Dopamine modulation of Ca²⁺ dependent Cl^- current regulates ciliary beat frequency controlling locomotion in Tritonia diomedea

Cilia are widely distributed in living organisms. In invertebrates, cilia function ranges from sensory structures to locomotor effectors. In mammals,ciliated epithelia line airways, brain ventricles, renal ducts and reproductive tracts. Yet there is no consensus as to how ciliary beating is controlled. A growing body of evidence suggests two possible mechanisms for cilia control. The first involves the remote release of hormones and peptides that travel through the circulatory system to target ciliated epithelia. These hormones are thought to act through second messenger systems to cause a change in ciliary beat frequency (CBF), often involving increases in intracellular free Ca²⁺ levels. The hormone prostaglandin E₂ excites CBF in rabbit oviduct through increasing internal free Ca²⁺(Verdugo, 1980). Extracellular ATP has been shown to be excitatory in esophageal cells from multiple species,including frog (Tarasiuk et al.,1995; Bar-Shimon et al.,1997), pig (Gertsberg et al.,2004), sheep (Salathe and Bookman, 1995; Lieb et al.,2002), rabbit (Korngreen et al., 1998) and human (Lieb et al., 2002). The mechanism of ATP excitation appears varied, but in at least frog esophageal cells, the excitation may be due to modulation of the membrane potential (Tarasiuk et al.,1995; Bar-Shimon et al.,1997). Finally, acetylcholine (ACh) is known to excite CBF via muscarinic receptors in frog esophagus(Zagoory et al., 2001; Zagoory et al., 2002) and serotonin (5-HT) excites CBF in rat ependymal cells(Nguyen et al., 2001).

A second possible ciliary control mechanism is by direct central nervous system (CNS) innervation of ciliated epithelia. The CNS could directly coordinate and control cilia through localized release of neurotransmitters and peptides across an entire epithelium. Nervous control of ciliary beating has been observed in the lateral gill cilia of Mytilus edulis, where stimulation of the branchial nerve causes depolarization of the ciliated cell membrane and arrest of ciliary beating(Paparo and Aiello, 1970; Murakami and Takahashi, 1975; Saimi et al., 1983a; Saimi et al., 1983b; Aiello, 1990). A similar mechanism has also been proposed in a number of gastropod larvae(Mackie et al., 1969; Mackie et al., 1976). Increases in CBF have also been linked to local release of neurotransmitter in Tritonia diomedea (Audesirk et al., 1979; Willows et al.,1997) and Helisoma embryos(Goldberg et al., 1994; Christopher et al., 1996; Christopher et al., 1999; Doran et al., 2004).

Unfortunately, recent work concentrates almost exclusively on hormonal control mechanisms, overlooking the possibility that ciliated cells may be electrically excitable. Little is known of the electrical properties of ciliated epithelia that may contribute to CBF control. Of the few patch clamp studies on ciliated epithelial cells possessing motile cilia(Machemer and de Peyer, 1982; Korngreen et al., 1998; Tarran et al., 2000; Nguyen et al., 2001; Ma et al., 2002), only one reported an underlying (hyperpolarization activated) voltage dependent current(Tarran et al., 2000).

In the present study we investigated the electrical properties of ciliated pedal (foot) epithelial (CPE) cells of Tritonia diomedea. Tritoniapermits access to both the CNS networks that control cilia, as well as the CPE cell's intracellular transduction mechanisms. The primary form of locomotion in Tritonia is a ciliary based crawling with CPE cells acting as locomotory effectors. In Tritonia two pairs of identified CNS neurons, Left and Right Pd 5 and 21(Audesirk, 1978; Audesirk et al., 1979; Popescu and Willows, 1999)control speed of locomotion. Further, CPE cells are innervated by neurons containing cilio-excitatory substances including Tritonia Pedal Peptide(TPep-NLS) (Willows et al.,1997). How those excitatory signals are transduced into a change in CBF remains unknown. In the present study we investigated the role of CPE cells as locomotor effector cells by describing their electrical properties and how cilio-excitatory substances exert their control on CBF through modulation of these electrical properties. Here, we describe new depolarization dependent currents in a ciliated epithelial cell. We found a voltage dependent proton current and a Ca²⁺ dependent Cl^- current (I_Cl(Ca)). Similar to other Ca²⁺ dependent conductances, I_Cl(Ca) may regulate the waveform of Ca²⁺ influx through voltage-gated channels by controlling the time course of depolarizing events. This is also consistent with the hypothesis that Ca²⁺ influx controls CBF.

In addition we investigated CBF control by a newly identified cilio-excitatory transmitter, dopamine, as well as previously described cilio-excitatory TPep-NLS (Lloyd et al.,1996; Willows et al.,1997). Specifically, we examined whether dopamine or TPep-NLS influenced the dominant I_Cl(Ca) in the locomotory CPE cells, and whether modulation of I_Cl(Ca) is directly responsible for control of CBF. Here we report that dopamine or TPep-NLS directly reduce or block I_Cl(Ca). Further, we found blocking I_Cl(Ca) alone is sufficient to increase CBF. These results suggest that the cilio-excitatory transmitters and peptides may be working directly to reduce I_Cl(Ca), leading to increases in CBF.

Bergh were trawled from Bellingham Bay, WA, USA, or collected using SCUBA in Puget Sound, WA, USA. They were maintained in 10°C seawater aquaria,and fed sea pens (Ptilosarcus gurneyi) either at Friday Harbor Laboratories (open circulation) or at the University of Washington, Department of Biology (recirculating aquaria).

