SEARCH

SEARCH BY CITATION

Keywords:

  • aging;
  • Alzheimer's disease;
  • basement membranes;
  • cerebral amyloid angiopathy;
  • cerebral vasculature

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgment
  8. Author contributions
  9. References
  10. Supporting Information

Development of cerebral amyloid angiopathy (CAA) and Alzheimer's disease (AD) is associated with failure of elimination of amyloid-β (Aβ) from the brain along perivascular basement membranes that form the pathways for drainage of interstitial fluid and solutes from the brain. In transgenic APP mouse models of AD, the severity of cerebral amyloid angiopathy is greater in the cerebral cortex and hippocampus, intermediate in the thalamus, and least in the striatum. In this study we test the hypothesis that age-related regional variation in (1) vascular basement membranes and (2) perivascular drainage of Aβ contribute to the different regional patterns of CAA in the mouse brain. Quantitative electron microscopy of the brains of 2-, 7-, and 23-month-old mice revealed significant age-related thickening of capillary basement membranes in cerebral cortex, hippocampus, and thalamus, but not in the striatum. Results from Western blotting and immunocytochemistry experiments showed a significant reduction in collagen IV in the cortex and hippocampus with age and a reduction in laminin and nidogen 2 in the cortex and striatum. Injection of soluble Aβ into the hippocampus or thalamus showed an age-related reduction in perivascular drainage from the hippocampus but not from the thalamus. The results of the study suggest that changes in vascular basement membranes and perivascular drainage with age differ between brain regions, in the mouse, in a manner that may help to explain the differential deposition of Aβ in the brain in AD and may facilitate development of improved therapeutic strategies to remove Aβ from the brain in AD.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgment
  8. Author contributions
  9. References
  10. Supporting Information

Cerebral amyloid angiopathy (CAA) results from the accumulation of insoluble amyloid-β (Aβ)peptides in the walls of cortical and leptomeningeal arteries and is observed in more than 90% of Alzheimer's disease (AD) brains (Premkumar et al., 1996; Attems et al., 2011). Vascular amyloid is composed predominantly of Aβ1-40, which is produced following proteolytic cleavage of the amyloid precursor protein (APP). In addition to resulting in degeneration of cerebrovascular smooth muscle and endothelial cells (Davis-Salinas et al., 1995; Van Nostrand et al., 2000), CAA inhibits angiogenesis, impairs vascular tone, and decreases total cerebral blood flow (Beckmann et al., 2003; Paris et al., 2004; Shin et al., 2007). Clinically, the degree of CAA severity correlates with cerebral hypoperfusion, increased risk of microhemorrhage, and degree of cognitive impairment (Pfeifer et al., 2002; Schrag et al., 2011).

CAA associated with aging and AD appears to result from a failure of clearance of Aβ from the brain (Mawuenyega et al., 2010). Aβ is cleared from the brain by enzymatic degradation, uptake by microglia and macrophages, receptor-mediated transport across the endothelium, and perivascular drainage along vascular basement membranes (Miners et al., 2008; Weller et al., 2008; Bell & Zlokovic, 2009). Although much work has been done to characterize the dysfunctional removal of Aβ by microglia and across the endothelium in AD, less is known about how efficient perivascular drainage of Aβ is in the aging brain. At the present time, the relative importance of the various mechanisms of clearance of Aβ remains unclear.

Based on experimental and pathological observations, perivascular drainage of Aβ from the brain seems to occur initially along the basement membranes of capillaries and arterioles, then along the walls of cortical and leptomeningeal arteries and out of the base of the skull to cervical lymph nodes (Goldmann et al., 2006; Carare et al., 2008; Ball et al., 2010). Theoretical modeling suggests that perivascular drainage may be driven by the contrary wave that follows arterial pulsations (Schley et al., 2006). Age-related arteriosclerosis and the resulting reduction in pulse amplitude appear to reduce the force and efficiency of drainage, leading to accumulation of Aβ in the basement membranes as CAA.

Cerebral vascular basement membranes are thin sheets of extracellular matrix that are composed of laminins, collagen IV, nidogens and heparan sulfate proteoglycans, such as perlecan, fibronectin, and agrin (Timpl, 1996). Several lines of evidence suggest that biochemical and morphological alterations to basement membranes contribute to the development of CAA. In vitro incubation of Aβ with laminin, collagen IV, and nidogen has been shown to prevent Aβ fibrillization, whereas agrin and perlecan promote and stabilize Aβ fiber formation (Castillo et al., 1997; Bronfman et al., 1998; Cotman et al., 2000; Kiuchi et al., 2002). Age-related changes in levels of laminin, fibronectin, and perlecan have been reported in mouse brain capillaries (Hawkes et al., 2011). Vacuolization, reduplication, and thickening of capillary basement membranes have also been shown to occur in the human aged and AD brain (Perlmutter, 1994; Shimizu et al., 2009).

