Influence of ethanol on thermoregulation: mapping quantitative trait loci

LARRY I. CRAWSHAW1,2, HELEN L. WALLACE1, ROBIN CHRISTENSEN1 and JOHN C. CRABBE2

1 Department of Biology, Portland State University, Portland, 97207
2 Department of Behavioral Neuroscience and Portland Alcohol Research Center, Oregon Health and Science University and Veterans Affairs Medical Center, Portland, Oregon 97201


    ABSTRACT
 TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
The genetic basis for the effects of ethanol on thermoregulation was investigated by utilizing recombinant inbred mouse strains from C57BL/6J and DBA/2J progenitor strains. Changes in core body temperature (Tc) and the degree of fluctuation of Tc were monitored in male mice following the administration of ethanol in an environment with cyclic changes in ambient temperature (Ta). Changes in Tc were utilized to assess ethanol-induced effects on regulated Tc, whereas fluctuations in Tc were utilized to assess thermoregulatory disruption. Ethanol was administered intraperitoneally at 1.5, 2.5, and 3.5 g/kg for all strains. Change in Tc and increase in tail temperature were also evaluated at 2.5 g/kg ethanol in a constant Ta of 26°C. Associations between the measured physiological responses and previously mapped genetic markers were used to identify quantitative trait loci (QTLs). This established probable chromosome locations for a number of genes for the responses. To our knowledge, this is the first report of QTLs that underlie changes in regulation as well as the disruption of a physiological regulatory system.

alcohol; temperature regulation; recombinant inbred strains; genetics


    INTRODUCTION
 TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
ETHANOL EXERTS SUBSTANTIAL effects on core body temperature (Tc), both by lowering the regulated temperature and by decreasing the accuracy of the thermoregulatory system. Effects on the regulated temperature can be demonstrated at low doses, and disruptive effects become greater as dose levels are increased (13, 14). Ethanol-induced alterations in Tc are important in both clinical and laboratory situations. Victims of fatal hyperthermia (often involving hot tubs) and hypothermia (typically outdoors) are often found to be intoxicated with ethanol (20, 27). In the laboratory, physiological systems such as glucose metabolism are often affected more by the ensuing change in Tc than by an initial ethanol treatment (23).

For mice and rats, individual differences in sensitivity to the thermoregulatory effects of ethanol are known to be partially controlled by genetics (20). The probability of identifying genes that underlie these different effects of ethanol has been greatly increased by a number of recently developed procedures. One approach is to "knock out" various genes and their products and assess each gene’s importance by measuring subsequent effects on the system of interest. A more global approach is to evaluate the likelihood of all genes’ involvement in a particular response. We used one such global approach, employing the statistical association of physiological responses with genetic markers to identify quantitative trait loci (QTLs).

Each physiological response that a population of animals exhibits typically reflects a varied contribution of many genes. In this study, such phenotypic traits include changes in Tc, variability in Tc, and changes in tail temperature (Tta) following ethanol administration. The continuous distributions of individual differences that underlie such responses confirm the influence of multiple genes, with each gene exerting a quantitative, rather than an all-or-none, effect on the trait. Sites on a chromosome that contain one or more genes that influence a measured phenotype are termed QTLs (10). One method for identifying and mapping QTLs involves the use of panels of recombinant inbred (RI) mouse strains. Because they are inbred, mice of the same sex from each of these strains are similar to monozygotic twins with identical alleles at each genetic locus.

A widely utilized RI panel (BXD) was derived from an F2 cross between C57BL/6J (B6) and DBA/2J (D2) mice. The development and use of this approach has been described (10). Briefly, inbred strains (in this case B6 and D2 mice) are crossed to form an F1 generation that is heterozygous for all genes that differ in the parents. Crossing F1 mice produces an F2 generation that is segregated genetically. Unique BXD RI strains were produced by inbreeding novel F2 individuals for many generations; each RI strain now has a fixed, different pattern of B6 and D2 alleles due to genetic recombination (3). Genetic markers at more than 1,500 loci signify the allelic status at these loci (B6 or D2) of each RI strain as well as the B6 and D2 progenitor strains.

The BXD RI panel is useful for locating any genes of interest that are polymorphic between the B6 and D2 progenitor strains. For the traits described in this study (Tc, etc.), polymorphism at some genes is likely since the two strains respond differentially to ethanol (14). B6 mice are more sensitive to the effects of ethanol on regulated temperature, whereas D2 mice are more affected by actions of ethanol that decrease the overall accuracy of the regulatory system. Differences between these two strains can thus be exploited by QTL methodology to identify genes most likely to be involved in mediating the different effects of ethanol on thermoregulation (9, 11).

To perform a QTL analysis using the BXD RI strains, each progenitor strain and RI strain is tested for the average degree of expression of phenotypic traits. The possession of specific alleles at genetic markers is then correlated with the mean strain phenotypic response in these 26 genotypes. Since each allele is known to derive from either a B6 or a D2 strain, if a particular marker allele is highly correlated with a phenotypic response characteristic, then it is inferred that a nearby (linked) gene affects the trait.

The purpose of the current study is to evaluate the genetic basis for the effects of ethanol on the thermoregulatory system using the above approach. A varying ambient temperature (Ta) regime was employed to allow simultaneous evaluation of the effects of ethanol on regulated body temperature (mean change in Tc) and on accuracy of temperature regulation (variation in Tc) after administration of 1.5, 2.5, and 3.5 g/kg ethanol. In addition, an index of a thermoregulatory effector response (Tta) was monitored after administration of 2.5 g/kg ethanol. Twenty-four RI strains were evaluated, and a QTL analysis was performed on the strain means.


