Various mouse models that have been used, or can be used in the future, to study the genetic influences on nicotine responses will be discussed in the sections that follow. Specific examples of experimental results will be included.
The initial screen for genetic effects almost universally involves the measurement of the effects of nicotine in several inbred mouse strains. Inbred mice can be obtained from several suppliers including Jackson Laboratories, which maintains a large number of inbred strains. Inbred mice are a stable breeding population and, barring mutations leading to genetic drift, maintain a constant genotype over time. Consequently, it is not necessary to measure all phenotypes of interest at once. Comparison of results obtained with inbred mice collected at different times and different places is feasible. However, care must be taken to consider that environmental factors such as housing conditions, season, and experimental differences among laboratories can affect the phenotype measured.34
Many experiments have been conducted using a screen of two or more inbred strains, and in virtually every study, the effects of nicotine varied with genotype. Several examples are cited here.
The C57BL/6 and DBA/2 strains of mice are among the most commonly studied of all inbreds; these two strains have been examined for the effects of nicotine on learning,26 locomotor activity,16 35 nicotine metabolism,32 seizure sensitivity,17 oral nicotine intake2836 and nicotine discrimination.29 Although the C57BL/6 and DBA/2 strains are excellent choices for an initial strain comparison, the information obtained may be applicable to only these two strains. In order to obtain more generalizable data, screening of several inbred strains is advisable. Several such screens have been conducted and will be briefly discussed here to illustrate the progression of such strain screens. In one study, four inbred mouse strains (BALB/cByJ, C3H/2Ibg, C57BL/6J, and DBA/2) were screened for their responses to nicotine for several behavioral or physiological tests, including body temperature, rotarod performance, open field activity, respiratory rate, heart rate, and acoustic startle response.15 Each test was administered independently and dose-response curves were constructed for each of the responses to nicotine. In this study, DBA and C57BL/6 mice displayed similar responses to nicotine, while responses of C3H mice differed qualitatively; C3H mice exhibited an increase in open field activity and startle response following nicotine whereas most other strains showed a decrease in these measures following nicotine administration. Inbred strains can also differ quantitatively from one another. C3H mice, for example, differ from C57BL/6, DBA/2, and BALB/cByJ mice in the nicotine dose required to decrease body temperature by 2° and in the dose required to decrease rotarod performance by 50%. Thus, even with this relatively modest comparison of four inbred strains, the advantage of examining strains other than C57BL/6 and DBA was illustrated.
A subsequent analysis of nicotine response used 19 strains.37 The strains studied included sets of closely related mice, such as the C57 family, as well as strains with no obvious genetic relationship with other strains, such as RIIIS/J, SWR/J, and BUB/BnJ. Before beginning these analyses, a multicomponent test battery was developed and validated so that several measures of nicotine response could be made with one mouse (respiratory rate, acoustic startle response, Y-maze crossing and rearing activities, heart rate, and body temperature),38 thereby reducing the number of animals required. Nicotine affected all 6 responses in each of the 19 inbred strains.37 Some responses differed quantitatively (same direction of response, but differences in the dose of nicotine required to elicit the response, e.g., locomotor activity, body temperature) while others differed qualitatively (different direction of response, e.g., acoustic startle). In a parallel study with these same strains, clonic seizures elicited by either ip or iv administration of nicotine were also evaluated.39
When several measurements are made within a large population — in this case strains of inbred mice — responses measured for each test can be compared to those measured for other tests. An initial comparison of responses can be made using regression analysis to evaluate the relationship between two variables. Regression analysis was used to compare the responses among the strains and revealed a wide range of relationships.40 For example, the relative sensitivities of the strains to nicotine-induced changes in Y-maze crosses and rears were highly correlated (r = 0.93), while relative sensitivities to nicotine's effects on Y-maze crossing activities and seizure sensitivity were not related (r = 0.07). A potential explanation for this observation is that the sets of genes influencing Y-maze activities and seizure responses are not the same. This conclusion is based on the assumption that these responses are mediated by multiple genes. It is possible, however, that one or two common genes modulate these responses to nicotine.
