Gradient echo (GRE) imaging, a magnetic resonance imaging (MRI) technique that is sensitive to changes in the magnetic susceptibility property of tissues, has recently revealed significant signal heterogeneity in white matter (WM) at high magnetic field B0 ≥ 3T. Various aspects of the underlying white matter microstructure have been linked to the observed contrast between white matter regions. This thesis investigates the origins of the observed differences in GRE signal behaviour. We proposed an explicit multi-compartmental model of WM that incorporates realistic representation of the geometry and magnetic susceptibility of the underlying microstructure that can be used to study the effects of WM microstructural changes on GRE signal characteristics. In particular, we looked at the apparent transverse relaxation rate (R2*) and the resonance frequency, as well as their respective deviations from mono-exponential decay and linear phase evolution. Next, we investigated the effect of WM fiber orientation on GRE signal using healthy human volunteers at 3T by correlating the GRE signal from different WM regions with WM fiber orientation information. Using literature-based parameters, we demonstrated that the geometric model predicted similar trends. Lastly, we studied the effect of myelin on GRE signal using a cuprizone mouse model at 7T . An ex vivo study was used to correlate GRE signal in fixed mouse brain with normalized myelin stain intensity. Simulated GRE signal from hypothetical scenarios of demyelination were then compared with the experimental results. R2* and resonance frequency were then used in an in vivo longitudinal study to track myelin changes during demyelination and subsequent remyelination.

Stroke is the 2nd leading cause of death in the UK and worldwide. 150,000 people have a stroke each year in the UK (ischaemic stroke being the most common) and a significant proportion of NHS resources go towards the treatment of these individuals (~£2.8 billion). Twin and family history studies have shown that having affected relatives makes you between 30 and 76% more likely to suffer a stroke, suggesting that there is a genetic component to the disease. So far, no genes have been convincingly associated with stroke. Intermediate traits may be useful tools for identifying genetic factors in complex disease. For stroke, two commonly used intermediate traits are carotid intima-media thickness (CIMT) and white matter hyperintensities (WMHs), which both show high heritabilities. These traits have both been studied widely for associations with many candidate gene polymorphisms. In this thesis I systematically reviewed the literature for all genetic association studies of these two traits. Where particular associations have been studied in large numbers I meta-analysed the available data, developing novel methods for meta-analysis of genetic association data. I found there was substantial heterogeneity and small study bias in the literature and most polymorphisms have still been studied in too small numbers to make accurate conclusions. Apolipoprotein E (APOE) ε is the only polymorphism which shows a consistent association with CIMT, even when only the largest studies are analysed (MD 8μm (95% CI 6 to 11) between E4 and E3, and E3 and E2). No polymorphism has shown a convincing association with WMHs and interestingly APOE appears unlikely to be associated with this trait. This is consistent with previous work that shows that APOE is associated with large artery but not small artery stroke. Taking this hypothesis I attempted to investigate the association of APOE comparing patients who have had a large artery stroke with those who have had a small artery stroke in the Edinburgh Stroke Study cohort. However, genotyping of this polymorphism failed and I present investigatory analyses of problems from the genotyping laboratory.

Bipolar disorder is a heritable psychiatric disorder, and several of the genes associated with bipolar disorder and related psychotic disorders are involved in the development and maintenance of white matter in the brain. Patients with bipolar disorder have an increased incidence of white matter hyper-intensities, and quantitative brain imaging studies collectively indicate subtle decreases in white matter density and integrity in bipolar patients. This suggests that genetic vulnerability to psychosis may manifest itself as reduced white matter integrity, and that white matter integrity is an endophenotype of bipolar disorder. This thesis comprises a series of studies designed to test the role of white matter in genetic risk to bipolar disorder by analysis of diffusion tensor imaging (DTI) data in the Bipolar Family Study. Various established analysis methods for DTI, including whole-brain voxel-based statistics, tract-based spatial statistics (TBSS) and probabilistic neighbourhood tractography, were applied with fractional anisotropy (FA) as the outcome measure. Widespread but subtle white matter integrity reductions were found in unaffected relatives of patients with bipolar disorder, whilst more localised reductions were associated with cyclothymic temperament. Next, the relation of white matter to four of the most prominent psychosis candidate genes, NRG1, ErbB4, DISC1 and ZNF804A, was investigated. A core haplotype in NRG1, and three of the four key single nucleotide polymorphisms (SNPs) within it, showed an association with FA in the anterior thalamic radiations and the uncinate fasciculi. For the three SNPs considered in ErbB4, results were inconclusive, but this was consistent with the background literature. Most notable however, was a clear association of a non-synonymous DISC1 SNP, Ser704Cys, with FA extending over most of the white matter in the TBSS and voxel-based analyses. Finally, FA was not associated with a genome-wide supported risk SNP in ZNF804A, a finding which could not be attributed to a lack of statistical power, and which contradicts a strong, but previously untested hypothesis. Whilst the above results need corroboration from independent studies, other studies are needed to address the cellular and molecular basis of these findings. Overall, this work provides strong support for the role of white matter integrity in genetic vulnerability to bipolar disorder and the wider psychosis spectrum and encourages its future use as an endophenotype.

