4-Octyl-Itaconate and Dimethyl Fumarate Inhibit COX2 Expression and Prostaglandin Production in Macrophages

This article is featured in Top Reads, p.

Prostaglandins are key lipid mediators, which exert a wide variety of physiological roles. Their synthesis begins with the release of arachidonic acid by cytoplasmic phospholipase A₂ (cPLA₂) from membrane phospholipids, after which it is converted to PGH₂ by the cyclooxygenase (COX) enzymes. COX1 is constitutively and ubiquitously expressed and is known to play several homeostatic roles, whereas COX2 is inducible by inflammatory stimuli, including TLR ligands and cytokines. PGH₂ can then be converted into various PGs and thromboxanes by a range of synthase enzymes (1).

The widespread and effective use of nonsteroidal anti-inflammatory drugs, which inhibit COX enzymes, highlights the clinical importance of blocking PG production in inflammation (2). The proinflammatory effects of PGE₂ in particular are well characterized. The capacity of PGE₂ to induce a wide range of physiological and often pathological phenotypes is largely due to its binding to four different E prostanoid receptors (EP1–4), which vary in their tissue expression and downstream signal transduction pathways (3). PGE₂ has been shown to activate mast cells (4), Th1 cells (5, 6), and Th17 cells (7–9). PGE₂ has been implicated in inflammatory diseases, such as psoriasis (9) and rheumatoid arthritis (10), in pain responses (11, 12), and recently in aging (13). However, PGE₂ has also been shown to exert anti-inflammatory functions, particularly in the environment of the lung (14–16). Another PG, PGD₂, has been reported to contribute to the allergic response. Engagement of PGD₂ with its receptors facilitates chemotaxis and activation of eosinophils and Th2 cells during allergic disease (17–19). Thromboxanes, which are also downstream of COX activity, promote vasoconstriction and platelet aggregation (20).

Itaconate is a metabolite that has emerged in recent years as an important immunomodulator (21). It is synthesized via the decarboxylation of the Krebs cycle intermediate cis-aconitate by the enzyme aconitate decarboxylase 1 (also known as IRG1), encoded by immune responsive gene 1 (22). The expression of IRG1 is predominantly restricted to macrophages and several other immune cell types and is markedly upregulated on stimulation with TLR ligands (23), thereby leading to an increase in intracellular itaconate. 4-Octyl itaconate (4-OI) is a cell-permeable derivative that is commonly used in the study of itaconate and has been shown to be converted into itaconate intracellularly in macrophages (24).

An important aspect of itaconate biology is that both the endogenous metabolite and 4-OI have been shown to function as cysteine modifiers (25). This cysteine alkylation was originally termed 2,3-dicarboxypropylation and is also referred to as itaconation. This posttranslational modification is the basis of many of the anti-inflammatory functions associated with itaconate and 4-OI (24, 26–30). 4-OI has been shown to modify cysteine residues on KEAP1 (25), which functions as a negative regulator of the master antioxidant transcription factor NRF2. Modifications of these cysteine residues cause KEAP1 to be degraded, which liberates NRF2 and permits translocation to the nucleus (31), where NRF2 induces transcription of antioxidant genes (32) and inhibits transcription of certain proinflammatory cytokines (33). The ability of 4-OI to activate NRF2 has been implicated in several of its anti-inflammatory and protective functions (25, 34–38).

Dimethyl fumarate (DMF) is a derivative of another Krebs cycle metabolite, fumarate, that is clinically approved for the treatment of multiple sclerosis (39). Like 4-OI, DMF is also a potent cysteine modifier (cysteine alkylation by fumarate is termed succination) (40), and DMF shares some of the same targets as 4-OI, such as GAPDH (27, 41), gasdermin D (28, 30, 42), and importantly, KEAP1 (25, 43), meaning that DMF is also a potent NRF2 activator. Therefore, these two metabolite derivates often have similar effects on biological pathways.

The effect of itaconate and its derivatives on PG production has not been studied to date. In this study, we show that 4-OI greatly reduces PG production in proinflammatory macrophages through transcriptional suppression of COX2. We demonstrate that PG production and COX2 transcription are unchanged by the deletion of IRG1, indicating a difference with endogenous itaconate. However, DMF replicates the decrease in COX2 expression and PG secretion observed with 4-OI. Finally, we show that 4-OI and DMF reduce COX2 expression in an NRF2-independent manner. We have therefore uncovered a novel anti-inflammatory role of 4-OI and DMF in macrophages.

