Ecotoxico-lipidomics: An emerging concept to understand chemical-metabolic relationships in comparative fish models

Abstract

Lipids play an essential role in development, homeostatic functions, immune signaling, reproduction, and growth. Although it is evident that changes in lipid biosynthesis and metabolism can affect organismal physiology, few studies have determined how environmental stressors affect lipid pathways, let alone alter global lipid profiles in fish. This is a significant research gap, as a number of environmental contaminants interact with lipid signaling and metabolic pathways. In this review, we highlight the utility of lipidomics as a tool in environmental toxicology, discussing the current state of knowledge regarding chemical-lipidomic perturbations. As with most oviparous animals, the processing and storage of lipids during oocyte development is also particularly important for embryogenesis in fish. Using largemouth bass (Micropterus salmoides) as an example, transcriptomics data suggest that various chemicals alter lipid metabolism and regulation, highlighting the need for more sophisticated investigations into how toxicants impact lipid responses. We also point out the challenges ahead; these include a lack of understanding about lipid processing and signaling in fish, tissue and species-specific lipid composition, and extraneous factors (e.g., nutrition, temperature) that confound interpretation. For example, toxicant exposure can lead to oxidative stress and lipid peroxidation, resulting in complex lipid byproducts that are challenging to measure. With the emergence of lipidomics in system toxicology, multi-omics approaches are expected to more clearly define effects on physiology, creating stronger linkages between multiple molecular entities (gene-protein-lipid/metabolite). The development and implementation of novel technologies such as ion mobility-mass spectrometry and ozone-induced dissociation support the complete structural elucidation of lipid molecules. This has implications in the adverse outcome pathway framework, which will enhance the application of lipidomics in toxicology by linking these cellular changes to effects at higher levels of biological organization.

Graphical Abstract

1. Lipids are a diverse group of molecules that serve a variety of physiological functions

Recent progress in lipid analytics and bioinformatics has increased the discovery, application and impact of lipidomics. Much of this momentum has been driven by our recent understanding of lipid diversity, which now rivals other fields of systems biology (Giles et al., 2018; Wenk, 2005). Lipids encompass several structural categories that include glycerolipids, glycerophospholipids, sphingolipids, sterol lipids, prenol lipids, saccharolipids, and polyketides (Sud et al., 2007). From this, it is evident lipids comprise a landscape of molecules with unique structural features and biological roles. In order to curate this diversity, the Lipid Metabolites and Pathways Strategy (LIPID MAPS) consortium was developed to provide online tools for structure prediction, annotation, and database development (Fahy et al., 2007, 2019; Liebisch et al., 2013). To date, over 40,360 unique lipid structures are included in the LIPID MAPS Structure Database (LMSD), and this number is still expanding. This is a testament to the immense diversity of these molecules. Additional databases include LipidWeb (https://www.lipidhome.co.uk/), Lipid Bank (http://lipidbank.jp/), and Cyberlipid (http://cyberlipid.gerli.com/), which offer more information about lipid structures, groups, and analysis methods.

Given the large diversity of lipid structures, it is no surprise that these molecules mediate a broad range of physiological functions. One of the most prominent examples of lipid function is the separation of cellular compartments through membranes. The early work of Luzzati described the formation of a bilayer, where lipids (e.g., phosphatidylcholine, among others) self-arrange so that the polar head-group of these molecules faces the aqueous milieu, while fatty acid domains interact and form the hydrophobic region (Luzzati and Husson, 1962). Indeed, other lipids such as cholesterol are involved in membrane structure, primarily to support fluidity and structural support (Raffy and Teissié, 1999). In select regions of the membrane, lipid rafts are used to organize protein complexes that provide important membrane signaling and trafficking functions. These rafts are primarily comprised of sterol and sphingolipids that support lipid-protein interactions (Lingwood and Simons, 2010). Finally, the role of lipids in cellular signaling has drawn great attention in recent years. Common lipids such as diacylglycerols and phospholipids are readily metabolized to release arachidonic acid, which serves as a precursor for eicosanoid synthesis. These molecules, which include prostaglandins, leukotrienes, and thromboxanes, provide several physiological functions, but they are most noted for their role in inflammation, immunity, and central nervous system function (Balazy, 2004). In addition to eicosanoids, sphingolipids have been implicated in cell signaling. Ceramides, which are categorized as sphingolipids, are potent signaling molecules for cellular growth and death (Obeid et al., 1993). Thus, lipids require special attention as they provide several important physiological functions related to exposure and health, energy metabolism and cellular compartmentalization.

In addition to these functions, it is important to understand how lipids are synthesized and regulated. Lipogenesis, the production of lipids, occurs when carbohydrate content exceeds a useful threshold for normal cellular function. In this process, acetyl-CoA is metabolized to create fatty acids, which are incorporated into phospholipids for membrane structures through the Kennedy pathway (Kennedy 1961) and remodeled through the Lands cycle (Lands 1958) or stored as triacylglycerols. Most lipids are synthesized in the Golgi apparatus, while others such as glycerolipids and sterols are produced in the endoplasmic reticulum or mitochondria. Once synthesized, lipids are moved through vesicular carriers to the appropriate location (Shevchenko and Simons, 2010).

