25 Feb 2017

Mouse models of Alzheimer’s disease develop massive amyloidosis without losing many neurons, making them incomplete models of AD. In the February 23 Neuron, researchers led by Bart De Strooper at KU Leuven and Pierre Vanderhaeghen at the Universite Libre de Bruxelles, both in Belgium, debut an AD mouse model that does feature dramatic cell death—of transplanted human neurons. The authors injected human neural precursor cells into newborn APPPS1 mice, whereupon the human cells integrated into the brain circuitry. As the chimeric animals aged, these human cells lost more synapses and accumulated more pathological tau than did neighboring mouse neurons. Eventually, the human cells died off in droves; the mouse neurons did not. The data suggest that human neurons may be particularly vulnerable to the harmful effects of amyloid, the authors note. The model may help researchers dissect mechanisms of neurodegeneration and identify new therapeutic targets, De Strooper suggested. “I think this is the closest you can get to a model for the human disease at the moment,” he told Alzforum. 

De Strooper had introduced these chimeric mice at the 2016 Society for Neuroscience conference in San Diego, where they generated buzz among scientists in the field (see Dec 2016 conference news). “This exciting work [reported in this paper] shows that Aβ accumulation in the brain can induce tau pathology and possibly neurodegeneration in human neurons. It represents another step toward comprehensively recapitulating AD pathology in a model system,” said Doo Yeon Kim at Harvard Medical School, Charlestown, Massachusetts.

Researchers have suggested that for certain applications, such as testing therapeutics, induced human neurons might better model the pathogenesis of AD than transgenic mice (see Oct 2014 news). Researchers led by Kim and Rudolph Tanzi at Harvard Medical School developed a three-dimensional culture model using a human neural cell line overexpressing pathogenic Aβ species (see Oct 2014 news). These cells develop tau pathology, driven by Aβ accumulation, but the cultures do not contain microglial cells or capture the complexity of the brain.

To see how human neurons would react to Aβ in a brain environment, joint first authors Ira Espuny-Camacho and Amaia Arranz differentiated embryonic stem cells from a single healthy donor into cortical precursor cells in vitro, then injected about 100,000 of them into the frontal cortices of immunodeficient newborn mouse pups. The latter were the offspring of APPPS1 animals crossed to NOD-SCID mice, which lack lymphoid cells (see Shultz et al., 1995). Normally, transplanted human cells survive poorly in mouse brain. In this immunodeficient environment, the human cells flourished, expressed mature neuronal markers, and formed synapses with mouse neurons. The cells did not migrate far from the injection site, maintaining a distinct cluster of human neurons in the mouse frontal cortex. Other than the presence of these cells, the chimeric mouse brains appeared normal. In NOD-SCID control mice, transplanted neurons survived into adulthood with no detectable cell loss (see image above and Espuny-Camacho et al., 2013).

The story was different for the APPPS1/NOD-SCID mice. While they maintained all transplanted neurons at two months of age, when plaques have just begun to form, by six months only one-quarter as many human neurons remained in the AD mice as in controls (see image above). Neighboring mouse neurons appeared healthy. Among the surviving human neurons, about one-third displayed signs of necrotic cell death, including swollen, disrupted mitochondria, large vacuoles, and broken nuclear membranes. The authors found no sign of apoptotic markers such as activated caspase.

The authors believe Aβ caused the demise of the human neurons. Amyloid pathology in the transplanted APPPS1/NOD-SCID mice developed in the typical fashion, with the small area containing human cells accumulating the same density of amyloid plaques as the rest of the frontal cortex, and recruiting as many activated microglia and astrocytes. However, in four-month-old APPPS1/NOD-SCID mice, dystrophic neurites were more abundant around plaques in human cell clusters than mouse cell clusters. Plaques near human neurons accumulated clumps of presynaptic and axonal markers, while postsynaptic markers vanished, suggesting synapse loss (see image at left). These accumulations of presynaptic markers have long been seen in AD brain (see Brion et al., 1991). Both mouse and human neurons accumulated hyperphosphorylated tau, but only the human cells contained pathological, misfolded tau, as detected by the MC1 antibody. However, the human neurons did not develop paired helical filaments or neurofibrillary tangles. 

