Ca²⁺-Dependent Enhancement of Release by Subthreshold Somatic Depolarization

Summary

In many neurons, subthreshold somatic depolarization can spread electrotonically into the axon and modulate subsequent spike-evoked transmission. Although release probability is regulated by intracellular Ca²⁺, the Ca²⁺-dependence of this modulatory mechanism has been debated. Using paired recordings from synaptically connected molecular-layer interneurons (MLIs) of the rat cerebellum, we observed Ca²⁺-mediated strengthening of release following brief subthreshold depolarization of the soma. Two-photon microscopy revealed that, at the axon, somatic depolarization evoked Ca²⁺ influx through voltage-sensitive Ca²⁺ channels (VSCCs) and facilitated spike-evoked Ca²⁺ entry. Exogenous Ca²⁺ buffering diminished these Ca²⁺ transients and eliminated the strengthening of release. Axonal Ca²⁺ entry elicited by subthreshold somatic depolarization also triggered asynchronous transmission that may deplete vesicle availability and thereby temper release strengthening. In this cerebellar circuit, activity-dependent presynaptic plasticity depends on Ca²⁺ elevations resulting from both sub- and suprathreshold electrical activity initiated at the soma.

Short-term alteration in the strength of neurotransmission can result from the electrotonic spread of subthreshold depolarization from the soma to presynaptic sites of release on the axon^1–3. Mechanisms that trigger or modulate neurotransmission are often mediated through control of VSCCs because the likelihood of vesicle fusion is largely determined by the intracellular [Ca²⁺]. Although high concentration Ca²⁺ elevations are required to rapidly trigger exocytosis^4,5, low concentration elevations can augment subsequent release via vesicle recruitment, priming, and sensitization⁶. Because of their large amplitude, action potentials are efficient triggers for VSCC activation^7–10 and release. However, direct recordings from calyceal nerve terminals show that even slight depolarization can open VSCCs, albeit at low probability, and strengthen subsequent spike-evoked neurotransmission^11,12. In contrast, in dentate granule cells of the hippocampus, short-term strengthening of release elicited by modest somatic depolarization may be Ca²⁺-independent^1,13. Whether somatic depolarization enhances action potential-evoked release by activating axonal VSCCs in inhibitory neurons has not been addressed, although asynchronous release is augmented in a Ca²⁺-dependent manner in young animals¹⁴.

MLIs of the cerebellum, including stellate and basket cells, make inhibitory GABAergic synapses onto Purkinje cells as well as other MLIs¹⁵. Depolarizing potentials originating in the somatodendritic compartment of MLIs passively propagate into their axonal arbor^16,17 and open VSCCs¹⁸. However, action potentials are required to rapidly coordinate release from presynaptic sites, thereby limiting the voltage range for VSCC activation to subthreshold depolarization. Whether axonal Ca²⁺ entry evoked by subthreshold somatic depolarization is sufficient to alter spike-evoked neurotransmission has not been directly determined in MLIs.

Here, we investigated the mechanisms that mediate somatic voltage control of axonal transmitter release between MLIs using paired electrical recording and two-photon laser scanning microscopy (2PLSM). Subthreshold somatic depolarization was sufficient to activate axonal VSCCs, elicit Ca²⁺ influx and strengthen both action potential-evoked and asynchronous transmitter release. Enhancement of release was diminished or eliminated by chelating intracellular Ca²⁺ with EGTA or by blocking VSCCs, indicating that there is a direct connection between somatic voltage-control of neurotransmission and Ca²⁺ entry at the site of release. This suggests that release plasticity elicited by subthreshold somatic depolarization depends on presynaptic VSCCs.

Results

Subthreshold depolarization enhances AP-evoked release

GABA_A receptor (GABA_AR)-mediated synaptic transmission was examined between pairs of MLIs in rat cerebellar slices using simultaneous whole-cell recording. Action potentials elicited in presynaptic cells by somatic current injection (100 – 600 pA, 3 – 5 ms) evoked time-locked inward currents in voltage-clamped postsynaptic cells (E_Cl- ≈ 0 mV) that were blocked by picrotoxin (100μM; 3.8 ± 1.8 % of control; n = 5). Irreversible rundown of evoked transmission, commonly observed in paired MLI recordings¹⁹, was eliminated by presynaptic perforated patch recording²⁰ (see Methods). To determine whether somatic depolarization was sufficient to alter the strength of neurotransmission, presynaptic neurons were briefly depolarized from rest (−73.3 ± 0.3 mV; n = 40) to a potential just subthreshold for spiking (−56.1 ± 0.9 mV, 300 ms). This depolarization was followed by action potential stimulation (10 ms delay). IPSCs evoked by action potentials preceded by subthreshold depolarization were larger than interleaved control IPSCs evoked by action potentials alone (121.1 ± 4.8% of control amplitude; n = 40, P < 0.001; Fig. 1a). Furthermore, the paired-pulse ratio (PPR = IPSC₂/IPSC₁) of IPSCs evoked by two action potentials (250 ms interval; PPR = 1.000 ± 0.064, n = 37; Fig. 1a, 1b) following subthreshold depolarization was significantly reduced compared to non-depolarized control trials (P = 0.002; Fig. 1b, 1c). This indicates that subthreshold somatic depolarization causes a short-term alteration in presynaptic release resulting from the electrotonic spread of the somatic depolarization into the axon, as observed in other neurons^1–3,13.

