Mining the Secretome of C2C12 Muscle Cells: Data Dependent


Mining the Secretome of C2C12 Muscle Cells: Data Dependent...

0 downloads 64 Views 971KB Size

Subscriber access provided by READING UNIV

Article

Mining the secretome of C2C12 muscle cells: Data dependent experimental approach to analyze protein secretion using label-free quantification and peptide based analysis Leonie Grube, Rafael Dellen, Fabian Kruse, Holger Schwender, Kai Stuehler, and Gereon Poschmann J. Proteome Res., Just Accepted Manuscript • DOI: 10.1021/acs.jproteome.7b00684 • Publication Date (Web): 11 Jan 2018 Downloaded from http://pubs.acs.org on January 13, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Journal of Proteome Research is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

Mining the secretome of C2C12 muscle cells: Data dependent experimental approach to analyze protein secretion using label-free quantification and peptide based analysis †Leonie Grube, ‡+Rafael Dellen, †Fabian Kruse, ‡+Holger Schwender, †#Kai Stühler, and †Gereon Poschmann* †Molecular Proteomics Laboratory, Biomedical Research Centre (BMFZ), Heinrich-HeineUniversity, Düsseldorf, Germany, ‡Mathematical Institute, Heinrich-Heine-University, Düsseldorf, Germany, +Center for Bioinformatics and Biostatistics, Biomedical Research Centre Heinrich-HeineUniversity, Düsseldorf, Germany, #Institute for Molecular Medicine, University Hospital Düsseldorf, Düsseldorf, Germany

*corresponding author Correspondence to: [email protected]; phone +49 0211 81-10567

KEYWORDS Secretome, skeletal muscle cells, ESI-LC-MS/MS, LSPFP, proteolytic processing, Plexin-B2 1 ACS Paragon Plus Environment

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ABSTRACT Secretome analysis faces several challenges, including detection of low abundant proteins and the discrimination of bona fide secreted proteins from false-positive identifications stemming from cell leakage or serum. Here, we developed a two-step secretomics approach and applied it to the analysis of secreted proteins of C2C12 skeletal muscle cells since the skeletal muscle has been identified as an important endocrine organ secreting myokines as signaling molecules. First, we compared culture supernatants with corresponding cell lysates by mass spectrometrybased proteomics and label-free quantification. We identified 672 protein groups as candidate secreted proteins, due to their higher abundance in the secretome. Based on Brefeldin A mediated blocking of classical secretory processes, we estimate a sensitivity of > 80% for the detection of classical secreted proteins for our experimental approach. In the second step, the peptide level information was integrated with UniProt based protein information employing the newly developed bioinformatics tool “Lysate and Secretome Peptide Feature Plotter” (LSPFP) to detect proteolytic protein processing events which might occur during secretion. Concerning the proof of concept, we identified truncations of the cytoplasmic part of the protein Plexin-B2. Our workflow provides an efficient combination of experimental workflow and data analysis to identify putative secreted and proteolytic processed proteins.

2 ACS Paragon Plus Environment

Page 2 of 40

Page 3 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

INTRODUCTION The skeletal muscle has been identified as an important active secretory organ releasing socalled ‘myokines’, mediating metabolic regulation, inflammatory processes, angiogenesis and 1-3

myogenesis.

So far, it has been shown that myokines have endo-, para- and autocrine

functions. A prominent example is interleukin-6 whose concentration increases during exercise. 4 This results in AMP kinase and/or phosphatidylinositol 3-kinase activation leading to enhanced glucose uptake and fat oxidation in the skeletal muscle. 5 Moreover, increased interleukin-6 levels enhance hepatic glucose production and whole body lipolysis.

6

In addition, myokine levels change in disease states like insulin resistance as

exemplified by increased matrix metalloprotease-2 levels

7

also found at higher concentrations

in plasma of diabetic patients. 8 To date, several studies have been performed using skeletal muscle cells of different species as model system to characterise and identify related myokines. Each experimental approach revealed a different number of identified myokines ranging from 189 proteins released from L6 rat myotubes

7

over 305

9

and 635 proteins from human skeletal muscle supernatants

10

to

1073 11 potential myokines released from mouse C2C12 cells. For the identification of myokines and discrimination of bona fide secreted proteins from contaminants, most proteomic approaches are based on data base annotation or prediction of signal peptides

12

and/or on

bioinformatical filtering steps. Beside these methodical aspects, the detailed characterization of the secretome is complex due to the different secretory pathways. So far, only the process details for the so-called ‘classical secretion’ pathway are well established and proteins secreted by this pathway can be predicted according to an N-terminal signal peptide. Furthermore, there are numerous endoplasmic reticulum-Golgi independent ‘non-classical’ secretion mechanisms of proteins without a 3 ACS Paragon Plus Environment

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

functional signal peptide.

13

Page 4 of 40

These non-classical secreted proteins are released by translocation

directly through the plasma membrane via microvesicle shedding, membrane transport channels, receptor shedding, exosomes, and secretion of membrane-anchored proteins via flip flop mechanism.14 So far, these unconventionally secreted proteins are difficult to predict and are prone to be assigned as false-positives or contaminants. Therefore, alternative strategies need to be developed. There are multiple bioinformatic tools available for in-depth secretome analysis. 15 A commonly used secretion prediction tool is SignalP

16

which predicts the presence and location of signal

peptide cleavage sites in the amino acid sequences of classically secreted proteins based on a combination of several artificial neural networks. SecretomeP uses sequence information common to all secreted proteins intending to forecast proteins’ non-classical secretion.

17

Its

prediction is based on a common feature hypothesis, but the value of the prediction is still under debate. Usually, the analysis of secreted proteins is restricted to a prediction of secretion and therefore the interpretation and validation of the data remains a challenging. 11 In addition to the identification of secreted proteins, proteolytically protein processing gained increased attention, as classical secreted proteins were frequently proteolytically processed during maturation.

18

Furthermore, extracellular protein domains can be released from the cell

surface into the cellular environment as soluble proteins by a proteolytic mechanism called ectodomain shedding. 19 Several methods already exist to identify protein proteolytic processing and truncations e.g. by sequence specific antibodies or via the “protein truncation test” which selectively detects translation-terminating mutations.20,

21

However, the available tools are

limited in the number of simultaneously analyzed proteins, the localization of the proteolytic processing sites in the protein sequence and the graphical representation of the results. Thus, bioanalytical tools need to be improved in order to identify, visualize, and assess the presence of 4 ACS Paragon Plus Environment

Page 5 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

biological significant protein processing events to fully exploit mass spectrometry generated data. Here, we used a combined approach of secretome and proteome analysis to characterise the secretome of C2C12 skeletal mouse muscle cells. By means of of label-free quantitative mass spectrometry in combination with the newly developed bioanalytical tool “Lysate and Secretome Peptide Feature Plotter” (LSPFP) we identified secreted skeletal muscle-derived proteins and gained insights into possible posttranslational modification processes during protein secretion.

5 ACS Paragon Plus Environment

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Experimental Methods Cell culture C2C12 mouse myoblasts (purchased from the European collection of Cell Cultures) were grown in Dulbecco's Modified Eagle's medium 4.5 g/L glucose supplemented with 20% sterile filtered fetal bovine serum (Pan-Biotech, Aidenbach, Germany) and 1% penicillin and streptomycin at 37 °C in humidified air containing 5% CO2. Undifferentiated myoblast were grown to 80-90% confluence. Differentiation of the C2C12 cells to myotubes was induced by 5 day incubation in medium supplemented with 2% horse serum. The medium was replaced with fresh medium every 2-3 days. If not stated otherwise, media and reagents were purchased from Sigma Aldrich (Taufkirchen, Germany).

