Ionic Currents of Kenyon Cells from the Mushroom Body of the Honeybee

The Journal of Neuroscience, August 1994, 74(8): 4800-4812

Sabine Schafer, Hendrik Rosenboom, and Randolf Menzel lnstitut fijr Neurobiologie, der Freien Universitzt, 14195 Berlin, Germany

The mushroom bodies have been suggested to be essen- mushroom body structure revealed that the mutants behave tially involved in learning and memory in insects. In the hon- poorly in olfactory conditioning tasks compared to wild-type eybee Apis mellifera they are composed of about 340,000 animals (Heisenberg et al., 1985). The gene products of Dro- intrinsic elements, called Kenyon cells, which can be easily sophila learning mutants dunce (dnc) and rutabaga (rut) have separated from all other neurons of the brain. Here we de- been shown to be preferentially expressedin the intrinsic ele- scribe a preparation in which we studied ionic currents in ments of the mushroom bodies, the Kenyon cells (Nighorn et the isolated Kenyon cell somata, using tight-seal whole-cell al., 1991; Han et al., 1992; see Davis, 1993, for review). Both recording. Several outward and inward currents were iden- mutationsinterfere with the CAMP pathway by affectinga CAMP- tified and investigated by the use of pharmacological agents specific phosphodiesteraseand a calcium/calmodulin-respon- and in ion substitution experiments: a rapidly inactivating sive adenylyl cyclase, respectively (Byers et al., 1981; Living- A-type potassium current that is completely blocked with 5 stoneet al., 1984). Behavioral pharmacologicalexperiments with mM 4-aminopyridine; a calcium-activated potassium current honeybees revealed that octopamine injected into the mush- that is blocked by l-100 nM charybdotoxin; a delayed rec- room bodiesfacilitates olfactory learning and memory retrieval tifier-type potassium current that is only weakly sensitive (Bicker and Menzel, 1989; Menzel et al., 1990), and biochemical to tetraethylammonium but is blocked by 100 MM quinidine; studies indicate that the CAMP cascademay be involved in a rapidly activating and inactivating, TTX-sensitive sodium mediating this effect (Menzel et al., 1991). In an electrophysi- current; a persistent sodium current that is both TTX and ological study, a presumably octopaminergic neuron has been cadmium sensitive; and a calcium current that is completely identified, which projects into the main input region of the bee blocked at 50 NM cadmium and is affected by verapamil and mushroom bodies, the calyces, and mediatesthe reward in ol- nifedipine only at high concentrations (100 PM). The currents factory learning (Hammer, 1993). Furthermore, Mauelshagen described here are very similar to currents found in other (1993) demonstrated changesin the responseof an identified insect neurons or muscle cells. mushroom body extrinsic neuron to an odorant after applying This preparation will not only facilitate studies concerning a classicalolfactory conditioning paradigm.Taken together,these the action of transmitters and neuromodulators that are con- resultsprovide evidence for an important role ofthe mushroom tained within neurons converging onto the Kenyon cells, but body intrinsic elements,the Kenyon cells, in olfactory learning will also allow a study of the role of the adenylyl cyclase and memory. pathway, elements of which are expressed in Kenyon cells, Unfortunately, due to the small size of cell bodies and pro- and are known to be essential for learning in invertebrates. cessesin flies and bees,electrophysiological investigations on [Key words: insect, whole-cell patch clamp, potassium the Kenyon cells cannot be performed in V~VO.We therefore currents, sodium currents, calcium currents] decided to study the isolated Kenyon cell somata of honeybee pupae usingtight-seal whole-cell recording(Hamill et al., 1981). The mushroombodies are probably the most peculiar structures This technique is particularly suitable for small cells and has in the insect brain, which has encouraged wide speculations already been successfullyapplied to a variety of insect neuronal about their possible function. Evidence for an involvement in preparations (Byerly and Leung, 1988; O’Dowd and Aldrich, learning and memory was initially derived from studiesin the 1988; Sole and Aldrich, 1988; Hardie and Weckstrijm, 1990; honeybee, which showed that selectively disturbing mushroom Hardie, 1991; Saito and Wu, 1991; Laurent et al., 1993; Pearson body function by cooling causesretrograde amnesiaafter olfac- et al., 1993). Anatomical studiesin the honeybee revealed that tory training (Menzel et al., 1974; Masuhr, 1976; Erber et al., each mushroom body contains about 170,000 Kenyon cells as 1980). Behavioral analysisof Drosophila mutants defective in the only intrinsic neuronal type (WitthGft, 1967; Mobbs, 1982). Becauseof their morphology, the mushroombodies in this insect can easily be dissectedout of the brain (Fig. 1) and their Received Sept. 13, 1993; revised Dec. 29, 1993; accepted Jan. 13, 1994. subsequentdissociation yields a pure preparation of Kenyon We thank Sabine Kreissl for introducing us to the cell preparation, Gerd Bicker cells. for sharing the cell culture facilities with us, Christine Jaeckel and Sybille Schaare With this article wegive a description ofthe variety ofvoltage- for technical assistance. and Paul Stevenson and Martin Hammer for comments on the manuscript. We are especially grateful to Tomaz Amon for assistance and gated and calcium-activated currents that are present in the a good spirit during the final phase of the experiments. This work was supported isolated somata of pupal Kenyon cells. The currents are com- by DFG Grant Pf 128/6-3 (S.S.) and a Leibniz award (R.M.). Correspondence should be addressed to Dr. Sabine Schlfer, Institut fir Neu- pared to those found in other insect neuronal and muscleprep- robiologie, Kiinigin-Luise-Strasse 28130, 14 195 Berlin, Germany. arations. Since we were the first to do voltage-clamp studies Copyright 0 1994 Society for Neuroscience 0270-6474/94/144600-13$05.00/O with theseneurons, we started to characterize their whole-cell

