Critique of pure prionopathy

If you follow the latest in neurodegenerative disease research, you’ve heard the “prion hypothesis” or “pathogenic spread hypothesis.”  For the past five or six years, it’s been widely claimed, and almost as widely accepted, that the proteins that mis-fold and aggregate in the brain cells in PSP, as well as in its big brothers Alzheimer’s and Parkinson’s and the rest, spread through the brain in a way similar to how prion protein spreads through the brain in the prion diseases such as Creutzfeldt-Jakob disease, mad cow disease and kuru.  A respectable body of experimental evidence supports — or at least is compatible with —  this idea.

But now a pair of highly respected Harvard neuroscientists, Dominic Walsh and Dennis Selkoe, have said not so fast.  In a very well-balanced and dispassionate review of the prion hypothesis in Nature Reviews / Neuroscience, they show that while the existing evidence is compatible with cell-to-cell spread of toxic protein aggregates, there is still plenty of room for a hypothesis that posits selective cell vulnerability with a more generalized toxic influence.  I won’t get into the technical weeds, but here are their major points:

  1. Even in the classical prion disorders, it is well-accepted that “cell-autonomous” factors, rather than just spread from nearby cells, determines which cells are and are not involved.  The salient example is that the asparagine-for-aspartate mutation at position 178 in the prion protein causes familial CJD when the person has a valine at position 129 in the same protein but causes fatal familial insomnia with there’s a methionine at 129.  (Neither of the latter substitutions by itself is pathogenic.)
  2. The “pathogenic spread” hypothesis rests in no small part on the observations of Braak and colleagues that early-stage Alzheimer’s or Parkinson’s pathology in people dying from other causes is confined to certain specific brain areas, suggesting that the process starts there and spreads.   But Walsh and Selkoe point out that those early sites of pathology may merely be the areas most sensitive to a generalized insult.  Furthermore, only about half of the cases of each of those diseases followed that pattern.
  3. Another buttress for the pathogenic spread hypothesis is the observation that 5-10% of fetal substantia nigra cells transplanted into the striatum of patients with Parkinson’s developed Lewy bodies themselves after a number of years. But this need not be the result of spread of pathogenic alpha-synuclein; it could be the result of a more generic insult such as inflammation in the injection site, where most of the injected cells have died.  They cite evidence that activation of microglia (the brain’s inflammatory cells) in other types of neural grafts can produce Lewy bodies in those grafts.
  4. The experiments showing that injected alpha-synuclein or tau protein can induce the formation of aggregates in host brain is incomplete because they do not adequately demonstrate actual cell loss or impairment of brain function in the host animal. We know from other lines of experiment that aggregates alone do not correlate well with neurological impairment in human or experimental neurodegenerative disease.
  5. The pathologic anatomy of rare, dominantly inherited forms of Alzheimer’s, Parkinson’s and frontotemporal dementia fits well within the spectrum of their corresponding sporadic conditions. A genetic cause, producing the same intense pressure for protein aggregation in many areas of the brain simultaneously, would not be expected to mimic the anatomic pattern of a single-anatomic-source process posited by the pathogenic spread hypothesis.
  6. There are still many questions left unanswered by the pathogenic spread hypothesis. This doesn’t directly contradict its other tenets, but it weakens its explanatory power. It cannot explain the initial protein misfolding; how the aggregates are released; how they remain aggregated in the interstitial fluid where the concentration of the protein is far less; why they don’t stick to the outsides of cells after being excreted, as their physical chemical characteristics suggest they should; how they choose only certain target cells to penetrate; and how the aggregates escape into the cytoplasm from the membrane vesicles that presumably would be the vehicles by which they penetrate their targets.


As a final point, Walsh and Selkoe make a case for avoiding the term “prion-like” or “prion-oid” with reference to neurodegenerative diseases unrelated to the prion protein itself.  They list several known features of prion protein spread in the known prion diseases that as far as we know are absent in PSP, Alzheimer’s, Parkinson’s, etc.  They also cite the absence of any known transmissibility of the non-prion-protein and point out that we don’t know enough about either group of diseases to equate them at that level of terminology.

Excellent scientists that they are, Walsh and Selkoe describe a set of experiments to undertake and new research tools to develop in order to strengthen or reject the pathogenic spread hypothesis.  Maybe I’ll get to that in another post.  But they end with the hope that the pathogenic spread hypothesis is true, for that would provide many potential therapeutic targets that would not otherwise exist.

