The tao of tau: Part 2

As promised, here’s the second of two installments on the latest in tau-ology, at least as of the Tau2020 conference, held in February 2020. Yesterday’s post covered treatment, and today’s, everything else. There was no Tau2021, but Tau2022 is planned for February.

Tau structure and function

More is becoming known about the N-terminal domain of tau, where exons 2 and 3 are alternatively spliced (i.e., some forms of tau have the amino acid sequence encoded in the MAPT gene’s exon 2, others have those of exons 2 and 3).  This end of the protein is now suspected of controlling the spacing between microtubules, which comprise the cell’s internal skeleton and transport system.  Next to that is the “proline-rich domain,” which interacts with enzymes and with other proteins that include the WW domain.  That’s where a protein includes the amino acid tryptophan occurring in a string that regulates signaling between proteins. This is part of the new body of evidence that tau is partly a signaling protein.  (Fun fact: The single-letter amino acid abbreviation system (allegedly) assigned “W” to tryptophan because T was taken and part of the tryptophan’s molecule has a W shape.)  

The highlight of the conference was probably the keynote talk from Dr. Michel Goedert describing his group’s work on high-resolution imaging of tau using cryo-electron microscopy. (See my recent post, “A frozen treat” for details and an update.)  The bottom line was that each tauopathy has its own, very different, pattern of tau misfolding that is uniform across patients with that disorder.  That’s true even for disorders with the same isoforms, like PSP and CBD, which are both 4R (i.e., 4-repeat; having 4 microtubule-binding areas) but have radically different tau misfolding patterns.  Better understanding of these structures may point to disease-specific diagnostic and therapeutic innovations.  This starts to undermine the hope that PSP will provide the key to all tauopathies, including Alzheimer’s. For this and his other work, Dr. Goedert was awarded the $250,000 Rainwater Prize at this conference.

Post-translational modifications of tau

You know about tau phosphorylation, and the textbooks say that it can result from “stress.”  But more recent research has identified some specific causes such as brain ischemia, as from atherosclerosis; brain trauma; and excessive sodium intake.  The last works through activation of the immune system.

A recently-identified role for tau is to protect DNA in the cell’s nucleus from oxidative stress, which can result from certain toxins or from mitochondrial dysfunction.  Tau seems to help the histone proteins in the nucleus do their job of regulating access to the genome.

Abnormally phosphorylated tau in the cytoplasm can indirectly affect the function of DNA in the nucleus in many ways.  These include deranging the function of actin and microtubules, which maintain the structure of the nucleus, damaging DNA, damaging the protein portion of chromosomes, affecting RNA handling, reducing ribosome stability and encouraging DNA code rearrangements.  The next problem is to determine exactly which PTM’s do what and to develop drugs specifically targeted at those that actually cause neurodegenerative diseases. In fact, we don’t yet even know for sure that hyperphosphorylated tau causes tau aggregation in humans.

Although we know of 50 different mutations in the MAPT (i.e., the tau) gene that cause or increase risk of tauopathies, none of them is known to act via causing PTM’s.  This confirms that the pathogenesis of PSP is multifactorial and that the combination of factors differs across individuals.

Tau genetics

It has been known since 1999 that PSP is associated with a genetic variant called the H1 haplotype.  This consists of a section of chromosome 17, comprising the MAPT gene and about 15 others, that is reversed relative to the rest of the chromosome.  More recently, variants, especially the H1c subhaplotype, have been discovered that associate more strongly and specifically with PSP. The mechanism may be to allow the protein transcription machinery access to other areas of the genome where the handful of other genes associated with PSP have been identified. 

