Acetyl-CoA production by Mediator-bound 2-ketoacid dehydrogenases boosts de novo histone acetylation and is regulated by nitric oxide

Summary

Histone-modifying enzymes depend on the availability of cofactors, with acetyl-CoA being required for histone acetyltransferase (HAT) activity. The discovery that mitochondrial acyl-CoA producing enzymes translocate to the nucleus suggests that high concentrations of locally synthesized metabolites may impact acylation of histones and other nuclear substrates, thereby controlling gene expression. Here we show that 2-ketoacid dehydrogenases were stably associated with the Mediator complex, thus providing a local supply of acetyl-CoA and increasing the generation of hyper-acetylated histone tails. Nitric oxide (NO), which is produced in large amounts in lipopolysaccharide-stimulated macrophages, inhibited the activity of Mediator-associated 2-ketoacid dehydrogenases. Elevation of NO levels and the disruption of Mediator complex integrity both affected de novo histone acetylation within a shared set of genomic regions. Our findings indicate that the local supply of acetyl-CoA generated by 2-ketoacid dehydrogenases bound to Mediator is required to maximize acetylation of histone tails at sites of elevated HAT activity.

Introduction

The deposition of histone modifications requires the availability of cofactors, such as acetyl-CoA and S-adenosyl-methionine (SAM) for acetylation and methylation reactions, respectively, thus explaining the multiple cross-talks between metabolic pathways and the activity of chromatin modifiers^1–3. In the simplest scenario, the global concentration of a cofactor in the cell impacts the activity of the enzyme(s) using it⁴. This relationship has been extensively analyzed for acetyl-CoA^5–7, given its central role in multiple metabolic pathways on the one hand, and as the sole provider of acetyl groups for lysine acetyl-transferases (KAT) on the other. The main sites of acetyl-CoA production in the cell are the mitochondria, where it is produced by the pyruvate dehydrogenase (PDH) using pyruvate molecules generated by glycolysis and reductive glutaminolysis, as well as by the ACSS1 (acyl-CoA synthetase short chain family member 1), which ligates acetate to CoA. However, mitochondrial membranes are impermeable to acetyl-CoA, and therefore the mitochondrial pool is not in equilibrium with the cytoplasmic and nuclear ones, which are instead generated mainly by the ATP-citrate lyase (ACLY) by cleaving the citrate exported from mitochondria into oxaloacetate and acetyl-CoA. Additionally, ACSS2 (acyl-CoA synthetase short chain family member 2) also generates acetyl-CoA starting from CoA and acetate molecules coming from outside of the cells or produced by deacetylation reactions catalyzed by lysine deacetylases. Importantly, ACLY⁸ and ACSS2⁹ are both cytoplasmic and nuclear enzymes, and at least in specific conditions their presence in the nucleus is required to support acetylation reactions that control gene expression and other nuclear transactions such as DNA repair^10–12. Therefore, while acetyl-CoA should in principle be in equilibrium between cytoplasm and nucleus because of its small size and ability to diffuse across the nuclear pores, its local production in the nucleus appears to provide an additional source of this cofactor that may be essential for specific regulatory events. Indeed, multiple lines of evidence suggest that the cytoplasmic and nuclear pools of acetyl-CoA may to some extent be functionally distinct and support the activity of different enzymes¹³.

Additional metabolic enzymes have been reported to be present in the nucleus and to generate acetyl-CoA and other acyl-CoA cofactors such as succinyl-CoA¹⁴. Notably, the mitochondrial pyruvate dehydrogenase complex, one of the largest macromolecular complexes in the cell, was also surprisingly shown to be delivered from mitochondria to nuclei and to be catalytically active in the nucleoplasm^14–17.

Since metabolites, such as acetyl-CoA and SAM, are highly diffusible molecules, the presence and functional relevance in the nucleus of the corresponding synthases imply that the transient generation of localized, elevated metabolite concentrations could enhance the activity of chromatin-modifying enzymes located in close proximity to their point of production.

Therefore, it is critical to determine if and how these enzymes are delivered to specific chromatin sites. However, only in a handful of cases a specific association between metabolic enzymes and transcription factors or chromatin-associated nuclear machineries has been reported^10,18.

Mediator is a 26 subunits megadalton-sized complex that together with general transcription factors of the preinitiation complex, controls transcription initiation by RNA Polymerase II (Pol II)^19–22, acting at any given time at tens of thousands of enhancers and promoters. As an intermediate between chromatin-bound sequence-specific transcription factors and Pol II, the Mediator complex integrates multiple regulatory inputs to enable enhancer-promoter communication²³, eventually controlling rates of transcription initiation as well as release of Pol II from promoter-proximal pausing. It has been noticed that compared to complexes of similar size, Mediator contains a much higher number of intrinsically disordered regions, which equip it with very high flexibility and thus with the potential ability to interact with a large number of different partners^22,24.

Indeed, a wide array of mammalian transcription factors (TFs) has been reported to interact with Mediator. By a variety of classical biochemical purification approaches, Mediator or some of its specific subunits have been shown to directly interact with distinct TFs and coactivators, ranging from the thyroid hormone receptor to PGC1a (peroxisome proliferator-activated receptor gamma coactivator 1-alpha) and many others^25–28, with observed differences being likely linked to cell types, biological contexts and purification procedures used.

Here, we report a direct association of Mediator with 2-ketoacid dehydrogenase enzymes that generate different acyl-CoA molecules, enhancing their deposition in chromatin, and specifically enabling the increased generation of hyper-acetylated histone tails. The acute increase in NO levels, such as the one occurring in lipopolysaccharide-stimulated macrophages, and the disruption of Mediator complex integrity, both resulted in a decrease in de novo histone acetylation across a common set of genomic regions. Our data suggest that the localized supply of acetyl-CoA produced by Mediator-associated 2-ketoacid dehydrogenases, contributes to maximize histone tail acetylation at genomic sites characterized by high HAT activity.

Results

The Mediator complex associates with acyl-CoA-producing enzymes

To characterize the composition of the Mediator complex and to identify its interactors in mouse macrophages, we set up an immunoprecipitation (IP)-mass spectrometry experiment with a rabbit polyclonal anti-MED1 antibody that was used at intermediate stringency binding and washing conditions (0,25M NaCl; 0.2% NP-40). All the Mediator complex components expressed in macrophages (32/32), including components of the kinase module, were highly enriched in the MED1 IP (Figure 1A and Table S1). Unexpectedly, the most enriched proteins in the MED1 IP included all the subunits of the mitochondrial 2-ketoacid (or 2-oxoacid) dehydrogenase complexes (Figure 1A and Table S1).

Figure 1. Interaction of the Mediator complex with 2-ketoacid dehydrogenases.

A) Left panel. Volcano plot showing proteins identified by MED1 IP/MS in unstimulated RAW264.7 mouse macrophages. Mediator complex subunits and 2-ketoacid dehydrogenase subunits are shown according to their statistical P-value (y-axis) and their relative abundance ratio (log2 fold change) in MED1 IP-IgG comparison. Biological replicates, n=4.

Right panel. Scheme depicting mitochondrial 2-ketoacid dehydrogenases and their relationship with the tricarboxylic acid (TCA) cycle.

B) Co-immunoprecipitation of DLAT with Mediator subunits. Antibodies specific for MED1, MED24 and MED16 were used to immunoprecipitate the Mediator complex in RAW264.7 and HeLa cells. Data are representative of n=4 independent experiments.

C) Co-immunoprecipitation of MED1 with MED24, OGDH and DLST in MED1-depleted HeLa cells. Data are representative of n=4 independent experiments.

D) Object-based colocalization analysis between MED1 and all three subunits of PDH in RAW264.7 and HeLa cells in DeepSIM super-resolution images.

On the left, scatter dot plot representing the percentage of each object (DLAT, DLD, PDHA1) colocalizing with MED1 on total number of objects (red bar indicates the mean). As control, the same analysis was performed for Pol2Ser5 and DAPI positive heterochromatic regions. The number of cells counted for RAW.264.7 analysis is the following: #159 (DLAT), #292 (DLD), #112 (PDHA1), #105 (Pol2Ser5), #191 (DAPI). For HeLa cells: #125 (DLAT), #107 (DLD), #125 (PDHA1), #71 (Pol2Ser5), #92 (DAPI).

On the right, representative image of cells (scale bar 5μm) and nuclei (scale bar 1μm) of RAW264.7 cells (upper panel) and HeLa cells (lower panel) stained with an anti-DLAT antibody (green) and an anti-MED1 antibody (red). Nuclear counterstaining with DAPI is also shown. The segmented DLAT (green) and MED1 (magenta) objects are shown on a representative nucleus on the third panel. White arrows indicate DLAT-MED1 colocalization.

E) Venn-diagram displaying the overlap between high confident OGDH-Flag, DLST-Flag and MED1 Chip-seq peaks in HeLa cells.

F) Meta profiles of MED1 (violet), DLST-Flag (red), OGDH-Flag (blue) and Empty-Flag (grey) ChIP-seq centered on the summit of the 5,132 DLST peaks (left) or the 4,609 OGDH peaks (right).

G) Representative snapshots reporting the signal for the samples as in panel F.

These multi-subunit enzymes catalyze the oxidative decarboxylation of 2-ketoacids, such as pyruvate and α-ketoglutarate, into their corresponding acyl-CoA esters (e.g., acetyl-CoA and succinyl-CoA), producing at the same time the reducing equivalent NADH²⁹. The family of 2-ketoacid dehydrogenases (Figure 1A, right panel) includes the pyruvate dehydrogenase (PDH), the α-ketoglutarate (2-oxoglutarate) dehydrogenase (OGDH) and the branched-chain α-ketoacid dehydrogenase (BCKDH). These complexes are composed of three distinct subunits that undergo coupled reactions facilitated by the lipoic acid cofactor (Figure 1A, right panel): the E1 subunit dehydrogenase (including E1-PDHA1, E1-OGDH, E1-BCKDHA), the E2 subunit dihydrolipoyl acyltransferase (E2-DLAT, E2-DLST, E2-DBT), and the E3 subunit dihydrolipoamide dehydrogenase (DLD), which is shared among these three complexes²⁹.

Because of the unusual character of such interactions between megadalton-sized complexes, we carried out several additional controls and orthogonal validations.

