Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

Ifosfamide is an oxazophosphorine used in the treatment of cancer in children and adults. Encephalopathy is a serious, sometimes fatal, and limiting side effect of ifosfamide therapy. This adverse reaction is particularly associated with the oral administration of ifosfamide and in the largest series of chemotherapy cycles performed to date (390 cycles in 65 patients), the incidence of encephalopathy was 30% (Cerny et al., 1989). Studies in smaller patient series have reported even higher incidences of encephalopathy up to 100% (Aeschlimann et al., 1998) Yet, ifosfamide remains a useful chemotherapeutic tool despite this poorly understood neurotoxicity.

Because of the dose-related nature and route of administration selectivity of the encephalopathy, it has been natural for investigators to pursue various metabolites of ifosfamide as the causative agents. Chloroacetaldehyde has hitherto attracted the most attention, and this small molecule has been implicated per se in the toxicity of ifosfamide (Goren et al., 1986). It has been proposed (Visarius et al., 1999) that the chloroacetaldehyde generated from ifosfamide may cause encephalopathy by its inhibition of long-chain fatty acid metabolism and/or depletion of hepatic glutathione, but this mechanism is still unverified. In recent years, research in this area has focused on the mitochondrion after the systematic investigation of a single case by Küpfer et al. (1994) demonstrated that a glutaric aciduria type II-like mitochondrial biochemical defect was present and that ifosfamide encephalopathy responded successfully to administration of methylene blue, both therapeutically and prophylactically. These same authors (Küpfer et al., 1996) have also proposed a metabolic basis of the encephalopathy that involves flavoprotein inhibition, an intracellular redox shift, and the formation of a complex cyclic amino acid metabolite, thialysine ketamine, which they hypothesized as the ultimate neurotoxin. However, this putative metabolite has never been determined in patient samples.

It is known, however, that chloroacetaldehyde may lead to the formation of chloroacetic acid and then to S-carboxymethylcysteine (SCMC), after conjugation with the amino acid cysteine, and that SCMC can account for about 80% of the administered dose of ifosfamide (Küpfer et al., 1996), which is then further degraded metabolically to thiodiglycolic acid (TDGA) (Hofmann et al., 1991). Examination of the SCMC chemical structure (Fig. 1) indicates that it shares a close structural similarity with the excitatory neurotransmitter glutamic acid and might therefore have glutamatergic effects on neurons. In this study, we proposed to examine some pharmacological effects of two ifosfamide metabolites, namely SCMC and TDGA, on single mouse cortical neurons in primary culture.

View larger version (26K): [in this window] [in a new window]   Fig. 1.   Structures of glutamic acid, lactic acid, pyruvic acid, SCMC, SCEC, TDGA, and ifosfamide.

We show that among the ifosfamide metabolites tested, SCMC selectively activates AMPA/kainate receptors in neurons. In addition, SCMC and the other compounds tested induce substantial cellular acidification that may involve neuronal transport through monocarboxylate transporters. The effect of SCMC on AMPA/kainate receptors and the observed cellular acidification will undoubtedly have important consequences on the central nervous system.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Cell Culture

Mouse neurons in primary cultures were obtained from brain cortices from 17-day mouse embryos mechanically dissociated in phosphate-buffered saline by successive aspiration through sterile fire-polished Pasteur pipettes. The dissociated cells were centrifuged at 600 rpm for 5 min and then resuspended at a density of 60 to 65,000 cells per cm² in Neurobasal (Invitrogen, Basel, Switzerland) culture medium complemented with 2% B27 solution (Invitrogen), 500 µM glutamine, and 30 µM glutamate according to Brewer et al. (1993). Cells were then plated on glass coverslips coated with poly-L-ornithine (Sigma, Buchs, Switzerland). Cells were used after 12 to 20 days of culture.

Microscopy

Experiments were carried out on the stage of an inverted epifluorescence microscope (Carl Zeiss GmbH, Jena, Germany) and observed through a 40× 1.3 numerical aperture oil-immersion Neofluar objective lens (Zeiss). Fluorescence excitation wavelengths were selected using a holographic monochromator (Polychrome II; Till Photonics, Planegg, Germany), and fluorescence was detected using a 12-bit cooled CCD camera (Micromax; Princeton Instruments, Trenton, NJ). Acquisition and digitization of images as well as time series was computer-controlled using the software Metafluor (Universal Imaging, West Chester, PA) running on a Pentium computer (Intel, Santa Clara, CA). The acquisition rate of ratio images was varied between 0.5 and 0.1 Hz. Once loaded with dye, cells were placed in a perfusion chamber designed for rapid exchange of perfusion solutions (Chatton et al., 2000). Up to ~10 individual neurons were simultaneously analyzed in the selected field of view.

