The Origin of Physiological Local mGluR1 Supralinear Ca²⁺ Signals in Cerebellar Purkinje Neurons

Distinct dendritic supralinear Ca²⁺ transients associated with paired PF-CF stimulation

Several studies have shown that, at sites of activated PF synapses, the Ca²⁺ transient associated with a CF-EPSP is significantly larger when it occurs after PF activation (Wang et al., 2000; Brenowitz and Regehr, 2005; Safo and Regehr, 2005; Canepari and Vogt, 2008). To initially analyze these supralinear Ca²⁺ signals, we used a protocol of stimulation for the PF input consisting of five stimuli delivered at 100 Hz with an electrode positioned near a PN dendritic site (Fig. 1A), with the initial V_m held between −75 and −85 mV, below the average V_m recorded in the soma when no current is injected (V_rest), which was always between −65 and −50 mV. The intensity of stimulation was set to attain a first EPSP in the range of 1–4 mV of somatic amplitude, and the protocol occasionally caused somatic action potentials during the last EPSPs. In cells filled with 2 mm low-affinity Ca²⁺ indicator OG5N (K_D = 35 μm; Canepari and Ogden, 2006), this protocol systematically produced a fast ΔF/F₀ signal, increasing at each individual PF-EPSP, followed by a slow ΔF/F₀ signal peaking at 100–150 ms after the first PF stimulus. As illustrated by the color-coded pictures in Figure 1A, both signals were localized within the same area of a few microns adjacent to the PF-stimulating electrode because the sharp depolarization associated with PF-EPSPs is also localized (Canepari and Vogt, 2008). However, at a submicron spatial scale, the two signals are probably not colocalizing, because the fast signal is mediated by VGCCs activated by the PF depolarization (Canepari and Vogt, 2008), whereas the slow signal is mediated by a nonselective cation conductance (Canepari et al., 2004), which is believed to be formed by C3-type transient receptor potential (TRPC3) channels (Hartmann et al., 2008).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Timing and localization of supralinear Ca²⁺ signals. A, Left, Reconstruction of PN filled with 2 mm OG5N, with position of two stimulating electrodes for the CF and PFs and two regions of interest indicated (R1 and R2). A, Right, Somatic V_m and Ca²⁺ ΔF/F₀ signals at R1 and R2 associated with PF stimulation (5 pulses at 100 Hz), with timing indicated by purple lines. Spatial distribution of ΔF/F₀ signal corresponding to the second-fifth PF-EPSPs and to the peak of the slow Ca²⁺ signal depicted using a color code. B, Somatic V_m and Ca²⁺ ΔF/F₀ signals at R1 and R2 associated with CF stimulation unpaired (light-blue traces) or paired (dark-blue traces) with PF stimulation at 100 Hz (timing indicated by purple lines) following at 60, 100, or 150 ms from the first PF pulse (timing indicated by purple triangles). Spatial distribution of the ΔF/F₀ signal associated with CF-EPSP paired with PF stimulation depicted using a color code. C, Ca²⁺ ΔF/F₀ signals at R1 associated with CF stimulation unpaired or paired with PF stimulation delayed by 60, 100, or 150 ms from the first PF pulse in control solution (blue traces) or after addition of 20 μM of the mGluR1 antagonist CPCCOEt (green traces). The time window outlined is reported below. D, Mean ± SD (N = 6 cells) of ΔF/F₀ ratio between the paired and unpaired signals in control solution or after addition of CPCCOEt. *Significant inhibition of supralinear Ca²⁺ signal by CPCCOEt (p < 0.01, paired t test). The dotted line depicts ΔF/F₀ ratio = 1. All optical signals were from averages of four trials.

The time course of the Ca²⁺ transient in the 2 × 2 pixel (∼11 × 11 μm²) region (R1) adjacent to the PF-stimulating electrode is reported together with ΔF/F₀ from a control region (R2) of the same size, with no Ca²⁺ signal to show the high signal-to-noise ratio of these measurements, permitting a precise quantitative analysis of Ca²⁺ signals. We investigated the Ca²⁺ signal associated with a CF-EPSP unpaired or paired with the PF-EPSPs at delays of 60, 100, or 150 ms from the first PF stimulus (Fig. 1B). In all three cases, the Ca²⁺ transient associated with the paired CF-EPSP was larger than that associated with the unpaired CF-EPSP, and these phenomena were restricted to the area where the two Ca²⁺ transients associated with the PF-EPSPs were observed, as illustrated by the color-coded frames in Figure 1B. Addition of the mGluR1 inhibitor CPCCOEt (20 μm) reduced the slow ΔF/F₀ signal associated with PF stimulation (Fig. 1C) and the supralinear Ca²⁺ signals associated with the CF-EPSP delayed by 100 and 150 ms from the first PF stimulus, but not that associated with the CF-EPSP delayed by 60 ms from the first PF stimulus. In N = 6 cells, we quantified the supralinear Ca²⁺ signal by calculating the ratio between the amplitude of the ΔF/F₀ signal associated with the paired CF-EPSP and the amplitude of the ΔF/F₀ signal associated with the unpaired CF-EPSP (ΔF/F₀ ratio; Fig. 1D). The ΔF/F₀ ratio significantly (p < 0.01, paired t test) decreased from 3.40 ± 0.73 to 1.93 ± 0.42 and from 3.38 ± 0.69 to 2.02 ± 0.27 in the cases of 100 and 150 ms delay of the CF stimulation, respectively. In contrast, it did not significantly (p > 0.1, paired t test) change (from 2.23 ± 0.47 to 2.19 ± 0.60) in the case of 60 ms delay of the CF stimulation. We concluded that supralinear Ca²⁺ signals associated with CF-EPSPs delayed by 100 and 150 ms were mGluR1-dependent, whereas the supralinear Ca²⁺ signal associated with the CF-EPSP delayed by 60 ms was mGluR1-independent.

