Abstract
In mouse cerebellar Purkinje neurons (PNs), the climbing fiber (CF) input provides a signal to parallel fiber (PF) synapses, triggering PF synaptic plasticity. This signal is given by supralinear Ca2+ transients, associated with the CF synaptic potential and colocalized with the PF Ca2+ influx, occurring only when PF activity precedes the CF input. Here, we unravel the biophysical determinants of supralinear Ca2+ signals associated with paired PF-CF synaptic activity. We used membrane potential (Vm) and Ca2+ imaging to investigate the local CF-associated Ca2+ influx following a train of PF synaptic potentials in two cases: (1) when the dendritic Vm is hyperpolarized below the resting Vm, and (2) when the dendritic Vm is at rest. We found that supralinear Ca2+ signals are mediated by type-1 metabotropic glutamate receptors (mGluR1s) when the CF input is delayed by 100–150 ms from the first PF input in both cases. When the dendrite is hyperpolarized only, however, mGluR1s boost neighboring T-type channels, providing a mechanism for local coincident detection of PF-CF activity. The resulting Ca2+ elevation is locally amplified by saturation of endogenous Ca2+ buffers produced by the PF-associated Ca2+ influx via the mGluR1-mediated nonselective cation conductance. In contrast, when the dendritic Vm is at rest, mGluR1s increase dendritic excitability by inactivating A-type K+ channels, but this phenomenon is not restricted to the activated PF synapses. Thus, Vm is likely a crucial parameter in determining PF synaptic plasticity, and the occurrence of hyperpolarization episodes is expected to play an important role in motor learning.
SIGNIFICANCE STATEMENT In Purkinje neurons, parallel fiber synaptic plasticity, determined by coincident activation of the climbing fiber input, underlies cerebellar learning. We unravel the biophysical mechanisms allowing the CF input to produce a local Ca2+ signal exclusively at the sites of activated parallel fibers. We show that when the membrane potential is hyperpolarized with respect to the resting membrane potential, type-1 metabotropic glutamate receptors locally enhance Ca2+ influx mediated by T-type Ca2+ channels, and that this signal is amplified by saturation of endogenous buffer also mediated by the same receptors. The combination of these two mechanisms is therefore capable of producing a Ca2+ signal at the activated parallel fiber sites, suggesting a role of Purkinje neuron membrane potential in cerebellar learning.
- calcium channels
- cerebellar Purkinje neuron
- climbing fibre
- endogenous calcium buffer
- metabotropic glutamate receptors
- parallel fibre
Introduction
In cerebellar Purkinje neurons (PNs), short-term and long-term synaptic plasticity of the parallel fiber (PF) inputs can be induced by pairing PF EPSPs with a climbing fiber (CF) EPSP (Wang et al., 2000; Brenowitz and Regehr, 2005; Safo and Regehr, 2005). These forms of plasticity, which are believed to underlie associative motor learning and control behaviors (Ito, 2001), are homosynaptic and require local Ca2+ elevation at the site of activated PF synapses. Precisely, the Ca2+ transient triggered by a CF-EPSP alone is not local (Knöpfel et al., 1991; Miyakawa et al., 1992; Canepari and Vogt, 2008; Ait Ouares et al., 2019) because it is caused by a depolarizing transient, originating in the soma and in the initial dendritic segment, that spreads to distal dendrites to open voltage-gated Ca2+ channels (VGCCs). When a CF-EPSP occurs after PF activity, the associated Ca2+ transient is larger than the summation of the two Ca2+ transients triggered by PF and CF inputs alone and is therefore referred to as “supralinear” Ca2+ signal (Brenowitz and Regehr, 2005). However, in contrast to the Ca2+ transient associated with an unpaired CF-EPSP, the supralinear Ca2+ signal associated with paired PF and CF inputs is localized exclusively at the site of activated PF synapses.
The ability of a spread physiological signal to trigger a local signal, in combination with synaptic activity, exclusively at the sites of activated synapses is a crucial requirement for coincident detection and associative plasticity in many systems. In hippocampal and neocortical pyramidal neurons, synchronous action potentials and excitatory synaptic activity can potentiate activated synapses (Debanne, 1996). In this case, action potentials transiently depolarizing the dendrites can provide a signal localized at activated synapses by unblocking NMDA receptors from Mg2+ (Ascher and Nowak, 1987). However, in the case of PF synaptic plasticity induced by coincident CF inputs, the underlying biophysical mechanism does not involve NMDA receptors but instead type-1 metabotropic glutamate receptors (mGluR1s), which are locally activated by glutamate release from PF terminals (Hartell, 1994; Wang et al., 2000; Brenowitz and Regehr, 2005; Safo and Regehr, 2005). Because mGluR1 activation is not voltage-dependent, the fundamental question is to understand how a CF-EPSP, which alone generates a depolarization and a Ca2+ influx everywhere in the dendrite, can trigger a strong localized Ca2+ elevation, in combination with mGluR1 activation, exclusively at the site of activated PFs.
