Differential expression of voltage-sensitive K⁺ and Ca²⁺ currents in neurons of the honeybee olfactory pathway

The formation of olfactory memory is a multistage process that involves different areas of the insect brain. Olfactory conditioning of the proboscis extension reflex of the honeybee has proved to be a very powerful system for studying learning-related neural mechanisms, enabling major neural elements of the olfactory and reward pathways in the honeybee brain to be identified (Erber et al., 1980; Hammer, 1993; Mauelshagen, 1993; Grünewald, 1999b; Hammer and Menzel, 1998) (for reviews, see Hammer, 1997; Menzel, 1999, 2001) and several molecular pathways involved in this behaviour to be elucidated (e.g. Hildebrandt and Müller, 1995; Müller, 1996, 2000; Grünbaum and Müller, 1998; Fiala et al., 1999) (for a review, see Menzel and Müller, 1996). Odors are perceived by sensillae on the honeybee antennae. The axons of the olfactory receptor neurons terminate within the primary olfactory neuropils of the honeybee brain, the antennal lobes. Olfactory information is processed here in a complex spatio—temporal fashion (Joerges et al., 1997; Stopfer et al., 1997; Sachse et al., 1999) and projection neurons transmit olfactory information from the antennal lobes to the lateral protocerebral lobes and the mushroom bodies within the protocerebrum (Homberg, 1984; Fonta et al., 1993; Abel et al., 2001). In the mushroom body calyces the projection neurons synapse onto mushroom body-intrinsic Kenyon cells, named after their discoverer (Kenyon, 1896). Olfactory information converges here with information from other sensory modalities, such as visual input in the mushroom body, and with reward-processing modulatory neurons from the suboesophageal ganglion (Erber et al., 1987; Hammer and Menzel, 1995). Intracellular recordings showed that odor applications induce complex spiking patterns in projection neurons, thus encoding olfactory and gustatory sensory information (Stopfer et al., 1997; Abel et al., 2001; Müller et al., 2002). The underlying ion channels of projection neurons have not yet been analysed, however, because the neurons could not be identified and distinguished from local interneurons, as is possible in other insects (Hayashi and Hildebrand, 1990; Oland and Hayashi, 1993).

The ionic currents of postsynaptic Kenyon cells, by contrast, have been analysed in vitro. They express functional nicotinic acetylcholine receptors, which may mediate fast synaptic transmission between antennal lobe projection neurons and mushroom body Kenyon cells (Goldberg et al., 1999; Deglise et al., 2002). Several voltage-sensitive ionic currents have been described (Schäfer et al., 1994; Pelz et al., 1999). Among the three different voltage-sensitive K⁺ currents is a transient A-type K⁺ current, which resembles the shaker-like current of other systems and interacts with a fast voltage-sensitive Na⁺ current during spike generation (Pelz et al., 1999). Similar voltage-sensitive currents have been described in honeybee antennal motor neurons, both in vitro and in situ (Kloppenburg et al., 1999a).

The antennal lobes and mushroom bodies are differentially involved in memory formation in the honeybee (Erber et al., 1980; Hammer and Menzel, 1998; Menzel, 2001) and our long-term goal is to analyse learning-related changes of cell physiology in the different olfactory neurons, beginning with an analysis of the ionic conductances of the different neurons types, antennal lobe projection neurons and mushroom body Kenyon cells. This, however, requires a staining technique that allows the identification of the neuron type prior to recording. Whereas cultures of mushroom bodies comprise somata of a homogeneous cell type (Kenyon cells), the dissociation of antennal lobes yields a heterogeneous mixture of projection neurons and local interneurons (Gascuel and Masson, 1991b; Devaud et al., 1994; Kirchhof and Mercer, 1997). To record from identified cell types, namely projection neurons and Kenyon cells, we have developed a labelling technique that enabled us to identify neurons in the culture dish prior to patch clamp recording. This study revealed pronounced differences between projection neurons and Kenyon cells.

Honeybee Apis mellifera L. pupae were collected from the comb between days 4 and 6 of pupal development, which lasts 9 days under natural conditions. At these stages projection neurons have already formed a branching pattern in the antennal lobe and the mushroom body, which is very similar to adult cell morphology (Schröter and Malun, 2000).

