The axonal ability to sustain impulse conduction highlights the mystery regarding the return pathway of sodium ions after entering the axoplasm, as the axolemmal sodium pump must extrude these ions into the periaxonal space, rather than returning them to the nodal gap. We have explored the pathway taken by these ions using in vivo confocal microscopy to observe axons during sustained impulse activity. Mice transgenically expressing yellow fluorescent protein in some axons were stimulated electrically at physiological frequencies or pharmacologically while observing their axonal structure by confocal imaging. A series of morphological changes ensued, starting with an expansion of the periaxonal space, separating the axolemma from the Schwann cell and compressing the axoplasm. The increase in axoplasmic pressure caused an inflation of the axonal morphology at the paranodes and a herniation of the enclosed axoplasm on either side of the nodal membrane, directed back over the outside of the axon, displacing the paranodes and widening the nodal gap. Concurrently, the fluid in the expanded periaxonal space accumulated into droplets that travelled to the paranode where they escaped by apparently parting the axolemmal attachment of the paranodal loops of myelin. These alterations occurred in virtually all axons, and none occurred in axons treated with sodium channel or sodium pump inhibitors. All these changes reversed spontaneously, and impulse conduction continued throughout.

The Ketogenic diet (KD) is an alternative treatment option for people with drug-resistant epilepsy. It can reduce seizure frequency, but it is resource-intensive and may cause adverse side effects. Predictors of response – which, in the absence of specific metabolic disorders, are unknown – would improve patient selection and may enhance understanding of how the KD exerts its antiseizure effect. This project is concerned with identifying possible genetic markers of response to the KD. DNA was extracted from capillary blood taken from individuals who were following the KD for their epilepsy, or who had done so in the past. Individuals were classed as responders if they achieved ≥50 seizure reduction. Response was classified at various follow-up points, as well as a summary of response over time. Association studies were conducted using candidate gene (KCNJ11 and BAD) sequencing, genome-wide single nucleotide polymorphism array, and whole exome sequencing data to determine whether there was an over-representation of specific gene variants in KD responders, compared to non-responders. Common variation in KCNJ11 and BAD was not significantly associated with KD response. rs12204701 reached significance in the array-based genome-wide association study including common variants, with 3-month diet response as the phenotype. No significant results were obtained when summary diet response was taken as the phenotype. Using the gene-based c-alpha test with the exome sequencing data, including all exonic and splicing variants, ANKRD36C reached significance; using a pathway-based count of case-unique alleles test, the ‘ERBB1 Internalisation’ pathway reached significance. No further significant results were obtained from the exome sequencing data when using other gene- and pathway-based tests or when variants were further filtered according to predicted functional consequence. Other genes with large differences in responder/non-responder minor or alternative allele counts are also of interest. It is unknown how these may contribute to variability in KD response. Some common themes were identified amongst the genes and pathways of significance and suggestive significance: cell cycling, apoptosis, glucose homeostasis, neurological processes and triglyceride biosynthesis. It is biologically plausible that these processes influence KD response, although it is likely that many genes play a role. A larger sample size is needed in order to improve power to detect genotypic-phenotypic associations and increase confidence in the importance of the genes of interest.

Over the past 5 years, whole exome sequencing (WES) ignited a revolution in genomic science, triggering a wave of genetic discoveries that have dramatically improved our understanding of the human genetic landscape. In my research, I have used this innovative approach to explore different facets of the complex genetic architecture of Alzheimer’s disease (AD). First, given the phenotypic overlap between different neurodegenerative dementias such as frontotemporal dementia (FTD), progressive supranuclear palsy (PSP), corticobasal degeneration (CBD) and Alzheimer’s disease, I tested the hypothesis that Mendelian dementia genes (APP, PSEN1, PSEN2, GRN, MAPT and PRNP) may influence sporadic Alzheimer’s disease. Second, genome-wide association studies (GWASs) have identified 9 main susceptibility loci for AD (CLU, PICALM, BIN1, EPHA1, ABCA7, MS4A6A, CD33, CR1, CD2AP). Although these have been consistently replicated, the risk variants underlying these GWASs hits still need to be identified, thus I have investigated the whole spectrum of protein-coding variability within these candidate loci. Third, the ‘Amyloid cascade hypothesis’ is the ‘Rosetta stone’ of AD pathology, therefore I have analyzed genes involved in APP-Aß metabolism. Fourth, the discovery of TREM2, the second most significant risk factor for AD. This tremendous finding has been the result of an extensive collaboration and the exome sequencing data that I have generated were the initial foundation of this work. Finally, an increasing body of evidence has shown that causative genes for complex Mendelian syndromes may harbor risk factors for common polygenic diseases. Thus, I have analyzed rare coding variants in Mendelian leukodystrophy genes, as potential susceptibility factors for Alzheimer’s disease. In conclusion, my research emphasizes the pivotal role of rare protein-coding variability in AD genetics. Whole exome sequencing has enabled researchers to explore the human genetic architecture in a comprehensive and unprecedented way, drawing maps of rare coding variability across the human genetic landscape. My doctoral thesis is a reflection of this exciting advancement.

