Of all the bad-actor molecules linked to neurodegeneration, the Alzheimer’s protein amyloid beta (Aβ) receives most of the press. But in recent years, scientists have been forced to confront the fact that Aβ itself has little if any direct toxicity to brain cells. “Aβ alone doesn’t cause dementia; it needs the presence of tau,” says John Q. Trojanowski, M.D., Ph.D., a Dana Alliance for Brain Initiatives (DABI) member who co-directs the Center for Neurodegenerative Disease Research at the University of Pennsylvania.
The tau protein has been linked to Alzheimer’s since the 1980s. To some extent in ordinary aging, but much more extensively in Alzheimer’s, tau is somehow corrupted and forms twists of fibril-like aggregates—known as neurofibrillary tangles and threads (NFTs)—within neurons. Although scientists have long debated whether tau’s corruption is a cause or effect of the Alzheimer’s disease process, that debate is now all but over. In fact, tau’s corruption seems to be a driver of disease not only in Alzheimer’s but in more than half a dozen other tau-linked maladies besides, including frontotemporal lobe dementia, progressive supranuclear palsy, and the head-knock syndrome known as chronic traumatic encephalopathy. Scientists also have been finding tau pathology in the brains of people with Huntington’s disease, Parkinson’s disease, and other conditions not traditionally considered tau-linked. Virtually all the major neurodegenerative diseases, and many minor ones, are now coming to be seen as “tauopathies” in one way or another.
Normal tau and corrupted tau
Tau normally functions within a neuron’s axon, or output stalk, and works there mostly as a binder and a stiffener for cylindrical, railroad-track-like constructions called microtubules. A neuron uses microtubules to enable the speedy transport of proteins and other cargoes along its axon. Without that transport, which brings nutrients and energy and carries away waste, the axon and the rest of the neuron would deteriorate.
Microtubules form and dissolve all the time in active neurons. To support these rapid changes, tau’s microtubule-binding properties are constantly being fine-tuned by the addition or removal of small molecules called phosphor groups. Enzymes called kinases add the phosphor groups—at any of several dozen sites on the tau protein—at least partly to weaken tau’s grip and make a microtubule less rigid. Other enzymes called phosphatases do the reverse, de-phosphorylating tau to increase its microtubule-gripping ability.
Somehow in Alzheimer’s and other tauopathies the tau phosphorylation process goes to an extreme, and isn’t reversed. Tau proteins become “hyperphosphorylated”—tagged with phosphor groups at multiple sites. They detach completely from microtubules and drift out of the axon, into the main neuronal body (soma) and even into the mesh of rootlike tendrils (dendrites) through which a neuron receives its principal inputs from other neurons. Eventually these hyperphosphorylated tau proteins begin to stick together to make NFTs—long, insoluble aggregates which, under a microscope, look like snips of coarse hair. The eponymous Alois Alzheimer noted these strange inclusions in his 1907 report on what would later be seen as the first-ever Alzheimer’s case:
In the centre of an otherwise almost normal cell there stands out one or several fibrils due to their characteristic thickness and peculiar impregnability.
After more than a century, scientists are starting to understand what these strange inclusions represent.
Tau dysfunction kills
Since the early 1990s, autopsy studies have found that the spread of tau NFTs through memory-related brain areas tracks the progress of Alzheimer’s dementia—and does so better than the spread of Aβ plaques. About fifteen years ago scientists also linked  a familial form of the dementia syndrome known as frontotemporal dementia with parkinsonism to a set of tau mutations—whose effects turned out to be very similar to what is seen in Alzheimer’s. “Some of the mutations impair the binding of tau to microtubules, while others cause tau to aggregate more readily,” says Trojanowski.