Figure 1. Subthreshold somatic depolarization enhances spike-evoked release in MLIs.

(a) In a MLI paired recording, action potentials in the presynaptic neuron evoked GABA_AR-mediated IPSCs in the postsynaptic neuron. Subthreshold somatic depolarization (300 ms) elicited prior to spiking enhanced release. Control action potential responses are in red and test responses preceded by subthreshold depolarization are in black. An amplified view of the first IPSC is shown in the inset. (b) Paired-pulse IPSCs from the same neuron normalized to the peak of the first response. (c) Summary data show that the PPR decreases with somatic depolarization of the presynaptic neuron. A significant difference (paired t-test, P < 0.05) between groups is indicated by asterisk with mean values ± SEM. (d) Linear regression analysis indicates that release enhancement elicited by subthreshold depolarization was correlated with the basal PPR. (e and f) Differences in release enhancement for facilitating and depressing responses measured in 1.5 and 3.0 mM extracellular Ca²⁺. ANOVA analysis with post-hoc Tukey’s multiple comparison tests indicated significance (P < 0.05) with mean values plotted ± SEM. For clarity, spontaneous IPSCs were blanked prior to averaging in a (inset) and b.

Control recordings included both facilitating and depressing responses (Fig. 1c), indicating that synapses between MLIs with different release probabilities co-exist in cerebellum. Synaptic strengthening caused by subthreshold depolarization depended on basal release probability, as the amount of enhancement of IPSC₁ was correlated with the PPR measured in control trials (Fig. 1d). Depressing synapses, on average, were not altered by subthreshold depolarization (P = 0.17). However, there was considerable variability within this group, with some pairs showing release enhancement while others were weakened (Fig. 1e). In 3 mM extracellular Ca²⁺ release enhancement caused by subthreshold depolarization was less than in 1.5 mM Ca²⁺ (Fig. 1e), but the increase was still significant (P = 0.03). High extracellular Ca²⁺ increased release probability, as reflected by larger control IPSCs (peak amplitude 306.0 ± 67.0 and 172.7 ± 22.5 pA, 3.0 and 1.5 mM Ca²⁺, respectively, n = 13 and 40, P = 0.02) and paired-pulse depression in all recordings (n = 11, Fig. 1f). Thus, strengthening of release by subthreshold depolarization is similar to other activity-dependent forms of release enhancement, such as spike-evoked facilitation and augmentation, in that these processes occur more readily in low release probability synapses.

Subthreshold depolarization and presynaptic Ca²⁺ entry

Intracellular Ca²⁺ is known to govern many aspects of transmitter release⁶. Therefore, we probed for presynaptic Ca²⁺ elevation evoked by subthreshold somatic depolarization using 2PLSM. MLIs were filled through the whole-cell patch pipette with Alexa Fluor 594 to visualize cell morphology and the Ca²⁺ indicator Fluo-5F to detect intracellular Ca²⁺ transients (Fig. 2a). Axon varicosities were targeted for optical recording. These varicosities are known to be hotspots of action potential-evoked Ca²⁺ entry and, therefore, are presumed to be en passant sites of release^15,18,21,22, though it remains possible that these specializations may include non-synaptic sites.

Figure 2. Subthreshold somatic depolarization evokes and enhances axonal Ca²⁺ entry.

(a) Two-photon fluorescence image of a representative MLI filled via patch pipette with Alexa 594 (13 μM) and Fluo-5F (200 μM). On the right, a higher-magnification view shows axon varicosities. (b) Somatic current injection elicited control action potential-evoked responses (red) and test spikes paired with 300 ms subthreshold depolarizations (black). Shown directly below are the resulting Ca²⁺ transients recorded in an axon varicosity. Integration of the Ca²⁺ signals accentuates the difference between control and depolarized responses prior to spike firing. At the bottom left, amplified view of Ca²⁺ transients with dashed lines (100 ms epoch average) demarcating the amplitude of the Ca²⁺ response evoked by subthreshold depolarization. On the bottom right, action potential-evoked Ca²⁺ transients baselined prior to spike firing (black bar) isolated depolarization-dependent facilitation of the spike-evoked Ca²⁺ transient. Exponential fits to the spike-elicited transients were used to determine peak amplitudes. (c) Comparison of depolarization-evoked Ca²⁺ elevation and facilitation of action potential-elicited Ca²⁺ entry. Line fit of the linear regression is indicated by the dashed line.

