Determination of the Excitatory Potencies of Fluoroquinolones in the Central Nervous System by an In Vitro Model

Abstract

Fluoroquinolones have been reported to induce central nervous system side effects, including seizures and psychiatric events. Although relatively rare in patients up to now, the proconvulsant activity depends on the chemical structure and might be a critical endpoint of some new representatives of this valuable class of antimicrobials. The electrophysiological determination of field potentials in the CA₁ region of the rat hippocampus slice allowed an assessment of the excitatory potential of fluoroquinolones and might be predictive for their neurotoxic potency in vivo. An optimization of this method and its extension to other fluoroquinolones resulted in a defined rank order. Well-known already-marketed quinolones as well as some fluoroquinolones under evaluation and development were used. The dose range tested was between 0.5 and 4 μmol/liter, which was comparable to the therapeutic concentration in the brain. All tested compounds increased the population spike amplitude in a concentration-dependent manner, and the resulting excitatory potency was highly dependent on the chemical structure, with compounds ranging from least to most excitatory as follows: ofloxacin, ciprofloxacin, nalidixic acid, moxifloxacin (= BAY × 8843), fleroxacin, lomefloxacin, enoxacin, clinafloxacin (much more excitatory than enoxacin), tosufloxacin, trovafloxacin, BAY 15-7828, and BAY × 9181 (much more excitatory than BAY 15-7828). The proposed hippocampus slice model not only is suitable for giving valuable alerts as to convulsive potential during candidate selection but also enables mechanistic investigations. These investigations pointed to the N-methyl-d-aspartate receptor as the probable target of the fluoroquinolone effects.

Quinolones have been reported to induce central nervous system (CNS) reactions in humans with an incidence of 0.9 to 2.1%. However, severe psychiatric or neurological reactions like hallucinations, periods of depression, nightmares, or convulsions are relatively rare (5). Convulsive seizures are reported after quinolone treatment mostly in the elderly or in patients with a history of epilepsy, cerebral trauma, or alcohol abuse (5). The proconvulsant activity of fluoroquinolones depends on the chemical structure and might be a critical endpoint of some new representatives of this valuable class of antimicrobials. It therefore has to be considered during preclinical development, either in animal studies or in vitro. Up to now, most animal models have failed to provide a reliable ranking of these compounds that is relevant to humans. As an in vitro model for assessing the excitatory potential of fluoroquinolones, the determination of evoked potentials, measured by extracellular recordings in the hippocampus slice of the rat, has been proposed (9). In this model, enoxacin, nalidixic acid, and ofloxacin were the most potent convulsants, whereas ciprofloxacin, norfloxacin, and pefloxacin were less active. With the exception of pefloxacin, these results were in agreement with the results of studies in DBA/2 mice, a strain genetically susceptible to sound-induced seizures (8), and also with data for CNS side effects in patients (20).

Different drugs, such as methylxanthine derivatives or nonsteroidal anti-inflammatory drugs, potentiate the convulsant activity of fluoroquinolones. The adenosine or γ-aminobutyric acid (GABA_A) receptor has therefore been proposed as a possible target for quinolones (12). In addition, interactions of quinolones with the dopamine and opiate receptors were also postulated (5, 27, 31). However, receptor binding studies using radioactive ligands for these receptors failed to identify a rank order for the different quinolones that was predictive of the in vivo situation (1, 12).

The structural similarities of fluoroquinolones to kynurenic acid and other similar compounds which are endogenous ligands of the glutamate receptor might suggest an interaction of quinolones with ligand-gated glutamate receptors. Accordingly, the proconvulsive action of quinolones is antagonized by AP-5 or AP-7, selective antagonists of the glutamate binding site of the N-methyl-d-aspartate (NMDA) receptor (27, 39). However, in receptor binding studies with [³H]glutamate, [³H]kainate, α-[³H]amino-3-hydroxy-5-methylisoxazole-4-propionic acid), and [³H]NMDA, no specific affinity of quinolones for the ion- or ligand-gated glutamate receptors has been found (12, 17, 34).