The CF-associated Ca²⁺ transient at initial hyperpolarized V_m (hyp; ∼−80 mV) is mainly mediated by Ca²⁺ influx via T-type VGCCs (Ait Ouares et al., 2019). Thus, we further investigated the supralinear Ca²⁺ signal at V_rest (rest; ∼−60 mV); i.e., when T-type VGCCs are mostly inactivated and the CF-associated Ca²⁺ transient is largely mediated by Ca²⁺ influx via P/Q-type VGCCs that can be activated because also most of A-type K⁺ channels are inactivated. Again, we used the protocol consisting of five PF stimuli at 100 Hz with an electrode positioned near a PN dendritic site (Fig. 2A), followed by a CF-EPSP delayed by 110 ms from the first PF stimulus. In this experiment, however, recordings were performed both at hyp and rest initial V_m (Fig. 2B). In both cases, the Ca²⁺ transient associated with the paired CF-EPSP was larger than that associated with the unpaired CF-EPSP, and both supralinear Ca²⁺ signals were colocalized with the PF active area (see the color-coded frames in Fig. 2B). Addition of CPCCOEt reduced both supralinear Ca²⁺ signals associated with the CF-EPSP (Fig. 2C). In N = 6 cells, the ΔF/F₀ ratio significantly (p < 0.01, paired t test) decreased from 3.02 ± 0.61 to 1.88 ± 0.50 and from 2.19 ± 0.38 to 1.46 ± 0.30 in the cases of hyp and rest initial V_m, respectively (Fig. 2D). We concluded that supralinear Ca²⁺ signals associated with CF-EPSPs, delayed by 110 ms from the first PF stimulus, were mGluR1-dependent both at hyp and at rest initial V_m (i.e., independently of whether the CF-associated Ca²⁺ transient is mediated by T-type or P/Q-type VGCCs).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Supralinear Ca²⁺ signals at different initial V_m. A, Reconstruction of PN filled with 2 mm OG5N with the position of two stimulating electrodes for the CF and PFs and a region of interest adjacent to the PF-stimulating electrode indicated. B, Somatic V_m and Ca²⁺ ΔF/F₀ signals at the region of interest associated with CF stimulation unpaired (light-blue traces) or paired (dark-blue traces) with PF stimulation (timing indicated by purple lines) delayed by 110 ms from the first PF pulse (timing indicated by purple triangle) in the case of hyperpolarized (hyp; ∼−80 mV) or V_rest (rest; ∼−60 mV). Spatial distribution of the ΔF/F₀ signal associated with CF-EPSP paired with PF stimulation depicted using a color code. C, Same Ca²⁺ ΔF/F₀ signals associated with CF stimulation unpaired or paired with PF stimulation in control solution (blue traces) or after addition of 20 μM of the mGluR1 antagonist CPCCOEt (green traces). The time window outlined is reported below. D, Mean ± SD (N = 6 cells) of the ΔF/F₀ ratio between the paired and unpaired signals in control solution or after addition of CPCCOEt. *Significant inhibition of supralinear Ca²⁺ signal by CPCCOEt (p < 0.01, paired t test). The dotted line depicts ΔF/F₀ ratio = 1. All optical signals were from averages of four trials.

