Cerebrovascular and blood–brain barrier impairments in Huntington's disease: Potential implications for its pathophysiology

Although the underlying cause of Huntington's disease (HD) is well established, the actual pathophysiological processes involved remain to be fully elucidated. In other proteinopathies such as Alzheimer's and Parkinson's diseases, there is evidence for impairments of the cerebral vasculature as well as the blood–brain barrier (BBB), which have been suggested to contribute to their pathophysiology. We investigated whether similar changes are also present in HD.

cytoplasm and nucleus. While the pathogenic effects of aggregate formation are still being debated, 2,3 the available evidence suggests that the oligomeric form of the protein is the most toxic (for review, see 4 ).
Given the genetic origin of HD, the process of mHtt oligomerization, aggregate formation, and associated pathology has been believed to be essentially cell autonomous. However, a number of recent studies have challenged this assumption. 5,6 In particular, our recent postmortem analysis of HD patients who had been in receipt of fetal striatal allografts has revealed the presence of mHtt aggregates within the genetically unrelated grafted tissue, in cells associated with blood vessels as well as within perivascular macrophages. 7 This not only provided the first in vivo demonstration of mHtt spread in patients with a monogenic human neurodegenerative disorder of the central nervous system (CNS), but also suggested the existence of non-cell-autonomous mechanisms of pathological protein transmission. These findings, along with recent work demonstrating that mHtt is expressed in peripherally circulating monocytes, 8 raise the intriguing possibility that mHtt might be transported into the CNS by peripheral blood-borne cells. Furthermore, There is increasing evidence that a number of neurodegenerative disorders are associated with alterations in the cerebrovasculature, including the neurovascular unit and the blood-brain barrier (BBB), 9,10 which may facilitate access of peripheral cells to the CNS compartment.
In this regard, we and others have recently reported that there are changes in cerebrovascular vessel density in HD patients and in transgenic mouse models of the disease. [11][12][13] However, aside from a few studies conducted in the 1990s showing reduced cerebral blood flow in the caudate of HD patients secondary to cell loss, [14][15][16][17] the vascular changes and integrity of the BBB have not been systematically investigated despite their obvious clinical and therapeutic relevance. We now present the first extensive study in both the R6/2 transgenic mouse model and patients with HD investigating mHtt expression in cells associated with blood vessels, as well as its impact on vascular changes and BBB permeability.