Thus, receptor binding studies of the relevant neuronal receptor types failed to predict the convulsive potency of the different fluoroquinolones up to now. A more sophisticated mechanistic model for investigating this effect, one which reflects the cell-cell interaction within the CNS properly, has been needed. The electrophysiological determination of the evoked field potentials in the CA₁ pyramidal cell layer of the hippocampus slice may represent such a mechanistic model, because (i) the complex functional architecture of the hippocampus is retained, (ii) it contains a well-defined principal cell population expressing the receptor types (GABA and NMDA) in question, and (iii) the endpoint determined in vitro is closely related to the seizure threshold responsible for some of the CNS side effects. The hippocampus slice model was therefore used, after some technical improvements, to investigate the excitatory potency of a broad range of fluoroquinolones in order to obtain a more-reliable ranking with respect to the different chemical structures and to get more insight into the possible mechanism of these effects.

MATERIALS AND METHODS

Chemicals.

The tested fluoroquinolones—ofloxacin, ciprofloxacin, fleroxacin, clinafloxacin, lomefloxacin, tosufloxacin, moxifloxacin, BAY × 9181 {7-[1-ami- nomethyl-2-oxa-7-aza-bicyclo(3.3.0)oct-7-yl]-1-cyclopropyl-6,8-difluoro-1,4- dihydro-4-oxo-3-quinolinecarboxylic acid hydrochloride}, BAY × 8843 {1-cyclopropyl-7-[(S,S)-2,8-diazabicyclo[4.3.0]non-8-yl]-6,8-difluoro-1,4- dihydro-4-oxo-3-quinolinecarboxylic acid hydrochloride}, BAY 15-7828 {7-[(3aS,7aR)-3a-amino-1,2,3,7a-tetrahydro-isoindol-2-yl]-8-chloro-6-fluoro- 1-[(1R,2S]-2-fluorocyclopropyl]-1,4-dihydro-4-oxo-3-quinolinecarboxylic acid}, and trovafloxacin (mesylate and hydrochloride)—were obtained from Bayer (Leverkusen, Germany). Enoxacin, nalidixic acid, and norfloxacin were purchased from Sigma (Deisenhofen, Germany). The purity of all fluoroquinolones was ≥99%. All compounds were dissolved in 0.1 N HCl to a concentration of 2 mmol/liter; afterwards, the stock solutions were diluted in artificial cerebrospinal fluid (ACSF) (see below) to their end concentrations (0.5 to 4 μmol/liter).

Equipment.

The orthodromic recordings of the evoked potentials in the hippocampus slice were driven by a specific computer program (Institute for Neurobiology, Madgeburg, Germany). Via a stimulation isolator (Hi-Med Isolator HG 203; List Electronics, Tamm, Germany), the electric stimulation occurred. The evoked field potentials were amplified by two systems (SEC-10L and EXT-01 preamplifiers and DPA 2F amplifier; npi Electronic GmbH, Tamm, Germany), were converted by a 1401 interface (Cambridge Electronic Design via Science Products, Hofheim, Germany), and were visualized by an oscilloscope (DRO 1604; Gould Electronics GmbH, Diezenbach, Germany).

Evoked potentials.

Hippocampus slice preparations were obtained from female rats (Wistar; Harlan Winkelmann). The animals were anesthetized with halothane (Hoechst, Frankfurt, Germany). The brains were removed immediately, the hippocampus was isolated, and the middle part was cut immediately in 450-μm-thick slices (four to six slices per middle part of each hippocampus). The slices were transferred to a chamber where they remained for 1 h before the experiment was started. The experiment was done in a superfusion chamber (modified McIlwain chamber; Fine Science Tools, Heidelberg, Germany) under defined conditions (34°C; 2-ml/min superfusion rate). The slices were stored and superfused with ACSF medium containing 124 mM NaCl, 5 mM KCl, 1.25 mM NaH₂PO₄, 26 mM NaHCO₃, 10 mM glucose, 2 mM MgSO₄, and 2 mM CaCl₂ (all constituents from Merck, Darmstadt, Germany). The medium was saturated with a 95% O₂–5% CO₂ mixture. Stimulation (stimulation electrode, SNEX-100; Science Products) occurred in the area of the Schaffer collaterals in the CA₂ area; the recording (glass capillaries, GB150TC-10; Science Products) was performed from the pyramidal cell layer of CA₁.

The electric stimulation was performed by 10 pulses with a pulse width of 200 μs, a pulse interval of 10 s, a potential of 3 to 8 V, and an amplitude of ∼1 mV. The following time schedule was used for every compound: 30 min of treatment with ACSF, 30 min of treatment with quinolone (0.5 to 4 μM), and 30 min in ACSF as wash out phase.

