Mass Spectrometric Measurement of Formaldehyde Generated in

MCF-7 breast cancer cells were obtained from American Type Culture Collection ... Following drug treatment, the drug/media mixture was collected and t...
1 downloads 0 Views 214KB Size
Download PDF
Chem. Res. Toxicol. 2000, 13, 509-516

509

Mass Spectrometric Measurement of Formaldehyde Generated in Breast Cancer Cells upon Treatment with Anthracycline Antitumor Drugs Shuji Kato, Patrick J. Burke, David J. Fenick,† Dylan J. Taatjes,‡ Veronica M. Bierbaum,* and Tad H. Koch* Department of Chemistry and Biochemistry, University of Colorado, Boulder, Colorado 80309, and University of Colorado Cancer Center, Denver, Colorado 80262 Received January 21, 2000

Selected ion flow tube-chemical ionization mass spectrometry was used to measure formaldehyde levels in human breast cancer cells in comparison with levels in cells treated with the antitumor drugs doxorubicin (DOX) and daunorubicin (DAU) and the daunorubicinformaldehyde conjugate Daunoform (DAUF). The measurement was performed on cell lysates and showed only background levels of formaldehyde in untreated cells and drug-treated resistant cells (MCF-7/Adr cells) but levels above background in DOX- and DAU-treated sensitive cells (MCF-7 cells). The level of formaldehyde above background was a function of drug concentration (0.5-50 µM), treatment time (3-24 h), cell density (0.3 × 106 to 7 × 106 cells/mL), and cell viability (0-100%). Higher levels of formaldehyde were observed in lysates of MCF-7 cells treated at higher drug levels, unless the treatment resulted in low cell viability. Elevated levels were directly related to cell density and were observed even with 0.5 µM drug. A lower limit for excess formaldehyde in MCF-7 cells treated with 0.5 µM DAU for 24 h is 0.3 mM. Control experiments showed that formaldehyde was not produced after cell lysis. Lysates of sensitive and resistant cells treated with 0.5 micromolar equiv of the formaldehyde conjugate (DAUF) for 3 h showed only background levels of formaldehyde. The results support a mechanism for drug cytotoxicity which involves drug induction of metabolic processes leading to formaldehyde production followed by drug utilization of formaldehyde to virtually crosslink DNA.

Introduction Recent results from several laboratories suggest that the clinically important antitumor drugs doxorubicin (DOX)1 and daunorubicin (DAU) catalyze the production of formaldehyde through induction of oxidative stress and utilize the formaldehyde for covalently linking the drugs to DNA (1-4); for recent reviews, see refs 5 and 6. The covalent linkage occurs with the drug intercalated in DNA and involves Schiff base chemistry with the formaldehyde linking the 3′-amino substituent of the drug to the 2-amino substituent of a G base. At a 5′-NGC-3′ site in DNA, the combination of intercalation, covalent bonding, and hydrogen bonding serves to virtually cross-link the DNA as shown in Scheme 1. The molecular structure of the virtual cross-link has been established by both X-ray crystallography and NMR spectroscopy (3, 7, 8). The virtual cross-link is hydrolytically unstable and appears to come apart without damage to DNA or drug (2, 9-11). At isolated G bases, intercalation and covalent bonding also occur, but the resulting DNA lesions are hydrolytically less stable, presumably because of the absence of some of the hydrogen bonding interactions * To whom correspondence should be addressed at the University of Colorado, Boulder, CO. † Current address: Peptide Technologies, Gaithersburg, MD 20877. ‡ Current address: Department of Molecular and Cell Biology, University of California, Berkeley, CA 94720. 1 Abbreviations: CI/MS, chemical ionization/mass spectrometry; DAU, daunorubicin; DAUF, Daunoform; DMSO, dimethyl sulfoxide; DOX, doxorubicin; DOXF, Doxoform; SIFT, selected ion flow tube.

(6, 10). Circumstantial evidence links drug-DNA covalent bond formation to tumor cell death, but the mechanism or mechanisms of cell death are not well-established (5, 6). In some tumor cells, drug-DNA virtual cross-linking appears to induce apoptosis (12-14). Drug catalysis of formaldehyde production has only been observed in extracellular experiments. It requires iron ion, a reducing agent, and a carbon source in the presence of dioxygen. The first observations of formaldehyde formation utilized ferric ion, dithiothreitol as the reducing agent, and Tris buffer as the carbon source (2). Subsequent experiments showed that glutathione or xanthine oxidase/NADPH preparations could serve as the reducing agent and that spermine could serve as a carbon source (9). The xanthine oxidase preparations also provided enough iron such that additional iron ion was not required. The iron chelator EDTA partially inhibited the oxidation of spermine to formaldehyde, and the stronger iron chelator desferal (desferrioxamine) completely inhibited the reaction (9). The mechanism for production of formaldehyde is thought to involve the Fenton reaction in producing reactive oxygen species culminating in the production of hydroxyl radical or its equivalent as shown in Scheme 1. The precise details of the mechanism have not been established, especially with regard to subsequent reactions of the reactive oxygen species with the carbon source. The Fenton oxidation to produce formaldehyde does not actually require the presence of a drug (9). The role of drug in cells now appears to be the sequestering of iron

