Inhibition of FoxM1-Mediated DNA Repair by Imipramine Blue Suppresses Breast Cancer Growth and Metastasis

Mammalian cells have evolved robust DNA damage surveillance mechanisms to maintain genome integrity. The defect in DNA damage response serves as a major factor predisposing normal cells to acquire oncogenic mutations (1). However, after tumor develops, malignant cells adapt various mechanisms such as repairing DNA damage resulting from unchecked DNA replication to manage their survival (2). Moreover, the ability of cancer cells to repair radiation or chemotherapy drug-induced DNA damage also serves as one of the mechanisms for therapy resistance. Consistent with that, several DNA repair factors are reported to be overexpressed in cancer patients (1, 3, 4). On the basis of these premises, the inhibition of DNA repair mechanisms has lately been considered to be a promising therapeutic target for treating cancers.

In this study, we report that a novel compound “Imipramine blue (IB)” may serve as a potent therapeutic agent as it induces DNA damage and inhibits the ability of breast cancer cells to repair DNA. IB is derived from imipramine, which is an FDA-approved drug for treating depression. IB was synthesized by a single-step process in which an electrophilic addition occurs at the ortho or para position to hydroxy or amine groups in the phenol or aniline compounds, or meta to nitro or carboxyl groups (5). IB was previously reported to inhibit invasion of glioma and head and neck squamous cell carcinoma (5, 6). However, IB's safety and viability to be used as a novel drug and its clinical relevance as well as utility in the treatment of breast cancers in particular is lacking. Most importantly, the mechanisms by which IB may inhibit breast cancer growth in particular are virtually unknown. Here, for the first time, using a clinically relevant ex vivo model of tumor explants from breast cancer patients, we show that IB inhibited breast cancer growth without any effect on normal mammary epithelial cells. In addition, the systemic delivery of IB suppressed breast cancer growth and metastasis in preclinical orthotopic mouse models without inducing any toxicity. Importantly, we report that IB inhibits breast cancer growth and metastasis by inhibiting homologous recombination (HR)–mediated DNA repair. Our results reveal that IB inhibits the levels and activity of DNA repair gene Forkhead Box M1 (FoxM1; ref. 7) and subsequently its transcriptional targets including S-phase kinase-associated protein 2 (Skp2; refs. 8, 9) and Exonuclease 1 (Exo1; ref. 10). Our interaction studies suggest that IB may affect the stability and transactivation function of FOXM1. Collectively, these findings indicate that IB may serve as a novel therapeutic lead compound with negligible toxicity for treating breast cancer patients. Furthermore, establishing the therapeutic potential of a compound that inhibits FoxM1, which is highly expressed and induces growth and progression of several cancers (11, 12), should exert much broader impact.

Breast cancer cells lines MDA-MB-231, MDA-MB-468, BT-549, MCF-7, and SKBR3 were purchased from the ATCC and cultured according to their guidelines. The cell lines were authenticated annually by using PCR for short tandem repeats.

For expression analysis and ex vivo explants, breast cancer tissues along with normal matched tissues were collected from Breast Cancer Clinic at UT Health Science Center San Antonio, TX (UTHSCSA) after obtaining approval (IRB #HSC20120041H).

FoxM1 cloning vector (pDNR-dual-FoxM1) was purchased from DNA repository at Arizona State University (DNASu, Arizona State University). FoxM1 insert was digested from pDNR-dual-FoxM1 vector and cloned in pCMV6 at EcoR1 and HindIII sites.

Breast cancer cells were seeded in 96-well plates at a density of 5 × 10³ cells per well and after 20 to 24 hours of incubation, cells were treated either with DMSO alone (0.02%, vehicle control) or with varying concentrations of IB (0.5–5 μmol/L) in DMSO for additional 24, 48, and 72 hours in CO₂ incubator at 37°C. Cell viability was assessed by using CellTiter-Glo (Promega Inc.) assay.

200,000 cells per well were plated in 6-well plates and after 20 to 24 hours of incubation, cells were treated either with DMSO alone or with varying concentrations of IB (1–5 μmol/L) in DMSO for another 24 hours. Next, 1,000 cells per well were re-seeded in 6-well plates for additional 7 days until colonies were clearly visible. Colonies were fixed with 4% paraformaldehyde and visualized by staining with 1% crystal violet and wells were scanned using scanner. Visible colonies were counted using image analysis software.