pH_i Measurements. pH_i was measured in single cells on glass coverslips after loading the cells with the pH sensitive fluorescent dye 2',7'-bis(carboxyethyl)-5,6-carboxyfluorescein (BCECF-AM; Teflabs, Austin, TX) as described previously (Chatton et al., 1997). Cell loading was performed at room temperature for 10 min using 1 µM BCECF-AM in a HEPES-buffered balanced solution (see composition below). Fluorescence was sequentially excited at 440 and 480 nm and detected through a 535 nm (35 nm bandwidth) interference filter. Fluorescence excitation ratios (F_{480 nm}/F_{440 nm}) were computed for each image pixel and produced ratio images of cells that were proportional with pH_i. In situ calibration was performed after each experiment using a nigericin technique (Thomas, 1984); cells were sequentially perfused with high K⁺ calibration solutions (see composition below) at pH 6.6, 7.1, 7.4, and 7.9 supplemented with 1 to 2 µM nigericin while BCECF ratios were measured. A calibration curve was then generated to convert ratio values into pH_i values for each individual cell under study.

Ca²⁺_i Measurements. Ca²⁺_i was measured using Fura-2 (Molecular Probes, Eugene, OR) loaded into cells by incubation with 5 µM Fura-2/AM for 30 min at 37°C. Experiments were performed in CO₂/bicarbonate-buffered solutions (see composition below). Calibration of cytosolic signal was accomplished in situ at the end of some experiments as previously described (Kao, 1994). Fluorescence was sequentially excited at 340 and 380 nm and detected at >515 nm.

Solutions

CO₂/bicarbonate-buffered experimental solutions contained 135 mM NaCl, 5.4 mM KCl, 1.3 mM CaCl₂, 0.8 mM MgSO₄, 0.78 mM NaH₂PO₄, 25 mM NaHCO₃, and 5 mM glucose bubbled with 5% CO₂/95% air. HEPES-buffered experimental solutions contained 160 mM NaCl, 5.4 mM KCl, 1.3 mM CaCl₂, 0.8 mM MgSO₄, 0.78 mM NaH₂PO₄, 20 mM HEPES, and 5 mM glucose bubbled with air. This saline was supplemented with 1% bovine serum albumin for dye loading. pH calibration solutions contained 20 mM NaCl, 120 mM KCl, 10 mM HEPES, 1.3 mM CaCl₂, 0.8 mM MgSO₄, and 0.78 mM NaH₂PO₄ and were adjusted to their respective pH by addition of NaOH.

Materials

Nigericin, TDGA, and sodium 2-mercaptoethanesulfonate were purchased from Fluka (Buchs, Switzerland). S-Carboxyethylcysteine (SCEC) was purchased from Aldrich (Buchs, Switzerland). CNQX, MK-801, and cyclothiazide were obtained from Tocris/ANAWA Trading (Wangen, Switzerland). SCMC and all other substances were obtained from Sigma.

Expression of Data and Statistics

Data are presented as means ± S.E. Student's t test was performed to assess the statistical significance of results with a p < 0.05 considered as significant. For estimation of EC₅₀ values, nonlinear curve fitting was performed using the Levenberg-Marquardt algorithm implemented in the Kaleidagraph software package (Synergy Software, Reading, PA). The dose-response analysis experiments were fitted using the following equation:
in which R_obs is the observed response; R_max and R_min are maximum and minimum parameters of the response, respectively. [S] is the concentration of transported compound, and K is the agonist concentration that yields half-maximum responses (i.e., EC₅₀), and n is the Hill coefficient.

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

Ifosfamide Metabolites and Glutamatergic Activity. Because the chemical structure of SCMC closely resembles that of the excitatory amino acid glutamate (Fig. 1), we tested the possibility that it could also interact with neuronal ionotropic glutamate receptors. Neurons were loaded with the Ca²⁺-sensitive probe Fura-2, and the changes in intracellular Ca²⁺ were monitored after application of the different compounds.

View larger version (22K): [in this window] [in a new window]   Fig. 2.   Single neuron Ca2+i response to SCMC, SCEC, TDGA, and MESNA. The Ca2+i changes induced by the individual compounds are shown parallel to the Ca2+i response to 100 µM glutamate. Representative experimental trace of 12 from two experiments.

