Human cerebral collateral arteriole function in subjects with normal cognition, mild cognitive impairment, and dementia

Abstract

Clinical and preclinical studies have suggested a link between cardiovascular disease and dementia disorders, but the role of the collateral brain circulation in cognitive dysfunction remains unknown. We aimed to test the hypothesis that leptomeningeal arteriole (LMA) function and response to metabolic stressors differ among subjects with dementia, mild cognitive impairment (MCI), and normal cognition (CN). After rapid autopsy, LMAs were isolated from subjects with CN (n = 10), MCI (n = 12), or dementia [n = 42, Alzheimer’s disease (AD), vascular dementia (VaD), or other dementia], and endothelial and smooth muscle-dependent function were measured at baseline and after exposure to β-amyloid (2 μM), palmitic acid (150 μM), or medin (5 μM) and compared. There were no differences among the groups in baseline endothelial function (maximum dilation to acetylcholine, CN: 74.1 ± 9.7%, MCI: 67.1 ± 4.8%, AD: 74.7 ± 2.8%, VaD: 72.0 ± 5.3%, and other dementia: 68.0 ± 8.0%) and smooth muscle-dependent function (CN: 93.4 ± 3.0%, MCI: 83.3 ± 4.1%, AD: 91.8 ± 1.7%, VaD: 91.7 ± 2.4%, and other dementia: 87.9 ± 4.9%). There was no correlation between last cognitive function score and baseline endothelial or smooth muscle-dependent function. LMA endothelial function and, to a lesser extent, smooth muscle-dependent function were impaired posttreatment with β-amyloid, palmitic acid, and medin. Posttreatment LMA responses were not different between subjects with CN/MCI vs. dementia. Baseline responses and impaired vasoreactivity after treatment with metabolic stressors did not differ among subjects with CN, MCI, and dementia. The results suggest that the cognitive dysfunction in dementia disorders is not attributable to differences in baseline brain collateral circulation function but may be influenced by exposure of the vasculature to metabolic stressors.

NEW & NOTEWORTHY Here, we present novel findings that brain collateral arteriole function did not differ among subjects with normal cognition, mild cognitive impairment, and dementia (Alzheimer’s disease and vascular dementia). Although arteriole function was impaired by vascular stressors (β-amyloid, palmitic acid, and medin), responses did not differ between those with or without dementia. The cognitive dysfunction in dementia disorders is not attributable to differences in baseline brain collateral circulation function but may be influenced by vascular exposure to metabolic stressors.

INTRODUCTION

Clinical and preclinical data show that vascular disease and cardiovascular risk factors are associated with Alzheimer’s disease (AD) and AD-related dementia disorders (2, 7). Impaired cerebrovascular hemodynamics found in patients with early AD (5), and pathological vascular changes suggesting altered arteriolar vasoregulation (11) and hemodynamic disturbances (13) point to the critical role of vascular dysfunction in disease pathogenesis. Leptomeningeal collateral arterioles (LMA) link the three major arterial territories over the brain surface. Although it is known that LMAs play a significant role in cerebrovascular autoregulation and represent an important compensatory mechanism in cerebrovascular disease and affect prognosis after stroke (3, 16), the role of LMA function in the pathophysiology of dementia disorders remains poorly understood despite pathological observations that LMAs are widely affected in AD (23). Additionally, vascular dysfunction may result from exposure to metabolic stressors such as fatty acids or amyloidogenic proteins such as β-amyloid (Aβ) or medin (aging-associated vascular amyloid protein) (15, 17, 21, 25, 26), but the differential susceptibility of the brain collateral circulation between patients with dementia disorders and those without remains unknown. Using a unique and novel model using ex vivo human LMAs isolated after rapid autopsy of brain donors (25), the present study aimed to test the hypothesis that LMA endothelium-dependent function and smooth muscle-dependent function are impaired in patients with dementia [AD, vascular dementia (VaD), or other causes] compared with patients with normal cognition (CN) or mild cognitive impairment (MCI). We also aimed to test the hypothesis that LMAs from patients with dementia have a worse response when exposed to vascular metabolic stressors [Aβ, palmitic acid (PA), or medin] compared with LMAs from those with CN/MCI.

