Journal of Neurophysiology

Activity-Dependent Increase of the AHP Amplitude in T Sensory Neurons of the Leech

Rossana Scuri, Riccardo Mozzachiodi, Marcello Brunelli


We identified a new form of activity-dependent modulation of the afterhyperpolarization (AHP) in tactile (T) sensory neurons of the leech Hirudo medicinalis. Repetitive intracellular stimulation with 30 trains of depolarizing impulses at 15-s inter-stimulus interval (ISI) led to an increase of the AHP amplitude (∼60% of the control). The enhancement of AHP lasted for ≥15 min. The AHP increase was also elicited when a T neuron was activated by repetitive stimulation of its receptive field. The ISI was a critical parameter for the induction and maintenance of AHP enhancement. ISI duration had to fit within a time window with the upper limit of 20 s to make the training effective to induce an enhancement of the AHP amplitude. After recovery from potentiation, AHP amplitude could be enhanced once again by delivering another training session. The increase of AHP amplitude persisted in high Mg2+ saline, suggesting an intrinsic cellular mechanism for its induction. Previous investigations reported that AHP of leech T neurons was mainly due to the activity of the Na+/K+ ATPase and to a Ca2+-dependent K+ current (I K/Ca). In addition, it has been demonstrated that serotonin (5HT) reduces AHP amplitude through the inhibition of the Na+/K+ATPase. By blocking the I K/Ca with pharmacological agents, such as cadmium and apamin, we still observed an increase of the AHP amplitude after repetitive stimulation, whereas 5HT application completely inhibited the AHP increment. These data indicate that the Na+/K+ATPase is involved in the induction and maintenance of the AHP increase after repetitive stimulation. Moreover, the AHP increase was affected by the level of serotonin in the CNS. Finally, the increase of the AHP amplitude produced a lasting depression of the synaptic connection between two T neurons, suggesting that this activity-dependent phenomenon might be involved in short-term plasticity associated with learning processes.


The afterhyperpolarization (AHP) following action potentials is an important determinant of neuronal firing (Gustafsson 1974). In many nerve cells, both early and late components of the AHP have been reported. The early component is primarily due to the activation of voltage-gated K+ currents. In some neurons, a fast Ca2+-activated K+ current (I K/Ca) also contributes to this component (Sah 1996; Storm 1987). The late component of the AHP is due to the activation of slower Ca2+-dependent K+conductances that are secondary to Ca2+ entry during the action potential (Gustafsson 1974;Hill et al. 1985; Sah 1996). In addition, in several neuronal types, a sustained burst of action potentials elicits a prolonged AHP due to an increase of the Na+/K+ ATPase activity as a result of changes in ion distribution after the sustained burst of action potentials. The involvement of the Na+/K+ ATPase in the AHP generation has been described in mammalian non-myelinated axons (Den Hertog and Ritchie 1969), in insect mechanoreceptors (French 1989), in terminals of lizard motor neurons (Morita et al. 1993), in lamprey sensory neurons (Parker et al. 1996), and in lobster olfactory receptors (Corotto and Michel 1998). The component due to the Na+/K+ ATPase is generally blocked by cardiac glycosides, such as ouabain (Baylor and Nicholls 1969a; Parker et al. 1996) and depends on temperature and extracellular potassium (Jansen and Nicholls 1973; Skou 1965) but is not markedly voltage sensitive (Morita et al. 1993). Trains of action potentials are generally the means through which neurons encode the transmission of signals along the axon to the synaptic terminals. Usually, the AHP has two major functions: it limits the firing frequency of the neuron and is responsible for generating spike-frequency adaptation. Therefore, because AHP is a key factor for setting the firing activity of several neurons (Corotto and Michel 1998; Pineda et al. 1998;Sanchez-Vives et al. 2000), it is often a strategic target for modulating the neural activity (Blitzer et al. 1994; Gerber and Gahwiler 1994; Pedarzani and Storm 1995; Pedarzani et al. 1998).

In T sensory neurons of the leech Hirudo medicinalis, a mechanical stimulation delivered onto their receptive field produces a burst of action potentials accompanied by an AHP that lasts for several seconds (Baylor and Nicholls 1969a). The amplitude and duration of the AHP depend on the number of action potentials: the longer the burst, the larger and longer the AHP will be. Two factors are responsible for this AHP: the activity of a Na+/K+ ATPase contributes to the AHP generation for the 75% of the total amplitude (Baylor and Nicholls 1969a; Van Essen 1973), whereas the residual 25% AHP is due to aI K/Ca (Jansen and Nicholls 1973), which is not yet classified in detail because of the difficulty finding specific blockers of different K+/Ca2+-dependent channels types in leech T neurons.

It has been previously demonstrated that both the stimulation of the serotonergic Retzius cells and bath application of serotonin (5HT) depress the AHP amplitude in T neurons without affecting other membrane properties (Belardetti et al. 1984; Catarsi and Brunelli 1991). In addition, Catarsi and Brunelli (1991) showed that 5HT does not affect either the amplitude or the duration of the action potential of T neurons. In that paper, the authors isolated the Na+/K+ATPase component by blocking the I K/Cawith Cd2+ and showed that the residual AHP was still reduced by 5HT. They also reported additional evidence of a direct effect of 5HT on the Na+/K+ ATPase. Finally,Catarsi et al. (1993) showed that 5HT depresses the Na+/K+ ATPase activity via the cAMP pathway. Thus 5HT selectively inhibits the Na+/K+ ATPase activity, so reducing the AHP amplitude of T neurons.