    METHODS
 TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 

Animals.
All protocols used male mice from 24 of the BXD RI strains. Data displayed for the progenitor strains (B6 and D2) are from a previously published study (14), wherein a more detailed description of the apparatus and protocol is available. The RI strains were bred from stock originally obtained from the Jackson Laboratories (Bar Harbor, ME). These were held 1–4 animals per cage, at 27.5 ± 1°C, with a 12:12-h light/dark photoperiod (lights on at 0700). All experiments were run between 0830 and 1700. The home cages contained corncob bedding at least 3 cm deep; when mice were individually housed (due to aggression or experimental protocol), their cages also contained shredded paper towels. All caged animals had ad libitum access to water and food (Mazuri 5663 Rodent Pellets or Purina Rodent Laboratory Chow). The mice were between 20 and 40 g and 8 and 26 wk of age when tested.

Injection protocols.
Ethanol was administered intraperitoneally as a 10% vol/vol solution. Each mouse was tested in two blocks of injections. The first block involved the alternating temperature experiments and was organized in the following sequence: saline, 1.5 g ethanol/kg, 2.5 g ethanol/kg, 3.5 g ethanol/kg. The second block comprised the constant temperature experiments, which consisted of a saline injection and a 2.5 g ethanol/kg injection. The saline injections contained the diluent for ethanol (0.9% NaCl) and were administered in a volume equivalent to the 3.5 g ethanol/kg injections. All injection solutions were warmed to 38°C. Saline injections were treated as the final stage of habituation and as confirmation that the mouse responded in a normal way to the injection procedure per se. These data were not used in the comparisons across the strains tested. Injections of NaCl alone were followed by at least 2 days of rest. A minimum of 3 days of rest followed each ethanol injection.

Alternating temperature experiments.
Partially restrained mice were exposed to cyclic changes in Ta as depicted in Fig. 1 and described in detail in Crawshaw et al. (14). Briefly, changes in Ta were produced by airflows that emanated from two air sources, regulated at 46.5 ± 0.3 and 10.0 ± 0.2°C. Entrance and exhaust fans directed the air from either of the two sources into the chamber that held the mouse. A timer switched the fans for one source off, and for the other source on, every 6 min. Passive one-way valves prevented air leakage between the two sources. The usual thermal extremes in the mouse chamber, as recorded by a thermocouple just in front of the mouse, were 14 and 42°C. The rate of airflow in the vicinity of the mouse varied from 1–3 m/s. An important function of the restraint was to maintain a relatively fixed orientation of the animal with respect to the airflow.



View larger version (14K):
[in this window]
[in a new window]
 
Fig. 1. The effect of a 2.5 g/kg injection on Tc for individual mice of the recombinant inbred (RI) strains BXD-14 and BXD-29. Injection was at time 0 and is indicated by the arrows. The continuous and dashed lines were added to facilitate comparison of the responses by the two different strains. The horizontal continuous line depicts the preinjection core body temperature (Tc), whereas the dashed lines connect two successive lows of Tc.

 
The rationale for this regime was to provide the mouse with symmetrical increases and decreases in Ta that, for a control (undrugged) animal, would produce little net hot or cold stress. The point midway between 14 and 42°C is 28°C. This latter temperature approximates a relatively neutral Ta over the entire 12-min thermal cycle. Usual values for thermoneutrality at basal levels of heat production are several degrees higher, but a partially restrained animal is likely to produce heat at levels somewhat above basal. The control animals had little problem maintaining Tc within the normal range. In addition, Tc showed little variation and did not "track" Ta. Following ethanol injections, Tc usually fell and began to mirror the changes in Ta. The mean change in Tc is likely most reflective of changes in the regulated body temperature, whereas increased variation in Tc likely is more reflective of generalized disruption of the regulatory system.

Partial restraint was accomplished by fitting mice with a rubber-lined plastic collar that was affixed to a track and could thus slide forward or backward (but not sideways) for 10 cm. To minimize stress during the procedures, mice were habituated to experimental conditions as follows: days 1 and 2, 10–15 min of handling; days 3 and 4, alone in cage with collar on for 2 h; days 5 and 6, in partial restraint system as used in experiments for 2 h; days 7 and 8, mice partially restrained and placed in an alternating temperature habituation environment (extremes of 39 and 23°C with a period of 12 min) for 2 h.

To begin an experiment, the animal was first weighed and then placed in the restrainer. Next, an epoxy tipped 36-gauge copper-constantin thermocouple (Pirelli Cable) was inserted 2.5 cm past the rectum and taped to the tail. A strip chart recorder (Linear model 0595) was used to monitor Ta and Tc as calibrated voltages. The instrumented and restrained mouse was then placed inside the chamber, and video recording was initiated. The video recording was continued throughout the experiment. After 15 min, the air flow was initiated. At this time, if Tc was 37.5°C or above, the experiment was started with a cooling interval. If Tc was below 37.5°C, the experiment was started with a heating interval. Injections were always made just after a cooling interval; normally the baseline period was thus 3 or 3.5 full cycles (36 or 42 min) depending on whether the experiment was initiated with a warming or a cooling interval. If the baseline was not stable at this time, another cycle was run. If the baseline remained unstable after 4 (or 4.5) cycles, the experiment was terminated. Otherwise, the run was continued for 3 cycles (36 min) after the injection.

Data analysis is best understood by reference to Fig. 1. Initially, Tc was transcribed at the end of each warming and cooling interval. These numbers were termed end values. To assess changes in the regulated temperature, postinjection Tc was subtracted from preinjection Tc. Postinjection Tc was the mean of the end values 24 and 30 min after the injection. Preinjection Tc was the mean of the end values 6 min before the injection and just before the injection (slightly before time 0). Disruption of the regulated temperature was evaluated by assessing the extent to which Tc varied between successive warming and cooling cycles. This number was derived from the mean of the absolute differences between the postinjection end values at 12 and 18 min, 18 and 24 min, 24 and 30 min, and 30 and 36 min.

