Recent evidence supports a bidirectional relationship between the eCBome and the gut microbiome. We previously reported that circulating levels of eCBome mediators are associated with total adiposity, fat distribution, recent dietary intakes, and gut microbiota composition (3, 12). Nevertheless, a knowledge gap persists on which eCBome mediators are in the gut lumen and how they are associated with host systemic context or with commensal bacteria metabolic activity. We thus aimed to profile NAE and 2-MAG eCBome mediators in the feces of healthy individuals to identify the host and gut microbiota determinants of these mediators in the feces. We showed that most NAEs and 2-MAGs are present in human feces, although omega-3–derived eCBome mediators were found at very low levels. Fecal eCBome mediators were strongly correlated within each family and between NAEs and 2-MAGs. However, fecal eCBome mediators were not strongly correlated with circulating eCBome mediators, except, somehow weakly, within NAEs. Indeed the fecal eCBome profile exhibited partial independence from the circulating eCBome profile, as also witnessed by the finding that fecal eCBome mediators showed a distinct association pattern with adiposity, metabolic parameters, and dietary intakes compared with the circulating metabolites (3, 21). We also highlighted an association between fecal eCBome mediators and gut microbiota composition. Indeed, increased levels of eCBome mediators in the feces were associated with reduced gut microbiota diversity and altered taxonomic relative abundance.

To our knowledge, this study is the first to profile a large array of NAE and 2-MAG congeners in human feces. Quantifying these lipid mediators in the fecal matrix remains complex but our extraction and liquid chromatography coupled to tandem mass spectrometry protocols proved effective in quantifying fecal levels of most of the congeners previously found in the plasma. Indeed, we found 2-LG and 2-OG in all fecal samples, while these mediators were detected in less than 40% of the samples by Goedert et al (22). Nevertheless, some of the mediators we have identified with our method remain undetectable in the feces of a large proportion of individuals. Notably, omega-3–derived NAEs and 2-MAGs were only detected in a few samples. At this stage, it is unclear whether these molecules are virtually absent or present in amounts just below the detection limit. Given that nearly all dietary fatty acids are absorbed in the small intestine, and that omega-3 fatty acids constitute only a small fraction of these lipids, we may hypothesize that the pool of luminal omega-3 fatty acids is insufficient to result in significant quantities of these mediators. By contrast, oleic and linolenic acids represent a substantial proportion of the dietary fatty acids, potentially favoring the production of their corresponding eCBome mediators in the gut lumen. Interestingly, the changes in omega-3–derived eCBome mediators following the short, high-quality dietary intervention tend to support this hypothesis. However, the most common omega-3–derived eCBome mediator in feces (ie, DHEA) decreased even when individuals had substantially increased their intakes of its direct precursor (ie, DHA) during the short-term intervention. By contrast, the increased intake of linoleic and, particularly, oleic acids during the 2-day dietary intervention was mirrored in the fecal levels of the corresponding 2-MAG congeners, but not of the NAE congeners. In view of our previous finding of a significant increase in circulating levels of DHEA and OEA following the same dietary intervention (3), these data confirm that the fecal eCBome mediator profile is not determined by the same factors as the host eCBome profile. The fecal eCBome, especially for what concerns NAEs, may also result from the biosynthetic and metabolic activity of the gut microbiota. In this sense, it is possible that (1) lower omega-3-PUFA levels in the gut lumen, (2) inability of commensal bacteria to synthesize omega-3-PUFA derived NAEs, and/or (3) higher catabolic activity of these mediators explain the diverging profile of eCBome mediators in the feces compared with the plasma. This possibility, which is currently being investigated in our laboratory, may have important functional consequences, given the points raised in (1) and below.

Following in the same line of reasoning, the intriguing decrease in DHEA levels following the high-quality diet, regardless of the precursor fatty acid intakes, may be the result of changes in the gut microbiota composition and metabolic activity caused by other components of this fiber-rich, plant-based diet. We found that the gut microbiota diversity and the relative abundance of specific taxa were associated with the levels of fecal eCBome mediator levels. More precisely, higher fecal levels of most NAE and 2-MAG eCBome mediators were associated with a lower diversity and a lower relative abundance of the Christensenellaceae, Clostridiales Family XIII, Rikenellaceae and Streptococcaceae families, but a higher relative abundance of the Burkholderiaceae family. Of these families, variedly associated with obesity, Christensenellaceae were consistently reported to be negatively associated with adiposity and dysmetabolic parameters and to play a potential protective function against obesity and its consequences (23-25).

The eCBome mediators derived from, or modulated by, gut bacteria may play a role in obesity, dysmetabolism, and intestinal function as previously proposed (1). If we assume that the levels of eCBome mediators in feces are primarily influenced by microbial activity, it can be inferred that a gut microbiota with a decreased relative abundance of families like Christensenellaceae, Clostridiales Family XIII, Rikenellaceae, and Streptococcaceae, along with an increased relative abundance of Burkholderiaceae, may favor the production over the breakdown of NAEs and 2-MAGs within the gut lumen. In the case of Burkholderiaceae, this could lead to enhanced overall fecal microbiota–eCBome signaling through 2-OG, 2-LG, AEA, and DHEA at receptors associated with protective functions, such as GPR55, GPR119, TRPV1, PPARα, and PPARɣ. Whereas, in the case of Christensenellaceae, their lower abundance could lead to higher predominance of signaling by AEA or 2-AG at CB₁ receptors playing an exacerbating role in dysmetabolism, thus possibly explaining, in part, why these 2 families are negatively associated with obesity. Nevertheless, it is worth considering the possibility that the fecal eCBome mediator influences the proliferation of these bacterial taxa. Given our current understanding, it would be premature to exclude any hypothesis, including the potential for a bidirectional relationship. Other studies have highlighted the association between gut microbiota composition and fecal eCBome mediators. Fornelos et al showed that feces from individuals with intestinal bowel disease contain higher levels of fecal NAEs, partly as a potential consequence of the predominance or reduction of gut microbial species that regulate their metabolism or cellular transport, and that in turn these NAEs produce microbiota changes in culture that reproduce those observed during intestinal bowel disease in both humans and mice (9). Thus, the association between disease or gut microbiota composition and the fecal eCBome suggests a potential contribution to disease symptoms or progression.

