Metabolism plays a central role in many cellular processes, and its dysregulation is a hallmark of many disease states, including cancer, obesity, and diabetes. Recent work (Cirulli et al., 2019) has shown that certain health effects, particularly cardiovascular disease, can be predicted from metabolic signatures prior to clinical manifestations. This suggests that metabolic dysregulation has causal influence over the disease state of an organism, and conversely disease may be preventable via metabolic intervention (Rubino et al., 2016).

An evolutionary system with particularly extreme changes in metabolic regulation is the Mexican tetra, Astyanax mexicanus, which has undergone considerable physiological and behavioral changes to colonize a number of subterranean caves in the Sierra de El Abra region of Mexico. Cavefish have evolved a suite of metabolic phenotypes to cope with the cave environment, including lower metabolic rate, increased appetite, fat storage, and starvation resistance (Aspiras et al., 2015; Hüppop, 1986; Xiong et al., 2018). Cavefish are also insulin resistant, hyperglycemic, and exhibit increased caloric intake (Riddle et al., 2018), a feature often associated with decreased longevity. A notable genomic feature in Pachón and Tinaja cavefish is a mutation in the insulin receptor (Riddle et al., 2018) that, in humans, is linked to Rabson-Mendenhall (RM) syndrome, a form of severe insulin resistance that causes many developmental abnormalities and typically progresses to ketoacidosis (Longo et al., 1999). Nevertheless, cavefish do not appear to suffer any of the adverse effects of RM, lack advanced glycation end products (AGEs) (Riddle et al., 2018) normally associated with hyperglycemia, and live long, healthy lives without ill-effects of metabolic disease (Riddle et al., 2018). A. mexicanus may provide natural solutions to overcome the challenges associated with adverse metabolic conditions (Krishnan and Rohner, 2019).

From an evolutionary standpoint, survival in the cave environment requires resistance to long periods of nutrient deprivation. This, in turn, leads to storage of excess energy in fat and glycerol, which are themselves potentially harmful to the host organism. We hypothesize that survival in the cave environment thus requires multiple, counterbalancing evolutionary changes and that the combined effect of these changes is to make cavefish resilient to a variety of metabolic conditions, of which starvation and triglyceride / sugar accumulation are discrete examples.

We thus sought to characterize the metabolic signature of resilience by examining the metabolome of Pachón and Tinaja cavefish (two recently derived cave populations) compared to surface-dwelling populations using untargeted mass spectrometry (MS) of primary metabolites and lipids. We characterized the response of energetically important tissues (the liver, muscle, and brain) of each population in short/long-term fasted, and fed conditions. We demonstrate that metabolite profiles in Pachón and Tinaja cavefish are more similar to each other than surface fish in each feeding state / tissue combination, highlighting the role of parallel evolution in shaping the metabolome of cavefish. We identify metabolic signatures and metabolites exhibiting the most pronounced regulatory changes within each tissue, population, and feeding state. We constructed inter-population and inter-feeding state comparisons and fit separate statistical models to each case (Figure 7—figure supplement 1). Cavefish exhibit many similarities with metabolic syndrome, but also differ from these conditions in terms of antioxidants, metabolites associated with cellular respiration / the electron transport chain, and unexpectedly reduced levels of cholesteryl esters.

Our results lay the groundwork to explore the mechanistic roles of metabolites and pathways in the nutrient availability-related adaptations of cavefish and suggest that natural evolutionary systems may offer insights into metabolic function by showing how disease states can be altered or counterbalanced under a genetic background more suited to a different set of parameters governing metabolic state.

In order to aid others in reproducing our analysis, we have provided a pipeline to reproduce all major figures and results on this paper that are derived from our metabolomics dataset at https://www.stowers.org/research/publications/libpb-1699 (ftp://odr.stowers.org/LIBPB-1699). We also provide a shiny app at https://cavefin.shinyapps.io/shiny to allow others to explore and visualize our dataset.

Our experimental design aimed to (a) characterize the response of the A. mexicanus metabolome to different feeding states in energetically expensive tissues and (b) utilize comparisons across populations and feeding states to identify metabolites conserved in cavefish populations. Food scarcity is one of the cave’s harshest evolutionary pressures. Cavefish have specialized feeding strategies and fat metabolism that helps them thrive in the cave environment (Jeffery, 2020; Aspiras et al., 2015; Hüppop, 1986; Xiong et al., 2018). We raised age-matched offspring of Surface (river) fish, and Pachón and Tinaja cavefish morphs originating from two independent cave colonizations. To understand how the cavefish metabolome adapts to ecologically relevant food challenges, we separated Surface, Pachón and Tinaja populations into three different groups at 4 months: 30 day fasted, 4 day fasted, and “Refed” (fed at 3 hr prior to collection after 4 days without food) (Figures 1 and 2). We also show RNA-Seq validation (Figure 6—figure supplement 1, Figure 6—figure supplement 2, Supplementary file 1-table s2) of the main themes observed from the metabolomic data.

Figure 1

Download asset Open asset

Figure 2

Download asset Open asset

Our untargeted metabolomics study yielded a total of 174 identified metabolites linked to KEGG (Kanehisa and Goto, 2000) / HMDB (Wishart et al., 2018) IDS and 483 identified lipids linked to LIPIDMAPS IDS. We examined the effect of normalizing identified peak values by the total sum of peaks (mTIC) and by sample weight (Fig supplement 2) and found that mTIC is more robust to variations in sample weight. Hence, we employed mTIC-normalized data for the remainder of the analysis.

Figure 3A/B shows that metabolites cluster primarily by tissue, in line with previous studies in mammals (Ma et al., 2015). Figure 3C/D shows how clustering patterns depend strongly on the chemical classification of identified metabolites. Some lipid and primary metabolite categories show a clear separation between different populations (e.g. carbohydrates, Figure 3C, and glycerophospholipids, Figure 3D, across most tissues), whereas other categories have a less pronounced change (amino acids in the brain, Figure 3C, and fatty acyls in most tissues). In order to quantify separation of feeding states as a function of population and metabolite category, we used a supervised machine learning method based on orthogonal projection of latent structures (O–PLS, Figure 7—figure supplement 3). We then used O–PLS to remove ‘orthogonal’ variation (Trygg and Wold, 2002) from each metabolite category and fit a Bayesian logistic regression model to the de-noised data (Supplemental Methods).

Figure 3

Download asset Open asset

In general, the metabolome of all three populations shows a large degree of similarity within a given tissue (Figure 7—figure supplement 4), highlighting the influence of evolutionarily conserved functions of individual tissues.

Sugar phosphate metabolism

Given the overall similarity at the tissue level for most classes of metabolites (Figure 7—figure supplement 4), phenotypic differences are likely to be linked to a relatively small subset of the metabolome. We sought to identify metabolites that could be responsible for the drastic change in phenotype of cave populations. Sugars and sugar phosphates are important energy metabolites and hence candidates for adaptations related to resistance to nutrient deprivation. This class of metabolites displays a dramatic change during short- and long-term fasting, particularly in the liver (Figure 4). Transcriptomics data from multiple studies by our group also displays an overall trend toward upregulation of sugar metabolism in cavefish (Figure 6—figure supplement 1).

