Travels with tau prions

Tau was originally identified as a microtubule associated protein, and subsequently recognized to constitute the fibrillar assemblies found in Alzheimer disease and related neurodegenerative tauopathies. Point mutations in the microtubule associated protein tau (MAPT) gene cause dominantly inherited tauopathies, and most predispose it to aggregate. This indicates tau aggregation underlies pathogenesis of tauopathies. Our work has suggested that tau functions as a prion, forming unique intracellular pathological assemblies that subsequently move to other cells, inducing further aggregation that underlies disease progression. Remarkably, in simple cells tau forms stably propagating aggregates of distinct conformation, termed strains. Each strain induces a unique and, in some cases, transmissible, neuropathological phenotype upon inoculation into a mouse model. After binding heparan sulfate proteoglycans on the plasma membrane, tau assemblies enter cells via macropinocytosis. From within a vesicle, if not trafficked to the endolysosomal system, tau subsequently enters the cytoplasm, where it becomes a template for its own replication, apparently after processing by valosin containing protein. The smallest seed unit is a stable monomer, which suggests that initial folding events in tau presage subsequent pathological aggregation. The study of tau prions has raised important questions about basic cell biological processes that underlie their replication and propagation, with implications for therapy of tauopathies.

glucocorticoid receptor (Diamond et al., 1990); Stanley Prusiner, M.D., in whose lab I originally worked as an undergraduate, and who won the Nobel prize alone for his discovery that prions are the cause of infectious neurodegenerative diseases; and Jonathan Weissman, Ph.D., who elegantly exploited reductionist approaches to study Sup35, a yeast prion.Prions are infectious pathological protein assemblies that serve as templates for their own replication.They adopt multiple, distinct, self-propagating structures, termed "strains," that are responsible for different disease presentations (Ayers et al., 2020).Like many before and after, I was fascinated by the idea that prion proteins are "shape-shifters," capable of faithful replication of unique amyloid structures in vivo to regulate cellular physiology, or, when uncontrolled, to cause disease.I wondered whether tau might be part of the prion club.
Others had made key discoveries that set up these ideas.First, electron microscopy revealed fibrillar assemblies (not yet identified as tau) in AD brain (Kidd, 1963).Tau was separately identified and purified by the Kirschner lab as a microtubule-associated protein (Cleveland et al., 1977;Weingarten et al., 1975).Brion et al.and colleagues subsequently used antibodies to identify it as a principal component of neurofibrillary tangles in Alzheimer's disease (Brion et al., 1985).Cloning of the tau gene followed, and its further characterization as a component of AD-associated fibrils (Goedert et al., 1988;Wischik, Novak, Edwards, et al., 1988;Wischik, Novak, Thøgersen, et al., 1988).The identification of multiple disease-associated dominant point mutations in the microtubule associated protein tau (MAPT) gene established it as an underlying cause of neurodegenerative tauopathies (Goedert et al., 2000;Hutton et al., 1998;Lee et al., 2001;Poorkaj et al., 1998).Since many of the mutations predispose tau to form amyloid structures in vitro and in vivo, this also implicated its aggregation as an underlying cause of neurodegeneration.The viability of tau knockout mice indicated tau is not required for basic functioning of a cell or the nervous system (Harada et al., 1994), consistent with pathological activity derived from gain, rather than loss of function.In 1991, Heiko and Eva Braak described an orderly, spreading progression of tau pathology in Alzheimer disease (Braak & Braak, 1991).As I considered these findings, I was also intrigued by amyotrophic lateral sclerosis (ALS), a networkbased neurodegenerative disorder that is highly restricted (in its initial presentation) to upper and lower motor neurons.In 2003, as I started my independent lab, I wondered whether prion mechanisms could underlie the progression and diversity of tauopathies and related disorders (Figure 1).

