A class of anti-inflammatory lipids decrease with aging in the central nervous system

Abstract

Lipids contribute to the structure, development, and function of healthy brains. Dysregulated lipid metabolism is linked to aging and diseased brains. However, our understanding of lipid metabolism in aging brains remains limited. Here we examined the brain lipidome of mice across their lifespan using untargeted lipidomics. Co-expression network analysis highlighted a progressive decrease in 3-sulfogalactosyl diacylglycerols (SGDGs) and SGDG pathway members, including the potential degradation products lyso-SGDGs. SGDGs show an age-related decline specifically in the central nervous system and are associated with myelination. We also found that an SGDG dramatically suppresses LPS-induced gene expression and release of pro-inflammatory cytokines from macrophages and microglia by acting on the NF-κB pathway. The detection of SGDGs in human and macaque brains establishes their evolutionary conservation. This work enhances interest in SGDGs regarding their roles in aging and inflammatory diseases and highlights the complexity of the brain lipidome and potential biological functions in aging.

Extended Data Fig. 1

Average levels of the co-expressed lipids in the 23 modules across different ages. n = 4 mice/group. Values are expressed with box plots. Boxes represent 25th to 75th percentile. Center lines represent the median values. Whiskers represent 1.5× interquartile range.

Extended Data Fig. 2

Association between module eigenlipid and age. Blue line represents fitted linear regression line between module eigenlipid and age. Pink shaded area represents 95% confidence interval of the regression line. Modules that are significantly associated with age are highlighted with red boxes.

Extended Data Fig. 3

Quantification of SGDGs in the brain of mice with three months of calorie restriction. Sex, male; Age, 6 months. n = 8 mice/group, data are means ± SEM. *P < 0.05 versus ad libitum-fed mice. P values were 0.0217, 0.0155, 0.0276, 0.0453, 0.0410, 0.0243, 0.2387, 0.0694, 0.1417, 0.1103, 0.0423, 0.0451 (left to right). Statistical significance was calculated using unpaired, two-tailed Student’s t-test with equal variance.

Extended Data Fig. 4

Fractionation of the mouse brain lysate into membrane and soluble fractions revealed that SGDGs are largely membrane-associated. Bar graph shows SGDG levels relative to the total lysate. n = 3, data are means ± SEM.

Extended Data Fig. 5

Quantification of SGAAGs in mouse tissues. (a) Quantification of SGAAGs in the brain and spinal cord of WT female mice at different ages. n = 4 mice/group, data are means ± SEM. (b) Quantification of SGAAGs in the brain, spinal cord, and testis of WT male mice at different ages. Data were collected from six 3-month-old, three 7-month-old, and six 22-month-old mice. Data are means ± SEM.

Extended Data Fig. 6

SGDGs are enriched in myelin. (a) Quantification of sulfatides in purified mouse brain myelin. n = 3 mice, data are means ± SEM. (b) Validation of myelin purification. Immunoblot shows that myelin proteins, PLP and MBP, are enriched in purified rat brain myelin compared with the brain lysate, whereas the abundant neuronal protein synaptophysin is depleted. (c) Quantification of SGDGs and sulfatides in purified rat brain myelin. n = 2 technical replicates, data are means ± SEM.

Extended Data Fig. 7

Quantification of SGDGs in oligodendrocyte precursor cells and myelinated oligodendrocytes at day 5 of differentiation.

Extended Data Fig. 8

Quantification of SGDGs in macaque and human brains. (a) Extracted ion chromatograms of SGDG(14:0_16:0) from the parietal cortex of a 6-year-old male macaque and the internal standard SGDG(¹³C₁₆-16:0/14:0). (b) Quantification of SGDGs in the parietal cortex of a 6-year-old male macaque. (c) Quantification of SGDGs in a frontal brain sample collected postmortem from a 5-month-old male subject.

