When Faced with Amyloidosis, Human Transplants Die by Necroptosis

When they find themselves trapped in an amyloid-ridden mouse brain, transplanted human neurons meet an untimely, necrotic demise. While this had been known before, a new-and-improved xenograft model has unveiled some of the dirty details behind this cascade. On September 15 in Science, researchers led by Bart De Strooper of the U.K. Dementia Research Institute, London, reported that, as amyloid accumulates in APP knock-in mice, pathological forms of tau start to amass inside the xenografted human neurons, ultimately forming neurofibrillary tangles.

• Human neurons xenografted into APPNL-G-F mice develop tau pathology, including tangles.
• Half of human neurons died within six months of transplant; likely killed by necroptosis.
• A steep rise in the lncRNA MEG3 might have tripped off the death cascade.

Half of the transplanted neurons die—via a cell death pathway called necroptosis—already during the early stages of amyloid deposition. The researchers tied the mayhem to MEG3—a long noncoding RNA that skyrocketed in the human cells in response to amyloid build-up, setting off the deadly cascade. Putting the kibosh on this particular flavor of cell death spared the human transplants. The findings illuminate an amyloid-instigated form of neurodegeneration that appears specific to human neurons.

“The findings suggest an explanation for why it has been challenging to produce Aβ-dependent tau pathology and neurodegeneration in mouse models,” wrote Daniel Sirkis and Jennifer Yokoyama of the University of California, San Francisco, in an editorial in Science. “The work also provides evidence for the importance of a cell-death pathway in AD that may be therapeuti­cally tractable.”

“The paper takes us a step closer to understanding the mechanism(s) involved in neuronal cell loss in AD,” wrote Domenico Praticò of Temple University in Philadelphia.

Death by Amyloid. Human neural progenitor cells transplanted into a mouse model of amyloidosis ultimately develop tau pathology and succumb to necroptosis, while their murine neighbors survive. [Courtesy of Sirkis and Yokoyama, Science Perspectives, 2023.]

Mouse cells don’t express the isoforms of tau that tangle in human cells, nor do they succumb to neurodegeneration in mouse models of amyloidosis. To model these aspects of AD, researchers have devised ways to incorporate human cells into their studies, for example by using ever-more elaborate three-dimensional co-culture systems of human-stem-cell-derived neurons and glia (Sep 2023 news). Another approach is to transplant human cells into the mouse brain. De Strooper’s group has been honing this chimeric mouse model approach, previously reporting that human neural progenitor cells transplanted into an APP/PS1 transgenic mouse died in apparently gruesome, necrotic fashion (Feb 2017 news). Before they died, the transplants had started to accumulate phosphorylated tau, although full-fledged neurofibrillary tangles never formed. This could be because the APP/PS1 mice used in the experiments were bred on a NOD-SCID background, a highly immunosuppressed strain that succumbs to tumors by 9 months of age. The researchers speculated that had the mice lived longer, perhaps a fuller spectrum of AD pathology would have unfolded.

To find out, first author Sriram Balusu and colleagues used Rag2 knockout mice—a less-immunosuppressed strain that lacks B and T cells but still churns out other types of immune cells and can live up to two years. The scientists also switched to a more physiological model of amyloidosis, crossing the Rag2 knockouts to APP^NL-G-F knock-ins. Shortly after the mice were born, the researchers transplanted 100,000 human cortical neural progenitor cells (NPCs) into the brains of both APP^NL-G-F Rag2 KO mice, as well as into Rag2 KO controls. Two months later, the transplanted cells appeared to have made themselves at home, forming healthy connections with their mouse neural neighbors.

Four months later, once amyloid deposition was underway, the situation was dramatically different. While the transplants still appeared healthy in control mice, roughly half had perished in the APP knock-in hosts. As plaques formed around them, both mouse neurons and the surviving human neurons became “disheveled,” with dystrophic neurites, amidst a hotbed of activated microglia and astrocytes. In the human, but not mouse neurons, aggregates of phosphorylated tau started to accumulate.

By 18 months, amyloidosis had overtaken the brain, and the surviving human neurons contained full-fledged neurofibrillary tangles of tau. Relative to xenografted control mice, or to ungrafted APP-KI mice, the APP-KI mice hosting human neurons had elevated p-tau181 and p-tau231 in their plasma. This suggested that, when faced with amyloidosis, the human neurons developed tau pathology and also churned out phosphorylated tau, akin to neurons in people with AD.

