Assessment of Phenylboronic Acid Nitrogen Mustards as Potent and Selective Drug Candidates for Triple-Negative Breast Cancer

Abstract

Triple-negative breast cancer (TNBC) has limited treatment options and the worst prognosis among all types of breast cancer. We describe two prodrugs, namely, CWB-20145 (1) and its methyl analogue FAN-NM-CH₃ (2) that reduced the size of TNBC-derived tumors. The DNA cross-linking of nitrogen mustard prodrugs 1 and 2 was superior to that of chlorambucil and melphalan once activated in the presence of H₂O₂. The cellular toxicity of 1 and 2 was demonstrated in seven human cancer cell lines. The TNBC cell line MDA-MB-468 was particularly sensitive toward 1 and 2. Compound 2 was 10 times more cytotoxic than chlorambucil and 16 times more active than melphalan. An evaluation of the gene expression demonstrated an upregulation of the tumor suppressor genes p53 and p21 supporting a transcriptional mechanism of a reduced tumor growth. Pharmacokinetic studies with 1 showed a rapid conversion of the prodrug. The introduction of a methyl group generated 2 with an increased half-life. An in vivo toxicity study in mice demonstrated that both prodrugs were less toxic than chlorambucil. Compounds 1 and 2 reduced tumor growth with an inhibition rate of more than 90% in athymic nude mice xenografted with MDA-MB-468 cells. Together, the in vivo investigations demonstrated that treatment with 1 and 2 suppressed tumor growth without affecting normal tissues in mice. These phenylboronic acid nitrogen mustard prodrugs represent promising drug candidates for the treatment of TNBC. However, the mechanisms underlying their superior in vivo activity and selectivity as well as the correlation between H₂O₂ level and in vivo efficacy are not yet fully understood.

Medical treatments of fast-growing cancers apply chemotherapy and radiotherapy. Most chemotherapeutic agents in use today are cytotoxins that are not selective toward tumor cells and have a poor therapeutic index. It is well-known that targeting unique biochemical alterations in cancer cells is a feasible approach to achieve a therapeutic activity and selectivity. Numerous studies have indicated that cancer cells exhibit an elevated intrinsic oxidative stress.¹⁻³ Compared to normal cells, most cancer cells have increased levels of reactive oxygen species (ROS), such as superoxide, hydrogen peroxide (H₂O₂), and hydroxyl radicals.^2,4,5 The increased amount of ROS offers a therapeutic opportunity because it is intrinsic to cancer cells.^6,7 Some ROS-targeting strategies attempted to enhance ROS production to inflict lethal cell damage or trigger apoptosis.^8,9 Other anticancer agents inhibit enzymes that are essential to maintain the redox potential of the cell.¹⁰

Over the past decade, several ROS-sensitive functionalities have been identified, which offer an ROS-responsive prodrug approach to improve drug selectivity and efficacy. Because of the highly reactive and transient nature of ROS, an accurate quantification of physiological concentrations of these species has been difficult. However, higher levels of oxidative DNA base damages were observed in cancerous tissues than in cancer-free surrounding tissues.^1,11 Different ROS probes have been employed to measure the level of endogenous H₂O₂, the most stable ROS, which indicated that many different types of human tumor cell lines produced large amounts of H₂O₂.^2,12−15 It has been estimated that the concentrations of H₂O₂ are up to 100-fold higher in cancer cells than healthy cells.^2,12−15 This has guided the development of H₂O₂-activated anticancer prodrugs to improve the selectivity and specificity of the cytotoxic agents toward cancer cells. Boronic acids and boronate esters selectively react with H₂O₂ and make an ideal target for prodrug activation.¹⁶ A number of arylboronic acid- and ester-based prodrugs have been developed, including prodrugs of DNA alkylating agents,¹⁷⁻²³ aminoferrocene analogues,²⁴⁻²⁶ estrogen receptor regulators,^27,28 antitumor nitric oxide,^29,30 and theranostic prodrugs.^31,32 Many of these prodrugs showed in vitro cytotoxicity and selectivity toward cancer cells. However, few reports have presented in vivo efficacy and selectivity.^20,24,32

DNA interstrand cross-linking (ICL) agents block DNA replication and/or transcription, leading to an inhibition of cell division and eventually resulting in cell death.^33,34 They are among the most frequently used antitumor agents in the clinic, such as cisplatins, mitomycin C, nitrogen mustards, and psoralens.³⁵ However, traditional DNA ICL agents induce serious side effects due to their poor selectivity toward cancer cells. Recently, our group developed two classes of H₂O₂-activated DNA cross-linking agents, including quinone methide precursors^17,18,23 and nitrogen mustard analogues,¹⁹⁻²² that contained a boronic acid or boronate ester as an ROS-responsive trigger unit (Figure 1).¹⁹⁻²³ The electron-withdrawing property of boronates reduced the toxicity of the DNA alkylating group while allowing its selective activation by H₂O₂ (A → B) (Figure 1a). These compounds spared normal cells but showed a selective cytotoxicity toward cancer cell lines, including chronic lymphocytic leukemia (CLL) and breast cancer cells with oxidative stress.^{4,6,9,36−40} For example, H₂O₂-activated ICL agents selectively killed CLL cells isolated from patients but not normal lymphocytes from healthy donors.^19,20,23 Some H₂O₂-activated ICL agents were more effective than clinically used DNA alkylating agents chlorambucil and melphalan.^19,20 More importantly, these reagents were particularly effective against triple-negative breast cancers (TNBC) that lack an expression of estrogen receptor alpha (ERα), progesterone receptor (PR), and HER2 for therapeutic targeting.²⁰ TNBC is difficult to treat and has the worst prognosis among all types of breast cancer. Efforts to find effective treatments for patients with TNBC have been ongoing for decades.

Figure 1.

Structures of antitumor agents.

All previous observations suggest that H₂O₂-activated DNA cross-linking agents are selective anticancer prodrug candidates for TNBC treatment, although a detailed mechanism for the ROS activation of these prodrugs in cancer cells has not yet been determined. It was reported that TNBC cells exhibit intrinsically higher H₂O₂ levels in association with the downregulation and decreased bioactivity of catalase.¹⁵ On the one hand, H₂O₂ production regulates the growth of aggressive breast cancer cells. On the other hand, normal human breast epithelial cells have a low basal ROS output and normal metabolic regulation. Since aromatic nitrogen mustard 1 (CWB-20145) and its methyl analogue 2 (FAN-NM-CH₃) exhibited the best biological and physicochemical properties, we investigated their pharmacokinetics and in vivo therapeutic efficacy and selectivity using a cell line-derived xenograft model of TNBC (Figure 1b).

Materials and Methods

Reagents and Drugs

Compounds 1 and 2 were synthesized as previously described.^19,20 To increase the purity, compound 2 was precipitated after column chromatography using dichloromethane (DCM) and hexane (v/v, 1:24). The purity of both compounds was confirmed by nuclear magnetic resonance (NMR), high-resolution mass spectrometry (HRMS), and liquid chromatography (Supporting Information Figures S1–5). The purity was greater than 97% as determined by an ACQUITY CSH C18 column (2.1 mm × 50 mm, 1.7 μm particle size) using the following gradient: 0–2.0 min 30%–60% MeCN in A, 2.0–3.0 min 60%–90% MeCN in A, 3.0–4.0 min 90% MeCN in A, 4.0–4.3 min 90%–30% MeCN in A, 4.3–6.0 min 30% MeCN in A, at a flow rate of 0.5 mL/min (Solution A: water) with 254 nm UV detection. Chlorambucil and melphalan were purchased from Sigma-Aldrich and directly used without any further purification.

