Examining the genetic and neural components of cognitive flexibility using mice

Abstract

This commentary summarizes the research presented during the symposium “Examining the genetic and neural components of cognitive flexibility using mice” at the annual meeting of the International Behavioral Neuroscience Society 2011. Research presented includes examining: 1) Corticostriatal networks underlying reversal learning using GluN2B knockout mice, cFos expression, and in vivo electrophysiological recording; 2) Cerebellar contribution to reversal learning using mutants with Purkinje cell loss and in vivo electrochemical recording; 3) Parvalbumin contribution to reversal learning and set-shifting using PLAUR mutants and in vitro recording to examine fast-spiking interneurones; and 4) Alpha 7 nAChR contribution to reversal learning, set-shifting, motivation, and the ‘eureka moment’ of rule acquisition. It is proposed that these studies revealed more about the neurobiology underlying these behaviors than could be discovered using pharmacological techniques alone. Together, the research presented stressed the importance of exploring the genetic contribution to neuropsychiatric disease and the important role that the mouse, coupled with robust behavioral measures, can play in understanding neurobiology underlying cognitive flexibility.

1. Introduction

Impairment in cognitive functioning has received increasing research focus in Psychiatry and Neuroscience over the past 10 years. This focus on cognitive research has stemmed in part from the recognition that improving functional outcome in patients with psychiatric disorders is likely to require improving their cognitive capabilities [1–3]. Behavioral flexibility is one cognitive domain that is adversely affected in numerous disorders including schizophrenia, bipolar disorder, Huntington’s disease, Alzheimer’s disease, attention deficit hyperactivity disorder (ADHD), and autism spectrum disorder. This domain refers to a subjects’ ability to alter behavior in response to changing environmental contingencies [4]. Examples of cognitive flexibility include reversal learning, set-shifting, and dynamic adjustment of control [5].

In order to improve cognitive flexibility in patients, an early step is to assess the efficacy of putative treatments in animals [6]. Moreover, because some disorders require the development of entirely novel treatments (i.e. no approved positive controls with which to compare against novel compounds), we must also be able to examine the genetic and neurobiological underpinnings of cognitive flexibility [7]. With the advent of sequencing of the human genome and the ability to precisely manipulate the mouse genome, the mouse has rapidly become the model species of choice for researchers investigating the genetic underpinnings of cognition. Given the strong genetic contributions to multiple neuropsychiatric diseases, assessing the behavior of genetic variants has become imperative. The development of paradigms to assess cognitive functioning in mice has only begun to catch up with rat cognitive paradigms [4, 8]. Publications investigating pharmacological-induced alterations in cognitive behavior using mice are dwarfed by the number of corresponding studies performed in rats [9]. This in part is due to mouse cognitive testing being a relatively young field largely based on rat protocols that have been developed over the past 50–60 years while cognitive testing in mice has a more recent history. In fact it is rare for a cognitive paradigm to be developed first in mice and not in rats, but such events are occurring [8].

This disparity between the numbers of pharmacological studies performed in rats compared to mice should be addressed given the interactions that are known to occur between genetic and pharmacological manipulations. The concern that mice cannot readily perform such cognitive tasks may have contributed to this disparity. We gathered experts in the field of assessing cognitive flexibility testing in mice to demonstrate that mice do perform complex tasks using neural circuitry analogous to the rat and primate. The power of assessing multiple genetic variants was highlighted within the contexts of several anatomical substrates, including examining corticostriatal networks and flexibility, cerebellar modulation of executive function, and fast spiking interneurons assist cortical tuning. The use of a battery of cognitive tests including flexibility has also been used to characterize the contribution of specific genes to cognitive performance (details of each talk are provided below).

2.1 Dr. Brigman: Corticostriatal networks underlying cognitive flexibility

In order for an organism to behave optimally in its environment it must automatize behaviors that consistently lead to positive outcomes while retaining the ability to shift its behavior in response to changing demands. Across species, corticostriatal networks are thought to underlie this balance of flexible and habitual behavior that is impaired in a wide range of neuropsychiatric disorders [10, 11]. In order to take advantage of the genetic, behavioral and pharmacological tools mouse models provide to investigate these impairments, it is essential to firmly establish the circuits underlying these behaviors in the species. We conducted a series of experiments using a touch-screen-based visual discrimination and reversal task [12] to explore cortical and striatal contributions to flexible and habitual action in the mouse. We first examined the activation patterns of cortical and striatal subregions during task performance via the immediate early gene c-Fos and then confirmed the functional role of these regions via targeted lesion. Patterns of cortical and striatal activation during discrimination and reversal learning in the mouse were consistent with the established role of these areas in flexible and well-learned behavior respectively and selective loss of function in these areas impaired task performance [13, 14].

