Genetic Strategies in Preclinical Substance Abuse Research

Kaitlin E. Browman, John Crabbe and Ting-Kai Li

   Genetic factors influence the predisposition of an individual to abuse drugs.  This is most clear for alcoholism, where studies suggest that different patterns of inheritance influence two (or more) diagnostically distinct forms of alcoholism (18).  As this review indicates, genetic control probably extends to other drugs of abuse as well.  However, one does not inherit predispositions; rather, one inherits genes.  The number and identity of genes whose gene products increase susceptibility to drugs of abuse, however, remain unknown.

   The expense and limitations inherent in human genetic research have led to a great increase in the use of genetic animal models to elucidate the pathways from gene to behavior.  This chapter deals with preclinical research, and almost entirely with genetic animal models.  The intent is not to review existing data exhaustively, as this has been the substance of a collection of reviews (25).  Rather, we wish to extend a previous version of this chapter (26) by indicating strategies fruitfully employed in recent years, and highlighting future applications in this area.  In roughly declining order regarding the amount and quality of interpretable genetic results, research has indicated that responses to drugs of abuse from these classes have substantial genotypic determinants: ethanol, opiates, and nicotine; psychostimulants and depressants; benzodiazepines; caffeine, hallucinogens, and organic solvents.  This review will primarily cite informative examples from studies with ethanol [see (59) for a review of nicotine, cocaine, and opiate studies and (84) for other studies].

   Pharmacogenetic studies are generally trying to address the question: are there heritable predispositions to abuse drugs?  Relevant responses to consider in analyzing genetic determinants of drug sensitivity include at least: innate (first-dose) sensitivity, which operationally often includes the development of within-session tolerance; chronic (2 doses or more) sensitivity; dependence (i.e., withdrawal severity); reinforcing efficacy; and metabolism.  For any given component response, there are several aspects of genetic influence to distinguish.  First, it is important to ascertain whether a single or multiple genes are involved.  Although biomedical disorders exist where a single gene is sufficient to induce disease, the apparent situation in psychopharmacology is considerably more complex, and multiple genes are likely operative.  Next, we need to ascertain the function of the genes' protein products.  It is also desirable to isolate their genomic location, which may allow for the identification of markers of drug susceptibility, even if these markers are not functionally important for drug abuse outcome.  Mapping drug susceptibility genes may lead one to the location of an apparent "candidate gene" in a chromosomal region, which may then lead to an understanding of function.  Knowledge of how these susceptibility genes are regulated, and how they are modulated by environmental conditions is also desirable.  If strictly genetic therapies are unlikely, such knowledge may allow therapeutic interventions at levels higher than the genome.

   The field of psychopharmacogenetics provides no perfect example beginning with a single gene and ending with a drug-related behavioral endpoint, with all the questions posed above answered.  Nonetheless, significant advances have been made, and new approaches offer great promise for reaching this goal.

TWO STANDARD METHODS IN GENETIC ANIMAL MODEL RESEARCH

   Artificial selection for traits relevant to understanding drug abuse represents a major historical contribution of genetic animal model research.  The application of techniques employed in the breeding of animals for desired agricultural or esthetic characteristics results in systematic mating of lines of mice and rats to respond characteristically to drugs of abuse.  The genetic result of this procedure is to increase the frequencies of most alleles affecting drug response positively in a sensitive line, while those alleles leading to low responsiveness become more frequent in a line bred for low response.  In contrast, the other standard method is to examine multiple existing inbred strains of animals.  An inbred strain is created after more than 20 generations of brother-sister matings.  After inbreeding, all same-sex members of an inbred strain are genetically identical, such that for each gene a single allele is fixed homozygous.  The particular alleles fixed in a given inbred strain are the result of chance, in contrast to selected lines where allele frequencies reflect the influence of systematic mating.  More than one hundred inbred strains of mice and rats are commercially available.  Uses of these techniques are complementary, and will be discussed in turn.

Selective Breeding

   The successful development of selected lines provides a population in which to test for genetically correlated traits.  By applying stringent criteria, any other difference between a pair of selected lines is presumed to be due to the common influence, or pleiotropism, of the genes determining the selected response. Furthermore, this technique enables one to capture genes with very small effects on the selected drug response.  Although most widely applied to select for ethanol-related traits, selective breeding has also been employed for opioids, benzodiazepines, and other psychoactive drugs.  Table 1 summarizes some of the currently available lines selectively bred for traits related to drug abuse.

   Although the technique of selective breeding is extremely powerful for pharmacological analyses, it also has limitations.  For example, it is not known which specific genes are mediating the selected behavior.  Furthermore, for complex traits such as those characterizing drug sensitivity, each selected line can model only a part of the whole trait domain.  A previous version of this chapter (26) discusses some of these limitations, so they will not be reviewed here.  The following insights gained from research with selected lines expands upon the previously published chapter.

WSP/WSR Mouse Lines

   One animal model developed for differential genetic susceptibility to ethanol withdrawal is the Withdrawal Seizure-Prone (WSP) and Withdrawal Seizure-Resistant (WSR) mouse lines (26).  Chronic exposure to ethanol vapor for three days was followed by assessment of withdrawal for 24 hours, using the severity of a handling-induced convulsion. Each generation, WSP mice with the most severe withdrawal scores were mated to form the next generation, while WSR mice with the lowest scores were similarly mated inter se to produce the WSR line.  There are two WSP and two WSR lines, and a direct comparison from the twenty-fifth selected generation demonstrated no overlap between the two pairs of selected lines in the severity of ethanol withdrawal.