Catarsi et al. (1990) and more recently Zaccardi et al. (2001), reported that at behavioral level, low-rate repetitive electrical stimulation delivered onto the skin of the leech leads to a decrement of the evoked swim response that conforms to the operational definition of habituation. In a swim habituation paradigm using a semi-intact leech preparation (Debski and Friesen 1985), a light-stroking stimulus onto the skin selectively activates T sensory cells (Debski and Friesen 1987). Moreover, swim habituation is likewise elicited by intracellular stimulation of T neurons (Brodfuehrer et al. 1995). To investigate the effects of repetitive stimulation into single T cells, we have first studied whether either repetitive intracellular stimulation of T neurons or repetitive mechanical stimulation of T receptive field could produce changes in the AHP amplitude. Our results indicate that repetitive stimulation brings about an increase of the AHP amplitude that is intrinsic to the T neuron. We have also investigated which component, generating the AHP in T cells, is involved in this positive modulation. Furthermore, we found that the modulation of the Na+/K+ATPase activity is essential for the AHP increase. Finally, we show that the AHP increase leads to a reduction of the synaptic strength between T neuron and its followers.


Animals and surgery

Adult leeches (Hirudo medicinalis) were purchased from a local supplier and from Ricarimpex (Eysines, France) and maintained at 16°C in artificial pond water (in mM, 0.48 NaCl, 0.0067 KCl, 0.0034 Ca(NO3)2· 4H2O, 0.001 MgSO4· 7H2O, and 0.046 Tris-maleate, buffered to pH 7.4 with HCl), under natural daylight rhythm. A short chain of ganglia was removed from mid-body level of anesthetized animals (dipped for 10 min in 10% ethanol in tap water) by cutting the body wall along the mid-line and by opening the ventral sinus. The surgery was carried out at room temperature. Ganglia were kept at 16°C in saline solution (see following text) for ≥1 h before the recording began.


Except where noted, all experiments were performed in leech saline solution consisting of (in mM) 115 NaCl, 4 KCl, 1.8 CaCl2, and 10 glucose, buffered at pH 7.4 with 10 Tris-maleate. A high-Mg2+ solution consisting of (in mM) 95 NaCl, 4 KCl, 1.8 CaCl2, 20 MgCl2, and 10 glucose, buffered at pH 7.4 with 10 Tris-maleate was obtained by replacing Na+ with Mg2+ to block chemical synaptic transmission in the ganglia (Baylor and Nicholls 1969b). Cadmium chloride, 5HT, and the peptide apamin (all purchased from Sigma, St. Louis, MO) were freshly dissolved in the leech saline just before their application.

Electrophysiological technique

A single ganglion (Fig.1 A) was pinned ventral side up to a silicone elastomer (Sylgard)-lined (Dow Corning, Midland, MI) recording chamber. T neurons were identified by the size and the location of their cell bodies as well as by their firing pattern (Nicholls and Baylor 1968). For intracellular recordings, the soma was impaled with borosilicate microelectrodes (Hilgerberg GmbH) filled with 4 M potassium acetate and having impedances of 60–80 MΩ. Recordings were performed in current-clamp mode. The signals were amplified with an appropriate electrometer and viewed on a storage oscilloscope (Tektronix, Wilsonville, OR). The data were filtered and digitized for analysis using National Instruments data-acquisition software with a BNC-2090 National Instruments series interface (National Instruments S.r.l., Milano, Italy). All the experiments were carried out at room temperature. AHP was induced by injecting 3-s trains of intracellular depolarizing pulses (200 ms, 2.5 Hz; Fig. 1 B). The discharge frequency of T cells during each series of trials was kept constant by adjusting the amount of injected current ranging from 0.4 to 0.8 nA. During a single 200-ms pulse, T neurons fired seven to eight spikes (data not shown) overlapping a pattern similar to that produced when their receptive field was stimulated. Because AHP is very sensitive to the quality of the impalement, only cells with resting potential greater than −40 mV and input resistance (measured with 200-ms hyperpolarizing pulses, 0.5 nA) > 60 MΩ were selected for this study.

Fig. 1.

Schematic drawings of the two preparations used in this study. A: a single ganglion was pinned in the recording chamber and the intracellular recording was performed with microelectrodes inserted into the soma of a T neuron. B: recording of the afterhyperpolarization (AHP) induced by a 3-s train of depolarizing impulses. The AHP amplitude was measured from the resting potential before the stimulus to the peak of the hyperpolarization.C: semi-intact preparation consisting of a patch of skin connected with the ganglion through the lateral roots. A glass capillary with the polished tip was used to stimulate the neurons receptive fields. D: recording of the AHP elicited by activation of a T neuron receptive field by 3-s brushing onto the skin with the tip of the glass capillary.

In one set of experiments, a segmental ganglion was removed together with a patch of skin corresponding to either a right or a left hemi-segment of the body wall (7–8 annuli, Fig. 1 C). The patch of skin was kept connected with the ganglion via the lateral roots. It contained the dorsal, lateral, and ventral portion of the skin where the receptive fields of the ipsilateral sensory neurons are located. The patch of skin and the ganglion were pinned in the recording chamber.

Experimental design

In the first part of this study, we investigated the ability of low-frequency repetitive intracellular stimulation to modulate AHP amplitude in T neurons. The stimulation protocol consisted of a series of 3-s trains delivered at different interstimulus intervals (ISI). We tested ISIs from 10 to ≤20 s. During each train, the AHP was recorded. The means of the AHP amplitudes were calculated. For convenience, only AHP1, AHP5, AHP10, AHP15, AHP20, AHP25, and AHP30 were plotted in the figures. Each mean value plotted in figures was normalized to the one measured during the first train (AHP1) taken as 100%.

In experiments with semi-intact preparations, T neurons were activated by mechanical stimulation of the skin (Fig. 1 C). Once a T sensory neuron was impaled, its major ipsilateral receptive field was identified by applying light mechanical pressure onto the skin with a glass capillary with rounded tip, connected with a piezo-electric manipulator. The pressure delivered was set to activate exclusively the T neurons and none of the two other classes of sensory cells (P and N neurons, data not shown). To generate a detectable AHP, the skin was mechanically stimulated for 3 s by dragging the capillary back and forth along the horizontal axis on the receptive field of a T neuron. The AHP elicited with this procedure was similar in amplitude and time course to the one induced by intracellular current injection (Fig.1 D).

