Differential Calcium Dependence of Axonal Versus Somatodendritic Dopamine Release, with Characteristics of Both in the Ventral Tegmental Area

Midbrain dopamine (DA) neurons in the substantia nigra pars compacta (SNc) and ventral tegmental area (VTA) exhibit somatodendritic release of DA. Previous studies indicate a difference between the Ca2+ dependence of somatodendritic DA release in the SNc and that of axonal DA release in dorsal striatum. Here, we evaluated the Ca2+ dependence of DA release in the VTA and nucleus accumbens (NAc) shell for comparison with that in the SNc and dorsal striatum. Release of DA was elicited by single-pulse stimulation in guinea-pig brain slices and monitored with subsecond resolution using carbon-fiber microelectrodes and fast-scan cyclic voltammetry. In dorsal striatum and NAc, DA release was not detectable at extracellular Ca2+ concentrations ([Ca2+]o) below 1 mM; however, a progressive increase in evoked extracellular DA concentration ([DA]o) was seen with [Ca2+]o ≥ 1.5 mM. By contrast, in SNc and VTA, robust increases in [DA]o could be elicited in 0.25 mM [Ca2+]o that were ∼60% of those seen in 1.5 mM [Ca2+]o. In SNc, a plateau in single-pulse evoked [DA]o was seen at [Ca2+]o ≥ 1.5 mM, mirroring the release plateau reported previously for pulse-train stimulation in SNc. In VTA, however, evoked [DA]o increased progressively throughout the range of [Ca2+]o tested (up to 3.0 mM). These functional data are consistent with the microanatomy of the VTA, which includes DA axon collaterals as well as DA somata and dendrites. Differences between axonal and somatodendritic release data were quantified using Hill analysis, which showed that the Ca2+ dependence of axonal DA release is low affinity with high Ca2+ cooperativity, whereas somatodendritic release is high affinity with low cooperativity. Moreover, this analysis revealed the dual nature of DA release in the VTA, with both somatodendritic and axonal contributions.

Previous comparison of somatodendritic DA release in the SNc vs. axonal DA release in dorsal striatum revealed a strikingly lower Ca 2+ dependence of somatodendritic DA release (Chen and Rice, 2001). That study had two limitations, however: first, the low Ca 2+ levels tested were undefined; second, DA release was evoked by pulse-train stimulation, which elicits the concurrent release of glutamate and GABA that can modulate DA release in both the dorsal striatum and SNc (Chen and Rice, 2002;Avshalumov et al., 2003;Patel et al., 2009). Moreover, the Ca 2+ dependence of DA release in NAc or VTA was not examined in that study.
In the present work, we used single stimulation pulses, which mimic single action potentials, to evoke DA release that is free from modulation by concurrently released glutamate and GABA (Chen et al., 2006). Release was examined in SNc, VTA, dorsal striatum, and NAc shell in a series of defined [Ca 2+ ] o . With this refined paradigm, we demonstrate the validity of previous work IntroductIon Brain dopamine (DA) arises primarily from two midbrain cell groups, the substantia nigra pars compacta (SNc; A9) and the ventral tegmental area (VTA; A10; Dahlström and Fuxe, 1964), both of which project ipsilaterally via the median forebrain bundle to forebrain regions (Ungerstedt, 1971;Fallon and Moore, 1978). Dopaminergic neurons of the SNc project to dorsal striatum and facilitate movement mediated by the basal ganglia, whereas dopaminergic neurons of the VTA project to the nucleus accumbens (NAc), prefrontal cortex, and other mesolimbic structures that underlie emotion, reward, and cognition.
To assess the Ca 2+ dependence of DA release as a function of defined [Ca 2+ ] o , Ca 2+ added to normal aCSF was titrated with the Ca 2+ chelator EGTA, according to Bers (1982), with all media containing 1 mM EGTA. In all experiments, D 2 autoreceptors were blocked by sulpiride (1 μM) and the DA transporter (DAT) inhibited by GBR-12909 (2 μM) to eliminate the influence of autoreceptor-mediated inhibition and DA uptake (Chen and Rice, 2001), both of which are more pronounced in striatum than SNc Cragg et al., 1997a).