Throughout the 2000s, scientists found more and more evidence that tau dysfunction kills and Aβ doesn’t—or rather that Aβ contributes to Alzheimer’s only indirectly, by causing tau dysfunction. The more conclusive findings have come only in the past few years. In 2011, for example, researchers in the Harvard Medical School laboratory of Dennis Selkoe, M.D. (also a Dana Alliance member) reported that small aggregates (“oligomers”) of Aβ, isolated from Alzheimer’s brains, triggered the hyperphosphorylation of tau as well as Alzheimer’s-like changes in neurons, including the loss of synapses, even at very low concentrations. This toxic effect was tau-dependent: no tau, no toxicity.
How do Aβ oligomers trigger tau hyperphosphorylation? Scientists have been busy teasing apart the process. When Aβ oligomers alight upon a memory-related neuron, they cause an excessive flow of calcium ions into the neuron—at least in part by overstimulating synaptic receptors, it is thought. In April, a team led by cell biologist Franck Polleux, Ph.D., at The Scripps Research Institute reported  that the Aβ-induced influx of calcium overactivates a chain of kinase enzymes, including one called AMPK that triggers the hyperphosphorylation of tau. “AMPK can phosphorylate the tau protein at a site that is important for the early effect of Aβ on synaptic maintenance,” says Polleux. Blocking AMPK entirely in Alzheimer’s mice, or even preventing it from making that one key phosphorylation of tau, prevents tau’s hyperphosphorylation and takes away Aβ oligomers’ ability to trigger the loss of synapses.
Too little where it should be
How does tau’s hyperphosphorylation harm a neuron? The most obvious possibility is that it keeps tau from doing its normal job of stabilizing microtubules. The affected microtubules become less stable, and ultimately are less able to bear essential cellular cargoes along neuronal axons. “Axonal transport can fail,” says Trojanowski, “and there are many consequences of axonal transport failure—the most serious of which are the loss of synapses, the degeneration of the axon, and eventually the loss of the neuron.”
One therapeutic approach to this has been the development of kinase inhibitors to reduce tau hyperphosphorylation. However, the relevant kinases tend to have multiple functions in cells, so that inhibiting them may cause unacceptable side effects. Trojanowski and his colleagues favor an alternative approach, microtubule-stabilization. Their results in Alzheimer’s mice models with one microtubule-stabilizing drug, Epithilone D, were successful enough to convince Bristol Myers Squibb, last year, to start a small clinical trial of the drug (now ongoing) in people with early-stage cognitive signs of Alzheimer’s.
Too much where it shouldn’t be
Tau phosphorylation and aggregation may also cause harm by increasing—inappropriately—the concentration of tau in other parts of the neuron. One of the earliest signs of damage to memory-related neurons, in mouse models of Alzheimer’s and in human patients, is the withering of tiny synapse-bearing structures on dendrites, known as dendritic spines. As noted above, Aβ oligomers are now thought to cause this loss of synapses and spines by corrupting tau. But how does tau, mainly resident in axons, end up harming dendrites? In 2010 the laboratory of Jürgen Götz, Ph.D., at the University of Sydney (now at the University of Queensland) reported that a small fraction of the tau population in a neuron normally resides within dendrites. These dendritic tau proteins serve, in effect, as amplifiers of the same synaptic receptors that are stimulated by Aβ oligomers. When ordinary tau in an axon becomes hyperphosphorylated, detaches from microtubules, and starts to drift out of the axon into the dendrites, it adds to the normally low supply of dendritic tau—exacerbating Aβ’s overstimulation of dendritic receptors. This in turn may exacerbate the ongoing hyperphosphorylation of axonal tau, in what could become a runaway process that leads to synapse loss and, later, the neuron’s demise.
Illustration courtesy of Nature Reviews Neuroscience from the paper: L. Ittner & J. Goetz, “Amyloid-β and tau–a toxic pas de deux in Alzheimer's disease,” 2011.
By blocking dendritic tau and thus damping this amplifier effect, Götz’s team was able to prevent cognitive impairments in Aβ-overproducing Alzheimer’s mice, and could even reverse those impairments after they had become evident. In contrast, Götz notes, when he and his team bred mice that overproduced hyperphosphorylated tau as well as Aβ, “we saw an accelerated pathology, in which all the mice died by the age of four months.”
