key: cord-0030253-tcgz5xql
authors: Anyaegbu, Chidozie C.; Szemray, Harrison; Hellewell, Sarah C.; Lawler, Nathan G.; Leggett, Kerry; Bartlett, Carole; Lins, Brittney; McGonigle, Terence; Papini, Melissa; Anderton, Ryan S.; Whiley, Luke; Fitzgerald, Melinda
title: Plasma Lipid Profiles Change with Increasing Numbers of Mild Traumatic Brain Injuries in Rats
date: 2022-04-02
journal: Metabolites
DOI: 10.3390/metabo12040322
sha: c2c2ebe9dd66574526ad956c4124434deaf3985f
doc_id: 30253
cord_uid: tcgz5xql

Mild traumatic brain injury (mTBI) causes structural, cellular and biochemical alterations which are difficult to detect in the brain and may persist chronically following single or repeated injury. Lipids are abundant in the brain and readily cross the blood-brain barrier, suggesting that lipidomic analysis of blood samples may provide valuable insight into the neuropathological state. This study used liquid chromatography-mass spectrometry (LC-MS) to examine plasma lipid concentrations at 11 days following sham (no injury), one (1×) or two (2×) mTBI in rats. Eighteen lipid species were identified that distinguished between sham, 1× and 2× mTBI. Three distinct patterns were found: (1) lipids that were altered significantly in concentration after either 1× or 2× F mTBI: cholesterol ester CE (14:0) (increased), phosphoserine PS (14:0/18:2) and hexosylceramide HCER (d18:0/26:0) (decreased), phosphoinositol PI(16:0/18:2) (increased with 1×, decreased with 2× mTBI); (2) lipids that were altered in response to 1× mTBI only: free fatty acid FFA (18:3 and 20:3) (increased); (3) lipids that were altered in response to 2× mTBI only: HCER (22:0), phosphoethanolamine PE (P-18:1/20:4 and P-18:0/20:1) (increased), lysophosphatidylethanolamine LPE (20:1), phosphocholine PC (20:0/22:4), PI (18:1/18:2 and 20:0/18:2) (decreased). These findings suggest that increasing numbers of mTBI induce a range of changes dependent upon the lipid species, which likely reflect a balance of damage and reparative responses.

Mild traumatic brain injury (mTBI) is increasingly recognised as a substantial public health problem, particularly in the sporting context, where repeated mTBI (rmTBI) is prevalent. At least 20% of the 3.2 million people worldwide that present to hospital each year experience persistent neuropsychological, cognitive, sleep or physical dysfunction [1] , with those experiencing rmTBI at greater risk [2, 3] . rmTBI is also associated with an cellular lipid alterations likely to be of clinical relevance. In this study, we used liquid chromatography-mass spectrometry (LC-MS) in conjunction with our rat rotational model of closed head mTBI [19] [20] [21] [22] to profile the plasma lipidome at day 11 following sham, one (1×) or two (2×) mTBI, hypothesising that changes would be observed with increasing numbers of mTBI relative to sham injury. Analyses were performed at day 11 because our previous studies showed that the cellular indicators of damage, including microglial reactivity and oxidative stress, were most apparent at this subacute time [19, 20] , as compared to the more acute (day 4; [21] ) and chronic (3 month; [22] ) time points. Female rats were used to enable comparison of findings to our previous work [19] [20] [21] [22] , and to address the substantial under-representation of female animals in TBI literature.

Following data cleaning, the final dataset consisted of 856 reproducible lipid species from 19 different lipid classes. A principal component analysis (PCA) model was created to assess the overall variation in the dataset ( Figure S1 ). From the PCA, the replicate quality control (QC) samples successfully clustered, demonstrating the reproducibility of the extraction and analysis.

To identify the metabolic signature associated with mTBI, supervised orthogonal projection to latent structure-discriminant analysis (OPLS-DA) was performed on the lipid data generated from the sham (n = 4), 1× mTBI (n = 6) and 2× mTBI (n = 7) groups ( Figure 1 ). The OPLS-DA model scores were as follows: sham control and 1× mTBI groups ( Figure 1A and Table S1 ) (R2 X = 0.637, R2 Y = 0.858, Q2 = 0.307); 1× mTBI and 2× mTBI groups ( Figure 1B and Table S2 ) (R2 X = 0.558, R2 Y = 0.84, Q2 = 0.206); sham control and 2× mTBI groups ( Figure 1C and Table S3 ) (R2 X = 0.603, R2 Y = 0.952, Q2 = 0.322). 