In current-clamped MLIs, subthreshold somatic depolarization (300 ms) often evoked small Ca²⁺ transients in axon varicosities (Fig. 2b; ΔSub Ca²⁺). The amplitude of these Ca²⁺ transients varied across recording sites, including many varicosities without any apparent Ca²⁺ elevation (Fig. 2c). However, on average, Ca²⁺ responses were significantly larger than corresponding measurements obtained from interleaved control trials without subthreshold depolarization (ΔG/R −0.0010 ± 0.0004 and 0.0026 ± 0.0010, control and depolarized, respectively; P < 0.0001; n = 35). In addition, we observed that Ca²⁺ transients evoked by action potentials following subthreshold depolarization were larger than those without prior depolarization (Fig. 2b). Subtraction of the Ca²⁺ elevation evoked during the subthreshold depolarization revealed that the facilitation of spike-evoked Ca²⁺ (24.5 ± 4.5% increase in peak amplitude, P < 0.001) was not merely the result of the shift in baseline fluorescence, but rather a true enhancement of spike-induced Ca²⁺ influx (Fig. 2b;ΔAP Ca²⁺). The amount of spike-evoked Ca²⁺ facilitation was not correlated with the amplitude of Ca²⁺ influx during subthreshold depolarization (Fig. 2c). Whether this indicates that the two components contributing to Ca²⁺ elevation are not causally related is unclear because the extremely small amplitude of the subthreshold response interferes with our ability to accurately measure it. However, in 3 mM extracellular Ca²⁺, the Ca²⁺ transient resulting from subthreshold depolarization was larger than that in control conditions (Peak ΔG/R 0.0036 ± 0.0008 and 0.0061 ± 0.0030; 1.5 and 3.0 extracellular Ca²⁺, respectively; n = 35 and 74, P = 0.04) and was correlated with the facilitation of the spike-induced Ca²⁺ influx (r² = 0.19; P = 0.0002), suggesting that the Ca²⁺ elevation during subthreshold depolarization caused the spike-induced facilitation.

Inhibition of VSCCs by a combination of antagonists (ω-conotoxin MVIIC 1μM, SNX 0.3 μM, nimodipine 20 μM, and mibefradil 10μM) confirmed that subthreshold depolarization-evoked Ca²⁺ entry occurs through VSCCs (Supplemental Fig. 1a, 1b) as previously reported¹⁸. The unblocked component presumably reflects toxin-resistant VSCCs^22–24, as spike-evoked Ca²⁺ transients were blocked to a similar extent (Supplemental Fig. 1b). Subthreshold depolarization-induced facilitation of spike-evoked Ca²⁺ entry was also diminished by VSCC inhibition (8.7 ± 5.0% of control; n = 9, P = 0.002), again suggesting that this facilitation was dependent on an increase in intracellular Ca²⁺ elicited by the preceding subthreshold depolarization.

Release strengthening caused by subthreshold depolarization in MNTB calyceal nerve terminals is inhibited by Ca²⁺chelation, implying an alteration in an underlying Ca²⁺ elevation^11,12. In MLI axonal varicosities, depolarization-evoked Ca²⁺ entry was not significantly altered by the cell-permeable Ca²⁺ buffer EGTA-AM (10–20 μM; >10 min; P = 0.08, Fig. 3a, 3b). However, in many cases subthreshold depolarization did not result in a measurable Ca²⁺ transient (see Fig. 2c) and, therefore, an effect of chelation overall could only be determined indirectly, using paired recording (see below). To test for a direct effect of chelation on Ca²⁺ elevation induced by subthreshold depolarization, we focused our analysis on large-amplitude responses (ΔG/R > 0.0050; n = 4; Fig. 3a). In this group, EGTA significantly reduced the Ca²⁺ transients (ΔG/R 0.0090 ± 0.0019 and 0.0043 ± 0.0027, control and EGTA, respectively, n = 4, P = 0.03). Facilitation of spike-evoked Ca²⁺ entry following subthreshold depolarization was greatly diminished by EGTA across all cells (Fig. 3c), indicating that, even in cases where depolarization did not produce a measurable increase in Ca²⁺, facilitation of the spike-evoked Ca²⁺ transient was Ca²⁺-dependent^25,26. EGTA also reduced Ca²⁺ transients evoked by single action potentials alone (70.1 ± 3.3% of control; n = 9, P = 0.008; Fig. 3a), suggesting a competition between Fluo-5F and EGTA for Ca²⁺ binding²⁷, that basal Ca²⁺ levels are sufficient to partially facilitate VSCC opening, or that there was a substantial rundown of VSCC opening.

Figure 3. Subthreshold depolarization-dependent Ca²⁺ signaling is EGTA-sensitive.

