Since Tau2020 was their last in-person conference, many attendees were hoping Tau2022 would be their first one back in person. Alas, it was not to be. Even so, the small virtual meeting hit some high notes as scientists focused on five major themes over two days. They debuted a phospho-tau 217 “clock” that predicts AD progression, identified LRRK2 as essential for neurons to internalize tau monomers from extracellular fluid, fingered toxic tau species as triggers of damaging immune responses, and linked tau haplotypes to errant transcriptional regulation and oxidative stress.
P-Tau217 Clock Predicts Alzheimer’s Progression Over 30 Years
Researchers need better tools to predict when people with preclinical Alzheimer’s pathology are likely to develop symptoms. At the Tau2022 conference, held virtually February 22-23, Suzanne Schindler at Washington University in St. Louis put forward a p-tau217 “clock” as one option. She analyzed the longitudinal change in p-tau217 cerebrospinal fluid levels in older adults, and found that after p-tau217 passes a “tipping point,” it rises at a consistent rate in everyone for the next 30 years. The age at which people reached this tipping point predicted the age when they would develop symptoms, with about the same accuracy as familial age of onset in autosomal dominant AD.
Schindler believes this p-tau217 clock could help researchers select secondary prevention trial participants with sporadic AD who are close to developing symptoms, thus improving trial power and efficiency. “P-tau217 seems to be a very early marker that is useful for a long time,” she told Alzforum.
Jeffrey Dage at Indiana University School of Medicine, Indianapolis, called the findings encouraging, and agreed there could be clinical trial applications. “If the relationship between biomarker level and time to AD symptom onset is influenced by treatment predictably and reliably in broad, diverse populations, then these measures could potentially serve as a useful way to study the effects of amyloid-removing therapies in preclinical AD,” he wrote (full comment below).
Disease Takes Off. In people who do not yet have AD, longitudinal CSF p-tau217 measurements (vertical lines) fluctuate up and down from the baseline number (horizontal line). However, after p-tau217 reaches a “tipping point” of 3.7 percent of total 217 tau fragments, the values only climb. Dotted line indicates a positive PiB PET scan, and colors reflects CSF Aβ42/40 values (blue-purple are normal, yellow-red abnormal). [Courtesy of Suzanne Schindler.]
Several studies have found that amyloid accumulates at a predictable rate (Jack et al., 2013; Villemagne et al., 2013; Oct 2019 news). Previously, Schindler had developed an “amyloid clock,” devising a way to predict the time of symptom onset from a single PiB PET scan (Sep 2021 news). In her latest work, she wanted to expand this approach to fluid biomarkers.
To build a tau clock, Schindler turned to CSF data from 385 participants in longitudinal studies at the Knight Alzheimer's Disease Research Center that had been generated by Nicolas Barthélemy and Randall Bateman at WashU. Using mass spectrometry, Barthélemy previously determined that rising CSF p-tau217 and p-tau181 reflected the growth of amyloid plaques in the brain, while other markers such as p-tau205 rose later and correlated with other aspects of disease (Mar 2020 news). Others have found similar relationships by immunoassay of plasma and CSF (Aug 2019 conference news; Jul 2020 conference news).
Schindler analyzed these longitudinal data to find the threshold level of each p-tau marker that heralded disease onset. In people without amyloid plaques, the percent phosphorylation fluctuated over time with no set pattern. However, after a certain tipping point, each p-tau climbed steadily, indicating the person was now on the path to AD. This occurred at a different age in each person. P-tau217 best mirrored the changes in the Aβ42/40 ratio, indicating a close relationship with plaques.
Notably, while the Aβ ratio dropped linearly after its tipping point, p-tau217 shot up exponentially after it reached its threshold, which occurred at about 3.7 percent phosphorylation (see image above). This may reflect a biological difference between the markers, Schindler told Alzforum. Aβ drops because the peptide is being sequestered in the brain, and thus depleted from CSF. P-tau217, on the other hand, is produced by the brain’s response to amyloid plaques. Injury responses often display an exponential pattern of change, Schindler noted.