Given the results in the OPLS-DA models and to further stratify the three groups, a non-parametric Kruskal-Wallis one-way analysis of variance (ANOVA) on ranks was performed using the individual lipid species in the data set to further clarify which metabo- Finally, to test for the existence of a progressive trend in the data following the 1× mTBI and 2× mTBI interventions, the 1× mTBI data were predictively projected onto the existing OPLS-DA model for sham control vs. 2× mTBI. This resulted in the 1× mTBI data projecting into the model space between the sham control and 2× TBI groups, indicating a progressive change in the lipid data: sham control → 1× mTBI → 2× mTBI ( Figure 1D ).

Given the results in the OPLS-DA models and to further stratify the three groups, a nonparametric Kruskal-Wallis one-way analysis of variance (ANOVA) on ranks was performed using the individual lipid species in the data set to further clarify which metabolites were driving the OPLS-DA models. This was performed on 856 individual lipid species identified in the data set, and a post hoc Dunn's test was then completed to report inter-class differences. A total of 18 plasma lipids were found to be significantly different (p ≤ 0.05) between the sham control, 1× mTBI and 2× mTBI groups, respectively (Table 1) . These lipids could be grouped into three distinct patterns of change following injury. The first pattern was lipids whose concentration was significantly altered by both 1× mTBI or 2× mTBI ( Figure 2 ): this included CE (14:0), which increased after 1× (p = 0.04) and 2× injuries (p < 0.01); PS (14:0/18:2) and HCER (d18:0/26:0), which decreased after 1× and 2× mTBI (PS (14:0/18:2): p < 0.01 for 1× and 2× mTBI; HCER (d18:0/26:0): p = 0.03 and p < 0.001 for 1× and 2× mTBI, respectively); and PI (16:0/18:2), which increased significantly after 1× mTBI (p < 0.01) but decreased in response to 2× mTBI (p = 0.01).

The second pattern of change was of lipids whose concentration was altered only in response to 1× mTBI ( Figure 3 ). This pattern was specific to FFA, with significant increases in concentration of both FFA (18:3) (p = 0.05) and FFA (20:3) (p < 0.01) with 1× injury only.

The third pattern was of lipids altered only in response to 2× mTBI ( Figure 4 ). This pattern comprised an increase in the concentration of HCER (22:0) (p = 0.04) and PEs (P-18:1/20:4) (p = 0.02) and (P-18:0/20:1) (p = 0.05), and decreased concentrations of LPE (20:1) (p = 0.01), PC (20:0/22:4) (p = 0.05) and PIs (18:1/18:2) (p = 0.04) and (20:0/18:2) (p = 0.05), after 2× mTBI. Some lipid alteration was also observed between 1× and 2× mTBI ( Figure S2 ). However, since there were no observed differences in comparison to the sham control, these findings were not considered in the scope of our hypotheses and current analyses.

Kruskal-Wallis effect size (eta 2 ), individual group means and their 95% confidence intervals ar also displayed. CE, cholesterol ester; FFA, free fatty acid; HCER, hexosylceramide; LPE, lysoph phatidylethanolamine; LPG, lysophosphoglycerol; mTBI, mild traumatic brain injury; PC, phos phocholine; PE, phosphoethanolamine; PI, phosphoinositol; PS, phosphoserine.

These lipids could be grouped into three distinct patterns of change following inj The first pattern was lipids whose concentration was significantly altered by both 1× m or 2× mTBI ( Figure 2 ): this included CE (14:0), which increased after 1× (p = 0.04) and injuries (p < 0.01); PS (14:0/18:2) and HCER (d18:0/26:0), which decreased after 1× and mTBI (PS (14:0/18:2): p < 0.01 for 1× and 2× mTBI; HCER (d18:0/26:0): p = 0.03 and p < 0 for 1× and 2× mTBI, respectively); and PI (16:0/18:2), which increased significantly afte mTBI (p < 0.01) but decreased in response to 2× mTBI (p = 0.01).

