Introduction
Breast cancer is one of the most prevalent cancers in women worldwide and it is estimated that a million women will develop this disorder each year. About 8% of breast cancer cases are inheritable, associated with mutations of highly penetrant breast cancer susceptibility genes, such as breast cancer-associated gene-1 and -2 (
BRCA1/2) and other tumor suppressor genes [
1‐
7]. In addition, it has been estimated that BRCA1 mutation carriers have a 50 to 80% risk of developing breast cancer before the age of 70 years [
8‐
11].
Both BRCA1 and BRCA2 play essential roles in many biological processes [
12‐
17]. A common feature of BRCA1/2-associated tumorigenesis is massive genetic instability, primarily due to the fact that cells lacking BRCA1 or BRCA2 have impaired ability to undergo homologous recombination (HR) [
15,
16,
18,
19], therefore these cells cannot effectively repair HR-mediated DNA damage, including DNA double-strand breaks (DSBs). The genetic instability often leads to altered expression of many genes and signaling pathways making it difficult to inhibit tumorigenesis and progression by targeting a single molecular target. Recently, significant work in the area of synthetic lethality has led to new approaches for the treatment of BRCA1/2-deficient cancers using high efficacy poly(ADP-ribose) polymerase 1 (PARP1) inhibitors (PARPi) with high efficiency [
20‐
24].
The PARP family plays important roles in DNA damage repair. For example, PARP1 is involved in the repair of DNA single-stranded breaks (SSBs) [
25‐
27]. Inhibition of PARP1 activity could also result in DSB formation when unrepaired SSBs meet the replication fork, causing its collapse. Because BRCA1/2 mutant cells are defective in repairing DSBs, PARPi inhibition may result in accumulation of DSBs in these cells and eventually lead to apoptosis. This may account for the molecular basis of why BRCA1- and BRCA2-deficient cells are extremely sensitive to PARPi [
20‐
22]. However, it was shown that several cell lines derived from mouse BRCA1 mutant mammary cancers [
28] and a human pancreatic BRCA2 mutant cancer cell line [
29] exhibited resistance to PARPi. While the exact cause for the resistance was unclear, it was hypothesized that some specific alterations/mutations might block the sensitivity of these cancer cells to PARPi [
30]. It was subsequently demonstrated that resistance to PARPi could occur through multiple mechanisms (reviewed in [
31]), such as impaired expression of 53BP1 [
32,
33], restoration of BRCA function [
34], and induction of P-glycoprotein expression [
35,
36]. To overcome the resistance, combinational therapies using multiple chemotherapeutic agents have been used to enhance the ability to kill BRCA1/2-related cancers [
35,
36].
In theory, all clinically used drugs have effects on biological systems other than those for which they were designed; therefore, drug repurposing consists of developing new applications for existing drugs. There has been an increase in the interest of drug repurposing due to the high cost of drug development and time involved in bringing new drugs to the market [
16,
37]. It has been estimated that it costs approximately more than USD 800 million to develop a new drug
de novo and the time estimated to develop a new drug that complies with the regulatory requirements for safety, efficacy and quality goes in the order of 10 to 17 years [
38].
In this study, a drug repurposing approach using the National Institutes of Health Chemical Genomics Center (NCGC) Pharmaceutical Collection (NPC) [
39], a library containing drugs approved for clinical use or that have been in clinical trials, was used to identify drugs that amplify the ability of AG14361, a potent PARP1 inhibitor [
21], to inhibit the growth of both human and mouse breast cancer cells, irrespective of their BRCA1 status.
Methods
Cell lines and viral vectors
Our initial study for human cell lines was performed in three isogenic models derived from the primary cell lines: 92 J, MDA-MB-231 (American Type Culture Collection, ATCC) and T47D (ATCC) and their BRCA1 mutant sublines 92 J-sh-BRCA1, MDA-MB-231-sh-BRCA1 and T47D-sh-BRCA1 respectively. The 92 J cell line, which is derived from a xenograft tumor of MDA-MB-231, forms mammary tumors much faster than the parent MDA-MB-231 cells when implanted into nude mice.
