Higher Siderophore Production and Fe Scavenging with Gluconeogenic Substrates Than with Glycolytic Substrates.

We investigated substrate- and species-dependent PVD production by the three Pseudomonas species following growth on four substrates: two carbohydrates (glucose and fructose) and two short-chain carboxylic acids (succinate and malate) (Fig. 1B). Acknowledging that multiple PVD structures may be secreted, in addition to the reported primary structure for each species, we monitored the chromophore common to all PVDs to quantify the total PVD concentrations (10) (SI Appendix, Fig. S1). Across all three strains, the biomass-normalized PVD concentrations were 34% to threefold higher following growth on the gluconeogenic substrates than following growth on the glycolytic substrates (Fig. 1B).

Based on the structural composition of the primary PVD structures, we estimated the relative contribution of different metabolic pathways for each PVD biosynthesis (Fig. 1B). The two P. putida strains, P. putida KT2440 and P. putida S12, have reportedly identical peptides of seven amino acids (aspartate-ornithine-hydroxyaspartate-diaminobutyrate-glycine-serine-cyclic hydroxyornithine) in their primary PVD structures (10, 40) but the P. protegens strain has a different peptide of eight amino acids of different composition (aspartatate-formylhydroxyornithine-lysine-threonine-alanine-alanine-formylhydroxyornithine-lysine) (40, 41) in its primary PVD structure (Fig. 1B). For these PVD structures, the highest carbon demand was from the TCA cycle (58 to 63%), followed by metabolites from downstream of the ED pathway (32 to 34%) and from the PP pathway (8 to 9%) (Fig. 1B). Therefore, PVD biosynthesis requires significant carbon investment from the TCA cycle to meet the carbon demands primarily from α-KG, followed by oxaloacetate (OAA) and succinate (Fig. 1B and SI Appendix, Fig. S2A).

Our data implied that the metabolic burden of PVD biosynthesis may be responsible for the preference of gluconeogenic substrates over the glycolytic substrates to favor PVD production by the Fe-limited Pseudomonas species. Furthermore, we found that the PVD-containing bacterial secretions resulted in 4.4 to 13.1 µM dissolved Fe (on average) following reactions with three crystalline Fe-bearing oxide minerals (goethite, hematite, and magnetite; 1 g L⁻¹) (Fig. 1C). Notably, relative to the glucose-derived bacterial secretions, the succinate-derived bacterial secretions led to up to sixfold higher dissolved Fe concentration from hematite and magnetite (Fig. 1C). Therefore, the enhanced secretion of PVD from gluconeogenic substrates was advantageous to Fe scavenging (Fig. 1C).

Increased Fluxes through the TCA Cycle Are Responsible for Enhanced Siderophore Production.

We performed kinetic isotopic profiling of P. putida KT2440 fed on [U-¹³C₄]-succinate or on [U-¹³C₆]-glucose to capture in vivo carbon fluxes into six PVD precursors: tyrosine, 3-phosphoglycerate (3-PG), pyruvate, aspartate, α-KG, and ornithine; we also obtained relative intracellular and extracellular levels of each metabolite (Fig. 1D). Compared to feeding on the labeled glucose, feeding on the labeled succinate resulted in a two- to fivefold higher isotopic enrichment in four of the metabolites (3-PG, aspartate, α-KG, and ornithine), three of which are derived from the TCA cycle (Fig. 1 A and D). The ¹³C enrichment of tyrosine and pyruvate was the same during feeding on both substrates (Fig. 1D). Tyrosine is made from metabolites derived from both the PP pathway and downstream of the ED pathway, which may explain the high incorporation of glucose-derived ¹³C carbons (Fig. 1 A and D). The low isotopic enrichment of pyruvate may be due to the high abundance of nonlabeled pyruvate in the extracellular medium (Fig. 1D). Taken together with the relative intracellular abundance of the metabolites, the kinetic labeling data demonstrated a faster metabolic flux of succinate-derived carbons than glucose-derived carbons for the PVD biosynthetic precursors derived from the TCA cycle (Fig. 1D).

