Abstract M1 macrophages enter a glycolytic state when endogenous nitric oxide (NO) reprograms mitochondrial metabolism by limiting aconitase 2 and pyruvate dehydrogenase (PDH) activity. Here, we provide evidence that NO targets the PDH complex by using lipoate to generate nitroxyl (HNO). PDH E2-associated lipoate is modified in NO-rich macrophages while the PDH E3 enzyme, also known as dihydrolipoamide dehydrogenase (DLD), is irreversibly inhibited. Mechanistically, we show that lipoate facilitates NO-mediated production of HNO, which interacts with thiols forming irreversible modifications including sulfinamide. In addition, we reveal a macrophage signature of proteins with reduction-resistant modifications, including in DLD, and identify potential HNO targets. Consistently, DLD enzyme is modified in an HNO-dependent manner at Cys^477 and Cys^484, and molecular modeling and mutagenesis show these modifications impair the formation of DLD homodimers. In conclusion, our work demonstrates that HNO is produced physiologically. Moreover, the production of HNO is dependent on the lipoate-rich PDH complex facilitating irreversible modifications that are critical to NO-dependent metabolic rewiring. Subject terms: Metabolic pathways, Immunology __________________________________________________________________ Nitric oxide has been shown to target mitochondrial aconitase 2 and pyruvate dehydrogenase to reprogramme macrophage metabolism. Here, the authors extend these findings to show that lipoate is used to generate nitroxyl in this process. Introduction Macrophages undergo profound metabolic changes that support their function in health and disease. The combination of lipopolysaccharide (LPS) and interferon-γ leads to differentiation of pro-inflammatory macrophages, often referred to as “M1” cells, with enhanced glycolytic flux and glutamine intake, and substantially reduced oxidative phosphorylation^[46]1. Glycolysis supplies the pentose phosphate pathway supporting the generation of NADPH to fuel the NADPH-oxidases for pathogen killing, as well biosynthetic pathways through the production of ribose^[47]2. Concurrently, nitric oxide (NO) following activation is directly lethal to microorganisms, while supporting glycolytic commitment by shutting down the use of pyruvate in the TCA cycle^[48]3–[49]5. In addition, NO suppresses mitochondrial aconitase (ACO2) activity, leading to the accumulation of citrate, which upon export to the cytoplasm supports inflammatory lipid production^[50]6–[51]8. Lastly, upregulation of aconitate decarboxylase (Acod)−1 leads to the production of the antimicrobial metabolite itaconate from the TCA intermediate cis-aconitate until NO concentration reaches levels that inhibit ACO2 function^[52]6,[53]9. Given that various metabolites such as NO drive unique phenotypic and metabolic characteristics^[54]8,[55]10, more study is required to investigate closely the molecular mechanisms driving metabolic adaptation during stimulation. A critical, but less studied, component of this metabolic re-organization is the inhibition of the pyruvate dehydrogenase (PDH) complex slowing glucose-derived carbon from entering the TCA cycle^[56]11,[57]12. Due to the relatively rapid kinetics of NOS2-dependent PDH shutdown during macrophage polarization, as well as the apparent lack of involvement of HIF-1α in enzyme inhibition^[58]6, we hypothesized that NO might directly target components of the PDH complex. The PDH complex consists of three different enzymes, E1, E2, and E3, which catalyze the conversion of pyruvate to acetyl-Coenzyme A (CoA), yielding NADH. Lipoate covalently bound to E2 is fundamental in accepting the acetyl group from E1, and transferring it to reduced CoA^[59]13. The resulting dihydrolipoate undergoes oxidation by the E3 enzyme (also referred to as dihydrolipoamide dehydrogenase; DLD) producing NADH. Interestingly, the DLD enzyme functions as a homodimer not only in PDH but also in α-ketoglutarate dehydrogenase (OGDH), branched-chain amino acid-dehydrogenase (BCKDH) and the glycine cleavage system (GCS), all within the mitochondrial matrix^[60]14,[61]15. Over the last 30 years, substantial attention has been given to then biological effects of NO and related oxidized nitrogen oxides such as NO[2] and N[2]O[3]. Nitroxyl (HNO) is more reduced than NO, and while the possibility of endogenous formation has long been hypothesized^[62]16–[63]20, the mechanisms have yet to be definitively established. Plausible routes for endogenous HNO production include the reaction of NO with hydrogen sulfide (H[2]S) and nucleophilic attack of an S-nitrosothiol (RSNO) by an adjacent thiol^[64]17,[65]21–[66]27. Compared with NO and other reactive nitrogen species (RNS), HNO has higher electrophilicity^[67]28, a distinct mechanism of action^[68]20,[69]29,[70]30 and high reaction rates with critical targets including thiols^[71]29,[72]31. Modification of thiols by HNO, to produce a reduction-resistant, irreversible sulfinamide (RS(O)NH[2]) has been observed in a varity of cellular systems^[73]17,[74]28. In part due to its ability to mediate reduction-resistant modifications driving enzyme inhibition^[75]32, there has been substantial interest in a physiological role for HNO as an alternative signaling agent to NO. HNO donors induce vasodilation and contribute to the prevention of atherosclerosis and vascular thrombosis, and HNO is a promising alternative clinical cardiac inotropic agent in heart failure^[76]33–[77]41. HNO donors also show efficacy in the treatment of alcoholism, cancer^[78]42 and pain^[79]43,[80]44. In sum, the distinct