Dopamine Receptors: Clinical Correlates

Philip Seeman

Brain dopamine receptors are the primary targets in the treatment of schizophrenia, Parkinson's disease, and Huntington's chorea. Dopamine receptor antagonists, or neuroleptics, are effective in blocking hallucinations (including L-DOPA-induced hallucinations) and delusions which occur in these diseases, while dopamine receptor agonists such as bromocriptine are effective in alleviating the signs of Parkinson's disease.

The discovery and direct detection of dopamine receptors originally depended on the existence of stereoselective antipsychotic drugs (154). While the antipsychotic drugs then permitted the discovery of dopamine receptors, the cloned dopamine receptors are now in turn facilitating the search and discovery of more selective antipsychotic drugs and antiparkinson drugs. A current hope is that more selective therapy can be achieved by developing drugs that are more selective for a particular type of dopamine receptor (151).

The original classification of two main groups of dopamine receptors, namely D1-like and D2-like dopamine receptors, still stands (178). All five known and cloned dopamine receptors fall into these two classes. The number of amino acids in each of the five dopamine receptors in the human is given in , along with the various mutations for each receptor.

The dopamine D1-like receptors include D1 and D5 (183,184), with two pseudogenes related to D5 (111). The pseudogenes are so named because they code for incomplete forms of the dopamine D5 receptor, wherein the protein stops at 154 amino acids instead of an expected full-length dopamine D5 receptor of 477 amino acids. These 154-amino acid proteins are not expected to have the usual receptor functions.

There are three types of dopamine D2-like receptors: D2, D3 and D4. As summarized in , the dopamine D2 receptor has two main variants, D2short and D2long, with an additional three variants which are less prevalent in the human population: Valine96Alanine (wherein Alanine replaces Valine), Proline310Serine, and Serine311Cysteine.

The difference between D2short and D2long is a 29-amino acid segment coded by an exon which is missing (i.e., spliced out) in D2short (125). In brain and endocrine tissues, the long form of D2 is more commonly expressed than the short form (74,83,173,192). Although the abundance of both D2short and D2long is increased by denervation (108) and by antipsychotic drug administration (17; see also 98,138.), the relative abundance of D2short and D2long in schizophrenic brain tissues is the same as human control brain tissues (137).

D2short and D2long differ in their ability to influence intracellular events. For example, D2long is more effective than D2short in inhibiting the production of inositol phosphate (4). D2long is also more effective than D2short in increasing intracellular calcium ions, because D2long is less readily inhibited by protein kinase C (91).

Dopamine, on the other hand, is more effective on D2short, compared to D2long, in stimulating the binding of [35S]guanosine-5'[g-thio]triphosphate to the receptor-associated G protein (45). It appears that the 29-amino acid segment inserted in D2long confers a specific interaction between the dopamine D2 receptor and the alpha-i-2 subunit of the Gi protein responsible for the inhibition of adenylate cyclase (58).

There are also some pharmacological differences between D2short and D2long. Although dopamine agonists and antagonists have a similar affinity for D2short and D2long (85), D2short has a 2- to 5-fold higher affinity for clozapine and several substituted benzamides (22,97). In addition, dopamine D2short receptors are more readily internalized into the cell (66).

No variant of D2 has been clearly linked or associated with schizophrenia (126), including the D2Serine311Cysteine variant, which has been extensively investigated (2,25,29,65,168,187).

As summarized in , the dopamine D3 receptor (50,175) has several variants in the human: D3long (50), D3short (43), D3Serine9Glycine, D3nf (nonfunctional), D3-transmembrane-4-region-deleted, and D3-transmembrane-3-region-deleted.

The D3nf variant is interesting insofar as it is consistently found in all post-mortem schizophrenic brain tissues (143), but it is also found in post-mortem brain tissues from patients who died with affective disorders (143).

Despite the extensive search for a D3 variant that may be associated with schizophrenia, none have been found (26,44,57,72,81,107,118,139,140,188,198,203).

The dopamine D4 receptor (190,191) has many variants in the human (88). Each variant has a different number of repeat units (located in the 3rd cytoplasmic loop of the receptor protein). Each repeat consists of 16 amino acids. Most humans (50–70%, depending on the ethnic composition) have four repeats, and this receptor is named the dopamine D4.4 receptor. No human has yet been found to have dopamine D4.0, D4.1 or D4.9 receptors (88).