We further investigated whether low-frequency repetitive stimulation of a T cell might bring about changes in the synaptic strength in a postsynaptic neuron. It is known that T neurons are interconnected each other within the ganglion and with adjacent ones via an excitatory synapse consisting of an electric and a chemical component (Acklin 1988; Baylor and Nicholls 1969b). Therefore we focused on the synapse between two T sensory neurons placed in the same side of the ganglion. The strength of the synaptic connection was detected by stimulating the presynaptic T neuron with a 200-ms depolarizing pulse (current injected: 0.2–0.4 nA, eliciting 5/6 action potentials) and the amplitude of the excitatory postsynaptic potential (EPSP) was recorded in the postsynaptic T neuron current clamped at −60 mV. Then, the presynaptic T neuron was stimulated with 15 trains of depolarizing pulses (3-s duration, 15 s ISI). The EPSP amplitude was tested again at the end of the stimulation and at different time points (5, 10, and 15 min) after training. The injected current was adjusted to ensure that the synapse between two T neurons was probed with the same number of spikes.

Statistical evaluation

AHP amplitudes were measured from the baseline of the resting potential before the stimulation, down to the peak of the maximal hyperpolarization (Fig. 1 B). The EPSP was measured from the resting membrane potential to the peak of the maximal depolarization during 200-ms stimulation of the presynaptic T cell. All values were expressed as means ± SE: however, because the data were nonparametric, in the figure, the SE will be not shown in the graphs and in the histograms. Statistical analysis was done using the Friedman test for repeated measures and the two-way repeated-measures ANOVA. The Mann-Whitney U test has been used to compare the means of two independent groups. Statistical significance was set atP < 0.05 and indicated with asterisks in all the graphs. Statistical analysis was performed by using the SPSS software package (Advanced Models 10.1).

Data fitting and correlation tests were carried out by using Mathematica 4.0 software (Wolfram Research, Champaign, IL).


Low-rate repetitive intracellular stimulation increases AHP amplitude

To investigate whether AHP could be modified by repetitive activity, we developed a protocol for intracellular stimulation of T sensory neurons. We delivered a series of trains of depolarizing pulses (see methods). The injected current was selected to elicit seven to eight spikes during each 200-ms pulse: from preliminary studies, we have observed that this number of action potentials well represented the physiological discharge of a T neuron when its receptive field was activated (data not shown). AHP amplitude gradually increased when a T neuron was stimulated with 30 trains of depolarizing pulses at 15-s ISI (Fig. 2, Aand B). After the fifth train, the AHP amplitude (AHP5, 8.5 ± 1.19 mV) was already significantly larger than the one recorded after the first train (AHP1, 6.75 ± 0.87 mV; Friedman test,P = 0.0016). The AHP amplitude following the 30th train (AHP30, 10.7 ± 1.26 mV) was ∼60% larger than AHP1 (Friedman test: P = 0.0001). However, we noticed that ∼10% of the T neurons analyzed did not exhibit any increase of AHP amplitude during repetitive stimulation. In the first part of this study, we did not include these cells in the statistical analysis. AHP did not return to its initial amplitude right after the end of the training; on the contrary, its recovery was prolonged: AHP amplitude (7.8 ± 1.11 mV) was still significantly larger than AHP1 (Friedman test:P = 0.0455; Fig. 2, B and C) 10 min after the presentation of the 30th train. Once the initial AHP amplitude was restored, a second training session was able to induce again an AHP increase (Fig. 3), indicating that T neurons have the ability to generate repetitive enhancement of AHP amplitude.

Fig. 2.

Repetitive intracellular stimulation induced a lasting enhancement of the T neuron AHP amplitude. A: raw data from a single neuron showing the gradual increase of the AHP amplitude during a training session consisting of 30 trains of depolarizing pulses with an interstimulus interval (ISI) of 15 s. B: traces from the same neuron in A showing the time course of the AHP recovery after training. C: summary data illustrating the time course of the AHP increase (left) and its recovery (right; n = 23). The associated fitted curves show the progressive increase of the AHP amplitude and the time course of the decay of the AHP enhancement. In this and in all the graphs and histograms of the following figures, each symbol represents the means of AHP amplitude normalized to the mean obtained after the first train (AHP1) taken as 100%. In this and in all the following figures, *, statistical significance,P < 0.05, evaluated with Friedman test.

Fig. 3.

AHP amplitude increase can be induced repetitively. The graph summarizes the increase of the AHP amplitude produced by 2 sessions of 15 trains with an ISI of 15 s spaced 20 min apart (n = 6). The amplitude and the time course of the AHP increase elicited during the 1st (■) and the 2nd session (○) does not differ statistically (2-way repeated-measures ANOVA: F = 0.0101, α = 0.05, P = 0.922).

The AHP increase can be fit with a function of the formy=a+b(1ecx) where y represents the AHP amplitude and xthe time measured in ISI (corresponding to 15 s) units. Using the AHP data during repetitive stimulation and a nonlinear regressiona, b, and c were estimated. The AHP data are best fit by the equationy=100.012+61.0752(1e0.124x) where τ = 1/0.124 . ISI is ∼120 s (Fig. 2 C).

Note that, the AHP increase was higher during the first 15 trains of stimulation then reaching a steady state (Fig. 2 C). In the subsequent experiments, we delivered only 10–15 trials.

