.
S O U R C E : Cell
HIGHLIGHTS
- A high-throughput method for investigating host-microbe-drug-nutrient interactions
- Metformin host effects are regulated by a bacterial nutrient signaling pathway
- Metabolic modeling of human gut microbiomes links metformin to microbial agmatine
- Metformin-bacterial interactions engage host lipid metabolism to extend lifespan
SUMMARY
Metformin is the first-line therapy for treating type 2 diabetes and a promising anti-aging drug. We set out to address the fundamental question of how gut microbes and nutrition, key regulators of host physiology, affect the effects of metformin. Combining two tractable genetic models, the bacterium E. coli and the nematode C. elegans, we developed a high-throughput four-way screen to define the underlying host-microbe-drug-nutrient interactions. We show that microbes integrate cues from metformin and the diet through the phosphotransferase signaling pathway that converges on the transcriptional regulator Crp. A detailed experimental characterization of metformin effects downstream of Crp in combination with metabolic modeling of the microbiota in metformin-treated type 2 diabetic patients predicts the production of microbial agmatine, a regulator of metformin effects on host lipid metabolism and lifespan. Our high-throughput screening platform paves the way for identifying exploitable drug-nutrient-microbiome interactions to improve host health and longevity through targeted microbiome therapies.
PLEASE WATCH ILLUSTRATIVE VIDEO HERE AT SOURCE.
The microbiota is widely acknowledged as a central regulator of host health (Kundu et al., 2017, Schmidt et al., 2018). Environmental cues, including drugs and diet, drive changes in microbial ecology and function (Maier et al., 2018, Rothschild et al., 2018) with important consequences for host health. However, the causal dynamics controlling these interactions are largely unknown. The biguanide metformin, a putative dietary restriction mimetic (Pryor and Cabreiro, 2015), is the most widely prescribed drug for type 2 diabetes. Unexpectedly, metformin treatment increases the survival of type 2 diabetic patients compared with matched healthy controls (Barzilai et al., 2016). The effects of metformin on host physiology are regulated by its interaction with the microbiota in an evolutionarily conserved manner, from C. elegans to humans (Bauer et al., 2018, Cabreiro et al., 2013, Forslund et al., 2015, Wu et al., 2017). For example, metformin treatment does not extend C. elegans lifespan in the absence of bacteria, when bacteria are metabolically impaired, or when bacteria develop resistance to the growth-inhibitory effects of metformin (Cabreiro et al., 2013). Nutrition also plays a key role in regulating both host and microbial physiology (David et al., 2014) as well as the efficacy of drugs in treating disease (Gonzalez et al., 2018). Indeed, the effects of metformin on host physiology are dependent on dietary intake (Bauer et al., 2018, Shin et al., 2014). However, the precise mechanisms by which microbes regulate these effects in a nutrient-dependent manner remain elusive.
Given the complexity of microbial metabolism and the myriad of metabolites of prokaryotic origin regulating host-related processes, understanding and harnessing their potential is a challenging task. Like humans, C. elegans hosts a community of gut microbes that acts as a central regulator of host physiology (Zhang et al., 2017). Recently, microbial metabolites of interest have been identified using animal models that allow direct high-throughput measurements of quantifiable and conserved host phenotypes that are directly regulated by microbes (Qi and Han, 2018). Moreover, similar to the human microbiota, C. elegans is dominantly colonized by enterobacteria (Lloyd-Price et al., 2017, Zhang et al., 2017), making it an ideal model for studying the effect of human gut microbes such as E. coli on host physiology and their function in mediating the response to host-targeted drugs (Cabreiro et al., 2013, Garcia-Gonzalez et al., 2017, Scott et al., 2017). Although many efforts have been made to develop techniques that further our understanding of the role of microbial genetics in host regulation, none exist to dissect the intricate relationships between nutrition, pharmacology, microbes, and host physiology.
Here we devise a high-throughput four-way screening approach to facilitate the evaluation of nutritional modulation of drug action in the context of the host-microbe meta-organism. Using this strategy, we identify a bacterial signaling pathway that integrates metformin and nutrient signals to alter metabolite production by the microbiota. Changes in metabolite production can, in turn, affect fatty acid metabolism in the host, altering the lifespan. Importantly, using a computational modeling approach, we show that these changes in metabolite production are also recapitulated in the microbiota of metformin-treated type 2 diabetic patients, providing a potential explanation for the pro-longevity effects of metformin in humans.
Four-Way Host-Microbe-Drug-Nutrient Screens Identify a Signaling Hub for the Integration of Drug and Nutrient Signals
We hypothesized that changing the nutritional context might alter the effects of metformin on bacterial growth and, in turn, modulate the metabolic and longevity response of C. elegans to metformin. Because metformin induces a dietary restriction-like state in C. elegans to regulate the organismal lifespan (Onken and Driscoll, 2010), we used the transgenic reporter C. elegans strain Pacs-2::GFP (Burkewitz et al., 2015), whose expression is an indicator of the transcriptional response under conditions of dietary restriction, to test this hypothesis. Acs-2 is an acyl-coenzyme A (CoA) synthase ortholog that mediates the activation of fatty acids for β-oxidation in response to dietary restriction. As predicted, the ability of metformin to impair bacterial growth (Figures 1A, S1A, and S1B), enhance host longevity (Figure 1B; Table S1), and increase the expression of Pacs-2::GFP (Figures 1C and S1C) varied dramatically according to metformin concentration. Critically, the magnitude of these effects differs depending on the growth medium, suggesting a nutritional input into this response (Figures 1B and 1C).