When regression analysis detects significant correlations among several variables, further analyses, such as factor analysis (also termed principal components analysis), may be applied. In general, a factor analysis assumes that relationships among several variables occur because these variables are influenced by a relatively few underlying common variables (factors). The goal of a factor analysis is to describe a set of experimental variables with as few underlying factors as possible. The resulting simplification of a complex set of observations may be useful in categorizing observations and improving understanding of the possible relationships among them. Of course, any relationship among variables and calculation of potential underlying factors does not prove a common biological basis for a set of responses, but only suggests such a relationship. When applied to responses to nicotine measured in the inbred mouse strains,37 factor analysis suggested a relatively simple relationship among the responses. Two major groups of responses (two factors), which accounted for about 70% of the variance for the interstrain differences for eight tests, were indicated: one for Y-maze crosses, Y-maze rears, body temperature and, to a lesser extent, heart rate and a second for ip and iv seizures. Respiratory rate and startle response shared properties in common with both of the other groups. Thus, the measurement of many nicotine effects in a relatively large number of strains suggested that several of the responses measured may have similar underlying influences reflecting common genetic influences.
When a large number of strains are tested, it may be useful to group mouse strains by their relative sensitivity, that is, to organize the strains into groups displaying similar characteristics. One such method of organization is cluster analysis, in which individuals or groups are combined with similar individuals. Cluster analysis of the data obtained in the 19-strain study37 suggested four clusters of strains that have similar sensitivity to nicotine's effects on Y-maze activity and temperature measures. The strains ranged from very sensitive (A, C57BL/6, C57BL/10) to very sensitive (ST/b) to very resistant (DBA/1, DBA/2). The value of clustering is that mouse strains can be selected, for future experiments, from each of the clusters to provide a range of responses without testing a large number of inbred mouse strains. It should be noted that strains in the same cluster do not necessarily have similar sensitivity because they are genetically identical, or nearly identical, at relevant loci.
One underlying goal of genetic studies is to evaluate whether a relationship exists between a phenotype measured in the whole animal and a particular molecular, biochemical, or cell physiological process. Such comparisons were made among several behavioral responses to nicotine and the levels of [3H]nicotine binding and [125I]a-bungarotoxin binding,40 which measure two major nicotinic receptor subtypes.4142 Regression analysis revealed a relationship between overall [3H]nicotine binding and Y-maze crossing and rearing activities and body temperature (r = -0.62) and between overall [125I]a-bungarotoxin binding and nicotine-induced seizures (r = -0.63). These results suggest that approximately 30 to 40% of the variance for the effects of nicotine on the Y-maze and body temperature responses may be explained by variability in the number of [3H]nicotine binding sites (i.e., most likely a402-type nAChR9 1043-45) and 30 to 40% of the variance in sensitivity to nicotine-induced seizures may be due to variability in [125I]a-bungarotoxin binding (i.e., a7-type nAChR46 47). In both cases, high basal expression of these receptor subtypes is associated with greater sensitivity to the effects of nicotine (lower ED50-like values). The finding that differences in receptor numbers "explain" only 30 to 40% of the variance implies that variables in addition to the density of binding sites contribute to the differential responses to nicotine measured in the 19 inbred mouse strains. Genetically-based variability in response to nicotine could also arise as a consequence of variation in nicotine distribution and metabolism as well as variability at any step between receptor binding and the ultimate manifestation of the response.
The cause of the strain differences in [3H]nicotine binding is unknown, but analyses for a7 nicotinic receptor differences revealed the presence of a restriction fragment linked polymorphisms (RFLP).48 RFLPs for this gene were identified initially in a screen of ten inbred strains. DBA/1 and DBA/2 mice displayed a different pattern from all other mice tested (including C3H/2 and ST/b). Subsequent analyses (unpublished results) determined that the polymorphism detected by the RFLP occurs in noncoding regions (intron 9), indicating that the polymorphisms did not affect the primary structure of the receptor protein. However, the RFLP may influence binding since an analysis of the association between the RFLP and brain [125I]a-bungarotoxin binding done in F2 mice derived from a DBA x C3H cross indicated that those F2 animals that were homozygous for the C3H-derived RFLP had higher levels of hippocampal [125I]a-bungarotoxin binding than did those animals that were homozygous for the DBA-derived RFLP; heterozygotes had intermediate levels of binding.
Studies with rats indicate that a-bungarotoxin sensitive nicotinic receptors may be important in the regulation of auditory gating. Given the observation that [125I]a-bungarotoxin binding varies among inbred strains, auditory gating was evaluated in nine inbred mouse strains that differ in the number of hippocampal [125I]a-bunga-rotoxin binding sites.49 Auditory gating varied among the inbred strains and a robust correlation (r = 0.72) was seen between test/conditioning ratio and the amount of hippocampal a-[125I]bungarotoxin binding. This is consistent with the proposal that mice expressing higher levels of a7-nAChR show higher auditory gating. This study is an example of using genetic comparisons to test a hypothesis originally proposed based on results obtained in pharmacological studies.
