J Appl Physiol 95: 1338–1351, 2003. First published April 18, 2003; 10.1152/japplphysiol.00132.2003.
mouse phenome project
Genotypic differences in ethanol sensitivity in two tests of motor incoordination John C. Crabbe,1 Pamela Metten,1 Chia-Hua Yu,1 Jason P. Schlumbohm,1 Andy J. Cameron,1 and Douglas Wahlsten2 1
Portland Alcohol Research Center, Department of Behavioral Neuroscience, Oregon Health & Science University, and Veterans Affairs Medical Center, Portland, Oregon 97239; and 2Department of Psychology and Centre for Neuroscience, University of Alberta, Edmonton, Alberta, Canada T6G 2E9 Submitted 7 February 2003; accepted in final form 16 April 2003
Crabbe, John C., Pamela Metten, Chia-Hua Yu, Jason P. Schlumbohm, Andy J. Cameron, and Douglas Wahlsten. Genotypic differences in ethanol sensitivity in two tests of motor incoordination. J Appl Physiol 95: 1338–1351, 2003. First published April 18, 2003; 10.1152/ japplphysiol.00132.2003.—Motor incoordination is frequently used as a behavioral index of intoxication by drugs that depress the central nervous system. Two tasks that have been used to assay incoordination in mice, the balance beam and the grid test, were evaluated to optimize aspects of apparatus and testing procedures for studying genetic differences. Mice of eight inbred strains were given one of several doses of ethanol or saline and tested for intoxication. Strains differed in sensitivity to ethanol in both tests, indicating a significant influence of genotype on ethanol sensitivity. For the balance beam, the width of the beam affected the strain sensitivity pattern, and only the widest beam worked well at all doses. For the grid test, both ethanol dose and the time after drug injection affected strains differentially. Although the behavioral sign of intoxication recorded for both tests was a foot-slip error, the correlations of strain means for ethanol sensitivity across the two tasks were generally not significant. This suggests that the genes influencing ethanol sensitivity in the two tasks are mostly different. These results make clear that a single set of task parameters is insufficient to characterize genetic influences on behavior. Several other issues affect the interpretation of data using these tests. inbred mouse strains; pharmacogenetics; ataxia; grid test; balance beam
of mice have been developed through systematic interbreeding of close family members, usually brother with sister, for many generations. As genetic diversity in an inbred strain is gradually eliminated, the result is a collection of genetically identical mice, where at each gene, two copies of the same allele are fixed. Because individual differences among animals within a strain are therefore nongenetic in origin by definition, when differences among strains, compared under carefully controlled environmental conditions, significantly exceed those within strains, clear evidence is obtained of the influence of
MORE THAN 100 INBRED STRAINS
Address for reprint requests and other correspondence: J. C. Crabbe, VA Medical Center (R&D 12), 3710 SW US Veterans Hospital Rd., Portland, OR 97239 (E-mail:
[email protected]). 1338
genes on behavior. In such studies, differences among strains in behavior are usually detected. There are many laboratory measures of motor coordination applied to studies of neurological disorders and drug-intoxicated states. These different measures are usually developed heuristically and generally do not target a specific neural pathway or behavioral function. Thus it is reasonable to assume that each such measure may reflect demand on a unique set of neurobiological systems, including those subserving attention, balance, vision, proprioception, locomotion, and patterned gait. Ability to perform in many assays shows evidence of genetic influence. For example, in a task frequently called the “balance beam,” mice are required to balance on or to traverse a narrow beam. Mice with null mutations or transgenic alterations in a number of genes show impaired performance on this task. How the task is configured and used, however, varies markedly from laboratory to laboratory. Three examples are given as illustration. Mice bearing a combination of presenilin-1 and amyloid precursor protein (APP) mutations fell from a suspended beam of unspecified width more rapidly than wild-type mice; the test was characterized as one assessing “balance/ agility” (1). In four experiments, Kiernan et al. (25) employed two widths of square beams and four widths of round rods and measured latency to traverse and foot-slip errors in mice with a null mutation of tenascin-C. They reported relative deficits in the null mutants, which were more pronounced for the easier conditions and which were absent when the mutation was tested in mice partially congenic on a 129 strain background. They characterized this task as assessing motor coordination, balance, and vestibular integrity. Finally, the balance beam reported here has been used to assess stress-induced ataxia in a mouse model of episodic ataxia type-1 (23). Assays of incoordination have also proven useful in studies of drug intoxication. For example, we have found that strains differ in sensitivity to ethanol intoxication in both the fixed speed and accelerating ver-
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked ‘‘advertisement’’ in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. http://www.jap.org
GENETICS OF ETHANOL INTOXICATION
sions of the commonly used rotarod task (37). Furthermore, the patterns of genetic influence depend on how the tasks are conducted. Under some conditions, many of the same genes appear to influence ethanol sensitivity in both rotarod tasks, but, under other test conditions, groups of genes are invoked that are unique to each task (37). Specific aspects of apparatus and testing parameters are not often considered in studies of different genotypes, but they can make a large difference in results. Rotarod parameters (e.g., diameter, rate of rotation) affect the efficacy of the task to detect ethanol’s effects and influence the pattern of strainspecific sensitivity to ethanol (37, 38). To enlarge the pool of assays available to assess genetic influence on motor incoordination in mice, it is desirable to begin with explorations of the range of normal mouse behavior in such tasks. In our studies, we have first used genetically heterogeneous (outbred) mice to characterize some of the apparatus and procedural variables important for coordination tasks. Having first established a range of suitable task parameters, we then explored the response of several standard inbred strains of mice, representing a diversity of genetic backgrounds. Strains were selected based on recommendations of the Mouse Phenome Project (34), a consortium effort to provide basic behavioral and physiological data on a variety of mouse genotypes. We believe that, under some conditions, inappropriate choice of apparatus and test parameters can invalidate attempts to characterize strain differences (12, 43). We report here systematic studies of two tasks: the balance beam and the grid test. We have previously used the balance beam to study the sensitivity to ethanol in mice with a null mutation of the serotonin 1B receptor subtype gene and their background genotype under a single test condition (4). The grid test was developed and originally used to show differences in sensitivity to ethanol between two inbred strains: C57BL/6 and DBA/2 (2, 3). A modified version of this task was subsequently used to show differences in ethanol sensitivity among multiple inbred strains (7), between genetically selected lines (39), among recombinant inbred strains derived from crosses between C57BL/6 and DBA/2 (11), and between serotonin 1B receptor null mutants and wild types (4). Those studies, however, were limited to a single test paradigm and often to a single dose of ethanol. We report here the results of a survey of eight inbred strains after multiple ethanol doses for both these apparatuses. METHODS
Animals and Husbandry Male and female mice from eight inbred strains were obtained from The Jackson Laboratory (Bar Harbor, ME) at 7 wk of age. Although the different strains had experienced different treatments before shipment (e.g., C57BL/6J mice were fed LabDiet 5K52, whereas DBA/2J mice were fed LabDiet 5K54; for other strains, see http://jaxmice.jax.org/ jaxmice-cgi/jaxmicedb.cgi), mice were treated as identically as possible after arrival in Portland, OR. All subsequent procedures surrounding husbandry and handling were apJ Appl Physiol • VOL
1339