Breast cancer cells were pre-treated with IB at different concentrations for 24 hours and subjected to invasion and migration assays as described previously (13, 14). For rescue experiments, breast cancer cells were pre-treated with IB for 3 hours followed by FoxM1 expression for 72 hours and then were subjected to migration and invasion assays.

All animal experiments were performed after obtaining UTHSCSA-IACUC approval and the animals were housed in accordance with the UTHSCSA's protocol for animal experiments. For experimental metastasis, model, MDA-MB-231-GFP-Luc cells (2 × 10⁵) were injected into the tail vein of athymic nude mice. Starting from 8 days after tumor cell injection, animals were randomized into two groups. Group 1 animals received DMSO and group 2 animals received IB weekly once for 3 weeks via tail vein. Group 2 animals initially received 1.5 mg/kg body weight of IB in the first week and 3 mg/kg body weight of IB for the remaining treatment period. The Xenogen Small-Animal Imaging System was used for subcellular imaging in live mice once a week. After 36 days, animals were sacrificed and lungs were fixed and analyzed for metastatic foci. Liver, spleen, and kidney tissues were also fixed and analyzed for toxicity studies.

For orthotopic tumor xenograft assays, 2 × 10⁶ MDA-MB-231 cells were mixed with an equal volume of matrigel and implanted in a mammary fat pad of 6-week-old female athymic nude mice as previously described (15, 16). Once tumors reached measurable size, mice were divided into control and treatment groups. The control group received vehicle (DMSO), and the treatment groups received IB-conjugated with liposome-nanoparticle (3 mg/kg/body weight in 20% intralipid) intravenously twice a week for 3 weeks. Tumor volumes and body weight were measured twice a week. After the 30th day, the mice were euthanized, and the tumors were isolated and processed for molecular and immunohistologic studies. Tumor volume was calculated by using the formula 0.5236L₁ (L₂)², where L₁ is long axis and L₂ is the short axis of the tumor. At the end of the experiment, mice were sacrificed, and tumors were excised, weighed, and fixed in buffered formalin for further analysis.

Excised breast tumor and normal matched tissues were provided by a pathologist in accordance with UTHSCSA Institutional Review Board (IRB) approval (Control# HSC20120041H). Culture media consisted of RPMI-1640 supplemented with 10% FBS, 1% antibiotic/antimycotic solution, 0.01 mg/mL hydrocortisone, and 0.01 mg/mL insulin. A core piece of 8-mm tissue was dissected into 1 mm³ pieces and cultured on pre-soaked gelatin sponges (Johnson and Johnson, New Research) in 500 μL RPMI containing vehicle or nanoparticle conjugated-IB for 72 hours in 24-well plates. All explant tissues were cultured at 37°C in a 5% CO₂ incubator for 72 hours and were subsequently either formalin-fixed and paraffin-embedded or preserved for RNA isolation.

For IHC analysis, explants and xenograft tumor tissues were processed, paraffin-embedded and incubated with antibody against Ki-67 (#NB500, Novus) or subjected to apoptosis assay using the TUNEL Assay Kit (#G3250, Promega Inc.) as described previously (13). Ki-67– and TUNEL-positive cells were counted at ten arbitrarily selected fields at ×40 magnification. The proliferation/apoptotic index (per ×40 microscopic field) was determined as (number of Ki-67/TUNEL-positive cells × 100)/total number of cells.

Total RNA was isolated from MDA-MB-231 cells following treatment with vehicle and IB for 24 hours, respectively. RNA samples were further processed at the UTHSCSA Genomics Core for gene-expression profiling using Illumina Human HT-12 v4 Expression BeadChip following the manufacturer's standard protocol (Illumina). Gene-expression data were quantified and normalized (quantile normalization) using the BeadStudio software (Illumina). Gene-expression data are publicly available at NCBI/GEO (GSE72363).

Total RNA extracted from tumors, normal tissues, cell lines, xenograft tumors, and explants were subjected to qRT-PCR analysis and Western blot analysis as described previously (13). Supplementary Table S1 lists primer sequences for all genes studied in the present study. Antibodies against β-actin (A3854) and Aurora kinase A (T6199) were purchased from Sigma-Aldrich. Antibodies against FOXM1 (sc-502), XRCC3 (sc-271714), EXO-1 (sc-19941), and GADD45-A (sc-797) were purchased from Santa Cruz Biotechnology. Antibody against PLK (#627701) was purchased from BioLegend and against NOX4 (NB110-58849) was purchased from Novus. Calnexin antibody was kindly provided by Dr. Hima Bansal, UTHSCSA (San Antonio, TX). Other antibodies, including SKP2 (#4358), cyclin D1 (#2978), E2 (#4132), CDK2 (#2546), CDK4 (#12790), p-JNK (#9251), total JNK (#9252), and NFκB p65 (#8242), were purchased from Cell Signaling Technology. RAD51 (#70005) antibody was purchased from BioAcademia.

