ARTICLE

https://doi.org/10.1038/s41467-020-15636-8 OPEN Glycogen metabolism regulates macrophage- mediated acute inflammatory responses

Jingwei Ma1,6, Keke Wei2,6, Junwei Liu3,6, Ke Tang2, Huafeng Zhang2, Liyan Zhu2, Jie Chen3, Fei Li1, Pingwei Xu2, Jie Chen2, Jincheng Liu2, Haiqing Fang2, Liang Tang1, Dianheng Wang2, Liping Zeng2, Weiwei Sun2, Jing Xie4,5, ✉ Yuying Liu4,5 & Bo Huang 1,2,4,5

Our current understanding of how sugar metabolism affects inflammatory pathways in

1234567890():,; macrophages is incomplete. Here, we show that glycogen metabolism is an important event that controls macrophage-mediated inflammatory responses. IFN-γ/LPS treatment stimu- lates macrophages to synthesize glycogen, which is then channeled through glycogenolysis to generate G6P and further through the pentose phosphate pathway to yield abundant NADPH, ensuring high levels of reduced glutathione for inflammatory macrophage survival.

Meanwhile, glycogen metabolism also increases UDPG levels and the receptor P2Y14 in

macrophages. The UDPG/P2Y14 signaling pathway not only upregulates the expression of STAT1 via activating RARβ but also promotes STAT1 phosphorylation by downregulating phosphatase TC45. Blockade of this glycogen metabolic pathway disrupts acute inflamma- tory responses in multiple mouse models. Glycogen metabolism also regulates inflammatory responses in patients with sepsis. These findings show that glycogen metabolism in mac- rophages is an important regulator and indicate strategies that might be used to treat acute inflammatory diseases.

1 Department of Immunology, Tongji Medical College, Huazhong University of Science & Technology, Wuhan 430030, China. 2 Department of Biochemistry & Molecular Biology, Tongji Medical College, Huazhong University of Science & Technology, Wuhan 430030, China. 3 Cardiovascular Surgery, Union Hospital, Huazhong University of Science & Technology, Wuhan 430071, China. 4 Department of Immunology & National Key Laboratory of Medical Molecular Biology, Institute of Basic Medical Sciences, Chinese Academy of Medical Sciences (CAMS) & Peking Union Medical College, Beijing 100005, China. 5 Clinical Immunology Center, CAMS, Beijing 100005, China. 6These authors contributed equally: Jingwei Ma, Keke Wei, Junwei Liu. ✉ email: [email protected]

NATURE COMMUNICATIONS | (2020) 11:1769 | https://doi.org/10.1038/s41467-020-15636-8 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15636-8

acrophages can be polarized to an inflammatory phe- cultured medium of macrophages (Fig. 1f). The 13C carbon tra- notype by interferon-γ (IFN-γ) and lipopolysaccharide cing showed a striking increase of m + 6 G6P/G1P, concomitant M fl + fl (LPS) or to an anti-in ammatory phenotype by with higher levels of m 6 UDPG in the in ammatory macro- interleukin-4 (IL-4) or other factors. Although inflammatory phages, compared to controls (Fig. 1g, h), indicating a potential macrophages play a crucial role in phagocytosis and eliminating presence of an active glycogen synthesis in the inflammatory bacteria, uncontrolled activation of inflammatory macrophages macrophages. To validate the above murine data in human may trigger systemic inflammatory response syndrome (SIRS) or macrophages, we additionally repeated the experiments in human even sepsis1–3. Therefore, precise regulation of inflammatory monocytic THP-1 cells. Similarly, IFN-γ/LPS-treated THP-1 cells macrophages is of paramount importance in guaranteeing displayed much higher glycogen levels, concomitant with the microbial clearance, injury limitation, and avoidance of serious upregulation of enzymes involved in glycogen synthesis (Sup- side effects4,5. It is known that inflammatory phenotype of plementary Fig. 1a–c). Moreover, we cultured bone marrow cells macrophages is profoundly regulated by glucose metabolism6–8, in the 13C-glucose culture medium and induced them to differ- however this metabolic regulation remains incompletely under- entiate to macrophages for 5 days. The cells were then stimulated stood. In addition to glycolysis as the mark for inflammatory with IFN-γ/LPS for 24 h, followed by the treatment with hydro- metabolic phenotype9,10, macrophages have recently been shown chloric acid to degrade polymer glycogen into monomer glu- to utilize glucose metabolic reprograms such as pentose phos- cose21. The released 13C-labeled glucose was determined by LC- phate pathway (PPP) and mitochondrial succinate oxidation to MS/MS, which showed a result of 70% 13C-labeled glucose drive their inflammatory phenotype11–13. But how the glucose (Supplementary Fig. 1d). These results suggest that inflammatory metabolic cues trigger inflammatory gene expression in macro- macrophages use glycogen as a metabolic way. phages remains unclear. In addition, whether and how other Glycogen synthesis is initiated from the transition of G6P to glucose-related metabolic pathway(s) regulates inflammatory G1P. Several pathways can be the source of G6P, including macrophages has not been well studied. gluconeogenesis, glycogenolysis, and glycolysis22,23. Previously, Glycogen is an important carbohydrate fuel reserve and gly- we have shown that CD8+ memory T cells use gluconeogenesis to cogen metabolism has been previously reported in myeloid cells generate G6P17. However, this seemed not to be the case in of the immune system including dendritic cells and macro- macrophages, because (1) the 13C-glucose tracing did not show phages14–16. Our recent studies additionally found that CD8+ m + 3 G6P/G1P to reflect gluconeogenesis (Fig. 1i); (2) 13C- memory T cells actively mobilize glycogen metabolism to gen- pyruvate or 13C-acetate tracing did not show m + 2 G6P/G1P erate glucose-6-phosphate (G6P) via glycogenolysis and channel (Supplementary Fig. 1e, f); and (3) key gluconeogenic enzymes the G6P to the PPP17. In this metabolic model, the glycogen is not phosphoenolpyruvate carboxykinase 1 (Pck1) and fructose-1,6- primarily used to store energy, but to provide antioxidant defense bisphosphatase (Fbp1) were not upregulated; instead they were through the generation of NADPH and subsequently reduced slightly downregulated (Fig. 1j, k). In addition, to avoid a futile glutathione. These findings imply that glycogen may exert metabolism, G6P should not be derived from glycogenolysis for important biological effects through metabolic pathways beyond glycogen biosynthesis. Thus, for inflammatory macrophages, G6P its role as a reservoir of glucose. In support of this notion, UDP- was likely derived from the glycolysis. Indeed, the macrophages glucose (UDPG), a glycogen metabolic intermediate, has been consumed more glucose, upregulated glucose transporter Slc2a1/2 shown to act as a ligand for purigenic receptor P2Y14, thus and enzyme hexokinase (Hk1/2/3), leading to the increased directly triggering signal transduction18–20. production of G6P (Fig. 1l, m). By contrast, knocking down Hk1/ In the present study, we provide evidence that glycogen 2/3 to inhibit glycolysis-derived G6P reduced the glycogen levels metabolism has a central function in controlling inflammatory in inflammatory macrophages (Fig. 1n and Supplementary pathways of macrophages by two related mechanisms: firstly, Fig. 1g). Also, the knockdown of Pgm1 or Gys1 resulted in the macrophages use glycolysis-derived G6P to synthesize glycogen, decreased glycogen levels in inflammatory macrophages (Fig. 1o and subsequent glycogenolysis regenerates G6P which is chan- and Supplementary Fig. 1g). Together, these data suggest that neled through the PPP to produce large amounts of NADPH inflammatory macrophages mobilize glycolysis-derived G6P to required for inflammatory macrophage survival; secondly, initiate glycogen synthesis. glycogenesis-derived UDPG activates the P2Y14 receptor, whose signaling transduction regulates the inflammatory phenotypes of macrophages in an autocrine fashion. Glycogenolysis-derived G6P is channeled to the PPP. Synthe- sized glycogen is stored in the cytoplasm or enters glycogenolysis for degradation24. Notably, glycogen-degrading enzymes such as Results glycogen phosphorylase Pygl (liver) and Pygm (muscle) were Glycogen is synthesized in inflammatory macrophages. To test found to be upregulated in IFN-γ/LPS-treated rather than the possible role of glycogen in inflammatory macrophages, we untreated or IL-4-treated macrophages (Fig. 2a, b). Consistent first determined whether glycogen was synthesized in the cells. results were also obtained from IFN-γ/LPS-treated human THP-1 We cultured murine bone marrow-derived untreated, IFN-γ/LPS- cells (Supplementary Fig. 2a, b), implying that inflammatory treated inflammatory and IL-4 treated anti-inflammatory mac- macrophages have glycogenolytic activity, leading to G6P pro- rophages, respectively. We found that glycogen was strikingly duction. In addition, we roughly calculated the glycogen turnover increased in the inflammatory macrophages, as evidenced by PAS rate, which was around 52% (Supplementary Fig. 2c). As a central staining and colorimetric assay (Fig. 1a, b), which was further metabolite, G6P can be channeled to different directions: confirmed by transmission electron microscope (TEM) (Fig. 1c). becoming glucose via dephosphorylation; being oxidized to pyr- In line with these results, enzymes involved in glycogen bio- uvate along glycolysis or to ribose-5-phosphate (R5P) via synthesis, including phosphoglucomutase 1 (Pgm1), UDP- PPP22,23. The 13C tracing showed that G6P could be channeled to glucose pyrophosphorylase 2 (Ugp2, the catalytic enzyme con- m + 5 R5P (Fig. 2c), which was blocked by glycogen phosphor- verting G1P to UDPG) and glycogen synthase 1 (Gys1), were ylase inhibitor (GPI), Gys1 or Pygl siRNA (Fig. 2d), suggesting upregulated in inflammatory macrophages and only Ugp2 was that glycogenolysis-derived G6P is channeled through the PPP. slightly increased a lesser extent by IL-4 (Fig. 1d, e). To further Consistently, two enzymes G6P dehydrogenase (G6pdx) and 6- confirm the above result, we added [U6]-13C-glucose to the phosphogluconate dehydrogenase (6Pgd) that mediate the

2 NATURE COMMUNICATIONS | (2020) 11:1769 | https://doi.org/10.1038/s41467-020-15636-8 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15636-8 ARTICLE

acUN IFN-γ/LPS IL-4 b UN IFN-γ/LPS IL-4 p < 0.0001 40 p < 0.0001 30

protein) 20 –1 g μ

Glycogen 10

(ng 0

UN IL-4 γ/LPS

IFN-

d p < 0.0001 p < 0.0001 p < 0.0001 e f 13C-Glucose p < 0.0001 p < 0.0001 p < 0.0001 4 6 15 kDa 3 61 Pgm1 m + 6 4 10 6AN 2 57 Ugp2 HK 2 mRNA (fold) mRNA (fold)

mRNA (fold) 5 1 84 Gys1 G1P Pgm1 G6P G6pdx 6PGL 0 0 0 43 β-actin Gys1 Ugp2 Pgm1 Ugp2 m + 6 m + 6 UN IL-4 UN IL-4 UN IL-4 γ/LPS γ/LPS γ/LPS UN IL-4 Pygl 6PG γ/LPS IFN- IFN- IFN- UDPG IFN- F6P Gys1 GPI Glycogen R5P g 13C-glucose h 13C-glucose i 13C-glucose G6P/G1P UDPG G6P/G1P m + 3 m + 5

1.2 p = 0.0016 0.8 p = 0.0002 p = 0.031 j 0.12 p < 0.0001 p = 0.0018 p = 0.0001 p = 0.015 4 UN S7P m + 0 0.6 m + 0 ) p = 0.01 γ –5 IFN- /LPS Pyruvate 0.8 m + 6 m + 6 3 0.08 Actb IL-4 m + 7 0.4 Glycolysis 0.4 0.04 2 0.2 E4P 1 ns ns 0 0 0 (×10 mRNA ns Relative abundance Relative abundance Relative abundance ns Relative to 0 m + 4 UN IL-4 UN IL-4 UN IL-4 γ/LPS γ/LPS γ/LPS Pck1 Fbp1 G6pase PPP IFN- IFN- IFN-

k lm n o γ kDa p < 0.0001 12 UN IFN- /LPS IL-4 40 1.5 p < 0.0001 **** 40 63 Pck1 **** **** **** **** 30 **** **** 30 = 0.002 p 1.0 8 = 0.002 = 0.001 39 Fbp1 **** p p protein) **** 20 protein) **** = 0.0004 –1 –1 **** **** 20 p g 40 G6pase 0.5 **** g μ cells per 24 h) 4 **** **** μ Glycogen 6 10 **** Glycogen β 10

43 -actin (ng (ng 0 Relative mRNA 0 0 0

UN Glucose consumption UN Slc2a1 Slc2a2 Hk1Hk2 Hk3 NC IL-4 IL-4 NC γ/LPS γ/LPS (mg per 10 siHk1#1siHk1#2siHk2#1siHk2#2siHk3#1siHk3#2 IFN- IFN- siPgm1#1siPgm1#2siGys1#1siGys1#2

