Macrophage-Derived Extracellular Succinate Licenses Neural Stem Cells to Suppress Chronic Neuroinflammation

Abstract

Neural stem cell (NSC) transplantation can influence immune responses and suppress inflammation in the CNS. Metabolites, such as succinate, modulate the phenotype and function of immune cells, but whether and how NSCs are also activated by such immunometabolites to control immunoreactivity and inflammatory responses is unclear. Here, we show that transplanted somatic and directly induced NSCs ameliorate chronic CNS inflammation by reducing succinate levels in the cerebrospinal fluid, thereby decreasing mononuclear phagocyte (MP) infiltration and secondary CNS damage. Inflammatory MPs release succinate, which activates succinate receptor 1 (SUCNR1)/GPR91 on NSCs, leading them to secrete prostaglandin E2 and scavenge extracellular succinate with consequential anti-inflammatory effects. Thus, our work reveals an unexpected role for the succinate-SUCNR1 axis in somatic and directly induced NSCs, which controls the response of stem cells to inflammatory metabolic signals released by type 1 MPs in the chronically inflamed brain.

Keywords: cell metabolism; experimental autoimmune encephalomyelitis; inflammation; macrophages; microglia; multiple sclerosis; neural stem cells; regenerative medicine; stem cells; succinate.

Graphical abstract

Figure 1

NSCs Transplantation Ameliorates Chronic Neuroinflammation and Reduces Succinate Levels in the CSF of EAE Mice (A–D) Representative images of fGFP⁺ iNSCs at 30 dpt expressing the proliferation marker Ki67 (A, arrowheads) and the neural marker Nestin (A), the mature neuronal marker NeuN (B, arrowhead), the astroglial lineage marker GFAP (C), or the oligodendroglial lineage marker OLIG2 (D, arrowhead). (E) Confocal microscopy image of a perivascular area with several fGFP⁺ iNSCs in juxtaposition to fGFP⁻/F4/80⁺ MPs. Nuclei in (A)–(E) are stained with DAPI (blue). (F) Behavioral outcome of iNSCs/NSCs-transplanted EAE mice. Data are mean EAE score (±SEM) from n ≥ 7 mice/group over n = 2 independent experiments. EAE mice injected icv with mouse fibroblasts (MFs) or PBS were used as controls. (G and H) Flow-cytometry-based ex vivo quantification of the expression levels of type 1 inflammatory (CD80) and anti-inflammatory (MRC1) markers in CX3CR1⁺ microglial cells (G) and CCR2⁺ monocyte-derived infiltrating macrophages (H) from the CNS of iNSC- and NSC-transplanted EAE mice at 30 dpt. Quantitative data are shown on the left, whereas representative density plots are shown on the right. Data are min to max % of marker-positive cells from n ≥ 4 pools of mice/group. (I) Representative confocal microscopy image and comparative histograms of a perivascular area with several fGFP⁺ iNSCs in juxtaposition to F4/80⁺ MPs. Low iNOS and prevalent MRC1 expression is detected in F4/80⁺ MPs close to fGFP⁺ iNSCs (inset on the left), whereas high iNOS expression is observed in the remaining MP infiltrate (inset on the right). Nuclei are stained with DAPI. (J) Expression levels (qRT-PCR) of pro- and anti-inflammatory genes in the brain and spinal cord of EAE mice. Data are mean fold change over HC from n ≥ 3 mice/group. (K and L) Quantification and representative 3D reconstructions of spinal cord damage in iNSC- and NSC-transplanted EAE mice. Data are mean % of Bielschowsky negative-stained axonal loss (K) or Luxol fast blue (LFB) negative-stained demyelinated (L) areas/spinal cord section (±SEM) from n ≥ 5 mice/group over n = 2 independent experiments. (M) Levels of CSF metabolites significantly changed during EAE (versus HC). Corresponding levels in matched plasma samples are also shown. Data are mean a.u. (±SEM) from n ≥ 3 mice/group. The scale bars represent 25 μm (A–E), 50 μm (I), and 2 mm (K and L). ^∗p ≤ 0.05 and ^∗∗p ≤ 0.01 versus PBS; ^#p ≤ 0.05 versus HC; dpt, days post-transplantation; FI, fluorescence intensity; HC, healthy controls; PD, peak of disease. See also Figures S1, S2, and S3 and Table S1.