Figure 4

Download asset Open asset

Hepatic glucose production is derived from gluconeogenesis and glycogenolysis, the latter relying on stored glycogen, which is quickly exhausted during fasting (Han et al., 2016), indicating that hepatic gluconeogenesis likely plays a role in sustaining survival under long-term nutrient deprivation in A. mexicanus. Surprisingly, surface fish also show stable (albeit generally lower) sugar levels in the liver under different feeding states (Figure 4), indicating that sugar production in the liver may be driven by overall demand rather than supply. This may point to a shift from oxidative to sugar-based metabolism as an energy source in energetically expensive tissues. Cavefish possess a larger amount of body fat (Xiong et al., 2018; Aspiras et al., 2015), and hence have a larger pool of glycerol to serve as a substrate for gluconeogenesis. We find that both cave populations exhibit decreased levels of glycerol in the 30-day fasted state, particularly in Tinaja (Supplementary file 1), indicating increased consumption of this intermediate as a substrate for gluconeogenesis may be the source of increased sugar / sugar phosphate abundance in cave populations.

Regardless of the substrates leading to sugar metabolite accumulation, it is clear that a large difference exists between cave and surface populations within this class of metabolites. However, the specifics of this alteration to sugar metabolism appear to be population-specific, with Tinaja showing a large increase in sugar phosphates and Pachón showing an increase in unphosphorylated sugars respectively in short/long-term fasted states in comparison to surface (Figure 4).

Other tissues show a mixed response, with muscle displaying increased levels of most sugar phosphates in Pachón but decreased levels of fructose-1-phosphate in both cave populations with respect to surface. The brain displays low levels of sugar / sugar phosphate metabolites overall but possesses increased sugar metabolite abundance in cave populations for certain metabolites and feeding states (Supplementary file 1).

While the levels of most simple sugars and sugar phosphates are increased in cavefish with respect to surface, the levels of gluconic acid and glucoronic acid show the opposite pattern (Figure 4). Gluconic acid and glucoronic acid belong to uronic acids, a class of sugar acids that are major building blocks of proteoglycans. It has previously been observed that A. mexicanus cave morphs lack advanced glycation end products (Riddle et al., 2018), which are a defining feature of diabetes and are normally associated with chronic hyperglycemia in humans. Altered metabolism of sugar acids in cavefish may play a role in inhibiting excessive protein glycation and the adverse health effects thereof.

Ascorbate

A highly unexpected and unexplained feature of our analysis is the abundance of vitamin C, particularly in its oxidized form dehydroascorbic acid (DHAA), across all tissues in cave populations (Figure 5). Ascorbic acid (AA), the reduced, active form, is also more prevalent in muscle tissue. DHAA can be recycled back to AA using reducing cofactors such as NADH and NADPH, which can in turn be regenerated from the pentose phosphate pathway and TCA cycle using simple sugars (which cavefish possess in great abundance). For this reason, vitamin C content in food labeling is usually reported as the sum of AA and DHAA (Wilson, 2002). Thus, cavefish possess a larger total ‘pool’ of vitamin C (including interconvertible oxidized and reduced forms, Figure 5).

Figure 5 with 2 supplements see all

Download asset Open asset

Many cavefish populations exhibit increased appetite and carry an allele of the melanocortin 4 receptor that predisposes them to hyperphagia (Aspiras et al., 2015). The increased appetite could cause cavefish to consume more overall food, which could be responsible for the AA/DHAA increase in the refed state. However, this does not pertain to the 30-day fasted state, where AA/DHAA levels are also elevated across all tissues. There is widespread consensus that teleosts, like humans, lack the ability to produce AA endogenously due to the absence of gulonolactone oxidase, which catalyzes the final step in AA biosynthesis (Ching et al., 2015). In humans, this enzyme is a pseudogene, whereas in teleosts the gene is absent entirely, thought to be lost in the distant evolutionary past. Thus, the additional AA/DHAA supply likely comes from selective reuptake in the kidney, a process that also occurs in humans to conserve AA/DHAA, or it may be produced by commensal microbiota in cavefish. Trace amounts of AA/DHAA in the feed used in the aquatics facility used to house the fish in this experiment may recirculate throughout the water filtration system and be redistributed to all tanks, including those housing fish in the fasted groups. Nevertheless, it remains that even in the case of circulating trace amounts of AA/DHAA, cavefish appear to exhibit selective retainment of AA/DHAA in larger quantities.

The advantages of AA conservation in adaptation to an environment where prolonged starvation is common are self-evident. AA is involved in collagen formation, and its deficiency leads major loss of integrity of connective tissue. Thus, the ability to retain what little ascorbate is present in underground cave environments would confer an enormous survival advantage to fish.

Another factor that could influence the AA/DHAA ratio is the effect of insulin resistance and hyperglycemia on the GLUT family of transporters, particularly GLUT4 in adipose/muscle tissue (Wilson, 2002; Huang and Czech, 2007). DHAA competes with glucose for transport across the membrane by GLUT4, whereas AA is taken up by Na⁺ transporters. GLUT4 activity is dependent on membrane translocation and this process is dysregulated in diabetes (Jaldin-Fincati et al., 2017). This combination of elevated blood sugar and insulin resistance suggests that GLUT4 could be less active in cavefish and cause DHAA to accumulate in the extracellular space. AA/DHAA have also been reported to influence C-reactive protein levels (Ford et al., 2003) and protein glycation (Franke et al., 2013), prompting further investigation into A. mexicanus as a model of diabetes-related resilience. Finally, Pachón cavefish possess a reduction in neutrophils, one cell type which are normally involved in the uptake of AA and reduction to DHAA, compared to surface (Peuß et al., 2020).

Obesity and inflammation-related metabolites

Chronic inflammation of adipose tissue is a common feature of obesity and can often lead to insulin resistance and eventually type 2 diabetes (Paschoal et al., 2020). Cave populations of A. mexicanus have been previously reported to exhibit pronounced insulin resistance (Riddle et al., 2018), but do not accumulate advanced glycation end products and do not appear to have diminished longevity.

In order to compare the metabolome of A. mexicanus cave populations to the known metabolic signatures of obesity (Cirulli et al., 2019), we calculated changes in lipid categories (the coarsest abstraction used in LipidMaps), classes (a more detailed partitioning scheme used in LipidMaps), and, within free fatty acids specifically, the degree of saturation. The metabolome displays a remarkable overlap with the proinflammatory signature associated with obesity that, in humans, leads to insulin resistance. This signature consists of Aspiras et al., 2015 the elevation of saturated fatty acids (SFAs) in muscle in most feeding states, which have a direct and pronounced proinflammatory effect in mammals through the recruitment of macrophages (Glass and Olefsky, 2012), although the importance of fatty acid release in insulin resistance is disputed (Morigny et al., 2019; Karpe et al., 2011; Xiong et al., 2018) abundance of ceramides in muscle in all feeding states, which are known direct mediators of insulin signaling (Glass and Olefsky, 2012). Indeed, the only feeding state for which skeletal muscle did not display increased SFA abundance was 30-day fasting, which could simply indicate the exhaustion of free SFA pools. Additionally, palmitate, a precursor of ceramide biosynthesis (Glass and Olefsky, 2012), is elevated in muscle in all feeding states. Sphingoid bases are significantly more abundant in muscle in all feeding states, suggesting generally upregulated sphingolipid biosynthesis in cave populations.