METHODS

Consecutive brain donor LMAs were isolated from 64 cadavers (88.7 ± 1.2 yr, 32 men and 32 women) after rapid autopsy (postmortem interval, or the time elapsed from official declaration of death to brain removal, 2.8 ± 0.1 h) who, before death, provided informed consent for brain donation (program information at https://www.brainandbodydonationregistration.org/) (1). Program operations were approved by the Banner-Sun Health Research Institute Institutional Review Board, and the study was approved by the Phoenix Veterans Affairs Institutional Review Board.

Each participant in the Banner-Sun Health Research Institute Brain and Body Donation Program undergoes general neurological examinations and neurological functional battery tests annually, and clinical diagnostic classification (CN, MCI, and specific dementia or neurological disorders) is performed after each annual assessment at a consensus conference attended by neurologists, psychiatrists, and neuropsychologists (1). Final clinicopathological diagnoses are assigned after death after review of all standardized clinical data, the most recent private medical records, and neuropathological examination findings. All subjects with CN (n = 10) and MCI (n = 12) did not meet criteria for dementia at the time of death and were without a major neuropathological diagnosis. AD (n = 34) was defined as at least intermediate or high National Institute on Aging-Reagan criteria (12). VaD cases (n = 13) were classified using modified National Institute of Neurological Disorders and Stroke-Association Internationale pour la Recherche et l’Enseignement Neurosciences criteria (24). All subjects had final pathological diagnoses. In 10 donors, AD and VaD were both present. Donors with dementia whose final pathological diagnoses were not AD or VaD were grouped under other dementia (n = 5: 2 donors with Parkinson’s disease, 2 donors with dementia with Lewy bodies, and 1 donor with Pick’s disease). Donors were also classified as to whether cerebral amyloid angiopathy (CAA) was present (n = 31) or not (n = 14). Clinical and pathological data are shown in Table 1.

Table 1.

Demographics, pathology, and leptomeningeal arteriole tone

	CN	MCI	ADa	VaDa	Other Dementia	P Value (CN/MCI/AD and Other Dementia)	P Value (CN/MCI/VaD and Other Dementia)
Demographics
n	10	12	34	13	5
Age, yr	90.6 ± 2.2	93.3 ± 2.1	88.4 ± 1.7	94.1 ± 1.5	75.8 ± 2.7	0.04b	<0.001b
Women/men, n (%)	3/7 (30)	5/7 (42)	19/15 (56)	5/8 (62)	3/2 (60)	NS	NS
Postmortem interval, h	3.1 ± 0.2	2.9 ± 0.2	2.7 ± 0.1	2.8 ± 0.1	3.4 ± 0.1	NS	NS
Last mini-mental state examination score	28.4 ± 0.4	25.2 ± 0.9	16.5 ± 1.4	21.1 ± 1.4	14.2 ± 5.6	<0.001c	<0.001c
Coronary artery disease/peripheral arterial disease, n (%)	5 (56)	9 (75)	12 (39)	7 (70)	2 (50)	NS	NS
Hypertension, n (%)	6 (67)	11 (92)	19 (63)	6 (60)	1 (25)	NS	NS
Diabetes mellitus, n (%)	3 (33)	5 (42)	4 (13)	2 (20)	1 (25)	NS	NS
Hyperlipidemia, n (%)	7 (78)	9 (75)	22 (71)	6 (60)	1 (25)	NS	NS
Pathology
Brain weight, g	1,097 ± 32	1,041 ± 95	1,125 ± 26	1,062 ± 34	1,101 ± 72	NS	NS
Brain total plaqued	8.4 ± 1.8	7.8 ± 1.7	11.7 ± 0.7	9.9 ± 1.7	1.3 ± 0.6	0.001e	NS
CAA present, n (%)	6 (67)	4 (80)	21 (84)	6 (67)	0 (0)	0.01f	NS
CAA scoreg	2.6 ± 0.9	2.8 ± 1.2	3.7 ± 0.7	2.9 ± 1.2	0 ± 0	NS	NS
Leptomeningeal arteriole tone
Diameter, μm
At no pressure	129 ± 7	149 ± 13	139 ± 6	138 ± 12	183 ± 20	NS	NS
At 30 mmHg	147 ± 13	167 ± 12	149 ± 6	144 ± 10	210 ± 12	0.03h	0.04h
At 60 mmHg	142 ± 11	168 ± 11	152 ± 6	155 ± 11	211 ± 9	0.01i	0.02i
Maximum diameter, μm	163 ± 14	199 ± 13	166 ± 7	166 ± 12	224 ± 10	0.005j	0.02j
Endothelin-1 dose used, × 10−9 M	9.7 ± 2.5	7.0 ± 2.3	8.5 ± 21.1	7.4 ± 1.3	12.5 ± 9.5	NS	NS