Kinetics of dendritic supralinear Ca²⁺ signals

The three forms of supralinear Ca²⁺ signals presented in the previous paragraph are time-correlated with the CF-associated Ca²⁺ transients, which are mediated by T-type and P/Q-type VGCCs (Ait Ouares et al., 2019). In principle, they can be due to the modulation of these Ca²⁺ sources, the activation of other Ca²⁺ sources, or both phenomena. If only the original Ca²⁺ source is larger, then the kinetics of the rising phase of the Ca²⁺ transient is expected to be the same for the paired and unpaired CF signal. Thus, useful information can be obtained by analyzing the kinetics of the rising phase of the Ca²⁺ transients in paired protocols with respect to the Ca²⁺ transients associated with unpaired CF-EPSPs, to reveal any possible subsequent component of the Ca²⁺ transients. To this aim, we performed Ca²⁺ recordings associated with the same stimulation protocols described in the previous paragraph, but at 20 kHz in this case to accurately analyze the kinetics of Ca²⁺ transients (Fig. 3A). We then analyzed the rising phase of the different Ca²⁺ transients exploiting the short relaxation time of OG5N (Jaafari et al., 2015; Ait Ouares et al., 2016; Jaafari and Canepari, 2016). To quantitatively analyze the fast kinetics of Ca²⁺ transients, we averaged nine recordings for each stimulating protocol (instead of four). As shown in Figure 3A, the individual somatic V_m signals and the associated Ca²⁺ transients were perfectly aligned, indicating no jitter in the timing of the CF stimulation. We then normalized the signals to the peak and applied a Savitzky–Golay smoothing filter (Fig. 3B) with a 20 point time window, which reduces the noise with minimal distortion of the kinetics of the fluorescence transient (Savitzky and Golay, 1964), as previously demonstrated (Jaafari et al., 2015). Finally, we calculated the time derivative (Jaafari et al., 2014) as shown in Figure 3C. The time to peak of the time derivative, from the CF stimulus, is an indirect kinetics measurement of the Ca²⁺ source and is typically longer for T-type VGCCs that principally mediate Ca²⁺ influx at hyp states, with respect to P/Q-type VGCCs that principally mediate Ca²⁺ influx at rest states (Ait Ouares et al., 2019). Thus, in the example shown in Figure 3C, the time difference between the two peaks (Δt_peak) at the hyp and rest states, in the case of unpaired CF-EPSPs, was three samples (150 μs). We then measured the times to peak in the cases of CF-EPSPs in pairing protocol and compared those with the cases of unpaired CF-EPSPs. In the example of Figure 3C, the time to peak of the paired CF signal at the hyp state was the same as that of unpaired CF signal at the hyp state (Δt_peak = 0). In contrast, the times to peak of the paired CF signal at the rest state or with the CF stimulation delayed by 60 ms from the first PF stimulus were the same as that of the unpaired CF signal at the rest state. We performed this analysis in N = 8 cells and calculated Δt_peak from the CF signal at both hyp (Fig. 3D, top) and rest states (Fig. 3D, bottom). The time to peak of the paired CF signal at hyp state was the same as the unpaired CF signal at hyp state, and the time to peak of the paired CF signal at rest state was the same as the unpaired CF signal at rest state. In contrast, the time to peak of the CF signals (paired and unpaired) at hyp state was significantly (p < 0.01, paired t test) longer (by 2.75 ± 0.71 samples) than the time to peak of the CF signals (paired and unpaired) at rest state. Finally, the time to peak of the CF signals at hyp state was significantly (p < 0.01, paired t test) longer (by 2.38 ± 0.75 samples) than the time to peak of the paired CF signals with the CF stimulation delayed by 60 ms from the first PF stimulus. Therefore, we concluded that the fast kinetics of Ca²⁺ transients associated with the CF-EPSP are identical in paired and unpaired conditions, both at hyp and rest states, when the CF stimulation is delayed by 110 ms from the first PF stimulus. This conclusion suggests that the principal Ca²⁺ source of the supralinear Ca²⁺ signal with the CF stimulation delayed by 110 ms from the first PF stimulus at hyp states is the T-type VGCC. In contrast, the principal Ca²⁺ source of the supralinear Ca²⁺ signals with the CF stimulation delayed by 110 ms from the first PF stimulus at rest states, or with the CF stimulation delayed by 60 ms from the first PF stimulus, is likely the P/Q-type VGCC.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Kinetics of supralinear Ca²⁺ signals. A, Reconstruction of PN filled with 2 mm OG5N; position of PF-stimulating electrode and adjacent region of interest indicated. Somatic V_m and Ca²⁺ ΔF/F₀ signals in the region of interest associated with different protocols: from left to right, a CF-EPSP unpaired at hyp and rest states, paired with a PF-EPSP train delayed by 110 ms from the first PF stimulus at hyp and rest states, and paired with a PF-EPSP train delayed by 60 ms from the first PF stimulus. On the bottom, V_m and Ca²⁺ signals associated with the CF-EPSP from the nine individual trials used for the averages aligned and displayed on a 3 ms time window to show that single acquisitions are not affected by jitters. B, Left, Ca²⁺ ΔF/F₀ signals associated with the CF-EPSP reported in A aligned with the CF stimulation (dotted line indicated by “stim”) on a time window of 5 ms. B, Right, Same signals smoothed by a 20 point Savitzky–Golay filter and normalized to the peak on a time window of 3 ms. Traces are reported twice, first superimposed to the signal corresponding to the unpaired CF at hyp state and second to the signal corresponding to the unpaired CF at rest state. C, Normalized time derivative [d(ΔF/F₀)/dt] of the filtered signals in B. Traces are reported twice, first superimposed to the time derivative corresponding to the unpaired CF at hyp state and second to the time derivative corresponding to the unpaired CF at rest state. The dotted lines indicate the time of the peak of the d(ΔF/F₀)/dt signals. Δt_peak, Time difference between the two peaks for unpaired CF at hyp state and at rest state. D, Mean ± SD (N = 8 cells) of Δt_peak expressed in samples from the signal associated with the unpaired CF at hyp state (top) and from the signal associated with the unpaired CF at rest state (bottom). *Significant difference from the signal of reference (p < 0.01, paired t test). All optical signals were from averages of nine trials.

Yet, several studies have linked PF-triggered activation of mGluR1s to Ca²⁺ release from internal stores, in particular via inositol trisphosphate (InsP3) receptors (Finch and Augustine, 1998; Takechi et al., 1998). To directly test whether Ca²⁺ release from internal stores plays a role in these supralinear Ca²⁺ signals, we performed experiments with the internal solution containing either 1 mg/ml of heparin (Fig. 4A) to inhibit Ca²⁺ store release via InsP3 receptors (Kim et al., 2008) or 100 μm ryanodine (Fig. 4B) to inhibit Ca²⁺ store release via ryanodine receptors (Kano et al., 1995). In groups of six cells tested with each pharmacological agent, mGluR1-dependent supralinear Ca²⁺ transients were consistent with those measured in control internal solution (Fig. 4C). The same result was obtained in a group of six cells, tested at hyp states only, injected with 10 μm phospholipase C inhibitor U73122 (data not shown), which was proven to prevent mGluR1-mediated Ca²⁺ release from stores in PN dendrites (Canepari and Ogden, 2006). Together, these results rule out any significant contribution of Ca²⁺ release from stores via either InsP3 or ryanodine receptors.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Supralinear Ca²⁺ signals in the presence of heparin or ryanodine. A, Left, Reconstruction of PN filled with 2 mm OG5N and 1 mg/ml of heparin sodium salt to block InsP3 receptors; position of PF-stimulating electrode and adjacent region of interest indicated. Somatic V_m and Ca²⁺ ΔF/F₀ signals in the region of interest associated with different protocols: from left to right, a CF-EPSP unpaired at hyp and rest states, paired with a PF-EPSP train delayed by 110 ms from the first PF stimulus at hyp and rest states, and paired with a PF-EPSPs train delayed by 60 ms from the first PF stimulus. B, Same as A, but with the internal solution containing 100 μm ryanodine to block ryanodine receptors. C, Mean ± SD (N = 6 cells for each group) of ΔF/F₀ ratio between the paired and unpaired signals in the presence of heparin (green columns) or ryanodine (purple columns). Blue columns are mean ± SD of ΔF/F₀ ratios in control conditions already reported in Figures 1 and 2. Values for heparin were as follows: 3.17 ± 0.41 for 110 ms CF delay at hyp states, 2.02 ± 0.27 for 110 ms CF delay at rest states, 2.72 ± 0.54 for 60 ms CF delay. Values for ryanodine were as follows: 3.06 ± 0.85 for 110 ms CF delay at hyp states, 1.93 ± 0.56 for 110 ms CF delay at rest states, 2.56 ± 0.80 for 60 ms CF delay. Experiments were performed 45 min after whole cell to obtain full equilibration of the internal solution. Used concentrations of heparin and ryanodine were the maximal tolerated by the cells for 1 h recording. All optical signals were from averages of four trials.

