Axons surrounding amyloid plaques swell, but what goes wrong inside, and with what consequences, remains murky. Using high-resolution imaging techniques, scientists led by Jaime Grutzendler at Yale University now report that around each amyloid plaque, hundreds of axons develop dystrophic neurites, or what the authors call plaque-associated axonal spheroids. In mice, these spheroids act as sinks for electric current, slowing down action potentials and disrupting long-range connectivity. The more these spheroids accumulated enlarged lysosomes, the bigger they got, and the weaker the action potentials. Knocking down expression of the phospholipase D3 gene shrank lysosomes and improved neurotransmission and network connectivity, the scientists report. The findings, published November 30 in Nature, suggest that independently of targeting amyloid plaques, modulating these spheroids could correct network dysfunction in AD.

“This report adds to the steadily emerging genetic and pathological evidence that the endosomal-lysosomal-autophagy axis is a likely primary catalyst and driver of AD pathogenesis,” wrote Ralph Nixon and Ju-Hyun Lee of New York University, Orangeburg (comment below). Nixon recently reported that faulty lysosomes stuffed with Aβ distort neurons, forcing them to burst, and leaving behind amyloid cores (Jun 2022 news).

To get a clearer picture of spheroid dynamics, co-first authors Peng Yuan, Mengyang Zhang, and Lei Tong used in vivo two-photon imaging to peer into the brains of 8-month-old 5xFAD mice. These animals develop amyloid plaques by 2 months and lose synapses starting at 4 months. Some spheroids remained stable in the mice over two months, while others shrank or even disappeared, leaving an intact axon behind. However, most of them grew in size.

How would these blebs affect axons? To find out, the scientists used viruses to express calcium and voltage reporters inside neurons as a means of measuring neurotransmission. Notably, the larger the axonal spheroids, the weaker the action potentials and the more slowly they traveled past the swellings. Signaling deficits appeared not only in axons next to the spheroids, but also in distant neurons, even those in the opposite hemisphere, suggesting that whole neuronal networks were disrupted. More current was needed to propagate action potentials between brain hemisphere in 5xFAD mice than in control animals. The researchers attributed this to the spheroids, hypothesizing that they act as electron sinks to siphon away the electrical signal, preventing it from passing along to the next neuron.

What’s in a Spheroid? Around amyloid plaques (light blue) in a mouse model of amyloidosis (left) or in people who had had AD (middle and right), axons swelled with multivesicular bodies containing the lysosomal proteins LAMP1 (green), v-ATPase (red), and PLD3 (purple). [Courtesy of Yuan et al., Nature, 2022.]

Confocal and electron microscopy identified enlarged endosomes and lysosomes within the spheroids (see image above). These were poorly acidified, had a dearth of the lysosomal protease, cathepsin D, and were stuffed with phospholipase D3. Ditto for spheroids in postmortem cortical tissue from six people who had had mild cognitive impairment, and in tissue from16 who had had AD. Spheroids in the AD tissue were larger and were engorged with lysosomes. Four of them carried the V232M PLD3 substitution, and they had even more of the swollen organelles per spheroid than noncarriers. This finding aligns with previous research suggesting that this variant, a potential risk allele for AD, hinders lysosome function (Dec 2013 newsDemirev et al., 2019).

Spheroid PLD3Swollen axons (green) around plaques (light blue) in 5xFAD mice (left) were even larger when the animal overexpressed PLD3 (middle) and they filled with swollen lysosomes (red). In contrast, knocking down PLD3 reduced the size of spheroids (right). [Courtesy of Yuan et al., Nature, 2022.]

