Is tau ready for admission to the prion club?

Abstract

Aggregation-prone proteins associated with neurodegenerative disease, such as α synuclein and β amyloid, now appear to share key prion-like features with mammalian prion protein, such as the ability to recruit normal proteins to aggregates and to translocate between neurons. These features may shed light on the genesis of stereotyped lesion development patterns in conditions such as Alzheimer disease and Lewy Body dementia. We discuss the qualifications of tau protein as a possible “prionoid” mediator of lesion spread based on recent characterizations of the secretion, uptake and transneuronal transfer of human tau isoforms in a variety of tauopathy models, and in human patients. In particular, we consider (1) the possibility that prionoid behavior of misprocessed tau in neurodegenerative disease may involve other aggregation-prone proteins, including PrP itself, and (2) whether “prionlike” tau lesion propagation might include mechanisms other than protein-protein templating.

Keywords: :

• Prion Hypothesis
• interneuronal lesion spread
• neurodegeneration
• prion
• prionoid
• tau
• templating

Prions and TSEs as a General Model for Neurodegenerative Disease

The formation of abnormal protein aggregates provides a common element in the pathogenesis of the vast majority of neurodegenerative diseases in humans, including Alzheimer Disease (AD), non-AD tauopathies, α synucleinopathies such as Lewy Body dementia, transmissible spongiform encephalopathies (TSEs), trinucleotide repeat diseases and amyotrophic lateral sclerosis. Since aggregate formation is generally considered (and has been modeled) as a cell-autonomous process, investigations into the pathogenesis of most neurodegenerative diseases have until very recently focused on cytotoxicity mechanisms associated with aggregate formation. As a consequence, intercellular aspects of most neurodegenerative diseases have been relatively neglected until very recently, and remain poorly understood. An exception to this pattern has been the study of TSEs, in which the consequences of prion protein (PrP) misfolding were addressed at the organismal level years before meaningful studies of cellular mechanisms could be attempted. This is because the transmissability and bizarre epidemiology of these diseases made a basic understanding of the transmission mechanism prerequisite to any meaningful studies of TSE cytopathogenesis. However, during the 1980s and 1990s, PrP was progressively established as the necessary and sufficient agent for TSE transmission via the promulgation, systematic testing and validation of the “Prion Hypothesis” ^1-3. Since then, the study of PrP biology and pathobiology at the molecular, cellular and organismal levels has led to a deepening appreciation of the subtlety with which specific misfolded conformers of PrP can reproduce and propagate disease-specific properties that define specific TSEs. Investigators have recently begun to use PrP propagation in TSEs as a model to address interneuronal aspects of non-prion neurodegenerative diseases such as AD, non-AD tauopathies and Lewy Body disease by focusing on two intriguing features common to these conditions: (1) the sequential involvement of synaptically connected parts of the brain with disease progression^4-7 and (2) synergistic interactions between α synuclein (Asyn), β amyloid (Abeta), tau and PrP, both in cellular disease models^8,9 and in the distribution of plaques and neurofibrillary lesions in TSE, Lewy Body disease and AD neuropathology.^10-12 This has raised the question of whether aggregation-prone proteins such as Asyn, Abeta and most recently tau,¹³ that have established roles in neurodegenerative disease cytopathogenesis, may also participate in interneuronal lesion propagation.¹⁴

The prion-like characteristics of Asyn and Abeta have been reviewed extensively elsewhere^13,15-18; this review will focus on the qualifications of tau as a prion—especially in the context of recent findings from our lab and elsewhere suggesting that tau secretion and interneuronal transfer may play a material role in the biogenesis of CSF tau and lesion propagation in AD. Here we will consider whether recent studies demonstrating tau secretion,^19-23 uptake,^19,24 interneuronal transfer,^19,25,26 toxicity,²⁷ and the modulation of these processes by tau aggregation state^24-27 provide sufficient reason to add tau to the growing list of “prionoid” proteins. One of the more interesting consequences of considering the “prionoid” characteristics of tau is that the multiplicity of toxic pathways exhibited by tau highlights the need for re-evaluating basic definitions of “prionoid” and “prionlike,” as is discussed further below.

