Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The undecapeptide cyclosporine currently is the basis of most immunosuppressive protocols in transplantation medicine. Cyclosporine has a narrow therapeutic index, and neurotoxicity is one of its most frequent and most serious side effects (DiMartini et al., 1991; McManus et al., 1992). Clinical magnetic resonance imaging and computed tomography studies have shown diffuse hypodensity of white matter, increase of the water contents in the white matter, and metabolic encephalopathy in patients with cyclosporine neurotoxicity (Pace et al., 1995; Schwartz et al., 1995). In previous in vitro studies, we showed that high concentrations of cyclosporine (5 mg/liter and more) induced significant astrocyte swelling, affected osmo- and volume regulation of brain cells, and increased turnover rates of membrane metabolism (Serkova et al., 1996; Serkova et al., 1997). However, these cyclosporine concentrations were more than 10-fold higher than the therapeutic target ranges in blood of transplant patients (Oellerich et al., 1995).

The macrolide sirolimus currently is in phase III of its clinical development as a new immunosuppressive drug in combination with cyclosporine after kidney transplantation (Kahan et al., 1995; Murgia et al., 1996; Kahan, 1997). Although not yet observed clinically, in vitro sirolimus exhibited toxic effects on brain cell metabolism similar to cyclosporine (Serkova et al., 1997). Possible implications for the side effect spectrum of the cyclosporine/sirolimus combination have not yet been assessed.

Because, at present, in vitro effects of cyclosporine or sirolimus on brain cell metabolism have been studied only at unphysiologically high concentrations, it was our objective to evaluate the concentration-dependent inhibition of brain cell metabolism including the therapeutic-relevant concentration range. In addition, because of its potential clinical impact, we studied the effect of the combination of cyclosporine and sirolimus on rat brain metabolism.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Materials and Reagents. Cyclosporine (Novartis, East Hanover, NJ) and sirolimus (Sigma Chemical Co., St. Louis, MO) stock solutions (1 g/liter) were prepared in acetonitrile/water (pH 3), 50:50 or 80:20 (v/v), respectively. Stock solutions were stored at 80°C until use. Various concentrations of each drug (100-10,000 µg/liter) were added directly to culture medium of perfused brain slices or cell cultures. The final concentrations of acetonitrile in the medium did not exceed 0.5%. The maximum acetonitrile concentration (0.5%) had no significant effects on cell metabolism. All animal protocols were approved by the University of California, San Francisco, Committee on Animal Research.

Perchloric Acid (PCA) Extraction of Cells. Neuroblastoma C1300 cells, subclone N1E-115 (Kimhi, 1981), were obtained from the Weizmann Institute (Tel Aviv, Israel). Cultures of primary astroglial cells were prepared from cortex regions of 3-day-old newborn Wistar rats as described previously (Serkova et al., 1996). The cells were incubated at 37°C in Dulbecco's modified Eagle's medium containing 10% fetal calf serum. Experiments were initiated when the cell layers were confluent (N1E-115 after 1 week; primary astrocytes, 4 weeks). Cyclosporine and/or sirolimus at concentrations ranging from 100 to 10,000 µg/liter were incubated with the cells for 12 h at 37°C. For the last 3 h culture medium containing 5 mM [1-¹³C]glucose was used. After removing the medium, the cells were extracted with PCA and the lyophilized extracts were redissolved in ²H₂O for NMR experiments (Serkova et al., 1996). NMR spectra were recorded on a Bruker AM 360 spectrometer (Karlsruhe, Germany). Each ¹H-NMR spectrum represented 200 accumulated scans, ³¹P-NMR spectra contained 5,000 scans, and ¹³C-NMR spectra contained 20,000 accumulated scans. ¹H-NMR and ³¹P-NMR spectra of PCA extracts were used for determination of absolute concentrations of water-soluble cellular compounds such as amino acids, neurotransmitters, osmoregulators, and high-energy phosphates. For the calculation of the metabolite concentrations from NMR spectra, (trimethylsilyl)propionic-2,2,3,3d₄ acid was used as an external standard. The results were correlated with the protein concentrations of cell extracts and given as milligram per kilogram of protein. Changes in glucose metabolism after incubating cells with cyclosporine and [1-¹³C]glucose were calculated from the ratio of ¹³C-labeled trichloroacetic acid (TCA) cycle products to ¹³C-labeled glycolysis metabolites using ¹³C-NMR spectra: [glutamate + glutamine + aspartate-lactate].

