Serotonin modulates the voltage dependence of delayed rectifier and Shaker potassium channels in drosophila photoreceptors

Neuron, Vol. 14, 845-856, April, 1995, Copyright© 1995by Cell Press

Serotonin Modulates the Voltage Dependence of Delayed Rectifier and Shaker Potassium Channels in Drosophila Photoreceptors Wulf Hevers and Roger C. Hardie University of Cambridge Department of Zoology Downing Street Cambridge CB2 3EJ England

Summary We describe the in situ modulation of potassium channels in a semi-intact preparation of the Drosophila retina. In whole-cell recordings of photoreceptors, rapidly inactivating Shaker channels are characterized by a conspicuously negative voltage operating range; together with a delayed rectifier, these channels are specifically modulated by the putative efferent neurotransmitter serotonin. Contrary to most potassium channel modulations, serotonin induced a reversible positive shift in the voltage operating range, of +30 mV for the Shaker channels and +10-14 mV for the delayed rectifier. The maximal current amplitudes were unaffected. Modulation was not affected by the subunit-specific Shaker mutations Sh e62and T(1;Y) W32 or a null mutation of the putative modulatory subunit eag. The modulation of both channels was mimicked by intracellularly applied GTPyS.

Introduction

Potassium channels are generally considered to be the most widespread and diverse of the ion channel families (Rudy, 1988). They are essential for determining many aspects of neuronal function (Hille, 1992), and their ability to be modulated provides a major mechanism for regulating neuronal excitability, synaptic plasticity, and learning (Kandel and Schwartz, 1982; Kaczmarek and Levitan, 1987; Alkon et al., 1991). The fast transient Shaker channel of Drosophila was the first potassium channel cloned (Baumann et al., 1987; Kamb et al., 1987; Papazian et al., 1987; Tempel et al., 1987) and represents the prototypic member of a large family of related potassium channels expressed in organisms as diverse as worms and man (Rudy, 1988; Wei at al., 1990; Salkoff et al., 1992). It is also the most extensively studied of the potassium channels, and site-directed mutagenesis studies in heterologous expression systems have revealed many of the structural features responsible for properties such as voltage dependence of channel gating, inactivation, and ion permeation (reviewed by Pongs, 1992; Jan and Jan, 1992; Hoshi and Zagotta, 1993; Bezanilla and Stefani, 1994). By contrast, very few studies have addressed the molecular basis of modulation (Moran et al., 1991; Zhong and Wu, 1993; Drain et al., 1994; Poulain et al., 1994). Indeed, the possibility that native Shaker channels can be physiologically modulated has only recently been suggested by the

inhibitory action of a calmodulin antagonist and a much weaker facilitation induced by 8-bromo-cGMP on Shakerdependent currents in Drosophila muscle fibres (Zhong and Wu, 1993), and more recently by the effect of calcium on the amplitude of the A-current in Drosophila larval muscles (Poulain et al., 1994). Until recently, the only native Shaker channels accessible to biophysical analysis were those on the muscle fibres (reviewed in Wu and Ganetzky, 1992), although evidence suggested widespread expression of Shaker channels in the nervous system (Schwarz et al., 1990; Baker and Salkoff, 1990). Recently, however, we discovered that Drosophila photoreceptors also express A-channels encoded by the Shaker gene that are amenable to single-channel and whole-cell patch-clamp analysis (Hardie et al., 1991; Hardie, 1991). A distinct physiological feature of the Shaker channels expressed in the photoreceptors is their voltage operating range, located 30 mV more negative than that reported in muscle or in heterologous expression systems. Since this would render the photoreceptor Shaker channels inoperative over most of the cell's physiological voltage response range, we wondered whether the voltage dependence might be subject to modulation. As in the majority of insects, the proximal retina in Drosophila is innervated by two prominent wide-field serotonergic neurons (N&ssel, 1987; reviewed in N&ssel, 1991; Homberg, 1994). In Drosophila photoreceptors we found that 5-HT (serotonin) induces a pronounced modulation of the Shaker channels. In contrast to virtually all previous reports of potassium channel modulation, which involve simple up- or downregulation of activity (e.g., Belardetti and Siegelbaum, 1988; Volterra and Siegelbaum, 1988; also reviewed in Levitan 1994; but see Atkins et al., 1991; Perozo et al., 1989), the modulation is manifested in a large ( - 3 0 mV) shift in the voltage dependence of operation for the Shaker channel, with the result that the voltage operating range now coincides with that reported for virtually all other Shaker channels (Salkoff and Wyman, 1983; Wu and Haugland, 1985; Zagotta and Aldrich, 1990a; Iverson et al., 1988). The voltage dependence of a delayed rectifierlike channel also found in these cells is shifted by about +10-14 mV, suggesting the possibilityof a common underlying mechanism. Despite the detailed structural knowledge mainly obtained by site-directed mutagenesis of Shaker, molecular mechanisms of modulation in any cloned potassium channel are just beginning to emerge (Hoger et al., 1991 ; Moran et al., 1991 ; Aiyar et al., 1993; Bosma et al., 1993; Zhong and Wu 1993; Drain et al., 1994; Huang et al., 1994; Esguerra et al., 1994). Drosophila photoreceptors provide the rare opportunity to study the modulation of a cloned potassium channel that can be genetically manipulated in situ. Thus this new preparation can be expected to provide valuable information on how fundamental properties of potassium channels, and in particular the voltage dependence of gating, are regulated under physiological conditions.

Neuron 846

a

~

0

I

+lOmV

-90

,

lnA

I

0

~l~ms

+lOmV -90

Figure 1. The Semi-IntactRetinaPreparation (a) Photographof a semi-intactpreparationof the Drosophilaretina (R). The retinawas fixedwith its optic gangliain a suctionpipette(P). The distal ends of the photoreceptorcells were accessibleto patch pipettes (magnification100x). (b) Whole-cell recordingof photoreceptorpotassiumcurrents in the semi-intact preparation.The potentialwas steppedto +10 mV after a rangeof hyperpolarizingprepulses(1 s) from 0 to -100 inV. With prepulsesdownto -40 mV, onlya delayedrectifier-likeconductance was activated(IKs).However,with prepulsesbelow-40 mV, a rapidly inactivatingA-currentencodedat leastin part by the Shakergenewas revealed(IA).In addition, a third componentnot present in everycell becamevisible with prepulsesbelow-60 mV (IKf). (c) In the mutant ShKs~, which does not expressfunctionalShaker channels,the fast component,IA, is completelyabsent.

Results The Semi-Intact Preparation We have developed a semi-intact preparation of the Drosophila retina in which the retina remained intact with at least the first optic ganglion, the lamina, still attached (Figure la). The distal parts of the photoreceptors were freed from adjacent primary pigment cells and were accessible to cell-attached and whole-cell patch recordings. The more proximal parts of the retina, including the photoreceptor axons, remained intact, and consequently, any functional interactions between photoreceptors, secondary pigment cells, and second order neurons can be expected to be preserved. The basic properties of the photoreceptors recorded in this semi-intact preparation showed no obvious differences to those recorded in dissociated ommatidia (see Hardie, 1991). The input resistance of these cells under dark-adapted conditions (with potassium gluconate in the electrode) was 500 MQ to 1 G~; the cell capacitance was 35-50 pF. The membrane resting potential was between

- 5 0 and - 6 0 mV when recorded under dim red light and about - 7 0 mV in total darkness. Under these conditions the physiological light response of these cells could easily be observed, both as flash responses and as single photon events (bumps) elicited under dim light. Compared with the dissociated preparation, however, cells in the semi-intact preparation were more stable with respect to run-down effects, which can include a pronounced negative drift in the voltage operating range of Shaker as well as delayed rectifier currents (Hardie, 1991). With the prolonged recording times required for application and washing out of any neuromodulators, the stability of voltage-dependent characteristics is critical. In the semi-intact preparation, we were routinely able to record from cells for more than 20 min without the voltage-dependent characteristics of the potassium channels changing by more than 3 inV. Most importantly, the semi-intact preparation showed little variability in the effects of 5-HT, which induced modulation in 100% of cells tested, as compared with about 45% with dissociated photoreceptors (see below).