Vascular, especially cerebrovascular, dysfunction may be a critical factor in ageing and dementia. Cerebrovascular impairment due to risk factors such as ageing, stroke, smoking, diabetes and cerebral hypoperfusion has a deterious impact on the normal supply of basic nutrients such as oxygen and glucose to the brain; their absence leads inevitably to neuronal death. The cerebral white matter lesions found in most forms of dementia are reportedly the result of chronic cerebral hypoperfusion. However the temporal and spatial evolution of damage remains unclear. Furthermore, any decrease in the integrity of the blood-brain barrier (BBB) has been hypothesised to be a precocious attack on white matter. The “milieu interieure” the most protected in the body, namely the extracellular fluid of the brain, is no longer maintained homeostatically. The cumulation of these various pathophysiological processes alters cerebral function and it has been postulated that, in the most extreme instances, the outcome of this cascade of nefarious events leads to dementia. This thesis examines the supposition that chronic cerebral hypoperfusion could be responsible for the time-related development of white and grey matter pathology and investigates the relationships between the disturbances in the integrity of the BBB and white matter pathology. Three studies addressed these aims. In the first, chronic cerebral hypoperfusion, induced in male Wistar rats by bilateral common carotid artery occlusion (BCCAo), was chosen as the model to study changes in axons, myelin, perikarya as well as microglial activation. The groups of rats that underwent BCCAo were examined at three hours as well as three, seven, 14 and 28 days after the induction of chronic cerebral hypoperfusion. The microscopic examination revealed that, after three hours post BCCAo, damage was detected only in axons and myelin. In contrast, no visible pathology to the neuronal perikarya or enhancement of activated microglia (compared to the sham group) was observable. Injury in both white and grey matter and enhancement of activated microglia was observed from three days post BCCAo and increased with time post BCCAo. The most severe damage to the white and grey matter and enhancement of microglial activation was detected at seven days post BCCAo. These results would indicate that white matter damage precedes grey matter pathology and the enhancement of activated microglia. In the second study, the integrity of the BBB at three hours (when only white matter pathology was found according to the results of the first study) and seven days post BCCAo (when more severe damage to the white and grey matter was shown) was assessed by the use of MRI on T1-weighted image acquisitions with gadolinium as a tracer for BBB permeability. White matter integrity was measured by MTR maps from MTI acquisitions in four brain structures (corpus callosum, caudatoputamen, the external and internal capsules). No differences in white matter integrity were detected between the BCCAo and sham group at three hours and seven days. No differences in signal enhancement of gadolinium were detected three hours post BCCAo. However, a significant signal enhancement of gadolinium was detected at seven days post BCCAo in the caudatoputamen and in the external capsule. Furthermore, immunohistochemistry revealed a significant enhancement of activated microglia seven days post BCCAo compared to the sham group. This functional and immunohistochemical finding, when taken together, might indicate that chronic cerebral hypoperfusion is not in itself responsible for BBB permeability. Rather, the damage to the white matter caused by cerebral hypoperfusion may be responsible for the dysfunction of the BBB over time. Another point of interest was the evidence that the enhancement of activated microglia may play a critical role in the increased permeability of the BBB. The final study in this thesis aimed to investigate the possible pathway and proteins potentially implicated in white matter damage and BBB permeability. To address this question, protein levels and the expression of genes involved in the apoptotic and nonapoptotic hypoxic pathways were compared to the sham groups (at three hours and seven days after BCCAo), in three brain structures (cortex, corpus callosum and caudatoputamen). The levels of HIF-1α, MMP-2, Caspase-3 and VEGF were unchanged compared to the sham group after BCCAo. However, VEGF mRNA expression was found to be significantly different to the sham group seven days post BCCAo in all the three structures examined. An overexpression of HIF-1α and a significant level of Caspase-3 would indicate the activation of the apoptotic pathway. However, neither of these criteria were met and these negative results suggest that the apoptotic pathway is not implicated in the mechanisms that lead to white matter pathology after cerebral hypoperfusion. Finally, the significant expression of VEGF mRNA, compared to the sham group seven days post BCCAo, may