Abstract

   Staphylococcus aureus is a leading cause of healthcare-associated
   pneumonia, contributing significantly to morbidity and mortality
   worldwide. As a ubiquitous colonizer of the upper respiratory tract, S.
   aureus must undergo substantial metabolic adaptation to achieve
   persistent infection in the distinctive microenvironment of the lung.
   We observed that fumC, which encodes the enzyme that converts fumarate
   to malate, is highly conserved with low mutation rates in S. aureus
   isolates from chronic lung infections. Fumarate, a pro-inflammatory
   metabolite produced by macrophages during infection, is regulated by
   the host fumarate hydratase (FH) to limit inflammation. Here, we
   demonstrate that fumarate, which accumulates in the chronically
   infected lung, is detrimental to S. aureus, blocking primary metabolic
   pathways such as glycolysis and oxidative phosphorylation (OXPHOS).
   This creates a metabolic bottleneck that drives staphylococcal FH
   (FumC) activity for airway adaptation. FumC not only degrades fumarate
   but also directs its utilization into critical pathways including the
   tricarboxylic acid (TCA) cycle, gluconeogenesis and hexosamine
   synthesis to maintain metabolic fitness and form a protective biofilm.
   Itaconate, another abundant immunometabolite in the infected airway
   enhances FumC activity, in synergy with fumarate. In a mouse model of
   pneumonia, a ΔfumC mutant displays significant attenuation compared to
   its parent and complemented strains, particularly in fumarate- and
   itaconate-replete conditions. Our findings underscore the pivotal role
   of immunometabolites in promoting S. aureus pulmonary adaptation.

   Subject terms: Pathogens, Infection, Bacterial infection, Bacterial
   infection
     __________________________________________________________________

   This study reveals how the pathogen Staphylococcus aureus adapts to the
   lung microenvironment, rich in host metabolites fumarate and itaconate,
   by using a key enzyme, FumC, to support its metabolic fitness,
   survival, and persistence.

Introduction

   Staphylococcus aureus frequently causes healthcare-associated
   pneumonia^[80]1. Antibiotic susceptible as well as resistant strains
   pose a significant risk for prolonged airway infection and associated
   morbidity^[81]2. This risk is heightened in patient populations with
   underlying pulmonary pathology, such as chronic obstructive pulmonary
   disease (COPD), cystic fibrosis (CF), and following viral
   infection^[82]3,[83]4. S. aureus is an especially formidable pathogen
   due to the expression of many surface proteins that thwart innate
   immune clearance^[84]5. While toxin-producing epidemic isolates can
   cause a fulminant necrotizing pneumonia^[85]6, typically the bacteria
   adapt to the airway milieu to cause a more indolent but intractable
   infection. These less immunostimulatory strains are better tolerated by
   the host as they fail to induce as robust an immune response that is
   detrimental to the host and pathogen alike. Of note, vaccine strategies
   that target classic virulence factors have failed^[86]7, indicating the
   need for alternative approaches to control staphylococcal infection.

   Among the common respiratory pathogens, S. aureus is notable for its
   substantial metabolic flexibility^[87]8. In contrast to more fastidious
   pathogens, S. aureus readily adapts to local carbon sources to optimize
   its metabolism. The infected airway provides a unique microenvironment
   with a distinctive set of metabolites and molecules produced by both
   the host and pathogen including reactive oxygen species, and DNA from
   lysed or damaged immune and stromal cells. By studying patterns of gene
   expression in clinical isolates of S. aureus from patients with
   persistent pneumonia, we observed staphylococcal adaptation to airway
   metabolites that accumulate during ongoing infection^[88]9. In a survey
   of methicillin resistant S. aureus (MRSA) strains isolated from chronic
   pneumonia, we noted over 100- to 1000-fold increases in the expression
   of fumC, which encodes the bacterial fumarate hydratase (FH) that
   converts fumarate to malate^[89]10. Fumarate, a potent proinflammatory
   metabolite that induces the production of TNFα and IFNβ in stimulated
   macrophages, is regulated by the host FH to limit
   immunopathology^[90]11. While the immunoregulatory function of host FH
   is well documented particularly in diseases associated with excess
   inflammation, such as systemic lupus erythematosus and renal cell
   carcinomas^[91]11–[92]13, less is known about the importance of S.
   aureus FumC in pneumonia and how it may contribute to the inflammatory
   response that determines host pathology induced by S. aureus infection.

   We hypothesize that the regulation of fumarate is an important
   component of S. aureus metabolic adaptation to the infected airway. In
   this report, we illustrate the unexpectedly central role of FumC and
   fumarate metabolism in orchestrating numerous metabolic processes in S.
   aureus relevant to its role as a pulmonary pathogen. Our findings raise
   the possibility of selectively targeting staphylococcal FumC to prevent
   chronic airway infection.

Results

Conservation of fumC across a global collection of S. aureus isolates

   We postulated that the importance of fumC in S. aureus adaptation to
   the lung is reflected in its conservation across clinical isolates.
   Specifically, we anticipate fewer non-synonymous substitutions and
   protein-truncating mutations in fumC compared to other genes. We
   analyzed a global dataset comprising 820 serially collected isolates
   from 181 patients with cystic fibrosis (CF) (Fig. [93]1A, Supplementary
   Data [94]1, 2). Using an established statistical genomics model of
   within-host evolution, we observed a relative mutation rate, λ, less
   than 1 in fumC from CF isolates (Fig. [95]1B, Supplementary Data [96]3,
   4). This indicates fewer mutations in fumC from pulmonary isolates than
   the mean number of mutations across the genome. Similarly, mqo,
   encoding the malate-quinone oxidoreductase enzyme involved in
   downstream malate metabolism to oxaloacetate^[97]13, displayed high
   conservation, without any mutations in pulmonary isolates (Fig. [98]1B,
   C). Conversely, lqo, which encodes a lactate-quinone oxidoreductase for
   lactate utilization^[99]14, exhibited a substantially higher relative
   mutation rate (λ = 5.49, p = 1 × 10^−5, Supplementary Data [100]4).

Fig. 1. S. aureus adapts to the metabolic airway milieu.

   [101]Fig. 1
   [102]Open in a new tab

   A Phylogenetic tree illustrating within-host evolution of S. aureus
   during persistent pulmonary infection (cystic fibrosis, blue clade).
   This model assumes genetic bottlenecks upon transmission and expansion
   of a single lineage during infection. Only mutations acquired within
   the host (i.e., blue branches) are considered in the analysis. B Output
   of the convergence analysis: the size of the dots is proportional to
   the number of independent (i.e., acquired de novo within the host)
   protein-altering mutations in genes of the TCA cycle and surrounding
   pathways. The relative mutation rate is equivalent to a rate ratio in
   Poisson models and calculated as: (mutations in gene x/length of gene
   x)/(mutations in all genes/length of all genes). A rate <1 (gray
   shading) indicates less mutations than the mean across the genome,
   suggesting that the gene is conserved during infection. GLY glycolysis,
   GLN gluconeogenesis, PPP pentose phosphate pathway, TCA tricarboxylic
   acid/Krebs cycle, UREA urea cycle. C Gene maps with position, type, and
   evolutionary niche of the de novo mutations. D Relative levels of
   fumarate and itaconate in sputum from healthy subjects (HS) and
   patients with cystic fibrosis (CF). Each data point is the mean ± SEM;
   n = 5 (HS) and n = 9 (CF). Statistical significance was determined
   using the Mann–Whitney U test (**p = 0.0040, ***p = 0.0010).

   We next evaluated how the only acquired single nucleotide polymorphism
   (SNP), E250G, identified in fumC from pulmonary isolates (Supplementary
   Fig [103]1A) may affect the enzymatic function. Our prediction of the
   tetrameric structure of FumC through homology modeling and molecular
   dynamics (MD) simulation indicated that the glutamate residue 250
   (E250) is positioned away from the active site (Supplementary
   Fig [104]1B), and that the SNP E250G is unlikely to inhibit its
   enzymatic activity. The highly negative surface charge (−78) of S.
   aureus FumC compared to the neutral human FH homolog (Supplementary
   Fig [105]1C) suggests differences in substrate affinity and catalytic
   efficiency. This observation, along with the conserved nature of fumC
   with minimal mutations in CF isolates, supports the pivotal role of
   FumC in chronic lung infections and its potential for selective
   inhibition.

Exogenous fumarate re-directs S. aureus metabolic activity

   We explored how high fumarate levels impact S. aureus. We confirmed the
   presence of substantially elevated levels of fumarate in the infected
   airways of CF patients as compared with healthy controls
   (Fig. [106]1D). S. aureus preferentially utilizes glucose as a carbon
   source to generate energy (ATP) through glycolysis^[107]15. In the
   infected lung, where the predominant recruited neutrophils compete with
   the bacteria for the already limited airway glucose^[108]16, we
   postulate that S. aureus adapts to fumarate using FumC. We assessed the
   impact of exogenous fumarate on the transcriptional response of S.
   aureus USA300 LAC by bulk RNA-seq analysis and observed a major impact
   on metabolic genes (Fig. [109]2A, B). The lacA/B/C/D/E/F genes were
   significantly downregulated by fumarate (Fig. [110]2A, B), indicating
   decreased lactose catabolism to galactose and glucose, two main carbon
   sources utilized by S. aureus for glycolysis. This was consistent with
   decreased glycolytic rates, which we confirmed by extracellular
   acidification rate (ECAR), in WT LAC and especially in the ΔfumC mutant
   that cannot catabolize fumarate (Fig. [111]2C). We also noted that
   exogenous fumarate inhibited bacterial oxidative phosphorylation
   (OXPHOS) represented by the oxygen consumption rate (OCR)
   (Fig. [112]2D). The purl/M/N/H genes were downregulated (Fig. [113]2A),
   suggesting decreased purine synthesis. In contrast, we observed
   increased expression of the argG/H genes, which are involved in the
   urea cycle and promote biofilm formation^[114]17.

