When it rains, sometimes it pours. In the second paper describing tau’s ability to undergo liquid-liquid phase separation in the past month, researchers led by Markus Zweckstetter of the German Center for Neurodegenerative Diseases in Göttingen report that phosphorylation of tau dramatically enhances droplet formation, and that tau’s repeat domains play a key role in the process. In their August 17 paper in Nature Communications, the researchers made the case that droplets are an essential precursor in the formation of toxic tau tangles, though experiments were all conducted in cell-free conditions. 

Tau is one of many proteins involved in neurodegenerative disease that have been spotted mingling in liquid droplets (Oct 2015 webinarOct 2016 news). Inside the cell, the process of liquid-liquid phase separation (LLPS) leads to membraneless organelles, including stress granules and the nucleolus. Interactions between proteins—especially those donning low-complexity domains—and nucleic acids trigger the process, and researchers have proposed the close quarters in the droplets could breed toxic aggregates, or derail essential cellular functions (May 2016 news). A recent study led by Kenneth Kosik and Songi Han of the University of California, Santa Barbara, reported that tau coalesced into liquid droplets in a dish, and that interactions between positively charged tau and negatively charged RNA made the magic happen (Jul 2017 news). Researchers led by Anthony Hyman of Germany’s Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, along with Bradley Hyman and Susanne Wegmann of Massachusetts General Hospital in Charlestown, have also reported that tau forms droplets in vitro and in neurons (May 2017 news).

Turbid Tau

Fluorescently labeled tau (green) forms droplets at body temperature (bottom), but not at 5C (top). [Image courtesy of Ambadipudi et al., Nature Communications, 2017]

In the current study, first author Susmitha Ambadipudi and colleagues investigated if, and under what conditions, tau undergoes LLPS, and how that relates to its propensity to aggregate into fibrils. They started by analyzing various regions of the tau protein with catGranule, a program that predicts the propensity of a given protein region to undergo phase separation. While much of tau’s N-terminus scored low, its repeat domains scored high. The researchers therefore conducted most of their experiments using the K18 fragment of tau, which contains only the four-repeat domains.

They reported that under reducing conditions similar to those inside a cell, K18 formed a turbid solution. Bright field and confocal microscopy revealed that K18 formed droplets under Goldilocks conditions—not too cold (5°C), not too hot (above 65°C), but just right (37°C). Under these conditions, K18 did not appear to form outright fibrils, although CD and NMR spectroscopy suggested the protein started exhibiting signs of β-sheet structure and that droplet-resident tau proteins formed tight molecular interactions, akin to a mesh.

Could this mesh facilitate fibril formation under certain conditions? To get closer to answering this question, the researchers toggled multiple parameters, including temperature and pH, and added the polyanion heparin into the mix. They found that heparin triggered fibril formation most efficiently and under the very same conditions that facilitate LLPS, suggesting the two processes are linked. Polyanions have long been used to promote formation of tau fibrils (Goedert et al., 1996).

In the cell, tau occurs in six isoforms due to alterative splicing, and can be further processed by proteolytic fragmentation and a variety of post-translation modifications. How might these permutations affect LLPS? They found that droplet formation correlated with the number of repeats, and did not occur at all in an N-terminal fragment that lacked repeats. They also found that phosphorylation of repeat domains by the MARK2 kinase promoted LLPS. Notably, phosphorylated tau underwent LLPS at just 2 μM, a concentration similar to that inside neurons. Interestingly, another recent study found that phosphorylation had the opposite effect on phase separation of the FUS protein, which plays a role in amyotrophic lateral sclerosis (Aug 2017 news).


In this proposed model of tangle formation, phosphorylation of tau promotes the formation of liquid droplets, which crowds tau, recruits polyanions, and triggers aggregation. [Courtesy of Ambadipudi et al., Nature Communications, 2017]


The in vitro findings mesh with Kosik and Han’s recent study, which found that tau formed droplets in the presence of RNA. Though Zweckstetter used heparin as a polyanion instead of RNA, both studies point to the importance of electrostatic interactions between tau and negatively charged molecules in promoting LLPS and aggregation. As a post-translational modification bearing a negative charge, phosphorylation may play a similar role, Zweckstetter pointed out. Zweckstetter added that phase separation of full-length, unphosphorylated tau did not occur in their hands. “Mostly likely phosphorylation or interactions with RNA would be needed to facilitate that,” he told Alzforum.

The latter finding contradicts the findings of Wegmann and colleagues, who did detect phase separation of full-length tau, and even of N-terminal tau completely devoid of repeat domains, she told Alzforum (May 2017 conference news). Wegmann was fascinated by this difference, adding that it underscored the complex process of phase separation, pointing to varying contributions of different regions of the tau protein in the process. For her part, Wegmann is working on pinning down the presence of liquid droplets of tau in cultured neurons, and in AD brain tissue.

In a joint commentary to Alzforum, Kosik and Han agreed that understanding the physiological significance of these droplets was crucial: “The in vitro studies lay the groundwork for the next big step: how might these phenomena operate in vivo, where life is not only many-fold more complicated but rife with emergent properties.”—Jessica Shugart


