1‐Deoxyceramides – Key players in lipotoxicity and progression to type 2 diabetes?

Ceramides are bioactive sphingolipids, comprised of sphingosine and a fatty acyl chain. They have been recognized as key mediators of lipotoxicity; a phenomenon where excess fat in adipose tissue leads to de novo synthesis of ceramides and their precursor dihydroceramides (DHCer). This occurs in adipose tissue as well as in the periphery. Accumulation of these ceramides is associated with insulin resistance, de novo lipogenesis, and inflammation,1 thus increasing the risk of cardiometabolic diseases such as type 2 diabetes (T2D) and atherosclerosis. Recently, in this journal, Hannich and colleagues reported that another, non-canonical class of ceramides, the 1-deoxyceramides (DoxCer), is highly enriched in visceral adipose tissue (VAT) as well as in the serum of obese patients with T2D.2 This study also clarified previously-reported discrepancies in the literature concerning the association of DHCer with the risk of T2D, as the signals from DHCer can be easily confused with the signals from DoxCer in mass spectrometric (MS) analysis of lipids. De novo synthesis of ceramide starts with palmitoyl coenzyme A (palmitoyl-CoA) and serine as precursors, and proceeds in four steps3 (Figure 1). Serine palmitoyltransferase (SPT) first produces 3-ketosphinganine, then 3-ketosphinganine reductase (KSR) produces sphinganine. The ceramide synthases (CerS) then add a second acyl chain to produce DHCer. There are six mammalian genes encoding CerS, with their products varying according to tissue, cellular localization, and acyl chain specificity.4 Dihydroceramide desaturase (DES) then adds a double bond into a sphingoid base and thus ceramides are produced. Both DHCer and ceramides are precursors of a variety of sphingolipids including complex glycosphingolipids, where various structural groups bind to the hydroxy group of (dihydro)ceramide originally derived from the hydroxymethyl group of serine. Various CerS contribute to the chemical, functional, and tissue-specific diversity of ceramides.4 Additionally, myristoyl-CoA and stearoyl-CoA can also serve as precursors, alongside serine, in de novo ceramide synthesis, further increasing their diversity. Additionally, alanine can also serve as a precursor of "ceramide" synthesis instead of serine, leading to the synthesis of a non-canonical class of 1-deoxyceramides through the same steps as synthesis from serine (Figure 1).3, 5 Since alanine contains a methyl group instead of a hydroxymethyl group of serine, DoxCer lack the hydroxyl group which could bind various other structural groups needed to form various complex sphingolipids. DoxCer are thus a metabolic "dead end" and it is not surprising, therefore, that these lipids have been found to have (lipo)toxic effects in multiple conditions, including in neuropathy, pancreatitis, and non-alcoholic steatohepatitis (NASH). In vitro studies in insulin-producing Ins-1 cells and in primary human islets have also shown that DoxCer are toxic for pancreatic beta cells in a dose-dependent manner.6 Specific serum 1-deoxysphingolipids, including precursors of DoxCer (albeit DoxCer were not assayed in the aforementioned study), were previously found to be elevated in patients with metabolic syndrome, impaired fasting glucose, and T2D, and were identified as independent predictors of progression to T2D.7 Hannich et al performed lipidomic analysis in two cohorts, one cohort comprising four groups (1- non-obese non-diabetic controls, 2 - obese non-diabetic, 3 - lean T2D, and 4 - obese T2D) and the second cohort comprised of three groups (1- controls, 2 - obese non-diabetic, 3- obese T2D).2 Serum samples were analysed in the first cohort, whilst serum, skeletal muscle, and VAT were studied in the second.2 The analytical protocol included direct-infusion MS for the analysis of phospholipids and sphingolipids, complemented by liquid chromatography (LC) – MS to separate and quantify DoxCer and DHCer. They observed alterations in the lipidome in the cases of obesity and T2D, as is generally in line with previous reports. The analysis did not, however, include neutral lipids such as di- and triacylglycerols. Serum phosphatidylethanolamines (PEs) were elevated in obese (including T2D) subjects, which is indicative of raised very-low-density lipoprotein (VLDL) particles. Moreover, several circulating choline-containing phospholipids were found to be decreased in obese subjects and patients with T2D, suggesting a possible global disturbance in choline metabolism. For example, serum lysophosphatidylcholines (LPCs) were reduced in obese subjects and (both lean and obese) patients with T2D. This included LPC(18:2), which has been one of the most consistently-replicated lipid biomarkers predictive of progression to T2D. Whether these LPCs are byproducts of, or contributors to, the pathogenesis