For reasons that are as yet unknown, vascular Aβ deposits form preferentially in certain brain areas. In the human brain and especially in AD, CAA is observed predominantly in the parietal and occipital cortices, whereas vessels in the basal ganglia and thalamus are rarely affected (Vinters & Gilbert, 1983). In most APP transgenic mouse models of AD, CAA develops in the cortex and hippocampus, although, in contrast to humans, TgAPPSwDI and TgAPP23 mice also develop CAA in the thalamus (Sturchler-Pierrat et al., 1997; Davis et al., 2004). The anatomical structure of the penetrating cerebral arteries may differ between regions of the brain. For example, arteries that supply the human basal ganglia are invested by two layers of leptomeninges that form the borders of an expandable perivascular space, whereas this space is not present around cortical arteries (Zhang et al., 1990; Pollock et al., 1997). This suggests that regional vulnerability or resistance to the development of CAA might be mediated in part by variations in the efficiency of perivascular drainage of Aβ from different brain areas, associated with differences in the composition or morphology of basement membranes and perivascular spaces.

In this study, we tested the hypotheses that (i) age-related changes in basement membrane thickness and composition differ between brain regions and correlate with the severity of CAA, (ii) the pattern of perivascular drainage of Aβ is dependent on age and brain region, with a more efficient clearance from the thalamus compared with the hippocampus.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgment
  8. Author contributions
  9. References
  10. Supporting Information

Regional differences in age-related basement membrane thickening

To determine if age-related basement membrane thickening is more pronounced in brain areas that are affected by CAA, tissue from the cortex, hippocampus, striatum, and thalamus of 2-, 7-, and 23-month-old mice was examined by transmission electron microscopy. Quantification of the mean thickness of capillary basement membranes revealed no differences between 2- and 7-month-old mice in any brain region (Table 1). Similarly, no differences were noted in the basement membrane thickness of capillaries in the striatum in mice of any age (Supporting Fig. 1A–C, Table 1). However, basement membrane thickness was significantly increased in 23-month-old mice compared with both 2- and 7-month-old mice in capillaries in the cortex (P < 0.05 and P < 0.01, respectively, Supporting Fig. 1D–F), hippocampus (P < 0.001 and P < 0.001, respectively) and thalamus (P < 0.05 and P < 0.01, respectively) (Table 1).

Table 1. Basement membrane thickness (nm) of capillaries by brain region and age group
Brain RegionAge group (months)MeanStandard error2 vs. 7 months2 vs. 23 months7 vs. 23 months
Mean difference (95% CI)a P a Mean difference (95% CI)a P a Mean difference (95% CI)a P a
  1. a

    Multiple comparison results were adjusted with Bonferroni correction.

Striatum267.034.34−7.91 (−26.25, 10.44)0.828−7.20 (−25.55, 11.15)0.9610.71 (−17.64, 19.06)1.000
774.946.62
2374.232.80
Cortex265.795.568.44 (−22.42, 39.30)1.000−31.70 (−62.56, −0.84)0.043−40.14 (−71.00, −9.28)0.008
757.354.33
2397.4912.27
Hippocampus262.582.70−0.92 (−29.48, 27.65)1.000−46.92 (−75.49, −18.35)0.001−46.00 (−74.57, −17.44)0.001
763.504.78
23109.5011.93
Thalamus270.113.6311.49 (−23.44, 46.42)1.000−38.32 (−73.25, −3.39)0.028−49.81 (−84.74, −14.88)0.004
758.625.17
23108.4314.51

Regional changes in basement membrane composition with age

To evaluate how the regional expression of cerebral basement membranes changes with age, brain samples from the cortex, hippocampus, striatum, and thalamus were quantified for collagen IV, laminin, nidogen 2, and fibronectin proteins by Western blotting. Due to difficulties in obtaining sufficient protein for Western blotting after antigen retrieval, expression of perlecan was assessed by immunocytochemistry.

In the cortex, there was a significant decrease in the levels of collagen IV in the 23-month-old mice, compared with both 2- and 7-month-old mice (P < 0.05 for both ages, Fig. 1A). The amount of laminin was also significantly decreased in the 23-month-old mice compared with the 2- and 7-month-old mice (P < 0.01 for both ages, Fig. 1B). Levels of nidogen 2, a proteoglycan with the role of cross-linking glycoproteins in basement membranes, were also significantly reduced in the cortex of 7- and 23-month-old mice when compared with 2-month-old mice (P < 0.05 and P < 0.01, respectively, Fig. 1C). In contrast, levels of the N-terminal fragment of fibronectin were significantly increased between 23- and 2- and 7-month-old mice (P < 0.01 for both ages, Fig. 1D). The immunoreactivity for perlecan was also upregulated in cortical vessels of 23-month-old mice vs. 2- and 7-month-old mice (P < 0.01 for both ages, Table 2, Supporting Fig. 2A–C).

Table 2. Perlecan immunoreactivity (% area covered) in each brain region and age group
Brain RegionAge (months)Mean (± SEM) P *
  1. Comparison with 23-month-old mice, **P < 0.01.

Striatum21.6 ± 0.3**<0.01
72.0 ± 0.3**<0.01
234.4 ± 0.4
Cortex22.3 ± 0.2**<0.01
72.2 ± 0.7**<0.01
234.6 ± 0.2
Hippocampus22.2 ± 0.4>0.05
72.2 ± 0.3*<0.05
234.1 ± 0.6
Thalamus23.8 ± 0.8>0.05
73.9 ± 0.6>0.05
234.4 ± 0.4
image

Figure 1. Changes in basement membrane protein levels in different brain regions with age. (A–P) Brain samples from the frontoparietal cortex (A–D), hippocampus (E–H), striatum (I–L), and thalamus (M–P) of 2-, 7-, and 23-month-old mice were processed for collagen IV (A, E, I, M), laminin (B, F, J, N), nidogen 2 (C, G, K, O), and fibronectin (D, H, L, P) levels by Western blotting. Collagen IV levels were significantly lower in the cortex and hippocampus of 23-month-old mice compared with 2- and 7-month-old animals, whereas levels were unchanged in the striatum and thalamus. Levels of laminin and nidogen 2 were also significantly reduced in the cortex and striatum of 7- and 23-month-old mice compared with 2-month-old mice. By contrast, levels of fibronectin were significant upregulated in all brain areas of aged mice. Bar charts with error bars represent mean optic density ratio of the protein level ± standard error of the mean (S.E.M.).

