Spatial profiles of store-dependent calcium release in motoneurones of the nucleus hypoglossus from newborn mouse

Abstract

Hypoglossal motoneurones (HMN) are selectively damaged in both human amyotrophic lateral sclerosis (ALS) and corresponding mouse models of this neurodegenerative disease, a process which has been linked to their low endogenous Ca²⁺ buffering capacity and an exceptional vulnerability to Ca²⁺-mediated excitotoxic events. In this report, we investigated local Ca²⁺ profiles in low buffered HMNs by utilizing multiphoton microscopy, CCD imaging and patch clamp recordings in slice preparations. Bath application of caffeine induced highly localized Ca²⁺ release events, which displayed an initial peak followed by a slow ‘shoulder’ lasting several seconds. Peak amplitudes were paralleled by Ca²⁺-activated, apamin-sensitive K⁺ currents (I_KCa), demonstrating a functional link between Ca²⁺ stores and HMN excitability. The potential involvement of mitochondria was investigated by bath application of CCCP, which collapses the electrochemical potential across the inner mitochondrial membrane. CCCP reduced peak amplitudes of caffeine responses and consequently I_KCa, indicating that functionally intact mitochondria were critical for store-dependent modulation of HMN excitability. Taken together, our results indicate localized Ca²⁺ release profiles in HMNs, where low buffering capacities enhance the role of Ca²⁺-regulating organelles as local determinants of [Ca²⁺]_i. This might expose HMN to exceptional risks during pathophysiological organelle disruptions and other ALS-related, cellular disturbances.

Cytoplasmic free calcium concentrations ([Ca²⁺]_i) are critical determinants of cell viability and contribute to a multitude of functions in neurones as well as other cell types (Friel & Tsien, 1992; Pozzan et al. 1994; Berridge et al. 1998). During neuronal activity, substantial fluctuations of [Ca²⁺]_i with time constants from milliseconds to hundreds of milliseconds can be observed (Bacci et al. 1999; Frermann et al. 1999; Kato et al. 1999). These dynamics of [Ca²⁺]_i reflect in part the spatio-temporal responses of postsynaptic signal cascades, including the regulation of Ca²⁺-dependent conductances and associated second messenger systems (Augustine et al. 1987; Zucker, 1999). For example, hypoglossal motoneurones discharge repetitive bursts of action potentials during rhythmic, respiratory-related activity (Smith et al. 1991; Mifflin, 1997; Lieske et al. 2000), and measurements in these motoneurones revealed large increases of [Ca²⁺]_i during inspiratory-related bursts (Ladewig & Keller, 2000). Of considerable interest is the heterogeneity between somatic and dendritic Ca²⁺ processing that is reflected by differences in the kinetics and voltage dependence of [Ca²⁺]_i.

The above findings raise the question: how are the dynamics of intracellular free Ca²⁺ concentrations regulated in motoneurones that are extremely active and whose burst-related Ca²⁺ oscillations have to follow maximum frequencies above 5 Hz. Earlier studies provided evidence that rapid Ca²⁺ oscillations in motoneurones are partially achieved by low endogenous Ca²⁺ buffering capacities (Lips & Keller, 1998), which enhance the efficiency of a given set of uptake and extrusion mechanisms. However, low buffering not only accelerates relaxation rates of Ca²⁺ transients, but also appears to increase the risk for Ca²⁺-mediated motoneurone damage during excitotoxic conditions and/or ALS-related pathophysiological states (DePaul et al. 1988; Alexianu et al. 1994; Reiner et al. 1995; Krieger et al. 1996; Vanselow & Keller, 2000). Previous studies therefore provided increasing evidence that cellular specializations that facilitate rapid dynamics of [Ca²⁺]_i also enhance the vulnerability of motoneurones during human ALS and related animal models of this neurodegenerative disease.

Until recently, the analysis of Ca²⁺ dynamics in hypoglossal motoneurones has mainly been directed at the temporal change of average [Ca²⁺]_i, in relatively large cellular compartments, such as the cell body (Lips & Keller, 1998; 1999). In the present investigation, we utilized a combination of multiphoton microscopy (MPM), CCD imaging and patch clamp recordings in slice preparations to examine Ca²⁺ signals in small compartments of hypoglossal motoneurones with high spatial and temporal resolution. More specifically, we evaluated the particular spatio-temporal pattern of store-operated Ca²⁺ release under different physiological and pharmacological conditions. Our measurements indicate that low endogenous Ca²⁺ buffering and a corresponding, low concentration of mobile Ca²⁺-binding proteins in the cytosol not only account for rapid relaxation times of Ca²⁺ transients, but also enhance the role of Ca²⁺-regulating organelles for localized control of [Ca²⁺]_i. During ALS-related cellular disturbances, this might enhance the selective vulnerability of motoneurones to organelle disruption and associated excitotoxic disturbances. Parts of this study have been published in abstract form (Lalley et al. 1999; Ladewig et al. 2000).

METHODS

Preparation of slices

In vitro brain stem slice preparations were obtained from newborn mice at postnatal day (P)2–5. Animal experiments were carried out in accordance with the recommendations of the European Commission (no. L358, ISSN 0378–6978). Animals were decapitated by a small guillotine, brains were removed and subsequently cooled to 4 °C. Transverse slices of the brain stem with a thickness of 200 μm were prepared (Lips & Keller, 1998, 1999). For simultaneous imaging and patch clamp recordings, preservation of functionally intact neurones close to the slice surface was essential. This was achieved by minimizing mechanical disturbances of tissue during slice preparations, performing isolation of slices at low temperatures (4 °C) and optimizing metabolic conditions by maximum oxygen supply. Slices were maintained at room temperature in continuously bubbled (95 % O₂, 5 % CO₂) bicarbonate-buffered saline (mm):118 NaCl, 3 KCl, 1 MgCl₂, 25 NaHCO₃, 1 NaH₂PO₄, 1.5 CaCl₂, 20 glucose) at pH 7.4 and investigated according to previously described methods (Lips & Keller, 1998). All experiments were performed at room temperature (21–23 °C).