Previous work (9) on Fe-limited P. putida cells grown on glucose has revealed that, in addition to a sevenfold decrease in biomass growth and a twofold decrease in glucose consumption relative to Fe-replete cells, enhanced gluconate secretion accounted for up to 50% of consumed glucose in the Fe-limited cells. The periplasmic oxidation of glucose to gluconate could be energetically beneficial to the Fe-limited cells by generating reducing equivalents (as reduced ubiquinone [UQH₂]) without exerting the expense of processing glucose intracellularly through a compromised metabolism (42). The metabolic overflow via gluconate secretion may explain both the slow metabolic flux of assimilated glucose toward siderophore precursors and the low siderophore yield observed here (Fig. 1 B and D). Metabolic overflow was also reported for Fe-limited E. coli (5), which secreted 70% of consumed glucose as acetate and lactate, and Fe-limited C. mineralivorans (29), which secreted over 40% of consumed glucose as gluconate. With respect to the physiological response of the succinate-grown cells to Fe limitation, we recorded a threefold decrease in biomass growth and a fourfold decrease in succinate consumption, indicating that both decreased carbon assimilation and compromised metabolism were also characteristic of succinate-grown Fe-limited cells (SI Appendix, Fig. S2B). We sought to understand how succinate metabolism still could achieve carbon metabolism to benefit siderophore biosynthesis in Fe-limited cells.

Enhanced Anaplerosis in Fe-Deficient Cells Promotes Succinate Carbon Retention in the TCA Cycle.

We investigated metabolic routing in Fe-replete and Fe-limited P. putida KT2440 grown on [1,4-¹³C₂]-succinate alone (Fig. 2A). This doubly ¹³C-labeled succinate would first generate doubly ¹³C-labeled fumarate and doubly ¹³C-labeled OAA through the TCA cycle and, after decarboxylation through gluconeogenic fluxes, singly ¹³C-labeled pyruvate; subsequent runs through the TCA cycle would yield nonlabeled metabolites (Fig. 2A). We measured nonlabeled (26%), singly ¹³C-labeled (29%), and doubly ¹³C-labeled (45%) fractions of OAA in Fe-limited cells, whereas Fe-replete cells only had 5% singly ¹³C-labeled fractions (Fig. 2A). The only metabolic route for singly ¹³C-labeled OAA would be the carboxylation of nonlabeled pyruvate by labeled CO₂ or carboxylation of singly ¹³C-labeled pyruvate by nonlabeled CO₂ (Fig. 2A). Using the labeling data and flux ratio analysis, we quantified the relative contribution of TCA cycle-derived flux versus anaplerotic flux from pyruvate to OAA biosynthesis (Fig. 2A). Compared to Fe-replete cells, the contribution from the TCA cycle was reduced by nearly half (from 87 to 49%) and the anaplerotic flux from pyruvate became the primary contributor to OAA biosynthesis (from 15 to 51%) in the Fe-limited cells, thus explaining the observed retention of succinate-derived carbons in PVD precursors from the TCA cycle (Figs. 1D and 2A).

Fig. 2.

Anaplerotic flux favored over gluconeogenic flux to promote succinate carbon retention in the TCA cycle. Carbon mapping, metabolite labeling, and flux ratio analysis of (A) assimilated [1,4-¹³C₂]-succinate or (B) assimilated [U-¹³C₄]-succinate with unlabeled glucose in exponentially growing P. putida KT2440 cells cultured in Fe-replete [(+)Fe] and intermediate Fe-limited [(−)Fe] conditions. Dashed arrow represents a minor flux. Addition of CO₂-derived carbon in OAA synthesis is shown with a green box. The measured data (mean ± SD) were from three biological replicates. The metabolite abbreviations are as described in the Fig. 1 legend.

When the cells were fed on a mixture of [U-¹³C₄]-succinate with unlabeled glucose, there was an increased incorporation of glucose-derived nonlabeled carbons into OAA and the oxidative side of the TCA cycle in the Fe-limited cells compared to Fe-replete cells, demonstrating again the preference of anaplerosis over gluconeogenesis in response to Fe limitation (Fig. 2B). Specifically, two-thirds of OAA were derived from the TCA cycle and one-third from pyruvate in Fe-replete cells, whereas, in the Fe-limited cells, only 10% of OAA was derived from the TCA cycle and 90% was from pyruvate (Fig. 2B). The preferential incorporation of ¹³C-glucose carbons indicated a preference for glucose utilization over succinate utilization in the Fe-limited cells (Fig. 2B). To shed light on Fe-dependent substrate selectivity and metabolic remodeling, we investigated different scenarios of substrate mixtures (Figs. 3 and 4).

Fig. 3.

