Folate, one of the water-soluble forms of vitamin B9, is essential for the synthesis of DNA and RNA. It has long been known that processes involving rapidly proliferating cells, including embryogenesis1 and cancer2, are particularly sensitive to folate levels, and drugs that interfere with folate metabolism, such as methotrexate, are used to treat cancer. Bacteria rely on de novo folate synthesis, and a number of antibiotics, including the sulfonamides, block essential enzymes in folate metabolism3. Trimethoprim is a well-characterized dihydrofolate reductase (DHFR) inhibitor that has been in clinical use since the 1960s. In an elegant study described in this issue, Kwon et al.4 use metabolomics to gain new insights into the effects of trimethoprim on the folate metabolic network.

The power of LC-MS–based “omic” technologies is in the ability to take a systems biology view of the gestalt of a pathway, while being able to measure individual molecular species with tremendous analytical precision. For example, lipidomics, the branch of metabolomics that focuses on profiling lipid changes, has been used to define in vivo substrate-product relationships and to characterize the sequential production of multiple lipid messengers5,6,7. Advances such as these in the fields of proteomics and metabolomics are beginning to provide a more comprehensive and quantitative picture of the interconnectivity of cell signaling and metabolic networks.

Hess et al.8 and Yuan et al.9 have previously developed a kinetic flux profiling method, involving 15N-labeled ammonia, to simultaneously measure the cellular levels of folate and its precursors in Escherichia coli. Using this metabolomics approach, Kwon et al. looked at the effects of trimethoprim on 31 different metabolites in the folate pathway. In addition to the expected inhibition of DHFR, the results suggested that folylpoly-γ-glutamate synthase (FP-γ-GS), another enzyme in folate metabolism, was also being inhibited. Trimethoprim did not directly inhibit FP-γ-GS. Instead, the authors found that dihydrofolate, the substrate of DHFR, is a potent inhibitor of FP-γ-GS (Fig. 1). These results suggest a model in which inhibition of DHFR leads to the accumulation of dihydrofolate, subsequently resulting in FP-γ-GS inhibition. To validate this mechanism, the authors developed an ordinary differential equation model, involving some reasonable simplifications, of folate metabolism. The model accurately reproduced the complex, biphasic responses that were observed for many metabolites, which suggests that the proposed mechanism accurately describes most of the effects of trimethoprim. These results illustrate the dynamic connections in metabolic networks and reveal new insights into the mechanism of action for this well-established antimicrobial drug.

Figure 1: Metabolomics reveals that DHF inhibits FP-γ-GS.
figure 1

The effects of the antimicrobial agent trimethoprim on dihydrofolate reductase are well known, but as a secondary effect the resulting elevations of dihydrofolate (DHF) levels were shown to inhibit FP-γ-GS. This report demonstrates the power of LC-MS–directed profiling to achieve very precise analytical measurements of folate, DHF, tetrahydrofolate (THF) and dozens of other metabolites of this pathway, while simultaneously identifying novel actions of pharmacological inhibitors. The application of this technology to drug discovery has the potential to improve specificity and allow fine tuning of the benefits of synergist drug interactions.

Full size image

Owing in part to the essential functions of folate, antifolates often have undesirable side effects. Some of the more toxic effects of these drugs might be ameliorated by the development of new inhibitors that modulate folate metabolism in different ways. This is where the current report reveals new opportunities. The indirect effects of trimethoprim on FP-γ-GS were previously unknown. Beyond using stable isotope labels to dynamically follow intermediate metabolites to understand drug action, it may be possible to use metabolomic approaches in drug discovery to identify new compounds that have cleaner effects on metabolic pathways. Lead compounds could then be optimized with regards to desired effects on upstream and downstream metabolites.

One of the fundamental properties of biochemical pathways are feedforward and feedback loops. Metabolomic profiling using LC-MS/MS is facilitating the discovery of complex interactions within networks. As shown in this report, metabolomics has tremendous potential for better understanding the molecular mechanisms of pharmacological agents and for future applications for development of more specific agents with carefully calibrated outcomes on cellular networks. This work also illustrates that mass spectrometry–driven technology is delivering on its early promise to expand our knowledge of metabolic networks and open new avenues of understanding drug action. The next several years are likely to see many more exciting insights of this kind with increasing application to drug discovery and personalized medicine.