Synthesis of 2

A solution of 4-bromo-N,N-bis(2-chloroethyl)-3-methylaniline (5 g, 16.1 mmol) in dry tetrahydrofuran (THF) (100 mL) was cooled to −78 °C under Ar. n-BuLi (25.6 mL, 2.5 M in hexane) was added slowly at −78 °C within 10 min. After 30 min, B(OⁱPr)₃ (10.6 g, 56.4 mmol) was added to the solution. The reaction mixture was allowed to warm to rt, stirred for another 6 h, quenched with NH₄Cl solution (saturated (sat)) at 0 °C, then extracted with DCM (75 mL × 3), washed with water, dried over Na₂SO₄, and concentrated under vacuum. The residue was purified by column chromatography (10:1 → 1:1 hexane/ethyl acetate) to afford a crude product. The crude product was dissolved in DCM (5.0 mL) in a 200 mL round bottle flask, followed by a slow addition of hexane (120 mL) with stirring, kept at rt overnight, and filtered to afford 2 as a white solid (3.0 g, 10.9 mmol, 67.7%). ¹H NMR (300 MHz, deuterated dimethyl sulfoxide (DMSO-d₆)/D₂O = 9:1): 7.39–7.37 (d, J = 6, 1H), 6.49–6.47 (d, J = 6, 2H), 3.70 (s, 8H), 2.36 (s, 3H). ¹³C NMR (75 MHz, DMSO-d₆/D₂O = 9:1): 147.57, 144.66, 136.44, 113.39, 108.61, 52.34, 41.77, 23.30. HRMS-ESI (+) (m/z): [M + H]⁺ calculated for C₁₁H₁₆NO₂BCl₂, 276.0726; found: 276.0732.

Detection of DNA Interstrand Cross-Linking

ICL formation and cross-linking yields were analyzed via a denaturing poly(acrylamide) gel electrophoresis (PAGE) with phosphorimager analysis. The DNA–DNA cross-linking capabilities were investigated using a ³²P-labeled 49mer DNA duplex (Figure 2A). The ³²P-labeled oligodeoxynucleotide (ODN) 3a (1.0 μM) was annealed with 1.5 equiv of the complementary strand 3b by being heated to 90 °C for 3 min in a buffer of 10 mM potassium phosphate (pH 7) and 100 mM NaCl, followed by cooling to rt. The ³²P-labeled ODN duplex (2 μL, 1.0 μM) was mixed with 1.0 M NaCl (2 μL), 100 mM potassium phosphate (2 μL, pH 8), 10 mM H₂O₂ (2 μL), and compounds in DMSO at the desired concentrations (0.2, 0.5, or 1.0 mM). Autoclaved distilled water was added to give a final volume of 20 μL. The reaction mixture was incubated at rt for 22 h, followed by quenching with an equal volume of 90% formamide loading buffer (90% formamide, 0.5% ethylenediaminetetraacetic acid (EDTA), 0.1% xylene cyanol, and 0.1% bromphenol blue). The resulting mixture was subjected to a 20% denaturing PAGE analysis.

Figure 2.

(A) DNA sequence (duplex 3) used for a cross-linking study. (B) The cross-linking reaction of ³²P-labeled 49-mer DNA duplex (3) and 10 μM, 100 μM, 200 μM, 500 μM, or 1 mM chlorambucil, melphalan, 1, and 2 with or without H₂O₂ (1.5 equiv of drug) at 25 °C for 22 h. All DNA ICL yields were obtained by triplicate experiments and shown as average ± standard deviation.

Cell Culture

The human tumor cell line MDA-MB-468 was purchased from the American Type Culture Collection. UO31, A498, SN12C, 786-0, TK-10, and CAKI-1 cells were purchased from the National Cancer Institute (NCI). MDA-MB-468 was cultured in L-15 Leibovitz media (HyClone No. SH30525.01, 500 mL) supplemented with fetal bovine serum (FBS, MIDSCI No. S01520HI, 50 mL), nonessential amino acids (NEAA 100X solution, HyClone No. SH30238.01, 5.0 mL), and penicillin and streptomycin (HyClone Penicillin Streptomycin 100X Solution, HyClone No. SV30010, 5.0 mL) at 37 °C with 100% air. Other cells UO31, A498, SN12C, 786-0, TK-10, and CAKI-1 were grown in RPMI 1640 medium (HyClone No. SH30027.01, 500 mL) supplemented as described for the L-15 media. Cells were grown at 37 °C and 5% CO₂ at 100% relative humidity.

Cytotoxicity Assay

Dose Response

Cells were plated into 384-well optical bottom plates (Nunc: 142762) in 20 μL at densities ranging from 5000 to 10 000 cells/well. The plates were incubated for 3 h prior to the addition of the compounds. Compounds were solubilized in dimethyl sulfoxide (DMSO) at 20 mM and serially diluted 10–12 times, decreasing 50% each time, in DMSO in a 384-well plate. 200 nanoliters of the serially diluted compounds was added to the cell plate (1:100 dilution) using a Tecan Freedom EVO liquid handling system equipped with a 100 nL pin tool (V&P Scientific). Plates were incubated for an additional 48 h followed by the addition of 20 μL of Celltiter-Glo Reagent (Promega). Luminescence was measured after 30 min of incubation at rt with an Infinite M1000 (Tecan) plate reader. Time response. The cell line (MDA-MB-468) was transferred to a 96-well plate (Nunc: 165306) that was pretreated with matrigel. The total number of cells was 3000 per well in 100 μL (3 × 10⁴ per mL). A solution of 2 or 4 mM compound in DMSO (0.5 μL) was added to each well to reach a final concentration of 10 or 20 μM, respectively. CellTiter-Glo reagent (100 μL) was added to the first row of the wells. The luminescence was then measured for day 0. Next, the plate was incubated at 37 °C for 24 h with 100% relative humidity. CellTiter-Glo Reagent (100 μL) was added to the second row of the wells, and the luminescence was measured for day 1. The procedure was continued for the other rows, over 5 d. A time-response curve was generated by plotting the percentage of viable cells against the time using GraphPad Prism5 software.

Hydrogen Peroxide Detection

The H₂O₂ level was determined using an Amplex Red Hydrogen Peroxide/Peroxidase Assay (Invitrogen, A22188) per the manufacturer’s protocol. Briefly (25–50) × 10³ cells were seeded in a 96-well black/clear bottom plate (Thermo Scientific Cat. No. 165305). Cells were treated with 20 μM 2, 250 μM N-acetyl cysteine (NAC), or a mixture of 20 μM 2 and 250 μM NAC for 24 h. A no-treatment control was also included. After 24 h of incubation, the cells were washed twice with 1X KRPG buffer (Krebs–Ringer phosphate consists of 145 mM NaCl, 5.7 mM sodium phosphate, 4.86 mM KCl, 0.54 mM CaCl₂, 1.22 mM MgSO₄, and 5.5 mM glucose, pH 7.35) and incubated for 5 h in 100 μL of KRPG buffer. Then, 50 μL of the standard curve samples, controls, and cell samples was transferred into another 96-well plate in triplicate and mixed with an equal amount of Amplex Red reagent (50 μM Amplex Red and 0.1 U/mL horseradish peroxidase (HRP) final concentrations). After 5 h of incubation at rt in the dark, the fluorescence was measured (Ex/Em: 560/590 nm) on an infinite M1000 microplate reader (Tecan). Each value in the figures represents a mean of triplicate samples. Note that the final concentration of H₂O₂ for each experimental sample would be twofold lower as the same volume of amplex red reagent was added.