Sample preparation for proteomic analysis Secretome and cells were harvested from five individual cell culture dishes, prepared and measured independently (n = 5 per group): After 5 days of differentiation the myotubes were washed five times with serum-free medium to reduce the amount of potential contaminating serum proteins and incubated for 5 h in serum-free medium. For Brefeldin A treatment a stock solution of 5 mg/mL in ethanol was prepared and stored at -20 °C until use. C2C12 myotubes were incubated at 37 °C for 5 h with Brefeldin A at 1 µg/mL final concentration in serum-free culture medium or as a control only with the solvent ethanol. Subsequenly, the conditioned medium was collected and the cells were harvested for proteomic analysis. This procedure was performed with five biological replicates (differentiated cells / medium supernatants from individual dishes). The cell viability after incubation with serum-free medium and after 6 ACS Paragon Plus Environment

Page 6 of 40

Page 7 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

treatment with Brefeldin A was monitored using trypan blue staining and cells analyzed by an automated cell counter (Countess II FL, Thermo Fisher Scientific, Schwerte, Germany). For mass spectrometric analysis, both, cells (referred to as cellular proteome) and culture supernatants (referred to as secretome) were prepared: 10 mL conditioned medium was centrifuged at 4 °C, 5 min at 1000 x g and the supernatants were sterile-filtered (pore size: 0.2 µm Acrodisc MS syringe filter, Pall, Dreieich, Germany) to remove cell debris and death cells. The proteins in the conditioned medium were precipitated for 1 h at 4 °C by addition of 50% (w/v) trichloroacetic acid and 0.1% sodium [dodecanoyl(methyl)amino]acetate in water, sedimented by centrifugation and washed with ice-cold acetone. After repeated centrifugation, the protein pellet was shortly air-dried at room temperature, and resolved in 50 µL lysis buffer (30 mM tris(hydroxymethyl)aminomethane, 2 M thiourea, 7 M urea and 4% (w/v) 3-[(3cholamidopropyl)dimethylammonio]-1-propanesulfonate, pH 8.5). Cellular proteomes were prepared from cells differentiated into myotubes. After harvesting cells by scraping them in phosphate buffered saline and buffer removal by centrifugation, lysis buffer in combination with a TissueLyser (Qiagen, Hilden, Germany) was used to extract proteins as described. 22 Briefly, the homogenates were sonicated and the supernatants were collected after centrifugation. The protein concentrations of secretome and cellular proteome samples were determined by the Pierce 660nm Protein Assay (Fisher Scientific, Schwerte, Germany). 5 µg protein per sample was loaded onto a 4-12% Bis-Tris sodium dodecyl sulfate (SDS)-polyacrylamide gel (Novex NuPAGE, Thermo Scientific, Darmstadt, Germany) and run for about 10 minutes for an one shot analysis, whereas 20 µg of protein was separated for about an hour for the analysis of 15 different gel slices per lane. After silver staining, protein bands were cut out and processed as 7 ACS Paragon Plus Environment

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

described. 22 Briefly, bands were destained and washed, proteins reduced by 10 mM dithiothreitol and alkylated with 55 mM iodoacetamide. Subsequently, proteins were digested for 16 h at 37 °C with 0.1 µg trypsin (Serva, Heidelberg, Germany) in 100 mM ammonium hydrogen carbonate in water. Tryptic peptides were extracted twice with an 1:1 (v/v) solution of acetonitrile and 0.1% trifluoroacetic acid and, after acetonitrile removal, resuspended in 0.1% (v/v) trifluoroacetic acid.

Liquid chromatography and mass spectrometric analysis For one shot analysis, 500 ng peptides per sample were subjected to an Ultimate 3000 Rapid Separation Liquid Chromatography system (RSLC, Thermo Fisher Scientific, Dreieich, Germany), whereas for the gel slice approach all peptides eluted from each of the 15 different slices were analyzed in 15 different runs (one run per slice). First, peptides were concentrated on a trap column (Acclaim PepMap100 trap column, 3 µm C18 particle size, 100 Å pore size, 75 µm inner diameter, 2 cm length, Thermo Fisher Scientific, Dreieich, Germany) for 10 minutes at a flow rate of 6 µL/min using 0.1 % trifluoroacetic acid as mobile phase. Then, the peptide mixture was separated on an analytical column (Acclaim PepMapRSLC, 2 µm C18 particle size, 100 Å pore size, 75 µm inner diameter, 25 cm length, Thermo Scientific, Dreieich, Germany) which was heated to 60 °C at a flow rate of 300 nL/min using a 2 h gradient from 4 to 40% solvent B (solvent A: 0.1% (v/v) formic acid in water, solvent B: 0.1% (v/v) formic acid, 84% (v/v) acetonitrile in water). For mass spectrometric analysis, peptides were injected into an online coupled Q Exactive plus hybrid quadrupole Orbitrap mass spectrometer (Thermo Fisher Scientific, Dreieich, Germany) 8 ACS Paragon Plus Environment

Page 8 of 40

Page 9 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

via a nanosource electrospray interface equipped with distal coated SilicaTip emitters (New Objective, Woburn, MA, USA). The mass spectrometer was operated in positive mode with a spray voltage of 1400 V and a capillary temperature set to 250 °C. A data dependent top ten method was applied: first, full scans were recorded in profile mode in the orbitrap analyser over a scan range from 350 to 2,000 m/z with a resolution of 70,000. The target value for automatic gain control was set to 3,000,000 and ions accumulated for a maximum of 80 ms. In a second step, tandem mass spectra were recorded for a maximum of ten precursors: Two and threefold charged precursors were isolated within a 2 m/z window, fragmented via higher-energy collisional dissociation and fragments analyzed in the Orbitrap over a maximal scan range from 200 to 2000 m/z at a resolution of 17,500. The maximum ion time was 60 ms and the value for automatic gain control was set to 100,000. Already fragmented precursors were excluded from repeated fragmentation for the next 100 s.

Computational mass spectrometric data analysis Peptides and proteins were identified and quantified using the MaxQuant environment (version 1.5.5.1, MPI for Biochemistry, Planegg, Germany). if not stated otherwise with standard parameters. As samples from mouse were analyzed, searches were conducted using the mouse proteome dataset (UP000000589, downloaded 2nd November 2016, 49838 entries) from the UniProt Knowledgebase using tryptic specificity (cleavage behind R and K) with a maximum of two missed cleavages sites. One analysis was also performed with semi-specific cleavage specificity. Carbamidomethylation at cysteines was considered as fixed, and methionine oxidation and acetylation at protein N-termini were set as variable modification. A first search was performed with 20 ppm precursor mass tolerance and identified high confidence peptides 9 ACS Paragon Plus Environment

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

were then used for recalibration using the ‘software lock mass’ feature of MaxQuant.

Page 10 of 40

23

Thereafter, a main search was conducted with a precursor mass tolerance of 4.5 ppm. The mass tolerance or fragment spectra was set to 20 ppm. Peptides and proteins were accepted with a false discovery rate of 1%. Label-free quantification

24

was enabled and based on unique and razor peptides. The minimal

ratio count was set to two, peptides with variable modifications were included in the quantification; the option for matching between the runs option was enabled. The mass spectrometry proteomics data has been deposited to the ProteomeXchange Consortium via the PRIDE 25 partner repository with the dataset identifier PXD007527. After MaxQuant based database search and quantification, quantitative protein level data was analyzed within the Perseus framework (version 1.5.5.3, MPI for Biochemistry, Planegg, Germany). Only proteins with at least 2 identified peptides and a minimum of 4 valid values in at least one group were considered and proteins excluded which were identified only by site or marked as contaminant (from the MaxQuant contaminant list). Calculations were done on normalized intensities (LFQ intensities) as provided by MaxQuant and missing data was imputed before statistical analysis by values from a normal distribution (width 0.3 standard deviations) of down-shifted 1.8 standard deviations. The significance analysis of microarrays (SAM) method

26

was applied on log2 transformed values using a S0 constant of 0.1 and a 5%

false discovery rate based cutoff. Presented fold changes have been calculated as difference from mean values of log2 transformed intensities. Two independent batches of individually processed samples were analyzed: batch one including lysates and secretomes harvested from 5 dishes of C2C12 cells and batch two including secretomes of 2 x 5 dishes (5 treated with ethanol, 5 with Brefeldin A). Subsequently, both 10 ACS Paragon Plus Environment

Page 11 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

batches were measured and analyzed independently. For a further, detailed comparison, proteins identifications were matched based on the identified UniProtKB accessions considering all accessions of the respective protein groups. For the second batch (Brefeldin A treatment versus control), information about potential signal peptides and transmembrane domains were mapped using protein identifiers from the UniProt KB release 2017_07 and signal peptides were predicted for the respective protein sequences using SignalP 4.1

16

with standard parameters for eukaryotes. The information mapping was

carried out for the first entry of each protein group and additionally for all entries of all protein groups, the latter one was used to compare signal peptide carrying and Brefeldin A affected proteins. Gene Ontology (GO) biological process (GOBP), cellular compartment (GOCC), Kyoto Encyclopedia of Genes and Genomes (KEGG), and Simple Modular Architecture Research Tool (SMART) were used for categorical annotations of identified proteins and annotation enrichments were calculated by Fisher’s exact tests. Enrichment associated p-values were adjusted via the method of Benjamini and Hochberg and the adjusted p-values presented. Two dimensional annotation enrichment

27

was carried out with Perseus. Enriched categories

showing an adjusted p-value of < 0.01 are presented.