chemistry of HNO compared to NO suggest that biological targeting by HNO may be selective and localized, with critical thiols being hot spots or privileged targets^[81]45. As such, HNO has high potential to function as a tightly controlled signaling molecule, which can be exploited therapeutically^[82]19,[83]32,[84]39. Although there are numerous chemical processes by which HNO could be generated physiologically, and ample evidence of its effects, until now endogenous HNO and its biochemical fingerprint has yet to be fully established, perhaps in part due to it propensity for self-reaction^[85]28. Here, we investigate the mechanism of inhibition of PDH and identify a model of intramolecular generation of HNO during macrophage M1 polarization. This mechanism involves NO attack of PDH-E2 associated lipoates. The unique thiol configuration of lipoate facilitates the subsequent generation of HNO in close proximity with PDH-distal DLD. Irreversible, HNO-derived modification of DLD results in potent inhibition of PDH and mitochondrial adaptation to stimulation. This mechanism is consistent with the parallel inhibition of other DLD-containing enzyme complexes including OGDH and BCKDH. Intramolecular HNO generation in biological settings may provide a mechanism to induce site-specific responses while preventing uncontrolled modifications. Results Metabolic reprogramming of macrophages at PDH is dependent on NO fluxes and is associated with alterations of PDH-E2 cofactor lipoate We previously showed that metabolic adaptation due to PDH inhibition is NOS2-dependent^[86]6 and we sought to examine this mechanism in more detail. Using bone-marrow derived macrophages (BMDM), we first measured PDH enzymatic activity and oxygen consumption rates (OCR) of permeabilized cells fueled with pyruvate; both decrease in an NO-dependent manner (Fig. [87]1a, b). As lipoate levels have been shown by us and others to be altered during macrophage activation^[88]46,[89]47, we supplemented cell cultures with lipoate and found that this ameliorated reduction of OCR by NO (Fig. [90]1c). Since NO-derived RNS such as N[2]O[3] interact with thiol groups, we considered that NO might target PDH through its E2-bound cofactor, lipoate (Fig. [91]1d). To test this possibility, we performed high resolution proteomics on the E2 enzyme of BMDMs (Fig. S[92]1a). We detected lipoyl-bound peptides containing Lys^131 and Lys^258 of E2 (Fig. [93]1e) and higher abundance and lipoylation rates for Lys^131 compared to Lys^258 was observed (Fig. S[94]1b–e). When NO was present, our analysis detected a form of lipoate consistent with S-nitrosation on one thiol and acetylation of the second thiol of the same molecule (K-C[10]H[16]O[3]S[2]N) (Fig. [95]1f). Although less abundant than on Lys^131, this acetyl-nitroso lipoate modification was also found on Lys^258 (Fig. S[96]1f). We did not detect substantial signal consistent with a mixed S-nitroso/reduced thiol (Fig. S[97]1g) or a dual S-nitroso modification on lipoate (Fig. S[98]1h). Due in part to steric interactions, the latter case is not expected, while S-nitrosation of thiols is commonplace^[99]48. The apparent lack of a mixed S-nitroso/reduced thiol lipoate, (K-C[8]H[15]O[3]S[2]N), compatible also with a lipoyl-sulfinamide or a lipoyl-hydroxyl-disulfenamide, is consistent with increased reactivity. In particular, when thiol/-SNO ratios are low such as when protein -SNOs are accessible only to other proximal protein thiols, HNO generation is a possibility^[100]17. Moreover, in the presence of NO, BMDM peptides accumulated that contain lysine bound to dithiol- and acetylated-lipoate (Fig. [101]1g, h). This suggests that the enzyme is inhibited in this configuration most likely because of lack of cofactor recycling. Thus, the S-nitroso-acetyl-lipoate modification likely represents S-nitrosation occurring after the enzyme is inhibited at the acetylated state. Fig. 1. NO fluxes regulate PDH and alter PDH-E2 cofactor lipoate in macrophages. [102]Fig. 1 [103]Open in a new tab a Enzymatic activity of PDH in total cell lysates from WT BMDMs from male and female mice (mixed sex in all groups) after 16 h activation with LPS + IFNγ or DETA/NO (100 μM) (n > 3 mice). b Representative Seahorse analysis of permeabilized WT BMDMs from male and female mice (mixed sex in all groups) where state 3-OCR was elicited by a mixture of pyruvate and malate to measure PDH flux. Bar graphs show quantified respiration in 16 h activated WT (n > 8 wells). Data in (a) and (b) were analyzed by one-way ANOVA with Dunnet’s multiple comparisons test. c Quantified pyruvate/malate respiration from Seahorse analysis of permeabilized WT BMDMs from male and female mice (mixed sex in all groups) after 8 h activation, performed in the presence of vehicle or lipoic acid (n > 3 mice) shown as percentage of OCR relative to untreated. Data were analyzed by two-way ANOVA (Sidak’s post-tests). d Schematic illustration of PDH enzyme with emphasis on E2 subunit. e Interrogated peptide of PDH-E2 predicted to bind lypoyl group on functional lysine (K). f–h Major products, possible structures and abundances of E2 K^131-lipoylated peptide in mitochondrial lysates from WT BMDMs from male mice after 16 h stimulation with LPS + IFNγ or DETA/NO (500 μM) (n = 3 technical repeats) Data were analyzed by two-way ANOVA with Dunnet’s multiple comparisons test. All error bars display mean ± SEM. Data shown are representative of two or more independent experiments. Source data are provided as a [104]Source Data file. Modifications on DLD and the E3 binding protein of PDH are compatible with HNO exposure We