The decay of AHP amplitude during recovery can be fit with a function of the formy=100+aebx where y represents the AHP amplitude and xthe time of the recovery. Using the AHP data during recovery and a nonlinear regression, we estimated the parameters a, b. Data are best fit by the equationy=100+65.9597e0.294058x where the time is in units of 2 min and τ = 1/0.294058 · 2 is ∼7 min (Fig. 2 C).

Increase of the AHP amplitude can be elicited also by repetitive stimulation of T neuron's receptive field

Mechanical stimulation of a T neuron's receptive field elicits a burst of action potentials in the cell body, followed by an AHP similar in amplitude and time course to the one elicited by intracellular injection of current (Baylor and Nicholls 1969a;Jansen and Nicholls 1973; Van Essen 1973). In semi-intact preparations (see methods), 15 stimuli (3-s train duration, 15 s ISI) were delivered onto the skin, while the neuronal activity was recorded from the soma. In the T cells examined, AHP amplitude increased during repetitive stimulation ≤135.6% (AHP15, 6 ± 1 mV) of the control value (AHP1, 4.5 ± 0.6 mV) (Fig.4). The amount of AHP increase was comparable with that induced by the injection of 15 trains of depolarizing pulses with 15-s ISI (2-way repeated-measures ANOVA between data in Figs. 2 and 4: F = 0.0135, α = 0.05, P = 0.909).

Fig. 4.

AHP amplitude increase can be elicited by repetitive mechanical stimulation of the T neuron receptive field in semi-intact preparation.A: raw data from a single T neuron showing the gradual increase of the AHP amplitude during repetitive activation of its receptive field with 15 mechanical stimuli of 3 s with an ISI of 15 s. B: the graph summarizes the activity-dependent increase of the AHP amplitude in a group of 11 T neurons.

AHP increase is sensitive to the ISI duration

We analyzed whether the duration of ISI could affect the induction of the AHP amplitude increase elicited by repetitive stimulation. To test whether a specific ISI was required for the increase of the AHP amplitude, we delivered 15 trains with ISIs between 10 and 20 s. We detected a significant increase of the AHP amplitude in T cells trained with ISI <20 s (Fig. 5,A and B). When the ISI was 20 s, the AHP increase was strongly impaired: after 5 trains, the AHP reached a plateau and did not increase any further (Fig. 5 A; Friedman test: AHP5 vs. AHP1,P = 0.0455; AHP10 vs. AHP1, P = 0.0956; AHP15 vs. AHP1,P = 0.1824). Indeed, the AHP increase was also induced for ISIs <10 s (data not shown): in this case, we were unable to analyze modulation of AHP because the membrane potential was often still hyperpolarized when the next stimulus occurred and this caused ambiguity in the measurement of the actual AHP amplitude.

Fig. 5.

ISI duration is a critical parameter for the induction of AHP. A: graph summarizing the effects of different ISI duration in the AHP increase. The AHP amplitude during repetitive stimulation performed in 3 different conditions is plotted: T cells trained with 15-s ISI (■, n = 23) and 17-s ISI (●, n = 8; Friedman test: AHP5 vs. AHP1, P = 0.0253; AHP10 vs. AHP1, P = 0.0207; AHP15 vs. AHP1, P = 0.0225) show an increase of the AHP amplitude; repetitive stimulation with 15 trains with an ISI of 20 s elicits only a transient increase of AHP amplitude (○, n = 10) that is significantly different from the one induced by the same training with ISI < 20 s (2-way repeated-measures ANOVA:F = 5.380, α = 0.05, P = 0.011). B: the scatter plot shows examples of the actual increase of the AHP amplitude at different ISIs .

Increase of the AHP amplitude depends on the intrinsic activity of T neuron

To test whether the enhancement of the AHP amplitude was due either to the intrinsic properties of T neuron or to the involvement of an extrinsic synaptic pathway recruited by T neuron activity, we performed experiments in high-magnesium (20 mM) solution to block chemical synaptic transmission (Baylor and Nicholls 1969b). First, 15 trains at 15-s ISI were delivered to the T cell in normal saline and the AHP significantly increased during the stimulation (Fig. 6 A; Friedman test: AHP5 vs. AHP1,P = 0.022; AHP10 vs. AHP1, P = 0.0412; AHP15 vs. AHP1,P = 0.0143). During 15-min rest, the ganglia were perfused with magnesium solution and then a second stimulation was delivered. Treatment with magnesium reduced AHP1(3.43 ± 0.4 mV) to 66.7% of the AHP1(5.14 ± 0.5 mV) recorded in normal saline, but an increase of the AHP amplitude still occurred (Fig. 6 A; Friedman test: AHP5 vs. AHP1,P = 0.020; AHP10 vs. AHP1, P = 0.0143; AHP15 vs. AHP1,P = 0.014). This result suggests the involvement of the intrinsic cellular machinery in the genesis of the AHP increase.

Fig. 6.

Effects of 20 mM MgCl2 and subthreshold stimulation on the AHP amplitude increase. A: the AHP increase induced by 15 trains with an ISI of 15 s in normal saline (■) does not differ from the increase induced by the same stimulation pattern when ganglia were perfused with saline containing 20 mM magnesium (○; n = 7; 2-way repeated-measures ANOVA: F = 1.537, α = 0.05, P = 0.239). B: the AHP increase did not occur when the injected current, used for repetitive stimulation of T neuron, was not sufficient to generate action potentials. The stimulation consisted of 15 trials with an ISI of 15 s. The injected current used was suprathreshold only during the 1st and the 15th train, whereas it was subthreshold for the rest of the training. The amplitudes of the AHPs recorded during the 1st and the last train were not statistically different (n = 12; Friedman test: P = 1). Note that in the same group of cells a previous training consisting of 15 suprathreshold trains induced an increase of the AHP amplitude (Friedman test:P = 0.0086).