For the analysis of proteolytic processing, the bioanalytical tool “Lysate and Secretome Peptide Feature Plotter (LSPFP 1.0.0)” was used. LSPFP was designed to facilitate peptide level data inspection and to narrow down the list of potential proteolytic modified proteins. For this purpose, several features have been implemented. LSPFP is written in R and freely available as 11 ACS Paragon Plus Environment

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

a package from the comprehensive R archive network (CRAN). LSPFP uses peptide output files from MaxQuant or Progenesis and plots the position of identified peptides within the complete sequence of the associated protein for each sample. The peptide plots contain information about peptide signal intensities (color coded), position, cleavage sites, protein accession, genename, protein name, and number of identified peptides. Semitryptic peptides are marked in the graphical output to visually detect potential proteolytic cleavage sites. Furthermore, the LSPFP analysis connects associated protein information from UniProt like secondary structure, protein topology (extracellular, cytoplasmic and transmembrane) or signal peptides. This information is added to the graphical output. The plots are saved in one output PDF-file.

Western blot analysis Immunoblot analysis was performed for the proteins low density lipoprotein receptor (LDLR) and Plexin-B2. Secretome and proteome samples were generated and prepared as described in the section describing sample preparation for proteomic analysis. To secure equal loading amounts, 20 µg protein of each sample were labeled with 5 pM BDP-FL-NHS-ester solution (Lumiprobe, Hannover, Germany) and subsequently separated on a 4-12% Bis-Tris SDS-gel (Thermo Fisher Scientific, Darmstadt, Germany) for 1 h at 200 V. Subsequently, proteins were transferred to a polyvinylidene fluoride membrane (Merck, Darmstadt, Germany). The blocking was performed in TBS-Tween 20 (0.1%) with 2% w/v BSA and 3% w/v skim milk powder for 30 min. Membranes were incubated with primary antibodies against LDLR (polyclonal protein A purified rabbit antibody, immunogen: human LDLR amino acid 811-860 (P01130), bs-0705R, Bioss, Woburn, USA) and Plexin-B2 (polyclonal sheep antibody, immunogen: mouse Plexin-B2 amino acid 20-1029 (NP_001152993), ABIN1982763, Antibodies-online, Aachen, Germany) at

12 ACS Paragon Plus Environment

Page 12 of 40

Page 13 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

4 °C overnight. After washing, an anti-rabbit Alexa Fluor® 555 coupled secondary antibody (A21429 Thermo Fisher Scientific, Darmstadt, Germany) was applied for LDLR detection and an anti-sheep Alexa Fluor® 555 coupled secondary antibody (A21436 Thermo Fisher Scientific, Darmstadt, Germany) for the detection of Plexin-B2. The blots were visualized by fluorescence detection using a Typhoon9400 laser scanner (GE Healthcare, Freiburg, Germany). The proteins were quantified using Image Studio Lite (version 5.2, LI-COR Biotechnology, Bad Homburg, Germany) based on either the ratio of signals from detected proteins bands (Plexin-B2) or the ratio between the antibody mediated signal and the BDP-FL loading control channel (LDLR) as in this case a second band signal for an intra-sample ratio determination is not available. Student’s t-tests were calculated on log transformed values.

13 ACS Paragon Plus Environment

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Results As the skeletal muscle is an active secretory organ which communicates with other tissues by secreting proteins – the so called myokines –

28

we were interested in identifying proteins

secreted by skeletal muscle cells. Therefore, we used differentiated C2C12 cells as model system and analyzed conditioned medium via quantitative mass spectrometry. Since the distinction of secreted proteins from protein contaminants released from dying cells is quite challenging, we performed a two-step approach to increase confidence in bona fide secreted proteins. In the first step, we compared quantitative data from the culture supernatants (referred to as secretomes) with data from corresponding cell lysates (referred to as cellular proteomes). We expected a higher relative abundance of secreted proteins in the secretome than in the respective cellular proteome. In the second step, peptide level information was considered to further investigate the data with regards to proteolytic protein processing events which might occur during protein secretion e.g. during protein maturation or shedding of extracellular protein domains. For the latter, we developed an R package called “Lysate and Secretome Peptide Feature Plotter” – LSPFP.

Comparison of cellular proteome and secretome as strategy to reveal high confidently secreted proteins Mass spectrometry-based workflows for the identification of secreted proteins commonly use prediction software such as SecretomeP or SignalP as well as database knowledge to filter the data based on the additional protein information and to characterize the secretome composition in detail.

29

There are some disadvantages of these approaches. For example database

annotations might be of ambiguous quality or not adequate for the analyzed cell type or 14 ACS Paragon Plus Environment

Page 14 of 40

Page 15 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

investigated condition. This also holds true for computational predictions which e.g. do not consider cellular specificities leading to false positive and false negative results which can hardly be controlled. Therefore, we suggest an approach for the determination of secreted proteins which is independent of database knowledge as well as predictions and instead relies merely on self-generated data. Here, we define proteins as secreted by C2C12 cells if a relatively higher abundance in conditioned media in comparison to their abundance in the corresponding cellular proteomes can be determined. Additionally, this approach provides the potential to exclude contaminant proteins released from dead or dying cells. It is assumed that these contaminants show a relatively lower abundance in conditioned medium if cell death is not massive. Initially, we monitored the influence of the serum-free medium on C2C12 cell viability. As we noticed that the viability after 24 h incubation in the serum-free medium dropped below 95%, we chose a relatively short incubation period of five hours in the serum-free medium (mean viability of 5 replicates: 98.7%, SD = 1.8, Figure S-1) prior to secretome harvesting. Then, we established a mass spectrometry based workflow providing the potential to reveal bona fide secreted proteins (Figure 1 A). Differentiated C2C12 cells were incubated for five hours with serum-free medium. Subsequently, proteins from five different culture dishes were individually extracted from medium supernatants as well as from remaining cells and analyzed via label-free mass spectrometry based quantification. Thus, using high-resolution mass spectrometry, we identified 2,432 protein groups in total including 86 protein groups showing valid quantitative values exclusively in the secretomes and 712 exclusively in the cellular proteomes (Figure S-2). 671 proteins showed a relatively higher abundance in the secretome compared to the cellular

15 ACS Paragon Plus Environment

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

proteome as revealed by the SAM method. (Figure 1 B, Table S-1) We suggest these proteins as candidate proteins released by the cells and term them “supernatant enriched proteins”. Furthermore, categorical enrichment analysis reveals that supernatant enriched proteins are associated with cellular compartment gene ontology categories related to extracellular/secreted components like vesicle (enrichment factor 1.5, p-value 1.3 x 10-36), extracellular space (enrichment factor 2.5, p-value 1.2 x 10-29) and extracellular matrix (enrichment factor 2.8, pvalue 6 x 10-18), whereas categories related to intracellular locations such as intracellular organelle (enrichment factor 0.8, p-value 1.1 x 10-12), mitochondrion (enrichment factor 0.6, pvalue 4.4 x 10-10) and ribonucleoprotein complex (enrichment factor 0.5, p-value 3.8 x 10-10) are underrepresented (Figure 1 C, Table S-2A). In contrast, proteins showing a higher abundance in the cellular proteome are associated with intracellular localization related categories like NADH dehydrogenase complex, endoplasmic reticulum and mitochondrion, whereas categories like extracellular space, extracellular region and extracellular matrix are underrepresented (Table S-2B).