Human Studies
HD patients and age-and sex-matched control subjects were recruited from the HD clinic at the John van Geest Centre for Brain Repair (Cambridge, UK) ( Table 1). None of the subjects had a history of vascular disease. Participants provided informed consent, and procedures were conducted under Research Ethics Committee approval (REC 12/EE/0528). Burden of disease score was calculated, as previously described. 18 3T MAGNETIC RESONANCE IMAGING. Subjects were scanned using a 3T Siemens TimTrio system (Siemens Medical Solutions, Erlangen, Germany). Structural imaging was achieved with an magnetization-prepared rapid acquisition gradient echo acquisition (1mm 3 isotropic resolution). Dynamic contrastenhanced (DCE) imaging used three-dimensional fast low-angle shot (1.8 3 1.8 3 5mm 3 spatial/6-sec temporal resolution). To measure T 1 images, flip angles [FA] from 5 to 35 were obtained. A fixed FA of 25 was used to acquire 120 frames. Gadobutrol (9ml; 1.0mmol/ml Gd-DTPA; Bayer Schering Pharma AG, Berlin, Germany) was injected into the medial cubital vein at 5ml/sec followed by a saline flush of 20ml. Image processing was performed using MATLAB (MathWorks, Natick, MA) and SPM12 (UCL, London, UK). The Parker arterial input function 19 was used to fit permeability (K trans ) and voxel-wise vessel fraction using Patlak analysis. 20 Statistical parametric mapping was used to produce voxel-wise maps for K trans differences between patients and controls, masked using segmented gray and white matter maps. Multiple comparisons were controlled using family-wise error correction. Hand-drawn regions of interest (ROIs) for the caudate and putamen were used to extract mean values. ROIs were produced using Analyze 12 (Mayo Clinic, Rochester, MN). Within each ROI, outlying values were excluded by taking the mean of values within the 1 st 299 th percentiles.
7T MAGNETIC RESONANCE IMAGING. Subjects were scanned on a 7T Philips Achieva system (Philips Healthcare, Andover, MA) with head transmit and 32-channel receive coil, using image-based shimming to correct for field inhomogeneities. A Look-Locker echo-planar imaging (LL-EPI), flowsensitive alternating inversion recovery arterial spin labeling (ASL) protocol 21-23 was used to generate arterial cerebral blood volume (aCBV) maps. Acquisition parameters: 21 LL-EPI readout pulses per 4.8-second label/control pair; gradient echo/echo planar imaging (EPI) readout (FA 55 o ; echo time [TE] 10ms; field of view 216 3 192mm and 2mm in-plane resolution; slice thickness 4mm); in-plane saturation; initial delay 150ms, LL-EPI readout spacing 100ms; 15/230mm selective/nonselective inversion; 20 averages. aCBV acquisitions were collected at 3 slice locations parallel to the anterior and posterior commissures line. A base magnetization image was collected to estimate M 0blood for aCBV quantification and inversion-recovery images to form a T 1 map. An 0.8-mm isotropic phase-sensitive inversion recovery (PSIR) anatomical image was acquired to define ROIs and assess atrophy. A turbo field EPI was used to generate magnetic resonance angiograms (EPI factor 3; repetition time/ TE 19.12/10.34ms; FA 7 ; voxel size 0.6mm isotropic; matrix 384 3 384 3 100). Caudate and putamen ROIs were defined from the PSIR scan, and a cortical gray matter ROI was formed by thresholding the T 1 map at 1.7 T 1 2.3 seconds. Average aCBV (ml blood/100ml tissue) and arterial transit time (seconds) were calculated using a two-compartment vascular kinetic model as described in a previous work. 23 POSTMORTEM HUMAN BRAIN TISSUE. Postmortem human brain tissue was obtained from the Cambridge Brain Bank Drouin-Ouellet: Vascular impairments in HD August 2015 (Cambridge, UK) and used under local ethical approval (REC 01/177). Severity of HD was graded by a certified pathologist according to the Vonsattel grading system. 24 All postmortem analyses conducted in Canada were approved by the Comit e d' ethique de la recherche du CHU de Qu ebec (#A13-02-1138). CAG repeat size in the HTT gene was determined for each HD case by DNA sequencing (Sanger Sequencing Services; Laragen Inc., Culver City, CA) (see Table 2 for details on the samples used). Note that initial data analysis included n 5 32 HD patients and n 5 16 controls, but we further restricted our analyses to include only patients with 40 hours post mortem delays (n 5 22 HD patients; n 5 9 controls). Similar results were obtained with both cohorts.
IMMUNOFLUORESCENCE IMAGING OF HUMAN TIS-SUE. Paraffin-embedded sections were stained as previously described. 25 For confocal microscopy, tissue preparation, and immunofluorescent staining, procedures were performed as previously described. 7,11 Images were acquired using either the simultaneous confocal/multichannel fluorescence mode of the TISSUEscope TM digital slide scanner and accompanying softwares from Huron Digital Pathology (Waterloo, Ontario, Canada) or a Fluoview FV1000 confocal microscope system equipped with 559-and 635-nm laser diodes and an Ar 488nm laser (Olympus Canada Inc., Richmond Hill, Ontario, Canada). For the quantification of the percentage of von Willebrand factor (vWF) 1 and laminin 1 blood vessels expressing EM48 1 aggregates, m 5 627 and m 5 173 were assessed in 8 and 6 fields of view, respectively. Diameter of blood vessels analyzed ranged from 5 to 94mm.

