The etiology and prevention of early-stage tau pathology in higher cortical circuits: Insights from aging rhesus macaques

Abstract

Aging rhesus macaques provide a unique model for learning how age and inflammation drive early-stage pathology in sporadic Alzheimer's disease, and for testing potential therapeutics. Unlike mice, aging macaques have extensive association cortices and inflammatory signaling similar to humans, are apolipoprotein E ε4 homozygotes, and naturally develop tau and amyloid pathology with marked cognitive deficits. Importantly, monkeys provide the unique opportunity to study early-stage, soluble hyperphosphorylated tau (p-tau), including p-tau217. As soluble p-tau is rapidly dephosphorylated post mortem, it is not captured in human brains except with biopsy material. However, new macaque data show that soluble p-tau is toxic to neurons and capable of seeding across cortical circuits. Extensive evidence indicates that age-related inflammatory signaling contributes to calcium dysregulation, which drives tau hyperphosphorylation and amyloid beta generation. Pharmacological studies in aged macaques suggest that inhibiting inflammation and restoring calcium regulation can reduce tau hyperphosphorylation with minimal side effects, appropriate for potential preventive therapeutics. HIGHLIGHTS: Aging monkeys provide a unique window into early stage, soluble phosphorylated tau (p-tau). Inflammation with advancing age leads to calcium dysregulation, p-tau, and amyloid beta (Aβ). Macaque research shows p-tau undergoes transsynaptic seeding early in the cortex. p-tau traps amyloid precursor protein-containing endosomes, which may increase Aβ and drive vicious cycles. Restoring calcium regulation in cortex reduced p-tau217 levels in aged macaques.

Keywords: Alzheimer's disease; association cortex; calcium dysregulation; calpain; glutamate carboxypeptidase II; inflammation; phosphorylated tau; seeding; tau.

FIGURE 1

The hypothesized, temporal pattern of pathological events that impact the structure of tau in sporadic AD. A, The amino acid sequence for tau is comprised of a MTBR, which has an overall positive charge and consists of 3 or 4 repeats (3R or 4R) in humans and rhesus macaques. Adjacent to the MTBR is the PRR, which under healthy conditions is also positively charged and thus repelled away from the MTBR. B, With aging and under conditions of inflammation, tau is subject to a plethora of PTMs, including phosphorylation in the PRR, which changes the charge configuration of the PRR to negative charge, potentially “pulling” the MTBR and causing it to detach from microtubules. The sequence of tau hyperphosphorylation has been recently elucidated with high‐resolution quantitative proteomics and is associated with early‐stage phosphorylation at epitopes including T181, T231, S235, S202, and T217. These phosphorylation events are accelerated by cleavage of the C‐terminus by proteases such as caspases, which may further allow phosphorylation in that region., C, Early‐stage soluble p‐tau causes detachment and aggregation on microtubules and disrupts homeostatic processes such as endosomal trafficking. It is pertinent to note that soluble p‐tau is likely toxic, abrogating neuronal integrity and is rapidly dephosphorylated post mortem and is unlikely to be captured in human brain tissue, unless biopsy material is available. D, After detachment from the microtubules, soluble p‐tau can bind to LRP1, which facilitates its seeding between excitatory neurons, propagating tau pathology in a distinct interconnected circuit of neurons. During transsynaptic seeding events, early‐stage soluble p‐tau likely interfaces with CSF and blood plasma to be captured as a fluid‐based biomarker. E, Soluble p‐tau subsequently undergoes further PTMs such as acetylation and ubiquitination of the MTBR, and then stacks to form fibrils, creating the paired helical filaments which constitute the neurofibrillary tangles labeled by AT8 used to diagnose AD. AD, Alzheimer's disease; CSF, cerebrospinal fluid; LRP1, low‐density lipoprotein receptor‐related protein 1; MTBR, microtubule‐binding region; PRR, proline‐rich region; p‐tau, phosphorylated tau; PTM, post‐translational modification