Values are means ± SE; n, number of subjects/group. CN, normal cognition; MCI, mild cognitive impairment; AD, Alzheimer’s disease; VaD, vascular dementia; CAA, cerebral amyloid angiopathy; NS, not significant.

^an = 10 were diagnosed with both AD and VaD;

^bOther dementia age was significantly less than CN, MCI, AD, or VaD by pairwise analyses;

^cCN was significantly higher than AD, VaD, or other dementia by pairwise analyses;

^dbrain total plaque: senile amyloid plaque density scored according to the Consortium to Establish a Registry for Alzheimer's Disease templates, with the highest possible total score being 15 (7);

^eAD was significantly higher than other dementia by pairwise analysis;

^fproportion of CAA present in AD was significantly higher than other dementia by Fisher’s exact test;

^gdensity of amyloidotic blood vessels (scored as 0 = none, 1 = sparse, 2 = moderate, and 3 = frequent) summed from frontal, temporal, parietal, and occipital regions (7);

^hOther dementia diameter was significantly larger than AD or VaD by pairwise analyses;

ⁱOther dementia diameter was significantly larger than CN or AD by pairwise analyses;

^jOther dementia diameter was significantly larger than CN, AD, or VaD by pairwise analyses; no imputation of missing data was performed.

Methodological details on vasoreactivity procedures have been previously reported (17, 18). In brief, LMAs were cannulated, pressurized in sequence to 30 mmHg of pressure and then to physiological 60 mmHg of pressure (30-min stabilization), and preconstricted to ~60% of maximum diameter with endothelin-1 (2–50 × 10⁻⁹ M). Baseline dilator responses to acetylcholine (10⁻⁹−10⁻⁴ M to measure endothelium-dependent function) and papaverine or diethylenetriamine NONOate (10⁻⁴ M to measure smooth muscle-dependent function) were measured by videomicroscopy, and LMA arteriole responses averaged for each donor. Endothelium-dependent function [maximum acetylcholine dilation and acetylcholine half-maximal effective concentration (EC₅₀)] (17, 18) and smooth muscle-dependent function were compared among 1) subjects with CN, MCI, AD, and other dementia and 2) subjects with CN, MCI, VaD, and other dementia (separate analyses were performed due to overlap in subjects having both AD and VaD). LMA function was also compared between donors with CAA versus those without CAA.

After washout, some LMAs were exposed for 1 h to one of the following: Aβ42 (2 µM, Anaspec, Fremont CA), PA (150 µM, the most common saturated fatty acid in the Western diet), or medin (5 µM, recombinantly produced as previously described) (17), and a second measurement of dilator responses was obtained; absolute differences with baseline LMA responses were computed and compared between subjects with CN/MCI versus those with AD/VaD. The doses selected represent physiologically relevant concentrations observed in vivo in humans (15, 17, 19).

Group comparisons were performed using one-way ANOVA with the pairwise Holm-Sidak method for normally distributed data and ANOVA on ranks with a pairwise Kruskal-Wallis test for non-normally distributed data. Baseline and posttreatment dilator responses were compared using a paired t-test. Posttreatment responses (changes in the dilator response) were compared between subjects with CN/MCI versus AD/VaD and between donors with CAA versus those without CAA using an unpaired t-test. Comparisons between two groups with non-normally distributed data were done using a Mann-Whitney rank sum test. Correlation analyses were done using Pearson’s correlation. χ²-Analysis or Fisher exact test was used for categorical variables. Data are expressed as means ± SE. In case of missing data, no imputation was performed. A significant P value was set at P < 0.05.