The extensive loss of human neurons in this chimeric model has not been achieved in any other mouse model to date. Commenters called it a significant advance. “This fascinating paper shows, for the first time convincingly in vivo, that human neurons behave differently in response to fibrillar Aβ than do mouse neurons,” Jochen Herms at the German Center for Neurodegenerative Diseases (DZNE), Munich, wrote to Alzforum (see full comment below). An earlier study had suggested primate neurons may be more vulnerable, reporting greater neuron death in aged Rhesus monkeys injected with fibrillar Aβ than is seen in mouse models. However, the possibility remained that other brain cells might contribute to this susceptibility (see Geula et al., 1998; Oct 2014 news). Seeing greater death of human neurons in the same environment where mouse neurons survive strengthens the idea that this is due to an inherent vulnerability of the human/primate cells.

Researchers were also impressed by how well the transplanted human cells modeled the types of tau seen in human brain. Adult human neurons express equal amounts of isoforms containing three and four microtubule-binding domain repeats, while mouse neurons express only 4R tau, and immature human neurons express only 3R. Induced human neurons mature poorly in vitro and tend to express mainly 3R, Kim noted. However, transplanted human neurons expressed high levels of 4R tau by six months of age, and nearly a 1:1 ratio of 3R:4R tau by eight months. “This is a great breakthrough, because now we can use this model not only to study Alzheimer’s, but other tau diseases,” Kim told Alzforum.

All the same, researchers puzzled over the lack of neurofibrillary tangles (NFTs), and suggested these might form if the mice lived longer. NOD-SCID mice are not viable past eight months because they develop thymic tumors. However, De Strooper believes NFTs may only form in these mice if they are seeded with a small amount of tau aggregate. He speculated that since the human neurons in these chimeric mice are packed with misfolded tau, they might be primed for tangle formation. In future work, he will inject the mice with small NFT seeds to see if that triggers the rapid appearance of tangles.

That neurons died in the absence of tangles was surprising, since tangles correlate with cell death in human postmortem studies, noted Robert Vassar at Northwestern University, Chicago. “That was fascinating to me,” Vassar told Alzforum. De Strooper said that this finding fits with the hypothesis that tangles themselves are not the toxic form of tau (see Mar 2013 conference news). To test the idea that tau is toxic, he plans to knock out the protein in the human cells before transplantation and measure whether those neurons survive better in the AD mouse brain.

He will also screen the human neurons for other factors that influence their survival in mice, and to glean clues to the mechanisms of cell death. Commenters cautioned, however, that the necrotic death seen here might not reflect the cell death seen in Alzheimer’s, since many papers report apoptotic markers in AD brain. “We also observe increased neuronal death in our three-dimensional culture model of AD, but we are still determining whether this is similar to cell death observed in human AD patients,” Kim told Alzforum.

The chimeric model has other limitations. Researchers agreed it would not be useful for high-throughput applications such as drug screening. They also wished for a longer-lived model. De Strooper and colleagues are investigating other ways to immunosuppress APPPS1 mice and promote survival of transplants without using NOD-SCID mice.

Nonetheless, researchers were enthusiastic about the potential of the model. Kim suggested that the chimeras will be particularly helpful for analyzing pathogenic mechanisms. De Strooper noted that because the approach uses wild-type human neurons, it models sporadic AD rather than the familial forms represented by most mouse models. Furthermore, researchers could test the effect of various genetic risk variants by editing the genes of the iPS cells before implantation, De Strooper said.

Genetic analysis of the human cells may turn up therapeutic targets with direct translational potential, he hopes. The authors reported the results of a preliminary genetic screen that found global gene expression changes in the transplanted neurons reminiscent of AD brain, including a drop in expression of genes responsible for synaptic transmission, learning, and memory, and a boost in those involved in myelination and cell death (see Zhang et al., 2013). The model might be useful for identifying human disease biomarkers in mouse cerebrospinal fluid, and for validating drug targets, De Strooper suggested.

