Introduction

Top Abstract Introduction Methods Results Discussion References

NT is a 13 amino-acid neurotransmitter found in both the CNS and gut. As with classical neurotransmitters, there are specific NTR. Genetic, pharmacological and behavioral studies indicate the existence of at least two types of neurotensin receptor (Pettibone et al., 1988; Erwin et al., 1993; Labbe-Jullie et al., 1994), named for their relative binding affinities: high and low. When administered centrally, NT acts through NTR to produce multiple pharmacological effects, including analgesia (Clineschmidt et al., 1979; Kalivas et al., 1982a; Hernandez et al., 1984), hypothermia (Nemeroff et al., 1977; Martin et al., 1980; Jolicoeur et al., 1981; Hernandez et al., 1984) and locomotor activation (Kalivas et al., 1981, 1982b) or inhibition (Nemeroff et al., 1977; Clineschmidt et al., 1979; Jolicoeur et al., 1981), depending on dose and site of administration. These effects are similar to the pharmacologic effects of acute ethanol (e.g., Erwin et al., 1990a).

In addition to having similar effects, NT and ethanol show interesting pharmacological and neurochemical interactions. NT potentiates the hypnotic, hypothermic and locomotor activity effects of ethanol (Frye et al., 1981; Luttinger et al., 1981; Widdowson, 1987; Erwin and Su, 1989). Acute doses of ethanol cause a brain region-specific decrease in NT-ir (Erwin et al., 1990b), although chronic ethanol produces increases in NT-ir levels (Erwin et al., 1992a) and changes in NTR. NTRs are up- or down-regulated in a brain region-specific manner after 2 wk of chronic ethanol (Erwin et al., 1992a; Campbell and Erwin, 1993) and the binding characteristics of NTR are changed after a similar regimen of chronic ethanol (Campbell and Erwin, 1992). Also, chronic ethanol or chronic NT each produce tolerance to the hypothermic and locomotor activity effects of acute doses of both drugs, i.e., cross-tolerance (Erwin et al., 1995). Therefore, it is likely that at least some of the effects of ethanol on the CNS are mediated by NT.

Low doses of ethanol have long been known to affect locomotor activity in rodents (Carlsson et al., 1972). For inbred mouse strains, consistent within-strain locomotor responses to ethanol are seen and these responses are dose- and genotype-dependent (Erwin et al., 1990a; Dudek et al., 1991; Phillips et al., 1995). Additionally, because locomotor activity is a postulated method for investigating the reward potential of drugs in animal models of addiction (Wise and Bozarth, 1987), this behavior provides a paradigm for studies in mice of alcohol sensitivity and the genetics of alcoholism. The LS and SS lines of mice, selected for differing sensitivities to the high dose hypnotic effects of alcohol, also differ in their locomotor responses to low dose ethanol (Erwin and Su, 1989; Phillips and Dudek, 1991). Recombinant inbred strains produced from the LS and SS mice (LSXSS RI) show a normal distribution of responses to low-dose ethanol and have been demonstrated to be a good model for study of genetic components of ethanol-related behaviors, including locomotor activation (Erwin et al., 1990a). Thus, ethanol-induced locomotor activity in the LSXSS RI is an excellent animal model for studying the genetics of ethanol's actions on the CNS.

