Automated in vivo compound screening with zebrafish and the discovery and validation of PD 81,723 as a novel angiogenesis inhibitor

Abstract

Angiogenesis is a critical process in tumor progression. Inhibition of angiogenesis by blocking VEGF signaling can impair existing tumor vessels and halt tumor progression. However, the benefits are transient, and most patients who initially respond to these therapies develop resistance. Accordingly, there is a need for new anti-angiogenesis therapeutics to delay the processes of resistance or eliminate the resistive effects entirely. This manuscript presents the results of a screen of the National Institutes of Health Clinical Collections Libraries I & II (NIHCCLI&II) for novel angiogenesis inhibitors. The 727 compounds of the NIHCCLI&II library were screened with a high-throughput drug discovery platform (HTP) developed previously with angiogenesis-specific protocols utilizing zebrafish. The screen resulted in 14 hit compounds that were subsequently narrowed down to one, with PD 81,723 chosen as the lead compound. PD 81,723 was validated as an inhibitor of angiogenesis in vivo in zebrafish and in vitro in human umbilical vein endothelial cells (HUVECs). Zebrafish exposed to PD 81,723 exhibited several signs of a diminished endothelial network due to the inhibition of angiogenesis. Immunochemical analysis did not reveal any significant apoptotic or mitotic activity in the zebrafish. Assays with cultured HUVECs elucidated the ability of PD 81,723 to inhibit capillary tube formation, migration, and proliferation of endothelial cells. In addition, PD 81,723 did not induce apoptosis while significantly down regulating p21, AKT, VEGFR-2, p-VEGFR-2, eNOS, and p-eNOS, with no notable change in endogenous VEGF-A in cultured HUVECs.

Figure 1

Top 5 hit compounds from a screen of the NIHCCLI&II for angiogenesis inhibitors. (A) Representative images of Tg(kdrl:EGFP) zebrafish treated with the top 5 hit compounds. Pixels containing endothelial cells are identified and enhanced with the HTP pixel count readout protocol as previously described . Fish are at 4 dpf. Scale bar is 1 mm. (B) Pixel count readout of Tg(kdrl:EGFP) zebrafish treated with various concentrations of the top 5 compounds compared to a negative control. PD 81,723 (4, 8, and 16 µM) and raclopride (1, 4, and 16 µM) yielded significantly (** = P < 0.01 and * = P < 0.05 vs. control) lower pixel counts when compared to the control group. The experiments were repeated in triplicate with an N = 16 for each condition in each experiment. The error bars represent the standard deviation of the means from the three experiments. (C) Survival curves. None of the compounds seemed to have a major impact on mortality. The error bars represent the standard deviation of the survival percentage from the three experiments.

Figure 2

Fluorescent images comparing the vascular phenotype of Tg(kdrl:EGFP) zebrafish treated with 0.05% DMSO, 64 µM PD 81,723, and 8 µM I3M. (A) The yellow arrow indicates the area next to the anus where blood islands have formed in the drug dosed fish instead of a mesh-like network that evolves into two separate vessels (caudal artery and caudal vein). The red arrows show a diminished EGFP signal coming from the ISVs. The green arrows are indicating edema surrounding an enlarged heart. The blue arrows are showing a diminished EGFP signal from the cranial vasculature. The purple arrow is pointing to the region where the sub-intestinal vessels (SIVs) have not formed in the fish dosed with a PD 81,723 concentration of 64 µM. Scale bar is 250 µm. (B) The 4 dpf images from A) with the SIV regions of the control and PD 81,723 dosed fish magnified on the right. The SIVs do not appear to be present in the PD 81,723 dosed fish. Scale bar for the images on the left is 250 µm. Scale bars for the images on the right are 100 µm.

Figure 3

PD 81,723 inhibits the development of SIVs, the VTA, and the PAV. (A) Confocal images comparing the vascular phenotype of Tg(kdrl:EGFP) zebrafish at 4 dpf treated with 0.05% DMSO and 64 µM PD 81,723 at the cranial end and focusing on the area surrounding the SIVs. The red arrows indicate a missing vertebral artery (VTA) and the blue arrow indicates a missing parachordal vessel (PAV). The yellow arrow indicates a reduction in the vasculature where the ISVs meet the posterior cardinal vein (PCV). The purple arrow is pointing to the region where the SIVs are malformed. Scale bars are 100 µm. (B) SIV counts at 4 dpf in zebrafish dosed with 0.05% DMSO, 64 µM PD 81,723, or 8 µM I3M. There was a significant decrease (* = P < 0.05 vs. control) in the number of vessels sprouting from the SIVs basket in the fish dosed with 64 µM of PD 81,723 when compared to the control. The experiment was repeated in triplicate with an N = 3 for each condition in each experiment. The error bars represent the standard deviation of the SIV counts from the three experiments. (C) Table showing that only 22% of the fish dosed with PD 81,723 had complete VTA and PAV, compared to 89% in the control, across all three trials.

Figure 4

Active Caspase-3 staining in SIVs and ocular area of 4 dpf zebrafish. (A) Representative images of SIVs and the ocular area of 4 dpf zebrafish treated with 0.05% DMSO, 64 µM PD 81,723, and 8 μM I3M at 10–11 hpf. Scale bars are 50 µm. (B) Graphical representations of the results from three different groups for the caspase-3 stains that were repeated in triplicate with an N = 3 for each group. The error bars represent the standard deviation of the means from the three experiments. The number of endothelial nuclei was significantly reduced in both regions of interest in all instances of this experiment (* = P < 0.05 for SIVs area and ** = P < 0.01 for eye area vs. control). There was no significant apoptotic activity in the tissue surrounding the ocular vasculature or the SIVs in PD 81,723 fish, when compared to the DMSO group.

