Contribution of calcium-dependent facilitation to synaptic plasticity revealed by migraine mutations in the P/Q-type calcium channel

Abstract

The dynamics, computational power, and strength of neural circuits are essential for encoding and processing information in the CNS and rely on short and long forms of synaptic plasticity. In a model system, residual calcium (Ca²⁺) in presynaptic terminals can act through neuronal Ca²⁺ sensor proteins to cause Ca²⁺-dependent facilitation (CDF) of P/Q-type channels and induce short-term synaptic facilitation. However, whether this is a general mechanism of plasticity at intact central synapses and whether mutations associated with human disease affect this process have not been described to our knowledge. In this report, we find that, in both exogenous and native preparations, gain-of-function missense mutations underlying Familial Hemiplegic Migraine type 1 (FHM-1) occlude CDF of P/Q-type Ca²⁺ channels. In FHM-1 mutant mice, the alteration of P/Q-type channel CDF correlates with reduced short-term synaptic facilitation at cerebellar parallel fiber-to-Purkinje cell synapses. Two-photon imaging suggests that P/Q-type channels at parallel fiber terminals in FHM-1 mice are in a basally facilitated state. Overall, the results provide evidence that FHM-1 mutations directly affect both P/Q-type channel CDF and synaptic plasticity and that together likely contribute toward the pathophysiology underlying FHM-1. The findings also suggest that P/Q-type channel CDF is an important mechanism required for normal synaptic plasticity at a fast synapse in the mammalian CNS.

Short-term facilitation of synaptic release has historically been attributed to enhanced vesicle release resulting from the accumulation of intracellular Ca²⁺ in presynaptic terminals during repetitive action potentials (APs), in which the buildup of residual Ca²⁺ enhances binding to sensor proteins that directly mediate vesicle fusion and transmitter release. Short-term depression of synaptic release has been traditionally accredited to vesicle depletion (1). However, at the large calyx of Held synapse, it has been shown that short-term synaptic plasticity is also achieved by a mechanism involving the regulation of presynaptic P/Q-type Ca²⁺ currents (2–4). Further, a recent study using recombinant P/Q-type channels expressed in cultured superior cervical ganglion neurons demonstrated that the Ca²⁺-dependent facilitation (CDF) and Ca²⁺-dependent inactivation (CDI) of P/Q-type channels are mediated through neuronal Ca²⁺ sensor proteins (CaSs) that bind the P/Q-type Ca_v2.1 subunit carboxyl terminus and induce short-term synaptic facilitation and rapid synaptic depression, respectively (5). Whether similar forms of CDF and/or CDI of P/Q-type channels are affected by mutations associated with human disease and whether these modulatory mechanisms represent crucial components of synaptic plasticity at typical intact central synapses have not been described to our knowledge.

In the current study, we explored the effects of Familial Hemiplegic Migraine type 1 (FHM-1) missense mutations on calmodulin (CaM)-mediated CDF and CDI of the P/Q-type channel. FHM-1 is an autosomal dominant form of migraine with aura, typified by hemiparesis and linked to missense mutations in the CACNA1A gene encoding the P/Q-type Ca_v2.1 α₁ subunit (6). Biophysically, FHM-1 missense mutations result in an overall gain-of-function P/Q-type channel phenotype as a result of an underlying shift in channel gating allowing increased Ca²⁺ influx at lower membrane potentials (7, 8). CaM-mediated CDF and CDI are robust forms of P/Q-type channel modulation in which CaM interacts with the Ca_v2.1 carboxyl terminus in a bipartite regulatory processes; CDF is mediated by a local increase in Ca²⁺ and CDI through a global increase in Ca²⁺ (4, 9–17). The underlying mechanisms of these forms of CDF and CDI are also attributed to changes in channel gating (10), and it was of interest to examine whether FHM-1 mutations affect these important modulatory properties of P/Q-type channels and to explore physiological implications by using transgenic models.

We find that FHM-1 gain-of-function missense mutations significantly occlude CDF in recombinant and native systems and correlate with a reduction in short-term synaptic facilitation. Collectively, the data support the notion that selective Ca²⁺-dependent regulation of presynaptic Ca²⁺ channels may underlie several key aspects of short-term plasticity at the parallel fiber-to-Purkinje cell (PF–PC) synapse in cerebellum, and also provide evidence that FHM-1 mutations directly affect the Ca²⁺-dependent regulation of P/Q-type channels (Fig. S1 shows the proposed model).