Fig. 2. Fumarate imposes metabolic stress on S. aureus.

   [115]Fig. 2
   [116]Open in a new tab

   A Volcano plot showing significantly differentially expressed genes in
   WT LAC in the presence (100 mM) and absence of fumarate above the
   dotted line; n = 3 biological samples (1 experiment). Count-based
   differential expression analysis was performed using DESeq2. Gene
   expression was modeled using a negative binomial distribution with
   gene-specific dispersion estimates. Statistical testing was conducted
   on raw counts. Two-sided Wald tests were used for hypothesis testing,
   and p-values were adjusted for multiple comparisons using the
   Benjamini-Hochberg method to control the false discovery rate (FDR). B
   Pathway enrichment analysis of genes from (A) by Gene Ontology
   depicting significant differences in metabolic pathways in the presence
   and absence of fumarate. Gene set enrichment analysis (GSEA) was
   conducted in R with ‘fgsea’ using the Wald statistic to assign rank and
   using ‘msigdbr’ to query Gene Ontology reference gene sets at MSigDB
   (Broad Institute, MIT). P values were adjusted for multiple comparison
   using Benjamini-Hochberg method. C Glycolysis and D oxidative
   phosphorylation (OXPHOS) of S. aureus, as measured by the extracellular
   acidification rate (ECAR) and oxygen consumption rate (OCR)
   respectively, using the Seahorse extracellular flux analyzer. Glucose
   was injected, followed by three sequential additions of either the
   vehicle (left panel) or itaconate (right panel). Each data point is the
   mean ± SEM; n = 4 biological samples (4 independent experiments) with
   at least 3 technical replicates. The asterisks denote statistical
   differences at each time point between left and right panels by one-way
   ANOVA. C ECAR; p values at 53, 66, 73, 79, 86 and 92 min for WT LAC ±
   fumarate = 0.0052, 0.0020, <0.0001, 0.0401, <0.0001, <0.0001, the ΔfumC
   mutant ± fumarate = 0.0063, 0.0115, <0.0001, not significant, 0.0002,
   <0.0001 and the complemented strain ± fumarate = 0.0012, 0.0007,
   <0.0001, 0.0004, <0.0001, <0.0001 respectively. D OCR; p values at 53,
   66, 73, 79, 86, and 92 min for WT LAC ± fumarate = 0.0086, 0.0046,
   0.0044, 0.0127, 0.0071, 0.0162 respectively and 0.0433 and 0.0346 for
   the ΔfumC mutant and complemented strain ± fumarate at 92 min. E
   Heatmap illustrating variations in gene expression in WT LAC cultured
   with (10 mM) and without fumarate, as determined by qRT-PCR. The
   differences in gene expression are relative to WT LAC without fumarate.
   Statistical analysis was conducted using the Mann–Whitney
   non-parametric U test for each gene, with significance denoted as
   *p < 0.05 (p = 0.0170 for lukS) and **p < 0.01 (p = 0.0081 and 0.0047
   for esxA and perR respectively). F Schematic diagram outlining the
   workflow for the chemoproteomic profiling of succinated S. aureus
   proteins by the fumarate analogue (FA) probe. Created in BioRender.
   Wong, T. (2025) [117]https://BioRender.com/rkq0y88. G Volcano plot of
   the quantified FA-captured sites that are competed for by fumarate in
   live S. aureus FPR3757. The threshold for the −log10(q value/FDR) is
   set at 1.3; n = 3 biological samples (3 independent experiments) (H)
   Gene ontology analysis for the succinated sites from the in situ
   profiling. I AcnA/CitB activity in WT LAC, the ΔfumC mutant and
   complemented strain. Each data point is the mean ± SEM.; n = 3
   biological samples (3 independent experiments) in triplicate.
   Statistical analysis was conducted by one-way ANOVA, p = 0.0002 when
   comparing the WT LAC or complemented strain to the ΔfumC mutant. J
   Carbon source assimilation of WT LAC, the ΔfumC mutant and complemented
   strain (PM1 Biolog; n = 3 biological replicates from 3 independent
   experiments). The color intensity in the heatmap corresponds to the
   absorbance of the bacterial strains (OD[590nm]) as a readout of
   bacterial respiration in the presence of the indicated carbon source,
   normalized to the absorbance (OD[590nm]) of WT LAC in the same carbon
   source. Asterisks denote statistically significant differences as
   determined by two-way ANOVA.

   The expression of genes associated with pathogenesis is often linked
   with bacterial metabolism^[118]18. We observed that expression of
   several such genes were altered in the presence of fumarate
   (Fig. [119]2A). The cap5 genes associated with capsule
   production^[120]19 were upregulated (Fig. [121]2A). The genes lukS and
   esxA, encoding a PVL toxin subunit and a Type Seven Secretion System
   (T7SS) effector respectively, were markedly upregulated upon fumarate
   exposure (Fig. [122]2A), as corroborated by qRT-PCR even at lower
   fumarate concentration (Fig. [123]2E). We also observed increased
   expression of perR and katA, genes important in orchestrating
   antioxidative defenses (Fig. [124]2E).

   The metabolic impact of fumarate in redirecting metabolic activity can
   be mediated by direct or indirect mechanisms. Fumarate can
   post-translationally modify both bacterial and mammalian proteins,
   through covalent binding or succination^[125]12,[126]20–[127]22. We
   performed global profiling of succinated proteins in live S. aureus
   using a biorthogonal fumarate analogue (FA) probe^[128]23
   (Fig. [129]2F). Among the proteins that were identified, many that were
   modified at specific cysteines are involved in antioxidative stress
   responses such as PerR (Fig. [130]2G, H, Supplementary Data [131]5). We
   were intrigued by targets involved in the biogenesis/repair of
   iron-sulfur (Fe-S) clusters, including Nfu and ScdA (Fig. [132]2G). We
   hypothesized that fumarate impairs Fe-S cluster assembly in S. aureus
   through succination, mirroring its effect in host cells^[133]12, and
   inhibits downstream Fe-dependent bacterial processes. A number of
   proteins require Fe-S prosthetic groups for function^[134]24, such as
   the TCA cycle enzyme, aconitase (AcnA/CitB). In S. aureus, defects in
   Fe-S synthesis result in decreased TCA cycle function and reduced
   survival within neutrophils^[135]25.

   We tested the inhibitory effect of fumarate on S. aureus AcnA using the
   ΔfumC mutant as a genetic model for fumarate accumulation. We observed
   a significant reduction in AcnA activity in the ΔfumC mutant compared
   with the parental and complemented strains (Fig. [136]2I), suggesting
   that fumarate accumulation may inhibit staphylococcal Fe-S assembly.
   Metabolite phenotype microarrays (Biolog) also highlighted the
   inability of the ΔfumC mutant to utilize numerous carbon sources,
   including TCA cycle intermediates and sugars (Fig. [137]2J). Thus, the
   accumulation of excess fumarate has multiple potentially detrimental
   effects on the bacteria, disrupting the generation of ATP through the
   TCA cycle, OXPHOS, as well as glycolysis. While these studies indicate
   the importance of the Fe-independent enzyme FumC^[138]26 in regulating
   the accumulation of fumarate, they also imply that it is involved in
   many bacterial biosynthetic pathways.

Deletion of fumC impairs major physiological processes in S. aureus

   We next addressed the importance of fumC in S. aureus physiology,
   particularly as relevant to pneumonia. We noted that under conditions
   that mimic the infected airway, the ΔfumC mutant had a significant
   growth defect. In artificial sputum media (ASM), which contains
   fumarate (Supplementary Fig [139]1D), mucin, DNA, and amino
   acids^[140]27, there was a significant growth attenuation of the ΔfumC
   mutant compared to the WT and complemented strains (Fig. [141]3A). This
   was also observed in Luria broth (LB) rich in peptides but lacking in
   glucose (Fig. [142]3B left panel). While the WT strain and the ΔfumC
   mutant could both grow equally well in the presence of glucose in
   chemically defined media (CDM) (Fig. [143]3C left panel), the ΔfumC
   mutant displayed significant growth impairment in the absence of
   glucose (Fig. [144]3C right panel). This growth defect was exacerbated
   by the addition of fumarate and restored upon supplementation with
   malate in either LB, ASM or CDM (Fig. [145]3B middle and right panels,
   Supplementary Fig [146]1E, F).

Fig. 3. FumC promotes bacterial metabolic fitness, antioxidative defense and
pathogenesis.