Projection neurons were identified by dye injection. For this, dextran-coupled rhodamine (MW 3000, Molecular Probes Inc., Eugene, OR, USA) was injected into the mushroom body calyces (see below), and the somata of projection neurons from the antennal lobes were retrogradely labelled and could easily be identified in vitro (see below). The procedure used was a modification of a protocol developed for confocal microscopical analyses of projection neuron morphology (Schröter and Malun, 2000).

Pupal honeybees were decapitated and a small window was cut into the head cuticle frontally between the compound eyes, the bases of the antennae, and the ocelli. After removal of trachea and glands the brain, with the prominent mushroom body calyces, was clearly visible. The head was flooded with sterile standard external saline (see below) with gentamycin (150 μl/50 ml, Gibco Life Technologies, Karlsruhe, Germany) added. Using a sterile quartz glass capillary (1.0 mm o.d., 0.5 mm i.d.) pulled with a horizontal laser puller (P2000, Sutter Instruments, Novato, CA, USA), the lip region of the calyces was repeatedly punctured in order to damage the neurons and allow dye uptake. A saturated paste from the dextran-coupled rhodamine was prepared by adding one drop of sterile distilled water to the dry dextran—rhodamine on a glass slide. Therefore, the exact dye concentration could not be determined. The lip regions of the calyces were briefly punctured with a sterile capillary, whose tip was coated with the paste, to insert a small amount of dye. After gently rinsing off excessive dye solution with standard external saline, the dye was allowed to diffuse to the somata for 3 h at room temperature.

To confirm that rhodamine—dextran labelled only projection neurons in the antennal lobe, whole-mount preparations were analysed using confocal microscopy. Brains were dissected out of the head capsule after labelling, briefly rinsed with standard external saline, and fixed for 1 h in 4% paraformaldehyde in 0.1 mol l^-1 phosphate-buffered saline (PBS), pH 7.2 at room temperature. Subsequently, specimens were dehydrated in graded ethanol and cleared in methyl salycilate. Whole-mounts were mounted in Permount (Fisher Chemicals, Springfield, NJ, USA) on slides and viewed with a confocal laser scanning microscope (Leica TCS-4D) equipped with a krypton/argon laser light source. At primary magnifications between 10-40×, several optical sections were imaged at 1-5 μm intervals. Series of images were stacked and two-dimensional projections of image stacks generated using the extended focus function of Imaris software (version 2.7, Bitplane).

Kenyon cells and projection neurons were dissected and cultured following a modified protocol of Kreissl and Bicker (1992). After the staining procedure brains were removed from the head capsule and transferred into a preparation medium (Leibovitz L15, Gibco BRL; see below). The glial sheath was gently removed and the mushroom bodies or the antennal lobes dissected out of the brains. After incubation (10 min) in calcium-free saline, mushroom bodies were transferred back to the preparation medium (2 mushroom bodies or 10-15 antennal lobes per 100 μl medium) and dissociated by gentle trituration with a 100 μl Eppendorf pipette. Cells were then plated in 10 μl samples on Falcon plastic dishes coated with polylysine (polylysine-L-hydrobromide, MW 150-300 kDa, Sigma) and allowed to settle and adhere to the substrate for at least 15 min. Thereafter, the dishes were filled with 2.5 ml of a supplemented culture medium (see below) and were kept at 26°C in an incubator at high humidity. Because the mushroom bodies can be mechanically dissected out of the brain and contain somata of Kenyon cells exclusively, this procedure yielded a culture of pure Kenyon cells. By contrast, antennal lobe neurons are a heterogeneous population consisting of two major classes, projection neurons and local interneurons. Thus, dissection of antennal lobes yielded cultures containing neurons of different classes (cf. Kirchhof and Mercer, 1997) and projection neurons needed to be identified in vitro. Since only projection neurons were labelled by dextran—rhodamine, their somata were easily identified in the culture dish by their fluorescence. Labelled projection neurons were located with epifluorescence illumination and photographed using a Zeiss inverted microscope (Axiovert 10, Zeiss, Jena, Germany), which was part of the patch clamp setup. Images were taken under phase-contrast optics at 32× primary magnification with a PC-controled digital camera (Olympus DP10; utility software C-2.1, Olympus). The digitized images were analysed and processed with Adobe Photoshop (version 5.0, Adobe Systems Inc.).

