A new APP knock-in mouse has hit the scene. Researchers led by Pascal Sanchez, Denali Therapeutics, San Francisco, California, described the details in a paper posted to bioRxiv on January 20. They humanized the Aβ sequence in the mouse APP gene and knocked in three familial Alzheimer’s disease mutations: Swedish, Arctic, and Austrian. TREM2 expression and cytokine concentrations spiked in brain tissue, microglia flocked to plaques, neurites swelled, and tau and neurofilament light collected in their cerebrospinal fluid. Notably, microglia surrounding plaques seemed distressed—the cells filled with lipids and revved up transcription of genes previously linked to a subset of microglia isolated from AD brain tissue. Denali, in collaboration with the Jackson Laboratory, will make this model open access for academics and companies alike.

  • Mice have humanized APP and Swedish, Arctic, Austrian mutations.
  • Microglial transcriptome resembles human AD profile.
  • Stressed microglia accumulate Aβ and lipids.

In 2014, researchers led by Takaomi Saido, RIKEN Brain Science Institute, Wako, Japan, had already created two APP knock-in mice: NL-F mice carrying the Swedish (KM670/671NL) and Iberian (I716F) mutations, and NL-G-F animals with the additional Arctic (E693G) mutation. Saido’s knock-ins make more Aβ than do wild-type mice, accumulate plaques in the brain, have obvious gliosis, and perform poorly in cognitive tests (April 2014 webinarSaito et al., 2014). 

Model Strategy. While healthy microglia protect neurons (bottom, left), those in SAA mice (top) become phagocytic (bottom, middle). Plaque-associated microglia swell with Aβ plaques and lipids, unable to cope with the high phagocytic load (bottom, right). [Courtesy of Pascal Sanchez, 2021.]

Alas, obtaining those mice for study proved difficult for many labs outside of Japan, so Denali scientists decided to make their own. Co-first authors Dan Xia, Steve Lianoglou, and colleagues used homologous recombination to humanize the Aβ sequence of the mouse APP gene, incorporating three FAD APP mutations, Swedish, Arctic, and Austrian (T714I), rather than the Iberian mutation used in the NL-G-F model. “This new knock-in model is conceptually identical to our NL-G-F,” Saido wrote to Alzforum. “The Swedish mutations increase the total amount of Aβ; the Arctic mutation makes Aβ prone to oligomerization and resistant to degradation; and the Austrian or Iberian mutation increases the ratio of Aβ42/40.” (Full comment below.)

Denali, in collaboration with the Jackson Laboratory, Bar Harbor, Maine, promises to allow unrestricted use of the SAA mice by academic and industry researchers alike. Mike Sasner from Jax noted the struggle company researchers especially often face to obtain models. “While the Jackson Lab has a lot of APP models, legal restrictions imposed by donating investigators prevent us from distributing them to companies,” he explained. “To get around this, companies tend to make their own mouse models, but keep them in house without sharing them.”

Christian Haass, German Center for Neurodegenerative Diseases (DZNE), Munich, Germany, agreed. “Model availability is a big issue, considering that many experiments were blocked or slowed due to legal issues,” he wrote to Alzforum (see comment below). “Making this mouse model freely available is a great sign for our research community.”

As expected, in the new homozygous SAA mice, the Aβ42/40 ratio in brain tissue extracts, CSF, and plasma rose higher than in wild-type controls in animals as young as 2 months of age, two months before plaques were detectable. Beginning when mice were 4 months old, plaques spread, starting in the cortex/hippocampus, moving to the entorhinal region, then spilling into the amygdala, thalamus, and striatum among other regions, similar to the phases of Aβ deposition see in the human brain (Thal et al., 2002). Plaques packed the mouse cortex and hippocampus and Aβ deposited in blood vessel walls of the meninges (see image below).

Amyloid on the Brain. Whole-brain heatmaps light up plaque distribution in 8-month-old SAA knock-in mice (left). Immunostaining revealed Aβ deposition in blood vessels of the meninges (arrows, right). [Courtesy of Xia et al., bioRxiv, 2021.]