Fe-limited cells trigger substrate hierarchy during mixed-substrate usage. (A) Substrate consumption during growth of Fe-replete [(+)Fe] and intermediate Fe-limited [(−)Fe] P. putida KT2440 on carbon-equivalent mixture with (Left) 1:1 glucose:succinate, (Middle) 1:1 glucose:citrate, or (Right) 1:1 glucose:acetate mixture; samples were obtained during lag phase, midexponential phase, and at the onset of stationary phase. Intracellular metabolite labeling in exponentially growing P. putida KT2440 cells after assimilation of (B) [U-¹³C₆]-glucose and unlabeled succinate or (C) [U-¹³C₆]-glucose and unlabeled citrate. (D) Electrospray-ionized monoisotopic positive ions ([M + H]⁺) of the primary PVD siderophore secreted by P. putida KT2440 during intermediate Fe-limited growth on unlabeled glucose (¹²C-glucose) alone, unlabeled succinate (¹²C-succinate) alone, unlabeled citrate (¹²C-citrate) alone, fully labeled glucose with unlabeled succinate (¹³C-glucose:¹²C-succinate), or fully labeled glucose with unlabeled citrate (¹³C-glucose:¹²C-succinate). Measured data (mean ± SD) were from biological replicates (n = 3). The metabolite abbreviations are as described in the Fig. 1 legend.

Fig. 4.

Enhanced siderophore production associated with preferential carbon usage and hierarchy in carbon metabolism. (A) Kinetics of substrate depletion (in % of total C) by P. putida KT2440 grown on 1:1 glucose:benzoate mixture under, from Left to Right, Fe-replete [(+)Fe], intermediate Fe-limited [Int(−)Fe], and Fe-limited [(−)Fe] conditions. (B–E) Intracellular metabolite labeling after assimilation of [U-¹³C₆]-glucose and unlabeled benzoate during (B) (+)Fe, (C) Int(−)Fe, (D) (−)Fe phase 1, and (E) (−)Fe phase 2 conditions; metabolites with glucose-derived ¹³C carbons and benzoate-derived unlabeled carbons are depicted in red and dark blue shades, respectively. Sampling times for the metabolite labeling data shown in B–E are shown by the white arrows in A; detailed graphs of the metabolite labeling data are in SI Appendix, Fig. S3. (F) Siderophore production rate (µmol PVD g_CDW⁻¹ h⁻¹) during (−)Fe growth on only glucose or growth on the glucose:benzoate (Gluc:Benz) mixture under int(−)Fe), (−)Fe phase 1, and phase 2 conditions (one-way ANOVA and Tukey’s studentized range test: ns, not statistically significant; ***P < 0.001). (G) Total dissolved Fe (µmol Fe L⁻¹) from the dissolution of Fe-bearing minerals (goethite, hematite, and magnetite; 1 g L⁻¹) following reactions with bacterial secretions obtained with P. putida KT2440 cells at the end of growth on the Gluc:Benz mixture under (+)Fe, (−)Fe phase 1, and (−)Fe phase 2 conditions (two-tailed t test: *P < 0.05; ns, not significant). In A and B, the (+)Fe consumption and labeling data used here as comparative reference data were obtained from previously conducted experiments (43). All data were obtained from three biological replicates. The metabolite abbreviations are as described in the Fig. 1 legend.

Iron Limitation Triggers Substrate Selectivity during Mixed-Substrate Growth.

We monitored substrate depletion during feeding on 1:1 carbon-equivalent mixture of glucose with succinate, citrate, acetate, or benzoate (Figs. 3A and 4A). There was equal or near-equal consumption of both substrates by Fe-replete cells but, in response to Fe limitation, we observed consistently that one substrate was consumed preferentially (Figs. 3A and 4A). With the Fe-limited cells, on one hand, glucose was depleted two times faster than succinate or 50% faster than acetate but, on the other hand, citrate was depleted six times faster than glucose and benzoate was depleted at least two times faster than glucose (Figs. 3A and 4A). The preferential consumption of citrate over glucose may be because citrate both served as a Fe chelator and provided carbon influx toward PVD biosynthesis (Fig. 1A). The role of citrate in Fe scavenging or in relieving Fe-limited growth has been reported in P. putida strains (44), the bacterium Bacillus cereus (45), the fungus Aspergillus niger (46), the bacterial fish pathogen Photobacterium damselae (47), and in the cyanobacterium Trichodesmium erythraeum (48). In our starting nutrient composition, we computed that 98% of the exogenous Fe was complexed with citrate, emphasizing the role of citrate as a notable Fe chelator (SI Appendix, Table S1).

