Traumatic brain injury changes fat storage in the fly brain depending on age and diet

To investigate LD dynamics following TBI, we used confocal microscopy of dissected fly brains to compare the size and number of LDs between injured and uninjured flies. These experiments used a standard laboratory fly strain, w¹¹¹⁸. Female flies were injured at 8 days old using a HIT device [47], and age-matched, uninjured female flies served as controls. Bodipy (boron-dipyrromethene), a fluorescent dye that has high specificity for neutral lipids, was used to stain LDs [56]; an antibody to Repo (Reversed polarity) marked glia [57]; and the fluorescent DNA stain DAPI (4',6-diamidino-2-phenylindole) labeled nuclei, including those of glia and neurons (Fig 1A). For each brain, images were obtained in four regions within the central brain at a plane with the highest concentration of nuclei (Fig 1B).

At 1 day post-injury, the size and number of LDs appeared to increase substantially (Fig 2). Near identical results were obtained with Nile Red, a lipophilic fluorescent dye that partitions into neutral lipid compartments [58], thereby independently confirming the identity of LDs (S1 Fig.). Quantitation of at least 11 brains and 21,000 cells per brain for each condition showed that the area of individual LDs was larger in injured compared to uninjured flies (Fig 3A, Table 1A, S1 Data). Assuming that LDs are spherical, injury increased their average diameter by 20%, from 0.95 μm in uninjured flies to 1.14 μm in injured flies. Correspondingly, injury increased the average volume of LDs by 73%, from 0.45 μm³ to 0.78 μm³. Injury also significantly increased the number of LDs per cell as well as the total area of LDs per cell (i.e., area of individual LDs X number of LDs per cell) (Figs 3B and 3C, Tables 1B and 1C, S1 Data). These data suggest that TBI impairs lipid breakdown and/or enhances lipid synthesis, resulting in larger LDs, and that it inhibits LD degradation and/or promotes LD formation, increasing their number. Additionally, the observed increases in size and number may involve altered fission and fusion dynamics of LDs [31]. Overall, these changes may reflect a coordinated metabolic response to cellular stress that promotes sequestering excess or peroxidized lipids, thereby reducing lipotoxicity.

LDs are dynamic organelles that interact with various cellular structures. In particular, LDs associate with microtubules to facilitate their directed movement and spatial distribution within the cytoplasm [59]. This microtubule-dependent trafficking is important for positioning LDs in proximity to organelles such as mitochondria. LD-mitochondria contacts are functionally significant, as mitochondria are a major source of ROS following TBI [5,60]. To determine the localization of LDs in relation to microtubules and mitochondria, we again used confocal microscopy. Microtubules were labeled using an antibody to α-tubulin, and mitochondria producing ROS were identified using a ubiquitously expressed hydrogen peroxide (H₂O₂) sensor targeted to the mitochondrial matrix (tub-mito-roGFP2-Orp1) [61]. In both injured and uninjured flies, LDs were found to co-localize with microtubules in what appear to be axon bundles, supporting a role for microtubule-based trafficking in LD distribution (Figs 4A and 4B). Additionally, in injured flies, LDs were frequently positioned adjacent to ROS-producing mitochondria, suggesting that LD-mitochondrial interactions are functionally relevant in the oxidative environment post-TBI (Figs 4C and 4D).

The observed increase in LDs following TBI suggests that triacylglycerol (TG) metabolism is altered in response to injury. To investigate this, we performed high-throughput, semi-quantitative lipidomic analysis to assess changes in lipid species between injured and uninjured flies in an unbiased manner. Eight-day old w¹¹¹⁸ male flies were subjected to injury using the HIT device, fed CMYD, and analyzed 1 day post-injury. Age-matched, uninjured flies served as controls. For each condition, fly heads (n = 150) were collected in triplicate and analyzed using liquid chromatography-tandem mass spectrometry (LC/MS/MS) in both positive and negative ion modes. Lipidomic datasets acquired in each mode were normalized and combined to generate a comprehensive profile for downstream analysis (S2 Data).