To determine DNA damage, immunofluorescence was performed with vehicle- and IB-treated cells using 53BP1 rabbit antibody (#A300-272AT; Bethyl Laboratories) as previously described (17).

The ISceI-based DR-GFP reporter assay was performed in cells following IB treatment (1, 2.5 or 5 μmol/L for 3, 12, or 24 hours) to evaluate frequency of double strand break repair by HR as described previously (18–20). For rescue experiments, cells were pre-treated with IB or vehicle control for 3 hours followed by transfection with control, FoxM1 expression plasmid and ISCeI expression vector, respectively, for 48 hours.

Both cell-cycle distribution and AnnexinV/PI-positive cells were analyzed using flow cytometry as described previously (21).

All values and error bars in graphs are means ± SEM; respective n values are indicated in figure legends; P values are determined by two-tailed Student t tests.

To address the effect of IB on breast cancer growth and progression, we first examined short- and long-term survival of IB-treated breast cancer cells. MDA-MB-231, MDA-MB-468, BT-549, SKBR3, and MCF-7 cells were treated with varying concentrations of IB (0–5 μmol/L) and cell proliferation was measured using CellTiter-Glo assay. IB treatment resulted in dramatically decreased breast cancer cell proliferation in a time and dose-dependent manner (Fig. 1A and Supplementary Fig. S1A). Interestingly, IB selectively targeted cancer cells without inducing any significant killing in normal human mammary epithelial cells (HMEC) that proliferate normally and maintain 24 hours doubling time (Fig. 1B). Next, we evaluated the long-term survival of IB-treated breast cancer cells. Indeed, pre-treatment with IB for 24 hours inhibited colony forming ability of breast cancer cells (Fig. 1C and Supplementary Fig. S1B). In addition to cell viability, IB treatment drastically reduced the migratory and invasive capabilities of the breast cancer cells (Fig. 1D and E and Supplementary Fig. S2A). To ensure that reduced migration and invasion of IB-treated breast cancer cells is not due to compromised viability, breast cancer cells were treated for a shorter duration (3 hours) and with lower concentration of IB that has minimum effect on cell viability followed by 12 hours of migration and invasion assays (Supplementary Fig. S2B). To further substantiate these findings, we performed Matrigel/3D invasion assay and showed that IB dramatically reduced the invasion of breast cancer cells (Supplementary Fig. S2C).

To test IB's tumor suppressor and metastasis suppressor-like activity in vivo and to address whether IB can be systemically delivered, we used two tumor xenograft models. First, we tested the efficacy of IB in an orthotopic xenograft model. MDA-MB-231 cells were implanted into the mammary fat pad of mice followed by treatment with specified concentrations of IB or vehicle control after a week when tumors reached measurable size. For systemic delivery of IB, we conjugated IB with liposome-nanoparticle (Nano-IB; 3 mg/kg body weight) and injected (intravenous) twice a week for 3 weeks. Interestingly, the systemic delivery of Nano-IB caused marked inhibition of mammary tumor growth when compared with vehicle control (Fig. 2A). To investigate whether IB can also inhibit tumor progression in vivo, we used a breast cancer experimental lung metastasis model, which recapitulates extravasation, seeding, and growth of metastatic process. MDA-MB-231 cells stably expressing GFP-luciferase reporters were injected through tail vein and mice were treated either with IB or vehicle control after 8 days when metastasis foci were first detected in the lungs. Intravenous (tail vein) injection of Nano-IB (1.5 mg/kg body weight of IB in the first week and to 3 mg/kg body weight of IB for the remaining treatment period for 4 weeks) into tumor-bearing mice resulted in the inhibition of metastasis, whereas vehicle-injected mice showed increased metastasis (Fig. 2B–D and Supplementary Fig. S3A). Although vehicle-treated mice showed aggressive multifocal metastasis outside of the vessel in the lung and with metastatic tissues attached to the diaphragm, IB-treated mice showed contrast effect as evident from the significant reduction or elimination of lung metastases (Fig. 2D). Importantly, analysis of tissues from IB-treated mice from both tumor models revealed that IB had no obvious toxicity because mice did not exhibit any weight loss and liver, kidney, spleen, or lung sections from the IB-treated group showed no evidence of toxicity (Fig. 2E and Supplementary Fig. S3B; a more detailed toxicity report is described in Supplementary Fig. S3 legend). These preclinical findings are highly encouraging as they suggest that IB may serve as a potent therapeutic for treating both primary and metastatic breast cancers with negligible toxicity.