When glucose was provided with benzoate (a gluconeogenic substrate that does not bind Fe strongly), there were near-equal consumption rates of both substrates in the Fe-replete cells (19.8 ± 4.4 mmol C g_CDW⁻¹ h⁻¹ and 27.2 ± 8.1 mmol C g_CDW⁻¹ h⁻¹ for glucose and benzoate, respectively; g_CDW is the cell dry weight in grams) as previously reported (43), but the Fe-limited cells preferably consumed benzoate over glucose, whereby decrease in Fe availability led to an increase in this carbon hierarchy (Fig. 4A). At the intermediate Fe-limited condition, benzoate consumption was 57% faster than glucose consumption and, for the severe Fe-limited condition, there were two clear phases of substrate depletion wherein benzoate depletion (64.5 ± 1.2 mmol C g_CDW⁻¹ h⁻¹) occurred rapidly prior to the start of the glucose depletion (15.2 ± 0.2 mmol C g_CDW⁻¹ h⁻¹) (Fig. 4A). To understand the intracellular metabolism associated with the different aforementioned substrate hierarchies, we performed ¹³C-metabolomics profiling of P. putida KT2440 cells grown on [U-¹³C₆]-glucose with unlabeled succinate, unlabeled citrate, or unlabeled benzoate (Figs. 3 B–D and 4 B–E).

Prioritizing Anaplerotic Flux versus Gluconeogenetic Flux Depends on Carbon Influx into the TCA Cycle.

In the mixture with ¹³C-glucose with unlabeled succinate, the assimilation flux of the glucose-derived ¹³C carbons in the TCA cycle was preferred over the flux of succinate-derived nonlabeled carbons, in accordance with enhanced flux from lower glycolysis (i.e., anaplerosis) toward the TCA cycle as already discussed (Figs. 2B and 3B). However, the assimilation flux of the citrate-derived nonlabeled carbons preferentially populated both the TCA cycle and lower glycolysis, implying a preference of gluconeogenic flux over anaplerotic flux when the carbon source was fed directly into the oxidative side of the TCA cycle (Fig. 3C). We confirmed that these different hierarchies in the carbon selectivity were reflected in the isotopic enrichment of the resulting PVD produced (Fig. 3D).

We compared the isotopologue distribution of the major [M + H]⁺ PVD ion in the bacterial secretions following growth on unlabeled substrate alone (glucose, succinate, or citrate) or a mixture of ¹³C-labeled glucose with unlabeled succinate or unlabeled citrate (Fig. 3D). The unlabeled PVD [M + H]⁺ ion (mass-to-charge ratio [m/z] of ∼1,073.4210) detected during growth on each unlabeled single substrate was replaced with variable ¹³C enrichment of this ion in agreement with the carbon selectivity (Fig. 3D). From the Gaussian isotopologue distribution of the PVD ion, the highest abundant peak had a m/z value of 1,092.4887 during feeding on the mixture with ¹³C-glucose and unlabeled succinate, but the corresponding peak had a m/z value of 1,080.4545 during feeding on the mixture with ¹³C-glucose and unlabeled citrate, indicating a higher incorporation of glucose-derived ¹³C carbons into the PVD structure in the presence of succinate than in the presence of citrate (Fig. 3D). Although the initial catabolism of succinate and citrate both require Fe-containing enzymes (respectively, succinate dehydrogenase and aconitate hydratase), our data implied that the assimilation of these two substrates was under different metabolic regulation.