Several proteins mediate lipogenesis and lipid trafficking, which are regulated through complex feedback loops that accompany differential gene and protein expression. For example, the Nobel Prize-winning work of Brown and Goldstein described a receptor-mediated pathway for cholesterol regulation, where low density lipoprotein (LDL) receptors regulate 3-hydroxy-3-methyl-glutaryl coenzyme-A (HMG-CoA) reductase in the mevalonate pathway, thus suppressing cholesterol biosynthesis (Brown and Goldstein, 1974; Goldstein and Brown, 1973). Another example includes the peroxisome proliferator-activated receptors (PPAR), which use fatty acids, among other molecules, as ligands and function as transcription factors for several genes, including those related to lipid metabolism (Berger and Moller, 2002). In cell membranes, there is growing evidence that lipid-protein interactions regulate sterol and sphingolipid content. From an evolutionary perspective, this mechanism may explain the concurrent presence of these lipids within raft structures (Guan et al., 2009). In addition to these mechanisms, there is also substantial evidence that nutritional status and hormone levels are important regulators of lipogenesis. For example, a high carbohydrate diet is expected to stimulate lipogenesis, in part through hormone feedback loops. In the presence of excess carbohydrates, insulin is a strong regulator that increases lipogenesis, whereas growth hormone and leptin inhibit this process (Kersten, 2001). It is clear that lipogenesis is tightly controlled through complex regulatory networks that respond to changes at the cellular and organism level to maintain homeostasis. Environmental factors, such as physical and chemical stressors, may also be expected to impact lipid synthesis and transport.

2. Lipids are important for fish reproduction

Lipids are an integral component of the reproductive axis in fish, particularly in the liver and gonad. In teleost species, as with most oviparous animals, the reproductive axis is comprised of the brain (hypothalamus), pituitary gland, gonad (either testis or ovary), and liver. In order to understand the effects of chemicals and other environmental stressors, this axis has been developed into a graphical systems model to visualize how effects in one component may cascade towards adverse reproductive effects (Villeneuve et al., 2007). Among components in the reproductive axis, a particular level of attention is given to the liver and gonad because these organs are responsible for egg yolk synthesis and oocyte development, respectively. Similarly, lipid synthesis, transport, and utilization in these organs may dictate the level of reproductive success in teleost fish species.

The liver is unique in oviparous animals because it provides several components for oocyte development including vitellogenin, choriogenin, and lipids. The comprehensive work of Wallace described the process of vitellogenesis in several species including the African clawed frog (Xenopus laevis), mummichog (Fundulus heteroclitus), and goldfish (Carassius auratus) (Wallace, 1985). In this process, granulosa cells in the ovary produce estrogens, primarily 17β-estradiol, which interact with estrogen receptors in the liver to induce vitellogenin synthesis. After synthesis and release, vitellogenin travels through the blood and is incorporated into growing oocytes, via receptor-mediated endocytosis (Wallace and Selman, 1981). As a lipoprotein, vitellogenin is modified with several lipids in the liver. The lipid content of vitellogenin has been characterized in several salmonids (Norberg and Haux, 1985) and in Atlantic halibut (Hippoglossus hippoglossus) (Norberg, 1995), finding that phospholipids comprise the majority (approximately 70%) of lipids in the vitellogenin lipoprotein complex. In addition, cholesterol and triacylglycerols were identified with vitellogenin, and it was found that lipids constitute approximately 20% of the overall lipoprotein mass. From this, it is clear that vitellogenin is a major carrier of lipids for oocyte development, especially for phospholipids and other polar lipids. Given their broad range of physiological functions, lipids play a diverse role in oocyte development. In addition to receptor-mediated endocytosis of vitellogenin, lipid uptake in the ovary can be accomplished through plasma lipoproteins (e.g., LDL) and other chaperones for free fatty acids. In the ovary, vitellogenesis can be categorized into endogenous and exogenous phases (or type I and type II vitellogenesis); each of these phases entail unique mechanisms for lipid uptake (Wiegand, 1996). Vitellogenesis is a stage characterized by the first entry of vitellogenin into the oocyte. During endogenous vitellogenesis (Type I), vitellogenin uptake is very low and can only be detected by electron microscopy, but lipid uptake and synthesis predominate and there is large accumulation of lipid globules (LeMenn et al., 2007). This occurs in the oocyte during primary growth stages (Stage IIIa) before high amounts of vitellogenin transport begins. In this process, dietary lipid uptake and lipid reserves provide several sources of lipids - predominately fatty acids - which can be transported by carrier proteins such as LDL. Exogenous vitellogenesis (Type II vitellogenesis) commences during Stage IIIb of oogenesis, where the proportion of yolk and lipid deposition are reversed, and large amounts of yolk come into the oocyte (LeMenn et al., 2007).

