Prenatal exposure to alcohol does not affect radial maze learning and hippocampal mossy fiber sizes in three inbred strains of mouse

It has long been known that prenatal exposure to alcohol can have devastating effects. In 1968, Lemoine et al. published a paper, in French, that described "des anomalies dans les infants de parents alcooliques" [1]. Five years later Jones and Smith [2] followed with their now classic paper on Fetal Alcohol Syndrome (FAS), which is the name given to a group of physical and mental birth defects that are the direct result of a woman's drinking alcohol during pregnancy. These defects may include mental retardation, growth deficiencies, central nervous system dysfunction, craniofacial abnormalities, and behavioral maladjustments (for an enumeration, see, among others, [3], and http://www.acbr.com/fas/index.htm).

Since then literally thousands of studies have been done and substantial advances in the understanding of FAS have been made, not in the least because of animal models, among them the mouse [4, 5]. These models strongly parallel the physiological responses to alcohol in human development and have led to valuable insights into the susceptibility to alcohol of specific parts of the developing embryo. One of the more vulnerable brain regions, for instance, is the hippocampus, which is believed to be a primary target of prenatal alcohol and therefore responsible for many neurobehavioral abnormalities [6]. Thus, Riley et al. [7] found the effects of prenatal alcohol exposure on behavior to be similar to those of hippocampal lesions, while hippocampal dysfunction as a consequence of prenatal alcohol exposure has also been assessed in spatial learning tasks such as the Morris water navigation task (see, among others, [8]). However, little attention has been paid to the genetic vulnerability of the developing embryo to prenatal alcohol exposure, particularly genetically determined differences in the sensitivity of the developing hippocampus. These genetically determined differences might, at least partly, explain one of the major questions in the etiology of FAS: why only a small percentage of alcoholic women give birth to children with fetal FAS, whereas other alcoholic women who drink the same amount do not.

The aim of this study was to determine whether different genotypes with varying hippocampal anatomy are oppositely affected by similar exposures to prenatal alcohol. To this end, pregnant females of three highly inbred mouse strains, C57BL/6J, BALB/cJ, and DBA/2J, which vary in hippocampal anatomy (e.g. [9, 10]), were exposed to a 12% alcohol solution as their only source of liquid throughout gestation. Following this, the male offspring was tested for their capacity to master an eight-arm radial maze, a spatial navigation task well known to depend on an intact hippocampus. Subsequently, the animals were sacrificed and the sizes of the hippocampal intra- and infrapyramidal mossy fiber (IIPMF) terminal fields were determined. Necessary control groups (untreated and pair-fed) were included.

Figure 1 depicts the number of total errors (summed from day 3 up to day 5). Only the strain origin affected this variable (F_2,96 = 5.0; p < 0.01). C57BL/6J made fewer errors than BALB/c (z = 3.2; p < 0.01) and DBA/2 (z = 2.5; p < 0.05). Treatment had no effect, neither alone, nor in interaction with strain. Figure 2 shows the running speeds in the radial maze. Similar to the number of errors, only the strain origin affected this variable (F_2,96 = 13.9; p < 0.001) with again C57BL/6 males differing from the other two strains (vs. BALB/c: z = 5.0; p < 0.001; vs. DBA/2: z = 4.8; p < 0.001).

Body and brain weights are presented in Figures 3 and 4, respectively. The origin of the strain influenced both variables (body weight: F_2,96 = 104.3; p < 0.001; brain weight: F_2,91 = 86.1; p < 0.001). BALB/c weighed more than C57BL/6 (z = 3.1; p < 0.01) and DBA/2 (z = 14.2; p < 0.001). The latter strain, in turn, weighed less than C57BL/6 (z = 7.2; p < 0.001). C57BL/6 males had higher brain weights than BALB/c (z = 5.0; p < 0.001) and DBA/2 (z = 11.7; p < 0.001). In addition, BALB/c had higher brain weights than DBA/2 (z = 9.1; p < 0.001). Prenatal exposure to alcohol affected both variables. Body weights were affected independently of the origin of the genotype (F_2,96 = 6.6; p < 0.01) with mice exposed to ethanol prenatally weighing less than control (z = 3.5; p < 0.01) and pair-fed (z = 3.2; p < 0.001) animals. The effect on brain weight depended on the background of the strain (strain × treatment: F_4,96 = 3.8; p < 0.01). This interaction effect was mainly caused by the BALB/c strain where prenatal alcohol decreased brain weight (vs. pair-fed: z = 3.6; p < 0.001; vs. control: z = 3.1; p < 0.01), whereas no significant effects were seen in the other two strains.