RESULTS

The donor pool consisted of elderly subjects, most whom had cardiovascular risk factors or disease (Table 1). Donors with other dementia (N = 5) were significantly younger than donors with CN, MCI, AD, or VaD. There were no significant differences in sex, postmortem interval, and presence of cardiovascular comorbidities (coronary or peripheral arterial disease, hypertension, diabetes, and hyperlipidemia) among the groups (Table 1). As expected, there were significant differences in cognitive function (last mini-mental state examination score) among the groups.

There were no significant differences in postmortem brain weight (Table 1). Total senile amyloid plaque density was significantly higher in donors with AD compared with donors with other dementia but did not differ compared with donors with CN or MCI. Although CAA score (density of amyloidotic vessels) did not differ among the groups, the proportion of donors with CAA was higher in the AD group versus the other dementia group but not with CN or MCI groups.

There were no significant differences in unpressurized vessel diameters among the groups, although subjects with other dementia showed a trend toward larger baseline values (Table 1). Average arteriole diameters increased from 0- to 30-mmHg intraluminal pressure. There were no significant differences in vessel diameters among subjects with CN, MCI, AD, and VaD; however, vessel diameter was significantly higher in subjects with other dementia at 30 mmHg, 60 mmHg, and maximum dilation versus other groups, although the changes in diameters after pressurization did not differ among the groups. The average maximum vessel diameter was 177 ± 5 μm.

There was no significant difference in endothelium-dependent function or smooth muscle-dependent function among subjects with CN, MCI, AD, and other dementia and among subjects with CN, MCI, VaD, and other dementia (Fig. 1). In separate analyses, donors with both AD and VaD were compared with CN, MCI, and AD without VaD and other dementia. The AD + VaD maximum endothelium-dependent dilation to acetylcholine (72.1 ± 6.2%), acetylcholine logM EC₅₀ (−6.0 ± 0.4), and smooth muscle-dependent dilation (91.1 ± 3.1%) did not significantly differ compared with the other groups.

Fig. 1.

Leptomeningeal arteriole vasoreactivity. A−C: there were no differences in baseline endothelial function subjects with normal cognition (CN), mild cognitive impairment (MCI), Alzheimer’s disease (AD), and other dementia were compared and when subjects with CN, MCI, vascular dementia (VaD), and other dementia were compared. D: dilator response to a single dose of papaverine or DETA-NONOate to reflect smooth-muscle dependent function also showed no significant difference among the groups. NS, not significant.

There was no significant correlation between last mini-mental state examination (MMSE) score and endothelial function or smooth muscle function (Fig. 2).

Fig. 2.

Leptomeningeal arteriole (LMA) function and cognitive function. There was no correlation between last cognitive function score [mini-mental state examination (MMSE) score] and measures of LMA endothelial (A) or smooth muscle-dependent function (B). NS, not significant.

There was no difference between subjects with CAA (N = 31) and those without CAA (N = 14) in terms of endothelium-dependent function (maximum acetylcholine dilation: 71.0 ± 3.2% vs. 81.1 ± 3.8% and acetylcholine log M EC₅₀: −5.8 ± 0.2 vs. −6.2 ± 0.2, respectively, both P = not significant) and smooth muscle-dependent function (91.5 ± 1.7% vs. 93.9 ± 1.7%, respectively, P = not significant).

Exposure of LMAs to physiologically relevant doses of Aβ42, PA, and medin caused a profound impairment in endothelium-dependent function (−31.4 ± 5.9%, −32.8 ± 8.9%, and −39.6 ± 7.2% vs. the baseline control response, respectively) while also causing an impairment in smooth muscle-dependent function (−11.3 ± 5.2%, −29.8 ± 8.7%, and −16.6 ± 6.1% vs. the baseline control response, respectively; Fig. 3). There was no significant difference in responses to Aβ42, PA, or medin between subjects with CN/MCI versus those with AD/VaD.

Fig. 3.