contribute to the time-relate increased permeability of the BBB. The results presented within this thesis provide a body of evidence to support the hypothesis that chronic cerebral hypoperfusion is - at least – causal to the damage to different components of the white matter which precedes either early ischaemic changes to the perikarya or enhancement of activated microglia following BCCAo. The increased permeability of the BBB, which can be related to the significant over-expression of VEGF mRNA (compared to the sham group seven days post BCCAo), does not appear to be primarily responsible for white matter pathology, because the MRI investigations indicated that BBB integrity was not affected after three hours of BCCAo. The increased permeability of the BBB, observed seven days post BCCAo with MRI, seems to be the consequence of increased brain damage; thereafter, there is a time-dependent relationship between increasing BBB permeability and increasing brain pathology. Overall, the studies reported herein, strengthen the initial working hypothesis. The conclusion – and direction for future studies – would be that minimising white matter pathology and protecting components of the BBB represent potential targets to decrease then incidence of neuropsychological function or to obtund the cerebral dysfunction in patients who suffer from chronic cerebral hypoperfusion.

Cerebral hypoperfusion can occur in normal ageing and is proposed to underlie white matter disturbances observed in the ageing brain. Moreover, cerebral hypoperfusion and white matter attenuation are early events in the progression of Alzheimer’s disease (AD). White matter mostly consists of myelinated axons which have distinct protein architecture, segregated into defined regions; the axon initial segment (AIS), the node of Ranvier, paranode, juxtaparanode, and internode. These sites are essential for action potential initiation and/or propagation and subsequently effective brain function. At the outset of the studies in the thesis there was evidence that the different regions within the myelinated axons are vulnerable to injury and disease. Thus it is hypothesised that in response to normal ageing and/or cerebral hypoperfusion these structures are altered and associated with cognitive impairment and that these effects are exacerbated in a transgenic mouse model (APPSw,Ind, J9 line) which develops age-dependent amyloid-β (Αβ) pathology. The first study aims to investigate the effect of normal ageing and Aβ deposition on myelinated axons and on learning and memory. To address this, the effects of normal ageing on the integrity of the AIS, nodes of Ranvier, myelin, axons, synapses and spatial working memory are examined in young and aged wild-type and TgAPPSw,Ind mice. A significant reduction in the length of nodes of Ranvier is demonstrated in aged wild-type and TgAPPSw,Ind mice. In addition, the length of AIS, is significantly reduced in the aged wild-type animals while the young TgAPPSw,Ind have significantly shorter AIS than the young wild-type mice. These effects are not influenced by the presence of Aβ. Myelin integrity is affected by age but this is more prominent in the wild-type animals whilst axonal integrity is intact. Moreover, there is an age-related decrease of presynaptic boutons only in the TgAPPSw,Ind mice. Contrary to the original hypothesis, working memory performance is not altered with age or influenced by increasing Aβ levels. The second study aims to examine the effects of cerebral hypoperfusion in combination with Αβ pathology and/or ageing on cognitive performance and the structure of myelinated axons. To address this, the effects of surgically induced cerebral hypoperfusion on the integrity of the nodes of Ranvier, paranodes, myelin, axons and spatial working memory performance are investigated in young and aged wild-type and TgAPPSw,Ind mice. A decrease in nodal length is observed in response to hypoperfusion in young and aged animals. This effect is shown to be exacerbated in the young TgAPPSw,Ind animals. Moreover, the disruption of the nodal domain is shown to occur without any gross alterations in myelin and axonal integrity. It is also demonstrated that in response to hypoperfusion, spatial working memory performance is defected in young and aged animals of both genotypes. This deficit is exacerbated in the young TgAPPSw,Ind. The observed changes in the nodal structure are associated with poor working memory performance indicating functional implication for the nodal changes. These data highlight that structures within myelinated axons are vulnerable to ageing and cerebral hypoperfusion. Therefore, the development of strategies that minimize injury or drive repair to these regions is necessary together with therapeutic approaches against the vascular insults that induce hypoperfusion and lead to white matter attenuation and cognitive decline. In the future, it would be interesting to investigate how alterations at the AIS/nodes of Ranvier affect neuronal excitability.