Figure 2

NSCs Reduce Succinate Levels and Reprogram the Metabolism of Type 1 Pro-inflammatory Mφ toward Oxidative Phosphorylation In Vitro (A) Experimental setup for in vitro Mφ^LPS co-cultures with iNSCs/NSCs. (B and C) Gene expression microarrays of Mφ^LPS-iNSCs/NSCs. (B) Venn diagram of differentially expressed genes (adjusted p value < 0.1). (C) Heatmap of genes differentially expressed (adjusted p value < 0.1) in Mφ^LPS-iNSCs or Mφ^LPS-NSCs. (D and E) qRT-PCR independent validation of differentially expressed inflammatory genes as in (C). (D) Expression of genes related to type 1 inflammatory (E) and anti-inflammatory Mφ phenotypes relative to Actb. Data are mean fold change (±SEM) versus Mφ^LPS from n ≥ 3 independent replicates per condition. (F) qRT-PCR of BV2^LPS-iNSCs/NSCs (±SEM) from n ≥ 3 independent experiments per condition. BV2 and BV2^LPS are shown as controls. (G and H) Extracellular flux (XF) assay of the oxygen consumption rate (OCR) (G) and extracellular acidification rate (ECAR) (H) in Mφ^LPS-iNSCs/NSCs. Data were normalized on total protein content and are expressed as mean values (±SEM) from n ≥ 3 independent experiments per condition. (I and J) Levels of significantly changed extracellular (EXTRA_Metab, I) and intracellular (INTRA_Metab, J) metabolites in Mφ^LPS versus Mφ at 25 hr. Data are mean a.u. (±SEM) from n ≥ 2 independent experiments per condition. (K and L) Hif-1α (K), PKM2 (K), and IL-1β (L) expression levels relative to β-actin. Data are mean fold change versus Mφ^LPS (±SEM) from n ≥ 3 independent experiments per condition. ^∗p ≤ 0.05 and ^∗∗p ≤ 0.01 versus Mφ^LPS. See also Tables S2 and S3.

Figure 3

Succinate Signals via SUCNR1 in Mouse and Human NSCs (A–C) Representative confocal microscopy images of meningeal perivascular areas with transplanted fGFP⁺ iNSCs (A) and NSCs (B) expressing SUCNR1 in the brain of a mouse with EAE. The image in (C) shows transplanted SUCNR1⁺ iNSCs in close vicinity to SUCNR1⁺/F4/80⁺ MPs. Nuclei are stained with DAPI. (D) SUCNR1 protein expression relative to β-tubulin in vitro. Data are shown as mean (±SEM) of n ≥ 3 independent replicates per condition. (E) Experimental setup for succinate treatment of iNSCs/NSCs in vitro. (F) Intracellular Ca²⁺ response after treatment with 500 μM succinate (live staining with Fluo-4AM). Representative images (baseline and during stimulation) are pseudocolored with red/blue according to high/low fluorescence intensity. Data are mean changes in fluorescence intensity as ΔF/F0 (±SEM) from n ≥ 3 experiments. (G) Phospho-p38 MAPK (P-p38) and total p38 MAPK (p38) protein expression after succinate treatment. Data are P-p38/p38 expression relative to β-tubulin and expressed as mean fold change (±SEM) versus untreated cells over n ≥ 3 independent experiments per condition. (H) qRT-PCR of SUCNR1 basal expression in human cells. Data are normalized on 18S and expressed as mean fold change (±SEM) versus NSCs from n ≥ 3 independent replicates per condition. (I) Representative blot of SUNCR1 basal protein expression in human cells. (J) P-p38 and p38 protein expression after stimulation with succinate ± pre-treatment with the irreversible inhibitor of the human SUCNR1 4c. The scale bars represent 25 μm. ^∗p ≤ 0.05 versus 0’. hBJFs, human BJ fibroblasts; ND, not detected. See also Figure S4.

