Separate Ca²⁺ Sources Are Buffered by Distinct Ca²⁺ Handling Systems in Aplysia Neuroendocrine Cells

Abstract

Although the contribution of Ca²⁺ buffering systems can vary between neuronal types and cellular compartments, it is unknown whether distinct Ca²⁺ sources within a neuron have different buffers. As individual Ca²⁺ sources can have separate functions, we propose that each is handled by unique systems. Using Aplysia californica bag cell neurons, which initiate reproduction through an afterdischarge involving multiple Ca²⁺-dependent processes, we investigated the role of endoplasmic reticulum (ER) and mitochondrial sequestration, as well as extrusion via the plasma membrane Ca²⁺-ATPase (PMCA) and Na⁺/Ca²⁺ exchanger, to the clearance of voltage-gated Ca²⁺ influx, Ca²⁺-induced Ca²⁺-release (CICR), and store-operated Ca²⁺ influx. Cultured bag cell neurons were filled with the Ca²⁺ indicator, fura-PE3, to image Ca²⁺ under whole-cell voltage clamp. A 5 Hz, 1 min train of depolarizing voltage steps elicited voltage-gated Ca²⁺ influx followed by EGTA-sensitive CICR from the mitochondria. A compartment model of Ca²⁺ indicated the effect of EGTA on CICR was due to buffering of released mitochondrial Ca²⁺ rather than uptake competition. Removal of voltage-gated Ca²⁺ influx was dominated by the mitochondria and PMCA, with no contribution from the Na⁺/Ca²⁺ exchanger or sarcoplasmic/endoplasmic Ca²⁺-ATPase (SERCA). In contrast, CICR recovery was slowed by eliminating the Na⁺/Ca²⁺ exchanger and PMCA. Last, store-operated influx, evoked by ER depletion, was removed by the SERCA and depended on the mitochondrial membrane potential. Our results demonstrate that distinct buffering systems are dedicated to particular Ca²⁺ sources. In general, this may represent a means to differentially regulate Ca²⁺-dependent processes, and for Aplysia, influence how reproductive behavior is triggered.

Introduction

Intracellular Ca²⁺ transduces electrical signals into biochemical cascades that control vital processes, including gene expression, excitability, and secretion (Clapham, 1995). Free Ca²⁺ is determined by the equilibrium between Ca²⁺ sources and Ca²⁺ removal (Catterall and Few, 2008). In neurons, Ca²⁺ enters primarily through voltage-gated Ca²⁺ channels, although the endoplasmic reticulum (ER) and mitochondria provide additional Ca²⁺ reservoirs for release (Armstrong and Hille, 1998; Rizzuto and Pozzan, 2006). Ca²⁺ removal relies on plasma membrane extrusion, controlled by the Na⁺/Ca²⁺ exchanger and plasma membrane Ca²⁺-ATPase (PMCA), and sequestration, mediated by the mitochondrial uniporter and sarcoplasmic/endoplasmic reticulum Ca²⁺-ATPase (SERCA) (Kim et al., 2003, 2005; Rizzuto and Pozzan, 2006). The goal of this study is to examine the contribution of extrusion and sequestration to handling Ca²⁺ from different sources.

It is well established that the involvement of a given Ca²⁺ buffering system to the removal of a Ca²⁺ load varies between neuronal types and cytological compartments (Thayer and Miller, 1990; Fierro et al., 1998; Krizaj and Copenhagen, 1998; Morgans et al., 1998; Juhaszova et al., 2000; Holthoff et al., 2002; Kim et al., 2003, 2005). As the Ca²⁺-dependent activation of downstream targets relies on a specific concentration and/or pattern of Ca²⁺, this disparity likely reflects the different Ca²⁺ requirements for particular processes in a given neuron (e.g., motor vs sensory neuron) or compartment (e.g., dendrites vs axon terminal) (Caride et al., 2001; Berridge et al., 2003). Similar to different compartments, discrete Ca²⁺ sources can have unique roles in controlling neuronal function (Deisseroth et al., 1998; Berridge et al., 2000). Despite this, the relative contribution of given removal systems to the clearance of Ca²⁺ from distinct sources remains largely unexplored.

To address whether Ca²⁺ sources are handled uniquely, we used the bag cell neurons of the marine mollusk Aplysia californica. On stimulation, these neurons undergo ∼30 min of action potential firing, known as an afterdischarge, during which egg-laying hormone (ELH) is secreted into the blood stream to initiate reproduction (Kupfermann and Kandel, 1970; Arch, 1972; Pinsker and Dudek, 1977). As the afterdischarge progresses, voltage-gated Ca²⁺ influx, Ca²⁺-induced Ca²⁺-release (CICR), and store-operated Ca²⁺ influx are engaged to provide the Ca²⁺ that orchestrates a sustained increase in excitability and neuropeptide secretion through various Ca²⁺-dependent mechanisms (DeRiemer et al., 1984; Loechner et al., 1990; Wilson et al., 1996; Michel and Wayne, 2002; Kachoei et al., 2006; Hung and Magoski, 2007). We show that voltage-gated Ca²⁺ influx is primarily sequestered by the mitochondria, which subsequently releases the Ca²⁺ to ultimately be extruded across the plasma membrane. A second plasma membrane Ca²⁺ source, store-operated Ca²⁺ influx, is preferentially cleared by the SERCA. Recent evidence from cervical ganglion neurons indicates that the differential contribution of removal systems can control which Ca²⁺ source activates a given intracellular pathway (Wheeler et al., 2012). By analogy, the differential Ca²⁺ clearance we observe in bag cell neurons may facilitate the production of source-specific Ca²⁺ profiles. This could ensure coupling to specific Ca²⁺ signaling pathways in the presence of multiple, spatio-temporally overlapping Ca²⁺ signals.

Materials and Methods

Animals and cell culture

Adult Aplysia californica (a hermaphrodite) weighing 150–500 g were obtained from Marinus. Animals were housed in an ∼300 L aquarium containing continuously circulating, aerated artificial sea water (Instant Ocean, Aquarium Systems) at 14–16°C on a 12/12 h light/dark cycle and fed Romaine lettuce 5 times/week. For primary cultures of isolated bag cell neurons, animals were anesthetized by an injection of isotonic MgCl₂ (∼50% body weight), the abdominal ganglion removed and incubated for 18 h at 22°C in neutral protease (13.33 mg/ml; 165859, Roche Diagnostics) dissolved in tissue culture artificial sea water (tcASW) (composition in mm: 460 NaCl, 10.4 KCl, 11 CaCl₂, 55 MgCl₂, 15 HEPES, 1 mg/ml glucose, 100 U/ml penicillin, and 0.1 mg/ml streptomycin, pH 7.8 with NaOH). The ganglion was then transferred to fresh tcASW and the bag cell neuron clusters were dissected from the surrounding connective tissue. Using a fire-polished Pasteur pipette and gentle trituration, neurons were dispersed onto 35 × 10 mm polystyrene tissue culture dishes (catalog #353001; Falcon, Becton Dickinson) filled with 2 ml of tcASW. Cultures were maintained in tcASW for 1–3 d in a 14°C incubator. Experiments were performed on neurons that were in vitro for at least 1 d. Salts were obtained from Fisher Scientific or Sigma-Aldrich.

Whole-cell, voltage-clamp recordings

Voltage-clamp recordings were made using an EPC-8 amplifier (HEKA Electronics) and the tight-seal, whole-cell method. Microelectrodes were pulled from 1.5 mm external, 1.2 mm internal diameter borosilicate glass capillaries (TW150F-4, World Precision Instruments) and had a resistance of 1–2.5 MΩ when filled with intracellular saline (see below). Pipette junction potentials were nulled, and subsequent to seal formation, pipette capacitive currents were cancelled. Following break-through, neuronal capacitance was also cancelled, and the series resistance (3–5 MΩ) compensated to 80% and monitored throughout the experiment. Current was filtered at 1 kHz with the EPC-8 Bessel filter and sampled at 2 kHz using a Digidata 1322A analog-to-digital converter (Molecular Devices), computer, and Clampex software (version 10.2, Molecular Devices). Voltage stimuli were delivered with Clampex.

Ca²⁺ currents were isolated using Ca²⁺-Cs⁺-tetraethylammonium (TEA) ASW, where the NaCl and KCl were replaced by TEA-Cl and CsCl, respectively, and the glucose and antibiotics were omitted (composition in mm: 460 TEA-Cl, 10.4 CsCl, 55 MgCl₂, 11 CaCl₂, 15 HEPES, pH 7.8 with CsOH). In some cases, the NaCl was not replaced by TEA to allow for Na⁺/Ca²⁺ exchanger activity. Whole-cell recordings used a Cs⁺-aspartate-based intracellular saline (composition in mm): 70 CsCl, 10 HEPES, 11 glucose, 10 glutathione, 5 ethyleneglycol bis (aminoethylether) tetraacetic acid (EGTA), 500 aspartic acid, 5 ATP (grade 2, disodium salt; A3377, Sigma-Aldrich), and 0.1 GTP (type 3, disodium salt; G8877, Sigma-Aldrich), pH 7.3 with CsOH. Certain experiments were performed with 0 mm EGTA in the internal saline. The majority of Ca²⁺-imaging (see below) was performed under whole-cell voltage clamp, during which the intracellular saline was supplemented with 1 mm fura-PE3 (0110; Teflabs) (Vorndran et al., 1995) to dye-fill neurons via passive dialysis. Cells were dialyzed for at least 20 min before experiments were performed.

Sharp-electrode current-clamp recording

For store-operated Ca²⁺ influx, current clamp was used to inject fura-PE3 into bag cell neurons using a PMI-100 pressure microinjector (Dagan). Neurons were filled with an optimal concentration of dye, similar to whole-cell conditions. Microelectrodes were pulled from 1.2 mm external, 0.9 mm internal diameter borosilicate glass capillaries (IB120F-4; World Precision Instruments) and had a resistance of 15–30 MΩ when the tip was filled with 10 mm fura-PE3 and backfilled with 3 m KCl. Recordings were made using an Axoclamp 2B amplifier (Molecular Devices) and the sharp-electrode, bridge-balanced method. All neurons used for imaging had resting potentials of −50 to −60 mV and action potentials that overshot 0 mV after depolarizing current injection (0.5–1 nA, from the amplifier). Store-operated influx recordings were performed in Ca²⁺-free ASW (composition as per tcASW but with added 0.5 mm EGTA and the glucose and antibiotics omitted) or Ca²⁺/Na⁺-free ASW [composition as per Ca²⁺-free ASW but with Na⁺ replaced with N-methyl-d-glucamine (NMDG)].

