Dopamine receptors (D1-D5)

The classical way to organise dopamine receptors is by their G protein coupling, because coupling sets the direction of the signal ^[1][2]. D1-like receptors (D1, D5) switch on adenylyl cyclase through Gs, or through the neuron-specific variant Golf that does the same job in the striatum, so activating them increases cAMP and the protein kinase A cascade downstream of it. D2-like receptors (D2, D3, D4) do the reverse through Gi/o; they decrease cAMP, and they also open inwardly rectifying potassium channels and close voltage-gated calcium channels, which makes the host neuron harder to fire and quieter at its terminals.

Why coupling matters in practice: the same dopamine surge can excite one population of neurons and silence another, depending on which family that population carries. In the striatum this maps onto the two main output routes. D1-rich neurons form the direct pathway that facilitates movement, D2-rich neurons the indirect pathway that suppresses it. The evidence for that separation is unusually clean; the two populations differ not only in receptor but in the peptide they co-release, striatonigral cells carrying D1 with substance P and dynorphin and striatopallidal cells carrying D2 with enkephalin, and a lesion of the nigrostriatal input moves the two sets of messenger RNA in opposite directions ^[9]. Blocking D2, as antipsychotics do, or refilling the synapse, as levodopa does, shifts the balance between those routes. The D1-like versus D2-like distinction is the lever nearly every dopamine drug pulls.

There is a gene-level echo of the same split that is worth carrying, because it explains a quirk further down the page. The D1-like genes carry no introns in their coding region, so D1 and D5 each come in exactly one protein form. The D2-like genes are interrupted by introns, which leaves room for alternative splicing, and the D2 gene uses that room to produce a short and a long isoform ^[2]. So the family that can be spliced is also the family that supplies the system's autoreceptors. The five genes sit on five different chromosomes and were cloned in a four year burst between 1988 and 1991, starting with D2 ^[5] and finishing with D4 and D5 in the same issue of the same journal ^[7][8].

One caution before the detail. The two-family scheme is a first approximation. All five receptors also signal through G protein-independent routes, chiefly beta-arrestin scaffolds, and a drug can favour one route over the other at the same receptor ^[1][3][12]. Treat cAMP as the headline, not the full account.

Each subtype has a characteristic coupling, distribution and drug relevance. The table below is the reference view; the notes after it cover the details that do not fit in a cell.

Abundance and affinity run in opposite directions. D1 is the most abundant dopamine receptor in the brain, and D5 is the scarcest, yet D5 binds dopamine with roughly ten times the affinity of D1 and is otherwise pharmacologically almost indistinguishable from it ^[8]. A scarce, high-affinity receptor and an abundant, low-affinity one respond to different things: the first can read low ambient dopamine, the second mostly reports phasic bursts.

D2 exists in two forms, and they do different jobs. Alternative splicing produces D2S (short) and D2L (long), differing by a 29 amino acid insert in the third intracellular loop. In mice engineered to lack the long form, D2L turns out to act mainly at postsynaptic sites while D2S serves the presynaptic autoreceptor role, and the catalepsy that Haloperidol normally causes disappears entirely ^[10]. That is about as direct a demonstration as exists that a single splice choice can decide which side of a synapse a receptor works on.

D3 is the limbic member. It was cloned as a D2-like receptor concentrated in the nucleus accumbens and the islands of Calleja rather than the motor striatum, and it was proposed as an antipsychotic target on exactly that basis: hit the limbic receptor, spare the motor one, avoid the movement side effects ^[6]. That promise is still only partly delivered.

D4 is the odd one. Its defining feature at discovery was an affinity for clozapine roughly an order of magnitude higher than D2 or D3, which is what made it interesting at all ^[7]. Its gene also carries a variable number tandem repeat in exon III that changes the length of the third intracellular loop between people, and that polymorphism has been chased through the personality literature for thirty years ^[19][20].

Structures now exist for the pocket itself. The human D3 receptor was crystallised with a D2/D3 selective antagonist in 2010 ^[21] and the D2 receptor with Risperidone bound in 2018 ^[22], the latter revealing an unexpectedly enclosed binding pocket. That matters to anyone designing a subtype-selective ligand rather than another broad blocker.

The core switch is cAMP. Through Gs or Golf, D1-like receptors activate adenylyl cyclase and raise cAMP, which activates protein kinase A, which phosphorylates DARPP-32. That last protein is the amplifier of the system: once phosphorylated on threonine 34 it becomes a potent inhibitor of protein phosphatase 1, so a modest cAMP rise is converted into a large, sustained change in the phosphorylation state of ion channels, transcription factors and glutamate receptors across the cell ^[11]. Through Gi/o, D2-like receptors inhibit adenylyl cyclase, lower cAMP, and push DARPP-32 the other way, while also gating potassium and calcium channels directly. A rise in synaptic dopamine therefore excites D1-bearing neurons and inhibits D2-bearing ones at the same moment.

