β₁ Integrin Inhibition Dramatically Enhances Radiotherapy Efficacy in Human Breast Cancer Xenografts

β₁ Integrin signaling has been shown to mediate cellular resistance to apoptosis after exposure to ionizing radiation (IR). Other signaling molecules that increase resistance include Akt, which promotes cell survival downstream of β₁ integrin signaling. We previously showed that β₁ integrin inhibitory antibodies (e.g., AIIB2) enhance apoptosis and decrease growth in human breast cancer cells in three-dimensional laminin-rich extracellular matrix (lrECM) cultures and in vivo. Here, we asked whether AIIB2 could synergize with IR to modify Akt-mediated IR resistance. We used three-dimensional lrECM cultures to test the optimal combination of AIIB2 with IR treatment of two breast cancer cell lines, MCF-7 and HMT3522-T4-2, as well as T4-2 myr-Akt breast cancer colonies or HMT3522-S-1, which form normal organotypic structures in three-dimensional lrECM. Colonies were assayed for apoptosis and β₁ integrin/Akt signaling pathways were evaluated using Western blot. In addition, mice bearing MCF-7 xenografts were used to validate the findings in three-dimensional lrECM. We report that AIIB2 increased apoptosis optimally post-IR by down-regulating Akt in breast cancer colonies in three-dimensional lrECM. In vivo, addition of AIIB2 after IR significantly enhanced tumor growth inhibition and apoptosis compared with either treatment alone. Remarkably, the degree of tumor growth inhibition using AIIB2 plus 2 Gy radiation was similar to that of 8 Gy alone. We previously showed that AIIB2 had no discernible toxicity in mice; here, its addition allowed for a significant reduction in the IR dose that was necessary to achieve comparable growth inhibition and apoptosis in breast cancer xenografts in vivo. [Cancer Res 2008;68(11):4398–405]

Radiation therapy is an effective modality used for breast cancer treatment; however, tumor resistance and recurrences remain significant clinical problems (1). Increasing evidence indicates that in addition to DNA damage, multiple cellular mechanisms such as cell interactions with neighboring cells and the microenvironment fundamentally influence cell fate in response to ionizing radiation (IR; ref. 2). Indeed, single doses of IR to single cells could modify cell-to-cell and cell-to-extracellular matrix (ECM) interactions and confer heritable epigenetic traits (3).

Adhesion to ECM has long been known to modify radiation sensitivity in several cell types, including cancer. Integrins, a major class of transmembrane molecules that mediate adhesion and cell-ECM interactions, are aberrantly expressed in many cancer cell types and are up-regulated post-IR (4, 5). In particular, β₁ integrins have been found to be up-regulated by clinically relevant doses of IR in cancer cell lines (4) and in human mammary epithelial cells in three-dimensional laminin-rich extracellular matrix (lrECM) cultures (3, 5). β₁ integrins have been implicated in mediating resistance to IR (6–10). However, the role of β₁ integrins as a strategic molecular target in conjunction with IR has not been fully investigated.

β₁ Integrins belong to a family of heterodimeric receptors whose ligands are arginine-glycine-aspartic acid–containing ECM molecules (11). Acting largely as mechanoreceptors, β₁ integrins transmit biochemical cues that can facilitate multiple cellular fates downstream, including apoptosis and survival (12). Increase in β₁ integrins seem to enhance cancer cell viability by promoting survival and confer resistance to chemotherapy in several tumor cell types (10, 13, 14). In addition, we have shown previously that in a subset of patients with early-stage breast cancer, high β₁ integrin expression is associated with poor overall survival, indicating that this subgroup of patients may benefit from more aggressive and targeted therapy (15).

