A new era of rationally designed antipsychotics

The ideal drugs for treating schizophrenia are postulated to selectively block the D2 dopamine receptor with optimum binding kinetics. The structure of D2 bound to an antipsychotic sheds light on how to design such drugs.
David R. Sibley is in the Molecular Neuropharmacology Section, National Institute of Neurological Disorders & Stroke, National Institutes of Health, Bethesda, Maryland 20892-3723, USA.

Search for this author in:

Lei Shi is in the Computational Chemistry and Molecular Biophysics Unit, National Institute on Drug Abuse, National Institutes of Health, Baltimore, Maryland 21224, USA.

Search for this author in:

Schizophrenia is a disorder that involves hallucinations, delusions and cognitive impairment, and that affects nearly 1% of the global population1. The mainstays of therapy have been drugs that block the activity of the D2 dopamine receptor (D2R), a member of the large G-protein-coupled receptor (GPCR) superfamily of membrane proteins. Unfortunately, most of these antipsychotic drugs come with a plethora of debilitating side effects, many of which are due to off-target interactions with other GPCRs. In a paper in Nature, Wang et al.2 now report the crystal structure of D2R in complex with the antipsychotic drug risperidone. The structure reveals features that might be useful for the design or discovery of drugs that have greater selectivity for D2R than existing therapeutics, and consequently have fewer side effects.

The naturally occurring ligand for D2R is a neurotransmitter called dopamine, which mediates various physiological functions, including the control of coordinated movement, cognition and the reinforcing properties of drugs of abuse. There are five receptors for dopamine, which fall into two subgroups on the basis of their associated intracellular signalling pathways and their affinities for various drugs3: D1-like receptors (D1R and D5R) and D2-like receptors (D2R, D3R and D4R). As early as the 1970s, it was hypothesized that the therapeutic effects of antipsychotic drugs were due to them blocking D2-like, rather than D1-like, receptors4,5, but the existence of multiple D2-like receptors was not discovered until they were cloned some 15 years later6.

Although it has been proposed that antipsychotic-drug action might involve the blocking of D3R or D4R, it is now generally agreed that D2R blockade is necessary, and probably sufficient, for the amelioration of the ‘positive’ symptoms of schizophrenia, such as delusions, hallucinations and disordered thinking7. (Antipsychotics currently in use are less effective at treating the ‘negative’ symptoms of this disorder, which include social withdrawal and cognitive impairment.) Progress has been made in the development of D3R-selective8 and D4R-selective9 compounds, but there remains a paucity of drugs with high selectivity for the closely related D2R10, despite its clear therapeutic importance.

Crystal structures of D3R bound to the drug eticlopride11 and of D4R bound to the antipsychotic nemonapride9 have previously been reported. Wang and colleagues’ structure now reveals that risperidone interacts with D2R in a different way from how eticlopride and nemonapride interact with D3R and D4R (Fig. 1). One part of risperidone (known as a benzisoxazole group) extends below the orthosteric site (the site at which dopamine binds) in D2R, and penetrates deep into a hydrophobic pocket that is not formed in the D3R and D4R structures. A second, extended binding pocket above the orthosteric site in D2R encloses another part of risperidone (a tetrahydropyridopyrimidinone group). This pocket consists of amino-acid residues from extracellular loop 1 (EL1) and three transmembrane helices (TMIII, TMVI and TMVII).

Figure 1 | Binding sites within crystal structures of D2-like receptors in complex with drug molecules. Drugs that block the activity of the D2 dopamine receptor (D2R) are used to treat schizophrenia, but also block the closely related D3 and D4 receptors (D3R and D4R), and exhibit debilitating side effects due, in part, to their interactions with other receptors. a, Wang et al.2 report the crystal structure of D2R in complex with the antipsychotic drug risperidone. They observe structural features and drug–receptor binding interactions not observed in the previously reported structure of D3R with the drug eticlopride11 (b), or of D4R with nemonapride9 (c). The drug molecules are shown as coloured space-filling structures, and the regions enclosed by dots make receptor contacts that are unique to each receptor. The identification of these contacts might help receptor-specific binding pockets to be delineated, which would aid the rational design of receptor-selective drugs. Receptors are shown in grey; thick ribbons are α-helices; thin regions are unstructured. EL1 and EL2 are extracellular loops. TMV is a transmembrane-spanning segment.

Strikingly, in D2R, a residue within another extracellular loop (EL2), and which is immediately adjacent to an evolutionarily conserved cysteine residue, is buried within the protein and faces the fourth transmembrane helix (TMIV). By contrast, the equivalent residues in D3R and D4R are oriented towards water in the extracellular milieu. EL2 therefore forms a short helical segment in D2R, but is largely extended and unstructured in D3R and D4R (Fig. 1). Consequently, the structural configurations near the EL1 and EL2 interface in D3R and D4R are different from those in D2R.