Genetic influences on additional responses potentially important for nicotinic pharmacotherapy and nicotine dependence have also been investigated using inbred mice. For example, eight inbred mouse strains differed in antinociception elicited by the potent nicotinic agonist, epibatidine.19 Three of the strains (A, BALB/c, and DBA/2) were much more sensitive than the other five strains. The relative rank order of sensitivity differed from that observed for either the locomotor/temperature or seizure effects of nicotine,3739 suggesting that genetic influences on epibatidine-mediated antinociception differed from those of the other responses. Inasmuch as nicotinic agonists are being investigated for the relief of pain, information about individual differences in the antinociceptive effects of these drugs will be very critical.
A central aspect of the study of nicotine abuse is the investigation of self-administration. As noted above, Stolerman et al.29 compared C57BL/6 and DBA/2 mice for drug discrimination. Stolerman argued that nicotine discrimination measures the cue that underlies the reinforcing effects of nicotine. Both strains readily learned to discriminate nicotine from saline, but the DBA mice were more sensitive to response-rate reducing effects of nicotine. The finding that these two strains differ only slightly in terms of nicotine discrimination is consistent with the findings of Robinson et al.,28 who reported that the C57BL/6 and DBA/2 strains consumed the highest daily dose of six strains analyzed when either water or 0.2% saccharin was used as the vehicle. Interestingly, a significant inverse relationship between seizure sensitivity and oral intake was observed across the six inbred strains,28 which suggests that a toxic reaction reflected by sensitivity to nicotine-induced convulsions may limit oral intake of nicotine in the mouse.
Inbred mouse strains also differ in the development of tolerance to nicotine.27 This result was obtained in a study that compared tolerance to nicotine using five inbred mouse strains. The strains were also differentially sensitive to the acute effects of nicotine. The mice were chronically infused with nicotine doses ranging between 0 (saline infusion) and 6 mg/kg/h through jugular cannulae and tested for tolerance 2 h after infusion was stopped. Genotype clearly influenced tolerance development. The amount of tolerance depended upon the initial sensitivity of the mice to nicotine: in general, mice initially more sensitive to the acute effects of nicotine developed tolerance following infusion of lower drug doses than did mice that were initially less sensitive. When maximum tolerance had developed, the sensitive strains had approximately the same sensitivity to nicotine as did those strains with low first dose sensitivity. The strains did not differ in the drug-induced upregulation of [3H]nicotine binding and [125I] a-bungarotoxin binding. This finding suggests that tolerance to nicotine is not inextricably due to upregulation of either of the two major neuronal nicotinic receptor subtypes.
The screening of inbred mouse strains will detect a genetic influence on behaviors and suggest relationships among the behaviors, but results from inbred strain comparisons do not provide information about the mode of inheritance of the traits, including information about additive or dominance components, or initial estimates of gene number. In order to obtain information concerning these issues, additional breeding experiments are required. One such breeding experiment, which is rarely used, is the diallel cross in which several inbred mouse strains are crossed to produce all possible F1 hybrids. These hybrids are isogenic (genetically homogeneous). All members of the F1 generation are homozygous at every locus where their parents carried the same allele and are heterozygous at every locus at which their parents differed. The hybrids can be tested and the subsequent results can be analyzed for the presence of additive genetic variance, dominance, and maternal effects.50-52 Diallel crosses allow a genetic analysis to be done in one generation. However, the phenotype will reflect an overall genetic architecture which is not amenable to the identification of specific genes. Diallel crosses have been used to evaluate the effects of nicotine on body temperature53 and open field activity.54 In both studies, F1 hybrid mice were generated by complete reciprocal crosses of five inbred mouse strains (A/J, BALB/ByJ, C3H/2Ibg, DBA/2J, and C57BL/6J) known to differ in response to nicotine.15 Both studies revealed that differential response to nicotine was heritable, with both additivity and dominance contributing to the heritability. Dominance following injection of a modest (for mice) dose of 0.75 mg/kg was toward an increased response to nicotine. This result may mean that intense response to nicotine provides a selective advantage, perhaps because an intense response to a toxic agent would tend to limit consumption of this agent, thereby preventing ingestion of a lethal dose. The diallel analyses also suggested that the regulation of nicotine's effects on open field activity and body temperature were, not surprisingly, polygenic with gene estimates for the nicotine effects of about seven. Such estimates should be regarded with caution because they tend to underestimate the actual gene number.
