Dissecting the structural organization of multiprotein amyloid aggregates using a bottom-up approach.

Deposition of fibrillar amyloid β (Aβ) in senile plaques is a pathological signature of Alzheimer's disease. However, senile plaques also contain many other components, including a range of different proteins. Although the composition of the plaques can be analyzed in post mortem tissue, knowledge of the molecular details of these multiprotein inclusions and their assembly processes is limited, which impedes the progress in deciphering the biochemical mechanisms associated with Aβ pathology. We here describe a bottom-up approach to monitor how proteins from human cerebrospinal fluid associate with Aβ amyloid fibrils to form plaque particles. The method combines flow cytometry and mass spectrometry proteomics and allowed us to identify and quantify 128 components of the captured multiprotein aggregates. The results provide insights in the functional characteristics of the sequestered proteins and reveal distinct interactome responses for the two investigated Aβ variants, Aβ(1-40) and Aβ(1-42). Furthermore, the quantitative data is used to build models of the structural organization of the multiprotein aggregates, which suggests that Aβ is not the primary binding target for all the proteins; secondary interactions account for the majority of the assembled components. The study elucidates how different proteins are recruited into senile plaques and establishes a new model system for exploring the pathological mechanisms of Alzheimer's disease from a molecular perspective.

Mass spectrometry. Protein concentrations were determined using a BCA kit from Pierce. Aggregates isolated from serum samples were analyzed as follows (some minor changes were made for the second batch of samples as indicated by the information within parentheses). Protein solutions were supplemented with 300 mM (100 mM) ammonium bicarbonate (Sigma-Aldrich), pH 8 to obtain 60 µL (40 µL) aliquots and 20 µL (40 µL) of 0.4% (0.2%) ProteaseMAX (Promega) in 200 mM (50 mM) ammonium bicarbonate/40% (20%) acetonitrile (Fischer Scientific) was added for incubation at 50 °C for 30 min while shaking at 350 rpm. The samples were then sonicated for 10 min in ultrasonic water bath.
Proteins were reduced with 2 µL of 200 mM DTT (Sigma) and incubated at 56 °C for 30 min. Alkylation was performed with 2 µL of 600 mM iodoacetamide (Sigma) at room temperature for 30 min in dark. Then 0.3 µg (0.1 µg) of sequencing grade modified trypsin (Promega) was added to each sample (1:20 trypsin:protein), followed by incubation at 37 °C for 16 h. The digestion was stopped by adding formic acid at final concentration of 5%. Then the samples were cleaned on a C18 StageTip (Thermo Scientific) and dried using a speedvac (MiVac; Thermo Scientific). Samples were dissolved in 20 µL (15 µL) of 2% AcN/0.1% formic acid and 5 µL (6 µL) volumes were injected on a 50 cm long C18 Easy-column (Thermo Scientific) connected to an Ultimate 3000 UPLC system (EASY-nLC1000 system, both from Thermo Scientific). Peptides were eluted in a 60 min (120 min) gradient at a flow rate of 400 nL/min (300 nL/min) connected to a Q Exactive HF Orbitrap mass spectrometer (Thermo Scientific). The gradient went from 4-26% (2-26%) of buffer B (2% acetonitrile, 0.1% formic acid) in 60 min (90 min), and up to 95% of buffer B in 5 min and the effluent was electrosprayed into the mass spectrometer directly from the column. The survey MS spectrum was acquired at the resolution of 120,000 (140,000) at m/z 200 in the range of m/z 300-1650. MS/MS data of the 16 most intensive precursors were obtained with a higher-energy S4 collisional dissociation (HCD) at 26% normalized collision energy (NCE) for ions with charge z>1 at a resolution of 30,000 (17,500).
Aggregates isolated from CSF samples were analyzed using TMT-10plex labeling as follows. Relatively mild extraction conditions were used to avoid overload the samples with Aβ from the added fibrils. Aliquots of protein solutions (1 µg proteins) were reduced with 16 µL of 200 mM DTT (Sigma) and incubated at 55 °C for 45 min. Alkylation was performed with 16 µL of 600 mM iodoacetamide (Sigma) at room temperature for 30 min in dark. Then 0.2 µg of sequencing grade modified trypsin (Promega) was added to each sample (1:5 trypsin:protein), followed by incubation at 37 °C for 16 h. The digestion was stopped by adding formic acid at final concentration of 5%. Then the samples were cleaned on a C18 HyperSep plate (Thermo Scientific) and dried using a speedvac (MiVac; Thermo Scientific). Samples were dissolved in 35 µL of 50 mM triethylammonium-bicarbonate (Sigma-Aldrich) pH 8 and 4 µg TMT-10plex reagents (Thermo Scientific), dissolved the in 15 µL of dry acetonitrile, were added. Samples were scrambled and incubated for 2 h at room temperature at 450 rpm. The reaction was stopped by adding hydroxylamine at final concentration of 0.5% and then incubated for 15 min at room temperature with 450 rpm shaking. Individual samples were combined to one analytical sample and dried in speedvac (MiVac; Thermo Scientific), followed by cleaned up on C18 Stage Tips and dried again in speedvac. Chromatographic separations of peptides were performed on a 50 cm long C18 EASY-spray column connected to an EASY-nLC1000 system (Thermo Fisher Scientific). Peptides were eluted in a 120 min gradient at a flow rate of 300 nL/min connected to a Fusion Orbitrap mass spectrometer (Thermo Scientific). The gradient went from 2-26% of buffer B (2% acetonitrile, 0.1% formic acid) in 110 min, up to 35% of buffer B in 10 min and up to 95% of buffer B in 5 min and the effluent was electrosprayed into the mass spectrometer directly from the column. The survey MS spectrum was acquired at the resolution of 120,000 (at m/z 200) in the range of m/z 375-1500. MS/MS data of the top 15 most intensive precursors were obtained with a HCD at 35% NCE for ions with charge z>1 at a resolution of 60,000 and first mass at m/z 100.