Experimental Animals

Six-week old female CD1 mice (Charles River Laboratory) were used for a safety and pharmacokinetics (PK) study. Immune-deficient female nude mice (Charles River Strain, Code 490) weighing 20–22 g were used for an in vivo efficacy study. The animals were housed under specific pathogen-free conditions, under standard conditions of humidity, temperature, and a controlled 12 h light and dark cycle, and had free access to food and water. All animals were allowed a period of adaptation (∼5 d) before experimental procedures. All animal experiments were in compliance with the University of Wisconsin–Milwaukee Institutional Animal Care and Use Committees (IACUC).

Safety Study

The maximum tolerated dose (MTD), defined as the highest dose not causing a serious adverse event (e.g., death, convulsion, ataxia, aberrant behavior, or evident pain) observed within 2 d of observation, was determined for 1 and 2 with female CD1 mice using groups of three animals per group. Compounds were formulated in a mixture of DMSO, poly(ethylene glycol) (PEG) 400, and phosphate-buffered saline (PBS) (volume ratio 2:19:19). The volume for an intraperitoneal (IP) injection was 100 μL. Approximately 18 mice were used with escalating IP dosages until serious adverse events were observed or the maximum dosage was reached (100 mg/kg). Once the dosing was completed, animals were observed for another 2 d to observe delayed-onset toxicity effects. Animals with the following signs were euthanized: weight loss of 20% from the initial weight or more, the inability to rise, ambulate, or reach food and water for over 3 d, and the presence of a labored respiration. To identify a safe dose of 1 and 2 for an in vivo efficacy study, decreased doses of compounds (IP injection) were given to the female CD-1 mouse (three mice for each dose) each day until a dose was administered with no signs of weight loss for all mice over a period of 5 d.

In Vivo Efficacy Study with Xenograft Models

Immune-deficient female nude mice were anesthetized with isoflurane and injected subcutaneously with cancer cells (MDA-MB-468) suspended in a 1:1 solution of matrigel and Dulbecco’s Modified Eagle Medium (DMEM) media. All cancer cells were obtained from the American Type Culture Collection (ATCC) and were negative for bloodborne pathogens. Cell numbers for each inoculation (100 μL per mouse to the subcutaneous area of the flank) was 5 × 10⁶. Animals were monitored daily for palpable tumors, and animal weights were recorded weekly before the compound was administered. When the tumors reached treatment size (200 mm³), the mice were randomized to treatment groups (11 per group). A compound or the control (vehicle) was given as single IP doses each day for seven weeks. The compound was formulated as specified for the safety study. The maximum volume of IP injection was 100 μL at a concentration of 5.0 mg/kg for compounds 1 or 2. Briefly, mice with palpable tumors were treated with a formulated compound in PBS/PEG400/DMSO (19:19:2) or control (11 mice per group). Mice were then weighed, and tumor sizes were measured using electronic calipers every 7 d. At the end of the study period, all tumors were harvested, weighed, and stored in −80 °C.

Microsomal Stability Assay

A master mix containing 282 μL of deionized water (18.2 mΩ), 80 μL of potassium phosphate buffer (0.5 M, pH 7.4), 20 μL of NADPH Regenerating System Solution A (Corning Life Sciences No. 451220), 4 μL of NADPH Regenerating System Solution B (Corning Life Sciences No. 451200), and 10 μL of human or mouse microsomes (with a final microsome concentration of 0.5 mg/mL) was preincubated at 37 °C for 5 min. Following the preincubation, 4 μL of test compound (1 mM in DMSO) was added for initiation of the reaction, and the reaction time was recorded. The reaction mixture was incubated at 37 °C, while aliquots of 50 μL of the reaction mixture were retrieved at the time intervals of 0 (without compounds), 10, 20, 30, 40, 50, and 60 min. Each aliquot was added to a vial containing 100 μL of ice-cold acetonitrile with 3 μM internal standard (compound 1 was used as the internal standard for compound 2, and vice versa), immediately followed by sonication for 10 s, and centrifugation at 10 000 rpm for 5 min. Then, 100 μL of the supernatant was transferred to Spin-X HPLC filter tubes (Corning Incorporated No. 8169) and centrifuged at 13 000 rpm for 5 min. 50 microliters of the filtrate was diluted 20-fold by liquid chromatography-mass spectrometry (LCMS) grade methanol (950 μL) in a 2 mL glass auto sampler vial (MicroSolv No. 95025-WCV) and analyzed by triple quadrupole LC-MS/MS (Shimadzu 8040). All samples were stored on ice until the MS analysis. The ratio of the peak areas of the internal standard and test compound was calculated for every time point, and the natural log of the ratio was plotted against time to determine the linear slope (k). The half-life equals 0.693/k. The experiment was repeated six times.

Plasma Pharmacokinetics and LC-MS/MS Analysis

CD-1 female mice (∼20–22 g) were dosed IP with 1 or 2 (10 mg/kg) formulated in PBS/PEG400/DMSO (19:19:2). Blood was collected by a cardiac puncture from three mice for each time point into tubes containing 50 μL of heparin (1 mg/mL in water). The samples were stored in liquid nitrogen until the analysis.

Blood samples were thawed on ice and vortexed for 10 s. A 100 μL aliquot of the blood sample was added to 300 μL of cold methanol containing 133.3 ng/mL of internal standard (IS). Samples were then vortexed for 30 s and centrifuged at 14 000 rpm for 5 min. Next, the supernatant layer was spin-filtered through a 0.22 μm nylon membrane centrifugal filter unit (Costar). Separation was accomplished using an ACQUITY CSH C18 column (2.1 mm × 50 mm, 1.7 μm particle size) under a flow rate of 0.6 mL/min, column temperature at 40 °C, with 254 nm UV detection, a mobile phase of 0.1% formic acid (v/v) (A) and MeCN (B), and a gradient time program of the following: 35% B (0) → 35% MeCN (0.5 min) → 40% B (1.0 min), hold at 40% B (1.5 min) → 90% MeCN (2.0 min), hold at 90% B (2.5 min), return to 35% B (3.0 min), hold at 35% B (4.0 min). Analytes were monitored under a positive mode using electrospray ionization (ESI). The following transitions are monitored in multiple reaction monitoring (MRM) mode. Ion pairs for 2 were m/z 275.90 > 164.10, m/z 275.90 > 213.10, and m/z 275.90 > 146.10. Transition ion pairs for 1 (IS) were m/z 262.10 > 150.10, m/z 262.10 > 199.10, and m/z 262.10 > 132.10. The collision energy was optimized for each transition to obtain the optimal sensitivity. The mass spectrometer was operated with the heat block temperature of 400 °C, drying gas flow of 15 L/min, desolvation line temperature of 250 °C, nebulizing gas flow of 2.0 L/min, and both needle and interface voltages of 4.5 kV.

