Deficiency in Mural Vascular Cells Coincides with Blood–Brain Barrier Disruption in Alzheimer's Disease

Human subjects

Written consent was obtained and approved by the University of Rochester Medical Center for all human subjects utilized in the study prior to death. The post‐mortem interval ranged between 4 and 16 h. Post‐mortem brain tissue samples, including frontal cortex (Brodmann area 9/10) and hippocampus (CA1 subfield), were obtained from subjects with a definite diagnosis of AD confirmed by neuropathological analysis [Braak stage III‐IV or V‐VI 8; CERAD, Consortium to Establish a Registry for Alzheimer's Disease 26—definite frequent] and neurologically intact controls with no AD pathology (Braak stage 0; CERAD—negative). Vascular risk factors, such as atherosclerosis, hypertension and/or myocardial infarction, were present in four out of six AD patients and six out of six control patients. The cause of death in all AD and control patients was either respiratory failure or cardiac failure (see Table 1 for details). Although the average age of death of neurologically intact controls was somewhat lower compared with AD cases, the difference was not statistically significant. Moreover, there was no significant difference in any of the studied parameters (ie, pericyte number and coverage, IgG and fibrin extravascular levels) between somewhat younger compared with older neurologically intact controls.

Immunofluorescence analysis

For all analyses, paraformaldehyde‐fixed, paraffin‐embedded human AD or control brain tissue samples were used. All tissue was cut to a thickness of 6 μm. Sections were deparaffinized with xylene and rehydrated to distilled water after serial ethanol washes. Subsequently, sections were incubated in 1:10 diluted target antigen retrieval solution, pH 9 (Dako, Carpinteria, CA, USA) for 20 minutes in a microwave. The tissue sections were blocked in 10% normal swine serum (Vector Laboratories, Burlingame, CA, USA) containing 0.05% Triton X‐100 (Sigma‐Aldrich, St. Louis, MO, USA) and then incubated with the following primary antibodies overnight at 4°C: goat anti‐human PDGFRβ (1:100, R&D Systems, Minneapolis, MN, USA), mouse anti‐human aminopeptidase N (CD13) (1:100, R&D Systems), goat anti‐human immunoglobulin G (IgG) (1:50, R&D Systems), rabbit anti‐human fibrinogen (1:500, Dako) and rabbit anti‐human Aβ (1:200, Cell Signaling, Boston, MA, USA). To visualize pericytes, sections were incubated in Alexa Fluor 488‐conjugated bovine anti‐goat (1:100, Jackson ImmunoResearch, West Grove, PA, USA) or Alexa Fluor 488‐conjugated donkey anti‐mouse secondary antibodies (1:100, Jackson ImmunoResearch) to detect PDGFRβ‐ or CD13‐positive pericytes, respectively. To visualize IgG, fibrin and Aβ sections were incubated in Cy3‐conjugated bovine anti‐goat (1:100, Jackson ImmunoResearch), Cy3‐conjugated donkey anti‐rabbit (1:100, Jackson ImmunoResearch) and Cy3‐conjugated donkey anti‐rabbit secondary antibody (Jackson ImmunoResearch) secondary antibodies, respectively. Blood vessels were stained by incubating sections in biotinylated ulex europaeus agglutinin I (1:100, Vector Laboratories) overnight at 4°C during the primary antibody incubation step, followed by incubating in Dylight 649‐conjugated streptavidin (1:100, Vector Laboratories) during the secondary antibody incubation step.

Tissue sections were mounted and coverslipped using fluorescent mounting media (Dako). All images were taken using a custom‐built Zeiss 510 meta‐confocal laser scanning microscope with a Zeiss Apochromat 259/0.8 NA water immersion objective (Carl Zeiss MicroImaging Inc., Thornwood, NY, USA). A 488‐nm argon laser was used to excite Alexa Fluor 488 and the emission was collected through a 500‐ to 550‐nm band pass (bp) filter. A 543‐nm HeNe laser was used to excite Cy3 and the emission was collected through a 560‐ to 615‐nm bp filter. A 633‐nm HeNe laser was used to excite DyLight 649 and the emission was collected through a 650‐ to 700‐nm bp filter.

