Dysregulation of mTOR by Tau in Alzheimer’s Disease

Abstract

Tau was discovered in the mid 1970’s as a microtubule-associated protein that stimulates tubulin polymerization, and subsequently was shown to be expressed primarily in neurons, where it is most concentrated in axons. Interest in tau rose by the late 1980’s, when it was shown to be the principal subunit of the neurofibrillary tangles (NFTs) that accumulate in Alzheimer’s disease (AD) brain, and achieved new heights by the late 1990’s, when numerous tau mutations were found to be highly penetrant for AD-related disorders that also are associated with NFTs and came to be known as non-Alzheimer’s tauopathies. The role of tau in neurodegeneration is far more complex than whatever effects on neurons may be caused by NFTs, however, and here we review our work on dysregulation of mTOR by tau in AD. mTOR is a protein kinase and master regulator of myriad aspects of cellular behavior. We have defined a complex signaling network whereby aberrant tau phosphorylation provoked by amyloid-β oligomers (AβOs), the building blocks of the amyloid plaques that from in AD brain, cause post-mitotic neurons to re-enter the cell cycle, but to die eventually instead of dividing, which may account for most neuron death in AD. Remarkably, we found that this same neuronal signaling network also poisons a fundamental cell biological process that we discovered, nutrient-induced mitochondrial activation, or NiMA. Tau-dependent cell cycle re-entry and NiMA inhibition occur in cultured neurons within a few hours of exposure to AβOs, and thus may represent seminal processes in AD pathogenesis.

1 |. mTOR dysregulation by tau

If not for its involvement in Alzheimer’s disease (AD) and other tauopathies, tau might be regarded as just another microtubule (MT) associated protein that excites only the most dedicated MT afficionados. How, then does tau promote the development of neurodegenerative diseases? One possible answer derives from the large body of evidence that AD-like tau phosphorylation can reduce tau’s affinity for MTs (Alonso et al., 1994; Alonso et al., 2010; Biernat et al., 1993; Fath et al., 2002; Rajbanshi et al., 2023). While there is no reason to doubt that aberrant tau phosphorylation contributes to neurodegeneration, tau’s role in AD, and probably in other tauopathies as well, extends far beyond its diminished MT binding and any consequent effects on MTs themselves. One especially important role for tau in AD is dysregulation of the mechanistic target of rapamycin, or mTOR, although how that might relate to reduced MT binding by tau remains to be determined.

mTOR is a serine-threonine protein kinase that is a subunit of two distinct, multimeric complexes, mTORC1 and mTORC2, which together are master regulators of cellular behavior. By sensing the extracellular environment for nutrients, like glucose and amino acids, and trophic factors, like insulin and growth factors, the mTOR complexes control a multitude of cellular responses, including but not limited to autophagy, cell proliferation and survival, mRNA translation and lipid biogenesis (Kim et al., 2019; Zoncu et al., 2011).

One of the first clues that mTOR is important for AD was the finding by Varvel and colleagues that rapamycin, which inhibits mTORC1 directly and mTORC2 indirectly, blocks re-entry into the cell cycle of cultured primary neurons exposed to extracellular amyloid-β oligomers (AβOs) (Varvel et al., 2008), the building blocks of the extracellular amyloid plaques that are a defining feature of AD brain. In humans, postnatal neurogenesis is important for memory consolidation, but normally occurs to a very limited degree, primarily in the dentate gyrus of the hippocampal formation (Toda et al., 2019). Nearly all neurons in a healthy adult brain are therefore permanently stuck in G0 of the cell cycle. In AD, however, large numbers of neurons in cortical brain regions display molecular markers of being in G1 or beyond (McShea et al., 1999; Seward et al., 2013; Yang et al., 2003), but instead of dividing, they apparently die eventually. In fact, ~30% of the brain’s neurons are lost by the end of typical AD cases, with up to 90% of that neuron death following ectopic cell cycle re-entry (CCR) (Arendt et al., 2010). Although a mechanistic link from CCR to neuron death in AD awaits formal proof, the ability of AβOs to induce rapamycin-sensitive CCR of cultured neurons suggests that AβOs are also responsible for neuronal CCR and subsequent neuron death in vivo in AD by an mTOR-dependent mechanism.

