A Specific Antagonist of the p110δ Catalytic Component of Phosphatidylinositol 3′-Kinase, IC486068, Enhances Radiation-Induced Tumor Vascular Destruction

The response of tumor microvasculature to radiation depends on the dose and time interval after treatment (1, 2, 3, 4, 5, 6). Tumor blood flow decreases when tumors are treated with doses in the range of 20–45 Gy (6). However, blood volume increases if doses <500 rads are administered (1, 3). Blood flow studies (interstitial Xe clearance) of irradiated mouse sarcoma showed that blood flow increased within 3–7 days (3). Radiation doses in the range of 10–20 Gy given as a single treatment induce apoptosis in tumor vascular endothelial cells in a manner that depends on acid sphingomyelinase (7). These studies demonstrate the importance of tumor vascular response to radiation in achieving tumor control.

Tumor blood vessels show less response to radiation doses in the range of 2–3 Gy, which is typically used during conventional radiation therapy (8, 9, 10). The dorsal skinfold window model of tumor blood vessels has shown that higher radiation doses in the range of 6 Gy are required to achieve tumor vascular destruction. The response of tumor blood vessels can be enhanced with inhibitors of receptor tyrosine kinases (RTKs; Refs. 8, 10, 11). These studies have demonstrated that RTK inhibitors administered before irradiation attenuate Akt-phosphorylation in vascular endothelium and improve tumor growth delay in response to radiation.

Phosphatidylinositol 3′-kinase (PI3k) activity is required for growth factor-mediated survival of various cell types (12, 13), suggesting that these growth factors and other RTK agonists exert their effects in a PI3k-dependent manner. PI3k catalyzes the addition of a phosphate group to the inositol ring of phosphoinositides (14). One target of these products is the serine/threonine protein kinase B (PKB; or Akt). Dominant negative genetic constructs that specifically antagonize PI3k-dependent signaling eliminate radiation-induced Akt-phosphorylation (15). Akt subsequently phosphorylates several downstream targets, including the Bcl-2 family member Bad and caspase-9, thus inhibiting their proapoptotic functions (16, 17). Akt also has been shown to phosphorylate the forkhead transcription factor FKHR (18). In addition, many other members of the apoptotic machinery and transcription factors contain the Akt consensus phosphorylation site (13), further suggesting that Akt plays a prominent role in inhibiting apoptosis.

Type IA PI3k is composed of a heterodimer of an M_r 85,000 (p85) or an M_r 55,000 (p55) adaptor subunit and an M_r 100,000 (p110) catalytic subunit (19, 20). The p85 subunit contains two Src homology 2 domains, which bind to tyrosine-phosphorylated receptors after stimulation of cells with growth factors and in this manner recruit p110 into the complex at the cell membrane (21). The type IA PI3k family of lipid kinases consists of three isoforms (p110α, p110β, and p110δ). IC486068 is a quinazolin-4-one that has a specific inhibitor of p110δ catalytic component of PI3k (22). This compound has IC₅₀ values for p110α = 400, p110β = 500, and p110δ = 20 nm. We studied the radiosensitizing effects of IC486068 in vascular endothelial cells of tumor blood vessels. This compound inhibits endothelial migration and tubule formation in irradiated endothelial cells. The radiation effect on tumor microvasculature was enhanced in the tumor vascular window model and Doppler blood flow analysis of hind limb tumors. Moreover, IC486068 enhanced radiation-induced tumor growth delay.

Human umbilical vascular endothelial cell (HUVEC) and human microvascular endothelial cell (HMVEC) lines were obtained from Clonetics (San Diego, CA), maintained in EBM-2 medium supplemented with EGM-2 MV Singlequots (Cambrex, East Rutherford, NJ). Only fourth or fifth passage cells were studied. The GL261 murine glioma cell line was obtained from Dr. Yancy Gillespie (University of Alabama, Birmingham, AL; Refs. 9, 10). GL261 cells were maintained in DMEM with Nutrient Mixture F-12 1:1 (Life Technologies, Rockville, MD) with 7% FCS, 0.5% penicillin-streptomycin, and 1% sodium pyruvate. Lewis lung carcinoma (LLC) cells were obtained from American Type Tissue Culture (Manassas, VA; Refs. 9, 10) and were maintained in DMEM supplemented with 10% FCS and 1% penicillin-streptomycin. MS1, RSV, HL60 leukemia, and mouse fibroblast cells were obtained from American Type Tissue Culture. All of the cells were incubated at 37°C in a 5% CO₂ incubator.

