Synergistic Memory Impairment Through the Interaction of Chronic Cerebral Hypoperfusion and Amlyloid Toxicity in a Rat Model
Background and Purpose—Vascular pathology and Alzheimer disease (AD) pathology have been shown to coexist in the brains of dementia patients. We investigated how cognitive impairment could be exacerbated in a rat model of combined injury through the interaction of chronic cerebral hypoperfusion and amyloid beta (Aβ) toxicity.
Methods—In Wistar rats, chronic cerebral hypoperfusion was modeled by permanent occlusion of bilateral common carotid arteries (BCCAo). Further, AD pathology was modeled by bilateral intracerebroventricular Aβ (Aβ toxicity) using a nonphysiological Aβ peptide (Aβ 25 to 35). The experimental animals were divided into 4 groups, including sham, single injury (Aβ toxicity or BCCAo), and combined injury (BCCAo-Aβ toxicity) groups (n=7 per group) . Cerebral blood flow and metabolism were measured using small animal positron emission tomography. A Morris water maze task, novel object location and recognition tests, and histological investigation, including neuronal cell death, apoptosis, neuroinflammation, and AD-related pathology, were performed.
Results—Spatial memory impairment was synergistically exacerbated in the BCCAo–Aβ toxicity group as compared to the BCCAo or Aβ toxicity groups (P<0.05). Compared to the sham group, neuroinflammation with microglial or astroglial activation was increased both in multiple white matter lesions and the hippocampus in other experimental groups. AD-related pathology was enhanced in the BCCAo–Aβ toxicity group compared to the Aβ toxicity group.
Conclusion—Our experimental results support a clinical hypothesis of the deleterious interaction between chronic cerebral hypoperfusion and Aβ toxicity. Chronic cerebral hypoperfusion-induced perturbation in the equilibrium of AD-related pathology may exacerbate cognitive impairment in a rat model of combined injury.
Alzheimer disease (AD) and vascular dementia are the most common causes of cognitive decline in the aging population.1 Accumulation of insoluble amyloid beta (Aβ) in the brain has been identified as the major culprit for the cognitive impairment observed in AD patients.2 Because senile plaques composed of the Aβ peptide have been found in the brains of AD patients,2 extensive research has focused on the amyloid hypothesis to explain AD pathology.
A hypothesis emphasizing the interaction between AD and vascular pathologies has recently emerged.3–5 The Nun study and other clinico-pathological studies6–8 have revealed that patients with AD exhibit concomitant vascular lesions in the brain. Further, epidemiological studies have shown that the major risk factors for AD mostly coincide with those for vascular dementia.4 The Rotterdam study,9 a large population-based prospective study, reported an increased prevalence of atherosclerosis in patients either with AD or with vascular dementia. In recent imaging studies, vascular lesions such as white matter ischemic changes or lacunar infarction have been suggested as risk factors for cognitive decline in patients with AD.10
A converging hypothesis involving chronic cerebral hypoperfusion and AD pathology originally has been suggested in clinical studies.11–13 Furthermore, the double-hit vascular hypothesis of AD, based on molecular and cellular experimental studies, has been proposed.14–17 An intact functional neurovascular unit is an active system prerequisite to maintain normal brain function.15 Entry of peptides in the brain is limited by the blood–brain barrier, unless there is a specific receptor-mediated or carrier-mediated transport, as in case of amino acids.17–19 The concept of blood–brain barrier dysfunction in a progressive age-dependent vascular-mediated secondary neurodegeneration has been suggested.16,17
However, how these 2 distinct vascular and AD pathologies interact has not been clearly established. Recent animal experiments using a rat model with combined injury of lacunar infarction and Aβ toxicity revealed that vascular insults exacerbated AD pathology.20 Aside from lacunar infarction, we hypothesized that chronic cerebral hypoperfusion may be an exacerbating factor for AD pathology and cognitive impairment. Modeling vascular dementia in animals has been demonstrated by permanent occlusion of the bilateral common carotid arteries in rats.21–23 Further, intraventricular injection of Aβ has been reported as a method for modeling of AD in rats.24,25 In the current study, we investigated how cognitive impairment could be synergistically exacerbated in a combined injury model of chronic cerebral hypoperfusion and Aβ toxicity in rats.
