Mitochondrial pyruvate carrier-mediated metabolism is dispensable for the classical activation of macrophages

Glycolysis is essential for the classical activation of macrophages (M1), but how glycolytic pathway metabolites engage in this process remains to be elucidated. Glycolysis leads to production of pyruvate, which can be transported into the mitochondria by the mitochondrial pyruvate carrier (MPC) followed by utilization in the tricarboxylic acid cycle. Based on studies that used the MPC inhibitor UK5099, the mitochondrial route has been considered to be of significance for M1 activation. Using genetic approaches, here we show that the MPC is dispensable for metabolic reprogramming and activation of M1 macrophages. In addition, MPC depletion in myeloid cells has no impact on inflammatory responses and macrophage polarization toward the M1 phenotype in a mouse model of endotoxemia. While UK5099 reaches maximal MPC inhibitory capacity at approximately 2–5 μM, higher concentrations are required to inhibit inflammatory cytokine production in M1 and this is independent of MPC expression. Taken together, MPC-mediated metabolism is dispensable for the classical activation of macrophages and UK5099 inhibits inflammatory responses in M1 macrophages due to effects other than MPC inhibition. Previous work using chemical inhibitors has reported the need for the mitochondrial pyruvate carrier for the classical activation of macrophages. In this study, Ran, Zhang et al. use a genetic approach to show that LPS-stimulated macrophage activation does not require the import of pyruvate into mitochondria.

Glycolysis is essential for the classical activation of macrophages (M1), but how glycolytic pathway metabolites engage in this process remains to be elucidated. Glycolysis leads to production of pyruvate, which can be transported into the mitochondria by the mitochondrial pyruvate carrier (MPC) followed by utilization in the tricarboxylic acid cycle. Based on studies that used the MPC inhibitor UK5099, the mitochondrial route has been considered to be of significance for M1 activation. Using genetic approaches, here we show that the MPC is dispensable for metabolic reprogramming and activation of M1 macrophages. In addition, MPC depletion in myeloid cells has no impact on inflammatory responses and macrophage polarization toward the M1 phenotype in a mouse model of endotoxemia. While UK5099 reaches maximal MPC inhibitory capacity at approximately 2-5 μM, higher concentrations are required to inhibit inflammatory cytokine production in M1 and this is independent of MPC expression. Taken together, MPC-mediated metabolism is dispensable for the classical activation of macrophages and UK5099 inhibits inflammatory responses in M1 macrophages due to effects other than MPC inhibition.
A growing body of knowledge demonstrates that the switch from oxidative phosphorylation to glycolysis plays a critical role in the inflammatory response of macrophages when stimulated with agents such as lipopolysaccharide (LPS) (classical activation) [1][2][3][4] . Pyruvate, the end product of glycolysis, can be transferred into the mitochondria to fuel the tricarboxylic acid (TCA) cycle as citrate or it can be metabolized into lactate. The latter route, known as aerobic glycolysis or the 'Warburg effect', is considered to be the main final pathway in classically activated macrophages, allowing them to use (repurpose) the mitochondria for reactive oxygen species (ROS) generation rather than ATP production [5][6][7] . Consequently, ROS production stabilizes hypoxia-induced factor 1α (HIF-1α) and promotes proinflammatory cytokine production [5][6][7] . The role of the alternative, mitochondrial route of pyruvate in this setting is less clear. The transport of pyruvate across the mitochondrial membrane is an active process, enabled by the MPC complex, which consists of two subunits, MPC1 and MPC2 (refs. 8,9). Notably, some studies found MPC-mediated metabolism to also be important for the activation of LPS-stimulated macrophages [10][11][12][13] . In this setting, Article https://doi.org/10.1038/s42255-023-00800-3 with different concentrations of UK5099. As shown in Fig. 2d, 100 μM UK5099, but not 2 and 10 μM UK5099, significantly reduced mitochondrial membrane potential. Impaired cellular respiration and mitochondrial membrane potential can lead to less accumulation of HIF-1α, which is critical for the activation of LPS-stimulated macrophages 5,7,18 . Indeed, 100 μM UK5099, but not 2 and 10 μM UK5099, strongly reduced HIF-1α protein levels in LPS-stimulated wild-type (WT) macrophages (Fig. 2e) by modulation of HIF-1α stabilization rather than gene expression (Fig. 2f). Besides being important for macrophage-mediated inflammation 18 , HIF-1α is an essential regulator of metabolic gene expression 19 . Indeed, herein high-dose UK5099 also reduced the expression of HIF-1α target glycolytic genes, including hexokinase 2 (Hk2) and phosphofructokinase (Pfk1) (Fig. 2g,h).
Collectively, these data show that the suppressive effect of UK5099 on the inflammatory response of M1 macrophages requires high concentrations and correlates with impairment in oxidative phosphorylation, mitochondrial membrane potential and HIF-1α levels, but seems to be unrelated to the suppressive effect on MPC.

