Cognition in PD and AD 245 9.3 Predicting Conversion from “Normal Aging” to MCI and from MCI to AD? . 272 9.4 Diagnosis of AD 273 9.5 Diet in AD 274 10 Imaging AD 274 10.1 Can Neuronal Dysfunction Be Visualized Before Cell Death? 274 11 APP Processing and Its Relation to Cognition 276 11.1 Amyloid Hypothesis 276 11.2 Aβ Extra- or Intracellular and in Which Compartment? 277 12 Revisiting the Unforgettable Tau 278 12.1 Aβ and Tau Interaction 278 13 Synaptic Dysfunction in AD 279 13.1 Is AD a Neuronal Disconnection Syndrome? 279 14 Future Perspectives 281 References 282 1 Introduction Parkinson’s disease (PD) is the second most common neurodegenerative disorder after Alzheimer’s disease. With increased longevity and improved health care, our society is experiencing an unprecedented challenge posed by these neurodegenera- tive disorders. Alzheimer’s disease alone is now the third most expensive disease to treat in the United States, costing close to $100 billion annually. The availability of genetically modified mice has advanced our understanding of several neurodegenerative processes. Cellular neurobiology experiments have informed us about mechanisms of neuronal dysfunction in AD and PD mouse mod- els. For instance, recent studies have identified that synaptic transmission is one of the earliest events in the cognitive abnormalities that characterize AD and PD. The integration of this information with data-based circuits modeling, in which neuronal electrical properties, synaptic transmission parameters, and brain oscillations can now be evaluated and it has been recently addressed in PD and AD. The recent availability of in vivo studies such as functional magnetic resonance imaging (fMRI) that can noninvasively obtain information about the metabolic state of the brain thus allows the use of longitudinal and cross-species studies designs in both neurodegenerative mouse models and patients. These types of work carry our knowledge from basic neuroscience to clinical neurology and back, to combine basic science with clinically used methods to address clinical issues in the neurobiol- ogy of aging and neurodegenerative diseases. Knowledge of basic aspects of cellular physiological changes induced by specific pathologies, that is, synuclein-related in PD or amyloid-related in AD is essential to identify potential therapeutic targets. The present chapter discusses data that exemplify relevant findings on synaptic and circuit abnormalities in relation to PD and AD and clinical information that may correlate with such abnormalities or complement it. 246 I. Bodis-Wollner and H. Moreno 2 Parkinson’s Disease: An Overview Parkinson’s disease is a progressive neurological disorder that affects critical domains of daily living. Since its first description as a “shaking palsy” it has been recognized that it affects many nonmotor functions. PD causes autonomic, sensory, cognitive, and behavioral problems all of which can significantly impair quality of life. The original observations by Carlsson and his collegues (1957) that 3-4- Dihydroxyphenylalanine reverses reserpine-induced akinesia in rabbits led to the first clinical use of levodopa by Birkmayer and Hornykiewicz (1961). Cognitive impairment, labeled as moderate to marked “dementia” was first reported in a large number of patients to occur tenfold more commonly than in controls (Lieberman et al., 1979). Demented PD patients in this study responded less well to levodopa therapy and it was suggested that PD with dementia may represent a different disorder from PD without dementia. We distinguish select cog- nitive impairments from dementia in PD and discuss neurochemical and molecular mechanisms of cognition in PD. Cognitive impairment in PD is traditionally specified with neuropsychological testing, performed with clinically validated methods. In the recent decade, however, considerable knowledge was gained from imaging (PET and fMRI) and EEG and “event”-related potentials (ERP) studies. They use specific experimental cognitive paradigms to tax select cognitive operations, such as executive functions, short-term memory, and visuospatial orientation. In the last decades, administration of new forms of therapy in addition to levodopa, has led to a better understanding of the role of dopaminergic and non- dopaminergic circuits in PD. Many scientists are engaged in trying to specify the neurochemical building blocks of PD and develop a rational pharmacotherapy of the whole disease and not just its motor manifestations. There are several promis- ing avenues of applying so-called “dirty drugs,” that is, medications that have more than one effect in the CNS. The circuitry responsible for cognitive and some other manifestations of the disease is still based on the core abnormality of