Membership in the Prion Club is Still Limited to its Founder – PrP

It is important to start by noting that the Prion Club still has only one undisputed member according to the strict “classic” membership criteria laid out 30 y ago in the Prion Hypothesis. This exclusivity is largely due to the controversial nature of protein-only disease transmission at the time and the intense scrutiny that it has received since. The “Prion Hypothesis” of protein–only TSE transmission proposed by Prusiner on the strength of his work^1,2 and that of others^28,29 in the early 1980s needed to both exclude a polynucleotide-based mechanism of infectivity and explain the key peculiarities of TSE transmission known at the time. It describes the experimental criteria to be met in demonstrating the transfer and recreation of a complete infective unit between organisms via a purely proteinaceous infectious agent (the “prion”). According to this hypothesis, the minimal infective unit is a protein that a) carries with it the information necessary to convert a resident non-toxic protein into a complete copy of the toxic prion and b) uses endogenous biological mechanisms of the recipient organism so as to permit the infection of additional host organisms. More recent investigations have implicated templated misfolding of PrP^C in recipient cells by exogenous PrP^Sc imported via various mechanisms as the actual propagation mechanism for PrP^Sc-mediated TSEs, However, this mechanism is not stipulated as a limiting criterion in the Prion Hypothesis. It should also be noted that because the Prion Hypothesis was formulated to account for the peculiar epidemiology of TSEs in mammals, a candidate “prion” also implicitly needs to meet an additional requirement imposed by the multicellular nature of the host —i.e., the ability to propagate between postmitotic cells (such as neurons) within the host organism. Although several prionlike yeast proteins^30-32 have provided invaluable insights into the pathogenesis of TSEs at the cellular level, this latter requirement appears to exclude yeast prions from full membership by virtue of the unicellular nature of their natural hosts, although they meet the other formal criteria for Prion Club membership.^32,33 Yeast “prions” share certain general structural features (i.e., enrichment in N, Q, P and related amino acids) with mammalian PrP, but have little specific sequence homology to it.³⁴ They have thus been particularly useful in distinguishing generally applicable prion characteristics from those that are PrP-specific and in validating the Prion Hypothesis as a concept. However, a complete understanding of TSE etiology and pathogenesis ultimately depends on elucidating the link between “normal” and pathological functions of PrP.

Insights into “Prionoid” Lesion Propagation Mechanisms from TSE Models

Cellular and transgenic PrP-based models have been successfully used to reproduce subtle characteristics of TSEs at the molecular (templated alterations in protein conformation), cellular (secretion, misprocessing) and organism (infectivity) levels. A prime example of this is the maintenance of strain-specific clinical and neuropathological differences between individual TSEs, which now appear to be associated with specific misfolded PrP conformations,^35-37 that are maintained even when the infection source is purely sporadic.³⁸ We now know that subtle differences in PrP^Sc tertiary structure between disease strains are modulated not only by polymorphisms at specific key loci in the PrP sequence (e.g., position 129 in familial TSEs induced by the D178N mutation) but also by polymorphisms elsewhere in the molecule associated with expansion of the octapeptide repeat region.³⁹ It is also becoming apparent that the propagation of misfolded PrP through the CNS of an infected individual and between different individuals occurs on a different timetable than the cytotoxic consequences of PrP misfolding,^40,41 with propagation to accessible vulnerable tissues appearing to precede the development of cytotoxicity.⁴⁰ This is consistent with the observed separation of templating, propagation and toxicity-mediating domains of PrP, and may shed light on the criteria for prion-like changes in other proteins with analogous structural and functional motifs. However, it also highlights the relative lack of information about the actual propagation mechanism of PrP^Sc between cells as opposed to the detailed characterizations available for the templating mechanism itself.

PrP^C is primarily localized to the plasma membrane and membrane-associated trafficking vesicles⁴² and plays a role in membrane-associated cellular functions such as process outgrowth.^43,44 Transformation of PrP^C to PrP^Sc occurs in close association with early endosome formation, thereby linking membrane trafficking with TSE cytopathogenesis and PrP^Sc propagation.⁴⁵ The GPI anchor mediated membrane link of PrP appears to be intimately involved with PrP^Sc propagation in particular, as even subtle modifications result in major alterations in propagation efficiency and pattern in transgenics.^46-50 The role of PrP^C in synaptic plasticity and signal transduction⁵¹ in particular may account for unique PrP characteristics (such as the subtlety of PrP templating in TSEs) and could explain (at least partly) why PrP is the only identified mediator of protein-only disease transmission among mammals. A possible analog for a prionlike normal PrP function is provided by a synaptic plasticity protein (cytoplasmic polyadenylation element binding protein or CPEB) in the marine snail Aplysia. The Aplysia CPEB has multiple highly stable conformations that confer stability to associated memory states and appear to involve localized templating interactions (model)⁵² that are highly reminiscent of abnormal PrP^Sc function in TSEs⁵³. Expression of the Q/N rich N-terminal domain of CPEB in yeast confers heritable conformation changes, thus making CPEB a prion in the same sense as yeast prions are.⁵⁴

