Decreased pyruvate kinase M2 activity linked to cisplatin resistance in human gastric carcinoma cell lines



Resistance to anticancer drugs is a major obstacle preventing effective treatment of disseminated cancers. Understanding the molecular basis to chemoresistance is likely to provide better treatment. Cell lines resistant to cisplatin or 5-fluorouracil (5-FU) were established from human gastric carcinoma cell lines SNU-638 and SNU-620. Comparative proteomics involving 2-dimensional gel electrophoresis (2-DE) and matrix-associated laser desorption ionization-mass spectroscopy (MALDI-MS) was performed on protein extracts from these parental and drug-resistant derivative lines to screen drug resistance-related proteins. Pyruvate kinase M2 (PK-M2) was identified as a protein showing lower expression in cisplatin-resistant cells compared to parental cells. Consistent with this finding, PK-M2 activity was also lower in cisplatin-resistant cells. Suppression of PK-M2 expression by antisense oligonucleotide resulted in acquired cisplatin resistance in SNU-638 cells. Furthermore, PK-M2 activity in 11 individual human gastric carcinoma cell lines positively correlated with cisplatin sensitivity. Taken together, PK-M2 protein and activity levels were lower in cisplatin-resistant human gastric carcinoma cell lines compared to their parental cell lines. Furthermore, suppression of PK-M2 expression using antisense oligonucleotides increased cisplatin resistance. These data clearly link PK-M2 and cisplatin resistance mechanisms. © 2003 Wiley-Liss, Inc.

Gastric cancer is one of the most common causes of cancer-related mortality worldwide.1 There are marked geographic variations in gastric cancer incidence, with the highest rates in Korea, Japan, China and South America, and much lower rates in Western countries.2 Contemporary chemotherapies for advanced gastric cancer, usually containing 5-fluorouracil (5-FU) and/or cisplatin, demonstrate response rates in the 20–40% range, with median survival between 6 and 12 months.2 However, cancer cells can be insensitive to drug treatment at the outset of therapy (intrinsic resistance) or they may become insensitive after exposure to the antitumor agent (acquired resistance).3 Drug resistance, especially acquired resistance, is a major obstacle preventing effective treatment of disseminated cancers. Despite extensive studies, the mechanisms underlying drug resistance have yet to be fully understood.

Greater understanding of the molecular basis of gastric cancer chemoresistance is likely to provide more effective treatment for patients. In the present study, we created human gastric carcinoma cell lines that were resistant to cisplatin and 5-FU and performed comparative proteomics to identify proteins involved in drug resistance. We herein report that reduced pyruvate kinase M2 (PK-M2) activity is linked to cisplatin resistance.


Drug-resistant human gastric carcinoma cell lines

Human gastric carcinoma cell lines SNU-1, SNU-5, SNU-16, SNU-216, SNU-484, SNU-520, SNU-601, SNU-620, SNU-638, SNU-668 and SNU-7194, 5 were obtained from the Korean Cell Line Bank (Seoul, Korea). Cell lines resistant to cisplatin (Ildong Pharm. Co. Ltd., Gyeonggi, Korea) or 5-FU (Choongwae Pharma Corporation, Gyeonggi, Korea) were induced from SNU-638 and SNU-620 cells as described previously.6 Briefly, cells were cultured in RPMI-1640 (Gibco/BRL, Grand Island, NY) containing 10% fetal calf serum, 2 mg/ml sodium bicarbonate, 100 U/ml penicillin and 0.1 mg/ml streptomycin (Gibco/BRL). Drugs were added at increasing concentrations, with the initial drug concentration being 10% of IC50 (defined as the concentration producing a 50% reduction in absorbance at 540 nm compared to untreated controls in an MTT assay [see below]). Every 4 weeks the drug concentration was increased by 10%. The final resistance concentrations for each cell line are shown in Table I and represent the mean and standard deviation of the IC50.

Table I. Human Gastric Carcinoma Cell Lines and Those Drug Resistant Cell Lines
Cell linesIC50 (μg/ml)1Relative resistance
  • 1

    IC50 is defined as a concentration of drug that produced a 50% reduction of absorbance at 540 nm compared with the untreated controls in MTT assay.

