Ar 190-30 pdf download

Therapies targeting late events in Alzheimer’s disease (AD), including aggregation of amyloid beta (Aβ) and hyperphosphorylated tau, have largely failed, probably because they are given after significant neuronal damage has occurred. Biomarkers suggest that the earliest event in AD is a decrease of cerebral blood flow (CBF). This is caused by constriction of capillaries by contractile pericytes, probably evoked by oligomeric Aβ. CBF is also reduced by neutrophil trapping in capillaries and clot formation, perhaps secondary to the capillary constriction. The fall in CBF potentiates neurodegeneration by upregulating the BACE1 enzyme that makes Aβ and by promoting tau hyperphosphorylation. Surprisingly, therefore, CBF reduction may play a crucial role in driving cognitive decline by initiating the amyloid cascade itself, or being caused by and amplifying Aβ production. Here, we review developments in this area that are neglected in current approaches to AD, with the aim of promoting novel mechanism-based therapeutic approaches.

Thirty years of research have given us a broad understanding of many mechanisms contributing to Alzheimer’s disease [99], but over 400 clinical trials of drugs targeting these pathways have largely failed to reduce cognitive decline [47, 109, 136]. Identification of the amyloid β protein (Aβ) as the major component of amyloid plaques, together with genetic evidence, initially indicated that dysfunction of the processing of amyloid precursor protein (APP) was the cause of Aβ plaque deposition and downstream tau tangle formation and neuronal dysfunction [59]. Subsequent work led to the conclusion that the level of soluble Aβ oligomers, and of hyperphosphorylation of the cytoskeletal protein tau that is induced by Aβ [62, 91], correlated better with cognitive decline than did plaque level [7, 57, 89, 123].

There are established mechanisms by which Aβ oligomers and hyperphosphorylated tau can contribute to neuronal dysfunction and cognitive decline before synaptic and neuronal damage, and even before Aβ plaque and tau tangle deposition (Fig. 1). Aβ oligomers reduce glutamate uptake [92, 94, 199]. This raises the extracellular glutamate level and increases neuronal excitability [19, 20], which alters synaptic plasticity [92, 94] and in extremis may induce excitotoxicity [60]. Tau phosphorylation leads to soluble tau relocating from axonal microtubules into dendritic spines, where it alters postsynaptic glutamate receptor trafficking or anchoring (of both AMPA and NMDA receptors) and thus suppresses excitatory postsynaptic currents and neuronal activity [21, 67]. These changes may be particularly important when they affect the function of interneurons, which play a key role in generating oscillatory activity that contributes to cognitive function [63, 70, 176].

Preclinical AD has therefore been conceptualised as a synaptic disease [157] driven by Aβ and downstream tau phosphorylation, with loss of synapses and cells occurring late in the disorder. However, individuals can be cognitively normal while having plaque levels as high as those in Alzheimer’s dementia patients, and the same is true for levels of soluble Aβ oligomers [39]. This could reflect the presence of compensating protective mutations or developmental differences in the subjects with high Aβ levels. Alternatively, together with the fact that attempts to prevent cognitive decline—by blocking Aβ production, removing Aβ with antibodies or preventing tau phosphorylation—have all failed clinically (with one possible exception [68]), these data may suggest that there is some other variable that is missing from our understanding of the Aβ-tau cascade. Previously it has been suggested that the vasculature might provide such a factor, in the form of hypertension, impaired blood–brain barrier function, decreased Aβ clearance to the blood, vascular oxidative stress and inflammatory damage, or reduced neurovascular coupling at the arteriolar level [71, 198]. In this review, we show that new evidence reveals that a major missing variable is cerebral blood flow—and specifically its control by capillary pericytes.

Cerebral blood flow and glucose metabolism are reduced, and the brain’s vascular resistance is increased, in human AD [17, 107, 112, 115, 144, 151, 163, 165, 188] and in mice overexpressing amyloid precursor protein (APP) to mimic AD [129]. This also occurs in humans and mice expressing the ApoE4 protein, which predisposes towards AD [111, 148, 162, 163, 172]. The CBF reduction reaches over 50% in some brain areas [5], which is expected to reduce the activity of the Na/K pump (the main consumer of ATP in the CNS: [8]) and all processes dependent on it (including maintenance of the resting potential and glutamate uptake). It will also lead to adenosine generation, which is known to suppress glutamate release [43], and will produce numerous cell biological changes including changes of the balance of protein synthesis and degradation [173].

