Cytotoxicity,crosslinking and biological activity of three mitomycins

Abstract

While interstrand crosslinks (ICLs) have been considered as one type of DNA damage in the past, there is mounting evidence suggesting that these highly cytotoxic lesions are processed differently by the cellular machinery depending upon the ICL structure. In this study, we examined the crosslinking ability of three mitomycins, the structure of the ICLs they produce and the cytotoxicity of the drugs toward three different cell lines. The drugs are: mitomycin C (1), decarbamoylmitomycin C (2), and a mitomycin-conjugate (3) whose mitosane moiety is linked to a N-methylpyrrole carboxamide. We found that, overall, both MC and compound 3 show strong similarities regarding their alkylation of DNA, while DMC alkylating behavior is markedly different. To gain further insight into the mode of action of these drugs, we performed high throughput gene expression and gene ontology analysis to identify gene expression and cellular pathways most impacted by each drug treatment in MCF-7 cell lines. We observed that the novel mitomycin derivative (3) specifically causes changes in the expression of genes encoding proteins involved in cell integrity and tissue structure. Further analysis using bioinformatics (IPA) indicated that the new derivative (3) displays a stronger downregulation of major signaling networks that regulate the cell cycle, DNA damage response and cell proliferation when compared to MC and DMC. Collectively, these findings demonstrate that cytotoxic mechanisms of all three drugs are complex and are not solely related to their crosslinking abilities or the structure of the ICLs they produce.

Graphical Abstract

INTRODUCTION

Mitomycin C (MC, 1, Fig. 1) is a DNA alkylating agent broadly used in chemotherapy, particularly with gastrointestinal related cancers (1–3). Mitomycins undergo a complex bioreductive process in vivo to yield reactive intermediates which form interstrand crosslinks (ICLs) between the exocyclic amines of opposing deoxyguanosine moieties (4–6). If ICLs are not repaired, they stall DNA replication which may lead to mitotic catastrophe and cell death. In addition to ICLs, mitomycins also generate monoadducts and intrastrand crosslinks in reactions with DNA, however, the scientific consensus is that ICLs are the major cytotoxic lesions produced by these drugs (7,8).

Figure 1.

Structure of mitomycin C (MC, 1), decarbamoylmitomycin C (DMC, 2) and a mitomycin-lexitropsin conjugate (MC-lex, 3).

Research in our laboratory has focused on Mitomycin C (MC, 1, Fig. 1) and two analogues of MC: Decarbamoylmitomycin C (DMC, 2, Fig. 1, a derivative of MC lacking the carbamoyl at C10) and a mitomycin conjugate (MC-lex, 3, Fig. 1) whose mitosane moiety is linked to the N-methylpyrrole carboxamide framework present in lexitropsins (9). Our interest in MC (1), DMC (2), and MC-lex (3) derives from the fact that they display significant differences in their cytotoxicity, even though all three drugs share common structural features. Since ICLs constitute the molecular basis for the cytotoxic effects of mitomycins, it is worth investigating if and how, differences in the local DNA structures of the ICLs play a role in the specific cellular responses triggered by MC, DMC and compound 3. MC and DMC can crosslink DNA in both 5’-CpG and 5’-GpC sequence contexts, and the crosslinking is diastereospecific: At 5’-CpG, mitomycins produce trans-ICLs (see 4a, Fig 2, the stereochemical configuration at C1” is R) whereas the cis-ICL is found at 5’-GpC steps (see 5a, Fig.2, the stereochemical configuration at C1” is S) (10,11). The major ICL formed by Mitomycin C in culture cells is the trans-ICL 4a, whereas decarbamoylmitomycin C yields the stereoisomeric cis-ICL 5a preferentially (Fig.2) (8). NMR solution structure and energy minimization computations have indicated that 4a (the trans-ICL) induces minimal distortion to the DNA helix which remains in a relatively undisturbed B-DNA structure (12). Conformational and CD studies have also showed that the cis-ICL, like the trans-ICL, only slightly affects the DNA structural parameters, and that its overall structure remains a B-DNA conformation (11). Until now, the structure of the ICL formed by 3 and the sequence context for crosslinking have not been elucidated.

Figure 2.

Structure of the major mitomycin C (MC) and decarbamoylmitomycin C (DMC) DNA-adducts.

Evidence of the importance of ICL structures in repair and recognition processes during replicative and non-replicative events has recently emerged. Kato et al. have found that ICL structure is a critical determinant of repair efficiency outside of DNA replication (13). The use of distinct mechanisms to repair structurally different ICLs has also been demonstrated in replication dependent processes. In yeast, diverse repair pathways are triggered in response to crosslinking damage according to different crosslinking drugs (14). In mammalian cells, the distortion of the DNA helix induced by ICLs influence the efficiency of ICL unhooking and repair (15). Moreover, in at least two cases involving Xenopus extracts, the structure of the ICL present determined which replication-coupled ICL repair pathway was activated (16). In the case of mitomycins, Paz et al. found that the cis and trans stereoisomeric ICLs (4a and 5a, Fig. 2) were removed at different rates in EMT6 cells, with the cis-ICL more rapidly eliminated than the trans-ICL (17). There is also some evidence that cis adducts produced by mitomycins signal differently to the cellular machinery than trans-adducts (18–23). In particular, cis-adducts produced by DMC are able to trigger a p53-independent mode of cell death more rapidly than MC trans-adducts (18–21). Our overarching goal is to identify cellular mechanisms triggered by chemotherapeutics such as DMC which are able to induce a rapid p53 independent cell death. This is particularly crucial in the case of triple-negative breast cancers (TNBCs), since the mutation of p53 is present in 80% of this type of cancers (24). MC is currently used to treat advanced breast cancer, alone or in combination with other drugs (25). Therefore elucidating the relationship between the ICL structure produced by MC (1), DMC (2), MC-lex(3) and the cellular responses on wild type (WT) p53 and mutant p53 breast cancer cell lines may pave the way to the design of more efficient chemotherapeutics to treat p53 mutant advanced breast cancer.

To investigate the relationship between the ICL structure produced by MC (1), DMC (2), MC-lex(3) and the biological activity of these three drugs we (a) elucidated the structure of the major ICL produced by MC-lex (3) as well as the crosslinking sequence targeted by this compound; (b) compared the crosslinking abilities of the three drugs; (c) carried out cytotoxic assays on wild type p53 and mutant p53 breast cancer cell lines; and (d) performed high throughput gene expression, gene ontology analysis and bioinformatics to identify both gene expression changes and cellular pathways most impacted by each drug treatment.

MATERIAL AND METHODS

Materials

MC was generously gifted by Professor Maria Tomasz. Reagents for the chemical synthesis and purification of DMC, mitomycin A and compound 3 were purchased from Sigma Life Sciences (St. Louis, MO) unless otherwise stated. Phosphodiesterase I (snake venom diesterase (SVD), Crotalus adamanteus venom, E.C. 3.1.4.1.) and alkaline phosphatase (Escherichia coli, EC 3.1.3.1) were obtained from Worthington Biochemical Corp (Freehold, NJ). Nuclease P₁ (penicillium citrinum, EC 3.1.30.1) was obtained from Sigma Life Sciences (St. Louis, MO). Sep-Pak C-18 cartridges were purchased from Waters Corp (Milford, MA). Oligonucleotides were purchased from Midland Certified Reagent (Midland, TX). M. lysodeicticus DNA (M. luteus DNA, type XI; 72% GC) and calf thymus DNA (type XV; 42% GC) were from Sigma Life Sciences (St. Louis, MO).

Synthesis of mitomycins

DMC, compound 3 and mitomycin A (MA) were synthesized from MC following published procedures (9,26,27).

Hydrolysis of compound 3 and isolation of the cis and trans apo-(3)

Compound 3 (15 mg, 0.032 mmol) was hydrolysed in 1 mM HCl. UV spectra of the hydrolysis products were recorded at 5 min interval during 180 min, after which no change in UV spectra was observed (supporting information; Fig. S1 and Table S1). The reaction mixture was neutralized using ammonium hydroxide and cis and trans mitosene products (apo-(3)) were isolated by prep TLC (SiO₂, 3% methanol in dichloromethane). The diastereomeric products were identified by their UV spectra and HRMS (Fig. S1–S3). HRMS m/z calcd for C₂₁H₂₄N₅O₇ [M+H]⁺: 458.1670; found: 458.1672 (Top spot isolated by prep TLC) and 458.1673 (Bottom spot isolated by prep TLC) (supporting information; Fig. S2, S3). Note: we did not identify which one of these two compounds was the cis nor trans apo-(3) since this information was not necessary for this study.

