Mouse genetic models in alcohol research

Mouse genetic models in alcohol research

Review TRENDS in Genetics Vol.22 No.7 July 2006 Mouse genetic models in alcohol research Beth Bennett1, Chris Downing1, Clarissa Parker1 and Thomas ...

172KB Sizes 0 Downloads 92 Views

Review

TRENDS in Genetics Vol.22 No.7 July 2006

Mouse genetic models in alcohol research Beth Bennett1, Chris Downing1, Clarissa Parker1 and Thomas E. Johnson1,2 1 2

Institute for Behavioral Genetics, 447 UCB, Boulder, CO 80309-0447, USA Department of Integrative Physiology, 354 UCB, Boulder, CO 80309-0354, USA

Animal models offer several advantages for the study of complex human disorders such as alcoholism. No animal model replicates all aspects of alcoholism but different components of the disorder can be investigated using various rodent models. In this article, we review a select subset of the most widely used mouse genetic models in alcohol research. Different genetically defined strains and stocks of mice are useful for genetic, physiologic, behavioral and pharmacological studies of this devastating disorder. In the past decade, numerous genomic regions associated with a tendency for various behavioral components of alcoholism have been identified; recent applications of new methods are shedding light on quantitative trait genes. Many of the underlying genes should be identified in the near future.

The impact and cause of alcoholism Alcoholism is a complex disease, with devastating effects on individuals, families and society. In the USA, O18 million adults (w7% of the population) have been diagnosed with alcohol abuse or dependence [1]. Worldwide, the incidence is thought to be 1.7% of adults [2]. In the USA, one in four violent crimes is believed to be committed by individuals who have been consuming alcohol [2], and the estimated economic costs approach $185 billion annually [1]. Given the alarming statistics on alcoholism, it is not surprising that alcohol studies encompass many disciplines, from molecular biology and neuropharmacology to family studies and epidemiology, spawning at least 17 specialty journals. Alcoholism has a significant genetic component: twin studies suggest a heritability of w50% [3]. Ethical concerns and experimental difficulties set limits on studies with human subjects, which can be circumvented by animal models. Various animal stocks, including inbred strains, selected lines and genetically engineered animals enable exploration of the genetic basis of this disorder. Here we introduce the general subject of animal models, and focus on the use of mice in alcohol research, highlighting their strengths and weaknesses. Additional details on animal models in psychiatric disorders are discussed by Geyer and Markou [4], and Hitzemann [5]; specific mouse models by Seong et al. [6]; animal models of drug addiction by Koob [7] and alcoholism specifically, by Corresponding author: Bennett, B. ([email protected]). Available online 26 May 2006

Tabakoff and Hoffmann [8]. A number of animal models exist, many of them designed to investigate molecular, physiologic, behavioral, environmental and treatment questions relating to alcoholism.

Animal models in human disease Goals An animal model attempts to mimic a human condition, with two goals: (i) to test mechanistic hypotheses and; (ii) to test and validate drug treatments [9]. Animal studies also help to illuminate normal function [10]. However, animals lack many of the attributes associated with psychopathologies in humans, such as suicidal behavior and affective disorders, making it difficult to generate an ideal model for these diseases [6]. This problem is particularly salient for diseases such as alcoholism and other addictive disorders, because, of course, animals lack the functional criteria relating to diagnoses (e.g. language, jobs or self-reflection abilities) [4,7,9]. Simulating all or part of a human syndrome depends on the ability to reproduce the defining symptoms of the disorder [4]. Therefore, it is essential to choose the appropriate behavioral assay to model a selected aspect of the disease [10]. For alcohol abuse and dependence, the fourth edition of the Diagnostic and Statistical Manual of Glossary Construct validity: assesses the accuracy with which a model measures the parallel condition in humans. Considered the litmus test of a model, construct validity is nonetheless difficult to assess and must continuously integrate additional data from both animal and human tests. Epistasis: nonlinear interactions between different loci, typically assessed by a statistically significant interaction term in ANOVA. Etiological validity: the causative phenomena of the model are similar or identical to those of the disorder. It is conceptually similar to construct validity, but is perhaps more difficult to assess because causal mechanisms for human psychopathologies are rarely known. Face validity: the phenotype of the animal model resembles that observed in the human disorder. Superficial similarity between animal and human measures might indicate convergence of an underlying etiology. This assessment is probably a good starting point for model development but requires additional validation. Haplotype: a set of tightly-linked genetic markers that are typically inherited together. Comparisons of haplotypes between strains can identify polymorphic regions, which can then be assessed for QTLs influencing traits which differ between the strains. Pleiotropy: multiple phenotypes influenced by the same genes. Predictive validity: the model enables predictions about the human condition; determination necessitates development of appropriate measures in humans to assess the reliability of the model. Quantitative trait locus (QTL): a gene or group of linked genes mapping to a single chromosomal region influencing a quantitative trait.

www.sciencedirect.com 0168-9525/$ - see front matter Q 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.tig.2006.05.005

368

Review

TRENDS in Genetics Vol.22 No.7 July 2006

Table 1. DSM criteria for alcoholism and matched animal model(s)a Diagnostic criteria in humans 1. Increased dose or consumption required for intoxication 2. Withdrawal symptoms (e.g. seizures); alleviated by alcohol 3. Persistent desire to control alcohol use 4. Unintended greater use of alcohol 5. Reduction in important activities as a result of the effect of alcohol 6. Excessive time devoted to alcohol use and recovery 7. Continued alcohol use despite a known alcohol problem 8. LRb a

Animal model Tolerance Withdrawal seizures Conditioned positive reinforcing effects Alcohol intake in dependent animals Choice paradigms Alcohol self-administration Binge intake following alcohol deprivation in selected lines Duration of LORE