6.4.3 Classical Genetic Cross Analysis of F1, F2, and Backcross Mice
The classical genetic cross is amenable to both quantitative genetic analysis and the subsequent mapping of specific loci using quantitative trait locus mapping (QTL) discussed later. In this analysis the F1 hybrid is generated by crossing mice of two inbred strains; F1 mice can subsequently be crossed with F1 mice to yield F2 mice or with each of the parental strains to yield backcross mice. The F2 and each backcross are a segregating population with different overall contributions of genes from each parental strain. Quantitative genetic analysis of the responses of the populations of mice generated by the classical genetic cross can subsequently be used to assess additive, dominance, and epistatic contributions to the heritability of the response and can be used to test certain genetic models.51 If several responses of individual mice from the segregating populations are measured, relationships among these responses can be evaluated to determine if they cosegregate. It should be emphasized that each mouse in the segregating populations is unique. As a consequence, the tests used to evaluate these animals must be of the sort where measuring one phenotype does not alter the value of any other phenotype.
Classical genetic crosses have been used to investigate the regulation of nicotine-induced seizures.18 Mice of C3H (sensitive) and DBA (resistant) strains differ in the ED50 for nicotine-induced seizures and were used as the parental strains for a classical genetic cross. Following ip administration of nicotine, dose-dependent increases in seizures were noted for mice of each generation (F1, F2, backcross), thereby demonstrating that seizure sensitivity is heritable. Substantial dominance toward resistance to nicotine-induced seizures was observed. Significantly higher [125I]a-bunga-rotoxin binding was observed in those animals obtained from F1 x DBA and F1 x C3H mice (the backcross generations) that seized following challenge with a test dose that elicited seizures in about 60% of the F2 animals. A similar trend was suggested for F2 mice, supporting the hypothesis that mice with high [125I]a-bun-garotoxin binding are sensitive to nicotine-induced seizures.55 Slightly different results were obtained when nicotine was infused intravenously (iv). As was the case following ip injection, C3H mice were more sensitive than DBA mice following iv administration of nicotine; however, no significant dominance component was observed in the F1 generation.56 Thus, although the relative sensitivity of the parental strains remained the same when the route of nicotine administration was changed, the genetic pattern of inheritance for the two routes of administration varied. These findings suggest that the rate of penetration of nicotine into the brain is important in regulating the seizure response and that different (presumably partially different) sets of genes regulate seizures depending on route of administration.
As noted previously, C3H and DBA/2 mice display an RFLP in the noncoding region of the a7 gene.48 C3H and DBA/2 mice also differ in a polymorphism for the a5 gene.57 This RFLP is also associated with variance in sensitivity to nicotine-induced seizures. In the segregating F2 population derived from a C3H x DBA/2 cross, F2 mice expressing the DBA variant for a5 were less sensitive to nicotine-induced seizures than were mice with the C3H variant. Heterozygotes displayed the same seizure sensitivity as DBA (i.e., there appeared to be dominance toward the DBA genotype). In this same F2 population, mice expressing the DBA variant of a7 were less sensitive to seizures and displayed lower hippocampal (but higher striatal) [125I]a-bungarotoxin binding than did mice expressing the C3H variant. The a7 heterozygotes were intermediate in seizure sensitivity and binding. The a5 and a7 genotypes appeared to act independently, indicating that each contributed to the differential seizure sensitivity. It should be noted here that the a5 RFLP was detected by hybridization with a probe to the DNA for this subunit. Inasmuch as, a5, a3, and P4 form a gene cluster, the association between the a5 RFLP and seizure sensitivity may be a measure of an association with another linked gene such as the a3 or P4 nicotinic receptor genes.