previously demonstrated a decline in DLD activity in macrophages stimulated with LPS + IFNγ that is dependent on NOS2 and occurs concurrently with altered migration in native gels^[105]6, suggesting direct modification of DLD by NO. This raises the possibility that other mitochondrial dehydrogenases sharing DLD (Fig. S[106]2a) would be similarly targeted in macrophages. Metabolomics from stimulated WT and Nos2 ^−/− macrophages suggest that BCKDH, OGDH and GCS are indeed targeted in a NO-dependent manner (Fig. S[107]2b). To further strengthen this conclusion, we performed isotope tracing analysis using [U-^13C6]Leucine to assess BCKDH activity (Fig. [108]2a). A 4-h pulse with [U-^13C6]Leucine resulted in >70% of intracellular leucine being labeled. Approximately 40% of α-ketoisocaproate was also labeled suggesting prompt transamination, with equal rates of fractional labeling between WT and Nos2 ^−/− that was independent of stimulation status (Fig. S[109]2c). As anticipated, we observed decreased [U-^13C6]Leucine-derived carbon incorporation downstream of BCKDH in stimulated WT BMDMs but not in stimulated Nos2 ^−/− cells (Figs. [110]2b, c and S[111]2d, e). The isotopologue incorporation pattern confirmed predominant changes in the M + 5 and M + 3 isotopologues of 3-hydroxy-3-methylglutarate (HMG) and acetoacetate, respectively, indicative of disruption of BCKDH activity in a NOS2-dependent manner (Fig. [112]2b, c). Fig. 2. DLD and the E3 binding protein of PDH are likely modified by HNO. [113]Fig. 2 [114]Open in a new tab a Schematic illustration of atom transitions using uniformly labeled 13C-Leucine ([U-13C]) (labeled carbons in blue) as tracer for determination of mass isotopologue distributions to infer relative intracellular fluxes. BCAT Branched-chain amino acid aminotransferase, BCKDH Branched-chain α-keto acid dehydrogenase. b, c WT and Nos2 ^−/− BMDMs from male and female mice (mixed sex in all groups) were activated for 16 h with LPS + IFNγ and cultured 4 h with labeled tracer. Bars show carbon incorporation into M + 0–6 isotopologues of indicated intermediates (n = 4 mice per genotype). Data were analyzed by two-way ANOVA with Sidak’s post-tests. d DLD in-gel activity and native IB of mitochondrial fractions from WT BMDMs untreated or 16 h-stimulated with LPS + IFNγ (n = 9 total mice). Treatments with DETA/NO/DTT/lipoate were performed post harvest on lysates. e Mechanism of generation of HNO. f Outputs of reaction of HNO with common thiols. g DLD in-gel activity of mitochondrial fractions from WT BMDMs stimulated for 16 h with LPS + IFNγ and 4 h with AS. Bar graphs show quantified diaphorase activity (n = 3 technical repeats). h DLD dehydrogenase activity in mitochondrial lysates from WT after 16 h-stimulation with LPS + IFNγ or either 400μM or 5 mM AS or DEA/NO for 4 h (n > 3 mice). i Lysates from  (h) were incubated 1 h at 37 °C with reducing agents (10 mM) before assessment of DLD dehydrogenase activity. Basal values of activity of LPS + IFNγ, AS and DEA/NO from (h) were set to 1 (n > 3 mice). Data in (g–i) were analyzed by one-way ANOVA with Dunnet’s multiple comparisons test. j Schematic illustration of PDH enzyme with emphasis on DLD and E3bp. k E3bp-IB on native gels of mitochondrial fractions from BMDMs after 16 h stimulation with LPS + IFNγ or DETA/NO (500 μM), or with AS or DEA/NO for 4 h. Lysates pre-native PAGE were incubated with 10 mM DTT for 1 h at 37 °C (n = 8 total mice). All error bars display mean ± SEM. Data shown are representative of two or more independent experiments. p values > 0.05 are reported as “ns” unless otherwise specified. Source data are provided as a [115]Source Data file. NO was sufficient to inhibit DLD activity (Fig. [116]2d), and treating intact cells with lipoate restored DLD functionality (Fig. S[117]2f), consistent with our OCR results (Fig. [118]1c). However, adding lipoate to lysates did not restore DLD diaphorase activity (Fig. [119]2d), suggesting its actions are not due to reduction. Accordingly, treatment with DTT was also not effective in restoring DLD activity, indicating non-reversible modification (Fig. [120]2d). Exogenous lipoate caused no substantial changes in GSSG/GSH levels but led to an increase in lipoyl-GMP, suggesting activation of lipoate consistent with use of the salvage pathway for protein lipoylation (Fig. S[121]2g, h)^[122]49,[123]50. Altogether, these data suggest that rather than acting as a reducing agent, exogenous lipoate is likely supporting lipoylation of newly formed E2. Having determined that NO modifies lipoate, and that lipoate adducts regulate DLD in association with loss of activity, we investigated the chemistry responsible for DLD modification. The reactivity was compatible with formation of a sulfinamide by HNO formed through nucleophilic attack of a lipoyl-SNO by the vicinal thiol (Fig. [124]2e). Reaction of HNO with a thiol leads to an intermediate that can spontaneously rearrange to a sulfinamide or in the presence of excess thiol can react to produce a disulfide and hydroxylamine (Fig. [125]2f)^[126]28. Generation of a sulfinamide in a biological system represents a nearly irreversible thiol modification^[127]51. Treatment with Angeli’s salt (AS), an HNO donor, was sufficient to inhibit DLD activity in BMDMs (Fig. [128]2g, h) and confirmed the propensity for HNO self-reactivity at high concentrations, where its effect on PDH is lost. Moreover, compared with an NO donor (DEA/NO) with a comparable half-life of release to AS, AS was a more effective inhibitor in culture (Fig. [129]2h). Similarly, OGDH activity is preferentially inhibited by HNO as