A project under the aegis of the Tau Consortium and the scientific leadership of Celeste Karch, Alison Goate and Sally Temple has created a collection of cells from 140 (and counting) volunteers with pure tauopathies such as PSP and some of their family members.  Some of the cells are fibroblasts taken from skin biopsies, others are stem cells (i.e., induced pluripotent stem cells, IPSCs) made from such fibroblasts and still others are neural cells made from those stem cells.  Some of the volunteers had single-gene causes of their illness and others had a single genetic variant that increases risk for PSP but is insufficient to cause it.  Some of the cells from the latter group with a single, identifiable mutation have had that mutation corrected using CRISPR, leaving a cell culture with only the poorly-understood genetic “background” necessary to cause the disease.  This represents a valuable tool for studying garden-variety, “sporadic” (i.e., apparently with no familial clustering) PSP. The cells are offered as a resource to carefully vetted researchers worldwide.

Mutated versions of a gene called LRRK2 (“lark-two”) are known as a group to be the strongest genetic contributors to Parkinson’s disease.  But strangely, a few people with some of LRRK2 mutations, including the most common one, develop PSP rather than PD. Now, Drs. Edwin Jabbari, Huw Morris and colleagues used samples from the UK’s Parkinson’s Disease Society Brain Bank to find that a genetic marker close to LRRK2 is associated with more rapid progression of PSP.  LRRK2 is known to affect disposal of dysfunction or excessive protein, possibly including tau. It also is involved in neuroinflammation. Both mechanisms are at or near the top of the current list of contributors to the pathogenesis of PSP.  The logical next step is to develop drugs suppressing the toxic activity of the enzyme produced by the LRRK2 gene.

The significance of the extreme predominance of 4-repeat tau in the tangles of PSP and CBD remains unclear.  (Most other tauopathies have an equal combination of 3-repeat and 4-repeat tau “isoforms,” mimicking tau in normal human brain, and a few tauopathies have predominantly 3-repeat tau.)  At Tau2020, The Rainwater Charitable Foundation awarded its early-career award to Dr. Patrick Hsu of the Salk Institute for a new technique that allows researchers to control the number of repeats in tau.  It’s based on the same general type of RNA-manipulating technology as CRISPR-Cas9, but in this case it’s called CRISPR-CasRx. It can be adapted to manipulate “alternative splicing” not only of tau, but also that of many other proteins that, like tau, have multiple isoforms. So far, the technique is only for neuronal cell cultures, but it opens up a world of potential experiments to fix the molecular variation in tau underlying PSP and CBD.

Prion-like tau propagation

We know that in cells growing in a researcher’s dish or in a mouse’s brain, misfolded tau introduced into the system can travel cell to cell, templating new copies of itself along the way.  But the details of the process and its relevance to the human tauopathies remain unclear.  In fact, in no human tauopathy has such a process been conclusively demonstrated, although it has been clearly observed in the prion protein disorders such as Creutzfeldt-Jacob disease or mad cow disease.  So, as we try to confirm the prion-like hypothesis in the tauopathies, we do have to remain open to alternative ideas to explain the spread of the disease within the brain.

A clue to why tau is secreted by brain cells could be its recently-discovered role as a local hormone or signaling molecule, to regulate the activity of brain cells and the sleep-wake cycle. However, even here the mechanisms are unclear.  There is recent evidence that tau is released from cells encapsulated in tiny membrane bubbles called vesicles.  In that case, the tau may be protected from therapeutic antibodies designed to slow the spread of tauopathies.  Additional recent evidence has found tau receptors on brain cells consisting of heparan sulfate proteoglycans, low-density lipoprotein receptor-related protein and even amyloid precursor protein.  For now, the science of tau spread remains, like most embryonic sciences, a collection of disconnected observations.

Independent of the specific molecular form or transit vehicle used by tau, imaging studies have recently shown that a prediction of the next brain region to become involved in a progressing tauopathy can be based on the involved area’s concentration of abnormal tau, its synaptic connections and its areas of direct non-synaptic contact with other cells.  Both of the latter two routes operate via active mechanisms; passive diffusion is no longer considered a factor.

Why are tau aggregates toxic?

This is a large, complicated issue, but one recently discovered clue is that brain cells containing tau aggregates, when finding themselves under stress for some other reason, signal the brain’s immune cells to come and engulf them, but without killing them. If this has the effect of protecting the tau-containing cells without preventing them from secreting their tau, then it could mean that a new tau-directed treatment could work better if coupled with a drug that inhibits the brain’s immune function.

