Biochemical Pharmacology of Midbrain Dopamine Neurons

Robert H. Roth and John D. Elsworth

Over the past quarter century, the dopamine cells of the ventral mesencephalon have been among the most intensively studied of the chemically defined neuronal groups. At first believed to be two rather homogeneous groups of tyrosine-hydroxylase-positive cells localized to the substantia nigra and ventral tegmental area, midbrain dopamine neurons have turned out to be heterogeneous populations consisting of dopamine neurons projecting to a variety of overlapping areas in the telencephalon. Since the original histochemical description of the monoaminergic neurons in brain during the mid-1960s, anatomical studies have focused on the delineation of the telencephalic projections of the midbrain dopamine neurons. Over the past 30 years, advances in neuroanatomical methodologies have led to an appreciation of the striking heterogeneity of the mesencephalic dopamine neurons both in terms of their connectivity and in reference to the transmitters and proteins localized to these neurons.

Neurochemical, pharmacological, and electrophysiological investigations have revealed corresponding heterogeneities in midbrain dopamine neurons and have led to an appreciation of the diversity of both intrinsic and extrinsic control mechanisms that regulate these systems. A conceptual approach that utilizes these anatomical and neurochemical heterogeneities to define functional subpopulations of dopamine cells seems most likely to yield information that may enable the rational pharmacological manipulation of dopaminergic function. The dopamine neurons of the ventral mesencephalon have been designated the A8, A9, and A10 cell groups. There are no clear anatomical boundaries between the neurons of the different cell groups (16, 25) and the dopamine neurons forming these populations appear at the same time during development (see Development of Mesencephalic Dopamine Neurons in the Nonhuman Primate: Relationship to Survival and Growth Following Neural Transplantation). These facts, coupled with the overlap in projection fields of the A8, A9, and A10 cell groups, have led to the suggestion that these neurons be collectively designated the mesotelencephalic dopamine system (16, 25). Such anatomical overlap in the innervation pattern of dopamine neurons on to terminal field targets suggest that there may exist a considerable degree of heterogeneity not only between but also within a given terminal field. Such heterogeneities are apparent in the striatum in which there are two distinct nerve terminal dopamine systems termed the islandic and diffuse systems. These two types of dopamine innervations originate from different populations of neurons within the midbrain dopamine cell groups and exhibit both different basal dopamine turnover rates and different responsiveness to dopamine agonists and antagonists (27, 30). Recent anatomical data suggest that the nucleus accumbens can also be divided into two compartments: (i) a core region related to the caudate-putamen and (ii) a shell region associated with the limbic system, receiving enriched innervation from A9 and A10 dopamine cell groups, respectively (33, 72). These two accumbal segments can be further distinguished on the basis of their response to both environmental and pharmacological challenges (15). Such heterogeneities in dopamine innervation, coupled with histochemically distinct compartments in the dorsal and ventral striatum with which the different dopamine systems are in register, suggest that there may be multiple levels of input/output organization that may be reflected in the functional characteristics of these pathways.

Midbrain dopamine neurons also project extensively to cortical sites, including the prefrontal cortices, cingulate cortex, and certain allocortical sites. Recent data indicate that dopaminergic fibers innervate neocortical sites that were previously thought to be devoid of dopamine input such as the visual and association cortices. Although it is not clear to what degree dopaminergic innervation patterns differ between species, the neocortical dopamine system appears to be considerably more extensive in primates than in rodents. There appear to be extensive connections among the dopamine terminal regions and also efferents from the terminal field regions on to the midbrain cell body areas forming feedback loops. Such pathways are currently the focus of considerable attention because these interconnections are believed to function in the integration of dopaminergic activity within the telencephalon.

While it is apparent that midbrain dopamine neurons exhibit anatomical, biochemical, and pharmacological heterogeneity, their functional organization generally reflects features of transmitter dynamics that are shared by all dopamine neurons. These features have been most thoroughly studied in the nigrostriatal pathway (52) and are summarized below (see Figure 1).

Dopamine synthesis, like that of all catecholamines in the nervous system, originates from the amino acid precursor tyrosine, which must be transported across the blood–brain barrier into the dopamine neuron. A number of conditions can affect tyrosine transport, diminishing its availability and thus altering dopamine formation. The rate-limiting step in dopamine synthesis once tyrosine gains entry into the neuron is the conversion of L-tyrosine to L-dihydroxyphenylalanine (L-DOPA) by the enzyme tyrosine hydroxylase. DOPA is subsequently converted to dopamine by L-aromatic amino acid decarboxylase. This latter enzyme turns over so rapidly that DOPA levels in the brain are negligible under normal conditions. Because of the high activity of this enzyme and the low endogenous levels of DOPA normally present in the brain, it is possible to dramatically enhance the formation of dopamine by providing this enzyme with increased amounts of this substrate. Because the levels of tyrosine in the brain are relatively high and are above the Km for tyrosine hydroxylase, it is not feasible, under most circumstances, to significantly augment dopamine synthesis by increasing brain levels of this amino acid. However, as noted below, there are some exceptions to this generality. Because tyrosine hydroxylase is the rate-limiting enzyme in the biosynthesis of dopamine, this enzyme sets the pace for the formation of dopamine and is particularly susceptible to physiological regulation and pharmacological manipulation. Endogenous mechanisms for regulating the rate of dopamine synthesis in dopamine neurons primarily involve modulation of the activity of this key enzyme.

Calcium-dependent release of dopamine from the nerve terminal is thought to occur in response to invasion of the terminal by an action potential. The extent of dopamine release appears to be a function of the rate and pattern of firing (8,29). The burst-firing mode leads to an enhanced release of dopamine. Dopamine release is also modulated by presynaptic release-modulating autoreceptors. In general, dopamine agonists inhibit, whereas dopamine antagonists enhance, the evoked release of dopamine. Much evidence gathered over the last 20 years has indicated that dopamine can be released not only from nerve terminals but also from dendrites of dopamine neurons that originate in the mesencephalon. The enzymes necessary for synthesis of the dopamine are located in perikarya and dendrites of the neurons, and dopamine release has been shown to be calcium-dependent and tetrodotoxin-sensitive. A variety of drugs (e.g., haloperidol, amphetamine) have been found to produce qualitatively similar effects on dopamine synthesis, release, or metabolism in terminal and cell body regions, although the magnitude of the response has tended to be greater in the terminals. Based on the effects of apomorphine or haloperidol on the rate of disappearance of dopamine in the substantia nigra after alpha-methylparatyrosine, it has been suggested that dopamine release in the substantia nigra is not regulated by local dopamine autoreceptors (44). More recently, however, it has been shown that intra-nigral administration of either a D2 dopamine receptor agonist or antagonist is capable of eliciting a synchronous decrease or increase in dopamine release, respectively, in both the substantia nigra and the striatum (68).