First, we checked the purity of our nuclei preparation in order to rule out the possibility of a post-lysis interaction between Mediator and dehydrogenase complexes released from contaminating mitochondria. Imaging of the nuclei purified from macrophages shows that they were completely devoid of mitochondria, as indicated by the virtually total absence of a mitochondria-selective fluorescent probe (MitoTracker) that after delivery to live cells is retained after fixation and permeabilization³⁰ (Figure S1A). The analysis of these highly pure macrophage nuclei by western blot confirmed the presence of DLAT and OGDH (Figure S1B). To accurately measure the relative abundance of nuclear vs. mitochondrial 2-ketoacid dehydrogenases, we set up a confocal immunofluorescence staining pipeline to determine their relative levels in the nuclear compartment and in the cytoplasm (Figure S1C). While, as expected, the majority of DLAT was detected in the cytoplasm (Figure S1C), a clear nuclear staining was visible that amounted to ca. 3% of the total, a finding in line with independent measurements obtained with the PDHA1-E1 and the DLD-E3 subunits (Figure S1D).

Second, we used antibodies recognizing two additional Mediator subunits (MED24 and MED16) in co-immunoprecipitation experiments in RAW264.7 macrophages. These experiments confirmed the presence of DLAT, the E2 subunit of PDH, in the immunoprecipitated complexes (Figure 1B, left panel). Similar results were obtained in HeLa cells, suggesting that the interaction between Mediator and 2-ketoacid dehydrogenases is not restricted to macrophages (Figure 1B, right panel). As additional validation of these data, we found that the partial depletion of MED1 decreased the amount of OGDH and DLST in the immunoprecipitated material (Figure 1C), together with a reduction of other Mediator subunits (Figure 1C).

Third, we measured the proximity of MED1 with each of the three subunits of the PDH complex (E1-PDH, E2-DLAT, E3-DLD) using a super-resolution microscopy technique, DeepSIM (Structured illumination microscopy), which achieves a resolution of ca. 100 nm. Analysis of the imaging data showed a high frequency, yet partial colocalization between MED1 and each of the three PDH subunits (Figure 1D) in the nuclei of both RAW264.7 macrophages and HeLa cells, indicating that Mediator and PDH are in close proximity in the nuclei.

Finally, we tried to determine whether nuclear PDH subunits were associated with chromatin and if their genomic distribution overlapped that of Mediator. As we could not identify ChIP-grade antibodies for any of the 2-ketoacid dehydrogenase subunits, we expressed Flag-tagged versions of several of them in Hela cells and eventually obtained good quality ChIP-seq data sets with two of the tested subunits, OGDH and DLST (Figure 1E-G). Whereas the number of peaks called in both cases was lower than that obtained in the anti-MED1 ChIP-seq, likely due to intrinsic technical difficulties of these ChIPs, we found a very high overlap between the genomic occupancies of MED1, OGDH and DLST (Figure 1E-G), indicating proximity between them at the level of active genomic regulatory elements.

Taken together, these data show that 2-ketoacid dehydrogenases physically interact with the Mediator complex in the nuclei.

LPS-stimulated nitric oxide production inhibits Mediator-bound PDH activity

We next aimed to determine whether 2-ketoacid dehydrogenases bound to the Mediator complex were active and functional enzymes. To this aim, we tested the enzymatic activity of Mediator-associated PDH by measuring the conversion of pyruvate to acetyl-CoA in the presence of MED1 immunoprecipitates. Briefly, paramagnetic beads coated with the anti-MED1 antibody were first used to immunoprecipitate Mediator from a macrophage nuclear extract and then incubated in vitro in the presence of pyruvate, NAD+, CoA and the required cofactor TPP (thiamine pyrophosphate) (Figure 2A). At the end of the incubation, the reaction products acetyl-CoA and NADH were quantified by liquid chromatography-tandem mass spectrometry (LC-MS/MS). We detected a linear relationship between the amount of the immunoprecipitated material and the levels of both reaction products (Figure 2B and Figure S2A), indicating that PDH complexes associated with Mediator are enzymatically active.

Figure 2. Effects of LPS-induced nitric oxide production on Mediator-bound PDH.

A) Schematic representation of the in vitro PDH enzymatic activity assay.

B) Acetyl-CoA and NADH production by Mediator-associated PDH. Acetyl-CoA (upper panel) and NADH (lower panel) levels measured by MS. In vitro PDH enzymatic activity assay was performed on MED1 immunoprecipitates from macrophages, as indicated in A.

C) Acetyl-CoA (left panel) and NADH (right panel) production by Mediator-associated PDH after 24h LPS treatment. The effects of the direct addition of a NO donor (Spermine NONOate, 5mM) in the reaction and of iNOS inhibition (1400W, 100μM) are shown. The NO donor was added directly to MED1 immunoprecipitated material before starting the in vitro enzymatic reaction described in A. n= 5 biological replicates. Bars represent P <0,05 by one-way ANOVA.

D) Acetyl-CoA global levels detected in total cell extract of RAW264.7 macrophage cells stimulated with LPS for 24h in the presence or absence of 1400W. n = 6 biological replicates.

E) Effects of LPS stimulation and NO on the association of Mediator with PDH. Lysates from unstimulated or LPS-stimulated RAW264.7 cells (100 ng/ml, 24 hours) or with NO donor Spermine NONOate (5mM for 1.5h), in the presence or absence of the iNOS inhibitor 1400W (100uμM) were immunoprecipitated with an anti-MED1 antibody and the abundance of DLAT was measured by western blot. Data are representative of n=4 independent experiments.

F) Effects of LPS treatment on PDH complex integrity. Coimmunoprecipitation of the PDH complex subunit DLAT with anti-DLD antibodies in nuclear lysates of RAW264.7 macrophages treated with LPS (100 ng/ml) for 24 hours.

G) Effects of LPS and NO treatment on PDH cellular localization. The DLAT fluorescence density was calculated in the nuclear compartment (as indicated in Methods) of RAW264.7 macrophages treated with LPS (100ng/ml) for 24h and with NO donor Spermine NONOate (500μM for 30’).

Recent findings indicated that PDH as well as other tricarboxylic acid (TCA) and electron transport chain (ETC) enzymes, are inhibited upon S-nitrosylation by nitric oxide (NO), which is abundantly produced by macrophages in response to LPS stimulation^31–33. NO production in LPS-stimulated macrophages arises from the conversion of arginine to citrulline catalyzed by the inducible nitric oxide synthase (iNOS, encoded by Nos2), which is transcriptionally activated by LPS starting ca. 4h after stimulation and peaking at 24h (Figure S2B)³⁴. Hence, we investigated whether PDH complexes bound to Mediator were sensitive to NO-mediated inhibition. To this end, macrophages were treated with LPS for 24h and the enzymatic activity of PDH in MED1 immuno-precipitates was tested as described before. Figure 2C shows that the production of both acetyl-CoA and NADH was inhibited when Mediator-associated PDH was isolated from LPS-treated macrophages (24h). The addition of a NO donor (Spermine NONOate) in the reaction had the same impact as LPS treatment (Figure 2C), indicating that nitric oxide suffices for LPS-mediated inhibition of PDH. NO-mediated inhibition of the enzymatic activity of nuclear DLAT was similar to that of the total (i.e., mitochondrial) DLAT, as expected (Figure S2C). Moreover, the pharmacological inhibition of iNOS in LPS-treated macrophages using the 1400W inhibitor, which efficiently blocks the conversion of arginine to citrulline (Figure S2B), restored PDH activity in the MED1 immunoprecipitated material to levels similar to those measured in untreated macrophages (Figure 2C). Taken together, these results suggest that the production of acetyl-CoA and NADH by Mediator-associated PDH is impaired by NO produced in LPS-treated macrophages. Notably, in spite of NO-mediated inhibition of PDH, LPS stimulation (24h) did not bring about global changes in acetyl-CoA levels (Figure 2D), likely due to the existence of additional acetyl-CoA production mechanisms, such as citrate cleavage in the cytoplasm and the nucleus by the ATP-citrate lyase (ACLY)^35–38. Conversely, another potential source of acetyl-CoA, the ACSS2 acetyl-CoA synthase, was not expressed in macrophages.

Next, we set out to determine whether the interaction between Mediator and 2-ketoacid dehydrogenases affected their enzymatic activity. To this aim, we used LC-MS/MS to measure acetyl-CoA production in purified nuclei from wild type cells or cells in which the integrity of the Mediator complex was disrupted by the depletion of the essential scaffold subunit MED14³⁹. To this aim, we inserted a dTAG-regulated FKBP12^F36V degron at the 3′ end of both copies of the MED14 gene in HCT116 cells⁴⁰ and we confirmed that dTAG treatment of these knock-in cells caused the extensive depletion of MED14 (Figure S2D). Next, we measured PDH enzymatic activity in isolated nuclei after adding pyruvate as a substrate, as reported before¹⁵. Measurement of acetyl-CoA production showed that nuclear acetyl-CoA generation was partially, yet significantly impaired in cells depleted of MED14, suggesting that interaction with Mediator supports the enzymatic activity of nuclear 2-ketoacid dehydrogenases (Figure S2D).

LPS treatment also attenuated the interaction between Mediator and the E2 subunit DLAT, while iNOS inhibition restored it (Figure 2E). However, LPS treatment did not affect PDH complex integrity, as the interaction between the PDH E2 and E3 subunits (DLAT and DLD, respectively) was not affected (Figure 2F). Moreover, the treatment with NO alone (Figure 2E, lower panel) did not decrease the interaction between Mediator and DLAT, indicating that it did not suffice to disrupt the physical association between them. Importantly, the effects of LPS on the interaction between Mediator and DLAT were not associated with any reduction in its nuclear abundance (Figure 2G).

Inhibition of the PDH enzymatic activity by NO may be caused by several mechanisms. First, the catalytic site of the lipoic acid synthase contains a Fe-S cluster that can be directly inhibited by NO, leading to reduced lipoic acid production³². Moreover, NO produced by iNOS can inactivate the PDH E3 subunit DLD by inducing S-nitrosylation of a critical cysteine residue³¹. Finally, reactive nitrogen species can covalently modify the lipoic acid thiol groups, generating adducts that impair enzymatic activity³³. To determine the possible contribution of these mechanisms, we investigated the lipoylation levels of the E2 subunits of both PDH (DLAT) and OGDH (DLST). We found that the level of catalytically active lipoic acid on DLAT did not change after a 24h LPS treatment, while lipoylated DLST levels progressively decreased starting from 6 hours after LPS treatment (Figure S2E). Conversely, the levels of total DLAT and DLST did not show any change (Figure S2E) during the LPS stimulation time course. Altogether, these data indicate that changes in the lipoylation state of the OGDH E2 subunit induced by LPS may be involved in OGDH inhibition by nitric oxide. However, this mechanism does not similarly affect PDH. PDH activity may instead be blocked by S-nitrosylation of cysteine residues of its E3 subunit, as previously reported³¹. Most notably, our data are consistent with the possibility that S-nitrosylation, in addition to blocking PDH enzymatic activity, may also influence its physical interaction with Mediator.