Patch clamp recording

Prior to recording, the nucleus hypoglossus was visually identified by its location close to the nucleus vagus and the fourth ventricle. Somata and dendrites of hypoglossal motoneurones were discerned by infrared differential interference contrast (IR DIC) optics connected to video equipment (Dodt & Zieglgänsberger, 1994; Lips & Keller, 1998; Stuart et al. 1993). Hypoglossal motoneurones were visually identified by their somatic diameters of approximately 20 μm and extensive dendritic arborizations, and their location close to the central canal (Lips & Keller, 1998, 1999; Ladewig & Keller, 2000). To assure acceptable optical conditions, patch clamp experiments were performed on cells within 60 μm below the slice surface. Patch clamp recordings from slice preparations, as previously described (Hamill et al. 1981; Edwards et al. 1989; Titz & Keller, 1997), were made with intracellular pipette solutions containing (mm): 140 KCl or potassium gluconate, 10 Hepes, 2 MgCl₂, 4 Na₂-ATP, 0.4 Na₂-GTP (adjusted to pH 7.3 with KOH or CsOH). Fura-2 and Calcium Green-1 were purchased from Molecular Probes (Eugene, OR, USA) and used in concentrations of 100 μm in the pipette solution. Patch pipettes were pulled from borosilicate glass tubing (Hilgenberg, Malsfeld, Germany) and heat polished before use. When filled they displayed resistances of 2.0–3.5 MΩ. Recordings were performed with an EPC-7 amplifier (HEKA Elektronik, Lambrecht, Germany) employing optimal series resistance compensation (Titz & Keller, 1997; Lips & Keller, 1998). Cells were selected for study if they had series resistances less than 15 MΩ. Whole-cell currents were recorded with sampling frequencies of 100 Hz to 5 kHz. Analysis of signals was performed off-line using Igor (Wavemetrics, Oregon, USA) plus Pulse and Pulse-Fit/v8.11 (HEKA) software.

Stimulus protocols

Unless otherwise stated, series of depolarizing voltage steps during voltage clamp from a holding potential of −60 mv (+10 mv depolarizing increments, 1 s⁻¹, to reach a maximum of +20 mv) were applied to activate plasma membrane Ca²⁺ channels and increase cytosolic Ca²⁺.

Intracellular Ca²⁺ measurements

Two different methodologies for detecting intracellular levels of cytoplasmic Ca²⁺ in the soma and dendrites were utilized. In the first, Ca²⁺ concentrations were measured by loading the cells from patch pipettes with fura-2 and measuring Ca²⁺-induced changes of fluorescence by charge coupled device (CCD) camera imaging. In the second procedure, cells were loaded from patch pipettes with Calcium Green-1 or fura-2, and Ca²⁺-induced changes in fluorescence were measured by MPM.

CCD camera imaging

As previously described (Ladewig & Keller, 2000) a modified version of a CCD camera system (TILL Photonics, Planegg, Germany) was employed in experiments carried out in the Göttingen laboratory. Calculations and analysis of intracellular Ca²⁺ concentrations and kinetics were subsequently performed off-line with the Igor analysis software.

Calibration constants for fura-2 were determined according to the method of Grynkiewicz et al. (1985). Calibration constants were adjusted after several days of experiments to account for small fluorescence changes in the microfluorometric system.

We defined selected regions of interest (ROI) to measure the Ca²⁺ signals in the cell compartments. The ROIs were always of the same size including a larger amount of pixels in the soma than in the dendrites (nearly the same number in the dendritic compartments). Background fluorescence was measured in corresponding out-of-focus ROIs, containing no obvious cellular structures. Values beyond a certain threshold calculated from the out-of-focus ROI were set to zero while the other pixel values were summarized, averaged and divided through the number of non-zero pixels.

Simultaneous measurement of mitochondrial and cytoplasmic [Ca²⁺]

In the present study, the excitation wavelength for rhod-2 was 510 nm. In order to monitor intramitochondrial calcium concentration ([Ca²⁺]_m), hypoglossal motoneurones were incubated with the Ca²⁺-sensitive fluorescent indicator rhod-2 AM (1–1.5 μm), for at least 1 h at room temperature. Rhod-2 has been used by several different laboratories to monitor changes in [Ca²⁺] within the mitochondrial matrix in different cell types (Babcock et al. 1997; Drummond & Tuft, 1999). Binding of Ca²⁺ to rhod-2 increases its fluorescence; however, since there is no significant shift in excitation or emission wavelengths upon Ca²⁺ binding, it cannot be used for ratio imaging. Rather, rhod-2 fluorescence is reported as 100 F/F₀, where F is the measured fluorescence, and F₀ is the fluorescence before application of the depolarizing stimulus.

Multiphoton excitation microscopy

Multiphoton imaging, originally described by Denk et al. (1990) was carried out in the Cornell laboratory with instrumentation and methodology previously described in detail (Kloppenburg et al. 2000). The multiphoton microscope consisted of a Spectra Physics Tsunami Ti:S laser with a 10 W Millenium (Spectra Physics, Mountain View, CA, USA), a retro-fitted Bio-Rad (Hercules, CA, USA) MRC-600 scanbox and a custom-made, fixed stage Olympus AX-70 upright microscope. A Hamamatsu (Bridgewater, NJ, USA) photomultiplier tube placed directly above the objective lens was used to collect the non-descanned emission (400–550 or 500–600 nm). The beam intensity was controlled using a ConOptics (Danbury, CT, USA) model 350–50 Pockels Cell, which also blanked the laser during fly-back (in between scan lines), eliminating unnecessary excitation of the preparation. Hypoglossal motoneurones loaded with Calcium Green-1 or fura-2 were imaged with 810 or 780 nm excitation respectively through a × 20/0.5 NA or a 40/0.8 NA water immersion objective lens. Ca²⁺ transients were acquired using scan lines at a rate of 2 or 4 ms line⁻¹. Voltage clamp data were simultaneously recorded with an Axopatch 200 amplifier (Axon Instruments, Foster City, CA, USA) on the second channel of the Bio-Rad scanner during the line scans to synchronize the start of depolarizing voltage steps with the Ca²⁺ signal.