Although these changes could simply reflect tissue atrophy in AD [30], with a corresponding loss of blood supply and metabolism, they are associated with hypoxia [114] and it has been reported that the decrease of metabolism is greater than would be expected for the amount of atrophy occurring [165]. Furthermore, the observations of focal constrictions in capillaries from human AD brains [83], constriction of capillaries near plaques in human AD brains [58], and reduced neurovascular coupling and cerebrovascular reactivity in AD mice [48, 174] suggest that blood flow may be reduced by decreases in vessel diameter, and not just by loss of blood vessels.

Chronic blood flow reductions of 50% are expected to cause significant cognitive changes: a sustained reduction in CBF beyond 20% in humans leads to loss of ability to sustain attention, while a reduction beyond 30% in rats impairs spatial memory [105, 177]. A causal influence of blood flow changes on the cognitive changes at the onset of Alzheimer’s disease, before synapses or neurons are lost, is suggested by the fact that the reduction of cerebral blood flow starts early in preclinical AD [107, 180], with a faster onset than the deposition of Aβ or tau [76], and the fall of metabolism is also an early event [81, 115]. Furthermore, these changes correlate with cognitive decline [17, 112, 151].

The brain is unusual in that most of the resistance in its vascular bed is in capillaries (Fig. 2a) rather than in arterioles or venules [49], and cerebral blood flow is controlled not only by vascular smooth muscle cells wrapped around arterioles, but also by contractile pericytes which enwrap at least the first 4 branch orders of capillaries from the penetrating arteriole [9, 56, 82, 84, 143, 152, 187]. Contraction of these pericytes produces localised capillary constrictions near the pericyte somata (where most of the circumferential processes of the pericytes are located [133]) and could account for the focal capillary constrictions seen anatomically in capillaries isolated from human AD brains [83].

Despite the award of the Nobel Prize to Krogh [87] for his discovery of contractile elements on capillaries which act independently of smooth muscle cells on arterioles, there has been some controversy in the literature about whether pericytes are in fact contractile. However, this debate has now largely been resolved. The Zlokovic group [127] assessed in vitro, ex vivo and in vivo studies on pericyte contractility and found that 37 out of 39 separate papers reported that pericytes display contractility (and one of the 2 remaining papers [65] actually showed pericytes contracting, but renamed these cells smooth muscle cells: see [9] for discussion). Furthermore, whereas contractility had previously been demonstrated most clearly for pericytes on the 1st–4th branch orders of capillary measured from a penetrating arteriole [56, 65] which express the highest levels of α-smooth muscle actin, innovations in histochemistry have revealed that even higher branch order pericytes express this contractile protein [3] and optogenetic experiments have shown that these higher branch order pericytes can also regulate capillary diameter and blood flow [www.biorxiv.org/content/10.1101/2020.03.26.008763v1].

Functional indications that capillary pericyte-mediated control of CBF is disrupted in AD have been provided by measurements of the capillary transit time of the blood, and its heterogeneity. Magnetic resonance imaging (MRI) experiments on humans and optical imaging experiments on AD mice have found that AD leads to both a prolongation of the capillary transit time and an increase in its heterogeneity, as if some capillary pericytes became more constricted than others [38, 54]. Furthermore, in humans, these changes correlate with cognitive decline (Fig. 2b), as measured by the Brief Cognitive Status Examination [128].

By analysing images of brain biopsies of patients who consulted neurologists for dementia of unknown cause (Fig. 2c), Nortley et al. [133] demonstrated that patients developing AD have capillary blood flow restricted as a result of capillary constriction. This was shown to be due to pericytes by examining how capillary diameter varied as a function of the distance along the capillary from the pericyte soma (Fig. 2d). Patients depositing Aβ and tau tangles showed a constriction at the pericyte soma relative to positions between pericytes on the capillary. This increased rapidly with the amount of Aβ deposited, suggesting a CBF reduction mechanism that occurs early in the development of the disease (before accumulation of Aβ in and around vascular cells—cerebral amyloid angiopathy—leads to pericyte loss), as is also seen in live imaging of CBF in AD patients [107]. In contrast, in patients lacking Aβ and tau deposition, capillaries showed a larger diameter near the pericyte soma, perhaps because pericytes normally induce growth of the endothelial tube. The difference in the spatial profile of capillary diameter between AD and non-AD patients was estimated to be able to generate a reduction in CBF of ~ 50%, similar to that found in AD patients in vivo [5].