4602 Schafer et al. * Ionic Currents of Kenyon Cells currents. The results provide the basis for future investigations the larval and early pupal stages. the smaller-diameter somata on the action of transmitter substances and neuromodulators appear not before the second half of pupal development. after that. with immunocytochcmical techniques. have been shown neurogcncsis of Kenyon cells has been completed. Judged by to be contained within neurons converging onto the Kcnyon their location in the center of the calycal cups. the 7 pm cells cells. Since the clrrc and YU[ gene products have been shown to arc most likely the latest progeny of Kenbon cell neuroblasts (S. be present in Kcnyon cells. we arc particularly interested in a Eichmiillcr and S. Schgfcr. unpublished observations). In con- possible involvement of the aden!,lyl cyclase pathway in mod- trast to Kcnyon cells ofyounger pupae. which in the dish usually ulating ionic currents of these cells. grow a single long and thin process within a few hours after A preliminary report of some of the work prcscntcd here has dissociation. the cells of late-stage pupae used in the present appcarcd in abstract form (Schil’cr et al.. 1993). study hardly differentiate within the first 2 d. Recordings were made only from IO pm cells without processes, 12-36 hr after Materials and Methods dissociation. The membrane capacitance estimated from the ,~lnirnals. Honeybee (:lp/s r&li/&il) pupae were collected from the comb capacitance compensation settings on the amplifier was 2.6 i between days 2 and 6 of the pupal development. which lasts 9 d under 0.5 pF (mean t- SD: II = 9 I) and did not change with the time natural conditions. They were reared at 28°C and 80% relative humidity the cells had spent in culture. As the cells did not show any In an incubator until they had reached 70-80% of adult development (i.e.. pupal day 7) judged on the basis of the body pigmentation. inward rectifying potassium currents. or any other gated currents C(J// prc~[)nratiotz. Kenyon cells of 7-d-old pupae were dissected and at \,oltagcs below -60 mV, their input resistance was deter- cultured following the protocol of Krelssl (1992; Kreissl and Bicker. mined cithcr from the slope of the current-voltage relationship 1992) with only minor modifications. In brief. brains were removed bctwccn -60 and - I20 mV. or. simpl>,. by calculation from from the head capsule in a Lcibovit/ L15 medium (GIBCO-Bethesda Research Laboratory) supplemented with sucrose, glucose. fructose. and the response to a step to -90 mV from a holding potential of prolinc (42.0, 4.0, 2.5. and 3.3 gm’500 ml. rcspcctively) to reach a final --60 mV. The input resistance was 6.6 +- 3.3 Ci12 (mean +- SD; osmolaritv of 500 mOsm (LI j-500). After the brains had been cxposcd n = 3 I ). With the excention ofa transient sodium current. which to a hypcrismolar medium (containing 59 gm instead of42 gm sucrose/ could not be detected during the first I8 hr. all of the other 500 ml: 600 mOsm). the elial sheath could be removed readilv. This I I currents described below were also seen in cells that were studied was followed by another step In Ll5-500 medium in which mushroom bodies were dlssected out of the brains. After a IO min incubation in a immediately after dissociation (n = 8). Howcvcr. with these calcium-free Ringer solution (in mM: 130 NaCI. 5 KCI, IO MgCI,. 25 cells it was extremely difficult to obtain a whole-ccl1 configu- glucose. I80 sucrose. IO HEPES; pH 7.2). mushroom bodics(MBs) were ration. transfcrrcd back to L15-500 medium (4 MBs/200 ~1) and finally dis- In rcsponsc to depolarizing voltage steps from a holding po- sociated by gentle trituration with a 100 PI sihcomLed Eppendorfpipettc. tential of -65 mV the cells showed a fast and transient inward Cells were then olatcd in 25 ul ahauots into 100 ul droplets. of Ll5-500 medium on uncoated Falcon plast;c dishes and allowed to attach for at current followed by a much larger outward current (Fig. 2). Both least I5 min. Thereafter. the dishes wcrc filled with 2.5 ml of L15-500 inward and outward currents arc a combination of several com- and were kept at 29°C in an incubator at high humidity. Under these ponents that can be separated by the USC of pharmacological conditions the cells can survive for up to 10 d in culture (Kreissl, 1992). tools and the application of appropriate pulse protocols. ~le~r~oph~~~iologicul technrqtrcs. Whole-cell gigaohm seal recording was performed at room temperature following the methods described by Hamill et al. (1981). Recordings wcrc made using a Axopatch 1D amplifier (Axon Instruments). Pulse generation. data acquisition, and The outward current is composed of a rapidly inactivating and analysis wcrc carried out using a TL-I interface in conjunction with a noninactivating phase. In most of the cells examined. the I- PCLAMP programs (Axon instruments) running on an AT-type micro- computer. Pipette and membrane capacitance were compensated, and I’curvc for the outward current had a characteristic N-shape series resistance compensation (80%) was routinely employed. Currents with a halfmtay peak around + 50 mV (Fig. 2. inset). Alterations were low-pass filtered with a 4-pole Bessel filter at 2-5 kH/ and digitally in the external K’ concentration and measurements of tail- sampled at 4-20 kH/. depending on the current under investigation. current reversal potentials indicate that the outward current is Voltages wcrc corrected for liqtnd junction potential: leakage currents primarily carried by K+ (not shown). Further analysis. per- were not subtracted. Electrodes were pulled from borosilicate glass cap- illaries (CC 150-l 5. Clark. Reading) and had tip resistances between 4 formed in the prcsencc of 200 IBM TTX in the external solution and 6 MQ in standard external solution (see below). For statistical anal- to block Na+ currents, rcvcaled that the outward current is ysis of the data we used ORIGIN (MicroCal Inc.). Where appropriate. composed of at least three different potassium currents. data are presented as mean -t standard error of the mean (SEM). unless (‘Ct/l.iltt)l-ucti~,ated K’ cwwnts. Eliminating calcium currents stated otherwise. So/ut/ons. The bath was continuously perfused at 2 ml/min with a by addition of 50 PM cadmium to the bath solution dramatically standard external solution that consisted of (in mM) 130 NaCI, 6 KCI, altered the size and shape of the outward current: both transient 4 MgCI,. 5 CaCI,. 160 sucrose, 25 glucose. IO HEPES/NaOH; pH 7.2. and sustained phase were considerably reduced in si/e. and the In some cxpcriments Ca?. was replaced by Ba” For the analysis of inactivation of the transient component became faster. Fur- outward currents 200 nM tetrodotoxin (TTX) and in some cases 50 PM thermore, the N-shape of the I- I’ relation was completely abol- CdCI, was added. The standard internal solution contained (in mM): 100 K-asoartate. 40 KF. 20 KCI. 2.5 MgCI,. I EGTA, 3 ATI’. 160 ished (Fig. 3). Switching to an external solution containing 2 sucrose. 1’0 HEPES: pH 7.2. For the recording of inward currents KCI mM cobalt or cxposurc to a calcium-free saline had the same was replaced by TEA-Cl and the remaining K was substituted by Cs.. effect. As these observations arc indicative for the presence of Totally synthetic, HPLC-purilicd charybdotoxin was obtained from a calcium-activated potassium conductance, we tested whcthcr Bachem (Heidelberg. Germany). All other chemicals were purchased from Sigma (St. Louis, MO). charybdotoxin (ChTX). a blocker for a subtype of calcium-activated potassium channels. also changes si7e and shape of the outward current. Indeed. ChTX (I-100 nM) blocked a transient Results outward current that activates at potentials above -30 mV and Based on the size of their cell bodies. two subtypes of Kcnyon has a maximum around +50 mV (Fig. 4). Since the current cells can be distinguished: one IO wrn, the other 7 pm in diamctcr gradually decreased with time and repeated application of de- (Fig. 1b.c). While the larger cell bodies are already present in polarizing pulse protocols. its block by ChTX could be shown The Journal of Neuroscience, August 1994, 74(&I) 4603

cadmium senslthfe current

command potential [mVj hm., A 0 I -lb0 -50 d 5b 160 command potential [mV]