10.1021/tx000008m CCC: $19.00 © 2000 American Chemical Society Published on Web 05/05/2000

510

Chem. Res. Toxicol., Vol. 13, No. 6, 2000

Kato et al. Scheme 1

from iron storage proteins. Although DOX and DAU chelate iron at their hydroxyquinone functionality, this chelation is not strong enough to shift iron from iron storage proteins (15, 16). However, recent evidence suggests that the drugs upset the iron homeostasis control mechanism, leading to release of iron from iron storage proteins (17). Chelation of the released iron ions by drug favorably locates the site of the Fenton reaction near the drug, thus allowing the drug to scavenge some of the resulting formaldehyde by reaction at its 3′-amino substituent. Drug induction of oxidative stress has been linked to tumor cell toxicity as well as to the important drug side effect of chronic cardiotoxicity. Some important experimental observations are summarized as follows. (1) Higher levels of enzymes which neutralize oxidative stress appear in DOX resistant MCF-7/Adr human breast cancer cells than in DOX sensitive MCF-7 breast cancer cells and in corresponding xenografts in nude mice. Overexpression of these enzymes has been proposed as a drug resistance mechanism (18-20). (2) The cell membrane permeable iron chelator ICRF-187 (Dexrazoxane) reduced the toxicity of doxorubicin to both MCF-7 cells (21) and rat heart cells (22). (3) Clinical studies indicate that Dexrazoxane is cardioprotective but that it may also interfere with the antitumor activity of doxorubicin (23). The discovery of drug-catalyzed production of formaldehyde and drug-formaldehyde virtual cross-linking of DNA prompted the synthesis and evaluation of drugformaldehyde conjugates as superior drugs for the treatment of resistant tumors with fewer side effects. The conjugates of DAU and DOX, Daunoform (DAUF) and Doxoform (DOXF), respectively (Scheme 2), contain two molecules of drug held together with three molecules of formaldehyde (24). In aqueous medium, they serve as prodrugs to the active metabolites of the clinical drugs

and form virtual cross-links to DNA. DAUF and DOXF are equally toxic to sensitive MCF-7 cells and to resistant MCF-7/Adr cells and substantially more toxic than DAU and DOX. The conjugates are proposed to overcome two drug resistance mechanisms: overexpression of the multidrug resistance protein, P-170 glycoprotein, which functions as a drug efflux pump and overexpression of enzymes which neutralize oxidative stress (5, 6). We now report direct measurement of elevated formaldehyde levels in DAU- and DOX-treated MCF-7 cells and only background formaldehyde levels in drug-treated MCF-7/Adr cells using chemical ionization mass spectrometry (CI/MS) coupled with selected ion flow tube (SIFT) techniques (25, 26). The CI/MS method employs gas phase reactions of ions in detecting neutral species. The combination of chemical ionization with flow tube techniques (27) has recently enjoyed a resurgence of interest and applications. For example, the method has been used to analyze air, breath, and urine samples (28-30) and to detect volatile organic emissions from cut plants (31). This approach has several powerful features: many organic and inorganic compounds, including formaldehyde, can be monitored simultaneously; the analysis is direct and free of interferences; the gases are removed from the cell samples and, in contrast to GC and HPLC assays, the formaldehyde is analyzed directly without the need for isolation or derivatization; the sampling and analysis are extremely rapid and highly sensitive; and the method can be quantified by using standard solutions.

Experimental Procedures Materials. All tissue culture materials were obtained from Gibco Life Technologies (Grand Island, NY) unless otherwise stated. MCF-7 breast cancer cells were obtained from American Type Culture Collection (Rockville, MD), and MCF-7/Adr Adriamycin resistant breast cancer cells were a gift from W. W. Wells

Mass Spectrometric Measurement of Formaldehyde

Chem. Res. Toxicol., Vol. 13, No. 6, 2000 511 Scheme 2

(Michigan State University, East Lansing, MI). Daunorubicin was a gift from the Nexstar Division of Gilead Sciences (San Dimas, CA), and doxorubicin was purchased from Tande (Houston, TX). Daunoform was prepared by reaction of formaldehyde with daunorubicin as described previously (24). Daunoform concentrations are reported in micromolar equivalents per liter to correct for the fact that Daunoform bears two active drug molecules per structural unit. Complete, Mini protease inhibitor cocktail tablets were obtained from Boehringer Mannheim (Indianapolis, IN). Maintenance of Cell Lines. MCF-7 and MCF-7/Adr cell lines were maintained in vitro by serial culture in RPMI 1640 medium supplemented with 10% fetal bovine serum (Gemini Bioproducts, Calbassas, CA), L-glutamine (2 mM), HEPES buffer (10 mM), penicillin (100 units/mL), and streptomycin (100 µg/ mL), henceforth called “media”. Cells were maintained at 37 °C in a humidified atmosphere of 5% CO2 and 95% air. Cell viability was assessed by trypan blue exclusion. Assessment of Cellular Formaldehyde Levels. Cultured cells were dissociated with trypsin/EDTA, plated in T-175 flasks, and grown to 80% confluence. The cells were treated at 37 °C with various concentrations of drug, which had been delivered to the media as a 100× solution in DMSO, for specified periods of time. Control samples were treated with 1% DMSO in media. Following drug treatment, the drug/media mixture was collected and the formaldehyde content analyzed via SIFT CI/MS. Cells were washed with Hank’s balanced salt solution and trypsinized for 5 min at 37 °C. Trypsinization was quenched with fresh media, and cells were counted and pelleted by centrifugation at 300g for 8 min at 20 °C. The supernatant was removed, and cells were resuspended in 5 mL of fresh (37 °C) media and protease inhibitors added per the manufacturer’s specifications. The cells were mechanically lysed at 1200 psi with an SLM Instruments, Inc., model FA-078 French Pressure Cell Press (American Instruments, Urbana, IL). Lysates were transferred to vacuum-tight glass flasks and the flasks sealed for immediate SIFT CI/MS analysis. The average diameter of spherical MCF-7 cells after release with trypsin was determined to be 23 ( 2 µm by comparison with known dimensions under magnification with a 10× objective lens. The cell diameter together with the cell count and formaldehyde measurement was used to calculate the concentration of formaldehyde above background in MCF-7 cells at the time of cell lysis. Further controls included the addition of daunorubicin immediately following cell lysis. Cultured cells were plated and grown as described. Samples were prepared as described above, omitting the drug treatment. Immediately following cell lysis, 50 µL of a 100× drug/DMSO solution was added to the lysates. Lysates were transferred to vacuum-tight glass flasks and formaldehyde levels assessed via SIFT CI/MS.