An Eldorado 8 Teletherapy 60Co Unit (Atomic Energy of Canada Limited, Mississauga, Ontario, Canada) was used to irradiate endothelial cell cultures at a dose rate of 0.84 Gy/min. Delivered dose was verified using thermoluminescence detectors. The number of cells undergoing apoptosis was quantified by microscopic analysis of apoptotic nuclei. Cells then were fixed and stained with H&E at 24 h after treatment with 6 Gy radiation and/or 100 nm IC486068. Cells then were examined by light microscopy. For each treatment group, five high power fields (×40 objective) were examined, and the number of apoptotic and total cells was determined. From these numbers, the percentage of apoptotic cells for each group was determined. Annexin V staining was quantified using FITC conjugation and flow cytometry. Experiments were performed in triplicate, and mean and SE were calculated.

Clonogenic survival analysis was performed as we have described previously (9, 10). Briefly, HUVEC culture plates were treated at each radiation dose level with or without 1–100 nm IC486068 for 1 h before irradiation. After treatment with radiation and/or antagonist, cells were trypsinized, counted by hemocytometer, and subcultured into fresh medium. After 14 days, the cells were fixed with cold methanol and stained with 1% methylene blue. Colonies with 50 cells were counted, and the surviving fraction was determined.

HUVECs were treated with IC486068, 100 nm with or without 3 Gy irradiation. At the indicated times, cells were washed twice with PBS and lysis buffer (20 nm Tris, 150 mm NaCl, 1 mm EDTA, 1% Triton X-100, 2.5 mm sodium pyrophosphate, 1 mm phenylmethylsulfonyl fluoride, and 1 μg/ml leupeptin) added. Protein concentration was quantified by the Bio-Rad method (Hercules, CA). Twenty μg of total protein were loaded into each well and separated by 8 or 12% SDS-PAGE gel, depending on the size of the target protein investigated. The proteins were transferred onto nitrocellulose membranes (Hybond ECL; Amersham, Piscataway, NJ) and probed with antibodies for the phosphorylated Akt and total Akt (New England Biolabs, Beverly, MA).

HUVECs were grown to 80% confluence in 100-mm dishes. IC486968 (100 nm) was added for 1 h, and the cells were treated with 3 Gy. Cells then were washed with PBS, detached with 1% trypsin, and seeded at 10⁵ cells/well onto Matrigel-coated (200 μl of 10 mg/ml) wells (BD Bioscience, Bedford, MA). The plate was allowed to sit at room temperature for 15 min and then incubated at 37°C for 30 min to the Matrigel to polymerize. The cells were incubated for 24 h to allow capillary-like structure formation. For optimal visualization of tubes, medium was removed carefully after incubation, and agarose was gently added to cells. After solidification of agarose, immobilized tubes were fixed and stained with Diff-Quick solution (BD Bioscience, Bedford, MA). Stained tubules were washed three times with PBS. The relative quantity of tubules was quantified by microscopic visualization and counting.

HUVECs were treated after reaching 80% confluence in 100-mm dishes. Cells were washed twice with sterile PBS. Trypsin buffer was added to the dish and incubated at 37°C for 3 min. Trypsin digestion then was inhibited by the addition of complete medium. A total of 2.5 × 10⁵ HUVEC cells were placed into a fibronectin-coated Boyden chamber in EBM-2 medium (Clonetics) with IC486068 (100 nm). Vascular endothelial growth factor was used as the growth factor in Boyden chamber migration assays. BSA-coated chambers served as negative controls for cell migration. Cells then were incubated at 37°C for 6 h. Cells then were removed from Boyden chambers using swabs. Media and cells again were swabbed from the inside of the chamber. Chambers then were placed into wells containing Cell Stain Solution (Chemicon International, Temecula, CA) and incubated for 30 min at room temperature. Cell stain then was removed from wells, and cells were washed three times with PBS. Boyden chambers were washed with distilled water. Cells migrating to the bottom of the membrane were counted by microscopy. Cell stain then was extracted using extraction buffer (Chemicon International) on a shaker for 5–10 min. One hundred μl of stained solution from cell extractions were placed into a microtiter plate, and absorbance was read at 550 nm.

