Amyloid β-responsive cells as a therapy in Alzheimer’s Disease

Introduction: Alzheimer’s Disease (AD) afflicts over 6.5 million Americans and has an estimated national cost of $321 billion annually. Biochemically, AD is characterized by the aggregation of amyloid-beta (Aβ) into plaques and tau into neurofibrillary tangles. Historically, Aβ accumulation has been posited as the major driver of disease and monoclonal antibodies (mAbs) targeting Aβ have been tested extensively in clinical trial; two (aducanumab, 2021 and lecanemab, 2022) have received accelerated FDA approval for AD. However, these therapies are associated with significant and potentially fatal side effects and, despite their ability to bind and clear Aβ from the brain, they have not demonstrated clear cognitive benefit. Additional factors of disease, such as chronic inflammation mediated by astrocytes and microglia, have more recently been posited as major drivers of disease. Thus, new treatment approaches, that do not rely solely on the clearance of Aβ protein aggregates, are needed. Here, we present a potential cell therapy for AD. Astrocytes are engineered to express a synthetic receptor that recognizes Aβ and allows them to respond in a therapeutic manner via expression of a programmed transgene. Critically, this transgene does not have to target Aβ. Instead, transgenes that address neuronal death or neuroinflammation can be selected. The use of this system results in cells that recognize Aβ as a marker of disease and that respond via autoregulated expression of a programmed therapeutic transgene.

Materials and

Methods: We leverage synthetic Notch (synNotch) to program cells. Building on the Notch regulatory core, we exchanged Notch domains for synthetic components: the extracellular domain for a single-chain variable fragment (scFv) of an anti-Aβ mAb and the intracellular domain for a synthetic transcription factor. Recognition of Aβ results in expression of programmed transgenes (Fig 1A). We built a receptor panel by repurposing clinically tested Aβ mAbs (gantenerumab, donanemab, and bapineuzumab). All three receptors were expressed in mouse fibroblasts by lentivirus transduction for receptor characterization; cells expressed luciferase and mCherry reporter transgenes in response to Aβ. Receptor expression was measured by flow cytometry probing for a c-myc epitope tag on the receptor. SynNotch cells were plated on synthetic Aβ42 or Aβ40, and luciferase and mCherry levels were analyzed by luminescence and fluorescence microscopy. We then expressed high-performing receptors (Gant-Notch, Bap-Notch) in human iPSC-derived astrocytes and treated them with human brain-derived Aβ seeds; mCherry was again analyzed by microscopy. Bap-Notch astrocytes were then programmed to express therapeutic transgenes in response to Aβ. Aβ-driven expression of the neuroprotective transgene brain-derived neurotrophic factor (BDNF) was measured by ELISA when Bap-Notch cells were plated on synthetic Aβ42. Finally, Bap-Notch astrocytes were programmed to express inflammatory cytokine antagonists soluble TNF receptor (sTNFr) and IL-1Ra in response to Aβ. A reactive astrocyte phenotype was induced in these Bap-Notch astrocytes by dosing with TNFα and IL-1α, and the effect of this cytokine treatment, as well as of the Aβ-dependent expression of sTNFr and IL-1Ra, were measured by qPCR.

Results, Conclusions, and Discussions: Aβ receptors constructed from gantenerumab, donanemab, and bapineuzumab (Gant-Notch, Don-Notch, Bap-Notch) recognize immobilized synthetic Aβ42 and Aβ40 based on expression of luciferase transgene (Fig 1B ). However, there are differences in receptor performance, with Gant-Notch and Bap-Notch resulting in higher luciferase expression than Don-Notch when cells are plated on Aβ42. All three receptors also recognize the shorter and less pathogenic Aβ40, though with lower fold changes overall. Bap-Notch also recognizes Aβ42 supplemented in the culture medium, likely due to the purported specificity of bapineuzumab for Aβ oligomers and motivating the selection of Bap-Notch as the lead candidate. As anticipated, mCherry expression correlates with luciferase expression.

Astrocytes programmed with the Bap-Notch receptor recognize human brain-derived Aβ, demonstrated by mCherry fluorescence (Fig 1C). Bap-Notch astrocytes programmed with a BDNF transgene produce BDNF upon recognition of immobilized synthetic Aβ42 (Fig 1D). BDNF is a promising therapeutic transgene to be used in a synNotch system, as it has shown great promise in repairing neuronal injury in non-human primate models; however, it is associated with adverse effects when delivered brain-wide. Its use in our synNotch platform would restrict its expression to areas of local Aβ deposition.

Treatment of the Bap-Notch astrocytes with TNFα and IL-1α resulted in a reactive astrocyte phenotype, as evidenced by increases in the expression of reactive genes Il6, CSF2, and SERPINA3 (Fig 1E ). This phenotype is attenuated by Aβ-dependent sTNFr and IL-1Ra expression in Bap-Notch astrocytes. When Bap-Notch cells are instead programmed to express a control SEAP reporter in response to Aβ, there is no attenuation in reactive gene expression.

Aβ-synNotch receptors represent a promising new approach to treat AD. Capitalizing on their native surveillance functions, astrocytes that express these receptors specifically recognize Aβ as a marker of disease and respond via direct upregulation of programmed therapeutic transgenes. SynNotch astrocytes thereby provide a means to localize disease-dependent production of biologic drugs that may be deleterious in the context of constitutive gene therapies. Future studies will investigate Aβ-synNotch in AD mouse models.

Acknowledgements (Optional):