How do neurons die in Alzheimer’s disease? Perhaps by necroptosis, a form of programmed cell death triggered by inflammation. In the December Acta Neuropathologica, researchers led by Dietmar Thal, Bart De Strooper, and Sriram Balusu at KU Leuven, Belgium, reported finding necrosomes, complexes of activated necroptotic proteins, inside granulovacuolar bodies in the Alzheimer’s brain. These bodies consist of large cytoplasmic vacuoles that contain dense granular material. Such vacuoles are common in AD. They develop early on in hippocampal pyramidal neurons and eventually appear in connected brain areas. Necrosomes inside these vacuoles correlated with tau pathology and neuronal loss, but not with amyloid plaques, the authors report, implicating granulovacuolar degeneration (GVD) and necroptosis in neuronal death in AD.

  • Activated necrosomes hide out in granulovacuolar bodies in AD brain.
  • The complexes associate with tau toxicity and neuron loss.
  • Causal relationships? TBD.

“The demonstration of necrosomes in GVD in AD brains is interesting and new, and opens up lines of research and potentially new therapeutic targets,” Steve Finkbeiner at the University of California, San Francisco, wrote to Alzforum (full comment below). Shijun Zhang at Virginia Commonwealth University in Richmond said the paper makes a valuable contribution. “The analyses are solid and convincing … the study adds further evidence to support the possible involvement of necroptosis as one of the mechanisms underlying neuronal degeneration in AD,” he wrote (full comment below).

Junying Yuan and colleagues at Harvard Medical School first described necroptosis 15 years ago (Jun 2005 news). Like apoptosis, the process is activated by tumor necrosis factor receptors, but it has different downstream mediators. First the TNF receptor phosphorylates receptor-interacting serine/threonine protein kinase 1 (RIPK1), switching it on. RIPK1 then binds and activates RIPK3, forming the necrosome. When this complex recruits mixed-lineage kinase-like pseudokinase (MLKL), which oligomerizes and punches holes in the cell membrane, the cell dies.

Tau and Necroptosis. Necroptotic proteins (green) occur together with phosphorylated tau (red) in some neurons (arrow; overlay looks yellow), but not in others (arrowhead). Nuclei are blue. [Courtesy of Koper et al., Acta Neuropathologica.]

Necroptosis previously has been linked to AD. Salvatore Oddo and colleagues at Arizona State University, Tempe, reported elevated levels of necroptotic proteins in AD brain, which correlated with tau tangles, brain atrophy, and cognitive decline (Jul 2017 news). 

De Strooper and colleagues became interested in necroptosis when they found it in human neurons they had transplanted into the brains of APPPS1 mice (Feb 2017 news). To learn if necroptosis figures in AD pathology, they turned to postmortem brain.

First author Marta Koper immunostained sections of brain tissue from 23 people who had had AD, from 24 who had pathologically confirmed preclinical AD, and from 16 healthy controls. Antibodies detected phosphorylated, activated versions of RIPK1, RIPK3, and MLKL in all of the AD brains, in 22 of the preclinical AD brains, and in two control samples. Activated necrosomes were only inside granulovacuolar bodies in neurons.

These vacuoles are bound by a double membrane and are believed to represent malfunctioning late-stage autophagosomes (Funk et al., 2011). They were first described in AD brain in 1911, and granulovacuolar degeneration is considered a characteristic of the disease (Kahn et al., 1985Okamoto et al., 1991; Kurdi et al., 2016). 

Koper and colleagues found necroptotic GVD bodies (GVDn+) scattered throughout all regions of AD brain they examined: hippocampus, entorhinal cortex, hypothalamus, amygdala, temporal cortex, and frontal cortex. They were most numerous in the hippocampus; sparse in frontal cortex.

Because granulovacuolar degeneration begins in the hippocampus, the authors homed in on this region. They found no spatial relationship between GVDn+ and plaques, but a close relationship with tangles. About 40 percent of GVDn+ neurons contained phosphorylated tau (see image above) and in a regression analysis, the extent of GVD correlated better with the Braak neurofibrillary tangle stage than with the amount of plaque or clinical status.