Geometry optimization and optimum energy calculations

Geometry optimization and optimum energy calculations were performed with Spartan software using the Density Functional Theory (DFT) at B3LYP level, with the 6–32G** basis set. Table S2 (supporting information) shows the optimized energies of reactants, starting materials and products in the reductive activation steps of MC and 3. Table S3 (supporting information) displays the energies of the reactions involved in the reductive activation process for MC and 3. Figures S4 and S5 (supporting information) represent compounds 8-14 from the reductive activation steps of MC and 3.

Synthesis and isolation of mitomycin A (MA) interstrand crosslink (7a) and monoadduct (7b)

This synthesis was performed using a modified published procedure (28). Self-complementary oligonucleotide 5’-ATATACGCGTATAT-3’ (300 A₂₆₀ unit scale; corresponding to 1.77 μmol) were dissolved in 100 mM potassium phosphate buffer, pH 7.5 (4.0 mL). Oligonucleotides were then annealed by heating (90°C, 10 min) followed by slow cooling. Mitomycin A (60 μmol) was added to the reaction mixture and was put under ice and deaerated via argon bubbling (30 min, 0°C). Sodium dithionite was added in four portions (18.25 μmol from a freshly prepared anaerobic 0.04 M solution in buffer) at 10 min intervals. A final addition of sodium dithionite was performed until the solution became colorless and allowed to stir over ice and under argon for 1 h. The solution was then open to air and stirred on ice for 30 min. The mixture was chromatographed on a 2.5 × 56 cm Sephadex G-25 column using 20 mM ammonium bicarbonate as eluent. Oligonucleotide containing fractions were lyophilized. The lyophilized fractions were redissolved in 20 mM ammonium acetate, pH 5.5 (1.25 A₂₆₀ units/mL). Nuclease P₁ (1.0 unit/A₂₆₀ unit of complex) was added to the mixture followed by incubation for 2 h at 37 °C. The pH was adjusted to 8.2 by addition of 200 mM sodium hydroxide solution (18 μL/mL of ammonium acetate solution) and 20 μL of 100 mM MgCl₂ was added. Addition of snake venom diesterase (2 units/A₂₆₀ unit of complex) and addition of alkaline phosphatase (1.6 units/A₂₆₀ unit of complex) were followed by incubation at 37 °C for 2 h. Semi-preparative isolation of the MA monoadduct 7b and interstrandcrosslink (dG-MA-dG, 7a) was completed using an Agilent 1200 HPLC system with a Phenomenex Clarity Oligo-RP column (5 um, 250 × 10 mm). Buffer A: 30 mM Potassium Phosphate pH 5.4; Buffer B: Acetonitrile. Flow Rate: 3 mL/min. The temperature was set to 30°C. The elution gradient was: t=0 to t=5 min: 6 to 12% B; t=5 min to t=10 min: 12% B; t=10 min to t=30 min: 12 to 16% B; t=30 min to t=32 min: 16 to 50% B; t=32 min to 37 min: 50% B; t=37 min to t=40 min: 50 to 6% B. Compounds were characterized by their UV spectra and by HRMS (Fig. S6 and S7). m/z calcd for 7b C₂₄H₂₈N₇O₈ [M+H]⁺: 541.1994; found: 542.1988; m/z calcd for 7a C₃₄H₃₉N₁₂O₁₁ [M+H]⁺: 791.2856; found: 791.2858 (Fig S6 and S7).

Conversion of mitomycin A interstand crosslink and monoadduct to the interstand crosslink and monoadduct formed by compound 3 (chemical conversion of 7a and 7b to 6a and 6b)

Mitomycin A monoadduct (7b) or ICL (7a) (0.033 μmol) were incubated with methyl 5-amino-1-methyl-1H-pyrrole-2-carboxylate (0.33 μmol, 51 μg) in methanol (1 mL) at 30°C. The conversion was monitored as follow: an aliquot (10 μL) of the reaction mixture was extracted with water and dichloromethane and the water layer was analyzed by HPLC using an Agilent 1200 HPLC system and a Kromasil C-18 reverse phase column (0.46 × 25 cm). Buffer A: 10 mM TEAA pH 7.0; Buffer B: Acetonitrile. Flow Rate: 0.7 mL/min. The temperature was set to 30°C. The elution gradient was: t=0 to t=5 min: 6 to 20% B; t=5 min to t=10 min: 20% B; t=10 min to t=100 min: 20 to 40% B; t=100min to t=105 min: 40 to 50% B; t=105 min to t=110 min: 50% B; t=110 min to t=113 min: 50 to 6% B. The final mixtures containing either DNA-monoadduct (6b) or ICL (6a) were also analyzed by LC-DAD-MS using the same HPLC method, column, elution gradient and flow rate. The MC-lex monadduct 6b (m/z calcd for C₃₀H₃₄N₉O₉ [M+H]⁺: 664) was detected at RT=33.2 min. The MC-lex ICL 6a (m/z calcd for C₄₄H₄₅N₁₄O₁₂ [M+H]⁺: 913) was detected at RT=29.9 min. (Fig. S8 and S9).

M. luteus and calf thymus DNA alkylation by compound 3 under bifunctional activation

A solution of sonicated DNA (either M. luteus or calf thymus; 12 mM) and (3) (1mM) was mixed in deaerated 10 mM potassium phosphate-1 mM EDTA, pH 5.8 (1 mL). Addition of a deaerated solution of Na₂S₂O₄ (2mM, from a freshly prepared anaerobic 120 mM solution in water) and of a solution of 3 was performed in 5 increments until a total of 5 mM of 3 and 10 mM of sodium dithionite were added, according to a published protocol for “bifunctional” activation of mitomycin developed in our laboratory (10). The drug-DNA complexes were extracted by adding 10 mM potassium phosphate, pH 9.0 to adjust the pH to 8.0 and then adding 1 mL of phenol:chloroform:isoamyl alcohol extraction solution (25:24:1 v/v). The solution was then vortexed vigorously and allowed to settle before transferring into vials and centrifuged (13,000 rpm, 10 min). DNA was isolated by applying only the top aqueous layer on a Sephadex G-25 column. DNA containing fractions were isolated and lyophilized.

Cell treatment and DNA/RNA extraction

Breast cancer cells (MCF-7 cells and MDA-MB 468 cells) and control cells (MCF10A) were cultured according to protocols previously described (23). MCF-7 is a human breast adenocarcinoma cell line expressing estrogen receptor (ER) and progesterone receptor (PR), but not human epidermal growth factor receptor (Her2). MCF 10A is a non-tumorigenic epithelial cell line. MDA-MB 468 cell line is a triple negative breast cancer cell line not expressing ER/PR/Her2 and has missense mutations of the TP53 gene. These cell lines were chosen because of their different p53 status. MCF-7 cells were cultured within completed media containing DMEM, 10% FBS, and 50 μg/mL gentamicin in a humidified atmosphere at 37°C and 5% CO₂. For chemical treatment (50 μM), cells were subcultured in Petri dishes or in a 24-well plate a day prior to the experiment. Cells were grown until medium density (~80% confluence) before chemical treatments. DNA and RNA were extracted from treated and untreated cells for the identification of DNA adducts after enzymatic digestion and for the analysis of changes in gene expression after treatment. Approximately ~4×10⁸ cells were grown in 150-mm Petri dishes. TRIzol^® reagent (Invitrogen Life Technologies) was added to the cell pellets for DNA and RNA extraction following the specific manufacturer’s protocol. For DNA extraction, cell pellets were incubated with TRIzol reagent for 5 min at room temperature. Following incubation, 0.2 mL of chloroform per 1 mL of TRIzol reagent was added and the tubes were shaken by hand vigorously for 15 seconds before incubating at room temperature for 3 min. Samples were then centrifuged at 12,000 ×g for 15 min at 4°C to separate the mixture into 3 layers. After removing the top aqueous layers, 0.3 mL of 100% ethanol per 1 mL TRIzol reagent was added to the remaining 2 layers and the tubes were inverted for mixing. After incubating at room temperature for 3 min, the mixture was then centrifuged at 2,000×g for 5 min at 4°C to precipitate DNA. DNA pellets were then washed twice with 1 mL of 0.1M sodium citrate in 10% ethanol for 30 min at room temperature with occasional mixing before centrifugation at 2,000 ×g for 5 min at 4°C. Following these 2 washes, 1.5 mL of 75% ethanol were added and the mixture was incubated at room temperature for 20 min prior to final centrifugation at 2,000 ×g for 5 min at 4°C. Dry DNA pellets were then subjected to enzymatic digestion and analysis. For extraction of total RNA, the cell pellets containing TRIzol were kept at −80°C and processed later. RNA extraction was conducted using the Direct-zol RNA Purification kit (Zymo Research) according to the manufacturer’s protocol. RNA samples were quantified using Nanodrop and purity was assessed by determining the 260/280 and 230/280 ratios.