Amplifying Aβ’s effect may be only one way in which mislocalized tau contributes to neurodegeneration. Tau is a relatively sticky molecule, and some researchers suspect that it serves as a binder and scaffold not only for microtubules but also for a variety of other neuronal structures. That strong tendency to grab other molecules could mean that tau wreaks widespread havoc whenever it drifts into the wrong places in a vulnerable neuron. “I suspect that there are multiple adverse effects of tau’s mislocalization,” Götz says.
Tau’s spreading dysfunction
Like the Aβ fibrils that make plaques in Alzheimer’s, the tau fibrils that form and clump together as NFTs in tauopathies belong to a category of aggregate known as amyloids. Almost every human neurodegenerative disease features amyloids—most prominently of Aβ and tau in Alzheimer’s, of the alpha synuclein protein in Parkinson’s disease, of the huntingtin protein in Huntington’s disease, and of the human prion protein in Creutzfeldt-Jakob Disease (CJD). A key feature of amyloids is that they tend to be self-propagating. An amyloid fibril, for example, can grow by grabbing individual proteins (monomers) in the vicinity and adding them to its two ends; eventually the lengthening fibril breaks to form two new fibrils, which continue to add monomers, and so on. By contrast, some oligomers—which are much smaller and can be ring-shaped or spherical—propagate their aggregated state in a more direct way: by inducing the monomers they contact to bend and join to form new, identical oligomers.
Prion protein aggregates are relatively hardy self-propagators, and appear to be the cause of CJD and other-species prion diseases such as “Mad Cow” disease. Tau aggregates are nowhere near as transmissible, but they can propagate within living brains, soaking up functional tau monomers wherever they go—as several recent studies have demonstrated. This likely accounts for the well-defined spread of NFTs in Alzheimer’s and other tauopathies. It may also be the process that ultimately defeats the brain’s ability to resist tau dysfunction, which would otherwise be more localized and easier to contain.
There is evidence that tau oligomers—those made from two or three tau monomers, especially—are more worrisome aggregation-inducers than the more obvious, NFT-forming tau fibrils. The mechanism that drives their self-propagation is so efficient that “all monomers can convert to oligomers very quickly,” says Rakez Kayed, Ph.D., an assistant professor of neurology at the University of Texas Medical Branch. By contrast, tau fibrils grow from monomers and oligomers much more slowly.
Kayed and others believe that tau oligomers, compared to tau fibrils, are also more harmful to neurons, via subtle toxic properties that may be common to all amyloid oligomers. “In tau-based mouse models, tau oligomers bind to neuronal membranes; they disrupt normal calcium levels; and they bind to synaptic receptors,” Kayed says. “But we think the most critical mechanism is that they impair axonal transport.”
At a conference last year, Kayed and his colleagues presented data on a preclinical study of a monoclonal antibody, TOMA, that targets tau oligomers. “In mice that develop an Alzheimer’s-like tau pathology and behavioral deficits, we were able to reverse the deficits by removing these tau oligomers,” Kayed says. The antibody treatment is oligomer-specific: it spares the larger, hyperphosphorylated tau fibrils. And yet it worked well in this test. That suggests that hyperphosphorylated tau fibrils are less important to the disease process, compared to the smaller, less phosphorylated oligomers—and may even be protective to the extent that they compete with oligomers. “It appears that once the tau oligomers are removed, individual tau proteins become more available to bind to microtubules, and axonal transport is restored,” Kayed says.
The expanding relevance of tau
Kayed hopes to get his anti-tau-oligomer monoclonal antibody into clinical trials within two or three years. Another tau aggregation inhibitor, based on methylene blue, a common laboratory stain and reagent, is already in advanced clinical trials for Alzheimer’s and other tauopathies, after showing promising results in early trials.