The results of the hippocampal data are presented in Table 2 and Figure 5. Strain effects were observed for the following variables: regio inferior (F_2,73 = 8.0; p < 0.001), stratum lacunosum-moleculare (F_2,73 = 18.9; p < 0.001), stratum radiatum (F_2,73 = 5.6; p < 0.01), suprapyramidal MF (F_2,73 = 6.0; p < 0.01) and IIPMF (F_2,73 = 132.0; p < 0.001). DBA/2 showed a smaller regio inferior than C57BL/6 (z = 3.7; p < 0.001) and BALB/c (z = 2.9; p < 0.001). C57BL/6 showed a larger stratum lacunosum-moleculare than DBA/2 (z = 6.1; p < 0.001) and BALB/c (z = 4.3; p < 0.001) while BALB/c males had a larger stratum lacunosum-moleculare than DBA/2 males (z = 2.4; p < 0.05). C57BL/6 exhibited smaller stratum radiatum and suprapyramidal MF than DBA/2 (z = 3.3; p < 0.01 and z = 3.4; p = 0.001) and BALB/c (z = 2.4; p < 0.05 and z = 2.8; p < 0.01). C57BL/6 showed larger IIPMF sizes than DBA/2 (z = 16.2; p < 0.001) and BALB/c (z = 10.8; p < 0.001). BALB/c had larger IIPMF sizes than DBA/2 (z = 7.0; p < 0.001).

Strain/Treatment	Regio inferior	Stratum oriens	Stratum pyramidale	Stratum radiatum	Stratum lacunosum-moleculare	Hilus	Suprapyramidal Mossy Fibers
C57BL/6J
Ethanol	753.9 ± 0.4	35.0 ± 0.05	15.2 ± 0.6	24.6 ± 0.1	8.1 ± 0.4	8.3 ± 0.9	8.7 ± 0.1
Pair-fed	775.4 ± 24.3	33.6 ± 0.8	14.7 ± 0.4	26.1 ± 0.4	7.9 ± 0.2	9.4 ± 0.2	8.5 ± 0.2
Control	747.8 ± 30.6	33.6 ± 0.5	14.8 ± 0.2	25.6 ± 0.6	8.3 ± 0.4	9.7 ± 0.3	8.3 ± 0.2
BALB/cJ
Ethanol	726.9 ± 18.5	34.5 ± 0.6	14.7 ± 0.2	26.6 ± 0.2	6.5 ± 0.2	8.4 v 0.3	9.3 ± 0.4
Pair-fed	742.9 ± 27.7	33.6 ± 0.9	15.2 ± 0.3	26.1 ± 0.5	7.2 ± 0.3	7.9 ± 0.3	9.6 ± 0.4
Control	690.3 ± 23.9	34.3 ± 0.9	14.3 v 0.3	26.6 ± 0.2	6.8 ± 0.2	8.8 ± 0.3	9.3 v 0.3
DBA/2J
Ethanol	638.0 ± 25.7	34.1 ± 0.5	15.0 ± 0.3	26.3 ± 0.3	6.6 ± 0.3	8.2 ± 0.3	9.7 ± 0.3
Pair-fed	649.9 ± 21.1	32.7 ± 0.6	15.3 ± 0.5	27.0 ± 0.3	6.0 ± 0.2	8.9 ± 0.4	10.0 ± 0.4
Control	699.6 ± 26.0	33.5 ± 0.5	15.2 ± 0.3	27.1 ± 0.5	6.4 ± 0.3	8.5 ± 0.2	9.4 ± 0.3

Prenatal alcohol exposure did not affect any of the hippocampal variables, neither as main factor, nor in interaction with the background.

These data demonstrate that, in three inbred strains of mice (C57BL/6, BALB/c, and DBA/2), prenatal exposure to alcohol does neither affect spatial memory nor hippocampal neuroanatomy. Prenatal alcohol exposure did influence body and brain weight in some strains and dramatically reduced live birth rates in C57BL/6 animals.

A first caveat is, of course, the observation that the low number (2) of C57BL/6 pups prenatally exposed to ethanol precludes any strong conclusions for that strain. We still decided to include these animals in the present report because simple visual inspection of the data shows that even for these two lone survivors there is not even a trend towards any differences in behavior or hippocampal morphology, just as is the case for the other two strains. Of course, sample sizes are much more adequate for strains BALB/c and DBA/2, so that the conclusions based on those strains are much stronger.