Tau-based brain imaging

Positron emission tomography (PET) imaging using a radio-labeled glucose analog as a tracer for energy production is a standard way to help distinguish the frontotemporal disorders such as PSP, CBD and FTD from Alzheimer’s disease.  PET using a dopamine analog can distinguish the atypical parkinsonisms from Parkinson’s disease but is only available for research use.  But we have no PET-based technique to specifically identify PSP or CBD.  PET using a tau tracer is showing excellent results in distinguishing AD from other dementias, but it works poorly for PSP.  One problem is that the spatial resolution of PET is insufficient to show the tau deposits because it sticks to some other normal and abnormal molecules in the same set of neurons.  But many drug companies are developing many tau tracers and a few of them are starting to show more validity for PSP.  One, called [3H]CBD-2115, has shown good accuracy but doesn’t cross the blood-brain barrier.  However, some tweaks to the molecule or to the BBB might solve that problem.  Hopefully, it will be safe to administer and once broken down in the brain, won’t leave any radioactive remains behind for any length of time.

An exciting but early-stage development is that Genentech and AC Immune are developing an anti-tau antibody called semorinemab in tandem with a tau-directed PET tracer called [18F]-GTP1.  The PET tracer, in addition to tracking any slowing of tau aggregation, may also be useful in allowing a measurement of tau aggregation at the study baseline to predict the individual’s disease progression absent any intervention.  That, in turn, could allow a more individualized interpretation of the neuroprotective drug’s benefit.  Although that seems a good model for drug testing, it was announced in September 2021 that in prodromal and mild AD, semorinemab failed to slow progression in most of the outcome measures but did slow progression by 44% in one critical bedside test, the ADAS-Cog11.  The sister trial in moderate AD continues.  The companies have not announced plans for testing in PSP.

The NIH and the Rainwater Charitable Foundation have each created consortia to develop tau-based PET tracers for non-AD tauopathies.  The RCF effort involves the Michael J. Fox Foundation in its efforts to distinguish PD from PSP and CBD.

The tao of tau: Part 1

Our knowledge of the tau protein and the tauopathies is expanding faster than ever.  In February 2020, just before the lockdown, the Alzheimer Association, the Rainwater Charitable Foundation and CurePSP hosted an international conference in Washington, DC to discuss the latest in tau-ology.  “Tau2020” drew 650 people from 21 countries – academic researchers, government regulators, lay organization leaders, philanthropists, and key players in the pharmaceutical and biotech industries. 

A 12,000-word article summarizing the two days of talks appeared last week in the journal Alzheimer’s and Dementia.  In this post and the next, I summarize that summary – with an eye in particular on PSP and CBD, though much of the conference’s attention, understandably, was on Alzheimer’s disease and frontotemporal dementia. What follows is not a textbook article nor literature review or even a summary of one, but only a list of recent advances presented at the Tau2020 conference.

Today’s post will cover treatment issues and tomorrow’s will cover everything else. Dickens, Dostoevsky, Hemingway, Joyce, and Nabokov published in installments – why can’t I?  

*****

The failure in clinical PSP trials of two monoclonal antibodies directed against tau’s N-terminal (i.e., the “beginning” or left end of the molecule as conventionally depicted) has not prevented researchers from developing antibodies against other parts of the tau molecule. Antibodies against (in conventional left-to-right order) the proline-rich region, the microtubule binding region or the C-terminal region are all under investigation. We now know that antibodies directed against the central regions of tau are more effective in preventing tau spread in cultured human neurons.  In fact, administering a mixture of antibodies directed against more than one of these regions is being contemplated. 

The two trials that ultimately failed in PSP chose the N-terminal as the target for their antibodies because in Alzheimer’s disease, misfolding starts in that area and because fragments of tau that include the N-terminal increase beta-amyloid production, an observation irrelevant to PSP.  Perhaps for these reasons, the two companies have continued their AD programs for their N-terminal-directed antibodies while discontinuing their PSP programs.  In my opinion, this is another potential difference between AD and PSP that weakens the hypothesis that a tau-directed prevention of either will be a prevention for both.

