Locust Medial Neurosecretory Cells in vitro: Morphology, Electrophysiological Properties and Effects of Temperature

The medial neurosecretory cells (MNSC) of the pars intercerebralis of insects influence a variety of physiological activities ranging from moulting (Wigglesworth, 1940) and reproduction (Girardie, 1966) to the regulation of metabolism (Goldsworthy, 1969; Goldsworthy et al. 1977) and they are therefore of considerable biological significance. The important role of this neurosecretory system was demonstrated by Kopec (1917) with the discovery of a brain hormonal effect in a caterpillar. Wigglesworth (1940, 1964) demonstrated the release of a brain hormone, from the pars intercerebralis, which regulates moulting in the insect Rhodnius prolixus. The morphology and anatomy of the pars intercerebralis and the retrocerebral neuroendocrine complex of locusts has been investigated in detail by Girardie and Girardie (1967) and Mason (1973).

Dissociated primary cell cultures provide a useful system to investigate the morphology, function and pharmacology of neurones in a controlled environment. Chen and Levi-Montalcini (1969) introduced an in vitro system able to promote axonal outgrowth and cell migration from the embryonic nervous system of cockroaches. Subsequent research on insect nerve cell cultures mainly focused on embryonic growth of the nervous system (for a review, see Beadle and Hicks, 1985). Seshan et al. (1974) established a culture system for larval and adult cockroach medial neurosecretory cells in combination with embryonic organs or ovaries of the same species. Recently, it has also been demonstrated that the presence of haemolymph in the culture medium can significantly improve neurite outgrowth in dissociated neuronal cell cultures of adult Periplaneta americana (Howes et al. 1991), Locusta migratoria and Schistocerca gregaria (Kirchhof and Bicker, 1992).

The release of neurohormones from neurosecretory cells in vivo is elicited by the arrival of action potentials in the terminals; in vitro it can be elicited by depolarization with high external potassium levels (Maddrell and Gee, 1974). Thus, the electrophysiological properties of neurosecretory cells are of great importance for their function. Neurosecretory activity is known to be affected by environmental temperature and light conditions (Lees, 1963, 1964). Cook and Milligan (1972) found that light affects the firing frequency and membrane potential of medial neurosecretory cells in Periplaneta americana. However, very little is known about the function of the pathways mediating between the environmental and internal inputs and the activity of neurosecretory cells.

The aim of the present work was to establish a primary culture system for larval and adult locust medial neurosecretory cells in order to facilitate electrophysiological investigations. The factors contributing to cell growth and morphology were studied. The function of isolated medial neurosecretory cells was investigated with electrophysiological recordings using the patch-clamp technique. Since temperature is an important variable affecting the release of neurohormones, we examined the direct temperature-dependence of some electrophysiological properties of medial neurosecretory cells.

All experiments were performed using both larvae (third to fifth instar) and adults of Locusta migratoria L. of both sexes reared in a crowded colony under constant conditions in the laboratory. The animals were kept at 35°C (day)/25°C (night) and 30% relative humidity with a 13h:11h light:dark cycle.

After the animals had been anaesthetized with CO₂, the head was severed from the body and fixed on a plate. The head and plate were then surface-sterilised by rinsing in 70% alcohol. Working in a laminar flow cabinet and using sterile dissecting techniques, a piece of cuticle was dissected away from between the compound eyes. The supraoesophageal ganglion (Fig. 1) was removed from the head and immediately transferred to a dish (Nunc organ culture dish, 35mm×10mm) containing fresh medium. The medium was similar to that used by Howes et al. (1991), consisting of five parts of Schneider’s Drosophila medium and four parts of Minimum Essential Medium (MEM) with Hanks’ salts (5+4 medium) and containing 25mmol l⁻¹ Hepes (Gibco). The only additives were 100i.u.ml⁻¹ penicillin/streptomycin and 0.25 μg ml⁻¹ Fungizone (Gibco). The ganglia were dissected free from adhering fat and tracheae, and the ganglion sheath was removed with very fine forceps. After rinsing, the ganglia were transferred to another dish containing fresh medium. In the desheathed ganglia, the two large groups of medial neurosecretory cells (MNSCs) in the pars intercerebralis become very distinct under dark-field illumination (Thomsen and Thomsen, 1954) (Fig. 1). The protocerebral lobes were carefully stretched to expose the pars intercerebralis. The two clusters of MNSCs were dissected out by cutting the fibre tracts with very fine forceps. The ganglionic debris was removed from the dish with forceps. The average size of one group of MNSCs in adults was 0.22mm in the longitudinal axis and the maximum lateral extension was 0.17mm; in third-instar larvae, a length of 0.15–0.17mm and a width of 0.1mm was measured. Similar values are described in histological investigations of the pars intercerebralis of Locusta migratoria carried out by Girardie and Girardie (1967), who counted a total of 570 medial neurosecretory cells in adult locusts and 492 in the fourth larval instar. The two groups of MNSCs were mechanically dissociated in the centre of the dish using very fine sterile insect mounting pins. The dishes were then transferred to a humid incubator chamber at 30°C; the cultures were incubated either in air or with 5% CO₂ in 95% air. The medium was partly changed for fresh medium every 7 days, in some cases every 14 days.