Potassium Currents and the Effect of 5-HT As described by Hardie (1991), Drosophila photoreceptors show at least three different voltage-gated potassium channels. Depolarizing the cells from prepulse potentials above - 4 0 mV, the whole-cell current is dominated by a slow delayed rectifier-like conductance (IKs; Figure lb). With prepulse potentials of - 5 0 mV and below, a fast transient A-current (IA) is revealed, along with a delayed rectifier with intermediate kinetics (IKf)found in only some cells. Mutant analysis indicates that the fast A-channel is encoded at least in part by the Shaker locus of Drosophila (Figure lc) (Hardie et al., 1991). The unusual operating range of this IA current, shifted by 30 mV to more negative values in respect to Drosophila muscles, renders the channels completely inactivated over most of the photoreceptore' operating range and suggested them as a target for neuromodulation. Whereas most putative neurotransmitters in the fly visual system had no influence on the potassium channels, 5 - H T was consistently able to modulate these conductances. Although 5-HT was effective at concentrations as low as 40 ~M, effects developed slowly at these concentrations, so that we routinely used 1 mM 5-HT in order to obtain rapid and reproducible effects. Following bath application of 1 mM 5-HT, a reversible shift of the voltage dependence of the Shaker IA conductance became obvious within 30 s to 2 min, with removal of the inactivation of IA now occurring at more positive prepulse potentials (Figure 2; see Figure 5). Modulation was not restricted to the Shaker current, and a pronounced effect on the delayed rectifier current IKsalso became apparent with longer application times (e.g., Figure 2a, 3.5'). Although 5-HT appeared to cause a large reduction in amplitude with these protocols, increasing the command voltage (to +30 mV; Figure 2b) revealed that the maximal amplitudes of both IA and IKswere largely unaffected. This indicated that the reductions in amplitude apparent in Figure 2b were largely due to a shift of the voltage dependence of activation (see further below).

Serotonin Modulates Shaker Channels 847

a a

Control

+ 5-HT (1.5')

-5o

50

-

SOs~

5oopA

+5.T I\\

L10os

^ - ~ \\

~ u s l ~ 2Control ph ~

i

5-HT (3.5')

Wash

Ls°°pA

-45 J"

10rns -

b~

90s

30/~

, 5-HT ~

800pA

s'

~

~

Control J ,Z Ms(39

+10rnV L _ _

5o

-

Figure 3. Modulation Is Specifically Induced by 5-HT Receptors 0rnV +10mY

OmV

--I

-100

+10mV

-100

b A ~l

Control 1000pA

~.

+20

N ---100

with 5-HT

- 20

+ 30 mV

+20

•

N

+ 30 mV

(a) The effect of 5-HT could best be demonstrated from a prepulse potential of -45 mV, which normally inactivates the Shaker channels. At 30 s after application of 5-HT, inactivation was already removed at this voltage and the fast Shaker conductance was progressively revealed; the delayed rectifier was slightly reduced. The effect was rapidly washed out. (b) The specific competitive 5-HT antagonist methysergide (Ms; 2 mM) was applied either after (above) or prior to (below) the application of 5-HT (1 mM). When applied after the effect of 5-HT was established (without washing out the 5-HF), it was able to reverse its effect over a time course of minutes (upper traces). Prior application of methysergide was able to block responses to subsequent application of 5-HT (lower traces). Whereas methysergide (Ms) does not modify the response itself (trace: 3' after Ms application), it blocked any response to 5-HT (trace: 10' after 5-HT application). Prepulse potential was -50 mV, therefore the Shaker conductance is already just visible.

-100

Figure 2. The Modulation of Photoreceptor Potassium.Currents by 5-HT (a) Following bath application of 5-HT (1 mM), the Shaker conductance was already elicited with more positive prepulses (-40 and -30 mV; upper right and lower left, respectively), whereas the overall current appeared to be reduced. Modulation was not restricted to IA, and the amplitude of IK, was also reduced following 5-HT application. Modulation of both I, and IK, could easily be reversed by washing out the 5-HT with fresh Ringer solution (lower right). Voltage protocols were performed as in Figure lb. (b) Although the current is reduced in the presence of 5-HT when stepping to a command voltage of +10 mV (a), when the cell was depolarized further (to +30 mV), the effect of 5-HT on the amplitude was much reduced.

Specificity and Localization of a 5-HT Receptor Modulation by 5-HT could best be demonstrated by using a hyperpolarizing prepulse of approximately -45 mV, which just fails to remove significantly inactivation of IA in the unmodulated state, while completely removing inactivation of IKs,which thus dominates the current elicited by a depolarizing step to +10 mV (Figure 3a). Following application of 5-HT, the fast transient Shaker component is progressively revealed (30, 60, and 90 s) as the voltage dependence of inactivation shifts to more positive values. In addition, although modulation is not established to its full extent (compare upper right and lower left traces in Figure 2a), the steady-state current due to the delayed rectifier is already slightly reduced owing to a shift in the voltage dependence of activation (see below):As with the other applications, the effect was easily reversed by washing out the 5-HT (Figure 3a). A similar protocol was used to test the specificity of the modulation. We applied methysergide, a specific, competitive antagonist of 5-HT receptors effective, for example, in biochemical preparations of

cloned 5-HT receptors from Drosophila (Witz et al., 1990; Saud0u et al., 1992). Prior application of methysergide could prevent the modulation of the Shaker conductance by 5-HT and could also reverse the voltage shift when applied after 5-HT (Figure 3b). Similarly, serotonergic modulation of IKswas blocked or reversed by methysergide (data not shown). The specificity of the action of 5-HT was further supported by the inability of a variety of putative transmitter substances to induce any effects at concentrations up to 2 mM. Apart from acetylcholine, y-aminobutyric acid, and glutamate, these included the closely related biogenic amines octopamine and dopamine. Only the photoreceptor neurotransmitter histamine shifted the voltage range in some experiments up to +9 mV (in 5/19 cells tested; noticeable only for the Shaker conductance). This effect was found only with high concentrations (>2 mM) and was not investigated further. To see whether the modulation was mediated by receptors on the photoreceptor cells themselves, or by interneurons in the first optic ganglion, the lamina, we used the dissociated ommatidia preparation (Hardie et al., 1991 ; Hardie, 1991). This preparation provides intact units of 8 photoreceptor cells (ommatidia) freed from any adjacent tissue and lacking the receptor axons and terminals. in this preparation, 5-HT induced reversible shifts in voltage dependence in both IA and IK, similar in magnitude to those observed in the intact preparation, although in only about 45% of cells tested (12/27 cells; data not shown). Since no additional neurons were present in this preparation, we can exclude the possibilityof other cells mediating the effect. Thus, it is reasonable to assume that native 5-HT receptors on the photoreceptor cells can mediate the shift in voltage dependency.

Neuron 848

a

11000pA

Control

C~0ms

~-30

with 5-HT

~-20

...... 1

+30 hiM

+20

-100

-100 12000 pA

_

b ~

+1o

W( _ ~ - ~ - " ~ - ~

+80

+80

s~c~ abs,currents~

i i i • ". Inactivation =j .,,= . . ~ Control~

t2\,

0.2 o.o

-80

\~ ~,. Control '~'~l,..~,_ . . . . .

-60 -40 -20 0 20 prepulse voltage (mV)

Activation 1,.o •

g

to, 1

Control ,~ ~; ~ J J r

t°'2 . . . . .

jo.o

-40 -20 0 20 40 60 command voltage (mV)

Figure 4. Modulation of the Delayed Rectifier Conductance IK, in the Mutant S h Ks'33 The S h a k e r mutant ShKS~s3 lacks any functional A-channels, leaving the delayed rectifier IK, in isolation, (a) Inactivation protocols with a constant command voltage of +30 mV revealed a shift of the voltage dependence on the order of 10 mV for the inactivation voltage dependence. Stepping to +30 mV left the current amplitudes of the isolated IK, much less affected than observed previously (Figure 2a), indicating a similar shift of the activation voltage dependence. (b) Activation protocols of the isolated IK, show a similar shift of about +14 mV for the activation voltage dependence. When stepping to +10 mV (arrows), 5-HT reduced the amount of channels activated by about 60°/0, confirming that the reduced amplitude of IK, (Figure 2b) is caused by a shift of the activation voltage dependence; however, at higher voltages the amplitude was not reduced• (c; left) Peak currents as recorded in (a) were corrected for series resistance errors and normalized. These data (symbols represent the data of 8 cells) were in many cases best fitted with the weighted sum of two Boltzmann distributions (dotted lines) according to equation 2 accounting for various amounts of the fast delayed rectifier channels (IKf). The voltage dependence of IKs is characterized by a V50inacL Of --25.7 mV with a slope of 6.4 and is shifted in the presence of 5-HT by about +10 mV (Vsoi..~t = -15.7; slope = 7.2; Table 1). The continuous line represents a Boltzmann distribution fitted to these values. (Right) The voltage dependence of the activation was determined from peak currents as in (b); they were converted to cord conductance assuming Erov = - 8 5 mV. In the absence of 5-HT, single Boltzman distributions (equation 1) described the voltage dependence by Vs0~o,.= -1.0 with slope ~- 9.1. This is shifted by 5 - H T to about +12.7 mV with slope = -9.5 (Table 1), The inset shows the absolute values obtained in two different preparations; the maximal current amplitude was not altered when stepping to saturating voltages.