For electrophysiological measurements, cells were used between culture days 3 and 6. Processes of those cells chosen for recordings did not overlap with neighbouring neurites.

The contents of the cell culture solutions were similar to those used during earlier physiological studies (Goldberg et al., 1999; Pelz et al., 1999).

To 1000ml of Leibovitz's L15 medium (Gibco) was added: sucrose, 30.0 g; glucose, 4.0 g; fructose, 2.5 g; proline, 3.3 g. The medium was adjusted to pH 7.2 with NaOH and to 500 mosmol l^-1; with sucrose.

Culture medium contained heat-inactivated fetal calf serum (Sigma, St Louis, MO, USA), 13% (v/v); yeast hydrolysate (Sigma), 1.3% (v/v); L-15 powder medium (Gibco BRL), 12.5% (w/v); glucose, 18.9 mmol l^-1; fructose, 11.6 mmol l^-1; proline, 3.3 mmol l^-1; 93.5 mmol l^-1 sucrose; Pipes, 2.1 mmol l^-1. The medium was adjusted to pH 6.7 and 500 mosmol 1^-1.

Calcium-free saline consisted of: NaCl, 147 mmol l^-1; KCl, 5 mmol l^-1; Hepes, 65 mmol l^-1; pH 7.2, 392 mosmol l^-1.

Whole-cell seal recordings (GΩ) were performed at room temperature following the methods described by Hamill et al. (1981). Recordings were made using a computer-controlled HEKA EPC9 patch-clamp amplifier (HEKA-Elektronik, Lambrecht, Germany). Data acquisition and online analyses were performed with PULSE software (version 8.50, Heka-Elektronik) running on a Pentium-based PC. Data were sampled at 10-20 kHz and were low-pass filtered with a four-pole Bessel filter. Voltages were corrected for liquid junction potential (4 mV); leakage currents were not subtracted. Series resistances were 5-20 MΩ and were compensated at about 80%. The cell capacitance for each cell was estimated from the capacitance compensation routine of the PULSE software. For some cells, cell capacitances and time constants were calculated from capacitive charging currents, which were measured during hyperpolarizing voltage pulses to a nonactive region of the membrane potential (-80 to -100 mV) prior to capacitance compensation. Electrodes were pulled from borosilicate glass capillaries (GB 150-8P, Science Products Germany) with a horizontal puller (DMZ Zeitz Instruments, München, Germany), and had tip resistances of 8-12 MΩ in standard external solution (see below). The holding potential was -80 mV throughout.

The bath was continuously perfused at about 2 ml min^-1 with a standard external solution that consisted of NaCl, 130 mmol l^-1; KCl, 6 mmol l^-1; MgCl₂, 4 mmol l^-1; CaCl₂, 5 mmol l^-1; sucrose, 160 mmol l^-1; glucose, 25 mmol l^-1; Hepes, 10 mmol l^-1. The external saline was adjusted to pH 6.7 with NaOH and to 510 mosmol l^-1. To record currents through K⁺ channels, the standard saline was replaced with one in which tetrodotoxin (TTX, 100 nmol l^-1) was added to block currents through the voltage-sensitive Na⁺ channels. Some experiments were performed with additional CdCl₂ (50 μmol l^-1) in the external solution to block voltage-sensitive Ca²⁺ currents. The pipette solution contained: K-gluconate, 87 mmol l^-1; KF, 40 mmol l^-1, KCl, 20 mmol l^-1; CaCl₂, 0.2 mmol l^-1; MgCl₂, 3 mmol l^-1; K-EGTA, 10 mmol l^-1; Na₂ATP, 3 mmol l^-1; Mg-GTP, 0.1 mmol l^-1; glutathione, 3 mmol l^-1; sucrose, 120 mmol l^-1; Hepes/bis-Tris, 10 mmol l^-1; pH 6.7, 500 mosmol l^-1. To record currents through Ca²⁺ channels, tetraethylammonium chloride (TEA-Cl, 10 mmol l^-1) was added to the external standard saline. In the pipette solution K⁺ ions were replaced by TEA or Cs²⁺; Cs-gluconate, TEA-Cl, Cs-EGTA and CsF replaced the corresponding K⁺ salts. All chemicals were purchased from Sigma (St. Louis, MO, USA).