Cancer cell resistance to IR has been shown previously to be mediated via Akt, a serine threonine kinase that lies immediately downstream of phoshatidyl-inositol-3 kinase (PI3K) and the integrin signaling pathways (16, 17). β₁ integrin has been shown to enhance cancer cell survival via the PI3K pathway in lung cancer (10) and in normal fibroblasts (18) in culture. We have shown previously that β₁ integrin inhibitory antibodies selectively induce apoptosis in breast cancer cells but not nonmalignant acini in three-dimensional lrECM cultures; these findings were validated in vivo with no toxicity to animals (19), indicating that β₁ integrin is a highly promising therapeutic target. In this report, we used the three-dimensional lrECM assay and in vivo models of breast cancer to test the hypothesis that β₁ integrins may facilitate IR-induced resistance by promoting Akt-mediated survival.

Cell culture. Nonmalignant human mammary epithelial cells, HMT-3522 (S-1), were originally derived from a woman with fibrocystic breast disease (20), and were cultured in H14 medium (21). S-1 cells were propagated on plastic in medium containing 10 ng/mL epidermal growth factor (EGF). Also derived from the same parental line, malignant HMT-3522 (T4-2) cells were propagated on collagen type I–coated flasks in the absence of EGF (21). T4-2 wild-type cells that were either stably transfected with a constitutively active myristoylated Akt (T4-2 myr-Akt) or empty vector control (T4-2 vc) were a gift from Hong Liu [University of California San Francisco (UCSF)]. Human breast cancer cell line, MCF-7, was obtained from the American Type Culture Collection. Three-dimensional lrECM cultures consisted of cells trypsinized from monolayer cultures and plated on top of commercially available gel produced from Englebreth-Holm-Swarm tumors (Matrigel; BD Sciences). Cell lines were maintained in H 14 medium as described previously (19) and transferred to the modified three-dimensional lrECM assay (19). This is referred to as day 0. Cultures were exposed to IR and/or AIIB2 on day 6 of culture after acinar formation for nonmalignant S1 cells and on days 4 and 5 of culture, sequentially for malignant cell lines. AIIB2 was added to culture medium on alternate days. All cultures were analyzed 72 h after the first treatment.

AIIB2. β₁ Integrin function-blocking antibody, clone AIIB2, was originally a purchased gift from Carolyn Damsky (UCSF). AIIB2 is a rat monoclonal IgG1 that was isolated from a human choriocarcinoma hybridoma that specifically binds β₁ integrin extracellular domain (22–24). Prior experiments using F(ab′)₂ fragments of enzyme-digested AIIB2 indicated that the epitope-binding portion of the antibody was active and resulted in down-modulation of β₁ integrin–mediated signaling (21, 25).

Apoptosis and proliferation assays in three-dimensional lrECM. Proliferating cells from three-dimensional lrECM cultures were detected by indirect immunofluorescence of Ki-67 nuclear antigen. Samples taken from cultures were fixed onto glass slides using methanol/acetone and blocked in 10% goat serum for 1 h at room temperature in a humidified chamber. Slides were then incubated in primary rabbit antibody against Ki-67, clone MIB-1 (Novocastra Laboratories), for 1 h. After washing in PBS, FITC-conjugated anti-rabbit secondary antibody (The Jackson Laboratory) was applied for 1 h in a dark humidified chamber. Slides were then washed and counterstained with 4′,6-diamidino-2-phenylindole (DAPI) before mounting with Vectashield mounting medium (Vector Laboratories). Apoptosis was assayed by detecting terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) in samples taken from three-dimensional lrECM cell culture using a commercially available kit (In Situ Cell Death Detection Kit, Fluorescein; Roche). Samples from cultures were fixed onto glass slides in 4% paraformaldehyde and permeabilized in 0.1% Triton X-100 in 0.1% sodium citrate. After washing, cells were incubated in TUNEL reaction mixture at 37°C for 60 min and mounted.

Fluorescence-activated cell sorting analysis. S-1 and T4-2 cells grown on tissue culture plastic were harvested using 0.25% trypsin. After resuspending cells in DMEM/F-12 medium containing trypsin inhibitor, cells were spun down and washed in 5% fetal bovine serum and 0.1% sodium azide in PBS on ice. Cells were incubated first in dilute primary Annexin V antibody (Trevigen) for at 4°C for 30 to 60 min, then washed and incubated with a fluorescein-conjugated IgG secondary antibody (1:100) for 30 to 60 min. After washing, the pellet was immediately suspended in 1% paraformaldehyde. Cells were analyzed using a Beckman-Coulter EPICS XL-MCL Analyzer. System II Data Acquisition and Display software, version 2.0, was used for data analysis.