With respect to the glucose:benzoate conditions, the extent of carbon hierarchy was enhanced upon worsening Fe deficiency (Fig. 4 A–E). In Fe-replete cells, the catabolism of glucose- and benzoate-derived carbons was nonuniform throughout the metabolic network, as detailed in a previous study (43) (Fig. 4B). The metabolites in initial glucose catabolism (gluconate, glucose-6-phosphate [G6P]), the ED pathway (6-PG, dihydroxyacetone-phosphate [DHAP]), the upper Embden–Meyerhof–Parnas (EMP) pathway (G6P, fructose-6-phosphate [F6P], fructose-1,6-bisphosphate [FBP]), and the PP pathway (xylulose-5-phosphate [Xu5P], ribose-5-phosphate [R5P], and sedoheptulose-7-phosphate [S7P]) were populated primarily by glucose-derived ¹³C-labeled carbons (by over 95%) (Fig. 4B). The metabolites in the TCA cycle (OAA, citrate, α-KG, fumarate, and malate) were derived from both nonlabeled benzoate carbons (60% or higher) and glucose ¹³C carbons (about 40% or lower) (Fig. 4B and SI Appendix, Fig. S3). The metabolites upstream of the TCA cycle (3-PG, PEP, and pyruvate) were enriched with glucose-derived ¹³C-carbons with some minor fractions of benzoate-derived nonlabeled carbons (Fig. 4B and SI Appendix, Fig. S3). With the intermediate Fe deficiency, the fraction of benzoate-derived nonlabeled carbons was increased in metabolites both upstream of and throughout the TCA cycle (Fig. 4 C and D and SI Appendix, Fig. S3). At the most severe Fe-limited condition, the first phase of substrate utilization led to preferential incorporation of nonlabeled benzoate carbons over ¹³C-glucose carbons, including an appreciable fraction (up to 25%) of partially ¹³C-labeled fraction of metabolites in initial glucose catabolism (G6P, F6P, and FBP) (Fig. 4D and SI Appendix, Fig. S3). In the second phase, the switch to glucose after benzoate depletion led to high ¹³C-labeled enrichment (over 70%) of all metabolites in the TCA cycle (Fig. 4 A and E and SI Appendix, Fig. S3). In sum, the metabolite labeling patterns from the different mixture scenarios (glucose with succinate, glucose with citrate, or glucose with benzoate) revealed that Fe deficiency led to a rerouted carbon metabolism in association with a hierarchy for gluconeogenic substrates with direct influx into the oxidative side of the TCA cycle over the anaplerotic flux from glucose-derived carbons (Figs. 3 B and C and 4 B–E).

Carbon Hierarchy Leads to Enhanced Siderophore Production and Fe Scavenging.

We evaluated explicitly the consequence of the carbon hierarchy of benzoate over glucose on siderophore secretion and Fe scavenging (Fig. 4 F and G). Compared to the PVD secretion rate under the intermediate Fe-limited condition (41.4 ± 3.59 µmol g_CDW⁻¹ h⁻¹), the lowest Fe availability led to a near 2-fold increase in the PVD secretion rate (91.8 ± 15.2 µmol g_CDW⁻¹ h⁻¹) during the first phase when benzoate was the primary substrate consumed from the mixture (Fig. 4 A and F). During the second phase when growth was supported only by glucose from the mixture, there was a 5-fold decrease in the PVD secretion rate (17.5 ± 2.86 µmol g_CDW⁻¹ h⁻¹) relative to the first phase; Fe-limited cells grown on glucose alone had a similar PVD secretion rate (13.0 ± 8.0 µmol g_CDW⁻¹ h⁻¹) (Fig. 4F). Given reported overflow of oxidized glucose as secreted gluconate (9), we compared the gluconate secretion rates during the two phases of substrate consumption in the Fe-limited cells and found a 200-fold increase in gluconate secretion (from 11.6 ± 4.6 µmol g_CDW⁻¹ h⁻¹ to 2.3 ± 0.3 mmol g_CDW⁻¹ h⁻¹) during the glucose-consumption phase compared to the benzoate-consumption phase (SI Appendix, Fig. S4A). Therefore, the preferential uptake of benzoate-derived carbons resulted in the highest rate of PVD production and bypassed the excess carbon loss via gluconate secretion (Fig. 4 A–F and SI Appendix, Figs. S4 A and B and S5).

Moreover, we found that the PVD-containing bacterial secretions from the glucose:benzoate growth condition led to significant levels of dissolved Fe, ranging from 2.8 to 20.3 µM of Fe on average, extracted from the aforementioned Fe-oxide minerals (Fig. 4G). Across the two consumption phases, the total extent of Fe scavenged from the minerals was largely achieved by the bacterial secretions following the benzoate-consumption phase; following the glucose-consumption phase, the dissolved Fe concentration either remained the same (for goethite and magnetite) or was increased by only 35% (for hematite) (Fig. 4G). Thus, the enhanced PVD secretion associated with the carbon hierarchy optimized Fe scavenging.

Enzymatic Regulation Versus Metabolic Regulation of Fe-Dependent Carbon Hierarchy to Enhance Siderophore Production.

We explored how changes in enzyme abundance versus changes in metabolic fluxes facilitated the aforementioned substrate hierarchy to enhance PVD production (Figs. 5 and 6). We compared data obtained with Fe-limited cells during the first (or benzoate-only) consumption phase from the glucose:benzoate mixture to those obtained with Fe-replete cells grown on the same carbon mixture (Figs. 5 and 6). Here we highlight changes in protein abundance that were 10% less or greater in the Fe-limited cells than in the Fe-replete cells (Fig. 5A and SI Appendix, Table S2). Along with the proteomics analysis, we used the ¹³C-metabolomics data illustrated in Fig. 4 B and C, to perform flux ratio analyses and characterize changes in the biosynthetic routes at four key metabolic nodes: OAA, GAP, PEP, and pyruvate (Fig. 6 A–D).

