Certain brain regions appear to be particularly susceptible to prenatal alcohol exposure, including the corpus callosum, basal ganglia, specific cerebral cortical regions, and – most relevant to this review – the cerebellum (13, 24–26). The cerebellum is associated with motor coordination, including balance, postural adjustment, and associative learning (e.g., eyeblink conditioning). Cerebellar dysmorphology in the form of region-specific hypoplasia and displacement is among the most widely reported anatomical abnormalities in the FASD literature (24, 27–31). Impairments in domains of motor functioning (i.e., balance; coordination) have been cited as well (12, 32, 33). Roebuck et al. (34) examined balance and postural sway in FASD children and controls to assess possible cerebellar-related impairments. The Sensory Organization Test was used to assess the influence of various modes of sensory information on the maintenance of balance. When visual and particularly somatosensory information was compromised or absent, FASD children performed more poorly (as measured by postural stability). However, no group differences emerged when only visual information was compromised, suggesting that cerebellar dysfunction may make FASD children particularly sensitive to altered somatosensory input.

These findings are significant in that they demonstrate the importance of task selection in identifying alcohol-induced impairments. Furthermore, while the deficits appear to be cerebellar dependent, possible effects on peripheral mechanisms cannot be ruled out. To obviate such potential confounds, cerebellar-dependent tasks that limit demands on multi-joint coordination may be ideal (c.f., 35). Toward that end, eyeblink classical conditioning may provide clearer indications of cerebellar dysfunction in FASDs (further details on eyeblink conditioning in the study of FASDs are presented below).

Early work was successful in achieving substantial cerebellar microcephaly (small brain relative to body size) and reductions in cerebellar neuronal populations (Purkinje cells; granule cells) at PD10 following postnatal (PD4–10) binge-like exposure to high alcohol doses (BACs > 450 mg/dl; 41). Later studies employing alcohol delivery over this period demonstrated that (a) more moderate doses (4.5 g/kg) are sufficient to produce brain damage, (b) the pattern of exposure is critical, as lower doses (4.5 g/kg) condensed into more frequent deliveries (binge-like) produce greater cerebellar damage than a higher dose (but lower BACs; 6.6 g/kg) gradually delivered over a longer period, and (c) brain damage persists into adulthood (42, 43).

Alcohol was delivered via artificial rearing in some of the early studies (e.g., 42, 43). Artificial rearing entails surgical implantation of feeding tubes into the stomach which justifiably raises concerns over unwanted surgery/stress effects. Another delivery method is exposure to ethanol vapors, which results in comparable BACs and Purkinje cell depletions (44, 45). Both delivery methods include extended periods of maternal separation. Intragastric intubation, a method in which a polyethylene tube is inserted down the esophagus and into the stomach for alcohol delivery (46, 47), avoids these concerns as maternal separation is minimal.

An outcome of neonatal alcohol exposure that is robust across a variety of doses, patterns of dosing, and delivery methods are findings of enhanced regional vulnerability. Deficits are consistently shown in earlier-maturing (i.e., I, II, XI, X) relative to later-maturing cerebellar cortical lobules (i.e., VI, VII; 4, 42, 43, 45, 48). Furthermore, neonatal binge-like alcohol effects upon Purkinje cells depend both on dose and exposure period within the PD4–9 window (the PD4–9 dosing period has often been used in “third trimester” equivalent models). While daily neonatal alcohol exposure to 4.5 g/kg⁺ (BACs generally > 200 mg/dl) is typically used to obtain Purkinje cell depletion (e.g., 42, 46, 49), slightly lower doses delivered in a binge fashion (i.e., 3.3 g/kg delivered across two feedings or 2.5 g/kg at once – BACs ∼ 150–200 mg/dl) are also capable of producing moderate neuronal loss (39, 50).