Data analysis was performed with laboratory-written software. Pixel values were extracted from the line scan images along the time axis in the area of interest (averaged across the spatial axis). The simultaneously acquired membrane potential was analysed by the software to determine the starting point of the voltage step, ensuring synchronization between the Ca²⁺ data and the voltage step. The fluorescence signal after the voltage pulse was fitted to a single exponential decay model using the Igor analysis software.

Drug solutions

Caffeine, carbonyl cyanide-m-chlorophenylhydrazine (CCCP), cyclopiazonic acid (CPA), and procaine hydrochloride were purchased from Sigma-Aldrich Chemie (Deisenhofen, Germany) and dissolved in bath solution just before use. Drug solutions were bubbled with 95 % O₂, 5 % CO₂ at room temperature immediately before application to brain slices.

Statistics

Summarized values related to [Ca²⁺]_i are expressed as standard error of the mean (s.e.m.) unless otherwise stated. The statistical significance of differences between control and treatment groups was determined by Student's paired t test. Differences were taken to be significant for P < 0.05.

RESULTS

Voltage-dependent Ca²⁺ signals in hypoglossal motoneurones as revealed by CCD imaging and MPM

Highly dynamic Ca²⁺ signals in brain stem and spinal motoneurones are mediated by a specialized Ca²⁺ homeostasis, where low endogenous buffering capacities facilitate rapid relaxation rates of cytosolic Ca²⁺ transients and predict large domains of ‘unbuffered’ Ca²⁺ concentrations around open Ca²⁺ channel pores (Lips & Keller, 1998, 1999; Vanselow & Keller, 2000). In this report, we investigated local profiles of [Ca²⁺]_i in hypoglossal motoneurones that are particularly vulnerable both during human amyotrophic lateral sclerosis (ALS) and in corresponding mouse models of this neurodegenerative disease, by utilizing whole-cell patch clamp recordings in slice preparations, CCD imaging and MPM. Figure 1A displays the anatomical landmarks of slice preparations utilized in our present study, where the arrow indicates the location of the nucleus hypoglossus in the brain stem. Figure 1B shows a multiphoton image of a hypoglossal motoneurone during whole-cell patch clamp recording, where the cell was pipette-filled with 100 μm fura-2 as a fluorescent marker of the soma and dendrites. Dendritic structures could be clearly imaged at distances larger than 100 μm from the soma and a notable absence of spines could be recognized in experiments similar to that pictured in Fig. 1B. During whole-cell patch clamp recordings, somatic depolarizations with voltage steps starting from a holding potential of −60 mV evoked rapid elevations of [Ca²⁺]_i that were detected in the soma and dendrites with either CCD imaging or MPM (Fig. 2). The amplitude of voltage-evoked [Ca²⁺]_i increases varied as a function of membrane depolarization as previously demonstrated (Ladewig & Keller, 2000).

Figure 1. Fura-2-loaded hypoglossal motoneurone and its location in the medulla oblongata.

A, schematic drawing of a transverse section showing the slice preparation of the medulla oblongata including hypoglossal regions (arrow) and the important landmarks for orientation, such as the fourth ventricle (IV) and the dorsal nucleus of vagus (AP, area postrema; Amb, nucleus ambiguus; 12, nucleus hypoglossus; 12n, axons of 12; 10, nucleus vagus dorsalis). B, fura-2-loaded cell. The MPM image is a projection of 45 sections taken with the × 40 objective and a 1 μm step size. Excitation was at 780 nm.

Figure 2. Heterogeneous Ca²⁺ elevations in dendritic compartments of a hypoglossal motoneurone evoked by depolarization.

Multiphoton imaging (single optical section) showing voltage-evoked increases of intracellular calcium ([Ca²⁺]_i) in a hypoglossal motoneurone loaded from a whole-cell patch pipette with fura-2. A, fluorescence imaging of a hypoglossal motoneurone. Line scan positions in the different cell compartments are marked with white lines (soma, 1; proximal dendrite, 2; and distal dendritic region, 3). B, electrophysiological response of a voltage-clamped hypoglossal motoneurone after somatic depolarization from −60 to +20 mV (100 ms). C, line scans taken at the points indicated in A. The temporal resolution was 4 ms per line. Line scan images reveal spatial (vertical) and temporal (horizontal) properties of Ca²⁺-sensitive fluorescence in the selected compartments. The brighter sections in the boxes below indicate the duration of the voltage pulse (100 ms), fura-2 fluorescence decreases while Ca²⁺ is binding, coinciding with soma step depolarization to +20 mV from a holding potential of −60 mV. D, data extracted from the images in C along the time axis. Decreasing fura-2 values were divided by the basal fluorsecence level (F₀) and inversed. The dashed lines show the voltage pulse duration. The decay time constants were fitted to a single exponential function using the analysis software Igor. Note that Ca²⁺ transients are of dissimilar amplitude, indicative of heterogeneous Ca²⁺ elevations. E, graphical summary showing basal [Ca²⁺]_i and [Ca²⁺]_i signals evoked by depolarization to +20 mV at different distances from the soma of nine hypoglossal motoneurones detected by ratiometric measurements with the CCD system. Note that Ca²⁺ transients reveal a high variability in amplitude in the discrete compartments of the compared cells. F, mean decay time constants in the different cell compartments (soma, proximal dendrite at 10 μm, and dendritic compartments at 20/40, 60 and 80 μm from the centre of the soma) as revealed by multiphoton imaging. Inset, depolarizing pulses induce heterogeneous Ca²⁺ signals in somatic and dendritic compartments. Two compartments displaying the same amplitude are compared; Ca²⁺ elevations were evoked by a 100 ms voltage pulse. Ca²⁺ responses were notably different in the soma and the distal dendritic compartment (at 80 μm from the centre of the soma). The histogram shows a much faster decline to resting [Ca²⁺] in dendritic compartments compared with the soma.