Figure S. Effect of 50 PM cadmiumon the outwardcurrent in the presenceof 100 nM TTX. I-V curvesfor the peak outwardcurrent before(A) and after (0) cadmium.+ indicatethe I-V relationof the cadmium-sensitivecurrent obtainedafter subtractingthe tracesob- tainedin the presencefrom thosebefore cadmium. The inset showsthe responseto a stepto +45 mV. Figure 2. Whole-cellcurrent in the absenceof any blockingagents. The membranepotential was stepped from a holdingpotential of -65 mV to a prepulsepotential of -95 mV for 100msec and then depo- larizedfor 250msec to commandpotentials between -65 and +85 mV determined according to the Nernst equation, assumingthe po- in 10 mV increments.Voltage steps were applied every 5 sec.The I-V tassium concentration within the cell to be identical to that of curvesfor the peakoutward (w) and sustainedoutward current (A) are the pipette filling solution. Even though the absolute size of the shownin the inset. A-type current was different from cell to cell (Fig. 7b), the re- spective curves for the relationship betweenthe fraction of con- to be reversible only for a few cells in which it was exceptionally ductance activated (g/g,,,,,) and voltage were almost identical. large at the very beginning. In addition to this transient current, From the normalized g-V curves a mean was calculated and ChTX in somecells also affected a small, sustainedcomponent fitted (Fig. 7~) with a Boltzmann distribution of the form ofthe outward current (Fig. 4d,e) that, in contrast to the transient g/&?,, = (2) calcium-activated potassium current, continuously increased l/l1 + ev[(V - v,,,YSl}, with increasingdepolarization. As its time courseresembled the where V,;2is the voltage at which half of the current isactivated delayed rectifier-type current describedbelow, this component and S is a factor determining the slopeofthe curve. The resulting of the ChTX-sensitive outward current may be due to an un- values were + 10.7 mV ( IJ’,,~)and - 16.76 (S). Steady state in- specificaction of the toxin. activation curves were obtained by measuringthe peak current ,4-tapecurrent. Superfusingthe cells with an external solution in responseto a test pulse (+30 mV) from different precondi- containing TTX, CdCl,, and quinidine (200 nM, 50 FM, and 100 tioning voltages (Fig. 7d,e). The mean inactivation curve and PM, respectively) revealed a very rapidly activating and inacti- its Boltzmann fit are shown in Figure 7J The I’,,, and S values vating outward current (Fig. 5). The time courseand sensitivity resulting from the fit were -42.33 mV and 8.15, respectively. to 4-aminopyridine (4-AP; Fig. 6) indicate this to be an A-type For a description of the kinetics of the current we measured current (Rudy, 1988). The block is voltage independent since time to peak for the activation, and examined the decay of the 4-AP reduced the current by the samepercentage over a large current by fitting a singleexponential to the falling phaseat each voltage range. To reveal the dose-responserelationship we ap- voltage. The time to peak rangedfrom 5-23 msecnear threshold plied activation pulse protocols before and in the presenceof to about 2 msec at +40 mV and above. The first-order time three different concentrations of 4-AP (0.2, 1, and 5 mM). The constant for the inactivation was l-4 msec above +40 mV. relative current persistingat a given concentration wasexpressed Delayed-rectijier current. Elimination of the A-type current asthe meanofthe values determined for five consecutive voltage by depolarizing prepulsesto -25 mV uncovered a third type steps(to 10,25,40, 55, and 70 mV). The inset ofFigure 6 shows of outward current (Fig. 8) that slowly activates with depolar- that the block is half-maximal at 0.8 mM, and is nearly complete izing pulsesabove - 30 mV. It was not affected by IO mM 4-AP, at 5 mM 4-AP. a concentration at which the A-type current was blocked com- Voltage activation curves were determined by measuringthe pletely. The effect of externally applied tetraethylammonium peak current activated by a given voltage (Fig. 7a,b). From these (TEA) was somewhat variable: in some cells the current was values the conductance(g) was calculated, usingthe relationship reduced only slightly by 20 mM TEA, and could not be completely blocked even at a concentration of 100 mM; in other g = I/(E - q. (1) cases20 mM TEA reduced the current to one-fourth or one- For E we usedthe potassiumequilibrium potential (-85 mV) fifth of the original size. In general, the sensitivity to TEA was 4604 Schafer et al. * Ionic Currents of Kenyon Cells

a control 3 b 1OnM 3 C wash 3 ChTX <2 C 9

d: a - b

Figure4. Effect of charybdotoxin (ChLY) on outward currents. The insetsshow the Z-V curves for the peak outward current. The cell was held at -65 mV, and after a 100 msec hyperpolarizing prepulse to -95 mV it was depolarized for 250 msec to command potentials between -80 and 100 mV, in 15 mV increments. a, Before application of ChTX the peak outward current has a maximum at +55 mV. b. In the presence of ChTX the peak outward current becomes apparently more transient and its Z-V curve looses the typical N-shape. c, The effect is partially reversed after a 2 min wash. d and e, The time course of the ChTX-sensitive current obtained by subtracting b from a (d), and b from c (e). related to the size of the current: the larger the current before PM (Fig. 5). In contrast to the A-type current, which persisted application, the stronger the TEA effect. On the basis of its without changein recordingsessions of over 30 min, the delayed relatively slow activation, and because it showed only little in- rectifier-type current showed massiverundown already within activation even with depolarizations lasting as long as 2 set, we the first 2 min after establishingthe whole-cell configuration. have classified it as a delayed rectifier-type current. Since we Whether this reflectsthe existenceof another subtypeof delayed observed some variability in the activation speedand inacti- rectifier-type channel, which may require intracellular condi- vation of the current, we believe that there are at least two tions that cannot be maintained during whole-cell recording, subtypesof channelswith different activation and inactivation remains to be seen. kinetics underlying the total current. However, as we were not able to separatethese components of the current any further, Inward currents we have not analyzed its kinetics in detail. Like its counterparts For an analysis of the inward current, outward currents were in other insectpreparations (Singh and Wu, 1990; Hardie, 199l), blocked by substituting KC1 with TEA-Cl and KF with CsF in it is sensitiveto quinidine, being completely abolishedwith 100 the pipette solution. Under theseconditions the remainingwhole-

+ 50 pM Quinidine

Figure 5. Effect of quinidine on outward currents. Same pulse protocol as in Figure 4. Quinidine significantly re- duces a sustained outward current component and leaves the very rapidly inactivating outward current largely unaffected. Inward currents and calcium-activated outward currents were blocked with 100 nM TTX and 50 FM cadmium. The Journal of Neuroscience, August 1994, M(8) 4605

a 100 pA

OmMTEA

20 mM TEA

Figure 6. 4-Aminopyridine sensitivity of the A-type current. The current was elicited by a step to +40 mV following a hyperpolarizing prepulse to -95 mV from a holding potential of -65 mV. It was recorded in the presence of TTX, Cd, and quinidine to block all other Figure 8. Delayed rectifier-type outward current and its partial block currents. The concentrations of 4-AP tested were 0.2, 1, and 5 mM. The by TEA. The current was recorded in the presence of 100 nM TTX and inset shows the dose-resoonse relation obtained from three different 50 PM Cd. The A-type current was totally inactivated by a 200 msec cells. depolarizing prepulse to -25 mV. Test pulses to -50, -35, -20, -5, and + 10 mV. a and b show the same traces with different sweep speed. c, In this example 20 mM TEA reduced the current to 40% of its original size. Time scale: a. 8 msec; b and c, 50 msec. b 3.0 - C 1 .o 2.5 - 1 0.8 T 1.5- 1 2 0.6 z 2.0 - WI p! h 0.4 l.O- 2 1 0.5 - 0.2 0 1

I I I 1 . I -100 -50 0 50 100 -100 -50 0 50 100 command potential [mV] command potential [mv] 25 e 1.21 f 1.0 -