SIFT CI/MS Measurements. A vacuum-tight glass flask (∼40 mL) containing a 5 mL liquid sample and a drop of Antifoam A (Sigma, St. Louis, MO) was attached to the inlet of the SIFT CI/MS reaction flow tube (Figure 1). Antifoam A prevented the liquid from bubbling under the reduced pressure during headspace sampling. With the inlet valve open to the low-pressure helium flow (0.5 Torr), formaldehyde was extracted from the liquid into the gas phase (headspace of the flask) and introduced into the flow tube, where it was detected by the rapid proton-transfer reaction with hydronium ion [k ) 3 × 10-9 cm3 molecule-1 s-1 (32, 33)]:

CH2O + H3O+ f CH2OH+ + H2O Water vapor in the headspace was concomitantly introduced, converting a small fraction of the H3O+ into water clusters; however, these clusters had little effect on the detection of formaldehyde. The sampling of headspace was optimized for the formaldehyde signal by adjusting the needle valve at the inlet to the flow tube. At thermal energies, CH2OH+ ion also forms a cluster with water which undergoes a subsequent ligand switching reaction to form H3O+(H2O) (34). Control experiments showed that neither the cell culture media nor antifoaming emulsion interfered with the formaldehyde detection. Operation of the SIFT CI/MS instrument (Figure 1) for these experiments was as follows (see ref 26 for a detailed description of the instrument). In the source flow tube, electron impact on helium (0.25 Torr) generates He+ and metastable He*, which then react with water vapor added downstream in a series of reactions to form primarily H3O+ ions. Electron ionization energy, water vapor pressure, and the relative positions of the ionizer and H2O inlet in the flow tube were optimized for H3O+ production. The source ions were extracted into the SIFT quadrupole mass filter to allow mass selection of H3O+, which was then injected into the second reaction flow tube containing helium carrier gas at 300 K (purified through a molecular sieve trap cooled to liquid nitrogen temperature). The SIFT injection energy, defined by the potential difference between the source flow tube and SIFT injection plate, was typically set to 50 eV to maximize the precursor ion signal while minimizing its collision-induced dissociation: He

H3O+ 98 H2O+ + H The resulting H2O+ ions (which represent 2-3% of the H3O+ signal) can react with oxygen via charge transfer:

H2O+ + O2 f O2+ + H2O This sensitive reaction was utilized to ensure that there was negligible contamination by room air that potentially contains

512

Chem. Res. Toxicol., Vol. 13, No. 6, 2000

Kato et al.

Figure 1. ambient formaldehyde. The injected ions are quickly thermalized by collisions with helium buffer, and only thermal energy ion-molecule reactions take place in the reaction flow tube region between the injected ions and the compounds introduced from the liquid sample. Reactant and product ions were extracted through the detection sampling orifice into the detection mass filters consisting of triple quadrupoles (Q1-Q3). In the study presented here, Q3 was used for mass selection while Q1 and Q2 provided efficient ion transport. Finally, massresolved ions were detected with the electron multiplier operating in an ion-counting mode. In the absence of a liquid sample, count rates for H3O+ ions were typically 100 000 counts/s with a stability of better than 5% over several hours. The ion signal at m/z 31 (CH2OH+) indicated the presence of formaldehyde; the noise signal at this mass was found to be virtually zero. The liquid samples were examined in a fixed sequence. The inlet valve was opened for a fixed period of time (at least 80 s) to evacuate the headspace air; this process was monitored by detecting the decreasing O2+ signal. Measurement of formaldehyde followed immediately, and the CH2OH+ ion counts were collected typically for 100 s. Cell lysate and drug/media samples were measured between “background” samples, i.e., 1% DMSO/ media or water (HPLC grade). This entire cycle of measurements was repeated two or three times, and the formaldehyde counts are reported in counts per 10 s. The background samples typically exhibited 12.5 ( 1.0 counts; however, the counts were primarily due to steady-state release of formaldehyde from the inlet surfaces upon contact with water vapor. Formaldehyde in drug/media samples was found to be at the background level. The detection of formaldehyde was calibrated using standard solutions of dilute formalin, and the long-term stability of the detection sensitivity is good to within 10% over a few years. This method of calibration cancels out the secondary chemistry of CH2OH+ ions with water vapor, which is commonly present in the headspace of both cell samples and formalin solutions. The only factor to be considered in using formalin as the calibrant is the secondary reaction of CH2OH+ with methanol which is present as a stabilizer in commercial formalin (CH2OH+ + CH3OH f CH2O + CH3OH2+ ). At higher formalin concentrations, the reaction with methanol (and also possibly with formaldehyde itself) becomes significant and the detection sensitivity was found to decrease. To derive the detection sensitivity in the absence of secondary reactions, CH2OH+ counts were measured for dilute formalin solutions at several concentrations spanning 2 orders of magnitude. The detection sensitivity was then extrapolated to the low-concentration

