Dynamic Brain Lipid Profiles Modulate Microglial Lipid Droplet Accumulation and Inflammation Under Ischemic Conditions in Mice

Abstract

Microglia are critically involved in post-stroke inflammation affecting neurological outcomes. Lipid droplet (LD) accumulation in microglia results in a dysfunctional and pro-inflammatory state in the aged brain and worsens the outcome of neuroinflammatory and neurodegenerative diseases. However, the role of LD-rich microglia (LDRM) under stroke conditions is unknown. Using in vitro and in vivo stroke models, herein accumulation patterns of microglial LD and their corresponding microglial inflammatory signaling cascades are studied. Interactions between temporal and spatial dynamics of lipid profiles and microglial phenotypes in different post-stroke brain regions are found. Hence, microglia display enhanced levels of LD accumulation and elevated perilipin 2 (PLIN2) expression patterns when exposed to hypoxia or stroke. Such LDRM exhibit high levels of TNF-α, IL-6, and IL-1β as well as a pro-inflammatory phenotype and differentially expressed lipid metabolism-related genes. These post-ischemic alterations result in distinct lipid profiles with spatial and temporal dynamics, especially with regard to cholesteryl ester and triacylglycerol levels, further exacerbating post-ischemic inflammation. The present study sheds new light on the dynamic changes of brain lipid profiles and aggregation patterns of LD in microglia exposed to ischemia, demonstrating a mutual mechanism between microglial phenotype and function, which contributes to progression of brain injury.

Keywords: cholesterol metabolism; ischemic stroke; lipid droplets; lipid metabolism; lipidomics; microglia; neuroinflammation.

Figure 1

OGD, LPS, and CM induce accumulation of LD in primary microglia in vitro. a) Immunofluorescence staining of LD (BODIPY + green) accumulation in microglia after exposure to OGD. b,c) Quantification of the percentage of BODIPY + Iba1+ cells of total Iba1 + cells and BODIPY mean fluorescence intensity (MFI) (n = 5). d) Immunofluorescence staining of LD (BODIPY + green) accumulation in microglia after OGD according to the RO time (Normoxia, 2, 6, 12, 24, 48 h). e,f) Quantification of the percentage of BODIPY+ cells of total cells and BODIPY MFI (n = 5). g) Accumulation of LD in microglia with treatment of LPS or with CM from primary neurons after OGD (OGD‐CM) can be inhibited by Trc (1 µM, ACAT inhibitor). h,i) Quantification of the percentage of BODIPY+Iba1+ cells and MFI of BODIPY in five different conditions (n = 5). j,k) Quantitative analysis of PLIN2 and IL‐1β expression of microglia in five different conditions: normoxia as control, OGD, OGD‐CM, LPS, and IL‐4 using western blot analysis normalized with the housekeeping protein GAPDH (n = 3). Statistical tests: Two‐tailed t‐tests were used (b, c). One‐way ANOVA followed by Tukey's post‐hoc‐tests were used (e, f, h, i, j, k). Data are expressed as mean ± SD, NS: no significance, ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001, ^**** p < 0.0001, ^### p < 0.001, and ^#### p < 0.0001. Scale bars, 50 µm in (a, d, g). Ctr, Control group; OGD, oxygen‐glucose deprivation; RO, reoxygenation; OGD‐CM, conditioned medium from primary neuron after OGD; LD, lipid droplet; MFI, mean fluorescence intensity; Trc, triacsin C; ACAT, Acyl‐CoA cholesterol acyltransferase; PLIN2, perilipin 2; IL‐1β, interleukin 1 beta; IL‐4, interleukin 4; LPS, lipopolysaccharide.