   Although WSP and WSR mice were selected strictly for differences in alcohol withdrawal severity, WSP mice have more severe acute and chronic withdrawal than WSR mice to diazepam, pentobarbital, acetaldehyde and tertiary butanol (27).  These results strongly suggest that some genes act to influence drug withdrawal severity not only to ethanol, but also to a number of other depressants.  Corroborative experiments are discussed in a later section.

   WSP and WSR mice do not differ in sensitivity to ethanol's locomotor stimulant, hypnotic or hypothermic effects.  More surprisingly, after several doses of ethanol, WSP and WSR mice developed tolerance to the same degree.  WSP (but not WSR) mice develop a conditioned preference for a location paired with ethanol injections; thus two-bottle choice preference drinking and conditioned place preferences may not tap the same underlying substrate (presumably, reward) in mice (30).  These patterns of results in WSP and WSR mice has made it clear that the genetic factors controlling ethanol sensitivity, tolerance, dependence and drinking are independent.  In turn, this strongly suggests that they are maintained to a significant degree by nonoverlapping neurobiological mechanisms.  Deitrich and Spuhler (33) have discussed the strength of selected line models for excluding common mechanisms in cases where there is no discernible correlated response.  Selection has been replicated by another group, and results from High (HW) and Low Withdrawal (LW) lines confirm some of the data obtained with WSP and WSR mice (V.G. Erwin, personal communication).

P/NP Rats, HAD/LAD and AA/ANA Rats

   Rats developed for Preferring (P) and Non-Preferring (NP) 10% ethanol vs. water are one of the older selection experiments in psychopharmacology.  The P and NP lines have been replicated through the development of two High Alcohol Drinking (HAD) and two Low Alcohol Drinking (LAD) lines (56).  Several other sets of rodent lines have been developed using similar paradigms, including the mouse lines HAP and LAP (see Table 1).  Thus, there is the opportunity to achieve convergent validity in the search for genetic correlates of ethanol drinking. We will largely confine our discussion to major findings in the P/NP, HAD/LAD and AA/ANA rat lines. Studies with Sardinian Preferring and Non-Preferring (sP/sNP), and UChA/UChB rat lines are reviewed elsewhere (25) and will not be covered here.

   Several differences between the P and NP rats have been reviewed [(26); see also (68) for comparisons].  For example, P rats will voluntarily drink 10-30% ethanol solutions in quantities that produce pharmacologically meaningful blood alcohol concentrations.  With chronic drinking, both metabolic and neuronal tolerance, and physical withdrawal signs develop in the P rats.  The P rats will perform an operant response (bar-pressing) to obtain ethanol in concentrations as high as 30% for reasons other than the caloric value, taste or smell of the ethanol solutions, and will self-administer substantial doses of ethanol (26).  Experiments using the conditioned place preference paradigm suggest that ethanol is less aversive to P than to NP rats (80).

   Although the most obvious neuroanatomic/neurochemical difference between the P and NP lines appears to involve serotonin (5-HT) neurons, other neurotransmitter systems are also important in regulating alcohol self-administration.  Ethanol drinking in the P rats is suppressed by 5-HT uptake inhibitors, 5-HT_1A antagonists, dopamine (DA) agonists, DA uptake inhibitors, the GABA inverse agonist Ro 15-4513, and opioid receptor antagonists(56, 62, 89).  It is important to note that no differences in densities of 5-HT_1A, 5-HT_1B, D1, D2 or D3 receptors within the mesolimbic dopamine system are found between the HAD and LAD rats (60), suggesting that these receptors are not important in mediating the differences in drinking.  A recent review on the neurobiology of high alcohol-seeking behavior in rodents has appeared (61).

   Initial sensitivity to the locomotor stimulant effects of ethanol, and within-session or acute tolerance to ethanol ataxia, are the most generalizable and robust responses to ethanol associated with ethanol preference.  P, HAD and AA rats exhibit increased spontaneous locomotor activity with low dose ethanol, and are more able to develop tolerance with exposure to a single sedative/hypnotic dose of ethanol, in contrast to NP, LAD and ANA rats (55, 56).  Tolerance and preference were also correlated in 8 inbred rat strains (79), and in individual HS/Ibg heterogeneous stock mice with high and low ethanol preference (37).  However, alcohol-preferring C57BL mice are insensitive to ethanol stimulation, and alcohol-nonpreferring DBA mice are very sensitive to this response, which is inconsistent with the rat studies, and studies of their relative propensities to develop tolerance to ethanol's ataxic effects have been variable (25).  Both P and AA rats also exhibit a greater preference for oral consumption of sweet solutions than the NP and ANA rats.  Clearly, more systematic study of the selected lines (and the HAP and LAP mouse lines) for associated behavioral and neurobiological traits can increase our understanding of mechanisms underlying ethanol reinforcement. 

LS/SS Mice and HAS/LAS Rats

   The LS and SS mice were among the first selected lines in pharmacogenetics.  LS and SS mice differ markedly in response to the hypnotic effects of ethanol (63), and numerous papers now report differences between these lines in behavioral, physiological, pharmacological and biochemical traits.  In the same way that the success of the P/NP selection led to its replication with HAD/LAD rats and HAP and LAP mice, interest in the LS/SS mice has been followed by the production of replicated pairs of HAS and LAS rats differing in an equivalent response.  The LS/SS mice have also been frequently studied for their nicotine responsiveness (19).  Studies with LS and SS mice have recently been comprehensively reviewed (32), and studies relating GABA function to ethanol's sedative effects are discussed in a later section.