SNU-6204.30 ± 0.300.03 ± 0.01 
SNU-620 cisplatin resistant25.78 ± 0.94 6.00
SNU-620 5-FU resistant 5.64 ± 2.57188.00
SNU-6380.81 ± 0.210.04 ± 0.05 
SNU-638 cisplatin resistant4.34 ± 1.47 5.36
SNU-638 5-FU resistant >500>12,500

MTT assay

A colorimetric assay using the tetrazolium salt MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide) was used to assess the cytotoxicity of anticancer agents cisplatin, 5-FU, doxorubicin (Dong-A Pharmaceutical Co. Ltd., Seoul, Korea) and mitomycin-C (MMC, Kyowa Hakko Kogyo, Japan). Single cell suspensions were prepared, the cell density measured and the MTT assay performed as previously described.7 Briefly, an equal number of cells was inoculated into each well in 0.18 ml culture medium, to which 0.02 ml of 10 times the normal concentration of drug or PBS (for untreated 100% survival control) was added. After 4 days of culture, 0.1 mg MTT was added to each well and incubated at 37°C for 4 hr. Plates were centrifuged at 450g for 5 min at room temperature and medium removed. DMSO (0.15 ml) was added to each well to solubilize the crystals, and the plates were read immediately at 540 nm on a scanning multiwell spectrometer (Bio-Tek instruments Inc., Burlington, VT). All experiments were performed 3 times and the mean and standard deviation of the IC50 (μg/ml) values were calculated.

Two-dimensional gel electrophoresis (2-DE) analysis

Human gastric carcinoma cell lines were harvested and suspended in 0.5 ml sample buffer consisting of 40 mM Tris, 5 M urea (Merck, Darmstadt, Germany), 2 M thiourea (Sigma, St. Louis, MO), 4% CHAPS (Sigma), 10 mM 1,4-dithioerythritol (DTT) (Merck), 1 mM EDTA (Merck) and a mixture of protease inhibitors (1 mM phenylmethylsulfonyl fluoride [PMSF] and 1 μg each of pepstatin A, chymostatin, leupeptin and antipain [Roche Diagnostic GmbH, Mannheim, Germany]). The suspension was sonicated for approximately 30 sec, centrifuged at 100,000g for 10 min and the supernatant collected and centrifuged at 150,000g for 45 min. The protein content in the supernatant was determined using the Coomassie blue method.8

2-DE was performed as previously reported9 with modifications. Briefly, a 0.15 mg sample was applied to 13 cm immobilized pH 3-10 nonlinear gradient strips (Pharmacia Biotechnology, Uppsala, Sweden). Proteins were focused at 8,000 V within 3 hr. The second-dimension separation was on 12% polyacrylamide gels (chemicals from Serva, Heidelberg, Germany and Bio-Rad, Hercules, CA). 2-DE gels were stained with Colloidal Coomassie Blue (Invitrogen, Carlsbad, CA) for 24 hr and destained with deionized water.

Matrix-associated laser desorption ionization-mass spectroscopy (MALDI-MS)

MALDI-MS analysis of 2-DE protein spots was performed as previously described10 with minor modifications. Briefly, spots were excised, destained with 50% acetonitrile in 0.1 M ammonium bicarbonate and dried in a speedvac evaporator. The dried gel pieces were reswollen with 3 μl 3 mM Tris-HCl, pH 8.8, containing 50 ng trypsin (Promega, Madison, WI) and after 15 min, 3 μl water was added. One μl was applied to the dried matrix spot. The matrix consisted of 15 mg nitrocellulose (Bio-Rad) and 20 mg α–cyano 4 hydroxycinnamic acid (Sigma) dissolved in 1 ml acetone:isopropanol (1:1, v/v). The matrix solution (0.5 μl) was applied to the sample, and samples analysed in a QSTAR® XL Hybrid LC/MS/MS System (Applied Biosystems, Foster City, CA) using a 20 kV accelerating voltage. Peptide masses were matched with theoretical peptide masses of all proteins from all species in the SWISS-PROT database.