Regions with GVDn+ tended to have fewer neurons, indicating degeneration. In the hippocampus, the proportion of GVDn+ neurons was none in control brain, 10 percent in preclinical brain, and 42 percent in AD. Neuronal density tracked in tandem, reaching 135 cells per square millimeter in control hippocampus, 109 in preclinical AD, and 81 in AD hippocampus. Granulovacuolar degeneration spreads to the frontal cortex late in disease. In this region, the authors counted no difference between preclinical and control, but in AD, 2 percent of frontal cortex neurons harbored necrosome-positive granulovacuoles and there were 99 neurons per square millimeter, compared with 113 in control cortex.

Do GVDn+ and neurofibrillary tangles underlie neuronal death? How do these two pathologies relate to each other? Tangles could induce granulovacuolar bodies. GVD occurs in mouse models of tauopathy, and seeding of tau tangles triggers GVD in cultured neurons and mouse models (Lewis et al., 2001; Köhler et al., 2014; Wiersma et al., 2019). However, the relationship could be indirect, as well. “There could be two parallel pathways that cause tau phosphorylation and RIPK1 phosphorylation,” De Strooper wrote to Alzforum.

Another puzzle is why necrosomes appear only inside GVD vacuoles. Perhaps these vacuoles sequester the necrosomes, preventing them from reaching the cell membrane and in that way prolong the life of the neuron. “We need to develop cellular and animal models to investigate causal relationships and molecular mechanisms,” noted De Strooper, who plans to investigate the effect of GVDn+ in chimeric mice. Finkbeiner suggested following single neurons that contain necrosome-positive vacuoles to see whether the pathology shortens or lengthens their lifespans.

Beyond GVD bodies, cells may have other ways of fending off necroptosis. A recent paper characterized two families with autosomal-dominant mutations in RIPK1. Even though the mutations rendered the protein constitutively active, predictably upping inflammation and cell death in blood, fibroblasts from carriers had low RIPK1 expression and resisted necroptosis (Tao et al., 2019, and related comments). “The compensation … may point to potential mechanisms for modulating cell-death speed, possibly in neurodegenerative disorders as well,” Thal suggested.

Does necroptosis play a role in glial cells? It’s unclear. The authors saw activated necroptotic proteins only in neurons, but Oddo, and later Veronique Miron’s group at the University of Edinburgh, did report active necrosome components in microglia (Jun 2019 news). Microglia might interact with necroptotic neurons, Miron noted. In the present study, microglia appear with necroptosing neurons in preclinical AD tissue. This would position the microglia to respond to damage wrought by necroptosis, which could, in turn, modify their behavior, she wrote (full comment below).

If necroptosis pushes vulnerable neurons toward destruction in AD, inhibiting this process could stave off disease progression. RIPK1 inhibitors are being developed. Partners Denali/Sanofi are testing DNL747 in AD and ALS, and DNL758 in auto-immune disease, and Rigel Pharmaceuticals started Phase 1 for its RIPK1 inhibitor R552 (see PRNewswire). GlaxoSmithKline ran eight Phase 1 and 2 trials of GSK2982772 in various inflammatory disorders, but removed this RIPK1 inhibitor from its clinical pipeline.—Madolyn Bowman Rogers

Comments

  1. I think the demonstration of necrosomes’ positive granulovacuolar degeneration (GVDn+) in AD brains is interesting and new, and opens up new lines of research and potentially new therapeutic targets.

    However, I think the mechanistic relationship between neurofibrillary tangles (NFTs) and GVDn+ should be investigated further. There is an inherent fundamental limitation with positive correlations based on snapshot data, such as neuropathology. I agree that a positive correlation does suggest that a relationship exists. But as we first showed many years ago for a different disease, it is very difficult to know the true nature of that relationship from snapshots.