(a) Somatic recording of an action potential (red) and an action potential preceded by a 300 ms subthreshold depolarization (black). The resulting Ca²⁺ transients in an axon varicosity are shown below in control and following EGTA-AM application (10–20 μM; >10 min). Integrals of the Ca²⁺ responses are provided for each condition. The inset shows a comparison of action potential-evoked responses, elicited without preceding subthreshold depolarization, in control and with EGTA. Dashed lines indicate the response amplitude prior to spiking (100 ms epoch) and the exponential fits of action potential-evoked Ca²⁺ transients used to determine peak amplitudes. (b) The effect of EGTA on Ca²⁺ entry evoked by subthreshold depolarization. (c) EGTA inhibits facilitation of spike-evoked Ca²⁺ entry caused by subthreshold depolarization. Mean values are presented with error bars (+SEM); asterisks denote significance (paired t-test, P < 0.05).

Enhancement of release is Ca²⁺-dependent

In paired MLI recordings, the enhancement of IPSC₁ by subthreshold depolarization was blocked by EGTA-AM (20 μM; >10 min; 122.2 ± 7.4 and 97.5 ± 6.9% of control amplitude, −EGTA and +EGTA, respectively, n = 9, P = 0.009; Fig. 4a, 4b). This indicates that the increase in spike-evoked release caused by subthreshold depolarization depends on the EGTA-sensitive Ca²⁺ elevation observed in axon varicosities. The block of release-strengthening by EGTA occurred mainly in recordings showing paired-pulse facilitation in control conditions. On average, IPSCs in depressing cell pairs were unaltered by subthreshold depolarization and remained unchanged with the addition of EGTA (Fig. 4b, top; see also Fig. 1e). It follows that the magnitude of the EGTA effect was correlated with the control PPR (Fig. 4b, bottom). This reinforces the conclusion that subthreshold depolarization differentially affects release at MLI synapses in a release probability-dependent manner.

Figure 4. EGTA inhibits enhancement of release elicited by subthreshold depolarization.

(a) Action potential-evoked IPSCs (red) recorded from pairs of connected MLIs in control and following EGTA-AM application. In alternating trials, action potentials were preceded by a 300 ms subthreshold depolarization (black). (b) Top summary graph shows that EGTA eliminates release strengthening caused by subthreshold depolarization in facilitating cell pairs. Release-strengthening was not apparent in cell pairs that depressed in basal conditions. In the bottom graph, linear regression analysis plots the difference between the percentage change in EGTA and in control relative to the basal PPR. (c) The effect of EGTA on paired-pulse responses from trials without preceding subthreshold depolarization. Amplified view of the first IPSC is shown on the inset. (d) EGTA altered paired-pulse responses in cell pairs that facilitated in basal conditions. Data are plotted as mean values with ± SEM and asterisks denote significance (paired t-test, P < 0.05). Spontaneous IPSCs were blanked prior to averaging in a and c.

EGTA did not affect the amplitude of IPSCs evoked without preceding subthreshold depolarization (Fig. 4c; IPSC amplitudes 97.5 ± 4.8 and 82.8 ± 11.3% of control; n = 5 and 4, P = 0.47 and 0.16; facilitating, and depressing responses, respectively). This implies that the Ca²⁺ domain responsible for triggering exocytosis is tightly associated with release-competent vesicles^28,29. However, EGTA altered the PPR, changing facilitation to depression, though it had no clear effect on depressing cell pairs (Fig. 4c, 4d; PPR for depressing cells 0.67 ± 0.09 and 0.97 ± 0.09, control and EGTA, respectively; n = 4, P = 0.13). This suggests that, in basal conditions, residual Ca²⁺ following an initial spike facilitates subsequent release²⁷ and may mask depression^30,31. EGTA did not induce a significant change in the PPR in these responses when subthreshold depolarization preceded spiking (PPR 0.79 ± 0.05 and 1.01 ± 0.11 EGTA and +EGTA, respectively; n = 5, P = 0.18), suggesting little additional or underlying modification of release in this condition.

Subthreshold depolarization and asynchronous release

At some synapses, asynchronous release can result from the low concentration of Ca²⁺ remaining after spike firing^32,33. In our paired recordings, there was often a small increase in the frequency of asynchronous IPSCs during subthreshold somatic depolarization (Fig. 5a) likely caused by the elevation of Ca²⁺ by the depolarizing stimulus. Although in many recordings, spike-evoked IPSCs from unclamped surrounding MLIs likely overwhelmed this small increase in asynchronous transmission, charge integration of averaged postsynaptic currents revealed a difference between control and depolarized trials across cells (Fig. 5b). The increase of asynchronous IPSCs by presynaptic somatic depolarization was much clearer when background action potential-driven release from spontaneously firing surrounding cells was blocked with TTX (1 μM) and the stimulus duration was prolonged (150 s; 3 mM extracellular Ca²⁺; Fig. 6a, 6b, and 6d). Block of VSCCs by Cd²⁺ and Ni²⁺ (100 μM) eliminated depolarization-evoked asynchronous release without affecting basal spontaneous transmission (Fig. 6c, 6e), demonstrating the connection between somatic depolarization, axonal VSCC activation, and asynchronous release.

Figure 5. Subthreshold depolarization evokes asynchronous release.

