Calcium dynamics in leech neurons were studied using a fast CCD camera. Fluorescence changes (ΔF/F) of the membrane impermeable calcium indicator Oregon Green were measured. The dye was pressure injected into the soma of neurons under investigation. ΔF/F caused by a single action potential (AP) in mechanosensory neurons had approximately the same amplitude and time course in the soma and in distal processes. By contrast, in other neurons such as the Anterior Pagoda neuron, the Annulus Erector motoneuron, the L motoneuron, and other motoneurons, APs evoked by passing depolarizing current in the soma produced much larger fluorescence changes in distal processes than in the soma. When APs were evoked by stimulating one distal axon through the root, ΔF/F was large in all distal processes but very small in the soma. Our results show a clear compartmentalization of calcium dynamics in most leech neurons in which the soma does not give propagating action potentials. In such cells, the soma, while not excitable, can affect information processing by modulating the sites of origin and conduction of AP propagation in distal excitable processes.
The nervous system of the leech Hirudo medicinalis is composed of a chain of 21 segmental ganglia, each consisting of ∼400 neurons (Muller et al. 1981; Nicholls and Baylor 1968). Two pairs of roots (anterior and posterior) that include branches of mechanosensory neurons and motoneurons emerge from each ganglion and innervate the corresponding body segment. Mechanosensory neurons have all been identified and electrophysiologically characterized (Muller et al. 1981). Each ganglion contains seven pairs of mechanosensory neurons, three of which are dedicated to the transduction of light mechanical stimulation (T cells), two signal a stronger mechanical stimulation (P cells), and two neurons are involved in the detection of noxious mechanical stimuli (N cells).
Most motoneurons and many interneurons have been identified and their role in mediating motor responses and behavior has been extensively studied (Kristan 1982; Lockery and Kristan 1990a,b; Muller et al. 1981; Stuart 1970). The great majority of leech neurons, like most neurons of invertebrates, do not have a conventional excitable axon and a distinct poorly excitable dendritic tree (Wessel et al. 1999a,b) but have a large single-process emerging from the soma where action potentials (APs) are initiated. When an electrode is inserted into the soma of leech neurons, APs with an amplitude >60 mV (Nicholls and Baylor 1968) are measured only in mechanosensory neurons and in very few other neurons (neuron 50, Retzius cells, Leydig cells, and S interneurons). In most leech neurons, the amplitude of APs recorded in the soma is <10 mV (Stuart 1970). By using laser axotomy and electrophysiological localization of APs, it has been shown that in some leech neurons (the Anterior Pagoda and Annulus Erector motoneurons), the initiation site for APs is located at the first major bifurcation (Gu et al. 1991; Melinek and Muller 1996), several hundred micrometers away from the soma.
These observations raise several questions: if most leech neurons appear electrically compartmentalized, will it be the same for intracellular messengers such as calcium? What are the functional implications of these biophysical properties?
To obtain experimental information about these issues, calcium dynamics during the generation of APs was studied in leech neurons (Ross et al. 1987). The membrane impermeant calcium indicator Oregon Green was pressure injected into individual leech neurons. After dye incubation, stained neurons were impaled with intracellular microelectrodes. Changes of the emitted fluorescence (ΔF) were monitored with a fast CCD camera (RedShirtImaging), and the spatial profile of optical signals (ΔF/F) caused by intracellular calcium dynamics was analyzed in single trials. In the Anterior Pagoda neuron, the Annulus Erector, L motoneurons, and other motoneurons, these optical signals originated and were the largest at the first major bifurcation. Our results show that in leech neurons other than mechanosensory neurons, calcium transients propagate throughout the arborization, but do not reach the soma.
Preparation and electrical recordings
Isolated ganglia of the leech (H. medicinalis) was used. The ganglion was kept in a silicone-elastomer (Sylgard)-coated dish at room temperature (20–24°C) and was bathed in a Ringer solution containing (in mM) 115 NaCl, 1.8 CaCl2, 4 KCl, and 12 glucose and buffered to pH 7.4 with 10 mM Tris maleate and NaOH (Muller et al. 1981). In some experiments, a piece of the skin was kept connected to the ganglion roots (Nicholls and Baylor 1968; Zoccolan et al. 2001; Zoccolan and Torre 2002⇓) and was flattened and fixed with pins to the bottom of the recording chamber. Ganglia were pinned in Sylgard-coated petri dishes. The connective tissue sheath over the neuronal soma was removed with fine tweezers to facilitate intracellular penetration and optical recording. Intracellular recordings were obtained with conventional sharp electrodes obtained from glass capillaries (Kwik-Fil 1B100F-4, WPI) pulled with a puller (Flaming/Brown Micropipette Puller Model P-97, Sutter Instruments). Electrodes had an input resistance between 15 and 50 MΩ. Two or three suction pipettes were used to obtain extracellular recordings from the anterior and the posterior roots and the connective entering into the leech ganglion (Arisi et al. 2001; Pinato and Torre 2000; Stent et al. 1978). Suction pipettes were used also to stimulate the roots and to analyze calcium dynamics initiated by the stimulation of distal processes (see Fig. 8). Roots were stimulated by applying a current pulse of 0.2 s with an intensity able to evoke a brief discharge of at least three APs.