In contrast to proinflammatory metabolites, omega-3 fatty acids (ω-3 FAs) such as DHA and EPA have protective effects against inflammation (Paschoal et al., 2020; Glass and Olefsky, 2012; Oh et al., 2010). These molecules bind to the GRP120 receptor on macrophages and adipocytes, and the activated receptor then modulates the activity of PPARγ and ERK (Paschoal et al., 2020; Oh et al., 2010). ω-3 FAs are less abundant in the liver under 4-day fasting and are generally not upregulated in most feeding states and tissues. Thus, ω-3 FAs do not appear to offset for the proinflammatory signature of cavefish SFA and ceramide signatures, suggesting that cavefish possess an alternate compensatory mechanism to prevent chronic tissue inflammation. Overall, cavefish appear to exhibit many metabolic similarities with obesity and health conditions associated with it.

However, this is not a universal trend. Cirulli et al. report a strong association between urate levels and BMI, likely due to insulin resistance interfering with uric acid secretion in the kidney Cirulli et al., 2019. In contrast, cavefish appear to have significantly reduced levels of uric acid in muscle, and in other tissues levels are comparable with surface except for a small but significant increase in Pachón liver during fasted states. Mannose, which is associated with obesity and insulin resistance (Cirulli et al., 2019), was abundant in the Pachón liver in all feeding states, but was reduced in Tinaja compared to surface fish.

Finally, cholesteryl esters and cholesterol in some feeding states (Table Supplementary file 1), were less abundant in cave populations. Using a previously published gene expression dataset (Krishnan et al., 2022), we investigated factors that might influence levels of cholesteryl esters. Cholesteryl ester transfer protein (CETP), which transports cholesterol in / out of lipoproteins, is downregulated in both Pachón and Tinaja compared to surface fish (Figure 7—figure supplement 5), suggesting a potential causal relationship between lower cholesteryl ester levels in cave populations and this important carrier protein. A human variant of CETP associated with decreased serum levels of the protein and larger low-density lipoprotein (LDL) and high-density lipoprotein (HDL) particle sizes has been linked to exceptional longevity (Barzilai et al., 2003). The LDL / HDL cholesterol ratio, mediated in part by CETP (Christison et al., 1995), is a major contributor to risk of atherosclerosis and coronary heart disease (Brousseau et al., 2004). The reliance on triglycerides as an energy source in cavefish may increase the risk of arterial disease by providing an abundance of free fatty acids and other lipids. Cholesteryl esters, in particular, are formed from esterification of a fatty acid and cholesterol, are a major constituent of foam cells in atherosclerotic lesions (Ghosh et al., 2010; Yu et al., 2013), and, for certain lipid species, show a large difference in abundance between cavefish and surface (Figure 6). Inhibition of CETP has been shown to reduce cardiovascular risk, ostensibly by altering the partitioning of cholesteryl esters between LDL and HDL (Barter and Rye, 2012). Indeed, lipid homeostasis in zebrafish exhibits strong similarities with human (Fang et al., 2014), and zebrafish express CETP whereas other model organisms such as mice do not. This raises the question of whether CETP and its cholesteryl ester substrates may be a locus under selection in cavefish to offset the risks associated with increased visceral fat.

Figure 6 with 2 supplements see all

Download asset Open asset

Large differences in CETP expression suggest that lipid homeostasis may be regulated differently between cave and surface populations. Lipoproteins, in the form of HDL, IDL, LDL, VLDL, and chylomicrons contribute to and regulate lipid homeostasis.

In summary, lipid metabolism in A. mexicanus represents a hub of evolutionary activity which clearly separates surface and cave populations. Cavefish must balance increased demands on energy storage with counter-adaptations to protect against pro-inflammatory and atherogenic metabolites. We observe elevated levels of most energy metabolites, with the notable exception of cholesterol and cholesteryl esters, and cavefish have a larger (V)LDL/HDL ratio as compared to surface fish. LDL and VLDL have higher triglyceride content as compared to HDL (Cox and García-Palmieri, 2011) and VLDL is a substrate for lipoprotein lipase (Freeman and Walford, 2016), which liberates free fatty acids from triglycerides. LDL/VLDL may thus be important for energy metabolism in cavefish, and hence selective pressure may contribute to the higher (V)LDL/HDL ratio in this population.

Resistance to nutrient deprivation

In order to determine the basis of cavefish adaptation to low-nutrient environments, we sought a statistical test that would be sensitive to metabolites that change significantly between refed and short/long-term fasted states and insensitive to metabolites that remain relatively stable across feeding states. We further hypothesized that certain metabolites may have an important role in cave adaptation. Pachón and Tinaja represent independently derived populations, and we reasoned that a test for parallel adaptation should be selective for metabolites that show the same differential feeding state response pattern across cave populations (e.g. differentially increased in both Pachón and Tinaja fasted states relative to surface). In order to construct this test, we fitted a Bayesian GLM (Gelman et al., 2008) to a linear combination $(P+T)/2-S$ of O-PLS-filtered z-score values (see Materials and methods), where $P$ stands for Pachón, $T$ stands for Tinaja, and $S$ stands for surface. We used this test to identify metabolites that might have a role in the fasting response of cavefish, that is metabolites that are differentially abundant in cave populations in the fasted versus refed state (Pachón and Tinaja are assigned equal weight), and generally show the opposite pattern in surface. Figure 7 shows the results of this test for 30-day fasting vs. refeeding (A, which corresponds to the most extreme experimental groups), and the two other possible comparisons between feeding states (B/C).

Figure 7 with 5 supplements see all

Download asset Open asset

Sugar metabolites do not appear to exhibit a strong differential feeding state response. However, long-chain fatty acids such as palmitate and stearate (Figure 7C) do show differential abundance between long- and short-term fasting, suggesting that cavefish may rely on increased usage of fat stores in long-term fasting. Furthermore, analysis of the 30-day fasting response in cavefish liver highlights orotic acid (OA, Figure 7A), an intermediate in pyrimidine synthesis that has been implicated in fatty liver condition (Standerfer and Handler, 1955). OA is suppressed in all feeding states in cave populations, but exhibits a sharp spike in refed surface fish (Figure 5).