Classical genetic crosses have also been used to evaluate the effects produced by lower nicotine doses. One such study used mice that were selectively bred, starting from a heterogenous stock of mice, for differences in duration of ethanol-induced loss of the righting response (sleep time).58 These lines, designated long sleep (LS) and short sleep (SS), are also differentially affected by nicotine,59 with LS mice generally more sensitive to both ethanol and nicotine. A classical cross analysis that used F1, F2, and backcross generations derived from the LS and SS mice60 demonstrated that nicotine responses measured in each of five tests (respiratory rate, Y-maze crosses, Y-maze rears, heart rate, and body temperature) were heritable and polygenic. The pattern of inheritance varied among the tests suggesting that the genetic regulation of these responses is not identical. Interestingly, sensitivity to ethanol and nicotine segregated together in the F1, F2, and backcross generations. This finding suggests that one or more genes may play an important role in regulating sensitivity to both drugs. This might relate to findings that common genes may influence vulnerability to alcohol and nicotine dependence in humans.6
It should be noted that segregation analyses do not prove that a specific gene contributes to variability in a phenotype. The RFLP strategy, for example, is a form of linkage analysis. Thus, even though it makes sense that variability in a nicotine-induced behavior is due, at least partially, to variability associated with a nicotinic receptor gene, it is premature to draw this conclusion. The association between nicotinic receptor RFLP and phenotype may arise because another gene closely linked to the RFLP is important in regulating variability in the trait. The RFLP data elevate the nicotinic receptor gene to a candidate gene. Other strategies will be required to demonstrate unequivocally the role of the candidate gene.
Recombinant inbred strains are generated by inbreeding through brother/sister mat-ings of mice of a segregating F2 population originally produced by the mating of two inbred strains.61 After 20 generations of brother-sister matings, new inbred strains (recombinant inbred strains, RI) are generated that have stable, recombined genotypes. RI strains are extremely useful for quantitative genetic studies, for uncovering major gene effects, and for QTL mapping, which will be discussed later. Because RI mice are genetically homogeneous, a reliable estimate of the value of the phenotype can be obtained by testing multiple animals from each strain. Thus, RI strains have many of the advantages of inbred mice in that phenotypes can be reliably measured and all relevant data required for analysis need not be collected from a single mouse. Results using the same RI strains can be compared even though the measurements are made at different times or even in different places and, therefore, are cumulative. The analysis of RI strains has limitations: the genetic architecture expressed in the RI strains is limited by the inbreds used to establish the lines, dominance relationships are not measured, and quantitative genetic analyses have reduced power compared to F2 analyses because the number of strains is small.
Construction of new RI strains is an enormous undertaking, but fortunately several sets of RI strains have already been established. Many are available from Jackson Laboratories. Among the well-known RI strains are: BXD (originally established from a C57BL/6 [B] female DBA/2 [D] male mating), AXB, and BXA (originally established from A female x C57BL/6 male and a C57BL/6 female x A male mating, respectively). A set of recombinant inbreds derived from the LS and SS mice are maintained at our home institute, the Institute for Behavioral Genetics.62
RI strain analyses of nicotine responses have only recently been initiated. A recent study63 used LS-SS RI mice to analyze the effects of acute nicotine injection on clonic seizures. An early study59 found that LS mice are more sensitive to nicotine-induced seizures than are SS mice, but the density of both [3H]-nicotine and a-[125I]bungarotoxin binding sites do not differ substantially. Thus, levels of a402-and a7-nAChR are not likely to be major factors in this difference. Stitzel et al.63 found RFLPs for a2, a3, a4, a5, and a6 in LS and SS mice; of these, only the a4 and a6 polymorphisms persist in the RIs. An analysis of seizure sensitivity in the LS x SS RI strains found a wide range of sensitivity in these strains and that those RI strains with the LS-like a4 genotype were more sensitive to nicotine-induced seizures than were RI mice with the SS-like a4 genotype. Conversely, male RI mice with the SS-like a6 genotype were more seizure sensitive than mice with the LS-like a6 genotype. The effect of the a6 genotype was not seen in female mice.
One of the best ways of establishing the heritability of a trait of interest is via selective breeding.64 Selective breeding for a trait of interest clearly establishes a heritable component for that trait. In addition, the main phenotypic value of the lines of animals that are generated generally exceed the extremes of the foundation population. Selective breeding must be started from a heterogeneous population such as an F2 or F3 generation or, even better, a heterogeneous stock (HS) such as that used by McClearn to derive the LS and SS mice.58 This HS stock was derived by interbreeding eight inbred strains and is maintained by outbreeding (avoiding mating animals who share a common grandparent). The goal of selective breeding is to obtain mice that display extreme responses to the drug in succeeding generations. A well-designed selection includes replicates of each selected line as well as control lines that have not been subjected to selective pressure. Selected lines are invaluable tools that can be used to identify genes that contribute to genetically-based variability in the selected phenotype.