Patch clamp data were analysed with PulseFit software (version 8.5, Heka) and Igor Pro (version 3.12; WaveMetrics Inc, OR, USA) on a Pentium-based PC. Statistical analyses were performed using Statistica (version 5.1, StatSoft Inc., Tulsa, OK, USA). Values are given as means ± standard errors of means (S.E.M.). Statistical significances between means were tested using Student's t-test; the level of significance was taken as P=0.05.

The staining procedure labelled somata and neurites of intrinsic and extrinsic mushroom body neurons, including axonal branches of antennal lobe projection neurons. The somata of projection neurons are located within the rim around the antennal lobes (Fig. 1). Whole-mount analyses revealed that rhodamine—dextran labelling of the axonal terminals yielded from 50 to several hundred stained somata in each antennal lobe. Each projection neuron sends one primary neurite into the neuropilar area of the antennal lobe, where it forms glomerular arborizations. The antennocerebralis tracts (ACT), which connect the antennal lobe with the protocerebrum, are densely stained. Neurites running within the median or the lateral ACT are labelled. Thus, somata of both ACT neurons were cultured.

Within the protocerebrum other mushroom body-extrinsic neurons, which innervate the mushroom bodies are stained as well. Among these neurons, feedback neurons from extensive branches within all calycal regions (Grünewald, 1999a). Accordingly, feedback neurons are well stained by rhodamine—dextran; two somata clusters were visible in most specimens. In addition, the dye labelled numerous Kenyon cells in each mushroom body. Cultures prepared from mushroom bodies contain only somata of Kenyon cells and some of them were labelled. In cultures prepared from antennal lobes only projection neurons were labelled, because only the antennal lobes were dissected out of the brain, and the somata of other mushroom body-extrinsic neurons, such as feedback neurons, were not cultured.

All cells were treated identically before, during and after the neurons were cultured. For cultures of Kenyon cells the yield from one mushroom body was plated onto 10 dishes, and the yield from 2-3 antennal lobes plated onto one dish. Thus, the cell density of the Kenyon cell and the projection neuron cultures were similar and adjusted so that the neurites of individual neurons did not overlap. The labelling remained visible in cultured neurons, allowing cell identification throughout the period of observation. Within each dish, 2-10 labelled projection neurons (or numerous Kenyon cells) were identified after 2-6 days in vitro. Neurons started to sprout new and fine processes after 1 day in vitro and continued to grow until day 6, the last day of observation (Fig. 2). The somata diameters of cultured Kenyon cells measured approximately 7-10 μm, as in the intact pupal mushroom body. The diameters of somata from projection neurons were larger (10-25 μm), similar to their size in the intact brain (Schröter and Malun, 2000), and the area covered by their branches were larger than those of Kenyon cells. Individual branches of cultured honeybee neurons were very thin, especially those of Kenyon cells, with a diameter of less than 1 μm. We did not quantify potential morphological differences between the neuron types in vitro, because most neurons were used for electrophysiological experiments before they grew elaborate neuritic branches.

Whole-cell patch clamp recordings were performed from projection neurons, identified from their fluorescence, and from Kenyon cells, which were derived from pure mushroom body cultures. Cells were measured electrophysiologically in vitro between days 3 and 5. Older neurons had rather elaborate neuritic branches, which would compromise the control of the membrane potential (space clamp). Thus, a total of 53 cells (22 Kenyon cells, 31 projection neurons) were measured. The capacitance of Kenyon cells was 4.1±0.27 pF (N=22) and 11.30±0.68 pF for projection neurons (N=31). Kenyon cells have significantly smaller cell capacitances because of their smaller somata diameters (P<0.001, t=-8.56, d.f.=51, Student's t-test).

