Keywords

Huntington’s Disease (HD) and Mouse Genetics

Memory disturbances due to brain damage or neuropathology have greatly contributed to our understanding of how memories are organized in normal brain. Many neurological diseases are associated with severe cognitive impairments, and some of them are caused by the mutation of one or multiple genes encoding for proteins or enzymes important for nerve cell functions. Powerful mouse genetics have been used to dissect the pathogenesis of these diseases, which provide an interesting setting for the investigation of the neurobiological basis of learning and memory, and behaviors in pathological conditions. This is the case, for example, for Alzheimer’s disease, Fragile X syndrome, and HD to cite only a few.

HD is particularly interesting because the disease is caused by the mutation of a single gene, which was cloned 20 years ago by the collaborative HD research group (1993). The mutation is transmitted by inheritance in an autosomal dominant manner. The mutated HD gene is located in chromosome 4 and encodes for Huntingtin protein whose functional roles are not well established. The mutation is characterized by an expansion of a trinucletide CAG repeat coding for glutamine, which results in toxic protein fragments that accumulate in the cell in the form of aggregates.

While the (mutated) protein is widely expressed (DiFiglia et al. 1995; Gutekunst et al. 1999), this inherited disease is associated with selective neurodegeneration of the basal ganglia and more specifically the striatum (Vonsattel et al. 1985). Accordingly, motor disturbances and chorea are cardinal clinical features. However, cognitive disturbances expressed in the form of cognitive inflexibility and/or working memory deficits as well as psychiatric problems often precede the motor impairments by several years (Josiassen et al. 1983; Lawrence et al. 1996; Lawrence et al. 2000; Lemiere et al. 2004; Vonsattel et al. 1985). Other cognitive impairments are reminiscent of deficits observed with striatal (and cortex) pathology, which is known to support stimulus–response (S–R) associations, habit formation, procedural, nondeclarative, or implicit memory (Graybiel et al. 1994; Squire 1992; Yin and Knowlton 2006); incremental processes taking place with repeated trials and errors. These findings are consistent with the neurobiological based attribute theory of memory in which the striatum subserves a response attribute which involves memory representations based on feedback from motor responses (often based on proprioceptive and vestibular cues) that occur in specific situations (Kesner and Rogers 2004).

The vulnerable striatum constitutes the main output structure of the basal ganglia, an ensemble of regions formed by the cortex, globus pallidus, subthalaminc nuclei, thalamus, and substance nigra (Fig. 14.1). The striatum contains heterogeneous cell populations. Principal cells, also called medium spiny neurons (MSNs), send axons to downstream regions in the basal ganglia loop, and several classes of local interneurons (IN) form intricate and complex regulatory and interactive networks. One subclass of MSN that forms so called “indirect pathway” is affected earlier than “direct pathway” MSN neurons in HD. The indirect pathway MSNs project via internal segment of pallidum and subthalamic nuclei to substantia nigra, and the direct pathway MSNs project directly to the substantia nigra (Reiner et al. 1988). While reasons for this differential vulnerability are not well understood, a new hypothesis suggests that the imbalance of activities between the two pathways, caused by dysfunctional indirect pathway and maybe compensatory activity of direct pathway, may constitute a core pathological process in HD (Andre et al. 2011b; Andre et al. 2011a).

Fig. 14.1
figure 1

Simplified basal ganglion loop with indirect and direct pathways. MSN medium spiny neuron, GPe external globus pallidus, STN subthalamic nuclei, SNc substance nigra compacta, GPi internal globus pallidus, SNr substance nigra reticulata

Full size image

As mentioned previously, the hereditary nature of the disease transmission has made the transgenic mouse the prime choice to model HD in animals. Upon cloning of the HD gene in 1993 (The Huntington’s Disease Collaborative Research Group 1993), several transgenic mouse lines (i.e., R6 lines, YAC128 mice, and knock-in mice) have been produced, with which the effects of the gene mutation could be explored from molecule to behavior. As such, a plethora of symptoms described in HD patients has been reproduced in HD mouse models, even though no one model reproduces all the symptoms seen in patients. Phenotypic changes in these transgenic mice concerned changes of biochemical and molecular properties in vulnerable cell populations (Cepeda et al. 2007; Cha 2007; Cha et al. 1999; Ferrante 2009), as well as behavioral and cognitive abnormalities. The latter domain is associated with deficits in visuospatial working memory, cognitive flexibility, emotionality, social interaction, and circadian rhythm among others (Brooks et al. 2006; Dumas et al. 2013; Lione et al. 1999; Morton et al. 2005; Naver et al. 2003; Nithianantharajah et al. 2008; Pietropaolo et al. 2011). Impaired synaptic plasticity and functional changes in the vulnerable corticostriatal pathway (Cummings et al. 2007; Cummings et al. 2006) have been associated with these symptoms.

