Pseudomonas aeruginosa reverse diauxie is an optimized, resource utilization strategy

Pseudomonas aeruginosa is a globally-distributed bacterium often found in medical infections. The opportunistic pathogen uses a different, carbon catabolite repression (CCR) strategy than many, model microorganisms. It does not utilize a classic diauxie phenotype, nor does it follow common systems biology assumptions including preferential consumption of glucose with an ‘overflow’ metabolism. Despite these contradictions, P. aeruginosa is competitive in many, disparate environments underscoring knowledge gaps in microbial ecology and systems biology. Physiological, omics, and in silico analyses were used to quantify the P. aeruginosa CCR strategy known as ‘reverse diauxie’. An ecological basis of reverse diauxie was identified using a genome-scale, metabolic model interrogated with in vitro omics data. Reverse diauxie preference for lower energy, nonfermentable carbon sources, such as acetate or succinate over glucose, was predicted using a multidimensional strategy which minimized resource investment into central metabolism while completely oxidizing substrates. Application of a common, in silico optimization criterion, which maximizes growth rate, did not predict the reverse diauxie phenotypes. This study quantifies P. aeruginosa metabolic strategies foundational to its wide distribution and virulence.


Introduction 1
substrates was measured and phenotypic characteristics, like the general lack of an overflow 23 metabolism, were quantified. Proteomics measured a constitutive core metabolism centered on 24 respiration and a dynamic set of enzymatic pathways that catabolized specific substrates, 25 Enzymes from the tricarboxylic acid (TCA) cycle and associated auxiliary enzymes had 1 largely, constitutive abundances regardless of the medium formulation and the growth phase 2 (Fig. 3). All TCA cycle enzymes except the membrane-associated succinate dehydrogenase 3 were detected and quantified. Additionally, the enzymes oxaloacetate decarboxylase (PA4872) 4 and PEP synthase which process metabolic intermediates for the TCA cycle, were expressed 5 constitutively. The abundance of ATP synthase subunits was also constitutive. Membrane-6 associated, electron transport chain enzymes were not detected. It was assumed that the 7 enzymes were also constitutively expressed based on the TCA cycle and the ATP synthase 8 protein abundances and the lack of an overflow metabolism. 9 Enzymes associated with the processing of specific substrates did change in abundance 10 based on presence and concentration of substrates, contrary to most TCA cycle enzymes (Fig.  11   4). Aspartate was plotted as a representative top tier amino acid (Table 2). Protein abundance 12 for aspartate ammonia-lyse (AspA), responsible for the catabolism of aspartate, was elevated 13 during the first exponential growth phase. When aspartate was exhausted, the abundance of 14 The order of substrate preference is hypothesized to reflect the ecological strategy used by P. 1 aeruginosa to thrive in environmental and medical niches. Computational systems biology 2 tested hypotheses regarding what phenotypic properties were being optimized in the Pa 215 3 cultures. Analysis utilized a genome-scale, stoichiometric metabolic model of P. aeruginosa 4 updated here with genome-supported, amino acid catabolism reactions [42-44](supplemental 5 material S11). 6 In silico testing of ecological strategies was applied first to amino acid preference and 7 included all experimentally measured amino acids except for aromatic and sulfur containing 8 amino acids due to their specialty chemistries. The amino acid preferences did not correlate 9 with the amino acid frequency in genome open reading frames (Fig. 5a) indicating the amino 10 acids were not consumed solely for protein assembly; amino acids were also used as anabolic 11 building blocks for other macromolecules and catabolized for cellular energy. Therefore, 12 simulations considered either the production of cellular energy (e.g. ATP) or cellular growth 13 which was quantified as carbon moles (Cmol) of biomass. 14 The first round of in silico analyses considered three separate, single dimension 15 optimization criterion 1) maximizing biomass or energy production rates based on electron 16 donor, 2) maximizing product rates based on electron acceptor (O2), or 3) minimizing nutrient 17 investment into the proteome required for product synthesis. 18