Smolen derived replicate lines of mice that differ in locomotion in the Y-maze following nicotine injection of 0.75 mg/kg nicotine.64 These lines were developed starting from the heterogenous stock of mice derived by McClearn.58 Replicate lines, called nicotine-activated (NA) and nicotine-depressed (ND), were developed by within-family selection for six generations. As selection proceeded the ND lines showed a progressive increase in sensitivity to the depressant effects of nicotine. However, selection did not work in the opposite direction: the NA animals were identical in response to nicotine to the unselected controls; i.e., selection did not work in the activated direction. Smolen also measured the effects of nicotine on body temperature and found that this effect was significantly correlated with the locomotor effect in the HS mice. These responses continued to be correlated throughout the selection, suggesting that these two measures are influenced by common genes, a result consistent with the inbred strain analyses discussed.37 Early generations of these selected lines were also used to measure conditioned place preference (CPP) following nicotine. Nicotine injection produced CPP in the NA and control lines, but the ND line did not develop CPP following nicotine treatment,30 indicating that these lines may have been useful for studies of the reinforcing properties of nicotine. Unfortunately, these selected lines no longer exist.
Most behavioral traits arise from the effects of many genes each of which contributes a small amount to variability in the phenotype. In other words, these traits are polygenic. Certainly some genes will have a larger impact on a given response than others, but complete description of the genetic architecture of any phenotype will require identification of the underlying genes. This is obviously an ambitious goal. Quantitative trait locus (QTL) mapping is one method developed which may help achieve this goal. The ever increasing identification of unique chromosomal markers throughout the mouse genome,66 has facilitated the use of QTL mapping.
In principle, QTL analysis is relatively straightforward and has been described in detail elsewhere.66-69 The first step is to identify strains of mice that differ quantitatively in a phenotype of interest. Subsequently, a segregating population derived from the progenitor strains is developed and tested for the phenotype. Optimally, the members of the segregating population should show widely different phenotypes so that identification of differences among mice in the population can be reliably assessed. Given that genetic differences underlie the phenotypic differences, the genes regulating the responses would be expected to differ among the mice displaying the differing phenotypes. The question, then, is where are these genes and how are they found? Initially with the discovery of RFLP differences among mouse strains and now with the availability of a large number of microsatellite markers distributed throughout the genome66 (see also www.informatics.jax.org and www.resgen.com), it has become possible to determine whether phenotypic differences segregate with markers that differ between the progenitor strains. Cosegrega-tion of the markers with the phenotype indicates a linkage between these two measurements and suggests the chromosomal location of the QTL. Relationships between the markers and the phenotype are analyzed by linkage analysis using a program such as MAPMAKER/QTL.70 QTLs that appear to regulate the response can be identified by comparison of phenotypic responses and markers distributed throughout the genome. Candidate genes can be identified or cloned by mapping the area of interest more closely.
Several suggestions for similar strategies to identify QTLs have been advanced. The following progression has been proposed for conducting a QTL analysis:68
1. Identify selected lines or inbred strains that differ in the phenotype of interest.
2. If available, measure the phenotype in RI strains generated from the chosen founders. As noted previously, RI mice are a stable breeding population, and chromosomal marker maps have been generated for several of them and are constantly being expanded. The availability of these maps avoids the need for genotyping of the RIs. A major problem for RI analyses is that most of the RI sets available are small (20 to 40). This limits the statistical power of RI stain analyses which, in turn, limits the power to detect QTLs.
3. Once candidate QTLs have been identified in RI strains, the phenotype can be measured in segregating populations (backcross or F2) with subsequent genotyping and mapping. Extension of the analysis to segregating populations is more labor intensive since each mouse is unique and must be individually tested and genotyped. The F2 analysis may be inappropriate for some studies, particularly one where the goal is to determine whether common genes regulate two phenotypes, and measuring one phenotype affects measurement of the other. However, a major advantage of the F2 approach is it has greater statistical power than does the RI approach because an unlimited number of animals can be produced and tested.
The first step in beginning a QTL analysis is choosing the strains to be analyzed. Eventually the power of the analysis depends upon the expression of a broad phenotype in the segregating populations. It has been suggested that the best progenitor strains are those selectively bred for differences in the trait of interest and subsequently inbred.68 This choice is not available for nicotine studies because no selectively bred lines are available.