In order to place in a broader context the consequences of NO-mediated inhibition of Mediator-associated 2-ketoacid dehydrogenases, we also analyzed the overall impact of increased NO production in LPS-stimulated macrophages on their metabolic and gene expression profiles. NO-dependent metabolic remodeling mainly occurs in mitochondria^41–46, where LPS-induced NO and NO-derived reactive nitrogen species nitrosylate and inactivate the iron-sulfur proteins in the ETC, leading to decreased mitochondrial ATP production^43,44. In addition, NO inhibits several TCA cycle enzymes in addition to PDH, including the mitochondrial aconitase (ACO2), isocitrate dehydrogenase (IDH), OGDH and succinate dehydrogenase (SDH)^{31,32,44,47–49}, thus contributing together with other mechanisms to the accumulation of metabolites in the right half of the TCA cycle^50–52.

Consistent with these reports, a targeted metabolomic approach in bone marrow-derived macrophages (BMDMs) stimulated with LPS for 24h with or without the iNOS inhibitor 1400W, showed that LPS induced the accumulation of TCA cycle metabolites, such as citrate, itaconate and succinate (Figure S3A). Inhibition of NO production by 1400W had a selective impact on a subset of them, notably citrate, itaconate, GABA and succinate (Figure S3A) likely due to the relief of NO-mediated inhibition on PDH, ACO2, IDH, OGDH, and SDH. Conversely, other TCA metabolites, such as fumarate, malate and oxaloacetate, which have also been reported to accumulate after LPS stimulation^35,43,52, did not show any change upon iNOS inhibition.

In parallel, we evaluated the impact of iNOS inhibition by 1400W on LPS-dependent transcriptional reprogramming of macrophages. Nascent chromatin-associated RNA-seq was performed in BMDMs either untreated or stimulated with LPS for 8h or 16h in the presence or absence of 1400W. This window of time corresponds to maximal iNOS levels and NO production. In total, we identified ca. 1,800 genes affected by 1400W pre-treatment, that could be grouped by unsupervised clustering into 9 clusters according to their relative level of expression across conditions (Figure S3B). No detectable changes were induced by 1400W in the absence of LPS stimulation, indicating that it did not cause effects unrelated to NO production. Six out of the nine clusters (clusters c-f and h-i in Figure S3B) comprised genes whose expression failed to be induced in the presence of 1400W. These clusters were mainly composed of genes involved in the adaptive cellular response to nitroxides, such as glycolytic genes activated upon inactivation of components of the mitochondrial respiratory chain (Figure S3C, D). On the other hand, two out of the nine clusters (a and b) included genes that were hyper-induced in the presence of 1400W (Figure S3B) and that mainly encoded mediators of the inflammatory response (Figure S3C, D). These results are consistent with a physiological role of NO production during the LPS response in attenuating the expression of pro-inflammatory genes.

Overall, these data indicate that NO produced in macrophages in response to LPS treatment, in addition to causing extensive metabolic changes in the TCA cycle, inhibited both the PDH enzymatic activity and its physical interaction with Mediator. Moreover, it induced large-scale transcriptional effects that on the one hand included an adaptive response to the metabolic reprogramming driven by NO, and on the other dampened the induction of the inflammatory gene expression program. While the precise mechanisms are likely multifaceted, the latter effect may, in part, result from the inhibition of nuclear 2-ketoacid acetyltransferases, as we aimed to address below.

Nitric oxide impairs de novo histone acetylation

A model compatible with our data is that Mediator-associated PDH may act as a local supplier of acetyl-CoA to be used for histone acetylation, thus providing an extra source of acetyl groups in addition to acetyl-CoA molecules generated by ACLY.

We reasoned that if Mediator-associated PDH is required to support histone acetyltransferase (HAT) activity, then PDH blockade by NO generated in response to LPS stimulation would impair de novo deposition of acetyl group on histones, which can be readily measured in conditions in which their removal by histone deacetylases is blocked by the HDAC inhibitor Trichostatin A (TSA). First, by measuring H3K27 acetylation (H3K27ac) levels by western blot, we found that NO impaired de novo H3K27 acetylation at early time points after HDAC blockade by TSA (Figure 3A). Such reduced deposition of acetyl groups on histones did not correlate with a global reduction in acetyl-CoA content in macrophages treated with TSA or NO+TSA (Figure 3B), suggesting that in this window of time the overall availability of acetyl-CoA in treated cell was not significantly altered.

Figure 3. Effects of nitric oxide on de novo histone acetylation.

A) Mouse bone marrow-derived macrophages were treated with TSA (500nM) or TSA and the NO donor Spermine NONOate (500μM) for the indicated times. H3K27ac in nuclear extracts was measured by western blot.

B) Acetyl-CoA content of the samples indicated in A.

C) Quantitative analysis of histone acetylation by LC-MS in macrophages treated as described in A for 20 minutes. The heatmap represents the log2 of the relative abundance of each modification vs. internal control normalized on the sample mean. Grey: not detectable.

D) Representative acetylated histone peptides from the analysis described in C. Plots represent the log2 ratio of the relative abundance of each modification in sample vs. internal control. “I” indicates the presence of the acetyl group on one residue or the other. Each bar represents the P <0,05 by one-way ANOVA. N=4 biological replicates.

We next analyzed acetylation of all measurable histone tail peptides by quantitative mass spectrometry (MS)-based proteomics. TSA-induced acetylation was partially decreased by NO at most of the acetylated peptides analysed (Figure 3C and Table S3). Notably, the impact of NO on histone acetylation was more pronounced and statistically significant when considering high-acetylated states rather than low-acetylated states of histone peptides (Figure 3D, right panel). For instance, while tetra-acetylation of the histone H4 N-terminal tail was highly significantly reduced, mono-acetylation of the same peptide was unaffected (Figure 3D).

Overall, these results indicate that the local supply of acetyl-CoA generated by Mediator-bound PDH supports de novo histone acetylation, being particularly relevant to achieve maximal acetylation of the histone tails.

Nitric oxide impairs histone acetylation on de novo acetylated genomic regions

We postulated that if the production of acetyl-CoA by Mediator-associated 2-ketoacid dehydrogenases plays a role in facilitating the localized deposition of acetyl groups onto histones, then the disruption of the Mediator complex caused by MED14 depletion and the immediate, short-term release of nitric oxide (NO) by Spermine Nonoate should induce overlapping effects on genomic histone acetylation patterns. Hence, we set out to assess the impact of Mediator complex integrity and NO release on genome-wide histone acetylation.

In our experiments, we induced acute histone hyperacetylation using two mechanistically different stimuli: TSA and LPS. TSA treatment allows for the accumulation of acetyl groups at constitutively active regions by altering the dynamic equilibrium between their deposition and removal, primarily because of the inhibition of deacetylases. Consequently, in TSA-treated cells, de novo acetylation can be quantified without the interference of deacetylase activity. In contrast, LPS stimulation leads to the activation and recruitment of signal-regulated transcription factors to chromatin, resulting in localized acetylation within the context of a dynamic equilibrium influenced by deacetylases.

We first generated H3K27ac calibrated ChIP-seq (cChIP-seq) data sets in primary mouse macrophages treated for 20 min with TSA (or vehicle as control) in two experimental settings: 1) in control BMDMs (infected with lentiviruses expressing a non-targeting shRNA) in the presence or absence of an NO donor to block 2-ketoacid dehydrogenases; and 2) in cells depleted of MED14 via lentiviral transduction of a specific shRNA. We identified ca. 22,000 high-confidence H3K27ac peaks that were consistent across conditions and replicates (FDR≤10^-5, see STAR Methods). TSA treatment increased H3K27ac at more than 2,000 cis-regulatory elements (CREs) (Figure 4A and Table S4) and this effect was pervasively and strongly impaired by both NO pre-treatment and MED14 depletion (Figure 4A). Although differences in the relative strength of the effects of NO and MED14 depletion were detected, these observations were consistent regardless of the magnitude of TSA-induced hyperacetylation, as shown by ranking the TSA-induced H3K27ac peaks into quintiles of increasing acetylation gain (Figure 4B, C). The identified acetylation changes were strongly biased towards gene promoters, as indicated by the analysis of ChIP-seq peaks distances from transcription start sites (TSSs) (Figure 4B, right panel). MED1 ChIP-seq data (Figure 4C) showed decreasing occupancy from the first to the fifth quintile, which was consistent with the different basal acetylation levels (Figure 4C, bottom).

Figure 4. Effects of nitric oxide and loss of Mediator complex integrity on histone acetylation.

A) Left panel: effects of the pre-treatment of macrophages with the NO donor Spermine NONOate (500 μM) on TSA-induced H3K27ac changes. The scatter plot shows the comparison between TSA-induced H3K27ac changes in control and NO pre-treated macrophages. Each CRE is color-coded based on the log2FC (TSA+NO vs. TSA). Shown are all CREs (n=2,257) that showed significant induction upon TSA treatment in control cells (adjusted p-value ≤ 0.05 and Log2Fold Change > 0).

Right panel: effects of shRNA-mediated MED14 depletion on TSA-induced H3K27ac changes. Shown is the comparison between TSA-induced Log2Fold changes in MED14-depleted macrophages relative to their counterpart transduced with Non-targeting (NT) shRNA. Each CRE is color-coded based on the log2FC (TSA shMED14 vs. TSA shNT). All data shown represent the average of the results obtained with two different NT or MED14-specific shRNAs (the same design was used for the experiments shown in panels B-D below).

B) Box plots showing H3K27ac peaks induced by TSA (as in panel A, n=2,257) divided into quintiles ordered from left to right on the basis of their increase in inducibility after TSA treatment. The box plots report the normalized read counts per kb in the individual conditions, as indicated. Friedman test with Dunn’s multiple comparison test (p<0.0001: ****; p<0.0002: ***).

The right panel shows the mean distance from the closest TSS of LPS-induced H3K27ac peaks divided into quintiles based on their inducibility in response to TSA. Kruskal-Wallis test is reported.

C) Metaplots of the putative CREs displaying TSA-induced H3K27ac ChIP-seq signals, divided into quintiles as above (n=2,257). The bottom panels show metaplots of the MED1 ChIP-seq signal in untreated cells.

D) Representative snapshots showing H3K27ac ChIP-seq profiles across the reported conditions as in panel A-D, together with the MED1 ChIP-seq signal in the untreated condition (bottom green track). shNT: Non-Targeting shRNA.

E) Heatmap (left) and violin plot (right) showing the transcriptional changes measured by nascent chromatin-associated RNA-seq in cells treated with TSA with or without NO pre-treatment (n=1,721 TSA-induced genes in control conditions, adj pval ≤0.01 and Log2FC ≥1). Friedman test with Dunn’s multiple comparison test (p<0.0001: ****) is reported for the violin plot at the bottom. The heatmap reports the z-score per gene across the displayed conditions and the DESeq2 calls when comparing control and NO pre-treated conditions (down-regulated: “blue”; up-regulated: “red”; not significant: “white” at adj pval ≤0.01).