Refs [20] [50] [44,45] [46] [44,45,63] [47,62] [14] [17,21]

For more information on these criteria in animal models, see Ref. [12]. The level of response has been shown to be an important predictor of risk [33,34].

b

Mental Disorders, (DSM-IV; [11]) criteria follow from ‘a maladaptive pattern of alcohol use, leading to clinically significant impairment or distress occurring at any time in the same 12-month period.’ (Table 1; [12]). A range of disparate criteria exist including tolerance, withdrawal symptoms, various reinforcing effects and behavioral problems (Table 1). Many of these symptoms replicate the characteristics of other substance-abuse disorders, supporting the trend to add alcoholism to the class of generalized disinhibitory disorders [13].

the nematode, Caenorhabditis elegans [19], have been developed. Mice are particularly well suited for genetic analyses and have long-been used as model organisms. Although they are not as suitable as primates for modeling all aspects of complex human behaviors, they have considerable genetic homology with humans (O99% of mouse genes have a clear human homolog) and share the complex central nervous system organization typical of mammals. The advantages of using mouse models are summarized in Box 1.

Criteria for evaluating models The behaviors listed in Table 1 seem appropriate for simulating a subset of human behavior linked to alcoholism; for example, that animals dependent on alcohol consume increasing doses [14]. But how do we know that a specified animal model truly mimics the human condition? The validity of a model assesses its utility for simulating a specified disorder [4]; a plethora of validity issues exist and are discussed in more detail by Geyer and Markou [4] and Hitzemann [5]. Face and predictive validity (see Glossary) are often used to validate animal models; however, construct and etiological validity are more difficult to apply to such models. A variety of animal models, which might represent endophenotypes, have been developed to model individual alcoholism phenotypes (Table 1). An endophenotype is a characteristic of a complex disorder that is more easily and reliably measured than the (often somewhat subjective) diagnosis. Desirable endophenotypes should be present in both affected individuals and unaffected relatives (face validity), and correlate with diagnosis, severity and heritability (etiological validity) [13]. Another key parameter of a model is its reliability [i.e. the stability and reproducibility of the variable(s) measured]. Many mouse behavioral models, including preference drinking, water maze performance and open field activity, are remarkably robust, reproducing well in different laboratories [15]; although changes in minor procedural details (e.g. rod diameter in ataxia tests) can have a large impact on test results [16]. It is not always desirable to assess reliability by using a test-retest analysis, because in some cases the first test affects the outcome of the second one [17].

Genetic mouse models A variety of genetically defined strains and stocks of mice have been used in alcohol research (Box 2). All of these have advantages and disadvantages for genetic analyses. Heterogeneous stocks (HS) are often used as a starting point for selection studies [20,21] and can be useful for fine-mapping [22], but genotypes are difficult to resolve if many alleles are segregating (B. Bennett, unpublished data). Bidirectional selection produces lines differing dramatically in the selected phenotypes, because alleles influencing the trait are separated in these lines [8]. This approach has been extensively exploited in rodent models of alcohol-related phenotypes. Segregating crosses are a nonrenewable resource for genetic mapping studies (i.e. all genotypes are unique), but numerous genotypes can be generated, increasing experimental power. Congenic

Mouse models In alcohol research, mice, rats and primates have historically been studied, and newer model organisms including the fruit fly, Drosophila melanogaster [18], and www.sciencedirect.com

Box 1. Advantages of using mouse models (i) The Inbred Mouse Strain Resource (IMSR) at the Jackson Laboratory (http://www.informatics.jax.org/imsr/index.jsp) lists 1770 live inbred strains currently available. (ii) The mouse genome has been fully sequenced in five commonly used inbred strains and is readily available (http://www.ensembl. org/Mus_musculus/index.html). Single nucleotide polymorphisms (SNPs) differentiating among these strains can be searched by gene or in 20-Mb intervals (http://www.genenetwork.org/cgi-bin/beta/ snpBrowser.py). (iii) Some experimental manipulations (e.g. implanting brain electrodes) can only be done in animals. (iv) Mice can be genetically manipulated by transgenic, knockout, knockin and siRNA technologies. (Many of the strains that are available are listed at http://www.informatics.jax.org/external/ko/). (v) Mouse maintenance costs are relatively cheap compared with those of primates or rats. (vi) Mutagenesis screens have the potential for identifying novel mutations; thousands have already been identified, phenotyped and mapped (http://www.informatics.jax.org/mgihome/other/phenoallele_commun_resource.shtml). Although none of these is specific to alcohol-related traits (only one such QTL has been identified [50]), genes mutagenized in QTL regions could be tested as candidates.

