Preclinical studies have highlighted the importance of improving our understanding of the biological and metabolic pathways involved in AUD to promote the development of new therapeutic and diagnostic methods. While many related studies have focused on the use of EV-resident microRNAs and proteins as plasma biomarkers, our results demonstrated, for the first time, that LPC and PC lipids (and enzymes such as phospholipases and acyltransferases) suffer from changes associated with cancer progression and neuroinflammation in female AUD patients. Moreover, male AUD patients exhibit dysregulation of Cer and SM lipid species (which involve sphingomyelinases, sphingomyelin phosphodiesterase, and sphingomyelin synthase), which potentially contributes to ethanol-induced hepatotoxicity. Additionally, computational analyses highlight sex-specific variations in EV lipids that play roles in vesicle fusion processes.
Considering that most, if not all, cells in the human body secrete EVs into circulating bodily fluids, the characterization of EV lipid profiles could provide information regarding the cell/tissue of origin and their functional state [32]. The distribution of lipid species in absolute amounts highlighted PC, SM, and TAG as the most abundant lipid subclasses. Whereas PC represents an abundant lipid subclass in EVs derived from neural cells [33], the SM subclass participates in EV biogenesis and is among the most abundant classes in brain-derived EVs [34, 35]; moreover, we identify the novel lipid species SM d18:2_24:0 as a potential biomarker in female and male AUD patients. The presence of the TAG lipid subclass in EVs could arise from the secretory autophagy pathway [36]; in addition, TAG could become transferred from lipoproteins to exosomes once released into the bloodstream [37], suggesting the absence of lipoprotein contamination during the EV isolation procedure [5].
We previously reported that acute ethanol intoxication induced the enrichment of distinct plasma EV lipid species (e.g., LPC, PA, FAHFA) in human female adolescents compared to males; these lipid classes participate in the formation, release, and uptake of EVs and immune response activation [5]. Following the same sex-based differential analysis in AUD patients, our current results indicate a lower abundance of the LPC and PE subclasses in AUD females than in healthy individuals. LPC, which is enriched in EVs, is related to pro-inflammatory functions and participates in EV biogenesis [38]; moreover, LPC promotes demyelination by activating CNS inflammatory responses and inducing microglia pyroptosis [39]. Indeed, alcohol-induced pro-inflammatory molecules in the periphery may provoke neuroinflammation by crossing the brain-blood barrier [40]. A general decline in plasmalogen lipids (mainly PC and PE subclasses) has been described in multiple brain regions in Alzheimer's disease [33], which could associate with increased oxidative stress, inflammatory responses, and neuronal cell death [41, 42]; however, additional studies have reported high and low levels of PC and PE in highly metastatic breast cancer, respectively [43]. In addition, our results demonstrated that most ceramide lipid species (e.g., Cer_NS d18:1_24:1, Cer_NS d18:1_22:0, and Cer_NDS d42:2 RT:12.673) exhibited sex-specific abundances. The subclasses Cer_AP and Cer_AS displayed a greater abundance in AUD females, whereas some lipids belonging to Cer_NS and Cer_NDS displayed lower abundance. An increase in Cer_AS species along with a decrease in Cer_NS and Cer_NDS has been previously described in a mouse model of metachromatic leukodystrophy, suggesting that alpha-hydroxylation of ceramides may play a role in the brain pathology of this disease (e.g., demyelination and motor dysfunction) [44].
The Fatty acids main class has been associated with inflammation [45] and neurotransmitter release [46] through cell surface and intracellular receptors, thereby being linked to the modification of membrane composition, cell signaling, gene expression, and lipid mediator production [45]. Our results revealed that unsaturated FA subclass (main class Fatty acids) had a negative LOR in the IS comparison, indicating class enrichment in AUD males compared with females. FAs have been implicated in neural cell pathology in lysosomal storage diseases, including metachromatic leukodystrophy, which is characterized by lipid accumulation in the brain, spinal cord, and peripheral nerves [44]. Furthermore, although the FAHFA subclass emerges as a significantly more abundant lipid in AUD males, we know little regarding the involvement of FAHFA in biological processes other than its anti-inflammatory role [47].