F) Violin plot of the distance between CREs displaying TSA-induced H3K27ac and TSS of either TSA-induced (red) or randomly sampled (grey) genes. Kolmogorov-Smirnov test is reported (p<0.0001 ****).

G) Left panel: effects of pre-treatment with the NO donor Spermine NONOate (500 μM) on LPS-induced H3K27ac changes. The scatter plot displays the comparison of LPS-induced changes between control and NO pre-treated conditions. Each CRE is color-coded based on the log2FC (LPS+NO vs. LPS). Shown are all CREs (n=3,328) that showed significant induction upon LPS treatment in the control condition (adjusted p-value ≤ 0.01 and Log2Fold Change ≥ 1). Note that the same design was used for the experiments shown below in panels H-J.

Right panel: effects of MED14 depletion on LPS-induced H3K27ac changes. The scatter plot shows the comparison between LPS-induced H3K27ac changes in control and MED14-depleted cells. Each CRE is color-coded based on the log2FC (LPS shMED14 vs. LPS shNT).

H) Box plots showing H3K27ac peaks induced by LPS (log2FC LPS vs. UT ≥ 1 in the control condition, n=3,328) divided into quintiles ordered from left to right based on H3K27ac inducibility in response to LPS stimulation. The box plots report the normalized and centered read counts per kb for the individual conditions, as indicated. Friedman test with Dunn’s multiple comparison test (p<0.0001: ****; p<0.03: *).

The right panel shows the mean distance from the closest TSS of the LPS-induced H3K27ac peaks divided into quintiles based on their LPS inducibility. Kruskal-Wallis test is reported (p<0.0001: ****).

I) Metaplots of the putative CREs displaying LPS-induced H3K27ac ChIP-seq signals divided into quintiles ranked from left to right based on their inducibility by LPS (Log2FC≥1 in control condition, n=3,328). The metaplot at the bottom shows the MED1 ChIP-seq signal in untreated and LPS-treated cells (bottom).

J) Representative snapshots showing H3K27ac ChIP-seq profiles across the reported conditions, with the MED1 ChIP-seq signals in untreated and LPS treated conditions shown at the bottom. shNT: Non-Targeting shRNA.

K) Heatmap (left) and violin plot (right) of the transcriptional changes measured by nascent chromatin-associated RNA-seq in macrophages stimulated by LPS with or without NO pre-treatment (n=3,605 LPS induced genes in control conditions, adj pval ≤0.01 and Log2FC ≥1). Friedman test with Dunn’s multiple comparison test (p<0.0001: ****) is reported for the violin plot at the bottom. The heatmap reports the z-score per gene across the displayed conditions and the DESeq2 calls when comparing control and NO pre-treated.

L) Violin plot of the distance between CREs displaying LPS-induced H3K27ac and TSS of either LPS-induced (red) or randomly sampled (grey) genes. Kolmogorov-Smirnov test is reported (p<0.0001 ****).

Representative examples including CREs assigned to each of the five quintiles in panel B are displayed in Figure 4D. Similar observations were retrieved in the same experimental setting by ChIP-seq with an antibody recognizing tetra-acetylated histone H4 (Figure S4A-D), which is the most TSA-induced and NO-inhibited peptide in the MS analysis shown above (Figure 3C).

Consistent with the widespread effects of NO pre-treatment on TSA-induced acetylation, transcription of nearly all TSA-induced genes, as detected by nascent chromatin-associated RNA-seq, was strongly affected by NO (Figure 4E). Furthermore, the ~2,000 CREs whose acetylation was induced upon TSA were significantly closer to the transcription start sites (TSS) of TSA-induced genes than to randomly sampled TSS (Figure 4F).

Next, we assessed the impact of NO pre-treatment and MED14 depletion on H3K27ac changes induced by a short (1h) LPS stimulation, keeping the same experimental design as the one outlined above. Of the ca. 22,000 high-confidence H3K27ac peaks detected, ~3,300 were induced by LPS stimulation (Figure 4G and Table S4). NO pre-treatment significantly reduced H3K27ac particularly at sites with low basal H3K27ac and high LPS-induced acetylation gain (4^th and 5^th quintiles in Figure 4H, I), that were biased towards TSS-distal regulatory elements (Figure 4H, right panel). Similar trends were observed in the tetra-H4ac ChIP-seq datasets we also generated (Figure S4E-G). MED14 depletion affected mostly those CREs with comparatively higher basal level of H3K27ac (1^st and 2^nd quintiles) (Figure 4G-I). Notably, the first two quintiles were those with the highest basal MED1 occupancy (Figure 4I, bottom). Note that no changes in Mediator recruitment were detectable when comparing cells treated with LPS alone or in combination with NO (Figure S4H-I). Representative snapshots of loci with LPS-induced and either NO-inhibited (Pkp, Jak2, Bltp3a) or unaffected (Cep55) H3K27ac levels are shown in Figure 4J. Coherently with the above observations, LPS-induced genes were partially affected by NO pre-treatment, albeit to a lower extent compared to the TSA-induced ones (Figure 4K). The ~3,000 CREs whose acetylation was induced upon LPS were significantly closer to the TSS of LPS-induced genes than to randomly sampled TSSs (Figure 4L).

Overall, these results show that acutely increased deposition of acetyl groups in chromatin are similarly affected by NO and Mediator complex disruption, hinting at a role of Mediator-associate 2-ketoacid dehydrogenases as local suppliers of acetyl-CoA.

Discussion

We report here the association of Mediator with nuclear 2-ketoacid dehydrogenase complexes. Our data hint at the possible role of Mediator-bound enzymes producing different acyl-CoA species, and particularly acetyl-CoA, in increasing the local availability of these cofactors, thus eventually enabling or maximizing the activity of lysine acetyltransferases and the de novo deposition of acetyl groups at a subset of chromatin sites. While previous data reported the interaction of nuclear KATs with acyl-CoA producing enzymes¹⁴, the association of 2-ketoacid dehydrogenases with Mediator in principle provides a simple strategy to locally enhance production of acyl-CoA species in the immediate proximity of several different KATs recruited at active chromatin sites. This model is consistent with the very high overlap of the genomic distribution of Mediator and that of 2-ketoacid-dehydrogenase subunits, as well as with the effects of the loss of Mediator complex integrity on de novo deposition of acetyl groups onto histones. We also propose that inhibition of 2-ketoacid dehydrogenases by NO produced in physiological conditions, such as macrophage activation by LPS, results in the local impairment of acetyl-CoA production and eventually in the reduced efficiency of deposition of acetyl groups onto histones. Among the other effects, this local inhibition of acetyl-CoA generation may contribute to dampen inflammatory gene expression at late time points after activation.

In this context, some outstanding issues remain to be clarified.

First, it is expected that individual nuclear lysine acetyltransferases may differentially rely on, and be impacted by, the high concentrations of acyl-CoA molecules generated in their proximity. The average cellular concentration of acetyl-CoA ranges between 2μM and 15μM, and most KATs have a dissociation constant for acetyl-CoA well within or even below this range (e.g., K_D=0.56μM in the case of human GCN5)^6,53, being also inhibited by high concentrations of their reaction products^46,54. Therefore, these KATs are likely capable of maximal activity at the acetyl-CoA concentrations present throughout the cell, and in the nucleus in particular. However, other KATs, such as p300 and CBP, have a relatively low affinity for acetyl-CoA (K_D=7.3-7.4μM)⁵⁵. Hence, their activity is likely affected by the global fluctuations in acetyl-CoA concentrations that can occur in common conditions such as transient or sustained changes in extracellular glucose⁶. Therefore, the local production of different acyl-CoA species may have a major regulatory impact on a subset of KATs and in fact regulate their activity. As several families of nuclear KATs are encoded in the mouse and human genomes, the role of Mediator-associated 2-ketoacid dehydrogenases on their activity is expected to be different depending on the specific enzymatic properties of each of them.

Second, as the association of Mediator with 2-ketoacid dehydrogenases was reduced by LPS stimulation in an NO-dependent manner, it appears possible that regulatory cues may control these interactions, enabling locus-specific or global regulation of acetylation of histones and non-histone substrates. Along the same line, while our data are compatible with the idea that the three different 2-ketoacid dehydrogenase complexes may have a similar ability to bind Mediator, we can’t rule out the existence of regulatory mechanisms that may differentially and specifically control their interactions with Mediator.

Finally, the molecular bases for the interaction between large macromolecular complexes such as Mediator and the 2-ketoacid dehydrogenases remain unclear at this stage, and their detailed molecular understanding will require complex efforts. The mammalian PDH complex is classically indicated as a giant 9.5 MDa assembly consisting of a central structural core of 60 E2 subunits that are lipoylated and surrounded by 30 E1 tetramers and 12 E3 dimers⁵⁶. However, at physiological salt concentrations much smaIler sub-megadalton complexes exist¹⁶, indicating high flexibility in complex organization. It is tempting to hypothesize that the protruding and highly flexible lipoyl moieties that move between E1 and E3 subunits during catalysis, may represent the shared component of different 2-ketoacid dehydrogenases that is involved in the interaction with Mediator. This possibility is consistent with the observation that the interaction between Mediator and PDH is sensitive to NO, since reactive nitrogen species can covalently modify the lipoic acid thiol groups, which on the one hand impairs enzymatic activity³³, and on the other may affect interaction with other complexes. As regards Mediator, it is remarkable that in comparison to complexes of comparable size and subunit composition, it possesses a notably greater number of intrinsically disordered regions (IDRs)^22,24. Functionality of IDRs is conferred by short linear motifs that are involved in different inter-molecular interactions, including protein-protein as well as protein-lipid interactions⁵⁷, and may thus enable binding and functional partnerships not only with other components of the transcriptional apparatus but also with nuclear metabolic enzymes.

Limitations Of The Study

Our data show that the Mediator complex and 2-ketoacid dehydrogenases interact in the nucleus and extensively overlap in their genomic distributions, leading us to propose that the interaction with Mediator tethers these metabolic enzymes to sites where they may support histone acetyltransferase activity. However, the molecular determinants of such an unconventional interaction between megadalton-sized complexes are still undefined, thus hampering both a comprehensive mechanistic understanding of their functional relationships and our ability to conduct precise genetic perturbation experiments that could conclusively discriminate between the mitochondrial and nuclear functions of 2-ketoacid dehydrogenases.