Fast-scan cyclIc voltaMMetry and experIMental desIgn
Evoked DA release was monitored in real time using fast-scan cyclic voltammetry (FCV) with carbon-fiber electrodes made from 7 μm carbon fibers (type HM, unsized, Courtaulds), which were spark-etched to a tip diameter of 2-4 μm (MPB Electrodes; St. Bartholomew's and the Royal London School of Medicine and Dentistry, University of London, UK). Instrumentation for FCV was a Millar Voltammeter (available from J. Millar, St. Bartholomew's and the Royal London School of Medicine and Dentistry), with data acquisition controlled by Clampex 7.0 software (Molecular Devices, Foster City, CA), which imported voltammograms to a PC via a DigiData 1200B A/D board (Molecular Devices). Scan rate for FCV was 800 V/s, with a sampling interval of 100 ms; scan range was −0.7 to +1.3 V (vs. Ag/AgCl). Because the sensitivity of carbon-fiber microelectrodes can change with changes in divalent cation concentration (Kume-Kick and Rice, 1998;Chen et al., 1999), electrodes were calibrated in the recording chamber at 32°C after each experiment with DA in aCSF with each Ca 2+ concentration tested. The DA calibration factor for each Ca 2+ concentration was used to calculate evoked [DA] o in that medium.
Single-pulse stimulation was used to evoke DA release. Although the stimulus pulse would be expected to elicit release of other transmitters, initial DA release is unaffected by subsequent receptor activation by these transmitters. Previous studies have confirmed the independence of single-pulse evoked [DA] o from regulation by glutamate and GABA by the lack of effect on signal amplitude and time course in striatum and SNc by a cocktail of receptor antagonists: AP5 (NMDA), GYKI-52466 (AMPA), picrotoxin (GABA A ), and saclofen (GABA B ) (Chen et al., 2006). Indeed, evaluation of the effect of a concurrently released transmitter on evoked DA release requires multiple-pulse stimulation (e.g., Chen and Rice, 2002;Avshalumov et al., 2003). The lack of effect of receptor antagonists on single-pulse evoked [DA] o also implies the absence of tonic glutamatergic or GABAergic regulation of DA in our slice preparations. Consequently, these antagonists were not added to the aCSF in the present studies. Stimulus pulse duration was 1 ms with amplitudes of 0.7-1.0 mA. In dorsal striatum and NAc, consistent evoked [DA] o can be elicited with repetitive local stimulation (Bull et al., 1990;Patel et al., 1995;Chen and Rice, 2001). Here, axonal DA release in dorsal striatum and NAc shell was evoked at 10-min intervals; the third of three consistent evoked increases in [DA] o was included in the data average for each [Ca 2+ ] o . In SNc and VTA, maximal release is often seen with the first stimulus, then decreases with repetition (Rice et al., 1997). Thus, DA release obtained from two sites in medial SNc and in VTA were averaged for inclusion in the data set for a given [ 2+ ] o and that in 2.0 mM (112 ± 13%; n = 7), 2.4 mM (118 ± 7%; n = 10), or 3.0 mM [Ca 2+ ] o (121 ± 13%; n = 6; p > 0.05 for each vs. control; Figure 1D).
In   The Ca 2+ -dependence data from each region was evaluated further using a three-parameter Hill analysis, in which the slope determined indicates the power of the Ca 2+ dependence (Figure 3).    (Figure 1) paralleled those seen with pulse-train stimulation reported previously (Chen and Rice, 2001). Given the absence of regulation of single-pulse evoked release by concurrently released glutamate and GABA acting at ionotropic receptors (Chen et al., 2006), this indicates that the basic patterns are not influenced significantly by the Ca 2+ -dependence of other transmitters. This is important because somatodendritic DA release in the SNc can be strongly inhibited by GABAergic input (Chen and Rice, 2002) (Chen and Rice, 2001). However, the new single-pulse DA release data in the present report demonstrate that the predominant contributing factor to the plateau is the Ca 2+ dependence of the DA release process itself.