Striatal Activity Changes in HD Transgenic Mice

While striatal degeneration is an incontestable hallmark of HD, some postmortem studies have shown only limited signs of cell loss in the brain including the striatum despite substantial clinical evidence of HD (Vonsattel et al. 1985). This suggests that neuronal dysfunction, rather than cell death, may contribute to behavioral manifestations, especially at early stage of HD. In line with this idea, most symptoms of HD have been reproduced in mouse models in which neuronal loss is not present at all. The nature of neural dysfunction in HD, in particular information coding process mediating cognitive and behavioral disturbances is not known, and mouse models with strong construct validity present a unique opportunity to explore this issue, which is not possible in human patients.

We thus recorded single-unit activity and local field potentials (LFPs) in the dorsal striatum of motor presymptomatic HD transgenic mice (i.e., R6/1 mice) and their nontransgenic littermates while they acquired and performed an operant conditioning task. In this task, mice learned to associate a nose-poke action with the arrival of a reward at the food port. Because it has been shown that striatal activity undergoes plasticity and displays activity changes in normal rodents throughout learning (Barnes et al. 2005; Jog et al. 1999), our recordings were initiated with behaviorally naïve mice and continued until they reached asymptotic performance levels (Cayzac et al. 2011). Since these HD mice displayed impaired synaptic plasticity in the cortico-striatal pathway, we hypothesized that learning related changes in cell firing may be absent in the transgenic mice (Cepeda et al. 2003; Cepeda et al. 2007; Cummings et al. 2007; Cummings et al. 2006; Cybulska-Klosowicz et al. 2004).

We chose to study the R6/1 line, which displayed delayed onset and progressive disease phenotypes, reminiscent of adult onset HD (Mangiarini et al. 1996). These mice exhibit impaired ability for spatial short-term (Grote et al. 2005; Naver et al. 2003; Nithianantharajah et al. 2008; Pang et al. 2009; Lebreton et al. 2015) and long-term memory (Nithianantharajah et al. 2008) at 10–12 weeks of age, abnormal social behaviors at 13 weeks (Pietropaolo et al. 2011), and impaired motor coordination and muscle strength at 18–20 weeks of age (Naver et al. 2003; Lebreton et al. 2015). The mice have a 6–7 month lifespan. Although mutant protein aggregates begin to appear at 8 weeks of age, no cell loss is observed even at the time of death (Naver et al. 2003; Nicniocaill et al. 2001). This progressive and sequential onset of phenotypic events is well adapted for studying early changes in neural processing associated with cognitive disturbances due to the HD mutation.

The activity of striatal cells recorded from wild-type (WT) littermates was tuned to the trajectory between nose-poke area and the food port, action preparation and initiation, and reaching the goal (food well) rather than the operant nose-poke behavior per se (Fig. 14.2). A majority of cells responded to these behavioral/task stimuli and action by increasing their discharge rates, and only a small (< 5 %) proportion of striatal cells decreased discharge rates especially during reward consumption period (Fig. 14.2). This was the case for not only phasically active MSN, but also IN, most of which are suspected to be the ones expressing parvalbumin because of their well-known tonic discharge characteristics (Berke et al. 2004; Mallet et al. 2005). Our data confirm the recent controversial findings that striatal IN do display firing patterns that are sensitive to behaviors (Berke 2008; Gage et al. 2010; Stalnaker et al. 2010). In addition, the proportion of cells responding to task elements increased as mice acquired the task from early (60–75 % correct) to late (75–100 % correct) stages of learning (Fig. 14.2f). Surprisingly, striatal cells recorded from R6/1 mice also displayed similar task sensitivity as well as plastic changes of firings; the proportion of task sensitive neurons increased as mice learned the task (Fig. 14.2e-f). This suggests somewhat normal integrity of striatal function despite the HD pathology in these mice.

