An ionic control of metabolic-epigenetic reprogramming links inammation to membrane potential-mediated nutrient acquisition

Metabolic and epigenetic reprogramming play pivotal roles in driving inammation, but the precise regulatory mechanisms remain minimally understood. Here we show an ionic control mediated by macrophage Kir2.1, an inwardly-rectifying K+ channel, promting lipopolysaccharide (LPS)-induced inammation. Kir2.1 blockade by the selective inhibitor ML133 or its specic deletion in macrophages suppressed the production of LPS-induced inammatory factors, such as interleukin-1β, and protected mice from LPS-induced sepsis in vivo. Kir2.1 loss-of-function led to a nutrient starvation phenotype, with impaired glucose and serine-glycine-one-carbon metabolism, whose synergy promotes the generation of S-adenosylmethionine (SAM). Accordingly, reduced SAM availability by Kir2.1 blockade decreased histone methylation at key inammatory effector loci, such as Il1b. Although the immunomodulatory effect of Kir2.1 was independent of modulation by Ca2+ ux and general signaling pathways, its loss-of-function led to a depolarized membrane potential (Vm) which decreased the surface expression of nutrient transporters, including GLUT1 and CD98. We thus identify an ionic control of metabolic-epigenetic reprogramming that links inammation to Vm-mediated nutrient acquisition and identies potential new strategies for anti-inammatory therapy. of LPS-induced inammatory effectors IL-1β by promoting nutrient import and utilization. Blockade of Kir2.1 glucose and serine-glycine-one-carbon (SGOC) metabolism, whose synergy fuels histone methylation for the expression of LPS-induced inammatory genes by supporting the generation of S-adenosylmethionine (SAM). We also uncovered an unexpected mechanism by which Kir2.1-mediated regulation of V m maintains surface nutrient transporters levels to meet the metabolic demands of inammatory macrophages. Thus, we identify an ionic-metabolic-epigenetic axis that links inammation to nutrient acquisition and provide potential new strategies of manipulating V m for anti-inammatory therapy. TLR-driven early glycolytic reprogramming the kinases TBK1-Page

