Though tau goes rogue in many neurodegenerative conditions, exactly how it contributes to disease still puzzles scientists. To gain a better idea, researchers led by Li Gan, now at Weill Cornell Medicine in New York City, used a relatively new technology to tag proteins that bumped up against wild-type or mutant tau in human neurons. In the January 14 Cell, they presented the first comprehensive tau interactome from human iPSC-derived neurons. The data reinforced previous findings in the field, but added some surprises. For example, the scientists found numerous interactions between mitochondrial proteins and wild-type, but not mutant, tau. Moreover, in the presence of mutant tau, mitochondria became sluggish. The finding hints at a role for tau in bolstering cellular energy production.

  • Comprehensive survey of tau’s “interactome” hints at multifarious roles.
  • In response to neuronal activity, tau binds synaptic proteins responsible for exocytosis.
  • Mutant tau poorly binds mitochondrial proteins, correlating with energy deficits.

The interactome data also bulked up tau’s synaptic portfolio. Not only did tau snuggle up to many proteins at the surface of synaptic vesicles, but when neurons were stimulated, tau dallied with additional proteins that regulated exocytosis. These data may help explain the activity-dependent release of tau that is thought to spread it among neurons.

Amy Pooler at Sangamo Therapeutics in California’s San Francisco Bay area called the findings exciting. “This is a wonderfully comprehensive analysis of interactions between neuronal proteins and tau,” she wrote. Giuseppina Amadoro at the Institute of Translational Pharmacology-National Resource Council, Rome, said it would be a valuable resource. “This interesting report not only provides a mechanistic explanation of the direct action of tau on mitochondrial bioenergetics and presynaptic function, both in health and in disease, but also reveals exhaustive mapping of potential microtubule-independent interactors of the protein that can be therapeutically targeted to slow down the progression of human tauopathies,” she wrote to Alzforum (full comments below).

Synaptic Menagerie. Tau contacts all the synaptic vesicle proteins shown, except synaptobrevin, reinforcing the idea that it helps regulate exocytosis. [Courtesy of Tracy et al., Cell.]

Most previous attempts to delineate the tau interactome have focused on mice (Liu et al., 2016; Wang et al., 2017; Maziuk et al., 2018). To extend these findings to people, Gan and colleagues used a human iPSC line modified to generate glutamatergic neurons (Wang et al., 2017). The APEX enzyme modifies tyrosine residues on any protein that comes within 10-20 nanometers of it, allowing these residues to be subsequently biotinylated. Biotinylated proteins can then be identified by mass spectrometry (Rhee et al., 2013). 

Joint first authors Tara Tracy, Jesus Madero-Pérez, and Danielle Swaney conjugated APEX to the 2N4R isoform of human wild-type tau, and expressed the hybrid protein in the iPSC-derived neurons. Mass spectrometry identified 246 proteins that had contacted tau. The list included known interactors, such as microtubule and cytoskeletal proteins, ribonucleoproteins, heat shock proteins, and α-synuclein. Nuclear and RNA-binding proteins also turned up, as did lysosomal and proteasome proteins involved in waste removal. There were also a large number of synaptic proteins responsible for vesicle docking and fusion, including SNARE complex proteins, dynamin, and syntaxin (see image above). Because the synaptic vesicle proteins were biotinylated only on their cytosolic side, never the luminal, the researchers concluded that tau bound to the surface of synaptic vesicles.

“I was excited to find the study corroborated several earlier findings, including, from our own work, the prominent association of tau with the ribonucleo-proteome. I look forward to digging deeper into the data files to see what else they may tell us about candidate interactors,” Gerold Schmitt-Ulms at the University of Toronto wrote to Alzforum. However, he cautioned that proteins that come close to each other do not necessarily interact, highlighting the importance of validating these findings by other methods (comment below).