Several distinct differences in dopamine biochemistry have been noted between the terminal and dendrites of nigrostriatal dopamine neurons. For instance, the substantia nigra possesses a pool of dopamine that has a much faster turnover than the terminals in the striatum. In addition, in contrast to the striatum, it appears that a considerable proportion of released dopamine in the substantia nigra is taken up into nondopaminergic cells. Another important difference is that nigral dopamine release is partially reserpine-insensitive, and it appears that dopamine in the dendrites of nigrostriatal dopamine neurons is stored both in classical storage vesicles and in smooth endoplasmic reticulum.

Dopamine is released from proximal dendrites in the substantia nigra para compacta of nigrostriatal neurons and from distal dendrites that arborize in the substantia nigra para reticulata. Dopamine released in the pars compacta can act at D2-like autoreceptors located on soma and dendrites of nigrostriatal neurons to reduce their firing rate. Dopamine released in the pars reticulata may act in a paracrine fashion on D1-like receptors to increase release of gamma-aminobutyric acid (GABA) from GABAergic striatal efferents that innervate the substantia nigra pars reticulata. Increased GABA release in the pars reticulata in turn is inhibitory to neurons that project to other regions (e.g., nigrothalamic neurons). A behavioral effect of dopamine-induced GABA release in the pars reticulata appears to be facilitated locomotion. In fact, evidence has accumulated recently to suggest that part of the antiparkinsonian effects of L-DOPA treatment may involve activation of D1 receptors in the substantia pars reticulata (50).

Dopamine nerve terminals possess high-affinity dopamine uptake sites which are important in terminating transmitter action and in maintaining transmitter homeostasis. Uptake is accomplished by a membrane carrier which is capable of transporting dopamine in either direction, depending on the existing concentration gradient. The dopamine transporter is a unique component of the functioning dopamine nerve terminal, which plays an important physiological role in the inactivation and recycling of dopamine released into the synaptic cleft by actively pumping extracellular dopamine back into the nerve terminal. The action of dopamine at synapses is terminated by a high-affinity reuptake into presynaptic nerve terminals via the action of Na+- and Cl--dependent transporter protein. There is no evidence to date to suggest that different midbrain dopamine neurons have unique dopamine transporters. Although some heterogeneity in dopamine transporter proteins have been detected, this heterogeneity may be the result of differential glycosylation of the same protein (46; see also The Dopamine Transporter: Potential Involvement in Neuropsychiatric Disorders).

Autoreceptors can exist on most portions of dopamine cells, including the soma, dendrites, and nerve terminals (53). Stimulation of dopamine autoreceptors in the somatodendritic region slows the firing rate of dopamine neurons, whereas stimulation of autoreceptors located on dopamine nerve terminals results in an inhibition of dopamine synthesis and release (70). Somatodendritic autoreceptors may also regulate dopamine release and synthesis by changing impulse flow. Thus, somatodendritic and nerve terminal autoreceptors work in concert to exert feedback regulatory effects on dopaminergic transmission. Three types of autoreceptors can be defined according to their functional effects: impulse modulating autoreceptors, release modulating autoreceptors, and synthesis modulating autoreceptors. In general, all dopamine autoreceptors can be classified as D2 dopamine receptors (see also Molecular Biology of the Dopamine Receptor Subtypes and Dopamine Receptors: Clinical Correlates).

Available data suggest that nerve terminal and somatodendritic autoreceptors have similar pharmacological properties. Both are relatively more sensitive to dopamine agonists than postsynaptic dopamine receptors and exhibit similar pharmacological profiles. It is conceivable that a single receptor protein might be responsible for modulating such diverse functions as transmitter release, tyrosine hydroxylation, and action potential generation. For example, all of these functions may share a common sensitivity to the initiating signal (hyperpolarization) triggered by autoreceptor occupation. However, it is more likely that different second messenger systems are ultimately responsible for transducing the signal of autoreceptor occupation into changes in dopamine release, synthesis, and impulse generation.

A number of differences also exist between D2 dopamine receptors in different brain regions. For example, D2 dopamine receptors in the nucleus accumbens do not inhibit adenylate cyclase, whereas the striatum appears to contain both D2 dopamine receptors that inhibit adenylate cyclase and D2 dopamine receptors that act independently of adenylate cyclase. The striatal D2 dopamine receptors which are not coupled to adenylate cyclase appear to be located on nerve terminals, whereas those D2 dopamine receptors that inhibit adenylate cyclase are located on intrinsic neurons. The D2 dopamine receptors not coupled to adenylate cyclase activity are likely those that mediate inhibition of dopamine and acetylcholine release in the striatum and nucleus accumbens. These releasemodulating D2 dopamine receptors may work by increasing potassium conductance. Indeed, if any signal transduction proves universal to all D2 dopamine receptors, it may be the ability to open potassium channels and thereby hyperpolarize target cells. This has been observed for postsynaptic D2 dopamine receptors on striatal neurons, for somatodendritic dopamine autoreceptors, and for pituitary lactotrophs (70). Terminal excitability studies also suggest that stimulation of nerve terminal autoreceptors may exert a hyperpolarizing effect (63). However, there are exceptions to this generalization. For example, stimulation of D2 dopamine receptors on the terminals of hippocampal accumbens neurons increases their terminal excitability, suggesting that D2 dopamine receptors at this site exert a depolarizing action (71). Thus, it appears that there is no consistent feature of D2 dopamine receptor function.

Autoreceptors also differ from postsynaptic D2 dopamine receptors with respect to their interaction with D1 dopamine receptors. It is now well-established that D1 and D2 dopamine receptors act synergistically in many behavioral and electrophysiological models (13). Stimulation of D1 receptors appears to play an enabling role; that is, D1 receptor occupancy is necessary for the full expression of the functional effects of postsynaptic D2 dopamine receptor stimulation. Interestingly, this does not appear to be the case for either synthesis-modulating or impulse-modulating autoreceptors (64), where synergy with D1 receptors is not observed.

There are many potential sites at which drugs can influence the function of midbrain dopamine neurons (69). The potential sites for modulation of dopamine function are illustrated in Figure 2. For the purpose of discussion, drug effects can be divided into three broad categories: (i) non-receptor-mediated effects exerted on presynaptic function; (b) dopamine-receptor-mediated effects; and (iii) effects mediated indirectly as a result of drug interactions with other chemically defined transmitter systems that modulate the function of dopamine neurons. The relative importance of the latter two potential sites of drug action will vary among different dopamine systems dependent upon factors such as the presence or absence of autoreceptors, the efficiency of postsynaptic-receptor-mediated neuronal feedback loops, and the nature of the afferent inputs impinging upon the dopamine neurons in question or upon their terminal fields.