. Another TCA cycle metabolite, itaconate, acts as an anti-in ammatory metabolite via nuclear factor erythroid 2-related factor 2 (Mills et al., 2018). In addition, we have previously demonstrated that LPS activates the pentose phosphate pathway, the serine synthesis pathway, and one-carbon metabolism, the synergism of which drives epigenetic reprogramming for interleukin-1β (IL-1β) expression . However, how in ammatory macrophage metabolic phenotypes re ect biochemical dependencies and how the nutrient availability orchestrates in ammatory responses are still elusive.
In contrast to rapidly-proliferating cells, macrophages are not prone to rapid proliferation after LPS stimulation, but instead produce diverse in ammatory cytokines (O'Neill et al., 2016). However, macrophages still require a constant supply of extracellular nutrients that fuel both anabolic metabolism and the synthesis of diverse in ammatory factors. To acquire extracellular nutrients such as glucose and amino acids, cells including macrophages rely on transporter proteins in the plasma membrane (McCracken and Edinger, 2013). For example, glucose transporter 1 (GLUT1)-mediated glucose Using RNA-sequencing (RNA-seq) analysis, we found that K Ca 3.1 (encoded by Kcnn4), THIK1 (encoded by Kcnk13), TWIK2 (encoded by Kcnk6), and Kir2.1 (encoded by Kcnj2) were most strongly expressed in bone-marrow-derived macrophages (BMDMs) (Fig. 1A). K Ca 3.1 plays a role in macrophage activation and migration (Feske et al., 2015), while TWIK2 and THIK1 have recently been shown to be required for NLRP3 in ammasome activation by promoting K + e ux (Di et al., 2018;Madry et al., 2018). We thus focused on the inwardly-rectifying K + channel Kir2.1. Under physiological conditions, Kir channels generate a large K + conductance at potentials negative to K + equilibrium potential (E K ) but also permit a relatively small current ow at potentials slightly more positive to E K which is essential to stabilize the resting membrane potential (V m ) (Hibino et al., 2010;Miyazaki et al., 1974;Sakmann and Trube, 1984). Among different tissues and cell types, the expression of Kir2.1 was relatively high in macrophages and other specialized macrophages including osteoclasts and microglia ( Figure S1A). To examine whether primary macrophages exhibit functional Kir2.1 channels, we performed whole-cell patch-clamp recordings from peritoneal macrophages subjected to voltage-ramps from − 120 to + 60 mV. Ba 2+ (BaCl 2 ) is usually used to block the Kir channels and we found a robust Ba 2+ -sensitive Kir current in the current-voltage relationship ( Figure S1B). ML133 is a selective Kir2 blocker, with an IC 50  Because knockout of Kir2.1 is lethal in the neonate (Zaritsky et al., 2000), its role in immune cells has not been reported using Kir2.1-de cient immune cells. We deleted Kir2.1 selectively in myeloid cells by generating Lyz2-cre-Kcnj2 f/f mice and found the ability of bone marrow cells to differentiate ex vivo into macrophages in the absence of Kir2.1 was normal in light of the expression of the macrophage surface markers CD11b and F4/80 ( Figure S1F). The proliferation of cultured Kir2.1-de cient BMDMs was also normal ( Figure S1G). However, Kir2.1 de ciency greatly impaired the Kir2.1 currents in both resting and LPS-stimulated peritoneal macrophages (Figs. 1D, 1E, 1F, and S1E), and ML133 showed no additional effect (Figs. 1D, 1E, and 1F).
Kir2.1 was reported to stabilize the resting V m of many cell types including cardiac myocytes (Sakmann and Trube, 1984), vascular smooth muscle cells (Karkanis et al., 2003;Park et al., 2008;Quayle et al., 1993), endothelial progenitor cells (Quayle et al., 1993;Zhang et al., 2019) and microglial cells (Gattlen et al., 2020). We rst adopted a real-time dynamic detection of peritoneal macrophage V m for 3 min by patch clamp experiments because of a relatively slow inhibitory effect of ML133 on Kir2.1 (Figs. 1B and 1C). The V m of peritoneal macrophages (-38.4 ± 2.8 mV ) was not changed when Kir2.1 was blocked by ML133 (-39.4 ± 3.4 mV) in the resting state. However, the V m of LPS-stimulated macrophages was dramatically changed from − 42.8 ± 2.9 mV to a much more depolarized V m of + 12.3 ± 1.9 mV by ML133, and to a lesser extent, to -29.7 ± 2.3 mV due to Kir2.1 de ciency (Figs. 1G and S1H). Moreover, the effect of ML133 was abolished in Kir2.1-de cient macrophages (Figs. 1G and S1H), further indicating the speci city of ML133 on Kir2.1. We also performed patch clamp recording for a short period of time of 30 seconds and found the V m of LPS-stimulated macrophages (-31.9 ± 1.6 mV) was depolarized to -17.3 ± 1.1 mV due to Kir2.1 de ciency (Fig. 1H). Of note, the V m of resting macrophages was also depolarized by Kir2.1 de ciency under this condition (Fig. 1H), suggesting a discrepancy in the depolarization of resting macrophages between ML133 (acute blockade) and Kir2.1 de ciency (long-term absence). Together, we conclude that Kir2.1 plays a critical role in regulating the membrane potential of in ammatory macrophages.
Kir2.1 loss-of-function suppresses LPS-and infectioninduced in ammatory cytokines and pathological in ammation To explore the role of Kir2.1 in LPS-induced in ammation, we rst treated LPS-stimulated peritoneal macrophages with ML133 and found little cytotoxicity of ML133 ( Figure S2A). Strikingly, ML133 dosedependently inhibited the LPS-induced IL-1β, but not the TNF-α (Figs. 2A and 2B). Blockade of Kir2.1 by Ba 2+ gave similar results ( Figure S2B). To more broadly explore the effect of ML133, we performed RNAseq analysis in BMDMs and found that, while Il1b was the gene most decreased by ML133, a cluster of LPS-induced in ammatory genes including Il1a, Il18, Il12a, and Il6 were also decreased, but Tnf was still not affected (Figs. 2C and S2C). Gene set enrichment analysis (GSEA) of the RNA-seq data showed a striking enrichment of 'in ammatory response' after ML133 treatment (Fig. 2D). Next, we used several genetic strategies to further investigate the role of Kir2.1 in LPS-induced in ammation. Silencing of Kcnj2 by small-interfering RNAs (siRNAs) in BMDMs or stably expressing shRNAs in immortalized BMDMs (iBMDMs) suppressed the LPS-induced IL-1β, but not the TNF-α ( Figures S2D-S2G). Peritoneal macrophages from Lyz2-cre-Kcnj2 f/f mice similarly showed a reduction in LPS-induced IL-1β compared to wild-type mice (Fig. 2E). We also treated Kir2.1-depleted peritoneal macrophages with ML133 and found the inhibitory effect of ML133 on LPS-induced IL-1α and IL-1β was greatly impaired, further suggesting the speci city of ML133 through Kir2.1 (Figures S2H). To address the function of Kir2.1 in vivo, we used an LPS-induced sepsis model which largely re ects the in ammatory functions of monocytes/macrophages. ML133 or Kir2.1 de ciency signi cantly decreased the serum levels of IL-1β and increased the survival of mice (Figs. 2F-2I). The decreased TNF-α levels in Lyz2-cre-Kcnj2 f/f mice are probably due to the contribution of IL-1β to TNF-α production in vivo (Fig. 2H) . When infected with Gram-negative bacteria, such as Escherichia coli and S. typhimurium (strain SL1344), ML133 or Kir2.1 de ciency similarly led to decreased IL-1β and IL-1α production both in vitro and in vivo, with less effect on the TNF-α (Figs. 2J, 2K, S2I, and S2J). Last, to strengthen the evidence for an antiin ammatory effect of Kir2.1 blockade in samples from patients, we used synovial uid cells from gouty patients. Gout is an in ammatory form of arthritis and IL-1 inhibitors are effective in treating patients with acute and chronic gout . We found that ML133 prevented IL-1β production in freshly-isolated synovial uid cells from gouty patients (Fig. 2L). These data together indicate that Kir2.1 stimulates LPS-induced in ammation and reveal a potential anti-in ammatory strategy by targeting Kir2.1.
