Dendrodendritic Inhibition Through Reversal of Dopamine Transport

See allHide authors and affiliations

Science  28 Sep 2001:
Vol. 293, Issue 5539, pp. 2465-2470
DOI: 10.1126/science.1060645


Synapses in the central nervous system are usually defined by presynaptic exocytotic release sites and postsynaptic differentiations. We report here a demonstration of dendrodendritic inhibition that does not engage a conventional synapse. Using amperometric and patch-clamp recordings in rat brain slices of the substantia nigra, we found that blockade of the dopamine transporter abolished the dendritic release of dopamine and the resulting self-inhibition. These findings demonstrate that dendrodendritic autoinhibition entails the carrier-mediated release of dopamine rather than conventional exocytosis. This suggests that some widely used antidepressants that inhibit the dopamine transporter may benefit patients in the early stages of Parkinson's disease.

Dopaminergic neurons of the substantia nigra are important modulators of basal ganglia function. Their degeneration leads to severe motor and cognitive deficits in Parkinson's disease (1, 2). They exert their influence distally, in the striatum, globus pallidum, and subthalamus (3), through the release of dopamine at axon terminals, and locally, in the substantia nigra, through the dendritic release of dopamine (4–6). Because dopamine hyperpolarizes dopaminergic neurons (7), it is widely accepted that the dendritic release of dopamine primarily leads to the autoinhibition of dopaminergic neurons (8), yet this has never been shown. Dopamine efflux has been documented in dendrites stimulated by high K concentrations (4, 9, 10), glutamate (11), amphetamine (10), and large electrical fields (12), and the molecular steps coupling dopamine autoreceptor activation to inhibition have also been identified (7). Inhibitory postsynaptic potentials (IPSPs) indicating a physiological response to released dopamine have not been seen in the substantia nigra.

We addressed this issue with patch-clamp and amperometric recording methods (13). To trigger release by a large population of dendrites, we chose to stimulate the subthalamic nucleus, because its neurons provide a robust glutamatergic input to nigral dopaminergic cells (14), and its activation in vivo increases extracellular dopamine in the substantia nigra (15,16).

Subthalamic neurons can be stimulated in vitro to produce measurable release of dopamine in the substantia nigra (Fig. 1A). Amperometric recordings in the pars reticulata revealed that subthalamic stimulation elicited small but reproducible current transients (2.3 ± 0.7 pA, n= 14), indicating an increase in extracellular dopamine of about 15 to 20 nM (17). These transients showed greater amplitude in the pars reticulata, consistent with the large number of subthalamic terminals reported in this region (7).

Figure 1

(A and B) Subthalamic afferents trigger dopamine release and autoinhibition in dopaminergic neurons of the substantia nigra. (A) Amperometric detection of extracellular dopamine upon subthalamic stimulation of dopaminergic dendrites. The subthalamic nucleus was stimulated by pressure-application of a solution containing carbachol (10 mM) and NaCl (154 mM). Dopamine oxidation currents were detected with a carbon fiber (diameter 30 μm, length 100 μm) set at a potential of +800 mV and positioned in the substantia nigra pars reticulata. The data were recorded digitally (1-Hz filter, sampling interval 9 ms). To increase the signal-to-noise ratio by a factor of 10, we increased the initial sampling interval from 9 ms to 1 s by averaging consecutive series of 111 points. Off-line analysis and illustration were performed with the program IGOR. (B) Autoinhibition of a dopaminergic neuron recorded in the substantia nigra pars reticulata. Top: The subthalamic nucleus was stimulated by passing small currents (10 μA to 1 mA) between two platinum wires (diameter 10 μm) positioned ∼200 μm apart. Current-clamp recordings of pars reticulata (SNr) and pars compacta (SNc) neurons monitored the response of individual cells to released dopamine (2-kHz filter, sampling interval 1 ms). Bottom: Summating EPSPs and late IPSP elicited by dual stimulation of the subthalamic nucleus, before and after addition of sulpiride (1 μM). The trace recorded in the presence of sulpiride was offset by +3 mV [baseline membrane potential (BMP) was –62 mV and –65 mV, before and after sulpiride]. (C) Autoinhibition of a dopaminergic neuron in the substantia nigra pars compacta. Trace 1: control conditions (BMP = –65 mV). Trace 2: after addition of D-AP5 (50 μM) (BMP = –68 mV). Trace 3: in the presence of D-AP5 and CNQX (5 μM) (BMP = –67 mV). Trace 4: after addition of saclofen (20 μM) to the previous antagonists (BMP = –67 mV). Trace 5: after addition of sulpiride (1 μM) to saclofen, D-AP5, and CNQX (BMP = –70 mV). The cell was injected with a constant current of –10 pA.

