Issues in Controlled Substance Use: Papers and Commentary, Conference on Issues in Controlled Substance Use (1980)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

ANIMAL MODELS AS PHARMACOGENETIC TOOLS: SOME INITIAL EXPLORATIONS INTO ALCOHOL CONSUMPTION Gerald E. McClearn Basic principles concerning the initiation and sustention of alcohol ingestion by laboratory animals might possibly be relevant to the issue of control or regulation of alcohol consumption in human beings. There is, of course, an enormous literature on animal consump- tion of alcohol and the environmental manipulations that increase or decrease it. I shall not review this extensive literature, but shall concentrate on the ways in which genetics has been used by our research group to increase understanding of the ethanol intake control system. This work can be fairly described as just a beginning; it is offered simply to illustrate the kinds of things that can be done. GENETICS AND VOLUNTARY ALCOHOL CONSUMPTION IN MICE To set a context, I shall first describe the studies that have demonstrated genetic influence on alcohol consumption in mice. Our first study compared several inbred mouse strains in a two-bottle choice situation in which the animals could choose between tap water and a l0-percent ethanol solution. The daily intake from each bottle was recorded for each mouse for a period of two weeks. Although based on rather small samples, the results of this first investigation (McClearn and Rodgers, l959) were quite persuasive in characterizing animals of the DBA/2, BALB/c, and A/Cal strains as largely alcohol avoiders; the C3H/NT strains as rather variable, with some animals showing moderate preference; and the C57Bl/Crgl strain as demonstrating a high preference for the ethanol solution. The results of this inves- tigation are illustrated in Figure l. NOTE: The research described in this report was supported by NSF grant G-l422l, NIAAA grant AA-00293, and NIGMS grants GM-l0735 and GM-l4547. The report was written while the author was a faculty fellow of the University of Colorado Council on Research and Creative Work. l42

l43 These strain differences have been replicated many times both in our laboratory and elsewhere. Figure 2 presents frequency distribu- tions from a representative replication. The DBA/2 animals are uni- formly low in alcohol intake; indeed, the low consumption displayed in this figure is very near the resolving power of the technique and pro- bably represents simply leakage from the ethanol solution bottle. The A animals are also characterized by relatively low alcohol intake, al- though they are somewhat more variable than the DBA/2s. C57BL mice have a high mean alcohol consumption; on the average, they will consume two-thirds to three-quarters of their total daily liquid intake from the ethanol solution bottle. However, the variability within this strain is noteworthy. It is not unusual, as in the example shown in Figure 2, to find an occasional abstaining C57BL mouse, at least when observations are made only for a two-week period. Most alcohol preference data on C57BL animals have been obtained during two-week observation periods, and the mean consumption has been found to increase from about 50 percent of the total fluid intake to 80 percent or more in that length of time. Longer-term ingestion of eth- anol might be expected to result in acquired tissue tolerance, so that the animals might be able to tolerate larger daily quantities. A pre- liminary study was therefore initiated for the purpose of observing trends in ethanol intake over a prolonged observation period. Five male C57BL mice were offered a choice of tap water and l0-percent eth- anol solution (0-l0 group), while five others were offered a choice between tap water and 20-percent ethanol solution (0-20 group). The first condition is that of the standard alcohol test; the second condi- tion, with the 20-percent ethanol solution, was included to make it possible for larger quantities to be ingested than would be possible with a l0-percent solution. All animals were l37 days of age at the start of the experiment. Daily readings were taken for l9 weeks, and bottle positions were interchanged every third day. Results of this study are shown on Figure 3. Although the num- bers are small, there is a clear increment in preference of the 0-l0 group from the first to the second week and a stable plateau from that time until the fourteenth week. For the 0-20 group, there is a sub- stantial reduction in alcohol preference ratio from the first until the fourth or fifth week, after which a stable level of intake prevails. The striking change for the 0-l0 group in the fifteenth and sixteenth weeks is attributable to a single animal whose preference dropped from 0.80 for week l4 to 0.l3 and 0.l9, respectively, for weeks l5 and l6, with a recovery to 0.72 for week l7 and approximately the same level thereafter. Examination of the protocol records reveals no obvious explanation for this abrupt change. The plateau level of intake for the 0-l0 group is so high as to represent almost exclusive choice of the l0-percent ethanol solution by the five animals tested, and it might be argued that the animals would have consumed more had it been physically possible. It is interesting that animals in the 0-20 group, which could have consumed substantially more ethanol because of the stronger concentration, did not ingest more ethanol. In fact, the respective levels of the mean preference ratios for the two groups (with that of the 0-20 group being roughly half that of the 0-l0 group)

l44 suggest that the animals were ingesting approximately equal quantities of absolute ethanol. In the foregoing study, increasing age was confounded with dura- tion of exposure to alcohol. For both theoretical and practical rea- sons, it is desirable to know if any consistent changes occur with respect to alcohol preference of C57BL mice as a function of age it- self. For example, insights into the physiological determinants of preference might be obtained by correlating enzymatic or hormonal changes with observed age-related preference changes. As a practical procedural matter, the discovery of a uniform preference over a broad range would permit flexibility in testing schedules. Also, if young animals display essentially the same preference characteristics as older animals, the generation interval in breeding experiments could be shortened. Three groups of eight animals each from the C57BL strain were tested in a standard l4-day, two-bottle preference situation. Animals in the three groups were 2l days old, 27 days old, and 33 days old, respectively, on the first day of testing. In order to control for possible litter-specific environmental effects, animals from four different litters were assigned in equal number to each of the three conditions. No significant differences were found among the three groups. All displayed moderate alcohol preference. Thus, the mech- anism which distinguishes the C57BL strain from other strains appears to be operative even at an early age. Some incidental observations on BALB/c mice suggested that the alcohol preference of younger animals of this strain might be consider- ably higher than that of adults. We therefore conducted a systematic investigation in which three groups of ten male BALB/c mice each were tested for alcohol preference at the ages of 6 weeks, 9 weeks, or l2 weeks. The mean alcohol preference ratios of these groups were 0.4l, 0.37, and 0.l3, respectively. More extensive observations on animals from 3 weeks of age through 20 weeks have given clear indication that alcohol preference of young BALB/c animals greatly exceeds that of adults (Kakihana and McClearn, l963). As shown in Figure 4, the tran- sition from alcohol-preferring to alcohol-avoiding appears to take place in a limited time span between the 7th and the 9th weeks of age. We then conducted a longitudinal study in which BALB/c mice were tested at 4 weeks and again at l6 weeks of age. Figure 5 shows the dramatic reduction in alcohol preference over this age range. To determine possible developmental changes during continuous exposure to a preference situation, Kakihana (l965) assigned 36 BALB/c male mice in equal numbers to three groups. In the first group (choice), the animals were placed in a two-bottle choice situation at 5 weeks of age; animals of the second group (forced) were given only l0- percent ethanol to drink beginning at 5 weeks of age; and control ani- mals were reared normally. They remained under these conditions for ll weeks. From the l2th through l5th weeks of age, all animals were given the standard preference test. The mean preference ratios for the three groups at the l5th week were: choice, 0.74; forced, 0.25; control, 0.08. It is clear that the return to typical low preference (or, more correctly, to alcohol avoidance) does not occur in BALB/c animals if they are allowed to remain in a choice situation from the stage at