4-OI was initially supplied by Prof. Richard Hartley, and results were later confirmed with commercially available 4-OI (Sigma-Aldrich). Pam3CSK4, DMF, diethyl maleate (DEM), indomethacin, and NS-398 (Sigma-Aldrich) were also used. Abs used were anti–β-actin (Sigma-Aldrich), anti-COX2 (Abcam), anti–phospho-cPLA₂ (Ser505), anti-cPLA₂, anti-NRF2, anti-KEAP1, anti-ATF4, anti–phospho-NF-κB p65 (Ser536), anti–NF-κB p65, anti–phospho-p38 MAPK (Thr180/Tyr182), anti-p38 MAPK, anti-phospho-p44/42 MAPK (Erk1/2) (Thr202/Tyr 204), and anti-p44/42 MAPK (Erk1/2) (Cell Signaling). Anti-mouse IgG and anti-rabbit IgG secondary HRP-conjugated Abs (Jackson Immunoresearch) were also used. A PGE₂ ELISA kit was used (Enzo Life Sciences). The Silencer Select siRNAs against NRF2 (s70522), ATF4 (s62689), and Anxa1 (s69299), as well as the Silencer Select negative control, were used (Thermo Fisher Scientific).

Bone marrow–derived macrophages (BMDMs) were isolated from C57BL/6J mice (Harlan UK). Legs from NRF2 knockout, KEAP1 knockdown, and matched wild-type mice were kindly provided by Prof. Albeena Dinakova-Kostova (University of Dundee). The animals were housed under specific pathogen-free conditions in keeping with Irish and European Union regulations. All experiments carried out required prior ethical approval by Trinity College Dublin Animal Research Ethics Committee and Health Products Regulatory Authority. Mice were euthanized in a carbon dioxide chamber, after which cervical dislocation was used to confirm death. The ends of the tibia, femur, and hip bones were cut, and the bone marrow was flushed. The cells were then differentiated in DMEM containing 10% (v/v) FCS, 1% penicillin/streptomycin, and 20% L929 supernatant for 6 d. After this time the macrophages were counted and replated for use in experiments.

Thirty milliliters of whole blood was layered onto 20 ml Lymphoprep (STEMCELL Technologies) in a 50-ml conical tube and centrifuged for 20 min at 2000 rpm with no brake. The PBMCs were then isolated from the middle layer and washed twice in PBS. PBMCs were cultured in RPMI supplemented with 10% (v/v) FCS and 1% penicillin/streptomycin for use in experiments.

The media on the cells were replaced with 500 μl DMEM containing no FCS or penicillin/streptomycin. The required amount of Lipofectamine RNAiMAX transfection reagent (Thermo Fischer Scientific) and the siRNAs were diluted in DMEM containing no FCS or penicillin/streptomycin and preincubated together for 15 min. A total of 500 μl of this mix was then added to each well so that the final dilution of Lipofectamine RNAiMAX was 5 μl/ml and the final concentration of siRNA was 50 nM.

Cells were lysed in sample buffer (0.125 M Tris [pH 6.8], 10% [v/v] glycerol, 0.02% SDS) and subsequently incubated at 95°C for a duration of 5 min. The samples were resolved on SDS-polyacrylamide gels, alongside the Spectra BR protein ladder (Thermo Fisher Scientific), so that proteins could be identified by m.w. The protein was then transferred to polyvinylidene fluoride membrane, and membranes were blocked for 1 h in 5% (w/v) dried milk in TBST. The blots were then incubated overnight at 4°C with the primary Ab. After incubation for 1 h with the secondary Ab, as well as three washes in TBST before and after secondary Ab, the blots were developed using chemiluminescent substrate (Thermo Fisher Scientific).