Fig. 14.2
figure 2

Task-sensitive activity in the striatum. a Photographs of a mouse with an electrode implant and operant chamber with nose-poke holes placed in its three inner walls. A food port was placed beneath the fourth wall. b–d Rastor plots color coded (with white to black indicating high to low activity levels) and average firing rates (circular peri-event time histogram) for three representative striatal cells across three (A, B, and C) sequential periods repeated during the nose-poke task. The three periods were bordered by nose-poke, animal’s arrival at the food port marked by first lick, exit of the reward port and nose-poke again. The period A was repeated to illustrate the circular nature of the behavioral sequences. e Percent modulated cells during the three periods in both genotypes. f Percent modulated projection cells (MSN) and interneurons (IN) throughout four different stages of the operant conditioning in both genotypes

Full size image
Fig. 14.3
figure 3

Relationship between the scarce recruitment of striatal projection cells (MSN) in R6/1 transgenic mice and learning rate. a Cell counts of recorded interneurons (IN) and projection cells (MSN) distinguished by spike width (ms) in both R6/1 and WT littermates. b Example of learning curb for representative WT and R6/1 mice across three consecutive daily learning sessions (30 min). c and d Cartoons showing a diminished recruitment of MSNs and a retarded learning rate in R6/1 mice. Filled circles in (c) symbolize activated MSN and empty circles represent inactivated MSN

Full size image

One intriguing observation was that the recorded neurons in the HD transgenics were mostly IN, while they represent only a half in proportion among the recorded neurons in WT mice (Fig. 14.3a). Even though MSNs represent 95–97 % of the striatal cell population, they are known to be silent in most physiological conditions, explaining such a low proportion (i.e., about 50 %) of MSN recorded in WT mice. As no cell loss has been reported in R6/1 mice, we suspected that most of MSN in transgenic mice did not reach spike firing threshold, and thus were dysfunctional. This activity pattern was associated with severe learning impairments. R6/1 mice required 4–5 times more sessions to attain an asymptote performance level, which in addition, remained significantly lower than the level reached by WT littermates. When examining behavioral performances closely, retarded learning of the transgenics was in part due to difficulties in maintaining learned information over days, such that they seem to learn freshly at each session even at an asymptote stage (Fig. 14.3b). We noticed that interrupting training for one or two daily sessions was enough to disrupt significantly the learned operant behaviors. This might suggest that incremental learning relying on long-term memory consolidation requires an intact recruitment of striatal projection cells, which ensures the proper information transfer and flow in the basal ganglion circuit (Fig. 14.3c, d). The local interneuronal activity, which appeared normal in our transgenic mice, may not be sufficient to support the normal information flow and learning. Abnormalities in other brain regions or other physiological changes (e.g., motivation level) may also contribute to the behavioral changes seen in our mice, because, for example, the hippocampus also undergoes the cellular abnormalities in HD patients and R6/1 mice. However, because hippocampal lesioned mice were able to perform this task perfectly (Cho and Jeantet 2010), a hippocampal contribution to the learning abnormalities seems unlikely.

In addition, when striatal cell firing patterns in HD transgenics were closely examined, their discharges were abnormally regular and presented rhythmic firings at high frequency (60–80 Hz) referred to as high γ oscillation (Fig. 14.4a, c). Both MSN and IN displayed this abnormal firing regime in HD transgenics, while no such phenomenon was present in WT littermates. Therefore, not only the scarce recruitment of the principal projection cells during task learning, but also excessive γ synchrony in intrinsic firings characterizes “abnormal” striatal activity in HD mice. Pairs of neurons recorded either from the same tetrodes or tetrodes placed bilaterally also showed common synchronous discharges at this γ range (Fig. 14.4b, d). This suggests a more global synchrony within (or maybe beyond) the cortico-striatal loop. This was also confirmed by an additional observation that motor cortical cells above the striatum also resonate in the γ frequency.

Fig. 14.4
figure 4

High γ-oscillation in the striatum in R6/1 a and b Autocorrelogram and crosscorrelogram of striatal cells displaying γ synchrony. c and d Relative peak γ frequency distributions of single (c) and paired (d) cells. e Spatial positions of an R6/1 mouse in the operant chamber when his striatal LFP expressed high γ synchrony (blue dots, green dots show visited pixels)

Full size image

When we looked at what precise moments during behavioral testing the γ oscillation in the population activity (reflected in the LFPs) becomes abnormally strong, we found that this activity was prominent when mice were immobile during reward consumption. It is known that dopamine (DA) is released while animals receive reward, and DA transmission is altered in HD (Andre et al. 2010; Cha et al. 1999; Chen et al. 2013; Jakel and Maragos 2000; Petersen et al. 2002). Therefore, it is tempting to speculate that the aberrant γ synchrony is associated with transient and abnormal DA transmission. Whether the γ oscillation is unique to our mice or this rhythm exists in HD patients remains yet unexplored. A few single-unit recording studies in humans limited to the pallidum, demonstrated increased or unchanged pallidal activity (Starr et al. 2008; Tang et al. 2005).