View larger version (20K): [in this window] [in a new window]   Fig. 3.   Glutamatergic nature of the SCMC Ca2+i response. Panel A, concentration dependence of the Ca2+ response to SCMC. The amplitude (nM) of the Ca2+ response (baseline to peak) is plotted against SCMC concentration (µM). Nonlinear fit using eq. 1 and a Hill coefficient of 2.1 yielded an apparent EC50 of 1.3 ± 0.1 mM (n = 74 cells from 12 experiments). Panel B, the Ca2+i response to 1 mM SCMC was completely abolished by 25 µM CNQX, an AMPA/kainate receptor antagonist, but only weakly affected by MK-801, a NMDA receptor antagonist. At the end of the experiment, a control application of SCMC shows a response comparable with the first application. Representative experimental trace of 34 from five experiments. Panel C, the Ca2+i response to SCMC in the absence (control application) and then in the presence of 100 µM CYZ, a compound that prevents the desensitization of AMPA/kainate receptors. CYZ alone did not alter baseline Ca2+i but reversibly enhanced the response to SCMC (11.8 ± 2.5-fold, n = 16 cells from three experiments, p < 0.001). A representative experimental trace is presented. Panel D, SCMC induced a cellular acidification that was measured using the fluorescent probe BCECF, which was strongly enhanced in the presence of CYZ (3.5 ± 0.6-fold, n = 14 cells from two experiments, p < 0.001). The paradigm of compound application is indicated by the horizontal bars in the graph. A representative experimental trace is represented.

Ifosfamide Metabolites and Intracellular Acidification. Because there is evidence (Hofer, 1995; Foxall et al., 1997) that SCMC interacts with the hepatic lactate transporter MCT2 (see Discussion) and because we observed (data not shown) that the acidification induced by SCMC was reduced but not abolished by CNQX as was the Ca²⁺ response, we investigated the possibility that SCMC and TDGA were substrates of the neuronal monocarboxylate transporters (MCT2).

View larger version (19K): [in this window] [in a new window]   Fig. 4.   Lactate- and pyruvate-induced acidification in single cortical neurons. Acidification induced by successive bath applications of lactate (panel A) and pyruvate (panel B) at increasing concentrations (representative experimental traces). Panels C and D, corresponding concentration-response analysis. The amplitude of H+ concentration change is plotted against bath lactate or lactate concentration. Apparent EC50 values obtained by nonlinear curve fitting yielded 1.3 ± 1.0 mM (n = 38 cells from nine experiments) and 2.8 ± 0.6 mM (n = 29 cells from four experiments) for lactate and pyruvate, respectively. Data are means ± S.E.M.; the number of experiments is indicated in the graph.

View larger version (15K): [in this window] [in a new window]   Fig. 5.   Cellular acidification induced by SCMC. Panel A, experimental trace comparing the acidification induced by 5 mM lactate with that induced by 1 mM SCMC alone or the combination of lactate plus SCMC. Panel B, concentration dependence of the SCMC-induced acidification yielding an apparent EC50 of 0.92 ± 0.27 mM (n = 51 cells from seven experiments).

View larger version (24K): [in this window] [in a new window]   Fig. 6.   Comparison of SCMC-, SCEC- and TDGA-induced acidification with the lactate-evoked acidification. The acidification (pHi) induced by 1 mM SCMC (panel A), 1 mM SCEC (panel B), and 1 mM TDGA (panel C) alone or in combination with 5 mM lactate is compared with that induced by 5 mM lactate alone. Data are means ± S.E.M. The number of cells and experiments are indicated in the graph.

View larger version (40K): [in this window] [in a new window]   Fig. 7.   Inhibition of the acidification induced by SCMC, SCEC, and TDGA by quercetin. The figure depicts the effects of 50 µM quercetin on the acidification induced by 5 mM lactate (panel A), 1 mM SCMC (panel B), 1 mM SCEC (panel C), and TDGA (panel D). Data are means ± S.E.M., ***, p < 0.001 using the paired t test (N.S.). The number of cells and experiments are indicated in the graph.

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

In the present study, we observed that the major metabolite, SCMC, of the alkylating antitumor agent ifosfamide has agonistic effects on AMPA/kainate receptors in mouse cortical neurons and, like the terminal metabolite TDGA, induces substantial cellular acidification. These effects would undoubtedly interfere with normal central nervous system functions and may provide a molecular basis for the observed encephalopathies associated with ifosfamide chemotherapy.