proved by the Institutional Animal Care and Use Committee at the Portland Veterans Affairs Medical Center and were in accordance with National Institutes of Health guidelines. On arrival, mice were transferred to plastic shoebox cages lined with Bed-o-Cob bedding, usually three mice per cage. Food (Purina 5001 brand rodent chow) and water were provided ad libitum. The lights in the colony room were on for 12 h (0600–1800), and the ambient temperature was 21 ⫾ 1°C. For balance beam testing, 356 mice of the following eight inbred strains were used: 129S1/SvImJ (129: n ⫽ 42), A/J (A: n ⫽ 40), BALB/cByJ (BALB: n ⫽ 48), BTBR T⫹tf/tf (BTBR: n ⫽ 43), C3H/HeJ (C3H: n ⫽ 42), C57BL/6J (B6: n ⫽ 46), DBA/2J (D2: n ⫽ 47), and FVB/NJ (FVB: n ⫽ 48). Four shipments were received, each comprising ⬃96 mice (⬃6 mice 䡠 sex⫺1 䡠 strain⫺1). The age of the subjects ranged from 58 to 112 days (mean ⫾ SE age ⫽ 82 ⫾ 1 days) at the time of testing. Bedding was changed weekly after testing. Strains were chosen from the group A list of the Mouse Phenome Database (http://aretha.jax.org/pub-cgi/phenome/ mpdcgi?rtn⫽docs/pristrains), which itself was selected because these strains are widely used, fertile, and relatively inexpensive and offer diverse pedigrees (34). The exception was the BTBR strain, which was once, briefly, on the group A list but is studied here because we have data on them for several other relevant tasks. We also studied B6D2F1 mice, which are the hybrid cross of B6 and D2 inbreds. Although B6D2F1 individuals are identical at each gene, they are heterozygotes for all genes where B6 and D2 mice possess different alleles. Therefore, although B6D2F1 data were collected because they might be of interest for studies of the genetic architecture of any strain differences between the popular B6 and D2 strains, they were neither analyzed nor presented but are available from the authors on request. Balance Beam The balance beams were built to custom specifications by Flair Plastics (Portland, OR). Four beams were constructed of gray PVC, at widths of 9.5, 12.7, 15.8, and 19.0 mm. All beams were 104.1 cm in length and mounted on supports that provided a vertical distance of 54.5 cm from the top of the beam to the table below. The beam was textured to provide slight traction. Pads were placed under the beam to provide cushioning if an animal fell. Within each shipment, the cages were randomized for testing order. Subjects were given one of three doses of ethanol (1.0, 1.2, or 1.4 g/kg) or saline. Doses were randomized such that within a home cage no two mice had the same dose. Ethanol (Pharmco Products, Brookfield, CT) was mixed 20% vol/vol in physiological saline (Baxter Healthcare, Deerfield, IL). On the day before data collection, subjects were trained on the 12.7-mm balance beam beginning at 0900. Each mouse was coaxed across the beam in both directions twice, ear punched, and returned to its home cage. If a mouse would not traverse the beam unaided, a nudge to the haunches or a gentle twist to the tail was used to induce movement. A cube of rodent chow held in front of the subject’s nose provided additional motivation for some mice. After all mice were trained, they were returned to the colony room. Twenty-four hours later, on the test day, the mice were again moved to the test room and habituated for 1 h. Subjects were injected with saline or ethanol (20% vol/vol in saline) intraperitoneally (ip) at 3-min intervals by one technician. After injection, each mouse was housed separately in a holding cage lined with cob bedding for 10 min to allow ethanol to reach maximum levels in the brain. A second technician,
95 • OCTOBER 2003 •
www.jap.org
1340
GENETICS OF ETHANOL INTOXICATION
unaware of strain, drug, or dose condition, then removed the mouse from the holding cage and commenced testing. Previous experience with this apparatus and test had shown that many mice would refuse to run during a second test; thus we elected not to use each mouse as its own (saline) control. Each mouse was run across three balance beams starting with the narrowest beam to the widest. Each trial on a beam was separated by an intertrial interval of 30 s. For the first pass, the three beams used in testing were the 9.5-, 12.7-, and 15.8-mm beams. However, the 9.5-mm beam proved too difficult for many mice, and much of the resulting data was uninterpretable (see RESULTS and DISCUSSION). Hence, subjects were tested serially on the 12.7-, 15.8-, and the 19.0-mm beam for the subsequent three passes of the experiment. Serial testing was adopted because the sample sizes would have been too small to detect a meaningful order effect in a counterbalanced design. Although data on the 15.8- and 19.0-mm beams could then also reflect learning during the test, the data on the 12.7-mm beam were taken with all mice naive to ethanol. We chose the 12.7-mm beam for the first test because our previous data had used this width. Hindfoot missteps and latency to cross the beam were recorded, and any aberrant behavior was noted. As soon as the mouse reached the opposite end of the beam, it was immediately returned to the holding cage. If a mouse fell in transit, it was picked up by the tail and placed back on the beam at approximately the same location from where it fell. In cases in which the mouse hung from the beam by its forepaws and could not right itself, the experimenter righted the mouse by picking it up by the tail and swinging its back legs back onto the beam. Each instance of hanging was recorded as a misstep as well as noted as an aberrant behavior. When a mouse turned around, the experimenter reoriented it by lifting its rear half by the tail and swinging it in the right direction. As during training, the mouse received a gentle pinch to the haunches or twist of the tail if it paused longer than 1 s. Because animals sometimes stop while traversing the balance beam, measures of latency to cross were deemed hard to interpret and are not reported here. After traversing all three beams, the subject was returned to its home cage where it remained until all mice were tested. Between testing of each mouse, all three balance beams were wiped down with a paper towel moistened with a 10% isopropyl alcohol solution mixed with tap water. At the end of testing for the day, the mice were all transported to the colony room. They remained under the conditions outlined above for ⬃1 wk before they were run on the grid test protocol. During this period, they were housed with their original cage mates in the same home cage.
mate for 60 min. On days 1, 2, and 3, the mice were habituated to the apparatus by placing them into the grid test immediately after an intraperitoneal saline injection. On the test day, the mice were given an ethanol injection of 1.5, 2.0, or 2.5 g/kg ip immediately before placement in the apparatus. The mice were returned to their home cages and colony after removal from the grid test on all four days. Eight to fourteen days after the initial test, the groups of mice given 2.0 g/kg were retested by using the same procedures. Only the test day was repeated. Blood Ethanol Concentrations Blood ethanol samples were not taken during these experiments. However, in other studies that used these strains, tail blood samples (20 l) were taken 35 min after injection of 1.5, 2.0, or 2.5 g/kg ip ethanol (7a) and 15 min after injection of 1.0 g/kg ip (unpublished observations). These values were analyzed by gas chromatography as previously reported (17). Statistics The statistical strategy for these experiments was somewhat unusual. Our general goal was to describe the inbred strains’ behavior and sensitivity to ethanol. In both tasks, because strains markedly differ, a particular set of test parameters (e.g., a specific dose of ethanol) might incapacitate some strains while barely affecting others. For the balance beam, some strains were only capable of managing the widest beam tested. Thus we were faced with empty cells in ANOVAs, and, to present meaningful data, we were forced to use multiple tests, sometimes eliminate strains or doses, and so forth. Complete data for all mice are, however, available from the authors on request. Our secondary goal was to identify a subset of conditions for each test (e.g., beam width, dose of ethanol) for which reasonable effect size due to strains and reliability could be demonstrated for the maximum number of strains (see RESULTS). An ␣-level of 0.05 was set for statistical significance. For most analyses, factorial ANOVA was used, where feasible (see RESULTS). Significant main effects or interactions were tested post hoc with Tukey’s honestly significant difference statistic following one-way ANOVA. Strain Effect Sizes and Measures of Genetic Commonality Inbred strain data from one-way ANOVAs on strain were used to estimate strain effect sizes (the proportion of individual differences attributable to genetic sources) for specific conditions (41). The proportion of individual differences in two traits attributable to a common set of genes was estimated from the correlation of inbred strain means (22).
Grid Test
Reliabilities
The ataxic effects of ethanol administration were then measured in the grid test apparatus, a 15 ⫻ 15 ⫻ 20-cm clear acrylic box placed in the center of a DigiScan activity monitor (AccuScan, Columbus, OH). The acrylic box had a 1.25-cm2 wire grid floor positioned 1 cm above a metal plate floor. When the mouse slipped on the wire mesh floor and stepped on the metal plate below, a circuit was completed and an error was recorded by custom software on a connected computer. Locomotor activity was recorded by the DigiScan apparatus by counting photo-beam interruptions. Locomotion data were collected in 5-min epochs, whereas foot-slip error data were collected in 1-min epochs, during the entire 15-min test. Before each day’s testing, the mice were moved from the colony room into the procedure room and allowed to accli-
We estimated reliability of mean strain values for selected behavioral indexes with the use of multiple methods. For the grid test, the groups given 2.0 g/kg ethanol were tested twice, so the correlation of strain means on the two tests was used to estimate test-retest reliability. Also for the grid test, because data were binned in 5-min epochs, we compared data from the second and third successive temporal epochs. For the balance beam, because mice were tested successively on the three beams, we used similar reasoning to obtain estimates by comparing strain values taken during the successive tests on the 15.8- and 19.0-mm-wide balance beams. However, it should be noted that neither of these latter methods was a true test-retest situation. For all the above, we also estimated the reliability of phenotypic scores by using individual data from all mice, ignoring strain.