Preparation of RNA Samples and RT-qPCR

MDA-MB-468 cells were incubated with 1, chlorambucil, or melphalan at 10 μM for 48 h in a six-well plate (1 M cells per well). The cells were harvested and resuspended in 350 μL of RLT buffer in the presence of 1% β-mercaptoethanol. The cells were lysed with QIAshredder (Qiagen) spin columns, and the total RNA was isolated and purified using an RNAeasy kit (Qiagen) and quantified by absorbance spectroscopy. Polymerase chain reaction (PCR) amplification and quantitation were performed using a Quantifast SYBR Green real-time (RT) PCR kit (Qiagen) following the manufacturer’s recommendations. The primers used in this these studies were as follows: GAPDH forward primer (FP) 5′-ACCACAGTCCATGCCATCAC-3′, GAPDH reverse primer (RP) 5′-TCCACCACCCTGTTGCTGTA-3′; P21 FP 5′-GGA AGA CCA TGT GGA CCT GT-3′, P21 RP 5′-GGC GTT TGG AGT GGT AGA AA-3′; P53 FP 5′-GTTCCGAGAGCTGAATGAG-3′, P53 RP 5′-TTATGGCGGGAGGTAGACTG-3′. Real-time PCR was performed on a Mastercycler (Eppendorf). cDNA synthesis and amplification were performed in a 96-well twin.tec real-time PCR plate (Eppendorf No. 951022015) with a volume of 20 μL comprised of 10 μL of SybrGreen Mix 2x, 4.8 μL RNase-Free water, 1 μL of each primer, 0.2 μL of quantifast RT, and 3 μL of RNA (50 ng) for each reaction. The target specificity of the assays was validated by a melting curve analysis. The expression of each gene was normalized relative to GAPDH expression levels for each sample. The expression of each gene relative to an untreated control was calculated using the 2^–ΔΔCt method of Livak and Schmittgen. Finally, standard deviations (SDs) were calculated from three biologically independent experiments performed in triplicate.

Results

Concentration-Dependent DNA Cross-Linking in the Presence of 1 and 2

Nitrogen mustards induce cell death by cross-linking DNA. Previously, we reported a DNA cross-linking of 1 and 2 in the presence of H₂O₂ at a concentration of 1 mM.^19,20 Therefore, we used a ³²P-labeled 49-mer DNA duplex (Figure 2A) and visualized an ICL formation and cross-linking with a denaturing PAGE gel and phosphor imager analysis. Herein, we incubated 10 μM, 100 μM, 200 μM, 500 μM, and 1 mM chlorambucil, melphalan, 1 or 2, and 1 or 2 with H₂O₂ at 25 °C for 22 h. The autoradiograms are presented in the Supporting Information (Figure S6). The ICL percentages were determined for each reaction in triplicate and are depicted in Figure 2B.

Without an addition of H₂O₂, 1 and 2 induced lower ICL yields than chlorambucil and melphalan, indicating that 1 and 2 are less reactive toward DNA. Even with the highest concentration (1.0 mM), less than a 2.5% ICL yield was observed for 1 and 2, which is less than those induced by chlorambucil or melphalan at the same concentration (5.7% or 4.9%, respectively). In contrast, the addition of H₂O₂ greatly increased the cross-linking capability of 1 and 2. For example, 1 induced 51.4% ICL, while 2 induced 53.4% ICL in the presence of H₂O₂, which is an ∼8–10-fold increase as compared to the ICL produced by chlorambucil or melphalan. Similar results were observed for all five concentrations tested.

Compounds 1 and 2 Are More Cytotoxic than Chlorambucil or Melphalan in Human Breast Cancer and Renal Cancer Cell Lines

Previously, the cytotoxicity of 1 was tested against 60 human cancer cell lines at the National Cancer Institute (NCI Developmental Therapeutics Program), which suggested that 1 showed a significant cytotoxicity in several cancer cell lines.¹⁹ Here, we determined the cytotoxicity of 1, 2, chlorambucil, and melphalan with breast cancer MDA-MB-468 cells and renal cancer cell lines including UO-31, A498, SN12C, TK-10, and CAKI-1. The results are depicted in Figure 3.

Figure 3.

Cellular toxicity of 1, 2, chlorambucil, and melphalan in (A) MDA-MB-468, (B) UO-31, (C) A498, (D) SN12C, (E) CAKI-1, and (F) TK-10 when incubated for 48 h (n = 4, IC₅₀ values were determined by a nonlinear regression). (G) Time-dependent toxic response of MDA-MB-468 cells when incubated with 10 μM of 1, 2, chlorambucil, or melphalan (n = 4). (H) Time-dependent toxic response of MDA-MB-468 cells when incubated with 20 μM of 1, 2, chlorambucil, or melphalan. The significance was determined by one-way ANOVA followed by a Tukey test to compare all pairs of columns (n = 4, (*) P < 0.05, (***) p < 0.0001.

The dose-dependent toxicity response after 48 h in the presence of compound concentrations ranging from 0.195 to 200 μM showed that 1 and 2 are more toxic than chlorambucil and melphalan. For example, the half-maximal inhibitory concentration (IC₅₀) of 1 for MDA-MB-468 is 16.7 μM, which is half that of chlorambucil (IC₅₀ = 34.4 μM) and one-third that of melphalan (IC₅₀ = 48.7 μM). Similarly, the IC₅₀ of 1 for UO-31 is 38.8 μM, while the IC₅₀ of chlorambucil and melphalan were higher than 100 μM. A similar result was obtained with A498, SN12C, TK-10, and CAKI-1 cells. More importantly, the introduction of a methyl group further improved the potency of 2. For example, 2 is four fold more potent than 1 for the MDA-MB-468 cell line (IC₅₀ = 3.1 μM for 2 vs 16.7 μM for 1). Similarly, lower IC₅₀ values were observed for 2 than 1 in other cell lines, including A498, SN12C, TK-10, and CAKI-1.

The time-dependent cytotoxicity of 1 and 2 was assessed with MDA-MB-468 over the period of 4 d (Figure 3G and Supporting Information Figure S7). The cells were treated with 10 or 20 μM of 1 and 2, respectively. More than a 50% toxicity was observed after a 2 d treatment of 1 at 10 μM (43% viability), and only a 28% viability was observed for 20 μM 1 in the same time period. A 4 d treatment of 1 led to a 93% toxicity. Furthermore, it was observed that MDA-MB-468 cells are more sensitive to 2 than to 1. For example, only a 42% viability was observed for the MDA-MB-468 cells after a 24 h exposure of 2 (20 μM), while a 77% viability was observed for 1 (20 μM). Similarly, 10 μM 2 led to a 65% viability at 24 h and 29% at 48 h for the MDA-MB-468 cells, which was lower than results obtained with 10 μM 1 (79% at 24 h and 43% at 48 h). Finally, our findings indicate that the cytotoxicity of 1 and 2 is time-dependent and exhibits a long-lasting effect on the viability of cancer cells.

MDA-MB-468 Cell Produced a High Level of H₂O₂

Having established that the MDA-MB-468 cell was particularly sensitive toward these ROS-activated prodrugs, we investigated the correlation between the efficacy and H₂O₂ level. Previously, Sen and coauthors reported that several breast cancer cell lines, including the MDA-MB-468 cell line, showed greatly increased H₂O₂ production rates than normal human breast epithelial cells, which is associated with the downregulation and decreased bioactivity of catalase in TNBC cells.¹⁵ Increased H₂O₂ production leads to an increased proliferation of aggressive TNBC cells. Here, we determined H₂O₂ level produced by MDA-MB-468 cells using the Amplex Red hydrogen peroxide assay kit (Invitrogen, A22188).⁴¹ To quantify the H₂O₂ released from MDA-MB-468 cells, we first prepared a H₂O₂ standard curve (Supporting Information Figure S10). Approximately 2.0–2.5 μM H₂O₂ was detected with (25–50) × 10³ MDA-MB-468 cells after being incubated at 37 °C for 5 h (Figure 4A) (Note: the H₂O₂ concentration of the cell samples is two-fold higher due to the prepared cell samples mixed with an equal amount of Amplex Red reagent before the assay). Given that the H₂O₂-activated prodrugs can be converted to active analogues in the presence of H₂O₂, we postulated that blocking the ROS production by NAC would inhibit the cytotoxicity induced by these agents. To test our hypothesis, we incubated MDA-MB-468 cells with 2 in the presence or absence of NAC and measured the end points such as cytotoxicity. On the one hand, our data showed that, in the presence of NAC (250–500 μM), the apoptosis induced by 2 (20 μM) was slightly decreased (Figure 4B). On the other hand, a much lower H₂O₂ level was detected with the cells treated with 2 alone or with 2/NAC (Figure 4A).