regime, where measurements of cell samples apply. The conversion factor thus derived is 1.7 µM formaldehyde for 1 count/10 s under the experimental conditions described here. Errors in the conversion factor affect all the cell measurements in the same direction. Therefore, the statistical significance of each cell measurement remains unchanged by these systematic errors. Conservative error limits of +100% to -50% are placed on the conversion factor, which primarily reflect the uncertainty in extrapolation. The instrumental stability and extremely low noise signal at CH2OH+, combined with the long signal accumulation time, rendered the detection sensitivity of formaldehyde comparable with that of the conventional Hantzsch reaction (∼5 µM) (35) and an order of magnitude more sensitive than a recent GC method which also involves a derivatization step (36). The SIFT CI/MS measurement is noninvasive, interference free, and significantly faster than these other methods.

Results Formaldehyde in Sensitive and Resistant Tumor Cells. Formaldehyde in MCF-7/Adr resistant cells was found to be at background level in all measurements except one which is described below. The formaldehyde level is reported as the difference between levels in sensitive MCF-7 cells and resistant MCF-7/Adr cells, ∆(formaldehyde counts) ) (MCF-7 counts) - (MCF-7/Adr counts), along with error bars reported as one standard deviation from the mean. This procedure effectively cancels out the background counts that are common to each sample. Weighted averages and associated error bars are reported when there were several measurements. In the computation described above, the cell density for MCF-7/Adr cells was normalized to that for MCF-7 cells; this correction has only a minor effect on the results since formaldehyde counts in MCF-7/Adr cells are nearly the same as in background samples (DMSO/ media or HPLC water). Results on the detection of formaldehyde in cell lysates without drug treatment are reported in Table 1. No excess formaldehyde was detected in MCF-7 sensitive cells under the experimental conditions described here. We note that the value of ∆(formaldehyde counts) was

Mass Spectrometric Measurement of Formaldehyde

Chem. Res. Toxicol., Vol. 13, No. 6, 2000 513

Table 1. Absence of Elevated Formaldehyde Levels above Background in Cell Lysates without Drug Treatment cell density (×106 cells/mL) MCF-7 MCF-7/Adr 1.5 1.5 0.44 1.8

∆(formaldehyde counts)a (MCF-7 - MCF-7/Adr) 2.9 ( 2.8 1.6 ( 2.1b -1.2 ( 1.6 2.3 ( 2.3

1.5 1.5 0.28 1.5

a The cell density for MCF-7/Adr was normalized to that for MCF-7; this correction has only a minor effect on the results. Absolute formaldehyde counts were at the background level for all cell samples. b No increase in the level of formaldehyde was observed after storing the sample for 48 h at ambient temperature.

statistically significant in one experiment with larger cell densities (3.5 × 106 and 1.9 × 106 cells/mL for MCF-7 and MCF-7/Adr cells, respectively); however, the formaldehyde counts for MCF-7 cells were at the background level, while those for MCF-7/Adr cells were anomalously below background. Therefore, this result is not included in our analysis or in Table 1. Formaldehyde in Drug-Treated Tumor Cells. Table 2 summarizes the detection of formaldehyde in cell lysates following treatment with drugs (DOX, DAU, or DAUF) at different concentrations and times. Cell viability was typically 70-100% for both sensitive and resistant cells after drug treatment. Two experiments with low cell viability are separately shown in entries 7 and 8. A large excess of formaldehyde was observed in MCF-7 sensitive cells following treatment with both DOX and DAU at the highest concentrations (50 µM). Elevated formaldehyde levels were also detected in sensitive cells treated with DAU at physiologically attainable concentrations (1 and 0.5 µM); in these experiments, cells were treated for a longer period of time so a high level of drug uptake could be attained and a high level of cell viability could still be maintained. With 0.5 µM DAU, excess formaldehyde in sensitive cells scaled with cell density but not linearly (entries 4 and 5). Less formaldehyde than anticipated was detected from sensitive cells following treatment for 24 h with 5 or 10 µM DAU (entries 7 and 8); this is likely due to low cell viability in these experiments. In contrast to experiments with cells treated with

