Back

CD33-CD45 Interaction Reveals a Mechanistic Link to Alzheimer's Disease Susceptibility

Vo, N.; Rillahan, C. D.; Chatila, Z. K.; Virga, D. M.; Hall, J. L.; Tamucci, K. A.; Rashid, M.; Connor, S. M.; Chintamen, S.; Guzman, G.; Taga, M.; Liu, M.; De Jager, P. L.; St George-Hyslop, P.; Lee, A. J.; Vardarajan, B. N.; Bennett, D. A.; Kohler, J. J.; Schenone, M.; Elyaman, W.; Carr, S. A.; Bradshaw, E. M.

2025-07-31 neuroscience
10.1101/2025.07.28.667311 bioRxiv
Show abstract

The innate immune gene CD33, encoding a myeloid inhibitory sialic acid-binding receptor, is associated with Alzheimers disease (AD) susceptibility. The AD-associated rs3865444CC risk variant reduces splicing of the sialic acid-binding domain and increases expression of the full-length (sialic acid-binding) CD33 isoform seven-fold compared to the rs3865444AA protective genotype. Here we identify CD45 as an immune cell-specific sialic acid-dependent cis CD33 binding partner, whose phosphatase activity is inhibited by CD33. Overexpression of CD33 or loss of CD45 contributes to impaired microglial clearance of amyloid beta and amyloid beta-induced loss of dendritic spines in microglial-neuronal co-cultures, aligning with a detrimental effect of CD33-mediated inhibition of CD45. CD33-CD45 interaction frequency was increased in monocytes from individuals with the rs3865444CC risk variant compared to rs3865444AA, as well as in AD compared to controls, independent of genotype. Furthermore, an interaction between CD33 and PTPRC (encoding CD45) gene expression in human brain tissue was associated with a pathological diagnosis of AD and global burden of AD pathology. Our findings thus establish a functional interaction between CD33 and CD45 relevant to AD susceptibility and systemic myeloid dysfunction in this disease.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.