|Home | About | Journals | Submit | Contact Us | Français|
Glutathione S-transferase (GST) and multidrug resistance-associated proteins (MRPs) play major roles in drug resistance in melanoma. In this study, we investigated caffeic acid phenethyl ester (CAPE) as a selective GST inhibitor in the presence of tyrosinase, which is abundant in melanoma cells. Tyrosinase bioactivates CAPE to an o-quinone, which reacts with glutathione to form CAPE-SG conjugate. Our findings indicate that 90% CAPE was metabolized by tyrosinase after a 60-min incubation. LC–MS/MS analyses identified a CAPE-SG conjugate as a major metabolite. In the presence of tyrosinase, CAPE (10–25 µM) showed 70–84% GST inhibition; whereas in the absence of tyrosinase, CAPE did not inhibit GST. CAPE-SG conjugate and CAPE-quinone (25 µM) demonstrated ≥85% GST inhibition via reversible and irreversible mechanisms, respectively. Comparing with CDNB and GSH, the non-substrate CAPE acted as a weak, reversible GST inhibitor at concentrations >50 µM. Furthermore, MK-571, a selective MRP inhibitor, and probenecid, a non-selective MRP inhibitor, decrease the IC50 of CAPE (15 µM) by 13% and 21%, apoptotic cell death by 3% and 13%, and mitochondrial membrane potential in human SK-MEL-28 melanoma cells by 10% and 56%, respectively. Moreover, computational docking analyses suggest that CAPE binds to the GST catalytic active site. Caffeic acid, a hydrolyzed product of CAPE, showed a similar GST inhibition in the presence of tyrosinase. Although, as controls, 4-hydroxyanisole and l-tyrosine were metabolized by tyrosinase to form quinones and glutathione conjugates, they exhibited no GST inhibition in the absence and presence of tyrosinase. In conclusion, both CAPE and caffeic acid selectively inhibited GST in the presence of tyrosinase. Our results suggest that intracellularly formed quinones and glutathione conjugates of caffeic acid and CAPE may play major roles in the selective inhibition of GST in SK-MEL-28 melanoma cells. Moreover, the inhibition of MRP enhances CAPE-induced toxicity in the SK-MEL-28 melanoma cells.
Melanoma is the most lethal cancer of skin, and the number of melanoma cases has doubled in the past 20 years . Surgical resection can be used to cure malignant melanoma in early stages, but once it has metastasized to other organs, chemotherapy is the only option available with limited success partly due to drug resistance. Among all the reasons that are significantly responsible for melanoma cancer drug resistance, the over-expression of glutathione S-transferase (GSTs) and multidrug resistance-associated proteins (MRPs) may have critical roles [2–4].
GST (glutathione S-transferase) is an important enzyme in detoxification of a broad range of compounds. An important role of GST is to biotransform xenobiotics and other endogenous toxic compounds. It initiates the conjugation of hydrophobic electrophilic toxic substances including drugs, carcinogens, herbicides, and insecticides with the tripeptide glutathione (GSH) . Over expression of GST may increase detoxification and circumvent the cytotoxic action of anticancer agents leading to multi-drug resistance (MDR) . For instance, the current alkylating agents for cancer therapy are substrates for GST in tumor which leads to the development of multi-drug resistance (MDR) . GSTs also play a role in the detoxification of superoxides, peroxides, and hydroxyl radicals . GST utilizes GSH to scavenge the toxic reactive xenobiotics, which are responsible for the production of oxidative stress and cell toxicity; this is one of the important parts of the defense mechanism against carcinogenic and toxic effects of toxic compounds . In a study of GST expression, it was shown that GST is highly expressed in melanoma cells when compared to the normal cells .
Furthermore, the co-expression of MRPs with GSTs may play a major role in protection of cancer cells from anticancer agents [8,9]. It was previously reported that MRP proteins are responsible for the active transport across biological membrane . MRPs were also shown to confer the resistance to several vinca alkaloids, anthracyclines and epipodophyllotoxins . Moreover, it was reported that the detoxification of several anti-neoplastic agents was due to mixed action of both GSTs and MRPs [8,9,12]. It was also shown that human melanoma cells express high levels of both GSTs and MRPs [2,13].
To enhance selective drug delivery to melanoma, we have recently used tyrosinase, as a primary molecular target for bioactivation of caffeic acid phenethyl ester (CAPE) [1,14]. This is because tyrosinase is over-expressed and up-regulated in melanoma . CAPE is an ester analog of caffeic acid and is an active component of propolis . CAPE also exhibits antibacterial, anti-inflammatory, anti-viral, and anti-cancer properties. Recently, we investigated CAPE selective toxicity towards melanoma cells [1,14]. Bioactivation of CAPE by melanoma tyrosinase leads to the formation of quinone which causes selective toxicity towards melanoma cells compared to non-melanoma cell lines [1,14].
In the current study, our aims were to investigate CAPE and its quinone and glutathione conjugate metabolites, which are formed as a result of the CAPE’s bioactivation by tyrosinase, as selective inhibitors of GST in melanoma cells, as a secondary molecular target, compared to non-melanoma cells, which do not express tyrosinase. The natures of GST inhibition, including reversibility, irreversibility, competitive and non-competitive inhibitions were investigated. We also tested the effect of MK-571, a selective MRP inhibitor  and probenecid, a non-selective inhibitor of MRP protein , in combination therapy with CAPE in human SK-MEL-28 melanoma cells.