BRCA1 short hairpin RNA (shRNA) constructs in the pLKO.1-based vector were obtained from Open Biosystems (GE Healthcare, Little Chalfont, UK). A control lentiviral shRNA vector, packaging vector pCMV-dR8.2, and envelope vector VSV-G was obtained from Addgene (Cambridge, MA, USA). The BRCA1 shRNA construct (TRCN0000039837) was used to produce lentiviral particles for generation of stable BRCA1 knockdown cells. Lentivirus was produced in 293 T cells and the media collected for later transduction of target cells. Cells were transduced with lentiviral supernatant and then selected with 2 μg/ml puromycin to generate cells with stable knockdown of BRCA1. The viral supernatant was used to infect 92 J, MDA-MB-231 and T47D cells
Mouse BRCA1 mutant cell line 69 derived from mammary tumor of
Brca1Co/Co; MMTV-Cre;p53+/− mice containing a targeted deletion of full-length BRCA1 [
40] and BRCA1 wild-type (wt) cell line Ras, derived from mammary tumor MMTV-Ras mice [
41].
Growth assays
For the growth curve assay, 5 × 104 cells were plated per well of a six-well plate and medium was changed every 24 hr. The plate was incubated at 37°C in 5% CO2 and every 12 hr cells were detached by trypsinization and counted with a Z1 Coulter counter (Beckman Coulter, Brea, CA, USA). Plating for each time point was done in triplicate for each 92 J isogenic pair. In order to eliminate artifacts that could be produced by the cell line, we validated the other two pairs of human isogenic cell lines following the same protocol.
NCGC Pharmaceutical Collection (NPC) and quantitative high throughput screening
The NPC drug library consists of 2,816 small molecule compounds at the time of this screening [
39]. Fifty-two percent of the compounds are drugs approved for human or animal use by the United States Food and Drug Administration (FDA), 22% are drugs approved in Europe, Canada or Japan, and the remaining 25% are drugs approved in other countries or compounds that have been tested in clinical trials.
For the initial screening, the library was prepared as 15 inter-plate titrations, which were serially diluted 1:2.236 in dimethyl sulfoxide (DMSO, (Thermo Fisher Scientific, Waltham, MA, USA) in 384-well plates. The stock concentrations of the test compounds ranged from 10 mM to 0.13 μM. Transfer of the diluted compounds from 384-well plates to 1,536-well plates was performed using an Evolution P3 system (Perkin Elmer, Wellesley, MA, USA). Each treatment plate included concurrent DMSO and positive control wells and concentration-response titrations of controls, all occupying columns 1 to 4. Cell viability was measured using a luciferase-coupled ATP quantization assay of metabolically active cells (ATPliteTM 1step Luminescence Assay System, Perkin Elmer). Cells were dispensed at 2,000 cells/5 μL/well in 1,536-well white, solid-bottom assay plates using a flying reagent dispenser (FRD). The assay plates were incubated at 37°C for 5 hr to allow for cell attachment, followed by addition of 5 μl of compounds via pin tool. After compound addition, plates were incubated for 48 hr at 37°C. At the end of the incubation period, 5 μL of ATPlite reagent was added, plates were incubated at room temperature for 20 to 30 min, and luminescence intensity was determined using a ViewLux plate reader (Perkin Elmer).
Data analysis
Analysis of compound concentration-response data was performed as previously described [
36]. Briefly, raw plate reads for each titration point were first normalized relative to the positive control compound (-100%) and DMSO-only wells (0%) as follows: % activity = ((V
compound – V
DMSO)/(V
pos – V
DMSO)) × 100, where V
compound denotes the compound well value, V
pos denotes the median value of the positive control wells, and V
DMSO denotes the median values of the DMSO-only wells, and then corrected by applying a NCGC in-house pattern correction algorithm using compound-free control plates (DMSO-only plates) at the beginning and end of the compound plate stack. Concentration-response titration points for each compound were fitted to a four-parameter equation yielding concentrations of half-maximal activity (AC
50) and maximal response (efficacy) values. Compounds were designated as class 1 to 4 according to the type of concentration-response curve observed [
42,
43]. Curve classes are heuristic measures of data confidence, classifying concentration-responses on the basis of efficacy, the number of data points observed above background activity, and the quality of fit. Compounds with class 1.1, 2.1, 1.2 or 2.2 (>50% efficacy) curves are considered active. Compounds with class 4 curves are considered inactive and compounds with all other curves classes are considered inconclusive. Compounds that were selectively active (showed a potency difference of >3-fold) in one cell line or with or without the combination compound were selected for confirmation and follow-up studies.
Determination of synergistic effect and additive effect
The theoretical addictive effect of compounds with AG14361 was based on the fractional inhibition of these compounds when used separately. If the 50% inhibition concentration (IC50) of each drug is administered together, by the union of two events, the predictive addictive killing is calculated as Etotal = E1 + E2 – E1 × E2 (where E1 is IC50 of drug 1 and E2 is IC50 of drug 2), which is 75%. This classifies a drug synergistic if, when treated with the 50% inhibition dose of each drug, the synergistic killing effect should be significantly greater than 75%.