0.8 -

?i 0.6 - .33 E 3 0.4 -

0.2 -

0 -

I -1w -100 -50 0 50 prepulse potential [mV] prepulse potential [mV] Figure 7. Characterization of the A-type current. The current was isolated by application of 100 nM TTX, 50 PM Cd, and 50 PM quinidine. u-c, Activation of the A-type current. Same pulse protocol as in Figure 4. d-1; Steady state inactivation determined by applying preconditioning voltage pulses between - 115 and +5 mV that preceded a test pulse to +25 mV. For further description, see Results. 4606 Schafer et al. l Ionic Currents of Kenyon Cells

cell current was mainly inward and showed a transient and a more slowly inactivating component (Fig. 9) the latter of which is normally masked by the massive outward currents. Both components are first activated by pulses to -40 mV and peak near - 10 mV. They can be separated pharmacologically and in ion substitution experiments. Sodium currents. Since we expected some of the inward current to be carried by calcium, we blocked calcium currents with 50 FM cadmium to study sodium currents in isolation, Under 0 L these conditions only a transient inward current remained, which IOms -0.2 was completely blocked with 50-100 nM TTX (see Fig. 116) a.4 and disappeared in Na+ -free external solution, indicating that AD.6 this component of the inward current was indeed a sodium current. Of all the currents examined, this current was the only 0 a.8 one for which we noticed an obvious correlation between the -1.0 size of the current and the time the cells had spent in vitro after -6040-20 0 2040 60 dissociation: during the first 18 hr the current could not be command potential detected at all and thereafter seemingly increased with the time VW the cells had been in culture. Since we restricted our investigations to cells that had not developed any processes, the ap- Figure 9. Total inward current. The pipette was filled with CsCl and pearance of the transient sodium current does not seem to re- TEA to block potassium currents. The inward current has at least two quire any morphological differentiation. components: one activating and inactivating rapidly, and one that de- Figure lOa-c shows an example of this inward sodium current cays more slowly. The inset shows the I-V curves for the peak (0) and the late (Orinward current. and the Z-V relation for 14 measurements. The current is activated at command potentials more positive than -45 mV

a b 0.41 0.2 - $0 - 2: 4.2. g xX4- f -0.6- ” X).8- XI.4 30 -1 .o - ‘\,/ -0.5 1 I.I.I-I.r’l’r 1. I’I - 1~1.I.I’l ...... _...... 40 -80-60 40-20 0 20 40 60 -80-60 40 -20 0 20 40 60 -7OmV command potential [mV] d e I f l.O- 1 .o -

0.8 - 0.8 - g 0.6- 0.6 - 5 5OOpA L 0.4 - Ims -1 OmV

1 I -100 -80 -60 40 prepulse potential [mV]

Figure 10. Transient sodium currents recorded in the presence of 50 FM Cd to block calcium currents. U-C, To measure the activation of the current, cells were held at -70 mV and depolarized to command potentials between -60 and +60 mV, in 10 mV increments. d-J; Steady state inactivation determined by application of preconditioning voltage pulses between -95 and -30 mV, in 5 mV increments, followed by a test pulse to - 10 mV. For further information, see Results. The Journal of Neuroscience, August 1994, 14(E) 4607

with a maximum around - 10 mV. With more positive test pulses the current decreasesas the potential approachesthe sodium equilibrium potential (+ 55 mV). Sincewe did not routinely use leak subtraction protocols, the sodium current is op- posedby leakageand incompletely compensatedcapacitive currents, both of which are outward at the command potentials applied. Therefore, the reversal potential of the current differed from cell to cell, and deviated from the sodium equilibrium potential, especially when the sodium current was relatively small. In a few caseswe did apply on-line leak subtraction using a P/N protocol, resulting in I-L’curves that reversedvery close to the sodium equilibrium. Neither activation threshold nor voltage at which the current was maximal differed from the b ’ values given above. Steady state inactivation was examined by ,Cd+TTX applying 1 set prepulsesto voltages between -95 and -30 mV Cd V in 5 mV increments (Fig. 1Od,e). From a set of nine normalized inactivation curves (Z/Z,,,,,), the mean was calculated and fitted (ZoopA (Fig. 1Of)with a Boltzmann distribution (Eq. 2) giving a I’,,? of -53 mV and a slope factor S of 5.4. 6ms The time courseof the sodium current was characterized by measuringtime to peak (t,) and inactivation rate. For the av- llX sensitive current erageof 14 cells measured,1, decreasedfrom 2.8 +- 0.3 at -40 in Ba-ES plus cadmium mV to 0.3 & 0.02 at +30 mV. The inactivation time constant (f,,) was determined from the fit of a single exponential to the Figure I I. Effect of 100 nM TTX on the inward current. Barium was falling phaseof the sodium current. It decreasedwith potential substituted for calcium in the external solution. Both a and b show the from 3.7 + 0.4 at -40 mV to 0.3 i 0.05 at +30 mV. response to a test pulse to - 10 mV from a holding potential of -70 Without Cd in the bath, application of TTX not only abol- mV. a, Application ofTTX blocks an inward current that has a transient ished the transient sodium current, but also eliminated a sus- and a persistent component (lower truce). b, After application of cad- tained component of the inward current (Fig. 1 la). Superfusing mium only a transient inward current, which is blocked by TTX (lower trace), is observed. the cells with an Na+-free solution had the same effect, indicating the existence of a persistent sodium current in addition to the transient sodium current describedabove. Sincewe never

b C

loo 100 mean (fSEM) 0 50 z B -100 - 0 E -200 3 *O -300 -100

400 -150

b-60 40-20 0 20 40 60 1 ‘,-,.,L’.,‘,.,. - -806040 -20 0 20 40 60 command potential [my

Figure 12. Barium currents recorded in the presence of 100 nM TTX to block sodium currents. Cells were held at -70 mV and depolarized to. command potentials between 0 and +60 mV, in 10 mV increments. b, I-Vcurves for the peak barium current obtained from 20 different cells. c, Mean barium current activation curve (?SEM). 4606 Schafer et al. * Ionic Currents of Kenyon Ceils