Cells were lysed and the RNA was extracted using the PureLink RNA minikit (Ambion). cDNA was subsequently prepared using the High Capacity cDNA Reverse Transcription kit (Applied Biosystems), according to the manufacturer’s instructions. Real-time quantitative PCR (qPCR) was then performed with the resulting cDNA using a 7500 Fast Real-Time PCR System with PowerUp SYBR Green Master Mix (Applied Biosystems). All genes were normalized to Rps18 expression for mouse or RPS13 for human. The sequences of the primer pairs for murine genes that were used are as follows; Rsp18, 5′-GGA TGT GAA GGA TGG GAA GT-3′ (forward) and 5′-CCC TCT ATG GGC TCG AAT TT-3′ (reverse); Ptgs2, 5′CGG ACT GGA TTC TAT GGT GAA A-3′ (forward) and 5′-CTT GAA GTG GGT CAG GAT GTA G-3′ (reverse); Ptges, 5′-GGA AGA AGG CTT TTG CCA ACC-3′ (forward) and 5′-CGA AGC CGA GGA AGA GGA AA-3′ (reverse); Nqo1, 5′-GCT GCA GAC CTG GTG ATA TT-3′ (forward) and 5′-ACT CTC TCA AAC CAG CCT TT-3′ (reverse); Il1b, 5′-GGA AGC AGC CCT TCA TCT TT-3′ (forward) and 5′-TGG CAA CTG TTC CTG AAC TC-3′ (reverse); Il6, 5′-CCA CAG TCC TTC AGA GAG ATA CA-3′ (forward) and 5′-CCT TCT GTG ACT CCA GCT TAT C-3′ (reverse); Tnf, 5′-GCC TCT TCT CAT TCC TGC TT-3′ (forward) and 5′-TGG GAA CTT CTC ATC CCT TTG-3′ (reverse); Il10, 5′-AGG CGC TGT CAT CGA TTT-3′ (forward) and 5′-CAC CTT GGT CTT GGA GCT T-3′ (reverse); Tgfb1, 5′-CGA AGC GGA CTA CTA TGC TAA A-3′ (forward) and 5′-TCC CGA ATG TCT GAC GTA TTG-3′ (reverse); Hmox1, 5′-CCT CAC AGA TGG CGT CAC TT-3′ (forward) and 5′-GCT GAT CTG GGG TTT CCC TC-3′ (reverse); Nos2, 5′-TTC ACC CAG TTG TGC ATC GAC CTA-3′ (forward) and 5′-TCC ATG GTC ACC TCC AAC ACA AGA-3′ (reverse); Ccl2, 5′-GTT GGC TCA GCC AGA TGC A-3′ (forward) and 5′-AGC CTA CTC ATT GGG ATC ATC TTG-3′ (reverse); and Cd86, 5′-TCT CCA CGG AAA CAG CAT CT-3′ (forward) and 5′-CTT ACG GAA GCA CCC ATG AT-3′ (reverse). The sequences of the primer pairs for human genes that were used are as follows; RPS13, 5′- TCA CCG TTT GGC TCG ATA TT-3′ (forward) and 5′-GGC AGA GGC TGT AGA TGA TT-3′ (reverse); PTGS2, 5′-TGC GCC TTT TCA AGG ATG GA-3′ (forward) and 5′-CCC CAC AGC AAA CCG TAG AT-3′ (reverse); IL1B, 5′-AGC TGA TGG CCC TAA ACA GA-3′ (forward) and 5′-TGT CCA TGG CCA CAA CAA CTG A-3′ (reverse); IL6, 5′-TCT GGA TTC AAT GAG GAG ACT TG-3′ (forward) and 5′-CTC AAA TCT GTT CTG GAG GTA CT-3′ (reverse); TNF, 5′-CCA GGG ACC TCT CTC TAA TCA-3′ (forward) and 5′-TCA GCT TGA GGG TTT GCT AC-3′ (reverse); IL10, 5′-CTG TCA TCG ATT TCT TCC CTG T-3′ (forward) and 5′-TGC CTT TCT CTT GGA GCT TAT T-3′ (reverse); TGFB1, 5′-CTG CAC TAT TCC TTT GCC C-3′ (forward) and 5′-TCT TCT TCA CTA TCC CCC AC-3′ (reverse); and HMOX1, 5′-CCC AGC CCT ACA CCC GCT AC-3′ (forward) and 5′-GGT GGC ACT GGC AAT GTT GG-3′ (reverse).

Cell supernatants were harvested and PG concentrations were quantified using an ELISA kit for PGE₂ (Enzo Life Sciences), according to the manufacturer’s instructions. This item is sold as a PGE₂-specific ELISA kit; however, our data indicated that other COX-derived oxylipins were also likely to be detected with this kit. Therefore, we refer to results obtained using this ELISA kit as measuring PGs instead of PGE₂.