A crucial wrinkle is the autoreceptor. D2, especially its short splice variant, and D3 sit on dopamine neurons themselves, on the cell body and on the presynaptic terminal. When dopamine binds them it feeds back to slow firing, reduce synthesis and cut further release; they are the system's own volume knob ^[1][10]. This is why low-dose D2/D3 drugs can behave paradoxically. A small dose of an antagonist such as Sulpiride preferentially blocks the more sensitive autoreceptors and can transiently raise dopamine output, while a larger dose blocks postsynaptic receptors and cuts transmission. The same logic runs in reverse for agonists, and it is a large part of why low-dose and high-dose regimens of one molecule can read as opposite drugs.

The second layer is not a G protein at all. Dopamine receptors, D2 in particular, also assemble a scaffold containing beta-arrestin 2, the kinase Akt and the phosphatase PP2A. Recruiting that complex deactivates Akt and so releases GSK-3, on a slower timescale than the cAMP response and through entirely separate machinery ^[12]. Two consequences follow. Lithium's known action on GSK-3 lands in the same pathway from another direction. And a ligand can be biased, favouring the G protein route or the arrestin route at one receptor, which is the idea behind attempts to keep antipsychotic benefit while shedding motor and metabolic side effects ^[1][3]. No biased dopamine ligand has yet proved that promise in people.

Two complications belong here rather than in a footnote. D5 shows meaningful constitutive activity, signalling to some degree with no agonist bound at all, so a neutral antagonist and an inverse agonist are not equivalent drugs at that receptor ^[8]. And dopamine receptors physically associate with others, including adenosine A2A, which sits alongside D2 on striatopallidal neurons and opposes it; that pairing is why an adenosine antagonist can act as an indirect dopaminergic. See adenosine receptors, and Istradefylline for the drug built on it.

Subtype identity only becomes useful once it is attached to an address. The brain's dopamine supply comes from a small number of cell groups with very different destinations, and a drug's side effect profile is usually just the list of pathways it hits that nobody wanted it to hit.

The periphery matters more than most summaries admit. Dopamine receptors sit on renal tubules and blood vessels, where D1-like activation promotes sodium excretion and vasodilation, and D2-like receptors sit on pituitary lactotrophs where they hold prolactin release in check ^[1][2]. They also sit in the area postrema, the vomiting trigger zone in the brainstem, which lies outside the blood brain barrier. That one anatomical fact explains a whole drug class: Metoclopramide blocks D2 there to stop nausea, and causes movement side effects precisely to the extent that it also reaches the brain proper.

Two clinical patterns fall straight out of the table. Blocking D2 everywhere at once buys antipsychotic effect from the mesolimbic route and pays for it with stiffness from the nigrostriatal route and raised prolactin from the tuberoinfundibular route; that is not three side effects but one action landing in three places. Activating D2-like receptors everywhere at once buys motor benefit from the nigrostriatal route and pays with impulse control problems from the mesolimbic one ^[18].

The prefrontal picture does not follow the striatal rules and deserves its own note. Prefrontal cortex is comparatively rich in D1 and D4 and poor in D2, and D1 signalling there follows an inverted U: too little and working memory suffers, too much and it suffers again, with a narrow optimum between. That is the honest reason a stimulant can sharpen one person and scatter another at the same dose, and why more dopamine is not a coherent goal on its own. See the dopaminergic system for the circuit-level view.

Antipsychotics: the cleanest dose-response relationship in psychiatry. The founding observation was that clinical antipsychotic potency across chemically unrelated drugs tracks their affinity for the D2 receptor almost perfectly ^[13]. Imaging later put numbers on it. In first-episode patients given haloperidol, the likelihood of clinical response rose sharply once D2 occupancy passed about 65%, raised prolactin appeared above about 72%, and extrapyramidal side effects above about 78% ^[14]. That is a therapeutic window measured in receptor occupancy rather than milligrams, and it is why dose-finding for this class is a different exercise from dose-finding for most others. Clozapine is the informative exception: at clinically effective doses it occupies only about 38% to 63% of D2, well under the supposed threshold, while occupying 38% to 52% of D1 ^[15]. Whatever makes clozapine work in treatment-resistant illness, plain D2 occupancy is not a sufficient explanation.

Newer agents refine the target rather than raise the dose. Cariprazine is a D3-preferring D2/D3 partial agonist, so it steadies signalling toward a middle value instead of shutting it off, and Aripiprazole applies the same partial-agonist logic at D2. Amisulpride and Sulpiride are nearly pure D2/D3 blockers with no serotonin component, which makes them useful reference compounds whenever a clinical effect is claimed to be serotonergic.

Stimulants and eugeroics. Amphetamine, Methylphenidate, Lisdexamfetamine and Modafinil do not bind dopamine receptors in any meaningful way. They raise synaptic dopamine, mostly by blocking or reversing the dopamine transporter, and let it act across all five subtypes at once ^[1]. The drug itself is agnostic about which receptor answers. See eugeroics for how that differs from a classical stimulant.