DOX and DAU, no formaldehyde above background was observed in sensitive cells following treatment with the daunorubicin-formaldehyde conjugate DAUF (entry 6). The elevated mass spectral counts of formaldehyde in MCF-7 cells were converted to concentration units with a calibration factor. Cell density data together with the measured cell volume were used to calculate the concentration of the formaldehyde within the volume occupied by the cells. The results are shown in the last column of Table 2 and represent a lower limit on the excess formaldehyde produced by the drugs in MCF-7 cells as explained in the Discussion. The results of several control experiments are relevant to data interpretation. Samples of lysed cells and drugtreated lysed cells sealed at ambient temperature for 48 h exhibited no increase in formaldehyde level with the additional time. Various concentrations of daunorubicin (50 nM, 500 nM, or 10 µM) were added immediately following cell lysis (cell density ∼ 2 × 106 cells/mL). This range covers the concentrations of the drug found in lysates of drug-treated cells, determined from UV absorption by DAU at 480 nm. Both sensitive and resistant cell lysates exposed to the drug exhibited no formaldehyde beyond the background level, indicating that the elevated formaldehyde levels observed in drug-treated sensitive cells (Table 2) were produced metabolically within the viable cells.

Discussion Prior circumstantial evidence from a series of extracellular and intracellular experiments points to a mechanism for cytotoxicity of the important antitumor drugs doxorubicin and daunorubicin which incorporates druginduced formaldehyde production (5, 6). Direct SIFT CI/ MS measurement of formaldehyde levels in lysates of sensitive MCF-7 and resistant MCF-7/Adr human breast cancer cells now strongly supports this mechanism. Lysates of both the sensitive and resistant human breast cancer cells showed only background levels of formaldehyde (Table 1), whereas lysates of DOX- and DAU-treated sensitive cells, but not resistant cells, exhibited elevated levels of formaldehyde. The level of excess formaldehyde

Table 2. Elevated Formaldehyde Levels in Lysates of MCF-7 Cells Relative to MCF-7/Adr Cells following Drug Treatment and Calculation of a Lower Limit on the Concentration of Excess Formaldehyde inside MCF-7 Cells

entry

drug

1 2 3 4 5 6

DOX DAU

7 8

DAU

DAUF

cell density (×106 cells/mL) MCF-7 (MCF-7/Adr)

∆(formaldehyde counts)a in lysates (MCF-7 MCF-7/Adr)

∆[HCHO] (mM) inside cells (MCF-7 MCF-7/Adr)

[drug] (µM)

treatment time (h)

50 50 1 0.5 0.5 0.5e

3 3 12 24 24 3

Experiments with High Cell Viability 1.5 (1.5) 1.5 (1.5) 0.32 (0.40) 0.62 (0.70) 7.2 (5.7) 1.8 (1.4)

16.7 ( 4.3b 20.9 ( 2.5 4.0 ( 1.3d 3.0 ( 1.3d 8.5 ( 1.9d -1.7 ( 1.5

3.0c 3.7 3.3 1.3 0.32 -

10 5

24 24

Experiments with Low Cell Viability 0.20 (1.4)f 1.1 (0.80)f

0.0 ( 1.4g 4.1 ( 1.2g

0.99

a Absolute formaldehyde counts for MCF-7/Adr samples were at the background level. The cell density for MCF-7/Adr samples was normalized to that for MCF-7; this correction has only a minor effect on the results. b Error bars represent statistical errors due to ion counting. c The concentrations inside cells have a statistical error of (40% arising from errors in ion counting, cell density, and cell volume measurements. In addition, the concentration data have an uncertainty of +100% to -50% as a result of a systematic error in the measurement of the conversion factor from counts to concentration units (see the text). d Weighted average over two measurements. e Units of micromolar equivalents. f Floating and adherent cells were combined due to the low viability of the sensitive cells following drug treatment (viability of 0% for 10 µM and 20% for 5 µM). g No increase in the level of formaldehyde was observed after storing the sample for 48 h at ambient temperature.

