BCAT1 controls metabolic reprogramming in activated human macrophages and is associated with inflammatory diseases

Branched-chain aminotransferases (BCAT) are enzymes that initiate the catabolism of branched-chain amino acids (BCAA), such as leucine, thereby providing macromolecule precursors; however, the function of BCATs in macrophages is unknown. Here we show that BCAT1 is the predominant BCAT isoform in human primary macrophages. We identify ERG240 as a leucine analogue that blocks BCAT1 activity. Selective inhibition of BCAT1 activity results in decreased oxygen consumption and glycolysis. This decrease is associated with reduced IRG1 levels and itaconate synthesis, suggesting involvement of BCAA catabolism through the IRG1/itaconate axis within the tricarboxylic acid cycle in activated macrophages. ERG240 suppresses production of IRG1 and itaconate in mice and contributes to a less proinflammatory transcriptome signature. Oral administration of ERG240 reduces the severity of collagen-induced arthritis in mice and crescentic glomerulonephritis in rats, in part by decreasing macrophage infiltration. These results establish a regulatory role for BCAT1 in macrophage function with therapeutic implications for inflammatory conditions.


Statistical analysis (or indications in figures) is lacking in
Dosing in vitro should be discussed further-20mM in vitro is a high dose if IC50 is in nM range? Why 20mM? ATP production should be marked as "estimated ATP production". The Y axis shows that it is a calculation from OCR experiment-not an actual measure of ATP. This is misleading.
A cartoon of metabolites regulated would aid the reader.
Add a citation for "as the public datasets for human immune cell expression indicate," Reviewer #2 (Remarks to the Author): In their manuscript, Papathanassiu et al. describe the role of BCAT1 during inflammation in macrophages. They show that BCAT1 activity is directly linked to the inflammatory response of macrophages and that loss of its activity reduces the inflammatory response determined based on synthesis of itaconic acid (a pro-inflammatory metabolite) and the expression of IRG1. Moreover, the authors describe an inhibitor, ERG240, that seems to specifically inhibits BCAT1 and not BCAT2. Experiments with ERG240 in mouse models of auto-immune diseases highlight the power of this compound as potential drug. The manuscript is of high quality and clearly written. The addressed topic is of high impact, novel and will be of importance to the field. However, there are a few concerns that should be addressed before publication.
1. To get a better understanding on the activation of macrophages it would be essential to see expression levels of other cytokines and markers such as TNFa, IL6 and iNOS.
2. Fig 1b: Based on matrix effects and peptide type, ionization efficiency can vary. Without proper (isotope labeled) control peptides, the abundances of BCAT1 and BCAT2 peptides can not be compared. Maybe a Western blot would do better.
3. The applied concentration of the inhibitor is very high (20 mM). The authors should demonstrate that the inhibitor does not impair cell viability. In addition, they should provide evidence that the inhibitor is not metabolized by the cells and this way causing secondary effects. 4. Fig 2D: BCAT1 expression with non-targeting and targeting siRNA is missing for LPS treatment. 5. Fig 2E and 3C: itaconate levels should be replicated in more than one experiment and error bars be added.
6. It would be interesting to know why the half-life in animals is very low. Again, is the drug metabolized by the organism? (see also point 3). 7. Leucine is degraded to acetyl-Coa and acetoacetate both fueling cellular acetyl-Coa pools. Does decreased activity of BCAT1 impact cellular acetyl-Coa and citrate levels? Itaconate synthesis is a sink for citrate and diverts carbon away from the TCA cycle. It would be of importance to see whether acetyl-Coa levels are effected by BCAT1 inhibition and if this translates through to the observed citrate levels.
8. One option to test whether the function of BCAT1 is independent of its transaminase activity would be to deprive cells of BCAA and see if the effect compares to BCAT1 silencing or inhibition through ERG240.