(a) Paired recording of MLIs show spontaneous IPSCs in unstimulated control responses (red) and recruitment of asynchronous events with a 300 ms presynaptic subthreshold depolarization (black). Three individual sweeps from the postsynaptic neuron are shown superimposed for each condition with the average response (27 and 28 sweeps, control and depolarized, respectively) at the bottom. (b) The average charge integration of the postsynaptic response across all cell recordings (± SEM). Asterisk denotes significance (paired t-test, P < 0.05).

Figure 6. Depolarization-dependent asynchronous release is triggered by VSCCs.

(a,b) Paired recording from MLIs in TTX with representative sweeps from the postsynaptic neuron showing recruitment of asynchronous IPSCs with presynaptic depolarization. The presynaptic holding potential is indicated in gray above each sweep. Responses were recorded in 3 mM extracellular Ca²⁺ to enhance release. (c) In the same cell, IPSC frequency is unaltered by presynaptic depolarization after VSCC block (100 μM Cd²⁺ and Ni²⁺). (d) Asynchronous IPSC frequency depends on the presynaptic holding potential. Mean values are plotted with error bars ± SEM. Asterisks denote significant differences (P < 0.05) between groups tests following ANOVA analysis (Tukey’s multiple comparison test). (e) Block of VSCCs eliminates the depolarization-dependent increase in asynchronous transmission. Mean values + SEM are shown with significance (asterisks P < 0.05) determined by ANOVA analysis.

Asynchronous release elicited by repetitive firing weakens subsequent action potential-evoked phasic transmission by depleting the pool of readily-releasable vesicles^33,34. Vesicle availability may, in part, regulate release strengthening induced by subthreshold depolarization. To explore this idea, we reduced the releasable pool of vesicles with a train of presynaptic action potentials (10 spikes, 60 Hz) and then tested for subthreshold depolarization-induced enhancement of release (Fig. 7a). The amplitude of spike-evoked test IPSCs following burst firing was reduced compared to interleaved trials without the burst (77.3 ± 7.0% of control; n = 10, P = 0.05, Fig. 7b), a result consistent with depression induced by vesicle depletion. In addition, burst firing eliminated release-strengthening elicited by subthreshold depolarization (Fig. 7a, 7c). This indicates that in conditions that diminish the releasable pool of vesicles, subthreshold depolarization is less effective in altering release probability. By extension, asynchronous release elicited by subthreshold depolarization likely limits the strengthening of spike-evoked transmission by depleting the readily-releasable pool of vesicles. This may explain why in some cell pairs (4 of 10) the additive effect of subthreshold depolarization and bursting resulted in a weakening of transmission rather than enhancement (Fig. 7a, 7c).

Figure 7. Burst firing reduces the capacity for release enhancement caused by depolarization.

(a) In a recording from connected MLIs, high-frequency bursts of action potentials (10 spikes, 60 Hz) preceded a 300 ms subthreshold stimulation and test spikes in interleaved trials. Amplified views on the right show release enhancement caused by subthreshold depolarization and weakening of transmission when subthreshold excitation was paired with burst firing. (b) The effect of burst firing on the first IPSC in control trials without subthreshold stimulation is shown. (c) Burst firing eliminated release strengthening induced by subthreshold depolarization. Mean values are illustrated ± SEM with significance denoted by an asterisk (paired t-test, P < 0.05). For amplified views (a and b) spontaneous events were blanked prior to averaging.

Discussion

We report that subthreshold somatic depolarization passively spreads into the axon arbor of MLIs and strengthens subsequent action potential-evoked release of GABA onto neighboring cells. Therefore, subthreshold depolarization enhancement of release is not limited to excitatory neurons, but also includes inhibitory neurons as well. We show that for MLIs, this enhancement of release depends on the opening of VSCCs, which results in Ca²⁺ elevation as well as facilitation of spike-evoked Ca²⁺ influx, a novel finding. An increase in asynchronous release that accompanies the subthreshold depolarization is Ca²⁺-dependent and may decrease the capacity for release strengthening owing to vesicle depletion. Our results suggests the possibility that subthreshold potentials may modify neurotransmission at synapses throughout the nervous system in a Ca²⁺-dependent manner, as do other forms of short-term facilitation.

Somatic depolarization and Ca²⁺ influx at sites of release

Axons are not electrotonically isolated from the somatodendritic compartment, but rather support the electrotonic spread of subthreshold somatodendritic potentials^1–3,16,18. As a result, the open probability of voltage-sensitive ion channels in axons can be modified by somatodendritic depolarization. Our Ca²⁺ measurements indicate that somatic depolarizations electrotonically spread into MLI axons and are sufficient to open VSCCs. Though not directly determined in this report, axonal length constants can exceed 400 μm in some neurons^1–3,35. This parameter must be convolved with the number and composition of VSCCs at sites of release to determine subthreshold depolarization-mediated Ca²⁺ entry³⁶. In calyceal terminals, P/Q-type VSCCs open at low probability following slight depolarization¹¹ (<−60 mV), an activation range below that previously reported^25,37. Although MLI varicosities express P/Q/N-type channels, among others²², the VSCC subtypes mediating depolarization-evoked Ca²⁺ influx remain unclear. It is possible that low-activation threshold VSCCs, such as R-type channels, preferentially mediate this Ca²⁺ influx, as predicted for hippocampal mossy fiber boutons³⁶. It is unknown whether Ca²⁺-induced Ca²⁺ release from intracellular stores³⁸ contributes to subthreshold depolarization-evoked Ca²⁺ transients.

