In this study we find that approximately one third of the anti-inflammatory downstream effects of SCFAs are statistically mediated by endocannabinoids but approximately two thirds of the effect of SCFAs on cytokines appears not to be related to ECs. Using an exercise intervention, we report that increases in SCFA-producing bacteria and decreases in the pro-inflammatory genus Collinsella are correlated with increases in endocannabinoid circulating levels.

Previous research has shown increased abundance of Collinsella to be strongly associated with increased intake of processed food and low vegetable intake (Menni et al., 2021; Wolf et al., 2019), whereas decreased abundance of Collinsella was observed following a Mediterranean diet intervention (Ghosh et al., 2020). Furthermore, it was found that this genus significantly increased the risk of nonalcoholic steatohepatitis, the most serious form of nonalcoholic fatty liver disease whereby inflammation causes liver damage that can progress to cirrhosis (Astbury et al., 2020). Moreover, increased levels of Collinsella have been detected in individuals with type 2 diabetes³⁶ and were noted to decrease in a weight loss study where insulin sensitivity improved during the course of the study.³⁷ Collinsella was also positively correlated with heightened inflammatory cell count profiles (i.e. lymphocytes).³⁸ The concomitant changes observed in EC levels, Collinsella and cytokines suggest that this genus might exert some of its pro-inflammatory effects via modulating the EC system and it is possible that some of the effects of this genus on insulin resistance may also be linked to ECs given the tight relationship between EC modulation and insulin resistance.³⁹

So far, no previous study examined the potential link between the EC system, exercise, and the gut microbiome. In a study with obese and normal weight women, high moderate to vigorous activity levels were not only higher in normal weight women but were also associated with higher OEA and AEA levels.⁴⁰ Studies have shown that acute physical exercise increases circulating AEA, but not 2-AG, levels in humans.^41,42 Two studies on elite rugby players showed that depending on diet and BMI the athletes had higher microbial diversity⁴³ and increased fecal SCFA levels⁴⁴ compared to non-active controls, whereas another study found cardiorespiratory fitness associated with higher levels of microbiome diversity, especially butyrate-producing bacteria independent of diet.⁴⁵ While using small numbers of participants, a longitudinal study using 6 weeks of progressive endurance exercise without changing diet in overweight women found an increase in SCFA-producing bacteria,⁴⁶ independent of age, weight, fat % as well as energy and fiber intake.⁴⁷ Therefore, our data confirm findings from previous studies that have shown that ECs and SCFAs increase with exercise and further reveal the strong correlation between increases in ECs and decreases in pro-inflammatory cytokines. It is possible therefore that improved EC tone induced by exercise may be mediating the shift in the gut microbiota to increased SCFA producers, thereby increasing the SCFA production without a dietary change. This is a hypothesis that needs testing in a controlled experimental setting.

Using formal mediation analysis, we find that ECs are statistically mediating up to a third of the effect of SCFAs on the circulating levels of pro and anti- inflammatory cytokines. We further show that EC levels, specifically AEA and OEA, are positively correlated not just with EC system genes such as cannabinoid receptor 2 (CNR2) but also with higher expression levels of the SCFA receptors FFAR2 but not with long fatty acid receptor like FFAR4. These data complement results from animal model studies showing that ECs can attenuate central and peripheral inflammation, can modulate gut microbiota composition and can reduce markers of gut permeability.⁴⁸ CNR2 deficient mice had higher serum levels of the anti-inflammatory cytokine IL-10 compared to controls following a bacterial lipopolysaccharide challenge.⁴⁹ The anti-inflammatory role of CNR2 was further demonstrated both in a chronic murine animal model and in IBD patients.⁵⁰ FFAR2 was found to be necessary for the inulin-induced reduction of food-intake and protection against diet-induced obesity.⁵¹ Overall, our data confirm that ECs and SCFAs are crucial modulators of the effects of the gut microbiome on human metabolism and physiology.

Consistent with previous studies,^52–54 we find anti-inflammatory effects of endocannabinoids, including positive associations of 2AG and OEA with anti-inflammatory markers (IL-10) and negative associations of most ECs with pro-inflammatory cytokines (TNFa and IL-6). At the same time, we show that the anti-inflammatory effects of ECs are partly mediated by increased levels of SCFAs and specifically butyrate. Elevated levels of butyrate have been shown to decrease mucosal permeability by increasing the secretion of mucins.^55,56 Furthermore, ECs have emerged as important players in modulating gut permeability via enhancing the production of the tight-junction protein, occludin-1, as well as decreasing the expression of claudin-1, that serves as a paracellular barrier.^57,58 Although further in-vitro and in-vivo studies are required to unravel the specific pathways involved, we think that the EC system and the gut microbiome play a role jointly in regulating an inflammatory status.

Our study has a number of strengths. Firstly, the findings from our study reinforce previous knowledge of an interaction between the EC system and the microbiome and add to previous literature by revealing an interaction between ECs, SCFAs and inflammatory system markers. Secondly, we have validated the cross-sectional associations between ECs and bacterial abundances in an independent cohort, and we have used gene expression data to annotate the pathways involved. Lastly, we have assessed the relationship between ECs, SCFAs and cytokines both cross-sectionally and longitudinally and have shown that simple lifestyle interventions such as exercise can modulate inflammatory markers via SCFAs and ECs.

We also note some study limitations. The transcriptomic assay we used did not include probes for FFAR1 and FFAR3 nor did it include CNR1 making our gene expression data only partially informative. The exercise intervention we carried out was performed in individuals with pain in knee osteoarthritis and may not be directly relevant to other groups. However, the associations between ECs, SCFAs and cytokines were validated in a healthy age and sex matched smaller cohort suggesting that our data are generalizable. The findings presented from the formal mediation analysis are purely statistical and do not indicate causality.

In conclusion, in this study we show that circulating levels of ECs are consistently associated with higher levels of SCFAs, with higher microbiome diversity and with lower levels of the pro-inflammatory genus Collinsella. We also show statistically that the anti-inflammatory effects of SCFAs are up to one third mediated by the EC system.