For most electrophysiology experiments, explants of the ciliated pedal epithelium were used. Pedal tissue hunks (2-3 mm square) were cut from the posterior two thirds of the ≈20 cm long foot surface (removal of small foot sections does not observably reduce the life span, nor alter eating or copulating behaviors). The pedal tissue was pinned out on a Sylgard™coated Petri dish and the epithelial layer dissected away from the foot musculature. This produced sheets of tissue consisting solely of epithelial cells. The sheets were further cut into explants containing 100-500 cells. The dissection was done at room temperature; however, afterwards the explants were cooled again to 10°C for >30 min before experimentation. Recovery of the explants was indicated by beating cilia and the absence of mucus secretion.

CPE cells were isolated from explants prepared as described above. The explants were treated in a divalent-free seawater(Table 1) for 30 min at 10°C, and then returned to artificial seawater (ASW) and kept at 10°C for another 30 min. During this time, a sharp glass electrode was raked across the ciliated edge loosening individual cells. Because of the pretreatment in the divalent free seawater the epithelial cells could be pulled easily out from the epithelium. Individual cells were considered viable if, once separated from the tissue, the cell was clearly ciliated, a photograph could be taken showing the cilia, and the cilia were still beating. All photographs were taken with a CCD camera (Carl Zeiss, ZVS-3C75DE, Thornwood, NY, USA),acquired and saved using Adobe Photoshop (Adobe Systems Inc., San Jose, CA,USA).

For ciliary beat frequency experiments, CPE cells were isolated from explants following published methods(Pavlova and Bakeeva, 1993):explants were dissected into pieces as small as possible, then were pipetted in and out repeatedly to disassociate mechanically the individual ciliated cells. The resulting cells were pipetted into 20 μl of seawater on a large coverslip. Cotton in the solution provided a substrate to which cells could stick, and a second coverslip was placed on top. Thus immobilized, cells and their beating cilia could be visualized and recorded for CBF measurements according to the protocol described below. Individual cells were considered viable if, once separated from the tissue, their cilia were still beating.

Seven different seawater bath combinations were used(Table 1). The recording pipette contained (in mmol l^-1) 180 potassium methyl sulfate (ICN Biomedicals, Irvine, CA, USA) 20 KCl, 10 NaCl, 1 MgCl₂, 400 d-Sorbitol, and either 2 or 20 Hepes, at a pH of 7.3. In some experiments 2 mmol l^-1 ATP and 0.1 mmol l^-1 cAMP were included in the recording pipette. Drugs were cooled during perfusion onto explants to match the bath temperature (10°C) upon delivery. Perfusion was ended after the delivery of 10 ml (into a 3 ml bath) of seawater containing drugs already at their final concentration. Inert dilute dyes were used to confirm adequate perfusion of the drug. All recordings were done with the perfusion off. For experiments comparing the effects of different Cl^- concentrations on the outward currents, we exposed one set of cells to each Cl^- concentration, not the same cells to both, to avoid ambiguities in the junction potential resulting from perfusing solutions with different Cl^- concentrations.

All CBF experiments were done in filtered (Millipore, 0.22 μm filter)seawater, with the exception of those in zero Cl^- seawater. Zero Cl^- seawater contains the following (in mmol l^-1): 400 Na⁺ gluconate, 10 K⁺ gluconate, 10 Ca²⁺gluconate, 50 Mg²⁺ gluconate, 10 Hepes, pH 8.0. Solutions were changed for the CBF experiments using a 100 μl pipette. The solutions were delivered at their final concentrations on one side of the coverslip, then an equal amount was drawn off the other side, wicking the solution over the explants or individual cells. A total of 500 μl was perfused for each solution change.

Dopamine (Sigma-Aldrich Corp., St Louis, MO, USA), TPep-NLS,ZnCl₂ and CdCl₂ were dissolved directly in ASW. The Cl^- channel blockers anthracene-9-carboxylic acid (9-AC), niflumic acid, and 5-nitro-2-(3-phenylpropylamino)benzoic acid (NPPB) (all from Sigma-Aldrich Corp.) were made up in a DMSO stock solution then diluted in seawater to their final concentrations (final solution contained <0.1%DMSO). We prepared the Cl^- channel blocker 4,4′-diisothiocyanatostilbene-2,2′-disulfonic acid (DIDS)(Sigma-Aldrich Corp.) in 0.1 mol l^-1 KHCO₃ creating a 50 mmol l^-1 stock solution. The stock was diluted in seawater until the final concentrations were achieved. Nifedipine (Sigma-Aldrich Corp.) was made up in ethanol in a 50 mmol l^-1 stock solution, which was then diluted in seawater to a final concentration of 50 μmol l^-1(final solution contained 0.1% ethanol).

Change in CBF was measured after the application of drugs to explants of CPE cells. Several explants were placed on a large coverslip in seawater. The tissue was immobilized using cotton fibers with a second coverslip placed on top. The slides were viewed with a Nikon inverted microscope scope (Nikon USA)at 40×. The slide rested on a stage cooled by a Peltier device, which maintained a temperature of 10-12°C. The beating cilia were filmed with an Elmo CCD camera #TSN401A (Plainview, NY, USA), which scans at 59.94 Hz. The video was then projected on a monitor. A Photonic sensor (fiber optic device that measures small changes in light intensity near its sensor probe) was placed in front of the video image of beating cilia and transduced the CBF into an oscillating voltage signal. The voltage signal was amplified with an inline 10× gain amplifier and digitally filtered with a bandpass filter(5-30 Hz) to reduce raster scan signal from the video monitor. The signal was recorded with a DASH -4U (Astro-Med Inc., West Warwick, RI, USA) digital oscilloscope, using a Fast Fourier Transform analysis to recover the dominant frequency of voltage oscillations. We validated the accuracy of CBF measurements by constructing a QuickTime (Apple Computer Inc., Cupertino, CA,USA) video with an artificial `cilium' beating at known rates (2-16 Hz).