Fig. 1 Glycogen is synthesized in inflammatory macrophages. a–c Intracellular glycogen levels in untreated, IFN-γ/LPS (20 ng mL−1 IFN-γ plus 100 ng mL−1 LPS) or IL-4 (10 ng mL−1) treated BMDMs were detected and observed by PAS staining (scale bar, 20 μm) (a), colorimetric assay (b) and TEM (scale bar, 0.5 μm) (c). The arrows point to intracellular glycogen deposits. d, e Pgm1, Ugp2, and Gys1 expression in untreated, IFN-γ/LPS or IL-4 treated BMDMs were determined by real-time PCR (d) and western blot (e). f Overview of three glucose metabolic pathways: glycogen metabolism (left), glycolysis (middle) and PPP (right) and the inhibitors (GPI/6AN) are shown. g–i BMDMs differentiated in normal 12C-glucose were stimulated with IFN-γ/LPS or IL-4 for 6 h and switched to 13C-glucose for 6 h, LC-MS/MS was performed for m + 6-labeled G6P/G1P (g), m + 6-labeled UDPG (h) and m + 3-labeled G6P/ G1P (i). j, k Pck1, Fbp1, and G6pase expression in untreated, IFN-γ/LPS or IL-4 treated BMDMs were determined by real-time PCR (j) and western blot (k). l Consumption of glucose in untreated, IFN-γ/LPS or IL-4 treated BMDMs were measured by enzymatic methods. m Relative mRNA expression of Slc2a1/2 and Hk1/2/3 in untreated, IFN-γ/LPS or IL-4 treated BMDMs were determined by real-time PCR. n, o Hk1/2/3, Pgm1 or Gys1 siRNA transfected BMDMs were stimulated with IFN-γ/LPS for 36 h. Intracellular glycogen levels were detected by colorimetric assay. Unless otherwise specified, n = 3 biologically independent experiments were performed. Data are presented as mean ± SEM. P values were calculated using one-way ANOVA, ****p < 0.0001. oxidation of PPP were upregulated in inflammatory macrophages Blocking glycogen synthesis by Ugp2 or Gys1 siRNA or blocking (Fig. 2e, f). Blocking PPP by G6pdx siRNA or G6pdx inhibitor 6- glycogenolysis by Pygl siRNA led to decreased S7P and E4P in aminonicotinamide (6AN) or blocking glycogenolysis by Pygl inflammatory macrophages (Supplementary Fig. 2f), suggesting siRNA or GPI led to accumulation of glycogen in inflammatory that glycogenolysis-derived G6P is channeled through the PPP in macrophages (Fig. 2g and Supplementary Fig. 2d, e). The PPP can inflammatory macrophages. Here, we also clarified how much be divided into oxidative and non-oxidative steps: G6P is first G6P was derived from glucose taken up by the macrophages oxidized to an intermediate molecule ribulose 5-phosphate versus how much G6P was generated from glycogenolysis. Bone (Ru5P); for the non-oxidative step, Ru5P is either converted to marrow cells were cultured with [U6]-13C-glucose medium for R5P for nucleotide synthesis25, or converted to R5P and xylulose 5 days in the presence of M-CSF, followed by 6-hour stimulation 5-phosphate (X5P), leading to the generation of intermediate with IFN-γ/LPS or IFN-γ/LPS + GPI and the switch of the products [sedoheptulose 7-phosphate (S7P) and erythrose 4- medium to 13C-glucose-free medium for the last 2- or 4 h. Cell phosphate (E4P)] and end products [glyceraldehyde 3-phosphate lysates were then analyzed by LC-MS/MS. Based on such m + 6 (G3P) and fructose 6-phosphate (F6P)]26. In line with the carbon G6P tracing, we calculated that 83.08% vs. 1.77% G6P at 2 h and flow from G6P to R5P, the 13C tracing assay further showed that 94.03% vs. 3.18% G6P at 4 h were generated by glycolysis vs. G6P could be channeled to m + 7 S7P and m + 4 E4P (Fig. 2h). glycogenolysis (Fig. 2i, j). In addition, we found that blockade of

NATURE COMMUNICATIONS | (2020) 11:1769 | https://doi.org/10.1038/s41467-020-15636-8 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15636-8

a b cd13 13 C-glucose C-glucose p = 0.03 p < 0.0001 p < 0.0001 R5P R5P p = 0.004 p < 0.0001 p < 0.0001 8 10 p = 0.001 p = 0.0002 kDa BMDM p = 0.0006 p = 0.0006 8 1.2 p = 0.0001 1.2 6 p = 0.03 97 Pygl 1.2 c and d 6 4 0.8 0.8 m + 0 4 97 Pygm 0.8 m + 5 mRNA (fold) mRNA (fold) 0.4 2 2 0.4 43 β-actin 0.4 Pygl 0 Pygm 0 0 0 Relative abundance 0 UN IL-4 Relative abundance 02050 Relative abundance UN IL-4 UN IL-4 /LPS UN IL-4 NC γ/LPS γ/LPS γ γ/LPS GPI (μM) IFN- IFN- IFN- IFN- siGys1#1siGys1#2siPygl#1siPygl#2

p < 0.0001 e p < 0.0001 f g h 13 13 p < 0.0001 p = 0.002 C-glucose C-glucose 8 10 p < 0.0001 p < 0.0001 kDa BMDM p = 0.0002 S7P E4P 100 p = 0.0009 p = 0.0025 6 8 p < 0.0001 58 G6pdx 1.2 0.4 p = 0.0009 6 75 p < 0.0001 m + 0 4 m + 5 4 46 6Pgd 0.3 mRNA (fold)

protein) 0.8 mRNA (fold) 50 m + 7 2 –1 2 β g 0.2 m + 4

43 -actin μ Glycogen 25 0.4 6Pgd

G6pdx 0 0 0.1 (ng UN UN UN IL-4 0 0 0 IL-4 IL-4 γ/LPS γ/LPS γ/LPS Relative abundance Relative abundance NC UN IL-4 UN IL-4 IFN- IFN- IFN- γ/LPS γ/LPS siPygl#1siPygl#2 siG6pdx#1siG6pdx#2 IFN- IFN-

i p = 0.0012 1.00 p = 0.0038 j 2 h 4 h 0.95 13C-glucose Glycolysis derived G6P/G1P 0.90 G6P/G1P Glycogenolysis derived G6P/G1P m + 0 Others 0.85 m + 6 Glycolysis Glycogenolysis 0.80 2 h 83.08%±0.73% 1.77%±0.20% Relative abundance 0 24 24 (h) 4 h 94.03%±0.40% 3.18%±0.18% IFN-γ/LPS IFN-γ/LPS + GPI

Fig. 2 Glycogenolysis-derived G6P is channeled to the PPP. a, b Pygl and Pygm expression in untreated, IFN-γ/LPS or IL-4 treated BMDMs were determined by real-time PCR (a) and western blot (b). c BMDMs differentiated in normal 12C-glucose were stimulated with IFN-γ/LPS or IL-4 for 6 h and switched to 13C-glucose for 6 h, LC-MS/MS was performed for m + 5-labeled R5P. d BMDMs were pretreated with GPI for 30 min or pretransfected with siRNA (Gys1 or Pygl) for 24 h prior to stimulation with IFN-γ/LPS for 6 h and switched to 13C-glucose for 6 h, LC-MS/MS was performed for m + 5-labeled R5P. e, f G6pdx and 6Pgd expression in untreated, IFN-γ/LPS or IL-4 treated BMDMs were determined by real-time PCR (e) and western blot (f). g Pygl or G6pdx siRNA transfected BMDMs were stimulated with IFN-γ/LPS for 36 h. Intracellular glycogen levels were detected by colorimetric assay. h BMDMs differentiated in normal 12C-glucose were stimulated with IFN-γ/LPS or IL-4 for 6 h and switched to 13C-glucose for 6 h, LC-MS/MS was performed for m + 7-labeled S7P, m + 5-labeled S7P and m + 4-labeled E4P. i, j BMDMs cultured and differentiated in 13C-glucose were pretreated with GPI for 30 min prior to stimulation with IFN-γ/LPS for 6 h and switched to 12C-glucose for 2 or 4 h, 13C-labeled G6P/G1P were detected by LC-MS/MS (i). The ratio of G6P/G1P from glycogenolysis or glycolysis was calculated. The glycogenolysis-derived G6P using the format of [m + 6 G6P (IFN-γ/LPS)–m + 6 G6P (IFN- γ/LPS + GPI)]/Total G6P (IFN-γ/LPS) and the glycolysis-derived G6P by the format of [m + 0 G6P (IFN-γ/LPS)] / Total G6P (IFN-γ/LPS) (j). Unless otherwise specified, n = 3 biologically independent experiments were performed. Data are presented as mean ± SEM. P values were calculated using one- way ANOVA. glycogenolysis by GPI led to the increase of 13C-labeled glucose in ratio and increased ROS levels (Fig. 3f–h and Supplementary glycogen from 70 to 84% and the decrease of m + 5 R5P from Fig. 3d, e). In line with increased ROS levels, lactate dehy- 95% to 84% (Supplementary Fig. 2g). This 14% increase was some drogenase (LDH) was strikingly present in the supernatants, consistent with 11% decrease, suggesting that glycogenolysis- reflecting a state of cell death (Fig. 3i). This release of LDH, derived G6P might flow to PPP. however, could be inhibited by the supply of exogenous GSH (Fig. 3j). Notably, disruption of either PPP or glycogenolysis by siRNA or inhibitor resulted in the downregulation of the PPP regulates macrophage phenotype, function, and survival. expression of iNOS, TNF, IL-6 and Il1b in IFN-γ/LPS-treated fi One biological signi cance of the PPP lies in the generation of macrophages (Fig. 3k-m and Supplementary Fig. 3f, g), con- NADPH, which is crucial for the reduction of oxidized glu- comitant with decreased ability to clear bacteria (Fig. 3n). tathione to GSH, thus maintaining the redox homeostasis of Together, these data suggest that the glycogen-PPP metabolic 27–29 cells . As expected, higher levels of NADPH were found in pathway regulates the phenotype, function and survival of γ IFN- /LPS-treated macrophages (Fig. 3a), concomitant with the inflammatory macrophages. higher ratio of GSH/GSSG relative to that in control macrophages (Fig. 3b). Blocking the PPP by either G6pdx siRNA or 6AN effectively decreased NADPH levels and the GSH/GSSG ratio UDPG regulates inflammatory macrophages via P2Y14 recep- (Fig. 3c, d and Supplementary Fig. 3a, b), but increased ROS tor. Although the above results showed that the inhibition of levels in the macrophages (Fig. 3e and Supplementary Fig. 3c). glycogenolysis downregulated the expression of proinflammatory Consistently, blocking glycogen metabolism by Gys1 or Pygl genes in IFN-γ/LPS-treated macrophages, blocking glycogen siRNA or GPI also led to decreased NADPH levels, GSH/GSSG synthesis at different stages produced conflicting results. Blocking

4 NATURE COMMUNICATIONS | (2020) 11:1769 | https://doi.org/10.1038/s41467-020-15636-8 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15636-8 ARTICLE

p = 0.002 a bcp = 0.002 d p < 0.0001 ef p = 0.006 g p < 0.0001 p < 0.0001 p < 0.0001 p = 0.005 1.5 30 p = 0.0497 1.5 30 p < 0.0001 3 p = 0.0001 + + p < 0.0001 p < 0.0001 ) p < 0.0001 p = 0.03 p = 0.017 4 + 1.5 p = 0.0004 p = 0.001 1.0 20 1.0 20 2 30 1.0 20 0.5 10 0.5 10 1 0.5 10 GSH/GSSG GSH/GSSG ROS (MFI×10 NADPH/NADP NADPH/NADP GSH/GSSG 0 0 0 0 0 NADPH/NADP 0 0 2 2 1 2 #1 # #1 # # # UN IL-4 UN IL-4 NC NC NC NC NC γ/LPS γ/LPS siGys1#1siGys1#2siPygl#1siPygl#2 siGys1#1siGys1#2siPygl#1siPygl#2 IFN- IFN- siG6pdxsiG6pdx siG6pdxsiG6pdx siG6pdxsiG6pdx

h p < 0.0001 ij p < 0.0001 NC siRNA#1 siRNA#2 UN UN p < 0.0001 20 μ 60 μ ) p = 0.0003 GPI 20 M 6AN 20 M 4 3 20 ns p < 0.0001 p < 0.0001 GPI 50 μM 6AN 50 μM p < 0.0001 p < 0.0001 ns p = 0.042 p < 0.0001 15 p = 0.0133 2 15 p < 0.0001 p = 0.0004 40 p = 0.0017 10 ns p = 0.0002 1 10 p < 0.0001 p = 0.0057 p = 0.0005 20 ns 5 5 p < 0.0001 p < 0.0001

ROS (MFI×10 0 p < 0.0001 p = 0.0003 LDH release (%) LDH release (%) 0 0 LDH release (%) 0 NC Gys1 Pygl G6pdx GSH 01 51020(mM)GSH 01 51020 (mM) siGys1#1siGys1#2siPygl#1siPygl#2 k 0.15 0.06 0.04 0.20 Nos2 Tnf Il6 Il1b = 0.0001 < 0.0001 p = 0.0002 < 0.0001 = 0.008 < 0.0001 p < 0.0001 < 0.0001 p p = 0.002 < 0.0001 p Actb Actb Actb Actb p < 0.0001

p 0.03 0.15 p = 0.005 p p < 0.0001 p < 0.0001 = 0.009 p < 0.0001

0.10 0.04 p p p p 0.02 0.10 0.05 0.02 0.01 0.05 Relative to Relative to Relative to Relative to 0 0 0 0 UN IFN-γ/LPS UN IFN-γ/LPS UN IFN-γ/LPS UN IFN-γ/LPS

l UN IFN-γ/LPS m kDa 600 600 k and m ) ) = 0.003 –1 –1 = 0.004 130 iNOS p NC p = 0.0004 = 0.004 = 0.0005 p p = 0.002 p = 0.0002

400 400 p < 0.0001

p siPygl#1 β p 43 -actin siPygl#2 200 200 siG6pdx#1 #2 #1 #2

NC NC IL-6 (pg mL TNF (pg mL siG6pdx#2 siPygl#1siPygl#2 siPygl#1siPygl#2 0 0 siG6pdx#1siG6pdx siG6pdxsiG6pdx UN IFN-γ/LPS UN IFN-γ/LPS

n Colony-forming Units NC siPygl#1 siPygl#2 siG6pdx#1 siG6pdx#2 010050 150

NC p < 0.0001 siPygl#1 p < 0.0001 siPygl#2 siG6pdx#1 p < 0.0001 p < 0.0001 siG6pdx#2

