The diabetes-associated perturbations of circulating lipid mediators and their relationships to plasma fatty acids are largely unexplored. Plasma NEFA concentrations reflect shift in NEFA uptake and release by adipose, liver and muscle [32], being dominated by adipose lipolysis in the fasted state [33]. Not unexpectedly, increases in plasma SFA and MUFA dominated the diabetes-associated changes in NEFA and were strong predictors of type 2 diabetes. However our targeted lipidomic survey uncovered subtle relationships between freely circulating NEFA, oxylipins, and endocannabinoids.

Based on the array of biosynthetically-related MUFA to SFA ratios, a net increase in systemic SCD activity was apparent in the diabetic cohort ( Table 3 and Figure 1 ). Increased SCD activity and/or expression have been reported in hypertriglyceridemia, obesity, nonalcoholic fatty liver disease (NAFLD), and the metabolic syndrome [34]–[37]. As the current study cohorts were BMI-matched and not hypertriglyceridemic [6], it is apparent that systemic SCD activity was elevated in the type 2 diabetic cohort beyond that expected by obesity alone. Such an observation is consistent with the increase in available glucoses, increasing de novo lipogenesis resulting in an elevation in SCD activity [35]

Despite the observed type 2 diabetes-associated increases in plasma NEFA, long and very long chain polyunsaturated fatty acids were not significantly elevated in this disease. In fact plasma markers of peroxisome-dependent VLCPUFA chain-shortening in both omega-3 and omega-6 pathways were reduced in diabetes, which also inversely correlated with SCD activity. Impaired VLCPUFA synthesis have previously been reported in the retina of streptozotocin-induced diabetic rats [38] and in NAFLD [39]. In fact, insulin resistance accelerates NAFLD in rodents [40] and observed changes in the VLCPUFA indices suggest an increased prevalence of NAFLD among the diabetic cohort.Diabetes and obesity are associated with pancreatic β-cell dysfunction, chronic inflammation, and vascular complications, conditions directly influenced by oxylipin metabolism [10]. While hyperglycemia and diabetes increase oxidative stress [41], and the total OxL concentration was positively correlated with fasting glucose, it did not increase in the diabetic compared to the non-diabetic cohort. Moreover, oxidative stress markers including the F2 isoprostanes were equivalent between cohorts. However, the concentration of multiple fatty acid epoxides, as well as several diols and ketones increased in association with diabetes. In particular, the epoxides of 18∶2n6 and 18∶3n3, and the 18∶2n6 ketone 13-KODE, were increased 2-fold and positively correlated with changes in SFA and MUFA. Changes in levels of 20∶4n6 and 20∶5n3-derived oxylipins were more subtle with ∼50% increases. Type 2 diabetes-associated changes in these species may arise from increased biosynthesis, decreased degradation, and or increased release of preformed metabolites. However, since fasting increases adipose lipolysis [33] and in this cohort epoxides and ketones are strongly correlated with plasma NEFA, we hypothesize that these species are derived from lipolysis of adipose. In support of this hypothesis, we have found that the adipose triglycerides of hamsters fed differing lipids in the diet preferentially accumulate 18 carbon species, with EpOMEs >5-fold over DiHOMEs, and KODEs ∼2-fold over HODEs in this pool [42].

Endocannabinoids and their related metabolites are important regulators of inflammation and energy balance, functioning through interaction with cannabinoid type 1 and type 2 receptors, transient receptor potential vanilloid type 1, and peroxisome proliferator activated receptors [11], [43], [44]. Previous investigations have reported increases in eCBs among both obese and diabetic humans [11]. Here plasma MAGs were inversely proportional to BMI, but unaffected by diabetes status. Conversely, plasma NAEs, including A-EA, were elevated and its concentrations correlated with SFA and MUFA in the diabetic cohort ( Figure 3 ). Given that insulin dependent suppression of plasma A-EA is inversely correlated with liver fat [14], the higher A-EA in this studies diabetic cohort is consistent with insulin resistance and an increased prevalence of NAFLD. Similarly, increases in the omega-3 DoHex-EA were strong predictors for the type 2 diabetes-associate phenotype ( Figure 2 ) and correlated with two 5-lipooxygenase metabolites, 5-HEPE and 5-KETE among diabetic participants. The basis for higher NAE metabolites in diabetes remains to be established, but an anti-inflammatory role is plausible. Specifically, the diabetes-associated elevation in SFA can activate Toll-Like Receptor 4 (TLR4), which has been reported to initiate inflammatory signaling and enhance NAE, but not MAG synthesis [45]. The NAEs may therefore act in a feedback loop to suppress inflammatory signaling through cannabinoid type 2 receptors [46].

In conclusion, to determine impacts of type 2 diabetes on the non-esterified plasma lipidome we quantified NEFA, oxylipins, acylethanolamides, lipoaminoacids and monoacylglycerides in weight-matched obese diabetic and non-diabetic cohorts. Diabetes-associated NEFA patterns indicate increases in SCD activity and decreases in VLCPUFA chain shortening, which may indicate impaired hepatic insulin sensitivity and/or fatty liver disease. Among diabetic participants, increases in 18 carbon epoxides and ketones correlate strongly with changes in SFA and MUFA, consistent with an enhanced release from adipose stores and/or suppressed degradation of these oxylipins. Type 2 diabetes is also associated with increases in NAEs and LAAs, but not MAGs. While the increases in NAE tone may constitute an adaptive mechanism to suppress inflammation developed in response to increases in circulating SFAs, MAG concentrations provide a diabetes-independent metabolic marker of body mass. Together, the observed changes in the plasma lipidome describe type 2 diabetes as a state of imbalance with respect to metabolic processes associated with fatty acid desaturation, VLCPUFA systhesis, adipose lipolysis, endocannabinoid tone and inflammatory activation. Oxylipin analysis revealed that these changes occur in the absence of systemic oxidative stress greater than that seen in obesity. Through this targeted lipidomic analysis, the following plasma markers were found as predictors of the type 2 diabetes phenotype: 1) 18∶1n9/18∶0, a marker of increased SCD activity; 2) 22∶5n6/22∶4n6, a marker of suppressed VLCPUFA synthesis; 3) 9(10)-EpOME and 9(10)-EpODE, putative markers of adipose lipolysis; 4) DoHex-EA, a marker of increased eCB system tone; 5) 1-LG, a metabolic marker of obesity.