CBF was sampled just after solution changes and then at 3 min intervals. We report the time when the largest effect was recorded unless otherwise stated. All experiments end with wash out of the applied drug, and only experiments demonstrating successful reduction in drug effect during wash out were considered for analysis.

The standard whole cell configuration of the patch clamp technique(Hamill et al., 1981) was used to record membrane currents. Explants of CPE cells were placed in a Petri dish lid with ≈3 ml of seawater. The cells were immobilized with a small cut piece of slide glass laid atop the explants, but only partially covering the explants. Cells were mounted and viewed on a Nikon Diaphot inverted microscope(Nikon USA). The bath was cooled to 10°C for all recordings using a Peltier device powered by a 12 V lead-acid storage battery to reduce electrical noise. An Ag-AgCl reference electrode was connected to the bath via an agar bridge. Pipettes were pulled from 50 μl hematocrit glass capillary tubes (VWR 53432-783, West Chester, PA, USA) using a Narishige two stage puller (PP-830, Long Island, NY, USA) to a resistance of 4-9 MΩ. Positive pressure was applied to prevent pipette clogging. Pipettes were moved laterally toward the CPE cells until a resistance change signaled contact. Individual identification of targeted cells was not possible, though only areas with high densities of cilia were targeted (see confirmation experiments below). After contact, suction was applied until a seal formed and the patch ruptured, initiating whole cell configuration. We were unable to measure seal resistances because the suction needed for seal formation also ruptured the membrane. Pipette capacitance was electrically compensated but the whole cell capacitance and the series resistance was not.

Whole cell currents were recorded with an Axopatch 200A amplifier, currents were digitized online using a Digidata 1200 digitizer, and visualized and saved using pClamp 8.01 software (all, Axon Instruments, Sunnyvale, CA, USA). The signal was low pass filtered at 1 kHz and sampled at least at 5 kHz. A holding potential of -50 mV (before junction potential correction) was applied to all cells.

Liquid junction potentials were measured using a Dagan 8900 differential amplifier (Dagan, Minneapolis, MN, USA) with a Dagan 8950 bath reference amplifier. This allowed use of a 3 mol l^-1 KCl reference electrode to measure the junction potentials for different bath/internal solution combinations (Neher, 1992). Only ASW produced a significant junction potential, i.e. +7 mV. This was subtracted post hoc from all ASW recordings. All other bath solutions produced junction potentials less than 1 mV, which were not corrected. All records were leak-subtracted offline. Left uncorrected was the non-linear leak that appears constant under all recording conditions. Input resistance was calculated by averaging the measured resistance from voltage pulses ±10 mV and -20 mV from the holding potential. The baseline was corrected to zero after leak subtraction. In reversal potential experiments a P/N offline leak subtraction protocol was used (pClamp 8.01). Eight sub pulses, each 1/8 of the final pulse and in the opposite polarity, were used to subtract linear leak.

Two types of I-V plots were constructed. The `slow' I-Vplots used the current measured 900 ms into the voltage pulse. The `fast'I-V plots used the current measured just 40 ms into the voltage pulse in an attempt to isolate fast activating currents. For the exponential analysis of current kinetics, using pClamp 8.01 software, we fit exponentials to currents generated with a 60 or 63 mV pulse. First, single exponentials were fit to the decaying slope of the capacitative transient, the point of its divergence from the current record marked the beginning of the fit segment. The end was 900 ms after the beginning of the pulse. A Levenberg-Marquardt fit was used with 100 iterations, using sum of squares minimization with no weighting. Fits were accepted only if accomplished in the 100 iterations with a correlation coefficient >0.900. In addition, the fits were extrapolated out to 4000 ms, and those fits observed to be incompatible with currents shapes actually observed at 4000 ms were discarded. This criterion excluded two observations, each with a slow component τ of over 8000 ms. The least number of exponentials that could be fit successfully was used in the analysis. Boltzman fits were also accomplished using the pClamp 8.01 software. The activation curve was created from the instantaneous tail currents generated by stepping from numerous potentials down to -57 mV, then normalized by the maximum tail current amplitude. Boltzman fits were acceptable with correlations >0.900.

All graphs were generated in Sigma Plot 2000 6.10 software (SPSS Inc.,Chicago, IL, USA) and all statistics in Microsoft Excel software (Microsoft Corporation, Redmon, WA, USA). For comparison of means from two different populations, a two-tailed Student's t-test was used. For comparisons of paired data including all before and after drug treatments and pH changes,a paired Student's t-test was used. For all multiple comparisons an ANOVA was employed with a Student-Newman-Keuls (SNK) test for multiple comparisons. All data are shown as means ± s.e.m., with Nvalues noted.

We used whole cell patch clamp to record currents from CPE cells(Fig. 1A). To begin, we compared currents from cells recorded with and without ATP and cAMP in the recording pipette. We were interested to know if differences in intracellular concentrations of ATP and cAMP affected whole cell currents. We found no differences (Fig. 1B). CPE cells were leaky with an input resistance of 127.5±3.4 MΩ(N=76). If the recording pipette contained ATP and cAMP the resistance rose non-significantly (P=0.07) to 138.3±5.4 MΩ (N=50). The composite voltage-gated currents were large and outward in direction, activating over a full 1 s. In addition the currents turned off slowly, requiring another 1 s to de-activate, judging by the substantial tail currents. The complexity of the tail currents also suggested the possible presence of multiple current types, each de-activating at a different rate. CPE cells' slow I-V relation (current measured after 900 ms of depolarization) (Fig. 1B), revealed that dominant non-leak currents have a strong voltage dependence, activating around -25 mV. This dependence was observed in cells with (N=50) or without (N=76) ATP and cAMP in the recording pipette.