Dendritic V_m associated with mGluR1-dependent supralinear Ca²⁺ signals

Because Ca²⁺ release from internal stores does not contribute to the supralinear Ca²⁺ signals reported above, the origin of these phenomena can be an increase in the Ca²⁺ influx through the plasma membrane during the CF-EPSP, an increase in free Ca²⁺ concentration (and, therefore, in the Ca²⁺ bound to the OG5N) due to a transient saturation of endogenous Ca²⁺ buffers, or a combination of both phenomena. If the supralinear Ca²⁺ signal includes an increase of Ca²⁺ influx through the plasma membrane, the additional charge influx is expected to increase the electrical current and, therefore, the dendritic V_m transient associated with the CF-EPSP. We initially measured dendritic V_m associated with a paired CF-EPSP delayed by 60 ms from the first PF stimulus using V_m imaging from a region adjacent to the PF stimulation electrode. V_m fluorescence transients were calibrated on an absolute scale (in mV) using prolonged hyperpolarizing pulses as previously described (Canepari and Vogt, 2008; Ait Ouares et al., 2019) to quantify the depolarization associated with the CF-EPSP under different conditions. In the experiment reported in Figure 5A, the initial V_m before the CF-EPSP shifted from −80 to −62 mV when the PF-EPSP train was applied (Fig. 5B). Consistent with the fact that at this initial V_m, A-type K⁺ channels are partially inactivated, and activation of P/Q-type VGCCs is boosted (Ait Ouares et al., 2019), the peak V_m shifted from −21 to −3 mV. Significant shifts in the initial and peak V_m were observed in N = 8 cells in which this experiment was performed (Fig. 5C), confirming that the mGluR1-independent supralinear Ca²⁺ signal, occurring when the CF-EPSP is at the end of the PF-EPSP train, is associated with a shift from T-type VGCCs to P/Q-type VGCCs, as suggested by the results reported in Figure 3.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Dendritic V_m associated with mGluR1-independent supralinear Ca²⁺ signals. A, Left, Dendrite of PN filled with the voltage-sensitive dye JPW1114; position of PF-stimulating electrode and adjacent region of interest indicated. A, Right, Somatic (top) and dendritic (bottom) V_m in the region of interest associated with a CF-EPSP unpaired or paired with a PF-EPSP train, with the CF stimulation delayed by 60 ms from the first PF stimulus. B, In a time window of 20 ms, dendritic V_m signals of A. V_init and V_peak, Initial and peak V_m, respectively, calibrated in mV. C, Mean ± SD (N = 8 cells) of V_init and V_peak for the signals associated with an unpaired CF-EPSP (−83 ± 3 and −18 ± 5 mV, respectively) and for the signals associated with a CF-EPSP paired with a PF-EPSP train, with the CF stimulation delayed by 60 ms from the first PF stimulus (−61 ± 4 and −5 ± 2 mV, respectively). *Significant differences of V_init and V_peak (p < 0.01, paired t test). All optical signals were from averages of four trials.

After this first set of V_m experiments, we performed combined V_m and Ca²⁺ imaging recordings with a similar approach recently used to characterize the dendritic V_m transient associated with the unpaired CF-EPSP (Ait Ouares et al., 2019) to measure dendritic V_m associated with mGluR1-dependent supralinear Ca²⁺ signals. The cell shown in Figure 6A was filled with the VSD JPW1114 and with 1 mm low-affinity Ca²⁺ indicator Fura-2FF (K_D = 6 μm; Hyrc et al., 2000), used to localize the area of supralinear Ca²⁺ signals both at hyp and rest initial V_m (Fig. 6A). Consistent with the observation that PF-ESPSs locally activate VGCCs at hyp initial V_m, the depolarization associated with the PF train was >−30 mV in the region (R1) adjacent to the stimulating electrode, whereas the depolarization was comparable with that in the soma in a different region (R2), as shown in Figure 6B. To excite VSD fluorescence >160 ms without causing photodamage, we used only 20% of the laser intensity applied for previous CF-associated V_m measurements (Ait Ouares et al., 2019). Using this weaker illumination, the SD of the photon noise was equivalent to ∼2.5 mV after calibration, and the peak-to-peak noise was equivalent to ∼10 mV in R1 (Fig. 6B). Therefore, to allow discrimination of V_m changes of <3 mV, we applied the following filtering procedures (Fig. 6C). For the unpaired CF recordings, the initial V_m was set to the averaged V_m signal before CF stimulation. For the paired CF recordings, the initial V_m was accurately estimated by applying a 64-point median filter to the original fluorescence signal after the end of the PF stimulation. Finally, for all recordings, a 4-point median filter was applied to the 20 ms fluorescence signal following CF stimulation. Figure 6D shows a small but detectable additional depolarization associated with the paired CF-EPSP at both hyp and rest initial V_m. In N = 7 cells, in which we successfully combined V_m and Ca²⁺ with filtered signals above the photon noise to discriminate V_m changes <3 mV, the additional depolarization, calculated as the difference between the paired and unpaired CF-associated V_m transients, was always detectable in the case of hyp initial V_m and corresponded to a significant increase of 5.0 ± 1.6 mV (p < 0.01, paired t test). In the case of rest initial V_m, a detectable increase (>3 mV) was observed only in four of seven cells. The mean ± SD of the V_m was 2.5 ± 2.5 mV (N = 7 cells). We concluded that small additional V_m transients are always associated with mGluR1-dependent supralinear Ca²⁺ signals at hyp initial V_m, a result consistent with the hypothesis that at least part of the supralinear Ca²⁺ signal under this condition is due to Ca²⁺ influx through the plasma membrane. In the case of rest initial V_m, additional V_m transients were occasionally observed, suggesting a smaller and possibly highly variable contribution of Ca²⁺ influx through the plasma membrane.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Dendritic V_m transients associated with mGluR1-dependent supralinear Ca²⁺ signals. A, Left, Dendritic PN area filled with the VSD JPW1114 and 1 mm Fura-2FF; position of PF-stimulating electrode and two regions of interest (R1 and R2) indicated. A, Right, Somatic V_m and Fura-2FF −ΔF/F₀ signals from single trials at R1 and R2 associated with CF stimulation unpaired (light-blue traces) or paired (dark-blue traces) with PF stimulation in the case of hyp (∼−80 mV) or rest (∼−60 mV) initial V_m, indicating supralinear Ca²⁺ signals at R1. B, V_m-calibrated recordings associated with CF stimulation unpaired (red traces) or paired (pink traces) with PF stimulation at R1 and R2 superimposed to somatic V_m, in the case of hyp or rest initial V_m. C, V_m recordings at R1 shown in B filtered as follows: for the unpaired CF, the average of the signals before the CF stimulation is used and in the next 20 ms a 4-points median filter is applied; for the paired CF, a 64-point median filter is applied after the PF train and before the CF stimulation, and a 4-point median filter is applied in the next 20 ms. D, Same signals as C in the outlined time window; the values of the V_m transients are indicated. E, Mean ± SD (N = 7 cells) of the V_m transient difference between the paired and unpaired CF in the case of hyp or rest initial V_m. *Significant difference (p < 0.01, paired t test). All optical signals were from averages of four trials.