Figure 4.

(A) H₂O₂ level released from MDA-MB-468 cells as measured by Amplex Red (data are represented as mean ± SD from three independent experiments) (Note: as the prepared samples mixed with an equal amount of Amplex Red reagent before the assay, the final H₂O₂ concentrations are two-fold lower). The significance was determined by one-way ANOVA followed by Dunnett to compare all columns (n = 6), (*) P < 0.05, (***) p < 0.0001 vs control (Amplex Red only). (B) Toxic response of MDA-MB-468 cells when incubated with NAC (250 μM), 2 (20 μM), 2 (20 μM) + NAC (250 μM), and DMSO (control). Data are represented as mean ± SD from three independent experiments. The significance was determined by one-way ANOVA followed by Dunnett to compare all columns (n = 4), (*) P < 0.05, (***) p < 0.0001 vs 20 μM 2.

In an effort to provide direct evidence for the selective transformation of a boronic acid prodrug to the phenol analogue, we used mass spectroscopy to analyze the transformed adducts obtained from a cell culture. Approximately 1–2 million cells were treated with 20 μM prodrug 2 and incubated at 37 °C for 2, 4, 6, 12, and 24 h. Then, the cell culture media and cells were separated, and the cells were trypsinized and then extracted with MeOH and analyzed by mass spectroscopy. Compound 2 was detected with the samples incubated for 2, 4, and 6 h but disappeared after 12 h of incubation, which suggested that 2 was completely transformed to other products (Supporting Information, Figures S11–S14). However, we could not detect the activated phenol form using mass spectroscopy. There are several possible reasons for an unsuccessful detection of the activated phenol form: (a) the phenol form might be too active to detect; (b) the conversion yield might be low due to the H₂O₂ level in cells being lower than that used in DNA cross-linking study; (c) other transformation products are possible, including hydrolysis, reactions, or association with cellular metabolites and/or biomolecules within the cells. All these would make the detection of prodrug transformation more difficult. Collectively, these data suggest that the MDA-MB-468 cell does produce a high level of H₂O₂ and that 2 may function through ROS-dependent mechanisms, but the detailed mechanism of function has not been fully understood yet.

Compounds 1 and 2 Reduced the Viability of Cancer Cells by Apoptosis Via Caspase 3/7

The ApoTox-Glo assay (Promega) measures viability, cytotoxicity, and apoptosis in the same sample well, which serves as an especially useful tool to better understand the mechanism of cellular cytotoxicity (https://www.promega.com/-/media/files/resources/protocols/technical-manuals/101/apotox-glo-triplex-assay-protocol.pdf?la=en).⁴² The assay simultaneously measures the activity of live-cell protease and dead-cell protease. A cell-permeant substrate (glycyl-phenylalanyl-aminofluorocoumarin (GF-AFC)) is used for measuring the live-cell protease activity, while a fluorogenic cell-impermeant peptide substrate (bis-alanylalanyl-phenylalanyl-rhodamine 110; bis-AAF-R110) is used to measure the activity of dead-cell protease released from cells that have lost membrane integrity. Furthermore, the assay measures the amount of caspase 3/7 activity using a luminogenic caspase-3/7 substrate. Caspase-3 and caspase-7 are two of the major effector caspases involved in the execution phase of apoptosis and are responsible for the breakdown of several cellular components involved in DNA repair and regulation.^43,44 MDA-MB-468 cells were exposed to different concentrations of 2 or chlorambucil for 6 h. ApoTox-Glo Triplex Assay was added to assess apoptosis and cytotoxic effects. All measurements were conducted on the same sample according to the manufacturer’s protocol. The results are depicted in Figure 5. Graphs with individual measurements can be found in the Supporting Information (Figure S8). No concentration-dependent cytotoxicity was noticed in the presence of 2 or chlorambucil for the range of 0.39–200 μM. Exposure of MDA-MB-468 cells to 2 or chlorambucil, however, led to a dose-dependent increase in caspase-3/7 activity. Because of this apoptotic effect, a dose-dependent decrease of cell viability was observed.

Figure 5.

Comparison of the viability, cytotoxicity, and apoptosis of 2-treated MDA-MB-468 cells (A) and chlorambucil-treated MDA-MB-468 cells (B).

In Contrast to Chlorambucil, 1 and 2 Did Not Show Adverse Effects at 80 and 100 mg/kg in Mice

The toxicity of 1 and 2 was further evaluated in vivo in comparison with chlorambucil. The initial 1 mg/kg IP dosage was escalated until serious adverse events were observed or the maximum dosage of 100 mg/kg was reached. The results are summarized in Table 1.

Table 1. Mouse Survival Rate after Drug Treatment^a.

	1 d treatment (mg/kg)					3 d treatment (mg/kg)
compounds	20	40	60	80	100	10	20	30	40	50
1	3/3	3/3	3/3	3/3	n.d	3/3	3/3	n.d.	3/3	3/3b
2	3/3	3/3	3/3	3/3	3/3	3/3	3/3	3/3	3/3	3/3b
chlorambucil	3/3	3/3	3/3	0/3	n.d	3/3	3/3	n.d	2/3b	0/3

^aThree mice were administered 1, 2, or chlorambucil IP in PBS/PEG400/DMSO (19:19:2).

^bSymptoms of toxicity were observed, including weight loss, loss of appetite, and reduced activity levels observed for survival mice.

The single-dose-treated mice survived at a maximal tolerated dose of 80 mg/kg (1) and 100 mg/kg (2). Chlorambucil, however, induced death at 80 mg/kg for all animals. After it was demonstrated that ROS-activated prodrugs 1 and 2 are less toxic than chlorambucil, a repeated-dose toxicity study was conducted. Chlorambucil induced death at a 40 mg/kg repeated dose on day 3. All mice treated daily with 50 mg/kg 1 or 2 survived. Thus, ROS-activated prodrugs 1 and 2 showed a better safety profile than chlorambucil.

To identify a safe dose for an in vivo efficacy study, three groups of mice were treated with vehicle [PBS/PEG400/DMSO (19:19:2)], 1, or 2, at doses of 5.0, 10, and 20 mg/kg. Mice were treated daily via IP for one week, and body weights were measured daily. The results are depicted in Figure 6.

Figure 6.

Changes of mice body weight after a 5 d treatment with 1 (A) and 2 (B) at doses of 5.0, 10.0, or 20.0 mg/kg. The significance was determined by two-way ANOVA (n = 3, ns P > 0.05, (**) P < 0.01, and (***) p < 0.001 vs control group).

A significant weight loss was observed for a 10 and 20 mg/kg dose of 1 at the last day of administration (Figure 6A). No obvious weight loss was observed for mice treated with 2 at all three doses (5.0, 10, and 20 mg/kg) (Figure 6B). More importantly, a slight increase of body weight was observed for mice treated with 5.0 mg/kg of 1 or 2. Thus, a dose of 5.0 mg/kg was used for the following in vivo efficacy study.