Kir2.1 promotes LPS-induced glucose uptake and consumption in in ammatory macrophages Given the inconsistent inhibition of LPS-induced IL-1β and TNF-α, we predicted that Kir2.1 blockade would not affect the general signaling pathways mediated by TLR4, including the NF-κB and MAPK activation essential for both cytokines upon LPS stimulation. As predicted, ML133 or Kir2.1 de ciency had little effect on LPS-induced NF-κB and MAPK activation ( Figures S3A-S3C). Accumulating evidence indicates that the Warburg effect of aerobic glycolysis plays critical roles in driving in ammatory macrophages, in particular the LPS-induced IL-1β production O'Neill et al., 2016). We thus further investigated whether Kir2.1 drives in ammatory macrophages by modulating LPS-induced metabolic reprogramming. We rst measured real-time changes in the extracellular acidi cation rate (ECAR) and found ML133 or Kir2.1 de ciency led to a decrease in the LPS-induced long-term commitment to glycolysis (Fig. 3A). Moreover, both unbiased metabolomics pro ling and a targeted metabolomics approach revealed that ML133 decreased the LPS-induced accumulation of metabolites representing aerobic glycolysis ( Figures S3A, S3B, and S3D). Moreover, GSEA of RNA-seq data showed that 'mTORC1 signaling' and 'HIF-1α targets' were not particularly enriched by ML133 ( Figures S3E and S3F). Given that macrophages require a constant supply of extracellular glucose to support their intracellular metabolic reprogramming (Freemerman et al., 2014;Gamelli et al., 1996), we next considered the possibility that Kir2.1 promotes glucose import during LPS stimulation. Strikingly, LPS-induced glucose uptake in peritoneal macrophages was dose-dependently inhibited by ML133 in vitro (Fig. 3D). Similar results were obtained in peritoneal macrophages from Lyz2-cre-Kcnj2 f/f mice compared to wild-type macrophages (Fig. 3E). Furthermore, we assessed this effect of Kir2.1 in vivo by measuring glucose uptake of macrophages in peritoneal exudate cells (PECs) and in ammatory monocytes in peripheral mononuclear cells (PBMCs) after intraperitoneal LPS challenge. Consistently, LPS-induced glucose uptake by these cells was signi cantly lower in Lyz2-cre-Kcnj2 f/f mice than that in wild-type mice (Figs. 3F and 3G). Together, these data suggest that Kir2.1 promotes glucose uptake and consumption in in ammatory macrophages.
Kir2.1 supports glycolysis offshoots and SGOC metabolism in in ammatory macrophages and its loss-of-function leads to an amino acid starvation phenotype A key mechanism for LPS-induced glycolysis is the induction of dimeric PKM2, which enters into a complex with HIF-1α to drive IL-1β expression (Palsson-McDermott et al., 2015a). As HIF-1α activation was minimally affected by ML133 ( Figure S3F), we considered other mechanisms underlying the impaired glycolysis by Kir2.1 blockade. While often represented as a linear metabolic ux, glycolysisderived carbons also feed into several offshoots and integrate into different biosynthetic pathways (Chaneton et al., 2012;Keller et al., 2012). The pentose phosphate pathway (PPP) generates pentose sugars for nucleotide synthesis, and another three-step offshoot, the serine synthesis pathway (SSP) diverts glucose-derived carbons into serine, which is further integrated into the serine, glycine, one-carbon (SGOC) metabolic network that includes the folate and methionine cycles (Newman and Maddocks, 2017; Yang and Vousden, 2016) (Fig. 4A). Strikingly, unbiased metabolomics pro ling revealed that the key metabolites 3-phosphoserine (3PS, representing the SSP) and ribose 5-phosphate (R5P, representing the PPP), increased in response to LPS but were decreased by ML133 (Fig. 4B). In addition, the enzymes mediating the three-step SSP -Phgdh, Psat1, and Psph - (Fig. 4A), and genes previously described as master regulators of the SSP (Yang and Vousden, 2016), such as Atf4 and Mdm2, were all upregulated by ML133 (Fig. 4C), showing a phenotype similar to a compensatory increase in the SSP upon serine starvation (Maddocks et al., 2013;Ye et al., 2012). Many enzymes involved in SGOC metabolism were also upregulated in ML133-treated in ammatory macrophages ( Figure S4A). Moreover, GSEA of RNA-seq data showed that ML133-treated in ammatory macrophages were enriched in 'amino acid transport' and 'SGOC metabolism' (Fig. 4D). Using strategies of pharmacological inhibitor (acute blockade) and genetic depletion (long-term absence) may lead to certain discrepancies in downstream cellular and molecular mechanisms. To reveal the common mechanisms of both ML133 and Kir2.1 de ciency, we analyzed together the RNA-seq data from both ML133-treated and Kir2.1-depleted macrophages upon LPS stimulation. We found 236 overlapped downregulated and 163 overlapped upregulated genes (Fig. 4E).
Pathway analysis revealed that the downregulated pathways affected by both ML133 and Kir2.1depletion were mostly related to 'in ammatory response' and 'response to LPS or IL-1' (Fig. 4F). Strikingly, the upregulated pathways were related to 'response to amino acid starvation' and 'regulation of translation' (Fig. 4F). Among these upregulated genes, we found several master sensors and regulators in response to amino acid starvation, including GCN2, PERK, IMPACT, and SLC38A2 ( Figure S4B) (Broer and Broer, 2017).
We next used U-[ 13 C]-glucose tracers to con rm the changes in the SSP and SGOC metabolism when Kir2.1 was blocked during LPS stimulation. ML133 signi cantly decreased intracellular m + 6 glucose, m + 3 serine, and m + 2 glycine, suggesting that Kir2.1 supports glucose uptake and the channeling of glucose-derived carbons into the SSP upon LPS stimulation (Figs. 4G and 4H). To assess the contribution of this suppressed SSP to the anti-in ammatory phenotype of ML133, we supplemented cell , and found the suppressed IL-1β production by ML133 was partially restored ( Figure S4C). Of note, most of the accumulation of intracellular serine and methionine was unlabeled (m + 0) ( Figures S4D and S4E), indicating their accumulation is largely fueled by exogenous import during in ammation. Strikingly, the LPS-induced accumulation of unlabeled serine, glycine, and methionine (key amino acids fueling SGOC metabolism (Locasale, 2013)) was also greatly blocked by ML133 (Fig. 4I). Similarly, we used U-[ 13 C]serine tracers and found a decrease of m + 3 serine, m + 2 glycine, and m + 0 methionine due to Kir2.1 de ciency, recapitulating the impeded serine and methionine uptake ( Figure S4F). These results together highlight a role of Kir2.1 in the nutrient supply fueling glycolysis offshoots and SGOC metabolism during LPS-induced in ammation.