Constant temperature experiments.
After completion of the experiments in the alternating temperature chamber, the mice were tested in a Ta of 26 ± 0.1°C. This Ta was employed because under the conditions of this experiment, 26°C was just below the threshold for active vasodilation. As Ta was increased above 26°C, vasodilation became more common, and it was difficult to obtain a steady baseline. As Ta was lowered below 26°C, the Tta response declined or disappeared. The air flow rate adjacent to the mouse was 0.5–0.6 m/s. If more than a week elapsed after the completion of the alternating temperature experiments, then the habituation procedure was repeated once before the mice were run in the constant temperature experiments.

To initiate an experiment, a mouse was partially restrained and outfitted with a rectal thermocouple (as above); in addition, a thermocouple was secured to the dorsal surface 1.5 cm distal to the base of the tail and covered with one layer of Johnson & Johnson clear tape. After 25 min at 26°C, Tta was monitored for a steady baseline. After at least 20 min of Tta within 1°C of Ta, the mouse was injected with either saline or ethanol. Ta, Tta, and Tc were recorded for 40 min after the injection. Changes in Tta were quantified by integrating the difference from baseline (in degree·minute) over the 18-min period following the injection. The response thus arrived at was termed "Tta response." Changes in Tc were quantified by taking the difference between mean Tc for the 18-min postinjection period and baseline Tc.

These latter experiments utilized Tta to evaluate blood flow changes in superficial tail regions. The goal was to use a thermoregulatory effector response for QTL analysis.

Statistical analyses.
Unless noted otherwise, statistical analyses utilized a two-way ANOVA for repeated measures or Student’s t-test for the difference between means. ANOVA values were utilized to produce estimates of heritability (h2), which were calculated as SSB/SST where SSB is the sum of squares between subjects for the factor strain, and SST is the sum of squares total. This estimate of h2 provides the proportion of the individual differences that is due to genetic influences. All results given in the text and in Figs. 2 and 3 are for the mean value ± one standard error of the mean (SE). The number of individuals tested from each strain was 14–16 with the following exceptions: B6, n = 24; RI lines 1 and 25, n = 13; RI line 31, n = 17; RI line 8, n = 8. BXD line 8 (BXD-8) mice do not reproduce well. They have low fertility and small litters. Consequently, few animals were available for this study. Coupled with this was their apparent sensitivity to some feature of the procedure which led to the loss of several BXD-8 mice during the early phases of testing. Nevertheless, eight individuals completed the entire series of tests without incident, and the BXD-8 line was included in the QTL analysis. Genetic correlations among strain means were estimated using the Pearson product moment correlation (19). Unless otherwise mentioned, differences were taken to be significant at an {alpha} level of P < 0.05.



View larger version (55K):
[in this window]
[in a new window]
 
Fig. 2. Various measures and doses of the study, for all BXD RI strains tested, as well as the C57BL/6J (B6) and DBA/2J (D2) progenitor strains. AF: responses in the alternating temperature experiments. A, C, and E: decrease in Tc, for increasing doses. B, D, and F: fluctuation of Tc, for increasing doses. The strains on the x-axis for C and D are ordered from the largest to the smallest response. The strains on the x-axis for A and E follow the same order as C, whereas those for B and F follow the same order as D. Values are means ± SE.

 


View larger version (21K):
[in this window]
[in a new window]
 
Fig. 3. Constant temperature experiments at 26°C of all BXD RI strains tested, as well as the B6 and D2 progenitor strains. Top: changes in Tc. Bottom: changes in the tail response. Values are means ± SE.

 
QTL analyses.
Details of the rationale for our approach and details involved in the QTL analysis have been described (4, 5, 9). Briefly, due to inbreeding, the BXD RI mouse strains are homozygous at all chromosomal locations for either B6 or D2 alleles. The BXD RI strains have been genotyped for genetic markers, and the chromosomal location is known for each of these markers (29). In the Portland Alcohol Research Center MAP MANAGER database, >1,500 markers were available when this analysis was run. By convention, D2 alleles were assigned a value of 1, and B6 alleles were assigned a value of 0. To identify possible QTLs, the strain mean of each RI strain for a given trait (such as change in Tc) was correlated with each marker’s value (either 1 for a D2 allele or 0 for a B6 allele). Correlation coefficients (r) and two-tailed P values for all significant associations are given in Table 1. A high correlation between the strain means of a given trait and a particular marker suggested that a gene somewhere nearby on the chromosome was influencing that trait.


View this table:
[in this window]
[in a new window]
 
Table 1. Significant and provisional QTLs affecting lowered core temperature, regulatory disruption, and tail vasodilation

 
Because of the large number of correlations calculated, the possibility of type I errors (false positive) was relatively large. We addressed this problem by reporting only clusters of markers in a particular chromosomal region that often show similar patterns of correlation. When this occurs, the probability of a QTL in that region is greatly increased. We employ a two-step procedure, in which we first provisionally identify and subsequently verify the existence of QTLs (5). The initial stage is presented here. Verification (i.e., elimination of type I errors) is achieved by testing additional populations of mice (B6D2F2 animals derived from an intercross of B6D2F1 mice, and short-term selected lines). In the second step, analysis of the genome is restricted to regions "nominated" as potentially of importance by the BXD RI associations. The probabilities of association estimated in all populations are then combined. At any stage, definitive mapping of a QTL requires a high level of statistical association. The usual accepted level for significant linkage in an experiment considering only RI strains is P < 0.00002 (22). This level of stringency assures that no value declared significant during the genome-wide search will be a false positive. Based on simulations, we have argued that the assumptions underlying this proposed standard are unreasonably high, and we have proposed that a better estimate of protection against type I error in the BXD RI strains would require a P < 0.0001 correlation in the RI data (5). Lander and Kruglyak (22) propose a P level of 0.0007 to demonstrate "suggestive" linkage, whereas we employ P < 0.002 (5). A suggestive QTL linkage assures that no more than one such genome-wide association will subsequently prove to be a type I error. In this paper we utilized the suggestive level as our primary statistical criterion.

