Limitations of Animal Models of Disinhibition
The nonhuman primate is an especially appropriate model for studying disinhibition at the behavioral and frontal brain level because the size of the monkey’s cerebral cortex is similar to that seen in humans (Grant and Bennett 2003). Other animal models, however, do not correspond as well.
For example, postmortem studies in rats suggest that the distributions of GABAA receptors differs from that of humans (Richards et al. 1987). This could have significant implications. A distinct distribution of receptors or differing subunit expression across species could lead to variations in the brain’s function at the molecular, cellular, and electrophysiological levels. For example, the P300, the most robust feature of event-related potentials (i.e., electrophysiological responses to stimuli with characteristic waveforms) (see Rangaswamy and Porjesz, pp. 238–242), manifested in response to unpredictable stimuli (Kaufmann et al. 1982) and emanating partially from the frontal cortex, is reduced in alcoholics (Begleiter et al. 1984; Johnson et al. 1984; Polich et al. 1994). Yet, in a recently developed mouse model of high alcohol consumption, the high-alcohol–preferring animals had an increased P3 latency when compared with the low-alcohol–preferring mice (Slawecki et al. 2003).
Limitations of Animal Models of Frontocerebellar Circuitry
As with other brain structures, the cerebellum as a whole is disproportionately enlarged in humans and nonhuman primates compared with lower species (Semendeferi and Damasio 2000; Sultan and Braitenberg 1993), and its volume of white matter is exponentially greater in more (phylogenetically) recent species (Bush and Allman 2003). The organization of cerebellar inputs from the cortex via the pons (i.e., mossy fibers) is significantly different in humans than in rats (Paula-Barbosa and Sobrinho-Simoes 1976). Cerebellar activation of cortical regions also has been shown to differ among the rat, cat, and monkey (Tolbert et al. 1978; Yamamoto et al. 2004). In addition, the GABAA receptor distribution in the cerebellum has been found to be different between humans and rats (Kume and Albin 1994). The distribution of dopamine receptors in the cerebellum also differs between the mouse, rat, guinea pig, cat, and monkey (Camps et al. 1990). Finally, the pattern of cerebellar pathology in response to alcohol in rodents is markedly different from that observed in humans (Tavares et al. 1987). Such ubiquitous evidence for structural differences in the cerebellum among various species has implications for function and suggests that the study of frontocerebellar circuitry disruption in alcoholism may be difficult in animal models.
Limitations of Animal Models of Reward
Humans and rodents react differently to pharmacological agents that target dopamine receptors located both locally in the VTA and distally in the striatum and prefrontal cortex (Wood et al. 2006). Even within a species, strains may have different dopamine receptor binding properties and distributions (Yaroslavsky et al. 2006; Zamudio et al. 2005); more “effective” receptors may be associated with innate deficits in dopamine levels. The subregional topography of the dopamine transporter, responsible for dopamine uptake after its release, also has been shown to be inconsistent across species (e.g., rodent, monkey, and human) (Smith and Porrino 2008), a finding that also may have a significant impact on extracellular dopamine levels and innate responses to rewarding stimuli.
At some point between initial exposure and dependence, the consumption of alcohol seems to proceed automatically, as a habitual response to antecedent stimuli. This transition may be the result of a complex interchange between executive and habit systems (Redish et al. 2008). Habitual drinking behavior becomes difficult to break using cognitive mechanisms because of an underperformance of executive systems (Jentsch and Taylor 1999), an overperformance of habit systems (Robbins and Everitt 1999), or because of an imbalance between the two systems (Bechara 2005).
Although not explored comprehensively, brain systems potentially contributing to habit formation include the striatum, cerebellum, amygdala, and, in limited conditions (e.g., trace conditioning; see below for more information), the hippocampus. Indeed, any system involved in “automatic” or implicit learning (i.e., learning without awareness) is fundamental for the establishment of habits (for reviews, Eichenbaum and Cohen 2001). Recent work in rodents has focused on the contribution of the corticostriatal network to habit formation. This work suggests that a switch occurs in the control of instrumental behavior so that the associative or medial striatum, important in the early, goal-directed stage of action, is overridden by the sensorimotor or lateral striatum at the later, more habitual stage (reviewed by Yin, Part 2). Furthermore, several types of classical conditioning/implicit learning paradigms, including eye-blink conditioning (McGlinchy-Berroth et al. 1994), visual discrimination learning (Rogers et al. 2000), and contextual cue discrimination learning (Greene et al. 2007), have been shown in both animal and human studies to be critically dependent on selective cerebellar sites.