Lacunar stroke accounts for one quarter of all ischaemic stroke and in the long term carries a greater risk of death and disability than was previously realised. Much of our current knowledge originated from neuropathological studies in the 1950s and 1960s. In the last thirty years, brain computed tomography (CT) and magnetic resonance imaging (MRI) have revolutionised our understanding of lacunar stroke and associated features of cerebral small vessel disease (SVD), namely white matter lesions (WML), enlarged perivascular spaces (EPVS) and brain microbleeds (BMB). The purpose of the projects which led to the writing of this thesis was to improve understanding of imaging characteristics of cerebral SVD. We aimed to assess (i) clinical and imaging features which might explain misclassification of lacunar infarcts as cortical infarcts and vice versa, (ii) the proportion of symptomatic lacunar infarcts progressing to lacunar cavities and associations of cavitation, (iii) completeness of reporting of lacunar lesions in the lacunar stroke literature, (iv) definitions and detection of lacunar lesions amongst SVD researchers, (v) the relationship between WML and carotid stenosis, (vi) clinical and imaging associations of EPVS and, (vii) observer variability in the assessment of EPVS and BMB, in order to develop visual rating scales. Section one describes neuroimaging of lacunar stroke. To investigate features which might explain clinical stroke subtype misclassification (‘clinical-imaging dissociation’), I used data from a stroke study. The main factor associated with clinical-imaging dissociation was diabetes, and in patients with acute lacunar infarction, proximity of the lacunar infarct to the cortex, age, diabetes and left hemisphere location. To investigate the proportion of symptomatic lacunar infarcts progressing to cavities, I used data from two stroke studies. A fifth of patients with acute lacunar ischaemic stroke showed definite cavitation on follow-up imaging at a median of 227 days; cavitation was associated with increasing time to follow-up. To assess completeness of reporting of lacunar lesions in the lacunar stroke literature, I reviewed 50 articles from three journals with a stroke focus. There was marked variation in terminology and descriptions of imaging definitions of lacunar lesions. To assess lacunar lesion definitions and detection amongst SVD researchers, I used an online survey consisting of case-based and non-case-based questions. There was marked variation in definitions and descriptions. Cavitated lesions were detected with the highest degree of confidence. Section two describes neuroimaging of associated features of cerebral SVD. Using data from two stroke studies, I examined the relationship between WML and ipsilateral carotid artery stenosis. There was no association between carotid stenosis and WML. I tested the association of EPVS with WML and lacunar stroke subtype using data from a stroke study. Total EPVS were associated with age and deep WML; basal ganglia (BG) EPVS were associated with age, centrum semiovale (CS) EPVS, cerebral atrophy and lacunar stroke subtype. Quantification of observer variability in EPVS rating was assessed on 60 MRI scans selected from a stroke study and an ageing cohort. Intrarater agreement was good and interrater agreement was moderate. Main reasons for interrater disagreement included the visualisation of very small EPVS and the presence of concomitant WML and lacunar lesions. Observer variability in BMB rating was quantified using MRI scans from a stroke study. Interrater agreement was moderate but improved following modification of the pilot rating scale (BOMBS; Brain Observer MicroBleed Scale), which had its main effect by differentiating ‘certain’ BMB from ‘uncertain’ BMB and BMB ‘mimics’. In conclusion, neuroimaging, particularly MRI, is a valuable tool for the investigation of lacunar stroke and associated features of cerebral SVD. With recent technological advances in both CT and MRI, neuroimaging will remain central to future SVD studies, hopefully leading to a much improved understanding of this important disease.