There are several stages in the life cycle of dopamine where drugs can influence transmitter dynamics (see Figure 1). Many useful pharmacological tools are available for modifying dopaminergic activity and manipulating dopaminergic function. However, most of these agents are not very selective for dopaminergic synapses and will interact with other catecholamines (norepinephrine and epinephrine) systems and in some cases with other monoamine (5-hydroxytryptamine) systems as well. The major exception to this generality is that there are drugs which target the specific monoamine transporters. Recently, uptake blockers in the GBR series have been characterized which appear to be highly selective for the dopamine transport complex and should prove to be valuable experimental tools perhaps leading to the development of therapeutic agents which can selectively augment dopamine function or be useful as diagnostic aids for visualization of the integrity of dopamine systems in vivo. In fact, striking results have been obtained with several new cocaine derivatives such as 3b-(4-fluorophenyl)tropane2b-carboxylate (CFT) and 3b-(4-iodophenyl)tropane2b-carboxylate (b-CIT), which exhibit high affinity for the dopamine transporter. These agents have been used in autoradiographic experiments and employed in PET and SPECT studies to image the striatal dopamine transporter in both normal and parkinsonian monkeys and humans (21, 26, 36, 40, 42). These studies have demonstrated (a) loss of striatal dopamine transporters in both experimental and idiopathic Parkinson's disease and (b) the restoration of dopamine transporter density associated with the presence of nigral grafts in the caudate of transplanted MPTP monkeys with improved behavioral functions.

Drugs that affect dopamine receptors can be subdivided into two groups: (i) receptors on nondopamine-containing cell types, which are usually referred to as postsynaptic receptors because they are postsynaptic to a dopamine-releasing cell (site 1), and (ii) receptors on dopamine cells or their processes, which are termed autoreceptors to indicate their sensitivity to the neurons own transmitter (sites 2 and 3). Dopamine agonists and antagonists may act on both types of dopamine receptors to elicit biochemical changes in the metabolism of dopamine and alter the functional output of dopaminergic systems (Figure 2). The overall effect of a drug on dopaminergic activity will depend on both its pre- and postsynaptic effects. This can be illustrated by considering the interaction of a reversible dopamine antagonist with various types of dopamine receptors. Blockade of postsynaptic dopamine receptors is usually considered to be the primary effect of neuroleptic administration; at a given synapse, this will result in a decrease in dopaminergic transmission. However, for some dopamine neurons (e.g., nigrostriatal cells) this immediate effect is opposed by compensatory feedback mechanisms. For example, blockade of presynaptic autoreceptors by neuroleptics will result in increased synthesis and release of dopamine from the dopaminergic terminal. The resulting increase in synaptic dopamine may act to competitively overcome the blockade of dopamine receptors produced by the antagonist. In addition to this local feedback mechanism, long-loop neuronal feedback pathways can also come into play. As noted above, striatal neurons communicate with dopamine cell bodies in the substantia nigra by way of striatonigral feedback loops. During periods of enhanced postsynaptic dopamine receptor stimulation, these loops exert negative feedback effects on dopamine cell firing. Conversely, then, blockade of postsynaptic receptors by antagonists will produce a compensatory increase in dopamine cell firing. Because dopamine synthesis, release, and turnover in dopamine neurons are coupled to firing rate and firing pattern under most conditions, this results in increased dopamine release. Similar to the effects elicited by autoreceptor blockade, the resulting changes in synaptic dopamine will ultimately contribute to lessening the effectiveness of the antagonist at blocking both pre- and postsynaptic dopamine receptors.

Even though the consequences of receptor blockade by an antagonist differ at autoreceptors and postsynaptic receptors, if the dopamine antagonist in question is equipotent in blocking receptors at both sites, the functional output of the dopamine system should be diminished. If the antagonist has a more potent effect on postsynaptic receptors, the blockade of dopaminergic function should be prolonged and more effective, because it will not be overwhelmed or competitively reversed by synaptic dopamine. On the other hand, if autoreceptor blockade predominates, the resultant increase in transmitter outflow will competitively antagonize the blockade at postsynaptic receptors and compromise the drug-induced blockade of dopaminergic function. These considerations suggest that neuroleptics can exert a multiplicity of effects on dopamine neurons and that the net outcome of these effects will depend, in part, on the relative importance of pre- and postsynaptic dopamine receptors in regulating the activity of a particular group of dopamine neurons. Unfortunately, less is known about the neuronal feedback pathways from mesolimbic and mesocortical target sites back to dopamine cell bodies in the A9 and A10 regions. Indirect biochemical and electrophysiological evidence suggests that feedback pathways are less important for mesocortical and mesolimbic dopamine neurons than for nigrostriatal dopamine cells. These considerations, as well as the lack of (or diminished number of) synthesis and impulse-modulating autoreceptors on subpopulations of dopamine neurons, are important to keep in mind when attempting to predict the biochemical responsiveness of particular dopamine systems to dopamine receptor blockade.

Dopaminergic function can be influenced by modulating afferent input to the dopamine cell soma or dendrites by blocking or mimicking the effects of transmitter release by the afferent terminals (site 4). Function can also be modulated by modifying afferent input to the dopamine nerve terminal (site 5), which can lead to alterations in the release of dopamine and colocalized peptides from the dopamine nerve terminals. Pharmacological modulation at sites 4 and 5 will depend upon the chemical nature of the afferent inputs to the dopamine neurons in question.

The previous section indicated that the functional output of a given dopamine neuron depends on both (a) intrinsic regulatory properties such as the presence of autoreceptors and (b) extrinsic regulatory properties such as the nature of afferent inputs to cell body regions. In the following section, consideration will be given to differences exhibited by subpopulations of mesotelencephalic dopamine neurons with respect to these parameters and how these heterogeneities might explain differences in the responsiveness of various dopamine systems to both pharmacological and environmental manipulations.

Many of the differences in responsiveness among various midbrain dopamine neurons can be explained in part by differences in their autoreceptor function. The terminals of all midbrain dopamine neurons examined to date—nigrostriatal, mesoaccumbens, mesoprefrontal, mesocingulate, mesoentorhinal, and mesoamygdaloid— have been found to possess release modulating autoreceptors (53, 70). This is not the case for synthesis- and impulse-modulating autoreceptors. For example, whereas most nigral and many ventral tegmental dopamine neurons possess somatodendritic-impulse-modulating and nerve-terminal-synthesis-modulating autoreceptors, the dopamine neurons which project to the prefrontal and cingulate cortices as well as those projecting to the amygdaloid nucleus appeared either to have a greatly diminished number of these receptors or to lack them entirely (5, 41, 53). Of interest is recent evidence which suggests that dopamine-synthesis-modulating autoreceptors appear transiently during development in the prefrontal cortex and that they may belong to the D1 class of receptors (62). The functional significance of the transient expression of the D1 autoreceptor in the prefrontal cortex is unknown. However, in adult rats the absence or diminished numbers of these important modulatory receptors on mesoprefrontal and mesocingulate dopamine neurons, and perhaps also on mesoamygdaloid neurons, appears in part to be responsible for some of the unique characteristics of these neurons when compared to other mesotelencephalic dopamine neurons which possess autoreceptors (nigrostriatal, mesolimbic, and mesopiriform systems) (51). For example, electrophysiological studies have revealed that midbrain dopamine neurons lacking somatodendritic autoreceptors possess a higher basal rate of physiological activity and exhibit a different pattern of activity characterized by a greater degree of burst firing (5, 6). These electrophysiological studies complement biochemical studies which demonstrate that the turnover rate of dopamine in the cingulate and prefrontal cortices is greater than that observed in the piriform cortex or in the projection fields of other midbrain dopamine neurons which possess autoreceptors (51).

