Brain and Behavioral Pathology in an Animal Model of Wernicke’s Encephalopathy and Wernicke-Korsakoff Syndrome

Abstract

Animal models provide the opportunity for in-depth and experimental investigation into the anatomical and physiological underpinnings of human neurological disorders. Rodent models of thiamine deficiency have yielded significant insight into the structural, neurochemical and cognitive deficits associated with thiamine deficiency as well as proven useful toward greater understanding of memory function in the intact brain. In this review, we discuss the anatomical, neurochemical and behavioral changes that occur during the acute and chronic phases of thiamine deficiency and describe how rodent models of Wernicke-Korsakoff Syndrome aid in developing a more detailed picture of brain structures involved in learning and memory.

Introduction

Studies in rodent models have provided greater understanding of the neurobiology of memory and memory dysfunction. In particular, rodent models have been developed and extensively validated for human memory disorders such as Alzheimer’s disease, age-related memory loss, and temporal lobe damage-associated amnesia. For example, transgenic mice carrying mutation/deletion of Alzheimer’s disease-associated genes are now revealing the anatomical and functional consequences of fibrillar amyloid deposits (Yang et al., 2005; Grutzendler et al., 2007). Likewise, studies with aged rats and mice are elucidating the anatomical and physiological changes that occur in hippocampal circuits which lead to age-related memory decline (Erickson and Barnes, 2003; Krause et al., 2008; Oler and Markus, 2000; Shen et al., 1997).

Despite extensive study on the role of the hippocampus and medial temporal lobe structures, identification of other brain areas critical for normal memory function are still debated. In the present review, we describe our work using a rodent model of Wernicke-Korsakoff Syndrome (WKS), a neurodegenerative disorder caused by thiamine (vitamin B1) deficiency, to study the role of various brain structures on memory and WKS dementia/learning deficits. This model recapitulates thiamine deficiency (TD) observed in WKS patients as well as the cognitive/memory deficits that are a cardinal feature of WKS. We discuss how the use of this model has revealed a number of neurochemical, neuroanatomical and behavioral changes that occur in various regions of the brain at the acute and chronic stages of TD.

Structural, Neurochemical and Behavioral Pathology Associated with Rodent Models of Wernicke’s Encephalopathy and Wernicke-Korsakoff Syndrome

TD induces severe CNS dysfunction leading to several disorders including: beriberi, Wernicke’s encephalopathy, and the amnestic disorder WKS. First described in the late 19th century (Thomson et al., 2008), the cause of TD was almost exclusively due to dietary insufficiency, a finding that was instrumental to the early discovery of vitamins and their role in human health and nutrition. Despite the fact that TD due to dietary insufficiency has been largely eradicated in industrialized countries following the establishment of food fortification and vitamin supplementation programs, TD is still a problem in adults with chronic alcoholism due to poor diet and nutritional malabsorption. In addition, TD continues to be problematic in developing nations where adequate nutrition is lacking (Harper, 2006). Genetic mutation of genes such as Slc19A2 or Slc19A3 which code for thiamine transporters (Guerrini et al., 2005; Kono et al., 2009) or the transketolase-like 1 gene (Tktl1) which affects thiamine pyrophosphate binding, is thought to potentially predispose individuals (especially alcoholics) who have diets with low/lacking thiamine to WE and WKS (Coy et al., 1996, 2005). More recently, TD has been observed in an increasingly diverse clinical spectrum (reviewed in Sechi and Serra, 2007) including after gastrointestinal surgery and in individuals with AIDS, Crohn’s disease, eating disorders, renal disease, cancer, etc (Parkin et al., 1991, 1993; Saad et al., 2010). Thus, brain and behavioral changes following TD is a clinically-relevant area of study with broad implications for a wide range of affected individuals.

Neurological manifestations of TD were first described by Carl Wernicke in the late 19th century and included a triad of symptoms including ataxia (loss of coordinated muscle movement), nystagmus (involuntary eye movement) or ophthalmoplegia (eye movement paralysis) and “confusion” or change in consciousness/mental status (Thomson et al., 2008; reviewed in Sechi and Serra, 2007). For his discovery, this condition now bears Wernicke’s name and is known as Wernicke’s encephalopathy (WE). Soon after Wernicke’s description, Sergei Korsakoff described a similar condition observed in chronic alcoholics that was characterized by severe memory deficits (Korsakoff, 1955), not knowing that these individuals suffered from WE. Thus, TD can result in WE which can culminate in death if left untreated (~20% of cases) or advance into WKS in ~85% of patients (Harper et al., 1986; Day et al., 2004).

A number of brain and behavioral changes accompany acute and chronic TD that can be investigated using rodent models (Table 1). Similar to that observed in humans, TD in rodents can be fatal and results in a neurological phenotype with striking similarity to that seen in humans (discussed below). In particular, histopathological studies in humans and rodents have demonstrated that prolonged TD produces similar time-dependent cell death, glial activation, inflammation, abnormalities in oxidative metabolism, and/or degeneration of neural tissue (Langlais, 1995; Zhang et al., 1995), suggesting a similar role played by thiamine in the human and rodent brain. In particular, our TD rodent model recapitulates the characteristic lesions seen in affected individuals with WE as well as the accompanying memory deficits that are characteristic of WKS. We review the brain and behavioral changes in rodent models that occur during (acute WE phase) and after recovery (chronic WKS phase) of TD and compare these results with findings in humans.

Table 1.