Vasoreactivity responses to vascular stressors. A−C: there was significant impairment in endothelial function when leptomeningeal arterioles (LMAs) were exposed to β-amyloid (Aβ42; 2 μM), palmitic acid (PA; 150 μM), and medin (5 μM) compared with the baseline control response. D−F: the impairment in response was not different between LMAs from subjects with normal cognition (CN)/mild cognitive impairment (MCI) vs. subjects with Alzheimer’s disease (AD)/vascular dementia. G−I: there was impaired smooth muscle-dependent function in LMAs exposed to Aβ42, PA, and medin. Posttreatment smooth muscle-dependent vasoreactivity responses were not different between LMAs from subjects with CN/MCI vs. subjects with AD/VaD. NS, not significant.

DISCUSSION

Cerebrovascular disease leading to brain hypoperfusion is believed to contribute to the pathophysiology of VaD, sporadic AD, and aging-related dementia (10). Collateral perfusion through LMAs is a critical component of cerebrovascular autoregulation; it is emerging as a key determinant of clinical outcome after ischemic stroke (16). A preclinical investigation showed impaired LMA vasoreactivity in a hypertensive rat model that may contribute to perfusion deficits (4). Leptomeningeal atherosclerosis was associated with AD neuropathological lesions, suggesting that atherosclerosis-induced hypoperfusion contributes to AD pathology (22). Our observations represent the first measurement and comparison of human LMA vascular function among subjects with CN, MCI, AD, VaD, and other dementia that we know of. Contrary to our hypotheses, endothelium-dependent function and smooth muscle-dependent function (within the limits of our experimental conditions) did not differ among subjects with CN, MCI, and dementia, suggesting that AD or VaD does not confer fixed phenotypic alteration in baseline LMA function. Consistent with this finding, there was also a lack of correlation between LMA vasoreactivity and cognitive function score. Although one could be tempted to interpret the results as being not supportive of the etiologic contribution of LMA dysfunction to cognitive dysfunction, our results after LMA exposure to stressors (Aβ, PA, and medin) suggest an alternative interpretation. The idealized, plasma-free, ex vivo conditions in our baseline vasoreactivity experiments likely do not reflect the metabolic milieu present in vivo in terms of the presence of metabolic abnormalities that could impair microvascular function (such as hyperlipidemia, exposure to circulating amyloid proteins, and proinflammatory state), and in the absence of vascular metabolic stressors as represented by our baseline conditions, no significant differences in the microvascular response were noted. However, the in vivo metabolic milieu may be different among the groups tested. Adverse metabolic milieu was simulated in our ex vivo model after exposure to Aβ, PA, or medin. Posttreatment arteriole vasoreactivity showed consistent impaired responses to a malignant metabolic milieu, conditions that are known to exist in vivo. Because the doses of vascular stressors used in this study are within levels reported in humans (15, 17, 19), living conditions/in vivo states where these conditions exist could recapitulate our posttreatment experimental conditions and cause microvascular dysfunction, which, especially in the setting of periodic and chronic exposure, could potentially lead to brain hypoperfusion. Future studies focusing on differences in the magnitude and duration of exposure in vivo to vascular metabolic stressors in subjects with or at risk for AD or VaD versus subjects with CN may yield insights on how microvascular dysfunction contributes to cognitive dysfunction that might also serve as therapeutic targets for modulation.

The present study is the first that we know of showing impaired brain collateral circulation function after exposure to PA, a well-established dietary cardiovascular risk factor. Earlier efforts by our group showed that Aβ42 (25, 26) and medin (17) caused LMA endothelial dysfunction. The present study vastly expanded on these data sets, and, in doing so, we not only validated prior observations but the larger sample size also allowed comparisons of differential responses between subjects with dementia and those without dementia, a novel aspect of this study. Interestingly, we also showed the lack of differences in LMA endothelium-dependent function and smooth muscle-dependent function between donors with and without CAA suggesting that soluble species, and not the insoluble amyloid forms, may confer vascular toxicity. Finally, the ability to measure arteriole function after rapid autopsy in a large series of brain donors points to the potential of this new human tissue model to study vascular determinants of neurodegenerative diseases that could bridge the translational gap between preclinical transgenic mouse models and human clinical trials.