UK5099 suppresses macrophage activation independent of MPC
To address the role of MPC in the anti-inflammatory actions of UK5099, we generated Mpc1-floxed mice by CRISPR/dCas9/gRNA (Mpc1 fl/fl ) (Extended Data Fig. 1b). Breeding with Lyz2-Cre mice yielded mice with myeloid cell-specific deletion of Mpc1 (Mpc1 ΔLysM ). As shown in Extended Data Fig. 1c,d, the Mpc1 messenger RNA level in BMDMs isolated from Mpc1 ΔLysM mice was dramatically reduced compared to those from Mpc1 fl/fl mice, whereas Mpc2 mRNA levels remained unchanged. Neither MPC1 nor MPC2 protein was expressed in Mpc1 ΔLysM BMDMs (Extended Data Fig. 1e,f) and both proteins are known to be indispensable to form a stable MPC complex 8,9,20 . To validate our knockout model in vivo, we tested Mpc1 and Mpc2 expression in freshly isolated alveolar macrophages (AMs), which are mainly LysM + cells 21 . Similarly to BMDMs, the expression of Mpc1 mRNA was abolished in AMs isolated from Mpc1 ΔLysM mice compared to those from Mpc1 fl/fl mice, whereas Mpc2 mRNA levels remained comparable (Extended Data Fig. 1g,h).
Of note, the suppressive effect of excess UK5099 on M1 macrophage proinflammatory cytokine production was not only seen in Mpc1 fl/fl macrophages but also in macrophages devoid of MPC expression ( Fig. 3a-f and Extended Data Fig. 1i). RNA-sequencing (RNA-seq) analysis confirmed that Mpc depletion had no impact on proinflammatory cytokine production contrary to UK5099 treatment ( Fig. 3g and Extended Data Fig. 2a). More than 1,700 genes including Tnf, Il6 and Il12b were changed by the treatment of excess UK5099 in LPS-stimulated Mpc1 fl/fl BMDMs, whereas only 487 genes were altered by Mpc1 depletion (Fig. 3g-i). Notably, the majority of UK5099-altered genes were observed in both Mpc1 fl/fl and Mpc1 ΔLysM BMDMs regardless of LPS stimulation (Extended Data Fig. 2b). Indeed, the patterns of UK5099-altered gene expression were similar between Mpc1 fl/fl and Mpc1 ΔLysM BMDMs (Extended Data Fig. 2c). These data indicate that UK5099 at a high dose has effects that are independent of MPC expression.
Even though we used a well-established, high-yield method for the generation of BMDMs in vitro (Extended Data Fig. 2d), BMDMs are well known to be a heterogeneous population [22][23][24][25] . For this reason, the theory has been that pyruvate transport by MPC is required for acetyl-CoA production, histone acetylation and epigenetic changes, which then facilitate inflammatory gene expression 10,11 ; however, these findings and interpretations are established based on pharmacological inhibition of MPC using α-cyano-β-(1-phenylindol-3-yl)-acrylate (UK5099) as such an inhibitor 10,12,13 . Therefore, closer examination of the MPC inhibitor UK5099 is warranted to clearly interpret the role of MPC in macrophage activation.
In this study, we confirmed that UK5099 reduces cytokine production of murine bone-marrow-derived macrophages (BMDMs) but found a clear dose dependency. UK5099 at high doses (100 μM) could effectively suppress the inflammatory response of M1 macrophages, yet low doses of 2-10 μM failed to do so, despite being equally potent in reducing the transport of glucose-derived pyruvate into the mitochondria. Studies on BMDMs derived from a murine Mpc1 conditional knockout model confirmed that the anti-inflammatory effects of UK5099 were independent of MPC. Even more, Mpc depletion in general had no impact on metabolic reprogramming, ATP generation and inflammatory cytokine production in M1 macrophages. Of further note, Mpc deficiency in myeloid cells had no impact on inflammatory responses and macrophage polarization toward M1 phenotype in the endotoxemia mice model. Based on these results, we conclude that MPC-mediated metabolism is dispensable for the classical activation of M1 macrophages and the anti-inflammatory effects of UK5099 seen at high concentrations relate to effects other than MPC inhibition.

UK5099 inhibits cytokine production in a dose-dependent manner
UK5099 has been shown to be effective in reducing the inflammatory response of macrophages at concentrations of 50-200 μM (refs. 10-13). Indeed, herein we found a clear dose-dependent effect of UK5099 on proinflammatory cytokine production above a threshold dose of 25 μM without any impact on viability (Fig. 1a-f and Extended Data Fig. 1a). The anti-inflammatory action of UK5099 has been attributed to its function as a potent MPC inhibitor, leading to the reduction of pyruvate entry into the TCA cycle, which in turn decreases the incorporation of glucose-derived carbons into histone acetylation and epigenetic modulation of inflammatory gene expression 10 . In this study, we found that UK5099 reduced the ratio of U-[ 13 C]-glucose-labelled TCA cycle metabolites starting at concentrations as low as 1 μM. This inhibitory effect reached a plateau starting at 5 μM without any further decrease in 13 C-labelling ratio at any higher concentrations ( Fig. 1g-m); however, its suppressive effect on MPC is not related with the suppressive effect on inflammatory response of M1 macrophages ( Fig. 1a-f), or the expression of global histone acetylation (Fig. 1n).
Low concentrations of UK5099 slightly reduced basal oxygen consumption rate (OCR), whereas high concentrations slightly increased it (Fig. 2a). The intracellular ATP levels were not affected by different concentrations of UK5099 (Fig. 2b,c). UK5099, however, did reduce FCCP-stimulated OCR in a dose-dependent manner (Fig. 2a) and this trend matched the dose response of UK5099 in inhibiting inflammatory cytokine production ( Fig. 1a-f).
As dysfunctional cellular respiration is known to be associated with impaired mitochondrial membrane potential [14][15][16][17] , we next tested the mitochondrial membrane potential of BMDMs after treatment an identical reaction of all cells to the LysM-Cre/loxP system may not be seen, which is pertinent for the interpretation of the results above. For instance, residual expression and function of MPC may be targeted by UK5099, affecting the average level of proinflammatory cytokines detected by RNA-seq, PCR and ELISA. To address this issue, we employed single-cell RNA-seq analysis of BMDMs in response to  LPS stimulation and UK5099 treatment. As shown in Extended Data Figs. 2e,f, bone marrow cells cultured for 7 d are mainly CD11b + F4/80 + macrophages, confirming that highly purified macrophages were generated in our system. Further, more than 99% of macrophages were LysM positive, which is in line with previous studies (Extended Data Fig. 2f ) 21 . We then compared the gene expression in CD11b, F4/80 and LysM triple-positive cells. As shown in Fig. 3j,k, most CD11b + F4/80 + LysM + Mpc1 ΔLysM BMDMs completely lost Mpc1 expression but have similar expression of Mpc2, as compared to Mpc1 fl/fl BMDMs. Of note, the expression of proinflammatory cytokines was similar between Mpc1 fl/fl and Mpc1 ΔLysM macrophages (Fig. 3l,m). UK5099 notably suppressed the expression of these cytokines in both WT macrophages  and macrophages completely void of Mpc1 expression (Fig. 3l,m). These results confirmed that UK5099 suppresses LPS-induced gene expression independently of MPC.
UK5099 at concentrations below 10 μM slightly reduced but at 100 μM UK5099 significantly decreased FCCP-stimulated OCR in both Mpc1 fl/fl and Mpc1 ΔLysM BMDMs (Fig. 4a,b). UK5099 influences Article https://doi.org/10.1038/s42255-023-00800-3 FCCP-stimulated maximum OCR more than basal and ATP-linked OCR, a portrait that is in line with oxidative phosphorylation (OXPHOS) inhibitors [26][27][28][29] . As shown in Fig. 4c-e, both Mpc depletion and UK5099 treatment at any dose effectively inhibited respiration driven by pyruvate. Of note, 2 μM UK5099 reduced respiration to similar levels to that of 100 μM UK5099 and Mpc depletion, confirming that low concentrations of UK5099 are also able to effectively block the entry of pyruvate into the TCA cycle. Therefore, the suppressive effect of 100 μM UK5099 on FCCP-stimulated respiration cannot be attributed to the inhibition of pyruvate entry into the TCA cycle, but instead may be due to the inhibition of other off-target substrates.
We then evaluated the impact of UK5099 on respiration driven by glutamate, whose transport into and subsequent metabolism in the mitochondria is independent of MPC 14, [26][27][28][29] . As expected, Mpc depletion had no impact on glutamate-driven respiration ( Fig. 4f-h); however, 100 μM UK5099, but not 2 and 10 μM UK5099, significantly reduced glutamate-driven respiration in both Mpc1 fl/fl and Mpc1 ΔLysM BMDMs, suggesting that high concentrations of UK5099 not only inhibit MPC but also suppress the utilization of glutamate. This effect is due to the inhibition of later steps of glutamate oxidation rather than reduction of glutamate/glutamine entry into the TCA cycle, as high concentrations of UK5099 did not reduce 13 C-glutamine-labelled TCA cycle metabolites ( Fig. 4i-m) and glutamine consumption (Fig. 4n).
Last, high-dose UK5099 reduced HIF-1α expression as well as the expression of HIF-1α target genes involved in glycolysis such as hexokinase 2 (Hk2) and nonmetabolic genes such as adrenomedullin (Adm) and matrix metalloproteinases (Mmps) in both Mpc1 fl/fl and Mpc1 ΔLysM macrophages ( Fig. 4o and Extended Data Fig. 2g).
Taken together, these data confirm that MPC is not an essential requirement for the anti-inflammatory and anti-metabolic effects of UK5099 on activated M1 macrophages.

