Fipronil, a phenylpyrazole insecticide introduced in 1993 (Moffat, 1993), is a potent blocker of GABA receptors, which are believed to be its primary target site. Because of its high insecticidal activity against many insects and other arthropod pests and low toxicity to mammals (Tingle et al., 2003), fipronil has found many uses in agriculture, veterinary medicine, and urban pest control. The oral LD₅₀ of fipronil in rats was estimated to be 91 mg/kg, and the LD₅₀ values in insects were 0.07 mg/kg in corn rootworm (Scharf and Siegfried, 1999) and 0.13 mg/kg in housefly (Hainzl and Casida, 1996). Thus, the sensitivity of insects to fipronil is 700- to 1300-fold higher than that of rats. At the level of GABA receptors, fipronil also displays significant insect selectivity: the IC₅₀ against cockroach GABA receptors of fipronil was 59-fold lower than that of rat GABA_A receptors (Zhao et al., 2003). Therefore, the differences in GABA receptor sensitivity to fipronil can only partially explain the selective toxicity between mammals and insects. We propose that differences in the fipronil sensitivity of another target site may also play a role in the selective toxicity.

The cyclodiene insecticide dieldrin is also known to act at the GABA receptor (Nagata and Narahashi, 1994). A single gene conferring 4000-fold resistance to dieldrin (Rdl) was identified and found to code for a GABA-gated chloride channel subunit (ffrench-Constant et al., 1993). Dieldrin resistance was associated with a point mutation of Ala to Ser at position 302 in the second transmembrane domain, which is thought to line the pore of the channel. Dieldrin resistance is often associated with much lower levels of fipronil resistance (Cole et al., 1995; Tingle et al., 2003). Furthermore, after several years of use in more than 60 countries, the field performance of fipronil has not been hampered by resistance. The Ala-Ser mutation of the Drosophila (DMRd1) GABA receptor is not associated with reduced fipronil sensitivity (Hosie et al., 1995), but equivalent mutation of Heliothis GABA receptor (HVDRL) showed reduced sensitivity to fipronil (Wolff and Wingate, 1998). Thus, a target site other than Rdl GABA receptor might be involved in the effectiveness of fipronil on dieldrin-resistant insects.

Glutamate-gated chloride channels (GluCls) are also a potential insect-specific target of fipronil because of their unique presence in invertebrates but not in vertebrates (Cleland, 1996; Raymond and Sattelle, 2002). Recent studies showed that fipronil exhibited inhibitory actions on GluCls (Cully et al., 1994; Horoszok et al., 2001; Ikeda et al., 2003), but the results are not quantitative enough to compare with the effects on GABA receptors. At least two types of GluCls, one picrotoxin-insensitive and the other picrotoxin-sensitive, were found to coexist in cockroach neurons (Raymond et al., 2000), and these were subsequently characterized as desensitizing and nondesensitizing, respectively (Ikeda et al., 2003). Fipronil only modestly inhibited the desensitizing GluCl with an IC₅₀ of 730 nM (Ikeda et al., 2003), but its action on the nondesensitizing GluCl has not yet been examined in detail.

The differential sensitivities of the desensitizing and nondesensitizing GluCls to fipronil were examined in cockroach thoracic ganglion neurons using the whole-cell patch-clamp technique. Fipronil blocked nondesensitizing GluCls much more potently than it blocked the desensitizing GluCls. Furthermore, the block of nondesensitizing GluCls by fipronil was shown for the first time to occur primarily through an open channel blocking mechanism. When coapplied with glutamate to nondesensitizing GluCls, fipronil blocked these receptors more potently than it blocked GABA receptors. It is concluded that the selective blocking action of fipronil on nondesensitizing GluCls is more significant than previously thought and is probably one of the potential mechanisms accounting for its selective toxicity between insects and mammals and its high activity against dieldrin-resistant insects.