dopaminer- gic deficiency in the basal ganglia, however, our understanding of the basic basal ganglia circuitry has undergone revision and refinement. One contribution to a better neuroanatomical understanding is based on the effects of nondopaminergic manipulations, in particular transcranial magnetic stimulation (TMS) and deep brain stimulation (DBS). DBS as a treatment for PD was introduced nearly two decades ago. The elec- trophysiological results obtained with DBS have contributed considerable new knowledge regarding the basal ganglia/cortex circuits responsible for the manifesta- tions of PD. Human intraoperative monitoring and electrophysiological recordings combined with microdialysis have yielded biochemical evidence on the role of GABA and the subthalamic and the pedunculopontine nucleus in the PD. We discuss the evolution of the development of PD based on the neuropatho- logical studies of Braak and collegues. Based on the distribution of Lewy bodies Cognition in PD and AD 247 Fig. 1 In the CNS, the pathological process of PD commences in the dorsal motor nucleus of the vagal nerve (medulla oblongata) from where it follows an ascending course, affecting additional nuclei in the lower brainstem, in basal portions of the mid- and forebrain, and eventually reaching the cerebral cortex (Braak et al., 2004). However, as summarized, several neuronal circuits that are relevant to cognitive/emotional changes in PD are also involved. In these circuits not all DA neurons have long axons at different stages of PD, Braak et al. (2002) suggested that PD progresses from peripheral to central long axon projecting neurons in a caudal-cranial direction (Fig. 1). Braak’s neuroanatomical model predicts relatively late occurrence of cog- nitive changes in PD, but not all clinical data and observations fit neatly into the model. One defined dopaminergic circuitry affected in PD is the retina. The retinal dopaminergic circuit is of interest from several points of view. For one, it provides an opportunity to study the logic role of diverse dopamine receptors. The retina is a multilayered structure with a single output line and three layers of neurons with lateral and feedback connections. Second, many patients with PD suffer from visual hallucinations and vision in PD can be to a great degree attributed to specifically impaired retinal processing. Anatomical and electrophysiological data show both in humans and in MPTP-treated monkeys that retinal dopaminergic deficiency may be one reason for visual dysfunction in PD. Neuropharmacological manipulations in the monkey model of PD have led to a better understanding of the “antagonistic” role of D1 and D2 type recptors. The retina has not been subjected to studies for evaluating the presence of alpha synuclein. On the other hand, animal and human studies suggest a correlation of motor disease progression with visual dysfunction. Affected retinal dopaminergic neurons do not have long projections. Recent imag- ing data, however, show neuronal thinning involving both inner nuclear layers and retinal ganglion cells. Retinal ganglion cells are among neurons having the longest axons in the CNS. Whether PD retinopathy starts in a subset of ganglion cells, before (consistent with the postulates of Braak) the process attacks dopaminergic neurons, is unknown at present. 248 I. Bodis-Wollner and H. Moreno 3 Neurobiology of Parkinson’s Disease 3.1 Etiology and Molecular Progression of PD The etiology of PD still eludes us; however, an understanding of cognitive impair- ment, linked to the classical concepts of PD, is emerging. Since the original description of Parkinson’s disease in 1817, as shaking palsy that does not affect the senses, the last decades have witnessed a slowly emerging consensus that PD affects movement, sensation, cognition, and mood. Furthermore these noncardinal features do not respond well to dopaminergic therapy. Recent pathological data (Braak et al., 2002) suggest a caudo-cranial evolution of the dis- ease progressing from nondopaminergic and nonmotor extracerebral neurons to basal ganglia to cortical neurons. Hence some nonmotor symptoms may be herald- ing PD. The possibility of precardinal and premotor diagnosis is challenging from a therapeutic point of view Although the MPTP model is nearly perfect once developed, it is a toxic model that occurs suddenly after the i ntroduction of the toxin, whereas PD is a progressive disease. Nevertheless the elucidation of the action of MPTP on the mitochondrial respiratory chain, more precisely on the step between I and II has led to the search to the search for agents with neuroprotective abilities in the mitochondrial respiratory chain. 