Caspase-2 and tau—a toxic partnership?

Abstract

Misfolded and hyperphosphorylated forms of the microtubule-associated protein tau are thought to be responsible for some degree of neurodegeneration. The demonstration of a novel toxic cleavage of tau by caspase-2 opens up new therapeutic avenues.

Under physiological conditions, the microtubule-associated protein tau¹ is confined largely to neurons, and more specifically, to their axons². Its normal function is to promote microtubule assembly and stability³. In Alzheimer’s disease (AD), there is abnormal phosphorylation and aggregation of tau⁴, which leads to the formation of paired helical filaments (PHFs)^5,6, which form the neurofibrillary tangles that are a hallmark of AD. Identical tangles are formed in fronto-temporal dementia (FTD) and other tauopathies⁷.

Among the possible mechanisms by which tau might exert an adverse influence on neuronal function is through the generation of toxic tau cleavage products produced by enzymes⁸. To date, interest has been focused largely on the cleavage product Δtau421, which is produced through the cleavage of tau by caspase-3 and caspase-7 in the AD-afflicted brain⁹. In this issue, Zhao and colleagues¹⁰ describe a novel tau cleavage product—a 35-kD fragment termed TCP35—that is generated by the cleavage of tau at a caspase site at Asp314 by caspase-2, which leaves a truncated tau that the authors designate Δtau314. Δtau314 is nonfibrillogenic and seems to promote the missorting of tau to the dendritic spines (Fig. 1).

Figure 1.

Caspase-2 cleavage of tau. The normal dendritic spine (left) contains little or no tau. Zhao et al.¹⁰ show in a mouse model of ALS that tau is cleaved by caspase-2 (right), which results in both full-length tau and the truncated Δtau314 (leaving a fragment termed TCP35) entering the dendritic spine, and thus contributes to neurodegeneration.

In their study, Zhao et al.¹⁰ use rTg4510 mice between 2 and 3 months old that overexpress human tau^P301L, a mutant tau form seen in FTD. These animals show impaired spatial memory in the water maze but have not lost synapses or neurons—events that occur only later in the lifespan of this model. They find that the previously unidentified Δtau314 is produced in the brains of these animals. By using mass spectrometry, they identified the cleaved tau peptide (Δtau314) from the mice and ascertained that it was probably produced by caspase cleavage of tau. To determine which caspase could produce this peptide, they carried out a caspase-cleavage assay in vitro and found that only caspase-2 generated Δtau314. A modest inhibition (~33%) of caspase-2 levels in the brains of these animals leads to a decrease in the levels of Δtau314 and an improvement in their memory deficits.

To investigate whether Δtau314 has a role in pathological tau mislocalization, the authors overexpressed enhanced green fluorescent protein (EGFP)-labeled tau constructs in primary hippocampal neuron cultures. They showed that whereas wild-type tau, tau^P301L (the form expressed in the rTg4510 mice) and Δtau314 filled the dendritic shafts of neurons, wild-type tau did not enter the dendritic spines; tau^P301L and tauΔ314, however, both entered the spines. Mutation of tau^P301L at Asp314 (D314E) to block the caspase cleavage site decreased tau spine localization,whereas mutation of the putative caspase-3 site at Asp421 did not alter tau^P301L entry into spines. Furthermore, neurons from mice lacking caspase-2 showed little labeled wild-type tau or tau^P301L in spines, whereas Δtau314 entered spines freely.

The above changes in localization were reflected in the synaptic function of the neurons. The expression of tau^P301L decreased the frequency and amplitude of miniature excitatory postsynaptic currents (mEPSCs), yet the cleavage-resistant mutant tau^P301L,D314E had no effect on these currents. The expression of tau^P301L in hippocampal neurons of mice resulted in behavioral deficits in these mice in the water maze, a measure of spatial memory, and loss of hippocampal volume, as compared to mice expressing wild-type tau or the caspase-2-cleavage-resistant tau^P301L,D314E. Surprisingly, the analysis of postsynaptic compartments from tau^P301L mice indicates that both tau^P301L and Δtau314 enter the dendritic spines, but that the expression of EGFP-Δtau314 alone does not induce behavioral changes, despite the fact that its levels in dendritic spines were almost tenfold higher than those of Δtau314 seen in mice overexpressing tau^P301L. It is unclear whether this suggests that only full-length tau^P301L is toxic, that both are required for damage or that neither is toxic and caspase-2 activity is the crucial element. Alternatively, the detection of Δtau314 in the spines might be an artifact of its high level of overexpression and its lack of effect on synaptic function, a result of the loss of microtubule-binding domains 3 and 4 after cleavage.

The authors conclude that the cleavage of tau^P301L at Asp314 and the generation of Δtau314 are crucial steps in giving tau access to the dendritic spine, where it affects synaptic function and behavior. They thus propose that the inhibition of caspase-2 might be an effective way to block the adverse effects of tau in tauopathies and AD. There are some other possible interpretations of the data. For example, it has been shown that caspase-2 is required for the induction of apoptosis by amyloid-β (Aβ), the putative toxic peptide in AD¹¹, and that mice expressing human amyloidogenic amyloid precursor protein do not develop behavioral or electrophysiological deficits in the absence of caspase-2 or when caspase-2 is inhibited¹². It is possible that tau^P301L itself leads directly or indirectly to caspase-2 activation, and that this activation simultaneously generates Δtau314 and opens the spine in a nonspecific manner for tau entry. Thus, in the study by Zhou et al.¹⁰, it is possible that the failure of wild-type tau and tau^P301L,D314E to enter the spine was because these permutations fail to activate caspase-2 or to dissociate it from a binding partner. This question could be addressed by direct measurement of caspase-2 activity.

The work in this study is based on human tau^P301L isolated from rTg4510 mice. No direct data on the cleavage of wild-type tau by caspase-2 is given, but it can be inferred from the data presented in this paper that wild-type human tau might not be an important substrate for caspase-2, and that wild-type tau levels are lower than mutant tau levels. This raises the possibility that Δtau314 is not the driver of the dendritic-spine changes, but is simply a passenger in a caspase-2-driven cascade. It would also be intriguing to know whether other mutant human tau forms generate the same cleavage products.

The authors underscore the potential importance of this work by demonstrating increased levels of Δtau314 in the human AD-afflicted brain postmortem. It is hard to reconcile the increased levels of TCP35 and Δtau314 in the postmortem human brain, which would be expected to have wild-type tau, with the relative resistance of wild-type tau to caspase-2 cleavage that is implicit in the failure of wild-type tau to enter spines reported in this paper. Clearly, a direct measure of wild-type tau cleavage by caspase-2 in vitro would be valuable in this case. It is certainly possible that caspase-2 could cleave wild-type tau at Asp314 in the aging human brain, but this still needs to be examined. Resolution of this point is relevant to determining whether the effects of Δtau314 are specific to tauopathies or relevant to AD as well, as is suggested by these data.

In summary, this work opens up a new avenue of research on the effects of tau in the tauopathies and potentially in AD, and adds urgency to the need to explore the potential therapeutic application of caspase-2 inhibition in these diseases.