Neuronal Calcium Sensor 1 and Activity-Dependent Facilitation of P/Q-Type Calcium Currents at Presynaptic Nerve Terminals

See allHide authors and affiliations

Science  22 Mar 2002:
Vol. 295, Issue 5563, pp. 2276-2279
DOI: 10.1126/science.1068278


P/Q-type presynaptic calcium currents (I pCa) undergo activity-dependent facilitation during repetitive activation at the calyx of the Held synapse. We investigated whether neuronal calcium sensor 1 (NCS-1) may underlie this phenomenon. Direct loading of NCS-1 into the nerve terminal mimicked activity-dependent I pCafacilitation by accelerating the activation time ofI pCa in a Ca2+-dependent manner. A presynaptically loaded carboxyl-terminal peptide of NCS-1 abolishedI pCa facilitation. These results suggest that residual Ca2+ activates endogenous NCS-1, thereby facilitating I pCa. Because both P/Q-type Ca2+ channels and NCS-1 are widely expressed in mammalian nerve terminals, NCS-1 may contribute to the activity-dependent synaptic facilitation at many synapses.

Neurotransmitter release is triggered by Ca2+ influx through presynaptic voltage-dependent Ca2+ channels (1). Modulation in the presynaptic calcium current (I pCa) results in robust alteration of synaptic efficacy because of their nonlinear relationship (2). At the calyx of Held nerve terminal, repetitive activation of Ca2+ channels increases the amplitudes of I pCa (3–5). The magnitude of I pCa facilitation is dependent on the extracellular Ca2+ concentration and is attenuated by intraterminal loading of Ca2+ chelating agents (4,5). This I pCa facilitation is distinct from the voltage-dependent relief of Ca2+ channels from tonic inhibition by heterotrimeric guanine nucleotide binding (G) proteins (6, 7), because presynaptic loadings of guanine nucleotide analogs have no effect (4). A Ca2+-binding protein may thus be involved in the activity-dependentI pCa facilitation.

Among neuron-specific Ca2+-binding proteins, frequenin was first cloned from Drosophila T(X;Y) V7mutants (8). Later, the frequenin homolog NCS-1 was cloned from a variety of species (9–14). NCS-1 (frequenin) is widely expressed in neuronal somata, dendrites, and nerve terminals (14–18) throughout embryonic and postnatal development (14, 17). Overexpression (19) or intracellular loading of NCS-1 in motoneurons (10) enhances neuromuscular transmission. We investigated whether NCS-1 is involved in the activity-dependent I pCa facilitation at the calyx of Held synapse.

Whole-cell voltage-clamp recordings were made from a calyceal nerve terminal (20), and I pCa was elicited by an action potential waveform command pulse at 0.1 Hz. The half-width and the peak amplitude of a prerecorded action potential were similar to those reported for afferent fiber–stimulated action potentials in 14-day-old rats (21). After a stable epoch ofI pCa, NCS-1 was infused into a nerve terminal through a perfusion tube (Fig. 1A). After infusion, amplitudes of I pCa gradually increased, reached a maximum in 5 min, and then gradually declined. This decline may be caused by “adaptation” in the mechanism of facilitation by NCS-1, because I pCa elicited at 0.1 Hz does not undergo rundown for more than 20 min (22). The mean magnitude of I pCa facilitation 5 min after the onset of NCS-1 infusion was 113 ± 37% (mean ± SEM, n = 3).

Figure 1

Facilitation of I pCa by NCS-1. (A) After stable I pCa recordings, NCS-1 (200 μM) was injected into the nerve terminal via pipette perfusion during the period indicated by a bar. Because of the large molecular weight of NCS-1 (20 kD) and the distance between the tips of the perfusion tube and the pipette, final concentration in the calyx could not be assessed. Sample traces below the plot are six averagedI pCa measurements before (a) and 5 min after the infusion (b). An action potential waveform used for the command pulse (V com) was obtained from another calyceal terminal by means of an intracellular recording amplifier. The spike was evoked by injecting a brief depolarizing current pulse (0.12 ms). The action potential had an overshoot of 36 mV and half-width of 0.20 ms. (B) The I-V relationships ofI pCa measured at 1 ms after the depolarizing pulse onset, normalized to the current amplitudes at +10 mV. Symbols: ○, control (no NCS-1); •, NCS-1 (10 μM); □, heat-inactivated (H.I.) NCS-1 (10 μM); ▵, NCS-1 plus BAPTA (16 mM). The peak of theI-V relationship in the presence of NCS-1 shows the negative shift compared with the others. Inset: In three calyces under voltage-clamp, I pCa was elicited by a depolarizing command pulse (5 ms) from a holding potential (–80 mV) to –10 mV every 10 s. Vertical dotted lines indicate the pulse onset and 1 ms after the onset. I pCatraces in the presence of NCS-1 (top) or H.I. NCS-1 (middle) are normalized in peak amplitude and superimposed for comparison (bottom). (C) The 10 to 90% rise time of I pCain the presence of NCS-1 (1.02 ± 0.09 ms, n = 5) is significantly faster (*P < 0.01, analysis of variance) than in the control (1.63 ± 0.13 ms, n= 6), H.I. NCS-1 (1.43 ± 0.11 ms, n = 6), and BAPTA + NCS-1 (1.37 ± 0.07 ms, n = 3) conditions. (D) V 1/2 ofI pCa in the presence of NCS-1 (–23.4 ± 1.5 mV, n = 4) was more negative (*P < 0.001) than in the control (–8.7 ± 0.7 mV, n = 5), H.I. NCS-1 (–11.0 ± 1.4 mV, n = 4), and BAPTA + NCS-1 (–9.3 ± 0.8 mV, n = 3) conditions. (E) In the presence of NCS-1, H.I. NCS-1, or no NCS-1, the peak amplitudes of I pCa measured at 5 ms from the command pulse onset are similar (P > 0.5).