We confirmed that recordings were from CPE cells, by showing that the voltage dependence and amplitude of currents recorded from isolated cells with clearly identifiable cilia (N=12) were identical to the presumptive ciliated cells of epithelial explants (N=76)(Fig. 2). The current amplitude at 63 mV after 900 ms of depolarization in both cell types is similar(I_explants=1163.5±46, N=76; I_{isolated cells}=1182.4±93, N=12; P=0.9) as was the range for current activation(Fig. 2C). Their input resistances differed slightly though not significantly (isolated cells:151.9±37.1, N=12; explant cells: 127.5±3.4, N=76. We conclude that this patch methodology using epithelial explants permits reliable differentiation and identification (based on electrical characteristics) of CPE cells from among muccal cells and others of the epithelium.

The large outward currents appear to be carried mainly by Cl^-ions. Reduction of external Cl^- concentration from 530 mmol l^-1 to 130 mmol l^-1 (LCSW) reduced current amplitude compared to control cells in ASW (Fig. 3A). In LCSW the whole cell currents display similar voltage dependence and kinetics, however overall current is reduced, seen in a comparison of the current voltage relationships(Fig. 3B). For instance, at 70 mV after 900 ms of depolarization, LCSW treated cells have smaller currents(P=2.4×10^-11, I_LCSW=516.0±54, N=18) than do the control cells (I_ASW=1414.8±56, N=76)(Fig. 3B inset). Further, when the I-V is compared at an earlier time(Fig. 3C) (40 ms into voltage step or `fast' I-V) the current in LCSW cells is again significantly decreased compared to control cells (P=3.1×10^-11, I_LCSW=156.5±13, N=18; I_ASW=754.6±38.3, N=76)(Fig. 3C inset), however, the reduction is much greater (75% at 40 ms versus 56% at 900 ms). The relative contribution of the component currents is therefore dependent on the time of activation, suggesting the Cl^- dependent conductance activates faster than do other component currents.

Two Cl^- channel blockers, DIDS (500 μmol l^-1) and 9-AC (50 μmol l^-1), each significantly reduced the total current at 63 mV (Fig. 3D): DIDS 36.5±5.3% reduction (N=4, P=0.01) and 9-AC 33.2±6.6% (N=4, P=0.01). The Cl^- channel blockers reduced the Cl^- conductance less than did LCSW treatment(64% reduction at 60 mV). Two other Cl^- channel blockers (50μmol l^-1 NPPB, N=2; and 50 μmol l^-1niflumic acid, N=2) failed to reduce current amplitude.

Reversal potential measurements confirmed the presence of a Cl^-current. The mean measured tail current reversal potential(E_rev) for control cells in seawater is -34.8±1.7 mV (N=13) (Fig. 4A),which is positive to the E_Cl- predicted by the Nernst equation of -68.5 mV. This measurement includes all conductances activated by our depolarizing pulse, not only the Cl^- current. Because of the ambiguity of each current's contribution to E_rev we relied on a comparison of the shift in E_rev in cells treated with LCSW compared to control cells. The LCSW treated cells have a E_rev of -0.1±5.2 mV (N=9)(Fig. 4B), representing a positive shift of +34.7 mV (Fig. 4C). This coincides well with the predicted shift in E_Cl- of +33.2 mV, calculated using the Nernst equation,when extracellular Cl^- is reduced from 530 mmol l^-1 to 130 mmol l^-1. Thus the fast activating Cl^- dependent conductance is likely a Cl^- current.

To investigate further the differential activation time between the Cl^- current and other residual conductances, exponentials were fit to the outward currents at 63 or 60 mV(Fig. 5A). In control ASW,activation kinetics of composite outward currents were well fit by two exponentials (τ₁=747±100,τ ₂=68.4±7.8; N=29). In LCSW, the rapid component was eliminated and the currents fit a single exponential similar to the slow component in ASW (Fig. 5B). Thus there appear to be at least two current types present in the CPE cells, a Cl^- current, activating quickly, and a second slowly activating conductance. As discussed below, this second component appears to be I_H.

The Cl^- currents depend on extracellular [Ca²⁺] and its ability to enter the cell. Cells in zero Ca²⁺ (0 Ca²⁺ SW) seawater showed marked reduction in current amplitude(Fig. 6A). Slow I-V(Fig. 6Bi) revealed 0 Ca²⁺ SW caused a general reduction in conductance, though not as profoundly as does LCSW or LCSW/0 Ca²⁺ SW. Comparisons of maximum current at 70 or 73 mV (Fig. 6Bii) showed that the 0 Ca²⁺ SW treatment, LCSW and LCSW/0 Ca²⁺ SW all caused a significant(P=8.7×10^-15) current reduction. The fast I-V (Fig. 6Ci)demonstrated that 0 Ca²⁺ SW treatments decreased total current primarily by reduction of the faster activating Cl^- current, and not by the other slower components. The comparisons of maximal current at 70 or 73 mV after 40 ms of depolarization(Fig. 6Cii) showed that in both 0 Ca²⁺ SW and LSCW, amplitudes were reduced significantly(P=3.39×10^-11), and the percent reduction in current for cells treated in 0 Ca²⁺ SW (53.2%) was greater after 40 ms of depolarization than after 900 ms of depolarization (40.8%).

