|Alzheimer's Disease: The Role of Tau's
Phosphatase-Activating Domain in Pathology Spreading
It has been well established that the Tau pathology propagates along anatomically connected neural networks (Liu et al., 2012), but its underlying mechanism remains elusive. The prion-like hypothesis posits that the pathology spreading is mediated by Tau aggregates which are first released from affected neurons to the extracellular space and then taken up by adjacent cells (Frost and Diamond, 2010). This hypothesis is supported by two findings: (1) extracellular short Tau fibrils, but not long filaments, can be internalized (Wu et al., 2013), and (2) injection of Tau fibrils into a mouse brain region leads to Tau pathology in anatomically connected regions (Guo et al., 2016). However, there is so far no convincing evidence that Tau aggregates can be secreted from a viable neuron (Hasegawa, 2016; Goedert and Spillantini, 2017).
Understanding the mechanism of pathology spreading is essential for guiding therapeutic strategies. If spreading is mediated by Tau aggregates, then intervention of Tau aggregation should produce tremendous beneficial results. Tau aggregation inhibitors (TAIs) were developed on the basis of this assumption. One of them, leuco-methylthioninium bis (LMTM), has managed to enter phase 3 clinical trial, but "the results do not suggest benefit of LMTM as an add-on treatment for patients with mild to moderate Alzheimer's disease" (Gauthier et al., 2016). On the other hand, there is evidence indicating that extracellular monomeric Tau protein is sufficient to initiate the spread of Tau pathology. Not only can monomeric Tau be internalized (Wauters et al., 2016), both full-length and truncated Tau may be released from neurons. The secretion of full-length Tau is facilitated by phosphorylation, while the truncated Tau must contain the N-terminal half (Kim et al., 2010; Plouffe et al., 2012). Monomeric Tau is released from presynaptic terminals (Sokolow et al., 2015), consistent with the synaptically connected spreading pattern. Furthermore, the secretion of monomeric Tau increases with increasing neural activity (Pooler et al., 2013; Yamada et al., 2014; Wu et al., 2016). That is, hyperexcitability should promote Tau release into the extracellular space (Mohamed et al., 2017).
According to the BDNF Cascade Hypothesis presented in previous papers (summarized in Paper 9), Alzheimer's disease (AD) is fundamentally caused by Ca2+ overload, which in turn arises from hyperexcitability. Therefore, the pathology spreading between neurons could result from "hyperexcitability transfer" mediated by certain substance. This paper will present evidence that the mediator is likely to be the pathological Tau species (phosphorylated and/or truncated) whose "phosphatase-activating domain" (PAD) is exposed. They have the propensity to form aggregates. However, the aggregation state is not required for pathology spreading. The key factor is "PAD exposure" (Combs and Kanaan, 2017). As long as the PAD is exposed, Tau or its fragment may spread pathology, whether it is monomeric or in an aggregate.
The Function of PAD
PAD is located in the N-terminal region of Tau, comprising amino acids 2 -18. As the name implies, PAD has the capacity to activate phosphatase. In a seminal work, Kanaan et al. (2011) demonstrated that PAD could activate protein phosphatase 1 (PP1), which in turn activates glycogen synthase kinase-3 (GSK-3). GSK-3 has two subtypes: GSK-3α and GSK-3β. They are constitutively active, but can be inactivated through the phosphorylation of a single residue: serine 21 for GSK-3α and serine 9 for GSK-3β. Dephosphorylation of serine 21 or 9 by PP1 will activate GSK-3. Alternatively, calpain may remove the N-terminal regulatory domain, rendering GSK-3 persistently active (Goñi-Oliver et al., 2007). GSK-3β is a major Tau kinase, capable of phosphorylating Tau at multiple sites (Wang et al., 2007). In addition to this well known function, the activation of GSK-3 also enhances neuronal excitability (see GSK-3, Valproic Acid and Epilepsy).