Fitting the normalized points for dorsal striatum to Hill's function (Schneggenburger and Neher, 2000) gave a slope of 3.2, which indicates an approximately third-power dependence on [Ca 2+ ] o ( Figure 3A). By contrast, Hill analysis of the Ca 2+ dependence of DA release in the SNc gave a slope of only 1.6 ( Figure 3A), confirming a more shallow dependence on [Ca 2+ ] o for somatodendritic DA release in the SNc than for axonal DA release in dorsal striatum. The slope for the NAc shell was 4.0 ( Figure 3B), which did not differ significantly from that determined for dorsal striatum (p > 0.05). Differing from all these regions, however, two distinct Hill fits were required for the VTA: the slope for evoked [DA] (Figure 3).

dIscussIon
We determined the Ca 2+ dependence of single-pulse axonal and somatodendritic DA release in the nigrostriatal and mesolimbic pathways using a defined series of [Ca 2+ ] o in brain slices. We show that DA release in the SNc and VTA has a lower [Ca 2+ ] o threshold and exhibits lower Ca 2+ cooperativity than seen with axonal DA release in the dorsal striatum and NAc shell. Moreover, somato- ence of DA release between dorsal striatum and either SNc or VTA was observed (Ford et al., 2010). Given that both sets of studies used similar methods, the simplest explanation for the discrepancy in SNc responsiveness may be a species difference. Mouse (and rat) SNc show a marked contribution from axonal serotonin (5-HT) input to the voltammetric signal (Cragg et al., 1997b;John et al., 2006;Ford et al., 2010). However, in the guinea-pig SNc, pure DA release is observed (Cragg et al., 1997b;Rice et al., 1997). Thus, the axonal-like Ca 2+ dependence of the release signal in mouse SNc (Ford et al., 2010) is likely to include a significant component of axonal 5-HT release, which could mask the [DA] o plateau reported here. Contributions from 5-HT would be amplified by the higher sensitivity of carbon-fiber electrodes for 5-HT vs. DA (O'Connor and Kruk, 1991;Patel et al., 1992). Interestingly, the Ca 2+ dependence reported for mouse VTA over a range of 0.5-2.5 mM [Ca 2+ ] o (Ford et al., 2010) is similar to that reported here for guinea-pig VTA ( Figure 2D). Given that the somatodendritic characteristics of DA release in the VTA are discerned only at [Ca 2+ ] o ≤ 0.5 mM, the evoked [DA] o response at [Ca 2+ ] o ≥ 0.5 mM would reflect primarily axonal contributions, as reported here for the first time, and therefore would not be expected to differ from that in striatum.