Contribution of the saturation of the endogenous buffer to the supralinear Ca²⁺ signals

In a previous study, we have shown that trains of PF stimuli at 100 Hz produce a local progressive saturation of endogenous Ca²⁺ buffers that boosts a Ca²⁺ fluorescent transient associated with a CF-EPSP occurring at the end of the PF train (Canepari and Vogt, 2008). Because mGluR1 activation induces a slow Ca²⁺ influx, saturation of the endogenous Ca²⁺ buffer may occur on a longer time scale and contribute to the mGluR1-dependent supralinear Ca²⁺ signals. To test whether this phenomenon occurs, experiments can be performed with the inclusion of the high-affinity UV-excitable indicator Fura2 (K_D ∼200 nm). We initially tested different Fura2 concentrations (100, 200, and 400 μm), and we finally chose the concentration of 400 μm because, under this condition, the Fura2 signal was not saturated by a CF Ca²⁺ transient, and the OG5N was still large enough to be accurately analyzed. To examine quantitatively how this type of measurement can provide an answer to the question on whether and to what extent the saturation of the endogenous buffer contributes to the supralinear Ca²⁺ signals, we ran computer simulations using the simple but realistic biochemical model described in the Materials and Methods. Specifically, we analyzed a scenario in which the same Ca²⁺ influx associated with a CF-EPSP occurs unpaired or paired with a previous slow Ca²⁺ influx, and another scenario in which the paired Ca²⁺ influx is 25% larger with respect to the unpaired Ca²⁺ influx (Fig. 7A). When the simulation is run without Fura2, the putative OG5N ΔF/F₀ exhibits supralinear Ca²⁺ signals under paired conditions in both scenarios (Fig. 7B). When the simulation is instead run with Fura2, while supralinear Ca²⁺ signals under paired conditions are still present in the putative OG5N ΔF/F₀, the putative Fura2 −ΔF/F₀ behaves differently in the two scenarios: it exhibits a small supralinear Ca²⁺ signal in the case of potentiated Ca²⁺ influx, but the paired Ca²⁺ signal is smaller than the unpaired Ca²⁺ signal when Ca²⁺ influx is the same (Fig. 7C). To quantitatively analyze this result, we introduce the variable S, defined as the ratio of the paired ΔF/F₀ peak against the unpaired ΔF/F₀ peak. The conclusive information that can be obtained, by the analysis of the variable S, is explained in detail in the Experimental design and statistical analysis section of the Materials and Methods. In brief, when the supralinear Ca²⁺ signal is exclusively due to buffer saturation, then the value of S for Fura2 (S_Fura2) must be <1 (S_Fura2 = 0.95 in the case of the simulation in Fig. 7C). In contrast, a value of S_Fura2 >1 indicates that larger Ca²⁺ influx contributes to the supralinear Ca²⁺ signal (S_Fura2 = 1.22 in the case of the simulation in Fig. 7C). Because a supralinear Ca²⁺ signal exclusively due to larger Ca²⁺ influx (i.e., no buffer saturation) leads to an increase in the absolute ΔF/F₀ signal, which is the same for the two indicators, the ratio of S for the two indicators (S_Fura2/S_OG5N) must be 1 in this case. It follows that a lower value of this ratio indicates that buffer saturation contributes to the supralinear Ca²⁺ signal (S_Fura2/S_OG5N = 0.54 in the case of the simulation in Fig. 7C). Thus, S_Fura2/S_OG5N also represents a quantitative estimate of the extent of buffer saturation in the two cases. With the premise that the analysis of the saturation of Fura2 provides an estimate for the saturation of the endogenous buffer occurring physiologically, we run experiments by adding 400 μm Fura2 in the patch electrode. In the cell shown in Figure 8A, we applied the same stimulation protocols of Figure 3 and measured separately the absolute fluorescent transients from the two indicators (OG5N ΔF/F₀ and Fura2 −ΔF/F₀). Under this condition, supralinear OG5N Ca²⁺ signals were observed (Fig. 8B), indicating that these signals are not prevented by the presence of Fura2 at this moderate concentration. In contrast, whereas a supralinear Fura2 Ca²⁺ signal associated with the paired CF-EPSP delayed by 110 ms at the hyp state was observed, the Fura2 Ca²⁺ signal associated with the paired CF-EPSP delayed by 60 ms was smaller than the Ca²⁺ signal associated with the unpaired CF-EPSP (Fig. 8B). We then calculated the variable S for the ratios of the ΔF/F₀ peaks in each protocol against the ΔF/F₀ peak associated with the unpaired CF stimulation (Fig. 8C). In N = 8 cells tested, S_Fura2 was significantly positive for the mGluR1-dependent supralinear Ca²⁺ signal at hyp states (p < 0.01, paired t test, S_Fura2 = 2.38 ± 0.05), confirming that larger Ca²⁺ influx through the plasma membrane contributes in this case (Fig. 8C). However, S_Fura2 was significantly smaller than S_OG5N (p < 0.01, paired t test, S_Fura2/S_OG5N = 0.58 ± 0.09), indicating that buffer saturation also contributes to this supralinear Ca²⁺ signal. Finally, S_Fura2/S_OG5N for the mGluR1-independent supralinear Ca²⁺ signal (0.31 ± 0.13) was significantly smaller than S_Fura2/S_OG5N for the mGluR1-dependent supralinear Ca²⁺ signal (p < 0.01, paired t test), indicating that, in this case, buffer saturation provides a larger contribution. The Ca²⁺ increase through P/Q-type VGCCs produced by switching from hyp states to rest states is essentially due to larger Ca²⁺ influx (S_Fura2/S_OG5N = 0.80 ± 0.23). Thus, as expected, the further mGluR1-dependent Ca²⁺ increase includes a significant contribution of buffer saturation (S_Fura2/S_OG5N = 0.54 ± 0.19, calculated using the unpaired CF signal at rest state). This analysis, however, does not allow estimating a possible contribution of larger Ca²⁺ influx in the case of mGluR1-dependent supralinear Ca²⁺ signal at rest states.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Computer simulations of the Ca²⁺ fluorescence signals in the presence of 2 mm OG5N and 400 μm Fura-2. A, Left, Right, Ca²⁺ influx used in the computer simulations associated with a CF-EPSP: Gaussian function of 15 μm/ms (left) or 20 μm/ms (right) and 1.5 ms time constant unpaired or paired with a preceding slow Ca²⁺ influx approximated by another Gaussian function of 4 μm/ms and 45 ms time constant. B, Computer simulations of the putative OG5N ΔF/F₀ signal in the absence of Fura2 using the model described in detail in the Materials and Methods and the Ca²⁺ influx reported in A. The ΔF/F₀ signal was extrapolated from the Ca²⁺ bound to the indicator using the dynamic range of 15 measured by Ait Ouares et al. (2016). C, Top, Bottom, Computer simulations of the putative OG5N ΔF/F₀ signal (top) and the putative Fura2 −ΔF/F₀ signal (bottom) in the presence of 400 μm Fura2 using the same model and Ca²⁺ influx as in B. The Fura2 −ΔF/F₀ signal was extrapolated from the Ca²⁺ bound to the indicator using the dynamic range of 0.9 used by Ait Ouares et al. (2019). The values of the variable S, described in detail in the Experimental design and statistical analysis section of the Materials and Methods, are reported for the two indicators.