Would tinkering with Pld3 expression alter the spheroids or neuron signaling? Overexpressing PLD3 in 7-month-old 5xFAD mice increased the size of the spheroids and dampened neuron signaling three months later, while knocking down expression shrank spheroids, restored neurotransmission to wild-type levels, and improved network connectivity (see image above). All told, the authors concluded that Pld3 hobbles lysosomes in axons around amyloid plaques, which disrupts neuronal connections.—Chelsea Weidman Burke

Comments

  1. This report by Yan and colleagues adds to the steadily emerging genetic and pathological evidence that the endosomal-lysosomal-autophagy (ELA) axis is likely a primary catalyst and driver of Alzheimer’s disease pathogenesis. Their work convincingly links the lysosomal protein, PLD3, an AD risk factor, to the disruption of this axis within neurons, leading to plaque-associated axonal spheroids and interruption of neuronal circuits. Notably, similar autophagy-related vesicles containing Aβ massively accumulate in the perikarya of selected neurons at a very early stage of disease in mouse AD models as a result of failing acidification of autolysosomes (Lee et al., 2022). This early pathobiology arising within neurons is the origin of the amyloid deposited extracellularly. Consistent with the conclusion in Yan et al., certain hallmark lesions in AD may develop due to earlier ELA dysfunction. It would be exciting for this investigative team to examine how the striking perikaryal autophagic stress visible at the earliest stages of disease in the 5xFAD model relates to the axonal dystrophy more prominent in the older animals.

    The fascinating new findings of Yan et al. confirm and extend our studies (Lee et al., 2011) demonstrating selective stalling of retrogradely moving autophagic/amphisomal vesicles, and their accumulation selectively within axonal swellings in primary cortical neurons treated with an inhibitor of lysosomal cathepsins or acidification. We also recently reported a similar phenomenon and mechanism in an FAD mutant PSEN1 knock-in mouse, which we showed develops a late-age onset axonal dystrophy, also characterized by selective accumulation of degradative vesicles within the axon swellings (Lie et al., 2022). Pearl Lie demonstrated selectively impaired retrograde transport of pH-deficient autophagic/amphisomal vesicles in primary cortical neurons from this mouse. These studies extend our earlier reports that PEN1-dependent lysosomal acidification deficiency arises from defective vATPase subunit maturation (Lee et al., 2010; Lee et al., 2015). This phenomenon could be reversed if proper lysosome acidity was restored by using one of several interventions. Since amyloid does not form in the PSEN1 KI mouse, the dynamic formation of dystrophic swellings is possible in this AD model in the absence of amyloid plaques (see also Lomoio et al., 2020). 

    References:

    . Faulty autolysosome acidification in Alzheimer's disease mouse models induces autophagic build-up of Aβ in neurons, yielding senile plaques. Nat Neurosci. 2022 Jun;25(6):688-701. Epub 2022 Jun 2 PubMed.

    . Lysosomal proteolysis inhibition selectively disrupts axonal transport of degradative organelles and causes an Alzheimer's-like axonal dystrophy. J Neurosci. 2011 May 25;31(21):7817-30. PubMed.

    . Presenilin 1 Maintains Lysosomal Ca(2+) Homeostasis via TRPML1 by Regulating vATPase-Mediated Lysosome Acidification. Cell Rep. 2015 Sep 1;12(9):1430-44. Epub 2015 Aug 20 PubMed.

    . Lysosomal proteolysis inhibition selectively disrupts axonal transport of degradative organelles and causes an Alzheimer's-like axonal dystrophy. J Neurosci. 2011 May 25;31(21):7817-30. PubMed.

    . Axonal transport of late endosomes and amphisomes is selectively modulated by local Ca2+ efflux and disrupted by PSEN1 loss of function. Sci Adv. 2022 Apr 29;8(17):eabj5716. PubMed.

    . Gga3 deletion and a GGA3 rare variant associated with late onset Alzheimer's disease trigger BACE1 accumulation in axonal swellings. Sci Transl Med. 2020 Nov 18;12(570) PubMed.

  2. The authors investigated plaque-associated axonal spheroids (PAASs) in relation to calcium signaling and PLD3 expression. They suggest that PLD3 is mechanistically linked to the accumulation of enlarged endolysosomal compartments: overexpression leads to an increase and to spheroid enlargement whereas knockdown reduced it, improving network function.

    Although the study of PAAS is highly relevant to better understand their relevance to AD disease pathogenesis, it has already been reported that PLD3 accumulates in dystrophic neurites; but this is as well true for other AD-relevant proteins, such as BACE1, APP, etc. Moreover, it is unclear, or not studied, how the observations relate to the function of PLD3. While the authors suggest that PLD3 may function in multivesicular body (MVB) maturation, this is not supported by mechanistic data. Moreover, the active (processed) PLD3 enzyme is confined to lysosomes, where it has been reported to exert nuclease activity and/or phospholipase activity, although the latter may be more unlikely given that the second PDE domain, which is required for phospholipase activity, is mutated in PLD3. How this activity contributes to PAAS and the accumulation of enlarged endo-lysosomes is not addressed.