FIGURE 2

The spatial and temporal pattern of tau pathology in the human sporadic AD brain, based on the work of del Tredici and Braak,, , , , , compared to that in the macaque. A, In humans, the earliest signs of fibrillated tau pathology in the cerebral cortex are observed in excitatory neurons of the TRE and then ERC cortices (NFT stages I–II, arrow in the lower portion of panel A points to the border between the two). Initially, the tau pathology is confined predominantly to layer pre‐α (i.e., layer II cell islands). B, Tau pathology then develops in the deeper layers of TRE and ERC, in the CA1 sector of the Ammon's horn (arrow in the lower portion of the middle panel points to the prosubiculum), and in the adjoining temporal association cortex (NFT stages III–IV). C, Tau pathology propagates to the secondary and then primary visual, auditory, somatosensory, and somatomotor cortices at the latest stages (NFT stages V–VI). D, Aging rhesus macaques recapitulate the same sequence of tau pathology as human AD with tau pathology first seen in layer II cell islands in ERC (33 years macaque), with stereotypical NFTs with AT8 labeling distinctly observed in extremely aged macaques (38 years macaque, inset). E, With advanced age, AT8 tau pathology is visualized dlPFC (30 years macaque), including the presence of extracellular deposits of amyloid plaques (Aβ) in rhesus macaques. F, Primary visual cortex is devoid of p‐tau (e.g., pS214Tau) immunolabeling in aged macaques (31 years macaque). Aβ, amyloid beta; AD, Alzheimer's disease; dlPFC, dorsolateral prefrontal cortex; ERC, entorhinal cortex; NFT, neurofibrillary tangle; TRE, transentorhinal region

FIGURE 3

Molecular signatures of vulnerability versus resilience in primate cortex. A, Schematic diagram highlighting how under healthy conditions, layer II ERC and layer III dlPFC postsynaptic compartments express the molecular machinery for feedforward cAMP–calcium signaling required for network connectivity necessary for higher‐order cognition. Feedforward calcium–cAMP signaling in layer II ERC and layer III dlPFC postsynaptic compartments is constrained by the phosphodiesterases (PDE4s), which hydrolyze cAMP, and by the Gi/o‐linked receptors mGluR3 (shown) and α2A‐AR (not shown), which are both predominantly localized on dendritic spines and inhibit the synthesis of cAMP. mGluR3 are stimulated by glutamate, but also by NAAG, which is co‐released with glutamate and is specific for mGluR3. mGluR3 are also expressed in their traditional location on astrocytes, where they promote glutamate uptake. Astrocytes also synthesize GCPII, an enzyme which catabolizes NAAG and reduces mGluR3 signaling. Under homeostatic healthy conditions, prominent NAAG stimulation of mGluR3 on dendritic spines, due to low levels of GCPII, enhances neuronal firing by inhibiting cAMP–PKA opening of K+ channels. In contrast, in classical glutamate synapses (e.g., primary visual cortex V1), neurotransmission depends on AMPA receptors, which provide permissive actions by depolarizing the postsynaptic membrane to eject Mg²⁺ from the NMDAR pore, thereby permitting NMDAR neurotransmission. The calcium entry through NMDAR can drive cAMP–PKA signaling to increase neuroplasticity and strengthen connections. cAMP is catabolized by PDE4s, which reduce memory formation. B, Under conditions of inflammation/advancing age, glia increase their expression of GCPII, which catabolizes NAAG and decreases mGluR3 regulation of cAMP–calcium signaling in layer II ERC and layer III dlPFC postsynaptic compartments. Aging/inflammation also reduce the expression of PDE4A/D and calbindin (layer II ERC might be devoid or have very sparse calbindin expression even in young age) within postsynaptic compartments, resulting in raised levels of cAMP–PKA–calcium signaling, including PKA‐mediated phosphorylation of ryanodine receptors to drive calcium “leak” from the SER. Excessive cAMP–PKA–calcium signaling opens K⁺ channels, which reduce neuronal firing, and high levels of cytosolic calcium activate calpain‐2, which cleave and activate kinases such as GSK3β to hyperphosphorylate tau, for example, at p‐tau217. cAMP, cyclic adenosine monophosphate; dlPFC, dorsolateral prefrontal cortex; ERC, entorhinal cortex; GCPII, glutamate‐carboxypeptidase II; GSK3β, glycogen synthase kinase‐3 beta; mGluR3, metabrotopic glutamate receptor 3; NAAG, N‐acetyl‐aspartyl‐glutamate; NMDAR, N‐methyl‐D‐aspartate; PKA, protein kinase A; p‐tau, phosphorylated tau; SER, smooth endoplasmic reticulum