In terms of basal myogenic tone, it is interesting to note that the gradual increase in LMA diameters from 0 to 30 mmHg was similar to responses observed in the rat posterior cerebral artery (20), but rat posterior cerebral arteries, unlike human LMAs, showed a constriction response (signifying development of myogenic tone) when exposed to 60 mmHg of intraluminal pressure. Future studies should investigate whether human collateral LMA responses differ from intraparenchymal or penetrating arteriole responses as they may have important implications in vasoregulation.

Our study is limited by the lack of detailed morphological assessment of the LMA vessels, which should be comprehensively studied in the future to gain insights on morphology-function relationships. We also did not study the role of penetrating and intracerebral arterioles and capillaries, which play critical roles in cerebral perfusion. We lack data on differential responses to shear stress as well as the in vivo status of the collateral brain circulation from imaging data that could have provided additional context to our findings. An additional study limitation is the lack of therapeutic intervention to restore microvascular function after exposure to vascular stressors that would inform mechanisms of injury and may uncover differential responses among subjects with CN, MCI, and dementia, if they exist. Despite this limitation, our previous work on a more limited number of donor LMAs did point to specific signaling mechanisms underlying vascular injury. We showed that Aβ42 (26) and medin (6, 17) induced endothelial cell oxidative and nitrative stress and reduced nitric oxide bioavailability, which were reversed by agents that have antioxidant effects. Additionally, we showed that inhibition of receptor for advanced glycation end product (RAGE) also restored LMA function after exposure to medin (17), suggesting that RAGE may be a common pathway by which Aβ (8, 9) and medin affect the microvasculature. Similar to the LMA response to PA shown in this study, human adipose arterioles using a similar experimental setup also showed profound endothelial dysfunction when exposed to fatty acid-rich very-low-density lipoprotein lipolysis products (15), and PA, like Aβ and medin, is known to impair endothelial cell nitric oxide production (14). Our findings support the need to systematically study the acute and chronic effects of PA and fatty acids, enriched in the Western diet, on cerebrovascular function.

The impaired treatment responses to Aβ42, PA, and medin are likely not due to impaired tissue viability after the baseline response. In our prior work, cotreatment of Aβ42 with phosphatidic acid-containing nanoliposomes (26) and cotreatment of medin with the antioxidant polyethylene glycol-superoxide dismutase or the RAGE inhibitor FPS-ZM1 (17) fully restored microvascular function, suggesting that vessels continue to be viable in these experimental conditions. It is important to note that in assessing the smooth muscle-dependent dilator response, we only used one high concentration (10⁻⁴ M) dose and did not perform full dose-response experiments, which might reveal differences not seen with the high dose.

In summary, our results show the novel finding that LMA endothelial and smooth muscle-dependent function did not differ among subjects with CN, MCI, and dementia, whereas impaired vasoreactivity was seen in, yet did not differ, among subjects with CN, MCI, and dementia after exposure to Aβ, PA, and medin. Unlike its emerging critical role in stroke pathophysiology, the role of the collateral cerebral circulation in the pathophysiology of cognitive dysfunction in AD and related dementia disorders remains to be fully explored and defined.

GRANTS

This work was supported by National Institutes of Health Grants R21-AG-044723, U24-NS-072026, P30-AG-19610, RO1-AG-019795, Veterans Affairs Merit Awards BX-003767 and BX007080, British Heart Foundation Grant FS/12/61/29877, Arizona Department of Health Services, Arizona Biomedical Research Commission, Grant 4001/0011/05–901/1001, Michael J. Fox Foundation for Parkinson’s Research, and United States Department of Defense Grant AZ160056.

DISCLAIMERS

This work does not represent the views of the Veterans Affairs or United States government.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

R.Q.M., J.M., and T.G.B. conceived and designed research; R.Q.M., S.T., N.K., H.A.D., and J.M. performed experiments; R.Q.M., S.T., G.S., and C.M. analyzed data; R.Q.M., C.M., P.R., and T.G.B. interpreted results of experiments; R.Q.M. prepared figures; R.Q.M. drafted manuscript; R.Q.M., S.T., N.K., G.S., C.M., H.A.D., J.M., P.R., and T.G.B. edited and revised manuscript; R.Q.M., S.T., N.K., G.S., C.M., H.A.D., J.M., P.R., and T.G.B. approved final version of manuscript.