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

Contribution of saturation of endogenous Ca²⁺ buffers to supralinear Ca²⁺ signals. A, Left, Dendritic PN area filled with two Ca²⁺ indicators: OG5N (2 mm, K_D = 35 μm) and Fura2 (0.4 mm, K_D = 0.2 μm); position of PF-stimulating electrode and adjacent region of interest are indicated. Somatic V_m and Ca²⁺ signals in the region of interest associated with different protocols: from left to right, a CF-EPSP unpaired at hyp and rest states, paired with a PF-EPSP train delayed by 110 ms from the first PF stimulus at hyp and rest states, and paired with a PF-EPSP train delayed by 60 ms from the first PF stimulus; signals from the two indicators (OG5N ΔF/F₀ and Fura −ΔF/F₀) were obtained by alternating the excitation wavelength. B, Same Ca²⁺ transients associated with the CF-EPSP in A, for the two indicators, superimposed either to the unpaired CF transient at hyp state or to the unpaired CF transient at rest state. C, S is defined as the ratio between the ΔF/F₀ peaks for each protocol and the ΔF/F₀ peak associated with the unpaired CF stimulation. C, Left, Mean ± SD (N = 8 cells) of S corresponding to Fura2 signals (S_Fura2) in the different cases as reported in the legend. *Significant difference (p < 0.01, paired t test) between the two ΔF/F₀ peaks used to calculate S_Fura2. C, Right, Mean ± SD (N = 8 cells) of S_Fura2/S_OG5N in the different cases as reported in the legend. *Significant difference (p < 0.01, paired t test) between S_OG5N and S_Fura2; **significant difference (p < 0.01, paired t test) between two S_OG5N/S_Fura2 ratios. All optical signals were from averages of four trials.

Origin of the mGluR1-dependent supralinear Ca²⁺ signals at initial hyperpolarized V_m

The previously presented results demonstrate that mGluR-dependent supralinear Ca²⁺ signals at the hyp state are due to a combination of Ca²⁺ influx increase through the plasma membrane and a transient saturation of the endogenous Ca²⁺ buffer. The rising phase of the paired CF Ca²⁺ signal has the same kinetics as the Ca²⁺ signal associated with the unpaired CF input (Fig. 3), which is principally mediated by T-type VGCCs (Ait Ouares et al., 2019). It has been reported that mGluR1 activation potentiates Cav3.1 T-type VGCCs in PNs (Hildebrand et al., 2009). Hence, we finally assessed directly whether the Ca²⁺ influx component of the mGluR1-dependent supralinear Ca²⁺ signal at hyp states is through these channels. In the cell reported in Figure 9A, we recorded the Ca²⁺ transient associated with a CF-EPSP unpaired or paired with PF stimulation at hyp or rest initial V_m in control conditions with a delay of 110 ms from the first PF stimulus and the CF stimulation. Addition of 30 μM of the Cav3.1 blocker NNC550396 inhibited spontaneous firing at rest initial V_m and strongly reduced Ca²⁺ transients associated with both PF-EPSPs and the CF-EPSP at hyp initial V_m (Fig. 9A). In contrast, Ca²⁺ transients associated with both PF-EPSPs and the CF-EPSP at rest initial V_m were only slightly affected by NNC addition, confirming that dendritic T-type VGCCs are mostly inactivated at this state (Fig. 9A). As shown in Figure 9B, the supralinear Ca²⁺ transient at hyp initial V_m was also strongly inhibited by addition of NNC. In N = 6 cells tested with this experimental protocol, the ratios between the CF Ca²⁺ signals in the presence of NNC and in control condition at initial hyperpolarized V_m were 0.44 ± 0.18 and 0.32 ± 0.13 for the CF unpaired or paired with PF stimulation, respectively (Fig. 9C). In both cases, the Ca²⁺ signal associated with the paired CF-EPSP was significantly smaller (p < 0.01, paired t test). In contrast, the ratios between the CF Ca²⁺ signals in the presence of NNC and in control condition at rest initial V_m were 0.93 ± 0.13 and 0.91 ± 0.10 for the CF unpaired or paired with PF stimulation, respectively, indicating that, in this state, T-type VGCCs contribute marginally to Ca²⁺ signals (Fig. 9C). The results reported in Figure 9 unambiguously confirm that the Ca²⁺ influx component responsible for mGluR1-dependent supralinear Ca²⁺ signals at hyp state is mediated by T-type Ca²⁺ channels.

