Activation of the pentose phosphate pathway in macrophages is crucial for granuloma formation in sarcoidosis

Abstract

Sarcoidosis is a disease of unknown etiology in which granulomas form throughout the body and is typically treated with glucocorticoids, but there are no approved steroid-sparing alternatives. Here, we investigated the mechanism of granuloma formation using single-cell RNA-Seq in sarcoidosis patients. We observed that the percentages of triggering receptor expressed on myeloid cells 2-positive (TREM2-positive) macrophages expressing angiotensin-converting enzyme (ACE) and lysozyme, diagnostic makers of sarcoidosis, were increased in cutaneous sarcoidosis granulomas. Macrophages in the sarcoidosis lesion were hypermetabolic, especially in the pentose phosphate pathway (PPP). Expression of the PPP enzymes, such as fructose-1,6-bisphosphatase 1 (FBP1), was elevated in both systemic granuloma lesions and serum of sarcoidosis patients. Granuloma formation was attenuated by the PPP inhibitors in in vitro giant cell and in vivo murine granuloma models. These results suggest that the PPP may be a promising target for developing therapeutics for sarcoidosis.

Keywords: Antigen-presenting cells; Dermatology; Immunology; Macrophages; Skin.

Figure 1. Identification of APC subsets.

(A) UMAP plot showing APCs from 3 sarcoidosis patients and 5 healthy subjects colored by subset. LC, Langerhans cells; cDC1, conventional type 1 dendritic cells; cDC2, conventional type 2 dendritic cells; mregDC, mature dendritic cells enriched in immunoregulatory molecules; resident, skin resident macrophages; TREM2, TREM2 macrophages. (B) Heatmap showing marker genes for each subset. Representative genes are labeled. (C) UMAP plot split by healthy and sarcoidosis samples. (D) An abundance of each APC subset. Šídák’s multiple-comparisons test was conducted between the healthy and sarcoidosis samples for each subset. *P < 0.05. (E) Volcano plot comparing TREM2 macrophages against other APCs. MAC, macrophages. (F) Violin plots showing expression of ACE, LYZ, CD163, TREM2, FBP1, FN1, APOC1, and APOE in APC subsets in all skin samples. (G) Representative immunofluorescence staining in healthy (n = 3) and sarcoidosis (n = 3) skin biopsy samples for expression of CD68 in gray, FBP1 in red, CD163 in green, and DAPI in blue. Scale bars: 100 μm.

Figure 2. FBP1 is expressed in macrophages in association with cutaneous and noncutaneous sarcoidosis.

(A and B) Representative immunofluorescence staining (A) and bar graph (B) in extradermal lesions (lung, n = 3; heart, n = 1; lymph node, n = 3) of sarcoidosis samples for expression of CD68 in gray, FBP1 in red, CD163 in green, and DAPI in blue. Scale bars: 100 μm. (C) Frequency of ACE⁺ and FBP1⁺ APCs in inflammatory skin diseases based on previously published data (–18). AD, atopic dermatitis; PSO, psoriasis; GA, granuloma annulare. *P < 0.05, Dunnett’s multiple-comparisons test.

Figure 3. FBP1 is expressed in macrophages in association with cutaneous granulomatous diseases.

(A and B) Representative immunofluorescence staining (A) and bar graph (B) in healthy and inflammatory skin disease samples (n = 3) for expression of CD68 in gray, FBP1 in red, CD163 in green, and DAPI in blue. Scale bars: 100 μm. *P < 0.05, Dunnett’s multiple-comparisons test. LMDF, lupus miliaris disseminatus faciei. (C and D) Box-and-whisker plots (C) and receiver operating characteristic curves (D) of serum FBP1 and ACE concentrations in healthy subjects (n = 32) and sarcoidosis patients (n = 14). Mann-Whitney U tests were conducted between healthy and sarcoidosis samples. *P < 0.05.

Figure 4. Metabolically active TREM2 macrophages express genes in association with giant cells.

(A) UMAP plots for all skin scRNA-Seq data (5 healthy subjects and 3 patients) showing macrophage subsets. (B) An abundance of each macrophage subset in all skin scRNA-Seq data (5 healthy subjects and 3 patients). Šídák’s multiple-comparisons test was conducted between the healthy and sarcoidosis samples for each subset. *P < 0.05. (C) Dot plot of genes upregulated in M1and M2 macrophages in macrophage fractions of scRNA-Seq data (5 healthy subjects and 3 patients). (D) Heatmap showing marker genes for each subset in all skin scRNA-Seq data (5 healthy subjects and 3 patients). Representative genes are labeled. (E) Heatmap showing metabolic pathway expression for each cell subset in all skin scRNA-Seq data (5 healthy subjects and 3 patients) via Reactome pathway database. (F and G) Representative H&E and immunofluorescence (F) staining and bar graph (G) in sarcoidosis skin samples (n = 3) for expression of CD68 (gray), FBP1 (red), CHI3L1 (green), and DAPI (blue). Dotted lines indicate giant cells, and arrowheads point to FBP1⁺CHI3L1⁺ cells. Scale bars: 50 μm. Mann-Whitney U tests were conducted between nongiant and giant cells. *P < 0.05.

