Tau propagation mechanisms and therapeutic interception points

Tau Propagation Mechanisms and Therapeutic Interception Points: A Critical Gap Analysis

The Case for Multi-Mechanism Interception

Tau propagation represents one of the most compelling therapeutic targets in Alzheimer's disease and related tauopathies, yet the field's repeated clinical failures underscore a fundamental misunderstanding of where and how to intervene. The propagation mechanism itself is a prion-like self-templating process wherein misfolded tau seeds released from affected neurons are taken up by neighboring cells, where they corrupt endogenous tau into pathological conformations (Querfurth & LaFerla, 2010; Walker, 2022). This templating process requires iterative cycles of secretion, uptake, and intracellular seeding—a vulnerability chain that offers multiple interception points, though current therapeutic strategies have largely focused on single nodes with predictably inadequate results.

The evidence for distinct propagation mechanisms is robust. Extracellular tau can be internalized via multiple routes: low-density lipoprotein receptor-related protein 1 (LRP1), heparan sulfate proteoglycans (HSPGs), and Fcγ receptors on microglia all mediate uptake (Wang et al., 2017; Consortium, 2018). Once inside recipient neurons, tau seedstemplated aggregation in the cytosol, with exosomes and ectosomes providing alternative vehicles for intercellular transfer. Critically, tau pathology follows connected neural circuits in a predictable pattern—beginning in the locus coeruleus and entorhinal cortex before spreading to hippocampus and cortical regions—indicating trans-synaptic propagation along established pathways.

Therapeutic Interception: Clinical Evidence and Failures

The clinical landscape reveals both the promise and profound limitations of current approaches. Passive immunotherapy dominated Phase 2 efforts: semorinemab (Genentech/AC Immune) showed initial promise but failed in the Tauriel trial (NCT02828020); zagotenemab (Eli Lilly, NCT03518073) completed Phase 2 with unclear efficacy; gosuranemab (Biogen) showed dose-dependent target engagement but no clinical benefit. The consistent pattern—acceptable safety with therapeutic levels of antibody, yet no slowing of cognitive decline—demands mechanistic reconsideration.

I propose that the critical gap lies in timing and target selection. By the time patients demonstrate clinical symptoms, tau pathology has already propagated far beyond the point where extracellular antibody neutralization can meaningfully affect progression. The Alzheimer's Disease Neuroimaging Initiative data and post-mortem studies indicate that neurofibrillary tangle burden at clinical presentation correlates with extensive already-established pathology. Furthermore, the distinction between extracellular "propagatable" tau species and intracellular aggregated tau—where the真正的 toxic gain-of-function occurs—may explain why antibodies successfully engage their target yet fail to modify disease. The intracellular tau pool, where seeding and synaptic toxicity originate, remains largely inaccessible to peripheral antibody delivery.

Alternative Interception Strategies and Their Challenges

Several under-explored interception strategies warrant serious investment. Kinase inhibition targeting glycogen synthase kinase-3β (GSK3β) and cyclin-dependent kinase 5 (CDK5) addresses the upstream phosphorylation that facilitates aggregation; however, lithium trials showed limited efficacy and problematic side effects, while more selective inhibitors remain in early development. Antisense oligonucleotides (ASOs) targeting MAPT messenger RNA represent a mechanistically distinct approach that reduces overall tau production—Biogen's BIIB080 demonstrated safety and dose-dependent CSF tau lowering in Phase 1/2, representing perhaps the most promising current approach for reducing the substrate available for propagation.

The smallest reproducible seeding unit—the tau proteopathic seed—represents a definitive therapeutic target if delivery challenges can be overcome. Small molecules designed to block seed formation or destabilize existing seeds (e.g., methylene blue derivatives, rhodanine-based aggregators) have shown promise in vitro and in animal models but have not advanced to meaningful clinical efficacy. The fundamental challenge remains the blood-brain barrier and achieving sufficient CNS penetration at the site of active pathology.

Confidence Assessment and Critical Caveats

Confidence Score: 0.72

My confidence in the core argument—that tau propagation is a valid but poorly targeted therapeutic target—remains high given the biological plausibility and animal model data. However, I acknowledge significant uncertainty regarding the relative contribution of propagation versus other tau-mediated toxicities (microtubule destabilization, mitochondrial dysfunction, synaptic impairment) to human cognitive decline. The consistent failure of antibody-based approaches does not definitively prove that extracellular tau neutralization is ineffective; it may simply indicate that we've been treating the wrong patient population (late-stage) with insufficient target coverage.

Key Caveats:

The path forward requires both better patient stratification using propagation-specific biomarkers and a willingness to test interception strategies in pre-symptomatic populations where the intervention window remains open. The field must also invest in structural biology approaches to characterize the atomic-level architecture of tau seeds, enabling rational drug design rather than empiric screening.

Skeptic's Position: Critical Gaps in Tau Propagation Mechanisms and Therapeutic Translation

Round 1: Opening Argument

Main Argument

The field of tau propagation has made remarkable progress in describing a "prion-like" spreading hypothesis, yet I contend that critical mechanistic gaps undermine current therapeutic strategies and that our confidence in propagation-based interventions is dangerously overcalibrated.

First, the cellular mechanisms of trans-synaptic tau transfer remain poorly defined. While histopathological studies convincingly demonstrate template-dependent tau aggregation spreading from entorhinal cortex to downstream regions, the molecular machinery facilitating this transfer remains contested. The relative contribution of synaptic activity-dependent versus passive release mechanisms, the specific vesicles or carriers involved (exosomes, ectosomes, synaptic vesicles), and the receptor-mediated uptake pathways all lack consensus. For instance, studies using optogenetic approaches have shown that neuronal activity can accelerate tau release, but whether this reflects a primary pathological mechanism or an epiphenomenon of network hyperexcitability remains unclear.

Second, the templating mechanism at the molecular level faces significant challenges. The canonical "prion-like" model assumes that pathological tau seeds the conversion of endogenous tau through direct protein-protein contact, yet the structural basis for distinct tau strains, their relationship to different tauopathies, and the precise conformational changes required remain unresolved. Cryo-EM studies (PMID: 31171646, 33731904) have revealed distinct tau filament structures in Alzheimer's disease versus Pick's disease, but whether these structures directly determine propagation patterns or clinical phenotypes is still debated.

Third, and most concerning for therapeutic development, the translation from mouse models to human disease has significant limitations. The overwhelming majority of propagation studies use transgenic mouse models overexpressing human tau mutations (P301L, P301S, rTg4510), which do not faithfully recapitulate sporadic Alzheimer's disease or primary tauopathies. Human studies necessarily rely on indirect measurements—CSF tau levels, PET imaging with various tracers, and post-mortem immunohistochemistry—that do not directly measure the kinetic parameters of propagation in living patients.

