New Partnership Will Probe the Connection Between Tau Protein and Alzheimer’s

While we often think of each gene giving rise to a single “protein,” the reality is that each protein exists in potentially thousands of variants known as proteoforms.

These variants are the active, functional forms of each protein and have a wide range of different modifications that impact how they fold, where they are located in the cell and how they perform enzymatic processes. Some of the modifications affect their primary sequence; for instance, through alternative splicing, they can have different sections cut out and pasted in. Likewise, as part of processing, after translation, they may have regions of sequence at the beginning or end cleaved. Additionally, proteins often have their individual amino acid building blocks modified, such as through phosphorylation or methylation.

The combinations of all these post-translational modifications give rise to the many distinct proteoforms derived from a given protein. Understanding that there are combinations of modifications and not just single modifications is key. These combinations vastly expand the number of possible proteoforms derived from a protein. Within a cell, there are likely many different proteoforms of each protein present. Furthermore, the proteoform composition of a cell is constantly changing every moment of every day.

First, we don’t know which of the near-infinite number of possible proteoforms actually exist. Building technologies to measure a vast, and unknown number of things is extremely challenging.

Furthermore, some proteoforms may be extremely rare – some may exist in only a handful of copies per cell. Consequently, extremely sensitive measurement tools are required – tools that can look at individual molecules. Also, for various practical, biophysical reasons, many protein measurement approaches digest proteins into fragments (peptides) and then measure those fragments. Unfortunately, digesting proteins before analyzing them destroys one’s ability to determine if multiple modifications are present on the same molecule.

Proteoforms are also hard to separate from each other. For example, two proteoforms of the same protein that are both triply phosphorylated – but perhaps phosphorylated at different sites – will all have nearly identical biophysical characteristics. Analytical techniques to separate these proteoforms from each other simply don’t exist.

Aggregating signals from many proteoforms to report “protein measurements” is like seeing a clump of green in a photo and calling it a “forest” rather than describing each unique tree.

On the Nautilus Platform, we use a method called Iterative Mapping to repeatedly analyze millions to billions of protein molecules at the single-molecule level. This method uses probes that bind specific modifications to identify which molecules have these modifications. By looking at millions to billions of molecules at once, we achieve quantitative views of the proteoform landscape. By changing how many and what kinds of probes we use, we can scale the resolution we get on each proteoform to suit researchers’ needs.

Neurofibrillary tangles (NFTs) were one of two protein pathologies originally described by Alois Alzheimer in 1907. Nearly 80 years later, we learned they were made up primarily of abnormally phosphorylated Tau protein (pTau).

A decade after that, Heiko and Eva Braak noted these NFTs progressively accumulate in specific brain regions along a staging criteria they created, which bears their name. While someone must have amyloid plaques to have AD by the National Institute on Aging-Alzheimer’s Association definition, Braak staging is more strongly associated with cognitive decline than measures of amyloid plaque burden.

How pTau aggregates has been intensely studied with bulk proteomics and seems to occur when portions of Tau that are positively charged become neutralized by the negative charge on phosphate. Once aggregated, pTau is thought to be toxic to neurons and lead to their eventual death, though how exactly is still an active area of research.

There’s a great perspective piece on this by Iqbal and Grundke-Iqbal, published in the Journal of Alzheimer’s Disease.

In that paper, they talked about their earliest research from 1986 that suggested abnormal phosphorylation of tau associated with AD. In subsequent papers, they noted that there was about eight times more staining for phospho-tau in brains from AD patients relative to controls. However, those studies were at a very macro level. They suggested “more” phosphorylation, but it’s always been a little ambiguous what “more” means. For example, it could mean that there are more tau molecules phosphorylated at one single site, or it could mean there are similar numbers of pTau molecules, but each is phosphorylated at many more sites, or it could be some mixture of both.

In vitro studies with synthetic peptides, recombinant proteins and proteins extracted from patient samples have shown that changes in phosphorylation can drive fibrilization and aggregation of tau very differently.

