Novel interconnections of HOG signaling revealed by combined use of two proteomic software packages
Modern quantitative mass spectrometry (MS)-based proteomics enables researchers to unravel signaling networks by monitoring proteome-wide cellular responses to different stimuli. MS-based analysis of signaling systems usually requires an integration of multiple quantitative MS experiments, which remains challenging, given that the overlap between these datasets is not necessarily comprehensive. In a previous study we analyzed the impact of the yeast mitogen-activated protein kinase (MAPK) Hog1 on the hyperosmotic stress-affected phosphorylome. Using a combination of a series of hyperosmotic stress and kinase inhibition experiments, we identified a broad range of direct and indirect substrates of the MAPK. Here we re-evaluate this extensive MS dataset and demonstrate that a combined analysis based on two software packages, MaxQuant and Proteome Discoverer, increases the coverage of Hog1 target proteins by 30%. Using protein-protein proximity assays we show that the majority of new targets gained by this analysis are indeed Hog1 interactors. Additionally, kinetic profiles indicate differential trends of Hog1-dependent versus Hog1-independent phosphorylation sites. Our findings highlight a previously unrecognized interconnection between Hog1 signaling and the RAM signaling network, as well as sphingolipid homeostasis.
Sample Processing Protocol
Mass spectrometric shotgun experiments were performed as described in Romanov et al., 2017. Briefly, cells were harvested by filtration and protein samples were prepared using Trizol extraction. Tryptic digest samples were subjected to TiO2 enrichment (M. Mazanek et al. 2007, T. E. Thingholm et al. 2006). To increase phosphoproteome coverage prior to MS analysis, phosphopeptides were fractionated offline by SCX chromatography. SCX fractions were analyzed by online reverse-phase chromatography with a nanoflow HPLC system coupled with an electrospray ionization interface to acquire MS and MS/MS scans. HeLa samples were kindly provided by Karl Mechtler. Briefly, cells were harvested, washed with 1M PBS, suspended in lysis buffer (8M urea, 50mM TrisHCl pH8, 150mM NaCl, 1mM PMSF, complete protease inhibitor, benzonase), and subsequently disrupted by sonification. Extracts were cleared by centrifugation (15000xg, 10min, 4°C) and proteins were precipitated by adding 5x excess of 100% ice-cold acetone (Applichem) (overnight, -20°C) and pelleted by centrifugation 15000xg, 30 min, 4°C). The pellet was re-suspended in 80% ice-cold acetone, centrifuged for 5min at 15000xg, air-dried for 5min and subsequently suspended in urea buffer (8M urea, 0.5M ammoniumbicarbonate). Soluble proteins were reduced with dithiothreitol (DTT) and alkylated using iodoacetamide (IAA), digested first with Lys-C for 2h at 30°C, and then with trypsin overnight at 37°C. HeLa samples were measured in an HPLC-MS/MS-setup using a Q Exactive (higher-energy collisional dissociation mode) mass spectrometer (Thermo Fisher Scientific). Mass spectrometry-based screen for probing of phosphorylation kinetics: SILAC-labelled cells were harvested by filtration, immediately deep frozen in liquid N2 and resuspended in TRIzol reagent (Invitrogen) for protein extraction [Reiter et al., 2012; Romanov et al., 2017]. Following TRIzol purification [Reiter et al., 2012], proteins were subjected to dithiothreitol (DTT) and iodoacetamide, and tryptic digestion. After desalting on Strata-X 33µm Polymeric Sorbent (8B-S100-TAK columns, Phenomenex) and drying, peptide carboxyl groups were esterified in methanolic HCl as described in [Ficarro et al., Nat Biotechnol 2002]. Esterified peptides were dried, redissolved in 30% ACN/30% methanol/40% H2O and incubated for one hour with 40μl PHOS-SelectTM iron affinity resin (Sigma), washed with 0.003% acetic acid, and eluted with 50-125mM Na2HPO4 (pH 6.0). Eluates were analyzed on an UltiMate TM 3000 Dual LC nano-HPLC System (Dionex, Thermo Fisher Scientific) coupled to a hybrid linear ion trap/Fourier transform ion cyclotron resonance mass spectrometer (LTQ-FT, Thermo Fisher Scientific), applying settings described previously [Reiter et al., 2012; Romanov et al., 2017].