To study the postsynaptic effects of released dopamine, we selected a subset of dopaminergic neurons whose cell bodies, located in the pars reticulata (3), may lie close to releasing dendrites (Fig. 1B, cell B in upper panel). These neurons were identified by their characteristic firing properties, morphology, or positive labeling by antibodies to tyrosine hydroxylase (13). In control conditions, they responded to repetitive subthalamic stimulation with summating monosynaptic excitatory postsynaptic potentials (EPSPs) that were followed by a slow, delayed hyperpolarization (Fig. 1B, lower panel, n = 152/173). This hyperpolarization reached a peak amplitude of 6.2 ± 3.0 mV (mean ± SEM, n = 51). It was reduced by the D2 receptor antagonist sulpiride (on average by 73.3 ± 15.1%,n = 15). Hence, for a small contingent of dopaminergic neurons, these findings establish a clear and reliable IPSP in response to dendritically released dopamine.

Dopaminergic neurons of the pars compacta never displayed such IPSPs in control conditions (Fig. 1C, left panel). We assumed that this apparent lack of response to dopamine might reflect the bias of our recording conditions, which favor detection of somatic synaptic potentials and actively amplified dendritic EPSPs over that of remote IPSPs that attenuate through passive propagation. Consistent with this hypothesis, in every tested cell, the inhibitory response to released dopamine was uncovered after partial blockade of the subthalamic EPSP with the glutamate receptor antagonists D(-)-2-amino-5-phosphonopentanoic acid (D-AP5, 50 μM) and 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX, 5 μM) (Fig. 1C, left panel, n = 21). These antagonists, which reduced the subthalamic EPSPs by 88.5 ± 5.1% (n = 21), did not interfere with the ability of the subthalamic afferents to stimulate dendrites and trigger dopamine release, presumably through the activation of metabotropic glutamate receptors (18). The hyperpolarization recorded in the presence of D-AP5 and CNQX represented a well-isolated dopamine-mediated inhibition, which was little affected by the γ-aminobutyric acid type B (GABAB) receptor antagonist saclofen but was almost completely blocked by sulpiride (93.5 ± 5.1% reduction, n = 5) (Fig. 1C, right panel). It resembles, in amplitude and duration, the hyperpolarization that is occasionally seen in dopaminergic neurons of the ventral tegmental area after direct electrical stimulation (19).

These data demonstrate that dopaminergic neurons of the substantia nigra respond to the dendritic release of dopamine with the expected autoinhibition. This finding applies to the small contingent of dopaminergic neurons scattered in the pars reticulata, as well as the main population of dopaminergic cells in the pars compacta, from which most of the dopaminergic innervation of the basal ganglia originates (3).

Reliable detection of autoinhibitory potentials allowed us to investigate the mechanisms by which dendrites release dopamine when they are stimulated synaptically. In dendrites exposed to high K concentrations (4), dopamine efflux is a Ca-dependent process that likely involves exocytosis. However, because dopamine-containing dendrites lack the morphological structures normally associated with exocytosis (6), other mechanisms have been considered. In particular, a Ca-independent efflux of dopamine occurs in neurons stimulated with veratridine (9) or amphetamine (10) through reversal of the dopamine transporter.

To differentiate between these two mechanisms, we used the dopamine transporter antagonist 1-[2-(diphenyl-methoxy)-ethyl]-4-(3-phenylpropyl)piperazine (GBR12935), a potent and selective inhibitor of dopamine bidirectional transport (20, 21). Because dopamine uptake contributes to the clearance of dopamine in the substantia nigra (22), blockade of the dopamine transporter is expected to decrease dopamine uptake and potentiate dopamine actions if release is exocytotic. On the contrary, if release occurs through a reversal of transport, inhibitors of dopamine transport will prevent release altogether, hindering any activation of dopaminergic IPSPs.