Preparation of Rat Brain Slices. Perfused rat brain slices were used for evaluation of effects of cyclosporine and/or sirolimus on high-energy phosphate metabolism. For each NMR experiment, 40 cerebrocortical slices prepared from 10 neonatal Wistar rats (weighing 10-15 g, 8 days old) were pooled. Slices (350-µm thick) were prepared as described previously (McIlwain and Buddle, 1953; Espanol et al., 1998), transferred to a 20-mm-diameter Wilmad NMR tube, and perfused with fresh medium that was in equilibrium with a 95% oxygen/5% CO₂ gas mixture. The perfusion medium was a modified Krebs' balanced salt solution containing 124 mM NaCl, 5 mM KCl, 1.2 mM KH₂PO₄, 1.2 mM MgSO₄, 1.2 mM CaCl₄, 26 mM NaHCO₃, and 10 mM glucose.

Method to Evaluate the Drug Interaction Between Cyclosporine and Sirolimus. Because linearity could not be assumed, the combined effects of cyclosporine and sirolimus on phosphate energy metabolism were analyzed using the algebraic algorithm described by Berenbaum (1977, 1978). This method is the mathematical surrogate of the geometric description of drug interactions by Loewe isobolograms (Berenbaum, 1977, 1978). Synergy indices were calculated using the following equation relating the effect of the doses of the drugs A, B... X to the corresponding equieffective doses A_e, B_e... X_e when the drugs are used alone: < 1 for synergism

This analysis was based on a checkerboard matrix combining the following cyclosporine and sirolimus concentrations: 0, 100, 500, and 5000 µg/liter. High-energy phosphate concentrations were measured in perfused rat brain slices by ³¹P-NMR spectroscopy as described above.

HPLC/Mass Spectroscopy (MS) Analysis of Immunosuppressants in Rat Brain Slices. Because of the low expected tissue concentrations, HPLC/MS was used. The slices were washed and homogenized with 1 M KH₂PO₄ buffer, pH 7.4 (2 ml). One milliliter of homogenate was removed and 28-, 40-diacetyl sirolimus and cyclosporine D were added as internal standards (Streit et al., 1996). The final concentration of the internal standards in the samples was 100 µg/liter. After addition of 2 ml of methanol/ZnSO₄ (80:20, v/v) for protein precipitation, the samples were vortexed for 30 s and centrifuged at 1500g for 3 min. The supernatant was loaded on C₁₈ extraction columns (Bond Elut LRC; Varian, Harbor City, CA) by drawing the samples through columns using a vacuum system. Immunosuppressants and internal standards were eluted using 1.5 ml of methylene chloride. The samples were evaporated to dryness under a stream of nitrogen. The residues were reconstituted in 120 µl of acetonitrile/water (pH 3; 75:25; v/v). The samples were transferred into micro-HPLC vials and were analyzed using a Hewlett-Packard (Palo Alto, CA) HPLC/electrospray-MS system. Concentrations of the immunosuppressants were calculated from an external standard curve and corrected on the basis of their internal standards. Experiments were carried out in triplicate.