Colony-forming assays. T4-2 and S-1 cells were propagated on two-dimensional culture as previously described (21). After reaching 50% to 70% confluence, culture flasks were exposed to IR (Sham, 2, 4, or 8 Gy). Six hours later, they were trypsinized and replated onto 35-mm per well dishes and allowed to grow for 10 to 12 days. Dishes were treated with crystal violet and colonies with >50 cells were counted. Six wells per condition were plated for n = 3 experiments. Plating efficiencies and surviving fractions were estimated based on binomial probability distribution.

Western immunoblot. To release cells from three-dimensional lrECM, cultures were first treated with ice-cold PBS/EDTA [0.01 mol/L sodium phosphate (pH 7.2) containing 138 mmol/L sodium chloride and 5 mmol/L EDTA] and then lysed in radioimmunoprecipitation assay buffer as previously described (21). Protein was aliquoted onto reducing SDS gels in equal amounts and separated using low voltage current. Protein bands were transferred onto nitrocellulose membrane (Invitrogen), and blots were blocked with 5% nonfat milk. Blots were probed with antibodies as listed below, then washed, incubated with secondary antibody, and exposed to X-ray film. Antibodies used included β₁ integrin, clone 18 (Oncogene); FAK, clone 77 (BD Transduction Laboratories); phospho-FAK, clone 14 (BD Transduction Laboratories); AKT, clone 7 (BD Transduction Laboratories); phospho-AKT (Cell Signaling); phospho-β₁ integrin (Biosource); TSC-2 clone c-20 (Santa Cruz Biotechnology, Santa Cruz, CA); phospho-Thr 1462-TSC-2 (Cell Signaling); β actin: clone AC-15 (Sigma); secondary enhanced chemiluminescence (ECL) anti-rabbit IgG/ECL, anti-mouse IgG (Amersham).

Tumor growth and toxicity assessment in vivo. Female BALB/c nu^−/− congenic mice were obtained from Simonsen Laboratories and kept in a controlled animal barrier at five per cage with water and chow as needed. After 1 week, animals were injected s.c. with 10⁷ MCF-7 cells into the posterior rear flank. Estradiol pellets were inserted s.c. above the tail for animals bearing MCF-7 xenografts. AIIB2 antibody or nonspecific rat IgG was injected into the i.p. cavity biweekly beginning on day 7 to 8 after cell implantation, before or after IR exposure. Before irradiation, animals were restricted in a Lucite container and the upper hemi-torso was shielded using 1.0-cm-thick cerrobend. IR was delivered to the exposed tumor and posterior hemi-torso using a 250 kVP source. Biweekly tumor dimensions (width, height, and depth) were recorded. At the time of sacrifice, animals were euthanized, and tumors were harvested and either immediately snap frozen or fixed in formalin. Animals were monitored for toxicity by measuring weight, assessing overall activity, and performing necropsy. All experimental procedures were received before approval by the Lawrence Berkeley National Laboratory Animal Welfare and Safety Committee.

Detection and quantification of Ki67 and TUNEL on in vivo tumor sections. We have developed methods to detect dual immunofluorescence of Ki-67 and TUNEL (adapted from Schafer and colleagues; ref. 26) in paraffin-embedded tissues that will allow simultaneous evaluation of the patterns of proliferation and apoptosis in situ. Five-micrometer sections from each tumor were baked in a 60°C oven for 30 min, dewaxed in xylene, and rehydrated through graded alcohols. Antigen retrieval was accomplished by heating slides in 10 mmol/L sodium citrate buffer (pH 6.0) for 45 min at 95 to 97°C. After cooling for 30 min at room temperature, sections were washed and then blocked in 10% goat serum/PBS for 1 h. Primary antibody against Ki67 (vector) is diluted in 10% goat serum/PBS, applied to the slide, and incubated overnight at 4°C. Slides were washed and incubated with Alexa Fluor 594–conjugated goat anti-rabbit IgG (Molecular Probes) and diluted in 10% goat serum/PBS for 40 min at room temperature. Slides are subsequently washed and labeled with DAPI. Primary antibody was withheld from negative controls.