Review

TRENDS in Genetics Vol.22 No.7 July 2006

Box 2. Mouse genetic resources † Isogenic stocks: inbred strains are derived by brother–sister mating for 20 generations, fixing O99% loci. Such strains are useful for defining behavioral extremes. F1 crosses of inbred strains are also isogenic, although they are heterozygous at all loci. Isogenic stocks eliminate genetic variability as a factor in phenotypic variability and enable repeated sampling of a specified genotype. † Heterogeneous stocks (HS): These genetically heterogeneous mice are either derived from either wild mice or by crossing multiple (four, six or eight) inbred strains. These animals provide additional allelelic diversity beyond that found in segregating crosses. An advantage of HS for genetic mapping is the extensive recombination that accumulates over numerous generations. † Selected lines: bidirectional selection in a nonisogenic founding population (e.g. HS), generates noninbred lines differing in the selected phenotype. Optimally, two replicates and a control, nonselected line, are produced. Selection for high and low responses concentrates the relevant high and low alleles in each line, which can be inbred to produce isogenic stocks. † Segregating crosses: intercrossing F1 (to produce an F2), or backcrossing to one or both inbred progenitors, yields new combinations of the parental alleles, with defined allele frequencies. These nonreplicable genotypes are useful in genetic mapping by testing for linkage of genotype and phenotype. Dominance can be estimated from F2 animals and epistasis from both types of crosses. † Congenic strains: phenotype- or genotype-based selection, combined with three-to-ten generations of backcrossing an F1 to one inbred parental strain (recipient), results in the introgression of a small region of the other (donor) strain onto the recipient genome. These strains are useful for testing hypotheses regarding influence of a single donor region on phenotype. † Consomic strains: these strains are similar to congenics except that an entire donor chromosome is introgressed. Testing a consomic panel for a specified phenotype can quickly localize QTL(s) to a single chromosome. Consomics, like congenic strains, do not enable testing hypotheses of epistasis or interactions between unlinked QTLs. † Recombinant inbred strains: panels of RI strains are generated by crossing two inbred strains. The resulting F2 animals are brother– sister mated in individual lineages; after 20 generations each strain is fully inbred. RI strains are useful for genetic mapping because each isogenic strain can be phenotyped repeatedly, whereas genotyping need only be done once. Their limitation is that most RI sets are small (20–30 strains). The LXS set, containing 75 extant strains, is the largest panel currently in existence. Additional BXD strains are currently being inbred. For more information, see Ref. [62].

strains are useful for testing background effects and confirming QTLs [23]. Recombinant inbred (RI) strains provide numerous advantages for genetic studies including estimation of gene by environment interactions [24]. Development and maintenance of large RI panels is expensive, thus few exist. Mice are particularly amenable to genetic manipulation. Genetically altered mice [9,25], can be produced by over-expression (transgenic), complete elimination (knockout), modification (‘knock in’) and ‘knockdown’ by short interfering RNA segments (siRNA; [26]), in a targeted gene. These mice are then screened for the phenotype(s) of interest [9]. These approaches are most powerful for studying the action of genes of large effect, but less useful for complex traits, genes of smaller effect and possibly epistatic interactions. Phenotypic changes between genetically-altered mice and control littermates can be due to the lack of the targeted allele or compensatory epigenetic changes, as a result of the targeted allele not being present during development [9]. www.sciencedirect.com

369

The importance of genetic background is shown in a striking example, in which a large increase in alcohol consumption was found in mice (with 129/SvPas background) that had a null 5HT1B receptor gene, but not in a replicate strain derived from 129/SvEvTac stem cells, suggesting the presence of substrain-specific genetic modifiers [27]. It might be possible to overcome some of the disadvantages of knockout mice by using RNA interference (RNAi). This technique can partially or completely eliminate the expression of an mRNA transcript (in vitro or in vivo), in defined tissues and developmental stages, although it should be stressed that the degree of silencing in mice is currently variable [26]. Gene identification strategies By definition, complex disorders are determined by numerous genetic and environmental factors [28,29]. Goals for genetic studies of complex traits include identifying genes (or QTLs) influencing a complex trait, assessing epistasis and pleiotropy [8,22,29]. In addition, it is important to define the role of environmental components and their interaction with genotype. Complex phenotypes are exquisitely sensitive to environmental effects, thus even small procedural differences can affect both the behavioral measure (i.e. reliability) and identification of QTLs. Initially, a panel of inbred strains can be screened for a given phenotype. The existence of diverse alleles in different strains, combined with interactions among and with background loci, will affect the phenotype. The traditional approach to QTL mapping is to use two strains that differ maximally in the phenotype as parental strains for genetic crosses, with the following caveats. QTL analysis based on a single cross will most likely reflect only a small portion of the net genetic variation, and QTL detection will be limited to regions where the two progenitor strains have functional polymorphisms. Data from multiple crosses, or from an HS, will overcome this limitation and can also be used to reduce QTL intervals [5,30]. An important goal of QTL mapping is identification of the underlying gene(s), and elucidation of the pathway(s) between gene and phenotype. Several approaches must be combined to reach this goal [22,29,30]; one is illustrated in Figure 1. Fine-mapping follows initial QTL mapping, optimistically resulting in candidate gene identification and eventual confirmation. Recent approaches, combining known QTL regions with differential expression data from microarray experiments and bioinformatic information, such as single nucleotide polymorphism (SNP)-based haplotypes (Box 1) or sequence data from various inbred strains, show promise in identifying candidate genes [30]. Quantitative complementation (QC) involves crossing strains containing a mutant (e.g. from a standard knockout mouse strain) and a wild-type allele of the candidate gene with strains containing the high and low QTL alleles, and then characterizing the F1 progeny for the trait. Failure of the mutation to complement the QTL allele is inferred by a significant cross by line interaction, such failure to complement provides evidence for the candidate gene being a QTL [31]. Here, we will discuss methods that

370

Review

TRENDS in Genetics Vol.22 No.7 July 2006

Inbred strain cross

Quantitative trait loci (QTL)

Congenic strain (confirmation)

Fine-mapping (interval specific congenics, other strategies) < 1 cM or ~ 1 Mb

Coding sequence differences? expression differences? Bioinformatics, haplotype analysis Candidate gene(s) Genetically engineered mice, functional ana lyses, tissue-specific hybridization TRENDS in Genetics

Figure 1. Strategies for identifying genes underlying complex traits. A segregating cross (F2 or backcross) is often a starting point, although selected lines, RI strains and HS have also been used. Confidence intervals for QTLs identified at this stage are often 20–30 cM, necessitating fine-mapping to a much smaller interval. Numerous possibilities then exist, in various combinations, to nominate candidate genes, such as screening public databases for polymorphisms in coding or regulatory regions, testing for differential expression of the candidate or correlating haplotypes in the candidate gene with phenotypes. Other combinatorial strategies can subsequently be used to confirm a candidate, with the most convincing being a targeted effect on the gene, such as conditional knockout or silencing.