Motivated by the study of Varvel and colleagues, our lab decided to explore further details of the CCR mechanism. We quickly discovered a cornerstone feature of this process: its dependence on tau, and furthermore on tau phosphorylation at a minimum of three sites - Y18, S409 and S416 - catalyzed respectively by fyn, protein kinase A and CaMKII (Seward et al., 2013). We also confirmed Varvel’s prior report that CCR can be blocked by rapamycin, while ruling out stimulation of insulin receptors by AβOs as a possible mechanism (Seward et al., 2013). The role of mTOR in promoting neuronal CCR was thus confirmed, but the question of which mTOR complex(es) and how remained unanswered.

Building on those observations, we next found that AβOs stimulate neuronal mTORC1 kinase activity, but in an unusual way. Activation of mTORC1 by nutrients, insulin or growth factors typically occurs at the surface of lysosomes, which in turn inhibits autophagy and sets in motion a variety of other downstream effects (Kim et al., 2019; Zoncu et al., 2011). In contrast, we found that AβOs activate mTORC1 at the plasma membrane, but not at lysosomes, by a mechanism dependent on intraneuronal tau and its rapamycin-sensitive phosphorylation at S262 (Norambuena et al., 2017). This AβO-stimulated activation of mTORC1 at the plasma membrane is normally blocked in primary neurons derived from tau knockout mice, but occurs robustly in such cells following lentiviral driven expression of wild type tau, but not of tau with a non-phosphorylatable S262A mutation. This toxic interplay between mTORC1 and tau may therefore represent a positive feedback loop whereby tau phosphorylation at S262 promotes activation of plasma membrane mTORC1, which in turn stimulates production of more tau_pS262, as illustrated in Figure 1 (Norambuena et al., 2017). Importantly, this pathogenic connection between Aβ and tau happens independently of their respective incorporation into amyloid plaques and neurofibrillary tangles. It remains to be determined if the neuronal CCR pathway described here also applies to non-Alzheimer’s tauopathies, in which perturbants other than AβOs probably promote aberrant tau phosphorylation.

FIGURE 1.

Induction of neuronal cell cycle re-entry (CCR) and inhibition of nutrient-induced mitochondrial activity (NiMA) by tau in Alzheimer’s disease (AD). Extracellular amyloid-β oligomers (AβOs) induce tau phosphorylation at multiple sites, most notably in the current context at Y18, S409 and S416, catalyzed respectively by fyn, protein kinase A (PKA) and calcium-calmodulin activated protein kinase II (CaMKII) (Seward et al., 2013). AβOs also stimulate mTORC1-dependent tau phosphorylation at S262, but probably indirectly (Norambuena et al., 2017). Together these tau phosphorylations are required for mTORC1 activation at the plasma membrane, which leads to CCR and eventual neuron death. NiMA is initiated by trophic factors, like insulin, and nutrients, like amino acids, which activate lysosomal mTORC1 (Norambuena et al., 2018). The activated mTORC1 then phosphorylates SOD1, which coincidently is a major pathogenic factor for amyotrophic lateral sclerosis (ALS), as a necessary step for mitochondrial activation (Norambuena et al., 2022). Remarkably, the AβO-induced, tau-dependent CCR pathway inhibits NiMA (Norambuena et al., 2018). Also noteworthy is the CCR pathway requirement for α-synuclein, a pathological factor in Parkinson’s disease, Lewy body dementia and AD (Khan et al., 2018). Additional proteins shown in the figure include NMDAR (N-methyl-D-aspartate receptor); NCAM (neural cell adhesion molecule); G_αs (subunit of the heterotrimeric G protein, G_s, that activates adenylyl cyclase); and the small GTPase, Rac1.

The same study also yielded evidence that shRNA-mediated knockdown of rictor, an essential subunit of the mTORC2 complex, prevents AβO-induced CCR in cultured neurons, prompting us to conclude at that time that mTORC2 is also critical for CCR (Norambuena et al., 2017). Our subsequent unpublished efforts to define mTORC2’s role in CCR more specifically eliminated mTORC2 kinase activity as a factor, however, indicating that the effects of rictor knockdown reflected one or more of that protein’s mTOR-independent activities (Agarwal et al., 2013; Hagan et al., 2008; McDonald et al., 2008). We therefore now believe that the ability of rapamycin to block AβO-induced, tau-dependent neuronal CCR signifies a requirement for mTORC1, but that mTORC2 is not involved.