UK5099 directly binds oxidoreductase activity-related proteins
To explore the potential targets of UK5099 besides MPC, we conjugated UK5099 with biotin and evaluated its bonded proteins using HuProt Proteome Microarray, which targets >21,000 human proteins encompassing 16,152 genes and covers ~81% of the human proteome 30 . A total of 156 candidate UK5099-binding proteins were identified from these microarrays (Extended Data Table 1). Gene Ontology (GO) analysis showed that these proteins are related to NAD/NADP metabolism and oxidoreductase activities (Fig. 4p), which is consistent with the inhibitory effect of high-dose UK5099 on cellular respiration. Notably, this is also in line with the Gene Set Enrichment Analysis of RNA-seq data, which showed that UK5099 treatment was positively associated with oxidoreductase activity in both Mpc1 fl/fl and Mpc1 ΔLysM BMDMs ( Fig. 4q and Extended Data Fig. 2h). These data show that UK5099 alters oxidoreductase activity by directly binding associated proteins.

Mpc depletion has no impact on metabolic reprogramming
To address the question of whether MPC-mediated transport of glycolysis-derived pyruvate is required for the activation of macrophages by LPS in general, we conducted further studies. U-[ 13 C]-glucose tracing showed that Mpc1 ΔLysM BMDMs had a significantly lower ratio of glucose-labelled TCA cycle metabolites compared to Mpc1 fl/fl BMDMs ( Fig. 5a-e), indicating effective reduction of glucose flux into the mitochondria. This was also confirmed by U-[ 13 C]-pyruvate-tracing experiments (Extended Data Fig. 3a-j). Notably, Mpc depletion resulted in an increased ratio of 13 C-labelled pyruvate but had no impact on 13 C-labelled lactate, suggesting pyruvate accumulation under conditions of MPC loss (Fig. 5f,g). While the upstream TCA cycle metabolites citrate and succinate were reduced (Fig. 5h,i), downstream metabolites were similar between Mpc1 ΔLysM and Mpc1 fl/fl BMDMs (Fig. 5j,k). There was also no difference in the total amount of intracellular pyruvate and lactate (Fig. 5l,m). Of note, Mpc depletion did not affect the metabolic reprogramming of macrophages stimulated by LPS, neither early increase in glycolysis (Fig. 6a), nor later decrease in oxidative phosphorylation (Fig. 6b). Likewise, glucose consumption and lactate production increased significantly after LPS stimulation and to similar degrees in Mpc1 ΔLysM and Mpc1 fl/fl BMDMs (Fig. 6c,d); however, more pyruvate was secreted into the medium from Mpc1 ΔLysM cells (Fig. 6e). Notably, there was no difference in ATP, ATP:ADP ratio, AMP:ATP ratio and energy charge between Mpc1 ΔLysM and Mpc1 fl/fl BMDMs before and after LPS stimulation ( Fig. 6f-i). Mpc depletion also had no impact on cell viability (Fig. 6j). The observation that oxidative phosphorylation and energy generation remained intact after Mpc depletion despite less pyruvate availability in the mitochondria to fuel the TCA cycle points to other compensatory sources. Indeed, Mpc1 ΔLysM BMDMs consumed more glutamine (Fig. 6k) and had higher glutamine-labelled TCA cycle  . P values were calculated using an unpaired, two-sided Student's t-test. Data are presented as mean ± s.e.m. Article https://doi.org/10.1038/s42255-023-00800-3 metabolites compared to Mpc1 fl/fl BMDMs (Fig. 6l-o), pointing to glutamine as a compensatory source and alternative substrate for oxidative phosphorylation.