Star Methods

Resource Availability

Lead contact

Further information and requests for resources and reagents may be directed to and will be fulfilled by the Lead Contact, Gioacchino Natoli (gioacchino.natoli@ieo.it).

Materials availability

This study did not generate new unique reagents.

Experimental Model And Study Participant Details

Cell culture and reagents

RAW 264.7 cells (from ATCC) were cultured in DMEM with 10% North American serum,1% pen/strep (cat. no. P4333, Sigma), and 1% L-Glutamax (cat. no. 35050061, Gibco). HeLa cells (from ATCC) were cultured in DMEM with 10% South American serum, 1% pen/strep and 1% L-Glutamax. MED14-dTAG HCT-116 cells were cultured in McCoy’s 5A Medium Modified with 10% South American serum,1% pen/strep. Cell lines were authenticated by the Tissue Culture Facility of the European Institute of Oncology using the GenePrint10 system (Promega) and routinely screened for mycoplasma contamination. Bone marrow isolation from C57BL/6J mice (males, 5-7 weeks old) was performed in accordance with the Italian Laws (D.lgs. 26/2014), which enforces the Directive 2010/63/EU (Directive 2010/63/EU of the European Parliament and of the Council of September 22, 2010, on the protection of animals used for scientific purposes). All animal procedures were approved by the Italian Ministry of Health (346/2017-PR, notification 75DA4.N.SXD). Bone marrow derived macrophage (BMDM) cultures were carried out as described before⁵⁹. Cells were treated with lipopolysaccharide (LPS) from E. coli, serotype EH100(Ra) from Alexis (cat. no. ALX-581-010-L002) at 10 ng/mL (BMDMs) and 100 ng/mL (RAW 264.7 cells). 100 mM iNOS inhibitor 1400W (1400W, cat. no. CAY-81520, Cayman Chemical) was added 30 min before macrophage activation. Trichostatin A (TSA, cat. no. S1045, Selleck) was added for 10, 20 and 30 min. Spermine NONOate (NO donor, cat, no. CAY-82150, Cayman Chemical) was added directly to MED1-immunoprecipitated material at 5 mM; in BMDMs it was used at 0.5 mM for 15 min before TSA or LPS treatment and in Raw 264.7 it was used at 5 mM for 1h and 30 minutes. Acetyl-CoA quantity was determined using the Acetyl-CoA Assay Kit (cat. no. MAK039, Sigma) following the manufacturer’s instructions. dTAG-13 (Tocris Bioscience cat. 6605) was added at a final concentration of 500 nM for 24h.

Method Details

Plasmid cloning and engineered cell lines

MED14-dTAG HCT-116 cells were generated using CRISPR/Cas9-mediated homology-directed repair (HDR). The following sgRNA were designed using Benchling:

Med14 sg_For: 5′-CTATTGTTTTTAAACCAGGA-3′

Med14 sg_Rev: 5′-TCCTGGTTTAAAAACAATAG-3′

To generate MED14-dTAG HCT-116 cells a pUC57-based donor vector, containing the MED14 left homology arm (chrX:40651808-40652109) a GGGS spacer, the 2xHA dTAG (FKBP_F36V) insert, a P2A sequence, the sequence for puromycin resistance and the MED14 right homology arm (chrX:40651505-40651805) was assembled using Gibson assembly (New England Biolabs). The dTAG insert (FKBP_F36V) derives from pCRIS-PITChv2-Puro-dTAG (BRD4) (# 91796, Addgene). In this case HCT-116 cells were co-transfected with the MED14 guide RNA expressing px330 plasmid and the MED14-dTAG donor plasmid. After transfection and puromycin selection (1ug/mL for 3 days), single cells were seeded in 96-well plates by limiting dilution and expanded. Clones containing puromycin-P2A-2xHA-dTAG cassette were screened by PCR using the following primers:

Fw_5′_M14: 5′-ctaactgggacttgcttgaagtt-3′

Rev_3′_M14: 5′-GTTTGGCATTTTCATAACTTTATAGT-3′.

and then validated by western blot.

Lentiviral shRNA delivery in BMDMs

The shRNAs targeting Med14 were purchased from Merck (TRCN0000096279 and TRCN0000096283) and cloned into the PLKO.1 lentiviral vector. PLKO.1-based non targeting vectors (NT1: MISSION pLKO.1-puro luciferase shRNA control and NT2: MISSION pLKO.1-puro nonmammalian shRNA control) were also purchased from Merck. Lentiviral infection was performed as described⁶⁰. Briefly, 2.5 × 10⁶ 293T cells were plated the day before transfection. Transfer vector PLKO.1-shMED14 or PLKO.1-shNT, packaging plasmid psPAX2, and envelope plasmid pMD2.G were cotransfected by using calcium phosphate’s method. Transfection medium was replaced by fresh BMDM growth medium 8 h after transfection. The virus-containing media were collected at 36 and 48 h after transfection. Viruses were then purified and concentrated by Lenti-X concentrator (Takara 631231). BMDM cells were cultured in BMDM medium for 5 days and subjected to two rounds of infection following puromycin (4 μg/mL) selection for 72 h. MED14 depletion was tested by RT-qPCR.

Overexpression of mitochondrial enzymes

To generate the DLST and OGDH lentiviral expression vectors, human DLST and OGDH were cloned into a 3XFlag-NLS-pScalps-Puro lentiviral expression vector. The 3XFLAG-NLS sequence derives from pX458 (# 48138, Addgene) and was assembled into the pScalps-Puro vector using Gibson assembly. A 3xFLAG-NLS-pSCALP-puro empty vector was used as negative control. HEK-293T cells were transfected with either the 3XFLAG-NLS DLST-pScalps_Puro vector or the 3XFLAG-NLS OGDH-pScalps_Puro vector or the 3xFLAG-NLS empty vector using CaPO4-coprecipitation method. Lentiviral supernatants from transfected HEK-293T cells were collected and immediately used for infection. HeLa underwent two cycles of infection over two days. Filtered lentiviral vectors containing supernatant is added directly to the HeLa cells and at the end of the second day of infection cells are split and the puromycin is added (1 ug/ml). Assays were carried out after 3-4 days of puromycin selection.

Immunoprecipitation (IP)-mass spectrometry (MS)

RAW 264.7 cell pellets were lysed with cytosolic Buffer A (50 mM Tris-HCl pH 8.0, 2 mM EDTA, 0,1% NP40, 10% glycerol), incubated for 5 minutes at 4°C and centrifuged 5 minutes at 4°C, 3000 RPM. The supernatant was kept as the cytosolic fraction. Nuclear pellets were then resuspended in Buffer B (250 mM NaCl, 50 mM Tris-HCl pH 8.0, 0,5 mM EDTA, 0,5 mM EGTA, 0,2% NP40), incubated by rotation 30 minutes at 4°C, and centrifuged 10 minutes at 4°C in microfuge at the maximum speed. The supernatant was collected as “first nuclear extract”. The pellet was resuspended in Buffer C (420 mM NaCl, 20mM Hepes KOH pH 7.9, 20% glycerol, 2 mM MgCl₂, 0,2 mM EDTA, 0,1% NP40), incubated 1 hr by rotation at 4°C and centrifuged 10 min at 4C at maximum speed. The supernatant was collected as “second nuclear extract” and joined to the “first nuclear extract”. A mixture of protease inhibitors (Complete EDTA-free 3x20 tablets, cat.no. 5056489001, Roche) and 1 mM PMSF were added to all lysis buffers. 2.5 mg of this combined nuclear extract (with a final concentration of 250 mM NaCl and 0,2% NP40) was used for IP-MS. 10 μg of anti-MED1 antibody (cat.no. A300-793A, Bethyl) or Rabbit IgG (cat. no. 011-000-003, ChromPure) were pre-bound to 100 μl of G protein-coupled paramagnetic beads (Dynabeads Protein G cat. no.10009D, Invitrogen) in PBS/BSA 0.5%. Beads were then added to the nuclear lysate and incubated for 45 min. Immunoprecipitates were washed extensively using Buffer B and eluted in Laemmli buffer (cat. no. 1610747, Bio-Rad). Samples were then loaded on SDS-PAGE (NuPAGE® 4-12% Bis Tris Gel, cat. no. NP0335BOX, Invitrogen), stained with Coomassie Brilliant blue (InstantBlue Coomassie Protein Stain, cat.no. ab119211, Abcam) and nine consecutive bands were excised and trypsin-digested following the procedure described by Shevchenko A et al.⁶¹ Briefly, proteins were reduced by treatment for 1h at 56°C in 10 mM DTT and then subjected to alkylation with 55 mM iodoacetamide at room temperature in the dark for 45 min. Trypsin digestion (12.5 ng/μL) was carried out overnight at 37°C. Tryptic peptides were extracted from the gel fragments using 3% trifluoroacetic acid (TFA) and 30% acetonitrile (ACN). Subsequently, peptides extracted from the gel fragments were cleaned up using homemade STAGE Tips microcolumns ⁶². Peptides were then eluted in 40 μl elution buffer B (80% ACN, 0.1% formic acid (FA). ACN was evaporated using a speedvac concentrator (Eppendorf) and the volume of the eluates was adjusted to 5 μl with 1% TFA, to be then analysed by LC-MS/MS using an EASY-nLC 1200 (cat. no LC140, Thermo Fisher Scientific,) connected to a Q-Exactive HF (Thermo Fisher Scientific) through a nano-electrospray ion source (EASY-SPRAY, Thermo Fisher Scientific). The nano-LC system was operated in one column set-up with an EasySpray PEPMAP RSLC C18 column (Thermo Fisher Scientific) kept at 45°C constant. Solvent A was 0.1% formic acid (FA) and solvent B was 0.1% FA in 80% ACN. Samples were injected in aqueous 1% TFA at a constant pressure of 980 Bar. Peptides were separated with a gradient of 3–30% solvent B over 55 min followed by a gradient of 30–40% for 10 min and 40–95% over 5 min at a flow rate of 250 nL/min. The MS instrument was operated in the data-dependent acquisition (DDA) mode. The 15 most intense peptide ions with charge states ≥2 were sequentially isolated to a target value of 3e6 and fragmented in the high collision dissociation (HCD) cell using a normalized collision energy setting of 28%. MS spectra were detected in the Orbitrap using a resolution R=60,000 at m/z 200 within an m/z range corresponding to 375-1650. The maximum allowed ion accumulation times were 20ms for full scans and 80ms for MSMS. The dynamic exclusion time was set to 15s.

siRNA-mediated depletion of MED1 in HeLa cells

For siRNA-mediated knockdown of MED1 in HeLa cells, siRNAs from Santa Cruz were used: siMED1 cat. no. sc-78161, and control siRNA cat. no. sc-37007. For transfection of the siRNAs, the Lipofectamine RNAiMAX reagent (cat. no. 13778150, Thermo Fisher) was used according to the manufacturer’s protocol.