Quantification of Voltage Dependence The Delayed Rectifier.Like Current IKs

To isolate the effects of 5-HT on the voltage-dependent characteristics of the delayed rectifier, we used the Shaker mutant Sh Ksl~3, which lacks any functional Shaker chan-

nels. Stepping to a command voltage of +30 mV from various prepulse potentials confirmed that the inactivation voltage dependence of IK. was shifted in the order of 10 mV to more positive potentials (Figure 4a). Although the amplitude in the presence of 5-HT was still slightly reduced compared with the control, stepping to increasing depolarizing voltages revealed that the 5-HT-induced modulation did not reduce the maximal current (peak amplitude with 5-HT = 103.2% _+ 5.3% of control [n -- 10]; see also Figure 4b and Figure 4c, inset). Instead, the voltage dependence of activation is similarly shifted by about +14 mV to more positive potentials. As indicated by the arrows in Figure 4b, this accounted for the - 6 0 % reduction in amplitude observed in Figure 2a. By fitting the peak amplitudes with Boltzmann distributions, we obtained the characteristic voltages for the half-maximal activation (Vs0act) and steady-state inactivation (Vso~,aot.)as well as the slope of the voltage dependence (Figure 4c; see also Table 1). Before the application of 5-HT, the inactivation is described by a Vs0~naot.of -25.7 mV (slope = 6.5), shifted by 5-HT to -15.7 mV (slope = 7.2); similarly, the Vsoaot. is shifted from -1.0 mV (slope = -9.1) to +12.6 mV (slope = -9.5). The intermediate, fast delayed rectifier (IKf) is difficult tO separate from IKs;limited data obtained in only 3 cells suggest that the voltage dependence of this conductance is also shifted by approximately +20 mV (V50inact. from approximately -68 mV to -49 mY; data not shown). The Shaker Current IA To quantitate the effects of 5-HT on the Shaker channels, we attempted to isolate the conductance by applying 200 I~M quinidine to the bath. As shown in Figure 5a, this reduced the delayed rectifier currents to almost negligible proportions (<10%; Singh and Wu, 1989; Hardie, 1991), whereas the voltage-dependent characteristics as well as the 5-HT-induced shift of IAboth appeared unaffected (Figure 5a). These recordings showed that the voltage dependence of activation is shifted in a way similar to the removal of inactivation and confirmed that the maximal amplitude remained unaffected when jumping to saturating depolarizing voltages (+40 mV; Figure 5b; Figure 6c). Again, the characteristic voltages for the half-maximal activation (Vs0act.)and steady-state inactivation (Vso~°ao,.)as well as the slope of the voltage dependence were obtained by fitting the peak amplitudes with Boltzmann distributions (Figures 6a and 6b). Before application of 5-HT, the voltage dependence is described by a Vso of -55.3 mV for the steady-state inactivation and -23.7 mV for activation (Table 1). These values are - 1 0 mV more positive than has been previously described in Drosophila photoreceptors (Hardie, 1991), reflecting the absence of significant run-down under the improved recording conditions (see Figure 6d). The application of 5-HT shifted V~o~,aot•of Shaker by about +28 mV to values of -27.1 mV (Figure 6a). A similar, symmetrical shift from -23.7 to +11.7 mV is observed for Vsoa~t(Figure 6b). Interestingly, these values obtained in the presence of 5-HT now resemble those reported for Drosophila muscles (Vso~,aot.= -27 mV: Wu and Haugland, 1985; -29 mV: Zagotta and Aldrich, 1990a) and could not be shifted any further by subsequent 5-HT application or prolonged application times.

Serotonin Modulates Shaker Channels 849

a

Figure 5. IsolationoftheShakerConductance and Its Modulation in the Presence of Quinidine

Control

~

- 50

,

-4oI

lOO0pA

+ 5-HT

-2o, +50mY

+50mV

I

+50mY

I--

-100

-100

-90 -60

a

Wash

]~om ~

~

-10(I

Wash

-100

Control

I

~/

0

-100

b

~

[ ~ ms

+10mY

0

200 pM quinidine

/

1000 pA

with 5-HT

(a) Quinidine (200 I~M) applied to the bath in the semi-intact preparation reduces delayed rectifier currents by over 90%, allowing the Shakerconductance to be studied in near isolation. Under these conditions, 5-HT was equally capable of shifting the voltage dependence, and it became obvious that inactivation was already removed at voltages as high as -20 mV. The effect readily reversed when 5-HT was washed out. (b) Voltage dependence of activation for the isolated Shaker conductance. From a holding potential of -100 mV (which removes all inactivation), the command voltage was stepped from -90 to +50 mV in 10 mV steps. Threshold for activation shifted from approximately -40 mV (control) to approximately -20 mV in the presence of 5-HT. The maximal amplitude, however, was nearly identical, and the effect reversed when 5-HT was washed out. A slow, maintained component was apparently enhanced in the presence of 5-HT with higher command voltages. This was not Shakerspecific but also present in Shakernull mutants and appears to be due to a partial relief of the quinidine block of delayed rectifier currents by 5-HT.

-.30

0

30

60

Activation

b

Figure 6. The Voltage Dependence of the Activation and Steady-State Inactivation

(a) Data from inactivation protocols as in Figure 5a; peak currents were corrected for series resistance errors and normalized. The data for 0.8 0.8 8 cells from four preparations are shown (small symbols). These data were best fitted with the E 0.6 0.6 weighted sum of two Boltzmann distributions (dotted lines) according to equation 2. A distri0.4 0.4 "~"~¢:~ ~ + 5-HT bution fitted to the average values obtained 0.2 0.2 from 23 cells is represented by the continuous line. Its main component is characterized by a 0.0 , ~ , , , , , 0.0 Vs0~a:t.of -55.3 mV with a slope of 3.9 and is 0 20 -80 -60 -40 -20 0 20 40 60 - 100 -80 -60 -40 -20 shifted in the presence of 5-HT by about +28 command voltage (mV) prepulse voltage (mV) mV (Vsoi,.= = -27.1; slope = 4.3). The second component is similarlyshifted (by +22 mV, from Time course C V%~,act = -74.8, s' = 10.7 to V's0~°~t. = -52.9, [ (pA) s ' = 10.5). -20 1raM 3000 / (b) The voltage dependence of the activation determined from traces as in Figure 5b; peak > -30 g currents were converted to cord conductances 2000 assuming E,~v = -85 mV. Recordings were ~ -40 also performed with elevated potassium con' + 5-HT centrations to give more accurate values below o -50 -50 mV (otherwise too close to EK). Under con5OO trol conditions single Boltzman distributions -60 (equation 1) described the voltage dependence -80-60-40-20 0 20 40 with a Vsoa=.of -23.7 and a slope of -12.8. The 5 10 15 20 25 30 35 command voltage (mV) application of 5-HT induced a positive shift to time ( m i n . ) about +11.7 mV and slope = -12.9. (Inset) A semi-logarithmic plot of data obtained with elevated potassium concentrations revealed a systematic aberration of the data at smaller voltages. This would indicate a component already activated at more negative voltages and probably represents the equivalent of the second component observed for the steady-state inactivation. (c) Absolute current-voltage relations of peak current in the absence (Control) and presence of 5-HT for 2 different cells. The maximum current values (/>+40 mV) are unaffected by 5-HT. (d) The time course of changes in voltage dependence (Vs0~nact).Under control conditions, Vs0 values were normally stable over at least 20 min, and any drift was always to slightly more negative values. Application of 1 mM 5-HT caused a pronounced and rapid shift within the time of the first measurement ( - 3 0 - 6 0 s). A similar but more slowly developing shift was induced by inclusion of GTPyS (300 p.M) in the pipette solution. 1.0

T

d

1.0

Neuron 860

Table 1. Propertiesof PotassiumCurrents in Wild Type and Shaker Mutants Vso(mY)

Shaker wild type Activation Inactivation Second Comp. Shaker mutants eag S h E62

T(1;Y)W32 Sh5 Delayed rectifier IK, Activation Inactivation

Slope

Control (n)

With 5-HT (n)

Control

With 5-HT

-23.7 -+ 3.5 (24) -55.3 ± 2.2 (23) -74.8 ± 6.7 (10)

+11.7 ± 2.9 (15) -27.1 ± 1.8 (18) -52.9 ± 4.2 (10)

-12.8 ± 2.6 +3.9 ± 0.24 +10.7 ± 1.2

-12.9 ± 1.8 +4.3 ± 0.25 +10.5 ± 1.7

-55.1 ± -54.6 ± =-54:8 ± -48.4 ±

-28.2 -26.6 -26.9 -24.4

+3.77 ± +3.85 ± +3.65 ± +4.21 ±

+4.36 +4.41 +4.05 +4.62

1.42 (6) 1.33 (8) 0~82(9) 3.76 (16)

-1.0 __. 1.48(17) -25.7 ± 1.1 (17)

± ± ± ±

0.56 (5) 0.84 (7) 0.73 (8) 1.32 (9)

+12.7 ± 1.75(17) -15.7 __. 1.12 (17)

0.58 0.48 0.52 0.37

-9.1 ± 0.64 +6.4 __. 0.40

± ± ± ±

0.29 0.55 0.23 0.30

-9.5 ± 0.53 +7.2 __. 0.54

The activationdataobtainedin the absenceor presenceof 5-HTwere bestfittedwith.asingle'BoltZmandistribution(equation1),with the half-maximal activation given by Vs0and the slope of the voltage dependencegiven by s. The inactivationdata were best fitted with the sum of two Boltzmann distributions(equation2; see ExperimentalProcedures).Vsoand s values for the major componentonly are presented.Eor the Shaker mutants, only the values for inactivationare shown. All data presentedas means ± SD with the number of cells (n) in brackets.