After verification of staining, coverslips were removed and sections were washed thoroughly with PBS. All samples were permeabilized in freshly prepared 0.1% Triton X-100, 0.1% sodium citrate in H₂O for 10 min at room temperature. Slides were then blocked for 1 h in 3% bovine serum albumin, 20% fetal bovine serum, 0.1 mol/L Tris-HCl (pH 7.5) and washed with PBS. To detect apoptosis, we used In Situ Cell Death Detection Kit, Fluorescein (Roche). A TUNEL reaction mixture of label and enzyme solution was incubated on each slide for 1 h in a dark, 37°C incubator. Negative controls were incubated with label solution only. After washing in PBS, slides were labeled again with DAPI, mounted with Vectashield. Lymph node tissue was used as a positive control for both Ki67 and TUNEL staining.

After staining, fluorescent images were acquired on a Zeiss Axioplan 2 Imaging microscope and AxioCam camera. Five representative images will be obtained for each tumor using a ×40 oil objective and ×10 eyepiece for a total magnification of ×400. All final images consist of a composite of three different exposures: DAPI (nuclei), FITC (TUNEL), and Cy3 (Ki67).

MetaMorph software was used to quantify the number of positive versus total nuclei present in each image composite. For each image, a DAPI “mask” was created by accounting for all the nuclei from the DAPI exposure that fit within an established size-exclusion variable. Using the mask as a template, cells with an overlapping signal in rhodamine for Ki67-positive nuclei are classified as positive or negative according to a threshold for signal intensity that is predetermined by eye. The advantages of this automated counting system are increased accuracy and decreased observer subjectivity. The potential limitation is that the signal intensity threshold is set and those nuclei that have positive but dim staining was not included. The error is systematic and still small compared with other biases introduced with manual counting. To minimize this potential error, signal intensities are set using negative and positive controls for batches of slides stained at the same time using the same reagents. Thus, each batch is processed independently. TUNEL-positive nuclei were manually verified to account for the variation in morphology of apoptotic cells, including chromosomal condensation, blebbing, and cell shrinkage.

Confocal microscopy. A Zeiss LSM 410 inverted laser scanning confocal microscope equipped with an external argon/krypton laser was used to acquire immunofluorescence images from thin sections of three-dimensional lrECM cultures and breast cancer xenografts. A Zeiss Fluor ×40 (1.3 numerical aperture) objective was used, and images were captured at the colony mid-section for three-dimensional lrECM cultures and breast cancer xenografts. The relative immunofluorescence intensity of images was standardized by comparing only slides that were processed simultaneously under identical conditions.

Statistical analysis. For Ki-67 and TUNEL counts in three-dimensional lrECM, each culture condition (e.g., each dose of AIIB2 or IgG and each dose of IR in both sequences) was compared in pair-wise fashion and χ² statistics were calculated for significant differences between groups. A P value of <0.05 was considered to be an association that had a 95% probability of not being due to chance alone.

We estimated the numbers of animals needed for each experiment based on our prior experience with the MCF-7 xenograft models (19). The coefficient of variation for tumor volume using MCF-7 xenograft tumors is ∼0.4. Assuming equal sample sizes and a two-sample Student's t test, we estimated that eight mice per group were needed to have 80% power (with a type 1 error of 5%) to detect a difference of 60% in mean volume between groups. In our first experiment, we started with 14 animals per group with planned animal sacrifice at different time points for biochemical analysis so that at 4 wk, 8 animals per group would be remaining. Repeat verification experiments were done with n = 8 animals per group (data not shown). For tumor volume data, multivariate analysis of variance (MANOVA) was used for analysis at each time point. For each dose of AIIB2 or control IgG in vivo, a two-sided pair-wise Student's t test or χ² comparison was used to analyze differences between TUNEL and Ki-67 expression. MINITAB (Minitab, Inc.) statistical software was used for all calculations.