Compounds 1 and 2 Reduced Tumor Size in MDA-MB-468-Derived Xenograft Mice

To investigate the in vivo efficacy of 1 and 2, the human breast cancer xenografts were established by a subcutaneous implant of MDA-MB-468 cancer cells in athymic nude mice. Tumors developed in all mice within one week. Groups of eight mice were treated with vehicle, 1 (5 mg/kg), or 2 (5 mg/kg). The size of tumors and the weight of mice were measured weekly. The data are presented in Figure 7.

Figure 7.

In vivo evaluation of a daily treatment of 1 and 2 in athymic nude mice. (A) Picture before and after treatment. (B) Time-dependent tumor growth measured by caliper. (C) Time-dependent body weight changes. (D) The mean of tumor weights at the end of treatment. (E) Appearance of animals after treatment with vehicle, 1, or 2. (F) Appearance of animal organs after treatment with vehicle, 1, or 2. The mice were administered IP with vehicle, 1, or 2 at a dose of 5 mg/kg. Data are expressed as mean ± SD (n = 8), (***) p < 0.001 vs control group.

Compounds 1 and 2 not only inhibited the tumor growth but reduced the tumor size after four weeks (Figure 7A,B). For drug-treated mice, tumor sizes decreased to 5%–50% of the initial size after seven weeks of treatment (Figure 7B). In contrast, the vehicle-treated mice showed a significant tumor growth, reaching 213%–1194% of the initial tumor size (Figure 7B). The average tumor volume of MDA-MB-468 xenografts after seven weeks of treatment with 1 (77.0 ± 17.4 mm³) was only 9% of the mean volume in the control group (861.2 ± 362.8 mm³). The calculated tumor growth inhibition rate [IR (%) = [1 – (mean volume of treated tumors)/(mean volume of control tumors)] × 100] was 91%. Similarly, 2 greatly inhibited the tumor growth in vivo, evidenced by 94.6% of IR and 80% of tumor shrinkage induced by 2. After seven weeks of treatment, the mean volume for 2-treated tumors was only 46.3 ± 17.7 mm³, that is, 5.4% of the mean volume of the control tumors.

The tumors were excised and weighed on the first day of week eight. The weight of tumors in the control group ranged from 100 to 370 mg, while those of 1-treated mice were in the range of 10–30 mg (Figure 7D and Supporting Information Figure S9). The average tumor weight for 1-treated mice was 16.3 ± 7.4 mg in comparison to 245 ± 103.6 mg for vehicle-treated mice. Similarly, a greatly reduced tumor size was observed for 2-treated mice (17.3 ± 7.4 mg vs 245 ± 103.6 mg for the control group). The results demonstrated that 1 and 2 effectively inhibited the tumor growth for MDA-MB-468 xenografts with a dosage of 5.0 mg/kg.

The three groups of athymic nude mice under an in vivo efficacy study were also monitored for symptoms of toxicity including changes in body weight, loss of appetite, reduced activity levels, treatment-related mortality, and changes (color and weight) in kidney, liver, spleen, and heart. Our observation suggested that the IP administration of 1 or 2 (5 mg/kg) for seven weeks (five injections per week) induced neither animal death nor weight loss. The 2-treated mice showed a similar increase in body weight as compared to the control group (Figure 7D). In addition, 1- or 2-treated mice did not show a loss of appetite, reduced activity levels, or color changes for lung, liver, spleen, heart, and kidney (Figure 7F). Together, the in vivo investigation demonstrated that a treatment with 1 and 2 potently suppressed the tumor growth without affecting normal tissues in the mice.

Quick Transformation of Prodrugs 1 and 2 In Vitro and In Vivo

The conversion of prodrugs 1 and 2 was investigated in mice as well as in the presence of liver microsomes (Figure 8). Initially, we investigated the conversion of 1 and 2 in mice over a period of 2 h (Figure 8A,B). The blood concentrations of 1 and 2 were quantified by LC-MS/MS following a single intraperitoneal injection of 10 mg/kg (Figure 8A,B and Supporting Information Table S1). Both compounds (1 or 2) were readily detected in plasma with a t_max of ∼3 min. The presence of a methyl group in 2 greatly increased the half-life to 8.84 min from 4.92 min for 1. The rate of transformation for 1 in the blood was fast (0.141 min^–1), which is 2 times faster than that of 2 (0.078 min^–1). Compound 2 showed an area under the curve (AUC) of 16 253 ng·min/mL, which is higher than that of 1 (10 883 ng·min/mL). Further in vitro microsomal stability studies revealed that prodrugs 1 and 2 are significantly more stable in a human (t_1/2 = 83.67 min for 1 and 59.38 min for 2) than in a mouse (t_1/2 = 16.26 min for 1 and 23.49 for 2) (Figure 8C).

Figure 8.

PK profile of prodrugs 1 and 2. (A) PK analysis of 1 in CD-1 mice following a single IP dose of 10 mg/kg. (B) PK analysis of 2 in CD-1 mice following a single IP dose of 10 mg/kg. (C) Stability of prodrugs 1 and 2 in the presence of human and mice microsomes. All experiments were conducted twice in triplicate.

ROS-Activated Prodrugs Regulated Oncogenes in TNBC

To understand the possible downstream signal transduction of ROS-activated prodrugs, a gene regulation in the presence of 1 was investigated. MDA-MB-468 cells were treated with 1, followed by a messenger RNA (mRNA) extraction and quantitative real-time polymerase chain reaction (qRT-PCR). The mRNA levels of specific genes for 1-treated cells were compared with those treated with chlorambucil and melphalan. The expression of two genes p21 and p53 were quantified after 48 h. p53 is one of the most frequently mutated tumor suppressors in human cancers that participated directly in the intrinsic apoptosis pathway. Chlorambucil was reported to induce cell cycle arrest and cellular apoptosis via the accumulation of cytosolic p53.⁴⁵p21 is tightly controlled by the tumor suppressor protein p53, which is an important tumor suppressor transcription factor that mediates apoptosis in response to DNA damage or other major cellular disruptions. The results are depicted in Figure 9. A strong upregulation of p53 and p21 was observed for 1-treated MDA-MB-468 cells. The data indicated that the ROS-activated prodrug 1 was able to significantly induce p53 expression that, in turn, activated expression of p21 and inhibited cell cycle progression. The gene regulation effected by chlorambucil and melphalan was similar but less pronounced at the same concentration.

Figure 9.

Regulation of genes involved in apoptosis and cell cycles. MDA-MB-468 cells were treated with the indicated compounds at 10 μM for 48 h, and mRNA levels of p21 and p53 were quantified by qRT-PCR. The y-axis indicates the relative gene expression against the expression levels of the untreated control being set at 1.0. Data are expressed as mean ± SD (n = 3), (***) p < 0.0001 vs control group.