dIFFerentIal ca 2+ sensItIvIty oF soMatodendrItIc versus axonal da release
Comparison of EC 50 values for the Ca 2+ dependence of DA release in each region revealed that the Ca 2+ sensitivity for somatodendritic DA release is six-to seven-fold higher than for axonal release, supporting our earlier hypothesis that a minimal level of Ca 2+ entry is enough to trigger pure somatodendritic DA release (Chen and Rice, 2001;Chen et al., 2006;Patel et al., 2009). The reasons for the greater sensitivity of somatodendritic DA release to [Ca 2+ ] o are not yet clear, but could reflect a combination of several contributing factors. One known factor is the involvement of intracellular Ca 2+ store activation in amplifying somatodendritic DA release in the SNc ). This amplification can involve store activation by either ryanodine receptors, which are localized near the plasma membrane in SNc DA neurons, or IP 3 (inositol-1,4,5triphosphate) receptors. Notably, facilitation of DA release in the SNc by activation of ryanodine receptors is enhanced in low vs. high [Ca 2+ ] o , which may contribute to the pattern of Ca 2+ sensitivity of DA release reported here. In addition, although the difference in the Ca 2+ sensitivity of somatodendritic and axonal DA release might suggest a difference in release mechanism, the goodness-of-fit of the Hill plots for midbrain and striatum (Figure 3) suggests a common Ca 2+ -dependent exocytotic process in both. Indeed, there is an absolute requirement for Ca 2+ -influx to trigger somatodendritic release, as release is abolished in Ca 2+ -free aCSF (Figures 1D and 2D) and in the presence of the Ca 2+ channel blocker Cd 2+ . Moreover, DA release from primary cultures is suppressed by disruption of SNARE proteins (Fortin et al., 2006), consistent with an exocytotic mechanism underlying somatodendritic DA release.
On the other hand, the higher Ca 2+ sensitivity and lower Ca 2+ cooperativity of somatodendritic vs. axonal DA release support the idea that exocytotic machinery may differ between these release sites (Bergquist et al., 2002;Fortin et al., 2006;Witkovsky et al., 2009). For example, the Ca 2+ -sensing protein primarily responsible for triggering What might contribute to the plateau in evoked [DA] o in SNc? Possible factors include saturation of Ca 2+ influx and depletion of the readily releasable pool (RRP) of DA (e.g., Südhof, 2000). Given that the relationship between [Ca 2+ ] o and [Ca 2+ ] i is usually assumed to be linear (Dodge and Rahamimoff, 1967;Reid et al., 1998), with saturation of Ca 2+ influx only at [Ca 2+ ] o  10 mM (see Church and Stanley, 1996), this leaves depletion of the RRP as the more likely explanation, as seen at the calyx of Held synapse (Schneggenburger et al., 1999). The observation that evoked [DA] o in the SNc plateaued at a ∼10-fold lower [Ca 2+ ] o than that predicted for dorsal striatum or NAc (Figures 3A,C) is consistent with the limited number of vesicles and presence of alternative DA storage sites in SNc DA somata and dendrites Wassef et al., 1981;Groves and Linder, 1983;Nirenberg et al., 1996), as well as the ∼10-fold lower DA content of SNc and VTA compared to striatum (Rice et al., 1997).  (Figure 3). Axonal DA release in dorsal striatum was dependent on the thirdpower of [Ca 2+ ] o , indicating that DA release in this region requires the cooperative action of three Ca 2+ ions. These data are strikingly consistent with the third-power dependence for DA release from striatal synaptosomes reported by Nachshen and Sanchez-Armass (1987), who used completely different methods of DA detection and data analysis than those used here. Similarly, axonal DA release in the NAc shell showed a fourth-power dependence in the present studies. The Ca 2+ dependence of axonal DA release in both regions was within range of other well-examined synaptic systems that use glutamate as their transmitter, with a second-power dependence on [Ca 2+ ] o at the squid giant synapse (Katz and Miledi, 1970), the cerebellar parallel fiber-Purkinje cell synapse (Mintz et al., 1995), and the hippocampal Schaeffer collateral synapses (Qian et al., 1997), but a fourth-power at the neuromuscular junction (Dodge and Ramahmimoff, 1967) and a fifth-power dependence at the calyx of Held synapse (Bollmann et al., 2000;Schneggenburger and Neher, 2000;Sun et al., 2007). By contrast, the dependence of somatodendritic release was less than to the second-power, and thus lower than the dependence for these established synaptic systems.