compared to NO (Fig. S[130]2i, j). Lastly, in experiments of potential rescue with DTT, glutathione (GSH) or Cysteine (Cys) in lysates^[131]52, these reducing agents did not substantially restore DLD activity in NO-rich cells (Fig. [132]2i). In addition, migration of the lipoate containing-E3-binding protein (E3bp) of PDH^[133]53 (Fig. [134]2j) on native gels changes during stimulation or treatment with AS, but not NO, and this migration is not affected by DTT (Figs. [135]2k and S[136]2k, l). These data suggest that HNO may be responsible for lipoate-related biochemical alterations in these physiological settings, and that components of PDH and other cellular dehydrogenases may be both HNO producers and targets of inhibition by HNO. HNO is generated by NO in the presence of reduced lipoate and is detectable in BMDMs Given the possible involvement of lipoate groups in endogenous HNO generation, we used chemical approaches to assess if the interaction of lipoate with NO can generate HNO in vitro. Although cellular dithiols can react with S-nitrosoglutathione (GSNO)^[137]54 to generate HNO, lipoate-mediated generation of HNO from NO and thiols has not been shown. We treated GSH with AS or DEA/NO in the presence of reduced or oxidized lipoate and GSH modifications were analyzed by ESI-LC/MS-MS (Fig. S[138]3a). In this system GSH functions as an HNO trap and GSH-sulfinamide (GS(O)NH[2]) is a selective biomarker for exposure to HNO^[139]17,[140]29,[141]51,[142]55. As expected GSH-sulfinamide was detected in the presence of AS (Fig. [143]3a). Remarkably, we measured substantial GSH-sulfinamide after incubation with DEA/NO and reduced lipoate but not oxidized lipoate (Fig. [144]3a), highly suggestive that exposure of reduced lipoate to NO can generate free HNO. We also assessed GSH modifications in reactions containing human recombinant DLD (rhDLD) and found that GSH-sulfinamide and sulfinic acid (-SO[2]H), a downstream hydrolyzed product, could be further increased (Figs. [145]3b, c and S[146]3c) with GSH concentrations dictating the amount of HNO-derived sulfinamide. Fig. 3. HNO is generated by NO in the presence of reduced lipoate and is detectable in BMDMs. [147]Fig. 3 [148]Open in a new tab a Test tube reactions were carried out in the presence of GSH and AS or mixtures of either oxidized or reduced-lipoate with DEA/NO for 1 h at 37 °C. GSH-sulfinamide was quantified by ESI-LC/MS-MS. Boxplots with floating bars (min to max) are shown with line at median. In (b) and (c) recombinant human DLD (rhDLD) was added to test tube reactions in the presence or absence of reduced lipoate and GSH-sulfinamide and sulfinic acid were quantified (n > 2 wells). Data were analyzed by one-way ANOVA with Dunnet’s multiple comparisons test. d WT BMDMs from male and female mice (mixed sex in all groups) were incubated with fluorescence probe for HNO for 30 min and stimulated with LPS + IFNγ for 4 h or with NOS2 inhibitor Aminoguanidine (AG) 1 h prior to stimulation. After PBS washing, fluorescence was measured in total cell lysates. For AS and DEA/NO conditions, cells were treated with donors for only 30 min. Data (n = 3 mice) were analyzed by one-way ANOVA with Dunnet’s multiple comparisons test. e Schematic illustration of mechanism of action of ML226 on ABHD11, which ensures stability of lipoate groups of PDH and OGDH enzymes. f Quantifications (percentage relative to LPS + IFNγ-stimulated cells) of fluorescence for HNO detection in cells treated with ML226 1 h prior to stimulation. Data (n = 3 mice) were analyzed by unpaired t test (two-tailed). All error bars display mean ± SEM. Data shown are representative of 2 or more independent experiments except (b) which is cumulative data from 2 independent experiments. Source data are provided as a [149]Source Data file. Next, we evaluated whether HNO was produced endogenously in our cell system. We incubated murine BMDM with a fluorescent probe for HNO quantification^[150]56 and readily detected an increase in fluorescence in LPS + IFNγ-treated vs unstimulated cells that was probe-specific (Figs. [151]3d and S[152]3d) and ablated with pre-treatments with the NOS2 inhibitor aminoguanidine (AG). Lastly, to interrogate the requirement of lipoate groups in endogenous HNO generation, we used ML226, an inhibitor of ABHD11, a mitochondrial hydrolase responsible for maintaining correct lipoylation of major α-keto-acid dehydrogenases^[153]57(Fig. [154]3e). ML226 treatment partially inhibited HNO-specific fluorescence (Fig. [155]3f). Taken together, these data demonstrate the ability of lipoate to generate HNO, that can be facilitated with additional DLD, and confirm that HNO generation within macrophages is NOS2- and lipoate-dependent, and capable of inducing reduction resistant modifications. Proteomic analysis reveals stable Cys-modifications in M1 BMDMs Next, we sought to detect the effects of endogenous HNO in BMDMs by surveying reduction resistant modified proteins at the level of Cys residues in the mitochondrial-enriched proteome. We undertook an approach with TCEP reduction followed by alkylation and tandem mass tag (TMT) proteomics analysis (Fig. [156]4a). The rationale included the possibility of detecting sulfinamides, sulfinic acid and sulfonic acid (-SO[3]H), which, unlike traditional -SNO modifications would be stable during reduction. Moreover, a decrease in alkylation-derived carbamidomethyl (CAM)-modification for a given peptide would indicate increased irreversible modification. Fig. 4. Proteomic analysis reveals stable cys-modifications in M1 BMDMs. [157]Fig. 4 [158]Open in a new tab a Experimental design