Ca²⁺ imaging

Imaging was performed using a TS100-F inverted microscope (Nikon) equipped with a Nikon Plan Fluor 10× [numerical aperture (NA) = 0.5], 20× (NA = 0.5), or 40× (NA = 0.6) objective. The light source was a 75 W Xenon arc lamp and a multiwavelength DeltaRAM V monochromatic illuminator (Photon Technology International) coupled to the microscope with a UV-grade liquid-light guide. Excitation wavelengths were 340 and 380 nm. Between acquisition episodes, the excitation illumination was blocked by a shutter, which along with the excitation wavelength, was controlled by a computer, a Photon Technology International computer interface, and EasyRatio Pro software (version 1.10, Photon Technology International). If image acquisition occurred at a frequency >0.2 Hz, the shutter remained open continuously. Emitted light passed through a 400 nm long-pass dichroic mirror and a 510/40 nm emission barrier filter before being detected by a Photometrics Cool SNAP HQ² charge-coupled device camera. Camera gain was maximized and exposure time adjusted on a per cell basis. Exposure times during 340 and 380 excitation were fixed to the same value. Background was removed by setting a minimal threshold value of 300 arbitrary units of fluorescence. Fluorescence intensities were typically sampled at 0.5 Hz. For longer recordings, sampling was switched to 0.2 Hz, after any fast periods of Ca²⁺ dynamics. Fluorescence signals were acquired using regions of interest measured over neuronal somata, at approximately the midpoint of the vertical focal plane and one-half to three-quarters of the cell diameter, then averaged eight frames per acquisition. The ratio of the emission following 340 and 380 nm excitation (340/380) was taken to reflect free intracellular Ca²⁺ (Grynkiewicz et al., 1985), and saved for subsequent analysis. Image acquisition, emitted light sampling, and ratio calculations were performed using EasyRatio Pro.

Reagents and drug application

Solution exchanges were accomplished by manual perfusion using a calibrated transfer pipette to first exchange the bath (tissue culture dish) solution. In most cases where a drug was applied, a small volume (≤10 μl) of concentrated stock solution was mixed with a larger volume of saline (∼100 μl) that was initially removed from the bath, and this mixture was then pipetted back into the bath. Carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP; 21857; Sigma-Aldrich), carboxyeosin (C-22803; Invitrogen), bafilomycin A (B1793, Sigma-Aldrich), and cyclopiazonic acid (CPA; C1530, Sigma-Aldrich or 239805, Calbiochem) all required dimethyl sulfoxide (DMSO; BP231, Fisher) as a vehicle. The maximal final concentration of DMSO was ≤0.5% (v/v) which, in control experiments as well as prior work from our laboratory, had no effect on membrane potential, various macroscopic or single-channel currents, resting intracellular Ca²⁺, or Ca²⁺ transients evoked by a train of action potentials (Kachoei et al., 2006; Lupinsky and Magoski, 2006; Hung and Magoski, 2007; Gardam et al., 2008; Geiger and Magoski, 2008; Tam et al., 2009, 2011; Hickey et al., 2010). Tetraphenylphosphonium chloride (TPP; 218790, Sigma-Aldrich) and lanthanum chloride (La³⁺; L-4131, Sigma-Aldrich) were prepared in water.

Analysis

Origin (version 7; OriginLab) was used to import and plot ImageMaster Pro files as line graphs. Analysis usually compared the steady-state value of the baseline 340/380 ratio with the ratio from regions that had reached a peak or new steady state. Averages of the baseline and peak regions were determined by eye or with adjacent-averaging. The rate of recovery of Ca²⁺ influx, after a stimulus train or store-operated Ca²⁺ entry and CICR, was quantified in different ways. To allow for comparison of data between our previous work on CICR (Geiger and Magoski, 2008), the rate of recovery from CICR was measured as the time required, after peak CICR, for the 340/380 ratio to return to 75% of the baseline ratio observed before the stimulus. The time at which the Ca²⁺ plateau first reached peak was considered time 0. Under circumstances where CICR was eliminated, time to 75% recovery was measured from the Ca²⁺ level at 1 min stimulation. This time was chosen because it reflects the typical point at which CICR responses peaked. Poststimulus area was used to quantify the magnitude and duration of CICR. Area was determined by integrating the region above the prestimulus baseline value from either 1 min post-train stimulus to 11 min or 11–21 min poststimulus. Again, measurements began at 1 min poststimulus to avoid including the initial recovery and capture peak CICR.

The degree of Ca²⁺ removal was also quantified by acquiring decay time constants and measuring the percentage recovery at 5 min poststimulus. Monoexponential decay functions were fit from the first point of decay to several minutes after complete recovery to baseline. The percentage recovery at 5 min was calculated by determining the degree of Ca²⁺ removed after the train stimulus or store-operated Ca²⁺ influx (340/380 peak–340/380 at 5 min post-peak) and dividing it by the peak rise during the response (340/380 peak–prestimulus baseline 340/380).

Summary data are presented as the mean ± SE). Statistics were performed using Instat (version 3.0; GraphPad Software). The Kolmogorov–Smirnov method was used to test datasets for normality. If the data were normal, Student's paired or unpaired t test (with the Welch correction as required) was used to test for differences between two means, whereas a standard one-way ANOVA with Dunnett's post hoc test was used to test for differences between multiple means. If the data were not normally distributed, a Mann–Whitney U test was used for two means, whereas a Kruskal–Wallis ANOVA with Dunn's post hoc test was used for multiple means. Fisher's exact test was used to test for differences in frequency between groups. A difference was considered significant if the two-tailed p value was <0.05.

For Figure 2, the rate of Ca²⁺ removal was determined for the post-train stimulus recovery period by deriving the slope of the Ca²⁺ decay ([Δ340/380]/Δt) at sequential time points using Microsoft Office Excel Plus 2010 (version 14). To prevent noise from influencing rate calculations, a fitted slope was measured starting at the initial decay point over 10 sequential time points ([Ca]_n − [Ca]_n+9/t_n − t_n+9) while incrementally shifting the start time (n+1) until the end of the decay phase. From this, a plot of Ca²⁺ decay rate versus 340/380 ratio was produced and fit with a second order polynomial function in Excel. Only traces that provided fits with r² > 0.9 were used for further analysis. Multiple polynomial fits were used to make an average polynomial function describing the dataset.

Figure 2.

Mitochondria remove voltage-gated Ca²⁺ influx and clear Ca²⁺ from repeated stimuli. A, Neurons voltage-clamped to −80 mV with 5 mm intracellular EGTA to allow for isolated measurement of voltage-gated Ca²⁺ influx and removal. A, Left, In DMSO, cytosolic Ca²⁺ transients evoked by the 5 Hz, 1 min train stimulus are followed by rapid recovery to baseline Ca²⁺. A, Right, Pretreatment with 20 μm FCCP, a protonophore that collapses the mitochondrial membrane potential and prevents Ca²⁺ sequestration, slows the recovery of Ca²⁺ following stimulation. A, Inset, The exponential decay time constant (τ) of the Ca²⁺ transient recovery phase is significantly larger in FCCP-treated neurons (unpaired Student's t test). B, Top, Relative Ca²⁺ clearance rate (R), calculated from the decay phase of Ca²⁺ transients shown in A, as a function of 340/380 ratio (rates normalized to the maximal value of the 340/380 range). Second-order polynomial fitted lines are plotted overtop of the data points. The difference between the control (R_total) and FCCP (R_FCCP) fits produce the estimated mitochondrial uptake (R_mit), represented by the light dashed line. B, Bottom, Second-order polynomial fit lines for R_control, R_FCCP, and R_mit representing averaged removal rates from multiple neurons. Sample sizes are different from the decay time constants shown in A due to quality of fit criteria required for rate functions (see Materials and Methods). C, Ca²⁺ influx from a train stimulus loads mitochondria with Ca²⁺. C, Left, FCCP (20 μm) elevates Ca²⁺ in neurons under voltage clamp at −80 mV with 5 mm EGTA in the pipette and 100 μm TPP to prevent potential release of mitochondrial Ca²⁺. C, Right, FCCP-induced Ca²⁺ release after a large influx of Ca²⁺ from a 5 Hz, 1 min train stimulus is increased. D, Train stimulation, before FCCP application, significantly enhances the peak percentage change upon FCCP-induced Ca²⁺ liberation (unpaired Student's t test). E, Mitochondrial Ca²⁺ clearance is necessary for recovery from repeated stimuli. E, Left, After a train stimulus in FCCP, a second stimulus produces a Ca²⁺ load that is largely unremoved. E, Right, To replicate the slow Ca²⁺ recovery in FCCP, a control cell is given a train stimulus, but then subsequently held at −30 mV to allow for a small persistent Ca²⁺ influx. Ca²⁺ levels are quickly restored following a second train stimulus when the cell is clamped at −80 mV. F, The ratio between the first and second percentage recovery at 5 min is significantly larger in FCCP-treated neurons (unpaired Student's unpaired t test).

Model development

Equations describing mitochondrial Ca²⁺ dynamics were adapted from Colegrove et al. (2000b) to produce a compartment model of bag cell neuron Ca²⁺.

Plasma membrane Ca²⁺ flux.

where J_influx is the rate of Ca²⁺ influx across the plasma membrane, k_influx refers to the Ca²⁺ permeability of the membrane, and [Ca²⁺]_i and [Ca²⁺]_e are the intracellular and extracellular Ca²⁺ concentrations, respectively. To produce Ca²⁺ influx in the model, k_influx was transiently increased and then reduced manually. J_efflux is the rate of plasma membrane efflux, V_max,efflux is the maximal rate of efflux, EC_{50, efflux} is the Ca²⁺ concentration at which efflux is half-maximal, and n_efflux is the Hill coefficient. J_pm is the net plasma membrane Ca²⁺ flux.

Mitochondrial Ca²⁺ dynamics.

where J_uptake is the rate of mitochondrial Ca²⁺ sequestration, k_{max, uptake} is the mitochondrial uptake rate constant, EC_{50, uptake} describes the Ca²⁺ concentration at which uptake is half-maximal, and n_uptake is the Hill coefficient. The factor, δ([Ca²⁺]_i), describes the inhibition of mitochondrial extrusion by cytosolic Ca²⁺. K_inhib is the Ca²⁺ concentration at which inhibition of J_release is half-maximal and n_inhib describes the sensitivity of inhibition to cytosolic Ca²⁺. V_{max, release} is the maximal rate of Ca²⁺ release from the mitochondria and EC_{50, release} is the concentration of mitochondrial Ca²⁺ ([Ca²⁺]_m) at which efflux rate is half of V_{max, release}. J_mito is the net Ca²⁺ flux of the mitochondria.