In Nautilus’ latest preprint, we – for the first time – show extensively exactly which combinations of tau phosphorylation events increase in AD. Perhaps our most intriguing but preliminary finding was that a sample from an Alzheimer’s patient with the most severe pathology in our small cohort contained proteoforms with the most phosphorylations per molecule of all the patient samples analyzed. In fact, one tau proteoform with four phosphorylations appeared to be specific to this patient.

This finding is highly preliminary, given that we only analyzed five Alzheimer’s patient samples, but no other technologies are capable of discerning differences in proteoforms like this.

In our Seattle Alzheimer’s Disease Brain Cell Atlas project (SEA-AD), we selected 84 aged donors that span the range of pathological burden (from no to very high levels) and cognitive impairment (from normal to having severe dementia). In these donors, we isolated nearly 7 million cells from 10 different brain regions that are impacted by pTau at different points in Braak staging. By looking at what genes were expressed in each cell, we could tell their cell type and the state they were in. We related that to each donor’s pathological burden, which we measured by staining for a specific pTau proteoform used in Braak staging (as well as other protein pathologies).

We found several kinds of neurons that appear to be selectively lost – including some when only a small amount of the pTau proteoform is present.

Our pilot with Nautilus is aimed at exploring if we can better measure other kinds of pTau proteoforms that are present in that early phase of the disease and see if any are more strongly associated with that early neuronal loss. If so, we may better understand that early loss and design therapies to stop it. We could also identify proteoforms that could serve as important biomarkers of early AD.

While measuring select tau modifications independently can improve Alzheimer’s diagnosis and deepen our understanding of the disease, as the Allen Institute’s research demonstrates, we saw a deeper opportunity in studying tau proteoforms with combinations of modifications.

These proteoforms are largely untapped when it comes to Alzheimer’s diagnostics and treatments, but identifying patients with different proteoform profiles may provide far more nuanced ways to stratify patients, detect Alzheimer’s at its earliest stages and guide treatment.

Further down the line, tau proteoforms uniquely associated with disease progression may become next generation drug targets – either eliminating them or encouraging their transformation into proteoforms associated with better outcomes may slow or even prevent Alzheimer’s progression. When researchers develop drugs targeting these proteoforms, our platform can also be used to identify patients with the targeted proteoforms. This will make the trials as efficient and informative as they can be.

Our project includes donors that span the pathological spectrum with early, middle and late affected brain regions. The absolute home run would be to chart the progression of pTau proteoforms that emerge as pathology spreads across regions within a donor and across donors with increasing pathological burdens, not dissimilar to the transformative staging criteria from the Braaks that is still used today.

This would give us a highly granular window into what pTau looks like before and as it aggregates, whether and how stereotyped it is, and if any proteoforms appear strongly associated with early cellular loss.

The Allen Institute is also committed to Open Science and to creating complete and permanent datasets that are as impactful, easy to use and high quality as possible. As part of that commitment, we have made our SEA-AD data freely and openly available as fast as possible (well ahead of any publications). We believe this facilitates and enables new research directions and collaborations that build on our dataset and the field’s understanding. Our collaboration with Nautilus is an excellent example of that. We look forward to everything we will learn in the coming months.

It’s early days in terms of understanding how proteoforms contribute to disease. It is my greatest hope that this collaboration:

1. Makes it clear that quantifying proteoforms vastly expands our understanding of the molecular landscape of health and disease.
2. Opens new research avenues – by seeing the results from our tau work, we hope people will begin to see new research paths paved by the insights proteoforms provide.
3. Identifies proteoform-based biomarkers and, perhaps further down the line, therapeutic targets that will form the basis of next-generation Alzheimer’s disease precision medicines.

We anticipate that the impact of this work will extend far beyond even something as critically important as tau and AD. Measuring tau proteoforms is the first application of our Iterative Mapping technology, and we’re definitely not stopping here. There’s evidence that proteoforms drive the biology of many diseases, including other neurodegenerative disorders, cancer, heart disease and more. We just haven’t had accessible tools for measuring these proteoforms.

We’re opening the proteoform landscape to researchers in a way that’s never been done before, and I am confident that proteoform analysis has the potential to vastly expand our understanding of biology. The ultimate result, hopefully, will be a treasure trove of new drug targets across all disease areas.