Wang et al. propose that such divergence contributes to the formation of distinct, extended binding pockets in these three receptors, as has been previously suggested9,11,12. Drugs designed to selectively engage the distinctive pockets in the D2R structure might display enhanced D2R selectivity. Analogous structure-based drug-discovery efforts have already proved useful in identifying high-affinity compounds13 that block D3R14 or that activate D4R9.

Notably, the receptor segments directly above the risperidone-binding site in D2R form a hydrophobic ‘patch’ composed of the side chains of three amino-acid residues, designated Leu942.64, Trp100EL1 and Ile184EL2. This patch potentially restricts the access of molecules to the D2R binding pocket. Wang and co-workers hypothesized that this feature might regulate the dissociation of risperidone from the D2R binding site, and thus affect its residence time at the receptor.

The authors tested this hypothesis by mutating single residues in the patch and by making a mutant D2R in which both Ile184EL2 and Leu942.64 were replaced. These mutations dramatically reduced risperidone’s residence time from 233 minutes in the wild-type receptor to as little as 6 minutes in the double mutant. This effect is notable because the kinetics of antipsychotic-drug binding to D2R might correlate with a tendency to produce debilitating extrapyramidal side effects (EPS), which include rigidity, tremors and involuntary movements. Antipsychotic drugs that cause fewer EPS, such as risperidone, are said to be atypical, and it has been suggested that antipsychotics with shorter D2R residence times exhibit greater ‘atypicality’15,16. Shorter residence times at D2R might enable a minimum level of dopaminergic stimulation, which lessens EPS. The current findings illustrate how elements of the D2R structure can regulate the kinetics of drug binding, which in turn might be associated with desirable therapeutic outcomes.

The hydrophobic patch in D2R is absent in the D3R and D4R structures, presumably because of the separation between the analogous EL1 and EL2 residues in the latter two receptors. Thus, an intriguing question is whether the kinetics of drug binding to D2R are fundamentally different from those to D3R and D4R, particularly for molecules that have similar affinities for the three receptors. In other words, are the kinetics of drug binding to these receptors patch-dependent?

Of further interest is the observation17 that risperidone is not selective between D2R, D3R and D4R, thus raising the question of how this drug can bind differently to these receptors and still have identical affinities for them. Additional structures (such as D3R or D4R in complex with risperidone) will probably be needed to answer this. Nonetheless, we expect that Wang and colleagues’ D2R–risperidone structure, along with the previous D3R and D4R structures, will accelerate the design and discovery of D2R ligands that have higher selectivity than current antipsychotics, and potentially greater therapeutic impact.

Nature 555, 170-172 (2018)

doi: 10.1038/d41586-018-02328-z


  1. 1.

    Perala, J. et al. Arch. Gen. Psychiatry 64, 19−28 (2007).

  2. 2.

    Wang, S. et al. Nature 555, 269–273 (2018).

  3. 3.

    Kebabian, J. W. & Calne, D. B. Nature 277, 93–96 (1979).

  4. 4.

    Creese, I., Burt, D. R. & Snyder, S. H. Science 192, 481–483 (1976).

  5. 5.

    Seeman, P., Lee, T., Chau-Wong, M. & Wong, K. Nature 262, 717–719 (1976).

  6. 6.

    Sibley, D. R. & Monsma, F. J. Jr Trends Pharmacol. Sci. 13, 61–69 (1992).

  7. 7.

    Kapur, S. & Remington, G. Biol. Psychiatry 50, 873–883 (2001).

  8. 8.

    Keck, T. M., Burzynski, C., Shi, L. & Newman, A. H. Adv. Pharmacol. 69, 267–300 (2014).

  9. 9.

    Wang, S. et al. Science 358, 381–386 (2017).

  10. 10.

    Moritz, A. E., Free, R. B. & Sibley, D. R. Cell. Signal. 41, 75–81 (2018).

  11. 11.

    Chien, E. Y. et al. Science 330, 1091–1095 (2010).

  12. 12.

    Newman, A. H. et al. J. Med. Chem. 55, 6689–6699 (2012).

  13. 13.

    Löber, S., Hübner, H., Tschammer, N. & Gmeiner, P. Trends Pharmacol. Sci. 32, 148–157 (2011).

  14. 14.

    Carlsson, J. et al. Nature Chem. Biol. 7, 769–778 (2011).

  15. 15.

    Kapur, S. & Seeman, P. Am. J. Psychiatry 158, 360–369 (2001).

  16. 16.

    Sykes, D. A. et al. Nature Commun. 8, 763 (2017).

  17. 17.

    Seeman, P. Clin. Neurosci. Res. 1, 53–60 (2001).

Download references

Nature Briefing

An essential round-up of science news, opinion and analysis, delivered to your inbox every weekday.