To identify individual lipid species associated with TBI, we applied Partial Least Squares Discriminant Analysis (PLS-DA) to the lipidomic dataset, which consisted of 908 lipid species, including 258 TG species. This multivariate statistical approach allowed visualization of overall differences between injured and uninjured samples and the identification of lipid species that most strongly contributed to group separation. The resulting PLS-DA scores plot revealed clear clustering of samples based on injury status, indicating distinct lipidomic signatures between conditions (Fig 5A). The top 20 differentiating lipid species were all TGs (Fig 5B). The probability of this occurring by chance is exceedingly small (6.8 x 10^-12), suggesting an important role for the TGs in the response to TBI. Furthermore, among these TGs, 97% of the fatty acid side chains had 12, 14, 16, or 18 carbon atoms, which are major energy substrates [62,63]. Seven of the lipid species were more abundant in injured than uninjured flies and the remaining 13 were more abundant in uninjured than injured flies, which may explain why the abundance of the TG class did not differ between injured and uninjured flies (S2A Fig., column 1). All seven TGs with increased levels post-injury contained at least two fatty acid side chains of 16:0, 16:1, 18:0, 18:1, or 18:2 (number of carbon atoms:number of double bonds) – the most abundant TG-associated fatty acids in LDs [64,65] – and four of these TGs had three side chains from this group. In contrast, only four of the 13 TGs that decreased post-injury contained two of these side chains, and none contained three. These findings suggest that TBI acutely promotes the synthesis of TG species that are preferentially incorporated into LDs.

To assess global changes in lipid profiles following TBI, we compared the abundance of the 24 identified classes of lipid species between injured and uninjured flies (., column 1). These data indicate that injury caused the greatest fold increase in lysophospholipids (LPs) and fatty acid esters of hydroxy fatty acids (FAHFA). Acylcarnitines (CARs) had the greatest fold decrease in abundance post-TBI. Thus, TBI rapidly and broadly alters lipid metabolism. S2A Fig

To investigate the long-term effects of TBI on LD dynamics, we repeated the microscopy assay at 7, 14, 21, and 35 days post-injury. Throughout this four week period, the area of individual LDs remained largely unchanged between injured and uninjured flies (Fig 3A, S1 Data). The dramatic decrease in LD size between 1 and 7 days in injured flies may result from TG breakdown through lipolysis or from LD fission. Lipolysis is more likely since the number of LDs per cell in injured flies also returned to the level of uninjured flies by 7 days post-injury (Fig 3B, S1 Data). Furthermore, while the substantial increase in LD size at 21 days in uninjured flies was also observed in injured flies (Fig 3A), the increase in LD number at 21 days was not (Fig 3B). However, by 28 and 35 days, the number of LDs in injured flies had risen to match that of uninjured flies (Fig 3B). Finally, total LD area per cell – reflecting both LD size and number – was substantially lower at 21 days post-injury and slightly lower at 35 days post-injury (Fig 3C, S1 Data). These findings suggest that over time, TBI selectively disrupts the formation of new LDs associated with aging, without altering the processes that regulate LD growth.

We used lipidomic analysis to investigate the long-term effects of TBI on lipid dynamics. The experiment performed at 1 day after injury (Figs 5A and 5B) was repeated at 7, 14, and 21 days post-injury (Figs 5C and 5D). Across all samples, 1401 lipid species were detected, including 365 TG species and 155 phosphatidylethanolamine (PE) species (S3 Data), some of which are components of LD membranes and contribute to membrane curvature and dynamics during LD formation and growth [66].

PLS-DA was used to identify lipids that differentiate injured and uninjured flies at 7, 14, and 21 days post-injury (Figs 5C and 5D). Component 1 of the scores plot organized the samples chronologically, suggesting that it reflects age-related lipid changes. Among the top 20 differentiating lipid species, 12 were TGs and 8 were PEs. The TGs were enriched in fatty acid side chains typically associated with LDs. In nearly all cases, these lipids decreased in abundance with age and at each time point were more abundant in injured than uninjured flies. This pattern contrasts with the age-related increase in LD abundance observed in uninjured flies (Fig 3C), suggesting that while TBI increases the abundance of lipid species typically stored in LDs as flies age, it also impairs LD formation, allowing these lipids to exert toxic effects.