Comparison of the behavioral presentation of TD in humans and in rodent models of TD

Behavioral Symptoms:	Clinical Presentation in Humans:	Rodent Model of ID:
Anorexia	NO	+
Apnea	+	NT
Ataxia	+	+
Bradycardia	+	tachycardia
Depressive symptoms	NO	+
Dysarthria	+	NO
Hearing loss	+	NT
Hypotension	+	NO
Loss of consciousness	+	+
Mental status change	+	+
Motor incoordination	+	+
Nystagmus	+	+
Oculomotor dysfunction	+	NT
Opisthotonus	NO	+
Postural dysfunction	+	+
Seizure	+	+
Startle response	NT	+
Anterograde memory loss	+	+
Retrograde memory loss	+	NO

NO: Not observed

NT: Not tested

Rodent Models of Wernicke Encephalopathy and Wernicke-Korsakoff Syndrome

There are two types of experimental rodent models of TD that have been studied in detail. The first approach induces TD via feeding of thiamine-free food to rodents for an extended period of time sufficient to deplete thiamine stores (3–4 weeks). The second approach generates TD more quickly by combining the feeding of a thiamine free diet in addition to administration (i.p. injection) of a thiamine pyrophosphokinase inhibitor, such as pyrithiamine. This second model, termed pyrithiamine-induced thiamine deficiency (PTD), induces a stereotyped progression of neurological/behavioral symptoms that have been mapped to specific pathological changes in anatomy and neurochemistry (Zhang et al., 1995; Hazell and Butterworth, 2009) and mimic the pathology described in humans with TD. Moreover, both human clinical reports (Cook and Thomson, 1997; Thomson and Cook, 1997, Cook, 2000; Thomson et al., 2002; Agabio, 2005; Kopelman et al., 2009;) and studies in rodent models of TD (Langlais, 1995) demonstrate that WE symptoms can be treated with high doses of thiamine (e.g. minimum of 500 mg thiamine three times daily, for at least 2 days in WE patients; Agabio, 2005) which may prevent the development WKS in both species. Thus, the validity of rodent models of TD center upon the replication of both acute neurological symptoms observed in the WE phase and the permanent cognitive changes that define the amnestic disorder of WKS. Therefore, the strength of these rodent models come from the ability to manipulate the timing and duration of TD, the timing and nature of therapeutic interventions, and the time at which anatomical and/or physiological indices are measured.

I. Cytopathological, Neurochemical and Behavioral Changes During the Acute WE Stage of Thiamine Deficiency

Cytopathology

Thiamine plays a key role in the maintenance of normal cellular function and because of its high oxidative metabolism, the CNS is particularly vulnerable to TD. Deficiencies of thiamine lead to disruption of several enzymatic pathways that are crucial for proper cellular and metabolic function (Butterworth, 1986a; Butterworth et al., 1986b, 1993a; Sheu et al., 1998). TD in humans and rodents reduces levels of thiamine pyrophosphate (TPP), a coenzyme necessary for the catalysis of enzymatic systems involved in cerebral glucose utilization (Heroux and Butterworth, 1995; Mancinelli et al., 2003). The reduction in TPP synthesis in turn disrupts the formation of TPP-associated enzymes. For example, decreased levels of the TPP-dependent enzyme alpha-ketoglutarate dehydrogenase (αKGDH) have been consistently observed in individuals with WKS as well as in the PTD model (Gibson et al., 1984; Butterworth et al., 1986a, b; Sheu et al., 1998). Serving as an important intermediate in the tricarboxylic acid (TCA) cycle, an enzymatic process essential for adenosine triphosphate synthesis, diminished αKGDH activity disrupts mitochondrial glucose oxidation resulting in cell death (Gibson et al., 1984). Consequently, oxidative stress as been implicated in the acute cytopathology induced by TD.

Recent studies further confirm the role of oxidative stress in the neuropathology associated with TD. Elevated levels of oxidative stress indicators such as hemeoxygenase-1 and intracellular cell adhesion molecule-1, have been reported in the brains of TD rats (Calingasan and Gibson, 2000b; Gibson and Zhang, 2002). TD-induced impairments in oxidative capacity also accelerate endothelial nitric oxide synthase expression leading to excessive nitric oxide (NO) production (Calingasan and Gibson, 2000a). This finding may be of particular significance as NO can interact with the free radical superoxide to form peroxynitrite, a powerful oxidant associated with cell toxicity and damage (Hazell and Butterworth, 2009). Importantly, both NO and peroxynitrite inhibit αKGDH activity which further exacerbates mitochondrial dysfunction and oxidative stress (Park et al., 1999; Hazell and Butterworth, 2009). Not surprisingly, concentrations of reactive oxygen species (ROS) were found to be significantly elevated in the thalamus and cortex in the PTD model (Langlais et al., 1997). Thus, TD results in a vicious biochemical cycle wherein decreases of αKGDH catalyze excessive ROS production which further inhibits the ability of αKGDH to regulate TCA cycle turnover (Hazell and Butterworth, 2009).

Unlike in the clinical setting where individuals suffering from TD are only be monitored upon seeking medical attention, rodent models of TD allow for near-continuous monitoring and assessment. In the thiamine-deficient food model and in the PTD model, lesions limited largely to the anteroventral ventrolateral (AVVL) and ventroposterolateral (VPL) thalamic nuclei (Zhang et al, 1995) are observed <1 hr after onset of seizures and opisthotonus (spasm of the muscles causing backward arching of the head, neck and spine). Specifically, there is an approximate 35% loss of neurons within the VPL thalamus with concomitant increases in reactive astrocytes and microglia activation (Todd and Butterworth, 1998, 1999). Three to five hours after the appearance of seizures, there is massive cell loss in the midline thalamic nuclei and mammillary bodies (Zhang et al, 1995). Approximately 6–14 hr of seizures, cell death extends throughout the periaquedutal grey and tegmentum. Furthermore, hemorrhagic lesions are also observed in the hypothalamus and thalamic geniculate nuclei (medial and lateral). Interestingly, there is little evidence of neurodegeneration in the hippocampus or cortex in the PTD model (Langlais and Mair, 1990).