“There Goes the Neighborhood”- The Case for Auxiliary Membership for “Prionoid” Proteins

The early brain inoculate studies in chimpanzees conducted by Gadjusek in the 1960s²⁹ demonstrated the unique transmissability of PrP-based TSEs relative to other neurodegenerative conditions, including AD and Parkinson Disease. However, it is the interneuronal transfer of PrP^Sc, rather than interorganismal transmissability per se that has attracted attention as a potential general model of non-TSE neurodegenerative disease pathogenesis. Recent studies of other proteins associated with aggregation-driven toxicity in neurodegenerative conditions have raised the question of whether an associated, less exclusive “prionoid”¹⁶ status should be designated in the Prion Club for proteins capable of mediating the interneuronal propagation of neurodegeneration-inducing toxicity rather than that of transmission between individuals. This would provide an attractive hypothetical framework for considering the mechanisms responsible for the stereotyped progression of neurofibrillary lesions through the brain in AD^5,7 Parkinson disease⁶ and other tauopathies.⁸ Of the 3 aggregation-prone proteins (e.g., Asyn, Abeta and tau) involved in the majority of human neurodegenerative conditions, Abeta and Asyn have received the most attention, and explicit cases have been made for their designation as prionlike agents in the pathogenesis of AD and Lewy Body dementias respectively.^55,56 Both proteins are secreted from neurons,^57,58 where they induce localized toxicity via either uptake^59,60 or receptor-mediated mechanisms.⁶¹ Some in vivo evidence for aggregation-mediated lesion propagation exists for both Abeta and Asyn; intraperitoneal injection of Abeta into mice transgenic for familial AD mutations in amyloid precursor protein have been shown to accelerate the onset of senile plaque formation⁵⁶ whereas long-term fetal grafts into Parkinson Disease patients have exhibited Lewy Body pathology that can only be plausibly accounted for by a lesion spreading mechanism.⁵⁵ It is worth noting that other neurodegenerative conditions driven by aggregation-prone proteins or protein sequences also share “prionoid” properties. Examples include SOD1, TDP-43, and polyQ containing proteins, with the latter being particularly interesting in the context of the high QN content of “consensus” regions identified in various yeast prions.³⁵ The degree to which key aggregation-prone proteins discussed satisfy Prion Club requirements is summarized in Table 1.

Table

	Inter-organismal transmission	Secretion Route	Uptake	Toxicity	Transport	Selective Templating (generation/transmission of multiple diseaseforms)	References
								Mammalian Prions	Yes	Exosomal	Yes	Yes	Anterograde and Retrograde	Templating, multiple forms	1,29,116
Yeast “Prions”	Yes*	Yes	Yes	N/A	N/A*	Templating, multiple forms	32
Asyn	Yes***	Exosomal and Microvesicle	Yes	Yes	Anterograde	Selective sequestration Templating possible	57,59
A β	Maybe**	Exosomal	Yes		Anterograde	Selective sequestration Templating possible	56,58
Tau	Not yet	Exosomal and Microvesicle	Yes	Yes	Anterograde and Retrograde	Selective sequestration Templating possible	19,21,22,24–27,75

* No lesion spread—single cell organism; **model requires priming by mutant β-Amyloid; ***direct (graft), model evidence of iatrogenic transmission

The Case for Tau as a Member in the Expanded Prion Club

Unlike Asyn and Abeta, tau is normally a cytosolic protein, expressed primarily in neurons and glia,⁶² with a significant role in the modulation of microtubule (MT) stability,^63,64 especially in axons, where it appears to play a role in establishing neuronal polarity and axonal identity.⁶⁵ While tau has been widely supposed to play an exclusively cytosolic/cytoskeletal role both normally and when misprocessed to form neurofibrillary aggregates, there has been increasing evidence that tau interacts with plasma membrane elements and becomes involved in membrane-associated mechanisms, including secretion, ever since the mid-1990s, when Lee and coworkers demonstrated that the tau N-terminal “projection” domain interacts with lipid raft-associated non receptor tyrosine kinases.^66,67 Tau has since been shown to have well defined roles in axonal outgrowth, guidance and myelination that involve tyrosine kinases (e.g., fyn) involved in signal transduction, endocytosis and vesicle trafficking,^68-70 and which appear also to be involved in tau secretion.²² However, while tau resembles PrP and Asyn in having a tertiary structure that is delicately balanced between an extended normal conformation and one prone to self-aggregation,¹⁵ it now appears that there are toxicity pathways associated with tau misprocessing that are not directly associated with classic “prionlike” templated misfolding. This raises fundamental questions about the meaning of “prionlike” that must be addressed when considering the Club membership of tau.