Enzymatic digestion of the alkylated DNA from the reaction between compound 3 and M-luteus or calf thymus DNA to adducted nucleosides

The lyophilized (3)-DNA complex (M. luteus or calf thymus) was dissolved in 20 mM ammonium acetate, pH 5.5 (2.5 A₂₆₀ units/mL). Nuclease P₁ (1.0 unit/A₂₆₀ unit of complex) was added to the mixture followed by incubation for 4 h at 37 °C. The pH was adjusted to 8.2 by addition of 200 mM sodium hydroxide (25 μL/mL of ammonium acetate), and MgCl₂ was added to a concentration of 0.9 mM. Addition of snake venom diesterase (2.25 units/A₂₆₀ unit of complex) and 2 h incubation at 37 °C were followed by the addition of alkaline phosphatase (1.6 units/A₂₆₀ unit of complex) and incubation overnight at 37 °C. Samples were lyophilized and redissolved for HPLC analysis.

Enzymatic digestion of the alkylated DNA from cells treated by compound 3 to adducted nucleosides

Cell DNA pellets were dissolved in 8 mM sodium hydroxide and 25 mM HEPES buffer using ultrasonication (⅛ inch tip; pulse setting with 30% amplitude for 30 seconds followed by 30 second wait for 15 mins). The dissolved DNA was then chromatographed on a 2.5 × 56 cm Sephadex G-25 column using 20 mM ammonium bicarbonate as eluent. DNA containing fractions were lyophilized. Lyophilized DNA was redissolved in dilute acetic acid pH 5.0 (2.5 A₂₆₀ units/mL). Nuclease P1 (1 unit/A₂₆₀ unit of DNA) was added to the mixture followed by incubation at 37° C for 4 h. The pH of the solution was adjusted to 8.2 by addition of 0.5 M Tris pH and MgCl₂ was added to a concentration of 1.0 mM. Snake venom diesterase was added to the mixture (2.25 units/A₂₆₀ unit of DNA) followed by 2 h incubation at 37°C. Lasty, alkaline phosphatase was added to the mixture (1.6 units/A₂₆₀ unit of DNA) followed by overnight incubation at 37°C. The digested mixture was concentrated and applied to a Sep-Pak C-18 cartridge previously washed consecutively with 2 mL acetonitrile, 2 mL water, and 1 mL 10 mM ammonium acetate pH 6.4. The unmodified nucleosides were eluted with 80 mL of water then 5 mL of 10% acetonitrile in water. Adducts were eluted with 30% acetonitrile in water (5 mL) and then 50% acetonitrile in water. Fractions were analyzed by HPLC.

HPLC analysis of DNA adducts after enzymatic digestion

Digestion mixtures were directly analyzed by HPLC. The HPLC system, column and elution method were the same as the one used for the detection of 6a and 6b produced during the conversion reactions of MA interstandcrosslink (7 a) and monoadduct (7b) described above (conversion of 7a and 7b to 6a and 6b). Digestion mixtures were also analysed by LC-DAD-MS using the same method.

In this case, during the first 10 minutes of the run, the detector was by passed to avoid salt contamination.

Crosslinking assessment using plasmid DNA

Plasmid pCMV6-Entry TAAR1 DNA from Origene (RC211034, 5.9 kb) was used as a DNA template to determine the crosslinking ability of mitomycins. After linearizing plasmid DNA with NOT I restriction enzyme, DNA was deaerated with N₂ gas for 15 min and then crosslinking reactions were carried out at 0°C, 22°C, and 37°C. First, all drugs (final concentration 50 μM) were activated using sodium dithionite under bifunctional activation conditions (2 equivalents of sodium dithionite relative to the amount of drug). Crosslinking reactions were monitored by denaturing 1.2% alkaline agarose gel electrophoresis using the difference of mobility between crosslinked plasmid DNA and linearized plasmid DNA. A published assay protocol was modified for this experiment (29). Briefly, 1.2% agarose gels prepared in 100 mM NaCl-2 mM EDTA (pH 8) solution were immersed in 40 mM NaCl-1mM EDTA electrophoresis running buffer for 1 h before used. Crosslinked and linearized plasmid DNAs were mixed with gel loading dye before loading onto agarose gels. Electrophoresis was conducted at room temperature for 3 h at 1 V/cm. After the run, gels were neutralized in 150 mM NaCl-100 mM Tris (pH 7.4) solution for 45 min, with the solution being changed every 15 min followed by staining with ethidium bromide (0.5 μg/ml) for 15 min. After rinsing the gels with distilled water for 15 min with shaking, DNA bands were visualized by Bio-Rad ChemiDoc^™ MP imaging System.

Crosslinking assessment using a comet assay

A modified comet assay protocol for quantifying DNA crosslinking was followed (30, 31). Briefly, cells were treated with mitomycins for 24 h. After treatment (50μM), cells were incubated with 100 μM hydrogen peroxide for 15 min before harvesting. Trypsinized cell pellets were treated according to the manufacturer’s protocol (Trevigen^® CometAssay^®, 4250–050-K). Comets were observed using a Nikon Eclipse 600 microscope and analysed by OpenComet (32). The degree of DNA crosslinking in mitomycins-treated cells was calculated as the percentage of the decrease of the tail moment. The percentage of the decrease of the tail moment is calculated as $\left[1-\left({\text{TM}}_{\text{drug}+{\text{H}}_2{\text{O}}_2}{\text{TM}}_{\text{contro}}\text{l}\right)/\left({\text{TM}}_{{\text{H}}_2{\text{O}}_2}-{\text{TM}}_{\text{control}}\right)\right]\times 100$. ${\text{TM}}_{\text{drug}+{\text{H}}_2{\text{O}}_2}$ is the tail moment of mitomycins + H₂O₂ treated samples. ${\text{TM}}_{{\text{H}}_2{\text{O}}_2}$ is the tail moment of H₂O₂ treated samples. TM_control is the tail moment of the control samples.

Cytotoxicity assays

Cell viability was determined by a neutral red assay which is based on the lysosome uptake of neutral red dye (33). After 24 h of chemical treatments (0.1 to 100μM), 20μL of 0.33% neutral red solution (Sigma Aldrich) was added onto wells. After 2 h incubation at 37°C in a 5% CO₂ incubator, the dye solution was carefully removed and cells were rinsed with 200μL neutral red assay fixative (0.1% CaCl₂ in 0.5% formaldehyde) twice. The absorbed dye was then solubilized in 200μL of Neutral red assay solubilization solution (1% acetic acid in 50% ethanol) for 10 min at room temperature on a shaker. Absorbance at 540 nm and 690 nm (background) was measured by BioTek Synergy Mx microplate reader. The IC₅₀ values obtained from MCF-7, MDA-MB 468 and MCF 10A cells treated with chemicals for 24 h are showed in Table 4.

Table 4:

IC₅₀ of Mitomycin C (MC, 1), decarbamoyl mitomycin C (DMC, 2), and Mitomycin C-lexitropsins conjugate (MC-lex, 3) towards breast cancer cell lines (MCF-7 and MDA-MB 468) and a non-tumorigenic epithelial breast cell line (MCF 10A

Cell line	IC50 (Mean ± SEM) μmol/mL
	MC (1)	DMC (2)	MC-lex (3)
MCF-7	13.52 ± 2.36	55.58 ± 8.17**	4.73 ± 0.98*
MDA-MB 468	28.42 ± 2.51†	2.95 ± 2.88** †	0.15 ± 0.14** †
MCF 10A	4.01 ± 0.78† Δ	27.88 ± 5.84** †Δ	6.38 ± 0.25** † Δ

^*vs MC, p<0.05;

^**vs MC, p<0.01;

^†vs MCF-7 cell line, p<0.05;

^Δvs MDA-MB 468 cell line, p<0.05

Gene expression and pathway prediction analysis

mRNA levels were determined using the nCounter system. Using ~250 ng of total RNA as input, the RNA samples were prepared by carrying out annealing, ligation and purification procedures according to the manufacturer’s instructions. This preparation entailed ligation of a specific DNA tag onto the 3’ end of each mature miRNA or mRNA thus normalizing the melting temperatures of the RNAs and providing identification for each mRNA present in the sample. Negative exogenous mRNA and the appropriate probes were added into the sample. The resulting sample was then hybridized with the nCounter mRNA PanCancer panel (Nanostring Technologies) tag-specific capture and reporter probes. These hybridization reactions were incubated at 65°C for approximately 18 h. Samples were then loaded into the nCounter SPRINT cartridge and run in the nCounter Analysis System. We used this system to profile the expression of 770 mRNA targets and 30 housekeeping genes per sample. nSolver was used to transform the data and determine gene expression in each sample. A Welch’s t-test was performed to determine which genes were differentially expressed between treated and untreated MCF-7 cells. All mRNAs expressed differently under each individual drug treatment were used for subsequent analysis. We used DAVID and KEGG to predict relevant biological pathways. To validate the results from the array, we conducted reverse transcription (RT) and real-time quantitative Polymerase Chain Reaction (RT-qPCR) in a subset of genes: MAP3K1, MDM2, JUN, MYB and GADD45. These were selected because the direction of the effect was similar for all drugs included in this study. The measurements were carried out using total RNA extracted from cells treated and untreated in triplicate experiments. cDNA was synthesized using the Omniscript RT kit (Qiagen) and 5μL of total RNA following the manufacturer instructions. From this reaction, we used 2μL to carry out RT-qPCR, using Taqman gene expression assays. Taqman runs were always done using duplicates and GAPDH was used as an endogenous control. The threshold cycle (Ct) values were used for analysis and the 2^−ΔΔCt method was used to quantify relative gene expression (34).

