29 March 2013. Neurofibrillary tangles of tau protein litter Alzheimer’s disease brains, but are they actually bad for neurons? Many researchers now think that soluble, prefibrillar forms of tau are more toxic, and some wonder if tangles could even be neuroprotective. At the 11th International Conference on Alzheimer’s and Parkinson’s Diseases, held 6-10 March 2013 in Florence, Italy, scientists added evidence for this view by showing that neurons loaded with tangles continue to function in a network circuit. Other speakers described similar data exonerating insoluble tau. The acquittal for tangles does not mean smaller aggregates are off the hook, however, as researchers pointed a finger at tau’s ability to aggregate as its key toxic feature. Others aimed to settle a revived controversy over where tau pathology begins in the brain, and whether some tau deposition is a normal feature of aging.

Insoluble Tau Not the Villain?
Numerous studies have shown that lowering soluble tau can improve memory in animal models even in the presence of neurofibrillary tangles (NFTs) (see, e.g., ARF related news story; Oddo et al., 2006; Berger et al., 2007). Do tangles themselves harm cells? To get at this question, Kishore Kuchibhotla at New York University School of Medicine examined how tangle-bearing neurons function in neural networks of live mice. In Florence, Kuchibhotla presented his recent work, conducted in the laboratories of Brad Hyman and Brian Bacskai at Massachusetts General Hospital, Charlestown, in collaboration with Susanne Wegmann and Tara Spires. He used Tg4510 mice, which express mutant human P301L tau, develop NFTs throughout the cortex, and eventually lose neurons. To visualize neuronal activity, he injected virally encoded Yellow Cameleon, a fluorescent calcium indicator, into layers II/III of visual cortex. He then placed awake animals under a multiphoton microscope and showed them pictures, while observing neuronal responses through a transparent cranial window.

Tangle-bearing neurons in visual cortex maintain their ability to respond to specific line patterns. Shown here are control neurons. Image courtesy of Kichore Kuchibhotla

Kuchibhotla’s preliminary findings suggest that, at a systems level, even neurons jammed with tangles still function. Visual cortical neurons in Tg4510 mice had similar receptive field properties to those in control mice. For example, when the animals looked at striped lines, their neurons fired in response to specific angles, a property called orientation tuning. Kuchibhotla then sacrificed the mice and stained for NFTs to confirm that these functioning neurons contained tangles.

“At a minimum, we can say that tangle-bearing neurons are integrated into the local network,” Kuchibhotla told Alzforum. He is now investigating other neuronal properties, such as the magnitude of the response and how well coordinated neuronal firing is, to see if the cells have more subtle deficits. In future work, Kuchibhotla and colleagues plan to stain for soluble forms of tau in the neurons as well. Then they will turn off tau production in the Tg4510 mice, allowing them to compare cells that contain only tangles, only soluble tau, both, or neither. Ultimately, Kuchibhotla hopes to dissect the contribution of soluble and insoluble forms of tau to cell dysfunction, he told Alzforum.

Catherine Cowan, working with Amritpal Mudher at the University of Southampton, U.K., went a step further in suggesting that insoluble tau could be neuroprotective. In Florence, Cowan described data obtained from a Drosophila tauopathy model (see Cowan et al., 2010). The flies express highly phosphorylated wild-type (0N3R) human tau. They have problems moving, but do not get tangles or neuronal death. This shows that soluble tau alone can cause neuronal dysfunction, Cowan said. When she inhibited kinase GSK-3β in these animals, tau phosphorylation dropped and the flies recovered normal movement. Intriguingly, the treatment also produced insoluble, granular tau aggregates. These electron-dense granules are globular, 20-50 nm in diameter, and contain about 40 units of tau each. Similar granular aggregates have been seen in AD brains at Braak stage 1, and are believed to be a precursor to fibrillar deposits, Cowan noted (see Maeda et al., 2006; Maeda et al., 2007). Because the appearance of the aggregates coincides with behavioral improvement, it is possible that these deposits help sequester abnormal, toxic tau and protect cells, Cowan speculated. This implies that treatments that dissolve tau deposits could prove counterproductive.

Soluble Aggregates Could Be Guilty
If not insoluble tau deposits, then what about small soluble aggregates as the toxic forms? Eckhard and Eva-Maria Mandelkow at the German Center for Neurodegenerative Diseases, Bonn, previously developed a strain of mice that expresses a mutant form of tau particularly prone to clump up, and another strain with tau that cannot aggregate. Pro-aggregant mice have learning and memory defects that vanish when tau production is shut off, even though tangles persist. Anti-aggregant mice are fine (see ARF related news story). Would an inhibitor of tau aggregation, such as methylene blue, similarly protect pro-aggregant mice? A derivative of this chemical, dubbed Rember®, is currently in clinical trials for AD and frontotemporal dementia (see ARF related news story; ARF news story). Recent studies have suggested methylene blue works by oxidizing tau, and thus prevents aggregation (see ARF related news story).