Other Selected Lines

   Reasonably comprehensive reviews of other selected lines and their uses have appeared (25, 32). High (HAR) and Low (LAR) Analgesic Response lines have been bred for high and low sensitivity to levorphanol using the hot plate assay of analgesia (5).  After more than 15 generations of selective breeding, the two lines differed by about 7-fold in their analgesic sensitivity to i.p. levorphanol or morphine, and about 67-fold to i.c.v. [D-Ala²-,NMePhe⁴,Gly-ol⁵]-enkephalin (DAMGO), a specific m-opioid receptor agonist.  These lines differ predominantly for m receptor-mediated responses, but differ relatively little in response to k or d agonists on the hot plate assay (5).  Receptor autoradiographic studies reveal a 150 to 200% greater [³H]DAMGO binding density in HAR mice in the dorsal raphé nucleus, but only small (16%) differences were seen in the periaqueductal gray, an area more traditionally associated with pain sensitivity (4).

   Seven generations of selective breeding have recently been completed for high acute functional tolerance (HAFT) or low (LAFT).  These lines do not differ in initial sensitivity to ethanol (35), but show large differences in ethanol tolerance.  Four generations have been completed for replicate lines.  These lines should be useful for identifying underlying mechanisms mediating acute functional tolerance.

Summary

   A number of selective breeding studies have revealed both the power of the approach to address psychopharmacological problems, and a number of pitfalls to be avoided in future research.  Human susceptibility to drug abuse has variously been described as failure to avoid, or uncontrollable tendency to approach the drug.  Animal models for reinforcing properties are notoriously difficult to work with, but the success of studies with oral self-administration suggests that it might be timely to undertake a selection project to develop rodent lines differing in sensitivity to the rewarding properties of drugs.  A number of choices exist (e.g., conditioned place preference and aversion; effects of drug on threshold for intracranial self-stimulation; and conditioned taste aversion).  Since most data support a prominent role for dopamine systems in reinforcement mechanisms, a set of mouse lines selected for their sensitivity to haloperidol catalepsy may prove to be useful (47).  A number of differences have been documented between Neuroleptic Responder and Nonresponder (NR/NNR) mice in dopaminergic and cholinergic systems of the basal ganglia (48).

   No models currently exist for a number of other interesting behavioral traits.  For example, it might be of interest to develop genetic animal models that recapitulate particular factors thought to enhance risk to drug abuse, such as impulsivity.  In this regard, the finding of Schuckit and colleagues (77) that sensitivity to an alcohol challenge in humans can predict future alcohol abuse/dependence suggests that this direction of research will be fruitful.

Inbred Strains

   Because each member of an inbred strain is genetically identical to all other members of that strain, if variation among inbred strain means exceeds within strain (i.e. non-genetic) variation, this demonstrates the presence of significant genetic control of the trait.  Genetic influence has been demonstrated for all drug responses studied [for reviews, see (25, 59, 78)].  Such findings imply that other genetic methods, such as selective breeding, may be fruitfully applied to study drug-related traits.

   A second use of inbred strains is to estimate the strength of genetic correlation.  For example, strains showing a high degree of initial sensitivity to ethanol-induced hypothermia also tend to develop tolerance to this effect when ethanol is chronically administered.  Patterns of correlation among inbred strains for a number of alcohol-related traits has revealed a pattern of genetic codetermination of sensitivity to some traits (e.g., hypnosis, depression of locomotor activity), and a fair degree of genetic independence of different groups of responses (22, 24).  The genetic stability of inbred strains (across laboratories, across time) offers an enormous advantage to the experimenter.  Data sets on a common battery of standard inbred strains are cumulative, and allow the pooled expertise and resources of multiple laboratories to accrue a common data base.  For three inbred strains of mice (C57BL/6, BALB/c, and DBA/2), a rather large data base has been accumulated for many responses to drugs of abuse (25).  A recent collaborative effort has undertaken to characterize several responses to ethanol, diazepam, pentobarbital, phenobarbital, and morphine in a panel of 15 mouse strains (22).  Sensitivity to morphine hypothermia and morphine-induced depression of activity were correlated (6).  In contrast, several responses to diazepam were genetically uncorrelated (23).  Preference drinking of ethanol and morphine were genetically correlated across strains, but preference for diazepam solutions was unrelated to either (2, 3).  Acute withdrawal severity from ethanol, pentobarbital and diazepam were correlated (64).  The main disadvantage of inbred strains is that there are no heterozygotes, so inbred strains are not very reflective of highly outbred individuals (such as individual humans).

   In general, such studies offer evidence for common genetic determination of specific clusters of drug response variables and provide insight into neurobiological mechanisms underlying mechanisms of action.  Inbred strain studies and studies with selected lines offer two independent methods for assessing genetic correlations, and the use of both methods is increasing.