Several studies have investigated the role of dietary lipid content on oocyte development (Leger et al., 1981; Mourente and Odriozola, 1990; Watanabe et al., 1984) and found a high selection for n-3 (Omega-3) unsaturated fatty acids (Wiegand, 1996). Poor egg quality and fertilization success were underlying factors in experiments where uptake and accumulation of these fatty acids were altered, indicating the importance of this specific lipid in reproduction (Ashton et al., 1993). Lipid reserves are also an important resource during endogenous vitellogenesis, particularly for perciformes, which heavily rely on these resources during oocyte development (Tanasichuk and Mackay, 1989). It is important to note that endogenous vitellogenesis is an important source of neutral lipid content, which contrasts with the polar lipid content of exogenous vitellogenesis (Wiegand, 1996). As mentioned previously, the vitellogenin lipoprotein is a major source of polar lipids, such as phospholipids. Thus, both endogenous and exogenous vitellogenesis are important sources for lipid uptake during oocyte development. The processing and storage of lipids during oocyte development is particularly important for embryogenesis in fish. In many species, nonpolar lipids such as triacylglycerols, which contain monounsaturated and essential fatty acids, are stored in oil droplets. This organization provides a storage mechanism for neutral lipids, as embryos predominately rely on yolk lipoproteins for immediate use. This yolk is rich in essential fatty acids, as well as polar and nonpolar lipids (Rønnestad et al., 1994), and has been found to be a metabolically-active site for lipid remodeling (Fraher et al., 2016). However, oil droplets still provide a valuable resource for neutral lipids. For example, embryos will leverage essential fatty acids from these stores, allowing the use of polar phospholipids for other catabolic processes, such as neurotransmitter synthesis (Wiegand, 1996). Additionally, lipid droplets are critical for maintaining embryonic energy homeostasis in the first five hours post-fertilization (Dutta and Sinha, 2017). From this, lipid stores are an important resource that are prioritized for use during embryogenesis.

3. Evidence of altered lipid metabolism from chemical exposure: A case study with largemouth bass

It is evident that lipids are vital for numerous physiological processes, and it is important to understand how these molecules are perturbed in synthesis, metabolism, and function by environmental stressors, such as chemical pollutants. As one approach to determine whether there are linkages between chemical exposure and the synthesis, degradation, or production of lipids in fishes, we reviewed transcriptomic data from a selection of largemouth bass (Micropterus salmoides) studies previously conducted by our research group. These studies involved transcriptomic profiling in various tissues (ovary, testis, and liver) in response to toxicant exposures, including pesticides, industrial chemicals, and pharmaceuticals. We reasoned that molecular pathways provide insight into whether certain contaminants affected lipids at the transcript level. Largemouth bass were selected for this case study because they offer an environmentally-relevant model to understand how oocyte development and chemicals affect the transcriptome, which may elucidate altered lipid metabolism and bioenergetics. Further, these studies leverage a similar microarray platform and analysis pipeline, thereby enabling comparable results. Here we focus on the toxicology of largemouth bass as a case study but acknowledge that additional studies may further elucidate relationships between pollution and transcript-level responses related to lipid metabolism.

As a baseline, transcriptomic responses in the largemouth bass ovary were characterized as a function of reproductive stage to better understand gene networks involved in reproduction (Martyniuk et al., 2013). This is important, as oocyte development is categorized by a series of unique reproductive stages related to vitellogenin uptake, oocyte maturation, and ovulation/atresia. Based on these transcript profiles, a sub-network enrichment analysis was used to understand how global gene expression patterns related to cellular pathways, including lipid metabolism and other pathways supporting bioenergetic homeostasis change during oocyte maturation (Martyniuk et al., 2013).

During oogenesis, there are several changes in transcripts related to lipid metabolism that determine the ultimate fate of an oocyte. During transition from primary growth stages (i.e., perinuclear to cortical alveoli), transcripts related to mitochondrial function, low density lipoprotein receptor, and apolipoproteins are differentially expressed. These changes are related to several phenotypic outcomes of primary and secondary growth stages, such as the accumulation of oil droplets and oocyte growth. Similarly, expression of apolipoproteins increases through secondary growth stages, marking the uptake of larger concentrations of vitellogenin and other very low-density lipoproteins. As oocytes reach maturity, there are unique transcript profiles that distinguish ovulation and atresia that are related, in part, to lipids. When energy resources are low, oocytes will undergo atresia instead of ovulation to protect against nutrient depravation. Indeed, there are unique gene networks involving lipids that support this decisive signaling cascade. For example, pathways related to arachidonic acid and sphingolipid metabolism increase during ovulation, while genes for oxidative phosphorylation decrease in atresia, signifying inadequate energy stores. At the global level, gene networks supporting activin and inhibin activity are differentially regulated during ovulation and atresia, respectively, and both networks include genes related to glucose metabolism, respiratory chain function, and lipid transport. It is apparent that lipids play a dynamic role in oocyte development, particularly for gene networks that support the actions of ovulation and atresia (Martyniuk et al., 2013).