The dorsal skinfold window is a 3-g plastic frame applied to the skin of the mouse before tumor implantation and remains attached for the duration of the study. The chamber was screwed together, and the epidermis was incised and remained open with a plastic covering. The midline was found along the back, and a clip was placed to hold the skin in position. A template, equivalent to the outer diameter of the chamber, was traced, producing the outline of the incision. A circular cut was made tracing the perimeter (7-mm diameter) of the outline, followed by a crisscross cut, thus producing four skin flaps. The epidermis of the four flaps then was removed using a scalpel with an effort to follow the hypodermis superior to the fascia. The area then was trimmed with fine forceps and iris scissors. The template was removed, and the top piece of the chamber was fixed with screws. During surgery, the area was kept moist by applying moist drops of PBS with 1% penicillin/streptomycin solution. The bottom portion of the chamber was put in place, and the top was carefully positioned on the cut side so that the window and the circular incision were fitted. Antibiotic ointment was applied at this time. The three screws that hold the chamber together then were positioned into the chamber holes and tightened so that the skin was not pinched, thus avoiding decreased circulation. Animals were placed under a heating lamp for several days. Tumor angiogenesis within the window was monitored by microscopy. Tumor blood vessels developed in the window within 1 week.

We studied the time- and dose-dependent response of tumor blood vessels to radiation using the window model. Vascular windows were treated with 2 Gy of superficial X-rays using 80 kVp (Pantak X-ray Generator; East Haven, CT). Five mice were studied in each of the treatment groups. IC486068 (25 mg/kg) was injected i.p. 30 min before irradiation. The window frame was marked with coordinates, which were used to photograph the same microscopic field each day. Vascular windows were photographed using a 4× objective to obtain a 40× total magnification. Color photographs were used to catalogue the appearance of blood vessels on days 0–7. Photographs were scanned into Photoshop software (Adobe, San Jose, CA), and vascular centerlines were positioned by ImagePro software (Media Cybernetics, Silver Spring, MD) and verified by an observer blinded to the treatment groups. Tumor blood vessels were quantified using ImagePro software, which quantifies the vascular length density (VLD) of blood vessels within the microscopic field. Centerlines were verified before summation of the VLD. The means and 95% CIs of VLD for each treatment group were calculated, and variance was analyzed by the general linear model and Bonferroni t test.

C57BL/6 mice received s.c. injections in the right thigh with 10⁶ viable cells of a murine glioblastoma (GL261) or lung carcinoma (LLC) suspended in 0.2 ml of a 0.6% solution of agarose. Each set of six mice was stratified into four groups on day 1 (vehicle control, radiation alone, IC486068 alone, or IC486068 + radiation) to control for mean tumor volume. An equal number of large- and intermediate-sized tumors were present in each group. Mouse tumors were stratified into groups so that the mean tumor volume of each group was comparable. The mean tumor volumes were 240 mm³ (range, 205–262) on day 1 for LLC and 260 mm³ (range, 240–285) for GL261. These volumes were reached at 12 and 14 days following implantation for LLC and GL261, respectively.

Irradiated mice were immobilized in Lucite chambers, and the entire body was shielded with lead except for the tumor-bearing hind limb. Radiation was administered within 30 min of IC486068 (25 mg/kg) i.p. injection. A total dose of 18 Gy was administered to the appropriate mice in six fractionated doses of 3 Gy on days 1–6. Tumor volumes were measured three times weekly using skin calipers (8, 10). The volumes were calculated from a formula (a× b× c/2) that was derived from the formula for an ellipsoid (πd³/6). Data were calculated as the percentage of original (day 1) tumor volume and graphed as fractional tumor volume ± SE for each treatment group.

