Dehydro-α-lapachone, a plant product with antivascular activity

Abstract

Antivascular agents have become a standard of treatment for many malignancies. However, most of them target the VEGF pathway and lead to refractoriness. To improve the diversity of options for antivascular therapy, we applied a high-throughput screen for small molecules targeting cell adhesion. We then assayed the resulting antiadhesion hits in a transgenic zebrafish line with endothelial expression of EGFP (Tg(fli1:EGFP)^y1) to identify nontoxic molecules with antivascular activity selective to neovasculature. This screen identified dehydro-α-lapachone (DAL), a natural plant product. We found that DAL inhibits vessel regeneration, interferes with vessel anastomosis, and limits plexus formation in zebrafish. Furthermore, DAL induces vascular pruning and growth delay in orthotopic mammary tumors in mice. We show that DAL targets cell adhesion by promoting ubiquitination of the Rho-GTPase Rac1, which is frequently up-regulated in many different cancers.

Regulation of blood vessel growth has proven to be an important strategy in the treatment of cancer and may provide new targets for the treatment of inflammatory disorders, asthma, obesity, diabetes, multiple sclerosis, endometriosis, and bacterial infections (1, 2). Antiangiogenic agents, most of which target VEGF or its receptors, are emerging as standard therapies for several major human cancers (3–5). Unfortunately, antiangiogenic therapy leads to modest efficacy, inherent or acquired resistance, and rare but life-threatening toxicity (6, 7). Considering these limitations, the development of antivascular agents that target pathways other than VEGF is urgently needed.

Cell adhesion pathways present attractive targets for antivascular agents in cancer therapy. Endothelial cell (EC) adhesion is crucial for blood vessel function (8, 9). Blood vessels in tumors are abnormal, featuring unusual leakiness, high tortuosity, and inefficient network structure (10). The immature ECs of this abnormal neovasculature in tumors have weak cell-cell junctions (10–13)—likely making them particularly sensitive to antiadhesion therapies. Candidate antiadhesion agents for antivascular therapy have been identified and have shown promising results in preclinical studies (14). Unfortunately, these early attempts at development of antiadhesion agents for antivascular therapy were met with failure stemming from issues with toxicity. We hypothesized that a nontoxic compound targeting cell adhesion would produce a safe antivascular effect specific to neovasculature. To this end, we devised a high-throughput cell adhesion screen to identify candidate nontoxic antiadhesion compounds with antivascular activity in tumors.

Results and Discussion

Screening Yields a Natural Product—DAL.

We developed a unique screening strategy to identify potential agents that target adhesion of ECs or cancer cells to their substrate. To this end, we began by screening 50,000 compounds in a high-throughput manner (Fig. 1A). Our initial screening step quantified the number of cells remaining attached to their wells after incubation with each compound and subsequent washing steps. Only 86 compounds affected cellular adhesion in our assay. Cell adhesion adaptor proteins have a domain for binding actin filaments (15, 16), thus adhesion molecules are directly linked to the actin cytoskeleton. We reasoned that the agents that affect cell adhesion may be monitored through remodeling of actin filaments. As our second screening step, we explored the effect of the selected compounds on actin assembly and redistribution by fixing and staining-treated cells with phalloidin. We observed changes in actin assembly and cell shape after treatment with 12 compounds. We next prioritized this set for compounds that were nontoxic to normal cells, cancer cells, and zebrafish. Of these 12 compounds, two were selected based on the in vitro and in vivo assessment. One of these compounds, dehydro-α-lapachone (DAL; Fig. 1B), showed structural similarities to β-lapachone, which has demonstrated antitumor and antitrypanosomal activities through DNA topoisomerase I inhibition and prevention of DNA repair (17, 18). We further tested DAL for toxicity in SCID mice, and found that mice are tolerant to i.p. doses of up to 100 mg/kg with no signs of toxicity. We therefore selected DAL—a natural product from the Tabebuia Avellaneda tree—for further study.

Fig. 1.

