The genome, the transcriptome and the proteome all receive a lot of attention, but there are also multitudes of small molecules in a cell, collectively making up the chemically diverse metabolome. Metabolites wear many hats, serving as enzyme substrates and products, cofactors, allosteric regulators, and as mediators of protein-complex assembly. But despite their essential roles in biological processes, their interactions with proteins—which are often transient and low-affinity—remain largely a mystery.

“Understanding how these interactions occur on a global scale is essential to understand mechanisms of cellular adaptation and ecosystems' dynamics,” says Paola Picotti of ETH Zurich in Switzerland. She and her colleagues recently designed a method to systematically discover which proteins bind to a metabolite of interest. The approach may also be useful for drug discovery, as a way to identify druggable sites in proteins and to test for off-target effects.

To identify protein–metabolite interactions in Escherichia coli, Picotti's team treated a whole-cell lysate with a metabolite of interest then added a low amount of the broad-specificity protease proteinase K for a short period of time, all under native conditions. This 'limited proteolysis' approach generates structure-specific protein fragments—metabolite binding can block proteinase K cleavage at locations which would otherwise be severed. Switching to denaturing conditions, the researchers then used the enzyme trypsin to completely digest the metabolite-treated sample and a reference untreated sample, generating peptides for label-free quantitative mass spectrometry analysis, and compared the resulting differential peptide spectral patterns. “The peptide-level resolution of our method enables pinpointing metabolite binding sites and effects on binding interfaces in case of protein complexes,” notes Picotti. “This information is very powerful because it enables predicting the nature of the interaction, especially if the structure of the proteins involved are known.”

Limited proteolysis (LiP) coupled with mass spectrometry reveals protein–metabolite interactions. Credit: Image reprinted with permission from Piazza et al. (Elsevier, 2018).

The researchers applied their method to assess 20 different E. coli metabolites, including amino acids, organic acids, sugar phosphates and nucleotides. They identified nearly 1,700 protein–metabolite interactions, of which more than 80% were novel, a particularly impressive result given that E. coli has probably the best-characterized metabolic network of any organism. The data provide clues about the biological activities of 76 proteins of completely unknown function.

A major challenge that Picotti's team faced was method validation. “Since our approach identifies a multitude of novel interactions for the metabolites we studied, an obvious question was: are they real?” says Picotti. Since it was not realistic to validate all novel interactions, they assessed method performance in several ways. They applied the approach to well-studied metabolites and a highly specific drug as a benchmark. They compared their results to a database of known enzyme-ligand interactions. They developed a scoring system to estimate the confidence of an interaction based on multiple lines of evidence. Finally, they performed biological experiments to validate several novel interactions.

Though the method appears to be unbiased with regards to metabolite chemistry, the researchers note that it will be difficult to identify metabolite interactions with low-abundance proteins. The compartmentalization found in eukaryotic cells could also make such analyses in higher-level organisms more complicated. Despite these challenges, the method has great potential to provide new insights into the essential roles of metabolites in the cell.