Recent clinical studies suggest that white matter pathology rather than grey matter abnormality is the major neurobiological substrate of age- related cognitive decline during “healthy” aging. According to this hypothesis, cerebrovascular (e.g. chronic cerebral hypoperfusion) and molecular (e.g. APOE, epigenetics) factors might contribute to age-related white matter pathology and cognitive decline. To test this, I used a new mouse model of chronic cerebral hypoperfusion and examined the following predictions: 1) hypoperfusion- induced white matter pathology might be associated with cognitive deficits, 2) APOE deficiency might be associated with white matter anomalies under normal physiological conditions and more severe hypoperfusion- induced white matter pathology, 3) chronic cerebral hypoperfusion might impact on hydroxymethylation (a newly discovered epigenetic marker) in white matter, via perturbations in associated epigenetic pathways, namely methylation and/ or TETs. I. Effects of chronic cerebral hypoperfusion on white matter integrity and cognitive abilities in mice To test the hypothesis suggesting that hypoperfusion- induced white matter pathology is associated with working memory and executive function impairment in mice, behavioural performance and neuropathology were systematically examined in two separate cohorts of sham and hypoperfused C57Bl6J mice. Spatial working memory, memory flexibility, learning capacity, short and long term memory recall were taxed using radial arm maze and water maze paradigms one month after surgery. At the completion of the behavioural testing white and grey matter integrity, inflammation were evaluated using standard immunohistochemistry with antibodies recognizing neuronal axons (APP), myelin sheath (MAG) and microglia (Iba1) as well as H&E histological staining to examine neuronal morphology and ischemic injury. In agreement with previous reports, the behavioral data indicated spatial working memory impairment in the absence of spatial memory flexibility, learning, short- and long- term memory recall deficits in hypoperfused mice However, in contrast to previous reports, a spectrum of white and grey matter abnormalities accompanied by an increased inflammation were observed in hypoperfused mice Although there was a significant association between hypoperfusion- induced inflammation in white matter and performance on a working memory radial arm maze task (p<0.05), the present pathological findings suggest that white matter abnormalities, neuronal ischemia and increased inflammation might be at the basis of hypoperfusioninduced cognitive impairment in mice. Further, chronic cerebral hypoperfusion might have affected alternative, non- examined brain processes (e.g. cerebral metabolism, neurotransmission) which might have contributed to the observed cognitive deficits in hypoperfused mice. II. Effects of APOE on white matter integrity under normal physiological and chronically hypoperfused conditions in mice To test the hypothesis suggesting that mouse APOE deficiency might be associated with white matter anomalies under normal physiological conditions and the development of more severe white matter pathology following chronic cerebral hypoperfusion, white and grey matter integrity, inflammation were examined in APOE deficient mice on a C57Bl6J background (APOEKO) and C57Bl6J wild- type (WT) counterparts one month after chronic cerebral hypoperfusion or sham surgery. A combined neuroimaging (MRI- DTI)/ immunochemical approach was attempted in these mice as an additional step towards translation of this research to human subjects. The ex vivo MRI- DTI findings demonstrated APOE genotype effects on the development of white matter abnormalities following chronic cerebral hypoperfusion in mice. Significant reductions in MRI metrics (FA and MTR) of white matter integrity were observed in examined white matter areas of APOEKO hypoperfused mice compared with WT hypoperfused counterparts (p<0.05). However, the neuroimaigng findings were not supported by the pathological analysis where no significant APOE differences were observed in hypoperfusion- induced axonal (APP), myelin (MAG, dMBP) pathology and inflammation (Iba1) (p>0.05). No significant differences in MRI parameters and pathological grades of white matter integrity were evidenced between APOEKO and WT sham mice (p>0.05). An absence of grey matter abnormalities was evidenced on T2- weighted scans and corresponding H&E stained brain sections in all experimental animals. However, significant reductions in MTR values and dMBP immunoreactivity (myelin pathology) (p<0.05) were observed in grey matter (the hippocampus) following chronic cerebral hypoperfusion in the absence of significant APOE genotype effect (p>0.05) suggesting the existence of both white and grey matter abnormalities in this animal model. Overall, the present neuroimaging data, but not pathological analysis, partially validated the main study hypothesis suggesting that APOE deficiency might be associated with the development of more severe white matter abnormalities in hypoperfused mice. III. Characterization of methylation and hydroxymethylation in white matter under normal physiological and chronically hypoperfused conditions in mice Lastly, I sought to test the hypothesis that chronic cerebral hypoperfusion might alter oxygen dependent DNA hydroxymethylation (5hmC) in white matter regions via perturbations in methylation (5mC) and/ or Ten- eleven translocation proteins (e.g. TET2) in mice. DNA methylation (5mC), hydroxymethylation (5hmC) and TET2 were immunochemically studied in white and grey matter of sham and chronically hypoperfused C57Bl6J mice a month after surgery. The immunochemical results demonstrated significant increases (p<0.05) in 5hmC in the hypoperfused corpus callosum (CC) in the absence of significant hypoperfusion- induced alterations in the distribution of 5mC and TET2 (p>0.05) in white matter. Significant hypoperfusion- induced increases were evident for TET2 in the cerebral cortex (Cx) (p<0.05). These data partially validated the main study hypothesis suggesting hypoperfusion- induced alterations in 5hmC in white matter. However, in contrast to the study hypothesis, the observed