In some experiments, a mechanical stimulus was delivered to the skin by rapidly pressing a nylon filament driven by a solenoid (Pinato and Torre 2000).
Sharp electrodes filled with 4 M potassium acetate were used to impale neurons and APs were evoked by passing depolarizing current pulses through the electrodes. APs were also evoked by applying an extracellular voltage pulse to one suction electrode in which one root was inserted. In this case, APs originating in the most distal branches propagated toward the soma. Intracellular currents were passed through a bridge circuit. APs were recorded as previously described (Pinato and Torre 2000) with an Axopatch 200 amplifier and analyzed with Clampex 8.0 (Axon Instruments).
Dye preparation and loading
3 mM Oregon Green dye (Oregon Green-1 488 BAPTA-1, hexapotassium salt, cell impermeant, Molecular Probes O-6806) was dissolved in 30 mM potassium acetate. The dye was pressure injected with intracellular microelectrodes (Kwik-Fil TW 100-4, WPI) into leech neurons impaled under visual control. A continuous pressure of 5–20 psi was applied to the microelectrode using a pressure injector system (PM2000B 4-Channel MicroData Instrument). Dye was injected for ∼5 min, depending on the cell, and the staining was controlled by visual inspection. After loading the neuron with the dye, the ganglion was left at room temperature in the dark for 2–3 h. In some control experiments, Calcium Green (Calcium Green-1, hexapotassium salt, cell impermeant, Molecular Probes C-3013) was used instead of Oregon Green and very similar results were obtained, as shown in Fig. 3. Calcium Green and Oregon Green have a Kd for Ca2+ in the absence of Mg2+ of 190 and 170 nM, but Oregon Green is more efficiently excited by a 488-nm spectral line than Calcium Green. As a consequence, with a similar dye loading, a larger emitted fluorescence is obtained by using Oregon Green.
The preparation was illuminated with an halogen lamp (Olympus) or a Xenon lamp (Optiquip). The Xenon lamp was ∼10 times brighter than the halogen lamp and was used when it was necessary to analyze calcium dynamics in fine dendrites. The leech ganglion was viewed with an upright microscope. The preparation was illuminated through the objective (×40/0.80 NA, water immersion, LUMPlanFI) and a dichroic mirror was used (U-MSWB filters: exciter filter BP 420–480 nm, barrier BA: >515 nm and dichroic mirror: 500 nm). Fluorescence emitted by the dye was measured with the fast CCD camera Neuro CCD-SM (RedShirtImaging). Images were acquired at a sampling rate of 1 kHz, and the sequence of fluorescence images F(n) was analyzed using NeuroPlex software. The spatial resolution of the fast CCD camera was 80 × 80. Fluorescence images F(n) obtained by the fast CCD camera, were averaged from F(n1) to F(n2), where F(n1) corresponded to the image acquired just after shutter opening and F(n2) corresponded to the image acquired just before stimulation. The averaged image F was used as the resting fluorescence light intensity (see Fig. 1A). Fractional fluorescence changes ΔF(n)/F were computed from the image sequence F(n) using the following procedure.