Is PSP the route to not just Alzheimer’s but also Parkinson’s?

We’ve known for many years that Parkinson’s disease, which the textbooks call an α-synucleinopathy, has some aggregated tau as well. It appears that each of the two proteins, once misfolded, not only induces its own normal brethren to misfold, it also induces copies of the other to misfold.
The first demonstration of this synergistic misfolding and resulting aggregation came in a series of three papers between 2002 and 2004 from the lab of John Trojanowski and Virginia Lee at Penn. The first authors were John Duda, Bernard Giasson and Paul Kotzbauer. (Disclaimer: I was one of their co-authors on all three.)

Now, Julia Gerson, a grad student in the lab of Rakez Kayed at the University of Texas Galveston, has presented work at the Society for Neuroscience that harnesses that finding of a PD/tauopathy overlap for therapeutic purposes. (Another disclaimer: Kayed has a related grant from CurePSP, where I chair the grant review.)

Gerson and friends created antibodies directed at oligomeric tau, which is tau in small aggregates of maybe 20 or 30 molecules, which are still small enough to remain in solution in the cytoplasm and therefore invisible to light microscopy, unlike mature neurofibrillary tangles. They didn’t want to target normal, non-aggregated tau for fear of disrupting the normal function of that protein, which is to stabilize microtubules.

They injected those anti-tau antibodies into mice that had a copy of a variant of the human gene encoding α-synuclein. The variation was an G209A mutation, producing an A53T alteration in the resulting protein. This is the mutation that my colleagues and I found in 1997 as the cause of PD in a large Italian-American family with autosomal-dominant PD, a finding that first linked PD with α-synuclein. (That’s Disclaimer #3. You’re starting to see why I’m so interested in this new finding.)

The antibody protected the mice against the loss of dopaminergic neurons that the α-synuclein mutation caused in the untreated mutant mice. Mice that received antibody against normal tau did even more poorly than the controls.

So here’s the take-home: Developing an anti-tau antibody for treatment of PSP may also help Parkinson’s. We already expect that it may help Alzheimer’s because that’s clearly a tau disorder. But now, the synergistic toxicity of tau and α-synuclein could also allow a single anti-tau antibody to protect against both Parkinson’s and dementia with Lewy bodies (which also has aggregation of both proteins).

If I were the drug companies, I’d be sitting up and taking notice. Two companies, Bristol-Myers Squibb and AbbVie (licensing an antibody from C2N) have already started anti-tau antibody trials in people with PSP. Others have anti-tau programs in progress.

This new report, which may extend the utility of those products to Parkinson’s, should give that snowball an extra push.

A new drug target from an epidemiologic observation

I think what jolted me out of my multi-month posting torpor is next week’s annual meeting of the Movement Disorders Society. I’ve been preparing a lecture on the treatment of PSP, CBD and MSA, and that got my juices flowing.
Speaking of treatment, an interesting paper that appeared in Plos One during my writer’s block came out of Günter Höglinger’s lab in Munich. Julius Bruch was first author. It builds on the observation that people with Guadeloupean tauopathy are far more likely than local controls to have consumed the fruits sweetsop and soursop, which contain annonacin, a mitochondrial Complex I inhibitor. Subsequent work with annonacin has suggested that it can cause a tauopathy in rats.
The new paper found that annonacin upregulates the production of 4R tau, the predominant form in PSP and some other tauopathies, by favoring the inclusion of the exon 10 peptide product into the finished tau molecule. Further experiments described in the same paper showed that annonacin upregulates the splicing factor SRSF2, which is one of a handful of factors known to regulate splicing of exon 10. So they used silencing RNA to knock down SRSF2. The result was a dramatic reduction in 4R tau.

They then took the next step and analyzed human PSP brain tissue for SRSF2, finding it markedly elevated compared to controls with no neurological disease.

To examine the possibility that the elevation of 4R tau and SRSF2 by annonacin was the result of mitochondrial Complex I inhibition rather than of nonspecific cellular stress or nutrient deprivation, they treated neuronal cultures with MPP+, a well-studied Complex I inhibitor, but not with 6-hydroxydopamine, a toxin that works independent of Complex I, or with nonspecific nutritional deprivation.