Electrophysiological recordings were carried out using the patch-clamp technique in the whole-cell configuration (Hamill et al. 1981) and a conventional patch-clamp tower (Luigs and Neumann Co.). We used fire-polished electrodes with a resistance of 2–5 MD. Measurements were performed in culture medium (five parts of Schneider’s Drosophila medium and four parts of MEM with Hanks’ salts containing 25mmol l⁻¹ Hepes obtained from Gibco). Our standard pipette solution (I) contained (in mmol l⁻¹): potassium gluconate, 180; NaCl, 10; CaCl₂, 1; MgATP, 2; EGTA, 10; MgCl₂, 2; and Hepes, 10; adjusted to pH7.3. In some experiments, we used a second pipette solution (II) containing in (mmol l⁻¹): KCl, 145; MgATP, 2; EGTA, 2; and Hepes, 10. The CED system hardware and software packages were used in combination with the EPC7 patch-clamp amplifier (List Electronics) for stimulus generation and for storage and analysis of data. For current–voltage relationships, voltage steps of 10mV were applied from different holding potentials. Current-clamp measurements were carried out in the current-clamp mode of the EPC7. For depolarization of cells, we injected currents of 0.1–0.5nA for 50ms. Membrane capacity was determined using the compensation circuit of the EPC7. Some experiments were performed with intracellular electrodes filled with 3 mol l⁻¹ KCl (resistance 10 MD). Temperature was controlled and regulated by a ‘Temperature Controller’ (Luigs and Neumann Co.) provided with two Peltier heat-exchange elements. The temperature was automatically regulated to a preset value by a feedback loop. The temperature controller was connected to a tape recorder (Sony instrumentation tape recorder) for permanent recording of the temperature.

Depending on the efficiency of the dissociation process, the MNSCs were either completely isolated or formed smaller or larger groups. The proportion of isolated cells was usually high. After dissociation, the soma were spherical and some of the cells retained a primary axonal stump (10–20% of the isolated cells). Short primary axonal stumps often retracted within the first 4 days in culture, while longer primary neurites often started to sprout new processes (Figs 2A and 3B). The majority of the cells adhered to the culture dish within the first hour of culture. In larvae (third to fifth instar) and adults, the diameter of the cell bodies ranged from 10 to 70 μm. Diameters between 60 and 70 μm were found in only a few cases; most of the cells had diameters between 15 and 30 μm (mean value 26 μm, N=58). The cells were maintained in culture for up to 2 months.

New processes started to sprout from the somata or from primary neurite stumps within the first 2 days in culture. The percentage of isolated cells showing outgrowth was around 80% in the third larval instar and about 60% in instars 4 and 5 and in adults. Since this was the only obvious difference between larvae and adults that we observed, most morphological observations on larval and adult tissue are considered together.

Incubation with 5% CO₂ in the atmosphere significantly improved viability and outgrowth of the cells in culture. The effect of CO₂ was compared in identically treated cultures of MNSCs from third-instar larvae observed for 30 days. The two samples were kept simultaneously in two incubators. During the first 5 days in culture, the percentage of cells showing outgrowth was around 80% in both cases. After 2 weeks in culture, about 90% of the isolated cells in the samples without CO₂ started to deteriorate and retract their processes. In the presence of CO₂, most of the isolated cells retained their processes over the whole observation period. In undissociated groups of MNSCs, survival of the cells and processes was better, even without CO₂ in the atmosphere.