Data Processing Protocol
Proteome Discoverer Analysis: Data analysis was performed using the SEQUEST algorithm (Proteome Discoverer 1.3 and 1.4) using the Saccharomyces Genome Database (SGD) (version February 2011) along with contaminants derived from common laboratory contaminants database (MQ). Fixed modifications included carbamidomethylation of cysteine, whereas variable modifications encompassed protein N-terminal acetylation, deamidation, oxidation of methionine, phosphorylation of serine, threonine and tyrosine, and heavy labels of arginine and lysine (Arg6, Lys6). Enzyme specificity was set to “Trypsin” and a maximum of 2 missed cleavages per peptide was allowed. For the assignment of phosphorylation sites we integrated the tool phosphoRS into the Proteome Discoverer pipeline, and considered 70% phosphorylation probability as an adequate threshold for phosphorylation site assignment. We performed the SEQUEST analysis against the SGD database, as well as a decoy database (reversed sequences) and calculated an empirical FDR<1% at the level of peptide spectrum matches (PSMs). Separately, we calculated an FDR at peptide and protein level as well (FDR<1%). To quantify phosphorylation events accurately, we performed a phosphorylation site group as explained in detail in the section “Phosphorylation site groups”. We considered potential arginine-to-proline conversion by calculating a correction factor based on the SILAC ratio biases observed for peptide groups that are differential in the number of prolines. SILAC Heavy-to-Light ratios were accordingly corrected, log2-transformed, and additionally summarized at the level of phosphorylation site groups. More details on the pipeline if required can be extracted from the individual search files deposited at PXD004294 to PXD004300. MaxQuant Analysis: MaxQuant (version 220.127.116.11) re-analysis was performed using default parameters, with following features: Saccharomyces Genome Database (SGD) (version February 2011) was used in combination with common laboratory contaminants database (MQ) for peptide spectrum matching. Modifications, such as protein N-terminal acetylation, deamidation of asparagine and glutamine, oxidation of methionine, and phosphorylation of serine, threonine and tyrosine were set as variable, whereas carbamidomethylation of cysteine was set as fixed. A maximum of 5 variable modifications per peptide was allowed. Enzyme specificity was set to “Trypsin/P” and a maximum of 2 missed cleavages per peptide was allowed. Heavy labels (‘Arg6’, ‘Lys6’) were specified, ‘Requantify’ and “Match between runs” was activated. The option to treat leucine and isoleucine as indistinguishable was activated. Computational processing, log2-transformation of SILAC ratios and correction for arginine-to-proline conversion was performed as described in Romanov et al., 2017. Phosphopeptides were filtered for phosphorylation site assignment probability ≥ 70% and grouped by phosphorylated residues. Note that for HeLa-cells we were not considering phosphorylations as variable modifications. Mass spectrometry-based screen for probing of phosphorylation kinetics: The obtained spectra were searched both by SEQUEST in the Proteome Discoverer 1.4 software package (Thermo Fisher Scientific) and MaxQuant 18.104.22.168 against the SGD database (version February 2011) plus contaminants, with similar settings as described above.
Wolfgang Reiter, MAX F. PERUTZ LABORATORIES - University of Vienna
Wolfgang L. Reiter, Mass Spectrometry Facility, Max F. Perutz Laboratories, University of Vienna, Vienna BioCenter, Vienna, Austria ( lab head )
Romanov N, Hollenstein DM, Janschitz M, Ammerer G, Anrather D, Reiter W. Identifying protein kinase-specific effectors of the osmostress response in yeast. Sci Signal. 2017 10(469) PubMed: 28270554