Several conference speakers emphasized that going forward, our ability to detect a neuroprotective benefit of any tau-based drug will improve if the clinical trials can manage to exclude non-tauopathy patients with disorders mimicking a tauopathy.  Such patients include those with dementia with Lewy bodies, multiple system atrophy, TDP-43-based frontotemporal dementia, normal-pressure hydrocephalus and vascular parkinsonism.  The speakers also agreed that new diagnostic markers will have to detect disease at a much earlier stage to give patients the greatest likelihood of responding while still able to live a good-quality life.  They also felt that future trials should avoid subjects taking many concomitant medications that could obscure the benefit of the drug being tested.  A further recommendation was to tailor the specific outcome measure to each patient’s combination of deficits and that the various subtypes of PSP, each of which may have its own mode of tau spread, are unlikely to respond equally to the same neuroprotective agent.  Finally, they noted that we do not yet know how far tau must be reduced to achieve slowing of disease progression, and that this may vary across patients.

Another active area, but so far applied only to AD, is active anti-tau vaccines.  This is where a person with or without disease receives an inactive fragment or form of tau, inducing the immune system to make antibodies directed against disease-causing tau that persist for months, years or decades.  This is the mechanism for most vaccines in common use.  It differs from the monoclonal antibody-based passive “vaccination,” where antibodies themselves are infused from the outset and typically wear off in a month or so.  Probably the active vaccine furthest down the clinical pipeline is ACI-35, from Axon Neuroscience.  Again, trials so far are in AD and no plans to extend them to PSP have been announced.

Anti-sense oligonucleotides (ASOs) have entered early clinical trials in both PSP and AD.  ASOs are short, single-stranded synthetic DNA molecules.  They bind to the messenger RNA that that assists in the production of tau, in turn inducing the enzyme RNase H1 to degrade that mRNA.  Unfortunately, ASOs can only be given by injection into the spinal fluid, as they cannot cross the blood-brain barrier.  The procedure must be repeated every three months. Another potential drawback is that ASOs reduce production of normal tau as much as they do abnormal tau. A Phase 1 safety and tolerability trial in 64 patients is being conducted at Mayo Clinic Rochester, Vanderbilt University and at two sites in Canada, three in Germany and one in the UK. For more information, visit http://www.clinicaltrials.gov (www.ClinicalTrials.gov Identifier: NCT04539041).

A highly innovative solution to the problem crossing the blood-brain barrier that limits utility of the ASOs and many other potential new neurological drugs.  Denali Therapeutics is testing a “transport vehicle,” a molecule that can escort a variety of drugs across the BBB.  It works by binding to an antibody with the “transferrin receptor” protein genetically engineered into its structure.  The therapeutic drug is then bound to this complex, which can penetrate the BBB.  Denali is experimenting with attaching anti-tau antibodies or tau-directed ASOs to this vehicle. Trials in humans have not yet begun.

Another approach that sounds like science fiction is “proteolysis-targeting chimeras (PROTAC) molecules, first developed in 2001.  These are hybrids of two small molecules, one of which can bind to, in this case, tau, and the other that is recognized by the brain cells’ internal garbage disposal called the ubiquitin-proteasome system.  Unlike antibodies, PROTACs can easily enter brain cells.  Their only human application so far has been in oncology, but experiments in mouse tauopathy models are proving successful.

A new methodology in clinical trial design is the “basket study,” common in oncology, where several rare types of cancer are tested as a group.  We still know too little about the differences among the tauopathies to insist that in evaluating a new drug, each disease must have its own trial.  The one basket study to date that included PSP was performed at UCSF for the drug TPI-287, which stabilizes microtubules.  AD, PSP and CBS were included.  Unfortunately, the drug proved inefficacious, but the trial method proved practical.