Cultured MNSCs showed a variety of morphologies. When a long primary axonal stump was retained after the dissociation process, a unipolar cell morphology was often promoted (Figs 2A and 3B; see also Kirchhof and Bicker, 1992); in other cases, most MNSCs tended to become multipolar (Figs 2B,C and 3A,C,D). Cells with short primary neurites usually became multipolar (Fig. 3C). After 4 days in culture, varicosities appeared on some of the processes (Figs 2D, 3D and 4). In many cases, the number and size of these axonal swellings increased with further growth. Morphological contacts between isolated MNSCs were often observed after about 3 days in culture (Figs 2B,E,F and 3D). Contacts occurred more frequently within clusters of MNSCs (Figs 2E,F and 3D). Processes of isolated cells were usually thin, while cells within or between groups of MNSCs formed larger neurite bundles. In undissociated tissue, cells with a flattened appearance and with granular inclusions were observed within cell groups and along large processes. Since any adhering tissue was carefully removed from the ganglion during dissection, we propose that these cells were of glial origin (Fig. 2I,J). In the presence of such glial cells, growth of large neurite bundles was promoted; the growth rate could be fast (up to 150 μm in a single day) and the overall length of processes was high (up to 800 μm). Groups of MNSCs containing these putative glial cells also survived for longer in culture (Fig. 4E). In one case, the two large processes emerging from a small group of MNSCs oriented along a scratch caused by an insect mounting pin on the bottom of the dish (Fig. 2H). The pattern of growth from a group of about 30MNSCs from day 7 to day 32 is shown in Fig. 4A–D. In isolated cells and groups of MNSCs, the growth rate was highest during the first 5 days in culture (up to 150 μm day⁻¹). After 5 days in culture, some of the processes continued to grow, while some higher-order side branches shortened. During this phase, the rate of growth was usually lower. In a group of MNSCs observed for 25 days after the initial growth phase, the average rate of growth was 10±5 μm day⁻¹.

To study the electrophysiological properties of single MNSCs, we performed recordings from isolated cells with the patch-clamp technique in the whole-cell configuration (Hamill et al. 1981). All experiments were carried out with isolated MNSCs of adult locusts in order to exclude variations caused by functional changes during larval development. After establishment of a giga-seal on the soma of the cell, the patch membrane was disrupted by suction. Successful recordings were obtained from 77 cells. Cells could be held for 1h or more. Most recordings were carried out between day 3 and day 7 in culture. The stability of the electrode potentials was checked outside the cell before and after recording. Measurements were performed directly in the medium. Most recordings were made with pipette solution I.

Both spiking cells and nonspiking cells were found. Immediately after dissociation and during the first 2 days in culture, most MNSCs did not show spiking activity and some had very low membrane potentials. Between day 3 and day 7, the ‘resting’ membrane potential was in most cases between −15 and −50mV and the percentage of cells showing spiking activity in response to a depolarizing current rose to become significantly higher (approximately 50% of the recorded cells). Membrane capacities between 6pF (small-diameter cells) and 100pF (large-diameter cells) were measured (N=21; 36±21pF). No obvious correlation between the membrane capacity (and the corresponding diameter of the cells) and spiking activity could be observed.

In spiking cells, the action potentials typically had long durations (between 3 and 12ms, depending on the temperature; see below), which is known to be a characteristic of neurosecretory cells in general (Gosbee et al. 1968; Cook and Milligan, 1972; Seshan et al. 1974; Rowell, 1976; Raabe, 1983). Spike amplitudes between 10mV and 80mV were recorded.

Fig. 5 demonstrates typical spike responses in current-clamp recordings from isolated MNSCs. The membrane potential was in all cases set to −30mV and the cells were stimulated with a depolarizing current pulse of approximately 0.5nA and of 50ms duration. About 50% of the spiking cells responded with a low-amplitude action potential (10–20mV; Fig. 5A), others discharged with a single high-amplitude spike (up to 80mV; Fig. 5B), and a number of cells showed continuous spiking activity when the membrane potential was set to −30mV (Fig. 5C). These cells often responded with more than one action potential to the stimulus pulse.

Since temperature is an important variable affecting neurosecretory activity (Lees, 1963), we examined its effect on the electrophysiological properties of MNSCs. In Fig. 6A,B, examples of current-clamp recordings from an MNSC are shown under similar stimulus conditions at temperatures of 20°C and 34°C. The ‘resting’ membrane potential of this cell was −40mV and the membrane potential was set to between −50mV and −20mV. A depolarizing current pulse of approximately 0.5nA with a duration of 50ms was applied. At 34°C, the cell discharged with 2–5 spikes per stimulus, depending on the membrane potential, and became spontaneously active at a potential of −30mV (spike frequency about 15Hz). The duration of action potentials at the ‘resting’ membrane potential level (−40mV) was between 3 and 4ms at this temperature. The time interval between the beginning of the fast rising phase and the corresponding point on the repolarizing phase of the action potentials was determined. At −20mV, the fast inward current for the generation of action potentials became inactivated (complete inactivation was between −15mV and −10mV in all cases). At 20°C, spike duration was increased to 12ms at a membrane potential of −40mV and the cell always responded to the current pulse with a single action potential. No spontaneous discharge was observed at a potential of −30mV. In Fig. 6B, the responses of the cell starting from the ‘resting’ membrane potential (−40mV) at the two temperatures are shown enlarged and superimposed. At 20°C, the spike number was lower, the spike amplitude was higher and the passive and active components of the response were clearly slower.