BLOOD VESSEL DENSITY AND MORPHOMETRIC AND
EXTRAVASATION ANALYSES. Images of the entire putamen were taken at 2-mm intervals using Stereo Investigator software (MBF Bioscience, Williston, VT) integrated to an E800 Nikon microscope (Nikon Canada, Mississauga, Ontario, Canada), and analyses were performed using ImageJ (National Institutes of Health [NIH], Bethesda, MD; http://imagej.nih. gov/ij). For blood vessel leakage analysis, masks from collagen IV and fibrin were obtained and assigned a color (magenta

MOUSE TISSUE PREPARATION AND IMMUNOFLUORES-
CENCE. Mouse brains were collected, sliced, and free-floating sections were stained as previously described 26 (see Table 3 for antibody list). For the lectin/albumin double labeling, DyLight 594-conjugated lectin (1:100; Vector Laboratories, Burlingame, CA) was used. Sections were observed as described above (cf. Immunofluorescence Imaging of Human Tissue).

BLOOD VESSEL DENSITY AND MORPHOMETRIC AND
EXTRAVASATION ANALYSES. Three striatal sections per mice, 250mm apart, were selected, and a 203 magnification image was taken in each quadrant of the striatum. Extravasation analysis of albumin was performed as described for fibrin in the human study.

BRAIN PERFUSION AND VASCULAR CASTING FOR SCAN-
NING ELECTRON MICROSCOPY. Mice were perfused first with 0.9% saline, followed by 100% ethanol. LR White Accelerator was added to LR White Embedding Medium (Electron Microscopy Sciences, Hatfield, PA), and mice were immediately perfused using this mixture. Tissue was incubated overnight at 37 C, and the striatum was dissected and soaked in a 4 M KOH solution for 4 weeks. Cast samples were then mounted onto a metallic support and observed with a scanning electron microscope (JEOL 6360LV; JEOL, Tokyo, Japan) set at 30kV.   CA). Relative gene-level differences between groups were evaluated with the DDC t method.
WESTERN BLOT ANALYSIS FOR MOUSE TISSUE. Striatal tissue (20mg) was homogenized and Western blot experiments were carried out and quantified as described for human tissues.
STATISTICAL ANALYSIS. For studies involving human tissue samples, the pathological grade was taken into account in all initial statistical analyses, but given that no significant effect was detected, data from HD cases were pooled and compared to control subjects. Statistical significance was set at p < 0.05. All statistical analyses were performed using Prism (

Expression of mHtt Pathology in Cerebral Blood Vessels
Building on our previous observation that mHtt aggregates were found in the basal membrane of cerebral blood vessels of HD patients, 7 we first assessed the expression of mHtt aggregates in all compartments of the neurovascular unit and observed such aggregates inside the basal membrane sheaths, as well as in the nuclei of cells embedded in the basal membrane in both small-(5-10mm) and large-caliber (>20mm) blood vessels (Fig. 1A-C). mHtt aggregates were found to colocalize with von Willebrand factor (vWF)labeled endothelial cells (Ecs; Fig. 1D-G), alphasmooth muscle actin (a-SMA) 1 cells within the basal membrane (Fig. 1H) and CD163 1 perivascular macrophages (Fig. 1I). EM48 1 aggregates were found in the basal membrane and endothelium of 3 and 7% of cortical blood vessels, respectively. Similar to our observations in postmortem samples from HD cases, EM48 1 aggregates could also be found both in the basal membrane and inside cells that were embedded in the basal membrane of small and larger blood vessels in R6/2 mice, specifically within a-SMA 1 pericytes or smooth muscle cells ( Fig. 2A-D), as well as within CD31 1 Ecs (Fig. 2E).