Exogenous Ca²⁺ buffers.

where J_EGTA is the rate of free cytosolic Ca²⁺ removal by EGTA (Nowycky and Pinter, 1993), k_off and k_on are the reverse and forward reaction constants, respectively, [CaB] is the concentration of the Ca²⁺-EGTA complex, [Ca²⁺]_i is the concentration of cytosolic Ca²⁺, and [B] is the concentration of free EGTA. Values for k_off and k_on (Table 1) were taken from Naraghi (1997), whereas [CaB] and [B] were calculated from the total EGTA concentration using MaxChelator (http://maxchelator.stanford.edu/CaEGTA-NIST.htm).

Table 1.

Parameter values used in compartment model of bag cell neuron Ca²⁺

Definition	Model variable	Standard value
Rate constant for PM Ca2+ influx	kinflux	5 × 10−6 (s−1)
Extracellular Ca2+ concentration	[Ca2+]e	11 mm
[Ca2+]i at half maximal rate of efflux	EC50, efflux	378.8 nm
Hill coefficient for efflux	nefflux	1.8
Maximal rate of efflux*	Vmax, efflux	4.06 ± 0.34 nm/s (n = 15)
[Ca2+]i at half-maximal rate of mitochondrial uptake	EC50, uptake	10 μm
Hill coefficient for mitochondrial Ca2+ uptake	nuptake	2
Rate constant for mitochondrial Ca2+ uptake*	kmax, uptake	10.3 ± 0.88 s−1 (n = 15)
[Ca2+]m at half-maximal rate of release	EC50, release	307 nm
Maximal rate of mitochondrial Ca2+ release*	Vmax, release	13.7 ± 4.0 nm/s (n = 15)
mitochondrial to cytosolic effective volume ratio	γ	2
[Ca2+]i at half-maximal release inhibition	EC50, inhib	500 nm
Hill coefficient for release inhibition	ninhib	6
Dissociation constant of EGTA	Kd, EGTA	180 nm
Forward rate constant of EGTA	kon	2.7 ± 106 M−1 · s−1
Reverse rate constant of EGTA	koff	0.5 s−1
Dissociation constant of fura	Kd	760 nm
Endogenous Ca2+ binding ratio	κe	60

*Indicates a free parameter estimated through data fitting (value ± SEM). All other parameters from the literature as indicated in text.

Ca²⁺-binding ratio.

The variable, κ, represents the mean Ca²⁺-binding ratio over the Ca²⁺ range experienced during typical neuronal excitation (Neher and Augustine, 1992). [B_t] is the total buffer concentration, and K_d is the dissociation constant of the exogenous buffer. [Ca²⁺]_{i, rest} and [Ca²⁺]_{i, peak} are the free intracellular Ca²⁺ concentrations in the bag cell neurons at rest and during peak stimulus-induced influx, respectively.

Total Ca²⁺ removal rate.

For model presentation, rates of change in free intracellular Ca²⁺ (d[Ca²⁺]_F/dt) were converted to rates of total Ca²⁺ removal (d[Ca²⁺]_T/dt). κ_B represents the average Ca²⁺-binding ratio for exogenous buffers (fura and EGTA) as calculated from Equation 9. Endogenous Ca²⁺-binding ratios (κ_S) were taken from estimates in Aplysia metacerebral neurons (Gabso et al., 1997).

Collective Ca²⁺ dynamics.

where d[Ca²⁺]_i/dt is the rate of change in cytosolic Ca²⁺, d[Ca²⁺]_m/dt is the rate of change of mitochondrial Ca²⁺, and γ is the ratio of effective mitochondrial and cytoplasmic volumes. The γ value used was taken from estimates in bullfrog sympathetic neurons (Colegrove et al., 2000b). For the estimates of model parameters, 340/380 ratios were converted to values of free intracellular Ca²⁺ based on Ca²⁺-sensitive electrode recordings and fura K_d measurements in Aplysia (Fisher et al., 1994; Gabso et al., 1997; Michel and Wayne, 2002). To fit individual traces, the V_{max, efflux}, k_{max, uptake}, and V_{max, release} were left as free variables, whereas the constants (EC₅₀ values) describing the Ca²⁺ sensitivity of plasma membrane extrusion and mitochondrial Ca²⁺ release were set to parameters established in bullfrog sympathetic neurons (Colegrove et al., 2000a,b). The components describing mitochondrial Ca²⁺ uptake (EC_{50, uptake} and n_uptake) were based on measurements in isolated mitochondria and used to determine k_max,uptake (Gunter and Pfeiffer, 1990; Gunter and Gunter, 1994; Colegrove et al., 2000b). Free parameters were then optimized to fit individual experimental records. Perhaps because of the time required for Ca²⁺ diffusion from the plasma membrane to the bulk of the cytosolic mitochondria, we found that proper fitting of bag cell neuron CICR often required a delayed onset of mitochondrial Ca²⁺ uptake. To account for this, our model implemented a time delay between the initial Ca²⁺ influx and the onset of mitochondrial buffering. The parameter estimates obtained from fitting were then collected from multiple neurons and averaged to obtain the values presented in Table 1. These values are shown as free, not total, intracellular Ca²⁺ to allow for comparison between similar models. Differential equations were solved numerically using Euler's method written in MATLAB with a time-step of 70 ms to produce graphical outputs of total cytosolic and mitochondrial Ca²⁺ over time.

Results

Mimicking the fast phase of the afterdischarge evokes distinct Ca²⁺ dynamics in bag cell neurons

A brief input to the bag cell neurons initiates the afterdischarge: a prolonged period of action potential firing consisting of a fast phase of ∼5 Hz for ∼1 min, which progresses into a slow phase of ∼1 Hz for ∼30 min (Kaczmarek et al., 1982; Fisher et al., 1994). To examine Ca²⁺ dynamics in response to a fast phase-like stimulus, a 1 min, 5 Hz train of 75 ms depolarizing steps to 0 mV was applied to fura-PE3-loaded, cultured bag cell neurons from a holding potential of −80 mV under whole-cell voltage clamp. Unless stated otherwise, all neurons were recorded using a Cs⁺-containing and TEA-containing external solution (to replace K⁺ and Na⁺, respectively) and a Cs⁺-containing internal pipette solution (to replace intracellular K⁺; see Materials and Methods for details).

Application of the train stimulus produced a large, transient rise in intracellular Ca²⁺ due to the activation of voltage-gated Ca²⁺channels, followed by an exponential decline, with recovery to baseline in 5–10 min (n = 8) (Fig. 1A, left). This response was measured with our Cs⁺-based intracellular saline containing 5 mm Ca²⁺ chelator, EGTA. As this buffer alters free intracellular Ca²⁺, we sought to apply the same stimulus when EGTA was omitted from the pipette solution. With 0 mm intracellular EGTA, excitation again resulted in a large Ca²⁺ transient; however, a prolonged Ca²⁺ plateau, often marked by a delayed peak, now followed the initial recovery (n = 6) (Fig. 1A, right). This Ca²⁺ plateau long outlasted the duration of the stimulus, and was followed by a slow return to baseline within 10–20 min. Similar sequences of changes to intracellular Ca²⁺ have been described as CICR in dorsal root ganglion neurons, bullfrog sympathetic neurons, adrenal chromaffin cells, and Aplysia neuron R15 (Gorman and Thomas, 1980; Friel and Tsien, 1994; Herrington et al., 1996; Colegrove et al., 2000a).

Figure 1.

A train of depolarizing stimuli induces a secondary Ca²⁺ rise sensitive to the Ca²⁺ chelator, EGTA. A, Simultaneous measurement of free intracellular Ca²⁺ and membrane current in cultured bag cell neurons using 340/380 fura PE3 fluorescence and whole-cell voltage clamp at −80 mV. A, Inset, A phase contrast image shows the recording pipette, bag cell neuron soma, and its neuritic processes. The bottom image shows the same neuron loaded with fura and the somatic region of interest (ROI) used for data collection. Scale bar applies to both images. A, Top left, Ca²⁺ influx indicated by a change in intensity of the 340/380 fluorescence ratio following a 1 min, 5 Hz train of 75 ms steps from −80 to 0 mV. With 5 mm intracellular EGTA, stimulation causes a large rise in Ca²⁺ followed by a rapid recovery to the prestimulus baseline. Top right, In the absence of intracellular EGTA, there is a prolonged Ca²⁺ plateau subsequent to the initial influx that greatly outlasts the stimulus duration, indicative of CICR. A, Bottom, Traces depict 300 overlaid Ca²⁺ currents from each pulse to 0 mV of the 1 min train stimulus in either 5 or 0 mm intracellular EGTA. The shifting band of traces is due to use-dependent inactivation of Ca²⁺ currents during the train stimulus. Unless stated otherwise, all cells were recorded in a Cs⁺-external and TEA-external (to replace K⁺ and Na⁺, respectively) and a Cs⁺-based intracellular solution (to replace intracellular K⁺). B, Left, The percentage change in 340/380 from baseline to the peak response during the train stimulus is significantly larger in 5 mm EGTA versus 0 mm EGTA (unpaired Student's t test). For this and subsequent bar graphs, data represents the mean ± SE, and the n-value is indicated within the bars. B, Middle and right, Zero mm EGTA significantly increases the total area measured from 1 min after stimulation to 11 min post-train stimulus (Mann–Whitney U test) and the time to reach 75% recovery to baseline Ca²⁺ from the peak of the plateau (Mann–Whitney U test).

Removing intracellular EGTA significantly reduced the peak percentage change in intracellular Ca²⁺ during the train stimulus compared with 5 mm EGTA (Fig. 1B, left). This is likely due to a facilitation of Ca²⁺-dependent inactivation of voltage-gated Ca²⁺ channels and an increase in resting Ca²⁺ levels (5 mm EGTA resting 340/380: 0.18 ± 0.004, n = 8; 0 mm EGTA resting 340/380: 0.26 ± 0.01, n = 6; p < 0.003, unpaired Mann–Whitney U test). The area from 1 to 11 min poststimulus train (10 min total) was used to quantify the magnitude and duration of the Ca²⁺ plateau (see Materials and Methods for details). The presence of the Ca²⁺ plateau in 0 mm EGTA was reflected by a significant increase in poststimulus train area from 1 to 11 min (Fig. 1B, middle) and the time to 75% recovery from peak post-train stimulus Ca²⁺ (Fig. 1B, right). Thus, mimicking an endogenous firing pattern evoked distinct rapid and slow periods of cytosolic Ca²⁺ dynamics in the bag cell neurons. As the transduction of a Ca²⁺ signal to activate unique biochemical pathways relies on the temporal and spatial properties of intracellular Ca²⁺, we sought to dissect the Ca²⁺ sources and removal processes that contributed to these Ca²⁺ responses.