Given its role in the plasticity underlying learning and memory [15, 16], the N-methyl-D-aspartate receptor (NMDAR) may be a potential molecular mechanism underlying these behaviors. To test this hypothesis we utilized both genetic and pharmacological techniques to deplete or inactivate GluN2B-containing NMDAR in either cortical, striatal or corticostriatal subregions. Loss of GluN2B in cortex was sufficient to impair reversal, while GluN2B in cortex and striatum impaired discrimination. Results from GluN2B antagonism were consistent with these findings.

Finally, in order to directly measure the organization of neuronal firing in the striatum during flexible and habitual learning, in vivo electrophysiological recordings [17] were combined with the touch-screen task. Striatal firing patterns showed robust learning related changes in association with changes in task behavior. Taken together, these findings demonstrate the utility of the mouse for examining cognitive dysfunction that accompanies neuropsychiatric disease as well as establish a role for GluN2B in regulating corticostriatal control of flexibility and habit.

2.2 Dr. Mittleman: Cerebellar modulation of executive function

Neuroimaging studies showing cerebellar activation during cognitive tasks as well as research demonstrating cerebellar pathophysiology coupled with deficits in executive functions (e.g., autism) provide mounting evidence for a cerebellar role in cognition. We investigated the relationship between developmental loss of cerebellar Purkinje cells (the sole outflow of cerebellar cortex) and cognitive dysfunction through a series of studies utilizing aggregation chimeras made from wildtype and lurcher mouse embryos. Lurcher mice lose 100% of their cerebellar Purkinje cells due to a gain-of-function mutation in the δ2 glutamate receptor (GRID2) gene while individual wildtype↔lurcher chimeras have between 0 and 100% of normal Purkinje cell numbers depending on the incorporation of each lineage.

Using operant techniques we assessed the performance of these chimeras on tasks designed to measure spatial working memory, sustained attention, perseverative lever pressing, and serial reversal learning of a conditional visual discrimination. The strategy in all experiments was to determine if Purkinje cell numbers correlated with behavioral measures in order to determine what cognitive abilities were, or were not affected by Purkinje cell loss. We found that Purkinje cell numbers were significantly correlated with serial reversal learning and perseverative lever pressing in that lower numbers were associated with more learning errors and enhanced perseveration [18, 19]. Interestingly, lower numbers of Purkinje cells were also associated with facilitated working memory performance [20]. We found no relationship between sustained attention and Purkinje cell numbers [21].

As behavioural deficits in serial reversal learning, working memory, and perseveration have been found to be related to frontal cortical function (e.g. Dalley, Cardinal, et al. 2004) we used in vivo electrochemistry to explore the hypothesis that disruption of cerebellar-prefrontal cortex (PFC) projections could lead to a dysregulation of dopaminergic-mediated activity in medial PFC. Dopamine (DA) efflux evoked by cerebellar or dentate nucleus electrical stimulation was recorded in the PFC of lurcher mice and wildtype controls. DA efflux following stimulation at either site was markedly attenuated in lurcher mice in comparison to controls [22]. Considered together, these experiments suggested that the cerebellum is involved in multiple cognitive functions, given Purkinje cell number was highly negatively correlated with serial reversal deficits and perseverative behaviour but was unrelated to sustained attention. These results also suggest that a loss of cerebellar modulation of PFC dopamine efflux may account for some of these behavioural deficits. These findings may explain why cerebellar and frontal cortical pathologies co-occur, and may provide a mechanism that accounts for the diversity of symptoms common to multiple neurological and psychiatric disorders.

2.3 Dr. Powell: Cortical tuning assisted by fast spiking interneurons

Research suggests there may be common underlying alterations to common neural circuits for the similar cognitive etiologies which are observed in multiple psychiatric disorders. One of these common anatomical manifestations involves deficits to the GABAergic system in the cerebral cortex, specifically to the fast spiking parvalbumin-expressing interneurons. The forebrain GABAergic interneurons originate embryonically from subcortical sources, in the precursor to the dorsal striatum [23]. The specification of phenotypically different interneuron subpopulations is an active area of research in developmental neurobiology. The migration paths and signalling mechanisms which govern the final location and function is largely unknown. Yet, perturbations in the number and location of the GABAergic interneurons, in particular the parvalbumin-expressing subpopulation, have been reported in human schizophrenia, autism and epilepsy. Therefore, abnormal interneuron ontogeny is hypothesized to lead to neuropsychiatric disorders.

One gene that is associated with interneuron development and human diseases of schizophrenia and autism is urokinase plasminogen activator receptor (PLAUR, also known as uPAR or CD87)[24]. Mice lacking Plaur have specific loss of parvalbumin expressing interneurons in frontal and parietal cortical areas [25]. To better understand the functional changes elicited by the interneuron deficit, we developed a mouse task to assess reversal learning and attentional set-shifting in which we could record in vivo electrophysiological activity. In the Plaur null mutant mouse, behavioral deficits in the ability to optimally modify learned associations and update decision making behaviors become apparent. The Plaur null mouse was impaired in reversal learning and unable to form the attentional set [26]. In vitro research with slice recordings indicated that parvalbumin-expressing interneurons are essential for the generation of high frequency oscillations, coordinating large ensembles of neurons. Local field potentials in the Plaur null mouse demonstrated decreased activity in the beta and gamma frequencies during reversal leaning. Our research works to illuminate a common neural substrate for learning and prefrontal decision related processes between mouse circuitry and behavior and human cortical function in psychiatric disease states.