The first mapping of genes mediating ethanol's effects on locomotor activity (Oliverio and Eleftheriou, 1976) identified a locus on chromosome 4 accounting for a major part of the inhibitory effects of ethanol on locomotor activity in the BALB/cBy X C57BL/6By recombinant inbred strains (CXB RI). More recent work has shown the effect of ethanol on locomotor activity to be a polygenically determined phenotype (e.g., Erwin and Jones, 1993) and a chromosomal map of polygenes mediating this effect of ethanol has been reported (Phillips et al., 1995). Phillips et al. (1995) located QTL associated with the locomotor effects of 2.0 g/kg ethanol in the C57BI/6J X DBA/2J recombinant inbred mouse strains (BXD RI). The QTL tentatively identified in that study have not yet been confirmed in other strains; they also do not provide advances toward a mechanistic theory of alcohol's actions on the CNS in eliciting locomotor effects. We have proposed that NT mediates, in part, the central actions of ethanol, including hypnotic sensitivity as well as locomotor effects (Erwin and Jones, 1993; Erwin et al., 1994). Common QTL for ethanol-induced locomotor activation and CNS neurotensin measures would support and advance this proposal. Thus, the research described herein had three aims: 1) to identify ethanol-induced locomotor activation QTL in the LSXSS RI, 2) to demonstrate the use of a new protocol for measuring locomotor activation and 3) to determine if any QTL for ethanol-induced locomotor activation are in common with QTL for neurotensin peptide and receptor levels. We chose to use the LSXSS RI strains of mice for these studies because there are endogenous differences in NT-ir and NTR measures in the LSXSS RI mice (Erwin et al., 1993) and NTR levels in these mice show significant correlations to ethanol-related behaviors (Erwin and Jones, 1993), thereby increasing the likelihood of identifying common QTL. Additionally, because the BXD RI were used by Phillips et al. (1995) for a QTL analysis of ethanol-induced locomotion, identification of similar QTL by us in a different RI panel (the LSXSS RI) decreases the probability that these QTL represent type I errors of analysis.

	Methods

Top Abstract Introduction Methods Results Discussion References

Animals. Male LSXSS RI strains of mice were obtained from the Institute of Behavioral Genetics, University of Colorado, Boulder, CO. All experiments were conducted with mice (60-80 days of age) which were maintained in a constant temperature (22°C), humidity (20%), and light (12L/12D) environment. Separate groups of mice were used for each ethanol dose.

Locomotor activity. Animals received injections i.p. with saline (day 1) and ethanol (15% v/v, day 2) at doses ranging from 1.0 to 2.0 g/kg and immediately placed in Omnitech Activity Monitors (Omnitech Electronics, Inc., Columbus, OH) to measure spontaneous locomotor activity as horizontal distance (cm) traveled after the injections. The activity monitors were enclosed in ventilated boxes (equipped with 5 watt light bulbs), and activity was monitored at 5-min intervals for 15 min by means of an MS-DOS computer. Distance traveled between 5 and 15 min was used for all analyses as previous studies have shown that blood ethanol concentrations peak within the first 5 min after ethanol administration, i.p. (V. G. Erwin, unpublished data). Ambient temperature was maintained at 22 to 23°C.

Neurotensin extraction, radioimmunoassays and neurotensin binding assays. Procedures for performing these experiments are presented in the accompanying manuscript (Erwin et al., 1997) and are described in detail by Erwin and Jones (1989) and Erwin et al. (1990b). The mean values for NT levels and receptor densities in specific brain regions from LSXSS RI mice, including hypothalamus, nucleus accumbens and ventral midbrain, have been published previously (Erwin et al., 1993).

QTL analysis. QTL analyses were performed as described in the companion manuscript by Erwin et al. (1997). LSXSS RI strains were genotyped as previously published (Markel and Johnson, 1994) using 120 SSLP markers (Research Genetics, Huntsville, AL) found to be polymorphic in LS and SS parental strains. These markers covered the mouse genome at an average marker interval of 15 cM. Strain distribution patterns were established for all 20 linkage groups in 22 to 24 RI strains (Markel et al., 1996). Because the LS and SS parental lines were not completely inbred, more than two alleles exist for some markers among the RI strains. Ten percent (12 of 120) markers gave three or more alleles and as a result of being unable to know the exact genotype (frequency of alleles) in the outbred LS and SS progenitors of the RI, we were unable to use the interval mapping method described by Markel et al., 1996, in identifying QTL. Therefore, one-way analyses of variance were carried out with the phenotypic measure as the dependent variable and the RI strains were grouped by allele type. All statistical analyses were performed with SPSS Version 6.0 for Windows. Because this was designed as an exploratory study to identify provisional QTL for ethanol-induced locomotor activity as well as common QTL for two independent phenotypes (ethanol-induced locomotor activation and NT measures), a level of significance, P < .05, was used. This  level was chosen specifically to decrease the type II error rate and improve the probability of identifying important QTL, similar to the procedures adopted by others in initial studies of alcohol behaviors in a limited number of RI strains (Crabbe et al., 1994; Phillips et al., 1995; Rodriguez et al., 1995). As a consequence, the type I error rate is increased, requiring confirmation of these provisional QTL in future studies with F₂ populations derived from the extreme responders among the LSXSS RI strains.