Pyruvate kinase activity assay

Harvested gastric carcinoma cells were sonicated for 30 sec and centrifuged at 15,000g for 15 min. Aliquots (5 μg) of the supernatant were assayed for pyruvate kinase activity using the coupling assay11 with modifications. Reaction buffer (50 mM Tris-HCl pH 7.4, 100 mM KCl, 5 mM MgCl2, 0.6 mM ADP, 0.9 mM phospho(enol)pyruvate [PEP], 0.3 mM NADH and 2.5 IU L-lactate dehydrogenase) was incubated in a spectrophotometer for 4–5 min to achieve temperature equilibrium and determine a “blank” rate. A sample aliquot was added and the rate of decrease in A340, resulting from equimolar oxidation of NADH to NAD+ during reduction of pyruvate from PEP by pyruvate kinase, was recorded for 5 min. Pyruvate kinase activity was calculated using ΔA340/min obtained from the initial linear portion of the curve, with one unit of activity defined as that required for the oxidation of 1 μmol NADH min−1 mg−1 at 25°C and pH 7.4 in the specified conditions described above.

Western blot analysis

Human gastric carcinoma cell lines were homogenized in 0.05% Tween 20-TBS buffer containing 10 mM Tris-HCl, pH 7.5, 150 mM NaCl in a Potter-Elvehjem homogenizer at 4°C in the presence of protease inhibitors (1 mM PMSF, 10 μg/ml apoprotein, 10 μg/ml leupeptin). Protein concentrations of homogenates were estimated by the dye binding method8 using bovine serum albumin as the standard. For Western blotting, 4,000g supernatants of homogenates containing equivalent amounts of protein were subjected to SDS-PAGE using the principle of Laemmli.12 After electrophoresis, proteins were transferred to polyvinylidene fluoride (PVDF) membranes (Millipore, Billerica, MA). Membranes were blocked by incubating overnight at 4°C in 1% Tween 20-TBS buffer containing 1.5% nonfat dry milk (Bio-Rad, Richmond) and 1 mM MgCl2. Membranes were washed 2 × 5 min with TBS buffer and put into a glass vessel containing diluted monoclonal mouse anti-human pyruvate kinase-M2 (ScheBo® Bioteck AG, Giessen, Germany). After 2 hr incubation with primary antibody at room temperature, membranes were washed 3 × 15 min in blocking solution and incubated with diluted HRP-conjugated secondary antibody for 1 hr at room temperature. Membranes were washed 3 × 15 min in blocking solution and incubated with NEN chemiluminescence reagent (NEN™ Life Science Products, Inc., Boston, MA) for 1 min and exposed to film (Kodak Blue XB-1, Rochester, NY). Developed films were scanned and densities of immunoreactive bands calculated by GelComparII software (Applied Maths, Kortrijk, Belgium).

Administration of PK-M2 antisense oligonucleotide to cells

Two antisense and 2 sense phosphorothiorated oligonucleotides were synthesized: PK-M2 AS(antisense)-1 5′-AGACGAGCCACATTCATTCC-3′, PK-M2 AS-2 5′-GTCCAGCCACAGGATGTTCT-3′, PK-M2 S(sense)-1, 5′-GGAATGAATGTGGCT CGTCT-3′ and PK-M2 S-2, 5′-AGAACATCCTGTGGCTGGAC-3′ (Bioneer, Daejeon, Korea). Cells were transfected according to the instructions provided by the manufacturer of Oligofectamin™ Reagent (Invitrogen, Carlsbad, CA). Briefly, the day before transfection, SNU-638 cells were trypsinised and plated into 6-well plates in order to be 30–50% confluent after about 24 hr. An amount of 10 μl of 20 μM stock sense and antisense S-oligonucleotide and 3 μl of Oligofectamin™ Reagent were diluted into medium without serum to give a final concentration of 200 nM S-oligonucleotide. Cells were incubated for 4 hr and concentrated serum media added without removing the transfection mixture. At 24 hr, cisplatin was added to a final concentration of 2.08 μg/ml. The number of cells and the level of PK-M2 expression were determined every 24 hr.

Statistical analyses

Between-group differences were calculated using the nonparametric Mann-Whitney U-test and within-group-correlations were done using the Spearman rank-coefficient. The significance was set at the p < 0.05 level.


Total protein extracts from normal and cisplatin-resistant SNU-638 cell lines were separated by 2-DE (Fig. 1a). Image analysis after Coomassie blue staining showed that the expression level of a 68-kDa protein with pI 8.0 was lower in the cisplatin-resistant compared to its parent cell line. The protein spot was excised from the 2-DE gel, trypsin digested and subjected to MALDI-MS analysis to determine peptide masses (Fig. 1b). A SWISS-PROT database search identified the protein as pyruvate kinase M2 isozyme (PK-M2) (Fig. 1b).