One unique feature of APEX is that it detects transient and even unstable interactions, providing a snapshot of cellular activity. Gan exploited this feature to examine the effects of neuronal activity on the tau interactome. When the scientists stimulated the cultures with potassium chloride, to depolarize them and trigger synaptic vesicle release, new proteins interacted with tau. Many were synaptic proteins that regulated exocytosis, such as synaptotagmin-1, Mint1, SV2C, and synapsin-1. Gan was intrigued by these data. She thinks the neuronal hyperexcitability seen in early AD could help drive the spread of tau between cells.

Mutant Tau Ignores Mitochondria. Some mitochondrial proteins (green) interact with wild-type but not V337M tau; others (purple) interact with wild-type but not P301L tau, and some (blue) interact with wild-type but neither mutant variety. The data hint at a role for wild-type tau in energy production. [Courtesy of Tracy et al., Cell.]

Although APEX can capture fleeting interactions, it has limitations. For one, it is less sensitive than other methods, and can miss protein partners that bind infrequently. To complement the APEX data, the authors expressed tagged wild-type or mutant tau in neurons, then lysed the cells and isolated the tau isoforms using beads that recognized the tag. Mass spectrometry identified any bound proteins. This technique largely replicated the APEX findings. However, when the authors compared interactomes, they were surprised to find that P301L and V337M mutant tau had far fewer contacts with mitochondrial proteins than did wild-type (see image above). Wild-type tau bound to a large variety of mitochondrial proteins, including cytochrome C, ATP transporters, amino acid transporters, fatty acid oxidizers, the protein importer TOMM40, chaperone TIMM13, and ATPase inhibitor ATP5IF1.

There are previous reports in the literature of mutant tau entering mitochondria, Gan noted, but this was usually assumed to be aberrant (Dec 2017 news). She believes the new data suggest tau has a physiological function at mitochondria.

However, Ben Wolozin at Boston University questioned if these interactions occur inside or even on mitochondria. Instead, he wondered if they reflect tau helping to traffic these proteins from the nucleus or endoplasmic reticulum to their final destination. “Tau is not a mitochondrial protein,” he noted (comment below).

Whatever explains the interactions, the authors found that mitochondrial function was impaired in neurons expressing mutant tau. More protons leaked from mitochondria, requiring the cells to use more oxygen to generate the same amount of energy as did wild-type cells. Because this occurred in the absence of any tau filaments or inclusions, the finding hints that weakened bioenergetics could be an early event in tauopathy, Gan said.

Does this happen in human brain? Analyzing mRNA and protein data from AMP-AD, the authors found that in the brains of people who had Alzheimer’s disease at the time of death, many mitochondrial tau partners were in short supply compared to levels in healthy age-matched controls. Intriguingly, the lower the protein level, the more amyloid and tau pathology that brain contained, as measured by CERAD and Braak scores, respectively. The authors found a similar pattern in postmortem brains from the University of Pennsylvania Brain Bank. Notably, levels of these mitochondrial interlocutors were low in tauopathies besides AD, as well, including frontotemporal dementia and progressive supranuclear palsy.

“Taken together, these findings suggest that defects in mitochondrial bioenergetics may represent a converging disease mechanism across primary tauopathies and AD,” noted Wilfried Rossoll at the Mayo Clinic in Jacksonville, Florida. Nicholas Seyfried at Emory University, Atlanta, called the human data a strength of the paper (full comments below).

Commenters believe the methods described here could be extended to gather more data. Pooler suggested looking at additional neuronal subpopulations besides glutamatergic, while Wolozin noted the importance of studying neurons containing oligomeric and fibrillized tau. Overall, they agreed the techniques will be broadly useful. “One can envision how this discovery platform can be used to map and compare interactomes under a variety of different conditions in neuronal health and disease,” Rossoll wrote.—Madolyn Bowman Rogers


  1. The authors are to be commended for a beautifully presented study that should serve as a useful resource in the years to come. The study showcases an elegant application of the APEX method to map binding partners of the 2N4R tau isoform in an in vitro human neuron paradigm, in the presence or absence of KCl stimulation. It nicely demonstrates both the power and limitations of this method, and how it can be paired with follow-on orthogonal techniques to reveal meaningful insights.