   [147]Fig. 3
   [148]Open in a new tab

   Growth curve of WT LAC, the ΔfumC mutant and complemented strain in (A)
   artificial sputum media (ASM) (n = 3 biological samples in triplicate
   for each strain), B Luria Bertani (LB) broth lacking glucose (left
   panel, n = 7 in triplicate for WT LAC and n = 6 in triplicate for the
   ΔfumC mutant and complemented strain) and supplemented with fumarate
   (middle panel, n = 5 in triplicate for each strain) or malate (right
   panel, n = 5 in triplicate for each strain) at a final concentration of
   10 mM or C chemically defined media (CDM) supplemented with (left
   panel, n = 3 in duplicate) and without (right panel, n = 3, 1 technical
   replicate for each biological sample) glucose. For (A–C), data are
   shown as mean ± SEM. Statistical analyses were conducted using the
   two-tailed t-student test with FDR correction, *p < 0.05,
   ****p < 0.0001 for (A, C) and two-way ANOVA for B. D Growth curve of WT
   LAC, the ΔfumC mutant and complemented strain in LB with and without
   hydrogen peroxide (H[2]O[2]) at a final concentration of 0.5 mM. Data
   are shown as mean ± SEM from n = 3 biological samples in triplicate for
   each strain in LB only and n = 3 biological samples in 2 technical
   replicates for each strain in LB + 0.5 mM H[2]O[2]. Statistical
   significance is determined by two-tailed t-student test with FDR
   correction. E Heatmap showing variations in gene expression in WT LAC,
   the ΔfumC mutant and complemented strain, as determined by qRT-PCR. The
   differences in gene expression are relative to WT LAC. Statistical
   analyses were conducted using the two-tailed t-student test with FDR
   correction for each gene, with significance denoted as **p < 0.01
   (p = 0.0038 and 0.0067 for lukS and icaB respectively, ***p < 0.001
   (p = 0.0006 for perR) and ****p < 0.0001 for hla, esxA, trx, aphF,
   msrA1, msrB, and capA between WT LAC and the ΔfumC mutant. F Expression
   of genes encoding the T7SS machinery by WT LAC (left panel) and the
   ΔfumC mutant (right panel) in the presence of increasing fumarate
   concentrations. The differences in gene expression are relative to WT
   LAC or the ΔfumC mutant grown in the absence of fumarate. Data for are
   shown as mean ± SEM from n = 9 (WT LAC + 0 mM fumarate, 9 biological
   samples in 2 technical replicates), n = 7 (WT LAC + 25 mM fumarate, 7
   biological samples in 2 technical replicates), n = 10 (WT
   LAC + 50/100 mM fumarate, 10 biological samples in 2 technical
   replicates), n = 3 (ΔfumC mutant + 0/50 mM fumarate, 3 biological
   samples in 3 technical replicates), and n = 3 (ΔfumC mutant + 25/100 mM
   fumarate, 3 biological samples in 2 technical replicates). Statistical
   analyses were conducted using two-way ANOVA with a multiple posteriori
   comparison, p = 0.025 (for essA, difference between LB and LB + 100 mM
   fumarate) and 0.0040 (for essA, difference between LB + 25 mM fumarate
   and LB + 100 mM fumarate).

   In addition to effects on growth, the absence of fumC rendered the
   bacteria more susceptible to oxidative stress (Fig. [149]3D),
   consistent with the downregulation of the antioxidative genes perR,
   trx, aphF, and msrA1/B, but not katA and sod that encode catalase and
   superoxide dismutase respectively (Fig. [150]3E). Similarly, the
   expression of biofilm-associated genes and those encoding α-hemolysin
   (Hla), LukS (PVL), and the T7SS EsxA effector were significantly
   downregulated in the ΔfumC mutant (Fig. [151]3E). As expected, the
   phenotypes associated with the ΔfumC mutant contrasted with those
   observed when S. aureus can degrade excess fumarate (Fig. [152]2E). Of
   note, the expression of genes encoding the T7SS machinery itself (essA,
   esaA, and essC) was induced by exogenous fumarate in a fumC-dependent
   manner (Fig. [153]3F). Our findings indicate that S. aureus, when
   exposed to fumarate, depends on fumC to express genes likely important
   in pathogenesis and in maintaining biosynthetic pathways crucial to
   preserve its physiology.

FumC orchestrates carbon flux to diverse pathways

   We elucidated the precise role of S. aureus FumC in the biosynthetic
   pathways critical for bacterial growth and fitness (Fig. [154]4A),
   properties essential for pathogens. We employed ^13C[4]-labeled
   fumarate and stable isotope tracing in WT LAC and the ΔfumC mutant to
   track the metabolic fate of fumarate. Of the total fumarate, 49.5% and
   59.4% were ^13C-labeled in WT LAC and the ΔfumC mutant respectively,
   particularly as the carbon 4 (C[4]) isotopologue (Fig. [155]4B). This
   translated into the incorporation of ^13C into malate as C[1]-C[4]
   isotopologues in WT LAC but not in the ΔfumC mutant (Fig. [156]4B). The
   accumulated ^13C was instead incorporated back into succinate in the
   ΔfumC strain as C[3]-C[4] isotopologues (Fig. [157]4B). We observed the
   assimilation of ^13C derived from labeled fumarate into the TCA cycle
   metabolites citrate, cis-aconitate and α-ketoglutarate (α-KG) in WT LAC
   but not the ΔfumC mutant, as predicted (Fig. [158]4B).

Fig. 4. FumC serves as a metabolic hub and directs fumarate to major
metabolic pathways.

   [159]Fig. 4
   [160]Open in a new tab

   A Schematic diagram depicting key metabolic pathways in S. aureus.
   ^13C[4]-fumarate labeling of metabolites involved in the B TCA cycle or
   C aspartate-argininosccuinate shunt, D gluconeogenesis, E hexosamine
   synthesis and H PPP in WT LAC and the ΔfumC mutant. Adjusted p-values
   (corrected for multiple comparisons using the Benjamini, Krieger, and
   Yekutieli method) between WT LAC and the ΔfumC mutant are shown for the
   unlabeled compound (C0) and for labeled carbon isotopologues that
   account for at least 1% of the total metabolite pool in both groups.
   Adjusted p-values are as follows: for fumarate C0, C3, and C4 − 0.0012,
   0.0072, and <0.0001 respectively; for malate C0 and C4 − 0.000002; for
   succinate C0, C3 and C4 < 0.000001, 0.000015, and 0.000002; for citrate
   C0 − 0.000043; for cis-aconitate C0 − 0.000002; for α-ketoglutarate C0
   − 0.00000.7; for argininosuccinate C0 and C4 − 0.000039 and 0.000002;
   for PEP, F6P, GlcNAc-1-P and UDP-GlcNAc C0 < 0.000001 and for R5P
   C0 = 0.000024. F Biofilm formation by WT LAC, the ΔfumC mutant, and the
   complemented strain in LB broth (without glucose) with increasing
   concentrations of fumarate, as evaluated by crystal violet staining.
   The p-values for WT LAC at 0 mM versus 25 mM and 50 mM fumarate are
   0.0245 and <0.0001, respectively. For the complemented strain, the
   p-value for 0 mM versus 50 mM fumarate is 0.0004. G Biofilm depth of WT
   LAC in the presence of increasing fumarate concentrations, measured
   using wheat germ agglutinin-Alexa Fluor 555 staining and confocal
   microscopy. For (B–H), data are shown as mean ± SEM from n  =  3
   biological replicates (1 independent experiment). Statistical
   significance is determined by two-tailed t-student test with FDR
   correction. I Diagram summarizing the anabolic pathways fueled by S.
   aureus FumC.

   We detected the substantial integration of labeled carbon from fumarate
   into argininosuccinate, particularly as the C[4] isotopologue
   (Fig. [161]4C). This may serve to rapidly replenish malate and
   oxaloacetate for gluconeogenesis, hexosamine synthesis and the pentose
   phosphate pathway (PPP) in the WT strain (Fig. [162]4A). Accordingly,
   there was increased integration of ^13C into the gluconeogenic
   intermediates, phosphoenol pyruvate (PEP) and fructose 6-phosphate
   (F6P), in WT LAC but not in the ΔfumC mutant (Fig. [163]4D). We also
   detected significant incorporation of ^13C into
   N-acetyl-glucosamine-1-phosphate (GlcNAc-1-P) and
   UDP-N-acetyl-glucosamine (UDP-GlcNAc) (Fig. [164]4E), which contribute
   to biofilm formation^[165]9,[166]28 in the WT strain. We confirmed that
   fumarate promoted biofilm formation in a dose-dependent manner in WT
   LAC (Fig. [167]4F), resulting in increased biofilm depth
   (Fig. [168]4G).

   In addition, we noted the incorporation of labeled carbon into the PPP
   intermediate, ribose 5-phosphate (R5P) in WT LAC (Fig. [169]4H). This
   was not observed in the ΔfumC mutant. Assimilation of ^13C into R5P
   correlated with the increased labeling of intermediates in the de novo
   purine nucleotide metabolism, such as AICAR, AMP, ADP, ATP, GMP, and
   GDP (Supplementary Fig [170]1G). In parallel, there was increased
   labeling of dihydroorate, orotate, and intermediates of the de novo
   pyrimidine nucleotide synthesis, including UMP, UDP, UTP and CTP, in WT
   LAC only (Supplementary Fig [171]1H). Consistently, the incorporation
   of ^13C into NAD and the reduced form NADH, which is crucial for energy
   production, occurs in the WT strain only (Supplementary Fig [172]1I).
   Overall, the stable isotope tracing experiment emphasized an
   unexpectedly central role of FumC as a metabolic hub, directing
   fumarate utilization into multiple biosynthetic pathways critical for
   S. aureus survival and proliferation (Fig. [173]4I).