Glutathione (GSH), 1-chloro-2,4-dinitrobenzene (CDNB), and all other materials, solvents and reagents used in this work were analytical grade with the highest degree of purity and were purchased either from Sigma-Aldrich, St. Louis, MO or Fisher-Scientific, Pittsburgh, PA. Glutathione S-transferase (GST) was purchased from Sigma-Aldrich (Cat. No. G8642). The isozyme is identified as hGSTP1-1 with 26 U/mg solid (55 U/mg protein). One unit will conjugate 1.0 µmol of CDNB with reduced glutathione per min at pH 6.5 at 25 °C and the source of this enzyme is human placenta [19,20]. Mushroom tyrosinase was used throughout this study, as the purified human tyrosinase is unavailable commercially. Mushroom tyrosinase was purchased from Sigma-Aldrich (Cat. No. T3824), with 4276 U/mg solid. One unit of tyrosinase leads to ΔA280 nm of 0.001 per in at pH 6.5 at 25 °C in 3 mL reaction mix containing l-tyrosine, with an isoelectric point of 4.7–5.0. Because the compounds were dissolved in DMSO, the final concentration of DMSO was 1% v/v in cell culture media of the cells treated with drugs. Therefore, the media for control cells contained 1% v/v DMSO in the experiment. Phosphate buffered saline (PBS) was used as a vehicle to dissolve glutathione.
Modified Eagle Medium Alpha (MEM) (1×) (Cat. No. 32571-036), Leibovitz’s L-15 Medium (Cat. No. 30-2008), fetal bovine serum (FBS) (Cat. No. 10082-139) were purchased from American Type Culture Collection (ATCC®), Manassas, VA Versene (1×, 0.2 g EDTA 4Na/L in phosphate buffered saline) (1:5000 Cat. No. 15040-066) were purchased from Invitrogen, Grand Island, NY. RPMI medium 1640 (1×) (Cat. No. 11875-119) was obtained from Invitrogen Corporation, Grand Island, NY. Human SK-MEL-28 cell line was obtained from ATCC®, Manassas, VA.
LC/MS–MS method  was used to identify a CAPE glutathione (CAPE-SG) conjugate. Tyrosinase (20 U/mL) was added to a mixture of CAPE (62.5 µM) and glutathione (200 µM) in a final volume of 2 mL phosphate buffer (0.1 M, pH 6.5). The mixture was incubated for 5 min at 37 °C. The CAPE-SG conjugate was extracted using solid phase strata-X 33 µm polymeric reversed phase extractor (Cat. No. 8B-S100-TAK, Strata, Torrance, CA, USA). Activation of the solid phase reversed phase extractor was performed by using 1 mL of water, 1 mL of methanol and 1 mL of water in sequence. One milliliter of the reaction solution was run through solid phase reversed phase extractor under a stream of nitrogen gas. Then, the solid phase extractor was washed with 100 µL water. The product was collected using 500 µL of methanol. The solvent was subsequently evaporated under a stream of nitrogen gas at 37 °C on a hot plate (N-Evap, Organomation Associates Inc., Berlin, USA). Five hundred microliter of acetonitrile was added to dissolve the sample. A similar procedure was performed for control solution containing CAPE and GSH in the absence of tyrosinase.
The above sample was analyzed by electrospray tandem mass spectrometry in positive mode with a Varian 1200L triple quadrupole mass spectrometer (Varian Inc., Palo Alto, CA, USA). Direct infusion of samples into the mobile phase stream was used to detect protonated parent ions and to determine precursor-product ion transitions. An isocratic flow of 50% water with 0.1% formic acid, and 50% acetonitrile at 0.1 mL/min was used for chromatographic separation on a Synergi 4u Max-RP column (75 × 2.1 mm) (Varian Inc., Lake Forest, CA, USA) protected by a Synergi 4u Max-RP guard column (Varian Inc., Lake Forest, CA, USA). The electrospray housing temperature was set at 40 °C, and the nitrogen drying gas temperature was held at 250 °C. Argon was used as the collision gas at 1.4 mTorr pressure, with collision energies between 8 and 22.5 V, depending on the precursor-product ion transition. The capillary voltage was 56 V and the detector multiplier voltage was 1500 V. Varian MS Workstation version 6.9 software was used for data acquisition and processing. Using selective/multiple reaction monitoring technique, each sample reaction was monitored for detection of parent ion m/z= 590 for CAPE-SG and parent ion m/z= 285 for CAPE, and their respective predicted daughter ions on the LC/MS/MS detector. Separate samples of CAPE and GSH were used as controls to predict possible daughter ions for CAPE-SG in selective/multiple reaction monitoring mode.
GST (0.1 U/mL) was added to a mixture of CAPE (50, 100 and 250 µM) and GSH (500 µM) in a final volume of 2 mL phosphate buffer (0.1 M, pH 6.5). The mixture was pre-incubated for 15, 30 and 60 min at 37 °C. A 250 µL aliquot was added to trichloroacetic acid (25 µL; 30% w/v), vortexed and left at room temperature for 5 min. A 100 µL aliquot of the supernatant was added to a mixture of Ellman’s reagent (DTNB) (62.5 µL; 2 mg/mL) and Tris/HCl buffer (875 µL; 0.1 M, pH 8.9), and then vortexed. The absorbance of the solution was monitored at 412 nm . Glutathione consumption was used as a marker to evaluate if CAPE were a substrate for GST. Similar experiments were performed on 4-HA, CA, and tyrosine to evaluate them as GST substrates. CDNB was used as a positive control. Ellman’s reagent (DTNB) method was used to measure the extent of GSH consumption as previously described [1,14,23].