Regardless of the preceding recording time, 5-HT was able to induce rapid shifts that could readily be repeated after the effect had been washed out (Figure 6d). In the absence of 5-HT, however, the voltage-dependent characteristics were stable over typical recording times of about 15-20 min. As a preliminary attempt to characterize the second messenger system responsible for modulation, we applied GTP~,S within the electrode (300 ~M) and found that this produced a similar shift in the voltage operating range over a time course of several minutes (Figure 6d) in all 6 cells tested. Again, this was not restricted to the Shaker channel, but also mimicked the 5-HT-induced shift of IKs (data not shown). We also evaluated the maximal peak current and the voltage dependence of the time to peak of the macroscopic Shaker current. Although time to peak values were shifted in the direction expected from the other voltage-dependent parameters, values at higher command voltages showed a pronounced additional deviation, with minimal times to peak increased from 1.1 _.+ 0.15 ms (n = 12) to 1.9 __+ 0.13 ms (n -- 9). The maximal peak currents remained unaffected in the presence of 5-HT (97.4% _+ 5.9% in respect to controls; n = 10). Further evaluation of the fast macroscopic inactivation was impeded by a varying amount of delayed rectifier currents revealed in the presence of 5 - H T (see also Experimental Procedures).

Mutant Analysis We studied a variety of mutants known to affect Shaker potassium conductances in Drosophila. The Shaker gene is capable of generating at least ten different transcripts (each encoding one subunit of a putative tetrameric channel) by alternative splicing (Iverson et al., 1988; Kamb et al., 1988; Pongs et al., 1988; Schwarz et al., 1988; Timpe et al., 1988a, 1988b). To test whether certain Shaker transcripts might be required for this modulation, we used the Sh E62and the T(1;Y)W32 mutants. As previously described (Hardie et al., 1991), in photoreceptors these both affect the macroscopic inactivation kinetics, but in opposite di-

rections (Figures 7c and 7d), indicating that the transcripts affected by these mutants (SHE62:class 1 ; T(1;Y)W32". ShB and ShD; see Experimental Procedures)(Hardie et al., 1991; Lichtinghagen et al., 1990) both contribute to the wild-type channel. Since the voltage dependence in both mutants was reversibly shifted just as readily as for the wild-type Shaker channels (Figures 7c and 7d), we conclude that these classes of transcript are not required for modulation. Recently, Zhong and Wu (1993) suggested that the eag gene product, which encodes a distinct potassium channel subunit with a putative cGMP binding domain (Warm ke et al., 1991 ; G uy et al., 1991 ), is responsible for mediating the modulation of various potassium channels, including Shaker, by cGMP or calcium/calmodulin. However, modulation of photoreceptor IA was not affected in mutants lacking the eag gene (Figure 7a). We also tested the mutant Sh 5, representing a point mutation in the $5 region (Lichtinghagen et al., 1990) that in muscle and oocytes shifts the voltage operating range to more positive values by about 5-10 mV (Wu and Haugland, 1985; Zagotta and Aldrich, 1990b; Baker and Salkoff, 1990; Gautam and Tanouye, 1990). We found that this mutation produced a similar positive shift in photoreceptors (Vso~,aot.now -48.4 _+ 3.7 mY, compared with -55.3 _ 2.2 mV in wild type). However, the Sh 5 mutation appeared to have no effect on the neuromodulatory action of 5-HT (Figure 5b), which still induced a positive shift to -24.4 _+ 1.32 mV (Table 1).

Discussion We have used a new semi-intact preparation of the Drosophila retina to study the in situ modulation of two classes of potassium channels, including a rapidly inactivating A-channel encoded by the Shaker gene. Although Shaker is probably the most intensively studied of all the ion channel genes, this represents the clearest demonstration of a physiologically implemented modulation of native Shaker channels (but see Zhong and Wu, 1993; Poulain et al.,

Serotonin Modulates Shaker Channels 851

a t

Control

I/~t

+ 5-HT ~

800 pA

b~ h " c

L~

/~_~o°~

300pA

Sh ~

- 50 - 20

d

800 pA

I " * " 3' 19-2116-18 H

~o~ 5,6 3

5' ~

2

I

1

Figure 7. The Modulation by 5-HT in Shaker-Related Mutants

Whole-cellcurrents elicitedby inactivationprotocols(as in Figurelb and Figure2a) in fourdifferentpotassiumchannelmutantsundercontrol conditions(left) and in the presenceof 1 mM 5-HT (right). In all mutants 5-HT induced a reversibleshift in voltage dependencythat was indistinguishablefrom that observedin wild type (seealsoTable 1). As shown in Figure3b and Figure 5b, the reduced amplitudeis explainedby a similar shift of the voltagedependenceof activation. Currents recordedin eag (a) appearidenticalto those from wild type (e.g., Figurelb; Figure2a),but noticethat the ShE62(c) and T(1;Y)W32 mutations (d), which excludecertain alternativesplicing variantsof the Shaker gene, both affect the time courseof Shaker inactivation. In the mutantSh5 (b) (andalso in T(1;Y)W32 [d]), inactivationappears more pronouncedowingto the lack of a fast delayedrectifierIKr(see Figurelc) normallycontributingsignificantlyto the maintainedcomponent on this time scale. (Inset)Organizationof the Shaker locus.A conservedcore regioncan be splicedto one of two alternative3' endsand one of five alternative 5' ends.ShE62is a pointmutationwith nonfunctionalclass 1 transcripts. In T(1;Y)W32 a chromosomalbreakpointexcludesall ShB and ShD transcripts at the 5' end. The Sh5 mutationis locatedwithin the core region itself. Terminologyaccording to Baumann et al. (1987) and Stocker at al. (1990).

1994). In addition, in contrast to virtually all previous reports on potassium channel modulation, we find that 5-HT affects the voltage operating range, shifting the operating range of a delayed rectifier-like channel by +10-14 mV, while the otherwise conspicuously negative voltage range of the photoreceptor S h a k e r channels is shifted by approximately +30 mV, thereby coinciding with values previously described from muscle recordings and heterologous expression systems. Specificity and Localization of 5-HT Action

In the present study, the only putative neurotransmitter able to modulate the voltage operating range consistently was 5-HT. The photoreceptor neurotransmitter histamine

(Hardie, 1987) also sometimes produced a small shift that might be explained by the structural similarities with serotonin and cross-reactivity at the receptor site (Berridge, 1972). Other transmitters believed to be present in the optic lobes failed to induce any noticeable effects. The identity of 5-HT as the natural agonist is further supported by the presence of 5-HT-immunoreactive neurons in the distal lamina (N&ssel, 1987) and the fact that the 5-HTinduced shift could be blocked by the specific 5-HT antagonist methysergide. A possible functional role is suggested by our recent finding that locust photoreceptors switch from a delayed rectifier-like potassium conductance during the day to a transient A-type potassium conductance at night. This endogenous, diurnal modulation is mimicked by application of 5-HT to the retina of the intact animal (Cuttle et al., 1995) and is consistent with our present understanding of potassium channel function in photoreceptors. Whereas a maintained delayed rectifier-like conductance reduces the membrane time constant and enhances the frequency response of the cells (Laughlin and WeckstrSm, 1993), a transient A-type conductance would still oppose transient depolarizations but maintain the high membrane time constant. Although our evidence indicates putative 5-HT receptors on the photoreceptor cells themselves, their exact localization remains unknown. The most likely serotonergic input to the photoreceptors comes from two 5-HT-immu noreactive neurons with widespread aborizations between the photoreceptor cell bodies and synaptic terminals in the lamina (N&ssel, 1987). Possibly, the receptors might be concentrated in the regions of the axons, leaving the cell bodies with variable amounts of receptors, or none at all, which might explain the reduced success rate in the dissociated preparation (<50%). Damage to the receptor or underlying messenger system during the dissociation appears less likely because all dissociated cells tested had low resting potentials and generated elementary light responses (bumps), characteristics that are most sensitive to any damage. A remote receptor site on the axon also receives some support from the latency of the 5-HT effect, which was greatly reduced after the use of enzyme treatment, consistent with the existence of a diffusion barrier. Modulation of the Voltage D e p e n d e n c e