The time constant τ of bleaching was computed by fitting with the equation F0 e-t/t the fluorescence observed between shutter aperture and electrical stimulation, where F0 is the fluorescence intensity measured immediately after the shutter opens. Typical values of τ varied between 1 and 2 s. Corrected fluorescence F*(n) images were obtained as F*(n) = F(n) e(n−m)Δt/τ where m is the number of the image corresponding to the shutter opening and Δt is the interval between two successive images (Δt is the inverse of the sampling frequency of image acquisition). ΔF(n)/F = [F*(n) − F]/F was then computed and to avoid the noise originating from dividing by small numbers, ΔF/F was multiplied by a factor equal to F2/(F2 + Th02). A value of Th0 corresponding to a fluorescence value of 5 was used for Th0. To reduce the noise further, an adaptive spatial filtering was also used. The value of ΔF/F was averaged over the entire image sequence and ΔF/F was obtained (see Fig. 1B). Pixels with a high value of ΔF/F are those where the optical signal is significant and relevant. The value of ΔF/F was spatially filtered in the following way: if ΔF/F was less than Th1, ΔF/F was spatially averaged over a mask of 5 × 5 pixels, if Th1 < ΔF/F < Th2, the mask was 3 × 3 and if ΔF/F was greater than Th2, no spatial filtering was used. Figure 1, C and D, compares an unfiltered and filtered image of ΔF/F, when the values of 1% and 3% were used for Th1 and Th2, respectively. The resulting image sequence ΔF/F was viewed with a color-coded map, where deep red was the highest fluorescence change.
Fluorescence changes (ΔF/F) originating from the calcium indicator, were measured in single trials. Because ΔF/F is related to changes of intracellular Ca2+ (Jaffe et al. 1992; Koester and Sakmann 2000; Stuart et al. 1997; Waters et al. 2005), spatial and temporal properties of calcium dynamics were analyzed at a fast temporal resolution (1 kHz) and at several simultaneous locations. Leech neurons were impaled with conventional fine electrodes, filled with 4 M potassium acetate, and APs were evoked by passing a depolarizing current into the soma, by stimulating the roots, or by touching the skin. This was done by applying a brief voltage pulse to the roots with a suction pipette. In the first case, APs traveled toward distal dendrites, in the last two cases APs traveled from distal dendrites toward the soma.
Mechanosensory neurons, T or P or N cells, innervate the skin with axons hundreds of micrometers long that pass out from the ganglion. Mechanosensory neurons also send numerous branches into the neuropile where they make synaptic contacts with interneurons and motoneurons (Gu et al. 1991). Their axons are able to produce full-size APs. When an AP was elicited in the soma of all mechanosensory neurons, a clear optical signal (ΔF/F) was observed, both in the soma and in distal processes. As shown in Fig. 2B, when three APs were elicited in a T cell (displayed in Fig. 2A), the optical signal ΔF/F was ∼6% both in the soma and in the two axons innervating the skin. ΔF/F increased in steps, each step associated with the occurrence of an individual AP. The rise of the optical signal followed the peak of the AP by ∼1 or 2 ms (see Fig. 2B). After reaching a peak, ΔF/F declined to zero with a time constant of about 1.6 s in the soma and 1.4 and 1 s in the axons. Figure 2C reproduces the spatial profile of ΔF/F in a color-coded scale at four selected times (indicated by t1, t2, t3, t4 in B) after current injection: the optical signal increases almost uniformly and simultaneously in the visible portion of the distal processes.
A more quantitative analysis of ΔF/F elicited by a single AP in mechanosensory neurons is shown in Fig. 3. In the P cell illustrated in Fig. 3A, optical signals ΔF/F of ∼3.5% were observed at the locations indicated by the same color and lettering in Fig. 3, A and B. ΔF/F followed the peak of the AP within 1 or 2 ms (see also Ross et al. 1987). The time constant τ of the decay of ΔF/F (see Fig. 3B) varied between 1 and 2 s similarly to results previously obtained in Retzius cells (Beck et al. 2001). Figure 3C reproduces the spatial profile of ΔF/F in a color-coded scale at four selected times (indicated by t1, t2, t3, t4 in B) after evoking a single AP in the P cell: the optical signal increased almost simultaneously in the soma and at some specific locations along the axon. A larger increase was observed in a sub-compartment of the soma (see red spots in images taken at t1 and t2 of C). When the optical signal in the soma was averaged over the circle indicated in A, ΔF/F had an amplitude and time course very similar (see B) to that observed at the two locations indicated in green, aa, and blue, ab, along the axon (see A).