NEW METHODS IN GENETIC ANIMAL RESEARCH

Candidate Gene Approaches

   Recent advances in molecular biology are of great potential importance for exploring existing genetic animal models and creating new ones.  By studying the animal models discussed above, responses to drugs can be assessed on a molecular, cellular and behavioral level.  These studies, combined with data from humans, have resulted in the identification of various genes (termed candidate genes) believed to play a role in mediating drug responses.  Modern molecular techniques have also nominated candidate genes.  Genes considered good candidates typically direct synthesis of an important component of the nervous system, such as a neurotransmitter, neurotransmitter receptor or a protein involved in cellular transmission of information.  Following the identification of a candidate gene it is possible to evaluate the contribution of that gene in modifying the response to drugs of abuse.  Current methods include use of transgenic/knockout mice (in which the candidate gene is deactivated or overexpressed) [see (49) for a review of these techniques].  In the following section selected candidate genes will be discussed.

The GABA_A-Benzodiazepine Receptor-Coupled Chloride Channel.

   The wealth of pharmacological and behavioral data available on some model systems (e.g., the inhibitory neurotransmitter gamma-aminobutyric acid, GABA) has suggested areas where particular genes appear to be good candidates for further study.  Recent reviews summarize a great deal of work implicating the GABA-benzodiazepine receptor complex, and in particular type A receptors, as important for mediating several effects of ethanol (45).  The GABA_A receptor complex comprises different subunits grouped into families (a, b, g, d, e, r) based on homology.  Some of the research implicating GABA in mediating the response to drugs of abuse has relied on pharmacological manipulations, while more recent studies have devoted attention to investigating different subunits of the GABA_A receptor to investigate the effects on the response to ethanol.  This section will briefly cover both types of studies, but see (45) and (50) for more thorough reviews. 

   Pharmacological manipulations with GABA agonists and antagonists affect ethanol sensitivity, and LS mice are more sensitive to such manipulations than SS mice.  This suggests that ethanol might produce depression in part by augmenting GABA's stimulant effects on chloride flux.  LS (and HAS) brain membrane preparations were more sensitive than those from SS (and LAS) to ethanol-potentiated muscimol-stimulated chloride flux (10).  Furthermore, the insensitivity to ethanol of GABA-activated chloride channels from SS mice can be seen in Xenopus oocytes in which mRNA is expressed; this finding is also apparent in HAS and LAS.  With chronic ethanol, alterations in levels of mRNA for specific GABA_A subunits are seen (88), and ethanol administered to oocytes chronically alters receptor function (9).  In other studies, naive WSR mice were found to have higher levels of mRNAs for a1, a6 and b2 subunits in cerebellum than WSP mice (53), and chronic feeding of ethanol to WSP and WSR mice also led to changes in specific GABA_A receptor subunit mRNA expression, which may be important to the development of withdrawal hyperexcitability (8).

   This chloride ion channel is also influenced by binding sites recognizing barbiturates and benzodiazepines.  Inactivation of the a6 gene abolished d subunit expression, suggesting that manipulating one subunit can dramatically affect the function of the entire receptor complex.  With regard to behavior, however, these mutant mice did not differ from wild-type in sensitivity to loss of righting reflex after ethanol administration (52), nor did they differ from wild-type mice in the hypnotic response to ethanol, midazolam, or pentobarbital (50).

   Chimeragenesis and mutagenesis approaches provide support for the notion that ethanol exerts a specific effect on proteins coding for subunits of the chloride ion channel.  A region of 45 amino-acid residues including two specific amino-acid residues in transmembrane domains 2 and 3 is critical for allosteric modulation of GABA and glycine receptors by ethanol (65).  By mutating Ser-267 located in transmembrane 2 of the Gly-R a1 subunit, a receptor was produced that was insensitive to the effects of ethanol.  In addition, mutating Ala-291 of the a1 but not b1 subunit of the GABA receptor also abolished ethanol sensitivity (65), demonstrating further how manipulation of one subunit can alter the function of the entire receptor complex. 

   The g subunit is necessary for sensitivity to benzodiazepines, and the g2 subunit is the major subunit in the majority of GABA_A receptors (50).  Mice lacking the g2 subunit do not show the benzodiazepine induced sedation and loss of righting reflex that wild type mice exhibit.  The g2 subunit exists in two splice-variant forms, a long (g_2L) and a short(g_2S), and studies with oocytes expressing different combinations of cloned GABA_A receptor subunits showed that ethanol sensitivity in this preparation specifically requires the presence of the g_2L subunit, which differs from the alternative g_2S subunit by only 8 amino acids.  This region contains a consensus phosphorylation site for protein kinase C.  Ethanol potentiation of GABAergic activity was prevented only if a g2L subunit sequence was administered.  Site directed mutagenesis to the phosphorylation site on this subunit prevents ethanol modulation of receptors containing the mutant g2L subunit (85).  In a complementary set of studies, null mutant mice lacking the g isoform of protein kinase C showed decreased sensitivity to ethanol on test of loss of righting reflex and hypothermia measures of ethanol sensitivity (44).  Electrophysiological studies demonstrate that ethanol potentiates GABA in both wild type and g2L knockout mice, suggesting that at least in a neuronal context native GABA_A receptors do not require the long variant to show an ethanol-potentiated response (50).

   Together, these studies suggest that GABA plays an important role in modulating some effects of ethanol and benzodiazepines.  Although the individual contribution of each subunit is not clear, genetic analysis with selectively bred lines and inbred strains have made a major contribution to clarifying the relationship of GABA and the response to ethanol.  Nonetheless, numerous drugs interact with the GABAergic neurotransmitter system.