Several studies have used transcriptomics to understand how largemouth bass respond to chemical stressors in the environment. In both the ovary and testis, it is evident that transcripts related to lipid metabolism and other bioenergetic pathways are altered by environmental contaminants (Table 1). For example, in the ovary, long-term mesocosm exposures to several organochlorine pesticides (OCPs) altered transcripts related to lipoprotein metabolism, hormone biosynthesis, bile acid metabolism, and retinoic acid metabolism in a sub-network enrichment analysis (Martyniuk et al., 2016b). These processes were common for two sampling dates (i.e., January and April), suggesting that contaminants can affect lipid metabolism independent of reproductive stage, as fish collected in January were in early vitellogenesis while fish collected in April were ready to spawn. In addition, a three month ingestion study with p,p’-DDE and methoxychlor altered retinoic acid metabolism and nutrient uptake, respectively, indicating that specific OCPs can affect lipid metabolism and bioenergetics in the ovary (Martyniuk et al., 2016a). These changes may be related to the endocrine activity of OCPs, as steroid hormones also affect lipid metabolism in largemouth bass gonad. For example, a 60-day feeding study with 17α-ethinylestradiol increased LDL oxidation and hormone metabolism in the ovary (Colli-Dula et al., 2014), suggesting that altered lipid metabolism may be linked to steroid hormone receptor activation.

Table 1.

Summary of largemouth bass transcriptomics data indicating altered lipid metabolism and bioenergetics by chemicals.

Study	Chemical stressor	Dose	Duration	Route of exposure	Gender	Tissue	Altered transcripts related to lipid metabolism and bioenergetics
Colli-Dula et al. 2016	Perfluorinated chemicals	NA (Tissue chemistry)	NA	Field study	Male	Liver	Increased carbohydrate metabolism, altered fatty acid metabolism, electron transport and ATP synthesis, and fatty acid elongation
						Testis	Increased lipid transport and cholesterol efflux, decreased carbohydrate metabolism, electron transport/ATP synthesis, fatty acid elongation
Martyniuk et al. 2016	Lake Apopka mesocosm (Jan)	NA (Tissue chemistry)	4 Months	Mesocosm	Female	Ovary	Increased lipoprotein metabolism, hormone biosynthesis, bile acid metabolism, neurosteroidogenesis, decreased retinoic acid metabolism
	Lake Apopka mesocosm (Apr)						Increased steroid metabolism, hormone biosynthesis, bile acid metabolism, amino acid catabolism, decreased retinoic acid metabolism
Martyniuk et al. 2016	p,p’-DDE	125 ppm	84 days	Ingestion	Female	Ovary	Decreased retinoic acid metabolism
	Methoxychlor	10 ppm					Increased nutrient uptake
	Flutamide	750 ppm					Decreased glucose metabolism
Mehinto et al. 2014	Cadmium	20 μg/kg	48 hours	IP injection	Male	Liver	Increased glycolysis, cholesterol biosynthesis, altered carbohydrate metabolism and steroidogenesis
						Testis	Decreased carbohydrate metabolism, steroidogenesis, tricarboxylic acid cycle
Colli-Dula et al. 2014	17α-ethinylestradiol	70 ng EE2/g feed	60 days	Ingestion	Female	Liver	Increased LDL oxidation, decreased hormone metabolism, sphingolipid biosynthetic process, altered lipid peroxidation
						Ovary	Increased LDL oxidation, hormone metabolism
Martyniuk et al. 2011	Methoxychlor	25 mg/kg	48 hours	IP injection	Male	Liver	Altered glycolysis, steroid biosynthesis and metabolism
Garcia-Reyero et al. 2008	17β-estradiol	1 mg/kg	48 hours	IP injection	Male	Liver	Decreased lipid metabolism, transport, glucose metabolism, glycolysis, monosaccharide metabolism, hexose metabolism, carbohydrate transport, cellular lipid metabolism, alcohol catabolism
						Testis	Altered mitochondrial function, cellular metabolism, steroid metabolism

Environmental contaminants also affect lipid metabolism and bioenergetics in the testis. For example, a field study with several lakes impacted by perfluorinated chemicals altered transcripts for selected biomarkers related to lipid transport, cholesterol efflux, carbohydrate metabolism, electron transport chain activity, and fatty acid elongation in the testis (Colli-Dula et al., 2016). It was suggested that these changes may be caused by PPAR activity, as many perfluorinated chemicals interact with this receptor and alter lipid metabolism (Janesick et al., 2016). However, these changes are not exclusive to PPAR activity in largemouth bass testis. For example, heavy metal exposure to cadmium affects carbohydrate metabolism, steroidogenesis, and the tricarboxylic acid cycle (Mehinto et al., 2014). Similar to the ovary, steroid hormone activity also affects lipid metabolism pathways, as exposure to 17β-estradiol impacts gene ontologies for several bioenergetic pathways (Garcia-Reyero et al., 2008), indicating that these responses are not sex-specific. From these studies, it is evident that several types of chemical stressors can affect lipid metabolism and bioenergetics in largemouth bass gonads.