Outside the realm of Alzheimer’s, it remains unclear what triggers tau dysfunction, and why that dysfunction starts at different places within the brain. Some researchers hypothesize that different tauopathies are driven by distinct types of tau aggregates—which, like different strains of a virus, have different toxicities, rates of spread, and other features. In any case, if tau dysfunction is the common denominator of these conditions, then tau-targeted treatments could work against them generally. Trojanowski, for example, says that he hopes to see good results for Epithilone D not just against Alzheimer’s but against all tauopathies.
Kayed also is looking at the roles of tau oligomers in various known and emerging tau-linked diseases. “We’ve found, for example, that tau oligomers form rapidly in two animal models of traumatic brain injury,” he says. “Levels of tau oligomers also are elevated in the brains of people with Parkinson’s disease and a related condition, Dementia with Lewy Bodies (DLB) as well as in mouse models of those conditions.” This may not be a mere coincidence. Studies by Trojanowski’s lab among others have shown that aggregates of the Parkinson’s- and DLB-linked alpha synuclein protein can trigger or “seed” the aggregation of tau. “It may be that in Parkinson’s and DLB, alpha synuclein seeds the formation of the first tau aggregate, which goes on to seed further tau aggregation,” says Kayed. “In this way tau, being so abundant in the brain, could become the main self-propagating toxic entity.”
Published May 2013
 “Tau-mediated neurodegeneration in Alzheimer’s disease and related disorders,” by Carlo Ballatorre, Virginia M.-Y. Lee, and John Q. Trojanowski, Nature Reviews Neuroscience 2007 (vol 8 (9), Sept): 663-72.
 Alzheimer A. Über eine eigenartige Erkrankung der Hirnrinde. Allgemeine Zeitschrift für Psychiatrie und Psychisch-Gerichtliche Medizin 1907; 64: 146–48.
 “Staging of Alzheimer-related cortical destruction,” by H. Braak, E. Braak and J. Bohl, Eur. Neurol. 1993 (vol. 33(6)): 403-8.
 “Segregation of a missense mutation in the microtubule-associated protein tau gene with familial frontotemporal dementia and parkinsonism,” by C. Dumanchin et al, Hum. Mol. Genetics 1998 (vol. 7(11); Oct): 1825-9.
 “Soluble amyloid β-protein dimers isolated from Alzheimer cortex directly induce Tau hyperphosphorylation and neuritic degeneration,” by Ming Jin et al, PNAS 2011 (vol 108 (14), Apr 5): 5819-24.
 “Abeta oligomers cause localized Ca(2+) elevation, missorting of endogenous Tau into dendrites, Tau phosphorylation, and destruction of microtubules and spines,” by H. Zempel et al, J. Neuroscience 2010 (vol. 30 (36), Sep 8): 11938-50.
 “The CAMKK2-AMPK kinase pathway mediates the synaptotoxic effects of Aβ oligomers through Tau phosphorylation,” by G. Mairet-Coello et al, Neuron 2013 (vol 77 (7), April 10): 94-108.
 “Dendritic function of tau mediates amyloid-beta toxicity in Alzheimer's disease mouse models,” by L.M. Ittner et al, Cell 2010 (vol. 142 (3), Aug 6): 387-97.
 “Transmission and spreading of tauopathy in transgenic mouse brain,” by F. Clavaguera et al, Nature Cell Biology 2009 (vol 11): 909-13.
 “Alzheimer brain-derived tau oligomers propagate pathology from endogenous tau,” by C.A. Lasagna-Reeves et al, Sci Rep. 2012 (vol. 2): 700.
 “Selective inhibition of Alzheimer disease-like tau aggregation by phenothiazines,” by C.M. Wischik et al, PNAS 1996 (vol. 93 (20), Oct 1): 11213-8.
 “Conformational diversity of wild-type Tau fibrils specified by templated conformation change,” by B. Frost et al, J. Biol. Chem. 2009 (vol. 284(6), Feb 6): 3546-51.
 “Initiation and synergistic fibrillization of tau and alpha-synuclein,” by B.I. Giasson et al, Science 2003 (vol. 300 (5619), Apr 25): 636-40.