Kir2.1 supports S-adenosylmethionine generation and con gures histone methylation in in ammatory macrophages
Through coupling with the methionine cycle, the SGOC metabolic network acts as an integrator of nutrient status to generate diverse outputs, including the primary methyl donor S-adenosylmethionine (SAM) ( Figure S5A). The total amount of SAM was also reduced by ML133 ( Figure S5B). Using U-[ 13 C]-serine tracers, we obtained similar results that ML133 or Kir2.1 de ciency decreased the LPS-induced incorporation of U-[ 13 C]-serine-derived carbons into m + 1 to 4 SAM, as well as the total amount of SAM (Figs. 5C, 5D, and S5C). Given that Kir2.1 blockade led to impaired nutrient uptake, exogenous glucose or amino acids (serine, glycine, or methionine) that donate carbons into SGOC metabolism only had mild rescue effects on the suppressed IL-1β production by ML133 ( Figures S5D-S5F). However, addition of SAM dose-dependently restored this suppressed IL-1β production (Figs. 5E and 5F), suggesting a role of Kir2.1 in supporting LPS-induced SAM generation by providing the supply of extracellular nutrients.
SAM is the universal methyl donor for all methylation reactions in cells, which plays critical roles in the chromatin state and gene transcription (Goll and Bestor, 2005). SAM availability can directly modulate several epigenetic methylation marks required for the maintenance of downstream gene transcription H3K36me3 is one of the most dynamic histone methylation marks and its main distribution appears in a wide range of gene body regions as a SAM 'sink' , making it more sensitive to SAM availability for downstream gene expression (Wagner and Carpenter, 2012). To test whether Kir2.1induced SAM generation fuels histone methylation such as H3K36me3 upon LPS stimulation, we coupled chromatin immunoprecipitation with quantitative PCR (ChIP-qPCR) and found that LPS-induced occupancy of H3K36me3 in Il1b gene-body regions farther from the 5' end  was signi cantly decreased by ML133 or Kir2.1 de ciency (Figs. 5I and S5H). In addition, LPS-induced H3K36me3 enrichment in the gene-body regions of other in ammatory factors including Il1a, Il18, and Cxcl10 were also decreased (Figs. 5J and S5H). In contrast, H3K36me3 enrichment in the gene-body regions of Tnf was unaffected in these experiments ( Figure S5I). We thus conclude that deregulation of histone methylation marks, at least H3K36me3, contributes to the anti-in ammatory outcome of Kir2.1 loss-of-function.
Kir2.1-mediated regulation of membrane potential orchestrates metabolic-epigenetic reprogramming in in ammatory macrophages Given that monovalent cations such as K + regulate the membrane potential, which indirectly controls the  (Fig. 6A), as well as glucose uptake detected by intracellular m + 6 glucose using U-[ 13 C]-glucose tracers (Fig. 6B). The effect of elevated [K + ] e was not due to an osmotic effect, as choline chloride or mannitol did not induce a similar suppression of IL-1β production ( Figure S6L). Gramicidin is another widely used strategy to depolarize V m by forming pores permeable to both K + and Na + in the plasma membrane (Munoz-Planillo et al., 2013). As expected, gramicidin depolarized the V m of in ammatory macrophages ( Figure S6K), and it recapitulated the suppressed LPS-induced IL-1β production and glucose uptake, but not the TNF-α (Figs. 6C and 6D). Furthermore, using U-[ 13 C]-glucose, we found that elevated [K + ] e and gramicidin led to a decrease in LPS-induced m + 3 serine, m + 5 to 9 ATP, and m + 5 to 9 SAM, as well as unlabeled m + 0 serine and methionine (Figs. 6E and 6F). Elevated [K + ] e also reduced the enrichment of H3K36me3 in the gene-body regions of Il1b, Il1a, Il18, and Cxcl10 loci (Figs. 6G and 6H). Together, these data suggest that the maintenance of membrane potential may be a common mechanism that orchestrates metabolic-epigenetic reprogramming in in ammatory macrophages.  Figure S7C), nystatin partially restored the decreased level of surface GLUT1, the suppressed glucose uptake, and IL-1β production but not TNF-α when Kir2.1 was blocked by ML133 or depleted (Figs. 7H-7K and S7D). Nystatin also rescued the enrichment of H3K36me3 in the gene-body regions of Il1a, Il1b, Il18, and Cxcl10 loci in Kir2.1-depleted in ammatory macrophages ( Figure S7E). Last, nutrient transporters including GLUT1, CD98, and LAT1 are ARF6/GRP1 cargo proteins that can be recycled back to the plasma membrane via the tubular recycling endosome (Eyster et