hypoperfusion- induced alterations in 5hmC occurred in the absence of changes in 5mC and TET2 in white matter. A subsequent correlation analysis between hydroxymethylation and 5mC, TET2 in the CC failed to show significant associations (p>0.05). In search of the cellular determinants of 5hmC in the CC, hydroxymethylation was examined in relation to some of the cell types in white matter- mature oligodendrocytes, oligodendrolial progenitors (OPC) and microglia both in vivo and in vitro. Specifically, a separate parametric correlation analysis between the proportion of 5hmC positive cells and the respective proportions of mature oligodendrocytes, OPC and microglia in the CC demonstrated that hydroxymethylation correlated significantly only with microglia in vivo (p<0.05). Following this, 5hmC immunochemical distribution was studied in vitro in oligodendroglia cells at different stages of maturation, and interferon γ/ lypopolisaccharide activated and nonactivated microglia. The in vitro analysis demonstrated that 5hmC is high in OPC, activated and nonactivated microglia, but it is low in mature oligodendrocytes. Taken together the in vivo and in vitro cellular analyses suggest that the processes of hydroxymethylation in white matter might be immunoregulated. However, it is possible that in vivo in addition to microglia, other cell types (e.g. astrocytes, OPC) contributed to the presently observed 5hmC upregulation in the hypoperfused CC. Conclusion The experimental work presented in this thesis further developed and characterized a new mouse model of chronic cerebral hypoperfusion by confirming previous behavioural findings (e.g. working memory deficits) and revealing previously undetected spectrum of white and grey matter pathology in this animal model. The thesis demonstrated for the first time by using a newly developed ex vivo MRI procedure that APOE might modulate hypoperfusion- induced white matter pathology in mice. Additional immunochemical analysis revealed important hypoperfusion- induced epigenetic alterations in white (5hmC) and grey (TET2) matter in this animal model. Future experiments on chronically hypoperfused mice would allow to get a better insight into the neurobiological determinants (e.g. white vs. grey matter) underlying the observed cognitive deficits in this animal model, the involved cellular and molecular pathways as well as the functional significance of genetic (APOE) and epigenetic (5hmC, TETs) alterations in the hypoperfused brain. Future experimental work on this animal model would potentially reveal new biological targets for the pre- clinical development of therapies for age- related cognitive decline. Further development and optimization of the newly developed ex vivo MRI procedure would allow its broader application in preclinical settings and would facilitate the translation of experimental findings to clinics.

Intact white matter is critical for normal cognitive function. In traumatic brain injury (TBI), chronic cerebral hypoperfusion and Alzheimer’s disease (AD) damage to white matter is associated with cognitive impairment. However, these conditions are associated with grey matter damage or with other pathological states and the contribution of white matter damage in isolation to their pathogenesis is not known. Furthermore, TBI is a risk factor for AD and cerebral hypoperfusion is an early feature of AD. It is hypothesised that white matter damage following TBI or chronic cerebral hypoperfusion will be associated with cognitive deficits and that white matter changes after injury contribute to AD pathogenesis. To investigate this, this thesis examined the contribution of white matter damage to cognitive deficits after TBI and chronic cerebral hypoperfusion and furthermore, investigated the role of white matter damage in the relationship between TBI and AD. Three studies addressed these aims. In the first, mild TBI was induced in wild-type mice and the effects on axons, myelin and neuronal cell bodies examined at time points from 4 hours to 6 weeks after injury. Spatial reference learning and memory was tested at 3 and 6 weeks after injury. Injured mice showed axonal damage in the cingulum, close to the injury site in the hours after injury and at 6 weeks, damage in the thalamus and external capsule were apparent. Injured and sham animals had comparable levels of neuronal damage and no change was observed in myelin. Injured animals showed impaired spatial reference learning at 3 weeks after injury, demonstrating that selective axonal damage is sufficient to impair cognition. In the second study mild TBI was induced in a transgenic mouse model of AD and the effects on white matter pathology and AD-related proteins examined 24 hours after injury. There was a significant increase in axonal damage in the cingulum and external capsule and parallel accumulations of amyloid were observed in these regions. There were no changes in tau or in overall levels of AD-related proteins. This suggests that axonal damage may have a role in mediating the link between TBI and AD. The third study used a model of chronic cerebral hypoperfusion in wild type mice and investigated white matter changes after one and two months of hypoperfusion as well as a comprehensive assessment of learning and memory. Chronic cerebral hypoperfusion resulted in diffuse myelin damage in the absence of ischaemic neuronal damage at both 1 and 2 months after induction of hypoperfusion. Hypoperfused animals also showed minimal axonal damage and microglial activation. Cognitive testing revealed a selective impairment in spatial working memory but not spatial reference or episodic memory in hypoperfused animals, showing that modest reductions in blood flow have effects on white matter sufficient to cause cognitive impairment. These results demonstrate that selective damage to white matter components can have a long-term impact on cognitive function as well as on the development of AD. This suggests that minimisation of axonal damage after TBI is a target for reducing subsequent risk of AD and that repair or prevention of white matter damage is a promising strategy for rescuing cognitive function in individuals who have experienced mild TBI or chronic cerebral hypoperfusion.