With a change of temperature, there was also a consistent membrane potential change. The cells hyperpolarized on heating and depolarized on cooling (Fig. 7A,B). In Fig. 7A, the change in membrane potential with a change in temperature was recorded in a single experiment over a period of 20min. The change in potential with temperature was on average 0.78+0.2mV°C⁻¹ (N=9). Fig. 7B demonstrates the distribution of membrane potentials recorded at constant temperatures of 20°C and 34°C. The electrode potential was checked outside the cell before and after recording. The mean membrane potential of the recorded MNSCs was −19.4±8.1mV at 20°C (N=34) and −30.8±10.3mV at 34°C (N=24). The change of potential with temperature calculated from the difference between these values is 0.8mV°C⁻¹. We found similar results using intracellular electrodes filled with 3mol l⁻¹ KCl and with a resistance of 10 MΩ.

In Fig. 8, a typical whole-cell voltage-clamp recording shows the inward and outward currents elicited in an MNSC producing high-amplitude action potentials at two temperatures (20°C and 34°C). The holding potential was −60mV and the membrane potential was set to between −40mV and +30mV for 100ms in 10mV voltage steps. The time course of the currents was clearly slower at the lower temperature.

Medial neurosecretory cells of the pars intercerebralis of larval and adult Locusta migratoria can be isolated from the supraoesophageal ganglion purely by mechanical dissociation and without treatment with enzymes. Depending on the efficiency of the dissociation process, the resulting population of isolated cells is relatively numerous. The culture medium described by Howes et al. (1991) for in vitro growth of differentiated neurones from the thoracic ganglia of Periplaneta americana also proved to be suitable for larval and adult locust MNSCs. MNSCs prepared in this manner are well-suited for patch-clamp recordings. When established, the culture system facilitates investigations on the electrophysiological properties of isolated neurosecretory cells.

In contrast to MNSCs in situ (Mason, 1973), many cells in vitro become multipolar. As has been reported by Kirchhof and Bicker (1992) for neurones isolated from meso-and metathoracic ganglia of Locusta migratoria, the presence of a primary neurite stump influenced the morphology of the cells in culture. We found that the percentage of cultured MNSCs showing outgrowth was significantly higher in third-instar larvae than in fourth and fifth instars and adults. Outgrowth and survival were promoted by incubation with CO₂ and also when large groups of MNSCs were grown together with cells of probable glial origin. In contrast to our present findings in Locusta migratoria, Seshan et al. (1974) demonstrated that significant outgrowth from groups of MNSCs from cockroaches was only achieved when MNSCs were co-cultured with embryonic nervous tissue, embryonic foregut explants or ovarian follicles. Kirchhof and Bicker (1992) also reported that, in cultured neurones from adult Schistocerca gregaria, less outgrowth occurred than for Locusta migratoria neurones and that growth was promoted by a haemolymph-conditioned medium, as previously described by Howes et al. (1991). In preliminary experiments, we found that haemolymph-conditioning of the medium using the method described by Howes et al. (1991) also seemed to promote growth of MNSCs isolated from adult locusts.

The electrophysiological results obtained from cultured MNSCs are in several ways consistent with reported findings from in situ recordings of insect MNSCs. After an initial phase of 3 days in culture, a considerable number of MNSCs produced action potentials (amplitudes between 10 and 80mV) with a typically long duration (3–4ms at 34°C, up to 12ms at 20°C) and some of the cells started to discharge continuously when the membrane potential was set to −30mV. ‘Resting’ membrane potentials between −15mV and −50mV were most frequent in our recordings. Cook and Milligan (1972) found spontaneous spiking activity in more than 50% of the MNSCs recorded in situ in Periplaneta americana (intracellular recordings from the soma) and reported ‘resting’ membrane potentials between −12mV and about −50mV (when the neuronal sheath potential was substracted). ‘Resting’ membrane potentials between −20mV and −50mV and spike amplitudes of about 60mV are mentioned by Raabe (1983). Action potentials with a relatively long duration and slow rise time are characteristic of neurosecretory cells (3–6ms and more compared with 0.6–2.5ms for other neurones) (Gosbee et al. 1968; Cook and Milligan, 1972; Normann, 1973; Rowell, 1976; Raabe, 1983). Seshan et al. (1974) also observed long spike durations (2–7ms at a temperature of 25°C) in extracellular recordings from large groups of cultured MNSCs from cockroaches; however, attempts to record from isolated MNSCs in vitro have been unsuccessful.