Tau Secretion and Toxicity in Models

Despite its cytosolic reputation, the possible involvement of secretion in disease-associated tau misprocessing was evident in the first successful “tauopathy” model,^71-73 in which human tau expressed cell autonomously in identified lamprey central neurons (ABCs) was released in the vicinity of degenerating dendrites. Further studies in this system established that tau secretion was fully separable from tau-induced toxicity by either the application of small molecule inhibitors of tau aggregation,^25,74 or by the removal of the MT binding repeat (MTBR)-containing C-terminal domain.¹⁹ Tau thus exhibits the full range of possible propagation routes necessary for Prion Club membership, at least in the lamprey model (Fig. 1). In all models in which tau secretion has been characterized, the presence of the N terminus has been essential for secretion, the efficacy of which is modulated by the presence of the alternatively spliced exon 2 insert.²⁰ By contrast, the uptake of tau from the extracellular space does not appear to require the tau N terminus, and rather appears to be linked to the aggregation state of the tau MTBR.²⁴ Extracellular tau toxicity can be either due to tau uptake²⁴ or via receptor mediated Ca²⁺ fluxes.⁷⁵ One of the two identified mechanisms of tau-induced toxicity is associated with MTBR mediated aggregation,^76,77 and disease-specific lesion characteristics at the molecular⁷⁸ cellular⁷⁹ and intercellular levels of analysis,^5-7,80 and thus appears analogous to PrP-style templated misfolding. Tau filament formation is itself catalyzed by interactions with membrane lipids⁸¹ and with perimembranous extracellular matrix molecules⁸² as well as via direct templating with other MTBRs,⁸³ suggesting a subtlety reminiscent of PrP-mediated pathogenesis. The other toxicity mechanism is specific to the N terminus^61,84-86 and mediates Abeta cytotoxity.⁸⁷ While N-terminal tau toxicity does not require the MTBR^85,86 and thus does not appear to involve templated misfolding, it could in theory propagate tau aggregate-containing lesions by modulating caspase and calpain cleavage of endogenous tau in recipient cells. This would generate both aggregation-prone tau fragments containing the MTBR^88,89 and additional MTBR-less N-terminal fragments, whose secretion could mediate further interneuronal lesion propagation (Fig. 2B).

Figure 1. Human tau is propagated transsynaptically, to adjacent neurons and to the CSF in the lamprey tauopathy model. (Panel A) Confocal images of a dendrodendritic apposition between 2 different identified lamprey reticulospinal neurons (ABCs) expressing the wild-type (WT) 4R0N human tau isoform in a transverse section through lamprey brain 20 d after plasmid injection. The right image shows the region inset at left. N-terminally tagged tau can be seen in exocytosing profiles (arrows) near the points of dendrodendritic contact (asterisks). Dendrodendritic anastmoses have not previously been described in lamprey reticulospinal neurons. (Panel B) Sonatodendritic (top left) and axonal (bottom left) cross-sectional profiles through ABCs expressing the 4R0N human tau isoform containing the P301L tauopathy-inducing mutation. P301L tau is diffusely secreted from the somata of expressing ABCs (asterisk, top left) crosses the IVth ventricle (IV) and enters adjacent neuronal somata (arrow, top left). Bottom Left: Somatic tau staining (s) can be seeen in a neuron that is postsynaptic to the axonal profile (a) of a tau-expressing ABC. Right image in B shows periventricular regions from the hindbrain of a lamprey that contained P301L-expressing ABCs (not present in section). Note that periventricular tau has moved into and bilaterally within the ventricle (IV, asterisk marks midline), crossed the ependymal cell layer (ep) and entered small neurons (n) on both sides of the brain (arrows). The gradient of immunolabel (inset) suggests that exogenous tau from the ventricle has penetrated brain tissue. (Panel C) Transverse sections through brains of lampreys that were treated with control vehicle (DMSO – leftmost image) and anti tau aggregant (NNI3 - right 3 images) as described in.²⁵ The rightmost section was taken at a point 100 microns rostral to the ABC dendritic field limit, and shows 3R0N tau-positive afferents to expressing ABCs (shown in center-right image) and thus illustrates retrograde transsynaptic transport of human tau.The center-right image shows a tau-expressing ABC and also tau-positive profiles of dendrites from adjacent non-expressing ABCs, suggesting dendrodendritic tau movement. Note that secretion, retrograde and transventricular tau movement are all enhanced by a tau anti-aggregation agent (NNI3), resulting in clearance of tau from overexpressing ABCs and uptake into periventricular ependymal cells (center left image). The section shown in A was immunostained for the N-terminal GFP epitope tag (red) a general tau marker (Tau5 – green) and filamentous actin (phalloidin – blue). The sections shown in B and C were immunolabeled with the human tau specific mAb Tau12 to the tau N terminus (residues 9–18) and revealed using standard HRP/DAB immunohistochemistry. Scale Bars: Panel A: 20 microns, Panel B 50 microns,(left): 100 microns (right), Panel C 100 microns