To turn these exponential data into a p-tau217 clock that keeps linear time, Schindler morphed the data with logarithms. Analyzing the WashU data by the log of the p-tau217 concentration, she found the clock maintained a steady rate of change over a period of 30 years, longer than the 18 years of the PiB PET amyloid clock (see animation below). Likely, this is because amyloid load plateaus around the time of symptom onset, while p-tau keeps rising.
From Chaos, Order. Aligning people's p-tau217 trajectories by their chronological age reveals no clear pattern, but anchoring them by the p-tau217 tipping point reveals a consistent rate of change for nearly everyone. [Courtesy of Suzanne Schindler.]
The beauty of this clock is that a person’s p-tau217 level at any time could be used to estimate where they are on this trajectory and, by extension, at which chronological age they reached the tipping point. This typically came about six years earlier than the PiB PET tipping point. This may be because brain-wide PiB PET positivity is not a very sensitive measure, Schindler noted. Regional plaques have already grown before the global scan turns positive. She believes the p-tau217 tipping point coincides with these initial regional amyloid deposits.
While the p-tau217 clock started earlier and ran longer than the PiB PET amyloid clock, it was not quite as accurate. In this dataset, 48 people progressed from cognitively healthy to AD dementia; in them, the p-tau217 tipping point predicted their age at symptom onset with a correlation of 0.66 and a variance of plus or minus five years. This is fuzzier than the PiB PET clock estimate, which varies by plus or minus three years. However, it is about the same as estimated year of onset (EYO) predictions based on mutation type or parental year of onset in autosomal-dominant AD.
Oskar Hansson at Lund University, Sweden, was intrigued by the evidence of a variable tipping point but a steady rate of change thereafter. “It implies to me that genetics and lifestyle factors mainly affect when the tipping point is reached in a certain individual, but once the aggregation process of Aβ has started in widespread cortical areas, the speed of continued aggregation is constant and not affected to a large degree by different risk or protective factors,” he wrote (full comment below).
In future work, Schindler will further analyze cognitive data to nail down the relationship between the p-tau217 tipping point, age, and symptom onset. She will also examine plasma p-tau217 to see if the relationship holds there, as a blood test would make a more clinically useful biomarker.—Madolyn Bowman Rogers
At Tau2022: Unknown Functions Emerge for Tau, LRRK2
The traditional view of tau is that of a rather dull microtubule-binding protein that occasionally goes rogue, wandering off into other cellular compartments where it stokes neurodegeneration. At Tau2022, a virtual meeting held February 22-23, scientists questioned this concept. Even in health, they said, tau leads a far more varied life. For one, it travels between neurons in a manner that commandeers the entire endolysosomal system. For another, cells require LRRK2 to internalize tau monomers, but not fibrils. Known for its association with Parkinson’s disease, LRRK2 was recently linked to primary tauopathies as well.
There’s more. In neurons from healthy human brain, tau was spotted within the nucleolus—the birthplace of ribosomes. It appeared right at home in this membraneless organelle, where it protects ribosomal DNA and keeps heterochromatin stable. Finally, tau appeared to play a role in ALS, as it shifted from the cytoplasm to synapses in people with this disease. There, tau cavorted with a protein called Drp1 to provoke mitochondrial fission.
Together, the findings suggest that tau plays a host of underappreciated and dynamic roles in health and in neurodegeneration, even in diseases not considered tauopathies.
Tau, A Frequent Traveler?
Mounting evidence suggests that in the context of neurodegenerative disease, aggregated forms of tau spread from cell to cell via templated misfolding. Scientists consider this propagation to be part of a pathological process, but Rick Livesey, who recently moved from University College London to Biogen, pointed out that tau has been spotted passing between healthy cells, as well. In previous studies, Livesey found that different forms of tau are actively secreted and taken up by iPSC-derived human excitatory neurons as part of a normal physiological process (Evans et al., 2018).
What cellular mechanism operates this shuffle? Lewis Evans in Livesey’s lab ran two genome-wide CRISPR screens in iPSC-derived neurons, asking which genes they needed to take up fluorescently labeled monomers and fibrils of tau. More than 200 popped out. Both screens identified genes throughout the entire endolysosomal trafficking system—everything from receptor-mediated endocytosis to endolysosomal fusion, autophagy, and even endoplasmic reticulum and Golgi trafficking.