Discussion
The coordination between metabolism and epigenetics has been proposed as a key mechanism in immunity (Chisolm and Weinmann, 2018; Phan et al., 2017), but how they are supported by nutrient availability remains poorly understood. In contrast to tumor cells or activated T cells, LPS stimulation dose not induce rapid proliferation of macrophages. However, it also creates increasing metabolic demands that would be supported by extracellular nutrients. The present work provides evidence for fundamental membrane potential control of optimizing nutrient acquisition and utilization for intracellular metabolic and epigenetic reprogramming in in ammatory macrophages ( Figure S7F). We However, given the expression of Kir2.1 also in other cell types such as endothelial cells, much work remains to determine the systemic effect of targeting Kir2.1 during in ammatory diseases.

Besides the suppressed LPS-induced in ammation, another key phenotype induced by both ML133 and
Kir2.1 depletion is nutrient starvation. Several master regulators in response to amino acid starvation such as GCN2, PERK, IMPACT, and SLC38A2 (Broer and Broer, 2017) were also upregulated. There is growing evidence that local nutrient limitations at immune effector sites can be obstacles to both antimicrobial and anti-tumor immunity (Olenchock et al., 2017). Understanding how the interactions among microenvironment factors, immune cell nutrient demands, and cellular metabolic state shape the "metabolic phenotyping" is critical to obtain a more complete understanding of immune cell metabolism.
Moreover, instead of acting alone, different types of nutrients would synergistically feed the generation of speci c immunometabolites to modulate immune functions. Although other metabolic pathways may also be regulated by Kir2.1 as the low speci city and selectivity of amino acid transporters, our study unveils an ionic control on the synergism of glucose and SGOC metabolism to epigenetically drive LPSinduced in ammation by promoting SAM generation. As a universal methyl donor, the dynamic production and utilization of SAM is critical for the regulation of gene expression by methylation reactions. As a SAM 'sink', H3K36me is one of the more dynamic histone methylation marks required for transcription elongation and splicing (Wagner and Carpenter, 2012; Ye et al., 2017). We found a different sensitivity to the SAM supply and H3K36me regulation between IL-1β (also a set of in ammatory genes) and TNF-α, which may provide an explanation for the distinct sensitivity to the metabolic control of these two important in ammatory cytokines during in ammation.
Our study additionally highlights a Kir2.1 control on the adaptations in nutrient uptake of in ammatory macrophages by dynamically regulating the surface expression of nutrient transporters, including GLUT1 (glucose) and CD98 (amino acids). Given the apical position of the nutrient transporters in metabolic pathways, these proteins are intriguing pharmacologic targets for cancer treatment. The strategy of starving cancer cells of required amino acids including serine and methionine has been proved to be effective both in mice and humans (Gao et  . Given that V m regulation on the surface retention of nutrient transporters may be a common mechanism during in ammation and cancer, it will be important to further investigate the roles of other ionic channels on nutrient uptake, as well as the nutrients acquired through those less well characterized transporters.
In sum, this current study unveils a Kir2.1 control acting on macrophage activation and shows that targeting ion channels may have implications for treating in ammatory diseases.