GBR12935, applied at low concentration (10 nM), inhibited the late IPSP triggered by subthalamic stimulation both in pars reticulata and pars compacta dopaminergic neurons (Fig. 2, n = 7). Figure 2A illustrates the effect of GBR12935 on the IPSP recorded in a dopaminergic neuron from the pars reticulata. In these cells, the magnitude of this effect was variable, because a significant fraction of the inhibition was often produced, after subthalamic stimulation, through GABAB receptor activation (23). Results in the pars compacta dopaminergic neurons were quantitatively more consistent, because we combined application of D-AP5 (50 μM), CNQX (5 μM), and saclofen (20 μM) to isolate their dopamine-mediated autoinhibitory potentials. GBR12935 nearly abolished the late IPSPs (which were reduced by 94.4 ± 5.0%, mean ± SEM,n = 3). This action was fully reversible. The hyperpolarization, which was recovered after prolonged washout of GBR12935, amounted to 97.8 ± 5.9% of the control value (n = 4, including 1 pars reticulata and 3 pars compacta neurons), and it was abolished by sulpiride (n = 4). The experiment of Fig. 2B recapitulates these findings in a single dopaminergic neuron from the pars compacta. The left panel illustrates the unmasking of the hyperpolarization, elicited by subthalamic stimulation, after application of D-AP5 and CNQX. The right panel describes, in the same cell, the IPSP as (i) insensitive to the GABAB antagonist saclofen, (ii) completely and reversibly blocked by GBR12935 (10 nM), and (iii) abolished after blockade of D2 receptors with sulpiride (1 μM).

Figure 2

Dopamine transporter antagonists block autoinhibition in dopaminergic neurons of the substantia nigra. (A) GBR12935 (10 nM) inhibition of the late IPSP elicited by subthalamic stimulation of a pars reticulata neuron. (B) GBR12935 inhibition of the late IPSP in a dopaminergic neuron of the pars compacta. Trace 1 illustrates the EPSPs triggered by subthalamic stimulation (BMP = –65 mV). To uncover the IPSP, the cell was sequentially exposed to D-AP5 (50 μM) (trace 2, BMP = –68 mV), D-AP5 + CNQX (5 μM) (trace 3, BMP = –67 mV), D-AP5 + CNQX +saclofen (20 μM) (trace 4, BMP = –67 mV). The late IPSP recorded in these conditions was blocked by GBR12935 (10 nM) (trace 5, in D-AP5 + CNQX + saclofen + GBR12935, BMP = –70 mV). Blockade by GBR12935 was reversed (trace 6, in D-AP5 + CNQX + saclofen, BMP = –68 mV) and the recovered IPSP was abolished by sulpiride (1 μM) (trace 7, in D-AP5 + CNQX + saclofen + sulpiride, BMP = –69 mV). Both cells in (A) and (B) were injected with a constant current of –10 pA. (C) GBR12909 (10 nM) inhibition of the IPSP elicited by subthalamic stimulation in a dopaminergic neuron of the pars compacta. APV (50 μM), CNQX (5 μM), and saclofen (20 μM) were present throughout the experiment. Fluoxetine (100 nM) was added 20 min before GBR12909 to verify that the late IPSP was unaffected by inhibition of the serotonin transporter (BMP = –66 mV in control, in the presence of fluoxetine, and after addition of GBR12909). The cell was steadily injected with a constant current of –50 pA. (D) β-CFT (10 nM) inhibition of the late IPSP evoked by subthalamic stimulation in another dopaminergic neuron of the pars compacta. APV (50 μM), CNQX (5 μM), and saclofen (20 μM) were present throughout the experiment. Desipramine (100 nM) was added 20 min before β-CFT to verify that the late IPSP was unaffected by inhibition of the norepinephrine transporter (BMP = –65 mV in control, BMP = –63 mV in the presence of desipramine, BMP = –68 mV after addition of β-CFT). The cell was steadily injected with a constant current of –20 pA.