Materials and Methods

Preparation of Cockroach Neurons. Adult American cockroaches, Periplaneta americana, were purchased from Carolina Biological Supply Company (Burlington, NC) and maintained at 29°C with free access to water and food. Isolation of neurons from cockroaches was performed using enzymatic digestion and mechanical dissociation (Alix et al., 2002). Briefly, a cockroach was immobilized with pins dorsal side up and was dissected to access the ventral nerve cord. Three thoracic ganglia were isolated and immersed in the normal cockroach saline solution containing 200 mM NaCl, 3.1 mM KCl, 4 mM MgCl₂, 20 mM d-glucose, and 10 mM HEPES acid (pH 7.3). The ganglia were then incubated for 30 min at room temperature (22–24°C) in cockroach saline solution supplemented with collagenase (type A, 0.5 mg/ml, Roche Diagnostics, Mannheim, Germany) and hyaluronidase (type I-S, 1 mg/ml, Sigma-Aldrich, St. Louis, MO). The ganglia were then rinsed twice in normal saline solution supplemented with 5 mM CaCl₂ and fetal calf serum (5% by volume) and were mechanically dissociated by gentle repeated trituration through a fire-polished Pasteur pipette. The dissociated neurons, suspended in the supplemented normal saline solution, were allowed to settle on cover slips coated with poly-l-lysine hydrobromide (mol. wt. >30,000, Sigma-Aldrich). The neurons were incubated at 24°C overnight before experiments. The resting potential of dissociated neurons was -75.2 ± 2.3 mV (n = 36). To prevent fipronil contamination, each cell culture dish was used only once, and the recording chamber was thoroughly washed with alcohol after each experiment.

Whole-Cell Current Recordings. Neurons were continuously perfused with cockroach external solution containing 167 mM NaCl, 3.1 mM KCl, 33 mM d-gluconic acid, 5 mM CaCl₂, 4 mM MgCl₂, and 10 mM HEPES acid (pH 7.4). The osmolarity was 420 mOsm. Ionic currents were recorded using the whole-cell patch clamp technique at room temperature (23°C). Pipette electrodes were made from 1.5 mm (o.d.) borosilicate glass capillary tubes and had a resistance of 2 to 3 MΩ when filled with the standard internal solution containing 15 mM NaCl, 170 mM KCl, 0.5 mM CaCl₂, 1 mM MgCl₂, 10 mM EGTA, 10 mM phosphocreatine diTris, 20 mM HEPES acid, and 3 mM ATP-Mg²⁺. The pH was adjusted to 7.4 with KOH, and the osmolarity was 420 mOsm. The membrane potential was clamped at -60 mV unless otherwise stated. The recording of whole-cell currents began 10 min following membrane rupture so that the cell interior milieu was adequately equilibrated with the pipette solution. Currents through the electrode were recorded with an Axopatch-200A amplifier (Axon Instruments, Inc., Union City, CA), filtered at 2 kHz, and stored by a PC-based data acquisition system that also provided preliminary data analysis. Data, when quantified, were expressed as the mean ± S.E.M.

Drug Application. A U-tube was used for rapid, controlled application of external solution containing glutamate alone or with blockers. The U-tube was a loop of fine tubing with a single hole that was placed near the cell. External solution containing glutamate alone or with blockers was fed through the tube by gravity from a container located above the bath. Closure of a computer-operated solenoid valve in the outlet side of the tube allowed the U-tube solution to flow out of the hole located near the cell. Another valve controlling a suction tube with an opening on the other side of the cell was also opened, allowing the test solution to be sucked away quickly. With this method, the external solution surrounding the cell could be completely exchanged with a test solution within 30 ms. Test compound was either coapplied with glutamate in the U-tube or added to the bath.

Chemicals. Monosodium glutamate (Sigma-Aldrich) was first dissolved in de-ionized water as the stock solution and then diluted with the cockroach external solution immediately before experiments. Fipronil (provided by Rhone-Poulenc Yuka Agro K.K., Akeno, Japan) was first dissolved in dimethyl sulfoxide to make stock solutions, and then diluted with the external solution shortly before experiments. The final concentrations of dimethyl sulfoxide in test solutions were 0.1% (v/v) or less, which had no effect on the glutamate-induced currents.

Analysis. Whole-cell currents were initially analyzed with the pClamp version 6.0.4 software (Axon Instruments, Inc., Union City, CA) to measure the current amplitudes and decay kinetics. The statistical analysis and the nonlinear regression analysis were carried out using the Sigmaplot 2001 software (SPSS Science, Chicago, IL). The dose-response relationship for glutamate to activate the glutamate receptor was evaluated by fitting the data to Hill equation: I = I_maxC₅₀^h/(C^h + C₅₀^h), where I is the current amplitude relative to the control maximum current, I_max, C is the concentration of chemical, C₅₀ is the concentration of chemical inducing a half-maximum response, and h is the Hill coefficient. The dose-response relationship for test chemicals to modulate the glutamate response was evaluated with a similar Hill equation to obtain EC₅₀ or IC₅₀ values.