Autoreceptors also appear to play a prominent role in controlling the response of various dopamine projections to acute and chronic treatment with dopamine antagonists such as haloperidol, as well as in controlling their response to dopamine agonists (6, 51). While dopamine neurons possessing autoreceptors exhibit dramatic alterations in parameters such as firing rate, dopamine synthesis, and dopamine metabolism in response to acute administration of dopamine antagonists and agonists, mesoprefrontal and mesocingulate dopamine cells are relatively unresponsive both biochemically and electrophysiologically to these agents. For example, administration of haloperidol produces large increases in dopamine synthesis and the accumulation of dopamine metabolites in nigrostriatal and mesoaccumbens and mesopiriform dopamine terminals but has only a modest effect on mesoprefrontal, mesocingulate, and mesoamygdaloid dopamine neurons. Dopamine cell bodies in A8, A9, and A10 areas also exhibit significant differences in their biochemical responsiveness to neuroleptics which may be related to autoreceptor distribution (52).

Acute treatment with haloperidol significantly increases dopamine synthesis in the medial and central substantia nigra, whereas the lateral substantia nigra is unaffected. Within the ventral tegmental area, the lateral area (primary source of the mesolimbic efferents) exhibits a significant increase in synthesis in response to haloperidol, whereas no increase is observed in the medial sector of the ventral tegmental area (primary source of the mesocortical innervation). This biochemical finding is most likely related to the fact that mesoprefrontal and mesocingulate dopamine neurons lack both synthesis-modulating and somatodendritic autoreceptors. No haloperidol-induced increase in dopamine synthesis is observed in either the medial or lateral sectors of the retrorubral field (A8 region), suggesting that A8 dopamine neurons may also lack synthesis-modulating autoreceptors. Failure of the lateral sector of the substantia nigra to respond to haloperidol may similarly reflect an absence or diminished number of somatodendritic autoreceptors in this region.

Although most studies have dealt with the acute effects of neuroleptics on dopamine neurons, chronic studies may be of considerably greater clinical relevance because both the therapeutic effects and extrapyramidal side effects produced by repeated administration of antipsychotic drugs in patients appear only after a latency of several weeks. Interestingly, dopamine neurons which lack autoreceptors respond very differently to chronic treatment with dopamine antagonists such as haloperidol (51). Following chronic treatment with haloperidol, tolerance develops to the elevation of dopamine metabolites levels elicited by a challenge dose of the dopamine agonist in nigrostriatal, mesolimbic, and mesopiriform dopamine neurons. Furthermore, the time course for the development of tolerance closely parallels the development of autoreceptor supersensitivity. In contrast, dopamine neurons projecting to the prefrontal and cingulate cortices appear to be relatively resistant to the development of biochemical tolerance. These studies in rodents have been extended to nonhuman primates, where tolerance development is observed in caudate, putamen, and olfactory tubercles but not in the dopamine systems projecting to the frontal and cingulate cortices (4). Thus, it seems likely that autoreceptors may be involved in the development of biochemical tolerance to antipsychotic drugs after long-term administration. The lack of biochemical tolerance observed in mesoprefrontal and mesocingulate dopamine neurons parallels the lack of tolerance to the therapeutic effects of antipsychotic drugs observed in patients and suggests that sustained biochemical alterations in mesocortical dopamine neurons may be related to the persistent therapeutic actions of these agents.

It is noteworthy that electrophysiological studies have also revealed differences in the responsiveness of dopamine cells to chronic antipsychotic drug administration (11). After repeated administration of antipsychotic drugs to rats, the number of actively firing dopamine cells in both A9 and A10 cell groups is decreased. This state of quiescence, which is thought to reflect depolarization block, appears to result from tonic depolarization of these neurons and can be reversed by hyperpolarizing current or neurotransmitters which hyperpolarize dopamine cells. These treatments help to repolarize the cell and thus restore its ability to fire. Certain dopamine cells appear to be resistant to depolarization block following repeated administration of antipsychotic drugs (11). These are the dopamine cells projecting to the prefrontal or cingulate cortices. Although this observation makes it tempting to speculate that autoreceptors play some role in depolarization block, numerous other mechanisms are likely operative. Dopamine metabolites or ratios of parent amine to metabolite have not proven to be useful biochemical markers of depolarization block. However, monitoring of in vivo tyrosine hydroxylation by following the short-term accumulation of DOPA after inhibition of DOPA decarboxylase has provided a useful biochemical correlate of depolarization block in dopamine projections following chronic administration of typical and atypical neuroleptics (18).

The enhanced susceptibility of mesoprefrontal and mesocingulate dopamine neurons to precursor control of transmitter synthesis may also be related to the absence of dopamine autoreceptors. These dopamine neurons selectively increase their synthesis rate following administration of physiologically relevant doses of precursor tyrosine (51). Precursor dependency has been suggested to be closely related to the physiological firing rate of catecholamine neurons. The mesoprefrontal and mesocingulate dopamine neurons exhibit the highest firing frequency and the most bursting of the mesotelencephalic dopamine cells, possibly because they lack impulse modulating somatodendritic autoreceptors (5). Thus, it is not surprising that this subset of dopamine neurons is most susceptible to precursor regulation of transmitter synthesis.

The findings outlined above have focused primarily on differences between mesocortical, mesolimbic, and nigrostriatal dopamine neurons. Recent findings, however, suggest that dopamine neurons terminating in the striatum must also be considered a heterogeneous population. Dopamine nerve terminals in the striatum can be divided into two distinct groups referred to as the islandic and diffuse systems (30). These two systems exhibit different basal rates of transmitter turnover and respond in a fashion quantitatively different from that of dopamine agonists and antagonists (27). Furthermore, because these two distinct dopamine systems appear to be in register with distinct histochemical compartments, it is likely that the two dopamine systems differentially regulate output characteristics of the striatum.