3.2 PD as a Synucleinopathy The degenerative process of idiopathic PD (iPD) is associated with the anatomical progression of CEB1-synuclein into select neurons. It is an aggregate of the mis- folded protein and appears in dendrites and axons, Lewy neurites (LNs), as well as punctuate structures and/or Lewy bodies (LBs) in the somata of involved nerve cells. It is thought that only projection neurons with a long axon become involved, whereas short-axoned cells resist the pathology. It has been proposed that vulnerable brain regions in PD are anatomically interconnected. Before the era of synuclein immunocytochemistry, Qualman et al. (1984) observed in a postmortem study LBs in the esophageal Auerbach plexus of two dysphagic PD patients but not in Meissner’s plexus. Subsequently, Wakabayashi et al. (1988) reported LBs and LNs in both plexuses of clinically diagnosed PD patients and asymptomatic incidental cases. In the gut, the bulk of the proteins were observed in cellular processes and cell bodies of vasoactive intestinal polypeptide (VIPergic) neurons (Wakabayashi et al., 1988, 1993). Braak and collegues (2006) suggested that the process starts in the neuronal plexus of the GI tract. They used immunocytochemisty to investigate gastric myen- teric and submucosal plexuses in five autopsy individuals, whose brains were also staged for Parkinson-associated synucleinopathy. CEB1-synuclein immunoreactive inclusions were found in neurons of the submucosal Meissner plexus, whose axons project into the gastric mucosa and terminate in direct proximity to fundic glands. They suggested that a yet to be identified environmental pathogen capable of passing Cognition in PD and AD 249 the gastric epithelial lining might induce CEB1-synuclein misfolding and aggre- gation in specific cell types of the submucosal plexus and reach the brain via a consecutive series of projection neurons. Some of the axonal aggregations occurred directly beneath the epithelial lining of the stomach, therefore they suggested that alpha synuclein pathology of the submucosal plexus of Meissner could represent the beginning of an “uninterrupted series of projection cells that ultimately link the Enteric Nervous System with the cerebral cortex.” In the CNS, the process appears to commence in the dorsal motor nucleus of the vagal nerve (medulla oblongata) and in the olfactory bulb from where it follows an ascending course, affecting additional nuclei in the lower brainstem, in basal portions of the mid- and forebrain, and even- tually reaching the cerebral cortex (Braak et al., 2004). However, as summarized in the following, several neuronal circuits that are relevant to cognitive/emotional changes in PD are also involved. In these circuits not all DA neurons have long axons. It should be noted that the Braak scheme is based on alpha synucleinopathy. More understanding is needed before PD is accepted as generalized alpha synucle- inopathy and in the Braak scheme a specific vulnerability of dopaminergic neurons, unless they have long axons. There is little knowledge at this point about what alpha-synuclein does in and to the cell and it is not yet clear why some long axon dopaminergic neurons are particularly affected in PD. 4 Basal Ganglia Circuit 4.1 Central Role of Dopamine in PD Parkinson disease is a clinical diagnosis based on four essential cardinal symptoms. These are rigidity, tremor, bradykinesia, and loss of postural reflexes. The clini- cal diagnosis is about 80% consistent with postmortem histopathological diagnosis. Clinically a number of other features are considered for postmortem diagnosis, but the essential pathology is the depigmentation and reduced number of tyrosine hydroxylase (TH)-labeled dopaminergic neurons in the area known as the substan- tia nigra. This deficit in turn causes impaired transmission between presynaptic dopaminergic neurons and postsynaptic dopamine receptors. This essential feature, a specific deficiency of dopaminergic neurons, is consistent with the spectacular clinical success of rational pharmacotherapy. This consists of treating patients with medications that promote presynaptic dopamine content in releasable synaptic vesi- cles and with ligands that directly bind to postsynaptic dopamine receptors. In the early 1980s the chance observation was made that methyl-phenyl tetrahydropyridine (MPTP) causes in man and monkeys a syndrome phenotypically almost identical to Parkinson disease. MPTP selectively affects dopaminergic neurons via the phys- iological high-affinity uptake system of TH-labeled neurons. These observations led to pathological studies showing degeneration of dopaminergic neurons in the substantia nigra as in PD and that the monkeys and humans affected by MPTP do respond to dopaminergic treatment. The selective effect of MPTP is well explained 250 I. Bodis-Wollner and H. Moreno by the high-affinity uptake