Blood flow within these tumors was quantified by power Doppler after the third fraction of irradiation. Tumor blood flow was imaged with a 5–10-MHz linear Entos probe attached to an HDI 5000 (probe and HDI 5000 from ATL/Philips, Bothell, WA) as we have described previously (8, 10). Power Doppler sonographic images were obtained with the power gain set to 82%. A 20-frame cineloop sweep of the entire tumor was obtained with the probe perpendicular to the long axis of the lower extremity along the entire length of the tumor. Intensity of blood flow was imaged as areas of color and quantified using HDI-lab software (ATL/Philips). This software allows direct evaluation of power Doppler cineloop raw. The color area was recorded for the entire tumor. Five mice were entered into each treatment group. Values for color area were averaged for each tumor set, and treated groups were compared with controls using the unpaired Student’s t test.

We used the general linear model (logistic regression analysis) to test for associations between the numbers of apoptotic cells present in culture, clonogenic survival, tumor blood flow, and tumor volumes. We applied the Bonferroni method to adjust the overall significant level to 5% for the multiple comparisons in this study. All of the statistical tests were two sided, and differences were considered statistically significant for P < 0.05. SAS software version 8.1 (SAS Institute, Inc., Cary, NC) was used for all of the statistical analyses.

To determine whether the δ isoform of p110 catalytic subunit of PI3k is present in endothelial cells, we extracted total protein from HUVECs and used antibodies specific for the δ isoform for Western immunoblots analyses. Fig. 1 shows that the p110δ isoform is present in HUVECs and HMVECs. Radiation induces the activation of Akt phosphorylation in a PI3k-dependent manner (9, 15). To determine whether the p110δ isoform contributes to radiation-induced Akt phosphorylation, IC486068 was added to cells 30 min before irradiation. Fig. 2 shows a Western immunoblot using the antibody that is specific for the phosphorylated form of Akt. Following treatment with 3 Gy, there was an increase in phosphorylation of Akt at 15 min after irradiation of HUVECs. The p110δ-specific inhibitor IC486068 attenuated radiation-induced activation of Akt phosphorylation (Fig. 2).

PI3k and Akt signaling contributes to endothelial cell viability (9, 23). To determine whether p100δ inhibition contributes to cell viability, we studied apoptosis and clonogenic survival in HUVECs treated with IC486068 and radiation. Fig. 3 shows an increase in the percentage of apoptotic endothelial cells to a total of 3.5% and 3% following treatment with radiation or IC486068, respectively. When cells were pretreated with IC486068 followed by irradiation, there was a significant increase in apoptosis to 9% (P = 0.04) compared with treatment with either radiation or IC486068 alone. Annexin V staining was quantified using FITC conjugation and flow cytometry. Fig. 3,B shows the mean and SE of three annexin V staining experiments. Treatment with IC486068 combined with radiation significantly increased the annexin staining compared with either agent alone (P = 0.02). Clonogenic assay was performed to determine whether IC486068 enhances radiosensitivity. HUVECs were treated with IC486068 (100 nm) for 60 min before irradiation. Fig. 3 C shows a significant reduction in clonogenic cell survival when cells are treated with IC486068 before irradiation compared with radiation alone (P = 0.01). IC486068 reduced plating efficiency to 90% compared with untreated control cells; IC486068 increased cytotoxicity of 2-Gy irradiation of endothelial cells 10-fold.

Cell migration and tubule formation in Matrigel were used to measure the physiologic effects of IC486068 on endothelial cells. Cells were treated with 100 nm IC486068 with or without 3-Gy irradiation before culturing cells in Boyden chambers. Fig. 4,A shows that cells treated with IC486068 and radiation showed reduction in cell migration compared with cells treated with radiation alone (P = 0.01). Fig. 4,B shows endothelial cell tubule formation in Matrigel. HUVECs formed tubules within 24 h of culturing in Matrigel. Radiation alone or IC486068 alone produced no significant attenuation of tubule formation of HUVECs in Matrigel (Fig. 4,B). Fig. 4 C shows the quantification of tubule formation as measured by microscopy. IC486068 combined with radiation nearly eliminated capillary-like tubules and significantly attenuated tubule formation (P = 0.03).