Results

FHM-1 Mutations Occlude CDF and CDI of Recombinant Human Ca_v2.1 Channels.

Human recombinant Ca_v2.1 channels transiently expressed in HEK cells (along with auxiliary subunits β_2a and α₂δ) is a well characterized system and allows for clear isolation and measurement of CaM-mediated CDF and CDI (12, 13). Consistent with previous findings, WT Ca_v2.1 channels showed typical CDF with prepulse-dependent facilitation when Ca²⁺ was used as the charge carrier (Fig. 1A, black squares). During a test pulse, Ca_v2.1 channels showed a rapid initial activation and then slowly transitioned to a facilitated mode as Ca²⁺ entered (Fig. 1E, gray traces). In contrast, an applied prepulse evoked Ca²⁺ entry increasing the fractional portion of channels in a facilitated mode of gating such that, during an ensuing test pulse, channels exhibited fast activation indicative of the facilitated gating state [Fig. 1E, black traces; facilitated mode is an enhancement of open-channel probability over the basal level in the normal gating mode (10)]. When barium (Ba²⁺) was used as the charge carrier, channels opened directly into the normal gating mode and did not transition to the facilitated state, regardless of the prepulse potential (Fig. 1A, gray symbols). A measure of “pure” CDF was calculated by subtracting relative facilitation (RF) determined using Ca²⁺ as the charge carrier from RF when using Ba²⁺ (Fig. 1A, denoted as g) (9). The FHM-1 mutations, R192Q and S218L, represent distinct ends of the clinical spectrum, with patients possessing the R192Q alteration exhibiting a mild, pure FHM-1 without other neurological symptoms (6, 18), whereas patients with the S218L mutation have an associated severe clinical migraine phenotype most often associated with ataxia or other cerebellar symptoms (19–21). Fig. 1 shows that the R192Q (Fig. 1A, triangles) and S218L (Fig. 1A, diamonds) mutations both occlude Ca_v2.1 CDF across multiple prepulse potentials and pure CDF relative to WT channels (Fig. 1B).

Fig. 1.

R192Q and S218L FHM-1 mutations inhibit Ca²⁺-dependent modulation of human recombinant Ca_v2.1 channels. (A) Using the paired-pulse protocol previously described (9), both the R192Q and S218L FHM-1 mutations are found to occlude CDF across several prepulse potentials shown as relative facilitation versus prepulse potential; results are means ± SEM. (B) Pure CDF (g) following a +20 mV prepulse is significantly reduced by the R192Q (g = 0.097 ± 0.042; *P < 0.05) and S218L (g = 0.023 ± 0.031; *P < 0.05) mutations compared with WT (g = 0.290 ± 0.046). (C) Using 1-s square test-pulses between −10 mV and +30 mV, the current remaining at 800 ms (r800) across several prepulse potentials is increased by both FHM-1 mutations relative to WT; data are means ± SEM. (D) Pure CDI (f) relative to WT (f = 0.418 ± 0.067) is modestly reduced by the R192Q mutation (f = 0.214 ± 0.082) but significantly reduced by the S218L mutation (f = 0.083 ± 0.038; *P < 0.05). (E) Paired-pulse protocol used and representative traces for peak CDF obtained following a +20 mV prepulse for WT, R192Q, and S218L Ca_v2.1 channels using Ca²⁺ (Top) and Ba²⁺ (Bottom) as charge carrier; traces normalized to the end of the test pulse. (F) Representative traces for CDI in which traces are normalized to the peak of a +10 mV test pulse. (G) APW used and representative traces for WT, R192Q, and S218L Ca_v2.1 channels. First hashed line represents the level of Ca²⁺ response during the first AP, and the second line represents the peak (maximum facilitation). The 100-Hz APW was derived from APs recorded in the calyx of Held (38, 54). n refers to the number of cells recorded. All statistics were obtained with use of one-way ANOVA.

The effect of FHM-1 mutations on CDI of exogenous Ca_v2.1 channels was tested using a 1-s test pulse to varying potentials in Ca²⁺ and Ba²⁺. WT Ca_v2.1 channels showed a typical CDI characterized by faster inactivation when Ca²⁺ was used as the charge carrier (Fig. 1C; black squares) relative to Ba²⁺ (Fig. 1C, gray, open squares; Fig. 1F, Left) (13, 17). A measure of pure CDI was determined by subtracting currents obtained using Ca²⁺ as the charge carrier from currents obtained using Ba²⁺ (Fig. 1C, denoted as f) (9). Although the R192Q mutation modestly reduced CDI relative to that of WT (Fig. 1C; triangles; Fig. 1F, Middle), the more severe S218L mutation (Fig. 1C, diamonds) significantly reduced the ability of channels to undergo CDI (Fig. 1D; Fig. 1F, Right). Importantly, because of its reliance on global Ca²⁺ levels (unlike that for CDF), CDI depends on Ca_v2.1 current density (22). Whereas the R192Q mutation has little effect on current density, the S218L mutation causes a large reduction in current density (23) (Fig. S2 shows similar findings). As such, at the mechanistic level, the reduced ability of mutant channels to undergo CDI is likely, at least in part, a result of reduced current density.

