Zempel H, Luedtke J, Kumar Y, Biernat J, Dawson H, Mandelkow E, Mandelkow EM.
Amyloid-β oligomers induce synaptic damage via Tau-dependent microtubule severing by TTLL6 and spastin.
EMBO J. 2013 Nov 13;32(22):2920-37.
Please login to recommend the paper.
To make a comment you must login or register.
This is a beautiful and important piece of work that sheds light on a couple of important topics: first, how tau contributes to Aβ toxicity, and second, how tau becomes mislocalized into dendrites and what effects it has once there. These are some of the most important unanswered questions regarding tau, and this manuscript gives us several new insights and new molecular players to investigate.
One interesting aspect is the suggestion that tau may be involved in more than one aspect of the cascade induced by Aβ. On one hand, tau appears to be involved in transporting TTLL6 (tubulin tyrosine ligase-like 6) into the dendrite, triggering a series of events leading to microtubule loss mediated by spastin. But it also seems to regulate the earlier step of calcium entry, which is decreased in tau knockout neurons.
Another important aspect of this work is the at least preliminary validation of the findings in vivo. There have tended to be some differences between tau knockout neurons in culture and in tau knockout mice. For example, tau knockout neurons have reduced neurite outgrowth but tau-deficient neurons in vivo have normal dendritic trees, suggesting different roles for tau in culture and in vivo. It will be important to further study these mechanisms in animal models and human tissue.
Finally, the work may help resolve an apparent paradox between the seemingly conflicting observations that tau reduction and microtubule stabilizers (which can be perceived as increasing one tau function) are both beneficial in preventing Aβ effects; this paper recapitulates the two observations in the same system. At least in this system, the answer seems to be that microtubule stabilization makes neurons more resistant to Aβ, but the loss of microtubule stability is not due to loss of tau function. Rather, it is due to the presence of tau recruiting TTLL6, and thus spastin to the microtubule—a process that is prevented in the absence of tau.
Using state-of-the-art technology, the Mandelkow team presents a very detailed analysis of how Aβ oligomers damage synapses. One of their most interesting findings is that the tau that accumulates in the somatodendritic domain of primary neurons exposed to Aβ is not of axonal origin, but de novo synthesized. One unanswered question is how this synthesis occurs.
In addition, I am surprised to find that neurons recover after a six hour-exposure to Aβ in these in vitro experiments. In an in vivo situation there is most likely a more constant exposure to Aβ with no time for recovery.
Zempel and colleagues propose a flowchart of the toxic events in AD where Aβ-induced spine loss is the first pathological event. In contrast, we and others found that an initial over-excitation of neurons caused tau to target the kinase Fyn to the spine and thereby mediate Aβ toxicity. How can these findings be integrated? I believe that we must consider the chronology of AD. In this respect, a recent review by Jannic Boehm Boehm (Boehm 2013) is very interesting as it proposes two steps, one involving activation of Fyn, and a second involving the phosphatase STEP that then inactivates Fyn. Activating Fyn would lead to over-activation of NMDA receptors at synapses, increasing the potential for glutamate excitotoxicity, while activating STEP would have the opposite effect. Boehm concludes that "Given the staggered activation of Fyn and STEP, and their aforementioned effects on NMDA receptor activation and synaptic transmission, one can speculate about the long-term effects this process has during AD development. The transient activation of Fyn and the NMDA receptor activity increase will, in the long run, contribute to the known effects of excitotoxicity seen in AD. The chronic overexpression and activation of STEP, on the other hand, will lead to a permanent decrease in synaptic transmission and number of synapses." The latter would be in keeping with Zempel and colleagues' findings.
A 'danse macabre': tau and Fyn in STEP with amyloid beta to facilitate induction of synaptic depression and excitotoxicity.
Eur J Neurosci. 2013 Jun;37(12):1925-30.
This study is a nice extension of Zempel et al., 2010, in which the authors demonstrated that treating primary neuron cultures with oligomeric Aβ resulted in a mislocalization of tau into the dendritic compartment, as seen in Alzheimer’s disease. The earlier report also showed that exposure to oligomeric Aβ led to elevation of Ca++, disruption of microtubules (MTs), and missorting of mitochondria in dendrites, as well as loss of dendritic spines. Importantly, this prior work revealed that tau localization, mitochondria sorting, and spine density could be normalized by treatment with the MT-stabilizing agent paclitaxel, suggesting that MT disruption after Aβ treatment was a key event linked to many of these other dendritic changes.
In this new paper, Zempel et al. have extended their prior studies and gained greater understanding of the neuronal changes induced by oligomeric Aβ. By showing that treatment with the MT-destabilizing agent nocodazole also led to an elevation of dendritic tau, they confirmed the importance of MT changes in the observed Aβ-induced tau mislocalization into dendrites.
Moreover, the authors have more closely examined the temporal sequence of dendritic changes that occur after Aβ oligomers are added. They show that increases in intracellular Ca++, destabilization of MTs, and a loss of dendritic spines precede the peak of tau mislocalization. Their data suggest that the Aβ-triggered increase in intracellular Ca++ may be a key initiator in this chain of events, as chelation of extra- or intracellular Ca++ prevented the loss of MTs and tau missorting, albeit with continued dendritic spine loss.