l45 which moderate preference is found. It appears that some intake con- trol system becomes "locked on" by ingestion during the critical devel- opmental period. Furthermore, results of this study showed that it is not simply presence of the substrate that is relevant. The "forced" group (which drank nothing but l0-percent ethanol from the 5th through the llth week) dropped to a low level of ethanol intake, although sig- nificantly higher than control, when offered a choice. In a further investigation of the lability of the BALB/c consump- tion pattern, alcohol intake observations of l50 days' duration were begun at various ages. Between one-third and one-half of these BALB/c mice displayed one or more episodes of high alcohol intake, often pre- ceded or followed by periods of low intake. Representative records of these episodic drinkers are shown in Figure 6. In order to determine if such abrupt changes of preference would be displayed by other strains, a similar experiment was undertaken with C57BL, A, and DBA mice. The latter two strains exhibited consistently low alcohol consumption without the episodic drinking shown by some BALB/c animals. As shown in Figure 7, however, some C57BL animals were found to display erratic drops in alcohol preference from their typi- cally high values, others showed a gradual decline with increasing age, and some showed an abrupt onset of alcohol consumption 20 or more days after their first exposure to the choice situation. Perhaps these rep- resent the "abstainers" that appear in the standard two-week testing situation. Whether such changes in alcohol intake are initiated by endogenous events or triggered by external stimuli, they would seem to offer opportunities for elucidating the mechanisms of alcohol intake regulation through studies of the physiological correlates of the changes. THE ETHANOL INTAKE CONTROL SYSTEM OF C57BL MICE The existence of intermediate levels of ethanol consumption in various strains and the fact that even C57BL mice do not drink exclu- sively from the ethanol solution indicate that the mechanism for con- trolling ethanol intake does not simply determine acceptance or avoid- ance. The ethanol intake control system, which will henceforth be abbreviated EICS for convenience, may well be found to be diffuse and to involve many tissues and organ systems—such as taste mechanism, neuroregulatory cells, osmotic systems, etc. Before a determined effort is made to analyze the biochemical, physiological, and anatom- ical nature of the EICS, however, it would likely be advantageous to have a better behavioral description of it. If the EICS operates to limit absolute amounts of alcohol in- gested, the preference ratio of animals offered a choice between water and a 20-percent solution should be lower than that of animals offered a choice between water and a l0-percent solution, but the total amount of alcohol ingested should be the same. It is also possible that adap- tation is important; if so, animals previously offered one solution might drink more or less when switched to another solution than they would have had they been tested on the latter solution throughout the

l46 test. Several experiments have been undertaken to investigate these possibilities. A total of 32 C57BL mice was used in the first experiment in this series. Half of these animals were tested for a two-week period with a choice of water and l0-percent alcohol solution, while the other half had a choice between water and 20-percent alcohol solution. At the end of this period, each group was subdivided into two groups, l0-percent solution vs. water or 20-percent solution vs. water, and tested for two weeks. The results are given in Figure 8. In terms of absolute amount of ethanol, the 20-percent group consumed more daily than the l0- percent group, and the amount of ethanol consumed was not constant when concentration of the solution was changed from l0 percent to 20 percent or vice versa. This seems to argue against the hypothesis of a simple regulatory mechanism responsive solely to amount of ethanol ingested. Surprisingly, these data suggested that there is an intake level char- acteristic of a given concentration and that the level increases when concentration increases. A shift in concentration is rapidly followed by an adjustment of intake to correspond to the new solution, with little evidence of a large or persistent residual effect from previous experience with the other concentration. In order to extend these observations over a broader concentra- tion range, a second study was performed in which l2 C57BL males were assigned to each of four choice conditions at 56 days of age. The conditions offered a selection between tap water and 3-percent, 6- percent, l2-percent and 24-percent ethanol solution, respectively. The standard preference testing routine was carried out for two weeks, at which time the ethanol solution offered to each animal was reduced to one-half of the previous concentration value and consumption was re- corded for an additional week. All data were corrected for leakage. An examination of the solid line in the preshift portion of Figure 9 shows that the absolute amount of ethanol ingested increased from 3 percent through l2 percent with a subsequent drop at 24 percent. The line separating the cross-hatched from the stippled areas of the figure represents the total volume of ethanol solution consumed (i.e., abso- lute amount of ethanol plus the water from the ethanol solution). Here it may be that the volume of ethanol solution consumed increased from 3 percent to 6 percent, with a subsequent drop at l2 percent and 24 per- cent. The situation for the postshift data is seen to be remarkably similar. In general, the amount of water ingested (stippled area) remained the same as in the preshift period. The total volume of ethanol solution consumed increased, however. The general description of the EICS obtained from the earlier study with l0-percent and 20-percent solutions applies quite well to these data over the range from l.5 percent to l2 percent. That is, the animals formerly on l2-percent choice consumed about as much ethanol when dropped to 6-percent choice as did the animals on 6-percent choice initially. In like manner, the 6-percent choice animals dropped close to the level of the initial 3-percent animals, and the 3-percent ani- mals dropped to a value close to what a linear extrapolation of the l2-percent, 6-percent, and 3-percent data points would suggest that the l.5-percent consumption level should be. These changes are shown on an

l47 expanded ordinate in Figure l0. However, the animals initially on 24- percent choice did not approach the level of the initial l2-percent animals when their test solution was reduced to l2 percent. These data thus suggest that (l) the monotonic increasing relationship between concentration and absolute amount ingested holds only to a concentra- tion value somewhere between l2-percent and 24-percent, and (2) the independence of consumption level from previous experience with alter- nate concentrations holds only for this range. Above this concentra- tion value, the absolute amount of ethanol ingested decreases and there is a carry-over effect of experience when the concentration is reduced. Shifting the groups as a whole from higher to lower concentra- tions permitted fairly accurate assessment of postshift ingestion levels, but this procedure provided no control data in the form of third-week measures on animals that had been tested continuously at a given concentration. Furthermore, only the effects of downward shifts were investigated. In order to provide appropriate control values and to observe the effect of both upward and downward shifts, an additional experiment was performed. Ten C57BL males (57 days of age) were as- signed to a 3-percent choice group, l0 were assigned to a 24-percent group, and l5 each were assigned to 6-percent and l2-percent choice groups. After l5 days under this preshift choice condition, each group was subdivided into groups of 5 animals each. One group of 5 animals previously on 3-percent choice was changed to 6-percent, while the other remained at 3-percent. One of the groups of 5 animals previously on 6-percent choice was changed to 3 percent, one remained at 6 percent, and one was increased to l2 percent, etc. Observation of postshift consumption continued for another l5-day period. All data were sub- jected to the appropriate correction for bottle leakage. Figure ll shows that the absolute amount of ethanol consumed increased regularly from 3 percent through 24 percent in both the preshift and postshift phases. In contrast to the results of the previous study, no decrease in absolute amount appeared. The total volume of ethanol solution in- gested increased from 3 percent through l2 percent, with a subsequent decrease. This pattern of increase followed by decrease is similar to that in the preceding study, but the decrease in this case was much smaller and occurred at a higher concentration value. In the postshift results shown in Figure ll, the circled numbers identify the mean levels of the various groups currently being tested at the concentration shown on the abscissa. Thus, a circled 3 above the 6-percent point identifies a group currently being tested at 6- percent choice but previously on 3-percent choice. Lines representing total liquid, volume of ethanol solution, and absolute amount of eth- anol consumed are drawn through the means at each point. Although there is a general consistency of results for the different shifted and control groups with respect to the total liquid measure and the ethanol solution measure, the most remarkable uniformity is found for the eth- anol amount measure. In this case, the data points for the various groups are practically indistinguishable. Thus, it is apparent that the absolute amount of ethanol ingested provides the most stable and reliable measure in terms of description and prediction. The outcome

l48 of this experiment differs somewhat from the previous results, princi- pally in the absence of a dropoff in absolute amount of ethanol in- gested between the l2-percent and 24-percent choice conditions. The explanation for this discrepancy is not obvious, but presumably must involve some unknown change in environmental conditions in the labora- tory from one experiment to the next or a random sampling fluctuation. Results of these three behavioral studies of the EICS agree in suggesting that the amount of ethanol ingested by C57BL mice is depen- dent on the concentration of the solution. The best evidence is that there is an increasing logarithmic function over at least a substantial range of concentrations. When shifted from a previous concentration condition, C57BL animals adjust their intake to correspond to the new concentration level (as defined by the control group) with no evidence of a carry-over influence from previous experience at a different con- centration. It should be most interesting to determine whether there is a threshold beyond which these relationships do not hold. Since concentrations greater than about 25 percent show an increasing ten- dency to drain from the cylinder, explorations of the higher concentra- tions will require use of either drinking tips with smaller diameter holes or some other type of apparatus. The availability of C57BL animals from the long-term study des- cribed above gave an opportunity to test whether the EICS operates after prolonged exposure as it does after short exposure. Ten animals (266±3 days of age) were tested with 5-percent, l0-percent, and 20- percent ethanol solutions. Each concentration was offered for three consecutive days, and each was repeated three times during the course of the experiment. The order of presentation of the different concen- trations was determined independently for each animal by random permu- tations of a 3 x 3 Latin square so that each condition would be balanced in terms of sequence effects. Standard procedures of daily readings with a change in position of the two drinking bottles every three days (coincident with concentration change) were followed. Under the 5-percent choice condition, the mean absolute amount of ethanol ingested over the nine days was l.l4 ml; under the l0-percent condition, it was 2.l8 ml; and under the 20-percent condition, it was 2.38 ml. The values for the 5-percent and l0-percent choice conditions are in very good agreement with previously obtained data. The rela- tively small increment between l0-percent and the 20-percent condition is also consistent with previous results showing that the general logarithmic function descriptive of the EICS is applicable only up to about l2-percent concentrations, after which it flattens or drops. The fact that all these studies revealed a quite orderly rela- tionship between concentration and amount consumed provided the begin- ning of a behavioral description of the EICS. It was also noted that individual C57BL animals showed a remarkable day-to-day stability in their intake level. We therefore conducted an experiment to determine if this stability in quantity of ethanol ingested is attributable to factors associated with taste or with physical aspects of ingestion (such as licking, swallowing, etc.) or to some more central factor, such as a possible pharmacological effect.