for mitochondrial-enriched proteomics with a focus on non-reversible modifications. b Volcano plot showing significant proteins with reduction resistant cysteine modifications with FC < 1 in stimulated WT BMDMs from male and female mice vs ctrl (mixed sex in all groups) (n = 4 mice). Data were corrected for total protein abundance. In blue are indicated proteins belonging to interactome of DLD together with DLD itself. c Total GO enrichment of proteins from (b) (protein number = 149). d Proteomics network of protein-protein interactions, depicting proteins as nodes and interactions as edges. e Node color is associated with the assigned network cluster. The most enriched GO term associated with each cluster is indicated. f Venn diagram indicating number of proteins belonging to interactome of DLD out of total (4b); g Enrichment analysis of subgroup of proteins from 4b-f (protein number = 31). h Volcano plot showing significant proteins with reduction resistant cysteine modifications with FC < 1 in stimulated Nos2 ^−/− BMDMs from male and female mice and treated with AS vs stimulated only (mixed sex in all groups). Data were corrected for total protein abundance (n = 4 mice). i Venn diagram indicating groups of proteins from (b) and (h) with respective overlap indicating possible protein modifications driven by endogenous HNO. Pathway enrichment analysis (GO molecular function) ( j) and network analysis (k) of merged list from (i). Data shown are representative of two or more independent experiments. Source data are provided as a [159]Source Data file. We detected stimulation-induced, reduction resistant modifications at Cys on more than 200 proteins in WT macrophages. GO analysis of the 149 significantly modified targets (Fig. [160]4b) revealed proteins involved in metabolic processes, cellular respiration and oxidative pathways (Figs. [161]4c and S[162]4a–c). Network analysis of protein-protein interactions showed clusters associated with distinct functions, most importantly RNA metabolism, signaling and cellular respiration (Fig. [163]4d, e). These results indicate that many metabolic proteins are oxidized in inflammatory conditions and these cysteine residues are susceptible to reduction-resistant, and therefore unlikely reversible, modifications. Among these modified proteins, we found DLD with a modification at Cys^477, together with multiple proteins (31 out of 149, ~20%) belonging to components of the known interactome of DLD (Fig. [164]4f, g). These are proteins that are likely to be in the proximity of PDH/lipoate-generated HNO. This highlights potential strong susceptibility of DLD, and proteins in its network, to be inhibited by HNO-mediated irreversible oxidation under inflammatory conditions. We then performed the same proteomic screen using stimulated Nos2 ^−/− BMDMs and attributed ~54% of the reduction-resistant proteins in WT BMDMs to NOS2 expression (Fig. S[165]4d). These proteins were enriched in pathways involving phagocytosis, cellular metabolism and respiration (Fig. S[166]4e, f), with a major subgrouping in organic acid metabolism and, reassuringly, nitrogen-compound metabolic processes (Fig. S[167]4g). Finally, we treated stimulated Nos2 ^−/− BMDMs with AS to identify possible direct HNO-modified proteins and found 75 proteins significantly modified (Fig. [168]4h), 32 of which overlap with reduction-resistant proteins in WT cells (Fig. [169]4i). Enrichment analysis of these targets showed dominant representation in mitochondrial transport as well as sugar and lipid metabolism related proteins (Fig. [170]4j, k). Within DLD, our proteome dataset detected only the CAM-versions of Cys^477- and Cys^484-containing peptides indicating reversible modifications. When interrogating NOS2 dependency of modifications, the data on Cys^477 did not pass our filtering parameters, although CAM-peptides of Cys^477 in Nos2 ^−/− cells indicated decreased modification (Fig. S[171]4h) in the absence of NO. In addition to identifying targets of modification we used this dataset to gain insight into the biological effects of macrophage exposure to HNO. We assessed relative total abundance of peptide/proteins across the proteome and we measured secreted cytokine/chemokine levels. We found 64 proteins to be significantly upregulated and 47 proteins downregulated in Nos2 ^−/− macrophages stimulated in the presence of AS (using a cutoff of FC = 1.25) (Fig. S[172]4i). This target list suggests that the addition of HNO alters the levels of proteins involved mainly in phagocytosis, intracellular transport and autophagy (Fig. S[173]4j, k). Our analysis of secreted factors showed that although cytokines classically associated with macrophage pro-inflammatory activation such as IL-6 and TNFα were not affected by the presence of the HNO, the production of CCL1, IL-1α, C5a, and GCSF was increased with HNO exposure. In contrast, the secretion of CXCL10 and KC was reduced with AS treatment (Fig. S[174]4l). Together these data show that exogenous HNO may tune the inflammatory status of macrophages. HNO drives susceptibility of modification at Cys^484 within DLD To look more closely at specific effects on DLD, we performed kinetic studies of DLD dehydrogenase activity upon incubation of rhDLD with either HNO or NO donors. These experiments revealed a stronger inhibitory effect of AS (Figs. [175]5a, b and S[176]5a, b) than DEA/NO, recapitulating what we found in BMDMs (Fig. [177]2h). This is generally consistent with the reactivity of these related nitrogen oxides. In these assays we also excluded the possible effect of nitrite, a major byproduct of AS