Lipids that change with one and two injuries vs. sham control. (A), increase after 1× o mTBI; (B), decrease after 1× or 2× mTBI; (C), increase with 1× mTBI and decrease with 2× m Boxplots represent the minimum, first quartile, median, third quartile, and maximum, for the trol, 1× mTBI and 2× mTBI groups. Significant differences between groups are indicated above boxplots by * (p < 0.05) or ** (p < 0.01). CE, cholesterol ester; HCER, hexosylceramide; mTBI, m traumatic brain injury; PI, phosphoinositol; PS, phosphoserine.

The second pattern of change was of lipids whose concentration was altered onl response to 1× mTBI ( Figure 3 ). This pattern was specific to FFA, with significant incre in concentration of both FFA (18:3) (p = 0.05) and FFA (20:3) (p < 0.01) with 1× injury o The third pattern was of lipids altered only in response to 2× mTBI ( Figure 4 ). This pat comprised an increase in the concentration of HCER (22:0) (p = 0.04) and PEs (P-18:1/2 (p = 0.02) and (P-18:0/20:1) (p = 0.05), and decreased concentrations of LPE (20:1) (p = 0 PC (20:0/22:4) (p = 0.05) and PIs (18:1/18:2) (p = 0.04) and (20:0/18:2) (p = 0.05), after 2× m Some lipid alteration was also observed between 1× and 2× mTBI ( Figure S2 ). Howe since there were no observed differences in comparison to the sham control, these findi were not considered in the scope of our hypotheses and current analyses. Lipids that change with one and two injuries vs. sham control. (A), increase after 1× or 2× mTBI; (B), decrease after 1× or 2× mTBI; (C), increase with 1× mTBI and decrease with 2× mTBI. Boxplots represent the minimum, first quartile, median, third quartile, and maximum, for the control, 1× mTBI and 2× mTBI groups. Significant differences between groups are indicated above the boxplots by * (p < 0.05) or ** (p < 0.01). CE, cholesterol ester; HCER, hexosylceramide; mTBI, mild traumatic brain injury; PI, phosphoinositol; PS, phosphoserine.

etabolites 2022, 12, x FOR PEER REVIEW Figure 3 . Lipids that change with one injury vs. sham control. Boxplots represent the min quartile, median, third quartile, and maximum, for the control, 1× mTBI and 2× mTBI g nificant differences between groups are indicated above the boxplots by ** (p < 0.05) or * FFA, free fatty acid; mTBI, mild traumatic brain injury. Lipids that change with one injury vs. sham control. Boxplots represent the minimum, first quartile, median, third quartile, and maximum, for the control, 1× mTBI and 2× mTBI groups. Significant differences between groups are indicated above the boxplots by ** (p < 0.05) or ** (p < 0.01). FFA, free fatty acid; mTBI, mild traumatic brain injury. Figure 3 . Lipids that change with one injury vs. sham control. Boxplots represent the minimum, first quartile, median, third quartile, and maximum, for the control, 1× mTBI and 2× mTBI groups. Significant differences between groups are indicated above the boxplots by ** (p < 0.05) or ** (p < 0.01). FFA, free fatty acid; mTBI, mild traumatic brain injury. , decrease after 2× mTBI. Boxplots represent the minimum, first quartile, median, third quartile, and maximum, for the control, 1× mTBI and 2× mTBI groups. Significant differences between groups are indicated above the boxplots by * (p < 0.05) or ** (p < 0.01). HCER, hexosylceramide; LPE, lysophosphatidylethanolamine; mTBI, mild traumatic brain injury; PC, phosphocholine; PE, phosphoethanolamine; PI, phosphoinositol.

The metabolic changes associated with mTBI are complex and can have central and peripheral physiological effects. LC-MS can provide unique insight into chemical perturbations which occur following TBI [23] [24] [25] . Lipids are broadly known to be disturbed following TBI. However, the characterisation of alterations in individual lipid species after mTBI is an emerging line of enquiry [10, 26, 27] . Here, we applied a target lipid screen to (B), decrease after 2× mTBI. Boxplots represent the minimum, first quartile, median, third quartile, and maximum, for the control, 1× mTBI and 2× mTBI groups. Significant differences between groups are indicated above the boxplots by * (p < 0.05) or ** (p < 0.01). HCER, hexosylceramide; LPE, lysophosphatidylethanolamine; mTBI, mild traumatic brain injury; PC, phosphocholine; PE, phosphoethanolamine; PI, phosphoinositol.