149 In this experiment (McClearn and Nichols, l970), 33 C57BL males (70 days of age) were placed in a standard two-bottle choice situation with water and a l0-percent ethanol solution. After daily consumption was recorded for a l5-day period, the animals were divided into experi- mental and control groups in such a way as to equate approximately the mean ethanol intake over the last 6 days. At various times thereafter, the control group received intraperitoneal injections of 0.5 cc of saline solution, while the experimental group received 0.5 cc of 20-percent ethanol in saline solution. The results are shown in Figure l2. PC designates a position change of the ethanol solution and water bottles, and INJ signifies injection. It should be noted that 0.5 cc of a 20-percent solution would equal l.0 cc of a l0-percent solution in terms of absolute amount of ethanol. Thus, the amount injected into the experimental animals was approximately equivalent to one vertical subdivision in Figure l2. It may be seen that the first injection caused a decrease in ingestion of ethanol solution by the experimental animals that was roughly this order of magnitude. Consumption returned toward previous levels on the subsequent two days and was reduced again in about the same amount by the second injection. This again was followed by a partial recovery and a subsequent drop after the next injection. Recovery was very incomplete after the third injection, however, and the drop after the fourth injection was small. A rest period of 8 days intervening before the next injection permitted recovery of ethanol solution ingestion almost to the pre-injection level. Two subsequent injections were followed by reduction and recovery as before. That the effect was not due simply to debilitation is indicated by the water intake curve for the experimental animals. It may be seen that this curve is very nearly the mirror image of the ethanol solution curve, so that the total amount of fluid ingested remained approximately constant over the various conditions. A seriously ill animal would be expected to have reduced fluid intake. The occurrence of two position change days without injection (days 27 and 30) and one injection day without position change (day 36) shows that position change alone does not elicit a drop in ethanol solution intake and that the injection effect does not depend on a coincident change in bottle position. The data for the control group show an effect of saline injection only on the first occasion, indicating that the animals apparently adapted quite well to the injection procedure. Since the normal ingestion route is bypassed when ethanol is introduced by intraperitoneal injection, these data suggest that the gustatory system or physical aspects of ingestion do not play an impor- tant role in the EICS. Some central mechanism apparently is involved, and this mechanism appears to be responsive to absolute quantities of ethanol. One intriguing question about the mode of operation of the EICS concerns the temporal interval over which the regulating or controlling is accomplished. For example, if the system is "forced" by exclusive presentation of ethanol for a certain length of time, how long a period will be required for consumption to return to baseline values? Figure l3 presents the results of such an experiment. The first three

l50 connected points give a pre-treatment reference value for ethanol con- sumption by l5 C57BL mice. Only a l0-percent ethanol solution was available for 6 days thereafter, and it can be seen that intake of eth- anol was clearly above baseline during this period. Results of four successive 3-day periods on a standard two-bottle choice situation are then presented. There is an obvious depression in the first period after forced consumption, and it is not until the fourth period that pre-treatment consumption levels are regained. Excess consumption over the 6-day period had a depressant effect that certainly continued for more than 24 hours. The behavioral studies of the EICS described thus far have shown a close relationship between amount of ethanol consumed and the concen- tration of the solution, and they have suggested that the EICS is a central mechanism responsive to the quantity of ethanol introduced into the system. If the EICS functions in such a way as to restrict daily ingestion to a certain level characteristic of a given concentration, then an increase in total daily fluid intake might be expected to come exclusively from the water bottle. Several methods (temperature in- crease, administration of diuretic drugs, etc.) could be used to in- crease daily fluid intake, but we chose the simple strategem of adding sucrose to both the water and the ethanol solution. C57BL males, ap- proximately l36 days of age when the experiment began, were presented with the standard two-bottle choice between tap water and l0-percent ethanol solution for l5 days. Mean daily intake of fluid was approxi- mately 6.0 ml during this period. On days l6 through 3l, l5 g of sucrose per l00 cc of liquid was added to each bottle. Mean daily fluid intake over this l5-day period was approximately 9.0 ml. Figure l4 shows that the increase in fluid consumption was approximately equally distributed over the two bottles. The mean absolute amount of ethanol ingested during the first l5-day period was 0.4l ml; for the second l5 days, after the addition of sucrose, the corresponding mean was 0.63. To determine if prior adaptation to a sucrose solution would have an influence on subsequent choice of ethanol vs. water, and also to attempt replication of the above results, the following experiment was performed. Two groups of 20 C57BL males, 52 days of age, were used. Group A was treated as in the previous experiment, with a choice be- tween water and l0-percent ethanol on days l-l5 and a choice between water + sucrose and ethanol + sucrose on days l6-30. On days l-l5, Group B was presented with two bottles, both containing the l5 g sucrose per l00 cc water solution. From day l6 through day 30, their choice was between ethanol + sucrose and water + sucrose. The results for Group A, as shown in Figure l5, are directly comparable to the results of the previous experiment. It can be seen that the outcomes were different; in this case, the results are very nearly as would be predicted on the assumption that the determining factor in the EICS is the ingestion of a certain amount of ethanol of a given concentration. Nearly all of the increased consumption after sucrose was added was from the water + sucrose bottle. There is no obvious difference between these two experiments except the age of the

l5l animals. It may be that there is a developmental change in the factors involved. A need for further study is clearly indicated. Figure l6 presents the results for both groups in 3-day blocks. It is clear that Group B animals consumed less of the ethanol + sugar solution than did the Group A animals, in spite of a greater total fluid consumption. A previously established preference for ethanol evidently can persist after the (presumably) more preferred incentive of sucrose is available, but a preference for ethanol cannot develop ab origine during periods when there is a high intake level of the more preferred substance. SOME HINTS AS TO MECHANISMS UNDERLYING THE EICS While work has proceeded on the behavioral description of the EICS, certain results have been obtained that give glimpses of the mechanisms underlying it. One early conjecture (Rodgers and McClearn, l962) was that alcohol preference might be related to ability to meta- bolize ethanol. Therefore, the relationship between strain means for alcohol preference and strain means for alcohol dehydrogenase (ADH) activity was plotted, and a positive relationship was found. It was generally thought that ADH, the enzyme mediating the first stage in the metabolism of ethanol, was the rate-limiting step; nevertheless, it remained to be established that direct measures of metabolism rate would differ among strains. Schlesinger (l963), in a study measuring rate of respiratory excretion of radiolabeled carbon dioxide, found C57BL mice to metabolize ethanol ll percent more rapidly than DBA/2 mice at one dose but not to differ at a higher dose. Thus, the results of this study are in only partial confirmation of the notion that ani- mals with higher preferences may have faster ethanol metabolism rates. Sheppard et al. (l968) subsequently repeated the comparison of ADH levels in C57BL and DBA/2 mice and also extended the observations to include activity of aldehyde dehydrogenase (ALDH), the enzyme mediating the second step in ethanol metabolism. It was discovered that C57BL mice showed 300 percent more ALDH activity and 30 percent more ADH activity than did DBA/2s. In this study, Fl hybrids were also evalu- ated, and their enzyme activity levels were found to be intermediate to those of the parental strains. An hypothesis arising from considerations of ethanol metabolism was that the alcohol-avoiding strains might show rapid metabolism of ethanol to acetaldehyde, but not be able to metabolize acetaldehyde to acetic acid readily. Since acetaldehyde is a toxin, its accumulation might have unpleasant effects and discourage further ingestion. Schlesinger (l963) also investigated this possibility by determining if blood acetaldehyde level was significantly higher in DBA/2s than in C57BLs after ethanol injection. Blood samples were taken l and 2 hours after intraperitoneal injection (l.4 pl/g body weight) and assayed for acetaldehyde by a spectrophotometrie technique. Ten animals of each strain were tested at each time interval, and it was found that DBA/2 animals did accumulate higher levels of acetaldehyde. This observation was replicated by Sheppard et al. (l970).