Figure 2. Candidate neuronal and interneuronal pathways of self-replicating tau toxicity and propagation. The generation of PrP^Sc from resident PrP^C occurs via a chaperone-like “templating” in association with mambrane-associated misprocessing events that eventiually result in exosome mediated PrP^Sc release to the extracellular space and transsynaptic propagation of infective prions to other parts of the CNS. (A and B) Hypothesized templated (A) and non-templated (B) tau lesion propagation pathways consistent with the Prion Hypothesis. (A) Recent studies of tau misprocessing suggest that tau misprocessing may occur via a similar mechanism (shown in red arrows) that permits synergistic interactions with PrP and other “prionoid” proteins such as Asyn and Abeta (shown in blue arrows). Either overexpression or toxicity-induced hyperphosphorylation causes the generation of non-MT associated tau which becomes associated with and phosphorylated by fyn kinase and other elements in the plasma membrane (e.g., PrP, Asyn, Abeta), resulting in the oligomerization, endocytosis and relocalization of tau to pre and postsynaptic sites within the neuron, where it accumulates and causes localized toxicity and exosome-mediated secretion. The ability of tau to selectively sequester other MAPs and bind/oligomerize with other aggregation prone proteins such as Asyn and PrP suggests that template-mediated misfolding might occur in this pathway (orange) (Panel B) Tau lesion propagation might also occur via the generation of toxic tau cleavage fragments by Abeta or tau-induced calpain and caspase activity and their secretion via a separate pathway. N-terminal truncated, MTBR containing, tau fragments exhibit an increased tendency to aggregate.^88,89 MTBR- fragments are readily secreted and may induce calpain⁸⁴/caspase⁸⁵ activation in recipient neurons increased resulting in self-regenerating tau lesion propagation. (C) Tau/PrP interactions in exosomal secretion. Two possible ways in which membrane associated tau misprocessing might result in diversion to the exomal secretion pathway. Usually considered cytosolic, non-MT-associated tau can be targeted to lipid raft domains in the plasma membrane by binding ligands such as fyn kinase or subcortical actin (1). Tau is phosphorylated by fyn (or a related Y kinase) and oligomerizes (2), possibly with the aid of polyanionic molecules within (fatty acids) or spanning (heparan sulfate proteoglycans (HSPG)) the membrane, causing raft patching, endocytosis (3) and the formation of multivesicular bodies (4) and exosome release (5). Although the GPI anchoring of PrP should direct it to the exterior of the plasma membrane, the presence of polyanions such as HSPG may bring it into direct contact with tau and permit synergistic co-oligomerization. Other aggregation-prone proteins such as Abeta and/or Asyn may play similar, possibly context-dependent roles with respect to the misfolding and interneuronal propagation of PrP and/or tau.

