Proteomic and phosphoproteomic characterisation of primary mouse embryonic fibroblasts

Fibroblasts are the most common cell type in stroma and function in the support and repair of most tissues. Mouse embryonic fibroblasts (MEFs) are amenable to isolation and rapid growth in culture. MEFs are therefore widely used as a standard model for functional characterisation of gene knockouts, and can also be used in co‐cultures, commonly to support embryonic stem cell cultures. To facilitate their use as a research tool, we have performed a comprehensive proteomic and phosphoproteomic characterisation of wild‐type primary MEFs from C57BL/6 mice. EIF2/4 and MTOR signalling pathways were abundant in both the proteome and phosphoproteome, along with extracellular matrix (ECM) and cytoskeleton associated pathways. Consistent with this, kinase enrichment analysis identified activation of P38A, P90RSK, P70S6K, and MTOR. Cell surface markers and matrisome proteins were also annotated. Data are available via ProteomeXchange with identifier PXD043244. This provides a comprehensive catalogue of the wild‐type MEF proteome and phosphoproteome which can be utilised by the field to guide future work.

Fibroblasts are of mesenchymal origin and are the most common cell type in connective tissue, functioning in the synthesis and remodelling of the extracellular matrix (ECM) and stroma.Fibroblasts are involved in the support and repair of almost every tissue, as well as in regulation of organ development, wound healing, innate immunity, inflammation, and cancer progression [1][2][3].Owing to their versatility, rapid growth, and ease of isolation, mouse embryonic fibroblasts (MEFs) have found widespread use as a standard model for characterisation of signalling pathways, and functional characterisation of gene knockouts.MEFs can also be used in co-cultures, commonly to support embryonic stem cell cultures [4].To further facilitate their use as a research tool, herein we provide a comprehensive proteomic and phosphoproteomic characterisation of wild-type primary MEFs.
Embryos were removed from pregnant female C57BL/6 mice at day E13.5.E13.5 is at the earlier end within the accepted time window for MEFs generation (E13.5-E15.5)[5], with fibroblasts from this earlier embryonic stage more likely to retain multipotency [6].All procedures were conducted with the approval of the University of Newcastle Animal Care and Ethics Committee, A-2015-512.The head, limb buds, and visible internal organs were removed from the embryos, and the remainder of each individual embryo was placed into 500 μL trypsin (0.25%) for 10 min at 37 • C [7].The trypsin was inactivated using DMEM supplemented with 10% FBS, and the mixture was pipetted up and down several times to dissociate the embryo into single cells.
The cells were suspended in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin, and then seeded into 10 cm culture dishes.
The dishes were incubated under standard cell culture conditions at 37 • C, 5% CO2.Following passage 0, MEFs were subcultured when they reached 70%-80% confluency.Once isolated, primary MEFs can reach senescence at passages 5 and above [8].Therefore, proteomic analysis was performed on cultures before their third passage.
Fibroblast cultures (n = 4, from 2 different litters) at 70%-80% confluence were harvested by scraping in guanidium hydrochloride (GdmCl) lysis buffer (6 M GdmCl, 100 mM Tris pH 8.5, 10 mM tris(2carboxyethyl)phosphine (TCEP), 40 mM 2-chloroacetamide (CAA)) [9].Collecting the lysates by scraping ensures that secreted ECM is retained, and removes the potential for proteomic alterations that may be induced by cell dissociation using trypsin.Lysates were heated at 95 • C for 5 min to inactivate endogenous phosphatases, before sonication using a tip-probe sonicator (Hielscher Ultrasonics GmbH) for a total of 40 s (10 s on, 30 s off).Samples were again heated at 95 • C for 5 min before determination of protein concentration by bicinchoninic acid assay, as per the manufacturer's instructions (Thermo Fisher Scientific).
Protein samples were prepared for mass spectrometry analysis using previously described procedures [10].In brief, samples were diluted using 25 mM Tris, pH 8.5, to dilute the GdmCl concentration below 1 M. Trypsin/LysC (Promega, V5072) was added to each sample 1:30 (protein:enzyme) and samples were digested overnight at 37 • C, 1500 rpm.The following day, TFA was added to a final concentration of 1%v/v before desalting using the Visiprep SPE Vacuum Manifold 12port model (Sigma) as previously described [11].The eluted peptides were speed-vacuumed to dryness before mass spectrometry analysis (Proteome) or phosphoenrichment (Phosphoproteome).Two-hundred micrograms of peptide per sample was used for TiO 2 enrichment, as previously described [11].