Fyn Kinase and Glutamate

   Mice lacking the fyn gene, which codes for a non-receptor type tyrosine kinase, are hypersensitive to ethanol-induced loss of righting reflex (45).  Following the administration of ethanol, tyrosine phosphorylation of the N-methyl-D-aspartate (NMDA) receptor was enhanced in wild-type mice compared to mice lacking the fyn gene.  Acute tolerance to the inhibition of NMDA receptor-mediated excitatory postsynapic potentials developed in the hippocampus of wild-type mice, but not in mice lacking the fyn gene (67).  Although the exact role of fyn in the response to ethanol is unclear, this is a plausible candidate for future research.  Other studies have shown that during withdrawal from chronic alcohol inhalation, WSP mice become more sensitive to convulsions elicited by NMDA and kainic acid than WSR mice.  Both selected lines show a non-specific increase in sensitivity to convulsions elicited by GABA antagonists, suggesting a role for glutamatergic receptors in alcohol withdrawal (38).

The 5-HT_1B Receptor

   With the knowledge that serotonergic (5-HT) systems influence drug and ethanol responses, genes mediating 5-HT functioning become obvious candidates for molecular biological study.  Of the numerous 5-HT receptor subtypes, the 5-HT_1B may be mediating a number of drug related behaviors.  Mice lacking the 5-HT_1B receptor display enhanced aggression and altered 5-HT release from some brain areas (71, 75, 83).  Null mutants also drink twice as much ethanol as their wild-type counter parts, and voluntarily ingested solutions containing up to 20% ethanol in water (31).  The intake of food, water, sucrose, saccharin and quinine solutions was normal in the null mutant mice, suggesting that differences due to taste or caloric properties were not responsible for the differences in ethanol intake. Mice lacking the 5-HT_1B receptor were less sensitive than wild-type mice in tests of ethanol-induced ataxia. Tests of ethanol withdrawal and metabolism resulted in equivalent responses between the null mutants and wild-types, however.  These combined results suggest that the 5-HT_1B receptor may modulate ethanol drinking, and that while serotonergic manipulations lead to reduced responsiveness to some ataxic effects of ethanol, these do not appear to affect physical dependence (31).

   The 5-HT_1B knockouts have also been tested for the acquisition of ethanol-induced conditioned taste aversion and ethanol-induced conditioned place preference.  In the place conditioning experiment, ethanol injections were paired with one environment and saline injections paired with another unique environment (71).  After repeated pairings of the substance with the given environment the mice were given the choice between the ethanol- and saline-paired environments.  The knockout mice showed no preference for the ethanol-paired environment, while wild-type animals showed ethanol conditioned place preference.  These results suggest that null mutants are less sensitive to the reinforcing effects of ethanol (72).  Taste conditioning experiments with these animals showed that both knockout and wild-type mice developed equivalent taste aversion at the same rate, suggesting that ethanol was an equivalent reinforcer.  Thus, three models of ethanol's reinforcing efficacy yielded three different answers with this animal model, demonstrating that additional studies will be required to establish the effect of the 5-HT_1B knockout.  Interestingly, the 5-HT_1B knockout mice acquire intravenous cocaine self-administration more rapidly than wild-type mice, and are more motivated to self-administer cocaine (73).

Dopaminergic system

   The mesolimbic dopamine system is important for mediating the effects of addictive drugs.  Disruption of the mouse dopamine transporter gene, of genes coding for dopamine D1, D2, D3, and D4 receptor subtypes, and inactivation of the tyrosine hydroxylase gene have all altered sensitivity to ethanol, the psychomotor stimulants or other addictive drugs [see (7) for a review of these studies].  For example, mice with the D2 dopamine receptor deleted consumed less ethanol, were insensitive to ethanol's locomotor depressant effects and exhibited reduced sensitivity to the ataxic effects of ethanol relative to their wild-type controls (70).  Null mutant mice did not exhibit differences in the locomotor stimulant response to morphine, but did show deficits in the development of conditioned place preference to morphine (57).  Combined, these studies suggest a potential role for the dopamine D2 receptor in sensitivity to drug reward.  It is important to keep in mind that the D2 knockouts in these two experiments were independently produced: thus, a comparison of the results over time may help to clarify the influence of genetic background on the sensitivity of knockouts to drugs of abuse.

   Other dopamine receptors may modulate an individual's sensitivity to drugs of abuse [see (28) for a recent review].  For example, dopamine D3 receptor null mutant mice exhibited enhanced locomotor stimulation to cocaine, and greater amphetamine-induced conditioned place preference compared to wild-type mice (87).  The D3 dopamine receptor is expressed in brain regions believed to influence motivation and motor functions, and studies with agonists selective for the D3 receptor decrease cocaine self-administration, supporting results from knockout studies (13).  Dopamine D4 null mutant mice also display enhanced sensitivity to the locomotor stimulant effects of ethanol and psychomotor stimulants (74).

Enzymes of Alcohol Metabolism

   In humans, genetic variants of the principal enzymes of alcohol metabolism, alcohol dehydrogenase (ADH) and mitochondrial aldehyde dehydrogenase (ALDH2) influence alcohol drinking behavior and rates of alcoholism, and evidence suggests that possession of variants of the ADH and ALDH2 gene tends to protect against the development of alcoholism in Asian populations (41).  The specific alleles that confer this "protective" effect on abusive drinking are ADH2*2 (produces a high activity b₂-ADH form) and ALDH2*2 (produces a low activity ALDH2 enzyme).  The net result is that acetaldehyde, a highly toxic substance, is produced more rapidly by the b₂-ADH variant and is metabolized less rapidly by the low-activity ALDH2 variant, producing the alcohol-flush syndrome.  Thus far, however, studies of ADH and ALDH2 as candidate genes in the rat lines selected for alcohol preference, (P vs. NP and AA vs. ANA) do not explain the difference in acetaldehyde metabolism and the difference in alcohol drinking behavior of these lines (15).