    Naturally, with this approach, its strength in mapping the spatial relationship of proteins also is its predominant weakness, because proteins that reside in proximity do not have to work together, or even influence each other. It therefore will still require a major effort to understand which of the proteins that reside in proximity to tau can truly be considered interactors before we can make biological sense of these data.

    The authors have begun this process by complementing their APEX data with affinity capture mass spectrometry analyses for a subset of their analyses, exploiting the inclusion of terminal FLAG-tags on their APEX-tau fusion constructs, and by showing that a subset of shortlisted candidate tau interactors have altered levels in human postmortem brains from individuals who succumbed to tauopathies.

    I was excited to find that the study corroborated several earlier findings, including, from our own lab, the prominent association of tau with the ribonucleo-proteome. I look forward to digging deeper into the data files to see what else they may tell us about tau candidate interactors.

    I would be remiss if I was not to leave a suggestion here for the authors that, in my mind, would greatly facilitate the task of integrating and comparing results from this and prior studies, i.e., to add to the Supplemental Data files the relative enrichment levels of the proteins that were observed in proximity to, or were co-purifying with, tau (in addition to the binary yes/no labels currently shown). An obvious next step would be to apply these cutting-edge methods to an in vivo paradigm, preferably based on a model that recapitulates more closely the tau isoform balance observed in human brains.

  2. This is an interesting and comprehensive analysis of the tau protein interaction network. The study identifies many cytoskeletal protein interactors, which have been previously noted to interact with tau.

    The group also identified a large network of proteins associated with vesicles and the presynaptic SNARE complex, which are part of the tau protein interaction network (PIN). This observation is interesting because tau is well-known to be secreted, which enables propagation of tau in disease. The identification of this tau synaptic/vesicular PIN potentially provides some targets for therapeutic intervention to reduce such propagation.

    This group includes multiple proteins involved in active zone docking, including DNM1, RABGAP3, Syntaxin 3, MINT1, etc. It’s interesting to note that N- and C-terminal APEX tags had about 55 percent overlap, which fits with prior results showing effects of chimeric tags producing varied results depending on whether they are on the N- or C-terminus.

    The authors also examined a large component of mitochondrial proteins as part of the tau PIN; this group includes multiple members of the inner mitochondrial membrane. I find the presence of proteins of the inner mitochondrial membrane surprising, because tau is not a mitochondrial protein; this suggests to me that tau is involved in trafficking these proteins from the nucleus to the mitochondria. The association of TOMM40 in this group is interesting because it is produced from a gene in the tau locus. Thus, the co-localization of these two genes might be functional. The authors find that many of these interactions decrease with pathology in human brain, which is consistent with a loss of the basal functioning of tau as it fibrillizes.

    A weakness of the study is that it omits any discussion of the actions of tau under conditions of stress or fibrillization. The iPSCs used in this study were not subject to any type of stress, and only stimulated with KCl. It is known that stress causes tau to become phosphorylated by proline-directed kinases, which causes dissociation of tau from microtubules and its oligomerization. The actions of pathologically phosphorylated or oligomeric tau are thought to be essential to its pathophysiology. This manuscript distinctly omits any discussion of these interactions.

    Interactions that involve RNA and DNA metabolism as well as the ribosome are presented in passing (e.g., Fig. 2 and 4) but are not discussed, even though these appear increased in the mutant tau (V337M and P301L) paradigm and are known to dramatically change with disease and contribute strongly to the pathophysiology of tauopathies (Jiang et al., 2021; Koren et al., 2019). 

    Despite the lack of evaluation of tau biology during the stress response, this study provides insight into basal functioning of tau protein and uses a powerful proteomic tool that strongly validates much of the work done by other groups previously. It is a solid manuscript that certainly represents an important addition to the literature on tau PINs.


