Assessing and modulating microglial activation states in Alzheimer's disease tissues and models
William J Meilandt, Karpagam Srinivasan, Tiffany Wu, Martin Weber, Hai Ngu, Kimberly Stark, Amy Easton, Brad A Friedman, David V Hansen
Depts. Neuroscience, Bioinformatics, and Pathology, Genentech, Inc., South San Francisco, CA 94080 USA
Our recent meta-analysis of microglial transcriptional profiles identified numerous modules of co-regulated genes and elevated numbers of microglia, distinct from so-called DAM or MGnD-expressing cells, that expressed the interferon-related module in two models of Alzheimer's disease (AD). Moreover, expression of pro-inflammatory and peripheral immune cell modules is more apparent in human AD tissues than in typical AD models. Human genetics support the notion that impaired myeloid cell activation contributes to the risk of late-onset Alzheimer's disease (AD). Mutations that impair the function of the activating receptor TREM2 increase AD risk, while a polymorphism that reduces the level of the inhibitory receptor CD33 reduces AD risk. In the PS2APP model, we have observed elevated amyloid plaque burden in aged Trem2-null mice. To learn whether deletion or inhibition of various inhibitory receptors could have the opposite effect, we used three approaches. 1) In AD models, microglial PD-1 expression is enhanced, as is PD-L1 expression in microglia and astrocytes. Low-dose, infrequent anti-PD-1 treatment was reported to reduce amyloid pathology in the 5xFAD model at advanced stages (Baruch 2016); however, we and others have been unable to replicate that finding (Latta-Mahieu 2017). Taking a different approach, we used our own anti-PD1 antibody, which is 50 times more potent than the one tested by others, and we began weekly high-dose treatments at an early age. 2) In AD models, microglial expression of PILRA ligands (COLEC12, PIANP) is enhanced. We tested whether Pilra deletion could enhance microglial activation and plaque clearance in PS2APP mice. 3) We tested whether Cd33 deletion could reduce plaque burden in PS2APP mice since CD33 is a known AD risk gene and plaque reduction has been reported in Cd33-null APP/PS1 mice (Griciuc 2013). Of the three approaches, only Cd33 deletion reduced amyloid pathology, which surprised us since murine Cd33 does not contain the inhibitory ITIM motif and is not a clear orthologue of human CD33. Reasons and mechanisms for these observations will be discussed.