These differences between axonal and somatodendritic DA release, although consistent with previous studies in guinea-pig brain slices (Chen and Rice, 2001), contrast with recent findings from mouse brain slices, in which no difference in the Ca 2+ depend-  and Rice, 2001;Patel et al., 2009;Witkovsky et al., 2009), because the SNc is devoid of dopaminergic collaterals and axonal synapses (Juraska et al., 1977;Wassef et al., 1981). The dual nature of DA release in the VTA also implies that factors in addition to Ca 2+ may differentially affect the two components. For example, DA neuron firing rate might be a key regulator of the somatodendritic component, whereas local presynaptic regulation might independently suppress or enhance axonal release consequent to neuronal activity.

conclusIon
The present studies demonstrate two key differences between axonal and somatodendritic DA release: low sensitivity, but high cooperativity of axonal release throughout striatum; and high sensitivity and low cooperativity in the midbrain. Based on these distinct features, we conclude that the VTA exhibits both somatodendritic and axonal DA release, providing new insight into DA signaling in this region. Additionally, these findings suggest that disease states that affect vesicle release or calcium entry could have differential consequences among the brain regions examined here. Conditions in which DA levels were depleted, but not eliminated, would not be expected to alter the Ca 2+ dependence patterns reported, assuming the only change was in the size of the releasable pool of DA. However, this might not be the case in disease states that affect transmitter release machinery, for example deletion or mutation of proteins that associate with synaptic vesicles. In support of this hypothesis, altered Ca 2+ dependence of axonal DA release in dorsal striatum has been reported following knockout of α-synuclein in a model of Parkinson's disease (Abeliovich et al., 2000) and with over-expression of mutant torsinA, the product of the DYT1 gene underlying early onset dystonia (Bao et al., 2010).

acknowledgMents
Supported by NIH/NINDS Grant NS036362 (MER) and NIH-NIDCD Training Grant 5T32 DC000063 (KAM). We are grateful to Dr Paul Witkovsky for a critical reading of the manuscript. exocytotic transmitter release is synaptotagmin (Südhof and Rizo, 1996;Südhof, 2000Südhof, , 2002Chapman, 2002) of which there are at least eight different isoforms with differing Ca 2+ affinities in vertebrate brain (Südhof, 2002). For example, synaptotagmin 1 and 2, which are the predominant isoforms found on vesicles involved in fast synaptic transmission (Xu et al., 2007), have a have a 10-20-fold lower Ca 2+ affinity than those of synaptotagmin 7 and 3 (Südhof, 2002;Sugita et al., 2002a,b). Notably, DA neurons in the SNc lack synaptotagmin 1 and 2 ), but express synaptotagmin 7 (Mendez et al., 2011), which may contribute to the higher Ca 2+ sensitivity of somatodendritic release. Thus, the presence of different synaptotagmin isoforms in different neurons or different compartments of the same neuron could alter the Ca 2+ sensitivity and kinetics of transmitter release. The somatodendritic compartment of DA neurons also expresses different complements of SNARE proteins than found at axon terminals (Bergquist et al., 2002;Witkovsky et al., 2009).

axonal as well as soMatodendrItIc da release characterIstIcs oF the vta
In the VTA, DA release showed high Ca 2+ sensitivity like that in SNc, consistent with somatodendritic release. However, the exponential increase in evoked [DA] o with increasing [Ca 2+ ] o indicated an axonal component, as well. Thus, the Ca 2+ dependence of somatodendritic DA release differentiates this process from axonal release. These distinctions reveal that the VTA exhibits both somatodendritic and axonal DA release, consistent with the microanatomy of this region, which includes DA axons, as well as DA somata and dendrites. Indeed, the VTA has DA synapses arising from its own axon collaterals, as well as minor synaptic input from the SNc (Deutch et al., 1988;Bayer and Pickel, 1990). Although sparse, the axonal component of VTA DA release reported here is robust, and is likely to be a contributing factor in the ability of the VTA to sustain DA release with repetitive stimulation (e.g., Iravani et al., 1996). Indeed, because of this potential confounding factor, mechanistic studies of somatodendritic DA release have often focused on the SNc (e.g., Chen