In addition, we have observed that the discharge of action potentials was necessary during the repetitive stimulation for inducing AHP increase. In fact, an intracellular stimulation ineffective in eliciting action potentials (subthreshold stimulation) failed to produce AHP increase. In these experiments, the usual protocol was modified as follows: 15 trains at 15-s ISI were still delivered, but after the first train, the injected current was switched from the suprathreshold current normally used (see methods) to a subthreshold current (0.1–0.3 nA) that did not induce action potentials. This current was used during all the training stimulation up to the last stimulus. Then the current was switched back to the suprathreshold value before the delivery of the 15th train. No significant change was detected when the amplitude of AHP15 (7.88 ± 0.92 mV) was compared with that of AHP1 (7.59 ± 0.94 mV; Fig.6 B). Therefore a subthreshold depolarization of the membrane is not sufficient to trigger the increase of AHP amplitude, suggesting that this phenomenon depends on the firing of action potentials during the stimulation.

AHP amplitude increase involves the Na+/K+ATPase activity

As estimated by Baylor and Nicholls (1969a), the contributions of the Na+/K+ATPase and I K/Ca for the AHP amplitude in T neurons are 75% and 25%, respectively. To investigate whether one of these two components was involved in the AHP increase produced by repetitive stimulation, specific pharmacological agents were used to block I K/Ca and 5HT was used to inhibit the Na+/K+ ATPase activity (see following text). We used CdCl2, which blocks the voltage-gated calcium channels from the extracellular side of the plasma membrane thus preventing the activation ofI K/Ca (Catarsi and Brunelli 1991; Stewart et al. 1989). Bath application of 0.2 mM CdCl2 reduced the AHP1 (5.78 ± 1.01 mV) to 68.5% of AHP1 (8.44 ± 1.39 mV) recorded in normal saline as previously described by Catarsi and Brunelli (1991). Nevertheless, 10 trains of depolarizing pulses delivered at 15-s ISI still elicited an increase of the AHP amplitude similar to the one induced in normal saline (Fig.7 A). In addition, 15-min incubation with 1 nM apamin, a more specific inhibitor ofI K/Ca involved in afterpotentials (Hugues et al. 1982; Zhang and Krnjevic 1987), produced a 25% reduction of the AHP1 as reported by Mozzachiodi et al. (2001) but did not prevent the increase in AHP amplitude (Fig.7 B). Then, ganglia were treated for 10 min with 50 μM 5HT, which has been demonstrated to depress AHP in T neurons through the selective inhibition of the Na+/K+ ATPase (Catarsi and Brunelli 1991). We observed that 5HT completely prevented AHP amplitude from increasing (Fig.8 A) during repetitive stimulation with 10 trains with an ISI of 15 s, whereas this stimulation protocol elicited an increase of the AHP amplitude in normal saline (Fig. 8 A). Moreover, when 50 μM 5HT was added to saline containing 1 nM apamin, the repetitive stimulation failed to induce AHP increase, while AHP enhancement occurred in saline containing only apamin (Fig. 8 B).

Fig. 7.

The I K/Ca is not involved in the increase of the AHP amplitude in T neurons. Perfusion with theI K/Ca blockers CdCl2 (0.2 mM;n = 9, A) or apamin (1 nM;n = 10, B) did not prevent the repetitive stimulation with 10 trains, with an ISI of 15 s, from inducing an increase of the AHP amplitude.

Fig. 8.

Serotonin (5HT) application prevents the increase of the AHP amplitude induced by repetitive stimulation. A: when ganglia were perfused with 50 μM 5HT for 10 min, repetitive stimulation with 10 trains with an ISI of 15 s did not produce any significant increase of the AHP amplitude (n = 7; Friedman test: AHP5 vs. AHP1, P = 1; AHP10 vs. AHP1, P = 0.7055). The same pattern of stimulation elicited an increase of the AHP amplitude during control in normal saline and after 15 min washout.B: 50 μM 5HT for 10 min impaired the AHP amplitude increase also when I K/Ca was blocked with 1 nM apamin (n = 7; Friedman test: AHP5vs. AHP1, P = 0.0736; AHP10vs. AHP1, P = 0.1797).

Na+/K+ATPase activity is also necessary for the maintenance of the AHP amplitude increase

In the next set of experiments, once AHP amplitude was successfully enhanced (15 trains at 15 s ISI; •, Fig.9), 50 μM 5HT was added to saline right after the delivery of the last stimulus. Then an AHP was elicited every 2 min. After 5HT was bath-applied, the AHP enhancement was completely abolished and the AHP amplitude was reduced to ∼50% of the control (●, Fig. 9). Recovery of the cells perfused with 5HT was significantly different from recovery in normal saline (○, Fig. 9).

Fig. 9.

5HT affects the recovery of AHP amplitude after activity-dependent enhancement. At the peak of the AHP amplitude enhancement induced by 15 trains with an ISI of 15 s (■,n = 18), 50 μM 5HT was added right after the end of the training. 5HT strongly reduced the AHP amplitude and removed any trace of potentiation (●, n = 11). Also note that the time course of AHP recovery in the presence of 5HT differed significantly from the one shown by neurons perfused with saline (○, n = 7; 2-way repeated-measures ANOVA: F = 4.840, α = 0.05,P = 0.042).

Seasonal variation in AHP amplitude enhancement elicited by repetitive stimulation

Catarsi et al. (1990) reported seasonal variations in the contents of 5HT in leech segmental ganglia with the highest levels of 5HT detected in December, the lowest in May. We investigated whether seasonal changes of 5HT levels in the ganglia could affect AHP increase. We analyzed T neurons that exhibited a significant increase of AHP during repetitive stimulation throughout a year. The largest AHP enhancement (evaluated as AHP10–AHP1 amplitude) was observed in May (AHP1 = 8.44 ± 0.55 mV; AHP5 = 10.8 ± 0.6 mV; AHP10 = 12.66 ± 0.8 mV), whereas the smallest AHP increase was detected in December (AHP1 = 4.86 ± 0.2 mV; AHP5 = 5.63 ± 0.8 mV; AHP10 = 6.12 ± 0.5 mV; Fig.10 A). Figure 10 Bshows the AHP increase measured in May (•) and in December (○), respectively. In May, the increase in the AHP amplitude was significantly larger than that recorded in December.

