The analysis of protein dynamics or turnover in patients has the potential to reveal altered protein recycling such as in Alzheimer disease, and to provide informative data regarding drug efficacy, or certain biological processes. The observed protein dynamics in a solid tissue or a fluid is the net result of protein synthesis and degradation, but also transport across biological compartments. We report a first model to simultaneously model protein dynamics observed in blood plasma and the cerebrospinal fluid (CSF) taking into account transport. We applied this model to 65 proteins of a single individual displaying similar or very different dynamics in plasma and CSF. The model structure indicates an active control at the blood-CSF barrier. This type of model has the potential to reveal altered transport or barriers resulting from disease in future studies.

1µL of plasma and 150µL of CSF were depleted with depletion columns (High Select™ Depletion Spin Columns, A36370, ThermoFisher). The filtrate was collected and evaporate to dryness on SpeedVac (50 °C). Samples were reconstituted with 20 µL Ammonium Bicarbonate (ABC) 100 mM, 1% SDS and transferred on Eppendorf™ twin.tec™ 96-Well (30129300). Samples were reduced, alkylated and digested with autoSP3 protocol (Muller et al. 2020 doi: 10.15252/msb.20199111). On AssayMap BRAVO (Agilent), SP3 protocol version 1.0.2 was used. Proteins were reduced with 5µL of Dithiothreitol 80mM during 1800s at 60°C. Then, there were alkylated with 5µL of Iodoacetamide 200mM during 1800s at 30°C. A 50/50 mix of Sera-Mag stock solution A and B was generated at 100 mg/mL and 5µL were added to the sample. 35µL of acetonitrile were added and sample were incubated 1080s. After this incubation, beads were wash two times with 200µL of 80% EtOH and one time with 180µL of acetonitrile. Proteins were digested by adding 35µL of ABC 100 mM, 5µL of Trypsin/LysC (0.05 µg/µL, Promega), and incubate overnight at 37°C well closed with sealing foil. Digestion was stopped with addition of 10µL 5% formic acid. Generated peptides were fractionated on C18 tips (AssayMAP 5 µL Reversed Phase (RP-S) cartridges, G5496-60033, Agilent T) at basic pH. 50µL of 200mM ammonium formate pH10 were added to samples and “Fractionation V2.0” was runned on AssayMap BRAVO (Agilent T.). Briefly, cartridges were primed with 100µL of 90% acetonitrile, equilibrated with 50µL 20mM ammonium formate pH10, before sample loading. Cartridges were washed with 50µL 20mM ammonium formate pH10 before sequential elutions with 35µL. To the CSF, 5 fractions were generated: 15%, 20%, 25%, 30% and 80% acetonitrile in 200mM ammonium formate pH10. To the plasma samples, 4 fractions were generated: 15%, 20%, 25% and 80% acetonitrile in 200mM ammonium formate pH10. In this condition, the fractions at 15% and 80% were mixed. Fractions were diluted with 0.1% formic acid and load on evotip following the manufacturer procedure. LC-MS acquisitions were performed on Evosep One using 8cm x 150µm, 1.5µm (EV1109, Evosep) with 60SPD method coupled to TIMS TOF HT (Bruker Daltonics) through a captive spray ion source. Ion source parameters were 1500V on capillary with 3.0L/min at 180°C for the drying gas. DDA-PASEF method was used in positive ion mode. MS1 range was 100-1700 m/z. TIMS settings were 1/K0 0.75-1.25, Ramp and Accumulation time of 100ms. At MS2 level, 10 PASEF Ramps were perform per cycle of 1.17s. Plasma fractions were analyzed in duplicate. Data Acquisition were in-live submitted and interrogate inside the Paser Box (Bruker D.). Uniprot databse from 2021 was used with human as the only taxonomy. Contaminants were added during the database indexation on Paser Box server. CID mode was selected to the fragmentation with monoisotopic precision at precursor and fragment level. Mass tolerance was 20ppm at precursor level and 30ppm et fragment level. Precursor mass range was between 600 and 6000 Da. Proteins were digested with trypsin with strict specificity and 2 maximum missed cleavages. Peptide minimal length was of 6 amino-acids with maximum of 2 potential variable modifications as deamidation (N, Q) and oxidation (M). Carbamidomethylation was used as fixed modification of cysteine. XCorr was used as primary score, Zscore in the secondary score, and TIMScore was used. Minimum of 1 peptide identified per protein is require. False discovery rate of maximum 1% was calculated at protein level. Identification results was export with mzIdent files (mzid and mgf files). The files were used to import Peptide Search in skyline. Peptides and precursor ions in the library were upload on skyline file. To the data importation, retention time tolerance was 1.5 minutes on MS/MS scan ID, 0.2 on ion mobility value coming from the experimental library, 3 isotopes at resolution 60000 at MS1 level. Isotope modification was added (13)C6-leucine.

Samples were generated following the clinical protocol “In Vivo Alzheimer Proteomics (PROMARA)” (ClinicalTrials Identifier: NCT02263235), which was authorized by the French ethical committee CPP Sud-Méditerranée IV (#2011-003926-28) and by the ANSM agency (#121457A-11). We applied the ethically approved (see above) original SILK 13C6-Leu infusion protocol (Bateman et al., 2006). Briefly, 13C6-Leu prepared per the European Pharmacopeia [19] was intravenously administered. After a 10 min initial bolus at 2 mg/kg, an 8h50 infusion at 2 mg kg/h was performed. Ventricular CSF or plasma EDTA samples were collected starting at the beginning of the 13C6-leucine infusion, roughly every 3h (3 to 6 mL). Samples were transported to the laboratory at 4°C, and centrifuged at 2000g for 10 minutes. CSF and plasma samples was aliquoted into 1.5-mL polypropylene tubes and stored at –80°C until further analysis.