Electrophysiological Properties of Midbrain Dopamine Neurons

Anthony A. Grace and Benjamin S. Bunney

The neurotransmitter, dopamine (DA), has had a comparatively short but momentous history (cf. ref. 20). The discovery of DA as an independent neurotransmitter occurred in 1958 by Dr. Arvid Carlsson. Five years after its discovery, this neurotransmitter was implicated in the mode of action of antipsychotic drugs, and the degeneration of DA neurons was discovered to be the etiological basis of Parkinson's disease. As a consequence, the DA system itself has been subjected to extensive analysis by investigators utilizing a variety of approaches to better understand this important but complex neurochemical system.

The electrophysiological analysis of DA-containing neurons began with a paper published in 1973 (4), in which combined physiological, neurochemical, and histochemical techniques were brought to bear in defining this system. As a result of this effort, substantial progress has been made in the past two decades in which the physiology of this system has been dissected in preparations ranging from in vitro recordings of isolated neurons to recording their activity in freely behaving primates. Therefore, although the physiology of the DA neuron has been reviewed in every volume of this series, including the last one published only 7 years ago, sufficient data have accumulated to devote two chapters to the analysis of this system: This chapter deals with the basic physiological properties of the DA neuron, and in Biochemical Pharmacology of Midbrain Dopamine Neurons the pharmacological responses of the DA neuron are examined in detail. In the present chapter, primary emphasis will be placed on the more recent in vitro recordings, with comparisons drawn to in vivo data reviewed in the last volume (see also Dopaminergic Neuronal Systems in the Hypothalamus and Mesocorticolimbic Dopaminergic Neurons: Functional and Regulatory Roles for differential functional emphases, and Pharmacology and Physiology of Central Noradrenergic Systems for noradrenergic neurons).

The electrophysiological analysis of neurons is capable of providing substantial insight into the function of these neuronal elements, and comparisons between their discharge properties and biochemical measures of neurotransmitter regulation can provide the basis for novel ideas regarding the multifaceted modes of regulation present in the nervous system. Indeed, such a coordinated biochemical and electrophysiological analysis of defined systems has led the way to substantial advances in our knowledge of the catecholamine systems. However, when the electrophysiological study involves neurochemically defined neurons, the physiologist has the burden of providing a level of proof that can be taken for granted in most biochemical investigations: that the neurons one is studying are indeed the cells that release the neurotransmitter of interest. Although identification of neurons on the basis of location or morphology may suffice when the investigator is studying the postsynaptic actions of these neurohumors, meaningful data cannot be collected from the cells that synthesize and release a given neurotransmitter unless this identification is definitive.

DA-containing neurons in the midbrain of rats have undergone a series of identification techniques ranging from indirect to direct, depending on the capabilities of the techniques employed. The first identification was accomplished in vivo, and it relied on the ability to cause DA in tissue slices to fluoresce when treated with substances that combine with DA to form a fluorescent product, such as formaldehyde vapor or glyoxylic acid. Thus, utilizing extracellular recording techniques, the DA precursor L-DOPA was locally applied by iontophoresis to midbrain neurons in the substantia nigra that exhibited a unique action potential waveform. The result was the recovery of a group of highly fluorescent, presumably dopaminergic neurons that had taken up the L-DOPA, converted it to DA, and as a result showed higher levels of fluorescence than observed in DA neurons located only a few hundred microns from the ejection site (4). This identification was later verified using in vivo intracellular recording techniques to inject the L-DOPA directly into a single DA-containing neuron, thereby providing precise identification of the single cell impaled as being of the dopaminergic cell type (22). This was further confirmed using other techniques to augment the DA content (i.e., intracellular injection of the tyrosine hydroxylase cofactor or intracellular injection of colchicine to cause somatic buildup of DA), and hence histochemical fluorescent intensity, of single DA neurons (23). Adjunct techniques including DA-specific lesions and antidromic activation provided further support for the accuracy of this identification (23, 31). As a result of these investigations, a unique action potential waveform and firing pattern could be accurately associated with identified DA-containing neurons, thereby enabling accurate studies of the physiology and pharmacology of this important neurochemical system.