Although rectangular depolarizations allow for optimal biophysical resolution of CDF and CDI, it was also important to test the effects of FHM-1 mutations under conditions more resembling neuronal functioning such as AP waveforms (APWs) (13). In response to application of 100 Hz APW, WT channels displayed typical CDF and CDI that shaped the Ca²⁺ currents; CDF caused an initial facilitation of Ca²⁺ currents in the first few APs, and CDI caused a cumulative reduction as the APs continued to be applied (Fig. 1G) (9, 10). The FHM-1 R192Q and S218L mutations both strongly suppressed the dynamics of the response to APWs, consistent with their effects on CDF and CDI (Fig. 1G; Fig. S2 shows average Ca²⁺ responses and those using Ba²⁺ as the charge carrier).

As the biophysical effects of FHM-1 mutations can be affected by a number of factors, including Ca_v2.1 splice-variation (24), β subunit coexpression (25), and the nature of the expression system (7, 26), it was important to test the FHM-1 mutations in the context of native P/Q-type current modulation. In this regard, we measured P/Q-type currents in acutely dissociated cerebellar PCs from WT and R192Q and S218L knock-in mice (27–29).

FHM-1 Mutations Occlude CDF of Native P/Q-Type Currents in Cerebellar PCs.

P/Q-type currents in acutely dissociated cerebellar PCs (approximately 90–95% of whole cell Ca²⁺ current; see refs 30, 31 and Fig. 2D) recapitulate the key features of CaM-mediated CDF observed in human recombinant Ca_v2.1 channels (17, 32, 33). In WT mice, P/Q-type currents in PCs showed similar CDF to that for WT human recombinant Ca_v2.1 channels (Fig. 2A). P/Q-type currents from dissociated PCs from homozygous R192Q and S218L knock-in mice had CDF reduced across both multiple prepulse potentials (Fig. 2A) and during APWs (Fig. 2E); however, only the more severe S218L mutation resulted in a statistically significant reduction in pure CDF (Fig. 2B). Fig. 2C shows that we did not detect significant CDI of endogenous P/Q-type currents in PCs from WT or R192Q and S218L mice. Of note, CDI of P/Q-type currents in dissociated PCs has been found to be variable under different recording conditions (17, 32, 33).

Fig. 2.

P/Q-type current CDF is altered in acutely dissociated PCs from FHM-1 R192Q and S218L knock-in mice. (A) Both the R192Q and S218L mutations occlude CDF across several prepulse potentials; results are means ± SEM (representative traces, far right). (B) Pure CDF (g) relative to WT (g = 0.210 ± 0.028) is not significantly reduced by R192Q (g = 0.136 ± 0.03), whereas the S218L mutation results in a significant reduction (g = 0.0686 ± 0.035; *P < 0.05). (C) There was no appreciable CDI of P/Q-type currents in PCs from WT (f = 0.058 ± 0.045) or R192Q (f = 0.074 ± 0.091) and S218L (f = 0.023 ± 0.028) mice. (D) P/Q-type currents were isolated from freshly dissociated cerebellar PCs identified by their characteristic large size and tear-shaped morphology. Exemplar trace below shows that the pharmacologically isolated P/Q-type currents were completely blocked with 0.2 μM ω-Aga-IVA. (E) APW used and representative traces for WT, R192Q, and S218L Ca_v2.1 channels. First hashed line represents the level of Ca²⁺ response during the first AP, and the second line represents the peak (maximum facilitation). n refers to the number of cells recorded. All statistics were obtained with use of a one-way ANOVA.

