Benzodiazepines act mainly at postsynaptic γ-aminobutyric acid type A (GABAA) receptors. In rat neocortical layer V pyramidal neurons, we found that midazolam (MDZ), a benzodiazepine, increases the frequency of GABAergic miniature inhibitory postsynaptic currents (mIPSCs) via insertion of α7 nicotinic acetylcholine receptors (nAChRs) at presynaptic GABAergic boutons. Although nicotine alone had no effect, MDZ plus nicotine dramatically increased mIPSC frequency. Neostigmine, an acetylcholinesterase inhibitor, mimicked the actions of nicotine. MDZ increased the number of α-bungarotoxin–bound boutons that were blocked by protein kinase C (PKC) inhibitors, as revealed by confocal imaging of a neuron-synaptic bouton preparation. Thus, MDZ may induce membrane translocation of α7 nAChRs on GABAergic boutons via activation of PKC, enabling endogenous acetylcholine to increase GABA release. The above actions seem unique to MDZ because neither other benzodiazepines (diazepam and flunitrazepam) nor zolpidem had this effect. The findings reveal both a novel cholinergic modulatory mechanism affecting GABAergic transmission and a novel action of some general anesthetics.
Benzodiazepines, like other anesthetics, act via the postsynaptic γ-aminobutyric acid type A (GABAA) receptor to potentiate the action of the major inhibitory neurotransmitter GABA. Midazolam (MDZ), a benzodiazepine, is used clinically as a general anesthetic and sedative. By comparison with other benzodiazepines, however, MDZ also causes episodes of relative awakening and has other unique actions indicating that an unknown mechanism might be involved in mediating some of its actions. An involvement of nicotinic acetylcholine receptor (nAChR) has been implicated in this unknown mechanism (Jastak 1985; van der Bijl and Roelofse 1991). Indeed, administration of the cholinesterase inhibitor physostigmine increases arousal and improves delirious states after MDZ-induced anesthesia (Caldwell and Gross 1982). Because nAChRs exist on GABAergic interneurons within the neocortex (Xiang and others 1998; Alkondon and others 2000), we may need to invoke not only postsynaptic GABAA receptors but also nAChR on presynaptic GABAergic neurons to account for all the actions of MDZ. However, when we began this study no such presynaptic actions of MDZ were known. Here, we demonstrate 1) the existence of a presynaptic action of MDZ that leads to a facilitation of GABA release onto layer V pyramidal neurons in the rat neocortex and 2) that this presynaptic action occurs via a PKC-mediated membrane translocation of α7 nAChR on the synaptic boutons of GABAergic interneurons.
Materials and Methods
Brain Slice Preparation and Patch-Clamp Recordings
The experimental procedures employed in this study were in compliance with the guidelines for animal research issued by Hamamatsu University School of Medicine. Neocortical slices were obtained from 2- to 3-week old Wistar rats. Each rat was deeply anesthetized with halothane, decapitated, and its brain rapidly removed. Coronal slices of 350-μm thickness were cut on a vibratome (VT1000S, Leica, Bensheim, Germany) at 4 °C in modified artificial cerebrospinal fluid (ACSF) (in mM): 220 sucrose, 30 glucose, 2.5 KCl, 1.25 NaH2PO4, 10 MgSO4, 0.5 CaCl2, and 26 NaHCO3. Individual slices were transferred to a recording chamber, perfused (2.5 mL/min) with pregassed (95% O2–5% CO2) standard ACSF (in mM): 126 NaCl, 2.5 KCl, 2 CaCl2, 2 MgSO4, 1.25 NaH2PO4, 26 NaHCO3, 20 glucose, and pH 7.4, and maintained at 30 °C. For miniature inhibitory postsynaptic currents (mIPSCs) recordings, blockers of excitatory amino acid receptors (D(-)-2-amino-5-phosphonopentanoic acid [D-AP5, 50 μM], 6-cyano-7-nitroquinoxaline-2,3-dione [CNQX, 10 μM]) as well as the GABAB receptor blocker CGP55845 (3 μM) and tetrodotoxin (TTX, 1 μM) were added to the ACSF. Patch electrodes had resistances of 4–5 MΩ when filled with a solution containing (in mM) 130 CsCl, 1 CaCl2, 2 MgCl2, 0.5 ethyleneglycol-bis(2-aminoethylether)-N,N,N′,N′-tetra acetic acid, 10 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 3 Mg(adenosine triphosphate)2, 0.4 guanosine triphosphate, and 5 QX314; pH 7.3. For miniature excitatory postsynaptic current (mEPSC) recordings, CGP55845 (3 μM), bicuculline methiodide (BMI, 10 μM), and TTX (1 μM) were added to the ACSF. Whole-cell patch-clamp recordings were made from visually identified pyramidal neurons and interneurons of the somatosensory cortex. With cells voltage clamped at −60 mV throughout this report, the GABAergic IPSC appeared as an inward current. The recorded currents were filtered at 2 kHz and digitized at 1–10 kHz, and data were analyzed using DigiData1322A and pCLAMP8 software (Axon Instruments, Sunnyvale, CA). mIPSCs and mEPSCs were examined by constructing cumulative probability distributions for 1-min epochs, immediately before (control) and from 4–5 min of drug application, and compared using the Kolmogorov–Smirnoff test (K–S test). Data are given as mean ± standard error of the mean, with each value being normalized with respect to the control. Possible differences in amplitude and frequency distribution were tested using a Student's 2-tailed paired t-test or the Steel–Dwass test for multiple comparisons. Absolute values, rather than normalized ones, were used for these analyses. Values of P < 0.05 were considered significant.