Fig. 3 PPP regulates macrophage phenotype, function and survival. a, b NADPH/NADP+ (a) and GSH/GSSG (b) ratio in untreated, IFN-γ/LPS or IL-4 treated BMDMs were analyzed. c–e G6pdx siRNA transfected BMDMs were stimulated with IFN-γ/LPS for 24 h, NADPH/NADP+ (c), GSH/GSSG (d) and ROS (e) were analyzed. MFI, mean fluorescence intensity. f–h Gys1 or Pygl siRNA transfected BMDMs were stimulated with IFN-γ/LPS for 24 h, NADPH/ NADP+ (f), GSH/GSSG (g) and ROS (h) were analyzed. i Gys1, Pygl or G6pdx siRNA transfected BMDMs were stimulated with IFN-γ/LPS for 24 h, lactic dehydrogenase (LDH) release was analyzed. j BMDMs were pretreated with GPI or 6AN for 30 min prior to stimulation with IFN-γ/LPS ± GSH (0, 1, 5, 10, 20 mM) for 36 h, and LDH release was analyzed. k–m Pygl or G6pdx siRNA transfected BMDMs were stimulated with or without IFN-γ/LPS for 24 or 36 h, Nos2, Tnf, Il6, and Il1b expression was determined by real-time PCR (k), iNOS, TNF and IL-6 expression was determined by western blot (l) and ELISA (m). n Pygl or G6pdx siRNA transfected BMDMs were stimulated with IFN-γ/LPS for 24 h and incubated with E.coli at ratio 1:1 for 3 h, then the E.coli was collected, coated and cultured with ampicillin-resistant agarose solid medium at 37 °C for 36 h, followed by colony-forming units count. Data are presented as mean ± SEM of n = 3 biologically independent experiments (a–d, f, g, j, k, and m)orn = 4 biologically independent experiments (e, h, i and n). P values were calculated using one-way ANOVA. the conversion of G6P to G1P by Pgm1 siRNA or G1P to UDPG Supplementary Fig. 4a), suggesting that glycogenesis-derived by Ugp2 siRNA resulted in the downregulation of the above- UDPG regulates the inflammatory phenotype of macrophages. In mentioned proinflammatory genes (Fig. 4a, b); unexpectedly, addition, UDPG also rescued the downregulated inflammatory those genes, however, were upregulated by knocking down Gys1, genes in the G6pdx siRNA group (Fig. 4g). Despite the increase of the enzyme transferring the glucose group from UDPG to gly- UDPG levels by the inhibition of glycogen synthesis, blocking cogen (Fig. 4c). This apparent paradox can likely be ascribed to glycogenolysis or PPP however resulted in decreased UDPG the metabolite molecule UDPG in that UDPG levels were levels, as evidenced by the ELISA detection and the 13C carbon increased by Gys1 siRNA but decreased by Pgm1 or Ugp2 siRNA tracing (Fig. 4h, i and Supplementary Fig. 4b). Intriguingly, we (Fig. 4d). Moreover, the supplement of exogenous UDPG not found that the inhibition of glycogen synthesis led to upregula- only upregulated those inflammatory genes expression in tion expression of Pgm1 and Ugp2, however the inhibition of inflammatory macrophages (Fig. 4e), but also rescued the glycogenolysis or PPP led to downregulation expression of Pgm1 downregulated genes in the Ugp2 siRNA group (Fig. 4f and and Ugp2 (Supplementary Fig. 4c, d), thus explaining the

NATURE COMMUNICATIONS | (2020) 11:1769 | https://doi.org/10.1038/s41467-020-15636-8 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15636-8

a b c d

NC siPgm1#1 siPgm1#2 NC siUgp2#1 siUgp2#2 NC siGys1#1 siGys1#2 200 **** cells) 1.5 p < 0.0001 p < 0.0001 p < 0.0001 1.5 p < 0.0001 p < 0.0001 4 6 **** p < 0.0001 p < 0.0001 p < 0.0001 p = 0.001 150 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 3 p = 0.02 p = 0.0002 p < 0.0001 p = 0.0002 1.0 1.0 100 p = 0.0003 p = 0.0003 p < 0.0001 2 50 ** 0.5 0.5 **** **** 1 0 **** Relative mRNA Relative mRNA Relative mRNA

0 0 0 UDPG (pg per 10 NC Nos2 Tnf Il6 Il1b Nos2 Tnf Il6 Il1b Nos2 Tnf Il6 Il1b siGys1#1siGys1#2siPgm1#1siPgm1#2siUgp2#1siUgp2#2

ehi 13C-glucose UN UDPG 100 UDPG 200 (μM)

) UDPG m + 6 9 p < 0.0001 800 p = 0.0003 p = 0.007 80 μ –1

UDPG ( M) cells)

p = 0.0049 p = 0.048 6 < 0.0001 p p < 0.0001 60 < 0.0001 10 < 0.0001

600 p p 6 p < 0.0001 p < 0.0001 kDa 0 100 200 < 0.0001 p < 0.0001 p < 0.0001 p 8 p = 0.0007 40 = 0.0016

400 = 0.0003 p

p = 0.004 p 130 iNOS < 0.0001 p 6 < 0.0001

3 20 p 200 4

Relative mRNA 43 β-actin 0

(×100000) 2

0 Cytokine (pg mL 0 NC Abundance (AU) Nos2 Tnf Il6 Il1b TNF IL-6 UDPG (pg per 10 0

siPygl#1siPygl#2 NC siG6pdx#1siG6pdx#2 siPygl#1siPygl#2 siG6pdx#1siG6pdx#2 f NC kDa j siUgp2 130 iNOS p = 0.0008 p < 0.0001 k

) p = 0.003 3 siUgp2 + UDPG 800 8 p < 0.0001 p < 0.0001 –1 p = 0.046 p = 0.009 kDa p = 0.0099 43 β-actin 600 6 p < 0.0001 2 p = 0.0002 p < 0.0001 38 P2Y14 p = 0.0003 4 p = 0.0003 p < 0.0001 400 NC 1 mRNA (fold) 2 43 β-actin

siUgp2 200 14

Relative mRNA 0 Cytokine (pg mL 0 0 P2ry UN IL-4 siUgp2 + UDPG TNF IL-6 γ/LPS Nos2 Tnf Il6 Il1b UN IL-4 γ/LPS IFN- IFN- g l m NC kDa siG6pdx 130 iNOS p = 0.0003 p < 0.0001 3 ) 800 p = 0.0001 6 siG6pdx + UDPG –1 μ p < 0.0001 p = 0.006 p = 0.023 UDPG ( M) 43 β-actin 600 4 2 p < 0.0001 p < 0.0001 p = 0.01 p = 0.007 p = 0.0002 kDa 0 100 200 p = 0.001 p = 0.0003 p = 0.0005 400 2 14 1 NC mRNA (fold) 38 P2Y 200 14 siG6pdx 0 β Relative mRNA 43 -actin P2ry

0 Cytokine (pg mL 0 0 100 200 Nos2 Tnf Il6 Il1b TNF IL-6 siG6pdx + UDPG UDPG(μM)

n p = 0.01 p = 0.01 p = 0.0002 p = 0.004 0.15 p = 0.0038 0.06 0.04 p = 0.008 0.20 p < 0.0001 p = 0.0004 NC 0.03 0.15 Actb 0.10 Nos2 Actb 0.04 TnfActb Il6Actb Il1b siP2ry14#1 0.02 0.10 siP2ry14#2 0.05 0.02 0.01 0.05

Relative to 0 Relative to 0 Relative to 0 Relative to 0 UN IFN-γ/LPS UN IFN-γ/LPS UN IFN-γ/LPS UN IFN-γ/LPS o p UN IFN-γ/LPS 600 p = 0.004 600 p = 0.0002

kDa ) p = 0.002 ) p = 0.0002

–1 NC 130 iNOS –1 400 400 siP2ry14#1 β 43 -actin siP2ry14#2 200 200 IL-6 (pg mL NC #1 #2 NC #1 #2 TNF (pg mL 14 14 14 14 0 0 γ γ siP2rysiP2ry siP2rysiP2ry UN IFN- /LPS UN IFN- /LPS

– inconsistence. It is known that UDPG could be released from P2Y14 receptor antagonist PPTN (Supplementary Fig. 4g i). In Golgi to the extracellular space through a secretory pathway30,31, addition, the addition of UDPG did not upregulate the expression fl where UDPG bound to the receptor P2Y14 for signal transduc- of P2Y14, STAT1 and pro-in ammatory cytokines in IL-4- 18–20,31,32 – tion . Notably, the expression of P2Y14 receptor was stimulated macrophages (Supplementary Fig. 4j l). Together, upregulated in IFN-γ/LPS-treated macrophages (Fig. 4j, k), and these results suggest that UDPG regulates inflammatory macro- further enhanced by exogenous UDPG addition (Fig. 4l, m). By phage phenotype via the P2Y14 receptor. contrast, the expression of P2ry14 was downregulated in Pygl or Ugp2 siRNA-treated inflammatory macrophages (Supplementary Fig. 4e). Moreover, when we used siRNA to knock down P2Y14 UDPG/P2Y14 regulates STAT1 expression and phosphoryla- receptor (Supplementary Fig. 4f), we found that those inflam- tion. Next, we investigated how the inflammatory macrophage γ matory genes were downregulated in IFN- /LPS-treated macro- phenotype was regulated by UDPG/P2Y14 signaling. STAT1 is phages (Fig. 4n–p). A similar result was also obtained by using known as a key transcription factor that mediates the macrophage

6 NATURE COMMUNICATIONS | (2020) 11:1769 | https://doi.org/10.1038/s41467-020-15636-8 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15636-8 ARTICLE

Fig. 4 UDPG regulates inflammatory macrophages via P2Y14 receptor. a–c Pgm1, Ugp2 or Gys1 siRNA transfected BMDMs were stimulated with IFN-γ/ LPS for 24 h, Nos2, Tnf, Il6 and Il1b expression was determined by real-time PCR. d Gys1, Pgm1 or Ugp2 siRNA transfected BMDMs were stimulated with IFN-γ/LPS for 24 h, UDPG in supernatants was determined by ELISA. e IFN-γ/LPS-stimulated BMDMs were treated with UDPG (0, 100 or 200 μM) for 24 or 36 h, Nos2, Tnf, Il6, and Il1b expression was determined by real-time PCR (left), iNOS, TNF and IL-6 expression was determined by western blot (middle) and ELISA (right). f, g Ugp2 or G6pdx siRNA transfected BMDMs were stimulated with IFN-γ/LPS ± UDPG (200 μM) for 24 or 36 h, Nos2, Tnf, Il6, and Il1b expression was determined by real-time PCR (left), iNOS, TNF and IL-6 expression was determined by western blot (middle) and ELISA (right). h Pygl or G6pdx siRNA transfected BMDMs were stimulated with IFN-γ/LPS for 24 h, UDPG in supernatants was determined by ELISA. i BMDMs were pretransfected with siRNA (Pygl or G6pdx) for 24 h prior to stimulation with IFN-γ/LPS for 6 h and switched to 13C-glucose for 6 h, lysed cells were analyzed by LC-MS/MS to determine m + 6-labeled UDPG. j, k P2Y14 receptor expression in untreated, IFN-γ/LPS- or IL-4-treated BMDMs was determined by real-time PCR (j) and western blot (k). l, m IFN-γ/LPS-stimulated BMDMs were treated with UDPG (0, 100 or 200 μM) for 24 or 36 h,

P2Y14 expression was determined by real-time PCR (l) and western blot (m). n–p P2ry14 siRNA transfected BMDMs were stimulated with or without IFN-γ/ LPS for 24 or 36 h, Nos2, Tnf, Il6, and Il1b expression was determined by real-time PCR (n), iNOS, TNF, and IL-6 expression was determined by western blot (o) and ELISA (p). Unless otherwise specified, n = 3 biologically independent experiments were performed. Data are presented as mean ± SEM. P values were calculated using one-way ANOVA. **p < 0.01, ****p < 0.0001.

inflammatory phenotype33,34. Indeed, Stat1 knockdown down- of STAT1 at Tyr701 (Fig. 6a, b) as well as its nuclear translocation regulated inflammatory gene expression in IFN-γ/LPS-treated (Fig. 6c), which could be reversed by the addition of UDPG macrophages (Fig. 5a). Surprisingly, the expression of Stat1 was (Fig. 6c, d and Supplementary Fig. 6a), suggesting that the markedly inhibited by Pgm1, Ugp2, Pygl or P2ry14 siRNA UDPG-P2Y14 signaling is required for STAT1 phosphorylation. (Fig. 5b). However, the addition of exogenous UDPG could res- Inflammatory macrophages use Janus kinases (JAK1/JAK2) to cue Stat1 expression in Pgm1-, Ugp2-orPygl-knockdown IFN-γ/ phosphorylate STAT139,40. We found that Pygl siRNA or GPI LPS-treated macrophages, but not in P2ry14-knockdown mac- treatment decreased the phosphorylation and the expression of rophages (Fig. 5b), suggesting that glycogen metabolism regulates JAK1 and JAK2 (Fig. 6e and Supplementary Fig. 6a). Such Stat1 expression via P2Y14 signaling. Several transcription factors decreased phosphorylation was not mediated by the proteasome including RARβ, ZNF-148 and IRF-1 have been reported to pathway, because the addition of proteasome inhibitor MG132 promote STAT1 expression35–38. Although RARβ, ZNF-148 and had no effect on JAK1/2 and STAT1 phosphorylation recovery fl fi IRF-1 were highly expressed in in ammatory macrophages (Fig. 6f). However, if we added Na3VO4, the non-speci c tyrosine (Supplementary Fig. 5a), the blockade of the glycogen-PPP phosphatase inhibitor, the above decreased phosphorylation of pathway by inhibitors or siRNAs only downregulated RARβ at JAK1/2 and STAT1 was rescued (Fig. 6g and Supplementary both mRNA and protein levels, but did not influence ZNF-148 Fig. 6b). T-cell protein tyrosine phosphatase 45 (TC45, encoded and IRF-1 (Fig. 5c–e and Supplementary Fig. 5b); and immu- by Ptpn2 gene), a nuclear protein tyrosine phosphatase, is the nofluorescent staining showed a consistent result of the decreased major enzyme that dephosphorylate JAK1/2 and STAT141–44.We nuclear location of RARβ (Fig. 5f). Intriguingly, knocking down found that TC45 was upregulated in inflammatory macrophages β P2Y14 receptor decreased RAR levels and overexpressing P2Y14 after GPI treatment (Fig. 6h); and Ptpn2 knockdown by siRNA receptor increased RARβ levels (Fig. 5g-i), suggesting that RARβ recovered the decreased phosphorylation of JAK1/2 and STAT1 is regulated by UDPG/P2Y14 signaling. Indeed, addition of exo- from the GPI treatment (Fig. 6i). Activation of MAP kinases ERK, genous UDPG induced RARβ and STAT1 expression in macro- JNK and P38 constitutes the core signaling pathway that mediates phages (Fig. 5j, k). Also, overexpression of RARβ induced STAT1 the inflammatory phenotype of macrophages45,46. Coincidently, fl and in ammatory gene expression (Fig. 5l-n and Supplementary the UDPG-P2Y14 signaling also induces the activation of MAP Fig. 5c). Then, we used siRNA to knock down Rarb, and this kinases47–49. We found that Pygl siRNA or GPI treatment resulted in the inhibition of the STAT1 and macrophage markedly inhibited the phosphorylation of ERK1/2, JNK and P38 inflammatory phenotype (Fig. 5o-q and Supplementary Fig. 5d). MAP kinases, however the addition of UDPG enhanced the Under this knockdown condition, the addition of exogenous phosphorylation of MAPKs (Fig. 6j and Supplementary Fig. 6a, UDPG could not rescue the inflammatory phenotype in the c). Moreover, using ERK, JNK and P38 inhibitors (U0126, macrophages (Supplementary Fig. 5e). In addition, Stat1 knock- SP600125 and SB203580) to treat UDPG and IFN-γ/LPS- down also invalidated the effect of the exogenous UDPG on the conditioned macrophages, we found that the expression of inflammatory phenotype in the macrophages (Supplementary TC45 was markedly upregulated, concomitant with the decreased β fi Fig. 5e). Thus, the UDPG-P2Y14-RAR axis was identi ed to phosphorylation of JAK1/2 and STAT1 as well as the down- regulate STAT1 expression. Moreover, analysis with the UCSC regulation of inflammatory gene expression (Fig. 6k, l and Genome Browser and JASPAR revealed the presence of multiple Supplementary Fig. 6d). Thus, P2Y14 signaling may activate consensus cis-elements for RARβ binding on the promoter of MAPKs so to downregulate TC45 expression. Together, these β Stat1, and a ChIP-PCR assay indicated that RAR directly bound data suggest that the UDPG-P2Y14 pathway regulates both to the Stat1 promoter (Fig. 5r), further confirming the above STAT1 expression and its activity. regulating axis. Despite the promoting effect of UDPG-P2Y14- RARβ on STAT1 in inflammatory macrophages, this axis seemed Macrophage glycogen metabolism regulates inflammation not to affect IL-4 stimulated macrophages, as evidenced by no in vivo. Next, we validated the in vivo significance of the above induction of STAT1 by the addition of UDPG or the over- elucidated glycogen metabolism in macrophages. In the LPS- β expression of P2Y14 or RAR (Supplementary Fig. 5f, g). induced acute peritonitis mouse model, an increased UDPG level In addition to STAT1 expression, we here also investigated was found, concomitant with the expression of inflammatory whether UDPG-P2Y14 signaling regulated the activity of STAT1. cytokines (TNF and IL-6) as well as NO (Fig. 7a, b); however, GPI STAT1 activation requires phosphorylation to form dimers, or 6AN treatment resulted in decreased UDPG levels and which are translocated into the nucleus where they exert their markedly suppressed expression of inflammatory genes (Fig. 7a, function33,34. We found that disruption of glycogenolysis by b). Meanwhile, the depletion of peritoneal macrophages by clo- either GPI or Pygl siRNA markedly inhibited the phosphorylation dronate liposomes (Clod) also led to the decrease of the