Taken together, the findings from the recombinant and endogenous P/Q-type channels support the notion that the FHM-1 R192Q and S218L mutations occlude CDF of P/Q-type channels. The effects on CDF suggest that FHM-1 mutations profoundly affect Ca²⁺-dependent regulation of P/Q-type channels and predicted the possibility of altering P/Q-type channel-dependent functions as they relate to synaptic signaling. Neither a contribution of CDF toward synaptic plasticity in the cerebellum or the demonstration whether FHM-1 mutations might affect fast synaptic cerebellar signaling has been reported. The PF–PC synapse is a well characterized (34), prototypical central synapse that relies predominantly on P/Q-type channels for neurotransmitter release and displays robust presynaptic forms of short-term facilitation via both paired-pulse and AP trains (35, 36). Although the specific modulators of facilitation at the presynaptic terminals of PFs have not been defined, we hypothesized that CDF of P/Q-type channels is an important means of short-term synaptic facilitation and further, that FHM-1 mutations affecting CDF would have a corresponding effect toward synaptic plasticity.

FHM-1 Mutations Attenuate Short-Term Synaptic Facilitation at the PF–PC Synapse.

Synaptic transmission at the PF–PC synapse was measured by using extracellular field recordings in transverse slices from WT and homozygous R192Q and S218L knock-in mice. A typical facilitation response was evoked by paired, 20-Hz, 180-μs stimulations of PFs, in which the second excitatory postsynaptic potential (EPSP) was facilitated relative to the first [i.e., paired-pulse facilitation (PPF) (36); Fig. 3A, Left]. Of note, the paired pulse ratio (PPR; size of the second pulse relative to the first) was significantly reduced by the R192Q and S218L mutations (Fig. 3A, bar graph), and in a manner quantitatively consistent with the reduction in Ca_v2.1 CDF in exogenous and native systems (compare with Figs. 1B and 2B). Following five 180-μs stimulations at 20 Hz, similar results were observed in that both the R192Q and S218L mutations significantly decreased the successive EPSPs relative to the first pulse (Fig. 3B). These results support the notion that CDF of Ca_v2.1 channels is a contributing and necessary component of synaptic plasticity in presynaptic terminals of PFs. However, an unresolved issue is whether FHM-1 mutant presynaptic channels are reluctant and/or unable to transition to a facilitated state resulting in decreased transmitter release, or whether they are in a constitutively facilitated state that basally increases presynaptic Ca²⁺ influx and transmitter release. Several lines of evidence led us to predict that the observed changes in synaptic plasticity at the PF–PC synapse results from a larger initial Ca²⁺ influx through basally facilitated mutant channels relative to unfacilitated WT channels in PF boutons (Fig. S3 and SI Discussion).

Fig. 3.

FHM-1 mutant mice exhibit attenuated PPF at the PF–PC synapse. (A, Left) Exemplar field recording of synaptic responses from the PCs evoked by two extracellular stimuli of approximately 15 V, 180 μs delivered at a 50-ms interval to PFs in the molecular layer (WT mouse). The PPR is a quantification of facilitation obtained by dividing response 2 (R2) by response 1 (R1). Right: PPR is significantly reduced relative to WT (2.36 ± 0.096) by both the R192Q (1.86 ± 0.13; *P < 0.05) and S218L mutations (1.60 ± 0.075; *P < 0.05); results are means ± SEM. (B) Left: Exemplar field recording of synaptic responses from the PCs evoked by five extracellular stimuli of approximately 15 V, 180 μs delivered at 20 Hz (WT mouse). (Right) RF was measured by dividing the peak response from each stimulus by the response obtained from the first stimulus, plotted versus pulse number; results are means ± SEM. Both mutations reduce RF during five pulses at 20 Hz. n refers to the number of slices recorded (from eight WT, four R192Q, and five S218L mice). All statistics were obtained with use of a one-way ANOVA.

FHM-1 Mutant P/Q-Type Channels Appear to Be in a Basally Facilitated State in PF Boutons.

The geometry and spatial equilibration of Ca²⁺ in the PF boutons is ideal for measuring the role of Ca²⁺ dynamics on a tens-of-milliseconds time scale by using Ca²⁺-sensitive fluorescent dyes (36, 37). To this end, we evoked a train of 5 PF responses (50-μs pulses at 20 Hz) in slices from WT or homozygous S218L mice and simultaneously monitored the fluorescent response of the Ca²⁺ indicator Rhod-2 in presynaptic boutons using two-photon microscopy in line-scan mode. We chose to examine the S218L mice because of the consistently larger effect on CDF. Fig. 4A shows an example of unstimulated (top) and stimulated (bottom) PFs in which the presynaptic boutons were detected as relatively bright regions. Ca²⁺ influx in boutons from S218L mice (Fig. 4B, gray diamonds) was enhanced during each of five PF responses compared with WT mice (Fig. 4B, black squares), suggesting channels containing the S218L mutation are in a basally facilitated state.