Experimental Model And Subject Details
Mice C57BL/6 mice were purchased from the Model Animal Research Center of Nanjing University, Lyz2-cre mice were purchased from the Jackson Laboratory, and the Kcnj2 f/f mice were kind gifts from by Professor Mark T. Nelson of the University of Vermont. Kcnj2 f/f mice were crossed with Lyz2-cre mice to obtain Lyz2-cre-Kcnj2 f/f mice. Animals were housed in a speci c pathogen-free facility in the University Laboratory Animal Center. The animal experimental protocols were approved by the Review Committee of Zhejiang University School of Medicine and were in compliance with institutional guidelines. Cells HEK293T cells were from ATCC and cultured in Dulbecco's modi ed Eagle's medium (DMEM).
To obtain Mouse peritoneal macrophages, in day 0, mice were injected peritoneally with 2.5 ml of 4% thioglycolate (Merck) medium, and in day 3 to 5, peritoneal macrophages were obtained by ush the peritoneal cavity with PBS or DMEM medium. Each mouse usually yields approx. 2 × 10 7 cells, and the non-adherent cells were discarded after macrophages adhere.
The iBMDMs were a kind gift from Prof. Shao (National Institute of Biological Sciences, China). J774.1 (from ATCC) and iBMDMs were cultured in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin. Synovial uid cells (5 × 10 5 /well) were seeded in 12-well plates in RPMI 1640 supplemented with 10% FBS. They were stimulated with 100 ng/ml LPS and inhibitors as indicated for 12 h. Then sample supernatants were used for IL-1β measurements using ELISA.