    Moreover, while the authors used overexpression of GFP-tagged PLD3, PLD3 does not tolerate tags that well (Gonzalez et al., 2018). Hence, overexpressed GFP-PLD3 may be mislocalized and/or not be processed/sorted to its active form in lysosomes, as evident from the broader distribution of GFP-PLD3 in spheroids, i.e., not restricted to Lamp1-positive organelles. The ectopic, non-lysosomal localization (and processing?) of PLD3 raises the question of whether interpretations would be physiologically relevant. Further, the authors show that overexpression of PLD3 does not impact APP processing. Although only addressed by one brief communication in Nature (Fazzari et al., 2017), most studies support an inverse correlation between PLD3 levels and dysregulation of APP processing (Mukadam  et al., 2018; Cruchaga et al., 2014; Nackenoff et al., 2021; Blanco-Luquin et al., 2018; Satoh et al., 2014).

    Moreover, when depleting PLD3 in their models, a reduction in spheroid size and decrease in endolysosomal compartments was noticed. Although PAAS specifically were not studied, this outcome seems contradictory to previous studies showing that PLD3 KO and PLD3 SNPs cause lysosomal dysfunction and enlargement (Demirev et al., 2019; Fazzari et al., 2017). Can the authors exclude that the sgRNAs they used in their study do not have off-target effects that could confound these observations?

    References:

    . PLD3 and spinocerebellar ataxia. Brain. 2018 Nov 1;141(11):e78. PubMed.

    . PLD3 gene and processing of APP. Nature. 2017 Jan 25;541(7638):E1-E2. PubMed.

    . Analysis of novel endosome-to-Golgi retrieval genes reveals a role for PLD3 in regulating endosomal protein sorting and amyloid precursor protein processing. Cell Mol Life Sci. 2018 Jul;75(14):2613-2625. Epub 2018 Jan 24 PubMed.

    . Rare coding variants in the phospholipase D3 gene confer risk for Alzheimer's disease. Nature. 2014 Jan 23;505(7484):550-4. Epub 2013 Dec 11 PubMed.

    . PLD3 is a neuronal lysosomal phospholipase D associated with β-amyloid plaques and cognitive function in Alzheimer's disease. PLoS Genet. 2021 Apr;17(4):e1009406. Epub 2021 Apr 8 PubMed.

    . PLD3 epigenetic changes in the hippocampus of Alzheimer's disease. Clin Epigenetics. 2018 Sep 12;10(1):116. PubMed.

    . PLD3 is accumulated on neuritic plaques in Alzheimer's disease brains. Alzheimers Res Ther. 2014;6(9):70. Epub 2014 Nov 2 PubMed.

    . V232M substitution restricts a distinct O-glycosylation of PLD3 and its neuroprotective function. Neurobiol Dis. 2019 Sep;129:182-194. Epub 2019 May 20 PubMed.

  3. This is an interesting paper, showing by high resolution imaging techniques that around each amyloid plaque hundreds of axons develop dystrophic neurites. It also provides pathological evidence that the endosomal-lysosomal-autophagy axis may be a primary catalyst and leads to pathogenesis in AD. The lysosomal protein, PLD3, also considered as an AD risk factor leading to plaque-associated axonal spheroids and interruption of neuronal circuits, is involved.

    In 2014 we published a similar study showing nuclear M78 and NeuN immunoreactivities are localized in the center of neuritic plaques, surrounded by M78 and APP-CTF positive dystrophic neurites in the 3xTg-AD mice model. Dystrophic neurites that are immunopositive for both M78 and APP-CTF are found surrounding DAPI and M78 positive cores. The idea that the chromatin at the center of the neuritic plaque is derived from neuronal nuclei is supported by the observation that most of the cores of neuritic plaques are positive for the neuronal nuclear marker, NeuN (Pensalfini et al., 2014).