Fig. 4.

The kinetics of Ca²⁺ influx at PF terminals from S218L mice suggests a basal facilitation of P/Q-type channels. (A) PFs were stimulated with a silver wire electrode and two-photon imaging was used to measure Ca²⁺ transients in presynaptic terminals of the PFs; postsynaptic activity was blocked with 100 μM MCPG, 20 μM CNQX, and 50 μM APV. Line scans were used to measure individual terminals (representative terminal indicated by hashed line). Four sets of five 50-μs stimuli at 20 Hz were given and the Ca²⁺ transients were averaged for each terminal. (B) The average Ca²⁺ influx at PF terminals (normalized to the background fluorescence) is enhanced in S218L mice relative to WT littermates; results are means ± SEM (arrows represent the five stimuli). A representative average Ca²⁺ transient response measured for one terminal in line-scan mode is shown (Lower). (C) An expanded scale of B to show Ca²⁺ transient elicited during the first AP. Ca²⁺ transients in mutant terminals are larger in magnitude than WT terminals (peak enhanced approximately 15% relative to WT), as expected if mutant channels are in a basally facilitated gating mode. Stimulus is shown by AP illustration. n refers to the number of presynaptic terminals recorded (from five WT and five S218L mice). (D) Expanded view of CDF of recombinant channels during APW from Fig. 1G. Ca²⁺ currents through WT recombinant Ca_v2.1 channels increases as Ca²⁺ influx causes CDF (Left), whereas recombinant Ca_v2.1 channels containing the S218L mutation have a maximal Ca²⁺ response regardless of APs (Right). Bottom: Ca²⁺ currents through facilitated P/Q-type channels (gray trace) are larger than through unfacilitated channels (black trace) during evoked APs. A 10–25% increase in Ca²⁺ current amplitude was the range obtained from recombinant Ca_v2.1 channels in HEK cells (Fig. 1G) and endogenous P/Q-type currents in PCs (Fig. 2E). The unfacilitated trace is the Ca²⁺ response of a WT channel obtained during the first AP, and the facilitated trace is the Ca²⁺ response obtained during the 10th AP (indicated by arrows, Top, Left).

Assuming presynaptic Ca²⁺ currents through Ca_v2.1 channels can be modeled by Hodgkin–Huxley equations, a shift to more negative activation voltages by the FHM-1 mutations (24) theoretically could generate larger Ca²⁺ currents during APs (38); however, it was recently shown that, during short 1-ms APs elicited at the calyx of Held in R192Q mice, a hyperpolarizing shift in the activation voltage by the R192Q mutation was insufficient to alter presynaptic Ca²⁺ influx. Only when the AP duration was prolonged by 1 to 2 ms was the shift in the activation by the R192Q mutation sufficient to cause greater presynaptic Ca²⁺ influx (39). Importantly, the AP duration at PF terminals closely resembles the normal, 1-ms APs at the calyx of Held (40, 41), and therefore a shift to more negative activation voltages observed under square pulse depolarizations does not likely translate to an increase in Ca²⁺ influx by the S218L mutation during the normal, short APs at PF terminals. In contrast, Ca_v2.1 channels in the facilitated state have an enhanced Ca²⁺ influx during APs with durations of 1 ms or less (Figs. 1 and 2) (9, 10, 17, 42), and as such, channels rendered toward a facilitated state by the S218L mutation would be expected to show enhanced Ca²⁺ influx at PF terminals during evoked short APs. In further support, the amplitude of Ca²⁺ transients in the presynaptic boutons in S218L mice were enhanced by approximately 15% (Fig. 4C, pulse 1) which is consistent with the 10% to 25% enhancement of Ca²⁺ currents through facilitated P/Q-type channels relative to unfacilitated channels in recombinant and endogenous systems (Fig. 4D). The results appear to represent a true gain in P/Q-type channel function, as there was no apparent compensation by other Ca²⁺ channel subtypes at these boutons in the S218L mice (Fig. S4).

Discussion

It has been known for some time that there exist at least two mechanisms of facilitation at the PF–PC synapse: the well understood mechanism described by the residual Ca²⁺ hypothesis, and an incompletely resolved mechanism driven by a Ca²⁺ sensor with high Ca²⁺ affinity that can detect modest, transient levels of Ca²⁺, likely near the pore of presynaptic Ca²⁺ channels (37). The exact mechanisms and molecular players involved in the latter form of facilitation have not been reported, although a role for high-affinity CaSs such as CaM have been predicted (37, 43, 44). In this report we provide supporting evidence for the hypothesis that Ca²⁺ influx at PF boutons induces CDF of Ca_v2.1 channels as a means to enhanced Ca²⁺ influx during subsequent APs and achieve synaptic facilitation at this central synapse.