J Appl Physiol • VOL
95 • OCTOBER 2003 •
www.jap.org
1341
GENETICS OF ETHANOL INTOXICATION RESULTS
Balance Beam Because each mouse crossed the same distance, the index reported is total foot-slip errors (missteps). Preliminary studies with genetically heterogeneous mice showed that the test was sensitive to lower doses of ethanol than other tasks such as the rotarod or grid test: doses higher than 1.5 g/kg rendered most mice incapable of performing the task, if not unwilling or unable to move (data not shown). Based on these findings, we selected the lower dose range for ethanol tests. Finally, other pilot studies performed immediately after injection or 30 min after injection produced data that were highly variable (data not shown), leading us to focus on the reported widths, doses, and time. As soon as testing with the inbred strains began, it became clear that some mice engaged in a variety of behaviors other than simply running across the beam. Particularly at the higher doses of ethanol, and particularly for some strains, two of these behaviors seemed potentially to confound interpretation of foot-slip errors as a direct index of intoxication. Many mice dragged one or both hind feet during a significant part of the test rather than placing their hind feet on top of the bar. When a foot was dragging, repeated errors
could not be counted, but it was clear that these mice were quite intoxicated. This behavior was most pronounced on the 12.7-mm beam after 1.4 g/kg ethanol (45% of mice) and decreased with increasing beam width (24 and 10% on 15.8- and 19.0-mm beams, respectively). It was very pronounced in strains 129, A, and C3H. Although individual 129, A, and C3H mice produced interpretable data that scored within the range of the other strains, there were so few of them that it was unreasonable to assume that they were representative of the strain’s capacity. Thus we did not feel that these three strains could be grouped with the other five strains meaningfully for four of the nine test conditions (see Table 1). BTBR mice were also quite affected by the 1.4 g/kg dose on the 12.7-mm beam but were able to avoid dragging their feet on the wider beams. A second unexpected behavior, seen less frequently, was when a mouse would stop, turn sideways on the bar, and its hindlimbs would slip from the bar (Table 1). It would hang by its front feet or even fall. In these cases, we replaced the mouse on the bar in the same location and allowed it to finish its trial. Hanging/ falling was also dose and strain dependent but was most pronounced on the 12.7-mm bar (Table 1). However, foot-slip data from these mice did not appear to
Table 1. Proportion of abnormal behaviors on the balance beam Hindlimb Dragging
Hanging by Forelimbs or Falling
Strain
EtOH Dose, g/kg
12.7 mm
15.8 mm
19.0 mm
12.7 mm
15.8 mm
19.0 mm
129S1/SvImJ A/J BALB/cByJ BTBR T⫹tf/tf C3H/HeJ C57BL/6J DBA/2J FVB/NJ 129S1/SvImJ A/J BALB/cByJ BTBR T⫹tf/tf C3H/HeJ C57BL/6J DBA/2J FVB/NJ 129S1/SvImJ A/J BALB/cByJ BTBR T⫹tf/tf C3H/HeJ C57BL/6J DBA/2J FVB/NJ 129S1/SvImJ A/J BALB/cByJ BTBR T⫹tf/tf C3H/HeJ C57BL/6J DBA/2J FVB/NJ
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.2 1.4 1.4 1.4 1.4 1.4 1.4 1.4 1.4
(6) 100 (6) 0 (6) 0 (6) 0 (6) 0 (6) 0 (6) 0 (6) 0 (13) 54 (11) 55 (14) 0 (11) 18 (11) 45 (12) 8 (13) 15 (14) 7 (11) 73 (12) 67 (13) 0 (12) 25 (13) 54 (14) 7 (13) 8 (14) 7 (12) 75 (11) 82 (15) 20 (14) 57 (12) 75 (14) 29 (15) 27 (14) 14
(6) 0 (6) 0 (6) 0 (6) 0 (6) 0 (6) 0 (6) 0 (6) 0 (13) 31 (11) 27 (14) 7 (11) 0 (11) 36 (12) 0 (13) 0 (14) 0 (11) 36 (12) 25 (13) 0 (12) 0 (13) 38 (14) 0 (13) 0 (14) 0 (12) 58 (11) 45 (15) 20 (14) 7 (12) 42 (14) 14 (15) 20 (14) 0
(6) 0 (6) 0 (6) 0 (6) 0 (6) 0 (6) 0 (6) 0 (6) 0 (9) 11 (7) 14 (10) 0 (8) 0 (7) 0 (8) 0 (9) 0 (10) 0 (7) 14 (8) 13 (9) 0 (8) 0 (9) 0 (10) 0 (10) 0 (10) 0 (8) 25 (7) 29 (11) 0 (11) 9 (8) 13 (10) 10 (11) 9 (10) 0
(6) 0 (6) 0 (6) 0 (6) 0 (6) 0 (6) 0 (6) 0 (6) 0 (13) 0 (11) 18 (14) 14 (11) 9 (11) 18 (12) 0 (13) 8 (14) 29 (11) 0 (12) 33 (13) 23 (12) 42 (13) 31 (14) 0 (13) 23 (14) 29 (12) 0 (11) 64 (15) 73 (14) 50 (12) 42 (14) 7 (15) 27 (14) 29
(6) 0 (6) 0 (6) 0 (6) 0 (6) 17 (6) 0 (6) 17 (6) 0 (13) 8 (11) 45 (14) 29 (11) 55 (11) 55 (12) 17 (13) 8 (14) 21 (11) 9 (12) 67 (13) 46 (12) 50 (13) 62 (14) 21 (13) 31 (14) 43 (12) 8 (11) 73 (15) 60 (14) 50 (12) 75 (14) 21 (15) 53 (14) 57
(6) 0 (6) 17 (6) 0 (6) 17 (6) 17 (6) 0 (6) 33 (6) 0 (9) 0 (7) 29 (10) 10 (8) 25 (7) 29 (8) 13 (9) 11 (10) 30 (7) 0 (8) 63 (11) 9 (8) 63 (9) 56 (10) 10 (10) 0 (10) 30 (8) 13 (7) 43 (11) 27 (11) 73 (8) 75 (10) 50 (11) 9 (10) 50
Numbers in parentheses are number of mice tested for the given strain, dose, and beam width (12.7, 15.8, and 19.0 mm). Proportions are percentage of mice exhibiting hindlimb dragging or hanging or falling. EtOH, ethanol. J Appl Physiol • VOL
95 • OCTOBER 2003 •
www.jap.org
1342
GENETICS OF ETHANOL INTOXICATION
differ from those of the same strain, dose, sex, and bar width condition who did not turn and hang or fall; therefore, we report the hanging/falling incidences but used the foot-slip data from these mice. We first analyzed the data obtained after saline injections to see whether strains differed in the absence of ethanol treatment. We were able to include all strains for both the 15.8- and 19.0-mm beams. Strains differed significantly (P ⬍ 0.05), but neither beam width nor the interaction was significant (data not shown). The 129 mice averaged 1.8 missteps, whereas the means for the other strains ranged from 0.3 (FVB) to 1.3 (C3H). We next examined the three beams separately. Strains (excluding the 129s) differed significantly for the 12.7-mm beam (P ⬍ 0.05): A and C3H mice averaged 3.2 missteps, whereas all other strains averaged ⬃1 misstep. Strains did not differ significantly for either of the two wider beams. For the analysis of ethanol effects, we elected to report corrected missteps for each strain, as done for other ethanol responses in the past (8). Corrected missteps were calculated as each ethanol-treated animal’s misstep score minus the strain mean saline score for that beam width, a method that allowed us to obtain estimates of strain variability. Next, we statistically analyzed corrected missteps from the four to eight strains producing meaningful data (the number depended on dose and beam width, but we show all data in Fig. 1). We first analyzed the levels of dose (1 and 1.2 g/kg) and beam width (15.8 and 19.0 mm) for which there were interpretable data for all strains. Our goal was to identify factors across levels of which the analyses could be collapsed to simplify further analysis and presentation. We saw significant effects of strain, beam width, dose, and significant interactions of strain ⫻ beam width, dose ⫻ sex, and beam ⫻ strain ⫻ sex (in descending order of effect size; all P ⬍ 0.05). Males made more errors than females after 1.2 g/kg, and mice had fewer missteps on wider beams and at the lower dose. Strain and beam width were clearly the most influential variables in the analysis (P ⬍ 0.01; see Fig. 1), so we next collapsed on sex and analyzed the strain, dose, and beam effects. Strain, beam width, dose, and the strain ⫻ beam width interaction were significant (all P ⬍ 0.05). In an analysis of three other responses to ethanolinduced incoordination, we had found that males were generally more intoxicated than females (7a). Because males are heavier than females at a given age, we asked whether the sex differences might simply reflect differences in body weight. In the dowel test, removal of the contribution of body weight by regression eliminated the sex differences. In contrast, for the screen test, the sex difference could not be entirely accounted for by the effect of body weight (7a). In our present experiment, we entered body weight into a linear regression for the data taken after 1.2 g/kg ethanol, separately for the 15.8- and the 19-mm beams. For the 15.8-mm beam, the effect of sex was no longer significant, indicating that all the sex differences could be accounted for by the greater weight of the male mice. J Appl Physiol • VOL
For the 19-mm beam, the effect of sex only tended to be significant (P ⫽ 0.06), and body weight was not a significant factor, unless the 129 strain was eliminated. Without the 129 strain, body weight also accounted for the apparent effect of sex. We therefore decided to report as our principal statistical analysis a series of one-way ANOVA of strain differences for each beam width and dose, collapsed on sex. This allowed us to estimate strain effect sizes as an aid to selecting the optimal test conditions for future studies. Significant effects of strain were found for the 19.0-mm beam width at all three doses (P ⬍ 0.01, 0.0001, and 0.05, respectively). A trend was observed for strain differences after the 1.2 g/kg dose on the 15.8-mm beam (P ⬍ 0.10). Strain was also significant after 1.4 g/kg on the 12.7-mm beam, but this was based on analysis of only four strains. Strain differences on all other dose ⫻ beam width combinations were not significant (P ⬎ 0.13). The most unusual responder of the five “normal” strains was D2. D2 mice were very sensitive to the combination of the highest dose and the narrowest beam width; however, on the widest beam, it showed the least ethanol sensitivity overall. When particular beam widths and doses were examined post hoc for significant strain differences, the 129 strain was found to make significantly more missteps than most of the other strains on the 19.0-mm beam at all doses. Genetic effect sizes. Proportion of variance in the corrected misstep data attributable to strain under each dose and beam width condition is given in Table 2. Greatest genetic contributions were seen on the widest beam after the lower two doses, suggesting that a substantial proportion of individual differences was due to strain under these test conditions. Data for the two narrower beams indicated less genetic influence. However, for the narrowest beam, only four or five of the eight strains were capable of the task, reducing the power to detect strain differences. Reliabilities. Table 2 also shows the reliability estimates, based on the correlation of performance on the 15.8- and 19.0-mm beams. Reliability of strain means was best after the 1.2 g/kg dose (0.45, 8 strains), followed by the 1.0 (0.28, 8 strains) and the 1.4 (0.15, 5 strains) g/kg doses. For phenotypic reliabilities, values were similar (0.25 ⱕ r ⱕ 0.41, all P ⱕ 0.05). Best measure for the balance beam. After genetic effect sizes and reliability have been jointly considered, these data would suggest that the best index of strainspecific sensitivity to ethanol was found on the 19-mm beam after the 1.2 g/kg ethanol dose, although data from the 1 and 1.4 g/kg doses on this beam width were also reasonable. Grid Test Habituation days. We tested a total of 186 mice. Within a strain, mice were reassigned to drug-dose groups pseudorandomly. We first analyzed the habituation days (Table 3) to determine whether strains differed in activity and coordination before ethanol. Be-
95 • OCTOBER 2003 •
www.jap.org
1343
GENETICS OF ETHANOL INTOXICATION
Fig. 1. Mean ⫾ SE corrected missteps on the balance beam. Three widths are shown: 12.7 mm (A), 15.8 mm (B), and 19.0 mm (C). Numbers of mice providing usable data are shown above bars. Mice were tested serially on the 3 beams starting 10 min after intraperitoneal injection of 1 of 3 doses of ethanol (listed in key). Mouse strain abbreviations are as follows: 129, 129S1/SvImJ; A, A/J; BALB, BALB/ cByJ; BTBR, BTBR T⫹tf/tf; C3H, C3H/HeJ; B6, C57BL/6J; D2, DBA/2J; and FVB, FVB/ NJ.