FumC is critical in pneumonia

   We expected S. aureus FumC to be crucial in chronic lung infection. We
   used a mouse pneumonia model with elevated fumarate levels to more
   adequately mimic the airway milieu in persistent infections
   (Fig. [174]1D). Ptenl^−/− mice, which were previously used to model the
   airway properties in CF^[175]29–[176]31, exhibited a baseline
   accumulation of the pro-inflammatory metabolite fumarate in the lung
   (~5 times higher than in BL/6 mice) but not succinate (Fig. [177]5A).
   As predicted, in settings with high fumarate levels (Ptenl^−/−), we
   observed that the ΔfumC mutant was significantly attenuated compared to
   the WT strain (Fig. [178]5B). This attenuation was more modest in the
   bronchoalveolar lavage fluid (BALF) and lung of BL/6 mice (3-fold and
   5-fold respectively) (Fig. [179]5B). Deletion of sucD, encoding the TCA
   cycle enzyme succinyl-CoA synthetase, had no effect on airway S. aureus
   load, while the ∆sucA mutant, lacking α-KG dehydrogenase, was
   attenuated in the lung but not in the BALF (Supplementary Fig [180]1J).
   This suggests that not all TCA cycle disruptions negatively impact
   staphylococcal airway survival and highlights the importance of
   fumarate metabolism in pulmonary pathogenesis.

Fig. 5. Host fumarate hydratase and S. aureus FumC both regulate fumarate and
pathogenesis.

   Fig. 5
   [181]Open in a new tab

   A Relative level of fumarate (left panel) and succinate (right panel)
   in the BALF of uninfected BL/6 and Ptenl^−/− mice. Fumarate levels
   differed between BL/6 and Ptenl^−/− mice, *p = 0.0190. B Bacterial
   burden from the BALF (left panel) and lung (right panel) of BL/6 and
   Ptenl^−/− mice infected with WT LAC, the ΔfumC mutant and complemented
   strain; BL/6 infected with WT LAC (n = 11), ΔfumC mutant (n = 13), and
   ΔfumC::fumC (n = 7), Ptenl^−/− infected with WT LAC (n = 6), ΔfumC
   mutant (n = 5). The above-mentioned total number of mice per group were
   from at least 2 independent experiments. The dotted line indicates the
   limit of detection. The bacterial load of WT LAC and the ΔfumC mutant
   differed significantly in the BALF and lung of BL/6 mice (p = 0.0009
   and 0.0338, respectively) and Ptenl^-/- mice (p = 0.0152 and 0.0173,
   respectively). Similarly, the burden of the ΔfumC mutant and
   complemented strain differed significantly in the BALF and lung of BL/6
   mice (p = 0.0009 and <0.0001, respectively). WT LAC burden also
   differed between BL/6 and Ptenl^−/− mice in the BALF (p = 0.0127) and
   lung (p = 0.0046). Innate immune cells (monocytes left panel,
   neutrophils middle panel and alveolar macrophages right panel) from the
   C BALF (*p = 0.0357) and D lungs of uninfected and infected mice from
   (A). E Cytokine measurements from the BALF of uninfected and infected
   mice from (A). Statistically significant differences were observed
   between groups in IL-1β (*p = 0.0472, **p = 0.0087), MIP2
   (**p = 0.0087), IL-6 (*p = 0.0062), KC (*p = 0.0140, **p = 0.0016),
   M-CSF (*p = 0.0140), MIG (*p = 0.0062) and IP-10 (*p = 0.0350). F
   Expression of murine Fh1, Ass1 and Hk2 from uninfected and infected
   mouse lungs (n = 5 per condition) by qRT-PCR (left panel) and schematic
   diagram showing the metabolic reactions catalyzed by the encoded
   enzymes fumarate hydratase, argininosuccinate synthase and hexokinase
   (right panel). A statistically significant difference in Hk2 expression
   was observed in the lungs of uninfected (PBS) and ΔfumC-infected mice
   (p = 0.0480). G Absolute level of malate in the BALF of uninfected
   (PBS) BL/6 mice and mice infected with WT LAC or the ΔfumC mutant. H
   Schematic diagram of reaction catalyzed by FumC. I FumC activity
   assessed by the absorbance of fumarate (left panel). Fumarate was
   supplied as a substrate for FumC, preincubated with and without 1 mM
   itaconate (chemical structure, right panel). For (A–G, I), data
   represent mean values ± SEM and statistical analyses were performed by
   Mann–Whitney non-parametric U test (two-tailed), *p < 0.05, **p < 0.01,
   ***p < 0.001 and ****p < 0.0001.

   To determine if differences in the immune response due to the loss of
   Ptenl might account for the enhanced ΔfumC clearance, we compared the
   immune cells (neutrophils, monocytes, alveolar macrophages) and
   cytokines during infection of both BL/6 and Ptenl^−/− mice. Overall,
   similar numbers of immune cells were observed in the infected BL/6 and
   Ptenl^−/− airway (Fig. [182]5C, D). In the Ptenl^−/− mice, where the
   difference in infection levels between WT LAC and the ΔfumC mutant was
   more pronounced (Fig. [183]5B), we observed a slight reduction in
   monocyte numbers in the BALF but not in the lung (Fig. [184]5C, D).
   While these mice also had lower levels of IL-1β and MIP2, there were no
   statistical differences in other proinflammatory cytokines such as IL6,
   KC and TNFα (Fig. [185]5E). In the BL/6 mice, similar numbers of immune
   cells and cytokine levels were also detected in response to infection
   with either the WT or ΔfumC strains (Fig. [186]5C–E). We confirmed that
   there were no differences in the uptake of WT LAC and the ΔfumC mutant
   by phagocytes (Supplementary Fig [187]1K). Thus, our results indicate
   that the attenuation of the ΔfumC mutant is not due to a lack of
   immunostimulation or to differences in the immune response between BL/6
   and Ptenl^−/− backgrounds. Instead, our findings underscore the
   critical role of S. aureus fumarate metabolism during pneumonia.

S. aureus exploits host metabolism

   The apparent disparity observed between the profound in vitro effects
   of fumC on S. aureus metabolism and the modest attenuation of the ΔfumC
   mutant in BL/6 mice led us to speculate that the bacteria may exploit
   the host FH to mitigate the absence of fumC in vivo. We observed
   comparable Fh1 transcript levels in both uninfected and infected murine
   lungs (Fig. [188]5F). This indicates that S. aureus sustains host Fh1
   expression, in contrast to Gram-negative pathogens and LPS, which
   typically downregulate it^[189]11,[190]32. The expression of Ass1,
   which encodes argininosuccinate synthase 1, was also unchanged, unlike
   Hk2 that was significantly upregulated upon infection with the ΔfumC
   mutant (Fig. [191]5F). We next measured the level of malate, the
   product of either S. aureus FumC or host FH, in the BALF harvested from
   BL/6 mice infected with WT LAC or the ΔfumC mutant. Similar levels of
   airway malate were detected during infection with either strain
   (Fig. [192]5G, H) despite the inability of the ΔfumC mutant to
   catabolize fumarate to malate. This suggests that host FH activity may
   protect S. aureus from the detrimental effects of fumarate in vivo,
   thereby limiting the impact of fumC deletion through the generation of
   malate. Our results emphasize that while S. aureus can exploit the host
   FH to degrade fumarate, it depends significantly on its own FumC enzyme
   in environments with accumulated fumarate levels typical of chronic
   lung infections.

   Itaconate, another dicarboxylate, accumulates in the chronically
   infected airway, along with fumarate (Fig. [193]1D). These two
   metabolites share structural similarities (Fig. [194]5I) and could
   potentially work in a concerted fashion. We wondered if the presence of
   itaconate might further enhance FumC activity. We monitored the
   conversion of fumarate to malate by purified FumC^[195]33 in the
   presence or absence of itaconate (Fig. [196]5H, I). The significant
   depletion of added fumarate in the presence of itaconate strongly
   suggests that itaconate potentiates FumC activity. Our data demonstrate
   that the metabolism of airway fumarate by S. aureus promotes pneumonia
   and that this effect is even more biologically significant in the
   context of other accumulated metabolites such as itaconate.

Discussion

   S. aureus is a well-studied pathogen which nonetheless remains among
   the top three causes of healthcare-associated infection, especially
   pneumonia^[197]1. Its extensive metabolic flexibility has long been
   recognized but not effectively targeted in either antibiotic or vaccine
   development. We uncovered a previously unappreciated role for fumarate
   metabolism in the overall physiology and pathogenesis of S. aureus,
   especially in microenvironments with limited glucose such as the
   airway. Our focus on FumC was initiated by studying clinical isolates
   that had successfully undergone long term adaptation to the lung. The
   increased expression^[198]10 and low mutation rate of fumC in S. aureus
   pneumonia isolates indicate its global epidemiological significance for
   pulmonary adaptation. This is particularly relevant in the chronically
   infected airway replete with the host immunometabolites, fumarate and
   itaconate.