Voltage-gated Ca²⁺ influx from a train stimulus is cleared by mitochondrial uptake

Mitochondria are an essential Ca²⁺ removal system in many neurons and neuroendocrine cells, particularly when Ca²⁺ concentrations are substantially higher than at rest (>500 nm) (Herrington et al., 1996). Prior work by our lab indicated a role for mitochondrial Ca²⁺ uptake after a train of action potentials (Geiger and Magoski, 2008). To test whether the mitochondria are key to removal of voltage-gated Ca²⁺ influx, the ability of mitochondria to clear Ca²⁺ was eliminated using FCCP. This protonophore collapses the mitochondrial membrane potential, releases stored Ca²⁺, and prevents subsequent Ca²⁺ uptake into the organelle (Heytler and Prichard, 1962; Babcock et al., 1997). EGTA (5 mm) was included in the pipette solution to eliminate the Ca²⁺ plateau and allow for isolated measurement of voltage-gated Ca²⁺ influx and removal. Post-train stimulus recoveries were well fit with monoexponential decay functions, from which time constants were derived to quantify changes in Ca²⁺ clearance. Percentage recovery at 5 min was measured to quantitate the degree of Ca²⁺ recovery. Any sample size discrepancy between percentage recovery and decay constants for the same dataset was because of the exclusion of poor exponential fits.

Initial observations showed that some neurons treated with FCCP had a reduced peak rise in Ca²⁺ during the train stimulus. This may be because of an increase in use-dependent inactivation of Ca²⁺ currents in the absence of mitochondrial Ca²⁺ clearance. To prevent this from impacting quantification of Ca²⁺ removal, additional FCCP-treated neurons were stimulated with 5 Hz, 1 min train of 175 ms pulses to enhance Ca²⁺ influx and match the peak levels seen in controls. Cells that presented peak Ca²⁺ amplitudes comparable to control were used in measuring the percentage recovery. As such, peak Ca²⁺ influx was not significantly different between DMSO-treated and selected FCCP-treated neurons (DMSO peak % Δ 340/380: 135.4 ± 8.8, n = 10; FCCP peak % Δ 340/380: 162.3 ± 14.5, n = 12; p > 0.05, unpaired Student's t test).

Compared with DMSO-treated cells, neurons stimulated after a 30 min exposure to 20 μm FCCP presented a slower Ca²⁺ recovery (Fig. 2A), as indicated by a significantly larger decay time constant (Fig. 2A, inset). These findings are consistent with the effects of FCCP in other systems, and indicate a role for mitochondrial Ca²⁺ clearance (Thayer and Miller, 1990; Friel and Tsien, 1994; Werth and Thayer, 1994). From the decay of these Ca²⁺ transients, we determined the relative rate of apparent mitochondrial uptake (R_mit) by subtracting the rate of Ca²⁺ removal in FCCP conditions (R_FCCP) from the total Ca²⁺ removal rate (R_total) at corresponding 340/380 ratio values (see Materials and Methods for details). Figure 2B (top) displays the relative cytosolic Ca²⁺ removal rate, normalized to peak rate, against the 340/380 ratio for the representative traces in Figure 2A, along with fitted polynomial functions. Fits from multiple neurons were used to produce averaged R_FCCP, R_total, and R_mit values (Fig. 2B, bottom). This suggests that mitochondrial uptake occurred over a wide range of Ca²⁺ levels, both at rest and at peak values during stimulation, with a corresponding increase in removal rate. In contrast, the nonmitochondrial Ca²⁺ buffer, represented as R_FCCP, had a shallower slope over the same range, indicating the presence of a relatively slow removal mechanism.

Central to the notion that the mitochondria buffer voltage-gated Ca²⁺ influx, is that there is an increase in mitochondrial Ca²⁺ after stimulation. To examine this, FCCP was used to liberate mitochondrial Ca²⁺, with or without a prior train stimulus. Assuming mitochondrial involvement, 100 μm tetraphenylphosphonium (TPP), a blocker of mitochondrial Ca²⁺ exchange in bag cell neurons (Karadjov et al., 1986; Geiger and Magoski, 2008), was applied to both DMSO-treated and FCCP-treated cells, to ensure no mitochondrial Ca²⁺ release followed the excitation. As per prior work suggesting the mitochondria of cultured bag cell neurons contain Ca²⁺ at rest (Jonas et al., 1997; Gardam et al., 2008; Geiger and Magoski, 2008), bath application of 20 μm FCCP increased cytosolic Ca²⁺ within 5 min (n = 6) (Fig. 2C, left). If a train stimulus was delivered before FCCP (n = 6), the Ca²⁺ release signal was significantly increased by ∼40%, consistent with voltage-gated Ca²⁺ influx enhancing the amount of Ca²⁺ stored in the mitochondria (Fig. 2C, right, D).

As mitochondria appear to be essential for Ca²⁺ removal after depolarization, we attempted to saturate Ca²⁺ clearance with and without active mitochondria. After a 5 Hz, 1 min train stimulus in FCCP (n = 5), Ca²⁺ often decayed slowly to a higher level than the prestimulus baseline (Fig. 2E). Application of a second train stimulus from this new baseline elevated Ca²⁺ to similar levels as the first train stimulus; however, the subsequent recovery was severely hindered (Fig. 2E, left). To control for differences between FCCP and control in Ca²⁺ after the first train stimulus, DMSO-treated neurons were held at a potential ranging from −20 to −30 mV immediately after the first train stimulus (n = 6). Because some neurons had smaller Ca²⁺ currents, a range of voltages was used to ensure that all Ca²⁺ plateaus were of comparable size. Voltage-clamping at the depolarized potential produced an elevated Ca²⁺ baseline similar to that seen in FCCP. After the second train stimulus, cells were again held at −80 mV, but unlike in FCCP, DMSO-treated cells recovered rapidly to the Ca²⁺ levels as seen at the start of the experiment (Fig. 2E, right). The ratio between the percentage recovery at 5 min after the first and second train stimulus was used to quantify the degree of buffer saturation in each condition. The percentage recovery ratio was significantly reduced in FCCP-treated compared with DMSO-treated neurons (Fig. 2F).

Contribution of the plasma membrane Ca²⁺ ATPase to the removal of voltage-gated Ca²⁺ influx

Although our results strongly suggest that mitochondria are the predominant buffer for voltage-gated Ca²⁺ influx, a residual contribution from other systems must exist, given the slow recovery even in the presence of FCCP. Most cells use the high-affinity, low-capacity PMCA to extrude Ca²⁺ across the plasma membrane (Sanchez-Armass and Blaustein, 1987; Blaustein and Lederer, 1999; Jeon et al., 2003; Tidow et al., 2012). The role of this pump in removing voltage-gated Ca²⁺ influx was tested with 2 mm extracellular La³⁺, a common PMCA inhibitor (Carafoli, 1991; Herrington et al., 1996; Zenisek and Matthews, 2000). Addition of La³⁺ ∼1 s subsequent to the end of the 5 Hz, 1 min train stimulus (n = 16) slowed the recovery to prestimulus baseline (Fig. 3A, left) compared with control neurons treated with water (n = 17). This manifested as a significantly smaller percentage recovery at 5 min poststimulus in La³⁺-treated neurons (Fig. 3C). Consistent with a minor role of the PMCA in removing Ca²⁺, La³⁺ appeared to have a smaller effect on percentage recovery than FCCP (Fig. 3C). Because La³⁺ was bath applied on the last pulse of the train stimulus, to avoid blocking Ca²⁺ currents, the onset of La³⁺ action was delayed for several seconds thereafter. Thus, post-train stimulus decay time constants were not determined for this experiment, as they would not have been an accurate reflection of PMCA inhibition.

Figure 3.

Removal of voltage-gated Ca²⁺ influx is slowed by inhibition of the PMCA, but is not influenced by acidic stores, ER, or the Na⁺/Ca²⁺-exchanger. A, Left, Blocking PMCA function by the addition of 2 mm extracellular La³⁺ on the last pulse of the 1 min, 5 Hz train stimulus, hinders Ca²⁺ removal (dark trace), whereas H₂O (control) (light trace) applied in the same manner has no effect. A, Middle left, In contrast, the inclusion of Na⁺ (dark trace) rather than TEA (light trace) in the extracellular saline, to permit Na⁺/Ca²⁺ exchanger activity, has no effect. Similarly, treatment with 20 μm CPA (middle right), to inhibit SERCA, or 100 nm bafilomycin A (baf) (right), to prevent uptake by acidic stores, does not alter Ca²⁺ removal after a train stimulus. B, Neurons treated with 20 μm FCCP present significantly larger exponential decay time constants (τ) relative to controls (reproduced from Fig. 2A), whereas cells in the presence of extracellular Na⁺, exposed to CPA, or baf do not have significantly different τ values (all comparisons using unpaired Student's t test). C, FCCP (Mann–Whitney U test) and La³⁺ significantly (Welch corrected unpaired Student's t test) reduce the percentage recovery at 5 min post-train stimulus whereas extracellular Na⁺, CPA, and baf have no effect (all comparisons using unpaired Student's t test).

The plasma membrane Na⁺/Ca²⁺ exchanger trades extracellular Na⁺ for intracellular Ca²⁺ (Blaustein and Lederer, 1999; Kim et al., 2003). In the bag cell neurons, Na⁺/Ca²⁺ exchanger activity can be eliminated by replacing extracellular Na⁺ with TEA (Knox et al., 1996). Our standard recording conditions used extracellular TEA, rather than Na⁺; thus, we tested the effect of supplementing extracellular Na⁺ in lieu of TEA on the rate of voltage-gated Ca²⁺ removal (Fig. 3A). Compared with TEA (n = 6), adding extracellular Na⁺ (n = 7) did not significantly alter the magnitude of the Ca²⁺ rise during stimulation (TEA peak % Δ: 174.9 ± 18.8, n = 6; Na⁺ peak % Δ: 150.5 ± 22.3, n = 7; p > 0.05, unpaired Student's t test), poststimulus Ca²⁺ decay time constant (Fig. 3B) or the percentage recovery at 5 min after peak Ca²⁺ (Fig. 3C).

In addition to the mitochondria, the ER is the other primary intracellular Ca²⁺ store in neurons (Berridge et al., 2000), and has been found to remove voltage-gated Ca²⁺ influx in neurons and neuroendocrine cells (Fierro et al., 1998; Kim et al., 2003). To test this in bag cell neurons, 20 μm CPA, a SERCA inhibitor found to be effective in bag cell neurons (Seidler et al., 1989; Kachoei et al., 2006; Gardam et al., 2008; Geiger and Magoski, 2008), was applied 30 min before stimulation. Post-train stimulus Ca²⁺ kinetics were not affected by the presence of CPA (n = 6) versus control (n = 7) (Fig. 3A). CPA did not alter the peak rise in Ca²⁺ during stimulation (DMSO peak % Δ: 174.3 ± 20.4, n = 7; CPA peak % Δ: 164.8 ± 26.2, n = 6; p > 0.05, unpaired Student's t test), the poststimulus Ca²⁺ decay time constant (Fig. 3B), or the percentage recovery to baseline following stimulation (Fig. 3C).