The striatum also exhibits other types of regional variations in pharmacological responsiveness. For example, the magnitude of the increase in dopamine metabolite accumulation and dopamine synthesis elicited by dopamine antagonists exhibits a twofold regional variability in the striatum: Areas in the striatum which exhibit the greatest increase in both parameters receive their dopamine innervation from the medial and central regions of the A9 dopamine cell group. Areas exhibiting low responsiveness are predominantly innervated by A10 and A8 neurons. Those areas intermediate in responsiveness receive mixed A9 and A10 inputs. Because differences in biochemical responsiveness of different dopamine cell body regions correlates with striatal heterogeneity, it appears likely that certain dopamine neurons which innervate the striatum may lack, or have a different density of, synthesis-modulating autoreceptors. This distinction is readily lost when the entire striatum is examined as a homogeneous structure. As indicated earlier, anatomical data suggest that the nucleus accumbens can also be parceled into two compartments, namely, a core and a shell region (33). These two compartments can be distinguished on the basis of anatomical markers, such as differences in the density of a number of transmitters and receptors and differences in their efferent projections. Recent anatomical findings suggest that the dopamine innervations of core and shell can also be distinguished on the basis of vulnerability to neurotoxic challenges (7). Using biochemical methods, dopamine innervations of the nucleus accumbens core and shell regions have been further characterized and shown to exhibit significant differences indicating that they can be distinguished on the basis of their response to both environmental and pharmacological challenges (15). The biochemical data are consistent with the anatomical data indicating that the dopamine innervation of the nucleus accumbens core is associated with the nigrostriatal system, whereas that of the nucleus accumbens shell is more closely related to the mesolimbic system.

Dopamine neurons also differ in the nature of their interaction with other neurotransmitter systems in the brain. A prototypic example of such interactions is that which occurs between dopamine and acetylcholine in the striatum. Axonal release of dopamine in the striatum inhibits acetylcholine release from cholinergic interneurons; conversely, acetylcholine appears to facilitate dopamine release within the striatum. While such presynaptic regulation of dopamine release is of great importance in the striatum, it appears that this type of interaction is regionally specific; although cholinergic interneurons are present in the nucleus accumbens, the olfactory tubercle, and the medial prefrontal cortex, there does not appear to be a functional dopamine/acetylcholine link in these mesoaccumbens or mesocortical areas. Thus, transmitter interactions such as those occurring at the presynaptic (i.e., independent of dopaminergic impulse flow) level can differ across various brain regions by virtue of the presence or absence of a particular neuron and receptors for its transmitter or physiological factors of local importance.

The ventral tegmental area receives a large number of afferents from telencephalic, diencephalic, and hindbrain regions (47). Moreover, afferents to the ventral tegmental area do not innervate the region diffusely but rather innervate specific nuclei within the ventral tegmental area (47), although dopamine neurons are distributed throughout the region. For example, the nucleus interfascicularis, situated in the most medial aspect of the ventral mesencephalic tegmentum, receives afferents from the medial habenulae and raphe, but does not appear to receive prominent projections from telencephalic dopamine-rich areas (47). Projections to the ventral tegmental area from the nucleus accumbens are relatively sparse, and GABAergic projections from the nucleus accumbens to the ventral tegmental area are predominantly restricted to the anteromedial ventral tegmental area (66). the relatively distinct termination pattern of afferents to the ventral tegmental area, coupled with the topographic organization of the A10 dopamine neurons onto the telencephalon (25), suggests that specific afferents to the ventral tegmental area may regulate subsets of midbrain dopamine neurons, such as those innervating the prefrontal cortex.

A number of recent studies have focused on the chemical nature and differences of afferent input to the somatodendritic region of subpopulations of A10 dopamine neurons in the ventral tegmental area (38). Many of these studies have suggested that subpopulations of A10 dopamine cells can be differentially modulated by afferent inputs. For example, experimental evidence suggests that substance P terminals in A10 may play an important role in modulating the activity of mesocortical dopamine neurons, whereas substance K, a related member of the tachykinin family of peptides, appears to be more important in modulating mesoaccumbens dopamine neurons (17). Thus, it is tempting to speculate that selective substance P or substance K antagonists might be useful in the differential modulation of mesoprefrontal and mesoaccumbens dopamine activity. Other endogenous peptide-containing afferents which innervate the ventral tegmental area also appear to selectively modulate distinct subsets of dopamine neurons. Thus, calcitonin gene-related peptide appears to selectively activate mesoprefrontal but not other mesocortical, mesolimbic, or nigrostriatal dopamine neurons (19).

The presence of GABAergic neurons in the substantia nigra and ventral tegmental area has been shown using immunocytochemical detection of both GABA and the GABA synthetic enzyme glutamic acid decarboxylase. In situ hybridization for glutamic acid decarboxylase mRNA also reveals a moderate density of GABAergic cells within the ventral tegmental area and substantia nigra. Whereas greater than 60% of the neurons in the substantia nigra pars reticulata are GABAergic, only 20% of the neurons in the ventral tegmental area express mRNA for glutamic acid decarboxylase. The intrinsic GABA neurons in the ventral tegmental area synapse on dopaminergic cells as well as project outside the ventral mesencephalon to influence other limbic and motor structures. GABAergic neurons projecting to the ventral tegmental area originate primarily in the nucleus accumbens and ventral pallidum. The GABAergic projection is topographically organized with the shell of the nucleus accumbens and ventromedial ventral pallidum projecting to the ventral tegmental area and the accumbal core and dorsolateral ventral pallidum projecting to the substantia nigra (33). GABAergic efferents to the substantia nigra and ventral tegmental area synapse on both dopaminergic and nondopaminergic neurons, which may explain the paradoxical electrophysiological observation that low doses of GABAA agonists increase the firing frequency of dopamine cells. This increase in dopamine cell firing is associated with a decrease in the firing frequency of nondopamine cells, and it is speculated that the inhibition of GABAergic interneurons results in the disinhibition of dopamine cells and their enhanced firing frequency. GABA does not appear to alter D2 receptor function. However, because GABAB-receptor-mediated membrane polarizations arise from the same potassium conductance as D2 receptors, an increase in GABAB receptor stimulation could compensate for D2 receptor desensitization. This modulatory effect might become most prevalent when D2 receptor desensitization arises from excessive somatodendritic release of dopamine, such as that observed during cocaine sensitization. Elevated synaptic dopamine would stimulate D1 receptors located presynaptically on GABAergic afferents to dopamine cells, thus augmenting GABA release.