Parkinson's disease. Loss of nigrostriatal dopamine is treated either by refilling the synapse with Levodopa or by activating the receptors directly with a D2-like agonist such as Pramipexole or Ropinirole. Amantadine sits alongside both; a weak dopaminergic layered on NMDA antagonism, valued mostly for taming levodopa-induced dyskinesia rather than for its own motor benefit.

Genetics reaches down to temperament, but less firmly than usually claimed. The D4 gene's exon III repeat was reported in 1996 to associate with novelty seeking ^[19], and it is one of the most cited findings in behaviour genetics. A later meta-analysis with a fresh replication sample found the picture messier: the repeat polymorphism itself did not hold up, a different D4 variant did, there was clear evidence of publication bias, and the effect that survived would account for at most about 3% of variance in the trait ^[20]. The receptor biology is real; the personality story built on it is much weaker than its citation count suggests.

Dopamine receptors are not fixed hardware. Their number, their affinity state and their coupling all shift in response to how much dopamine has been arriving, and the direction of the shift is generally opposite to the drug. That is the whole of tolerance in one sentence, but the detail is where the clinically important surprises live. See tolerance and dependence for the general framework.

Chronic blockade raises the receptor's sensitivity, sometimes enough to defeat the drug. In rats given clinically relevant doses of haloperidol or olanzapine continuously, both drugs progressively lost their behavioural effect while still occupying a high fraction of D2 receptors. The failure tracked a 20% to 40% rise in D2 receptor number and a 100% to 160% rise in the proportion of receptors sitting in the high-affinity state for dopamine, and it was at least temporarily reversible by increasing the dose ^[16]. This is the animal model of the clinical pattern where an antipsychotic that worked stops working, and it is the same biology that underlies tardive dyskinesia and the withdrawal-emergent rebound that follows stopping a D2 blocker abruptly.

Chronic flooding lowers receptor availability. Imaging in heavy stimulant users shows the opposite adjustment: methamphetamine users had about 16% lower D2 receptor availability in the caudate and 10% lower in the putamen than comparison subjects, and the size of that reduction tracked reduced glucose metabolism in the orbitofrontal cortex ^[17]. Similar reductions are reported in cocaine, alcohol and opioid dependence, which makes low D2 availability look less like the signature of one drug and more like a common consequence of sustained overstimulation. Whether it is entirely a consequence, or partly a pre-existing trait that raises risk, these cross-sectional studies cannot say.

Agonists cause a category of problem that is not tolerance at all. In 3,090 Parkinson's patients an impulse control disorder was present in 13.6% overall, and in 17.1% of those taking a dopamine agonist against 6.9% of those not; gambling, compulsive buying, compulsive sexual behaviour and binge eating all appeared, and the risk was the same for pramipexole and ropinirole ^[18]. That is a class effect of sustained D2-like and D3 agonism in limbic territory, not an idiosyncratic reaction. It usually resolves when the agonist is reduced, and a distinct withdrawal syndrome can follow that reduction.

What follows practically. Receptor number, receptor affinity state, transporter density and presynaptic synthesis all adapt on different timescales, so tolerance to one effect of a dopaminergic drug does not imply tolerance to the others, and neither does recovery. A drug that has stopped producing its wanted effect may still be producing its unwanted one at full strength.

What D4 is for. It has been known since 1991, it is the receptor clozapine prefers, it is enriched in prefrontal cortex, and no selective D4 drug has ever shown clinical benefit. Selective D4 antagonists were tested as antipsychotics and failed. The receptor is clearly doing something; nobody has demonstrated what, in a way that translates into a treatment ^[1][7].

What D5 contributes on its own. Because D5 shares nearly all of D1's pharmacology, almost no experiment separates them with drugs alone, and the field has leaned on knockout mice instead ^[8]. Its constitutive activity and high dopamine affinity suggest a distinct role in reading low ambient tone, but that is a reasonable hypothesis rather than an established function.

Whether the high-affinity state is a real target. The rise in the proportion of D2 receptors in the high-affinity state is a central plank of the supersensitivity account ^[16], and the measurement itself has been contested on methodological grounds. It appears here because the behavioural finding stands on its own; the molecular explanation for it does not carry the same weight.

Why D1 agonists have not become drugs. D1 signalling is the arm most directly tied to prefrontal working memory, and full D1 agonists have existed for decades. They have not translated, partly because of the inverted U, partly because of cardiovascular effects, partly because tolerance to D1 agonism develops fast. That failure is the clearest evidence available that raising a signal is not the same as improving a function.

Whether biased signalling delivers anything. The beta-arrestin arm is well characterised biochemically ^[12], and the argument that separating it from the G protein arm would yield cleaner antipsychotics is a decade old. No ligand has yet demonstrated that separation in a human trial. Treat it as an active hypothesis, not a property of anything currently sold.

And a framing caution. Dopamine is routinely called the reward chemical or the motivation chemical. Neither holds at the level of these five receptors, which also govern prolactin, nausea, blood pressure and sodium handling. The receptor is the unit that does the work; the transmitter is only the message. For the circuit view see the dopaminergic system, and for the same one-to-many design elsewhere see serotonin receptors.