Alcohol-induced brain damage is truly problematic insofar as it compromises quality of life, vis-à-vis behavioral and cognitive impairments. The primary behavioral deficits reported as a result of neonatal exposure in rodents are generally in the spatial and motor domains. Neonatal alcohol exposure also impairs traditionally recognized cerebellar-dependent tasks that place high demands on motor control (i.e., parallel bar and rotating rod; 51–53). Though Purkinje neuronal loss has been correlated with motor performance deficits (53), establishing links between alcohol-induced cerebellar damage and behavior is difficult in these instrumental tasks that require complex multi-joint coordination. Cerebellar-dependent Pavlovian tasks like eyeblink classical conditioning which utilize an easily measurable reflexive response and engage a well-defined neural circuit (54) provide a more promising avenue for investigation of causal relationships between alcohol-induced brain damage and behavior (35, 55).

The extensive behavioral (56) and neurobiological (54) data compiled for the first two decades of research on animal models of eyeblink conditioning has focused on the rabbit model, due in part to their tolerance for restraint. Because rats do not accept restraint as well as rabbits, their testing in eyeblink conditioning requires a freely moving preparation. This was established by Ronald Skelton in the late 1980 s using adult Long Evans rats (57). Mark Stanton adapted this freely moving preparation to developing rats. Specifically, the seminal report on this body of work (58) showed that eyeblink conditioning in the rat using a short-delay paradigm (280 ms interstimulus interval – ISI – between CS and US onset – Fig. 1.1) develops gradually between postnatal days (PD) 17 and 24. Furthermore, acquisition differences across ages cannot be attributed to differences in CS or US processing (58), increasing US intensity or varying the ISI does not overcome failures of conditioning at PD17 (59), and the learning (conditioned eyeblink responses (CRs)) present in weanlings is cerebellar dependent (60, 61). John Freeman and colleagues have replicated the behavioral emergence of eyeblink conditioning in rats (PD17–24; see 62) and, through electrophysiological recording of various components of the putative cerebellar–brainstem “eyeblink” circuit, have offered insights into the underlying mechanisms of the ontogeny of eyeblink conditioning.

Vascular risk factors such as hypertension, type 2 diabetes, abnormal lipid levels, and obesity are risk factors common to Alzheimer’s and cardiovascular disease. Observations of Aβ-containing senile plaques in the brains of human patients with cardiovascular disease led Sparks and colleagues (104) to examine the heretofore unexplored brains of a long-standing animal model of heart disease, the cholesterol-fed rabbit.

An extensive elaboration of this rabbit model of AD has demonstrated that the cholesterol-fed rabbit is a valid model of AD on a number of levels. At a molecular level, the brain of this animal model has numerous (>12) features similar to the pathology observed in the AD brain including elevated Aβ concentration, neuronal accumulation of Aβ immunoreactivity, extracellular Aβ plaques, reduced levels of acetylcholine, elevated brain cholesterol, apolipoprotein E immunoreactivity, breaches in the blood–brain barrier, microgliosis, and neuronal loss.

Sparks also discovered that increased levels of Aβ in the brains of cholesterol-fed rabbits were dependent on the quality of the water the animals were administered. Animals given tap water accumulated considerably more Aβ in the brain than animals administered distilled water (105). It was in a collaborative attempt to demonstrate cognitive validity of cholesterol-fed rabbits that Sparks discovered that it was trace amounts of copper in the drinking water that were responsible for the higher levels of Aβ in the brain in comparison to the brains of rabbits administered the cholesterol diet and distilled water. In this issue, he elaborates his position that cholesterol in the diet causes accumulation of Aβ, whereas copper introduced in the drinking water impairs elimination of Aβ.

Sparks and Schreurs’ (106) demonstration that eyeblink classical conditioning was impaired in cholesterol-fed rabbits, in the light of our own work on eyeblink conditioning in human AD and classical conditioning in normal aging rabbits and humans, was a stimulating discovery that led us to initiate work with AD model rabbits. Our aim was to replicate AD neuropathology in the cholesterol-fed rabbit model, test eyeblink conditioning in this model, and determine if galantamine (Razadyne^®) would ameliorate impaired conditioning. Rabbit chow with 2% cholesterol and drinking water with 0.12 ppm copper sulfate were administered for 10 weeks. Control rabbits received normal food and distilled water. Rabbit brains were probed for neuropathology. AD model rabbits had significant neuronal loss in frontal cortex, hippocampus, and cerebellum. Changes in neurons in the hippocampus were consistent with neurofibrillary degeneration and cytoplasmic immunoreactivity for β-amyloid and tau. In a second experiment, cholesterol-fed model rabbits were injected daily with vehicle or 3.0 mg/kg galantamine and tested on 750-ms trace and delay eyeblink conditioning (animals were conditioned first in trace, then in a delay paradigm matched for the ISI between CS and US onset). Galantamine improved eyeblink conditioning significantly over vehicle. We concluded that this animal model of AD has validity from neuropathological to cognitive levels and offers a promising addition to the available animal models of AD. Galantamine ameliorated impaired eyeblink conditioning, extending the validity of the model to treatment modalities (107).

