Multiplex in‐vivo quantification of tau by mass spectrometry: Method optimization for comprehensive detection of tau proteoforms and biomarker identification

Abstract Background Neurodegenerative diseases diagnosis is still challenging, especially at the early stages. The diagnosis of Alzheimer’s disease (AD) relies on the identification of tau proteoforms (i.e., phosphorylation) in the cerebrospinal fluid (CSF) and plasma. However, pathophysiological fluid biomarkers are still lacking for non‐AD tauopathies. Multiplex liquid chromatography (LC)‐mass spectrometry (MS/MS) offers the opportunity to quantify the vast diversity of tau proteoforms (3R and 4R isoforms, truncations, phosphorylations, acetylations). The current gold standard for tau extraction before LC‐MS/MS analysis of very low concentrations in biofluids relies on immunoprecipitation (IP) approaches. Nonetheless, IP may miss some proteoforms due to its high selectivity. Here, we tested anti‐body free and IP approaches and compared their ability to detect tau proteoforms with LC‐MS/MS. Method We analyzed samples of artificial CSF (recombinant tau protein in diluted serum), and human CSF (from AD and non‐AD patients). Fully 15N‐labeled tau was added as an internal standard to the samples. Regarding IP, we first assessed a panel of commercial tau antibodies (in line with the literature) coated on magnetic beads. For the antibody‐free approach, we performed perchloric acid precipitation, followed by micro‐Solid Phase Extraction (Pc‐µSPE) using Oasis hydrophilic‐lipophilic balance sorbent. Purified proteins were digested into peptides using trypsin before multiplex quantitative analysis on reversed‐phase capillary LC, coupled with a qExactive high‐resolution mass spectrometer operated in the Parallel Reaction Monitoring (PRM) acquisition mode. Result IP optimization led to a combination of antibodies targeting the central and N‐ter domains of tau. IP and Pc‐µSPE comparison in artificial CSF showed a higher sensitivity of IP: limits of detection decreased up to a factor 5. In the human CSF, Pc‐µSPE allowed for better detection of peptides from the N‐ter to the C‐ter than IP, and a higher signal for tau proteoforms (e.g., 3R and 4R isoforms and phosphorylated tau species). Conclusion Despite better performances when dealing with canonical tau, IP was overcome by the antibody‐free sample preparation when identifying tau proteoforms with LC‐MS/MS in human CSF. The protocol is readily applicable to samples of clinical cohorts. This result opens the possibility of better detecting tau proteoforms and identifying pathophysiological biomarkers for non‐AD tauopathies.