Midbrain dopamine cells are also modulated by excitatory amino acids (EAAs). Glutamatergic innervation of the ventral tegmental area arises from three potential sources: EAA projections from the medial prefrontal cortex; the pedunculopontine region; and the subthalamic nucleus. Numerous data demonstrate that EAA input to the ventral tegmental area is at least partially responsible for converting pacemaker-like firing in dopamine cells into burst firing patterns. Prefrontocortical afferents appear to play a major role in regulating N-methyl-Daspartate (NMDA)-dependent burst firing of ventral tegmental area dopamine cells. Electrical stimulation or direct application of glutamate in the prefrontal cortex converts dopamine neuronal activity in the ventral tegmental area into bursting patterns, and cooling the prefrontal cortex converts spontaneous burst firing dopamine cells back to pacemaker-like firing. Thus afferent EAA tone appears to be an important determinant of whether the firing pattern of dopamine cells will be pacemaker-like or bursting (12). This in vivo activity contrasts strikingly with the pacemaker activity of the same dopamine neurons recorded in vitro when the afferent inputs are disrupted. Burst firing of dopamine neurons in vitro can be induced by application of NMDA (37). Pharmacological studies also indicate that the burst firing of midbrain dopamine neurons in vivo result from tonic activation of NMDA receptors by endogenous EAAs (12). It has been suggested that the EAA inputs to midbrain dopamine neurons may constitute a major physiological mechanism in the control of synaptic dopamine levels in target projection areas. In fact, the neuronal discharge pattern, independent of the mean discharge rate, has been shown to exert a potent control over the release of dopamine and colocalized peptides (such as neurotensin) from terminal fields (8, 29). The observation that bursting activity elicits more terminal dopamine release per action potential than pacemaker activity and that bursting elicits release of copeptide transmitters more readily than nonbursting patterns has focused attention on the functional significance of bursting activity in midbrain dopamine neurons (see also New Developments in Dopamine and Schizophrenia).

Some of the differences in pharmacological responsiveness between subpopulations of dopamine neurons appear to be due to differences in the responsivity of these cells to afferent inputs. For example, recent neurochemical studies have directly demonstrated that EAA receptors may be involved in the selective metabolic activation of subsets of dopamine neurons within the ventral tegmental area which project to the prefrontal cortex (38, 39). Stimulation of NMDA receptors more selectively activates the dopamine projection to the prefrontal cortex, whereas stimulation of non-NMDA receptors activates mesoaccumbens and nigrostriatal neurons. The NMDA receptor complex is an EAA-ligand-gated ion channel which is selectively activated by NMDA and regulated at several pharmacologically distinct sites including a high-affinity, strychnine-insensitive glycine binding site. Competitive antagonists of the strychnine-insensitive glycine site which cross the blood–brain barrier have become available, making possible in vivo pharmacological manipulation of the NMDA receptor through this regulatory site. One such high-affinity selective antagonist at the NMDA receptor glycine site, (+)-HA-966, is the recently resolved enantiomer of the drug (±)-3-amino-1-hydroxy-2-pyrolidinone (59). Systemic administration of this agent to the rat has been shown to normalize dopamine neuron firing patterns (57). The neurochemical and behavioral effects of (+)-HA-966 have been studied in several paradigms (restraint stress and conditioned fear) that are known to cause a metabolic activation of mesoprefrontal and mesoaccumbens dopamine neurons. (+)-HA-966 given systemically or injected into the ventral tegmental area prevents the stress-induced increase in dopamine metabolism in the prefrontal cortex without altering the response in the nucleus accumbens (43). Similarly, systemic administration of the noncompetitive antagonist of the NMDA receptor, MK-801, blocked the stress-induced rise in dopamine metabolism in the prefrontal cortex but not the nucleus accumbens. The negative enantinomer of HA-966 did not produce a selective antagonism of the stress-induced dopamine metabolism in the prefrontal cortex. The role of the NMDA receptor and its glycine modulatory site in the rat conditioned-fear paradigm has also been investigated (28). Aversive conditioning results in selective dopamine and serotonergic metabolic activation in the medial and lateral prefrontal cortex, elevation in serum corticosterone, ultrasonic vocalization, and freezing behavior. Pretreatment with (+)-HA-966 abolished the dopamine metabolic activation in the medial and lateral prefrontal cortex. Serotonergic metabolic activation in the medial prefrontal cortex and dopamine metabolic activity in nucleus accumbens were not affected by (+)-HA-966 pretreatment, indicating neurochemical and regional specificity of the (+)-HA-966 blocking effect. Pretreatment with (+)-HA-966 did not effect serum corticosterone elevation, but facilitated extinction of the freezing response. These data indicate that the NMDA receptor complex and associated glycine modulatory site may play an important role in the afferent control of the mesoprefrontal cortical dopamine system during restraint or conditioned fear, suggesting a potential target for pharmacological modulation of this dopamine projection. These results also support previous data, which suggest that the mesocortical and mesoaccumbens dopamine neurons respond to excitatory input through different glutamate receptor mechanisms. Recent studies have demonstrated that microinjection of NMDA antagonists into the ventral tegmental area can prevent the initiation of behavioral sensitization to systemic cocaine (39), which poses the interesting possibility that the NMDAdependent induction of burst firing in ventral tegmental area dopamine cells may be a critical factor in establishing sensitized motor behavior. In view of the observation that EAA afferents from the prefrontal cortex play an important role in burst firing, these data argue that EAA afferents from the prefrontal cortex may have a permissive function in initiating behavioral sensitization. Midbrain dopamine neurons also differ in the types of neurotransmitter systems which influence dopamine turnover at the level of the nerve terminals (seeFigure 2), although comparatively little is known about this type of interaction in mesolimbic and mesocortical projection fields compared to the striatum.

It is now well-documented that the mesoprefrontal cortical dopamine system is activated by certain stress conditions (mild footshock stress and conditioned fear) which do not produce an appreciable metabolic activation of other mesotelencephalic dopamine systems (nigrostriatal mesolimbic or other mesocortical) or central noradrenergic systems. The stress-induced activation of mesoprefrontal dopamine neurons is antagonized by anxiolytic agents such as lorazepam or diazepam, which target the benzodiazepine site of the GABAA receptor. Furthermore, the anxiolytic properties of these agents as well as their ability to block the stress-induced activation of mesoprefrontal dopamine neurons are reversed by benzodiazepine antagonists, implicating the involvement of benzodiazepine GABA receptors in the stress-induced alterations in mesoprefrontal dopamine function (20). The observation that mesoprefrontal dopamine neurons differ from nigrostriatal and mesolimbic dopamine neurons in that they lack impulse modulating and synthesis modulating autoreceptors prompted the speculation that the lack of autoreceptors might account in part for the unique responsiveness of this system to mild stress. However, the experimental evidence to date is not entirely consistent with the correlation between lack of autoreceptors and enhanced responsiveness to stress. For example, dopamine neurons that innervate the cingulate cortex also lack synthesis- and impulse-modulating autoreceptors but are not activated by mild stress. Thus, extrinsic influences over midbrain dopamine neurons such as peptidergic, EAA, serotonergic, and noradrenergic inputs are thought to play an instrumental role in the production of this phenomena (20, 38). Consistent with this speculation, it has been demonstrated that infusion into the ventral tegmental area of monoclonal antibodies directed against substance P effectively prevents the footshock-induced increase in prefrontal cortical DOPAC. It therefore seems likely that distinct afferents to different areas within the mesencephalon may, in part, account for the selective response of various dopamine systems to stress rather than solely the presence or absence of somatodendritic and synthesis-modulating nerve terminal autoreceptors. Also, more recently, it has also been shown, as alluded to above, that administration of the glycine site antagonist of the NMDA receptor, HA966, can block the metabolic activation of the mesoprefrontal dopamine neurons induced by restraint stress and aversive conditioning without altering the metabolic activation which occurs in the mesoaccumbens, suggesting a differential regulation of mesocortical and mesoaccumbens dopamine neurons by NMDA receptors (28; see also Excitatory Amino Acid Neurotransmission, Intracellular Messenger Pathways as Mediators of Neural Plasticity, and Stress).