Results

Differential Actions of Fipronil on Two Types of GluCls. In isolated cockroach neurons, two types of GluCl currents were recorded in response to application of 100 μM glutamate at a holding potential of -60 mV: the desensitizing current and the nondesensitizing current (Fig. 1). The desensitizing current (Fig. 1A) was characterized by a fast rising phase and a fast decay phase (observed in 18 of 145 cells), and the nondesensitizing current (Fig. 1B) was characterized by a slower rising phase and a very slow decay phase (observed in 56 of 145 cells). Most neurons exhibited a mixed-type current with a fast transient component followed by a slow component upon application of 100 μM glutamate (Fig. 1C). The kinetic properties of the mixed-type current and its differential modulation by drugs suggested that the mixed-type current represented a composite of desensitizing and nondesensitizing currents. The EC₅₀ values determined from the dose-response relationship were 37 μM for the desensitizing current (Ikeda et al., 2003) and 116 ± 4 μM (n = 6) for the nondesensitizing current.

The two types of GluCl currents responded differently to the blocking action of fipronil. Following coapplications of 1 μM fipronil with 100 μM glutamate via a U-tube, the peak amplitude of the desensitizing current was gradually decreased (Fig. 1A). The rising phase of the current was not changed, but the decay phase was slightly accelerated. As a result, the time to the peak of the current was shortened. After three consecutive coapplications at an interval of 2 min, the fipronil-induced inhibition reached a steady-state level of 57% of the control and the decay time constant was reduced from 425 ms to 297 ms.

In another neuron that generated mainly the nondesensitizing current, coapplication of 1 μM fipronil with 100 μM glutamate greatly accelerated the current decay without changing the rising phase (Fig. 1B). At the end of an 8-s application of fipronil, the current was inhibited almost completely. In the second and third applications of fipronil, the glutamate-induced currents remained small. The steady-state currents evoked by 100 μM glutamate were inhibited by 96.3 ± 3.5% (n = 7). The differential action of fipronil on the two types of glutamate-induced currents could also be seen in neurons generating the mixed-type current (Fig. 1C). Thus, a use-dependent inhibition is seen with fipronil in both the desensitizing and the nondesensitizing current, and the nondesensitizing current is much more sensitive than the fast transient current to fipronil block.

Blocking Action of Fipronil on Desensitizing GluCls. The potency and reversibility of the inhibitory action of fipronil on the desensitizing GluCl currents were analyzed in detail. As shown in Fig. 2, when 1 μM fipronil was applied via bath perfusion and U-tube, the peak amplitudes of the desensitizing currents evoked by 100 μM glutamate at an interval of 2 min were gradually inhibited to 26.2% of the control (b over a). The degree of inhibition was similar to that previously reported (Ikeda et al., 2003). After washing with fipronil-free external solutions for 10 min, only a slight recovery was seen (c versus b traces) in the absence of repetitive glutamate stimulation. However, recovery occurred rapidly when the receptors were activated by glutamate. Most of the recovery must have occurred during the first application of glutamate to activate GluCls (c) because the current (d) in response to the second application of glutamate reached near the maximum recovery level during repeated activation by glutamate. The recovery from the fipronil block was also discernible as a sustained inward current following the transient peak current in trace c. The results suggested that the fipronil molecule was trapped in the channel of the resting receptor and that unblocking of the desensitizing GluCls required the activation or opening of GluCls. This implies that the fipronil-blocked receptors do not undergo fast desensitization.

Fipronil block of the desensitizing GluCls was concentration-dependent. Desensitizing currents were evoked by a series of 30-s applications of 100 μM glutamate at an interval of 2 min. After several stable control recordings were established, fipronil at different concentrations ranging from 10 nM to 10 μM was coapplied with glutamate. The peak current amplitude was gradually decreased, reaching a steady-state level of block after four consecutive coapplications of each fipronil concentration. The concentration-response curve for steady-state block of desensitizing currents caused by fipronil is illustrated by closed circles in Fig. 3. The fitting to the data with a sigmoid curve gave an IC₅₀ of 801 ± 207 nM (n = 3–5), a value similar to that of the previous report (Ikeda et al., 2003).