β₁ Integrin inhibition and IR are proapoptotic in malignant breast cell colonies and nonmalignant cells in two-dimensional, but not in acini-like structures in three-dimensional, lrECM. The original three-dimensional assay developed by Petersen and Bissell laboratories (26) consisted of breast cells suspended in a lrECM. Under these conditions, nonmalignant and malignant cells can be rapidly and robustly distinguished from each other (26). We modified this assay to use as a screen for novel therapies and showed that it could be used as a surrogate to discriminate between normal and malignant tissue response to β₁ integrin inhibitory antibody in vivo (19).

In the present study, we used the modified three-dimensional lrECM assay to compare apoptotic response of malignant colonies and nonmalignant acini to IR with and without β₁ integrin inhibitory antibody. We treated the preformed S1 and T4-2 structures with 8 Gy IR or 0.08 mg/mL β₁ integrin inhibitory antibody (AIIB2) alone or in combination, where AIIB2 was added to the culture medium for 24 hours and the medium was changed before IR exposure. The T4-2 cultures exhibited significantly more apoptosis compared with S-1 cultures (4.8% versus 0.75%, P < 0.05) after a single dose of 8 Gy, and the response was enhanced with the addition of AIIB2 (Supplementary Fig. S1B). In contrast, on standard two-dimensional tissue culture plastic, fluorescence-activated cell sorting analysis for TUNEL or Annexin V showed that both nonmalignant S-1 and malignant T4-2 cells had similar apoptotic responses to 8 Gy of IR (20% versus 14.2%, P > 0.05), which were both enhanced with the addition of AIIB2 (Supplementary Fig. S1A). Furthermore, cellular reproductive capacity following IR exposure (0–8 Gy) assessed by colony forming assays showed even more radiosensitivity in S1 compared with T4-2 cells (Supplementary Fig. S2).

β₁ Integrin inhibition administered after IR optimally enhances apoptosis associated with a down-regulation of Akt activity in three-dimensional lrECM. IR has been shown to up-regulate β₁ integrin expression in lung cancer cells in two-dimensional (nonmalignant cells were not tested in this study) and breast cell lines in three-dimensional lrECM (3, 5). Because we found previously that continuous AIIB2 treatment for 72 hours compared with 24-hour exposure was more effective in killing cancer cells, with little or no effect on S-1 colonies (19), we added AIIB2 to the medium each time the medium was changed until cultures were harvested (Fig. 1A). This protocol significantly increased apoptosis in T4-2 breast cancer cells in three-dimensional lrECM post-IR (>80% increase, P < 0.05; Fig. 1C) without significantly affecting S-1 nonmalignant acini (Fig. 1B).

Using two-dimensional cultures of small-cell lung cancer and MDA-MB-231 breast cancer cell lines (did not include normal controls), others have shown that β₁ integrin mediates resistance to apoptosis after cytotoxic insults, an effect mediated by a number of mechanisms, including up-regulation of Akt signaling (10, 13). Akt is a serine-threonine kinase that acts immediately downstream of PI3K as a mediator of cell survival post-IR (27–29). Akt activity was up-regulated following IR in T4-2 breast cancer cells in three-dimensional lrECM measured by the increase in phosphorylation of serine 473 on Akt (p-S473 Akt) using Western blots. Up-regulation post-IR (2-fold post 8 Gy IR) was dose dependent (Fig. 1E), indicating that IR induced an increase in Akt activity that could potentially be targeted by β₁ integrin inhibitory antibodies. To address this possibility further, we assayed three-dimensional lrECM cultures for levels of p-S473 Akt after exposure either to IR and control IgG antibodies or a combination of IR and AIIB2 with the latter applied both before and after IR. We found that down-regulation of Akt activity occurred when AIIB2 was administered post-IR (Fig. 1E). Interestingly, Akt activity after 6 hours of AIIB2 treatment alone did not decline once the cancer cells had already formed colonies (Fig. 1C).