Figure 2

The formation of LD in microglia in the mouse middle cerebral artery occlusion (MCAO) stroke model. a–d) Immunofluorescence of cortical infarct core staining at different reperfusion times (sham, 3, 7, 28 days) after MCAO indicated that LD mainly accumulated in microglia. e,f) Quantification of the percentage of BODIPY+ Iba1+ cells and mean fluorescence intensity (MFI) of BODIPY (n = 5). g) Quantitative analysis of CD45intCD11b+BODIPY+ microglia in three groups (sham, ipsilateral, and contralateral side of post‐ischemia day 7) by flow cytometry (n = 7). h,i) Density plots of FACS showed a significant increase of LD‐rich microglia (CD45intCD11b+BODIPY+) within the MCAO lesion site group (h), which matched the BODIPY fluorescence intensity (i). j,k) Quantitative analysis of PLIN2 and IL‐1β expression in MCAO mice with different reperfusion times (sham, 1, 3, 5, 7, 9, 11, and 14 days) by Western blot analysis of the ischemic hemisphere. Western blot was normalized with the housekeeping protein GAPDH (n = 3). Statistical tests: One‐way ANOVA followed by Tukey's post‐hoc‐tests were used for (e‐g, j, k). Data are expressed as mean ± SD, NS: no significance, ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001, ^**** p < 0.0001, ^# p < 0.05, ^## p < 0.01, and ^#### p < 0.0001. Scale bars, 1000 µm (i) and 50 µm (ii) in (a, b, c, d). MCAO, middle cerebral artery occlusion; LD, lipid droplet; MFI, mean fluorescence intensity; FACS, Fluorescence‐activated cell sorting assay; PLIN2, perilipin 2; IL‐1β, interleukin 1 beta.

Figure 3

Primary microglia (PM) with different treatments affect the cell viability of primary neurons (PNs) after OGD in a co‐culture system. a) Schematic diagram of the PM/PNs co‐culture system. b) Cell viability was analyzed in PNs exposed to 4 h of OGD followed by 24 h of reoxygenation with co‐cultures of pre‐treated PM by MTT assay (n = 5). Cells incubated under normoxic conditions were defined as 100% cell survival. c) OGD‐induced neuronal cell toxicity was further assessed in the lactate dehydrogenase (LDH) release assay (n = 5). d,e) The LIVE/DEAD assay displays representative immunofluorescence stainings of calcein AM (LIVE cells, green) and ethidium homodimer‐1 (DEAD cells, red), followed by cell viability analysis via the percentage of LIVE cells to total cells (n = 5). The assay used the same conditions for PM and PNs as mentioned for the MTT assay. f,g) Immunofluorescence stainings and quantitative analysis of apoptotic cell (TUNEL, red) rate in PNs (NeuN, green) by TUNEL assay in the four groups: normoxia, OGD, OGD followed by co‐culture with OGD‐CM pre‐treated microglia, and OGD followed by co‐culture with OGD‐CM+Trc pre‐treated microglia (n = 5). Statistical tests: One‐way ANOVA followed by Tukey's post‐hoc‐tests were used for (b, c, e, g). Data are expressed as mean ± SD, NS: no significance, ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001, ^**** p < 0.0001, ^# p < 0.05, ^## p < 0.01, ^### p < 0.001, and ^#### p < 0.0001. Scale bars, 20 µm in (d, f). OGD, oxygen‐glucose deprivation; RO, reoxygenation; PM, primary microglia; PN, primary neuron; OGD‐CM, conditioned medium from primary neuron after OGD; LD, lipid droplet; IF, Immunofluorescence staining; WB, Western blot assay; MFI, mean fluorescence intensity; Trc, triacsin C; IL‐4, interleukin 4; LPS, lipopolysaccharide; Trc, triacsin C, LDH, lactate dehydrogenase.

Figure 4

Lipid droplet‐rich microglia (LDRM) exhibit distinct lipid metabolism‐related gene expression as well as phenotype polarization and inflammatory factor levels after stimulation with LPS or OGD‐conditioned medium (OGD‐CM). a–d) Quantitative measurement of pro‐inflammatory factors (IL‐1β, TNF‐α, and IL‐6) and anti‐inflammatory factors (TGF‐β1) in microglia treated with LPS or OGD‐CM (w/ or w/o Trc) followed by different incubation periods (2, 6, 12, 24, and 48 h) using ELISA assay (n = 5). e) Heatmap of gene expression comparisons between six groups: control, LPS, LPS+Trc; OGD‐CM, OGD‐CM+Trc, and IL‐4. The heatmap was produced by pheatmap package in Rstudio. Data were log2 transformed and shown as red to blue: red (up‐regulated), blue (down‐regulated) and white (no change). Rows and columns are clustered using correlation distances and average associations. f Immunofluorescence co‐staining of LD (BODIPY, green) and M1‐like (iNOS, red) polarization of microglia in four different treatments: Control, OGD‐CM, OGD‐CM+Trc, IL‐4. g) Quantitative analysis of BODIPY+iNOS+ microglia of total cells (n = 5). h) Immunofluorescence co‐staining of LD (BODIPY, green) and M2‐like (CD206, red) polarization of microglia in the aforementioned four groups. i) Quantitative analysis of BODIPY+ microglia of total cells (n = 5). j–l) Quantitative analysis of PLIN2 (j), SREBP2 (k), and p65 (l) expression in six groups: normoxia as control group, OGD‐CM, LPS, LPS+Trc, OGD‐CM+Trc and IL‐4 group using Western blot analysis normalized with the housekeeping protein GAPDH (n = 3). Statistical tests: Two‐way ANOVA followed by Tukey's post‐hoc‐tests were used for (a‐d). One‐way ANOVA followed by Tukey's post‐hoc‐tests were used for (g, I, j‐l). Data are expressed as mean ± SD, NS: no significance, ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001, ^**** p < 0.0*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ^# p < 0.05, ^## p < 0.01, ^### p < 0.001, and ^#### p < 0.0001. Scale bars, 20 µm (f, h). Ctr, Control group; OGD, oxygen‐glucose deprivation; RO, reoxygenation; OGD‐CM, conditioned medium from primary neuron after OGD; LD, lipid droplet; LDRM, lipid droplet‐rich microglia; Trc, triacsin C; PLIN2, perilipin 2; IL‐1β, interleukin 1 beta; IL‐4, interleukin 4; SREBP2, sterol regulatory element‐binding protein 2.