Reports from the front

Last week the International Parkinson and Movement Disorder Society held its annual meeting – by Zoom, natch.  (A pity – in normal times, that meeting is held each year in a different, interesting international city, letting us append a culturally-oriented vacation to a consistently great conference.  But COVID has done far worse to others, so I shouldn’t complain, and the Zoom format allowed clinicians working in remote or poor countries to participate, a major plus.)

There were 1,320 original research presentations, all in the form of posters.  Of those, 37 came up in my search on “progressive supranuclear palsy.” Presented now for your delectation, in random order, are summaries of the top five presentations of original research in PSP along with my own take on their importance and implications: 

  • Imaging/Gait Correlates: 

A group from the Mayo Clinic led by Dr. Irene Sintini performed detailed gait analysis in 19 people with PSP and imaged their brains in three ways: non-contrast MRI (looking for atrophy in specific areas), diffusion tensor imaging (an advanced MRI technique showing fibers connecting the various parts of the brain and the speed of fluid slowly flowing through in them), and flortaucipir PET (showing tau aggregation, a research technique not quite ready for routine clinical diagnostic use in PSP, but working quite well for Alzheimer’s).  Then they looked for associations among the gait and imaging measures.    

They identified two general patterns of association:  One was that better gait velocity, stride length, and gait stability were associated with larger frontal lobe volumes and less flortaucipir uptake in the precentral gyrus, which is a part of the frontal lobe that controls movement.  The other was that worse postural imbalance was related to greater flortaucipir uptake in the left paracentral lobule, which helps control movement and sensation on the right side as well as bowel and bladder function.

Admittedly, this was kind of a “fishing expedition,” an unkind term we use for a research project that’s just looking for patterns of abnormalities without a specific hypothesis in mind.  But that’s how the process leading to eventual breakthroughs can begin.  In this case, associating specific gait-related abnormalities of PSP with specific brain regions could point the way to deep-brain stimulation techniques, cell replacement therapies, or transcranial (i.e., non-invasive, painless) magnetic or electrical stimulation treatment.  Besides, who knows what gene therapy might come along in a few years to take advantage of this groundwork in some still-undreamt-of, anatomically-directed way?

  • Diagnostic MicroRNA:

Dr. Ravi Yadav and colleagues at India’s National Institute of Mental Health and Neurosciences in Bangalore compared microRNA in plasma (blood without its cells) from 18 patients with PSP and 17 healthy controls matched for age and sex.  They used a type of polymerase chain reaction (PCR) test that, unlike PCR for forensic purposes or COVID testing, provides quantitative measurements.  MiRNA regulates many things in cells, acting much like enzymes.

They found five kinds of miRNA where the difference between PSP and controls was large enough to serve as a diagnostic test.  A commonly used measure of the ability of a diagnostic test to work well at the individual level rather than merely to distinguish groups by their averages is called the “area under the receiver operating characteristic curve” (AUC).  A good AUC is at least 0.8, with a perfect test being 1.0.  These five miRNAs’ AUC’s ranged from 0.78 to 0.86. 

This is promising, and if refined and perhaps combined with another moderately accurate test, could provide an excellent test for PSP.  But first, we need replication in a larger study and a different lab; PSP has to be compared by this test to other neurodegenerative disorders, not just to healthy controls; and cases with early, diagnostically-equivocal signs of PSP should receive this test and then be followed until a diagnosis declares itself.

  • Astrocyte Proteomics:

Astrocytes are the main type of glial cells in the brain.  They are largely non-electrical but perform many supportive functions for the neurons and may actually process information.  They are where the abnormalities of PSP start.  These cells don’t normally make tau but they can accumulate it in the tauopathies, and their resulting appearance, called “tufted astrocytes,” is the pivotal diagnostic feature of PSP through the microscope. 

Dr. Felipe Ravagnani and co-workers from the University of São Paolo, Brazil were interested in what genes are expressed into protein more intensively in such cells from patients with PSP relative to the same cells from healthy controls.  You can’t take astrocytes from a living person’s brain, so they created them in a dish by taking fibroblast cells from the lowest layer of a skin biopsy from each subject and treated them with various things to first remove their skin specializations and to become stem cells.  Then they treated with other things to turn them into astrocytes and compiled and compared the proteins in PSP-patient-derived astrocytes to those from controls.

