Amyloid-β-induced disruption of axon-initial-segment mitochondria localization: consequences for TAU missorting in Alzheimer’s disease pathology

TAU is a neuronal microtubule-associated protein preferentially located in axons. In a battery of neurodegenerative diseases termed “tauopathies,” including Alzheimer’s disease (AD), TAU is missorted and abnormally phosphorylated, leading to filamentous accumulations of hyperphosphorylated TAU, a pathological hallmark and potential disease driver of AD and related tauopathies (Zempel, 2024). The other hallmark of AD is the aberrant production of amyloid-β (Aβ), which accumulates in extracellular plaques. The axon initial segment (AIS) is a 20 to 60 µM long neuronal compartment separating the axon from the axon hillock and cell body. Its molecular organization is characterized by a complex protein scaffold, consisting of transmembrane proteins, Ankyrin G (ANKG), spectrins, actin structures (rings/patches composed of filamentous actin (F-actin), microtubules, and microtubule-associated proteins (Leterrier, 2018). Alterations of the AIS structure have been observed as a result of Aβ exposure: F-actin is lost at the AIS after exposure to amyloid-β-oligomers (AβO; AD-like stress), likely due to aberrant activation of the F-actin severing enzyme cofilin (actin depolymerizing factor). Subsequent rapid remodeling and repolarization of F-actin might induce TAU missorting by blocking the physiological anterograde sorting of TAU into the axon (Zempel et al., 2017, Bell-Simons et al., 2023, Buchholz and Zempel, 2024). Recently, we also provided evidence for a cluster of mitochondria in the proximal AIS, which has a functional role in the maintenance of TAU sorting: While mitochondria show relative absence in the central AIS, they cluster at the proximal AIS. The mitochondria in this cluster are largely immobile and only sparsely participate in axonal mitochondria-trafficking. Importantly, (local) impairment of this AIS-mitochondria-cluster (AMC) leads to detectable increases of somatic TAU, resembling AD-like TAU missorting (Tjiang and Zempel, 2022). While it is thus clear that both proper maintenance of F-actin in the AIS, as well as proper function of AIS-based mitochondria, may be crucial for neuronal function and maintenance of TAU sorting and function, the effect of AD-like stress on mitochondria in the AIS is underexplored.

Here we provide preliminary data showing that AβO (as a surrogate of AD-like stress) induces disruption of the proximal AIS-located mitochondrial barrier and causes mislocalization of the proximal mitochondria to the proximal and central AIS. This effect might occur in response to the AβO-dependent activation of cofilin and subsequent aberrant repolarization of F-actin in the AIS, which corresponds well to the observed TAU missorting. We aim to address how mitochondrial translocation and AIS remodeling are implicated in pathological TAU missorting and share first insights hinting towards an interdependence of mitochondria and TAU mislocalization here.

The missorting of TAU plays an important role in AD and other neurodegenerative diseases. We have previously reported that the AIS-located F-actin and the AMC are involved in pathophysiological TAU sorting. Here we present preliminary data, showing disintegration of the AMC after AD-like stress and propose a multi-step patho-cascade upon AβO-insult involving cofilin, F-actin remodeling, and dissolution of the AMC leading to breakdown of the axonal TAU diffusion barrier.

To evaluate the effects of Aβ treatment on the AMC, we co-stained primary rat hippocampal neurons at day 21 for MAP2, ANKG, and mitochondria (as in Zempel et al., 2017; Zempel and Mandelkow, 2017), and treated them with AβO to introduce AD-like stress. Untreated cells confirm previously reported observations of mitochondria clustering in the proximal AIS but not within the central AIS (Tjiang and Zempel, 2022). Upon AβO insult, we observe that: (i) the mitochondrial cluster located at the AIS dissociates, and (ii) mitochondria relocate from their proximal position to the central AIS. However, cells treated with 1 µM AβO as early as 30 minutes exhibit a decrease in ANKG levels, indicating AIS destabilization (Zempel et al., 2017). Additionally, there is mitochondrial translocation from the proximal AMC to the central AIS (Figure 1A and B). The increase in mitochondria levels in the central AIS at approximately 25% after 1 hour, indicating a brief period of regeneration 30 minutes after Aβ stress, followed by a sustained rise in mitochondria within the central AIS throughout our 3-hour measurement period. At the same time, we see an increase in mitochondrial influx into the axon, as indicated by up to 75 % increase in mitochondria at the proximal AIS (Figure 1C). To support our hypothesis by live-imaging data, we also transfected rat hippocampal neurons at 16 days with mitoRFP and visualized the effect of AβO on the AMC in time-lapse imaging. Dissolution of the AMC is seen as early as 10 minutes after exposure to AβO and appears highly dynamic with multiple translocation events of mitoRFP-tagged mitochondria across the AIS boundary (Figure 1D). It is unusual that the mitochondrial localization at the AMC changes rapidly, because we have previously demonstrated that the AMC is a rather stable accumulation of mitochondria that do not contribute significantly to the axonal trafficking of mitochondria (Tjiang and Zempel, 2022). The fast changes after AβO-exposure are an indication of AD-like stress-induced AMC disruption, which we hypothesize here to be of importance and that we believe warrants future investigation.