Additional experiments with different concentrations of Mg²⁺ (1.75 to 3.5 mM), MK 801 (dizocilpine) (0.1 to 10 μg/ml), or d-serine (100 and 400 μM) alone or in combination with clinafloxacin were done according to the same time schedule described above: 30 min of treatment with ACSF, 30 min under experimental conditions (Mg²⁺, MK 801, or clinafloxacin), and 30 to 60 min in ACSF as wash out phase. Each experiment was done with six slices from two individual animals (three slices from each animal).

The influence of fluoroquinolones on the evoked field potentials in the hippocampus slice of rats was investigated according to the method of Dimpfel et al. (9). In principle, the Schaffer collaterals projecting to the neurons in the CA₁ region of the hippocampus were stimulated extracellularly in the CA₂ region by single pulses. The response of the neurons in the CA₁ region, the amplitude of the field potential (population spike), was recorded by an extracellular electrode. An increase of the amplitude compared to that with untreated control was indicative of increased excitability of these neurons.

We modified the method in order to obtain improved reproducibility. The main differences from the original method described by Dimpfel et al. (9) were as follows. We used a modified McIlwain chamber instead of a Haas chamber, which allowed a more intense superfusion of the slice with the substance-containing buffer. As a consequence, a deeper recording position (250 to 300 μm from the surface) in the center of the slice was possible, and its reproducibility was improved by searching the optimal somatic layer stepwise with a nanostepper system (SPI, Oppenheim, Germany). Additionally, the study design was modified. After a stabilization period of 30 min, a control phase of 30 min with ACSF superfusion was recorded. After this time, the superfusion with the quinolone-containing ACSF was started and lasted for a period of 30 min, followed by a wash out phase of 30 to 60 min with ACSF. Depending on the physicochemical properties of the substance, the peak of the excitatory effect was even reached in the wash out period.

Statistical analysis.

A statistical analysis was performed using a one-way analysis of variance (ANOVA) followed by a t test (Student-Newman-Keuls method) (Sigma Stat; Jandel, Erkrath, Germany).

RESULTS

Extracellular evoked field potentials.

The fluoroquinolone concentrations in the in vitro model were defined according to pharmacokinetic considerations. The maximum therapeutic concentration of the fluoroquinolones in plasma is <30 μmol/liter. In cerebrospinal fluid approximately 10% of this concentration was reached (6, 23, 25, 30). Therefore, concentrations of 0.5 to 4 μmol/liter were used in our investigations.

The fluoroquinolones tested increased the population spike amplitude of the pyramidal cells in the CA₁ region of the hippocampus in a concentration-dependent manner (Fig. 1). At 4 μmol/liter, some of the compounds (clinafloxacin, trovafloxacin, and BAY × 9181) were strongly active in this model and irreversibly damaged the system or were unsoluble in the superfusion fluid (BAY 15-7828). For comparing an extended number of compounds in the screening model, a fixed concentration of 2 μmol/liter was therefore used (Table 1). Most of the fluoroquinolones as well as the new 8-methoxyfluoroquinolone moxifloxacin increased the population spike amplitude only moderately. The resulting ranking was ofloxacin, ciprofloxacin, nalidixic acid, moxifloxacin, BAY × 8843, fleroxacin, lomefloxacin, and enoxacin. Some other fluoroquinolone derivatives increase the population spike amplitude more distinctly, with the ranking clinafloxacin, tosufloxacin, trovafloxacin (mesylate and hydrochloride), and BAY 15-7828. The greatest population spike (increase of more than 400% of the control) was observed for the experimental fluoroquinolone BAY × 9181, indicating that its convulsive potential might be extremely high. This also demonstrates that the effect range of the hippocampus slice model (130 to 400%) with different fluoroquinolones was markedly higher than reported before (9).

FIG. 1.

Dose-response curves of selected fluoroquinolones. The concentration range tested was 0.5 to 4 μmol/liter. Each compound and concentration was tested with six individual brain slices from two animals. A statistically significant (P < 0.01) increase of the population spike amplitude (more than 125 to 130% in relation to the control level) could be shown in all experiments. Statistical analysis was done by one-way ANOVA followed by a t test.

TABLE 1.