| EXOGENOUS TAU TRIGGERS INTRACELLULAR AGGREGATION
Mice were the standard research tool to study prion protein (PrP) infectivity, but as a new assistant professor I could not afford them to test my ideas about tau.Fortunately, our prior work using fluorescent protein tags to study intracellular polyglutamine protein aggregation (Diamond et al., 2000;Pollitt et al., 2003;Welch & Diamond, 2001) allowed us to monitor tau-YFP aggregation in cells.A decade earlier, biochemists had developed purification and in vitro fibrillization protocols for tau (Goedert et al., 1996;Montejo de Garcini et al., 1988;Perez et al., 1996;Wille et al., 1992), which we replicated to form our own fibril preparations.Ordinarily, a tau-YFP fusion colocalizes perfectly with microtubules (Frost et al., 2009).As a fundamental test of the prion hypothesis, we tested whether exogenous tau fibrils could gain entry to a cultured cell to trigger tau-YFP aggregation.Remarkably, this worked: fibrils added to the cell media entered cells, converting the diffuse intracellular tau-YFP to punctate amyloid structures.Moreover, we observed newly formed tau-YFP aggregates transferred to co-cultured cells (Frost et al., 2009).These rudimentary experiments for the first time linked tau and prion mechanisms.Of course we were not the only ones with this idea, and the Tolnay lab contemporaneously observed Tau monomer in solution adopts a variety of conformational ensembles.Amyloid-forming motifs are represented by colored regions.In the inert form(s) of tau the amyloid motifs are relatively shielded, and tau does not aggregate.In the seed-competent form(s), the motifs are exposed to differing degrees, enabling assembly into distinct strains, schematized as Strains 1 and 2. (b) Once formed in a vulnerable cell, tau assemblies (dark blue) are mobile and migrate through brain networks, possibly retrograde within neurons (Ramirez et al., 2023).Assemblies then serve as templates for their own replication in second order cells, recruiting native monomer to the growing aggregate (light blue) via unknown mechanisms.(c) In human brain, depending on the predominant strain that forms and the connectivity of the region in which it does so, seeding occurs in unique patterns, spreading pathology along defined brain networks.This accounts for distinct clinical phenotypes associated with each tauopathy.Figure created using BioRender.com.similar effects after injecting tau preparations into mouse brain (Clavaguera et al., 2009).Our cell-based studies turned out to be foundational.They enabled the development of ever more sensitive "biosensor" cells, engineered so that intracellular aggregation could be quantified and used to read out the concentration of tau "seeds" derived from biological or recombinant sources (Hitt et al., 2021;Holmes et al., 2014).Simple cell models of seeding by protein amyloids have now allowed laboratories around the world to study fundamental mechanisms of aggregation, and to develop new therapeutic leads.

| PRION PROPERTIES OF TAU
The prion model predicted that the diversity of human tauopathies would be linked to tau strains, defined as unique aggregate assembly structures that propagate faithfully in vivo, and, consequently, have distinct biological effects.To study this problem we created tau fibrils in vitro and transduced them into biosensor cells expressing a fusion of YFP to the tau repeat domain (the central core of the protein responsible for its aggregation).The cells formed aggregates, and rapidly cleared them as they divided.Yet a small fraction maintained inclusions indefinitely.We consequently used limiting dilution to establish clonal aggregate-containing cell populations, and observed that each featured inclusions of distinct morphology (Sanders et al., 2014).The inclusions stained positive for thioflavin, indicating beta sheet structure, and each had distinct patterns of protease cleavage, seeding potential into biosensor cells, and detergent solubility.
Furthermore, extraction of the aggregated tau and re-introduction into aggregate-free biosensor cells recreated the same assembly structures (Sanders et al., 2014).In other words, the tau aggregates were "infectious" in cells, and had "strain-like" properties.
We then tested the effects of two highly characterized strains in vivo by inoculating them into the brain of a mouse model created by Virginia Lee's lab (PS19) that expressed a full-length human tau isoform (1N4R) containing the disease-associated P301S mutation (Sanders et al., 2014).Amazingly, each strain produced a unique pattern of neuropathology that was transmissible to subsequent generations of mice.Tau thus had fundamental properties of a prion: it could be induced to form an "infectious" conformation in vitro; it adopted multiple distinct, self-propagating structures in simple cell models (strains); and each strain created a unique, transmissible disease (Sanders et al., 2014).We also determined that tauopathies classified by a neuropathologist contained distinct strain compositions that correlated with diagnosis (Sanders et al., 2014).This work anticipated subsequent identification of unique tau filament structures using cryo-electron microscopy (Falcon, Zhang, et al., 2018;Fitzpatrick et al., 2017;Scheres et al., 2020).Depending on how one defines "infectious," this work suggested tau could be considered a prion, or, at a minimum, "prion-like."Subsequently, we studied 18 strains separately injected into the PS19 mouse model, and evaluated them blinded to the original inoculum.Each strain produced unique neuropathology, with a specific anatomic pattern of spread and cellular vulnerability (Kaufman et al., 2016).These experiments indicated that individual tau strains were a cause, not a consequence of diverse human pathologies, and likely accounted for distinct patterns of disease progression in each tauopathy (Figure 1c).