Data Analysis. All results are given as means ± S.D. The results were compared by using either unpaired Student's t test (procedure TTest, SAS, Version 6.12; SAS Institute, Cary, NC) or ANOVA in combination with Duncan grouping (procedure GLM; SAS).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Concentration-Dependent Effects of Cyclosporine on Brain Metabolism. The metabolic pathways in neuroblastoma N1E-115 cells and primary astroglial cells showed different sensitivities to cyclosporine (Tables 1 and 2). Incubation with the lowest cyclosporine concentrations, 100 µg/liter for 12 h, led to a significant reduction of intracellular osmoregulators in both neuroblastoma and primary astroglial cells: aspartate decreased by 27 ± 3% [n = 4, P < .05; half-maximal inhibition concentration (IC₅₀) = 50 ± 30 µg/liter] and taurine decreased by 23 ± 9% (n = 4, P < .05; IC₅₀ = 60 ± 70 µg/liter) for neuroblastoma cells; hypotaurine decreased by 14 ± 5% (n = 4, P < .05; IC₅₀ = 330 ± 320 µg/liter) and taurine decreased by 11 ± 7% (n = 4, P < .05; IC₅₀ = 4380 ± 470 µg/liter) for astrocytes. In addition, 100 µg/liter cyclosporine also reduced high-energy phosphate concentrations: NTP by 11 ± 5% (n = 4, P < .05; IC₅₀ = 1110 ± 420 µg/liter) in neuroblastoma cells and PCr by 10 ± 2% (n = 4, P < .05; IC₅₀ = 1850 ± 600 µg/liter) in astroglial cells. Cyclosporine concentrations of 500 µg/liter additionally affected intracellular neurotransmitter concentrations in neuroblastoma cells (Table 1): glutamate concentrations decreased by 22 ± 9% (n = 4, P < .05; IC₅₀ = 360 ± 220 µg/liter) and N-acetylaspartate (NAA), a putative neuronal marker, decreased by 13 ± 7% (n = 4, P < .05; IC₅₀ = 1100 ± 330 µg/liter). Cyclosporine inhibited the TCA cycle and stimulated glycolysis. In neuroblastoma cells in the presence of 10,000 µg/liter cyclosporine, the ratio of ¹³C-labeled TCA cycle products (glutamate, glutamine, aspartate) to ¹³C-labeled glycolysis products (lactate and alanine) was reduced by 66 ± 12% (n = 4, P < .05; IC₅₀ = 3030 ± 150 µg/liter) (Table 1). In primary astrocytes, the [TCA/glycolysis] ratio decreased by 59 ± 9% (n = 4, P < .05; IC₅₀ = 2350 ± 700 µg/liter) (Table 2).

Effects of the Combination of Cyclosporine and Sirolimus on Energy Metabolism of Perfused Rat Brain Slices. High-energy phosphate concentrations were sensitive to sirolimus as well as to cyclosporine in cell cultures. Therefore, the effects of cyclosporine and sirolimus alone and in combination were investigated using ³¹P-NMR analysis of perfused rat brain slices. Figure 1 shows five representative ³¹P-NMR spectra of brain slices from one typical experiment (n = 4). As shown in the control spectrum (Fig. 1A), the signals of PCr, NTP, P_i, and intracellular compounds of cell membrane biosynthesis PME were detected in the brain slices. Perfusion with sirolimus (Fig. 1B) as well as with cyclosporine (Fig. 1C) reduced the concentration of PCr. The combination of sirolimus and cyclosporine reduced the high-energy phosphate concentration to a larger extent than each of the drugs at the same concentration alone (Fig. 1D). After perfusion of the brain slice preparation with fresh medium without immunosuppressive drugs for 30 min, the high-energy phosphate concentrations returned to their original values and were not different from the control (Fig. 1E). The final [PCr/PME] reduction, as calculated from ³¹P-NMR spectra of perfused rat brain slices (representative spectra shown in Fig. 1), was 27 ± 2% after 500 µg/liter cyclosporine and 17 ± 2% after 100 µg/liter sirolimus versus 46 ± 6% after a 2-h combined treatment (n = 6, P < .02, ANOVA). Also, the [NTP/PME] ratio after perfusion with both drugs was significantly more affected than by perfusion with cyclosporine or sirolimus alone at the same concentrations and for the same time: 31 ± 7% after a 2-h perfusion with both drugs in combination versus 16 ± 2% with cyclosporine and 10 ± 3% with sirolimus (n = 6, P < .02, ANOVA). Compared with the untreated controls, perfusion of brain slices with cyclosporine and sirolimus alone or in combination significantly decreased the [NTP/PME] ratio.

