Spatial mapping of the AA-PGE2-EP axis in multiple sclerosis lesions

Abstract

Bioactive lipid mediators (LMs) derived from polyunsaturated fatty acids (PUFAs) are key molecules in both the initiation and resolution of inflammatory responses. Previous findings suggest that a dysregulated LM balance, especially within the arachidonic acid (AA) pathway, may contribute to an impaired resolution response and subsequent chronic neuroinflammation in multiple sclerosis (MS). However, to date, the local biosynthesis and signaling of LMs within the brain of people with MS (PwMS) remains unexplored. In this study, we, therefore, mapped the distribution of AA and its key downstream LM prostaglandin E₂ (PGE₂) in white matter MS brain tissue and of non-neurological controls (NNCs) for the first time using mass spectrometry imaging. We found that AA levels are lower in MS cases compared to NNCs and reduced in MS lesions compared to peri-lesional tissue. Furthermore, the PGE₂/AA ratio, indicating the PGE₂ synthesis from the AA substrate, was increased in lesion areas compared to fully myelinated regions in MS. In line with that, the expression of prostaglandin synthesizing enzymes as measured by RT-qPCR was partially increased in MS tissue compared to NNCs. In addition, the expression of prostaglandin E2 receptor 4 (EP4) decreased, while prostaglandin E2 receptor 2 (EP2) showed increased expression levels in MS lesions compared to NNCs and localized specifically to microglia. We also found that PGE₂ addition to pro-inflammatory human-induced pluripotent stem cell (iPSC)-derived microglia resulted in enhanced cytokine signaling pathways, but also the upregulation of its synthase PTGES and homeostatic/resolving signaling, the latter of which might mainly occur through EP2 signaling. Collectively, our results provide detailed information about the region-specific levels of AA and PGE₂ in MS lesions and we propose enhanced PGE₂-EP2 signaling in inflamed microglia in MS.

Keywords: Arachidonic acid; Mass spectrometry imaging; Microglia; Multiple sclerosis; Neuroinflammation; Prostaglandin E2.

Fig. 1

Broad and in-depth tissue classification of NNC and MS human brain tissue. a Study sampling of three NNCs and seven MS human post-mortem brain tissues. b ROIs were defined by differences in neuropathology, assessed by PLP and HLA-DR reactivity. Within the MS tissues, broad tissue classes encompass a representative ROI for a lesion, adjacent lesion rim and distant peri-lesional tissue; scale bar: 5 mm. For in-depth quantification, ROIs were stratified by PLP reactivity (A full, B partial and C absent/demyelinated), and HLA-DR density and morphology (1: ramified, low density and inactive lesion; 2: ramified, high density, mixed active/inactive lesion; 3: rounded, ameboid, high density and active lesion); scale bar: 50 µm c Immunohistochemical depiction of PLP and HLA-DR in MS tissues (color-coded) and annotated ROIs for broad and in-depth tissue classification; scale bar: 5 mm.

Fig. 2

Decreased AA levels in WM MS lesions. a Ion images of AA [M + 107Ag] + (m/z 411.1448 ± 5 ppm) normalized to internal standard AA-d8 in human NNC (n = 3) and MS brain tissues (n = 7). The spatial distribution of AA is visualized by a min–max intensity scale within each ion image; hence, the images are not related to each other. White arrows point out exemplary demyelinated/lesion areas; red arrows indicate myelinated/peri-lesional tissue. Pixel size is 25 µm × 150 µm; scale bar: 5 mm. b AA signal intensities from the entire tissue section were extracted and normalized to the internal standard AA-d8 to yield the average detected concentration per pixel (µM/pixel). Each dot represents the tissue of one donor. c Paired analysis of AA within MS tissues based on broad classification (one representative ROI per tissue area). Each colored dot represents the tissue of one donor. d Paired analysis of AA levels based on in-depth classification based on PLP or HLA-DR (averages of multiple ROIs per class). Each colored dot represents the tissue of one donor. e mRNA expression of various biosynthesizing enzymes in human brain tissue block lysates of NNC (N = 8) and MS tissues (N = 12). Tissues used both in MSI and qPCR are marked in red. f Representative images of COX2, CD45 (immune cell marker), and Collagen IV (Coll IV, vascular marker) immunoreactivity in MS WM tissue; scale bar: 50 µm, zoom in: 25 µm. g Quantification of COX2 mean fluorescent intensity (MI) measured within the nuclei and percentage of COX2⁺ cells in NNC (N = 4) and MS lesions (N = 6). h Paired tissue analysis of COX2⁺ cells in MS tissues (HLA-DR in-depth) and Spearman correlation (r_s) of detected AA concentration with COX2⁺ cells (HLA-DR in-depth). Data is shown as box plots with median ± quartiles; whiskers extend to minimum and maximum. Data have been statistically tested for three groups by Friedman test (paired) for non-normally distributed data and Dunn’s post hoc analysis. For two groups, an unpaired student t-test with Welch’s correction was used when the variance of the groups was significantly different or the Mann–Whitney test for non-parametric datasets. Exact p-values are reported and statistical significance is set at p < 0.05 (red)