Curiously, many of these genes also enable uptake of viruses, including influenza A, Zika, and SARS-CoV2. These three viruses start with receptor-mediated endocytosis and from there hijack the entire endosomal trafficking system to invade cells. A notable exception is Ebola, which uses a different mechanism—micropinocytosis—to barge into cells. In keeping with this distinction, Livesey reported virtually no overlap between genes involved in tau uptake and Ebola infection. Viruses have evolved to exploit the endolysosomal system to infect cells, but it is still unclear why tau uses the same machinery, or what process its cellular entry serves.
While the endolysosomal system seems necessary for both tau monomers and fibrils to get inside cells, Livesey identified differences in specific genes involved in the internalization of each. For example, LRP1 was required in the uptake of the former, but not the latter, confirming findings first reported at Tau2020 by researchers at Kenneth Kosik’s lab at the University of California, Santa Barbara (Mar 2020 conference news).
Secondly, Livesey reported that LRRK2—a multicatalytic protein infamous for its role in familial and sporadic Parkinson's disease—was essential for the entry of tau monomers, but not fibrils. This rang bells for Livesey, because Virginia Lee and colleagues at the University of Pennsylvania, Philadelphia, had found extensive AD-like tau tangle pathology in people with PD caused by LRRK2 mutations, and at Tau2020, Huw Morris and Edwin Jabbari at UCL had linked high LRRK2 expression to a genetic risk variant for the primary tauopathy progressive supranuclear palsy (Henderson et al., 2019; Mar 2020 conference news; Jabbari et al., 2021). A case-control study had also tied LRRK2 mutations to PSP and corticobasal degeneration (Sanchez-Contreras et al., 2017).
Using iPSC-derived neurons without LRRK2 to probe further, Livesey found that the cells not only decline to ingest tau monomers, they also accumulated LRP1 on their surface. This suggested a link between LRRK2 and LRP1 in tau monomer internalization, but Livesey does not know how LRRK2 mediates internalization of LRP1.
When Livesey blocked LRRK2 kinase activity with an inhibitor rather than by knocking it out, cells stopped taking up both monomeric and fibrillar tau, despite retaining a functionally intact endolysosomal system. Why the inhibitor has a different effect than does deleting the gene remains to be seen. Complicating matters even further, neurons expressing the PD-causing G2019S LRRK2 mutation also ingested less of both forms of tau. This variant boosts LRRK2 kinase activity and compromises endolysosomal function.
Complex findings are par for the course for LRRK2, a large protein that has multiple functions in different cell types. While these studies were limited to neurons, Livesey believes that LRRK2 might facilitate passage of tau between glia as well.
Li Gan of Weill Cornell Medical College in New York was intrigued by the overlap between genes mediating uptake of tau and viruses. Once inside cells, viruses escape the confines of the endolysosomal system, and Gan wondered if tau makes it into the neuronal cytosol the same way. Livesey has not investigated this, but said others have proposed ideas ranging from tau buddying up with specific proteins that whisk it across the endolysosomal membrane, to tau seeds punching holes to escape into the cytosol (Jan 2021 news).
Is cellular release and uptake of tau part of its life cycle? Is this hand-off somehow beneficial, and if so, under what circumstances might it become pathological? Researchers posed different versions of this question during discussion. Hui Zheng of Baylor College of Medicine in Houston said previous studies from several labs have found that secreted monomers have been truncated by lysosomal processing, rendering them incompetent for seeding aggregation (Mar 2018 news; Xu et al., 2020). This suggests that cellular uptake and release serves as a beneficial clearance mechanism, Zheng said, adding that distinct pathways may be involved in proteopathic propagation of tau aggregates.
Maria Grazia Spillantini of the University of Cambridge, U.K., took a therapeutic view of these tau shenanigans. “If you were developing a LRRK2-based treatment for tauopathies, would you prefer that tau is taken into the cell, or left out?” she asked Livesey. Livesey has been mulling the question. “What it comes down to is, are there truly toxic extracellular species of tau, and if so, what is their target? I’m not sure we have answers to either of those questions,” he said.