Figure 13. Effect of calcium channel blockers on the barium current. The same cell was consecutively exposed to verapamil (100 by; a), cadmium (IO ELM;b), and nifedipine(100 PM; c) with a 5 min washbetween the applications. Upper traces show the response to a de- polarizingpulse to -IO mV (holding potential- 70 mV) beforeand after ap- plicationof the drug.Lower traces were obtainedafter subtraction,and depict - ...... 1.’ p ...... - -I ...... 7 the time courseof the verapamil-,cadmium-, and nifedipine-sensitivecur- -pp4 rent, respectively. G-’ 6ms observed a persistent component of the inward current in 50 ison of the Z-V curves of the nifedipine-sensitive current with PM Cd (Fig. 1 lb), this persistentsodium current could either be that of the current that persisted in nifedipine did not reveal Cd sensitive or require calcium influx into the cell for its acti- any difference with respectto the voltage at which the current vation. activated, and where the current had its maximum. Further- Calcium current. In an external solution containing 200 nM more, multiplying the current persisting in nifedipine with a TTX, an inward current was observed that activates and in- variable factor would give a current with a time coursethat was activates more slowly than the transient sodium current. It was more or lessidentical to that ofthe current before the application completely blocked by 50 FM cadmium (Fig. 1 lb) and also of the drug. Thus, it seemsunlikely that there are two different disappearedwhen calcium in the external solution wasremoved calcium currents, one of which is blocked by nifedipine and one or replaced by cobalt (not shown). We therefore regard this of which is not. current to be a calcium current. With calcium asa chargecarrier, the current rapidly decreased Discussion to one-half to one-fifth of its original size within a few minutes. Recordings were performed on Kenyon cell somata isolated When calcium was substituted by barium this rundown was from the brain of honeybee pupae that had nearly completed much lesspronounced, suggestingthat the calcium current may adult development. Using the same dissection and culturing be inactivated or regulated in a calcium-dependent manner. protocol, Kreissl(1992) showedthat by injection of depolarizing However, this processmust be fairly slow sincethe inactivation current the cells can be induced to spike. As the pupal cells do kinetics, as judged from the 50 msecpulses, were not obviously not noticeably differ from adult Kenyon cells in their compo- altered after switching to a barium-containing external solution. sition of ionic currents (unpublished observations), we believe Replacingcalcium with barium alsoresulted in an instantaneous that the complement of ion channelsexpressed in Kenyon cells increaseofthe current without a changein the reversal potential, is complete at the end of the pupal stage.The inward and out- indicating that the calcium channels underlying the whole-cell ward currents observed are similar to those describedin other current are more permeableto barium than to calcium. insect neuronal (Byerly and Leung, 1988; Sole and Aldrich, Figure 12 gives an example of the barium current and shows 1988; Baker and Salkoff, 1990; Saito and Wu, I99 1) or muscle the I-Vrelations for the peak current derived from 20 different preparations (Gho and Mallart, 1986; Zagotta et al., 1988; Singh cells from a holding potential of - 70 mV. An average(+ SEM) and Wu, 1989, 1990). In contrast to the former, which included of all 20 measurementsis shown in Figure 12~.The current is a variety of different cell types dissociatedfrom entire brains or first activated at -40 mV and peaks between - 10 and 0 mV. ganglia, the preparation introduced here allows the selective Changing the holding potential to more negative values in- study of a distinct neuronal population. Investigations were re- creasedthe current significantly. The barium current was max- stricted to cells that had not grown any processesduring their imal at - 100 mV, reduced to three-fourths to two-thirds at - 70 short-term culture lasting up to 2 d. Even though the pupal mV, and was fully inactivated at - 30 mV. Lowering the holding Kenyon cell somatahave roughly the samediameter and whole- potential to - 100 mV did not uncover a secondtype of calcium cell capacitanceas Drosophila embryonic neurons (Byerly and (or barium) current. It simply increasedthe size of the overall Leung, 1988; O’Dowd and Aldrich, l988), their currents are current without affecting its kinetics or the shape of the Z-V about one order of magnitude larger. This may be becausewe relation. studied a very special neuronal type, or, more likely, because For a further pharmacologicalcharacterization we studied the embryonic neurons express fewer ionic channels than pupal effects of organic calcium channel blockers verapamil (a phen- neurons. With the Kenyon cell preparation this question may ylalkylamine) and nifedipine (a dihydropyridine), both ofwhich, be addressed,because it allows investigation of the expression in the 20 nM to 50 FM range, are thought to selectfor the L-type of ionic currents in this particular neuronal population not only calcium channel in vertebrates (Tsien et al., 1988; Hille, 1992). in the late pupa, but also at earlier stagesof postembryonic At 100 PM, verapamil significantly reduced (by SO-90%) the development, and in the adult. barium current in every cell tested (n = 14; seeFig. 13~ for an In Drosophila muscle two types of calcium-activated potas- example). The effect of 100 PM nifedipine (Fig. 13~) was less sium currents have been described: a fast transient current (ZcF pronounced (up to 50%); in several casesno effect was observed or I,,,) and a slow noninactivating current (I,, or I,) (Salkoff, .at all. Since we did not try any concentrations other than 100 1983; Elkins et al., 1986; Gho and Mallart, 1986; Singh and I.IM we do not know whether this is due to a difference in the Wu, 1989). Neither one appeared to be affected by apamin, a dose-responserelationships of the two compounds. A compar- bee venom polypeptide (Gho and Mallart, 1986) that has been The Journal of Neuroscience, August 1994, 14(8) 4609 shown to block selectively a special type of calcium-activated pulses,and sensitivity to TEA. It is very similar to the delayed potassium channel (see Cook and Quast, 1990, for review). I,, rectifier current found in other insectneurons (Byerly and Leung, has been reported to be blocked by charybdotoxin, a peptide 1988; Sole and Aldrich, 1988; Hardie and Weckstrom, 1990; from scorpion venom (Elkins et al., 1986). Because of its time Hardie, 1991; Saito and Wu, 1991). The variability of its ac- course and its sensitivity to charybdotoxin, the transient cal- tivation rate and the degreeof inactivation observedduring 250 cium-activated current described here closely resembles Zc,. msectest pulsessuggest that this current is composedof at least The sustained component of the ChTX-sensitive current seen two kinetically different conductances.Several potassiumchan- in Figure 4 may be due to an unspecific block of some other nels have been describedthat may contribute to sustainedout- outward current. In contrast to the transient ChTX-sensitive ward currents in insect neuronal and muscle preparations(Sole current, the I-Vrelation of which has an N-shape characteristic and Aldrich, 1988; Zagotta et al., 1988;Gorczyca and Wu, 1991; for calcium-dependent currents, the I-T/curve for the sustained Hardie, 1991). In the absenceof single-channeldata it is difficult component is more or less linear. In some cells ChTX has been to decide which and how many types of channelsunderlie the reported not to be perfectly selective, affecting not only calcium- delayed rectifier-type current in Kenyon cells. As the current activated potassium channels but also delayed rectifier currents showed only little inactivation during 2 set voltage steps, a (Cook and Quast, 1990). This may also be the case in our prep- noninactivating potassium channel (K, or L), in addition to aration. the slowly inactivating K,, must significantly contribute to this In addition, Kenyon cells may possess a slow, noninactivating current. calcium-activated potassium current: in many cells the 1-V re- The most striking feature of the delayedrectifier-type current lation for the persistent component of the outward current also was its susceptibility to washout, which occurred within a few had a characteristic N-shape (Fig. 1). However, this component minutes after establishingthe whole-cell configuration. The frac- of the sustained outward current, and hence the N-shape of its tion of the current that persistedbeyond this time was usually 1-V curve, usually disappeared within a few minutes after es- stable for the rest of the recording session.This observation, tablishing the whole-cell configuration (see, e.g., Fig. 4). There- and the variability in the sensitivity of the current to TEA, could fore, we were not able to characterize this current further. Re- be explained with the presenceof two subtypesof delayed rec- cently, Wegener et al. (1992) described a calcium-dependent tifier channels, one of which is more sensitive to TEA and is potassium channel in antenna1 receptor neurons of the locust readily “washed out,” the other being only weakly sensitiveto that is inhibited by millimolar concentrations of intracellular TEA, but resistantto washout. We have recently started to apply ATP and is blocked neither by apamin nor by charybdotoxin. the perforated-patch technique to our cells usingAmphotericin As our pipette solution contained 3 mM ATP, inhibition by B as an ionophore. Under theseconditions the