2.4 Dr. Young: Alpha 7 nicotinic acetylcholine receptor contribution to learning

There are numerous psychiatric disorders for which impaired cognitive functioning is being given increasing relevance. The primary neuropathologies for some diseases are clear, such as dopaminergic neuronal loss in the caudate/putamen in Parkinson’s disease [27] or repeat of the CAG allele in Huntington’s disease [28]. Schizophrenia, on the other hand, has a wide ranging pathology affecting > 20 brain regions and numerous genetic abnormalities contributing to the heterogeneity of the disease [29]. It is perhaps for this reason that no drug has been approved for treating cognitive dysfunction in this disorder despite many attempts and increasing research. The field has turned in part toward understanding the impact of psychosocial (cognitive remediation) techniques which assist patients learning to cognate at a higher level. Not all cognitive remediation studies have been positive however, which may in part be due to poor learning in patients with schizophrenia [30, 31]. Thus, improving learning in patients with schizophrenia may be useful to maximize the effects of cognitive remediation and may lead to improved functional outcome for these patients.

Identifying the mechanisms underlying learning may provide a target for pro-learning treatments to be developed and used in conjunction with cognitive remediation. Nicotine enhances learning across species but its mechanism of action remains unclear. Earlier studies characterizing the cognitive performance of alpha 7 nicotinic acetylcholine receptor (nAChR) knockout (KO) mice identified impaired attention and working memory of these mice [32, 33]. One commonality of these studies were the ad hoc references to learning deficits in these mice compared to their wildtype (WT) littermates [34–36]. Thus, we began to systematically assess the learning capabilities of these mice using an operant-based holepoking task, a foraging task, and the ethologically relevant attentional set-shifting task to assess simple and reversal learning as well as set-shifting behavior [37]. We also examined the motivation of these mice using the progressive ratio breakpoint paradigm [38, 39]. It became apparent that null mutation of the alpha 7 nAChR resulted in impaired initial learning in each paradigm across all modalities despite normal motivation. Assessing the cognitive flexibility of these mice provided the most compelling evidence that the learning deficit of these mice was limited to initial learning of an operant acquired rule. Despite impaired simple learning in the initial stages of the attentional set-shifting task, KO mice exhibited normal reversal learning and set-shifting performance. When subjected to repeated reversals, the WT mice ‘learned to reverse’, while the KO also did so but took longer to acquire this rule. Thus, it became clear that the alpha 7 nAChR is involved in learning the concept of a rule and that removing this receptor delays this ‘eureka’ moment whereby the rule becomes clear and can be followed irrespective of changing contingencies [40]. It is possible therefore, that by administering an agonist at the alpha 7 nAChR might improve this reward-related learning, providing a viable target for pharmacological augmentation to cognitive remediation.

2.5 Dr. Feldon: Discussant

Dr. Feldon summarized each of the talks, describing where similarities between studies and findings were observed, as well as where discrepancies existed. He also emphasized the strength of an integrated approach to investigating cognitive flexibility, combining not only genetic, behavioral and pharmacological approaches, but also intracellebellar recording in cognitive task performing animals. Dr. Feldon also noted that each of the presentations described reversal learning performance of mice during each of these manipulations and stressed that going forward a major goal should be to standardize both procedures for assessment and just as importantly, methods for analyzing and presenting results from reversal learning tasks. In particular, he recommended that in future analyses of data, the speakers could learn as much from an in-depth analysis of error types (regressive, perseverative) made as was observed from total errors.

3. Conclusion

The research described here highlighted the ability of modern researchers to assess cognitive flexibility in mice. Moreover, these studies can be performed across different testing environments – visual, olfactory, lever, and nosepoking – as well as using in vivo EEG recording. Combining these techniques with genetic manipulations available in mouse research offers an opportunity to examine the genetic contribution to cognitive flexibility. The effects of such genetic manipulations on basic cognitive functioning and neuronal activity may reveal more about the neurobiology underlying these behaviors than could be discovered using pharmacological techniques alone [41]. In addition, genetic animal models of disease can be created and the similarity of their behavioral profile to the disease in question can be tested. In an age where it is being recommended that pharmacological interventions be tailored to genetic background, such capabilities will prove invaluable.

Highlights.

• A commentary on the application of genetic and pharmacological manipulation effects on cognitive flexibility in mice.

• Striatal, cortical, and cerebellar in vivo recording of animals performing reversal learning tasks.

• Utilizing a battery of tasks to determine the contribution of receptors to specific behavioral aspects of cognition.

• Utilizing genetic animal models of disease in order to develop treatments to be tailored to specific genotypes.