By 8 months, the brain was packed with plaques, and neurites swelled with phosphorylated tau detected by the AT-8 antibody. Neurofilament light (NfL), and the lysosomal marker LAMP1 ticked up—all characteristic of AD. These changes correlated with increased total tau and NfL in the CSF. TREM2 expression and cytokine concentrations spiked in brain tissue. Microglia also flocked to plaques.

Were these microglia different from those in wild-type mice? Expression of more than 600 genes was different in microglia from SAA mice compared to wild-type. Transcripts related to cholesterol metabolism, glycolysis, and phagocytic/lysosomal function ticked up in the knock-ins. Disease-associated microglia (DAM) genes that were previously linked to pathology in amyloid models were particularly upregulated (Jun 2017 news). 

What about microglia that swarmed around plaques? The researchers labeled microglia with methoxy-X04, a fluorescent dye that binds to Aβ fibrils, then sorted fibril-positive from fibril-negative cells. Those that had swallowed Aβ fibrils altered expression of more than 800 genes, including those of the DAM and plaque-induced gene (PIG) variety. PIGs were found when researchers in Bart De Strooper’s lab in KU Leuven, Belgium, carried out a spatial transcriptomics study in brain tissue from NL-G-F mice and in postmortem tissue taken from people who had had AD. They found a suite of genes that were up- or down-regulated in microglia adjacent to plaques (July 2020 news). 

Some of the differentially expressed genes in the SAA phagocytic microglia are orthologs of those in a sub-cluster of microglial transcripts found in people with AD (May 2019 news). In SAA mice, expression of neurotrophic genes, such as those involved in neuronal development, axon guidance, and spine morphogenesis, was reduced (see image below). Overall, the microglia surrounding plaques in SAA mice had a transcriptional signature that tracked with those in other mouse models and in the human brain.

DAM Microglia! In SAA mice, plaque-associated microglia expressed higher amounts of disease-associated microglia (DAM) genes (red) and less neurotrophic genes (blue) than non-plaque-associated microglia. [Courtesy of Xia et al., bioRxiv, 2021.]

Plaque-associated microglia also highly expressed genes involved in lipid clearance and metabolism. What did this have to do with their function? Despite the ramp-up, lipids, such as ganglioside GM3, filled the cells, hinting at lysosomal dysfunction. They also accumulated spermine, a molecular distress signal that cells release when they are overworked. “Highly phagocytic, distressed microglia had difficulty coping with getting rid of so much debris, which is reflected in their mishandled lipid metabolism,” suggested Sanchez. “We are now investigating whether this metabolic dysregulation will lead to microglia dysfunction.”

Sasner wants to cross these SAA mice with different genetic backgrounds to create new combinations of alleles, such as with APOE4 and humanized tau. “Because the SAA model does not have tau pathology, we are hoping that crossing it with a humanized tau model will give us both phenotypes,” he told Alzforum.

Initial mouse distribution is estimated for August 2021. Interested researchers can pre-order on the Jackson Laboratory website.—Chelsea Weidman Burke

Comments

  1. This is a mouse model for amyloidosis, which, finally, is freely available for the entire research community. This is a big issue, considering that many experiments were blocked or slowed due to legal issues with mouse models. I think this is a great sign for our research community to freely interact and to finally stop the very unproductive or even counterproductive Materials Transfer Agreement business. After all, our mission is primarily to provide the best data and to translate them as soon as possible to preclinical and clinical research. Many thanks to the scientists at Denali for making this possible!

    We are using this model to further test the preclinical properties of our TREM2 modulating antibody 4D9.

    There are as always still open questions. Due to the unique combination of three mutations, this model has its limitations. In vitro data suggest that the Arctic mutation used in this model could drive Aβ38 production, which is a bit puzzling, since that species may not be deposited and is thought be produced at the expense of Aβ42. One may also have to consider that some antibodies could fail to recognize amyloid deposits, due to the potential structural changes caused by that mutation. 