Figure 1.

Amyloid-β insult induces mitochondria displacement into the central AIS.

(A) Overview image of untreated rat hippocampal neurons stained for ANKG (red), as an AIS marker, and MAP2 (blue) as a dendritic marker. cb = cell body, arrowheads mark AIS. (B) Rat hippocampal neurons (21 DIV) were stained for MAP2 (blue), ANKG (red), and mitochondria (green). Cells were vehicle-treated (top, ctrl) or treated with 1 µM AβO for 30 minutes (bottom). Left: image showing the proximal AIS (boxed area) and the cell body (cb). Right: Magnification of boxed area. (Proximal) axons are indicated by dotted lines, arrowheads point to the void space of the AIS in control conditions, and to mitochondrial invasion of the AIS after treatment. Scale bar: 2 µm. (C) Quantification of B, the relative change of mitochondrial fluorescence intensity within the AIS after Aβ insult for up to 3 hours. Changes are detected in independent cultures at time points 0, 5, 15, 30, 60, and 180 minutes after fixation of AβO-treated cultures. Central AIS: red, proximal AIS, grey. Mitochondria invade the AIS minutes after insult. Error bars show SEM. N = 2, n = 10. (D) Rat hippocampal neurons (16 DIV) were transfected with mitoRFP for 5 days, treated with 1 µM AβO, and imaged via time-lapse imaging. The axon is shown at 4, 10, 17, and 31 minutes (left to right). Dotted lines indicate the axon, arrowheads indicate the central AIS, cb = cell body, scale bar corresponds to 5 µm. Areas in red boxes are magnified 1.7 times. Note that mitochondria invade the AIS minutes after insult, but show also rapid reorganization starting within half an hour. (E) Proposed mechanism: AD-like stress results in cofilin/ADF-based F-actin disassembly, mitochondrial displacement into the AIS, and aberrant restructuring of F-actin. Healthy neurons exhibit organized/structured AIS, containing regular Ankyrin G (orange) and actin rings (purple), and AMC (i.e. mitochondria are excluded from the central AIS) and the TAU diffusion barrier (TDB) is intact, TAU (red) is sorted into the axon and largely excluded from the AIS and the cell body. After Aβ insult, the cofilin/Actin depolymerizing factor (ADF, blue) is activated via dephosphorylation at S203 and cleaves essential F-actin structures at the AIS. This loss of F-actin leads to destabilization of the AIS, as can be seen, e.g. by loss of Ankyrin G. The AMC destabilizes and mitochondria translocate to the central AIS. After aberrant activation of cofilin and loss of F-actin structures at the AIS, F-actin may be aberrantly regenerated within the AIS, leading to loss of the TAU diffusion barrier and pathological TAU missorting, a hallmark of AD. Unpublished data. Artworks have been designed with Microsoft PowerPoint. AbO: Amyloid-beta oligomers; AD: Alzheimer’s disease; ADF: actin depolymerization factor (cofilin); AIS: axon inital segment; AMC: AIS-located mitochondria cluster; ANKG: ankyrin G; CB: cell body; ctrl: control; DIV: days in vitro; MAP2: microtubule-associated protein 2; SEM: standard error of the mean; TDB: TAU diffusion barrier; WT: wildtype.

These results (both life-imaging/pulse-chase and immunostaining-based microscopy) show that Aβ insult to neurons may have a direct effect on mitochondrial dynamics within the AIS. As the AIS and the AMC have both been shown to be involved in the pathogenesis of TAU mislocalization (Li et al., 2011; Zempel et al., 2017; Tjiang and Zempel, 2022) the alteration of the AMC might play an important role in TAU and Aβ related pathogenesis. In cell culture conditions, the disintegration of the AMC appears as early as 15 minutes after Aβ insult (Figure 1D), concurrent with dephosphorylation of cofilin (and hence activation of its F-actin severing activity) and loss of F-actin at the AIS (Zempel et al., 2017), but much earlier than TAU pathology appears. While the data presented here are preliminary and require further confirmation, they are in line with our previous results and the data from others (Zempel et al., 2017; Woo et al., 2019; Tjiang and Zempel, 2022). We believe that in combination we can now propose the hypothesis that at least in paradigms of AD-like stress, AIS dysfunction and (AIS-based) mitochondrial impairment may well be an early and determining event in disease, and warrants further investigation.

