Outer Membrane Vesicles Released from Garlic Exosome-like Nanoparticles (GaELNs) Train Gut Bacteria that Reverses Type 2 Diabetes via the Gut-Brain Axis

Abstract

Gut microbiota function has numerous effects on humans and the diet humans consume has emerged as a pivotal determinant of gut microbiota function. Here, a new concept that gut microbiota can be trained by diet-derived exosome-like nanoparticles (ELNs) to release healthy outer membrane vesicles (OMVs) is introduced. Specifically, OMVs released from garlic ELN (GaELNs) trained human gut Akkermansia muciniphila (A. muciniphila) can reverse high-fat diet-induced type 2 diabetes (T2DM) in mice. Oral administration of OMVs released from GaELNs trained A. muciniphila can traffick to the brain where they are taken up by microglial cells, resulting in inhibition of high-fat diet-induced brain inflammation. GaELNs treatment increases the levels of OMV Amuc-1100, P9, and phosphatidylcholines. Increasing the levels of Amuc-1100 and P9 leads to increasing the GLP-1 plasma level. Increasing the levels of phosphatidylcholines is required for inhibition of cGas and STING-mediated inflammation and GLP-1R crosstalk with the insulin pathway that leads to increasing expression of Insulin Receptor Substrate (IRS1 and IRS2) on OMV targeted cells. These findings reveal a molecular mechanism whereby OMVs from plant nanoparticle-trained gut bacteria regulate genes expressed in the brain, and have implications for the treatment of brain dysfunction caused by a metabolic syndrome.

Keywords: Akkermansia muciniphila; GLP‐1R; Garlic exosome‐like nanoparticles (GaELN); Gut bacteria; Gut‐Brain axis; Obesity; Outer Membrane Vesicles (OMVs); Type 2 diabetes (T2DM); cGas‐STING signaling.

Figure 1.. GaELNs gavage-given to HFD fed mice shapes the composition of gut bacteria and increases the abundance of gut A. muciniphila.

(A). GaELNs were labelled with PKH26 and orally gavaged into naïve male and female mice. After 3 h, gut bacteria were isolated and FACS sorted for PKH26 positive bacteria that took up the GaELNs. DNA was isolated from these bacteria and subjected to 16s sequencing. Heatmap shows the composition of bacteria in family level. PKH26 positive bacteria were listed in supplementary table S1. (B). Calculated percentage of bacteria that take up GaELNs in phylum level. (C). Fecal DNA isolated from lean, HFD and GaELN treated HFD fed mice. The level of A. muciniphila, E. rectale and F. prasuntzii was determined by real-time PCR. The fold changes were normalized based on expression of 16s RNA. n = 3. (D). A. muciniphila was treated with GaELNs for 1 h and the bacteria were plated on agar plates. Colony forming units (CFU) were determined. n = 3. (E). A. muciniphila were treated with GaELNs for 24 h. The growth of A. muciniphila was determined by measuring the OD at 600 nm (n = 3). P values were calculated by means of an ANOVA test. *P < 0.05, **P < 0.01, and ***P < 0.001.

Figure 2.. GaELNs increase OMV secretion in A. muciniphila via induction of OmpH gene expression.

(A). The size and number of OMVs released from control and GaELN treated A. muciniphila was determined using a NanoSight 300 (left panel) and A. muciniphila was treated with GaELNs for 1 h and the bacteria were plated on agar plates. Colony forming units (CFU) were determined (right panel). n = 5. (B). Total lipids were extracted from OMVs released from A. muciniphila and separated by TLC. (C). Total RNA isolated from WT and OmpH knockout A. muciniphila was subjected to real-time PCR for OmpH gene expression. The fold changes were normalized based on expression of 16s RNA. n = 3. (D). WT and OmpH knockout A. muciniphila was treated with and without GaELN for 24 h. OMVs were released from these bacteria and the number was determined by NTA. n = 8 per treatment. P values were calculated by means of an ANOVA test. ***P < 0.001.

Figure 3.. OMVs released from A. muciniphila trafficking to the brain are taken up by microglial cells.