	Results

Top Abstract Introduction Methods Results Discussion References

The frequency distribution for locomotor activity after an i.p. injection of saline (data not shown) showed that the LSXSS RI produce a normal distribution (Kolmogorov-Smirnov goodness of fit, Z = 0.792, P = .56) indicating that locomotion after saline is a quantitative trait and that the responsible genes have segregated independently in the RI strains. For this study, we wished to identify only those genetic loci mediating the effects of ethanol on locomotor activation and not confound our results with loci involved in stress-induced locomotor activity, i.e., activity after saline administration, i.p. Previous studies have used differences scores, obtained by subtraction of saline activity from ethanol activity (Phillips and Dudek, 1991; Tritto and Dudek, 1993; Phillips et al., 1995) or analysis of covariance of saline and ethanol activity (Erwin et al., 1990a), to separate stress-induced activity from that caused by the pharmacological effects of ethanol. We chose to use difference scores, so to determine locomotor activity for each ethanol dose, day 1 activity after saline was subtracted from day 2 activity after ethanol. These difference scores were then used to calculate the ethanol activation slope.

Determination of ethanol-induced locomotor activation is complicated by ethanol's biphasic dose-response curve for locomotor activity (Dudek et al., 1991). A common dose used to measure alcohol-induced locomotor activation is 2.0 g/kg (Phillips and Dudek, 1991; Phillips et al., 1995). As shown by representative dose-response curves in figure 1 and the data in table 1, this dose poorly defines locomotor activation in the LSXSS RI. Some of the RI strains have peaked for locomotor activation at lower doses and have activity levels equivalent to saline activity at a doseage of 2.0 g/kg, indicating an equilibrium has been approached between the activating and inhibiting effects of ethanol. If the genes controlling the stimulatory effects of alcohol are different from those controlling alcohol's inhibitory effects, as seems likely (Crabbe et al., 1982; Dudek et al., 1991), identifying only the genes mediating ethanol's stimulatory effects is made difficult in a QTL analysis using the activity induced by 2.0 g/kg ethanol. Subsequently, we have devised an alternative measure of alcohol-induced activation based on the slope of the line formed by a semi-log plot of activity after three low doses of ethanol: 1.0, 1.25 and 1.5 g/kg. All of the LSXSS RI show only locomotor activation at 1.5 g/kg; for many of the RI, 1.5 g/kg produces the peak level of activation (table 1). The use of the slope from the stimulatory phase of the ethanol dose-response curve makes for a more robust measure of ethanol-induced locomotor activation than a single-dose metric. This ethanol activation slope theoretically enabled us to identify only those genes involved in the activation phase of the ethanol dose-response curve. The activation slopes for most of the RI were linear (representative slopes are shown in fig. 1) with a mean for regression coefficients of 0.91 ± 0.03. Table 1 lists all the ethanol activation slope values for each RI strain as well as mean activities after saline, and mean difference scores for 1.0, 1.25, 1.5, 1.75 and 2.0 g/kg ethanol.

View larger version (19K): [in this window] [in a new window]   Fig. 1.   Ethanol dose-response curves for locomotor activity. Three representative RI responses are shown for total distance traveled after 1.0, 1.25, 1.5, 1.75 and 2.0 g/kg ethanol. Difference scores (ethanol score on day 2saline score on day 1) were plotted against the log ethanol dose. The dotted line is the ethanol activation slope formed by the 1.0, 1.25 and 1.5 g/kg data points.

It is important to determine if this ethanol activation slope measure produces a skewed or normal distribution when applied to the LSXSS RI, as the common methods of quantitative trait analysis assume a normal distribution of the phenotype in question. Figure 2 shows that the LSXSS RI form a normal distribution when using slope to measure ethanol activation (Kolmogorov-Smirnov goodness of fit, Z = 0.754, P = .62). Similarly, the activity of the RI after 2.0 g/kg ethanol was normally distributed (data not shown). Significant correlations were observed between activities after saline and after various doses of ethanol. For example, activation after 2.0 g/kg ethanol (difference score) correlates with activity after saline (r = 0.54, P = .007) indicating a significant saline activity component in this measure of ethanol-induced activation. However, the slope measure is not significantly correlated with activity after saline (table 2); it has effectively removed the stress component (activity after saline) from the locomotor activity measure. Thus, the ethanol activation slope appears to offer a more accurate measure of the pharmacologic stimulation by ethanol than using a single dose test.