Reviewer #1 (Remarks to the Author):
Papathanassiu et al. present a straightforward study using a novel BCAT1 inhibitor, complemented by siRNA for BCAT using human monocyte-derived macs as well as murine and rat models. Authors report that BCAT1 is primary transporter in macs. They  Authors' reply. We thank the reviewer for the thorough and accurate review of the findings. Since this is a crucial point, we performed an in vivo experiment where mice were injected with i) LPS, ii) LPS and ERG240 at the same time or iii) LPS for 3 hours, followed by ERG240, and peritoneal macrophages were collected 24 hours after LPS injection. Itaconate levels as well as Il6, Gbp2, Gbp4, expression levels (representative transcripts of IFNinduced GTPases and M1-like activation) were measured and presented below. Itaconate levels and Il6, Gbp2, Gbp4 mRNA expression in peritoneal macrophages isolated from mice injected either with LPS or LPS + ERG240 or a sequential treatment where LPS was first injected for 3 hours, followed by ERG240 (LPS + ERG240 (sequential)). Peritoneal macrophages were collected 24h following the LPS injection and subjected to GC/MS (itaconate) and qRT-PCR (Il6, Gbp2, Gbp4 expression). n=3 mice / group; All expression values are normalized to those obtained for Actinb mRNA levels. *, P < 0.05, ***, P < 0.001, when compared with LPS-treated samples. These results show that ERG240 can still reduce significantly itaconate production when injected after an initial inflammation is established in the mouse peritoneum. In addition, when ERG240 is administered following 3 hours of LPS challenging, there is an inhibitory effect on Il6, Gbp4 expression levels, suggesting that the drug can maintain its antiinflammatory properties, at least in part. This is also in line with the CIA experiment where ERG240 alleviates the severity of CIA when given orally after arthritis is established ( Fig 5C in the revised manuscript, 4C in the submitted manuscript). As suggested by the reviewer, ERG240 is therefore likely to affect the intrinsic changes in macrophage metabolism. As a result, we now include the findings describing the sequential treatment (Itaconate measurements by GC/MS in mice treated with LPS for 3 hours followed by ERG240) in the revised version of the main manuscript ( Figure 4C, lower panel).

Several concerns about the interpretation of the metabolic assays should be addressed. The inhibitor appears to inhibit both ECAR and OCR for Seahorse bioenergetic studies such as glycolytic rate, etc. (although no stats are presented, and should be). This means that both oxygen consumption (OCR) and glycolysis (ECAR) are reduced by the inhibitor
Authors' reply. Statistics are now added on all ECAR and OCR measurements' results as separate bar graphs and included in the revised manuscript's Figure 2A and B (see also the response to point 4 where statistics are also included for the revised siRNA experiments showing that indeed both ECAR and OCR are reduced).

First, authors mention that oxidative phosphorylation is inhibited but it should be
discussed that in macrophages, NADPH oxidase can also use oxygen in the "oxidative burst", along with the electron transport chain (oxphos). Therefore, saying just "oxidative phosphorylation" is potentially misleading. Authors should say "oxygen consumption" because that is all that the Seahorse measures unless further studies are conducted to specifically show that it is specifically inhibition of oxphos.
Authors' reply. The reviewer is correct that what was measured is oxygen consumption and the term 'oxidative phosphorylation' has now been replaced by 'oxygen consumption' throughout the revised manuscript (see highlighted text and figure legends).  Figure 2D.

Second
These results show that BCAT1-siRNA treatment significantly reduces OCR and ECAR, confirming the results obtained with pharmacological blockade (Figures 2A and B). These results are now updated in the abstract, results and discussion.

Third, authors state the importance of BCAT1 in CD4+ T cells regulating glycolysis, but fail to discuss their findings on glycolysis (ECAR) here. Typically, glycolytic cells are associated with pro-inflammation while inhibition of mitochondrial metabolism (Vats/Chawla
Cell Met and others) show that the "M2/alternative-like" phenotype of macrophage is dependent upon mitochondrial metabolism. The role of BCAT1 in glycolysis as a potential mechanism must be discussed.
Authors' reply. The role of BCAT1 in glycolysis is now discussed by adding the following paragraph (see also highlighted text in the discussion): Metabolism in M(LPS) macrophages is characterised by high levels of glycolysis and decreased respiration, a fragmented TCA and relatively high levels of some of its intermediates (succinate, citrate and itaconate). By contrast, metabolism in M2 (or M(IL-4)) corresponds to high levels of fatty acid oxidation and oxidative phosphorylation. Our findings indicating (i) the reduced ECAR levels as a result of BCAT1 inactivation in human macrophages, (ii) the anti-inflammatory effects of ERG240, reversing the M1-like transcriptome in mouse peritoneal macrophages, argue in favour of the involvement of BCAA in M1/pro-inflammatory macrophage function.