In some control experiments, the calcium indicator Calcium Green was used instead of Oregon Green and very similar results were obtained, as shown in Fig. 3D. In this experiment, a P cell was loaded with Calcium Green, and a single AP was evoked (see D, top); Optical signals ΔF/F were measured on the cell body (red trace and s in D) and along the axon at ∼40 μm (green trace and aa in D) and at ∼150 μm (blue trace and ab in D) from the cell body. As Oregon Green emits fluorescence more efficiently than Calcium Green and the same light intensity was used in the two experiments, a similar emitted fluorescence can be obtained only by a larger loading of the dye Calcium Green. As shown in Fig. 3, amplitude and time course of optical signals ΔF/F obtained with Oregon Green or Calcium Green were similar, suggesting that dye loading had a minor effect on calcium dynamics inside leech neurons. Collected data from four P cells (red symbols), one T cell (black symbols), and one N cell (blue symbols) are shown in Fig. 3E: the peak of the optical signal ΔF/F evoked by a single AP had almost the same amplitude in the soma and along the axons at distances of 250 μm from the soma.
APs in mechanosensory neurons can be initiated either by direct mechanical stimulation of the skin (Nicholls and Baylor 1968) or by extracellular stimulation of the roots (Wittenberg et al. 1990). The amplitude and time course of ΔF/F when the AP was initiated either in the soma or in the distal processes were very similar and almost indistinguishable.
The results in Figs. 2 and 3 show that calcium transients, elicited by APs in mechanosensory neurons, have similar properties in the soma and in the axons. A specific initiation site for APs was not identified possibly because of the limited time resolution of our optical system (1 kHz) or the calcium indicator dye. It is concluded that voltage-gated Ca2+ channels are not differentially clustered along processes of mechanosensory neurons; and voltage signals and Ca2+ transients are observed throughout the entire neuron.
Comparison of averaged and single trial responses
The data shown in Figs. 2 and 3 were obtained from single trials and were not averaged. To analyze reproducibility of Calcium transients, mechanosensory neurons were loaded with Oregon Green and a single AP was evoked in successive trials repeated every 2 min.
Figure 4 illustrates a stained T cell in which a single AP was evoked in successive 15 trials. Optical signals from the cell body (shown in red and with the letter s) and the two axons emerging from it (shown in blue, a1, and green, a2) were recorded and their variability analyzed. The amplitude of the peak of ΔF/F was rather stable over most trials as shown in A–C, bottom. In two trials (4 and 5), large optical signals were observed, but in the remaining 13 trials, the amplitude of ΔF/F was remarkably reproducible. The SD of the peak of ΔF/F was 0.95, 0.48, and 0.53 in the soma and in the anterior and posterior axon. The average optical signal <ΔF/F> and its coefficient of variation CVΔF/F obtained from the three colored regions of the neuron were computed. As shown in A–C, top, the amplitude and time course of the averaged ΔF/F in the three regions were quite similar. The CVΔF/F in the axons decreased to ∼0.4 at the peak of the optical signal and was 0.55 in the cell body. Very similar results were obtained in two other T cells, in three P cells, and in two N cells. A coefficient of variation between 0.3 and 0.5 is often observed for electrical signals in many neuronal networks (Shadlen and Newsome 1994, 1998) and is usually taken as an indication of a good reproducibility. It is concluded, therefore, that optical signals ΔF/F and the associated calcium dynamics is rather reproducible in mechanosensory neurons.
The Annulus Erector (AE) motoneuron innervates muscles immediately under the skin, and its electrophysiological properties have been extensively investigated in previous studies (Gu et al. 1991). The AE motoneuron has a unipolar process emerging from the soma between 100 and 200 μm long (see Fig. 5A). From the trunk, many fine branches emerge rich in postsynaptic contacts (Gu et al. 1991). The trunk divides at the first major bifurcation into two branches. The two branches project into the left and right contralateral roots and innervate specific muscles in the leech body. APs recorded in the soma have an amplitude varying between 5 and 10 mV, and the site of initiation of APs has been identified at the first major bifurcation. Figure 5B illustrates APs recorded in the soma when a pulse of depolarizing current was injected into the soma. These APs had an amplitude of ∼5 mV. The red and blue traces reproduce the changes of ΔF/F recorded at the two different-colored locations shown also by corresponding lettering in Fig. 5A. In contrast with the results in mechanosensory neurons, ΔF/F was significantly larger in the dendritic tree than in the soma where ΔF/F was negligible. At the first major bifurcation (indicated in blue and with the letter b), ΔF/F increased in discrete steps, each evoked by an individual AP. The size of each step was ∼1.5%. The increase of ΔF/F in the first major bifurcation occurred with a delay of some milliseconds from the peak of the AP recorded in the soma.
Figure 5C illustrates images of changes of ΔF/F in a color-coded scale at selected times after the current injection. ΔF/F is very small in the soma and in the initial portion of the trunk originating from it. On the contrary, ΔF/F is visible and significant at the first major bifurcation and at all more distal locations. The amplitude of ΔF/F had a hot spot at the first major bifurcation and was rather uniform in the more distal regions.
