Ian P. Stolerman
Section of Behavioural Pharmacology, Institute of Psychiatry P048, King's College London, London, UK
Cueing properties of drugs; Drugs as cues
Drug discrimination refers to behavioral paradigms in which the subjects learn to recognize the effects of a drug and report its presence by behavioral responses emitted to obtain reward or to avoid aversive stimuli. More formally, the presence of a drug generates an interoceptive ► discriminative stimulus (cue) signaling that an appropriate behavioral response will be reinforced. Drugs belonging to a wide range of pharmacological classes can serve as discriminative stimuli, and the subjects participating in a study may be either animal or human.
Most psychoactive drugs have effects that can be recognized and reported by people who have consumed them. Such self-reports maybe seen conceptually as a consequence of an individual's natural history of learning to describe moods during his or her lifetime. Responses acquired during drug discrimination experiments with human subjects also reflect the reporting of subjective states, but under more carefully controlled conditions and with known drugs given in defined doses (Schuster and Johanson 1988). By imposing a analogous history of conditioning upon animal subjects, it is possible to implement drug discrimination experiments that establish responses that are similarly controlled by the recognition of drug effects. The development of the field owed much to the pioneering work of Donald Overton and was initially documented in a series of books (e.g., Colpaert and Rosecrans 1978; Glennon et al. 1991; Lal 1977), after which the area has become more fully integrated in the broader field of behavioral pharmacology. The experimental paradigms that emerged assist the development of novel drugs for treating psychiatric conditions, for characterizing the receptors and neurotransmitter systems through which psychoactive drugs act, and for evaluating the ► abuse liability of substances that might produce dependence or addiction (Ator and Griffiths 2003; Solinas et al. 2006; Stolerman 1992). Drug discrimination should be considered not as one technique but as a family of closely related methods. One of its distinctive features is a unique bibliographic database (www. drugrefs.org).
Simple, drug versus vehicle discrimination studies using two-lever ► operant conditioning methodology and food reinforcers account for about three-quarters of all reports of original investigations in the field (www.drugrefs.org). In a typical animal discrimination study with exterocep-tive stimuli, the subject is placed in a test chamber and lever pressing responses result in the delivery of food. If a stimulus light is turned on only during periods when lever presses produce food pellets, then lever pressing occurs at a high rate when the light is on and at a relatively low rate when it is off. The light is thereby established as a discriminative stimulus that controls the lever-pressing behavior. In drug discrimination, the effects of a drug serve a discriminative function in relation to behavior that is similar to that of the light in the preceding example. Thus, animals are trained to press one of two levers to obtain food after receiving drug injections, and to press the other lever after control injections in which no drug is given. Intermittent schedules of reinforcement facilitate the development of optimal performance. After learning the discrimination, the animal identifies the presence or absence of the drug effect and reliably selects the appropriate lever. Discriminative stimulus effects of the drug are distinguished from those of the food by using data from test sessions where differential reinforcement of responses on the two levers is temporarily withheld. Figure 1a illustrates the acquisition of drug discrimination behavior in a typical study using animal subjects. Human studies are carried out in a functionally equivalent way but with responses appropriate for the species and with monetary reinforcers instead of food or water.
Once a subject has been trained with a given drug, further investigations are possible, where the dose of the substance is varied, or other drugs are administered. Figure 1b shows a typical dose-response curve, from which the potency of the drug can be estimated by the ► ED50 value. These dose-response curves are also psychophysical generalization gradients along the dimension of stimulus intensity, through which the subjects indicate how they perceive similarities between the test stimuli (drug doses) and those used for training. When a different drug is given during test sessions, a dose-response curve similar to that for the training drug may be obtained (► stimulus generalization); this typically occurs when the two substances belong to the same pharmacological class and therefore have similar psychoactive effects. For example, ► amphetamine is generalized with ► cocaine because both drugs act as psychomotor stimulants. In contrast, if an ► anxiolytic or sedative-hypnotic drug is administered to rats trained to discriminate cocaine, then they do not identify any psychomotor stimulant effects,
Drug Discrimination. Fig. 1. (a) Typical pattern of acquisition of simple drug discrimination by a group of eight rats. Initially responding was at the chance level of 50% regardless of drug or saline injection. Over 20 saline training days responding on the drug-appropriate, lever decreased progressively. In the 20 randomly interspersed training days with nicotine (0.4 mg/kg) responding on the drug-appropriate, lever increased progressively. Data shown as percentage of sessions in which rats initially chose to press the drug-appropriate lever averaged across successive blocks of five sessions. (b) Typical dose-response curve for nicotine obtained in a group of eight trained rats. Different doses of nicotine and saline were given in random order before test sessions that took place twice weekly, with normal training on intervening days. Results shown as percentage of responses on the drug-appropriate lever during 5-min sessions when no food was available (means ± s.e.m.).
and they press predominantly upon the nondrug lever. Such observations show that in typical two-lever drug discrimination experiments, the discrimination that is acquired is not between abnormal and normal states, but between the presence and absence of the specific pharmacological effects of the training drug. Thus, the trained rats can serve as a "litmus-paper" for a test drug that has pharmacological effects shared with those of the training drug. In such cases, the generalization gradient (dose-response curve) reflects the extent of qualitative similarity between training and test drugs as well as the intensity of the drug stimuli.