In a recent study, we demonstrated that the dendritic depolarization produced by a CF-EPSP activates two types of VGCCs, T-type and P/Q-type VGCCs, and that the activation of both channels is determined by the dendritic membrane potential (Vm) at the beginning of the CF-EPSP (Ait Ouares et al., 2019). Specifically, at initial hyperpolarized Vm, the dendritic Ca2+ transient is largely mediated by T-type VGCCs, whereas activation of A-type voltage-gated K+ channels limits the opening of P/Q-type VGCCs. In contrast, at initial depolarized Vm, T and A channels are inactivated, and the larger dendritic depolarizing transient activates P/Q-type VGCCs that mediate Ca2+ spikes. It was also shown that both T-type VGCCs (Hildebrand et al., 2009; Isope et al., 2012) and P/Q-type VGCCs (Otsu et al., 2014) can be boosted by mGluR1 activation, providing two potential biophysical mechanisms underlying localized supralinear Ca2+ signals that are required for coincident detection and PF synaptic plasticity.
In the present study, we addressed the question of how a CF-EPSP can generate a supralinear Ca2+ signal exclusively at the sites of activated PF synapses. We used a protocol of moderate PF stimulation mimicking a physiological burst to reveal the biophysical mechanism that permits a CF-EPSP to trigger a localized signal at the site of activated PF synapses in realistic scenarios. We used ultrafast Vm and Ca2+ imaging, using indicators of different affinity and a series of selective pharmacological manipulations, and we performed a detailed analysis to elucidate the different mechanisms underlying supralinear Ca2+ signals under different conditions.
Materials and Methods
Slice preparation, solutions, electrophysiology, and pharmacology
Experiments were ethically performed in accordance with European Directives 2010/63/UE on the care, welfare, and treatment of animals. Procedures were reviewed by the ethics committee affiliated with the animal facility of the university (D3842110001). We used mice aged 21–35 postnatal days (C57BL/6j) of either sex. Animals were anesthetized by isoflurane inhalation, and the entire cerebellum was removed after decapitation. Cerebellar sagittal slices (250 μm thick) were prepared following established procedures (Vogt et al., 2011a,b; Ait Ouares et al., 2016) using a Leica VT1200 (Leica). Slices were incubated at 37°C for 45 min before use. The extracellular solution contained the following (in mm): 125 NaCl, 26 NaHCO3, 1 MgSO4, 3 KCl, 1 NaH2PO4, 2 CaCl2, and 20 glucose, bubbled with 95% O2 and 5% CO2. The intracellular solution contained the following (in mm): 125 KMeSO4, 5 KCl, 8 MgSO4, 5 Na2-ATP, 0.3 Tris-GTP, 12 Tris-phosphocreatine, 20 HEPES, adjusted to pH 7.35 with KOH. In Vm imaging experiments, PNs were loaded with the voltage-sensitive dye (VSD) JPW1114 as previously described (Canepari and Vogt, 2008), and in some experiments, PNs were also loaded with the Ca2+ indicator Fura-2FF (at 1 mm) using a previously described procedure (Vogt et al., 2011a). In experiments of Ca2+ imaging only, Oregon Green BAPTA-5N (OG5N) was added to the internal solution at 2 mm concentration, and in some experiments, Fura-2 (Fura2) at 0.4 mm concentration was also included. Patch-clamp recordings were made at 32°C–34°C using a MultiClamp 700A amplifier (Molecular Devices), and electrical signals were acquired at 20 kHz using the analog-to-digital board of the CCD camera, regardless of the imaging acquisition rate. The measured Vm was corrected for junction potential (−11 mV) as previously estimated (Canepari et al., 2010). PF-EPSPs and CF-EPSPs were elicited by current pulses of 5–20 μA amplitude and 100 μs duration delivered by separate electrodes. To avoid jitters, the stimulation protocols were precisely controlled with a Master-8 pulse generator (A.M.P.I.). All recordings were performed at least 30 min after establishing whole cell. This minimum time before recording started was 45 min in experiments with Fura2, heparin, and ryanodine. In Vm imaging experiments, cells were initially loaded with the voltage-sensitive dye JPW1114 and repatched with a solution without this dye to avoid toxicity due to dye overloading. 