A MICROGLIAL ACTIVITY STATE BIOMARKER PANEL DIFFERENTIATES FTD-GRANULIN AND AD FROM CONTROL CASES

Abstract

SummaryO_ST_ABSBackgroundC_ST_ABSWith the emergence of microglia-modulating therapies there is an urgent need for reliable biomarkers to evaluate microglial activation states. MethodsUsing mouse models and human induced pluripotent stem cell-derived microglia (hiMGL), which were genetically modified to yield the most opposite homeostatic (TREM2- knockout) and disease-associated (GRN-knockout) states, we identified microglia activity-dependent markers. Non-targeted mass spectrometry was used to identify changes in microglial and cerebrospinal (CSF) proteome of Grn- and Trem2-knockout mice. Additionally, we analyzed the proteome of GRN- and TREM2-knockout hiMGL and their conditioned media. Candidate marker proteins were tested in two independent patient cohorts, the ALLFTD cohort with 11 GRN mutation carriers and 12 non-carriers, as well as the proteomic data set available from the European Medical Information Framework Alzheimers Disease Multimodal Biomarker Discovery (EMIF-AD MBD). FindingsWe identified proteomic changes between the opposite activation states in mouse microglia and cerebrospinal fluid (CSF), as well as in hiMGL cell lysates and conditioned media. For further verification, we analyzed the CSF proteome of heterozygous GRN mutation carriers suffering from frontotemporal dementia (FTD). We identified a panel of six proteins (FABP3, MDH1, GDI1, CAPG, CD44, GPNMB) as potential indicators for microglial activation. Moreover, we confirmed three of these proteins (FABP3, GDI1, MDH1) to be significantly elevated in the CSF of AD patients. In AD, these markers differentiated amyloid-positive cases with mild cognitive impairment (MCI) from amyloid-negative individuals. InterpretationThe identified candidate proteins reflect microglia activity and may be relevant for monitoring the microglial response in clinical practice and clinical trials modulating microglial activity and amyloid deposition. Moreover, the finding that three of these markers differentiate amyloid-positive from amyloid-negative MCI cases in the AD cohort suggests that these marker proteins associate with a very early immune response to seeded amyloid. This is consistent with our previous findings in the DIAN (Dominantly Inherited Alzheimers Disease Network) cohort, where soluble TREM2 increases as early as 21 years before symptom onset. Moreover, in mouse models for amyloidogenesis, seeding of amyloid is limited by physiologically active microglia further supporting their early protective role. The biological functions of some of our main candidates (FABP3, CD44, GPNMB) also further emphasize that lipid dysmetabolism may be a common feature of neurodegenerative disorders. FundingThis work was supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germanys Excellence Strategy within the framework of the Munich Cluster for Systems Neurology (EXC 2145 SyNergy - ID 390857198 to CH, SFL and DP) and a Koselleck Project HA1737/16-1 (to CH).