  2. Xia et al. present a new APP knock-in (KI) mouse model for Aβ amyloidosis that expresses humanized Aβ with three familial, early onset AD mutations and recapitulates the spatial patterning of plaque deposition observed in humans under physiological conditions. The homozygous KI mice accumulate appreciable parenchymal plaques and leptomeningeal vascular deposits (CAA) after 8 months of age, which makes this model practical enough for use in basic and preclinical studies. 

    I applaud Denali for making this model accessible to all—something the AD research community should be trying to reconcile for existing animal models, and striving toward in the release of future ones. One consideration, however, is that the E22G Aβ (E693G APP) “Arctic” mutation will produce a distinct conformational variant of Aβ in these mice that does not reflect the predominant conformation(s) formed by wild-type Aβ in sporadic, late-onset AD patients. 

    Using fluorescent structure-sensitive dyes in brain samples from human subjects bearing the very rare E22G mutation, we have shown that E22G Aβ deposits have a unique conformation that does not overlap much with wild-type Aβ deposits in sporadic AD; similarly, synthetic E22G Aβ fibrils formed in vitro exhibit a distinct dye-emission signature compared to wild-type (Condello et al., 2018). 

    Moreover, the Aβ isoform composition of plaques is distinguishable in E22G mutation carriers compared to wild-type Aβ in sporadic AD. For example, the Aβ38 isoform is not commonly found in amyloid deposits, but in intra-Aβ mutation carriers (e.g., also in E22Q “Dutch” and D23N “Iowa” cases) this seems to be a hallmark feature (Moro et al., 2012). Notably, using an inoculation paradigm in susceptible wild-type Aβ mice, we demonstrated that injecting synthetic E22G Aβ fibrils or patient brain extract from an E22G mutation carrier induced a distinct plaque conformation with increased Aβ38 deposition (Condello et al., 2018; Watts et al., 2014), which argues that the conformation of E22G deposits more efficiently template the Aβ38 isoform. 

    This new KI model also bears the “Austrian” mutation near a γ-secretase cleavage site, and familial mutations in this region have been reported to increase Aβ38 production (Suárez-Calvet et al., 2013). While it remains unclear how the Aβ38 isoform contributes to pathogenicity, it would be interesting to know if this new KI model also exhibits such molecular features. Interestingly (at least based on the images displayed in the preprint), Xia et al. show Aβ plaques that appear more filamentous or diffuse in shape, resembling the morphology described for the E22G familial AD human cases (Kalimo et al., 2013). 

    As the evidence grows for Aβ conformational heterogeneity within and between AD patients and etiologies (Cohen et al., 2015; Qiang et al., 2017; Rasmussen et al., 2017; Condello et al., 2018), there is an emerging hypothesis that distinct conformational variants may underlie different phenotypic manifestations of AD. So, perhaps we should no longer generalize that a plaque is a plaque—it’s not all the same, at least at the molecular level. Thus, it seems reasonable to question if the plaques composed of mutant Aβ in this new KI model are representative of those in sporadic AD and lead to similar consequent pathobiological pathways found in mouse models or humans producing wild-type Aβ. 

    Beyond the histological validation of several plaque-associated features such as microglial activation and dystrophic neurites, Xia et al. present RNA-sequencing data from sorted microglia, and show that there is a set of differentially expressed disease-associated (DAM) and plaque-induced (PIG) microglial genes that partially overlaps with DEGs reported in the popular 5xFAD overexpression model. This supports the notion that there is a generalized microglial transcriptional programming toward extracellular and/or phagocytosed Aβ.

    It would be fascinating to know if the non-overlapping microglial DEGs in this new APP KI model manifest because of unique signaling induced by mutant Aβ, or the failure to degrade it. Curiously, histological studies have observed a muted plaque-associated microglial phenotype in E22G familial AD brain samples compared to sporadic AD (Kalimo et al., 2013). Because we now appreciate that microglia are heterogeneous cells (Stratoulias et al., 2019) that elicit specialized responses to different pathologies (Friedman et al., 2018), it is plausible that distinct microglia function (and dysfunction) occurs because of Aβ plaque variants. If true, this has great implications for the discovery and development of targeted molecular therapies.