7-(Hydroxyimino)cyclopropa[b]chromen-1a-carboxylate ethyl ester (CPCCOEt), (1S,2S)-2-[2-[[3-(1H-benzimidazol-2-yl)propyl]methylamino]ethyl]-6-fluoro-1,2,3,4-tetrahydro-1-(1-methylethyl)-2-naphthalenyl-cyclopropanecarboxylate-dihydrochloride (NNC550396 or NNC), and N,N,N-trimethyl-5-[(tricyclo[3.3.1.13,7]dec-1-ylmethyl)amino]-1-pentanaminium bromide hydrobromide (IEM1460) were dissolved in extracellular solution and applied to the bath via the perfusion system. Heparin sodium salt, 1H-pyrrole-2-carboxylic acid (3S,4R,4aR,6S,7S,8R,8aS,8bR,9S,9aS)-dodecahydro-4,6,7,8a,8b,9a-hexahydroxy-3,6a,9-trimethyl-7-(1-methylethyl)-6,9-methanobenzo[1,2]pentaleno[1,6-bc]furan-8-yl ester (ryanodine), and 1-[6-[[(17β)-3-methoxyestra-1,3,5(10)-trien-17-yl]amino]hexyl]-1H-pyrrole-2,5-dione (U73122) were dissolved in intracellular solution and applied via the patch recording. (S)-3,5-Dihydroxyphenylglycine (DHPG), dissolved in external solution, was applied through a pipette positioned near the PN dendrite using 20 ms pulses of pressure provided by a pressure ejector, PDES-2DX (npi). AmmTx, dissolved in external solution, was applied through a pipette positioned near the PN dendrite by continuous gentle pressure application. CPCCOEt, heparin sodium salt, ryanodine, U73122, NNC550396, and DHPG were purchased from Tocris Bioscience. IEM1460 was purchased from Hello Bio. AmmTx was purchased from Smartox Biotechnology.
Optical recording
Invitrogen OG5N (Thermo Fisher Scientific) was excited at 470 nm with an OptoLED (Cairn Research) as previously described (Jaafari et al., 2014; Ait Ouares et al., 2016, 2019; Jaafari and Canepari, 2016). Fura-2FF (Santa Cruz Biotechnology) and Invitrogen Fura-2 (Thermo Fisher Scientific) were excited at 385 nm using the OptoLED. Vm and Ca2+ optical measurements were achieved sequentially (Canepari and Vogt, 2008) by alternating excitation of Fura-2FF and excitation of the voltage-sensitive dye Invitrogen JPW1114 (Thermo Fisher Scientific) at 528 nm using a LDI laser (89 North). OG5N Ca2+ fluorescence was recorded at 530 ± 21 nm. Fura-2FF and Fura2 Ca2+ fluorescence was recorded at 510 ± 41 nm. JPW1114 Vm fluorescence was recorded at >610 nm. Image sequences were recorded using a NeuroCCD-SMQ camera (RedShirtImaging). Images were demagnified by ∼0.2× to visualize an area of ∼150 μm diameter and acquired at 5 or 20 kHz with a resolution of 26 × 26 pixels or 26 × 4 pixels, respectively. In all experiments, four trials, with a 20 s interval between two consecutive trials, were obtained to assess the consistency of the signals and the average of these four trials was used to improve the signal-to-noise ratio. The number of averaged trials was nine in 20 kHz recordings. However, in all figures, the somatic Vm recording used in the illustrations was from a single trial, corresponding to the first trial of the series. Fluorescence was corrected for bleaching using a filtered trial without signal. Ca2+ fluorescence transients were expressed as fractional changes of fluorescence (ΔF/F0), using the average of the first four images of the sequence to calculate F0. In the color-coded images, all pixels with intensity value below twice the minimum are set to 1 to exclude pixels with large photon noise threshold, and only those pixels having signal peak >5% are illustrated to appreciate the signal localization. Vm fluorescence transients were calibrated in mV using prolonged hyperpolarizing steps as described in a previous article (Ait Ouares et al., 2019).