l52 In addition to these essentially descriptive approaches, there have been several studies in which the effects of biochemical inter- ventions on voluntary ethanol consumption have been explored. Schlesinger et al. (l966) administered Antabuse to animals of the C57BL, RIII, and C3H strains and compared their ethanol consumption with that of untreated controls. Antabuse, as an inhibitor of ALDH, leads to the accumulation of blood acetaldehyde; it is widely used in the treatment of alcoholism. In all tested strains, Antabuse-treated animals showed lower mean ethanol consumption than did their controls. In subsequent studies, Sanders et al. (l976, l977) have investigated the effects of several monomine oxidase inhibitors on ethanol prefer- ence in C57BL/6 mice and have concluded that the reduction in ethanol preference subsequent to administration of these drugs is due to their effect in increasing acetaldehyde after ethanol ingestion. Taken together, these studies are strongly suggestive that circu- lating acetaldehyde levels are an important signal in the EICS of the mice. Although the behavioral and biochemical portraits of this system are still very hazy, enough of an outline is emerging to make possible the generation of hypotheses to be tested directly in studies with human subjects. A PROGRAM FOR FUTURE RESEARCH The work on inbred strains summarized above can perhaps best be characterized as establishing a few benchmarks in preparation for a proper survey of the domain. There are many more arrows in the genet- ics quiver than techniques using inbred strains, and more of these arrows need to be employed. For example, Henry and Schlesinger (l967) demonstrated that a particular type of C57BL mouse, differing from the ones described heretofore only in respect to one gene (albino locus), display a reduced alcohol preference. Are there implications from what is known of the physiological functioning of the albino gene for our description of the EICS? What other single-gene effects might be dis- covered by systematic exploration of those that are available for research? A particularly important disadvantage of inbred strains arises with respect to hypotheses concerning correlated characters. Within a single inbred strain, any covariance between characters must arise from environmental sources. Possible genetic contributions to covariance have been eliminated. Comparison of two inbred strains that differ on attribute A to determine if they also differ on attribute B does not provide strong evidence concerning the causal relatedness of the two traits. Because the process of inbreeding causes fixation of all loci, animals of a particular inbred strain are stabilized geotypically for all traits subject to genetic influence. Therefore, given a difference between two strains on trait A, the expectation is that they will show a significant mean difference with respect to a very large number of other traits that may have no connection whatsoever with trait A. Therefore, our findings that mice of the C57BL and DBA/2 strains differ, for example, with respect to activity of the ADH or ALDH

l53 enzymes, or with respect to accumulated acetaldehyde, are permissive, not persuasive. That is, they permit us to entertain further our hypothesis that alcohol preference might be related to these particular enzyme systems or metabolic products. More efficient correlational procedures become available when properly constituted, genetically heterogeneous stocks are available (McClearn et al., l970). Genetically segregating populations derived from inbred strains, such as F2s or subsequent generations, comprise one sort of heterogeneous stock. An example of the use of such a group in the present context is provided by Anderson (l974), who examined the interrelationships among measures of ADH, ALDH, ethanol preference, and ethanol acceptance (another measure of voluntary consumption). One salient finding was a correlation of 0.25 between ADH activity (ex- pressed in terms of liver weight) and ethanol acceptance. Thus, while the result confirms in a general way that there may be some influence of enzyme activity on ethanol consumption, only about 8 percent of the variation in the latter is "explainable" in terms of variation in the former. This serves as a reminder of the pitfalls in interpreting mean differences among inbred strains. Significant mean differences might actually indicate relatively small effects. Whether the effect is small in the case of the Anderson study depends on perspective. Cer- tainly, only a small proportion of variability in ethanol consumption was relatable to activity of this particular enzyme. On the other hand, if one assumes from the beginning the complexity of the EICS, locating one biological factor (ADH/g liver)—itself due presumably to one or a few genes—that accounts for as much as 8 percent of the vari- ability is noteworthy. Another pitfall is identified in the Anderson (l974) study. The correlation between the measure of ethanol preference and that of eth- anol acceptance is only 0.27. Clearly, the behavioral domain of eth- anol consumption is a complex one, and the particular probes (i.e., indices of consumption) that have been used are very arbitrary and may be representing idiosyncratic features of the total system. The use of multivariate analyses to characterize this behavioral domain and its relationships to pharmacological effects seems to be a particularly powerful approach and one not hitherto exploited in the pharmacoge- netics literature. Finally, it is worth noting one other genetic procedure of great promise in generating animal models for the study of drug-related behavior. Selective breeding offers a means of tailoring animals' attributes to some specified end. By mating together animals of like extreme with respect to some attribute, a shift in that attribute can be accomplished if there is any genetic basis for the variation in the trait in question. Repeated over successive generations, selective breading may lead to very substantial changes in the selected trait. If selection is undertaken in both the high and the low direction, selected lines with a large difference between their means are avail- able for research on mechanisms underlying the trait. Selective breeding has already generated mouse lines differing in sensitivity to the hypnotic effects of ethanol (McClearn and Kakihana, l973) and in ethanol acceptance (McClearn and Anderson, in press); rat

l54 lines differing in alcohol preference (Eriksson, l968) and reactivity to alcohol (Riley et al., l977) have also been generated. These selec- tion studies have all been rather unidimensional in their objectives, seeking to obtain line divergence with respect to a single attribute. By using multivariate procedures to clarify our understanding of the dimensions of drug avidity, drug sensitivity, and drug metabolism as well as the acquisition of tolerance and the development of dependence, we should be able to formulate sophisticated selection objectives and generate animal models of greatly enhanced value. REFERENCES Anderson, S. M. (l974) Ethanol consumption and hepatic enzyme activity. Unpublished Master's thesis, University of Colorado. Eriksson, K. (l968) Genetic selection for voluntary alcohol consump- tion in the albino rat. Science l59:739-74l. Henry, K. R., and K. Schlesinger (l967) Effects of the albino and dilute loci on mouse behavior. Journal of Comparative and Physiological Psychology 63:320-323. Kakihana, R. Y. (l964) Developmental study of preference for and tolerance to ethanol in inbred strains of mice. Unpublished doctoral dissertation, University of California, Berkeley. McCLearn, G.E. (l968) Genetics and motivation of the mouse. In W. J. Arnold (ed.) Nebraska Symposium on Motivation. Lincoln: University of Nebraska Press. Pp. 47-83. McClearn, G. E. (l972) The genetics of alcohol preference. In Proceedings of the International Symposium on Biological Aspects of Alcohol Consumption. Helsinki: The Finnish Founda- tion for Alcohol Studies. Pp. ll3-ll9. McClearn, G. E., and S. M. Anderson (in press) Ethanol consumption: Selective breeding in mice. Behavior Genetics. McClearn, G. E., and R. Kikihana (l973) Selective breeding for ethanol sensitivity in mice. Behavior Genetics 3:409-4l0. (Abstract) McClearn, G. E. and D. Nichols (l970) Effects of intraperitoneal in- jection of ethanol on ethanol ingestion of C57BL mice. Psychomonic Science 20:55-56. McClearn, G. E. and D. A. Rodgers (l959) Differences in alcohol preference among inbred strains of mice. Quarterly Journal of Studies on Alcohol 20:69l-695.

l55 McClearn, G. E., J. R. Wilson and W. Meredith (l970) The use of isogenic and heterogenic mouse stocks in behavioral research. In G. Lindzey and D. D. Thiessen (eds.), Contributions to behavior-genetic analysis: The mouse as a prototype. New York: Appleton-Century-Crofts. Pp. 3-22. Riley, E. P., E. D. Worsham, D. Lester, et al. (l977) Selective breeding of rats for differences in reactivity to alcohol: An approach to an animal model of alcoholism. II. Behavioral measures. Journal of Studies on Alcohol 38:l705-l7l7. Rodgers, D. A. and G. E. McClearn (l962) Alcohol preference of mice. In E. L. Bliss (ed.), Roots of behavior. New York: Hoeber. Pp. 68-95. Sanders, B., A. C. Collins, D. R. Peterson, and B. S. Fish. (l977) Effects of three monoamine oxidase inhibitors on ethanol preference in mice. Pharmacology, Biochemistry and Behavior 6:3l9-324. Sanders, B., A. C. Collins, and V. H. Wesley. (l976) Reduction of alcohol selection in pargyline in mice. Psychopharmacologia 46:l59-l62. " Schlesinger, K. (l963) Genetic and biochemical determinants of alcohol preference and alcohol metabolism in mice. Unpublished doctoral dissertation, University of California, Berkeley. Schlesinger, K., R. Kakihana and E. L. Bennett (l966) Effects of tetraethyIthiuramdisulfine (Antabuse) on the metabolism and consumption of ethanol in mice. Psychosomatic Medicine 28:5l4-520. Sheppard, J. R., P. Albersheim and G. E. McClearn (l968) Enzyme activities and ethanol preference in mice. Biochemical Genetics 2:205-2l2. Sheppard, J. R., P. Albersheim, and G. E. McClearn (l970) Aldehyde dehydrogenase and ethanol preference in mice. Journal of Bio- logical Chemistry 6l:l65-l69.