View larger version (32K): [in this window] [in a new window]   Fig. 1.   Effect of cyclosporine and sirolimus alone and in combination on high-energy phosphate metabolism in rat brain slices. Five representative 31P-NMR spectra of a typical experiment (n = 4) with perfused rat brain slices are shown. A, 30 min after perfusion with immunosuppressant-free medium (control). B, 120 min after perfusion with 100 µg/liter sirolimus. C, 120 min after perfusion with 500 µg/liter cyclosporine. D, 120 min after perfusion with 100 µg/liter sirolimus and 500 µg/liter cyclosporine. E, 30 min after perfusion with immunosuppressant-free medium (wash-out and recovery).

Cyclosporine and Sirolimus Concentrations in Brain Slices. HPLC/MS analysis of drug tissue concentration in brain slices indicated changes in drug tissue distribution when both drugs were present in the perfusion medium (Fig. 2). After perfusion with 100 µg/liter cyclosporine for 2 h, cyclosporine concentrations in the rat brain slices were 65 ± 13.2 ng/g tissue (mean ± S.D., n = 3). Addition of 100 µg/liter sirolimus increased cyclosporine tissue concentrations 3-fold to 183.4 ± 38.6 ng/g (n = 3, P < .01) (Fig. 2A). In contrast, at the same concentrations, cyclosporine lowered sirolimus tissue concentrations (Fig. 2B), or, at higher concentrations, cyclosporine only slightly increased sirolimus tissue distribution. Perfusion of rat brain slices with 100 µg/liter sirolimus for 2 h resulted in a sirolimus concentration of 363.7 ± 192.8 ng/g tissue. Addition of 500 µg/liter cyclosporine slightly increased sirolimus tissue concentration by 25% to 511.4 ± 118.5 ng/g tissue (n = 6, P < .05). Cyclosporine and sirolimus were washed out during perfusion of the brain slices with drug-free buffer. After 30 min, however, although drug tissue concentrations were reduced significantly, the washout was not yet complete (Fig. 2).

View larger version (30K): [in this window] [in a new window]   Fig. 2.   Concentrations of cyclosporine (A) and sirolimus (B) in brain slices during perfusion with 100 µg/liter cyclosporine, 100 µg/liter sirolimus or a combination of both and during the wash-out period. All values are means ± S.D. (n = 3). Concentrations were measured by HPLC/MS as described in Materials and Methods.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Although well defined experimental models exist for the assessment of immunosuppressive activity (Fahr et al., 1990), no method has been available to measure a wider range of toxic effects of immunosuppressants on cell metabolism and to identify the underlying mechanisms. Because NMR spectroscopy techniques to evaluate biochemical pathways are commonly believed to be characterized by low sensitivity, previous NMR spectroscopy studies on cyclosporine toxicity in vitro used extremely high and therapeutically irrelevant cyclosporine concentrations (Ruiz-Cabello et al., 1994; Serkova et al., 1996).

All effects of cyclosporine on brain cell metabolism using extremely high cyclosporine concentrations as reported previously (Serkova et al., 1996, 1997) were confirmed by our study. However, our study showed that the cyclosporine concentrations required to significantly affect the various biochemical pathways differed. Cyclosporine concentrations as low as 100 µg/liter, which is at the lower limit of cyclosporine trough blood concentrations in transplant patients (Oellerich et al., 1995), reduced the concentrations of aspartate, taurine, and hypotaurine, which are important cell volume regulators and osmoregulators. In addition, high-energy phosphate metabolism was decreased at these low concentrations. Five-fold cyclosporine concentrations were required to cause an additional reduction of intracellular concentrations of neurotransmitters such as glutamate or a putative neuronal marker NAA and to cause an increase in PME concentrations, a marker of intracellular phospholipid catabolites suggesting membrane decomposition. Cyclosporine concentrations more than 20-fold higher than those affecting osmoregulation, as well as high-energy phosphate metabolism and concentrations more than 2-fold higher than cyclosporine peak in the blood of transplant patients, were necessary to alter glucose metabolism by inhibition of the TCA and an increase of glycolysis. Whereas previous studies did not allow for differentiation of the clinically relevant mechanisms (Ruiz-Cabello et al., 1994; Serkova et al., 1996, 1997), our study indicated that, in patients, cyclosporine causes neurotoxicity most likely by affecting volume regulation and high-energy phosphate metabolism. Our in vitro results correlated well with clinical findings in cyclosporine-treated transplant patients with cyclosporine neurotoxicity. In those patients, cyclosporine caused cerebral brain edema, metabolic encephalopathy, and hypoxic neuropathy (Hawley et al., 1990; Pace et al., 1995; Schwartz et al., 1995).