The pathology seen in rodent TD models largely mimic those seen in humans with WE as described by functional imaging or postmortem histology. However, human imaging studies often describe typical and atypical neuroimaging findings, (reviewed in Zuccoli and Pipitone, 2009) which likely relates to the diversity of clinical presentations of individuals with WE as well as whether or not the patient is an alcoholic (Zuccoli and Pipitone, 2009; Zuccoli et al., 2009, 2010b). Typical neuroimaging findings are present in the mammillary bodies, periaqueductal grey, and a number of thalamic nuclei including dorsomedial, anterior, pulvinar, and midline nuclei. In contrast, atypical neuroimaging findings of WE include signal intensity changes in regions such as cerebellum, caudate, and cerebral cortex (Lapergue et al., 2006; reviewed in Zuccoli and Pipitone, 2009). Signal intensity changes in the basal ganglia have typically only been observed in pediatric cases of WE (Zuccoli et al., 2010c) but were recently reported in two adult cases (Zuccoli et al., 2010a). Only recently have magnetic resonance imaging (MRI) methods been applied to the study of TD in rodent models. A comparison between results from human neuroimaging studies and those using rodent models of TD is shown in Table 2. An advantage of rodent models of TD is the reproducibility and reduced variability in the resultant brain and behavioral pathology. In a recent study in PTD-treated rats, regions showing signal intensity increases in T2-weighted MRI included the neocortical white matter, tectum, cerebellar peduncles, as well as anterior, dorsal, and midline thalamic nuclei and during the early acute (WE) phase (Dror et al., 2010). No changes were observed in the basal ganglia in two MRI studies of PTD treated-rats (Pfefferbaum et al., 2007; Dror et al., 2010). As shown in Table 2, these neuroimaging data resemble many, but not all, of that seen in individuals with WE.

Table 2.

Comparison of neuroimaging and neuropathology findings in humans with TD and in rodent models of TD

Brain Structure	Neuroimaging Findings in Humans	Rodent Neuropathology	Rodent Neuroimaging
Cerebellum	Atypical	+	+
Corpus callosum	Atypical	+	+
Cranial nerve nuclei	Atypical	+	NO
Frontal cortex	Atypical	+	+
Mammilary bodies	Typical	+	+
Medulla	Atypical	+	NO
Parietal cotex	Atypical	+	NO
Periaqueductal gray	Typical	+	+
Red Nucleus	Atypical	NO	NO
Striatum	Atypical	NO	NO
Tectum	Typical	+	+
Thalamus	Typical	+	+

NO: Not observed

Neurochemical Changes

Rodent models have been influential in describing the neurochemical changes in the TD brain. All major neurotransmitter systems are altered during the acute stage of TD, including transmitter synthesis and/or release mechanisms. Selective regions of the thalamus, such as the midline and lateral posterior nuclei, demonstrate increased extracellular glutamate levels 4–5 hours after the onset of seizures (see below) which suggests that thalamic lesions seen in the PTD model may result from excitotoxicity (Hazell et al., 1993; Langlais and Zhang, 1993; Todd and Butterworth, 1998). Mechanistically, increased glutamate levels during TD might be a consequence of impaired glutamate buffering by the astrocytic transporters GLT1 and GLAST (Hazell et al., 2001) and/or increased presynaptic transmission of glutamate via pathological reductions of complexin I and II (Hazell and Wang, 2005). These finding are also consistent with impaired mitochondrial oxidation as increased oxidative stress is known to inhibit glutamate transport via peroxynitrite-induced transporter breakdown (Volterra et al., 1994; Trotti et al., 1996, 1998). Increased glutamate levels may play a role in seizure generation observed during the acute stages of TD in rodent models.

Gamma-aminobutyric acid (GABA) activity is reduced during the acute phase of TD in rodent models (Thompson and McGeer, 1985; Heroux and Butterworth, 1988). This might be due to impaired TCA cycle turnover as reduced de novo synthesis of GABA is seen in the medial thalamus of PTD-treated rats (Navarro et al., 2008). Deficits of C-glucose incorporation into GABA and glutamate have also been observed (Gaitonde et al., 1974, 1975). Thus, the altered balance of GABA and glutamate levels in the brain may be responsible for the seizures observed during the acute stages of TD in rodent models.

TD alters norepinephrine (NE) and dopamine (DA) levels, which might result from vesicular dysfunction via impaired catecholamine metabolism (Mousseau et al., 1996). In the frontal cortex, hypothalamus, and thalamus of PTD-treated rats, impaired vesicular transport leads to intracellular release and breakdown of DA (Mousseau et al., 1996). This may lead to a reduction in synaptic/extracellular release of DA and may blunt NE production through impaired metabolic function.

As described above, TD results in a reduction of TPP activity as well as decreases in another TPP-dependent enzyme, pyruvate dehydrogenase (Gibson et al., 1982; Pekovich et al., 1998). This enzyme is critical for the synthesis of acetyl-CoA from pyruvate during glycolysis (Butterworth, 1993a; Todd & Butterworth, 1999). The TD-induced decrease in levels of acetyl-CoA (Jankowska-Kulawy et al., 2010), which is a critical co-factor of acetylcholine (ACh), consequently results in a decrease ACh synthesis (Gibson et al., 1984). These data combined with the documented neurodegeneration of cholinergic neurons after chronic TD or PTD treatment (see below) reflect major changes in ACh function both during and after TD.

Dorsal and medial raphe neurons show reduced levels of tryptophan hydroxylase in mice fed a thiamine-free diet for 20 days (Nakagawasai et al., 2007), indicative of impaired synthesis of serotonin (5-HT). Interestingly, head-twitch responses in TD mice elicited by intracerebroventricular administration of a 5-HT2A receptor agonist is enhanced, suggesting that TD alters the function of 5-HT2A receptors (Nakagawasai et al., 2007). In the PTD rat model, increased 5-HT synthesis has been observed in the cerebellum, hippocampus, striatum, hypothalamus and cortex after administration of a monoamine oxidase inhibitor (Van Woert et al., 1979). These neurochemical effects as well as behavioral manifestations of TD are completely reversed within two hours after thiamine administration (Van Woert et al., 1979).

There is also evidence that histamine is a key mediator of cell death in the acute TD brain (Steinbusch and Verhofstad, 1986; Langlais et al., 2002). Altered levels of histamine have been observed in brain regions vulnerable to neurodegeneration in PTD-treated rats (Langlais et al., 1994; McRee et al., 2000). Moreover, Langlais and colleagues (2002) demonstrated that therapeutic depletion of brain histamine levels prior to TD symptom onset was capable of attenuating thalamic neurodegeneration in the PTD model. While these findings suggest an integral role for histamine in the pathogenesis of TD, the exact mechanisms remain unknown. Histamine may influence TD-induced neurodegeneration by increasing blood-brain barrier permeability (Steinbusch and Verhofstad, 1986), glutamate-related excitotoxicity (Colwell and Levine, 1997; Shelton and McCarthy, 2000) and/or proinflammatory cytokine release (Abbot, 2000). Further research is needed to elucidate the role of histamine in the neuropathology associated with TD.