The Anterior Pagoda is a neuron extensively analyzed in previous investigations (Melinek and Muller 1996). Its function is not known. Like other leech motoneurons, it has a unipolar process with a major bifurcation 150/200 micrometers distally from the soma. The process divides into two major branches projecting into the anterior and posterior contralateral roots (see Fig. 5D). APs recorded in the soma have an amplitude varying between 5 and 10 mV with a characteristic shape, reminiscent of the roof of a pagoda. As with the AE motoneuron, the site of initiation of APs has been identified at the first major bifurcation (Melinek and Muller 1996). When a pulse of depolarizing current was injected through the microelectrode into the soma, a train of small APs was evoked, as shown in Fig. 5E. The colored traces in Fig. 5E indicate that ΔF/F in the soma was almost undetectable, while it became visible at the first major bifurcation 150 μm distal from it. At the first major bifurcation (indicated in blue and with the letter b), ΔF/F increased in discrete steps, each evoked by an individual AP. The size of each step was ∼0.5%. Figure 5F illustrates images of changes of ΔF/F in a color-coded scale at selected times after the current injection. ΔF/F is very small in the soma and in the initial portion of the major axo-dendritic tree originating from it. ΔF/F was significantly larger on locations more distal than the first major bifurcation. Similar results were obtained from a total of five Anterior Pagoda neurons.
The decline of ΔF/F along the initial trunk connecting the first bifurcation to the soma is analyzed in Fig. 5G, where red and black symbols refer to data collected from the AE and the Anterior Pagoda neurons, respectively. ΔF/F declined rather sharply with a space constant varying from 25 to 90 μm (see the solid lines in red and black in G). The decline of ΔF/F along the initial portion within the trunk is in sharp contrast with what was observed in mechosensory neurons (see Fig. 3D) in which the amplitude of ΔF/F in the soma and in the axons was almost uniform. The change of ΔF/F along the branches from the first major bifurcation toward the roots is reproduced in Fig. 5H for the two branches (open and filled symbols) of Anterior Pagoda and AE neurons. In some cases, ΔF/F clearly declined, but in other neurons, ΔF/F in distal branches was large and comparable with that observed at the bifurcation.
The small optical signal recorded in the soma compared with that observed at the first major bifurcation and in distal processes could be due to an inhomogeneous loading of the dye and/or to a dye sequestration into intracellular organelles of the soma of the Anterior Pagoda and AE neurons. As shown in Figs. 1 and 5, the soma of motoneurons is well stained, and its resting fluorescence was comparable to that of mechanosensory neurons as those shown in Figs. 2 and 3. In two experiments, optical signals from the soma were measured while injecting the dye. In these two experiments, ΔF/F in the soma was also small and almost undetectable. In these experiments, optical signals were measured as soon as the soma appeared to be stained when the dye was presumably in the cytoplasm before it could have been absorbed by intracellular stores. These observations suggest that the compartmentalization of calcium transients in the Anterior Pagoda and AE neurons is physiological and is not due to inhomogenity of dye loading and/or to its sequestration into intracellular stores.
Differential invasion of APs in branches
The results described in the previous section were obtained from single trials, and it is possible that the invasion of distal processes following stimulation of the cell body may vary from trial to trial. Therefore the reproducibility of calcium transients in the AE motoneuron was studied in successive trials.
Figure 6 illustrates a stained AE motoneuron in which trains of six or seven APs were evoked in successive trials. Optical signals from the first major bifurcation (shown in blue and with the letter b), the trunk connecting the bifurcation to the cell body (shown in cyan and with a1), and the two processes emerging from the first major bifurcation (shown in red, a2, and green, a3) were recorded and their variability analyzed. In all trials, calcium transients were larger at the first bifurcation and were detected in both processes: indeed, the amplitude of the peak of ΔF/F was very similar in all trials as shown in A–D, left. The average optical signal <ΔF/F> and its coefficient of variation CVΔF/F obtained from the four colored regions of the motoneuron were computed. As shown in A and B, right, in all trials <ΔF/F> in the trunk was approximately three times smaller than at the first major bifurcation. In all trials, optical signals were observed with an average amplitude of 5.9 and of 8.4% in the anterior and posterior branches, respectively. The CVΔF/F at the first major bifurcation decreased to ∼0.3 at the peak of the optical signal and was <0.5 in the two branches. Similar results were obtained in two other AE motoneurons and in one AP motoneuron. These results indicate that calcium transients are detected equally well in the two branches emerging from the first major bifurcation and that calcium dynamics is rather reproducible in leech motoneurons.