Discriminative stimuli produced by drugs with well-defined actions at known receptors are blocked by the appropriate specific antagonists (e.g., Fig. 2) and agents that influence the release, reuptake, and degradation of neurotransmitters can show orderly and comprehensible profiles of generalization according to the drug used for training. The influence of variables such as drug dose used for training, timing, and route of drug administration and ► schedule of reinforcement have been studied and are well understood. Such information has established the validity of the approach for investigating the neurophar-macological mode of action of drugs. The simple, drug versus vehicle discrimination using two-lever operant conditioning methodology and food reinforcers accounts for about three-quarters of all reports of original investigations in the field. Drug discrimination may be related to ► state-dependent learning, but some experts argue on theoretical and empirical grounds that the two phenomena are fundamentally dissimilar.
Discriminative drug effects nearly always originate within the central nervous system, as has been shown by the typical lack of generalization to agents that share the peripheral actions of psychoactive substances but do not penetrate into the central nervous system. Similarly, receptor antagonists that penetrate poorly into the brain are rarely effective. However, generalization and blockade can be seen with agonist and antagonist drugs that are poorly brain-penetrant if they are administered intracerebrally.
Drug discrimination methodology has been much used in the later stages of searches for new and improved pharmaceuticals. It may be used to identify compounds with specific actions at novel receptor sites when there is a prototypical agonist available for use as the training drug. Alternatively, putative antagonist drugs can be evaluated against a range of agonists to determine whether they produce full or partial block and whether the effect is specific for the targeted agonist. By such means it is possible to verify that substances initially identified
Drug Discrimination. Fig. 2. (a) Dihydro-b-erythroidine (DHbE) blocked the discriminative stimulus effect of a 0.1 mg/kg training dose of nicotine in a group of eight rats. The nicotine stimulus did not generalize to DHbE when the antagonist was administered alone. (b) DHbE shifted the dose-response curve for the nicotine (0.1 mg/kg) discriminative stimulus to the right in a dose-related manner (nine-fold increase in the nicotine ED50 after 1.6 mg/kg of DHbE). Horizontal dashed line represents performance after saline. All results shown as means ± s.e.m. for a group of eight rats. (Redrawn from Stolerman et al (1997) Psychopharmacology 129:390-397.)
mainly by in vitro methods or with quicker but less specific in vivo approaches have the intended effects. The 5-HT2 antagonist drugs are an example of such substances, some of which have proven valuable as ► antipsychotics. It was shown that whereas older compounds such as ► methergoline acted as ► partial agonists in relation to ► LSD, novel agents such as ► risperidone and ritanserin, acted as pure antagonists (Colpaert 2003).
Academic researchers have also made very extensive use of discrimination methodology in studies of the mode of action of an extraordinarily wide range of drugs. Prior to the development of robust ► self-administration techniques for ► nicotine in the 1980s and for ► cannabinoids from 2000 onwards, drug discrimination was the main approach for studying their dependence-related effects in animals. Thus, discrimination studies using a variety of agonist and antagonist drugs suggest that the nicotine discriminative stimulus is mediated mainly through high-affinity heteromeric nicotinic receptor subtypes and this is verified by studies with knockout mice lacking different nicotinic-receptor subunits; the findings are concordant with results of studies on nicotine self-administration. However, improvements in self-administration techniques and the increased range and sophistication of such approaches are contributing to a decline in use of discrimination methods in the current millennium.
Both American and European authorities regard drug discrimination as a key technique in the evaluation of ► abuse liability. The fact that a substance supports discrimination learning is not by itself evidence for abuse potential because some non-abused substances have that ability. The key principle underlying its use in assessing abuse liability is stimulus generalization; only if a test compound generalizes with a known abused compound is there evidence for abuse (Ator and Griffiths 2003). Tests may be made by training with the test compound itself and then testing for generalization to prototypical agents of each class of abused substance. This is the more cost-efficient approach as less work and fewer subjects are required, but it may fail if the test substance does not produce a robust discriminative stimulus. Alternatively, separate groups of subjects are trained with abused drugs from different classes and then generalization to the novel compound is examined.