A high-throughput drug screen for the DAL. (A) Of the 50,000 compounds screened, we selected 86 to test in the secondary assays. We only selected compounds that made MDA-MB-231 cells more rounded, with less spreading than control cells, and interfered with actin regulation for further study. We further passed all these selected small molecules through a number of filtering toxicity assays to eliminate toxic compounds. To confirm the absence of toxicity in the cellular model for each candidate molecule, we repeated the experiments and reassayed at three different concentrations (1.67 M, 5 M, and 10 M). We then analyzed the nontoxic compounds in a zebrafish model. We used small molecules that did not alter zebrafish survival during embryonic development and adult life for further experiments. Finally, of 50,000 compounds, we selected two nontoxic compounds that demonstrated antivascular properties. One of these compounds, DAL, was chosen for further experiments. (B) The structure of DAL, a natural product. (C) DAL-treated cells lost their adhesion properties. (D) Untreated MDA-MB-231 control cells spread more rapidly on fibronectin-coated plates than cells treated with DAL. In contrast, cells treated with DAL were significantly smaller than control cells and had a rounded shape. (E) DAL is involved in actin cytoskeleton assembly regulation. For F-actin visualization, we plated MDA-MB-231 cells on fibronectin and treated them with each compound (5 μM) for 2 h. After fixing and permeabilizing these cultured cells, we subsequently stained them with phalloidin. The untreated control cells spread more rapidly on fibronectin-coated plates than those treated with DAL. In contrast, cells treated with DAL were significantly smaller than control cells and had a rounded shape. In the control cells, actin stress fibers were clearly visible and aligned with the long axis of the cells. Meanwhile, in cells treated with DAL, actin was accumulated close to the membrane. We used cytochalasin D as a positive control that inhibits actin assembly.

DAL Shows Antivascular Effects in Zebrafish Models.

To elucidate the potential impact of DAL on the process of vascular network formation, we next assessed its effects in zebrafish embryos at different stages of development. To visualize vessel defects in detail, we used transgenic fish expressing EGFP in endothelial cells (Tg(fli1:EGFP)^y1) (19, 20). This model expresses EGFP in blood vessel ECs throughout normal development and during fin regeneration (21). We first characterized the effects of DAL on normal vascular development in zebrafish embryos. In control embryos, developing vessels migrated from the lateral plate mesoderm to the midline, where they coalesced into a vascular cord. These endothelial clusters subsequently established the pattern of the dorsal aorta and posterior cardinal vein. Intersomitic vessels sprouted at designated branch sites in control embryos after dorsal aorta formation. In contrast, after treatment with DAL, the sprouting of intersomitic vessels at their designated branch sites failed to occur and the embryos developed unusual wave-like vessel structures (Fig. 2A). DAL treatment resulted in a profound reduction in the complexity of the arterial network. These findings demonstrate that DAL can interfere with the regulation of vascular formation and branching morphogenesis during development. Interestingly, DAL did not show any effects on the vasculature in adult fish, probably due to the stable junctions and postmitotic nature of the mature vasculature.

Fig. 2.

DAL impairs vessel development and regeneration in zebrafish. (A) Macroscopic view of 48 h after fertilization Tg(fli1:EGFP)^y1 zebrafish embryos, comparing control embryos and embryos treated with DAL (5 μM). Treated embryos failed to form vessel branches. (B and C) Regeneration experiments with zebrafish, depicting the vasculature of the caudal fin on 2 and 9 d after amputation in control and treated zebrafish. White lines indicate the amputation plane. (B) Vascular plexus formation, plexus remodeling, and late regenerative angiogenesis. (C) A higher magnification of the reconnected vessels. Nine days after amputation, the regenerating vasculature of each fin ray (three fin rays are shown) consists of a plexus with dense unstructured vessels extending distally from the amputation plane (white arrowheads). Regenerating blood vessels in the wild-type control (fli1:EGFP) form plexuses. Fish treated with DAL have defects in anastomosis and plexus formation. (Scale bars: 200 μm.)

To examine whether DAL has a similar antivascular effect on neovasculature in adult zebrafish, we induced neovascularization by amputating the caudal fin at the ≈50% proximal–distal level and imaged fin rays and vessels from each fish over time. In control zebrafish caudal fins, amputated blood vessels healed their ends by 24 h after amputation and then reconnected arteries and veins through anastomosis, with blood flow resuming at wound sites by 48 h after amputation (Fig. 2B). Meanwhile, regenerating vessels in fish treated with DAL had defects in anastomosis and plexus formation. All control fish developed normal blood vessels and formed anastomotic bridges at the amputation plane by 9 d after amputation; in contrast, we did not find such bridges in fish treated with DAL (Fig. 2C).

DAL Prunes Tumor Vasculature.