So it looks like a drug that inhibits SRSF2 could correct the abnormal 4R/3R ratio in PSP and potentially prevent cell loss. But a lot of work remains to determine how important this particular pathway is in causing the cell loss. The highly variable 4R/3R concentration across different brain areas in PSP and the existence of tauopathies with normal or low 4R/3R ratios show that the story isn’t so simple. But with the recent explosion of interest from drug companies in PSP as a route to Alzheimer’s disease, any new approach could attract interest, and this one deserves a place on the list.  I don’t know if any existing or approved drugs inhibit SRSF2, but that could be a good job for a lab that’s tooled up for high-throughput screening.

Not Your Father’s PSP

As it turns out, PSP comes in many clinical flavors. Back in the 80s I remember some patients whose illness looked like Parkinson’s until I realized that they weren’t responding to my levodopa prescriptions, at which point I repeated a careful ocular motor exam and found square wave jerks and slow downward saccades. I also remember one member of my first series of 41 patients with PSP from 1988 with severe gait apraxia and freezing as his most disabling feature.
Then, in 2005, David Williams and colleagues, mentored by Andrew Lees at Queen Square, published what is probably the most important clinical paper on PSP in the half-century since Steele, Richardson and Olszewski. That work delineated and named PSP-Richardson syndrome (PSP-RS ) and PSP-parkinsonism (PSP-P). This wasn’t just a new way to slice a clinical spectrum sharing the same basic pathology; the two variants actually had statistical differences by cluster analysis. This suggests that they differ at the pathoanatomic level. They even differed in the ratio of 4R/3R tau. (It turns out that the predilection of PSP’s tangles for 4R tau is driven by RS.)

Since then, a cornucopia of low-frequency clinical variants meeting pathoanatomic criteria for PSP has been described. In approximate descending order of prevalence after RS and PSP-P are corticobasal syndrome, postural Instability, pure akinesia with gait freezing, frontotemporal dementia, ocular motor predominance, progressive non-fluent aphasia, semantic dementia, and a cerebellar variant.
The clinicopathologic studies are only starting to appear, but it’s likely that they will all turn out to emphasize different cells, nuclei and brain regions. We will also probably see some subtle molecular differences among them (presaged by the 4R/3R difference between RS and PSP-P).

That sounds like different diseases to me. Different diseases shouldn’t be combined in treatment trials, genetic analyses or descriptive studies. What a mess.

Or is it? Maybe we don’t need to find causes and cures for each PSP variant individually. As they’re all tau aggregation disorders, maybe they will all yield to the same prevention. Maybe the mechanism of prion-like spread, by now pretty much a textbook verity, will apply not only to all of the “pure tauopathies” (and it’s not yet clear that all of the PSP variants are in fact pure tauopathies) but to all of the protein-aggregation-based neurodegenerative disorders. If it does, then poisoning that process could be the grand unified answer to Alzheimer’s, Parkinson’s, ALS, and PSP in all its malign variety.

The Gathering

Now’s the time to arrange your October schedule around CurePSP‘s annual International Scientific Symposium.  The proceedings will be dawn to dusk on Saturday, October 18, 2014 in Baltimore at Johns Hopkins’ Mt. Washington Conference Center.  As usual, I’m privileged to be its scientific director and master of ceremonies.

As always for this shindig, the technical level will be very high, with scientists talking to scientists.  No lay-language summaries. There’s always time for everyone to contribute ad libitum during the ample discussion periods.  A lot of the meeting’s material is unpublished and it’s a great place to form new ideas and collaborations.

The agenda has two categories of speakers: recent CurePSP grantees presenting the results of their funded work completed within the past year; and internationally-known authorities describing the state of the art in their own areas.  This year, the grantees are Diana Apetauerova, Nilufer Ertekin-Taner, Stuart Feinstein, Pau Pastor, Michael Wolfe and Benjamin Wolozin .  The invited speakers are Adam Boxer, Günter Höglinger, Virginia Lee and John Trojanowski.  As always, Dennis Dickson will update the group on the operations of CurePSP’s brain bank, which he directs, and on the projects of his arising from that collection.  There will also be a few posters, with brief, live presentations by those PIs.

But perhaps the star attraction will be Jerry Schellenberg presenting the preliminary results of his much-anticipated whole-exome sequencing project in PSP.  We have reserved a large chunk of the schedule for his talk and for talks by a number of experts in the most promising genes identified by the WES.