Reliable Ca²⁺ responses in the soma and dendrites were detected when membrane potentials were depolarized to -50 mV and reached maximal levels at 0 to +20 mV. They exhibited different kinetics in the two cell compartments. A depolarizing voltage step with a step amplitude of +80 mV evoked transient, voltage-dependent increases in neuronal Ca²⁺ in the soma and dendritic compartments. CCD camera imaging revealed peak amplitudes with values of 362 ± 133 nm (+10 mV) in the soma and 312 ± 97 nm, 214 ± 59 nm (n = 9) for dendritic distances of 10 and 60 μm, respectively. Decay time constants of [Ca²⁺]_i were faster in dendrites (τ= 1075 ± 113 ms; n = 34) compared to somata (τ= 2631 ± 649 ms) although the buffer capacities (κ_B= 42 ± 14 in dendrites, n = 5 cells; see also Lips & Keller, 1998) and the basal Ca²⁺ levels (Fig. 2E) were similar. Voltage-evoked Ca²⁺ signals were locally defined and exhibited a heterogeneous pattern in dendritic compartments as revealed by MPM (Fig. 2). Under physiological conditions, depolarization-induced Ca²⁺ responses are an important determinant of Ca²⁺-dependent afterhyperpolarizations that follow each action potential and control motoneurone firing rates during burst discharges (Lape & Nistri, 2000). In hypoglossal motoneurones, depolarization-evoked AHPs are due to activation of Ca²⁺-dependent, SK-type K⁺ channels, and could therefore be blocked by apamin (10 μm, Fig. 3).

Figure 3. Caffeine-induced intracellular release of Ca²⁺ and outward current in a hypoglossal motoneurone.

A, Ca²⁺ signals measured by CCD camera imaging of a fura-2-loaded, voltage-clamped hypoglossal motoneurone. Fura-2 Ca²⁺ signals evoked in the soma first (left) by a step depolarization from −60 to +20 mV, and second (right) by bath application of 5 mm caffeine after blocking high voltage-activated Ca²⁺ channels with bath-applied 2 mm nickel (Ni²⁺). Response to caffeine consists of fast and slow components. Inset (upper right), outward current measured in the soma of the motoneurone during caffeine and Ni²⁺ application. The caffeine-induced outward current in 19 cells displayed a mean amplitude of 85.6 ± 21.8 pA. B, superimposed tail currents (occurring at the end of a voltage step protocol) which could partially be blocked by apamin, showing the presence of SK channels in neonatal mouse hypoglossal motoneurones.

The heterogeneity of Ca²⁺ signals could be explained by the different surface-to-volume ratio in the soma and dendrites or by the variability of Ca²⁺ channel expression (Ladewig & Keller, 2000). Another possible explaination for this effect is the contribution of ryanodine receptor (RyR)-dependent Ca²⁺ release to Ca⁺² signalling in motoneurones (Sah & McLaclan, 1991).

Store-operated Ca²⁺ release in hypoglossal motoneurones

To evaluate the contribution of store-operated Ca²⁺ release in hypoglossal motoneurones, we bath applied pharmacological agents which trigger or prevent release from intracellular storage sites (Kostyuk & Verhratsky, 1998). For the investigation of endoplasmic reticulum (ER)-dependent Ca²⁺ release, caffeine has been established as a specific activator of Ca²⁺ efflux from RyR-dependent stores (Kostyuk & Verhratsky, 1998). In a standard experiment, we transiently elevated [Ca²⁺]_i in voltage clamp mode with a 1 s depolarization to 0 mV with the intention of refilling empty Ca²⁺ stores. Subsequently, RyR-dependent release was probed by 5 mm caffeine in the presence of 2 mm Ni³⁺ in the bath solution to block the potential entry of extracellular Ca²⁺ through voltage-activated Ca²⁺ channels (McFarlane & Gilly, 1998).

Caffeine evoked substantial increases of [Ca²⁺]_i in the somatic compartment as illustrated in Fig. 3. The Ca²⁺ signal was biphasic, consisting of a relatively large and fast initial transient followed by a smaller and slower component (‘shoulder’). In contrast to depolarization-evoked Ca²⁺ responses, caffeine-induced Ca²⁺ transients were more restricted to the soma than to the proximal dendrites as revealed by CCD imaging (Fig. 4A). The initial fast responses were significantly larger in the soma (758 ± 263 nm, n = 13) compared to dendritic compartments at branching points (312 ± 98 nm, n = 13). Caffeine responses were absent after bath application of membrane-permeable compounds ryanodine (500 nm) or procaine (1 mm), supporting the notion that they resulted from RyR receptor-controlled Ca²⁺ stores.

Figure 4. Heterogeneous distribution of intracellular Ca²⁺ release sites in somatic and dendritic compartments of hypoglossal motoneurones.