In the absence of selected lines, progenitor strains can be chosen after screening inbred strains for the trait of interest and determining that the trait of interest is heritable. It would seem necessary intuitively to choose progenitor strains that differ in the phenotype of interest, but such differences are not obligatory. It may be, for example, that two inbred strains have similar values for a phenotype of interest because each strain has several genes (alleles) that contribute to an increase in the phenotypic value ("plus" alleles) as well as several alleles that contribute to a decrease in the phenotypic value ("minus" alleles). If this is the case the extreme of the segregating population will far exceed the means of the parental strains. Inbred strains that are phenotypically similar are especially useful if analysis of RI strains is to be included.
The second stage of analysis must use segregating populations, such as F2 animals generated from the chosen founder strains. A major disadvantage of testing F2 animals is that the analysis is much more labor intensive than testing RIs because every mouse is unique and must be tested phenotypically. Every animal must also be genotyped. The major advantage of testing an F2 population is that a large number of mice can be tested and the power of the subsequent analysis can be extended beyond that which is possible with an RI strain analysis. With increased power, the number of candidate QTLs can be expanded and candidate genes that explain a smaller fraction of the variance in the value of the phenotype can be identified. It is important to remember that each F2 animal is unique genetically and, consequently, measurement of the phenotype must be reliable or effects, especially small genetic effects, may be lost in measurement error. If several traits are to be evaluated, it is imperative that measuring one phenotype does not alter the value of the others.
Although testing and genotyping of each individual in a segregating population provides the most extensive information about the genetic basis for the behavior, such an analysis can be massive and massively expensive. As an alternative to genotyping each animal, strategies have been proposed that involve measuring the phenotype in large numbers of mice and genotyping only those animals that have the lowest and highest phenotypic values.67 6971 This strategy is especially applicable if the effort involved in phenotypic testing of large numbers of subjects is less than that required to genotype all of the animals.
QTL analyses have not yet been undertaken for the effects of nicotine, but an interesting example for a drug of abuse has been obtained for morphine. A screen of several inbred mouse strains revealed marked strain differences in oral morphine intake.7273 Among the most extreme strains were the DBA/2 (low intake) and C57BL/6 (high intake). Consequently, oral intake of morphine was measured in the BXD RIs.74 QTL mapping methods were not widely available when this analysis was done, but when marker maps and quantitative methods became available, the results from the BXD RI analysis of oral morphine intake study were reexamined.75 This analysis identified several candidate QTLs (on chromosomes 2, 4, 8, and 9). A short time after this a QTL analysis of oral morphine intake of C57BL/6 x DBA/2 F2 mice was done.76 Intriguingly, the results obtained from the F2 analysis differed markedly from those obtained with with the BXD RI strains. The F2 analysis76 identified highly significant QTL on chromosomes 6 and 10, and a less significant QTL on chromosome 1. A potential explanation for these discrepant findings is that RI-strain QTL analyses have low statistical power. This means that any QTL identified in an RI study must be viewed as provisional. The F2 analysis has greater statistical power and the results obtained in such an analysis should be given greater credence. Another reason for giving the F2 analysis greater credence is that (at least for this example), data were obtained that seem to make sense. Specifically, the largest QTL on chromosome 10 found in the F2 study maps precisely to OPRm, the ^ opioid receptor which pharmacological studies and null mutant analysis have identified as the major site of action of morphine.77 Interestingly, the other significant QTLs found in the F2 study do not correspond to the chromosomal locations of 5-, K-, or o-opioid receptors. Therefore, QTL mapping provides some expected results (mapping of oral intake at the chromosomal location of a likely candidate gene, the ^-opioid receptor) but also identifies QTLs that do not correspond to a known opioid receptor.
The pattern of QTLs will likely depend on the inbred strains chosen to establish the F2 mice. It is absolutely imperative to understand that QTL mapping identifies only chromosomal sites harboring a gene that contributes to variability in response. Thus, if two progenitor strains have the same allele for a critical gene that modulates a phenotype, a QTL analysis of the phenotype using F2 mice derived from these strains will not identify a QTL near the chromosomal location of the critical gene. Another QTL analysis using a segregating population derived from inbred strains that are polymorphic at the critical gene should detect a QTL at the map site of this critical gene.
Identifying a QTL is the first step. The next, and more difficult step, is to identify the chromosomal location of the QTL. One method that shows promise for achieving this goal involves using additional segregating populations, for example, outbred mice such as the HS78 or more extensively crossed segregating populations,71 in which linkage between the QTL of interest and markers not tightly coupled to the region of interest have been broken. If these advanced intercross methods provide a more precise location of the QTL, the gene(s) that correspond to the QTL may be cloned from this tightly mapped region of the chromosome.