Starvation has detrimental effects on an organism in many ways. One detrimental effect is the depletion of antioxidant substances and the resulting oxidative stress through increasing levels of reactive oxygen species (ROS) (Furne and Sanz, 2017). Studies that focus on the impact of food deprivation on oxidative stress in fish show that prolonged starvation decreases the capacity of fish to ameliorate oxidative stress (Furne and Sanz, 2017). Glutathione is a major antioxidant that detoxifies ROS and thereby prevents cellular damage from oxidative stress (Pompella et al., 2003). Cavefish face prolonged periods of nutrient deprivation in their natural environment (Aspiras et al., 2015). Adaptation to the cave environment may have led to changes in glutathione metabolism in cavefish to protect against oxidative stress under prolonged fasting. Indeed, in an earlier study we were able to demonstrate that cavefish show an increased expression of genes that are involved in the metabolism of glutathione, which is indicative of an increased stress level compared to surface fish in their natural habitat (Krishnan et al., 2020).Here, we can confirm that these trends in gene expression are accompanied by elevation of reduced glutathione in the liver and brain (Figure 5, Supplementary file 1). We did not observe a significant increase of glutathione in the surface fish in the fasted states (Figure 5, Supplementary file 1).

Guided by these observations, we further attempted to characterize the role of altered antioxidant and cholesterol metabolism in cavefish. In particular, we hypothesized that cavefish resilience to widely varying nutrient levels is driven by a robust antioxidant system that prevents the accumulation of ROS under conditions of stress (e.g. induced by fasting).

To test this hypothesis, we examined ROS state in the liver under a subset of our original fasting experiment using similarly aged fish. ROS accumulation has been shown to cause lethal levels of cell damage in flies, but such damage can be prevented via oral administration of antioxidants or over-expression of antioxidant-producing enzymes (Vaccaro et al., 2020). This caused us to ask whether naturally elevated antioxidant levels in cavefish might protect against starvation-induced ROS accumulation. We reasoned that the most extreme changes are observed between 30-fasted and 4-day fasted +refed groups, and that given the similarity in Pachón and Tinaja antioxidant profiles, one cave population would be sufficient for validation. We thus repeated the fasting experiment using only 30-day fasted and 4-day fasted +refed groups, and using only Pachón and surface populations.

Using dihydroethidium (DHE) staining, we examined different populations and feeding states for ROS level. While liver sections of fasted / refed fish did not show differences in cytoplasmic / nuclear localization of DHE (Figure 8B–E), a quantitative analysis showed that, under 30-day fasting, juvenile Pachón cavefish exhibit lower levels of ROS in cytoplasm as compared to surface under the same conditions (Figure 8A), supporting the hypothesis that elevated antioxidants in cavefish may have a protective effect by neutralizing active ROS and may have arisen as an adaptation allowing cavefish to tolerate ROS produced by long periods of nutrient deprivation. We did not observe significant differences in DHE nuclear intensity for either the cytoplasm or the nucleus, suggesting that 30 days of fasting may not be sufficient to produce salient differences in the nucleus for juvenile fish. The known resilience properties of antioxidants may also help explain the metabolic resilience of cavefish by providing a buffer against different forms of stress.

Figure 8

Download asset Open asset

Cavefish thus exhibit decreased ROS abundance in the liver in 30-day fasted conditions, which correlates with the trend we observed in glutathione. Analysis of three separate transcriptomic experiments conducted by our group also points to important changes in key genes, such as glutathione S-transferase, being consistently upregulated in cavefish (Figure 6—figure supplement 1, Supplementary file 1-Table S2). Combined with the findings presented here, this suggests that response oxidative stress is indeed an important factor in cavefish physiology.

Antioxidant levels in cavefish may have evolved as a strategy to allow cavefish to tolerate variation in nutrient availability, as a means of controlling inflammation caused by deleterious energy metabolites, as a way of inhibiting atherogenesis, or several of these effects.

A. mexicanus has been advanced as a model of resilience under ostensibly pathological conditions including hyperglycemia, diabetes (Krishnan and Rohner, 2019), and insulin resistance (Riddle et al., 2018). Here, we have provided a large, untargeted study of the metabolome of A. mexicanus surface fish and two cave populations in order to investigate the molecular underpinnings of these adaptations.

We were particularly interested in the role of metabolism in cave adaptation of the two A. mexicanus cavefish populations in this study: Pachón and Tinaja. This suggests that parallel adaptation to cave environments requires satisfying certain common metabolic needs that are an inherent part of the niche. The obvious candidate for this evolutionary conflux is adaptation to a low-nutrient environment. However, metabolic strategies for survival in such environments are not currently well-understood. We found that drastic alterations in energy metabolism, together with shifts in mediators of redox metabolism and ascorbate, an essential vitamin which is lacking in the cave environment, constitute a major feature of cave adaptation in these populations.

Cavefish appear to have substantially altered sugar metabolism, and exhibit higher levels of sugars and sugar phosphates. However, the opposite trend occurs for uronic acids, which are the oxidized forms of simple reducing sugars and can be formed enzymatically or non-enzymatically. This incongruency can be resolved by noting the overall trend to decreased reliance on oxidative metabolism (and enzymes that catalyze oxidative processes) and increased reliance on sugar metabolism. This trend stems from sugars and sugar phosphates, antioxidants such as ascorbate and glutathione, and $\alpha$-KG (which has been shown to inhibit the electron transport chain in C. elegans). Due to drastic fluctuation in oxygen level in the subterranean niche, cavefish may rely on a shift from oxidative to predominantly sugar-derived energy metabolism, as compared to their surface-dwelling cousins. Reduction in uronic acids, which are derived from sugars using oxidative processes, can thus be seen as part of this trend. However, the specific reduction of uronic acids in particular may have an additional survival benefit for cavefish by inhibiting protein glycation and thus preventing accumulation of advanced glycation end products. Further investigation is required to fully understand the evolutionary and physiological implications of these metabolic changes.

Further work is required to establish the extent of hypoxic conditions in A. mexicanus evolution. However, we also find that certain redox-related metabolites, including $\alpha$-KG, glutathione, and ascorbate, all exhibit distinctive abundance patterns in cavefish. These patterns may be in response to hypoxia, poor nutrient conditions, differences in metabolic rate, or some other aspect of the cave niche.

Our data indicate that upregulation of glucose and long-chain fatty acid production is a common feature shared by Pachón and Tinaja cave populations, suggesting that certain cave habitats do require considerable changes in energy metabolism. Pachón and Tinaja likely have a greater reliance on fat stores for locomotion, as evidenced by increased SFA content in muscle in fasting and refeeding. The decrease of $\omega \text{-}3$ FAs during fasting (Supplementary file 1) coupled with the increase of palmitate (Supplementary file 1) in long- vs short-term fasting suggests that cavefish metabolism may be preferentially biased toward storing caloric intake as energy-rich saturated fatty acids. Whether cavefish possess adaptations to counteract the deleterious effects of high body fat, such as suppression of orotic acid (a metabolite implicated in steatosis in the liver), requires further investigation.

In summary, A. mexicanus troglomorphic populations share many metabolic similarities with known pathological conditions such as obesity, but also display important differences which may help to explain their resistance to negative effects of these pathologies. We also found important differences in ascorbate, which is known to serve diverse physiological roles, nicotinamide, which is a precursor to $NA{D}^{+}$ synthesis and hence is related to oxidative metabolism, $\alpha$-ketoglutarate, which has been implicated in longevity in C. elegans (Chin et al., 2014). We have shown via ROS staining that Pachòn cavefish do possess lower levels of superoxide radicals in the liver after 30 days of fasting. This confirms that increased antioxide levels in cavefish do indeed correlate with physiological state, and suggests that selection in cave environments favors resistance to oxidative stress. Whether selective pressure is driven chiefly by starvation resistance, or whether it is a mechanism to offset oxidized LDL and other harmful metabolites is an open question.