Mpc expression is dispensable for cytokine production
We next tested whether Mpc depletion influences the inflammatory response of macrophages stimulated by LPS. Previous studies reported that MPC inhibition by UK5099 reduced the production of proinflammatory cytokines, including tumor necrosis factor (TNF)-α, interleukin (IL)-6 and IL-12 (refs. [10][11][12][13]. Genetic Mpc depletion, as in this study, however, did not change the expression of these cytokines (Fig. 7a-c) or their secretion ( Fig. 7d-f).
As the LysM-Cre/loxP system depletes genes of interest in early stage myeloid cells 21  transgenic mice to generate mice with an inducible myeloid-specific deletion of Mpc1. As shown in Extended Data Fig. 4a-c, treatment with 4-hydroxytamoxifen (4-HT) effectively induced the depletion of both mRNA and protein level of MPC1 in BMDMs isolated from Mpc1 ΔLysM-ERT mice but not those from Mpc1 fl/fl mice, while treatment with vehicle had no such effect. The mRNA level of Mpc2 was comparable after 4-HT treatment, but the deletion of the MPC2 protein in Mpc1 ΔLysM-ERT BMDMs was achieved due to the stability of MPC complex 8,9,20 . These data confirmed that the inducible knockout model was successfully established using a CreERT2/loxP system. Of note, we did not observe any differences in proinflammatory cytokine expression between control BMDMs and BMDMs that acutely lost Mpc1 expression (Extended Data Fig. 4d-i); however, UK5099 inhibited the production of these cytokines in both Mpc1 fl/fl and Mpc1 ΔLysM-ERT BMDMs (Extended Data Fig. 4j-o). Our data suggest that MPC-mediated metabolism is dispensable for the classical activation of macrophages and that metabolic adaptation does not play a role in this setting. Previous studies indicated that the conversion of glycolysis-derived pyruvate to acetyl-CoA supports histone acetylation, which is deemed critical for inflammatory gene expression in M1 macrophages 10 . In the present study, Mpc depletion in BMDMs did not decrease histone acetylation ( Fig. 7g and Extended Data Fig. 4p). Collectively, these data indicated that MPC-mediated transport of glucose-derived pyruvate into the mitochondria is not mandatory for histone acetylation or the activation of macrophages by LPS. Unlike LPS-stimulated macrophages, IL-4-induced macrophage activation relies on oxidative phosphorylation for their bioenergetic demands [1][2][3][4][5] . We thus also tested whether MPC-mediated metabolism is required for alternative macrophage polarization. Notably, we found that the depletion of Mpc1 had no impact on alternative macrophage polarization under regular culture condition as indicated by similar expression of key M2-related genes, including Arg1 and Fizz1, in Mpc1 fl/fl and Mpc1 ΔLysM BMDMs ( Fig. 7h-j). We then asked whether MPC-mediated metabolism is necessary for tumor-associated macrophage (TAM) polarization. Low glucose and high lactate levels are characteristic of the tumor microenvironment and lactate in particular has been recognized as a key driver of TAM polarization 31 . A previous study showed that the transport of lactate into the mitochondria by MPC is required for acetyl-CoA production and histone acetylation to promote TAM polarization, as pharmacological inhibition of MPC reduced key TAM-related gene expression 32 . By using a genetic approach, we obtained similar findings; Mpc1 depletion significantly decreased the expression of key TAM-related genes (Arg1 and Fizz1) under culture conditions that included the replacement of glucose by lactate before IL-4 stimulation ( Fig. 7k-m). Of note, using 13 C-lactate-tracing experiments, we confirmed that Mpc1 depletion did indeed block lactate entry into the TCA cycle (Extended Data Fig. 5a-d).

Mpc1 ΔLysM mice display intact inflammatory responses
Although we demonstrated that MPC-mediated metabolism is dispensable for macrophage activation by LPS in vitro, cell culture systems differ significantly from in vivo conditions 33 . Therefore, we tested the role of MPC-mediated metabolism in inflammatory responses and macrophage polarization in an endotoxemia in vivo model with peritoneal injection of LPS using both Mpc1 ΔLysM and Mpc1 fl/fl mice. As shown in Fig. 8a-f, the levels of both serum and peritoneal proinflammatory cytokines, including TNF-α, IL-6 and IL-12, were comparable between Mpc1 ΔLysM and Mpc1 fl/fl mice. Crossing floxed mice with Lyz2-Cre mice leads to conditional gene depletion in myeloid cells, including peritoneal tissue-resident macrophages 34 . We then tested whether Mpc1 depletion affects the polarization of peritoneal macrophages toward the M1 phenotype in vivo. As shown in Fig. 8g and Extended Data Fig. 5e, there was no difference in the expression of CD86, which is a classical marker of macrophages in response to LPS stimulation, in peritoneal macrophages of Mpc1 fl/fl and Mpc1 ΔLysM mice after LPS challenge. Therefore, Mpc depletion in myeloid cells has no impact on inflammatory response and macrophage differentiation in vivo.

Discussion
Metabolic reprogramming of immune cells has been investigated intensively in recent years, but a better understanding of how metabolism precisely controls the immune responses of these cells is still needed [1][2][3][4][5] . One aspect in question has been the role of MPC, which mediates pyruvate transfer into the mitochondria. This question has been addressed in previous studies by the use of UK5099 as a specific MPC inhibitor. In this study, we provide new data outlining that the inhibitory effects of UK5099 on M1 macrophage activation are dose-dependent and MPC-independent. Indeed, in general, MPC is not required for M1 macrophage activation in vivo and in vitro.