Such a technique ordinarily produces a long list of differences that’s hard to draw any conclusions from.  So it pays to classify the proteins that differ between the two groups into the kinds of cellular and biochemical pathways in which they participate.  This incriminated two pathways.  One was for cell cycle activation, which is how the cell decides when to divide and to stop dividing.  The other was for one of the chaperone pathways, in this case CTT/TriC, which is important to axonal transport and to degradation of abnormal or excessive proteins.  Both pathways involve tau.

It’s still too early to know what to make of this, and proteomics research is notoriously subject to methodologic variables.  But the chaperone pathways, which are important in regulating protein folding, have been suspected for many years as part of the cause of PSP.  Cell cycle abnormalities are critical to cancer, but the opposite problem – insufficient cell division – could contribute to glial pathology and start the more general neurodegenerative process.  If the new results are confirmed, they would present new targets for drug development.

  • CBD Diagnostic Accuracy:

Just as PSP has multiple variants, so does corticobasal degeneration (CBD).  The most common and the classic form is called CBD-corticobasal syndrome (CBD-CBS).  Other common ones are CBD-PSP syndrome, CBD-frontal behavioral/spatial syndrome (CBD-FBS) and CBD-nonfluent/agrammatic variant aphasia (CBD-NAV).  A set of clinical diagnostic criteria for CBD was published in 2013. 

Now, Drs. Danielle Lux and colleagues at University College London have evaluated the accuracy of the CBD criteria in predicting actual CBD pathology at autopsy in 133 cases.  They found the positive predictive value (PPV) of the “probable CBD” criteria was only 33% and the PPV for the “possible CBD” criteria was only 51%.  (PPV is equal to the number who actually have the disease on autopsy divided by the number who have tested positive during life by satisfying the clinical criteria.)  They also found that CBD-NAV had a better PPV than the other variants.  As an aside, CBD-PSP had the most rapid course of all of the variants assessed.

These results confirm and extend previous reports in the literature using smaller sample sizes.  They better elucidate CBD’s wide variety of clinical presentations. This variety is the main reason why almost all of the clinical trials so far testing tau-directed treatment enroll people with PSP-Richardson syndrome, not CBD — the PSP-RS diagnostic criteria have a much greater PPV for actual PSP pathology at autopsy.  We hope that tau PET imaging and fluid biomarkers in CSF or blood will soon correct this situation for folks with CBD.

  • A Survival Model:

Until the fine day comes when we can prevent or slow the progression of PSP, predicting survival is important.  Patients and families need to plan psychologically and financially.  Designers of long-term treatment trials need to know the likely dropout rate due to death. 

Dr. Tao Xie and colleagues at the University of Chicago have added usefully to the considerable existing literature on this topic.  In 23 patients who had died with PSP, they recorded the time from onset of the first PSP symptom to onset of downgaze palsy; the severity of downgaze palsy at that point using a scale ranging from 10 to 100; sex; age at PSP onset; motor function; and use of medication for parkinsonism or for pulmonary or cardiovascular diseases.  They used those data to create a formula by which to calculate survival from onset to death.

They found that total survival duration in years can be predicted by the equation: 5.76 + (1.11 x disease duration at the assessment) – (0.03 x downgaze palsy severity at the time of the assessment) – (0.03 x the age of onset).  The result predicted total survival duration reasonably accurately, with an average error of 0.82 (standard deviation 0.67) years.

The method of measuring the downgaze palsy was not described in the brief presentation and may be somewhat subjective, so an even more accurate prediction may be feasible using other measures.  Also, it’s not clear from the material presented that other features of PSP might perform as well as downgaze palsy as a predictor of survival.  For example, last year my colleagues and I published a very different method of estimating PSP survival that works about as well as this new one but requires administration of the whole 28-item PSP Rating Scale. 