The consideration of a possible “prionoid” role for tau in lesion propagation is more recent and more tentative than for Abeta and Asyn, despite spatiotemporal lesion development in AD and non-AD tauopathies suggestive of transsynaptic tau movement^6,7 and demonstrations of toxicity-associated secretion^71,73, interneuronal mobility²⁵ and active release to the CSF in non-murine tauopathy models (Fig. 1). This is very likely due to unique difficulties of modeling tau toxicity and secretion in more widely used cell culture and murine transgenic tauopathy models. In cultured neurons, overexpression of human tau isoforms has consistently failed to produce tau specific toxicity⁹⁰ without the application of toxic, broad spectrum inducers of protein phosphorylation⁹¹ or the use of hyperaggregating mutants,⁷⁶ although overexpression at similar levels in situ readily induces tau hyperphosphorylation, fibrillogenesis and neurodegeneration.^71-73,92,93 This is presumably due to a dependence of tau-induced toxicity either on terminal differentiation of the neuronal state⁹⁴ or on the retention of toxic secreted tau in the perineuronal extracellular space,^27,95 neither of which can occur in cell culture. Initial attempts to model human tau induced toxicity in murine transgenics used wild-type tau isoforms that were only mildly overexpressed relative to human tau levels and did not induce neurodegeneration, although they did exhibit somatic accumulation of phosphorylated, non-MT-associated human tau.^96,97 Signs of neurofibrillary degeneration were only observed with stronger (5x human) overexpression of wild-type human tau isoforms, and then only in aged animals.^92,93 The first widely used mouse model of familial tauopathy was generated using the overexpression of a tauopathy mutant tau isoform (P301L).⁹⁸ Tau secretion was not identified in any of these models, presumably owing to the difficulty of determining the source of secreted tau against a tau–positive background, a problem that was not encountered in cell-autonomous, non-transgenic models such as the lamprey. This problem was eventually elegantly circumvented in a study that implicitly demonstrated tau-mediated lesion propagation by showing that tau-induced aggregate formation could be propagated over significant distances within the brain of a mouse already transgenic for human tau.²⁶

Despite widespread reservations about the use of overexpression in disease models, the nature of tau-induced toxicity makes it likely that in situ overexpression of human tau faithfully replicates the downstream disease mechanisms (i.e., toxicity and secretion) that drive AD and tauopathy pathogenesis. Externally applied Abeta 1–42 induces tau hyperphosphorylation and MT loss⁹⁹ as does the presence of familial tauopathy mutations,^100,101 effectively raising the cytosolic concentration of non-MT-associated tau (as does tau overexpression) and thus favoring tau secretion. The repeated observation that tau overexpression induces AD-like tau hyperphosphorylation is also consistent with the cytopathological effects of Abeta-mediated tau hyperphosphorylation in AD, and suggests that both hyperphosphorylation and overexpression both result in the accumulation of nonMT-associated tau in somatodendritic locations, albeit by different mechanisms. Tau secretion in the lamprey model occurs preferentially from dendritic loci where non MT-associated tau has accumulated and MTs are being lost prior to the onset of degeneration,^22,72,73 suggesting that such tau is both locally toxic^27,76 and prone to secretion.^19,22 Tau secretion that disposes of intracellular accumulations may be neuroprotective to the neuron expressing tau,^25,95 but may also generates extracellular tau that is toxic to other neurons^27,75 The neuroprotective and pro-secretory effect of the anti-tau aggregant NNI3²⁵ may be due to the ability of NNI3 to to block high level tau oligomer/polymer formation,⁷⁴ which could induce accumulation of lower level oligomers in membrane-associated tau and possibly favor its secretion,¹⁰² thereby preventing the intracellular buildup of toxic, non-MT-bound tau.

Secreted, Oligomerized Phosphotau in Human CSF

The recent finding that elevated CSF tau in early AD appears to be secreted²¹ constitutes the first direct link between tau secretion and human disease pathogenesis and replicates the transventricular propagation of tau that has been demonstrated in the lamprey model (Fig. 1) and suggested in recent mouse model studies.²³ It also raises novel possible mechanisms for prionoid lesion propagation. For instance, CSF-tau levels are significantly more elevated in AD than in non-AD tauopathies,¹⁰³ suggesting that Abeta-tau interactions – possibly involving the tau N-terminal projection domain^61,85 - might enhance any tau secretion responsible for the genesis of elevated CSF-tau. This is consistent with the overrepresentation of N-terminal tau species in CSF from AD patients¹⁰⁴ and the restriction of significantly elevated CSF tau to AD (relative to most non-AD tauopathies), but is not necessarily indicative of a classical “prionlike” mechanism, since the MTBR of tau is not needed for either tau secretion or Abeta-mediated toxicity.^19,85 However, recent findings that exposure to CSF from Parkinson disease (which contains oligomeric Asyn) is cytotoxic¹⁰⁵ raise the possibility that elevated CSF tau may play a direct role in AD pathogenesis that may not be present in non-AD tauopathies. The finding that some of the membrane-associated tau in AD may be oligomeric is especially interesting in this context, as it is more consistent with a PrP-like templated toxicity mechanism.²¹

Is Tau Lesion Spreading due to Synergistic Interactions with PrP?