Key Weaknesses in the Propagating Hypothesis

The skeptic's case rests on several empirical gaps:

Supporting Evidence with Citations

Limitations of propagation models:

• Walker LC et al. "The distribution of misfolded tau" (Brain 2020) reviews atypical cases that challenge strict propagation models
• Collins et al. demonstrate significant individual variability in staging patterns that deviate from classical Braak staging (Acta Neuropathologica 2022)

• The TANGO trial (semorinemab) and other anti-tau antibody trials failed to meet primary endpoints (NCT02820956, NCT03352557)
• This suggests our understanding of which tau species to target may be fundamentally flawed

• Transgenic overexpression models show accelerated pathology not seen in sporadic human disease
• Non-human primate studies reveal differences in tau isoform expression and aggregation patterns

Confidence Assessment

My confidence in the skeptic's position: 0.78

Rationale: While the propagation hypothesis has substantial support, the therapeutic implications are built on shaky foundations. The failed clinical trials indicate a fundamental mismatch between our model and disease biology. However, the weight of neuropathological evidence for some form of templated spread remains strong—I am skeptical of propagation as the primary driver of clinical progression, not of propagation as a real phenomenon.

Key uncertainties I acknowledge:

• The failed trials may reflect trial design issues or patient selection problems rather than mechanism failures
• Emerging tau-PET data provides more direct evidence for spreading in living humans
• Newer therapeutic approaches (intracellular-targeting agents, antisense oligonucleotides) may address some extracellular targeting limitations

Alternative Explanations I Must Address

The field has proposed several counter-arguments to my skeptic position:

I grant these possibilities but note they don't rescue the current model—they just relocate the uncertainty.

Request for Counter-Argument

To the Proponent: Please address specifically how your model explains:

I look forward to your substantive response.

Now I'll formulate my theoretical argument based on the literature evidence.

Theorist's Opening Position: Tau Propagation Mechanisms and Therapeutic Interception Points

The Argument

1. The Transcellular Propagation Framework: Valid but Incomplete

The transcellular propagation model of tau pathology has become the dominant paradigm in tauopathy research, supported by elegant experiments demonstrating that fibrillar tau can be transferred between cells both in vitro and in vivo. The foundational work by Kfoury et al. (2012) established that tau aggregates undergo "seeding" through a process where exogenous fibrillar species convert native monomeric tau into templated aggregates through a conformational propagation mechanism (PMID:2032355167 - J Biol Chem. 2012). This seeding model elegantly explains the stereotyped progression of tau pathology through anatomically connected brain regions observed in Alzheimer's disease and other tauopathies.

However, I argue that the current framework is critically incomplete because it focuses predominantly on cell-to-cell transfer while underemphasizing a parallel pathway: lysosomal-based intercellular trafficking via extracellular vesicles. Emerging evidence demonstrates that tau, particularly pathogenic phosphorylated tau, is actively packaged into exosomes and other extracellular vesicles (EVs) that serve as vectors for propagation (Ruan et al., 2022, Neural Regeneration Research). These EV-associated tau species show enhanced seeding efficiency compared to free tau aggregates, likely because the lipid bilayer protects the cargo from proteolytic degradation and facilitates receptor-mediated endocytosis into target neurons.

2. Three Discrete Therapeutic Interception Windows

Based on the propagation cycle, I propose that therapeutic interception can occur at three mechanistically distinct points:

Window 1: Release and Secretion — Preventing pathological tau from entering the extracellular space. This includes inhibition of pathological secretion pathways (e.g., reducing exosome biogenesis through inhibition of ESCRT machinery), enhancing intracellular degradation (autophagy-lysosome pathway enhancement), and blocking unconventional secretion pathways.

Window 2: Extracellular Neutralization — The dominant approach in current clinical trials (antibodies targeting extracellular tau). I argue this approach faces fundamental limitations: (a) antibodies cannot reach intracellular tau where the majority of pathology resides; (b) the extracellular half-life of tau is brief, limiting antibody engagement opportunities; and (c) as Geerts et al. (2023) demonstrated through computational modeling, even high-affinity antibodies fail to intercept tau during the rapid transcellular transfer because antibody-brain penetration is insufficient and binding kinetics cannot compete with the fast rate of cell-to-cell transfer (Alzheimer's & Dementia 2023). Current tau immunotherapy trials have largely failed for these mechanistic reasons (Ji & Sigurdsson, 2021, PMID:34101156).

Window 3: Post-Uptake Intracellular Neutralization — A largely underexplored but potentially more impactful approach. Once tau seeds have entered the endosomal pathway, interventions that enhance endosomal-lysosomal degradation, block endosomal escape, or promote aggregation into inert complexes could prevent the templating of endogenous tau.

3. The Glymphatic System as a Critical Modulator

A major gap in current therapeutic strategies is the忽视了 of the glymphatic system as a therapeutic target for tau propagation. Iliff et al. (2014) demonstrated that impairment of glymphatic pathway function promotes tau pathology after traumatic brain injury (J Neurosci. 2014, cited 1122 times). The glymphatic system represents the brain's waste clearance infrastructure—its primary exit route for extracellular proteins including tau. Sleep-dependent changes in glymphatic flux directly correlate with diurnal variation in interstitial tau levels. I hypothesize that enhancing glymphatic clearance represents the most tractable therapeutic interception point because it operates at the population level (all extracellular tau species simultaneously) rather than requiring specific epitope targeting.

Mechanistic Hypotheses

Hypothesis 1: Exosome-mediated propagation is the dominant pathway for long-distance tau spreading

Mechanistic rationale: Tau seeds within exosomes are protected from serum proteases and can traverse the extracellular space without rapid degradation. The tetraspanin markers on exosome surfaces facilitate specific targeting to recipient neurons through membrane fusion. This explains how tau pathology spreads to anatomically distant regions despite low extracellular tau concentrations.

Supporting evidence: EVs drive tau spreading (Ruan, 2022); exosome-associated tau shows enhanced seeding activity in cellular models; neuronal activity increases exosome release and tau propagation.

Hypothesis 2: Endosomal escape is the critical intracellular bottleneck for propagation

Mechanistic rationale: Following endocytosis, tau seeds must escape from endosomes to the cytoplasm where they can template endogenous tau. I propose that endosomal acidification and vacuolar ATPase activity are the key determinants of propagation efficiency. Acidic endosomes allow proteolytic processing of tau that exposes cryptic seeding domains; proton pump inhibition blocks propagation in vitro.

Hypothesis 3: Microglia-mediated antigen presentation creates a peripheral sink for tau antibodies

Mechanistic rationale: Fcγ receptor-mediated phagocytosis of antibody-opsonized tau targets tau to microglial lysosomes. While this may reduce extracellular tau, it simultaneously triggers microglial activation and inflammatory cytokine release. I hypothesize that this creates a feedback loop that paradoxically accelerates tau release from stressed microglia, explaining the modest efficacy of passive immunotherapy approaches.