| TAU MONOMER AS A SHAPE SHIFTER
Tau monomer in solution is often termed "unstructured", although NMR has defined local protein sub-structures (Eliezer et al., 2005;Mukrasch et al., 2005).In a search for the minimal size of a tau seed, our lab and that of Lukasz Joachimiak, Ph.D. identified monomer as the smallest unit capable of inducing intracellular aggregation (Mirbaha et al., 2018).Consistent with earlier NMR studies, we found that tau exists in distinct conformational states, in which local unfolding converts it to a seed-competent form (Chen et al., 2019;Hou et al., 2021;Mirbaha et al., 2018;Mirbaha et al., 2022;Sharma et al., 2018).Importantly, in mice, the formation of the "seed competent" monomer precedes oligomers and eventual neuropathological change (Mirbaha et al., 2022).Moreover, seed-competent monomer derived from different tauopathies encodes subsets of strains in cells that produce distinct pathological patterns after inoculation into mouse brain (Sharma et al., 2018).These findings have been supported by structural studies of the tau filaments from different tauopathies, each of which indicates a unique monomer (protofilament) fold (Shi et al., 2021).This suggests that initial events in each tauopathy derive from particular folding patterns in the monomer.Rather than being "unstructured," we have concluded that tau probably occupies an ensemble of structures that interconvert based on ligand binding and/or stochastic events (Figure 1a).

| MOVEMENT OF TAU SEEDS
If tau propagates pathology in vivo, it must exit one cell and enter the cytoplasm of another (Figure 1b).We have found that all tau seeds bind heparan sulfate proteoglycans (HSPG) at the cell surface, followed by micropinocytosis (Holmes et al., 2013;Stopschinski et al., 2018).At this point the internalized tau must either enter the cytoplasm via membrane rupture, as has been proposed (Falcon, Noad, et al., 2018), or via membrane translocation without rupture, which we have observed (Dodd et al., 2022).We have further determined that tau assemblies within vesicles either traffic to the endolysosomal system for degradation, or escape to the cytoplasm, where they function as seeds, and are eventually cleared by the proteasome (Kolay et al., 2022).The mechanism of how a macromolecule such as a tau assembly might cross a membrane is unknown, and raises important new questions about membrane topology.

| ASSEMBLY REPLICATION MACHINES
Seeding into cells requires that an assembly encounter additional monomer to replicate amidst a dense sea of diverse proteins.Yeast prions have informed us about a potential aggregate amplification machinery, as the chaperone Hsp104, an AAA+ ATPase, plays a critical role in replication of yeast prions: too much or too little of this protein and the Sup35 assemblies will disappear (Doyle & Wickner, 2009).There is no human homolog to Hsp104, but several studies point to a protein of similar function, VCP/p97, a AAA+ ATPase that was recently linked to dominantly inherited tauopathy (Darwich et al., 2020), and which we have found to co-purify with aggregates derived from biosensor cell lines (Perez et al., 2023;Saha et al., 2023).VCP could potentially play two opposing roles: extraction of monomer to degrade seeds (Darwich et al., 2020), and creation of smaller seed-competent fragments by fibril fragmentation (Saha et al., 2023).

| GENERAL FUNCTION(S) OF TAU AND PRIONS
Most diseases derive from normal physiology gone awry.Furthermore, tau is by no means prone to aggregation.Quite the contrary, it is highly thermostable, and will not form fibrils without very specific inducers.Therefore, the efficiency of tau seeding into cells and the maintenance of aggregates suggest that these are linked to an undefined cellular function.Prions carry information within their structure, and, as replicating agents, they confer "transmissibility" of this information-between cells and even organisms.Could the multiplicity of tau strains represent a diversity of potentially useful biological signals?One prediction of this model is that tau seeds should not be confined to disease states.Indeed, we have detected seeding activity in control brain tissues over a range of ages, and in the absence of any evidence of underlying tau pathology (LaCroix et al., 2022).
Taken together, the study of tau, originally identified as a microtubule binding protein, has raised many new ideas: (1) some "intrinsically disordered" proteins probably exist in conformational ensembles and adopt distinct shapes to carry out specific functions; (2) cells carry machinery that facilitates amyloid protein replication, implying an evolutionary utility for this process; (3) mechanisms of tau aggregate propagation (exit, entry, and faithful replication of complex assemblies) may yield important insights about cell biology; (4) interaction of ordered tau assemblies with other intracellular proteins or nucleic acids may have physiological functions that become uncontrolled in disease.As our journey with tau teaches us about each of these topics, we will better understand both normal and abnormal physiology, with important implications for how to diagnose and treat tauopathies and related disorders.