Unfortunately, the criteria used to identify this class of neurons in the in vivo preparation proved to be insufficient to be utilized in most in vitro recording studies. This was due to three factors: (i) The neurons recorded in the substantia nigra in the in vitro midbrain slice did not fire with the same discharge pattern as that linked with identified DA neurons recorded in vivo, (ii) identification based on antidromic activation from DA axon target sites was not possible in a 400-mm-thick slice, and (iii) because most in vitro studies involved intracellular recordings, the unique extracellular action potential waveform of the DA cell was not an appropriate identification criterion. Indeed, one of the most unusual features of the spike discharge of the DA neuron in vivo was the highly variable nature of the spike waveform generated from a single neuron; however, because this variability was apparently derived from the interaction with afferent fibers, this characteristic was not observed in the in vitro preparation. Furthermore, because studies have shown the presence of nondopaminergic neurons within the zona compacta region (32), localization of the recorded neuron was not sufficient to identify these neurons. Indeed, reliance on location within the zona compacta has led several investigators to ascribe characteristics to DA neurons, such as rebound burst firing and prominent low-threshold calcium events, even though these properties have never been observed in any of the studies that have used reliable identification techniques. In fact, studies that relied on histochemical identification of neurons have universally ascribed these properties to the nondopaminergic neurons within the substantia nigra (29, 69). As a result, a different set of criteria based on new identification procedures were required.

Because of the high variability in the intraneuronal concentrations of DA present within neurons located at different depths within the midbrain slice recorded in vitro, it was difficult to identify DA neurons using the techniques previously utilized in vivo. Instead, the first direct identification of DA neurons in vitro relied on immunocytochemistry. Thus, because only catecholamine-synthesizing neurons contain the enzyme tyrosine hydroxylase, and given the absence of the enzyme DA-b-hydroxylase (a norepinephrine- and epinephrine-specific enzyme) within neurons in this brain region, DA cell identification was accomplished using combined intracellular staining with immunocytochemical localization of tyrosine hydroxylase. In order to reliably separate the fluorescence of the intracellular stain from that of the tyrosine hydroxylase immunolocalization within the same brain section, the putative DA neuron was injected intracellularly with the highly fluorescent dye Lucifer yellow, which could be readily separated from the rhodamine-labeled secondary antibody employed in the immunocytochemical localization of tyrosine hydroxylase. Using this method, DA-containing neurons within the midbrain ventral tegmental area and substantia nigra were accurately identified during in vitro recordings in the midbrain slice (29, 69). Unlike that reported in vivo, in both cases the DA-containing cells were identified as the non-burst-firing cell class, with the cells showing prominent rebound burst firing shown to be the nondopaminergic zona reticulata neurons. Furthermore, this study provided the basis for another distinct criterion that could be used for the electrophysiological identification of DA-containing neurons during intracellular recordings in brain slices: the presence of a large-amplitude, slow depolarization that was responsible for driving spike activity in these cells (18, 29). More recently, DA-containing neurons have been identified using direct histochemical detection of catecholamine fluorescence or immunocytochemistry in dissociated substantia nigra tissue (33, 64) as well as in primary neuron cultures (8, 54).

Intracellular injection of Lucifer yellow was also used for examining the morphology of DA neurons in thick slices. In general, DA neurons of the substantia nigra could be divided into two classes: The DA neurons located in the more dorsal regions of the zona compacta of the substantia nigra were typically fusiform in appearance, with somata averaging 15–25 mm in diameter and with 2–5 dendrites emanating from the poles of the neuron. The dendrites branched sparsely within the substantia nigra, but remained confined to the zona compacta region of this nucleus. In contrast, the more ventrally located DA neurons were multipolar in shape, having somata that were approximately 20–35 mm in diameter. The dendrites also emanated from the poles of the soma and extended laterally within the zona compacta region. However, unlike the dorsal neurons, DA cells in the ventral region also had a long dendritic process that entered into and branched within the zona reticulata. The DA neurons stained within the ventral tegmental region ranged from large fusiform to multipolar in appearance. They also had 3–5 dendrites emanating from the soma that extended in a radial array from the soma, probably owing to the comparative lack of spatial constraints within the ventral tegmentum. Analysis of neurons using an image analysis system revealed that the neuron and its dendritic field typically extended over a mediolateral and dorsoventral region of approximately 400–2500 mm; however, the neurons were comparatively constrained in the anteroposterior direction, typically extending only 100–200 mm in this dimension (). Similar morphological properties of zona compacta neurons have been described using intracellular injection of horseradish peroxidase (41, 67).

DA neurons in both the ventral tegmentum and the substantia nigra have several atypical morphological characteristics that may contribute to their distinctive physiological properties. One of these unusual properties is that DA neurons appear to be capable of storing and releasing DA within the substantia nigra and ventral tegmentum in a calcium-dependent, tetrodotoxin (TTX)-sensitive manner (6). However, this release does not appear to be derived from local axonal collaterals. The axon of the DA neuron in almost all cases is not derived from the soma of the neuron, but instead emanates from a major dendrite or from a somatic appendage. The thin, unmyelinated axon projects out of the somatodendritic field before turning in an anterior direction and exiting the plane of the slice. Local axonal branching or collaterals have not been observed. On the other hand, the long, sparsely branched dendrites of the DA neuron appear in all cases examined to end in a forked process. The forks of this process are thin and highly recursive, giving the appearance of an axonal branch encircling a postsynaptic element. Although speculative, the axon-like morphology of the dendrite and the absence of recurrent axonal terminals suggest that these distally located, specialized dendritic arborizations may represent the dendritic DA release sites associated with these neurons.