UV–Vis spectroscopy method was used to elucidate the inhibition of GST by CAPE. Human placenta GST inhibition was assayed by the method of Tuna et al. . For a typical assay, the reaction mixture of 2.5 mL contained potassium phosphate buffer (100 mM, pH 6.5), glutathione (1 mM) and CDNB (200 µM), with 0.05 U/mL of human placenta GST . Various concentrations of CAPE (10, 25 and 50 µM) were added to the above reaction mixture to investigate the inhibition of GST. The cuvette containing the reaction mixture had a water-jacket supported by a pump circulating water at a constant temperature of 25 °C. The time-dependent changes in A340 nm were recorded for 6 min using a GBC UV–Vis spectral spectrophotometer (GBC Scientific, Victoria, Australia). The absorbance difference between 6 and 1 min was used to calculate the extent of GST inhibition. Absorbance was less than 1.0 at 6 min. The GST activity was calculated using the following formula: GST activity (mM/min) = [(ΔA340 nm/min of sample) – (ΔA340 nm/min of blank)] divided by 9.6 mM−1 cm−1 (extinction coefficient of DNP-SG at A340 nm) . Where ΔA was the absorbance difference between from 6 and 1 min at 340 nm. However the unit was converted to µM/min (nmol/mL/min) to graph Lineweaver–Burk plot. All experiments were determined in triplicate in buffers equilibrated at constant room temperature. Similar procedures were performed for 4-hydroxyanisole (4-HA), caffeic acid (CA), and tyrosine.
In order to investigate the inhibition of GST by CAPE-quinone, tyrosinase (20 U/mL) was added to a mixture of CAPE (25 µM) and GST (0.1 U/mL) in a final volume of 2.5 mL phosphate buffer (0.1 M, pH 6.5). The mixture was incubated for 30 min at 37 °C to permit CAPE metabolism to CAPE-quinone by tyrosinase. To investigate the reversible or irreversible nature of GST inhibition by CAPE-quinone, mixture was filtered through Millipore centrifugal filter units with 10 K molecular weight cutoff to isolate GST from the reaction mixture (UFC801024, Amicon Ultra, Carrigtwohill, Ireland). Then, glutathione (1 mM) and CDNB (200 µM)  were added to the solution recovered from the Millipore centrifugal filter unit and made the final volume to 2.5 mL with phosphate buffer (100 mM, pH 6.5). The cuvette containing the reaction mixture had a water-jacket supported by a pump circulating water at a constant temperature of 25 °C. This mixture was used to investigate the irreversible inhibition of GST by CAPE-quinone using UV–Vis spectroscopy method as mentioned in Section 2.5. Control solutions containing GST or GST/tyrosinase were also used to assess the recovery for GST enzyme activity in the absence of tested compounds. The percentage inhibition of the enzyme activity by various inhibitors was calculated by comparing the results with GST activities in the controls. A similar approach was used to investigate the nature of GST inhibition by caffeic acid (CA).
CDNB method was used to measure GST activity . This experiment was performed to investigate the reversible and irreversible nature of GST inhibition by CAPE-SG conjugate. Tyrosinase (20 U/mL) was added to a mixture of CAPE (25 µM), glutathione (100 µM) and GST (0.1 U/mL) in a final volume of 2.5 mL phosphate buffer (0.1 M, pH 6.5). The mixture was incubated for 30 min at 37 °C to allow the formation of CAPE-SG conjugate in the presence of tyrosinase and glutathione. The rest of experiment was similar to what described in the previous section. A similar approach was used to investigate the nature of GST inhibition by caffeic acid glutathione conjugate.
CDNB method  was used to investigate the nature of GST inhibition and the inhibition constant (Ki)  of CAPE, CAPE quinone, and CAPE-SG conjugate by Lineweaver–Burk plots. Briefly, tyrosinase (20 U/mL) was added to a mixture of CAPE (10 µM), and GSH (1 mM) in phosphate buffer (0.1 M, pH 6.5). The mixture was incubated for 30 min at 37 °C to allow the formation of CAPE-SG conjugate. Then, GST (0.02 U/mL), CDNB (0.2–1 mM) were added to the solution mixture. The cuvette containing the reaction mixture had a water-jacket supported by a pump circulating water at a constant temperature of 25 °C. The absorbance at 340 nm was monitored for 6 min using UV–Vis spectroscopy method at 340 nm. The change in absorbance between 6 and 1 min was used as the indication of the reaction rate. The GST activity was calculated as described in Section 2.5. All the experiments were performed on three different days using freshly prepared GST, CDNB, CAPE, GSH and tyrosinase stock solutions on a daily basis. The averages of data points were used to graph the Lineweaver–Burk plot to determine the nature of competitive and non-competitive inhibition. The Ki values were determined according to a method published previously . The control reaction contained GST (0.02 U/mL), GSH (1 mM) and CDNB (0.2–1 mM). The concentration of CDNB was 1 mM, when the nature of GST inhibition was investigated with respect to GSH (0.2–1 mM).
Similar experiments were performed to investigate the nature of GST inhibition with respect to GSH and CDNB by CAPE-quinone (10 µM), CAPE (100 µM) and EA (20 µM) with concentrations of GSH and CDNB ranging from 0.2 to 1 mM as discussed above. The inhibitory constants Ki of competitive and non-competitive inhibitions were calculated from following formulas .