The first observation of this study is that SCMC is an agonist of AMPA/kainate receptors, whereas it only weakly activates NMDA receptors. This element is undoubtedly of high relevance for the understanding of the encephalopathies associated with ifosfamide treatment; as such, SCMC could exert excitotoxic actions (Ambrosio et al., 2000). It should be pointed out that SCMC is routinely prescribed as an oral mucolytic agent, which under normal dosage has not been described to cause neuropathies. It is thus unlikely that SCMC freely crosses the blood-brain barrier. Furthermore, the EC₅₀ of ~1.3 mM Ca²⁺ response to SCMC indicates that relatively high concentrations have to reach the parenchyma to elicit significant effects. Doses of ifosfamide given to cancer patients are frequently 1 to 17.5 g per chemotherapy cycle, and therefore, plasma ifosfamide concentrations reached during chemotherapy might well result in brain interstitial SCMC concentrations in the range where it has agonistic effects on AMPA/kainate receptors. It has been observed in rats (A. Küpfer and J. R. Idle, unpublished observations) that administration of high doses of SCMC, unlike those of TDGA, do not produce observable CNS effects, confirming the suspicion that this amino acid derivative does not cross the blood-brain barrier.

Likely sources of the SCMC in the brain are reactions in situ between cysteine and/or glutathione and two-carbon donor molecules such as chloroacetaldehyde and chloroacetic acid. Recently, Saghir et al. (2001) demonstrated that, in male rats administered with [¹⁴C]chloroacetic acid, the mean residence time for radioactivity was much higher in the brain than any other tissue measured. It is plausible that chloroacetic acid metabolism in the brain to polar and resident metabolites, presumably SCMC, explain the mean residence times of around 30 h compared with that for the plasma of less than 4 h. At intravenous doses of chloroacetic acid of 60 mg/kg or above, a high proportion of rats entered coma and then rapidly died.

Data on ifosfamide and SCMC have indirectly indicated that SCMC interferes with the hepatic monocarboxylate transporter. Foxall et al. (1997) performed high-resolution proton nuclear magnetic resonance spectroscopy on the urine of 10 nonencephalopathic and 5 encephalopathic patients during their ifosfamide treatment. In the encephalopathic patients, characteristic time-related changes in the excretion profiles of some low molecular weight molecules were observed. Of particular note was a decrease in hippuric acid excretion with a concomitant increase in glycine excretion. Hippuric acid is formed from dietary benzoic acid after MCT2-mediated transport into the liver and subsequent conjugation with glycine. Unfortunately, they did not determine benzoic acid excretion. These observations, together with an enhanced urinary excretion of lactate in the encephalopathic patients, suggest that the encephalopathy is associated with the inhibition of MCTs at the level of the liver. In support of this concept, Hofer (1995) showed that the metabolism of the anticancer drug thiotepa in both adult and pediatric cancer patients proceeded to SCMC and then to TDGA, analogously to ifosfamide. Patients had elevated urinary excretion of benzoic acid, normally only found in trivial concentrations in the urine, with values up to 50 times normal.

Monocarboxylates are transported by MCTs along with protons, the stoichiometry of the cotransport being one monocarboxylate molecule for one proton (Halestrap and Price, 1999). Transport through MCTs is therefore associated with cellular acidification.

In line with the evidence from studies in liver, we found in the present study that SCMC and TDGAthe latter having no effect on Ca²⁺_iinduced significant cellular acidification that could be partly inhibited by the MCT transport inhibitor quercetin, suggesting the possible involvement of lactate transporters in the acidification. However, the additivity of the pH_i response to lactate and to the tested substances does not support transport through MCTs as the sole mechanism responsible for the observed acidifications.

Neurons are known to rely mainly on oxidative metabolism to sustain their activity (Magistretti et al., 1999), and glucose has long been thought to be the sole metabolic substrate of neurons in the brain. However, it has been shown also that neuronal activity could be completely preserved in the absence of glucose if neurons were given lactate or pyruvate as metabolic substrates (Schurr et al., 1988). Astrocytes have been shown to release substantial amounts of lactate, originating from the glycolytic processing of glucose (Pellerin and Magistretti, 1994). This set of observations has been encapsulated in an operational model of brain energy metabolism at the cellular level (Magistretti et al., 1999), which suggests the production of lactate by astrocytes in register with synaptic activity and subsequent use of lactate by neurons to fuel the tricarboxylic acid cycle and respiration. Cell-specific expression of MCTs has been found in the membrane of both astrocytes (MCT1) and neurons (MCT2) (Pierre et al., 2000).