The dendrites of the DA neurons are mostly smooth in morphology. However, the dendrites are also observed to possess short cytoplasmic extensions that occur infrequently along the length of the process. These branchlets are approximately 0.5–2.0 mm in diameter and extend from 5 to 50 mm from the dendrite without branching. These branchlets are not unlike those associated with axon terminal attachment sites in other brain regions (see also Electron Microscopy of Central Dopamine Systems for ultrastructural progress).

The membrane properties of DA neurons appear to depend, to some extent, on the preparation used. Stable intracellular recordings from DA neurons in vivo have shown them to have a moderate input resistance, averaging 31 ± 7 megohms (range: 18–45 megohms) (23). In contrast, identified DA neurons recorded in vitro exhibit input resistances approximately three- to fivefold greater, with input resistances averaging 168 ± 61 megohms (range: 80–320 megohms) (16, 29, 45). Nonetheless, in both cases the membranes are characterized by a prominent anomalous rectification when hyperpolarized () (16, 18, 23, 41, 43, 45). This anomalous rectification is comprised of two components: an instantaneous component and a time-dependent component that showed maximal activation at membrane potentials of -63 mV and -78 mV, respectively (18). In addition, at the termination of a brief membrane hyperpolarization the DA cell exhibits a pronounced voltage-dependent delayed repolarization (13, 16, 18, 29, 64). Both in terms of its voltage dependency and its effects on membrane repolarization, this conductance resembles that described in other preparations as the IA current. However, unlike the IA described in other regions, this conductance is not affected by application of the potassium blocker 4-aminopyridine ().

One of the unusual properties of DA neurons noted during extracellular recording is the prominent negative phase of the spike and the highly variable shape of their spike waveform, even when recording a series of spikes from a single, well-isolated neuron. Intracellular recordings performed from DA neurons both in vivo (23, 25) and in vitro (17, 18, 29) have provided a potential explanation for this observation. First, in both preparations the DA neuron action potential is triggered from a comparatively depolarized membrane potential, averaging -41 mV in vivo and -36 mV in vitro. Thus, the DA neuron membrane must be depolarized approximately 15–25 mV in order to depolarize the membrane from its resting potential (average RMP = -57 ± 4 mV) to this high threshold for spiking. In fact, it is these depolarized spike thresholds that likely prevent antidromic activation of DA neurons from eliciting a full-amplitude action potential, since even the initial segment (IS) spike is not of sufficient amplitude to provide this level of depolarization (17), 23). The spontaneous discharge of the DA neuron is therefore highly dependent on the presence of a slow, large-amplitude pacemaker depolarization (). This pacemaker conductance is a voltage-dependent depolarization that drives spontaneous spike generation in DA neurons as shown in both in vivo (23, 25) and in vitro (14, 16, 18, 29, 41) preparations. It is the alternation of this pacemaker with the spike and its associated afterhyperpolarization that underlies the highly regular firing pattern observed in DA neurons in the in vitro brain-slice preparation (29) as well as in dissociated dopaminergic neurons (33). Although DA neurons recorded in vivo rarely fire in a pacemaker-like pattern, intracellular injection of the calcium chelator ethyleneglycol bis(aminoethyl ether)tetraacetate (EGTA) has been shown to change an irregularly discharging or a burst-firing neuron in vivo to one firing in a highly regular pacemaker pattern identical to that found in vitro. This pacemaker conductance appears to be comprised of both sodium- and calcium-dependent conductances, because administration of either TTX or cobalt will block spontaneous spike generation of DA neurons in vitro.

Blockade of sodium conductances in DA neurons by administration of TTX readily blocks spontaneous spike activity and reveals the presence of two cobalt-sensitive (and hence presumably calcium-mediated) active membrane events. Following TTX treatment, a low-amplitude, brief hyperpolarization of the membrane will evoke a rebound depolarizing all-or-none wave (16, 29, 41, 52) that in many ways resembles the low-threshold calcium spike (LTS) reported in other brain regions. The presence of this rebound event is highly dependent on the membrane potential: It cannot be evoked by brief hyperpolarizing pulses injected into a hyperpolarized neuron, whereas depolarizing the neuron will typically reveal this rebound event. However, if the amplitude of the hyperpolarizing pulse is increased, the rebound event is obscured by the IA-like delayed repolarization (). Indeed, the constraints placed on the amplitude of the LTS by its voltage-dependency and that of the putative IA most likely underlie the inability to evoke large-amplitude rebound depolarizations or bursts, but appear to be sufficient to enable small-amplitude hyperpolarizing events to elicit sustained rebound spike firing (). Because this LTS can be evoked by injecting very-small-amplitude depolarizing pulses into the soma, is highly dependent on the membrane potential of the soma, and has an activation threshold that is not altered by potassium blockers, this event is most likely generated proximal to the soma (17, 44).