Over time, the levels of select lipid classes, including cardiolipin (CL), lysophosphatidylinositol (LPI), phosphatidylinositol (PI), and phosphatidylserine (PS), steadily increased in injured compared to uninjured flies (., columns 2–4). Analyses of lipid class abundance at 7, 14, and 21 days post-injury relative to 1 day post-injury (and., columns 2–4) indicate that the increase in CL, LPI, PI, and PS was driven by a decline in the level of theses lipids in uninjured flies, while levels in injured flies remained stable. These findings indicate that TBI has long-term effects on lipid metabolism in the head, which likely reflects changes in the brain because it occupies a substantial fraction of the head volume. S2A Fig S2B S2C Figs

We previously found that the type of food consumed within 24 h after TBI influences early mortality, with significantly higher mortality for flies fed high-carbohydrate diets – CMYD or >1 M glucose – compared with those fed low-carbohydrate diets – ketogenic diet (KD) or water [48,55]. To investigate whether diet-related effects on early mortality might involve LD metabolism, we assessed LD size and number at 2 and 24 h following TBI of 8-day old, w¹¹¹⁸, injured and uninjured, female flies fed one of four diets post-injury: CMYD, 2 M glucose, a commercial KD, or water. The 2 h time point was selected because it is the earliest time at which effects of CMYD and water on early mortality have been observed [52].

As expected, diet affected early mortality after TBI: 24% for CMYD, 28% for glucose, 13% for KD, and 11% for water. However, early mortality among uninjured flies was < 1% across all diets, indicating that carbohydrate-rich diets do not directly cause early mortality but instead contribute to TBI-induced mechanisms that lead to early mortality. Accordingly, quantitation of at least 15 brains and 11,000 cells per brain for each condition, revealed that in both injured and uninjured flies, diet significantly affected individual LD size, LD number per cell, and total LD area per cell at 2 and 24 h (Fig 6 and Table 2). These data indicate that diet influences LD abundance through injury-independent and -dependent mechanisms. At 24 h, injury caused the most significant increase in LD area per cell in flies fed a carbohydrate-rich diet – CMYD or glucose (Table 2C) – consistent with the results shown in Figs 2 and 3 and Table 1C. The increase in total LD area per cell was largely driven by an increase in the number of LDs per cell (Table 2B). The underlying mechanism was activated by 2 h for glucose but not CMYD. The delay with CMYD may be due to the additional time required to digest complex carbohydrates to glucose. At 24 h, TBI had little or no effect on LD area per cell with KD and water diets, respectively (Table 2C), despite significant effects of these diets on LD size and number at 2 and 24 h (Tables 2A and 2B). These findings reveal a correlation between TBI-induced LD accumulation and early mortality across the different diets, suggesting that LDs either contribute directly to early mortality or arise as a result of processes that induce early mortality. However, it is important to note that absolute LD abundance per cell did not correlate with early mortality following TBI: flies fed KD or CMYD had similar LD area per cell at 24 h (Table 2C), yet experienced different levels of early mortality. This may reflect the dual functions of LDs: under CMYD, increased LD abundance may result from the sequestration of lipid peroxides, which are associated with early mortality; in contrast, under KD, the increase in LDs may reflect storage of undamaged lipids for β-oxidation, serving as an energy source that helps prevent early mortality [67,68].