Tau: Card-Carrying Member of the Nucleolus?
Tau’s reach does not end in the lysosomes, cytosol, or even the synapse. Mahmoud Maina, University of Sussex, Brighton, U.K., reported that tau also helps organize the heart of the cell—the nucleus and nucleolus. The protein had been spotted in these nucleic-acid-rich regions 30 years ago, but what it does there is a mystery (Loomis et al., 1990; Brady et al., 1995). Recently, tau has been linked to nuclear dalliances, for example disruption of the spliceosome, deformation of the nucleus, and a dangerous liaison with RNA (Apr 2021 news; Jan 2019 news; Jul 2017 news).
At Tau2022, Maina argued that tau’s presence in so many cellular compartments means it could have diverse physiological functions. “To truly understand tauopathies and find effective therapies, we need a deep understanding of tau’s function in these different locations,” Maina said.
Many Haunts. Tau (red) is found in myriad locations of the cell, including the nucleus and nucleolus. [Courtesy of Mahmoud Maina, University of Sussex.]
A membraneless organelle within the nucleus, the nucleolus is where ribosomes are born. There, ribosomal DNA is transcribed into rRNA, which is then readied to join the massive ribonucleoprotein conglomerate that translates RNA into protein. What business might tau have in this protein-making hub of the cell?
Maina addressed this first by asking whether tau resided within the nucleus or nucleolus in neurons. Using transmission electron microscopy and immunolabeling, he spotted tau within both compartments in human brain samples, SHSY5Y neurons, and iPSC-derived cortical neurons. Within the nucleus, tau occupied dense regions of heterochromatin surrounding the nucleolus, and also co-localized with fibrillarin, a nucleolar protein (Maina et al., 2018). Notably, Maina reported that nucleolar tau was not phosphorylated, suggesting its presence there is not a consequence of pathological modification.
Tau in the Nucleolus. Non-phosphorylated tau (green) and the nucleolar protein fibrillarin (arrowhead, red) co-mingle (yellow) within the nucleolus of human iPSC-derived neurons. [Courtesy of Mahmoud Maina, University of Sussex.]
Maina got a hint that tau might be a regular in the nucleolus when he noticed that it co-localized with the nucleolar remodeling complex (NoRC), which is an essential component of the nucleolar machinery responsible for silencing rDNAs. These repetitive sequences within the genome loop into the nucleolus, where they are transcribed in a highly regulated manner that moves to the bioenergetic beat of the cell. Because their repeats make them prone to recombination, rDNAs are protected by being tightly packaged into heterochromatin by NoRC until they are needed. Knocking down either of NoRC’s two main components—Snf2h or TIP5—relaxes the heterochromatin, ramping up rDNA transcription and jeopardizing its stability.
Maina was surprised when he realized that knocking down tau, which directly associated with TIP5, had a similar effect. In SHSY5Y cells, loss of tau led the heterochromatin to relax, rDNA transcription to rise, and more nucleoli to form per cell. Could tau be an integral part of the nucleolus, not merely a passerby? In line with this idea, Maina had previously reported that tau behaved like a classical nucleolar protein in response to stress, that is, it rapidly redistributed to the nucleoplasm and cytoplasm when cells were rattled by Aβ oligomers or glutamate (Maina et al., 2018). Maina believes tau shows some of the hallmark features of a nucleolar protein, and noted that dysfunction of the nucleolus has been implicated in multiple neurodegenerative diseases.
Curious about the unphosphorylated nature of nucleolar tau, Zheng asked Maina whether phosphorylation might be a barrier for tau’s entry into the nucleolus. Maina considers tau phosphorylation a stress response. He thinks that under conditions such as exposure to Aβ oligomers or synaptic toxicity, phosphorylated tau enters the nucleus, but is excluded from the nucleolus. He plans to study the dynamic interplay between stress, tau phosphorylation, and nucleolar function.