• Download figure
• Open in new tab
• Download powerpoint

Figure 9.

Supralinear Ca²⁺ signals after blocking T-type VGCCs. A, Left, Dendritic area of a PN filled with 2 mm OG5N, with position of PF-stimulating electrode and an adjacent region of interest indicated. A, Right, Somatic V_m (black traces) and Ca²⁺ ΔF/F₀ (blue or green traces) signals at the region of interest associated with CF stimulation unpaired (light traces) or paired (dark traces) with PF stimulation (timing indicated by purple lines) delayed by 110 ms from the first PF pulse (timing indicated by purple triangle) in the case of hyp or rest initial V_m. Signals were acquired in control conditions (blue traces) or after addition of 30 μM of the T-type VGCC blocker NNC (green traces). B, Same Ca²⁺ signals as in A over a 20 ms time window. C, Mean ± SD (N = 6 cells) of the ratios between the ΔF/F₀ peak after addition of NNC and the peak under control conditions; dark-blue columns are for paired signals; light-blue columns are for unpaired CF signals. Ratios are calculated both for unpaired and paired CF signals. *Significant reduction of paired Ca²⁺ signal after addition of NNC (p < 0.01, paired t test). The dotted line depicts ratio = 1. All optical signals were from averages of four trials.

As shown in the example of Figure 9A, addition of NNC reduced the fast Ca²⁺ transient associated with the PF-EPSP train, whereas it did not affect the slow Ca²⁺ transient mediated by mGluR1s. Furthermore, although both unpaired and paired Ca²⁺ transients were inhibited by NNC, the paired Ca²⁺ transient was still larger (Fig. 9A), suggesting that local amplification of the residual component by buffer saturation is still present when T-type VGCCs are blocked. Thus, endogenous buffer is likely saturated by the slow mGluR1-mediated Ca²⁺ influx. To test this hypothesis, we measured mGluR1-dependent supralinear Ca²⁺ signals at hyp states while blocking the slow mGluR1-mediated Ca²⁺ transient with the channel blocker IEM1460 that does not affect the mGluR1-dependent boosting of T-type VGCCs (Hildebrand et al., 2009). In the example in Figure 10A, addition of 250 μm IEM1460 produced a use-dependent block of the slow mGluR1-dependent Ca²⁺ transient over a period of 20 min, in which stimulation protocols were repeated every 5 min. Whereas a change in the somatic V_m recordings was observed after IEM1460 application, the fast PF-mediated Ca²⁺ transients did not change, indicating that VGCCs were not directly affected by the blocker. Whereas the Ca²⁺ transient associated with the unpaired CF-EPSP was also not affected, the Ca²⁺ transient associated with the paired CF-EPSP was progressively reduced consistently with the slow mGluR1-mediated Ca²⁺ transient that was extrapolated using a simple fitting of the decay phase of the fast PF-mediated Ca²⁺ transient (Fig. 10B). We performed this experiment in N = 6 cells, obtaining a significant reduction of the slow mGluR1-mediated Ca²⁺ transient (p < 0.01, paired t test) 20 min after IEM1460 application, corresponding to a ΔF/F₀ ratio of 0.49 ± 0.16 (Fig. 10C). Consistently with this result, the supralinear Ca²⁺ signal, quantified again as the ratio between the paired and the unpaired ΔF/F₀ peaks, significantly (p < 0.01, paired t test) decreased from 3.37 ± 0.83 to 1.86 ± 0.56 (Fig. 10D). We concluded that the mGluR1-dependent supralinear Ca²⁺ signal at hyp states is caused by boosting of T-type VGCCs amplified by endogenous buffer saturation produced by the slow mGluR1-dependent PF-mediated Ca²⁺ influx.

• Download figure
• Open in new tab
• Download powerpoint

Figure 10.

Use-dependent block of mGluR1 slow Ca²⁺ transients and of supralinear Ca²⁺ signals. A, Left, Dendritic area of a PN filled with 2 mm OG5N, with position of PF-stimulating electrode and an adjacent region of interest indicated. Somatic V_m and Ca²⁺ ΔF/F₀ signals at the region of interest associated with an unpaired CF-EPSP, with PFs stimulated with five pulses at 100 Hz or with a CF-EPSP paired with PF stimulation at hyp state. Black and dark-blue traces were from recordings in control solution. Traces at different gray or blue tones were from recordings performed a time (t) after addition of 250 μM of the channel blocker IEM1460. B, Left, Same recordings as in A for PF stimulation only (PF) in control condition and 20 min after addition of IEM1460; the exponential fit of the Ca²⁺ ΔF/F₀ decay after the fifth PF-EPSP is also reported with a dotted line. B, Right, Ca²⁺ transients associated with the slow mGluR1-mediated component calculated as the difference between the Ca²⁺ ΔF/F₀ signals associated with PF stimulation and the fit trace on the left. C, Mean ± SD (N = 6 cells) of the ratio between the slow mGluR1-dependent PF-mediated Ca²⁺ transients (PF-fit) 20 min after addition of IEM1460 and in control condition. *Significant reduction of the signal (p < 0.01, paired t test). The dotted line depicts ratio = 1. D, Mean ± SD of the ΔF/F₀ ratio between the paired and unpaired signals in control solution or after addition of IEM1460. *Significant inhibition of supralinear Ca²⁺ signal by IEM1460 (p < 0.01, paired t test). The dotted line depicts ΔF/F₀ ratio = 1. All optical signals were from averages of four trials.