Before TAU pathology manifests, activated cofilin may play a dual role, not only in severing F-actin but also in destabilizing microtubules, subsequently impairing neuronal function through various mechanisms. Active cofilin, distinct from its inactive form, selectively interacts with tubulin, leading to the destabilization of microtubules. Moreover, it competes with TAU for direct microtubule binding, thereby inhibiting TAU-induced microtubule assembly (Woo et al., 2019). Recently it has been demonstrated that cofilin is cleaved by an asparagine endopeptidase at N138 in AD brains. The resulting fragment cofilin 1–138 not only interacts with TAU and promotes its aggregation, but also localizes to mitochondria upon its dephosphorylation and enhances mitochondrial-dependent apoptosis, leading to neuronal loss (Yan et al., 2023). Thus, we hypothesize that Aβ-induced activation of cofilin leads to a multilevel patho-cascade in which dephosphorylated/activated cofilin not only leads to transient loss of anchoring of the AMC and disruption of the TAU diffusion barrier via F-actin severing and aberrant restructuring as demonstrated by us, but possibly also to direct destabilization of microtubules and, in its cleaved form, to direct mitochondrial-impairment. We can currently not exclude the proteins that anchor mitochondria to F-actin, e.g. the proteins four and a half LIM domains protein 2 (Basu et al., 2021), Spire1C (Manur et al., 2015), or the more widely involved proteins syntaphilin or one of the mitochondria immobilizing myosins (in particular Myo19) are also involved. But because of the well-matching timecourse of cofilin activation and loss of F-actin, and mitochondrial invasion into the AIS, we currently hypothesize that indeed the cofilin-actin axis is the primary cause of the AMC disintegration.

We propose the following pathway: In AD and related tauopathies, chronic Aβ-based stress (mimicked here by AβO exposure) leads to dephosphorylation (and thereby activation) of cofilin (Zempel et al., 2017; Woo et al., 2019). At the AIS, activated cofilin cleaves stabilizing actin rings, leading to loss of AIS stability, also evidenced by loss of ANKG and impaired microtubule dynamics at the AIS (Zempel et al., 2017). This loss of key structural elements in turn leads to impairment of the anchoring mechanism of the AMC and subsequent aberrant translocation of mitochondria into the central AIS, as shown here (Figure 1A–D) and previously by us (Tjiang and Zempel, 2022). While the precise pathophysiological effect of this translocation remains elusive, TAU mislocalization may thus be downstream of AMC/AIS impairment, and TAU likely does not directly impact the AMC. However, it is unclear whether AMC impairment constitutes a driver of AD-related pathology, a side effect of aberrant cofilin activity or fragments (e.g. its 1–138 fragment) and subsequent structural impairment at the AIS, or whether different mechanisms are involved. Our data suggest the breakdown of the TAU diffusion barrier after deactivation of cofilin due to aberrant actin remodeling (Zempel et al., 2017) at the distal AIS, impeding physiological TAU sorting (Figure 1E). However, the remodeling of actin warrants further investigation. In healthy neurons, actin is polymerized into periodical rings along the AIS, as demonstrated using STED-nanoscopy (Leterrier, 2018). Super-resolution imaging will have to reveal the structure of aberrant actin after Aβ-insult and may shed light upon the loss of physiological TAU sorting. Furthermore, these results have not been replicated in-vivo so far and it is unclear, whether the deactivation of cofilin is a pathological feature, or whether it is a cell culture artifact after decreased Aβ-toxicity.

Here, we provide preliminary data that indicate that AβO induces the disintegration of AIS-based mitochondrial clustering and the translocation of mitochondria into both the proximal and central AIS. This translocation phenomenon has not been previously studied and its role remains unclear. Additionally, it remains unclear whether this translocation constitutes only a secondary effect that is not strictly pathophysiological. However, we here hypothesize that it may play a crucial role in the development of neurodegeneration in Alzheimer’s disease. While future studies must provide more evidence, translocation of mitochondria is more plausible than other phenomena, e.g. spawning or elimination of mitochondria, as this was observed previously (Tjiang and Zempel, 2022) and also occurs an order of magnitude slower (~2 hours) than the changes observed here (as soon as 10–15 minutes). Future studies must address whether the reorganization of the AIS actin structure, mitochondria anchoring proteins, or specific, yet to be postulated AIS-specific signaling, are an important mechanism for the maintenance of AIS function and neuronal cell polarity and TAU anterograde sorting, and, most importantly, whether this proposed mechanism is a druggable target in the many detrimental neurodegenerative diseases.

Additional file: Open peer review report 1 ^{(87.2KB, pdf)} .