Because the zebrafish data indicated that DAL is a potential antivascular agent selective for neovasculature, we sought to determine the effects of DAL on tumor vasculature in mammals. To study these effects quantitatively, we conducted fluorescent angiographies in female SCID mice bearing orthotopic 4T1 mammary tumors in mammary fat pad windows via intravital multiphoton microscopy (Fig. 3A) (22–24). We treated these mice with 37.5 mg/kg DAL or saline daily for 5 d by i.p. injection. We found that DAL treatment decreases tumor vascular volume fractions—a measure of vascular density—compared with saline treatment (Fig. 3B; P = 0.002, day 4). Furthermore, DAL treatment lowers total tumor vascular length (normalized to tumor volume) versus saline treatment (Fig. 3C; P = 0.007, day 2; P = 0.02, day 4), whereas mean tumor vascular diameter remains the same (Fig. 3D). These data indicate that DAL reduces vascular density in tumors through vessel pruning.

Fig. 3.

DAL has antivascular effects in mouse models. (A) Representative intravital images of tumor vessels in mice bearing orthotopic 4T1 mammary tumors through a treatment course. We imaged the tumors with multiphoton microscopy on days 0, 2, and 4; daily treatments started on day 0 after the first images were collected. Images are 3D projections of tissue from 0 to 200 μm, with pseudocolor labeling by depth. (B–D) Measured vascular parameters in the mice over the course of treatment. Tumors in mice treated with DAL have lower vascular volume fractions than tumors in saline-treated mice at day 4 (P = 0.0017). Total tumor vascular length (normalized to tumor volume), but not mean tumor vascular diameter, is lower in DAL-treated mice than in saline-treated mice (P = 0.0069, day 2; P = 0.024, day 4), suggesting vessel pruning as the mechanism for decreased vascular volume fraction. n = 5 for the saline group, n = 4 for the DAL group.

DAL Inhibits Tumor Growth.

Considering that DAL has an antivascular effect in orthotopic mammary tumors in mice, we expected that it might have an antitumor effect. Pharmacokinetic studies for DAL indicated that its half-life is 1.7 h in plasma after i.p. dosing. We evaluated the effect of DAL on the growth of orthotopic 4T1 and E0771 mammary tumors with daily treatment of DAL or saline (Fig. 4A). We used 37.5 mg/kg DAL for these experiments based on the pharmacokinetics results as well as in vitro and in vivo data. In E0771 tumors, with treatment initiated at a ≈100 mm³ volume, DAL increased the tumor volume doubling time from 2.65 ± 0.27 d to 4.77 ± 0.58 d (P = 0.01). In 4T1 tumors, with treatment starting at a larger tumor volume of ≈200 mm³, DAL increased the doubling time from 2.20 ± 0.07 d to 11.21 ± 3.53 d (P = 0.04; Fig. 4B). Importantly, DAL induced no weight loss or other noted signs of toxicity in these mice. Together, these results demonstrate that DAL is a safe chemical agent with prominent antitumor effect.

Fig. 4.

DAL has antitumor effects in orthotopic mouse models. (A) Tumor growth curves for 4T1 (n = 4) and E0771 orthotopic mammary tumors (n = 5). We treated tumor-bearing female SCID mice with daily i.p. injections of saline or 37.5 mg/kg DAL. Throughout the study, we measured tumor size on alternating days. Because all tumors grew at different rates to the start size, we time- and size-matched tumors for their initial treatment on day 0 just after the first measurement. (B) DAL increased the doubling time for 4T1 tumors from 2.20 ± 0.07 d to 11.21 ± 3.53 d (P = 0.044) and that for E0771 tumors from 2.65 ± 0.27 d to 4.77 ± 0.58 d (P = 0.010).

DAL Interferes with Adhesive Properties of Endothelial Cells.

To gain further insight into the antivascular effects of DAL and to test the involvement of DAL in modulating actin assembly, we plated normal human umbilical vein endothelial cells (HUVEC) and treated them with 10 μM DAL for 3 h. We fixed, permeabilized, and subsequently stained the cells with phalloidin to visualize F-actin (Fig. 5A). Cells treated with DAL were smaller than untreated control cells and had a rounded shape. Actin stress fibers were clearly visible and aligned with the long axis in control cells, whereas DAL-treated cells featured actin present only as small fibers and spots throughout the cells or accumulated close to the nuclei. Together, these results show that DAL destroys or significantly changes the normal organization of the actin cytoskeleton and further inhibits actin-dependent processes such as cell spreading on the extracellular matrix.

Fig. 5.