have yielded substantial progress in identifying the genetic architecture of the behavioral phenotype. Mouse models of alcohol phenotypes Numerous genetic models have been developed to investigate specific aspects of alcoholism in mice (for a comprehensive review, see Downing et al. [25]). We focus on four commonly investigated phenotypes with strong evidence of genetic control: tolerance, withdrawal, motivational effects and high-dose sensitivity (Box 3). For a summary of significant QTL regions, see Figure 2. The strains used to map these QTLs were derived primarily from LS and SS, or B6 and D2 progenitors; consequently, some regions might be unique to some crosses. Tolerance Tolerance is typically defined as a state of progressively decreasing responsiveness to a particular drug effect. In alcoholism the relationship between initial sensitivity and tolerance is important [32,33]. Schuckit and colleagues have shown that low initial sensitivity, or low level of response (LR), is a good predictor of risk of alcoholism in sons and daughters of alcoholics [33–35], possibly because those with lower LR to alcohol might lack the warning signs that indicate it is time to stop drinking, which could lead to a greater level of alcohol intake and subsequent behavioral and pharmacological tolerance [33]. An alternative hypothesis, that tolerance can be a compensatory response to a homeostatic perturbation [32], predicts that individuals that are more sensitive to a drug will exhibit a greater homeostatic perturbation, and thus have the capacity to develop greater tolerance. Inbred strain differences, selective breeding and QTLmapping studies have clearly demonstrated a genetic effect on tolerance for many ethanol-related phenotypes, including ataxia, hypothermia and loss of the righting reflex (reviewed in Ref. [8]). We limit this discussion to acute tolerance, typically measured as decreased sensitivity to ethanol on a motor coordination task [36], because most mapping studies have been limited to this phenotype. The High Acute Functional Tolerance (HAFT) and Low Acute Functional Tolerance (LAFT) www.sciencedirect.com

Box 3. The phenotypes used in behavioral tests in mouse models of alcoholism (i) Tolerance is typically defined as a state of progressively decreasing responsiveness to a drug effect. Two types of tolerance have been examined, rapid or acute (within-session) and chronic tolerance. Acute tolerance is measured after one or two drug exposures within the same testing session. It is assessed as an improvement in performance of a given task at the same point on the rising and falling blood ethanol concentration curves (BEC). Slight differences in methodological details in this phenotype can produce large differences in behavioral responses, and hence QTLs might also differ. Chronic tolerance is measured after repeated drug exposure. (ii) Withdrawal follows cessation of drinking in dependent individuals producing a constellation of clinical symptoms. Animal models rely on a high level of alcohol intake over a period ranging from a single acute exposure to many days, introducing the same issue discussed above regarding QTL identification. Ethanol administration is typically via vapor inhalation chamber, liquid diet or multiple daily doses administered by injection or by intragastric infusion, then tested hourly for handling-induced convulsions. These models possess good face validity for human withdrawal, including tremors, autonomic hyperactivity, anxiety, seizures and sleeplessness [57]. Withdrawal seizures, which are easy to measure and are scored by severity, are displayed in all species tested. (iii) Motivational effects can be positive (e.g. euphoria, stress reduction) or negative (e.g. hangover) [8]. Animal models that assess motivation fall into two categories: self-administration models, in which the mouse controls its alcohol intake; and conditioning models, in which a fixed dose is administered [44]. (iv) Self-administration models. The two-bottle choice test is the most commonly used protocol [48]; variations in temporal availability enable assessment of intake patterns. One potential problem is that even strains that voluntarily consume large amounts of ethanol will regulate their intake to produce BECs that are much less than those found in human drinkers. (v) Level of response (LR), measured as initial sensitivity, can be measured for virtually any behavioral response to ethanol, including low-dose activation, ataxia, hypothermic effects and high-dose sedative effects. In humans, body sway following low-dose is measured. We measure this phenotype in mice by injecting a high dose of ethanol (20% w:v) into the intraperitoneal cavity and recording the duration of the loss of the righting reflex (LORR) (i.e. the length of time an animal will stay on its back in a V-shaped plastic tray).

Review

TRENDS in Genetics Vol.22 No.7 July 2006

1

3

2

10.0

371

7

9

8

WS2**

AFT4

AFT4

AFT1

2

WS **

AFT1

cM

30.0

50.0

70.0

VEC5 LORE

VEC9*

AFT1

AFT4

LORE7*** WS2** WS2***

LORE8*** LORE8*

*

AFT6 LORE6 AFT4

90.0 AFT4

10.0

VEC5

LORE7** LORE8***

8

VEC2**

AFT4

VEC2** LORE7*

AFT

8

LORE

3

**

LORE6 WS2**

LORE6* AFT1 AFT3 AFT3

WS2** AFT3

WS2** WS2** LORE7*

WS2** WS2**

30.0 cM

WS2*** LORE *** LORE8*

50.0

WS2**

WS2*** WS2***

8

VEC5

LORE6 LORE 8*

WS2***

LORE8**

LORE8**

AFT1

LORE6

LXS RI*

70.0

90.0 11

12

13

14

* P –<0.05; ** P–< 0.01; ***P–< 0.001

15

18

TRENDS in Genetics

Figure 2. A summary of significant mouse QTL chromosomal regions identified for AFT, withdrawal seizures (WS), ethanol consumption (VEC) and LORE. These regions represent work on strains of various origin and from many different laboratories. Thus, it is important to note that these regions almost certainly map loci that can be polymorphic between some strain pairs (e.g. B6 and D2, but not others; e.g. ILS and ISS). In addition, because procedures vary somewhat between labs, what we call a single phenotype is almost certainly a cluster of related phenotypes that can be influenced by different genomic regions. The location is expressed in centiMorgans (cM) because many of the mapping studies were performed using this metric. Peak LOD or P-value positions of the most significant marker are shown (horizontal line); with no indication of a confidence interval. References are shown as superscripts: 1, [16]; 2, [62]; 3, [36]; 4, [37]; 5, [48]; 6, [52]; 7, [17]; 8, [23], 9, [47].