delayedrectifier- ATP of a calcium-activated potassium current would explain type current was stable for a much longer period. Further ex- the rundown of a sustained outward current with N-shaped I- periments must now be performed to characterize the current I’ relation. So far, we have not performed any experiments in greater detail. without ATP in the pipette solution in order to see whether The transient sodium current describedhere is virtually in- under such conditions this particular component ofthe outward distinguishable,in terms of voltage operating range and kinetic current would be more stable. parameters,from that ofDrosophila embryonic neurons(O’Dowd In an immunological study Schwarz et al. (1990) showed that and Aldrich, 1988). Since we studied only cells without any Drosophila mushroom bodies expresshigh levels of Shaker- apparent processes,the source of the current is obviously the encodedpotassium channels.Thus, it was not surprising for us soma membrane. The fact that the sodium current does not to find a prominent A-type current in honeybee Kenyon cells arise in neurites(which might have been very thin and therefore that resemblesthe A-current of Drosophila muscle encodedby overlooked) can also be seenin the current records: we never the Shaker potassium channel gene, with respect to both its observed a significant delay between the onset of the test pulse kinetic parametersand its voltage operating range (Salkoff and and rising phaseof the sodium current, which should occur, if Wyman, 1983; Wu and Haughland, 1985; Zagotta and Aldrich, there was improper voltage control over processes. 1990). This is not the casein most other insect neuronal prep- It is unlikely that Kenyon cell somatagenerate sodium-driven arations in which the operating range of this current appearsto action potentials in vivo, as is the casefor the majority of insect be shifted by about 30 mV negative compared to that in Dro- nerve cell bodies.We observed no sodium currents immediately sophila muscle(Sole and Aldrich, 1988; Hardie and Weckstrom, after dissociation and up to 18 hr thereafter. The same occurs 1990; Hardie, 1991; Hardie et al., 1991; Saito and Wu, 1991). with isolated adult DUM neurons, where the development of It remainsto be seenwhether the A-type current found in Ken- the transient sodium current, following axotomy and deaffer- yon cells differs with respectto its voltage operating range from entation, can be abolishedby protein synthesisinhibitors (Tri- that of other neuronsin the bee brain. but et al., 1991). Immunocytochemical studies indicate that In pupal Drosophila neurons Baker and Salkoff (1990) re- sodium channelsin insect neuronsare primarily located in axon ported a considerablevariation in the inactivation kinetics and membranes,but restricted to the cytoplasmin cell bodies(French voltage dependenceof the A-type current, and suggestedthat et al., 1993). Our observations suggestthat isolatedKenyon cells in somecells two populations of channelscoexist. The shapeof retain their ability to synthesize sodium channels, which are, someof the inactivation curves (e.g., solid diamonds in Fig. 7e) however, incorporated into the soma membranein the absence is indeed suggestiveof the presenceof a secondtype of A-type of an axon. In contrast to the transient sodium current, all other current with a more negative voltage operating range. Further currents described here did not changewith time after disso- experiments are necessaryto substantiatethis impression. ciation. We thus conclude that the channels underlying these The second type of voltage-activated outward current en- currents are also present and functional in Kenyon cell somata countered in our preparation is characterized by a slow acti- in vivo. vation, only little inactivation during 250 msec depolarizing Persistent, TTX-sensitive sodium currents have been found 4610 Schafer et al. * Ionic Currents of Kenyon Cells in cockroach (Christensen et al., 1988; Lapied et al., 1989) and current seemsto undergosteady state inactivation at fairly neg- Drosophila (Saito and Wu, 1991) neurons. A similar current ative holding potentials, being 100%available at - 100 mV and also seems to be present in Kenyon cells. However, we were not below, about 70% available at -70 mV, and fully inactivated able to study this current in detail, since it could not be effec- at -30 mV and above. As is the case with all other insect tively separated from the other inward currents. The persistent preparations studied so far, the calcium current first activates sodium current was absent when the bath contained micromolar at -40 mV and peaksbetween - 10 and 0 mV. Given that the concentrations of cadmium to block the calcium current; sub- partial block by 100 FM verapamil and nifedipine may be a stituting calcium by cobalt had the same effect. Isolating both nonspecific effect, the calcium current describedhere has many sodium currents together by the use of a calcium-free external characteristicsin common with the vertebrate “N-type” calcium solution was also impracticable, as recordings were instable un- current, such as sensitivity to very low concentrations of cad- der these conditions. Therefore, we do not know whether the mium, activation at relatively high voltages, and steady state persistent sodium current differs with respect to activation inactivation at rather negative holding potentials. threshold and steady state inactivation from the transient so- It is yet unknown whether and how the above describedcur- dium current, as is the case in Drosophila neurons (Saito and rents are modulated, and how such a modulation may be related wu, 1991). to neural plasticity. Indications of such have, however, emerged Calcium currents have been describedin Drosophila embry- from studieson the Drosophila learning and memory mutants onic neurons(Byerly and Leung, 1988; Leung and Byerly, 1991; dunce (dnc) and rutabaga (rut), both of which are characterized Saito and Wu, 199l), Manduca larval and pupal motoneurons by abnormal levels in intracellular CAMP (Byers et al., 1981; (Hayashi and Levine, 1992) and locust adult (Pearsonet al., Livingstone et al., 1984). A clue as to how thesealtered CAMP 1993) and embryonic interneurons (Laurent et al., 1993). In levels might causedeficient learning and memory was provided Drosophila, calcium currents with different inactivation kinetics by an electrophysiological study (Zhong and Wu, 1991) which have been observed (Byerly and Leung, 1988; Saito and Wu, showed that in contrast to the wild type, dnc larva1 neuromus- 199l), which were differentially affected by two spider toxins, cular junctions do not exhibit facilitation and posttetanic po- indicating the existence of two different types of neuronal cal- tentiation. As Delgado et al. (1992) demonstrated, this kind of cium currents (Leung et al., 1989). In contrast, the waveform synaptic plasticity can be restored by application of potassium of the calcium current describedhere was rather consistentfrom channel blockers. Drosophila larval muscle fibers express a cell to cell, presumably becausewe recordedonly from Kenyon CAMP-activated potassium channel, which is persistently ac- cells, and not from a variety of different neuronal types. Laurent tivated in dnc (Delgado et al., 1991). In the larval musclefibers et al. (1993) made a similar observation studying calcium cur- of dnc, both A-current and delayed rectifier are increased,pos- rents in nonspiking interneurons from the locust. sibly as a result of elevated CAMP levels (Zhong and Wu, 1993). Substituting calcium with barium had two effects. First, it Obviously, the samemechanisms underlying lack of synaptic increasedthe current without changing the reversal potential, plasticity of the dnc neuromuscularjunction may also be rele- indicating that the underlying calcium channels,like those pres- vant in the CNS. In the brain of Drosophila, both dnc and rut ent in adult locust neurons (Pearson et al., 1993) are more gene products are preferentially expressedin the intrinsic ele- permeable to barium than to calcium. Second, it very much ments of the mushroom bodies, the Kenyon cells (Nighorn et slowedthe rundown ofthe current, which would normally result al., 1991; Han et al., 1992). In the fly, however, the role of the in a loss of the calcium current within 5-10 min, possibly be- CAMP pathway in these cells has so far not been investigated. causeof a calcium-dependentinactivation or regulation of the Due to their small size (< 5 Hrn) they appear to be inaccessible current. Barium, however, could not reverse rundown once it to electrophysiologicalstudies in vivo. Furthermore, becauseof had occurred. Unlike Drosophila embryonic neurons(Saito and anatomical constraints,a preparation of the fly mushroombod- Wu, 199l), barium did not changethe waveform of the current, ies and hence an isolation of their intrinsic elementsdoes not indicating that the inactivation observed during 50 msec de- seemfeasible. The honeybee preparation describedhere allows polarizing pulsesis most likely not calcium dependent, and that selective study of this cell type, not only with electrophysiolog- the calcium-dependentregulation of the calcium current leading ical but also with biochemical and other techniques.A potential to rundown must be a rather slow process. mechanismunderlying olfactory learningand memory in insects Organic calcium channel blockers, such as verapamil and may be a changein Kenyon cell excitability through a CAMP- nifedipine, which block certain types of vertebrate calcium chan- mediated modulation of ion channels.Most of the currents de- nels, have been shown to have little effect on insect calcium scribed here