Figure 5

Ischemia‐induced up‐regulation of SREBP2 and PLIN2 leads to the formation of pro‐inflammatory lipid droplet‐rich microglia (LDRM) and the activation of the NF‐κB pathway. a–d) Quantitative measurement of IL‐1β (a), TNF‐α (b), IL‐6 (c) and TGF‐β1 (d) levels in the six regions: lesion core of cortex (L Core), ipsilateral cortex out of lesion (L Cortex), white matter area of lesion edge (L Edge), and corresponding areas on the contralateral non‐lesion side (NL Core, NL Cortex, and NL Edge) in the brain (sham, post‐ischemia 3, 7, and 28 d) using ELISA assay (n = 6). e–k) Quantitative analysis of SREBP2 (f), PLIN2 (g), p65 (h), IκBα (i), IL‐1β (j), and TGF‐β1 (k) expression in the aforementioned six regions with 3 days post‐ischemic mice by Western blot analysis normalized with the housekeeping protein (n = 3). i–r) Quantitative analysis of SREBP2 (m), PLIN2 (n), p65 (o), IκBα (p), IL‐1β (q), and TGF‐β1 (r) expression in the aforementioned six regions with 7 days post‐ischemic mice by Western blot analysis normalized with the housekeeping protein (n = 3). Statistical tests: Two‐way ANOVA followed by Tukey's post‐hoc‐tests were used for (a–d). One‐way ANOVA followed by Tukey's post‐hoc‐test was used for (f–k, m–r). Data are expressed as mean ± SD, NS: no significance, ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001, ^**** p < 0.0001, ^# p < 0.05, ^## p < 0.01, ^### p < 0.001, and ^#### p < 0.0001. L Core, lesion core of cortex; L Cortex, ipsilateral cortex out of lesion; L Edge, white matter area of lesion edge; NL Core, contralateral non‐lesion core of cortex; NL Cortex, contralateral non‐lesion cortex out of lesion; NL Edge, white matter area of contralateral non‐lesion edge; LD, lipid droplet; LDRM, lipid droplet‐rich microglia; PLIN2, perilipin 2; IL‐1β, interleukin 1 beta; IL‐6, interleukin 6; SREBP2, sterol regulatory element‐binding protein 2.

Figure 6

Phenotype patterns of PLIN2+ resident microglia in different lesion areas of the post‐stroke brain. a) Immunofluorescence staining of M1‐like microglia (iNOS, red) and LD surface protein marker (PLIN2, green) at post‐ischemia days 3 and 7 in the aforementioned six regions. b) Quantitative analysis of mean fluorescence intensity (MFI) of PLIN2 (n = 5). c) Quantitative analysis of the percentage of PLIN2+iNOS+ cells of total iNOS+ cells (n = 5). c Immunofluorescence staining of polarization of M1‐like microglia (iNOS, red) and M2‐like microglia (CD206, green) at post‐ischemia days 3 and 7 in the aforementioned six regions. d) Quantitative analysis of the number of iNOS+ and CD206+ cells per 0.2 mm² area (n = 5). Statistical tests: Two‐way ANOVA followed by Tukey's post‐hoc‐tests were used for (b, c, e, f). Data are expressed as mean ± SD, NS: no significance, ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001, ^**** p < 0.0001, ^# p < 0.05, ^## p < 0.01, ^### p < 0.001 and ^#### p < 0.0001. Scale bars, 50 µm (a, d). MCAO, middle cerebral artery occlusion; LD, lipid droplet; PLIN2, perilipin 2; L Core, lesion core of cortex; L Cortex, ipsilateral cortex out of lesion; L Edge, white matter area of lesion edge; NL Core, contralateral non‐lesion core of cortex; NL Cortex, contralateral non‐lesion cortex out of lesion; NL Edge, white matter area of contralateral non‐lesion edge; MFI, mean fluorescence intensity.