Effects of different fluoroquinolones on population spike amplitude^a

Compound	Population spike amplitudeb	SD	95% Confidence interval
Ofloxacin	130	9.5	120.0–140.0
Ciprofloxacin	155	12.1	142.3–167.7
Nalidixic acid	165	1.7	163.2–166.8
Moxifloxacin	170	4.7	165.1–174.9
BAY × 8843	170	8.6	161.0–179.0
Fleroxacin	172	4.2	167.6–176.4
Lomefloxacin	180	8.6	171.0–189.0
Enoxacin	192	13.8	177.5–206.5
Clinafloxacin	233	12.6	219.8–246.2
Tosufloxacin	250	2.3	247.6–252.4
Trovafloxacin (mesylate)	250	12.0	237.4–262.6
Trovafloxacin (hydrochloride)	270	9.0	260.5–279.5
BAY 15-7828	281	15.5	264.7–297.3
BAY × 9181	419	16.2	402.0–436.0

^aThe concentration of each test compound was 2 μmol/liter. Each compound was tested with six individual brain slices from two animals. A statistically significant (P < 0.01) increase of the population spike amplitude (more than 125 to 130% in relation to the control level) could be shown in all experiments. Statistical analysis was done by one-way ANOVA followed by a t test. 

^bPercent of control. 

Mechanistic investigations of fluoroquinolone effects on field potentials.

The increase of the population spike amplitude could in principle be due to activation of the NMDA receptor on the neurons of the CA₁ region or, alternatively, to a reduction of the activity of the GABAergic inhibitory interneurons by GABA_A antagonism.

The activity of the NMDA receptor-gated Ca²⁺ channel is modulated by Mg²⁺ and may be blocked by the specific ligand MK 801. In order to substantiate the influence of fluoroquinolones on the NMDA receptor, the effects of Mg²⁺ and MK 801 in combination with fluoroquinolones were investigated.

As expected, a reduced concentration of Mg²⁺ alone in the ACSF increased the population spike amplitude, whereas an increased concentration depressed the response (Fig. 2a). Clinafloxacin (2 μmol/liter) induced an increase of the population spike amplitude of 233% related to the control level (100%) at the physiological Mg²⁺ concentrations of 2 mmol/liter. In combination with increasing concentrations of Mg²⁺, the population spike amplitude induced by clinafloxacin (2 μmol/liter) was reduced to the control level (Fig. 2b). It is important that this happened already at Mg²⁺ concentrations, which affected the control population spike amplitude only slightly (Fig. 2a). In contrast, a slight decrease of the Mg²⁺ concentrations (1.75 mmol/liter) potentiated very strongly the clinafloxacin effect.

FIG. 2.

Effects of Mg²⁺ concentration on population spike amplitude. The standard concentration of Mg²⁺ in ACSF was 2 mM. The Mg²⁺ concentration was varied between 1.75 and 3.5 mmol/liter. Each experiment was done with six individual brain slices from two animals. Statistical analysis was done by one-way ANOVA followed by a t test. (a) Mg²⁺ alone; (b) Mg²⁺ with 2 μM clinafloxacin.

d-Serine (100 and 400 μmol/l) alone and in combination with different concentrations of Mg²⁺ (1.75 and 3 mmol/liter) (Fig. 3) was used to demonstrate the effect of Mg²⁺ on the effects of ligands of the glycine binding site of the NMDA receptor. A possible effect of fluoroquinolones on this binding site was postulated by Dimpfel and coworkers (10). In contrast to these authors, we detected no effects with d-serine alone, even at very high concentrations. This was in agreement with the findings of Peeters and Vanderheyden (29) and Huettner (21), who demonstrated that glycine or d-serine had no effect by itself but could modulate the NMDA receptor channel opening times. Only in combination with low Mg²⁺ concentrations and at high d-serine concentrations could a slight modulative action of d-serine be demonstrated. This effect was not comparable to the results with clinafloxacin in combination with low Mg²⁺ concentrations (Fig. 2b), where a threefold increase of the population spike amplitude was observed. This might indicate that the underlying mechanisms were different.

FIG. 3.

Effects of d-serine and Mg²⁺ on population spike amplitude. d-Serine concentrations of 100 and 400 μmol/liter and control concentrations were tested. Each experiment was done with six individual slices from two animals. Statistical analysis was done by one-way ANOVA followed by a t test.