The metabolic changes associated with mTBI are complex and can have central and peripheral physiological effects. LC-MS can provide unique insight into chemical perturbations which occur following TBI [23] [24] [25] . Lipids are broadly known to be disturbed following TBI. However, the characterisation of alterations in individual lipid species after mTBI is an emerging line of enquiry [10, 26, 27] . Here, we applied a target lipid screen to measure 856 lipid species to investigate the effect of single vs. repeated mTBI on the plasma lipidome in a rat mTBI model at 11 days post-injury, a time at which the oxidative and inflammatory mediators of secondary pathology are evident [21, 22] . We identified 18 significantly altered lipids in rat plasma that could accurately distinguish between sham, 1× mTBI and 2× mTBI, indicating that the lipidome remains significantly altered in the subacute phase of mTBI, and that this pathology increases variably with repeated injury.

Multivariate OPLS-DA (sham vs. 1× mTBI; sham vs. 2× mTBI; 1× mTBI vs. 2× mTBI) resulted in models with distinct separations between the classes. Furthermore, projection of the 1× mTBI intervention group onto the sham vs. 2× TBI model resulted in the 1× mTBI group being predicted centrally between the sham group and the 2× mTBI group, indicating that overall, lipid alterations follow progressive trends with increasing mTBI insults. These findings may provide further insight into the pathological consequences of increasing numbers of mTBI, as seen in contact sport and military injuries. Following multivariate modelling of the data, univariate analysis indicated eighteen lipids that distinguished between the sham, 1× mTBI and 2× mTBI groups. Further analysis of concentration ratios demonstrated three principal patterns of change across the 18 lipid biomarkers identified: (1) altered lipid concentration with both 1× or 2× mTBI vs. sham control; (2) altered lipid concentration with 1× mTBI (but not 2× mTBI); (3) lipid concentrations that significantly differed from sham control concentrations only after 2× mTBI (See Table S5 ).

Cholesterol esters (CE) conformed to the first pattern, being increased after both 1× and 2× mTBI. CEs have been shown to increase transiently in a controlled cortical impact rat model of moderate/severe TBI, with CE (18:1) peaking at 3 days in the lesion and perilesional tissue [28] . However, altered CE (14:0) has not previously been reported in either a closed-head or an mTBI model. Our results suggest that the time course of CE alteration may be substantially different in closed-head injury, with significant elevations in plasma still detected at 11 days post-injury in our closed-head rotational mTBI model after 1× and 2× injuries. Indeed, the subacute elevations of lipids following this identified pattern warrant further investigation for their capacity as injury biomarkers, given that they were readily detectable in plasma beyond the acute phase. In contrast, we found that phosphoserine (PS) (14:0/18:2) and hexosylceramide (HCER) (d18:0/26:0) significantly decreased with both 1× and 2× mTBI. As an ester, PS is an important intermediary for phosphoric acid and serine. L-serine has received increasing attention as a multifunctional agent capable of restoring cognitive function, reducing inflammation and promoting remyelination in several neurological conditions [29] , suggesting that downstream metabolite loss may reduce the production of this neurotrophic factor, with deleterious consequences. Lipid species from the phosphoinositol (PI) class (16:0/18:2) were increased in concentration in the 1× mTBI group when compared to the sham control group. However, the same species decreased in the 2× mTBI group. This finding is paradoxical, although it mirrors our previous findings of microglial reactivity and oxidative damage in brain tissue in this model [21, 22] , lending support to its biological plausibility. While these opposing responses to single and repeated injuries are currently under investigation, they reveal that pathology may not follow a linear pattern with increasing injury, but may instead involve distinct responses.

The second observed lipid pattern comprised free fatty acids (FFA), which were also increased only after 1× mTBI. FFAs are precursors to eicosanoids, which are important modulators of inflammation, cell signalling and vascular response (among others) throughout the central nervous system [30] . These products of lipid peroxidation have been increasingly reported to accumulate after TBI in both tissue and biofluids, potentiating the pathological response [25, 31, 32] . As such, FFA concentrations have been suggested as a putative clinical biomarker of outcome [32] . In the present study, our finding of increased plasma concentrations of both FFA (18:3) and FFA (20:3) with 1× mTBI supports these clinical findings, potentially demonstrative of persisting lipid peroxidation and inflammation in this subacute post-injury period. It is noteworthy that this pattern was not observed after 2× mTBI, and further investigation is required in a larger cohort to determine whether this FFA alteration is truly specific to single injury.