Chromatin RNA-seq

Chromatin-associated RNA was extracted as described⁶³. Briefly, 5-15x10⁶ BMDM cells were lysed with ice-cold 10 mM Tris-HCl (pH 7.5), 0.05% NP-40, and 150 mM NaCl buffer for 5 min. The lysate was then layered on 2.5 vol of a chilled sucrose cushion and centrifuged at 14,000 rpm for 10 min at 4°C. The nucleus pellet was gently washed with ice-cold PBS/1 mM EDTA and then resuspended in a 20 mM Tris-HCl (pH 7.9), 75 mM NaCl, 0.5 mM EDTA, 0.85 mM DTT, 0.125 mM PMSF, and 50% glycerol buffer by gentle flicking of the tube. An equal volume of a cold 10 mM HEPES (pH 7.6), 1 mM DTT, 7.5 mM MgCl₂, 0.2 mM EDTA, 0.3 M NaCl, 1 M urea, and 1% NP-40 buffer was added. The tube was gently vortexed and then centrifuged at 14,000 rpm for 2 min at 4°C. The chromatin pellet was then dissolved in TRIzol (Invitrogen). Chromatin-associated RNA was purified with TRIzol with an additional phenol/chloroform extraction step prior to precipitation. We used 60-100 ng of the nascent RNA for cDNA-library synthesis using the TruSeq Stranded Total RNA Sample Preparation kit (Illumina RS-122-9007) with the ribosomal depletion step. Libraries were quantified with the Quantifluor reagent (Promega cat. E2670) and the quality of the size was controlled with the Tapestation instrument (Agilent) using the high-sensitivity assay HD5000 (Agilent cat. 5067-5592). cDNA libraries were sequenced on an Illumina NovaSeq platform with 51-bp paired-end reads.

Cell lysates and Western blots

For whole cell extracts, cells were harvested, washed once with cold PBS 1X, and centrifuged at 1500 rpm for 5 min. Cell pellets were resuspended with lysis buffer (250 mM NaCl, 50 mM Tris-HCl pH 8.0, 0,5 mM EDTA, 0,5 mM EGTA, 0,2% NP40) and incubated for 30 minutes at 4°C. Then lysates were centrifuged in microfuge tubes at 13,000 rpm for 10 minutes at 4°C. Nuclear extracts were obtained with a 2-step nuclear extraction protocol. Briefly, cells were lysed in cold cytoplasmic lysis buffer (10% sucrose, 0,5 mM EGTA, 15 mM NaCl, 60 mM KCl, 15 mM HEPES and 0,5 % Triton) and incubated on ice for 10 min. The lysate was layered on a cold sucrose cushion and centrifuged in microfuge tubes at 13,000 rpm for 10 minutes at 4°C. The supernatant from this spin represented the cytoplasmic fraction. The nuclear pellet was gently washed using cytoplasmic lysis buffer and then centrifuged at 3500 rpm for 5 min. Nuclear pellets were then resuspended in Buffer B and Buffer C as mentioned above. Cocktail of protease inhibitors and 1 mM PMSF were added to all lysis buffers used. Protein extracts were resolved on SDS–polyacrylamide gel, blotted onto nitrocellulose membranes, and probed with the following antibodies: LaminB1 (cat.no Ab16048, Abcam); Lipoic acid (cat.no. 437695, Sigma); H3K27Ac (cat.no. Ab4729, Abcam); Vinculin (cat.no. sc-73614, Santa Cruz Biotechnology); DLAT (cat. no. 13426-1-AP, Proteintech); OGDH (cat. no. 15212-1-AP, Proteintech); ACO2 (cat. no. 6922S, Cell Signaling); HA (cat.no. 26183, Invitrogen).

Co-immunoprecipitation

Nuclear cell lysates were obtained from: (1) wild-type RAW264.7 cells; (2) MED1-depleted HeLa cells as described for the IP-MS experiment. 2.5 mg of these nuclear extracts were used for the co-immunoprecipitation experiments. 10 μg of anti-MED1 (cat.no. A300-793A, Bethyl), anti-MED16 (cat.no. A303-668A, Bethyl), anti-MED24 (cat.no. A301-472A, Bethyl) and anti-DLD (cat.no. A304-733A, Bethyl) antibodies or Rabbit IgG (cat.no 011-000-003, Jackson ImmunoResearch) were pre-bound to 100 μl of G protein-coupled paramagnetic beads in PBS/BSA 0.5%. Beads were then added to the nuclear lysate and incubated for 45 min (MED1 IP) or overnight (MED24, MED16 and DLD IP). Immunoprecipitates were washed extensively using Buffer B, eluted in Laemmli buffer (cat. no. 1610747, Bio-Rad), resolved by SDS–PAGE and immunoblotted with the following antibodies: MED1 (cat. no. A300-793A, Bethyl); MED24 (cat. no. A301-472A, Bethyl); MED16 (cat. no. A303-668, Bethyl); DLAT (cat. no. 13426-1-AP, Proteintech); OGDH (cat. no. 15212-1-AP, Proteintech); DLST (cat. no. A304-309A, Bethyl); DLD (cat. no. A304-733A, Bethyl).

RT-qPCR

Total RNA (from UT, 1400W-treated cells, 24hLPS-treated cells and 1400W+24h LPS-treated cells, 3 biological replicates) was extracted from BMDMs using the Zymo Quick-RNA kit (cat. no. R1055, Zymo Research). 500 ng of total RNA was reverse-transcribed with ImProm-II reverse transcription system (cat. no. A3800, Promega) following the manufacturer’s instructions. RT-qPCR was assembled with Fast SYBR Green master mix (cat. no. 4385614, Applied Biosystems) and run on a QuantStudio 6 realtime PCR machine (Applied Biosystems). Analysis was done on the Thermo Fisher Cloud platform, and Nos2 primers (Forward: CCATCATGAACCCCAAGAGT and Reverse: CATCCAGAGTGAGCTGGTAGG) were designed using Primer3.

Confocal microscopy and image analysis

To check the purity of our nuclear preparation, we stained live cells with MitoTracker Red CM-X2Ros (200 nM) for 30 min at 37°C (cat. no. M7513, Thermofisher) and Hoechst 33342 (cat. #4082, Cell Signaling Technology), that were used as mitochondrial and nuclear markers respectively. Nuclei were then purified as described above and images of live whole cells and purified nuclei were acquired by a Leica SP8 FSU AOBS confocal microscope (Leica Microsystems). Hoechst and MitoTracker Red CM-X2Ros were excited with the 405 nm and the 561 nm laser lines and the emitted signals acquired between 415 and 480 nm, and between 570 and 700 nm, respectively, in both cases with a hybrid detector and the HC PL APO CS2 63x/1.40 oil immersion objective lens (Leica Microsystems). The transmitted signal was also acquired in differential interference contrast (DIC) mode. To measure the presence of the three PDH subunits inside the nuclei of macrophage cells, immunofluorescence staining and confocal acquisitions were performed on Raw264.7 cells (UT, 24h LPS, 30 min NO) grown onto glass coverslips. Briefly, PFA fixed cells where permeabilized with 0.1% Triton X-100, blocked and incubated overnight with anti-DLAT antibody (1 mg/ml, cat. no. 12362S, Cell signaling) or anti-DLD antibody (4 mg/mL cat. no. sc-365977, Santa Cruz Biotechnology) or anti PDHA1 antibody (4 mg/mL cat.no sc-377092, Santa Cruz Biotechnology). Alexa Fluor Plus 488 conjugated anti-mouse secondary antibody (cat. #A32766, Thermo Fisher Scientific) and Phalloidin (cat.no. AB176759, Abcam) were used to visualize the three PDH subunits and F-actin, respectively. Nuclei were counterstained with DAPI and samples were mounted with Glycerol mounting medium (70% Glicerol in PBS with 1.5% DABCO, cat. no. D27802, Sigma). Z-stacks of 12 microns in depth were acquired with a Nikon CSU-W1 spinning disk field scanning confocal system (Nikon Europe B.V.) equipped with a L6Cc laser sources (Oxxius), a multiband dichroic mirror, single-band emission filters and a Prime BSI sCMOS camera (Teledyne Photometrics). 3-D multi-channel stacks were acquired through a 100x/1.5 oil immersion objective lens with a voxel size of 70x70x200 nmPrior to perform image analysis, the multichannel Z-stacks were deconvolved using the Lucy-Richardson algorithm with NIS software, and a background subtraction was carried out in FiJi⁶⁴. To calculate the intensity of the PDH subunits signals in the nucleus and in the cytoplasm, a custom Python-based image segmentation pipeline exploiting the deep learning CellPose algorithm ⁶⁵ was used. The nuclear over cytoplasmic (nu/cyt) ratio of DLAT (Figure S1D) was calculated by the total intensity of the pixels with grey levels above 80 in the two subcellular compartments. The threshold was established to eliminate any remaining background noise or signals arising from non-specific antibody labeling from the fluorescence quantification. For the calculation of the DLAT fluorescence density in the nucleus, as shown in Figure 2G, the total intensity above threshold (80 grey levels) was normalized by the nuclear area (mm²).

DeepSIM imaging and analysis

Raw 264.7 and HeLa cells were grown onto glass coverslips before PFA fixing step. Cells were then permeabilized with 0.1% Triton X-100, blocked and stained with DAPI, Phalloidin (cat.no. AB176759, Abcam), MED1 antibody (2 ug/mL cat.no. HPA052818, Sigma) and DLAT antibody (1 ug/ml, cat. no. 12362S, Cell signaling) or DLD antibody (4 ug/mL cat. no. sc-365977, Santa Cruz Biotechnology) or PDHA1 antibody (4 ug/mL cat.no sc-377092, Santa Cruz Biotechnology) or Pol2-Ser5p antibody (4 ug/mL cat. no. C15200007, Diagenode). Anti-rabbit CF-568 (SAB4600400, Sigma) and anti-mouse Alexa Fluor Plus 488 (cat. no. A32766, Thermo Fisher Scientific) conjugated secondary antibodies were used to visualize the MED1 and the PDH subunits or the Pol2-Ser5p signals, respectively. Samples were mounted with Glycerol mounting medium (70% Glicerol in PBS with 1.5% DABCO, cat. no. D27802, Sigma). Cells were imaged by the DeepSIM super-resolution module (CrestOptics S.p.A.) mounted on an Eclipse Ti2 fluorescence microscope (Nikon Europe B.V.) equipped with solid-state lasers (Lumencor Celesta light engine), a sCMOS camera (Kinetix, Teledyne Photometrics) and a 100x/1.49 NA oil immersion objective lens (Nikon Europe B.V.). The standard structured illumination mask (CrestOptics S.p.A.) was used and 37 images per channel of a single Z plane at the middle height of the nucleus were acquired. The super-resolution reconstructed images were analyzed with Arivis4D scientific image analysis software (Carl Zeiss Microscopy Software Center Rostock GmbH). Briefly, the DAPI (to segment the nuclei), MED1, PDH subunits, Pol2-Ser5p and heterochromatic DAPI signals were segmented and the number of DLAT, DLD, PDHA1 colocalizing objects over the total number of the corresponding objects was calculated. In the case of MED1-Pol2-Ser5p and the MED1-heterochromatin samples, the ratio of MED1 colocalizing objects over the total number of MED1 objects was calculated. A colocalizing object was considered as such when at least 5% of its area intersected with a partner object. Only the objects localized inside the nuclei were considered for the analysis.