Mechanical Dissociation of Neurons
In brief (Akaike and Moorhouse 2003), the slices were transferred to a 35-mm glass-bottomed dish filled with a standard external solution (in mM): 150 NaCl, 5 KCl, 1 MgCl2, 2 CaCl2, 10 HEPES, and 10 glucose (pH 7.4). A fire-polished glass pipette was placed lightly on the surface of cortical layer V and vibrated horizontally at 60 Hz for about 2 min using a custom-made vibration device. The dissociated neurons were allowed to settle and adhere to the bottom of the dish for about 10 min. Pyramidal neurons were visually identified by the somatic size and the cell shape.
Fluorescence Labeling of Surface α7 nAChRs on Synaptic Boutons
Mechanically dissociated pyramidal neurons were incubated at room temperature for 15 min with 500 nM Alexa 488 α-bungarotoxin (Alx α-Bgt) (Molecular Probes, Carlsbad, CA) in a standard external solution containing (in μM) 50 D-AP5, 10 CNQX, 3 CGP55845, and 1 TTX. After washing, neurons were then incubated with a solution containing 1 μM MDZ and 500 nM Alx α-Bgt for 15 min. FM1-43 (Molecular Probes) was used at a concentration of 10 μM for loading into synaptic boutons during a 60-s perfusion with 15 mM high-K+ external solution.
Imaging by Confocal Microscopy
Confocal images were obtained using a confocal laser microscope (IX 70, Olympus, Tokyo, Japan) equipped with a microlens-attached Nipkow-disk scanner and real time 3-dimensional system (CSU-21, Yokogawa, Tokyo, Japan). An argon laser was used for excitation (488 nm) and a long-pass filter (>520 nm) for emission. Confocal fluorescence images were recorded using a 3CCD color camera in combination with an image intensifier (C7780/C2400-21, Hamamatsu Photonics, Shizuoka, Japan) and analyzed using IPLab software (Scanlytics). Multiple focal planes at intervals of 1 μm were superimposed.
MDZ but Not Other Benzodiazepines Increased the Frequency of mIPSCs
To elucidate the effect of MDZ on GABAergic synaptic transmission in the somatosensory cortex, we examined the effects of MDZ on GABAA receptor–mediated IPSCs in layer V pyramidal neurons in brain slices. MDZ increased both the frequency and the amplitude of spontaneous IPSCs (which were sensitive to BMI, a GABAA receptor antagonist). When the sodium channel blocker TTX was added, to enable us to record mIPSCs, MDZ significantly increased both the amplitude and frequency of mIPSCs (Fig. 1). The values (% control) obtained for mIPSC amplitude during MDZ application were 100.0 ± 2.8 (40 nM, n = 4), 113.7 ± 8.4 (100 nM, n = 5, P < 0.01), and 122.6 ± 12.3 (1 μM, n = 13, P < 0.001, Fig. 1C). The values (% control) obtained for mIPSC frequency in the same epoch were 96.6 ± 11.8 (40 nM), 128.9 ± 6.0 (100 nM, P < 0.001), and 153.7 ± 11.0 (1 μM, P < 0.001, Fig. 1D). However, other benzodiazepines (viz. diazepam [DZP] and flunitrazepam [FNP]) and zolpidem (ZPD) all failed to increase the frequency of mIPSCs even at relatively high concentrations (0.5–1 μM). The values (% control) obtained for mIPSC frequency following DZP, FNP, and ZPD applications were 103.5 ± 4.4 (n = 6), 107.5 ± 5.0 (n = 5), and 102.4 ± 8.5 (n = 5), respectively (Fig. 1D), whereas the amplitude values (% control) were 133.4 ± 12.5 (P < 0.001) with DZP, 115.8 ± 2.7 (P < 0.01) with FNP, and 108.3 ± 2.9 (P < 0.05) with ZPD (Fig. 1C). Thus, 0.5 μM DZP and 1 μM FNP caused increases in the amplitude of mIPSCs equivalent to that induced by 1 μM MDZ (P > 0.05, Fig. 1C) but failed to increase mIPSC frequency, suggesting that the MDZ-induced increase in mIPSC frequency was not simply due to “buried” mIPSCs, becoming detectable due to an increase in their amplitude.