Behavioral Changes

In addition to (or instead of) the cardinal symptoms described above, patients with WE can present with a diverse number of symptoms which may also vary depending on the age of the affected individual (Kornreich et al., 2005; Fattal-Valevski et al. 2009; Sechi et al. 2007; Sugai and Kikugawa, 2010). Some of presenting symptoms of individuals with TD have not been directly assessed in rodent models of TD such as apnea, hearing loss, and dysarthria. Nevertheless, several behavioral comparisons between TD in rodents and humans are available and are described below.

A number of abnormal behaviors are observed once rodents reach TD status. However, the temporal progression of behavioral change differs between the thiamine-free diet model and the PTD model. In the thiamine-free diet model (no PTD), mice display performance deficits in the Y-maze as early as 9 days after the start of treatment (Zhao et al., 2008). In this same model, rats display an increase in startle response along with bradycardia, lack of appetite, and hypothermia after 14 days of treatment (Onodera et al., 1981, 1991; reviewed in Nakagawasai, 2005). As the TD state progresses (day 25) in the thiamine-free diet model, there is memory impairment on tasks such as passive avoidance (Nakagawasai et al., 2001) and changes in emotional behavior, including increased muricide and impairment on the forced swim task (Nakagawasai et al., 2001). Once these behavioral symptoms develop, they are not suppressed by the administration of thiamine, indicative of a significant change in brain physiology and neurochemistry (Nakagawasai, 2005). Interestingly, some 5-HT and NE reuptake inhibitors can decrease some of the observed emotional behavior changes of acute TD (Onodera et al., 1981; Nakagawasai et al., 2001).

In contrast to the thiamine-free diet model, PTD treatment generally lasts for approximately 16 days, at which point seizures and opisthotonus begin and continue until death unless parenteral thiamine is administered in order to begin restoration of thiamine levels. The first behavioral symptoms of the PTD model include weight loss which becomes evident approximately on day 10 of PTD treatment followed by ataxia between days 13–15 of treatment (Zhang et al., 1995). Within 1–2 days of the development of ataxia, PTD-treated rats display a loss of righting reflex and the emergence of seizures and opisthotonus (day 16–17; Zhang et al., 1995). These later symptoms are associated with significant brain lesion and the degree and extent of neural injury is directly related to when parenteral thiamine treatment is administered (typically 2 doses of 100 mkg/kg i.p. of thiamine hydrochloride; Anzalone et al. 2010).

II. Cytopathological, Neurochemical and Behavioral Changes Following Recovery from Thiamine Deficiency in the PTD model

The PTD model has been the most widely used in the examination of brain and behavior changes after recovery from TD, and is an accepted model of WKS. The approach generally used in our lab and others is to continue PTD treatment until behavioral manifestations of TD appear such as seizures (~16 days). Upon the onset of seizures, parenteral thiamine is administered to prevent death and begin the TD recovery process. Therefore, unlike in most clinical scenarios where affected individuals seek medical attention and receive treatment at varying times after the onset of TD symptoms, the PTD model allows for very stereotyped and reproducible effects. Studies using this approach with the PTD model have revealed numerous neuroanatomical, neurochemical, and behavioral changes that persist from weeks to months as well as others which are likely permanent. In the following sections, we review these neurobiological changes and their relevance to behavioral dysfunction following recovery from TD with particular emphasis on the memory deficits observed in WKS.

Cytopathology

i. Subcortical Pathology in PTD

Following PTD treatment and recovery from TD, hallmark neuropathology is observed in multiple thalamic nuclei (anterior, medial dorsal, intralaminar [ILn], posterior) similar to that seen in WKS patients. In addition to cell loss, there is evidence of hemorrhaging, neural atrophy, and intense gliosis (Mair et al., 1988). As shown in Figure 1, there is a significant loss of neurons in the anterior nuclei and ILn as well as the medial mammillary nuclei (Langlais and Savage, 1995). Much like what has been described in individuals with WKS (Victor et al., 1989), the PTD model produces degeneration of key limbic system fiber tracts such as the mammillothalamic tract and fornix (Markowitsch, 1988; Langlais and Zhang, 1997). Together, these diencephalic structures constitute integral components of the limbic-forebrain memory system and disruption of this circuit following TD is strongly associated with the amnestic phenotype of WKS (discussed below).

Figure 1.

Neurodegeneration of thalamic nuclei after PTD treatment. Coronal sections of the thalamus and staining of NeuN immunoreactive neurons in the thalamus of Control (A, C) and PTD-treated rats (B, D). Rostral (A, B) and more caudal (C, D) levels of the thalamus are shown for comparison. Abbreviations; AD: Anterodorsal nucleus, ADDM: Anteroventral thalamic nucleus/dorsomedial division; ADVM: Anteroventral thalamic nucleus/ventrolateral division; 3V: third ventricle; D3V: Dorsal third ventricle stria medullaris; PC: Paracentral thalamic nucleus; CM: Central median nucleus; CM: Central lateral nucleus.

A number of studies have attempted to relate the amnestic symptoms of WKS with damage to specific thalamic regions, especially the anterior nuclei, medial nuclei, and ILn complex (Mair et al., 1979; Mayes et al., 1988; Harding et al., 2000). Following postmortem histological analyses, Harding and colleagues found that the extent of damage to the anterior nuclei predicted the degree of memory impairments in chronic alcoholics with WKS (Harding et al., 2000). Using this same approach of postmortem analyses of alcoholics with and without memory deficits, damage to subcortical structures including the hypothalamus (Harding et al., 1996), brainstem serotonergic nuclei (Halliday et al., 1993), the noradrenergic locus coeruleus (Halliday et al., 1992), basal forebrain (Cullen et al., 1997) and the hippocampus (Harding et al., 1997) were found to be uncorrelated with the clinical presentation of memory impairments. In contrast, Gold and Squire (2006) reported similar amnestic profiles in three patients that had lesions to one of three different limbic regions including the hippocampus, ILn/internal medullary lamina (IML), or anterior thalamus. Comparisons between postmortem (Cullen et al., 1997; Halliday et al., 1992, 1993) and functional neuroimaging (Gold and Squire 2006) are difficult but suggest that the thalamus and hippocampus are part of a limbic-diencephalic memory circuit and that damage to a single component of this circuit is sufficient to disrupt memory functioning (see also Aggleton and Brown, 1999).