J Appl Physiol • VOL
95 • OCTOBER 2003 •
www.jap.org
1344
GENETICS OF ETHANOL INTOXICATION
Table 2. Genetic effect sizes and reliabilities of corrected foot slip errors on the balance beam Genetic Effect Size EtOH Dose, g/kg
12.7-mm beam width
15.8-mm beam width
19.0-mm beam width
1.0 1.2 1.4
0.07 (5) 0.01 (5) 0.13 (4)
0.06 (8) 0.08 (8) 0.00 (5)
0.19 (8) 0.35 (8) 0.13 (8)
Reliabilities: Correlation Between 15.8-mm and 19-mm Beams EtOH Dose, g/kg
Genetic
Phenotypic
1.0 1.2 1.4
0.28 (8) 0.45 (8) 0.15 (5)
0.32† (62) 0.25* (61) 0.41† (58)
Numbers in parentheses ⫽ no. of strains or individual mice included in estimates. Genetic effect sizes (proportion of individual differences attributable to strains) ⱖ0.13 were based on significant ANOVAs (P ⬍ 0.05). In reliability results: * P ⬍ 0.05; † P ⬍ 0.01.
cause mice had not yet been given ethanol, we ignored the subgroups based on subsequent ethanol dose, collapsed on sex, and cumulated activity across all three epochs on each day. Strains differed significantly in distance traveled during the 15-min test (P ⬍ 0.0001). Activity declined by about one-half over the 3 days (P ⬍ 0.0001), and the strain differences interacted with day (P ⬍ 0.0001). Strain, day, and their interaction were also statistically significant for ataxia scores (for definition, see next section; all P ⬍ 0.001). Because all mice had been given saline injections, very few foot-slip errors were recorded, and ataxia scores were negligible. When we computed the correlations between strain mean ataxia on the last habituation day and the test day for each epoch and dose, we found only one of the nine correlations to be significant (P ⬍ 0.05), so we disregarded the habituation data for the analyses of postethanol ataxia on the test day. Ethanol-induced ataxia: preliminary analyses. Two measures were recorded for analysis: activity (distance traveled in the DigiScan apparatus in cm) and foot-slip errors (recorded separately with custom software). Because these variables are positively correlated and because strains differed significantly in activity during the habituation phase, the ratio errors per centimeter was used to index incoordination (henceforth, “ataxia”). This has been our custom in previous analyses using this task (4, 11, 39). The DigiScan apparatus also reports beam interruptions (from which distance traveled is calculated). The data were first inspected epoch by epoch before statistical analysis. Beam interruptions can be accumulated by an animal not engaged in locomotion. For example, an animal may be repeatedly rearing near a beam, its tail or a paw may occasionally or repeatedly interrupt a beam even if it has lost righting reflex after a high ethanol dose, and so forth (8). Occasionally, an animal accumulated “foot-slip errors” in the absence of any distance traveled, traveled some centimeters “distance” in 0 s, or 20 cm/s or more for 5 s or fewer. We eliminated some data like these J Appl Physiol • VOL
(from 35 mice) because we deemed them uninterpretable. Most cases (a total of 24) were from mice given the 2.5 g/kg dose of ethanol (9 from 2.0 g/kg and 2 from 1.5 g/kg). For example, one BTBR female displayed zero beam breaks and 0 cm traveled in the third 5-min epoch, yet 76 foot-slip errors were recorded. We think it likely that she had lost the ability to move and was lying with her foot through a grid, repeatedly making contact with the plate beneath. The 129 strain was most affected by this confound, and data were removed from all three dose groups for this strain. In these cases, data for the whole 5-min epoch were removed. Data were removed for one or more epochs (n removed/n tested, for the 1.5, 2.0, and 2.5 g/kg dose groups, respectively) as follows: 129 (2/7, 6/8, 6/8); A (0/6, 1/7, 6/8); BALB (0/8, 0/8, 1/8); BTBR (0/7, 0/8, 4/7); C3H (0/7, 0/7, 2/7); B6 (0/7, 2/7, 5/7); D2 (0/8, 0/8, 0/8); and FVB (0/8, 0/8, 0/8). Only data from the first epoch following the 2.0 g/kg dose were interpretable for six of eight 129 mice and were retained in the analysis; both subsequent epochs had too few mice to provide interpretable data for this strain. The A (6/8), BTBR (4/7), and B6 (5/7) strains were also sensitive to the 2.5 g/kg dose during some epochs. All of the affected BTBR mice were females; for other strains, both males and females were affected. Finally, foot-slip data for three C3H and three B6 mice (1 mouse 䡠 strain⫺1 䡠 dose⫺1) were not collected due to apparatus malfunction, but their activity data were retained. We then performed a set of initial analyses on the ataxia scores after ethanol was administered to determine how best to assess and characterize any strain differences. Initial analysis included the repeatedmeasure factor, time epoch, but this eliminated the 129 strain and the 2.5 g/kg dose. This revealed that the Table 3. Activity and ataxia during habituation in the grid test Activity, cm Strain
Day 1
Day 2
Day 3
A/J BALB/cByJ BTBR T⫹tf/tf C3H/HeJ C57BL/6J DBA/2J FVB/NJ 129S1/SvImJ
430.2 ⫾ 35.1 1,338.1 ⫾ 70.1 1,289.0 ⫾ 101.7 608.5 ⫾ 36.1 1,067.4 ⫾ 76.4 1,009.3 ⫾ 72.5 1,353.5 ⫾ 83.2 798.2 ⫾ 60.5
310.9 ⫾ 29.1 1,023.8 ⫾ 99.4 933.4 ⫾ 87.4 483.2 ⫾ 35.3 757.5 ⫾ 62.8 688.6 ⫾ 66.6 925.3 ⫾ 77.0 305.1 ⫾ 41.4
221.8 ⫾ 18.3 887.8 ⫾ 67.3 513.2 ⫾ 59.4 477.1 ⫾ 42.6 559.9 ⫾ 42.2 635.0 ⫾ 48.5 899.3 ⫾ 74.3 164.2 ⫾ 21.2
Ataxia, errors/cm Strain
A/J BALB/cByJ BTBR T⫹tf/tf C3H/HeJ C57BL/67 DBA/2J FVB/NJ 129S1/SvImJ
Day 1
Day 2
Day 3
0.009 ⫾ 0.002 0.007 ⫾ 0.001 0.022 ⫾ 0.004 0.010 ⫾ 0.001 0.011 ⫾ 0.001 0.008 ⫾ 0.001 0.023 ⫾ 0.003 0.008 ⫾ 0.002
0.009 ⫾ 0.003 0.006 ⫾ 0.001 0.012 ⫾ 0.002 0.007 ⫾ 0.001 0.008 ⫾ 0.002 0.005 ⫾ 0.001 0.025 ⫾ 0.003 0.007 ⫾ 0.002
0.006 ⫾ 0.001 0.005 ⫾ 0.002 0.007 ⫾ 0.002 0.010 ⫾ 0.004 0.005 ⫾ 0.001 0.006 ⫾ 0.001 0.016 ⫾ 0.003 0.004 ⫾ 0.002
Values are means ⫾ SE distance traveled or foot-slip errors/cm traveled, cumulated over the 15-min test.