Figure 7

Lipid‐class composition of distinct mouse brain regions after stroke. a) Principal Component Analysis (PCA) of lipid composition patterns in different regions of brains (sham, post‐ischemia 3 days and 7) to identify the most important variables for lipid components. b) Heatmap for Core regions representing color‐coded Z scores of lipid‐class distribution of the 24 most abundant lipid classes using the lipid species concentrations (pmol) with log2 normalization. c) Volcano map of lipid profiles in sham core and 3 days L Core. d) Orthogonal partial least squares discriminant (OPLS‐DA) analysis of lipid profiles with VIP score in Sham Core and 3 days L Core. e) Volcano map of lipid profiles in 3 days L Core and 7 days L Core. f) OPLS‐DA analysis of lipid profiles with VIP score in 3 days L Core and 7 days L Core. Source data are provided as a source data file. PCA, Principal component analysis; OPLS‐DA, orthogonal partial least squares discriminant analysis; PLIN2, perilipin 2; L Core, lesion core of cortex; L Cortex, ipsilateral cortex out of lesion; L Edge, white matter area of lesion edge; CE, cholesteryl ester; Cer, ceramide; CL, cardiolipin; SM, sphingomyelin; DAG, diacylglycerol; TAG, triacylglycerol; PA, phosphatidate; PC, phosphatidylcholine; PC O‐, ether‐linked phosphatidyl‐choline; PE, phosphatidylethanolamine; PE O‐, ether‐linked phosphatidyl‐ethanolamine; PS, phosphatidyl‐serine; PI, phosphatidylinositol; PG, phosphatidylglycerol; LPA, lyso‐phosphatidate; LPC, lyso‐phosphatidyl‐choline; LPC O‐, ether‐linked lyso‐phosphatidyl‐choline; LPE, lyso‐phosphatidyl‐ethanolamine; LPE O‐, ether‐linked lyso‐phosphatidyl‐ethanolamine; LPI, lyso‐phosphatidyl‐inositol; LPG, lyso‐phosphatidyl‐glycerol; LPS, lyso‐phosphatidyl‐serine; Sulf, sulfatide; HexCer, hexosylceramide.

Figure 8

The connection between enrichment of lipid species and inflammation after stroke. a–i) The Radar plots display the logarithmic mean concentrations of lipids sorted by classes: cholesteryl ester (a), ceramide (b), cardiolipin (c), sphingomyelin (d), diacylglycerol (e), triacylglycerol (f), sulfatide (g), hexosylceramide (h), and phosphatidylglycerol (i). Each color represents different groups. The scale of the plot ranges from 0% to the highest concentration observed within each lipid class (100%). For lipid classes with over 50 sublipids, only the highest abundances are shown for clarity. The color scheme applied to the lipid subspecies names indicates significance based on the modified t‐test of the volcano plot (blue = p < 0.05 for comparison of Sham core versus 3 days L core; red = p < 0.05 for comparison of 3 days L core versus 7 days L core). j Dynamic changes of genes related to inflammation in different areas after MCAO (n = 6). k Dynamic changes of genes related to lipid metabolism in different areas after MCAO (n = 6). Data are expressed as mean ± SD, NS: no significance, ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001, ^**** p < 0.0001, ^# p < 0.05, ^## p < 0.01, ^### p < 0.001 and ^#### p < 0.0001. In each bar plot, p‐values < 0.05, VIP‐score > 1. CE, cholesteryl ester; Cer, ceramide; CL, cardiolipin; SM, sphingomyelin; DAG, diacylglycerol; TAG, triacylglycerol; Sulf, sulfatide; HexCer, hexosylceramide; PG, phosphatidylglycerol; PLIN2, perilipin 2; IL‐1β, interleukin 1 beta; L Core, lesion core of cortex; L Cortex, ipsilateral cortex out of lesion; L Edge, white matter area of lesion edge.