Presynaptic Action of MDZ Was Mediated by α7 nAChRs
We studied a possible mediation by presynaptic benzodiazepine receptors by using a specific benzodiazepine receptor antagonist, flumazenil (1 μM) (Fig. 2). Flumazenil failed to block the MDZ-induced increases in mIPSC frequency (120.0 ± 7.4, n = 6, P < 0.05, Fig. 2B, right), while blocking increases in the amplitude (93.5 ± 5.5, P > 0.05, Fig. 2B, left). The results indicate that postsynaptic but not presynaptic benzodiazepine receptors were mediating the MDZ actions.
Next, we studied a possible mediation by presynaptic nAChRs. The MDZ-specific effect on mIPSC frequency was blocked by the α7 nAChR blocker, methyllycaconitine (MLA, 0.1 μM, n = 5) (Fig. 3) but not by either the α4β2 nAChR blocker, dihydro-β-erythroidine (50 μM, n = 5), or the muscarinic AChR blocker, atropine (1 μM, n = 4) (not shown). Thus, MDZ-induced increases in mIPSC frequency were mediated by α7 nAChR. On the other hand, MLA did not affect the MDZ-induced increases in mIPSC amplitude (119.9 ± 4.7%, n = 5, P < 0.05).
Interestingly, when applied alone, nicotine failed to increase the mIPSC frequency (Fig. 4A,B). However, nicotine application during MDZ treatment led to a dramatic augmentation of the frequency of mIPSCs (Fig. 4C,D). The values (% control) obtained for mIPSC frequency were 145.9 ± 15.7 (MDZ) and 275.2 ± 53.7 (MDZ + nicotine) (n = 9). On the other hand, nicotine did not significantly affect mIPSC amplitude with (119.4 ± 11.0, n = 9, P > 0.05) or without (108.7 ± 9.4%, n = 6, P > 0.05) MDZ.
Insertion of α7 nAChRs Was Induced by MDZ on Synaptic Boutons
We therefore inferred that GABA release onto postsynaptic layer V pyramidal cells is facilitated by MDZ through insertion of silent α7 nAChRs on GABAergic boutons, thus providing receptors for activation by endogenous acetylcholine (ACh). As expected, the exposure of slices to an acetylcholinesterase inhibitor, neostigmine (1 μM), alone failed to increase mIPSC frequency (98.4 ± 6.1% control, n = 5, P > 0.05). However, neostigmine administration during MDZ treatment led to a significant enhancement of the MDZ effect on mIPSC frequency in 6 of 8 neurons tested (Fig. 5). The values (% control) obtained for mIPSC frequency were 146.2 ± 12.4 (MDZ) and 161.2 ± 16.4 (MDZ + neostigmine) (n = 6, P < 0.05).
Further to test the above hypothesis that MDZ induces translocation of α7 nAChRs over the membrane surface in presynaptic GABAergic terminals, we utilized acute mechanical dissociation of layer V pyramidal cells, a technique that is supposed to preserve synaptic boutons (Akaike and Moorhouse 2003). The binding of the fluorescence probe Alx α-Bgt was significantly increased in the presence of MDZ (% control: 202.2 ± 30.0 when analyzed by cluster number per cell, P < 0.05; 313.2 ± 100.0 when analyzed by fluorescent area [pixels] per cell, P < 0.01; n = 6), suggesting that membrane translocation of nAChRs was indeed induced by MDZ (Fig. 6). Further, staining of synaptic boutons by depolarization-induced loading of FM1-43 revealed that 53.9 ± 5.1% (n = 3) of FM1-43–labeled boutons possessed Alx α-Bgt–labeled α7 nAChRs (Fig. 6Ad).