The association of LDs with microtubules in axons of injured brains (Fig 4B) provides the first direct evidence, to our knowledge, that disrupted LD trafficking may contribute to the development of CTE pathology. CTE is characterized by microtubule disruption, impaired intracellular transport of cargoes, and abnormal hyperphosphorylation of tau, which destabilizes microtubules and compromises neuronal structural integrity [8]. Our prior studies implicated the microtubule-associated proteins Lissencephaly-1 (Lis-1) and Patronin in TBI outcomes [49]. Lis-1 regulates dynein, the microtubule motor responsible for retrograde transport of many cargoes, including LDs [69], while Patronin stabilizes microtubules by protecting their minus ends from depolymerization [70]. SNPs in the Lis-1 and Patronin genes are associated with early mortality following TBI, and mutations in Lis-1 enhance early mortality and neurodegeneration following TBI [49]. Furthermore, reduced expression of Lis-1 is associated with a late stage of CTE [71]. Together, these findings suggest that microtubule-dependent regulation of LD localization – particularly their trafficking to mitochondria (Fig 4D) – is a key component of the cellular response to TBI and defects in the process may contribute to the development of CTE pathology.

We found that a primary brain injury induces dynamic changes in LD abundance that can manifest weeks after a primary injury. LD size and number increased significantly at 24 h post-injury but returned to baseline within 7 days (Fig 3 and Table 1). This resolution likely reflects recovery from the conditions that initially promoted LD accumulation – such as mitochondrial dysfunction, oxidative stress, and neuroinflammation – and may involve lipid breakdown by lipases or lipophagy, where LDs are engulfed by autophagosomes and degraded by lysosomes [72]. In the weeks following injury, LD number increased, consistent with prior observations in flies and mice [73,74]. However, the typical age-related increase in LD abundance was attenuated in injured flies compared to uninjured controls (Figs 3B and 3C). This difference was most pronounced at 21 days post-injured (i.e., in 29 day old flies), when uninjured flies had significantly more LDs per cell (Fig. 3B) but similar LD size (Fig. 3A), suggesting that TBI impairs the formation of new LDs without affecting the growth of existing ones. This time point aligns with related physiological events in aging flies, suggesting that they may be functionally linked. First, brain lipid turnover peaks at around 25 days of age, as indicated by the ratio of newly synthesized lipids to total lipids during aging from 5 to 45 days [32]. Second, a transient spike in 4-HNE levels is observed in fly heads at approximately 38 days of age [75]. Third, flies injured at 0–7 day old and surviving the first 24 h exhibit normal survival for about 17 days, but begin dying at a higher rate than uninjured flies between 17 and 25 days of age [47,55]. Collectively, these findings suggest that TBI disrupts programmed changes in lipid and LD metabolism during aging, potentially contributing to neurodegeneration and other late-onset consequences of injury. In particular, a primary injury may acutely trigger LD metabolic events that typically occur later in life (i.e., LD size and number in injured flies at 1 day post-injury was comparable to that of uninjured flies at 21 days post-injury) (Figs 3A and 3B and Tables 1A and 1B). This early onset may contribute to accelerated, age-associated neurodegeneration.

Our findings indicate that reducing carbohydrate intake – either by feeding KD or water immediately after TBI – may protect against injury-related damage by limiting production of toxic lipids, as reflected by the absence of new LD formation (Fig 6 and Table 2B). Similarly, carbohydrate-rich Western diets lead to worse TBI outcomes than KD in rodents [76]. Moreover, in a rat TBI model, KD-fed animals showed reduced oxidative stress and lower levels of 4-HNE-modified proteins at 6 and 24 h post-injury compared to those fed a high-carbohydrate, standard chow diet [77]. The protective effects of KD appear to be mediated by ROS scavenging and enhanced mitochondrial electron transport chain activity, suggesting that the absence of TBI-induced LD formation in KD-fed flies results from reduced upstream signals linked to mitochondrial dysfunction and oxidative stress.

Our prior data show that flies given only water post-injury experience substantially lower early mortality than those fed CMYD, with the protective effect most pronounced when water is provided between 2 and 8 h after injury [52]. Similarly, Davis et al. (2008) showed in a rat TBI model that 24 h of water-only fasting after injury compared to unrestricted access to a balanced diet reduced lesion volume, mitochondrial dysfunction, ROS production, and lipid peroxidation and improved cognitive function [78].