Vascular Changes in the Putamen of HD Patients
Given that mHtt can impair mitochondrial integrity and induce oxidative stress, which may affect the physiology of cells forming the cerebral vasculature, 32,33 we next investigated morphological, structural, and functional changes in blood vessels in the context of HD using MRI and postmortem HD tissue. There was a significant increase in aCBV in cortical gray matter in the HD cohort (p < 0.05; Fig. 3A), but not in the putamen nor caudate, and the aCBV did not correlate with gray matter volume (data not shown). No significant difference was found in arterial transit time between HD and control participants (Fig. 3B, C). Qualitative assessment of 7T MRI of patients did not yield overt differences in blood vessel density for largecaliber blood vessels (>600mm), although it was not possible to perform proper quantifications given the limited number of collected angiograms (Fig. 3D). We thus used postmortem tissue samples to specifically assess potential morphological differences in blood vessels of the putamen of HD patients. Measurements of cerebral blood vessel density revealed an increase in HD subjects (p < 0.001; Fig. 4A, B), dependent of putaminal atrophy (R 2 5 0.24; p 5 0.032; Fig. 4C). We found no difference in the diameter of small-or mediumsized blood vessels (5-10 and 10-20mm diameter, respectively) between HD and matched controls (Fig. 4D), nor in the Feret diameter (data not shown). However, we found an increased proportion of small compared to medium-sized blood vessels, in the putamen of HD patients (p < 0.001; Fig. 4E). Interestingly, a significant positive correlation was obtained between the size of blood vessels and the surface area of the putamen (R 2 5 0.49; p < 0.001; Fig. 4F), as well as between the Feret diameter and the surface area of the putamen (R 2 5 0.34; p < 0.01; Fig. 4G), suggesting that putaminal atrophy was accompanied by a reduction in blood vessel size. Finally, we measured the wall thickness of the basal membrane, given that this is among the characteristic features of vascular pathology in other neurodegenerative diseases, such as Alzheimer's disease, but did not detect any differences (Fig. 4H). Overall, although the morphology of large-caliber blood vessels remains intact in the brain of HD patients, there is an increase in the number of small blood vessels, an effect that appears to occur concurrently with putaminal degeneration.

Morphological Changes in Striatal Blood Vessels of R6/2 Mice
Given the similar pattern of expression of mHtt aggregates within the neurovascular units of the R6/2 mouse and HD patients, we further used the R6/2 mouse to investigate blood vessel changes. In line with our observations in human postmortem samples, the density of striatal blood vessels in R6/2 mice was increased (p < 0.01; Fig. 5A, B). This was accompanied by a reduction in the average Feret diameter of blood vessels (p < 0.05; Fig. 5C), although apparent differences in size vanished upon distinguishing arterioles from capillaries and venules (Fig. 5D), suggesting that the decrease in Feret diameter of blood vessels in R6/2 mice is not specific to particular a blood vessel subtype(s). Scanning electron microscopy revealed further , as compared to controls (n 5 6). (B) Collagen IV immunofluorescent staining revealed an increase in the density of blood vessels in the putamen of an HD case, as well as the matching black-and-white masks used for blood vessel measurements. (C) Negative correlation of blood vessel density and putaminal atrophy in HD patients. (D) Quantification of blood vessel diameters of small-and medium-sized blood vessels did not show any differences, but uncovered an increase in the proportion of small blood vessels (5-10mm) and a decrease in the proportion of larger-caliber blood vessels (10-20mm) in the HD putamen (E). Average size (F) and average Feret diameter (G) positively correlated with the surface area of the putamen. (H) Wall thickness of small (5-10mm) and larger-caliber blood vessels (10-20mm) did not differ from controls. Data are expressed as means 6 standard error of the mean. Statistical analysis was performed using an analysis of covariance with age and gender as covariates followed by a Bonferroni post-hoc test. Coefficients of correlation were determined using the Pearson correlation test. *p < 0.05; **p < 0.01 versus controls. CTL 5 controls; HD 5 Huntington's disease. Scale bar B 5 100mm.
abnormalities in the striatal vascular network, which was characterized by a smaller ratio of the diameter of branches to that of the parent vessel, compared to wild-type (WT) littermates (p < 0.05; Fig. 5E, F). Additionally, several aborted branch buds were observed in the striatum of the R6/2, but not of WT littermates (Fig. 5E). We next assessed whether these structural changes had an impact on the volume of the striatal vascular network using in situ cerebral perfusion, and found that the V vasc of the striatal network in R6/2 mice and WT littermates was similar. Brain uptake clearance (Cl up ) of [ 3 H]diazepam, which measures the surface area of the vascular network, was also similar in R6/2 mice and controls (Fig. 5G). This supports the scanning electron microscopy and immunofluorescence data, showing that the brain vasculature anomalies were restricted to smaller vessels.