Basal Ganglia Network Activity Across the Disease Progression in HD Mice

We previously formulated the hypothesis that dysfunctional striatal MSNs, which did not meet threshold for action potentials during recording, might be part of the indirect pathway vulnerable in HD. As a first step for testing this long standing but untested hypothesis, we employed immunocytochemistry for labeling the neural activity marker, c-Fos, to determine age-dependent changes in the basal ganglia activity.

Our data were surprising for two reasons. First, contrary to our expectation, in R6/1 mice, there was no diminution of striatal activity revealed by Fos-immunoreactivity (IR) subsequent to behavioral stimulation at any ages studied (Bassil et al. 2012). This suggests that the cellular machinery necessary for c-Fos transcriptional activity remains relatively intact in the transgenic mice even at symptomatic ages. Second, contrary to the expected diminution, when compared with WT counterpart, the level of Fos-IR in the dorsomedial striatum was unexpectedly increased at an early asymptomatic age well before the appearance of motor symptoms. These early cellular and possibly compensatory changes by the intact direct pathway were correlated with significant degradation of cognitive and social behaviors (Pietropaolo et al. 2011; Lebreton et al. 2015) in these young R6/1 mice. The increase of Fos-IR disappeared at older (4 and 6 months) ages in R6/1 mice, annulling differences between the genotypes. The pattern of Fos activation of infra- and prelimbic prefrontal cortex that project directly to the dorsomedial striatum (Voorn et al. 2004) studied here, followed closely the activation pattern of the neurons in the striatum. This confirms that changes in the integral cortico-striatal pathway are an early event in HD (Andre et al. 2011a). The exact mechanisms responsible for these unique and paradoxical changes at early ages remain to be investigated.

To dissociate the MSN activity of the indirect from that of the direct pathway we used a transgenic mouse line expressing enhanced green fluorescent protein under the control of a D1 receptor promoter. D1 receptors are expressed mainly in MSNs of the direct pathway, while D2 receptors are mainly expressed in MSNs of the indirect pathway (Gerfen et al. 1990; Le Moine and Bloch 1995). MSNs constitute about 95 % of striatal cell population as mentioned above, and MSNs of the two pathways are equivalent in numbers (Matamales et al. 2009), even though a negligible proportion of them coexpresses both D1 and D2 receptors. Using colabeling of Fos and DARP32 (labeling all MSNs) in crossbred mice (R6/1 x D1-GFP), we quantified the proportions (and difference) of D1 labeled MSNs among Fos-IR neurons. Preliminary results of this ongoing experiment confirmed an increased Fos-expression in the striatum at presymptomatic age in HD transgenics. In addition, the percent of MSN expressing D1 receptors among Fos-IR cells was significantly higher in the R6/1 mice as compared to WT mice, suggesting a hyperactivity of the direct D1 pathway in these asymptomatic transgenic mice.

β Synchrony and Its Paradoxical Relationship with Sleep–Wake Vigilance States

When analyzing striatal activity during the previous procedural/habit learning, we have noticed subtle changes in striatal LFPs in the frequencies of 25–35 Hz, which we referred to as β oscillations. Albeit different in the frequency, this oscillatory activity present in both normal and transgenic mice was associated with specific behaviors, for example, at the mouse’s arrival at the food port. Previous recording studies performed in healthy rats also reported such oscillatory activity in the cortico-striatal circuit (Howes et al. 2011; Leventhal et al. 2000). Because the β bursts arising at discrete moments during habit learning increased significantly when animals acquired the task, this transient synchrony has been proposed to be a marker of the habit memory formation (Howes et al. 2011).

The procedural/habit memory formation requires over night sleep, and slow wave sleep (SWS) or rapid eye movement (REM) sleep deprivation perturbs the retention of the learned procedural memory and retards the task acquisition (Marshall and Born 2007; Walker and Stickgold 2004). Since HD patients and HD transgenic mice display sleep and circadian cycle disturbances (Wiegand et al. 1991; Petersen et al. 2005; Arnulf et al. 2008; Goodman et al. 2011; Morton et al. 2005), the examination of brain rhythmic activities during sleep might be informative for explaining cognitive and behavioral changes seen in HD.

We, therefore, performed electrophysiological recording of LFPs of different regions of the basal ganglia circuitry throughout disease onset and progression according to a longitudinal plan. The analysis of different LFPs during sleep revealed an abnormal β synchrony (20–40 Hz) in all R6/1 mice studied, but not in any of WT littermates across ages. This β oscillation thus constituted a hallmark brainwave of HD in R6/1 mice. Ages at which the β rhythm was first detected preceded neurological hind limb clasping by 1.5 months in most of the transgenics, announcing the later appearance of motor impairments (Jeantet et al. 2013).