A. mexicanus cave populations clearly exhibit altered lipoprotein composition and expression of CETP, an important component of lipid homeostasis. Together with increased abundance of antioxidants, this may contribute to the ability of cavefish to withstand high triglyceride levels.

Finally, in order to make our data and methods maximally available to other researchers, we have implemented a transparent pipeline that can be used to regenerate all main figures and tables presented here. We have also used the structured data library ObjTables (Karr et al., 2020) to provide machine-readable, semantically accurate representations of the results presented here.

This study provides a large, untargeted metabolomics and lipidomics study of A. mexicanus surface and cave morphs. However, there are limitations to our approach. Our analysis was based on juvenile, pre-sexually mature fish. We chose this developmental time point under the hypothesis that evolutionary changes in metabolism would tend to act early to respond to the selective pressure of the cave environment and positive juvenile fish for robust starvation resistance. However, an age of 5 months does not necessarily coincide with seasonally-correlated food shortages, and hence manifestations of starvation adaptations may not occur until a later developmental stage. We also did not examine a fasting duration of greater than 1 month, despite the potential for longer periods of nutrient deprivation in cave environments.

Source data and a complete pipeline are linked in the submission: http://www.stowers.org/research/publications/libpb-1699 (ftp://odr.stowers.org/LIBPB-1699), and has also been deposited to the github repository https://github.com/stowersinstitute/libpb-1699-metabolomics (copy archived at swh:1:rev:f4ac2098697e1d3e263dcaa72cc2761beb08bdba). An interactive Shiny application (https://cavefin.shinyapps.io/shiny/) can be used to explore and visualize our dataset.

1. 1. Kyle Medley J
   2. Persons J
   3. Biswas T
   4. Olsen L
   5. Peuß R
   6. Krishnan J
   7. Xiong S
   8. Rohner N

   (2022) Stowers Institute

   ID LIBPB-1699. The Metabolome of Mexican Cavefish Shows a Convergent Signature Highlighting Sugar, Antioxidant, and Ageing-Related Metabolites.

   https://www.stowers.org/research/publications/libpb-1699

1. 1. Asadi Shahmirzadi A
   2. Edgar D
   3. Liao C-Y
   4. Hsu Y-M
   5. Lucanic M
   6. Asadi Shahmirzadi A
   7. Wiley CD
   8. Gan G
   9. Kim DE
   10. Kasler HG
   11. Kuehnemann C
   12. Kaplowitz B
   13. Bhaumik D
   14. Riley RR
   15. Kennedy BK
   16. Lithgow GJ

   (2020) Alpha-Ketoglutarate, an Endogenous Metabolite, Extends Lifespan and Compresses Morbidity in Aging Mice

   Cell Metabolism 32:447–456.

   https://doi.org/10.1016/j.cmet.2020.08.004

   • PubMed
   • Google Scholar
2. 1. Aspiras AC
   2. Rohner N
   3. Martineau B
   4. Borowsky RL
   5. Tabin CJ

   (2015) Melanocortin 4 receptor mutations contribute to the adaptation of cavefish to nutrient-poor conditions

   PNAS 112:9668–9673.

   https://doi.org/10.1073/pnas.1510802112

   • PubMed
   • Google Scholar
3. 1. Barter PJ
   2. Rye K-A

   (2012) Cholesteryl ester transfer protein inhibition as a strategy to reduce cardiovascular risk

   Journal of Lipid Research 53:1755–1766.

   https://doi.org/10.1194/jlr.R024075

   • PubMed
   • Google Scholar
4. 1. Barzilai N
   2. Atzmon G
   3. Schechter C
   4. Schaefer EJ
   5. Cupples AL
   6. Lipton R
   7. Cheng S
   8. Shuldiner AR

   (2003) Unique lipoprotein phenotype and genotype associated with exceptional longevity

   JAMA 290:2030–2040.

   https://doi.org/10.1001/jama.290.15.2030

   • PubMed
   • Google Scholar
5. 1. Brousseau ME
   2. Schaefer EJ
   3. Wolfe ML
   4. Bloedon LT
   5. Digenio AG
   6. Clark RW
   7. Mancuso JP
   8. Rader DJ

   (2004) Effects of an inhibitor of cholesteryl ester transfer protein on HDL cholesterol

   The New England Journal of Medicine 350:1505–1515.

   https://doi.org/10.1056/NEJMoa031766

   • PubMed
   • Google Scholar
6. 1. Cajka T
   2. Fiehn O

   (2014) Comprehensive analysis of lipids in biological systems by liquid chromatography-mass spectrometry

   Trends in Analytical Chemistry 61:192–206.

   https://doi.org/10.1016/j.trac.2014.04.017

   • PubMed
   • Google Scholar
7. 1. Chin RM
   2. Fu X
   3. Pai MY
   4. Vergnes L
   5. Hwang H
   6. Deng G
   7. Diep S
   8. Lomenick B
   9. Meli VS
   10. Monsalve GC
   11. Hu E
   12. Whelan SA
   13. Wang JX
   14. Jung G
   15. Solis GM
   16. Fazlollahi F
   17. Kaweeteerawat C
   18. Quach A
   19. Nili M
   20. Krall AS
   21. Godwin HA
   22. Chang HR
   23. Faull KF
   24. Guo F
   25. Jiang M
   26. Trauger SA
   27. Saghatelian A
   28. Braas D
   29. Christofk HR
   30. Clarke CF
   31. Teitell MA
   32. Petrascheck M
   33. Reue K
   34. Jung ME
   35. Frand AR
   36. Huang J

   (2014) The metabolite α-ketoglutarate extends lifespan by inhibiting ATP synthase and TOR

   Nature 510:397–401.

   https://doi.org/10.1038/nature13264

   • PubMed
   • Google Scholar
8. 1. Ching B
   2. Chew SF
   3. Ip YK

   (2015) Ascorbate synthesis in fishes: A review

   IUBMB Life 67:69–76.

   https://doi.org/10.1002/iub.1360

   • PubMed
   • Google Scholar
9. 1. Christison JK
   2. Rye K-A
   3. Stocker R

   (1995) Exchange of oxidized cholesteryl linoleate between LDL and HDL mediated by cholesteryl ester transfer protein

   Journal of Lipid Research 36:2017–2026.

   https://doi.org/10.1016/S0022-2275(20)41119-8

   • PubMed
   • Google Scholar
10. 1. Cirulli ET
    2. Guo L
    3. Leon Swisher C
    4. Shah N
    5. Huang L
    6. Napier LA
    7. Kirkness EF
    8. Spector TD
    9. Caskey CT
    10. Thorens B
    11. Venter JC
    12. Telenti A

    (2019) Profound Perturbation of the Metabolome in Obesity Is Associated with Health Risk

    Cell Metabolism 29:488–500.

    https://doi.org/10.1016/j.cmet.2018.09.022

    • PubMed
    • Google Scholar
11. Book

    1. Cox RA
    2. García-Palmieri MR

    (2011)

    Clinical Methods: The History, Physical, and Laboratory Examinations

    Boston: Butterworth Publishers.