Having established the therapeutic efficacy of IB in mouse models, we asked whether IB is indeed a safe and viable therapeutic alternative for treating breast cancer patients. To address this, we tested the efficacy of IB in ex vivo explants using tumor tissues collected from breast cancer patients. Our results show that ex vivo explants recapitulate the structural complexity and individual heterogeneity of human breast cancers, indicating that explants can be used to assess the therapeutic response of cancer treatment drugs (Fig. 3A). Interestingly, Nano-IB reduced the growth of tumor cells as revealed by dramatically reduced Ki-67 staining and increased apoptosis in explants from breast cancer patients (Fig. 3B and C and Supplementary Fig. S4). Importantly, IB did not affect the growth of the normal cells as revealed by comparable Ki-67 levels and absence of apoptotic cells in untreated and Nano-IB–treated explants from normal adjacent mammary tissues (Fig. 3B and C). These findings are highly significant as they indicate that explants recapitulate the microenvironment of human tumors (22), and therefore, can serve as a bona fide clinical model to determine therapy response in patients. Furthermore, it underscores the therapeutic potential of IB for treating breast cancers in most physiologically relevant milieu.

To understand the mechanism by which IB may inhibit breast cancer growth and progression, we performed gene-expression analyses on breast cancer cells treated with or without IB. Interestingly, gene ontology enrichment analysis using DAVID of differentially expressed genes revealed that genes involved in the cell-cycle and DNA repair pathways were highly enriched (Supplementary Table S2). Examples of those genes included FoxM1, cell division cycle 25 (Cdc25), stathmin1 (Stmn1), aurora kinase (Aurak), polo-like kinase 1 (Plk1), CyclinD1, Skp2, Exo1, and Xrcc3 that are known to play important role in cell-cycle progression and DNA damage surveillance or repair (7–10, 23–26; Fig. 4A and Supplementary Fig. S5A). To confirm our microarray results, we determined the target gene expression at the RNA and protein levels. IB treatment resulted in significant inhibition of target genes both at the RNA and protein levels (Fig. 4B and C). To further substantiate these findings, we determined the levels of IB-targeted genes in our tumor xenograft tissues and ex vivo explants from breast cancer patients. As expected, the levels of target genes were significantly reduced in IB-treated tumors and explants when compared with vehicle-treated tumors (Fig. 4D and E). We also determined the levels of these genes in breast cancer patients. Our analysis revealed that these genes are overexpressed in tumor tissues from breast cancer patients when compared with normal control tissues (Supplementary Fig. S5B).

Previous reports have shown that IB inhibits invasion of glioma and head and neck squamous cell carcinoma by reducing the levels of reactive oxygen species (ROS), NADPH oxidase 4 (NOX4), and NF-κB (5, 6). Therefore, in addition to target genes altered in our expression analysis, we determined the levels of ROS, NADPH oxidase 4 (NOX4), and NF-κB in breast cancer cells. Interestingly, we did not observe any significant difference in NF-κB levels in IB-treated breast cancer cell when compared with vehicle-treated cells (Supplementary Fig. S5C). Although IB treatment resulted in reduced ROS and NOX4 levels in some breast cancer cells, the effect was not consistently observed in all breast cancer cell lines (Supplementary Fig. S5D and S5E). Our findings suggest that NOX4-ROS-NF-κb signaling cascade may not be the predominant mechanism by which IB affects breast cancer growth.