A, CCD imaging series showing the propagation of the caffeine-evoked response in a hypoglossal motoneurone. Time course of the different Ca²⁺ responses in the soma (1) and in the proximal dendrite (2) evoked by bath application of 5 mm caffeine (black traces) and depolarization to +20 mV from a holding potential of −60 mV (grey traces) are shown. Boxes in the inset indicate the selected regions in the peripheral soma and proximal dendrite at a distance of 20 μm. The caffeine response is reduced to a minimum in the proximal dendrite compared to the soma. B, multiphoton imaging (single optical section) showing non-uniformly distributed Ca²⁺‘hot spots’ in the more distal dendrites evoked by bath application of 5 mm caffeine and depolarization (inset). White squares identify regions of interest (ROIs) before and after the branching point where transients related to caffeine (5 mm)-induced intracellular Ca²⁺ release were measured. C, occurrence of both signal types in different compartments reveal their heterogeneous distribution. The graphical summary of 23 hypoglossal motoneurones shows the frequency of significant Ca²⁺ elevations in the defined compartments. Caffeine-evoked responses were selected by evaluating the distance from the initial release site and their typical time course. The total number (N) of hot spots that showed Ca²⁺ increases after either depolarization or caffeine application was: soma, 23; primary dendrite, 22; 1st order branches, 23; secondary dendrite, 20; and 2nd order branches, 19. The frequency (f=n/N) of the above hot spots that responded to depolarization but not caffeine in these compartments was 0.25, 0.73, 0.30, 0.75 and 0.58 and of those reacting to both caffeine and depolarization was 0.75, 0.27, 0.70, 0.25 and 0.42, respectively. Caffeine responses without the related depolarization-induced responses could not be observed. The frequency of a Ca²⁺ influx through voltage-gated Ca²⁺ channels without any inducible caffeine responses was very high for two compartments (0.73 for primary and 0.75 for secondary dendrites). The soma and the primary branching point both showed a high probability for the coexistence of voltage-gated Ca²⁺ channels and caffeine-induced responses (f_depol-caff»f_depol for the soma branching points, about 40 and 70 μm from the soma, and in dendritic regions > 30 μm from the soma (*).

Although caffeine was usually administered into the bath at a concentration of 5 mm, it was possible to evoke graded Ca²⁺ responses by varying the caffeine perfusion time. Exposure times of 30–45 s evoked submaximal increases in [Ca²⁺]_i, on average 200 nm, whereas maximal increases (500–900 nm) occurred following exposure to caffeine for 60 s. Large caffeine-induced Ca²⁺ signals were accompanied by outward currents (85.6 ± 21.8 pA, n = 19) recorded at a holding potential of −60 mV. These outward currents coincided with Ca²⁺ signals and displayed a reversal potential close to the Nernst potential for K⁺, consistent with the idea that they were carried by Ca²⁺-dependent, apamin-sensitive K⁺ channels. Indeed, amplitudes and localization of Ca²⁺ signals were important determinants of outward current magnitudes, and Ca²⁺-dependent K⁺ currents were small or absent when Ca²⁺ increases were less than 200 nm (data not shown).

The high spatio-temporal resolution of MPM revealed that caffeine-induced release originated from defined sites in dendrites and somata. This is exemplified in Fig. 4B, where caffeine was bath applied at a concentration of 5 mm. This dose induced detectable Ca²⁺ release events in two locations labelled ‘1’ and ‘2’, but failed to evoke a response in a third location that was less than 15 μm away from ‘active’ release sites. This demonstrated that hypoglossal motoneurones: (i) displayed highly localized Ca²⁺ release events and (ii) were able to maintain substantial Ca²⁺ gradients over a distance of several micrometres for more than 20 s (see below), indicating a surprising functional separation of neighbouring dendritic compartments.

MPM was also used to investigate whether RyR-dependent Ca²⁺ release was located close to sites where voltage-dependent Ca²⁺ influx occurred. As shown in Fig. 4C, sites of caffeine-induced release could indeed be found in close vicinity of depolarization-activated Ca²⁺ influx pathways but only in distinct places like the somatic region and branching points. Such a co-localization could be observed in the soma and at branching points in 70 % of all cells investigated. On the other hand, a series of experiments (n = 5) with procaine (1 mm) showed no significant effect on kinetics of depolarization-evoked Ca²⁺ transients in the soma, dendrites or branching points of hypoglossal motoneurones (data not shown).

Involvement of ER and mitochondria in cytoplasmic Ca²⁺ regulation

To investigate the relative impact of endoplasmic reticulum and mitochondria, we tested pharmacological agents that selectively block Ca²⁺ uptake into different compartments. For example, CCCP was used as an uncoupler of the electrochemical potential across the inner mitochondrial membrane to abolish mitochondria-mediated regulation of [Ca²⁺]_i (Hirsch et al. 1989). CPA was utilized as a sarcoplasmic/endoplasmic reticulum (SERCA) pump inhibitor.

As shown in Fig. 5A, Ca²⁺ stores were initially preloaded by voltage activation of plasma membrane Ca²⁺ channels. Subsequent bath application of 2 μm CCCP evoked a substantial, slow increase of cytoplasmic free [Ca²⁺]_i, from 78 ± 28 to 256 ± 54 nm (n = 5). This increase could not be explained by a reduced level of ATP, as 4 mm ATP was continuously provided by pipette perfusion of the cytosol. As expected from the dominant role of K_Ca channels in hypoglossal neurones, CCCP-evoked [Ca²⁺]_i increases were paralleled by outward currents with mean amplitudes of 39.8 ± 14.7 pA, n = 5 cells (Fig. 5B). Figure 5C shows a recording from a cell, where after a short depolarization to +20 mV to reload stores, initially CCCP and subsequently, within 3 min, caffeine were applied to the bath. A second application of CCCP without caffeine, within 3 min did not evoke a further basal increase in [Ca²⁺]_i, indicating that the underlying mechanisms were still inactivated. Very important, however, was the finding that CCCP fully eliminated the fast component of the caffeine-evoked response (5/5 cells, Fig. 5C). In this case, the caffeine response was depressed to levels under 200 nm, and no outward current could be observed after caffeine application. Taken together, these results indicate a significant role of mitochondria for both caffeine-induced Ca²⁺ responses and the control of K_Ca-dependent motoneurone excitability.

Figure 5. Depression of caffeine-induced intracellular Ca²⁺ signals after depletion of mitochondrial Ca²⁺.