Depleted of dopamine, the dynamics of the Parkinsonian brain impact on both “action” and “resting” motor behaviour. Subthalamic nucleus deep brain stimulation (STN DBS) has become an established means of managing these symptoms, although its mechanisms of action remain unclear. Functional magnetic resonance imaging (fMRI) using the blood oxygen level dependent (BOLD) contrast provides the opportunity to study the human brain in vivo, collecting indirect measures of neural activity across the whole brain. To date, technical difficulties and safety concerns have precluded the use of fMRI in DBS patients. Previous work from this department has demonstrated that scanning patients with certain DBS systems and MRI equipment is both safe and feasible. This thesis explores the neuromodulatory actions of STN DBS on both action and resting motor behaviours in patients with Parkinson’s disease (PD) using fMRI. In brief, I present two fMRI studies conducted on STN DBS patients, one task-based, and one resting, collected under a previously approved protocol. I then present experiments exploring the safety of scanning DBS patients using an improved protocol, and then detail two further experiments collected under this new protocol, again one task-based, and one resting. Specifically, I employ statistical parametric mapping to determine DBS-induced changes in motor evoked responses. Using dynamic causal modelling (DCM) and Bayesian model selection, I compare generative models of cortico-subcortical interactions to explain the observed data, inferring which connections DBS may be affecting, and which modulations predict efficacy. I proceed to use stochastic DCM to model the modulatory effects of DBS on endogenous (resting-state) dynamics. Abstract | 4 4 This work casts DBS in terms of modulating effective connectivity within the cortico-basal ganglia motor loops. I discuss how this may explain its current usage in PD, as well as exploratory uses to treat other pathological brain states.

Goal-relevant information maintained in working memory is remarkably robust and resistant to distractions. However, our nervous system is endowed with exceptional flexibility; therefore such information can be updated almost effortlessly. A scenario – not uncommon in our daily life – is that selective maintaining and updating information can be achieved concurrently. This is an intriguing example of how our brain balances stability and flexibility, when organising its knowledge. A possibility – one may draw upon to understand this capacity – is that working memory is represented as beliefs, or its probability densities, which are updated in a context-sensitive manner. This means one could treat working memory in the same way as perception – i.e., memories are based on inferring the cause of sensations, except that the time scale ranges from an instant to prolonged anticipation. In this setting, working memory is susceptible to prior information encoded in the brain’s model of its world. This thesis aimed to establish an interpretation of working memory processing that rests on the (generalised) predictive coding framework, or hierarchical inference in the brain. Specifically, the main question it asked was how anticipation modulates working memory updating (or maintenance). A novel working memory updating task was designed in this regard. Blood-oxygen-level dependent (BOLD) imaging, machine learning, and dynamic causal modelling (DCM) were applied to identify the neural correlates of anticipation and the violation of anticipation, as well as the causal structure generating these neural correlates. Anticipation induced neural activity in the dopaminergic midbrain and the striatum. Whereas, the fronto-parietal and cingulo-operculum network were implicated when an anticipated update was omitted, and the midbrain, occipital cortices, and cerebellum when an update was unexpected. DCM revealed that anticipation is a modulation of backward connections, whilst the associated surprise is mediated by forward and local recurrent modulations. Two mutually antagonistic pathways were differentially modulated under anticipatory flexibility and stability, respectively. The overall results indicate that working memory may as well follow the cortical message-passing scheme that enables hierarchical inference.