In a normal free Tau, both N and C termini fold back so that they are in close proximity (Jeganathan et al., 2006). This structure is known as the "paper-clip conformation" (Figure 1), where the C-terminal domain prevents PAD from interacting with PP1. When the paper-clip conformation is opened, PAD could become accessible to PP1. The resulting Tau phosphorylation and hyperexcitability by aberrant GSK-3 activation may have detrimental effects on neuronal functions. It has been shown that PAD is exposed in early pre-tangle Tau aggregates, but not in late neurofibrillary tangles (NFTs) (Combs et al., 2016), in agreement with the findings that Tau oligomers, not NFTs, are the true toxic species (Shafiei et al., 2017).
There are two mechanisms that can open up Tau to expose PAD: phosphorylation and truncation. Jeganathan et al. (2008) showed that pseudo-phosphorylation at the epitope of the antibody AT8 (S199E + S202E + T205E) moves the N-terminal domain away from the C-terminal domain. Kanaan et al. (2011) provided further evidence that the AT8 pseudo-phosphorylated Tau activates the PP1/GSK-3 cascade.
TauC3 and Other Truncated Tau
"TauC3" refers to the N-terminal fragment of Tau after cleavage by caspase-3 (C3) at D421. Thus, it contains 421 amino acids if cleaved from the longest Tau isoform (with 441 residues). TauC3 co-localizes with NFTs in both AD and vascular dementia (Day et al., 2015). It may undergo further cleavage at E391. Accumulation of D421- and E391-cleaved Tau in NFTs has been shown to correlate with AD progression (Basurto-Islas et al., 2008). This demonstrates the importance of Tau N-terminal fragments in pathology spreading. Further evidence comes from the studies on high-molecular-weight (HMW) Tau (500 - 1000 kDa) which contains various truncated and phosphorylated forms of Tau, including TauC3 (Nicholls et al., 2017). HMW Tau is soluble in phosphate-buffered saline (PBS), distinct from insoluble Tau aggregates. The HMW Tau species accumulate in the cerebrospinal fluid of AD brains. They can be taken up by neurons and induce intracellular aggregates (Takeda et al., 2015; Takeda et al., 2016). This "seed-competent" property may explain the formation of Tau aggregates although aggregation is not required for pathology spreading. TauC3 and other N-terminal fragments are likely to exert neurotoxicity via PAD exposure.
When the C-terminal domain is truncated, as in N-terminal fragments, the PAD becomes more exposed, which could result in elevated excitability by activating the PP1/GSK-3 pathway. Remarkably, in a study of Bright et al. (2015), the N-terminal fragments of Tau did cause neuronal hyperactivity. This study also revealed that the hyperactivity increases amyloid-beta (Aβ) production, supporting the notion that Aβ is a by-product of Tau pathology.
The Effects of PAD Exposure on BDNF Signaling
The PAD exposure of Tau can influence not only the amount of BDNF-containing synaptic vesicles, but also the BDNF release into synaptic cleft. After BDNF is produced in the cell body, they are packaged into vesicles and transported to presynaptic axon terminals (Andreska et al., 2014). PAD Exposure has been demonstrated to impair anterograde fast axonal transport (FAT) (Kanaan et al., 2011). Therefore, the number of BDNF vesicles could be reduced by PAD exposure. In addition, there is a report that the PAD exposure reduces neurotransmitter release (Moreno et al., 2016), possibly arising from the binding between phosphorylated Tau and vesicles (Moreno et al., 2011; Zhou et al., 2017). These reports are consistent with the observation that Tau hyperphosphorylation resulting from treatment of okadaic acid leads to BDNF reduction (Atasoy et al., 2017).