Metabolite-MS and Data analysis

BMDM cells (UT, LPS 24h, 1400W, 1400W+ LPS 24h) were grown in 10 cm plates and harvested in ice-cold PBS. Pellets were then resuspended in 250 μl of ice-cold methanol/acetonitrile 1:1 containing 1ng/μl [U-¹³C₆] glucose (internal standard, cat. no. 389374 Sigma-Aldrich) and 1ng/μl [U-¹³C₅] glutamine (internal standard, cat. no. 605166 Sigma-Aldrich), and spun at 20,000×g for 5min at 4°C. Supernatants were then passed through a regenerated cellulose filter, dried, and resuspended in 100μl methanol for subsequent analysis. Amino acid quantification was performed through previous derivatization. Briefly, 50μl of 5% phenyl isothiocyanate in 31.5% ethanol and 31.5% pyridine in water were added to 10μl of each sample. Mixtures were then incubated with phenyl isothiocyanate solution for 20 min at RT, dried under nitrogen flow, and suspended in 100μl of 5 mM ammonium acetate in methanol/H2O 1:1. The identities of all metabolites were confirmed by using pure standards. Quantification of different metabolites was performed with a liquid chromatography/tandem mass spectrometry method using a C18 column (Biocrates) for amino acids and a cyano-phase LUNA column (50 mm × 4.6 mm, 5μm; Phenomenex) for the TCA cycle metabolites, respectively. Methanolic samples were analyzed by 10 min run in positive (amino acids) and 5 min run in negative (metabolites) ion mode. The mobile phases for the amino acid analysis were phase A: 0.2% formic acid in water; and phase B: 0.2% formic acid in acetonitrile. The gradient was T0 100% A, T5.5 min 5% A, T7 min 100% A with a flow rate of 500μl/min. The mobile phase for negative ion mode analysis (TCA cycle metabolites) was phase A: water; and B: 2 mM ammonium acetate in methanol. The gradient was 90% B for all analyses with a flow rate of 500μl/min. Metabolomic data were acquired on an API-4000 triple quadrupole mass spectrometer (AB Sciex) coupled with an HPLC system (Agilent Technologies), a CTC PAL HTS autosampler (PAL System), and a Sciex Triple Quad 3500 (AB Sciex) with an HPLC system (AB Sciex) and a built-in autosampler (AB Sciex). MultiQuant software (version 3.0.2) was used for data analysis and peak review of chromatograms. Metabolomic data were normalized by defining $x_n^N$ (relative metabolite area) as:

$$x_n^N=\frac{Xn}{{{\displaystyle \sum }}_{n=a}^{z}n}$$

where x_n represents the peak areas of metabolite n for samples a, b, …, z and ${{\displaystyle \sum }}_{n=a}^{z}n$ represents the sum of peak areas of metabolite n for samples a, b, …, z.

Relative metabolite area $\left(x_n^N\right)$ was then divided by the sum of relative metabolite areas analyzed in each sample to obtain relative metabolite abundance (m), as:

$$m_a^N=\frac{x_n^N}{{{\displaystyle \sum }}_{a=1}^{n}a}$$

where ${{\displaystyle \sum }}_{a=1}^{n}a$ represents the sum of relative metabolite areas 1, 2, …, n for sample a. Internal standards were used to control instrument sensitivity ⁶⁶. Normalized data are reported in Table S2.

In-vitro PDH enzymatic activity assay by MS

For the MS-analysis of acetyl-CoA and NADH following MED1 or DLAT-immunoprecipitation, nuclear and total extracts and MED1 or DLAT IPs from RAW 264.7 cells (UT, LPS 24h, 1400W+LPS, NO donor samples for MED1-IP and UT samples for DLAT-IP) were performed as described above. For the DLAT-IP, 4 ug of DLAT antibody were pre-bound to 100 μl of G protein-coupled paramagnetic beads in PBS/BSA 0.5%. Beads were then added to the nuclear and total lysates and incubated for 3h. For both DLAT and MED1 immunoprecipitates were then washed extensively and incubated with 5 mM Sodium Pyruvate (cat. no. P5280, Sigma) in the presence of PDH buffer (50 mM MOPS HCl, pH 7.4, 0.2 mM MgCl₂, 0.01 mM CaCl₂, 0.3 mM cocarboxylase, 0.12 mM CoA, 2 mM β-nicotinamide adenine dinucleotide, 2.6 mM l-cysteine) at 30°C for 30 min (buffer and protocol adapted from Wang et al.) ¹⁴. 10μl of this eluate was mixed with 250μl of MeOH/ACN. Then, samples were centrifuged at 4°C for 5 min at 12000 rpm and filtered using a regenerated nitrocellulose filter (0.22μm). The samples were dried under nitrogen flow and resuspended in 100μl of MeOH. 10μl of each sample was injected into a mass spectrometer using the same chromatographic method described earlier for TCA cycle metabolites, adding the MRMs to identify acetyl-CoA and NADH. MultiQuant software (version 3.0.2) was used for data analysis and peak review of chromatograms.

Isolated nuclei for Acetyl-CoA measurement

MED14-dTAG HCT-116 cells were maintained in glucose free medium 24h prior to nuclear isolation. Immediately following nuclear isolation, nuclei were incubated with 5 mM Sodium Pyruvate (cat. no. P5280, Sigma) in the presence of PDH buffer (50 mM MOPS HCl, pH 7.4, 0.2 mM MgCl₂, 0.01 mM CaCl₂, 0.3 mM cocarboxylase, 0.12 mM CoA, 2 mM β-nicotinamide adenine dinucleotide, 2.6 mM l-cysteine) at 37°C for 8h. Supernatant was collected and Acetyl-CoA and NADH were measured by MS as described above.

Histone post-translational modifications (PTMs) MS analysis

Histones were enriched from primary BMDM cells by resuspending 5× 10⁶ cells in PBS buffer containing 0.1% Triton X-100, protease inhibitors, Na butyrate and PMSF 1mM. Nuclei were isolated through a 10 min centrifugation at 2300×g, resuspended in RIPA buffer (150 mM NaCl, NP40 1%, Na-Deoxycholate 0,1%, 50 mM Tris HCl pH 8.0, 0,1% SDS) and incubated for 1h at 4°C in the presence of 250 U of benzonase (cat. no. E1014, Sigma) to digest nucleic acids. The yield of histones extracted was estimated by SDS-PAGE gel by comparison with known amounts of recombinant histone H3.1, following protein detection with Coomassie Brilliant blue. Approximately 4 μg of histone octamer were mixed with an equal amount of heavy stable isotope labelled histones, which were used as an internal standard ⁶⁷ and separated on a 17% SDS-PAGE gel. Histone bands were excised, chemically acylated with propionic anhydride and in-gel digested with trypsin, followed by peptide N-terminal derivatization with phenyl isocyanate (PIC)⁶⁸. Peptide mixtures were separated by reversed-phase chromatography on an EASY-Spray nano-column (Thermo Fisher Scientific), 25-cm long (inner diameter 75 μm, PepMap C18, 2 μm particles), which was connected online to a Q Exactive Plus instrument (Thermo Fisher Scientific) through an EASY-Spray™ Ion Source (Thermo Fisher Scientific), as previously described ⁶⁸. Briefly, Solvent A was 0.1% formic acid (FA) in ddH₂O and solvent B was 80% ACN plus 0.1% FA. Peptides were injected in an aqueous 1% TFA solution at a flow rate of 500 nl/min and were separated with a 55-min linear gradient of 10–45% solvent B. The Q Exactive instrument was operated in the data-dependent acquisition (DDA) mode. Survey full scan MS spectra (m/z 300–1350) were analyzed in the Orbitrap detector with a resolution of 70,000 at m/z 200. The 12 most intense peptide ions with charge states comprised between 2 and 4 were sequentially isolated to a target value for MS1 of 3×10⁶ and fragmented by HCD with a normalized collision energy setting of 28%. The maximum allowed ion accumulation times were 20 ms for full scans and 80 ms for MS/MS, and the target value for MS/MS was set to 1×10⁵. The dynamic exclusion time was set to 20 sec, and the standard mass spectrometric conditions for all experiments were as follows: spray voltage of 1.8 kV, no sheath and auxiliary gas flow.

ChIP-seq

H3K27ac and tetra-acetylated histone H4 (cat. no. 06866, Millipore) ChIP-seq were carried out as previously described⁶⁰. Briefly, 5-10x10⁶ BMDM cells were fixed with 1% formaldehyde (cat. no. 252549, Merck) for 10 min at room temperature and lysed to prepare nuclear extracts. For MED1 ChIP, 200x10⁶ RAW264.7 cells or 70 x10⁶ HeLa cells, for FLAG-ChIP-seq (mitochondrial enzymes) 70-80 x10⁶ HeLa transduced cells were fixed with a double cross-linking protocol as previously described⁶⁹. Briefly, cells were scraped in PBS, incubated for 45 min with 2 mM DSG in PBS, washed twice, and then incubated with 1% formaldehyde in PBS for 10 min followed by quenching with 125 mM glycine. For H3K27ac, tetra-acetylated histone H4 and MED1 ChIP-seq, after chromatin shearing by sonication with a Branson Ultrasonics Sonifier, lysates were incubated overnight at 4°C with protein G Dynabeads (cat. no.10009D, Invitrogen) previously coupled with 3 μg of H3K27Ac antibody (cat. no. ab4729, Abcam), or 5ul of tetra-acetylated histone H4 antibody or 10ug of MED1 antibody (cat.no. A300-793A, Bethyl). For FLAG-ChIP, after chromatin shearing by sonication with a Branson Ultrasonics Sonifier, lysates were incubated overnight at 4°C with Dynabeads M-280 Sheep anti-mouse IgG (cat. no.11202D, Thermo Fisher) coupled with 8ug of FLAG antibody (cat. no F1804, Sigma). For H3K27ac and tetra-acetylated histone H4 ChIP-seq, 1 μg of chromatin was supplemented with 5% spike-in of S2 Drosophila chromatin (prepared in the same manner). After immunoprecipitation, beads were recovered using a magnetic stand and washed with RIPA buffer (50 mM HEPES-KOH at pH 7.6, 500 mM LiCl, 1 mM EDTA, 1% NP-40, 0.7% Na-deoxycholate). Immunoprecipitated chromatin was eluted and cross-link-reverted overnight at 65°C. DNA was purified with QIAquick PCR purification kit (cat. no. 28106, Qiagen) and then quantified with QuantiFluor (cat. no. E2670, Promega). DNA libraries were prepared as previously described ⁷⁰ using 1 ng of DNA and the NEBNext® Ultra™ II DNA Library Prep Kit from Illumina (cat. no. E7645L). The purified DNA libraries were quantified with the Quantifluor reagent (cat. no. E2670, Promega) and the quality of the size was controlled with the Tapestation instrument (Agilent) using the high-sensitivity assay HD5000 (cat. no. 5067-5592, Agilent). DNA libraries were diluted to a working concentration of 4 nM and 50-bp paired reads (for H3K27ac, tetra-acetylated histone H4 and Hela MED1 ChIP-seq libraries) or 76-bp single-reads (for Raw 264.7 MED1 ChIP-seq libraries) were sequenced on the Illumina Novaseq6000 or on the NextSeq500 platforms respectively.