As the site of vitellogenesis and xenobiotic biotransformation, the liver is an important component of the reproductive axis where chemical stressors can affect lipid metabolism. In the female largemouth bass liver, a feeding study with 17α-ethinylestradiol increased sub-networks for LDL oxidation (Colli-Dula et al., 2014), which may affect lipid modification of vitellogenin. There are several studies investigating the effects of chemicals in male largemouth bass liver, particularly as exposure to estrogens can induce vitellogenin synthesis (Heppell et al., 1995; Sumpter and Jobling, 1995). For example, exposure to the steroid hormone 17β-estradiol downregulated several gene ontologies for lipid metabolism and transport pathways in male largemouth bass liver including lipid metabolism, as well as other bioenergetic pathways including glucose metabolism, glycolysis, monosaccharide metabolism, hexose metabolism, carbohydrate transport, cellular lipid metabolism, and alcohol catabolism (Garcia-Reyero et al., 2008). Similarly, methoxychlor, which is an OCP and a weak estrogen, also altered gene networks for glycolysis, and steroid biosynthesis and metabolism in largemouth bass liver (Martyniuk et al., 2011), indicating that these changes may be related to steroid hormone receptor activation. As before in the gonad, heavy metals and perfluorinated chemicals also affect lipid metabolism pathways in the male liver (Mehinto et al., 2014), suggesting that several modes of action may exist for altered bioenergetics in the liver. These changes should be further investigated as the liver fulfills several roles related to reproduction, metabolism, and energy storage, where altered lipid metabolism and bioenergetics may result in fitness costs related to these physiological functions.

4. Chemical stressors in fish and lipidomics

Although there is growing evidence that chemical stressors can alter gene networks related to lipid signaling and metabolism, a small number of studies have investigated the effects of chemical exposure on lipid composition in fish. Most of this development has occurred in zebrafish (ZF), a popular model for biomedical research. For example, exposure to the mitochondrial uncoupler 2,4-dinitrophenol decreased phospholipid and glycolipid fatty acid contents, as well as the phospholipid fatty acid unsaturation index in ZF embryos (Hachicho et al., 2015). Similarly, exposure to trichloroethylene, perfluorooctane sulfonate, bisphenol A, tributyltin, and chiral pesticides such as cis-bifenthrin resulted in alterations in lipid profiles in ZF embryos (Ortiz-Villanueva et al., 2018; Pirro et al., 2016; Xiang et al., 2019). Adult zebrafish have also been used to examine lipid responses; liquid chromatography-quadrupole time-of-flight mass spectrometry (LC-qTOF/MS) has been used to measure lipids in zebrafish gill, liver, intestine, and brain following a 30-day exposure to microcystin toxins from cyanobacteria. A KEGG pathway analysis of the metabolite data determined that lipid metabolism was most affected in the gills. Among various lipid pathways, cholesterol had the highest number of altered metabolites, and there were also significant differences for pathways related to essential fatty acids and lipid oxidation (Pavagadhi et al., 2013).

In addition to ZF, other fish species have been used to explore lipid responses to chemical exposure. Japanese medaka (Oryzias latipes) is another common toxicology model, and lipids have been characterized in this species following exposure to the organophosphate pesticide chlorpyrifos (Jeon et al., 2016). Using matrix assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF/MS), it was found that several phospholipids were significantly affected by chlorpyrifos exposure, and the phosphatidylcholine “PC(18:1(11E)/0:0)” was a notable biomarker that decreased in abundance with exposure. In another study using juvenile rainbow trout (Oncorhynchus mykiss) exposed to 10 to 40 ug chlorpyrifos/L for 7 days, brain tissues were examined for alterations in lipid metabolism and fatty acid profiles by LC-MS/MS (Greer et al., 2019). The hypothesis tested in the study was that chlorpyrifos altered serine lipase activity in fish brains, thus compromising neuronal signaling and impacting neurobehavioral responses. The study showed that phospholipid lipase activities were limited by chlorpyrifos, as expected. The critical signaling lipids that were significantly reduced included lysophosphatidylethanolamines with a concomitant increase in their respective phosphatidylethanolamines, (e.g., PE (P16:0/18:1)), suggesting that chlorpyrifos also altered lipid metabolism (Greer et al., 2019).