Membrane Translocation of α7 nAChRs Was Mediated via Activation of PKC
In cloned mammalian cells, PKC stimulation increases translocation of nAChRs to the cell surface (Nashimi and others 2003). Indeed, application of the PKC activator, phorbol 12,13-dibutyrate, at 10 μM significantly increased the mIPSC frequency to 169.6 ± 18.7% of control (n = 5, P < 0.05; not shown). Therefore, to elucidate whether PKC is involved in the action of MDZ, we examined the effect of PKC inhibitors. Bath application of the cell-permeable, specific PKC inhibitor calphostin C (Calp, 1 μM) for 20 min did not change the mIPSC frequency but blocked the increase induced by MDZ (Fig. 7A,B). The values (% control) obtained for mIPSC frequency were 93.4 ± 5.9 (Calp) and 92.7 ± 2.4 (Calp + MDZ) (n = 5, P > 0.05). In contrast, Calp did not affect the incremental effect of MDZ on mIPSC amplitude, the values (% control) for which were 101.6 ± 5.2 (Calp) and 130.9 ± 6.2 (Calp + MDZ) (P < 0.05). Another PKC inhibitor, bisindolylmaleimide (BiM1, 5 μM), gave similar results, the values (% control) obtained for mIPSC frequency being 97.6 ± 11.6 (BiM1) and 100.9 ± 14.9 (BiM1 + MDZ) (n = 6, P > 0.05).
Incubation of isolated cells with Calp (100 nM) for 20 min also inhibited the MDZ-induced incremental effect on Alx α-Bgt binding (% control: 131.7 ± 23.2 when analyzed by cluster number per cell, P > 0.05; 138.9 ± 27.7 when analyzed by fluorescent area [pixels] per cell, P > 0.05; n = 6; Fig. 7C,D).
MDZ Action on Presynaptic α7 nAChRs Was Unique to GABAergic Terminals Innervating Layer V Pyramidal Cells
In contrast to their effects on mIPSCs, neither MDZ alone nor MDZ + nicotine altered mEPSC frequency, either in layer V (Fig. 8A,B) or in layer II/III (data not shown). The values (% control) obtained for mEPSC frequency were 93.4 ± 2.7 (MDZ) and 93.6 ± 5 (MDZ + nicotine) (n = 5, P > 0.05). Although α7 nAChRs may exist on the glutamatergic terminals on cortical pyramidal neurons (Levy and Aoki 2002), electrophysiological effects of nicotine on mEPSC frequency are observed only briefly during development (P8–16) (Aramakis and Metherate 1998) and are exclusively on thalamocortical terminals (Gil and others 1997). The lack of nicotinic modulation of mEPSC frequency in our study may thus be attributable to our choice of material (i.e., rat 2–3 weeks old and layers V/II), and hence the lack of observed MDZ effects on mEPSCs may be due to a lack of presynaptic α7 nAChRs on glutamatergic terminals in such material.
We also examined the effects of MDZ and nicotine on mIPSCs recorded in layers II/III (n = 6), IV (n = 6), and VI (n = 5) pyramidal neurons. Neither MDZ alone nor MDZ + nicotine altered mIPSC frequency, although MDZ alone induced a significant increase in amplitude. For instance in layer II/III, the values (% control) obtained for mIPSC amplitude (MDZ) were 124.6 ± 9.0 (P < 0.05) and those for frequency were 114.0 ± 8.4 (MDZ) and 108.4 ± 14.6 (MDZ + nicotine) (P > 0.05, Fig. 8C,D). Thus, presynaptic α7 nAChRs may exist exclusively on GABAergic terminals innervating layer V pyramidal cells; otherwise, they are exclusively sensitive to such a unique action of MDZ.