For a drug that exhibits as complex a pattern of behaviors as does nicotine, QTL analysis may be particularly valuable in determining whether and which of the most likely candidate genes, nicotinic receptor subunits, are implicated in that phenotype. In addition, QTL analyses may suggest additional candidate genes important for manifestation of various phenotypes. Possible candidates that can be envisioned include other receptors activated by neurotransmitters such as dopamine and y-aminobutyric acid — the release of which is stimulated by nicotine — as well as hormones and hormone receptors affected by nicotine such as prolactin, ACTH, or corticosterone.
6.4.7 Reverse Genetics: Studies of the Effects of Candidate Genes
Each of the approaches discussed earlier involved a screen of naturally occurring mouse populations for genetic influences that might explain phenotypic variability in the population. An alternative, and increasingly more common, approach is to identify a candidate gene and investigate the effect of changing the expression or structure of that gene. A very powerful and informative method with which to evaluate the effects of a candidate gene is to construct a null mutant or transgenic mouse targeted to the gene of interest. If alteration of a phenotype is achieved by gene deletion or alteration, the gene in question may be important.
The obvious candidates for primary genetic control of nicotine-mediated phe-notypes are the nicotinic receptor genes. To date, nine mammalian nicotinic receptor genes expressed in the central nervous system have been cloned: a2,79 a3,80 a4,81 a5,80 a6,82 a7,83 P2,84 P3,85 and P4.86 In addition, receptor genes that seem not to be expressed in the brain have also been identified: a9.87 Each of the genes encoding the subunits of the receptor at the neuromuscular junction have also been identified88 as well as their chromosomal localization.89 The chromosomal location of many of the neuronal genes has also been mapped in mice (gene code: chrn, e.g., chrna4 is the a4 gene); see www.informatics.jax.org for current map locations). Given this diverse gene family and the differential expression of its members, examination of the effects of null mutation or of expression of mutated subunits is likely to provide insights into the regulation of the diverse responses to nicotine.90
Several null mutants for nicotinic receptor subunits have been produced and phenotypic analyses of these mutants are in progress: P2,45 P4,91 a3,92 a4,910 and a7.47 Phenotypic analysis of several of these mutants suggests potential roles for these subunits. For example, a3 null mutants are born, but all of the homozygous null mutants die at or near weaning.92 The mice suffer severe deficiencies in the function of the autonomic nervous system, consistent with the crucial role of the a3 subunit in the receptors in the autonomic ganglia. Similarly, mice lacking both the P2 and P4 subunits die due to severe autonomic malfunction,91 while animals lacking either of these subunits are viable. Studies with P2 null mutants suggest that receptors requiring this subunit account for almost all high-affinity nicotine binding in brain and that P2-requiring receptors are important for regulating dopamine release and maintenance of nicotine self-administration.93 The P2 subunit, as well as the a4 subunit, has also been implicated in nicotine-mediated antinociception.9 Null mutation of the a7 subunit eliminates the fast desensitizing response to nicotinic agonists in hippocampal cells, confirming that this response is mediated by a receptor that requires this subunit.47 Further studies of the acute and chronic effects of nicotine with these and other nicotinic receptor null mutants may help to identify the role of defined subunits in response to nicotine.
Changes in receptor function can also affect phenotype and investigations of mice expressing altered receptors may provide additional insights into the regulation of responses to nicotine. For example, variants of the a4 subunit mutated in the channel domain have been identified in human populations and linked to certain seizure disorders.94 95 The two mutations alter the kinetics of the ion channel, leading to a change in desensitization kinetics and Ca2+ permeability.96 Seizure activity is not uniquely mediated by changes in a4 subunit, since analysis of a human population revealed potential QTL near the a3, a5, and 04 gene cluster.74 Directed mutation of the channel domains of several nicotinic receptor subunits has been demonstrated to change receptor function.97-99 A mutation of the a7 gene has been introduced into mice by homologous recombination to generate a receptor that appears to desensitize much more slowly than the wildtype a7 gene.7 Mice homozy-gous for this mutated gene do not survive, while the heterozygotes display receptors with markedly different function. Phenotypic analysis of these mice may prove interesting and also may illustrate that receptors displaying either naturally occurring or artificially introduced mutations may prove to be very useful in unraveling the genetic basis for differential response to nicotine. As is the case with all null mutants, care must be taken when interpreting the results, since failure to express a receptor subunit throughout development may have an unanticipated effect on a phenotype, owing to compensation for the missing gene.100
Was this article helpful?