Modulation of potassium channels is a widespread and well described phenomenon; however, so far it has appeared largely restricted to up- and down-regulation (reviewed in Levitan, 1994; Timpe and Fantl, 1994; White et al., 1994; Poulain et al., 1994; Esguerra et al., 1994). Arguably the most closely related example is the 5-HTinduced modulation of potassium channels in Hermissenda photoreceptors, widely cited as a model system for neuronal plasticity and learning (Alkon et al., 1985). Despite recent detailed studies of molluscan type-B photoreceptors describing at least three different kinds of potassium currents affected by 5-HT, including a delayed rectifier-like current, a transient "A"-potassium current, and an inward rectifier (Acosta-Urquidi and Crow, 1993), as well as further reports on the serotonergic modulation of potassium channels in Aplysia sensory neurons (Kandel and

Neuron 852

Schwartz, 1982; White et al., 1994), effects on voltage dependence have not been reported in these preparations. To our knowledge only two convincing demonstrations of a shift in voltage dependency in potassium channels have previously been reported in situ. Atkins et al. (1991) reported that muscarinic agonists induced a negative shift (-13 mV) in the voltage operating range of a transient potassium channel in rat neostriatal neurons, while Perozo et al. (1989, 1991a, 1991b) reported a positive shift (+1016 mV) for the delayed rectifiercurrent in squid giant axons induced by a PKA-dependent phosphorylation. In addition, Moran et al. (1991) reported that phorbol esters could induce a small shift ( - 3-4 mV) in Shaker H4 channels heterologously expressed in oocytes. Although very weak, this effect was the first indication that Shaker channels might be subject to modulation. Interestingly, modulation of voltage dependency has been more widely reported among a variety of other channel types (voltage-gated calcium channels: Bean, 1989; Elmslie et al., 1990; Lopez and Brown, 1991; calcium-activated potassium channels: Esguerra et al., 1994; heart pacemaker if: DiFrancesco and Mangoni, 1994; muscarinic metabotropic cation channels: Zholos and Bolton, 1994), suggesting that this mode of regulation may be more widespread than is commonly thought. Molecular Basis of Modulation

The present paper provides further evidence that the voltage operating range of potassium channels is a physiological target of modulation and raises the question of how this crucial functional property might be regulated in situ. The electric field sensed by the channel's voltage sensor is influenced by local surface charge (for review, see Hille, 1992), which is determined both by the distribution of charged amino acids (including any additional charged residues such as phosphates) within the channel protein itself and by other closely associated membrane proteins or charged headgroups of surrounding lipide (e.g., Ji et al., 1993; see also Lopez and Brown, 1991; Mayer and Sugiyama, 1988; MacKinnon et al., 1989). The observation that the voltage dependence of two and probably three different classes of potassium channels is shifted, albeit by different amounts, raises the possibility of a common underlying mechanism. In this respect at least two general possibilities may be envisaged. First, the modulatory site might be located on an accessory protein or channel subunit coded by a separate gene such as eag. Although eag forms functional channels when expressed alone in Xenopus oocytes (Br~iggemann et al., 1993), allele-specific interactions suggest that, in situ, eag may coassemble as an integral "promiscuous" subunit, required for calcium/ calmodulin- and cGMP-dependent modulation of different potassium channels in Drosophila muscle (Zhong and Wu, 1993). Although we failed to find any influence of the eag mutation on the photoreceptor potassium channels or their modulation by 5-HT, this does not exclude the possible involvement of homologous proteins, such as the eag-like potassium channel (elk; Warmke and Ganetzky, 1994). Second, a more general effect on the membrane surface

charge could be suggested, and in particular a net increase in negative surface charge to account for the large positive shift in voltage dependence observed. A precedent is suggested by negative shifts in the voltage dependence of sodium and calcium currents caused by arachidonic acid, which may be due to the screening of negative charges by rises in cytosolic calcium (Meves, 1994). However, (un)screening by divalent ions seems an unlikely candidate in the photoreceptors, since buffering internal free calcium by BAPTA or EGTA (5-10 raM) in the 10 nM to 5 I~M range (in the presence and absence of external calcium) as well as removal of external and internal magnesium had only minor effects on the voltage dependence of the photoreceptor potassium channels (<5-10 mV shifts in IA) or on their ability to be modulated by 5-HT (W. H. and R. C. H., unpublished data). Another mechanism for altering local charge distribution is phosphorylation of the channel itself. This has been reported to influence the voltage operating range of IK in squid giant axons (Perozo et al., 1989, 1991a, 1991b), probably by electrostatic interactions in close vicinity to the voltage-sensing region (Perozo and Bezanilla, 1990). The sequence of the channels responsible for photoreceptor IKs is unknown; however, the Shaker gene sequence is reported to include only three putative phosphorylation sites: one, on the N-terminal (Drain et al., 1994), is excluded by the T(1; Y)W32 breakpoint (Papazian et al., 1987; Baumann et al., 1989) and presumably cannot account for modulation of photoreceptor IA, which is unaffected in the T(1;Y)W32 mutant (Figure 7). Like this site, the other two phosphorylation sites (one on the C-terminal [Schwarz et al., 1988; Tempel et al., 1987; Pongs et al., 1988] and one on the loop joining $4 and $5 [Pongs et al., 1988]) are serine residues and potential cAMP-dependent phosphorylation sites; however, preliminary experiments with cyclic nucleotides and their analogs revealed no influence on the photoreceptor Shaker channels or their ability to be modulated by 5-HT (W. H. and R. C. H., unpublished data). In all voltage-gated channels, the voltage dependence is assumed to reside within the $4-$5 transmembrane helices, where site-directed mutagenesis can profoundly influence voltage operating range and related characteristics (Papazian et al., 1991; Lopez et al., 1991; McCormack et al., 1991). However, as membrane-spanning domains, the $4-$5 helices themselves are presumably not directly accessible to physiological modifications, and it remains a challenge to understand how a presumably remote modulatory site (or subunit) can so profoundly affect the channel's voltage sensitivity. Conclusion

Modulation of potassium channels is a widespread phenomenon underlying many aspects of short- and long-term plasticity in the nervous system. Despite spectacular progress in the understanding of the structure-function relationships of potassium channels, this has thus far been largely restricted to basic biophysical properties such as ion permeation, gating, and subunit assembly. By contrast, molecular mechanisms underlying modulation are

Serotonin Modulates Shaker Channels 853

l a r g e l y u n k n o w n , a l t h o u g h virtually e v e r y k n o w n m e s s e n ger pathway has been implicated. The demonstration and c h a r a c t e r i z a t i o n o f m o d u l a t i o n in p h o t o r e c e p t o r potassium c h a n n e l s p r o v i d e s a rare o p p o r t u n i t y to s t u d y m o d u lation o f a c l o n e d p o t a s s i u m c h a n n e l in situ. In particular, the possibility o f g e n e t i c

manipulation,

including

site-

d i r e c t e d m u t a g e n e s i s in t r a n s g e n i c flies, will a l l o w the mol e c u l a r b a s i s o f m o d u l a t i o n to be a d d r e s s e d in situ, while

our d e t a i l e d k n o w l e d g e o f s i g n a l i n g in D i p t e r a n p h o t o r e c e p t o r s ( H a r d i e , 1985; L a u g h l i n a n d W e c k s t r S m , 1993) can p r o v i d e f u r t h e r insight into the f u n c t i o n a l s i g n i f i c a n c e o f p o t a s s i u m c h a n n e l m o d u l a t i o n in n e u r a l signaling.