In addition to being selectively activated by mild stress, mesoprefrontal dopamine neurons also exhibit increased dopamine turnover in response to administration of anxiogenic beta carbolines. These agents are believed to act as inverse agonists at the benzodiazepine recognition site on the GABAA benzodiazepine receptor complex and are known to antagonize the various central effects of benzodiazepines. One of the anxiogenic beta carbolines, FG7142 (N-methyl-b-carboline-3-carboxamide), produces a dose-dependent increase in dopamine metabolism in the prefrontal cortex and ventral tegmental area of rats without causing any significant increase in dopamine metabolism in other mesocortical, mesolimbic, or nigrostriatal sites (51). The metabolic activation observed in the prefrontal cortex is blocked by anxiolytic benzodiazepines and by benzodiazepine antagonists, implicating the involvement of benzodiazepine/GABA receptors in this response (51). Because all midbrain dopamine cells examined to date have been shown to be inhibited by GABA, the selective effects of this beta carboline on mesoprefrontal dopamine neurons suggests that cell bodies of mesoprefrontal dopamine neurons may be wired in a unique manner to the GABA-benzodiazepine-dependent mechanisms in the ventral tegmental area or be enriched in a particular subtype of GABAA receptor. The fact that mesoprefrontal dopamine neurons are uninfluenced or less responsive to autoreceptor-mediated regulation of firing rate may explain why these cells might be more sensitive to the inhibitory effects of GABAergic inputs.

Other evidence suggests that dopamine systems may also be differentially modulated by serotonin and noradrenergic neurons. For example, dopamine utilization is selectively enhanced in the nucleus accumbens but decreased in the prefrontal cortex in a temporally specific fashion following electrolytic lesions of the median raphe. Lesions of the dorsal raphe increase dopamine utilization in the nucleus accumbens but do not alter dopamine turnover in the prefrontal cortex (34). Systemic administration of 8-hydroxy-dipropylaminotetralin (8-OH-DPAT), a 5-HT1a agonist, increases the firing rate and bursting of A10 dopamine neurons (2) and also increases the in vivo release of dopamine from the prefrontal cortex (1). A postmortem increase in dopamine turnover as assessed by measures of the DOPAC/dopamine ratio is also observed in the prefrontal cortex (Rasmusson, Goldstein, and Roth, unpublished observations). Furthermore, dopamine utilization is decreased in the prefrontal cortex but remains unaltered in the nucleus accumbens following 6-OHDA-induced degeneration of the norepinephrine fibers projecting to the ventral tegmental area. Although the ventral tegmental area is known to receive both serotonergic and noradrenergic afferents from brainstem monoamine neurons, it is not clear whether the regulatory interaction described above occurs at the level of the cell bodies of origin of the various dopaminergic projections; interactions at the level of the terminal field are equally possible.

The possibility of an additional type of interaction between dopamine and neuropeptides is suggested by recent findings which indicate that dopamine, like other classical neurotransmitters, is often colocalized in single neurons with neuropeptides and nontransmitter proteins. For example, certain midbrain dopamine neurons contain the peptide cholecystokinin (CCK) (14), whereas other subpopulations of mesencephalic dopamine neurons contain the peptide neurotensin and a third group of dopamine neurons in the ventral tegmental area contains CCK, neurotensin, and dopamine (56). Similarly, dopamine is colocalized in certain mesencephalic neurons with nontransmitter proteins, including acetylcholinesterase, protein-O-carboxymethyltransferase, cytochrome P-450 reductase, and a vitamin-D-dependent calcium-binding protein (see Colocalization in Dopamine Neurons).

The functional significance of such colocalization is not clear. However, because receptors for certain colocalized peptides such as neurotensin are present on dopamine neurons in the midbrain, activity of these dopamine neurons may be regulated by peptide autoreceptors in a fashion analogous to somatodendritic dopamine autoreceptor modulation of impulse flow and dopamine synthesis. Furthermore, because it appears that release mechanisms for dopamine and colocalized peptides may be dissociated under certain conditions (e.g., dependent on the firing pattern and firing frequency of the neuron (8), colocalized peptides may serve as part of a hierarchical array of neuronal regulatory features. In this regard, it is of interest that nerve terminal autoreceptors in the prefrontal cortex have been shown to exert reciprocal effects on dopamine and neurotensin release. Stimulation of dopamine autoreceptors diminishes dopamine release and enhances neurotensin release, whereas blockade of dopamine receptors augments dopamine release and diminishes neurotensin release. The functional implications of these findings for the activity of follower cells in the prefrontal cortex is presently uncertain, but it could allow the prefrontal cortex dopamine neurons to differentially modulate the physiological activity of cortical postsynaptic follower cells.

Although the functional correlates of peptide-amine colocalization in mesencephalic dopamine neurons remains to be clearly established, it appears likely that colocalization of peptides or nontransmitter proteins and dopamine will prove to reliably define certain subpopulations of dopamine neurons. Thus, CCK-dopamine colocalized neurons of the ventral tegmental area project to the caudal, but not rostral, nucleus accumbens. Such distinctions may have important implications for regionally specific function of dopamine in psychiatric and neurological disorders, as well as for the response of specific dopamine systems in these pathological conditions or to pharmacological treatment. The coexistence with neurotensin in particular mesencephalic dopamine neurons has not been observed in primates (9). However, this might not be a permanent phenotype. The possibility of a transient coexpression either in pathological states like schizophrenia or during ontogeny similar to the transient multi- colocalization of tyrosine hydroxylase with peptides observed in the rodent amygdala (somatostatin and substance P) seems plausible. In fact, Hökfelt and collaborators (55) have recently found that the CCK gene is expressed in the midbrain of humans and specifically in the substantia nigra of schizophrenic patients, whereas CCK mRNA is low or nondetectable in the mesencephalon of normal subjects. Although neurotensin and CCK appear to modulate the function of mesotelencephalic dopamine neurons (14, 58), their role in normal brain function or their possible dysregulation in neurological or psychiatric disorders or in stress- or drug-induced sensitization is still unclear. However, the availability of potent, bioavailable antagonists should lead to new insight concerning their importance and help to elucidate the role played by these neuropeptides in both normal and abnormal brain function (14), 32).

Much of the latter part of this chapter has been concerned with the heterogeneity of central dopamine neurons and the implications of these differences in the pharmacology and pathology of dopamine neurons. A summary is presented in .