Quantification And Statistical Analysis

IP-MS data analysis

Acquired raw data were analysed using MaxQuant version 1.6.2.3 integrated with Andromeda search engine⁷¹. False discovery rate (FDR) was set to a maximum of 1% both at the peptides and protein level. Carbamidomethylcysteine and Methionine oxidation were selected as fixed and variable modifications, respectively. The UniProt mouse Fasta database UP000000589 (55315entries) was specified for the search. The LFQ intensity calculation was enabled. The “protein groups” output file from MaxQuant was first inspected using Perseus software 1.6.14.0⁵⁸ to filter out common contaminant proteins (keratin, desmoplakin, plectin and actin) and false positive hits (reverse hits from the Decoy database). LFQ intensity were log2 transformed and missing values were replaced by random numbers drawn from a normal distribution with a width of 0.3 and down shift of 1.8. In order to determine significantly enriched proteins (#129 of 257 proteins identified) a Welch’s t-test was used and hits with a minimum corrected p-value of 0.05 (FDR 5%) were considered for further analysis. Volcano plot was used for data visualization (Figure 1A) using R-statistical programming language version 4.0.3 (https://www.R-project.org/) and ggplot2 package. A list of all proteins confidently identified in each experiment are also reported in Table S1.

Histone PTM-MS data analysis

The acquired RAW data were analyzed using EpiProfile 2.0⁷², selecting the SILAC option, followed by manual validation. For each histone modified peptide, a % relative abundance (%RA) value for the sample (light channel - L) or the internal standard (heavy channel - H) was estimated by dividing the area under the curve of each acetylated peptide for the sum of the areas corresponding to all the observed forms of that peptide and multiplying by 100. Light/Heavy (L/H) ratios of %RA were then calculated and are reported in Table S3.

ChIP-seq data analysis

Spike-in H3K27ac and tetra-acetylated histone H4 ChIP-seq samples were analyzed similar to previous reports⁷³. Briefly the genome sequence of both the mouse (mm10 assembly) and the Drosophila genomes (assembly dm6) were concatenated to produce a chimeric combined genome, with the Drosophila chromosomes re-named to include the “_dm6” suffix. Custom Bowtie2 index was generated using the chimeric genome and used to sequence in pair-end mode each sample (with the following commands “-X 1000 -k 2 -t --phred33 -p 8 - q”). Mapped reads were then filtered by keeping reads mapped in proper pair and removing those unmapped, having mate unmapped, and failing quality. Duplicated reads were then removed using samtools markdup function. Finally reads split into mm10 and dm6 based on the chromosome suffix which was at the end removed. ChIP-seq peaks were identified using MACS2 (version 2.2.6) with parameters “‐‐keep-dup ‘all’ ‐‐call-summits ‐‐nomodel ‐‐nolambda ‐‐extsize 300”. Reference set of peaks was identified selecting significant peaks per-samples in addition to being consistent between replicated (overlapping area of at least 50% between replicates and FDR ≤ 10-5). ChIP-seq scaling factors were computed using the reads mapping to the dm6 genome. Differentially regulated peaks were identified using DESeq2 (R/Bioconductor package version 1.26.0; R version 3.6.2). While statistically significant changes were evaluated, we exploited the estimated Log2FC by DESeq2 to rank H3K27ac peaks and tetra-acetylated histone H4 according to the induction observed in control conditions. The experimental design for either ChIP-seq readout, specifically focusing on H3K27ac or tetra-acetylated histone H4, encompassed two replicates for each condition. The H3K27ac dataset encompassed two non-targeting (NT) and two MED14-targeting shRNA oligos. Each condition was performed with two replicates. The analysis was conducted while keeping the TSA and LPS stimulations separate. As a result, the untreated (UT) and nitric oxide (NO) conditions were common between the control and shMED14 conditions in both sets. The two distinct non-targeting and two MED14-targeting shRNAs were considered as pseudo-replicates (see GSE227964). Concordance between replicates and pseudo-replicates was evaluated, and the pairwise Spearman correlation coefficient in all samples ranged from 0.92 to 0.98.

MED1 ChIP-seq (in RAW264.7 mouse macrophages or Hela cells) and Flag ChIP-seq of the DLST and OGDH mitochondrial enzymes (in Hela cells) were analyzed using the nextflow workflow ⁷⁴ freely available at https://github.com/fgualdr/mmchipseq. Briefly, FASTQ samples were first filtered and trimmed using FASTP (v. 0.23.2) and then aligned against the mouse (mm10 assembly) or the human (hg38 assembly) genome using STAR (v. 2.7.9a) with the following parameters '--runMode alignReads', '--alignIntronMax 1', '--alignMatesGapMax 100000', '--alignEndsType EndToEnd', '--outSAMtype BAM Unsorted', '--readFilesCommand zcat', '--runRNGseed 0', '--outSAMattributes All'. Duplicate reads were marked using PICARD (v. 2.27.4). Bigwig coverage files were generated using deeptools (v. 3.5.1 see PMID: 24799436). ChIP-seq peaks were identified using MACS2 (v. 2.2.7.1). For Flag-tagged ChIP-seq of DLST and OGDH we selected peaks with an associated FDR ≤ 0.01, we further excluded any peak overlapping peaks called in the Empty-FLAG condition and further selected those with at least twice the coverage if compared to the Empty-Flag condition.

Chromatin-associated RNA-seq analysis

Chromatin-associated RNA-seq (ChAR) samples were analyzed following the nextflow workflow freely available at https://github.com/fgualdr/mmrnaseq. Briefly, FASTQ samples were first filtered and trimmed using FASTP (v. 0.23.2) and then aligned against the mouse (mm10 assembly) using STAR (v. 2.7.9a) with the following parameters ‘--alignEndsType EndToEnd’, ‘--quantMode TranscriptomeSAM’, ‘--twopassMode Basic’, ‘--outSAMtype BAM Unsorted’, ‘--readFilesCommand zcat’, ‘--runRNGseed 0’, ‘--outFilterMultimapNmax 100’, ‘--winAnchorMultimapNmax 100’, ‘--alignEndsProtrude 100 DiscordantPair’, ‘--outFilterScoreMinOverLread 0.4’, ‘--outFilterMatchNminOverLread 0.4’, ‘--alignSJDBoverhangMin 1’, ‘--outSAMattributes All’, ‘--quantTranscriptomeBan Singleend’, ‘--outSAMstrandField intronMotif’. Duplicate reads were marked using PICARD (v. 2.27.4). Bigwig coverage files were generated using deeptools (v. 3.5.1 see PMID: 24799436). Per-gene read counts were retrieved using the R/Bioconductor (v4.0.3) library GenomicAligment (v1.26.0) and the function summarizeOverlaps (Lawrence, 2013) against the UCSC_Mus_musculus.GRCm38.102 RefSeq annotation. Total read counts per gene were collected using the summarizeOverlaps method and used for downstream analysis. Sample normalization was achieved by selecting invariant genes across samples/conditions using the R library GeneralNormalizer (see https://github.com/fgualdr/GeneralNormalizer). Differentially regulated genes were selected using DESeq2 (R/Bioconductor package version 1.38.0; R version 4.2.2) after turning off the default normalization that DESeq2 applies. Genes differentially expressed with an associated absolute Log2FoldChange greater than 1 and an adjusted p-value less than or equal to 0.01 were considered (refer to individual plots and figures). Unsupervised clustering presented in Figure S3B was achieved using dimensionality reduction coupled with Louvain community detection. Gene set enrichment analysis was conducted against the MSigDB (msigdb_v7.5.1) computing the hypergeometric enrichment and the Standardized Jaccard index. GSEA analysis was conducted using the R/Bioconductor fgsea library (v. 1.26 see https://doi.org/10.1101/060012). The experimental design related to Figure 4E and K, encompassed three replicates for each condition. The analysis was conducted keeping the TSA and LPS stimulation separate, resulting in untreated (UT) and nitric oxide (NO) conditions being common between the two sets (see GSE227964).

Supplementary Material

Table S2. List of TCA and urea cycle metabolites measured by LC-MS, Related to Figure 2.

Bone-marrow primary macrophages were treated with LPS for 24h (10 ng/ml) with or without the iNOS inhibitor 1400W (100 μM) 30 minutes before LPS treatment.

Table S3. Quantitative analysis of histone post-translational modifications measured by LC-MS, Related to Figure 3.

Mouse bone marrow-derived macrophages were treated with TSA (500 nM) or TSA and the NO donor Spermine NONOate (500 μM) for 20 minutes

Table S4. ChIP-seq analysis, Related to Figure 4.

Summary of the H3K27ac ChIP-seq data reporting the 2,257 TSA-induced and the 3,328 LPS-induced ChIP-seq peaks. The mean read counts per conditions are reported, followed by the DESeq2 differential analysis in shNT infected cells (TSA vs. UT and LPS vs. UT respectively) (adjpval and Log2FoldChange).

Data and code availability

• Raw and processed sequencing data were deposited in the Gene Expression Omnibus (GEO) repository under the accession number GSE227964. The mass spectrometry data referring to the IP-MS experiments and the histone PTM analyses have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository ⁵⁸ with the dataset identifiers PXD040979 and PXD040937, respectively. All deposited data are publicly accessible as of the date of publication.

• This paper does not report any original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.