  1. Tau Packs Heat

    No doubt tau protein is the culprit in its namesake diseases—the tauopathies, which include Alzheimer’s disease. Among the longstanding head-scratching problems in tau pathobiology is how tau, one of the most soluble proteins known, transforms into a neurofibrillary tangle, one of the most insoluble protein complexes known. Like a master magician who deceives us into questioning the laws of physics as objects appear and disappear over impossible distances, so tau undergoes this elusive transition in what, until now, has been obscured from view. This new work from Ambadipudi et al. and our previous study (see Zhang et al., 2017) build strong arguments that the tau transition state in vitro involves liquid-liquid phase separation (LLPS). The discovery that tau can phase separate into droplets opens the door to the rich world of coacervation. Complex coacervation requires an associative interaction between oppositely charged macromolecules or macromolecular domains to create a distinct phase that is immiscible in, while in equilibrium with, the surrounding liquid (De Jong and Bonner, Protoplasma, 24:198, 1935). Complex coacervates reside in a finely tuned state stabilized by sufficiently strong (intra- or inter-) polymer-polymer interactions, often of electrostatic nature, but not so strong as to cause polymer dehydration and precipitation. Although the concentration of macromolecules in complex coacervates can reach levels not sustainable in aqueous solutions, protein structure, dynamics, and function appear to be minimally altered as illustrated in one example by the retention of enzymatic activity (Xia et al., Biopolymers, 41:359, 1997). Many biopolymers are also capable of LLPS by self-coacervation, whose process, however, is facilitated when the polymer contains oppositely charged blocks of domains, and can be considered a mechanism closely related to complex coacervation (Das and Pappu 2013).

    The key tuning parameters conducive to coacervation of the K18 tau construct (the repeat region of 4R-tau) according to Ambadipudi et al. and of the N-terminus truncated 4R-tau according to Zhang et al. include temperature, ionic strength, pH, excluded volume and protein concentration, as well as an important kinetic parameter, the time course over which coacervation conditions occur. Salt tunes the effective polyelectrolyte charges and pH affects the degree of ionization of the functional groups. When taken together, complex coacervation of tau variants is found to be an entropy-driven process by displaying a lower critical separation temperature (LCST). In the case of tau, when the conditions are finely tuned, these critical parameters approach physiological conditions in terms of temperature, ionic strength, pH, and crowding pressure. Ambadipudi et al. excluded the influence of intramolecular and intermolecular cross-linking through two native cysteine residues by mimicking the reducing environment inside cells. Monitoring the formation of droplet-like structures and their characteristic fusion due to their low interfacial tension is quite straightforward by differential interference contrast microscopy. The presence of K18 in the droplets is determined by labeling this tau variant with the fluorescent dye Alexa-488.

    The discovery that tau can liquid-liquid phase separate immediately raises the question whether this phase transition is on pathway to tau fibrils. In both Ambadipudi et al. and in our study, a small increase in thioflavin T (ThT) fluorescence suggests the formation of β-sheet structures, but only after long incubation times. Using two different methods—NMR in the case of Ambadipudi et al. and EPR (electron paramagnetic resonance) in ours—the studies show that tau appears to retain its solution structure inside the droplet consistent with coacervation. The NMR spectra leaves some ambiguity as to whether tau is dynamic within the droplets or fast exchanging in and out of the droplets. With its faster timescale on the order of nanoseconds for dynamics versus micro- to milliseconds for exchange, EPR suggests that tau is actually in a dynamic state within droplets. Importantly with regard to the potential for fibril formation, the observed molecular contacts of K18 are direct evidence for crowding of the repeat domain of tau and its aggregation-prone hexapeptides. Ambadipudi et al. build on this observation by adding heparin, which is known to promote tau aggregation, to the droplets. In an elegant set of experiments they show that optimal temperature, pH, and ionic conditions for heparin-induced fibril formation resembled those of droplet formation, and that even though heparin is a very strong aggregator, its potency falls off sharply when temperature and ionic conditions are no longer conducive to droplet formation. Finally, they show that phosphorylation within the repeat domain by MARK2 enhances LLPS. This finding is important because it is known that phosphorylation of tau precedes the formation of insoluble tau deposits.

    The in vitro studies lay the groundwork for the next big step: how might these phenomena operate in vivo, where life is not only many-fold more complicated but rife with emergent properties. Our previous work on tau droplets suggested that RNA could serve as a counter ion and thereby loosely tied tau LLPS to intracellular membraneless organelles such as RNA granules (Knowles et al., 1996), as well as to a set of RNA-binding proteins involved in neurodegeneration. However, Ambadipudi et al. show that tau is capable of self-coacervation with its phosphorylation domain serving as the counter ion. In moving this work to an in vivo setting, those sites where tau can become tightly packed and demix from the surrounding cytoplasm to a membraneless oragnelles is the next hurdle.

    Songi Han was the co-author of this comment.


    . RNA stores tau reversibly in complex coacervates. PLoS Biol. 2017 Jul;15(7):e2002183. Epub 2017 Jul 6 PubMed.

    . Conformations of intrinsically disordered proteins are influenced by linear sequence distributions of oppositely charged residues. Proc Natl Acad Sci U S A. 2013 Aug 13;110(33):13392-7. Epub 2013 Jul 30 PubMed.

    . Translocation of RNA granules in living neurons. J Neurosci. 1996 Dec 15;16(24):7812-20. PubMed.

Make a Comment

To make a comment you must login or register.


Webinar Citations

  1. Fluid Business: Could “Liquid” Protein Herald Neurodegeneration?

News Citations

  1. ALS Research ‘Gels’ as Studies Tie Disparate Genetic Factors Together
  2. Stress Granule Protein Entwines and Misfolds Tau
  3. Tau Hooks Up with RNA to Form Droplets
  4. Protein Liquid-Liquid Phase Transitions: The Science Is About to Gel
  5. Phosphorylation of FUS Does Away with Droplets

Paper Citations

  1. . Assembly of microtubule-associated protein tau into Alzheimer-like filaments induced by sulphated glycosaminoglycans. Nature. 1996 Oct 10;383(6600):550-3. PubMed.

Further Reading

No Available Further Reading

Primary Papers

  1. . Liquid-liquid phase separation of the microtubule-binding repeats of the Alzheimer-related protein Tau. Nat Commun. 2017 Aug 17;8(1):275. PubMed.