Brain damage in humans with TD is seldom restricted to individual thalamic nuclei leading researchers to rely on animal lesion studies to further identify the individual contributions of specific brain nuclei to memory function and memory dysfunction in WKS. For example, selective damage to the anterior thalamic nuclei in rodents produces impairments in both working and reference forms of spatial memory (Mitchell and Dalrymple-Alford, 2005), a result likely due to the strong reciprocal connectivity of these nuclei with the hippocampus (Gabriel and Sparenborg, 1986; Gabriel et al., 1987; van Groen and Wyss, 1995). The underlying mechanisms of spatial memory deficits associated with anterior thalamic lesions remains to be clarified, but might involve: 1] failed spatial information storage (Bryatt and Dalrymple-Alford, 1996; Warburton and Aggleton, 1999; Warburton et al., 2000), 2] perturbed thalamic theta activity (Vertes et al., 2001) and/or 3] a loss of thalamic head direction cells (Aggleton and Brown, 1999; Taube and Bassett, 2003). These data are relevant toward a greater understanding of how anterior thalamic lesions in individuals with WKS results in memory deficits (discussed below).

Lesions of the ILn/IML complex, especially the rostral component (i.e., central median, paracentral, and central lateral nuclei), are observed in the PTD model of TD. These nuclei are proposed to be involved in attentional processes that can modulate working memory performance such as short-term shifting and sustained attention, (Van der Werf et al., 2000, 2002; Schiff, 2008). Consistent with a role in attention, rats with ILn/IML lesions do not show improvements on a signal-detection task when the inter-trial interval is increased, compared to control rats that do show improved performance (Newman and Burk, 2005). Thus, the primary results of ILn/IML lesions appear to relate to attentional deficits that contribute to working memory impairment. In support of these conclusions are data from a recent human functional magnetic resonance imaging (fMRI) study that demonstrated increased signal intensity of the thalamus with increases in working memory or visual attention load (Tomasi et al., 2007). These data and those described above, point to separate roles played by anterior thalamus and ILn/IML lesions in the cognitive phenotype of WKS.

Neuronal loss is observed in the medial dorsal nucleus of the thalamus in the PTD model of WKS (LaRoche et al. 1987). Unlike lesion to other regions of the thalamus such as the anterior nuclei, damage to the medial dorsal thalamus in rats and nonhuman primates has little effect on spatial memory (Neave et al. 1993). Instead, lesion of medial dorsal nuclei has been observed to impair memory for reward location (Mitchell and Dalrymple-Alford, 2005) as well as the acquisition and expression of reward-guided behaviors (Yu et al., 2010) in rats. In nonhuman primates lesions of the mediodorsal thalamus have been shown to impair reward-based decision making on simple discrimination tasks (Gaffan and Murray, 1990) as well as in reward-devaluation studies (Mitchell et al. 2007; Izquierdo and Murray, 2010). Interestingly, the mediodorsal thalamus receives projections from the amygdala (Aggleton and Mishkin, 1984) and amygdala lesions produce similar behavioral deficits to that seen after mediodorsal lesions (Izquierdo and Murray, 2010), suggesting that the medial dorsal thalamus is part of the extended-amygdala system that is critical for reward-based decision making. These data point to a distinct role played by the medial dorsal thalamus in processing reward-related information compared to the anterior nuclei or the ILn which participate more in spatial memory and attention, respectively (described above). Thus, these different thalamic nuclei are critical components of different learning/memory, reward, and attentional systems in the intact brain and each contribute to the cognitive deficits observed in WKS.

Findings of structural damage to the hippocampus in WKS patients has been mixed (Caulo et al., 2005; Mair, 1994; Harding et al., 1997). Similarly, early reports on the structural integrity of the hippocampal formation following PTD treatment revealed no neuropathology (Langlais et al., 1992). However, recent in vivo MRI animal studies have revealed some hippocampal shrinkage following PTD treatment (Pfefferbaum et al., 2007). Despite equivocal anatomical findings, there is good evidence of hippocampal dysfunction in both human and rodents after TD. For example, studies from our laboratory (Roland et al., 2007; Savage et al., 2003, 2007; Vetreno et al., 2008) indicate that acetylcholine release in the hippocampus is reduced in PTD rats during spatial navigation and is accompanied by spatial memory deficits. Likewise, an fMRI study with WKS patients revealed reduced activation of the hippocampus during both the encoding or recognition phases of a visual matching-to-sample task despite no observed structural damage to the hippocampus of these patients (Caulo et al., 2005). These data in humans and rodent models suggest that hippocampal dysfunction seen after TD may be due to damage to structures functionally connected with the hippocampus.

The basal forebrain is vulnerable to TD-induced neurodegeneration. Cell death and dendritic reorganization in the nucleus basalis (nBM) as well as the medial septum (MS) and vertical limb of the diagonal band (DB) was reported in postmortem analyses of WKS brains (Arendt et al., 1995). Studies using TD rodent models have reported similar neurodegeneration. In particular, a 30% reduction in choline acetyltransferase (ChAT) immunopositive neurons (Pitkin and Savage, 2001; 2004) was observed in the MS/DB of PTD-treated rats and this cell loss was correlated with the degree of memory impairment (Roland and Savage, 2009b). Consistent with a loss of MS/DB ChAT neurons, there is also a loss of cholinergic fibers in the cortex (Anzalone et al., 2010) and hippocampus (Nakagawasai et al., 2000) of PTD-treated rats. Moreover, in vivo microdialysis studies with the PTD model have demonstrated reduced ACh activity in the hippocampus and cortex (Savage et al., 2003; Roland and Savage, 2007; Savage et al., 2007; Vetreno et al., 2008). Thus, loss of basal forebrain cholinergic projection neurons likely contributes to the decline in hippocampal function and memory impairments in individuals with WKS as well as in the PTD model.