    References:

    . Structural heterogeneity and intersubject variability of Aβ in familial and sporadic Alzheimer's disease. Proc Natl Acad Sci U S A. 2018 Jan 23;115(4):E782-E791. Epub 2018 Jan 8 PubMed.

    . APP mutations in the Aβ coding region are associated with abundant cerebral deposition of Aβ38. Acta Neuropathol. 2012 Dec;124(6):809-21. PubMed.

    . Structural heterogeneity and intersubject variability of Aβ in familial and sporadic Alzheimer's disease. Proc Natl Acad Sci U S A. 2018 Jan 23;115(4):E782-E791. Epub 2018 Jan 8 PubMed.

    . Serial propagation of distinct strains of Aβ prions from Alzheimer's disease patients. Proc Natl Acad Sci U S A. 2014 Jul 15;111(28):10323-8. Epub 2014 Jun 30 PubMed.

    . Autosomal-dominant Alzheimer's disease mutations at the same codon of Amyloid Precursor Protein differentially alter Aβ production. J Neurochem. 2013 Oct 11; PubMed.

    . The Arctic AβPP mutation leads to Alzheimer's disease pathology with highly variable topographic deposition of differentially truncated Aβ. Acta Neuropathol Commun. 2013 Sep 10;1:60. PubMed.

    . Rapidly progressive Alzheimer's disease features distinct structures of amyloid-β. Brain. 2015 Apr;138(Pt 4):1009-22. Epub 2015 Feb 15 PubMed.

    . Structural variation in amyloid-β fibrils from Alzheimer's disease clinical subtypes. Nature. 2017 Jan 12;541(7636):217-221. Epub 2017 Jan 4 PubMed.

    . Amyloid polymorphisms constitute distinct clouds of conformational variants in different etiological subtypes of Alzheimer's disease. Proc Natl Acad Sci U S A. 2017 Dec 5;114(49):13018-13023. Epub 2017 Nov 20 PubMed.

    . Structural heterogeneity and intersubject variability of Aβ in familial and sporadic Alzheimer's disease. Proc Natl Acad Sci U S A. 2018 Jan 23;115(4):E782-E791. Epub 2018 Jan 8 PubMed.

    . The Arctic AβPP mutation leads to Alzheimer's disease pathology with highly variable topographic deposition of differentially truncated Aβ. Acta Neuropathol Commun. 2013 Sep 10;1:60. PubMed.

    . Microglial subtypes: diversity within the microglial community. EMBO J. 2019 Sep 2;38(17):e101997. Epub 2019 Aug 2 PubMed.

    . Diverse Brain Myeloid Expression Profiles Reveal Distinct Microglial Activation States and Aspects of Alzheimer's Disease Not Evident in Mouse Models. Cell Rep. 2018 Jan 16;22(3):832-847. PubMed.

  3. The new knock-in mice described by Xia and colleagues carry Swedish (NL), Arctic (G) and Austrian (I) mutations (NL-G-I mice) are conceptually identical to our mice that carry Swedish, Arctic, and Beyreuther/Iberian (F) mutations (NL-G-F mice) (Saito et al., 2014). 

    The Swedish mutations increase the total amount of Aβ; the Arctic mutation renders Aβ oligomerization-prone and resistant to degradation by the major Aβ-degrading enzyme, neprilysin (Iwata et al., 2001; Tsubuki et al., 2003); the Austrian or Beyreuther/Iberian mutation increases the ratio of Aβ42/Aβ40 production. As a result, their pathological phenotypes, including microglial responses, appear similar. So many papers have already been published using our mice, and, for instance, the following two appear very important: Chen et al., 2020, and Sobue et al., 2021). 

    The mutant mice, patented in Japan, the United States, and EU, are available to industry and pharma after a license contract has been signed. More than 10 companies worldwide are using our mice. The license fee depends on the users’ financial situation. Here are some advantages of using our knock-in mice, although researchers are free to use any model.

    1. Our knock-in mice are available from RIKEN BioResource Center, a Japanese counterpart of Jackson Laboratory: RIKEN BioResource Research Center.