514

Chem. Res. Toxicol., Vol. 13, No. 6, 2000

in the drug-treated sensitive cells was consistent with the level of drug treatment (Table 2, entries 1-5), cell density (Table 2, entries 4 and 5), and the viability of the cells (Table 2, entries 7 and 8), and an elevated formaldehyde level was observed even at drug concentrations approaching physiological levels (Table 2, entries 4 and 5). An elevated formaldehyde level was not observed in the drug/media solution in contact with drug-treated MCF-7 cells (data not shown). To observe the elevated formaldehyde level, lysing the cells was required and was achieved physically with a French press rather than with detergent to minimize contamination. Presumably, lysing resulted in release of formaldehyde chemically bonded to nucleophilic sites within cells, including the drug virtual cross-links to DNA. Such covalent bonding is reversible (2, 3, 6, 10), and the lysing process, which increased the volume to 5 mL, likely shifted the equilibrium toward measurable free formaldehyde. Hence, the elevated formaldehyde levels represent a lower limit. Control experiments established that lysing cells in the presence of protease inhibitors did not cause any additional formaldehyde synthesis but only released formaldehyde such that a portion appeared in the headspace above the lysates for sampling by SIFT CI/MS. Addition of drug to lysates of untreated cells produced no formaldehyde above the background level. Further, elevated formaldehyde levels observed in drug-treated MCF-7 cells were a function of cell viability. Hence, neither constituents of lysed cells in the presence of drug nor nonviable cells produced formaldehyde via the steps shown in Scheme 1. Entries 4 and 5 of Table 2 show that more mass spectral counts were observed with higher cell density; however, the increase was not linear in cell density. A 12-fold increase in cell density resulted in only a 3-fold increase in counts, and the calculated concentration of formaldehyde within the volume occupied by the cells differed by a factor of 4. This observation is consistent with formaldehyde being released from bound sites upon lysing of the cells. The experiment carried out at low cell density reported in entry 4 involved a 270-fold dilution of the cell volume, whereas the experiment carried out at high cell density reported in entry 5 involved only a 23-fold dilution of the cell volume. The higher dilution likely resulted in a larger percentage of the formaldehyde being released from bound sites. This might not be the only factor contributing to the lack of linearity of formaldehyde counts with cell density. It is possible that less drug per cell is taken up at high cell densities. The observation of elevated formaldehyde levels in drug-treated, sensitive cells but not resistant cells is consistent with several earlier observations. Higher formaldehyde levels parallel higher drug levels in sensitive cells measured by flow cytometry (37). Lower drug levels in resistant cells result at least in part from overexpression of an active drug efflux pump, the multidrug resistance protein P-170 glycoprotein (19, 20). The elevated formaldehyde level in sensitive cells is also consistent with the observed lower levels of enzymes which neutralize oxidative stress, in particular, glutathione peroxidase, in sensitive cells (19, 20). The combined effect of higher drug levels and lower antioxidant enzyme levels logically leads to significantly higher levels of formaldehyde.

Kato et al.

Production of formaldehyde by tumor cells also parallels drug toxicity. The concentration of DAU which inhibited half the growth of MCF-7 cells (IC50 value) with a 3 h drug treatment period is 60 nM for MCF-7 cells but 2000 nM for MCF-7/Adr cells (24). MCF-7 and MCF7/Adr cells in the absence of drug treatment exhibit equal background levels of formaldehyde as shown in Table 1. Hence, the dramatic difference in IC50 values is not explained by a possible lower level of formaldehyde being naturally present in resistant cells, at least within the sensitivity of the measurement. Such a possibility was suggested by earlier detection of excess formaldehyde in breath, tissues, and bodily fluids of cancer patients and tumor-bearing mice (30, 36, 38). In the case of the cancer patients, prior or current treatment with DOX was not addressed. Only background levels of formaldehyde were observed in both MCF-7 and MCF-7/Adr cells treated with the daunorubicin-formaldehyde conjugate Daunoform (DAUF) (Table 2, entry 6), despite DAUF carrying its own formaldehyde into the cells. In addition, previous experiments indicate that DAUF is taken up in greater amounts in both sensitive and resistant cells with respect to DAU (37). Further, DAUF is equally toxic to MCF-7 and MCF-7/Adr cells with IC50 values in the range of 9 nanomolar equiv with a 3 h drug treatment (24). The observation of only background levels of formaldehyde in lysates of DAUF-treated cells is rationalized as follows. DAUF rapidly hydrolyzes to the anthracycline active metabolite (Scheme 2) which rapidly forms virtual crosslinks in DNA as observed by flow cytometry and fluorescence microscopy (37). When covalently bound to DNA, DAU is unable to sequester iron ion and induce formaldehyde production (Scheme 1). The amount of formaldehyde released by the DAUF and the virtual cross-links upon cell lysis is not above the background level within experimental error. The observation of excess formaldehyde in DAU- and DOX-treated MCF-7 cells but not in DAUF-treated MCF-7 cells indicates that DAU and DOX induce the synthesis of excess formaldehyde, much more than necessary to form DNA virtual cross-links. This is not surprising since some of the formaldehyde likely escapes the local environment before it can be captured at the 3′-amino substituent of the drug to form the active metabolite (Scheme 1). Production of elevated formaldehyde levels is also indicated from the observation of formaldehyde among other indicators of oxidative stress in the urine of DOX-treated rats (39). The calculated concentration of excess formaldehyde within the volume of MCF-7 cells treated with 0.5 µM DAU for 24 h is in the range of 0.3-1 mM (Table 2, entries 4 and 5). Literature reports suggest that this level of formaldehyde may be toxic to some cells but not others. Treatment of Chinese hamster V79 lung cells with 0.3 mM paraformaldehyde for 24 h inhibited growth (40). Treatment of rat hepatocytes with even 7.5 mM formaldehyde did not affect cell viability, although it did lower the level of cytosolic glutathione (41). Glutathione in conjunction with formaldehyde dehydrogenase is thought to play a protective role in cells against formaldehyde (42). The level of formaldehyde in mouse tissue has been measured and found to be approximately 1 mM in liver tissue but not detectable in other tissues (36).

Mass Spectrometric Measurement of Formaldehyde

Conclusions The chemically noninvasive technique of SIFT CI/MS showed formaldehyde levels above the background level in DOX- and DAU-treated, sensitive, human breast cancer cell lysates but not in the corresponding resistant cell lysates. Lysates of untreated sensitive and resistant cells exhibited only background levels of formaldehyde. These observations, in the context of earlier studies, support a mechanism for drug toxicity which involves drug induction of metabolic processes leading to formaldehyde synthesis and utilization of the formaldehyde for drug virtual cross-linking of DNA.