A major limitation of this study is that it was conducted in a single genetic background. Our studies in flies, along with studies in rodents and humans, demonstrate that genetic background strongly influences the outcomes of TBI [48,49,52,79,80]. Rodents receiving identical primary injuries can exhibit markedly different outcomes depending on their genotype [81 –84]. This variation underpins our GWAS analyses that identified associations between early mortality following TBI and SNPs in lsd-1, Lis-1, and Patronin [48]. In our experiments using 30 inbred fly lines from the Drosophila Genetic Reference Panel (DGRP) [85,86], early mortality following TBI varied substantially across genotypes [52]. Furthermore, factors such as age at the time of injury and post-injury diet influenced genetically distinct secondary injury pathways. For instance, while most DGRP fly lines showed higher early mortality when injured at an older age and when fed CMYD rather than water, DGRP439 was sensitive to age but not diet – indicating genetic-based resistance to the carbohydrate-driven early mortality pathway. In contrast, DGRP381 was diet-sensitive but not age-sensitive. These findings suggest that lipid and LD metabolism responses following TBI in w¹¹¹⁸ flies are unlikely to generalize across other genetic backgrounds. Understanding how genetics differences affect lipid and LD metabolism in flies may ultimately inform the development of personalized therapies for genetically diverse human populations.

Additionally, because flies that died were not examined, it remains unclear whether LD accumulation is protective or detrimental in the context of early mortality. The elevated LD levels observed in flies that survived one day post-injury is consistent with a protective role, potentially by sequestering toxic lipids and supplying energy to support neuronal recovery. However, it is also possible that excessive LDs accumulation is detrimental, and that survivors had lower LD levels than those that died. To resolve this, it will be necessary to compare animals that will survive to those that are fated to die within one day following TBI. This is feasible in flies, as increased intestinal permeability after TBI reliably marks individuals that will die within one day [48,87]. Therefore, to determine the functional role of LDs, their levels should be measured in flies exhibiting intestinal barrier dysfunction to determine whether LD accumulation correlates positively or negatively with early mortality.

Finally, an important confounding factor in the interpretation of both the confocal microscopy and lipidomic data is that these analyses were performed on bulk tissue – whole brains in the case of microscopy and whole heads for lipidomic analysis. As a result, these data represent an average signal across multiple cell types, potentially obscuring cell type-specific responses to TBI. LDs are known to vary in size, abundance, and function across different brain cell types [37,41,88]. Following TBI, it is possible that LDs increase in size and/or number in some cell populations while decreasing in others or that certain lipid species are differentially regulated depending on the cell type. Such heterogeneity could not only lead to misinterpretation of the directionality or magnitude of the response but also mask meaningful biological effects. Our microscopy data (Figs 1, 2, 4, 6, and S1 Fig.) do not clarify whether TBI-, aging-, and diet-induced LD changes in LD size and number occur in neurons or glia. Demonstrating neuronal localization would be paradigm-shifting, as glia are currently considered the primary site of stress-induced LD accumulation [35,37,38,46,68,73,74]. To resolve these ambiguities, future studies incorporating cell type-specific labeling, sorting, or single-cell approaches will be critical to accurately characterize how TBI alters lipid and LD metabolism. Nonetheless, bulk tissue analyses presented here have been tremendously informative, providing foundational understanding of lipid and LD alterations after TBI and offering critical impetus for pursing more refined, cell type-specific investigations.

Flies were maintained on a standard fly food, cornmeal-molasses-yeast diet (CMYD), at 25°C, unless otherwise stated. CMYD contained 30 g Difco granulated agar (Becton-Dickinson), 44 g YSC-1 yeast (Sigma), 328 g cornmeal (Lab Scientific), 400 ml unsulfured Grandma's molasses (Lab Scientific), 3.6 L water, 40 ml propionic acid (Sigma), and tegosept (8 g Methyl 4-hydroxybenzoate in 75 ml of 95% ethanol) (Sigma). w¹¹¹⁸ flies were maintained in our lab for many years, and w¹¹¹⁸; tub-mito-roGFP2-Orp1 flies [61] were obtained from the Bloomington Drosophila Stock Center (stock #67673).