NATURE COMMUNICATIONS | (2020) 11:1769 | https://doi.org/10.1038/s41467-020-15636-8 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15636-8

abp = 0.0005 p < 0.0001 p = 0.0004 NC siStat1#1 siStat1#2 p = 0.0002 p = 0.02 p < 0.0001 p = 0.0001 p < 0.0001 2.0 p < 0.0001 2.0 4 p = 0.002 p < 0.0001 2.0 p = 0.0002 p < 0.0001 1.5 p < 0.0001 p < 0.0001 Ctrl p = 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 1.5 p < 0.0001 1.5 3 1.5 UDPG p < 0.0001 1.0 p = 0.001 1.0 1.0 p = 0.0001 2 1.0 ns ns 0.5 mRNA (fold) 0.5 mRNA (fold) 0.5 mRNA (fold) 1 mRNA (fold) 0.5 Stat1 Stat1 Stat1 Stat1 Relative mRNA 0 0 0 0 0 NC #1 #2 NC #1 #2 NC #1 #2 NC #1 #2 Nos2 Tnf Il6 Il1b siPgm1 siUgp2 siPygl siP2ry14 c d e f kDa kDa – ++ ++ + IFN-γ/LPS RARβ DAPI Merge 91 p-STAT1 91 p-STAT1 84 84 1.2 UN 91 STAT1 91 STAT1 84 84 < 0.0001

β β 0.8 < 0.0001

50 RAR 50 RAR < 0.0001 < 0.0001 p p p p IFN-γ/LPS 89 ZNF148 89 ZNF148

mRNA (fold) 0.4 45 IRF-1 45 IRF-1 IFN-γ/LPS

Rarb + GPI 43 β-actin 43 β-actin 0 020502050(μM) γ – + ++IFN- /LPS IFN-γ/LPS NC NC GPI 6AN – – + – GPI + 6AN –––+ 6AN siPygl#1siPygl#2 siG6pdx#1siG6pdx#2

g h i RARβ DAPI Merge j k p = 0.0005 UN UDPG 100 μ 1.2 p = 0.0004 Mock p < 0.0001 UDPG ( M) 8 p = 0.006 UDPG 200 30 p = 0.0002 kDa 0 100 200 6 0.8 50 RARβ P2Y14-OE 15 4 p < 0.0001 91 0.4 6 p-STAT1 mRNA (fold) mRNA (fold) 2 84 NC 4 p < 0.0001 91 STAT1 Rarb Rarb 2

0 0 Relative mRNA 84 #1 #2 0 NC -OE 43 β-actin 14 14 Mock 14 siP2ry14 Rarb Stat1 siP2rysiP2ry P2Y l m n o kDa NC siRarb#1 siRarb#2 p = 0.002 p = 0.0003 Mock 50 RARβ p = 0.0001 6 p = 0.0003 ) 800 1.5 RARβ-OE p = 0.001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 –1 91 p < 0.0001 p < 0.0001 p < 0.0001 p-STAT1 600 p = 0.01 p < 0.0001 4 p = 0.006 84 1.0 p = 0.0005 91

STAT1 p = 0.04 84 (pg mL 400 2 0.5 130 iNOS 200 Relative mRNA Relative mRNA

0 43 β-actin Cytokine 0 0 Stat1 Nos2 Tnf Il6 Il1b TNF IL-6 Stat1 Nos2 Tnf Il6 Il1b Mock RARβ-OE

pqr CAAAGTCAAGCAGTCA ATG Rarb siRNA –1042 –1027 p = 0.0004 p < 0.0001 Input kDa NC #1 #2 ) 600 –1 p < 0.0001 p < 0.0001 HRE 50 RARβ RARβ 226 bp 400 91 p-STAT1 UN 84 p < 0.0001 (pg mL IFN-γ/LPS 91 STAT1 IgG p < 0.0001 84 200 IFN-γ/LPS + GPI p < 0.0001 130 iNOS IFN-γ/LPS – + ++ IFN-γ/LPS + 6AN GPI + Cytokine – – – 43 β-actin 0 % Input 0 0.51.0 1.5 2.0 TNF IL-6 6AN –– –+

Fig. 5 UDPG-P2Y14 pathway regulates STAT1 expression. a Stat1 siRNA transfected BMDMs were stimulated with IFN-γ/LPS for 24 h, Nos2, Tnf, Il6, and Il1b expression was determined by real-time PCR. b Pgm1, Ugp2, Pygl or P2ry14 siRNA transfected BMDMs were stimulated with IFN-γ/LPS ± UDPG for 24 h, Stat1 expression was determined by real-time PCR. c–f BMDMs were pretreated with inhibitor (GPI or 6AN) for 30 min or pre-transfected with siRNA (Pygl or G6pdx) for 24 h prior to stimulation with IFN-γ/LPS for 24 or 36 h, STAT1, ZNF-148, IRF-1, and RARβ expression was determined by western blot (c, d),

RARβ expression and location were analyzed by real-time PCR (e) and confocal microscope, scale bar, 10 μm(f). g–i P2ry14 siRNA or P2Y14-overexpression vectors (P2Y14-OE) transfected BMDMs were stimulated with IFN-γ/LPS for 24 or 36 h, RARβ expression and location were determined by real-time PCR (g, h) and confocal microscope, scale bar, 10 μm(i). j, k IFN-γ/LPS stimulated BMDMs were treated with UDPG for 24 or 36 h, RARβ and STAT1 expression was determined by real-time PCR (j) and western blot (k). l–q RARβ-overexpression vectors (RARβ-OE) or Rarb siRNA transfected BMDMs were stimulated with IFN-γ/LPS for 24 or 36 h, Stat1, Nos2, Tnf, Il6, and Il1b expression was determined by real-time PCR (l, o), RARβ, STAT1, iNOS, TNF, and IL-6 expression was determined by western blot (m, p) and ELISA (n, q). r Schematic representation of the promoter region on the upstream of the transcription start site of Stat1. BMDMs were pretreated with GPI or 6AN for 30 min prior to stimulation with IFN-γ/LPS for 24 h, RARβ enrichment around the promoter of Stat1 were analyzed by ChIP-PCR and IgG was used as a negative control. ChIP-qPCR were used to Stat1 quantitative detection and total genomic DNA was used as input. Data are presented as mean ± SEM of n = 3 biologically independent experiments (a, b, e, g, j, l, n, o, q)orn = 4 biologically independent experiments (h)orn = 12 from two independent experiments (r). P values were calculated using one-way ANOVA (a, b, e, g, j, o, q, and r) and two-tailed unpaired Student’s t-tests (h, l, and n).

8 NATURE COMMUNICATIONS | (2020) 11:1769 | https://doi.org/10.1038/s41467-020-15636-8 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15636-8 ARTICLE

a Ctrl GPI 20 μM GPI 50 μM b NC siPygl#1 siPygl#2 c p-STAT1 DAPI Merge kDa 15 30 60 120 15 30 60 120 15 30 60 120 (min) kDa 15 30 60 120 15 30 60 120 15 30 60 120 (min) 91 91 NC 84 p-STAT1 84 p-STAT1 91 STAT1 91 STAT1 84 84 43 β-actin 43 β-actin siPygl

d NC siPygl siPygl + UDPG kDa 15 30 60 120 15 30 60 120 15 30 60 120 (min) e NC siPygl#1 siPygl#2 siPygl 91 +UDPG 84 p-STAT1 kDa 15 30 60 120 15 30 60 120 15 30 60 120 (min) 91 STAT1 130 p-JAK1 84 h 43 β-actin 130 JAK1 Nucleus Cytosol Ctrl GPI Ctrl GPI 125 p-JAK2 kDa 0 30 60 0 30 60 0 30 60 0 30 60 (min) f Ctrl GPI 20 μM GPI + MG132 125 JAK2 45 TC45 kDa 15 30 60 120 15 30 60 120 15 30 60 120 (min) β 91 43 -actin 84 p-STAT1 17 Histone H3 91 84 STAT1 43 β-actin g Ctrl GPI 20 μMGPI + Na3VO4 130 p-JAK1 kDa 15 30 60 120 15 30 60 120 15 30 60 120 (min) 130 JAK1 i Ptpn2 siRNA Ptpn2 siRNA 91 p-STAT1 84 kDaNC #1 #2 #3 kDa NC #1 #2 #3 125 p-JAK2 91 STAT1 130 p-JAK1 45 TC45 84 125 JAK2 130 p-JAK1 130 JAK1 91 p-STAT1 84 43 β-actin 130 JAK1 125 p-JAK2 91 STAT1 84 125 p-JAK2 125 JAK2 43 β-actin j NC siPygl siPygl + UDPG 125 JAK2 kDa 15 30 60 120 15 30 60 120 15 30 60 120 (min) p < 0.0001 l p = 0.0005 p < 0.0001 44 43 β-actin p = 0.0005 p-ERK 4 p < 0.0001 42 p = 0.0007 44 ERK 3 Nos2 Tnf 42

2 < 0.0001 k UDPG UDPG UDPG = 0.002 p 54 p-JNK p 46 Ctrl UDPG + U + SP + SB 1 1313131313(h)

kDa Relative mRNA 54 JNK 0 46 45 TC45 UN UN 43 p-P38 130 p-JAK1 UDPG U0126 UDPG U0126 SB203580SP600125 U + UDPG SB203580SP600125 U + UDPG SB + UDPGSP + UDPG SB + UDPGSP + UDPG 40 P38 130 JAK1 p = 0.0001 p = 0.016 130 p-JAK1 p < 0.0001 125 p-JAK2 p = 0.0002 8 p < 0.0001 130 JAK1 p = 0.005 125 JAK2 6 Il6 Il1b 125 p-JAK2 91 p-STAT1 4 < 0.0001 p

84 = 0.0002

125 JAK2 p 91 STAT1 2 45 TC45 84 Relative mRNA 0 43 β-actin 43 β-actin UN UN UDPG U0126 UDPG U0126 SB203580SP600125 U + UDPG SB203580SP600125 U + UDPG SB + UDPGSP + UDPG SB + UDPGSP + UDPG

Fig. 6 UDPG-P2Y14 pathway regulates STAT1 phosphorylation. a, b BMDMs were pretreated with GPI (20 or 50 μM) for 30 min or pretransfected with Pygl siRNA for 24 h prior to stimulation with IFN-γ/LPS, followed by western blot analysis of STAT1 and its phosphorylation from 15 to 120 min after stimulation. c, d Pygl siRNA transfected BMDMs were treated with IFN-γ/LPS ± UDPG, STAT1 and its phosphorylation was analyzed by two-photon confocal microscope, scale bar, 10 μm(c) and western blot (d). e BMDMs were pretransfected with Pygl siRNA for 24 h prior to stimulation with IFN-γ/ LPS, followed by western blot analysis of JAK1 and JAK2 from 15 to 120 min after stimulation. f, g BMDMs were pretreated with GPI alone or combined with MG132 (f)orNa3VO4 (g) for 30 min prior to stimulation with IFN-γ/LPS, followed by western blot analysis of STAT1, JAK1, and JAK2 from 15 to 120 min after stimulation. h BMDMs were pretreated with GPI for 30 min prior to stimulation with IFN-γ/LPS, the cytoplasmic and nuclear protein fractions were blotted for TC45, β-actin (cytoplasmic marker) and Histone H3 (nuclear marker). i Ptpn2 siRNA transfected BMDMs were stimulated with IFN-γ/LPS, STAT1, JAK1, JAK2, and TC45 were analyzed by western blot. j Pygl siRNA transfected BMDMs were stimulated with IFN-γ/LPS ± UDPG, ERK, JNK, P38, JAK1, JAK2, and TC45 were analyzed from 15 to 120 min after stimulation by western blot. k, l IFN-γ/LPS stimulated BMDMs were treated with UDPG alone or combined with U0126, SP600125 or SB203580 for 1 and 3 h, JAK1, JAK2, STAT1, and TC45 were analyzed by western blot (k). Nos2, Tnf, Il6, and Il1b expression was determined by real-time PCR (l). Data are presented as mean ± SEM of n = 3 biologically independent experiments. P values were calculated using one-way ANOVA. inflammatory cytokines and NO in the above LPS-treated mice did not show different levels of TNF and IL-6 in the serum (Fig. 7a, b). Next, we isolated peritoneal macrophages from the (Supplementary Fig. 7b), suggesting that the glycogen metabolism GPI- or 6AN-treated mice, which showed an accumulation of and PPP activity in macrophages might be the relevant target of glycogen and a marked downregulation in the expression of Rarb, these inhibitors in vivo. Besides LPS-induced peritonitis, similar Stat1 and Nos2 (Supplementary Fig. 7a and Fig. 7c). Moreover, results were obtained in the ConA-induced hepatitis model, while all the untreated mice died from LPS-induced peritonitis, where the treatment with GPI, 6AN or clodronate liposomes most of those who were treated with GPI, 6AN or clodronate inhibited inflammatory cytokines (TNF and IL-6), reduced liver liposomes survived (Fig. 7d). Here, we also treated the mice with damage, and decreased the mortality rate (Supplementary the GPI plus Clod or 6AN plus Clod, however the combination Fig. 7c–f). To further convince the role of glycogen metabolism in