Notably, the authors now show that the MT loss that occurs after Aβ addition likely results from cleavage by the MT-destabilizing protein spastin. Spastin recruitment to MTs is facilitated by post-translational polyglutamylation of MTs, and both the extent of MT polyglutamylation and the amount of MT-bound spastin appeared to increase after Aβ treatment. Similarly, the enzyme TTLL6 (tubulin tyrosine ligase-like 6), which can polyglutamylate MTs, was also localized to dendritic MTs, suggesting that TTLL6 was critical to the process of MT cleavage after Aβ treatment. Importantly, knockdown of spastin prevented Aβ-induced loss of MTs.
Interestingly, the tau that invades dendrites after Aβ oligomer treatment does not seem to originate from the axonal pool, but rather is newly synthesized. The authors demonstrated this by showing that Dendra2-tagged tau did not migrate to dendrites after Aβ addition, and that the appearance of dendritic tau could be prevented by treatment with tau shRNA.
Another notable finding was that most of the Aβ-induced dendritic changes, including MT loss, were not observed in neurons from tau knockout mice. Moreover, reintroduction of tau into the knockout neurons sensitized them to Aβ treatment, resulting in recruitment of TTLL6 and spastin, with MT cleavage. Thus, tau seems to be necessary for the Aβ-mediated effects, including the alteration of MTs.
It is not entirely clear from these studies how tau contributes to the Aβ-mediated effects, but these observations are consistent with work from other laboratories (Roberson et al., 2007; Ittner et al., 2010) that has implicated tau as a necessary component for Aβ-induced toxicities. In this regard, Zempel et al. could provide new insights into the consequences of oligomeric Aβ on neurons.
Of particular note is the observation that dendritic MT alterations seem to be a critical consequence of Aβ exposure, and that adding paclitaxel to Aβ-treated neurons can prevent various dendritic changes. There is evidence of reduced MT density in brains with Alzheimer’s (Cash et al., 2003; Hempen and Brion, 1996), and this has also been demonstrated in tau transgenic mouse models (Ishihara et al., 1999; Brunden et al., 2010). Although the changes in the transgenic mouse models of tauopathy suggest alterations of axonal MTs, the work of Zempel et al. indicates that Aβ-induced MT deficiencies in dendrites may also be of importance in Alzheimer’s. This provides further rationale for testing brain-penetrant MT-stabilizing agents, such as epothilone D, which has been demonstrated to be efficacious in multiple tau transgenic mouse models (Brunden et al., 2010; Zhang et al., 2012; Barten et al., 2012). Epothilone D is under investigation in a Phase 1b study in Alzheimer’s patients (BMS-241027; clinicaltrials.gov), but it would seem important to develop additional MT-stabilizing candidates appropriate for testing.
This study by Zempel et al. reveals a potential mechanism by which Aβ oligomers exert tau-mediated toxic effects on microtubules. In this study, Aβ oligomers (ADDLs) were prepared from a mixture of Aβ1-40 and Aβ1-42 at a ratio of 7:3 and were used at 1 µM. Although the in-vitro findings are solid, a healthy dose of caution is warranted in extrapolating the results to human AD brains. First, there are no good data to show that Aβ oligomers produced in test tubes ever exist in AD brains, whereas ample data show that amphipathic Aβ peptides are susceptible to artifacts when manipulated in vitro (see Alzforum webinar). Second, analysis of CSF Aβ, which is perhaps the best indicator of how much soluble Aβ exists in AD brains (Aβ oligomers are never detected in CSF, see Alzforum news), shows that Aβ1-40 is present at ~2 nM and Aβ1-42 ~100 pM (Hata et al., 2012; Shaw et al., 2009), which is several orders of magnitude lower than used here. Finally, the ratio of CSF Aβ40:42 in AD brains is between ~25:1 and 50:1. Thus, the ratio of 7:3 used in this study is at least 10 times higher than that found in AD brains.
In-vitro studies are essential for understanding potential molecular mechanisms, but as a recent study from the Holtzman group shows, such studies can lead to incorrect conclusions (see Alzforum news). Previous studies based on the use of abnormally high ratios of Aβ to ApoE had concluded that ApoE sequesters Aβ and prevents its clearance from the brain. However, Holtzman's group came to the conclusion that under more physiological conditions Aβ does not bind ApoE, regardless of isoform (Verghese et al., 2013).
This is not to criticize the Zempel et al. study, since almost all studies use unrealistically high levels of Aβ, but to underscore the possibility that the continued practice of using unrealistic conditions—µM levels of Aβ oligomers whose in-vivo existence essentially cannot be proven—to document Aβ toxicity can lead to incorrect conclusions.
Hata S, Taniguchi M, Piao Y, Ikeuchi T, Fagan AM, Holtzman DM, Bateman R, Sohrabi HR, Martins RN, Gandy S, Urakami K, Suzuki T.
Multiple gamma-secretase product peptides are coordinately increased in concentration in the cerebrospinal fluid of a subpopulation of sporadic Alzheimer's disease subjects.
Mol Neurodegener. 2012 Apr 25;7(1):16.
Shaw LM, Vanderstichele H, Knapik-Czajka M, Clark CM, Aisen PS, Petersen RC, Blennow K, Soares H, Simon A, Lewczuk P, Dean R, Siemers E, Potter W, Lee VM, Trojanowski JQ, .
Cerebrospinal fluid biomarker signature in Alzheimer's disease neuroimaging initiative subjects.
Ann Neurol. 2009 Apr;65(4):403-13.
Verghese PB, Castellano JM, Garai K, Wang Y, Jiang H, Shah A, Bu G, Frieden C, Holtzman DM.
ApoE influences amyloid-β (Aβ) clearance despite minimal apoE/Aβ association in physiological conditions.
Proc Natl Acad Sci U S A. 2013 May 7;110(19):E1807-16.