Acknowledgment. This work was supported by Grants CA24665 and CA78756 from the National Cancer Institute of the NIH. We thank the Nexstar Division of Gilead Sciences for a sample of daunorubicin and William Wells for MCF-7/Adr cells. We thank David Smith and Werner Lindinger for valuable discussions concerning CI/ MS methodology and Shelley Copley for the use of her French press.

References (1) Cullinane, C., van Rosmalen, A., and Phillips, D. R. (1994) Does adriamycin induce interstrand cross-links in DNA? Biochemistry 33, 4632-4638. (2) Taatjes, D. J., Gaudiano, G., Resing, K., and Koch, T. (1997) A redox pathway leading to the alkylation of DNA by the anthracycline, anti-tumor drugs, adriamycin and daunomycin. J. Med. Chem. 40, 1276-1286. (3) Zeman, S. M., Phillips, D. R., and Crothers, D. M. (1998) Characterization of covalent adriamycin-DNA adducts. Proc. Natl. Acad. Sci. U.S.A. 95, 11561-11565. (4) Luce, R. A., Sigurdsson, S. T., and Hopkins, P. B. (1999) Quantification of formaldehyde-mediated covalent adducts of adriamycin with DNA. Biochemistry 38, 8682-8690. (5) Taatjes, D. J., Fenick, D. J., Gaudiano, G., and Koch, T. H. (1998) A redox pathway leading to the alkylation of nucleic acids by doxorubicin and related anthracyclines: application to the design of antitumor drugs for resistant cancer. Curr. Pharm. Des. 4, 203218. (6) Taatjes, D. J., and Koch, T. H. (2000) Nuclear targeting and retention of anthracycline antitumor drugs in sensitive and resistant tumor cells. Curr. Med. Chem. 7 (in press). (7) Wang, A. H. J., Gao, Y. G., Liaw, Y. C., and Li, Y. K. (1991) Formaldehyde cross-links daunorubicin and DNA efficiently: HPLC and X-ray diffraction studies. Biochemistry 30, 3812-3815. (8) Podell, E. R., Harrington, D. J., Taatjes, D. J., and Koch, T. H. (1999) Crystal structure of epidoxorubicin-formaldehyde virtual crosslink of DNA and evidence for its formation in human breastcancer cells. Acta Crystallogr. D55, 1516-1523. (9) Taatjes, D. J., Gaudiano, G., and Koch, T. H. (1997) Production of formaldehyde and DNA-adriamycin or -daunomycin adducts, initiated through redox chemistry of DTT/iron, xanthine oxidase/ NADH/iron, or glutathione/iron. Chem. Res. Toxicol. 10, 953-961. (10) van Rosmalen, A., Cullinane, C., Cutts, S. M., and Phillips, D. R. (1995) Stability of adriamycin-induced DNA adducts and interstrand crosslinks. Nucleic Acids Res. 23, 42-50. (11) Leng, F., Savkur, R., Fokt, I., Przewloka, T., Priebe, W., and Chaires, J. B. (1996) Base specific and regiospecific chemical crosslinking of daunorubicin to DNA. J. Am. Chem. Soc. 118, 47314738. (12) Skladanowski, A., and Konopa, J. (1993) Adriamycin and daunomycin induce programmed cell death (apoptosis) in tumour cells. Biochem. Pharmacol. 46, 375-382. (13) Skladanowski, A., and Konopa, J. (1994) Relevance of interstrand DNA crosslinking induced by anthracyclines for their biological activity. Biochem. Pharmacol. 47, 2279-2287. (14) Skladanowski, A., and Konopa, J. (1994) Interstrand DNA crosslinking induced by anthracyclines in tumour cells. Biochem. Pharmacol. 47, 2269-2278. (15) Fiallo, M. M. L., Drechsel, H., Garnier-Suillerot, A., Matzanke, B. F., and Kozlowski, H. (1999) Solution structure of iron(III)anthracycline complexes. J. Med. Chem. 42, 2844-2851.