To determine whether these findings were generalizable to other breast cancer cell lines, we used MCF-7 cells, a widely used model for breast cancer. In addition to determining the optimal sequence of therapy, we treated three-dimensional lrECM cultures with AIIB2 both before and after exposure to IR (Fig. 2A). Compared with colonies that were treated with IR and a nonspecific antibody control, colonies treated with combined IR and AIIB2 in either sequence showed an appreciable decrease in total cell numbers (data not shown) and a significant (>70% increase, P < 0.05) in apoptosis (assayed by TUNEL) compared with IR alone (Fig. 2B). The observed differences with untreated control were larger when IR was given before AIIB2; however, the data did not reach statistical significance to a level of 95% certainty in two of four experiments. We further verified the down-regulation in Akt activity associated with β₁ integrin inhibition post-IR (Fig. 2C).

IR-induced Akt effects could be overcome by application of β₁ integrin inhibitory antibody. To evaluate whether apoptosis induced by β₁ integrin inhibition or IR operated directly through Akt signaling, we used T4-2 cells transfected with a constitutively activated, myristoylated form of Akt (myr-Akt; refs. 30, 31; the construct was a kind gift of R. Roth; ref. 32) Fig. 3A. We measured apoptosis using TUNEL in T4-2 myr-Akt colonies after either IR alone or in combination with 0.08 mg/mL AIIB2—the dose used previously to induce apoptosis in several breast cancer cell lines in three-dimensional lrECM (19). In contrast to T4-2 vector controls, which showed increased apoptosis after IR plus 0.08 mg/mL AIIB2, T4-2 myr-Akt colonies were refractory to this treatment (Fig. 3B).

We reasoned that if Akt activity was responsive to β₁ integrin signaling, then increasing β₁ integrin inhibition could abrogate Akt signaling even in myr-Akt cells. A 3-fold increase in AIIB2 (0.24 mg/mL) elicited a 2-fold increase in apoptosis in T4-2 myr-Akt cells in an IR dose-dependent manner in three-dimensional lrECM (Fig. 3C). To determine if this apoptotic response correlated with down-regulation of Akt signaling, we measured p-S473 Akt and phosphorylated threonine 1462 TSC-2 (p-Thr1462 TSC-2), a downstream target of Akt activity (32). Concordant with the results observed in MCF-7 cells, Akt was up-regulated 6 hours after IR alone in T4 myr-Akt (Fig. 3D). Moreover, treatment with these higher doses of AIIB2 corresponded to down-regulation in p-S473 Akt and p-Thr1462 TSC-2. Taken together, these data strongly suggest that IR resistance is in part mediated by activation of β₁ integrin and a subsequent increase in Akt signaling both of which could be abrogated by β₁ integrin inhibition.

Combining β₁ integrin inhibition with IR treatment in vivo allows reduction of the effective dose of IR. Given the promising results in three-dimensional lrECM, we asked whether these data could be validated in vivo. Following implantation of MCF-7 cells into the rear flank of female athymic nu−/− mice (n = 14 per group), animals were subsequently randomized to receive either IR and control IgG, IR followed by AIIB2 or AIIB2 followed by IR, as described in Materials and Methods (Fig. 4A). Overall, the addition of AIIB2 to IR-enhanced tumor growth inhibition compared with either treatment alone (Supplementary Fig. S3). Remarkably, we found that animals that received 2 Gy followed by AIIB2 had similar tumor growth inhibition to those that received 8 Gy with control antibody (P < 0.014; Fig. 4B). We measured apoptosis at 24 hours after the final treatment in each group and found that tumors treated with 2 Gy plus AIIB2 also had a higher rate of apoptosis compared with other groups (data not shown).

To verify whether Akt activity was down-regulated by combined treatment as we had observed in three-dimensional lrECM, we assayed whole tumor cell lysates for evidence of Akt activity by probing for phosphorylation of S473 Akt. Again, the in vivo results replicated the findings in three-dimensional culture: We observed an increase in AKT activity associated with IR dose, which was down-modulated by the addition of AIIB2 after IR (Fig. 4C).

