Gut microbiome-derived glycine lipids are diet-dependent modulators of hepatic injury and atherosclerosis

Oral and gut Bacteroidetes produce unique classes of serine-glycine lipodipeptides and glycine aminolipids that signal through host Toll-like receptor 2. These glycine lipids have also been detected in human arteries, but their effects on atherosclerosis are unknown. Here, we sought to investigate the bioactivity of bacterial glycine lipids in mouse models of atherosclerosis. Lipid 654 (L654), a serine-glycine lipodipeptide species, was first tested in a high-fat diet (HFD)-fed Ldlr^−/− model of atherosclerosis. Intraperitoneal administration of L654 over 7 weeks to HFD-fed Ldlr^−/− mice resulted in hypocholesterolemic effects and significantly attenuated the progression of atherosclerosis. We found that L654 also reduced liver inflammatory and extracellular matrix gene expression, which may be related to inhibition of macrophage activation as demonstrated in vivo by lower major histocompatibility complex class II gene expression and confirmed in cell experiments. In addition, L654 and other bacterial glycine lipids in feces, liver, and serum were markedly reduced alongside changes in Bacteroidetes relative abundance in HFD-fed mice. Finally, we tested the bioactivities of L654 and related lipid 567 in chow-fed Apoe^−/− mice, which displayed much higher fecal glycine lipids relative to HFD-fed Ldlr^−/− mice. Administration of L654 or lipid 567 for 7 weeks to these mice reduced the liver injury marker alanine aminotransferase, but other effects seen in Ldlr^−/− were not observed. Therefore, we conclude that conditions in which gut microbiome-derived glycine lipids are lost, such as HFD, may exacerbate the development of atherosclerosis and liver injury, whereas correction of such depletion may protect from these disorders.Supplementary key words: liver, dietary fat, cholesterol, heart, inflammation, high-fat diet, L654, L567, alanine aminotransferase, mouse models, Bacteroidetes, Bacteroidota

Driven by recent advancements in high-throughput sequencing technology, there has been a dramatic increase in research supporting associations between the gut microbiome, diet, and metabolic diseases. The Bacteroidetes and Firmicutes phyla typically comprise more than 90% of the bacteria found in the human gut microbiome (1). Some studies report obese mice, and humans have an increased relative abundance of fecal Firmicutes and a lower abundance of Bacteroidetes compared with lean controls (2, 3). Lower microbial diversity and a higher Firmicutes/Bacteroidetes ratio often characterize the gut dysbiosis seen in high-fat diet (HFD)-induced obesity and metabolic diseases (4, 5, 6, 7, 8). Even though a higher prevalence of Bacteroidetes appears to be associated with beneficial health outcomes, some species of the phylum, such as Porphyromonous gingivalis, can invade the endothelial layer of arteries and translocate to the circulatory system, allowing for potential inflammatory signaling (9). Alongside changes in gut microbiota composition and translocation of whole microbes, exposure to microbiota-produced metabolites and bioactive compounds may also drive disease development. Thus, characterizing the host response to metabolic products of Bacteroidetes is considered critical for the understanding of gut microbiome-host interactions in human metabolic diseases (10).Besides the well-known role of lipopolysaccharide (LPS) in Toll-like receptor (TLR) activation, emerging classes of gut microbiota-derived lipids relevant to inflammation-related disease are the glycine lipids first identified in P. gingivalis and found to be broadly expressed in the Bacteroidetes phylum (11, 12, 13). These unique classes of bacterial lipids signal through TLR2 and include the glycine amino lipid species, lipid 342 (L342) and lipid 567 (L567) (12), named for their negative ion mass, as well as the serine-glycine lipodipeptides, lipid 430 (L430), lipid 654 (L654), and lipid 1256 (L1256) (Fig. 1A) (11, 13). These glycine lipids are related to each other through the constituent core L342 structure consisting of a 3-OH iso C_17:0 fatty acid, which is amide linked to a glycine (12). The addition of an ester-linked fatty acid (iso or anteiso C_15:0) to L342 yields the other glycine amino lipid species, L567 (12). L654 consists of a terminal serine that is amide linked to the glycine of L567, thus making it a serine-glycine lipodipeptide (12). Removal of the ester-linked fatty acid from L654 produces L430, which could occur via phospholipase A2-mediated hydrolysis in humans (14). L654 can be modified further through the attachment of a diacylated phosphoglycerol to its serine moiety to form L1256 (13). The synthesis and bioactivities of these related glycine lipids are still being resolved, although they are likely shed from bacteria through outer membrane vesicles or cell turnover, and have been shown to activate TLR2 in vitro (11, 12, 13, 15), and induce acute inflammation in mice via TLR2 (11). Notably, it was recently reported that some of these glycine lipids are found in human serum, brain, and carotid artery atheroma (14). Because of the localization of these lipids in atherosclerotic lesions and signaling through inflammatory TLRs, it is thought that these bacterial-derived lipids may contribute to the pathogenesis of atherosclerosis (9, 14). Furthermore, it seems plausible that dietary factors known to influence the relative abundance of Bacteroidetes (e.g., dietary fat) may influence host exposure to these bacterial glycine lipids. Thus, we sought to investigate the bioactivities of bacterial glycine lipids in mouse models of atherosclerosis and characterize the glycine lipid content in the feces of mice fed different diets. We hypothesized that chronic exposure to glycine lipids would exacerbate atherosclerosis progression through increasing inflammation.Open in a separate windowFig. 1Western-type HFD markedly decreases fecal bacterial glycine lipids and fecal microbiota diversity. Structures of bacterial glycine lipids used for experiments (L567, L654, and L1256) depicting the dominant lipid species within each lipid class (A). Cecal feces were aseptically collected for characterization of 16S V4 region (B, C). Measures of fecal microbiota alpha diversity (B) and fecal Bacteroidetes (C) of Ldlr^−/− mice housed in different cages and fed either a standard low-fat chow diet or Western-type HFD for 14 weeks (n = 6–11, mean ± SEM). Lipids were extracted from feces and L654 (D), and total bacterial glycine lipids (E) were quantified by UPLC-MS/MS (C, D) (n = 5–20, mean ± SEM). Serum L654 was quantified by UPLC-MS/MS after pooling three individual animals per sample (n = 2–3 pooled, mean ± SEM). Statistical significance determined by two-tailed Student’s t-test (∗P < 0.05, ∗∗P < 0.01, and ∗∗∗∗P < 0.0001).