Fig. 10.

Seasonal variations of AHP increase. A: histogram summarizing the changes in the increase of the AHP amplitude detected along a year time (January, n = 8; March,n = 9; April, n = 7; May,n = 14 and December, n = 12). In the y axis, the normalized increase of the AHP amplitude (evaluated as AHP10 - AHP1 amplitude) is plotted vs. the month of the year in which T cells were studied. The largest net amount of the AHP amplitude enhancement has been found in May, whereas the smallest has been detected in December.B: the graph shows that the AHP increase detected in May (■) was significantly larger (2-way repeated-measures ANOVA: F = 28.734, α = 0.05,P < 0.001) than the one measured in December (○).

AHP increase changes the strength of the synaptic connection between a T neuron and its followers

At the aim of studying whether the AHP increase affected the synaptic transmission between a T neuron and a postsynaptic cell, we selected another T neuron as the most reliable postsynaptic element (Baylor and Nicholls 1969b) (see alsomethods). The presynaptic T neuron was stimulated with a 200-ms depolarizing pulse (test stimulus), and the EPSP (control) was recorded in the postsynaptic T neuron while its membrane potential was current clamped at −60 mV (Fig.11 A1). Then, a repetitive stimulation (15 trains, 3-s duration, 15-s ISI) was delivered to the presynaptic element, and EPSPs in response to the test stimulus were recorded immediately after the last train and 5, 10, and 15 min after the end of the stimulation. The EPSP evoked right after the end of the stimulation was strongly depressed in the synapses in which the presynaptic element showed AHP increase during repetitive stimulation (Fig. 11, A2 and B). The EPSP recovered to its initial amplitude in ∼15 min (Fig. 11, A3 andB). In Fig. 11 B, we plotted together the graphs of the AHP amplitude increase and of EPSP amplitude depression and the time courses of the respective recoveries. A significant linear correlation was found between the decrease of the EPSP amplitude and the increase of the AHP amplitude (r = 0.954;P = 0.033). Interestingly, the time courses of the two events were very similar: the EPSP recovered from depression while the AHP returned to its initial amplitude. No change of input resistance was detected in either the presynaptic or the postsynaptic T neuron after the training (data not shown). Furthermore, when the presynaptic T neurons did not exhibit AHP increase, the EPSP amplitude after repetitive stimulation (Fig. 11, C2 and D) did not differ from the control (Fig. 11, C1 and D).

Fig. 11.

The AHP amplitude increase depresses the synaptic connection between T neurons. A: raw data from a pair of T neurons. The synapse was probed with 6 action potentials elicited in the presynaptic T neuron (test stimulus, top) and an excitatory postsynaptic potential (EPSP) was recorded from the postsynaptic T neuron (bottom, A1). The EPSP was reduced when tested after the activity-dependent increase of AHP in the presynaptic neuron (A2). The EPSP recovered its initial amplitude after 15 min (A3). B: the graph summarizes the results on 7 pairs of cells in which repetitive stimulation elicited an increase of the AHP amplitude (■) in the presynaptic element and a reduction of the EPSP (○) recorded in the postsynaptic cell.C: the EPSP amplitude recorded in the postsynaptic cell after the training (C2) does not differ from the control (C1), in pairs of cells where repetitive stimulation fails to potentiate the AHP in presynaptic cell. D: the graph summarizes the effect of repetitive stimulation on neurons that exhibited (■, n = 7) or not (▧, n = 7) an increase of the AHP amplitude (the amplitude of AHP1 did not differ significantly between the 2 groups, Mann-Whitney U test,P = 0.569). Note the difference of normalized EPSP after training.


In T sensory neurons of the leech, the action potential firing discharge is followed by an AHP that is mainly due to the increased activity of the Na+/K+ATPase (Baylor and Nicholls 1969a; Jansen and Nicholls 1973). Repetitive intracellular stimulation led to an increase of the AHP amplitude. When 30 trains of depolarizing pulses were delivered at 15-s ISI, the increment of the AHP amplitude reached a plateau after 15 trains, suggesting a mechanism of saturation. In the further experiments, the T neurons were stimulated with only 10 or 15 trains of depolarizing pulses: this training protocol still induced an increase of the AHP amplitude as well as minimized the stress produced in the neuron by massive repetitive stimulation. The AHP enhancement is a lasting phenomenon. This suggests the involvement of an intracellular pathway for the onset and maintenance of the AHP increase. A gradual increase in the AHP amplitude was also induced when T neuron's receptive field was activated with a mechanical stimulus. These results suggest a physiological role of the AHP increase and provide the underlying mechanisms. Namely, this form of activity-dependent modulation of the AHP in T neurons might account for their reduced ability to communicate with postsynaptic followers during repetitive activation.

Increase of AHP amplitude and ISI duration

The activity-dependent increase of the AHP amplitude in T neurons occurs within a specific time window. The ISI has to be ≤20 s for the repetitive stimulation to effectively enhance the AHP. This suggests that progressive increases of the AHP amplitude might be due to a cellular mechanism that is triggered by repetitive stimulation and has to be periodically sustained with proper timing to fully occur.