FIGURE 4

Calcium dysregulation is associated with tau hyperphosphorylation in the aging macaque dlPFC. A, The calcium buffering protein, calbindin, is expressed in pyramidal cells as well as interneurons in layer III of macaque dlPFC. B and C, Biochemical assays show a loss of calbindin with advancing age from macaque dlPFC. The asterisk denotes an exceptional aged monkey who retained high levels of calbindin and is shown by a green rectangle in (C). D, There was a high correlation between calcium dysregulation, as assayed by pS2808‐RyR2 levels, and p‐tau levels in aged macaque dlPFC. Note the animal with preserved calbindin (green rectangle) had no p‐tau expression. Data in (B—D) from Datta et al. dlPFC, dorsolateral prefrontal cortex; p‐tau, phosphorylated tau

FIGURE 5

A, Schematic illustration of how calcium dysregulation in the aging association cortex leads to activation of calpain‐2 and an exacerbation in sporadic AD pathology. Calcium regulatory mechanisms that constrain feedforward cAMP–calcium signaling, for example, PDE4, calbindin, mGluR3, are lost with age and/or inflammation (shown in gray). Activated calpain‐2 leads to tau hyperphosphorylation by disinhibiting GSK3β and cdk5. Disinhibited cdk5 also increases β‐secretase production of Aβ. Calpain‐2 cleavage of hsp70.1 leads to cathepsin‐induced cellular necrosis, identical to the pattern by which neurons degenerate in sporadic AD, as well as increased fibrillation of p‐tau. Furthermore, activated calpain mediates synapse loss through activation of the p38‐MK2‐LIMK1 signaling pathway and cleavage of ROCK1. As APOE ε4 exacerbates calcium dysregulation, and rhesus monkeys are all APOE ε4 homozygotes, they may be particularly susceptible to sporadic AD‐like neuropathology. B, The rhesus monkey model provides the opportunity to test treatments that restore calcium regulation that act upstream of Aβ and tau pathology, and that may be especially involved in the etiology of sporadic AD. This includes: (1) 2‐MPPA and 2‐PMPA that reduce GCPII expression to potentiate mGluR3‐mediated regulation of calcium signaling within postsynaptic compartments to attenuate tau pathology, including p‐tau217 in brain and blood plasma; (2) S107, which stabilizes pS2808‐RyR2 with downstream effector protein calstabin‐2 to prevent excess calcium “leak” from the SER to the cytosol; (3) guanfacine, a selective noradrenergic α2A‐AR agonist that can strengthen network connections and improve cognitive functioning by regulating cAMP–calcium signaling; (4) NA‐184, a highly selective calpain‐2 inhibitor, which is currently under investigation in aged rhesus macaques as a potential therapy. Our hypothesis posits that administration of NA‐184 early in the aging process will prevent tau and amyloid pathology prior to irreversible neuronal degeneration. AD, Alzheimer's disease; Aβ, amyloid beta; α2A‐AR, α2A‐adrenoceptor; APOE, apolipoprotein E; cAMP, cyclic adenosine monophosphate; cdk5, cyclin‐dependent kinase 5; GCPII, glutamate‐carboxypeptidase II; GSK3β, glycogen synthase kinase‐3 beta; mGluR3, metabrotopic glutamate receptor 3; PDE4, phosphodiesterase 4; p‐tau, phosphorylated tau