Blocking Action of Fipronil on Nondesensitizing GluCls. To obtain the concentration-response relationship for fipronil block of the nondesensitizing GluCls, two protocols were used. In the first protocol, the effect of bath-applied fipronil on the nondesensitizing current was monitored at a 30-s interval by a 2-s pulse of glutamate and fipronil coapplication. The steady-state inhibition curve is plotted as open circles in Fig. 3. The fit to the data gave an IC₅₀ of 10 ± 0.9 nM (n = 3–10). Thus, fipronil is about 80 times more potent against the nondesensitizing current compared with the desensitizing current.

In the second protocol, fipronil was coapplied with 100 μM glutamate for 20 s to observe the steady-state fipronil block of the activated receptors. The sustained currents were measured at the end of 20-s coapplications of glutamate and fipronil (Fig. 4A). Although the currents shown in Fig. 4A may contain desensitizing GluCl currents, the nondesensitizing components can be obtained by subtraction: since the desensitizing current was minimally affected by 100 nM fipronil and also since the nondesensitizing current was almost completely inhibited, the nondesensitizing current can be obtained by subtracting the current after the third application of fipronil from the total current. The subtracted nondesensitizing currents are illustrated in Fig. 4B. The inhibition of the sustained current following the first, second, and third coapplications of fipronil with glutamate was calculated and is shown in Fig. 4C. The IC₅₀ values were estimated to be 27.0 ± 5.5, 17.3 ± 4.2, and 11.9 ± 3.5 nM fipronil for the first, second, and third coapplications, respectively (n = 3–9). The IC₅₀ value of 11.9 nM after the third coapplication is similar to the value obtained from combined bath and U-tube application (Fig. 3). Thus, fipronil blocked the nondesensitizing current in a use-dependent manner.

The use-dependent block of nondesensitizing GluCls by fipronil was further examined. Nondesensitizing currents were evoked by 2-s applications of 100 μM glutamate. Fipronil at a concentration of 100 nM was bath-perfused for 10 min, during which a series of repetitive coapplications of fipronil and glutamate were performed at an interval of 30 s (Fig. 5A). Currents were suppressed gradually during repeated coapplications and eventually reached very low levels at the end of 10-min fipronil perfusion. In another protocol shown in Fig. 5B, nondesensitizing currents were also evoked by 2-s applications of 100 μM glutamate, but, during bath perfusion of 100 nM fipronil, no glutamate was applied to the neuron until the end of 10-min fipronil perfusion. The results showed that the nondesensitizing steady-state currents were inhibited to 10.4 ± 1.4% (n = 7) of the control by fipronil when GluCl receptors were repeatedly activated by glutamate, whereas the steady-state currents were decreased only to 90.6 ± 2.0% (n = 6) of the control by fipronil when GluCl receptors were kept in a resting state (Fig. 5C). The activation of the receptors significantly enhances the blocking action of fipronil on nondesensitizing GluCls (p < 0.01).

Kinetic Analysis of Fipronil Block of Nondesensitizing GluCls. To estimate the rate of fipronil block of the activated glutamate receptor, fipronil at various concentrations was coapplied with 100 μM glutamate for 20 s via a U-tube perfusion system. In the absence of fipronil, the current evoked by a 20-s application of glutamate decayed slowly. The time constant for decay of the nondesensitizing current could be determined from neurons generating the mixed current 3 s after its onset, as the desensitizing current, which decayed with a time constant of 400 ms, was essentially over by then. The glutamate current decayed faster as the fipronil concentration was increased from 10 to 1000 nM (Fig. 6). The time constants thus determined decreased from a control of 29.4 s to 23.6, 15.8, 9.7, 5.1, and 2.5 s, respectively, by 10, 30, 100, 500, and 1000 nM fipronil. The reciprocal of these time constants is plotted as a function of fipronil concentration in Fig. 6. The slope of the linear relationship gave the binding rate constant of 3.04 × 10⁵ M^-1s^-1 and the unbinding rate constant of 5.77 × 10^-2 s^-1 (n = 3–7). These values are of the same order of magnitude as those for blocking the cockroach GABA-activated chloride channel (Zhao et al., 2003).