In addition, an intriguing finding was that the β synchrony owned a unique and unexpected relationship with sleep–wake vigilance states. More precisely, the β rhythm was mainly associated with sleep state because it was present barely during waking state. However, when entering REM sleep, which is usually associated with intense desynchronized cortical activity similar to the waking state, the β oscillation continued to intensify its amplitude instead of disintegrating as major cortical synchrony (Fig. 14.5a-b). Therefore, characteristics of the β oscillation in our HD mice enabled a clear dissociation between REM sleep and waking states.

Fig. 14.5
figure 5

β (20–40 Hz) oscillation in R6/1 mice a and b Power variation in the β frequency range across four vigilance (active wake, quiet wake, slow wave sleep (SWS), and rapid eye movement sleep (REM)) states in a representative symptomatic R6/1 mouse (a) and an age-matched littermate (b). Blue, green, purple, and brown lines in a and b represent active wake, quiet wake, SWS, and REM sleep, respectively. c and d REM sleep power spectral variations across disease progression (2–6 months (M), symptomatic to symptomatic, respectively) in the same R6/1 (c) and WT littermate (d) shown in a and b

Full size image

Finally, this pattern of β oscillation variation across vigilance levels became more pronounced with the disease progression. The β peak shifted with age toward lower frequency but higher amplitude within a given brain state (Fig. 14.5c-d). The phenomenon thus became more exaggerated with the disease progression. However, the frequency and amplitude of 7–10 Hz θ rhythm also strongly present in REM sleep spectra in both R6/1 and WT mice, remained constant over ages even in R6/1 mice. However, its frequency remained 0.5–0.75 Hz lower in R6/1 mice than in WT mice. These data identify the β, but not θ synchrony, as a dynamic in vivo neurophysiological marker of HD, which accompanies sleep abnormalities (i.e., fragmentation and diurnal activity changes) and cognitive impairments in these mice (Lebreton et al. 2015). The fact that the β oscillation is present during SWS and REM sleep, periods critical for offline information processing and consolidation, and the aberrant oscillations may disturb these critical neural processes and information transfer among neural regions involved in learning, providing network “correlates” of the observed learning deficits mentioned earlier (Cayzac et al. 2011). This also highlights the interrelationship among the cortico-striatal and basal ganglia network activity, sleep and cognitive functions in general and especially in HD.

During typical sleep–wake cycles, two types of phenomena are known to announce REM sleep onset following SWS: (1) the return of certain brain (cortex and hippocampus) activities to awake-like state and (2) a further intensification of certain SWS characteristics such as increased threshold for awakening by sensory stimuli (during eye movement) (Ermis 2010) and loss of muscular tone (atonia). Since, in R6/1 mice, the β synchrony that appears during SWS and grows further during REM sleep, the cortico-striato–thalamo–cortical network at the origin of this synchrony may be subject to the second type of phenomenon, i.e., the further intensification of SWS characteristics. Neither the exact constituents of the β generating circuit (i.e., striatum, subthalamic nuclei, thalamus, cortex, etc.), nor the way in which this circuitry is modulated during REM sleep has been identified. However, our data suggest that the machinery for β, but not for θ synchrony, that is distinctively active during sleep, is dysfunctional in HD mice. This specificity may point to defective brain activation systems (e.g., DA, Acetylcholine, Serotonin, Orexin, etc.) or other mechanisms within the restricted circuit affected in HD, which when dissected, may offer new targets for therapeutic interventions.

Conclusion

Basal ganglia form a complex neural network involved in the selection and execution of response/action through interactions with multiple brain areas that process sensorimotor, emotional, and cognitive information. Here, we summarized how the functional network activity in the basal ganglion is compromised in a mouse model of HD, which expresses the HD mutation. Growing evidence suggests that synchronization of neuronal activity within and across different brain regions is a fundamental property of cortical and subcortical networks and serves a variety of functions in cognitive processes (Fries 2005; Singer 1999). Because remarkable and unique synchronies in the β and γ frequencies were associated with cognitive and behavioral perturbations in HD mice, it is tempting to speculate that these aberrant neural synchronies play critical roles in cognitive and motor dysfunctions associated with HD. Further work should aim at understanding of how these peculiar synchronies involving large populations of neurons perturb the proper information coding and flow in the basal ganglia loop and related limbic structures. These studies may ultimately bring a new insight into not only the pathophysiology of HD, but also the nature of the cognitive operations performed in/by the intact striatum, the main component of habit and procedural learning system.