The electrophysiological results indicate that MNSCs from the pars intercerebralis of Locusta migratoria are rather heterogeneous with respect to membrane potential and excitability. This seems also to be the case for MNSCs of Periplaneta americana as shown in in situ recordings from the somata (Cook and Milligan, 1972). It may be that our pipette solution (I) containing calcium and EGTA was not appropriate for buffering intracellular calcium to an optimal level in all recordings. This is especially important in the context of the investigations of Pitman (1979) and Thomas (1984), who demonstrated that the properties of some insect neurones are very dependent upon intracellular calcium concentration. Chelation of intracellular calcium by injection of citrate or EGTA can enable somata that are normally passive to generate calcium-dependent action potentials. Experiments are in progress to investigate in detail the ionic current components of the action potentials of locust MNSCs.

The influence of temperature on electrophysiological properties was also investigated in locust motoneurones by Heitler et al. (1977). The reported effects of temperature on spike variables and membrane potential are consistent with our results. Heitler et al. (1977) calculated slopes of potential versus temperature of between 0.46mV°C⁻¹ and 0.57mV°C⁻¹, depending on the type of neurone. In locust MNSCs, we measured a mean value of 0.78±0.2mV°C⁻¹ when the temperature was shifted while recording from single MNSCs. The mean value calculated from two samples of membrane potentials recorded at constant temperatures (0.8mV°C⁻¹) corresponds well with the value found in single recordings. However, the large spread of ‘resting’ membrane potentials may indicate that the cells cannot be regarded as a single population and the calculation of a mean value from such a sample may, therefore, be inappropriate. The effects on the spike variables and the membrane potential indicate that an increase in temperature mediates an increase in the excitability of the MNSCs. In cells showing spiking activity at low temperature, the spike number is increased with increasing temperature (Fig. 6). The spread of ‘resting’ membrane potentials in different cells may play an important role in this context. MNSCs that have a very low ‘resting’ membrane potential (−10mV to −20mV) and that are non-excitable at low temperatures may become excitable at higher temperatures. At low membrane potentials, all fast inward currents become inactivated. Complete inactivation occurs between −15mV and −10mV. An increase in temperature of 15°C causes an increase in the membrane potential of −12mV, which would be high enough to activate the fast inward currents. The total number of cells in the pars intercerebralis capable of generating action potentials may therefore be significantly increased by raising the temperature.

Since the release of neurohormones is elicited by spike activity of neurosecretory cells, the dependence of the electrophysiological properties of MNSCs on temperature is especially important for poikilotherms such as insects. Temperature and photoperiod are known to have a decisive role in insect growth and reproduction, two processes that are triggered by the activity of neurosecretory cells (Lees, 1963, 1964). Lees (1964) has suggested that neurosecretory cells in the pars intercerebralis may be implicated as both photoperiodic receptors and humoral effectors. Cook and Milligan (1972) demonstrated that light affects both the ‘resting’ and spike potentials of MNSCs of cockroaches, suggesting that MNSCs receive synaptic inputs. However, the effects of light may also be enhanced by transformation of light into heat and the change in temperature might contribute to the change in activity of neurosecretory cells. The pars intercerebralis and neighbouring regions in the supraoesophageal ganglion of larval and adult Locusta migratoria are extensively covered by a red-brown pigment which might play a role in this context. Our results indicate that, in addition to synaptic inputs of sensory information, the direct influence of temperature on the electrophysiological properties of MNSCs may contribute to changes in the release of neurohormones with changes in temperature.

We thank Professor Dr K. Kalmring and Professor Dr H. Wiegand for supporting this work. We are grateful to Professor Dr G. Heldmaier for the use of his cell culture laboratory, to D. Encke for help with the cell culture technique and to Mrs E. Schäfer and L. Koschubat for technical help. Two anonymous referees improved the paper with their constructive criticism. This work was supported by the DFG, Sonderforschungsbereich Ökophysiologie 305, A7.