Experimental Procedures Fly Stocks Most experiments were performed on the white-eyed (w) strain Oregon R. However, for some recordings we used the mutant norpA P24(w), in which the light response is abolished owing to the lack of the photoreceptor-specific phospholipase C (Bloomquist et al., 1988). We also studied two mutations that affect specific splicing alternatives of the Shaker gene (see Figure 7): Sh E~2and T(1;Y)W32. Owing to a chromosomal breakpoint, the mutant T(1;Y)W32 excludes all ShB and ShD transcripts at the 5' end of the Shaker gene (Papazian et al., 1987; Baumann et al., 1987), whereas Sh E62is a point mutation resulting in nonfunctional class 1 transcripts at the 3'end (Figure 7) (Lichtinghagen et al., 1990; for nomenclature, see Baumann et al., 1987; Stocker et al., 1990). In addition, we tested In(1)sc 29 wa, which is a null allele of the eag gene that represents an additional potassium channel subunit affecting delayed rectifier as well as Shaker currents in muscle (Wu et at., 1983; Drysdale et al., 1991; Zhong and Wu, 1991, 1993). Finally, we used f5 Sh s, representing a point mutation within the $5 region and shifting the voltage dependence of Shaker channels in muscles by about 5-10 mV (Wu and Haugland, 1985; Baker and Salkoff, 1990; Gautam and Tanouye, 1990; Salkoff and Wyman, 1981 ; Lichtinghagen et al., 1990). To isolate the delayed rectifier IK,, we used the Shaker mutant Sh Ks,~3, a missense mutation in the core region resulting in nonfunctional Shaker channels (Lichtinghagen et al., 1990).

Preparation Animals were reared at - 2 4 ° C under a 12 hr day/night cycle. The preparation was adapted from Hardie et al. (1991); in short, adult Drosophila up to 3 hr after emergence were immobilized by cooling and decapitated in chilled Ringer solution. The eyes were removed with a razor chip, and the retinae were dissected out carefully and transferred to 2 mg/ml collagenase in Ca2÷/Mg2+-free Ringer solution at 22°C for 3-4 min. They were then gently triturated using fire-polished glass pipettes in enzyme-free Ringer solution supplemented with 10% fetal calf serum until the primary pigment cells dissociated. For the semi-intact preparation, the retina with at least the lamina still attached was fixed under an upright microscope with a glass suction pipette mounted in a micromanipulator (Figure la). The ommatidia were now accessible to patch-clamp recordings. For the dissociated preparation, retinae were further triturated until individual ommatidia dissociated. These were transferred to a microscope chamber, the bottom of which was formed by a coverslip. The bath was perfused with fresh Ringer solution and circulated by a slight flow of oxygen applied over the surface of the solution.

Solutions Standard Ringer solution contained 120 mM NaCI, 5 mM KCI, 8 mM MgSO4, 1.5 mM CaCI2, and 10 mM N-Tris-(hydroxymethyl)-methyl-2amino-ethanesulphonic acid (TES; pH 7.15). The preparation itself was performed in Ca2+/Mg2+-free Ringer solution supplemented with 20 mM sucrose and 10% fetal calf serum, whereas the final bath solution contained in addition 20 mM proline and 5 mM alanine. To evaluate the voltage dependence near the potassium equilibrium potential, recordings were performed in high potassium Ringer solution (50 mM KCI, 75 mM NaCI or 125 mM KCI instead of NaCI). The electrode solution contained 135 mM potassium gluconate, 2 mM MgCI2,

10 mM TES, 2 mM Mg-ATP, and 0.4 mM GTP; this induced a junction potential of -10.8 mV, which was corrected for arithmetically. Most recordings were performed without any EGTA/calcium, although the free calcium concentration in earlier recordings was adjusted with 5 mM EGTA to a final concentration of 20 nM (0.68 mM CaCI2). GTPTS (300 I~M; Sigma) was applied via the recording electrode. Various putative neurotransmitters (reviewed in N~ssel, 1991; Hardie, 1989) were applied within the bath in concentrations of 0.1-2 raM. In earlier preparations performed without enzyme treatment, 1 mM 5-HT had noticeable effects only after at least 8 rain; however, following enzyme treatment this time was much reduced (<3 rain). Although concentrations around 40 p.M already shifted the voltage dependence, these shifts occurred over a slower time scale (4-6 mV shift over 10 min). We therefore routinely used a higher concentration of 5-HT (1 mM), which induced the maximal inducible shift within reasonable recording times.

Electrophysiology Recordings Recordings were made at a constant room temperature of 22°C ( _ 1°C) under dim red light on the stage of an upright, fixed-stage microscope (M2, Oxford Instruments, UK). Patch pipettes were pulled from fiber-filled borosilicate glass ( - 1 ram; Clark Electromedical Co.) and had resistances of 4-10 M~. Whole-cell recordings were achieved as previously described (Hardie et al., 1991; Hardie, 1991). Signals were amplified using an Axopatch-1 D patch-clamp amplifier (Axon Instruments) and filtered by a 4 pole Bessel filter (10 kHz). Data were sampled and analyzed using pCLAMP 5.5 software in conjunction with the Labmaster-TL1 interface (Axon Instruments). Clamp Quality Seal resistances were typically 10 GQ;the input resistance of the cells was in the range of 500 M~ to 1 G~. Assuming the photoreceptor to be 100 p.m long and 4 I~m in diameter, input resistances of/>500 MQ give length constants of more than 1 mm. Although this should ensure adequate space clamp under most conditions, there will inevitably be some deterioration with the largest currents evoked in this study (e.g., length constant during 2 nA current estimated to fall to - 3 2 0 ~m). The series resistances (typically 8-16 MQ) were monitored repeatedly during the recordings and 85%-90% compensation was applied. The linear leak fraction was estimated from small hyperpolarizing voltage steps (below - 8 0 mV) and subtracted using the software facilities of pCLAMP. Evaluation of the Voltage.Dependent Conductences Measurements were started at about 2 rain after establishing the whole-cell configuration, by which time any initial small drift due to equilibration of cell contents had largely stabilized (Figure 6d). To determine the voltage dependence of the steady-state inactivation, we evaluated peak currents during a command step to -10, +10, or +30 mV after decreasing hyperpolarizing prepulses of 1 s duration. For activation, peak values were taken during increasing depolarizing command voltages from a holding potential of -100 mV and converted to cord conductances, assuming a reversal potential of -85 mV (Hardie, 1991; measured from tail currents and found not to be altered by 5-HT); calculated this way, the voltage dependence of activation showed no difference from data obtained from tail currents. In all cases the command voltage was corrected for the remaining series resistance error by subtracting the voltage drop across the series resistance ( - 4-8 mV for the largest currents). To exclude the possibility of deteriorating clamp condition with large currents compromising our results, we compared steady-state inactivation data obtained using different command voltages (-10, +10, and +30 mV) as well as elevated external potassium (which reduces the driving force for potassium); however, varying the total current in this way did not significantly affect the estimated voltage dependencies (<2 mV). In addition, we note that the slope of the steady-state inactivation curve of the Shakercurrents, a parameter particularly sensitive to imperfect voltage-clamp conditions, wasverysimilar(slope = 3.9)tothatdeterminedfromShakerchannels in single-channel recordings in which clamp control is essentially perfect (e.g., Zagotta and Aldrich, 1990a). All data were corrected for series resistance errors, normalized, and fitted using least-square fit routines from Origin (Microcal). Activation data were fitted with a Boltzmann distribution:

Neuron 854

g/g,,,= = a/J1 + exp[(Vsoa=- V)/S]/

(1)

of single K÷ channel function by distinct second messengers. Trends Neurosci. 11,232-238.

where Vsoa=.is the half-maximal activation and s is the slope of the voltage dependence. The inactivation data were best fitted with the weighted sum of two Boltzmann distributions:

Berridge, M. J. (1972). The mode of action of 5-hydroxytryptamine. J. Exp. Biol. 56, 311-321. Bezanilla, F., and Stefani, E. (1994). Voltage-dependent gating of ionic channels. Annu. Rev. Biophys. Biomol. Struct. 23, 819-846. Bloomquist, B. T., Shortridge, R. D., Schneuwly, S., Perdew, M., Montell, C., Stellar, H., Rubin, G., and Pak, W. (1988). Isolation of a putative phospholipase C gene of Drosophila norpA, and its role in phototransduction. Cell 54, 723-733. Bosma, M. M., Allen, M. L., Martin, T. M., and Tempel, B. L. (1993). PKA-dependent regulation of mKv1.1, a mouse Shaker-like potassium channel gene, when stably expressed in CHO cells. J. Neurosci. 13, 5242-5250. Br0ggemann, A., Pardo, L. A., Stuehmer, W., and Pongs, O. (1993). Ether-&-go-go encodes a voltage-gated channel permeable to K÷ and Ca2÷ and modulated by cAMP. Nature 365, 445-448.

g / g ~ = a/[1 + exp[(V.......