Gene ontology, molecular, and cellular pathways quantitative enrichment analysis

Canonical biochemical pathways and cellular signaling networks were generated using IPA (Ingenuity Systems) using the gene expression profiles identified from (Supplementary Table S9). Specifically, the experimentally determined gene expression ratios corresponding to data set: 1) [MC-Lex/ctr]; 2) [MC/Ctr] and 3) [DMC/Ctr] were used to calculate the fold changes by rescaling their values using a log2 transformation, such that positive values reflected fold increases while the negative values reflected fold decreases. For network generation, datasets containing gene identifiers (gene symbols) for each of the above-mentioned data set were uploaded into the IPA application together with their rescaled log2 transformation of the gene expression’s average ratios. These molecules were overlaid onto a global molecular network contained in the Ingenuity Knowledge Base and the networks were then algorithmically generated based on their connectivity index using the built-in IPA algorithm. The probability of having a relationship between each IPA indexed biological function and the experimentally determined genes was calculated by a right-tailed Fisher’s exact test. The level of significance was set to p < 0.05. As such, the IPA analysis identified the molecular and cellular pathways from the IPA library of canonical pathways that were most significant to the dataset (−log (P value) > 1.3). For the quantitative analysis of the expression profiles, IPA assigned the z-score function to all eligible canonical biochemical and cellular pathways (where z < −2 represents significant down-regulation while z > 2.0 represents a significant up-regulation of the selected pathways; Table S10, supporting information).

Cell cycle analysis

After chemical treatments, the cells were washed twice with ice cold phosphate buffered saline (PBS) and fixed in 70% ethanol at 4°C overnight. Then the fixed cells were stained with FxCycle^™ PI/RNase Staining Solution (Thermo Fisher Scientific) at room temperature for 30 minutes, and the cell cycle distribution was analyzed by Attune® NxT Cytometer using Attune® Cytometric Software.

Mass spectrometry

The analysis was performed in a Nexera X2 LC-30AD pump, SIL-30AC autosampler, CTO-20AC oven and SPD-M30A diode array detector (DAD). The mass spectrometer (MS) was the LCMS-8030 (triple quadrupole). The instrument was acquired from Shimadzu (Columbia, MD). The chromatographic separation (column, gradient and flow rate) was previously described above in the section “Conversion of mitomycin A interstand crosslink and monoadduct to the interstand crosslink and monoadduct formed by Compound 3”. The DAD wavelength range was 190–700 nm. The MS interface parameters were the following: electrospray positive, nebulizing gas flow 2 L/min, desolvation temperature 250°C, heat block temperature 400°C, and drying gas flow 15 L/min. The MS data was acquired in SIM (single ion monitoring) mode. The m/z monitored were 664 and 913.

RESULTS

Unambiguous structural identification of expected interstrand crosslink and monoadduct produced by 3

The structure of the major DNA adducts produced by MC and DMC have been elucidated. To compare the crosslinking ability and the cellular responses of the 3 mitomycins (MC, DMC and 3), we had to first identify the major DNA adducts produced by compound 3. The established mechanism of action for MC is initiated by a cellular reduction of the quinone moiety, followed by elimination of methanol to form a leucoaziridinomitosene (9, Fig 3A). This is followed by protonation and opening of the aziridine (11, Fig 3A) and culminates with mono-or bisalkylation of DNA resulting in the formation of 4a and 4b (Figure 2). Accordingly, 3 was expected to yield DNA adducts structurally identical to 4a and 4b except for the substituent at C(7). These probable major DNA adducts were labeled as trans-ICL (6a) and trans-monoadduct (6b) (Fig. 3B) and the UV-spectra of 6a and 6b were anticipated to display a combination of the chromophores present in both deoxyguanosine and in apo-3 (apo-3 is easily obtained from acidic hydrolysis of 3; Fig 3C, Fig. S1 and Table S1).

Figure 3:

(A) Bifunctional reductive activation mechanism of MC and 3. (B) Expected (and detected) major DNA adducts from the reaction of 3 with DNA. (C) Hydrolysis of compound (3) to cis and trans apo-3 monitored by UV-vis spectroscopy (0.163 mM, t=0 to t=180 min).

Identification of the expected crosslink and monoadduct produced by 3 was possible by taking advantage of the facile displacement of the 7-methoxy group of mitomycin A (MA) by aromatic primary amines (35,36). This involved treating both mitomycin A ICL (7a) and monoadduct (7b) with 5-amino-1-methyl-1H-pyrrole-2-carboxylate to obtain 6a and 6b (Scheme 1); the expected monoadduct and crosslink formed by 3 (7a and 7b were previously characterized by Maria Tomasz and synthesized in our laboratory: Fig.S6 and S7) (28).

Scheme 1:

Conversion of MA crosslink (7a) and MA monoadduct (7b) into the crosslink produced by 3 (6a) and the monoadduct produced by 3 (6b).

The conversions were monitored by HPLC and completed after 24 h at 30 °C (Fig. 4). The UV spectra of 6a and 6b were similar and showed two absorption bands at λmax=258 nm and 577 nm with a slight shoulder around 285 nm indicating the presence of apo-3 chromophore (Fig. 4) and deoxyguanosine. The structures of 6a and 6b were further confirmed by mass spectrometry (LC-DAD-MS) with a detected mass of 913 (M+H) for 6a, and 664 (M+H) for 6b (Fig S.6 and S.7). One should note that since the conversion of 7a to 6a, and 7b to 6b (i.e. displacement of the methoxy group on C7 by 5-amino-1-methyl-1H-pyrrole-2-carboxylate) does not affect the stereochemistry at C1”, the stereochemical configuration at C1” of 6a and 6b is also trans.

Figure 4:

HPLC chromatograms of the conversion reactions 7a to 6a and 7b to 6b. (*): primary amine 5-amino-1-methyl-1H-pyrrole-2-carboxylate.

Adducts 6a and 6b are the major adducts produced in the reaction of 3 with calf thymus and M. Luteus DNA

Calf thymus and M. luteus DNA were then reacted with fully reduced 3 according to a protocol developed in our laboratory (10,11). Under these conditions (i.e. bifunctional activation and low concentration of drug relative to the DNA concentration Fig. 3A), reactions produce an adduct profile that closely resembles the adduct profile observed in cell cultures treated with mitomycins (37). Alkylated DNA was digested by nucleases and phosphatases to the individual nucleosides, and two major adducted nucleosides were detected. The UV spectra, retention time, and mass of the two adducts matched that of 6a and 6b obtained from the MA conversion reactions described above (Fig. 5 and Fig. S10) (Note: the same HPLC and LC-DAD-MS system, equipment and elution method were used for both experiments). This data proves unambiguously that the two major adducts identified in the reaction between 3 and CT DNA, or M. Luteus DNA are identical to 6a and 6b. Additionally, we observed that the CG composition of the DNA (42% for CT DNA versus 72% for M. Luteus DNA) did not affect the nature of the adducts produced, which is also the case with MC, but in contrast to what happens with DMC.

Figure 5:

HPLC chromatograms of the enzymatic digests from the reaction of 3 with calf thymus DNA (A) and M. Luteus DNA (B) under bifunctional reductive activation.

Adducts 6a and 6b are produced in the reaction of 3 with cells

After treatment of MCF-7 cells with compound 3 (50 μM) under aerobic conditions, nuclear DNA was isolated and enzymatically digested to nucleosides. The adducted nucleosides were then separated from the unmodified ones (Sep-pak C-18 cartridge). HPLC analysis of this pre-purified fraction containing adducted nucleosides showed the presence of monoadduct 6a and crosslink 6b as well as uncharacterized compounds (*) whose UV spectra did not display absorption bands in the 260 nm region (Fig. 6). Other minor peaks present in the chromatogram may be minor DNA adducts formed by 3 in cells but were not characterized in this work.