Cell proliferation assays for validation of synergistic effect
In order to validate the synergistic effect of the selected drugs in vitro we performed cell viability assay using a luciferase-coupled ATP quantization assay of metabolically active cells (ATPliteTM 1step Luminescence Assay System, Perkin Elmer) in a 96-well plate and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT). For MTT, 1 to 2 × 104 cells were plated per one well of a 24-well plate. Target drugs at various concentrations were dissolved in DMSO and then added to the cells in 10% fetal bovine serum-containing Dulbecco’s modified Eagle’s medium (DMEM), IC50 concentration of AG14361 were also added to each well. The final DMSO concentration was kept at 0.1% after the addition to medium. After 48 hr medium was removed and 0.3 ml of 0.1% MTT in phosphate-buffered saline (PBS) was added in each well. After incubation for 30 min in a 37°C CO2 incubator, MTT solution was removed and 0.8 ml of 2-propanol was added. After shaking for 30 min, OD560 was measured using a plate reader. Plating for each time point was done in triplicate.
Histological and immunohistochemical analysis of tumor samples
For immunohistochemistry procedures, the tumors were fixed in phosphate-buffered formalin, embedded in paraffin, cut in 4-μm thickness, and stained. Immunohistochemical analysis of proliferating cell nuclear antigen (PCNA) was performed using a labeled streptavidin-biotin technique described previously. Anti-PCNA monoclonal antibody PC 10 (Dako, Carpenteria, CA, USA), which reacts exclusively with nuclei, was used at a dilution of 1:200. The number of PCNA-positive cells was counted in five high-power fields (0.135 mm2 fields at × 200 magnification) selected at random, and the PCNA labeling index for each field was calculated as the percent of PCNA-positive cells (relative to the total). Apoptosis in tumor cells was detected using the terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick end-labeling (TUNEL) assay, as described previously. In the same manner as PCNA, five fields (0.135 mm2 fields at × 200 magnification) were selected at random, and the apoptotic index of each field was calculated as the percentage of TUNEL-positive cells.
RT-PCR and real-time PCR
Total RNA from cells or tissues were extracted with RNA STAT-60™ following the manufacturer’s protocol (Tel-Test, Inc., Gainesville, FL, USA), and cDNA was generated by Cells-to-cDNA™II (Ambion, Inc., Austin, TX, USA). Quantitative RT-PCR was performed using a SYBR green PCR Master Mix (Applied Biosystems, Carlsbad, CA, USA) and the 7500 Real-Time PCR system (Applied Biosystems). Primers used are listed below: BRCA1F 5′-ctgatgtgctttgttctgga-3′ BRCA1R: 5′-ggctatcctctaagagtgaca-3′, β-actinF: 5′gatggagttgaagtgagtttcgtg-3′, β-actinR 5-gcgggaaatcgtgcgtgcgtgacatt-3′, IL8F: 5′-aatctggcaaccctagtctgcta-3′, IL8R: 5′-aaaccaaggcacagtggaaca-3′, p50F: 5′-cagctcttctcaaagcagca-3′, p50R: 5′tccaggtcatagagaggctca-3′, MMP9F: 5′-cctgtgtgttcccgttcatct-3′, MMP9R: 5′-cgctggaatgatctaagccca-3′, COX2F: 5′-gaagtggggtttaggatcatc-3′, COX2R: 5′-cctttcactttcggataacca-3′, p65F: 5′-gcaggctcctgtgcgtgtct-3′, p65R: 5′-ggtgctcagggatgacgtaaag-3′, IL6F: 5′-cactgggcacagaacttatgttg-3′, IL6R: 5′-aaaataattaaaatagtgtcctaacgctcat-3′.