In TTX-treated neurons, injection of large-amplitude depolarizing current pulses fails to evoke any additional active membrane events. However, after administration of the potassium blocker tetraethylammonium (TEA) to the bath, comparatively small-amplitude depolarizations of the membrane will evoke large-amplitude (i.e., 60–85 mV), long-duration (i.e., 7–20 msec), cobalt-sensitive spikes with prominent afterhyperpolarizations () (16, 17, 29, 41) that resemble the high-threshold calcium spikes (HTSs) observed in other preparations. TEA is known to block the delayed rectifier of neurons, allowing current to spread more effectively into the distal dendrites and making them more isopotential with the soma. The fact that these events cannot be evoked by direct somatic depolarization without the presence of TEA suggests that they are generated distally to the soma, possibly at the distal dendritic regions (44). Indeed, the calcium-dependent nature of these spikes and their generation at distal dendritic sites suggest that these spikes may at least contribute to the dendritic release of DA from these neurons (29, 44). These HTSs apparently give rise to the prolonged afterhyperpolarizations associated with the slow firing frequency of these neurons. The afterhyperpolarization itself appears to be composed of two components: (i) a fast component that inactivated near -63 mV independent of the membrane potential and (ii) a slower, larger-amplitude component that increased in amplitude with hyperpolarization of the membrane (18). A long-latency afterhyperpolarization that exhibited summation with subsequent spikes in a train similar to that observed in vivo (25) was not observed in the in vitro preparation (18).

The pharmacology of these calcium-dependent events has recently been investigated in detail (53). It appears that the oscillatory calcium conductance that underlies the TTX-insensitive portion of the oscillatory pacemaker potential can be abolished by treatment with nifedipine, whereas the HTS is little affected by this blocker. Furthermore, both nifedipine and apamin also attenuated the slow component of the spike afterhyperpolarization (53, 63). In contrast, the HTS can be partially blocked by administration of w-conotoxin.

As found with blockade of sodium conductances with TTX, administration of the calcium blocker cobalt to DA neurons recorded in vitro causes a cessation of spontaneous spike activity. However, strong depolarizations of the membrane will trigger a rapid burst of fast, small-amplitude spikes (). These spikes have little or no afterhyperpolarizations and can be blocked by TTX. Although administration of TEA does not affect these events, application of 4-aminopyridine (4-AP) causes a substantial lowering of their threshold, enabling these spikes to be evoked by moderate depolarization of the membrane. In terms of their amplitude, duration, and absence of an afterhyperpolarization, these spikes resemble the IS spikes triggered during antidromic activation of DA-containing neurons in vivo.

Perhaps the most distinctive feature of DA cells is the irregular but unique spike waveform they produce. These action potentials are unusual in having an atypically depolarized spike threshold (i.e., -41 mV in vivo and -36 mV in vitro at RMP) that varies with the membrane potential; indeed, the threshold decreases by approximately 5 mV with a 7-mV hyperpolarization of the membrane. Studies have shown that, as with most other neurons, the action potential is comprised of a short-duration (0.8–1.5 msec) initial segment (IS) spike followed by a longer-duration (1.5–3 msec) somatodendritic (SD) spike; however, in DA cells the IS spike is only approximately one-fourth to one-third the size of the SD spike. The IS is the lowest-threshold region of the neuron, and therefore it determines the action potential threshold. The IS spike in DA neurons is likely to be generated at a site that is comparatively distal and not isopotential with respect to the soma, because (a) the morphological studies reviewed earlier showed that in DA neurons, the axon does not arise from the soma but instead arises at a more distal location on a primary dendrite, (b) the IS spike is smaller in amplitude than the SD spike, (c) its threshold varies with the membrane potential, and (d) administration of 4-AP lowers the action potential threshold as well as the threshold of the IS spike, without affecting the SD spike. TEA, on the other hand, lowers the threshold of the calcium-mediated HTS spike and increases the duration of the SD spike without affecting the IS spike. Furthermore, as shown during antidromic activation of DA neurons in vivo, activation of the IS spike is not sufficient to trigger the SD spike unless the membrane is depolarized by the pacemaker potential.