95 • OCTOBER 2003 •
www.jap.org
GENETICS OF ETHANOL INTOXICATION
main effects of strain and ethanol dose were highly significant (P ⬍ 0.0001) but that epoch was also important (P ⬍ 0.005). Significant interactions of epoch with strain, dose, and strain ⫻ sex were also present (P ⬍ 0.05). A second preliminary analysis focused on the 1.5 g/kg dose, allowing the 129 strain back into the analysis. This revealed significant main effects of strain and epoch (P ⬍ 0.05). The two- and three-way interactions of epoch with strain and strain ⫻ sex were also significant (P ⬍ 0.05). A third preliminary analysis examined the five strains for which valid data were obtained across all epochs and doses but did not include the factor of sex. This analysis revealed a significant main effect of ethanol dose and an interaction of epoch and dose (P ⬍ 0.0001). Given this pattern of results and the lack of a complete factorial design, we elected to report separate analyses for each dose and epoch, with the 129 data eliminated. As with the balance beam data, this strategy also allowed us to estimate strain effect magnitudes for the specific test conditions as an aid to identifying the condition most sensitive to strain differences. Ethanol-induced activity: preliminary analyses. On the test day, ethanol generally stimulated activity relative to the third habituation day (compare Figs. 2, A–C, Table 3). Main effects of strain and epoch, as well as their interaction, were significant (all P ⬍ 0.001), and a significant interaction of epoch with sex was found (P ⬍ 0.05). Females had significantly greater activity than males in the first epoch (P ⬍ 0.05) but not subsequently. Because our primary interest was in the ataxia data, we used the same analytic scheme for reporting, i.e., separate analyses for each dose and epoch. As noted, for the 129 strain, we eliminated data for the 2.0 g/kg dose after the first epoch and all data for the 2.5 g/kg dose. Ethanol-induced activity and ataxia. After the 1.5 g/kg ethanol dose, strains differed in distance traveled in all three epochs (all P ⬍ 0.0001), and activity generally declined across epochs. The most active strains after 1.5 g/kg ethanol were FVB and D2, and the least active strains were A and 129. The same strain-specific pattern of activity across epochs was seen after the 2.0 g/kg dose (P ⬍ 0.0001, 0.0001, and 0.05, respectively). During the latter two epochs, the B6 strain was one of the least active strains, consistent with their known lack of locomotor stimulation by ethanol (10). After the 2.5 g/kg dose, strains differed significantly during epochs 1 and 3 (P ⬍ 0.01) only. For ataxia scores (Fig. 2, D–F), strains also differed significantly after 1.5 g/kg ethanol in all three epochs (P ⬍ 0.05, 0.0001, 0.01, respectively). Mice of the A strain were most sensitive to intoxication by ethanol, and BTBR and FVB mice were the least sensitive. Ataxia scores appeared to decline across epochs. At the 2.0 g/kg dose, although the pattern was generally similar, strains only differed significantly during the second and third epochs (both P ⬍ 0.01). Ataxia scores tended to be sustained across all three epochs for most strains at this dose. Strains did not differ after 2.5 g/kg during any epoch (but data from only 5 strains were J Appl Physiol • VOL
1345
analyzed). Inclusion of the 129 strain in the analyses conducted for the 1.5 g/kg dose did not markedly affect the pattern of results but reduced significance levels. The 129 strain was generally the second most responsive strain (after A). Genetic effect sizes. The proportion of individual differences due to strain is shown in Table 4. For ataxia, strain effects were greatest for the 1.5 g/kg dose during the second epoch of testing and for the 2.0 g/kg dose during the second and third epochs. Strain effects for distance traveled were nearly all significant and were uniformly higher for the two lower doses. Reliability. Table 4 also shows the reliabilities for ataxia and distance traveled. For ataxia, the reliabilities of strain mean scores across epochs 2 and 3 were significant for the two lower doses (r ⱖ 0.90, P ⬍ 0.01) but was only 0.43 for the 2.5 g/kg dose. The strain mean test-retest reliabilities for the mice tested twice after 2.0 g/kg were substantial for all three epochs (0.61 ⱕ r ⱕ 0.78). For activity, genetic reliabilities between epochs 2 and 3 were also high for the 1.5 and 2.0 g/kg doses and a bit lower for the 2.5 g/kg dose. Test-retest reliabilities for activity after 2.0 g/kg were very high for all three epochs. The individual phenotypic reliabilities for these measures were also substantial, and, due to the larger number of observations, all were significant. Interestingly, test-retest phenotypic reliabilities for ataxia and activity in the 2.0 g/kg dose group were generally lower than the genotypic reliabilities. Blood ethanol concentrations. Blood ethanol concentrations for the inbred strains after various ethanol injections showed significant strain differences in blood ethanol concentration (all P ⬍ 0.05). Because these values were unrelated to behavioral response (see next section), they are not reported here (but see Ref. 7a). Best measures for grid test. On the basis of strain effect sizes and reliability, data from the grid test seemed most sensitive to strain-specific responses to lower doses (1.5 and 2.0 g/kg) of ethanol and during the second and third 5-min epochs after injection. Estimates of Genetic Codetermination Strain mean correlations were determined among the various indexes of strain sensitivity discussed above. With only four to eight strain means contributing to each estimated genetic correlation, there was little statistical power to detect significant associations; therefore, high levels of both type I and type II error were to be expected. However, the pattern of such correlations often reveals evidence for underlying genetic association. Therefore, to avoid rejecting the possibility of genetic relationships, we report correlations here. As is our usual practice, scatter plots for all correlations discussed were examined to ensure that obvious outliers were not leading to spurious interpretations (8). We concentrate our discussion on relationships among those specific test conditions shown to have the greatest strain effect sizes and reliability,
95 • OCTOBER 2003 •
www.jap.org
1346
GENETICS OF ETHANOL INTOXICATION
Fig. 2. Mean ⫾ SE activity (cm traveled) in the grid test during epochs 1 (A), 2 (B), and 3 (C) starting immediately after injection of 1 of 3 doses of ethanol (listed in key). D, E, and F depict ataxia scores (errors/cm) for the same mice. J Appl Physiol • VOL
95 • OCTOBER 2003 •
www.jap.org
1347
GENETICS OF ETHANOL INTOXICATION
Table 4. Genetic effect sizes and reliabilities of foot-slip errors in the grid test Ataxia, errors/cm Dose, g/kg
Epoch 1
2
1.5 2.0 2.5
0.17 (7) 0.09 (7) 0.05 (7)
0.50 (7) 0.25 (7) 0.01 (6)
Distance, cm 3
Epoch 1
2
3
0.24 (7) 0.51 (7) 0.22 (7)
0.33 (7) 0.42 (7) 0.07 (6)
0.40 (7) 0.19 (7) 0.24 (7)
Genetic effect sizes 0.23 (7) 0.27 (7) 0.06 (7)
Ataxia, errors/cm Dose, g/kg
Genetic
1.5 2.0 2.5
0.90† (8) 0.98† (7) 0.43 (6)
Distance, cm Phenotypic
Genetic
Phenotypic
0.92† (8) 0.82* (7) 0.63 (6)
0.81† (58) 0.86† (54) 0.62† (39)
Reliability: correlation between epochs 2 and 3 0.77† (56) 0.75† (52) 0.65† (38) Ataxia, errors/cm Epoch
Genetic
1 2 3
0.78* (8) 0.61 (7) 0.63 (7)
Distance, cm Phenotypic
Genetic
Phenotypic
0.98† (8) 0.92† (7) 0.83* (7)
0.60† (60) 0.64† (53) 0.61† (49)
Reliability: test-retest for 2.0 g/kg groups 0.36† (57) 0.62† (51) 0.34* (47)
Numbers in parentheses ⫽ number of strains or individual mice included in estimate. Genetic effect sizes (proportion of individual differences attributable to strains) ⱖ0.14 were based on significant ANOVAs (P ⬍ 0.05). In reliability results: * P ⫽ 0.02; † P ⬍ 0.01.