Figure 5

Phosphorylated H3 staining in SIVs and ocular area of 4 dpf zebrafish. (A) Representative images of SIVs and the ocular area of 4 dpf zebrafish treated with 0.05% DMSO, 64 µM PD 81,723, and 8 μM I3M at 10–11 hpf. Scale bars are 50 µm. (B) Graphical representations of the results from three different groups for the phosphorylated H3 stains that were repeated in triplicate with an N = 3 for each group. The error bars represent the standard deviation of the means from the three experiments. The number of endothelial nuclei was significantly reduced in both regions of interest in all instances of this experiment (** = P < 0.01 vs. control). A significant reduction in mitotic activity in the tissue surrounding the ocular vasculature (** = P < 0.01 vs. control) and the SIVs (* = P < 0.05 vs. control) in PD 81,723 fish was observed, when compared to the DMSO group.

Figure 6

PD 81,723 inhibits angiogenic potential, migration, and cell proliferation. (A) PD 81,723 reduced endothelial branching and outgrowth in a matrigel assay. Representative images of HUVECs in matrigel treated with 0.05% DMSO and 50 µM PD 81,723 at 9 h after seeding with a graphical representation of the results from the assay. The error bars represent the standard deviation of the means from the six wells for each condition. The total capillary tube length was significantly reduced in the presence of PD 81,723 (*** = P < 0.001). Scale bars are 100 µm. (B) PD 81,723 inhibits wound closure in HUVECs. The cells were treated with 0.05% DMSO or 50 µM PD 81,723 in either complete endothelial growth media or endothelial growth media without VEGF-A. Five sites near the center of each gap were imaged every 30 min for 24 h. The wounds treated with PD 81,723 in complete media and DMSO in media without VEGF-A took significantly longer to close (* = P < 0.05) than the wounds treated with DMSO in complete media. The wounds treated with PD 81,723 in VEGF-A deficient media did not close within a 24 h period. The error bars represent the standard deviation of the means from three wells. (C) PD 81,723 inhibits proliferation in HUVECs. A bromodeoxyuridine (BrdU) incorporation cell proliferation kit was used to look at cell proliferation. The cells were treated with 0.05% DMSO or 50 µM PD 81,723 in either complete endothelial growth media or endothelial growth media without VEGF-A, with five wells per condition in a 96-well plate. There was no notable difference between DMSO groups, while PD 81,723 significantly (** = P < 0.01 vs. complete media control) reduced cell proliferation in both cases. This experiment was repeated in triplicate. The error bars represent the standard deviation of the means from the three experiments.

Figure 7

Annexin V/propidium iodide (PI) flow cytometry apoptosis assay with HUVECs. (A) Representative scatter plots divided into quadrants showing the distribution of HUVECs marked with annexin V, PI, and double stained with annexin V and PI. The apoptotic activity is quantified by the amount of cells in Q2 and Q2-1. The top plots are controls and the bottom plots are for cells treated with PD 81,723. The plots on the left are for cells grown in complete media and the plots on the right are for cells grown in VEGF-A deficient media. (B) Bar graph exhibiting the amount of apoptotic activity for each condition. There was no significant change in the levels of apoptotic events in PD 81,723 treated cells, independent of the addition of VEGF-A to the growth media. This experiment was repeated three times and the error bars represent the standard deviation from three experiments.

Figure 8

PD 81,723 significantly reduced the expression of p21, AKT, VEGFR-2, and eNOS in HUVECs. HUVECs at 60–70% confluency in 6-well plates were treated with 0.05% DMSO or 50 µM PD 81,723 in either complete endothelial growth media or endothelial growth media deficient in VEGF-A for 24 h. The cells were harvested and protein was extracted 24 h post treatment, with the experiment repeated in triplicate. (A) PD 81,723 treated cells expressed significantly (*** = P < 0.001 vs. complete media control) lower levels of p21, independent of the addition of VEGF-A in the growth media. (B) There was no statistically significant change in endogenous VEGF-A mean protein levels in PD 81,723 treated cells. (C) AKT was decreased in the PD 81,723 treated cells; with a statistically significant (* = P < 0.05 vs. complete media control) decrease in PD 81,723 treated cells in complete media. (D) p-AKT appeared unchanged from observing the blots, but when comparing the ratio of p-AKT/AKT there was an increase in PD 81,723 treated cells; with a statistically significant (* = P < 0.05 vs. complete media control) increase in PD 81,723 treated cells in VEGF-A deficient media. (E) VEGFR-2 and p-VEGFR-2 did not present any detectable bands in PD 81,723 treated cells. (F) The levels of eNOS, and p-eNOS were not enough to present detectable bands in PD 81,723 treated cells. The band intensities for p21, VEGF-A, and AKT were divided by those of the corresponding loading control (β-actin) and p-AKT was divided by the total AKT result. The data was normalized to the results of the 0.05% DMSO treated samples in complete endothelial growth media. The error bars represent the standard deviation from three experiments. The blots for VEGFR2, p-VEGFR2, eNOS, and p-eNOS showed a definitive near absence of protein in PD 81,723 treated cells which did not require mean intensity quantification. Original blots are presented in Supplementary Fig. S4.