Competitive Inhibition :
Non-competitive Inhibition :
Mixed Inhibition :
where Km and values are Michaelis–Menton constants, Vmax and values are maximum reaction rates, and Ki is the inhibition constant. Km and Vmax values were obtained in the absence of the inhibitor. and were obtained in the presence of inhibitors (CAPE, CAPE-SG, CAPE-quinone, or EA). [I] is the concentration of the inhibitor.
The cell culture experiments were performed as described previously . Human SK-MEL-28 melanoma cells were cultured in MEM Alpha media supplemented by 10% FBS. After reaching a confluency of 90%, the cells obtained were used for cytotoxicity studies.
CDNB method was used to investigate the GST inhibition in melanoma cells [24,27]. A suspension of 200,000 cells in 500 µL MEM Alpha media supplemented with 10% FBS was used for this study. The cell suspension was sonicated for 3–5 s and then centrifuged for 15 min at 13,000 rpm. The supernatant was taken into a tube to which glutathione (1 mM), CAPE (15, 25, 50, 100 µM), CDNB (1 mM) and tyrosinase (10 U/mL) were added in a final volume of 2.5 mL phosphate buffer (100 mM, pH 6.5). The reaction mixture was used to investigate the inhibition of GST in human SK-MEL-28 melanoma cells by CAPE using UV–Vis spectroscopy method as described in Section 2.5. A similar procedure was used to study GST inhibition in the presence of 4-hydroxyanisole (4-HA), caffeic acid (CA), and tyrosine.
The cytotoxicity assay was evaluated using yellow tetrazolium dye (3-(4, 5-dimethylthiazolyl-2)-2,5-diphenyl tetrazolium bromide) (MTT) [28,29]. SK-MEL-28 cells were obtained from 90% to 95% confluent cell cultures, seeded at 40,000 cells/well in 24 well plate, and grown in 500 µL fresh MEM Alpha media (supplemented by 10% FBS). After 24 h of seeding, cells were treated with CAPE, CA, 4-HA and tyrosine at 15, 30, 50 and 100 µM for 48 h before measuring the cell viability using MTT assay . An equal volume of DMSO was added to control cells so that final concentration of DMSO was 1%. An analysis of variance (ANOVA) was carried out to compare the percentage of surviving cells for each compound at various concentrations followed by Bonferoni’s post t-test.
The biochemical mechanism of CAPE toxicity in SK-MEL-28 cells was performed using various modulators  using MTT assay as described in the previous section. After 24 h of cell seeding, ethacrynic acid (EA) (2 µM) a GST inhibitor [30,31], and MK-571 (10 µM), a selective MRP inhibitor  and probenecid (500 µM), a non-selective MRP inhibitor  were added to cell culture media. After 1 h cells were treated with CAPE (15 and 30 µM) for 48 h. MTT assay was used to assess the effect of modulators on CAPE toxicity as described in Section 2.11.
Apoptotic cells were identified by FITC-conjugated Annexin V and propidium iodide using Annexin V-FITC apoptosis assay kit (Cat. No. PF032, Calbiochem, La Jolla, CA, USA) according to a method published by us previously . Briefly, SK-MEL-28 cells were seeded at 160,000 cells/well in six-well plates in MEM media and allowed to attach overnight. After seeding, MK-571 (10 µM), a selective MRP inhibitor  and probenecid (500 µM), a non-selective MRP inhibitor  were added to the cell culture media. After 1 h, CAPE (15 and 30 µM) was added to the wells. 48 h after incubation, cells were detached using versene solution (300–350 µL) (Cat. No. 15040, GIBCO, Carlsbad, CA). Collected cells were incubated with 5 µL of Annexin FITC and 5 µL of propidium iodide in 100 µL of binding buffer for 15 min as described in Calbiochem Annexin V-FITC apoptosis assay kit (Cat. No. PF032, Calbiochem, La Jolla, CA, USA) . The experiments were performed in triplicates.
To investigate the role of MRP in the biochemical mechanism of CAPE toxicity in SK-MEL-28 cells, mitochondrial membrane potential was determined in the presence and absence of MK-571, a selective MRP inhibitor  and probenecid, a non-selective MRP inhibitor , using tetramethyl rhodamine methyl ester (TMRM) fluorescent dye . Using a method published by us previously . Briefly, SK-MEL-28 melanoma cells were seeded at 160,000 cells/well in six well plates in MEM media and allowed to attach overnight. After seeding, MK-571 (10 µM) and probenecid (500 µM) were added to the respective wells 1 h prior to the addition of CAPE (15 and 30 µM). After 48 h incubation, 20 µL (50 nM) tetramethyl rhodamine methyl ester (TMRM) (Cat. No. T668, Invitrogen A/S, Taastrup, Denmark) was added . The stained cells were analyzed using flow cytometer (Accuri C6 flow cytometer, Ann Arbor, MI). The 10,000 events were acquired. Arithmetic mean values of fluorescence signal in arbitrary units were determined for each sample in triplicate.
AutoDock tools 1.5.2 and AutoDock 4.0 [34,35] were used to predict possible interactions between GST and CAPE or EA. During the calculations, the protein and ligands were treated as rigid groups. The predicted bound configuration with the lowest free energy was chosen for each ligand. Crystal structures of both wild-type GST (PDB ID 11GS) [36,37] and the C47S/Y108V GST double mutant (PDB ID 3KM6)  were applied for this study.
Results from three replicate experiments have been reported as mean ± SD. An analysis of variance (ANOVA) was carried out to compare the percentage of surviving cells and percentage of GST inhibition for each compound at various concentrations followed by Bonferoni’s post t-test.
