Illustration of a tau protein fragment. Atoms are represented as spheres with conventional color coding.

Using one of this year’s most popular technologies, a team of researchers with decades of experience has—for the first time—revealed the atomic structure of tau filaments, the protein linked to a variety of neurodegenerative diseases, including Alzheimer’s and CTE.

In work begun almost 30 years ago, the researchers, whose study was published last week in Nature, say understanding the structures of tau filaments is key to developing drugs to prevent their formation.

The groundbreaking work was facilitated by two critical components—tau filaments from the brain of a patient who died with Alzheimer’s disease, as well as the novel use of cyro-electron microscopy (cryo-EM).

Previous research had only imaged samples of tau that were assembled in the lab. But because amyloid structures can form in many different ways, researchers were unsure how closely the structures resembled the disease.

“Many amyloid aggregates are thought to be able to adopt multiple different conformations. In the prion field (where amyloids form infectious protein aggregates), these different conformations are called 'strains'. It could be that different strains of certain amyloids play different roles in different diseases. By purifying the tau filaments straight from a diseased human brain, we know that these structures are the ones relevant for Alzheimer's disease,” study author Sjors Scheres explained to Laboratory Equipment.

The extracted tau filament samples were then imaged using cryo-EM—an up-and-coming technique Laboratory Equipment identified as a 2017 lab trend in March. Cryo-EM is a form of transmission electron microscopy where the sample is studied at cryogenic temperatures in its native form, often at near-atomic resolution. Used for decades in pharmaceutical research, the technique is now seeing a second life in structural biology.

“Cryo-EM was essential for this project,” said Scheres, group leader of the Medical Research Council’s Laboratory of Molecular Biology. “Alternative structure determination techniques are X-ray crystallography and nuclear magnetic resonance. For crystallography, you need 3-D crystals, which are very hard, if not impossible to obtain for these types of samples. For NMR you need isotope-labeled samples, which precludes working with samples from human tissues. Both techniques require much larger amounts of protein than cryo-EM. Cryo-EM was therefore in a unique position to work with the small amounts of unlabeled protein one can purify out of a human brain.”

That being said, cryo-EM presented its own set of unique challenges. Amyloid reconstructions are notoriously difficult to image since they are helical structures that are smooth along the helical axis—making it extremely hard to superimpose noisy cryo-EM images at the atomic level.

Thus, the team had to rely on a software Scheres created called RELION (REgularised LIkelihood OtimisatioN) that implements a statistical approach to the cryo-EM structure determination problem. The approach infers parameters about signal and noise from cryo-EM data, thereby replacing decisions from expert users in alternative software solutions.

According to Scheres, it has proven to be highly efficient at separating images from many different 3-D conformational states, and in this research, RELION managed to complete amyloid cryo-EM reconstructions to resolutions sufficient for atomic model building.

“[The software] has contributed to the rapid uptake of the resolution revolution in cryo-EM by many labs new to electron microscopy,” Scheres said. “We have used collaborations with various groups to work on cryo-EM samples that were challenging to existing methods in order to drive methods development forward, while also learning exciting new things about biology.”

One such collaboration is this tau research, in which Scheres worked with co-author Michel Goedert, head of the Neurobiology Division at the Medical Research Council’s Laboratory of Molecular Biology.

Goedert has been working with tau for almost 30 years. In 1988, a team including Goedert first identified the tau protein as an integral component of the lesions found in Alzheimer’s and a range of other neurodegenerative diseases. In 1992, Goedert showed that all six brain tau isoforms are present in paired helical and straight filaments. Six years later, Goedert and team described mutations in MAPT, the tau gene, that cause neurodegeneration and dementia. And in 2009, Goedert demonstrated the prion-like behavior of aggregated tau in transgenic mouse brain.

“Until now, the high-resolution structures of tau or any other disease-causing filaments from human brain tissue have remained unknown,” Goedert said. “This new work will help to develop better compounds for diagnosing and treating Alzheimer's and other diseases that involve defective tau.”

Currently, most pharmaceutical companies use assays that are based on different parts of tau. However, this research could change that. Seeing the atomic structure of tau allows scientists to learn which parts of the protein form the seed for aggregation—thus learning how to develop drugs to stop it.

Abnormal amyloid structures also suggest how the same protein may form different filaments in other neurodegenerative diseases, potentially opening a pathway to mechanism-based therapies.

“There are many different neurodegenerative diseases that have tau aggregates in the brain,” Scheres said. “These filaments look different in low-resolution, negative-stain microscopy. We would like to understand how they are different at the atomic level. Apart from helping us understand the differences between these diseases, it would also provide fundamental insights into the different structures this protein can adopt.”