In addition to the number of repeats varying in the human dopamine D4 receptor, there are at least 21 different types of repeat units (88), each repeat type being identified as a, b, q, h, e, z, g, k, n, s, r, m, i, d, i, l, o, p, or x. For example, one person may have a dopamine D4.4 receptor with repeat types a, b, q and z, such that his/her D4.4 receptor may be named D4.4abqz. The first and last repeat types (repeat types a and z) are the same in all individuals (88).

The different phenotypes and genotypes of the human dopamine D4 receptor preclude a simple nomenclature for the D2-like receptors, such as D2A for D2, D2B for D3, or D2C for D4, as has been suggested by others (33,170). Otherwise, the D4.4 receptor would be D2C.4 or D2C.4abqz, both excessively complex to remember or to use verbally. Moreover, there is at present considerable confusion on the use of D2A and D2B. For example, D2B has been used to refer to D2short (30), to D3 (170), and to D4 (33).

As summarized in , there are three interesting variations of the D4 receptor in humans. First, in approximately 10–15% of the world population of black or African individuals, there is a single amino acid change (from valine to glycine) at position 194 (162). This mutation causes the D4 receptor to be insensitive to dopamine, clozapine and olanzapine (90). Second, in about 8% of the population there is an absence of four amino acids (alanine, serine, alanine, glycine) near the beginning of the receptor (24). Third, in about 2% of the population the D4 receptor has a frame-shift in the coding sequence such that the receptor is truncated down to 98 amino acids (119); this D4 variant may, therefore, be a pseudogene.

At present, none of the variants or mutations of the dopamine D4 receptor have been clearly linked to any disease, including schizophrenia (176), Tourette's or bipolar disorder, or clearly associated with clinical response to psychotropic medications.

A summary of the human dopamine receptor amino acid sequences and their human variations is given elsewhere (146,164).

The sensitivities of the cloned dopamine receptors to dopamine agonists and antagonists are generally the same as those for dopamine receptors in native tissues. These potencies are summarized in and .

For dopamine agonists, however, it has been difficult to obtain accurate comparisons of the agonist sensitivities of the cloned receptors with those in native tissues. This is because dopamine receptors in native tissues can adopt either a high-affinity state or a low-affinity state for an agonist, with the high-affinity state being the functional state.

Various tissue culture cells, moreover, vary in their ability to reveal the high-affinity state. For example, African green monkey kidney tissue culture cells, which are often used to express the cloned dopamine receptors, do not have sufficient or appropriate G protein subunits to allow the high-affinity state of the receptor to exist. Thus, the high-affinity data for agonists in are incomplete at present. However, an important consistent difference between D1 and D5 is that dopamine itself is about five to ten times more potent at the D5 receptor.

Accurate values for many of the agonist potencies on the dopamine D2, D3 and D4 receptors are not currently available, for the reason mentioned above. Nevertheless, some important selectivities are clearly emerging. For example, bromocriptine is about two orders of magnitude more potent at D2 and D3, compared to D4, as indicated in .

A D3-selective antagonist is nafadotride (141), and PD 128,907 is a selective D3 agonist (59). Although (+)7-OH-DPAT has often been used to label D3 receptors, (+)7-OH-DPAT has only about a 7-fold higher affinity for D3 compared to D2 (see ). Hence, it is not possible to label D3 receptors selectively in vivo by means of [3H]7-OH-DPAT (18,84).

In the case of antagonists, clozapine is 28-fold more potent at D4 than at D2, as shown in . Raclopride, on the other hand, is about two orders less avid for D4, compared to D2 or D3. Therefore, because raclopride discriminates between D2/D3 and D4 receptors, it is useful for the selective blockade of D2 and D3, permitting the study of D4 or D4-like receptors or binding sites (165). A further difference between D2 and D4 is the action of sulpiride. Although S-sulpiride is ten times more potent at D2 compared to R-sulpiride, both of these enantiomers have equal affinity for D4 (152).

The dopamine hypothesis of schizophrenia proposes that brain dopamine synapses are overactive in schizophrenia (146). This overactivity may stem from either an excess release of dopamine (1,15) or an overactive response by the dopamine receptors.