>_._.-^^ " i . : ^ » 1 4 i i I III "K=CC-' j^ - NjJ: llii Kl 0 K O I 1 1 c _ .- o i- 3 .c *- «.§ 8 ° a> '= .= 15 i~» -2 c i- Q) <U -c 5 ,« S y K O o°S§ I s<:i = _ eu I«s a (DO) .>* ^I-S1^ .C ra O -O .2 g tr » .E c = I S • « & - c^j a; 2 t=d « D ^ O) g -5 c «o t o 3 o o e " — -~ 8og| CO o 3 JO <-. .2 o i O '.P 3J M | fl^ ll^l o 2 € ° .^ <u .£ "0 O " O c -a S « ^ -S1 ° -^ S 03 .^ H- ^ =: c 6 p o <- .£ — <u o (O U *r O ^^ o *" CO ".M ^^ ^ _ ^ 2 S TB 2 5 8 5 §«z = .?'«? .; ai° ?J >- <u O g g I ' I < £ 2 S S . > to y o e jft £ tf 2 o> -D " .£ i> -o ° 'CB-'S il m £ a w l56

l57 1 •• 5 - DBA/2 - J ' • i i i . i i i i i i I i j 1 -i 5 A D u. J r 1 i i i i i i i i i i i i i i i C57BL Kn I MEAN PREFERENCE RATIO I Figure 2. Frequency distributions of mean preference ratios (ratio of consumed alcohol solution to total consumed liquid during a two-week period) for three inbred mouse strains.

l58 I.OOr- .80 0) crv o .60 o> o> •4- 0> c o 0) .20 0-20 Group .x-.-x. -.i- J.-_L._L 9 l1 13 Week 15 17 19 Figure 3. Mean weekly preference ratios of male C57BL mice offered a choice between tap water and l0 percent ethanol solution (0-l0 group) or between tap water and 20 percent ethanol solution (0-20 group) over a l9-week period.

l59 .50 £ o 30- fc -20- .10- .00 10 15 AGE (WEEKS) 20 Figure 4. Mean preference ratios of BALK/c mice as a function of age at the beginning of a two-week test period. Source: Kakihana, R. and McClearn, G.E. (l963). Development of alcohol preference in BALB/c mice. Nature l99: 5ll-5l2. Reprinted by permission.

l60 20. cf cf 18 . Cf 16 ' Cf 14 . cf cf 12. Cf ID- & 16 WEEKS Cf Cf S' Cf 0 6- £j z C( Ul o 3 • 3k O V UJ o ^f £T • 9cf 9 b. 00 9 6 • 4 WEEKS 4 - £ 2 • cf or cf or cfocfcfcf .00- '.21- '.46- '.71- '.96- 0 cTcfcf Q cfc/5 9. cro 9 .05 .25 .50 .75 1.00 PREFERENCE RATIO Figure 5. Frequency of mean preference ratios in a group of BALB/c mice tested for two weeks beginning at 4 weeks of age and again at l6 weeks of age. Source: Kakihana, R. and McClearn, G.E. (l963). Development of alcohol preference in BALB/c mice. Nature l99: 5ll-5l2. Reprinted by permission.

l6l o X o o 4 u S 1.0 0 1.0 ,5 0 1-0 .5- o i.o I.O .9' Bolb/cCrjl i.o .5 O 40 6O 80 100 AGE (DAYS) 120 140 160 Figure 6. Mean preference ratios of individual BALB/c mice that displayed episodes of high alcohol intake during a l50-day observation period.

l62 xs- 0* CBTiL ALCOHOL MKFIIUNCC RATIOS x)«A8t AT DAY ONE •a, 40 •o OAT* Figure 7. Mean preference ratios of individual C57BL mice that displayed changes in alcohol intake during a 200-day observation period.

5.0 .-S4JO 3.0 Q. S 10 .80 .60 .40 .20 Ethonol Solution I H oo oo — CVJ —CM II II oo oo CVJCVJ oo oo OJCVJ ii it OO OO — cvj — cvj Preference Ratio Water oo oo oo oo — cvj —cvr cvjcvj oo oo oo oo CVJCVJ —CVJ —CVJ Absolute Ethanol .r Total Water (read ordinate as ratio) n i Total Liquid I OO OO — cvj —cvj OO OO -- cvjevj OO OO OO OO — — cvjcvj —cvj —cvj Absolute Amount Ethanol (read ordinate in ml.) I Q 00 oo oo oo oo oo oo oo — cvj —cvj cvjcvj —cvj —cvj cvjcvj 1 i i i i i i i i i i i i i i i OO OO OO OO OO OO OO OO cvjcvj —cvj —cvi cvjcvj —cvj —cvj OO 00 00 OO — cvj —cvj cvjcvj ii ii ii ii OO OO OO OO CVJCVJ —cvj —cvj Figure 8. Various consumption measures for C57BL mice of four groups (l0-l0, l0-20, 20-l0, and 20-20) during two observation periods (I and II) as described in the text Source: McClearn, G. E. (l968). Genetics and motivation of the mouse. In W. J. Arnold (Ed.), Nebraska symposium on motivation. Lincoln: Univer- sity of Nebraska Press, p. 69. Reprinted by permission. l63

l64 Total liquid — Ethonol solution Ethanol Water Water from ethanol solution Ethanol 24 1.5 Concentration (per cent by volume) Figure 9. Liquid consumption measures for male C57BL mice offered a choice between tap water and various concentrations of ethanol solution during preshift and postshift periods as described in the text. Source: McClearn, G.E. (l968). Genetics and motivation of the mouse. In W.J. Arnold (Ed.), Nebraska Symposium on Motivation. Lincoln: University of Nebraska Press, p. 7l. Reprinted by permission.

l65 § o S ^ o in A e tn 0> O Preshift • Postsh.-t 1.5 3 6 Concentration 12 24 Figure l0. Absolute amount of ethanol consumed by male C57BL mice as a function of concentration of the ethanol solution during preshift and postshift periods as described in the text. Source: McClearn, G.E. (l968). Genetics and motivation of the mouse. In W.J. Nebraska Symposium pjn Motivation. University of Nebraska Press, p. by permission. Arnold (Ed.), Lincoln: 72. Reprinted

l66 Total liquid - —— Ethanol solution Ethanol Water Water from ethonol solution Ethonol 12 24 3 Concentration (per cent by volume) 12 24 Figure ll. Liquid consumption measures for male C57BL mice as a function of concentration of ethanol solution and previous experience. In the postshift results, the circled numbers identify the mean levels of the various groups currently being tested at the concentration shown on the abscissa. Source: McClearn, G.E. (l968). Genetics and motivation of the mouse. In W.J. Arnold (Ed.) Nebraska Symposium on Motivation. Lincoln: University of Nebraska Press, p. 74. Reprinted by permission.

l67 10 35 Figure l2. Mean consumption of tap water and l0 percent ethanol solution by C57BL mice in groups receiving injections of 20 percent ethanol in saline solution (experimental) or saline solution alone (control) at various times from Day l5 through Day 36 of a 38-day observation period. PC designates a position change of the ethanol solution and water bottles; INJ signifies injection.

l68 4- J 3 2 *-. 1 2 - *A ALCOHOL BLOCK Figure l3. Mean consumption of l0 percent ethanol solution by C57BL mice before, during, and after six days of forced alcohol consumption. Source: McClearn, G.E. (l972). The genetics of alcohol preference. In Proceedings of the International Symposium on_ Biological Aspects of_ Alcohol Consumption. Helsinki: The Finnish Foundation for Alcohol Studies, p. ll3-ll9. Reprinted by permission.

l69 4 o u e> 2 Ethanol + sugar + sugar.—\ Period Figure l4. Mean consumption of tap water and l0 percent ethanol solution by male, l36-day-old C57BL mice during l5-day periods before (Period I) and after (Period II) the addition of sucrose to both the water and the ethanol solution.

l70 o £2 =! I _l tthonol •thonol * sugar^ II PERIOD Figure l5. Mean consumption of tap water and l0 percent ethanol solution by male, 52-day-old C57BL mice under conditions comparable to those described for Figure l4.