Figure 1.

Identification of pyruvate kinase M2 as differently expressed in cisplatin-resistant human gastric carcinoma cell lines. (a) 2-DE gels of protein extracts from SNU-638 and its cisplatin-resistant derivative cell line. Protein (0.15 mg) was applied to an immobilized pH 3-10 nonlinear gradient strip and, after focusing, the second dimension separation was performed on 12% homogenous polyacrylamide gels. Typical electrophoretic patterns are shown, with enlarged partial 2-DE images showing the different protein pattern in cisplatin-resistant cell extracts. (b) MALDI-MS analysis of the protein spot differently expressed in SNU-638 cisplatin-resistant cells. The protein spot was excised from the gel, digested with trypsin and peptides analyzed by MALDI-MS. Figure shows the mass spectrum, and the identified protein spot was characterized using SWISS-PROT accession numbers. Theoretical and approximate observed Mr and pI values, as well as the number of matching and total peptides, are given.

PK-M2 expression was measured in cisplatin-resistant cells using Western blot analysis. When normalized to actin levels, PK-M2 protein expression was found to be significantly decreased in both SNU-638 and SNU-620 cisplatin-resistant cell lines (Fig. 2). PK-M2 enzyme activity was also found to be lower in cisplatin-resistant cells, with activity in SNU-638 and SNU-620 cisplatin-resistant cells lines being less than half of that in their parent lines (Fig. 2).

Figure 2.

Decreased levels of both PK-M2 protein and activity in human gastric carcinoma cell lines correlated with cisplatin resistance but not with 5-FU resistance. Western blot analysis and enzyme activity assay were performed in 2 different parental cell lines (SNU-638 and SNU-620) and their cisplatin- and 5-FU-resistant derivative cell lines.

SNU-638 parental cells were transfected with 4 S-oligonucleotide types, 2 sequences being “sense” PK-M2, namely S-1 and S-2, and 2 being “antisense” PK-M2, namely AS-1 and AS-2. Western blot analysis showed PK-M2 expression was markedly decreased 24 hr after AS-2 transfection (Fig. 3b). AS-1 did not affect PK-M2 expression and neither did S-1 or S-2 (Fig. 3a). When cisplatin (2.08 μg/ml) was added 24 hr after AS-2 transfection, cells exhibited increased resistance, with the relative survival (the number of cells proliferating after transfection with antisense or sense oligonucleotide per the number of cells proliferating without transfection.) increasing about 200% up to 72 hr after transfection (Fig. 3b). In contrast, cells transfected with S-1, S-2 or AS-1 exhibited about 75% relative survival at 72 hr after transfection after cisplatin addition (Fig. 3).

Figure 3.

Cisplatin resistance in SNU-638 cells after administration of antisense S-nucleotide suppressing PK-M2 expression. Two PK-M2 sense (S-1 and S-2) and antisense (AS-1 and AS-2) S-oligonucleotides were administrated to SNU-638 cells. The effects of AS-1 (a) and AS-2 (b) on cisplatin resistance and PK-M2 are shown. PK-M2 protein expression and activity was monitored by Western blot analysis and enzyme activity assay, respectively. aRelative survival = the number of cells proliferating after transfection with antisense or sense oligonucleotide per the number of cells proliferating without transfection.

We investigated whether there was a correlation in PK-M2 activity and drug sensitivity in 11 human gastric carcinoma cell lines. We found a positive correlation between PK-M2 activity and cisplatin sensitivity (p = 0.044; Fig. 4a). However, sensitivities to 2 other drugs, 5-FU and mitomycin-C (MMC), did not significantly correlate with PK-M2 activity (p = 0.203 and p = 0.248, respectively; Fig. 4c). Significant correlation of PK-M2 activity with doxorubicin sensitivity (p = 0.031) appeared only when the SNU-520 cell line was included (doxorubicin IC50 > 100 μg/ml; Table II) (Fig. 4b).

Figure 4.

Correlation of PK-M2 activity with drug sensitivity in gastric carcinoma cell lines. (a) Cisplatin, (b) doxorubicin, (c) 5-FU and MMC. PK-M2 activity and sensitivity to the drugs were measured 3 times and mean value was used for calculating correlation. Sensitivities of 11 cell lines to the 4 drugs were determined as shown in Table II.