Figure 1Four-Way Host-Microbe-Drug-Nutrient Screens Identify a Signaling Hub for the Integration of Drug and Nutrient Signals
(A–C) The effects of metformin on bacterial growth (A), wild-type N2 worm lifespan (B), and metabolism © are dependent on drug dose, nutrients, and bacteria. OP50-MR is an E. coli OP50 strain that developed metformin resistance. As observed previously (Cabreiro et al., 2013), metformin does not extend the lifespan when worms are grown on OP50-MR. In (B), each data point corresponds to the mean lifespan of 80–154 worms. See also Table S1. In ©, each panel shows 8 individual worms.
(D) Diagram of the four-way host-microbe-drug-nutrient interaction screen.
(E) Nutrient effects on bacterial phenotype (growth, x axis) and on wild-type N2 worm phenotype rescue (Pacs-2::GFP expression, y axis) in response to metformin. The red fit line shows the correlation between metformin and nutrient effects in bacteria and worms. Antagonistic or synergistic refers to the type of interaction determined by linear modeling observed between metformin and nutrient effects, leading to an overall effect that is significantly greater than the sum of the effects of the two components alone either in C. elegans Pacs-2::GFP levels or E. coli growth. Positive fold changes indicate nutrient suppression of the effect of metformin in bacterial growth or C. elegans Pacs-2::GFP expression. Error bars represent SE. FDR < 0.05 for significance. All colored circles are statistically significant. Gray circles are non-significant. Effects of highlighted nutrients are provided in detail in Figures S1 and S2.
(F and G) EcoCyc metabolite class (F) and KEGG pathway (G) enrichment for the effects of nutrients on E. coli OP50 growth and worm Pacs-2::GFP expression in the context of metformin treatment.
Data are represented as mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.
See also Table S1 for lifespan statistics and Table S2 for screen statistics.
Figure S1Four-Way Host-Microbe-Drug-Nutrient Screens Identify a Signaling Hub for the Integration of Drug and Nutrient Signals, Related to Figure 1
(A) Bacterial growth of E. coli OP50 on different types of media with increasing concentrations of metformin. Shaded area shows mean growth OD ± SD.
(B) Bacterial growth of E. coli OP50-MR (metformin resistant) on different types of media with increasing concentrations of metformin. Shaded area shows mean growth OD ± SD.
© Pacs-2::GFP expression of worms grown on E. coli OP50 with different types of media and increasing concentrations of metformin. Significance stars represent comparison with 0 mM metformin for each media type.
(D) Comparison of nutrient effects on E. coli OP50 growth and worm Pacs-2::GFP expression in the context of metformin treatment.
(E and F) Correlation between nutrient rescue of worm Pacs-2::GFP fluorescence and nutrient effect on E. coli OP50 growth in control (E), and metformin treatment conditions (F). Nutrient supplementation without metformin (r2 = 0.057, p = 8.8 × 10−6) (E) nor nutrient supplementation with metformin (r2 = 0.097, p = 4.9 × 10−9) (F) does strongly predict the effects of metformin on host physiology.
(G) Strong correlation (r2 = 0.76, p = 6.0 × 10−6) between effects of nutrient supplementation on E. coli OP50 growth in control (x axis) versus metformin treatment conditions (y axis).
(H) Venn diagram of nutrients with significant effects on E. coli and/or worms in the context of metformin treatment.
(I) Top panel: Bacterial growth curves on base NGM media and with nutrient supplementation. Shaded area represents mean growth OD ± SD. Here and in following panels, red corresponds to control and purple to metformin treatment conditions. Middle panel: Examples of worm Pacs-2::GFP expression with the corresponding nutrient supplementation and the type of drug-nutrient interaction in worm response. Bottom panel: Histograms of worm Pacs-2::GFP expression in log2 scale, with distribution density shown on y axis. Shaded area shows worm brightness distribution SD for individual worms. Vertical lines indicate Q90 worm Pacs-2::GFP expression values. Red- Control and Blue – 50 mM metformin. Full lines- NGM control and dotted lines – NGM plus indicated nutrient supplementation. Full lines are represented in all conditions as a reference for direct comparison.
(J) Bacterial growth estimates based on log2 transformed AUC values (top) and worm Pacs-2::GFP expression estimates based on log2 transformed fluorescence brightness Q90 values (bottom). Dashed lines indicate bacterial growth on NGM and a worm Pacs-2::GFP expression level used as a reference. Arrows indicate metformin treatment and significant interaction effects (FDR < 0.05).