decomposition, on DLD activity (Fig. S[178]5a), strengthening our hypothesis that HNO itself is responsible for DLD inhibition. Fig. 5. HNO drives susceptibility of modification at Cys^484 within DLD. [179]Fig. 5 [180]Open in a new tab a Dose response regression curves for dehydrogenase activity of rhDLD in the presence of AS or DEA/NO (n = 10 wells). Data were analyzed by non linear fit (dose-response -inhibition) (R squared = 1 for AS, 0.95 for DEA/NO). b Native in-gel activity assay of rhDLD incubated with 500 μM or 2 mM AS or DEA/NO for 1 h at 37 °C. Bar graphs show quantified diaphorase activity (n = 3 wells). Proteomics analysis of modifications of rhDLD at Cys^477 (c) and Cys^484 (d) in the presence of AS or DEA/NO (n = 3 wells). Data in (b, c, d) were analyzed by one-way ANOVA with Dunnet’s multiple comparisons test. e Targeted proteomics assay (PRM) on DLD Cys^484 -containing peptide on mitochondrial lysates from WT and Nos2 ^−/− BMDMs from male and female mice (mixed sex in all groups) stimulated with LPS + IFNγ and AS, indicating Log2FC values (relative to unstimulated cells) of carbamidomethyl (CAM)- Cys^484 containing peptide (n = 3 mice). f TMT global mitochondrial proteomic analysis on mitochondrial lysates from WT cells from male mice stimulated with LPS + IFNγ for 16 h and treated with BSO for 4 h, indicating normalized abundance of CAM- Cys^484 containing peptide. Data (n = 4 mice) were analyzed by unpaired t test (two-tailed). g Proteomic LFQ analysis of Cys-sulfinic acid modifications of rhDLD at Cys^484 in the presence of AS or DEA/NO and equimolar GSH (n = 3 mice). h PCA plots of proteomics data of IAM-PEG[2]/TMT labeling of mitochondrial lysates from WT and Nos2 ^−/− BMDMs from male and female mice (mixed sex in all groups) stimulated with LPS + IFNγ and AS, for identification of free cysteines before (h) and after (i) peptide enrichment. j Quantification of IAM-PEG2-enriched (free) Cys^484-containing peptide of DLD in conditions from (h, i). Basal values in ctrl were set to 1. Data (n > 3 mice) were analyzed by two-way ANOVA (Sidak’s post-tests). All error bars display mean ± SEM. Data shown are representative of two or more independent experiments. p values > 0.05 are reported as “ns” unless otherwise specified. Source data are provided as a [181]Source Data file. We next sought to identify post-translational modifications (PTM)s on DLD using high-resolution proteomics and to confirm the presence of HNO-dependent PTM(s). We confirmed that AS-driven modifications would appear as sulfinic acid rather than the Cys sulfinamide; oxidation from sulfinamide to sulfinic acid during proteomics sample prep is indeed a previously described phenomenon^[182]58(Fig. S[183]5c, d). Experiments on recombinant DLD (Fig. S[184]5e) using both label free and TMT, showed five cysteines modified by RNS: Cys^80, Cys^302, Cys^312, Cys^477 and Cys^484. Cys^80, together with Cys^302 and Cys^312, were only seen in the sulfonic acid state (+48) (Fig. S[185]5d, f–h) which was increased with all treatments tested when compared to control. These Cys modifications were not specific to HNO, but rather reflect a sensitivity to oxidation. In contrast, Cys^477 was detected carrying either a sulfinamide (+31), sulfinic acid (+32) or sulfonic acid in the presence of AS or DEA/NO (Figs. [186]5c and S[187]5i) with sulfinamide and sulfonic acid modifications specific for AS. Assessments of Cys^484 showed a greater increase in sulfinic acid with AS vs DEA/NO (Fig. [188]5d). Importantly these two cysteines (Cys^477 and Cys^484) were also detected by our TMT assay on BMDMs mitochondrial-enriched proteomes, revealing the importance of these two sites for oxidative modifications in physiological settings. Here also we excluded nitrite to be responsible for modifying DLD (Fig. S[189]5j). It is important to note that the pattern of modifications of Cys^477 and Cys^484 was different suggesting distinct reactivity. Although both Cys are clearly modified in an AS-dependent manner, Cys^477 seemed to be particularly sensitive to degrees of oxidation which affected detectability (Fig. S[190]5k). In contrast, Cys^484 was consistently detected as sulfinic acid-modified across our approaches. Moreover, analysis of the unmodified peptides showed reductions on Cys^484 in treatments with AS or DEA/NO consistent with reduction-resistant modification (Fig. [191]5d), while this was not the case for Cys^477 (Fig. [192]5c). As a result of these observations, we focused further analysis of physiologically relevant modifications in BMDMs on Cys^484 specifically. We used labeled, chemically synthesized Cys^484 peptides of DLD (Fig. S[193]5a) to track PTM status through our targeted proteomics pipeline. We found that concentrations of AS higher than 0.2 mM result in the highest HNO-specific oxidation state, peaking at modification of ~6% of total protein (Fig. S[194]6b, c). When applying this assay to stimulated BMDMs, we failed to detect Cys^484 sulfinic acid modified peptides. However, we readily measured reductions in the levels of unmodified peptide when macrophages were stimulated or treated with donors (Fig. [195]5e) indicating a conversion of these peptides to a reduction-resistant modification state. Importantly, levels of unmodified peptide fall in a NOS2-dependent manner and crucially, Nos2 ^−/− cells restored likelihood of modification of Cys^484 during exposure to HNO (Fig. [196]5e). PTMs like lipid peroxidation or modifications with phosphine groups derived from TCEP-RNS interactions were not detected, suggesting they are unlikely to account for the peptide loss of unmodified