Here, we report promising results combining β₁ integrin inhibitory antibody with IR to decrease the total effective radiation dose and to maximize the efficacy of treatment by sensitizing the tumor, but not normal organs, to IR. Our increasing awareness of the global effects of IR on tissues, and the role of the microenvironment on modifying cellular response to IR (2, 33), are pointing to the necessity of reducing IR effective doses if at all possible by advancing our knowledge and providing novel therapeutic targets. β₁ Integrin plays a multifaceted role in cancer progression and resistance to cytotoxic treatment including IR. We have shown previously that the β₁ integrin inhibitory antibody, AIIB2, can revert the malignant phenotype of cells in three-dimensional lrECM (21) and enhance breast cancer cell death in xenografts while sparing nonmalignant acini in culture and normal organs in vivo (19). Here, we show that AIIB2 post-IR selectively enhances apoptosis in breast cancer cells but not nonmalignant acini in three-dimensional lrECM organotypic cultures, and that antibody combined with low doses of IR can achieve similar inhibition in tumor growth to single, but much larger doses of IR in vivo. In addition, we find that enhanced apoptosis after combined treatment is associated with down-modulation in Akt signaling, again both in three-dimensional cultures and in vivo. Importantly, when we assayed for apoptosis and reproductive death in standard two-dimensional tissue culture, we were unable to discriminate between malignant and nonmalignant cells. These data corroborate our previously published data on treatment with single chemical agents (19, 34) and indicate that the intrinsic radiosensitivity of both normal and cancer cells are modified also in the context of the three-dimensional structures. Importantly, we verify here that the three-dimensional lrECM assay is a useful and faithful surrogate for predicting differences in normal and malignant tissue response in vivo.

We had shown previously that IR induces heritable and global changes in cell-ECM interactions, and specifically in β₁ integrin expression, in human mammary epithelial cells cultured in three-dimensional lrECM. We also have observed activation of β₁ integrins in breast cancer cells post-IR,⁴

corroborating reports by others of integrin activation in irradiated tumors (4, 5, 35). We reasoned that if β₁ integrin was up-regulated by IR, then AIIB2 would be more effective post-IR than pre-IR because of increased epitope accessibility to antibody binding. To test whether there was an optimal sequencing of treatment, we used the three-dimensional lrECM assay where cancer cells were cultured to form tumor-like colonies and then treated with either AIIB2, 24 hours before IR or 24 hours post-IR. Both sequences resulted in significantly greater cell death (P < 0.05) compared with IR alone without significant toxicity to nonmalignant cells. Although there was a trend that favored sequencing IR followed by AIIB2, this did not achieve a statistical significance between the two treatment regimens over multiple repeated experiments. One of the likely reasons for this finding is that the magnitude of effect of AIIB2 alone was greater than a measurable difference attributed solely to the sequence of treatment. Importantly, AIIB2 was maintained continuously in the three-dimensional lrECM cultures as it would be also in vivo. Studies that examined the effect of sequencing PI3K inhibitor, LY294002, in breast cancer cells in two-dimensional culture (36) reported an optimal increase in apoptosis after combined IR and LY when LY was given both pre- and post-IR. However, the effects of the inhibitor and IR in nonmalignant cells were not reported. Another group (37) showed that LY294002 given pre-IR in HeLa cells was significantly better than post-IR. These results are not directly comparable with ours because the culture models and mechanism of action of Akt pathway inhibition are different in two-dimensional and three-dimensional models; however, they indicate that sequencing may play an important role in optimally delivering combined treatments based on the mechanisms of interaction of the inhibitor and IR and the context in which IR and the drug are delivered.

PI3K inhibitors have long been known to sensitize to, or work in combination with, IR to enhance apoptosis in several cancer cell types in cultured cell lines. Akt is emerging as a central mediator of resistance both from studies showing the radiosensitizing effect of PI3K inhibition (5, 38) and as a major effector of PI3K, the activity of which is modified by multiple cell surface receptors, including growth factors and integrins. In our three-dimensional lrECM model, AIIB2 post-IR resulted in down-regulation of Akt signaling, which was associated with enhanced apoptosis.