Based on the data presented in this section, a model of spike generation in DA neurons may be derived (18): The action potential sequence is initiated when the membrane is depolarized by the pacemaker potential. The IS segment is depolarized fastest because of its smaller volume and proximity to the pacemaker potential source. The depolarization at the soma lags behind the IS potential due to a 4-AP sensitive conductance. Indeed, a 4-AP-sensitive IA located between the IS and the soma could contribute to this lag by shunting the slow depolarization. Such a 4-AP-sensitive IA conductance has been identified in the more electrotonically compact, higher input resistance dissociated DA neuron preparation (64), in which normally electrotonically distal events may be more easily assessed by a whole-cell clamp at the soma. As a consequence, the apparent threshold of the IS as measured at the soma would be substantially higher than at the IS, and would vary depending on the voltage-dependent activation of the putative 4-AP-sensitive IA. The IS would then spread across the somatodendritic region already depolarized by the slow depolarization, and thereby cause a sufficient level of depolarization in the distal dendritic regions (possibly the site of the recursive dendritic arbors and putative DA release sites) to trigger the HTS/SD spike component. The Ca-mediated HTS then hyperpolarizes the cell by activating a IK(Ca), which upon decay activates the LTS and the next slow depolarization. Another potential function of the HTS would be to trigger calcium-dependent dendritic DA release. Note that this is only one of several possible ways to fit the data into a sequential model. Nevertheless, it can serve as a basis for analyzing the functional compartmentalization of the DA neuron until more substantial data can be derived.

Studies have shown that the pharmacological or physiological activation of DA neurons occurs across three dimensions (3, 16, 35): (i) firing rate, (ii) spike discharge pattern (25, 26), and (iii) the proportion of neurons that are spontaneously active. Thus, systemic administration of DA antagonists (3, 4), iontophoretic application of glutamate (26), or large depletions of striatal DA (35) will increase the firing rate of active neurons, increase the number of spontaneously active neurons recorded per electrode track in the substantia nigra, cause the DA neurons to fire in bursts, and cause them to change from a single-spiking to a burst-firing pattern of discharge.

One activity state that has received increased attention recently is that of firing pattern. DA cells show a comparatively limited range of activity, with spontaneous firing rates averaging approximately 4.5 Hz in the anesthetized rat, but are only capable of firing at a maximum of approximately 10 Hz. However, an increase in DA cell activity is typically accompanied by a shift from an irregular single-spiking pattern to one of burst firing (25, 26). Burst firing found to occur in DA cells is distinct from that observed in areas more typically associated with burst discharge, such as the hippocampus or thalamus, in that the DA cells have an exceptionally long interspike interval between spikes in the burst, the bursts can consist of 15 spikes or more, and the bursts are associated with depolarization of the DA neuron membrane (26). Stimulation studies have shown that activation of the DA neuron axon in patterns resembling burst discharge will release two to three times more DA than is released by an equivalent number of evenly spaced stimuli (15). Therefore, an alteration in firing pattern is likely to cause a more substantial alteration in synaptic DA release than that produced by an increase in discharge rate alone. For this reason, the regulation of DA cell discharge pattern has been the focus of considerable attention in recent years.

DA neurons recorded in vivo will fire in bursts upon depolarization of their membrane. One of the most potent activators of burst firing in these neurons is the direct application of glutamate to the neurons (26), or by activating glutamatergic afferents that innervate DA neurons (65). Furthermore, glutamatergic stimulation appears to be a prerequisite for inducing burst firing in vivo. Thus, inactivation of glutamatergic afferents to DA neurons (65, 66) or administration of a glutamate antagonist such as kynurenic acid (5) or an N-methyl-D-aspartate (NMDA) blocker (7, 55) will cause DA neurons to enter a single-spiking mode. Nonetheless, glutamatergic stimulation alone does not appear to be sufficient to cause burst firing. Several studies have now shown that positively identified DA neurons do not fire in bursts when recorded in vitro in midbrain slices, nor can they be induced to fire by depolarization or hyperpolarization of the membrane (18, 29, 69). Furthermore, application of glutamate or NMDA agonists alone will not cause DA neurons recorded in vitro to fire in bursts (62). Indeed, the only instances in which activity partially resembling burst firing has been induced in the in vitro preparations are cases in which one or more potassium channel blockers have been applied to the neuron (39, 63).

Therefore, the current data are insufficient to adequately account for the mechanisms underlying burst firing in DA neurons. However, one thing that is apparent about the mechanism responsible is that, although burst firing is typically associated with increased firing rate, studies have shown that these characteristics appear to be regulated by different systems. Thus, the correlation between firing rate and burst firing is reported to be substantially different when nicotine instead of glutamate is used to stimulate these neurons (30). This can also be illustrated in studies in which the subthalamic nucleus is activated. Stimulation of the subthalamic nucleus appears to activate direct glutamatergic afferents to DA neurons as well as afferents to GABAergic neurons in the substantia nigra that inhibit DA cell firing. Thus, activation of the subthalamus nucleus will inhibit the firing rate of DA neurons. However, upon cessation of the stimulation, the DA cell shows a significant activation of burst firing as it recovers toward its basal rate of firing (65). Thus, burst firing appears to be regulated by a system that has a different time course from that controlling firing rate.