Human Subjects
Synovial uid (approximate 4-5 ml) was obtained from two gouty patients (a 36-year-old man and a 56year-old man) with serum uric acid levels > 500 mmol/l and knee effusion. The patients were not involved in previous procedures or drug tests. To use these clinical materials for research purposes, prior patient written informed consent and approval from the Institutional Research Ethics Committee of The Second A liated Hospital of Zhejiang University School of Medicine were obtained (approval no. 2018-064).
They were stimulated with 100 ng/ml LPS and inhibitors as indicated for 12 h. Then sample supernatants were used for IL-1β detection using ELISA.

Method Details
Immunoblot Analysis Cells were lysed in 2 × SDS buffer (100 mM Tris-HCl, 4% SDS, 20% glycerol, 2% 2-mercaptoethanol, and 0.05% bromophenol blue) followed by boiling for 10 min. Then samples were separated by SDS-PAGE on 12% gels, after which the proteins were transferred to nitrocellulose membranes (#28637358, Pall). The membranes were blocked for 1 h in blocking buffer (5% skimmed milk and 0.1% Tween 20 in TBS), and then incubated with primary antibodies in 5% BSA overnight. The membranes were incubated with secondary antibodies in 0.1% Tween 20 in TBS at room temperature for 1 h. To detect proteins, we used ECL blotting reagents (Thermo Fisher).

In vivo LPS challenge
Mice were injected intraperitoneally with LPS (25 mg/kg body weight) alone or along with ML133 (30 mg/kg body weight). In the sepsis model, mice were sacri ced 4 h after LPS challenge, and the serum levels of IL-1β and TNF-α were measured by ELISA (Thermo Fisher) according to the manufacturer's instructions. For mouse survival rate analysis, mice were injected intraperitoneally with LPS (20 mg/kg body weight) alone or along with ML133 (30 mg/kg body weight), then survival rates were analyzed.

Bacterial infection
For in vitro bacterial infection assay, mouse peritoneal macrophages were seeded in 12 well plate (5 × 10 5 /well) and infected with 5 × 10 6 E.coli or salmonella SL1344 for 6 h in the presence or absence of ML133 (25 µM) followed by qPCR analysis of in ammatory genes transcription. For in vivo bacterial infection assay, 8-week-old mice were injected with 1 × 10 7 bacteria peritoneally in the presence or absence of ML133 (30 mg/kg), 6 hours later, the mice were sacri ced and the serum levels of IL-1β and TNF-α were measured by ELISA (Thermo Fisher) according to the manufacturer's instructions.