    . Interaction of tau with HNRNPA2B1 and N6-methyladenosine RNA mediates the progression of tauopathy. Mol Cell. 2021 Oct 21;81(20):4209-4227.e12. Epub 2021 Aug 27 PubMed.

    . Tau drives translational selectivity by interacting with ribosomal proteins. Acta Neuropathol. 2019 Apr;137(4):571-583. Epub 2019 Feb 13 PubMed.

  3. This very interesting work led by Dr. Li Gan and a team of collaborators creatively used proximity labeling and pull-down experiments in human iPSC-derived neurons to map tau interactomes with high spatiotemporal resolution under different conditions. The Gan lab has done pioneering work on the role of tau acetylation in disease, and the current work identifies additional potential disease mechanisms and targets for future therapy development.

    The first part of this study used peroxidase-based proximity labeling and interactome mapping via mass spectrometry to identify neuronal activity-dependent changes in the tau interactome. This method allows for a high degree of spatiotemporal resolution. Tagging tau on either the N- or C-terminus identified a surprisingly large percentage of unique interactors. A wealth of additional spatial information comes from the enrichment for biotin-modified peptides, which made it possible to not only identify proximal proteins but also map proximal biotinylated tyrosine residues at high resolution. One can envision how this discovery platform can be used to map and compare interactomes under a variety of different conditions in neuronal health and disease.

    In the present study, the precise mapping of biotinylation sites led to the discovery that neuronal activity enhanced interactions with specific synaptic vesicle-associated proteins, providing a potential mechanism for activity-dependent tau release from neurons via the presynaptic vesicle fusion machinery. It will be interesting to see how this mechanism vs. exosome-mediated tau release may contribute to the observed spread of tau pathology across brain regions.

    A more conventional co-immunoprecipitation approach was used in the second part of this study to compare the tau-associated proteome between wild-type (WT) and FTD-associated mutant tau. It would have been interesting to see how this alternative approach compares to the proximity labeling and interactome mapping method used before.

    The main finding here was a weakened interaction of tau carrying FTD-causing mutations with mitochondrial proteins, potentially impairing mitochondrial bioenergetics as shown in isogenic iPSC-derived neurons carrying the V337M mutation. While this finding suggests a potential disease mechanism related to compromised energy metabolism, it remains to be seen how this loss-of-function phenotype of mutant tau in vitro relates to an autosomal-dominant mutation causing early onset dementia in human patients. 

    Finally, the analysis of existing proteomic datasets revealed a significant reduction in the levels of WT tau interactors modified by FTD-causing mutation in AD, FTD, and PSP brain tissue compared with those in control cases. This highlights a major strength of this study in combining different approaches to probe tau interactomes and to explore potential connections to the disease process.

    While the mechanism behind the reduced abundance of WT tau interactors is unclear at this point, taken together these findings suggest that defects in mitochondrial bioenergetics may represent a converging disease mechanism across primary tauopathies and AD.

  4. In this elegant study, the authors generated an atlas of tau-interacting proteins using proximity labeling and immunoprecipitation approaches to capture both wild-type (WT) and mutant tau interacting partners.

    One novel aspect of this paper was that the authors engineered human iPSC-derived neurons to express APEX2 either N- or C-terminal to full-length tau for proximity labeling of tau-interacting proteins with biotin and identification by mass spectrometry. Remarkably, nearly half of all tau interactions were uniquely detected in either N-APEX or C-APEX tau neurons, revealing unique domain-specific functions of tau and corresponding proteolytic fragments that are produced in disease. Mapping sites of biotinylation also allowed the authors to resolve contact regions between tau and the interacting proteins, highlighting both unique cytosolic and nuclear partners. The authors further leveraged these engineered human neurons to capture activity-dependent changes in tau-interacting partners following KCl treatment, which revealed enrichment of proteins that regulate synaptic vesicle exocytosis, and may provide new mechanisms and drug targets that underlie tau secretion.