Because several cell cycle–associated proteins were altered in IB-treated cells, we reasoned that IB may affect a specific phase of cell-cycle progression. Interestingly, IB treatment resulted in S-phase arrest when compared with control (Fig. 4F). The observed accumulation of cells in S-phase was accompanied by a concomitant decrease in G₁-phase cell population in IB-treated breast cancer cells (Fig. 4F). Consistent with this, FoxM1, which is known to regulate both G₁–S and G₂–M progression via several proteins, including cyclin D1, CDC25, AURAK, STMN1, and PLK1 (27), showed significantly reduced level in IB-treated breast cancer cells (Fig. 4C and G). To further substantiate these findings, we determined the levels of cyclin-dependent kinases (CDK) and other cyclins, which are the key regulators of cell-cycle progression. By qRT-PCR and Western blot analyses, we observed that IB treatment resulted in the inhibition of cyclin E2 as well as CDK2 and CDK4 (Fig. 4G and Supplementary Fig. S6A). In addition to CDKs and cyclins, we tested the levels of growth arrest and DNA damage-induced protein GADD45A, which is required for proper S-phase control and checkpoints (28). IB-treated breast cancer cells showed increased levels of GADD45A when compared with vehicle-treated cells (Fig. 4G). Because GADD45 is known to activate c-Jun N-terminal kinase (JNK)–dependent programmed cell death (29), we examined the effect of IB on JNK and apoptosis. Breast cancer cells treated with IB showed increased levels of JNK and apoptosis when compared with vehicle-treated cells as revealed by Western blot analysis and annexin V staining followed by FACS analysis, respectively (Fig. 4H and Supplementary Fig. S6B).

Given that GADD45a is known to be activated in response to damage, we hypothesized that IB may be inducing DNA damage, either directly or indirectly. To address this, we assessed the amount of DNA strand breaks induced by IB by determining the levels of p53-binding protein (53BP1) nuclear foci (30). Immunofluorescence analysis showed significantly increased number of 53BP1 foci in IB-treated breast cancer cells when compared with vehicle-treated breast cancer cells (Fig. 5A). Because IB can induce apoptosis, we reasoned that IB-induced DNA strand break may not be properly repaired. Considering the accumulation of cells in S-phase, we examined HR repair activity, a key DNA strand break repair pathway during S- and G₂-phases of the cell cycle (10). To address this, first, we assessed the kinetics of repair by examining the levels of RAD51, which is a FoxM1 transcriptional target (31) and is known to play critical role in HR (32). Western blot analysis showed significantly reduced levels of RAD51 in breast cancer cells treated with IB (Fig. 5B and Supplementary Fig. S6C) supporting the notion that IB may inhibit DNA repair by HR. To confirm these findings, we performed a functional assay to monitor HR events. We used ISceI-based DR-GFP reporter assay that measures the frequency of double strand break repair by HR (18–20). FACS analysis showed significantly reduced number of GFP-positive cells in IB-treated cells stably expressing DR-GFP reporter (Fig. 5C). Interestingly, we observed a time- and dose-dependent inhibition of GFP in IB-treated cells (Fig. 5C). These findings suggest that inhibition of FoxM1/FoxM1-associated signaling and HR-mediated DNA repair may be one the specific mechanisms by which IB affects growth and progression of breast cancers in particular and FoxM1-overexpressing cancers in general. Consistent with this, IB treatment resulted in reduced levels of FoxM1 and RAD51 and subsequently inhibition of long-term growth of lung, cervical, and prostate cancer cells that express high levels of FoxM1 (Supplementary Fig. S7).