Fourth, if the drug is blocking ECAR and OCR, how long can the cells live?
Authors' reply. According to the reviewer's suggestion, a cell viability assay was performed on human MDMs (Please see also the response to Reviewer 2, point 3). ERG240 did not affect cell viability at 3h and 8h LPS stimulation neither in control or LPS-treated human macrophages. The results are presented in the revised manuscript, as well as in response to Reviewer 2's point below (Reviewer 2, point 3). Add additional explanation for "extracellular space" findings because this is unclear.

Si-Control
Authors' reply. 'Extracellular space' has now been replaced by 'extracellular matrix genes' in the revised manuscript ( Figure 3 in the submitted manuscript which is now Figure 4 in the revised one). We have also added a Volcano plot showing the differentially expressed genes ( Figure 4E in the revised manuscript). Authors' reply. This paper is now cited and discussed by adding this following paragraph to the discussion (highlighted text).
Besides its antimicrobial effects, itaconate has been proposed as an inhibitor of substrate level phosphorylation (SLP) through its effect on succinate-CoA ligase in RAW-264.7 cells 1 . Fig 1C. Fig 4C. Authors' reply. Statistical analyses are added for Fig 1C and 4C. Please note that Figure  4C is now Figure 5C in the revised manuscript. (See also highlighted text in figure legends).

Dosing in vitro should be discussed further-20mM in vitro is a high dose if IC50 is in nM range? Why 20mM?
Authors' reply. ERG240 is a structural analogue of leucine, designed based on the X-ray crystal structure of BCAT1 and, as the reviewer correctly states, it can inhibit BCAT1 within 0.1-1nM range of IC50. The concentration of 20 mM used in the cellular assays is because of the dependence of ERG240 on cell transporters for its intracellular inhibitory effect and its possible competition with leucine, which is in mM concentration (0.381 mM) in the RPMI media used throughout the study. Analysis of samples of human macrophage lysates using GC/MS revealed that 1x10 6 cells treated with 3.28 mg ERG240 (corresponding to 20 mM of the drug) will uptake about 0.8-3 ng of the compound with an estimated intracellular ERG240 concentration of ~70-260 nM (See also Author's reply to Point 3 made by Reviewer 2 below). Considering that only a fraction of the intracellular drug level will reach the target enzyme (with the rest associated with the transporting machinery and other cell sites), the estimated intracellular ERG240 levels is in line with the range of concentrations required to inhibit the enzymatic activity of the BCAT1.
ATP production should be marked as "estimated ATP production". The Y axis shows that it is a calculation from OCR experiment-not an actual measure of ATP. This is misleading.
Authors' reply. ATP production is now labelled as 'estimated ATP production' and Figures 2 and S2 were modified together with their respective legends as well as the results/methods section. Fig S3.

Add a citation for "as the public datasets for human immune cell expression indicate,"
Authors' reply. The sentence was deleted from the revised manuscript. -

To get a better understanding on the activation of macrophages it would be essential to see expression levels of other cytokines and markers such as TNFa, IL6 and iNOS.
Authors' reply. We thank the reviewer for his/her careful and constructive review. According to the reviewer's suggestion, we measured the expression of Il6, NOS2, TNF and PTGS2 as well-established early LPS-response transcripts. The results are presented below and in the revised manuscript (Fig S1-F, 3 hours LPS stimulation; n=6 donors used per group). Fig 1b:   backed up by quantitative proteomics by LC-MS/MS (with maximum of the 12 most abundant multiply-charged ions registered in each survey spectrum selected in a datadependent manner) confirms that increased BCAT1 mRNA levels reflect equally increased protein levels. We follow the reviewer's advice and withdraw LC-MS/MS results from the revised manuscript.