Fipronil Preferentially Blocks Nondesensitizing GluCls in the Activated State. To assess the activated receptor block, a 30-s coapplication of fipronil with glutamate was used (Fig. 7A). To assess the block when receptors are repetitively activated, brief 2-s pulses of glutamate and fipronil were coapplied at 30-s intervals while fipronil was perfused in the bath (Fig. 7B). The block of the activated receptor developed with a time constant of 18.0 ± 3.1 s (n = 4) (Fig. 7C), whereas the block during repetitive receptor activations occurred much more slowly, with a time constant of 83.1 ± 12.4 s (n = 4) (Fig. 7B). However, if the time courses of fipronil block using the two protocols were expressed in terms of the duration during which the receptors were activated, the fipronil block induced by brief repetitive 2-s coapplications of glutamate and fipronil became faster, with a time constant of 5.1 ± 1.2 s (n = 4) (Fig. 7C). Thus, repetitive openings of the glutamate receptor are more effective than a single long opening for fipronil block. This result together with that shown in Fig. 5B suggests that the activated receptors are more sensitive to the blocking action of fipronil than desensitized receptors.

Recovery of Desensitizing and Nondesensitizing GluCls from Fipronil Block. To compare the recovery of desensitizing and nondesensitizing currents from fipronil block, experiments depicted in Fig. 8 were performed. During a 30-s coapplication of 100 μM glutamate and 1 μM fipronil, the peak was reduced to about 40% of the control while the steady-state current was decreased to 5.8 ± 2.4% (n = 5) of the control (Fig. 8, A and D). After washing the neuron with fipronil-free external solution for 10 min, the peak current remained decreased in response to the application of 100 μM glutamate while the steady-state current recovered to 24.9 ± 4.9% (n = 5) of the control (Fig. 8, B and D). A second application of glutamate 30 s later revealed that the peak current completely returned to the control level while the steady-state current increased slightly (Fig. 8B, washout-2). The rapid recovery of the peak current during the second test pulse after washout was similar to that described in Fig. 2B,d.

The recovery of the nondesensitizing current from the block caused by 1 μM fipronil was incomplete (Fig. 8B). However, since the nondesensitizing current was 80 times more sensitive to fipronil block than the desensitizing current (Fig. 3), it was necessary to compare the recovery after washing using the equivalent concentrations of fipronil that cause the same degree of block. Thus, additional experiments were performed for the recovery of the nondesensitizing current using 10 nM fipronil (Fig. 8C). The steady-state current was decreased to 71.6 ± 5.2% (n = 5) of the control during coapplication of 100 μM glutamate and 10 nM fipronil. After washing the neuron with fipronil-free external solutions for 10 min, during which no glutamate was applied, the steady-state currents recovered to 93.0 ± 3.1% (n = 5) of the control (Fig. 8D). It was concluded that nondesensitizing GluCls, when blocked by a moderate concentration of fipronil, could fully recover in the absence of receptor activation.

Interaction of Fipronil and Picrotoxinin in Blocking the Nondesensitizing Current. Picrotoxinin and fipronil block rat GABA_A receptors by binding to different receptor sites (Ikeda et al., 2001). Although picrotoxinin had little or no effect on the desensitizing GluCls, it blocked the nondesensitizing GluCls potently, with an IC₅₀ on the order of 4 μM. Thus, a question arose whether picrotoxinin and fipronil interacted with each other in blocking the nondesensitizing GluCls. The time-dependent block of the nondesensitizing current by 100 μM picrotoxinin coapplied with 100 μM glutamate is illustrated in Fig. 9A. The block of nondesensitizing current by picrotoxinin was nearly complete at the end of the first 5-s glutamate/picrotoxinin coapplication (Fig. 9A,b), as evidenced by the complete block by the second coapplication of picrotoxinin and glutamate (Fig. 9A,c). The recovery from picrotoxinin block is illustrated in Fig. 9B,a and 9B,b. After washing the neuron with picrotoxinin-free external solution for 2 min, the first glutamate pulse induced a current with a slow rising phase (Fig. 9B,a). The recovery was essentially complete at the end of a single 5-s glutamate pulse, as indicated by the near normal current evoked by the second application of glutamate applied 1 min later (Fig. 9B,b). These results suggest that the picrotoxinin molecule is trapped in the closed channel when the receptor is at rest and that the picrotoxinin molecule egresses the channel only when it is open. Thus, picrotoxinin is an open channel blocker.