Mass spectrometry analysis was performed using an Orbitrap
Exploris 480 coupled to a Dionex Ultimate 3000RSLC nanoflow HPLC (Thermo Fisher Scientific).Peptides were concentrated with an Acclaim PepMap 100 C18 75 μM x 20 mm trap column (Thermo Fisher Scientific) prior to separation on a 75 μM x 25 cm EASY-Spray PepMap C18 column (Thermo Fisher Scientific).Phosphoproteome samples were separated over 70 min using a 5%−32% acetonitrile gradient.
Full MS scans of m/z range 350-1200 were acquired at a resolution of 60,000, with an automatic gain control of 300% and maximum injection time 50 ms.The 20 most intense multiply charged precursors were selected for fragmentation.MS/MS scans were acquired using a resolution of 15,000, standard automatic gain control, a stepped collision energy (30,36), and automatic maximum injection time.Proteome samples were separated over 90 min using a 5%−35% acetonitrile gradient.
Full MS scans of m/z range 360-1500 were acquired at a resolution of 60,000, with an automatic gain control of 300% and maximum injection time 100 ms.The 20 most intense multiply charged precursors were selected for fragmentation.MS/MS scans were acquired using a resolution of 15,000, standard automatic gain control, a stepped collision energy (30,36), and automatic maximum injection time.
Raw files were searched against the UniProt Mus musculus database (25,342 sequences, September 27, 2021), using SEQUEST HT and Proteome Discoverer 2.5 software [12] (Thermo Fisher Scientific) as previously described [13].Search parameters allowed up to two missed cleavages, with a precursor mass tolerance of 10 ppm and fragment mass tolerance set to 0.02 Da.Cysteine carbamidomethylation was set as a fixed modification while dynamic modifications were phosphorylation (S/T/Y), acetylation (N-terminus, K), oxidation (M), methylation (R, K), deamidation (N/Q), and N-terminal methionine loss.Percolator [14] was used to filter the results to a 1% false discovery rate at the peptide level, using the target-decoy strategy."Minora Feature Detector" and "Feature Mapper" nodes were used to detect chromatographic peaks and match them across files as described [15].
The "Precursor Ions Quantifier" node was used for label-free precursor ion quantification.Abundances were normalised for variation in sample loading using the total peptide abundances in each sample.The protein list was filtered for entries with ≥1 unique peptides.Protein and phosphopeptide abundances were further median-normalised by taking the fold change over the sample median expression level.
At the phosphoproteome level, 7482 phosphopeptides from 2318 proteins were quantified in at least 3 replicates, again with high reproducibility (Pearson's correlation > 0.92, Figure 1B, Table S2).Expected ratios of serine, threonine, and tyrosine phosphosites were observed (Figure S1).Thirty-two percent of proteins quantified in the proteome were also identified in the phosphoproteome (Figure 1C).The quantified proteins were predominantly cytoplasmic, followed by nuclear, plasma membrane, and extracellular proteins (Figure 1D).Enzymes,   a Annotated using the matrisome database [37].
transporters, transcription regulators, and kinases were the most common functional categories (Figure 1E,F).
To identify the processes that are enriched in primary MEFs, we performed Ingenuity Pathway Analysis (IPA, Qiagen) on the proteins in the top quartile of abundance in the proteome and the phosphoproteome (Figure 2A, Tables S3,S4).The top 15 significantly enriched overlap between the proteome and the phosphoproteome, with eIF2/4 and mTOR signalling pathways enriched in both.mTOR signalling is critical in the fibroblast response to wound healing and inflammation [16][17][18].As expected, ECM and cytoskeleton related pathways were enriched in both the proteome and phosphoproteome; Integrin Signalling, Actin Cytoskeleton Signalling, Germ Cell-Sertoli Cell junction Signalling, and Signalling by Rho Family GTPases, including RhoA, important in actin polymerisation [19] (Figure 2A).Other significantly enriched pathways in the proteome included protein homeostasis pathways BAG2 signalling, FAT10 signalling, and the Protein Ubiquitination pathway (Figure 2A).Both the proteome and phosphoproteome contained enrichment of pathways associated with regulation of gene expression, with inhibition of ARE-mediated mRNA degradation pathway (proteome), and spliceosomal cycle (phosphoproteome).The phosphoproteome further showed enrichment of pathways associated with fibroblast proliferation and migration (Protein kinase A signalling [20], Insulin receptor signalling [21]).
The PhosphositePlus human and mouse databases [22] were used to map the upstream kinases for phosphosites in the top quartile of abundance.Kinases were classified as enriched if ≥ 5 substrates were identified [13].In accordance with the pathway analysis, PKA was enriched, along with mTOR signalling kinases mTOR, P70S6K, P90RSK, and RSK2; with P90RSK displaying the second highest average substrate abundance (Figure 2B).P38A kinase was enriched with the highest average substrate abundance (Figure 2B).P38A is integral to fibroblast function, with previously described roles in regulation of MMP expression [23], contact inhibition [24], fibrosis [25], reprogramming [26], invasion [27], migration and proliferation [28].Casein kinase 2 (CK2) subunit alpha had the third highest average substrate abundance (Figure 2B).CK2 has been previously described in TGF-β activation of fibroblasts [29,30].Protein-protein interactions between the upstream kinases were analysed using STRING [31], and visualised using Cytoscape 3.10.1 [32].This identified P38A (MAPK14) and mTOR as the upstream kinases with the highest number of interactions with other kinases (15 each, Figure S2), supporting that these kinases are integral to fibroblast function.
Given the cell surface proteins expressed by MEFs are of interest for co-culture experiments and for selecting cell markers for labelling cell types, we next analysed cell surface protein expression.Surface-Genie [33] was used to identify proteins with predicted cell surface localisation (Figure 2C, Table S5).As expected, integrin cell adhesion receptors were highly expressed, with Itgb1 and Itgav in the top 3 most abundant cell surface proteins (Figure 2C).The αvβ1 integrin heterodimer has been previously identified in fibroblasts, with a role in lung and liver fibrosis [34].Also involved in cell adhesion, Cd44, Ncam1, and Mcam featured in the top 15 most abundant cell surface proteins (Figure 2C).Solute carriers/ion transporters/channel proteins Atp1a1, Slc3a2, Atp1b3, and Gja1 were also highly abundant (Figure 2C).LDL receptor related protein 1 (Lrp1) and mannose receptor c type 2 (Mrc2) were highly expressed and also displayed cell type enhanced expression in fibroblasts in the Human Protein Atlas single cell atlas ( [35] Figure S3), suggesting potential application as fibroblast markers.Previously identified mesenchymal cell surface marker Pdgfra [36] was quantified but did not fall inside the top 15 most abundant (Table S5).
Given the enrichment of adhesion proteins identified in the pathway analysis, the widespread use of MEFs as feeder cell lines, and that fibroblasts are responsible for secretion of most of the ECM within tissues, we next annotated matrisome proteins identified in the proteome using the matrisome database [37] (Table 1).Of the core-matrisome proteins, top expressed collagens included previously identified fibroblast markers Col1a1, Col1a2, and Col5a1 (Table 1, [38]).The integral basement membrane protein heparan sulfate (Hspg2) was the topmost abundant proteoglycan (Table 1).The top-most abundant glycoproteins were fibronectin (Fn1), Thbs1, Fbln2, and Tnc (Table 1).Fbln2 has been previously reported as a fibroblast marker [38].Matrisomeassociated proteins include secreted factors, ECM regulators and ECM affiliated proteins.Of the secreted factors, calcium binding proteins S100a6, S100a4, and S100a11 were the most abundant (Table 1).
Other commonly used mesenchymal markers include the intermediate filament cytoskeletal protein vimentin [40] and cadherin adhesion protein cadherin-11 [41].In line with this, vimentin was the second most abundant protein identified overall (Table S1).Cadherin-11 was identified with medium level abundance (Table S1).
In conclusion, we have utilised sensitive, label-free proteomics to present a comprehensive proteome and phosphoproteome profile of WT primary MEFs.We have annotated active kinases, cell surface markers, and matrisome proteins, which include known and potentially novel fibroblast markers.Together, this work contributes to our knowledge of fibroblast function and provides a framework for future studies.