View larger version (18K): [in this window] [in a new window]   Fig. 2.   Distribution of LSXSS RI for ethanol activation slopes. Slopes for each RI, determined as in figure 1, were averaged and the frequencies of mean values plotted.

If NT and NTR are involved in ethanol-related behaviors, we would expect significant genetic correlations between NT and NTR measures and expression of the low-dose ethanol activation phenotype. Table 3 shows the significant correlations between ethanol activation and NT-ir in the hypothalamus and nucleus acumbens and as well as those between activity after saline, ethanol activation slope, or activity after 2.0 g/kg ethanol and NTR densities in frontal cortex and striatum. There were 7 correlations of 27 with values of P < .05 and the cumulative Poisson distribution was used to determine the probability of this observation when only 1.35 (0.05 × 27) are expected by chance. The probability was calculated to be P = .0004 [p(6 or more) = 1-.99959]. Thus, it is highly likely that more than one of these genetic correlations indicate a common underlying mechanism regulating locomotor effects of ethanol and NT measures. Given these significant genetic correlations, it was reasonable to look for QTL in common between CNS NT measures and ethanol-induced locomotor activation. Table 4 lists the QTL we identified for the two measures of ethanol-induced activity, ethanol activation slope and activity after 2.0 g/kg ethanol. Only one of the QTL is in common between the two activity measures, linked to the D12Mit44 marker at 4cM on chromosome 12. One of the five QTL for 2.0 g/kg ethanol activity and one of the seven QTL for ethanol activation slope are similar to those identified by Phillips et al. (1995) using 2.0 g/kg ethanol in the BXD RI. They are located at 25 cM on chromosome 17 (2.0 g/kg activity) and 71 cM on chromosome 10 (ethanol activation slope).

Comparing the QTL for ethanol-induced locomotor activity to those in our companion paper for NT-ir levels and NTR_H and NTR_L densities (Erwin et al. 1997) gives six QTL in common for at least one of the activity measures and one of the NT measures. These are listed in table 5. QTL were determined to be in common if they mapped ± 15 cM of each other, 15 cM being the average density of our marker coverage of the LSXSS RI genomes. Five of the seven ethanol activation slope QTL have in common at least one NT measure QTL, although only two of the five QTL for activity after 2.0 g/kg ethanol are in common with NT measure QTL. These common QTL do not include the one mentioned above that is similar to a QTL published previously by Phillips et al. (1995) for activity after 2.0 g/kg ethanol in the BXD Ri.

As in the accompanying manuscript (Erwin et al., 1997), the question whether the number of common QTL between pairs of phenotypes (NT measure and ethanol activation) exceeds that expected by chance was addressed. The probability of one common QTL due to chance would be the product of the relative frequency of QTL for the two phenotypes. The frequency for ethanol activation (slope) is 6% (6 QTL of an estimated 100 unlinked markers) and the frequency for NTR_L is 7 or 11% for STR or VMB, respectively. The probability of markers at P < .05 for these NT measures being in common with significant markers for activation is 0.0042 (0.06 × 0.07) for STR and .0066 (0.06 × 0.11) for VMB. Assuming 100 independent markers, the expected number of false positives is 0.42 for STR and 0.66 for VMB. Thus, 0.42 or 0.66 markers would be expected to attain P < .05 for both ethanol activation (slope) and the respective NTR_L densities by chance. Similar calculations showed that 0.36 markers would be expected to attain P < .05 for both NT-ir levels in NA and ethanol activation (slope). The cumulative Poisson distribution was used to determine the probability of observing three common markers at P < .05 (QTL) between ethanol activation (slope) and STR or VMB NTR_L binding when only .42 or .62 are expected by chance; the P(3 or more) = .02 or .05, respectively. Similarly, the P(2 or more) = .05 for observing two common QTL between ethanol activation (slope) and NT-ir levels in the NA. These results show, by conservative estimates, that the number of common QTLs is significantly greater than that due to chance. Common QTL between other NT measures and ethanol activation are not significantly greater than those expected by chance at P < .05.