    References:

    . Intracellular amyloid and the neuronal origin of Alzheimer neuritic plaques. Neurobiol Dis. 2014 Nov;71:53-61. Epub 2014 Aug 1 PubMed.

  4. We appreciate the previous comments on our paper.

    We would like to clarify a few things, especially with regards to comments by Annaert, Van Acker, and Damme. We agree that there is a lot to learn about how PLD3 dysfunction or accumulation in spheroids might lead to lysosomal abnormalities and subsequently to spheroid enlargement.

    However, a point made in the comment was that in our viral-mediated expression experiments, PLD3-GFP fusion might disrupt normal PLD3 function, therefore clouding the interpretability of our results. This statement is inaccurate, as our studies never used a PLD3-GFP fusion protein but rather a gfp-p2a-pld3 construct, where GFP and PLD3 are expressed independently.

    The commentators also raised the possibility that, in our CRISPR/Cas9 ko experiments, the results might not be specific but rather an off-target effect of the sgRNA. This point is a general concern for any experiment using CRISPR; however, in our case in separate experiments, we used two independent sgRNAs, showing target engagement, and the manipulations led to reduced spheroid size, reduced aberrant lysosomes and improved electrical conduction, consistent with our other findings. Thus, we think it is extremely unlikely that the improvement in axonal phenotype we observed is due to some other gene than the desired target, which was PLD3.

    With regard to whether our results agree or not with previous studies, the only thing we can say for certain is that the few functional papers available on PLD3 are in very different contexts, models, age, and cell types, and cannot be directly compared to our in vivo Alzheimer’s conditions. This does point to the complexity of these mechanisms and the importance of further research to better understand the endolysosomal and autophagy pathways in axons and their role in neurodegenerative disease.

    We do think that our study provides a nice framework for understanding axonal pathology in AD in an in vivo model. Dystrophic neurites have been known to exist for more than 100 years, and people routinely use them as a pathological hallmark in their studies. Our study has, for the first time, used a single-axon imaging, molecular, optical electrophysiological and computational approach to study these structures comprehensively. We think we provide compelling evidence of the importance of endolysosomal abnormalities associated with amyloid plaques in the disruption of axonal conduction and downstream neural circuits.

    Our study also highlights the potential reversibility of plaque-associated axonal spheroids and beneficial effects on interconnected neural circuits. Therefore, we think that current strategies for amyloid plaque removal are an insufficient approach, and addressing axonal spheroids as a therapeutic target is critical. Hopefully our study stimulates further mechanistic investigations and innovation on the best therapeutic approach.

  5. Dr. Grutzendler's study is an important contribution to the field. Its strength is that the findings link, through defects in endo-lysosomal handling, two seemingly distinct features of AD pathology—abnormal protein aggregates and neural circuit dysfunction. I'm intrigued because this nicely aligns with a recent study in my lab, which demonstrated that the lysosomes in AD lack proteolysis capacity due to a more alkaline pH, and it is consistent with engorged lysosomes and lack of corresponding protein degradation of amyloid (Mustaly-Kalimi et al., 2022).

    It is a bit harder to say if the lysosomes solely serve as the current sink, and/or the pathogenic protein aggregates compromising membrane integrity in axons/neurites cause ionic leaks—both could look similar physiologically in terms of AP conduction defects. But in concert with Randy Nixon's comments, there is accumulating evidence pointing to endo-lysosomal pathways as central components of pathophysiology in AD, and this study adds a compelling piece to that puzzle.

    References:

    . Protein mishandling and impaired lysosomal proteolysis generated through calcium dysregulation in Alzheimer's disease. Proc Natl Acad Sci U S A. 2022 Dec 6;119(49):e2211999119. Epub 2022 Nov 28 PubMed.

  6. PAAS, PANTHOS, and TRIAD: The Trinity of Early Stage AD pathology?

    This highly interesting work makes a great contribution for understanding unsolved enigmas in Alzheimer’s disease. It reveals that accumulation of LAMP1-positive vesicles in the autophagy-lysosomal degradation pathways (Extended Data Fig 5 of Yuan et al.) causes axonal spheroids, which has been observed in various neurodegenerative diseases including Alzheimer’s, and that underlying this phenotype is PLD3, a protein abundant in endoplasmic reticulum, endosome, and lysosome. Importantly, these findings have close relationships with recent results on the early stage pathologies of Alzheimer’s disease (Fujita et al., 2016; Tanaka et al., 2020; Lee et al., 2022), as suggested by some comments on this page.