    Additional studies highlighted differences between tau interacting partners of WT and mutant tau (TauP301L or TauV337M). Here, they observed a loss of interaction between mutated tau and mitochondrial proteins compared to WT tau, suggesting that energy defects in disease are linked to tau dysfunction.

    Finally, a strength of the paper is that the authors used proteomic data from human brain tissue to show that in AD, and other tauopathies, levels of tau interactors correlate with disease severity. Overall, this paper not only confirms many of the known components of the tau interactome, but also provides a strong resource of novel tau-protein interactions that can be followed up for future mechanistic studies.

  5. This is a wonderfully comprehensive analysis of interactions between neuronal proteins and tau. We’ve known for a long time that tau is distributed among many different compartments in neurons, but there’s still a lot that’s unknown about how this process is regulated, particularly in the context of disease-causing tau mutations.

    I find the new interactions with mitochondrial proteins particularly exciting. Mitochondrial dysfunction, including reduced bioenergetics, is well described in the early stage of tauopathies, and abnormal tau may disrupt coupling between mitochondria and the ER. I look forward to further studies on how these novel interactions may play a role in these processes.  

    One limitation of this study is the focus on iPSC-derived neurons. In future it will be important to understand whether these interactions vary between the many different neuron subpopulations, which may shed light on regional vulnerabilities to tau toxicity in the brain.

  6. Aβ has been reported to enter mitochondria, and it is certainly worth considering if tau does the same, especially in the setting of dysregulated mitophagy.

    Given the gradual failure of cellular energetics across numerous cell types in Alzheimer's, it is important to find out more about this phenomenon.

    I haven't read the paper yet, but I wonder if the majority of mitochondrial proteins in this experiment incidentally contacted tau in the cytoplasm before they were imported to mitochondria, as opposed to contacting tau within the inner membrane? For example, TOMM40 on the outer membrane is a nuclear gene on chromosome 19 which has been previously associated with Alzheimer's disease, and would be readily accessible to tau.

    My lab's unpublished RNA-Seq data from a number of non-neuronal cell types also implicates exocytosis of Aβ and tau as potentially causative mechanisms in AD, as well as implicating a number of mitochondrial pathways.

    Whatever the explanation for these mitochondrial contacts with tau, I think the authors have made an important observation, and this technique is quite elegant.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Is There No End to Tau’s Toxic Tricks?

Paper Citations

  1. . Co-immunoprecipitation with Tau Isoform-specific Antibodies Reveals Distinct Protein Interactions and Highlights a Putative Role for 2N Tau in Disease. J Biol Chem. 2016 Apr 8;291(15):8173-88. Epub 2016 Feb 9 PubMed.
  2. . Tau interactome mapping based identification of Otub1 as Tau deubiquitinase involved in accumulation of pathological Tau forms in vitro and in vivo. Acta Neuropathol. 2017 May;133(5):731-749. Epub 2017 Jan 12 PubMed.
  3. . RNA binding proteins co-localize with small tau inclusions in tauopathy. Acta Neuropathol Commun. 2018 Aug 1;6(1):71. PubMed.
  4. . Scalable Production of iPSC-Derived Human Neurons to Identify Tau-Lowering Compounds by High-Content Screening. Stem Cell Reports. 2017 Oct 10;9(4):1221-1233. Epub 2017 Sep 28 PubMed.
  5. . Proteomic mapping of mitochondria in living cells via spatially restricted enzymatic tagging. Science. 2013 Mar 15;339(6125):1328-1331. Epub 2013 Jan 31 PubMed.

External Citations

  1. AMP-AD

Further Reading

Primary Papers

  1. . Tau interactome maps synaptic and mitochondrial processes associated with neurodegeneration. Cell. 2022 Feb 17;185(4):712-728.e14. Epub 2022 Jan 20 PubMed.