DLD. While these data do not definitively prove specific HNO-mediated modifications, when cellular GSH was depleted with BSO, pushing the equilibrium of HNO chemistry towards sulfinamide (Fig. [197]2f), even greater decreases in unmodified Cys^484 peptide were seen, consistent with increases in modification specifically at this site (Figs. [198]5f and S[199]6d). We further confirmed this using rhDLD (Fig. [200]5g) demonstrating that GSH levels control the degree of bioavailability of HNO and likely limit the detectability of modified Cys^484 in our cellular systems. Considering this, we contemplated the possibility that the cellular pool of DLD may encompass forms with both reversible and reduction-resistant modifications driven or perturbed by RNS. Therefore, we used IAM-PEG[2]-biotin tags to alkylate reduced (free) Cys thiols to assess sites that may undergo modification in macrophages, and enriched conjugated peptides were assessed by MS (Fig. S[201]6e). Samples clustered differently at PCA analysis (Fig. [202]5h, i), highlighting differences induced by stimulation and the presence of NO and HNO. Peptides containing Cys^69 and Cys^484 of DLD showed enrichment in this assay. LPS + IFNγ stimulated WT cells had higher levels of IAM-PEG[2] tagged Cys^484 peptides than unstimulated, indicating higher free Cys (Fig. [203]5j). Interestingly stimulated Nos2 ^−/− macrophages showed even higher levels of IAM-PEG[2] Cys^484 peptides, while AS treatment brought these levels down to the stimulated WT. This effect was specific for Cys^484 (Fig. S[204]6f). Together these data support the notion that during macrophage stimulation, Cys^484, and to a lesser extent Cys^477, of DLD can be modified by HNO resulting in a reduction-resistant PTM detectable as Cys-sulfinic acid. Although this exact modification in intact stimulated cells appears difficult to detect, our data suggest that Cys^484 may be substantially oxidized in resting macrophages, likely to a disulfide, that may be reduced to modify enzyme activity during stimulation. Moreover, it appears that HNO may dictate the extent to which the disulfide is present. Once available, the increased free Cys^484 in LPS + IFNγ treated cells is subject to direct modification by HNO resulting in a reduction-resistant drop in PDH activity. Cys^477 and Cys^484 sulfinamide impair DLD homodimer formation To assess how the Cys^484 DLD modifications might affect PDH activity we undertook an in silico modeling approach utilizing available structural data. Current human DLD (hDLD) structures show that known disease-causing mutations map to three locations in the human enzyme: the dimer interface, the active site, and the FAD and NAD (+)-binding sites^[205]59,[206]60. We used PyMOL to build energetically minimized structural models of the hDLD ([207]PDB_ID: 3rnm) modified at Cys^484 by adding a sulfinamide functional group in place of the native thiol. Our analysis showed that Cys^484 likely forms intra- and inter-chain reciprocal H-bond interactions at the monomer-monomer interface^[208]61 with residues His^364; Asp^368; Cys^388; Pro^390; Ala^457; Arg^482; Val^483; His^485. All the above cited residues, alternatively from both chains, locate within 4 Å from Cys^484 from chain A or chain B. Notably, along conformation samplings, Cys^484 and Cys^388 occupy positions less than 2.5 Å distant, making them likely involved in a possible (transient) interchain disulfide bond. Our structural analysis shows that introduction of a sulfinamide group at Cys^484 perturbs the local H-bond interaction network and would prevent the formation of a disulfide bridge between Cys^484 and Cys^388. Specifically, interactions between Cys^484 (chain A) and His^364 (chain B) are weakened (their distance becomes greater than 4 Å) and new interactions with Arg^495 (chain B) are strengthened (Fig. [209]6a, b). Notably, Cys^484 is the cysteine closest to the FAD cofactor, being only 8 Å away from it (see also Fig. S[210]6g). Cys^477 is the cysteine residue closest to the DLD-E2b subunit (chain E), mapping to 10 Å from it. Fig. 6. Cys^477 and Cys^484 sulfinamide impair DLD homodimer formation. [211]Fig. 6 [212]Open in a new tab a Top view of dihydrolipoamide dehydrogenase (DLD_E3 dimer) in complex with the subunit binding of human dihydrolipoamide transacylase (E2b). DLD_E3 dimer is reported in gray (chain A) and blue (chain B) cartoon representation, whereas E2b is reported in pink cartoon representation. FAD and NAD+ are reported in orange and sticks representation, respectively. Residues within 4 Å from C484 are reported in white (chain A) and blue (chain B) sticks and labeled. Dashed lines indicate the inter-chain interaction between the indicated Cys^388 and Cys^484 residues. The reported distances are lower than 2.5 Å. b Residues within 4 Å from Cys^484sulfinamide (sC484) are reported in white (chain A) and blue (chain B) sticks and labeled. Dashed lines indicate the inter-chain distance at the level of the indicated Cys^388 and sCys^484 residues. The reported distances are greater than 5.5 Å. c HEK293T cells were transfected with either empty vector/WT DLD/DLD C484A/DLD C484W. SDS IBs were perfomed probing for the expression tag Myc-DDK, DLD and β-actin as loading control. d Whole cells lysates where assessed for DLD dehydrogenase activity (n = 5 technical repeats). Data were analyzed by one-way ANOVA with Dunnet’s multiple comparisons test. e DLD in-gel activity assay and IBs on native gels of whole cells lysates from HEK293T transfected as in (c). f Lysates from (d) were used for kinetic assay for DLD dehydrogenase activity with increasing