Extended Data Fig. 9

Expression profile of MGDGs, SGAAGs, SGDGs, MGMGs, and lyso-SGDGs in the spinal cord module which shows progressive decrease during aging.

Extended Data Fig. 10

SGDG exhibits anti-Inflammatory effects in RAW 264.7 cells and BV2 cells. (a) RAW 264.7 cells were incubated with media alone, 100 ng/mL of LPS, or co-treated with 100 ng/mL of LPS and 5 μM of SGDG(14:0/16:0) for 4 hours. mRNA levels were determined by RT-PCR. n = 4 replicates. P values were 3.24e-13, 9.54e-11, 3.44e-9, 3.22e-7, 9.35e-11, 5.41e-7, 1.36e-10, 0.00004, 7.10e-10, 0.00001, 2.17e-6, 0.06078 (left to right, top to bottom). (b) BV2 cells were incubated with media alone, 100 ng/mL of LPS, or co-treated with 100 ng/mL of LPS and 25 μM of SGDG(14:0/16:0) or dexamethasone (Dex) for 24 hours. IL-6 and TNF-α levels in the media were measured by ELISA and normalized to cell viability. n = 3 replicates. P values were 0, 0, 0, 0, 0.00005, 0.00038 (left to right). In a, b, data are means ± SEM. *P < 0.05 versus LPS. Statistical significance was calculated using one-way ANOVA followed by post hoc Dunnett’s test to correct for multiple comparisons.

Figure 1.

Global lipidomics of mouse brain during aging. (a) Principal component analysis (PCA) of brain samples from mice at 4, 12, 25, 48, and 78 weeks of age. The PCA analysis was performed with identified lipids. (b) Hierarchical clustering of co-expression modules by module eigenlipid. Modules statistically significantly associated with age at FDR < 0.05 are marked with asterisks (*). Blue, downregulated with age; red, upregulated with age. Darker color indicates larger change with age. Lipid enrichment analyses are shown at the bottom with asterisks (*) corresponding to FDR < 0.05. (c, d) Average levels of the co-expressed lipids in the turquoise (c) and the black (d) modules across different ages. n = 4 mice/group. Box, 25th to 75th percentile; center line, median; whisker, 1.5× interquartile range. (e, f) Expression profile of the identified lipids in the turquoise (e) and the black (f) modules across different ages.

Figure 2.

Discovery of a class of lipids that progressively decrease during aging. (a) Representative MS/MS spectrum of a sulfoglycosyl diacylglycerol and a sulfoglycosyl alkylacylglycerol identified in the mouse brain at HCD 30 V, featuring ions at m/z 96.9587 and m/z 241.0023 corresponding to sulfate and glycosyl sulfate. (b) Expression profile of sulfoglycosyl diacylglycerols and sulfoglycosyl alkylacylglycerols in the black module. (c) Enrichment analysis of lipid classes in the black module. Dashed line, FDR 0.05; node size, number of lipid species of the indicated lipid class. (d) Endogenous sulfoglycosyl diacylglycerol contains a galactose group. SGDG and SGlcDG are synthetic sulfoglycosyl diacylglycerol standards containing galactose and glucose, respectively. (e) The sulfate group is attached to the alcohol at the 3-position on SGDG. Sulfation mix was prepared by loading the sulfate unselectively. Shown in (d) and (e) are extracted ion chromatograms of SGDG(14:0/16:0). Endogenous lipid is crude lipid extract from mouse brain.

Figure 3.

Quantification of SGDGs in mouse tissues. (a, b) Quantification of SGDGs in the brain (a) and spinal cord (b) of WT female mice at different ages. n = 4 mice/group. (c) Quantification of SGDGs in the testis of WT mice. Data were collected from six 3-month-old, three 7-month-old, and six 22-month-old male mice. (d) Quantification of SGDGs in the brain and spinal cord of shiverer mice. Sex, male; Age, 8–9 weeks. n = 4 mice/group. (e) Left panel, validation of myelin purification. Myelin proteins, PLP and MBP, are enriched in purified mouse brain myelin versus the brain lysate, whereas the abundant neuronal protein synaptophysin is depleted. Right panel, quantification of SGDGs in purified mouse brain myelin. n = 3 mice. Data in bar graphs are means ± SEM. In d and e, data were obtained using the untargeted method and ms1-based quantification.