Curiously, at least 37 protein kinases are known to phosphorylate tau, but mTOR is not among them (https://bit.ly/2JyZTbS). We thus suspect that AβO-stimulated tau phosphorylation at S262 is catalyzed by a different enzyme, p70/S6 kinase, which is known to phosphorylate tau at that site (https://bit.ly/2JyZTbS) and itself is phospho-activated by mTORC1 (Harrington et al., 2004; Um et al., 2004). Regardless, our results revealed an AβO-triggered, bi-directional pathway, whereby mTORC1-dependent tau phosphorylation at S262 is required for pathogenic activation of mTORC1 at the plasma membrane and neuronal CCR (Norambuena et al., 2017).

Much remains to be learned about the neuronal CCR pathway in AD, but a more recent study from our lab uncovered a previously unknown, fundamental cell biological phenomenon common to many cell types, nutrient-induced mitochondrial activity, or NiMA, that is inhibited in neurons by a mechanism that closely parallels the neuronal CCR pathway in AD (Norambuena et al., 2018). NiMA involves activation of lysosomal mTORC1 by nutrients or insulin (Norambuena et al., 2022), which leads to a still mysterious lysosome-to-mitochondrion signaling process manifest as upregulation of mitochondrial activity and downregulation of mitochondrial DNA synthesis (Norambuena et al., 2022). Remarkably, NiMA is inhibited in neurons exposed to AβOs in a tau-dependent manner because of the aforementioned activation of plasma membrane mTORC1. In contrast to neuronal CCR, however, NiMA can be overridden by simultaneous activation of lysosomal mTORC1 (Norambuena et al., 2018).

It is important to note that both neuronal CCR and NiMA inhibition occur in cultured neurons within hours of their initial exposure to AβOs. By extension, these closely intertwined phenomena may represent seminal tau-dependent processes in AD pathogenesis. Consistent with that idea are reports that mTOR haploinsufficiency in AD model mice (Caccamo et al., 2014) or their long term treatment with rapamycin (Spilman et al., 2010) reduces AD-like histopathology and cognitive decline.

Tau’s ability to facilitate pathological signaling by AβOs, including disruption of NiMA, exemplifies the role of tau in neurodegeneration. In a mature neuron, tau is most concentrated in axons and synaptic terminals, and is sparser in the soma, where the nucleus is located, and endosomes, lysosomes and mitochondria are most highly concentrated. Mitochondria not only deliver ATP and other key metabolites needed to support cellular functions in regions of high energy demand, but also are the main source of reactive oxygen species (ROS), which serve as second messengers, but can be harmful when in excess. It possible that the normally low concentration of tau in the soma represents an evolutionary advantage by establishing a perinuclear environment in which mitochondrial functioning and ROS production are maintained in balance, and excessive production of ROS and oxidative damage is avoided in a region crowded with organelles. It follows naturally that abnormally high levels of somatic tau, as occurs in AD neurons, could cause mitochondrial dysfunction and oxidative damage.

2 |. Summary

The unique susceptibility of neurons to functional impairment and cell death in AD reflects, at least in part, a pathogenic signaling network from AβOs to the neuron-specific protein, tau. Two net effects of the AβO-tau connection are induction of CCR, which precedes most neuron death in AD, and mitochondrial poisoning as a consequence of NiMA inhibition. A graphical summary of what we currently know about how tau promotes CCR and inhibits NiMA in neurons is shown in Figure 1. An especially noteworthy feature of this signaling network is that its constituent proteins include not only two of the most notorious factors germane to AD, Aβ and tau, but in addition α-synuclein, which forms Lewy bodies in Parkinson’s disease, Lewy body dementia and frequently in AD itself, and SOD1, a major pathogenic factor for amyotrophic lateral sclerosis (ALS). The involvement of these signature proteins for neurodegeneration in a single, albeit complex signaling network, may symbolize the mechanistic relatedness of the individual diseases for which each of the proteins is best known. As summarized in Figure 1, tau may serve as a glue that ties together this pathogenic signaling network, at least for AD.