Last but not least, Ghazaleh Sadri-Vakili of Massachusetts General Hospital in Boston presented recently published findings implicating tau in mitochondrial dysfunction in amyotrophic lateral sclerosis (Petrozziello et al., 2022). Studies reported hyperphosphorylated tau in the primary motor cortices of people with ALS, and altered ratios of p-tau to total tau in their cerebrospinal fluid. Since tau is known to mislocalize to synapses in Alzheimer’s disease, Sadri-Vakili wondered if it does the same in ALS. Tiziana Petrozziello at Mass. General and colleagues in her lab investigated postmortem motor cortex samples from 55 people with ALS and 26 controls. Six ALS cases had a known disease mutation, including five with a C9ORF72 and one with an SOD1 mutation. Sadri-Vakili reported that, regardless of sex or mutation, neurons from cases showed tau to have shifted from the cytosol into synaptosomes, specifically isoforms phosphorylated on serines 396 and 404.
To find out what these isoforms might be doing there, Petrozziello isolated synaptoneurosomes from ALS brain samples, and added them to primary neuronal cultures. The ALS synaptosomes triggered release of reactive oxygen species, suggesting that something within these synaptic compartments, possibly tau that got taken up by the neurons, stressed out mitochondria. Sadri-Vakili and colleagues found profound mitochondrial dysfunction in their postmortem ALS samples, such as low levels of electron transport chain proteins and fewer, smaller mitochondria in people with ALS than controls.
Does this dysfunction come from tau? Previous studies have reported that phospho-tau binds to, and promotes activity of, the mitochondrial fission protein, DRP1 (Manczak and Reddy, 2012). In further experiments, Sadri-Vakili found both p-tau396 and DRP1 to be enriched within ALS synaptoneurosomes. When added to neuroblastoma cells in culture, these p-tau- and DRP1-laden packages triggered fission, resulting in smaller mitochondria. Finally, Sadri-Vakili reported that silencing DRP1, or reducing p-tau with the ubiquitin ligase linker QC-01-175, prevented mitochondrial destruction and release of reactive oxygen species (Silva et al., 2019). QC-01-175 stimulated tau degradation by the proteasome.
While mechanistic questions remain, the findings hint at an important connection between hyperphosphorylated tau and mitochondrial dysfunction in ALS, Sadri-Vakili said. They also dovetail with a recent study by Gan, which identified synaptic and mitochondrial proteins as key constituents of the tau interactome (Jan 2022 news).—Jessica Shugart
Tau Triggers Neuroinflammation, But Mechanisms Vary by Disease
Inflammation flares up in the brain in all neurodegenerative diseases, but how so? At the Tau2022 conference, held virtually February 22-23, researchers elucidated some links between tauopathy and neuroinflammation. Jessica Rexach of the University of California, Los Angeles, compared primary tauopathies and Alzheimer’s disease, a secondary tauopathy, using both mouse and human data. She found strikingly different inflammatory profiles. In primary tauopathies, natural killer cells infiltrated the brain, whereas AD featured a microglial antiviral response orchestrated by interferon-γ. “Tauopathies invoke the immune response in different ways,” Rexach told Alzforum.
Lennart Mucke of the Gladstone Institute of Neurological Disease, San Francisco, came at the question of tau and neuroinflammation from a different angle. He tied activated microglia in AD brain to the electrical imbalance caused by excessive tau. Lowering tau dampened excitation and suppressed many inflammation-related AD risk genes. “Epileptiform activity drives aberrant microglial responses,” Mucke concluded.
Some previous studies have blamed activated microglia for sparking tau pathology, but few have looked in the other direction, i.e., at how abnormal tau itself kicks off inflammation (Feb 2007 news; Oct 2019 news; Nov 2019 news).
Rexach earlier examined this in rTg4510 and TPR50 mice, which express human mutant P301L tau (Onishi et al., 2014). She purified microglia at different points along the disease course and analyzed their gene expression to identify co-expression modules that characterized each stage. These modules reflect the biological response of microglia to tauopathy as pathology worsens, Rexach told Alzforum. She then identified the same co-expression modules in published human transcriptomic data from AD brain, as well as in two primary tauopathies, progressive supranuclear palsy (PSP) and frontotemporal dementia (FTD). This suggested the modules were relevant to human disease.