BBB Leakage in HD Patients
We next assessed whether the observed morphological and structural changes in cerebral blood vessels of HD patients are associated with functional impairments of the BBB. Decreases in expression of proteins forming the TJs of the BBB, such as occludin and claudin-5, have been associated with increased permeability. 34 Western blot analysis revealed a significant decrease in both proteins (p < 0.001 and p < 0.01, respectively) in the putamen of HD patients (Fig. 6A). While we did not detect a significant change in vascular endothelial growth factor and PDGFb expression, a significant increase in other markers associated with increased BBB permeability, such as hepatocyte growth factor (p < 0.01), interleukin-8 (p < 0.05) and tissue inhibitor of metalloproteinase 1 (p < 0.05; Fig. 6D) was observed. To confirm that BBB breakdown occurred in the putamen of HD patients, we next measured the levels of extravascular fibrin deposition 35 and found it to be increased 2.5-fold, when compared to controls (p < 0.05; Fig. 6B, C). Finally, we assessed whether BBB leakage might be detected in mild-to-moderate stage HD patients using a DCE MRI method. Overall, we found a trend toward an increased leakage in the caudate and putamen of HD patients, as compared to age-and sex-matched control subjects (Fig. 7A, B). Interestingly, BBB leakage in the right caudate significantly correlated with the burden of disease score (R 2 5 0.69; p < 0.05; Fig. 7C), suggesting that BBB permeability increases with disease progression in HD.

Paracellular and Transcytotic Alterations in the Striatum of R6/2 Mice
We further characterized paracellular transport and transcytosis across the BBB in R6/2 mice. While TJ length in the striatum of R6/2 and WT mice was similar, the extracellular cleft on the side of the lumen was wider in the transgenic animals (p < 0.05, Fig. 8A). Additionally, TJs had a wider angle with respect to the lumen in R6/2 mice (p < 0.05, Fig. 8B, C), suggesting a random junctional alignment in the HD model. 36 Moreover, claudin-5 and occludin were both significantly decreased in R6/2 mice (p < 0.05; Fig. 8D). Finally, given that TJs may facilitate leukocyte diapedesis through weakened interendothelial contact points, 37 we looked for the presence of perivascular cells and found an increased number of CD45 1 leukocytes attached to the blood vessel walls in R6/2 (Fig. 8E).
In addition to the paracellular barrier formed by the TJs, the BBB provides a transcellular hindrance. 38 We thus assessed transcytosis in R6/2 mice and found an increase in the total number of vesicles in the endothelium of striatal blood vessels (p < 0.05; Fig. 9A). 31 Upon evaluating the different subtypes of vesicles, we found a and occludin presented as a GAPDH and surface area of blood vessel ratio revealed a significant decrease of both proteins in the putamen of HD patients (n 5 13), as compared to controls (n 5 5). (B) Method used for the quantification of the extravascular fibrin staining. A yellow mask of the fibrinogen staining and a magenta mask of the collagen IV staining were generated using ImageJ (NIH, Bethesda, MD) and merged using Adobe Photoshop. The magenta and white (from merging magenta and yellow) were then deleted, leaving only the extravascular fibrinogen staining (yellow), the surface area of which was quantified for each sample using imageJ. (C) Quantification of the extravascular fibrin positive staining showed an increase in fibrin levels outside the blood vessels in the putamen of HD patients (n 5 9), compared to controls (n 5 4). (D) Expression of protein markers associated with increased BBB permeability and angiogenesis was assessed using a multiplex ELISA assay in the human putamen of HD patients (n 5 9-18) and matched controls (n 5 5-9). Statistical analyses were performed using an analysis of covariance with age and sex as covariates, followed by a Bonferroni post-hoc test. *p < 0.05; **p < 0.01; ***p < 0.001, as compared to controls. CTL or C 5 controls; Coll IV 5 collagen IV; HD or H 5 Huntington's disease. Scale bar C 5 100mm.
5-fold increase in luminal type II vesicles-connected vesicles pinching from the luminal plasma membranesuggesting an increase in pinocytotic events (p < 0.05; Fig. 9B, C). Immunogold labeling of EM48 also revealed the presence of gold particles in the endothelium of R6/ 2 mice, which were further detected within all the different subtypes of vesicles (Fig. 9D). In accord with increased transcytosis in the striatum of R6/2 mice, we found a nearly 20-fold increase in extravascular albumin in transgenic mice (p < 0.05; Fig. 9E, F), suggesting a disruption of BBB integrity. 39 Because pericytes play an important role in BBB integrity and pericyte-deficient mice (pdgfrb -/mice) display increased transcytosis, 36 we further measured pericyte coverage, but did not find significant differences (Fig. 9G). However, messenger RNA expression of PDGFb was significantly decreased (p < 0.01; Fig. 9H). Taken together, these findings suggest an increased BBB permeability in R6/2 mice, which includes increased transcytosis and paracellullar transport.