    Before addressing the critical commonness shared with recent results, we would like to point out some small issues in Yuan et al. First, the authors used ThioflavinS for detection of Aβ, and showed various sizes of Aβ accumulation. For instance, in Figure 1b it is definite that enlarged axonal terminals arising from the contralateral hemisphere surround Aβ accumulation. However, the ThioflavinS stain is very small and could be a single-cell size. The authors also showed some larger Aβ accumulations (Extended Data Fig 1b). As 5xFAD mice show intracellular Aβ accumulations (Oakley et al., 2006), we wonder how different sizes of “Aβ plaques” correspond to intracellular Aβ accumulation in a single neuron or at a few neurons or to extracellular Aβ accumulation derived from multiple neurons (Tanaka et al., 2020). We feel it might be better not to call a single-cell level of Ab accumulation an Aβ plaque, while extracellular versus intracellular Aβ accumulation remains a matter of discussion.

    Second, we wonder whether the axonal swelling surrounding a spot of Aβ accumulation (“stable” in Figure 1c) and the axonal swelling at a certain distance from Aβ accumulation (“dynamic” in Figure 1c) could be considered as a similar process, because an AAV-GFP-positive axon does not seem positive for LAMP1 (Extended Data Figure 1b), though this image might represent an exception.

    More broadly, the similarity of Yuan et al.'s findings to some recent results is very impressive. Our group revealed that degenerative neurites full of autophagosome were surrounding TRIAD necrosis of a single neuron caused by intracellular Aβ accumulation in 5xFAD and APP-KI mouse models as well as in human AD postmortem brains (Fujita et al., 2016; Tanaka et al., 2020). We also showed the molecular mechanism that HMGB1 released from necrotic neurons induces neurite degeneration via phosphorylation of a submembrane molecule MARCKS and neuronal death via phosphorylation of Ku70 (Fujita et al., 2016; Tanaka et al., 2020; Jin et al., 2021; Tanaka et al., 2021).

    Aβ accumulation/aggregate develops from the level of a single cell at 3 months of age to that of multiple cells at 6 months of age. Immunostaining with anti-phospho-Ser46-MARCKS antibody reveals degenerative/dystrophic neurites (white arrow) around TRIAD necrosis similar to PAAS. Axon swellings distal from neuronal necrosis are stained with anti-phospho-Ser46-MARCKS antibody (yellow arrow).

    Interestingly, we observed phospho-MARCKS-positive neurites (see image at right) like axon swellings reported in Yuan et al. The group of Ralph Nixon also observed a similar autophagosome-lysosome change and called it PANTHOS (Lee et al., 2022). They proposed a model that the structures full of autophagosome-lysosome protruded from a centrally located dying neuron with Aβ accumulation. According to Yuan et al., most of these structures are degenerative neurites coming from other neurons, such as neurons in contralateral hemisphere (see Figure 1b). This observation of degenerative neurites from other neurons full of autophagosome-lysosome matches very well with the findings described in Greenfield’s Neuropathology Textbook (Graham and Lantos, 2002) and observed by our previous reports (Fujita et al., 2016; Tanaka et al., 2020; Jin et al., 2021; Tanaka et al., 2021). However, we also think that some of these processes represent protrusions from the dying neuron located inside degenerative neurites.

    In Yuan et al., the authors injected Aβ and observed a similar phenomenon. This does not mean that this phenomenon is derived from cell signaling triggered by extracellular Aβ. Instead it could be a phenomenon induced by intracellular Aβ, because we previously revealed that injected Aβ is taken up by neurons and subcellularly transported to the autophagosome-lysosome degradation system to finally induce neuronal necrosis (Chen et al., 2015). 