 DAL interferes with adhesive properties of endothelial cells in vitro. (A) DAL is involved in actin cytoskeleton assembly regulation. For F-actin visualization, we plated HUVECs and treated them with saline or DAL (10 μM) for 3 h. After fixing and permeabilizing these cultured cells, we subsequently stained them with phalloidin. (B) DAL interferes with adherens junctions in networks of HUVEC cells cultured on matrigel. After 24 h, control HUVEC cells organized into a network of cordlike structures, whereas DAL-treated HUVECs show disjoined networks. Existing cordlike structures, when exposed to 10 μM of DAL for 3 h, were also disrupted. (C) A wound-healing motility assay with HUVECs. We disrupted HUVEC monolayers mechanically with a pipette tip and followed the migration of cells to “heal” the wound using microscopy. DAL (5 μM) slowed healing compared with the control, shown at 24 h after wounding.

To test whether DAL affects EC functions, we used in vitro models of vascular network formation from HUVECs. After overnight cultivation in matrigel, HUVECs spontaneously assemble into cord-like structures that resemble vascular networks. Treatment with DAL not only blocks HUVEC network formation, but also leads to complete reorganization of existing networks over a few hours (Fig. 5B). These results indicate that DAL likely interferes with cell-cell junction integrity and actin assembly in ECs as a mechanism for its antivascular effects. Because DAL affects actin cytoskeleton assembly, we speculated that DAL might also affect cell motility. We used a scratch wound assay to study HUVEC motility based on the healing rate of a wounded monolayer. Cells treated with DAL did not move and failed to close the wound (Fig. 5C). These results demonstrate that DAL can interfere with EC motility as well, likely representing a secondary mechanism for inhibiting angiogenic sprouting.

DAL Decreases Rac1 Activity and Promotes Degradation.

Considering that DAL brings about its antivascular effects by destabilizing cell adhesion through the actin cytoskeleton, we sought to identify which specific adhesion pathway it targets. Mechanical attachment at cell-cell adherens and tight junctions is regulated by dynamic changes in the actin cytoskeleton. Adherens junctions are composed of E-cadherin/β-catenin complexes and are connected to the actin cytoskeleton via α-catenin (25, 26), whereas tight junctions consist of transmembrane claudins and occludin and are associated with the actin cytoskeleton via ZO proteins (27, 28). We therefore assayed for potential targets of DAL in signaling pathways involved in actin remodeling. We found that DAL had no significant effects on 14 kinases (AKT1, AKT2, AKT3, PKA, PKCα, PKCδ, PAK1, EPHB2, EPHB4, p38α, PI3K-α, PI3K-β, PI3K-γ, PI3K-δ). We then suspected that Rho-GTPase family members might be targets for DAL. In particular, we reasoned that Rac1 might be the target because of the combination of disrupted adhesion and actin networks along with inhibited motility. Rac1 plays an important role in actin cytoskeletal organization and is involved in the coordination of cell adhesion (29). Furthermore, Rac1 is responsible for stabilizing endothelial cell junctions by opposing actin remodeling by Rho (30). To test whether DAL might modulate Rac1, RhoA, and Cdc42 activities, we used affinity purification assays to monitor RhoA and Cdc42/Rac1 activity. We applied the Rho-binding domain (RBD) of the Rho effector protein rhotekin and the p21 binding domain (PBD) of the Cdc42/Rac effector protein p21 activated kinase 1 (PAK) to pull down active forms of each protein. We found that DAL decreased Rac1 activity in a concentration-dependent manner (Fig. 6A)—and seemed to decrease Rac1 levels—but had no effect on Cdc42 or RhoA activity, indicating Rac1 regulators or Rac1 itself as potential targets of DAL.

Fig. 6.

DAL decreases Rac1 activity and activates Rac1 degradation. (A) Detection of RhoA activity in HUVEC cells treated with DAL. (B) DAL decreases level of Rac1 in HUVECs. We treated HUVEC cells with 10 μM DAL for 24 h and were lysed. A Western blot with anti-human Rac1 antibody is shown (Upper), whereas Lower represents a loading control. (C) Treatment with DAL increases Rac1 ubiquitination in HUVEC cells. To enhance ubiqutination events, we pretreated HUVEC cells with the proteasome inhibitors lactacystin (2.5 mM) or MG-132 (0.2 mM) for 1 h and subsequently treated cells with DAL (10 μM) for 24 h. We used Duolink in situ for the detection of ubiqutinated Rac1 in DAL-treated and -untreated HUVEC cells. For the two Duolink PLA probes, we used primary mouse antibodies raised against human Rac1 and primary rabbit anti-ubiquitin. Fluorescent spots represent ubiqutinated Rac1. We additionally stained the nuclei with DAPI.