Alternatively MK 801 was used to block the NMDA receptor-gated ion channel. MK 801 at 1 and 10 μg/ml had no effect on the control population spike amplitude, whereas at a concentration of 50 μg/ml it decreased the amplitude strongly (Fig. 4a). Therefore, MK 801 concentrations of 0.1 to 10 μg/ml were used in order to influence the excitatory effects of clinafloxacin (2 μmol/l) (Fig. 4b). Blocking the ion channel of the NMDA receptor by MK 801 counteracts the effects of quinolones in a concentration-dependent manner. MK 801 at 1 and 10 μg/ml abolished the excitatory effect of clinafloxacin completely, whereas MK 801 at 0.1 μg/ml already reduced it by about 50%. Although not conclusive, both experiments point to a direct involvement of the NMDA-gated ion channel in the exitatory effects of fluoroquinolones.

FIG. 4.

Effects of MK 801 on population spike amplitude. Each experiment was done with six individual slices from two animals. Statistical analysis was done by one-way ANOVA followed by a t-test. (a) MK 801 alone (at 1, 10, and 50 mg/kg); (b) MK 801 (at 0.1, 1, and 10 μg/ml) with clinafloxacin (2 μmol/liter).

DISCUSSION

The hippocampus exhibits a unique sensitivity to cell injury resulting from hypoxia, seizures, and neurotoxic compounds. Hippocampal vulnerability has been shown anatomically, biochemically, electrophysiologically, and behaviorally (13, 14). Therefore, the hippocampus is considered to be a relevant model for toxicological or pharmacological investigations in vitro and in vivo. The method using mammalian brain slices as physiologically intact models dates back to 1966 (40). The interest in the technique has increased in recent years, because in this in vitro model the complex physiology, e.g., the interconnections between neurons as well as between neurons and glial cells, is maintained. The hippocampus is thought to be the structure in the brain with the lowest seizure threshold (3). It has therefore seen widespread use as a fairly simple model for the study of cellular mechanisms of acute epilepsy by measuring extracellular field potentials.

Field potentials reflect the summated output of an entire population of neurons, including both excitatory and inhibitory influences. Excitatory neurotransmission in the hippocampus is mediated mainly by glutamate, for which different receptors exist. The most abundant excitatory acting receptor in the CA₁ region of the hippocampus is the NMDA subtype. The NMDA ionophore is blocked by Mg²⁺ in a voltage-dependent fashion. The Mg²⁺ block is relieved by a large or prolonged cell depolarization (11, 26). The inhibitory interneurons in the hippocampus are predominantly GABAergic, with two subclasses, GABA_A and GABA_B receptors, both hyperpolarizing the cells (4).

The CA₁ region of the hippocampus is considered an optimal region for studying the excitatory potency of test substances, because the neurotransmitter receptors involved in the generation of field potentials are well characterized and mechanistically related to the endpoint in question in vivo.

By an extensive number of fluoroquinolones tested in this model it was shown that all fluoroquinolones dose dependently increased the population spike amplitude of the neurons in the CA₁ region of the hippocampus. The observation is qualitatively in agreement with the observed convulsant potential of some fluoroquinolones in humans (20). It is, however, important to state that proconvulsant activity might be only one aspect of the CNS effects of fluoroquinolones. With respect to this endpoint, a ranking of the substances tested based on the in vitro data (population spike, 130 to 400% of control) was possible. This ranking indicated that considerable differences between the fluoroquinolones exist. However, a detailed correlation of the in vitro activity to the in vivo findings is hampered by the fact that in vivo data under comparable experimental conditions are largely lacking. Therefore, for two fluoroquinolones, moxifloxacin and BAY 15-7828 (population spikes, 170 and 281%, respectively, of control in vitro) an orientating study in a rhesus monkey was performed as an example. Following intravenous administration of 30 mg of BAY 15-7828/kg of bodyweight, the monkey showed clear signs of CNS toxicity, with somnolence, nystagmus, and seizures, whereas intravenously administered moxifloxacin at 45 mg/kg induced no clinically detectable signs of neurotoxicity, indicating that the ranking found in vitro is also of relevance in vivo. Unfortunately, results in humans are available only for narrow-spectrum fluoroquinolones, which showed a very tight clustering of population spike amplitudes in vitro (130 to 192% of control) and in fact are also relatively similar in their incidence of CNS side effects (<2%) (20). Given the narrow range of the intrinsic excitatory potency of these compounds, other factors, such as the steepness of the concentration-response curve and pharmacokinetics, significantly influence their in vivo response. Therefore, the in vitro model used allowed no further differentiation of the expected in vivo convulsant activity for these compounds with relatively similar intrinsic activities. This similarity may also account for some inconsistencies in the ranking obtained in different studies either in vitro or in vivo with these fluoroquinolones (8, 10, 20). However, the hippocampus slice model indicated a higher excitatory potency for some newer fluoroquinolones, such as clinafloxacin and trovafloxacin. By accumulating clinical experience with these compounds, further insight into the relevance of this model to the situation in humans should become available. For the time being, the proposed hippocampus slice model represents a rapid in vitro model for quantifying the excitatory potency of fluoroquinolones, which may give valuable alerts as to convulsive potential during drug candidate selection. In addition, this model will in principle allow more-detailed studies for structure-activity relationship for this effect. Unfortunately, the marked structural diversity of the fluoroquinolones tested did not yet allow final conclusions on the structural prerequisites influencing the excitatory potency.