We used two Ca²⁺ channel blockers to investigate further the mechanism of extracellular Ca²⁺ influence on Cl^-current. We found that the non-specific Ca²⁺ channel blocker Cd²⁺ (2 mmol l^-1) significantly reduced total current amplitude (Fig. 6D,F) by 29.56±5.9% (P=0.01, N=7). The more specific L-type Ca²⁺ channel blocker nifedipine(Fig. 6E,F) reduced total current by 36.13±1.9% (P=1.9×10^-5, N=5) a reduction very similar to the reduction seen in 0 Ca²⁺ SW. The action of Ca²⁺ channel blockers to inhibit the Cl^- current to levels similar to those seen in cells treated in LCSW suggests that Ca²⁺ influx from extracellular sources enhances the Cl^- current.

Proton currents are commonly seen in epithelial cells(Decoursey, 2003) and are well described in molluscs (Byerly et al.,1984; Byerly and Suen,1989). Proton currents contribute to the regulation of internal pH, a function that may be particularly important for the layers of cells exposed to the seawater environment of Tritonia diomedea. We studied the slowly activating conductance in isolation by removing all external Cl^- and all external Ca²⁺(Fig. 7Ai). The cells treated in 0 Cl^-/0 Ca²⁺ SW had currents that activated slowly and the resulting tail currents displayed a transient increase in amplitude after the voltage returned to rest. At the beginning of each step there was also a small inward current that was likely the residual Cl^-current. A plot of the activation curve, derived from instantaneous tail current measurements, can be fit with a Boltzman curve(correlation=0.949, N=7)(Fig. 7Aii). Half activation was 13.5 mV (N=7) with a K value of 20.5 (N=7),corresponding to the movement of 1.2 elementary charges across the membrane field, assuming the simple open/closed model for channel activation. The small number of moving charges puts this current in the range of known H⁺channels, but not similar to any other voltage dependent channels(Decoursey, 2003). The fast I-V (Fig. 7Aiii) (to limit any slowly activating conductance) confirms that all fast activating Cl^- current is removed in 0 Cl^-, 0 Ca²⁺SW.

A general property of many voltage dependent H⁺ currents, and specifically molluscan proton currents, is their sensitivity to Zn²⁺ (Mahaut-Smith,1989; Decoursey,2003). We tested the sensitivity of the possible H⁺current to Zn²⁺ and found it reduced current amplitude(Fig. 7Bi,Bii). 10 μmol l^-1 Zn²⁺ reduced the current amplitude at 70 mV by 63.69±6.7% (P=0.00007, N=7)(Fig. 7Biii).

Finally, we examined the effect of altering pH on the putative H⁺ current. We found that after lowering the external pH to 7.0,the currents were reduced (Fig. 7Ci), as would be expected for H⁺ currents. We also observed a shift in the reversal potential after reducing the external pH to 7.0 (Fig. 7Cii). The average reversal potential at external pH 8.0 was -24.43±12 (N=7) and in external pH 7.0 the reversal potential shifted to +27.44±9.5 mV,corresponding to a mean positive shift of 51.87±6.7 mV (N=5). The Nernst equation predicts a shift in E_H+ with a tenfold reduction in external H⁺ concentration of 56 mV.

Dopamine immunoreactivity is found in neurites innervating CPE cells (S. D. Cain and G. A. Pavlova, unpublished observations). Dopamine applied directly to explants of CPE cells increased CBF dramatically(Fig. 8A). Dopamine is excitatory at mid-range concentrations: 10 μmol l^-1 increases CBF 161.4±12% (N=5, P=0.0001); 100 μmol l^-1 increases CBF 198.9±16% (N=5, P=0.0003); and 1 mmol l^-1 increases CBF 123.7±11%(N=6, P=0.0001) (Fig. 8B). The dose-response is not linear; at both very low and very high concentrations the excitatory effect disappears resulting in no significant change in CBF from control. 1 μmol l^-1 dopamine non-significantly increases CBF 26.7±15% (N=5, P=0.137), and 10 mmol l^-1 increases CBF only 18.4±9.9% (N=6, P=0.12). To control for the possibility that oxidation of dopamine could be driving the excitation of CBF we included 50 μmol l^-1 ascorbic acid, an antioxidant, with 100μmol l^-1 dopamine. We found that ascorbic acid made no appreciable difference to the excitatory properties of dopamine. Dopamine (100μmol l^-1) and 50 μmol l^-1 ascorbic acid together increased CBF 199.8±7.6% (N=3), an increase not significantly different (P=0.97) than with 100 μmol l^-1 dopamine alone.

Dopamine also excites CBF in isolated single CPE cells(Fig. 8C). Dopamine raised CBF 95.4±8.7% from 8.4±0.5 Hz to 16.2±1.3 Hz (N=5, P=0.0002). Further, we also found at 10 μmol l^-1,(N=5), dopamine excites CBF more than does TPep-NLS (10 μmol l^-1, N=8, P=0.0005).