Our third observed pattern of lipids responsive only to 2× mTBI included HCER (22:0), PE (P-18:0/20:1), and PE (P-18:1/20:4), which increased with injury. These two PE species have been linked to multiple neurodegenerative diseases, including Parkinson's disease and Alzheimer's disease [33, 34] , where they were found at increased concentrations in the post-mortem brain. Findings of PE alterations in experimental injury have been mixed: in moderate-severe TBI, PE levels are increased in lesioned brain tissue at three days [35] . Moreover, in moderate-severe TBI, decreased serum PE levels have been reported across the first seven days [25] and at three months, coinciding with decreases in brain tissue [36] . A recent study also described serum PE decreases at 24 h after single mTBI and repeated mTBI [37] . In contrast, in a mouse model of repeated mTBI, total PE concentrations significantly increased in the hippocampus at 24 h after injury, and remained significantly elevated to 12 months post-injury [38] . Given the roles for PE in myelin maintenance at the axonal node and vesicular formation [39, 40] , this generalised mTBI-mediated increase in PE species at 11 days post-injury may reflect heightened transport for reparative myelination at the node of Ranvier, for which we have previously described significant alteration after rmTBI in this model [19] . In contrast, several lipids were decreased only in response to 2× mTBI, including lysophosphatidylethanolamine (LPE) (20:1), phosphocholine (PC) (20:0/22:4), and phosphoinositol (PI) (18:1/18:2) and (20:0/18:2). LPEs have recently been shown to stimulate neurite outgrowth and reduce glutamate-mediated excitotoxicity in neuronal culture [41] , indicating a beneficial role after injury. Likewise, PIs are key regulators of cytoskeletal function and myelination, with tissue depletion linked to demyelination and axonal loss [42] , and PCs have a vital role in the structural integrity of the neuronal and glial membranes [43] .

While future mechanistic studies are required to identify the damage processes responsible for the subacute lipid alterations reported in this study, we previously found increased lipid peroxidation in cortical neurons at day four following 2× mTBI [21] , suggesting that oxidative stress-induced lipid peroxidation may play a role. Lipid peroxidation is a hallmark of neurodegenerative conditions, such as TBI [44] . The prolonged peroxidation of phospholipids causes cell membrane disintegration [45] . As PC and PE are the major lipids in neuronal somas and neuritic processes [46] , their oxidative damage in the brain could potentially stimulate increased recruitment of PC, PE and their metabolites from the periphery for repairing neurons. The observed plasma decreases in LPE, PC and PI concentrations after 2× mTBI may therefore reflect a higher brain demand for reparative purposes. Further investigation of brain tissue lipid species will be required to investigate this possibility.

The findings of this study need to be considered in the context of several limitations. We cannot be sure of the extent to which the lipids detected in extracranial plasma samples reflect the true nature of lipid alteration within brain tissue, or whether they are altered by peripheral factors. Future research should compare lipid profiles in plasma to those in the brain parenchyma and/or cerebrospinal fluid to establish lipid similarities or differences after mTBI. This will also be particularly valuable for studies of temporal lipid changes. We chose 11 days post-mTBI for our exploratory lipid analyses, as this time point is known to have particular pathological relevance in this model. However, the characterisation of alterations at additional acute and chronic time points is a key future direction to delineate whether lipid alteration is indicative of injury or repair responses.

Our study also included small cohorts of rats in each group, and thus may have been underpowered to detect additional statistical differences between groups. Further work is needed to determine if these preliminary findings are replicable in a larger cohort, and to discover whether any key lipid changes were missed in this initial analysis.

Finally, we cannot speak to whether our findings in female rats are translatable to males. Studies in females are lacking in TBI research, with comparatively little being known about female biological responses to injury [47] . Since female sex hormones may be neuroprotective following TBI [48] , future studies including male rats are required to identify sex-dependent changes in lipid profiles. However, it is noteworthy that our study identified differences in lipid species between uninjured controls and 1× or 2× mTBI, indicating that the neuroprotective effects of female hormones are not absolute, and allowing the detection of injury-induced changes in female rats.