Registration is free!  (but required) If you’re a student (medical or grad or other), fellow (clinical or research) or resident (neurology or other), CurePSP offers travel scholarships  Breakfast and lunch will be provided to all registrants.  I’ll post the exact schedule as soon as I can, but you can assume that the first speaker will start at about 7:45 AM and that we’ll all head for the nearest bar at about 5:00.

Poster submissions are welcome.  Please submit them by September 5 to with a copy to me (  I’ll have a decision for you by Sept. 10.

For more information, go to


Is PSP a Disease?

The neurodegenerative diseases are starting to merge. The most obvious level of commonality lies at the cellular level of pathogenesis, where each disease is now hypothesized to include protein misfolding, templating, intercellular spread and damage by oligomers. Within the tauopathies, there is major overlap among “diseases,” as shown in this superb diagram from David Williams and Andrew Lees (Lancet 2009).

The blue, green and purple areas are pathological syndromes and the reddish ones are clinical syndromes. Note that all of the patients with Richardson’s syndrome and PSP-parkinsonism have classic PSP pathology, but the reverse is not true. Corticobasal syndrome is only about half explained by corticobasal degeneration pathology (though the diagram suggests about 85%), most of the rest being PSP and frontotemporal dementia pathology. Similar shortfalls in clinicopathological correlation underlying our traditional definition of a “disease” plague the rest of the tauopathy diagram. A similar diagram can be made for the α-synucleinopathies.

How to explain this to our patients? Our students? Ourselves? I like to think of neurodegenerative diseases as a set of spectrums. As there are only a limited number of neural systems available to damage, inevitably some parts of some of the spectrums will overlap in their anatomical, therefore clinical, phenotypes. This idea may seem unsatisfying to our traditional, neat system of clinicopathological pigeonholes. It’s not as easy to digest as, for example, the “autism spectrum,” where we don’t yet have the messy variable of pathological correlates to contend with. But this state of neo-nosologic confusion is only temporary. Before too long, we will have a long list of genomic, epigenetic, toxic, proteomic variants along with just plain stochastic events that in combination produce neurodegenerative disease. We will then have an understanding of such diseases that is more sophisticated and rational than the current combination of microscopical, biochemical and clinical abnormalities. These insights will render our present concept of “a disease” obsolete and make it much easier to devise prevention for most of these conditions.

PSP markers in CSF? Not yet

As a PSP-ologist, it takes a lot to discourage me, but the excellent review of CSF markers in the diagnosis of PSP did it. Nadia Magdalinou, Andrew Lees and Henrik Zetterberg of University College London, writing in the JNNP, point out that no CSF measure has been consistently or reproducibly found to differentiate PSP from all of the relevant competing diagnostic considerations.
An excellent study cited in the review found low levels of CSF α-synuclein in Parkinson’s, DLB and MSA relative to PSP and other brain disorders. A value less than 1.6 pg/μl had good (91%) positive predictive value for any synucleinopathy but higher concentrations had poor (20%) negative predictive value.  So that measure is of some small value.
Neurofilament light chain in CSF is elevated in PSP, MSA and CBD, according to another study, with an area under the ROC curve of 0.93. This has been confirmed by others since. This is useful in distinguishing PSP from PD, but when your patient has a poor levodopa response and downgaze problems, PD isn’t really the issue; PSP, MSA and CBD are.
One study of neurofilament heavy chain found that it can differentiate PSP from CBD but not from MSA. That study was published in 2006 and we’re still awaiting confirmation.
You’d think that tau would be the object of intense scrutiny in the differential diagnosis of PSP by CSF, but there’s been relatively little on that. One good study found that the ratio of phospho-tau to total tau is lower in PSP and MSA than in PD. The other studies of phospho-tau in PSP have been negative.
So the winner so far for PSP, limping across the finish line, seems to be neurofilament light chain. It’s not available commercially as far as I can tell; nor should it be, without further study.
Adding to this discouraging picture is the fact that most or all of the studies of CSF markers in PSP have sampled patients in a stage of PSP that allowed clinical diagnosis. By that time, the CSF picture may be more diagnostic than in the earlier stages, when a state marker would be most useful. In other words, the studies were retrospective rather than prospective.
For now, I’m putting my money on imaging.