(A). The number of OMVs released from control and GaELN treated A. muciniphila was determined using a NanoSight 300. n = 5 per treatment. (B). OMVs released from A. muciniphila treated with or without GaELN, lysed with SDS lysis buffer and the proteins were separated by SDS-PAGE and stained with Coomassie brilliant blue. Cell lysates were subjected to MS/MS analysis. (C). OMVs released from A. muciniphila were labelled with DiR dye. The labelled nanoparticles (1×10¹⁰) were orally gavage given to HFD fed mice. Distribution of OMVs in different organs. (D). OMVs released from A. muciniphila treated with or without GaELN were labeled with PKH26 and orally given to lean, HFD and GaELN treated HFD fed mice for 3 h. Microglial cell uptake of OMVs was visualized by confocal microscopy. Microglial cells were stained for IBA-1 expression and the nucleus was stained with DAPI. (E). Blood brain barrier function was determined. FITC-BSA was intraperitoneally injected into lean, HFD and GaELN treated HFD fed mice. 1 h after the injection, localization of FITC-BSA was visualized using confocal microscopy. (F). Total cell lysates were obtained from these mice and the fluorescence intensity quantified (n = 5 mice per group) (G). Total RNA isolated from lean, HFD and GaELN treated HFD mice brains were subjected to real-time PCR for Caludin-1, Claudin-3 and ZO-1 mRNA expression. The mRNA expression was normalized by actin mRNA expression (n =5). P values were calculated by means of an ANOVA test. *P < 0.05, **P < 0.01, and ***P < 0.001.

Figure 4.. A. muciniphila OMVs traffick to microglial cell mitochondria and inhibit the expression of inflammatory cytokines.

(A). Pro-inflammatory and anti-inflammatory cytokines levels in the brain of mice treated with controls and OMVs released from A. muciniphila treated with/without GaELNs were quantified by ELISA. n = 5. (B). U937 cells were treated with LPS (10 ng/ml) and stimulated with OMVs released from A. muciniphila for 24 h. Pro-inflammatory and anti-inflammatory cytokines levels in the brain were quantified by ELISA. n = 5 per treatment. (C). BV2 cells were treated with PKH26 labelled OMVs released from controls and GaELN treated A. muciniphila. Mitochondrial localization of OMV was visualized by confocal microscopy. Mitochondria were stained with a mitochondrial marker and the nucleus was stained with DAPI. (D). Identification of OMV binding protein in mitochondria of microglial cells. Total lipids were isolated from A. muciniphila OMVs and bound to the LIP-1 sensor chip. Mitochondrial protein was extracted from BV2 cells and used as the analyte. The bound protein was eluted using 200 mM NaOH and bound mitochondrial protein was identified by MS/MS analysis. P values were calculated by means of an ANOVA test. **P < 0.01, and ***P < 0.001.

Figure 5.. GaELNs modulate lipid composition of A. muciniphila OMVs that inhibit the expression of cGAS-STING and inflammatory cytokines in microglial cells.

(A). Total lipids from control and GaELN treated A. muciniphila OMVs were extracted and separated by TLC and stained using iodine vapor as described in the Methods. (B). Lipid nanoparticles were made from lipids extracted from TLC plates by removing each band and pooling together the remaining lipids. The lipid nanoparticles were labelled with PKH26 and incubated for 24 h with BV2 cells. Uptake of lipid nanoparticles was determined by flow cytometry. (C). GaELN treatment modulates lipid composition of OMVs from A. muciniphila. (D). BV2 cells were treated for 24 h with brain metabolites (100 μL/mL) of HFD fed mice and band 2 lipids from OMVs and PC (16:0) (1 μg/ml). Total RNA isolated from these cells was subjected to real-time PCR for cGAS and STING mRNA expression. mRNA expression was normalized to actin mRNA expression. cGAMP level was quantified by ELISA (n = 5). (E). BV2 cells were treated for 24 h with brain metabolites (100 μL/mL) of HFD fed mice and band 2 lipids from OMV and PC (16:0) (1 μg/ml). Total RNA isolated from these cells was subjected to real-time PCR for IFN-γ, IL-1β, TNF-α and IL-10 mRNA expression. mRNA expression was normalized to actin mRNA expression (n = 5). P values were calculated by means of an ANOVA test. *P < 0.05.

Figure 6.. Oral administration of GaELNs trained A. muciniphila contributes to reversing insulin resistance.