The majority of α7 nAChRs in the neocortex may exist on GABAergic neurons as postsynaptic receptors (Kawai and Berg 2001). Indeed, we could record nicotine-evoked inward currents from visually identified interneurons in layers IV and V (−11.1 ± 2.7 pA, n = 4). However, these currents were not increased by MDZ (−11.4 ± 2.9 pA, P > 0.05; Fig. 8E,F), indicating that MDZ has no effects on postsynaptic α7 nAChRs already on the surface and/or that no intracellular pool of α7 nAChRs is available in the somatic region of interneurons.
Some of the Unique MDZ Actions May Be Attributable to This Presynaptic Action
MDZ is thought to act mainly via the postsynaptic GABAA receptor to potentiate the actions of GABA. The increases in mIPSC amplitude induced by MDZ, and other benzodiazepines, confirmed this idea (Fig. 1). In addition, we have now demonstrated that MDZ acts not only on the postsynaptic GABAA receptor but also by enhancing presynaptic GABA release. Clinically, MDZ is found not only to provide sedation and anxiety relief but also to have some unique effects of its own (such as memory impairment, episodes of relative awakening, and agitation) that are enhanced by physostigmine (Caldwell and Gross 1982; Jastak 1985; van der Bijl and Roelofse 1991). This raised the possibility that MDZ interacts with ACh receptors. Here, we provide direct evidence that the nAChR on GABAergic synaptic boutons do indeed interact with MDZ. The finding that of the benzodiazepines tested only MDZ had this action supports the idea that some of the unique effects of MDZ may be attributable to this presynaptic action.
The actions of MDZ within the neocortex would seem to be restricted to GABAergic synapses because it had no effects on mEPSCs. Interestingly, this action of MDZ on GABAergic neurons appears to be specific to layer V (it was not found in the other layers). Although this could be due to a lack of α7 nAChRs in the other layers, MDZ might not induce membrane translocation of α7 nAChRs in GABAergic terminals in other than layer V. The postsynaptic α7 nAChRs on GABAergic “interneurons” were also not affected by MDZ. Thus, α7 nAChRs from different sources may be subject to different types of regulation, as reported previously (Kawai and Berg 2001). In any case, these topographical characteristics of the actions of MDZ, which have also been reported in the hippocampus (Poncer and others 1996; Rovira and Ben-Ari 1999; Bai and others 2001), may contribute to the observed clinical effects specific to MDZ.
Layer V pyramidal neurons can be functionally subdivided into the regular spiking and the bursting neurons (Franceschetti and others 1998). However, in the material we used, bursting neurons were not observed possibly due to the younger age (Franceschetti and others 1998). In any case, MDZ increased mIPSC frequency in both nonadapting (6 out of 7) and adapting (3 out of 4) regular-spiking neurons (not shown). Our results suggest that mainly somatic region of these pyramidal neurons were innervated by GABAergic terminals bearing α7 nAChRs (see Figs 6Ab and 7Cb). Because the basket cell is known to innervate preferentially the somatic region of pyramidal neurons in layer V (Freund 2003; Karube and others 2004), the basket cells may be the cellular targets of this unique MDZ action on α7 nAChRs.
Physiological Implications for the Presynaptic Modulation by α7 nAChRs on GABAergic Terminal
In the sleep–waking cycle, the ACh release from cholinergic basalocortical neurons onto postsynaptic pyramidal cells (Cape and others 2000) is greater during waking and paradoxical sleep than during slow-wave sleep (Celesia and Jasper 1966; Jasper and Tessier 1971). Atropine-resistant θ activity in electroencephalography can be stimulated (Cape and others 2000) through nicotinic receptors located on GABAergic interneurons (Xiang and others 1998; Porter and others 1999). Thus, the possibility exists that insertion of α7 receptors at GABAergic synapses in layer V may modulate both cortical activity and the sleep–wake state as a previously unknown unique mechanism.