Immediately on harvesting, cell supernatants were snap frozen in liquid nitrogen. Samples were spiked with 2.1–2.9 ng of PGE₂-d4, PGD₂-d4, 20-HETE-d6, and thromboxane B₂ (TXB₂)-d4 standards (Cayman Chemical). Lipids were extracted by adding a 2.5 ml solvent mixture (1 M acetic acid/isopropanol/hexane; 2:20:30, v/v/v) to 1 ml supernatants in a glass extraction vial and vortexed for 30 s. A total of 2.5 ml hexane was added to samples and after vortexing for 30 s, tubes were centrifuged (500 × g for 5 min at 4°C) to recover lipids in the upper hexane layer (aqueous phase), which was transferred to a clean tube. Aqueous samples were re-extracted as described earlier by addition of 2.5 ml hexane, and upper layers were combined. Lipid extraction from the lower aqueous layer was then completed according to the Bligh and Dyer technique. Specifically, 3.75 ml of a 2:1 ratio of methanol:chloroform was added followed by vortexing for 30 s. Subsequent additions of 1.25 ml chloroform and 1.25 ml water were followed with a vortexing step for 30 s, and the lower layer was recovered after centrifugation as described earlier and combined with the upper layers from the first stage of extraction. Solvent was dried under vacuum, and lipid extract was reconstituted in 100 μl HPLC-grade methanol. Lipids were separated by liquid chromatography (LC) using a gradient of 30–100% B over 20 min (A: Water:Mob B 95:5 + 0.1% acetic acid; B: acetonitrile: methanol, 80:15 + 0.1% acetic acid) on an Eclipse Plus C18 Column (Agilent) and analyzed on a Sciex QTRAP 6500 LC tandem mass spectrometry (LC-MS/MS) system. Source conditions were temperature 475°C, ion spray voltage −4500, gas 1 60, gas 2 60, curtain gas 35. Lipids were detected using multiple reaction monitoring with the following parent-to-daughter ion transitions: PGE₂ and PGD₂ [M-H]-351.2/271.1, 15-deoxy-PGJ₂ [M-H]-315.2/271.1, and TXB₂ [M-H]-369.2/169.1. Deuterated internal standards were monitored using precursor to product ions transitions of TXB₂-d4 [M-H]-373.2/173.1, PGE₂-d4 and PGD₂-d4 [M-H]-355.2/275.1, and 20-HETE-d6 [M-H]-325.2/281.1. Chromatographic peaks were integrated using Multiquant 3.0.2 software (Sciex). The criteria for assigning a peak were signal-to-noise ratio of at least 5:1 and with at least 7 points across a peak. The ratio of analyte peak areas to internal standard was taken, and lipids were quantified using a standard curve made up and run at the same time as the samples.

Statistical significance was established by the one-way or two-way ANOVA methods as indicated in the figure legends. Data are expressed as mean ± SEM. Significance was designated as follows: *p < 0.05, **p < 0.005, ***p < 0.0005, and ****p < 0.0001. GraphPad Prism version 9 software was used for statistical analysis.

To investigate whether 4-OI might play a role in modulation of PGs, we treated murine BMDMs with 4-OI before stimulation with the TLR1/2 agonist Pam3CSK4 for 24 h, which strongly induces PGs. We first used a PGE₂ ELISA and observed a strong upregulation of cell supernatant PGs by Pam3CSK4, which was inhibited by 4-OI at concentrations as low as 25 μM (Fig. 1A). 4-OI also reduced Pam3CSK4-induced PG secretion by human PBMCs (Fig. 1B). To validate this result, we performed a lipidomic screen of oxylipins in BMDMs using quantitative LC-MS/MS. Again, stimulation with Pam3CSK4 led to a robust increase in PGE₂ that was blocked by 4-OI pretreatment (Fig. 1C); however, the concentrations were notably lower than those measured by ELISA (Fig. 1A). This is likely because of Ab-based measurements for lipid measurements displaying lower specificity and reporting on COX-derived oxylipins more broadly. From the LC-MS/MS screen, we also observed that 4-OI potently inhibited Pam3CSK4-induced increases in PGD₂ (Fig. 1D)_, 15-deoxy-PGJ₂ (Fig. 1E), and TXB₂ (Fig. 1F), in addition to PGE₂. Therefore, 4-OI was inhibiting all oxylipins downstream of COX activity. As confirmation that the ELISA was detecting COX-derived PGs in general, we found that the COX inhibitor indomethacin completely inhibited the Pam3CSK4-induced lipids from BMDMs detected by the ELISA (Fig. 1G). In addition, we used the COX2-specific inhibitor NS-398, which also inhibited all Pam3CSK4-induced lipids detected by the ELISA (Fig. 1H), indicating that PG secretion from BMDMs is almost entirely dependent on COX2.