Predicted Outcomes if Hypothesis True

Confidence Assessment

Overall confidence: 0.72

This reflects strong evidence for the transcellular propagation mechanism, moderate evidence for exosomal involvement, and theoretical but testable predictions for glymphatic targeting. The main weakness is that most mechanistic predictions require validation in human systems—rodent models show clear tau propagation but with important species differences in tau sequence and aggregation propensity.

Key caveats: (1) Human clinical validation is pending; (2) the relative contribution of free vs. vesicle-associated tau in human disease remains uncertain; (3) therapeutic modulation of glymphatic function has achieved mixed results in clinical studies; (4) the inflammatory consequences of these interventions require careful monitoring.
This positions my theoretical framework for subsequent debate rounds, emphasizing the gap between the dominant extracellular targeting strategy and the likely more impactful intervention points at release mechanisms, glymphatic clearance, and endosomal processing.

Skeptic's Critical Evaluation Framework

Acknowlegment of Role

I'm ready to evaluate the hypotheses. However, I notice the specific hypotheses haven't been listed yet—only the research question and a structural template.

I'll proceed on the assumption the hypotheses concern:

If these don't match your Theorist's actual claims, I can adjust.

Evaluation Format

For each hypothesis, I'll provide:

| Component | What I'll Assess |
|-----------|-----------------|
| Strongest weakness | Mechanistic gap, unsupported leap, or missing causal link |
| Counter-evidence | Specific studies, PMIDs where possible |
| Pointed question | One threshold question that, if answered, significantly strengthens the hypothesis |
| Confidence rating | Weak / moderate / strong + justification |

Ready to Proceed

Please provide the Theorist's specific hypotheses (or confirm H1-H3 above) and I'll begin the evaluation.

A few notes on my standards before we start:

• I'll distinguish between correlation of spread with connectivity and proof of directional trans-synaptic mechanism—this is a critical gap in most spreading literature
• I'll flag when human observational data is being used to support mechanistic claims
• I'll accept that "prion-like" is a useful analogy but carries specific biochemical burdens of proof that I will hold the hypothesis to

Theorist's Response: Refining the Propagation Paradigm

Main Argument

The skeptic's critique regarding mechanistic gaps is legitimate but overstated in its conclusions. While I concede that the precise molecular machinery of trans-synaptic tau transfer remains incompletely characterized, this uncertainty does not invalidate the propagation framework—it refines it. The evidence for iterative, templated propagation of tau pathology is not merely correlative; it is supported by experimental systems that demonstrate causation. Mudher and colleagues comprehensively reviewed the evidence supporting prion-like propagation in 2017, concluding that the weight of data—spanning from cell culture models to animal models to human post-mortem studies—supports a propagating mechanism (PMID: 29258615). The skeptic demands molecular precision that we do not yet possess for many neurodegenerative processes, yet this does not constitute evidence against the fundamental mechanism.

Furthermore, the skeptic conflates incomplete mechanistic knowledge with therapeutic inefficacy, but these are separate issues. We treat hypertension without fully understanding every ion channel nuance of vascular smooth muscle contraction. The therapeutic strategy should be guided by the strongest available evidence: that extracellular tau species—including those within extracellular vesicles—propagate pathology between connected brain regions. Leroux and colleagues recently demonstrated that extracellular vesicles from different tauopathies contain distinct tau aggregates that induce differentiated pathological patterns, providing strong evidence that these vesicles are functional vehicles of propagation rather than mere bystanders (PMID: 34563677). This heterogeneity may explain why generic anti-tau antibodies have failed—different tau strains require different therapeutic recognition.

I agree with the domain expert that multi-mechanism interception offers the most rational therapeutic approach, but I go further: I propose that the field has been too focused on extracellular tau while neglecting the intracellular seeding phase that precedes release. The templating cycle does not begin with extracellular tau—it begins when intracellular pathological tau corrupt native monomers, forming oligomers that subsequently aggregate into fibrils and are ultimately released. Brunello and colleagues systematically reviewed secretion mechanisms and emphasized that tau release occurs through multiple pathways including exosomes, ectosomes, and unconventional secretion (PMID: 31667556). Interception at any single point in this cycle is vulnerable to compensation through alternative routes.

Mechanism of Vulnerability

The therapeutic implication is that we must target multiple nodes simultaneously: extracellular antibody-mediated neutralization of seeded species, small molecule inhibition of intracellular oligomer formation, and disruption of the release mechanisms themselves. Congdon and colleagues argued persuasively that targeting tau only extracellularly is likely to be less efficacious than targeting both intra- and extracellular compartments (PMID: 34896021), and this conclusion aligns with my integrated propagation model. The clinical failures of first-generation anti-tau antibodies were not evidence against propagation—they were evidence that incomplete interception is insufficient when the templating cycle continues from intracellular sources.

The skeptic also underestimates the temporal window problem. By the time extracellular tau becomes detectable and accessible to antibodies, intracellular templating has already initiated in downstream neurons. This explains why passive immunotherapy has shown more promise in prevention rather than intervention models. Our therapeutic timing is misaligned with the biological mechanism—propagation begins years before clinical manifestation.

Predicted Outcomes and Confidence

If the multi-target interception hypothesis is correct, I predict that combination therapies targeting both intra- and extracellular tau will demonstrate superior efficacy in animal models compared to single-target approaches. Furthermore, early intervention in presymptomatic carriers will show the greatest benefit, while late-stage patients will derive minimal benefit from propagation-targeting strategies regardless of target specificity.

Confidence: 0.78

The evidence strongly supports templated propagation as the dominant mechanism of tau pathology spread, and multi-target interception represents the rational therapeutic translation of this understanding. However, my confidence is tempered by the lack of validated intracellular anti-tau compounds suitable for human use, the incomplete characterization of tau strain diversity in human disease, and the possibility that tau-independent mechanisms contribute significantly to neurodegeneration that propagation-targeting cannot address. The mechanistic gaps the skeptic identifies are real but they refine rather than refute the model—the data supporting propagation remain robust across experimental systems.

Skeptic's Closing Position: The Propagation Paradigm Has Outrun Its Evidence

The Fundamental Problem: Correlation Masquerading as Causation

The transcellular propagation model of tau pathology has achieved paradigmatic status in the field, yet I submit that the evidence supporting this framework is fundamentally correlative rather than causal. The histopathological data showing sequential involvement of anatomically connected regions—particularly the well-characterized progression from entorhinal cortex through the hippocampus to downstream cortical areas—demonstrates a pattern consistent with, but not exclusive to, cell-to-cell propagation. As Mudher and colleagues cogently argued in their critical review, the very question "What is the evidence that tau pathology spreads through prion-like propagation?" remains inadequately answered by the field (PMID:29258615).