The applied concentration of the inhibitor is very high (20 mM). The authors should demonstrate that the inhibitor does not impair cell viability. In addition, they should provide evidence that the inhibitor is not metabolized by the cells and this way causing secondary effects.
Authors' reply. According to the reviewer's suggestion, we have performed cell viability for ERG240 and/or LPS in hMDMs for the stimulation time points used in the manuscript. We have also included a 24h stimulation time point and the results are now presented below and in the revised  The results show that for the stimulation time points used in our study (3h and 8h), there is no significant effect of ERG240 in both control (basal) and LPS-treated primary human macrophages. There was a modest but significant reduction in cell viability at 24h only in the LPS-stimulated cells.
The drug does not undergo any cellular metabolic reaction. ERG240 can easily be detected in lysates of human macrophages by GC/MS. In the graph below, overlapping chromatographs of a standard concentration of 10 ng ERG240 and cell lysates of untreated and ERG-treated human macrophages are shown. Fig 2D: BCAT1 expression with non-targeting and targeting siRNA is missing for LPS treatment.

4.
Authors' reply. According to the reviewer's suggestion, we have now included siRNA results for LPS treatment. Figure 2C and D in the submitted manuscript were also changed to take into account Reviewer 1' suggestions (see also the response to Reviewer 1, point 4) Fig 2E and 3C: itaconate levels should be replicated in more than one experiment and error bars be added.

Authors' reply.
According to the reviewer's suggestion, we have repeated the BCAT1 siRNA and measurements of itaconate by GC/MS in 4 donors. We have also repeated the in vivo LPS + ERG240 injection in mice to have an independent set of biological replicates (Please note that we have also included a group where there was a sequential treatment of LPS and ERG240 according to the reviewer 1's suggestion). The results are shown below and included in the main revised manuscript Figure 2E  Authors' reply. The pharmacokinetic study that determined the half-life of the drug in mice also showed that the oral bioavailability of the drug was 100%. The latter suggests that ERG240 is completely absorbed and not subject to first-pass metabolism. Since the drug is structurally a leucine analogue, it is likely to be uptaken by the ubiquitously expressed amino acid transporter LAT1. In that light, the low half-life of the drug is related to its complete and swift absorption by various sites rather to a phase I or II metabolic reaction. ERG240 is also capable of crossing the blood brain barrier (BBB) although it does not accumulate preferentially in the brain.
As previously mentioned, the drug does not undergo any metabolic reaction. For the metabolism of the drug in human macrophages, please see the response to point 3, which shows the GC/MS chromatograms with comparative analysis of human macrophages treated with ERG240 (20 mM) and 10 ng of ERG240. (sequential)). Peritoneal macrophages were collected 24h following the LPS injection. n=3 mice / group. *, P < 0.05.

 Humans  Mice
The results show that there is relative accumulation of citrate in LPS+ERG240-treated human macrophages when compared to LPS only. This is in line with the inhibitory role of ERG240 on IRG1, which converts cis-aconitate derived from citrate into itaconic acid (See also Fig S3-B for a representative cartoon of metabolites). Despite our efforts, we failed to detect Acetyl-coA in human macrophages by GC/MS, probably due to very instable properties of this metabolite. Nevertheless, the effect of BCAT1 inhibition on the broken TCA cycle in human macrophages lies predominantly downstream citrate at the IRG1 site with a relatively minor effect through inhibition of BCAA transamination and reduction in the amount of acetyl-coA entering the cycle (Fig S3B).

One option to test whether the function of BCAT1 is independent of its transaminase activity would be to deprive cells of BCAA and see if the effect compares to BCAT1 silencing or inhibition through ERG240.
Authors' reply. According to the reviewer's suggestion, we have performed an experiment where human macrophages isolated from healthy donors were stimulated with LPS in media deprived from BCAA, followed by measurement of IRG1 expression and itaconate levels.
The results are shown below and are included in the revised manuscript. (Fig S1-C and D).  As the reviewer predicted, the results are similar to pharmacological inhibition by ERG240 and the silencing experiments' findings. This also suggests that BCAT1 must be enzymatically active to have its effect on IRG1-mediated itaconate production in human macrophages. We thank the reviewer for suggesting this experiment, which added significantly to the understanding of our findings.