Luciferase reporter assay
The reporter plasmid, pNF-κB-luc, containing the κB-enhancer consensus sequences ((TGGGGACTTTCCGC) × 5) and nuclear factor κB (NF-κB)-dependent firefly luciferase gene was purchased from Stratagene (La Jolla, CA, USA). 92 J, and MDA-MB-231 isogenic cells were transiently transfected with two plasmids (pNF-κB-luc plasmid and renilla) using the LipofectAMINE 2000 Plus regent. Cells were seeded in a 24-well plate the day prior to transfection to achieve 80 to 85% confluence on the following day. Twelve hours after transfection, cells were incubated for an additional 24 hr in medium containing IC50 of AG14361, and lestaurtinib as mono-treatment and in combination and harvested for luciferase reporter assays. NF-κB transcription activity assay was also performed in a HeLa cell line, which carries a stably integrated luciferase reporter (Signosis, Inc., Santa Clara, CA, USA) after treatment with IC50 of AG14361 and/or lestaurtinib for 24 hr. Luciferase activity was measured with a luminometer using the Luciferase Assay System (Signosis). Renilla activity was detected to normalize any variations.
Caspase 3/7 activity
The 92 J isogenic cell lines were dispensed in culture medium at 2,000 cells/5 μl/well in 1536-well white/solid-bottom assay plates. The cells were incubated a minimum of 5 hr at 37°C. The compounds (23 nl/well) were added via the pin tool and then Resveratrol or AG14361 were added. The treated cells were incubated for 5 or 24 hr at 37°C, followed by the addition of the Caspase-Glo 3/7 (Promega, Madison, WI, USA) reagent at 5 μl/well. After 30 min incubation at room temperature, the luminescence intensity of the assay plates was measured using a ViewLux Plate Reader (Perkin Elmer).
Cell cycle analysis
After the 12, 24, 36 and 48 hr of drug treatment as mono-treament and combination, 92 J cells were trypsinized and washed twice in PBS (pH 7.4). Cells were fixed in 2 mL 70% ethanol (stored at −20°C), vigorously vortexed and incubated at 4°C for 4 hr. The cells were then washed with ice-cold PBS and resuspended in 200 μl PBS. Subsequently, cell suspension was incubated with 20 μl DNase-free RNase (10 mg/mL) and 1 ml of DNA intercalating dye PI (50 μg/ml, Triton-X 100 1.0%) at 4°C for 30 min. Cell cycle phase analysis was performed by flow cytometry using Epics-XL II FACS Caliber flow cytometer (Beckman Coulter, Brea, CA, USA), and data were analyzed by Multicycle AV software (Phoenix Flow Systems, San Diego, CA, USA).
Ninety-two J-shBRCA1 and 92 J-PLK cells after trypsinization were resuspended in PBS, the cells were injected into the fourth mammary fat pad on both sides of female nude mice at 1 × 106 cells/100 μl/spot. There were four groups of mice per cell line injected defined by the drug they were treated by. The first group was injected with PBS, second group with lestaurnitib, the third with AG14361, and the fourth with the combination of lestaurnitib and AG14361. Each group was formed of seven mice, with a total of twenty-eight mice per studied cell line. The drug treatment started when tumors became palpable (about 14 days after cell implantation). The mice were injected with AG14361 (30 mg/kg) intraperitoneally five times per week and lestaurtinib (10 mg/kg) three times per week, respectively. The recipient mice were housed in pathogen-specific facility, kept in a 12 hr light and dark cycle, and fed with a regular diet. Mice were monitored for tumor formation, and were euthanized when the tumors were 3.5 cm, which required harvesting, or there were tumor ulcerations. Tumor size was measured every three days before day 15 and every four days thereafter with a caliper, and tumor volume was calculated by using the formula V = 2/3πrxryrz (r is radius and x, y, z refer to each axis, and π = 3.14). All animal experiments were approved by the Animal Care and Use Committee of National Institute of Diabetes, Digestive and Kidney Diseases (ACUC, NIDDK).
Statistical analyses
All analyses were performed with the assistance of GraphPad Prism software version 4.0a (GraphPad Software, San Diego, CA, USA). A P value of less than 0.05 was considered statistically significant.
Discussion
The majority of BRCA1/2-related breast cancers exhibit high grade and are insensitive to most available hormonal or targeted therapeutic agents [
1‐
7,
48,
49]. Moreover many sporadic breast cancers also exhibit reduced or diminished expression of BRCA1 [
50,
51]. Thus the finding that BRCA1/2-associated breast cancers are highly sensitive to PARPi has been considered as a very promising approach for breast cancer [
20‐
24,
52]. However, like many other therapeutic agents, PARPi treatment is also associated with drug resistance after initial response to the treatment (reviewed in [
31]). For example, in phase II trials, 400 mg twice daily exposure of PARPi olaparib only resulted in the delay of breast cancer and ovarian cancer progression (both with median progression-free survival of about six months), and all patients with BRCA1/2 mutations eventually died of cancers [
53,
54].