Materials and Methods
Male Wistar rats were used in these experiments (Charles River, Gapeung, Republic of Korea). All rats resided in the vivarium at Konkuk University for 2 weeks before the beginning of the experiment. The rats were housed at a controlled temperature (22°C±1°C) and humidity (50%±10%) on a 12-hour alternate light–dark cycle. Food and water were provided ad libitum throughout the experiment. All animal experimental procedures were in accordance with the approved guidelines of the Institutional Animal Care and Use Committee of Konkuk University.
The experimental animals were divided into 4 groups using a 2-by-2 table of permanent occlusion of bilateral common carotid arteries (BCCAo) or bilateral Aβ 25 to 35 intracerebroventricular injection (Aβ toxicity), including sham, Aβ toxicity, BCCAo, and BCCAo–Aβ toxicity groups (Figure 1A). Detailed procedures are described in the Supplemental Methods (http://stroke.ahajournals.org). A timeline for surgical procedures for single (Aβ toxicity or permanent occlusion of bilateral common carotid arteries) or combined injury (permanent occlusion of common carotid arteries and Aβ toxicity) and other tests were described (Figure 1B).
Water maze task and novel object location and recognition behavior tests were performed to evaluate cognitive impairment. Before the water maze task, a blind test was performed to exclude blinded rats. Detailed procedures are described in the Supplemental Methods.
Cerebral Blood Flow and Metabolism Measurement
Cerebral blood flow and metabolism were measured by 18F-fluorodeoxyglucose positron emission tomography scanner in the acute and chronic phases after permanent occlusion of bilateral common carotid arteries. First 5-minute and between 60- and 80-minute scans of 18F-fluorodeoxyglucose uptake were used as indicators of cerebral blood flow and metabolism, respectively.26,27 Detailed procedures are described in the Supplemental Methods.
Six weeks after sham operation or permanent occlusion of bilateral common carotid arteries (ie, 3 weeks after Aβ or phosphate-buffered saline injection), thionin staining, terminal deoxynucleotidyl transferase dUTP nick end labeling, and immunohistochemisty, including OX-6, Iba-1, glial fibrillary acidic protein, NeuN, MAP-2, laminin, amyloid β 1 to 42, amyloid precursor protein (APP), and tau-2, were performed. Quantitative analysis was performed when it was needed. Detailed procedures and antibody information are described in the Supplemental Methods.
Parameters for spatial memory, including search error, time latency, and path length, as well as swimming speed, were analyzed by 1-way repeated-measure ANOVA followed by a post hoc Fisher protected least-significant differences test. One-way ANOVA was conducted to compare results of the probe trials, cued behavior, percent exploratory preference, and the quantitative data of the immunohistochemstry (mean±SE). A value of P<0.05 was considered to be statistically significant. Data analyses were performed with the SPSS software version 12.0.
Mortality and Blindness
All procedures for the permanent occlusion of bilateral common carotid arteries were completed within 15 minutes. By the time of Aβ or phosphate-buffered saline injection, the mortality rate was 22.9% (8 of 35) in the permanent occlusion of bilateral common carotid arteries group and 0% (0 of 14) in the sham-operated group (Figure 1A). Before Aβ or phosphate-buffered saline injection, rats with escape latency longer than the blindness cut-off were considered blind and were excluded from further experiments (Figure 1C). The blindness rate among surviving rats was 48.1% (13 of 27) in the permanent occlusion of bilateral common carotid arteries group and 0% (0 of 14) in the sham-operated group (Figure 1A). Fourteen surviving nonblinded rats from each group were equally allocated for an additional Aβ or phosphate-buffered saline injection procedure. No mortality related to the Aβ or phosphate-buffered saline injection procedure developed, and no major neurological deficits impairing motor function, such as forelimb flexion or unilateral circling, were observed in any group. Ultimately, 7 rats per group were used for the behavioral test and 4 per group were used for the histological study.
Cerebral Blood Flow Measurement
After permanent occlusion of bilateral common carotid arteries, the first 5-minute uptake of 18F-fluorodeoxyglucose indicating cerebral blood flow in the cortex was decreased to 96.8% of the baseline at 3 days, 98.3% at 5 weeks, and 74.5% at 8 weeks (Figure 2A). Decreased cerebral blood flow in the cortex was not evident at 3 days and 5 weeks after permanent occlusion of bilateral common carotid arteries; however, decreased pattern was shown at 8 weeks. Between 60- and 80-minute uptake of 18F-fluorodeoxyglucose, indicating cerebral metabolism in the cortex, was decreased to 83.1% of the baseline at 3 days, 80.1% at 5 weeks, and 80.3% at 8 weeks after permanent occlusion of bilateral common carotid arteries (Figure 2B). Compared to the slightly recovered pattern of the metabolism in the striatum and hippocampus, decreased metabolism in the cortex was sustained over the course of 8 weeks.