Table II. Different Drug Sensitivities of Human Gastric Carcinoma Cell Lines
Cell linesIC50 (μg/ml)1
  • 1

    IC50 is defined as a concentration of drug that produced a 50% reduction of absorbance at 540 nm compared with the untreated controls in MTT assay.

SNU-11.23 ± 0.170.38 ± 0.090.04 ± 0.010.13 ± 0.03
SNU-50.21 ± 0.030.05 ± 0.010.08 ± 0.010.02 ± 0.00
SNU-161.59 ± 0.500.22 ± 0.020.02 ± 0.010.10 ± 0.04
SNU-2164.23 ± 0.7440.70 ± 14.600.25 ± 0.061.57 ± 0.15
SNU-4842.61 ± 2.3712.90 ± 3.220.25 ± 0.040.92 ± 0.04
SNU-5207.66 ± 1.630.65 ± 0.37>1002.49 ± 0.89
SNU-6011.52 ± 0.290.03 ± 0.000.06 ± 0.010.25 ± 0.07
SNU-6204.30 ± 0.300.03 ± 0.010.45 ± 0.190.22 ± 0.08
SNU-6380.81 ± 0.210.04 ± 0.050.14 ± 0.020.08 ± 0.01
SNU-6686.52 ± 0.842.38 ± 0.110.20 ± 0.010.72 ± 0.03
SNU-7193.35 ± 0.640.40 ± 0.060.04 ± 0.010.01 ± 0.00


Although the biochemic basis underlying cellular resistance to cisplatin has yet to be fully determined, several mechanisms that can contribute to resistance have been defined.13, 14 Recently, expression of the ERCC1 (excision repair cross complementation group 1) nucleotide excision repair gene has been widely studied in the context of cisplatin resistance in several tumor types. Poor response to cisplatin-based chemotherapy was correlated with high pre-treatment levels of ERCC1 mRNA in gastric adenocarcinoma patients,15 although high ERCC1 mRNA levels may not universally predict poor responsiveness to chemotherapy.16

In our present study, comparative proteomics using human gastric carcinoma cell lines resistant to cisplatin or 5-FU was able to identify proteins associated with drug resistance. One such protein, PK-M2, showed a decrease in both activity and expression in cisplatin-resistant cell lines (Figs. 1 and 2). Pyruvate kinase catalyzes the last step in the process of glycolysis, conversion of the high-energy intermediate PEP to pyruvate accompanied by conversion of ADP to ATP. Pyruvate kinase activity also can determine the relative amount of glucose that is channeled into synthetic processes or used for glycolytic energy production.17, 18 Pyruvate kinase occurs in 4 isozymic forms—L, R, M1 and M2—and these are encoded by 2 different genes, PKL and PKM.19 The L and R isozymes are generated from the PKL gene by means of alternative promoters, while the M1 and M2 forms are produced from the PKM gene by differential splicing.19 PK-M2 is associated with tumors20 but its expression has also been reported in normal tissues such as brain.21 PK-M2 occurs in an active tetrameric form and a less active dimeric form,18, 22, 23 and tumor cells are usually characterized by a high amount of the dimeric form.24 Our microarray results showed that of the isoforms of pyruvate kinase, only PK-M2 is expressed in the human gastric carcinoma cell lines used in our study (unpublished data). This indicates that the pyruvate kinase activity measured in the current study is most likely due to PK-M2. We found that expression of PK-M2 protein correlated with PK-M2 activity (Fig. 2), and that levels of both were significantly lower in most cisplatin-resistant human gastric carcinoma cells.

We found that when PK-M2 expression and activity were suppressed by administration of antisense oligonucleotide in cisplatin-sensitive SNU-638 parent cells, these cells displayed increased drug resistance (Fig. 3b). These results appear to confirm the clear link between decreased PK-M2 and cisplatin resistance in human gastric carcinoma cell lines.

We tested whether cisplatin has an inhibitory effect on pyruvate kinase activity. Cisplatin showed very weak inhibitory effect on PK-M2 activity in vitro, i.e., The concentration of a drug required to inhibit enzyme activity by 50% was about 50 μg/ml (∼ 167 μM) cisplatin (data not shown). This dose is much higher than the amount of cisplatin used in Figure 3 and cisplatin IC50 values of cell lines obtained from MTT assay. Thus, it demonstrates no inhibitory effect of cisplatin on the PK-M2 activity described in Figure 3.