If fipronil and picrotoxinin were mutually exclusive in blocking the nondesensitizing GluCls, one would expect that picrotoxinin would alter the time course and degree of fipronil block, because fipronil could not affect the picrotoxinin-blocked channel until the channels were free of picrotoxinin. This was indeed the case as shown in Fig. 9C. After picrotoxinin block had been established, the neuron was washed with picrotoxinin-free external solution for 2 min and subsequently exposed to 1 μM fipronil coapplied with glutamate. The glutamate-induced current had a slow rising phase, reaching a peak near 3 s and then began to show slow decay. These changes in current kinetics are consistent with the hypothesis that fipronil cannot affect the channel until it is free of picrotoxinin. In the absence of picrotoxinin, fipronil blocked the nondesensitizing GluCl in a time-dependent manner without affecting the rising phase (Fig. 9D). As a consequence of their mutual exclusion, the fipronil block measured at the 5-s time point was less in the channel previously blocked by picrotoxinin than in the control channel. On average, the fipronil blocks were 59.6 ± 2.2% (n = 10) and 80.0 ± 2.7% (n = 6), respectively, for the picrotoxinin-blocked channel and the control channel (p < 0.01) (Fig. 9E).

Discussion

The present study has demonstrated that the desensitizing and nondesensitizing currents produced by glutamate in acutely dissociated cockroach neurons differ not only in their activation and decay phase but also in their sensitivity to picrotoxin and fipronil. Fipronil was found to block the desensitizing current with an IC₅₀ of 801 nM, whereas it more potently blocked the nondesensitizing current, with an IC₅₀ of 10 nM.

Two components of glutamate-gated chloride currents have previously been reported in cockroach dorsal unpaired median neurons (Raymond et al., 2000). These two types of currents exhibited different pharmacological profiles in terms of ibotenate and picrotoxin sensitivity. One component was sensitive to picrotoxin and fipronil, but insensitive to ibotenate, whereas the other was insensitive to 100 μM picrotoxin, but sensitive to ibotenate. We identified two ibotenate-sensitive GluCls in cockroach neurons, only one of which was sensitive to picrotoxin (unpublished observation). The picrotoxin-insensitive current undergoes rapid desensitization while the picrotoxin-sensitive current is nondesensitizing. In the present study, we have further characterized the mechanism of the blocking action of fipronil on both desensitizing and nondesensitizing currents.

Fipronil blocked the desensitizing current in a use-dependent manner. One of the requirements for use-dependent block is a state-dependent change in affinity for the blocker or its access to the binding site. When fipronil was coapplied with glutamate, the rising phase of the current was not altered, but the decaying phase was accelerated, indicative of an open channel blocking action. Once fipronil block had been established, recovery from block did not occur until the channel was opened by activation of the receptor. This result suggests that the fipronil molecule is trapped in the closed channel. Both open channel block and blocker trapping phenomena are necessary and sufficient to account for the use-dependent block.

The unblocking rate constant of fipronil for the desensitizing current can be estimated from the slow time constant of recovery from fipronil block (Fig. 2A,c). The slow time constant of 5.7 s corresponds to an unblocking rate constant of 1.7 × 10^-1 s^-1, which, in combination with a K_d of 801 nM, would lead to the blocking rate constant of 2.1 × 10⁵ M^-1s^-1. This value for fipronil block of the desensitizing current is similar to that for fipronil block of the nondesensitizing current (3 × 10⁵ M^-1s^-1).

Fipronil also blocked the nondesensitizing current in a time- and use-dependent manner. Although having no effect on the rising phase of the current, fipronil induced a prominent decay and caused a profound reduction in the glutamate-induced current at the end of a 30-s pulse. The current remained small upon subsequent coapplication of fipronil with glutamate, since the blocked channels did not recover during a 2-min interval between fipronil applications.