Recently, lipidomics has also been deployed in the field to support environmental monitoring. Whole body lipid profiles in mosquitofish (Gambusia holbrooki) inhabiting metal(loid)-contaminated wetlands revealed deregulation of cellular membrane lipids (e.g., glycerophospholipids, cholesterol, sphingolipids) and increased energy storage molecules (e.g., triacylglycerols and fatty acids) via nuclear magnetic resonance (NMR) spectroscopy (Melvin et al., 2019). Non-invasive approaches have also been employed to measure lipids in skeletal muscle of two Mediterranean cyprinidae (Barbus meridionalis, Squalius laietanus) exposed to urban and industrial wastewater. Ultrahigh performance liquid chromatography coupled with high-resolution mass spectrometry (UPLC-HRMS) determined there was clear separation of lipid profiles between polluted and reference sites, driven by phosphatidylcholines and phosphatidylethanolamines. Specifically, the polluted site had a decrease in PCs, PC-plasmalogens, and highly unsaturated PCs, as well as an increase in plasmanyl-PCs, lyso-PCs, and cholesterol esters, which suggests oxidative stress and altered cholesterol homeostasis (Marqueño et al., 2019). Disturbances in lipid homeostasis have also been reported in rainbow trout exposed to perfluorooctane sulfonate (Stefanovic, 2018), yellow croaker (Pseudosciaena crocea) and Japanese seabass (Lateolabrax japonicus) following a tropical storm (Yan et al., 2012), and Mozambique tilapia (Oreochromis mossambicus) affected by pansteatitis in suspected polluted waters (Koelmel et al., 2019), among others (Al-Salhi et al., 2012; Arukwe and Mortensen, 2011; Li et al., 2010; Qiao et al., 2016). These studies highlight the efficacy of lipidomics to identify early molecular responses and potential biomarkers for chemical exposure in fish.

5. Advances in lipidomics technologies

Recent advances in mass spectrometry and bioinformatics have provided unprecedented opportunities for lipid analysis. In previous years, these analyses were primarily limited to radioactive labeling to understand lipid fate, as well as compound-specific isotope analysis (Hixson et al, 2014; Katan et al., 2019). Similarly, gas chromatography and high-performance liquid chromatography have been used to study lipids for many years, but these approaches were inherently limited and lacked the capability to profile multiple lipid classes or low-abundance lipids. Several advances including soft ionization methods (electrospray ionization, matrix-assisted laser desorption/ionization), tandem mass spectrometry, multiple reaction monitoring and high-resolution/accurate mass analyses have fueled the growth of the field, providing improvements in lipid separation and profiling across several lipid classes (Wenk, 2005, 2010). These technologies have greatly increased the versatility of both direct infusion-mass spectrometry (also known as shotgun lipidomics) and chromatography-based lipidomics workflows. In shotgun lipidomics, the intrinsic electrochemical properties of different lipid classes/species are exploited to achieve separation in the ion source of the mass spectrometer at a constant concentration of the sample solution (via infusion), without prior chromatographic separation (Han and Gross, 2003, 2005). This workflow enables a virtually unlimited number of experiments to be performed at a constant analyte/solvent ratio, and since spectra can be obtained in just a few minutes, this technique is amenable for high-throughput applications. However, the concurrent ionization of lipids from all lipid classes, which is the characterizing feature of this technique, can result in ionization suppression of lipids with low electric potentials and/or low endogenous concentrations, lowering instrument sensitivity and potentially preventing their detection (Khoury et al., 2016). Thus, the analysis of lipids in the low-abundance regime remains a challenge that can only be accomplished via liquid chromatography/mass spectrometry-based methods. Additionally, the interpretation of lipidomic data from shotgun analyses can be challenging due to the highly convoluted spectra that are obtained as a result of contributions from fatty acid isomers and double-bond or positional isomers. In contrast, the increasing diversity and technological advancements of liquid chromatography columns that are now commercially-available and the advent of two-dimensional liquid chromatography, can enable compound separations based on a myriad of physical/chemical characteristics including degree of polarity (Yang et al., 2017), unsaturation (Yang et al., 2012), chirality (Lee and Blair, 2009) and charge (Bang and Moon, 2013) permitting the deepest interrogations of complex lipid samples yet. Despite the quantitative robustness of liquid chromatography/mass spectrometry, longer run times and the need for skilled users with a background in compound separations have limited the adoption of this technique in certain applications, particularly in high-throughput screening. Nevertheless, both direct infusion and liquid chromatography/(tandem) mass spectrometry-based methods have a place within the lipidomics community; the selection of the approach will be largely driven by the application and/or specific hypotheses.