16 ACS Paragon Plus Environment

Page 16 of 40

Page 17 of 40

A Protein processing

Tryptic digestion

Bioinformatic analysis (LSPFP)

Conditioned medium Secretome LC-MS/MS C2C12 Myotubes

B

Tryptic digestion

C

Supernatant enriched Proteins

12

Identification & label free quantification

Cell lysate Cellular proteome Protein processing

3.0 Enrichment factor

10 -log10 p-value

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

2.5

8

2.0

6

1.5

4

1.0

2

0.5

0

0.0 -10

-5

0

5

10

Fold change secretome / cellular proteome

Figure 1 A) Workflow for the analysis of proteins secreted from C2C12 myotubes. For the proteomic analysis of the secretome in comparison to the cellular proteome, the conditioned media and also the cells were collected after 5 h incubation in the serum-free medium. To reveal features of secreted proteins, the identification and quantification data was analyzed on protein as well as on peptide level. Furthermore, a newly introduced tool for peptide level information visualization, the LSPFP, was used for further data analysis. B) Volcano plot showing proteins detected in secretomes and the cellular proteome (n = 5 / group). Proteins marked with blue circles appear to be higher in abundance in the secretome compared to the cellular proteome (5% FDR). C) Categorical enrichment analysis. Protein annotations associated with secretome 17 ACS Paragon Plus Environment

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 40

related cellular compartments are increased in the subgroup of supernatant enriched proteins as revealed by Fisher’s exact tests.

Inhibition of classical secretion by Brefeldin A reveals that the majority of putative classical secreted proteins is detected by secretome / cellular proteome comparison To validate our approach and investigate which proportion of the determined supernatant enriched proteins might be secreted by classical secretion processes, we inhibited the classical secretion pathway by Brefeldin A (a blocker of endoplasmic reticulum to Golgi transport)

30, 31

and analyzed secretome samples with regards to a decreased abundance in the C2C12 secretome after Brefeldin A treatment. Treatment with Brefeldin A did not negatively affect the cell viability (Figure S-1). Secretomes (n = 5 Brefeldin A treated and n = 5 control) were analyzed via label-free mass spectrometry. In total, 1,557 protein groups were identified in this experiment, 875 proteins less than in the previous comparison where also cellular proteomes had been considered. 165 proteins showed a relatively lower abundance in the Brefeldin A samples compared to controls (Figure S-3, Table S-3). These proteins might represent classically secreted proteins whose secretion has been inhibited by Brefeldin A. 127 of these proteins were also quantified in the secretome / cellular proteome comparison and 83% (105 proteins) have been defined here as supernatant enriched proteins (Figure 2 A, Figure S-4, Figure S-6A). We therefore estimate for our workflow a sensitivity of > 80% for the detection of classically secreted proteins. The overlap of Brefeldin A dependent secreted proteins and supernatant enriched proteins is linked with secretion associated ontology categories like ‘extracellular region’ or ‘extracellular matrix’ (Figure 2 B, Table S-4). Most of the 22 proteins showing a Brefeldin A dependent 18 ACS Paragon Plus Environment

Page 19 of 40

secretion but were not found in the supernatant enriched proteins, show a medium to high abundance in the cellular proteome which might hamper their detection as supernatant enriched proteins. Interestingly, based on our data most of the proteins identified as supernatant enriched proteins (566 proteins, 84%) do not belong to the group of classically secreted proteins if we assume that the secretion of the classical secreted proteins is blocked by Brefeldin A.

A 8 6 4 2 0 -7.5

-5.0

-2.5

0.0

2.5

5.0

Fold change Brefeldin A / control

B Secretion blocked by Brefeldin A, supernatant enriched Secretion blocked by Brefeldin A

10 -log10 p-value

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

0.4 0.2 0.0 -0.2

cell part ribonucleoprotein complex nucleus macromolecular complex protein complex extracellular organelle intracellular organelle vesicle intracellular membrane-bounded organelle membrane

vacuole

-0.4

lysosome

extracellular space extracellular region

-0.6 -0.8 -1.0

extracellular matrix proteinaceous extracellular matrix

-1.0

-0.5

0.0

0.5

1.0

1.5

Fold change secretome / cellular proteome

Fold change Brefeldin A / control

Figure 2 Comparison of secretomes from Brefeldin A and control treated C2C12 cells A) Volcano plot after SAM analysis showing proteins from culture supernatants of C2C12 myotubes treated with Brefeldin A or solvent ethanol. The vast majority of proteins whose secretion was blocked by Brefeldin A appeared to be enriched in the culture supernatant as revealed by secretome and cellular proteome comparison. B) Two dimensional annotation enrichment analysis of gene ontology cellular compartment categories associated with proteins from secretome / cellular proteome comparison as well as the comparison of secretomes from control and Brefeldin A treated C2C12 cells.

Furthermore, we also investigated the cellular proteomes of Brefeldin A treated cells and controls to look for proteins which might accumulate in the cell after inhibition of classical secretion. Here we revealed 47 protein groups in total showing a Brefeldin A treatment mediated altered abundance (data not shown). Of the 25 proteins showing a higher abundance in 19 ACS Paragon Plus Environment

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Brefeldin A treated cells, 21 were also quantified in the analyzed C2C12 secretomes. In addition, 19 (e.g. App, Bgn, Col1a1, Serpinf1, Sparc) of these proteins were detected in secretome samples at lower abundance levels after Brefeldin A treatment, indicating that blocking of classical secretion leads to an enrichment of a subgroup of secreted proteins within the cell (Figure S-6B). For defining secreted proteins studies often rely on workflows combining protein identification from conditioned media with subsequent filtering steps to remove contaminants. Here, mainly information from databases or from prediction tools like SecretomeP or SignalP is used. Since cells often release proteins in response to certain stimuli, we do not expect that cells continuously release all proteins containing a signal peptide which is predicted e.g. from SignalP. To investigate this for classically secreted proteins, we annotated protein groups with information about signal peptides collected from SignalP and UniProt KB and checked whether their secretion was blocked by Brefeldin A. A signal peptide was predicted by SignalP for 93 % (154 proteins, annotated by UniProt KB for 95%, 156 proteins) of the 165 proteins whose secretion is reduced by Brefeldin A. In contrast, SignalP predicts a signal peptide for 10 % (140 proteins, UniProt annotates for 11%, 147 proteins) of the 1392 proteins which show no significant reduction in the supernatant after Brefeldin A treatment (Figure S-5). The prediction and annotation of classically secreted proteins by SignalP and UniProt classifies the vast majority of the Brefeldin A blocked proteins as signal peptide containing ones. However, there are some minor issues with this approach. For example, the basis for information retrieval is not well specified in several studies. Therefore, we compared systematically the prediction and annotation of signal peptide and transmembrane regions in 20 ACS Paragon Plus Environment

Page 20 of 40

Page 21 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

case all identified proteins for a certain protein group are used or only the first representative from the output. If sequences from all identifiers in a protein group are used, SignalP predicts a signal peptide for 9% more protein groups (294 versus 270 proteins). In line with this, UniProt KB annotates a signal peptide for 10% more protein groups (303 versus 275 proteins) and a transmembrane region for 20% more protein groups (136 versus 113 proteins) (Figure S-7A).