Origin of the mGluR1-dependent supralinear Ca²⁺ signals at V_rest

The experiments reported in Figures 8 and 10 suggest that the mGluR1-dependent supralinear Ca²⁺ signal also at rest state is largely due to saturation of the endogenous buffer, but the results reported in Figure 9 further suggest that an increase in Ca²⁺ influx may contribute in a variable manner. It has been reported that mGluR1 activation shifts the inactivation curve of A-type K⁺ channels, boosting the opening of P/Q VGCCs at initial V_m >−65 mV in PNs (Otsu et al., 2014). Thus, consistent with the results of the kinetics analysis reported in Figure 3, an increase in Ca²⁺ influx via P/Q-type VGCCs may contribute to this supralinear Ca²⁺ signal. In contrast with the direct local regulation of T-type VGCCs by mGluR1s (Hildebrand et al., 2009) occurring at hyp states, the indirect regulation of P/Q-type VGCCs is produced by a V_m modulation, and therefore, it is not colocalized with mGluR1s. By analyzing cells in which V_rest was <−60 mV (i.e., where A-type K⁺ channels are not yet fully inactivated), we observed some occasional small supralinear Ca²⁺ signals at sites far away from the PF-activated region at rest states. An example of this wide supralinear Ca²⁺ signal is reported in Figures 11A and 10B. Although this far supralinear Ca²⁺ signal was smaller than the supralinear Ca²⁺ signal colocalized with the PF-mediated signal (Fig. 11A), the kinetics were qualitatively similar to those of a “supralinear” Ca²⁺ signal obtained by blocking A-type K⁺ channels with the toxin AmmTx3 (Ait Ouares et al., 2019; Zoukimian et al., 2019), as shown in the example in Figure 11C.

• Download figure
• Open in new tab
• Download powerpoint

Figure 11.

Wide mGluR1-dependent supralinear Ca²⁺ signals at rest states. A, Left, Dendrite of PN filled with 2 mm OG5N; position of PF-stimulating electrode and two regions of interest indicated (R1 and R2). R1 is adjacent to the PF-stimulating electrode; R2 is ∼60 μm away from the PF-stimulating electrode. A, Right, Somatic V_m (top) and dendritic Ca²⁺ signals (bottom) in R1 and R2 associated with a CF-EPSP unpaired or paired with a PF-EPSP train, with the CF stimulation delayed by 110 ms from the first PF stimulus at rest state. B, In a time window of 20 ms, somatic V_m and dendritic Ca²⁺ signals in R2 reported in A, exhibiting a supralinear Ca²⁺ transient smaller than that in R1. C, Somatic V_m and dendritic Ca²⁺ signal associated with an unpaired CF-EPSP in control condition or after addition of the A-type K⁺ channel inhibitor AmmTx3. The two supralinear Ca²⁺ signals reported in A and B are qualitatively very similar. All optical signals were from averages of four trials.

Finally, in many studies, physiological mGluR1 activation has been mimicked by bath application of the agonist DHPG (for example, see Tempia et al., 2001; Canepari et al., 2001; Maejima et al., 2005; Otsu et al., 2014). We examined this artificial stimulation protocol, which in principle can also produce Ca²⁺ release from internal stores, by applying 100 μm DHPG, using a pipette positioned near the dendrites, with a short (20 ms) pressure ejection as shown in the example in Figure 12A. At hyp state, DHPG application triggered a slow Ca²⁺ signal, but in contrast to the physiological PF stimulation, a corresponding slow V_m depolarization was observed in the soma (Fig. 12A). Thus, although a substantial supralinear Ca²⁺ signal associated with DHPG pairing was observed, the kinetics of the Ca²⁺ transient indicate a dominance of Ca²⁺ influx via P/Q-type VGCCs (Fig. 12B), consistent with the fact that the initial V_m for the CF-EPSP is ∼−60 mV. The size of the supralinear Ca²⁺ signal associated with DHPG application is comparable with those associated with PF stimulation (Fig. 12C). Nevertheless, because all of these phenomena are largely due to transient saturation of the endogenous buffer, and the spatial and temporal dynamics of Ca²⁺ influx causing this saturation are different in the different protocols, we conclude that experiments using DHPG application cannot realistically mimic the physiological scenarios associated with mGluR1 activation.

• Download figure
• Open in new tab
• Download powerpoint

Figure 12.

Supralinear Ca²⁺ signals associated with DHPG application. A, Left, Dendrite of PN filled with 2 mm OG5N; a region of interest and the position of a pipette delivering 100 μm DHPG dissolved in external solution are indicated. DHPG application was achieved by applying a 20 ms pulse of pressure using a pressure ejector. A, Right, Somatic V_m (top) and dendritic Ca²⁺ signals (bottom) in the region of interest associated with a CF-EPSP unpaired (light-blue trace) or paired with DHPG application (dark-blue trace); the timing of DHPG application and of CF stimulation is indicated with a purple line and a purple arrow, respectively. A ΔF/F₀ artifact is observable during the short DHPG application. B, In a time window of 20 ms, dendritic Ca²⁺ signals reported in A, exhibiting a supralinear Ca²⁺ transient. C, Left, Mean ± SD (2.20 ± 0.80, N = 7 cells) of ΔF/F₀ ratio between the CF-associated signal paired with DHPG application and unpaired. C, Right, Mean ± SD of ΔF/F₀ ratios between CF-associated signals paired with PF stimulation and unpaired; the statistics relative to 60 ms delay between the first PF stimulus and the CF stimulation are the same as reported in Figure 1D. The two statistics relative to 110 ms delay between the first PF stimulus and the CF stimulation at hyp and rest states are the same as reported in Figure 2D. The dotted line depicts ΔF/F₀ ratio = 1.

In summary, at the rest state, regulation of P/Q-type VGCCs by mGluR1 activation may contribute to the supralinear Ca²⁺ signal, but the localization of this signal, produced by Ca²⁺ influx via these channels, is entirely due to the previous Ca²⁺ entry that saturates the endogenous buffer.