Fig. 5.

Fe-dependent carbon selectivity and siderophore biosynthesis facilitated by protein abundance changes. (A) Percent changes (greater than or less than 10%) in proteins involved in the central carbon metabolism of P. putida KT2440 cells grown on the glucose:benzoate mixture under Fe-limitation versus Fe-replete conditions. (B) Fold change (expressed in log2 of ratio of Fe-limitation data versus Fe-replete data) in the abundance of proteins involved in PVD biosynthesis and transport, proteins involved in Fe regulation, and Fe-containing metabolic proteins. The Fe-limited data were obtained during the first phase of Fe-limited substrate consumption depicted in Fig. 4A. The total profiling of protein abundance ratios and data statistics is in SI Appendix, Table S2. Data were obtained from three biological replicates.

Fig. 6.

Metabolic remodeling of carbon fluxes in response to Fe deficiency. (A–D) Scheme of the metabolic routes for the biosynthesis of four metabolic nodes: (A) GAP derived from 3-PG or the ED pathway, (B) OAA derived from pyruvate or malate via the TCA cycle, (C) PEP derived from 3-PG, Pyr, or OAA via the TCA cycle, and (D) Pyr derived from the ED pathway or from the TCA cycle (via OAA and malate). (E–H) (Top) Fe-dependent percentage (mean ± SD) of each precursor compound or pathway for each metabolic node and (Bottom) estimated fold-change increase or decrease (mean ± SD) in ATP, UQH₂, and NAD(P)H in Fe-limited cells relative to Fe-replete cells of P. putida KT2440 during growth on the glucose:benzoate mixture. The Fe-limited cell data were obtained during the first phase of Fe-limited substrate consumption depicted in Fig. 4A. Analyses were conducted using metabolite labeling data from three biological replicates (SI Appendix, Fig. S3). The metabolite abbreviations are as described in the Fig. 1 legend.

Our proteomics profiling quantified a total of 52 proteins involved in initial substrate catabolism and central carbon metabolism (Fig. 5A). Consistent with the decreased glucose uptake, 9 of the 10 identified enzymes involved in initial glucose catabolism were depleted (by up to −38%) in Fe-limited cells relative to the Fe-replete cells (Figs. 4A and 5A). However, glucose dehydrogenase (gcd) was elevated by 17%, in accordance with reported increase in gluconate secretion (9) in Fe-limited cells fed on glucose (Fig. 5A and SI Appendix, Fig. S4A). With respect to initial benzoate catabolism, five enzymes (benK, catB, catC, pcaJ, and pcaF) were elevated (by up to 20%) in the Fe-limited cells (Fig. 5A). Of the 7 enzymes quantified in the upper EMP, PP, and ED pathways, the levels of 4 were depleted (by 30%) in the Fe-limited cells (Fig. 5A). Notably, in accordance with the depletion (by 29%) in the aldolase enzyme (eda) that splits 6-PG into GAP and pyruvate in the ED pathway, the flux analysis determined a decreased contribution of the ED pathway in GAP biosynthesis in the Fe-limited cells (59%) compared to the Fe-replete cells (83%) (Figs. 5A and 6 A and E). Thus, the carbon selectivity of benzoate over glucose in the Fe-deficient cells was facilitated by changes in the abundance of metabolic enzymes involved in transport and initial catabolism of each substrate (Fig. 5A).

For OAA biosynthesis, the metabolic flux analysis determined a near-equal investment of pyruvate (58%) and the TCA cycle via malate (42%) in the Fe-replete cells, whereas there was a significant dependency on the TCA cycle (82%) with a small contribution from pyruvate (18%) in the Fe-limited cells (Fig. 6 B and F). In agreement with these flux distributions, the level of malate:quinone oxidoreductase (mqo3) for the conversion of malate to OAA was increased (by 62%) and the level of pyruvate carboxylase (accC2) for the carboxylation of pyruvate to OAA was decreased (by 21%) in the Fe-limited cells compared to the Fe-replete cells (Fig. 5A). An increased contribution of gluconeogenic reactions from the TCA cycle at the expense of anaplerosis was also evident in the flux ratio analyses for PEP and pyruvate biosynthesis (Fig. 6 G and H).