Model and computer simulations of calcium binding to the indicators
A simplified version of the model described by Ait Ouares et al. (2016) was used to estimate the Ca2+ fluorescence signals in the presence of 2 mm OG5N and 400 μm Fura2. The model consists of a set of equations:
This model comprises four buffers, with association and dissociation constants KONBj and KOFFBj, respectively. Buffers 1 and 2 are the indicators OG5N and Fura2, respectively, having KONFura2 = 570 μm−1 s−1 (Canepari and Mammano, 1999) and KD (KOFF/KON) = 35 and 0.2 μm, respectively. Buffers 3 and 4 are the two pairs of binding sites of calbindin d-28k, having KON = 217.5 and 27.5 μm−1 s−1 and KOFF = 35.8 s−1 nm and 2.6 s−1, respectively, as used by Ait Ouares et al. (2019), to account for the difference in temperature and the radial diffusion from the published values (Nägerl et al., 2000). Calbindin d-28k dominates Ca2+ binding in the millisecond time scale (Schmidt and Eilers, 2009) and has an estimated concentration of 120 μm (Kosaka et al., 1993). The Ca2+ influx associated with the CF was approximated by a Gaussian function of 15 or 20 μm/ms and a 1.5 ms time constant, consistent with the detailed model described by Ait Ouares et al. (2019). We simulated 200 ms of recording using the function “ode45” in MATLAB (MathWorks). In these 200 ms, the Gaussian function approximating the CF-associated Ca2+ influx was delayed by 180 ms from the beginning of the simulation, and in the simulations corresponding to the paired scenarios, it was preceded by a slow Ca2+ influx approximated by another Gaussian function of 4 μm/ms and a 45 ms time constant to estimate the effect of saturating the high-affinity buffers. The OG5N 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). The Fura2 −ΔF/F0 signal was extrapolated from the Ca2+ bound to the indicator using the dynamic range of 0.9 as used by Ait Ouares et al. (2019).
Experimental design and statistical analysis
Data were processed and analyzed using MATLAB. Changes produced by the pairing protocol with respect to the CF stimulation alone, or by the pharmacological action of an agent, were assessed by performing the paired Student's t test on the signal parameter under the two different conditions. A change in the signal was considered significant when p < 0.01. In all figures, a significant change was indicated with *. In the specific case of data reported in Figure 1, the conclusion that the effect of adding CPCCOEt is different when the CF stimulation is delayed by 60 ms from the first PF stimulus, with respect to the cases of 100 and 150 ms delays, is supported by a paired t test performed on the ΔF/F0 ratios at two different delays that gave a value of p < 0.01.
For the quantitative interpretation of the results of combined Ca2+ imaging with the low-affinity indicator OG5N and with the high-affinity indicator Fura2, we calculated the variable S, defined as follows:
It must be noticed that S is principally used to analyze the supralinear Ca2+ signal at the hyperpolarized and resting states, and therefore, both the numerator and the denominator are different according to the different initial Vm. If a supralinear Ca2+ signal occurs when a CF-EPSP is evoked under two different conditions, such as unpaired and paired, then S(OG5N) is obviously positive because a larger free Ca2+ concentration implies a larger fraction of Ca2+ bound to the low-affinity indicator OG5N. The parameter S(Fura2) can unambiguously discriminate whether a supralinear Ca2+ signal is or is not exclusively due to the saturation of Fura2, and whether it is or is not exclusively due to a larger Ca2+ influx through the plasma membrane.
Case of supralinear Ca2+ signal exclusively due to the saturation of Fura2.
In the presence of Fura2, the larger signal under paired conditions can, in principle, be exclusively due to the saturation of the high-affinity indicator by the Ca2+ associated with the PF stimulation. Thus, if less free Fura2 is available to buffer the Ca2+ entering the cell during the CF-EPSP, more Ca2+ will bind to OG5N and less Ca2+ will bind to Fura2 under paired conditions, implying that S(Fura2) will be <1. Thus, if S(Fura2) is >1, it unambiguously proves that the supralinear Ca2+ signal is not exclusively due to the saturation of Fura2 (i.e., that at least a fraction of Ca2+ bound to the low-affinity indicator OG5N originates from Ca2+ influx through the plasma membrane).
Case of supralinear Ca2+ signal exclusively due to Ca2+ influx through the plasma membrane.
If the concentration of Fura2 bound to Ca2+ associated with the PF stimulation is negligible with respect to the free Fura2 concentration at the time of the CF-EPSP, then the Ca2+ transient entering the cell will linearly bind to OG5N and to Fura2, implying that S will have the same value for the two indicators. Thus, if the ratio S(Fura2)/S(OG5N) = 1, then Fura2 is not saturated. Alternatively, if S(Fura2)/S(OG5N) < 1, then at least a fraction of Ca2+ bound to the low-affinity indicator OG5N originates from the saturation of Fura2.