Download figure to PowerPoint

In the hippocampus, the levels of collagen IV were significantly decreased between 2- and 7-month-old mice (P < 0.05), and remained stable between 7 and 23 months of age (Fig. 1E). No significant differences were noted in the levels of laminin (Fig. 1F) or nidogen 2 (Fig. 1G) between any age groups. However, fibronectin levels were significantly upregulated in the brains of 23-month-old mice compared with both 2- and 7-month-old animals (P < 0.001 for both ages, Fig. 1H). Perlecan immunoreactivity was also higher in hippocampal blood vessels of 23-month-old mice vs. 7-month-old animals (P < 0.05, Table 2, Supporting Fig. 2D–F).

In the striatum, the quantity of collagen IV did not differ between the three age groups examined (Fig. 1I). Levels of laminin (Fig. 1J) and nidogen 2 (Fig. 1K) were significantly reduced in the 23-month-old mice compared with the 2- and 7-month-old mice (laminin P < 0.05 for both ages, nidogen 2 P < 0.001 and P < 0.05, respectively). Fibronectin levels were significantly increased in the brains of 23-month-old mice (P < 0.001, Fig. 1L). Immunocytochemistry results indicated that perlecan expression was higher in 23-month-old mice compared with 2- and 7-month-old mice (P < 0.01 for both ages, Table 2, Supporting Fig. 2G–I).

In the thalamus, measurement of the levels of collagen IV and laminin did not demonstrate differences between mice at any age (Fig. 1M and N). Nidogen 2 levels were significantly decreased in 7-month-old mice, compared with 2-month-old animals (P < 0.05), but remained stable thereafter (Fig. 1O). Fibronectin levels were increased in 23-month-old mice vs. both 2- and 7-month-old mice (P < 0.05 for both ages, Fig. 1P). Perlecan immunoreactivity also showed a trend toward higher expression in the aged mice, compared with young and adult mice (Table 2, Supporting Fig. 2J–L).

Differences in perivascular drainage of Aβ from the hippocampus vs. the thalamus

To determine if perivascular drainage differs between brain regions that are affected by or spared from CAA in young and old animals, fluorescently conjugated, soluble human Aβ40 was injected into the left thalamus or the left hippocampus of 7- and 20-month-old mice. The number of arteries containing Aβ in their walls 5 min after intracerebral injections was counted in the ipsi- and contralateral cortex, striatum, hippocampus, and thalamus of serial coronal brain sections at 200-μm intervals, up to 1000-μm anterior and posterior to the injection site.

Aβ was noted in arterial walls in all the brain regions examined following injection into either the hippocampus or thalamus. When Aβ was injected into the hippocampus of 7-month-old mice, there was strong overlap in the pattern of Aβ distribution within the ipsilateral cortex and striatum, which showed little variation in the number of Aβ-containing arteries across all distances (Fig. 2A). Similarly, the number of arteries containing Aβ in the ipsilateral thalamus was consistent across all distances (Fig. 2B). By contrast, the pattern of Aβ distribution in the ipsilateral hippocampus was varied, with the maximal number of hippocampal arteries containing Aβ observed at 600-μm anterior to the injection site (P < 0.05, ANOVA with Dunnett's Multiple Comparison, Fig. 2B). In the contralateral hemisphere, Aβ was also present in the walls of arteries in the hippocampus, cortex, thalamus, and striatum (Fig. 2E and F). Although the number of Aβ-containing arteries was similar between ipsi- and contralateral hemispheres, there was less overlap in the pattern of Aβ distribution between brain regions in the contralateral hemisphere, particularly in brain sections that were posterior to the site of injection (Fig. 2E and F).

image

Figure 2. Pattern of perivascular drainage of Aβ40 along arterial basement membranes following injection into the hippocampus or the thalamus of 7-month-old mice. (A, B, E, F) The mean number of arteries (± S.E.M.) per total tissue area (μm2) containing Aβ after injection into the hippocampus was quantified in the ipsilateral (A, B) and contralateral (E, F) cortex (A, E), striatum (A, E), hippocampus (B, F), and thalamus (B, F). In the ipsilateral hippocampus, the number of Aβ-containing arteries varied as a function of distance from the site of injection, with lowest numbers at the site of injection, and reaching a maximum at 600-μm anterior to the injection site. The number of Aβ-containing arteries was similar between ipsi- and contralateral hemispheres, although there was less overlap in the pattern of Aβ distribution between brain regions in the contralateral hemisphere. (C, D, G, H) The mean number of arteries (± S.E.M.) per total tissue area (μm2) containing Aβ after injection into the thalamus in the ipsilateral (C, D) and contralateral (G, H) cortex (C, G), striatum (C, G), hippocampus (D, H), and thalamus (D, H). Analysis of the pattern of Aβ drainage within the ipsilateral thalamus showed little variation in the number of Aβ-containing arteries at any distance from the injection site. The highest number of arteries with Aβ in their basement membranes following injection into the thalamus was noted in the hippocampus at 200-μm anterior to the injection site. In the contralateral hemisphere, there was a strong overlap in the pattern of Aβ distribution throughout the cortex, striatum, hippocampus, and thalamus.