Considerable support for the hypothesis of dopamine overactivity in schizophrenia relies on the fact that neuroleptics block dopamine D2 receptors in direct relation to their clinical antipsychotic potencies (146,154), as shown in .

The therapeutic concentrations of antipsychotic drugs in spinal fluid or in the water phase of the serum used in have previously been summarized (146), with the exception of clozapine for which new data have appeared (113). The concentration of clozapine in the spinal fluid is 20% of that in the plasma (113; see also Refs. 87,186). Hence, the total concentration, C, of free clozapine and norclozapine in the spinal fluid is given by C = 20% x 274 ng/mL x 3.06 nM/ng/mL (1 + 56% x 44 nM/100 nM) = 208 nM, where 20% is the free and unbound fraction of clozapine in the plasma (113), 274 ng/mL is the average concentration of clozapine in plasma in patients taking clinically effective doses of clozapine (60,61,89,102,127,130), 3.06 nM/ng/mL is the factor to convert ng clozapine/mL plasma into nM, 56% is the average norclozapine fraction of clozapine (60,61,89,102,127,130), and 44 nM and 100 nM are the dissociation constants of clozapine and norclozapine, respectively, at the dopamine D2 receptor (), and indicating that norclozapine has a potency at the D2 receptor which is 44% (i.e., 44 nM/100 nM) that of clozapine.

The correlation in , therefore, indicates that all the clinical neuroleptic potencies correlate with their ability to block D2. No such correlation exists between the clinical antipsychotic doses and the dissociation constants at the dopamine D1, D3 or D4 receptors.

With respect to the D1 receptor, it is known that a number of antipsychotic compounds, such as haloperidol, occupy few (less than 5%) or none of the dopamine D1 receptors at clinically effective doses (40,41). Dopamine D1 receptor antagonists, moreover, are not clinically useful as antipsychotics (31,32,70).

Concerning the dopamine D3 receptor, most antipsychotic drugs have less affinity for D3 than for D2 receptors. For example, haloperidol has an affinity for D3 which is one-twentieth that of D2 (); clozapine has one-fourth the affinity; and loxapine, melperone, molindone, perlapine, seroquel and sertindole have about one-half to one-third the affinity (). Remoxipride is D2-selective insofar as it is twenty times more avid at D2 compared to D3 (). Hence, the clinical action of remoxipride may be appropriately attributed to D2 and not D3 (or D4), in accordance with the data in . In addition, some evidence indicates that clinically relevant doses of antipsychotic drugs apparently do not occupy D3 receptors in vivo in animals (144).

One of the reasons for the wide range in dissociation constants for a particular neuroleptic () is that the dissociation constant depends on the radioligand used (153,164,166). This is shown in , using several different radioligands of differing lipid solubilities (or tissue/buffer partition coefficients). The neuroleptic dissociation constant rises with highly fat-soluble radioligands, presumably because the more fat-soluble radioligand adheres more avidly to the receptor and is difficult to displace by the competing neuroleptic.

Hence, the radioligand-independent dissociation constant is that value which is extrapolated to a low partition value, as illustrated in . This extrapolated value exactly agrees with the dissociation constant obtained by using the [3H]neuroleptic directly, as shown in . The radioligand-independent dissociation constants for neuroleptics are listed in .

The radioligand-independent dissociation constants () place the neuroleptics into groups: those with "loose" binding and those with "tight" binding at dopamine D2 receptors ().

In addition, these data () reduce the range of variation between the clinical doses and the neuroleptic dissociation constants for D2, as shown in . This figure (right side) also illustrates that the important clinical feature is not so much the clinical dose but the final free concentration of neuroleptic in the plasma water, as indicated by the data for chlorpromazine and thioridazine.

As a result of many studies using positron emission tomography to measure the occupancy of D2 receptors in neuroleptic-treated patients (16,37,38,40,115,132), there is a clear relation between clinical signs and D2 block.