MPC-mediated metabolism is dispensable for the activation of LPS-stimulated macrophages
After stimulation by LPS for 18-24 h, one key metabolic signature of macrophages is reduced oxygen consumption, increased mitochondrial membrane potential and reactive oxygen production generated by the electron transport chain 5,7 . This process promotes HIF-1α stabilization to facilitate proinflammatory cytokine production 5,7 . Notably, recent studies found that short-term stimulation of BMDMs by LPS increases maximal respiratory capacity, while maintaining an intact basal respiration, and that increased respiration is required for proinflammatory cytokine production 10,11 . How glycolytic metabolites participate in this metabolic rewiring of LPS-stimulated macrophages, however, remains unclear.
MPC is located at the intersection of glycolysis, the TCA cycle and OXPHOS. MPC-mediated transport is a critical step to facilitate glucose oxidation in the mitochondria for energy production; however, little is known regarding the role of this physiologically important metabolic pathway in macrophage activation. Previous studies demonstrated that pharmacological inhibition of MPC by UK5099 attenuated glycolytic fueling of the mitochondria and the inflammatory response in LPS-stimulated macrophages [10][11][12][13] . Herein, however, we found that genetic depletion of Mpc has no effect on metabolic reprogramming and proinflammatory cytokine production in LPS-stimulated macrophages. Although Mpc depletion substantially reduced the transport of glycolysis-derived pyruvate into the TCA cycle, TCA cycle metabolites can be compensated for by other sources, such as glutamine and therefore maintained a functional oxidative phosphorylation to support inflammatory responses. Indeed, previous studies indicated that the main source of accumulated TCA cycle metabolites in LPS-stimulated macrophages was glutamine rather than glucose 6 . In addition, macrophages that lack MPC expression show intact glycolysis with similar level of increase in response to LPS stimulation as control macrophages. Stimulation of glycolysis generates many building blocks to support macrophage activation by increasing several     Article https://doi.org/10.1038/s42255-023-00800-3 metabolic pathways, including the pentose phosphate pathway and de novo serine synthesis 35,36 . Collectively, our study supports the notion that the metabolic reprogramming of macrophages is flexible and does not depend on MPC in LPS-activated macrophages.

Histone acetylation and macrophage activation
The role of histone acetylation in macrophage activation by LPS is controversial. Previous studies observed that LPS stimulation increases the incorporation of glucose-derived carbons into histone acetylation and that pharmacological inhibition of glycolysis by 2-DG or MPC by UK5099 reduces proinflammatory cytokine production. These observations led to the conclusion that the transport of glycolysis-derived pyruvate into the mitochondria is essential for acetyl-CoA production, histone acetylation and the activation of LPS-stimulated macrophages 10,11 ; however, whether these inhibitory effects impact global levels of (rather than glucose-incorporated) histone acetylation is unknown. Our study found that both Mpc depletion and UK5099 treatment does not decrease histone acetylation, even though intracellular citrate levels were reduced in both conditions. Coincidently, recent studies showed that genetic knockdown or depletion of ATP citrate lyase (Acly), which catalyzes citrate into acetyl-CoA in the cytoplasm, even increases cytokine production in LPS-stimulated macrophages [37][38][39] . Further, the different results between present and previous studies cannot be attributed to metabolic adaptation, as acute loss of Mpc1 in macrophages resulted in similar findings. There are two possible reasons for these data. First, cytosolic acetyl-CoA rather than mitochondrial acetyl-CoA is the substrate for histone acetylation. Although Mpc depletion resulted in less intracellular citrate, cytosolic acetyl-CoA can be compensated by accumulated pyruvate. Further, pyruvate can be catalyzed into acetate and, in turn, to acetyl-CoA in the cytosol 40 . Indeed, we found that Mpc depletion blocks the entry of pyruvate into the mitochondria and leads to more pyruvate release. Second, it is unknown whether histone acetylation is strictly determined by the level of acetyl-CoA, which requires further investigation.
One of the main agents used in these studies is UK5099, a potent inhibitor of MPC, thought originally invented as the analog of α-cyano-4-hydroxycinnamate (CHC) long before the discovery of MPC [45][46][47] . Studies found that it could reduce the entry of pyruvate into the mitochondria (with a half-maximum inhibitory concentration (IC 50 ) of 50 nM in rat heart mitochondria) 45  could suppress inflammatory cytokine production 10,12,13 , leading to the conclusion that mitochondrial entry of pyruvate produced from enhanced glycolysis is critical for the activation of macrophages by LPS. In the present study, by using mice with genetic depletion of Mpc, we found that the inhibition of inflammatory cytokine production in LPS-stimulated macrophages by UK5099 is independent of MPC but shows concentration dependency. Selective MPC inhibitory effects of UK5099 are seen only in a narrow, low-dose range (<10 μM). Thus, conclusions drawn from studies on mitochondrial pyruvate transfer using higher doses of UK5099 than this should be evaluated carefully.

Excessive UK5099 impairs OXPHOS and HIF-1α stabilization
Although UK5099 suppresses the inflammatory responses of LPS-stimulated macrophages due to its off-target effects rather than MPC inhibition, elucidating its true targets and mechanism of action is important to identify new compounds to treat inflammatory diseases. UK5099 at high concentrations (for example 100 μM) effectively inhibits the production of key proinflammatory cytokines, including TNF-α, IL-6 and IL-12, suggesting that appropriate structural modification of this drug could be a promising strategy to curb excessive inflammatory responses. Of note, we found that UK5099 at high concentrations not only reduces the transport of glucose-derived pyruvate into the TCA cycle but also suppresses the oxidation of other substrates, such as glutamate. As the transport and metabolism of glutamate is independent of MPC, the inhibition of glutamate oxidation indicates that UK5099 might target cellular respiration. Notably, proteome microarray data and RNA-seq analysis showed that UK5099 alters oxidoreductase activity by directly binding to associated proteins. On the other hand, it is unknown whether MPC inhibition is required in combination with the off-targets of UK5099 to suppress inflammatory responses. Although MPC inhibition alone is unable to reduce proinflammatory cytokine production in LPS-stimulated macrophages, it is possible that it further decreases available substrates for OXPHOS when cellular respiration is impaired by high concentrations of UK5099. Further studies are required to investigate how macrophages precisely control metabolic reprogramming in mitochondria to fine tune inflammatory responses.
In conclusion, we found that MPC-mediated metabolism is dispensable for the activation of LPS-stimulated macrophages. UK5099 inhibits the inflammatory response in macrophages, but this is due to its off-target effects rather than MPC inhibition.