The emergence of novel acquisition strategies including data dependent and data-independent acquisition and precursor ion discovery have also increased the analytical potential of lipidomics workflows for untargeted or discovery-based examinations of complex biological matrices. Specifically, data-independent acquisition is an unbiased approach that relies on retention-time matching of low-energy (i.e. full-scan MS) and high-energy scans (i.e. MS/MS) without quadrupole pre-selection of precursor ions. This technology is, in principle, the best solution for untargeted analyses of compounds of low abundance, as this is a known disadvantage of common data-dependent strategies (Fenaille et al., 2017). Precursor ion discovery is a semi-targeted approach that combines features from both data-independent and data-dependent methods in order to identify known or unknown compounds that share defined functional groups or structures (e.g., complex lipids that contain a specific fatty acid) (Zhou et al., 2009). This latter technology can be particularly useful in the discovery of novel lipids as biomarkers of disease that may not yet exist in curated lipid libraries. Ion mobility-mass spectrometry has more recently been made amenable to these and other analytical strategies and has increased the practical usability of both targeted and untargeted methods. By introducing an additional dimension of separation that considers the size, shape, charge and/or mass of ions as they travel through a gas-filled cell, lipids that have been traditionally difficult to resolve by conventional chromatographic methods can now be profiled. This includes the discrimination of sn-glycerol regioisomers, carbon-carbon double bond positional isomers, and cis-/trans- geometrical stereoisomers (Jeanne Dit Fouque et al., 2019). Alternative applications that incorporate ozone-induced dissociation for isomer elucidation with or without ion mobility have also been reported (Marshall et al., 2019; Poad et al., 2017), providing robust results that can enhance our abilities to interrogate complex samples and confidently characterize and measure lipid structures. Untargeted (or nontargeted) high-resolution mass spectrometry approaches are hypothesis-driving strategies that promise to be advantageous in an era where we know that many endocrine-active chemicals are promiscuous (i.e., they can affect various physiological mechanisms and pathways). Further, the adoption of multi-omics approaches is also a necessary evolution. In order to characterize and appreciate the overall exposure effects, methodologies capable of synchronizing omics data sets from the gene, transcript, protein, metabolite and system levels need to be realized and employed. Consequently, the growth of the field of lipidomics will benefit from implementing the strategies noted above and further technological and analytical developments, in addition to expanding bioinformatics resources and lipid libraries that include structures unique to aquatic species.

6. Challenges and future applications of lipidomics in toxicology

With the advancements of lipidomics in systems and molecular biology, unique applications in toxicology are also beginning to emerge. For example, there has been a growing interest in lipid peroxidation following toxicant exposure, as this process has been implicated with altered cellular homeostasis and several diseases (Berliner and Zimman, 2007). Lipid oxidation products are increasingly utilized as biomarkers to identify mitochondrial dysfunction, oxidative stress, and inflammation. Similarly, lipids have been used to develop biomarkers that identify modes of action for chemicals in the environment. Jungnickel and colleagues (Jungnickel et al., 2014) used lipidomics in MCF-7 cells exposed to unique toxicants to develop metabolomic biomarkers for different toxicological signaling pathways. While these studies demonstrate the recent emergence of lipidomics in toxicology, there remain several research needs that have been overlooked, particularly for ecotoxicology. Strong associations between nutrient availability and altered tissue lipid levels have also been reported, but the mechanisms underlying these responses remain a topic of ongoing research. For example, alpha-tocopherol (vitamin E) has been demonstrated to play a critical role in the prevention of highly-unsaturated fatty acid peroxidation in ZF embryos (Choi et al., 2015), with deficiency resulting in significant morphological abnormalities and impaired locomotor responses (McDougall et al., 2016). In another study, a diet that included fucoidan (a sulfate-containing polysaccharide found in various species of brown seaweed) resulted in significant reductions in fat mass in yellow head catfish (Pelteobagrus fulvidraco) (Xu et al., 2017). Furthermore, fish mercury levels have been negatively associated with omega-3 fatty acid content (Laird et al., 2018; Reyes et al., 2017; Strandberg et al., 2016), and industrial effluents have also been associated with altered omega-6 metabolism in whitefish (Coregonus lavaretus) liver and gonad (Toivonen et al., 2001). While these relationships have not been fully characterized, it is evident that the implementation of lipidomics has the potential to enable the identification of specific metabolic targets for the elucidation of physiological mechanisms.

Moving forward, it will be important to integrate lipidomics data with responses at various levels of biological organization. Adverse outcome pathways (AOPs; Ankley et al., 2017) have provided a useful framework to integrate omics data (Brockmeier et al., 2017), and lipidomics is no exception. These pathways consist of a series of causal key events (KE) that link a molecular initiating event (MIE) to an adverse outcome (AO) (Villeneuve et al., 2014). In this context, lipidomics may be useful to elucidate important KEs that underlie apical responses that are considered an AO (Figure 1). As summarized in this review, there are numerous proteins that regulate lipid biosynthesis and metabolism, and these proteins can be affected by environmental chemicals. In the AOP framework, these interactions are characterized as MIEs, which can lead to subsequent KEs related to lipid metabolism and bioenergetics. By extension, dynamic energy budgets can be used to understand how these changes lead to subsequent effects on apical endpoints, such as survival, growth and reproduction (Murphy et al., 2018).