Lysate and Secretome Peptide Feature Plotter (LSPFP) includes peptide level information to detect proteolytic processing of secreted proteins A major drawback of comparing secretomes and cellular proteomes in order to identify bona fide secreted proteins is that certain secreted proteins might be missed by this approach. Reasons could be that proteins are secreted at lower level or proteins fragments released by ectodomain shedding, might not be detected as their amount in the analysis might not exceed their amount in the cell. In our dataset the 22 proteins whose secretion was blocked by Brefeldin A but which were not found in the secretome / cellular proteome comparison as secreted showed a significantly lower abundance (mean log2 intensity 24) compared to the other secretome proteins (mean log2 intensity 25.7, p-value = 0.0003, Student’s t-test). To address this issue and to additionally identify potential proteolytic processing events, we developed a tool called “Lysate and Secretome Peptide Feature Plotter” (LSPFP) using peptide level information which is inherently available from bottom up proteomics experiments and can add further information. The main function of LSPFP is to plot the position of identified peptides within its respective associated protein from samples of secretomes and cellular proteomes. This offers the possibility to compare the patterns of identified peptides and to select candidate proteins showing a characteristic processing pattern. It might also allow the identification of protein 21 ACS Paragon Plus Environment

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

cleavage and processing events. LSPFP has been designed to assist scientist by manual data inspection and to narrow down the list of candidate proteins. For this purpose, several features have been implemented in the software. LSPFP reads peptide output files from MaxQuant or Progenesis QI for proteomics and plots identified peptides from each sample, separately for each protein. In the current version, LSPFP plots tryptic and semitryptic peptides with color coded intensities. Semitryptic peptides are marked in the graphical output to visually detect potential proteolytic cleavage sites (Figure S-8). This offers the possibility to directly visually refine e.g. signal peptides cleavage or other proteolytic protein processing events. Furthermore, information downloaded from UniProt is added to the graphical output: secondary structure information as well as information about the subcellular localization of a protein part (extracellular, cytoplasmic, and transmembrane) or if a signal peptide is annotated (Figure 3). Cellular localization information will help to evaluate if a protein might be processed e.g. by shedding. LSPFP is written in R and freely available as a package from the comprehensive R archive network (CRAN).

22 ACS Paragon Plus Environment

Page 22 of 40

Page 23 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

Figure 3 Graphical output from LSPFP. Exemplarily, 21 peptides identified for the candidate protein Plexin-B2 are shown. For five individual secretome (S1-S5) and cellular proteome (L1L5) samples (y-axis) each identified and quantified peptide is mapped by a color coded line (mass spectrometric intensity, blue: low intensity, red: high intensity) to its respective position in the protein sequence (x-axis). Furthermore, the accession, gene and protein name as well as the number of identified peptides is displayed and, if available, annotation information from UniProtKB. Topology information (light red: extra cellular, light blue: signal peptide, light green: cytoplasm, yellow: trans-membrane) is visualized as well as information about secondary structure (upper part of the plot; green: α-helix, yellow: β-strand, dark blue: turn).

Validation of proteins predicted by LSPFP to be proteolytically processed during secretion Using LSPFP, we identified several proteins in the secretome which might have been proteolytically processed during protein secretion, e.g. Dystroglycan (Figure S-9 A). Furthermore, in the secretom samples Plexin-B2 (PLXB2, Figure 3) and the low-density lipoprotein receptor (LDLR, Figure S-9 B) showno peptide signals in the cytoplasmic part of the protein. However, in the extracellular part a coverage of at least six peptides for LDLR and three for Plexin-B2 in each samples is visible. Next, we validated our findings in an independent set 23 ACS Paragon Plus Environment

Journal of Proteome Research

of samples (n = 3 per group) by western blot analysis. Using an antibody directed to the extracellular part of the Plexin-B2, an additional protein species of lower molecular weight was detected predominantly in the secretomes representing a truncated variant of the protein (p = 0.0049, Figure 5). There was only an antibody directed to the cytoplasmic protein part which was available for LDLR and we detected a lower amount of this protein in secretome samples compared to the cellular proteomes (p = 0.03, Figure S-10).

B

kDa 206 140 65

 

2.0 Ratio upper / lower band

Lysate

A

Secretome

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 40

1.5

Secretome Lysate

**

1.0 0.5 0.0

25

Figure 4 Western blot analysis of Plexin-B2 protein species present in C2C12 cell secretomes and cellular proteomes. A) A protein variant showing a lower molecular weight compared to full length Plexin-B2 is predominantly present in secretome samples. B) Quantitative analysis of signals (marked by arrows) revealing a significantly (p=0.0049, n = 3/group) higher abundance of the truncated Plexin-B2 variant at 140 kDa compared to the 206 kDa variant in secretome samples. Bars represent the mean values of intensity ratios and error bars the standard deviation.

24 ACS Paragon Plus Environment

Page 25 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

Moreover, to reach higher protein sequence coverage and to validate our findings of proteolytically processed protein variants in an independent-experimental set-up, we separated one cellular proteome sample and one secretome sample of C2C12 cells in a polyacrylamidgel. This strategy enabled us to analyze a higher amount of protein in the subsequent mass spectrometric analysis. Each lane was cut into 15 pieces (Figure S-11) which were individually analyzed. Subsequently, the resulting data was processed with LSPFP to plot peptide distributions for every secretome and cellular proteome gel fraction. Using this approach, we were able to confirm our results for Plexin-B2 (Figure 5) and LDLR (Figure S-12). For both proteins, only in cellular proteome samples peptides were detected in both, extra- and intracellular protein regions, whereas in secretome samples the cytoplasmic protein part was missing. By measurements of gel-fractions, the protein sequence coverage for LDLR increased from 17.3% (10 peptides) to 22.7% (14 peptides) and for Plexin-B2 from 18.4% (21 peptides) to 23.3% (32 peptides). Furthermore, we supported our previous data and confirmed the missing peptide coverage in cytoplasmic protein regions for Plexin-B2 and LDLR in secretome samples. Therefore, we suggest that lower molecular weight species detected for both, LDLR and PlexinB2, predominantly in the secretome samples might represent a truncated variant of the respective protein.

25 ACS Paragon Plus Environment

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 5 Graphical output from LSPSP. C2C12 myotube cell culture supernatants and cellular proteomes were separated using SDS gel electrophoresis and 15 individual gel fractions per lane (S1-S15: secretome, L1-L15: cellular proteomes) were analyzed using quantitative mass spectrometry. Afterwards, identified peptides were visualized for each gel fraction. This approaches results in a higher protein sequence coverage compared to a one-shot analysis and confirms our findings that in the cell culture supernatant only Plexin-B2 peptides were found which have been annotated to be extracellular located, but not peptides from the cytosolic region. In the cellular proteomes peptides from both intracellular and extracellular regions have been identified.

26 ACS Paragon Plus Environment

Page 26 of 40

Page 27 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

Discussion Over the past decade, secretome analysis has been reported in various organisms and cell types. The combination of secretome data and bioinformatical tools enables the potential to predict protein secretion based on sequence information. 29 In the present study, we investigated protein secretion of C2C12 skeletal muscle myotubes in a two-step approach based on self-generated data. First, we compared corresponding secretomes and cellular proteomes. In the second step we used the peptide information to identify proteolytic protein processing events using the bioanalytical tool LSPFP. The composition of secretomes depends on the cell system (cell lines, primary cells, tissue explants), but also on the cell culturing and secretome harvesting conditions.