Antipsychotic action occurs at about 65–80% D2 block, while extrapyramidal Parkinsonian signs and akathisia occur when at least 80% of D2 receptors are occupied (). The imaging data are primarily based on the binding of [11C]raclopride to the striatal D2 receptors in volunteers or patients. This binding primarily reflects D2 because there are negligible amounts of D3 and D4 receptors in the human striatum. At present, the measurement of D2 receptors in non-striatal regions of the brain is not feasible because these regions have low D2 densities.

Supplementing and independently confirming the brain imaging data, the occupancy of brain dopamine D2 receptors, under clinically therapeutic conditions, can now also be indirectly determined from the antipsychotic drug concentrations in the spinal fluid and by using the radioligand-independent dissociation constants in , as follows.

The dissociation constant, K, in is defined as the antipsychotic concentration required for 50% occupation of the receptor in the absence of dopamine or any other ligand. In reality, however, because the antipsychotic drug must compete with dopamine within the synaptic space, the antipsychotic concentration to block 50% of the receptors in the presence of dopamine (C50%) will be higher than that in the absence of dopamine, in accordance with the equation C50% = K x [1+D/DHigh], where D is the effective dopamine concentration in the synapse and DHigh is the dissociation constant of dopamine at the high-affinity state of the dopamine D2 receptor.

(Although the effective concentration of dopamine in the synapse is not known, it is considered to be between 4 nM and 45 nM. The basal level of synaptic dopamine in the rat nucleus accumbens has been estimated to be 4 nM (128). At a firing frequency of five impulses per second, the synaptic dopamine in rat striatum has been estimated to be about 200 nM in the first few milliseconds and then rapidly averaging to 6 nM within 200 milliseconds (71). The resting synaptic dopamine concentration in the human striatum has been indirectly estimated to be about 45 nM (80). The dopamine D2 receptor can exist in either a high- or a low-affinity state for dopamine, wherein the high-affinity state, DHigh, is the physiologically functional state (48). The value for DHigh is approximately 6 nM (23,27,150,96,50,175). It appears, therefore, that the synaptic concentration of dopamine, D, as well as the value for DHigh, are both approximately around 6 nM.)

Despite the unknown value for the dopamine concentration in the synaptic space, a reasonable assumption is that the effective concentration of dopamine in the synapse is similar to the dissociation constant of dopamine at DHigh. If this single assumption is valid, namely that D = DHigh, the final result from the above equation is that the antipsychotic concentration for 50% block in the presence of dopamine, C50%, will be equal to 2 x K.

In addition, the fraction, f, of D2 receptors occupied by an antipsychotic at any concentration C will be C/(C+K). Using this formula, the concentration of antipsychotic drug needed to occupy 75% of the D2 receptors (i.e. C75%) is about three times higher than that required to occupy 50% of the receptors, C50%. In other words, C75% = 3 x C50%, or C75% = 6 x K.

Hence, using the K values in , the antipsychotic C75% concentrations were calculated according to the latter equation and graphed in . These values were graphed vs. the therapeutic concentrations of the antipsychotic drugs in the cerebrospinal fluid or in the plasma water (i.e., corrected for drug binding to the plasma proteins).

The therapeutic concentrations used in have previously been summarized (146), with the exception of clozapine, for which new data have appeared (113). As outlined in a previous section, the combined concentrations of free clozapine and norclozapine in the spinal fluid averages 208 nM in patients taking clinically effective doses of clozapine.

The heavy line in is the line for identical values between the C75% and the therapeutic free concentration of neuroleptic in the spinal fluid or plasma water. It may be seen that the C75% values for all the antipsychotic drugs (for which data are available) fall on this line of identity. In other words, based on the single assumption that the effective synaptic dopamine concentration is similar to the K value of dopamine for the high-affinity state of D2 (for which there is reasonable evidence, as indicated above), one may conclude that 75% of dopamine D2 receptors are occupied by neuroleptics under therapeutic conditions.

A corollary of these arithmetic considerations is that D2 blockade can be higher than 75% in those brain regions where there is less dopamine in the synapse, such as might occur in the non-striatal regions of the brain.

The results in are not a coincidence, because similar calculations for the dopamine D3 and D4 receptors did not reveal any constant percent occupancy for all the neuroleptics, with some antipsychotics (such as remoxipride or S-sulpiride) occupying less than 10% of D3 or D4 receptors.

