Comparing Brain and Blood Lipidome Changes following Single and Repetitive Mild Traumatic Brain Injury in Rats

Abstract

Traumatic brain injury (TBI) is a major health concern in the United States and globally, contributing to disability and long-term neurological problems. Lipid dysregulation after TBI is underexplored, and a better understanding of lipid turnover and degradation could point to novel biomarker candidates and therapeutic targets. Here, we investigated overlapping lipidome changes in the brain and blood using a data-driven discovery approach to understand lipid alterations in the brain and serum compartments acutely following mild TBI (mTBI) and the potential efflux of brain lipids to peripheral blood. The cortices and sera from male and female Sprague-Dawley rats were analyzed via ultra-high performance liquid chromatography-mass spectrometry (UHPLC-MS) in both positive and negative ion modes following single and repetitive closed head impacts. The overlapping lipids in the data sets were identified with an in-house data dictionary for investigating lipid class changes. MS-based lipid profiling revealed overall increased changes in the serum compartment, while the brain lipids primarily showed decreased changes. Interestingly, there were prominent alterations in the sphingolipid class in the brain and blood compartments after single and repetitive injury, which may suggest efflux of brain sphingolipids into the blood after TBI. Genetic algorithms were used for predictive panel selection to classify injured and control samples with high sensitivity and specificity. These overlapping lipid panels primarily mapped to the glycerophospholipid metabolism pathway with Benjamini-Hochberg adjusted q-values less than 0.05. Collectively, these results detail overlapping lipidome changes following mTBI in the brain and blood compartments, increasing our understanding of TBI-related lipid dysregulation while identifying novel biomarker candidates.

Keywords: Traumatic brain injury; blood biomarkers; brain biomarkers; lipidomics; ultra-high performance mass spectrometry.

Figure 1

Overview of the workflow from study design to interpretation. (A) Experimental groups included males (n = 14) and females (n = 18). Rats were assigned to a sham control group that received no injuries (n = 11), a single impact group that received one impact (n = 10), or a repeat impact group that received three separate impacts (n = 11). (B) Injury groups received closed head impacts, and the blood and brains were collected at 24 h postinjury. Lipids were extracted with homogenization in isopropanol. (C) Samples were analyzed in random order with high-resolution LC-MS. (D) Spectral alignment, peak detection, isotope and adduct grouping, gap filling, and drift correction were accomplished using Compound Discoverer v.3.0. (E) Lipids detected in the brain and serum data sets were selected and annotated by spectra matching to in-house databases. (F) PCA score plot of brain and serum data sets. (G) Features selected by machine learning algorithms were combined to create an oPLS-DA model. (H) Annotated compounds selected in the models were imported into LIPEA to explore biological pathways altered following TBI.

Figure 2

Annotated features observed in the brain and serum compartments. (A) Pie chart of number of annotated lipids that overlapped in the cortex and serum data sets. (B) PCA score plot of 250 annotated lipids that overlap in the brain and serum compartments. Data show clear separation along PC1 of the brain and serum metabolites. (C) Spider plot of annotated features in both brain and serum data sets based on lipid subclasses depicts the number of lipids that either increased or decreased due to injury in each compartment. Dark colored bars signify number of significant features between injury severity and sham. Welch’s t test p < 0.05. Abbreviations: Car – acyl carnitines, CE – cholesteryl esters, Cer – ceramides, DG – diacylglycerols, FFA – free fatty acids, LPC – lysophosphatidylcholine, LPE – lysophosphatidylethanolamine, PC – phosphatidylcholine, PE – phosphatidylethanolamine, PI – phosphatidylinositol, PS – phosphatidylserine, SM – sphingomyelin, TG – triacylglycerols.

Figure 3

Heatmap of fold change values of the lipid classes. Fold change values of overlapping lipids injury relative to sham control. (A) Sphingolipids. (B) Glycerolipids. (C) Free Fatty Acids. (D) Phospholipids.

Figure 4

Abundances of overlapping lipids in the cortex and serum. (A) Volcano plot of cortex relative to serum abundances after mTBI. Red dots denote enriched in the brain, blue dots denote enriched in the serum, and green dots denote brain-specific lipids. (B) Heatmap of lipids that are either enriched in the cortex or serum based on log2(FC) of 5 between brain and blood samples. FFA class was removed from analysis due to limitations in the lipid internal standard.

Figure 5

PCA and oPLS-DA models of brain and serum compartments comparing SHAM and rmTBI. (A) PCA score plot of 21 features in the brain that have high fold change values (± 1.5). (B) oPLSDA model of 11 features that distinguish between SHAM and 3X with a sensitivity of 87.5% and specificity of 100%. (C) PCA score plot of 69 features in the brain that have high fold change values (± 1.5). (D) oPLSDA model of 13 features that distinguish between SHAM and 3X with sensitivity 88.9% and specificity of 100%. Principal component analysis (PCA), orthogonal partial least-squares discriminant analysis (oPLS-DA). (E) LIPEA pathway analysis for lipids in final panels. The Rich factor, shown along the x-axis, represents the number annotated lipids belonging to a specific pathway in the data set out of total of known lipids in the pathway. The y-axis represents logarithmic value using the Benjamini–Hochberg correct factor. The size of bubbles represents the percentage of lipids converted from the final panel.

Figure 6

PCA and oPLS-DA score plots of cortex and serum compartments comparing SHAM and smTBI. (A) PCA score plot of 70 features in the brain that have high fold change values (± 1.20). (B) oPLS-DA model of 13 features that distinguish between SHAM and 1X with sensitivity and specificity of 100%. (C) PCA score plot of 136 features in the serum that have high fold change values (± 1.20). (D) oPLS-DA model of 12 features that distinguish between SHAM and 1X with sensitivity and specificity of 100%. Principal component analysis (PCA), orthogonal partial least-squares discriminant analysis (oPLS-DA). (E) LIPEA pathway analysis for lipids in final panels. The Rich factor, shown along the x-axis, represents the number annotated lipids belonging to a specific pathway in the data set out of total of known lipids in the pathway. The y-axis represents a logarithmic value using the Benjamini–Hochberg correct factor. The size of the bubbles represents the percentage of lipids converted from the final panel.