Measurement of extracellular acidi cation rate (ECAR)
The Seahorse XF96 analyzer (Agilent Technologies) was used to measure extracellular ECAR. BMDMs were seeded on XF96 plates at 8 × 10 4 cells/well one day prior to the XF assay. On the day of assay, the medium was replaced with assay medium composed of XF Base Medium without phenol red (Agilent Technologies, 103335-100) supplemented with 10 mM glucose, 2 mM L-glutamine, 1 mM sodium pyruvate, adjusted to pH 7.4 and incubated at 37 °C without CO 2 45 min prior to XF assay. The assay protocol was as follows: baseline measurement with 5 cycles (mix 3 min, wait 0 min, measure 3 min); then LPS and/or ML133 was injected with the nal concentration of 500 ng/ml or 25 µM respectively and measurements continued with 10-99 cycles (mix 3 min, wait 0 min, measure 3 min). Data shown are the mean ± SD, n = 6 technical replicates. (Sigma) (supplemented with 10% dialyzed fetal bovine serum (FBS), 1% penicillin/streptomycin). Cells were treated with 500 ng/ml LPS in the presence or absence of compounds as indicated for 6 hours. For metabolite extraction, cells were washed twice with PBS and once more with 0.9% NaCl. After completely aspirating the liquid, the plates were put on dry ice, 1 ml of 80% (v/v) methanol (pre-chilled to − 80 °C) was added, and the plates were kept at − 80 °C for 2 h. The plates were scraped on dry ice, the cell lysate/methanol mixture was transferred to a 2-ml tube on dry ice, then another 0.8 ml of 80% methanol was added to wash the plate and transfer the mixture to a tube. Each tube was centrifuged at 14,000 g for 20 min at 4 °C and the metabolite-containing supernatant was transferred to a new tube and lyophilized. Metabolites were analyzed using a TSQ Quantiva Ultra triple-quadrupole mass spectrometer coupled with an Ultimate 3000 UPLC system (Thermo Fisher, CA) equipped with a heated electrospray ionization probe. Chromatographic separation was done by gradient elution on a reversed-phase UPLC HSS T3 column (  Differential expression analysis. Differential expression analysis of two conditions/groups (two biological replicates per condition) was performed using the DESeq2 R package (1.16.1). DESeq2 provide statistical routines for determining differential expression in digital gene expression data using a model based on the negative binomial distribution. The resulting P-values were adjusted using the Benjamin and Hochberg's approach for controlling the false discovery rate. Genes with an adjusted P-value < 0.05 found by DESeq2 were assigned as differentially expressed.
GO and KEGG enrichment analysis of differentially expressed genes. Gene Ontology (GO) enrichment analysis of differentially expressed genes was implemented by the cluster Pro ler R package, in which gene length bias was corrected. GO terms with corrected P-value less than 0.05 were considered signi cantly enriched by differential expressed genes.
Gene Set Enrichment Analysis (GSEA) (http://www.broadinstitute.org/gsea/index.jsp) of the expression data was used to assess enrichment of the KEGG as well as the SGOC genesets.
Stable gene-knockdown with short-hairpin RNAs We constructed pLVX-shRNAs plasmids targeting kcnj2. iBMDMs were spinfected with retrovirus encoding short-hairpin RNA for 90 min at 2500 rpm and 32 °C. Forty-eight hours after infection, the cells were selected by culture with 5 mg/mL puromycin (Sigma).
shRNAs used in the study. For continuously monitoring changes of membrane potential (Vm), whole-cell con guration was adopted using current clamp (I = 0). we recorded a total of 3 min with or without ML133 and 5 s for Highpotassium (50 mM

Membrane protein biotinylation assay
Mouse peritoneal macrophages were seeded at 4 × 10 6 /dish in 60-mm dishes and treated as indicated.
Cells were washed three times with ice-cold PBS to remove any contaminating proteins. Then cell surface proteins were biotinylated by incubating cells with Sulfo-NHS-SS-Biotin solution on ice with a concentration of 1 mg/ml in PBS. 2 h later, cells were washed three times with ice-cold PBS to remove non-reacted biotinylation reagent. Alternatively, 25-50 mM Tris (pH 8.0) may be used for the initial wash to quench any non-reacted biotinylation reagent. After being lysed and centrifuged at 12000 rpm, 10 min, the supernatant was then transferred to a 1.5 ml microcentrifuge tube containing pre-washed streptavidin magnetic beads and incubate at 4℃ overnights with rotation. Streptavidin magnetic beads were washed and lysed in SDS-PAGE Reducing Sample Buffer and the samples were used for western blot.
Flow cytometry 6 × 10 5 mouse peritoneal macrophages were seeded and stimulated under the indicated conditions. Then cells were collected by cell scraper, followed by antibody staining for 30 min at 4℃. After washing with PBS solution to exclude nonspeci c staining, the surface expression of CD98 or GLUT1 was detected by ow cytometry (ACEA NovoCyte).

Statistical Analysis
The results for q-PCR are expressed as the mean ± SD. The mouse sepsis model and the LC/MS experiments are expressed as the mean ± SEM and analyzed using two-tailed Student's t-test for two groups. The q-PCR results are representative of at least three independent experiments. For mouse survival rate analysis, GraphPad Prism7 was used to plot Kaplan-Meier survival curves and to compare survival using log-rank tests.

Declarations CONTACT FOR REAGENT AND RESOURCE SHARING
Further information and requests for resources and reagents should be directed to and will be ful lled by the Lead Contact, Di Wang (diwang@zju.edu.cn).

DATA AND CODE AVAILABILITY
The RNA-seq data in this paper have been deposited in the Gene Expression Omnibus (GEO) with accession number GEO: GSE146158.