Effect of steady currents
The initiation site of APs in the AE and in the Anterior Pagoda neurons can be modulated at some extent by steady currents injected into the soma (Gu et al. 1991; Melinek and Muller 1996): in fact, depolarizing currents move the initiation site toward the soma and hyperpolarizing current moves it toward the roots. It is possible that calcium transients could be modulated in a similar way by currents injected into the soma.
A large optical signal was recorded at the first major bifurcation of the AE neuron shown in Fig. 7A, but no detectable fluorescence changes were measured in its soma (see Fig. 7B). This was also true when a train of seven APs was elicited by passing a depolarizing current pulse through the microelectrode inserted in the soma. Having established that optical signals in the soma were negligible, the preparation was moved by ∼100 μm, and a large portion of the distal dendrites was visualized (see Fig. 7C). When a dozen APs were evoked in the neuron (see Fig. 7E), a large optical signal ΔF/F with an amplitude of ∼15% was observed at the first major bifurcation (see red trace in Fig. 7E). As shown in D and E, a much smaller optical signal was detected along the trunk. When a steady depolarizing current was injected through the microelectrode, the AE neuron fired a train of APs with a frequency of 10 Hz, and the firing rate increased to 30 Hz when the same depolarizing current pulse was superimposed to the steady current (see Fig. 7G). Under these conditions, as shown in Fig. 7, F and G, the optical signal ΔF/F evoked by the same depolarizing pulse was very similar in the bifurcation and in the trunk. When a steady hyperpolarizing current was injected, optical signals elicited by a depolarizing pulse evoking a brief train of APs, were primarily located at the first bifurcation, in accordance with what was observed in the absence of a steady current. Similar results were obtained from a total of six AE motoneurons, showing that the region of calcium transients was moved toward the soma by a steady depolarization.
These results clearly show that APs evoked in the AE motoneuron cause intracellular Ca2+ to increase at locations more distal than the first major bifurcation at a position identified as the initiation site of APs in previous studies (Gu et al. 1991). In the presence of a steady depolarizing current, small but detectable Ca2+ transients were found in the trunk with an amplitude larger than those measured at more distal locations.
L motoneuron and other motoneurons
The L motoneuron innervates longitudinal muscles in the leech skin and projects contralaterally into the anterior and posterior roots. Therefore its dendritic arborization divides into several major branches. Electrophysiological and functional properties of the L motoneuron have been well characterized. APs recorded in the soma have an amplitude slightly larger than in the AE and Anterior Pagoda neurons and vary between 5 and 10 mV (Stuart 1970). In three L motoneurons, we repeated the experiments described in Fig. 5 obtaining similar results: also in L motoneurons, when APs were evoked, ΔF/F was larger in distal processes than in the soma. Similar experiments were repeated in two motoneurons No. 5 and in one motoneuron No. 2, with very similar results. In all analyzed motoneurons, ΔF/F was significantly larger in distal processes than in the soma.
Comparison of calcium transients elicited by stimulation of the soma and distal dendrites
To establish the degree of compartmentalization of optical signals in the Anterior Pagoda neuron, calcium transients originated by direct intracellular stimulation of the soma and by extracellular stimulation of the roots (see Fig. 8A) were compared. In the first case, APs travel toward the roots and in the second case, travel toward the soma. The first major bifurcation and the distal dendrites were visualized (see Fig. 8A). We compared ΔF/F when APs were initiated by passing a depolarizing current pulse in the soma (see Fig. 8, B and C) and when APs were evoked by extracellular stimulation of the anterior (see Fig. 8, D and E) or posterior root (see Fig. 8, F and G). As shown in the Fig. 8, C, E, and G, the stimulations evoked a train of a similar number of APs clearly detected by the intracellular microelectrode impaling the soma.
For the three stimulations, the time course and spatial profile of ΔF/F was similar: in all three cases, ΔF/F was significantly larger at the first major bifurcation (see B, D, and F) and was small and almost undetectable in the soma. Calcium transients evoked by the stimulation of the anterior root propagated to the processes entering into the posterior root and vice versa but hardly reached the soma.