Summary

   Although the candidate gene approach is being actively pursued, these studies cannot conclusively demonstrate that a gene is mediating a given behavior.  One limitation of gene targeting technology is that any response to ethanol and drugs of abuse is determined by several genes and their interactions.  The effect of disrupting one gene may not be entirely detectable, or may alter only one aspect of a given phenotype.  It is easy to imagine that if multiple genes are affecting a given behavior, the elimination of one gene might result in developmental compensation by the other genes involved.  This might result in a negative finding in the knockout mouse when the gene would be influencing the behavior in an intact animal.  A second complication with knockout technology is that many genes of interest are present in multiple cells in the body (i.e., not just in the brain region of interest).  The genes in knockout mice are disrupted in all cells, and thus it is impossible to determine what the location of the cells involved in a given phenotype are.  Another issue of concern is genetic background.  Transgenes may not exert similar effects when expressed on different genetic backgrounds (39).  The complication of developmental compensations may be addressed by the use of conditionally-regulated transgenics. These transgenics have a sequence inserted which allows the experimenter to alter gene expression, commonly by using an antibiotic treatment (16).  Finally, targeted gene expression may allow for the manipulation of gene expression in a specific brain region, which could increase our knowledge not only of the specific neurotransmitter system or receptor subtype involved in a response to a drug of abuse, but the identification of a brain region mediating the phenotype of interest. 

   Another technique potentially useful in elucidating the gene(s) mediating the response to drugs of abuse is the use of antisense oligodeoxynucleotides (ODNs).  Antisense ODN strategies have traditionally been used to produce more discrete deletions of gene function, such as at a particular phase of development or even in a specific region of the brain.  Although ODN strategies are useful for blocking gene expression after development and enable brain regional specificity, the timing of protein suppression compared to their effect on drug response is uncertain.  Also, many antisense ODNs have proven to be toxic, limiting their application.  A discussion of strengths and weaknesses of ODN studies has appeared (86), as has a review of some studies using these techniques (28).

Gene Identification Approaches

   Until recently there was no way of conclusively identifying genes that might be playing a role in an individual’s predisposition to addiction.  Genome mapping and DNA sequencing in the mouse has led to the development of a mouse genetic map that can be related to the human genetic map.  Specific regions of the mouse chromosome have been identified which are believed to contain specific genes mediating sensitivity to effects of alcohol and other drugs of abuse.  A variety of modern molecular biological techniques have allowed for the identification of such genes.  These include quantitative trait loci (QTL) analysis (and congenic verification), immediate-early gene expression, subtractive hybridization and differential display.  These will be discussed in turn below.

Quantitative Trait Loci Gene Mapping.

   Use of this technique in psychopharmacological studies has often depended upon the analysis of drug responses in recombinant inbred (RI) strains. RI strains are derived from two inbred strains by inbreeding from their F₂ (genetically heterogeneous) cross. They are called RI strains because the progenitor chromosomes are recombined several times per chromosome during their development, resulting in a unique pattern of recombinations of the progenitor chromosomes in each RI strain.  Each RI strain thus represents a somewhat random sample of the genetic variability available in the two parent strains.  Like any other inbred strain, all members of each RI strain are genetically identical, and each RI must have inherited any given allele from one of the two parents.  The power of RI analysis lies in the existence of many Strain Distribution Patterns identifying the genetic map location of previously typed genes and markers.

   When both progenitor strains and a battery of their RIs are tested for a drug response trait under controlled environmental conditions, strains will differ due to their unique genotypes. The strain means can then be referred to the pattern of previously mapped marker loci (1, 42).  Methods have been developed to detect the influence of genes with relatively small influences on a response, and then assign a tentative map location (54).  These genes are known as polygenes, and their locations are called quantitative trait loci (QTLs). When a significant association is found between RI strain mean patterns and a pattern for a previously mapped marker, this suggests linkage between a gene affecting the trait of interest and the marker, and a QTL is mapped.

   Of the several sets of RI strains, the best studied in psychopharmacology is the BXD RI series, derived from the C57BL/6 and DBA/2 inbred strains.  The BXD RI strains have a relatively large number of markers (>1500) that have been previously mapped.  It is important to keep in mind that most drug responses appear to be determined by several QTLs, as several recent studies have shown for morphine, ethanol, amphetamine and methamphetamine sensitivity (1, 42, 43).  The majority of the QTL mapping work, however, has been conducted on the response to ethanol, reflecting the pharmacogenetic literature [see (7, 29) for a review of studies with other drugs of abuse].

   One area of interest has been the possibility that susceptibility to withdrawal from different drugs of abuse may have common genetic determinants.  Results discussed earlier with WSP and WSR mice and inbred strains indicate substantial genetic communality of influence on withdrawal from multiple drugs of abuse.  QTLs for acute alcohol withdrawal were found on chromosomes 1, 4 and 11 accounting for more than 68% of the genetic variance in this response (11).  The QTL on chromosome 11 mapped near genes coding for the a1, a6 and g2 subunits of the GABA_A receptor.  These loci were confirmed using F₂ mice derived from C57 and DBA parental crosses, and in lines selectively bred for acute withdrawal.  This confirmation of the BXD findings in independent, genetically-segregating populations is important to rule out chance associations. 