Subthreshold depolarization not only induces direct influx of Ca²⁺ in MLI axons but also facilitates action potential-evoked Ca²⁺ entry. This effect is probably dependent on the preceding influx of Ca²⁺, because both inhibition of VSCCs and Ca²⁺ chelation diminished the facilitation of spike-evoked Ca²⁺ entry. At some synapses, VSCC-mediated currents are enhanced following Ca²⁺ binding by neuronal calcium sensor 1 (NCS-1), which induces shortening of the activation phase of VSCCs^26,39. VSCC facilitation is elicited by both supra- and subthreshold depolarization-evoked Ca²⁺ elevation and is mitigated by exogenous Ca²⁺ buffering^12,25,26. In MLIs, enhancement of spike-evoked Ca²⁺ entry following subthreshold depolarization could reflect a comparable Ca²⁺-dependent facilitation of VSCC activity, as MLIs express NCS-1 at presynaptic sites of release^40,41. It remains possible that facilitation of action potential-evoked Ca²⁺ entry does not depend directly on the Ca²⁺ elevation evoked by the subthreshold depolarization. Alternatively, because spike-evoked Ca²⁺ influx in presynaptic terminals is, in part, controlled by the amplitude and duration of the action potential waveform^9,42, facilitation of spike-evoked Ca²⁺ entry following subthreshold depolarization may reflect an alteration of the underlying action potential. Prolonged subthreshold somatic depolarization inactivates axonal Kv1 potassium channels in L5 pyramidal cells of visual cortex, leading to spike broadening³, which should in turn enhance Ca²⁺ entry⁴². However, potassium channel inactivation is not expected to be EGTA-sensitive or depend on VSCCs, arguing against this possibility.

Somatic depolarization and enhancement of release

Presynaptic receptor-mediated depolarization of calyceal nerve terminals elicits VSCC Ca²⁺ elevation and strengthening of release^11,43. Our results extend this finding; subthreshold excitatory potentials need not only be elicited on or near terminals to enhance release. Depolarizing potentials evoked in the dendrites or soma spread through the axon, opening VSCCs, elevating Ca²⁺ at presynaptic specializations, and thereby increasing the likelihood of transmission. Small Ca²⁺ transients increase the probability of fusion by increasing microdomain Ca^2+44,45, recruiting or sensitizing vesicles through Ca²⁺ binding by a high-affinity sensor distinct from the exocytosis trigger^27,46, and saturating endogenous Ca²⁺ buffers^44,45. If these mechanisms are mediated by residual Ca²⁺ following action potential firing, then they may also be triggered by Ca²⁺ influx during subthreshold depolarization¹². Given that release probability is steeply dependent on the intracellular [Ca²⁺]^4,5, enhancement of spike-evoked Ca²⁺ entry will also strengthen transmission^12,25,26,42 unless the release process is saturated. If this occurs in MLIs, it may explain why high release probability synapses have a diminished capacity for release-strengthening induced by subthreshold depolarization.

We show that axonal VSCC-mediated Ca²⁺ entry elicited during presynaptic depolarization is sufficient to evoke asynchronous release in MLIs. Ca²⁺-induced asynchronous fusion has a relatively high sensitivity to intracellular [Ca²⁺], with an elevation to as little as 1 μM triggering release^4,5. If the Ca²⁺ sensor for asynchronous exocytosis at MLIs has a similar sensitivity, then the intracellular [Ca²⁺] reached at MLI varicosities may be much higher than depolarization-evoked Ca²⁺ elevation in the calyx¹¹ (100 nM). The supra-linear Ca²⁺ cooperativity for exocytosis^4,5,47 might suggest a greater amount of spike-evoked release enhancement following presynaptic depolarization and Ca²⁺ entry than we observed. However, asynchronous transmission can deplete the pool of release-ready vesicles, diminishing availability for phasic transmission^33,34. The resulting short-term depression of release can develop concomitantly with facilitation, one effect masking the other^30,31. In this way, the capacity for depolarization-evoked strengthening of transmission in MLIs is limited by the superimposition of Ca²⁺-dependent depression mediated by vesicle pool depletion. Though IPSC amplitudes were unchanged by subthreshold depolarization in EGTA, it may be that release enhancement was merely balanced by depression rather than fully blocked. EGTA only partially diminished depolarization-induced Ca²⁺ elevation suggesting that the Ca²⁺-dependent effects on release persisted in this condition. The expression of facilitation and depression will depend on the relative Ca²⁺ sensitivity of each mechanism.