Activity-dependent changes in AHP are often intrinsic to neurons: for example, Coulter et al. (1989) showed a classical conditioning-specific reduction in the amplitude and duration of calcium-dependent AHP in rabbit hippocampal pyramidal cells. The AHP depression was intrinsic to CA1 pyramidal cells, i.e., it persisted in the absence of synaptic transmission. In leech ganglia, we observed an increase in the amplitude of the AHP during repetitive stimulation that occurred during perfusion with saline containing 20 mM magnesium, namely in the absence of chemical synaptic transmission. This indicates that in the leech nervous system modulation of the AHP is intrinsic to single neurons. The AHP reduction, measured in high-magnesium saline, might be due to a suppressive effect of the Mg2+ onto I K/Ca. However, these data require further investigation. In addition, we found that the firing of action potentials by the T neuron during training is necessary to enhance the AHP amplitude.

Role of the Na+/K+ATPase in the AHP amplitude enhancement

Our results provide evidence for the involvement of the Na+/K+ ATPase in the AHP amplitude enhancement of T neurons of the leech. We analyzed the role of the two components generating AHP in these cells: theI K/Ca and the Na+/K+ ATPase. In the presence of CdCl2, we still obtained an increment of the AHP amplitude during repetitive stimulation. Because CdCl2 inhibitsI K/Ca by blocking Ca2+ channels but also affects the input resistance and produces a shunt of the AHP current (Catarsi and Brunelli 1991; Mozzachiodi et al. 2001), we used a more specific I K/Ca inhibitor, the peptide apamin that has been successfully utilized to blockI K/Ca involved in afterpotentials in several preparations (Hugues et al. 1982; Zhang and Krnjevic 1987), including leech T neurons (Mozzachiodi et al. 2001). Apamin did not prevent the AHP increase occurring during repetitive stimulation. Altogether, these data show that although the I K/Cacontributes to generation of AHP, it does not play a role in the enhancement of the AHP amplitude and suggest the involvement of the Na+/K+ ATPase in the induction of this activity-dependent phenomenon. To confirm this hypothesis, it was necessary to selectively block the Na+/K+ ATPase. We could apply specific blockers of the sodium pump such as ouabain or strophantidin, which are routinely used to inhibit Na+/K+ ATPase activity (Baylor and Nicholls 1969a). Nevertheless, the application of these drugs in leech ganglia (ouabain has a nonreversible action) made it very difficult to keep the T cells firing discharge constant throughout the training, presumably because of an excess of sodium in the T cell that cannot be pumped out (Catarsi and Brunelli 1991). Consequently, the AHP amplitude and its increase could not be reliably measured. Therefore we used 5HT to reversibly inhibit the Na+/K+ ATPase in T sensory neurons. 5HT modulates several membrane conductances in leech neurons. For example it modulates Cl channels (Ali et al. 1998; Lessmann and Dietzel 1995) and K+ channels (Goldermann et al. 1994) in P sensory neurons. In addition, 5HT modulates voltage-activated K+ channels (Acosta-Urquidi et al. 1989) and a Cl conductance (Munsch and Schlue 1993) in the Retzius cells. Nevertheless, Catarsi and Brunelli (1991) demonstrated that 5HT blocks the Na+/K+ ATPase when Na ions were iontophoretically injected in T cells. In addition, they showed that 5HT did not affect either the amplitude or the duration of the action potential of T neurons, indicating that the inhibitory effect on the Na+/K+ ATPase is the only action exerted by 5HT on the T neuron's plasma membrane. A possible effect of 5HT on the Na+/K+ ATPase might be to change the exchanging ratio of Na+ and K+ transport through the membrane, making the pump less electrogenic. Moreover, we cannot so far exclude that 5HT reduces sodium influx as well. In our experiments, treatment with 50 μM 5HT for 10 min prevented the activity-dependent increase of the AHP amplitude when I K/Ca was either active or blocked. Moreover, 5HT strongly shortened the recovery from the AHP potentiation. These data suggest that the Na+/K+ ATPase plays a key role not only in the expression but also in the maintenance of AHP enhancement. In addition, we have found a correlation between basal levels of 5HT in the nervous system and the ability of T neurons to increase AHP during repetitive stimulation. High contents of 5HT in the ganglia seem to act like a cutoff filter that reduced the AHP amplitude enhancement.

Which molecular mechanisms can induce the potentiation of the Na+/K+ ATPase activity? Because the Na+/K+ ATPase is activated by intracellular [Na+], one could explain the augmentation of the activity of the Na+/K+ ATPase with a massive influx of Na+ into T cells during the repetitive stimulation. However, because the AHP amplitude increase is a lasting event, we can suppose that a cytosolic molecular cascade, triggered by the stimulation, produces intracellular messengers capable of positively modulating the Na+/K+ ATPase. This might also explain the presence of a time window in the induction of the AHP increase. One candidate is arachidonate and its metabolites. It has been reported that in the mouse diaphragm, arachidonate have a positive effect on the Na+/K+ ATPase (Vyskocil et al. 1987), and some preliminary results have shown that arachidonic acid metabolites are produced in T neurons during repetitive stimulation (Scuri et al. 1998).

Functional significance of the AHP amplitude enhancement

There is a large body of evidence that learning-related changes in neuronal excitability can be associated with changes of AHP amplitude and duration due to the modulation of ionic currents underlying the AHP (Coulter et al. 1989; Moyer et al. 1996;Saar et al. 1998, 2001). Saar et al. (1998) reported that pyramidal neurons in the olfactory piriform cortex of rats trained to discriminate positive and negative cues in pairs of odors, exhibited reduced spike AHP persisting 3 days after training. More recently, the same authors have shown that the learning-related decrease of AHP is due to a reduction in the acetylcholine-sensitive, Ca2+-dependent K+ current, I AHP(Saar et al. 2001). Recently, Sanchez-Vives et al. (2000) examined, in neurons from slices of ferret primary visual cortex, properties and mechanisms for spike-frequency adaptation and for prolonged AHP generation in response to neuronal activation similar to those observed during typical visual adaptation protocols in vivo. This result shows that the activation ofI K/Ca and Na+-activated K+ currents generate the AHP and seem to be essential for contrast adaptation in vivo.