MS data analysis.
The raw data files were converted to Mascot Generic Format (mgf) using the in-house written Raw2mgf program. Proteins were identified by searching mgf files against the SwissProt database (HUMAN) using Mascot v 2.5.1 (MatrixScience Ltd., UK) database search engine. For the qualitative analysis of serum samples, the list of hits was filtered to remove all entries with only a single peptide identified and all keratins (contaminations). Then a threshold was set for each sample to achieve a false discovery rate (FDR) of less than 3%.
Data with the TMT-labeled samples were analyzed on Proteome Discoverer v2.2 (Thermo Scientific) using Mascot v2.5.1 (MatrixScience Ltd., UK) database search engine. Keratins (contamination) were removed. Isoforms the same proteins were grouped and the 'Master protein' entry were kept when redundant. The abundance was set to zero for proteins that were not detected in a specific sample. As we observed a large variation in the total amount of proteins in the replicate samples we normalized the abundances using the abundance of Aβ (APP) in each sample. This approach is justified by the observation that there is a linear S5 correlation (R 2 = 0.83) between the abundance of Aβ (APP) and the average protein abundance for the samples (Fig. S6A). The normalization implicated that samples without detected Aβ were not included in the analysis and the final set of samples were: seven of Aβ  in control CSF, six of Aβ  in control CSF, five of Aβ  in AD CSF, and six of Aβ  in AD CSF. We also note that the Aβ (APP) abundance detected by MS does not correlate with the Aβ content of the CSF samples measured before out experiments (Fig. S6B).
Data analysis. P-values were calculated using Student's T-Test (using the T.TEST function in Microsoft Excel). For each protein, the differences in abundances of AD and control CSF samples were analyzed in terms of relative changes calculated as log 2 (Abundance in AD/Abundance in control). GO annotations for the identified proteins were extracted from UniProt database (January 2019). The amino acid sequences from the Uniprot entries were used to predict the pI (from Proteome Discoverer v2.2), charge at neutral pH (Expasy ProtParam), GRAVY (Expasy ProtParam), Intrinsic solubility at pH 7 (CamSol 3 ), and the propensity of amyloid aggregation (TANGO 4-5 , pH 7, 37 °C, ionic strength = 0.02 M, concentration = 1 M). Sequences longer than the maximum allowed for TANGO (500 amino acids) were analyzed in fragments (at least three different ways) and summarized to get the values for the whole protein. Pairwise correlations between the protein abundances were computed using Matlab R2014b (MathWorks). All samples (14 for each Aβ variant) were included in this analysis and the abundances were not normalized. STRING analysis 6 was performed using the web interface (string-db.org, v11.0, Aug 27, 2019). Figure S1. Formation of Aβ amyloid fibrils during 48 h incubation at 37 °C in sodium phosphate buffer (pH 7.4). A) ThT fluorescence intensities at 487 nm measured for the samples before and after fibril formation, respectively. B) CD spectra of the samples after incubation. C, D) Representative AFM images of the formed Aβ(1-40) (C) and Aβ(1-42) (D) fibrils, respectively.   S9 Figure S4. STRING protein-protein interaction network analysis with colour-coding according to the layer model for Aβ . APP is shown in magenta. Figure S5. STRING protein-protein interaction network analysis with colour-coding according to the layer model for Aβ . APP is shown in magenta.  The results originate from one sample per method and composition and is the same data as presented in Figure 3 in the main text.         Table S9. Positioning of the identified proteins in Figure 4E. The list is sorted from highest to lowest RMS of the log 2 (AD/Control) values for Aβ     Table S11. GO: Biological process statistics for the layers of the proposed multilayer model (n-values correspond to the numbers given in figure 6B). Only the highest ranked processes in each layer are listed.  Table S12. GO: Molecular function statistics for the layers of the proposed multilayer model (n-values correspond to the numbers given in figure 6B). Only the highest ranked processes in each layer are listed.   figure 6B). Only the highest ranked processes in each layer are listed.