As shown in , the neuroleptic potencies at D2 appear to be clustered into three groups. One group of antipsychotic compounds has dissociation constants below 1 nM, indicating that they bind tightly to the dopamine D2 receptor. This group includes trifluoperazine, chlorpromazine, thioridazine, haloperidol, fluphenazine, risperidone and raclopride. These compounds are known to elicit extrapyramidal signs and symptoms.

A second group has dissociation constants between 1.5 and 7 nM, in the range of the values for the dissociation constant of dopamine for the high-affinity state of the D2 receptor. This group includes molindone, loxapine, olanzapine and sertindole, all of which elicit mild levels of Parkinsonism.

A third group of antipsychotics has very high dissociation constants ranging from 30 to 100 nM, indicating that these compounds bind loosely to D2. These compounds include the so-called atypical neuroleptics, such as melperone, seroquel, perlapine, clozapine and remoxipride, all of which elicit few or no Parkinsonian signs and symptoms.

Because the physiologically functional state of D2 is its high-affinity state (48) [for which dopamine has a dissociation constant of 2–7 nM], it appears that the tightly bound neuroleptics (with dissociation constants below 1 nM) readily elicit Parkinsonism by tightly blocking D2. The neuroleptics with K values above 30 nM elicit little or no Parkinsonian symptoms because they are not effective in competing against dopamine at the high-affinity state of D2. A dominant factor, therefore, in determining whether a particular neuroleptic elicits Parkinsonism is its radioligand-independent dissociation constant relative to that for dopamine at the D2 receptor.

In the case of clozapine, however, the occupancy of D2 receptors in patients is much lower than that for haloperidol or other traditional neuroleptics. For example, antipsychotic doses of haloperidol occupy 50–80% of dopamine D2 receptors (40,68,120-122), using [11C]raclopride, [18F]methylspiperone, [11C]methylspiperone or [18F]fluorethylspiperone (28,69,92,93).

The situation is somewhat different for clozapine, however, because it has a very high dissociation constant at D2 (), making clozapine weak in competing against radioligands used in brain imaging. For example, summarizes the reports on the proportion of D2 receptors occupied by clozapine, using different radioligands in different clinical laboratories. The D2 occupancy is graphed vs. the dissociation constant of the radioligand. Specifically, the percent of D2 receptors occupied by clozapine is very low (between 0% and 22%) when using [11C]methylspiperone, [18F]methylspiperone or [18F]fluorethylspiperone (28,69,92,93) but is progressively higher when using [123I]iodobenzamide or [11C]raclopride with their higher dissociation constants (117).

The data in , therefore, indicate that extrapolation to the dissociation constant for dopamine at its high-affinity state reveals that clozapine occupies high levels of dopamine D2 receptors under ordinary clinical conditions in the absence of any radioligand. This extrapolation is similar in principle to that for the in vitro data in . The conclusion is that clozapine occupies about 70–80% of dopamine D2 receptors (in the striatum, but this proportion is possibly higher in the non-striatal regions) under normal therapeutic conditions.

The density of dopamine D2-like receptors is elevated in post-mortem schizophrenic brain tissues (146,156), which is compatible with the idea of dopamine overactivity in schizophrenia. For example, control human striata (putamen and caudate nucleus) have an average D2 receptor density of 12.9 pmol/g, using [3H]spiperone. In striata from schizophrenic patients, the density of [3H]spiperone sites is elevated by 56%, to a value of 20.2 pmol/g (156).

Such elevation of D2 receptors, however, is only found in vivo using [11C]methylspiperone (51,114,116,201) but not with [11C]raclopride (39). The scientific basis underlying these discordant findings is not clear. The simplest explanation is that radiolabeled spiperone congeners attach to a monomer form of D2 (109), while radiolabeled benzamide congeners (such as raclopride or nemonapride) bind to both the monomer and dimer forms of the dopamine D2 receptor, as shown in for [125I]iodoazido-nemonapride attaching to the D2 monomer (MW of 48,000) and to the D2 dimer (MW of 98,000) [(82,109,204)].