The amygdala is another brain structure implicated in habit formation. It plays a role in emotional regulation and behavioral control (for review, see McBride 2002). It has been connected to a specific type of conditioned learning—Pavlovian fear conditioning (Volkow et al. 2002)—in which a neutral conditioned stimulus is paired with a fear-inducing unconditioned stimulus, so that animals come to exhibit a conditioned fear response to the conditioned stimulus. Extensive evidence indicates that the basolateral amgydala is critical for experimental extinction of this acquired fear (Akirav and Maroun 2007).
Although there is support for (Alvarez et al. 1989) and against (Kril et al. 1997) neuronal loss in the amgdala of chronic alcoholics, several in vivo MRI studies provide evidence for volume deficits in the amygdala of abstinent, long-term chronic alcoholics (Fein et al. 2006; Makris et al. 2008). Furthermore, modifications of the GABAA receptor in the basolateral amygdala have been reported in Cynomolgus macaques exposed to alcohol for 18 months (Anderson et al. 2007). How altered GABA receptor function, loss of neurons, or volume reductions in the amygdala contribute to the formation of an alcohol habit remains to be seen.
In another specific form of classical conditioning—termed trace conditioning—a silent period elapses between the occurrence of the conditioned stimulus and the delivery of the unconditioned stimulus (i.e., the conditioned stimulus and unconditioned stimulus are not paired at precisely the same moment, but rather, there is a silent period between the presentation of the conditioned stimulus and unconditioned stimulus).
Evidence from animal (Weible et al. 2006) and human (Cheng et al. 2008) research suggests that the hippocampus plays a critical role during trace eye-blink conditioning. MRI provides in vivo evidence for volume deficits in the anterior hippocampus of chronic alcoholic individuals (Agartz et al. 1999; Sullivan et al. 1995). However, other than its effect on volume shrinkage, alcohol does not appear to have an effect on the number of hippocampal neurons, per se, as shown in studies using postmortem human hippocampal tissue (Harding et al. 1997; Kril et al. 1997). In contrast to the human condition, chronic exposure to alcohol in rodents induces a decrease in neuronal counts in CA1 to CA4 regions of the hippocampus in female Sprague-Dawley (Bengoechea and Gonzalo 1991) and male Long-Evans (Walker et al. 1980) rats and a decrease in the number of pyramidal neurons in CA1 and CA2 regions of the hippocampus of mice (Pawlak et al. 2002). Compared with humans, rodents have a disproportionately larger hippocampal volume, which may account for the notable differences in neuronal loss observed between humans and rodents. Nonetheless, modified hippocampal anatomy may contribute to impaired trace eye-blink conditioning in rats exposed to a binge-like patterns of alcohol in the early postnatal period (Tran et al. 2007) and in nonamnesic alcoholic individuals (McGlinchey et al. 2005).
In humans, both the striatum and the cerebellum have been shown to participate in the automatization process during the late learning stage of a repeated visuomotor sequence (Doyon et al. 1997) and of a sequence of finger movements (Doyon et al. 1998). Yet the collaborative contributions of the striatum, cerebellum, amygdala, and hippocampus to the formation of an alcohol consumption habit have yet to be demonstrated.
Limitations of Animal Models of Stress
CRF initializes the synthesis of corticosteroid hormones, which, in turn, act on glucocorticoid receptors in the brain. Glucocorticoid receptors act as nuclear transcription factors and contribute to the regulation of brain cell properties by modifying the transcription of responsive genes, and therefore, protein synthesis (de Kloet et al. 2005).
In adulthood there is high consistency across animal species in terms of the brain regions that express glucocorticoid receptors, although the levels of expression can differ (e.g., rodents exhibit relatively high glucocorticoid receptor expression in the CA1-2 subfields of the hippocampus and primates exhibit relatively high glucocorticoid receptor expression in the neocortex) (Pryce 2008). Significantly, the relative densities of these receptors differ considerably during postnatal development, creating species-specific periods of critical vulnerability. For example, early life stress in a species that exhibits low glucocorticoid receptor expression in infancy could be less harmful than early life stress in a species that exhibits high glucocorticoid receptor expression, because there are fewer receptors to mediate the effects of elevated cortisol (Fuchs and Flugge 2002). These findings are relevant when modeling alcoholism in animals, especially in light of evidence that the onset of stress-related disorders is age dependent.