9

Although

respective models are useful to answer functional questions, a validation in vivo is necessary. Nevertheless, the analysis of cell culture supernatants offers the potential to relate secreted proteins directly to the cell of origin which is often not possible in more complex systems. One major drawback of cell line secretome analysis is the possible contamination of the conditioned medium by death cells and serum contaminants. Therefore, we propose an approach for C2C12 myotube secretome analysis comparing the abundance of proteins present in culture supernatants with corresponding cellular proteome protein abundances. However, the amount of secretory proteins in the conditioned medium depends on their secretion rate, the incubation length in serum-free medium before harvesting and their degradation rate. Here, we pragmatically loaded equal protein amounts of secretome and cellular proteome samples for LC-MS/MS analysis. Here, it is necessary to be careful if the analyzed cells show a low protein secretion rate. Furthermore, it is possible that due to cell-type specific different proteomic compositions, masking effects might be pronounced and the different 27 ACS Paragon Plus Environment

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

intensity distributions might impact the quantitative comparison of secretome and cellular proteome samples. Including several peptides for the quantitative analysis might partly solve these issues. Another aspect is the comparability of the secretome and cellular proteome samples because of necessitated deviant sample preparation for proteomic analysis. Due to the often low concentration of secreted proteins in culture supernatants there might be the need of concentrating them before further protein analysis. Therefore, proteins of the secretome were precipitated by addition of trichloroacetic acid, whereas proteins of the cellular proteome not. Other methods for purification and concentration of secretory proteins are ultrafiltration and dialysis

32

and besides trichloroacetic acid, acetone, chloroform and methanol as well as

ammonium sulfate can be used for protein precipitation. Each method has its merits and demerits, but the major advantage of using trichloroacetic acid for precipitation is the high protein recovery and number of identifiable proteins, as well as the easy handling, although it requires an additional step with acetone to remove trichloroacetic acid. 32, 33 Nevertheless, the recovery of proteins from different sample preparation procedures might have an impact on the result of the quantitative analysis. Unfortunately, there is no gold standard to assess which proteins are really released by cells and therefore it is not trivial to benchmark the chosen approach of comparing secretomes and cellular proteomes. Our Brefeldin A experiment suggests, that database (UniProt KB) and prediction (SecretomeP) based filtering approaches might detect at least classical secreted proteins with a sensitivity of 93 % and a specificity of 89 %. Nevertheless, the precision lays only around 50 %. Even if it must be considered that signal peptide containing proteins might also be released by Golgi bypass mechanisms

34, 35

, the high number of potential false positives hits suggests that our

proposed data dependent approach might be an advantage. Furthermore, each prediction tool 28 ACS Paragon Plus Environment

Page 28 of 40

Page 29 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

and database provides a different number of proteins e.g. for containing a signal peptide (Figure S-7 B, C), which has also been shown by others. 16, 36 On the one hand, we found that the identified supernatant enriched proteins from C2C12 myotubes show an enrichment of categorical annotations associated with extracellular or vesicular localization, whereas a lower number of annotations associated with intracellular localization were found in this group of proteins. This result supports the usefulness of the chosen approach. On the other hand, we identified proteins whose secretion was blocked by Brefeldin A as a group of proteins which are probably classically secreted and therefore might serve as a positive control for classically secreted proteins for benchmarking. Within this group of proteins 83% were also identified as supernatant enriched proteins revealing that a large amount of classically secreted proteins can be detected. Nevertheless, 566 additional proteins were found to be enriched in the supernatant. We suppose, that at least a part of those proteins are released by processes like ectodomain shedding or unconventional secretion e.g. by Golgi independent vesicular secretion processes.

37

The protein content of exosome-like vesicles has already been

characterised in C2C12 cells in proliferation and differentiation experiments. 38 Furthermore, some of the identified proteins indicate that differentiation might impact the secretome of C2C12 cells. Interestingly, there are several proteins (MIF, IGFBP-2, -6, -7, TIMP2, Cathepsin B) among the supernatant enriched proteins which have been previously associated with the senescence-associated secretory phenotype of cells.

39

This might provide a

link to cellular senescence associated with C2C12 differentiation. 40 Additionally, there were several proteins in the list of supernatant enriched proteins which are mainly known for their function within the cell like histones (Histone H1.1, H1.5, H 1.2, H1.4).

29 ACS Paragon Plus Environment

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 30 of 40

Nevertheless, histones have been described to be released by several cell types accompanied with extracellular functions including interaction with toll-like receptors. 41 Concerning the secretome of C2C12 cells two other lists of secreted proteins have been described by Deshmukh et al. (1073 proteins)

11

and Henningsen et al. (635 proteins)

10

. We

compared the lists of identified proteins to our data qualitatively and quantitatively (Figure S13). Whereas in each study over 600 proteins have been identified, the overall overlap of 142 proteins appears not to be quite large. The relatively low number of commonly found candidate secreted proteins might be due to the two different approaches used for filtering secreted proteins – prediction and database information based filtering in the reports by Deshmukh and Henningsen and a quantitative comparison of secretome and proteome in our approach. If the unfiltered list of in the secretome samples from our study is used for the comparison with the lists of secreted proteins from the other reports, the number of common proteins increases from 187 to 406 (Deshmukh et al.) and 184 to 383 (Henningsen et al.). Furthermore, we suggest that also different criteria for positive protein identification as well as different cell culture conditions (12 h incubation with serum free medium in the other two reports and 5 h in our study) explain the low overlap between the compared reports. In addition the applied incubation duration in serum free medium as well as the treatment with palmitate and BSA might also explain the low correlation (mean Pearson’s correlation coefficient of 0.61) between the quantitative data from Deshmukh et al. with our data. The idea to reveal secreted proteins by comparing secretomes with proteomes using LC-MS/MS analysis shas also been followed by Luo et al.. In this study based on spectral count data, the cellular proteome of the A549 lung-cancer cell line was used as a reference to refine secreted proteins. Using this workflow which did not include replicate analysis, they refined 382 proteins in the so called high-quality secretome of which more than 85% were annotated as secreted. 42 30 ACS Paragon Plus Environment

Page 31 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

In another report by Stiess et al. used a dual SILAC strategy and reported relative extra- to intracellular ratios for each protein. One advantage of this method is that serum contaminants or other high abundant proteins which have been added during the experiment can be easily excluded.

43

In our setup, we used a label-free quantification based approach to compare the

relative abundance of proteins in and outside the cell. Here, each sample has to be analyzed in separate mass spectrometric runs which might impact the quantification as cell lysates and secretomes show quite a different protein composition which might affect co-eluting peptide as well as feature matching. Nevertheless, main advantages of our approach are that no protein labeling is necessary and statistical analysis and dealing with missing values are straight forward as no SILAC ratios have to be considered for the calculations. A quite interesting approach for the determination of secreted proteins from cultured cells has been proposed by Eichelbaum et al.

44

They used pulse labeling with azidohomoalanine and

heavy labeled amino acids. The azidohomoalanine is used for selective enrichment of newly synthesized proteins by click chemistry and the heavy amino acids to facilitate quantification. Using this approach it was possible to quantify over 600 putative secreted proteins even in the presence of high serum amounts. Beside the proposed way of determining secreted proteins by a comparison of secretome and cell-lysates we developed a new tool – LSPFP to mine our data also on peptide leve. LSPFP has primarily been developed to visualize data from a secretome and cellular proteome comparison and to enable the use of peptide level data for getting insights in protein processing and modification. In our study, datasets profit from a high coverage which can be reached by recent LC-MS instrumentation, long gradients, fractionation and a combination of datasets generated with different proteases. Nevertheless, it is important to keep in mind that detection on peptide level might also depend on protein modification, overall protein abundance in a 31 ACS Paragon Plus Environment

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 32 of 40

sample type, matrix effects which might differ between cellular proteomes and secretomes and the measurement sensitivity. By data visualization and comparing the peptide patterns from secretomes and cell lysates, several proteins could be identified as candidates for proteolytic processing or shedding. For several protein candidates, shedding from the cell surface has already have been described in other cell types. For Dystroglycan for example, shedding of the extracellular domain has been shown for skin keratinocytes and fibroblasts which might be mediated by metalloproteases.

45

Interestingly, both skin and muscle cells are known to connect their cytoskeleton via a Dystroglycan containing complex to the basement membrane. The semaphoring receptor PlexinB2 has already been shown to be proteolytically processed in several cell lines presumably by subtilisin-like proprotein convertases. 46 Furthermore, LDLR shedding – has been demonstrated in brain endothelial cells which might here be mediated by Adam10. Interestingly, LDLR shedding has been linked to beta-amyloid endocytic transport

47, 48

and amyloid beta has also

been found in as supernatant enriched proteins in C2C12 cells in our study. The LSPFP based peptide level data supported shedding of extracellular protein domains from Plexin-B2 and LDLR from muscle cells has been successfully validated in our study using Western blot analysis. Beside the detection of shedding by peptide level secretome / lysate comparison, LSPSP can additionally be used to detect internal proteolytic cleavage introduced during protein maturation by visualizing semitryptic peptides. For example an easy validation of proposed signal peptide cleavage is feasible as well as the detection of so far unknown proteolytic processing events. As already exemplified, LSPSF can also be used for peptide level visualization of other experiment types like gel slice wise peptide visualization. This setup might help to characterise

32 ACS Paragon Plus Environment

Page 33 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

proteoforms running at different positions in the gels or – as shown – for getting data with higher peptide coverage to enable an extended characterization of protein isoforms.