A, fura-2 CCD camera imaging of Ca²⁺ transients evoked in the soma of a hypoglossal motoneurone by a single step depolarization to +20 mV from a holding potential of −60 mV (1 s, *); by mitochondrial Ca²⁺ release evoked by 2 μm CCCP and finally by a second voltage pulse. B, CCCP evoked a small outward current (indicated by dashed lines in A), which displayed a mean amplitude of 39.8 ± 14.7 pA (n = 5). C, ER Ca²⁺ release triggered by 5 mm caffeine, 4 min after CCCP application and washout. The initial rapid [Ca²⁺]_i transient typically evoked by caffeine is absent, while the slow transient is depressed (Ca²⁺ entry through voltage-gated Ca²⁺ channels was blocked by 2 mm Ni²⁺). D, graphical summary of results from 4 motoneurones showing the decrease of caffeine responses after CCCP application.

The role of mitochondria and ER for cytosolic Ca²⁺ sequestration is further illustrated in Fig. 6, where CCCP as well as CPA prolonged the decay of voltage-evoked Ca²⁺ signals in somatic and dendritic compartments. CCCP increased τ, the relaxation time constant of [Ca²⁺]_i, from 3.2 ± 0.6 s to 6.3 ± 0.6 ms (P < 0.05 n = 6) in somatic and from 2.1 ± 0.4 to 6.3 ± 0.6 s in dendritic compartments (P < 0.05; n = 6). Again, this prolongation could not be explained by ATP depletion, as it was continuously supplied via the pipette. Also, the retardation of relaxation times was not explained by a potential ‘rundown’ of extrusion or uptake mechanisms, as previous studies have demonstrated a robust stability of Ca²⁺ recovery kinetics for up to 1 h during whole-cell recordings from hypoglossal motoneurones (Lips & Keller, 1998, 1999). As shown in the inset of Fig. 6A, the CCCP-mediated prolongation of Ca²⁺ transients was paralleled by a retardation of K_Ca-dependent tail currents measured after a standard depolarization protocol (500 ms, +20 mV). Interestingly, the effect of CCCP on decay times (approximately 2-fold prolongation) was significantly greater than that of CPA (increase of τ by 0.8 ± 0.1 s over control, P < 0.05, 1.2-fold prolongation; n = 4). Both drugs had only small effects on the amplitude of voltage-evoked Ca²⁺ signals (< 15 %).

Figure 6. Relative effectiveness of mitochondrial and ER uptake mechanisms in shaping Ca²⁺ responses to plasma membrane depolarization.

A, fura-2 CCD camera imaging of Ca²⁺ transients evoked in the soma by step depolarization to +20 mV from a holding potential of −60 mV, before (Control) and after bath application of 2 μm CCCP. Inset, prolongation of the tail current (depolarization to +20 mV) of the Ca²⁺-activated K⁺ conductance after CCCP application. B, graphical summary of results from 6 motoneurones showing effects of 2 μm CCCP on the decay time constants of Ca²⁺ signals (τ, ms) in different cell compartments (decay time constants were prolonged from 3.2 ± 0.6 to 6.2 ± 1.2 s for the soma; 2.7 ± 0.4 to 5.8 ± 0.8 s for the proximal dendrite; 1.9 ± 0.4 to 3.9 ± 0.7 s for distance from centre of soma (d) = 20 μm; and 1.7 ± 0.04 to 3.5 ± 0.1 s for d = 30–40 μm. Effective prolongation of decay time constants after CCCP: 2.1 ± 0.25 for somatic and dendritic sites). C, fura-2 CCD camera imaging of Ca²⁺ transients evoked in the soma by step depolarization to +20 mV from a holding potential of −60 mV, before (Control) and after blocking sarcoplasmic-endoplasmic reticulum (SERCA) pumps with bath application of cyclopiazonic acid (CPA), 50 μm. D, graphical summary of results from 4 motoneurones showing increased time constant of decay (P < 0.05) in different cell compartments (from 3 ± 0.2 to 3.9 ± 0.3 s in soma and proximal dendrite, 1.9 ± 0.24 to 2.6 ± 0.29 s for d = 20–30 μm, effective prolongation of decay time constants: 1.3 ± 0.2 for somatic and dendritic sites). E, kinetics of cytosolic and mitochondrial Ca²⁺ influx as revealed by the fluorescence of fura-2 and rhod-2. Hypoglossal motoneurones were loaded with 1–1.5 μm rhod-2 AM for 1 h to enable monitoring of [Ca²⁺]_m and fura-2 was included in the patch pipette to allow simultaneous monitoring of [Ca²⁺]_i. Representative recording showing overlay of [Ca²⁺]_i and [Ca²⁺]_m transients during a 2 s depolarizing pulse (from −60 to +30 mV) re-plotted on an expanded time scale ([Ca²⁺]_i indicated by fura-2 label and F/F₀ of calcium-dependent intramitochondrial fluorescence by rhod-2 label). F, mitochondrial depolarization (perfusion with 2 μm CCCP) partially (by 50–70 %) blocked Ca²⁺ accumulation by the mitochondria after depolarization (n = 2).

A further observation supporting the notion that mitochondrial Ca²⁺ uptake is an important regulator of cytosolic Ca²⁺ homeostasis results from experiments using the mitochondrial indicator dye rhod-2. After loading motoneurones with rhod-2 ester, its positive charge results in significant dye accumulation within the mitochondrial matrix. Following hydrolysis of the acetoxymethylester group, rhod-2 becomes trapped in the mitochondrial matrix. To wash out residual, cytosolic rhod-2 cells were then filled with fura-2 via the patch pipette. During a depolarizing step from −60 to +30 mV (duration of 2 s), [Ca²⁺]_i increased to 814 ± 87 nm (n = 5; Fig. 6E). The total time to peak for the increase in [Ca²⁺]_i was 2.0 ± 0.2 s and the decay (τ) for recovery following the end of the depolarizing pulse was 3.9 ± 0.8 s. The same stimulus also increased mitochondrial [Ca²⁺] as indicated by the increase in rhod-2 fluorescence. When the cytoplasmic and mitochondrial [Ca²⁺] records from the same cell were viewed on the same time scale, no significant difference in the rising phases of [Ca²⁺]_i or intramitochondrial calcium dynamics could be detected. However, the recovery time constant τ of rhod-2 fluorescence was 7.1 ± 1.7 s which was significantly longer compared to the recovery of [Ca²⁺]_i (P < 0.05).