Download figure to PowerPoint

Analysis of the pattern of drainage of Aβ following injection into the thalamus of 7-month-old mice showed an overlap between the ipsilateral cortex and striatum (Fig. 2C). No significant variation was noted in the number of Aβ-containing arteries in the thalamus at any distance from the injection site (Fig. 2D). Rather, the highest number of arteries with Aβ following injection into the thalamus was noted in the hippocampus at 200-μm anterior to the injection site (P < 0.05, two-tailed Student's t-test; Fig. 2D). Moreover, the peak number of arteries containing Aβ that was observed in the ipsilateral hippocampus following injection into the thalamus was equivalent to that observed after injection into the hippocampus itself (injection into thalamus = 0.013 ± 0.001 vs. injection into hippocampus = 0.012 ± 0.002, P > 0.05, two-tailed Student's t-test; Fig. 2B and D). In the contralateral hemisphere, the pattern of Aβ distribution throughout the cortex and striatum (Fig. 2G) was similar to that observed in the hippocampus and thalamus (Fig. 2H).

In the brains of 20-month-old mice that received an Aβ injection into the hippocampus, the pattern of Aβ distribution within the ipsilateral cortex and striatum was similar (Fig. 3A), with no statistical differences in the number of Aβ-containing arteries between 7- and 20-month-old mice in either area (Fig. 4A and B). The pattern of drainage of Aβ in the ipsilateral hippocampus and thalamus was similar to that observed in the younger mice (Fig. 3B). Analysis of Aβ distribution in the ipsilateral hippocampus in 20-month-old mice compared with 7-month-old mice showed a significant decrease in the number of Aβ-positive arteries in the aged animals at 600-μm anterior from the site of injection (7 months = 0.012 ± 0.002 vs. 20 months = 0.003 ± 0.001, P < 0.05, two-tailed Student's t-test; Fig. 4C). However, there was a strong degree of overlap in the pattern of Aβ distribution in the ipsilateral hippocampus between 7- and 20-month-old mice at and posterior to the site of injection (Fig. 4C). The number of Aβ-containing arteries was significantly decreased in the ipsilateral thalamus of 20-month-old mice compared with young mice (P < 0.05, two-tailed Student's t-test; Fig. 4D). In the contralateral hemisphere, the overall pattern of drainage remained similar between the cortex, striatum, and hippocampus (Fig. 3E and F), although there was a significant decrease in the number of Aβ-containing arteries in the cortex and striatum of aged mice (P < 0.05 for both areas, two-tailed Student's t-test; Fig. 4E–G). Aβ distribution in the thalamus of 20-month-old mice closely matched that seen in the 7-month-old mice (Fig. 4H).

image

Figure 3. Pattern of perivascular drainage of Aβ40 along arterial basement membranes following injection into the hippocampus or the thalamus of 20-month-old mice. (A, B, E, F) The mean number of arteries (± S.E.M.) per total tissue area (μm2) containing Aβ after injection into the hippocampus was quantified in the ipsilateral (A, B) and contralateral (E, F) cortex (A, E), striatum (A, E), hippocampus (B, F), and thalamus (B, F). The ipsilateral hippocampus remained the area with the highest arterial count, although there was overlap in the pattern of Aβ distribution within the ipsilateral cortex, striatum, thalamus, and hippocampus. A similar relationship was noted in the contralateral hemisphere. (C, D, G, H) The mean number of arteries (± S.E.M.) per total tissue area (μm2) containing Aβ after injection into the thalamus in the ipsilateral (C, D) and contralateral (G, H) cortex (C, G), striatum (C, G), hippocampus (D, H), and thalamus (D, H). Following intracerebral injections of Aβ into the thalamus, the number of arteries containing Aβ was similar in the ipsilateral cortex, striatum, and thalamus. The ipsilateral hippocampus contained the highest number of arteries with Aβ in their walls. In the contralateral hemisphere, there was a strong overlap in the pattern of Aβ drainage in the cortex and striatum, and a higher vessel count in the hippocampus than the thalamus.

Download figure to PowerPoint

image

Figure 4. Comparison between the pattern of perivascular drainage of Aβ40 along arterial basement membranes following injection into the hippocampus of 7-month-old mice vs. 20-month-old mice. (A, B, E, F) The mean number of arteries (± S.E.M.) per total tissue area (μm2) containing Aβ after injection into the hippocampus was quantified in the ipsilateral (A, B) and contralateral (E, F) cortex (A, E), striatum (A, E), hippocampus (B, F), and thalamus (B, F). There was a significant decrease in the number of Aβ-positive arteries in the ipsilateral hippocampus of 20-month-old mice compared with 7-month-old mice at 600-μm anterior from the site of injection (C). In the contralateral hemisphere, there was a significant decrease in the number of Aβ-containing arteries in the cortex and striatum of aged mice, whereas Aβ distribution in the thalamus of 20-month-old mice closely matched that seen in the 7-month-old mice.