We report all provisional associations in Table 1, defining provisional to include those correlations significant at P < 0.01. Of the associations we have reported in other work achieving P < 0.01 in the RIs only, 50% have subsequently been found to represent false-positive associations when further populations were studied (8, 10). We present the larger group of provisional associations, because it is difficult to know in advance which will prove to be true and which will be false-positive associations. We believe it is important to reveal such provisional associations, even if some prove eventually to be false positive, because this protects us against a second type of error (type II error), which Lander and Kruglyak did not consider (5, 22). That is, if we prematurely reject a potentially important association because it does not achieve the very high level of significance demanded by Lander and Kruglyak, then we may miss an important gene when we turn to additional studies in other populations.

To evaluate the possibility of important covariates within the RI strain battery, the mean responses of the RI lines obtained in this study were correlated with the mean body weight of each RI line at the time of testing. The mean responses were also correlated with the rate of ethanol elimination for each strain (Grisel JE, Metten P, Wenger CD, Merrill CM, and Crabbe JC, unpublished observations).


    RESULTS
 TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 

Physiological findings.
Typical changes in Tc elicited by 2.5 g/kg ethanol during the course of an alternating temperature experiment are shown in Fig. 1 for two individuals of the BXD RI strains. Although Tc fell substantially below the preinjection baseline for BXD-14, Tc increased for BXD-29. On the other hand, fluctuations in Tc following injection were greater for BXD-29 than for BXD-14. For all the trials run on these two strains, the mean change in Tc following an ethanol injection of 2.5 g/kg was -1.30 ± 0.15°C for BXD-14 and +0.13 ± 0.13°C BXD-29. Corresponding values for fluctuations in Tc were 0.43 ± 0.05°C for BXD-14 and 0.46 ± 0.05°C for BXD-29. Responses like those illustrated in Fig. 1 were averaged across all mice within each RI strain to obtain a measure of changes that ethanol injection induced in both the mean Tc and in the excursion of Tc during thermal fluctuations in Ta. These latter derived measures are depicted in Fig. 2, wherein the means of each RI strain are shown for each of the measures. Changes in Tc at the three doses utilized are depicted in divisions A, C, and E of Fig. 2. Divisions B, D, and F depict the thermal fluctuation (range) seen following injection of the three ethanol doses. Figure 3 depicts, for the various strains, responses of Tc and Tta at a constant Ta of 26°C for a single dose of ethanol (2.5 g/kg).

For change in Tc during alternating Ta, all doses produced a statistically significant difference across the strains tested [1.5 g/kg, F(25,368) = 2.62, P <= 0.001; 2.5 g/kg, F(25,363) = 4.96, P <= 0.001; and 3.5 g/kg, F(25,357) = 4.18, P <= 0.001]. For these same doses, respectively, genotype (h2) accounted for 15%, 26%, and 25% of the total variance.

For changes in the excursion, or range, of Tc following ethanol injections, only the two highest doses were statistically significant [1.5 g/kg, F(25, 369) = 1.40, P <= 0.1; 2.5 g/kg, F(25,363) = 2.11, P <= 0.002; and 3.5 g/kg, F(25,358) = 2.72, P <= 0.001]. Genotype accounted for 9%, 13%, and 16% of the total variance for these range values.

For the mice tested at a constant Ta of 26°C after 2.5 g/kg ethanol, neither the changes in Tc [F(25, 340) = 1.33, P <= 0.14], nor the changes in Tta [F(25,340) = 1.19, P <= 0.24] were statistically significant across strains. Genotype accounted for 9% and 8% of the total variance for these two measures.

Because of low heritability, QTLs located for changes in the range of Tc at 1.5 g/kg, and the two measures of the constant temperature experiments would be more likely to reflect type I errors. Therefore, QTLs for changes in the range at 1.5 g/kg and changes in Tta are not reported. Because of relatively strong correlations among the strain means of the decreases in Tc in the alternating temperature experiments and the constant temperature experiments (see below), QTLs located for decreases in Tc at a constant Ta were included as an alternate means of validating the QTLs involving thermal variables in this and other studies.

Live video observations of the mice during experiments and later reviews of the video recordings gave a good indication of the animals’ responses to the apparatus. There was invariably an initial brief struggle, usually followed by acquiescence. The intensity and duration of the initial struggle was relatively similar across BXD lines. If an animal struggled continuously or violently, then the experiment was terminated.

Correlations among RI strain means.
The data collected in this study made it possible to estimate genetic correlations by correlating RI strain means among measures. These correlations were performed for the alternating temperature experiments and are displayed in Table 2. Correlations with P <= 0.15 only occurred between closely related measures. Strain means for change in Tc were significantly correlated for both pairs of adjacent doses. Changes in Tc after 1.5 and 3.5 g/kg ethanol were moderately correlated. A similar pattern of correlations pertained between strain means for changes in the range of Tc. Finally, the decreases in Tc at a constant Ta (only 2.5 g/kg was used) were significantly correlated with the decreases in Tc with the changing Ta for all doses given.


View this table:
[in this window]
[in a new window]
 
Table 2. Genetic correlations (among strain means)

 
Ethanol-induced changes in range of Tc were not significantly correlated with decreases in Tc under either alternating or constant Ta. This lack of correlation suggests that it is appropriate to utilize increase in range and decrease in Tc as separate measures of the effects of ethanol.

