Fig 1.
Workflow of the proteometabolomics experiment.
(a) Cell lysate from Chaetomium thermophilum was separated by size exclusion chromatography and fractions with molecular weight between 200k Da and 5,000k Da were collected. (b) The collected fractions were divided into two parts, one part was digested by trypsin and analyzed by protein MS, the other part was extracted by methanol and analyzed by metabolite MS. Protein–metabolite interactions were inferred from the resulting intensity profiles.
Fig 2.
Identified proteins and metabolites.
(a) For proteins (top part) and metabolites (bottom part), the relation between theoretical molecular weight and observed molecular weight (according to their elution time) is shown. When proteins are annotated with the molecular weight of heteromeric complexes that they participate in, there is a good correlation between the theoretical and observed molecular weights. Notably, metabolites are observed at molecular weights far above their actual molecular weight. (b) Relative intensity profiles for one example each from the five classes of identified metabolites. (c) Relative intensities of the 20S and 26S proteasomes and co-eluted peptides.
Fig 3.
Distribution of correlations and intensities for protein–metabolite interactions.
(a) The distribution of protein−metabolite correlations is bimodal, with known interactions showing increased correlations. We chose a cutoff of 0.5 for interaction predictions. (b) Proteins of higher intensity (i.e. abundance) are enriched among the known interactions. In both panels, the difference between the distributions is highly significant (p ≤ 2e-15 using Mann-Whitney U test).
Fig 4.
The top 10 scoring protein–metabolite interactions.
Known interactions are shown in red, and proposed novel interactions are shown in black. One of the identified compounds, namely cyromazine, is an insecticide. It is not clear whether this compound has been introduced as part of the growth medium, or if this is a mis-annotation of an endogenous compound.
Fig 5.
The predicted protein–metabolite interactions.
(a) For three example proteins, predicted and known interactions with metabolites are shown. (b) The intensity score and correlation of all possible proteins which can interact with the metabolite isopentenyl adenine. Among all those proteins, the ribosome has the highest score. (c) Correlations between candidate proteins and isopentyl adenine are shown after independently performing IEX and SEC. Among all proteins, the ribosome has the highest score. (d) The experimental verification of the interactions between the ribosome and isopentenyl adenine: In the in vitro transcription system, luciferase was used as reported protein. Ribosome activity was measured by calculating the ratio of luciferase/non-specific protein.