   We observed a fascinating dynamic regulation of airway fumarate, an
   important pro-inflammatory mediator^[199]11, by both host and S. aureus
   fumarate hydratase. For the host, the production of the dicarboxylates
   fumarate and itaconate offers homeostatic protection against the
   detrimental effects of excessive inflammation. Fumarate, a product of
   activated macrophages, induces type I interferon and TNFα production in
   response to mitochondrial damage^[200]11. In contrast, itaconate
   exhibits immunomodulatory properties; it stimulates major antioxidative
   mediators such as Nrf2 and ATF3^[201]34 and inhibits NLRP3 inflammasome
   activation^[202]35. Just as excess fumarate is detrimental to the host,
   we show that it is toxic to the bacteria, inhibiting primary metabolic
   pathways such as glycolysis, the TCA cycle and OXPHOS. The presence of
   fumC in S. aureus isolates not only enables fumarate degradation,
   circumventing its negative effects, but also promotes its utilization
   in anabolic pathways such as gluconeogenesis, and hexosamine
   production, which are essential for biofilm formation and defense
   against oxidative stress. We also found that several genes expected to
   function more directly in pathogenesis were upregulated by fumarate
   such as those involved in toxin and capsule production and the T7SS.

   Having documented the pervasive effects of fumarate on S. aureus
   biology and the critical role of FumC in numerous metabolic pathways,
   we also demonstrate its role in vivo, particularly in environments
   abundant in fumarate, such as the Ptenl^−/− airway. Consistently, we
   show that the ΔfumC mutant had a more modest limitation during acute
   pulmonary infection in BL/6 mice. No substantive differences in the
   immune response to the WT or ΔfumC strains were identified. These
   observations can be explained by compensatory host FH activity, as
   confirmed by quantifying equivalent levels of malate in the airway of
   BL/6 mice. Thus, the host can regulate local fumarate levels to a
   certain extent, even in the absence of S. aureus FumC, enabling the
   microorganisms to proliferate. However, in the Ptenl^−/− background
   with excessive airway fumarate, the lack of fumC resulted in an
   appreciable loss of bacterial fitness in vivo. This further supports
   the critical role of FumC in chronic pulmonary infections.

   In addition, we observed that the host metabolite itaconate enhances
   the activity of S. aureus FumC and further promotes the degradation of
   fumarate and its utilization as a carbon source. Thus, in vivo, ambient
   fumarate amounts are regulated by both the host FH and S. aureus FumC
   whose activity seems likely to be boosted by itaconate. It remains to
   be determined exactly how itaconate accelerates FumC and whether this
   is unique to S. aureus. Given the overall structural similarities
   between the host and S. aureus FH, FumC may initially appear to be an
   unsuitable target for anti-staphylococcal therapy. However, some of the
   peculiarities of S. aureus FumC structure and charge characteristics
   may be sufficiently distinct to generate selective bacterial toxicity.
   These distinctive features could be harnessed for the development of
   adjuvant therapy to prevent chronic pneumonia.

   We highlight two limitations to provide appropriate context for
   interpreting our findings. Although fumarate is a central focus of this
   work, its physiological concentrations in the human airway have not yet
   been determined either in healthy individuals or patients with chronic
   pulmonary infection. We tested a wide range of fumarate concentrations
   across multiple media types, including LB, glucose-depleted basal
   medium, and artificial sputum medium to reflect the likely range of in
   vivo conditions and to account for the influence of staphylococcal
   carbon catabolite repression by preferred substrates. Importantly, S.
   aureus demonstrated FumC-dependent adaptation across this broad range
   of fumarate concentrations, including levels likely exceeding
   physiological norms; this adaptation was impaired in the absence of
   fumC, where fumarate instead exerted toxic effects. Another challenge
   consists of the variability in expectorated sputum from individuals
   with CF and induced sputum from healthy controls, an issue inherent to
   the many studies using clinical materials. These factors may alter
   sample composition, in part due to dilution effects. We include these
   limitations here to ensure transparency and support the interpretation
   of our results within the experimental conditions used.

Methods

   All experimental procedures were conducted in accordance with relevant
   ethical regulations and approved by the appropriate institutional
   review boards or ethics committees.

Human subjects

   All human samples (healthy individuals and CF patients) were obtained
   from adults (22–44 years of age) in the CF program at Yale University
   under the Yale IRB Protocol 0102012268 and Columbia Protocols AAAR1395.
   Males and females are represented in an ~50–50% ratio. Sputum samples
   from healthy adult individuals were collected after nebulization with
   3% hypertonic saline for five minutes on three cycles. To reduce
   squamous cell contamination, subjects were asked to rinse their mouth
   with water and clear their throat. CF subjects expectorated sputum
   spontaneously for our studies. Expectorated sputum samples were
   collected into specimen cups and placed on ice. Of the 7 CF patients
   that were chronically infected with Staphylococcus aureus (either MSSA
   or MRSA), 5 patients were co-infected with other agents, such as
   Pseudomonas (3 patients), Achromobacter (1 patient) or human influenza
   (1 patient). No CF patients were infected with Aspergillus, a known
   itaconate producer. None of the CF patients studied were under CFTR
   modulator therapy (e.g., Kaleydeco, Orkambi). An informed consent was
   signed by all subjects.

Mouse experiments

   All animal experiments were performed following institutional
   guidelines at Columbia University and Rutgers New Jersey Medical School
   (NJMS). Animals were housed and maintained at Columbia University
   Irving Medical Center (CUIMC) and Rutgers Medical Science Building
   (MSB) under regular rodent light/dark cycles at 18–23 °C and fed with
   irradiated regular chow diet (Purina Cat#5053, distributed by Fisher).
   Animal health was routinely checked by an institutional veterinary. The
   animal work protocols (AC-AABD5602 and PROTO202400003) were approved by
   the Institutional Animal Care and Use Committee (IACUC). Animal
   experiments were carried out in strict accordance with the
   recommendations in the Guide for the Care and Use of Laboratory Animals
   of the NIH, the Animal Welfare Act, and US federal law. C57BL/6 mice,
   purchased from the Jackson Laboratory (stock number 000664), were
   acclimated in our facilities. Ptenl^−/− mice^[203]29 were bred in our
   facilities. The animals were infected at 8 to 10 weeks of age. In vivo
   experiments were performed using roughly 50% female and 50% male
   animals. During studies, animals were randomly assigned to cages.

In silico fumC conservation

   We assembled a collection of serially sequenced S. aureus strains from
   seven published within-host evolution studies of S. aureus during
   cystic fibrosis (CF)^[204]36–[205]42. The studies were added if: they
   included at least two S. aureus strains per patient; sequencing reads
   were publicly available; minimal metadata were reported: patient
   identifier, specimen source, date of collection. We supplemented the
   published studies with unpublished data from a CF study at the
   Children’s Hospital of Philadelphia. Raw reads and metadata were
   downloaded from NCBI using the Bioproject identifier provided in the
   publications (Supplementary Data [206]2). Metadata were additionally
   retrieved from the publications. Quality control for reads data was
   performed as described in ref. ^[207]43. Reads were excluded if the
   coverage was below 20, reads classification using Kraken2,
   v2.1.1^[208]44 did not yield S. aureus or <50% reads were classified as
   S. aureus, or if the assembly size exceeded 3.5 million bp. Reads were
   assembled using Shovill, v.1.1.0, which implements the Spades
   assembler^[209]45. Multi-locus sequence type and resistance genes were
   inferred from the assemblies using Mlst, v2.19.0 and Abricate, v1.0.1,
   both available at [210]https://github.com/tseemann.

   Within-host evolution analysis was carried out as described in ref.
   ^[211]46: for each patient, we used Snippy, v4.6.0
   ([212]https://github.com/tseemann/snippy) to map isolate reads to the
   draft assembly of the earliest strain (if multiple strains were
   available the reference was selected randomly). The output was filtered
   by removing variants from reference reads and variants at positions for
   which the reads coverage of the reference was below 10. The amino-acid
   sequences of the mutated genes across all episodes were clustered using
   CD-HIT, v4.8.1^[213]47 and for each cluster representative we searched
   for a S. aureus FPR3757^[214]48 homolog (NCBI accession [215]CP000255)
   using Blastp, v2.10.1. This generated a list of FPR3757 genes with
   counts of independent episodes with at least one mutation arising
   within the host. We then excluded synonymous substitutions and limited
   the list to genes involved in the tricarboxylic acid cycle (TCA) and
   surrounding pathways (glycolysis, gluconeogenesis, urea cycle) and
   calculated the relative mutation rate λ for each gene i as (mutations
   in gene i/length of gene i)/(mutations in all genes/length of all
   genes). We inferred the significance of the mutations enrichment by
   fitting two Poisson models: model 0 as the null hypothesis (neutral
   evolution), where the number of variants per gene was the product of
   the mean mutation rate across the genome and the gene length; model 1
   estimated the number of variants per gene base on a gene-specific
   mutation rate while variants in the remaining genes are the product of
   a mean mutation rate. Model 0 and model 1 were compared using a
   likelihood ratio test. Scripts for the within-host evolution analysis
   are available at [216]https://github.com/stefanogg.

Bacterial strains

   The bacterial strains used in this study are shown in Supplementary
   Data [217]6. WT LAC and its derivative strains, ∆fumC and ∆fumC::fumC
   were constructed as previously described^[218]49. The transposon
   insertions sucA::Tn(ermB) and sucD::Tn(ermB) were sourced from the
   NARSA transposon mutant library^[219]50. The alleles were transduced
   into USA300_LAC using bacteriophage 80α^[220]51. The mutant strains
   were confirmed via PCR using sucAB and sucCD primers. All strains were
   grown at 37 °C on plates of Luria-Bertani lacking glucose (LB, Becton
   Dickinson (BD), 244610) supplemented with 1% agar (w/v, Acros Organics,
   400400010). Overnight cultures and subcultures (1/100) were grown in LB
   broth at 37 °C with shaking. Bacterial inocula were estimated based on
   the optical density at 600 nm, OD[600nm], and verified by retrospective
   plating on LB agar plates to determine colony forming units (CFUs). S.
   aureus recovered from the mouse airways were selected on LB agar plates
   and BBL CHROMagar Staph aureus plates (BD).