LC–MS/MS was used to identify the oxidized product of CAPE. The retention times on HPLC for the CAPE-SG conjugate and CAPE were 2.2 and 7.8 min, respectively (Fig. 1A and B). This was expected as CAPE-SG conjugate should be more hydrophilic than CAPE; and hence to elute faster. Using selective/multiple reaction monitoring, the parent ions were monitored for CAPE-SG and CAPE, simultaneously. Fig. 1A and B illustrate two overlaid detection windows for m/z = 590 (CAPE-SG) peak and m/z = 285 (CAPE) peak on the LC/MS/MS detector. There was no peak observed at 2.2 min when CAPE and glutathione incubated in the absence of tyrosinase implying that the peak at 2.2 min was formed only after CAPE bioactivation by tyrosinase and a subsequent reaction with glutathione (Fig. 1A and B). The 2.2 min peak was not formed when CAPE was incubated with tyrosinase (data not shown). The CAPE incubation with GSH also showed one peak at 7.8 min indicating that CAPE does not react with GSH in the absence of tyrosinase. These results indicated that incubation of CAPE, tyrosinase, and glutathione formed a major product that was eluted at 2.2 min. To characterize the product, LC–MS/MS analysis of parent ion was carried out. Further analysis of the peak at 2.2 min using tandem mass spectrometry in positive ion mode indicated a mono CAPE–SG conjugate at m/z [MH]+ 590. Individual samples of CAPE and GSH were used as controls to predict possible daughter ions for CAPE-SG conjugate in selective/multiple reaction monitoring using LC–MS/MS analyses. Subsequent LC–MS/MS analyses of the parent signal [MH+] = 590 m/z exhibited parent CAPE-SG conjugate ion at m/z 590 [MH]+ and daughter ions at 515 [M-glycine]+, 468 [M-phenethyloxy]+, 461 [MH-glut+H]+, 393 [M-phenethyloxy-glycine+H]+, 264 [M-phenethyloxy-glycine-glu]+, and 145 [glut+NH]+ (Fig. 1C).
GSH consumption was used as a biomarker to evaluate CAPE, CA, 4-HA, and tyrosine as substrate for GST. The study found that none of these tested compounds, including CAPE, 4-HA, tyrosine, and CA, was a substrate for GST. CDNB was reported previously to be a substrate of GST  and was used as a positive control. On a molar basis, 0.6 mol glutathione was consumed per mole of CDNB, when CDNB was metabolized by GST at 60 min incubation.
CAPE alone did not inhibit GST activity at concentrations <25 µM; however, it marginally inhibited GST activity by 13% at a higher concentration of 50 µM (Fig. 2A). Caffeic acid (Fig. 2B), a hydrolyzed product of CAPE, 4-HA, a substrate for tyrosinase  and tyrosine, a natural substrate of tyrosinase  did not show any inhibition of GST at concentrations of 10–50 µM. In contrast, CAPE-quinone, formed by bioactivation of CAPE in the presence of tyrosinase was a potent GST inhibitor, which decreased the human placenta GST activity by 70% and 93% at concentrations 10 and 50 µM, respectively (Fig. 2A). Similarly, it was found that caffeic acid-quinone at concentrations of 10–50 µM inhibited GST activity by 23–67% (Fig. 2B), whereas 4-HA-quinone (50 µM) and tyrosine–quinone (50 µM) showed no significant GST inhibition (data not shown).
Interestingly, it was found CAPE-SG conjugate 10–50 µM, formed as a result of CAPE bioactivation by tyrosinase in the presence of glutathione, inhibited GST activity by 68–96% (Fig. 2A). Similarly, caffeic acid glutathione (CA-SG) conjugate also inhibited GST activity by 19–61% (Fig. 2B). Ploemen et al. also reported similar findings on CA-SG conjugate . In contrast, neither 4-HA-SG conjugate nor tyrosine-SG conjugate inhibited GST activity (data not shown). The order of the GST activity inhibition for CAPE in descending order was CAPE-quinone ≥ CAPE-SG conjugate >>>> CAPE. The order of GST activity inhibition for caffeic acid, a hydrolyzed product of CAPE, in descending order was CA-Quinone > CA-SG conjugate >>>> CA (Fig. 2).
The 10 K Millipore filter was used to separate GST from the reaction mixture. Although CAPE-SG conjugate (25 µM) showed significant GST inhibition (Fig. 3A), the activity of GST was recovered after filtering the reaction mixture through 10 K Millipore filter (Fig. 3B), indicating that CAPE-SG conjugate inhibited GST in a non-covalent binding fashion that were reversible. As shown, CAPE-quinone inhibits GST significantly (Fig. 3A). In contrast, when the reaction mixtures were filtered through 10 K Millipore filter, the recovered GST from the filter did not show enzymatic activity (Fig. 3B), suggesting that CAPE-quinone inhibited GST through irreversible covalent binding. Tyrosinase alone did not have an inhibition effect on GST (data not shown). A similar result was observed for CA-SG and CA-quinones indicating that CA-SG inhibited GST reversibly whereas CA-quinone inhibited GST irreversibly (Fig. 3).