³¹P-NMR spectroscopy of perfused brain slices allows for noninvasive, continuous measurements to analyze time-dependent effects of drugs on high-energy phosphate metabolism of viable cells or organs. Previous studies demonstrated that the results of ³¹P-NMR studies with perfused rat brain slices correlate well with in vivo findings with regard to metabolic activity of brain (Espanol et al., 1992; Hasegawa et al., 1997). In our study, brain slices of newborn rats were used because, in comparison with adult animals, these are more resistant to the ischemic conditions during preparation and because they have better metabolic stability. In a previous study (Serkova et al., 1997), a sirolimus concentration of 5 mg/liter showed effects on brain cell metabolism that were comparable to those caused by the same cyclosporine concentration. This led to the conclusion that, although never reported in clinical studies, sirolimus possesses neurotoxic potential. Our present study showed that the concentrations used in the previous study (Serkova et al., 1997) were far above the linear range of most sirolimus effects on brain cell metabolism. The lack of clinically significant sirolimus neurotoxicity can be explained by the results of our present study. For sirolimus, a minimum concentration of 100 µg/liter was required to significantly affect osmoregulation and high-energy phosphate metabolism, which is higher than the peak blood concentration in most patients (Zimmerman and Kahan, 1997).

Sirolimus is under clinical investigation as an immunosuppressant in combination with cyclosporine. Combination of both drugs led to larger decreases in high-energy phosphates than did each drug alone at the same concentration. In this study, we demonstrated the synergistic nature of the combination of cyclosporine and sirolimus. The decrease in high-energy phosphate concentrations when both drugs were combined, compared with cyclosporine alone, can, at least in part, be explained by the higher intracellular cyclosporine concentration in the presence of sirolimus. In comparison with the concentrations in the perfusion medium, sirolimus and cyclosporine accumulated in the brain tissues. In the body, distribution of cyclosporine and, probably, sirolimus is governed mainly by immunophilins. Immunophilins are present in relatively high concentrations in the central nervous tissue (Steiner et al., 1992), and, therefore in our study, binding to intercellular immunophilins may explain the distribution of cyclosporine and sirolimus from the perfusion medium into brain tissues. Both drugs were clearly washed out of the brain tissues during perfusion with drug-free medium, and the high-energy phosphate concentrations returned to the level of the controls. However, after 30 min, the washout was incomplete. It remains unclear whether the concentrations of sirolimus and/or cyclosporine remaining were insufficient to cause detectable changes of high-energy phosphate metabolism or whether metabolism accommodated to the drug with time. Our study demonstrated the importance of tissue concentration measurement when the effects of drug combinations are studied in tissue slices. Both sirolimus and cyclosporine are p-glycoprotein substrates (Wacher et al., 1995), and it has been shown that p-glycoprotein is a transporter eliminating drugs from brain tissue (Mayer et al., 1997). It can be hypothesized that inhibition of cyclosporine efflux by sirolimus may be involved in the increase of cyclosporine concentration in brain slices. However, no data about the influence of sirolimus on cyclosporine tissue concentrations are available presently.

Brain tissue concentrations are more important than blood concentrations to decide whether or not the results of our in vitro study are of clinical relevance. Lensmeyer et al. (1991), who used an HPLC/UV assay with a minimum quantitation limit of 50 µg/liter, and Sangalli et al. (1989) failed to detect cyclosporine and/or its metabolites in the central nervous system of transplant patients and of experimental animals. However, in a study in monkeys using a much more sensitive and specific HPLC/MS assay, cyclosporine concentrations of 55 ± 23 ng/g cyclosporine (n = 4, mean ± S.D.) were found in the brain (N. Serkova, B. Hausen, R. E. Morris, L. Z. Benet, U. Christians., unpublished data). The corresponding blood concentrations were 207 ± 11 µg/liter and in the target range of transplant patients (Oellerich et al., 1995). The monkeys did not show any signs of neurotoxicity, but the brain tissue concentrations were only 2-fold lower than those required to significantly affect osmolarity and high-energy phosphate concentrations in our in vitro study.