To begin to address the mechanism by which IB may regulate its targets, we focused on FoxM1 because several IB target genes, including Skp2 (33), Exo1 (31), Rad51 (31), Stmn1 (34), and Aurak (27), are reported to be transactivated by FoxM1. The dramatically reduced expression of FoxM1 in IB-treated breast cancer cells prompted us to test whether IB may directly interact with FoxM1. We explored the surface of FoxM1 using SiteMap tool from Schrodinger suite to identify potential landing spots for IB. Interestingly, we found at least one potential binding site for IB. Our results reveal that residues Met296-Tyr317, which form a β-hairpin, are major contributors to the cavity, whereas residues Arg254-Arg256 complement its structure. The docking of IB showed considerable complementarity between shape of the cavity and topology of the IB (Fig. 6A). Our interaction model suggests that dibenzazepine part of the IB binds to the carboxylic group of Glu298, whereas the die moiety lands on side chains of Val296, Glu298, His311, Pro312, and the flexible dimethyl-amino tail of IB binds to the side chain carbons of Lys255 and Pro312. Supporting this interaction model, we observed that β-hairpin, which participates in the binding site, is directly connected to the α-helix H3 that binds to the DNA groove (Fig. 6A). Because β-hairpin exhibits flexibility, it is likely that specific binding of a small molecule (like IB) will constrain this flexibility. Given that interdomain interactions (DNA-binding forkhead domain, N-terminal auto-inhibitory domain, and a c-terminus transcriptional transactivation domain) are critical for FoxM1 stabilization and activity. Moreover, FoxM1–DNA interaction is reported to be weak and any minor adjustment to FoxM1 structure (such as interference with β-hairpin flexibility by IB) may affect both FoxM1 transactivation function as well its stability. There are two lines of evidences that further support the notion that interaction of IB with FoxM1 will result in FoxM1 transcriptional and translational instability. First, FoxM1 is reported to induce cyclin D1 transcription (27), whereas cyclin D1-CDK4 is reported to phosphorylate and stabilize FoxM1 protein (35). Therefore, it is likely that inhibition of FoxM1 by IB results in reduced cyclinD1-CDK4 levels (Fig. 4), which in turn affect FoxM1 levels. Second, FoxM1 transcriptional target Skp2 is reported to activate ATM via Mre11/Rad50/NBS1 (MRN) complex and ATM is known to transcriptionally activate FoxM1 (8, 9). Therefore, IB–FoxM1 interaction resulting in FoxM1 inhibition may result in Skp2 and subsequently ATM downregulation, which in turn may affect FoxM1 transcription. Indeed our expression and Western blot analyses revealed that Skp2, Nbs1, and ATM levels were significantly reduced in IB-treated cells (Fig. 4 and Supplementary Figs. S5A and S6D). To further substantiate these findings, we performed rescue experiments. Transfection of FoxM1 expression vector rescued the IB's inhibitory effect on HR-mediated DNA repair as revealed by ISceI-based DR-GFP reporter assay (Fig. 6B). In addition, FoxM1 expression rescued the migration and invasion of breast cancer cells pre-treated with IB (Fig. 6C and D). Though these results indicate a possible FoxM1–IB interaction, future in-depth structural and biophysical studies will be required to unambiguously establish IB binding to FoxM1 protein.

Our results reveal that IB may serve as a novel cancer treatment agent with negligible toxicity as it selectively inhibited breast cancer growth and metastasis without targeting normal mammary epithelial cells. We demonstrate that IB inhibits breast cancer growth and progression by regulating genes involved in DNA damage repair.

The presence of DNA damage from dysregulated growth control and upregulated DNA damage response are two common features that characterize cancer development as well as progression (2). Therefore, overexpression of factors that are capable of inducing both DNA damage and DNA repair responses may promote cancer cell survival and successful targeting of these factors will yield to promising therapeutic outcomes. One such factor is FoxM1, which is highly expressed in breast cancer patients (12) and is reported to regulate DNA damage response to support the uncontrolled proliferation associated with cancer (7, 36, 37). The advantage conferred by overexpression of FoxM1 by increasing DNA repair capacity is highlighted by the finding that cells depleted of FoxM1 have increased DNA strand breaks (31). Because FoxM1 is a transcriptional activator, it is likely that FoxM1 regulates a network of genes to mediate DNA damage sensing and repair in breast cancer cells (33). An example of such a gene includes Skp2, which is a transcriptional target of FoxM1 and is significantly downregulated in IB-treated cells. Skp2 plays a critical role in the cellular response to DNA damage and the maintenance of chromosome integrity by activating ATM via Mre11/Rad50/NBS1 (MRN) complex that results in the recruitment of ATM to the DNA damage foci to initiate HR-mediated repair (8, 9). Another factor that may play an equally important role in IB-mediated and FoxM1-dependent HR event is cyclin D1 (26), which is highly overexpressed in many human cancers, including breast cancers (26). Consistent with this, loss of cyclin D1 expression in cancer cells has been shown to inhibit the recruitment of RAD51 to damaged DNA and consequently inhibit the DNA repair by HR (38). Because Rad51 is a direct transcriptional target of FoxM1 (31), is functionally linked with Skp2 (39) and physically and functionally interacts with cyclin D1 (26), it is likely that the FoxM1–cyclin D1–Skp2–RAD51 signaling cascade is a central component mediating IB's effect on HR (Fig. 6E). In addition to FoxM1–RAD51/Skp2 signaling cascade, Xrcc3 is another gene that may mediate DNA repair events in IB-treated breast cancer cells as Xrcc3 is known to participate in HR to maintain chromosome stability and repair DNA damage (23). Importantly, increased Xrcc3 levels are associated with increased HR (25) and variants of Xrcc3 were recently shown to be associated with increased risk for breast cancer patients (40).