lines were selected for differential ethanol tolerance using a dowel balancing task, which is a measure of motor coordination [20]. Tabakoff and colleagues combined QTL mapping for acute functional tolerance (AFT) in the BXD RI panel (RI lines derived from a cross between C57BL/6J [B6] and DBA/2J [D2] mice) with results from microarray analysis of HAFT and LAFT mice [37,38], to identify five candidate genes with differential expression levels in a QTL interval on chromosome 11. Withdrawal The B6, D2 and derived strains, are strikingly different in their susceptibility to withdrawal after both chronic (vapor-chamber exposure) and acute (high-dose intraperitoneal cavity injection) ethanol [39]. B6, a high ethanolpreferring strain, has a low level of withdrawal severity, whereas D2, a low ethanol-preferring strain, is characterized by extreme withdrawal severity. This pattern holds for other strains that differ in their alcohol preference [39]. B6 and D2 are an ideal strain pair (i.e. they are well-differentiated in the phenotype of interest) www.sciencedirect.com

for genetic mapping studies. In an elegant series of studies, Buck and colleagues used F2 intercross animals [40], RI strains [41] and interval-specific congenic strains [42] to fine-map a large-effect QTL underlying the withdrawal phenotype to a 1-cM region on chromosome 4. Sequencing of coding region from B6 and D2 indicated that only one gene (Mpdz) in this interval had allelic variants with coding sequence and/or expression differences. Mpdz encodes the multiple PDZ domain protein [42] and was also identified by its differential expression between B6 and D2, and its haplotype correlation using 13 inbred strains rated for withdrawal sensitivity [42]. Another genetic model, the withdrawal seizure-prone (WSP) and -resistant (WSR) mice, was generated by this group using selective breeding for increased and decreased HIC severity, respectively, after withdrawal from 72 hours of exposure to alcohol vapor [43]. WSP mice exhibited a tenfold increase in withdrawal severity compared with WSR mice. Selected lines such as these also represent an ideal starting point for crosses aiming to identify underlying genes. The Mpdz genotype cosegregates with withdrawal severity in mice

372

Review

TRENDS in Genetics Vol.22 No.7 July 2006

that are selectively bred for phenotypic differences in the severity of acute withdrawal from alcohol or pentobarbital [42]. Motivational effects Alcohol exerts both positive [e.g. euphoria and anxiolysis (or sedation)] and negative (e.g. hangover) effects that are believed to be important motivators (rewarding or aversive) in alcohol consumption [8]. In self-administration models, consumption and/or preference is completely dependent on the animal. By contrast, in conditioning paradigms [operant self-administration, conditioned place preference (CPP) and conditioned taste aversion (CTA)], the experimenter determines alcohol delivery [44]. Self-administration models yield a simple, quantitative phenotype [ethanol (in grams) consumed per day per kilogram of body weight], lending this approach to genetic mapping. However, one problem with voluntary consumption models is that even the highest-consuming strains do not approach the high blood ethanol concentration (BEC) levels of human alcoholics. Conditioning models, many of which have been developed and refined in rats, yield valuable information on patterns of consumption and associations between site and taste of consumed solutions [44]; however, only one mapping study has been undertaken to map CPP in BXD RIs [45]. In addition, recent animal models incorporating reinstatement and deprivation seem to provide good face and predictive validity for dependence and craving [46], but as yet there are no genetic studies involving these models. Self-administration models This phenotype is typically studied in B6, D2 and derivative strains, because these strains show a dramatic and consistent difference in voluntary consumption. B6 females consume w16 g/kg/day and D2 w0.2 g/kg/day [47]. It is difficult to relate this amount to the amount consumed by human alcoholics, because intake, tolerance and metabolism vary in humans, and rodent and human metabolic rates are different. For example, excessive drinking is typically defined as four or more drinks per day. At an average of 15 g per drink and assuming a 150 lb (68 kg) body weight, this figure in humans would be just under 1 g/kg/day. Importantly, the absolute amount of alcohol consumed by humans is not diagnostic; the downstream behavioral effects are the relevant criteria (Table 1), arguing that face or predictive validity of mouse models should not be based solely on amount of ethanol consumed. One of the most consistently identified ethanol QTLs is a region on chromosome 2 that is associated with alcohol preference (Alcp1) (reviewed in Ref. [48]). Several groups have identified the syntaxin-binding protein 1 (Stxbp1) at 32.7 Mb as a potential candidate for Alcp1. Worst and colleagues [49] screened two selection sets of preferring and nonpreferring rats for differential gene expression, finding Stxbp1 upregulated significantly in one set but not the second set. Fehr and colleagues [50] correlated genotype in BXD RIs (B6 and D2 are polymorphic for a SNP in the coding sequence that changes the secondary structure of the protein), an F2 intercross, two short term www.sciencedirect.com