Primers

   The primers used in this study are listed in Supplementary Data [221]7.
   Full-length fumC was amplified from the genome of WT USA300 FPR3757
   using KOD One PCR Master Mix -Blue- (TOYOBO, KMM-201) and cloned into
   the pET-21b(+) vector by Gibson assembly.

Growth curves

   A U-bottomed, clear 96-well plate (Greiner Bio-One #650161) was
   prepared with LB, chemically defined media (CDM)^[222]52 or artificial
   sputum media^[223]27. For the experiments that used media supplemented
   with itaconic acid (Sigma, [224]I29204), L-malic acid (Sigma, M1000),
   sodium fumarate (Sigma, F1506), or hydrogen peroxide, H[2]O[2], (Sigma,
   H1009), the pH of the media was corrected to 7.0 prior to filter
   sterilization using 0.20 μm filters (Thermo Fisher
   Scientific,725-2520). Each well was inoculated with 1.5 × 10^5
   bacteria. Absorbance at 600 nm was read every 10 min for 36 h on the
   Varioskan™ LUX multimode microplate (Thermo Fisher Scientific,
   VL0000D0), as the plate was maintained at 37 °C with shaking.

Measurement of biofilms

   Log phase cultures were diluted to OD[600nm] of 0.1 in LB broth
   supplemented with or without fumarate (Sigma, F1506) prior to
   incubation in either 96-well plates (Greiner Bio-One, 650161,
   100 µl/well) or 4-well chamber slides (Sarstedt, D-51588, 750 µl/well)
   for 24 h and 12 h respectively. The plates were processed as follows:

   1. Crystal violet staining in 96 well-plates: The OD[600nm] was
   measured prior to crystal violet staining. The plates were fixed with
   100% methanol, stained with 1% crystal violet (w/v, Sigma, C6158),
   washed and dried. Subsequently, 33% acetic acid (v/v, Acros Organics,
   222140010) was added to solubilize the dye. The OD[540nm] was measured
   and normalized to the bacterial growth at OD[600nm].

   2. Wheat germ agglutinin-Alexa Fluor 555 staining and confocal
   microscopy: Wheat germ agglutinin (WGA) coupled to Alexa Fluor 555
   (5 µg/ml, Thermofisher, [225]W32464) was added into each well and the
   chamber slides were covered with foil. Each chamber slide was placed on
   a shaker (horizontal orbit, 60 rpm) in a 37 °C room for 5 min, before
   being kept stationary in a 37 °C room for 12 h. Confocal imaging of
   viable biofilms was performed using a Zeiss LSM 980 Airyscan with a
   37 °C cage incubator. 16-bit images of WGA-positive cells (1.5x crop
   area, 1024 × 1024 px frame size) were captured using a 40X
   oil-immersion objective with the 488 nm laser (670 V gain) set at 1%
   intensity. Z-stack slices were collected at 1 µm intervals. The z-depth
   of high-density biofilm layers were calculated as the number of slices
   (1 µm apart) where extracellular spaces occupied no more than 20% of
   the image. ImageJ 1.54 f. (National Institutes of Health, Bethesda, MD,
   USA) was used to calculate the % of extracellular space in an image
   using thresholding.

Isolation of RNA from bacterial cultures

   Bacterial cultures grown in LB with and without 10 mM fumarate were
   standardized to an OD[600nm] of 1.0 prior to centrifugation. Bacterial
   pellets were incubated in lysis buffer (50 μM Tris-EDTA pH 7.5,
   8 U ml^−1 mutanolysin (Sigma-Aldrich), 0.018 mg ml^−1 lysostaphin
   (Sigma-Aldrich), 0.05 mg ml^−1 lysozyme (Sigma-Aldrich)) at 37 °C for
   45 min and TRK lysis buffer (Omega Bio-tek) was added. After 10 min at
   room temperature, 70% ethanol was added, and samples were transferred
   to E.Z.N.A. RNA isolation columns (Omega Bio-tek). RNA was isolated
   following the manufacturer’s instructions and treated with DNase using
   the DNA-free DNA removal kit (Invitrogen).

Complementary DNA synthesis and qRT-PCR

   Multiscribe reverse transcriptase (Applied Biosystems) was used to
   generate cDNA for qRT–PCRs with Power SYBR Green PCR Mastermix (Applied
   Biosystems). qRT-PCR was performed using primers for hla, lukS, esxA,
   per, katA, sod, trx, aphF, msrA1, msrB, capA, icaB and 16S
   (Supplementary Data [226]7) and a StepOne Plus thermal cycler (Applied
   Biosystems). The data were analyzed using the ΔΔC[T] method.

RNA-sequencing

   WT LAC was grown in LB with or without 100 mM fumarate to late
   exponential phase. Bacterial pellets were incubated in a cell wall
   lysis mixture (described above) at 37 °C for 45 min. TRK lysis buffer
   (Omega Bio-tek, R6834-02) and 70% ethanol were added, and samples were
   transferred to E.Z.N.A. RNA isolation columns (Omega Bio-tek,
   R6834-02). RNA was isolated following the manufacturer’s instructions
   and treated with DNase using the DNAfree DNA removal kit (Fisher
   Scientific, AM1906). The RNA was precipitated with 0.1 volume 3 M
   sodium acetate (Thermo Fisher, S209) and 3 volumes of 100% ethanol,
   recovered by centrifugation, and washed with ice cold 70% ethanol. A
   ribosomal RNA-depleted cDNA library was prepared according to the
   manufacturer’s instructions using the Universal Prokaryotic RNA-Seq
   Prokaryotic AnyDeplete kit (NuGEN, 0363-32) and sequenced with Illumina
   HiSeq. Raw base calls were converted to fatsq files using Bcl2fastqs.
   Filtered reads were aligned to the LAC_FPR3757 reference genome using
   STAR-Aligner v2.7.3a. The mapped reads were annotated for read groups
   and marked for duplicates using the Picard tools v2.22.3. The raw
   counts were quantified using Subreads:FeatureCounts v1.6.3 and
   processed for differential gene expression using DEseq2 in R v3.5.3.

In silico protein structure simulation and charge prediction

   The predicted structure of S. aureus FumC was constructed using the
   homology modeling tool from the Schrödinger software package
   (Schrödinger Release 2022-4: Prime; Schrödinger, Inc: New York, NY,
   2022)^[227]53. We used the human fumarate hydratase (PDB: 5UPP) with
   57% identity as template. The structure of E. coli FumC, which shares a
   higher identity (60%), could have been utilized, but would have
   necessitated the de novo construction of longer loops; one loop
   consisting of 3 amino acids and another loop consisting of 8 amino
   acids instead of two loops of 3 amino acids and one loop of 1 amino
   acid. The obtained structure was replicated into four copies, which
   were superimposed on the four chains of the template tetramer structure
   and merged. Protonation states of the predicted tetramer were assigned
   using PROPKA and the structure subjected to a constrained minimization
   using the OPLS4 force field (Schrödinger Release 2022-4: Protein
   Preparation Wizard, Schrödinger, LLC, New York, NY, 2022)^[228]54.

Synthesis of FA-alkyne probe

   Monomethyl fumarate (1 g, 1 eq, 7.69 mmol) was dissolved in
   acetonitrile and cooled to 0 °C. EDCl·HCl (1.6 g, 1.1 eq, 8.46 mmol),
   HOBt (1.1 g, 1.1 eq, 8.46 mmol) and Et[3]N (1.39 mL, 1.3 eq, 2.29 mmol)
   were added and then stirred for 40 min at 0 °C. Subsequently,
   propargylamine (0.6 g, 1.3 eq, 10 mmol) was added. The solution was
   allowed to warm up to room temperature and stirred overnight. The
   solvent was removed in vacuo, the residue was redissolved in
   ethylacetate and washed with 1 M HCl, saturated sodium bicarbonate
   solution and brine. The organic layer was dried over anhydrous sodium
   sulfate, the resulting crude was dried, and the solvent was removed in
   vacuo. The resulting crude was purified by flash column chromatography
   (SiO[2], DCM:MeOH=40:1) to afford the pure compound 1 (820.8 mg, 63.9%)
   as white solid (Supplementary Fig. [229]2A). Lithium hydroxide
   (330.0 mg, 7.86 mmol) was added to a stirring solution of compound 1
   (820.8 mg, 4.91 mmol) in a 1:1 mixture of water and tetrahydrofuran
   (Supplementary Fig. [230]2B). The reaction mixture was stirred at room
   temperature and monitored by TLC. After 1 h, the reaction mixture was
   acidified with 1 M HCl and the solvent was removed in vacuo. The crude
   product was redissolved in ethylacetate and filtered. The solution was
   evaporated in vacuo and the crude product was purified by flash column
   chromatography (SiO[2], DCM:MeOH=20:1 to 10:1) to obtain the pure FAyne
   (FA-alkyne, 622.1 mg, 82.8%) as white solid. ^1H NMR (400 MHz,
   DMSO-d[6]) δ 8.96 (t, J = 5.5 Hz, 1H), 6.94 (d, J = 15.5 Hz, 1H), 6.58
   (d, J = 15.5 Hz, 1H), 4.01 (dd, J = 5.5, 2.6 Hz, 2H), 3.21 (t,
   J = 2.5 Hz, 1H) (Supplementary Fig. [231]2C). ^13C NMR (101 MHz,
   DMSO-d[6]) δ 166.28, 163.35, 135.86, 131.37, 81.77, 72.78, 28.64
   (Supplementary Fig. [232]2D).