We next examined the effect of NCS-1 onI pCa elicited by a 5-ms depolarizing pulse. When NCS-1 was included in the presynaptic pipette solution, the rise time of I pCa was significantly faster than rise times in the presence of heat-inactivated (H.I.) NCS-1 or in the absence of NCS-1 [Fig. 1, B (inset) and C]. The current-voltage (I-V) relationship ofI pCa measured at 1 ms after the onset of the command pulse had a peak at –10 mV in the presence of NCS-1, whereas the peaks were at 0 mV in the presence of H.I. NCS-1 or in the absence of NCS-1 (Fig. 1B). Similarly, in the presence of NCS-1, the half-activation voltage (V 1/2) calculated from the modified Boltzmann equation (20) was significantly more negative than those in the presence of H.I. NCS-1 or in the absence of NCS-1 (Fig. 1D). However, NCS-1 had no effect on the magnitude of plateau Ca2+ currents (Fig. 1E).

NCS-1 has four helix-to-helix Ca2+-binding architectures (EF-hands) and binds three Ca2+ ions (12). To examine whether the facilitatory effect of NCS-1 on I pCa is Ca2+-dependent, we loaded the Ca2+ chelator 1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid (BAPTA) into calyces together with NCS-1. In the presence of BAPTA, NCS-1 no longer shortened the rise time ofI pCa (Fig. 1C), nor did it cause the negative shift in V 1/2 (Fig. 1, B and D).

The effect of NCS-1 on I pCa mimics the activity-dependent facilitation induced by repetitive stimulation (3–5) with respect to (i) the shortened rise time ofI pCa, (ii) the negative shift inV 1/2, and (iii) its Ca2+ dependency. To further examine whether the NCS-1–induced facilitation and the activity-dependent facilitation of I pCa share a common mechanism, we carried out an occlusion experiment. When NCS-1 was included in the presynaptic pipette, the magnitude ofI pCa facilitation induced by a pair of brief depolarizations was significantly smaller than those in the presence of heat-inactivated NCS-1 or in the absence of NCS-1 (Fig. 2); these findings indicate that NCS-1 partially occluded the activity-dependent I pCafacilitation.

Figure 2

Occlusion of activity-dependentI pCa facilitation by NCS-1. (A) In a paired-pulse protocol, I pCa was evoked every 10 s by a pair of 1-ms square-wave command pulses to –10 mV at interpulse intervals of 5, 10, 20, and 50 ms (superimposed). Left, a calyx in the absence of NCS-1 (control); right, another calyx in the presence of NCS-1. Horizontal dotted lines indicate the amplitudes of the first I pCa (*) in the paired-pulse protocol. (B) Summarized data of paired-pulse facilitation at various interpulse intervals in the presence of NCS-1 (n = 5, •), H.I. NCS-1 (n = 7, □), and no NCS-1 (n = 5, ○). Error bars are merged into some data points.

Although mRNAs encoding NCS-1 are expressed in the rodent brainstem (14, 18), it is not known whether NCS-1 is expressed at this terminal. Immunocytochemical examinations using antibody to NCS-1 (20) revealed NCS-1 immunoreactivity both in the medial nucleus of the trapezoid body (MNTB) cell somata and in the calyceal terminal, the latter being colocalized with the immunoreactivity of the presynaptic marker syntaxin (Fig. 3A). The specificity of the NCS-1 immunofluorescence was confirmed by absorption experiments (Fig. 3B). The immunocytochemical resolution distinguishing the pre- and postsynaptic signals was confirmed by a separation in the signal of the postsynaptic cell body marker microtubule-associated protein 2B (MAP2B) from that of syntaxin (Fig. 3C).