Conclusion The presented approach relies on self-generated data to characterise proteins secreted from C2C12 skeletal muscle cells. By comparing relative protein amounts from cell lysates and cell supernatants, supernatant enriched proteins can be defined, representing probably secreted proteins. Furthermore, the LSPFP software package assists by visualizing peptide level data in detection of proteolytic processing events like protein shedding.

33 ACS Paragon Plus Environment

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Supporting Information The following files are available free of charge. Figure S-1 Cell viability after incubation with serum free medium and after treatment with Brefeldin A Figure S-2 Quantitative analysis of C2C12 myotube secretomes and cellular proteomes Figure S-3 Analysis of secretomes from Brefeldin A treated C2C12 myotubes Figure S-4 Brefeldin A affected proteins in the comparison of secretomes and cellular proteomes Figure S-5 Brefeldin A affected proteins showing an UniProt KB annotated signal peptide S-7 Figure S-6 Overlap of proteins from different C2C12 cell related experiments Figure S-7 Different data analysis workflows provide different signal peptide annotations Figure S-8 LSPFP output: example for semitryptric cleavage Figure S-9 LSPFP output: examples for potentially proteolytically processed proteins Figure S-10 Western blot analysis of low-density lipoprotein receptor Figure S-11 Analyzed protein containing gel slices Figure S-12 Output from LSPFP from the analysis of secretomes and cell lysates previously separated in polyacrylamide gels. Figure S-13 Comparison of the lists of proteins putatively secreted from C2C12 cells from this analysis and from other published studies. 34 ACS Paragon Plus Environment

Page 34 of 40

Page 35 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

Table S-1 Supernatant enriched proteins as revealed by secretome / cellular proteome comparison Table S-2 Annotation enrichment analysis of supernatant enriched proteins Table S-3 Proteins showing abundance changes in C2C12 secretome after Brefeldin A treatment Table S-4 Two dimensional annotation enrichment of gene ontology cellular compartment categories

35 ACS Paragon Plus Environment

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Acknowledgements We would like to thank Anja Fink for proofreading the manuscript and greatly acknowledge the Strategic Research Fund of the Heinrich-Heine-Universität Düsseldorf for funding; grant 10/2014 to G.P..

Conflict of Interest Disclosure The authors declare no competing financial interest.

36 ACS Paragon Plus Environment

Page 36 of 40

Page 37 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

References

1. Hartwig, S.; Raschke, S.; Knebel, B.; Scheler, M.; Irmler, M.; Passlack, W.; Muller, S.; Hanisch, F.-G.; Franz, T.; Li, X.; Dicken, H.-D.; Eckardt, K.; Beckers, J.; de Angelis, M. H.; Weigert, C.; Häring, H.-U.; Al-Hasani, H.; Ouwens, D. M.; Eckel, J.; Kotzka, J.; Lehr, S., Secretome profiling of primary human skeletal muscle cells. Biochimica et Biophysica Acta (BBA) - Proteins and Proteomics 2014, 1844, (5), 1011-1017. 2. Eckardt, K.; Görgens, S. W.; Raschke, S.; Eckel, J., Myokines in insulin resistance and type 2 diabetes. Diabetologia 2014, 57, (6), 1087-1099. 3. Evers-van Gogh, I. J. A.; Alex, S.; Stienstra, R.; Brenkman, A. B.; Kersten, S.; Kalkhoven, E., Electric Pulse Stimulation of Myotubes as an In Vitro Exercise Model: Cell-Mediated and Non-CellMediated Effects. Scientific Reports 2015, 5, 10944. 4. Pedersen, B. K.; Febbraio, M. A., Muscle as an endocrine organ: focus on muscle-derived interleukin-6. Physiol Rev 2008, 88, (4), 1379-406. 5. Carey, A. L.; Steinberg, G. R.; Macaulay, S. L.; Thomas, W. G.; Holmes, A. G.; Ramm, G.; Prelovsek, O.; Hohnen-Behrens, C.; Watt, M. J.; James, D. E.; Kemp, B. E.; Pedersen, B. K.; Febbraio, M. A., Interleukin-6 increases insulin-stimulated glucose disposal in humans and glucose uptake and fatty acid oxidation in vitro via AMP-activated protein kinase. Diabetes 2006, 55, (10), 2688-97. 6. Febbraio, M. A.; Hiscock, N.; Sacchetti, M.; Fischer, C. P.; Pedersen, B. K., Interleukin-6 is a novel factor mediating glucose homeostasis during skeletal muscle contraction. Diabetes 2004, 53, (7), 1643-8. 7. Yoon, J. H.; Yea, K.; Kim, J.; Choi, Y. S.; Park, S.; Lee, H.; Lee, C. S.; Suh, P. G.; Ryu, S. H., Comparative proteomic analysis of the insulin-induced L6 myotube secretome. Proteomics 2009, 9, (1), 51-60. 8. Derosa, G.; D'Angelo, A.; Scalise, F.; Avanzini, M. A.; Tinelli, C.; Peros, E.; Fogari, E.; Cicero, A. F., Comparison between metalloproteinases-2 and -9 in healthy subjects, diabetics, and subjects with acute coronary syndrome. Heart Vessels 2007, 22, (6), 361-70. 9. Hartwig, S.; Raschke, S.; Knebel, B.; Scheler, M.; Irmler, M.; Passlack, W.; Muller, S.; Hanisch, F. G.; Franz, T.; Li, X.; Dicken, H. D.; Eckardt, K.; Beckers, J.; de Angelis, M. H.; Weigert, C.; Haring, H. U.; Al-Hasani, H.; Ouwens, D. M.; Eckel, J.; Kotzka, J.; Lehr, S., Secretome profiling of primary human skeletal muscle cells. Biochim Biophys Acta 2014, 1844, (5), 1011-7. 10. Henningsen, J.; Rigbolt, K. T.; Blagoev, B.; Pedersen, B. K.; Kratchmarova, I., Dynamics of the skeletal muscle secretome during myoblast differentiation. Mol Cell Proteomics 2010, 9, (11), 2482-96. 11. Deshmukh, A. S.; Cox, J.; Jensen, L. J.; Meissner, F.; Mann, M., Secretome Analysis of LipidInduced Insulin Resistance in Skeletal Muscle Cells by a Combined Experimental and Bioinformatics Workflow. J Proteome Res 2015, 14, (11), 4885-95. 12. Bortoluzzi, S.; Scannapieco, P.; Cestaro, A.; Danieli, G. A.; Schiaffino, S., Computational reconstruction of the human skeletal muscle secretome. Proteins 2006, 62, (3), 776-92. 13. Nickel, W., Unconventional secretory routes: direct protein export across the plasma membrane of mammalian cells. Traffic 2005, 6, (8), 607-14. 14. Prydz, K.; Tveit, H.; Vedeler, A.; Saraste, J., Arrivals and departures at the plasma membrane: direct and indirect transport routes. Cell Tissue Res 2013, 352, (1), 5-20. 15. Caccia, D.; Dugo, M.; Callari, M.; Bongarzone, I., Bioinformatics tools for secretome analysis. Biochimica et Biophysica Acta (BBA) - Proteins and Proteomics 2013, 1834, (11), 2442-2453. 16. Petersen, T. N.; Brunak, S.; von Heijne, G.; Nielsen, H., SignalP 4.0: discriminating signal peptides from transmembrane regions. Nat Methods 2011, 8, (10), 785-6. 17. Bendtsen, J. D.; Jensen, L. J.; Blom, N.; Von Heijne, G.; Brunak, S., Feature-based prediction of non-classical and leaderless protein secretion. Protein Eng Des Sel 2004, 17, (4), 349-56.