Cerebral small-vessel damage manifests as white matter hyperintensities and cerebral atrophy on brain MRI and is associated with aging, cognitive decline and dementia. We sought to examine the interrelationship of these imaging biomarkers and the influence of hypertension in older individuals. We used a multivariate spatial covariance neuroimaging technique to localize the effects of white matter lesion load on regional gray matter volume and assessed the role of blood pressure control, age and education on this relationship. Using a case-control design matching for age, gender, and educational attainment we selected 64 participants with normal blood pressure, controlled hypertension or uncontrolled hypertension from the Northern Manhattan Study cohort. We applied gray matter voxel-based morphometry with the scaled subprofile model to (1) identify regional covariance patterns of gray matter volume differences associated with white matter lesion load, (2) compare this relationship across blood pressure groups, and (3) relate it to cognitive performance. In this group of participants aged 60-86 years, we identified a pattern of reduced gray matter volume associated with white matter lesion load in bilateral temporal-parietal regions with relative preservation of volume in the basal forebrain, thalami and cingulate cortex. This pattern was expressed most in the uncontrolled hypertension group and least in the normotensives, but was also more evident in older and more educated individuals. Expression of this pattern was associated with worse performance in executive function and memory. In summary, white matter lesions from small-vessel disease are associated with a regional pattern of gray matter atrophy that is mitigated by blood pressure control, exacerbated by aging, and associated with cognitive performance.

Cette thèse est consacrée à la cartographie de la microstructure du cerveau humain par IRM quantitative et de diffusion. L'imagerie quantitative T1/T2 repose sur des séquences dédiées à la cartographie des temps de relaxation T1 et T2. Leurs variations au sein du tissu sont liées aux différents compartiments d'eau issus d'organisations spécifiques à l'échelle cellulaire. Mesurer ces paramètres quantitatifs permet donc de mieux caractériser la microstructure du tissu cérébral. L'IRMd étudie le mouvement brownien des molécules d'eau dans le tissu cérébral dans lequel leur mouvement est contraint par des barrières naturelles, telles que les membranes cellulaires. Ainsi, les informations sur leurs déplacements contenues dans le signal de diffusion permettent de révéler la cytoarchitecture sous-jacente. La combinaison de ces deux modalités donne donc une unique possibilité de mieux sonder la microstructure du tissu cérébral. Ce travail vise à mettre en place la méthodologie permettant d’étudier chez l'homme, in vivo, la microstructure de la matière blanche cérébrale. La première partie inclut l'acquisition d'une base IRM unique de 79 sujets sains (base Archi/CONNECT) incluant des données anatomiques à haute résolution spatiale, des données de relaxométrie, des données de diffusion à haute résolution angulaire, et fonctionnelles. Ces données ont permis dans un premier temps de construire le premier atlas de la connectivité anatomique du sujet sain grâce à la segmentation automatique des grands faisceaux de la substance blanche de tous les sujets. Cet atlas fournit un repère anatomique au sein de la substance blanche pour ensuite étudier pour chaque faisceau les paramètres quantitatifs caractérisant son organisation microstructurelle. L'accent a d’abord été mis sur la construction du premier atlas des profils des temps de relaxation T1/T2 le long des grandes voies de la matière blanche. Ces profils ont ensuite été corrélés avec les profils quantitatifs issus de l'imagerie de diffusion (fraction d'anisotropie, diffusivités radiales et longitudinales, coefficient de diffusion apparent) pour mieux comprendre leurs relations et d'expliquer la variabilité le long des faisceaux et les asymétries interhémisphériques. La deuxième partie de cette thèse fut centrée sur la modélisation du tissu cérébral à l'échelle cellulaire pour extraire des paramètres quantitatifs caractérisant la microstructure, tel que le diamètre et la densité des axones. Une séquence d’IRMd a été développée sur les imageurs 3T et 7T cliniques de NeuroSpin, permettant de jouer n'importe quelle forme de gradients et ainsi de s'inscrire dans une démarche où cette forme résulte d'une optimisation sous l'hypothèse d'un modèle géométrique du tissu et sous contraintes matérielles et temporelles liées aux applications cliniques. Cette séquence a été utilisée pour scanner 14 sujets sains afin de construire le premier atlas du diamètre et de la densité locale des axones. Nous avons également proposé un nouveau modèle géométrique de l'axone, divisant le compartiment