The same result occurs in clinical PET of dopamine D2 receptors in control human subjects. The density of D2 sites labeled by [11C]raclopride is 30–80% higher than the density of spiperone-labeled sites in the human striatum. For example, the density of D2 sites labeled by [11C]methylspiperone or [3H]spiperone is 16 pmol/g (13 to 24 pmol/g; refs. (51,116,202), while the density of D2 sites labeled by [11C]raclopride or [3H]raclopride is 25 pmol/g (16–35 pmol/g; Refs. 39,51,62,202). In other words, because benzamides attach to both monomers and dimers of D2, the density of [11C]raclopride sites in human subjects exceeds the density of D2 monomer sites labeled by [11C]methylspiperone.

In schizophrenia, therefore, the density of [11C]methylspiperone sites rises (116,201), reflecting an increase in monomers, while the density of [11C]raclopride sites remains the same, indicating that the total population of D2 monomers and dimers does not change.

Although schizophrenic tissues reveal elevated D4-like binding sites, found by subtracting the densities of two different radioligands (105),161,163,182), this elevation does not reflect true D4 receptors but could represent an increase in the number of dopamine D2 receptor monomers.

Recently, D4-selective radioligands ([125I]NGD 94-1 [99] and two other new [3H]ligands [Seeman, unpublished]) indicated few or no d true D4 dopamine receptors in either human control or schizophrenic striata (163,166). Hence, the elevated D4-like sites in schizophrenia, although not representing genuine D4 receptors, may possibly reflect altered features of D2 or D2-like receptors, as a result of biochemical modification of the receptor after its synthesis.

An increase in dopamine D4 receptors has been found, however, in the entorhinal cortex and the substantia nigra from post-mortem schizophrenic brains, using [3H]NGD 94-1 (78).

It is important to note that the densities of many receptors in the human nervous system progressively fall by 1–6% per decade. , for example, illustrates that the elevated density of D2 receptors in striata from schizophrenics falls as the patients age (155,156). The rate of decline for schizophrenic men is about three times faster than that for the schizophrenic women. Clinically, these data may be related to the slow but steady clinical improvement found in schizophrenic men as they age (145).

Patients with schizophrenia turn more often towards their left side, in contrast to control individuals, who turn to their left or right equally often (9,10,94). The same is true for a small subgroup of nonschizophrenic but severely psychotic patients (95). Such rotational preference is not related to handedness in normal controls (11) and is not related to handedness or medication in psychotic patients (9,94,95). There is, however, a correlation between the severity of delusions and the extent of left turning (14).

On a neurochemical basis, it has been found that rotation of the body is commonly toward the brain side containing less dopaminergic activity. This holds for both animals (53) and humans (12,13,52).

In order to study why patients with schizophrenia turn more often toward their left side, we examined whether there was an asymmetry in the densities of dopamine D2-like receptors in the left and right post-mortem schizophrenic brain striata. Using [3H]nemonapride to label dopamine D2-like receptors (112), we found that the density of receptors on the right side was higher than that of the left side in 13 out of 16 pairs (81%) of striata from schizophrenics (). This right-side finding with [3H]nemonapride was identical to that found previously with [3H]spiperone (135). Thus, if the extra dopamine D2-like receptors on the right are functional and active, then the individual would turn toward the left, in keeping with the general principle that both animals and humans rotate toward the side where the brain hemisphere is relatively hypodopaminergic.

There are many psychomotor activities where the dopamine D1 and D2 receptors are cooperative or synergistic (158). A functional example of this D1-D2 interaction is shown in for the treatment of Parkinson's disease, wherein D1 and D2 agonists cooperate to elicit a smoother and maintained antiparkinson effect on patients with Parkinson's disease.

One molecular basis for the D1-D2 interaction may be through the bg subunits of the G proteins that are associated with both of these dopamine receptors (158). That is, the bg subunit of the D1-associated Gs protein is identical to the bg subunit of the D2-associated Gi protein. It is thought that the bg subunits "shuttle" between the alpha subunits of the Gs and the Gi proteins. This shuttling would result in a D1 and D2 interaction.

Most interesting is the observation that the D1-D2 link is reduced or absent in approximately two-thirds of post-mortem striatal tissues from schizophrenic patients and from late-stage Huntington's disease patients (158). Normally, the D1 receptor appears to keep D2 in its low-affinity state, possibly via the shuttling of the bg subunits, mentioned above. Any reduction in the D1 influence on D2, therefore, would be expected to result in D2 retaining its high-affinity (or functional) state. Hence, clinically, a reduced influence of D1 on D2 would be expected to result in psychotic symptoms.