Incorporating LINEX2 lipid network enrichment into our data provided the basis for a knowledge-driven integration of lipidomics with proteomics data by connecting enzymatic activity to lipid species [7]. The resulting network analysis revealed more significant substrate-product changes in AUD females for reactions involving the LPC and PC subclasses, including phospholipases and acyltransferases (e.g., LPCAT3/4). The upregulation of enzymes such as LPCAT1 has been reported in human colorectal adenocarcinoma [48] and metastatic prostate cancer [49], suggesting the involvement of the LPC metabolism in cancer progression. In addition, PLA2-activated neuroinflammatory pathways (through the upregulation of the oxidative stress status) become induced by binge alcohol drinking in adult rats and in organotypic hippocampal-entorhinal cortical slice cultures [50]. Our results also demonstrated that PLA2G2A becomes upregulated in AUD females; this enzyme, which possesses lysophospholipase, transacylase, and PLA2 activities [51], plays an antimicrobial role by degrading bacterial membranes and releasing pro-inflammatory eicosanoids from inflammatory cell EVs [52].
We also observed the enzymatic dysregulation of Cer and SM lipid species in AUD males. Previous studies reported alterations in the levels of various sphingolipids (including Cer and SM) in human chronic alcohol-related liver disease [53] and individuals with high alcohol consumption [54]. The enzymes involved in these substrate-product reactions - the sphingomyelinases (e.g., ASM, ENPP7, SMPD family, and SGMS1) – have been linked to chronic alcohol consumption [55]. In addition, recent studies revealed increased sphingomyelinase activity in ethanol-treated microglial cells [56] and high sphingomyelinase protein levels associated with alcoholic liver disease [57]. Since the enzymes involved in sphingolipid metabolism may mediate ethanol's hepatotoxic effects [58], ASMase activation and C16-ceramide generation could sensitize hepatocytes to the effects of TNF-α [59]. In agreement with our results, a sex-based evaluation by Mühle et al. reported high levels of serum ASMase activity in alcohol-dependent male patients [60].
As lipids exhibit many structural and signaling functions, the biosynthesis of lipids and changes to biophysical properties must be considered. We performed a comprehensive computational lipidomic analysis using network-based and lipid property-related methods through the LION algorithm to evaluate membrane remodeling and lipid-mediated signaling in EVs. Interestingly, our results demonstrated LION-term enrichment featuring "positive intrinsic curvature" in AUD females but "negative intrinsic curvature" in AUD males. Lipids with positive intrinsic curvature (such as LPC) hinder stalk formation during vesicle fusion [61] to facilitate fusion pore expansion [62]. While lipids with greater negative curvature (such as PE and DAG) represent critical players in fusion, lipids of lesser negative curvature (such as phosphatidic acid) generally play modulatory roles [63]. Lipids with negative curvature (such as oleic acid or DAG) significantly influence vesicle fusion processes [63, 64] and tend to promote stalk formation and inhibit pore expansion [65]. Notably, the formation and expansion of fusion pores during SNARE-dependent vesicle fusion remain essential for neurotransmitter release and vesicle recycling during exocytosis [66].
This study aimed to provide data regarding individual lipid species to support a rigorous lipidomic pathway analysis, as lipid species of the same class can behave differently, leading to distinct biological functions; however, this analysis does suffer from certain limitations. For instance, i) a lack of standardization in lipid nomenclature and integration into computational tools (e.g., FAHFA displays significant abundance but may not be included in the LINEX2 software); ii) lipid databases (e.g., LIPID MAPS and HMDB) contain general information regarding lipid class biology; and iii) LINEX2 details lipid species and their enzymatic activity, although this software package does not allow complete control and provides aleatory results based on the algorithm. Of additional note, the EVs used in this study have sizes and protein marker expression profiles similar to exosomes; however, we cannot currently specifically identify them as exosomes.
Perspectives and significance
Our findings underscore the presence of sex-based differences in EV lipidomic profiles induced by AUD. These distinctions, evident in lipid network analysis and enzymatic dysregulation, highlight the innovative nature of our study. It employs a comprehensive bioinformatic strategy to explore the sex-specific effects of ethanol on lipidomic profiles, providing new insights into lipid metabolism.