The perfusion of 100 μmol l^-1 dopamine reduces the whole cell current amplitude in CPE cells (Fig. 9A,B). Specifically at 73 mV the maximal current is significantly reduced 41.2±5.8% (N=7, P=0.0004) after the application of 100 μmol l^-1 dopamine(Fig. 9C inset). The normalized instantaneous I-V shows that current reduction is similar at all voltages where the current is active (Fig. 9C), and the voltage dependence does not shift. The representative currents in Fig. 9A,B also suggest that the faster components of the whole cell currents during the voltage pulse and in the tail current are disproportionately affected. This observation is further supported when current size is compared after only 40 ms of depolarization. The current activated at 40 ms is predominantly made up of the faster I_Cl(Ca). At 40 ms, the dopamine induced decrease in current size increases to 50.3±6.0%, an indicator that dopamine may be preferentially acting on I_Cl(Ca).

TPep-NLS elicits responses similar to those of dopamine. 10 μmol l^-1 TPep-NLS reduces the whole cell current amplitude(Fig. 9D,E). The normalized instantaneous I-V, like dopamine, shows TPep-NLS affects the currents at all voltages where they are active (Fig. 9F) without shifting the voltage dependence. Maximal current at 73 mV is inhibited (Fig. 9F,inset)by 29.5±5.6% (N=10, P=0.0005) after the application of 10 μmol l^-1 TPep-NLS. The representative traces again suggest disproportionate effect on the faster components, and interestingly TPep-NLS seems to enhance the size of the slowly deactivating tail currents(Fig. 9E). Again, like dopamine, at 40 ms TPep-NLS induced decrease in current size is greater(32.0±3.4%), possibly suggesting specific action on I_Cl(Ca). The inhibition by dopamine and by TPep-NLS is similar, suggesting a common mechanism of action upon CBF excitation.

Outward currents in CPE cells consist of two components: a rapidly activating I_Cl(Ca) (τ=68.4±7.8 ms) and a slowly activating I_H (τ=747±100 ms). Because of the difference in τ values we used an exponential analysis to determine specifically which current of the CPE cells was inhibited by dopamine and TPep-NLS. The currents before and after 100 μmol l^-1 dopamine exposure were both successfully fit with a double exponential(Fig. 10A). Fig. 10A further illustrates qualitatively that the faster components are disproportionately affected by dopamine. The time constants for the two currents remained unchanged, as did the amplitude of the slower I_H; however, the amplitude of I_Cl(Ca) was reduced by 55%(Fig. 10B). This suggests that dopamine selectively inhibits I_Cl(Ca).

Again TPep-NLS seemed to mimic dopamine, selectively inhibiting I_Cl(Ca) (Fig. 10D). Currents from both controls and cells exposed to 10 μmol l^-1 TPep-NLS were successfully fit with double exponentials(Fig. 10C). The time constants for both currents and the amplitude for I_H remained unchanged after TPep-NLS exposure; only the amplitude of I_Cl(Ca) was reduced, again by roughly half (49%)(Fig. 10D).

One puzzling result was the increase in input resistance observed after exposure to dopamine and TPep-NLS. Though not reflected in our leak-subtracted voltage dependent current data, we found that dopamine increased input resistance 34.0±15% (N=7), reflected as a decrease in I_leak at our holding potential of -57 mV. TPep-NLS also increased input resistance 25.9±18% (N=10). To attempt to establish if this decrease in I_leak was due to blocking specific ions, we used the Cl^- channel blocker 9-AC, hypothesizing that dopamine and TPep-NLS specifically block Cl^- leak. 9-AC increased input resistance by 26.9±9.8% (N=7), an amount similar to dopamine and TPep-NLS, suggesting that they block both I_Cl(Ca) and the Cl^- components of I_leak.

Dopamine and TPep-NLS both excite CBF and inhibit I_Cl(Ca), an outward Cl^- current, and possibly I_(Cl-)leak. We tested the possibility that blocking Cl^- influx could mimic the excitatory effects of dopamine and TPep-NLS on CBF. DIDS significantly increased CBF(Fig. 11) at both 500 μmol l^-1 (95.3±14% increase from control, N=5) and 1 mmol l^-1 (152.3±15% increase, N=5) concentrations. To confirm, we removed all external Cl^- from the seawater bath, and again observed a significant 117.1±17% (N=5) increase in CBF(Fig. 11), similar to excitation seen with dopamine and TPep-NLS. Taken as a whole, our results suggest that dopamine and TPep-NLS block Cl^- currents directly,resulting in CBF excitation.

There is physiological and anatomical evidence that ciliated epithelia and ciliary beating are under neural control(Aiello and Guideri, 1964; Murakami and Takahashi, 1975; Mackie et al., 1976; Audesirk, 1978; Audesirk et al., 1979; Willows et al., 1997; Mathew, 1999; Nguyen et al., 2001). However,there are no previous studies characterizing voltage dependent currents in epithelial cells with motile cilia. The focus of our experiments was an investigation of the underlying electrical characteristics of ciliated pedal epithelial cells of Tritonia diomedea to determine whether modulation of membrane properties of these cells is a pathway for CBF control.

We found that ciliated epithelial cells possess ionic currents triggered by depolarization. Tritonia CPE cells have a fast activating Cl^- current dependent on the influx of Ca²⁺, I_Cl(Ca), and a slowly activating proton current, I_H. Cl^- currents similar to Tritonia's I_Cl(Ca) have been reported in Xenopus oocytes,vertebrate olfactory receptors, cardiac muscle, smooth muscle and secretory epithelia (Hartzell et al.,2005), but we report here the first instance of a Ca²⁺dependent Cl^- current in a ciliated epithelial cell. Our results suggest an electrical mechanism for neuronal control of cilia beating. The presence of I_Cl(Ca) provides a mechanism to control depolarization and recovery.