namely, response to 1 and 1.2 g/kg on the widest (19 mm) balance beam and response to 1.5 and 2 g/kg during epochs 2 and 3 of the grid test. Blood ethanol concentrations after the three higher doses were significantly correlated with each other (all r ⱖ 0.72, all P ⬍ 0.05). The only significant correlation between a blood ethanol level and a behavioral measure was the negative relationship between blood ethanol levels 35 min after 1.5 g/kg ethanol and corrected missteps on the 15.8-mm balance beam after 1 g/kg (r ⫽ ⫺0.75, P ⬍ 0.05). Blood ethanol levels taken 15 min after the 1 g/kg dose correlated significantly only with activity during epoch 3 after 2.5 g/kg in the grid test (r ⫽ 0.78, P ⬍ 0.05). Thus, strain differences in behavioral intoxication could not be attributed to differences in ethanol levels. For the balance beam, responses to the three ethanol doses on the widest beam were highly correlated (r ⱖ 0.76, P ⬍ 0.05). However, inspection of scatter plots revealed that these correlations were driven by inclusion of the 129 strain, and they were no longer significant when 129 was eliminated. The relationship between the 1 and 1.2 g/kg doses was also substantial for the other two beam widths (r ⱖ 0.78), and these correlations did not depend on 129. The pattern of strain differences after 1.4 g/kg was less well correlated with that observed after the lower doses for the narrower two beams (r ⱕ 0.54, P ⬎ 0.10). Interdose correlations for the grid test followed a similar pattern but depended on epoch. The dose-to-dose correlations for ataxia during epoch 2 were all r ⱖ 0.73. These associations, however, depended on the high scores of the A strain (129 mice were already eliminated) and were no longer significant with A data removed. For epochs 1 J Appl Physiol • VOL
and 3, however, only the correlations for 1.5 with 2 g/kg were significant (r ⱖ 0.82) and interpretable. Within each dose, strains responded similarly during each epoch (for the 1.5 g/kg dose, all r ⱖ 0.86, and for the 2.0 g/kg dose, all r ⱖ 0.92, P ⬍ 0.01). This pattern was not maintained as strongly for the 2.5 g/kg dose, where epochs correlated to r ⱕ 0.59 (P ⬎ 0.10). Strain-specific activity scores intercorrelated very well for the 1.5 and 2.0 g/kg doses, with each other and across epochs (0.82 ⱕ r ⱕ 0.98), but correlations for the highest dose were again not significant. The possibility of correlation of intoxication scores across the two tasks was our primary interest, but there was little evidence for genetic association. Corrected missteps on the widest balance beam after both 1.0 and 1.2 g/kg appeared to be significantly correlated with responses during epoch 1 of the grid test after 2.0 g/kg (r ⫽ 0.74 and 0.78, respectively, both P ⬍ 0.05), as well as with responses during epoch 3 after 1.5 g/kg (r ⫽ 0.69, P ⬍ 0.10, and r ⫽ 0.74, P ⬍ 0.05, respectively). However, all these apparent relationships became nonsignificant when the 129 strain was eliminated. Similarly, missteps after 1.5 and 2.0 g/kg in epoch 2 of the grid test were associated with corrected missteps after 1.2 g/kg on the 19-mm beam (r ⱖ 0.56, P ⬍ 0.10), but neither relationship was still present without the 129 and A strains. Grid test responses to the 2 g/kg dose during all three epochs were correlated with responses on the 15.8-mm beam after 1.4 g/kg dose was given (r ⱖ 0.84, P ⬍ 0.05, 0.05, and 0.10, respectively). However, these relationships were based on only five strains because 129, A, and C3H mice could not perform on the balance beam after this dose. Thus the
95 • OCTOBER 2003 •
www.jap.org
1348
GENETICS OF ETHANOL INTOXICATION
patterns of strain sensitivities on these tasks were rather different. We thought that strains highly activated by ethanol might perform more poorly on the balance beam. Thus we also explored the relationships between activity assessed in the grid test and intoxication on the balance beam (activity and missteps/cm in the grid test are computationally correlated and cannot be interpreted). Contrary to our prediction, distance traveled after the 1.5 g/kg dose in the grid test was significantly negatively related to errors on the 19-mm beam after the 1.4 g/kg dose (all r ⱖ ⫺0.81, P ⬍ 0.01), as well as the 1.2 g/kg dose (r ⱖ ⫺0.74, P ⬍ 0.05). The relationship was also apparent after the 1.0 g/kg dose (r ⫽ ⫺0.60) but did not reach significance. Activity after all three doses in the grid test during the first epoch was negatively correlated with 19-mm balance beam corrected missteps after 1.2 g/kg ethanol (r ⫽ ⫺0.75, ⫺0.64, ⫺0.88 and P ⬍ 0.05, 0.10, 0.01, for 1.5, 2.0, and 2.5 g/kg, respectively). DISCUSSION
Both the balance beam and the grid test appear to be somewhat useful tasks insofar as their ability to detect genetic differences in sensitivity to ethanol-induced intoxication. For the grid test, strain effect sizes (and reliabilities) for the activity measures were quite high. Previous studies have suggested that mouse strain differences in activity are quite heritable, even when compared across apparatus and test conditions (14). The ataxia measures in this task also were quite reliable after the lower doses (⬎0.9), and genetic contributions were moderately strong (see Table 4). The ataxia data from the grid test clearly show that reliable strain-specific data could be obtained even for measures with modest genetic influence. The reliabilities for corrected foot-slip errors on the widest (19 mm) balance beam were modest at best (0.2–0.5) and then only after the lower ethanol doses. Genetic contributions were even less pronounced but indicate that a significant proportion of individual differences were due to genotype. That said, each task presented specific difficulties. Specifics of the task apparatus and parameters matter a great deal. With the balance beam, doses too high, times too long after injection, or beams too narrow resulted in uninterpretable data for many animals from some strains. This is similar to what was found when we and our colleagues explored parameters and doses appropriate for the accelerating and fixed-speed rotarod (37, 38). Even when we adapted our test conditions to a narrower range of doses and beam widths, for the group of inbred strains studied here, three strains engaged in behaviors incompatible with the use of simple foot-slip errors as an index of ethanol intoxication. The appearance of behaviors other than walking along the beam has been reported before. In a series of experiments comparing heterozygous lurcher mutant mice (Lc/⫹) and their normal littermate controls (⫹/⫹) J Appl Physiol • VOL
of the B6CBACa-Aw⫺J/A strain, Lalonde and his collaborators (27) first reported that lurcher mutants and wild types did not differ in latency to fall from a 10-mm wide wall. Lurcher is a cerebellar mutation leading to losses in cerebellar Purkinje and granule cells, as well as inferior olive cells, and the mutants have difficulty with some tasks assessing ataxia and equilibrium. In a subsequent study, lurcher mutants more frequently hung from the bar by their forepaws or by the ipsilateral fore and hind paws than wild types, although they still did not fall from the beam sooner (28). These behaviors are similar to those that we report in Table 1 as “hanging/falling.” Subsequent studies with beams modified with sticky tape to improve traction for the lurcher mutants or with wider (107-mm) beams showed that lurchers learned to stay on the beam longer with practice but did not reach levels of performance attained by wild types (26, 29). Transgenic mice with excess copies of the 751 amino acid isoform of APP could not traverse even the widest of a series of round and square bridges without wrapping themselves around the bars (35). Others have employed multiple beam widths to study mouse behavior. Contet et al. (6) compared 129S2/SvHsd and C57BL/6JOlaHsd mice for latency to traverse multiple round dowels (33-, 27-, 21-, 14-, or 8-mm diameter). The C57 strain traversed all but the smallest dowel much more quickly than the 129 strain; on the smallest dowel, most mice simply hung upside down (6). Kiernan et al. (25) employed either 28- or 23-mm beams or 28-, 18-, 12-, or 9-mm round rods in their study of tenascin C-null mutants. Different genetic backgrounds were tested. Mutants showed more foot-slip errors on all widths on one background, only on the narrowest width round dowels (9 mm) for a different background, and no deficit at all when the background was essentially a 129 strain. APP transgenics for the 695 and 751 amino acid isoforms were more impaired on more difficult round and square bridges (35). These studies emphasize the importance of test conditions but also of strain background. The background strains of mice included at least B6, CBA, and an unspecified substrain of 129. In studies with CD-1 outbred mice, beam widths of 6, 9, 12, and 21 mm were progressively less difficult, but only latency to cross was measured and estrogen treatment was equally effective across beam widths (32). Purkinje cell degeneration mutants did not differ from controls in latency to cross round dowels of either 46 or 85 mm (30). To characterize the developmental onset of motor deficits in mice transgenic for exon 1 of the human Huntington’s disease gene, Carter et al. (5) used 28-, 12-, or 5-mm beams or 28-, 17-, or 11-mm round dowels. Mice were tested serially from widest to narrowest, and latency and hindfoot slips were recorded. As the animals aged from 5 to 14 wk, the transgenics began to show increased latencies and more foot slips. The deficits appeared earliest on the most difficult beams. Also, they noted the emergence in the transgenics of the hindlimb dragging that we saw in 129, A, and C3H mice by 10–11 wk. In this study,
95 • OCTOBER 2003 •
www.jap.org
GENETICS OF ETHANOL INTOXICATION
use of more difficult task variants clearly allowed the investigators to detect the motor deficit at an earlier age (5). Voikar et al. (42) used a 20-mm round dowel to test Niemann-Pick type C mutants vs. their BALB/c (substrain unspecified) wild types. Mutants developed deficits seen as shorter latencies to fall by 5–6 wk (42). Most of these other studies employing balance beamlike apparatus reported latency measures to traverse the beam. In our tests, mice often stopped in transit, confounding latency measures. Use of extrinsic motivation (e.g., food or water deprivation) might overcome this limitation. However, the bigger problem we found was that too many standard inbred strains simply could not perform the task under most conditions, making it difficult to recommend for widespread application. In particular, the 129 substrain we used had difficulty with narrower beams. The selection of 129 strains as the genetic background for null mutations may render this task too difficult even for wild-type 129s. Nonetheless, there are a number of 129 substrains with rather different genetic backgrounds, and the particular 129 substrain employed may behave differently (40). The grid test presented somewhat fewer barriers to interpretation. Nonetheless, at higher ethanol doses, data from many animals needed to be eliminated because the animals were so intoxicated that they were not moving in the apparatus. Strains could be clearly discriminated in their sensitivity to ethanol intoxication, but a broad-based screen of strains will definitely need to employ more than one dose of ethanol. The grid test offers a good cautionary note about the appropriate use of automation in collection and analysis of mouse behavior. Driven by the desire for rapid throughput behavioral screens that can be used to test randomly mutagenized mice, some have argued for highly automated data collection (18). We have ourselves argued elsewhere that, when feasible, automated data collection can improve the quality of the data by reducing the role of the experimenter who may potentially elicit strain-specific responses (43, 44). In the grid test, however, it turned out to be necessary to examine the temporal pattern of infrared beam interruptions, distance traveled computations (in cm), and foot-slip errors to determine whether the mouse was actually moving. Had we not done so, we would have interpreted the data from the recording devices incorrectly, mistaking some of the most intoxicated mice for the least. Although the balance beam elicited a number of other behaviors that rendered data from some strains invalid, at least the data were visible to the experimenter on line, making their interpretation more straightforward. It seems unlikely that the genetic influence on ethanol sensitivity in these two tasks shows substantial pleiotropism. Genetic correlations between strain sensitivities under different conditions on the same task were reasonably good, which offers face validity. However, even though the behavioral sign of intoxication recorded for both tests was a foot-slip error, the correlations of strain means for ethanol sensitivity across J Appl Physiol • VOL