Mouse strains
Mpc1 fl/fl mice were generated by CRISPR/Cas9, which targets exon 3-5 of the Mpc1 gene of mice in C57BL/6J background. Mpc1 fl/fl mice were then crossed with Lyz2-Cre or Lyz2-CreERT2 transgenic mice (The Jackson Laboratory) to generate mice with a myeloid-specific deletion of MPC1 (Mpc1 ΔLysM or Mpc1 ΔLysM-ERT ). Female Mpc1 ΔLysM or Mpc1 ΔLysM-ERT (6-8 weeks of age) and their littermate control mice (Mpc1 fl/fl ) were used in this study. WT BMDMs were isolated from female WT C57BL/6J mice (6-8 weeks of age). The mice were bred and maintained under pathogen-free conditions with ad libitum access to food and water. This study was approved by the Tongji University Institutional Animal Use and Care Committee.

Bone-marrow-derived macrophage isolation
BMDMs were isolated as previously described [22][23][24] . Briefly, mice were individually killed by CO 2 inhalation and both the femur and tibia were isolated and placed in PBS after removing the skin and muscle. Bone marrow cells were then flushed out from the bone, resuspended and grown in RPMI-1640 medium (10% heat-inactivated FBS, and 1% penicillin and streptomycin) containing 20 ng ml −1 M-CSF. Half the volume of fresh growth medium was added on day 4. To induce loss of Mpc1, 4-hydroxytamoxifen (2 μM) and vehicle were added to Mpc1 fl/fl and Mpc1 ΔLysM-ERT cells at day 4 to allow the Cre recombinase to delete Mpc1. After 7 d in culture, macrophages were collected and plated for further experimentation.

Macrophage activation
BMDMs were seeded at 1 × 10 6 ml −1 for experiments unless stated otherwise. To induce the classical activation of macrophages, the cells were treated with or without different concentrations of UK5099, followed by stimulation with LPS (100 ng ml −1 ) for the indicated times. To induce the alternative activation of macrophages, Mpc1 fl/fl and Mpc1 ΔLysM BMDMs were incubated in glucose-free medium supplemented with 10 mM glucose or 10 mM lactate for 1 h, followed by stimulation with IL-4 (20 ng ml −1 ) for the indicated times.

Metabolism assay
ECAR and OCR of BMDMs were analyzed with an XF96 Extracellular Flux Analyzer (Agilent) as described previously 23,24 . Briefly, WT, Mpc1 fl/fl or Mpc1 ΔLysM BMDMs were seeded in XF96 Cell Culture Microplates at 8 × 10 4 cells per well for 24 h before the experiments. Cells were treated with or without different concentrations of UK5099 for 1 h, followed by stimulation with LPS for 4 h or 24 h. Cells were then switched to Seahorse medium and ECAR and OCR were measured at baseline and after the injection of different inhibitors (oligomycin, FCCP, rotenone and antimycin A).
To measure substrate specific respiration, BMDMs were permeabilized and provided with different oxidizable substrates (26)(27)(28). Briefly, Mpc1 fl/fl and Mpc1 ΔLysM BMDMs were seeded in XF96 Cell Culture Microplates at 8 × 10 4 cells per well for 24 h before the experiments. Cells were treated with or without different concentrations of UK5099 for 1 h, followed by stimulation with LPS for 4 h. Cells were then washed twice and switched to MAS buffer (70 mM sucrose, 220 mM mannitol, 10 mM KH 2 PO 4 , 5 mM MgCl 2 , 2 mM HEPES and 1 mM EGTA; pH 7.2) containing 1 nM XF Plasma Membrane Permeabilizer (XF PMP, Agilent) to selectively permeabilize the plasma membrane. For pyruvate-dependent respiration, cells were provided with 5 mM pyruvate, 0.5 mM malate and 2 mM DCA. For glutamate-dependent respiration, cells were given 10 mM glutamate and 5 mM malate. ECAR and OCR were measured at baseline and after the injection of different inhibitors (oligomycin, FCCP, rotenone and antimycin A). Intracellular metabolites were measured by GC-MS 23,24 . Briefly, 2 × 10 6 WT, Mpc1 fl/fl or Mpc1 ΔLysM BMDMs were seeded in 35-mm Petri dishes for 24 h. Cells were treated with or without different concentrations of UK5099 for 1 h, followed by stimulation with LPS for 4 h. After removal of culture medium, cells were washed three times with ice-cold saline, quenched with −20 °C 1:1 water:methanol and flash frozen by rapid immersion of dishes in liquid nitrogen. Frozen cells were scraped from culture dishes kept on dry ice, frozen in liquid nitrogen and thawed on ice for 20 min. After vortexing, samples were centrifuged at 10,000g for 10 min at 4 °C. The supernatant was then collected. The extraction procedure was repeated, followed by the supernatant combined and completely dried in a SpeedVac concentrator, methoximated using 10 μl MOX Reagent at 30 °C for 90 min and then derivatized using 40 μl MSTFA + 1% TMCS (N-methyl-N-trimethylsilyltrifluoroacetamide with 1% trimethylchlorosilane) at 37 °C for 30 min. Metabolite levels were determined using GC-MS (Hewlett-Packard, HP 5980B) with a DB5-MS column. GC-MS spectra were deconvoluted using AMDIS software, then SpectConnect software was used to create the metabolite peaks matrix. The Agilent Fiehn GC/MS Metabolomics RTL Library was used for metabolite identification. Ion count peak area was used for analysis of the relative abundance of the metabolites.
Metabolite levels in the culture medium were analyzed by NMR spectroscopy to determine the consumption or the production of certain substrates 23,24 . Briefly, WT, Mpc1 fl/fl or Mpc1 ΔLysM BMDMs were treated with or without different concentrations of UK5099 for 1 h, followed by stimulation with LPS for 24 h. Original and cell culture medium were collected and deproteinized before NMR analysis. In Article https://doi.org/10.1038/s42255-023-00800-3 150 μl medium samples, 450 μl cold methanol was added. The samples were spun down at 10,000g for 10 min at 4 °C, supernatants were collected and dried down in centrifugal vacuum evaporator for 12 h. To dried samples, 500 μl 0.1 M phosphate buffer and 50 μl 1 mM TSP-d 4 solution in D 2 O were added. Samples were vortexed for 20 s and transferred to 5-mm NMR tubes. NMR spectra were acquired on a Bruker 500 MHz Avance III HD spectrometer equipped with a BBO cryoprobe and SampleCase auto sampler (Bruker Biospin). 1 H-NMR spectra were recorded using 1D NOESY pulse sequence with pre-saturation (noesygp-pr1d), with a 90-degree pulse (~13 μs), 4.68-s acquisition time and 4-s relaxation delay. Spectra were phase and baseline corrected using Topspin v.3.5 software. Metabolites were identified and quantified, using Chenomx NMR Suite v.8.2 by fitting the spectral lines of library compounds into the recorded NMR spectrum of the cell extracts. The quantification was based on peak area of TSP-d 4 signal and metabolite concentrations were reported as μM in the medium. Samples were collected and processed as described for intracellular metabolite measurement. Metabolites were quantified based on total ion count peak area of specific mass ions. To determine 13 C-labelling, mass information for known fragments of labelled metabolites was retrieved. These fragments contained either the whole or partial carbon skeleton of the metabolite. For each fragment, the retrieved data consisted of mass intensities for the lightest isotopomer (without any heavy isotopes, M + 0) and isotopomers with increasing unit mass (up to M + 6) relative to M0. These mass distributions were normalized by dividing by the sum of M0 to M6 and corrected for the natural abundance of heavy isotopes, using matrix-based probabilistic methods as described previously 46 and implemented in Microsoft Excel. 13 C-labelling data are expressed as fractional abundance of each isotopolog of a measured metabolite pool or relevant enrichment of each metabolite. Data were from three biological replicates.