To determine whether IC486068 enhances radiation-induced destruction of tumor vasculature, IC486068 was administered to mice before irradiation with 2 Gy. Tumor VLD was measured using intravital tumor vascular window. Fig. 5,A shows the vasculature within LLC tumors implanted into the dorsal skinfold window in C57BL6 mice. Representative photographs of tumor vasculature before and 48 h after treatment with IC486068, 2 Gy, or IC486068 followed by 2 Gy indicate that p110δ inhibition increases tumor vascular destruction compared with either agent alone. Five mice were treated in each of the treatment groups, and the VLD after treatment was quantified (Fig. 5 B). Mean vascular length densities for 4 days are shown as a bar graph. At 48 h after treatment with IC486068 and 2 Gy, VLD in tumors was significantly reduced to 8% of that at 0 h (P < 0.01). In comparison, tumors treated with either 2 Gy or IC486068 alone showed an insignificant reduction in VLD to 75% and 84% compared with VLD at the 0-h time point, respectively. Combined IC486068 and 2 Gy achieved significantly greater reduction in VLD compared with either agent alone (P = 0.011). VLD in untreated mice showed no significant change in 48 h.

To determine whether IC486068 enhances tumor growth delay in irradiated tumors, mice bearing LLC and GL261 hind limb tumors were treated with i.p. injection of 25 mg/kg IC486068 or drug vehicle 30 min before each 3-Gy dose of radiation for a total of seven administrations. The inhibitor and radiation were discontinued after day 6 (Fig. 6). The mean fold-increase in tumor volumes in five mice in each of the treatment groups (vehicle, IC486068, vehicle + 18 Gy, and IC486068 + 18 Gy) is shown. The number of days for GL261 tumor growth to increase fivefold compared with day 1 tumor size was 8, 12, 19, and 33 days for each treatment group, respectively. LLC and GL261 tumors showed a significant delay in tumor growth when IC486068 was added before daily 3-Gy fractions compared with either agent alone (P < 0.05).

To determine whether prolonged growth delay correlated with a reduction in tumor blood flow, amplitude-modulated power Doppler was used to monitor blood flow. Fig. 7 shows representative images of signal intensity of blood flow in GL261 tumors on day 5 of treatment. As tumor volumes increase during the 4 days of therapy (days 1–5), there is a slight, insignificant increase in blood flow. Reduced blood flow in tumors treated with IC486068 and radiation correlated with the improved tumor growth delay that was found in Fig. 6. The bar graph shows the average blood flow within GL261 tumors at day 5 of daily treatment in each of the treatment groups. Tumors treated with IC486068 and radiation approached a significant reduction in blood flow compared with tumors treated with radiation alone (P < 0.05).

The objective of this study was to evaluate the role of PI3k, specifically p110δ, in the response of tumor blood vessels to ionizing radiation. Radiation induces phosphorylation of Akt, which depends on PI3k activity (9, 23). The nonspecific inhibitors of p110, Wortmannin and LY294002, prevent radiation-induced Akt phosphorylation, induce apoptosis, and enhance radiation-induced cytotoxicity. These inhibitors are nonspecific and also are potent inhibitors of DNA-dependent protein kinase, FRAP-mTOR, smooth muscle myosin light chain kinase, and casein kinase 2 (24, 25, 26). In contrast, IC486068 achieves no inhibition of DNA-PK (22). We found that the p110δ isoform is expressed in endothelial cells and overexpressed in HL60 leukemia cells, suggesting that the δ isoform also is a molecular target in acute myelogenous leukemia. IC486068 enhanced apoptosis and cytotoxicity induced by irradiation of endothelial cells. Cytotoxicity was enhanced to a greater degree following 2-Gy irradiation compared with 6 Gy. One possible explanation for this change in slope of the clonogenic survival curve could represent a small percentage of cells that do not express the δ isoform of p100. Moreover, IC486068 inhibited endothelial tubule formation when combined with radiation. These findings suggest that the p110δ-specific inhibitor prevents tubule formation in part by impacting on radiation-induced cytotoxicity.