All neurons of the leech CNS are unipolar, i.e., there is only one major trunk emerging from their soma. When an intracellular microelectrode is inserted in the soma of a leech neuron, it is possible to record APs of the usual size of 60–70 mV only from mechanosensory neurons, the S interneuron, the Leydig cell, and the Retzius cells, whereas APs with a lower amplitude are recorded from the soma of the great majority of other neurons. Similar observations have been made in neurons of other invertebrate nervous systems such as round worms (White et al. 1986) and crustacea (Combes et al. 1999a,b). These observations raise several questions: why is the soma of most leech neurons not excitable? Why do mechanosensory neurons have APs of the usual size in the soma?
Calcium indicators have been often used to image AP initiation in a variety of neurons, and although they have some disadvantages (see Waters et al. 2005 for a thorough discussion), they are a powerful tool to investigate compartmentalization at a cellular level and to monitor changes of intracellular Ca2+ concentration. Our results show that in the leech nervous system Ca dynamics in mechanosensory neurons compared with the great majority of other neurons are different. Ca dynamics in the Anterior Pagoda neuron, in the AE, L, No. 2 and No. 5 motoneurons and presumably in most other motoneurons and interneurons are highly compartmentalized and calcium transients are restricted to the distal processes and hardly reach the soma (see Figs. 5–8). By contrast, Ca transients are observed over the entire arborization of mechanosensory neurons (see Figs. 2–4). This differential compartmentalization in leech neurons is observed consistently in different trials and is affected by a small variability (see Figs. 4 and 6). Therefore compartmentalization of Ca dynamics mirrors the compartmentalization of electrical events (Gu et al. 1991; Melinek and Muller 1996).
Ca dynamics in leech neurons
A single AP in mechanosensory neurons evoked an optical signal ΔF/F with peak amplitude varying between 2 and 6%. Optical signals ΔF/F of similar amplitude caused by a single AP backpropagating in the dendrites of CA1 pyramidal neurons (Callaway and Ross 1995; Frick et al. 2003; Spruston et al. 1995) and of neocortical pyramidal neurons (Markram et al. 1995) have been observed. Larger optical signals evoked by a single AP have been measured in the intact retina (Denk and Detwiler 1999). As shown in Figs. 2–4, the value of ΔF/F is similar in the soma and in the distal processes and in the branches innervating the skin. A similar behavior is observed in many other mammalian neurons of the CNS, where calcium transients initiated by APs are observed with approximately the same size in the cell body and in the dendrites such as in CA1 pyramidal neurons (Frick et al. 2003). In L2/3 pyramidal neurons, the peak value ΔF/F evoked by a single AP is ∼10% in the soma, increases by two or three times at a distance of ∼100 μm along the apical dendrites and then decreases at larger distances from the soma (Callaway and Ross 1995; Spruston et al. 1995; Waters et al. 2003,2005).
As shown in Figs. 5–8 in the Anterior Pagoda neuron, in the AE, L, No. 5 and No. 2 motoneurons ΔF/F associated with APs occur at the first major bifurcation and sharply decline with a space constant varying between 25 and 80 μm along the trunk connecting to the soma. In these neurons, Ca transients can be observed only in distal processes and are strongly compartmentalized. This observation is in agreement with the analysis of Ca transients in Retzius cells of the leech (Beck et al. 2001) where large optical signals ΔF/F were observed primarily along the cell dendrites. Ca transients were not observed in the soma of these neurons even when large depolarizing currents were injected into the soma itself (Fig. 5), and therefore the absence of Ca transients in the presence of APs cannot be attributed only to the small electrical signals reaching the soma. These observations show that the density of voltage-gated Ca channels in the soma of these neurons is significantly lower than in distal processes. The sharp decline of Ca2+ transients along the initial trunk is caused by the combination of several mechanisms: the restricted diffusion of Ca2+ ions along a narrow structure; the absorption and buffering of diffusing Ca2+ ions by intracellular stores; and a differential distribution of Na+ and Ca2+ voltage-gated channels along the trunk. A Ca2+-activated K+ conductance that has been described in the processes of the Anterior Pagoda neuron (Wessel et al. 1999a) is likely to be a major determinant also for the attenuation of the amplitude of APs along the same structure and therefore will reduce the Ca2+ inflow through voltage-gated Ca2+ channels along the trunk. APs in motoneurons are expected to have the usual amplitude of ∼70 mV, and their initiation site has been located at the first major bifurcation (Melinek and Muller 1996). An AP, traveling from the first bifurcation to the soma is attenuated between 10 and 20 times over a distance of 100–200 μm, indicating an effective space constant of ∼50 μm, similar to the decline of the optical signal ΔF/F (see Fig. 5). A space constant of 50 μm in an entirely passive cable can originate from a very high axial resistivity and an unusually high membrane conductance (Rall and Agmon-Snir 1998). Therefore it is also likely that active properties such as the Ca2+-activated K+ conductance (Wessel et al. 1999a,b) contribute to the large attenuation of voltage and the associated calcium transients along the major trunk.