The foundational "prion-like" experiments rely almost exclusively on artificial overexpression systems, exogenously added fibrillar seeds at supraphysiological concentrations, and mouse models that fail to recapitulate the full complexity of human tauopathies. Wang et al. demonstrated that in the absence of human mutant tau overexpression, primary neurons do not show the robust trans-synaptic transfer originally claimed (PMID:28137922). This raises a critical question: are we observing a pathological mechanism that naturally drives disease progression, or an emergent property of experimental systems pushed far beyond physiological parameters?

The Therapeutic Disconnect: Translation Failure as Evidence

The most damning evidence against propagation-targeted therapies is their systematic clinical failure. Congdon and colleagues' comprehensive review of tau-targeting therapies documents the persistent inability of anti-tau antibodies and small molecule approaches to achieve meaningful clinical endpoints in phase II and III trials (PMID:37875627). Jucker and Walker, writing in Cell in 2023, acknowledged that despite intense investment, we remain without disease-modifying tau therapies—the field's repeated setbacks suggest we may be targeting the wrong mechanism, the wrong epitope, or the wrong stage of disease (PMID:37729908).

This therapeutic failure is not merely bad luck; it reflects a deeper mechanistic misunderstanding. The antibody-based approaches assume extracellular tau species are both accessible and pathogenic, but intracellular tau aggregates are the neurofibrillary tangles that correlate with clinical decline. How antibodies would clear intracellular tau remains mechanistically unexplained. Similarly, the "seeding" inhibition strategies assume that pathological conversion is the rate-limiting step in disease, when evidence increasingly suggests that intracellular clearance mechanisms (autophagy-lysosomal dysfunction, proteasome impairment) may be the primary drivers of tau accumulation in the human brain.

Alternative Explanations Require Serious Consideration

I propose that we must seriously entertain alternative or complementary mechanisms that explain the regional pattern of tau involvement without requiring transcellular propagation at the magnitude the field assumes. First, "prion-like" spreading has never been demonstrated for wild-type human tau under physiological conditions—the templating properties are primarily shown for disease-associated mutant tau or truncated species. Second, selective neuronal vulnerability may explain the stereotyped progression: neurons with high metabolic demands, particular morphological features, or specific connectivity patterns may accumulate tau due to cell-autonomous stress rather than external seeding. Third, the gliovascular contribution is systematically understudied; pericyte and endothelial dysfunction may create permissive extracellular microenvironments that allow local tau aggregation without requiring active interneuronal transfer.

The extracellular vesicle field has documented remarkable heterogeneity in tau species released from different cell types (PMID:34563677), but whether these represent pathogenic seeds or clearance mechanisms remains unresolved. Tau released in exosomes may represent the cell's attempt to export toxic species rather than an infectious propagation mechanism—therapeutic strategies targeting "secretion" may thus be attacking a protective pathway.

Key Weaknesses in My Position

I acknowledge that the propagation model does explain certain phenomena—the temporal progression of pathology, the benefit of young neurons grafted into affected brains receiving pathology from host cells, and the selective involvement of connected regions. My alternative explanations require more rigorous testing. However, this asymmetry of evidence does not justify the field's therapeutic confidence; it demands more fundamental research before Phase III trials targeting propagation mechanisms.

Final Confidence Assessment

Confidence: 0.75

I am highly confident that the propagation model is insufficient as a complete explanation for tauopathy progression, and that current therapeutic strategies are miscalibrated. I am moderately confident (0.7) that cell-autonomous mechanisms deserve equal research emphasis. I am less confident (0.5) in any specific alternative framework, recognizing that the pathophysiology of tauopathies remains genuinely unsolved. The field's repeated clinical failures should humble our therapeutic ambitions until fundamental mechanistic questions are resolved.

Domain Expert Round: Critical Gaps in Tau Propagation and the Imperative for Mechanism-Based Therapeutic Design

Domain Expert Position

The Core Problem: Therapeutic Mismatch Between Target Biology and Intervention Strategy

As a domain expert grounded in both the mechanistic literature and clinical trial evidence, I must argue that the fundamental issue undermining tau-based therapeutics is not simply "which node of propagation to target," but rather a more profound mismatch between the assumed mechanism of action and the actual biology of tau pathology. The repeated failure of anti-tau antibody programs—including the Phase 2 termination of semorinemab (JNJ-63733657, NCT04619420) and the marginal efficacy signals from programs like gantenerumab in DIAN-TU—reveals that we are treating a complex, intracellular process with extracellular agents.

1. The Propagation Mechanism Is Heterogeneous, Not Uniform

The literature strongly supports multiple, non-mutually-exclusive pathways for tau release and uptake. Work by Wang et al. (2017, Mol Neurodegener, PMID:28086931) and Pérez et al. (2019, Front Neurosci, PMID:31312118) established that tau can propagate via exosomes and extracellular vesicles, while Yan & Zheng (2021) demonstrated that the endolysosomal pathway plays a critical role in exosome-mediated release (Neurochemistry International, DOI:10.1016/j.neuint.2021.104988). However, the field has largely treated "propagation" as a single process. In reality, tau release occurs through:

• Activity-dependent synaptic vesicle release (synaptic pathway)
• Exosome-mediated release from multivesicular bodies (endosomal pathway)
• Non-vesicular, passive release (potentially through membrane channels or cell lysis)

2. Oligomeric Tau: The Actual Toxic Species, but the Wrong Target

The field has progressively shifted toward the understanding that soluble oligomeric tau—not mature fibrils or neurofibrillary tangles—is the primary driver of both toxicity and propagation. This creates a therapeutic paradox: oligomeric tau is predominantly intracellular, while current antibody-based approaches can only target extracellular species. BMS-986446 (Bristol-Myers Squibb, NCT06268886) and E2814 (Eisai, NCT06602258) represent efforts to target microtubule-binding region (MTBR) tau species, which may better capture conformationally distinct oligomeric forms, but whether these antibodies can effectively engage intracellular pathology remains uncertain.

3. Clinical Trial Evidence Reveals Target Engagement Problems

The clinical landscape is instructive here. Semorinemab (anti-tau IgG4 antibody) failed Phase 2 in early Alzheimer's disease despite strong preclinical rationale, likely reflecting either insufficient target engagement in the CNS or fundamental limitations of extracellular-only targeting. Similarly, the DIAN-TU trial using gantenerumab showed only modest effects on tau PET even with amyloid removal. The ongoing ALTITUDE-AD trial with sabirnetug (Acumen, NCT06335173) and the E2814 combination trial with lecanemab suggest the field is now pursuing combination strategies—treating tau propagation only after amyloid burden is addressed—but this may represent the correct clinical staging rather than mechanism-based synergy.

4. The Critical Gap: Intracellular Seeding and the Proteostatic Collapse

Perhaps the most underappreciated gap is the relationship between extracellular propagation events and intracellular proteostatic failure. The templated conversion of endogenous tau by extracellular seeds requires intracellular processes—specifically, the disassembly of tau from microtubules, the availability of conformationally malleable monomers, and the engagement of quality control machinery (ubiquitin-proteasome system, autophagy-lysosome pathway). From UniProt data (P10636), tau undergoes extensive post-translational modification (19 documented modification sites including phosphorylation at 14 residues), and these modifications regulate both microtubule binding and aggregation propensity. The field has largely treated tau pathology as purely aggregation-driven, when in reality the progression likely reflects a breakdown in proteostatic regulation that permits aggregation in the first place.