Flies were collected at 0–1 days post-eclosion, aged 7 days, and subjected in groups of 30 to TBI (4 strikes from a HIT device with 5 min between strikes), as described in Katzenberger et al., 2013 [47]. Uninjured flies were manipulated the same as injured flies, but they were not subjected to strikes from HIT device. Injured and uninjured flies were transferred to vials with CMYD or a Whatman paper disc at the bottom that contained 200 μl of H₂0, 2 M glucose (Sigma), or a commercial ketogenic diet (KD) (Teklad ketogenic diet, TD.96355, Envigo) at 0.3 calories/200 μl. KD contained 173.3 g/kg casein, 2.6 g/kg DL-methionine, 586.4 g/kg vegetable shortening (Crisco), 86.2 g/kg corn oil, 88.0 g/kg cellulose, 13.0 g/kg vitamin mix (Teklad 40060), 2.5 g/kg choline bitartrate, 0.1 g/kg tertiary butylhydroquinone (TBHQ), 20.0 g/kg mineral mix (calcium phosphate deficient), 19.3 g/kg dibasic calcium phosphate, 8.2 g/kg calcium carbonate, and 0.4 g/kg magnesium oxide. KD at 0.3 calories/200 μl was prepared by diluting 11 g of KD in 50 ml of H₂O and stirring for 1 min at about 95°C.

Brains were dissected under a light microscope in a drop of fresh, ice cold 4% formaldehyde on a Sylgard plate with #5 Dumont forceps (Fine Science Tools) and transferred to an Eppendorf tube containing 1 ml 4% formaldehyde on ice for approximately 30 min. Brains were washed 3 x 20 min in 1.5 ml 1X phosphate buffered saline (PBS) with 0.3% Triton-X (PBS-T), blocked in 500 μl PBS-T and 5% normal goat serum (Sigma) overnight at 4°C. Block was removed and a primary antibody solution was added and incubated overnight at 4°C. Brains were washed 3 x 20 min in 1.5 ml PBS-T at room temperature. Secondary antibody solution was added and incubated overnight at 4°C. Brains were washed 3 x in 1.5 ml PBS-T at room temperature. The first wash (10 min) contained DAPI in PBS-T, the second wash (30 min) contained Bodipy or Nile Red in PBS-T, and the third wash (10 min) was only PBS-T. During washes and incubations, tubes were placed in a light tight box and agitated. Brains were mounted in Vectashield (Vector Laboratories) and imaged at 100x magnification on a Nikon A1R-SI+ confocal microscope (Optical Imaging Core, University of Wisconsin, Madison, WI). Primary antibodies used were α-Repo (1:100; Developmental Studies Hybridoma Bank), α-GFP (1:500; Invitrogen), α-tubulin (12G10 1:100; Developmental Studies Hybridoma Bank). Fluorescently labeled secondary antibodies used were α-mouse Alexa Fluor 488 (1:500; Invitrogen), α-chicken Alexa Fluor 488 (1:500; Invitrogen), α-mouse Alexa Fluor 594 (1:500; Invitrogen). Fluorescent dyes used were DAPI (1 mg/mL in H₂O, diluted 1:1000 in PBS-T, Santa Cruz Biotechnology), Bodipy 493/503 (0.5 mg/mL in DMSO, diluted 1:1000 in PBS-T, Invitrogen), and Nile Red 515/560 (1 mg/mL in DMSO, diluted 1:1000 in PBS-T, Sigma).

Nikon Elements Imaging Software (Optical Imaging Core, University of Wisconsin, Madison, WI) was used to determine the area of individual LDs as well as the number of LDs and nuclei (cells) in each image. Cellpose software was used to segment and count the number of DAPI-positive cells per image [89]. 100x single channel confocal images were used for this analysis from at least 11 brains and 11,000 cells. Approximately half of the brains were imaged from the posterior side and the other half from the anterior side. LDs were sorted by area using Microsoft Excel, and those between 0.2 and 15.0 μm² were counted. GraphPad Prism (version 8.4.3) was used for graphing and statistical analyses of data in Fig 3 and Tables 1 and 2.