ii. Cortical and White Matter Pathology in PTD

While contributions of diencephalic structures to the amnestic condition associated with WKS have been extensively studied (Langlais et al., 1996; Mair, 1994), cortical and white matter neuropathology has received less attention. Histopathological (Kril et al., 1997) and in vivo neuroimaging (Colchester et al., 2001; Jernigan et al., 1991) studies of humans with WKS have revealed significant tissue shrinkage of the neocortex including temporal, orbitofrontal, frontal and parietal cortices. Similarly, PTD treatment culminates in degeneration of the frontal, parietal and retrosplenial cortices consisting of axonal and terminal atrophy and cortical thinning due to laminar specific (i.e., layers III and IV) cellular degeneration (LaRoche et al., 1987; Langlais and Zhang, 1997). Pires and colleagues (2001, 2005) demonstrated a reduction in acetylcholinesterase (AChE) activity in the neocortex of TD rats. Work by our group has also revealed a reduction in AChE-stained fibers in the cortex of PTD rats; the extent of AChE fiber loss was correlated with deficits in memory performance (Vetreno et al., 2010). Therefore, cortical ACh dysfunction may occur as a consequence of a loss of basal forebrain cholinergic cells (described above) and may underlie some of the cognitive impairments common in the PTD model and WKS.

Selective white matter loss in the frontal and parietal cortices is common (Kril et al., 1997) as is a reduction in the size of the corpus callosum in alcoholics with WE (Lee et al., 2005). Similar cortical and callosal white matter loss is observed in the PTD model along with degeneration of the anterior commissure as well as external and internal capsules (Langlais and Zhang, 1997). Finally, thinning of myelin sheaths and reduced axon thickness occur in the corpus callosum of a rat TD model (He et al., 2007).

iii. Cerebellum, Midbrain and Brainstem Pathology in PTD

Neuroimaging (Sullivan et al., 2000; reviewed in Sullivan and Pfefferbaum, 2009) and postmortem (Baker et al., 1999) studies of alcoholics with WKS have revealed reductions in grey and white matter of the cerebellum including a reduction in the number of Purkinje cells in the vermis in individuals with WKS (Butterworth, 1993b). Pathological changes associated with WKS have also been reported to occur in the periaqueductal grey, superior and inferior colliculus, trochlear nucleus, and tegmental and pretectal regions of the human midbrain (Victor et al., 1989; Kril, 1996). Similar midbrain pathology is observed in rodents TD models (Troncoso et al., 1981; Irle and Markowitsch, 1982; Aikawa et al., 1984; Pfefferbaum et al., 2007). Pontine edema, especially within the superior and inferior vestibular nuclei, has also been reported both in individuals with alcoholism-associated WKS (Sullivan and Pfefferbaum, 2001) as well as in the PTD model (Takahashi et al., 1988; Watanabe and Kanabe, 1978). In particular, the superior and lateral pontine vestibular nuclei are damaged following PTD treatment, as are the medullary inferior olives (Troncoso et al., 1981). Damage to these structures likely contributes to symptoms like ataxia seen in humans with WKS as well as the PTD model. Further research is required to delineate the contributions of these brain regions to the diverse pathological profile of TD.

Neurochemical Changes

i. Amino Acids

Long-term neurochemical alterations have been reported in the PTD model 1–4 months after the restoration of thiamine levels. First, there are significant reductions in overall GABA and glutamate levels in the medial thalamus (Langlais et al., 1988). In addition, the amnestic effects of the glutamate antagonist MK-801 are increased four weeks after recovery from TD, indicative of disturbances in glutamatergic signaling (Savage et al., 1999). Carvalho and colleagues (2006) showed that 16-weeks after recovery from TD, rats were impaired on a water maze task and demonstrated reduced glutamate uptake in the prefrontal cortex (PFC). This is similar to the hypoactivity of PFC and impaired performance on memory and executive function tasks reported in fMRI studies of individuals with WKS (Paller et al., 1997; Reed et al., 2003; Caulo et al., 2005). Together these data strongly point to long-term alterations in glutamatergic function after TD.

Decreases in thalamic GABA levels have been reported soon after recovery from TD in the PTD model (3-days; Heroux and Butterworth, 1988). Similar reduction in GABA levels in the thalamus was observed 9 weeks after recovery from TD in the same model (Langlais et al., 1988). These observations suggest permanent alterations in the synthesis and/or utilization of both GABA and glutamate (described above) within the thalamus. Alterations in the balance between glutamate and GABA likely result in excitability changes within the thalamus, as well as signaling to thalamic efferents (i.e. thalamocortical projections), and processing of thalamic afferents (i.e. corticothalamic projections).

ii. Acetylcholine

Deficits in cholinergic function have been implicated in the long-term cognitive and behavioral impairments associated with WKS as well as the memory deficits seen in the PTD model (Savage et al., 2003; Pires et al., 2005; Savage et al., 2007; Roland et al., 2008). For example, one month after recovery from TD in the PTD model, rats display reduced hippocampal ACh efflux that is correlated with impaired behavioral performance on a spontaneous alternation task (Savage et al., 2003, 2007; Vetreno et al., 2008) and a delayed matching-to-sample task (Roland and Savage, 2007). Similar associations between impaired behavioral performance and reduced ACh efflux have been detected in other brain areas including the frontal cortex and retrosplenial cortex (Pires et al., 2005; Anzalone et al., 2010) suggesting widespread dysfunction of forebrain cholinergic activity. Pires and colleagues (2005) reported decreased levels of AChE activity in the cortex both during and after recovery from TD (3-months) in PTD-treated animals as well as reduced ACh release from cortical slices (Pires et al., 2001, 2005). Furthermore, PTD-induced decreases in the cortex were associated with learning impairments on spatial memory tasks such as the Morris water maze (MWM), suggesting that cortical ACh activity plays an important role in spatial learning (Pires et al., 2005). Interestingly, in the PTD model, ACh release is unaffected in other brain regions outside of the hippocampus and cortex such as the striatum (Vetreno et al., 2008) and amygdala (Savage et al., 2007).