The immunosuppressive action of cyclosporine results from binding to the intracellular immunophilin cyclophilin. The cyclosporine/immunophilin complex inhibits activity of the calcium-activated phosphatase calcineurin, resulting in accumulation of phosphorylated calcineurin substrates in the cell including the nuclear factor of activated T cells, which is active only in the nonphosphorylated state. Sirolimus binds to a different family of intracellular immunophilins from cyclosporine, the FK-binding proteins. In contrast to cyclosporine, the sirolimus/immunophilin complex does not interact with calcineurin but binds to the mammalian target of rapamycin, which regulates p70 S6 kinase. Because sirolimus and cyclosporine both reduce the concentration of high-energy phosphate, this may indicate that reduction of high-energy phosphate concentrations in brain slices was independent of calcineurin. It is well established that cyclosporine in the presence of calcium inhibits mitochondrial ATP synthesis and enhances mitochondrial calcium uptake and storage (Salducci et al., 1992, 1995, 1996). The effect of sirolimus on high-energy phosphate concentrations has not yet been studied in detail. Therefore, it is unclear whether or not the underlying mechanism is similar to that of cyclosporine. The enhancement of cyclosporine distribution into the brain tissue in the presence of sirolimus most likely contributed to the synergistic effect of both drugs on high-energy phosphate concentrations. However, the results of our study do not allow for further identification of the mechanism underlying a potential pharmacodynamic component of the synergistic interaction between sirolimus and cyclosporine.

An effect of cyclosporine on ATP concentrations has been reported for several organs, such as kidney (Ruiz-Cabello et al., 1994) and lymphocytes (Karlsson et al., 1997). As in brain tissue, sirolimus also may enhance the cyclosporine-induced reduction of high-energy phosphate concentrations in other tissues. Although in our study, as discussed above, the sirolimus concentrations required to cause a significant reduction of high-energy phosphate concentrations were higher than the sirolimus peak blood concentrations usually found in transplant patients, the possibility that sirolimus enhances cyclosporine toxicity because of the synergistic effect must be taken into account. Andoh et al. (1996) reported synergistic nephrotoxicity of sirolimus and cyclosporine at subtherapeutic concentrations in the rat, and a synergistic effect on high-energy phosphate concentrations may be a possible explanation. However, these findings have not yet been confirmed by the results of phase I and II clinical cyclosporine/sirolimus studies (Kahan, 1997).

With respect to transferring our results of a synergistic toxic interaction between sirolimus and cyclosporine from this in vitro study to the in vivo situation in patients, several restrictions must be taken into account. As discussed, brain slices from newborn rats were used because of their greater resistance against ischemia, indicating significant age-dependent metabolic differences in rats. In addition, species-specific differences of sirolimus toxicity are well established (Collier et al., 1990, 1991), and in a perfused in vitro system potential factors such as pharmacokinetics, tissue distribution, protein binding, and an intact blood-brain barrier are ignored. Our study showed that sirolimus possesses neurotoxic potential at concentrations above, but close to, those found in blood of sirolimus-treated patients and has a synergistic neurotoxic effect in combination with cyclosporine. However, a decision about whether or not our results are of clinical relevance for patients treated with a combination of cyclosporine and sirolimus is not warranted.

It is concluded that NMR spectroscopy is a potent and sensitive method with which to investigate neurotoxicity of immunosuppressive drugs in cell cultures and tissue slices. The biochemical mechanisms involved in cyclosporine toxicity are concentration dependent. Cyclosporine only has a significant effect on osmoregulation and high-energy phosphate metabolism in rat brain tissue at concentrations that are of therapeutic relevance. Addition of sirolimus synergistically enhanced the effects of cyclosporine on brain cell metabolism, which, in our study, could be explained partially by an increase of cyclosporine tissue concentrations in the presence of sirolimus.