	Discussion

Top Abstract Introduction Methods Results Discussion References

It is well known that ethanol has both stimulatory and inhibitory effects on many different performance tasks in both laboratory animals and man (Pohorecky, 1977). If one is interested in only the stimulatory effects of ethanol on locomotor activity, experiments should be designed with care. A major factor in separating the two effects on locomotor activity is selection of ethanol dose. The proper dose to use for locomotor activation is a function of genotype. The LSXSS RI results demonstrate that the best measure of locomotor activation is given by the slope of the line formed by plotting difference scores of total distance traveled over 10 min (starting from 5 min postinjection) against the log of ethanol dose.

The slope measure of ethanol activation was obtained by determining locomotor activity after three low-doses of ethanol (1.0, 1.25 and 1.5 g/kg). Data for each dose in each RI strain was collected using animals naive to ethanol. Because the log ethanol dose-response curves were linear for all RI, the resulting slopes appear to be the best estimates of ethanol-induced activation unconfounded by ethanol-induced locomotor inhibition. This method is recommended when examining activation in inbred strains. However, when testing animals from a genetically heterogeneous population, such as an F₂ population or an outbred line, the repeated exposures to ethanol needed to obtain a dose-response and a slope measurement are not ideal. The correlation between ethanol activation slope and activity at each ethanol dose presented in table 2 was highest at 1.5 g/kg ethanol (r = 0.87). These results, taken together with those reported by others (Dudek et al., 1991), indicate that in mice that show activation, 1.5 g/kg ethanol is preferred in studies of ethanol stimulation where only one dose is possible.

In our studies, we also examined the locomotor effects of ethanol at 2.0 g/kg because a previous study has reported QTL for these effects at this dose in the BXD RI strains (Phillips et al., 1995). One of the goals of our study was to compare those QTL to the ones we identified in the LSXSS RI strains. As noted in the results, two QTL we found in the LSXSS RI for activity after 2.0 g/kg ethanol were similar to QTL reported for the BXD RI. Thus, our study confirms the following previously reported QTL for the effects of 2.0 g/kg ethanol on locomotor activity: chromosome 10, at 71 cM, and chromosome 17, at 25 cM. Differences in the other QTL might be the result of both type 1 and type 2 errors caused by a lack of statistical power in the limited RI strain numbers used for the two studies. Also, differences in fixed alleles in founder populations, i.e., LS and SS vs. C57BL/6 and DBA, could have contributed to differences in the QTL. The QTL we identified for the slope measure of locomotor activation will need to be confirmed in an independent experiment, such as in an F₂ population of mice derived from extreme responders in the LSXSS RI strains.

It is surprising that there was only one common QTL among the 11 total QTL for activity at 2.0 g/kg ethanol and the ethanol activation slope (table 4) because these measures were highly correlated (r = .73) (table 2). One factor contributing to this apparent discrepancy may be a difference in the underlying mechanisms involved in these two ethanol-induced behaviors. The ethanol activation slope appears to measure true activation and it has the highest correlation with activity induced by 1.5 g/kg (table 2). In the LS and SS mouse lines, 1.5 g/kg is also the cut-off dose for observing the effects of ethanol tolerance on locomotor activity. Chronic ethanol treatment does not affect locomotor activity at challenge doses up to and including 1.5 g/kg, but at doses beyond this, development of ethanol tolerance unmasks additional stimulatory effects of ethanol on locomotor activity (Erwin et al., 1992b). Because chronic tolerance develops only to the inhibitory effects of ethanol and not the stimulatory (Crabbe et al., 1982; Tabakoff and Kiianmaa, 1982; Masur et al., 1986), this further defines 1.5 g/kg ethanol as the dose demarking the end of mainly stimulatory effects and the beginning of increased inhibitory effects on locomotor activity. Thus, activity induced by 2.0 g/kg ethanol involves both the stimulatory and inhibitory influences of ethanol, making it a mixture of two different quantitative traits, and rendering the QTL identified by using this dose difficult to interpret. Further support for this comes from a comparison of QTL identified using the single dose data for ethanol-induced locomotor activity. Four of eight QTL are in common between 1.25 g/kg activity and locomotor activity after 1.5 g/kg, although none are in common between 1.25 and 2.0 g/kg (data not shown). This is again suggestive of different subsets of genes becoming involved as the ethanol dose increases and supports the contention that the stimulatory and inhibitory effects of ethanol are controlled by two different sets of genes (Crabbe et al., 1982; Dudek et al., 1991).