concentration of lipoic acid (n = 4 technical repeats). Non linear regression model of Michealis-Menten is shown, together with Lineweaver-Burk plot. All error bars display mean ± SEM. Data shown are representative of two or more independent experiments. Source data are provided as a [213]Source Data file. We next used the FOLDX repair tool and the Yasara Minimization server to optimize and relax the 3D models, as described in the Methods section. The interaction energies calculated at the monomer-monomer interface within the DLD dimer or at the dimer-monomer interface within the DLD E3-E2b trimer, both in the unmodified and in the DLD E3-E2b Cys^484 sulfinamide modified complex, gave a negative binding energy value (Table [214]1) that was lower in the native configuration than in the modified complex. This confirmed the expectation of a weakened binding interaction in presence of the modified Cys^484 sulfinamide residue. This was further confirmed also by relaxing the models in the Rosetta force field (Table [215]2). Table 1. Energy calculations about the investigated 3D models Interaction energies (FoldX AnalyseComplex) WT C484sulfinamide modified DLD Evaluated parameters InterChain.A_B InterChain.AB_E InterChain.A_B InterChain.AB_E Group1 A AB A AB Group2 B E B E IntraclashesGroup1 29,5201 68,3825 30,0418 70,7886 IntraclashesGroup2 34,9679 2,75887 36,619 2,90248 Interaction Energy −90,6667 −9,56689 −88,5912 −9,45471 Backbone Hbond −20,4434 −1,89106 −18,8878 −1,892 Sidechain Hbond −40,3367 −6,84229 −37,4774 −6,75942 Van der Waals −73,7061 −7,66983 −73,7899 −7,63457 Electrostatics −8,97054 −2,11671 −9,12802 −2,12716 Solvation Polar 93,908 14,3846 93,3047 14,3999 Solvation Hydrophobic −100,265 −9,20305 −100,58 −9,10423 Van der Waals clashes 2,42316 0,207425 2,54317 0,201281 entropy sidechain 42,492 4,53423 40,836 4,41639 entropy mainchain 15,6268 1,0743 15,9141 1,07426 torsional clash 1,47145 0,221081 1,58459 0,205195 backbone clash 14,7392 1,9886 14,7351 1,98845 helix dipole −1,81962 −1,55306 −1,6856 −1,53745 electrostatic kon −1,76226 −0,712563 −1,92266 −0,696887 energy Ionization 0,715411 7,77E−16 0,698004 −5,55E−16 Entropy Complex 2384 2384 2384 2384 Number of Residues 994 994 994 994 Interface Residues 195 29 197 29 [216]Open in a new tab Chain A and B indicate DLD_E3 chain within [217]3rnm.pdb and in the corresponding Cys^484-sulfinamide modified protein. Chain E indicates the E2b subunit within 3rnm.pdb and in the corresponding Cys^484-sulfinamide modified protein. Bold numbers indicate interdomain interaction energies of energetically relaxed protein complexes. “PDB.Chain” indicates the chain of the PDB used within the indicated analyses on the cited crystallized structures or models obtained as described in the Methods section. Table 2. The PDB.Chains used in the analyses with Rosetta were the same indicated in Table [218]1 INTERFACEANALYZER APP WT.A_B WT.AB_E C484.sulfinamide.A_B C484.sulfinamide.AB_E complex_normalized 0829 0921 0878 0932 dG_separated −164,508 −14.133 −157.316 −14.048 dG_separated/dSASAx100 −2066 −1.237 −1.975 −1.235 dSASA_hphobic 4412,5 631.604 4.469.150 621.870 dSASA_int 7962,872 1.142.113 7.967.294 1.137.917 dSASA_polar 3550,372 510.508 3.498.144 516.047 delta_unsatHbonds 34 3.000 33.000 1.000 dslf_fa13 1139 1.139 1.139 1.139 fa_atr −6083,011 −6.081.043 −6.096.449 −6.096.449 fa_dun 2707,8 2.707.800 2.712.322 2.712.322 fa_elec −1566,988 −1.567.587 −1.562.165 −1.562.165 fa_intra_rep 12,356 12.357 12.705 12.705 fa_intra_sol_xover4 215,991 215.991 216.419 216.419 fa_rep 5734,169 5.732.690 7.159.078 7.159.078 fa_sol 3821,65 3.816.065 3.828.241 3.828.241 hbond_E_fraction 0242 0204 0257 0205 hbond_bb_sc −150,436 −150.436 −150.722 −150.722 hbond_lr_bb −238,149 −238.149 −238.149 −238.149 hbond_sc −78,691 −78.691 −77.714 −77.714 hbond_sr_bb −287,354 −287.354 −287.354 −287.354 hbonds_int 43 4.000 42.000 4.000 lk_ball_wtd −61,909 −61.907 −63.055 −63.055 nres_all 940 991.000 940.000 991.000 nres_int 274 51.000 274.000 51.000 omega 137,884 167.863 167.863 167.863 p_aa_pp −168,903 −167.453 −167.453 −167.453 packstat 0636 0513 0618 0618 per_residue_energy_int 0,39 0826 0424 0857 pro_close 684,611 684.611 684.611 684.611 rama_prepro 159,539 170.234 170.234 170.234 ref 453,216 453.216 453.216 453.216 sc_value 0666 0491 0656 0489 side1_normalized 0249 −0212 0322 −0159 side1_score 34,109 −6.157 44.114 −4.615 side2_normalized 0531 2.195 0526 2.197 side2_score 72,793 48.285 72.095 48.342 [219]Open in a new tab A list of the energy terms taken from [220]https://www.rosettacommons.org/docs/latest/application_documentati on/analysis/interface-analyzer follows. Energy term abbreviation meanings: dslf_fa13 indicates disulfide geometry potential; fa_atr indicates Lennard-Jones attractive between atoms in different residues; fa_dun indicates the internal energy of sidechain rotamers; fa_elec indicates coulombic electrostatic potential with a distance-dependent dielectric; fa_intra_rep indicates Lennard-Jones repulsive between atoms in the same residue; fa_rep indicates Lennard-Jones repulsive between atoms in different residues; fa_sol indicates Lazaridis-Karplus solvation energy; hbond_bb_sc indicates sidechain-backbone hydrogen bond energy; hbond_lr_bb indicates backbone–backbone hbonds distant in primary sequence; hbond_sc indicates sidechain–sidechain hydrogen bond energy; hbond_sr_bb indicates backbone–backbone hbonds close in primary sequence; pro_close indicates Proline ring closure energy and energy of psi angle of preceding residue; rama indicates Ramachandran preferences; ref indicates reference energy for each amino acid;