Adult female Piebald Viral Glaxo rats (160-200 g; 3 months of age) from the Animal Resource Centre (Murdoch, WA, Australia) were used for this study. Animals were housed in pairs under specific pathogen-free conditions and a 12:12 h light:dark cycle. Rats had free access to food and water. They were acclimatised to housing conditions for a minimum of one week prior to any procedures. All procedures involving rats were approved by the University of Western Australia Animal Ethics Committee (Approval Number RA/3/100/1699), in accordance with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes, issued by the National Health and Medical Research Council.

Twenty rats were randomly assigned to the sham (n = 6), 1× mTBI (n = 7) or 2× mTBI (n = 7) groups. Injuries were delivered at 24 h intervals, using a closed head, rotational acceleration weight-drop model of mTBI which has been described in detail previously [21] . Briefly, rats were anaesthetised with 4% isoflurane in 4 L/min oxygen and their heads were shaved before they were laid prone on a delicate task wiper (Kimwipes, Kimberly-Clark, Irving, TX, USA). The rat's head was aligned to ensure that the 250 g weight released from a 1 m height would impact midline 2-3 mm anterior to the front of the ears, at the lambda suture line. Sham procedures were identical except for the weight drop (i.e., no injury). Animals that received 1× mTBI received a sham procedure when mTBI was not administered, ensuring equal anaesthesia for all animals to control for the effect of anaesthesia. All rats received analgesia (Carprofen, 4 mg/kg, s.c., Norbrook Laboratories, Tullamarine, VIC, Australia) immediately after the sham or mTBI procedures, before they were returned to their housing.

At day 11 after the first sham or mTBI (day 1), rats were deeply anesthetised using pentobarbital sodium (160 mg/kg i.p., Virbac; supplied by Provet, Malaga, WA, Australia). Peripheral blood (2 mL) was collected by cardiac puncture using a 23 G needle in a syringe, transferred to an EDTA vacutainer (BD Vacutainer; REF 367839; BD Biosciences, North Ryde, NSW, Australia) and mixed by gentle inversion of the tubes. Plasma was separated from whole blood by centrifugation at 1500× g for 15 min at 4 • C, and stored at −80 • C until use.

Lipidomic analysis was completed using an established method previously described [49] , resulting in the quantification of 856 lipid species from 10 µL of rat plasma. Sample extraction was completed using a Biomek i5 sample automation system (Beckman Coulter, Mount Waverley, VIC Australia). To each 10 µL plasma sample, 90 µL of isopropanol (Fisher Scientific, Malaga, Western Australia) containing stable isotope-labelled internal standards (LipidyzerTM Internal Standards Kit from Sciex, Framingham, MA, USA, and SPLASH LipidoMIXTM, Lyso PI 17:1, Lyso PG 17:1, and Lyso PS 17:1, Avanti Polar Lipids, Alabaster, AL, USA) was added, and then mixed for 20 min. Samples were then centrifuged (3500× g for 10 min at 4 • C). Supernatant aliquots of 70 µL were then transferred into an Eppendorf 350 µL 96-well plate for LC-MS analysis. Samples were analysed within 24 h of preparation. QC samples were prepared using a commercial pooled human plasma sample (BioIVT, New York, NY, USA) and underwent replicate extraction (n = 5) using the method described. These replicate samples were periodically analysed throughout the analytical sequence.

Lipidomic analysis of the plasma sample extracts was performed using a targeted tandem mass spectrometry (MS/MS) approach using predefined multiple reaction monitoring (MRM) transitions (Sciex sMRM Pro Builder, Framingham, MA, USA) and in-house defined chromatographic retention time windows.

The analytical system consisted of a Sciex ExionLC TM coupled to a Triple Quadrupole Linear Ion Trap (QTRAP 6500+) (SCIEX, Concord, ON, Canada). Lipid metabolites were chromatographically separated using a Waters Acquity BEH C18 reverse phase column (1.7 µm, 100 × 2.1 mm particle size; Waters Corp., Milford, MA, USA), maintained at 60 • C.

The mobile phase consisted of water, acetonitrile and isopropanol (all Optima grade), which were all purchased from Thermo Fisher Scientific (Malaga, Western Australia). Mobile phase composition and gradients are described in detail in the Supplementary Materials, along with additional instrument settings. Specific MS settings, MRM transitions and chromatographic retention times for the lipids of interest are reported in the results section and Tables S2-S4 are included in Table S4 .