At the scientific level, it’s interesting that in the mathematical model of Xie et al, downgaze palsy is an important factor in predicting death despite involving only a small portion of the total brain pathology.  I say that because in PSP, death is not particularly related to visual problems, but is usually the result of overall immobility, poor nutrition, aspiration and bladder infections.  But even if the measure of downgaze palsy only provides the model with an easily-measured proxy for those other disabilities, the model would still be a convenient and useful service for patients and families. 

——————————

Keep in mind that despite clearing the bar for acceptance at this conference, these research reports have not been subjected to a detailed peer-review process.  In fact, most original presentations at most conferences are never published in anything like the same form.  But I chose to relay these five to you because I think that in the end, they’re likely to stand up to scrutiny and to influence scientific thinking or bedside practice. 

Post-post modifications

Since yesterday’s post about the discovery of the detailed pattern of tau misfolding in the various tauopathies went live, some discussions with colleagues and laypersons have prompted me to bang out the following Q/A:

Q: Does the newly discovered, major difference in tau folding between PSP and CBD mean that CurePSP and researchers interested in PSP will devote fewer resources to CBD going forward?

A: Absolutely not.  The two diseases still have a lot more commonalities than differences and each will continue to benefit from research and patient care insights into the other.

Q: Does this new information mean that PSP has less value as a test bed for treatments aimed at Alzheimer’s disease?

A:  Only very slightly.  All of the tau-based treatments in the research pipeline are aimed at tau in general, not any specific folding pattern.  Over time, it may emerge that the parts of the tau molecule that are more “buried” within the folding structure are less able to interact with antibodies or drugs.  If that inaccessible section is different in PSP compared with in Alzheimer’s, then a drug company or researcher interested in AD may decide to direct their efforts accordingly.  But most of the tau molecule will probably prove to be equally accessible, for practical purposes, in all of the folding conformations, in which case, PSP will remain as valid as ever as a test bed.

Q: How definite is the new discovery?

A: Like any scientific discovery, it needs to be confirmed by different researchers in different labs using slightly different techniques. 

Q: Will the new discovery accelerate finding a prevention or cure*?

A: Possibly, and it’s hard to say by how much.  An effective prevention or cure may not care about the specific tau folding pattern.  For example, a very promising type of prevention now entering clinical trials is anti-sense oligonucleotides (ASO’s), which are small strands of RNA that prevent the genetic code for tau from being translated into tau protein.  The misfolding occurs at a later stage, so an ASO directed at tau should help reduce tau production in any tauopathy.  (Whether the amount of tau production is the key to the disease is another question.)

*Technically, a “cure” requires not only halting of the disease progression but also repair of the existing damage.  Those are two different things, and the latter is more difficult to achieve for a complex structure like the brain.  If we can find a way to diagnose PSP in its earliest stages, before there’s enough damage to cause symptoms, then all we need is a way to prevent further damage.

Q: What determines the tau folding pattern in the first place?

A: Any of several things that we know of, and there are undoubtedly more that we don’t yet know of.  The leading candidate is the attachment of small molecules to the tau protein that affect its ability to bend and to stick to itself.  The most common such molecule by far is phosphate (one phosphorus atom and three oxygen atoms) but others are various sugars, lipids, acetyl groups, sulfyl groups, methyl groups and dozens of others (Wikipedia has a nice list). These are called “post-translational modifications” (PTMs) and they’re normal, but like any normal biological process, they can go awry.

Other things causing protein misfolding in some disorders are environmental toxins and genetic abnormalities.  We don’t yet know the relevance of these to PSP or CBD, but there are some indications that it’s not zero.

We don’t know why certain diseases have abnormalities of PTMs.  We don’t even know, in most cases, if the abnormal PTMs are a cause of the tissue damage or a collateral effect of the same toxic process that causes that damage.  The new information about tau folding patterns will certainly intensify the already intense search for those answers.

A frozen treat

I wish I could report on some breakthrough in the treatment or prevention of PSP, but that hasn’t happened yet. But what I can do is to report on a major advance in understanding an “upstream” step in the process that’s killing brain cells in PSP. The abnormal tau molecule in PSP has now been imaged.