12 II 10 9 8 o° o co 5 ec UJ §4 2 I 0 T- . -T- w ,s W- WATER A - ALCOHOL 8-8UOM T- TOTAL GROUP A 345678 BLOCKS OF 3 DAYS 10 Fiqure l6 Mean li(1uid consumption measures for male, 52-day-old C57BL mice in a group treated as described for Figure 14 (Group A) and 1n a group exposed to sucrose solution alone during the initial l5-day period (Group B). l7l

RESPONSE TO "ANIMAL MODELS AS PHARMACOGENETIC TOOLS" William Meredith First of all, let me remind you that all of Professor McClearn's remarks illustrate the significance and centrality of individual dif- ferences not only in behavior but in biochemical and biological reac- tions as well, in the study of response to and control of substance use. Although researchers usually profess to be interested in individ- ual differences, in practice they are treated as an embarrassment or a source of error. Animal researchers often neglect individual differ- ences entirely. Yet all the problems of drug and substance abuse with- in a given cultural context are essentially problems of individual dif- ferences. You might say that many are called but few are chosen. For example, in our society nearly all of us are at risk for alcoholism or are perhaps at risk for teetotalism. Yet relatively few become alco- holics or tetotallers. If we are to understand this problem we must investigate the basic domain of relative individual variability both in the social and environmental precursors of a given set of behaviors and in their biological substrates. Ultimately, we must learn how life history variation and biological variation interact. I want to take up first Professor McClearn's points that behav- ioral variables that could be employed in evaluating a response to a drug are manifold, and seemingly related or similar behaviors or bio- logical indices, such as enzyme activity, often turn out to be statis- tically independent. This points up, on one hand, the importance of having reliable measures of the attributes under study, inasmuch as unreliability can seriously statistically attenuate an otherwise sig- nificant relationship. Furthermore, a population must show a substan- tial degree of variability with respect to the attributes studied if significant relationships are to be obtained. I'm not speaking about experimental intervention and the kinds of relationships you get in it but simply naturally occurring kinds of relationships. The thrust of Professor McClearn's observations seems to me to suggest that behav- ioral variables and related physiological and biochemical variables should be studied in their full roultivariate complexity whenever pos- sible instead of in isolation. There are, of course, by now an enor- mous variety of statistical techniques to utilize in such studies— for example, multivariate analysis of variance; multivariate analysis of variance with step-down procedures that sequentially partial out the l72

l73 influences of different variables; and multivariate analysis of vari- ance with canonical variates. I'd like to suggest how multivariate model theory, specifically structured factor analysis, can reveal use- ful structure in a body of data. Why be interested in factor analysis in a context like this? Well, it seems to me that often the kinds of variables we choose to observe are but pale reflections of what we suppose might actually be going on inside an organism. Take, for example, acetyldehydrogenase activity in the liver. You must sacrifice an animal, remove the liver, and go through a bioassay technique. You can't repeat the observation on that animal and you cannot, at least with present technology, moni- tor enzyme activity during the course of alcohol injection. When we look at a variable like enzyme activity we're looking at individual differences in enzyme activity in a really gross way. That is, if you can imagine that the organism can produce a greater or lesser amount of that particular enzyme on demand when it's needed then we're not really getting at that kind of enzyme activity. What we hope, I suppose, when we remove a liver and get enzyme activity is that what we observe there is correlated with the kinds of processes that we think might be going on during the course of the activity under study. But again, for alco- hol preference, alcohol acceptance, etc., there are a number of dif- ferent situations that one could imagine or devise to evaluate such processes, and they're likely to give different results. It is attractive, at least, to imagine that underlying a complex of variables like this there may be some sort of behavioral, biologi- cal, or psychological reality. It is at least the pious hope for us factor analysts that using factoring techniques we can reveal some parts of that underlying biological reality. A lot has been happening in this field that is very interesting and very applicable to these kinds of studies. There's an enormous explosion in the development of new techniques for fitting very strongly specified, very strongly structured factor analytic models longitudinally; that is, studying the same individuals at different times during their lives, or fitting the models to the same individual studied under different conditions, or fitting them to different individuals chosen from different populations at random. I'll mention two examples, neither of which has anything to do with drug abuse, but they illustrate the point that I want to make. In the area of cognitive psychology, we have by now a reasonably well-developed theory about some of the higher-order factors underlying cognitive abilities, due to John Horn and Ray Cattell. Their model calls for three factors called crystallized intelligence, fluid intel- ligence, and general speediness. Given a particular battery of meas- ures, we know reasonably well how to define the nature of these three higher-order factors. We have collected data in Hawaii from Caucasian families, Chinese families, and part-Hawaiian families. We have col- lected data in Korea from Korean families using the same test battery translated into Korean. We're finding that the hypothesized fluid, crystallized, and general speediness factors emerge in all of these populations in almost exactly the same form unless you care to quibble about second and third decimal points. It's really quite amazing how

l74 similar the factor structures are and how supportive they are of the Horn-Cattell model. Having seen this kind of analysis of data implies to me that the model of fluid intelligence, crystallized intelligence, and general speediness in intelligence and cognitive abilities is more real than it was before. These factors have great stability from one population to another, from one cultural context to another, and I think it's safe to assume that we're getting some sort of a handle on an underlying, per- haps biological reality. A further implication is that the kinds of pathways that serve to unite variables into these factor composites are the same in a variety of different cultural contexts, despite differ- ences in education, educational style, educational level, and family background, etc. Another example is a study that Professor McClearn and I collabo- rated on along with several other people, in which we had a number of different measures of activity in what is called a diallel-cross exper- iment, starting out with seven inbred strains and all possible Fl generations derived from those inbred strains; we had 98 groups of animals. Now, the experiment is such that we can calculate an environ- mental variance-covariance matrix, a genetic variance-covariance matrix, and the total variance-covariance matrix for a set of 30 or 40 variables. We get a set of apparatus-specific factors, reflecting the multiple measurements from the same apparatus or test situation plus two superordinate or higher-order general factors that I think we could agree on labeling emotionality and something like exploratory activity or exploratory drive. So here are two superordinate factors of explor- atory activity and emotionality. There are as many apparatus-specific factors as there are different apparatuses. These factors appear in both the genetic and environmental variance-covariance matrixes and in exactly the same form, again, give or take some fiddling with second or third decimal points. Another interesting point is that the apparatus- specific factors have a much larger proportion of variances that are environmental than is the case of the two superordinate factors. This is fairly natural in that all the measures for a given apparatus are obtained on the same day. So whatever day-to-day variation in mood or temperament or temperature might be affecting animal behavior will affect all the scores on the apparatus. More important, this suggests that the biological pathways underlying these behaviors are shared. The genes affect these pathways; they don't affect behavior themselves, directly. The next step one would suppose would be to find and iden- tify some of the biological pathways that underlie this kind of activ- ity. Certainly, the same thing could be done in any sort of research on substance or control. Now, with the issue of selective breeding there are a couple of ideas that arise naturally. There are three important things that one can do with selective breeding studies or with a series of selective breeding studies. One is to look at correlated response. If you are selectively breeding for alcohol preference or alcohol acceptance you can in each generation or every third generation or so examine some of the animals for sleeping time. (For a more complete discussion, see G. E. McClearn and S. M. Anderson. Drug and Alcohol Dependence l979,

l75 4:6l-76.) You can do this after they've been tested for alcohol ac- ceptance and have been bred, or it could be done with siblings of the animals that were used in actual breeding processes. The question is, is selective breeding for alcohol acceptance dragging sleeping time along after it? In a regular and predictable way? If it is, that says something about the fact that there is some underlying biological mech- anism that they share. If the one isn't being carried along, they are independent. The second kind of experiment which would be interesting to do, I should think, after having selected animals for certain characteris- tics, like alcohol acceptance, is to then ask yourself, how has the acceptance response to injection of alcohol been affected by selection? There is evidence showing how the acceptance of alcohol is moderated by injection in a particular series of strains. What about after you've done selective breeding for alcohol acceptance? At this point we begin getting a handle on the kinds of interaction that might be present in the system. Finally, I'd like to propose the idea of doing a particular experiment; I think really Professor McClearn implicitly proposed it. To the best of my knowledge this sort of study has never really been done. I have coined a term for it. I would call it high/low-low/high simultaneous double selection. Consider an attribute like acetaldehyde dehydrogenase activity in the liver. (Well, that's a little problem- atic because you have to sacrifice one animal to get the alcohol dehy- drogenase activity. You'd have to use a sibling as the animal for breeding). But then simultaneously select for that and, let's say, that alcohol senstivity as measured by sleep time. If you can break them apart you should be able to simultaneously select high/high, high/low, low/high, or low/low and get four different lines. If you can break them apart, different biological pathways are involved. If you cannot break them apart then you know there's a linkage in the genetic sense. This kind of experiment could be an extraordinarily useful technique for exploring genes and/or biological mechanisms.