In order to evaluate the nature of GST inhibition by CAPE, CAPE-quinone, CAPE-SG conjugate and EA, the activity of GST enzyme was measured with respect to CDNB (0.2–1 mM) and GSH (0.2–1 mM). As shown in (Fig. 4A and B), the reciprocal plots of 1/[rate] versus 1/[CDNB] have the inhibitor and no inhibitor lines intercepting above the 1/[CDNB] axis, suggesting that CAPE and CAPE-quinone are mixed inhibitors of GST with respect to CDNB with an apparent Ki and of 233 and 451 µM for CAPE, and 2.7 and 8.8 µM for CAPE-quinone (Fig. 4A and B and Table 1). However, CAPE-SG conjugate and EA inhibited GST via a competitive mechanism as the “no inhibitor” and “inhibitor” lines intercepted on the [1/rate] axis with an apparent Ki of 3.1 and 3.0 µM, respectively (Fig. 4C and D and Table 1).
With respect to GSH (0.2–1 mM, Fig. 4E and G), CAPE and CAPE-SG conjugate showed mixed inhibitions of GST with an apparent Ki and of 942 and 2990 µM for CAPE, respectively, and 2.6 and 14 µM for CAPE-SG conjugate, respectively. However, CAPE-quinone acted as a competitive inhibitor of GST with an apparent Ki of 0.7 µM (Fig. 4F and Table 1). On the other hand, EA acted as a non-competitive inhibitor of GST with an apparent Ki of 12 µM (Fig. 4H and Table 1).
In the presence of tyrosinase, CAPE (15–100 µM) showed considerable and selective inhibition of GST in human SK-MEL-28 melanoma cell homogenate resulting in 61–86% inhibition (Fig. 5). Similarly, caffeic acid (15–100 µM) showed 21–53% GST inhibition whereas 4-HA (15–100 µM) showed only 4–18% GST inhibitions. Tyrosine (15–100 µM) did not show any significant GST inhibition (Fig. 5). These findings suggest that CAPE and its hydrolyzed product caffeic acid were significantly more potent in inhibiting GST in human SK-MEL-28 melanoma cell homogenate than 4-HA and tyrosine.
MTT assay was performed to evaluate the anti-proliferative effects of CAPE, CA, 4-HA and tyrosine. CAPE showed significant cytotoxic effects towards human melanoma SK-MEL-28 cells in comparison to CA, 4-HA and tyrosine. After 48 h incubation, CAPE (15 µM) showed 48% cell toxicity towards SK-MEL-28 melanoma cells whereas CA, 4-HA and tyrosine at 30 µM demonstrated 2–27% cell toxicity (data not shown). The IC50 (48 h) values of CAPE and 4-HA were 15 and 60 µM in SK-MEL-28 melanoma cells, respectively, whereas the IC50 (48 h) values of CA and tyrosine were 2.3 mM and >5 mM, respectively. These findings suggested that CAPE was more effective as a cytotoxic agent towards SK-MEL-28 melanoma cells than other tested compounds.
To investigate the biochemical mechanism of CAPE toxicity in SK-MEL-28 melanoma cells, the cell viability was determined in the presence and absence of modulators using MTT assay, Annexin V-PI apoptosis assay, and mitochondrial membrane potential measurement.
UsingMTT assay,MK-571 (10 µM),aselective MRP inhibitor  and probenecid (500 µM), a non-selective MRP inhibitor , decreased the IC50 value of CAPE (15 µM)by13% and 21%, respectively, and of CAPE (30 µM) by 12% and 9%, respectively (Table 2). These findings suggest that CAPE or its cytotoxic metabolites are substrate for MRP and hence can be detoxified by this efflux pump. Therefore, the inhibition of MRP may enhance CAPE induced toxicity in SK-MEL-28 cells. On the other hand, ethacrynic acid (EA) (2 µM), a GST inhibitor [30,31] did not affect the IC50 value of CAPE towards melanoma cells (Table 2). This could be because both CAPE-quinone and CAPE glutathione conjugate are potent GST inhibitors (Table 1), which may explain why ethacrynic acid did not enhance CAPE toxicity in melanoma cells.
Using Annexin V apoptosis assay kit, MK-571 (10 µM) and probenecid (500 µM), MRP inhibitors, increased the CAPE (15 µM) induced apoptotic cell death by 3% and 13%, respectively. MK-571 and probenecid also increased the CAPE (30 µM) induced apoptotic cell death by 7% and 16%, respectively (Fig. 6). Probenecid (500 µM), a non-selective MRP inhibitor, approximately doubled the CAPE-induced apoptotic cell death at 15 or 30 µM (Fig. 6).
The combination of MK-571 and probenecid with CAPE further decreased mitochondrial membrane potential in SK-MEL-28 cells for CAPE (15 µM) by 10% and 56%, respectively and for CAPE (30 µM) by 29% and 68%, respectively (Fig. 7). Probenecid (500 µM), a non-selective MRP inhibitor, approximately increased the effect of CAPE (15–30 µM) on mitochondrial membrane potential for 8–11-fold. On the other hand, MK-571 (10 µM), a selective MRP inhibitor, approximately increased the effect of CAPE (15–30 µM) on mitochondrial membrane potential for 3–4-fold. The modulators had no significant toxicity effect when incubated alone with SK-MEL-28 cells (Table 2 and Figs. 6 and and77).
Computer prediction of the interactions between wild-type GST and CAPE or EA showed that the bound CAPE overlaps with both glutathione-binding site (G site) and substrates-binding (H site) of the enzyme (Fig. 8E); while bound EA only overlaps with the H site (Fig. 8C). Similar results were obtained from the wild type GST and C47S/Y108V GST mutant (Fig. 8D and F). The molecular modeling also revealed that the binding sites of CAPE and EA could potentially overlap, which may explain why EA, a GST inhibitor, did not enhance CAPE toxicity in melanoma cells.