Male w¹¹¹⁸ flies were collected at 0–1 days, aged 7 days and were subject to TBI via the method described in Katzenberger et al., 2013 [47]. Flies were transferred to vials with CMYD for 1, 7, 14, or 21 days. Heads were isolated by freezing flies in liquid nitrogen, vortexing to cause decapitation, and separating heads from bodies by passing them through a sieve. Heads from injured and uninjured flies were stored at −80°C until all samples were ready for analysis. For each time point, three samples of 150 uninjured and three samples of 150 injured heads were submitted for lipidomic analysis (Mass Spectrometry Facility, University of Wisconsin, Madison, WI).

Lipid extraction was performed using a method based extensively on Matyash et al. 2008 [90]. A solution containing 250 μl PBS, 225 μl methanol containing internal standards and 750 μl methyl tert-butyl ether (MTBE) was used for the extraction. The internal standard mix contained Avanti SPLASH LipidoMix (cat #330707−1EA) at 10 μl per sample, supplemented with C18 ceramide-d7 (d18:1-d7/18:0, Cayman Chemical cat #22788), EOS-d9 (d18:1-d9/32:0/18:2, Cayman Chemical cat #24423), linoleic acid-d11 (Cayman chemical cat #9002193), and heptadecanoyl-L-carnitine-d3 (Cayman Chemical cat #35459). Samples of 150 fly heads, held on dry ice, were transferred to bead-beating tubes. To each sample was added 250 μl PBS, 10 μl of internal standard mix, and 215 μl of methanol. The methanol containing internal standards was made up in sufficient volume for all the samples. Samples were then subjected to 2 cycles of homogenization (TissueLyzer II, Qiagen) at 30 Hz for 40 s, rested for 5 min at 4°C, then subjected to 2 more cycles of homogenization. After homogenization, 750 μl MTBE was added and two cycles of extraction, again at 30 Hz for 40s each with 5 min at 4°C in between, were performed. Following extraction there was a final rest at 4°C for 15 min. Samples were then centrifuged at 17,000 X g for 5 min at 4°C. 700 μl of the upper MTBE phase was transferred to a new 1.5 ml tube and evaporated to dryness using a speed-vac concentrator. Dried lipid samples were then reconstituted in 150 μl isopropyl alcohol (IPA). After reconstitution in IPA samples were again centrifuged at 17,000 X g for 15 min at 4°C. Concurrently with sample extractions, a process blank and a process blank spiked with internal standard mix were prepared. During data collection, an aliquot of the NIST standard reference material SRM-1950 (http://www-s.nist.gov/srmors/view_detail.cfm?srm=1950↗), Metabolites in Plasma, was analyzed along with samples and process blanks to evaluate instrument performance. All reconstituted extracts were stored at −80°C prior to analysis. LC/MS/MS analysis of the lipid samples was carried out as described by Landowski et al. 2023 [91].

Lipidomic data were analyzed using MetaboAnalyst (metaboanalyst.ca). Raw data for positive and negative ion modes were transferred separately to Microsoft Excel. Only the columns containing the compound names and lipidomic data for the samples were retained; all others were removed. The data were sorted alphabetically by compound name. Group labels were added for each sample. All internal standard species with (d7) and (d9) labels were removed. Duplicate samples were removed, and unique identifiers (e.g., _1 and _2) were added for lipid species with the same name but different peak intensities. Peak intensities were normalized to internal standards by averaging peak intensities for each lipid species (i.e., each row was averaged) and dividing each sample by the average. Normalized values for samples were then averaged (i.e., each column was averaged) and each lipid species was divided by the average. Data from positive and negative ion modes were combined into a single file, and unique identifiers were added for lipid species with the same name. Normalized data were saved as a CSV file and analyzed by Partial Least Square-Discriminant Analysis (PLS-DA) using MetaboAnalyst 6.0, which generated Figs 5A–5D. Data for 1 day (Figs 5A and 5B) were analyzed separately from data for 7, 14, and 21 days (Figs 5C and 5D). S2 and S3 Data were also used for the analyses shown in S2 Fig. GraphPad Prism (version 8.4.3) was used to analyze and graph the data in S2 Fig.