Figure 4

SUCNR1 Expression Is Necessary for the Anti-inflammatory Effect of NSCs on Type 1 Pro-inflammatory Mφ In Vitro (A) Heatmap showing the microarray expression profile of the 50 most upregulated genes in NSCs after treatment with succinate. Data are shown as Z scores. (B) qRT-PCR independent validation of Ptgs2 expression as in (A). Data are calculated relative to Actb and shown as mean fold change (±SEM) versus untreated cells over n ≥ 3 independent experiments per condition. (C) PGE2 secretion following 1 hr treatment with succinate ± pre-treatment with the selective PTGS2 blocker SC-58125. Data are mean values (±SEM) over n ≥ 3 independent experiments per condition. (D) PGE2 secretion by hiNSCs treated with succinate ± pre-treatment with either SC-58125 or 4c. Data are mean values (±SEM) over n ≥ 3 independent experiments per condition. (E) PGE2 secretion in Mφ co-cultures. Data are mean values (±SEM) over n ≥ 3 independent experiments per condition. (F) Il1b expression relative to Actb in Mφ co-cultures. Data are mean fold change versus Mφ^LPS (±SEM) from n ≥ 3 independent experiments per condition. (G) XF assay of the OCR of Mφ as in (E) and (F). Data are normalized on total protein content and expressed as mean values (±SEM) over n ≥ 3 independent experiments per condition. (H) Il1b expression relative to Actb of Mφ co-cultures with hiNSCs. Data are mean fold change versus Mφ^LPS (±SEM) from n ≥ 3 independent experiments per condition. (I) XF assay showing the OCR of Mφ as in (H). Data are normalized on total protein content and expressed as mean values (±SEM) over n ≥ 2 independent experiments per condition. (J) PGE2 secretion in Mφ co-cultures as in (H) and (I). Data are mean values (±SEM) over n ≥ 3 independent experiments per condition. ^∗p ≤ 0.05 versus untreated cells (B); ^∗p ≤ 0.05 and ^∗∗p ≤ 0.01 (C and D); ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, and ^∗∗∗p ≤ 0.001 versus Mφ^LPS (E–J); ^#p ≤ 0.05 and ^##p ≤ 0.01 versus Mφ^LPS-NSCs (E–G) or versus Mφ^LPS-hiNSCs (H and I). See also Figure S4 and Table S4.

Figure 5

SUCNR1 Regulates the Uptake of Succinate by NSCs In Vitro (A) SLC13A3 and SLC13A5 protein expression levels after 2 hr of succinate treatment. (B) SLC13A3 and SLC13A5 protein expression levels after 6 hr of succinate treatment in hiNSCs. Data in (A) and (B) are relative to β-actin and expressed as mean fold change (±SEM) versus untreated cells over n ≥ 3 independent experiments per condition. (C and D) Uptake assay of [¹⁴C]-labeled succinate at 0 and 6 hr. (C) Intracellular [¹⁴C] labeling and (D) extracellular [¹⁴C] signal in tissue culture media are shown. Box-whiskers plots ± min to max value from n ≥ 4 technical replicates per group from n = 2 independent experiments are shown. (E) Succinate release in Mφ co-cultures. Data are mean values versus Mφ (±SEM) from n ≥ 2 independent experiments per condition. ^∗p ≤ 0.05 versus untreated cells (A and B) or versus Mφ^LPS (E); ^∗p ≤ 0.05 and ^∗∗p ≤ 0.01 versus 0 hr, Mann-Whitney test (C and D). See also Figure S5.

Figure 6

Transplantation of Sucnr1 Loss-of-Function NSCs Shows Impaired Ability to Ameliorate Chronic Neuroinflammation In Vivo (A) Behavioral outcome of EAE mice. Data are mean EAE score (±SEM) from n ≥ 5 mice/group. (B and C) Flow-cytometry-based ex vivo quantification of the expression levels of type 1 inflammatory (CD80) and anti-inflammatory (MRC1) markers in CX3CR1⁺ microglial cells (B) and CCR2⁺ monocyte-derived infiltrating macrophages (C) at 30 dpt. Quantitative data are shown on the left, whereas representative density plots are shown on the right. Data are min to max % of marker-positive cells from n ≥ 4 pools of mice/group. (D and E) Pathological outcomes of experiments as in (A). Data are mean % Bielschowsky negative-stained axonal loss (D) or LFB negative-stained demyelinated (E) areas/spinal cord section (±SEM) from n ≥ 4 mice/group. The scale bars represent 400 μm. (F) PGE2 levels in the CSF and plasma of EAE mice at 30 dpt. Data are mean values (±SEM) from n ≥ 3 samples/group. (G) Succinate levels in the CSF and plasma of EAE mice at 30 dpt. Data are mean values (±SEM) from n ≥ 4 mice/group. Kruskal-Wallis followed by Mann-Whitney post-test is shown. ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, and ^∗∗∗p ≤ 0.001 versus PBS; ^#p ≤ 0.05 versus NSCs. See also Figure S6.