--

V)/s]t + bill + exp[(V%,°a¢,.- V)/sl/(2)

Vsoinact. is the voltage at which half of the steady-state inactivation was removed. For IA the major component represents - 8 0 % of the total current and has the more positive voltage dependency. The values for Vr5oinact. and s' describe a second component with a more negative voltage dependence (Figure 6a). Since both components were also observed with quinidine, but not in ShKs~ (data not shown), we assume there is a Shakersubpopulation characterized by more negative voltage dependence. Whether these two components represent different channels or different physiological states of the same channels, or are artificially caused by the preparation conditions remains open. In case of the delayed rectifier currents in S/'ds13~,the second Boltzmann component accounted for the intermediate fast delayed rectifier IKf, not necessarily present in every cell. where

Acknowledgments We are thankful to Dr. S. B. Laughlin for critical discussion and reading of the manuscript. Mutants were generously supplied by Professors W. Pak, O, Pongs, B. Ganetzky, and A. Ferrus. This work was supported by a rolling grant from the S. E. R. C (to R. C. H. and W. H.) and the Royal Society (to R. C. H.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 USC Section 1734 solely to indicate this fact. Received July 27, 1994; revised December 30, 1994.

References Acosta-Urquidi, J., and Crow, T. (1993). Differential modulation of voltage dependent currents in Hermissenda type B photoreceptors by serotonin. J. Neurophysiol. 70, 541-548. Aiyar, J., Grissmer, S., and Chandy, K. G. (1993). Full-length and truncated Kv1.3 K÷ channels are modulated by 5-HT(lc) receptor activation and independently by PKC. Am. J. Physiol. 265, C1571-C1578. Alkon, D. L., Sakabibara, N., Forman, R., Harrigan, J., Lederhendler, I., and Farley, J. (1985). Reduction of two voltage-dependent K÷ currents mediates retention of learned association. Behav. Neural Biol. 44, 278-300. AIkon, D. L., Amaral, D. G., Bear, M. F., Black, J., Carew, T. J., Cohen, N. J., Disterhoft, J. F., Eichenbaum, H., Golski, S., and Gorman, L. K., (1991). Learning and memory. Brain Res. Rev. 16, 193-220. Atkins, P. T., Surmeier, D. J., and Kitai, S. T. (1991). Muscarinic modulation of a transient K÷ conductance in rat neostriatal neurons. Nature 344, 240-242. Baker, K., and Salkoff, L. (1990). The Drosophila Shaker gene codes for a distinctive K÷ current in a subset of neurons. Neuron 4, 129-140. Baumann,. A., Krah-Jentgens, I., Muller, R., Muller-Holtkamp, F., Seidel, R., Kecskemethy, N., Casal, J., Ferrus, A., and Pongs, O. (1987). Molecular organization of the maternal effect region of the Shaker complex of Drosophila: characterization of an IA channel transcript with homology to vertebrate Na-channel. EMBO J. 6, 3419-3429. Baumann,. A., Krah-Jentgens, I., Muller, R., Muller-Holtkamp, F., Canal, J., Galceran, J., Ferrus, A., and Pongs, O. (1989). Molecular basis of biological diversity at the Shaker locus of Drosophila. In Molecular Biology of Neuroreceptors and Ion Channels, NATO ASI Series, Vol. 32, A. Maelicke, ed. (Heidelberg, Germany: Springer Verlag), pp. 231243. Bean, B. P. (1989). Neurotransmitter inhibition of neuronal calcium currents by changes in channel voltage dependence. Nature 340,153156. Belardetti, F., and Siegelbaum, S. A. (1988). Up- and down-modulation

Cuttle, M. F., Hevers, W., Laughlin, S. B., and Hardie, R. C. (1995). Diurnal modulation of photoreceptor potassium conductance in the locust. J. Comp. Physiol., in press. DiFrancesco, D., and Mangoni, M. (1994). Modulation of single hyperpolarization-activated channels (if) by cAMP in the rabbit sino-atrial node. J. Physiol. 474, 473-482. Drain, P., Dubin, A. E., and Aldrich, R. W. (1994). Regulation of Shaker K÷ channel inactivation gating by the cAMP-dependent protein kinase. Neuron 12, 1097-1109. Drysdale, R., Warmke, J., Kreber, R., and Ganetzky, B. (1991). Molecular characterization of eag: a gene affecting potassium channels in Drosophila melanogaster. Genetics 127, 497-505. Elmslie, K. S., Zhou, W., and Jones, S. W. (1990). LHRH and GTP-y-S modify calcium current activation in bullfrog sympathetic neurons. Neuron 5, 75-80. Esguerra, M., Wang, J., Foster, C. D., Adelman, J. P., North, R. A., and Levitan, I. B. (1994). Cloned Ca2+-dependent K* channel modulated by a functionally associated protein kinase. Nature 369, 563-565. Gautam, M., and Tanouye, M. A. (1990). Alteration of potassium channel gating: molecular analysis of the Drosophila Sh5 mutation. Neuron 5, 67-73. Guy, H. R., Durell, S. R., Warmke, J., Drysdale, R., and Ganetzky, B. (1991). Similarities in amino acid sequences of Drosophi/a eag and cyclic nucleotide-gated channels. Science 254, 730. Hardie, R. C. (1985). Functional organisation of the fly retina. Prog. Sens. Physiol. 5, 1-79. Hardie, R. C. (1987). Is histamine a neurotransmitter in insect photoreceptors? J. Comp. Physiol. (A) 161,201-213. Hardie, R. C. (1989). Neuretransmitters in compound eyes. In Facets of Vision, D. G. Stavenga and R. C. Hardie, eds. (Heidelberg, Germany: Springer Verlag), pp. 235-256. Hardie, R. C. (1991). Voltage-sensitive potassium channels in Drosophila photoreceptors. J. Neurosci. 11, 3079-3095. Hardie, R. C., Voss, D., Pongs, O., and Laughlin, S. B. (1991). Novel potassium channels encoded by the Shaker locus in Drosophila photoreceptors. Neuron 6, 477-486. Hille, B. (1992). Ionic channels of excitable membranes. (Sunderland, Massachusetts: Sinauer Association). Hoger, J. H., Walter, A. E., Vance, D., Yu, L., Lester, H. A., and Davidson, N. (1991). Modulation of a cloned mouse brain potassium channel. Neuron 6, 227-236. Homberg, U. (1994). Distribution of neurotransmitters in the insect brain. Prog. Zool. 40, 1-88. Hoshi, T., and Zagotta, W. N. (1993). Recent advances in the understanding of potassium channel function. Curt. Opin. Neurobiol. 3, 283290. Huang, X. Y., Morielli, A. D., and Peralta, E. G. (1994). Molecular basis of cardiac potassium channel stimulation by protein kinase A. Proc. Natl. Acad. Sci. USA 91,624-628. Iverson, L. E., Tanouye, M. A., Lester, H. A., Davidsen, N., and Rudy, B. (1988). A-type potassium channels expressed from Shaker locus