Consistent with the role of cortical and hippocampal ACh in memory, administration of AChE inhibitors or herbal compounds that increase ACh release can rescue the cognitive performance deficits observed on tasks such as passive avoidance, water maze, and non-matching-to position after recovery from TD in rats (Nakagawasai, 2005, Roland et al., 2008). Reversal of memory deficits can also be achieved via blockade of GABA receptors in the medial septal region and suggests that balancing ACh and GABA levels in the septohippocampal circuit may prove to be an effective therapeutic approach for the treatment of cognitive impairments of TD in rodent models and potentially an avenue for the development of future pharmacological treatment of individuals with WKS.

iii. Monoamines

Reductions of NE metabolites in the cerebral spinal fluid of WKS patients are correlated with cognitive deficits (McEntee and Mair, 1978) and can be reversed by administration of α-adrenoreceptor agonists (McEntee and Mair, 1980; Mair and McEntee, 1986). Similarly, neurochemical analysis of PTD brains revealed long-term (13–15 weeks after TD) decreases in NE levels in the cortex and olfactory bulbs, with surprising increases in NE levels in the cerebellum (Mair et al., 1985; Langlais et al., 1987). To reconcile these disparate findings, it has been posited that cortical NE deficits may reflect select degeneration of long-range projection fibers from the locus coeruleus, whereas increased NE content in the cerebellum is produced via sprouting of proximal cerebellar afferents (Langlais et al., 1987).

Change in serotonin activity and its relation to memory impairment in WKS is not well understood. Although there is evidence for the loss of serotonergic neurons in the raphe nucleus of WKS patients (Halliday et al., 1993; Baker et al., 1996), there are limited case studies that document the recovery of cognitive performance following drug therapy with serotonin reuptake inhibitors (Numata, et al., 2005). In the PTD model, there is also a loss of serotonergic neurons in the raphe nucleus (Matsushita et al., 1999) but increases in serotonin and/or its metabolites in the thalamus and striatum following recovery from TD (Langlais et al., 1987, 1988; Vigil et al., 2010). A recent study demonstrated moderate correlation between high levels of the serotonin metabolite 5-hidroxyindolacetic acid in the thalamus and spatial memory performance on the MWM (Vigil et al., 2010). Such results suggest that altered serotonin levels may contribute to cognitive dysfunction associated with TD.

In conclusion, during the acute stages of TD as well as after recovery from TD there are significant changes in a number of neurotransmitters systems. Moreover, several neurochemical changes have been shown to correlate with cognitive deficits observed in both humans with WKS as well as animal models (Anzalone et al., 2010; Mair and McEntee., 1986; McEntee and Mair, 1978, 1980; Roland and Savage, 2009; Savage et al., 2003; Vigil et al., 2010). There is emerging evidence that restoring NE or ACh levels can lead to improved cognitive outcomes in WKS patients and TD animal models (Cochrane et al., 2005; Iga et al., 2001; McEntee and Mair, 1980; Roland et al., 2009, 2010) which are relevant to future design of therapies for individuals with WKS.

Behavioral Changes

After recovery from TD (1–12 months), studies using the PTD model have consistently demonstrated deficits on working memory tasks that are dependent on hippocampal function such as spatial matching-to-position (MTP) and nonmatching-to-position (NMTP). Impaired spatial memory and navigation on tasks such as the MWM and the radial arm maze are also observed. In particular, PTD-treated rats exhibit longer latencies to reach the hidden platform and demonstrate increased thigmotaxic searching behavior (circling the rim of the maze; Langlais et al., 1992) on the MWM which is similar to the type of memory deficits observed after experimental hippocampal lesion (Morris et al., 1982). In the radial arm maze task, PTD-treated rats show significant deficits in working memory for maze arms baited with food compared to modest reference memory impairments for unbaited maze arms (Robinson and Mair, 1992). In the radial-arm maze task, memory for baited arms is thought to model human episodic/declarative memory. In contrast, memory for unbaited arms is thought to model reference/non-declarative memory. Thus, the PTD model demonstrates greater impairment of episodic/declarative memory than reference/non-declarative, which is a behavioral finding very similar to that observed in WKS patients (see Gold & Squire, 2006; Kopelman et al., 2009). Recently, our lab has also demonstrated that PTD-treated rats are impaired on a spontaneous alternation task, a non-reinforced spatial memory task. Behavioral deficit on this task was correlated with impaired ACh release in the hippocampus (Savage et al., 2003; Savage et al., 2007; Vetreno et al., 2008) suggesting that TD results in deficits in cholinergic function in the hippocampus that manifests in memory deficits. Thus, despite a lack of neuropathological evidence indicating hippocampal damage in the PTD model, hippocampal function is severely compromised. Consistent with results observed in rodent TD models, memory deficits in WKS patients are similar to patients with hippocampal damage (after stroke, encephalopathy, etc) despite a lack of clear evidence of pathological damage to the hippocampus of WKS patients. Therefore, we posit that memory dysfunction in WKS stems from lesion to the thalamus (reviewed above) as well as loss of hippocampal cholinergic function.

Damage to midline thalamic nuclei is a key predictor of working memory impairment in the PTD model (Knoth and Mair, 1991; Mair et al., 1991; Langlais et al., 1992, Langlais and Savage, 1995; Mumby et al., 1999; Roland and Savage, 2007) as well as other diencephalic regions, such as the anterior thalamus and mammillary bodies (Langlais and Savage, 1995; Savage et al., 1997). However, not all memory function is altered by TD in the PTD model. For example, similar to WKS patients, there are some learning and memory tasks on which PTD-treated rats show preserved performance. These include simple discrimination tasks that are not dependent on hippocampal function such as light-dark (Mair et al., 1988) or left-right discrimination tasks (Vetreno et al., 2008). In contrast, performance is impaired on an object-based, delayed nonmatch to sample discrimination task in WKS patients (Squire et al., 1988) as well as after PTD treatment in nonhuman primates (Witt and Goldman-Rakic, 1983) and rats (Mumby et al. 1995). These effects are thought to be mediated by thalamic lesions caused by TD.