Because we had observed that 4-OI suppressed all detected oxylipins downstream of COX activity, we next investigated whether 4-OI had an effect on COX2 expression. COX2 is potently upregulated by TLR ligands, such as Pam3CSK4, as seen by upregulation of Ptgs2 transcript, the gene that encodes COX2, as well as a strong boost in COX2 protein levels. 4-OI (200 μM) potently blocked Pam3CSK4-induced Ptgs2 levels at 4 and 8 h, the time points with the greatest induction of Ptgs2 (Fig. 2A). 4-OI also greatly decreased COX2 protein levels at both 6 and 24 h (Fig. 2B). Concentrations as low as 25 μM 4-OI reduced COX2 protein at 24 h (Fig. 2C), whereas concentrations of 50 μM or higher decreased Ptgs2 transcript at 6 h (Fig. 2D). 4-OI also reduced transcript levels of PTGS2 in human PBMCs (Fig. 2E). The reduction of COX2 expression with 4-OI is likely the reason that 4-OI inhibits PG secretion, because COX2 is often referred to as the rate-limiting enzyme for PG synthesis. 4-OI did not alter the phosphorylation levels of cPLA₂, nor did it affect total cPLA₂ (Fig. 2F). 4-OI also had no significant effect on the mRNA levels of Ptges (Fig. 2E), the gene that encodes PGE₂ synthase, the enzyme that catalyzes conversion of PGH₂ to PGE₂. Therefore, on the PGE₂ biosynthetic pathway, 4-OI seems to specifically suppress COX2 expression, which leads to a decrease in downstream PG production.

As has been previously reported (25, 27), we observed that 4-OI modulated the mRNA expression of several cytokine genes in BMDMs and PBMCs that were induced by Pam3CSK4 stimulation (Supplemental Fig. 1A–E, 1J–N), such as yielding a reduction in Il1b, Il6, and Il10 transcripts. 4-OI also reduced mRNA levels of Nos2, which encodes NO synthase (Supplemental Fig. 1G), and the chemokine Ccl2 (Supplemental Fig. 1H). Cd86 levels were unchanged (Supplemental Fig. 1I). As expected, 4-OI increased transcript levels of the NRF2-dependent gene Hmox1 in both BMDMs and PBMCs (Supplemental Fig. 1F, 1O, respectively).

We next tested whether endogenous itaconate, as well as the derivatized 4-OI, would impact COX2 expression and PG synthesis. For this we used BMDMs lacking Irg1, the gene that encodes the enzyme responsible for itaconate synthesis. However, when Ptgs2 transcript levels between Irg1^+/+ BMDMs and Irg1^−/− BMDMs are compared, there is no difference in COX2 induction by Pam3CSK4 (Fig. 3A). In addition, there were no changes in the COX-derived oxylipins PGE₂ (Fig. 3B), PGD₂ (Fig. 3C), 15-deoxy-PGJ₂ (Fig. 3D), and TXB₂ (Fig. 3E) when measured by MS/MS. This shows that endogenous itaconate does not affect COX2 expression or PG production.

We next tested whether DMF might have similar effects on COX2 and PG production to 4-OI. Pretreatment of BMDMs with concentrations as low as 5 μM DMF decreased levels of Ptgs2 transcript (Fig. 4A). DMF significantly downregulated Ptsg2 mRNA levels at 4 and 8 h (Fig. 4B), similar to 4-OI. COX2 protein levels were also attenuated by DMF, using concentrations as low as 5 μM (Fig. 4C). DMF also decreased Pam3CSK4-induced PG secretion by BMDMs (Fig. 4D). Given that 4-OI and DMF are both potent cysteine modifiers and have very similar effects regarding COX2 expression and PG synthesis, it is likely that they are acting through a shared mechanism.