NATURE COMMUNICATIONS | (2020) 11:1769 | https://doi.org/10.1038/s41467-020-15636-8 | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15636-8

a b **** **** **** **** **** **** 100 **** 2.5 **** 50 **** **** 150 **** **** ) **** **** **** **** ) ) ) –1 –1 2.0 –1 –1 75 40 1.5 30 100

50 mol L (pg mL μ 1.0 (ng mL 20 50 25 0.5 10 IL-6 NO ( TNF (ng mL UDPG 0 0 0 0

PBS GPI 6AN Clod LPS PBS GPI 6AN Clod LPS PBS GPI 6AN Clod LPS PBS GPI 6AN Clod LPS

LPS LPS+ GPI LPS+ 6AN + Clod LPS LPS+ GPI LPS+ 6AN + Clod LPS LPS+ GPI +LPS 6AN + Clod LPS LPS+ GPI +LPS 6AN + Clod

c **** **** d **** **** **** **** **** **** 200 **** **** 3 **** **** 3 100 PBS 150 GPI 2 2 p = 0.0001 6AN 100 p = 0.0007 Clod 1 50 LPS mRNA (fold)

mRNA (fold) 1 mRNA (fold) 50 p < 0.0001 LPS + GPI LPS + 6AN Rarb Stat1

0 0 Nos2 0 LPS + Clod Survival ratio (%) 0 GPI LPS GPI 6AN LPS GPI6AN LPS PBS 6ANClod PBS Clod PBS Clod 020406080(h)

LPSLPS + GPI LPS+ 6AN + Clod LPSLPS + GPI +LPS 6AN + Clod LPSLPS + GPI LPS+ 6AN + Clod e fg 100 400 300

Sham ) Ileocecal artery ) –1 Ileum CLP 300 –1 p = 0.0081 200 Colon = 0.0004 < 0.0001 p

p = 0.0188 CLP + GPI = 0.0006 < 0.0001 < 0.0001 p < 0.0001 p

200 p p

< 0.0001 = 0.0001 p = 0.0434 p p p CLP + 6AN 50 100 CLP + Clod 100 ALT (IU L AST (IU L 0 0

Survival ratio (%) 0 CLP Cecum 0 2 468 (days) Sham CLP Sham

CLP CLP+ GPICLP + 6AN + Clod CLPCLP + GPI CLP+ 6AN + Clod h j **** **** **** **** **** **** **** **** **** **** 300 **** **** 400 **** **** 20 4 **** **** 150 ) **** **** **** **** –1 )

–1 15 3 200 100

(IU L 200 10 2 mRNA (fold) mRNA (fold) (×1000)

100 mRNA (fold) 50 5 1 CPK BUN (mmol L Rarb Nos2 0 0 0 Stat1 0 0

CLP CLP CLP CLP CLP Sham Sham Sham Sham Sham

CLP CLP+ GPI CLP+ 6AN + Clod CLP CLP+ GPI CLP+ 6AN + Clod CLP CLP+ GPI CLP+ 6AN + Clod CLP CLP+ GPI CLP+ 6AN + Clod CLP CLP+ GPI CLP+ 6AN + Clod

i **** k l 2.0 **** 12 **** LPS induced peritonitis CLP-induced sepsis ) **** ) –1 –1 **** **** 100 100 1.5 **** **** 8 1.0

(ng mL WT 4 0.5 50 WT 50 P2Y14 (+/–)

IL-6 14 TNF (ng mL P2Y (–/–) 0 0 P2Y14 (+/–) p = 0.0065 P2Y14 (–/–) p = 0.0055 p < 0.0001 p < 0.0001 Survival ratio (%) CLP CLP Survival ratio (%) Sham Sham 0 0 020406080(h) 0 4 8 12 (days) CLP CLP+ GPI CLP+ 6AN + Clod CLP CLP+ GPI CLP+ 6AN + Clod m n o γ 14 14 γ WT-UN WT+IFN- /LPS P2Y (–/–)-UN P2Y (–/–)+IFN- /LPS WT P2Y14 (–/–) kDa p = 0.0008 ) p < 0.0001 0.09 **** 1.2 **** 50 RARβ –1 600 91 Actb Actb p-STAT1 0.06 0.8 84 400 91 STAT1 **** **** 84 0.03 0.4 200 **** **** 130 iNOS Relative to Relative to 43 β-actin

0 0 Cytokine (pg mL 0 Rarb Stat1 Nos2 Tnf Il6 Il1b – + – + IFN-γ/LPS TNF IL-6 acute inflammatory disorders, a cecal ligation and puncture released inflammatory cytokines such as TNF and IL-6 were (CLP)-induced sepsis mouse model was used (Fig. 7e)50–52. reduced (Fig. 7i). Consistently, Rarb, Stat1 and Nos2 was down- Blocking glycogen metabolism by either GPI or 6AN effectively regulated in the peritoneal macrophages after GPI or 6AN inhibited sepsis, as evidenced by (1) 80% mice were rescued from treatment (Fig. 7j), implicating that P2Y14 signaling regulates the fl septic death (Fig. 7f); (2) CLP-induced organic damage (liver, in ammatory responses in vivo. We thus used P2Y14-/- mice to fi kidney and heart) was relieved (Fig. 7g, h); and (3) the levels of con rm this, it was found that P2Y14 knockout effectively

10 NATURE COMMUNICATIONS | (2020) 11:1769 | https://doi.org/10.1038/s41467-020-15636-8 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15636-8 ARTICLE

Fig. 7 Macrophage glycogen metabolism regulates inflammation in vivo. a–d C57BL/6 J mice were treated with GPI, 6AN or clodronate liposomes, followed by i.p. injection of 20 μgg−1 body weight LPS. Four hours later, serum levels of UDPG were detected by ELISA, n = 5 mice per group (a). Serum levels of TNF and IL-6 (left and middle, n = 5 mice per group) and NO (right, n = 4 mice per group) were measured by ELISA (b). Rarb, Stat1 and Nos2 expression in peritoneal macrophages was determined by real-time PCR, n = 5 mice per group (c). The long-term survival of LPS-induced peritonitis was assessed, n = 10 mice per group, p values are presented relative to LPS group (d). e Schematic illustration showing the cecal ligation and puncture (CLP). f–j C57BL/6J mice were treated with GPI, 6AN or clodronate liposomes, followed by CLP procedure, the long-term survival of CLP-induced sepsis was recorded, n = 10 mice per group, p values are presented relative to CLP group (f). Serum levels of transaminase AST and ALT (g, n = 4 mice per group), BUN (h, left, n = 5 mice per group), CPK (h, right, n = 4 mice per group), TNF and IL-6 (i, n = 6 mice per group) were determined by ELISA. Peritoneal macrophages were isolated and Rarb, Stat1 and Nos2 expression was determined by real-time PCR, n = 5 mice per group (j). k, l Wide type (WT), P2Y14 −1 (+/−) or P2Y14 (−/−) mice were injected intraperitoneally with 20 μgg body weight LPS or were subjected to a sham or CLP procedure (n = 8 mice per group), the long-term survival was recorded, p values are presented relative to P2Y14 (−/−) group. m–o BMDMs from WT or P2Y14 (−/−) mice were stimulated with IFN-γ/LPS for 24 or 36 h, Rarb, Stat1, Nos2, Tnf, Il6, and Il1b expression was determined by real-time PCR, n = 3 mice per group (m), RARβ, STAT1, iNOS, TNF, and IL-6 expression was determined by western blot (n) and ELISA, n = 6 mice per group (o). Unless otherwise specified, n = 3 independent experiments were performed. Data are presented as mean ± SEM. P values were calculated using one-way ANOVA (a–c, g–j, m and o) and two-sided log-rank (Mantel-Cox) test (d, f, k and l). ****p < 0.0001.

prevented the death of the mice with peritonitis or sepsis Discussion (Fig. 7k, l). In line with these results, macrophages isolated from Remodeled metabolism is known to be vital for a macrophage- − − β fl P2Y14 / mice downregulated RAR , STAT1 and pro- mediated in ammatory response, but the underlying mechanism inflammatory cytokine expression in response to IFN-γ/LPS sti- remains largely unclear, especially for the regulation of inflam- mulation (Fig. 7m-o). However, this pro-inflammatory gene matory gene expression by the altered metabolism. In this study, downregulation could be rescued by RARβ or STAT1 over- we provide evidence that macrophages mobilize glycogen meta- expression (Supplementary Fig. 7g-i). Together, these results bolism, which governs macrophage-mediated inflammatory suggest that the glycogen metabolism in macrophages regulates response. On one hand, glucose through the PPP via glycogenesis inflammatory responses. and glycogenolysis, thus providing antioxidative NADPH for the survival of activated inflammatory macrophages; on the other Macrophage glycogen metabolism regulates sepsis in patients. hand, glycogen metabolism produces an intermediate metabolite To validate the above murine data in human inflammatory UDPG which triggers the P2Y14 signaling pathway, which then responses, we first repeated the results in the human monocytic regulates the key inflammatory transcription factor STAT1 THP-1 cells, the most widely used model for human macro- expression and activity. phages53. Similarly, IFN-γ/LPS-treated THP-1 cells displayed the Glycogen, the long-term reservoir of glucose, is known to be upregulation expression of inflammatory phenotypes NOS2, TNF, primarily produced by liver and muscle cells, thus providing IL6, and IL1B (Fig. 8a), However, disruption of either glycogen- energy for cells and maintaining blood sugar homeostasis of the olysis or PPP by inhibitors (GPI or 6AN) resulted in marked body24. Notwithstanding this original understanding, the physio- pathological role of glycogen metabolism may be more complex. decrease of the expression of NOS2, TNF, IL6, and IL1B (Fig. 8b). + In line with the subdued inflammatory phenotype, marked We previously reported that CD8 memory T cells use a glycogen decrease of UDPG levels were also observed, concomitant with metabolic program to regulate memory formation and main- the downregulation of RARB and STAT1 expression (Fig. 8c-e). tenance17. In the present study, we further show that an active By contrast, the addition of exogenous UDPG led to the upre- glycogen metabolism governs the inflammatory phenotype of gulation of inflammatory gene expression as well as the RARβ macrophages. In these macrophages, the end product of glycogen and STAT1 expression (Fig. 8f, g). In addition, treatment with metabolism, G6P, is not channeled to glycolysis but to the PPP. fl P2RY14 or RARB siRNA also downregulated STAT1 expression Previously, an active PPP was observed in in ammatory and inflammatory gene expression (Fig. 8h, i and Supplementary macrophages11,13,55. However, those studies considered that the Fig. 8a, b). These data suggest that glycogen metabolism regulates PPP was directly branched from glycolysis and used glycolysis- inflammatory phenotype in human macrophages. derived G6P. Nevertheless, by conducting 13C tracing assay, here Excessive activation of innate immune cells may lead to SIRS or we provide clear evidence that the G6P is not directly derived even sepsis1–3,54. Given the crucial role of macrophages in human from glycolysis but from glycogenolysis. Following glucose inflammatory conditions and in the above-described sepsis mouse phosphorylation by hexokinase, the formed G6P is first used in model, we here further investigated whether glycogen metabolism glycogenesis. Then, the glycogen is degraded to produce G6P was involved in human SIRS or sepsis. For this purpose, blood again and the latter is channeled through the PPP. Identification samples from patients with sepsis (n = 25) or SIRS (n = 28) and of this metabolic pathway undoubtedly provides new insight into healthy controls (n = 30) were collected (Supplementary Table 1). how glucose metabolism regulates macrophage phenotype, but We found that glycogen levels in CD14+ monocytes were higher raises the question that why macrophages use G6P to synthesize in SIRS and sepsis patients compared to healthy donors (Fig. 8j). glycogen and then degrade glycogen to produce G6P, an overt In accordance with this, the expression of glycogenic and futile glycogen synthesis/degradation cycle. This might be due to glycogenolytic enzymes was also upregulated (Fig. 8k), concomi- the distinctive compartmentalization of the involved enzymes. tant with increased expression of RARB and STAT1 (Fig. 8l). Another explanation is that glycogen metabolism triggers the + fl When we cultured the CD14 monocytes and treated them with UDPG-P2Y14 signaling pathway that regulates in ammatory gene GPI or 6AN, we found that blocking glycogenolysis or PPP led to expression in macrophages. UDP-glucose (UDPG) and other the decrease of UDPG levels, downregulation of inflammatory UDP-sugars are essential for glycosylation. Following the synth- gene expression as well as RARB and STAT1 genes (Fig. 8m, n). esis in the cytosol, UDPG is translocated to the ER/Golgi via ER/ Together, these results suggest that glycogen metabolism of Golgi-resident SLC35 transporters, where UDPG is used for human macrophages regulates inflammatory responses in septic glycosylation. On the other hand, UDPG in the ER/Golgi lumen patients. can also be released as cargo to the extracellular space via the