Competitive ABPP workflow for S-succinated cysteines in S. aureus

   S. aureus USA300 FPR3757 was cultured overnight in tryptic soy broth
   (TSB), washed with PBS and concentrated to OD[600nm] of 40. For the
   competition group, 800 μl of bacterial suspension was incubated with
   80 μl of fumarate (100 mM stock was dissolved in NaOH and pH was
   adjusted to neutral) on the ThermoMixer (950 rpm, 30 min, 37 °C). For
   the control group, fumarate was replaced with the same volume of PBS.
   After competition, both groups were incubated with 8 μl of 1 M
   FA-alkyne probe on the ThermoMixer (950 rpm, 37 °C, 1 h). The bacterial
   suspension was then centrifuged (12,000 rpm, 5 min, 4 °C), the
   supernatant was decanted, and the pellets were washed once with 1 ml of
   pre-chilled PBS. Washed pellets were resuspended with 200 μl of 0.1%
   (v/v) Triton X-100 (Amresco, 0694-1 L) in PBS (hereafter, 0.1% PBST).
   Bacterial suspension was incubated on the ThermoMixer (1200 rpm,
   15 min, 37 °C) after the addition of 3 μl of lysostaphin (5 U/μl,
   Coolaber, CL6941-1mg). Then, the lysate was supplied with 6 μl of 10%
   SDS (Thermo Fisher Scientific, AM9822) and sonicated (35% intensity).
   The lysate was centrifuged (20,000 g, 10 min, RT) to remove the debris
   and the supernatant was transferred into a new 1.5 ml tube (30 μl of
   supernatant was also aliquoted for the gel-based ABPP assay). Click
   reaction was carried out with 106 μl of Click reagent mix (60 μl of
   50 mM TBTA ligand, 20 μl of 50 mM CuSO[4], 20 μl of freshly prepared
   50 mM TCEP and 6 μl of 20 mM acid-cleavable azide-biotin tag
   (Confluore, BBBD-19-25 mg) and incubated on the ThermoMixer (1200 rpm,
   1 h, 29 °C). The samples were then subjected to streptavidin enrichment
   and dual enzyme digestion (trypsin and Glu-C) steps before mass
   spectrometry analysis as previously described^[233]55.

   After the dual enzyme digestion, the peptides were further labeled with
   a dimethylation reagent (“light” for the control group (i.e., FA probe
   + vehicle) and “heavy” for the fumarate competition group (i.e., FA
   probe + fumarate))^[234]56. Streptavidin beads were washed three times
   with MS-grade water to thoroughly remove dimethylation reagents. The
   labeled peptides were released from streptavidin beads by incubating
   with 200 μl of 2% formic acid on the ThermoMixer (1200 rpm, 25 °C,
   90 min). Peptides from the control group and fumarate competition group
   were combined and dried in a SpeedVac (30 °C). They were then analyzed
   by HPLC-MS/MS and subjected to quantitative proteomic analysis. A
   competitive light-to-heavy ratio for each target was calculated. A
   ratio greater than 1 (or greater than 0 after log2 transformation)
   suggests that the cysteine site (or protein) undergoes succination by
   fumarate. The corresponding competition ratios for each cysteine site
   (or protein) were averaged across three biological replicates, and the
   q-value (false discovery rate/FDR) was calculated.

Protein purification

   Plasmid pET21b-FumC-His[6] was transformed into BL21(DE3). Protein
   expression was performed overnight at 16 °C in the presence of 0.2 mM
   IPTG (VWR, 0487-100 G). Bacterial pellets were resuspended in
   suspension buffer (50 mM Tris-HCl, pH 8.0 (Thermo Fisher Scientific,
   15568-025), 150 mM NaCl) and disrupted with sonication. After the
   lysates were clarified via centrifugation (12,000 rpm, 30 min, 4 °C),
   the supernatant was loaded onto 5 ml HisSep Ni-NTA 6FF column (YEASEN,
   20504ES25) and washed with W1 (50 mM Tris, pH 8.0, 150 mM NaCl, 20 mM
   imidazole, pH 7.4) and W2 (50 mM Tris, pH 8.0, 150 mM NaCl, 40 mM
   imidazole, pH 7.4). Protein was eluted with elution buffer (50 mM Tris,
   pH 8.0, 150 mM NaCl, 300 mM imidazole, pH 7.4) and the eluate was
   centrifuged (4000 g, 4 °C, swing-out) using Amicon® Ultra-15
   Centrifugal Filter Unit to remove imidazole and concentrate protein.
   The concentrated protein was mixed with 20% glycerol and subjected to
   BCA assay to measure concentration (Thermo Fisher Scientific, 23225).
   Finally, the protein was supplied with 1 mM DTT, aliquoted, flash
   frozen in liquid nitrogen, and stored at −80 °C.

FumC activity

   Recombinant FumC was diluted with Tris-HCl (20 mM, pH 8.0) to 1 mg/ml,
   and incubated with either 1 mM itaconate (pH was adjusted with NaOH to
   7.4) and 1 mM DTT or PBS and 1 mM DTT as the negative control (2 h,
   37 °C). For the enzymatic assay, the reaction was conducted in a 1.5 ml
   polystyrene cuvette (BIOFIL, CUV010015). The reaction was initiated by
   mixing 100 μg of post-incubated FumC and the reaction mixture
   containing 20 mM Tris-HCl (pH 8.0) and 15 mM fumarate (Sigma-Aldrich,
   47910) in a total volume of 1 ml at 25 °C for 30 min. The maximal
   absorbance of fumarate was determined using a spectrophotometer
   (IMPLEN, P330) in a 1.5 ml cuvette. Absorbance was read at 289 nm. This
   enzymatic assay was performed on three biological replicates.

AcnA/CitB activity

   AcnA assays were conducted as previously described with slight
   modifications^[235]57,[236]58. Strains were cultured overnight in TSB
   before diluting them in fresh TSB to an OD[600nm] of 0.1. The cultures
   were diluted in 0.5 ml or 4 ml of TSB in 10 ml culture tubes. The cells
   were cultured for 8 h, before they were harvested by centrifugation,
   and the cell pellets were stored at −80 °C. The cells were thawed
   anaerobically, resuspended with 200 μl of AcnA buffer (50 mM Tris,
   150 mM NaCl, pH 7.4), and lysed by the addition of 4 μg lysostaphin and
   8 μg DNase. The cells were incubated at 37 °C until full lysis was
   observed (~1 h). The cell debris was removed by centrifugation, and
   AcnA activity was assessed by monitoring the conversion of isocitrate
   to cis-aconitate spectrophotometrically.

Metabolite phenotype microarray (Biolog)

   For Carbon Source Phenotype Microarray™ (Biolog) assays, a stock
   solution of 2 × 10^7 bacteria/ml was prepared in 1X IF-Oa buffer
   (Biolog, 72268) supplemented with 1X Redox Dye Mix A (Biolog, 74221).
   100 μl of this stock solution (delivering 2 × 10^6 bacteria) was added
   to each well of a PM1 Microplate™ (Biolog, 12111) and the plate was
   incubated at 37 °C overnight with shaking. Absorbance was read at
   590 nm on the SpectraMax M2 plate reader (Molecular Devices).

Bacterial extracellular flux analyses

   The XFe24 sensor cartridge (Agilent #102340-100) was calibrated as per
   the manufacturer’s instructions overnight at 37 °C without CO[2].
   500 μl of XF base medium (Agilent #102353-100) was added to each well
   of a Seahorse XF24 well plate (Agilent #102340-100) and inoculated with
   3 × 10^7 bacteria for a 3 h incubation at 37 °C. The oxygen consumption
   rate (OCR, indicative of oxidative phosphorylation) and extracellular
   acidification rate (ECAR, indicative of glycolysis) were again measured
   on a Seahorse XFe24 Analyzer (Agilent Technologies) using Seahorse Wave
   Desktop v2.6.0. Glucose (Sigma #G7021) was added at a final
   concentration of 10 mM followed by the addition of fumarate (Sigma,
   F1506) to a final concentration of 10 mM.

^13C[4]-fumaric acid labeling and stable isotope tracing

   WT LAC and the ΔfumC mutant were grown overnight in LB, then inoculated
   (1/100) into fresh LB supplemented with ^13C[4]-fumaric acid (Sigma,
   606014, pH 7.0), grown at 37 °C to late exponential phase and
   standardized to an OD[600nm] of 14. For metabolite extraction, each
   culture was diluted with 3 volumes of PBS and centrifuged at 2000 × g
   for 10 min at 1 °C. The supernatant was discarded, and the pellet was
   washed with PBS. The pellet (30 μl in volume) was resuspended in a 3:1
   methanol:water extraction solution and lysed with 10 freeze-thaw cycles
   by alternating emersion in liquid nitrogen and a dry-ice/ethanol bath.
   The debris was removed by centrifugation at 14,000 × g for 5 min at
   1 °C and the supernatant was stored for analysis. Targeted LC/MS
   analysis was performed on a Q Exactive Orbitrap mass spectrometer
   (Thermo Fisher Scientific) coupled to a Vanquish UPLC system (Thermo
   Fisher Scientific). The Q Exactive operated in polarity-switching mode.
   A Sequant ZIC-HILIC column (2.1 mm i.d. × 150 mm, Merck) was used for
   separation of metabolites. Flow rate was set at 150 μl/min. Buffers
   consisted of 100% acetonitrile for mobile A, and 0.1% NH[4]OH/20 mM
   CH[3]COONH[4] in water for mobile B. Gradient ran from 85 to 30% A in
   20 min followed by a wash with 30% A and re-equilibration at 85% A.
   Metabolites were identified based on exact mass within 5 ppm and
   standard retention times. Relative quantitation was performed based on
   peak area for each isotopologue. All data analysis was done using MAVEN
   2011.6.17. The data in Fig. [237]4 are presented with correction for
   natural isotope abundance.