Download figure to PowerPoint

Following intracerebral injections of Aβ into the thalamus of 20-month-old mice, the number of arteries containing Aβ was similar in the ipsilateral cortex and striatum (Fig. 3C). The number of arteries with Aβ in their walls in the ipsilateral hippocampus overlapped with the thalamus in distances posterior to the injection site (Fig. 3D). Overlay of the pattern of Aβ drainage following injection into the thalamus of young vs. aged mice, showed a similar pattern of distribution of Aβ in the ipsilateral cortex and striatum and a significant reduction in the number of Aβ-containing arteries in the cortex of 20- vs. 7-month-old mice (P < 0.01, two-tailed Student's t-test; Fig. 5A and B). By contrast, the same maximal number of Aβ-containing arteries was noted in the hippocampus of 7- and 20-month-old mice following injection into the thalamus (7 months = 0.013 ± 0.001 vs. 20 months = 0.012 ± 0.003, P > 0.05 two-tailed Student's t-test; Figs 3D and 5C). The same pattern of drainage and number of arteries was also observed between 7- and 20-month-old mice in the ipsilateral thalamus (P > 0.05 two-tailed Student's t-test; Figs 3D and 5D). In the contralateral hemisphere, there was a strong overlap in the pattern of Aβ drainage in the cortex and striatum (Fig. 3G), with no significant difference between young and aged mice (Fig. 5E and F). There was also a strong degree of overlap in the pattern and number of arteries containing Aβ in the hippocampus and thalamus of 7- and 20-month-old mice (Figs 3H and 5G and H).

image

Figure 5. Comparison between the pattern of perivascular drainage of Aβ40 along arterial basement membranes following injection into the thalamus of 7-month-old mice vs. 20-month-old mice. (A, B, E, F) The mean number of arteries (± S.E.M.) per total tissue area (μm2) containing Aβ after injection into the hippocampus was quantified in the ipsilateral (A, B) and contralateral (E, F) cortex (A, E), striatum (A, E), hippocampus (B, F), and thalamus (B, F). The same maximal number of Aβ-containing arteries was noted in the hippocampus of 7- and 20-month-old mice following injection into the thalamus (C). No difference was noted between the pattern of drainage and number of arteries between 7- and 20-month-old mice in the ipsilateral thalamus (D). In the contralateral hemisphere, there was a strong overlap in the pattern of Aβ drainage in all brain areas between 7- and 20-month-old mice.

Download figure to PowerPoint

To determine if differences in perivascular drainage of Aβ between the hippocampus and thalamus were due to regional and/or age variations in arterial vessel density, the density of vessels with diameters over 10 μm was quantified in the cortex, hippocampus, striatum, and thalamus of 7- and 20-month-old mice. No significant differences were noted in density of arteries between any brain region at any age (Supporting Fig. 3A and B).

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgment
  8. Author contributions
  9. References
  10. Supporting Information

In the aging and AD brain, CAA develops primarily in cortical and leptomeningeal blood vessels, whereas vessels in the deep gray matter are less affected. In this study of the mouse brain, we found that basement membrane thickness in capillaries of the cerebral cortex, hippocampus, and thalamus, but not in the striatum, increased with age. Examination of individual basement membrane protein levels revealed a significant decrease in collagen IV in the cortex and hippocampus of 23-month-old mice compared with young and adult animals, whereas levels were unchanged in the striatum and thalamus. Laminin and nidogen 2 levels were also significantly reduced in the cortex of 7- and 23-month-old mice compared with 2-month-old mice, but laminin was also reduced in the striatum. In contrast, levels of fibronectin and perlecan were significant upregulated in all brain areas of aged mice.

Analysis of perivascular drainage of Aβ40 from the hippocampus and thalamus of 7- and 20-month-old mice showed that the maximal number of Aβ-containing arteries was observed in anterior sections of the ipsilateral hippocampus, regardless of the site of injection. With age, the peak number of arteries containing Aβ was decreased following injection into the hippocampus, but not when injected into the thalamus. Collectively, these results suggest that the topographic distribution of CAA is due in part to regional differences in the efficiency of perivascular drainage along cerebrovascular basement membranes, and underlying variations in basement membrane protein expression with age.

Aging, the strongest risk factor for the development of CAA and AD, is associated with morphological and functional changes in the cerebral vasculature. Arteries become rigid, elongated, and tortuous, while vascular basement membranes show abnormal inclusions, splitting, duplication, and thickening (Wisniewski et al., 1992; Vinters et al., 1994; Kalaria, 2002; Hunter et al., 2012). Increased basement membrane thickness has been observed in brain capillaries of transgenic mice that overexpress TGF-β as early as 4 months of age, approximately 5 months before CAA is detected, suggesting that basement membrane thickening may contribute to Aβ deposition (Wyss-Coray et al., 2000). In this study, we found that capillary basement membrane thickness increased significantly with age in the frontoparietal cortex, hippocampus, and thalamus, but not in the striatum of wild-type mice. Given that CAA typically develops in blood vessels of the neocortex, hippocampus, and thalamus in mouse models of AD (Sturchler-Pierrat et al., 1997; Davis et al., 2004) and that capillary basement membrane thickening has been reported in the temporal gyrus, but not in the cerebellum, of AD patients (Zarow et al., 1997), our results suggest that regional changes in the ultrastructure of vascular basement membranes are associated with the development of CAA.