Diverse models for drug discrimination broaden its applicability. Three variants are discussed here and others are known (Stolerman 1993). In drug versus drug discrimination procedures one response is associated with administration of one psychoactive drug and a second response is associated with a different substance. This procedure facilitates finer distinctions between closely related substance, such as full and partial agonists acting at the same receptor site. In practice the most useful model is a three-response variant in which the third response is associated with the undrugged state; although training takes longer, it avoids a problem with the simpler procedure in which test results for substances different from each of the training drugs are not readily interpretable.
Discrimination experiments where mixtures of dissimilar drugs were used as stimuli have shed light on how subjects process the different elements in a complex drug-induced stimulus. These studies have shown that in most cases the stimuli produced by the component drugs in binary mixtures are perceived and processed independently. Figure 3 illustrates such findings for a mixture of stimulant and depressant drugs that
Drug Discrimination. Fig. 3. Identification of the components of a drug mixture. Dose-response curves for a group of ten rats trained to discriminate a mixture of (+)-amphetamine (0.5 mg/kg) plus pentobarbitone (12 mg/kg) from saline. Each drug administered alone increased drug-appropriate responding in a dose-related manner, and at the largest doses tested their effects were not significantly below those of the drug mixture. Results shown as mean percentage of responses on the drug-appropriate lever (±s.e.m.) during 5-min sessions when no food was available. (Redrawn from Mariathasan et al (1991) Behav Pharmacol 2:405-415.)
produce unique effects in less refined procedures; the drug mixture did not produce a distinctive, novel discriminative stimulus.
Finally, novel-response procedures for drug discrimination in human subjects facilitate the interpretation of placebo-appropriate response choices (Bickel et al. 1993). In standard procedures placebo-appropriate responding indicates either the absence of any drug effect or a drug effect unlike the training drug. Instructions to subjects exposed to novel-response procedures inform them that responses on a novel-appropriate device should be made if an active test substance unlike the training drug is detected. This methodology increases the selectivity of placebo-appropriate responding and has been used in studies with psychomotor stimulants, opioids, nicotine, sedative-hypnotics, and other agents.
Advantages and Limitations of Drug Discrimination Procedures
The major advantages of discrimination procedures have resided in (1) applicability to an extremely wide range of psychoactive substances; (2) typically excellent pharmacological specificity as shown in generalization tests between agents from different classes; (3) the robust, highly reproducible nature of effects seen in many studies; (4) relative ease of collecting high-quality quantitative data. On the other hand, the time taken to train animals militates against use in primary screens or in combination with procedures that cannot be carried out repeatedly in the same subjects, such as those involving intracerebral drug administrations. Intermediate results from generalization tests can be difficult to interpret. Although a strong case has been made on both theoretical and empirical grounds that drug discrimination is functionally equivalent to the self-reporting of drug-induced subjective states, attempts to link discriminative drug effects to specific changes such as relief of anxiety, as distinct from sedation, have not been successful. Perhaps paradoxically, the greatest success with the procedures has been attained in analyses of the neuropharmacological mode of action of drugs where in many cases drug action at the molecular level has been firmly related to behavior.
► Abuse Liability
► Discriminative Stimulus
► State-Dependent Learning
Ator NA, Griffiths RR (2003) Principles of drug abuse liability assessment in laboratory animals. Drug Alcohol Depend 70(Suppl 1):S55-S72 Bickel WK, Oliveto AH, Kamien JB, Higgins ST, Hughes JR (1993) A novel-response procedure enhances the selectivity and sensitivity of a triazolam discrimination in humans. J Pharmacol Exp Ther 264:360-367
Colpaert FC (2003) Discovering risperidone: the LSD model of psycho-
pathology. Nat Rev Drug Discov 2:315-320 Colpaert FC, Rosecrans JA (1978) Stimulus properties of drugs: ten years of progress. Elsevier-North Holland, Amsterdam Glennon RA, Jarbe TUC, Frankenheim J (eds) (1991) Drug discrimination: applications to drug abuse research. NIDA Research Monograph 116. U.S. Government Printing Office, Washington Lal H (ed) (1977) Discriminative stimulus properties of drugs. Advances in behavioral biology, vol 22. Plenum, New York Schuster CR, Johanson C-E (1988) Relationship between the discriminative stimulus properties and subjective effects of drugs. In: Colpaert FC, Balster RL (eds) Transduction mechanisms of drug stimuli. Springer-Verlag, Berlin, pp 161-175 Solinas M, Panlilio LV, Justinova Z, Yasar S, Goldberg SR (2006) Using drug-discrimination techniques to study the abuse-related effects of psychoactive drugs in rats. Nat Protoc 1:1194-1206 Stolerman IP (1992) Drugs of abuse: behavioural principles, methods and terms. Trends Pharmacol Sci 13(170):176 Stolerman IP (1993) Drug discrimination. In: Van Haaren F (ed) Methods in behavioral pharmacology. Elsevier, Amsterdam, pp 217-243
Was this article helpful?