Notably, many risk genes for PSP and FTD fell into a single module characterized by the suppression of microglial viral defense genes. This module activates at early preclinical stages of tauopathy and includes genes like Trim21, which has been associated with the prion-like spread of proteins and has been linked to TDP-43 forms of FTD (Dec 2018 conference news). AD genes, on the other hand, popped up in a module that cranks up viral defense genes late in tauopathy, led by interferon-γ (Rexach et al., 2020). “We need to dig further into these parallels between tauopathy and viral response,” Rexach told Alzforum.
At Tau2022, Rexach added data on noncoding GWAS hits. She used chromosome conformation capture to predict what genes these noncoding risk variants affected. Again, the diseases diverged: PSP heritability affected genes involved in glia-lymphocyte interactions, particularly protection from natural killer cells, while AD heritability affected complement, cytokine, and myelin genes. Other studies have linked complement proteins to synapse loss in AD (Aug 2013 news; Dec 2014 news; Apr 2016 news).
Did brain tissue support this genetic dichotomy? In collaboration with Daniel Geschwind at UCLA, William Seeley at the University of California, San Francisco, and Dheeraj Malhotra at Roche, Rexach examined postmortem tissue from AD, PSP, and Pick’s disease brains. Pick’s disease is a subtype of FTD characterized by three-repeat deposits of tau, making it distinct from PSP, which has four-repeat tau. Tissue samples came from 40 brains and seven brain regions, comprising the angular gyrus, dentate of the cerebellum, insula, entorhinal cortex, globus pallidus, calcarine cortex, and precentral gyrus. These regions were selected because they each selectively degenerate in one or two of the three tauopathies, but not in the others. The researchers analyzed samples by both bulk and single-cell RNA-Seq.
The data dovetailed with Rexach’s previous findings. In PSP and Pick’s brain, single-cell data revealed the presence of infiltrating natural killer cells in the insula. These lymphocytes are cytotoxic, selectively attacking sick or damaged cells. Notably, in all regions of PSP brain, genes that protect against NK cells were suppressed. Meanwhile, in every region of Pick’s brain, the MICB gene that activates NK cells was massively expressed. Together the data suggest that in both PSP and Pick’s, brain tissue may become more vulnerable to cytotoxic damage from invading lymphocytes.
In addition, microglia in Pick’s brain had a distinct gene-expression profile driven by the transcription factor IKZF1. This gene is also active in lymphocytes and is pro-inflammatory, suggesting it could contribute to microgliosis in Pick’s.
In AD brain, by contrast, microglia displayed a familiar inflammatory profile driven by the SPI1 gene that encodes the master regulator protein PU.1. PU.1 controls AD risk genes and affects Alzheimer’s onset (Jun 2017 news).
AD microglia also expressed many genes found in mouse disease-associated microglia (DAM). Evidence for DAM in human brain has been mixed, with some studies not finding them and others reporting DAM genes scattered across several microglial subtypes (May 2019 news; Dec 2020 news). Intriguingly, Rexach found the strongest DAM signature in microglia around plaques in brain regions without much degeneration. This fits with a model in which DAM genes help microglia contain amyloid and ameliorate tissue damage (Jun 2017 news). In brain regions where the disease has advanced, microglia may lose this ability, Rexach suggested. She plans to follow up on these genetic data with functional studies in cellular and animal models.
While this study characterized the immune responses to tauopathy, it remains unclear how tau might trigger inflammation. Mucke provided some clues by focusing first on network activity. He has long been interested in links between tau and epileptic activity, and recently reported that deleting the protein in mice affects excitatory and inhibitory activity differently, with the former firing less and the latter firing more (Oct 2021 news). But which comes first?
At Tau2022, Mucke demonstrated that tau exerts its effects in excitatory cells. Conditionally deleting the protein only in excitatory neurons prevented epileptic activity after exposure to a stimulant, while deleting tau only in inhibitory neurons did not. The same thing happened when Mucke crossed these conditional knockouts with mice that model Dravet syndrome, which combines epilepsy and autism-like behaviors. Deleting tau in excitatory neurons ameliorated stereotyped behaviors and improved mouse survival to nearly wild-type levels, while deleting it in inhibitory neurons had no effect.