Each of the activity states reviewed above appears to be under a homeostatic regulatory influence. Thus, when DA demand is increased by administering a DA blocker, such as haloperidol, each of these dimensions of DA cell activity are increased (3, 28)). Another method that has been used to place a demand on the DA system is the production of 6-OHDA-induced partial lesions of the nigrostriatal DA system. Studies using this technique provide a substantial example of the enormous recovery potential present in this system. Thus, rats depleted of 95% or more of their striatal DA will, after approximately 1 month, exhibit spontaneous recovery of function (70). This recovery is mediated by several homeostatic changes within the nigrostriatal DA system, including decreased DA uptake sites, increased DA synthesis, DA receptor supersensitivity, and so on (cf. refs. 19 and 70). However, the electrophysiological activity of the remaining DA neurons is at its basal level; indeed, even the proportion of neurons active remains constant after correcting for cell loss (35). On the other hand, all of these parameters of DA cell activity are increased if the lesion size exceeds approximately 97%, at which point the rat fails to recover. Furthermore, in lesion-recovered rats, introduction of additional demand on the DA system, such as a low dose of haloperidol, causes a rapid depolarization blockade of DA cell firing and a return of profound motor deficits (37). Therefore, it appears that DA neurons are required to be at their basal level of activity in order to maintain a normal behavioral condition. On this basis, it appears that the homeostatic changes take place over time at the biochemical and receptor level in order to maintain a maximal dynamic range of electrophysiological response, thereby enabling a maximal, rapid increase in DA cell firing and DA release in response to an immediate behavioral demand (35).

An additional state of DA cell activity is depolarization block of spike generation (3, 28). In depolarization block, DA cells are in a state of overexcitation, resulting in the inactivation of spike firing. However, unlike the other states of activity, depolarization block appears to only occur in association with pharmacological manipulation of the nigrostriatal system. Thus, repeated administration of DA blockers (3, 28), or acute administration of a DA blocker in DA-lesioned and recovered rats (37), will induce this state in DA neurons. Interestingly, decreased DA receptor stimulation by repeated haloperidol administration will readily induce depolarization block, whereas the diminished DA receptor stimulation occurring after a partial DA lesion fails to induce block. One potential explanation for this difference could be that haloperidol administration also pharmacologically circumvents a major homeostatic influence on DA neurons—that is, somatodendritic autoreceptor-mediated inhibition. Thus, with 6-OHDA-induced depletion of only 50% or more of striatal DA, the DA cells in the substantia nigra show evidence of autoreceptor supersensitivity (56). Furthermore, the subsequent pharmacological blockade of these autoreceptors by acute administration of a DA blocker immediately induces a state of depolarization block (37). Therefore, the DA system appears to be capable of compensating for physiological deficits; however, pharmacological interference with its autoreceptor-mediated feedback regulatory system causes the system to lose stability and inactivate due to overdepolarization.

In the previous section, recent advances into the membrane mechanisms that influence the discharge properties of individual DA neurons were reviewed. In this section, evidence for DA neuron regulation via interaction with neighboring dopaminergic neurons is summarized. This interaction has been shown to occur via two distinct and independent mechanisms: (i) electrical interactions and (ii) dendritic release of DA.

This topic has been reviewed in detail in prior publications of this volume, and it will only be mentioned briefly here. Autoreceptor-mediated inhibition via dendritic interactions between DA neurons have been known for some time. However, this type of inhibitory interaction cannot account for data showing that neighboring DA neurons often fire simultaneously with very short interspike intervals (24). Studies using in vivo intracellular recording and injection of the dye Lucifer yellow provided evidence that subsets of DA neurons appear to be interconnected by gap junctions (24). This hypothesis has received support by the recent demonstration of the presence of connexin immunoreactivity within the zona compacta region of the substantia nigra (51), because connexin has been shown to be a monomeric component of the gap junction structure.