Correlations between RI strain means and other factors.
All correlations between strain means for weight and thermal responses to ethanol were low and ranged from -0.29 (P <= 0.15) to +0.17 (P <= 0.42). Correlations between strain means for ethanol elimination rates and thermal responses were also low, ranging from -0.38 (P <= 0.06) to 0.16 (P <= 0.45). Thus, of the 24 correlations (the 8 measures of this study and body weight or ethanol elimination at 2.0 and 3.0 g/kg), none was statistically significant.

QTL analysis for BXD RI strains.
Table 1 lists correlations indicating suggestive and provisional QTLs for the various measures of the study. In each case, only the genetic marker in each region showing the highest correlation with each trait is shown. In the alternating temperature experiments, for change in Tc, there were 14 suggestive and provisional QTLs: 6 at 1.5 g/kg, 5 at 2.5 g/kg, and 3 at 3.5 g/kg ethanol. For the range of Tc, there were 7 such QTLs: 3 at 2.5 g/kg and 4 at 3.5 g/kg ethanol. For the experiments run at a constant Ta, there were 6 such QTLs for change in Tc. Eight QTLs achieved suggestive levels of linkage, and these are identified by an asterisk in Table 1. Figure 4 illustrates locations on schematic mouse chromosomes for all suggestive and provisionally mapped QTLs and also includes provisional QTLs from other studies investigating the hypothermic effect of ethanol.



View larger version (43K):
[in this window]
[in a new window]
 
Fig. 4. Representation of quantitative trait loci (QTLs) from this and other studies on a set of schematic mouse chromosomes. Chromosomes 10 did not contain QTLs and therefore is absent. The dark dots at the top of the chromosome scales represent the telomeres, and the hash marks on the scales represent the distance in centimorgans from the telomeres. The whole numbers ("1 2 3 4") along the top represent the dose of ethanol that was utilized to obtain the QTL on each vertical line. Doses utilized in other studies and in this study were either whole numbered (1.0, 2.0, 3.0 and 4.0 or 4.2 g/kg) or whole numbers plus 0.5 (1.5, 2.5, 3.5 g/kg). The former were placed on the dose lines, while the latter were placed between the appropriate lines. For the QTLs from this study, the large symbols represent suggestive associations (P <= 0.01), whereas the small symbols represent provisional associations (P <= 0.05). For the data of this study depicted with solid symbols, {blacksquare} represent QTLs for a decrease in Tc, • represent QTLs for an increased variability of Tc, and {blacklozenge} represent QTLs for a change in Tc at a constant temperature (°C). For other studies involving hypothermia, depicted with unfilled symbols, {triangleup} represent QTLs from the study by Crabbe et al. (9), and {triangledown} represent QTLs from the study of Erwin et al. (15).

 

    DISCUSSION
 TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
The main purpose of this study was to evaluate the genetic components of physiological alterations that affect Tc following the administration of ethanol. There are two major means by which ethanol alters Tc. The first involves changes in the regulated level of body temperature. Ethanol decreases the regulated temperature, which triggers heat loss mechanisms such as vasodilation and the seeking out of a cooler microenvironment. The ensuing decrease in Tc appears adaptive; lower body temperatures substantially decrease the toxic effects of ethanol (13). The second involves the disruption of nervous function that ethanol causes. Disruption of the thermoregulatory system causes Tc to increase when the Ta is above thermoneutrality and decrease when the Ta is below thermoneutrality. The alternating Ta experiments allowed us to address the genetic basis for both of these phenotypes, which are differentially affected by ethanol in B6 and D2 mice (14).

The phenotype of regulated temperature is represented by the mean change in Tc across several cycles of the Ta. The mean Ta over a full cycle is very close to the thermoneutral Ta for an awake, alert mouse and without appropriate activation of thermoregulatory effectors, little change in Tc is expected. However, with coordinated activation of heat loss mechanisms, the thermoregulatory system can cause a rapid decrease in Tc.

The phenotype of thermoregulatory disruption is represented by the extent to which Tc is increased and decreased during the warming and cooling portions of the Ta cycle. Greater disruption of the thermoregulatory system causes Tc to increase more rapidly when Ta is above thermoneutrality and to decrease more rapidly when Ta is below thermoneutrality. The experiments at a constant Ta were designed to measure the effects of ethanol on the phenotype represented by a prominent thermoregulatory effector system, i.e., the alteration of thermal conductivity by vasomotor changes.

The phenotypes described above are all complex and could well be confounded with other variables. Ethanol metabolism, body weight, energy reserves, arousal state, stress, motor activity, and postural adjustments all influence Tc. Although none of these influences can be discounted with certainty, available evidence indicates that, with the exception of energy reserves, such interactions were minimal.

In a study on ethanol pharmacokinetic differences among RI strains, mice were given intraperitoneal ethanol injections of 2.0 or 3.0 g/kg ethanol and had blood levels of ethanol monitored periodically over the next several hours. From the rates of metabolism in each strain, we estimated the pseudo-zero-order elimination rates. We then subjected the elimination constant, as well as the blood level at 60 min after injection, for each dose to a QTL analysis following the methods described in this paper. In this analysis, two significant and several suggestive QTLs were found. However, when the locations of the metabolism-related QTLs were compared with the locations of the four QTLs with strongest associations in the current study, there was no overlap. As an additional test, the strain mean values for ethanol elimination rates were correlated with the values obtained in this study for the BXD RI strains. The range of correlations between elimination rates after 2 or 3 g/kg ethanol and the seven temperature regulation variables was from -0.38 to +0.16, and all correlations were nonsignificant. We conclude that ethanol metabolism did not play a role in the thermoregulatory responses of the QTLs we report.

Body weight of the RI lines showed very small correlations with any of the responses of this study. Thus it is unlikely that factors directly related to body weight affected the results of this study. However, selection experiments indicated that important energetic aspects can vary independently of body weight. Thus, although mice selected for high and low heat loss did not differ in body weight, the high heat loss line showed an increased expression of ribosomal protein L3 in both the hypothalamus and in the brown adipose tissue (2).