How might tau’s effect on excitation dovetail with autism-like behavior? Mucke and others previously reported that tau inhibits PTEN. This phosphatase suppresses a PI3 kinase pathway that promotes autism-like behavior. Network hyperactivity triggers this PI3K pathway, but when tau is absent, PTEN is able to intervene to dampen it. Excess tau, however, prevents PTEN from doing its job and leaves the brain vulnerable to hyperexcitability (Marciniak et al., 2017; Tai et al., 2020).
At Tau2022, Mucke reported massive activation of the PI3K pathway in the Dravet syndrome mice; dampening tau suppressed it. It is unclear how this mechanism relates to the selective vulnerability of excitatory neurons to tau. Mucke plans to dissect the molecular mechanisms to find out if PTEN or other pathway components are preferentially expressed in excitatory neurons.
And where does inflammation come in? Lowering tau in an amyloidosis mouse model also drops expression of many inflammation-related AD genes such as C1q, Tyrobp, and TREM2, suggesting a direct link, Mucke noted (Das et al., 2021).
“The way we put this together is that network dysfunction exists in a vicious cycle with immune dysfunction,” Mucke said. This could play out in various ways in a diseased brain. In some cases, microinfarcts might bring on local overexcitation that triggers an immune response; in others, a TREM2 variant might promote immune dysfunction that leads to excitation. “So this circle keeps churning and leads to synaptic dysfunction and loss,” he said.
Mucke believes both anti-epileptic therapy and tau reduction have the potential to break this cycle. In Dravet mice, lowering tau with antisense oligonucleotides improved survival to nearly wild-type levels.—Madolyn Bowman Rogers
Tau Haplotypes Hint at Transcriptional Changes, Ferroptosis
Tau aggregation lies at the heart of many neurodegenerative diseases, but why this protein goes rogue remains mysterious. At the Tau2022 conference, held virtually February 22-23, speakers homed in on the role of the tau haplotype in neurodegeneration. The larger region of the genome that includes the tau gene, MAPT, comes in two forms, H1 and H2, with the former associated with neurodegenerative disease risk, and the latter protective. Because the region contains around a dozen genes, the causal factors are unknown. Kathryn Bowles at the Icahn School of Medicine, Mount Sinai, New York, described a multicenter effort to dissect the source of this risk. Although the project is in its infancy, it has already turned up previously unknown differences between the haplotypes.
Meanwhile, Peter Heutink at the German Center for Neurodegenerative Diseases in Tübingen, suggested that one culprit in H1 might be weaker binding of a master regulator of oxidative stress, which leaves cells more vulnerable to ferroptosis, a type of cell death caused by iron toxicity.
H2 is an evolutionarily ancient 1.5 MB inversion of H1 that contains duplicated genes. Because of the inversion, H1 and H2 are unable to recombine with each other, and so these haplotypes are inherited separately as blocks. Researchers have long known that H1 associates with several primary and secondary tauopathies, including progressive supranuclear palsy, corticobasal degeneration, Parkinson’s disease, and APOE4-negative Alzheimer’s disease (Sep 2005 news; Mar 2019 news; Strickland et al., 2020).
Mirror Images. The H2 haplotype (bottom) is a 1.5 MB inversion of H1 (top) and also contains some duplicated genes. This large region, which contains the MAPT gene, is inherited as a block. [Courtesy of Bowles et al., 2019.]
At Tau2022, Bowles described a large consortium, called the “Tau Centers Without Walls.” It is trying to track down the source of this risk by making iPSC lines from people with each haplotype and performing omics analyses. Bowles, working in the lab of Alison Goate at Mount Sinai, has made lines from people with European ancestry. The H2 inversion is most common in this population, occurring in 10 to 36 percent of people, depending on the exact group studied. It is less common in African populations and nearly nonexistent in South Asians. In fact, some have suggested this inversion is a legacy from Neanderthal ancestors (Hardy et al., 2005).
