Targeting activin receptor-like kinase 1 inhibits angiogenesis and tumorigenesis through a mechanism of action complementary to anti-VEGF therapies

Abstract

Genetic and molecular studies suggest that activin receptor-like kinase 1 (ALK1) plays an important role in vascular development, remodeling, and pathologic angiogenesis. Here we investigated the role of ALK1 in angiogenesis in the context of common proangiogenic factors [PAF; VEGF-A and basic fibroblast growth factor (bFGF)]. We observed that PAFs stimulated ALK1-mediated signaling, including Smad1/5/8 phosphorylation, nuclear translocation and Id-1 expression, cell spreading, and tubulogenesis of endothelial cells (EC). An antibody specifically targeting ALK1 (anti-ALK1) markedly inhibited these events. In mice, anti-ALK1 suppressed Matrigel angiogenesis stimulated by PAFs and inhibited xenograft tumor growth by attenuating both blood and lymphatic vessel angiogenesis. In a human melanoma model with acquired resistance to a VEGF receptor kinase inhibitor, anti-ALK1 also delayed tumor growth and disturbed vascular normalization associated with VEGF receptor inhibition. In a human/mouse chimera tumor model, targeting human ALK1 decreased human vessel density and improved antitumor efficacy when combined with bevacizumab (anti-VEGF). Antiangiogenesis and antitumor efficacy were associated with disrupted co-localization of ECs with desmin(+) perivascular cells, and reduction of blood flow primarily in large/mature vessels as assessed by contrast-enhanced ultrasonography. Thus, ALK1 may play a role in stabilizing angiogenic vessels and contribute to resistance to anti-VEGF therapies. Given our observation of its expression in the vasculature of many human tumor types and in circulating ECs from patients with advanced cancers, ALK1 blockade may represent an effective therapeutic opportunity complementary to the current antiangiogenic modalities in the clinic.

Conflict of interest statement

Disclosure of Potential Conflicts of Interest

DDH, EC, LZ, PL, TAS, BS, JW, JHC, ZF, SB, GFC, WJL, CGS, and DRS are full-time

Pfizer Inc. employees and own Pfizer Inc. stock.

PM, KDW, RS, and AE declare no competing interests.

GW, KA, XJ, and JL disclose work performed as employees of Pfizer Inc.

KWF has received research funding from Pfizer Inc.

WF has received research funding, and fees for advisory board meetings and invited speeches from Pfizer Inc.

FB has received compensation from Pfizer Inc. for a consultant/advisory board role.

©2011 AACR.

Figures

Figure 1. Modulation of ALK1 signaling PAFs and Anti-huALK1 in HUVECs

A, top: cells were starved overnight and treated with Anti-huALK1 for 2–4 hours followed by a 45-minute stimulation with 2.5% FBS. pSmads were detected by Western blotting. β-Actin was used as loading control (n = 2). Bottom: cells were treated as abov and the Id-1 transcript in the presence of 0.3× of EGM®-2 BulletKit® (Lonza; 30-minute stimulation) was measured by quantitative RT-PCR. Shown are fitted curves of dose-dependent inhibition of Id-1 by Anti-huALK1 from three experiments. B, cells were starved for 2–4 hours and incubated with Anti-huALK1 (200 nM) for an additional 1–2 hours before a 45-minute stimulation with VEGF (10 ng/mL), bFGF (30 mg/mL), or TGFβ (1 ng/mL). pSmads and Id-1 were detected by Western blotting. GAPDH (glyceraldehyde 3-phosphate dehydrogenase) was probed as loading control (n = 3). C, cells were treated as in B, permeabilized, and stained for pSmads (green). Nuclei were counterstained with DAPI (blue) to indicate the presence of cells after Anti-huALK1 (n = 2). D, cells and the experiment details were the same as in B, except that VEGF and TGFβ were used as stimuli, and sunitinib (50 nM) and ALK1/ECD (200 nM) were also used as inhibitors in the assay (n = 3).

Figure 2. HUVEC function can be inhibited by Anti-huALK1

A, Anti-huALK1 inhibited PAF-induced EC attachment and spreading. HUVECs (in triplicates) in a 96-well plate equipped with electro-sensor probes (RT-CESTM) were incubated with PAFs ± Anti-huALK1 (200 nM) in basal media containing 1% FBS for 2 hours. Cellular phenotypical changes were recorded in real time and reported as an arbitrary unit (activation index). Data are average ± SD. B, HUVECs were co-cultured with MRC-5 fibroblast cells. Shown are representative photoimages of the tubule network in the presence of VEGF (20 ng/mL), bFGF (30 ng/mL), or TGFβ (1 ng/mL) ± Anti-huALK1 (2, 20, and 200 nM). A representative photoimage of the tubule network in basal media absent of any exogenous growth factors is also shown (EBM-2 + 5% FBS). C, tubule area was quantified as described in the Materials and Methods (8 fields/well; 2 wells for each concentration). Data under each PAF were normalized to that of PAF alone (control) and expressed as average ± SEM.