CAPE is a caffeic acid phenethyl ester derivative which is a naturally occurring hydroxycinnamic acid found in propolis . It is also known to be highly stable in human plasma . In our earlier work, we have used tyrosinase as a primary molecular target to bioactivate CAPE to cytotoxic agents [1,14]. We also showed that CAPE was metabolized by tyrosinase to CAPE-quinone, depleted GSH, and selectively toxic towards melanoma cells in vitro and in vivo [1,14].
In the current work, we have used GST as a second molecular target in melanoma as GST plays an important role in multidrug resistance [9,43] due to the over expression in melanoma [2,13]. GST is one of the essential targets in the current development of cancer therapy because of resistance to various anticancer agents . GST also has a vital role in glutathionylation of cellular proteins in cancer, which considered as an important protective mechanism against oxidative stress . GSTP1 enzyme cooperates with multidrug resistance-associated protein (MRP) to protect the melanoma tumors from the anti-melanoma agents. Synergetic effects of both GSTP1 and MRP lead to melanoma drug resistance [9,45]. The important element in the failure of chemotherapy of melanoma is drug resistance . Melanoma is one of the most aggressive form of skin cancers and resistant to all current modalities of cancer therapy . Among all resistance mechanisms involved in the melanoma therapy, the over expression of GST and MRP may play critical roles [2,13,47].
A study by Laio reported that CAPE has anti-metastatic activity in mouse colon carcinoma CT26 cells . Recently, Chung group showed that CAPE also had inhibitory effects on human hepatocellular carcinoma HepG2 cell line . Grunderger group showed that CAPE exhibited a significantly greater sensitivity to human tumor cell lines than normal cell lines . Our previously reported work also showed that CAPE was an anti-melanoma agent [1,14]. However, none of these studies demonstrated that CAPE selectively inhibits GST in melanoma cells. Moreover, these previous studies did not investigate the role of tyrosinase in the inhibition of GST by CAPE.
The current study is mainly focused on investigating CAPE as a selective GST inhibitor in the presence of tyrosinase. Such a selective inhibitor will be useful in the treatment of melanoma. Our findings indicate that CAPE-SG conjugate (≥ 10 µM) and CAPE-quinone (≥ 10 µM) significantly inhibit GST whereas CAPE (<50 µM) did not inhibit GST. We have found that CAPE and caffeic acid were weak and reversible inhibitors of GST, whereas CAPE-SG conjugate, CA-SG conjugate , CAPE-quinone, and CA-quinone all were potent inhibitors of GST. It was also found that CAPE-SG conjugate was a reversible competitive and mixed GST inhibitor with respect to CDNB and GSH, respectively (Table 1). On the other hand, CAPE-quinone was an irreversible mixed and competitive inhibitor of GST with respect to CDNB and GSH, respectively. For the first time, we also identified a mono glutathione conjugate of CAPE when CAPE was incubated with tyrosinase and glutathione. Previously, we also reported a mono glutathione conjugate of 4-HA  and a mono glutathione conjugate of caffeic acid . Tyrosine was also investigated as an endogenous substrate of tyrosinase . Our results indicate that all four compounds CAPE, caffeic acid (a hydrolyzed product of CAPE), 4-HA, and tyrosine were substrates for tyrosinase but not for GST. CAPE-SG conjugate and CAPE-quinone inhibited GST significantly, clearly indicating an additional role for tyrosinase in selective inhibition of GST in melanoma cells by intracellularly bioactivating CAPE (Scheme 1). However, among all tested compounds only CAPE demonstrated significant cytotoxic effects towards SK-MEL-28 cells at low concentrations, which could be due to its high lipid solubility (LogP 3.38) and cell membrane permeability.
Literature search on previous works has also shown that glutathione conjugates are involved in inhibition of GST [53,54]. One recent study revealed that GSH conjugate of doxorubicin inhibited GST activity . Doxorubicin showed an increased cytotoxicity not only towards MDR cells but also doxorubicin sensitive cells suggesting that glutathione conjugates of drugs such as doxorubicin may play a vital role in the GST expression and hence drug induced cytotoxicity. Lyttle group also reported that a number of compounds which are coupled to the thiol group of glutathione showed significant inhibition of GST activity . Tyrosinase catalyzes the metabolism of quercetin to quinone and glutathione adducts . Thilakavathy group reported that the quinone metabolite is more potent in inhibiting GST than quercetin itself [55–57]. Previously, Ploemen et al.  also reported that caffeic acid glutathione conjugate as a potent inhibitor of GST. Similarly, our data suggest that the bioactivation of CAPE by tyrosinase leads to the formation of CAPE-quinone, which is a more potent inhibitor of GST than CAPE alone.
In addition, % glutathione consumption mediated by human placenta GST in the absence of tyrosinase was used to investigate if CAPE, CA, 4-HA, and tyrosine were substrates for GST. At 60 min incubation, none of the above compounds found to be a substrate for GST.
Ethacrynic acid (EA) is known as a GST inhibitor  because of its α,β-unsaturated carbonyl functional group [30,31]. EA can inhibit cell proliferation at higher concentrations, and is also able to enhance the cytotoxicity of many anticancer agents . Because our data suggested that CAPE acted as a selective GST inhibitor in melanoma cells due to its bioactivation with tyrosinase, the co-incubation of EA with CAPE did not enhance CAPE induced toxicity in SK-MEL-28 melanoma cells, which express tyrosinase (Table 2) .