An I_H in CPE cells is not unexpected, and Tritonia's I_H is similar in many respects to those described in other epithelial cells(Decoursey, 2003). Unfortunately, I_H activation does not reach a steady state after 1 s of depolarization. Therefore, we have only used I_H amplitude and τ as estimates to suggest that relative to I_Cl(Ca), I_H activates more slowly. Possible small errors associated with these current fits will not alter this conclusion. Similarly, fits of the I_Hactivation curve do not perfectly describe the activation curve because the currents have not reached a steady state, a problem common when fitting slowly activating proton currents (Decoursey,2003).

Identifying the Cl^- channels underlying I_Cl(Ca) is problematic because I_Cl(Ca)properties do not fit neatly into any of the most probable of the epithelial Cl^- channel categories. CLC channels are voltage dependent(Jentsch et al., 2002; Maduke et al., 2000) like I_Cl(Ca) but do not exhibit Ca²⁺ dependence. CLC channels are also commonly blocked by μmolar concentrations of extracellular Zn²⁺ (Jentsch et al., 2002), something we do not observe for I_Cl(Ca). CFTR channels(Jentsch et al., 2002) are activated by cAMP and are voltage independent, but neither of these are traits of I_Cl(Ca). In fact the presence of cAMP in our recording pipettes had no effect on current size or kinetics. Swelling activated Cl^- channels can be ruled out because I_Cl(Ca)shows no apparent intracellular ATP dependence(Jentsch et al., 2002). This leaves only Ca²⁺ activated Cl^- channels (CaCC). CaCC currents and I_Cl(Ca) share many common characteristics. Both have steady state current voltage relations showing strong outward rectification(Begenisich and Melvin, 1998; Hartzell et al., 2005) and both open channel current voltage relations are nearly linear(Begenisich and Melvin, 1998). Both can be blocked by DIDS (Hartzell et al., 2005; Qu and Hartzell,2001) and 9-AC (Jentsch et al., 2002; Qu and Hartzell,2001). CaCC currents are activated by increases in[Ca²⁺]_in, which may come either by Ca²⁺influx or from release of internal stores, or both(Hartzell et al., 2005). I_Cl(Ca) also appears to be potentiated by increases of[Ca²⁺]_in, demonstrated by the reduction in I_Cl(Ca) by Ca²⁺ channel blockers Cd²⁺ and nifedipine. Therefore, we suggest that I_Cl(Ca) is carried by CaCC-like channels.

Our conclusion that I_Cl(Ca) is carried by CaCC-like channels is in part based on our findings that by removing[Ca²⁺]_out or by blocking Ca²⁺ influx we can significantly reduce I_Cl(Ca). These results are consistent with the presence of Ca²⁺ channels. Specifically the sensitivity of these channels to nifedipine suggests the presence of voltage-gated calcium channels. It was therefore unexpected that we did not record any fast inward currents in our whole cell recordings that could be carried by Ca²⁺ions. Even the replacement of Ca²⁺ with Ba²⁺ to increase conductance did not reveal any Ca²⁺ currents. These are similar to Barish's findings (Barish,1983), who described the I_Cl(Ca) in Xenopus oocytes, and found that the Cl^- current depended on Ca²⁺ influx through voltage-gated Ca²⁺ channels,though they did not appear in the whole cell recordings. Small Ca²⁺currents could be masked by I_Cl(Ca), I_H, or large capacitance transients, each rendering Ca²⁺ currents invisible under our recording conditions.

We found the E_rev of I_Cl(Ca) to be-34.8±1.7 mV, compared with a predicted value of -68.5 mV for E_Cl-. The measured E_rev of I_Cl(Ca) in LCSW represented a shift almost exactly as predicted (measured=+34.7 mV, predicted=+33.2 mV), implying that Cl^- is the dominant contributor to I_Cl(Ca). Why then the disagreement between the measured and predicted values for E_rev? First, although we tried to limit it, I_H is still contributing to E_rev. E_H in 8.0 pH seawater is -24.43±12 mV, more positive than either the predicted or measured E_rev for I_Cl(Ca). Thus its contribution will shift E_rev to more positive values. The magnitude of the E_rev error, however, is great enough to suggest the existence of additional sources of error. A second possibility arises from the fact that Cl^- channels do not discriminate well among anions or cations and consequently are believed to have relatively large pore diameters. CaCCs exhibit only a tenfold selectivity between ions that differ in radii by 1.5 Å (Qu and Hartzell,2000; Hartzell et al.,2005). This results in small but significant permeability to anions used as Cl^- replacements. For example, in maxi-Cl^- channels the normalized permeability coefficients for common Cl^- replacements are: glucuronate 0.78(Woll et al., 1987; Bosma, 1989), aspartate 0.62(Bosma, 1989), gluconate 0.25(Ravesloot et al., 1991) and methyl sulfate 0.71 (Ravesloot et al.,1991). These values are large enough to substantially alter E_rev measurements for Cl^- currents away from predicted E_Cl-. Here we use 180 mmol l^-1 of methyl sulfate in the recording pipette, a molecule shown to be highly permeable in at least one type of Cl^- channel(Ravesloot et al., 1991).

I_Cl(Ca) in CPE cells would permit several physiological adaptations. For example, I_Cl(Ca) could control the waveform of depolarization as well as Ca²⁺ entry through voltage-gated channels. Other outward, Ca²⁺ dependent conductances that modulate Ca²⁺ entry by regulating depolarization include the large-conductance Ca²⁺ activated K⁺ channel (BK). BK channels have been described as modulators of voltage-gated Ca²⁺channel activity (Vergara et al.,1998), and I_K(Ca) has also been shown to play a role in transmitter release by regulating the amount of Ca²⁺influx (Yazejian et al.,1997). Ca²⁺ influx has been found to be a trigger for increasing CBF in many different types of ciliated cells(Eckert, 1972; Christopher et al., 1996; Korngreen et al., 1998; Barrera et al., 2004; Nguyen et al., 2001; Zagoory et al., 2001; Doran et al., 2004). Our results are consistent with the hypothesis that I_Cl(Ca)acts to regulate Ca²⁺ entry and CBF.