Parameters for spatial memory were significantly different among groups with respect to search error (P<0.001), time latency (P<0.001), and path length (P<0.001; Figure 3A). Swimming speed was not different among groups (P=0.27; Figure 3B). Among the 4 groups, the rats with BCCAo–Aβ toxicity showed maximally increased search error, time latency, and path length. On post hoc analysis, the rats with BCCAo exhibited increased search error (P<0.05), time latency (P<0.01), and path length (P<0.01) compared to rats with sham operation or Aβ toxicity (Figure 3A). When Aβ toxicity was added to the permanent occlusion of bilateral common carotid arteries injury (ie, the BCCAo–Aβ toxicity group), there was a synergistic increase in both search error (P<0.05) and time latency (P<0.05) compared to the BCCAo group. In the first probe trial, the time in the target quadrant was not different among groups (P=0.22); however, in the second probe trial, the time in the target quadrant increased up to >40% in sham-operated rats, implicating effective learning and a difference among groups (P<0.05). On post hoc analysis, the time spent in the target quadrant significantly decreased in rats with Aβ toxicity (P<0.05), BCCAo (P<0.05), and BCCAo–Aβ toxicity (P<0.01) compared to sham-operated rats (Figure 3C). Cued time latency to reach the visible platform was not different among groups (P=0.27; Figure 3D).
On the novel object location test, percent exploratory preference was significantly different among groups (P<0.05). On post hoc analysis, percent exploratory preference significantly decreased in rats with Aβ toxicity (P<0.05), BCCAo (P<0.05), and BCCAo–Aβ toxicity (P<0.05) compared to sham-operated rats (Figure 1D). On the novel object recognition test, percent exploratory preference was not different among groups (P=0.80; Figure 1D).
Laminin staining showed no definite changes of vascular basement membrane in the capillaries or total length of the capillaries in the cortex among experimental groups (Figure 2C).
Apoptosis or Changes in Neuronal Morphology
Thionin staining of the hippocampus demonstrated no significant morphological changes of neurons in the CA1, CA3, and dentate gyrus in any group (Figure 4A). Additional staining including transferase dUTP nick end labeling (Figure 4B), NeuN, and MAP-2 staining (Figure 4C) showed no significant apoptosis or any neuronal morphological changes in the CA1. Further, no definite cell death was observed in the cortex and striatum in any group (data not shown).
As reported in previous studies using rats with permanent occlusion of bilateral common carotid arteries, a microglial marker of the neuroinflammation (OX-6) increased in the white matter region, including the corpus callosum and fimbria of the hippocampus, as well as in the internal capsule (Figure 5A). Microglial activation was also induced in rats with Aβ toxicity and BCCAo–Aβ toxicity. Microglial activation in other region, including the cortex, striatum, and hippocampus, was not as prominent as in the white matter region (data not shown). OX-6 staining also can mark peripheral monocytes or macrophages that have entered the brain through a damaged blood–brain barrier. However, OX-6–positive activated microglial cells, displaying a larger amoeboid form with stubby processes, were colocalized with Iba-1, suggestive of the central nervous system origin (lowest column in Figure 5A).
An astroglial marker of the neuroinflammation (glial fibrillary acidic protein) increased in the CA1, CA3, and dentate gyrus subfields of the hippocampus in rats with Aβ toxicity, BCCAo, or BCCAo–Aβ toxicity (Figure 5C). Further, astroglial activation also was induced in the white matter lesions, where microglial activation was enhanced (data not shown).
Quantitative analysis revealed that compared to the sham group, neuroinflammation with microglial (P<0.01 in the corpus callosum; P<0.05 in the fimbra of the hippocampus) or astroglial (P<0.05 in the CA1 and CA3; P<0.01 in the dentate gyrus) activation was increased both in multiple white matter lesions and the hippocampus in other experimental groups (Figure 5B, D). However, no synergistic increase in neuroinflammation was observed in rats with BCCAo–Aβ toxicity compared to rats with Aβ toxicity or BCCAo.