Peak picking was conducted using SkylineMS [50] (Version 21.1) MacCoss Lab Software, University of Washington, Seattle, WA, USA). Data preprocessing and feature filtering was performed using R (version 4.4.1; R Foundation, Vienna, Austria) in R Studio [51] (version 1.4.1; R Studio, Boston, MA, USA) using in-house scripts; first, the lipid features were kept in the dataset if the relative standard deviation across the replicate QC samples was <30%; signal drift correction was then performed using the package statTarget [52] ; the data then underwent logarithmic transformation to normalise variance and to estimate the normal distribution required for the univariate and multivariate analyses.

For multivariate statistics, the final data matrix was transferred to SIMCA 17.0.1 (Sartorius AG, Goettingen, Germany). Data were pareto-scaled prior to multivariate modelling. PCA was performed to assess the measurement precision of the QC data, as well as to identify and remove the outlier samples. Three outlier samples (sham group = 2, 1× mTBI group = 1) were identified within the PCA via Hotelling's T2 (95% CI) and removed from subsequent statistical analysis. Following PCA analysis, a supervised OPLS-DA was performed. Variable Importance for Projection Plots (VIP) were used to determine the importance of each lipid in the OPLS-DA model.

Following multivariate analysis, and to further stratify the three groups, univariate tests were performed using a Kruskal-Wallis one-way ANOVA on ranks, with Dunn's test applied post-hoc to determine intergroup differences (sham, 1× mTBI and 2× mTBI). Statistical significance was deemed where p ≤ 0.05. As the study was of a discovery design and to avoid the potential loss of true positives, no correction for multiple testing was performed. Univariate analysis was completed using R (version 4.4.1) in R Studio [51] (version 1.4.1). Box plots were produced using GraphPad Prism v8.0.0 (GraphPad Software, San Diego, CA, USA, www.graphpad.com (accessed on 28 October 2021)).

Lipids are key molecules in a plethora of biochemical and physiological processes, including cellular signalling, immune responses, membrane lipid structure, and inflammatory pathways. Measuring the perturbations of the lipid species in clinically-relevant mTBI models provides further insight into the underlying pathophysiology, and highlights lipid alteration as a potential therapeutic target to improve outcomes following mTBI. The findings from our lipidomic analysis build on reported TBI insights in the literature. Furthermore, we provide novel insights into the subacute lipidomic signature of single and repeated mTBI. Given our finding of bi-directional plasma lipid concentration changes with single and repeated injury, the alteration of lipid species may be a new avenue for biomarker research in mTBI. Future work is needed to establish the degree to which pathological brain changes are reflected by plasma lipids, and to characterise the temporal lipid profile after single and repeated injury.

Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/metabo12040322/s1, Figure S1 : Unsupervised Principal Component Analysis (PCA) with 19 samples (6 from sham controls, 6 from rats with 1× mTBI, and 7 from rats with 2× mTBI). Figure 1C (1× mTBI vs. 2× mTBI). The table includes the top 20 lipid features (ordered by VIP value). Figure S2 . Lipids that decreased between one mTBI and two mTBIs. Boxplots represent the minimum, first quartile, median, third quartile, and maximum, for the control, 1× mTBI and 2× mTBI groups. Table S4 

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

The authors declare no conflict of interest. 

Post-concussion states: How do we improve our patients' outcomes? An Australian perspective

Neuropsychological performance following a history of multiple self-reported concussions: A meta-analysis

Long-term effects of sport concussion on cognitive and motor performance: A review

Emerging Histomorphologic Phenotypes of Chronic Traumatic Encephalopathy in American Athletes

Long-term consequences of recurrent sports concussion

The New Neurometabolic Cascade of Concussion

Induction of Oxidative and Nitrosative Damage Leads to Cerebrovascular Inflammation in an Animal Model of Mild Traumatic brain Injury Induced by Primary Blast. Free Radic

Disruption of Axo-Glial Junctions Causes Cytoskeletal Disorganization and Degeneration of Purkinje Neuron Axons

Lipid profiling of brain tissue and blood after traumatic brain injury: A review of human and experimental studies

Mild TBI Results in a Long-Term Decrease in Circulating Phospholipids in a Mouse Model of Injury

Organization and Evolution of Brain Lipidome Revealed by Large-Scale Analysis of Human, Chimpanzee, Macaque, and Mouse Tissues