Laminin, nidogens, and collagen IV are vascular basement membrane proteins that inhibit aggregation of Aβ and destabilize preformed fibrils of Aβ (Bronfman et al., 1998). Vessels under 50 μm in diameter from AD patients show a decrease in the amount of collagen IV, compared with aged-matched controls (Christov et al., 2008). Heparan sulfate proteoglycans such as fibronectin and perlecan provide stability and flexibility to the basement membrane, accelerating the aggregation of Aβ by high-affinity interactions (Castillo et al., 1997; Cotman et al., 2000). In AD brains, the levels of heparan sulfate proteoglycans are increased compared with age-matched controls (Berzin et al., 2000; Shimizu et al., 2009). In this study in the mouse, we found that the levels of collagen IV was significantly decreased in the cortex and in the hippocampus, but not in the striatum and thalamus of aged mice. Laminin was reduced in the cortex and striatum; nidogen 2 was reduced in the cortex thalamus and striatum. By contrast, fibronectin and perlecan were significantly increased with age in blood vessels in the cortex, striatum, and hippocampus. These findings support the suggestion that with increasing age, there may be a relative decrease in the ratio of individual basement membrane proteins that inhibit and promote Aβ aggregation, leading to a failure of perivascular drainage of Aβ and other solutes which accumulate in the cerebrovascular basement membranes (Revesz et al., 2003; Thal et al., 2007; Weller et al., 2008).

Aβ contained within interstitial fluid enters capillary basement membranes, drains into the basement membranes of cerebral arterioles and arteries, reaching leptomeningeal arteries, and out of the base of the skull to cervical lymph nodes (Carare et al., 2008; Ball et al., 2010). Increasing age leads to arteriosclerosis, diminishing the driving force for perivascular clearance, and resulting in the development of CAA (Schley et al., 2006). In this study, we found that intracerebrally injected human Aβ40 was visible within the basement membranes of arteries in the cortex, hippocampus, striatum, and thalamus within 5 min of injection in the hippocampus or thalamus of 7- and 20-month-old mice. Detection of Aβ at 1000 μm from the injection site in both ipsi- and contralateral brain regions suggests that Aβ distribution is mediated by a combination of diffusion and perivascular drainage along arterial basement membranes and white matter tracts (Cserr & Ostrach, 1974; Carare et al., 2008).

Our results show that following injection into either the hippocampus or the thalamus of the 7-month-old mice, the highest number of arteries with Aβ in their basement membranes was observed in the ipsilateral hippocampus. This suggests not only that perivascular drainage differs between brain regions under normal physiologic conditions, but also that perivascular drainage of Aβ from the thalamus is directed predominantly toward the hippocampus. This is unlikely to be due to diffusion of Aβ into anatomical regions of close physical proximity to the thalamus because (i) there was little variation in the number of labeled striatal arteries following injection of Aβ into the thalamus and (ii) diffusion through the extracellular spaces from the thalamus into the hippocampus would be hindered by the presence of the hippocampal fissure. In mice and humans, blood supply to the lateral aspects of the thalamus is provided by the ventral and dorsolateral thalamic arteries, which are derived from the internal carotid artery (Schmahmann, 2003; Dorr et al., 2007). The ventrolateral thalamus is also supplied by the thalamoperforating artery, which is a branch of the posterior cerebral artery (Supporting Fig. 4). The posterior cerebral artery also gives off the longitudinal hippocampal artery, which follows the length of the hippocampus and gives rise to the transverse hippocampal arteries (Erdem et al., 1993; Dorr et al., 2007) (Supporting Fig. 4). This anatomical connection between the thalamus and hippocampus of the mouse and human brains provides a route by which solutes may drain quickly from the thalamus into the hippocampus. Moreover, drainage of the majority of solutes from the hippocampus may be restricted to the longitudinal hippocampal artery, whereas ISF from the thalamus may clear more efficiently along multiple different arteries from anterior and posterior circulations. Anastomoses between thalamoperforating arteries in the two brain hemispheres may also provide a rapid route for solutes to reach the contralateral side (Dorr et al., 2007).

We have previously shown that perivascular drainage from the hippocampus is less efficient in the aged mouse brain, resulting in retention of solutes in capillaries (Hawkes et al., 2011). Our current data strengthen this finding, as we observed that the number of arteries labeled with Aβ was significantly lower in the injected hippocampus of old mice compared with young mice. In contrast, the same number of Aβ-containing arteries was observed in the ipsilateral hippocampus of both adult and old mice after injection into the thalamus, indicating that age did not affect perivascular drainage of Aβ from the thalamus. Age-related failure of perivascular drainage from the hippocampus may also be exacerbated by the continuous flow of solutes from the thalamus. Furthermore, the presence of an expandable perivascular space around thalamic arteries may help to counteract age-related decreases in the pulsatile forces that drive perivascular drainage and allow for greater exposure to perivascular macrophages that reside within perivascular spaces and mediate CAA severity (Zhang et al., 1990; Pollock et al., 1997; Bechmann et al., 2001; Hawkes & McLaurin, 2009).