The high degree of sensitivity of DA neurons to their own transmitter (4) and the release of DA from dendritic stores (6) have been known for some time; however, clear evidence of a functional role for this system has been lacking. Thus, several investigators cast doubt about whether tonic autoreceptor stimulation is present on DA neurons, because iontophoretic application of DA antagonists (1) or the systemic administration of DA blockers in rats following kainic acid-induced lesions of the striatum (42) have been reported to cause an activation of DA neuron firing similar to that observed after systemic DA antagonist administration to intact rats (4). This observation suggested that systemically administered DA antagonists activated DA cells via the blockade of this neurotransmitter at postsynaptic sites. However, more recent data suggested that this was not the case. Thus, infusion of large amounts of DA blockers directly onto DA neurons using pressure ejection was found to activate DA cell firing (68). Furthermore, acute transection of the striatonigral feedback pathway 1 hour prior to testing did not alter the ability of DA antagonists to activate DA cell firing (57). In addition, recent evidence obtained during in vitro intracellular recordings from DA neurons demonstrated that L-DOPA administration inhibited DA cell firing in control media but was ineffective when the decarboxylation of L-DOPA to DA is blocked by carbidopa administration (46). Furthermore, DA antagonists are capable of activating DA neuron firing when administered to the in vitro slice (58). Therefore, this evidence provided substantial support for the tonic inhibition of DA neuron activity via autoreceptor stimulation by locally released DA within the confines of the substantia nigra. One question that remained, however, was whether the dopaminergic inhibition occurred through autoinhibition of the cell releasing the DA, or whether the DA released by individual cells formed a "pool" of neurotransmitter that inhibited neighboring DA neurons as well. Although not conclusive, studies showing the development of DA cell autoreceptor supersensitivity following partial 6-OHDA-induced lesions (56) suggests that lesion-induced loss of neighboring DA neurons will decrease the biophase concentration of DA within the substantia nigra, leading to supersensitivity of the residual neurons. The pharmacology and biophysics of this mode of regulation is covered in detail in the chapter by Chiodo, Freeman, and Bunney.

DA neurons receive afferent inputs from a number of brain regions. However, the majority of the inputs to the substantia nigra arise either directly or indirectly from the striatum. DeLong and co-workers (2) have described two efferent projection systems from the striatum: a direct pathway and an indirect pathway. Recent studies show that these afferent systems may form a parallel feedback regulatory input to the substantia nigra DA-containing neurons as well.

As reviewed in detail previously, stimulation of the striatum will evoke a short-latency inhibitory postsynaptic potential (IPSP) in zona compacta dopaminergic neurons and in zona reticulata neurons (27). However, stimulation of the striatum in trains will actually cause an activation of DA neuron firing. This DA cell activation occurs in concert with an inhibition of zona reticulata GABAergic neurons that normally inhibit DA neuron activity (21, 27). These GABAergic inhibitory neurons may represent collaterals of nigrothalamic neurons (9) or a short-axon interneuron located near the zona compacta (10). Therefore, striatal stimulation appears to exert two actions on DA neuron electrophysiology: a direct GABAergic inhibition and an indirect disinhibition. A similar dichotomy between interneurons and dopaminergic cells has been proposed to exist within the ventral tegmentum as well (38).

The striatum is known to send a large number of GABAergic inhibitory projections to the globus pallidus, which, in turn, sends GABAergic fibers to the subthalamic nucleus (36, 40, 59). Recent physiological and metabolic studies have shown that extensive lesions of the nigrostriatal DA system activate the striatopallidal pathway, thereby disinhibiting the subthalamus (49, 50). Single-pulse stimulation of the subthalamus has been shown to produce short-latency excitation of both dopaminergic and nondopaminergic neurons within the substantia nigra (34, 52, 65), and glutamatergic excitatory postsynaptic potentials (EPSPS) have been evoked in DA neurons recorded in vitro (38, 47). However, maintained stimulation of the subthalamic nucleus has been shown to exert substantially different effects. During early periods of subthalamus activation, there is an excitation of zona reticulata neuron firing and a concomitant inhibition of DA cell activity. However, as the activation diminishes, the DA neuron regains its baseline firing rate. Moreover, this is accompanied by a significant activation of burst firing (65). In contrast, lesions of the subthalamus were found to produce a regularization of firing pattern primarily in the DA cells located in more lateral regions of the substantia nigra () (65). The activation of this feedback system in rats with large DA depletions (49, 50) may therefore underlie the increase in burst firing without a similar activation of firing rate that occurs in the residual DA neurons recorded after such lesions (35).

The above evidence suggests that the striatonigral feedback system has at least two major modes of regulation over the DA neuron, with each of these modes apparently having opposing actions on DA cell activity. However, it should be noted that in each case, stimulation of the striatum or the subthalamus was done in such a manner as to activate a large number of neurons projecting to the nigra. Instead of causing opposing influences, a more likely scenario is that the system is capable of independently activating feedback inhibitory or feedback excitatory influences in order to rapidly and potently alter DA neuron activity. For example, during periods of DA demand the striatal system that activates the direct subthalamic excitatory input to the DA neuron may be co-activated with the striatonigral GABAergic projection that preferentially inhibits GABAergic interneurons, thereby disinhibiting the DA neuron. Conversely, during periods of decreased DA demand, the striatum may activate the excitatory projections from the subthalamus to the GABAergic interneurons while also activating inhibitory GABAergic afferents to the DA neurons (65). Whether the anatomy of the system coincides with this model remains to be examined using highly specific tract tracing techniques.