Chem. Res. Toxicol., Vol. 13, No. 6, 2000 515 (16) Gelvan, D., and Samuni, A. (1988) Reappraisal of the association between adriamycin and iron. Cancer Res. 48, 5645-5649. (17) Minotti, G., Recalcati, S., Mordente, A., Liberi, G., Calafiore, A. M., Mancuso, C., Preziosi, P., and Cairo, G. (1998) The secondary alcohol metabolite of doxorubicin irreversibly inactivates aconitase/iron regulatory protein-1 in cytosolic fractions from human myocardium. FASEB J. 12, 541-552. (18) Sinha, B. K. (1989) Free radicals in anticancer drug pharmacology. Chem.-Biol. Interact. 69, 293-317. (19) Sinha, B. K., and Mimnaugh, E. G. (1990) Free radicals and anticancer drug resistance: oxygen free radicals in the mechanisms of drug cytotoxicity and resistance by certain tumors. Free Radical Biol. Med. 8, 567-581. (20) Mimnaugh, E. G., Fairchild, C. R., Fruehauf, J. P., and Sinha, B. K. (1991) Biochemical and pharmacological characterization of MCF-7 drug-sensitive and Adr multidrug-resistant human breast tumor xenografts in athymic nude mice. Biochem. Pharmacol. 42, 391-402. (21) Doroshow, J. H. (1986) Prevention of doxorubicin-induced killing of MCF-7 human breast cancer cells by oxygen radical scavengers and iron chelating agents. Biochem. Biophys. Res. Commun. 135, 330-335. (22) Voest, E. E., van Acker, S. A., van der Vijgh, W. J., van Asbeck, B. S., and Bast, A. (1994) Comparison of different iron chelators as protective agents against acute doxorubicin-induced cardiotoxicity. J. Mol. Cell. Cardiol. 26, 1179-1185. (23) Abbott, B. D., and Ippolitti, C. M. (1998) Dexrazoxane: a new cardioprotective agent. J. Pharm. Technol. 14, 182-190. (24) Fenick, D. J., Taatjes, D. J., and Koch, T. H. (1997) Doxoform and Daunoform: anthracycline-formaldehyde conjugates toxic to resistant tumor cells. J. Med. Chem. 40, 2452-2461. (25) Adams, N. G., and Smith, D. (1976) The selected ion flow tube; a technique for studying ion-neutral reactions. Int. J. Mass Sepctrom. Ion Phys. 21, 349-359. (26) Van Doren, J. M., Barlow, S. E., DePuy, C. H., and Bierbaum, V. M. (1987) The tandem flowing afterglow-SIFT-drift. Int. J. Mass Spectrom. Ion Processes 81, 85-100. (27) Graul, S. T., and Squires, R. R. (1988) Advances in flow reactor techniques for the study of gas-phase ion chemistry. Mass Spectrom. Rev. 7, 263-358. (28) Lindinger, W., Hansel, A., and Jordan, A. (1998) On-line monitoring of volatile organic compounds at pptv levels by means of proton-transfer-reaction mass spectrometry (PTR-MS): medical applications, food control and environmental research. Int. J. Mass Spectrom. Ion Processes 173, 191-241. (29) Smith, D., and Spanel, P. (1996) Application of ion chemistry and the SIFT technique to the quantitative analysis of trace gases in air and on breath. Int. Rev. Phys. Chem. 15, 231-271. (30) Spanel, P., Smith, D., Holland, T. A., Singary, W. A., and Elder, J. B. (1999) Analysis of formaldehyde in the headspace of urine from bladder and prostate cancer patients using selected ion flow tube mass spectrometry. Rapid Commun. Mass Spectrom. 13, 1354-1359. (31) de Gouw, J. A., Howard, C. J., Custer, T. G., and Fall, R. (1999) Emissions of volatile organic compounds from cut grass and clover are enhanced during the drying process. Geophys. Res. Lett. 26, 811-814. (32) Adams, N. G., Smith, D., and Grief, D. (1978) Reactions of HnCO+ ions with molecules at 300 K. Int. J. Mass Spectrom. Ion Phys. 26, 405-415. (33) Mackay, G. I., Tanner, S. D., Hopkinson, A. C., and Bohme, D. K. (1979) Gas-phase proton-transfer reactions of the hydronium ion at 298 K. Can. J. Chem. 57, 1518-1523. (34) Hansel, A., Singer, W., Wisthaler, A., Schwarzmann, M., and Lindinger, W. (1997) Energy dependencies of the proton-transfer reactions H3O+ + CH2O T CH2OH+ + H2O. Int. J. Mass Spectrom. Ion Processes 167/168, 697-703. (35) Nash, T. (1953) The colorimetric estimation of formaldehyde by means of the Hantzsch reaction. Biochem. J. 55, 416-421. (36) Ebeler, S. E., Hinrichs, S. H., Clifford, A. J., and Shibamoto, T. (1994) Volatile carbonyl levels in tissues of transgenic mice with nerve sheath tumors. J. Chromatogr., Biomed. Appl. 654, 9-18. (37) Taatjes, D. J., Fenick, D. J., and Koch, T. H. (1999) Nuclear targeting and nuclear retention of anthracycline-formaldehyde conjugates implicates DNA covalent bonding in the cytotoxic mechanism of anthracyclines. Chem. Res. Toxicol. 12, 588-596. (38) Ebeler, S. E., Clifford, A. J., and Shibamoto, T. (1997) Quantitative analysis by gas chromatography of volatile carbonyl compounds in expired air from mice and human. J. Chromatogr. B702, 211215.

516

Chem. Res. Toxicol., Vol. 13, No. 6, 2000

(39) Bagchi, D., Bagchi, M., Hassoun, E. A., Kelly, J., and Stohs, S. J. (1995) Adriamycin-induced hepatic and myocardial lipid peroxidation and DNA damage, and enhanced excretion of urinary lipid metabolites in rats. Toxicology 95, 1-9. (40) Saito, S. (1987) Effect of paraformaldehyde on cultured V79 cells. Shigaku 75, 75-87. (41) Ku, R. H., and Billings, R. E. (1986) The role of mitochondrial glutathione and cellular protein sulfhydryls in formaldehyde

Kato et al. toxicity in glutathione-depleted rat hepatocytes. Arch. Biochem. Biophys. 247, 183-189. (42) Naylor, S., Mason, R. P., Sanders, J. K. M., Williams, D. H., and Moneti, G. (1988) Formaldehyde adducts of glutathione. Biochem. J. 249, 573-579.

TX000008M