selected lines differing in ethanol consumption and an inbred strain panel, with ethanol consumption. Genotypephenotype correlations were impressive [50]; however, there was no differential expression of Stxbp1 between B6 and D2, suggesting a complex role of this gene in determining the phenotype [49]. This QTL was confirmed in congenics in our laboratory and dissected in 14 interval-specific congenic strains, which spanned the QTL interval, dividing the original D2 region of 35 cM into much smaller regions [47]. Three of these strains had a D2 phenotype for reduced alcohol consumption. One strain included Stxbp1 but its decreased ethanol consumption was not significant [47]. The overlap of two of these strains reduced the QTL interval to 3.4 Mb (71.8–75.2 Mb [47], 40 Mb distal to Stxbp). This reduced region is still gene-rich, with 31 known genes (Ensembl v 36: http://www.ensembl.org/ Mus_musculus/index.html). A useful approach for identifying candidate genes in strains derived from B6 or D2 involves correlating gene expression data with phenotypic data in the BXD RI strain at WebQTL (http://www. genenetwork.org/search.html). Approximately a third of these genes showed significant correlations between their expression levels and ethanol preference phenotypes (B. Bennett, unpublished data). This finding further implicates this region, although no particular gene is necessarily supported, because linked genes could be coordinately regulated. Level of response (LR) The inbred Long Sleep (ILS) and inbred Short Sleep (ISS) mouse strains, like other selected strain pairs, are a valuable resource for addressing the genetic bases of this behavior [21]. The ILS and ISS and their progenitors have been widely used in alcohol research, and they have been cited in some 400 publications over the past 30 years. These strains were derived by selection for differential sensitivity to high-dose ethanol, assayed by duration of loss of righting reflex (LORE), but also show differential responses to many other ethanol-induced behaviors including low-dose activation [51], acute functional tolerance [16], hypothermia [52] and anxiolysis [53]. The selected phenotype is a measure of initial sensitivity, and as such suggests an overlap with LR (Table 1) in humans [33,34]. LORE is a high-dose response, whereas LR is determined by low-dose ethanol, but both of them measure initial sensitivity. LORE correlates with other, low-dose phenotypes in ILS and ISS (B. Bennett, unpublished data), although this is not true in other, nonselected strain comparisons [54]. The role of LR as a risk factor for alcoholism is unclear (discussed previously), but low LR could be mediated by larger development of acute tolerance in these individuals. The original derivation of the ILS and ISS strains has been described in detail by Bennett and colleagues [23]; briefly, these strains trace their descent from a heterogeneous stock [21] derived from an eight-way cross of inbred strains (A, AKR, BALB/c, C3H/2, C57BL, DBA/2, Is/Bi and RIII). To generate the LXS RI strains, ILS and ISS mice were reciprocally intercrossed, with 75 fully inbred strains still extant [24].

Review

TRENDS in Genetics Vol.22 No.7 July 2006

373

Box 4. Identifying human orthologs of ethanol QTLs Candidate genes for ethanol sensitivity, loss of righting induced by ethanol 1 and 2 (Lore1 and Lore2), identified in mouse QTL regions can be assessed in human populations when a similar phenotype (e.g. LR) has been mapped. The COGA study http:// www.niaaa.nih.gov/ResearchInformation/ExtramuralResearch/SharedResources/projcoga.htm) has assembled a collection of O300

extended families affected by alcoholism, consisting of O3000 individuals. Clinical, neuropsychological, electrophysiological, biochemical and genetic data have been collected and analyzed in numerous studies. A summary of human QTL regions, or candidate genes, which overlap with those in mouse, is given in Table I.

Table I. Human orthogs of ethanol QTLsa Human chromosomal location (Mb) 2 (216.0) 5 (131.7) 1 (125-175) 2 (110-150)

Mouse region

Gene

Refs

Evidence

2 (72.6) 11 (53.7) 1 (158–175); Lore1 [55] 2 (127–129); Lore2 [55] 1 (118–120); Lore1 [55]

Xrcc5 (encodes an X-ray repair protein) Slc22a4 (encodes a cation transporter) NA NA

[60,61] [59,61] [35,54] [35,54]

DE in ILS, ISS DE in ILS, ISS Overlap of QTLs Overlap of QTLs

a

Abbreviation: NA, not available.

We have focused on initial sensitivity to the sedative effect of ethanol, a phenotype with large heritability (w40%), making it amenable to genetic characterization, substantial test-retest reliability (rZ0.69, P!0.001) [17] and face validity to LR. We mapped QTLs for sensitivity to a sedative dose of ethanol, using a small panel of 25 RI strains (LSXSS), derived from the non-inbred LS and SS [55], and, subsequently, a large F2 intercross between ILS and ISS of (1000 mice [17], and the LXS RI [52]. We confirmed and captured four of the major QTLs for this trait identified in congenic strains on each background [23] and narrowed the interval surrounding the QTL on two chromosomes to O12 Mb [56]. We have confirmed five previously identified QTLs [23,52] and found one additional significant region in the LXS RI panel [56], which, with 75 strains and w5000 SNP markers, provides substantial power for identifying and fine-mapping smalleffect QTLs [57]. One such candidate, the norepinephrine transporter (NET; [58]) was identified by a combination of approaches including RI testing, sequence polymorphisms, and expression and activity differences between ILS and ISS (Figure 1). Concluding remarks and future directions Identification of the QTL gene(s) and elucidation of their biochemical pathways will facilitate basic scientific understanding of complex phenotypes such as alcoholism. The use of genomic resources, such as inbred strain sequence information, expression arrays, haplotype analysis and bioinformatics databases, will ultimately parse the genetic architecture of the model systems described here. One such approach using human QTL regions identified by the Collaborative Study on the Genetics of Alcoholism (COGA) study [35,59,60], that sought to identify overlapping mouse and human QTL regions [61] and candidate genes therein, is illustrated in Box 4. If these models are valid endophenotypes, this approach will result in candidate gene identification and, in the future, diagnostic and treatment options for alcoholism. Acknowledgements This work was supported by the University of Colorado and the NIH (RO1 AA08940 and AA11984). Our collaborators at the University of Tennessee at Memphis (R.W. Williams, L. Lu, E. Chesler and J. Peirce) and the Wellcome Trust (J. Flint and R. Mott) have provided material and www.sciencedirect.com

intellectual support. Colleagues too numerous to mention at the University of Colorado have done likewise; we especially appreciate the institutional support (AA014666 and DA015663) from the Institute for Behavioral Genetics for our mouse colony.

