0.05 versus Fc.

Figure 5

Mechanistic consequences of Ang2 and Ang1 inhibition in the Colo205 tumor model.