AD-related pathology was scarcely observed in the white matter region in any group. Deposition of Aβ in the cortex and striatum was not different among rats with a single or combined injury (data not shown). Both APP and tau-2 pathology were scarcely and randomly observed in the cortex and striatum in sham-operated rats. In rats with BCCAo–Aβ toxicity, APP staining showed enhanced immunoreactivity with sporadic cellular or clustered plaque patterns in the cortical or deep gray matter (Figure 6A) and tau-2 staining with sporadic cellular patterns in the cortical gray matter (Figure 6B) compared to rats with Aβ toxicity. Quantitative analysis revealed that, compared to the sham group, AD-related pathology was increased in other experimental groups (P<0.01; Figure 6C). The tau-2 pathology was especially significantly increased in rats with BCCAo–Aβ toxicity compared to rats with a single injury (ie, Aβ toxicity or BCCAo).
Spatial Memory Impairment Synergistically Exacerbated by the Combined Injury of Chronic Cerebral Hypoperfusion and Amyloid Toxicity
Chronic cerebral hypoperfusion has been assumed as a pathogenic mechanism in patients with not only vascular dementia but also AD.4,5 Cognitive impairment has been demonstrated in animals with chronic cerebral hypoperfusion, modeling patients with vascular dementia.21,28 Results of our experiment were in accordance with previous studies that replicated spatial memory impairment in rats with permanent occlusion of bilateral common carotid arteries. With respect to swimming path analysis, the search error, time latency, and path length to find a hidden platform were increased in rats with BCCAo, although the swimming speed was not different among groups. Rats with a visual disturbance were excluded by a blind test before the water maze task and cued time latency dependent on the visual information was not different among rats included in our study. Additionally, other cognitive tests relatively unrelated to the visual ability, such as novel object location and recognition test, also showed some cognitive impairment in rats with BCCAo. Taken together, delays in finding a hidden platform may have occurred because of the spatial memory impairment, rather than being produced by motor or visual dysfunction.
In previous studies, diverse animal models with Aβ toxicity showed cognitive impairment in various behavioral tests.24,25,29 However, in our experiment, rats with Aβ toxicity did not show any spatial memory impairment through an analysis of the swimming path. According to the post hoc analysis of the swimming path, spatial memory impairment that was initially induced by the permanent occlusion of bilateral common carotid arteries became significantly exacerbated by additional Aβ toxicity in rats with combined injury (BCCAo–Aβ toxicity). In the second probe trial, the time to swim in a target quadrant was >40% in sham-operated rats, demonstrating normal memory consolidation with respect to spatial learning (ie, rats remembering where the hidden platform was). According to the post hoc analysis of the second probe trial, spatial memory was significantly impaired in all 3 injury groups (Aβ toxicity, BCCAo, and BCCAo–Aβ toxicity). Interestingly, this included the Aβ toxicity group in which spatial memory impairment was not initially evident in the swimming path analysis compared to sham-operated rats. In general, a trend for severe spatial memory impairment in rats with combined injury was demonstrated in probe trials.
Another behavioral test including novel object location and recognition test was performed to confirm memory impairment in our animal models.16,30,31 The novel object location test, indicating more hippocampal specific memory, showed significant difference among groups.30 Compared to sham-operated rats, other experimental groups showed significant memory impairment, although a synergistic pattern was not observed in rats with BCCAo–Aβ toxicity. In contrast, the novel object recognition test, indicating perirhinal cortex-mediated memory consolidation,16,31 showed no significant difference among groups.
Aβ toxicity in our model may not be enough to induce full-blown cognitive impairment, because rats with Aβ toxicity showed memory impairment only in the analysis of probe trials and the novel object location test. Too short time interval between Aβ injection and behavioral tests or too low concentration of used Aβ may have been the reason why the rats with Aβ toxicity failed to show definite cognitive impairment. However, our Aβ toxicity model with subtle cognitive impairment may be more appropriate to test the hypothesis of synergistic exacerbation of chronic cerebral hypoperfusion and Aβ toxicity than already full-blown AD models. Regarding synergism, when minimal Aβ toxicity was combined by the permanent occlusion of bilateral common carotid arteries, the BCCAo–Aβ toxicity group showed statistically significant synergism of memory impairment beyond the baseline memory impairment initially induced by the permanent occlusion of bilateral common carotid arteries. Similarly, AD-related pathology that was initially minimal in rats with Aβ toxicity was also enhanced when Aβ toxicity was combined with permanent occlusion of bilateral common carotid arteries. Therefore, although Aβ toxicity itself in our model may not have been enough to induce spatial memory impairment independently, subclinical Aβ toxicity could have played a synergistic role when it was combined with critically impaired cerebral hypoperfusion.