In mouse models, prolonged treatment of the PARPi, AZD2281 also resulted in drug resistance presumably due to upregulation of Abcb1a/b genes encoding P-glycoprotein efflux pumps [
36]. Do Soto
et al. [
28] found that while a PARPi was able to kill naïve BRCA1 mutant cells with high specificity both
in vitro and
in vivo, it exhibited minimal specificity in inhibiting several cell lines derived from mouse Brca1 mutant mammary tumors. Altogether, these observations reinforced the need for screens for additional drugs that efficiently kill BRCA1/2-associated cancer cells when combined with PARPi. Of note, it has been demonstrated that PARPi, when combined with agents that impair DNA repair, are also effective in killing cancer cells containing wild-type BRCA1/2 [
36,
55,
56].
In this study, we screened a library that contains 2,816 small molecules, most of which are approved for human or animal use by the FDA or other countries [
39,
47], in the presence of AG14361 at a constant sublethal dose in order to identify compounds that kill breast cancer cells synergistically with PARPi. Our initial screen identified seventeen compounds that have similar levels in killing both shBRCA1 and wtBRCA1 cells, six compounds that are more specific for killing shBRCA1 cells, and nine compounds that kill wtBRCA1 cells better. After validation, lestaurtinib was selected for further investigation. Lestaurtinib is an orally bioavailable multikinase inhibitor for a number of kinases including protein kinase C-related kinase 1 (PRK1) [
57], FMS-like tyrosine kinase 3 (FLT3) [
44,
45], JAK2 [
46,
58], Trk-A/B/C [
59,
60]. Lestaurtinib has been used in clinical trials for myeloproliferative disorders, and acute myelogenous leukemia [
44‐
46], but there has been no report of its application for breast cancer treatment yet. Our data indicated that lestaurtinib is highly potent against tumor cells derived from both mouse and human breast cancers as a mono-treatment agent. In combination with AG14361, this effect is synergistically enhanced as reflected by further delay of tumor progression. We also found that four out of fourteen tumors completely regressed during combination treatment, while no regression was observed in the other three groups of mice (control, mono-treatment of either AG14361 or lestaurtinib) carrying a total of 42 tumors derived from each of cell line tested. Therefore, the synergy between AG14361 and lestaurtinib treatment is significant in these cancer cells. The complete tumor regression in these four groups of animals may reflect a differential threshold response of these mice to the treatment compared with the other recipients. In the clinic setting, complete tumor regression upon the therapeutic treatment is the most desirable outcome, however it does not always happen. In most cases, patients display partial response at different degrees, perhaps, due to individual difference in response to the treatment [
53,
54]. Nonetheless, the significant delay of tumor progression could prolong the life of patients and provide valuable time for further therapeutic therapies. Our data are reminiscent of this feature. We are in the process of screening further drug combinations in order to achieve the most desirable outcome in the near future.
The effect of lestaurtinib are primarily on G2 arrest, apoptosis and reduced proliferation of cancer cells irrespective of their BRCA1 status. Mono-treatment of AG14361 exhibited a similar, yet mild effect on apoptosis and proliferation; however, it affected all phases of the cell cycle. Of note, the combination of both drugs results in dramatic expansion of cells in the G2 phase at the expense of the S phase. This may account for the much more severe growth retardation and markedly enhanced cell death. Of note, we found that four out of eleven compounds, including lestaurtinib, which exhibits synergy with PARPi, could inhibit NF-κB activity based on our previous study [
47]. NF-κB is a transcription factor that plays important roles in cell cycle progression, cell survival and inflammation [
1,
52,
59,
60]. Therefore we tested the effect of AG14361 on NF-κB and found it could also inhibit NF-κB activity, albeit to a less extent compared with lestaurtinib. When combined together, AG14361 and lestaurtinib exhibited a much stronger inhibitory effect on the expression of a number of genes in the NF-κB signaling pathway, such as
p50,
p65,
IL6,
IL8,
COX2 and
MMP9 that are involved in cancer cell proliferation, inflammation, invasion and/or cell death [
45,
50,
54,
58,
59].
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
GVO designed and performed the experiments, analyzed and interpreted the data and wrote the manuscript. CC, SS and TJL performed the experiments. XX developed the cell lines. CL performed the mice surgeries, CJT synthesized the AG14361. RH and MHX designed the high-throughput experiments and performed the data analysis. CXD designed the experiments, interpreted the data and wrote the manuscript. CJT, MHX and CXD revised the manuscript for important intellectual content. All authors read and approved the final manuscript.