While improved techniques have overcome most limitations to the infectivity and in vivo propagation of artificial prions,¹⁰⁶ this propagation still requires cytosolic and membrane-associated cofactors capable of stabilizing the transient intermediate conformations that occur in PrP^C to PrP^Sc conversion to complete the transformation of PrP^C into fully infectious PrP^{Sc 107–108}. In particular, it is still unclear whether the strain-specific PrP conformations that give rise to specific TSEs require particular cellular contexts (and ligands) to express the disease specific features of any given TSE.^46-49 One possible context-dependent ligand that might modulate PrP-mediated disease is tau, which exhibits significant neuropathological and neurochemical abnormalities in at least some TSEs.¹⁰ NFT formation is a prominent aspect propagated lesions in Creutzfeld-Jakob Disease (CJD) in as well as in AD, while CSF-tau levels are even higher in CJD than they are in AD, raising the possibility that PrP^Sc in at least some TSEs may induce tau misprocessing, especially since tau has recently been shown to bind to PrP on a disease-associated basis.¹⁰⁹ However, the implications of such interactions are unclear, as a recent mouse model study suggests that knockout of endogenous tau does not modulate conversion of PrP^c to the toxic PrP^Sc form,¹¹⁰ while another study in a cellular overexpression model demonstrated a tau-mediated depletion of endogenous PrP at the plasma membrane that was correlated with PrP toxicity.¹¹¹ That said, there are numerous similarities between tau and PrP misprocessing pathways that would seem to present ample opportunity for these proteins to interact in ways that would favor their mislocation, aberrant phosphorylation, oligomerization and secretion (see Fig. 2). Both PrP^Sc generation from PrP^C and membrane-associated events in tau misprocessing^20,21 are localized to subcortical actin-rich structures that include early endosomes.^42,112,113 Both tau and PrP exhibit disease-associated interactions with the lipid raft-associated tyrosine kinase fyn in endocytotic pathways that characteristically involve the oligomerization of fyn-associated membrane-associated proteins^{70,102,114,115,} in some cases via substrate.⁷⁰ This last is of particular interest since both tau and PrP have now been shown to undergo exosome-mediated secretion^21,116 in association with disease-associate misfolding¹¹⁷ and since oligomerization may cause the diversion of a broad range of membrane-associated proteins into the exosomal secretion pathway.¹¹⁸ Moreover, tau and PrP colocalize and interact with common polyanionic ligands^82,119-122 including HSPGs associated with unconventional secretion pathways.¹²³ The relocalization of either tau or PrP with respect to membrane topology via polyanion binding could be important in increasing synergistic interaction between tau and PrP¹¹¹ and their endocytosis,^70,124 hypothetically leading to oligomerization of tau and/or PrP as well as their secretion (Fig. 2C). Finally, while most tau is MT-bound and axonally localized, tau is typically disassociated from MTs and relocalized to the soma and dendrites in AD and non-AD tauopathies, and has recently been identified in dendritic spines, where it has a tauopathy mutation-specific effect on synaptic function^125–127. Thus the specific association and modification of tau with/by abnormal ligands may well play a key role in disease-associated tau misprocessing. Similarly, differential processing of PrP^Sc in various TSEs suggest that TSE disease identity is to some extent dependent on differential interactions between PrP and ligands capable of stabilizing critical conformation intermediate states between PrP^C and PrP^Sc. Candidate ligands (e.g., Asyn, 14–3–3 proteins) have known chaperone-like functions and are known to interact with membrane-associated polyanions¹²⁸ as well as both PrP and tau.^120,129,130 Conversely, it is also possible that disease-specific elements of tauopathies (e.g., propagation rate¹³¹ might be due to different affinities of tau for PrP imposed by disease mutations or combined with isoform specific misprocessing elements. Conformational changes induced in tau by the P301L tauopathy mutation may alter the relationship between tau and specific tau-binding proteins (e.g., Asyn¹³²) that are known to have effects on tau conformation and MT-binding.⁸ This may also account for the disproportionate presence of specific tau isoforms (i.e., exon 2+, exon 10+) in the elevated CSF-tau levels typically present in CJD.¹³³