In the specific case in which the unpaired Ca2+ signal is the same (i.e., when the initial Vm is at the hyperpolarized state and the delay of the CF stimulation is either 60 or 110 ms from the first PF stimulus), the ratio S(Fura2)/S(OG5N) provides a comparative estimate of the Fura2 saturation. Indeed, regardless of the contribution of the Ca2+ influx that is different at the two different delays, the more Fura2 is saturated, the smaller S(Fura2) will be and, conversely, the larger S(OG5N) will be. Thus, the smaller S(Fura2)/S(OG5N) is, the larger the contribution of Fura2 saturation will be.
Results
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 Vm held between −75 and −85 mV, below the average Vm recorded in the soma when no current is injected (Vrest), 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 (KD = 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).
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 Vm 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 Vm 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 Vm (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 Vrest (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 Vm (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 Vm, 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 Vm (i.e., independently of whether the CF-associated Ca2+ transient is mediated by T-type or P/Q-type VGCCs).
Supralinear Ca2+ signals at different initial Vm. 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 Vm 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 Vrest (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 Vm 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 (Δtpeak) 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 (Δtpeak = 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 Δtpeak 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.
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 Vm 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, Vm 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. Δtpeak, Time difference between the two peaks for unpaired CF at hyp state and at rest state. D, Mean ± SD (N = 8 cells) of Δtpeak 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.
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 Vm 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 Vm 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 Vm transient associated with the CF-EPSP. We initially measured dendritic Vm associated with a paired CF-EPSP delayed by 60 ms from the first PF stimulus using Vm imaging from a region adjacent to the PF stimulation electrode. Vm 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 Vm 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 Vm, A-type K+ channels are partially inactivated, and activation of P/Q-type VGCCs is boosted (Ait Ouares et al., 2019), the peak Vm shifted from −21 to −3 mV. Significant shifts in the initial and peak Vm 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.
Dendritic Vm 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) Vm 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 Vm signals of A. Vinit and Vpeak, Initial and peak Vm, respectively, calibrated in mV. C, Mean ± SD (N = 8 cells) of Vinit and Vpeak 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 Vinit and Vpeak (p < 0.01, paired t test). All optical signals were from averages of four trials.
After this first set of Vm experiments, we performed combined Vm and Ca2+ imaging recordings with a similar approach recently used to characterize the dendritic Vm transient associated with the unpaired CF-EPSP (Ait Ouares et al., 2019) to measure dendritic Vm 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 (KD = 6 μm; Hyrc et al., 2000), used to localize the area of supralinear Ca2+ signals both at hyp and rest initial Vm (Fig. 6A). Consistent with the observation that PF-ESPSs locally activate VGCCs at hyp initial Vm, 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 Vm 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 Vm changes of <3 mV, we applied the following filtering procedures (Fig. 6C). For the unpaired CF recordings, the initial Vm was set to the averaged Vm signal before CF stimulation. For the paired CF recordings, the initial Vm 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 Vm. In N = 7 cells, in which we successfully combined Vm and Ca2+ with filtered signals above the photon noise to discriminate Vm changes <3 mV, the additional depolarization, calculated as the difference between the paired and unpaired CF-associated Vm transients, was always detectable in the case of hyp initial Vm and corresponded to a significant increase of 5.0 ± 1.6 mV (p < 0.01, paired t test). In the case of rest initial Vm, a detectable increase (>3 mV) was observed only in four of seven cells. The mean ± SD of the Vm was 2.5 ± 2.5 mV (N = 7 cells). We concluded that small additional Vm transients are always associated with mGluR1-dependent supralinear Ca2+ signals at hyp initial Vm, 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 Vm, additional Vm transients were occasionally observed, suggesting a smaller and possibly highly variable contribution of Ca2+ influx through the plasma membrane.