NATURE COMMUNICATIONS | (2020) 11:1769 | https://doi.org/10.1038/s41467-020-15636-8 | www.nature.com/naturecommunications 11 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15636-8

a b UN GPI 20 μM GPI 50 μM UN 6AN 20 μM 6AN 50 μM c UN p < 0.0001 p = 0.001 IFN-γ/LPS p < 0.0001 p < 0.0001 1.5 p = 0.002 1.5 p < 0.0001 80 800 IL-4 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p = 0.0007 p < 0.0001 p = 0.0002 cells) 6

p < 0.0001 p = 0.0001 p = 0.001 p < 0.0001 p = 0.0021 p < 0.0001 60 < 0.0001 p 400 < 0.0001 < 0.0001 p < 0.0001 p < 0.0001 p

1.0 1.0 p 9 p < 0.0001 40 p < 0.0001 p < 0.0001 6 p < 0.0001 0.5 0.5 20 3 0 Relative mRNA Relative mRNA Relative mRNA 0 0 0 020502050(μM) NOS2 TNF IL6 IL1B NOS2 TNF IL6 IL1B NOS2 TNF IL6 IL1B UDPG (pg per 10 GPI 6AN d e f 1.2 1.2 UN UDPG GPI 6AN 8 p < 0.0001

= 0.01 p = 0.0004 = 0.0001 p < 0.0001 p μ = 0.01

< 0.0001 ( M) p 0 20 50 20 50 = 0.0001 p 0.8

0.8 < 0.0001 p = 0.0002 p kDa

p 6 < 0.0001 p p = 0.0005 p β 50 RAR p = 0.0006 p = 0.002 mRNA (fold) mRNA (fold) 0.4 4 0.4 91 p-STAT1 84 2 91 STAT1 0 Relative mRNA RARB 0 STAT1 84 020502050(μM) 020502050(μM) 0 43 β-actin GPI 6AN GPI 6AN RARB STAT1 NOS2 TNF IL6 IL1B g h i NC siP2RY #1 siP2RY #2 1.2 14 14 NC siRARB #1 siARB #2 1.5 1.5 kDa UN UDPG p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001

< 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 p < 0.0001 < 0.0001 β p p 50 RAR 0.8 < 0.0001 < 0.0001 p

p 1.0 1.0 91 p-STAT1 mRNA (fold) 0.4 84 0.5 0.5 91 STAT1 84 0 Relative mRNA Relative mRNA STAT1 β 43 -actin NC #1 #2 0 0 14 14 NOS2 TNF IL6 IL1B NOS2 TNF IL6 IL1B

siP2RYsiP2RYsiRARBsiRARB #1 #2

p < 0.0001 j k p < 0.0001 l p < 0.0001 p < 0.0001 p < 0.0001 p = 0.0007 0.06 p = 0.0212 p = 0.1343 150 0.008 p = 0.0682 0.004 p = 0.4734 0.05 p = 0.0803 p < 0.0001 p < 0.0001 p = 0.0143 0.006 0.003 0.04 p = 0.0024 ACTB ACTB 100 0.04 ACTB ACTB 0.03 mRNA mRNA mRNA mRNA protein) 0.004 0.002 –1

g 0.02

μ 50 0.02 Glycogen 0.002 0.001 GYS1 PYGL RARB 0.01 STAT1 (ng relative to relative to relative to 0 relative to 0 0 0 0 HC SIRS Sepsis HC SIRS Sepsis HC SIRS Sepsis HC SIRS Sepsis HC SIRS Sepsis

m n Patient 1 Patient 2

12 p < 0.0001 UN p = 0.0002 p = 0.001 p = 0.0005 UN p < 0.0001 1.5 p < 0.0001 p < 0.0001 p < 0.0001 1.5 p = 0.001 p < 0.0001 p = 0.01 p < 0.0001 p < 0.0001 p < 0.0001 GPI p < 0.0001 p < 0.0001 GPI p < 0.0001 p < 0.0001 p = 0.0002 p = 0.0004 p < 0.0001 p < 0.0001 6AN p = 0.0004 p = 0.0004 p < 0.0001 p = 0.0004 p < 0.0001 p < 0.0001 6AN

cells) 8 5 1.0 1.0

UDPG 4 0.5 0.5 Relative mRNA Relative mRNA (pg per 10 0 0 0 Patient 1 Patient 2 RARB STAT1 NOS2 TNF IL6 IL1B RARB STAT1 NOS2 TNF IL6 IL1B

Fig. 8 Macrophage glycogen metabolism regulates sepsis in patients. a THP-1 cells were cultured with PMA (100 ng mL−1) for 3 days and differentiated into macrophages, followed with IFN-γ/LPS or IL-4 stimulation for 24 h. NOS2, TNF, IL6 and IL1B expression was determined by real-time PCR, n = 3 biologically independent experiments. b–e PMA incubated THP-1 cells were pretreated for 30 min with GPI or 6AN prior to stimulation with IFN-γ/LPS for 24 or 36 h, NOS2, TNF, IL6, and IL1B expression was determined by real-time PCR, n = 3 biologically independent experiments (b). UDPG in supernatants was determined by ELISA, n = 3 biologically independent experiments (c). RARβ and STAT1 expression was determined by real-time PCR, n = 3 biologically independent experiments (d) and western blot (e). f, g PMA and IFN-γ/LPS-stimulated THP-1 cells were treated with or without UDPG for 24 or 36 h, RARB, STAT1, NOS2, TNF, IL6, and IL1B expression was determined by real-time PCR, n = 3 biologically independent experiments (f), RARβ and STAT1 expression was determined by western blot (g). h, i P2RY14 or RARB siRNA transfected THP-1 cells were stimulated with IFN-γ/LPS for 24 h, STAT1, NOS2, TNF, IL6, and IL1B expression was determined by real-time PCR, n = 3 biologically independent experiments. j–l Blood samples from patients with sepsis (n = 25) and SIRS (n = 28) and healthy controls (n = 30) were collected. Intracellular glycogen levels in human peripheral blood CD14+ monocytes were determined by colorimetric assay (j). GYS1, PYGL, RARB, and STAT1 expression was determined in human peripheral blood CD14+ monocytes by real-time PCR (k, l). m, n Two sepsis patients’ CD14+ monocytes were isolated and cultured with M-CSF (20 ng mL−1) for 4 days, and then treated with GPI or 6AN respectively. UDPG in supernatants was determined by ELISA, n = 3 biologically independent experiments (m) and RARB, STAT1, NOS2, TNF, IL6, and IL1B expression was determined by real-time PCR (n), n = 3 biologically independent experiments. Data are presented as mean ± SEM. P values were calculated using one-way ANOVA (a–d and h–n) and two-tailed unpaired Student’s t-tests (f).

30,31 constitutive secretory pathway . This released UDPG can bind we further identify that UDPG-P2Y14 signaling regulates the with the P2Y14 receptor via an autocrine or paracrine pattern. expression and phosphorylation of STAT1 via upregulating β P2Y14 is a purinergic G-protein coupled receptor, which can be transcription factor retinoic acid receptor RAR and down- expressed by immune cells including macrophages56,57. Notably, regulating tyrosine phosphatase TC45. In this study, we did not UDPG can act as the agonist to activate P2Y14 signaling, which elucidate how RARβ was regulated through the UDPG-P2Y14 may include phosphatidylinositol 3-kinase-γ, GPCR kinases 2 signaling. RARs, composed of three subtypes α, β and γ, act as and 3, phospholipase C and MAPKs47–49,58. In the present study, ligand-activated vfactors through dimerizing with retinoid X

12 NATURE COMMUNICATIONS | (2020) 11:1769 | https://doi.org/10.1038/s41467-020-15636-8 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15636-8 ARTICLE receptors (RXR) and binding to retinoic acid response elements Cells. Human monocytic THP-1 cells were purchased from the China Center for (RARE) in the promoter region of target genes. Intriguingly, Type Culture Collection (Wuhan, China) and cultured in complete RPMI-1640 β 59 medium containing 10% fetal bovine serum, 10 mM glucose, 2 mM L-glutamine RAR itself can be one of the target genes . More recently, a −1 β and 100 U mL penicillin-streptomycin. THP-1 cells were differentiated into M0 study showed that RAR was upregulated by the MAPK (ERK1/2 macrophages by incubation with 100 ng mL−1 phorbol 12-myristate 13-acetate and P38) signaling, which was consistent with an earlier report (PMA) (P8139, Sigma-Aldrich) for 72 h. Once the cells were adherent, they were β transferred to PMA-free media to obtain resting macrophages. These cells were that nerve growth factor-activated Ras signal pathway for RAR − 60 incubated with 20 ng mL 1 human macrophage colony-stimulating factor (M- upregulation . Notably, P2Y14 signaling also activates MAPK. − β CSF) (300-25, PeproTech) and then co-stimulated with 100 ng mL 1 LPS (L2630, Thus, P2Y14 signaling might regulate RAR levels through the Sigma-Aldrich) plus 20 ng mL−1 IFN-γ (300-02, PeproTech) or stimulated with − MAPK pathway. All in all, the elucidation of the UDPG-P2Y14 10 ng mL 1 IL-4 (200-04, PeproTech) for 24 h to generating inflammatory or anti- signaling provides an example insight into how glucose meta- inflammatory macrophages. bolism is involved in the signaling pathway of cells. In addition to fl in ammatory macrophages, other immune cell type(s) might also Human samples. Human peripheral blood was obtained from patients at ICU of mobilize this signaling pathway to respond innate stimulation. Union Hospital (Wuhan, China). Ethical permission was granted by the Ethics Neutrophils are some similar to macrophages in that they have Committee of the Huazhong University of Science and Technology. All patients the common progenitors and neutrophils also have the phago- provided written informed consent to participate in the study. cytotic function. Whether neutrophils use glycogen metabolism to trigger the UDPG-P2Y14 signaling is worthy of further Reagents. GPI (PZ0189), 6AN (A68203), UDPG (U4625), ConA (L7647), PPTN fl investigation. tri uoroacetate salt (SML1809), MG-132 (M8699), Na3VO4 (S6508), SB203580 Inflammatory macrophages play a crucial role in mediating (S8307), SP600125 (S5567), U0126 (19-147) were purchased from Sigma-Aldrich. acute immune responses that cause pathological tissue damage in Clodronate liposomes was purchased from Liposoma. Glucose (GO) Assay Kit various inflammatory diseases1–3,54,61. The glycogen-PPP and (GAGO20) and Periodic Acid-Schiff (PAS) Assay Kit (395B) were purchased from fi Sigma-Aldrich. The primary antibodies were purchased from Cell Signaling UDPG-P2Y14 pathways identi ed in this study may have Technology: anti-iNOS (D6B6S#13120, 1:1000), anti-phospho-STAT1-Tyr701 important clinical significance and provide potential therapeutic (58D6#9167, 1:1000), anti-STAT1 (9H2#9176, 1:1000), anti-JAK1 (D1T6W#50996, targets to block acute inflammatory responses. Indeed, in murine 1:1000), anti-phospho-ERK-T202/Y204 (194G2#4377S, 1:1000), anti-ERK (137F5#4695S, 1:1000), anti-phospho-JNK-T183/Y185 (9251S, 1:1000), anti-JNK models of both acute peritonitis and liver damage, blocking (9252S, 1:1000), anti-phospho-P38-Thr180/Tyr182 (3D7#9215, 1:1000), anti-P38 glycogen-PPP metabolic or UDPG-P2Y14 signaling pathway (9212S, 1:1000), anti-Fbp1 (D2T7F, 1:1000), anti-G6pdx (#8866, 1:1000), anti-Gys1 effectively inhibits the inflammatory response and averts certain (3893S, 1:1000), anti-TC45 (D7T7D#58935, 1:1000) and anti-IRF-1 (D5E4#8478, death for the mice. More importantly, such treatment also pro- 1:1000). The following primary antibodies were purchased from Abcam: anti-Pgm1 duces an efficacious outcome in the mouse sepsis model. Sepsis is (ab192876, 1:1000), anti-Ugp2 (ab154817, 1:1000), anti-Pygl (ab190243, 1:1000), anti-Pygm (ab88078, 1:1000), anti-P2Y (ab136264, 1:1000), anti-G6pase a serious clinical issue, causing millions of deaths worldwide each 14 – (ab83690, 1:1000) and anti-Histone H3 (ab8284, 1:1000). Anti-phospho-JAK1- year61 65. Analysis of clinical samples of SIRS/Sepsis patients also Y1022/1023 (YP0154, 1:1000), anti-phospho-JAK2-Tyr570 (YP0306, 1:1000) and indicates that macrophages activate glycogen-PPP and UDPG- anti-JAK2 (YT2428, 1:1000) were purchased from Immunoway. Anti-Pck1 (Z6754- P2Y pathways to polarize an inflammatory phenotype, unco- Z-AP, 1:1000) was purchased from Proteintech. Anti-6Pgd (A7710, 1:1000), anti- 14 HK1 (A1054, 1:1000), anti-HK2 (A0994, 1:1000) and anti-HK3 (A8428, 1:1000) vering a potential strategy against clinical sepsis. Upon activation, were purchased from ABclonal. Anti-ZNF148 (QC7801, 1:1000), anti-RARβ macrophages can be polarized to an inflammatory or anti- (310315, 1:1000) and anti-β-actin (A1978, 1:10000) were purchased from Sigma- inflammatory phenotype. Although the glycogen-PPP or UDPG- Aldrich. The secondary antibody goat anti-rabbit IgG Dylight®594 (ab96885, P2Y pathway is required for an inflammatory phenotype, using 1:400) was purchased from Abcam, HRP-goat anti-rabbit and HRP-goat anti- 14 mouse (1:10000) were purchased from EARTH. siRNA (Pgm1, Ugp2, Pygl, G6pdx or P2ry14) to block the γ glycogen-PPP or UDPG-P2Y14 pathway did not switch IFN- / LPS-stimulated macrophage development toward anti- Preparation of mouse macrophages. Bone marrow cells isolated from C57BL/6J inflammatory phenotype such as increased levels of arginase 1 mice were cultured for 5 days in the complete RPMI-1640 medium containing 20 ng mL−1 recombinant mouse M-CSF (315-02, PeproTech), 10% fetal bovine and IL-10 (Supplementary Fig. 9). In addition to sepsis, these −1 fi serum, 10 mM glucose, 2 mM L-glutamine and 100 U mL penicillin- ndings have potential value to break immunological tolerance in streptomycin. On day 6, macrophages were co-stimulated with 100 ng mL−1 LPS cancer. For instance, it is possible for us to use glycogen synthase plus 20 ng mL−1 IFN-γ (315-05, Sigma-Aldrich) or stimulated with 10 ng mL−1 IL- activator to reset tumor-associated macrophages toward inflam- 4 (214-14, Sigma-Aldrich) for 24 h to generating inflammatory or anti- fl matory phenotype through the regulation of glycogen in ammatory macrophages. Mouse peritoneal macrophages were harvested by peritoneal lavage. Cold PBS was injected into the peritoneal cavity and extracted metabolism. after gentle agitation. The peritoneal cell suspension was centrifuged at 1300 rpm, In summary, the data in this study clearly show that macro- and the cell pellet was mixed with 2 mL Red blood cell lysis buffer for 5 min at phages, by virtue of their mobilizing glycogen metabolism, are room temperature. After washing, the cells were cultured on six-well plate for 3 h. polarized to an inflammatory phenotype. This metabolic pathway The adhesion cells were collected as peritoneal macrophages. not only enhance the PPP that provides NADPH for antioxida- tion and cell survival, but also triggers the UDPG-P2Y14 signaling Isolation of human monocytes. Human peripheral blood mononuclear cells were pathway to promote the expression and activity of the key isolated from human peripheral blood using density gradient separation. Mono- inflammatory transcription factor STAT1 (Fig. 9). Therefore, we cytes were purified by human CD14 Micro-Beads (130-050-201, MACS) and then −1 identify a previously unknown key metabolic pathway that fun- cultured in complete RPMI 1640 medium containing 20 ng mL recombinant fl human M-CSF for the induction of macrophages. Seven days later, human mac- damentally regulates the in ammatory phenotype of macro- rophages were harvested and stimulated with 100 ng mL−1 LPS plus 20 ng mL−1 phages, which provides novel strategies against inflammatory human IFN-γ or 10 ng mL−1 human IL-4 for inflammatory and anti-inflammatory diseases. macrophages.