Discussion
Our data provide clear evidence for ultrastructural, morphological, and functional changes in cerebral blood vessels with alterations of the BBB in HD, as evidenced both in an animal model of disease, the R6/2 mouse, and in humans as shown by MRI in mild-to-moderate stage HD patients as well as in postmortem tissue.
Although the specific pathophysiological effects of mHtt expression within these neurovascular compartments are as yet unknown, neural activity has been shown to affect vascular structure. 40,41 It is possible that the changes reported here are secondary to the effects of mHtt on neural activity, although this seems unlikely given the extent of changes observed. 42,43 In addition, brain Ecs contain large numbers of mitochondria, as compared to peripheral endothelium, and are thus more susceptible to oxidative stress, 32,33 which has been proposed as an effector of BBB damage and endothelial dysfunction through alterations in the expression of TJassociated proteins. 44 An intrinsic dysfunction of Ecs secondary to their accumulation of mHtt is thus more likely to be the main determinant of the morphological and functional changes observed in our study. We have recently shown that allografts of normal tissue are not well vascularized in the brain of HD patients, 11 which could reflect an intrinsic limitation of the cerebral blood vessels in HD in the remodeling and proper vascularization of new tissue. Moreover, mHtt is expressed in astrocytes, and it has been reported that mHtt expression in this cell type could mediate signs of neurological impairments associated with HD. 5 Given that astrocytes are an important component of the BBB, the expression of mHtt in these cells could also participate in the vascular impairments that occur in HD.
The finding that the size of blood vessels decreases concomitantly with the atrophy of the putamen is surprising, but might be a consequence of the increased number of smaller blood vessels observed here, along with the decrease in the number of arteries. Oxygen diffusion is different in arterioles, venules, and capillaries, with most oxygen diffusion taking place across the capillary bed. An increase in the total surface of capillary beds could help maintain proper oxygenation of the surrounding tissues, despite a decrease in the number of larger blood vessels. This hypothesis may hold, provided that the increase in capillary surface compensates for the reduced primary blood flow and dissolved oxygen flow rate that results from the decreased number of large vessels. Alternatively, a reduced rate of oxygen delivery across a compromised vascular network in HD might be the primary event that triggers a compensatory, secondary increase in vasculature. Furthermore, aCBV-which, in essence, measures the volume of blood in the arterial, noncapillary component-was increased in the gray matter of HD patients, in line with previous observations in R6/2 mice. 13,45 However, the R6/2 mouse model is not characterized by overt neuronal death, 27 which argues against a direct involvement of neuronal loss in the various changes in the vasculature, as does the fact that the increase in density and decrease in the size of blood vessels have been previously reported to develop as early as 5-7 weeks of age in these mice. 13 Overall, the available evidence supports the existence of a vascular pathology in HD that is independent from the neuronal loss characteristic of the disease.
The HD-associated abnormalities in the vasculature we observed have functional consequences, in that the BBB is compromised in both R6/2 mice and HD patients, as we show here for the first time. While lipopolysaccharide-induced BBB leakage has been reported in the YAC128 model, BBB disruption using Evans Blue dye injection was not detected in absence of the inflammatory challenge. 12 Two previous studies using DCE-MRI in R6/2 mice with Gd-DTPA as a contrast agent have also failed to show any BBB leakage. 