Dendritic Vm 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 Vm 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 Vm, indicating supralinear Ca2+ signals at R1. B, Vm-calibrated recordings associated with CF stimulation unpaired (red traces) or paired (pink traces) with PF stimulation at R1 and R2 superimposed to somatic Vm, in the case of hyp or rest initial Vm. C, Vm 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 Vm transients are indicated. E, Mean ± SD (N = 7 cells) of the Vm transient difference between the paired and unpaired CF in the case of hyp or rest initial Vm. *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 (KD ∼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 (SFura2) must be <1 (SFura2 = 0.95 in the case of the simulation in Fig. 7C). In contrast, a value of SFura2 >1 indicates that larger Ca2+ influx contributes to the supralinear Ca2+ signal (SFura2 = 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 (SFura2/SOG5N) 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 (SFura2/SOG5N = 0.54 in the case of the simulation in Fig. 7C). Thus, SFura2/SOG5N 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, SFura2 was significantly positive for the mGluR1-dependent supralinear Ca2+ signal at hyp states (p < 0.01, paired t test, SFura2 = 2.38 ± 0.05), confirming that larger Ca2+ influx through the plasma membrane contributes in this case (Fig. 8C). However, SFura2 was significantly smaller than SOG5N (p < 0.01, paired t test, SFura2/SOG5N = 0.58 ± 0.09), indicating that buffer saturation also contributes to this supralinear Ca2+ signal. Finally, SFura2/SOG5N for the mGluR1-independent supralinear Ca2+ signal (0.31 ± 0.13) was significantly smaller than SFura2/SOG5N 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 (SFura2/SOG5N = 0.80 ± 0.23). Thus, as expected, the further mGluR1-dependent Ca2+ increase includes a significant contribution of buffer saturation (SFura2/SOG5N = 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.
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.
Contribution of saturation of endogenous Ca2+ buffers to supralinear Ca2+ signals. A, Left, Dendritic PN area filled with two Ca2+ indicators: OG5N (2 mm, KD = 35 μm) and Fura2 (0.4 mm, KD = 0.2 μm); position of PF-stimulating electrode and adjacent region of interest are indicated. Somatic Vm 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 (SFura2) 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 SFura2. C, Right, Mean ± SD (N = 8 cells) of SFura2/SOG5N in the different cases as reported in the legend. *Significant difference (p < 0.01, paired t test) between SOG5N and SFura2; **significant difference (p < 0.01, paired t test) between two SOG5N/SFura2 ratios. All optical signals were from averages of four trials.
Origin of the mGluR1-dependent supralinear Ca2+ signals at initial hyperpolarized Vm
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 Vm 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 Vm and strongly reduced Ca2+ transients associated with both PF-EPSPs and the CF-EPSP at hyp initial Vm (Fig. 9A). In contrast, Ca2+ transients associated with both PF-EPSPs and the CF-EPSP at rest initial Vm 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 Vm 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 Vm 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 Vm 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.
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 Vm (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 Vm. 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 Vm 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.
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 Vm 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 Vrest
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 Vm >−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 Vm modulation, and therefore, it is not colocalized with mGluR1s. By analyzing cells in which Vrest 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.
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 Vm (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 Vm and dendritic Ca2+ signals in R2 reported in A, exhibiting a supralinear Ca2+ transient smaller than that in R1. C, Somatic Vm 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 Vm 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 Vm 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.
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 Vm (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.
Discussion
Determinants of the different dendritic supralinear Ca2+ signals in PNs
In this study, we identified the biophysical determinants of the dendritic supralinear Ca2+ signals, observed when a CF-EPSP is preceded by PF activation, that are believed to trigger PF synaptic plasticity (Hartell, 2002; Jörntell and Hansel, 2006; Vogt and Canepari, 2010). In a recent report, we demonstrated that isolated CF-EPSPs essentially trigger Ca2+ influx via T-type VGCCs when the initial Vm is <−70 mV and via P/Q-type VGCCs when the initial Vm is >−60 mV (Ait Ouares et al., 2019). In both cases, these signals are spread throughout the dendritic arborization at different extents. Here, we demonstrate that bursts of PF-EPSPs are capable of locally amplifying these two Ca2+ signals at the sites of activated PF synapses, driving the CF “error” signal specifically to these inputs (Safo and Regehr, 2008). The biophysical mechanisms that permit these phenomena of amplification are summarized in Figure 13. As shown in Figure 6B, a burst of PF-EPSPs locally depolarizes the dendritic area comprising the activated spines. Thus, if a CF-EPSP occurs during this depolarizing