have been shown to be modulated by CAMP in currents (Byerly and Leung, 1988; Hayashi and Levine, 1992). other preparations (Rudy, 1988; Tsien et al., 1988). The effect Thus, it was not surprising to find that, even at a concentration of CAMP on ion channelsmay be direct, or involve the activity as high as 100 WM, both verapamil and nifedipine reduced the of a CAMP-dependent kinase (PKA), an enzyme that is also calcium current to some extent, but would not abolish it com- found in the bee brain (Altfelder and Miiller, 1991). In a bio- pletely. Pearsonet al. (1993) reported that 1 PM verapamil re- chemical assay,dopamine, noradrenaline, and octopamine have duced the calcium current in adult locust neuronsby 85%. This been shown to stimulate adenylyl cyclaseactivity in bee brain was about the maximal effect of verapamil seenin our prepa- tissue, which is particularly high in the mushroombodies (Men- ration with 100 FM. However, we did not test any concentrations zel et al., 1991). With immunocytochemical techniques, do- lower than that. At the same concentration, nifedipine, a di- pamine- and octopamine-containingneurons were found, which hydropyridine that is also thought to selectfor L-type channels project into the main input regions of the mushroom bodies, in vertebrates (Tsien et al., 1988) was even lessefficient.The the calyces (Schafer and Rehder, 1989; Kreissl et al., 1991). strongestblocking effect was seenwith cadmium, which com- Most if not all of the octopamine immunoreactivity within the, pletely abolishedthe calcium current at 50 PM and significantly calyces could be attributed to a neuron, which in an electro- reduced it at concentrations well below 10 WM. The calcium physiological study has been shown to mediate the reward in The Journal of Neuroscience, August 1994. 74(E) 4611 olfactory learning (Hammer, 1993). Furthermore, there is good room body mutants are deficient in olfactory learning. J Neurogenct evidence that acetylcholine may be transmitter of a subset of 2: l-30. neurons conveying olfactory information from the antenna1 lobes Hille B (1992) Ionic channels of excitable membranes. Sunderland, MA: Sinauer. to the mushroombodies (Kreissl and Bicker, 1989). Thus, future Kreissl S (1992) Zellbiologische und immuncytochemische Untersu- studieswill be directed toward determining the action of trans- chungen am zentralen Nervensystem der Honigbiene in situ und in mi’tters and neuromodulators on Kenyon cell currents, and vitro. PhD thesis, Freie Universitit, Berlin. whether CAMP is involved as a second messenger. Kreissl S, Bicker G (1989) Histochemistry of acetylcholinesterase and immunocytochemistry ofan acetylcholine receptor-like antigen in the brain of the honevbee. J Coma Neurol 286:7 l-84. Kreissl S, Bicker G -( 1992) Dissociated neurons of the pupal honeybee References brain in cell culture. J Neurocytol 21:545-556. Altfelder K, Miiller U (1991) Cyclic nucleotide-dependent protein Kreissl S, Eichmiiller S, Bicker G, Rapus J, Eckert M (1991) The kinases in the neural tissue of the honeybee Apis melliferu. Insect distribution of octopamine-like immunoreactivity in the brain of the Biochem 2 1:487-494. honeybee. In: Synapse-transmission-modulation. Proceedings of the Baker K, Salkoff L (1990) The Drosophila shaker gene codes for a 19th GGttingen Neurobiology Conference (Elsner N, Penzlin H, eds), distinctive K+ current in a subset of neurons. Neuron 2: 129-140. abstr 407. Stuttgart: Thieme. Bicker G, Menzel R (1989) Chemical codes for the control of behav- Lapied B, Molecot 0, Pelhate M (1989) Ion species involved in the iour in arthropods. Nature 337:33-39. electrical activity of single adult aminergic neurons isolated from the Byerly L, Leung H-T (1988) Ionic currents of Drosophila neurons in sixth abdominal ganglion of the cockroach Periplaneta americana. J embryonic cultures. J Neurosci 8:4379-4393. Exp Biol 144:535-549. Byers D, Davis RL, Kiger JA (198 1) Defect in cyclic AMP phospho- Laurent G, Seymour-Laurent KJ, Johnson K (1993) Dendritic excit- diesterase due to the dunce mutation of learning in Drosophila me/- ability and a voltage-gated calcium current in locust nonspiking local anogaster. Nature 289:79-S 1. interneurons. J Neurophysiol 69:1484-1498. Christensen BN, Larmet Y, Shimahara T, Beadle D, Pichon Y (1988) Leung H-T, Byerly B (199 I) Characterization of single calcium chan- Ionic currents in neurones cultured from embryonic cockroach (Per- nels in Drosophila embryonic nerve and muscle cells. J Neurosci 11: iakzneta americana) brains. J EXD Biol 135: 193-2 14. 3047-3059. Cobk NS, Quast U (i 990) Potassium channel pharmacology. In: Po- Leung H-T, Branton WD, Philips HS, Jan L, Byerly L (1989) Spider tassium channels (Cook NS, ed), pp 181-255. Chichester: Horwood. toxins selectively block calcium currents in Drosophila. Neuron 3:767- Davis R’L (1993) Mushroom bodies and Drosophila learning. Neuron 172. 1l:l-14. Livingstone MS, Sziber PP, Quinn WG (1984) Loss of calcium/cal- Delgado R, Hidalgo P, Diaz F, Latorre R, Labarca P (1991) A cyclic modulin responsiveness in ihe adenylate cyclase of rutabaga, a Dro- AMP-activated K+ channel in Drosophila larval muscle is persistently sophila learning mutant. Cell 37:205-2 15. activated in dunce. Proc Natl Acad Sci USA 88:557-560. Masihr T (1976) Lokalisation und Funktion des Kurzzeitgedlcht- Delgado R, Latorre R, Labarca P (1992) K+ channel blockers restore nisses der Honigbiene, Apis mellifera L. PhD thesis, Freie Universitlt, synaptic plasticity in the neuromuscular junction of dunce, a Dro- Berlin. sophila learning and memory mutant. Proc R Sot Lond [Biol] 250: Mauelshagen J (1993) Neural correlates of olfactory learning para- 181-185. digms in an identified neuron in the honeybee brain. J Neurophysiol Elkins T, Ganetzky B, Wu C-F (1986) A Drosophila mutation that 69~609-625. eliminates a Cal+-deoendent K+ current. Proc Nat1 Acad Sci USA Menzel R, Erber J, Masuhr T (1974) Learning and memory in the 83:8415-8419. - honeybee. In: Experimental analysis of insect behaviour (Barton-Brown Erber J, Masuhr TH, Menzel R (1980) Localization of short-term L, ed), pp 195-217. New York: Springer. memorv in the brain ofthe bee Atzs mellifera. Phvsiol Entomol5:343- Menzel R, Wittstock S, Sugawa M (1990) Chemical codes of learning 358. . and memory in the honeybee. In: Medica hoechst, Vol23, The biology French AS, Sanders EJ, Duszyk E, Prasad S, Torkkeli PH, Hashins J, of memory (Squire L, Lindenlaub K, eds), pp 335-359. Stuttgart: Murphy RA (1993) Immunocytochemical localization of sodium Schattauer. channels in an insect central nervous system using a site-directed Menzel R, Hammer M, Braun G, Mauelshagen J, Sugawa M (1991) antibody. J Neurobiol 24:939-948. Neurobiology of learning and memory in honeybees: In: The behav- Gho M, Mallart A (1986) Two distinct calcium-activated potassium iour and Dhvsioloav of bees (Goodman L. Fisher RC. eds).I._. DD 323- currents in larval muscle fibers of Drosophila melanogaster. Pfluegers 353. 0x0;: CAB &emational. Arch 4011526-533. Mobbs P (1982) The brain of the honeybee Apis mellifera. I. The Gorczyca MG, Wu C-F (1991) Single-channel K+ currents in Dro- connections and spatial organization of the mushroom bodies. Philos sophila muscle and their pharmacological block. J Membr Biol I2 1: Trans R Sot Lond [Biol] 298:309-354. 237-248. Nighom A, Healy MJ, Davis RL (1991) The cyclic AMP phospho- Hamill OP, Marty A, Neher E, Sakman B, Sigworth F (198 1) Improved diesterase encoded by the Drosophila dunce gene is concentrated in patch-clamping techniques for high resolution current recordings from the mushroom body neuropil. Neuron 6:455-467. cells and cell free membrane patches. Pfluegers Arch 391:85-100. O’Dowd DK, Aldrich RW (1988) Voltage-clamp analysis of sodium Hammer M (1993) An identified neuron mediates the unconditioned channels in wild-type and mutant Drosophila neurons. J Neurosci stimulus in associative olfactory learning in honeybees. Nature 366: 8:3633-3643. 59-63. Pearson HA, Lees G, Wray D (1993) Calcium channel currents in Han P-L, Levin LR, Reed RR, Davis RL (1992) Preferential expres- neurones from the locust (Schistocerca gregaria) thoracic ganglia. J sion of the Drosophila rutabaga gene in mushroom bodies, neural EXD Biol 177:201-221. centers for learning in insects. Neuron 9:6 19-627. Rudy B (1988) Diversity and ubiquity of K channels. Neuroscience Hardie RC (I 99 I) Voltage sensitive potassium channels in Drosophi/u 251729-749. ohotoreceotors. J Neurosci I 1:3079-3095. Saito M, Wu C-F (199 1) Expression of ion channels and mutational H&die RC, Weckstrijm M (1990) Three classes of potassium channels effects in giant Drosophila neurons differentiated from cell division- in large monopolar cells of the blowfly Culliphoru vicina. J Comp arrested embryonic ieuroblasts. J Neurosci 11:2 135-2 I 50. Physiol [A] 167:723-736. Salkoff L (1983) DrosoDhila mutants reveal two comoonents of fast Hardie RC, Voss D, Pongs 0, Laughlin SB (1991) Novel potassium transient‘curr&t. Nat&e 302:249-25 1. channels encoded by the shaker locus in Drosophila photoreceptors. Salkoff L, Wyman RJ (1983) Ion currents in Drosophila flight muscle. Neuron 6:477-486. J Physiol (Lond) 337:687-709. Hayashi JH, Levine RB (1992) Calcium and potassium currents in Schsfer S, Rehder V (1989) Dopamine-like immunoreactivity in the leg motoneurons during postembryonic development in the hawk- brain and suboesophageal ganglion of the honeybee. J Comp Neurol moth Manduca sexta. J EXD Biol 17 1: 15-42. 280:43-58. Heisenberg M, Borst A, Wagner S, Byers D (1985) Drosophila mush- Schafer S, Rosenboom H, Menzel R (1993) Ionic currents in identified 4612 Schafer et al. * Ionic Currents of Kenyon Cells