FIGURE 6

Sequence of tau pathology in macaque dlPFC and ERC dendrites. A, Under healthy conditions, tau plays a crucial role in regulating microtubule dynamics and providing anterograde and retrograde transport of cargo within cells via endosomal trafficking. B, PTMs, including early‐stage phosphorylation, results in soluble p‐tau that induces detachment from microtubules and aggregation. ImmunoEM in the lower panel shows p‐tau217 aggregating on dendritic microtubules “trapping” of endosomes. C, Nanoscale examination in aging rhesus monkeys reveal a mechanism by which phosphorylated tau may aggravate Aβ generation within neurons, similar to how genetic predispositions in retromer (e.g., SORL1) signaling pathways may increase the risk of sporadic AD by triggering “endosomal traffic jams,” increasing the time APP spends in endosomes where it is cleaved to Aβ. Using dual‐label immunoEM (lower panel), our data show evidence of p‐tau217 surrounding endosomes containing Aβ42 in dendrites in aged macaque dlPFC, suggesting that p‐tau217 may be involved in the etiology of amyloid pathology. D, As tau becomes hyperphosphorylated, dendritic soluble p‐tau is associated with hallmarks of neurodegeneration, including autophagic vacuolar degeneration, and abnormal mitochondria, that is, MOAS profiles. Aggregations of p‐tau217 (lower panel) are often seen in dendrites containing autophagic vacuoles with multilamellar bodies (pseudocolored in orange) and near dysmorphic mitochondria. E, With later‐stage fibrillated tau pathology, the neuronal dendrite and soma are associated with extensive autophagic vacuoles, where normal neuronal organelles are now lost, and the cellular engine for Aβ production has deteriorated. A degenerating dendrite (lower panel) from an AT8‐labeled neuron in aged monkey layer II ERC. The dendrite is devoid of normal organelles and is filled with autophagic vacuoles similar to those in sporadic AD. AD, Alzheimer's disease; APP, amyloid precursor protein; Aβ, amyloid beta; dlPFC, dorsolateral prefrontal cortex; ERC, entorhinal cortex; MOAS, mitochondria‐on‐a‐string; p‐tau, phosphorylated tau; PTM, post‐translational modification

FIGURE 7

A cartoon depicting interacting vicious cycles in sporadic AD in which aggregations of soluble p‐tau on microtubules in dendrites drives Aβ production via “trapping” of endosomes. Aβ in turn drives more tau phosphorylation, and both p‐tau and Aβ drive further calcium dysregulation. These interactive vicious cycles may be initiated by inflammation driving calcium dysregulation in sporadic AD. AD, Alzheimer's disease; Aβ, amyloid beta; p‐tau, phosphorylated tau

FIGURE 8

A, Hypothetical mechanisms by which inhibition of GCPII inflammatory signaling with 2‐MPPA would restore mGluR3 regulation of cAMP–calcium signaling and attenuate tau hyperphosphorylation in aged macaque cortex. B, The levels of GCPII activity correlated highly with levels of p‐tau217 in aged macaque dlPFC (r = 0.96, p = 0.0019); blue = 2‐MPPA‐treated monkeys; red = vehicle‐treated monkeys. C, Chronic daily treatment with the GCPII inhibitor, 2‐MPPA, was associated with reduced levels of p‐tau217 in the brain. Aged macaques receiving 2‐MPPA treatment had significantly lower levels of p‐tau217 in dlPFC compared to aged macaques treated with vehicle. (*P = 0.016; n = 4/group). p‐tau217 levels were assayed by western blot and are expressed as a ratio of GAPDH expression. D, Aged macaques receiving 2‐MPPA treatment also had significantly lower levels of p‐tau217 in ERC compared to aged macaques treated with vehicle (*P = 0.04; n = 4/group). E, Plasma levels of p‐tau217 measured by nanoneedle assays were reduced from baseline levels after 6 months of daily treatment. with 2‐MPPA in three aged macaques; the fourth animal with the lowest levels at baseline was unchanged after treatment, resulting in an overall trend level reduction from baseline in the group (P = 0.1; n = 4/group). Data in (B—D) are from Bathla et al. 2‐MMPA, 2‐(3‐mercaptopropyl) pentanedioic acid; cAMP, cyclic adenosine monophosphate; dlPFC, dorsolateral prefrontal cortex; ERC, entorhinal cortex; GAPDH, glyceraldehyde‐3‐phosphate dehydrogenase; GCPII, glutamate‐carboxypeptidase II; mGluR3, metabrotopic glutamate receptor 3; p‐tau, phosphorylated tau