Finally, FCCP also collapses other stores with proton gradients, including lysosome, endosomes, and secretory vesicles (Goncalves et al., 1999; Christensen et al., 2002). The contribution of these stores to the removal of voltage-gated Ca²⁺ influx was tested by treating cells with bafilomycin A, a H⁺-ATPase inhibitor that prevents the sequestration of Ca²⁺ by acidic stores (Bowman et al., 1988; Goncalves et al., 1999). Our earlier work demonstrated that bafilomycin A causes a slow, steady increase in bag cell neuron cytosolic Ca²⁺, distinct from the response to other liberating agents (Kachoei et al., 2006; Hickey et al., 2010). Pretreatment with 100 nm bafilomycin A (n = 6) did not alter the post-train stimulus Ca²⁺ removal compared with DMSO-treated neurons (n = 6) (Fig. 3A). Bafilomycin did not change the post-stimulus decay time constant (Fig. 3B), the percentage recovery at 5 min after peak Ca²⁺ (Fig. 3C) or the peak Ca²⁺ rise (DMSO peak % Δ: 235.58 ± 18.3, n = 6; bafilomycin A peak % Δ: 233.0 ± 23.7, n = 6; p > 0.05, unpaired Student's t test). Thus, the effect of FCCP on Ca²⁺ removal was due its action on mitochondrial function.

The PMCA, but not the SERCA or Na⁺/Ca²⁺ exchanger, clear somatic Ca²⁺ in the absence of mitochondrial function

The inhibition of a dominant Ca²⁺ clearance system can unveil the activity of other, formerly uninvolved removal mechanisms (Zenisek and Matthews, 2000; Kim et al., 2005). This presumably reflects a compensatory property that ensures normal Ca²⁺ homeostasis. We examined this possibility in bag cell neurons by exploring the contribution of nonmitochondrial clearance mechanisms in the presence of FCCP. Addition of the PMCA inhibitor, La³⁺, to the extracellular solution on the last pulse of the 5 Hz, 1 min train stimulus (n = 7), significantly blunted the recovery from peak compared with FCCP alone (n = 5). In the presence of FCCP, once full PMCA inhibition manifested, little to no recovery occurred and Ca²⁺ remained at a much higher plateau than in FCCP alone (Fig. 4A, left). The percentage recovery at 5 min was significantly reduced under these conditions (Fig. 4C). These data indicates that in the absence of both mitochondrial and PMCA function, voltage-gated Ca²⁺ removal is largely occluded. In contrast, exchanging extracellular Na⁺ (n = 7) for TEA (n = 8), or pretreatment with CPA (control, n = 8; CPA, n = 8) remained ineffective at influencing post-train stimulus Ca²⁺ removal in the absence of mitochondrial function (Fig. 4A, middle, right). Poststimulus decay time constants (Fig. 4B) and percentage recoveries at 5 min (Fig. 4C) were unchanged by including extracellular Na⁺ or pretreatment with CPA.

Figure 4.

In the absence of mitochondrial function, PMCA, but neither SERCA nor the Na⁺/Ca²⁺-exchanger, influence the removal of voltage-gated Ca²⁺ influx. A, Left, Post-stimulus addition of the PMCA inhibitor, La³⁺ (2 mm) (dark trace), in the presence of 20 μm FCCP, strongly attenuates the subsequent recovery relative to FCCP alone (light trace). The inclusion of extracellular Na⁺ (middle) or pretreatment with 20 μm CPA (right) does not influence Ca²⁺ removal in the presence of FCCP. B, For neurons pretreated with FCCP, the mean post-train stimulus recovery τ values are not significantly altered by the inclusion of extracellular Na⁺ or exposure to CPA (unpaired Student's t test and Mann–Whitney U test, respectively). C, Summary of mean percentage recovery at 5 min post-train stimulus. In the presence of FCCP, La³⁺ significantly increases the percentage recovery at 5 min, whereas the application CPA (Mann–Whitney U test) or the inclusion of extracellular Na⁺ remains ineffective (all other comparisons using unpaired Student's t tests).

The EGTA-sensitive Ca²⁺ plateau is caused by mitochondrial Ca²⁺ release

In many neurons, it is common for brief periods of action potential firing to evoke sustained Ca²⁺ release from mitochondria or the ER with similar characteristics as the EGTA-sensitive Ca²⁺ plateau shown in Figure 1 (Gorman and Thomas, 1980; Smith et al., 1983; Neering and McBurney, 1984; Tang and Zucker, 1997; Lee et al., 2007). Prior research from our lab has found that prolonged stimulation of bag cell neurons under sharp electrode recording elicited a CICR plateau that was sensitive to FCCP as well as TPP, an inhibitor of mitochondrial Ca²⁺ exchangers (Geiger and Magoski, 2008). To determine whether the EGTA-sensitive Ca²⁺ plateau we observed under whole-cell conditions was also due to mitochondrial Ca²⁺ release, 100 μm TPP was applied to cells 30 min before stimulation. In the presence of TPP (n = 8), the post-train stimulus response transformed from a slow, large Ca²⁺ plateau under control conditions (n = 6) to a rapid exponential recovery indistinguishable from that seen in 5 mm intracellular EGTA (compare Figs. 5A, 1A). This was apparent from the significantly reduced post-train stimulus area from 1 to 11 min (Fig. 5C) and time to 75% recovery from peak Ca²⁺ (Fig. 5D). TPP had this effect without altering the peak rise in Ca²⁺ during stimulation (control peak % Δ: 127.5 ± 7.4, n = 6; TPP peak % Δ: 127.6 ± 11.5, n = 8; p > 0.05 unpaired Student's t test).

Figure 5.

The EGTA-sensitive Ca²⁺ plateau is caused by mitochondrial Ca²⁺ release. A, A post-stimulus Ca²⁺ plateau is evoked by a 5 Hz, 1 min train of depolarizing steps under voltage clamp. Intracellular EGTA is absent (0 mm) to allow for CICR. The prolonged Ca²⁺ plateau in control conditions (light trace) is substantially smaller in 100 μm the mitochondrial Ca²⁺ exchange blocker, TPP (dark trace). B, Treatment with CPA, to inhibit SERCA and prevent ER Ca²⁺ release, does not affect the post-train stimulus Ca²⁺ plateau (dark trace) when compared with DMSO-treated neurons (light trace). C, D, TPP but not CPA, significantly reduces the 1–11 min area relative to controls (unpaired Student's t test in both cases). TPP also significantly reduces the time to reach 75% recovery of baseline Ca²⁺ from peak CICR (Welch corrected, unpaired Student's t test). In contrast, CPA has no significant effect on the recovery time (unpaired Student's t test). Sample sizes for the time to 75% recovery in CPA are smaller than those used for post-stimulus area measurements as some neurons did not achieve 75% recovery.

To determine whether CICR from the ER contributed to the plateau, CPA was used to deplete the ER of Ca²⁺ before stimulation. Cells treated with 20 μm CPA (n = 7) presented a similar Ca²⁺ plateau magnitude as in control conditions (n = 8) and recovered to prestimulus baseline with a comparable time course (Fig. 5B). CPA did not significantly alter post-train stimulus area from 1 to 11 min (Fig. 5C) and did not affect the time to 75% recovery from peak Ca²⁺ (Fig. 5D). These results indicate that the EGTA-sensitive Ca²⁺ plateau is mitochondrial, but not ER, Ca²⁺ release, and that removal of CICR is independent of the SERCA.

CICR is removed by plasma membrane extrusion via Na⁺/Ca²⁺ exchange and the PMCA

As the Ca²⁺ plateau from mitochondrial release had a different magnitude and kinetics than voltage-gated Ca²⁺ influx during the train stimulus, it is possible that the handling mechanisms responsible for CICR are different from those for voltage-gated Ca²⁺. Full recovery from CICR typically required >10 min; therefore, we presumed that a relatively slow clearance system was responsible for its removal. To examine this, we first tested for the contribution of the Na⁺/Ca²⁺ exchanger by substituting extracellular Na⁺ for TEA. In contrast to its ineffectiveness in removing voltage-gated Ca²⁺ influx, extracellular Na⁺ significantly reduced the time to 75% recovery from peak CICR to baseline (Fig. 6A, left) compared with TEA. However, the post-train stimulus area from 1 to 11 min was not significantly different between Na⁺ and TEA conditions (Fig. 6B). This occurred because the first stage of CICR is marked by the rising Ca²⁺ plateau, which was not significantly different in magnitude between TEA and Na⁺ conditions (TEA external, peak % Δ: 86.3 ± 2.0, n = 7; Na⁺ external, peak % Δ: 79.6 ± 5.0, n = 12; p > 0.05, Mann–Whitney U test). Nevertheless, the area from 11 to 21 min post-train stimulus, where the recovery from peak is most prominent, was significantly smaller in the presence of extracellular Na⁺ (Fig. 6B). Sample size between time to 75% recovery and post-train stimulus area for Na⁺-conditions was different because one cell did not recover to 75% by the end of the recording.

Figure 6.

Mitochondrial Ca²⁺ release is removed by the Na⁺/Ca²⁺ exchanger and the PMCA. A, Left, In the absence of extracellular Na⁺ (TEA, light trace), the mitochondrial Ca²⁺ plateau evoked by a 1 min, 5 Hz train stimulus from −80 mV is prolonged compared with cells bathed in extracellular Na⁺ (dark trace). A, Inset, The presence of extracellular Na⁺ significantly accelerated the time to 75% recovery for CICR (Mann–Whitney U test). A, Middle, In Na⁺-free extracellular saline (TEA), the application of La³⁺ (dark trace) during the onset of CICR occludes recovery. In these experiments, cells are voltage-clamped at −80 mV following the stimulus. A, Right, As a control, a 1 min 5 Hz train stimulus from −80 mV, followed by a prolonged step depolarization evokes a Ca²⁺ plateau similar to CICR (at break, a portion of the Ca²⁺ influx is omitted for clarity to emphasize the Ca²⁺ plateau). The recovery of this Ca²⁺-influx plateau is insensitive to the replacement of extracellular Na⁺ (dark trace) for TEA (light trace). For this experiment, neurons are pretreated with 100 μm TPP to prevent any influence of mitochondrial Ca²⁺ release and are recorded with 0 mm intracellular EGTA. B, The post-train stimulus area from 1 to 11 min during mitochondrial Ca²⁺ release is not significantly different between Na⁺ and TEA conditions (unpaired Student's t test). However, the area from 11 to 21 min is significantly larger in TEA-containing extracellular saline (Mann–Whitney U test). C, In the absence of Na⁺/Ca²⁺ exchanger activity, the application of 2 mm La³⁺ during CICR does not alter the post-train stimulus area from 1 to 11 min (Mann–Whitney U test), but significantly increases the area from 11 to 21 min (Welch corrected Student's t test, respectively). D, Area summary data for the Ca²⁺ plateau evoked by post-train stimulus depolarization at (ranging from −10 to −20 mV). Replacing extracellular Na⁺ with TEA does not significantly alter the post-train stimulus area from time 1–11 min or 11–21 (Student's t test for both).