From the holding potential of -80 mV a hyperpolarizing prepulse was applied to -120 mV to completely remove inactivation of the transient A type K⁺ current (cf. Pelz et al., 1999). Whole-cell membrane currents induced by depolarizing voltage pulses of Kenyon cells and projection neurons were dominated by large voltage-sensitive outward currents (Fig. 3). Depolarizing voltage pulses also induced a rapidly activating transient inward current in most cells, which was blocked in all experiments by switching to an external saline containing 100 nmol l^-1 TTX. Kenyon cells and projection neurons expressed different types of outward K⁺ currents. Schäfer et al. (1994) identified that these outward currents are carried by K⁺ ions, by altering the external K⁺ concentration and determining the tail current reversal potential. The outward K⁺ currents may comprise a rapidly activating transient and a sustained component (delayed rectifier, I_K,V). The transient K⁺ current (A type current, I_K,A) of the Kenyon cells was more pronounced than the I_K,A of the projection neurons, as can be seen when comparing traces of typical currents of two representative cells (Figs 3, 5). The transient current component can be blocked in Kenyon cells by a depolarising prepulse to -20 mV (Pelz et al., 1999). Accordingly, inactivating prepulses reduced the amplitude of the transient K⁺ currents in Kenyon cells, but did not affect whole cell K⁺ currents of projection neurons (Fig. 3). Subtraction of these currents yielded the pure transient K⁺ current (Fig. 3C). The I_K,A was a large portion of the total K⁺ current of Kenyon cells, whereas such transient K⁺ currents were negligible in projection neurons. The ratio of transient over sustained K⁺ current, as revealed by calculating the quotient of the maximum current at the current onset and at the end of the depolarizing pulse (command potential +50 mV), was higher in Kenyon cells than in projection neurons (P<0.0001, t=8.95, d.f.=30, Student's t-test, N=10 Kenyon cells, 22 projection neurons; Fig. 4C).

The K⁺ current amplitudes of the projection neurons were higher than those of Kenyon cells (Fig. 4A). Measured at a pulse potential of +50 mV, the mean peak outward current (measured a few ms after current onset) of projection neurons measured 7376.5±117.2 pA and was significantly higher than those of Kenyon cells (2907.2±67.1 pA; P<0.001, t=3.65, d.f.=30, Student's t-test). Similarly, when measured at the end of the depolarizing pulse, the amplitude of the sustained current component was higher in projection neurons (P<0.001). This was only partially due to the larger soma diameter of projection neurons, since the comparison of the current densities (pA/pF) revealed differences for the sustained K⁺ current (567.0±46.0 pA/pF for projection neurons versus 326.9±50.2 pA/pF for Kenyon cells, P<0.005, t=3.18, d.f.=29), but not for the transient K⁺ current (Fig. 4B, P=0.29). This indicates that the total K⁺ current consisted mainly of non-inactivating currents and only a small transient component, whereas that of Kenyon cells was a mixture of inactivating and non-inactivating currents with a pronounced transient current.

The I-V relationship of the outward currents showed current activation at approximately -45 mV (Figs 5C, 6). The I-V curves of projection neurons and Kenyon cells differed when whole-cell currents were measured without blockade of the calcium currents by external Cd²⁺. Kenyon cells expressed a linear voltage-dependency of K⁺ currents at positive membrane potentials. By contrast, projection neurons showed a pronounced N-shaped I-V curve with a local current peak at +60 mV and a local minimum at +80 mV (Figs 5C, 6). When voltage-sensitive Ca²⁺ currents were blocked with external 50 μmol l^-1 Cd²⁺ the outward currents of projection neurons were significantly reduced. By contrast, Cd²⁺ did not affect or only slightly increased the outward K⁺ currents of Kenyon cells (Fig. 5B,C). In addition, external Cd²⁺ irreversibly transferred the N-shaped form of the I-V curve of projection neurons into a linear one (Fig. 5C), indicating that Cd²⁺ blocked a calcium-sensitive outward current in projection neurons but not in Kenyon cells. Typically, such a calcium-dependent N-shaped I-V curve is caused by the activation of calcium-dependent K⁺ currents.