    It is conventional to interpret positive correlations as pathogenic responses/mechanisms because it is intuitive. But coping mechanisms can actually be positively correlated, too, if the cellular abnormality is a response to stress and part of some mechanism to stave off cell death. The only way we could distinguish positive correlations due to pathogenic mechanisms from coping responses was to invent a method to follow single cells longitudinally and then quantify the predictive relationship of cytopathology for neurodegeneration using the same tools that are used for clinical trials.

    Mind you, this is no criticism of the paper. As I mentioned, it’s an inherent limitation in all data produced using snapshot approaches versus longitudinal single-cell analyses. I dare say that it was likely the positive correlation between amyloid plaques and AD and the interpretation of that relationship that led to so much investment by the pharmaceutical industry into therapies that target amyloid. The results of those clinical trials would suggest that we need to at least be aware of the inherent limitation of inferring biological mechanisms from snapshots and suggest a measure of caution.

    It would be exciting to follow up a study like this with longitudinal single-cell analysis of NFT formation and its relationship to GVDn+ to bring additional clarity.

  2. This was a very nice study of AD patients and control brain samples, showing the presence of pRIPK1, pRIPK3, and pMLKL in granulovacuolar degeneration (GVD) of neuronal cells. The analyses are solid and convincing. Although differences exist with regard to the co-localization of RIPK1 and RIPK3, as well as the localization of pMLKL, compared with a previous study the findings add further evidence that necroptosis is involved in neuronal degeneration in AD (Caccamo et al., 2017). 

    Although activated necrosomes were demonstrated to be present in GVD of neuronal cells, the role of necroptosis in neurodegeneration still needs to be validated. The ongoing clinical studies of highly selective RIPK1 inhibitors in AD may provide clinical evidence to support the pathological roles of necroptosis.

    References:

    . Necroptosis activation in Alzheimer's disease. Nat Neurosci. 2017 Sep;20(9):1236-1246. Epub 2017 Jul 24 PubMed.

  3. This study raises three major questions: Do neurons undergo a delayed slow necroptosis in AD? If so, how do microglia respond? And do microglia undergo necroptosis in AD?

    The authors demonstrate co-localization of activated necrosome components within cytoplasmic vacuolar lesions (GVD lesions) in neurons, and hypothesize that the sequestration of these components could be protective because it would prevent their pore-forming activity at the plasma membrane. Although the density of GVD neurons testing positive for phosphorylated MLKL, a necrosome marker, is higher in areas of increased neuronal loss in AD brains, it’s not clear whether these neurons will eventually die, or if they are detected because they’ve protected themselves from necroptosis. It would be interesting to follow up by tracking neuronal behavior over time in experimental models that manifest GVD lesions.

    These findings also raise the interesting question as to how microglia respond to necroptosing neurons, and whether this differs if the necroptosis is delayed or protracted. Interestingly, in this study, microglia appear to associate with necroptosing neurons in tissue from people with preclinical AD. This would position microglia to respond to the DAMPs, aka damage-associated molecular patterns, which are robustly released through necroptosis, which could in turn modify microglial behavior. Necroptosis has been associated with both pro-inflammatory and anti-inflammatory responses in different contexts, yet how microglia respond to necroptosis in AD, and how this impacts pathology, remain to be investigated.

    As for necroptosis in microglia, Koper et al. did not detect activated necroptosis component expression in microglia in AD, however, a previous study found that almost 30 percent of p-MLKL+ cells in AD brain were positive for the microglial marker Iba1+ (Caccamo et al., 2017). This raises the question as to why there is such a discrepancy between the studies, and how prevalent microglia necroptosis is in AD. If microglia do indeed undergo necroptosis in AD, does this lead to repopulation of the cells? Microglia repopulation following their death (experimental or occurring following injury) is robust, and can change their activation profile, which influences neural cells in their vicinity. For instance, we’ve shown that microglia undergo necroptosis following acute demyelination of white matter, followed by repopulation associated with pro-regenerative function (Lloyd et al., 2019). However, experimental microglial depletion can lead to repopulation of a neurodegenerative phenotype in select areas of gray matter (Rubino et al., 2018). Whether microglial necroptosis and repopulation are important regulators of AD pathology is unknown, but would be of paramount interest to the AD community.