    • Google Scholar
12. 1. DeFelice BC
    2. Mehta SS
    3. Samra S
    4. Čajka T
    5. Wancewicz B
    6. Fahrmann JF
    7. Fiehn O

    (2017) Mass Spectral Feature List Optimizer (MS-FLO): A Tool To Minimize False Positive Peak Reports in Untargeted Liquid Chromatography-Mass Spectroscopy (LC-MS) Data Processing

    Analytical Chemistry 89:3250–3255.

    https://doi.org/10.1021/acs.analchem.6b04372

    • PubMed
    • Google Scholar
13. 1. Fang L
    2. Liu C
    3. Miller YI

    (2014) Zebrafish models of dyslipidemia: relevance to atherosclerosis and angiogenesis

    Translational Research 163:99–108.

    https://doi.org/10.1016/j.trsl.2013.09.004

    • PubMed
    • Google Scholar
14. 1. Fiehn O
    2. Wohlgemuth G
    3. Scholz M
    4. Kind T
    5. Lee DY
    6. Lu Y
    7. Moon S
    8. Nikolau B

    (2008) Quality control for plant metabolomics: reporting MSI-compliant studies

    The Plant Journal 53:691–704.

    https://doi.org/10.1111/j.1365-313X.2007.03387.x

    • PubMed
    • Google Scholar
15. 1. Fiehn O

    (2016) Metabolomics by Gas Chromatography-Mass Spectrometry: Combined Targeted and Untargeted Profiling

    Current Protocols in Molecular Biology 114:30.

    https://doi.org/10.1002/0471142727.mb3004s114

    • PubMed
    • Google Scholar
16. 1. Ford ES
    2. Liu S
    3. Mannino DM
    4. Giles WH
    5. Smith SJ

    (2003) C-reactive protein concentration and concentrations of blood vitamins, carotenoids, and selenium among United States adults

    European Journal of Clinical Nutrition 57:1157–1163.

    https://doi.org/10.1038/sj.ejcn.1601667

    • PubMed
    • Google Scholar
17. 1. Franke SIR
    2. Müller LL
    3. Santos MC
    4. Fishborn A
    5. Hermes L
    6. Molz P
    7. Pereira CS
    8. Wichmann FMA
    9. Horta JA
    10. Maluf SW
    11. Prá D

    (2013) Vitamin C intake reduces the cytotoxicity associated with hyperglycemia in prediabetes and type 2 diabetes

    BioMed Research International 2013:896536.

    https://doi.org/10.1155/2013/896536

    • PubMed
    • Google Scholar
18. Book

    1. Freeman MW
    2. Walford GA

    (2016) Chapter 41 - lipoprotein metabolism and the treatment of lipid disorders

    In: Jameson JL, Groot LD, Giudice LC, Grossman AB, Melmed S, Potts JT, Weir GC, editors. Endocrinology: Adult and Pediatric. Philadelphia: WB Saunders. pp. 715–736.

    https://doi.org/10.1016/B978-0-323-18907-1.00041-X

    • Google Scholar
19. Book

    1. Furne M
    2. Sanz A

    (2017)

    Starvation in fish – sturgeon and rainbow trout as examples

    In: Preedy VR, Patel VB, editors. Handbook of Famine, Starvation, and Nutrient Deprivation. Berlin, Germany: Springer. pp. 401–411.

    • Google Scholar
20. 1. Gelman A
    2. Jakulin A
    3. Pittau MG
    4. Su Y-S

    (2008) A weakly informative default prior distribution for logistic and other regression models

    The Annals of Applied Statistics 2:1360–1383.

    https://doi.org/10.1214/08-AOAS191

    • Google Scholar
21. 1. Ghosh S
    2. Zhao B
    3. Bie J
    4. Song J

    (2010) Macrophage cholesteryl ester mobilization and atherosclerosis

    Vascular Pharmacology 52:1–10.

    https://doi.org/10.1016/j.vph.2009.10.002

    • PubMed
    • Google Scholar
22. 1. Glass CK
    2. Olefsky JM

    (2012) Inflammation and lipid signaling in the etiology of insulin resistance

    Cell Metabolism 15:635–645.

    https://doi.org/10.1016/j.cmet.2012.04.001

    • PubMed
    • Google Scholar
23. 1. Han H-S
    2. Kang G
    3. Kim JS
    4. Choi BH
    5. Koo S-H

    (2016) Regulation of glucose metabolism from a liver-centric perspective

    Experimental & Molecular Medicine 48:e218.

    https://doi.org/10.1038/emm.2015.122

    • PubMed
    • Google Scholar
24. 1. Huang S
    2. Czech MP

    (2007) The GLUT4 glucose transporter

    Cell Metabolism 5:237–252.

    https://doi.org/10.1016/j.cmet.2007.03.006

    • PubMed
    • Google Scholar
25. 1. Huang D
    2. Stein N
    3. Rubin DB
    4. Kou SC

    (2020) Catalytic prior distributions with application to generalized linear models

    PNAS 117:12004–12010.

    https://doi.org/10.1073/pnas.1920913117

    • Google Scholar
26. 1. Hüppop K

    (1986) Oxygen consumption of Astyanax fasciatus (Characidae, Pisces): a comparison of epigean and hypogean populations

    Environmental Biology of Fishes 17:299–308.

    https://doi.org/10.1007/BF00001496

    • Google Scholar
27. 1. Imam TS
    2. Bala U
    3. Balarabe ML
    4. Oyeyi TI

    (2010)

    Length-weight relationship and condition factor of four fish species from Wasai reservoir in Kano, Nigeria

    African Journal of General Agriculture 6:125–130.

    • Google Scholar
28. 1. Jaldin-Fincati JR
    2. Pavarotti M
    3. Frendo-Cumbo S
    4. Bilan PJ
    5. Klip A

    (2017) Update on GLUT4 Vesicle Traffic: A Cornerstone of Insulin Action

    Trends in Endocrinology and Metabolism 28:597–611.

    https://doi.org/10.1016/j.tem.2017.05.002

    • PubMed
    • Google Scholar
29. 1. Jeffery WR

    (2020) Astyanax surface and cave fish morphs

    EvoDevo 11:1–10.

    https://doi.org/10.1186/s13227-020-00159-6

    • Google Scholar
30. 1. Kanehisa M
    2. Goto S

    (2000) KEGG: kyoto encyclopedia of genes and genomes

    Nucleic Acids Research 28:27–30.

    https://doi.org/10.1093/nar/28.1.27

    • PubMed
    • Google Scholar
31. 1. Karpe F
    2. Dickmann JR
    3. Frayn KN

    (2011) Fatty acids, obesity, and insulin resistance: time for a reevaluation

    Diabetes 60:2441–2449.