Figure 6:

HPLC chromatogram from the enzymatic digest of cellular DNA. MCF-7 cells were treated with 3 at 50 μM for 24 h.

DMC, but Neither MC or 3, requires physiological or higher temperature for crosslinking

Compound 3, MC (1) and DMC (3) were incubated with plasmid DNA under reductive conditions to compare their ability to crosslink DNA. The crosslinked DNA products were then separated (agarose gel) and visualized under UV. MC (1) and MC-lex (3) crosslinked plasmid DNA at 0°C, 22°C and 37°C with similar efficiency. However, DMC (2) only crosslinked plasmid DNA at 37°C (Fig. 7).

Figure 7:

Crosslinking ability of plasmid DNA by MC (1), DMC (2) and 3 under bifunctional reductive activation at different temperature.

All three mitomycins show a similar amount of crosslinking with cellular DNA at room temperature

A modified comet assay was performed to monitor the cellular DNA crosslinking ability of the three mitomycins in MCF-7 cells (30). The decreased tail moment obtained from the modified comet assay correlates with the retardation of the electrophoretic mobility of alkaline denatured cellular DNA due to DNA crosslinking. Control cells treated with PBS showed no DNA damages and the decrease in tail moments was taken as 100% (100% decrease). Cells treated with H₂O₂ triggered DNA fragmentations, but no crosslinking. The decrease in tail moments of H₂O₂ treated cells showed almost no decrease in tail moments. When cells were treated with MC (1), DMC (2) and 3, the tail moments were decreased by approximately 30% (30%, 30%, and 32%, respectively). This shows that all three mitomycins crosslink cellular DNA to a similar extent in vitro at room temperature.

Cell cycle analysis

Compound 3 is more easily reduced than mitomycin C

The energies of the reactions involved in the reductive activation process for mitomycin C and compound 3 are displayed in the supporting information section (Table S2 and S3). The data indicates that the energy of the reduction step in the activation mechanism of 3 is lower by 3.82 kcals/mole, than in the case of mitomycin C (Table S3. Compare the energies of the initial reduction steps for MC and 3 respectively) and that the second step of the reductive activation process (loss of methanol, Table S3) is exothermic for mitomycin C, but endothermic for compound 3. Interestingly, while the structure of the intermediate protonated leuco-aziridinomitosene could be geometrically optimized in the case of the MC (10, Fig. 3A and Table S2), the equivalent structure for 3 is not stable. Instead, the aziridine opens spontaneously leading to the formation of the final reduced carbonium ion (14, Fig.3A). Geometrically optimized representative figures of compounds 8-14 produced during the reductive activation steps of MC and 3 (Fig. 3A) are shown in the supporting information (Fig. S4 and S5). One should note that these calculations are for the gas phase so there may be some differences with the energies of the reactions in solution.

Compound 3 is more cytotoxic toward breast cancer cell Lines MCF-7 and MDA-MB 468 than MC and DMC at concentration <15μM. Compound 3 is less toxic toward the non-cancer cell line MCF 10A than MC at low doses (<7μM)

Cytotoxicity results (neutral red assay) indicate that the novel derivative 3 is more cytotoxic toward triple negative breast cancer cell lines independently of their p53 status (MCF-7, wild type p53, MDA-MB 468, p53 mutant) at concentration lower than 15 μM, but less toxic toward non-cancer cell lines (MCF 10A) than MC at low doses (<7 μM).

Gene expression profiles

We used a PanCancer expression array to identify similarities and differences between the changes in gene expression induced by 3 (MC-lex) and the other two mitomycins, MC (1) and DMC (2). The array included 770 targets involved in tumorigenesis related pathways as well as targets related to immune response and tumor microenvironment. Figure 10A displays an unsupervised hierarchical cluster analysis of the mRNA levels in 513 targets with normalized values above the threshold. The heat map represents untreated and treated MCF-7 cells with the three different drugs (Figure 10A and Table S4 in Supporting Information). A total of 195 genes were differentially expressed after treatment with MC. Most of these genes (118) were underexpressed in treated cells when compared to untreated cells. A similar number of targets (201) were expressed differently after treatment with 3 (MC-lex); however, in this case, most targets were overexpressed (117). DMC (2) had the least number of differentially expressed genes when compared to untreated cells. A total of 83 targets were differentially expressed with the majority (65) being overexpressed. The Venn diagram shows the overlapping targets among expressed genes after each drug treatment (Figure 10B). Fifty-five (55) genes are uniquely differentially expressed after treatment with 3 (MC-lex), including some important genes which encode proteins involved in pathways related to genomic stability such as RAD51, RAD21 and MLH1. Only 33 common genes are differentially expressed after treatment with each drug (when compared to control), i.e. these genes are differentially expressed regardless of the nature of the mitomycin. This list includes MDM2, MAP3K1, JUN, and other genes which have been previously identified as important actors in DNA damage response.

Figure 10:

Differential expression patterns after treatment of MCF-7 cells with the three drugs, MC (1), DMC (2) and 3 (MC-lex), in MCF-7 cells. A. Unsupervised hierarchical clustering (heat map) showing the expression changes in untreated and treated cells. B. Venn diagram of differentially expressed mRNA targets.

We conducted a validation experiment by measuring the levels of six genes using real-time qPCR. We determined the relative gene expression of MDM2, MAP3K1, GADD45, JUN, and MYB using GAPDH1 as an endogenous control and calculated the expression percentage of each gene in reference to their levels in untreated cells (Fig. 11). We selected these genes because they respond similarly to all mitomycins. In comparison to the array, GADD45 and MYB both showed the same direction of change with the quantitative PCR as with the array. Levels of GADD45 were significantly higher after treatment with all drugs when compared to untreated cells. GADD45 levels were 16478 ± 11927 % (p_{MC vs CTRL}=0.010), 470 ± 128 % (p_{DMC vs CTRL}=0.004), and 607 ± 349 % (p_{MC-lex vs CTRL}=0.012) after exposure to MC (1), DMC (2), and MC-lex (3) respectively. Similarly, as in the array, levels of MAP3K1 and MYB in MCF-7 cells were significantly lower after treatment with all drugs when compared to untreated cells. Expression percentages of MAP3K1 after treatment with MC (1), DMC (2), and MC-lex (3) were 9 ± 3% (p_{MC vs CTRL}=0.0003), 10 ± 15 % (p_{DMC vs CTRL}=0.0006), and 10 ± 9 % (p_{MC-lex vs CTRL}=0.00003), respectively. For MYB, expression levels were 6 ± 1 % (p_{MC vs CTRL}=0.012), 4 ± 3 % (p_{DMC vs CTRL}=0.012), and 15 ± 8 % (p_{MC-lex vs CTRL}=0.007) after treatment with MC (1), DMC (2), and MC-lex (3). We found that, for the other two genes, results of the validation experiments showed partial agreement. In accordance with the results from the array, higher MDM2 levels (136 ± 14 % (p_{MC vs CTRL}=0.010) and 200 ± 40 % (p_{MC-lex vs CTRL}=0.005)) were observed after exposure to MC (1) and MC-lex (3); however no change in expression of MDM2 was detected after exposure to DMC (2) (115 ± 55 %, p_{DMC vs CTRL}=0.338). Similarly, expression of JUN after treatment with MC (1) and MC-lex (3) was consistent with levels found in the array: 136 ± 14 % for MC (p_{MC vs CTRL}=0.0005) and 200 ± 40 % for MC-Lex (p_{MC-lex vs CTRL}=0.036); in contrast to JUN expression after treatment with DMC (2): in this case, no change in expression percentage was detected (154 ± 148 %, , p_{DMC vs CTRL}=0.280).

Figure 11:

Gene expression levels by quantitative PCR. Relative gene expression in percentage (%) of MDM2, MAP3K1, GADD45, JUN, and MYB was determined using GAPDH1 as the endogenous control and using real time qPCR. Levels were measured in MCF-7 cells after treatment with MC (1), DMC (2), and MC-lex (3) with at least three replicates per experiment. t-tests were used to determine differences between treatments, and statistically significant differences are indicated in the graphs, *p<0.05, **p<0.01 and ***p<0.0001.

Pathway analysis using DAVID: Most predicted pathways are related to cell proliferation and survival. Compound 3 affects some specific biological pathways related to tissue structure.

