Towards mechanism-based tau-targeted therapies

Tau plays a crucial role in several neurodegenerative diseases, collectively referred to as tauopathies. Therefore, targeting potential pathological changes in tau could enable useful therapeutic interventions. However, tau is not an easy target because it dynamically interacts with microtubules and other cellular components, which presents a challenge for tau-targeted drugs. New cellular models could aid the development of mechanism-based tau-targeted therapies.

Tauopathies are a group of neurodegenerative diseases characterized by increased phosphorylation and aggregation of the microtubule-associated protein tau (Arendt et al., 2016). Certain tauopathies, such as Frontotemporal Dementia with Parkinsonism linked to Chromosome 17 (FTDP-17), are caused by mutations in the tau gene, while in the most common tauopathy, Alzheimer’s disease (AD), neurofibrillary tangles composed of tau aggregates are accompanied by the presence of senile plaques composed of amyloid-beta peptide. However, even in AD, tau pathology correlates much better with progressive neurodegeneration than the formation of senile plaques, indicating that alterations in tau protein are pivotal in the neurodegeneration process. Thus, focusing on tau and its pathological modifications could offer a promising strategy to mitigate neurodegenerative progression.

Yet, tau is not an easy target. The tau gene arose through a gene duplication event early in vertebrate evolution from a common precursor gene with microtubule-associated protein 2 (MAP2), another microtubule-associated protein (Sundermann et al., 2016). While MAP2 is predominantly present in the somatodendritic compartment of mature neurons, tau is enriched in the axon. Sequencing data indicates that all vertebrates possess at least one tau gene, highlighting its essential role across species. Therefore, it was surprising that knocking out tau in mice did not result in major phenotypic changes and the animals developed functional axons (Harada et al., 1994). This also implied that contrary to the traditional view, the main role of tau is not to stabilize microtubules (Baas and Qiang, 2019). Consequently, it cannot be assumed that the loss of tau function due to tau aggregation in disease directly leads to a breakdown of the axonal microtubule array, as is still portrayed in many schematic representations of disease mechanisms.

What then is the biological activity and function of tau? In the axon, tau exhibits a highly dynamic interaction with microtubules, which we termed “kiss and hop” (Janning et al., 2014). While more than 80% of tau is present on microtubules at any given time, the residence time of a single tau molecule at a microtubule-binding site is only about 40 ms in cultured neurons. Despite this short interaction time, tau is still able to modulate microtubule dynamics. The “kiss and hop” interaction is likely crucial for tau’s ability to modulate microtubule dynamics without affecting vital axonal transport processes that also occur on axonal microtubules. Indeed, modifications of tau that increase its residence time on microtubules impair the efficiency of axonal transport, leading to neuronal atrophy (Conze et al., 2022). In addition to the changes that could be a direct consequence of the microtubule interaction activities of tau, several more subtle changes have been observed in mice lacking tau protein, including alterations in synaptic plasticity, memory formation, and anxiety-related behavior, suggesting additional functions of tau (Brandt et al., 2020). Indeed, in addition to the microtubule-binding region (MBR), tau carries a long N-terminal projection region containing intrinsically disordered regions. Functionally, intrinsically disordered regions could be relevant because they offer a large interaction surface, thus allowing the binding of many interaction partners. Indeed, several binding partners of tau’s N-terminal projection region have been identified, including plasma membrane components and members of various signaling cascades.

The most obvious approach to prevent tau-dependent neurodegeneration would be to inhibit or at least reduce disease-related changes in tau protein, such as increased phosphorylation or the formation of tau aggregates. However, several aspects regarding the potential physiological role of tau need to be taken into account (Figure 1). Regarding tau phosphorylation, most of the sites showing increased phosphorylation (hyperphosphorylation) during disease are highly conserved serine and threonine residues flanking tau’s MBR (Trushina et al., 2019). These include binding sites for commonly used antibodies to detect the hyperphosphorylated state of tau protein, such as the AT8 antibody (phosphorylation at Ser202/Thr205) or the PHF1 antibody (phosphorylation at Ser396/Ser404). However, the interaction of tau with microtubules is also regulated by phosphorylation at regions flanking the MBR of tau, and in many cases, reduced phosphorylation of these sites leads to increased interaction of tau with microtubules. As mentioned above, the highly dynamic interaction of tau with microtubules is required for tau not to disrupt axonal transport. Therefore, reduced phosphorylation below physiological levels may generate hyperactive tau proteins that disrupt axonal transport and cause neuron degeneration. Furthermore, tau is not the only target for kinases and phosphatases and affecting phosphorylation-modulating proteins can also affect the phosphorylation states of many other target proteins, potentially having functional consequences.

Figure 1.

Dynamics of the tau protein in healthy and diseased states.

Reactions whose modulation could be therapeutically useful are marked in red. See text for details. Hyperphosphorylation at the proline-rich region is marked by “PPP”. k_on, k_off: On- and off-rate constants of the tau-microtubule interaction; NFTs: neurofibrillary tangles; PHFs: paired helical filaments.

Undesirable consequences could also be caused by approaches aimed at reducing tau aggregates. The MBR of tau is involved in both binding to microtubules and the formation of tau aggregates. Drugs that bind to the MBR of tau and thus prevent the formation of tau aggregates may therefore also disrupt the microtubule interaction of tau and thus the physiological regulation of microtubule dynamics. Furthermore, the MBRs of tau and MAP2 are highly conserved, so drugs that bind to the MBR of tau most likely also affect the interaction of MAP2 with microtubules, which may lead to dendritic defects. Additionally, it is still controversial what constitutes the bioactive and toxic species of the aggregation process. Experimental results suggest that soluble tau aggregates are the cause of toxicity (Lasagna-Reeves et al., 2011). In this case, higher tau aggregates such as the neurofibrillary tangles may represent nothing more than a kind of garbage bin where pathological tau is collected in an inactive form. In such a scenario, drugs that dissolve higher tau aggregates could even accelerate the degenerative process by increasing the amount of toxic soluble tau species.