In hippocampal slices from aging rats, it has been recently observed that L-type Ca2+channels can impair LTP induction during high level of synaptic activation via an increase in the Ca2+-dependent AHP (Norris et al. 1998). Moreover, CA1 neurons in aging rabbits show increased AHPs and a decrement of excitability, suggesting that this age-related change may underlie the learning deficits in aging rabbits (Disterhoft et al. 1996).

The activity-dependent increase of AHP might be a putative mechanism underlying habituation, a simple form of nonassociative learning. It has been previously shown that intracellular stimulation of T neurons induces swim habituation in leech semi-intact preparations (Brodfuehrer et al. 1995).

Because no difference in the number of spikes evoked in T cells by current injection in swimming versus nonswimming trials was detected, the inability of T cells stimulation to continue to evoke swimming was not due to a failure of T cells to follow an intracellular current pulse (Brodfuehrer et al. 1995). This suggests that during habituation, the switch from a swimming to a nonswimming response likely results from a reduced ability of T neurons to drive their postsynaptic cells. Here, we report that the repetitive stimulation that induced an increase in AHP also produced a sustained depression of the synaptic connection between T cells; the synapse remained depressed as long as the AHP was enhanced. The synaptic depression could be due to a reduction of the neurotransmitter released from the presynaptic terminals brought about by repetitive stimulation. This is not the case in our experiments because repetitive stimulation unable to induce AHP amplitude increase did not cause reduction in EPSP amplitude. The correlation between changes in AHP amplitude and synaptic depression suggests that the AHP amplitude increase can act as a presynaptic mechanism producing a decrease in the synaptic efficacy, although a direct evidence of the effect of AHP increase in the modulation of the synaptic strength has not been provided yet.

Although previous results indicate that AHP is an active process that can be generated by action potentials traveling along the T cell's neurite upstream of the cell body (Van Essen 1973;Yau 1976), there is no direct evidence so far about how an altered somatic AHP can affect synaptic transmission in our experimental design. However, we can speculate about the mechanism(s) through which AHP enhancement can modify synaptic efficacy. The morphology of T cells shows an extensive arborization of the neurite (Yau 1976). Some bifurcation points with particularly low safety margins for conduction can operate as low-pass filters limiting the frequency of impulses capable of invading following branches (Yau 1976). There are theoretical and experimental lines of evidence that the propagation of action potentials along axonal branches can be delayed or even fail to reach the terminal because of the geometric structure of the arborization (Goldstein and Rall 1974; Luescher and Shiner 1990a,b; Manor et al. 1991). The geometric structure of T cell can act as a switch for the signal transmission to different terminals (Gu 1991), and changes of AHP amplitude can dynamically modulate the transmission at the branch points (Baccus et al. 2000; Krauthamer 1990). The block of conduction along the neuritic tree produced by the AHP has been reported in T neurons by Van Essen (1973) and Yau (1976). In addition, in T neurons, the conduction block is modulated by 5HT (Mar and Drapeau 1996). In our protocol, the synaptic strength was tested by probing the synapse with a constant number of action potentials evoked in the cell body. However, assuming that the mechanisms observed in the cell body are also present in axonal branch points, some action potentials might not invade the synaptic terminals (i.e., conduction block) after training-induced increases in AHP so affecting the release of neurotransmitter.

In summary, our data suggest that repetitive firing in the soma generates and sustains an activity-dependent increase of the AHP amplitude. This AHP increase is a symptom of the underlying Na+/K+ ATPase activity increase and might decrease the safety factor at axonal branch points, thus reducing the total amount of current that can successfully activate the synaptic terminal. Although this hypothesis is reasonable and an evident correlation between AHP increase and synaptic depression exists, further investigations are required to establish the actual contribution of AHP potentiation to the modulation of the synaptic strength at the terminals.


We are grateful to Dr. E. Cataldo for valuable support for the mathematical analyses of data. We gratefully acknowledge Dr. Gregg A. Phares for valuable discussion and Drs. Paola Lombardo and Leonardo Lami for support in surgical procedures.

This study was supported by the University Grant Research Support 60%.

Present address of R. Mozzachiodi: Dept. of Neurobiology and Anatomy, University of Texas, School of Medicine, 6431 Fannin, Houston, TX 77030.


  • Address for reprint requests: M. Brunelli, Dept, of Physiology and Biochemistry “G. Moruzzi,” University of Pisa, Via S. Zeno, 31, 56127 Pisa, Italy (E-mail: marbru{at}


To analyze the existence of a linear correlation between the decrease of the EPSP amplitude and the increase of the AHP amplitude, we considered the 5 data points showed in Table1.

View this table:
Table 1.

Correlation data points

We estimated a value of −0.954 for the linear correlation coefficientr. A common test to analyze r consists in evaluating the probability to obtain the r value from a probability distribution for a parent population completely uncorrelated (correlation coefficient ρ = 0). In this way, we calculated the probability to have ‖r‖ ≥ 0.954 from a completely uncorrelated set of data. The smaller the probability was, the more likely our data were correlated. This probability was given by the following equationP[r,n]=r12/π(Γ[(ν+1)/2]/Γ[ν/2])(1ρ2)(ν2)/2dρ In the equation, Γ(n +1) = n!represents the integral arguments in the case of n = 0, 1, 2, … and Γ(n +1) =n(n − 1)(n−2) … (3/2)(1/2) π represents the half-integral arguments in the case of n = 1/2, 3/2, 5/2.(Bevington 1970). The number of degrees of freedom (ν) was expressed as the number of data points (n) minus 2. The calculated P value was 0.033. By considering the level of confidence of 95%, the AHP increase and the depressed EPSP can be considered strongly correlated.


View Abstract