Implications for quantitative models of Ca²⁺ homeostasis in hypoglossal motoneurones

For a more detailed understanding of cellular function, it is useful to develop quantitative models of Ca²⁺ profiles and their spatio-temporal dispersion in the cytosol. Earlier studies established a linear, quantitative model of Ca²⁺ homeostasis in hypoglossal motoneurones that, among others, identified a low endogenous buffering capacity, κ_S= 41, as a critical determinant of [Ca²⁺]_i (Lips & Keller, 1998). Under physiological conditions, low κ_S presumably reflects low levels of mobile, Ca²⁺-binding proteins (CaBP), like calbindin or parvalbumin, and is thought to facilitate rapid relaxation times of global Ca²⁺ transients during high-frequency, rhythmic activity (Lips & Keller, 1999). As diffusible CaBPs also accelerate the dispersion of spatial gradients, low κ_S tends to stabilize localized Ca²⁺ signal domains in cytosolic compartments. This interpretation was underlined by observations using high-resolution MPM. As exemplified in Fig. 4, application of caffeine evoked detectable Ca²⁺ responses in two neighbouring locations that lasted many seconds, but failed to induce a response in a third location that was less than 15 μm away. Such results indeed illustrate a remarkable separation of Ca²⁺ profiles in neighbouring compartments.

To analyse the impact of ‘buffered diffusion’ in motoneurones in a more quantitative way, it is useful to consider a patch-clamped cell where endogenous buffers are either fixed or mobile. In this case, the ‘effective’ diffusion constant for Ca²⁺ can be estimated from elementary parameters of the underlying buffers by using the equation (Zhou & Neher, 1993; Klingauf & Neher, 1997):

where D_Ca= 220 μm² s⁻¹ and D_fura-2= 170 μm² s⁻¹ represent values from the literature for diffusion constants of Ca²⁺ and fura-2; and κ_B′, κ_mob and κ_S denote Ca²⁺-binding capacities of fura-2, mobile and fixed buffers, respectively (Zhou & Neher, 1993; Klingauf & Neher, 1997). A direct implication of this relation is that highly mobile buffers such as fura-2 account for a large product D_fura-2κ_B′′ and thus notably accelerate the diffusion of Ca²⁺. Accordingly, caffeine-induced transients are expected to be substantially more localized under undisturbed conditions compared to the whole-cell recording situation employed in this report. Moreover, earlier studies established that mobile endogenous buffers were either absent or reduced to minimal levels in low buffered motoneurones (Lips & Keller, 1999; Palecek et al. 1999), suggesting that κ_mob«κ_B′. For the experimental conditions in patch clamped hypoglossal motoneurones (κ_S= 41, [fura-2]= 100 μm), we therefore calculate an expected value D_eff= 158 μm² s⁻¹ based on the motoneurone-specific profile of buffers.

This value can be directly compared to actual experimental results from CCD imaging and MPM. For example, the effective diffusion constant can be measured by observing the distance r that Ca²⁺ diffuses in a time interval t by using the equation (Neher, 1995):

To illustrate this relation, we consider experimental conditions similar to those presented in Fig. 4, where a notable gradient occurs over a distance r = 15 μm for t > 20 s. Although several distinct molecular mechanisms might account for this spatial heterogeneity, it is still useful to consider the elementary diffusion model. By using the above equation, we obtain D_eff < 5.4 μm² s⁻¹, which is a factor of 30 smaller compared to the diffusion constant of 158 μm² s⁻¹ expected from theoretical considerations. Such results have two interesting implications. First, low levels of endogenous, mobile buffers in hypoglossal motoneurones presumably retard the dispersion of localized Ca²⁺ gradients around open membrane channels and release sites. Second, this effect is apparently enhanced by an additional, functional separation of dendritic compartments that further prevents the rapid dispersion of Ca²⁺ gradients in the cytosol.

In this context, it is interesting to note that multiphoton images of hypoglossal motoneurones such as those exemplified in Fig. 1 demonstrated a remarkable absence of dendritic spines, which are thought to restrict the dispersion of Ca²⁺ gradients in other cells like hippocampal neurones or cerebellar Purkinje cells. As these neurones are known to display substantially higher Ca²⁺ buffering capacities with notably increased concentrations of diffusable CaBP in the cytosol, it is reasonable to interpret low buffering capacities and functional separations of dendritic compartments in motoneurones as two complementary, cellular adaptations that facilitate localized Ca²⁺ signal domains in the absence of specific anatomical formations (e.g. dendritic spines).

DISCUSSION

Ca²⁺ signalling in motoneurones is a critical determinant of synaptic and electrical properties and has been associated with many aspects of neuronal plasticity (Rekling et al. 2000). Failure of precise regulation of free intracellular Ca²⁺ levels may lead to severe pathophysiological conditions and/or cell death (Krieger et al. 1996). In the present study, we investigated the local profile of Ca²⁺ signals in hypoglossal motoneurones of newborn mice by employing MPM, CCD imaging and whole-cell patch clamp recording in brain stem slices (Ladewig & Keller, 2000). We show the existence of dynamic intracellular Ca²⁺ storage and release mechanisms, and demonstrate physiological interaction between RyR-dependent Ca²⁺ release, mitochondrial processes and K_Ca-dependent control of electrical activity. The high spatio-temporal resolution of MPM offered the unique advantage to analyse highly localized Ca²⁺ signals in somatic and dendritic compartments.