   In similar studies, a QTL for acute pentobarbital withdrawal convulsions was also identified in the same region of chromosome 1 as that for acute ethanol withdrawal, suggesting that some specific genes may be found which confer susceptibility to withdrawal from multiple drugs of abuse (12).  Studies using alcohol to condition a taste aversion mapped a QTL to this same region of chromosome 1 [see (29) for a review of these studies].  Chronic alcohol withdrawal QTLs have also been provisionally mapped (21), and a QTL on chromosome 1 was in the same region as definitively mapped for acute alcohol withdrawal.  When QTLs for different behaviors are mapped to the same chromosomal region, it indicates the potential for pleiotropism (i.e., that one gene mediates more than one trait).  Determining the specific gene responsible for each response, however, requires much additional research.

   The principles of RI analysis can also be used in RIs developed from selectively-bred lines. This application of RIs to detect single genes is strengthened by the fact that the progenitor stocks differ markedly on the traits analyzed because they were specially bred to have those differences. On the other hand, such RIs have the limitation that they are almost completely uncharacterized with regard to marker loci.  For one system, a sufficiently dense map has been produced to allow detection of significant QTLs affecting alcohol-induced loss of righting reflex in Long-Sleep (LS) and Short-Sleep (SS) mice (51).  Five significant QTLs affecting the loss of righting reflex response were located in RIs and then verified in an F2 cross from the LS x SS mice.  These QTLs were located on chromosomes 1, 2, 8, 11 and 15, and account for more than 50% of the genetic variability in this trait (57A).  Potential candidate genes include the gene Ntsr (coding for the high-affinity neurotensin receptor; chromosome 2), and Acrg and Acrd (acetylcholine subunit genes coding for g and d subunits on chromosome 1).  A QTL for neurotensin receptor density was also provisionally mapped in LS x SS RI strains (36).

   Traits thought to reflect the reinforcing effects of ethanol have been investigated using ethanol preference drinking paradigms, conditioned place preference, conditioned taste aversion, locomotor stimulation and sensitization to the locomotor stimulant effects (28).  Provisional QTLs for these traits have been identified, and confirmation is being sought in additional genetic populations.  For ethanol preference drinking and acceptance, QTLS were provisionally identified on chromosomes 2, 9, 11 and 15.  The regions on chromosomes 2 and 9 have been verified using an F2 cross of the BXD RI progenitor strains (69).  Of particular interest is the QTL on chromosome 9, near Drd2 (coding for the dopamine D2 receptor), because D2 knockouts were found to drink less alcohol than their wild-type counterparts (70).  These studies cannot exclude the D2 dopamine receptor as a candidate, but as noted above, cannot prove that it is the QTL.  An important caveat is that in human studies there is no consensus for a role of the DRD2 gene for susceptibility to alcoholism (40).

   A comparative linkage map of the mouse and human genomes has been published which estimates that at least 61%, and possibly as much as 80%, of the linked regions in mouse are conserved in humans (20).  The genetic map of the rat is rudimentary compared with that of the mouse.  Nonetheless, recent interest has led to the founding of a large collection of polymorphic markers (simple sequence repeats).  F2 intercross progeny from the P and NP rat lines were generated to investigate QTLs influencing alcohol consumption (14).  A QTL on chromosome 4 was identified which is near the gene for neuropeptide Y, a peptide shown to be anxiolytic and participate in control of appetitive behavior.  Interestingly, rat chromosome 4 is partly syntenic with human chromosome 7, on which a genome-wide search in humans has identified one provisional locus for alcohol dependence (40).  A recent paper has reported that deleting the gene coding for neuropeptide Y production in mice leads to an increase in ethanol consumption compared to wild-type controls whereas overexpression of the gene leads to decreased drinking (81).

   In summary, these results emphasize the utility of QTL analysis as a powerful hypothesis-generating approach to identify genes influencing drug sensitivity, both in animal models and humans. While some of the QTLs for drug responses have been verified, for most verification testing is still in progress.  In addition, verifying the influence of a QTL requires substantial time and effort, and there is always the problem of striking an appropriate balance between false-positive associations ("finding" a QTL that is not, in fact, there) and false-negative rejections (failure to find a real QTL effect).  Nonetheless, knowing the approximate chromosomal map location of a provisional QTL represents an important step towards identifying a specific gene mediating a trait.  Provisional loci can immediately suggest candidate genes that can be further studied using techniques discussed above.  However, the ultimate goal is to identify the effective gene and clone it.  One way to approach this goal is discussed in the next section.

   Congenics.  Innovation at the level of molecular biology is paralleled by the application of more finely-tuned genetic techniques at the level of behavior.  One example is the notion of creating new congenic lines based on QTLs.  Congenic mice are identical to their inbred background strain at all loci except for a small region bracketing a QTL in question.  They are created by transferring a portion of the genome (identified by a QTL)  of a donor strain to a recipient strain through a series of backcrosses.  Congenic strains can be especially useful in investigating a region of interest identified in a QTL analysis.  In congenic strains, genetic variation between two strains is restricted to a particular segment of the chromosome of interest.  If a series of partially overlapping congenic strains are created, identifying which still retain an effect on the trait of interest can reduce the size of the QTL and allow for finer mapping.  Eventually, this should facilitate positional cloning of the gene mediating the phenotype of interest.