13,45 However, the sensitivity of this method in small animals is lower and may thus not afford detection of subtle . (E) A greater number of CD45 1 perivascular macrophages (white) were associated with blood vessels (collagen IV; red) in the R6/2 mouse, as compared to WT. Data are expressed as means 6 standard error of the mean. Statistical analysis was performed using an unpaired Student t test, with a Welch correction if variances were not equal. *p < 0.05 versus controls. Coll IV 5 collagen IV; DAPI 5 4',6-diamidino-2-phenylindole; L 5 lumen; TJ 5 tight junction; WT 5 wild type. Scale bars C 5 100nm; E 5 30mm. changes in BBB permeability. In contrast, our study now shows that in HD patients (using the same contrast agent), there is a trend toward an increased permeability of the BBB-that is, a loss of BBB integrity-with a concomitant increase in the burden of disease score. The reason why only a trend could be definitively established probably relates to the small sample size (n 5 7) and the fact that we restricted our selection of participants to mild-to-moderate stage HD to avoid issues related to movement-dependent artefacts in the scanner secondary to their chorea. However, the significance of this trend is reinforced by our postmortem analyses, which revealed unequivocal signs of BBB leakage both in R6/2 mice and HD patients. (D) mHtt aggregates were detected using EM48 immunogold labeling in the endothelium of R6/2 mice (inset 3) as well as in luminal type I (insets 1, 2, and 5), type II (inset 6), cytoplasmic (inset 8), and abluminal vesicles (inset 4). Gold particles were also frequently found in pericytes of R6/2 mice (inset 7). (E, F) Quantification of albumin 1 staining (green) showed an increase of albumin in the striatal parenchyma of R6/2 mice (n 5 8), as compared to WT littermates (n 5 5). Blood vessels were stained using lectin-594 (red). (G) Transmission electron microscopy analysis of pericyte coverage did not unveil any difference between R6/2 (n 5 6) and WT littermates (n 5 6). (H) qPCR quantification of PDGFb in the mouse striatum showed a decrease in R6/2 mice (n 5 10), as compared to WT (n 5 10). Data are expressed as means 6 standard error of the mean. Statistical analysis was performed using an unpaired Student t test, with a Welch correction when variances were unequal. *p < 0.05; **p < 0.01, as compared to controls. E 5 endothelium; L 5 lumen; PDGFb 5 platelet-derived growth factor beta; WT 5 wild type. Scale bars B 5 100nm; D 5 200nm; E 5 100mm.
Claudin-5 is a major functional constituent and a critical determinant of BBB paracellular permeability in mice by creating size-(<800Da) and charge-selective hydrophilic paracellular pores, 46,47 whereas occludin enhances TJ tightness. Our finding of decreases in the expression of these TJ proteins in both R6/2 mice and the HD putamen, along with the increase found here in the number of leukocytes in the perivascular space of cerebral blood vessels, are likely to reflect an increase in circulating blood cells transiting into the brain of HD patients. Moreover, the extravascular accumulation of albumin in the R6/2 brain suggests an increase in transcytosis, given that this protein can be transported through the endothelium by transcytotic vesicles, although increased paracellular transport might also contribute to the process. 48 The extravascular accumulation of fibrin deposition that we observed in HD brains might be in accord with this hypothesis, given that it is a well-known marker of BBB leakage in human autopsy tissues. 49,50 The BBB is vital for protecting the CNS from systemic perturbations and from elements of the peripheral immune system, so that any impairment of its integrity could have farreaching consequences on the health of the CNS. Our data, demonstrating that the integrity of the BBB is compromised and that the cerebral vasculature is considerably disrupted in HD, indicate that these impairments may contribute in driving the disease process. Indeed, these data provide evidence linking the welldescribed peripheral immune changes of HD to the CNS pathology. 51,52 In particular, the changes reported herein in both mice and humans could allow peripheral blood leukocytes to enter the CNS more easily, where they might act as vehicles for mHtt transfer into the brain, or exacerbate neuronal death by participating in the cerebral inflammatory response. The detection of mHtt in transcytotic vesicles, in conjunction with the increased transcytosis observed in R6/2 mice, suggests that an exchange of mHtt between the blood and the brain is possible (Fig. 10). Overall, the present study not only reveals a new pathogenic process underlying HD, but also uncovers potentially novel therapeutic avenues.