Amyloid Triggers Tau. Only when exposed to amyloid (blue, bottom), did transplanted human neurons (green) express markers of pathological tau (red). [Courtesy of Balusu et al., Science, 2023.]

To search for clues about how the xenografted neurons were degenerating, the scientists performed transcriptomics at two, six, and 18 months after transplantation. As amyloidosis developed, the transplanted neurons assumed a gene-expression profile that resembled that reported in postmortem brain samples from people with AD. This included an uptick in expression of genes involved in MAPK signaling, TNF-α and interferon signaling, cell proliferation, aging, tissue regeneration, and myelination.

Curiously, the amyloid-exposed human neurons also shifted into an immature state, ramping up expression of genes involved in cell cycle re-entry, and turning down mature neuronal pathways. This hypomature phenotype had also been reported in human AD brain samples (Jun 2021 news).

Another finding stood out in the transcriptomes of amyloid-exposed human neurons. MLKL, the so-called executor of necroptosis, was a dramatically upregulated. Immunostaining the mouse brains with necroptosis markers p-RIPK1, p-RIPK3, and p-MLKL revealed that, in 18-month-olds, the death cascade was in full swing in transplanted human neurons but not in their murine neighbors.

Specifically, the researchers spotted the necroptotic protein trio huddled into punctate, vesicular structures called necrosomes. In some cases, casein kinase 1 δ, a marker of granulovacuolar degeneration bodies, mingled within these structures. GVBs are vesicles of autolysosomal origin loaded with granules of densely packed proteins. They have been tied to tau and α-synuclein toxicity (Wiersma and Scheper, 2020; Jorge-Oliva et al., 2022). De Strooper and colleagues previously found these structures in neurons from AD brain samples, where they are thought to reflect a last-ditch effort to survive by sequestering the death proteins (Jan 2020 news).

Wiep Scheper of Amsterdam University Medical Center in the Netherlands, whose work has explored the link between GVBs and tau pathology, called the findings from the xenograft model impressive. However, she noted that the partial colocalization between CK1δ and necrosome structures in the study appeared to be different from GVBs in the human brain, and that more work is needed to confirm that the punctate structures observed in the transplanted human neurons are bona fide GVBs. That can be done by staining for a wider array of GVB core as well as lysosomal membrane markers, and in more cells, she said.

Notably, researchers spotted the GVB-like structures in some neurons with, and some without, tau tangles. To Balusu, this supports the idea that tangles are not the toxic form of tau pathology. He believes other types of tau aggregate, such as oligomers, might be involved in setting off GVD. This idea jibes with new findings from Scheper’s lab, which tied GVBs to shorter tau filaments within the neuronal soma (Jorge-Oliva et al., 2023). 

Finally, the researchers hunted for culprits that may have instigated the necroptotic cascade in the human transplants. They zeroed in on MEG3, a long noncoding RNA that was ramped up by a full order of magnitude in the amyloid-exposed human, but not mouse, neurons, and two- to threefold in human AD brain samples. This lncRNA has been implicated in cell death pathways and neurodegeneration (Jiang et al., 2021; Chanda et al., 2018). Overexpressing MEG3 in cultured human neurons hastened their death by necroptosis, a fate that could be avoided by treating the cells with necroptosis inhibitors.

Strikingly, counteracting the necroptosis cascade in the human transplanted neurons, either by knocking down their expression of MEG3 or RIPK-3 prior to xenografting, or by treating the mice with necroptosis inhibitors, significantly improved their survival within the amyloid-ravaged hosts.

Scheper looks forward to future studies with the model that might disentangle the relationship between tau pathology and necroptosis. In particular, she asked, does blocking necroptosis also get rid of tau pathology, and/or GVB-like structures? Sirkis and Yokoyama made a similar point in their editorial: “Another open question is whether the improved neuronal survival that results from reduction of MEG3 expression occurs by modulating the emergence of tau pathology, or despite such pathology.”

Taking a step even further back, Praticò noted that the question of whether tau pathology even triggers the MEG3-necroptotic cascade remains unsettled.—Jessica Shugart