4-OI and DMF are both known to be potent NRF2 activators through their ability to modify crucial cysteine residues on KEAP1, a negative regulator of NRF2. We also tested whether another NRF2 activating compound, DEM (44), would also modulate COX2. Indeed, pretreatment of BMDMs with DEM resulted in decreased COX2 protein levels (Supplemental Fig. 2A) and transcription (Supplemental Fig. 2B), in addition to suppression of PG secretion (Supplemental Fig. 2C). This indicated that perhaps NRF2 was somehow involved in the modulation of COX2 and PGs. To test this, we used BMDMs isolated from wild-type, NRF2 knockout, and KEAP1 knockdown mice. As expected, expression of the NRF2-dependent gene Nqo1 was completely ablated in the NRF2 knockout BMDMs, whereas the KEAP1 knockdown BMDMs displayed enhanced Nqo1 transcription compared with wild-type cells (Fig. 5A, 5B, Supplemental Fig. 2D). Both 4-OI and DMF maintained the capacity to suppress Ptgs2 transcription in NRF2 knockout and KEAP1 knockdown BMDMs (Fig. 5C, 5D, respectively). 4-OI and DMF also reduced COX2 protein levels in NRF2 knockout and KEAP1 knockdown BMDMs (Fig. 5E, 5F, respectively). NRF2 knockdown using siRNA also did not abolish the capacity of 4-OI to reduce COX2 expression (Supplemental Fig. 3A). Furthermore, 4-OI and DMF both blocked PG production in all genotypes (Fig. 5G, 5H, respectively). It is also worth noting that cells lacking NRF2 have lower levels of COX2 and PGs, whereas the KEAP1 knockdown cells, which exhibit augmented NRF2 activation, display elevated COX2 expression and PG production (Fig. 5E–H). Interestingly, DEM, which is considered to be a well-characterized NRF2 activator, could also still block Ptgs2 transcription in NRF2 knockout and KEAP1 knockdown BMDMs (Supplemental Fig. 2E), indicating that its ability to reduce COX2 is separate to its NRF2-activating function. We also observed that 4-OI had no effect on NF-κB p65 phosphorylation, p38 phosphorylation, or ERK phosphorylation (Supplemental Fig. 3B), all of which are signaling pathways that are known to modulate Ptgs2 transcription. We also investigated whether ATF4 was involved in the observed effect, given that ATF4 was recently shown to bind directly to the Ptgs2 promoter and induce its transcription (45). However, 4-OI actually increased ATF4 expression, and ATF4 silencing had no effect on the inhibition of COX2 by 4-OI (Supplemental Fig. 3C). We also tested whether the effect of 4-OI on COX2 might be dependent on annexin A1, which has been shown to be modified by itaconate and itaconate derivatives (25, 28, 29) and is known to function as a negative regulator of cPLA₂, which catalyzes the first step of PG biosynthesis. However, silencing annexin A1 did not alter the inhibition of PG production by 4-OI (Supplemental Fig. 3D). These results indicate that the capacity of 4-OI and DMF to suppress COX2 expression and downstream PG production is independent of NRF2 activation and other known signals that regulate COX2.

The itaconate derivative 4-OI has recently garnered much attention as an immunomodulator. The findings of this study suggest a role for 4-OI in the inhibition of PG production in macrophages activated with the TLR1/2 ligand Pam3CSK4. We show that 4-OI potently reduces several COX-derived oxylipins and provide evidence that 4-OI suppresses COX2 transcription. We demonstrate that although endogenous itaconate derived from IRG1 activity does not affect COX2 levels, DMF attenuated COX2 expression and PG secretion in a similar manner to 4-OI. This implies that a cysteine modification may be involved, but we also provide evidence that suggests that the effect of 4-OI and DMF on PG production is not via KEAP1 degradation and NRF2 activation. Further work is required to fully elucidate the mechanism by which 4-OI and DMF impair COX2 transcription.