Methods Electron microscopy. The untreated, IFN-γ/LPS or IL-4-treated macrophages Mice. Female wild-type C57BL/6J mice with 5–7-weeks old were purchased from were washed with PBS three times, then fixed in 2.5% glutaraldehyde in 0.1 M PBS the Center of Medical Experimental Animals of Hubei Province (Wuhan, China) and processed for routine electron microscopy as described previously17.Briefly, +/− −/− and P2Y14 and P2Y14 mice were purchased from the Cyagen Biosciences the samples were post fixed in 1% osmium tetroxide (OsO4) for 100 min at room Inc. For studies, all mice were kept under SPF conditions at the Animal Care and temperature, and rinsed with distilled water three times. The pellets were then Use Committee of Tongji Medical College. All animal experiments were conducted dehydrated in a graded ethanol series, treated with propylene oxide and embedded in accordance with a protocol approved by the Animal Care and Use Committee of with Spurr’s epoxy resin. Cut sections were stained with uranyl acetate and lead Tongji Medical College (ethics number: S1891). citrate, and then imaged using a JEM1010 electron microscope (JEOL).

NATURE COMMUNICATIONS | (2020) 11:1769 | https://doi.org/10.1038/s41467-020-15636-8 | www.nature.com/naturecommunications 13 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15636-8

Glucose UDPG

UDPG P2Y14 Golgi Glucose

G6pdx Pgm1 6PG 6PGLG6P G1P TC45 NADPH NADP+ Ugp2 NADP+ RARβ P 6Pgd Pygl UDPG NADPH STAT1 Gys1 Pyruvate ROS Glycogen R5P TC45 RARβ S7P STAT1 P STAT1 RARβ STAT1 E4P Nucleus Inflammatory gene expression

Fig. 9 Glycogen metabolism regulates macrophages. In inflammatory macrophages, glycogen is synthesized and then channeled through glycogenolysis to generate G6P and further through the pentose phosphate pathway to yield abundant NADPH, ensuring high levels of reduced glutathione for inflammatory macrophage survival. Glycogen metabolism also increases UDPG levels and the receptor P2Y14 in macrophages. The UDPG/P2Y14 signaling pathway not only upregulates the expression of STAT1 via activating RARβ but also promotes STAT1 phosphorylation by downregulating phosphatase TC45.

Glucose consumption. The untreated, IFN-γ/LPS or IL-4-treated macrophages CellROX Green for 30 min at 37 °C, protected from light. Cells were washed and after complete medium exchange, a cell-free control group were set. Following a scraped in PBS and immediately analyzed by flow cytometry, using 488 nm exci- 24-h incubation, the supernatants were collected and glucose concentration was tation for the CellROX Green. measured using Glucose Assay Kit according to the manufacturer’s instructions. Cells were lysed for protein quantification using a BCA Protein Assay Kit (23227, Plasmid constructs and transfection. Recombinant vectors encoding murine Thermo Fisher). Glucose consumption rate was determined as the glucose con- fi P2ry14, Stat1 and Rarb were constructed by PCR-based ampli cation from cDNA centration in the supernatants minus that of the cell-free control group, and + normalized to total protein levels. of BMC and then were subcloned into the GV230 (P2ry14), pcDNA3.1( ) × Flag (Stat1) eukaryotic expression vectors or GV287 (Rarb) lentivirus expression vector. All constructs were confirmed by DNA sequencing. Plasmids were transiently Metabolite analysis. For 13C tracing experiments, bone marrow cells were cul- transfected into BMM with Lipofectamine 2000 transfection reagent (11668019, tured in 5% CO2 conditions and differentially cultured into macrophages. On day- Invitrogen). 6 of culture, cells were washed, then cultured with [U6]-13C glucose (389374, Sigma-Aldrich), [U3]-13C pyruvate (490717, Sigma-Aldrich) or [U2]-13C acetate (282014, Sigma-Aldrich) for 24 h. Cells were washed twice in saline and lysed in Gene silencing experiments. siRNAs targeting mouse Hk1, Hk2, Hk3, Pygl, extraction solvent (80% methanol/water) for 30 min at −80 °C. After centrifugation P2ry14, Ugp2, Stat1, Gys1, G6pdx, Rarb, Pgm1, Ptpn2 and human P2RY14, RARB at 13,000 × g, 10 min at 4 °C, supernatant extracts were analyzed by LC-MS/MS as and negative control siRNAs (NC) were purchased from RiboBio (Guangzhou, ™ described previously. Briefly, liquid chromatography was performed using a HPLC China). siRNA (50 nM) was transfected into BMC using Lipofectamine RNAi- (Ultimate 3000 UHPLC) system (Thermo Fisher) equipped with An X bridge MAX Transfection Reagent (13778150, Invitrogen) according to the manu- ’ amide column (100 × 2.1 mm i.d., 3.5 μm; Waters). The column temperature was facturer s instruction. The siRNA sequences are shown as Supplementary Table 2. maintained at 10 °C. The mobile phase A is 20 mM ammonium acetate and 15 mM ammonium hydroxide in water with 3% acetonitrile, pH 9.0, and mobile phase B is Real-time PCR. Total RNA extraction was prepared with TRIzol reagent acetonitrile. The linear gradient is as follows: 0 min, 85% B; 1.5 min, 85% B, 5.5 fl (15596026, Invitrogen) and the cDNAs were generated by ReverTra Ace qPCR RT min, 30% B; 8 min, 30% B, 10 min, 85% B, and 12 min, 85% B. The ow rate was Kit (FSQ-101, Toyobo). Real-time PCR was performed for all genes with primers −1 μ 0.2 mL min . Sample volumes of 5 l were injected for LC-MS/MS analysis. LC- on a Bio-Rad CFX Connect and the data was captured using Bio-Rad CFX Man- MS/MS analysis was performed on a Q-exactive mass spectrometer (Thermo ager 2.0 software. The expression of mRNA for genes of interest was normalized to Fisher) equipped with a HESI probe, and the relevant parameters are as listed: Actb (Mus) or ACTB (Homo). The entire procedure was repeated in at least three heater temperature, 120 °C; sheath gas, 30; auxiliary gas, 10; sweep gas, 3; spray −1 biologically independent samples. The primer sequences are shown as Supple- voltage, 2.5 kV for the negative mode. A full scan ranges from 80 to 350 (mz ) was mentary Table 3. used. The resolution was set at 70,000. Data were captured using the Xcalibur™ software, version:3.0 (Thermo Fisher) and quantified by integrating the area underneath the curve of each compound using Xcalibur Qual browser (Thermo Western blot analysis. Cell lysates and pre-stained molecular weight marker were Fisher). Each metabolite’s accurate mass ion and subsequent isotopic ions were separated by SDS-PAGE, followed by transfer onto nitrocellulose membranes. The extracted (EIC) using a 10 ppm window. membranes were blocked in TBST (Tris-buffered saline with 0.5% Tween 20) containing 5% bull serum albumin (BSA) and probed with specific antibodies overnight at 4 °C. The membranes were washed three times and incubated with The level of UDPG, glycogen, and LDH. The level of UDPG, glycogen or LDH horseradish peroxidase-conjugated secondary antibodies. The immunoreactivity was measured by UDPG Detection Kit (mouse RY-03799, human RY-11791, was visualized by enhanced chemiluminescence according to the manufacturer’s Shanghai RunYu Biotech), Glycogen Assay Kit (KA0861, Abnova) or CytoTox 96 protocol (ECL Kit, 34577, Pierce). Non-Radioactive Cytotoxicity Assay (G1780, Promega), respectively, according to the manufacturer’s Instructions. Two-photon confocal microscopy. For intracellular staining, bone marrow cells were fixed in 2% paraformaldehyde for 10 min at room temperature, permeabilized NADPH/NADP+ and GSH/GSSG assay. The NADPH/NADP+ ratio was with 100 μM digitonin and blocking with 1% BSA for 1 h at 25 °C. Samples were determined with the NADP/NADPH Quantification Colorimetric Kit (KA1663, incubated with primary antibody (in PBST and 1% BSA) overnight at 4 °C. Cells Abnova). Measurements were performed according to the manufacturer’s were washed three times in PBS and incubated with secondary antibodies for 1 h at instructions. The GSH/GSSG ratio was measured by LC/MS/MS. room temperature. Nuclei were stained in DAPI solution (1 μgmL−1). The con- focal images were observed under confocal microscope (Leica SP8) and captured Detection of ROS. ROS levels were measured using CellROX™ Green Flow using LAS X Life Science Software, version: 3.0.16120.2 (Buffalo Grove, IL 60089 Cytometry Assay Kit (C10444, Invitrogen). Cells were loaded with 500 nM United States).

14 NATURE COMMUNICATIONS | (2020) 11:1769 | https://doi.org/10.1038/s41467-020-15636-8 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15636-8 ARTICLE