Fig.3

PGE₂/AA levels are increased in demyelinated WM MS tissue. a Ion images of PGE₂ (first row) and PGE₂/AA concentration ratio images (second row). PGE₂ ion images were constructed by visualizing m/z 459.1301 ± 5 ppm ([M + ¹⁰⁷Ag]⁺) normalized pixel-by-pixel to internal standard PGE₂-d9 in human NNC (n = 3) and MS lesions (n = 7). The spatial distribution of PGE₂ is visualized by a min–max intensity scale within each single ion image; hence, the images are not related to each other. In contrast, all the PGE₂/AA concentration ratio images are fitted to the same scale and can be compared. White arrows point out exemplary lesion areas; red arrows indicate peri-lesional tissue with full myelination; scale bar: 5 mm. b PGE₂ signal intensities were extracted as average pixel intensity per tissue and normalized to the internal standard of PGE₂ to yield average detected concentration per pixel (µM/pixel). Each dot presents one donor. c Paired analysis of PGE₂ concentration and PGE₂/AA concentration ratio based on broad classification (one representative ROI) within MS tissues. Each colored dot represents the tissue of one donor. d, e Paired analysis of PGE₂ and PGE₂/AA ratio based on PLP or HLA-DR classification (in-depth classes). Each colored dot represents the tissue of one donor. f Representative images of HLA-DR immunoreactivity in MS WM tissue; scale bar: 50 µm, zoom in: 10 µm. g Spearman correlation (r_s) of PGE₂/AA ratio values with HLA-DR MI in present, partial, and absent PLP tissue categories. Data is shown as box plots with median ± quartiles; whiskers extend to minimum and maximum. Data have been statistically tested for three groups by Friedman test (paired) for non-normally distributed data and Dunn’s post-hoc analysis. Exact p-values are reported and statistical significance is set at p < 0.05 (red)

Fig. 4

Increased microglial EP2 expression in MS lesions. a Schematic overview of PGE₂ receptors EP1-4 and their G-protein coupled signaling pathways. EP2 (blue) and EP4 (yellow) are highlighted. b Representative cropped immunoblots of EP2, EP4 (upper panel each), and β-actin (lower panel) from human brain tissue homogenates of NNC and MS lesions. c Densitometric quantification of EP2 and EP4, normalized to β-actin in NNC (N = 9) and MS lesions (N = 14). d Representative images of EP2 (cyan), Iba1 (magenta) and TMEM119 (yellow) immunoreactivity in WM of NNC and mixed A/I MS lesions (peri-lesion and lesion). Panels show outlined excerpts at higher magnification; scale bar: 50 µm. e Quantification of EP2⁺ cells and EP2 mean fluorescent intensity within microglia (Iba1⁺TMEM119⁺ cells) in NNC and MS tissues (N = 5). f mRNA expression of PTGER2 (encoding EP2) and EP2 protein were measured in human iPSC-derived microglia (hiPSC microglia) non-stimulated (resting) or stimulated with LPS + IFNγ for 24 h (pro-inflam). Data is shown as box plots with median ± quartiles; whiskers extend to minimum and maximum. Data have been statistically tested for three groups by ordinary one-way ANOVA with Dunnett’s correction or Kruskal–Wallis test for non-normally distributed data and Dunn’s post hoc analysis. For two groups, an unpaired student t-test with Welch’s correction was used when the variance of the groups was significantly different or the Mann–Whitney test for non-parametric datasets. Exact p-values are reported and statistical significance is set at p < 0.05 (red)

Fig. 5

PGE₂ signaling adds to immunological response in microglia. a Schematic overview of PGE₂ and EP inhibitor treatment strategy on resting and pro-inflammatory hiPSC microglia. b Venn diagram of significantly differentially expressed genes (DEG) presenting upregulated (top) and downregulated genes (bottom) in resting (green) vs. pro-inflammatory (blue) microglia treated with PGE₂ vs. vehicle; N_TR = 5. c Volcano plot representing significant DEGs comparing PGE₂ treatment versus vehicle in resting microglia with respect to -log₁₀ p adjusted in the y-axis and log₂ fold change in the x-axis. d Volcano plot representing significant DEGs comparing PGE₂ treatment vs. vehicle in pro-inflammatory microglia. e Volcano plot visualizing DEGs regulated by EP2i + PGE₂ vs EP4i + PGE₂ in pro-inflammatory microglia. f mRNA expression of CXCR4, PTGES, TREM2, CCL13, and PAK1 in pro-inflammatory microglia treated with PGE₂ + EP inhibitors; N = 3. g Over-representation analysis of enriched KEGG pathways in pro-inflammatory iPSC microglia treated with PGE₂ + EP2i vs. PGE₂ + EP4i. Data is shown as box plots with median ± quartiles; whiskers extend to minimum and maximum. Data have been statistically tested for four groups by Friedman test for non-normally distributed data and Dunn’s post hoc analysis. Exact p-values are reported and statistical significance is set at p < 0.05 (red)