The term ‘antibiotics’ seems self-explanatory — these drugs are intended to kill or inhibit the growth of pathogenic bacteria and other microorganisms. However, as data accumulate on the broad, off-target effects of antibiotics and host-targeted drugs against the human microbiota1, and on the direct impact of antibiotics on mammalian cells2,3, it is becoming clear that a more comprehensive view of the mechanism of action and off-target effects of small-molecule drugs is urgently needed4. Reporting in this issue of Nature Microbiology, Park and colleagues show that antibiotics can activate cryptic biosynthetic pathways in gut bacteria, leading to the production of immunomodulatory compounds5 (Fig. 1). This work extends the list of fundamental mechanisms through which antibiotics alter immune cells independently from their inhibition of pathogenic microbes: through direct effects of antibiotics on host signalling pathways2; biotransformation of antibiotics to immunomodulatory metabolites6; depletion of microbial colonization and metabolite levels7; and now the activation of cryptic biosynthetic pathways (Fig. 1).

Fig. 1: General mechanisms through which antibiotics shape the immune system.
figure 1

a, Major categories include: direct effects on host signalling pathways (1); drug biotransformation to immunomodulatory metabolites (2); decreased microbial metabolite levels (3); and activation of microbial biosynthetic pathways (4). b, We highlight specific examples where each of these types of interactions has been implicated in a downstream immune response, including: ciprofloxacin increasing the host’s production of adenosine monophosphate (AMP) resulting in increased pathogen engulfment2 (1); E. coli metabolizing ampicillin into a metabolite that activates NF-κB6 (2); vancomycin disrupting the gut microbiota leading to decreased levels of propionate and thereby lower colonic regulatory T cells7 (3); and sub-inhibitory levels of SMX inducing a stress response in E. coli Nissle1917 resulting in the production of colipterins5 (4). Colipterins 4 and 6 promote IL-10 in macrophages and protect from the dextran sodium sulfate model of colitis. Figure created with BioRender.com.

Full size image

Through a series of elegant in vitro experiments, Park et al. show that sub-inhibitory levels of the antibiotic sulfamethoxazole (SMX) re-routes cellular metabolism in Escherichia coli towards a stress response which produces a set of previously uncharacterized immunomodulatory molecules, which they name the ‘colipterins’. However, the precise molecular mechanisms responsible remain to be determined. Prior structural characterization of the presumed target of SMX, dihydropteroate synthase, has focused on pathogenic bacteria such as E. coli8. The data in the current study by Park et al. are consistent with the hypothesis that SMX inhibits dihydropteroate synthase, based upon the shunting of compounds from the folate pathway to the monapterin pathway, resulting in the production of the colipterins. However, the source of the observed strain-level variation in SMX sensitivity and colipterin production remains unclear, as does the full scope of direct and indirect cellular targets. Of note, folate biosynthesis varies among gut bacterial taxa, motivating studies into the degree to which this pathway predicts the potential for SMX to more broadly shape the structure and function of the human gut microbiome.

Having discovered the colipterins, Park and colleagues sought to better understand their biological relevance to host immune responses. Unlike the gene-centric methods that have been more widely applied in human microbiome research9, the authors took a ‘compound-first’ approach. Gene- or compound-led strategies both suffer from similar limitations, as it is difficult to predict or identify the phenotypes that will arise from the expression of biosynthetic gene clusters or from the administration of purified compounds. In silico methods for choosing the phenotypes, cell types and assay conditions to test are lacking, forcing a more empirical ‘brute-force’ approach. Park et al. screened 12 human primary cell-based co-culture systems that model multiple tissues and disease states against the 6 different colipterins. Using immune markers as a read-out, they showed that colipterin treatment resulted in increased production of interleukin (IL)-8 and IL-10. They validated these findings in a macrophage cell line (THP-1) and primary bone marrow-derived macrophages. These results are fortuitous in two respects — both that there was any phenotype given the limitations of the screening used, and that the phenotype was specific to a small set of markers with clear links to disease. An alternative approach10 is to start with a phenotypic assay from cell-free supernatants followed by the use of activity-guided biochemical fractionation methods to identify the compound(s) responsible. More studies using each of these three complementary approaches (gene-, compound- and activity-first) are needed to decipher the chemical languages between the microbiome and host cells.

The cellular mechanisms responsible for the observed increase in IL-10 production in response to colipterins warrant additional studies. Based upon prior work on pterin family members, enzyme inhibition, redox cofactors and/or antioxidant activities could all play a role. Park et al. provide initial evidence for the radical scavenging activity of colipterins and propose that their antioxidant activity upregulates IL-10. Multiple host signalling pathways could also be involved, including the Toll-like receptor (TLR) and extracellular-signal-regulated kinase (ERK) pathways that regulate IL-10 levels in macrophages.

The authors go on to show that colipterins 4 and 6 lead to a significant decrease in the severity of colitis in mice due to increased IL-10 production (Fig. 1). These results are consistent with experiments with wild-type and mutant E. coli strains that vary in their ability to produce colipterins in response to SMX, suggesting that sufficient levels of colipterins are produced in vivo to ameliorate disease. However, more work is needed to determine if the observed effects of SMX are relevant in the context of a complex human gut microbial community, or if colipterin production leads to clinically meaningful changes to treatment outcomes for patients suffering from autoimmune disease. SMX is typically prescribed for short periods of time to treat acute infections, so any potential increase in colipterin levels is likely transient. Strain-level variations in the E. coli population may also affect the degree to which this phenomenon occurs: analyses of three representative strains by Park et al. (a probiotic, pathogen and commensal) support the conservation of this pathway in E. coli species but also highlight strain-level differences in SMX sensitivity and colipterin production.

The identification of the colipterins adds to the arsenal of microbiome-derived compounds that might be leveraged in drug development for the treatment of autoimmune disease. Given that the widely used probiotic E. coli Nissle1917 produces colipterins, it is conceivable that either cell therapies or drug formulations of colipterins could provide a much-needed source of stress relief for ourselves and for our microbial co-conspirators.