Compared to the contributions from the TCA cycle (43%) and pyruvate (53%) for PEP biosynthesis in the Fe-replete cells, the majority of PEP was synthesized from the TCA cycle via OAA (88%) in the Fe-limited cells (Fig. 6C). Of the 12 quantified enzymes in the gluconeogenic/anaplerotic junction pathways, the 2 enzymes involved in the gluconeogenic synthesis of PEP via the OAA node of the TCA cycle were elevated in the Fe-limited cells compared to the Fe-replete cells: phosphoenolpyruvate carboxylase (ppc, by 25%) and oxaloacetate decarboxylase (PP1389, by 19%) (Fig. 5A). For pyruvate biosynthesis, pyruvate was made from both the ED pathway (41%) and the TCA cycle (59%) in Fe-replete cells but only 4% of pyruvate was derived from the ED pathway and the major biosynthetic source was the TCA cycle (96%) in the Fe-limited cells (Fig. 6H). Incongruent with the metabolic flux change in pyruvate biosynthesis, the level of oxaloacetate decarboxylase (oadA) involved in the decarboxylation of OAA to pyruvate was decreased (by 32%) in the Fe-limited cells compared to the Fe-replete cells (Fig. 5A).

Within the TCA cycle, the abundances of 6 metabolic enzymes (mqo3, acnA-1, PP4011, fumC, lpd, and lpdV) were elevated in the Fe-limited cells (Fig. 5A). Most significant was a threefold increase in fumC, which converts fumarate to malate (Fig. 5A). The levels of the enzymes involved in succinate synthesis were also elevated: lpd (by 56%), lpdV (by 22%), and pcaF (by 14%) (Fig. 5A). Notably, the 2 Fe-containing enzymes in the TCA cycle were depleted: aconitate hydratase (acnB, by −15%) and succinate dehydrogenase (sdhB, by −42% decrease) (Fig. 5 A and B). Additionally, in the Fe-limited cells, there was an increased abundance of isocitrate lyase (acea, by 10%) in the glyoxylate shunt, which was consistent with reported induction of the glyoxylate shunt from elevated acetyl-CoA that would be expected from benzoate catabolism (49). Furthermore, from the global proteomics profiling, we found that the levels of 17 Fe-containing enzymes either remained unchanged or were depleted (by up to −82%) in the Fe-limited cells relative to the Fe-replete cells (Fig. 5B). Toward promoting Fe acquisition, proteins involved in PVD biosynthesis and transport were elevated by over 200% in Fe-limited cells, in agreement with reported up-regulation of genes associated with PVD in Fe-limited Pseudomonas species (6) (Fig. 5B).

Using the metabolic flux ratio analyses, we estimated the changes in the energetic yields afforded to the Fe-limited cells due to the metabolic rerouting relative to the Fe-replete cells (Fig. 6 E–H). The increased gluconeogenic flux from the TCA cycle in the biosynthesis of OAA and pyruvate in Fe-limited cells led to a greater than fivefold increase in adenosine triphosphate (ATP) yield and a nearly 150% increase each in reduced nicotinamide adenine (diphosphate) [NAD(P)H] and UQH₂ (Fig. 6 F and H). However, the change in the biosynthetic route for GAP was energetically unfavorable but, since the net consumption of ATP and NAD(P)H from 3-PG to GAP was less than their net generation from the biosynthesis of OAA and pyruvate (43), the overall remodeled metabolic network would still lead to favorable energy fluxes in the Fe-limited cells (Fig. 6E).

Carbon Flux Demand for Siderophore Biosynthesis Drives Metabolic Flux Rerouting.

Thus far, our data have demonstrated that hierarchy in carbon metabolism promoted fluxes in the TCA cycle and led to enhanced PVD production, despite the depletion of Fe-containing and other proteins in the TCA cycle and connected metabolic pathways (Fig. 4F). We conducted a quantitative analysis of the metabolic flux demands for biomass growth versus siderophore biosynthesis during P. putida KT2440 growth on the glucose:benzoate mixture across the different Fe conditions. We combined measured data (growth rates, carbon consumption rates, siderophore production rates, and metabolite secretion rates) and computed data (biomass biosynthetic demand, PVD biosynthetic demand, and CO₂ efflux) (Fig. 7A). The metabolic flux demand to sustain biomass growth was based on genome-scale cellular stoichiometry (50) and the PVD biosynthetic demand was based on the pre-PVD structure of the primary PVD secreted by P. putida KT2440 (10, 40) (Fig. 7A). Guided by previous metabolic flux analyses (9, 43), the CO₂ efflux was estimated by performing flux balance analysis using the ¹³C-labeling patterns obtained at the different Fe conditions (SI Appendix, Table S3 and Fig. S2 and Fig. 7A).