The chimeric approach is not restricted to human neurons. Next, De Strooper and colleagues will differentiate iPS cells into astrocyte precursor cells and inject those to find out how these glial cells react to amyloid accumulation. Mouse and human astrocytes vary greatly and might be responsible for some species differences (see Jan 2016 news).—Madolyn Bowman Rogers

Comments

1. • Jochen Herms
     Ludwig-Maximilians-Universität Munich
   • Posted: 25 Feb 2017

   It was a pleasure to read this major and excellent paper that documents very elegantly the great potential of human PSCs for the development of better mouse models for AD. I did not expect the technique would be this powerful. This fascinating paper shows, for the first time convincingly in vivo, that human neurons behave differently in response to fibrillary Aß than mouse neurons. It is tempting to speculate that this is due to the expression of 3R/4R tau splice forms in the human neurons, but there could also be other reasons.

   It is also difficult to judge if the necrotic nerve cell degeneration of iPSC neurons within the AD mouse brain is a phenomenon that is indeed responsible for nerve cell loss in patients with AD. From a neuropathological point of view, neuronal necrosis to an extent observed in this study is not a typical feature observed in human AD brains, but rather a finding seen in occasions of more acute damage.

   View all comments by Jochen Herms

Make a Comment

To make a comment you must login or register.

References

News Citations

1. Next-Generation Mouse Models: Tau Knock-ins and Human Chimeras 8 Dec 2016
2. Forget Mice—Are Human Cells Better for Drug Testing? 11 Oct 2014
3. Alzheimer’s in a Dish? Aβ Stokes Tau Pathology in Third Dimension 10 Oct 2014
4. Can Monkeys Model Alzheimer’s Better than Rodents? 17 Oct 2014
5. In Pursuit of Toxic Tau 29 Mar 2013
6. Purification of Adult Human Astrocytes Shows: They Are Unique 15 Jan 2016

Research Models Citations

1. APPPS1

Antibody Citations

1. Tau (MC1)

Paper Citations

1. Shultz LD, Schweitzer PA, Christianson SW, Gott B, Schweitzer IB, Tennent B, McKenna S, Mobraaten L, Rajan TV, Greiner DL. Multiple defects in innate and adaptive immunologic function in NOD/LtSz-scid mice. J Immunol. 1995 Jan 1;154(1):180-91. PubMed.
2. Espuny-Camacho I, Michelsen KA, Gall D, Linaro D, Hasche A, Bonnefont J, Bali C, Orduz D, Bilheu A, Herpoel A, Lambert N, Gaspard N, Péron S, Schiffmann SN, Giugliano M, Gaillard A, Vanderhaeghen P. Pyramidal neurons derived from human pluripotent stem cells integrate efficiently into mouse brain circuits in vivo. Neuron. 2013 Feb 6;77(3):440-56. PubMed.
3. Brion JP, Couck AM, Bruce M, Anderton B, Flament-Durand J. Synaptophysin and chromogranin A immunoreactivities in senile plaques of Alzheimer's disease. Brain Res. 1991 Jan 18;539(1):143-50. PubMed.
4. Geula C, Wu CK, Saroff D, Lorenzo A, Yuan M, Yankner BA. Aging renders the brain vulnerable to amyloid beta-protein neurotoxicity. Nat Med. 1998 Jul;4(7):827-31. PubMed.
5. Zhang B, Gaiteri C, Bodea LG, Wang Z, McElwee J, Podtelezhnikov AA, Zhang C, Xie T, Tran L, Dobrin R, Fluder E, Clurman B, Melquist S, Narayanan M, Suver C, Shah H, Mahajan M, Gillis T, Mysore J, MacDonald ME, Lamb JR, Bennett DA, Molony C, Stone DJ, Gudnason V, Myers AJ, Schadt EE, Neumann H, Zhu J, Emilsson V. Integrated systems approach identifies genetic nodes and networks in late-onset Alzheimer's disease. Cell. 2013 Apr 25;153(3):707-20. PubMed.

Further Reading

News

1. Building Better Mouse Models for Late-Onset Alzheimer’s 20 Jan 2017
2. Knock-In Alzheimer’s Mice Catch on More Broadly in the Field 8 Dec 2016
3. Do APP Knock-ins Call Overexpression Models of AD into Question? 22 Sep 2016
4. Can Monkeys Model Alzheimer’s Better than Rodents? 17 Oct 2014

Webinars

1. Good-Bye Overexpression, Hello APP Knock-in. A Better Model? 11 Apr 2014