In AD mouse models, live cortex imaging through a cranial window, or reconstructing the hippocampal vasculature of fixed brains, also showed a reduction of mean capillary diameter compared to normal mice [55, 133, 193], which in cortex reflected capillary constriction near pericyte somata [133]. Nortley et al. [133] further demonstrated that, in the AD model mouse they used, neither arterioles nor venules had an altered diameter, implying that the reduction of CBF is generated by capillaries (although this still remains to be shown for human AD and other AD mouse models).

Although the mechanism of the long-term pericyte-mediated constriction of capillaries that occurs in human AD brains has not yet been definitively identified, short-term application of Aβ oligomers (both Aβ_1–42 and Aβ_1–40, at nanomolar concentrations similar to those present in AD) to human or rodent brain slices evoked capillary constriction [133] mediated by reactive oxygen species (ROS) generation and activation of endothelin A (ET_A) receptors (Fig. 3). It is plausible that this signalling pathway is also responsible for capillary constriction in the human AD brain, since the concentrations of both ROS and endothelin-1 are known to be elevated in human AD [10, 114, 135]. The locus of ROS generation is debated, with Park et al. [141] suggesting it to be perivascular macrophages, while Nortley et al. [133] found that ROS are generated by microglia and pericytes. ET_A receptors are known to be expressed on all classes of pericyte [190] and their activation in AD is consistent with the elevated level of extracellular endothelin-1 (ET) found in post-mortem AD brains [113, 135].

Release of inflammatory mediators generated during AD may also contribute to the decrease of CBF occurring. Interleukin-1β is generated when microglial and astrocyte inflammasomes are activated by oligomeric Aβ, and (in the context of ischaemia) this cytokine has been shown to decrease CBF by releasing ET [125], although it is unknown whether this decrease is generated by pericytes. Similarly, a mutation in the microglial TREM2 receptor (an AD susceptibility gene) that increases the production of inflammatory mediators also leads to a decrease of CBF [85]. The neuroinflammation occurring in multiple sclerosis can also be associated with hypoperfusion that is correctable by blocking ET_A receptors or voltage-gated calcium channels [33, 34].

Constrictions of rodent cerebral arterioles and middle cerebral artery, resulting in a decrease of cerebral blood flow, have been reported to be evoked by application of exogenous Aβ_1–40 [130, 169], but interestingly—at least in the APP^NL−G−F rodent model of AD—the level of Aβ that occurs in AD is sufficient only to constrict capillaries and not arterioles [133]. Nevertheless, in some AD mice, neurovascular coupling is impaired at the arteriole level [131]. Furthermore, changes in the properties of arteries and arterioles upstream of the brain’s capillary beds, and of the downstream venous system, could contribute to the onset of AD. Possible contributing changes include atherosclerosis [69, 182] leading to partial occlusion of large vessels, an increase in arterial stiffness [69] and hypertension [45, 72] (discussed below) resulting in microvascular damage. It is possible that, rather than directly reducing CBF, these changes may promote Aβ generation or reduce its clearance [45, 69].

The graded constriction of capillaries by pericytes is predicted to reduce CBF by 50% even in the absence of cells in the blood [133]. In addition, two mechanisms that can produce complete occlusion of vessels have been reported to reduce CBF in AD.

By imaging cell movements in cerebral capillaries, Cruz Hernández et al. [26] observed that in AD (APP/PS1), mice capillaries could become blocked by neutrophils (Fig. 2e). In the AD mice 1.8% of capillaries—predominantly of smaller diameter—became blocked, whereas in wild-type mice only 0.4% of capillaries were blocked. It will be important to reproduce these results in human AD patients. In wild-type mice, capillary block increases with ageing and can lead to vessels being pruned [159]. Remarkably, although modelling suggested that the increased block in AD would lead to a decrease of CBF of less than 5%, applying intraperitoneally a high concentration of an antibody to a neutrophil surface marker (Ly6G) led to a relief of capillary block, an increase of blood flow by 26–32% and improved memory. This is surprising because, at least in conditions of inflammation, antibody to Ly6G promotes neutrophil adhesion and aggregation, coagulation and decreased blood flow [132]. The large effect of the antibody on CBF compared with the modelling predictions for relief of capillary block alone may indicate either that the modelling is over-simplified or that the antibody has effects beyond simply preventing neutrophil blocking of capillaries, perhaps on the effective viscosity of the blood (which leukocytes significantly affect [2, 16