    The important takeaway here is that three groups have independently observed the similar phenomenon (see image below) of a single cell, or a few cells, of necrosis surrounded by degenerative and swelling axons/neurites full of structures of the autophagosome-lysosome system. This indicates that this is in fact occurring in early stages of AD pathology.

    TRIAD, PANTHOS, and PAAS could reflect three aspects of a single phenomenon, i.e., degenerative/dystrophic neurites (red) around and at a little distance from neuronal necrosis (purple) caused by intra-neuronal Ab accumulation (green). Such neurites are full of organelles of the autophagosome-lysosome system.

    One of the remaining issues would be which molecules mediate axon swelling in the proximal and distal regions from neuronal necrosis. Yuan et al. revealed that PLD3 plays an essential role for axon swelling within axons. On the other hand, our group revealed that HMGB1 released from necrotic neurons triggers TLR4-MARCKS and TLR4-Ku70 signals to induce neurite degeneration (Fujita et al., 2016) and neuronal death (Tanaka et al., 2021), respectively. Interestingly, PLD3 has an activity of endolysosomal ssDNA exonuclease (Gavin et al., 2018) and plays a role in ssDNA-sensing by TLR9 in endosome and lysosome (Gavin et al., 2021). 

    It is possible that damage-associated molecular patterns (DAMPs) released from necrotic neurons, such as HMGB1, DNAs and Aβ, trigger axon swelling and other phenotypes of neurite degeneration (see image above). Further studies in the future will likely unravel the relationship between DAMPs and associated neurite changes.

    References:

    . Intraneuronal beta-amyloid aggregates, neurodegeneration, and neuron loss in transgenic mice with five familial Alzheimer's disease mutations: potential factors in amyloid plaque formation. J Neurosci. 2006 Oct 4;26(40):10129-40. PubMed.

    . YAP-dependent necrosis occurs in early stages of Alzheimer's disease and regulates mouse model pathology. Nat Commun. 2020 Jan 24;11(1):507. PubMed.

    . HMGB1, a pathogenic molecule that induces neurite degeneration via TLR4-MARCKS, is a potential therapeutic target for Alzheimer's disease. Sci Rep. 2016 Aug 25;6:31895. PubMed.

    . Prediction and verification of the AD-FTLD common pathomechanism based on dynamic molecular network analysis. Commun Biol. 2021 Aug 12;4(1):961. PubMed.

    . HMGB1 signaling phosphorylates Ku70 and impairs DNA damage repair in Alzheimer's disease pathology. Commun Biol. 2021 Oct 11;4(1):1175. PubMed.

    . Faulty autolysosome acidification in Alzheimer's disease mouse models induces autophagic build-up of Aβ in neurons, yielding senile plaques. Nat Neurosci. 2022 Jun;25(6):688-701. Epub 2022 Jun 2 PubMed.

    . Greenfield’s Neuropathology (2 Vol. Set). 7th Ed. vol. 2, 210

    . Fasting activates macroautophagy in neurons of Alzheimer's disease mouse model but is insufficient to degrade amyloid-beta. Sci Rep. 2015 Jul 14;5:12115. PubMed.

    . PLD3 and PLD4 are single-stranded acid exonucleases that regulate endosomal nucleic-acid sensing. Nat Immunol. 2018 Sep;19(9):942-953. Epub 2018 Aug 13 PubMed.

    . Cleavage of DNA and RNA by PLD3 and PLD4 limits autoinflammatory triggering by multiple sensors. Nat Commun. 2021 Oct 7;12(1):5874. PubMed.

Make a Comment

To make a comment you must login or register.

References

News Citations

  1. Behold PANTHOS, a Toxic Wreath of Perinuclear Aβ That Kills Neurons
  2. Phospholipase D3 Variants Double Sporadic AD Risk

Research Models Citations

  1. 5xFAD (B6SJL)

Paper Citations

  1. . V232M substitution restricts a distinct O-glycosylation of PLD3 and its neuroprotective function. Neurobiol Dis. 2019 Sep;129:182-194. Epub 2019 May 20 PubMed.

Further Reading

No Available Further Reading

Primary Papers

  1. . PLD3 affects axonal spheroids and network defects in Alzheimer's disease. Nature. 2022 Dec;612(7939):328-337. Epub 2022 Nov 30 PubMed.