    2. Approximately 500 groups worldwide use our knock-in mice as we started distribution of the mice before publication, so experiments performed by different researchers can be compared in a relatively unbiased manner. Search PubMed for “Saido T” to identify published papers after 2014.

    3. Other line-ups are also available: NL and NL-F lines. The NL-F line shows less aggressive pathology than NL-G-F line but can be used to analyze how Aβ is anabolized and catabolized without the interference of the Arctic mutation.

    4. Human tau (hMAPT) knock-in mice in which the entire murine MAPT gene is humanized (Hashimoto et al., 2019; Saito et al., 2019), and double knock-in mice, i.e., NL-G-F X hMAPT and NL-F X hMAPT, are also from RIKEN BRC.

    5. Wild-type humanized Aβ mice and G-F mice will soon be made available. G-F mice, which accumulate pathological Arctic Aβ, can be used to characterize β-secretase and β-secretase inhibitors without the interference of the Swedish mutation.

    References:

    . Single App knock-in mouse models of Alzheimer's disease. Nat Neurosci. 2014 May;17(5):661-3. Epub 2014 Apr 13 PubMed.

    . Metabolic regulation of brain Abeta by neprilysin. Science. 2001 May 25;292(5521):1550-2. PubMed.

    . Dutch, Flemish, Italian, and Arctic mutations of APP and resistance of Abeta to physiologically relevant proteolytic degradation. Lancet. 2003 Jun 7;361(9373):1957-8. PubMed.

    . Dutch, Flemish, Italian, and Arctic mutations of APP and resistance of Abeta to physiologically relevant proteolytic degradation. Lancet. 2003 Jun 7;361(9373):1957-8. PubMed.

    . Spatial Transcriptomics and In Situ Sequencing to Study Alzheimer's Disease. Cell. 2020 Aug 20;182(4):976-991.e19. Epub 2020 Jul 22 PubMed.

    . Microglial gene signature reveals loss of homeostatic microglia associated with neurodegeneration of Alzheimer's disease. Acta Neuropathol Commun. 2021 Jan 5;9(1):1. PubMed.

    . Tau binding protein CAPON induces tau aggregation and neurodegeneration. Nat Commun. 2019 Jun 3;10(1):2394. PubMed.

    . Humanization of the entire murine Mapt gene provides a murine model of pathological human tau propagation. J Biol Chem. 2019 Aug 23;294(34):12754-12765. Epub 2019 Jul 4 PubMed.

Make a Comment

To make a comment you must login or register.

References

Research Models Citations

  1. APP NL-F Knock-in
  2. APP NL-G-F Knock-in
  3. AppSAA Knock-in

Mutations Citations

  1. APP K670_M671delinsNL (Swedish)
  2. APP I716F (Iberian)
  3. APP E693G (Arctic)
  4. APP T714I (Austrian)

Webinar Citations

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

News Citations

  1. Hot DAM: Specific Microglia Engulf Plaques
  2. Paper Alert: Those PIGs! Spatial Transcriptomics Add Human Data
  3. When It Comes to Alzheimer’s Disease, Do Human Microglia Even Give a DAM?

Paper Citations

  1. . Single App knock-in mouse models of Alzheimer's disease. Nat Neurosci. 2014 May;17(5):661-3. Epub 2014 Apr 13 PubMed.
  2. . Phases of A beta-deposition in the human brain and its relevance for the development of AD. Neurology. 2002 Jun 25;58(12):1791-800. PubMed.

Other Citations

  1. AT-8

External Citations

  1. Jackson Laboratory website

Further Reading

Papers

  1. . Microglial gene signature reveals loss of homeostatic microglia associated with neurodegeneration of Alzheimer's disease. Acta Neuropathol Commun. 2021 Jan 5;9(1):1. PubMed.

Primary Papers

  1. . Fibrillar Aβ causes profound microglial metabolic perturbations in a novel APP knock-in mouse model. bioRxiv. January 20, 2021.
  2. . Novel App knock-in mouse model shows key features of amyloid pathology and reveals profound metabolic dysregulation of microglia. Mol Neurodegener. 2022 Jun 11;17(1):41. PubMed.