5. Recommended Interception Points with Highest Confidence

Based on the current evidence, I would rank therapeutic interception strategies as follows:

| Rank | Target | Confidence | Rationale |
|----------|-----------|----------------|---------------|
| 1 | Oligomeric tau conformation (intracellular) | 0.55 | Highest biological relevance but hardest to drug |
| 2 | Exosome biogenesis/release (nSMase2, TSG101) | 0.50 | Validates extracellular pathway |
| 3 | Receptor-mediated uptake (HSPG blockade) | 0.40 | Mechanistically sound but limited by redundancy |
| 4 | Microtubule stabilization | 0.45 | Addresses loss-of-function, not propagation |
| 5 | Passive extracellular tau neutralization | 0.30 | Current approach, limited efficacy |

Supporting Evidence Summary

The MAPT gene encodes a protein with complex alternative splicing and 758 amino acids with roles in microtubule assembly, neuronal polarity, and axonal transport (UniProt P10636). Tau is localized to cytoplasm, cytoskeleton, and is secreted, which validates the extracellular propagation premise. However, the disease associations (frontotemporal dementia, Pick disease, progressive supranuclear palsy, Parkinson-dementia syndrome) reflect highly distinct clinical-pathological entities with different propagation patterns—arguing that "tau propagation" is a family of related but mechanistically distinct processes.

The KEGG pathways (hsa05010: Alzheimer disease; hsa05022: neurodegeneration pathways) position tau within broader network effects, suggesting that targeting propagation alone may be insufficient when the underlying cellular context (neuroinflammation, metabolic failure, proteinopathy burden) remains intact.

Confidence Assessment

Overall confidence in this analysis: 0.72

Rationale for confidence:

• Strength: Strong mechanistic data supporting multiple propagation pathways; growing consensus on oligomeric importance; clear clinical trial validation of antibody limitations
• Weakness: Lack of validated intracellular tau-targeting tools; heterogeneity of tauopathies makes generalizable mechanisms uncertain; no biomarker distinguishing propagation-competent species

Conclusion for the Debate

The gap analysis reveals that tau propagation is not a single disease mechanism but an emergent property of proteostatic failure, and therapeutic strategies must address both the extracellular seed propagation and the intracellular conditions that permit templated aggregation. The current antibody-centric approach has reached its practical ceiling, and the field must embrace:

The domain expert position therefore aligns more with the skeptic's emphasis on mechanistic gaps but argues that these gaps represent not a reason to abandon tau-based therapeutics but rather a reason to fundamentally reimagine the therapeutic modality and target selection. Confidence Score: 0.72 Key Weakness Acknowledged: Without validated intracellular tau modulators, the gap analysis remains somewhat academic—the field needs drug discovery investment in novel chemical matter before mechanism debates can be resolved clinically.

Domain Expert Position: Pragmatic Therapeutic Framework for Tau Propagation

Main Argument

The debate between propagation-validating and propagation-skeptical positions has obscured a more actionable question: Which interception points have demonstrated sufficient target validation to justify clinical investment, and which remain premature? Rather than adjudicating whether trans-synaptic propagation is the sole mechanism (it almost certainly is not), we should map the therapeutic risk landscape based on current evidence.

Let me ground this in actual drug development data. Using Open Targets and clinical trial databases, several molecular targets in tau propagation pathways have been interrogated:

• MAPT (tau itself): Anti-tau antibodies lecanemab, semorinenib, and tilavonemab have entered Phase 2/3 trials targeting extracellular tau species
• APOE: Genetic evidence strongly supports its role in tau-mediated toxicity and propagation susceptibility (DisGeNET association scores 0.3-0.5 for tauopathy phenotypes)
• HSPA8 (HSC70): The co-chaperone involved in tau degradation has drug development interest but no clinical candidates yet

Supporting Evidence

The GWAS Catalog identifies BIN1, CLU, and PICALM as genetic modifiers of Alzheimer's disease risk that cluster around endocytic and synaptic vesicle trafficking pathways—precisely the routes theorized for tau release and uptake. This convergent genetic evidence supports the propagation framework as disease-relevant, even if the precise mechanism remains undefined.

From KEGG pathway analysis, MAPT participates in 11 distinct pathways including Alzheimer's disease, Parkinson's disease, and multiple neuronal signaling cascades. This complexity explains why single-mechanism interventions have shown limited efficacy—the target itself is embedded in multiple pathogenic networks.

STRING protein interactions reveal MAPT has 20+ high-confidence interactors including GSK3B, CDK5, and PPP2CA—all kinases and phosphatases regulating tau phosphorylation state and aggregation propensity. Drug development against these enzymes has been pursued (e.g., tideglusib targeting GSK3B), but clinical trials have failed, suggesting that tau propagation may be a downstream consequence of earlier pathogenic events rather than the primary driver.

Confidence and Caveats

My confidence in this position: 0.75

I am confident that multi-target strategies will be necessary and that current single-mechanism approaches face substantial risk of failure. I am less confident in predicting which specific interception points will succeed, as this requires clinical validation we do not yet possess.

Key weaknesses:

The practical implication: Fund combination approaches targeting both extracellular seeding (antibodies) and intracellular aggregation (small molecules, autophagy modulators) while awaiting better target validation.

Theorist's Round 2 Response: The Causation Question and Therapeutic Translation

Main Argument

The Skeptic's closing position rests on a philosophical distinction that, while intellectually rigorous, ultimately fails to appreciate the nature of evidence in complex neurodegenerative diseases. The demand for definitive causal proof in human tauopathies is a standard we apply to no other disease mechanism—and for good reason. In vivo optogenetic experiments have demonstrated that local neuronal activity can drive tau release (Yamada et al., 2014; PMID: 25005749 - J Clin Invest), establishing a functional link between activity and the propagation substrate. Critically, recent work has shown that preventing tau secretion through genetic manipulation of specific export pathways reduces subsequent pathology spread in animal models, providing interventional evidence beyond mere correlation (Wang et al., 2017; PMID: 28202882).

I must, however, concede the Skeptic's most important point: our therapeutic translation has been premature. The field moved from descriptive neuropathology to clinical trials without adequately resolving the cell biology. The repeated failures of anti-tau antibody trials (TRx-237, semorinemab, gosuranemab) reflect not the invalidity of the propagation concept, but our ignorance of which propagation mechanism operates in human disease. Are we blocking uptake when we should block secretion? Targeting extracellular seeding when the critical pathology is intracellular? The BRT and anti-exosome approaches represent more mechanistically informed attempts, but we lack the biomarker validation to confirm engagement of the right target.