We used the Kyoto Encyclopedia of Genes and Genomes (KEGG) and DAVID to predict which pathways would more likely be altered after treatment with MC (1), DMC (2) and 3 (MC-lex). A total of 98 pathways were predicted to be affected after treatment with 3 (MC-lex) with Benjamini-values (Q-values) below 0.05. In the case of MC (1) and DMC (2), DAVID software predicted 105 and 87 KEGG pathways respectively (Tables S5–S7 in supporting information). Most of the identified pathways are signaling mechanisms relating to cell proliferation and survival. Table 5 shows five representative pathways which are shared by the three drugs, the number of targets found to play a role in each pathway, as well as the corresponding Benjamini-values. These pathways include PI3K-Akt, MAPK, RAS, p53 and apoptosis. An example of how the three compounds target similar or diverse genes in the AKT pathway is shown in supporting information (Fig. S11). We then analyzed in detail the p53 signaling pathway for its relevance in DNA damage response. 81% and 67% of the targets are predicted to promote p53 signaling leading to apoptosis for MC and DMC respectively, while 67 % of the targets are predicted to inhibit p53 signaling after treatment with 3 (MC-lex) (Table 6). This suggests compound 3 does not promote p53 signaling to induce apoptosis/cell cycle arrest/senescence and that these biological events may be triggered through alternate pathways in response to DNA damage caused by 3.

Table 5:

Top pathways predicted by KEGG pathway analysis using DAVID shared by MC (1), DMC (2) and MC-lex (3).

KEGG Pathway Term	Main Pathway Characteristics	MC (1)		DMC (2)		MC-lex (3)
		Gene Counts	Benjamini value	Gene Counts	Benjamini value	Gene Counts	Benjamini value
PI3K-Akt signaling pathway	The PI3K-AKT signaling pathway is activated by cellular stimuli or toxic insults and regulates cellular functions such as transcription, translation, proliferation, growth, and survival	48	2.6×10−20	20	1.2×10−7	43	1.3×10−16
MAPK signaling pathway	The mitogen-activated protein kinase (MAPK) cascade is involved in cell proliferation, differentiation and migration.	43	2.9×10−21	19	8.4×10−9	36	1.8×10−15
Ras signaling pathway	RAS proteins are GTPases that function as molecular switches for pathways regulating cell proliferation, survival, growth, migration, differentiation or cytoskeletal dynamics	20	6.0×10−12	7	3.5×10−6	22	6.7×10−13
p53 signaling pathway	p53 activation is induced by a number of stress signals, including DNA damage. The p53 protein is employed as a transcriptional activator that results in three major outputs; cell cycle arrest, cellular senescence or apoptosis	11	5.3×10−7	3	1.4×10−2	9	4.3×10−7
Apoptosis	Apoptosis is a genetically programmed process for the elimination of damaged or redundant cells by activation of caspases.	17	2.0×10−11	8	3.4×10−5	7	1.1×10−2

Table 6:

Effect of by MC (1), DMC (2) and MC-lex (3) on p53 signaling pathway.

MC (1)		DMC (2)		MC-lex (3)
Gene Count	Overall effect	Gene Count	Overall effect	Gene Count	Overall effect
11	81% of targets promote p53 signaling log2(TP53)=1.64	3	67% of targets promote p53 signaling log2(TP53)=1	9	67 % of targets inhibit p53 signaling log2(TP53)=−2.09
Biological effect
Cell cycle arrest, cellular senescence or apoptosis are likely to be promoted via p53		Cell cycle arrest, cellular senescence or apoptosis are likely to be promoted via p53		Cell cycle arrest, cellular senescence or apoptosis are not likely to be promoted via p53

While there are shared biological targets and mechanisms impacted by some or all the mitomycins studied here, some specific targets and biological pathways are only affected by 3 (MC-lex). Our analysis showed that 3 causes changes in gene expression encoding proteins involved in integrity of cells and tissues (COL3A1, DKK1, LAMA5, PRKCA, FZD8, CTNNB1 and COL5A2) as well as others responsible for the maintenance of genome stability and cellular response to DNA damage (MDC1, NBN, TFDP1, RAD21, RAD51, CHEK1 and MLH1). As expected, 3 (MC-lex) also causes changes in gene expression involved in the regulation of apoptosis; proliferation and migration (BIRC7, SMAD4, NPM1, MAPK8, FEN1, IGF1R, DTX3, RET, PIM1, SYK, MAP2K1, BID, RASA4, and SHC1). The predicted pathways that are unique to 3 (MC-lex) are described in details and shown in supporting information (Table S8). They include pathways related to development (Notch signaling pathway) and cell and tissue structure such as the Hippo, ECM-receptor interaction, and Adherens junction signaling pathway. The Hippo and Adherens junction pathways share several of the same targets including SMAD3, SMAD4, and CTNNB1. These two pathways play important roles in maintaining structural cell integrity and interconnect with other mechanisms that regulate cell proliferation and growth. The last pathway that is uniquely affected by 3 (MC-lex) is the DNA replication pathway in which the chromosome maintenance complex components (MCM4 and MCM7), replication factors (RFC2, RFC3, RFC4 and RFC5) and the FEN1 endonuclease encoding genes display changes in expression compared to no treatment.

Bioinformatics analysis using IPA: cell cycle, DNA damage response and cell proliferation pathways are strongly downregulated by MC-lex (3)

We conducted further bioinformatics analysis using IPA on the main pathways related to cell cycle control, cell proliferation and DNA damage response. Figure 12 shows a hierarchical clustering of these main pathways. The variation in z score is represented in color (blue: downregulation, red: upregulation). All fold expressions and GO annotations for all genes subjected to quantitative IPA analysis as well as z-scores for each pathway are available in the supporting information section (Table S9 and S10). All gene expressions were normalized to untreated control. These data show that MC-lex derivative (3) has a very distinct activity in downregulating the expression of many genes involved in pathways mentioned in Fig. 12. Remarkably, MC-lex derivative (3) displays a stronger downregulation of major signaling networks that regulate cell cycle, DNA damage response and cell proliferation when compared to MC and DMC. These different regulations are depicted in detail for one pathway corresponding to each effect: Fig 13 shows the effect of the three drugs on the regulation of cell cycle through the “Role of CHK proteins in cell cycle checkpoint control”, Fig 14 shows the effect of the three drugs on the DNA damage response through the “G2/M DNA damage checkpoint regulation” and Fig. 15 shows the effect of the three drugs on cell proliferation through the “JAK/STAT” signaling. Other significant pathways are depicted in supporting information (Fig S12–18).

Figure 12:

Pathways involved in regulating cell cycle control, cell proliferation and DNA damage impacted by MC (1), DMC (2) and MC-Lex (3). IPA analysis displays the heat map for the biochemical pathways using the activation z-scores: up-regulated (orange, z-score >2) or down-regulated (blue, z-score < −2). Note: although some gene expression changes in two pathways related to DNA damage response were recorded (DNA damage-induced 14–3-3σ signaling and DNA methylation and transcriptional repression signaling), IPA was not conclusive regarding the up or down regulation of these pathways.

Figure 13:

View of IPA-predicted Role of CHK proteins in cell cycle checkpoint control showing MC (1), DMC (2) and MC-Lex (3) mediated quantitative changes in the gene expression profiles as compared with the untreated (control) samples. The gene expression profiles that were experimentally determined to be upregulated are depicted in red, and those that were downregulated are shown in green. Gene expression profiles predicted by IPA to be inhibited due to their networking relationship with the experimentally identified genes are shown in blue, while the genes that were predicted to be activated are shown in orange.

Figure 14:

View of IPA-predicted G2/M DNA damage checkpoint regulation showing MC (1), DMC (2) and MC-Lex (3) mediated quantitative changes in the gene expression profiles as compared with the untreated (control) samples. The gene expression profiles that were experimentally determined to be upregulated are depicted in red, and those that were downregulated are shown in green. Gene expression profiles predicted by IPA to be inhibited due to their networking relationship with the experimentally identified genes are shown in blue, while the genes that were predicted to be activated are shown in orange.

Figure 15:

View of IPA-predicted Jak/Sat signaling showing MC (1), DMC (2) and MC-Lex (3) mediated quantitative changes in the gene expression profiles as compared with the untreated (control) samples. The gene expression profiles that were experimentally determined to be upregulated are depicted in red, and those that were downregulated are shown in green. Gene expression profiles predicted by IPA to be inhibited due to their networking relationship with the experimentally identified genes are shown in blue, while the genes that were predicted to be activated are shown in orange.

DISCUSSION

Although MC is a potent antitumor antibiotic, its use has been restricted because of dose-limiting toxicity and delayed myelosuppression among other side effects (2). Over 1000 analogues of MC have been previously synthesized in efforts to prepare derivatives with improved therapeutic properties (36). Variations include direct modifications at N7, C6, C10, and N1a (Fig. 1) (38–40). However, to date, only a few derivatives have shown improved efficacy and/or decreased cytotoxicity, and none is commercialized yet.