Figure 5. FBP1-positive macrophages switch to PPP as the dominant route for glucose metabolism.

(A) Schematic diagram of glucose metabolized by the glycolytic system or via the PPP. (B) Heatmap showing metabolomic changes for each macrophage subset in all skin scRNA-Seq data (5 healthy subjects and 3 patients). Mono, monocytes. (C) Representative immunofluorescence staining in healthy (n = 3) and sarcoidosis samples (lung, n = 3; heart, n = 1; lymph node, n = 3) for expression of G6PD in gray, FBP1 in red, CD163 in green, and DAPI in blue. Scale bars: 100 μm. (D) H&E-stained sections and imaging for MS analysis of NADPH in the skin of sarcoidosis patients. Dotted lines indicate giant cells Scale bars: 500 μm (left); 100 μm (right).

Figure 6. Treatment of in vitro giant cell model by inhibition of the PPP.

(A) Fold induction of FBP1 and G6PD mRNA in Con A, IFN-γ, and anti-CD40 antibodies stimulated (stim) monocytes, as analyzed by quantitative PCR (n = 3). (B) Measurement of intracellular NADP and NADPH levels in the in vitro giant cell model (n = 4). (C) Schematic diagram of metabolic pathways and inhibitors of glucose and NADPH metabolized by the glycolytic system or by the PPP. (D and E) Giemsa staining (D) and bar graphs (E) of monocytes stimulated with Con A, IFN-γ, and anti-CD40 antibodies, with or without DMSO, DPI, FBPi, and 6AN (n = 4). (F and G) Giemsa staining (F) and giant cell count (G) of 3 day cultures stimulated with Con A, IFN-γ, and anti-CD40 antibodies followed by incubation with DMSO, FBPi, or 6AN for 3 days (n = 3–5). (H and I) Giemsa staining (H) and bar graphs (I) of monocytes stimulated with Con A, IFN-γ, and anti-CD40 antibodies, with or without DMSO, 6AN, or G6PDi (n = 5). Data are representative of at least 3 independent experiments. Scale bars: 100 μm. (A and B) *P < 0.05, unpaired t test. (E, G, and I) *P < 0.05, Dunnett’s multiple-comparisons test.

Figure 7. Metabolic analysis of in vitro giant cell model.

(A) Giemsa staining (left) and bar graphs (right) for monocytes stimulated with Con A, IFN-γ, and anti-CD40 antibodies, with or without vehicle, 2-DG, and etomoxir (n = 5). Scale bars: 100 μm. Data are represented as mean ± SD. *P < 0.05, Dunnett’s multiple-comparisons test. Data are representative of at least 3 independent experiments. (B–D) Bar graph of analysis by MS of NADPH (B) and metabolites associated with PPP (C) and ATP (D) 1 hour after addition of ¹³C₆ glucose to a giant cell model cultured for 3 days (n = 3). Data are represented as mean ± SD. *P < 0.05, unpaired t test. D, DMSO; 6, 6AN; F, FBPi.

Figure 8. PPP inhibition reduces in vivo granuloma formation.

(A and B) Representative photographs (A) and bar graph (B) of H&E-stained sections of ear skin from DMSO-, 6AN-, and FBPi-treated mice with Con A and bead injection (n = 6). Red lines indicate granulomas, and dotted lines indicate beads. Scale bars: 100 μm. (C) Time course of ear-swelling response (n = 6). (D and E) Dot plot and bar graph of flow cytometric analysis of the number of macrophages in the whole ear skin before and 72 hours after treatment (n = 4). (F and G) Representative photographs (F) and bar graphs (G) of H&E-stained sections of mouse ear skin (n = 6). Two days after subcutaneous injection of Con A and beads into mouse ear, DMSO, 6AN, or FBPi was administered intraperitoneally. Red lines indicate granulomas. Scale bars: 500 μm. (H) Time course of ear-swelling response (n = 6). Data are represented as mean ± SD. *P < 0.05, Dunnett’s multiple-comparisons test. Data are representative of at least 3 independent experiments.