1349
the two tasks were generally not significant. When significant, they generally turned out to be spurious, in that they simply reflected the high sensitivity of 129 (and sometimes A) on both tasks. This suggests that the genes influencing sensitivity are mostly different, an interesting result. This is not entirely surprising. We previously tested mice with a null mutation for the serotonin 1B receptor subtype and their wild-type controls for sensitivity to ethanol by using several measures of intoxication, including the grid test and balance beam. The null mutants were less sensitive to ethanol in both of those tests. However, there was no difference between genotypes using several other tests, including one version of the accelerating rotarod, one version of the fixed-speed rotarod, a test of grip strength, one of dowel balancing, and other tests (4). If we had based our estimate of the effects of this serotonin receptor subtype gene on ethanol sensitivity on a single test, we could have erroneously concluded that it had a general effect on intoxication, rather than the more subtle effect it appears to have. Furthermore, in a test of several variants of the rotarod following different ethanol doses, we found strain sensitivities to differ depending on the specific task parameters and doses employed (37). Although the 129 strain had difficulty with both the tests of ethanol ataxia that we employed here, they displayed middling sensitivity to ethanol on the accelerating rotarod in a comparison of 20 inbred strains. That study also included BTBR, which was by far the most sensitive to ethanol on the rotarod (37). Therefore, it seems unlikely that 129 mice will show a generalized sensitivity to ethanol intoxication as more tasks are investigated. Rather, they seem specifically responsive on the tasks employed here. The data we report seem to reflect different brain sensitivity to ethanol across strains. Although the strains differed in blood ethanol levels, those levels did not correlate significantly with sensitivity scores. Earlier studies of strain differences in sensitivity to ethanol have shown a similar independence of blood or brain ethanol concentrations (7–9, 13, 37). Of course, strain sensitivity to other effects of ethanol may reflect metabolic differences, and there may be strains other than those we tested here that are sensitive because they achieve higher brain ethanol concentrations. The neurobiological target of ethanol in these tasks is not known. The tasks employed may involve balance, proprioceptive feedback, muscle strength, gross motor sequencing, and/or spatial planning. Many ataxias result from disorders of cerebellum and/or its connections, but circuitry integrating motor cortex and striatum may also be involved (32). The sex difference in the effects of ethanol on performance on the balance beam appeared to be entirely accounted for by body weight. In other tasks, we have also seen that the heavier males are more impaired by ethanol, but whether these differences can be entirely attributed to body weight depends on the task; only for the screen test have we found any strain-dependent sex differences not due to body weight (7a, 37). Another way body size could
95 • OCTOBER 2003 •
www.jap.org
1350
GENETICS OF ETHANOL INTOXICATION
contribute is through differences in forelimb or hindlimb stance width. Both the present tasks required movement, and genetic differences in gait dynamics, which would assess stance and gait patterning, seem a likely target for future studies. A new system for assessing gait dynamics in mice has recently been used to characterize a mouse model of Down syndrome (20, 21). Preliminary studies of ethanol’s effects on gait in B6 and D2 mice have shown differential effects in these two strains and also suggest that ethanol affects hindlimb function more potently than forelimbs. Future studies will explore ethanol effects on gait in the eight strains we studied here (T. G. Hampton, personal communication). As noted, learning may have played a role in performance on the two wider balance beams. Given the negative genetic relationship between ethanol-induced activity and balance beam intoxication, the mesolimbic dopamine system may also be involved. Ethanol effects on rotarod performance have been postulated to result from interactions with adenosine in cerebellum and striatum (15, 16, 33). However, other aspects of motor learning and performance involve at least basal ganglia and frontoparietal cortex (19, 24). It is reasonable to assume that the genetic differences in sensitivity to ethanol ataxia will ultimately be traceable to these circuits. In other studies, we have seen that patterns of mouse inbred strain differences are often highly reliable both within and across laboratories, but this does not apply to all behaviors. An index of ethanol selfadministration was highly reliable across three laboratories, whereas an index of anxiety derived from the elevated plus maze was reliable within each of the three laboratories but was less so across laboratories (12). Sensitivity to ethanol intoxication on the accelerating rotarod was also highly reliable across laboratories (37). In the studies reported here, the sensitivity to ethanol on the grid test was reliable within one of our laboratories (that of J. C. Crabbe). It will be interesting to move to the next step in our studies, which will involve testing a subset of the ethanol doses and conditions reported here in two laboratories (those of J. C. Crabbe and D. Wahlsten) simultaneously. This next step also involves extending this analysis to at least 20 of the 40 strains recommended by the Mouse Phenome Project to obtain a broader idea of the relevance of our findings for Mus musculus domesticus. Finally, despite our best efforts, we cannot rule out the possibility of gene-environment correlations as contributors to the patterns of strain differences in behavior (36). One source of such gene-environment correlation is the repeated testing of mice on three different beams because, for example, strain differences in motor learning might contribute to the apparent strain differences on the beams tested second and third. Another clear source of gene-environment correlation is the immediate postweaning feeding of the strains at The Jackson Laboratory. Because different strains are given different dietary formulations until shipping, the postweaning nutrition may differ among strains. We do not believe that this is a major problem. The goal of J Appl Physiol • VOL
using different diets is to try to bring all strains up to a similar level of nutritional adequacy, while preventing such unwanted side effects as allowing obesityprone strains to develop incipient diabetes. Another potential source of gene-environment correlations is strain-specific (and different) patterns of maternal behavior. To avoid this, all strains would need to be cross-fostered to a common strain of dams. In any case, for all practical purposes, there is nothing we can really do about it! All comparative studies from all laboratories that use purchased Jackson Laboratory mice are comparing mice that arrive with the same built-in gene-environment correlations; thus the best strategy is simply to remember that strain differences do not represent strictly genetic differences (36). Nonetheless, careful strain comparisons in standardized test environments can offer insight into gene-behavior pathways. We thank Karyn L. Best, Christina J. Cotnam, Janet D. Dorow, Jason R. Sibert, and Jessica M. Zuraw for technical assistance and Dr. Justin S. Rhodes for help with statistical analyses. DISCLOSURES These studies were supported by National Institute on Alcohol Abuse and Alcoholism Grants AA-10760 and AA-12714, a grant from the Department of Veterans Affairs, National Sciences and Engineering Research Council Grant 45825, and the Mouse Phenome Project of The Jackson Laboratory. REFERENCES 1. Arendash GW, King DL, Gordon MN, Morgan D, Hatcher JM, Hope CE, and Diamond DM. Progressive, age-related behavioral impairments in transgenic mice carrying both mutant amyloid precursor protein and presenilin-1 transgenes. Brain Res 891: 42–53, 2001. 2. Belknap JK. The grid test: a measure of alcohol- and barbiturate-induced behavioral impairment in mice. Behav Res Meth Instrumen 7:1: 66–67, 1975. 3. Belknap JK, Berg JH, Ondrusek G, and Waddingham S. Barbiturate withdrawal and magnesium deficiency in mice. Psychopharmacology (Berl) 59: 299–203, 1978. 4. Boehm SL II, Schafer GL, Phillips TJ, Browman KE, and Crabbe JC. Sensitivity to ethanol-induced motor incoordination in 5-HT1B receptor null mutant mice is task-dependent: implications for behavioral assessment of genetically altered mice. Behav Neurosci 114: 401–409, 2000. 5. Carter RJ, Lione LA, Humby T, Mangiarini L, Mahal A, Bates GP, Dunnett SB, and Morton AJ. Characterization of progressive motor deficits in mice transgenic for the human Huntington’s disease mutation. J Neurosci 19: 3248–3257, 1999. 6. Contet C, Rawlins JN, and Deacon RM. A comparison of 129S2/SvHsd and C57BL/6JOlaHsd mice on a test battery assessing sensorimotor, affective and cognitive behaviours: implications for the study of genetically modified mice. Behav Brain Res 124: 33–46, 2001. 7. Crabbe JC. Sensitivity to ethanol in inbred mice: genotypic correlations among several behavioral responses. Behav Neurosci 97: 280–289, 1983. 7a.Crabbe JC, Cotnam CJ, Cameron AJ, Schlumbohm JP, Rhodes JS, Melten P, and Wahlsten D. Strain differences in three measures of ethanol intoxication in mice: the screen, dowel, and grip strength tests. Genes Brain Behav. In press. 8. Crabbe JC, Gallaher ES, Phillips TJ, and Belknap JK. Genetic determinants of sensitivity to ethanol in inbred mice. Behav Neurosci 108: 186–195, 1994. 9. Crabbe JC, Janowsky JS, Young ER, Kosobud A, Stack J, and Rigter H. Tolerance to ethanol hypothermia in inbred mice:
95 • OCTOBER 2003 •
www.jap.org
GENETICS OF ETHANOL INTOXICATION
10.