As implied in our burst-firing experiment, the capacity for release-strengthening by subthreshold depolarization may be strongly limited by depletion. Non-homogenous pools of vesicles may populate presynaptic sites at MLIs, as they do at the calyx of Held⁴⁸. Subthreshold depolarization may help recruit more reluctant vesicles, distinguished by their relative Ca²⁺ sensitivity and/or distance from VSCCs, to participate in release during subsequent action potential-evoked transmission. Bulk Ca²⁺ elevation evoked by repetitive firing may also recruit these vesicles, depleting their availability and thereby decreasing the capacity for subthreshold depolarization induced enhancement of release. That is, multiplicative non-linear processes are competing for overlapping vesicle pools to determine the expression of short-term plasticity. We expect that any reduction in the vesicle pool(s), mediated by preceding sub- or suprathreshold activity, would diminish enhancement of release induced by subthreshold depolarization.

Functional implications

Subthreshold depolarization-induced regulation of spike-evoked release is clearly important in a number of cell types, though the mechanisms underlying the regulation may vary across brain regions^{1–3,13,18,35}. In MLIs, the phenomenon appears to be entirely dependent on axonal Ca²⁺, as observed for other forms of release-strengthening⁶. In contrast, at hippocampal mossy fiber synapses, it may be that the Ca²⁺-dependence is restricted to proximal portions of the axon, whereas at more distal sites other mechanisms of strengthening are involved¹³, such as potassium channel inactivation^3,42. One clear difference between these two cell types is axonal length, mossy fibers being much longer than MLI axons. Even in MLIs, the increases in axonal Ca²⁺ are, on average, larger in proximal than in distal portions of the axon, as expected from cable properties¹⁸. However, there was no correlation between the depolarization-induced enhancement of release and distance between the somata of synaptically connected pairs (r² = 0.06, P = 0.23, n = 27), though this comparison is likely compromised by the inaccuracy of this measurement in predicting axonal length. The present results suggest that Ca²⁺-dependent plasticity induced by somatic depolarization will affect the gain of most synapses in short axon cells and at least proximal synapses in other cell types. However, it is likely that depolarization-induced plasticity does not simply monotonically decay with distance from the soma given the amount of variance we observed in our Ca²⁺ imaging and paired recording experiments. The control of transmitter release by subthreshold synaptic input that precedes action potential initiation adds a new dimension to synaptic signaling and requires consideration when analyzing information transfer across synapses.

Methods

Slice preparation

Parasagittal slices from cerebellar vermi were prepared from Sprague Dawley rats (PND 14–20) in accordance with OHSU IACUC-approved protocols. Rats, following isoflurane anesthesia, were decapitated and the cerebellum was isolated. Slices (250–300 μm) were cut on a vibroslicer (Leica Instruments, Nussloch, Germany) in an ice-cold solution containing (in mM) 87 NaCl, 25 NaHO₃, 2.5 KCl, 1.25 NaH₂PO₄, 7 MgCl₂, 0.5 CaCl₂, 10 glucose, and 75 sucrose. Slices were subsequently transferred to a holding chamber containing (in mM) 119 NaCl, 26.2 NaHO₃, 2.5 KCl, 1 NaH₂PO₄, 2 MgCl₂, 1 CaCl₂, 11 glucose and maintained at 34°C for 30 min, thereafter, at 22–25°C until use. For whole-cell recording, slices were placed in a submersion chamber and superfused with the same solution (22–25°C) altered by elevating CaCl₂ to 1.5 and reducing MgCl₂ to 1.5 mM, except where noted. All solutions were oxygenated by pre- or continuous equilibration with carbogen gas (95% O₂, 5% CO₂).

Electrophysiology

Whole-cell recordings were obtained from MLIs identified with gradient-contrast infrared video microscopy⁴⁹ based on their location within the molecular layer and distinct morphology¹⁵. Patch pipettes used for current-clamped MLIs contained a solution of (in mM) 128 K-gluconate, 2 KCl, 9 HEPES, 4 MgCl₂, 4 NaATP, and 0.5 NaGTP and 9 K₄-BAPTA. For paired recordings, perforated access⁵⁰ to the presynaptic MLI was achieved by including amphotericin B (300 μg/ml, previously aliquoted in DMSO). Alexa Fluor 488 hydrazide (30 μM; Molecular Probes; Eugene, OR) was also included in the pipette to fluorescently monitor the integrity of the perforated recording with light microscopy. If the patch membrane ruptured, resulting in cell fluorescence, recording was terminated. For Ca²⁺ imaging experiments, K₄-BAPTA was replaced with 200μM Fluo-5F and 13μM Alexa Fluor 594 hydrazide (Molecular Probes, Eugene, OR). Current-clamped cells were held near −70 mV with constant current injection. Action potentials were stimulated between 0.3 to 0.1 Hz (~25–60 trials per condition). MLIs were voltage-clamped with a pipette solution containing (in mM) 135 CsCl, 10 HEPES, 4 Mg₂ATP, 0.3 NaGTP and 5 EGTA. The holding potential was kept at −60 mV giving rise to inward GABA_AR-mediated currents (E_Cl- ≈ 0 mV). All pipettes had open-tip resistances of 3–12 MΩ.