In the European lines, Bowles identified previously unreported differences in the noncoding, antisense versions of some genes. These may serve a regulatory function, she noted. Expression of the noncoding variants varied by brain cell type as well as haplotype. For example, MAPT antisense was more highly expressed in H1 than H2 neurons, but no differently expressed in other brain cell types. On the other hand, an antisense version of the chromatin-modifying gene KANSL1 was up in all three H2 cell types examined: neurons, astrocytes, and microglia.
Bowles also identified numerous splicing differences between genes in H1 and H2. The biggest was for KANSL1, where H2 cells tended to retain an intron. These findings have to be confirmed in human brain, as the immature neurons generated from iPSCs can splice genes differently than do mature neurons.
Overall, H2 neurons and astrocytes suppressed genes involved in protein translation compared to H1 cells, whereas H2 microglia boosted them. “H2 is associated with differences in splicing that affect protein localization and trafficking,” Bowles concluded. She expects disease risk will ultimately involve far more genes than just MAPT. “I believe there are effects in multiple genes in both neurons and glia that all interact to contribute to influencing disease risk,” Bowles wrote to Alzforum.
Heutink focused on one particular aspect of risk, i.e., the susceptibility of H1 cells to oxidative stress. He generated neural precursor cells from H1 and H2 iPSCs, then allowed them to mature into neurons in culture. When antioxidants were omitted from the culture medium, H1 neurons developed swollen axons and died, while H2 neurons stayed healthier. Heutink believes this is due to the transcription factor NRF2, a master regulator of the oxidative stress response. When a cell experiences oxidative stress, NRF2 enters the nucleus and binds an antioxidant response element (ARE) upstream of MAPT. Notably, this ARE is mutated in the H1 haplotype, resulting in weaker binding of NRF2 (Wang et al., 2016).
Heutink screened a library of 1,430 chemical compounds for any that could improve the oxidative stress response. He found 87 compounds that made H1 neurons more viable; about half generated consistent dose-response curves, suggesting a robust effect. Reviewing the literature on these compounds, Heutink found a number of them were involved in ferroptosis, a type of cell death that results from iron accumulation and oxidative stress (for review, see Li et al., 2020). These included curcumin, ethinyl estradiol, deferasirox, idebenone, lapatinib, zileuton, and Vitamin E.
Did the H1 neurons die by ferroptosis? Indeed, staining techniques confirmed that dying H1 neurons displayed features of ferroptosis, such as axon swelling, mitochondrial damage, and the presence of lipid peroxidase. In addition, death swept through the cultures in a wave, which is characteristic of ferroptosis. In contrast, apoptosis hits cells in a random pattern. Also, compounds that suppressed apoptosis did not prevent neuron death.
Heutink is now examining downstream effects of these compounds on specific genes in the H1 haplotype, including MAPT and KANSL1, to determine the mechanisms involved in the increased vulnerability to ferroptosis. In answer to an audience question, he said he plans to check for iron accumulation in the dying cells as well.
Bowles noted that several different MAPT mutation lines, all of which have the H1 haplotype, also show increased susceptibility to oxidative stress compared with wild-type H1 tau. “This could demonstrate a similar risk-associated mechanism in both sporadic and familial FTD and PSP,” she told Alzforum.—Madolyn Bowman Rogers
Bowles KR, Pugh DA, Farrell K, Han N, TCW J, Liu Y, Liang SA, Qian L, Bendl J, Fullard JF, Renton AE, Casella A, Iida MA, Bandres-Ciga S, Gan-Or Z, Heutink P, Siitonen A, Bertelsen S, Karch CM, Frucht SJ, Kopell BH, Peter I, Park YJ, Crane PK, Kauwe JSK, Boehme KL, Höglinger GU, PART working group, International Parkinson’s Disease Genomics Consortium, Progressive Supranuclear Palsy Genetics Consortium, Charney A, Roussos P, Wang JC, Poon WW, Raj T, Crary JF, Goate AM.
17q21.31 sub-haplotypes underlying H1-associated risk for Parkinson’s disease are associated with LRRC37A/2 expression in astrocytes.
bioRxiv. November 30, 2019
bioRxiv
Comments
No Available Comments
Make a Comment
To make a comment you must login or register.