The released BDNF may bind with its receptor, TrkB, on the postsynaptic membrane and stimulate miR-132 production in the postsynaptic neuron. miR-132 regulates the expression of total Tau and the ratio of 4-repeat (4R) and 3-repeat (3R) Tau. Its deficiency has been shown to increase total Tau, especially the 4R isoform, which in turn leads to hyperexcitability (Paper 4). Hence, the decreased BDNF vesicles and impaired neurotransmitter release in the presynapse will attenuate BDNF-TrkB signaling in the postsynaptic neuron, resulting in hyperexcitability.
The Mechanism of Hyperexcitability Transfer
Figure 2 summarizes the mechanism of hyperexcitability transfer from an originally affected neuron to a synaptically connected neuron. In AD, the neurons in entorhinal cortex (EC) were affected first, possibly because (1) they are intrinsically very excitable (Alonso and Klink, 1993), and (2) they are enriched with Tau proteins (Shukla and Bridges, 1999). Thus, EC is vulnerable to the risk factors that can increase excitability further (Paper 5 and Paper 9). The hyperactive neuron may cause Ca2+ overload, triggering two main toxic pathways: calpain activation and mitochondrial dysfunction.
Calpain has the capacity to directly activate GSK-3 and Cdk5 (Paper 6) which are the major Tau kinases capable of phosphorylating Tau at multiple sites, including AT8 epitope: S199, S202 and T205 (Wang et al., 2007). Tau phosphorylation at AT8 sites causes the N-terminal domain to move away from the C-terminal domain (Jeganathan et al., 2008), resulting in PAD exposure (Kanaan et al., 2011). The PAD exposure of Tau in the presynaptic neuron (e.g., EC neuron) may attenuate BDNF-TrkB signaling in the postsynaptic neuron (e.g., hippocampal neuron) through two mechanisms: (1) impaired fast axonal transport (FAT) (Kanaan et al., 2011) leading to decrease of BDNF-containing synaptic vesicles, and (2) reduced BDNF release due to Tau-vesicle binding (Moreno et al., 2016). Hence, calpain activation in the presynaptic neuron may cause hyperexcitability in the postsynaptic neuron via Tau phosphorylation and the BDNF/miR-132/4R-Tau axis (Paper 4).
The oxidative stress resulting from mitochondrial dysfunction can induce a variety of damaging processes, including Aβ production (Ganguly et al., 2017) and caspase-3 activation. Generally, caspase-3 can be activated by two different pathways: intrinsic and extrinsic. In the extrinsic pathway, the extracellular signals stimulate death receptors on the cell membrane, resulting in the activation of caspase-8, which then activate caspase-3. In the intrinsic pathway (also called mitochondrial pathway), the oxidative stress stimulates the release of cytochrome c from mitochondria into cytosol, leading to the activation of caspase-9, which can also activate caspase-3 (Snigdha et al., 2012). Low level of Ca2+ ions through NMDA receptors may cause moderate increase in caspase-3 activity, which plays an important role in long term depression (Li et al., 2010). However, Ca2+ overload may increase caspase-3 activity dramatically, resulting in the production of TauC3 and other N-terminal fragments. The phosphorylated or truncated monomeric Tau can be secreted into the extracellular space and directly taken up by connected neurons. They could also be internalized in the form of HMW Tau, where PAD is still exposed. As described above, the PAD exposure in either monomeric or HMW Tau can cause hyperexcitability through the PP1/GSK-3 axis.
The above mechanism of pathology spreading suggests that PAD should be a promising therapeutic target. Very recently, Dai et al. (2017) demonstrated that the antibody against PAD inhibited not only Tau but also Aβ pathology. Another group found that the PAD antibody inhibited cytotoxic effect of Tau oligomers (Agadjanyan et al., 2017). These studies were performed in animals and cells. In the past three decades, numerous chemical compounds exhibited beneficial effects in animal and cellular models, but eventually failed in clinical trials for the treatment of AD. However, these compounds targeted Aβ pathology and Tau aggregation. The PAD antibody is entirely different. It has chance to hit the bull's-eye.
Author: Frank Lee