Electrophysiological potentials and currents were recorded with a Multiclamp 700B amplifier (Molecular Devices, Union City, CA). Electrode series resistance was compensated by bridge balance in current-clamped cells and was uncompensated in the voltage-clamp configuration. Analog signals were filtered at 3–10 kHz and digitized at 20–50 kHz with a 16-bit A/D converter (Instrutech Corp., Port Washington, NY). High access resistance (40–80 MΩ) in the presynaptic perforated recording also likely contributed to filtering and action potential attenuation, though, access often improved during an experiment due to continued partitioning of the antibiotic. Data were collected using custom software (J.S. Diamond) written in IgorPro (Wavemetrics, Lake Oswego, OR) and analyzed using Axograph (Axograph Scientific, Sydney, Australia). GABA_AR-mediated synaptic responses were isolated with NBQX (10 μM) and D-AP5 (50 μM) (Tocris Cookson, Ellisville, MO) to block fast glutamate receptor-mediated responses. Picrotoxin (100 μM) was also included during imaging experiments to eliminate GABA_AR-mediated transmission. EGTA-AM (10–20 μM, Molecular Probes, Eugene, OR), made from a stock 100 mM solution in DMSO, was applied for at least 10 minutes during which data acquisition ceased. Upon resumption, EGTA-AM was continuously applied until the end of the experiment. In amplified views shown in figures, spontaneous IPSCs that were not time-locked to the presynaptic action potential were blanked prior to averaging for illustrative purposes; events were only included within 1 ms after the action potential peak. In our analysis, however, spontaneous events were left unmodified.

Two-photon laser-scanning microscopy

Fluorescence was monitored with a lab-built 2PLSM using an Olympus (Melville, NY) upright microscope, objective (60X, 1.0 NA), and oil-immersion condenser (1.4 NA). A Ti:sapphire laser (Coherent, Santa Clara, CA) tuned to 810 nm was used for excitation. Emitted green and red fluorescence were simultaneously collected by photomultiplier modules (H8224, Hamamatsu, Hamamatsu City, Japan) in both epi- and transfluorescence pathways using a 565 nm dichroic and 525/50 and 620/60 bandpass filters (Chroma, Battleboro, VT). Images were acquired using ScanImage software(K. Svoboda). Line scans were obtained at 500 Hz. Cells were filled for a minimum of 20 minutes prior to acquisition to allow for dye equilibration (axon varicosities <150 μm from the soma). For varicosities >150 μm from the soma, longer equilibration times were used.

Imaging analysis was performed offline using ImageJ (National Institutes of Health) and Axograph. Fluorescence changes were quantified as increases in green fluorescence normalized to red fluorescence (ΔG/R). Peak amplitude measurements of subthreshold depolarization-evoked Ca²⁺ transients were determined by averaging a 100 ms epoch prior to action potential firing. Alternating trials in which no stimulation occurred were used to subtract background fluorescence levels; the difference reflecting the peak response. The peak of action potential-evoked Ca²⁺ transients were determined from exponential fits of the fluorescence decay following stimulation (measured at t = 0). Changes in the peak amplitude of action potential-evoked responses might be underestimated if the indicator were in the non-linear range. The Ca²⁺ transients evoked by subthreshold depolarization were usually very small. Therefore, for illustrative purposes only, we integrated these responses to accentuate the rise in Ca²⁺ from the baseline noise during the subthreshold stimulus (Fig. 2b, 3a). This method was not included in any quantification.

Statistical Analysis

Reported values are mean ± SEM. Excel (Microsoft) and InStat (GraphPad Software) were used for statistical analysis using paired and unpaired t-tests (two-tailed). ANOVA with post-hoc Tukey’s tests were used when making multiple comparisons. Differences were considered statistical significance with α values of P < 0.05. In figures, asterisks denote statistical significance.

Supplementary Material

1. Supplemental Figure 1.

Subthreshold depolarization evokes and enhances Ca²⁺ entry through VSCCs

(a) An action potential with (black) and without (red) a preceding subthreshold depolarization (300 ms) recorded in the soma. Shown below are the resulting Ca²⁺ transients recorded in an axon varicosity (3 mM extracellular Ca²⁺) in control and following inhibition with a combination of VSCC blockers (ω-conotoxin MVIIC 1μM, SNX 0.3 μM, nimodipine 20 μM, and mibefradil 10μM). (b) Inhibition of axonal Ca²⁺ transients by VSCC blockers. Mean values are presented with error bars (+SEM); asterisks denote that the reductions from control responses were significant (paired t-test, P < 0.05).