11.
12. 13. 14.
15. 16. 17.
18. 19. 20.
21.
22. 23. 24. 25.
26.
genotypic correlations with behavioral responses. Alcohol Clin Exp Res 6:4: 446–458, 1982. Crabbe JC, Johnson NA, Gray DK, Kosobud A, and Young ER. Biphasic effects of ethanol on open-field activity: sensitivity and tolerance in C57BL/6N and DBA/2N mice. J Comp Physiol Psychol 96: 440–451, 1982. 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. Crabbe JC, Wahlsten D, and Dudek BC. Genetics of mouse behavior: interactions with laboratory environment. Science 284: 1670–1672, 1999. Crabbe JC, Young ER, and Kosobud A. Genetic correlations with ethanol withdrawal severity. Pharmacol Biochem Behav 18, Suppl 1: 541–547, 1983. Crawley JN, Belknap JK, Collins A, Crabbe JC, Frankel W, Henderson N, Hitzemann RJ, Maxson SC, Miner LL, Silva AJ, Wehner JM, Wynshaw-Boris A, and Paylor R. Behavioral phenotypes of inbred mouse strains: implications and recommendations for molecular studies. Psychopharmacology (Berl) 132: 107–124, 1997. Dar MS. Antagonism by intracerebellar Ro15–4513 of acute ethanol-induced motor incoordination in mice. Pharmacol Biochem Behav 52: 217–223, 1995. Dar MS, Mustafa SJ, and Wooles WR. Possible role of adenosine in the CNS effects of ethanol. Life Sci 33: 1363–1374, 1983. Gallaher EJ, Jones GE, Belknap JK, and Crabbe JC. Identification of genetic markers for initial sensitivity and rapid tolerance to ethanol-induced ataxia using quantitative trait locus analysis in BXD recombinant inbred mice. J Pharmacol Exp Ther 277: 604–612, 1996. Gerlai R. Phenomics: fiction or the future? Trends Neurosci 25: 506–509, 2002. Graybiel AM. The basal ganglia and chunking of action repertoires. Neurobiol Learn Mem 70: 119–136, 1998. Hampton TG, Kale A, Gladstone K, Bridges J, Costa ACS, and Amende I. How to characterize gait dynamics in mouse models of neuromuscular diseases. Program No. 22.60. 2002 Abstract Viewer and Itinerary Planner. Washington, DC: Society for Neuroscience, 2002. Online. Hampton TG, Stasko MR, Amende I, Kale A, and Costa ACS. Gait dynamics in a mouse model of Down syndrome. Program No. 664.15. 2003 Abstract Viewer and Itinerary Planner. Washington, DC: Society for Neuroscience, 2003. Online. Hegmann JP and Possidente B. Estimating genetic correlations from inbred strains. Behav Genet 11: 103–114, 1981. Herson PS, Virk M, Rustay NR, Bond CT, Crabbe JC, Adelman JP, and Maylie J. A mouse model of episodic ataxia type-1. Nat Neurosci 6: 378–383, 2003. Hikosaka O, Nakamura K, Sakai K, and Nakahara H. Central mechanisms of motor skill learning. Curr Opin Neurobiol 12: 217–222, 2002. Kiernan BW, Garcion E, Ferguson J, Frost EE, Torres EM, Dunnett SB, Saga Y, Aizawa S, Faissner A, Kaur R, Franklin RJ, and ffrench-Constant C. Myelination and behaviour of tenascin-C null transgenic mice. Eur J Neurosci 11: 3082–3092, 1999. Lalonde R, Botez MI, Bontemps R, and Loron P. Effects of an ectodermal microceptor preparation on motor coordination in cerebellar mutant mice. Pharmacol Biochem Behav 49: 777–779, 1994.
J Appl Physiol • VOL
1351
27. Lalonde R, Botez MI, Joyal CC, and Caumartin M. Motor abnormalities in lurcher mutant mice. Physiol Behav 51: 523– 525, 1992. 28. Lalonde R, Joyal CC, Guastavino JM, and Botez MI. Hole poking and motor coordination in lurcher mutant mice. Physiol Behav 54: 41–44, 1993. 29. Lalonde R, Joyal CC, and Thifault S. Beam sensorimotor learning and habituation to motor activity in lurcher mutant mice. Behav Brain Res 74: 213–216, 1996. 30. Le Marec N and Lalonde R. Sensorimotor learning and retention during equilibrium tests in Purkinje cell degeneration mutant mice. Brain Res 768: 310–316, 1997. 31. Massaquoi S and Hallett M. Ataxia and other cerebellar syndromes. In: Parkinson’s Disease and Movement Disorders, edited Jankovic J and Tolosa E. Baltimore, MD: Williams & Wilkins, 1998, p. 623–686. 32. McDermott JL, Kreutzberg JD, Liu B, and Dluzen DE. Effects of estrogen treatment on sensorimotor task performance and brain dopamine concentrations in gonadectomized male and female CD-1 mice. Horm Behav 28: 16–28, 1994. 33. Meng ZH and Dar MS. Possible role of striatal adenosine in the modulation of acute ethanol-induced motor incoordination in rats. Alcohol Clin Exp Res 19: 892–901, 1995. 34. Paigen K and Eppig JT. A mouse phenome project. Mamm Genome 11: 715–717, 2000. 35. Perry TA, Torres E, Czech C, Beyreuther K, Richards SJ, and Dunnett SB. Cognitive and motor function in transgenic mice carrying excess copies of the 695 and 751 amino acid isoforms of the amyloid precursor protein gene. Alzheimers Res 1: 5–14, 1995. 36. Rowe DC. Assessing genotype— environment interactions and correlations in the postgenomic era. In: Behavioral Genetics in the Postgenomic Era, edited by Plomin R, DeFries JC, Craig IW, and McGuffin P. New York: Am. Psychol. Assoc., 2002, p. 71–86. 37. Rustay NR, Wahlsten D, and Crabbe JC. Assessment of genetic susceptibility to ethanol intoxication in mice. Proc Natl Acad Sci USA 100: 2917–2922, 2003. 38. Rustay NR, Wahlsten D, and Crabbe JC. Influence of task parameters on rotarod performance and sensitivity to ethanol in mice. Behav Brain Res 141: 237–249, 2003. 39. Schafer GL and Crabbe JC. Sensitivity to ethanol-induced ataxia in HOT and COLD selected lines of mice. Alcohol Clin Exp Res 20: 1604–1612, 1996. 40. Simpson EM, Linder CC, Sargent EE, Davisson MT, Mobraaten LE, and Sharp JJ. Genetic variation among 129 substrains and its importance for targeted mutagenesis in mice. Nat Genet 16: 19–27, 1997. 41. Sokal RR and Rohlf FJ. Biometry. San Francisco, CA: Freeman, 1981. 42. Voikar V, Rauvala H, and Ikonen E. Cognitive deficit and development of motor impairment in a mouse model of NiemannPick type C disease. Behav Brain Res 132: 1–10, 2002. 43. Wahlsten D, Metten P, Phillips TJ, Boehm SL II, Burkhart-Kasch S, Dorow J, Doerksen S, Downing C, Fogarty J, Rodd-Henricks K, Hen R, McKinnon CS, Merrill CM, Nolte C, Schalomon M, Schlumbohm JP, Sibert JR, Wenger CD, Dudek BC, and Crabbe JC. Different data from different labs: lessons from studies of gene-environment interaction. J Neurobiol 54: 283–311, 2003. 44. Wahlsten D, Rustay NR, Metten P, and Crabbe JC. In search of a better mouse test. Trends Neurosci 26: 132–136, 2003.
95 • OCTOBER 2003 •
www.jap.org