Image analysis

All images were analyzed using NIH‐developed ImageJ (Bethesda, MD, USA). A field size of 420 × 420 μm and a maximum z‐projection of 5 μm were utilized for all images. For each analysis, five images were randomly taken from three nonadjacent tissue sections per specimen (n = 6/group). The analysis of pericyte number and coverage on brain capillaries was performed using double immunostaining for PDGFRβ and endothelial‐specific lectin, and/or CD13 and endothelial‐specific lectin, as previously described 1, 4, 34, 36, 37. In addition to expression in brain capillary pericytes, PDGFRβ and CD13 can also be expressed in other mural cell types such as vascular smooth muscle cells of small cerebral microvessels that are larger in diameter than brain capillaries 35. Therefore, the present quantification analysis of pericyte number and coverage was restricted to PDGFRβ and CD13‐positive mural cells covering exclusively brain capillaries, which, by definition, are pericytes, not vascular smooth muscle cells 35. In this study, brain capillaries were defined as microvessels <6 μm in diameter, as in previous studies on pericyte coverage of brain capillaries 1, 4, 36, 37. Briefly, PDGFRβ‐ and CD13‐positive area was measured using the ImageJ Area tool and divided by the total lectin‐positive brain capillary area to represent percent pericyte coverage of the capillary wall. Pericyte number was determined by manually counting the PDGFRβ‐positive cell bodies per image and was represented as number of pericytes per mm² of lectin‐positive capillary endothelial cell area 1, 4, 36, 37. To quantify capillary leakage, the levels of extravascular fibrin and IgG were measured as previously described 4, 5, 36, 37. In short, the ImageJ Area tool was used to measure the total area of fibrin‐ and IgG‐positive signal. Any fibrin or IgG signal that co‐localized with the lectin‐positive signal was subtracted from the total area of leakage, yielding a value representing extravascular levels of each plasma‐derived protein. Post‐image thresholding was utilized to account for variations in background signal. All images were analyzed by a blinded investigator.

Statistical analysis

The Student's t‐test was used to analyze differences between AD and control groups. Correlations were determined using Pearson's correlation analysis and fit to a linear regression. A P‐value <0.05 was considered statistically significant in all experiments. All values are expressed as a mean ± standard error of the mean.

Pericyte number and coverage are reduced in AD

Figure 1 illustrates the relationship between glial‐fibrillar acidic protein (GFAP)‐positive astrocytes (green), PDGFRβ‐positive pericytes (red) and lectin‐positive brain endothelial cells (blue) within the neurovascular unit in the human cortex. As shown in the inset (Figure 1), pericyte processes sheathe most of the outer side of the brain capillary endothelial cells and astrocytic end foot processes unsheathe the microvessel wall, which is made of endothelial cells and pericytes.

Figure 2 shows that mural cells positive for PDGFRβ or CD13 are substantially reduced in AD cases compared with neurologically intact controls. Using double PDGFRβ and lectin immunostaining restricted to brain capillaries <6 μm in diameter 1, 5, 34, 35, 37, we found 59% and 60% decrease in pericyte numbers (P < 0.01) and 29% and 30% (P < 0.01) decrease in PDGFRβ‐positive pericyte coverage of capillary microvessels in human AD cortex and hippocampus, respectively, compared with neurologically intact controls (Figure 2A–C). To confirm these results, we next utilized another marker that recognizes pericytes, that is, CD13 2, 35, 36, which indicated independently a similar 32% and 33% (P < 0.01) reduction in pericyte coverage of brain capillaries <6 μm in diameter in AD cortex and hippocampus compared with controls, respectively (Figure 2D,E).

Increased extravascular IgG deposition correlates with pericyte reduction in AD

To determine the extent of BBB breakdown, first we studied extravascular deposition of serum IgG in the cortex and hippocampus of AD cases and controls. We observed increased levels of extravascular IgG in AD individuals by 2.3‐ and 2.1‐fold (P < 0.01) in the cortex and hippocampus, respectively, compared with the low background levels detected in neurologically intact controls (Figure 3A,B). Moreover, Pearson's correlation analysis indicated that extravascular IgG accumulation in the cortex and hippocampus strongly correlated with a deficiency in pericyte coverage of brain capillaries in AD cases compared with controls, that is, the Pearson's correlation coefficient r‐values were −0.96 (P < 0.01) and −0.86 (P < 0.01) for the cortex and hippocampus, respectively (Figure 3C,D).

Increased extravascular fibrin deposition correlates with pericyte reduction in AD

To confirm that BBB breakdown in AD is not molecule specific, we next studied extravascular accumulation of fibrin in AD cases and controls. Extravascular fibrin levels were increased by 2.2‐ and 1.9‐fold (P < 0.01) in human AD cortex and hippocampus, respectively, compared with the low background levels determined in controls (Figure 4A,B). As for the IgG, there was a significant negative correlation between extravascular fibrin levels and pericyte coverage of brain capillaries in AD cases compared with controls, as indicated by the Pearson's coefficient r‐values of −0.81 (P < 0.01) and −0.98 (P < 0.01) in the cortex and hippocampus, respectively (Figure 4C,D).

Increased Aβ deposition correlates with pericyte reduction in AD

We performed Aβ immunostaining in the hippocampus and observed an increase in Aβ load in AD individuals by approximately 4.2‐fold (P < 0.01) compared with barely detectable background levels of Aβ in neurologically intact controls (Figure 5A,B). The Pearson's correlation analysis indicated that Aβ accumulation in the hippocampus correlated with a deficiency in pericyte coverage of brain capillaries in AD cases compared with controls, that is, the Pearson's correlation coefficient r‐value was −0.80 (P < 0.01) (Figure 5C). There was also an overlap between Aβ immunostaining and IgG immunostaining (Figure 5D).