Other blockers of the dopamine transporter, (1-{2-[bis(4-fluorophenyl)methoxy]-ethyl}-4(3-phenylpropyl)piperazine (GBR12909, n = 4) and 2-β-carboxymethoxy-3-β-(4-fluorophenyl)tropane (β-CFT,n = 4), also abolished the subthalamic IPSP recorded in the same conditions (Fig. 2, C and D). In contrast, this IPSP was not affected by fluoxetine, an antagonist of the serotonin transporter (Fig. 2C, n = 3), or by desipramine, an antagonist of the norepinephrine transporter (Fig. 2D, n = 3) (24).

Amperometric detection of extracellular dopamine provided a separate demonstration that the dopamine transporter mediates dopamine release evoked by subthalamic inputs (Fig. 3A). The oxidation currents elicited (as in Fig. 1A) by subthalamic stimulation were inhibited reversibly by GBR12935 (n = 3/3). After application of 20 nM GBR12935 for 90 min, they were reduced by 76 ± 6.8% and they fully recovered within 2 hours after washout of GBR12935. In contrast, GBR12935 (20 nM) enhanced currents elicited by exogenously applied dopamine (by 42 ± 20% after 30 min, n = 4, Fig. 3B), due to inhibition of uptake. A similar enhancement was seen inFig. 3A (bottom), which suggests that the effects of GBR12935 on release and uptake can be dissociated (25).

Figure 3

GBR12935 inhibits the dendritic release of dopamine evoked by subthalamic stimulation and enhances the response to exogenously applied dopamine. (A) Amperometric currents were evoked in the pars reticulata after carbachol stimulation of the subthalamic nucleus (as in Fig. 1A), in control condition (top), after the application of 20 nM GBR12935 (middle) and 15 min after washout of the antagonist (bottom). Calibration of the carbon-fiber electrode was 9 nM dopamine/pA. (B) Amperometric currents evoked by pressure application of dopamine (1 μM, in 154 mM NaCl, with 1 mMd-cysteine to prevent dopamine oxidation), in control (top), 25 min (middle), and 60 min (bottom) application of 20 nM GBR12935. Calibration of the carbon-fiber electrode was 16 nM dopamine/pA.

In addition to their distinct response to dopamine transporter antagonists, the exocytotic and carrier-mediated release of dopamine can be distinguished by distinct calcium requirements. Unlike exocytosis, dopamine transport is unaffected by removal of external Ca2+ ions (9, 20). To emulate the subthalamic stimulation of dendrites in calcium-free conditions, we used local pressure application of glutamate (1 mM). In the pars reticulata, glutamate applications resulted in robust and reproducible amperometric currents (10.6 ± 1.8 pA, n = 12) (Fig. 4). These transient currents, measured at a carbon fiber that was positioned in the pars reticulata, signaled transient increases in dopamine levels with little contamination by other amines (17). Whereas the transient responses in the presence of 2 mM extracellular Ca decreased with each application of glutamate, the amperometric signals in calcium-free conditions were remarkably stable (26), demonstrating that glutamate receptor activation reliably triggers the calcium-independent efflux of dopamine.

Figure 4

GBR12935 inhibits the Ca-independent release of dopamine triggered by glutamate application. Traces represent individual amperometric responses to glutamate (1 mM) application in control calcium-free condition, in the presence of GBR12935 (20 nM), and after washout of the blocker. The final trace is superimposed with the control response to illustrate the full recovery of the evoked signal and the baseline to their initial values. Amperometric signals were detected and recorded as in Fig. 1A, but dendrites were stimulated directly by pressure-application of glutamate through a glass pipette that was positioned in the substantia nigra pars reticulata and filled with glutamate (1 mM) and NaCl (154 mM). Each pulse (9 s, 30 psi, every 30 min) applied ∼60 nl of the glutamate-containing solution. The whole experiment was carried out in calcium-free solution (26). Calibration of the carbon fiber electrode was 140 nM dopamine/pA.