Taken together, results of behavioral tests and histological investigations in our experiment support the clinical hypothesis that chronic cerebral hypoperfusion and AD pathology could interact in a synergistically exacerbating manner.
Mechanism of Combined Injury From Chronic Cerebral Hypoperfusion and Amyloid Toxicity
A converging hypothesis has emerged to explain the interaction between vascular and AD pathologies.3–5 However, how these pathologies interact with each other has not been clearly studied. In the current study, we investigated neuronal cell death in the hippocampus to explain spatial memory impairment. Cell death and diffuse atrophy of the hippocampus with behavioral impairment have been reported in a rat model of transient global ischemia or chronic cerebral hypoperfusion.21,32 On the contrary, no definite morphological change in neurons or atrophy of the hippocampus was observed in any of our experimental groups. Our results suggested that a neuronal toxic effect from the permanent occlusion of bilateral common carotid arteries, Aβ toxicity, or even a combined injury was not strong enough to induce neuronal cell death in the hippocampus.
Second, we hypothesized that the combined injury of the permanent occlusion of bilateral common carotid arteries and Aβ toxicity may exacerbate neuroinflammation. Neuroinflammation or demyelination with neuroinflammation has been suggested as a key pathological mechanisms of chronic cerebral hypoperfusion.21–23,33 Further, microglial or astroglial activation in the hippocampus has been reported in animal models of chronic cerebral hypoperfusion or Aβ toxicity.20,21 A rat model with a combined injury of lacunar infarction and Aβ toxicity showed increased neuroinflammation in the hippocampus.20 In our experiment, neuroinflammation was increased by the permanent occlusion of bilateral common carotid arteries and Aβ toxicity, as compared to that in sham-operated rats; however, an additional synergistic exacerbating effect was not definite in the combined injury group. Neuroinflammation could not fully explain the synergistic exacerbation that was observed in the spatial memory test in the combined injury group.
Third, we investigated AD-related pathology, including Aβ, APP, and tau. Increased APP deposition has been demonstrated not only in animal models of AD but also in acute or chronic ischemic lesions.33–35 In addition, APP deposition in the cortex and hippocampus in an ischemic rat model and alteration of Aβ metabolism by chronic cerebral hypoperfusion have been suggested.36,37 Chronic cerebral hypoperfusion has been shown to accelerate Aβ deposition in an APP transgenic mouse model with chronic stenosis of the bilateral carotid arteries.38 Furthermore, Aβ deposition in a transgenic mouse model of cerebral amyloid angiopathy has been demonstrated to impair cerebral blood flow regulation.39 Chronic cerebral hypoperfusion and Aβ toxicity may interact with each other in a deleterious way, rather than a 1-sided cause-and-effect relationship. In our experiment, deposition of Aβ itself in the cortex, striatum, or hippocampus was not different among rats with a single or combined injury. Further, both APP and tau pathology were scarcely and randomly observed in both the cortex and striatum of sham-operated rats. However, APP and tau pathology were enhanced in rats with BCCAo–Aβ toxicity compared to rats with Aβ toxicity. Patterns of APP staining were shown to be sporadic neuronal or conglomerated plaque forms. Therefore, chronic cerebral hypoperfusion may play a key role in triggering Aβ toxicity.
The importance of cerebral blood flow in AD has been well-documented in clinical studies.11,12 A clinical hypothesis has suggested that regional cerebral hypoperfusion is one of the earliest clinical manifestations in patients with AD.12 It has been shown that amyloid is in dynamic equilibrium in terms of clearance and kinetics between the brain and blood compartments.15 In a previous study, mutant Aβ peptides showed reduced clearance from the brain because of lack of affinity for the Aβ peptide carrier, which eliminates Aβ from the cerebrospinal fluid and brain into the vascular circulation.17 It also has been suggested that perturbation in the equilibrium of amyloid deposition between brain parenchymal and vascular compartments may induce pathological changes.15,40,41 Further, permanent occlusion of bilateral common carotid arteries-induced suboptimal cerebral perfusion may generate ultrastructural damage of the cerebral capillary walls.13 In our experiment, this capillary damage may have had some effect on the normal clearance of AD-related pathology. However, no definite morphological changes of the vascular basement membranes in the capillaries were observed. Instead, our serial positron emission tomography study showed decreased cerebral blood flow and metabolism in the various areas of the cerebral hemispheres, especially in the cortex. Although vascular injury in our chronic cerebral hypoperfusion model may not have been enough to induce strong structural capillary damages as in acute ischemic models, functional impairment in the cerebral blood flow and metabolism may be the main pathogenic mechanisms in our chronic cerebral hypoperfusion model.