An additional indication of a significant abnormality in the coupling between D2 and G proteins may be seen in , where the binding of [3H]raclopride to D2 in schizophrenic tissue was not affected by guanine nucleotide, in contrast to post-mortem tissues from patients with other brain diseases (161).

The D1-D2 interaction may take place through a neuron-neuron interaction, where D1 and D2 reside on different neurons, or through an intracellular interaction where D1 and D2 coexist in the same cell. There is evidence for both the cell-cell interaction hypothesis (3,5,46,63,129, 177,193) and for an intracellular mechanism (42,76,77,104,134,142,148,169,185,189,193,194,205). In those instances or diseases where the D1-D2 interaction is based on a neuron-neuron interaction, degeneration of one group of neurons could then result in a loss of the D1-D2 interaction.

For the present purpose, an atypical antipsychotic drug is here defined as one that elicits little or no Parkinsonism at doses that are clinically effective in reducing psychotic symptoms.

What is the receptor basis of these atypical antipsychotic drugs, such as clozapine?

As outlined above in connection with , the neuroleptic affinities for D2 are clustered into three groups.

The antipsychotics with dissociation constants below 1 nM (trifluoperazine, chlorpromazine, thioridazine, haloperidol, fluphenazine, risperidone and raclopride) bind tightly to the dopamine D2 receptor and thus readily elicit extrapyramidal signs and symptoms.

A second group has dissociation constants between 1.5 and 7 nM (molindone, loxapine, olanzapine and sertindole), close to the dissociation constant of dopamine for the high-affinity state of the D2 receptor. Thus, the competition between these antipsychotics and endogenous dopamine is in balance, and these antipsychotics, therefore, elicit mild or dose-dependent levels of Parkinsonism.

The third group (the atypical neuroleptics: melperone, seroquel, perlapine, clozapine and remoxipride) has high dissociation constants (between 30 and 100 nM), thus binding loosely to D2 and eliciting little or no Parkinsonian signs and symptoms.

Hence, whether a particular antipsychotic elicits Parkinsonism depends on its radioligand-independent dissociation constant relative to the value of 2–7 nM for dopamine at the high-affinity state of the D2 receptor. Thus, a low affinity for D2 is a common factor for atypical antipsychotic drugs.

There are also other hypotheses which attempt to explain the receptor basis of the atypical antipsychotic drugs. One such hypothesis is the "SDA" (serotonin-dopamine-antagonism) hypothesis, which suggests that the block of serotonin 2A receptors in addition to dopamine D2 receptors helps prevent or minimize Parkinsonism (64,86,100,101,102,103,179).

As summarized elsewhere (123,167), however, the relevance of serotonin 2A receptor blockade in the treatment of schizophrenia is controversial and has not been confirmed.

The following findings support the SDA hypothesis:

Much evidence, however, does not support a therapeutic role for serotonin-2A receptor blockade in the treatment of psychosis or in the alleviation of Parkinsonism/catalepsy, as follows:

Practical benefits are emerging from the dopamine hypothesis and from the cloning strategies (147,148). As just noted above, there is a clear relation between clinical signs and D2 receptor blockade. Hence, the art of psychiatry is rapidly becoming the quantitative science of psychiatry. In addition, the development of new clozapine-like neuroleptics (which do not cause Parkinsonism or tardive dyskinesia) is both desirable and possible. A new generation of selective neuropsychopharmacology is on the horizon.

I thank Dr. S. Kapur for generously providing unpublished data, and I thank Dr. Teresa Tallerico for reviewing this manuscript. Supported by Mr. and Mrs. Robert Peterson of The Peterson Foundation (University Park, FL) and NARSAD (National Alliance for Research in Schizophrenia and Depression), the Ontario Mental Health Foundation, the Medical Research Council of Canada, the Medland family (the late J. Aubrey Medland and the late Helen Medland, Pamela O'Rorke, Janet Marsh and David Medland), the Stanley Foundation of the NAMI Research Institute, and the National Institute on Drug Abuse, USA.

published 2000