References 1 Li, T-K. et al. (2004) Alcohol use disorders and mood disorders: a national institute on alcohol abuse and alcoholism perspective. Biol. Psychiatry 56, 718–720 2 Grant, B.F. et al. (2004) The 12-month prevalence and trends in DSMIV alcohol abuse and dependence: United States, 1991-1992 and 20012002. Drug Alcohol Depend. 74, 223–234 3 Reich, T. et al. (1999) Genetic studies of alcoholism and substance dependence. Am. J. Hum. Genet. 65, 599–605 4 Geyer, M.A. and Markou, A. (1995) Animal models of psychiatric disorders. In Psychopharmacology: The Fourth Generation of Progress (Bloom, F.E. and Kupfer, D.J., eds), pp. 787–797, Raven Press 5 Hitzemann, R. (2000) Animal models of psychiatric disorders and their relevance to alcoholism. Alcohol Res. Health 24, 149–158 6 Seong, E. et al. (2002) Mouse models for psychiatric disorders. Trends Genet. 18, 643–650 7 Koob, G.F. (1995) Animal models of drug addiction. In Psychopharmacology: The Fourth Generation of Progress (Bloom, F.E. and Kupfer, D.J., eds), pp. 759–772, Raven Press 8 Tabakoff, B. and Hoffman, P.L. (2000) Animal models in alcohol research. Alcohol Res. Health 24, 77–84 9 Crawley, J.N. (2000) What’s wrong with my Mouse? Behavioral Phenotyping of Transgenic and Knockout Mice, Wiley-Liss 10 Crabbe, J.C. and Morris, R.G. (2004) Festina lente: late-night thoughts on high-throughput screening of mouse behavior. Nat. Neurosci. 7, 1175–1179 11 American Psychiatric Association (2000) Diagnostic and Statistical Manual of Mental Disorders, (4th edn) American Psychiatric Association 12 Koob, G.F. (2004) A role for GABA mechanisms in the motivational effects of alcohol. Biochem. Pharmacol. 68, 1515–1525 13 Porjesz, B. et al. (2005) The utility of neurophysiological markers in the study of alcoholism. Clin. Neurophysiol. 116, 993–1018 14 Becker, H.C. and Lopez, M.E. (2004) Increased ethanol drinking after repeated chronic ethanol exposure and withdrawal experience in C57BL/6 mice. Alcohol. Clin. Exp. Res. 28, 1829–1838 15 Crabbe, J.C. et al. (1999) Genetics of mouse behavior: interactions with laboratory environment. Science 284, 1670–1672 16 Deitrich, R.A. et al. (2000) Phenotypic and genotypic relationships between ethanol tolerance and sensitivity in mice selectively bred for initial sensitivity to ethanol (SS and LS) or development of acute tolerance (HAFT and LAFT). Alcohol. Clin. Exp. Res. 24, 595–604 17 Markel, P.D. et al. (1997) Confirmation of quantitative trait loci for ethanol sensitivity in long-sleep and short-sleep mice. Genome Res. 7, 92–99 18 Heberlein, U. (2000) Genetics of alcohol-induced behaviors in Drosophila. Alcohol Res. Health 24, 185–188

374

Review

TRENDS in Genetics Vol.22 No.7 July 2006

19 Crowder, C.M. (2004) Ethanol targets: a Bk channel cocktail in C. elegans. Trends Neurosci. 27, 579–582 20 Erwin, V.G. and Deitrich, R.A. (1996) Genetic selection and characterization of mouse lines for acute functional tolerance to ethanol. J. Pharmacol. Exp. Ther. 279, 1310–1317 21 McClearn, G.E. and Kakihana, R. (1981) Selective breeding for ethanol sensitivity. Short-sleep and long-sleep mice. In Development of Animal Models as Pharmacogenetic Tools (McClearn, G.E., Deitrich, R.A. and Erwin, V.G., eds), pp. 81–113, US Government Printing Office 22 Flint, J. et al. (2005) Strategies for mapping and cloning quantitative trait genes in rodents. Nat. Rev. Genet. 6, 271–286 23 Bennett, B. et al. (2002) Reciprocal congenics defining individual quantitative trait loci for sedative/hypnotic sensitivity to ethanol. Alcohol. Clin. Exp. Res. 26, 149–157 24 Williams, R.W. et al. (2004) Genetic structure of the LXS panel of recombinant inbred mouse strains: a powerful resource for complex trait analysis. Mamm. Genome 15, 637–647 25 Downing, C. et al. (2005) Alcoholism: QTL mapping. In Comparative Genetics and Genomics (Peltz, G., ed.), pp. 195–248, Humana Press 26 Glaser, S. et al. (2005) Current issues in mouse genome engineering. Nat. Genet. 37, 1187–1193 27 Phillips, T.J. et al. (1999) Complications associated with genetic background effects in research using knockout mice. Psychopharmacology 147, 5–7 28 Crabbe, J.C. (2002) Alcohol and genetics: new models. Am. J. Med. Genet. 114, 969–974 29 Abiola, O. et al. (2003) The nature and identification of quantitative trait loci: a community’s view. Nat. Rev. Genet. 4, 911–916 30 Hitzemann, R. et al. (2004) On the integration of alcohol-related quantitative trait loci and gene expression analyses. Alcohol. Clin. Exp. Res. 28, 1437–1448 31 Rikke, B.A. et al. (1998) Towards the cloning of genes underlying murine QTLs. Mamm. Genome 9, 963–968 32 Kalant, H. et al. (1971) Tolerance to, and dependence on, some nonopiate psychotropic drugs. Pharmacol. Rev. 23, 135–191 33 Schuckit, M.A. (1994) Low level of response to alcohol as a predictor of future alcoholism. Am. J. Psychol. 151, 184–189 34 Eng, M.Y. et al. (2005) The level of response to alcohol in daughters of alcoholics and controls. Drug Alcohol Depend. 79, 83–93 35 Schuckit, M.A. et al. (2005) Autosomal linkage analysis for the level of response to alcohol. Alcohol. Clin. Exp. Res. 29, 1976–1982 36 Gehle, V.M. et al. (2000) The genetics of acute functional tolerance and initial sensitivity to ethanol for an ataxia test in the LSXSS RI strains. Alcohol. Clin. Exp. Res. 24, 579–587 37 Kirstein, S.L. et al. (2002) Quantitative trait loci affecting initial sensitivity and acute functional tolerance to ethanol-induced ataxia and brain camp signaling in BXD recombinant inbred mice. J. Pharmacol. Exp. Ther. 302, 1238–1245 38 Tabakoff, B. et al. (2003) Selective breeding, quantitative trait locus analysis, and gene arrays identify candidate genes for complex drugrelated behaviors. J. Neurosci. 23, 4491–4498 39 Metten, P. et al. (2005) Alcohol withdrawal severity in inbred mouse (Mus musculus) strains. Behav. Neurosci. 119, 911–925 40 Buck, K.J. et al. (2002) Mapping murine loci for physical dependence on ethanol. Psychopharmacology (Berl.) 160, 398–407 41 Crabbe, J.C. et al. (1994) Genetic determinants of sensitivity to ethanol in inbred mice. Behav. Neurosci. 108, 186–195 42 Fehr, C. et al. (2002) Congenic mapping of alcohol and pentobarbital withdrawal liability loci to a !1 centimorgan interval of murine chromosome 4: Identification of Mpdz as a candidate gene. J. Neurosci. 22, 3730–3738