Our data complement those of others who showed that survival of bladder cancer cells harboring a mutation in ras was significantly decreased after combined treatment with LY294002 and IR, although these investigators reported no significant benefit of LY alone either in two-dimensional cultures or in vivo (16, 39). Others also have reported a synergistic effect of LY294002 and IR for glioblastoma multiforme xenografts (27). Although dose-limiting toxicity and pharmacokinetics limit LY294002 as a viable drug in humans, these studies highlight the importance of targeting Akt in conjunction with IR. In addition, Akt has been shown to be a promising target in cancer cells downstream of integrins and growth factor pathways post-IR (17). Thus, multiple pathways at the cell surface converge at Akt via PI3K, a central mediator of cancer cell survival, which may be targeted simultaneously to enhance therapeutic efficacy. Our data point to the existence of an IR-dependent survival pathway mediated by Akt that is down-modulated with β₁ integrin inhibition. It is quite possible that IR affects polarity pathways involving Rac 1 (32), which was shown also to affect radiation-induced apoptosis (40, 41).

Of note, β₁ integrin is ubiquitously expressed, and, thus, there are many potential targets for AIIB2 in vivo, in both malignant and normal tissue (42). We previously showed that AIIB2 was not toxic in vivo at doses up to 20 mg/mL/kg. Thus, although the antibody may recognize nontumor targets, our data indicated that the context of β₁ integrin expression, for example, within an organized structure, determined the apoptotic response to the antibody, as we showed in three-dimensional lrECM (19). In our present studies, we focused on whether the combination of IR and AIIB2 produced an antitumor growth effect and whether this therapy was toxic to animals. Further studies are warranted to explore other potential therapeutic targets for AIIB2 in malignant tissue, including neoangiogenic vasculature (42).

We found that the addition of AIIB2 treatment to IR in vivo resulted in significant improvement in inhibition of tumor growth by allowing a much smaller dose of IR (2 Gy) but reaching the same end point of higher doses of IR alone (8 Gy). To our knowledge, this is the first report of a β₁ integrin inhibitor used in conjunction with IR in a therapeutic context in vivo. There was a modest improvement in tumor growth inhibition associated with the addition of AIIB2 to 8 Gy compared with 8 Gy alone. However, improvement on 2 Gy of radiation was considerable. These data suggest that there may be an IR dose threshold above which the addition of AIIB2 has decreasing value in enhancing tumor cell death. We acknowledge that the doses and single fractions used in these experiments may not be the optimal dosing and sequencing in humans and that these data do not establish a synergistic relationship between β₁ integrin inhibition and IR; however, these data provide proof of principle that the combination of β₁ integrin inhibition and IR are a promising novel strategy for breast cancer radiotherapy.

Several questions remain unanswered and are currently undergoing active investigation by others and us. Clinically, fractionated IR is used to minimize normal tissue damage. It is unclear how β₁ integrin inhibitory agents may be optimized with fractionated therapy, and to what degree the total necessary IR dose may be modified to achieve maximal tumor control. A role for β₁ integrin– and Akt-mediated resistance post-IR is an active area of investigation. Our data indicate that there are alternate pathways that act in the presence of IR that could promote survival via Akt. Together, our findings show great promise for the combination of β₁ integrin inhibitory agents and IR and provide a rationale for combining multiple agents that target the Akt/PI3K pathway post-IR.

No potential conflicts of interest were disclosed.

Grant support: UCSF Clinical Investigator Research Program (C. Park) NIH P50 Specialized Programs of Research Excellence grant CA CA58207-08 (C. Park); the U.S. Department of Energy, Office of Biological and Environmental Research (DE-AC03-76SF00098; M.J. Bissell); NIH/National Cancer Institute (CA64786-09; M.J. Bissell); and an Innovator Award from the U.S. Department of Defense Breast Cancer Research Program (DAMD17-02-1-0438; M.J. Bissell).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Stephanie Kim and Richard Y. Lee for technical assistance.