Membrane Lipids and Cell Signaling

Fatty Acid Uptake and Incorporation in Brain: Studies with the Perfusion Model

Delivery and Turnover of Plasma-Derived Essential PUFAs in Mammalian Brain

Metabolomics Profiling of Concussion in Adolescent Male Hockey Players: A Novel Diagnostic Method

Plasma Lipidomic Profiling in a Military Population of Mild Traumatic Brain Injury and Post-Traumatic Stress Disorder with Apolipoprotein E 4-Dependent Effect

Disruption in Brain Phospholipid Content in a Humanized Tau Transgenic Model Following Repetitive Mild Traumatic Brain Injury

Biomechanics of Concussion

The Effects of a Combination of Ion Channel Inhibitors in Female Rats Following Repeated Mild Traumatic Brain Injury

Simultaneous flow cytometric characterization of multiple cell types and metabolic states in the rat brain after repeated mild traumatic brain injury

Repeated mild traumatic brain injury in female rats increases lipid peroxidation in neurons

Differential responses to increasing numbers of mild traumatic brain injury in a rodent closed-head injury model

Phospholipid profiling of plasma from GW veterans and rodent models to identify potential biomarkers of Gulf War Illness

Plasma metabolomic biomarkers accurately classify acute mild traumatic brain injury from controls

Discovery of Lipidome Alterations Following Traumatic Brain Injury via High-Resolution Metabolomics

Plasma Lipidomic Analyses in Cohorts With mTBI and/or PTSD Reveal Lipids Differentially Associated with Diagnosis and APOE ε4 Carrier Status

The Role of Bioactive Lipids in Attenuating the Neuroinflammatory Cascade in Traumatic Brain Injury

Mass Spectrometry Imaging of Rat Brain Lipid Profile Changes over Time Following Traumatic Brain Injury

Is a Potential Neuroprotective Agent for Neurological Disease and Injury

Arachidonic-Acid-Derived Eicosanoids: Roles in Biology and Immunopathology

Global assessment of oxidized free fatty acids in brain reveals an enzymatic predominance to oxidative signaling after trauma

Eicosanoids and Inflammation

Severe Alterations in Lipid Composition of Frontal Cortex Lipid Rafts from Parkinson's Disease and Incidental Parkinson's Disease

Plasmalogen Deficiency in Early Alzheimer's Disease Subjects and in Animal Models: Molecular Characterization Using Electrospray Ionization mass Spectrometry

Matrix-Assisted Laser Desorption/Ionization-Mass Spectrometry Imaging of Lipids in Experimental Model of Traumatic Brain Injury Detecting Acylcarnitines as Injury Related Markers

Lipidomic Analyses Identify Injury-Specific Phospholipid Changes 3 mo after Traumatic Brain Injury

Lipidome Alterations following Mild Traumatic Brain Injury in the Rat

Converging and Differential Brain Phospholipid Dysregulation in the Pathogenesis of Repetitive Mild Traumatic Brain Injury and Alzheimer's Disease

Ethanolamine and Phosphatidylethanolamine: Partners in Health and Disease

Formation of Tubules and Helical Ribbons by Ceramide Phosphoethanolamine-Containing Membranes

Abundant Oleoyl-Lysophosphatidylethanolamine in Brain Stimulates Neurite Outgrowth and Protects against Glutamate Toxicity in Cultured Cortical Neurons

Phosphoinositides: Regulators of Nervous System Function in Health and Disease

Dynamics of Choline-Containing Phospholipids in Traumatic Brain Injury and Associated Comorbidities

Lipid Peroxidation and Neurodegenerative Disease. Free Radic

Lipid Peroxidation in Cell Death

Lipid Composition of Neuronal Cell Bodies and Neurites from Cultured Dorsal Root Ganglia

Sex Differences in Traumatic Brain Injury: What We Know and What We Should Know

Neuroprotection by Estrogen and Progesterone in Traumatic Brain Injury and Spinal Cord Injury

Diagnostic Potential of the Plasma Lipidome in Infectious Disease: Application to Acute SARS-CoV-2 Infection

Skyline for Small Molecules: A Unifying Software Package for Quantitative Metabolomics

R: A Language and Environment for Statistical Computing; R Foundation for Statistical Computing

statTarget: A streamlined Tool for Signal Drift Correction and Interpretations of Quantitative Mass Spectrometry-Based Omics Data