Tetracycline Antibiotics Induce Biosynthesis of Pro-Inflammatory Metabolites in the Immunobiotic Bacteroides dorei

Abstract

The human gut microbiome consists of diverse microbes that communicate through small molecules. Numerous recent studies have demonstrated links between gut microbiota and host physiological processes; however, the underlying metabolites remain elusive in part because laboratory conditions do not replicate the native environment of these bacteria. Herein, we focused on Bacteroides dorei, a predominant and representative member of human gut microbiota, to interrogate the chemical composition and possible biological functions of its secondary metabolome. Using UPLC-MS-guided high-throughput elicitor screening (HiTES), we examined how the metabolome of this commensal bacterium responds to hundreds of FDA-approved drug molecules that the host may intake. We identified low-dose tetracyclines as pleiotropic inducers of the B. dorei secondary metabolome, leading to the identification and structural elucidation of six serine-glycine dipeptide lipids, named doreamides A-F, and two 6-N-acyladenosines. The induced doreamides and N-acyladenosines exhibited pro-inflammatory activities, upregulating tumor necrosis factor α (TNFα), interleukin (IL)-1β, IL-6, and IL-10 in macrophages. Doreamides also triggered production of cathelicidin, which inhibits the growth of multiple bacteria tested but not B. dorei. Our results show that low-dose antibiotics can perturb the secondary metabolome of gut bacteria, and that these induced metabolites can exert immunomodulatory effects and restructure the microbiome.

1

DMC serves as an inducer of secondary metabolism in B. dorei. (A) 3D plot of the secondary metabolome of B. dorei in response to 400 FDA-approved drugs. Metabolites are characterized by m/z and abundance as a function of the drug library. The m/z 417.2931 signals are highlighted with the red box. (B) Extracted ion chromatogram of m/z 417.2931 (doreamide A) in the presence or absence of 16 μM DMC. (C) Extracted ion counts of doreamide A as a function of elicitors, which are numbered. (D) Chemical structures of the top five elicitors (1–5) of doreamide A. (E) Dose-dependent induction of doreamide A by DMC. (F) IC₅₀ analysis of DMC against B. dorei. (G–H) Dose-dependent induction of two isomeric metabolites with m/z 352.16 by DMC. *, **, and *** denote differences between DMC-untreated and treated conditions at p < 0.05, p < 0.01, and p < 0.001, respectively. (I) Extracted ion chromatogram of m/z 352.16 in the presence or absence of 16 μM methacycline (6). Compounds 16 and 17, and conf -16 and conf -17 coelute. (J) Extracted ion counts of m/z 352.16 as a function of elicitors, which are numbered. (K) Chemical structures of the top five elicitors of m/z 352.16, including 6–9. See Table S2 and Figure S2 for elicitors labeled with Roman numerals in panels C and J.

2

Characterization of DMC-induced metabolites. (A) Chemical structures for doreamides A–C. Key NMR correlations used to solve the structures of 10–12 are shown (Tables S5, S8). (B) Structures of three additional derivatives (doreamides D–F, 13–15). The structure of variant E was elucidated by 1D/2D NMR, while those of variants D and F were deduced based on HR-MS and HR-MS/MS analysis. (C) Catabolic pathway of BCAAs and their incorporation into dipeptide lipids and adenosine. (D) Chemical structures for adenosine variants. Key NMR correlations used to solve the structures of 16 and 17 are shown (Tables S10, S11).

3

Doreamides induce a pro-inflammatory response in RAW 264.7 macrophages. (A) Effect of doreamides A–C (10–12) on the expression of genes associated with pro-inflammation, anti-inflammation, and immunogenetic regulators. (B–E) Change in serum levels of (B) TNFα and (C) IL-1β as well as (D) MCP-1 and (E) CAMP upon exposure to 10–12 or LPS for 24 h. (F) Effect of 24 h treatment of macrophages with 10–12 on the expression of TLR2 and TLR4. (G) Reduced expression of genes associated with pro-inflammation by treatment of CU-CPT22, a specific TLR2 antagonist, in the presence of 20 μg/mL 12. TLR2 agonist Pam3CSK4 was used as positive control. Data represent mean ± SEM. The averages of three and two independent biological replicates are shown for PCR analysis and ELISA assays, respectively. *, **, and *** denote differences between control and the indicated condition at p < 0.05, p < 0.01, and p < 0.001, respectively. #, ##, and ### indicate differences between Pam3CSK4- or 12-stimulated cells and their respective cotreatments with CU-CPT22 at p < 0.05, p < 0.01, and p < 0.001, respectively. Both sets of analyses were performed via a one-way ANOVA followed by a Bonferroni post hoc test. “ns” indicates not significant compared to unstimulated control.

4

N-Acyladenosines induce a pro-inflammatory response in RAW 264.7 macrophages. (A) Effects of N-acyladenosines 16 and 17 on the expression of genes associated with pro-inflammation, anti-inflammation, and immunogenetic regulators. (B) Effect of 17 or adenosine (Ado) on cytokine expression in LPS-activated RAW 264.7 macrophages. (C–F) Increase in serum levels of (C) TNFα and (D) IL-1β cytokines as well as (E) MCP-1 and (F) CAMP upon exposure to 16, 17, or Ado in LPS-activated RAW 264.7 macrophages. Data represent means ± SEM (n = 2–3). *, **, and *** denote the difference between control and the indicated point at p < 0.05, p < 0.01, and p < 0.001, respectively. #, ##, and ### denote the difference between LPS-stimulated cells and the indicated condition at p < 0.05, p < 0.01, and p < 0.001, respectively. Both sets of analyses were performed via a one-way ANOVA followed by a Bonferroni post hoc test.

5

Prevalence of the glsAB BGC in bacterial genomes. (A) The two-gene cluster coding for doreamide in the genome of B. dorei. (B) Proposed biosynthetic pathway for 10–15 and their acylated and dipeptide derivatives. The gene involved in the ligation of Ser remains to be identified (‘??’). Note, acylated derivatives, like lipid 654, are not observed in B. dorei. Moreover, dipeptide compounds 22 and 23 are not observed when heterologously expressing B. dorei glsAB in E. coli. (C) Phylogenetic tree of GlsB proteins and distribution of 841 glsA and/or glsB sequences; the two-gene operon is almost exclusively encoded in Bacteroidota.

6

Model for the pro-inflammatory effects mediated by B. dorei cryptic metabolites. Low-dose tetracycline antibiotics, identified using a forward chemical genetic screen, pleiotropically induce production of doreamides and 6-N-acyladenosines from B. dorei, all of which in turn induce synthesis of pro-inflammatory cytokines. The doreamides additionally trigger CAMP production, thereby augmenting the innate immune response and possibly reshaping the local microbiome.