Our findings may also provide insight toward the molecular mechanisms of FHM-1 pathophysiology. We find that FHM-1 mutations likely render Ca_v2.1 channels in a basally facilitated state and that the facilitated channels result in a larger Ca²⁺ influx during APs in PF boutons relative to WT channels (Fig. 4). Of note, in previous experiments using single-channel recordings, some FHM-1 missense mutations (including R192Q and S218L) were shown to cause an overall gain-of-function Ca_v2.1 channel phenotype as a result of enhanced open-channel probability (7, 8). Likewise, the mechanism that underlies CDF of Ca_v2.1 channels has been determined to be a Ca²⁺/CaM-mediated transition of channels to a functional state with an enhanced open-channel probability and a facilitated mode of gating (10). As such, we predict that the basally facilitated state induced by the R192Q and S218L mutations reflects a state of enhanced open-channel probability that is the same as (or analogous to) the facilitated state resulting from CDF and that precludes any further CDF-mediated facilitation. In addition to affecting synaptic efficacy, we predict that this gain-of-function state represents an additional mechanism in migraine pathophysiology. To date there has not been an adequate explanation for the cerebellar ataxia associated with the FHM-1 phenotype. We hypothesize that disruption of Ca_v2.1 CDF and synaptic efficacy at the PF–PC synapse may explain the cerebellar ataxia often associated with FHM-1. PCs have regular intrinsic pacemaking that is shaped by various inhibitory and excitatory inputs onto PCs, and fine tuning (increase or decrease) of the interspike interval from that of intrinsic pacemaking conveys information to the deep cerebellar nuclei relevant to motor coordination (45). The significant increase in presynaptic Ca²⁺ influx and presumably enhancement in glutamate release from PF induced by the S218L FHM-1 mutation (Fig. 4) may alter the precise balance of excitatory and inhibitory inputs onto PC cells under certain conditions. As such, the fine tuning of PC firing may be compromised sufficiently to lead to cerebellar motor deficits including ataxia. In support, other mutations in the Ca_v2.1 subunit that increase, decrease, and/or alter the dynamics of excitatory inputs to PC have been correlated with aberrant motor phenotypes in other models involving episodic ataxia (46, 47).

The data presented in this report provide strong evidence that Ca_v2.1 CDF plays a critical role in synaptic plasticity at the PF-PC synapse. Considering that Ca_v2.1 channels are the predominant Ca²⁺ channel underlying synaptic transmission at most other fast synapses in the mammalian CNS (48), we predict Ca_v2.1 CDF is an unrecognized but significant contributor to short-term synaptic facilitation at other fast CNS synapses.

Materials and Methods

SI Materials and Methods provides details of solutions and procedures. In brief, HEK 293 cells were transiently transfected with WT, R192Q, or S218L human Ca_v2.1 in combination with β2, α₂δ, and CD8 as previously described (24). Cerebellar PCs from WT or homozygous R192Q and S218L mice between postnatal days 15 to 25 were enzymatically isolated using similar dissociation techniques previously described (49–51). In both cases, macroscopic Ba²⁺ and Ca²⁺ currents were recorded using the whole-cell patch-clamp technique (52). CDF was analyzed using a two-pulse protocol in which 50-ms test pulses to +5 mV (−10 mV in PCs) were given subsequent to no prepulse or prepulses ranging from −90 mV to +100 mV (or −50 to +40 mV in PCs), or during APWs (9). CDI was measured using a square pulse protocol including a 1-s test depolarization (500 ms in PCs) ranging from −10 mV to +30 mV, in 10-mV increments, from a holding of −90 mV (−60 mV holding in PCs). For extracellular field recording and two-photon imaging experiments, cerebellar vermi were cut into 300-μm transverse slices and maintained for 1 to 6 h at room temperature in artificial cerebral spinal fluid (aCSF). For two-photon imaging experiments, neurons in cerebellar slices were bulk-loaded with Rhod-2 AM using a modified Cremophor-loading technique adapted from Trevelyan et al. (53). PFs were stimulated with a silver wire electrode and presynaptic imaging performed with a two-photon laser scanning microscope directly coupled to a 10 W Chameleon ultrafast laser (Coherent).