Somewhat more than a dozen QTLs have been identified relating to body fat content. These QTLs are described or mentioned in Refs 21, 25, and 30. Of these, three overlap with significant or provisional sites of the current study. These QTLs will be mentioned as the sites are discussed.

The remaining potential confounds were not quantified and could only be subjectively assessed by viewing the video tape recordings. Motor activity was minimal and similar in all lines, once the initial struggle was over. Arousal level was also similar in the various lines. The mice remained alert throughout the experiments except for the expected tranquilizing and anesthetizing effects of the ethanol. Likewise, posture and location in the apparatus was similar. Although the level of stress likely differed to some degree across the RI lines, there were no overt indications that stress was a major factor in the analysis.

For all responses, there was separation between the B6 and D2 progenitor strains within the BXD RI strain distribution. The position of these progenitor strains in the distributions varied in an interesting pattern. For decreases in Tc, the B6 mice always evidenced a greater decrease than the did the D2 mice, but the difference decreased at the higher doses. As illustrated in Fig. 2, the relationship between magnitude of response and strain order for the RI strains showed similarity across doses, although there was notable variation.

For the alternating temperature experiments, there was no overlap between the 14 associations for regulatory changes and 7 associations implicated in the disruption of regulation. These results indicate the genetic distinctiveness of these two effects. The associations also differed across dose. For decreases in the regulated Tc, twice as many associations were identified at the lowest dose compared with the highest dose of ethanol. For decreased thermoregulatory accuracy, there were four associations at the highest dose and three at the middle dose. These results are consistent with the concept that the genetic contribution to level of regulation is more affected by low doses of ethanol, whereas the disruption of regulation is more affected by higher doses.

Unlike locomotor activity, which is augmented by low doses of ethanol and decremented by high doses (15, 16), all responses monitored in this study depicted monotonic increases in magnitude as the dose of ethanol was increased. For both of the measures in the alternating temperature experiments, there were significant differences among the RI and progenitor strains which reflected an underlying genetic variation.

The elaborate genetic map that exists for the BXD RI battery of strains allowed us to detect a number of suggestive and provisional QTLs. An advantage of this approach is that these chromosome locations were identified without the necessity of prior hypotheses or specific genotyping. A further advantage of the current technique is that the results of this study are based on genetically stable strains and will be valid for comparison with past and future investigations. The ensuing discussion will focus on associations that attained the suggestive level as defined in the METHODS.

For decreases in Tc at a constant Ta of 26°C, we located six suggestive or provisional QTLs, of which two overlapped with such QTLs for decreases in the regulated Tc in the alternating temperature experiments. None of these six overlapped with the suggestive or provisional QTLs for decreased accuracy of thermoregulation. In the following discussion, unless otherwise mentioned, changes in Tc will refer to results from the alternating temperature experiments.

Of the genetic loci identified in this study, four sites appear particularly interesting. These sites are more likely to represent true QTL associations because, even with the limited number of genotypes tested (26 genotypes), these associations were well beyond our criterion for suggestive linkage. In our studies with other phenotypes, all regions originally identified in RI strains at this level of significance have subsequently been verified as true associations when other populations have been studied. Some of the four sites described above were also the locations of provisional QTLs for other thermoregulatory traits assessed in this study or from other thermoregulatory studies. Three of the sites were associated with a decrease in Tc in the alternating or constant temperature experiments, and these will be discussed first. Next, a site associated with disruption of thermoregulation will be addressed. Consideration of potentially relevant genes will be limited to genes mapping within 10 cM of the QTL under discussion.

A strong association was found on chromosome 2 at 69 cM, which involved a fall in Tc for all three doses of ethanol. In addition, markers in this region were also highly associated with a decrease in Tc at a constant Ta. This locus is also the site of a QTL influencing heat loss (25). Genes identified in this area code for interleukins-1{alpha} and -1ß, {alpha}2b-adrenergic receptors, and arginine vasopressin. These systems have all been implicated in different aspects of the control of body temperature. For example, the interleukin-1 system has hypnogenic properties and affects the regulated temperature, particularly during fever (28), the {alpha}-adrenergic system has been implicated in the maintenance of Tc in both thermoneutral and cool environments (17), and arginine vasopressin has been suggested as a possible mediator for the hypothermia that accompanies severe hypoxia (31).

A second highly significant association for decreased Tc was found in mid-chromosome 8 at 33 cM, for both the 1.5 and 2.5 g/kg doses. The 2.5 g/kg dose was found to have a provisional association at this locus. Potential candidate genes in this region include the neuropeptide Y receptor gene, Npy1r.

The strongest association in this study was found on chromosome 7 at 53 cM, for a decrease in Tc at a constant Ta. Provisional QTLs at this locus were also identified by Crabbe et al. (11) for hypothermia following doses of both 2 and 4 g/kg ethanol. This is also the locus of the tubby (tub) gene, which influences fatness in mice (see Ref. 21), as well as the locus of QTLs associated with heat loss (25) and the amount of body fat (30). Gene products coded for at this site include two subtypes of non-voltage-gated sodium channels.

The strongest associations between genetic markers and the disruption of thermoregulation by ethanol occurred on chromosome 11 at 59 cM. The association was found only for the lowest dose of ethanol (1.5 g/kg). This site is strongly associated with markers D11Mit14 and Tstap91a and is near genes coding for a voltage-gated sodium channel and a calcium channel.

Although no individual marker on chromosome 1 reached our suggestive criterion, the proximal portion had a series of strong provisional associations at ~20 cM. Decreases in Tc for all three doses of ethanol as well as decreases in Tc in the constant temperature experiments were all highly linked with this area. In this same area, Crabbe et al. (11) identified a provisional QTL for ethanol-induced hypothermia after a dose of 4.0 g/kg ethanol. Genes coding for interleukin receptors 1 and 2 have been identified in this area. As mentioned earlier, the interleukin-1 system is involved with both sleep and fever (28).