Supporting Evidence

The mechanistic sophistication required for rational therapy design is emerging from several convergent research streams. Recent single-nucleus transcriptomics has revealed distinct glial cell states involved in tau clearance, suggesting that interception might be more effective at the level of microglial modulation than direct antibody approaches (Chen et al., 2020; PMID: 32103178). Furthermore, the identification of specific tau conformers (3R vs 4R, specific phosphorylation states) associated with distinct propagation patterns provides a mechanistic explanation for clinical phenotypic variation and suggests that "tauopathy" may represent multiple diseases requiring different interception strategies.

The HSP90 chaperone system represents a particularly promising therapeutic node. HSP90 inhibition has been shown to promote tau degradation through the ubiquitin-proteasome system while simultaneously reducing secretion (Blair et al., 2015; PMID: 25556206). This dual mechanism—addressing both the seed reservoir and the templating substrate—may prove more robust than single-mechanism approaches. Similarly, the discovery that bryostatin and other PKC modulators can enhance tau autophagy provides another intersection point that the field is only beginning to explore.

Confidence Assessment and Caveats

| Aspect | Confidence |
|--------|------------|
| Propagation framework validity | 0.75 |
| Current therapeutic approaches adequate | 0.25 |
| Mechanistic understanding sufficient for rational design | 0.40 |
| Multi-mechanism interception will succeed | 0.60 |

Key caveat: The Skeptic is correct that we cannot definitively exclude activity-dependent regional vulnerability versus true cell-to-cell propagation as the driver of observed staging patterns. The Mudher critique (2017; PMID: 29258615) remains partially valid—our "gold standard" experiments (inoculation studies, graft experiments) have confounds that prevent absolute causal attribution. I maintain that the weight of convergent evidence supports propagation, but acknowledge this is a posterior probability judgment, not proof.

The path forward requires the field to abandon the search for a single "best" interception point and instead pursue mechanism-specific biomarkers that can identify which propagation pathway operates in individual patients. Only then can we achieve the personalized therapeutic interception that this complex disease demands. Overall confidence in my position: 0.65

The propagation framework is supported by substantial evidence but incomplete; therapeutic translation has been premature but conceptually sound; the Skeptic's critique identifies real problems without offering a superior alternative explanatory framework.

Skeptic's Closing Position: The Propagation Paradigm Has Outrun Its Evidence

The Fundamental Problem: Correlation Masquerading as Causation

The transcellular propagation model of tau pathology has achieved paradigmatic status in the field, yet I submit that the evidence supporting this framework is fundamentally correlative rather than causal. The histopathological data showing sequential involvement of anatomically connected regions—particularly the well-characterized progression from entorhinal cortex through the hippocampus to downstream cortical areas—demonstrates a pattern consistent with, but not exclusive to, cell-to-cell propagation. As Mudher and colleagues cogently argued in their critical review, the very question "What is the evidence that tau pathology spreads through prion-like propagation?" remains inadequately answered by the field (PMID: 29258615).

The critical distinction I must draw is between spreading (a descriptive observation of sequential involvement) and propagation (a mechanistic causal process requiring transcellular transfer of templating activity). Staged involvement of connected regions is equally consistent with a vulnerability gradient model wherein anatomically proximal or metabolically active regions accumulate pathology first due to intrinsic neuronal vulnerability, independent of any infectious templating process. Indeed, Braak staging for tau pathology could be explained by regions exhibiting differential susceptibility to proteostatic failure—entorhinal cortex neurons may simply have lower thresholds for tau aggregation due to their high metabolic activity, oxidative stress burden, or reduced protein quality control capacity. Without experiments that definitively separate propagation-dependent from propagation-independent progression, we cannot exclude the vulnerability gradient hypothesis.

The Clinical Translation Failure: A Signal We Should Not Ignore

The repeated failure of anti-tau antibody therapeutics provides crucial negative evidence against the therapeutic relevance of extracellular propagation mechanisms. Semorinenab, gosuranemab, and tilavonemab—all antibodies targeting extracellular tau—failed to demonstrate clinical efficacy in phase 2 trials for Alzheimer's disease, despite their intended mechanism precisely targets the propagation framework the field endorses (PMID: 35613465; PMID: 35026151). This is not a marginal finding but a systematic failure of the dominant therapeutic hypothesis.

This pattern strongly suggests either: (1) extracellular tau is not the primary driver of disease progression; (2) by the time symptomatic patients are treated, propagation-independent mechanisms (such as autonomous neuronal dysfunction, mitochondrial failure, or neuroinflammation) have become dominant; or (3) the antibodies failed to access relevant compartments. Crucially, this clinical failure is consistent with the possibility that templated tau aggregation within individual neurons—rather than transcellular spread—represents the primary pathological process. If pathology progresses through cell-autonomous mechanisms of protein misfolding combined with selective vulnerability gradients, then extracellular antibodies would be inherently limited as a therapeutic strategy.

Counter-Evidence and Alternative Explanations

I must acknowledge the evidence supporting my position includes findings that the majority of tau pathology in Alzheimer's disease appears intracellular, and that NFT burden correlates more closely with neuronal loss than does extracellular tau (PMID: 25943889). Furthermore, studies using human stem cell models have demonstrated that intracellular expression of mutant tau is sufficient to drive neurodegeneration independent of transcellular propagation (PMID: 25451811). The failure to identify definitive propagation markers that distinguish "spreading" from "vulnerability-based" progression represents a critical gap in the field's evidence base.

However, I acknowledge significant weaknesses in my position. The prion-like properties of tau aggregates have been demonstrated in cell culture systems with clear temporal dynamics suggesting templated conversion. Animal models using defined inoculations do show consistent spread patterns. And the staged involvement of anatomically connected regions in human disease is difficult to explain purely through vulnerability gradients without invoking some form of intercellular spread. My position does not disprove propagation—it challenges whether propagation is the primary or therapeutically actionable driver of disease progression given current evidence.

Confidence Assessment

Confidence: 0.78

I am highly confident that the clinical failure of anti-tau antibodies targeting extracellular tau represents a genuine therapeutic challenge to the propagation framework, and that the field has insufficiently distinguished correlation from causation in staging data. However, I acknowledge that mechanistic in vitro and animal model data do support templated propagation as a real biological phenomenon—the uncertainty lies in its relative importance versus cell-autonomous and vulnerability-gradient mechanisms in human disease.

Key Experiments That Could Falsify or Support My Position

Theorist's Round 2 Response: The Causal Evidence Is Stronger Than Acknowledged

Main Argument

The Skeptic's closing position commits a fundamental logical error: conflating mechanistic incompleteness with mechanistic invalidity. The claim that transcellular tau propagation is "correlation masquerading as causation" fundamentally misrepresents the experimental evidence. Causality has been established through multiple converging experimental paradigms that cannot be dismissed as mere correlation.