AS-2 decreased PK-M2 expression, whereas AS-1 did not (Fig. 3). Although a bias in the nucleotide composition of antisense oligonucleotides has been suggested,25 it is not possible to predict effective targets for antisense oligonucleotide. Considering that typically several antisense oligonucleotides complementary to a given mRNA must be prepared and tested in order to find an effective one in cells, the reason why AS-2 worked only might be explained.

At this stage, the mechanism(s) of resistance to cisplatin caused by low PK-M2 activity is not clear. However, one of the possible links may exist in the increase of NADPH production within the oxidative pentose phosphate pathway as well as glutaminolysis. PK-M2 catalyzes the last step in the process of glycolysis but it is also involved in a novel metabolic strategy for energy generation during increased cell proliferation.20 The amount of dimeric PK-M2 in tumor cells closely correlates with the degree of malignancy.26 A high level of this less active form can lead to accumulation of glycolytic phosphometabolites upstream of PK-M2,24 and these metabolites can be used in synthesis of biomacromolecules for cell proliferation. Thus, PK-M2 is able to direct glucose carbons to the pentose phosphate pathway for the biosynthesis of nucleic acid sugar moieties.27 For cell proliferation, tumor cells with low PK-M2 activity obtain the energy from glutaminolysis.28 Therefore, low PK-M2 activity may increase the NADPH production within the oxidative pentose phosphate pathway as well as glutaminolysis.28 It is well accepted that cisplatin is inactivated by covalent linking to glutathione (GSH) after nucleophilic attack of the GSH thiolate anion.14 NADPH as a cofactor of GSH reductase is absolutely necessary for the reduction of oxidized GSH (GSSH) back to 2 moles of GSH.29 Furthermore, several studies have demonstrated a correlation between toxicity of cisplatin and intracellular activity of the thioredoxin (Trx) system, i.e., Trx, Trx reductase and NADPH.30, 31 Conversely, activation of the Trx system by increased NADPH is also responsible for the development of cellular resistance to cisplatin, possibly by scavenging intracellular toxic oxidants generated by cisplatin.32

Taken together, low PK-M2 activity (perhaps due to an increased proportion of dimers or decreased amount of PK-M2 protein) may cause the increase of NADPH production within the pentose phosphate pathway as well as glutaminolysis. The increased NADPH may exert effects not only on reduction of GSSH to GSH but also on activation of the Trx system to confer cisplatin resistance.

Correlation between PK-M2 activity and sensitivities to anticancer drugs (Fig. 4) was examined in 11 different individual human gastric carcinoma cell lines. PK-M2 activity showed a positive correlation with cisplatin sensitivity (p = 0.044) (Fig. 4a), supporting the link between PK-M2 and cisplatin resistance. However, the sensitivities to other drugs such as 5-FU and MMC (p = 0.203 and p = 0.208, respectively) (Fig. 4c) did not show any correlation with PK-M2 activity. Schneider and coworkers33, 34, 35 reported that immunohistochemic detection of PK-M2 in tissue sections of lung cancer specimens exhibited selective staining of tumor cells, independent of the histologic classification of the tumor. In addition, in EDTA plasma of lung cancer patients, the dimeric form of PK-M2 correlated well with tumor load during follow-up, showing significantly increasing dimeric form with progressive tumor stages and decreased dimeric form during tumor remission from cisplatin/gemcitabine combinational chemotherapy.33, 34, 35 Thus, PK-M2 may be considered a target protein in reducing cisplatin resistance. Some biochemic approaches may provide a way to modulate PK-M2 activity in order to reduce cisplatin resistance, for example, regulation of the tetramer-dimer ratio using ATP, FBP and serine, or direct interaction with oncoproteins pp60v-src36 and HPV-16 E7.37

In conclusion, we observed that cisplatin resistance correlated with decreased levels of PK-M2 protein and activity in human gastric carcinoma cell lines, and that lowering PK-M2 expression through antisense transfection increased cisplatin resistance. These data clearly link PK-M2 activity and cisplatin-resistance mechanisms.