Taken together, our results suggest that regional differences in perivascular solute drainage and the accumulation of Aβ as CAA are due to a combination of biochemical and anatomical factors. Increased basement membrane thickness and decreased expression of anti-amyloidogenic basement membrane proteins may increase the vulnerability of certain brain areas to the development of CAA, particular in areas where Aβ production and/or levels are high, such as the cerebral cortex and hippocampus (McLean et al., 1999; Yasojima et al., 2001; Lehman et al., 2003; Lopresti et al., 2005; Ma et al., 2011). This may also be exacerbated in regions that have a restricted vascular supply and where arteries are not invested with an expandable perivascular space. These data shed important light on the mechanisms that underlie the topographic distribution of CAA in the aging and AD brain that may be useful for the development of novel therapeutics.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgment
  8. Author contributions
  9. References
  10. Supporting Information

Animals

Male and female C57BL/6 mice were kept on a standard 12-h light/dark cycle and allowed food and water ad libitum. Young (2 months old) and adult (7 months old) mice were obtained from the Biomedical Research Facility (University of Southampton), whereas aged mice (20–23 months old) were obtained from the Centre for Comparative Biology, Newcastle University and Harlan Labs UK (Bicester, Oxforshire, UK). All experiments were performed in accordance with animal care guidelines stipulated by the Animal Care and Use Committee at the University of Southampton and Newcastle University.

Transmission electron microscopy

2-, 7-, and 23-month-old mice (n = 3/group) were deeply anesthetized with sodium pentobarbital and perfused intracardially with 0.01 m phosphate-buffered saline (PBS) followed by a formaldehyde-glutaraldehyde-picric-acid fixative. Brains were removed, postfixed for 2 weeks, and microdissected for frontoparietal cortex, hippocampus, striatum, and thalamus. Brain areas were processed according to Šišková et al. (Siskova et al., 2009). Tissue blocks were trimmed and semithin 0.5 μm sections were cut and stained in 1% v/v toluidine blue in 1% w/v borax. These sections were used as a guide to cut ultrathin 80-nm sections of areas containing capillaries. The ultrathin sections were collected onto copper grids, stained in Reynolds lead stain, and viewed under a Hitachi H7000 transmission electron microscope. Three capillaries in transverse section with visible and well-defined basement membranes from each brain area were digitally photographed at a magnification of ×9000 using iTEM software (Universal TEM Imaging Platform, Soft Imaging System, Münster, Germany) operating a MegaView III digital camera (Soft Imaging System, Münster, Germany). Basement membranes were then photographed at a magnification of ×50 000 to provide two higher powered images of each capillary.

Western blotting

2-, 7-, and 20-month-old mice (n = 4/group) were perfused with PBS, brains rapidly removed and dissected for frontoparietal cortex, hippocampus, striatum, and thalamus. Brain tissues were sonicated in Ripa lysis buffer [20 mm Tris-HCl (pH 8.0), 150 mm NaCl, 1 mm EDTA, 0.1% SDS, 1% Igepal, 50 mm NaF, 1 mm NaVO3] containing a protease inhibitor cocktail (Merck, Nottingham, UK), spun down (13 000 rpm, 10 min, 4°C), and supernatants collected, aliquoted, and frozen at −80°C until further use. Proteins (15–45 μg) were separated by gel electrophoresis on 3–8% Tris-acetate gels (Invitrogen, Paisley, UK) and transferred onto a nitrocellulose membrane. For laminin, fibronectin, and nidogen-2 blots, samples were diluted in NuPAGE® LDS sample buffer (Invitrogen, Paisley, UK) containing 2.5% β-mercaptoethanol and heated at 75°C for 5 min. Collagen IV samples were prepared without β-mercaptoethanol or heat. Membranes were incubated overnight at 4°C with anti-collagen IV (1:500), anti-laminin (1:500), anti-nidogen 2 (1:1000), or anti-fibronectin (1:2000; AbD Serotec, Kidlington, UK) antibodies. Membranes were stripped and reproved with anti-glyceraldehyde-3-PDH (GAPDH) antibody (1:50 000; Sigma-Aldrich, Gillingham, Dorset, UK) to ensure equal protein loading. Immunoblots were quantified by densitometry using Image J software (NIH, Maryland, USA) and calculated as an optical density ratio of protein levels normalized to GAPDH levels.

Enzyme-linked immunocytochemistry

2-, 7-, and 23-month-old mice (n = 4/group) were perfused with 0.01 m PBS, brains rapidly removed, and snap frozen. Tissue sections (20-μm thickness) were fixed in 4% paraformaldehyde, washed with PBS, blocked with 3% H2O2 and 15% goat serum, and incubated overnight with anti-perlecan (1:500; Millipore, Watford, UK) or biotinylated solanum tuberosum (Potato) lectin (1:250, Vector Labs, Peterborough, UK). Sections were washed with PBS, incubated with anti-rat horseradish peroxidase conjugates (1:400; Vector Labs) and developed with 3,3′-diaminobenzidine as chromogen.

Preparation of Aβ peptide

HiLyte Fluor™ TR-labeled human Aβ40 (100 mg, Cambridge Biosciences, UK) was solubilized in 50 μL 1% NH4OH, vortexed for 30 s, and made up to 100 μm with ice-cold sterile PBS. Aβ was aliquoted, immediately frozen on dry ice, and kept at −80°C until use.