Comparison of crosslinking and alkylating properties of the three Mitomycins

In most studies involving Mitomycin C derivatives, DNA adducts formed by synthetic compounds were not fully characterized and were presumed to have a trans stereochemical configuration at C1” (as in MC trans ICL 4a and monoadduct 4b, Fig. 2), inferring that the chemical modifications on the mitosane framework did not alter the side of DNA attack on the reduced mitomycin analogues during alkylation (Fig 16, first alkylation step: Guanine-N² nucleophilic attack was presumed to occur on the “alpha” face to generate “trans” adducts). This paradigm was contradicted by the finding that decarbamoylmitomycin C (DMC, Fig. 1), (a synthetic analogue of MC where the carbamoyl on O10 is replaced by a hydroxy group) generates predominantly adducts whose stereochemical configuration at C1” is cis rather than trans (Fig. 2) (8). In addition, the crosslinking of mitomycins was found to be diastereospecific: Trans-ICL are found at CpG steps whereas cis-ICLs are generated at GpC steps (10). This implies that DMC targets GpC sequences in mammalian DNA in contrast to MC, which targets CpG steps. The study presented here includes a mitomycin derivative (compound 3, Fig. 1) with a N-methylpyrrole carboxamide moiety appended at N7. Our results show that analogue 3, like MC, yields trans-adducts in its reaction with DNA. Since the crosslinking of DNA by mitomycins is sequence dependent, this also implies that 3 targets CpG sequences for crosslinking in mammalian cells. Additionally, our data demonstrate that MC and 3 crosslink plasmid DNA to a similar extent both at cold and biologically relevant temperature, whereas DMC crosslinking requires temperature higher than ~ 20°C (Fig. 7). This agrees with the established mechanistic for crosslinking by mitomycins where the second alkylation step is dependent on the expulsion of the leaving group on C10 (i.e. the carbamoyl on C10 present in MC and 3 is a better leaving group than the hydroxy group in DMC, Fig. 13). Lastly, treatment with DMC generates a greater frequency of monoadducts compared to ICLs than similar treatment with MC or compound 3 (Table 7, R=5.48 for DMC, 2.5 for 3 and 1.5 for MC). Overall, MC and 3 show strong similarities regarding their alkylation of DNA, while DMC alkylating behavior is markedly different.

Figure 16:

Alkylation mechanism for MC (R=CONH₂; R’=H), DMC (R, R’=H) and 3 (R=CONH₂; R’=methyl 1-methyl-1H-pyrrole-2-carboxylate).

Table 7:

Summary of the crosslinking properties and cytotoxicity of MC, DMC, and 3.

	MC	DMC	MC-lex
	(1)	(2)	(3)
Stereochemical configuration of major adducts in mammalian DNA (low CG content)	Trans *	Cis §	Trans
Stereochemical configuration of major adducts in M. Luteus DNA (high CG content)	Trans *	Trans §	Trans
Target sequence for crosslinking in mammalian DNA	CpG*	GpC§	CpG
Temperature for crosslinking	0°C	>20°C	0°C
R= (amount of major monoadduct generated)/(amount of major ICL generated)	1.50∥	5.48∥	2.5Δ
Toxicity toward wild type p53 cells (MCF-7, MCF 10A) versus mutant p53 (MDA-MB 468)	Higher	Lower	Lower

^(*):from reference 6.

^(§):from references 8,10,36.

^(∥):from reference 20, measured in MCF-7 cells.

^(Δ):calculated in this work as the ratio between the UV absorption of 6a and 6b at 577nm from the HPLC analysis of the cells’ digest (in MCF-7 cells).

Comparison of the drugs’ cytotoxicity toward wild type and p53 mutant breast cancer cell lines in relation to the stereochemical configuration of the major ICL they produce

Results from our study and prior research, summarized in Table 7, allow for a comparison between the crosslinking properties of the three mitomycins and their toxicity to MCF-7, MCF 10A (both wild type p53) versus MDA-MB 468 (mutant p53) cell lines. The cytotoxicity of MC and DMC in cell lines with either a wild type or mutant p53 has been compared in prior research. In these studies, DMC was consistently more cytotoxic in all cell lines regardless of their p53 status (18–21). Our results partially agree with these studies (Table 4 and Fig. 9). Here, we found that DMC is more cytotoxic than MC toward breast cancer cells with a mutant p53 (MDA-MB 468) but not in wild type p53 cells (MCF-7 and MCF 10A) (Table 4). This difference in results may be due to the difference in assay used: MTT assay in previous work versus neutral red assay here. Our experiments also show that the IC₅₀ of 3 (MC-lex) is overall lower for all cell lines than the IC₅₀ of other mitomycins studied here (Table 4). Interestingly, results indicate that the novel derivative 3 is more cytotoxic toward the triple negative breast cancer cell line MDA-MB 468 (p53 mutant) than MC or DMC (Table 4). DMC also shows a stronger cytotoxic effect on the MDA-MB 468 cell line, while MC lowest IC₅₀ corresponds to the MCF 10A normal epithelial cells. Furthermore, our survival analysis shows that at concentration lower than 20 μM, the loss of viability of MDA-MB 468 after exposure to 3 (MC-lex) is significantly larger than in the case of the other two cell lines. Derivative 3 is also less toxic toward non-cancer cell lines (MCF 10A) at lower concentrations (<7 μM) than any of the other mitomycins investigated here. This indicates that 3 may have improved anti-cancer efficacy and selectivity toward p53 mutant MDA-MB 468 cells compared to the other two mitomycins.

Figure 9:

Cytotoxicity curves of Mitomycin C (MC, 1), decarbamoyl mitomycin C (DMC, 2), and compound 3 (MC-lex) towards breast cancer cell lines (MCF-7 and MDA-MB 468) and a non-tumorigenic epithelial breast cell line (MCF 10A).

It was previously hypothesized that the stereochemical configuration at C1” (cis) of the major ICL produced by DMC may play a role in the toxicity of the drug toward cells with a mutant p53 by inducing a p53 independent form of cell death (18). Our study challenges this hypothesis. Although limited to three cell lines, our results suggest that the cytotoxicity of mitomycins toward mutant p53 cell lines is not directly nor solely related to the stereochemical configuration of their ICL, the sequence they target, or to the amount of crosslinking generated.¹

Comparison of changes in gene expression between the three drugs. Discussion on changes in gene expression and relevance to survival and proliferation.

To better understand the biological mechanisms that are impacted by treatment with 3, MC and DMC, we analyzed the change in gene expression levels after exposure to each drug in MCF-7 cells. We found that 285 targets (37%) showed differential expression after treatment with at least one of the mitomycins studied here. Thirty-three targets (10%) were affected after treatment with all mitomycins. Using pathway prediction analysis (DAVID) we found that over 60% of the pathways predicted to be impacted by each drug were shared by all three, which is likely due to the structural similarities between the three mitomycins. We analyzed in detail how the p53 signaling pathway is impacted by drug treatment. Results showed that, unlike MC and DMC, treatment with 3 downregulates TP53 (Table S9) and does not promote p53 signaling (Table 6); which implies that cell cycle arrest/cell death/senescence may be triggered through alternate pathways in response to DNA damage caused by 3. Since more that 50% of cancers have a p53 mutation, the mechanistic through which compound 3 is able to trigger these biological events without promoting p53 signaling will require further attention. Future work focusing on biological mechanisms mediated by compound 3 should also explore regulators of p53 such as MDM2. The MDM2 protein acts as an E3 ubiquitin ligase that binds and sequesters proteins for degradation. Although it has multiple targets, the most known is p53 (41–44). Here, we found that levels of MDM2 increase significantly after treatment with 3 (MC-lex), which could be the result of p53 activation of MDM2 expression or can result independently through the action of this mitomycin.

Bioinformatic analysis using IPA revealed that pathway regulations triggered by 3 is distinctly different than in the case of MC and DMC. Remarkably, compound 3 displays a stronger downregulation of major signaling networks that regulate cell cycle, DNA damage response and cell proliferation when compared to MC and DMC. Cell cycle analysis showed that the three compounds triggered a shift of MCF-7 cell cycle distribution pattern from the G2/M to the G1/G0 phase, however, IPA results show that this effect is mediated similarly by MC and DMC through the upregulations of most pathways involved in cell cycle control (except for the “G1/S Checkpoint Regulation” and Cyclins and “Cell Cycle Regulation pathways) whereas compound 3 triggers cell cycle arrest via downregulating most cell cycle related pathways except for the “”Cyclins and Cell Cycle Regulation” pathway. Compound 3 also downregulates all major pathways related to cell proliferation in contrast to MC and DMC. Biologically, this result may seem surprising since the three compounds (not just compound 3) are cytotoxic toward all cell lines studied here. These results warrant further investigation. The protein expression profile (and post translational modifications of these proteins) of these three drugs may explain the discrepancy between gene expression governing cell growth/proliferation and cytotoxicity in the case of MC and DMC. Finally, analysis of the “G2/M DNA damage checkpoint regulation”, a major signaling network that regulates DNA damage response, predicts that MCF-7 cells rely on the ATM/ATR-p53 or ATM/ATR-CHEK1 axis in response to MC and DMC induced DNA damage in contrast to what happens with compound 3. This aligns with the fact that treatment with 3 does not promote p53 signaling.