In short, our results on both behavior and neuroanatomy appear not to be in line with those, for instance, summarized in Berman and Hannigan [11], who concluded that prenatal alcohol exposure consistently produced significant deficits in spatial learning and/or memory. A closer look, however, revealed that most animal studies on the effects of prenatal alcohol exposure have been performed in rats and not mice. For instance, a survey of the recent literature, using PubMed, revealed only few studies that used mice as experimental models for prenatal alcohol exposure, and, to our knowledge, no studies involved multiple (>2) inbred mouse strains in the analysis of brain-behavior relations with regard to prenatal alcohol exposure. We could only find one study [12] that included two inbred strains (C57BL/10 and DBA/1), which reacted differently to early alcohol exposure. For instance, open field activity was decreased in C57BL/10, but not in DBA/1 mice, whereas aggression was more affected in DBA/1s. Most mouse studies, however, investigated the effects of prenatal alcohol exposure on various neurobehavioral aspects in only one strain, namely C57BL/6 mice. Thus, Opitz et al. [13] intubated pregnant C57BL/6 females with alcohol from gestational day 14–18 and investigated whether this affected radial maze performance in their offspring. As was the case in the present study, no effect of prenatal alcohol exposure on radial maze performance was found and it might, therefore, be concluded that radial maze performance is not affected by prenatal alcohol exposure in mice. One should keep in mind, though, that there might be other reasons why such an effect is not observed. One possibility is that blood alcohol levels were low in this study, an argument also raised by Opitz et al. in their discussion of their data [13]. Since we decided not to disturb pregnancies by taking blood samples (the rationale being that prenatal stress would then be a confounding factor), we were not able to determine blood ethanol concentrations (BAC). Hence, 'our' BACs might have been too low to have an effect. However, brain weights as well as body weights were affected in these experiments by the exposure to prenatal alcohol, a finding reminiscent of Wainwright et al. [14]. In addition, live births were dramatically reduced in C57BL/6, but the two male pups that we did obtain were phenotypically completely normal. It should furthermore be noted that the alcohol concentration used in the present study was almost three times as high as those used by previous authors [15, 16], who reported significant effects on BACs (35-100mg/dl) in their female mice. A more recent study in which ethanol concentrations were slowly increased to 10% over pregnancy showed even higher BACs in the mother (varying from 50 to 150 mg/dl) [17]. Taken together these findings make it very unlikely that not sufficient ethanol reached the developing embryos to have significant effects.

Another explanation for the lack of an effect on learning might be that this type of radial maze is not sufficiently demanding and that other spatial learning tests would be more appropriate to detect prenatal alcohol effects. However, Wainwright et al. [14] did not observe any apparent effects of prenatal alcohol exposure on any measures of performance in a water navigation task in an F2 cross between C57BL/6 and DBA/2 males. Hence, other types of learning tests, less spatial ones such as the puzzle box [18], might perhaps reveal prenatal alcohol effects. In this respect it is interesting to note that in adult C57BL/6 mice prenatal alcohol exposure weakens the efficacy of reinforcers [15], impairs the development of conditioned taste aversion [13], and enhances the sensitivity to amphetamine [16]. We would also like to point out that the radial-maze task used here is quite sensitive and has been used successfully in the past to show learning defects that were not, or only barely, detectable in a water navigation task [19]. An alternative explanation for the absence of effects could be species-specific characteristics with mice being less vulnerable to alcohol per se than rats. It should be noted, however, that other treatments affecting hippocampal mossy fiber projections, such as early postnatal hyperthyroidism, have similar neuroanatomical (and behavioral) effects in rats and mice [20, 21]. A final possibility is that effects depend on the age of testing. For instance, there might be a specific time window, in which the effects become visible. Either the effects might be transient or they might only appear at a later age (see [11] for a discussion).

Another striking result of this study is the low delivery rate of C57BL/6J females. Out of 13 pregnancies only one female gave birth to a viable litter. The other twelve pregnancies resulted in still-born pups, or the mother died while giving birth, or the pups were eaten by their mothers immediately after birth. Whether this finding reflects a higher sensitivity in developing C57BL/6J embryos to alcohol or higher blood alcohol concentrations in the mother or a combination of both cannot be inferred from these experiments. Although the only litter born might not be a representative sample, males from this litter appeared to perform equally well in the radial maze as their pair-fed and control counterparts; neither were the sizes of the IIPMF terminal fields affected.

Summarizing, in this experimental design, which used three distinct inbred strains of mice, prenatal alcohol exposure influenced neither radial maze performance nor the sizes of the IIPMF terminal fields. We believe that future research should be pointed either at different targets when using mouse models for FAS (e.g. more complicated behavioral paradigms, different hippocampal substructures, or other brain structures) or involve different animal models, including zebrafish [22] or guinea-pig [23].