To investigate how specific nutrients affect metformin action on the host in a bacterium-dependent way, we developed a high-throughput four-way host-microbe-drug-nutrient screen that allowed us to map these interactions at an extensive scale (Figure 1D; see STAR Methods for details). Briefly, we determined the propensity of 337 specific nutrients to modify the effect of metformin on bacterial growth. This provides a simple readout of nutrient-metformin interactions at the bacterial level. Similarly, measuring the expression levels of the Pacs-2::GFP C. elegans reporter line, we determined the propensity of these 337 nutrients to modify the dietary restriction-like transcriptional and metabolic response in the host induced by metformin in the presence of bacteria. This allowed us to identify nutrients that act in the host in the context of metformin in a bacterium-dependent and -independent manner (Figures 1D, 1E, and S1D; Table S2). Fold change values in the metformin-dependent Pacs-2::GFP fluorescence response of worms in the presence of specific nutrients (y axis) were plotted relative to the bacterial growth fold change values (x axis) for the same condition (Figure 1E). Although E. coli growth and C. elegans phenotype rescue by nutrients were not fully predictive of each other (r2 = 0.08, p = 1.1 × 10−7; Figures 1E and S1E–S1G), a large subset of the nutrient-bacterium interactions strongly predicted the effects of metformin on host physiology (Figure 1E, red circles). We observed that 37 of 79 nutrients that significantly rescued metformin-induced impairment of bacterial growth also suppressed Pacs-2::GFP activation by metformin in worms, and 25 nutrients that suppressed Pacs-2::GFP activation by metformin had either a neutral or synergistic interaction with the effects of metformin on bacterial growth (Figure S1H). Taken together, these data suggest specific nutritional tuning of metformin effects on host metabolism through the bacteria.
Next we performed a nutrient EcoCyc metabolite class enrichment analysis on both the bacterial growth and Pacs-2::GFP host data (Figure 1F; Table S2) with the aim to identify nutrients that specifically rescue metformin effects on host physiology through the bacteria. From our metabolite enrichment analysis, nutrients belonging to the classes of amino sugars, peptides, amino acids (e.g., L-serine), and nucleotides (e.g., adenosine) significantly rescued the effects of metformin on bacterial growth without affecting the effects of metformin on host metabolism and lifespan (Figures 1E, 1F, S1I, S1J, S2A, and S2B). Conversely, carbohydrates, aldehydes, or carboxylates (e.g., D-glucose, D-ribose, and glycerol) rescued E. coli growth and abolished both the upregulation of Pacs-2::GFP and lifespan extension in worms in a bacterium-dependent manner, as demonstrated by the specific deletion of bacterial genes responsible for nutrient catabolism (Figures 1E, 1F, S1I, S1J, and S2C–S2L). This suggests the presence of specific processes in bacteria integrating the effects of nutrients and metformin to regulate host physiology. To identify these processes, we performed Kyoto Encyclopedia of Genes and Genomes (KEGG) ontology pathway analysis for both E. coli and C. elegans (Figure 1G; Table S2). This analysis revealed enrichment for the galactose and the phosphotransferase system (PTS) as key metabolic and signaling bacterial pathways, respectively, mediating the effects of metformin on the host. Altogether, our findings suggest a mechanism whereby the presence of specific metabolic and signaling pathways in bacteria function to integrate signals from both nutrition and drugs to regulate host metabolism.
Figure S2E. coli Integrates Drug and Nutritional Cues to Regulate Host Physiology, Related to Figure 1
(A and B) Supplementation with L-serine (A) or adenosine (B) does not suppress worm lifespan extension by metformin.
© Supplementation with glycerol rescues inhibition of bacterial growth by metformin in control E. coli OP50 but not in OP50 ΔglpK mutants unable to catabolize glycerol.
(D–G) Glycerol supplementation suppresses metformin-induced upregulation of Pacs-2::GFP expression (D-E) and abolishes lifespan extension (F-G) in worms in a bacteria-dependent manner. Nutrient effects are rescued by an E. coli OP50 ΔglpK mutant unable to catabolize glycerol. In (E), each panel shows 5 individual worms.
(H) Supplementation with D-ribose rescues inhibition of bacterial growth by metformin in control E. coli OP50 but not in OP50 ΔrbsK mutants unable to catabolize D-ribose.
(I–L) D-ribose supplementation suppresses metformin-induced upregulation of Pacs-2::GFP expression (I-J) and abolishes lifespan extension (K-L) in worms in a bacteria-dependent manner. Nutrient effects are rescued by an E. coli OP50 ΔrbsK mutant unable to catabolize D-ribose. In (I), each panel shows 5 individual worms.
Data are represented as mean ± SEM unless otherwise stated. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001. For C, D, H, and J, significance stars represent metformin effect (purple) and metformin-nutrient interaction (green). See also Table S1 for lifespan statistics and table S2 for screen statistics.
.../...
FOR THE REST OF THE STUDY, PLEASE VISIT THE SOURCE