We recently found that during learning of a left-right discrimination task, there was reduced hippocampal but increased striatal release of ACh in the in PTD-treated rats compared to controls (Vetreno et al., 2008), suggesting that procedural memory systems display increased ACh activity in the face of pathological decreases in hippocampal ACh activity (described above). Similarly, we have found that there is no deficit when PTD-treated rats are trained using a Pavlovian procedure that pairs unique rewards (rat chow vs. chocolate milk) in a MTP task. More specifically, both PTD-treated and control rats are able to learn this version of the task in about 100 trials (Langlais and Savage, 1995). In contrast, without specific task-dependent reward pairing, PTD-treated rats require twice as many trails to learn the MTP task than control rats. Therefore, training protocols that incorporate Pavlovian-based learning reverse the effects of TD, suggesting that these cognitive demands recruit neural circuits that are resistant to TD-induced brain injury (Savage and Ramos, 2009). These data are important toward the development of behavioral learning approaches and strategies for use with individuals with WKS.

In addition to the anterograde memory deficits, individuals with WKS display retrograde memory loss - the inability to recall information learned prior to brain insult. Moreover, the hallmark feature of retrograde amnesia in WKS patients is that it is temporally graded; memory loss is greater for events that occurred immediately preceding brain damage whereas memory for events that occurred well-before the brain injury remains intact. In rodent models such as the PTD model, retrograde memory can be assessed by retesting animals on a task that was learned prior to PTD-treatment. Using this approach with the MWM, PTD rats re-tested after treatment did not display any deficits (Langlais et al., 1992; Pires et al., 2005). Lack of retrograde amnesia following PTD treatment was also reported on an object discrimination task (Mumby et al., 1999). Furthermore, pre-training rats prior to PTD treatment did not impair retrograde memory on a passive avoidance task (Langlais and Savage, 1995). This lack of retrograde memory loss was also reported in rats following electrolytic lesion to the midline thalamic nuclei (Langlais et al., 1992). Thus, although the PTD model recapitulates the anterograde memory impairment seen in WKS, to date, TD in rodents has not been shown to produce retrograde memory loss. Thus, rodent models of TD may not produce sufficient hippocampal dysfunction to create retrograde amnesia as is observed in humans. Alternatively, task-specific features used to measure cognitive performance in rodents such as the amount of training or the interval between learning, brain injury and retesting are not optimal for revealing retrograde amnesia (see Sutherland et al., 2010). Further investigation is needed to understand these differences between TD-induced retrograde memory deficits in humans and rodent models.

Concluding Remarks

Early understanding of human memory was derived from postmortem observation of human brain damage associated with symptoms of memory impairment. The seminal works of Wernicke (Thomson et al., 2008) and Korsakoff (1889) on cognitive impairment associated with TD not only provided insight into TD-induced brain pathology, but were also the impetus for the discovery of multiple memory systems within the brain. More recent technologies such as fMRI and positron emission tomography have allowed physicians and scientists to examine both structural and functional changes that occur in individuals with WE and WKS. Animal models, however, have allowed for the precise description of neurochemical changes that occur in particular brain regions both during and after recovery from TD. Together, these approaches have provided greater understanding of the cellular mechanisms, neuropathology, and clinical profiles of WE and WKS.

The PTD model of WKS has permitted greater understanding of brain structures involved in learning and memory deficits characteristic of WKS. Work from our laboratory employing the PTD model has identified hippocampal hypofunction as a major factor contributing to the cognitive impairments associated with TD despite a lack of gross anatomical pathology in the hippocampus compared to the thalamus and mamillary bodies (Roland and Savage, 2007; Savage et al., 2003; Vetreno et al., 2008). Work from our group (Pitkin and Savage, 2001, 2004; Roland and Savage, 2009b) and others (Arendt et al., 1995) has revealed a compromised basal forebrain cholinergic system as causal to hippocampal cholinergic dysfunction in the PTD model, an effect demonstrated by neurodegeneration of basal forebrain cholinergic neurons as well as reduced hippocampal ACh efflux in vivo. Future studies are necessary to understand why cholinergic neurons of the medial septum but not nucleus basalis are vulnerable to TD-induced degeneration (Savage et al., 2007) and to determine methods to protect these neurons at the earliest stages of injury.

A number of areas merit further exploration toward greater understanding of the relationship between brain and behavioral changes seen after TD. For example, recent work has demonstrated reductions in postnatal neurogenesis in the dentate gyrus in a TD rodent model (Zhao et al., 2008). Therefore, further research into the cellular consequences of TD is required to elucidate the potential involvement of reduced neurogenesis in cognitive dysfunction associated with TD. In light of the fact that numerous pharmacological and behavioral methods have been identified to increase neurogenesis in the rodent dentate gyrus such as following fluoxetine administration (Malberg et al., 2000) or voluntary exercise (van Praag et al., 2003), there is the potential to test the effect of these neurogenesis-stimulating treatments in the PTD model.

Although TD-induced deficits of cognition have been generally emphasized in both clinical studies and those using rodent models, less attention has been paid to those cognitive faculties that remain intact after TD. Using the PTD model, we have demonstrated preserved functional memory systems of the amygdala and striatum as well as increased ACh release in the striatum compared to controls (Vetreno et al., 2008). Future studies must focus greater attention toward understanding of the plastic brain changes that emerge after TD as they relate to learning and memory function and dysfunction.

Abbreviations

5-HT

serotonin

αKGDH

alpha-ketoglutarate dehydrogenase

ACh

acetylcholine

AChE

acetylcholinesterase

AVVL

anteroventral ventrolateral

DA

dopamine

DB

diagonal band

GABA

gamma-aminobutyric acid

ILn

intralaminar nucleus

IML

internal medullary lamina

MRI

magnetic resonance imaging

MS

medial septum

MTP

matching-to-position

MWM

Morris water maze

NE

norepinephrine

NO

nitric oxide

NMTP

nonmatching-to-position

PFC

prefrontal cortex

PTD

pyrithiamine-induced thiamine deficiency

ROS

reactive oxygen species

TCA

tricarboxylic acid

TD

thiamine deficiency

TPP

thiamine pyrophosphate

WE

Wernicke’s encephalopathy

WKS

Wernicke-Korsakoff Syndrome

VPL

ventroposterolateral