Isolation of murine BMDMs and infection

   Murine BMDMs were differentiated as described previously^[238]9. The
   femurs and [239]tibias were surgically removed from mice. The bone
   exterior was sterilized with 70% ethanol and the bone marrow was
   recovered by flushing with phosphate-buffered saline (PBS, Corning,
   20-031). The cell suspension was centrifuged for 6 min at 500 × g and
   resuspended in ACK lysis buffer (Thermo Fisher Scientific, A1049201) to
   remove the red blood cells. The lysis solution was quenched with PBS,
   and the cells were centrifuged again and resuspended in DMEM medium
   (Corning, 10-013-CV) containing 10% heat-inactivated fetal bovine serum
   (FBS v/v, Sigma, F4135) and 1% penicillin/streptomycin (P/S, v/v,
   Corning, 30-002-CI)) supplemented with 20 ng/ml rM-CSF (PeproTech
   #315-02). The rM-CSF-supplemented media was replenished 3 days after
   isolation, and the BMDMs were mechanically detached on day 7,
   centrifuged, resuspended in DMEM medium containing 10% FBS (Sigma), and
   counted using trypan blue stain (Invitrogen, [240]T10282). The cells
   were seeded at 1 × 10^6 cells/ml in a 24 well plate (Agilent,
   102340-100) and incubated at 37 °C with 5% CO[2] overnight. The cells
   were infected at a multiplicity of infection (MOI) of 10 for 30 min and
   10 μg ml^−1 lysostaphin (Sigma) was then added until the desired time
   point. The cells were washed, detached using TrypLE Express (Life
   Technologies), serially diluted, and plated on LB agar. Cell viability
   was determined by counting live cells using trypan blue (Life
   Technologies).

Mouse pneumonia model

   Mice were infected intranasally with 1 × 10^7 CFUs of S. aureus in
   50 μl PBS. PBS alone was used as a control. The mice were sacrificed at
   18 h post infection and the [241]BALF and lung were collected. The lung
   was homogenized through 40 μm cell strainers (Falcon, 352340). Aliquots
   of the BALF and lung homogenates were serially diluted and plated on LB
   agar plates and BBL CHROMagar Staph aureus plates (BD) to determine the
   [242]bacterial burden. The BALF and lung homogenates were spun down and
   the BALF supernatant was collected for cytokine and untargeted
   [243]metabolomic analysis. After hypotonic lysis of the red blood
   cells, the remaining BALF and lung cells were prepared for
   fluorescence-activated cell sorting (FACS) analysis as described below.

Flow cytometry of mouse BALF and lung cells

   For the identification of immune cell populations, mouse BALF and lung
   cells were stained in the presence of counting beads (15.45 μm
   DragonGreen, Bangs Laboratories Inc., FS07F) with LIVE/DEAD stain
   (Invitrogen, L23105A) and an antibody mixture of anti-CD45-AF700
   (BioLegend, 103127, clone 30-F11), anti-CD11b-AF594 (BioLegend, 101254,
   clone M1/70), anti-CD11c-Bv605 (BioLegend, 117334, clone N418),
   anti-SiglecF-AF647 (BD, 562680, clone E50-2440 RUO), anti-Ly6C-Bv421
   (BioLegend, 128032, clone HK1.4), and anti-Ly6G-PerCP-Cy5.5 (BioLegend,
   127616, clone 1A8), each at a concentration of 1:200 in PBS, for 1 h at
   4 °C. After washing, the cells were stored in 2% paraformaldehyde (PFA,
   [244]Electron Microscopy Sciences, 15714-S) until analysis on the BD
   LSRII (BD Biosciences) using FACSDiva v9. Flow cytometry was analyzed
   with [245]FlowJo v10. Mouse BAL and lung cells were identified as
   follows:

   [246]alveolar macrophages: CD45^+CD11b^+/−SiglecF^+[247]CD11c^+;

   monocytes: CD45^+CD11b^+SiglecF^−MHCII^−CD11c^−Ly6G^−Ly6C^+/−;

   neutrophils: CD45^+CD11b^+SiglecF^− MHCII^−CD11c^−Ly6G^+Ly6C^+/−.

Cytokine analysis

   Cytokine concentrations in mouse BALF supernatants were quantified by
   Eve Technologies (Calgary, Canada) using a bead-based multiplexing
   technology.

Semi-targeted metabolomics

   Metabolites in the BALF were identified and quantified by
   high-resolution [248]liquid chromatography mass spectrometry (LC-MS) at
   the Calgary Metabolomics Research Facility (Calgary, Canada). The
   metabolites were extracted in a 50% methanol (Alpha Aesar #22909):water
   (v/v) solution. Sample runs were performed on a Q Exactive HF Hybrid
   Quadrupole-Orbitrap mass spectrometer (Thermo Fisher Scientific)
   coupled to a Vanquish [249]UHPLC System (Thermo Fisher Scientific).
   Chromatographic separation was achieved on a Syncronis [250]HILIC UHPLC
   column (2.1 mm × 100 mm × 1.7 μm, Thermo Fisher Scientific) using a
   binary solvent system at a flow rate of 600 μl/min. Solvent A consists
   of 20 mM [251]ammonium formate pH 3 in mass spectrometry grade water
   and solvent B consists of mass spectrometry grade [252]acetonitrile
   with 0.1% [253]formic acid (v/v). The following gradient was used:
   0–2 min, 100% B; 2–7 min, 100–80% B; 7–10 min, 80–5% B; 10–12 min, 5%
   B; 12–13 min, 5–100% B; 13–15 min, 100% B. A sample injection volume of
   2 μl was used. The mass spectrometer was run in negative full scan mode
   at a resolution of 240,000 scanning from 50 to 750 m/z. Metabolites
   were identified by matching observed m/z signals (±10 ppm) and
   [254]chromatographic retention times to those observed from commercial
   metabolite standards (Sigma-Aldrich). The data were analyzed using
   E-Maven and are listed in Supplementary Data [255]8-[256]9.

Isolation of RNA from mouse lungs

   Total RNA was isolated from excised mouse lungs using TRIzol reagent
   (Thermo Fisher Scientific, 15596026) according to the manufacturer’s
   instructions. Residual DNA was then degraded using the TURBO DNA-free
   Kit (Thermo Fisher Scientific, AM1907) prior to reverse transcription,
   which was performed using the High-Capacity cDNA Reverse Transcription
   Kit (Thermo Fisher Scientific, 4374966). Real-time qRT-PCR was
   performed on the cDNA generated in the previous step using the primers
   listed in Supplementary Data [257]7. The data were analyzed using the
   ΔΔC[T] method.

Quantification and statistical analysis

   We modeled the number of independent experiments required to reach
   significance between groups using the computer program JMP. This
   simulation was based on experimental design, preliminary data and past
   experience. Analyses were performed assuming a 20% standard deviation
   and equivalent variances within groups. Significance <0.05 with power
   0.8 was used. Experiments in this study were not performed in a blinded
   fashion. All analyses and graphs were performed using the GraphPad
   Prism 7a and 8 software. Values in graphs are shown as average ±SEM and
   data were assumed to follow a [258]normal distribution. For comparison
   between average values for more than two groups, we performed one-way
   ANOVA with a multiple posteriori comparison. When studying two or more
   groups along time, data was analyzed using two-way ANOVA with a
   multiple posteriori comparison. Differences between two groups were
   analyzed using a student’s t test for normally distributed data or
   Mann–Whitney or Kruskal–Wallis test otherwise. Differences were
   considered significant when a P value under 0.05 (p < 0.05) was
   obtained. Statistical details of experiments are indicated in each
   figure legend. No data points were excluded.

Supplementary information

   [259]Supplementary Information^ (3.6MB, pdf)
   [260]41467_2025_62453_MOESM2_ESM.pdf^ (61.6KB, pdf)

   Description of Additional Supplementary Files
   [261]Supplementary Data 1^ (15.7KB, xlsx)
   [262]Supplementary Data 2^ (127.5KB, xlsx)
   [263]Supplementary Data 3^ (26.3KB, xlsx)
   [264]Supplementary Data 4^ (11.3KB, xlsx)
   [265]Supplementary Data 5^ (30.1KB, xlsx)
   [266]Supplementary Data 6^ (10.9KB, xlsx)
   [267]Supplementary Data 7^ (12.5KB, xlsx)
   [268]Supplementary Data 8^ (9.9KB, xlsx)
   [269]Supplementary Data 9^ (11.8KB, xlsx)
   [270]Transparent Peer Review file^ (1.4MB, pdf)

Source data

   [271]Source data^ (867.8KB, xlsx)

Acknowledgements