www.sciencedirect.com

43 Crabbe, J.C. et al. (1993) Selective breeding for alcohol withdrawal severity. Behav. Genet. 23, 171–177 44 Cunningham, C.L. et al. (2000) Animal models of alcohol’s motivational effects. Alcohol Res. Health 24, 85–92 45 Cunningham, C.L. (1995) Localization of genes influencing ethanolinduced conditioned place preference and locomotor activity in BXD recombinant inbred mice. Psychopharmacology 120, 28–41 46 Spanagel, R. (2003) Alcohol addiction research: from animal models to clinics. Best. Pract. Res. Clin. Gastroenterol. 17, 507–518 47 Ruf, C. et al. (2004) Confirmation and genetic dissection of a major quantitative trait locus for alcohol preference drinking. Alcohol. Clin. Exp. Res. 28, 1613–1621 48 Belknap, J.K. and Atkins, A.L. (2001) The replicability of QTLs for murine alcohol preference drinking behavior across eight independent studies. Mamm. Genome 12, 893–899 49 Worst, T.J. et al. (2005) Transcriptome analysis of frontal cortex in alcohol-preferring and nonpreferring rats. J. Neurosci. Res. 80, 529–538 50 Fehr, C. et al. (2005) The syntaxin binding protein 1 gene (stxbp1) is a candidate for an ethanol preference drinking locus on mouse chromosome 2. Alcohol. Clin. Exp. Res. 29, 708–720 51 Downing, C. et al. QTL mapping for low-dose ethanol activation in the LXS recombinant inbred strains. Alcohol. Clin. Exp. Res. (in press) 52 Bennett, B. et al. (2004) Genetic mapping for ethanol-related behaviors in the LXS recombinant inbred strains from ILS and ISS. Alcohol. Clin. Exp. Res. (Suppl.) 28, 87 53 Parker, C. et al. (2004) 2. Ethanol-mediated anxiety reduction in inbred long-sleep and inbred short-sleep mice on the elevated zero maze: a pilot study. Alcohol. Clin. Exp. Res. (Suppl.) 28, 90 54 Crabbe, J.C. et al. (2005) An analysis of the genetics of alcohol intoxication in inbred mice. Neurosci. Biobehav. Rev. 28, 785–802 55 Markel, P.D. et al. (1996) Quantitative trait loci for ethanol sensitivity in the LSXSS recombinant inbred strains: Interval mapping. Behav. Genet. 26, 447–458 56 Bennett, B. et al. (2002) Genetic dissection of quantitative trait loci specifying sedative/hypnotic sensitivity to ethanol: Mapping with interval-specific congenic recombinant lines. Alcohol. Clin. Exp. Res. 26, 1615–1624 57 Bennett, B. et al. (2005) Replication of small effect QTLs for behavioral traits: Estimation of effect size from independent cohorts. Genes Brain Behav. DOI 10.1111/j.1601-183X.2005.00174.x (http://www.blackwellsynergy.com/loi/gbb) 58 Haughey, H.M. et al. (2005) Norepinephrine transporter: A candidate gene for initial ethanol sensitivity in Inbred Long-Sleep and ShortSleep mice. Alcohol. Clin. Exp. Res. 29, 1759–1768 59 Wilhelmsen, K.C. et al. (2003) The search for genes related to a lowlevel response to alcohol determined by alcohol challenges. Alcohol. Clin. Exp. Res. 27, 1041–1047 60 Schuckit, M.A. et al. (2001) A genome-wide search for genes that relate to a low level of response to alcohol. Alcohol. Clin. Exp. Res. 25, 323–329 61 Maclaren, E.J. et al. (2006) Expression profiling identifies novel candidate genes for ethanol sensitivity qtls. Mamm. Genome 17, 147–156 62 Crabbe, J. et al. (1998) Quantitative trait loci: Mapping drug and alcohol related genes. Adv. Pharmacol. 42, 1033–1037 63 Becker, H.C. (2000) Animal models of alcohol withdrawal. Alcohol Res. Health 24, 105–113