Mycobacterial Ser/Thr kinases play a critical role in bacterial physiology and pathogenesis. The challenge now lies in linking kinases to the physiological substrates they phosphorylate in vivo, thereby elucidating their exact functions. The aim of this work was to associate protein phosphorylation in mycobacteria with important subsequent macro cellular events by identifying the physiological substrates of PknG in Mycobacterium bovis BCG. The study compared the phosphoproteome dynamics during the batch growth of M. bovis BGC versus the respective PknG knock-out mutant (ΔPknG-BCG) strains. We employed TiO2 phosphopeptide enrichment techniques combined with label free quantitative phosphoproteomics work flow on LC/MS/MS. The comprehensive analysis of label free data identified 603 phosphopeptides on 307 phosphoproteins with high confidence. 55 phosphopeptides were differentially phosphorylated, of these, 23 phosphopeptides were phosphorylated in M. bovis BCG wild type only and not in the mutant. These were further validated through targeted mass spectrometry assays (PRM’s). The kinase-peptide docking studies based on a published crystal structure of PknG in complex with GarA revealed that the majority of identified p-sites presented docking scores close to that seen in previously described PknG substrates, GarA and ribosomal protein L13. Six out of the 23 phosphoproteins had higher docking scores than GarA, suggesting that the proteins identified here are truly PknG substrates. Based on protein functional analysis of the PknG substrates identified, the study confirms that PknG play an important regulation role in mycobacterial metabolism, but also indicated its association with the machinery of protein translation and folding.

Identified PknG substrates were validated by targeted mass spectrometry. Briefly, the discovery phosphoproteomic data was used to identify peptides with confidently localised phosphosites (phospho probability of >0.75) that were present in the M.bovis BCG Wt and absent in the PknG knock-out mutant. Fourteen of these peptides, derived from 7 proteins, were selected to validate the phosphoproteomic data. A spectral library was generated using the discovery phosphoproteomic data in Skyline (version 3.6.0.10162), with the best representative spectrum for each identified peptide and phosphopeptide. Retention times were calculated based on the average retention time observed in the discovery phosphoproteomic analysis. An isolation list was generated with a 10-minute retention time window around each peptide’s calculated retention time. This isolation list was used to carry out a 2-plex scheduled PRM analysis with 100 ms injection time and a total cycle time of 2 seconds on a QExactive hybrid Orbitrap mass spectrometer (Thermo). The AGC target was set to maximum, and a 2 m/z mass error window was allowed. Targeted MS2 data was acquired at a resolution of 35000. The chromatography setup was identical to that of the discovery phosphoproteomic analysis. The resulting PRM data was analysed in Skyline with the background M. bovis BCG database obtained from Uniprot (www.uniprot.com). The spectral library was used to confirm the identity of the targeted peptides.

M. bovis BCG reference strain (Pasteur 1172) and M.bovis BCG PKnG knock-out mutant strain generously donated by Prof Jean Pieters (Walburger et al., 2004) were grown n 7H9 DifcoTM Middlebrook liquid media (Becton Dickinson; BD), supplemented with OADC and Tween 80 to prevent clumping at 37 °C while shaking. Growth was monitored daily by measuring optical Density (OD600) and the growth curve plotted. Cells were harvested at mid-log (OD~ 0.6) by centrifugation for 10 min (4000 g) and washed twice in phosphate buffered saline (PBS) pH 7.4 (Sigma).