Fig. 7.

Metabolic pathway flux demand for biomass growth versus siderophore biosynthesis. (A) Measured carbon influx, measured effluxes (gluconate secretion, siderophore production, and biomass production), and estimated CO₂ production (mmol C g_CDW h⁻¹) in P. putida KT2440 grown on 1:1 glucose:benzoate mixture under Fe-replete [(+)Fe], intermediary Fe-limited [Int(−)Fe], Fe-limited [(−)Fe] phase 1 (P1), and (−)Fe phase 2 (P2) conditions. Color code: CO₂ production (green), gluconate secretion (yellow), siderophore production (orange), biomass production (blue), and substrate carbon influx (gray). (B) Calculated pathways flux demand (mmol C g_CDW⁻¹ h⁻¹) required from central carbon metabolism to sustain (Left) measured biomass growth and (Right) biosynthesis of measured PVD secretion rate (one-way ANOVA, Tukey’s studentized range test: ns, not statistically significant; ***P < 0.001). (C) Percentage (%) increase in metabolic flux demand for PVD biosynthesis relative to flux demand for biomass growth alone. Color code for B and C: EMP pathway, dark blue; PP pathway, red; downstream ED pathway, light orange; and TCA cycle, light blue. All data, shown as mean ± SD, were obtained from three biological replicates.

Compared to the carbon influx rate in the Fe-replete cells (47.0 ± 9.2 mM C g_CDW⁻¹ h⁻¹), the corresponding rate was 56% higher in the intermediate Fe-limited condition (73.5 ± 11 mM C g_CDW⁻¹ h⁻¹), 41% higher in first phase of the severe Fe-limited condition (66.1 ± 1.4 mM C g_CDW⁻¹ h⁻¹), and threefold lower in the second phase (Fig. 7A). Siderophore secretion rate accounted for only 3.5%, 8.6%, and 7.1% of the carbon influx rate for the intermediate Fe-limited condition, first phase, and second phase of the severe Fe-limited condition, respectively (Fig. 7A). The primary carbon efflux was instead for biomass production (about 50% or higher) in all conditions, except for the latter condition (i.e., second phase of the Fe-limited condition) wherein 52% of the carbon efflux was due to gluconate secretion (Fig. 7A). Compared to the CO₂ efflux estimated for the Fe-replete cells (19.6 ± 0.98 mM C g_CDW⁻¹ h⁻¹) (43), this value was twofold higher in the intermediate Fe-limited condition (39.1 ± 2.1 mM C g_CDW⁻¹ h⁻¹), 20% higher in the first phase of severe Fe-limited condition (23.4 ± 1.2 mM C g_CDW⁻¹ h⁻¹), and 73% lower in the second phase of severe Fe-limited condition (5.24 ± 0.3 mM C g_CDW⁻¹ h⁻¹) (Fig. 7A). These values implied that siderophore biosynthesis involved CO₂-generating metabolic reactions. For the second phase of the severe Fe-limited condition, there was a large discrepancy between the total carbon efflux (26.7 ± 2.7 mM C g_CDW⁻¹ h⁻¹) and the carbon influx (15.2 ± 0.2 mM C g_CDW⁻¹ h⁻¹), which we attributed to the unaccounted contribution of carbon retention from the first phase beyond the measured carbon influx in the second phase (Fig. 7A). We also considered the possible secretion of other organic acid metabolites but their summed concentration was negligible (<1 mM C g_CDW⁻¹ h⁻¹) across all conditions (SI Appendix, Table S4).

Metabolic flux requirements to support biomass biosynthesis were satisfied primarily by the TCA cycle (∼40%), downstream of the ED pathway (∼40%) and the PP pathway (∼20%); contribution of the EMP pathway was minimal (Fig. 7B). Despite the relatively low fractional amount of the PVD biosynthetic efflux within the total carbon efflux, the relative metabolic flux requirement for PVD biosynthesis was significant. Specifically, due to the metabolic requirements for the PVD siderophore biosynthesis in the Fe-limited cells (Fig. 1B), the metabolic flux demand was increased by up to 30% from the TCA cycle and less than 10% from either the downstream ED pathway or from PP pathway (Fig. 7C). Therefore, the substrate hierarchy and metabolic remodeling in Fe-limited cells determined from our metabolomics data were aimed at satisfying the excess carbon flux demand required from the TCA cycle to sustain PVD biosynthesis (Figs. 5A, 6 E–H, and 7C).