37 ACS Paragon Plus Environment

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

18. Ngounou Wetie, A. G.; Sokolowska, I.; Woods, A. G.; Wormwood, K. L.; Dao, S.; Patel, S.; Clarkson, B. D.; Darie, C. C., Automated mass spectrometry-based functional assay for the routine analysis of the secretome. J Lab Autom 2013, 18, (1), 19-29. 19. Hayashida, K.; Bartlett, A. H.; Chen, Y.; Park, P. W., Molecular and cellular mechanisms of ectodomain shedding. Anat Rec (Hoboken) 2010, 293, (6), 925-37. 20. Roest, P. A.; Roberts, R. G.; Sugino, S.; van Ommen, G. J.; den Dunnen, J. T., Protein truncation test (PTT) for rapid detection of translation-terminating mutations. Hum Mol Genet 1993, 2, (10), 171921. 21. BeaufrÈRe, L.; Tuffery, S.; Hamel, C.; Bareil, C.; Arnaud, B.; Demaille, J.; Claustres, M., The Protein Truncation Test (PTT) as a Method of Detection for Choroideremia Mutations. Experimental Eye Research 1997, 65, (6), 849-854. 22. Poschmann, G.; Seyfarth, K.; Besong Agbo, D.; Klafki, H. W.; Rozman, J.; Wurst, W.; Wiltfang, J.; Meyer, H. E.; Klingenspor, M.; Stuhler, K., High-fat diet induced isoform changes of the Parkinson's disease protein DJ-1. J Proteome Res 2014, 13, (5), 2339-51. 23. Cox, J.; Michalski, A.; Mann, M., Software lock mass by two-dimensional minimization of peptide mass errors. J Am Soc Mass Spectrom 2011, 22, (8), 1373-80. 24. Cox, J.; Hein, M. Y.; Luber, C. A.; Paron, I.; Nagaraj, N.; Mann, M., Accurate proteome-wide label-free quantification by delayed normalization and maximal peptide ratio extraction, termed MaxLFQ. Mol Cell Proteomics 2014, 13, (9), 2513-26. 25. Vizcaino, J. A.; Csordas, A.; del-Toro, N.; Dianes, J. A.; Griss, J.; Lavidas, I.; Mayer, G.; PerezRiverol, Y.; Reisinger, F.; Ternent, T.; Xu, Q. W.; Wang, R.; Hermjakob, H., 2016 update of the PRIDE database and its related tools. Nucleic Acids Res 2016, 44, (D1), D447-56. 26. Tusher, V. G.; Tibshirani, R.; Chu, G., Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci U S A 2001, 98, (9), 5116-21. 27. Cox, J.; Mann, M., 1D and 2D annotation enrichment: a statistical method integrating quantitative proteomics with complementary high-throughput data. BMC Bioinformatics 2012, 13 Suppl 16, S12. 28. Benatti, F. B.; Pedersen, B. K., Exercise as an anti-inflammatory therapy for rheumatic diseases[mdash]myokine regulation. Nat Rev Rheumatol 2015, 11, (2), 86-97. 29. Caccia, D.; Dugo, M.; Callari, M.; Bongarzone, I., Bioinformatics tools for secretome analysis. Biochim Biophys Acta 2013, 1834, (11), 2442-53. 30. Doms, R. W.; Russ, G.; Yewdell, J. W., Brefeldin A redistributes resident and itinerant Golgi proteins to the endoplasmic reticulum. The Journal of Cell Biology 1989, 109, (1), 61-72. 31. Zhao, J.; Li, L.; Leissring, M. A., Insulin-degrading enzyme is exported via an unconventional protein secretion pathway. Molecular Neurodegeneration 2009, 4, (1), 4. 32. Cao, J.; Shen, C.; Zhang, J.; Yao, J.; Shen, H.; Liu, Y.; Lu, H.; Yang, P., Comparison of alternative extraction methods for secretome profiling in human hepatocellular carcinoma cells. Sci China Life Sci 2011, 54, (1), 34-8. 33. Jiang, L.; He, L.; Fountoulakis, M., Comparison of protein precipitation methods for sample preparation prior to proteomic analysis. J Chromatogr A 2004, 1023, (2), 317-20. 34. Grieve, A. G.; Rabouille, C., Golgi bypass: skirting around the heart of classical secretion. Cold Spring Harb Perspect Biol 2011, 3, (4). 35. Tveit, H.; Akslen, L. K.; Fagereng, G. L.; Tranulis, M. A.; Prydz, K., A secretory Golgi bypass route to the apical surface domain of epithelial MDCK cells. Traffic 2009, 10, (11), 1685-95. 36. Choo, K. H.; Tan, T. W.; Ranganathan, S., A comprehensive assessment of N-terminal signal peptides prediction methods. BMC Bioinformatics 2009, 10 Suppl 15, S2. 37. Rabouille, C.; Malhotra, V.; Nickel, W., Diversity in unconventional protein secretion. J Cell Sci 2012, 125, (Pt 22), 5251-5. 38. Forterre, A.; Jalabert, A.; Berger, E.; Baudet, M.; Chikh, K.; Errazuriz, E.; De Larichaudy, J.; Chanon, S.; Weiss-Gayet, M.; Hesse, A. M.; Record, M.; Geloen, A.; Lefai, E.; Vidal, H.; Coute, Y.; Rome, S., Proteomic analysis of C2C12 myoblast and myotube exosome-like vesicles: a new paradigm for myoblast-myotube cross talk? PLoS One 2014, 9, (1), e84153. 38 ACS Paragon Plus Environment

Page 38 of 40

Page 39 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

39. Coppe, J. P.; Desprez, P. Y.; Krtolica, A.; Campisi, J., The senescence-associated secretory phenotype: the dark side of tumor suppression. Annu Rev Pathol 2010, 5, 99-118. 40. Sancho, S. C.; Ouchi, T., Cell Differentiation and Checkpoint. Int J Cancer Res Mol Mech 2015, 1, (2). 41. Chen, R.; Kang, R.; Fan, X. G.; Tang, D., Release and activity of histone in diseases. Cell Death Dis 2014, 5, e1370. 42. Luo, X.; Liu, Y.; Wang, R.; Hu, H.; Zeng, R.; Chen, H., A high-quality secretome of A549 cells aided the discovery of C4b-binding protein as a novel serum biomarker for non-small cell lung cancer. J Proteomics 2011, 74, (4), 528-38. 43. Stiess, M.; Wegehingel, S.; Nguyen, C.; Nickel, W.; Bradke, F.; Cambridge, S. B., A Dual SILAC Proteomic Labeling Strategy for Quantifying Constitutive and Cell-Cell Induced Protein Secretion. J Proteome Res 2015, 14, (8), 3229-38. 44. Eichelbaum, K.; Winter, M.; Berriel Diaz, M.; Herzig, S.; Krijgsveld, J., Selective enrichment of newly synthesized proteins for quantitative secretome analysis. Nat Biotechnol 2012, 30, (10), 984-90. 45. Herzog, C.; Has, C.; Franzke, C. W.; Echtermeyer, F. G.; Schlotzer-Schrehardt, U.; Kroger, S.; Gustafsson, E.; Fassler, R.; Bruckner-Tuderman, L., Dystroglycan in skin and cutaneous cells: betasubunit is shed from the cell surface. J Invest Dermatol 2004, 122, (6), 1372-80. 46. Artigiani, S.; Barberis, D.; Fazzari, P.; Longati, P.; Angelini, P.; van de Loo, J. W.; Comoglio, P. M.; Tamagnone, L., Functional regulation of semaphorin receptors by proprotein convertases. J Biol Chem 2003, 278, (12), 10094-101. 47. Shackleton, B.; Crawford, F.; Bachmeier, C., Inhibition of ADAM10 promotes the clearance of Abeta across the BBB by reducing LRP1 ectodomain shedding. Fluids Barriers CNS 2016, 13, (1), 14. 48. Bachmeier, C.; Shackleton, B.; Ojo, J.; Paris, D.; Mullan, M.; Crawford, F., Apolipoprotein E isoform-specific effects on lipoprotein receptor processing. Neuromolecular Med 2014, 16, (4), 686-96.

39 ACS Paragon Plus Environment

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

for TOC only 81x60mm (300 x 300 DPI)

ACS Paragon Plus Environment

Page 40 of 40