We found six QTL in common between low-dose ethanol activity and measures of CNS NTR or NT-ir (table 5). All but two of the QTL for ethanol activation slope had in common a QTL for at least one NT measure. Included in table 5 are some candidate genes that might provide mechanistic links between NTR density and ethanol activation. On chromosome 1, at 103 cM, are QTL for ethanol activation slope and NTR_L densities in the VMB. Mapped at 101 cM on chromosome 1 is the Akp1 gene encoding alkaline phosphatase-1 (Committee on the Mouse Genome, 1994). Phosphorylation states of many different proteins are known to be altered by alcohol (Deitrich et al., 1989) and phosphorylation is also known to influence gene expression (Karin, 1991; Ghosh et al., 1994; Kuiper and Brinkmann, 1994). A polymorphism in the Akp1 gene could affect expression of the NTR_L gene that then subsequently alters responsiveness to alcohol. For the same reasons, Pkca, which encodes the  subunit of protein kinase C and maps near a common QTL on chromosome 11 (Committee on the Mouse Genome, 1994), might affect both NTR_H densities and ethanol activation. Additionally, Pkca is an intriguing candidate gene because NTR_H is coupled to PI hydrolysis in brain tissues (Erwin and Radcliffe, 1993). Finally, on chromosome 12, in a linkage group from 4 to 17 cM, are QTL for ethanol activation slope, activity after 2.0 g/kg ethanol, and NTR_L densities in the frontal cortex and striatum. Located at 4 cM on chromosome 12 is also the Pomc1 gene, encoding POMC (Committee on the Mouse Genome, 1994). Adrenocorticotropin and -endorphin, derived from the POMC polypeptide, are known to be differentially regulated by ethanol in the LS and SS lines (Wand, 1989) and POMC peptides are known to stimulate locomotor activity (Spanagel et al., 1991; van Erp et al., 1991). Thus, a polymorphism in the POMC gene promoter could produce different levels of POMC peptides and alter locomotor responses to ethanol. It is less clear how the POMC gene might be involved in regulation of NTR_L gene expression, although nothing is yet known about the NTR_L gene or its regulation.

We previously have reported that four genes are involved in ethanol-induced locomotor activation based on activity data collected after 1.875 g/kg ethanol administration in the LSXSS RI (Erwin et al., 1990a). Recalculating the number of loci involved using the ethanol activation slope and 2.0 g/kg data provides estimates of 3.9 and 4.5 loci, respectively. These numbers are derived from the formula n = (R²)/8V_A where R² is the square of the difference of the extreme responses and V_A is estimated as the variance of the means for all the RI (Falconer, 1981). This calculation is based on the assumption of equal effect for all genes and would be an underestimate if this assumption is untrue. Thus, with the expectation that some of the loci listed in table 4 may represent type I errors, the six QTL identified by the slope measure and the five QTL identified using 2.0 g/kg ethanol are in reasonable agreement with our estimation of the number of loci involved in ethanol-induced locomotor activity.

Although QTL mapping is not a recent innovation, dating at least to 1923 (Sax, 1923), modern molecular biology techniques have enabled QTL analyses of low-fecundity species and greatly enhanced the method's utility. Consequently, QTL mapping is being used enthusiastically by scientists studying alcohol sensitivity in rodent models. To realize the method's full potential, however, some problems need solutions. Identification of a 10 cM region containing a locus of interest is now relatively simple, and increasing the map to a 1 to 2 cM resolution is becoming easier with the growing library of markers. However, even with a 1 cM resolution, isolation and cloning of an allelic polymorphism contributing a fraction of the variability in any given phenotype will be a daunting task. Bridging the chasm between genetic mapping and molecular cloning will require luck, years of perserverance or development of new methods of positional cloning. In the meantime, identification and confirmation of chromosomal regions linked to genetic differences in behaviors and correlated neurochemical phenotypes, and hunting for polymorphisms in candidate genes mapped within these regions, is a substantial first step in identifying genes mediating ethanol-related behaviors.