   Recently, Markel and colleagues (58) have initiated the development of congenic strains for inbred LS and SS mice using a "speed congenic" approach, using multiple genetic markers in the selection process for each backcross generation.  These mice should be useful in elucidating genetic influences on ethanol-sensitivity induced loss of righting reflex.  Congenics are also being developed by other groups involved with QTL mapping. 

Immediate-early gene expression

   Increased expression of immediate-early genes, such as c-fos and junD, is believed to reflect neuronal activity.  A wide variety of stimuli (including pharmacological agents) can induce expression of immediate-early genes in neurons.  By quantifying c-fos and junD expression it is possible to map cell populations and brain regions that are involved in eliciting a particular drug response.  Mapping of areas of interest using gene expression is a useful technique for comparing the behavioral responses of different mouse lines in their sensitivity to ethanol. 

   Several doses of ethanol increased Fos-like immunoreactivity in selective limbic areas of the brain, including the central amygdala.  The DBA/2J strain (a strain very sensitive to many effects of ethanol) exhibited greater Fos-like immunoreactivity in the central amygdala, but not in basal ganglia regions relative to the non-responsive C57BL/6J strain (46).  When the P and AA lines were compared with the NP and ANA, several brain areas responded with increases in Fos-like immunoreactivity and increases in some brain areas were also dose-dependent (82).  Other genes are also potentially involved with the response to alcohol (45).  In the past few years immediate-early gene expression has attracted much attention with regard to drugs of abuse, and the mapping of brain areas potentially mediating the response to drugs of abuse.  These techniques are clearly useful for an initial indication of where changes might be occurring in the brain mediating at least the acute response to drugs of abuse.  Their principal difficulty is their sensitivity to non-specific aspects of experimental procedures, requiring care in interpreting results.  With future technological advances immediate early gene expression may indeed be able to help elucidate drug mechanisms.

Subtractive Techniques

   Techniques based on molecular hybridization have only recently been applied to the study of drug sensitivity responses.  These techniques are useful for identifying differentially expressed mRNA, which has been instrumental in understanding gene function and molecular mechanism(s) underlying a particular biological system.  The techniques include subtractive hybridization, subtraction (including serial analysis of gene expression, or SAGE) and differential display.  Because these techniques select anonymous mRNAs that are regulated by drug treatments, they enable the identification of genes that were previously unknown mediators of a drug response.  Two of the more widely used and best characterized are subtractive hybridization and differential display.  Subtractive hybridization combined with 2-D gel electrophoresis showed that exposure of neuroblastoma cells to ethanol induced the expression of the molecular chaperones GRP78 and GRP94 (66).  Differential display can detect both increased and decreased gene expression, and was used to identify ethanol-regulated genes in rats that had been chronically exposed to ethanol vapor (17).  This technique has also been applied to the psychomotor stimulants, and has resulted in the identification of a novel brain peptide which appears to be modulated transcriptionally by this class of drug (34).  The application of this technique to genetic animal models for specific ethanol phenotypes may become a powerful method for identifying trait-relevant changes in gene expression.  In a recent extension of these techniques, differential display was used to identify ethanol-regulated genes in the brains of the WSP and WSR selected lines (76).  The results of this study identified a gene homologous to the human neuroendocrine-specific protein as a gene having increased expression after chronic exposure in the WSP mice.

   Subtractive hybridization and differential display are still in their infancy, and direct evidence linking changes in gene expression with the response to drug are not always available.  As automated microarrays or DNA chips enter routine use, this will allow for the simultaneous determination of many mRNAs.  With chip technology, thousands of cDNAs could be used to screen mRNAs from (for example) drug-exposed and control animals concurrently.  As mentioned several times in this review, it is likely that a large number of genes influence the response to drugs of abuse, so the identification of many at once will aid in the determination of which are critical for tolerance and dependence.

CONCLUSIONS

   The studies discussed above represent only a small fraction of the available pharmacogenetic data related to abused substances. There is a considerable amount of evidence for a heritable contribution to drug sensitivity for all drugs of abuse studied thus far. All potentially important pharmacological features of drug action appear to be influenced by genes, including acute sensitivity, tolerance development, dependence liability, and reinforcing properties.  Evidence suggests that there are commonalties of genetic influence among drugs of abuse, and some genotypes appear to be susceptible to dependence on multiple drugs of abuse. There are also individual genes which appear to predispose to particular drug responses, or to multiple drug responses.  Given the similarity between mouse and human genome, genes that have been mapped to specific locations on the mouse genome are being tracked to the analogous human site.

   Traditionally, progress in elucidating genetic determinants of drug sensitivity relied on behavioral and biochemical pharmacological levels of analysis.  Recently, molecular-based techniques are contributing to our knowledge.  As candidate genes are identified as having important functions in mediating the biological effects of drugs, the tools of molecular biology are allowing the genetic diversity underlying their expression and function to be explored. The recent advances in QTL mapping strategies and the application of new molecular techniques in selected lines are especially promising, because they suggest that a merger of molecular biological techniques and use of genetic animal models is beginning to occur.  It is increasingly evident that to fully understand the genetics of drug addiction the interface of these methods in the study of drugs of abuse is necessary, especially given the long history of use of genetic animal models in pharmacogenetic research.

ACKNOWLEDGMENTS

   Some of the studies reviewed in the MS were supported by a grant from the Department of Veterans Affairs, and by NIH Grants AA10760, DA05228, AA06243, AA08553, AA07611.  The authors would like to thank Dr. Pamela Metten for comments on an earlier draft of this manuscript.

published 2000