Our knowledge of how Krebs cycle activity impacts inflammation has been rapidly expanding over the last decade. Some Krebs cycle intermediates are known to exert proinflammatory actions, such as succinate, which has been shown to stabilize HIF-1α, thereby upregulating IL-1β transcription (46). Succinate has also been reported to exacerbate certain inflammatory diseases, such as arthritis (47) and type 2 diabetes (48). Another Krebs cycle intermediate, citrate, has also been shown to alter the inflammatory response. The mitochondrial export, in addition to the breakdown of citrate, has been reported to be essential for NO and PG production (49, 50). Several anti-inflammatory roles have recently been described for both itaconate (25, 30, 51) and fumarate (41, 42), and in this article we report a novel function for derivatives of these two Krebs cycle metabolites in macrophages.

This work also highlights the importance of bearing in mind that metabolite derivatives do not always truly represent the action of the corresponding endogenous metabolites. In the case of 4-OI, the use of this derivative replicates the biological effects of endogenous itaconate in a number of cases, such as the inactivation of the NLRP3 inflammasome (24) and impairment of glycolysis (27). However, there are also incidences where 4-OI and endogenous itaconate differ in their effects, such as type I IFN production, which is boosted by endogenous itaconate (52) but inhibited by 4-OI (25). Although itaconate has also been shown to modify proteins in the same manner as 4-OI, the targets are not always the same. Our previous study (25) found that, although there were some overlapping proteins modified by both 4-OI and endogenous itaconate, many targets were mutually exclusive. This could potentially be because of differences in electrophilicity between itaconate and 4-OI.

Although DMF is known to be a potent NRF2 activator (43, 53), NRF2-independent anti-inflammatory functions of DMF have begun to emerge, such as the impairment of glycolysis through GAPDH succination (41) and inhibition of pyroptosis via gasdermin D (42). It is also true for 4-OI that a number of its currently known anti-inflammatory effects are NRF2 dependent (25, 34, 35), although others are not (24, 27, 30, 51). Interestingly, DEM, which is predominantly used as an experimental tool to pharmacologically activate NRF2, also suppressed COX2 expression in an NRF2-independent manner. Because DEM has been shown to modify cysteines on KEAP1 in a similar way to 4-OI and DMF (44), perhaps DEM also possesses the capacity to modify other reactive cysteines in the cell with further reaching implications. An observation that adds complexity to this system is that 15-deoxy-Δ^12,14-PGJ₂, which is downstream of COX activity and therefore would be reduced by 4-OI, has been shown to activate NRF2 (54). Nonetheless, our data reveal another NRF2-independent anti-inflammatory role for DMF, 4-OI, and even DEM.

There is also the possibility that the effect of 4-OI on COX2 and PGs will contribute to the attenuation of other inflammatory markers. For example, macrophages express EP receptors (55); therefore, PGE₂ can signal in an autocrine manner. We have previously shown that endogenous PGE₂ acting via the EP2 receptor is required for induction of pro–IL-1β (56). Hence the inhibition of PGE₂ secretion from the macrophage and subsequent binding to EP2 could potentially contribute to the 4-OI–induced decrease in pro–IL-1β that has been reported (25). PGE₂ has also been shown to augment IL-6 production in macrophages (57, 58), another cytokine that is blocked by 4-OI treatment (25). Therefore, inhibition of PGE₂ by 4-OI could potentially be a factor in the downregulation of IL-6. Treatment of macrophages with PGE₂ was shown to boost production of IL-10 in macrophages (59), an anti-inflammatory cytokine that is also downregulated by 4-OI. It is also conceivable that a reduction in PGs secreted by macrophages could contribute to the anti-inflammatory effects of 4-OI during in vivo models (24, 25, 27). Conversely, the capacity of 4-OI to attenuate the production of other proinflammatory mediators could potentially affect the induction of COX2. For example, IL-1β and NO, both of which are inhibited by 4-OI (25), have been reported to contribute to COX2 induction (60, 61).

Our results therefore identify 4-OI and DMF as inhibitors of PG synthesis in macrophages. This work further highlights the potential of 4-OI as a therapeutic anti-inflammatory treatment and supplements the current knowledge of the anti-inflammatory effects of DMF.

We thank Prof. Richard Hartley (University of Glasgow) for the synthesis of 4-OI. We thank Prof. Albeena Dinkova-Kostova (University of Dundee) for providing bones from NRF2 knockout, KEAP1 knockdown, and matched wild-type mice.

The authors have no financial conflicts of interest.