Summary and Conclusion

In summary, current research in models and very recently human patients suggests that toxic forms of tau can (A) be secreted to ISF and other body fluid compartments (CSF, blood) and be taken up by adjacent neurons and (B) be retrogradely and anterogradely transported between neurons across synapses. Specific secretion pathways have been identified that are modulated by defined regions of the tau molecule. There is also clear evidence that oligomerized tau is itself toxic,¹³⁴ with extracellularly applied tau being more toxic,²⁷ more susceptible to uptake²⁴ and able to “seed” new aggregates⁶⁰ in an oligomeric state than it is as monomer. Misprocessed (e.g., hyperphosphorylated, oligomerized, mutant) tau shows features that are consistent with templating, including an altered conformation¹³⁵ the selective sequestration of tau family MAPs by misprocessed tau⁸³ and the multiple filament morphologies encoded by aggregated tau isoforms.⁷⁸ These characteristics suggest that tau might be capable of disease-specific templating under ideal conditions, some of which may involve poorly understood interactions with other aggregation-prone proteins, including PrP itself. However, it remains possible that tau toxicity transfer could be an important disease mechanism without involving templated misfolding, especially in AD, where established toxicity routes for the known causative agent (β amyloid) include MTBR-independent toxicity^61,86 It is still unclear whether the features characterizing tau secretion and toxicity described by our lab and others in various tauopathy models will apply directly to AD and other tauopathies, but the recent association of exosomal tau secretion and elevated CSF-tau in AD²¹ and the generation of a mouse model for early-stage tau lesion spread¹³⁶ suggests that they will have significant clinical relevance. Tau’s credentials for membership in the “prionoid” auxiliary of the Prion club may therefore be said to be currently under review, but appear to be broadly similar to those of A β and asyn, even if only templating-mediated mechanisms are considered.

While the lack of direct evidence for interneuronal as well as inter-organismal transmission for tau in human neurodegenerative disease is likely due to a more limited ability of tau proteins to template and propagate than is true for PrP, it may also reflect the current rudimentary understanding of the role of tau transfer in neurodegenerative disease pathogenesis. Our assessment is further complicated by what is still a relatively poor understanding of the prion propagation mechanism – how PrP^Sc is actually transferred between neurons - beyond the established ability of PrP^Sc to move between neurons. While it is clear that PrP^Sc can be secreted via the exosome pathway⁵⁶ in a number of cell types including neurons¹³⁷ and it appears likely that this mechanism is operative in gut-transmitted TSEs (such as vCJD)^117,138 it is as yet unclear whether this is actually how disease transmission occurs between neurons in TSEs. PrP possesses a “signal” sequence and can be incorporated into extracellular aggregates via the classical secretion pathway via cleavage of its GPI anchor,⁴⁸ but there is as yet no evidence that this is an important pathway in the cytopathogenesis of TSEs. Finally, the multiplicity of toxicity routes (some of which do not involve templated misfolding) available to tau,^86,87 and other “prionlike” protein candidates (e.g., TDP-43)¹³⁹ raise issues that are central to both tau membership (e.g., are both toxic tau and PrP^Sc generated and propagated in a common exosomal pathway in TSEs and/or tauopathies?) and a more fundamental question – what does “prionlike” mean? Does it merely require that the original Prion Hypothesis be satisfied in the context of interneuronal lesion transfer, or does it also require that the toxicity and propagation mechanism involve templated protein misfolding, the mechanism now generally accepted as mediating PrP pathogenesis? It is possible to envision a scenario (diagrammed in Fig. 2B) by which neurofibrillary lesions are spread via the secretion of N-terminal tau fragments¹⁹ that act via amplifying excitotoxicity^75,85 in the recipient cell,^61,87 especially in AD, where such fragments constitute the majority of CSF-tau species.¹⁰⁴ This is of course highly speculative (as is the case with the now much remarked-upon similarities between aggregation-prone proteins such as tau and PrP) but it again raises the possibility that the original concept of protein mediated toxicity transfer may need to be updated.

So in short, the background check on tau is incomplete at the Prion Club, but fraternization between applicants and members has already begun, with possible widespread consequences for the meaning of Club membership.

Abbreviations:
PrP	=	prion protein
PrPSc	=	misfolded disease-associated PrP
PrPC	=	Normal (cellular) PrP
TSE	=	transmissible spongiform encephalopathy
CJD	=	Creutzfeld Jakob Disease
AD	=	Alzheimer’s Disease
MTBR	=	microtubule binding repeat
Asyn	=	alpha synuclein
Abeta	=	beta amyloid peptide
CPEB	=	cytoplasmic polyadenylation element binding protein
GPI	=	glycophosphoinositol
HSPG	=	heparan sulfate proteoglycan
MT	=	microtubule
CSF	=	cerebrospinal fluid
CNS	=	central nervous system
4R0N	=	tau isoform with 4 microtubule binding repeats
Exon 2,3 –	=	tau lacking exons 2 and 3