F I G U R E 1
Characterisation of wild-type primary mouse embryonic fibroblasts (MEFs).(A) Scatterplots of the log2 abundances and Pearson correlation coefficients demonstrate proteome reproducibility between MEF samples.(B) Scatterplots of the log2 abundances and Pearson correlation coefficients demonstrate phosphoproteome reproducibility between MEF samples.Correlations performed using Perseus [43].(C) Overlap in proteins represented in the proteome and phosphoproteome.Ingenuity Pathway Analysis revealed.(D) Cellular location of proteins identified.(E) Functional category of proteins in the proteome.(F) Functional category of phosphorylated proteins.Proteins assigned as "other" were excluded from E and F.

F I G U R E 2
Functional analysis of the wild-type primary MEF proteome and phosphoproteome.(A) Ingenuity pathway analysis identifies pathways enriched in proteins within the top quartile of abundance in the proteome and the phosphoproteome.The top 15 pathways for proteome and phosphoproteome are shown, excluding disease specific pathways.The full list of pathways is given in Tables S3,S4.(B) Kinase enrichment analysis was performed using the phosphosites in the top quartile of abundance.Bars, average log2 abundance of kinase substrates, numbers = number of substrates.(C) The top 15 most abundant cell surface proteins annotated by SurfaceGenie.TA B L E 1 Core matrisome and matrisome-associated proteins.a