In Florence, Eva-Maria Mandelkow reported that 15-month-old pro-aggregant mice that received methylene blue for one to three months showed no cognitive improvement. However, when given the drug from birth, the adult mice performed as well as wild-type animals. The therapy works as a preventative, but not a treatment, Mandelkow concluded. She then looked to see how late in life therapy could be started and still rescue cognition. She found that if she treated pro-aggregant mice with methylene blue for six months, starting at eight months of age and before cognition falters, their memory stayed sharp. Treated animals had fewer prefibrillar tau aggregates and less hyperphosphorylated tau, but more soluble tau. They also had more markers of autophagy, a cellular process that helps dispose of unwanted protein. With this drug, treatment needs to begin before cognitive decline kicks in, she said, adding, “We have to find that point in patients.” Biomarker studies in people who have inherited a familial AD gene show that tau levels rise a few years before clinical symptoms (see ARF related news story).

In contrast to the pro-aggregant protein, excessive amounts of non-aggregating tau seem to have few ill effects. In Florence, Maria Joseph in the Mandelkow's group highlighted the differences in the two mouse strains. Using hippocampal slice cultures from neonatal animals, she saw that slices from pro-aggregant mice had massive pathology and mislocalized tau (see Messing et al., 2013), while those from anti-aggregant mice looked healthy, without hyperphosphorylated tau or microgliosis, and still with all their dendritic spines. Intriguingly, these anti-aggregant slices had an abundance of neurons, more than in wild-type mice. They contained more proliferating cells and more cells that stained with markers of neurogenesis compared to wild-type, suggesting more neurons are being born. Likewise, live anti-aggregant mice sported enlarged hippocampi. Mandelkow told Alzforum she plans to investigate the mechanisms behind this finding.

Where Does Tau Pathology Start?
Current ideas about AD progression propose that amyloid pathology initiates the disease and kicks off tau tangles (see, e.g., ARF Webinar). However, a recent study by Heiko Braak at Goethe University, Frankfurt, Germany, cast doubt on this sequence by reviving an old debate about whether tau pathology comes earlier (see Braak et al., 2011). While classic Braak staging shows AD pathology beginning in the entorhinal cortex, the new work found pre-tangle tau deposits in the brainstem of the majority of people under 30. This has led some researchers to suggest that the disease starts very early in life, with tau pathology in the brainstem. To investigate this issue, neuropathologist Kurt Jellinger at Vienna University School of Medicine, Austria, examined the brains of 239 unselected autopsy cases from ages 55 to 102. Almost half had AD, one-third were healthy control brains, 16 percent had AD with Lewy body pathology, 5 percent had Parkinson’s disease, and 2 percent, dementia with Lewy bodies.

In Florence, Jellinger reported that he did see a subtle tau pathology in subcortical nuclei, such as the locus coeruleus, substantia nigra, and olfactory bulb, in brains at Braak stage 0 or 1. However, only about half of people at stage 1 had it. The prevalence gradually rose with increasing Braak stages to reach 100 percent at stage 6. The amount of pathology also correlated with stage, with brains at stage 5 or 6 displaying massive tau tangles in subcortical areas. “This suggests that these regions become increasingly involved during AD progression, rather than representing sites initially affected by AD-associated tau pathology,” Jellinger said (see Attems et al., 2012). This idea is not new. For example, John Morris at Washington University, St. Louis, Missouri, has long held that some modest tau pathology is a normal part of aging (see ARF interview). It is present in almost all older brains regardless of cognitive status. Once amyloid pathology starts, these incipient tangles accelerate and spread, leading to the extensive pathology seen in AD, Morris suggested.

Tackling Tauopathy by Reactivating Plasticity
As a side note, Jellinger’s studies also revealed that neurons surrounded by perineuronal nets containing aggregan seem to be protected against tau pathology. Perineuronal nets consist of negatively charged, sugary proteins called chondroitin sulfate proteoglycans (CSPGs), of which aggrecan is one specific type. The nets form a cage around neurons and help to stabilize synapses and preserve memories. This extracellular matrix develops as critical periods end, preventing further plasticity (for a review, see Wang and Fawcett, 2012; McRae and Porter, 2012). Chewing up the perineuronal net with enzymes has been shown to reactivate plasticity in the adult visual cortex (see Pizzorusso et al., 2002).

Maria Spillantini at the University of Cambridge, U.K., wondered if enhancing plasticity in this way might ameliorate some of the memory defects caused by tauopathies. In Florence, she described work done in collaboration with Michel Goedert, also at Cambridge, and Patrick Aebischer at the École Polytechnique Fédérale de Lausanne, Switzerland. She used a strain of transgenic mice that express P301S mutant human tau and start forming neurofibrillary tangles at five weeks of age, increasing up to four to five months (see Allen et al., 2002). These mice have trouble remembering objects they have seen before, and do worse with age. Spillantini injected chondroitinase ABC, which breaks down the perineuronal net, into six sites in the perirhinal cortex of impaired mice. This brain region is crucial for visual perception and memory. She found that the treatment restored object recognition memory. By contrast, an enzyme that does not digest the CSPGs did nothing for memory. The effect of the treatment was temporary, however. Over the next five weeks the perineuronal net returned and the mice once again became impaired. This serves as a proof of principle that boosting neuronal plasticity might provide a temporary therapy for tauopathy, Spillantini said. Safer methods would have to be found to do it, as this enzymatic approach is not suitable for use in people.––Madolyn Bowman Rogers, with reporting by Gabrielle Strobel.