axonal, habituellement modélisé par un simple cylindre en deux compartiments: le premier correspondant aux molécules d’eau à diffusivité lente, entourant la membrane de l'axone et le second correspondant aux molécules plus loin des membranes à plus forte diffusivité, moins restreinte. Nous avons mené une étude théorique montrant que sous des conditions cliniques, utiliser ce nouveau modèle pourrait aider à limiter la surestimation des petits axones que l’on observe dans les études actuelles. Pour aller plus loin dans la physiopathologie de l'autisme, nous avons rajouté au protocole d'imagerie à 3T déjà en place la séquence que nous avons développée afin de cartographier le diamètre et la densité des axones et ainsi mieux comprendre les atrophies au sein du corps calleux, initialement observées via des paramètres moins spécifiques tels que la fraction d'anisotropie. D'autres applications cliniques sont à venir. / This thesis is focused on the human brain microstructure mapping using quantitative and diffusion MRI. The T1/T2 quantitative imaging relies on sequences dedicated to the mapping of T1 and T2 relaxation times. Their variations within the tissue are linked to the presence of different water compartments defined by a specific organization of the tissue at the cell scale. Measuring these parameters can help, therefore, to better characterize the brain microstructure. The dMRI, on the other hand, explores the brownian motion of water molecules in the brain tissue, where the water molecules’ movement is constrained by natural barriers, such as cell membranes. Thus, the information on their displacement carried by the dMRI signal gives access to the underlying cytoarchitecture. Combination of these two modalities is, therefore, a promising way to probe the brain tissue microstructure. The main goal of the present thesis is to set up the methodology to study the microstructure of the white matter of the human brain in vivo. The first part includes the acquisition of a unique MRI database of 79 healthy subjects (the Archi/CONNECT), which includes anatomical high resolution data, relaxometry data, diffusion-weighted data at high spatio-angular resolution and functional data. This database has allowed us to build the first atlas of the anatomical connectivity of the healthy brain through the automatic segmentation of the major white matter bundles, providing an appropriate anatomical reference for the white matter to study individually the quantitative parameters along each fascicle, characterizing its microstructure organization. Emphasis was placed on the construction of the first atlas of the T1/T2 profiles along the major white matter pathways. The profiles of the T1 and T2 relaxation times were then correlated to the quantitative profiles computed from the diffusion MRI data (fractional anisotropy, radial and longitudinal diffusivities, apparent diffusion coefficient), in order to better understand their relations and to explain the observed variability along the fascicles and the interhemispheric asymmetries. The second part was focused on the brain tissue modeling at the cell scale to extract the quantitative parameters characterizing the geometry of the cellular membranes, such as the axonal diameter and the axonal density. A diffusion MRI sequence was developed on the 3 Teslas and 7 Teslas Siemens clinical systems of NeuroSpin which is able to apply any kind of gradient waveforms to fall within an approach where the gradient waveform results from an optimization under the hypothesis of a geometrical tissue model, hardware and time constraints induced by clinical applications. This sequence was applied in the study of fourteen healthy subjects in order to build the first quantitative atlas of the axonal diameter and the local axonal density at 7T. We also proposed a new geometrical model to model the axon, dividing the axonal compartment, usually modelled using a simple cylinder, into two compartments: one being near the membranes with low diffusivity and one farer from the membranes, less restricted and with higher diffusivity. We conducted a theoretical study showing that under clinical conditions, this new model allows, in part, to overcome the bias induced by the simple cylindrical model leading to a systematic overestimation of the smallest diameters. Finally, in the aim of going further in the physiopathology of the autism, we added to the current 3T imaging protocol the dMRI sequence developed in the framework of this thesis in order to map the axonal diameter and density. This study is ongoing and should validate shortly the contribution of these new quantitative measures of the microstructure in the comprehension of the atrophies of the corpus callosum, initially observed using less specific diffusion parameters such as the generalized fractional anisotropy. There will be other clinical applications in the future.