Endothelial cell tubule formation involves several physiologic processes. Endothelial cells cultured in Matrigel form tubules within several hours. This requires intercellular signaling, cytokinesia, and tubule differentiation. Treatment with a radiation dose of 2 Gy does not inhibit tubule formation. Likewise, IC486068 (100 nm) produces no significant reduction in tubule formation, whereas IC486068 followed by irradiation abolished tubule formation. We found no recovery of tubule formation during the course of 36 h in treated cells. Together with studies from the vascular window and Doppler blood flow, these findings suggest that p110δ-specific inhibitors enhance the antiangiogenic effects of ionizing radiation.

Tumor blood vessels develop during the course of 6–8 days within the vascular window before a single dose of 3 Gy is administered. This dose has little effect on these tumor blood vessels (8, 10). Previous studies have shown that inhibition of RTKs or nonspecific inhibitors of PI3k significantly enhance radiation-induced tumor blood vessel destruction (8, 10). It is unclear whether this effect is caused by enhancement of cytotoxicity from radiation or inhibition of vascular repair because both processes can contribute to enhanced tumor blood vessel destruction. In the present study, IC486068 prevented tubule formation following irradiation. These findings suggest that p110δ participates in tubule formation and repair of microvasculature following cytotoxic therapy.

The focus of these studies has been on the antiangiogenic events that occur through IC486068 inhibition of p110δ within endothelial cells without considering any effects that the drug may have on tumor cells. We have found that this compound has little effect on the LLC and GL261 cancer cell models used within these experimental tumor models.⁶ Therefore, the regression of new vessel formation and decreased growth of flank tumors are likely to be the result of inhibition of p110δ in endothelial cells and not tumor cells.

Previous studies with nonspecific inhibitors, Wortmannin and LY294002, may have demonstrated a combined inhibition of DNA-PK and PI3k (27). Attenuation of programmed cell death occurs through several independent mechanisms. One such mechanism involves the Bcl gene family members, which are directly phosphorylated by Akt and participate in the Akt-mediated antiapoptotic effect (28). This is supported by our findings that the PI3k inhibitors or radiation alone can promote HUVEC apoptosis (9, 23). Most importantly, we found that the number of apoptotic bodies within HUVECs increased dramatically when treated by the PI3k inhibitor and radiation together compared with treatment with radiation or a PI3k dominant negative alone (23). In the present study, we found a radiation activation of Akt phosphorylation in endothelial cells. The p110δ inhibition resulted in enhancement of endothelial apoptosis and decreased viability. These data indicate that the p110δ participates in radiation-induced phosphorylation of Akt and subsequent enhancement of cell viability through prosurvival pathways.

Two measures of tumor vasculature (Doppler blood flow and vascular window) were used to assess the effectiveness of IC486068 at enhancing radiation-induced tumor vascular destruction. Intravital tumor vascular window provided a measure of VLD that was significantly reduced when IC486068 was added before irradiation compared with either agent alone. VLDs are followed longitudinally and compared with those of tumors before treatment (0-h time point). VLDs of untreated tumors do not vary during the 72 h of observation. This finding was supported by amplitude-modulated Doppler blood flow measurement, which showed reduced blood flow when tumors were treated with IC486068 together with radiation compared with either agent alone. Although power Doppler may underestimate the response of tumor microvasculature, when taken together with vascular window data, these findings suggest that p110δ inhibition enhances the therapeutic effect of radiation in part through enhanced tumor vascular destruction. The contribution of tumor vascular destruction on tumor control also is supported by previous studies that show antiangiogenic agents enhance radiation-induced tumor control (8, 10). Mechanisms of interaction between IC486068 and radiation could involve increased cytotoxicity within vascular endothelium or within tumor cells. This finding indicates that a primary mechanism of interaction between this p110δ antagonist and radiation is within the tumor blood vessels.

Decreased tumor blood flow results in hypoxia, which could contribute to radioresistance (29). The effect of hypoxia could be overcome by the concomitant reduction in nutrients and growth factors achieved by ischemia. To evaluate whether the ischemia induced by IC486068 and radiation has an impact on tumor control, we studied growth delay in hind limb tumors. This showed that destruction of tumor vasculature correlates with improved growth delay compared with tumors treated with radiation alone. Although hypoxia could have a lessening effect on the interaction between these agents, there was a significant improvement in tumor growth delay.