    https://doi.org/10.2337/db11-0425

    • PubMed
    • Google Scholar
32. Preprint

    1. Karr JR
    2. Liebermeister W
    3. Goldberg AP
    4. Sekar JA
    5. Shaikh B

    (2020) Structured Spreadsheets with Objtables Enable Data Reuse and Integration

    arXiv.

    https://arxiv.org/abs/2005.05227

    • Google Scholar
33. 1. Kind T
    2. Liu K-H
    3. Lee DY
    4. DeFelice B
    5. Meissen JK
    6. Fiehn O

    (2013) LipidBlast in silico tandem mass spectrometry database for lipid identification

    Nature Methods 10:755–758.

    https://doi.org/10.1038/nmeth.2551

    • PubMed
    • Google Scholar
34. 1. Krishnan J
    2. Rohner N

    (2019) Sweet fish: Fish models for the study of hyperglycemia and diabetes

    Journal of Diabetes 11:193–203.

    https://doi.org/10.1111/1753-0407.12860

    • PubMed
    • Google Scholar
35. 1. Krishnan J
    2. Persons JL
    3. Peuß R
    4. Hassan H
    5. Kenzior A
    6. Xiong S
    7. Olsen L
    8. Maldonado E
    9. Kowalko JE
    10. Rohner N

    (2020) Comparative transcriptome analysis of wild and lab populations of Astyanax mexicanus uncovers differential effects of environment and morphotype on gene expression

    Journal of Experimental Zoology. Part B, Molecular and Developmental Evolution 334:530–539.

    https://doi.org/10.1002/jez.b.22933

    • PubMed
    • Google Scholar
36. 1. Krishnan J
    2. Seidel CW
    3. Zhang N
    4. Singh NP
    5. VanCampen J
    6. Peuß R
    7. Xiong S
    8. Kenzior A
    9. Li H
    10. Conaway JW
    11. Rohner N

    (2022) Genome-wide analysis of cis-regulatory changes underlying metabolic adaptation of cavefish

    Nature Genetics 54:684–693.

    https://doi.org/10.1038/s41588-022-01049-4

    • PubMed
    • Google Scholar
37. 1. Longo N
    2. Wang Y
    3. Pasquali M

    (1999) Progressive decline in insulin levels in Rabson-Mendenhall syndrome

    The Journal of Clinical Endocrinology and Metabolism 84:2623–2629.

    https://doi.org/10.1210/jcem.84.8.5902

    • PubMed
    • Google Scholar
38. 1. Ma S
    2. Yim SH
    3. Lee S-G
    4. Kim EB
    5. Lee S-R
    6. Chang K-T
    7. Buffenstein R
    8. Lewis KN
    9. Park TJ
    10. Miller RA
    11. Clish CB
    12. Gladyshev VN

    (2015) Organization of the Mammalian Metabolome according to Organ Function, Lineage Specialization, and Longevity

    Cell Metabolism 22:332–343.

    https://doi.org/10.1016/j.cmet.2015.07.005

    • PubMed
    • Google Scholar
39. 1. Matyash V
    2. Liebisch G
    3. Kurzchalia TV
    4. Shevchenko A
    5. Schwudke D

    (2008) Lipid extraction by methyl-tert-butyl ether for high-throughput lipidomics

    Journal of Lipid Research 49:1137–1146.

    https://doi.org/10.1194/jlr.D700041-JLR200

    • PubMed
    • Google Scholar
40. 1. Morigny P
    2. Houssier M
    3. Mairal A
    4. Ghilain C
    5. Mouisel E
    6. Benhamed F
    7. Masri B
    8. Recazens E
    9. Denechaud P-D
    10. Tavernier G
    11. Caspar-Bauguil S
    12. Virtue S
    13. Sramkova V
    14. Monbrun L
    15. Mazars A
    16. Zanoun M
    17. Guilmeau S
    18. Barquissau V
    19. Beuzelin D
    20. Bonnel S
    21. Marques M
    22. Monge-Roffarello B
    23. Lefort C
    24. Fielding B
    25. Sulpice T
    26. Astrup A
    27. Payrastre B
    28. Bertrand-Michel J
    29. Meugnier E
    30. Ligat L
    31. Lopez F
    32. Guillou H
    33. Ling C
    34. Holm C
    35. Rabasa-Lhoret R
    36. Saris WHM
    37. Stich V
    38. Arner P
    39. Rydén M
    40. Moro C
    41. Viguerie N
    42. Harms M
    43. Hallén S
    44. Vidal-Puig A
    45. Vidal H
    46. Postic C
    47. Langin D

    (2019) Interaction between hormone-sensitive lipase and ChREBP in fat cells controls insulin sensitivity

    Nature Metabolism 1:133–146.

    https://doi.org/10.1038/s42255-018-0007-6

    • PubMed
    • Google Scholar
41. 1. Oh DY
    2. Talukdar S
    3. Bae EJ
    4. Imamura T
    5. Morinaga H
    6. Fan W
    7. Li P
    8. Lu WJ
    9. Watkins SM
    10. Olefsky JM

    (2010) GPR120 is an omega-3 fatty acid receptor mediating potent anti-inflammatory and insulin-sensitizing effects

    Cell 142:687–698.

    https://doi.org/10.1016/j.cell.2010.07.041

    • PubMed
    • Google Scholar
42. 1. Paschoal VA
    2. Walenta E
    3. Talukdar S
    4. Pessentheiner AR
    5. Osborn O
    6. Hah N
    7. Chi TJ
    8. Tye GL
    9. Armando AM
    10. Evans RM
    11. Chi NW
    12. Quehenberger O
    13. Olefsky JM
    14. Oh DY

    (2020) Positive Reinforcing Mechanisms between GPR120 and PPARγ Modulate Insulin Sensitivity

    Cell Metabolism 31:1173–1188.

    https://doi.org/10.1016/j.cmet.2020.04.020

    • PubMed
    • Google Scholar
43. 1. Peuß R
    2. Box AC
    3. Chen S
    4. Wang Y
    5. Tsuchiya D
    6. Persons JL
    7. Kenzior A
    8. Maldonado E
    9. Krishnan J
    10. Scharsack JP
    11. Slaughter BD
    12. Rohner N

    (2020) Adaptation to low parasite abundance affects immune investment and immunopathological responses of cavefish

    Nature Ecology & Evolution 4:1416–1430.

    https://doi.org/10.1038/s41559-020-1234-2

    • PubMed
    • Google Scholar
44. 1. Pompella A
    2. Visvikis A
    3. Paolicchi A
    4. De Tata V
    5. Casini AF

    (2003) The changing faces of glutathione, a cellular protagonist

    Biochemical Pharmacology 66:1499–1503.

    https://doi.org/10.1016/s0006-2952(03)00504-5

    • PubMed
    • Google Scholar
45. 1. Riddle MR
    2. Aspiras AC
    3. Gaudenz K
    4. Peuß R
    5. Sung JY
    6. Martineau B
    7. Peavey M
    8. Box AC
    9. Tabin JA
    10. McGaugh S
    11. Borowsky R
    12. Tabin CJ
    13. Rohner N