system of dopaminergic neurons; in this case MPTP finds its targets via the same system. In this chapter we review the classical model of PD and summarize the two “direct” and “indirect” loops of the dopaminergic circuit of the striato-caudal system and the “antagonistic” effects of D1 and D2 type receptors. The retina contains both types of receptors in humans and in the monkey model of PD. We summarize the retinopathy of PD, the retinal dopaminergic circuit, and the functional significance of D1 and D2 antagonism in the retina. Then we review some newly described roles of adenosine in the classic basal ganglia circuit and the molecular mechanisms associated with adenosine in the circuit. The antagonistic effects of D1 and D2 type receptors in neuronal responses have been considered an enigma for a long time. Recent biochemical studies revealed a cascade of molecular events that explain in part t he role of D1 and D2 receptors in cognitive dysfunction in PD. 4.2 The Classical Basal Ganglia Circuit The classical model of the cortico–basal ganglia–cortical circuit, with its indirect and direct pathways, was developed to explain the phenomenon of hypokine- sia in Parkinson’s disease (Albin et al., 1989; Alexander and Crutcher, 1990; DeLong, 1990). The most researched cortico–subcortical circuit is the “motor circuit” because of its importance for movement disorders. The motor circuit is composed of several subcircuits that originate from the motor cortex and several premotor areas (Fig. 2). In a general sense, tonic output from this circuit, arising in motor portions of the GPi and SNr, may regulate the overall amount of movement. Increased basal ganglia output could translate into less movement through inhibi- tion of thalamocortical projection neurons, whereas reduced basal ganglia output could translate into increased movement because of disinhibition of these neurons. Although no direct evidence is available, it has been proposed that the combined action of information traveling via the direct and indirect pathways may scale or focus movements. To achieve scaling of movement parameters or termination of movements, striatal output would initially inhibit specific neuronal populations in the GPi and SNr via the direct pathway, hence facilitating movement, followed by disinhibition of the same GPi and SNr neuron via input over the indirect pathway, thus inhibiting ongoing movement. In the alternative focusing model, inhibition of relevant pallidal and nigral neu- rons via the direct pathway would allow intended movements to proceed, whereas unintended movements would be suppressed by concomitant increased excitatory input to other GPi and SNr neurons via the indirect pathway. The balance between direct and indirect pathways is regulated by the differential actions of dopamine on striatal neurons from terminals of neurons in the substantia nigra pars compacta. Release of dopamine in the striatum increases activity along the direct pathway (acting on D1 receptors in striatal neurons) and reduces activity along the indirect pathway (acting on D2 receptors). Together these actions result in a net reduction Cognition in PD and AD 251 Fig. 2 A simplified diagram, based on Wichmann and DeLong (2003). The cortical motor areas give rise to a specific motor subcircuit. Red arrows: inhibitory (γ-aminobutyric acid [GABA]–ergic) connections; green arrows: excitatory (glutamatergic) connections. GP t he globus, substantia nigra pars compacta; SNr, substantia nigra pars reticulata; STN, The sketch is only a skeleton of the much more intricate interconnected system of the basal ganglia with the cortex. These principal basal ganglia–thalamocortical circuits establish a balance between excitatory and inhibitory neurotransmission. The categorical division between the so-called direct (D1-receptor linked) and the indirect (D2-receptor linked) striatal output pathways may be complicated though (not shown here) by pre-synaptic receptor mechanisms. Presynsaptic dopamine receptors have higher affinity to dopaminergic ligands than do postsynaptic receptors in GPi and SNr activity. Conversely, a decrease in striatal dopamine release would result in an increase in GPi and SNr activity. The two pathways regulating basal ganglia output via a balance of D1 and D2 receptors bears a certain logic similar- ity to their role in the retina regulating its output neurons, the retinal ganglion cells (Fig. 3). Increased understanding of the anatomy and function of the basal ganglia and their role in motor and nonmotor disorders (Bodis-Wollner et al., 1983) now posits the basal ganglia at the core of cortical connections. The basal ganglia are now seen (DeLong and Wichmann, 2007) as components of parallel, re-entrant cortico– subcortical circuits, which originate from individual cortical areas, traverse the basal ganglia and thalamus, and connect to a number of separate neuronal groups termi- nating in the frontal cortices. Dopamine (DA) is a powerful neuromodulator for a wide variety of behaviors and some sensory processing. The working hypothesis of cognitive deficits in PD is that they result from impairment of specific cortico–subcortical circuits. These circuits in part depend on dopamine-linked synapses. Although the original model (see Fig. 1) has been modified, the essential role of dopamine deficiency of the basal ganglia remains at the core of cognitive deficiencies in PD involving the frontal cortices. 