Functional interaction of Ca²⁺ storage and release sites

During bath application, caffeine evoked localized RyR-mediated increases in [Ca²⁺]_i which varied substantially between different locations within a given cell. As our indicator dye (50–100 μm fura-2 in standard experiments) is thought to substantially accelerate local dispersion of [Ca²⁺]_i, Ca²⁺ profiles are expected to be even more localized under physiological, undisturbed conditions. A close interaction between Ca²⁺ release and mitochondrial Ca²⁺ homeostasis was suggested by our observation that caffeine-induced responses were significantly suppressed after disruption of the mitochondrial electrochemical potential by CCCP. As mitochondria are not thought to contain RyR-channels (Szalai et al. 2000), the most likely interpretation is that RyR-induced Ca²⁺ release is modulated by mitochondrial processes. It is interesting to note that studies on the interaction of RyR and mitochondrial homeostasis performed on smooth muscle cells (Nassar & Simpson, 2000), cardiac cells (Szalai et al. 2000) and HeLa cells (Rizzuto et al. 2000) displayed results that were similar to those observed in hypoglossal neurones. This indicates that a tight coupling between Ca²⁺ regulation in mitochondria and RyR-sensitive Ca²⁺ storage sites represents a more general molecular scheme. Our interpretation is clearly different from that developed for dissociated basal forebrain neurones of rats (Murchison & Griffith, 2000), where a similar, biphasic caffeine-induced signal was transformed by co-application of CCCP into a much larger monophasic Ca²⁺ signal lacking a late, slower ‘shoulder’. In this case, the rapid initial decline of the caffeine-induced signal in forebrain neurones was rather associated with avid mitochondrial uptake of cytoplasmic Ca²⁺.

In addition to RyR-mediated Ca²⁺ responses, caffeine evoked outward currents that were strongly correlated with caffeine-evoked Ca²⁺ signals. This indicated that they were primarily mediated by apamin-sensitive, Ca²⁺-dependent K⁺ channels known to be present in hypoglossal motoneurones of that age (Lape & Nistri, 2000). Similarly, in hippocampal pyramidal neurones, caffeine activates K_Ca currents that hyperpolarize membrane potentials (Uneyama et al. 1993) and induces spontaneous hyperpolarizations in dopaminergic neurones of neonatal rats (Seutin et al. 2000). In a corresponding scheme, Ca²⁺-induced membrane hyperpolarization might reduce build-up of [Ca²⁺]_i in hypoglossal motoneurones during hypoxia or ischaemia (Zhang & Lipton, 1999; Semenov et al. 2000), and thus avoid excess Ca²⁺ influx through voltage-activated Ca²⁺ channels during persistent depolarization.

Functional significance of localized Ca²⁺ profiles during physiological motoneurone activity

Our findings clearly demonstrate the presence of efficient Ca²⁺ stores in the ER of hypoglossal motoneurones, which can be readily activated by established pharmacological tools. Together with the strong impact on the decay of voltage-activated Ca²⁺ signals, these observations indicate that Ca²⁺-storing organelles may play a significant role in physiological Ca²⁺ regulation, and as the mitochondrial Ca²⁺ clearance is highly temperature sensitive the effect will be much higher at body temperature (David & Barrett, 2000). It is interesting to note that in low buffered cells the presence of dynamic, Ca²⁺-sequestering organelles could have several advantages. First, this arrangement offers a mechanism for offsetting the potentially dangerous consequences of low endogenous Ca²⁺ buffering resulting from small concentrations of Ca²⁺-binding proteins in the cytosol. A second advantage might be related to the fact that hypoglossal motoneurones exhibit highly variable and rapidly changing discharge patterns which control the genioglossus muscles during mastication, deglutition, phonation, vomiting and ventilation. In the latter case, precise postural control of the tongue must be quickly adjusted to optimize upper airway resistance and adjust ventilation across a wide range of breathing frequencies. In rodents, for example, frequencies may be as high as 5 breaths s⁻¹ (Jaquin et al. 1996) and rapid adjustments in [Ca²⁺]_i are necessary to fine-tune activity-dependent changes in muscle force and corresponding tongue position.

Potential implication of localized Ca²⁺ profiles for pathophysiological conditions and ALS-related motoneurone disease

Previous investigations indicated that motoneurones which are particularly vulnerable during ALS are also characterized by exceptionally low endogenous buffering capacities and relatively large microdomains around open Ca²⁺ channels (Lips & Keller, 1999). By using MPM and CCD imaging, we now present strong evidence that this cell-specific specialization of Ca²⁺ homeostasis is accompanied by efficient control of local [Ca²⁺]_i by Ca²⁺-regulating organelles. Although major changes in morphological and electrophysiological properties have been documented during the first 2 postnatal weeks in hypoglossal motoneurones (Viana et al. 1994), our findings have several functional implications for motoneurones during human ALS and related animal models of this neurodegenerative disease. For example, localized Ca²⁺ regulation becomes particularly vulnerable to disruptions of mitochondrial function, a phenomenon that is a well-known symptom of motoneurone degeneration during ALS. In such pathophysiological situations, bursts of action potentials might increase local Ca²⁺ amplitudes to high concentrations above the micromolar domain, due to the combination of massive Ca²⁺ influx and low buffering (Lips & Keller, 1999; Herrington et al. 1996). Without efficient Ca²⁺ sequestration machinery, the local Ca²⁺ concentration might easily reach levels that are sufficient to activate mitochondria-related apoptotic mechanisms and associated signal cascades (Jacotot et al. 1999). Taken together, our measurements therefore indicate that cellular specializations, which facilitate highly dynamic and localized Ca²⁺ profiles in motoneurones under physiological conditions, might also enhance their vulnerability during ALS-related motoneurone disease and associated cellular disturbances.