Even in the presence of TEA, when the Na⁺/Ca²⁺ exchanger was inhibited, CICR recovery still occurred, albeit at a slower rate, indicating involvement of another removal system. To determine whether the PMCA was responsible, 2 mm La³⁺ was applied at the peak of CICR, ∼1 min post-train stimulus, while in the presence of extracellular TEA. PMCA inhibition by La³⁺ halted CICR recovery (n = 7) whereas in control cells (n = 8) Ca²⁺ still returned to baseline (Fig. 6A, middle). As with the Na⁺-replacement experiment, the post-train stimulus area from 1 to 11 min was not significantly different in La³⁺-treated neurons, whereas the area from 11 to 21 min post-train stimulus was significantly enhanced (Fig. 6C).

The activation of Ca²⁺-dependent processes is highly sensitive to magnitude, duration, and frequency of cytosolic Ca²⁺ change (Clapham, 1995; Berridge et al., 2003). Therefore, the source specific involvement of the Na⁺/Ca²⁺ exchanger may be related to the difference between the rapid, large voltage-gated Ca²⁺ influx, and the slow, moderately sized CICR from the mitochondria. To test this, we examined whether a voltage-gated Ca²⁺ influx plateau, that was similar in amplitude and kinetics to CICR, was sensitive to extracellular Na⁺. Neurons were stimulated with a 5 Hz, 1 min train stimulus from −80 mV; however, unlike control conditions, cells were immediately voltage-clamped at a potential ranging from −10 to −20 mV for the remainder of the recording. Holding at a depolarized potential produced a steady Ca²⁺ influx comparable to that elicited during CICR. As with rapid, train stimulus-induced voltage-gated Ca²⁺ influx, the inclusion of extracellular Na⁺ (n = 5) had no apparent effect on the recovery of the slow, persistent voltage-gated Ca²⁺ influx plateau compared with controls (n = 5) (Fig. 6A, right). The post-train stimulus areas from 1 to 11 min, and 11–21 min were not significantly different between TEA and Na⁺ external conditions (Fig. 6D).

A model of bag cell neuron Ca²⁺ dynamics recapitulates EGTA-sensitive CICR

The data concerning bag cell neuron Ca²⁺ removal and release after prolonged stimulation provided the information necessary to create a model of Ca²⁺ dynamics. Previous work from bullfrog sympathetic neurons demonstrated that patterns of Ca²⁺ influx and release can be accounted for by a 3 component model consisting of extracellular, cytosolic, and mitochondrial compartments (Friel and Tsien, 1994; Colegrove et al., 2000b). Therefore, the parameter framework from these models was used in this study. Our three-compartment Ca²⁺ model included an extracellular Ca²⁺ influx source (J_influx), uptake (J_uptake) and release (J_release) by a mitochondrial store, and plasma membrane Ca²⁺ efflux (J_efflux) (Fig. 7A, inset). For simplification, extrusion by the PMCA and Na⁺/Ca²⁺ exchanger was represented by a single component. Estimates of these rates in the bag cell neurons were derived by fitting raw measurements of free intracellular Ca²⁺ (Fig. 7A) (see Materials and Methods). Values derived from individual traces were collected and averaged to produce the model parameters used for the subsequent graphs (Fig. 7B–D). To account for the buffering of Ca²⁺ by EGTA, fura, and endogenous Ca²⁺-binding proteins, rates of free Ca²⁺ removal and extrusion were converted to rates of total Ca²⁺ removal. Thus, the model output is presented as concentrations of total cytosolic Ca²⁺ ([Ca²⁺]_T) (Fig. 7B–D) (see Materials and Methods).

Figure 7.

A model of bag cell neuron Ca²⁺ dynamics demonstrates the sensitivity of mitochondrial Ca²⁺ release to the Ca²⁺ chelating agent, EGTA. A, The inset illustrates the components involved in the 3 compartment model of Ca²⁺. Ca²⁺ influx is represented by J_influx, cytosolic Ca²⁺ removal is mediated by mitochondrial uptake (J_uptake) or plasma membrane efflux (J_efflux), and mitochondrial Ca²⁺ release is denoted as J_release. Collectively, these variables determine the cytosolic and mitochondrial Ca²⁺concentrations (see Materials and Methods). Raw data from a bag cell neuron (light trace) presenting stimulus-induced Ca²⁺ influx, CICR, and recovery is fit by the model (dark trace) to estimate values for free intracellular Ca²⁺ fluxes. Train stimulus-induced Ca²⁺ influx (bottom) is simulated in the model by transiently increasing the Ca²⁺ influx rate constant (k_influx). Directly after reducing k_influx there is a rapid recovery leading to a prolonged Ca²⁺ plateau that readily fits the raw data (root mean square error: 2.3 × 10⁻⁸). Parameters for this fit are (± 95% confidence interval): V_{max, efflux}: 6.2 ± 0.26 nm/s, k_{max, uptake}: 4.45 ± 0.82 (s⁻¹), V_{max, release}: 6.2 ± 1.14 nm/s. Unless otherwise stated, all subsequent model presentations were derived using the average values presented in Table 1. The values in Table 1 reflect parameters measured from changes to free Ca²⁺, before converting to rates of total Ca²⁺. B, For this and subsequent model graphs, Ca²⁺ levels reflect total, rather than free intracellular Ca²⁺ to correct for the presence of exogenous (fura and/or EGTA) and endogenous Ca²⁺ buffers present when estimating rates (see Materials and Methods). Serially reducing the rate constant of mitochondrial Ca²⁺ uptake (k_{max, uptake}) from 10 to 3.3 then 0 (s⁻¹) attenuates the degree of mitochondrial Ca²⁺ uptake (right, light traces), slows the post-stimulus removal, and reduces CICR (left, light traces). C, Left inset, to include EGTA (0.5 mm), the bag cell neuron model of Ca²⁺ dynamics is increased to four components by adding Ca²⁺ removal by an exogenous Ca²⁺ binding agent. The clearance of Ca²⁺ by EGTA is determined by its forward (k_on) and reverse (k_off) rate constants, respectively. In the absence of EGTA, evoking Ca²⁺ influx causes a rise in Ca²⁺ and a subsequent Ca²⁺ plateau (left, dark trace). Under these conditions, mitochondrial Ca²⁺ increases then falls as Ca²⁺ is released into the cytosol (right, dark trace). The cytosolic Ca²⁺ response in the presence of EGTA slightly reduces peak Ca²⁺ influx magnitude and eliminates mitochondrial CICR (left, light traces). EGTA also partially attenuates the magnitude of mitochondrial Ca²⁺ influx after stimulation (right, light traces). D, In the presence of an EGTA component (0.5 mm), increasing the mitochondrial uptake rate constant (k_{max, uptake}) from 10 to 40 then 160 (s⁻¹), potentiates the degree of mitochondrial Ca²⁺ loading (right, light traces), speeds the rate of cytosolic Ca²⁺ recovery after influx, and produces very limited CICR (left, light traces).

Transiently increasing the plasma membrane influx rate constant (k_influx), to replicate voltage-gated Ca²⁺ influx, produced a fast increase in cytosolic Ca²⁺, followed by a rapid recovery and a prolonged cytosolic Ca²⁺ plateau with a similar time course as that seen in actual bag cell neurons (Fig. 7B, left). Over the same time period, mitochondrial Ca²⁺ rapidly increased and decayed to prestimulus levels (Fig. 7B, right), corresponding to the uptake and release phases of cytosolic Ca²⁺, respectively. To demonstrate the necessity of mitochondrial Ca²⁺ uptake and release, J_uptake was tempered by serially reducing the uptake rate constant (k_max,uptake). This produced a progressively smaller Ca²⁺ increase in the mitochondria, slowed the rate of cytosolic Ca²⁺ removal from peak influx, and reduced the CICR magnitude (Fig. 7B, left). These results are qualitatively similar to those seen in ours and other experiments when mitochondrial uptake is eliminated (Thayer and Miller, 1990; Friel and Tsien, 1994; Colegrove et al., 2000a; Geiger and Magoski, 2008).

Having established parameters that replicate our experiments, we sought to address the sensitivity of CICR to intracellular EGTA. The effects of EGTA on CICR could be due to competition with mitochondrial uptake for Ca²⁺ influx, causing a reduction in mitochondrial loading and subsequent release, or by EGTA binding the Ca²⁺ as it is extruded. To discern between these possibilities, a fourth component was added to the model, representing the buffering of Ca²⁺ by EGTA, which was determined by its forward (k_on) and reverse reaction rate constants (k_off; see Materials and Methods) (Nowycky and Pinter, 1993; Naraghi, 1997). Inclusion of an EGTA component (0.5 mm) in the compartment model caused a small reduction in the peak cytosolic Ca²⁺ rise during influx, while eliminating CICR (Fig. 7C, left). This in silico result is strikingly comparable to that observed in vitro (Fig. 1). In the EGTA-containing conditions, peak mitochondrial Ca²⁺ levels were slightly reduced following cytosolic Ca²⁺ influx, and consequently, decayed to prestimulus Ca²⁺ levels at a faster time course than in the absence of EGTA (Fig. 7C, right).

To determine whether increasing the degree of mitochondrial Ca²⁺ loading could rescue CICR in the presence of the EGTA component (0.5 mm), mitochondrial uptake was serially enhanced. Stepping the k_{max, uptake} from our standard value of 10 to 40 then 160 (s⁻¹) increased mitochondrial Ca²⁺ to concentrations as large or higher than those seen in the absence of EGTA (compare Fig. 7C, right, D, right). Despite increasing the degree of mitochondrial Ca²⁺ available for release by enhancing mitochondrial uptake, CICR was only weakly rescued in the presence of EGTA. Thus, the sensitivity of CICR to EGTA is largely attributable to competition for Ca²⁺ released from the mitochondria.