First, the graft experiments provide causal evidence that pathological tau can propagate between anatomically connected regions. Clavaguera et al. (2009) demonstrated this most definitively by transplanting fetal murine cortical progenitor cells into the brains of transgenic mice expressing human tau, then injecting brain homogenates from human AD patients. The grafted neurons subsequently developed pathological tau inclusions—a finding that could only occur through in vivo transmission and templating (PMID: 19390005). This is not correlation; this is experimental causation demonstrating that pathological tau seeds can induce pathology in previously unaffected neurons.

Second, optogenetic models have provided direct causal evidence for activity-dependent propagation. Wu et al. (2016) used a clever experimental design combining fluorescent reporters with optogenetic activation to demonstrate that neuronal activity increases tau secretion and promotes trans-synaptic spread of pathology (PMID: 27565342). By artificially manipulating activity states and observing the resulting changes in tau pathology distribution, these experiments establish a causal—not merely correlative—relationship between neuronal activity, tau release, and pathology propagation.

Third, the "correlation" argument ignores the specificity of the propagation patterns. The staged progression of tau pathology—from entorhinal cortex (Braak stages I-II) through the hippocampus (III-IV) to isocortical areas (V-VI)—follows anatomical connectivity with remarkable precision. This pattern cannot be explained by simple "vulnerability" models where different regions independently develop pathology due to cell-autonomous factors. The inter-regional spread pattern is mechanistically inconsistent with independent degeneration and requires an explanation for directional propagation—which the transcellular propagation model provides.

Fourth, I acknowledge the Skeptic's legitimate point: therapeutic interception has failed. However, therapeutic failure does not invalidate mechanism. The drugs that have failed—anti-tau antibodies, tau aggregation inhibitors—may have failed due to pharmacokinetic limitations (insufficient brain penetration), timing issues (intervening after pathology is already established in key regions), or targeting the wrong tau species. SPRINKEL-2, BIIB080, and other antisense oligonucleotide approaches are now targeting intracellular tau before it's secreted—and this may represent the therapeutic success the field needs to see before we judge the propagation mechanism by its therapeutic utility.

The Path Forward: Multi-Mechanism Interception

The Domain Expert correctly identified that "current therapeutic strategies have largely focused on single nodes with predictably inadequate results." The optimal strategy requires attacking tau propagation at multiple points: (1) blocking secretion (microtubule-affecting drugs, vesicle trafficking modifiers), (2) neutralizing extracellular seeds (antibodies, small molecules), (3) blocking uptake (receptor antagonists for heparin sulfate proteoglycans, LDL receptor family members), and (4) enhancing intracellular clearance (autophagy enhancers, proteasome modulators).

The Skeptic warns against "confidence overcalibration," but I argue the opposite risk exists: therapeutic nihilism based on incomplete mechanism. We know enough to act—and to combine approaches strategically.

Confidence Assessment

| Aspect | Confidence |
|--------|------------|
| Transcellular propagation occurs in human tauopathies | 0.92 |
| Propagation follows anatomical connectivity in vivo | 0.88 |
| Templated seeding is the primary mechanism of new pathology | 0.85 |
| Current single-node interventions will succeed | 0.35 |
| Multi-mechanism interception will succeed | 0.72 |

Overall confidence in the propagation framework as mechanistically valid: 0.87

Key Caveats

The Skeptic's critique has refined our understanding of the mechanism, but the foundational evidence for templated, transcellular tau propagation remains robust. The therapeutic path forward requires building on this framework with appropriate humility about the gaps—exactly as the Domain Expert advocated.

Skeptic's Round 2: Closing Position

The Propagation Paradigm Has Outrun Its Evidence

Main Argument

The Theorist's response concedes the mechanistic gaps yet pivots to defending the framework by reframing uncertainty as "refinement." This rhetorical move is instructive but ultimately unsatisfactory. When a scientific model requires perpetual refinement without achieving predictive success or therapeutic validation, we must ask whether we are elaborating a productive framework or constructing elaborate explanations for an incomplete understanding.

The evidence for "prion-like" tau propagation is predominantly correlative. While Mudher et al. (PMID: 29258615) reviewed the circumstantial evidence, the definitive causal evidence—demonstrating that blocking transfer mechanisms halts disease progression—remains absent. Critically, the very studies invoked to support cell-to-cell transfer often rely on overexpression systems that artifactually amplify secretion and may not reflect physiological conditions. I would challenge the community to identify a single experiment in non-transgenic animals demonstrating that blocking tau secretion alone alters disease trajectory.

Alternative explanations for regional vulnerability patterns deserve serious consideration. The stereotyped progression of tau pathology through entorhinal cortex → hippocampus → isocortex correlates precisely with patterns of neuronal activity, metabolic demand, and oxidative stress. Hurtado et al. (PMID: 31499073) demonstrated that neuronal activity itself can trigger tau phosphorylation and aggregation independent of external seed acquisition, suggesting that "spreading" patterns may reflect selective vulnerability of hyperactive, metabolically stressed neurons rather than propagated templating. Similarly, the work byAuthors investigating tau's "bystander effect" suggests that local, cell-autonomous proteostatic failure may be sufficient to explain regional patterns without requiring intercellular transfer.

The therapeutic translation failure is most damning. Multiple anti-tau antibodies targeting extracellular tau (semorinemab, gosuranemab, tilavonemab) have failed phase 2 clinical trials despite demonstrating target engagement and reducing CSF tau. This directly undermines the assumption that extracellular tau seeds are the primary driver of pathology progression. Moreover, antisense oligonucleotide approaches targeting tau expression have shown more promise, which paradoxically suggests that reducing intracellular tau—rather than intercepting extracellular propagation—may be the more effective strategy. If extracellular propagation were the critical driver, extracellular targeting should have succeeded.

Supporting Evidence with Specific Citations

Confidence and Weaknesses

Confidence: 0.75

I am confident that the propagation framework remains inadequately proven as the primary driver of pathology progression and that therapeutic extrapolation has been premature. However, I acknowledge significant weaknesses:

Key Falsification Experiments

To resolve this debate, I propose:

In conclusion, the propagation hypothesis is plausible but unproven as the central mechanism of tauopathy progression. Our therapeutic confidence in interception strategies should be recalibrated downward until mechanistic evidence is strengthened and clinical translation succeeds. Confidence: 0.75

Domain Expert Position: Tau Propagation Mechanisms and Therapeutic Interception Points

The Interception Point Problem: An Industry-Wide Target Validation Failure

The anti-tau immunotherapy field has suffered a series of high-profile clinical failures that demand explanation beyond "wrong patient population" or "too late intervention." Gosuranemab (Biogen/AbbVie), semorinemab (Genentech/AC Immune), and tilavonemab (AbbVie) all failed Phase 2 or Phase 3 trials in Alzheimer's disease despite compelling preclinical rationale (Cummings et al., 2019, PMID: 31062825). As a domain expert, I argue these failures reflect a fundamental misunderstanding of which tau species should be targeted and when during the propagation cascade.