This compartmentalization is reminiscent of what is observed in several invertebrate preparations, such as in barnacle neurons (Krauthamer and Ross 1984; Ross and Krauthamer 1984) and in Aplysia neurons (Gorman and Thomas 1978), where the initiation site of APs is located at 100–200 μm from the soma. In vertebrate and in particular in mammalian neurons, the site of AP initiation is usually located in the axon at some distance from the soma (Stuart and Sakmann 1994; Stuart et al. 1997). In cerebellar Purkinje rat neurons, the site of AP initiation has been localized at ∼75 micrometers from the soma at the first major axonal branch (Clark et al. 2005). In these neurons, however, the amplitude of the AP recorded in the soma is large, almost identical to that measured at the site of AP initiation because it back propagates.
Mechanosensory neurons (see Fig. 9A) differ from other leech neurons in many aspects: indeed they integrate sensory input in the skin and send signals to other neurons in the same ganglion and in neighboring ganglia. Nervous signals generated in the skin are transmitted to the soma, at a distance of some millimeters apart, by usual action potentials. These APs have to reach presynaptic endings in the ganglion and in neighboring ganglia and travel at short distance from the soma. Therefore the soma of leech mechanosensory neurons is close to the pathway where action potentials must travel.
Leech interneurons and motoneurons (see the scheme of Fig. 9B) need to integrate synaptic inputs within the ganglion itself. Leech interneurons and motoneurons obtain a functional segregation of their processes by having many of their synaptic inputs concentrated between the soma and the first major bifurcation of their arborization. Their presynaptic contacts to other neurons and/or muscles are primarily located more distally than the first major bifurcation. To process properly some of the synaptic inputs, it is useful and possibly necessary to have a poorly excitable membrane with a low density of voltage-gated Na and Ca channels, avoiding the initiation of action potentials that would temporarily wipe out all synaptic signals being passively propagated. Therefore leech neurons have a low density of voltage-gated channels in those regions primarily destined for synaptic integration, i.e., between the soma and the first major bifurcation.
Leech motoneurons and possibly also interneurons achieve a functional segregation by controlling the density of voltage-gated Na and Ca channels: their density is low in the soma up to its first major bifurcation. The density of voltage-gated channels becomes high at the first major bifurcation, where APs are initiated. In leech interneurons and motoneurons, the flow of information goes from the trunk to the first bifurcation, where action potentials are initiated, and finally to the presynaptic endings in the ganglion or onto the muscles. The soma of the great majority of leech neurons has a low density of Na- and Ca-gated channels, and therefore it is poorly excitable and does act as a threshold for incoming synaptic signals as the soma of more conventional neurons does. Nonetheless passive properties of the soma can influence the way in which APs travel on excitable distal processes, for instance by modulating their reflection at bifurcations (Baccus et al. 2000, 2001) and by modifying the pattern of APs trains (Amir and Devor 2003; Amir et al. 2005) with the occasional insertion of “extra APs.”
Functional and anatomical properties of leech and vertebrate neurons, apparently so different, can be reconciled. The first major bifurcation has the same role of the axon hillock of conventional vertebrate neurons. Vertebrate neurons, however, have developed a morphology better suited to their functions and have the genetic and biochemical machinery (Craig and Banker 1994; Wodarz 2002) able to produce a polarized structure with distinct functional properties. In vertebrate neurons, the soma of unipolar leech neurons has been moved to the major bifurcation, obtaining a better functional segregation with an excitable axon on one side and the dendritic tree on the other side (see Fig. 9C).
This work was supported by the Fondo per gli Investimenti della Ricerca di Base and Comitato internazionale Programmazione Economica (Center for Genetics of Regeneration and Neurodegenerative Diseases Frivli Venezia Giulia) grants.
We thank Prof. John Nicholls and Prof. Larry Cohen for comments on the manuscript and for helpful suggestions. We also thank W. Vanzella and P. Bonifazi, who assisted in the development of the software for image analysis and for the preparation of the figures.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2005 by the American Physiological Society