The hypothermic response to ethanol has been examined in two other QTL studies. In these studies, both a decrease in the regulated body temperature and disruption of the thermoregulatory system presumably contributed to the degree of hypothermia observed. Crabbe et al. (11) employed female B6, D2, and BXD RI strains of mice and administered doses of 1, 2, 3, and 4 g/kg ethanol. Erwin et al. (15) utilized male LS, SS, and LSXSS RI strains after 4.2 g/kg ethanol. Both other studies used protocols that involved manually recording Tc before and after ethanol injection in mice maintained at a Ta of 22–23°C. For mice, this is a condition of moderate cold stress.

We evaluated the commonality of the suggestive and provisional QTLs identified in these other studies by comparing them with those identified in this study. QTLs were considered common if they were derived from reasonably similar doses of ethanol and were within 10 cM, which is the approximate accuracy range of mapping with these RI strains. There was a correspondence on the order of 30% between the suggestive and provisional QTLs of the other studies and those of this study. This low level of correspondence appears reasonable, given the relative specificity of the effects of particular genes (see Ref. 1) and the fact that changes in Tc can be produced a number of ways, many of which are not directly involved in the thermoregulatory system (see Ref. 12). Such effects include alterations in arousal state, emotional responses, motor activity, and postural adjustments. These responses were probably elicited to different degrees by the different protocols and thus implicated different genes in the resultant alteration of Tc. In addition, the genetic makeup of the mice utilized by Erwin et al. (15) differed from that of the mice in this study. Although the LS and SS mice were created from a foundation stock that included DBA/2 and C57BL/6 inbred strains, the foundation stock also involved intercrosses of six other inbred mouse strains (24). Finally, sex differences may have played a role.

The provisional identification of QTLs in this study is only the initial step in localizing the set of genes that are involved in mediating the effects of ethanol upon thermoregulatory processes. One example of the steps that will produce a better resolution can be found in Buck et al. (8). Further confirmation or rejection of these QTLs could come from the characterization of F2 animals (the offspring of B6 and D2 intercrosses) or the study of phenotypically selected individuals from the F2. In these populations, phenotyping and the subsequent genotyping of individuals can provide increased evidence of association.

The confidence intervals surrounding these provisional QTLs are large, and each region contains hundreds of genes, any one (or group) of which could be responsible for the association. Therefore, discussions of potential candidate genes would be highly speculative, but as confidence intervals are reduced, many fewer candidate genes will remain.

Several steps would still be necessary to achieve the goal of cloning a gene responsible for a QTL. One method that is currently in progress involves creation of congenic mice, created from B6 and D2 genotypes. By backcrossing, mice are being created with a background genotype of 99% D2 or B6, with a small region of chromosome derived from the other progenitor strain containing the verified QTL. This region will then be made progressively smaller through backcrossing, genotyping, and selection of recombinant individuals, thus further delimiting the area containing the QTL (10).

In conclusion, we have utilized different response measures to locate potential genetic loci responsible for two separate effects of ethanol on the thermoregulatory system. Alterations of the regulated temperature and disruptive actions on physiological regulation are represented in different portions of the mouse genome. Further work is needed to clarify which of these potential QTLs are valid contributors to the effects of ethanol on this physiological system.

Perspectives
The QTL approach is one of many ways to further understanding of physiological function. This particular approach links the responses of a whole system with the genes that code for component parts exerting the greatest influence on measured variables. Because of the high degree of synteny between the mouse and the human genome, genes implicated in mice will often lead directly to corresponding genes in humans. Although the path from QTL to gene is not currently direct, data that produce QTL will remain available for a future time when more complete genetic maps will allow a more accurate QTL localization and the procedures for identifying genes from these discrete QTL are standardized.


    ACKNOWLEDGMENTS
 
A large number of hardworking, intrepid undergraduate students provided invaluable help in the completion of this study. Marc Neisenfeld and Leroy Lausch provided valuable fabrication and electronics support.

This research was supported by National Institute of Alcohol Abuse and Alcoholism Grant AA-10760, by a grant from the Department of Veterans Affairs, and by support from Portland State University.


    FOOTNOTES
 
Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).

Address for reprint requests and other correspondence: L. I. Crawshaw, Dept. of Biology, Portland State Univ., PO Box 751, Portland, OR 97207 (E-mail: crawshl{at}pdx.edu).

10.1152/physiolgenomics.00041.2001.