DNA repair and replication are closely linked processes (10); therefore, we wondered whether increased DNA strand breaks in response to IB may be due to aberrant replication and increased replication stress. Supporting this, our results revealed that Exo1, which is a FoxM1 transcriptional target (31), is significantly downregulated in IB-treated breast cancer cells. In addition to being involved in HR, EXO1 processes single-stranded DNA arising due to stalled and collapsed replication forks (10). Additional evidence for the involvement of replication errors following IB treatment comes from our observation that level of RAD51, which prevents replication error-associated DNA strand breaks and has been shown to promote restart/repair of stalled/collapsed fork (41), is significantly reduced in IB-treated breast cancer cells. Because DNA strand breaks often occur during S-phase as a result of replication fork collapse (10), which should be repaired before mitosis by HR during S- and G₂-phases, it is likely that reduced FoxM1 expression resulting in reduced RAD51 levels and S-phase arrest is causal in IB-induced inhibition of DNA repair process.

In summary, IB mediates its antitumor growth and metastasis functions by inhibiting critical DNA repair genes such as FoxM1 and subsequently HR-mediated DNA repair (Fig. 6E). FoxM1 is highly expressed in breast cancers (12) and are known to play a causal role in promoting breast cancer growth and metastasis. In addition, FoxM1 is known to transcriptionally regulate the expression of Vegf, which is an important angiogenic factor that promotes growth, invasiveness, and metastasis of several tumors (42–46). Therefore, pharmacologic inhibition of FoxM1 expression and its associated signaling is of significant clinical interest and holds great promise for treating cancer. Collectively, this study provides a strong rationale for developing IB as a novel therapeutic for treating cancer in general and breast cancers in particular and establishes several aspects that are needed for future drug development.

J.L. Arbiser holds ownership interest (including patents) in ABBY. No potential conflicts of interest were disclosed by the other authors.

Conception and design: S. Rajamanickam, B. Onyeagucha, R.K. Vadlamudi, A.J. Bishop, J.L. Arbiser, M.K. Rao

Development of methodology: S. Rajamanickam, S. Panneerdoss, B. Onyeagucha, J.L. Arbiser, M.K. Rao

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Rajamanickam, S. Panneerdoss, A. Gorthi, S. Timilsina, B. Onyeagucha, M.K. Rao

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Rajamanickam, S. Panneerdoss, A. Gorthi, D. Kovalskyy, D. Ivanov, M.A. Hanes, Y. Chen, M.K. Rao

Writing, review, and/or revision of the manuscript: S. Rajamanickam, S. Panneerdoss, B. Onyeagucha, D. Kovalskyy, D. Ivanov, R.K. Vadlamudi, Y. Chen, A.J. Bishop, J.L. Arbiser, M.K. Rao

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Rajamanickam, S. Panneerdoss, S. Timilsina, M.K. Rao

Study supervision: S. Rajamanickam, A.J. Bishop, M.K. Rao

The authors thank UTHSCSA microarray and GCCRI Bioinformatics core facilities for performing microarray analysis and bioinformatics analysis, respectively. The authors thank Dr. Luzhe Sun, UTHSCSA (San Antonio, TX) for MDA-MB-231-GFP-luc cells, Drs. Maria Jasin and Jeremy Stark for DR-GFP integrated U2OS cells and the pCAGGS vector with ISceI/GFP, Drs. Michael T. Drake and Zhenze Zhao at UTHSCSA for their help in the experimental metastasis experiment, and Hung I Harry Chen for microarray data analysis. The authors also thank Dr. Hima Bansal at UTHSCSA for Calnexin antibody, Dr. Paul Hasty at UTHSCSA for critically reading the manuscript, and Nourhan Abdelfattah for helping with the figures.

M.K. Rao is supported by NIH (NCI) grant R01CA179120-01A1, J.L. Arbiser is supported by NIG grant R01AR47901, Y. Chen is supported by NIH CTSA 1UL1RR025767-01. Y. Chen and M.K. Rao are supported by NCI P30 CA054174 grants and B. Onyeagucha is supported by CPRIT training grant RP140105.

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.