In addition, chronic cerebral hypoperfusion has been reported to disturb endogenous AD-related pathology, including APP.35,37,38 In our experiment, exogenous Aβ in the Aβ toxicity group may not have been enough to induce AD-related pathology and was likely cleared by normal cerebral perfusion. Thus, memory impairment and AD-related pathology may be minimal in this Aβ toxicity group. However, when Aβ toxicity was combined with the permanent occlusion of bilateral common carotid arteries, AD-related pathology was exacerbated. The vascular clearance mechanism may be disrupted in rats with permanent occlusion of bilateral common carotid arteries because of downregulation of low-density lipoprotein receptor-related protein 1, the key Aβ clearance receptor, which has been recently shown to be reduced in AD cerebral vessels via hypoxia-mediated overexpression of vascular specific genes.42 Further study focused on the Aβ clearance receptor in this rat model of combined injury is needed.
Limitations of Aβ 25 to 35
Aβ 25 to 35 used in our experiment may not have contained the Aβ sequences that play a critical role in both transport and interactions with Aβ putative receptors.17 The Aβ brain extract produced in vivo showed particular properties that are essential for disseminating AD pathology.43 Concentration of Aβ 25 to 35 in our experiment was significantly higher than observed levels of endogenous Aβ in patients with AD.14 Therefore, further study using different confirmations of Aβ peptide (Aβ 1 to 40 or Aβ 1 to 42) with the concentration relevant to clinical AD pathogenesis is needed, because artificial Aβ 25 to 35 may not have the same pathological activity as the endogenous Aβ proteins found in AD patients.
In summary, chronic cerebral hypoperfusion-induced perturbation in the equilibrium of AD-related pathology may exacerbate cognitive impairment in a rat model of combined injury. Thus, maintaining optimal cerebral perfusion may be crucial to prevent further cognitive impairment in patients with AD.
Sources of Funding
This work was supported by the Korea Research Foundation Grant funded by the Korean Government (KRF-2008-331-E00298).
The authors thank Kwang Sun Woo for his technical assistance.
The online-only Data Supplement is available at http://stroke.ahajournals.org/cgi/content/full/STROKEAHA.111.620179/DC1.
- Received March 11, 2011.
- Accepted March 31, 2011.
- © 2011 American Heart Association, Inc.
- Masters CL,
- Simms G,
- Weinman NA,
- Multhaup G,
- McDonald BL,
- Beyreuther K
- Yoshita M,
- Fletcher E,
- Harvey D,
- Ortega M,
- Martinez O,
- Mungas DM,
- et al
- Monro OR,
- Mackic JB,
- Yamada S,
- Segal MB,
- Ghiso J,
- Maurer C,
- et al
- Whitehead SN,
- Hachinski VC,
- Cechetto DF
- Whitehead SN,
- Cheng G,
- Hachinski VC,
- Cechetto DF
- Prediger RD,
- Franco JL,
- Pandolfo P,
- Medeiros R,
- Duarte FS,
- Di Giunta G,
- et al
- Mullani NA,
- Herbst RS,
- O'Neil RG,
- Gould KL,
- Barron BJ,
- Abbruzzese JL
- Nishio K,
- Ihara M,
- Yamasaki N,
- Kalaria RN,
- Maki T,
- Fujita Y,
- et al
- Shibata M,
- Yamasaki N,
- Miyakawa T,
- Kalaria RN,
- Fujita Y,
- Ohtani R,
- et al
- Winters BD,
- Bussey TJ
- Shin HK,
- Jones PB,
- Garcia-Alloza M,
- Borrelli L,
- Greenberg SM,
- Bacskai BJ,
- et al
- Deane R,
- Wu Z,
- Zlokovic BV
- Eisele YS,
- Obermüller U,
- Heilbronner G,
- Baumann F,
- Kaeser SA,
- Wolburg H,
- et al