Computational docking of EA into the dimer interface of GST was previously reported and tested experimentally . Our molecular docking studies also showed that CAPE and EA bind to the active site of GST (Fig. 8), suggesting that EA and CAPE may act as GST inhibitors. The binding modes of both CAPE-quinone and CAPE-SG conjugates may differ from that of the un-conjugated CAPE molecule. These molecular modeling suggest that CAPE and CAPE-SG conjugate may share the same binding sites with EA and EA-SG conjugate on GST.
The GST inhibition results suggested that the nature of GST inhibition by CAPE and CAPE-quinone are mixed whereas the nature of GST inhibition by CAPE-SG and EA are competitive with respect to CDNB (0.2–1 mM) (Fig. 4A – D and Table 1). With respect to GSH, the nature of GST inhibition by CAPE and CAPE-SG is different from ethacrynic acid (EA) and CAPE-quinone. GST inhibition studies at 0.2–1 mM concentration of GSH (Fig. 4E – H) suggested that CAPE and CAPE-SG are mixed inhibitors of GST with respect to GSH whereas CAPE-quinone and EA are competitive and non-competitive inhibitors of GST with respect to GSH, respectively. Awasthi et al. has suggested that EA is a non-competitive inhibitor of GST with Ki of 11.5 µM whereas EA–SG conjugate is a competitive inhibitor of GST inhibitor of GST with Ki of 1.5 µM with respect to both CDNB and GSH .
The over-expression of GST is not always conferring a significant protection from the anticancer agents. GST must be co-expressed with MRPs to protect cells from anticancer agents [8,9,60]. It was previously reported that the detoxification of anti-cancer agents is a combined effect of both GSTs and MRPs [8,12]. This study also investigated the role of MRP in CAPE induced toxicity in human SK-MEL-28 melanoma cells. Our results showed that CAPE induced toxicity in SK-MEL-28 cells are enhanced by BSO, a glutathione biosynthesis inhibitor  (data not shown), and MK-571, a selective MRP inhibitor  and probenecid, a non-selective MRP inhibitor . An increase in toxicity of CAPE in the presence of BSO in SK-MEL-28 sounds reasonable because glutathione depletion plays a vital role in CAPE toxicity towards melanoma cells as reported previously [1,14]. Most anticancer agents become resistant due to high levels of GST and high levels of glutathione in tumor cells, which makes anti-neoplastic agents inactive. Melanoma is one of the most chemo-resistant cancers, which expresses high levels of GST [2,13] and high levels of MRP . MK-571, a selective MRP inhibitor , and probenecid, a non-selective MRP inhibitor , increased CAPE induced cell toxicity, apoptosis, and significantly decreased mitochondrial membrane potential in SK-MEL-28 melanoma cells when co-incubated with CAPE. However, the effect of probenecid, a non-selective MRP inhibitor, was more significant than MK-571, a selective MRP inhibitor, suggesting other transporters may also play a role in CAPE induced toxicity. Previous literature also suggested that GST and MRP can act in synergy to protect cancer cells from toxicity of anticancer agents .
In summary, for the first time, we showed that CAPE acts as a selective inhibitor of GST in the presence of tyrosinase. Our investigation describes tyrosinase-catalyzed activation of CAPE to a quinone intermediate that reacts spontaneously with GSH to form a CAPE-SG conjugate that competitively and reversibly inhibits GST with respect to CDNB, while CAPE alone is not inhibitory. The CAPE quinone also irreversibly inhibits GST in the absence of GSH, likely by inactivation via alkylation of a key cysteine residue on GST. The irreversible inhibition by CAPE of GST was shown in cell lysates, but not in intact cells, which contain GSH in mM levels . Similar results were found with caffeic acid, albeit with less potent dose–response, and in agreement with results with caffeic acid reported by Ploemen et al. . GST and MRP play vital roles in the protection of cancer cells against cancer therapy. The toxicity of CAPE was enhanced by MK-571, a selective MRP inhibitor, and probenecid, a non-specific MRP inhibitor, suggesting MRP as a CAPE resistance factor, presumably via efflux of the CAPE-SG conjugate.
Because tyrosinase is expressed in melanoma, this may allow selective inhibition of GST as a secondary target in melanoma cells compared to non-melanoma cells that do not express tyrosinase, in addition to the direct toxicity of CAPE due to the selective activation and cytotoxicity of the CAPE quinone. Therefore, bioactivated CAPE and its GSH conjugates selectively inhibit GST in the presence of tyrosinase. It is expected that these metabolites selectively inhibit GST in melanoma cells as they contain tyrosinase. Here, the selective inhibition refers to selective inhibition of GST in melanoma cells versus non-melanoma tissue or cells. In addition to selective GST inhibition in melanoma cells, CAPE and its derivatives may also mediate a variety of biologically and toxicologically relevant processes and reactions, which were not investigated in this study.
Our study suggests that tyrosinase plays a major role in the bioactivation of CAPE, which leads to significant and selective inhibition of GST in human SK-MEL-28 melanoma cells. It also suggests a role for MRP in the biochemical mechanism of CAPE induced toxicity in melanoma cell line.
The work was supported by NCI/NIH, 1R15CA122044-01A1 (to M.M.). The work was also partially supported by National Institutes of Health Grants R21HL087895 and R01GM095538 (to L.G.) and Texas Norman Hackerman Advanced Research Program 010674-0034-2009 (to L.G.). L.G. and M.S.Y. performed the computational docking and prepared the related figure and text. H.T. helped with LC-MS/MS experiments.
Conflict of interest statement
The authors state no conflict of interest.