As a first functional consequence, access to the metabolic substrate lactate by neurons could be hindered in the presence of SCMC and TDGA, and normal neuronal activity would be inhibited. Indeed, inhibition of lactate transport has been shown to decrease synaptic activity in hippocampal slices (Izumi et al., 1997). In addition, regardless of the mechanism responsible for the acidification, a decreased cellular pH will have the consequence of weakening lactate/proton cotransport. These metabolites once taken up in cells may also be cytotoxic [e.g., by interfering with mitochondrial function (Marthaler et al., 1999) or with mitochondrial pyruvate transporter]. Thus, SCMC and TDGA may interfere with lactate uptake and energy metabolism in neurons through different mechanisms.

It is possible that SCMC may exist at background levels in various body compartments due to human exposures to environmental toxicants. Analysis of normal urine from persons with no history of exposure to noxious chemicals revealed that 39 of 40 volunteers excreted 0.2 to 2.0 mg/day TDGA in their urine (Vågbø et al., 1998). Likewise, Müller et al. (1979) described excretion values of TDGA in the same range. It is not known if "endogenous" TDGA derived from SCMC. Nevertheless, a second potential source of TDGA is the two cyclic sulfur-containing dicarboxylic acids, hexahydro-1,4-thiazepine-3,5-dicarboxylic acid (cyclothionine) and thiomorpholine-3,5-dicarboxylic acid, which are also present in normal human urine (Matarese et al., 1987). Ring opening of these acids followed by decarboxylation might be expected to yield TDGA as has been demonstrated for thiomorpholine in Mycobacterium aurum cultures (Combourieu et al., 1998). By what other means might the human brain be exposed to this potential neurotoxin SCMC? Electrophilic two-carbon donors can react with cysteine to yield S-substituted cysteines, which will ultimately be oxidized to SCMC. For example, the industrial chemicals and solvents 2,2'-bis-(chloroethyl)ether (Lingg et al., 1982), vinyl chloride (Chen et al., 1983), 1,2-dichloroethane (ethylene dichloride; Cheever et al., 1990), and acrylonitrile (Fennell et al., 1991) are all metabolized to SCMC and TDGA. Interestingly, the toxicity profiles of all of these compounds comprise a significant component of neurotoxicity (e.g., vegetative dysfunction; Liubchenko et al., 1997), and this compound has been evaluated by a working group to be a possible human neurotoxin (Simonsen et al., 1994). All of these low molecular weight compounds may act by readily accessing the brain and generating in situ the resident metabolites SCMC and/or TDGA.

Finally, it should be stated that a scheme could be envisaged whereby SCMC and TDGA are able to interact with lactate and pyruvate in the brain. Figure 8 shows the metabolic biotransformations necessary to convert SCMC to TDGA. The intermediates 3-(S-carboxymethylthio)lactic acid and 3-(S-carboxymethylthio)pyruvic acid are formed from SCMC by the processes of oxidative deamination and transamination, respectively, and 3-(S-carboxymethylthio)lactic acid is a principal metabolic intermediate of SCMC (Meese and Fischer, 1990; Hofmann et al., 1991). An overload of chloroacetaldehyde and its conversion by aldehyde dehydrogenase to chloroacetic acid will generate an excess of NADH at the expense of NAD⁺, as discussed by Küpfer et al. (1996), and may promote the formation not only of lactate from pyruvate but also of 3-(S-carboxymethylthio)lactic acid from 3-(S-carboxymethylthio)pyruvic acid. The consequence of this may be to hinder further the uptake of the essential neuronal fuel, lactate, by MCTs.

View larger version (16K): [in this window] [in a new window]   Fig. 8.   Biotransformation of SCMC to TDGA via oxidation or transamination pathways. The figure shows the common role of lactate dehydrogenase in the interconversions of 3-(S-carboxymethylthio)lactic acid with 3-(S-carboxymethylthio)pyruvic acid, lactic acid with pyruvic acid, and the sensitivity of both these interconversions to the cellular NAD+/NADH ratio.

The understanding of the potential neurochemical mechanisms of ifosfamide encephalopathy, in particular the identification of discrete receptor and transporter types with which the major ifosfamide metabolites SCMC and TDGA interact, opens new horizons of investigation for the pharmacological manipulation of this adverse drug reaction, both therapeutic and prophylactic.

In conclusion, this study shows that certain ifosfamide metabolites, in particular SCMC, cause activation of AMPA/kainate receptors and induce cellular acidification, which may constitute the mechanisms responsible for the encephalopathy that is frequently associated with ifosfamide treatment. This may represent a novel pathway of neurotoxicity for a number of different compounds that are able to generate these same metabolic products in vivo.