Intracerebral injections

7- and 20-month-old mice (n = 4/group) were injected stereotaxically with 0.5 μL ice-cold Aβ40 into the left hippocampus (coordinates from Bregma: AP = −1.9 mm; ML = 1.5 mm; and DV = 1.7 mm) or left thalamus (coordinates from Bregma: AP = −1.3 mm; ML = 1.0 mm; and DV = 3.3 mm). Injection pipettes were left in situ for 2 min to prevent reflux and mice were sacrificed 5-min postinjection. Mice were intracardially perfused with phosphate-buffered saline (PBS, pH 7.4), followed by 4% paraformaldehyde, and brains were processed for double-labeling immunocytochemistry. To assess the pattern of Aβ40 drainage following injection into the left hippocampus or thalamus, the number of arteries and arterioles containing Aβ in the frontoparietal cortex, striatum, hippocampus, and thalamus was counted manually. Quantification was carried out in two pictures/region at 200-μm intervals anterior and posterior to the site of injection (up to 1000 μm away from the injection site), in both the ipsi- and contralateral hemispheres and divided by tissue area per picture.

Double-labeling immunocytochemistry

Brain sections (20-μm thickness) were incubated overnight with anti-laminin (1:500) and anti-α smooth muscle actin (1:500; Sigma-Aldrich), washed in PBS and developed with AlexaFluor 488-conjugated anti-rabbit and AlexaFluor 633-conjugated anti-mouse (1:200; Invitrogen, Paisley, UK). Photomicrographs were captured using a Leica SP5 confocal laser scanning microscope (Milton Keys, UK) and exported to Photoshop CS software.

Statistical analysis

All mice were coded, and measurements were carried out in a blinded fashion. Basement membrane thickness – Capillary thickness was quantified using iTEM software. Perpendicular measurement bars were used to measure the distance between the inside and outside edge of the basement membrane. This was repeated ten times for each high-powered micrograph for each capillary. The measurements expressing the thickness of the basement membranes of capillaries were averaged per mouse and analyzed using one-way ANOVA test, with Bonferroni correction for multiple comparisons (significance set at P < 0.05). Western Blots – Mean ± S.E.M. densitometry values were compiled from two blots per antibody and analyzed using one-way ANOVA with Newman–Keuls correction for multiple comparisons (significance set at P < 0.05). Arterial density and perlecan immunocytochemistry – Micrographs were converted into binary images (four sections/mouse), evaluated by densitometry using Image J software and were compared between age groups using one-way ANOVA test with Newman–Keuls correction (significance set at P < 0.05). Regional differences in perivascular drainage – Mean ± S.E.M. of the number of arteries containing Aβ at each distance away from the single injection site were analyzed using one-way ANOVA with Dunnett's Multiple Comparisons test. Other analyses were carried out using two-tailed Student's t-test with Bonferroni correction (significance set at P < 0.05). Analyses were carried using SPSS (v. 19) and GraphPad Prizm software.

Acknowledgment

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgment
  8. Author contributions
  9. References
  10. Supporting Information

The authors wish to thank Dr Anton Page for excellent assistance in electron microscopy.

Author contributions

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgment
  8. Author contributions
  9. References
  10. Supporting Information

C. H. performed experimental design, tissue collection, Western blotting, intracerebral injections, data analysis, and manuscript preparation. M. G. carried out histology and data analysis. M. S. performed electron microscopy experiments and quantification. A. D. prepared vascular anatomy diagram. H. M. Y. performed statistical analyses. R. K. provided aged mice and data analysis. R. W. and R. C. contributed to experimental design, data analysis, and manuscript preparation.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgment
  8. Author contributions
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgment
  8. Author contributions
  9. References
  10. Supporting Information
FilenameFormatSizeDescription
acel12045-sup-0001-FigS1.tifimage/tif4767KFig. 1 Regional differences in basement membrane thickness with age. (A–F) Electron micrographs of basement membranes (arrows) of capillaries from the brains of 2- (A and D), 7- (B and E), and 23- (C and F)-month-old mice did not show differences in thickness in the striatum with increasing age (A–C). By contrast, there was a significant increase in basement membrane thickness with age in capillaries in the frontoparietal cortex (D–F). Original magnification = ×50 000.
acel12045-sup-0002-FigS2.tifimage/tif5343KFig. 2 Increased perlecan expression with age. (A–L) Photomicrographs of perlecan immunoreactivity in blood vessels in the cortex (A–C), hippocampus (D–F), striatum (G–I), and thalamus (J–L) of 2- (A, D, G, J), 7- (B, E, H, K), and 23- (C, F, I, L)-month-old mice.
acel12045-sup-0003-FigS3.tifimage/tif398KFig. 3 Arterial blood vessel density by brain region and age. (A and B) Mean (± S.E.M.) density of potato lectin staining of arteries in the brains of 7- (A) and 20-month-old (B) mice. No differences were noted in percent area coverage of lectin between the cortex, striatum, hippocampus, or thalamus in either young or aged mice.
acel12045-sup-0004-FigS4.tifimage/tif2398KFig. 4 Vascular anatomy of the thalamus and hippocampus of the adult mouse brain. (A and B) Anterior (A) and lateral view (B) showing the posterior cerebral artery blood supply (purple) to the thalamus (green) and the hippocampus (blue) in the mouse. Yellow arrows indicate the flow of solutes moving from the thalamus to the hippocampus. Vasculature was imaged with micro-CT and registered to MRI of brain structures. Posterior cerebral artery, thalamus, and hippocampus segmentations and 3D images made using Display software (Montreal Neurological Institute Brain Imaging Centre, Montreal, Canada).

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.