Store-operated Ca²⁺ influx refills the ER Ca²⁺ store and is primarily cleared by the SERCA

In addition to voltage-gated Ca²⁺ influx and CICR, a third Ca²⁺ source prominent in the bag cell neurons is store-operated Ca²⁺ influx (Kachoei et al., 2006). It is well established that signaling cascades release ER Ca²⁺ through ryanodine- and inositol triphospate (IP₃)-receptors in the bag cell neurons (Fink et al., 1988; Fisher et al., 1994; Geiger and Magoski, 2008). Here we evoked store-operated influx to determine whether a third Ca²⁺ source uses distinct removal mechanisms.

To measure store-operated Ca²⁺ influx, bag cell neurons were fura-filled by sharp-electrode pressure injection rather than recorded under whole-cell voltage clamp. This was possible because ER depletion and store-operated Ca²⁺ influx do not significantly alter the membrane potential, and therefore do not require voltage clamp (Kachoei et al., 2006). Store-operated influx was activated by applying 20 μm CPA in Ca²⁺-free medium to deplete the ER by inhibiting the SERCA (Seidler et al., 1989). After depletion and washout of CPA, the addition of Ca²⁺ back to the bath caused a rapid rise in intracellular Ca²⁺ as a result of Ca²⁺ influx through depletion-activated store-operated channels (Fig. 8A). A second application of CPA after influx and recovery elicited another rise in intracellular Ca²⁺ (Fig. 8A); however, its magnitude was significantly smaller than the first (Fig. 8B). In contrast, when store-operated influx recovery occurred in the presence of CPA (see below), a second application of CPA had no apparent effect on resting Ca²⁺ (data not shown).

Figure 8.

The store-operated Ca²⁺ influx pathway is cleared by the SERCA to replete the ER. A, Addition of 20 μm CPA depletes ER Ca²⁺ in a cultured bag cell neuron pressure-injected with fura-PE3. Because neurons are not recorded under voltage clamp, a normal Na⁺-containing and K⁺-containing external solution is used. After the first depletion, CPA is washed out using bath exchange (at break); upon recording resumption, the addition of extracellular Ca²⁺ results in a rapid elevation of intracellular Ca²⁺. Delivering CPA a second time again evokes a Ca²⁺ rise, albeit smaller than the first, indicating repletion of the CPA-sensitive store. B, The magnitude of Ca²⁺ release to the second CPA exposure, after store-operated influx, is significantly smaller than the response elicited during the first CPA-induced depletion (unpaired Student's t test). C, Left, Washout of CPA, before addition of extracellular Ca²⁺, speeds the recovery of influx back to baseline. The removal of store-operated Ca²⁺ recovers at a slower rate in the presence of CPA (dark trace). Both neurons previously depleted in Ca²⁺-free ASW with 20 μm CPA. C, Right, Ca²⁺ influx, similar in size to that evoked by store-operated influx, caused by a short 5 Hz train of depolarizing stimuli from −80 to 0 mV in the presence (dark trace) and absence (light trace) of CPA. CPA has no effect on the speed of recovery from the short stimulus. D, Left, Mean percentage change in 340/380 (left) indicates that the increase in cytosolic Ca²⁺ during store-operated Ca²⁺ influx and the short train stimulus Ca²⁺ influx are not significantly different within and between conditions (ANOVA). D, Middle, CPA significantly increases the mean store-operated Ca²⁺ influx decay time constant (Welch corrected unpaired Student's t test) although having no effect on the mean τ of the similarly sized train stimulus-induced Ca²⁺ influx (unpaired Student's t test). D, Right, The percentage recovery at 5 min (right) after the application of extracellular Ca²⁺ is significantly shorter with prior CPA washout (Welch corrected unpaired Student's t test). However, the percentage recovery at 5 min after the short train stimulus Ca²⁺ influx is not significantly altered by CPA.

These results indicate that store-operated Ca²⁺ influx refills the ER via the SERCA. To determine the degree of Ca²⁺ removal by the SERCA, the recovery from peak store-operated Ca²⁺ influx was measured in the presence and absence of CPA. After ER depletion, washout of CPA and addition of extracellular Ca²⁺ evoked a transient rise in Ca²⁺ followed by a monoexponential decay, typically reaching pre-influx baseline within 10 min (Fig. 8C, left). If CPA remained in the bath after depletion (n = 17), store-operated influx was similar in size compared with controls (n = 17); however, Ca²⁺ recovery slowed dramatically, often requiring well over 15 min for full recovery (Fig. 8C, left). Furthermore, in CPA, following a brief initial recovery, Ca²⁺ levels often decayed to a new baseline at a Ca²⁺ concentration higher than pre-influx levels.

Our previous data established that the ER was ineffective at removing rapid voltage-gated Ca²⁺ influx. As the activity of the SERCA pump was demonstrated to be a relatively slow during store-operated influx, the effect of CPA on the recovery of a large Ca²⁺ influx may have gone undetected. Therefore, we produced a voltage-gated Ca²⁺ influx, sized-matched to the average store-operated influx, and tested its sensitivity to CPA (control n = 10, CPA n = 10). A brief (1.5–3 s) 5 Hz train stimulus from −80 to 0 mV applied to fura-loaded bag cell neurons under whole-cell voltage clamp elicited response magnitudes that were not significantly different than those during store-operated influx (Fig. 8D, left). Despite having the same absolute magnitude as store-operated influx, voltage-gated Ca²⁺ influx had a dramatically faster exponential recovery (Fig. 8D, middle). Consistent with this, the handling of voltage-gated Ca²⁺ influx showed no sensitivity to CPA, and rapidly declined to prestimulus levels within 2 min (Fig. 8C, right, D, middle, right).

Role for the mitochondria, but not the PMCA or Na⁺/Ca²⁺ exchanger, in store-operated Ca²⁺ influx

In other systems, mitochondria have been shown to be functionally coupled to store-operated influx (Gilabert and Parekh, 2000; Glitsch et al., 2002; Parekh and Putney, 2005; Naghdi et al., 2010). Therefore, influx was measured after 20 min pretreatment with 20 μm FCCP, to collapse the mitochondrial membrane potential. Compared with controls (n = 8), FCCP-treated neurons (n = 9) had a significantly reduced peak percentage change in response to bath application of Ca²⁺ after CPA-induced depletion (Fig. 9A, left, B). Furthermore, exposure to FCCP reduced the percentage of cells presenting a measurable influx signal. In the DMSO-treated group, 8 of 9 cells showed a response, whereas after FCCP, influx was observed only in 9 of 24 neurons (p < 0.02, Fisher's exact test). The contribution of acidic stores to the removal of store-operated influx was tested by exposing cells with 100 nm bafilomycin A (Fig. 9A, right). Incubation in bafilomycin A (n = 20) before the addition of extracellular Ca²⁺ did not change the peak store-operated influx (Fig. 9B), Ca²⁺ decay time constant (Fig. 9C), or the percentage recovery at 5 min post-peak Ca²⁺ compared with control (n = 16) (Fig. 9D).

Figure 9.

Store-operated Ca²⁺ influx is reduced in magnitude by FCCP but is not influenced by acidic stores, the PMCA, or the Na⁺/Ca²⁺ exchanger. A, Left, Neurons exposed to 20 μm FCCP before the addition of extracellular Ca²⁺ (dark trace) causes a reduction in store-operated influx relative to control (light trace). A, Right, Exposure to 100 nm bafilomycin A (baf) before the addition of extracellular Ca²⁺ (dark trace) does not influence the size or recovery of store-operated Ca²⁺ influx. Neurons were previously depleted in Ca²⁺-free ASW with 20 μm CPA. B, Treatment with FCCP significantly reduces the peak Ca²⁺ influx after the addition of extracellular Ca²⁺ (left), whereas bafilomycin has no effect (both comparisons using Mann–Whitney U test). C, Summary data demonstrating that the mean decay time constant (τ) is significantly larger in CPA (Welch corrected unpaired Student's t test), but not in bafilomycin (unpaired Student's t test), 20 μm carboxyeosin (CE) (Welch corrected Student's t test), or where extracellular Na⁺ is replaced with NMDG (Mann–Whitney U test). D, CPA pretreatment significantly reduces the percentage recovery from peak Ca²⁺, whereas incubation with bafilomycin, carboxyeosin, or extracellular NMDG does not (all comparisons using unpaired Student's t test).

Contributions from the Na⁺/Ca²⁺ exchanger and the PMCA were examined by including extracellular NMDG in lieu of Na⁺ and 20 μm carboxyeosin, respectively. We used NMDG to inhibit the Na⁺/Ca²⁺ exchanger (Blaustein and Lederer, 1999; Zhang et al., 2004), rather than TEA, because TEA blocks K⁺ currents and would depolarize bag cell neurons in non-voltage-clamped conditions (Hagiwara and Saito, 1959; Kaczmarek and Strumwasser, 1984). Carboxyeosin, a different PMCA inhibitor (Shmigol et al., 1998), was used because the onset time of store-operated influx varied among individual cells during a recording, making proper timing of La³⁺ application impossible. Again, SERCA function was inhibited by 20 μm CPA following prior ER depletion. Blocking the Na⁺/Ca²⁺ exchanger (Na⁺ n = 12; NMDG n = 7) or the PMCA (control n = 12; carboxyeosin n = 16) did not significantly alter the decay time constants (Fig. 9C) or the percentage recovery at 5 min (Fig. 9D) despite the ongoing presence of CPA.

Discussion

Although neuron-specific or compartment-specific expression of Ca²⁺ clearance systems has been reported, there is sparse evidence of differential handling mechanisms between Ca²⁺ sources. White and Reynolds (1995) found that a glutamate-evoked Ca²⁺ response is more sensitive to the disruption of the Na⁺/Ca²⁺ exchanger and mitochondria than voltage-gated Ca²⁺ influx. Conversely, in mouse taste receptor cells voltage-gated Ca²⁺ influx, but not Ca²⁺ release, is Na⁺/Ca²⁺ exchanger sensitive (Szebenyi et al., 2010). This work is the first to fully characterize the disparate contribution of handling mechanisms to multiple Ca²⁺ sources in a single neuronal species. We demonstrate: (1) voltage-gated Ca²⁺ influx is primarily removed by the mitochondria with a secondary contribution from the PMCA; (2) CICR arises from subsequent mitochondrial Ca²⁺ release, which is then handled by the Na⁺/Ca²⁺ exchanger and the PMCA; and (3) store-operated Ca²⁺ influx is sequestered via the SERCA (Fig. 10). This is profound, given that each Ca²⁺ source provides unique contributions to neuronal function.

Figure 10.