The prevailing, and probably correct, theory is that brain cells are damaged somehow by tau protein that has folded upon itself, causes its healthy sibling tau molecules to similarly misfold via a templating process. The tau then forms stacks like checkers, called “fibrils,” which are toxic to the cells. Tau doesn’t normally have any folding pattern at all – if it’s not attached to the cells’ internal skeleton, doing it’s normal job, it’s floating around in the cytoplasm (the cells’ internal fluid) like overcooked spaghetti in boiling water. We’ve known for decades that through the electron microscope, the fibrils of PSP look different from those of Alzheimer’s, which in turn are different from those of corticobasal degeneration, which are different from those of Pick’s disease, and so on. We’ve hypothesized that this is because the pattern of misfolding determines the geometry of the stacks of tau, and those stacks determine which cells are attacked, and that determines the patient’s symptoms. But we had no details.


Well, now we do, thanks to a new lab technique called “cryo-electron microscopy” or “cryo-EM.” That’s where the sample is imaged at a temperature cold enough to keep it from wiggling around as much, allowing a far greater visual resolution than was possible using other methods. It still does wiggle a bit, which is why the technique starts by making a video of a tau molecule and then computer-averages the image. The resulting still photo shows the protein molecule’s pattern of loops and folds; even the individual amino acids are visible as fuzzy bumps on the chain. The resolution of the images was 2.7 Angstroms, about the same as the diameter of a water molecule.

Over the past couple of years, researchers applied cryo-EM to tau in a few tauopathies including AD, CBD and Pick’s disease. Only now has PSP been studied by this technology, along with a bunch of much less common tauopathies. The new paper is from an international team led by Michel Goedert, Sjors H. W. Scheres, Yang Shi and Wenjuan Zhang, all of Cambridge University. It was published in Nature, probably the world’s most prestigious and selective biomedical journal.

Misfolded tau from a patient with PSP (Shi Y., et al. Nature 2021)

They imaged tau from 11 diseases, a number large enough to justify a classification system using a few different features such as the number of layered folds and hairpin turns. They found that the folding patterns of the various PSP subtypes were identical except for one case of a rare type, PSP-F, a variant with disproproportionate frontal lobe behavior and cognitive problems. Under the new classification system based on tau folding, the tauopathy most similar to PSP was globular glial tauopathy, a rare cause of dementia diagnosable only at autopsy.


They found the folding pattern of Alzheimer’s tau to be identical to that of primary age-related tauopathy (PART), familial British dementia (FBD) and familial Danish dementia (FDD). The pattern in chronic traumatic encephalopathy (CTE) was similar but not identical to that. Corticobasal degeneration (CBD) was similar to argyrophilic grain disease (AGD) and aging-related tau astrogliopathy (ARTAG). Finally there was Pick’s disease, which was similar to none of the other ten.

SHI Y ET AL. NATURE 2021


Why is all this important? I’ll let the authors themselves explain:


“The presence of a specific tau fold in a given disease is consistent with its formation in a small number of brain cells, followed by the prion-like like spreading of tau inclusions. This may underlie the temporo-spatial staging of disease. Knowledge of the tau folds in the different diseases provides a framework for studying tauopathies that will lead to a better understanding of disease pathogenesis. At a diagnostic level, our findings will inform ongoing efforts to develop more specific and sensitive tau biomarkers.”

SHI Y ET AL. NATURE 2021


By “temporo-spatial spreading,” they mean where in the brain the damage spreads and how fast. So the next big step is to figure out just how the molecular structure presented on the surface of each kind of misfolded tau interacts with healthy brain structures. Once we do that, we can find a monkey wrench to throw into that process – one monkey wrench for PSP and globular glial tauopathy; another for AD and its pals PART, FBD and FDD and possibly CTE; one for Pick’s; and yet another for CBD, AGD and ARTAG.

I’ll try to keep you updated on this, and soon I’ll opine on the collateral question of whether this new work casts doubt on our long-held position that PSP, as a “pure tauopathy,” is a good test-bed for all of the tauopathies, including AD.