The differences in Ca²⁺-dependent K⁺ currents may reflect differences in the voltage-sensitive Ca²⁺ currents. Therefore, we measured the voltage-sensitive Ca²⁺ currents of Kenyon cells and projection neurons. For this, currents through calcium channels were isolated by blocking outward K⁺ currents with external TEA and Cs²⁺ in the patch pipette and voltage-sensitive Na⁺ currents with external TTX. The remaining inward currents were currents through Ca²⁺ channels (Schäfer et al., 1994). Under these conditions, depolarizing command potentials activated inward currents at command potentials higher than -40 mV in both neuron types (Figs 7, 8). These currents through calcium channels showed rapid activation and slow inactivation (Fig. 7A). The peak current amplitude ranged between -38.6 and -419.9 pA (mean -197.3±30.9 pA, N=12) for Kenyon cells. Projection neurons expressed Ca²⁺ currents with peak amplitudes of -292.6 to -1166.2 pA (mean -683.0±30.9 pA, N=9, Fig. 8B). When equimolar barium was used a charge carrier instead of calcium, the whole-cell currents through calcium channels did not change substantially (not shown). Instead, current amplitudes were only slightly increased and the current decay during depolarizing was almost indistinguishable (not shown). The Ca²⁺ currents were sensitive to externally applied 50μmol l^-1 Cd²⁺, a concentration that was shown to be sufficient to completely and irreversibly block Ca²⁺ currents in Kenyon cells (Schäfer et al., 1994). However, in projection neurons, a small Cd²⁺-resistant inward current remained unblocked at this Cd²⁺ concentration (Fig. 7B).

The voltage threshold for activation of inward Ca²⁺ currents was approximately -35 mV. The current—voltage relationship of the currents through Ca²⁺ channels peaked at command potentials between 0 and 10 mV and had a reversal potential at approximately 35-45 mV (Figs 7C, 8A). The overall shapes of the voltage-sensitive Ca²⁺ currents and the I-V curves of this current in Kenyon cells and projection neurons were similar (Fig. 8A). The mean peak current amplitudes, however, were significantly higher in projection neurons than in Kenyon cells (P<0.0001, t=-5.57, d.f.=19, Student's t-test; Fig. 8B). This is probably due to their larger soma diameter, because the mean peak Ca²⁺ current densities (pA/pF) did not differ (P=0.29, t=-1.09, d.f.=19; Fig. 8C). Thus, Kenyon cells and projection neurons express similar voltage-sensitive Ca²⁺ currents with similar current densities.

In this study we have analysed voltage-sensitive ionic currents of identified neuron types of the honeybee brain, namely mushroom body Kenyon cells and antennal lobe projection neurons, which had been selectively labelled and identified prior to recording. The analysis revealed pronounced differences between these neurons, in particular between transient voltage-sensitive K⁺ currents and a calcium-dependent K⁺ current.

In order to assign physiological properties to a certain neuron class it is crucial to identify the neuron type. This is particularly true for antennal lobe cultures, which contain different types of neuron. Cultured projection neurons or Kenyon cells do not retain their in vivo morphology. This is probably not a consequence of the labelling procedure, because it has been described previously for cultures of unlabelled Kenyon cells (Kreissl and Bicker, 1992) and unidentified antennal lobe neurons (Devaud et al., 1994; Kirchhof and Mercer, 1997). Also, in the present study branching patterns of labelled and unlabelled neurons within a given culture dish were indistinguishable. Thus, in contrast to cultures of, for example, antennal lobe neurons of the moth Manduca sexta (Hayashi and Hildebrand, 1990; Oland et al., 1996; Mercer and Hildebrand, 2002a,b), labelling is required to differentiate between cultured local interneuron and projection neurons in the honeybee. The identification of antennal lobe projection neurons was achieved by rhodamine dye injections into the mushroom body calyces. The staining within the antennal lobes presented here is very similar to previous stainings observed with rhodamine—dextran into the mushroom body (Schröter and Malun, 2000). The morphology of the projection neurons within the honeybee brain has been described in detail elsewhere (Homberg, 1984; Gascuel and Masson, 1991a; Fonta et al., 1993; Schröter and Malun, 2000; Abel et al., 2001). The staining procedure that we used labelled other neurons as well as antennal lobe projection neurons and Kenyon cells, including several known mushroom body extrinsic neurons (Mobbs, 1982, 1985; Rybak and Menzel, 1993), mushroom body feedback neurons (Bicker et al., 1985; Grünewald, 1999a) and, occasionally, a few somata within the suboesophageal ganglia. However, staining of these neurons did not interfere with the identification of projection neurons in vitro, because in the antennal lobes only projection neurons could take up the tracer and were, therefore, the only labelled cells in these cultures.