DISCUSSION OF "ANIMAL MODELS AS PHARMACOGENETIC TOOLS" Kurt Schlesinger My task in this report is to discuss Professor McClearn's paper on Animal Models as Pharmacogenetic Tools. It is a pleasure to do so, and for a number of personal and scientific reasons. Personally, it is very gratifying to discuss this paper because Dr. McClearn was my major professor during my tenure as a graduate student, and because I was his research assistant when some of the early work on the genetics of alco- hol preference was being performed. Scientifically, it is very pleas- ing to discuss this research because idiosyncratic drug reactions that depend on an individual's genotype are very widespread, and in the past such individual differences in the response of organisms to drugs have not been given the attention they deserve. Pharmacogenetic studies have the potential of telling us much about the mechanisms that mediate many drug-related behaviors. Dr. McClearn's presentation and paper focus on what can be des- cribed as individual differences in alcohol preference in mice of sev- eral inbred strains. In particular, the paper outlines some experi- ments that deal with a possible alcohol intake control system by which these animals limit the amount of ethanol they consume when placed into a two-bottle free-choice situation. He also discussed some of the pos- sible mechanisms that might mediate genetically determined differences in alcohol preference behavior. Before specifically discussing alcohol preference behavior, I would like to say a few words about pharmacogenetics in general. The first point that I would like to stress as strongly as possible is that heterogeneity in the response of animals to drugs is a biological fact of life. Goldstein, Aronow, and Kalman comment on the extent of this variability in their textbook titled Principles of Drug Action (l974). In fact, they devote an entire chapter to a discussion of pharmacoge- netics and drug idiosyncracies, and they give many examples of such variability. Let me briefly illustrate the extent of this hetero- geneity by citing two examples from their chapter. One example is an experiment in which guinea pigs were pretreated with an antihistamine drug; the animals were then given a lethal dose of histamine. The results were that the dose required to protect 80 percent of the animals was fully l0 times greater than that required to protect l3 percent of the guinea pigs. l76

l77 A second example, and a rather surprising one, is the response of human patients to digitoxin. In this study l0 patients suffering from atrial fibrillations were given digitoxin and the dependent measure was a 40-50-percent decrease in heart rate. The results were as follows: Some patients showed this response when they were given 6 micrograms of the drug per kilogram of body weight, whereas other patients required as much as 50 micrograms of the drug per kilogram of body weight to exhibit a similar response. There are many examples of such heterogeneity; animals are indeed variable in their responses to a wide range of pharmacological agents, including those that act on the central nervous system. The point, however, and Goldstein et al. stress this point in their textbook, is that such idiosyncratic drug reactions should not be thought of simply as unexpected responses. Rather, they should be considered as drug effects that are altered in genetically variant animals. With respect to genetically determined differences in alcohol preference behavior, I would like to begin my discussion by emphasizing the following general points: First, the amount of alcohol that ani- mals of the high preference strains consume is anything but trivial. Calculated on a per unit body weight basis, and making the average human weight about 75 kilograms, then these mice consume the human equivalent of approximately 2 fifths of l00 proof whiskey per day. At the same time it is necessary to point out that the metabolic rate of mice is considerably greater than that of human beings, and that mice consume substantial amounts of ethanol. Second, these mice consume large quantities of alcohol in free-choice situations, which all text- books of psychology refer to as methods for studying the motivation of animals. Thus, one can consider McClearn's experiments as studies of motivation or, more specifically, of genetically mediated specific appetites. Third, McClearn's studies show that alcohol preference is determined, at least in part, by genetic factors, findings that have been substantiated by studies in many other laboratories. The evidence that makes such a strong interpretation possible comes from experiments that have utilized a wide variety of genetic techniques: These include (l) strain comparison studies and subsequent breeding experiments, (2) selective breeding studies, and (3) single gene studies. Nearly all of this research has been performed on two species of rodents, the rat and the mouse. Quite naturally, this raises the ques- tion of whether this research tells us anything about human alcohol consumption, particularly about the etiology of human alcoholism. One similarity between alcohol preference behavior in animals and human alcoholism is the fact that in animals the behavior is determined in part by genetic factors and there now exists a substantial literature to suggest that this may also be the case in human beings. Rutstein and Veech (l978) have recently reviewed the evidence concerning the inheritance of alcoholism; this evidence comes from (l) twin studies, (2) family correlation studies and, (3) studies of adopted children. One of the more recent and convincing studies of this type was reported by Schuckit (l972). In this study the incidence of alcoholism was studied in male half-siblings from marriages between alcoholic and non- alcoholic spouses. The findings were that the incidence of alcoholism

l78 was significantly greater in half-siblings who had an alcoholic parent and that the incidence was not increased by living with that alcoholic parent. These studies are very suggestive that genetic mechanisms might be involved in the etiology of alcoholism. Such experiments do not prove this point, but in any event it is at least possible that both animals and human beings inherit the propensity to drink alcohol from their parents. Another important question in pharmacogenetic studies of alcohol preference is how one thinks of preference and avoidance. That is, whether one thinks of these two behaviors as oppostie sides of the same coin or as two quite different behaviors. If one thinks of preference and avoidance as the opposite sides of the same coin then one would search for a single mechanism to explain both behaviors. On the other hand, if one thinks of the two as separate phenomena then one would be forced to look for at least two explanatory hypotheses. With respect to this last issue, consider the following example: Assume that one reason that C57BL mice consume large quantities of alcohol in a two-bottle free-choice test is that they are able to metabolize ethanol very efficiently. This seems a possibility since, as McClearn and others have shown, mice of this genotype have approxi- mately 30 percent higher alcohol dehydrogenase activity in liver than do avoiding DBA/2 animals. This enzyme catalyzes the rate-limiting reaction in the oxidation of ethanol, converting it to acetaldehyde. But even if this were the case, it does not necessarily follow that DBA/2 mice avoid alcohol because they metabolize it less efficiently. Other possibilities exist: For example, a number of experiments have shown that C57BL mice have about 300 percent more acetaldehyde dehy- drogenase activity in liver than do DBA/2 animals. This enzyme cata- lyzes the oxidation of acetaladehyde to acetic acid. Therefore, it is possible to speculate as follows: DBA/2 mice have low acetaldehyde dehydrogenase activity and it is possible that mice of this genotype accumulate acetaldehyde even after drinking only small amounts of eth- anol. If this were the case, and since acetaldehyde is highly toxic, its accumulation could render DBA/2 mice sick teaching them to avoid substances that smell and taste like ethanol. The differences in acetaldehyde dehydrogenase activity between C57BL and DBA/2 mice are quite large; it even seems possible that two different proteins catalyze the oxidation of acetaldehyde in animals of these two strains. Examination of the kinetic data published by Sheppard et al. (l970) are in accord with such a hypothesis. In addi- tion, Schlesinger et al. (l966) have found that Antabuse, an inhibitor of acetaldehyde dehydrogenase, is a much more potent drug in C57BL than in DBA/2 mice. In fact, a given dose of the drug inhibits this enzyme twice as much in C57BL than in DBA/2 mice. These data are also in accord with the hypothesis that two different proteins catalyze the oxidation of acetaldehyde in animals of these two genotypes. McClearn has reported on the results of several experiments in which he has studied the consumption of alcohol over very long periods of time. One might have expected that long-term consumption of ethanol might result in tissue tolerance and that the animals might consume