of T2D is, at present, still unclear. Serum ether lipids, such as those containing a choline headgroup (O-PC), were previously found to be reduced in multiple (pre-) disease conditions, including, eg, in obesity, T2D, type 1 diabetes (T1D), and Alzheimer's disease. Hannich et al confirmed the association of these circulating lipids with obesity, while also identifying a similar decrease of these lipids in lean patients with T2D. This observation did not hold true for all tissue types, as in skeletal muscle the pattern of changes in O-PCs was, in fact, opposite to that observed in circulation. As ether lipids require peroxisomes for their synthesis, they are primarily produced in tissues and organs where peroxisomes are abundant, such as in the liver. Ether lipids have been previously linked to multiple, non-exclusive possible roles in health and disease, including as endogenous antioxidants, as reservoirs of long-chain polyunsaturated fatty acids (eg, arachidonic acid, docosahexaenoic acid), and as lipids modulating the stability of cellular membranes. Despite the fact that certain lipid classes, such as LPCs and ether lipids, have been consistently found to be associated with T2D, the understanding of their potential pathophysiological role(s) remains lacking, highlighting the need for further studies. Since LPCs and ether lipids are primarily found in cellular membranes, such studies would need to adopt a systems biology approach, addressing not only the complex biochemical and signalling networks the lipids are part of, but also the impact they have on the key properties of cellular membranes. Also in line with some previous reports, Hannich et al found that putatively-identified DHCer are elevated in T2D, specifically, those detected by MS as intact molecules with water loss, eg, [DHCer(22:0) – H2O]. However, when first reported in mammals,5 intact DoxCer molecules, eg, DHCer(22:0), were found to produce the same ions as the corresponding DHCer – H2O molecules, thus they are indistinguishable in MS analysis. For this reason, the authors then further characterized ceramides by LC-MS and, indeed, confirmed that it was DoxCer and not DHCer that were elevated in T2D. For comprehensive characterization of ceramides, LC-MS is required due to the typical fragmentation patterns of ceramides in electrospray, as ceramides form not only protonated ions [M + H]+ but also ions produced by the loss of water ([M + H - H2O]+ and [M + H - 2H2O]+). This, in turn, means that the ionized species of the corresponding DHCer and DoxCer have identical masses, including the characteristic fragment in tandem MS mode. Thus, for robust quantitation, chromatographic separation between the Cer, DHCer, and DoxCer prior to MS is needed. Reversed-phase LC allows detailed separation of the individual ceramide molecular species, including Cer, DHCer, DoxCer, and DoxDHCer species, while some other LC-based approaches, such as appropriately-optimized hydrophobic interaction chromatography (HILIC) as applied by Hannich et al, can provide ceramide-class specific separation. Tissue lipidomics was also performed in their second cohort,2 which revealed marked elevation of DoxCer and DoxDHCer in the VAT of obese and obese/T2D subjects, while no such changes were detected for ceramides and DHCer. Sphingomyelins and hexosyl-ceramides were mostly decreased in obese T2D subjects. Notably, the DoxCer from VAT had shorter acyl chain lengths than those found as elevated in the serum of T2D patients, suggesting that circulating DoxCers do not directly reflect the DoxCer pool in VAT. Further in vitro studies in a mouse adipocyte cell line showed that adipocytes have the capacity to independently produce DoxCer, thus confirming that VAT may not only act as storage for these toxic lipids, but may actively produce them. In conclusion, the work by Hannich et al provides further evidence that lipidomics is in the process of transforming metabolic research. Recently, comprehensive MS analysis of lipids, combined with sophisticated bioinformatics and systems biology strategies, has uncovered more of the complexity and diversity of some lipid classes, such as N-acyl amides (which includes the endocannabinoids) and bile acids. Sphingolipids were known to be highly diverse due to the plethora of carbohydrates that can attach to the sphingosine backbone. Recent research, however, suggests that the core sphingolipid metabolic network is also more complex than previously known, with potentially important contributions to make to our understanding of the pathogenesis of various diseases, particularly cardiometabolic diseases. The authors thank Dr Aidan McGlinchey (Örebro University) for language editing and to the members of the Systems Medicine research group (Örebro University and University of Turku) for stimulating discussions. The authors declare no conflict of interest exists.