Although activation profoundly enhanced fipronil block of nondesensitizing GluCls with little recovery during a 30-s interpulse interval (Fig. 5), there was substantial recovery from block in the absence of channel opening after 10 min of washing (Fig. 8). This observation suggests that the binding site of fipronil in nondesensitizing receptors is not as strongly protected by the activation gate as it is in the desensitizing receptors. Fipronil block of activated nondesensitizing GluCls can be estimated from the time-dependent block. The blocking rate constant estimated from the concentration dependence of time constants of fipronil block was 3.04 × 10⁵ M^-1s^-1 and the unblocking rate constant was 5.77 × 10^-2s^-1 (Fig. 6). After washing out fipronil, the unblocking rate constant would predict a substantial time-dependent increase in the current during a 30-s activation of the receptor by glutamate, but no such time-dependent recovery was seen. Thus, the fipronil molecule escapes from the blocked receptor at a rate much slower than predicted from the unblocking rate constant (Fig. 6B). This result suggests that there exists more than one blocked state, and that Y-intercept of the concentration dependence of the on-rate is not simply reflecting the unblocking rate. That is, the affinity of the second blocked state of the channel for fipronil is much higher than that of the first blocked state of the channel. This would also explain the discrepancy between the IC₅₀ of 10 nM determined at the steady-state block and the K_d value of 190 nM calculated from the blocking and unblocking rate constants for block of the open channel. However, if one were to use the apparent K_d of 10 nM and the blocking rate constant of 3 × 10⁵ M^-1s^-1 to calculate the unblocking rate constant, the unblocking rate constant would become 3 × 10^-3 s^-1, which is small enough to explain the negligible increase in the current during a 30-s application of glutamate to activate the receptor after washout of fipronil. In contrast, there was a sizable recovery of the desensitizing current after the application of glutamate (Fig. 2).

It is interesting to note that the nondesensitizing current is more sensitive than the desensitizing current to the blocking action of fipronil and picrotoxinin. This situation is opposite to what has been reported for GABA_A receptors where the desensitized state favors picrotoxin block (Newland and Cull-Candy, 1992; Pribilla et al., 1992; Zhang et al., 1994). An allosteric mechanism was proposed in which picrotoxin binds to the channel lumen inducing a rapid conformational change to the desensitized state (Zhang et al., 1994). According to this allosteric mechanism, the lower sensitivity of the desensitizing GluCls to fipronil block can be explained by two mechanisms. First, the channel opens transiently and second, the blocked channel appears not to undergo desensitization (Fig. 2A,c).

The open channel block by fipronil is also inferred from the experiment in which the interaction of fipronil with picrotoxinin in blocking the nondesensitizing current was analyzed. Picrotoxinin blocked and unblocked the nondesensitizing current when the receptors were activated. The analysis of the kinetics and steady-state block of the picrotoxinin-blocked channel by fipronil suggests that once the channel was blocked by picrotoxinin, fipronil could not bind to the channel until the channel was free of picrotoxinin (Fig. 9C). This interaction could arise from direct competition for the same binding site or from allosteric interactions.

This is the first study in which the potency of fipronil against the nondesensitizing GluCl has been measured and analyzed. The IC₅₀ of 10 nM indicates that this target is at least as sensitive as the GABA receptors of the same species (Zhao et al., 2003). In that study, an IC₅₀ of 28 nM for fipronil block of GABA receptors was measured but might be considered an underestimate of its potency because, due to the extremely slow action of fipronil on GABA receptors, the measurements were not made at equilibrium. Thus, GABA receptors and GluCls represent two high-affinity target sites for fipronil in insects.

The existence of two target sites for fipronil could be an important factor limiting the development of resistance. Dieldrin blocks only the GABA receptors in insects (Ikeda et al., 2003; Zhao et al., 2003), and is subject to very high levels of target-site resistance, conferred by a point mutation of Ala to Ser in transmembrane domain II of the Rdl ligand-gated chloride channel subunit (ffrench-Constant et al., 1993). Native GABA receptors from insects containing this mutation are highly resistant to dieldrin and picrotoxinin (Zhang et al., 1994; Cole et al., 1995), as are heterologously expressed Rdl homomultimers, which are activated by GABA (Buckingham et al., 1996). On the other hand, Rdl is also a component of GluCls in insects (Ludmerer et al., 2002), and although this may be unimportant in terms of dieldrin resistance, it could play a role in the development of resistance of fipronil, which also potently blocks GluCls. Heterologously expressed Rdl homomultimers containing the resistance mutation may display reduced sensitivity to fipronil in some cases (Wolff and Wingate, 1998) that, although much less than the resistance to dieldrin, could lead to moderate levels of resistance to fipronil in some species, which is in fact seen (Cole et al., 1995). Studies of the action of fipronil against native GABA and nondesensitizing GluCls in insects with the Rdl resistance mechanism are needed to fully understand the moderate resistance to fipronil.

Fipronil block of the GluCls of cockroach neurons is of paramount importance from the toxicological point of view. Because GluCls are not found in mammals (Cleland, 1996; Raymond and Sattelle, 2002), these receptors are an excellent target for developing insecticides with a high selective toxicity for insects over mammals.