To gain a broader understanding of gene regulation by 1, we conducted a microarray analysis of mRNA extracted from MDA-MB-468 cells using an Affymetrix whole human genome expression array. With a cutoff value of less than −2 and greater than 2, a total of 13 genes were found to be upregulated, and 62 genes were downregulated (Supporting Information Table S2). Table 2 shows all upregulated genes and the top 20 most downregulated genes and their expression levels. Among these differentially expressed genes, several upregulated genes (e.g., p53, ANKRD1,⁴⁶SERPINB,⁴⁷DKK1,⁴⁸SFTA1P⁴⁹) and a few downregulated ones (e.g., CYP4Z1,^50,51DIAPH2,⁵²GABRA3,⁵³FER,⁵⁴SEMA3E,⁵⁵S100A7,⁵⁶⁻⁵⁸PLCB4⁵⁹) have been reported to play an important role in proliferation, migration, and invasion in breast cancer cells. Similar gene regulations have been observed for direct drug-induced DNA damage or through DNA damage-signaling molecules like DDR, ATM, and C-Abl.⁶⁰⁻⁶² The top upregulated ANKRD1, as a transcriptional coactivator, enhanced the p53 activity to suppress tumor growth and promote apoptosis.⁴⁶ It was reported that SERPINB overexpression inhibited malignant cancer cell survival and suppressed invasion and migration of malignant cancer cells such as breast cancer cells.⁴⁷ Also, most of the other upregulated genes were involved in promoting a DNA damage-induced cytotoxicity.^48,49 Meanwhile, CYP4Z1, the top downregulated gene, promoted ERK1/2 phosphorylation and activated the PI3K-AKT pathway.^50,51 As a proto-oncogene, FER’s inactivation dramatically inhibited α6- and β1-integrin-dependent adhesion to cause anoikis.⁵⁴ Taken together, these upregulated and downregulated genes played an important role in cancer cell survival and growth and mediated drug 1-induced cancer cell cytotoxicity.

Table 2. Differentially Expressed Genes in 1-Treated MDA-MB-468 Cells Relative to Nontreated Cells: Upregulated and Top 20 Downregulated Genes.

top 20 downregulated genes		upregulated genes
gene symbol	fold change	gene symbol	fold change
FER	–2.22	ANKRD1	3.48
SBF2	–2.68	DKK1	2.91
PLCB4	–2.69	SFTA1P	2.56
NBAS	–2.75	MIR3143	2.49
SEMA3E	–2.78	OLR1	2.47
STX8	–2.79	SERPINB7	2.30
S100A7	–2.83	TMEM27	2.27
CLCA2	–2.85	CPA4	2.22
S100A7A	–2.86	ACTBL2	2.17
AKR1C2	–2.89	KRTAP2-3	2.10
CADPS2	–2.89	ROS1	2.09
CDKAL1	–2.90	ZNF699	2.09
DPYD	–3.00	IL1RL1	2.08
FAM172A	–3.00
GABBR2	–3.13
CYP4Z2P	–3.15
FMO6P	–3.18
GABRA3	–3.25
GLYATL2	–3.25
CRISP3	–3.70
DIAPH2	–3.93
CYP4Z1	–4.30

Discussion and Conclusion

The ability to target tumor cells selectively is a central aim in cancer therapy. Consequently, the unique biological processes of cancer cells have been exploited to design safer cancer therapies. The use of an ROS responsive trigger to induce the production of a cytotoxin to kill cancer cells offers a therapeutic advantage, as cancer cells have an increased level of ROS in comparison to normal cells.²⁻⁵ ROS-activated anticancer prodrugs have been sought for some time, but very few showed an in vivo efficacy and selectivity.^20,24,32 Here, we demonstrated the therapeutic utility of two ROS-activated DNA interstrand cross-linking agents using a xenografted mouse model. DNA-alkylating agents like cyclophosphamide, chlorambucil, and bendamustine are some of the most widely used anticancer drugs. They are effective against fast-dividing cancer cells because they interfere with DNA replication and transcription, stall mitosis, and/or induce apoptosis. However, many nonmalignant cells also divide rapidly, such as cells in bone marrow, the lining of the mouth and intestines, and hair follicles. Thus, most DNA-targeting anticancer drugs have serious side effects, including weight and hair loss, nausea and vomiting, fatigue, low blood-cell counts, easy bruising or bleeding, and the risk of cardiotoxicity. The side effects due to their toxicity are dose-limiting. Nevertheless, in the absence of improved agents, alkylating agents are still needed for cancer treatments. However, more selective DNA-targeting agents are needed to reduce unwanted side effects.

The in vivo evaluation suggested that ROS-activated DNA cross-linking agents, CWB-20145 (1) and its methyl analogue FAN-NM-CH₃ (2), showed an improved in vivo efficacy and selectivity in comparison with the clinically used DNA alkylation agents chlorambucil and melphalan. CWB-20145 and FAN-NM-CH₃ were not only more toxic than chlorambucil and melphalan in several cancer cell lines but also demonstrated an enhanced in vivo efficacy, better safety, and reduced side effects. Both compounds led to a significant tumor shrinkage in mice xenografted with the MDA-MB-468 cell line (up to 80% shrinkage in tumor size) without obvious signs of general toxicity. We further demonstrated that, in comparison with the parent compound CWB-20145, a methyl analogue FAN-NM-CH₃ showed improved drug-like properties (e.g., increased duration time and absorption) and a superior in vivo efficacy with a favorable safety profile. This provides valuable guidance for the further design of compounds with optimized drug-like properties that can be ultimately used as a human therapeutic.

Importantly, our study indicated that CWB-20145 and FAN-NM-CH₃ are the most effective against TNBC cells, such as MDA-MB-468 cells. Among different subtypes of breast cancers, TNBC, which lacks an expression of an estrogen receptor, progesterone receptor, and HER2, is particularly difficult to treat and often has poor prognoses.^63,64 The revolution that has transformed the treatment of many breast cancers has largely bypassed patients with triple-negative tumors. Because of the absence of a recognizable therapeutic target, the systemic treatment options for TNBC are still limited to cytotoxic chemotherapy.⁶⁵⁻⁶⁹ CWB-20145 and FAN-NM-CH₃ showed increased in vivo efficacy and selectivity toward TNBC cells, which may lead to a selective chemotherapy with phenyl boronic acid-modified DNA cross-linking prodrugs as a new treatment option for patients with TNBC. Although the in vivo mechanism of function for this type of molecule has not been fully understood yet, a high level of H₂O₂ was detected with TNBC cells, such as the MDA-MB-468 cell, which might be one of the factors that accounted for an improved efficacy and selectivity of these molecules toward TNBC cells. As the toxicity of the alkylating effectors is masked by the presence of electron-withdrawing boronic acid, these prodrugs are unlikely to be activated in normal cells with a low level of H₂O₂ but are expected to be activated specifically in cancer cells under an oxidative stress. However, a correlation between the ROS level and an in vivo efficacy and selectivity has not been defined yet, which is under investigation.

DNA alkylating agents, such as chlorambucil, produce anticancer effects by interfering with DNA replication and damaging the DNA in a cell. DNA damage induces a cell cycle arrest and cellular apoptosis via the accumulation of tumor suppressor protein p53. Caspase-3 and caspase-7 are two of the major effector caspases involved in the execution phase of apoptosis and are responsible for the breakdown of several cellular components involved in DNA repair and regulation.^43,44 The ApoToxGlo assay demonstrated that CWB-20145 caused a significant apoptosis evaluated by a caspase 3/7 protein expression. A treatment of MDA-MB-468 cells with CWB-20145 or chlorambucil resulted in a dose-dependent decrease in the apparent viability with no obvious increased cytotoxicity but an enhancement of caspase-3/7 activity, a profile consistent with cell cycle arrest and early-phase apoptosis. These results suggest that an apoptosis induced by CWB-20145 or chlorambucil is associated with the activation of caspase-3/7. Gene regulation indicated that CWB-20145 was able to significantly induce the p53 expression that in turn activated the expression of p21 and inhibited the cell cycle progression. A gene regulation effected by chlorambucil and melphalan was similar but less pronounced at the same concentration. An increased upregulation of p53 by the ROS-activated prodrugs suggests their increased DNA-damaging capability in cells.