Only 2 studies have reported associations between fecal eCBome mediators and health parameters, body mass index and inflammatory bowel disease, respectively (9, 26). Here we report that higher fecal 2-AG, PEA, LEA, and DHEA levels were associated with higher total and visceral fat mass. Higher fecal 2-AG levels were associated with a deteriorated metabolic profile, while fecal OEA levels were negatively associated with HbA1c. The associations between several metabolic parameters and 2-AG levels in the feces mirror those observed in circulation. These results are not surprising considering that 2-AG is notably associated with adverse inflammatory, metabolic, and gut permeability effects through its preferential activation of the CB₁ receptor (1, 27). Unlike 2-AG, DHEA is generally associated with beneficial anti-inflammatory and metabolic effects through the activation of PPARα, GPR119, and TRPV1 receptors (28). The positive association of DHEA with adiposity and insulin resistance is thus intriguing and may not be merely attributable to the fact that luminal DHEA levels could be already below the minimum effective concentration to be clinically relevant. We have previously reported that circulating OEA levels were positively associated with adiposity measures (eg, fat mass) (3, 12), while we report here that fecal OEA is not associated with adiposity. These results suggest that, unlike other NEAs, fecal and circulating levels of OEA seems to diverge and be associated differently with adiposity and metabolic measures. Further assessment of eCBome mediators in the feces of subjects with obesity and cardiometabolic disease such as type 2 diabetes or dyslipidemia will be needed to address all the above hypotheses. Yet, it is interesting to note how a recent study showed that the administration of Akkermansia muciniphila to obese subjects increases the plasmatic levels of 2-PG, an eCBome PPAR-α agonist, while producing beneficial effects on dysmetabolism (29).

Dietary intakes as macronutrients, food groups and patterns are known to shape the gut microbiota composition and activity, but also determine the availability of precursor nutrients for metabolite synthesis. As mentioned above, arachidonic, oleic, and omega-3–derived eCBome mediators were associated with intakes of their corresponding fatty acids in the circulation (3, 30), while only levels of 2-OG and 2-LG show such direct association in the feces. We have previously reported the importance of food patterns in determining the circulating eCBome (3, 21). Similarly, several food groups known as key drivers of the gut microbiota composition (ie, fruits, nuts and legumes, animal proteins) were found here to be associated with NAE and 2-MAG levels in the feces. Moreover, a high-quality dietary intervention provokes changes in the fecal eCBome that, unlike reported previously for the circulating eCBome, appear to be partly independent of their respective fatty acid intakes.

In sum, we can infer that the eCBome mediators we measured in feces may be synthesized or metabolized by the microbial communities or released from epithelial cells being shed into the gut lumen, or both. Previous research highlighted the production of eCBome mediators by gut bacteria (31). Moreover, Baxter et al found that LEA, PEA, and OEA levels are higher in colonoscopy samples from the ascending colon than from the descending colon (26). As gut microbiota composition and metabolic activity differ along the intestinal segments, it would not be surprising to observe site-specific difference in luminal eCBome tone if gut bacteria contributed to NAEs and 2-MAG metabolism. We must emphasize, however, that our results reflect eCBome signaling in feces, and we have currently no data to determine whether our observations can be generalized to the proximal or the distal segment of the colon. Nevertheless, our results support the concept of the contribution of the gut microbiota to the production or catabolism of fecal eCBome mediators. Cytokines, metabolic proteins, and other metabolites are already known to be released through intestinal epithelium shedding in health and in diseases (32). If fecal eCBome mediators originated uniquely from this phenomenon, we would expect that the eCBome profile in the feces, and its response to the diet, reflected those of the host, which seems to be true only in part from our results. Changes in either the intestinal or the circulating eCBome mediator profile in response to a dietary intervention are also not always similar in animal models, which reinforces the present observation (15, 33). Additionally, we have recently shown that levels of NAEs and 2-MAGs in ileal biopsies from individuals with type 2 diabetes do not reflect—and, in fact, in the case of 2-MAGs, are inversely correlated with—the corresponding circulating levels (34). In contrast to molecules that can only be synthesized by host cells, the exact respective contributions of intestinal epithelial shedding and gut microbiota to the fecal eCBome is challenging to assess precisely in humans.

In conclusion, our results seem to support the contribution of the gut microbiota to the production or the catabolism of fecal eCBome mediators. We show that the fecal profile of eCBome mediators is associated with dietary patterns, gut microbiota diversity, and the relative abundance of bacterial families. Considering the crucial role of eCBome mediators in host homeostasis, notably in metabolic and inflammatory functions of the intestinal epithelium, our results should foster further studies aimed at clarifying the link between diet, gut microbiota, and the host to fully understand the role of the fecal eCBome in health and disease.