l79 larger and larger amounts of the drug as they gain more and more expe- rience with it. However, this does not seem to occur. McClearn's data indicate that animals consume rather characteristic amounts of absolute ethanol pretty much regardless of their experience with it. This does not mean that tissue tolerance to alcohol does not develop; in fact, quite the opposite seems to be the case. Some years ago, Schlesinger et al. (l966) reported on the results of an experiment in which the activity of alcohol dehydrogenase was measured in the livers of C57BL mice that had been fed a l0-percent solution of ethanol for a period of 90 days. Control animals were maintained under similar laboratory conditions except they were given tap water as their only liquid choice. The results of this experiment were as follows: The activity of alcohol dehydrogenase in the animals that had been maintained on alcohol was significantly greater than that of animals that drank only tap water. However, we also observed an increase in total liver proteins in the animals that had experience with ethanol. Therefore it seemed possible that the increase in liver alcohol dehydrogenase activity that we had observed was merely a secondary reflection of the total increase in liver protein. In an attempt to resolve this question the activity of other enzymes not directly involved in ethanol metabolism was measured. For example, we measured the activity of hexokinase and found that alcohol consumption had no significant effect on the activity of this enzyme. In summary, our results convinced us that the increase in alcohol dehydrogenase activity was not merely a secondary reflection of the increase in total liver protein, but was due to the experience of these animals with alcohol. The increase in alcohol dehydrogenase activity that we had observed in the animals with experience with ethanol was of the order of 20 percent. In order to ascertain whether this increase in enzyme activity was of any physiological significance we measured the rates of ethanol metabolism in animals with and without experience with alco- hol. To perform this experiment we injected animals with radioactively labelled alcohol and measured the rates of appearance of l^C02 with a vibrating reed electrometer. Our results showed that the animals with experience with alcohol metabolized ethanol significantly more quickly than did our control subjects. Thus, we were able to show that experience with alcohol does indeed increase an animal's ability to metabolize the substance, probably because of an increase in the activity of alcohol dehydrogenase. A similar question has been asked of human alcoholics. Do such individuals who have had a great deal of experience with alcohol also metabolize ethanol more quickly than do individuals without such experience? In human subjects this is not an easy question to answer experimentally, for the following reasons: Experience with alcohol causes liver damage, and in order to determine whether drinking alcohol increases the rate at which individuals metabolize this substance one would like to know the rates at which ethanol is metabolized per func- tioning liver cell. In a very clever experiment in l965 Mendelson and his colleagues were able to show that experience with alcohol does

l80 indeed increase the rates of ethanol metabolism. The way this experi- ment was performed was as follows: Alcoholics were placed on a closely controlled hospital ward and they were not given alcohol for a period of time. Alcohol metabolism was then measured in exactly the same way described in the mouse experiments. The patients were injected with trace amounts of radioactively labelled alcohol and the rates of ap- pearance of radioactive carbon dioxide was used as a measure of alcohol metabolism. When these measurements had been made, the second phase of the experiment was begun. During this phase the subjects were fed sub- stantial amounts of alcohol for a period of 7 days. At the end of this time each subject was again injected with trace amounts of radio- actively labelled alcohol and metabolic rates were again determined on the basis of the rates of appearance of radioactive carbon dioxide. Rates of metabolism were compared for each individual subject and the results were as follows: Alcohol metabolism was significantly more rapid after experience with alcohol than when the subjects were dry. Since subjects served as their own controls, the issue of liver damage as a confounding variable in the experiment was eliminated. Thus, both human beings and mice metabolize alcohol more quickly when they have had experience with the drug. In the case of mice this increased metabolic rate is probably due to the increase in liver alco- hol dehydrogenase activity subsequent to the ingestion of alcohol for prolonged periods of time. Whether this is the mechanism responsible for the increased rate of alcohol metabolism in the case of human beings is not known. In neither case do we know whether the increase in alcohol dehydrogenase activity represents true enzyme induction, but experiments that would allow us to answer this question are certainly possible with animal subjects. In summary, McClearn's experiments represent an attempt to under- stand genetically determined individual differences in alcohol prefer- ence behavior in inbred and selectively bred mice. From these experi- ments two very important conclusions have emerged: Namely, that alco- hol preference behavior is determined in large measure by genetic vari- ables and, second, that many genes are involved in mediating these behaviors. With respect to the second point, namely that polygenic systems underlie alcohol preference behavior, the following considera- tion emerges: Since alcohol preference and avoidance behaviors are controlled by many genes obviously interacting with environmental fac- tors, no single variable will explain why mice of certain strains drink alcohol whereas animals of other genotypes avoid this substance. REFERENCES Goldstein, A., L. Aronow and S. Kalman (l974) Principles of Drug Action: The Basis of Pharmacology, 2nd Edition. New York: John Wiley and Sons. Rutstein, D., and R. Veech (l978) Genetics and Addiction to Alcohol. New England Journal of Medicine 298(20).-ll40-ll4l.

l8l Schuckit, M. (l972) Family history and half-sibling research in alcoholism. Annals of New York Academy of Science l97:l2l-l25. Schlesinger, K., R. Kakihana, and E. L. Bennett (l966) Effects of tetraethylthiuramdisulfine (Antabuse) on the metabolism and consumption of ethanol in mice. Psychosomatic Medicine 28:5l4-20. Sheppard, J. R., P. Albersheim, and G. E. McClearn (l970) Aldehyde dehyrogenase and ethanol preference in mice. Journal of Biological Chemistry 6l:l65-l69.

ANIMAL MODELS AS PHARMACOGENETIC TOOLS; GENERAL CONFERENCE DISCUSSION Charles P. O'Brien In his formal presentation, McClearn presented an animal model that he modestly claimed was "not completely without relevance" to the human interaction with alcohol. The model was not presented in a nature-versus-nurture dichotomy, but rather as a genetically determined baseline on which environmental influences can act. Using inbred strains of mice, McClearn and colleagues developed strains with con- sistent relatively high preference for alcohol. Those in the high- preference strains seemed to "sense" intake of absolute alcohol and regulate their intake as a logarithmic function of concentration of alcohol in the solution offered them. Their intake of absolute alcohol could be proportionately reduced by pre-treatment with intraperitoneal alcohol injections or by injections of drugs with sedative properties such as diazepam. In the long-term studies, McClearn showed data in- dicating that the quantity of alcohol ingested by the high-preference mice was remarkably consistent over time. (This is in contrast to studies of alcohol self-administration in human alcoholics in whom an episodic pattern is usually observed). During the general discussion of McClearn's paper, the group dis- cussed the possibility of testing humans in advance of exposure to al- cohol to determine who would be likely to become alcoholic. McClearn stated that work on the membrane effects of alcohol was under way and perhaps an accessible system such as the skin or red blood cells might be used for such a predictive test if differences in alcohol membrane effects can be specified. It was pointed out that while the quantity of alcohol consumed by the high-preference mice was not trival, it could not really be com- pared on a weight for weight basis with quantities in humans. Mice are high metabolizers and typically require a high mg/kg drug does to get effects seen with smaller doses in humans. It was again emphasized that genetic factors can provide the baseline on which environmental factors operate. McClearn noted that the blood alcohol levels in the high-preference mice with spontaneous drinking were 60-80 mg/kg, not really intoxicated levels, but they are pharmacologically significant. There followed a long discussion about the legal definition of intoxi- cation and how strict adherence to blood alcohol levels in defining intoxication is not always helpful in light of individual variations in l82

l83 tolerance. Some sensitive individuals may be unfit to operate a vehi- cle at quite low blood alcohol levels. A question was raised concerning the specificity of the low- preference/high-preference differentiation. Could it be a taste phenomenon? McClearn responded that bitter and sweet taste studies were in progress to determine the effects of taste on preference. He also noted that diazepam injections reduce intake in high-preference mice, suggesting that the signal is based on neurological effects rather than taste. It was pointed out that taste versus effects as a signal could be studied in animals with an esophageal fistula. There was a discussion of conditioned taste-avoidance phenomena in the low-preference strains, a bait-shyness effect studied by Garcia and colleagues. McClearn pointed out that the low-preference mice showed aversion to alcohol with no prior exposure to its taste or effects. It was then pointed out that there are enormous individual variations not only in drug responses but in most responses to bacteria and viruses and in reactions to all environmental toxins studied. The possibility of a biological definition of alcoholism was raised. It was pointed out that identification of phenylketonuria (PKU) as one subcategory of alcoholics could be biologically identi- fied. This concept was extended to nicotine during the discussion. Could a biological test be developed to identify smokers who would have less difficulty than others in giving up their addiction? The discussion closed on the topic of clinical genetic strate- gies. The use of a flagged proband in studying the relationship of two conditions was described. This technique was successfully followed in identifying hyperkenetic children and showed that the incidence of al- coholism in their first-degree relatives was significantly increased compared with control subjects.