Consistent with the hypothesis that glutamate stimulation of dopaminergic dendrites leads to a reversal of dopamine transport, the amperometric currents recorded in these conditions were significantly reduced (by 70 ± 1.3%, n = 4) by GBR12935 (20 nM). The amperometric signals fully recovered, within 2 hours, after washout of the antagonist (n = 3), ruling out inhibition of the responses to glutamate through reduction in the carbon fiber sensitivity, depletion of the releasable pool of dopamine, or degradation of the brain slice condition. In the experiment of Fig. 4, selected for the stability of the carbon fiber electrode over the 4-hour length of the recordings, GBR12935 nearly abolished the transients evoked by glutamate applications. Interestingly, it produced dissociated effects on the basal levels of dopamine and the transient increases triggered by glutamate application (25). As predicted if low concentrations of GBR12935 preferentially affect dopamine uptake, GBR12935 initially increased the baseline levels of dopamine and enhanced the amperometric signal evoked by glutamate application. However, the subsequent responses to glutamate were reduced. After 90 min of GBR12935 application, their inhibition amounted to ∼90% of the control current. Recovery from block was slow but complete and was accompanied by a parallel return of the dopamine baseline level to its initial value. These data demonstrate that, upon glutamate receptor activation, dopaminergic dendrites release dopamine in calcium-free conditions through reversal of dopamine transport.

These data show that under physiological conditions, reversal of dopamine transport occurs when dopamine-containing dendrites are stimulated by their subthalamic afferents. Although they explain the lack of dendrodendritic synapses reported in the pars reticulata (6), our findings were largely unexpected. First, recent amperometric studies (27) clearly establish the occurrence of dopamine exocytosis in the substantia nigra. Second, inside-out transport of dopamine had only been documented in nonphysiological conditions that seemed unlikely to be matched in dendrites stimulated synaptically (9, 20).

The direction and magnitude of dopamine transport depends on membrane potential and the Na+ and Cl ion gradients (28). Our results imply that subthalamic EPSPs depolarize dendrites beyond the reversal potential for dopamine transport. Considering the geometry of the dendrites (their large surface-to-volume ratio), it is likely that increases in intracellular Na and membrane depolarization both contribute to reversal of dopamine transport. One can easily imagine that Na influx through glutamate-gated channels and voltage-gated Na channels (29) raises the intracellular Na concentration significantly, bringing the reversal potential for dopamine transport closer to the cell resting membrane potential.

The carrier-mediated release of dopamine bypasses the classic requirements for transmission at conventional synapses. Considering the wide distribution of the dopamine transporter (30), every dendritic compartment is a potential release site for dopamine. Rather than being defined structurally by their exocytotic machinery, the dendritic release sites for dopamine may therefore be allocated dynamically, in space and time, by propagating Na action potentials (29) and other dendritic regenerating responses (31). Locally, it is unclear whether the activity of the dopamine transporter itself shortens dopamine efflux; released dopamine would be expected to hyperpolarize releasing dendrites, thereby promoting its own uptake (32). Because the release of dopamine through the transporter is intrinsically transient, it is well suited for shaping the phasic responses of dopaminergic neurons in the substantia nigra during reward and attention-shifting tasks (33), as well as their bursting behavior (7).

Dopamine transporter inhibitors are widely used in the treatment of depression, attention-deficit hyperactivity disorder, and addiction (34). Disruption of carrier-mediated release of dopamine (in midbrain nuclei) could be another mechanism by which dopamine transporter antagonists exert their therapeutic effects, in addition to their known facilitation of dopaminergic transmission at axon terminals. Such a mechanism of action predicts a novel clinical application for dopamine transporter inhibitors as neuroprotective agents in the early stages of Parkinson's disease. Because subthalamic afferents trigger the carrier-mediated release of dopamine in the substantia nigra, their hyperactivity in Parkinson's patients (2) likely results in excessive release of dopamine. Considering dopamine neurotoxicity (35), inhibition of such release by dopamine transporter blockers may slow the degeneration of dopaminergic neurons in Parkinson's disease.

  • * To whom correspondence should be addressed. E-mail: imintz{at}


Stay Connected to Science

Navigate This Article