The retrosplenial cortex (RSC) comprises Brodmann areas 29/30 and is an integral part of a brain system that is engaged by spatial navigation, scene processing, recollection of the past and imagining the future. Damage involving the RSC in humans can result in significant memory and navigation deficits, while the earliest metabolic decline in Alzheimer's disease is centred upon this region. The precise function of the RSC, however, remains elusive. In this thesis I sought to determine the key contribution of the RSC in a series of six studies that each comprised behavioural and functional magnetic resonance imaging (fMRI) experiments. Specifically, I discovered that the RSC is acutely responsive to landmarks in the environment that maintain a fixed, permanent location in space, and moreover is sensitive to the exact number of permanent landmarks in view. Using a virtual reality environment populated with entirely novel ‘alien’ landmarks I then tracked the de novo acquisition of landmark knowledge and observed the selective engagement of the RSC as information about landmark permanence accrued. In three further studies I established the parameters within which the RSC operates by contrasting permanent landmarks in large- and small-scale space, by comparing landmark permanence with orientation value, and by investigating permanence in non-spatial domains. In parallel lines of inquiry, I uncovered evidence that a fully functional RSC may be a prerequisite for successful navigation, while also characterising RSC interactions with other brain regions, such as the hippocampus, that could have importance for constructing reliable representations of the world. Together my findings provide new insights into the role of the RSC in a range of cognitive functions. The RSC’s processing of permanent predictable features may represent a key building block for spatial and scene representations that are central to navigation, recalling past experiences and imagining the future.

Fast movements have traditionally been avoided during rehabilitation after stroke for fear that they might increase spasticity. However movements in daily life require alterations in speed. In healthy individuals (n=14), 1 week of intensive reaching training in a robotic manipulandum biased participants movement speed during other tasks performed in the manipulandum: people who trained to move slowly tended to move more slowly whereas the opposite was true for the fast training group. Yet the improvement in movement accuracy was the same in both groups. In chronic stroke survivors (n=37) functional ability before training was determined by muscle stiffness and weakness. Endpoint accuracy improved after training at both movement speeds in all participants apart from those with the greatest sensory impairment. However, training at the different speeds modified different kinematic parameters of the task (e.g. movement trajectory versus velocity profile) so that training at one speed did not generalise well to movements performed at non-trained speed. Most interestingly muscle stiffness was not increased by training at high velocity. Unexpectedly, it was reduced and functional ability as well as the ability to reach outside of the robotic manipulandum improved. Transcranial magnetic stimulation on a sample of these stroke participants (n=19) was used to test the hypothesis that ipsilateral drive to proximal muscles from the non-stroke hemisphere is important in determining recovery of arm and shoulder movement. MEPs could be elicited in at least 25% of muscles from both hemispheres but contrary to expectation, arm function was only correlated to the drive from the affected hemisphere, even in patients with the greatest impairment. Muscle activation patterns were not significantly different between the affected and unaffected limb and training did not alter these. Training at a fast speed is not detrimental in patients after stroke and should be encouraged to increase the variety of movements patients can perform.

A long-standing dichotomy in neuroscience pits automatic or reflexive drivers of behaviour against deliberate or reflective processes. In this thesis I explore how this concept applies to two stages of action control: decision-making and response inhibition. The first part of this thesis examines the decision-making process itself during which actions need to be selected that maximise rewards. Decisions arise through influences from model-free stimulus-response associations as well as model-based, goal-directed thought. Using a task that quantifies their respective contributions, I describe three studies that manipulate the balance of control between these two systems. I find that a pharmacological manipulation with levodopa increases model-based control without affecting model-free function; disruption of dorsolateral prefrontal cortex via magnetic stimulation disrupts model-based control; and direct current stimulation to the same prefrontal region has no effect on decision-making. I then examine how the intricate anatomy of frontostriatal circuits subserves reinforcement learning using functional, structural and diffusion magnetic resonance imaging (MRI). A second stage of action control discussed in this thesis is post-decision monitoring and adjustment of action. Specifically, I develop a response inhibition task that dissociates reactive, bottom-up inhibitory control from proactive, top-down forms of inhibition. Using functional MRI I show that, unlike the strong neural segregation in decision-making systems, neural mechanisms of reactive and proactive response inhibition overlap to a great extent in their frontostriatal circuitry. This leads to the hypothesis that neural decline, for 4 example in the context of ageing, might affect reactive and proactive control similarly. I test this in a large population study administered through a smartphone app. This shows that, against my prediction, reactive control reliably declines with age but proactive control shows no such decline. Furthermore, in line with data on gender differences in age-related neural degradation, reactive control in men declines faster with age than that of women.