(A). Germ-free (GF) mice were orally given A. muciniphila (1×10⁹) pretreated with or without GaELNs for 1 h and washed. After 48 h, small and large intestinal content was collected, plated on agar plates and colony forming units (CFU) were determined. n = 5 per group. (B). A. muciniphila were treated for 24 h with GaELNs. The growth of A. muciniphila was determined by measuring the OD at 600 nm. n = 3 per treatment. (C). Different concentrations of PKH26 labeled GaELNs (0–2.5 ×10⁷) were incubated with A. muciniphila for 1 h at 37°C in an anaerobic chamber. The uptake of GaELNs by A. muciniphila was determined using flow cytometry. (D). Human fecal bacteria were isolated from healthy volunteers and patients with type 2 diabetes and 10¹⁰ bacteria were orally given to GF mice. The mice were treated with or without GaELNs for 2 weeks. The level of A. muciniphila, E. rectale and F. prasuntzii was determined by real-time PCR. The fold changes were normalized based on the expression of 16s RNA (n = 5). (E). Fecal microbial content was collected from the healthy volunteers and patients with T2DM and orally given to GF mice for 2 weeks. Glucose tolerance and insulin resistance assays were conducted (n = 5). (F). Fecal bacteria was collected from HFD fed mice and GaELN treated HFD fed mice and orally given to GF mice for 2 weeks. Glucose tolerance and insulin resistance assays were conducted (n = 5). (G). Normal B6 mice were treated with an antibiotic cocktail for two weeks then orally given the fecal microbial content from lean, HFD fed mice and GaELN treated HFD fed mice for 2 weeks. Then mice were orally given GaELNs for 2 weeks and the glucose and insulin tolerance tests were carried out as described in the Methods (n = 5). P values were calculated by means of an ANOVA test. *P < 0.05, **P < 0.01, and ***P < 0.001.

Figure 7.. OMV Amuc_1100, P9, and PC contribute to reversing insulin resistance in HFD fed mice.

(A-B). A. muciniphila was treated with and without GaELNs for 24 h. Total RNA was isolated from these bacteria and subjected to real-time PCR for Amuc_1100 and P9 mRNA expression. n = 3 per treatment. (C). BV2 cells were stimulated with metabolites derived from HFD mice brain (HFD-meta) and treated with OMV derived from WT and P9 KO A. muciniphila for 24 h. cGAMP level was determined by ELISA from total cell lysates from the BV2 cells. n = 3 per treatment. (D). HFD fed mice were orally gavaged with OMVs (1 × 10¹⁰ particles/ml) released from WT and Amuc_1100 and P9 knockout A. muciniphila for 6 weeks. Glucose tolerance and insulin resistance assays were determined in these mice as described in the Materials and Methods (n = 5). (E). Quantification of GLP-1 level in the plasma, of lean, HFD and OMV treated HFD fed mice (n = 5 mice /group). (F). GLP-1R mRNA and protein expression in lean, HFD and OMV treated HFD mouse brains by real-time qPCR and western blot analysis, respectively. (G). BV2 cells were treated with HFD brain metabolites (100 μl/ml) with and without OMV derived from A. muciniphila. Total RNA isolated from these cells were subjected to real-time PCR for GLP-1R mRNA expression (n = 5). (H). BV2 cells were transfected with CRISPR/cas9 GLP-1R plasmid to knockout GLP-1R expression and cells were treated with HFD brain metabolites with and without OMV derived from A. muciniphila. Total cell lysates were subjected to western blot analysis for IRS1, IRS2 and HIF-1α expression. (I). BV2 cells were transfected with scrambled plasmid and GLP-1R CRISPR/Cas9 plasmid and stimulated with metabolites (100 μl/ml) derived from HFD mice and OMVs for 24 h. Total mRNA isolated from these cells were subjected to real-time qPCR analysis for IRS-1 and IRS-2 expression (n=3). (J). Glucose uptake was determined in these cells. (K). BV2 cells were treated for 24 h with PKH26 labelled lipid nanoparticles from band 2 lipids from OMVs and PC (16:0). Cells were fixed and stained with anti-GLP-1R antibody and lipid nanoparticles binding to GLP-1R were visualized using confocal microscopy. The nucleus was stained with DAPI. (L). BV2 cells were treated with HFD metabolites (100 μL/mL) and band 2 lipid nanoparticles from OMV and PC (16:0) nanoparticles (1 μg/ml) for 24 h. Total RNA isolated from these cells was subjected to real-time PCR for IRS1 and IRS2 mRNA expression. mRNA expression was normalized to actin mRNA expression (n = 3). (M). BV2 cells were treated with HFD metabolites (100 μL/mL) and band 2 lipids from OMV and PC (16:0) (1 μg/ml) for 24 h and glucose uptake was determined in these cells (n = 3). P values were calculated by means of an ANOVA test. *P < 0.05, **P < 0.01, and ***P < 0.001.