As outlined above, the hippocampus slice might be used for mechanistic studies on the excitatory effects of fluoroquinolones since the relevant receptors are present and functionally active. The increased excitability may in principle be due to inhibition of the activity of the GABAergic interneurons or to activation of the NMDA receptor. It has been shown in various receptor binding studies as well as patch-clamp investigations that quinolones have no or only a weak affinity to the GABAergic system (2, 15, 16, 18, 22, 33, 37). In contrast it was shown by our experiments that MK 801, a selective channel blocker of the NMDA receptor, abolishes the excitatory effects of clinafloxacin, thus strongly suggesting the involvement of the NMDA channel in its effects in the hippocampus slice. MK 801 has also been reported to antagonize the proconvulsive action of fluoroquinolones in male mice (19, 28, 39).

However, it has also been shown that fluoroquinolones did not bind to the glutamate or glycine binding site of the NMDA receptor (12, 17). A functional agonism to the glycine binding site has therefore been postulated for the fluoroquinolones (10) but could not be reproduced in the present study. In contrast to Dimpfel’s findings, no excitatory effects of glycine or other agonists of this site, such as d-serine or d-alanine, could be demonstrated electrophysiologically (21, 29). With very high concentrations (10 mmol/liter), a reduction of the population spike amplitude occurred (38). The modulatory action of agonists to the glycine binding site should become obvious by enhancing normal signals related to the channel. This was shown by the slight increase of the population spike amplitude under low Mg²⁺ concentrations. However, this effect was not comparable at all to the strong effect of clinafloxacin under these conditions. Interestingly, MK 801 failed to antagonize the effects of the agonists of the glycine site, whereas 7-chlorokynurenic acid was an excellent antagonist of this site (32, 38).

Therefore, based on the observations that (i) fluoroquinolones do not bind to the glycine site, (ii) glycine site agonists do not or only slightly potentiate the population spike, even under low Mg²⁺ concentrations, and (iii) fluoroquinolone effects but not glycine site agonist effects are completely antagonized by MK 801, we conclude that fluoroquinolones probably do not act via an agonism of the glycine binding site.

Interestingly, Tanaka et al. (36) showed that fluoroquinolones decreased the blocking effects of Mg²⁺ and MK 801 binding to the receptor channel. They therefore characterized the fluoroquinolones as “open channel blockers.” This is supported by our findings on the effect of Mg²⁺ on the population spike amplitude, again underlining the involvement of the NMDA receptor in the fluoroquinolones’ convulsive action. Considering the Mg²⁺ chelating properties of fluoroquinolones, which have been also postulated as a mechanism for fluoroquinolone action in juvenile cartilage (7, 24, 35), it is tempting to speculate that the excitatory potency of fluoroquinolones might be based on activation of the NMDA receptor by abolishing the Mg²⁺ block in the ion channel. This would prolong the opening time of the channel, thus increasing intracellular Ca²⁺ concentrations and the excitability of the neuron.

In summary, the determination of field potentials in the hippocampus slice offers a fairly simple mechanistic model for the convulsive properties of fluoroquinolones. It not only is useful for a rapid screening of fluoroquinolones with regard to this relevant endpoint but also enables more-detailed insight into the underlying mechanisms of this effect.