Nucleotide measurements
Intracellular nucleotides were measured by HPLC 24 . WT, Mpc1 fl/fl or Mpc1 ΔLysM BMDMs were plated at 0.5 × 10 6 cells per ml in 35-mm Corning dishes for 24 h. Cells were then treated with or without different concentrations of UK5099 for 1 h and subsequently stimulated with LPS for 4 h. After culture medium was removed, cells were quickly rinsed with PBS, quenched with 6% ice-cold HClO 4 and flash frozen by rapid immersion of dishes in liquid nitrogen. The cells were stored at −80 °C before nucleotide analysis. Frozen cells were scraped from culture dishes and transferred to a micro-centrifuge tube chilled in liquid nitrogen. The frozen cells were homogenized using a Kimble homogenizer (Sigma) to extract the nucleotides. The homogenate was centrifuged at 10,000g for 10 min at 4 °C. The supernatant was collected and neutralized to pH 7.4 with 2 M KHCO 3 (100:35, v/v). After centrifuging at 10,000g for 10 min at 4 °C, nucleotides in the supernatant were separated on a reverse-phase Discovery C18 column (Sigma) with the Agilent 1290 HPLC system (Agilent), as described previously 24 . Briefly, a phosphate buffer with tetrabutylammonium sulfate and methanol mixture was used as mobile phase at 0.7 ml min −1 flow rate. Gradient elution was applied and the separation of nucleotides was completed in 10 min. The nucleotide levels were normalized by cell number.

Mitochondrial membrane potential measurement
Cells were plated at 0.5 × 10 6 cells per ml in 12-well plates and treated with different concentrations of UK5099 for 1 h, followed by stimulation with LPS for 4 h. After removing culture medium, cells were washed and incubated in PBS containing 50 nM MitoTracker Red (Thermo Fisher Scientific) at 37 °C in the dark for 30 min. After three washes, cells were then collected in 500 μl PBS by using a cell scraper and transferred to polypropylene FACS tubes. Aqua fluorescent reactive dye (Thermo Fisher Scientific) was applied to excluded dead cells. Cells were then analyzed using a LSR Fortessa flow cytometer and data were analyzed using FlowJo software.

Gene expression analysis by 3′ RNA sequencing
Mpc1 fl/fl and Mpc1 ΔLysM BMDMs were seeded in 12-well plates (1 × 10 6 per well) and stimulated with or without LPS for 4 h after pre-treatment with UK5099 (100 μM) or dimethylsulfoxide (DMSO) for 1 h. After washing with PBS, cells were collected in Trizol (Thermo Fisher Scientific) and stored in liquid nitrogen. RNA was extracted and a total amount of 1 μg RNA per sample was used as input material for the RNA sample preparations. Briefly, mRNA was purified from total RNA using poly-T oligo-attached magnetic beads. Fragmentation was carried out using divalent cations under elevated temperature and the library fragments were purified with AMPure XP system (Beckman Coulter). Next, PCR was performed with Phusion High-Fidelity DNA polymerase, Universal PCR primers and Index (X) Primer. Last, PCR products were purified (AMPure XP system) and library quality was assessed on the Agilent Bioanalyzer 2100 system. The clustering of the index-coded samples was performed on a cBot Cluster Generation System using TruSeq PE Cluster Kit v3-cBot-HS (Illumina) according to the manufacturer's instructions. After cluster generation, the library preparations were sequenced on an Illumina Novaseq platform and 150-bp paired-end reads were generated.

scRNA-seq library preparation and data processing
BMDMs were plated at 1 × 10 6 cells per ml in six-well plates and treated with 100 μM UK5099 for 1 h, followed by stimulation with LPS for 4 h. Cells were then collected and made into cell suspension and the single-cell library was constructed on the 10x Genomics platform. After sequencing, fastq files were quantified to expression matrix through Cell Ranger (v.7.0.0) in Linux. Barcode, features, matrix files were regarded as input files of Seurat 48 (v.4.1.1). Considering the presence of double droplet, empty droplet, dead cell, low quality cells with the nFeature_RNA > 4,000, percent.mt > 10 and nFeature_RNA < 1,000 were dropped out. UMAP computation was performed with the RunUMAP function in the dims range from 1 to 10. The difference between groups was calculated using the FindMarker function in the Seurat package using the Wilcoxon rank-sum test. Other calculations and visualizations were performed in R-base (v.4.2.1).