neurons from the mushroom body of the honeybee, Apis mellutira. Wegener JW, Tareilus E, Breer H (1992) Characterization of calcium- In: Gene-brain-behaviour. Proceedings of the 2 1st Gijttingen Neu- dependent potassium channels in antenna1 receptor neurones of Lo- robiology Conference (Elmer N, Heisenberg M, eds), abstr 634. Stutt- custa migratoria. J Insect Physiol 38:237-248. gart: Thieme. Witthijft W (1967) Absolute Anzahl und Verteilung der Zellen im Schwarz TL, Papazian DM, Carretto RC, Jan Y-N, Jan LY (1990) Him der Honigbiene. Z Morph01 Tiere 6 1: 160-l 84. Immunological characterization of K+ channel components from the Wu C-F, Haugland FN (1985) Voltage clamp analysis of membrane Shaker locus and differential distribution of splicing variants in Dro- currents in larval muscle fibers of Drosophila: alteration of potassium sophila. Neuron 2: 119-I 27. currents in Shaker mutants. J Neurosci 5:2626-2640. Singh S, Wu C-F (1989) Complete separation of four potassium cur- Zaeotta WN. Aldrich RW (1990) Voltaae-deoendent aatina of Shaker rents in Drosophila. Neuron 2: 1325-l 329. A-type channels in Dros~phila’muscle~J Gen Physiol95:>9-60. Singh S, Wu C-F (1990) Properties of potassium currents and their Zagotta WN, Brainard MS, Aldrich RW (1988) Single-channel analysis role in membrane excitability in Drosophila larval muscle fibers. J of four distinct classes of potassium channels in Drosophila muscle. Exp Biol 15259-76. J Neurosci 8:4765-4779. Sole CK, Aldrich RW (1988) Voltage-gated potassium channels in Zhong Y, Wu C-F (199 1) Altered synaptic plasticity in Drosophila larval CNS neurons of Drosophila. J Neurosci 8:2556-2570. memory mutants with a defective cyclic AMP cascade. Science 25 1: Tribut F, Lapied B, Duval A, Pelhate M (1991) A neosynthesis of 198-201. sodium channels is involved in the evolution of the sodium current Zhong Y, Wu C-F (1993) Differential modulation of potassium cur- in isolated adult DUM neurons. Pfluegers Arch 4 19:665-667. rents by CAMP and its long-term and short-term effects: dunce and Tsien RW, Lipscombe D, Madison, DV, Bley KR, Fox AP (1988) rutabaga mutants of Drosophila. J Neurogenet, in press. Multiple types of neuronal calcium channels and their selective modulation. Trends Neurosci 11:431-438.