Most cholinergic fibers identified in the cortex originate from the basal forebrain nuclei (Mesulam and others 1983) and also from cholinergic neurons within the cortex (Parnavelas and others 1986), with layer V receiving the most abundant cholinergic input (Eckenstein and others 1988) and those to pyramidal cells being mainly presynaptic (Turrini and others 2001). Functional cholinergic receptors on GABAergic interneurons in layer V have also been reported (McCormick and Prince 1986; Nicoll and others 1996; Xiang and others 1998), and ACh application does not increase mIPSC frequency in layer V (Xiang and others 1998), a result compatible with ours (see Fig. 4A,B). Previous studies suggested that nAChRs are present on GABAergic synapses onto neocortical GABAergic interneurons (Xiang and others 1998; Alkondon and others 2000). However, the presence of nAChR on GABAergic synapses onto pyramidal neurons has not previously been reported. Because neither postsynaptic currents nor increases in mEPSCs frequency were observed upon application of nicotine to pyramidal neurons (Fig. 8A,B), most preexisting surface α7 nAChR clusters may be on presynaptic GABAergic boutons. Despite this observation, nicotine alone failed to increase mIPSC frequency. The size of each α7 nAChR cluster was enlarged by MDZ because the MDZ-induced increase in Alx α-Bgt binding was significantly greater when analyzed by fluorescent area than when analyzed by cluster number (P < 0.01, Wilcoxon paired-sample test). Thus, preexisting surface α7 nAChRs on presynaptic GABAergic boutons may be incompletely inserted and not functional enough fully to conduct nicotinic currents with a high conductance for Ca2+ (Broide and Leslie 1999; McGehee 1999). Indeed, the existence of uninserted α7 nAChRs has been predicted to explain a discrepancy between the density of α-Bgt–binding sites and the magnitude of α7 nAChRs-generated currents (Dineley and Patrick 2000; Fabian-Fine and others 2001).
Mechanistic Implications for the Membrane Translocation of α7 nAChR on GABAergic Terminals
Our results clearly reveal that the α7 nAChRs on GABAergic terminals innervating layer V pyramidal neurons are normally silent and that an insertion of such receptors is induced by MDZ (see Figs 4 and 6). The finding that the effect occurs in boutons on dissociated cells suggests a rapid local insertion, rather than an effect on the somatodendritic compartment, with transport to terminals. It was reported recently that PKCγ activation translocates assembled α4β2 nAChR from the cytoplasm to the cell surface, so that the peak amplitude of ACh currents is increased in transfected HEK293T cells (Nashimi and others 2003). Although this was a rather slower translocation, the authors speculated that phosphorylation at key residues may inhibit recognition of the endoplasmic reticulum retention sequence in α4β2 nAChR. Dineley and Patrick (2000) identified key amino acid residues within α7 nAChR subunits that determine surface expression by inducing receptor redistribution between surface and intracellular pools. These surface-expression determinants are suggested to act through a receptor-transport mechanism. Therefore, it is conceivable that PKC may phosphorylate these amino acid residues in α7 nAChR. Although the mechanism responsible for the “rapid” α7 nAChR translocation observed here remains unclear, our results suggest that PKC activation is involved.
Another issue is the identity of the factor mediating the PKC activation induced by MDZ. The involvement of an action downstream of GABAA receptor activation seems unlikely because benzodiazepines other than MDZ (viz. DZP and FNP) and ZPD had no effects. Supporting this idea, a specific benzodiazepine receptor antagonist, flumazenil (1 μM), failed to block the MDZ-induced increases in mIPSC frequency, while blocking increases in the amplitude (Fig. 2). The α1 adrenoceptor (Waugh and others 1999), κ opioid receptor (Cox and Collins 2001), and A2A adenosine receptor (Seubert and others 2000) could be activated by MDZ. Of these, the α adrenoceptor (Bennett and others 1998) and κ opioid receptor (Svingos and Colago 2002) do indeed exist on the presynaptic terminals of cortical GABAergic neurons, and activation of the former could stimulate PKC. Although the mechanism needs further elucidation, our results are novel in showing that α7 nAChR insertion is induced in the synaptic boutons of GABAergic neurons by at least one general anesthetic, allowing the endogenous ACh present in the milieu to increase the release of GABA.
In conclusion, this nAChR-mediated presynaptic modulation of GABA release may be an important mechanism by which the cholinergic system and MDZ (or some as yet unknown endogenous factor) modulate the cerebral cortical circuitry. Further studies will be needed both to elucidate exactly how MDZ might induce translocation of nAChRs and to establish the identity of any endogenous factors that might be involved.
We thank Drs T. Narahashi and D. A. Prince for critical readings of this paper, Dr S. Terakawa for encouragement during the confocal imaging experiment, and Dr R. Timms for language editing. This work was supported by grants from the Japan Society for the Promotion of Science, the COE program, and the Core Research for the Evolution Science and Technology from the Japan Science and Technology Agency to AF and by a JSPS Research Fellowship (SY). Conflict of Interest: None declared.
Funding to pay the Open Access publication charges for this article was provided by Hamamatsu University.