Figure 1.

Adverse outcome pathways involving key events with lipidomics.

In addition to this general mechanism, there are also specific examples where lipids are linked to apical endpoints in fish. For example, a number of chemicals affect mitochondrial function by binding or inhibiting specific protein complexes in the electron transport chain (e.g., Complex I) (Dreier et al., 2019). This activity can lead to oxidative stress, lipid peroxidation, and atresia in oocytes, thereby affecting reproduction. Similarly, because lipid stores are the primary source of energy for developing embryos (Brooks et al., 1997), lipid profiling may offer a useful indicator for egg quality, fertility, and/or potential hatchability. Lipids are also important signaling molecules for reproduction; prostaglandins are responsible for ovulation (Stacey et al., 1982), and there is an established AOP linking cyclooxygenase inhibition to decreased prostaglandin synthesis, ovulation, and by extension fecundity and reproduction (Martinović-Weigelt et al., 2017). Because lipids are implicated in multiple AOPs, lipidomics offers a useful approach to inform key events (blue KEs in Figure 1) that affect reproductive outcomes in fish.

Integrating lipidomics data in the AOP framework may also complement other sources of information that describe related mechanistic, metabolic, or apical responses. For example, the present review of transcriptomic data in largemouth bass strongly suggests that chemical stressors affect genes related to lipid metabolism and other bioenergetic pathways. Similarly, provided an established AOP, mechanistic data can be used to identify chemicals associated with altered lipid metabolism. One resource is the publicly-accessible Comparative Toxicogenomics Database (http://ctdbase.org/), hosted by North Carolina State University, which collects chemical-gene interactions for numerous targets (Davis et al., 2019), including those related to lipids. For example, there were numerous chemical-gene interactions for the peroxisome-proliferator active receptor gamma (PPARγ; Figure 2, Table 2, Supplemental Table 1), which is a molecular initiating event in several AOPs published (Tsakovska et al., 2014) or under development in the AOP wiki (https://aopwiki.org/) (e.g., Epigenetic modification of PPARG leading to adipogenesis; PPARgamma activation leading to sarcomas in rats, mice, and hamsters; Peroxisome proliferator-activated receptors γ inactivation leading to lung fibrosis). Interestingly, many environmental chemicals had gene interactions with PPARγ, including bisphenol A, ethanol, tributyltin chloride, and benzo[a]pyrene. While AOPs themselves are not chemical-specific (Villeneuve et al., 2014), it remains important to identify MIEs affecting lipid metabolism, as well as chemicals that may potentially affect these targets. Similarly, AOPs should include lipid responses from non-chemical stressors, such as temperature or hypoxia, as well as multiple stressor responses (Groh et al., 2015; LaLone et al., 2017) to more completely understand the variables driving lipid profiles observed. Lipidomics experiments will be necessary to ascertain the strength of these stressor-MIE-lipidome interactions and determine whether they lead to apical outcomes through causal key event relationships. In conclusion, there is a promising future for lipidomics in the field of ecotoxicology, and it is expected that these technologies will provide useful information to support AOPs and evaluate the safety of chemicals in the environment.

Figure 2.

Selected chemicals (small molecules) that regulate expression of the peroxisome proliferator-activated receptor gamma in the Comparative Toxicogenomic Database (accessed 2019–11-23).

Table 2.

Selected chemicals affecting gene expression of the peroxisome proliferator-activated receptor gamma. Gene interactions were counted from the Comparative Toxicogenomic Database (accessed 2019–11–23).

Chemical Name	Chemical ID	CAS RN	Interaction Count	Organism Count
Bisphenol A	C006780	29348	71	7
Ethanol	D000431	64–17–5	34	3
Tributyltin	C011559	688–73–3	34	4
Sodium arsenite	C017947	13768–07–5	33	5
Mono-(2-ethylhexyl)phthalate	C016599	4376–20–9	30	5
Carbon tetrachloride	D002251	56–23-5	25	2
Diethylhexyl phthalate	D004051	117–81–7	25	3
Cadmium chloride	D019256	10108–64–2	24	4
Tetrabromobisphenol A	C020806	79–94–7	20	6
Perfluorooctanoic acid	C023036	335–67–1	17	3
2,2’,4,4’-tetrabromodiphenyl ether	C511295	5436–43–1	16	4
Arsenic trioxide	D000077237	1327–53–3	16	2
Benzo(a)pyrene	D001564	50–32–8	15	3
Lead acetate	C008261	301–04–2	15	1

Highlights:

• Lipids are diverse molecules and perform several essential roles in fish

• Transcriptomics data suggest that chemicals alter lipid metabolism and regulation

• Analytical technologies in mass spectrometry have advanced the field of lipidomics

• Lipidomics can inform adverse outcome pathways