Figure 3

NCS-1 immunoreactivity at calyceal nerve terminals. (A) NCS-1 immunofluorescence signal (green, left) and syntaxin immunofluorescence signal (red, center) partially overlapped (yellow, right), indicating that NCS-1 is expressed in both the presynaptic and postsynaptic compartments. (B) Pretreatment of antibody to NCS-1 with exogenous NCS-1. No clear NCS-1 immunofluorescence signal was observed, whereas the syntaxin signal remained unchanged. (C) Immunofluorescence signal for MAP2B (green, left) and syntaxin signal (red, center) showed no overlap when superimposed (right). Scale bar, 10 μm.

A unique feature of NCS-1 is the large conformational shift of its COOH-terminal region, creating a wide hydrophobic crevice for the target recognition (13). Thus, a COOH-terminal fragment peptide (20) might inhibit the interactions between NCS-1 and P/Q-type Ca2+ channels. When the COOH-terminal peptide was loaded into the calyces, the facilitation ofI pCa elicited by a paired-pulse protocol was abolished (P < 0.001), whereas the scrambled peptide (20) had no such effect (Fig. 4A). The COOH-terminal peptide loaded into calyces had no appreciable effect on I pCawith respect to current kinetics or I-Vrelationships (Fig. 4B). The COOH-terminal peptide also blocked theI pCa facilitation induced by a 100-Hz train (Fig. 4C). Because the COOH-terminal peptide lacks the flanking sequence required for efficient Ca2+ binding, it cannot work by reducing the Ca2+ concentration in calyces. Thus, the COOH-terminal fragment of NCS-1 is likely to interfere with the interaction between NCS-1 and P/Q-type Ca2+ channels, thereby blocking the activity-dependent I pCafacilitation.

Figure 4

COOH-terminal fragment peptide of NCS-1 inhibits the activity-dependent I pCafacilitation. (A) The facilitation ratio at various intervals in the presence of the COOH-terminal peptide (50 μM,n = 6, •, sample traces in inset), scrambled peptide (50 μM, n = 3, □), or no peptide (n= 5, ○). (B) The I-V relationships of I pCa in the presence of the COOH-terminal peptide (V 1/2 = −6.5 ± 0.7 mV, n = 3, •, P = 0.08, sample trace in inset) or no peptide (V 1/2= −8.7 ± 0.7 mV, n = 5, ○). In the presence of scrambled peptide, V 1/2 was similar (−6.8 mV, n = 2, not shown). (C) Mean magnitude of I pCa facilitation during a train of short depolarizing pulses (to −10 mV, 1 ms, 10 ms interval). COOH-terminal peptide abolished theI pCa facilitation induced by a tetanic stimulation (100 Hz) every 20 s. Ten consecutive data points in control (○, n = 4) and in the presence of the COOH-terminal peptide (•, n = 3) are shown.

The Ca2+ channel subtype expressed at the calyceal terminal of rats used in this study (14 to 18 days old) is purely P/Q-type (3, 23). The presynaptic P/Q-type Ca2+currents have a single fast activation phase (<2 ms), which can be shortened by a conditioning depolarization (4) or presynaptic loading of NCS-1 (Fig. 1, B and C). In contrast, currents of recombinant P/Q-type Ca2+ channels expressed in human embryonic kidney–293 cells show a slow activation phase (>10 ms) after a fast activation phase (24). A conditioning depolarization makes this slow activation phase merge into the fast one, and this effect has been attributed to calmodulin (24). At the calyx of Held, however, a calmodulin inhibitory peptide has no effect on I pCa (25). Thus, calmodulin does not likely play a role in the I pCafacilitation, at least at this synapse. NCS-1 mediates the facilitatory effect of glial cell line–derived neurotrophic factor (GDNF) on N-type Ca2+ currents in motoneurons and on transmitter release at the neuromuscular junction (26). This phenomenon is also distinct from the activity-dependent I pCafacilitation, because GDNF increases the peak amplitude of N-type Ca2+ currents without affecting their activation kinetics.

At the calyx of Held nerve terminal, overall Ca2+concentration increases to 0.4 μM in response to a single action potential (27). NCS-1 can bind three Ca2+ ions with an affinity of 10 μM (second EF-hand) and 0.4 μM (third and fourth EF-hands) (12). Thus, residual Ca2+ after an action potential may bind with NCS-1, thereby accelerating the activation time of P/Q-type Ca2+ channels.

In mammals, synaptic transmission is largely mediated by P/Q-type Ca2+ currents both in the peripheral (28) and central (29) nervous system, and their contribution to transmitter release increases with postnatal development (30). Given the wide distribution of NCS-1 in the nerve terminals (14–18), the activity-dependentI pCa facilitation may be mediated by NCS-1 at various synapses, thereby mediating activity-dependent synaptic facilitation. The residual Ca2+ hypothesis for the synaptic facilitation has been widely accepted, but its detailed mechanism is still unknown (31). One of the downstream effects of residual Ca2+ is the facilitation ofI pCa (3–5). Hence, our results suggest that NCS-1 may be a key molecule for the activity-dependent synaptic facilitation.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: tujimoto-tky{at}


Stay Connected to Science

Navigate This Article