A number of neurotoxins have been reported to damage certain mesotelencephalic dopamine neurons. Among these toxins are (a) heavy metals such as manganese and lead and (b) a variety of organic compounds such as 3-acetylpyridine. The toxic effects of such compounds, however, are relatively nonspecific and result in the loss of nondopaminergic as well as dopaminergic neuronal elements throughout the brain. 1-Methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) is a neurotoxin which appears to exert its effects in a relatively selective manner to effect a depletion of striatal dopamine stores, and thus result in a parkinsonian syndrome in primates (see Parkinson’s Disease).

Rats are relatively resistant to the neurotoxic effects of systemic MPTP administration. The susceptibility of primate species to MPTP is a striking example of the differences that exist in the biochemistry or pharmacology of rodent and primate central dopamine neurons (). In fact, due to the relative paucity of research on primates, there are doubtless other discrepancies that are not yet known. In areas where distinct differences occur, caution is needed when relating the significance of observations obtained in rodents to human or nonhuman primates.

Initial reports of MPTP-induced neuropathology indicated that the toxin resulted in a selective loss of the nigrostriatal dopaminergic system of primate species, whereas mesolimbic and mesocortical dopamine systems were spared. However, more recent data (22,23) indicate that certain mesolimbic/cortical dopamine neurons are also damaged by MPTP. In contrast to idiopathic Parkinson's disease, in which other monoamine systems are impacted, the susceptibility of serotonergic and noradrenergic neurons to the toxic effects of MPTP appear to be minimal.

The primary toxicity of MPTP appears to be manifested at the terminal field level, because most of the midbrain dopamine cell bodies of origin of the striatal dopamine innervation do not disappear until well after the striatal dopamine depletion is maximal. It has been possible to exploit the temporal disparity between terminal and somatodendritic degeneration to examine the possible preferential vulnerability of different populations of midbrain dopamine neurons. Several studies indicate that the lateral A8 and central dorsal A9 dopamine neurons are preferentially vulnerable to MPTP, and that only later do medial A9 neurons of the substantia nigra show degenerative changes. The dorsomedial A10 dopamine neurons of the ventral tegmental area are also lost following MPTP treatment, consistent with the partial loss of the septal dopamine innervation seen following exposure to the neurotoxin. In addition, there is heterogeneity of dopamine loss in striatal regions after MPTP (22).

The key question generated by the studies of MPTP has been to determine the mechanisms whereby MPTP exerts such selective effects on central dopamine neurons. Not only is there remarkable specificity in the neuronal damage induced by MPTP, but, furthermore, MPTP toxicity is manifested in its complete form only in primate species. Administration of MPTP to infraprimate species, including mice, cats, and dogs, can result in striatal dopamine depletion; however, the loss of striatal dopamine in these species is not of the magnitude observed in primates treated with far lower doses of MPTP. These data suggest that some feature restricted to specific subpopulations of dopamine neurons of primates may render these dopamine neurons more vulnerable to the toxic effects of MPTP. In fact, recent data are consistent with the possibility that the calcium-binding protein, calbindin, might exert a neuroprotective effect on dopamine neurons. The dopamine neurons which do not express calbindin are preferentially vulnerable to Parkinson's disease and to the neurotoxic effects of MPTP. Identification of the corresponding biochemical heterogeneity may allow the development of specific strategies aimed at the prevention of Parkinson's disease. In the interim, the MPTP-treated parkinsonian primate has provided a very useful model in which to examine therapeutic strategies for the treatment of Parkinson's disease. In fact, this model has already been successfully exploited to design, refine, and evaluate neuronal transplantation techniques (23, 49, 60) and to test new pharmacological strategies for the therapeutic management of Parkinson's disease (61).

On the basis of biochemical, anatomical, and electrophysiological studies, there is now compelling evidence that the organization of ascending dopamine neurons arising from the substantia nigra, ventral tegmental area, and retrorubral field is much more complex than originally suspected. The initial classification scheme indicating the existence of three major systems, namely the nigrostriatal, mesolimbic, and mesocortical systems, no longer seems fully appropriate. It is now recognized that distinct dopamine subsystems are responsible for the dopamine innervations of well-defined cortical areas as well as mesolimbic structures. It is also clear that the substantia nigra contains dopamine cells that project not only to the striatum, but to certain cortical and mesolimbic sites as well; furthermore, certain ventral tegmental area neurons contribute to the striatal dopaminergic innervation. These interdigitating dopamine subsystems can be distinguished not only on the basis of their origin with the midbrain dopamine cell groups, but also by virtue of the type of dopamine neurons they comprise (e.g., shape of the cell body and nature of the dendritic arborization, characteristic axonal morphology, presence of axon collaterals, coexistence of identified peptides or enzymes, existence of autoreceptors), by their physiological profile (e.g., firing rate, degree of bursting, transmitter turnover), by their respective specific afferents, and by their specific reactivity to pharmacological treatments or environmental stimuli. In mesocortical dopamine systems, such distinctions have been examined extensively in relation to the medial prefrontal cortical dopaminergic innervation of the rat. Much remains to be done to characterize in a similar way the properties of dopamine neurons innervating other cortical regions as well as to extend these studies in rodents to primates.

The findings discussed above suggest that it may be plausible to design drugs which would selectively target specific dopamine cells through interaction with distinct regulatory features of these neurons. For example, certain subsets of dopamine neurons exhibit striking differences in intrinsic regulatory properties such as the presence of autoreceptors or coexistence with neuropeptides. They also possess distinct differences in extrinsic regulatory properties such as the nature of afferent inputs to their cell body regions or terminal fields. It is conceivable that drugs might be targeted to exploit these differences. It may be possible to take advantage of interactions between specific dopamine systems and other neurotransmitters in order to selectively alter the activity of subpopulations of midbrain dopamine neurons. For example, it appears that substance K and substance P preferentially modulate mesolimbic and mesocortical dopamine neurons, respectively. Thus, if stable, bioavailable specific substance P agonists or antagonists were developed, it might prove possible to selectively manipulate mesocortical dopamine function. Mesoprefrontal and mesolimbic dopamine neurons may also be differentially modulated by serotonergic and noradrenergic afferents originating from brainstem sites as well as by enkephalinergic and other peptidergic afferents. Some midbrain dopamine neurons in the ventral tegmental area appear to be differentially sensitive to excitatory glutamatergic input, thereby allowing their activity to be differentially modulated by NMDA receptors. Differences in afferent input or the responsivity of heteroreceptors on dopamine neurons could provide the rationale for the design of novel agents which would interact selectively with subpopulations of dopamine neurons, or they suggest conditions in which traditional antipsychotic agents might be more effective if used in combination with drugs which affect other transmitter systems or target selective dopamine systems.

In conclusion, there have been considerable advances in the last decade in the understanding of the diversity of central nervous system dopamine neurons and the chemical basis of this diversity. The next challenge is to apply this knowledge to the development of useful diagnostic agents and more selective therapeutic agents for both the evaluation and treatment of neurological and psychiatric disorders which involve dopamine dysfunction.

published 2000