ELISA. TNF (900-K54, PeproTech) and IL-6 (900-K50, PeproTech) production in 9. Xie, M. et al. PKM2-dependent glycolysis promotes NLRP3 and AIM2 the supernatants were quantified by ELISA kits according to the manufacturer’s inflammasome activation. Nat. Commun. 7, 13280 (2016). protocol. Alanine amino transferase (ALT), aspartate amino transferase (AST), 10. Palsson-McDermott, E. M. et al. Pyruvate kinase M2 regulates Hif-1alpha blood urea nitrogen (BUN), creatinephosphokinase (CPK) and serum concentra- activity and IL-1beta induction and is a critical determinant of the warburg tions were measured by ELISA according to manufacturer instructions. effect in LPS-activated macrophages. Cell Metab. 21,65–80 (2015). 11. Mills, E. L. et al. Succinate dehydrogenase supports metabolic repurposing fl – ChIP-qPCR. Chromatin immunoprecipitation (ChIP) Assay Kit (53009, Active of mitochondria to drive in ammatory macrophages. Cell 167, 457 470 Motif) was utilized to examine the binding of RARβ to the Stat1 promoter. (2016). Macrophages treated with GPI or 6AN fixed with 1% formaldehyde on ice to cross- 12. Tannahill, G. M. et al. Succinate is an inflammatory signal that induces IL- link the proteins bound to the chromatin DNA. After washing, the chromatin DNA 1beta through HIF-1alpha. Nature 496, 238–242 (2013). was sheared by enzymatic force to produce DNA fragments of around 200–1000 13. Haschemi, A. et al. The sedoheptulose kinase CARKL directs macrophage bp. The same amounts of sheared DNA were used for immunoprecipitation with a polarization through control of glucose metabolism. Cell Metab. 15, 813–826 RARβ antibody (SAB4502581, Sigma-Aldrich) or an equal amount of pre-immune (2012). rabbit IgG (Gene Tex). The immunoprecipitate then was incubated with protein G 14. Thwe, P. M. et al. Cell-intrinsic glycogen metabolism supports early glycolytic Magnetic Beads, and the antibody-protein G Magnetic Beads complex was col- reprogramming required for dendritic cell immune responses. Cell Metab. 26, lected for subsequent reverse cross-linking. The same amount of sheared DNA 558–567 (2017). without antibody precipitation was processed for reverse cross-linking and served 15. Park, S. H., Park-Min, K. H., Chen, J., Hu, X. & Ivashkiv, L. B. Tumor necrosis as input control. DNA recovered from reverse cross-linking was used for PCR. PCR factor induces GSK3 kinase-mediated cross-tolerance to endotoxin in was performed with primers for the Stat1 promoter (forward, 5′-GACCAAG macrophages. Nat. Immunol. 12, 607–615 (2011). CAAGAAAAGCCATCAT-3′ and reverse, 5′-CTCGAGCCTCGGTTGGTCTA-3′) 16. Ceperuelo-Mallafre, V. et al. Adipose tissue glycogen accumulation is flanking the RARβ binding site at 59 °C for 36 cycles. associated with obesity-linked inflammation in humans. Mol. Metab. 5,5–18 (2016). Animal studies. In the LPS-induced acute peritonitis mouse model, C57BL/6J 17. Ma, R. et al. A Pck1-directed glycogen metabolic program regulates forma- mice were treated with GPI (10 μgg−1, i.p., 12 and 4 h before LPS injection), 6AN tion and maintenance of memory CD8(+) T cells. Nat. Cell Biol. 20,21–27 (15 μgg−1, i.p., 12 and 4 h before LPS injection) or clodronate liposomes (Clod, (2018). 10 μLg−1, i.p., 36 and 4 h before- and 36 h after- LPS injection), followed by i.p. 18. Meister, J. et al. The G protein-coupled receptor P2Y14 influences insulin injection of 20 μgg−1 body weight LPS. In the ConA-induced hepatitis model, release and smooth muscle function in mice. J. Biol. Chem. 289, 23353–23366 C57BL/6J mice were treated with GPI (10 μgg−1, i.v., 12 and 4 h before ConA (2014). injection), 6AN (15 μgg−1, i.v., 12 and 4 h before ConA injection) or clodronate 19. Sesma, J. I. et al. UDP-glucose promotes neutrophil recruitment in the lung. liposomes (Clod, 10 μLg−1, i.p., 36 and 4 h before- and 36 h after- ConA injection), Purinergic Signal 12, 627–635 (2016). followed by i.v. injection of 15 μgg−1 body weight ConA. In the cecal ligation and 20. Jokela, T. A. et al. Extracellular UDP-glucose activates P2Y14 Receptor and puncture (CLP)-induced sepsis mouse model, C57BL/6J mice were treated with Induces Signal Transducer and Activator of Transcription 3 (STAT3) Tyr705 − − GPI (10 μgg 1, i.p., 12 and 4 h before CLP procedure), 6AN (15 μgg 1, i.p., 12 and phosphorylation and binding to hyaluronan synthase 2 (HAS2) promoter, − 4 h before CLP procedure) or clodronate liposomes (10 μLg 1, i.p., 36 and 4 h stimulating hyaluronan synthesis of keratinocytes. J. Biol. Chem. 289, before- and 36 h after- CLP procedure), followed by CLP procedure. 18569–18581 (2014). 21. Long, C. P. & Antoniewicz, M. R. High-resolution (13)C metabolic flux – Statistics. All experiments were performed at least three times. Results were analysis. Nat. Protoc. 14, 2856 2877 (2019). expressed as mean ± SEM and analyzed by two-tailed unpaired Student t test or 22. Agius, L., Centelles, J. & Cascante, M. Multiple glucose 6-phosphate pools or one-way ANOVA. In all tests, p values of <0.05 were considered statistically sig- channelling of flux in diverse pathways? Biochem. Soc. Trans. 30,38–43 nificant. The analysis was conducted using the GraphPad Prism 8.0 software. (2002). 23. Gomis, R. R. et al. Glucose 6-phosphate produced by gluconeogenesis and by glucokinase is equally effective in activating hepatic glycogen synthase. J. Biol. Reporting summary. Further information on research design is available in Chem. 278, 9740–9746 (2003). the Nature Research Reporting Summary linked to this article. 24. Roach, P. J., Depaoli-Roach, A. A., Hurley, T. D. & Tagliabracci, V. S. Glycogen and its metabolism: some new developments and old themes. Data availability Biochem. J. 441, 763–787 (2012). The authors declare that the data supporting the findings of this study are available 25. Curtis, M. et al. Fibroblasts mobilize tumor cell glycogen to promote within the article and its Supplementary Information files and from the corresponding proliferation and metastasis. Cell Metab. 29, 141–155 (2019). author on reasonable request. The raw data for the figures and supplementary figures are 26. Puleston, D. J., Villa, M. & Pearce, E. L. Ancillary activity: beyond core presented in the Source Data file. metabolism in immune cells. Cell Metab. 26, 131–141 (2017). 27. Cracan, V., Titov, D. V., Shen, H., Grabarek, Z. & Mootha, V. K. A genetically encoded tool for manipulation of NADP(+)/NADPH in living cells. Nat. Received: 22 August 2019; Accepted: 11 March 2020; Chem. Biol. 13, 1088–1095 (2017). 28. Xiao, W., Wang, R. S., Handy, D. E. & Loscalzo, J. NAD(H) and NADP(H) Redox Couples and Cellular Energy Metabolism. Antioxid. Redox Signal 28, 251–272 (2018). 29. Blacker, T. S. & Duchen, M. R. Investigating mitochondrial redox state using NADH and NADPH autofluorescence. Free Radic. Biol. Med. 100,53–65 References (2016). 1. Orozco, L. D. et al. Unraveling inflammatory responses using systems 30. Lazarowski, E. R. & Harden, T. K. UDP-sugars as extracellular signaling genetics and gene-environment interactions in macrophages. Cell 151, molecules: cellular and physiologic consequences of P2Y14 receptor 658–670 (2012). activation. Mol. Pharmacol. 88, 151–160 (2015). 2. Chousterman, B. G., Swirski, F. K. & Weber, G. F. Cytokine storm and sepsis 31. Sesma, J. I. et al. Endoplasmic reticulum/golgi nucleotide sugar transporters disease pathogenesis. Semin. Immunopathol. 39, 517–528 (2017). contribute to the cellular release of UDP-sugar signaling molecules. J. Biol. 3. Lord, J. M. et al. The systemic immune response to trauma: an overview of Chem. 284, 12572–12583 (2009). pathophysiology and treatment. Lancet 384, 1455–1465 (2014). 32. Abbracchio, M. P. et al. Characterization of the UDP-glucose receptor (re- 4. Wolf, D. et al. A ligand-specific blockade of the integrin Mac-1 selectively named here the P2Y14 receptor) adds diversity to the P2Y receptor family. targets pathologic inflammation while maintaining protective host-defense. Trends Pharmacol. Sci. 24,52–55 (2003). Nat. Commun. 9, 525 (2018). 33. Toshchakov, V. et al. TLR4, but not TLR2, mediates IFN-beta-induced 5. Huang, X. et al. PD-1 expression by macrophages plays a pathologic role in STAT1alpha/beta-dependent gene expression in macrophages. Nat. Immunol. altering microbial clearance and the innate inflammatory response to sepsis. 3, 392–398 (2002). Proc. Natl Acad. Sci. USA 106, 6303–6308 (2009). 34. Murray, P. J. et al. Macrophage activation and polarization: nomenclature and 6. Blagih, J. & Jones, R. G. Polarizing macrophages through reprogramming of experimental guidelines. Immunity 41,14–20 (2014). glucose metabolism. Cell Metab. 15, 793–795 (2012). 35. Kolla, V., Weihua, X. & Kalvakolanu, D. V. Modulation of interferon action by 7. O’Neill, L. A., Kishton, R. J. & Rathmell, J. A guide to immunometabolism for retinoids. Induction of murine STAT1 gene expression by retinoic acid. J. Biol. immunologists. Nat. Rev. Immunol. 16, 553–565 (2016). Chem. 272, 9742–9748 (1997). 8. Wang, A., Luan, H. H. & Medzhitov, R. An evolutionary perspective on 36. Bai, L. & Merchant, J. L. Transcription factor ZBP-89 is required for STAT1 immunometabolism. Science 363, eaar3932 (2019). constitutive expression. Nucleic Acids Res. 31, 7264–7270 (2003).

NATURE COMMUNICATIONS | (2020) 11:1769 | https://doi.org/10.1038/s41467-020-15636-8 | www.nature.com/naturecommunications 15 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15636-8

37. Shang, Y., Baumrucker, C. R. & Green, M. H. The induction and activation of 62. Fernando, S. M., Rochwerg, B. & Seely, A. Clinical implications of the Third STAT1 by all-trans-retinoic acid are mediated by RAR beta signaling International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). pathways in breast cancer cells. Oncogene 18, 6725–6732 (1999). CMAJ 190, E1058–E1059 (2018). 38. Wong, L. H. et al. Isolation and characterization of a human STAT1 gene 63. Cecconi, M., Evans, L., Levy, M. & Rhodes, A. Sepsis and septic shock. Lancet regulatory element. Inducibility by interferon (IFN) types I and II and role of 392,75–87 (2018). IFN regulatory factor-1. J. Biol, Chem. 277, 19408–19417 (2002). 64. Fan, S. L., Miller, N. S., Lee, J. & Remick, D. G. Diagnosing sepsis - The role of 39. Wang, L. et al. ‘Tuning’ of type I interferon-induced Jak-STAT1 signaling by laboratory medicine. Clin. Chim Acta 460, 203–210 (2016). calcium-dependent kinases in macrophages. Nat. Immunol. 9, 186–193 (2008). 65. Singer, M. et al. The Third International Consensus Definitions for Sepsis and 40. Martinez, F. O. & Gordon, S. The M1 and M2 paradigm of macrophage Septic Shock (Sepsis-3). JAMA 315, 801–810 (2016). activation: time for reassessment. F1000Prime Rep. 6, 13 (2014). 41. Pelzel, C., Begitt, A., Wenta, N. & Vinkemeier, U. Evidence against a role for beta-arrestin1 in STAT1 dephosphorylation and the inhibition of interferon- Acknowledgements gamma signaling. Mol. Cell 50, 149–156 (2013). This work was supported by Natural Science Foundation of China (81788101, 91742112, 42. Chen, Z. et al. Negative regulation of interferon-gamma/STAT1 signaling 81661128007, 81530080, 81802847), CAMS Initiative for Innovative Medicine (2016- through cell adhesion and cell density-dependent STAT1 dephosphorylation. I2M-1-007) and The National Key Research and Development Program of China Cell Signal 23, 1404–1412 (2011). (2017YFA0105603). 43. Ten, H. J. et al. Identification of a nuclear Stat1 protein tyrosine phosphatase. Mol. Cell Biol. 22, 5662–5668 (2002). Author contributions 44. Shields, B. J., Court, N. W., Hauser, C., Bukczynska, P. E. & Tiganis, T. Cell B.H. and J.W.M. designed the study, performed experiments and data analysis, and wrote cycle-dependent regulation of SFK, JAK1 and STAT3 signalling by the protein the manuscript. K.K.W., J.W.L., J.C.L., H.Q.F., and L.P.Z. performed and/or contributed tyrosine phosphatase TCPTP. Cell Cycle 7, 3405–3416 (2008). to most of the experiments. L.Y.Z. and W.W.S. performed key mass spectrometry 45. Rao, K. M. MAP kinase activation in macrophages. J. Leukoc. Biol. 69,3–10 experiments. L.T. and J.C. (Tongji Medical College) performed RNA and ChIP- (2001). sequencing experiments. F.L., D.H.W., and P.W.X. performed the animal studies. J.C. 46. DeFranco, A. L., Hambleton, J., McMahon, M. & Weinstein, S. L. Examination (Union Hospital) collected human blood samples. K.T., H.F.Z., and Y.Y.L. performed of the role of MAP kinase in the response of macrophages to some data collection and analysis. J.X. provided reagents input. All authors read and lipopolysaccharide. Prog. Clin. Biol. Res. 392, 407–420 (1995). approved the manuscript. 47. Carter, R. L. et al. Quantification of Gi-mediated inhibition of adenylyl cyclase activity reveals that UDP is a potent agonist of the human P2Y14 receptor. Mol. Pharmacol. 76, 1341–1348 (2009). Competing interests 48. Gao, Z. G., Ding, Y. & Jacobson, K. A. UDP-glucose acting at P2Y14 receptors The authors declare no competing interests. is a mediator of mast cell degranulation. Biochem. Pharmacol. 79, 873–879 (2010). Additional information 49. Scrivens, M. & Dickenson, J. M. Functional expression of the P2Y14 receptor Supplementary information is available for this paper at https://doi.org/10.1038/s41467- in human neutrophils. Eur. J. Pharmacol. 543, 166–173 (2006). 50. Dejager, L., Pinheiro, I., Dejonckheere, E. & Libert, C. Cecal ligation and 020-15636-8. puncture: the gold standard model for polymicrobial sepsis? Trends Microbiol. Correspondence and requests for materials should be addressed to B.H. 19, 198–208 (2011). 51. Piliponsky, A. M. et al. Neurotensin increases mortality and mast cells reduce Peer review information neurotensin levels in a mouse model of sepsis. Nat. Med. 14, 392–398 (2008). Nature Communications thanks Karsten Hiller, Alan Saltiel 52. Singleton, K. D. & Wischmeyer, P. E. Distance of cecum ligated influences and the other, anonymous, reviewer(s) for their contribution to the peer review of mortality, tumor necrosis factor-alpha and interleukin-6 expression following this work. cecal ligation and puncture in the rat. Eur. Surg. Res. 35, 486–491 (2003). Reprints and permission information 53. Niu, Z. et al. Caspase-1 cleaves PPARgamma for potentiating the pro-tumor is available at http://www.nature.com/reprints action of TAMs. Nat. Commun. 8, 766 (2017). fl Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in 54. Bosmann, M. & Ward, P. A. The in ammatory response in sepsis. Trends fi Immunol. 34, 129–136 (2013). published maps and institutional af liations. 55. Nagy, C. & Haschemi, A. Time and demand are two critical dimensions of immunometabolism: the process of macrophage activation and the pentose phosphate pathway. Front. Immunol. 6, 164 (2015). Open Access This article is licensed under a Creative Commons 56. Myrtek, D. & Idzko, M. Chemotactic activity of extracellular nucleotideson Attribution 4.0 International License, which permits use, sharing, human immune cells. Purinergic Signal 3,5–11 (2007). adaptation, distribution and reproduction in any medium or format, as long as you give 57. Lattin, J. E. et al. Expression analysis of G Protein-Coupled Receptors in appropriate credit to the original author(s) and the source, provide a link to the Creative mouse macrophages. Immunome Res. 4, 5 (2008). Commons license, and indicate if changes were made. The images or other third party 58. Scrivens, M. & Dickenson, J. M. Functional expression of the P2Y14 receptor material in this article are included in the article’s Creative Commons license, unless in murine T-lymphocytes. Br. J. Pharm. 146, 435–444 (2005). indicated otherwise in a credit line to the material. If material is not included in the 59. Balmer, J. E. & Blomhoff, R. Gene expression regulation by retinoic acid. J. article’s Creative Commons license and your intended use is not permitted by statutory Lipid Res. 43, 1773–1808 (2002). regulation or exceeds the permitted use, you will need to obtain permission directly from 60. Cosgaya, J. M. & Aranda, A. Nerve growth factor activates the RARbeta2 the copyright holder. To view a copy of this license, visit http://creativecommons.org/ – promoter by a Ras-dependent mechanism. J. Neurochem. 76, 661 671 (2001). licenses/by/4.0/. 61. Kaukonen, K. M., Bailey, M., Pilcher, D., Cooper, D. J. & Bellomo, R. Systemic inflammatory response syndrome criteria in defining severe sepsis. N. Engl. J. Med. 372, 1629–1638 (2015). © The Author(s) 2020

16 NATURE COMMUNICATIONS | (2020) 11:1769 | https://doi.org/10.1038/s41467-020-15636-8 | www.nature.com/naturecommunications