phase, the Ca2+ influx associated with the CF-EPSP will be principally mediated by P/Q-type VGCCs, independent of the initial Vm. The fast Ca2+ influx associated with the PF-mediated depolarization can transiently saturate the endogenous Ca2+ buffer, and therefore, a supralinear free Ca2+ signal is produced in the dendritic region where Vm is depolarized. It must be pointed out that this type of supralinear Ca2+ signal depends on the number and arrangement of activated PFs that determine the spatial pattern of depolarization, but it is not clearly established whether a scenario of “beams” of PF axons activating spines in close proximity occurs in vivo (Bower, 2002). During the short PF-mediated depolarization, mGluR1s do not affect the Ca2+ transient associated with the CF-EPSP (Fig. 1D). After the end of the PF train, Vm is repolarized and mGluR1 effects take action. In this time window, if the repolarized Vm is at the hyp state (<−70 mV), the Ca2+ influx associated with the CF-EPSP will be principally mediated by T-type VGCCs which are potentiated by mGluR1s (Fig. 9). In addition, the slow mGluR1-mediated Ca2+ influx is also capable of saturating the endogenous buffer, a mechanism that further amplifies this supralinear free Ca2+ signal (Fig. 10). Interestingly, both T channels potentiating and the slow mGluR1-mediated Ca2+ influx are triggered by the same mGluR1 pathway (Hildebrand et al., 2009) involving a protein tyrosine phosphatase (Canepari and Ogden, 2003), suggesting that the two factors contributing to this supralinear Ca2+ signal colocalize with mGluR1s at a submicron scale. Eventually, if the repolarized Vm is >−65 mV, the Ca2+ influx associated with a CF-EPSP will be principally mediated by P/Q-type VGCCs. The Ca2+ influx via P/Q-type VGCCs can also be boosted by mGluR1s through inhibition of A-type K+ channels (Otsu et al., 2014), but with two major differences with respect to the T channel boosting discussed earlier. First, this mechanism is highly variable because, at these initial Vm ranges, A-type K+ channels are already partially inactivated. Second, the potentiated P/Q channels do not colocalize with the site of activated PFs (Fig. 11A,B). Thus, the only mechanism providing colocalization in this case is saturation of the endogenous buffer produced by the slow Ca2+ influx that is independent of the initial Vm.
Illustration of the determinants of the different dendritic supralinear Ca2+ signals in PNs. Top, A representative PN dendritic region indicating the position of the PF stimulation electrode and a hand-drawn schematic of dendritic Vm associated with an unpaired CF-EPSP and the consequent Ca2+ transient at hyp and rest states. The Vm transient opens principally T and A channels at the hyp state (box b) and principally P/Q (and possibly A) channels at the rest state (box a). In both cases, Ca2+ entering through VGCCs binds to the endogenous Ca2+ buffer (ECB) without saturating it. On the bottom, a hand-drawn schematic of dendritic Vm associated with paired CF-EPSPs and the consequent Ca2+ transient. During the PF-EPSPs, Ca2+ entering through both VGCC types saturates the ECB (box c). At 60 ms delay from the first PF-EPSP, the CF-Ca2+ transient principally mediated by P/Q channels produces a supralinear Ca2+ increase because the ECB is saturated (box d). At 110 ms delay and at hyp state, the CF-Ca2+ transient produces a supralinear Ca2+ increase because T channels are potentiated by active mGluR1s, and the ECB is saturated by Ca2+ entering through the mGluR1-activated channel (box f). At 110 ms delay and at rest state, the CF-Ca2+ transient produces a supralinear Ca2+ increase because the ECB is saturated by Ca2+ entering through the mGluR1-activated channel and possibly because of a larger Ca2+ influx via P/Q channels enhanced by stronger inactivation of A channels by mGluR1s (box e).
Role of the initial Vm and implication for synaptic plasticity
The evidence that mGluR1s locally potentiate a specific Ca2+ source (the T-type VGCC) and concomitantly amplify the same signal suggests that the occurrence of a hyp state episode is important for synaptic plasticity. Yet, in vivo electrode recordings have shown that, in PNs, Vrest is similar to that observed in brain slices (Kitamura and Häusser, 2011), indicating that the dendritic Vm is normally not at the hyp state. Also, spontaneous bursts of dendritic Ca2+ spikes are observed by recording Ca2+ and Vm signals simultaneously (Roome and Kuhn, 2018), again indicating that depolarizing transients occur when the dendrite is relatively depolarized. These observations suggest that the hyp state might occur exclusively during rare but crucial episodes in which synaptic plasticity may occur. Episodes of hyp state can be due, in principle, to synaptic inhibition coactively occurring during activity patterns associated with cerebellar learning (Suvrathan and Raymond, 2018). Indeed, it was recently demonstrated that optogenetic activation of interneurons in the molecular layer strongly affects PF synaptic plasticity and motor learning, as well as CF-mediated signaling (Gaffield et al., 2018; Rowan et al., 2018). Another possibility is that hyp state episodes are intrinsically generated in PNs by a “priming” signal that hyperpolarizes the cell, such as by activating a K+ conductance. Interestingly, in a recent study, it was proposed that PF synaptic depression may physiologically occur when two CF events occur, the first one concomitant with a PF-EPSP burst and the second one after 100 ms (Bouvier et al., 2018). Notably, the concomitant activation of PF and CF inputs hyperpolarizes the PN, as also visible in the examples with 60 ms delay reported in Figures 3A and 8A.