Treatment of cells with DAL did not change expression levels of Rac1 mRNA measured with quantitative PCR. We confirmed that DAL reduces protein levels of Rac1 (Fig. 6B), and we therefore suspected that DAL promotes Rac1 protein degradation. We questioned whether DAL might enhance Rac1 ubiquitination and therefore speed up degradation (31–34). For the detection of Rac1 ubiquitination in HUVECs, we identified Rac1-ubiquitin binding in situ by using a Duolink molecular proximity assay. In this assay, molecules thought to interact are targeted with fluorescent antibodies in situ, and individual interactions generate a signal detectable by microscopy only if secondary antibodies are located in close proximity to each other. For detection of ubiqutinated Rac1 proteins, we used a mouse antibody raised against Rac1 protein and a rabbit antibody raised against a ubiquitin. We used low concentrations of irreversible (lactacystin) and reversible (MG-132) proteasome inhibitors to block the ubiquitin proteasome system and enrich the fraction of ubiquitinated proteins in our assay. Cells treated with DAL contained more ubiquitinated Rac1 protein compared with control cells (Fig. 6C). These results strongly suggest that DAL promotes Rac1 ubiquitination, leading to rapid Rac1 degradation.

Conclusion

Angiogenesis inhibitors targeting the VEGF signaling pathway have proven to be an efficacious anticancer treatment strategy. Novel targets for drug discovery are needed to diversify antiangiogenic treatment and enhance safety for the benefit of cancer patients. To improve the diversity of options for antivascular therapy, we applied a high-throughput screen for small molecules that affect cell adhesion. We identified a nontoxic natural plant product, DAL, with antivascular properties leading to an antitumor effect. Intriguingly, DAL affects cell adhesion by promoting the ubiquitination and proteasomal degradation of the Rho-GTPase Rac1, which has been noted to play a role in cell-cell adhesion stability. This small molecule may serve as a lead compound for the development of a novel class of anticancer drugs.

Materials and Methods

Compound Library.

We used a compound library from the Partners Center for Drug Discovery (PCDD) high-throughput assay. This library includes 50,000 compounds from different sources: (i) marketed drugs from Prestwick; (ii) purified natural products; (iii) 3,000 end-blocked tetrapeptides; (iv) small molecules purchased from Peakdale, Maybridge, CEREP, Bionet Research Ltd., Chemical Diversity Lab, ENAMINE, and I.F.Lab Ltd; and (v) small molecules from PCDD chemists and from different academic institutions. Compounds are stored at a stock concentration of 2 mg/mL to ≈5 mM. For the screen, the 5 mM solutions were diluted to ≈1.67 mM in 100% DMSO, and 0.4 μL of each was spotted in wells in 384-well plates by using a Beckman Coulter Multimek 96/384 Channel Automated Pipetter. The average molecular weight of the compounds in the library is 400 Da (range = 225–600), and this library has been tested successfully in a number of screening assays.

High-Throughput Screening (HTS).

We used a cell-based readout system for compound selection. For our cell-based screening, we selected an optimal cell density of 10,000 breast cancer MDA-MB-231 cells per well to produce the most prominent signal. We incubated the plated cells with each compound, washed with PBS, and used the Cell Titer-Glo (Promega) reagent to detect the number of cells remaining attached in each well. This ATP-based detection has a linear relationship between the luminescent signal and the number of cells per well. Of the 50,000 compounds screened, we selected 86 (0.172% of screened compounds) at 1.67 μM concentration for the second screening step. We treated breast cancer cells with the compounds at 3 μM for 3 h and then examined for changes in cell shape by microscopy. We selected 12 of the 86 compounds for further study. We passed all these selected compounds through a number of toxicity assays to eliminate any toxic compounds. We confirmed the absence of toxicity for the candidate compounds in the cellular model in repetitive experiments and reassayed at three different concentrations (1.67, 5, and 10 μM). We analyzed the compounds that showed no toxicity at any concentrations tested in a zebrafish model and then used compounds at concentrations that did not affect zebrafish survival during embryonic development and adult life for further experiments. Of the 50,000 compounds, we selected the compound DAL for further characterization. As a further test of toxicity, we injected DAL i.p. at 100 mg/kg in SCID mice and monitored for signs of toxicity including cachexia, ruffled fur, diarrhea, anorexia, skin ulceration, and toxic death.