Serotonin Modulates Shaker Channels 855

cDNA. Proc. Natl. Acad. Sci. USA 85, 5723-5727. Jan, L. Y., and Jan, Y. N. (1992). Structural elements involved in specific K÷ channel functions. Annu. Rev. Physiol. 54, 537-555. Ji, S., Weiss, J. N., and Langer, G. A. (1993). Modulation of voltagedependent sodium and potassium currents by charged amphiphiles in cardiac ventricular myocytes. J. Gen. Physiol. 101,355-375. Kaczmarek, L. K., and Levitan, I. B. (1987). Neuromodulation: The Biochemical Control of Neuronal Excitability (New York: Oxford University Press). Kamb, A., Iverson, L. E., and Tanouye, M. A. (1987). Molecular characterization of Shaker, a Drosophila gene that encodes a potassium channel. Cell 50, 405-413. Kamb, A., Tseng-Crank, J., and Tanouye, M. A. (1988). Multiple products of the Drosophila Shakergene may contribute to potassium channel diversity. Neuron 1, 421-430. Kandel, E. R., and Schwartz, J. H. (1982). Molecular biology of learning: modulation of transmitter release. Science 218, 433-443. Laughlin, S. B., and Weckstr6m, M. (1993). Fast and slow photoreceptors: a comparative study of the functional diversity of coding and conductances in the Diptera. J. Comp. Physiol. (A) 172, 593-609. Levitan, I. B. (1994). Modulation of ion channels by protein phosphorylation and dephosphorylation. Annu. Rev. Physiol. 56, 193-212. Lichtinghagen, R., Stocker, M., Wittka, R., Boheim, G., St(Jhmer, W., Ferrus, A., and Pongs, O. (1990). Molecular basis of altered excitability in Shaker mutants of Drosophila melanogaster. EMBO J. 9, 43994407. Lopez, G. A., Jan, Y. N., and Jan, L. Y. (1991). Hydrophobic substitution mutations in the $4 sequence alter voltage-dependent gating in Shaker K÷ channels. Neuron 7, 327-336. Lopez, H. S., and Brown, A. M. (1991). Correlation between G protein activation and reblocking kinetics of Ca2+channel currents in rat sensory neurons. Neuron 7, 1061-1068. MacKinnon, R., Latorre, R., and Miller, C. (1989). Role of surface electrostatics in the operation of a high-conductance Ca2÷-activated K÷ channel. Biochemistry 28, 8092-8099. Mayer, M. L., and Sugiyama, K. (1988). A modulatory action of divalent cations on transient outward current in cultured rat sensory neurones. J. Physiol. 369, 417-433. McCormack, K., Tanouye, M. A., Iverson, L. E., Lin, J. W., Ramaswami, M., McCormack, T., Campanelli, J. T., Mathew, M. K., and Rudy, B. (1991). A role for hydrophobic residues in the voltage-dependent gating of Shaker K÷ channels. Proc. Natl. Acad. Sci. USA 88, 2931-2935. Meves, H. (1994). Modulation of ion channels by arachidonic acid. Prog. Neurobiol. 43, 175-186. Moran, O., Dascal, N., and Lotan, I. (1991). Modulation of a Shaker potassium A-channel by protein kinase C activation. FEBS Left. 279, 256-260. N~ssel, D. R. (1987). Serotonin and serotonin containing neurons in the nervous systems of insects. Prog. Neurobiol. 30, 1-85. N&ssel, D. R. (1991). Neurotransmitters and neuromodulators in the insect visual system. Prog. Neurobiol. 37, 179-254. Papazian, D. M., Schwarz, T. L., and Tempel, B. L. (1987). Cloning of genomic and complementary DNA from Shaker, a putative potassium channel gene from Drosophila. Science 237, 749-753. Papazian, D. M., Timpe, L. C., Jan, Y. N., and Jan, L. Y. (1991). Alteration of voltage-dependence of Shaker potassium channel by mutations in the $4 sequence. Nature 349, 305-310. Perozo, E., and Bezanilla, F. (1990). Phosphorylation affects voltage gating of the delayed rectifier K÷ channel by electrostatic interactions. Neuron 5, 685-690. Perozo, E., Bezanilla, F., and Dipolo, R. (1989). Modulation of K channels in dialyzed squid axons: ATP-mediated phosphorylation. J. Gen. Physiol. 93, 1195-1218. Perozo, E., Jong, D. S., and Bezanilla, F. (1991a). Single channel studies of the phosphorylation of K÷ channels in the squid giant axon. I1. Nonstationary conditions. J. Gen. Physiol. 98, 19-34. Perozo, E., Vandenberg, C. A., Jong, D. S., and Bezanilla, F. (1991b). Single channel studies of the phosphorylation of K ÷ channels in the

squid giant axon. I. Steady state conditions. J. Gen. Physiol. 98, 1-17. Pongs, O. (1992). Molecular biology of voltage-dependent potassium channels. Physiol. Rev. 72, $69-$88. Pongs, O., Kecskemethy, N., Muller, R., Krahjentgens, I., Baumann, A., Kiltz, H. H., Canal, I., Llamazares, S., and Ferrus, A. (1988). Shaker encodes a family of putative potassium channel proteins in the nervous system of Drosophila. EMBO J. 7, 1087-1096. Poulain, C., Ferrus, A., and Mallart, A. (1994). Modulation of type A K+ current in Drosophila larval muscle by internal Ca2+: effect of the overexpression of frequinin. PflSgers Arch. 427, 71-79. Rudy, B. (1988). Diversity and ubiquity of K channels. Neuroscience 25, 729-749. Salkoff, L, and Wyman, R. J. (1981). Genetic modification of potassium channels in Drosophila Shaker mutants. Nature 293, 228-230. Salkoff, L., and Wyman, R. J. (1983). Ion currents in Drosophila flight muscles. J. Physiol. 337, 687-709. Salkoff, L., Baker, K., Butler, A., Covarrubias, M., Pak, M. D., and Wei, A. (1992). An essential 'set' of K÷ channels conserved in flies, mice and humans. Trends Neurosci. 15, 161-166. Saudou, F., Boschert, U., Amlaiky, N., Plassat, J.-L., and Hen, R. (1992). A family of Drosophila serotonin receptors with distinct intracellular signalling properties and expression patterns. EMBO J. 11, 717. Schwarz, T. L., Tempel, B. L., Papazian, D. M., Jan, Y. N., and Yan, L. Y. (1988). Multiple potassium-channel components are produced by alternative splicing at the Shaker locus in Drosophila. Nature 331, 137-142. Schwarz, T. L., Papazian, D. M., Carretto, R. C., Jan, Y.-N., and Jan, L. Y (1990). Immunological characterization of K÷ channel components from the Shaker locus and differential distribution of splicing variants in Drosophila. Neuron 4, 119-127. Singh, S., and Wu, C.-F. (1989). Complete separation of four potassium currents in Drosophila. Neuron 2, 1325-1329. Stocker, M., Stuhmer, W., Wittka, R., Wang, X., Muller, R., Ferrus, A., and Pongs, O. (1990). Alternative Shakertranscripts express either rapidly inactivating or noninactivating K÷ channels. Proc. Natl. Acad. Sci. USA 87, 8903-8907. Tempel, B. L., Papazian, D. M., and Schwarz, T. L. (1987). Sequence of a probable potassium channel component encoded at Shaker locus of Drosophila. Science 237, 770-775. Timpe, L. C., and Fantl, W. J. (1994). Modulation of a voltage-activated potassium channel by peptide growth factor receptors. J. Neurosci. 14, 1195-1201. Timpe, L. C., Jan, Y. N., and Jan, L. Y. (1988a). Four cDNA clones from the Shaker locus of Drosophila induce kinetically distinct A-type potassium currents in Xenopus oocytes. Neuron 1,659-667. Timpe, L. C, Schwarz, T. L, Tempel, B. L, Papazian, D. M, Jan, Y. N, and Jan, L. Y. (1988b). Expression of functional potassium channels from Shaker cDNA in Xenopus oocytes. Nature 331, 143-145. Volterra, A., and Siegelbaum, S. A. (1988). Role of two different guanine nucleotide-binding proteins in the antagonistic modulation of the S-type K÷ channel by cAMP and arachidonic acid metabolites in Aplysia sensory neurons. Proc. Natl. Acad. Sci. USA 85, 7810-7814. Warmke, J. W., and Ganetzky, B. (1994). A family of potassium channel genes related to eag in Drosophila and mammals. Proc. Natl. Acad. Sci. USA 92, 3438-3442. Warmke, J., Drysdale, R., and Ganetzky, B. (1991). A distinct potassium channel polypeptide encoded by the Drosophila eag locus. Science 252, 1560-1562. Wei, A., Covarrubias, M., Butler, A., Baker, K., Pak, M., and Salkoff, L. (1990). K÷ current diversity is produced by an extended gene family conserved in Drosophila and mouse. Science 248, 599-603. White, J. A., Baxter, D. A., and Byrne, J. H. (1994). Analysis of the modulation by serotonin of a voltage-dependent potassium current in sensory neurons of Aplysia. Biophys. J. 66, 710-718. Witz, P., Amlaiky, N., Plassat, J.-L., Maroteaux, L., Borrelli, E., and Hen, R. (1990). Cloning and characterization of a Drosophila serotonin

Neuron 856

receptor that activates adenylate cyclase. Proc. Natl. Acad. ScL USA 87, 8940-8944. Wu, C.-F., and Ganetzky, B. (1992). Neurogenetic studies of ion channels in Drosophila. In Ion Channels, VoL 3, T. Narahashi, ed. (New York: Plenum Press), pp. 261-314. Wu, C.-F., and Haugland, F. N. (1985). Voltage clamp analysis of membrane currents in larval muscle fibers of Drosophila: alteration of potassium currents in Shaker mutants. J. Neurosci. 5, 2626-2640 Wu, C.-F., Ganetzky, B., Haugland, F. N., and Liu, A.-X. (1983). Potassium currents in Drosophila: different components affected by mutations of two genes. Science 220, 1076-1078. Zagotta, W. N., and Aldrich, R. W. (1990a). Voltage-dependent gating of Shaker A-type potassium channels in Drosophila muscle. J. Gen. Physiol. 95, 29-60. Zagotta, W. N., and Aldrich, R. W. (1990b). Alterations in activation gating of single ShakerA-type potassium channels by the Sh5mutation. J. Neurosci. 10, 1799-1810. Zholos, A. V., and Bolton, T. B. (1994). G-protein control of voltage dependence as well as gating of muscarinic metabotropic channels in guinea-pig ileum. J. Physiol. 478, 195-202. Zhong, Y., and Wu, C.-F. (1991). Alteration of four identified K÷currents in Drosophila muscle by mutations in eag. Science 252, 1562-1564. Zhong, Y., and Wu, C.-F. (1993). Modulation of different K÷ currents in Drosophila: a hypothetical role for the eag subunit in multimeric K÷ channels. J. Neurosci. 13, 4669-4679.