    References:

    . Necroptosis activation in Alzheimer's disease. Nat Neurosci. 2017 Sep;20(9):1236-1246. Epub 2017 Jul 24 PubMed.

    . Central nervous system regeneration is driven by microglia necroptosis and repopulation. Nat Neurosci. 2019 Jul;22(7):1046-1052. Epub 2019 Jun 10 PubMed.

    . Acute microglia ablation induces neurodegeneration in the somatosensory system. Nat Commun. 2018 Nov 1;9(1):4578. PubMed.

  4. This extensive and well-performed study by Koper et al. shows an interesting novel feature of GVBs, as their core is immunopositive for necrosomes. As the authors and commentators indicate, caution is warranted with mechanistic interpretation based on observations in neuropathological tissue, which are indeed “snapshots.”

    Mechanistic insight may be facilitated by the neuronal GVB cell model recently developed in my lab (Wiersma et al., 2019). Using this model we could go beyond correlations and clearly establish causality between intracellular tau aggregation and GVB formation.

    In addition, this allowed a better characterization of GVBs, and we demonstrated that (contrary to what is stated in the commentary above) GVBs are single-membrane, proteolytically active lysosomes. The presence of necrosomes in GVBs further stresses the importance to investigate whether GVBs are a protective or degenerative response in tau pathogenesis (Wiersma and Scheper, 2019).

    References:

    . Granulovacuolar degeneration bodies are neuron-selective lysosomal structures induced by intracellular tau pathology. Acta Neuropathol. 2019 Dec;138(6):943-970. Epub 2019 Aug 27 PubMed.

    . Granulovacuolar degeneration bodies: red alert for neurons with MAPT/tau pathology. Autophagy. 2020 Jan;16(1):173-175. Epub 2019 Oct 23 PubMed.

Make a Comment

To make a comment you must login or register.

References

News Citations

  1. A New Program for Cell Death: Necroptosis Premiering in a Neuron Near You
  2. Necroptosis Rampant in the Alzheimer’s Brain?
  3. Chimeric AD Model Shows Human Neurons Are Uniquely Vulnerable to Aβ
  4. Dead Microglia Pave the Way for Myelin Regeneration

Research Models Citations

  1. APPPS1

Paper Citations

  1. . Granulovacuolar degeneration (GVD) bodies of Alzheimer's disease (AD) resemble late-stage autophagic organelles. Neuropathol Appl Neurobiol. 2011 Apr;37(3):295-306. PubMed.
  2. . Immunohistological study of granulovacuolar degeneration using monoclonal antibodies to neurofilaments. J Neurol Neurosurg Psychiatry. 1985 Sep;48(9):924-6. PubMed.
  3. . Reexamination of granulovacuolar degeneration. Acta Neuropathol. 1991;82(5):340-5. PubMed.
  4. . Granulovacuolar Degeneration in Hippocampus of Neurodegenerative Diseases: Quantitative Study. J Neurodegener Dis. 2016;2016:6163186. Epub 2016 Oct 23 PubMed.
  5. . Enhanced neurofibrillary degeneration in transgenic mice expressing mutant tau and APP. Science. 2001 Aug 24;293(5534):1487-91. PubMed.
  6. . Granulovacuolar degeneration and unfolded protein response in mouse models of tauopathy and Aβ amyloidosis. Neurobiol Dis. 2014 Nov;71:169-79. Epub 2014 Jul 27 PubMed.
  7. . Granulovacuolar degeneration bodies are neuron-selective lysosomal structures induced by intracellular tau pathology. Acta Neuropathol. 2019 Dec;138(6):943-970. Epub 2019 Aug 27 PubMed.
  8. . A dominant autoinflammatory disease caused by non-cleavable variants of RIPK1. Nature. 2019 Dec 11; PubMed.

External Citations

  1. PRNewswire
  2. clinical pipeline

Further Reading

Primary Papers

  1. . Necrosome complex detected in granulovacuolar degeneration is associated with neuronal loss in Alzheimer's disease. Acta Neuropathol. 2019 Dec 4; PubMed.