Additionally, a microarray analysis indicated that 13 genes were upregulated by CWB-20145 and that 62 genes were downregulated. Most of the upregulated genes, including ANKRD1, DKK1, SFTA1P, MIR-3143, SERPINB7, ROS1, and IL1RL1, mediate upregulation of the p53 tumor suppressor protein. For example, ANKRD1 is a proapoptotic gene that has been reported to be a transcriptional coactivator of the p53 tumor suppressor protein.⁴⁶ The increased activity of p53 enhanced the affinity of YAP1 to bind with p73, leading to an overexpression of ANKRD1, which in turn increased the p53 activity.⁶⁰⁻⁶² It has been shown that an overexpression of SFTA1P can lead to increased levels of TP53 mRNA and protein, therefore suppressing cell proliferation, migration, and invasion.⁴⁹ An overexpression of p53 could also lead to an expression of MIR-3143 that inhibits the expression of two oncogenes AKT1 and PIK3CA.⁷⁰⁻⁷³ Mohammadi-Yeganeh et al. demonstrated that miR-3143 targets both PI3CA and AKT1 oncogenes, which is an effective factor to inhibit breast cancer progression and metastasis.⁷³ It has been shown that tumor suppressor miRNAs, such as miR-3143, were often downregulated in breast cancer cells, in particular, TNBC cells.⁷² An upregulation of miR-3143 may suggest a novel strategy based on ROS-activated prodrugs for miRNAs-based therapies for a TNBC treatment. The overexpression of the SERPINB gene has been reported to effectively suppress the invasiveness and motility of malignant cancer cells.⁴⁷ p53 also participates in pathways that lead to higher levels of ROS, which then further leads to DNA oxidative damage and an expression of the gene SERPINB7 that inhibits proliferation.⁴⁷IL1RL1 is induced through an immune response via IL-33 that increases numbers and IFNγ production by CD8⁺ and NK cells in tumor tissue.⁷⁴ It has been shown that IFNγ expresses NADPH oxidase, which enhanced ROS levels that are important for a prodrug activation and pro-apoptotic gene expression. Collectively, these data suggested that the ROS-activated prodrug CWB-20145 causes an apototic cell death in MDA-MB-468 breast tumors by a p53-dependent pathway as a result of drug-induced DNA damages. However, to provide more detailed signal transduction pathways will require more in-depth study, which is part of our ongoing efforts.

Most downregulated genes do not directly interact with p53. However, it has been reported that many of the genes are downregulated because of the corresponding inhibitor genes that are highly expressed due to DNA damage, such as CYP4Z1,^75,76CYP4Z2P,^75,76DIAPH2,^52,77,78 and GABRA.^79,80 Several of the downregulated genes, such as CYP4Z1,^51,81GABRA3,⁵³S100A7,⁵⁶⁻⁵⁸FER, and SEMA3E, are strongly overexpressed in breast cancer cells and in breast cancer metastases, which promotes tumor angiogenesis and growth in breast cancer and is associated with a poor prognosis of TNBC. For example, the most downregulated gene is CYP4Z1, a family member of cytochrome P450.⁸¹ It has been reported that the downregulation of CYP4Z1 promotes cell apoptosis.⁵⁰ Downregulation of CYP4Z1 induced by 1 suggests that these ROS-activated prodrugs may represent a novel approach to prevent a breast cancer progression by targeting CYP4Z1.⁸²DIAPH2 is one of the genes involved in the actin cytoskeleton pathway. Blocking the expression of DIAPH2 significantly inhibits breast cancer cell migration.^52,77,78GABRA3 is highly expressed in breast cancer, which inversely correlates with breast cancer survival by promoting breast cancer cell migration, invasion, and metastasis.⁵³FER kinase promotes breast cancer growth and metastasis by regulating cell adhesion and migration. FER is highly expressed in aggressive breast carcinomas, which correlates with high-grade basal/triple-negative tumors and worse overall survival. It has been shown that inducible FER downregulation in vivo inhibited tumor growth and the formation of distant metastases.⁵⁴SEMA3E is expressed in murine mammary adenocarcinoma cells that regulate tumor survival and correlates with the metastatic progression of human breast cancers. It was reported that silencing SEMA3E in breast cancer cells induced apoptosis.⁵⁵S100A7 is elevated in estrogen receptor (ER)/PR negative breast cancer, which is strongly correlated to an increased tumor growth, metastatic capacity, and a poor prognosis.⁵⁶⁻⁵⁸PLCB4 is a top-ranking upregulated gene in aggressive cancer associated with tumor progression.⁵⁹ Downregulation of those genes suggests that these ROS-activated prodrugs may represent a novel approach to prevent a breast cancer progression by targeting these genes.

In conclusion, following an earlier development of ROS-activated DNA alkylating agents to improve the selectivity and reduce the side effects of anticancer agents, we now report a more potent and selective drug candidate FAN-NM-CH₃ that is effective in vivo. This compound has a greatly improved in vivo efficacy and selectivity in addition to its reduced side effects and improved pharmacokinetic properties. The promising therapeutic activity and selectivity of FAN-NM-CH₃ against TNBC cells suggests that the H₂O₂-activated cross-liking agents have a potential for use in the treatment of TNBC patients and warrants a further evaluation in clinical setting. Current studies are focused on the identification of metabolites, defining the correlation between the in vivo efficacy and ROS level, understanding signal transduction pathways, and evaluating the immunotoxic effects of this novel class of compounds in addition to future Investigational New Drug enabling studies with FAN-NM-CH₃.

Glossary

Abbreviations Used

ROS

reactive oxygen species

DNA

DNA

ICL

interstrand cross-linking

CLL

chronic lymphocytic leukemia

TNBC

triple-negative breast cancers

ER

estrogen receptor

PR

progesterone receptor

HER2

human epidermal growth factor receptor 2 (MDA-MB-468)

DCM

dichloromethane

NMR

nuclear magnetic resonance

HRMS

high resolution mass spectrometry

DMSO

dimethyl sulfoxide

Bu

butyl

Pr

propyl

PAGE

polyacrylamide gel electrophoresis

ODN

oligodeoxynucleotide

FBS

fetal bovine serum

NEAA

nonessential amino

RPMI

Roswell Park Memorial Institute

IACUC

Institutional Animal Care and Use Committees

MTD

maximum tolerated dose

PEG

polyethylene glycol

PBS

phosphate-buffered saline

DMEM

Dulbecco’s modified eagle medium

ATCC

American Type Culture Collection

NADPH

nicotinamide adenine dinucleotide phosphate hydrogen

HPLC

high-performance liquid chromatography

LC

liquid chromatography

MS

mass spectrometry

PK

Pharmacokinetics

IP

intraperitoneal

ESI

electrospray ionization

MRM

multiple reaction monitoring

RNA

ribonucleic acid

RT-qPCR

quantitative reverse transcription polymerase chain reaction

GAPDH

glyceraldehyde 3-phosphate dehydrogenase

FP

forward primer

RP

reverse primer

IC₅₀

the concentration of inhibitor required to produce 50% inhibition

IR

inhibition rate

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsptsci.0c00092.

• Figures illustrating the purity of the compounds (HPLC profile and NMR and HRMS spectra), representative phosphor image autoradiogram of denaturing PAGE analysis of DNA ICL formation, time-dependent toxic response of MDA-MB-468 cells, graphs for the ApoTox-Glo assay, comparison of tumor weight in control mice and drug-treated mice, H₂O₂ detection, LC-MS analysis of compound 2 in cell culture media, the data obtained with the blood sample analysis, and gene expression data from microarray experiments (PDF)

Author Contributions

^† (H.F., M.A.U.Z., and W.-B.C.) These authors contributed equally to this work.

The authors declare no competing financial interest.