Figure 3. Anti-muALK1 inhibited angiogenesis and tumor growth in mice

A, PAF-enriched AngioReactorsTM were implanted in mice and murine angiogenesis was assessed 10 days later (n = 6–8/group). Anti-muALK1, Amu-VEGF, or A1D16 (a non-specific mouse IgG1 control) was administered either systemically (4 days post implant) or pre-mixed in the devices (30 μg/mL each). *, P < 0.001 versus either of the control groups. Results were reproducible in two other experiments. B, Anti-muALK1 inhibited the growth of mammary fatpad-implanted MDA-MB-231 xenograft tumors in mice (n = 2 studies). Top left: colocalization (yellow) of muALK1 (red) with muCD31 (green) in the tumor vasculature. Bottom left: Anti-muALK1 (QW and twice weekly [BIW]) and Amu-VEGF produced 59%, 68%, and 81% TGI compared with the control group, respectively (n = 10/group). Top right: photoimages of immunofluorescent staining of murine CD31+ blood vessels (green) and LYVE-1+ lymphatic vessels (red) from the xenograft tumors. Bottom right: quantification of the positive staining areas of the above two markers. †, P < 0.05 and ‡, P < 0.01 versus control. C, inset: representative photoimages of immunofluorescent staining of huCD31+ and muCD31+ vessels in tumors of the control and Anti-huALK1 (10 mg/kg) groups. Single injection of Anti-huALK1 (1–50 mg/kg) dose-dependently inhibited huCD31 in chimera tumors. Bars: group mean ± SEM (6–10 animals/group, 3–5 hotspots/tumor section, two independent viewers). †, P < 0.05 and ‡, P < 0.01 compared with IgG2 isotype control. D, Anti-huALK1 inhibited huCD31+ vessel area to a similar degree as that by sunitinib and bevacizumab in the chimera model (n = 3). JBS5 (anti-human integrin α5β1) was used as a reference mAb. Bars: group mean ± SEM (5–6 animals/group; 3–5 hotspots/tumor section; two independent viewers). †, P < 0.05 versus control.

Figure 4. ALK1 inhibition disrupted interaction between ECs and PCs in the M24met/R model (A) and the chimera tumor model (B D)

A, (i) ELISA (enzyme-linked immunosorbent assay) showed increased human VEGF expression in PF-00337210-treated compared with untreated M24met/R tumors; vascular staining (immunofluorescent) was performed and showed co-localization of ALK1 and CD31 (ii) and muCD31 and desmin (iii, iv). (v) M24met/R tumors lacked response to PF-00337210 (▲) compared with untreated tumors (●); addition of Anti-muALK1 (10 mg/kg, QW) to PF-00337210 (▼) delayed tumor growth by 57% compared with PF-00337210 alone (*, P < 0.01; 10 animals/group). Vessels in the combination group showed increased sprouting (arrows) and tortuosity, and reduced CD31+/desmin+ co-staining (vi, vii). Rx, treatment. B, chimera tumors were treated with Anti-huALK1, bevacizumab, or a combination of the two agents (treatment started when average TV was 50 mm3 and lasted for 10 days). Data are group average tumor weight ± SEM. †, P < 0.05 compared with control (5–7 animals/group). C, quantification of huCD31 and muCD31 in tumors from B. Anti-huALK1 had no effect on muMVD due to lack of cross-reactivity with muALK1. Data are group average ± SEM (n = 5–7/group; 3–5 hotspots/tumor). *, P < 0.01 compared with control; ‡, P < 0.01 compared with bevacizumab. D, representative images of huCD31 (green) and desmin (red) double staining of chimera tumors from B. Panels iv, viii, xii, and xvi are Z-stack three-dimensional images. Arrowheads: desmin+/huCD31− vessels; §, small vessels were mostly huCD31+/desmin− arrows: huCD31+ vessels devoid of desmin staining. Details are discussed in the Results section.

Figure 5. Effect of Anti-huALK1 on vascular BF and perfusion

A, CE-US images of chimera tumors at baseline and 60 hours after treatment. The color bar represents time required for T20% (0–10 second). The formation of fast-flow vessels (T20%<1.5 seconds; yellow) is evident in the control, but not the Anti-huALK1-treated tumor (arrowheads). Fast flow in the kidney cortex is shown below the tumor (arrow). B and C, quantitation of fast-flow (B) and slow-flow blood vessels (C; T20% in 3–10.5 seconds) in the control and Anti-huALK1 groups on days 1 and 4 of treatment. *, P = 0.011 (n = 4–5/group).

Figure 6. Expression of ALK1 on human tissue specimen and assessment of ALK1+/CECs in clinical samples

A, rank order of percentage of patients with vascular ALK1 expression scores of 1+–3+ (by IHC with TMA samples) in the top 20 common cancer types. Numbers of tumor samples assessed are in brackets. *, Occasional presenceof ALK1 in prostate tumors. B, representative photoimages of ALK1 vascular staining (brown) in the carcinoma tumors and normal tissues of colon (i, ii), lung (iii, iv), and pancreas (v, vi). Normal tissues are generally devoid of specific ALK1 staining compared with the corresponding tumors. C, percent of viable ALK1+/CECs in evaluated cancer samples was significantly greater than those from healthy volunteers. Viable ALK1+/CEC counts were calculated from total ALK1+/CECs and percentage of total apoptotic cells (Supplementary Table C). †, P < 0.05 and ‡, P < 0.001 compared with healthy volunteers (Mann Whitney U test).