Cultured dextran-labelled neurons have been used successfully for patch clamp recordings; their viability has repeatedly been noted (e.g. Kloppenburg and Hörner, 1998; Kloppenburg et al., 1999a; Dugladze et al., 2001) and was confirmed here. Projection neurons grew extensive neurites in culture throughout the period of observation (3-6 days), as has been described for unidentified antennal lobe neuron cultures (Gascuel and Masson, 1991a; Devaud et al., 1994; Kirchhof and Mercer, 1997), which suggests that they were healthy and that the dye did not impair the viability of the cells. The applied mass staining technique is therefore a very reliable method for the identification of honeybee antennal lobe projection neurons in vitro and may be used in the future for analyses of other identified neurons of the bee brain, such as mushroom body feedback neurons.

The major finding of this study is that antennal lobe projection neurons and mushroom body Kenyon cells express different sets of ionic currents.

Kenyon cells express at least three different voltage-gated outward K⁺ currents. The transient component comprises a rapidly inactivating (A-type) current and a slowly inactivating current, whose kinetic parameters have been described in great detail (Pelz et al., 1999). The amplitude of the sustained (delayed rectifier) current is relatively small compared to the transient K⁺ currents, as described in various insect or crustacean neurons (e.g. Byerly and Leung, 1988; Saito and Wu, 1991; Hayashi and Levine, 1992; Delgado et al., 1998; Kloppenburg and Hörner, 1998; Benkenstein et al., 1999; Kloppenburg et al., 1999b; Schmidt et al., 2000 (for a review, see Wicher et al., 2001) and in antennal motoneurons within the honeybee deutocerebrum (Kloppenburg et al., 1999a). By contrast, projection neurons do not express such prominent transient K⁺ currents. This is interesting, because computer simulations using Hodgkin—Huxley-derived equations indicated that A-type K⁺ currents are mainly responsible for the membrane repolarisation during an action potential in honeybee Kenyon cells (Pelz et al., 1999; Ikeno and Usui, 1999). Similarly, shaker mutations in Drosophila impair spike repolarisation (Tanouye and Ferrus, 1985). Therefore, other outward currents (calcium-dependent or delayed rectifier currents) must be responsible for spike repolarisation in honeybee projection neurons. Supporting evidence for different expressions of K⁺ channels comes from immunohistochemical studies of the Drosophila brain showing that shaker channel proteins are highly expressed in the mushroom body neuropil, but not in the antennal lobes (Rogero et al., 1997). Furthermore, axons of motoneurons and mechanosensory neurons within the antennal nerve of the fly are shaker-immunoreactive, consistent with the presence of transient K⁺ currents in antennal motoneurons in honeybees (Kloppenburg et al., 1999a).

Ca²⁺-dependent K⁺ currents have been described in a variety of insect neurons (e.g. Thomas, 1984; Nightingale and Pitman, 1989; David and Pitman, 1995; Grolleau and Lapied, 1995; Mills and Pitman, 1999; Hewes, 1999) (for reviews, see Saito and Wu, 1991; Wei et al., 1994; Grolleau and Lapied, 2000; Wicher et al., 2001). They may provide many functional roles within the nervous system, including spike repolarisation (e.g. Lapied et al., 1989) and afterhyper-polarisation (Saito and Wu, 1993; Hu et al., 2001). Ca²⁺-dependent K⁺ currents can be blocked by Ca²⁺ channel blockers (typically Cd²⁺ in insect neurons). This treatment reduced the amplitude of outward currents in projection neurons, but not in Kenyon cells. In addition projection neurons, which express voltage and Ca²⁺-dependent K⁺ currents, show a nonlinear I-V relationship at positive command potentials, because the K⁺ current amplitude is influenced by the activation of voltage-sensitive Ca²⁺ channels. Assigning the honeybee Ca²⁺-dependent K⁺ current to any of the identified insect or vertebrate channels is currently not possible. It remains to be analysed whether the currents are both calcium- and voltage-dependent and whether honeybees express two separate Ca²⁺-dependent K⁺ channels as in DUM neurons of Periplaneta (Grolleau and Lapied, 1995) or Drosophila `giant' neurons (Saito and Wu, 1991).