Recent approaches aim to reduce the overall level of tau protein in patients with mild AD using an antisense approach (Mummery et al., 2023). This seems to be justified by the observation that the overall amount of tau increases during the development of AD and pathological consequences could then be attenuated by counteracting this change. However, as mentioned above, tau is a multifunctional protein that can affect many neuronal processes and signaling mechanisms in subtle ways. Therefore, a general reduction of tau protein below a critical level could have adverse effects, for example in terms of memory processes or mood disorders.

What conclusions can be drawn from this for the development of mechanism-based tau therapies (Figure 1)? (a) Successful drugs must reduce tau hyperphosphorylation and reduce tau aggregation without affecting the dynamic interaction of tau with microtubules and possible other tau activities. (b) Successful drugs targeting tau aggregates should prevent the formation of new aggregates but not dissolve existing ones. (c) Modulation of a multifunctional protein such as tau could benefit from a polypharmacological drug approach.

A first consequence is the need to establish neuronal models that allow quantifying both tau aggregation and the interaction of tau with axonal microtubules. Particularly helpful is the identification of tau mutations that increase tau aggregation, such as the single amino acid deletion mutant TauΔK280 reported in tauopathies (Momeni et al., 2009). Indeed, we were able to show that recombinant TauΔK280 exhibited more than 50% increased aggregation compared to wild-type tau in cell-free aggregation assays, and long-term expression of this construct led to progressive tau amyloid formation in primary neurons (Pinzi et al., 2024). Furthermore, developments in single-molecule tracking and quantitative live-cell imaging allow the determination of the kinetics of the tau-microtubule interaction in axons of living neurons (Janning et al., 2014). A combination of these approaches revealed that the aggregation-prone TauΔK280 had reduced interaction with microtubules compared to wild-type tau, consistent with the interpretation that aggregate formation reduces the amount of tau protein available for binding to microtubules. Indeed, the progressive formation of tau aggregates in primary neurons also led to a progressive decrease in the interaction of tau with axonal microtubules (Pinzi et al., 2024). Therefore, expression of pathologically aggregation-prone tau combined with live-cell imaging to assess the extent of tau microtubule interaction may provide a cellular model to identify drugs that both reduce tau aggregation and restore physiological tau-microtubule interaction.

As a proof of concept, we used this neuronal model to screen a panel of small molecules predicted to bind to tau protein. Indeed, one of the compounds increased the binding of aggregation-prone tau to microtubules similar to wild-type tau levels in model neurons, suggesting that it reduced pathological tau aggregation in the cells. Accordingly, it also reduced tau amyloid formation in primary neurons after long-term expression (Pinzi et al., 2024).

Such a model allows identifying candidate factors that inhibit tau aggregation and restore physiological tau-microtubule interaction. However, as described above, a successful drug candidate should prevent tau aggregation without affecting existing neurofibrillary tangles to avoid increased formation of potentially toxic soluble tau aggregates. This activity can be assessed in vitro; our findings indicate that this small molecule candidate reduced tau aggregate formation in cell-free assembly assays but did not dissolve pre-existing tau aggregates or insoluble paired helical filaments isolated from AD patients. Accordingly, the compound also diminished the formation of tau amyloids in cells without causing a decrease in their overall levels.

In tauopathies, the formation of pathological tau aggregates is accompanied by increased phosphorylation at selected sites. A drug candidate that modulates both aspects, tau aggregation and tau phosphorylation, could therefore be superior to more mechanistically restricted candidates. Experimentally, this would require detecting the kinome activity of neurons by systematically determining the effect of potential drug candidates on gene expression and protein phosphorylation. This can be done by proteomic and phosphoproteomic analysis of the cells, which would allow, on the one hand, identifying changes in tau phosphorylation, and, on the other hand, by kinase enrichment analysis to link the identified phosphorylation sites to reduced activity of the kinases most likely responsible for the reduced protein phosphorylation. Indeed, a corresponding analysis of our drug candidate showed that it reduces phosphorylation of tau in the proline-rich region of tau, which is involved in regulating tau aggregation and tau-microtubule interaction, and induces reduced activity in tau kinases such as glycogen synthase kinase-3β, which has previously been implicated in mediating tauopathies (Pinzi et al., 2024).

Although the model is promising, there is still room for improvement. First, the cellular model is based on the exogenous expression of a human tau construct in rodent cells, meaning that the cells still express a background of endogenous rodent tau. The sequence and expressed isoforms of rodent tau differ from human tau, which may be relevant for the development of tauopathies, and rodent tau may modulate the effect of the exogenously expressed human tau construct. In addition, human neurons may respond differently than rodent cells, as they also have a very different timeline of their development. Therefore, the neuronal model could be improved by using human neurons differentiated from patient-derived induced pluripotent stem cells with pathogenic tau mutations, for example from FTDP-17 patients. However, even with such an improvement, another limitation arises: tauopathies are age-related diseases and increasing age is the main risk factor. Currently, induced pluripotent stem cell-based models do not reflect aged neurons and also the culture time cannot be extended indefinitely. Of course, it must also be taken into account that dissociated neuronal cultures will never reflect the complexity of interactions between neurons and non-neuronal cell types that occur in a real brain, and it is still controversial why tau pathology develops stereotypically in selected brain regions. In this respect, cell-to-cell tau spread could also play an important role (Figure 1). If this is the case, this aspect would also have to be taken into account by a successful therapy.