As shown by the data reviewed above, the striatum can potentially exert substantial feedback inhibitory control over the substantia nigra dopaminergic neurons. On the other hand, transection of striatonigral interconnections has been reported to cause little or no change in DA neuron activity (56, 57). In contrast, transection of striatonigral connections appears to exert prominent effects on nigral cell activity in a rat in which a demand has been placed upon the DA system, such as during chronic haloperidol administration (3). Therefore, it appears that the striatonigral feedback system exerts little exogenous regulation on DA neuron activity except in cases in which a pharmacological perturbation has caused an increased DA demand on the system.

For some time, studies of the activities of DA neurons in awake animals and its correlation with behavior have involved studies of the rodent or in several cases the cat. In these studies, DA cells have been shown to behave similar in some respects to those recorded in the anesthetized animal, showing a somewhat similar proportion of neurons spontaneously discharging action potentials and firing in both single-spiking and burst-discharge modes (11, 12). However, one property that was somewhat unusual is the report that DA neurons in awake rats are capable of rapid switching from a single-spiking to a burst firing pattern of activity (12), as contrasted to their comparatively stable firing patterns recorded in the anesthetized animal (25, 26). Furthermore, several studies have correlated this burst-driven activation with the presentation of behaviorally relevant stimuli—that is, stimuli requiring a behavioral response (48).

Recent advances in the behavioral relevance of DA neuron discharge have been made in studies of DA cells recorded in the primate. By using an organism that is capable of acquiring and performing complex tasks, specific questions regarding the precise role of the DA system in behavioral processes may be elucidated. In general, studies have shown that, as in the rat, DA neurons are activated when the primate is presented with a behaviorally relevant stimulus requiring a response. However, the DA system appears to be primarily involved during the acquisition phase of this event, with little or no activation present when the animal is overtrained on the task (61). Therefore, as suggested by studies of the DA system in reward behavior (60), DA neuron discharge appears to be necessary during the phase in which the behavioral relevance of the stimulus is being defined.

The DA neuron exhibits at least three electrophysiological dimensions along which it is capable of activating postsynaptic transmitter release: a change from inactive to spontaneously discharging state, an increase in the rate of spontaneous spike discharge, and an alteration from single-spiking to a burst-discharge mode of activity. A fourth state, one of depolarization block, occurs when an abnormally large demand, typically mediated by an exogenous pharmacological agent, causes the system to be overdriven to the state of inactivity. DA neurons have been shown to be under a number of complex regulatory influences. These factors range from characteristics related to their membranes and their morphology to properties derived from the influences of their afferents. Thus, the membrane properties that drive spike activity in the DA neuron enable it to fire spontaneously in the absence of afferent activation, and to accommodate rapidly to maintained excitatory influences and regain its basal level of activity. Dendrodendritic interactions appear to underlie synchronization of activity as well as provide autoreceptor-mediated local feedback modulation of the activity of the neuron, with drug administration or lesions causing alteration in the sensitivity of the autoreceptor, resulting in the restoration of basal activity states within this system. The influences of afferent and feedback processes further define the responsivity of this system: They apparently play a minor role when the system is in a basal state, but provide a substantial impact on the discharge pattern and level of activity when the system is challenged pharmacologically or physiologically.

In each case, these multiple homeostatic factors appear to act in concert to enable the DA neuron to maintain a basal level of activity. This would be consistent with experimental evidence showing that the DA system is under extensive homeostatic regulation and can readily compensate for maintained changes in afferent drive or residual capacity. Therefore, it would appear that the discharge of DA neurons is not involved in the maintenance of long-term changes in the nigrostriatal system, but instead only alters their activity states phasically in response to short-term demand (35). Therefore, by preserving basal levels of electrophysiological activity, the DA neuron conserves the dynamic range of electrophysiological response that can be rapidly drawn upon when required by the system. Indeed, the property of the DA system to maintain basal activity levels when faced with a tonic demand but respond rapidly with a massive phasic increase when required is consistent with the behavioral studies showing that DA neuron discharge is only altered when the system is changing state, as occurs with the learning of a novel stimulus paradigm. On the other hand, when the DA system is driven in a manner such that the biochemical compensations are inadequate to respond to the demand [resulting in a tonic activation of DA cell firing (e.g., with chronic neuroleptic treatment or near maximal DA depletions)] or if one of the primary regulatory responses is thwarted (e.g., administering a DA blocker to a system recovered from a lesion), the ultimate consequence appears to be depolarization block of DA cell discharge and a reinstatement of the deficit state () (37)). However, under conditions in which the DA system appears to be showing an abnormal increased responsivity, such as schizophrenia (19), the induction of depolarization block may be the most effective way to circumvent behaviorally mediated activation of this system (19, 20, 37).

published 2000