Figure 4.

SGDGs are present in human. Alignment of MS/MS spectra of SGDG(14:0_16:0) from human brain and the heavy-labeled internal standard SGDG(¹³C₁₆-16:0/14:0). The brain tissue was collected from the frontal lobe of a male subject at 5 months of age. The ¹³C-labeled carbons of the standard are indicated by asterisks.

Figure 5.

Lyso-SGDGs are the potential degradation products of SGDGs in the brain. (a) lyso-SGDG(14:0) and lyso-SGDG(16:0) were produced after incubating 100 μM of SGDG(14:0/16:0) with mouse brain lysate at 37 °C for 30 min. As a negative control, SGDG(14:0/16:0) was incubated with heat-denatured brain lysate. The left panel shows lyso-SGDG levels relative to the control. n = 4 replicates, data are means ± SEM. The right panels show MS/MS spectra of lyso-SGDGs at HCD 30 V. (b) Quantification of lyso-SGDGs in the spinal cord of WT female mice. n = 4 mice/group, data are means ± SEM. (c) Average levels of the co-expressed lipids (top panel) and the enrichment analysis of lipid classes (bottom panel) in the spinal cord module which shows progressive decrease during aging. SGDG and SGDG-related lipids are shown in red.

Figure 6.

SGDG exhibits anti-Inflammatory effects via the NF-κB pathway. (a) RAW 264.7 cells were incubated with media alone, 100 ng/mL of LPS, or co-treated with 100 ng/mL of LPS and 5 μM of SGDG(14:0/16:0), MGDG(14:0/16:0), or dexamethasone (Dex) for 24 hours. n = 4 replicates. P values were 0, 8.26e-06, 0, 0, 0, 0.00029, 0, 0 (left to right). (b) BMDMs were incubated with media alone, 100 ng/mL of LPS, or co-treated with 100 ng/mL of LPS and indicated concentrations of SGDG(14:0/16:0), MGDG(14:0/16:0), or Dex for 24 hours. n = 3 replicates. P values were 0, 0.52182, 3.11e-11, 0.01466, 0.00002, 3.45e-11, 5.60e-14, 0.75155, 1.30e-6, 0.02537, 0.00078, 9.20e-6 (left to right). In a and b, IL-6 and TNF-α levels in the media were measured by ELISA and normalized to cell viability determined by MTT assay. (c) BMDMs were incubated with media alone, 100 ng/mL of LPS, or co-treated with 100 ng/mL of LPS and 5 μM of SGDG(14:0/16:0) for 4 hours. mRNA levels were determined by RT-PCR. n = 4 replicates. P values were 2.24e-7, 9.62e-6, 4.54e-9, 8.60e-8, 2.84e-6, 0.00004, 0.00023, 0.02761, 7.85e-10, 0.00002, 2.86e-6, 0.00007, 2.62e-9, 0.00037 (left to right, top to bottom). (d) NF-κB-dependent luciferase activity in HEK293T cells transfected with TLR4 and MD-2 and treated with media alone, 100 ng/mL of LPS, or co-treated with 100 ng/mL of LPS and indicated concentrations of SGDG(14:0/16:0) or MGDG(14:0/16:0) for 6 hours. n = 3 replicates. P values were 9.43e-6, 0.94641, 0.23496, 0.00366 (left to right). In a, b, c, d, data are means ± SEM. *P < 0.05 versus LPS. Statistical significance was calculated using one-way ANOVA followed by post hoc Dunnett’s test to correct for multiple comparisons.