One of the results highlighted in our report was the evident loss of Purkinje neurons and staining for Aβ and tau in the cerebellum of cholesterol-fed rabbits (confirmed by Sparks, unpublished observations). We examined the cerebellum in cholesterol-fed rabbits primarily because the cerebellum is the essential substrate for eyeblink classical conditioning, a behavior on which they were impaired. Although the cerebellum has traditionally been viewed as one of the last structures to be impacted by classical AD pathology in humans, evidence has mounted to indicate that the cerebellum is indeed affected by AD (108).

Sparks and Schreurs (106) had tested trace eyeblink conditioning in cholesterol-fed rabbits because medial temporal lobe structures affected in AD are essential for trace conditioning. In their study, rabbits tested in a short-delay paradigm were not impaired. We used the same interval between the tone conditioned stimulus and airpuff unconditioned stimulus in the delay and trace paradigms and found both to be impaired (see Fig. 1.2 for a comparison of a representative control rabbit (a) and a representative cholesterol-fed rabbit (b) trained in the 750-ms trace eyeblink conditioning paradigm). However, both Sparks and Schreurs (106) and our study tested the trace paradigm first. We followed up with a study of cholesterol-fed rabbits to determine if delay eyeblink conditioning would be impaired when tested before trace conditioning (109). Cholesterol-fed rabbits were significantly impaired in the 750-ms delay eyeblink conditioning paradigm (Fig. 1.1) when that paradigm was tested first, after 10 weeks of the 2% cholesterol and trace copper diet. Just as humans patients diagnosed with AD are significantly impaired in delay eyeblink conditioning, so are cholesterol-fed rabbits.

The eyeblink conditioning apparatus is commercially available from San Diego Instruments and consisted of a portable, automated system with an infrared eyeblink detector and airpuff jet attached to headgear and input to an interface box containing a microprocessor and a miniature air compressor. The interface box is interactive with a computer. The headgear has an adjustable headband holding the airpuff jet and infrared eyeblink monitor. Filtered air is compressed in the interface box and delivered through a jet set approximately 2 cm from the cornea.

The voltage from the infrared device is amplified and differentiated; eyelid data are transferred to the computer for storage and analysis. From the speaker of the interface box, a 1,000-Hz, 80-dB sound pressure level (SPL) tone CS is delivered through headphones. Air for the corneal airpuff US is compressed in the interface box, filtered, and delivered at a pressure of 5–10 psi. The timing of stimulus presentations is controlled by the microprocessor in the interface box.

The conditioning apparatus consists of four separate sound-attenuating chambers, permitting four rabbits to be trained simultaneously. A speaker mounted to the wall of each chamber delivers a pure tone (1 KHz, 85 dB SPL) used as the conditioned stimulus (CS). Each box has a fan running throughout the experiment that provides low-decibel background noise. The headpiece, affixed behind the rabbit’s ears and under its muzzle, holds a plastic tube to deliver 7–8 psi corneal-directed airpuff unconditioned stimulus (US) and an infrared emitter and detector (San Diego Instruments, San Diego, CA) to measure the rabbit’s eyeblink response. This device is positioned approximately 2 cm from the rabbit’s eye. Eyelid retractors keep the rabbit’s eye open. Output from the infrared detector is stored and analyzed using a PC. This system also controls the timing and presentation of the stimuli. The intertrial interval is randomized and ranges between 20 and 30 s. A single session lasts approximately 45 min and consists of 90-paired CS–US trials.