Prion diseases are fatal, transmissible neurodegenerative diseases which are associated with the conversion of host encoded prion protein, PrPC, into an aggregated, proteinase-resistant isoform, termed PrPSc. Coding polymorphisms within Prnp, the gene encoding PrPC, are known to affect disease incubation times and susceptibility in human, mouse, and sheep and the most prominent example, codon 129 polymorphism in humans, has major disease modifying effects. However, significant differences in incubation times for scrapie in mice with the same Prnp genotype indicate a major role of PrP-independent genetic factors. To identify these factors, prion-resistant revertants were isolated from a highly prion-susceptible cell clone and the respective transcriptomes were analysed. Remarkably, incubation of revertants with the differentiation agent retinoic acid led to a forty-fold increase in prion propagation rates and downregulation of previously identified differentially expressed genes. This approach led to identification of a gene signature of eighteen genes associated with susceptibility to prion propagation and regulated by differentiation. The experimental validation of the relationship between gene expression and prion propagation by transcriptional silencing confirmed the role of nine genes, expressed in revertants. Several susceptibility genes encode proteins that are involved in the regulation of the extracellular matrix (ECM), a site where disease associated PrP (PrPd) is found. Further experimentation revealed novel insights on how expression of these encoded proteins modulate prion replication mechanistically. Inhibition of fibronectin1 binding to integrin α8 by the RGD peptide significantly decreased the activation of matrix metalloproteinase (MMP)-2/9 whilst prion propagation rates increased. Moreover, Papss2 loss-of-function led to undersulphation of heparan sulphate with a concomitant increase in prion propagation. Remarkably, under these conditions, PrPC was deposited at the ECM. The observation that Neural cell adhesion molecule (Ncam) colocalised with PrPd at the basement membrane of infected, but not of uninfected cells led us to scrutinise membrane microdomains in uninfected and chronically infected cells. Co labelling experiments provided evidence for colocalisation of PrPd with tetraspanin-containing microdomains and GM1 domains. The significance of this finding has yet to be determined. In summary, we have identified a gene regulatory network associated with prion propagation at the ECM which is governed by the cellular differentiation state. Work is currently in progress to determine the functional relevance of the identified genes in vivo.

This thesis describes the electrophysiological and anatomical distribution of the projections from the rubrospinal tract (RST) in the cervical spinal cord of the rat, in control animals and animals with a dorsal column lesion. It also describes the behavioural changes following the lesion. The RST originates from the magnocellular region of the red nucleus (RNm), from where its axons project to the contralateral spinal cord, mostly in the dorsolateral funiculus (DLF). Previous work has shown that the RST is implicated in precise limb movements. It has also been demonstrated that where the corticospinal tract (CST) is impaired, the RST is capable of motor control and will compensate for the damage to the CST. In terminal experiments, electrical stimulation of the RNm elicited two orthodromic descending volleys recorded from the contralateral DLF. The latency of the early one indicated that it is produced by direct activation of fibres in or close to the RNm. Temporal facilitation demonstrated that the second volley is elicited by synaptic excitation of RNm neurons. Postsynaptic responses related to the second volley were seen only in the intermediate zone of the contralateral spinal cord. In animals with a dorsal column lesion, the distribution of RST actions is similar to that of the controls. Functional recovery in lesioned animals was assessed using three behavioural tests: pellet retrieval, cylinder, and the sticker removal test. The cylinder and sticker removal tests have both shown to be ineffective in providing important information on the recovery. However, the pellet retrieval test, which requires skilled motor control, has enabled correlation of recovery to size of lesion. The results obtained demonstrate that lesion to dorsal columns alone has little effect on the success rate. The anatomical and electrophysiological experiments provide information on the distribution of RST actions, which will be correlated in the thesis with the behavioural results. Evidence for plasticity will be discussed.