An optimized quantitative proteomics method establishes the cell type‐resolved mouse brain secretome

Abstract To understand how cells communicate in the nervous system, it is essential to define their secretome, which is challenging for primary cells because of large cell numbers being required. Here, we miniaturized secretome analysis by developing the “high‐performance secretome protein enrichment with click sugars” (hiSPECS) method. To demonstrate its broad utility, hiSPECS was used to identify the secretory response of brain slices upon LPS‐induced neuroinflammation and to establish the cell type‐resolved mouse brain secretome resource using primary astrocytes, microglia, neurons, and oligodendrocytes. This resource allowed mapping the cellular origin of CSF proteins and revealed that an unexpectedly high number of secreted proteins in vitro and in vivo are proteolytically cleaved membrane protein ectodomains. Two examples are neuronally secreted ADAM22 and CD200, which we identified as substrates of the Alzheimer‐linked protease BACE1. hiSPECS and the brain secretome resource can be widely exploited to systematically study protein secretion and brain function and to identify cell type‐specific biomarkers for CNS diseases.

A Comparison of the novel hiSPECS (green) and the previous SPECS (blue) protocol (Kuhn et al, 2012). hiSPECS uses lectin-based glycoprotein enrichment followed by covalent binding to magnetic beads which improves sample processing. On-bead tryptic digestion is followed by mass spectrometry analysis and label-free quantification (LFQ) using data-independent or data-dependent acquisition (DIA vs DDA). In contrast, SPECS depends on biotinylation of the azide-functionalized glycoproteins, followed by streptavidin pull-down, SDS-gel-based fractionation into 14 gel slices, and DDA analysis only. B Coomassie-stained gel showing the prominent reduction in albumin as a result of glycoprotein enrichment with ConA. Left lane: secretome (conditioned medium) of primary neurons before glycoprotein enrichment. 10% of the volume was loaded that was used for the ConA enrichment. Right lane: eluate after glycoprotein enrichment using ConA beads. The albumin band is highlighted with a red arrow. C A representative peptide density plot of a neuronal secretome using the hiSPECS method and analyzed in DDA mode. The m/z ratio is plotted against the retention time. One MS1 scan (purple) is followed by 20 MS2 scans covering a range of 300-1,400 m/z with an overlap of 1 m/z between adjoining m/z windows. The m/z windows were adjusted to achieve equal numbers of peptides. . E Venn diagram illustrating in which sources the proteins of (D) were found to contain a glycosylation site. As expected for the glyco-secretome, more than 85% of the quantified secretome proteins were annotated as glycoproteins. F Representative Pearson correlations of log 2 -transformed protein LFQ intensities of four biological replicates (BR) processed with either the SPECS or hiSPECS method.
In the previous SPECS studies, sample pairs were separated on the same gel to achieve high reproducibility, whereas the correlation of biological replicates run on different gels was rather low. Thus, protein LFQ ratios of the individual replicates were used for statistical evaluation.  A Heat map of the top 50 differentially secreted proteins (Bonferroni's P.adj < 0.05 using the R package Limma (Ritchie et al, 2015)) across the four cell types from hierarchical clustering. For the missing protein quantification data, an imputation approach was undertaken using data missing at random within a left-shifted Gaussian distribution by 1.8 standard deviation. The rows represent the differentially secreted proteins, and the columns represent the cell types with their replicates. The colors represent log-scaled protein levels with blue indicating the lowest, white indicating intermediate, and red indicating the highest protein levels. B Functional annotation clustering with DAVID 6.8 (da Huang et al, 2009a; da Huang et al, 2009b) for gene ontology term biological process (FAT) of the cell typespecific secretome proteins (Table EV4). All proteins detected in the hiSPECS brain secretome study have been chosen as the background. The dot sizes indicate the enrichment score.

Specifically secreted from neurons
Specifically secreted from astrocytes Specifically secreted from microglia % enrichment relative to max. Secretome Lysate Figure EV4. Protein levels in the brain cell secretome vs lysate proteome.
Comparison of the iSPECS secretome resource and lysate data by Sharma et al (2015). The % enrichment is indicated normalized to the average of the most abundant cell type. For example, APLP1 is similarly abundant in lysates of neurons and oligodendrocytes, but only secreted to a relevant extent from neurons. Thus, APLP1 is classified as a cell type-specifically secreted protein. Horizontal lines indicate the mean. ▸ Figure EV5. Substrate candidate identification of the Alzheimer's protease BACE1 as proof of principle of the hiSPECS method.
A The same experiment as in Fig 5A and B, but using DDA instead of DIA for mass spectrometric data acquisition. The volcano plot shows changes in protein levels in the secretome of primary cultured neurons (hip + ctx) upon BACE1 inhibitor C3 treatment using the hiSPECS DDA method. The negative log 10 -transformed P-value (two-sample t-test) of each protein is plotted against its log 2 fold change comparing inhibitor-treated and control condition (N = 11). The gray hyperbolic curves depict a permutation-based false discovery rate estimation (P = 0.05; s0 = 0.1). Significantly regulated proteins (P < 0.05) are indicated with a dark blue dot, known BACE1 substrates are indicated with blue letters, and unknown ones are indicated with black letters. The two newly validated BACE1 substrates CD200 and ADAM22 are indicated in red. SEZ6 and IL6ST are not depicted in the volcano plot because they were only identified in the control condition. B Correlation of the hiSPECS DDA to DIA method. Plotted are the log 2 fold changes between C3 treatment and DMSO control samples (N = 11). The two newly validated BACE1 substrates CD200 and ADAM22 are indicated in red, known BACE1 substrates in blue, and unknown ones in black. C The volcano plot (details described in (A)) shows changes in protein levels in the secretome of primary cultured neurons isolated either from the hippocampus (hip) or from cerebral cortex (ctx) using the hiSPECS DIA method. Proteins exclusively detected in one group are indicated. D Correlation analysis of the average log 2 LFQ values of the secretome proteins of (C). E Venn diagram indicating the overlap of quantified proteins in the secretome of hippocampal vs cortical neurons. Proteins were considered robustly quantified if detected with at least 5 of 6 biological replicates. F The same experiment as (A), however, using only hippocampal neurons and hiSPECS DIA (N = 8). Proteins were considered if quantified in at least 6 of 8 biological replicates in one group. G Correlation of the hiSPECS DIA method comparing hippocampal to mainly cortical (mixed ctx and hip, as described in legend to panel A neuronal secretome; Fig 5B).
Plotted are the log 2 fold changes between C3 treatment and DMSO control samples. The newly validated BACE1 substrate CD200 is indicated in red, known BACE1 substrates in blue, and unknown ones in black.  Figure EV5.