We next investigated the possible modulation of I_Cl(Ca)by neurotransmitters and peptides to ascertain if I_Cl(Ca)plays a role in controlling CBF and locomotion. We found that dopamine and TPep-NLS act directly to reduce I_Cl(Ca) (not viainternal Ca²⁺), and reduce I_(Cl-)leak,producing an excitatory effect on CBF similar to Cl^- channel blockers. This evidence is consistent with our hypothesis that dopamine and TPep-NLS depolarize the cell by reducing I_Cl(Ca) and I_(Cl-)leak, which in turn causes Ca²⁺ influx and change in CBF.

We found, in addition to 5-HT (Aiello,1990; Goldberg et al.,1994; Christopher et al.,1996; Christopher et al.,1999; Willows et al.,1997) and TPep-NLS (Willows et al., 1997; Popescu and Willows, 1999), that dopamine also produces significant increases in the CBF of CPE cells. All previous reports of dopamine action on cilia beating in molluscs (Catapane et al.,1978; Aiello,1990), marine invertebrates(Wada et al., 1997) and even vertebrates (Maruyama et al.,1983) found that dopamine inhibits CBF, often acting in opposition to an excitatory 5-HT pathway (Catapane et al., 1978). Tritonia appears to be the first example where dopamine excites ciliary beating. However, the CPE used in the present study are innervated by neurons of the pedal ganglia, a developmentally and functional divergent central nervous system structure(Chase, 2002) that may possess a unique repertoire of dopamine receptors.

We also report that both dopamine and TPep-NLS decrease the size of I_Cl(Ca), preventing the Cl^- current from repolarizing the cell membrane after depolarization, thus increasing the amount of Ca²⁺ influx. In addition we found that blocking Cl^- channels or removing external Cl^- alone was sufficient to cause dopamine/TPep-NLS-like increases in CBF. Because Ca²⁺ increases I_Cl(Ca) amplitude and because I_Cl(Ca) is decreased in the presence of dopamine or TPep-NLS, we suggest that dopamine and TPep-NLS are acting directly to block or reduce I_Cl(Ca). Dopamine modulation of Cl^-currents has been reported in a number of different animals and cell types,including leech neurons, where dopamine leads to an increase in Cl^-channel activity (Ali et al.,1998), and in rod cells of salamander retina, where again dopamine leads to an increase in I_Cl(Ca) and Cl^- efflux(Thoreson et al., 2002).

Dopamine and TPep-NLS also increase input resistance by selectively blocking I_(Cl-)leak. Blocking I_(Cl-)leak can dramatically increase cell excitability and decrease the input current necessary to cause depolarization or cause depolarization outright. Spontaneous single channel I_Cl(Ca) currents in rabbit pulmonary artery smooth muscle cells were rare in solutions with low or no[Ca²⁺]_out; however, in [Ca²⁺]_outof 10 mmol l^-1, spontaneous currents could be evoked(Piper and Large, 2003). This suggests that in seawater, Ca²⁺ activated Cl^- channels may contribute to I_(Cl-)leak and that both I_(Cl-)leak and I_Cl(Ca) can be blocked by a common mechanism.

In conclusion, we propose that receptor binding of dopamine and TPep-NLS leads directly to reduction of I_Cl(Ca)(Fig. 12). This reduction increases duration of any depolarization and Ca²⁺ influx via voltage-gated channels, and may also provide a mechanism of depolarization by blocking I_(Cl-)leak contribution to the resting potential. Blockage of I_Cl(Ca) and I_(Cl-)leak leads to increases in CBF, possibly by causing Ca²⁺ influx through voltage-gated Ca²⁺ channels. The ability of dopamine and TPep-NLS to block I_Cl(Ca) may not be the only pathways used to affect CBF and locomotion. Neurotransmitter action on Cl^- channels and the entire CBF control cascade warrants further study. Excitable membranes give locomotory CPE cells the capability for graded CBF change under central nervous system control.

List of abbreviations
 
• 5-HT

  serotonin

 
• 9-AC

  anthracene-9-carboxylic acid

 
• ACh

  acetylcholine

 
• ASW

  artificial seawater

 
• BK

  Ca²⁺ activated K⁺ channel

 
• CaCC

  Ca²⁺ activated Cl^- channels

 
• CBF

  ciliary beat frequency

 
• CNS

  central nervous system

 
• CPE

  ciliated pedal epithelial cells

 
• DA

  dopamine

 
• DIDS

  4,4′-diisothiocyanatostilbene-2,2′-disulfonic acid

 
• NPPB

  5-nitro-2-(3-phenylpropylamino)benzoic acid

 
• SCSW

  low Cl^- seawater

 
• TPep-NLS

  Tritonia Pedal Peptide

Funded by NIH Fellowship 5 F31 NS047922-02 (O.M.W.). We appreciate numerous discussions with Drs R. Wyeth and S. Cain, and thank Drs W. Moody and M. Bosma for reading earlier versions of the manuscript. We thank Dr W. Moody for donation of equipment and space necessary for the completion of this work, and P. Hunt for photography support. A final thanks goes to Tatia Chay Woodward for logistical support.