Limitations of Animal Models of Inflammation
The neuroinflammation theory of alcohol-related neuronal loss and brain atrophy is relatively new. As a result, there have been few studies designed to specifically test the hypothesis. With respect to the effects of neuroinflammation on neurogenesis, major differences exist between the rat and mouse stem/progenitor cells that are involved in neurogenesis (Ray and Gage 2006), which warrants caution when drawing inferences from one species to another.
Evidence for Recovery with Abstinence
From the earliest computed tomography (CT) studies to current MRI studies aimed at tracking evidence for brain structural recovery, there has been positive support for at least partial reversal of brain tissue shrinkage with abstinence from alcohol (CT studies: Cala et al. 1983; Carlen et al. 1984, 1986) (MRI studies: Cardenas et al. 2007; Pfefferbaum et al. 1995, 1998).
Indeed, alcoholic brain pathology can be subsumed under Carlen’s two-component hypothesis (Carlen et al. 1984), one reflecting permanent change (i.e., irreversible neuronal cell death), notably in the superior frontal association cortex, and one reflecting a transient change, such as shrinkage without cell death, thereby permitting volume to change (up or down) without long-term damage. Indeed, the majority of shrinkage with drinking does not necessarily reflect “neuronal loss.” Rather, the controlled longitudinal imaging studies demonstrating volume reductions likely reflect nonneuronal loss and neuronal cell body and process shrinkage. That brain volume can increase and that this increase predicts improvement in neuropsychological test performance (Cardenas et al. 2007; Rosenbloom et al. 2007; Sullivan et al. 2000b) supports the contention that little neuronal death occurs with alcoholism.
Animal Models of Recovery
In aged Fisher 344 rats, recovery after long-term treatment with alcohol was associated with a restoration of the total number of synapses on Purkinje neurons of the cerebellum lost during exposure (Dlugos and Pentney 1997). Furthermore, abstinence for 5 weeks indicated a twofold increase in new neurons formed in neurogenetic zones of abstinent animals compared with alcohol-naive animals (Nixon and Crews 2004). This increase in neurogenesis during abstinence from chronic alcohol exposure may be related to the recovery of brain volume deficits (Pfefferbaum et al. 1995) and cognitive deficits in abstinent alcoholics (Sullivan et al. 2000c).
Together, studies in humans and animal models provide support for the involvement of specific brain structures over the course of alcohol addiction. Researchers have identified genetic variants of key inhibitory receptors in the prefrontal cortex that may produce a heritable vulnerability to alcohol, perhaps accounting for the disinhibited personality type observed in certain alcoholics and which leads to a predisposition to develop alcoholism.
The prefrontal cortex and its complex circuitry with the basal ganglia also is likely involved in the acute reinforcing (or rewarding) effects of alcohol. Furthermore, modified prefrontal inhibitory receptors may contribute to dysregulation in other brain regions targeted by the prefrontal cortex, such as the cerebellum. The basal ganglia, cerebellum, amygdala, and hippocampus may collectively contribute to the formation of an alcohol habit. The HPA axis additionally has a role in the development of dependence, as well as the vulnerability to stress-induced relapse. Inflammatory cascades initiated by chronic alcohol consumption are a factor that may contribute to alcohol-induced neuropathology.
Each theory, linked to specific brain structures, has helped to describe the mechanisms associated with the transition from controlled drinking to compulsive consumption or dependence. The development of each theory depended critically on information acquired from animal models, whether they met all the criteria necessary for an animal model of alcoholism or not. Figure 7 is a simplified schematic of the brain structures modified by alcohol and illustrates reciprocal connections between basal ganglia, limbic structures (i.e., hippocampus and amygdala), and cerebellum, each driven by inputs from the cortex, with reciprocal connections to the cortex via the thalamus. Also illustrated are the reciprocal connections between basal ganglia, limbic structures, and cerebellum with the hypothalamus. Not illustrated but germane to the course of alcohol addiction are modifying aminergic (dopamine and norepinephrine), cholinergic, serotonergic, peptidergic, and hormonal influences on the various structures.
In moving forward, a challenge will be to develop a theory that accounts for the brain structures uniquely targeted by alcohol. Perhaps different neural circuits are important at different stages across the time course from first drink to dependence. Alternatively, differential involvement of these circuits across alcoholics could contribute to heterogeneity in brain regions affected. A theory that unifies the brain circuitries modified by alcohol may very well have a major impact on our understanding of brain function in general.