    REFERENCES
 TOP
 ABSTRACT
 INTRODUCTION
 METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 

  1. Alcock JC. Animal Behavior: An Evolutionary Approach. Sunderland, MA: Sinauer, 1998.
  2. Allan MF, Nielsen MK, and Pomp D. Gene expression in hypothalamus and brown adipose tissue of mice divergently selected for heat loss. Physiol Genomics 3: 149–156, 2000.[Abstract/Free Full Text]
  3. Bailey DW. Recombinant-inbred strains: an aid to identify linkage and function of histocompatibility and other genes. Transplantation 11: 325–327, 1971.[ISI][Medline]
  4. Belknap JK, Metten P, Helms ML, O’Toole LA, Angeli-Gade S, Crabbe JC, and Phillips TJ. Quantitative trait loci (QTL) applications to substances of abuse: physical dependence studies with nitrous oxide and ethanol in BXD mice. Behav Genet 23: 213–222, 1993.[ISI][Medline]
  5. Belknap JK, Mitchell SR, O’Toole LA, Helms ML, and Crabbe JC. Type I and type II error rates for quantitative trait loci (QTL) mapping studies using recombinant mouse strains. Behav Genet 26: 149–160, 1996.[ISI][Medline]
  6. Belknap JK, Richards SP, O’Toole LA, Helms ML, and Phillips TJ. Short-term selective breeding as a tool for QTL mapping: ethanol preference drinking in mice. Behav Genet 27: 55–66, 1997.[ISI][Medline]
  7. Browman KE and Crabbe JC. Alcohol and genetics: new animal models. Mol Med Today 5: 310–318, 1999.[ISI][Medline]
  8. Buck KJ, Metten P, Belknap JK, and Crabbe JC. Quantitative trait loci involved in genetic predisposition to acute alcohol withdrawal in mice. J Neurosci 17: 3946–3955.
  9. Crabbe JC, Belknap JK, Mitchell SR, and Crawshaw LI. Quantitative trait loci mapping of genes that influence the sensitivity and tolerance to ethanol-induced hypothermia in BXD recombinant mice. J Pharmacol Exp Ther 269: 184–192, 1994.[Abstract]
  10. Crabbe JC, Phillips TJ, Buck KJ, Cunningham CL, and Belknap JK. Identifying genes for alcohol and drug sensitivity: Recent progress and future directions. Trends Neurosci 22: 173–179, 1999.[ISI][Medline]
  11. Crabbe JC, Phillips TJ, Gallaher EJ, Crawshaw LI, and Mitchell SR. Common genetic determinants of the ataxic and hypothermic effects of ethanol in BXD/Ty recombinant inbred mice: genetic correlations and quantitative trait loci. J Pharmacol Exp Ther 277: 624–632, 1996.[Abstract]
  12. Crawshaw LI, Rausch RN, and Wallace HL. Thermoregulation. In: Wilderness Medicine (4th ed.), edited by Auerbach PS. St. Louis: Mosby, 2001, p. 112–128.
  13. Crawshaw LI, Wallace HL, and Crabbe JC. Ethanol, body temperature and thermoregulation. Clin Exp Pharmacol Physiol 25: 150–154, 1998.[ISI][Medline]
  14. Crawshaw LI, Wallace HL, Crabbe JC, Ramos C, Duerr J, and O’Connor CS. Ethanol disrupts and decreases the regulated body temperature differentially in C57BL/6J and DBA/2J mice. Am J Physiol Regulatory Integrative Comp Physiol 273: R696–R702, 1997.[Abstract/Free Full Text]
  15. Erwin VG, Markel PD, Johnson TE, Gehle VM, and Jones BC. Common quantitative trait loci for alcohol-related behaviors and central nervous system neurotensin measures: hypnotic and hypothermic effects. J Pharmacol Exp Ther 280: 911–918, 1997.[Abstract/Free Full Text]
  16. Erwin VG, Radcliffe RA, Gehle VM, and Jones BC. Common quantitative trait loci for alcohol-related behaviors and central nervous system neurotensin measures: locomotor activation. J Pharmacol Exp Ther 280: 919–926, 1997.[Abstract/Free Full Text]
  17. Frank SM, Raja SN, Wu PK, and El-Gamal N. {alpha}-Adrenergic mechanisms of thermoregulation in humans. Ann NY Acad Sci 83: 101–110, 1997.
  18. Hegmann J and Possidente B. Estimating genetic correlations from inbred strains. Behav Genet 11: 103–114, 1981.[ISI][Medline]
  19. Kalant H and Lê AD. Effects of ethanol on thermoregulation. Pharmacol Ther 23: 313–364, 1984.[ISI]
  20. Keightley PD, Morris KH, Ishikawa A, Falconer VM, and Oliver F. Test of candidate gene: quantitative trait locus association applied to fatness in mice. Heredity 81: 630–637, 1998.[ISI][Medline]
  21. Lander ES and Kruglyak L. Genetic dissection of complex traits: guidelines for interpreting and reporting linkage results. Nat Genet 11: 241–247, 1995.[ISI][Medline]
  22. Masur J and Souza MLO. Does hypothermia play a relevant role in the glycemic alterations induced by ethanol? Pharmacol Biochem Behav 16: 903–908, 1982.[ISI][Medline]
  23. McClearn GE and Kakihana R. Selective breeding for ethanol sensitivity: Short-sleep and long-sleep mice. In: Development of Animal Models as Pharmacogenetic Tools (NIAAA Research Monograph 6), edited by McClearn GE, Detrich RA, and Erwin VG. Washington, DC: NIAAA, 1981, p. 147–169.
  24. Moody DE, Pomp D, Nielsen MK, and Van Vleck LD. Identification of quantitative trait loci influencing traits related to energy balance in selection and inbred lines of mice. Genetics 152: 699–711, 1999.[Abstract/Free Full Text]
  25. Moore JA and Kakihana R. Ethanol-induced hypothermia in mice: influence of genotype on development of tolerance. Life Sci 23: 2331–2338, 1978.[ISI][Medline]
  26. Paton BC. Accidental hypothermia. Pharmacol Ther 22: 331–377, 1983.[ISI][Medline]
  27. Rechtschaffen A and Siegel J. Sleep and dreaming. In: Principles of Neuroscience (4th ed.), edited by Kandel ER, Schwartz JH, and Jessell TM. New York: McGraw-Hill, 2000, p. 936–947.
  28. Silver LM. Mouse Genetics: Concepts and Applications. New York: Oxford Univ., 1995.
  29. Warden CH, Fisler JS, Shoemaker SM, Wen PZ, Svenson KL, Pace MJ, and Lusis AJ. Identification of four chromosomal loci determining obesity in a multifactorial mouse model. J Clin Invest 95: 1545–1552, 1995.[ISI][Medline]
  30. Wood SC. Interactions between hypoxia and hypothermia. Annu Rev Physiol 53: 71–83, 1991.[ISI][Medline]