    (2018) Insulin resistance in cavefish as an adaptation to a nutrient-limited environment

    Nature 555:647–651.

    https://doi.org/10.1038/nature26136

    • PubMed
    • Google Scholar
46. 1. Rubino F
    2. Nathan DM
    3. Eckel RH
    4. Schauer PR
    5. Alberti KGMM
    6. Zimmet PZ
    7. Del Prato S
    8. Ji L
    9. Sadikot SM
    10. Herman WH
    11. Amiel SA
    12. Kaplan LM
    13. Taroncher-Oldenburg G
    14. Cummings DE
    15. Delegates of the 2nd Diabetes Surgery Summit

    (2016) Metabolic Surgery in the Treatment Algorithm for Type 2 Diabetes: A Joint Statement by International Diabetes Organizations

    Surgery for Obesity and Related Diseases 12:1144–1162.

    https://doi.org/10.1016/j.soard.2016.05.018

    • PubMed
    • Google Scholar
47. 1. Skogerson K
    2. Wohlgemuth G
    3. Barupal DK
    4. Fiehn O

    (2011) The volatile compound BinBase mass spectral database

    BMC Bioinformatics 12:321.

    https://doi.org/10.1186/1471-2105-12-321

    • PubMed
    • Google Scholar
48. 1. Standerfer SB
    2. Handler P

    (1955) Fatty liver induced by orotic acid feeding

    Proceedings of the Society for Experimental Biology and Medicine. Society for Experimental Biology and Medicine (New York, N.Y.) 90:270–271.

    https://doi.org/10.3181/00379727-90-22005

    • PubMed
    • Google Scholar
49. 1. Trygg J
    2. Wold S

    (2002) Orthogonal projections to latent structures (O-PLS

    Journal of Chemometrics 16:119–128.

    https://doi.org/10.1002/cem.695

    • Google Scholar
50. 1. Tsugawa H
    2. Cajka T
    3. Kind T
    4. Ma Y
    5. Higgins B
    6. Ikeda K
    7. Kanazawa M
    8. VanderGheynst J
    9. Fiehn O
    10. Arita M

    (2015) MS-DIAL: data-independent MS/MS deconvolution for comprehensive metabolome analysis

    Nature Methods 12:523–526.

    https://doi.org/10.1038/nmeth.3393

    • PubMed
    • Google Scholar
51. 1. Vaccaro A
    2. Kaplan Dor Y
    3. Nambara K
    4. Pollina EA
    5. Lin C
    6. Greenberg ME
    7. Rogulja D

    (2020) Sleep Loss Can Cause Death through Accumulation of Reactive Oxygen Species in the Gut

    Cell 181:1307–1328.

    https://doi.org/10.1016/j.cell.2020.04.049

    • PubMed
    • Google Scholar
52. 1. Westerhuis JA
    2. van Velzen EJJ
    3. Hoefsloot HCJ
    4. Smilde AK

    (2008) Discriminant Q2 (DQ2) for improved discrimination in PLSDA models

    Metabolomics 4:293–296.

    https://doi.org/10.1007/s11306-008-0126-2

    • Google Scholar
53. 1. Wilson JX

    (2002) The physiological role of dehydroascorbic acid

    FEBS Letters 527:5–9.

    https://doi.org/10.1016/s0014-5793(02)03167-8

    • PubMed
    • Google Scholar
54. 1. Wishart DS
    2. Feunang YD
    3. Marcu A
    4. Guo AC
    5. Liang K
    6. Vázquez-Fresno R
    7. Sajed T
    8. Johnson D
    9. Li C
    10. Karu N
    11. Sayeeda Z
    12. Lo E
    13. Assempour N
    14. Berjanskii M
    15. Singhal S
    16. Arndt D
    17. Liang Y
    18. Badran H
    19. Grant J
    20. Serra-Cayuela A
    21. Liu Y
    22. Mandal R
    23. Neveu V
    24. Pon A
    25. Knox C
    26. Wilson M
    27. Manach C
    28. Scalbert A

    (2018) HMDB 4.0: the human metabolome database for 2018

    Nucleic Acids Research 46:D608–D617.

    https://doi.org/10.1093/nar/gkx1089

    • PubMed
    • Google Scholar
55. 1. Xiong S
    2. Krishnan J
    3. Peuß R
    4. Rohner N

    (2018) Early adipogenesis contributes to excess fat accumulation in cave populations of Astyanax mexicanus

    Developmental Biology 441:297–304.

    https://doi.org/10.1016/j.ydbio.2018.06.003

    • PubMed
    • Google Scholar
56. 1. Yu X-H
    2. Fu Y-C
    3. Zhang D-W
    4. Yin K
    5. Tang C-K

    (2013) Foam cells in atherosclerosis

    Clinica Chimica Acta; International Journal of Clinical Chemistry 424:245–252.

    https://doi.org/10.1016/j.cca.2013.06.006

    • PubMed
    • Google Scholar

Author details

1. J Kyle Medley

   Stowers Institute for Medical Research, Kansas City, United States

   Contribution

   Conceptualization, Data curation, Formal analysis, Resources, Software, Validation, Visualization, Writing – original draft, Writing – review and editing

   Contributed equally with

   Jenna Persons

   For correspondence

   medjk@comcast.net

   Competing interests

   No competing interests declared

2. Jenna Persons

   Stowers Institute for Medical Research, Kansas City, United States

   Present address

   Newman University, Wichita, United States

   Contribution

   Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Writing – review and editing

   Contributed equally with

   J Kyle Medley

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-8807-9678

3. Tathagata Biswas

   Stowers Institute for Medical Research, Kansas City, United States

   Contribution

   Data curation, Resources

   Competing interests

   No competing interests declared

4. Luke Olsen

   1. Stowers Institute for Medical Research, Kansas City, United States
   2. Department of Molecular and Integrative Physiology, University of Kansas Medical Center, Kansas City, United States

   Contribution

   Methodology, Resources, Validation

   Competing interests

   No competing interests declared

5. Robert Peuß

   Department of Molecular and Integrative Physiology, University of Kansas Medical Center, Kansas City, United States

   Present address

   University of Münster, Münster, Germany

   Contribution

   Resources, Supervision

   Competing interests

   No competing interests declared

6. Jaya Krishnan

   Stowers Institute for Medical Research, Kansas City, United States

   Contribution

   Resources, Supervision, Validation

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-2302-5748

7. Shaolei Xiong

   Stowers Institute for Medical Research, Kansas City, United States

   Contribution

   Data curation, Investigation

   Competing interests

   No competing interests declared

8. Nicolas Rohner

   1. Stowers Institute for Medical Research, Kansas City, United States
   2. Department of Molecular and Integrative Physiology, University of Kansas Medical Center, Kansas City, United States

   Contribution

   Conceptualization, Funding acquisition, Project administration, Supervision, Visualization, Writing – review and editing

   For correspondence

   nro@stowers.org

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-3248-2772

• 1,674

  Page views

• 247

  Downloads

• 12

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.