Zebrafish Maintenance and Drug Eexposure.

We raised zebrafish in accordance with established protocols (19, 20). The age of embryos is indicated by the hours after fertilization and days after fertilization for all experimental data shown. For the screening study, we incubated zebrafish embryos with the selected compounds at a concentration of 5–10 μM. We then examined treated embryos under a 100× PlanNeofluor objective mounted on a Nikon TE-200 epifluorescence microscope. Images of embryos bearing GFP-positive cells were captured with a Zeiss stereomicroscope.

Regeneration Experiments.

For the regeneration experiments, we amputated caudal fins at approximately the 50% proximal-distal level (21). We kept the amputated fish in individual 250-mL fish tanks and used at least five fish in each group. We carried out staging for normal blood vessel regeneration at 25 °C with TG(fli1:EGFP)y1 fish, with the same fin ray and vessels of each fish photographed at different time intervals during the course of the time-lapse study.

Determination of Rac1 Activity.

We measured Rac1 and cdc42 activities with a Rac1 (BK035) activation assay biochem kit (Cytoskeleton) according to the manufacturer protocol. In brief, we bovine brain extract-starved HUVECs for 12 h in 1% FBS-EGM (Lonza) and pretreated with various concentrations of DAL for 30 min at 37 °C. We then stimulated the HUVECs with VEGF at 50 ng/μl for 15 min at 37 °C. After stimulation, we immediately incubated the cells on ice, washed twice with PBS, and lysed with lysis buffer. We precleared the lysates and added 15 μL of PAK-PBD beads for pull-down of activated Rac1 or cdc42. After rocking at 4 °C for 1 h, we washed the beads once and boiled them at 100 °C for 2 min. We then resolved the protein lysates in a 8–16% gradient gel (Invitrogen) and transferred them to a PVDF membrane. We added primary antibodies (anti-Rac1, 1:1,000, cytoskeleton or anti-cdc42, 1:1,000, Cell Signaling) to bind to the protein overnight at 4 °C, followed by incubation with HRP-conjugated secondary antibodies. We measured protein activity by the standard ECL method (Amersham).

Measurement of Vascular Parameters in Tumors in Mice.

We treated female SCID mice bearing orthotopic GFP-labeled 3-mm 4T1 mammary tumors with five daily i.p. injections (day 0–day 4) of 37.5 mg/kg DAL or saline. We surgically implanted mammary fat pad windows (35) to facilitate intravital imaging through multiphoton microscopy with a custom-built multiphoton microscope with a 20× 0.95 N.A. objective (Olympus) and imaged mice by fluorescent angiography after retro-orbital injection of 60 μL of 10 mg/mL TAMRA-dextran (2 MDa molecular mass). We collected images on days 0, 2, and 4, with the first images collected immediately before the first treatment. We imaged 2 × 2 grids in the sample plane (630-μm field of view at each position) tissue from 0 to 200 μm depth. To measure vascular parameters in the mice over the course of treatment, we applied a custom 3D vascular tracing and analysis software package based on MATLAB (Mathworks) to the images.

Tumor Growth Delay Study.

To assess the antitumor effects of DAL, we treated female SCID mice bearing orthotopic 4T1 or E0771 mammary tumors with daily i.p. injections of 37.5 mg/kg DAL or saline. To calculate tumor volumes, we measured the major and minor axis diameters with calipers and calculated the tumor volume assuming a spherical shape based on the average of these two dimensions. We implanted the tumors in the mammary fat pad and allowed them to grow to a starting size (small ≈100 mm³ tumors for E0771, large ≈200 mm³ tumors for 4T1) over ≈2 wk. To ensure that tumors in the two groups were of similar growth rates, we began the treatment with saline or DAL in pairs of tumors with similar sizes and growth times. We measured tumor volumes every other day until the tumors reached and average of ≈10 mm in diameter. We quantified tumor growth with the doubling rate, calculated by fitting an exponential growth curve to the data for tumor size versus time.

Kinase Inhibition Screening.

To determine whether DAL might target kinase activity, we screened against kinases (AKT1, AKT2, AKT3, PKA, PKC-α, PKC-δ, PAK1, EPHB2, EPHB4, p38-α, PI3K-α, PI3K-β, PI3K-γ, PI3K-δ) in a 10-dose IC50 mode with threefold serial dilutions starting at 100 μM. We used staurosporine as a control starting at 10 μM and carried out the reactions at 1 μM ATP.