Probably all insect neurons express voltage-sensitive Ca²⁺ channels. These currents may contribute to action potential generation, synaptic transmission or neuromodulation (for reviews, see Jeziorski et al., 2000; Wicher et al., 2001). The voltage-sensitive Ca²⁺ currents of the honeybee projection neurons and Kenyon cells are similar (with respect to steady-state activation, Cd²⁺-sensitivity and inactivation) to those described in other insect preparations, e.g. Drosophila (Byerly and Leung, 1988; Saito and Wu, 1993; Schmidt et al., 2000), Periplaneta (Grolleau and Lapied, 1996; Wicher and Penzlin, 1997; Mills and Pitman, 1997), Manduca (Hayashi and Levine, 1992), Gryllus (Kloppenburg and Hörner, 1998) and locusts (Laurent et al., 1993; Pearson et al., 1993). They activate rapidly and show a slow inactivation. The functional properties of the Ca²⁺ currents of Kenyon cells and projection neurons presented here are thus consistent with earlier descriptions of Kenyon cells (Schäfer et al., 1994). The Ca²⁺ current densities of Kenyon cells and projection neurons are similar, which implies that the observed differences in Ca²⁺-dependent K⁺ currents are not due to differing densities of Ca²⁺ channels, but may indeed represent differential expression of Ca²⁺-dependent K⁺ channel proteins in Kenyon cells and projection neurons.

Although the present results are largely consistent with earlier investigations (Schäfer et al., 1994), there are differences as well. The finding that Kenyon cells did not show pronounced calcium-dependent K⁺ currents, contradicts the report of these currents by Schäfer et al. (1994). What could be the reasons for this discrepancy? We exclude differences in Ca²⁺ channel expression between both studies, because the Ca²⁺ currents are very similar with respect to activation threshold (approximately -40 mV) and peak current amplitudes (means -160 pA and -197 pA). The culture conditions have been changed since the study by Schäfer et al. (1994), e.g. the components and pH of the medium. However, it is implausible that these changes abolish Ca²⁺-dependent K⁺ currents only in Kenyon cells and not in projection neurons, because both neuron types were cultured and measured in parallel under the same conditions. The recordings by Schäfer et al. (1994) were performed solely from the somata, because only cells without any processes were used, and the neurons were cultured for only 12-36 h. In the present study neurons were allowed to grow for 3-6 days and formed small neurites. Therefore, the whole cell currents may comprise somatic and neuritic currents. Accordingly, Kenyon cells may express somatic, but not neuritic Ca²⁺-dependent K⁺ currents. It is interesting to note here that Pelz et al. (1999) also observed differences in the pharmacology and kinetics of the A type K⁺ current as compared to Schäfer et al. (1994), but the ultimate reasons for these differences remain unclear.

Differences in the ionic currents between Kenyon cells and projection neurons indicate that a specific expression pattern is maintained throughout the culturing procedure and for several days in vitro. Therefore, the observed differences may in fact represent physiological differences of these neuron types in vivo. In the living honeybee brain neither the function nor the dendritic or axonal localization of the various voltage-sensitive ionic currents has yet been satisfyingly unraveled. Both Kenyon cells and projection neurons generate action potentials (Homberg, 1984; Hammer and Menzel, 1995; Abel et al., 2001; Müller et al., 2002), but differ with respect to the outward currents, which probably mediate spike repolarisation. Thus, different ionic currents may interact during spike generation in the two neuron classes. Both the antennal lobe and the mushroom body are involved in olfactory learning and memory formation (Masuhr and Menzel, 1972; Hammer and Menzel, 1998; Menzel, 1999) and both are innervated by modulatory octopaminergic neurons (Hammer, 1993, 1997; Kreissl et al., 1994). Additional work is required to reveal the functional significance of the differential expression of K⁺ currents of the different neuron types within the central olfactory pathway of the honeybee and whether these currents are differentially modulated during olfactory learning.

The author wishes to thank Holger Anlauf and Marion Ganz for skilful technical assistance with neuron staining protocols and cell culture techniques. I am grateful to Dr Einar Heidel for critically reading the manuscript, Mary Wurm for help with the English version, and Dr Randolf Menzel for continuous support and many valuable discussions at all steps of this work. This work was supported by the Deutsche Forschungsgemeinschaft, SFB 515/C5.