Role of the mGluR1-activated nonselective cation conductance
An mGluR1-activated nonselective cation conductance, permeable to Ca2+, is the channel that mediates the slow mGluR1 Ca2+ influx (Canepari et al., 2001, 2004). This channel has been identified as TRPC3 because, in mutant mice lacking this channel, all signals associated with the mGluR1-activated nonselective cation conductance are absent (Hartmann et al., 2008). This mouse, as well as another mouse model carrying a point mutation in the TRPC3 channel (Becker et al., 2009), exhibit severe changes in behavioral cerebellar functions. More recently, it was shown that mGluR1s can also activate GluD2 delta “orphan” glutamate receptors, raising the question of whether the mGluR1-activated nonselective cation conductance is actually composed of different channels (Ady et al., 2014). Notably, critical mutations of GluRD2 impair synaptic plasticity and motor learning and cerebellar behavioral functions (Hirano, 2006). Regardless of the identity of the channel underlying the slow mGluR1 Ca2+ influx, in this study we report a clear role for this conductance. During the time window of mGluR1 action, Ca2+ entering the cell after a PF-EPSP burst binds to mobile proteins forming the rich endogenous buffer of PNs (Bastianelli, 2003). Therefore, in this critical time window, the buffer capacity of PN dendrites, which is exceptionally high at rest (Fierro and Llano, 1996), can locally decrease to a level that enables Ca2+ influx associated with a CF-EPSP to activate molecular pathways that would not be activated by an unpaired CF-EPSP. The transient saturation of the endogenous buffer has been already identified as a crucial computational element that allows local integration of Ca2+ signals (Maeda et al., 1999). However, experimental evidence of endogenous buffer saturation was associated with dendritic depolarization (Canepari and Vogt, 2008), colocalized with activated VGCCs which go beyond the spatial pattern of activated spines. In contrast, endogenous buffer saturation caused by the slow Ca2+ influx can be segregated to activated synapses. In general, the spatial pattern of activated spines (Schmidt et al., 2007) and their geometry (Schmidt and Eilers, 2009) are expected to be determinants of endogenous buffer saturation in the case of synaptic Ca2+ influx. Finally, endogenous buffer saturation caused by the slow mGluR1 Ca2+ influx is likely colocalizing with the potentiated T-type VGCCs when the dendrite is at hyp state, because the two mGluR1-mediated mechanisms share the same molecular pathway (Hildebrand et al., 2009).
Perspectives
This study was achieved by monitoring fluorescence simultaneously in large portions of PN dendrites with a spatial resolution of ∼10 μm. Thus, the dendritic sites of recording comprise several spines and the parent dendritic segment. A full understanding of the signals in terms of individual spines and dendritic bulks, which is not available in this study, should be achieved in the near future either by reducing the temporal resolution using rapid multisite confocal imaging (Filipis et al., 2018) or by reducing the number of recording spots using two-photon random access microscopy (Otsu et al., 2008). A second question that should be explored in detail is the relation between the initial Vm state and the induction of PF synaptic plasticity. Numerous studies have investigated other parameters as crucial determinants of PF synaptic plasticity, including timing, size of Ca2+ signals, spatial arrangements of synaptic inputs, and others (Vogt and Canepari, 2010). In this study, we show that a scenario, in which a specific Ca2+ source (the T-type VGCC) is locally boosted and amplified by mGluR1s, occurs only when the initial Vm is hyperpolarized, a phenomenon that can translate into an important information-processing rule associated with cerebellar function.
Footnotes
This work was supported by the Agence Nationale de la Recherche through three grants (ANR-14-CE17-0006–WaveFrontImag; Labex Ion Channels Science and Therapeutics: program number ANR-11-LABX-0015; and National Infrastructure France Life Imaging “Noeud Grenoblois”) and by the Federation pour la recherché sur le Cerveau (Grant Espoir en tête, Rotary France). We thank Dr. Boris Barbour for useful discussions during the project, for reading the manuscript before submission, and for advice on improving the clarity of the figures.
The authors declare no competing financial interests.
- Correspondence should be addressed to Marco Canepari at marco.canepari{at}univ-grenoble-alpes.fr
