Integrative Phosphoproteomics Links IL-23R Signaling with Metabolic Adaptation in Lymphocytes

Interleukin (IL)-23 mediated signal transduction represents a major molecular mechanism underlying the pathology of inflammatory bowel disease, Crohn’s disease and ulcerative colitis. In addition, emerging evidence supports the role of IL-23-driven Th17 cells in inflammation. Components of the IL-23 signaling pathway, such as IL-23R, JAK2 and STAT3, have been characterized, but elements unique to this network as compared to other interleukins have not been readily explored. In this study, we have undertaken an integrative phosphoproteomics approach to better characterise downstream signaling events. To this end, we performed and compared phosphopeptide and phosphoprotein enrichment methodologies after activation of T lymphocytes by IL-23. We demonstrate the complementary nature of the two phosphoenrichment approaches by maximizing the capture of phosphorylation events. A total of 8202 unique phosphopeptides, and 4317 unique proteins were identified, amongst which STAT3, PKM2, CDK6 and LASP-1 showed induction of specific phosphorylation not readily observed after IL-2 stimulation. Interestingly, quantitative analysis revealed predominant phosphorylation of pre-existing STAT3 nuclear subsets in addition to translocation of phosphorylated STAT3 within 30 min after IL-23 stimulation. After IL-23R activation, a small subset of PKM2 also translocates to the nucleus and may contribute to STAT3 phosphorylation, suggesting multiple cellular responses including metabolic adaptation.

Label free quantitation and data analysis with Progenesis.

Analysis by tandem mass spectrometry
Dried peptide samples were resuspended in 2 % acetonitrile and 0.1 % trifluoroacetic acid and analyzed by nano-liquid chromatography tandem mass spectrometry (nano-LC-MS/MS) as described previously (1). In brief, samples were separated using a nanoUPLC (Easy spray C18 column with a 75 μm × 500 mm, 2.1 μm particle size; Thermo-Fisher) coupled to a Q Unstimulated/stimulated samples were digested as described above and the peptides labelled with two different dimethyl labels per the table below; note the inverted label strategy between dataset A and datasets B and C. Analysis of the MS data for these phosphopeptide datasets was performed using Proteome  Table   S1. Finally, stimulated/unstimulated ratios were normalized to correct for mixing error by dividing all values by the median of all peptides exported from PD (normalization factor for datasets: A: 1.10, B: 1.14, C: 0.95).

Statistical analysis of differential phosphorylation at the peptide & protein level
We first compared the phosphopeptide and the phosphoprotein datasets like-for-like, calculating the log2 fold change for 30 min versus 0 min. In order to apply a consistent method of quantitation thresholding for both the IMAC phosphopeptide and the phosphoprotein datasets (30 versus 0 min IL-23 stimulation), we used the 'significance A' pvalue described for the MaxQuant quantitation package (2). In brief, for each measured changed (phosphopeptide or phosphoprotein), we computed the probability of obtaining a log-ratio of at least that magnitude under the null hypothesis that the distribution of log-ratios had normal upper and lower tails ( Figure S1a, b). As per the MaxQuant method, separate estimators of standard deviation were used for the upper and lower tails to allow thresholding to be more robust in the face of bias toward up-or down-regulation. We defined as significant any points with a false discovery rate <= 0.01, calculated from the p-values using the Benjamini-Hochberg-Yekutieli (BHY) procedure for controlling error rate / correction for multiple testing (3). Additional statistical analysis was performed taking into account the full phosphoprotein dataset that consisted of a time-course with data points at 0, 5, 15 and 30 min. Log2-fold changes were standardized (normalized to zero mean and unit variance). We first analyzed variance in the data set by PCA, marking outliers based on BHY-corrected chisquare p-values based on the Mahalanobis distance from the center of the projection ( Figure   S1c) (4,5). Secondly, to cluster phosphoproteins by similarity across the whole time course, we used the Mfuzz soft clustering package in R (6) based on its previous deployment in phosphoprotein time-course analysis (7). The included 'mestimate' function was used to calculate the fuzzifier parameter (8), and a cluster number parameter of 16 was selected from a range of target cluster numbers from 4 through 40 as the point at which the minimum centroid distance for each cluster number moved from rapid decline to slow decline.
Clustering was repeated for 100 iterations and the iteration with the highest group membership score for STAT3 was selected ( Figure S1d-f). Primary and secondary antibodies were prepared in Odyssey Blocking buffer (LICOR Biosciences) and washing steps were performed using 1 x PBS/0.05 % Tween-20.

Subcellular fractionation, immunoprecipitation and immunoblotting
Phosphospecific and total antibodies were incubated 1 h RT in sequential steps followed by incubation with IRDye labelled secondary antibodies for 1 h, RT, with three washes between.
When using 2 antibodies generated in the same species, phosphospecific and secondary antibody were stripped off with ReBlot Plus strong antibody stripping solution (Millipore) before blocking and incubation with total antibody and respective secondary antibody as described. Blots were scanned using the Odyssey CLx infrared imaging system and band intensities analyzed with the Image Studio software (LICOR Biosciences).

Gene expression analysis
For gene expression analysis, 5 x 10 6 cells were rested as described above, resuspended in full medium at a cell density of 1 x 10 6 cells/ml and either left unstimulated or stimulated with IL-23 or IL-2 (10 ng and 20 ng (360 U)/5 x 10 6 cells). After 24 h RNA was isolated using RNeasy Mini Kit (Qiagen) and reverse transcribed using a mix of oligo(dT) and random pentadecamer primers followed by amplification of target genes using unlabelled primers and SYBR® Green RT PCR Kit (Applied Biosystems). Primers were designed to span exon intron barriers and specificity of product amplification ensured by measuring melting curves. Gene expression data were analysed using the Delta C(t) method (9).
Statistics applied was ratio paired t test using log transformed data with p < 0.05 being statistically significant.

Lactate measurement by GCxGC-MS
For intracellular lactate measurement 2.5 x 10 6 rested cells were left unstimulated or stimulated either with IL-23 (see above) or with Dynabeads coated with αCD3/αCD28 beads (Invitrogen, 1.25 beads per cell) for 24 h. There was no difference in cell numbers between the different conditions (unstimulated versus stimulated) after 24 h. Extraction of metabolites was performed by treating cells with 400µl methanol/H2O (1:1) and homogenization using washed glass beads in a bead beater (Precellys 24, Bertin Technologies) for four cycles (6500 Hz, 45s), followed by the addition of 1 ml tert-butyl ether (MTBE) to extract metabolites.
After vortexing for 5 min and centrifugation for 20 min at 13000 g and 4°C, the organic phase was transferred to a glass vial and dried under vacuum. To the remaining aqueous phase, 800µl of methanol was added, samples vortexed for one cycle (6500 Hz, 45s), kept at -80°C for one hour and centrifuged for 20 min at 13000 g and 4°C. 1ml of aqueous phase was removed, added to the glass bead vial containing the organic phase and the samples dried under vacuum. Chemical derivatization was performed essentially as described (10). In brief, samples were resuspended in 50µl of a 20µg/µl solution of methoxyamine hydrochloride in pyridine and incubated for 90 min at 30°C and shaking (1200 rpm). 70µl of N-Methyl-Ntrimethylsilyltrifluoroacetamide (MSTFA) with 1% chlorotrimethylsilane (TMCS) were added to the samples, followed by incubation for one hour at 60°C and shaking (1200 rpm).
The samples were immediately analysed using a GCxGC-MS system comprising of a gas chromatograph coupled to a quadrupole mass spectrometer (Shimadzu GCMS QP2010 Ultra)