252 I. Bodis-Wollner and H. Moreno Fig. 3 The imbalance of striatal GABAergic output pathways in PD (after Alexander and Crutcher, 1990. Blue arrows: inhibitory (γ-aminobutyric acid [GABA]–ergic) connections; red arrows: exci- tatory (glutamatergic) connections. GP, globus pallidus; SNc, substantia nigra pars compacta; SNr, substantia nigra pars reticulate) and presumed sites of commonly applied therapies (courtesy of A. Mori) 5 Frontal Cortices, Striatum, and Cognition in PD 5.1 Fontostriatal Circuits in PD The frontostriatal circuits connect the basal ganglia with cortical areas that are involved in cognitive, motor, and emotional processes. Furthermore, a correlation between cortical DA innervation and expression of cognitive capacities, including learning, has been shown by a number of studies (Nieoullon, 2002). Therefore a role of dopamine in impaired cognitive processing in PD is not surprising, Considerable evidence accumulated from rodent and monkey experiments over the last two decades suggests that DA activity in the frontal cortex is reciprocally linked to that in functionally related basal ganglia (BG) structures. However, the functional importance of this in humans is still unknown. To address this issue, we measured endogenous DA release using positron emission tomography in 15 healthy sub- jects as they practiced the first training session of a finger sequence learning task. Significant results were observed not only in striatal areas but also i n extrastriatal “motor” regions, bilaterally. Faster learning was specifically coupled to lower DA Cognition in PD and AD 253 release in the sensorimotor part of the globus pallidus pars interna (GPi) contralat- eral to the moving hand, which was paralleled by a higher increase in DA levels in the presupplementary motor area (pre-SMA). This finding provides original evi- dence supporting a motor-learning-related interaction between DA release in left GPi and pre-SMA, a mechanism that may also apply to other anatomically and functionally interconnected BG and frontal cortical areas as a function of behavior. Cortical activity during executive processing in PD depends on striatal mecha- nisms even in early stages of the disease, as shown with functional imaging (Monchi et al., 2007). They have developed a new card-sorting task t hat is known to require frontal involvement and represent executive functions including planning and set- shifting. They have first shown that in young healthy adults, the caudate nucleus is specifically required when a set-shift must be planned. They studied a group of early-stage Parkinson’s disease patients (seven right-handed patients at Hoehn and Yahr stages 1 and 2; mean age 62 years, range 56–70) and matched control subjects. Decreased cortical activation was observed i n the patient group in the condition significantly involving the caudate nucleus. Their study showed a pattern of either reduced or increased activation depending on whether the caudate nucleus was involved in the task. This activation pattern included not only the prefrontal regions but also posterior cortical areas in the parietal and prestriate cortex. These findings are not in agreement with the traditional model, which proposes that the nigrostri- atal dopamine depletion results in decreased cortical activity. These observations provide further evidence in favor of the hypothesis that not only the nigrostriatal and frontal mechanisms are affected in PD executive dysfunction. 