Ca²⁺ from multiple sources is cleared by distinct sets of Ca²⁺ removal systems in the bag cell neurons. A conceptual model of Ca²⁺ dynamics during an afterdischarge based on the current study and prior work by our laboratory and others (Fink et al., 1988; Fisher et al., 1994; Michel and Wayne, 2002; Kachoei et al., 2006; Geiger and Magoski, 2008). Top, right inset, A sample trace of intracellular Ca²⁺ presumed to reflect afterdischarge dynamics in vivo. Numbers correspond to a series of chronological events depicted in the main illustration: 1, During the fast phase of the afterdischarge there is a large Ca²⁺ influx through action potential-evoked voltage-gated Ca²⁺ channels (VGCC). 2, This early Ca²⁺ influx is predominantly removed by rapid mitochondrial uptake, with ancillary assistance from the PMCA. 3, Ca²⁺ accumulates in the mitochondria, after which it is slowly extruded into the cytosol through a TPP-sensitive Ca²⁺-exchanger. Additionally, second messenger cascades are activated that initiate Ca²⁺ liberation from the ER through IP₃ and ryanodine receptors. 4, The release of Ca²⁺ from intracellular stores causes a prolonged rise in cytosolic Ca²⁺ that is removed by the Na⁺/Ca²⁺ exchanger and the PMCA. 5, Last, sustained Ca²⁺ release during the afterdischarge depletes the ER and initiates store-operated Ca²⁺ influx: a distinct Ca²⁺ source that is removed by the SERCA into the ER lumen.

Voltage-gated Ca²⁺ entry is the primary means to increase cytosolic Ca²⁺ (Catterall and Few, 2008). In the bag cell neurons, this gates Ca²⁺-dependent cation channels that promote the afterdischarge (Whim and Kaczmarek, 1998; Lupinsky and Magoski, 2006; Hung and Magoski, 2007). We show that voltage-gated Ca²⁺ influx is handled by the mitochondria, similar to some neurons and neuroendocrine cells (Werth and Thayer, 1994; Kaftan et al., 2000). However, unlike chromaffin cells, where mitochondrial uptake is engaged only at high Ca²⁺ (Herrington et al., 1996), bag cell neuron mitochondria function near resting and peak Ca²⁺ levels. Neuroendocrine cells provide hormones to orchestrate fundamental behaviors, such as feeding, drinking, and reproduction (Arnauld et al., 1974; Lincoln and Wakerley, 1974; Kawakami et al., 1982). Considering the importance of energy status to these behaviors, it is unsurprising that mitochondria are the first recipients of Ca²⁺ during bag cell neuron excitation. Essentially, mitochondria are situated to act as gatekeepers of the afterdischarge and reproductive status.

In the bag cell neurons, a 1 min stimulus evoked CICR that is sensitive to TPP or intracellular EGTA. A compartment model of bag cell neuron Ca²⁺ suggests that the effect of EGTA on CICR was partially due to a reduction in mitochondrial Ca²⁺ loading, but mainly the result of EGTA binding extruded mitochondrial Ca²⁺. Likely, this is possible because of slow Ca²⁺ release versus fast uptake across the mitochondria. Similarly, maximal uptake in sympathetic neuron mitochondria approaches ∼120 nm/s, whereas release is far slower (∼35 nm/s) (Colegrove et al., 2000a). Such kinetics render mitochondrial Ca²⁺ release sensitive to competitive cytoplasmic buffers. CICR transduces short-lived events into protracted Ca²⁺ signals, which impact biochemical pathways more effectively than the initial response alone. For example, CICR promotes posttetanic transmitter release at mouse neuromuscular junctions and hippocampal synapses (Garcia-Chacon et al., 2006; Lee et al., 2007), and prolongs Ca²⁺-dependent transcription in dorsal root ganglion neurons (Kim and Usachev, 2009). During an afterdischarge, CICR may serve analogously to sustain hormone release. Indeed, some ELH secretion is independent of extracellular Ca²⁺, implicating the involvement of Ca²⁺ stores (Wayne et al., 1998; Michel and Wayne, 2002). The prolonged duration of bag cell neuron CICR reflects a combination of persistent Ca²⁺ release, and removal by the PMCA and Na⁺/Ca²⁺ exchanger, providing an efficient form of recycling Ca²⁺ to the extracellular space. These extrusion mechanisms, which are slow (∼4 nm/s), may shape CICR but not necessarily dampen its effectiveness. The mitochondria cope with Ca²⁺ influx and transduce it into a lengthy release that is extruded without saturating the maximal rate of plasma membrane transport.

Store-operated Ca²⁺ influx was exclusively handled by the SERCA. Accordingly, this pump refilled the ER, consistent with similar roles of store-operated Ca²⁺ influx in other cells (Hoth and Penner, 1992; Putney, 2003). In the bag cell neurons, ER Ca²⁺ release occurs via IP₃ and ryanodine receptors, and can activate a BK-like K⁺ channel and cation current (Fink et al., 1988; Knox et al., 1996) and may contribute to ELH secretion (Michel and Wayne, 2002). Activation of store-operated influx requires ER depletion. As such, it likely presents a delayed activation, once the majority of Ca²⁺ liberation has occurred (Fig. 10), and could sustain processes activated by ER Ca²⁺. Store-operated influx was also reduced when mitochondrial function was absent. This is also seen in T-lymphocytes and retinoblastoma-1 cells, where mitochondrial Ca²⁺ sequestration prevents Ca²⁺-dependent inhibition of store-operated channels (Glitsch et al., 2002; Naghdi et al., 2010). In the bag cell neurons, although neither the Na⁺/Ca²⁺ exchanger nor the PMCA were involved, there was a residual recovery from peak store-operated influx when the SERCA was inhibited. Therefore, a similar explanation involving mitochondrial Ca²⁺ clearance is plausible.

The specificity between Ca²⁺ sources and clearance mechanisms seen in this study could be attributed to the duration of Ca²⁺ exposure, the affinity of the Ca²⁺ handling systems, and/or the distribution of the buffers and influx sources (Gabso et al., 1997; Berridge et al., 2003). The proportion of removal by each handling mechanism depends on influx magnitude. For example, the PMCA and SERCA are reported to have high Ca²⁺-binding affinities and remove small Ca²⁺ loads, whereas the low affinity Na⁺/Ca²⁺ exchanger and mitochondria are engaged only by high intracellular Ca²⁺ (Herrington et al., 1996; Blaustein and Lederer, 1999; Berridge et al., 2003). Such is not the case here, where voltage-gated Ca²⁺ influx of similar size as store-operated influx or CICR does not engage the Na⁺/Ca²⁺ exchanger or SERCA. This is also confirmed by our prior work demonstrating the CPA-insensitivity of a small voltage-gated Ca²⁺ influx elicited in sharp-electrode recorded bag cell neurons (Geiger and Magoski, 2008). It is doubtful that differences in Ca²⁺ signal kinetics contributed to our results. A prolonged poststimulus Ca²⁺ influx showed no sensitivity to the Na⁺/Ca²⁺ exchanger, despite demonstrating a similar duration as CICR. Furthermore, CICR occurred with a duration and magnitude comparable to store-operated Ca²⁺ influx, and yet presented a different involvement of the SERCA, PMCA, and Na⁺/Ca²⁺ exchanger. Thus, the differential control we observe seems attributable to variation in the location of Ca²⁺ sources and/or handling systems. Indeed, Ca²⁺ channel types, including those in the bag cell neurons, can occupy discrete areas within the soma (White and Kaczmarek, 1997). Furthermore, in cortical neurons and astrocytes, the Na⁺/Ca²⁺ exchanger parallels the position of intracellular organelles and can occupy distinct regions relative to the PMCA (Juhaszova et al., 1996, 2000). Organelles can also occupy unique cellular loci. In bullfrog sympathetic neurons, SERCA clusters near the plasma membrane, whereas the mitochondria form a centralized inner ring (McDonough et al., 2000). Preferential localization can have functional implications, such as in cervical ganglion neurons, where Ca²⁺ handling is biased toward a given source, by the ER and mitochondria associating with Ca_v2 but not Ca_v1 channels (Wheeler et al., 2012).

Unlike in the calyx of Held and retinal bipolar cells (Zenisek and Mathews, 2000; Kim et al., 2005), we find that, in the absence of a dominant removal system, the contribution of an otherwise uninvolved clearance mechanism is not enhanced. Compensation likely occurs because Ca²⁺ gradients disperse after influx (Hua et al., 1993), diffusing to buffers that are displaced from the source. Therefore, even when considering a heterogeneous distribution of removal systems, our results are surprising. Potentially, Ca²⁺ signals are compartmentalized by diffusional barriers such as partitions formed by membrane junctions (e.g., ER-plasma membrane) (Carrasco and Meyer, 2011) or “Ca²⁺-sponging” by immobile Ca²⁺-binding proteins (Nowycky and Pinter, 1993; Weiss et al., 2012), which are found in other Aplysia neurons. Estimates of the Ca²⁺-diffusion coefficient in cultured Aplysia neurons (≤16 μm²/s) are similar to those of specialized Ca²⁺ signaling structures, such as dendrites (10–50 μm²/s) or photoreceptor outer segments (15 μm²/s) (Gabso et al., 1997; Murthy et al., 2000; Nakatani et al., 2002). Therefore, auxiliary systems may have contributed to removal, but their effects were slow or delayed.

Although differential clearance between sources may reflect an efficient form of Ca²⁺ homeostasis or Ca²⁺-metabolism coupling (Rizzuto et al., 1998; Chouhan et al., 2012), a more intriguing possibility is that it shapes Ca²⁺ signals to determine which Ca²⁺-dependent pathway is activated. In the aforementioned superior cervical ganglion neurons, the mitochondria and ER preferentially remove Ca²⁺ influx from Ca_v2, but not Ca_v1 channels, permitting Ca_v1-specific activation of transcription (Wheeler et al., 2012). For bag cell neurons, Ca²⁺ entry through a cation channel, but not voltage-gated Ca²⁺ influx or Ca²⁺ release, induces a period of refractoriness to further stimulation (Magoski et al., 2000). Furthermore, following the afterdischarge, there is a Ca²⁺-influx-dependent increase in ELH synthesis, which could be brought about by differences in the clearance of Ca²⁺ from distinct sources (Wayne et al., 2004). Differential handling between Ca²⁺ sources could underlie the ability of discrete Ca²⁺ sources to activate specific downstream targets. Our work reinforces the shifting view of Ca²⁺ buffers as passive determinants of intracellular Ca²⁺ to dynamic regulators that can control excitability, secretion, synaptic plasticity, and gene expression (Krizaj and Copenhagen, 1998; Holthoff et al., 2002; Kim and Usachev, 2009; Simons et al., 2009).