Unique signalling pathways activated by the novel MC derivative 3

Although 60% of the cellular pathways triggered by the three mitomycins are identical, our analysis using DAVID predicted that 3 altered distinct pathways through its effect on specific targets. Genes which are uniquely impacted after treatment with 3 correspond to (more commonly than for MC and DMC) targets encoding proteins involved in cell integrity and cell and tissue structure. This is exemplified in Notch signaling where 66% of target genes in this pathway are regulated to promote the inhibition of cell development (Table S8, supporting information). While not explored in depth, previous experiments suggest that MC treatment can affect cell migration in at least some cell lines (45). The extent of compound 3 effect in cell structure and migration needs to be confirmed in future experiments, however, the number of targets and predicted pathways identified here suggest this might be a relevant and specific biological mechanism of action for this drug. Other important pathways impacted after treatment with 3 are part of different DNA repair and replication pathways. An inability to repair damaged DNA plus a disruption in the DNA replication machinery might explain the specific high cytotoxicity resulting from the action of 3 in comparison to the other drugs included in this study.

Factors potentially responsible for the different biological responses observed

There are several potential explanations for the differences in changes of gene expression between the three mitomycins. Our favored hypothesis is that the different structure of the ICLs produced by the drugs are mostly responsible for these differences, particularly in all genes involved in DNA damage response since ICLs are the most cytotoxic lesions produced by these compounds. Interstrandcrosslinks (ICLs) have been characterized in the past as one type of DNA lesion. Although several studies have shown a direct correlation between the degree of DNA interstrand crosslinking and biological responses; both in vitro and in vivo (46–48), recent research has shown that the structure of the ICL itself has an effect on the processing of ICLs by the cellular machinery and therefore, impacts the cellular response (18–23).

Other factors, however, may also be responsible for the observed differences such as the structure and number of DNA-monoadducts generated by these drugs. In particular, DMC treatment results in 12 times more monoadducts than MC in its reaction with DNA in MCF-7 cells (10μM treatment) and treatment with DMC generates a greater frequency of monoadducts compared to ICLs than similar treatment with MC or compound 3 (Table 7). Changes in gene expression may also be due to the interaction of these drugs with other biological targets than DNA. This is supported by the fact that MC was shown to form adducts with ribosomal RNA (49) and enzymes (50) although the biological relevance of these adducts requires more investigation.

Regarding cytotoxicity, in agreement with previous research, our data show that the energy of the reduction step in the activation mechanism of 3, which displays a more electronegative group on C(7) than MC, is lowered by 3.82 kcals/mole (compared to MC). This ease of reduction is associated with an enhancement of the cytotoxicity of compound 3, a phenomenon that was observed previously with other easily reduced MC analogues (51,52), although no strict correlation between reduction potential and anti-tumor activity has been evidenced (53). Another reason for the increased cytotoxicity of compound 3 toward both cancer cell lines studied could be a greater cell uptake of compound 3 (compared to MC and DMC), a consequence of greater lipophilicity. Finally, the reduced form of 3 may have enhanced binding to DNA compared to MC/DMC, allowing it to target DNA more efficiently. In this report, we did not detect adducts resulting from the alkylation of DNA by inactive metabolites derived from compound 3, whereas these species alkylate DNA significantly in the case of MC (54).

Conclusion

In this study, we examined the crosslinking ability of three mitomycins (MC, DMC and compound 3) as well as the structure of the ICLs they produce, in correlation with their cytotoxicity. We first elucidated the structure of the major DNA adducts formed by the recently synthesized mitomycin-conjugate (3) whose mitosane moiety is linked to a N-methylpyrrole carboxamide. We found that overall, MC and compound 3 show strong similarities regarding their alkylation of DNA, while DMC alkylating behavior is strikingly different. The gene expression profiles of the three drugs were compared and we observed that compound 3, in contrast to MC and DMC 1) targets encoding proteins involved in cell integrity and cell and tissue structure and 2) does not promote p53 signaling in MCF-7, which suggests that cell cycle arrest/cell death/senescence may be triggered through alternate pathways in response to DNA damage caused by 3. This may also explain the high cytotoxicity of compound 3 toward p53 mutant MDA-MD 468 cell lines. Pathway analysis predicted that over 60% of the molecular pathways impacted by each drug were shared by all three mitomycins, but our data also demonstrate that there are unique effects associated with 3. Compound 3, remarkably, displays a stronger downregulation of major signaling networks that regulate cell cycle, DNA damage response and cell proliferation when compared to MC and DMC. Future experiments will have to explore in more depth the specific biological mechanisms affected by 3 to shed light into its potential use as a chemotherapeutic drug. The prediction analysis will be used as a guide for these biological experiments. They will include validation of the changes in gene expression, protein levels, and activity to confirm these initial findings. Future research should also explore how the response to a specific ICL only, rather than to the drug themselves, can lead to the over expression or inactivation of gene expression specifically.

Figure 8:

Decreased in tail moment (modified comet assay) after treatment of MCF-7 cells by MC (1), DMC (2) and MC-lex (3).

Table 1:

Cell cycle analysis of MCF-7 cells

Cell cycle profiles of MCF-7 cells treated with mitomycins at 50 μM for 24 hours
Chemical	Cell Cycle Phase
	G1/G0	S	G2/M
Control	61.9%	5.3%	32.8%
MC (1)	76.0%	2.1%	21.9%
DMC (2)	82.9%	1.9%	15.2%
MC-lex (3)	83.4%	1.9%	14.7%

The cell cycle analysis of untreated MCF-7 cells shows that most cells halted at the G1/G0 phase (61.9% cell population at the G1/G0 phase, 5.3% at the S phase and 32.8%, at the G2/M phase). All three testing compounds triggered a shift of MCF-7 cell cycle distribution pattern from the G2/M to the G1/G0 phase with DMC and MC-lex (3) showing more than 20% increase in the G1/G0 phase.

Table 2:

Cell cycle analysis of MDA-MB 468 cells

Cell cycle profiles of MDA-MB 468 cells treated with mitomycins at 50 μM for 24 hours
Chemical	Cell Cycle Phase
	G1/G0	S	G2/M
Control	50.7%	5.7%	43.6%
MC (1)	60.9%	6.8%	32.4%
DMC (2)	69.1%	5.4%	25.5%
MC-lex (3)	87.2%	1.5%	11.4%

The cell cycle analysis of untreated MDA-MB 468 cells shows a similar amount of cells halted at the G1/G0 phase and the G2/M phase (50.7% cell population at the G1/G0 phase, 5.7% at the S phase, and 43.6% at the G2/M phase). All three testing compounds triggered a shift of MDA-MB 468 cell cycle distribution pattern from the G2/M phase to the G1/G0 phase with MC-lex (3) showing more than 30% increase in the G1/G0 phase.

Table 3:

Cell cycle analysis of MDA-MB 468 cells

Cell cycle profiles of MCF-10A cells treated with mitomycins at 50 μM for 24 hours
Chemical	Cell Cycle Phase
	G1/G0	S	G2/M
Control	84.0%	8.1%	7.8%
MC (1)	42.8%	10.2%	47.0%
DMC (2)	20.5%	4.7%	74.8%
MC-lex (3)	23.2%	3.8%	73.0%

The cell cycle analysis of untreated MCF-10A cells shows that most cell halted at the G1/G0 phase (84.0% cell population at the G1/G0 phase, 8.1% at the S phase, and 7.8% at the G2/M phase). All three testing compounds triggered a shift of MCF-10A cell cycle distribution pattern from the G1/G0 phase to the G2/M phase with DMC and MC-lex (3) showing more than 65% increase in the G2/M phase.

Highlights.

• We elucidated the structure of major DNA-adducts produced by a novel mitomycin C derivative.

• We compared the DNA crosslinking abilities of three mitomycins: Mitomycin C, decarbamoylmitomycin C and a novel mitomycin C analogue.

• We carried out cytotoxic assays with the three drugs on three different cell lines.

• We performed high throughput gene expression and gene ontology analysis with the three mitomycins on MCF-7 cells.

• We identified both gene expression changes and cellular pathways most impacted by each drug treatment.

• We discovered that the novel mitomycin C analogue specifically alters the expression of proteins involved in in cell integrity and cell and tissue structure and displays a stronger downregulation of major signaling networks that regulate cell cycle, DNA damage response and cell proliferation when compared to MC and DMC.