5.2 Impaired Memory in PD: Thalamocortical Circuitry Aggleton and Brown (1999) proposed two parallel brain systems with qualitatively different contributions to memory. The proposed functional-anatomical division is of relevance to cognition in PD. Hay et al. (2002) investigated memory performance in patients with either mild Parkinson’s disease, moderate Parkinson’s disease, or amnesia using measures of habit (automatic memory) and conscious recollection (intentional memory). Patients with amnesia displayed the expected dissociation between (intact) habit learning and (deficient) conscious recollection, patients with moderate PD were impaired on both measures whereas the mild PD patients had no abnormalities on either one. Hay et al. (2002) attributed the deficiencies in habit learning to striatal dysfunction, whereas conscious recollection is thought to result from disruption of prefrontal cortical processing. Evidence of executive dysfunction Parkinson’s disease patients is also consis- tent with proposals that frontostriatal circuit damage produces widespread prefrontal dysfunction (e.g., Buytenhuijs et al., 1994; Leplow et al., 1997; Owen et al., 1998; Dujardin et al., 2001). It has been recognized recently that the thalamic dopaminergic system degen- erates in PD. In addition to the GABAergic thalamic input to the cortex it has now been recognized that the nigrothalamic connection is affected in humans 254 I. Bodis-Wollner and H. Moreno (Garcia-Cabezas et al., 2007a) and in MPTP monkeys (Sanchez-Gonzalaez et al., 2005; Garcia-Cabezas et al., 2007b). However, the contribution of the dopaminergic thalamocortical system to cognition has not been elucidated. Deep brain stimulation (DBS) has yielded quantitative information on the diver- sity of different functional loops between cerebral cortex and the subthalamic area in PD. These studies suggest that functional subloops between the subthalamic area and cerebral cortical motor regions can be distinguished by their frequency, cortical topography, and temporal relationships (Fogelson et al., 2006). Tuning to distinct frequencies may provide a means of marking and segregating related processing, over and above any anatomical segregation of processing streams. They recorded EEG and local field potentials (LFPs) from macroelectrodes inserted into the subthalamic nucleus area in nine awake patients following func- tional neurosurgery for PD. Patients were studied after overnight withdrawal of medication. Coherence between EEG and SA LFPs was apparent in the theta (3– 7 Hz), alpha (8–13 Hz), lower beta (14–20 Hz), and upper beta (21–32 Hz) bands, although activity in the alpha and upper beta bands dominated. Theta coherence predominantly involved mesial and lateral areas, alpha and lower beta coherence the mesial and ipsilateral motor areas, and upper beta coherence the midline cor- tex. SA LFPs led EEG in the theta band. In contrast, EEG led the depth LFP in the lower and upper beta bands. SA LFP activity in the alpha band could either lead or lag EEG. In PD patients undergoing surgical procedures one may record enhanced beta range oscillatory rhythms in the BG and effective l evodopa medication attenuates beta activity. The dominant cortical synaptic input to the BG and to the ST is gluta- matergic. Given this fact, it is likely that local BG oscillatory rhythms derive from cortical inputs, but not dopamine. Dopamine is not known to regulate quick synap- tic currents via phasic ionotropic mechanisms (Traub et al., 2008) hypothesize that abnormal oscillations which occur in PD arise at the cortex, even though recorded deeper. Trottenberg et al. (2006) r ecorded oscillatory activity in the gamma frequency (60–100 Hz) band in local field potentials (LFPs) recorded from the region of the subthalamic nucleus (STN) in PD patients. Spike-triggered averages of LFP activ- ity suggested that the discharges of neurons in this region were locked to gamma oscillations in the LFP. They suggested that gamma band oscillations in the LFP are likely to represent local neuronal discharge. Gamma activity (Gray and Singer, 1989; Singer, 1993; Singer, 1999) reflects syn- chronization of thalamo-cortical neuronal groups in order for them to act together, a necessary prerequisite for voluntary motor action and for forming a coherent percept (Joliot et al., 1994; Tallon-Baudry et al., 1997). Fixation on visual stimuli that are optimal for foveal processing results in a time-locked increase in the power of the gamma component of the human EEG (Tzelepi et al., 2000; Bodis-Wollner et al., 2001). When saccades are performed either in the presence of visual stimuli or in the dark, gamma range activity is enhanced in the intrasaccadic period over the pari- etal and occipital cortices (Bodis-Wollner and Tzelepi, 2002; Forgacs et al., 2008). . dysfunction. Affected retinal dopaminergic neurons do not have long projections. Recent imag- ing data, however, show neuronal thinning involving both inner nuclear layers and retinal ganglion cells. Retinal ganglion. passing Cognition in PD and AD 249 the gastric epithelial lining might induce CEB1-synuclein misfolding and aggre- gation in specific cell types of the submucosal plexus and reach the brain via. movement, followed by disinhibition of the same GPi and SNr neuron via input over the indirect pathway, thus inhibiting ongoing movement. In the alternative focusing model, inhibition of relevant