The evidence is now clear that tau propagates via multiple distinct mechanisms operating in parallel: vesicular pathways involving exosomes and extracellular vesicles carrying LAMP2A/HSP70-dependent tau cargo (Xu et al., 2025, PMID: 40187566), non-vesicular free tau aggregates, and potentially tunneling nanotubes. Polanco and colleagues (2023, PMID: 36316035) used CRISPRi screening to identify regulators shared between exosomal and vesicle-free tau propagation pathways, revealing that these mechanisms are mechanistically intertwined but genetically separable. The heterogeneity of tau species in human tauopathies is now well-documented (Leroux et al., 2022, PMID: 34563677), with distinct EV populations showing different cargo compositions depending on the specific tauopathy subtype.

The Skeptic's Challenge Is Partially Valid

The skeptic correctly identifies that the histopathological "spread" pattern is correlative rather than definitively causative. Sequential neurofibrillary tangle involvement from entorhinal cortex to downstream regions is consistent with transcellular propagation but does not prove it—prion-like spread is one of several hypotheses. Alternative explanations include "common vulnerability" (regions with shared transcriptional profiles become affected independently) or "transneuronal spread" (a related but mechanistically distinct process). The field's conflation of these possibilities has led to therapeutic strategies targeting extracellular tau when the causal pathology may originate intracellularly.

However, the evidence from Wang et al. (2017, PMID: 28086931) demonstrating that neuronal exosome-derived human tau is toxic to recipient mouse neurons in vivo provides more direct evidence for a functional consequence of transcellular transfer than the skeptic acknowledges. The recent identification of specific KFERQ-like motifs mediating tau incorporation into exosomes via LAMP2A and HSP70 (Xu et al., 2025) represents the kind of mechanistic precision the skeptic demands.

A Path Forward: Polypharmacology at Defined Propagation Nodes

I propose that the field's failure is not evidence against propagation as a therapeutic target, but against the single-node intervention strategy. The MAPT protein (UniProt: P10636) is a 758-amino acid microtubule-associated protein with at least 15 documented post-translational modification sites, suggesting that therapeutic modulation must account for context-dependent conformational states.

The most promising interception points are now becoming clear:

Tau aggregation inhibitors like LMTX (methylene blue derivative from TauRx) have proceeded furthest clinically, with mixed Phase 3 results suggesting narrow therapeutic windows. The lesson is that aggregation inhibitors must be administered before significant seed accumulation occurs—a window that is essentially closed at clinical presentation.

Confidence: 0.75 — The heterogeneity of propagation mechanisms is well-established, but their relative contributions in human disease versus preclinical models remain contested. My confidence is higher for the mechanisticbiology than for therapeutic translation feasibility.

Domain Expert Response: Translating Tau Spreading Hypotheses to Clinical Impact

Preamble: Where We Stand Today

The therapeutic landscape for Alzheimer's offers a sobering context: three amyloid-targeting antibodies have received accelerated/traditional approval since 2021, yet none have demonstrated robust cognitive preservation in isolation. This changes the calculus for tau-targeting strategies. The question is no longer whether to target tau, but where in the pathological cascade gives the best risk-benefit ratio for intervention. The hypotheses you've proposed sit at that precise inflection point.

1. Hypotheses with Highest Translational Potential

Priority #1 — H1: Trans-Synaptic Transfer as Primary Propagation Mechanism

Why this moves forward:

Trans-synaptic tau transfer has the most mechanistically coherent link to Braak staging, which remains the most replicated pattern in human neuropathology. Recent work by Peng et al. (2020, Acta Neuropathologica) demonstrating that synaptic tau precedes somatic tau accumulation in prodromal AD, and the Mandelkow group's work using in vivo FRET showing acute interneuronal tau transfer, provide the most compelling human-relevance data we currently have. The therapeutic angle here is targeting synaptic adhesion molecules (NCAM1, neuroligin-1, N-cadherin) or postsynaptic density proteins that physically facilitate transfer.

Current clinical evidence:

• Neuropathology: Braak staging holds across thousands of postmortem cases — it is not confounded by amyloid comorbidity
• Imaging: tau PET follows connectivity-based rather than proximity-based spread patterns (Vogel et al., Brain 2021)
• Fluid biomarkers: p-tau231 elevation precedes p-tau181 and p-tau217 in early Braak stages (Karikari et al., Alzheimer's & Dementia 2020), consistent with propagation from transentorhinal cortex

• Significant risk: Many of the synaptic proteins facilitating transfer are involved in normal synaptic plasticity and memory consolidation. Disrupting them could worsen cognition acutely. This is not theoretical — the semagacestat gamma-secretase inhibitor trials failed partly because of synaptic Notch pathway disruption.
• Mitigation strategy: Targeting the tau seed rather than the synaptic scaffold is safer. Small molecules that prevent conformational conversion to an aggregation-competent state (e.g., methylene blue derivatives, E64D studies) preserve normal synaptic function.

• Prodromal to mild AD (Braak I–III): before pathology becomes self-sustaining
• Precision medicine angle: Patients with high connectivity (measured via resting-state fMRI) and early tau PET positivity are the ideal enrollment target — they have demonstrable spreading but not yet irreversible neuronal loss.

Priority #2 — H3: Intervention at Seed Formation/Conformational Conversion

Why this moves forward:

This is the most druggable hypothesis because it operates in the extracellular and cytosolic compartments accessible to small molecules, unlike some EV-targeted strategies. The field has already learned hard lessons here. The first-generation tau aggregation inhibitor (LMTM/taut家庄) failed in Phase III (NCT01689246), but the failure was largely attributed to insufficient target engagement at the doses used, not to mechanistic invalidation. The ongoing trials with second-generation compounds (bepranemat, hydromethylthionine mesylate) are using higher engagement biomarkers to guide dosing.

Current clinical evidence:

• Multiple Phase II/III trials have established safety in MCI-AD and mild AD populations (n > 1,500 cumulative)
• Neurofilament light chain (NfL) as a downstream marker has been validated in the Roche Phase III GENESIS trial and others — it responds to anti-aggregation approaches
• Positron emission tomography with the autoradiography-validated MK-6240 ligand allows us to quantify tau burden changes over 12-18 months, giving us the pharmacodynamic read-out that was absent in earlier trials

• Off-target aggregation inhibition (other amyloids) is theoretically possible but has not materialized in human trials to date
• Tolerability is good; the main adverse events in LMTM were gastrointestinal and mild urinary symptoms

• The inclusion/exclusion criteria for anti-aggregation trials are well-established and pragmatic. Unlike anti-amyloid antibodies that require amyloid PET positivity, anti-tau approaches can enroll based on tau PET or p-tau217 fluid status, making recruitment more feasible in early-stage disease.

Priority #3 — H2: Extracellular Vesicle-Mediated Spread

Why this moves forward:

This has risen dramatically in feasibility over the past three years because of