Western blotting
After different treatments and stimulation, cells were lysed in RIPA buffer containing Complete Mini EDTA-Free protease inhibitor cocktail and phosphatase inhibitor cocktail (Roche). Protein concentrations of cell lysate were measured by Pierce BCA protein assay kit (Thermo Fisher Scientific) and equal amounts of total protein from different samples were used. Cell lysate was then boiled in SDS sample buffer for 5 min at 95-100 °C, separated by SDS-PAGE and transferred to PVDF membranes (Bio-Rad). The membranes were blocked in TBS plus 5% nonfat dry milk (Bio-Rad) for 1 h and then incubated with primary antibodies overnight at 4 °C. After three washes, the membranes were incubated with HRP-linked secondary antibodies for 1 h at room temperature in TBS-T plus 5% nonfat dry milk. ECL western blotting Article https://doi.org/10.1038/s42255-023-00800-3 chemiluminescent substrates (Thermo Fisher Scientific) were added for 5 min after washing three times. Bands of interest were developed by using an autoradiographic film. Blot bands were quantified by densitometry using ImageJ software.

ELISA
Cells were seeded in 24-well plates (5 × 10 5 per well) and treated with different reagents before stimulation with or without LPS for the indicated times. The supernatants were collected and stored at −80 °C for further measurements. Concentrations of cytokines in supernatants were determined by ELISA kits (R&D systems) according to the manufacturer's instructions.

Quantitative real-time PCR
Cells were seeded in 24-well plates (5 × 10 5 per well) and treated with different reagents before stimulation with or without LPS for indicated time. Total RNA was extracted by using RNeasy Plus Mini kit with DNase treatment (QIAGEN). Equal amounts of RNA from different samples were reversely transcribed into cDNA using the RevertAid First Strand cDNA Synthesis kit with random hexamer primer (Thermo Fisher Scientific). Quantitative real-time PCR was performed using the PowerUp SYBR Green Master Mix (Thermo Fisher Scientific) on a QuantStudio 6 Flex System (Applied Biosystems). Gene expression levels were normalized to the housekeeping gene GAPDH. The primer sequences used for gene expression analysis were listed in Extended Data Table 2.

Cell viability
Cell viability was tested by the XTT assay (Cell Signaling Technology). Briefly, 1 × 10 5 cells per well of BMDMs were seeded in 96-well plates and treated with different reagents before stimulation with or without LPS for the indicated times. Each well was added with 50 μl XTT detection solution and absorbance was read at 450 nm after incubation at 37 °C for 3 h.

Proteome microarray assays
HuProt human proteome microarrays were obtained from the Johns Hopkins Medical Institutions Protein Microarray Core (CDI Laboratories). For microarray proteome analysis, the microarrays were blocked by 5% BSA in TBS-T (pH 7.5) buffer for 1.5 h. Biotin and Biotin-UK5099 (10 μM) were then added to the proteome microarrays and incubated at room temperature for 1 h. After three washes, Cy5-Streptavidin was added (1:1,000 dilution) and microarrays were scanned by Axon GenePix 4000B. The data were extracted and analyzed by GenePix Pro v.6.0 software (Axon Instruments). GO analysis was performed using cluster Profiler package in RStudio.

Endotoxin-induced model of sepsis
Male and female mice aged 10-12 weeks were used for in vivo LPS challenge experiment. Briefly, sex-and age-matched Mpc1 fl/fl and Mpc1 ΔLysM mice were randomly assigned to the vehicle control and LPS groups. Mice were peritoneally injected with LPS (15 mg kg −1 ) to induce an endotoxemia model. Serum and peritoneal lavage fluid samples were collected at 6 h after injection. Proinflammatory cytokine levels in serum and peritoneal lavage fluid were determined by ELISA.

Flow cytometry
BMDMs were collected at day 7 of in vitro culture and peritoneal cells were collected from endotoxemia mice models. Cells were centrifuged and resuspended in PBS containing 1% BSA and 1 mM EDTA. After incubating with anti-mouse CD16/32 for 10 min at room temperature to block FC receptors, BMDMs were stained with F4/80 and CD11b and peritoneal cells were stained with CD45, Ly6G, F4/80, CD11b and CD86 antibodies on ice for 20 min in the dark. After washing twice, cells were assayed by an Arial II-Optics flow cytometer and data were analyzed using FlowJo v.10.0. BMDMs were analyzed for F4/80 and CD11b expression. Peritoneal macrophages were gated as CD45 + Ly6G-CD11b + F4/80 + and CD86 expression was compared between Mpc1 fl/fl and Mpc1 ΔLysM mice.

Mitochondrial ROS and mitochondrial membrane potential measurements
After different treatments and stimulation, cells were washed with warmed PBS and then incubated with MitoSOX (Invitrogen) at 5 μM or MitoTracker Red FM (Invitrogen) at 100 nM for 30 min in PBS at 37 °C. After washing with warm PBS (37 °C), cells were collected and analyzed by FACS immediately. Unstained cells were processed with the same procedure but without staining with MitoSOX red or MitoTracker Red FM. For the negative control for mitochondrial membrane potential, carbonyl cyanide 3-chlorophenylhydrazone was added 5 min before the incubation with MitoTracker Red FM.

Statistical analysis
Results are presented as mean ± s.e.m. An unpaired Student's t-test was used to test the differences between two groups, based on the assessment of variance of the data. One-way or two-way ANOVA was used to test the differences between three or more groups. All data were analyzed by GraphPad Prism software (v.9). Detailed statistical values are provided in the figure legends.

Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability
The raw sequence data reported in this paper have been deposited in the Genome Sequence Archive (GSA) in the National Genomics Data Center, China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences, under accession code GSA: CRA009169 (RNA-seq data) and GSA: CRA009183 (single-cell RNA-seq data), which are publicly accessible at https://ngdc.cncb.ac.cn/gsa (refs. 49,50). Other data that support the findings of this study are available from the corresponding authors upon reasonable request. Source data are provided with this paper.  Differentially identified proteins from human proteome microarray between biotin-UK5099 and biotin.