Amino acid preference does not correlate with in silico maximization of rates 19
Computational approaches for studying metabolism often assume cells utilize metabolic 20 potential to maximize growth rate [45][46][47]. Amino acid preference was analyzed using this 21 theory. Simulations identified the optimal conversion of each individual amino acid into cellular 22 energy or biomass. The in silico phenotypes which maximized yields, oxidized completely the 23 substrates, consistent with in vitro cultures which did not utilize overflow metabolisms. 24 Maximizing cellular energy yield for each amino acid, on a mole substrate or Cmol 25 substrate basis, did not correlate with amino acid preference, having r 2 values of 0.06 and 0.15 26 respectively (Fig. 5b). The phenotypes which maximized yields were used to calculate 1 maximum product rates using enzyme parameters from survey studies [48-50] and experimental 2 medium composition (supplemental material S1). Maximizing the rate of cellular energy 3 production (or growth) did not predict amino acid preference, as the correlation was r 2 = 0.17 4 (Fig 5c)(supplemental material S12-S17). 5 P. aeruginosa has a respiration-centric metabolism. The rate maximization criterion was 6 also applied to O2 which is required to fully oxidize the amino acids under the experimental 7 conditions. This alternative, rate maximization criterion predicted substantial overflow 8 metabolisms for most of the amino acids. This predicted phenotypic trait was not consistent with 9 the experimental data indicating the criterion was not relevant for Pa 215 metabolism 10 (supplemental material S12). 11 Amino acids with high cellular energy yields (mol ATP (mol amino acid) -1 ) also had high 12 biomass yields (Cmol biomass (mol amino acid) -1 ); the two yields correlated with an r 2 value of 13 0.99 (supplemental data S17). Therefore, maximizing rates for cellular energy production or 14 biomass production had similar trends and neither predicted rCCR phenotype (supplemental 15 material S12). 16

Single dimension optimization of resource investment predicts overflow metabolism, 17 inconsistent with rCCR phenotypes 18
Computational analysis was used to identify phenotypes that minimized resource investment 19 into catabolic pathways. Explicit investment models were not possible in non-model organism, 20 P. aeruginosa. Therefore, two previously developed resource proxies were applied to estimate 21 relative, proteome investment into central metabolism [23,26,51,52]. The flux minimization 22 proxy [23, 26, 53] assumes the total network flux is proportional to the enzymatic resources 23 required to synthesize the necessary proteome. Another proxy for protein investment minimizes 24 the number of enzyme catalyzed reactions (a.k.a. minimal proteome investment) identifying the 25 smallest proteome required to realize an in silico phenotype [23]. These hypotheses assumed 1 central metabolism enzymes could be approximated as having the same molecular weight with 2 the same amino acid distribution. 3 Both proxies for resource investment, when applied as the single optimization criterion, 4 predicted overflow metabolisms for most amino acids (supplemental material S14, S15). The in 5 vitro experimental cultures did not demonstrate substantial overflow metabolisms indicating this 6 single criterion was not relevant for Pa 215 phenotypes. 7

Substrate preference is consistent with a resource utilization strategy optimizing 8 substrate oxidation and proteome investment 9
Life occurs in multifactorial environments with multiple stressors influencing phenotypes 10 [24, 54, 55]. Two dimensional, optimizations of in silico phenotypes were performed where the 11 first dimension considered the optimal conversion of substrate into cellular energy which 12 completely oxidized the substrate. The second dimension quantified the nutrient investment 13 necessary to synthesize the central metabolism proteome (supplemental material S14, S15). 14 Both the flux minimization and minimal proteome investment proxies were tested. Two-15 dimensional optimization (2-DO) using the flux minimization proxy had poor correlations with the 16 observed amino acid preferences (Fig 6a). Alternatively, 2-DO using complete substrate 17 oxidation and the minimal proteome proxy predicted the experimental preference for amino 18 acids (r 2 = 0.67) (Fig. 6b). 19 2-DO was further refined using experimentally measured proteomics data. The 20 constitutively expressed TCA cycle, anaplerotic enzymes, ATP synthase, and electron transport 21 chain (Fig. 3) were considered part of a core, constitutive proteome, independent of substrate. 22 The refined, 2-DO theory considered only the resource investment in addition to the conserved, 23 core proteome. This theory lead to improved predictions of amino acid preference with the 24 minimal proteome investment theory but not the flux minimization theory (Fig. 6a, 6b). The 25 outlier amino acid in Fig. 6b was serine. Serine is catabolized via the L-serine dehydratase 26 enzyme which is O2-labile suggesting higher cell densities and lower O2 concentrations are 1 necessary for its functionality [56,57]. The predictive accuracy of the analysis improved to a 2 correlation of r 2 = 0.88 if serine data were excluded. 3 2-DO, considering complete oxidation of substrate and minimal proteome investment, 4 was extended to the other available substrates including organic acids and glucose. Analysis 5 applied the minimal proteome investment with conserved core proteome assumption and 6 considered both cellular energy production as well as the more complex biomass production 7 (Fig. 7a). The in silico analysis accurately predicted experimental, substrate preference for 8 amino acids, citrate, succinate, lactate, acetate, and glucose. Correlations had an r 2 of 0.94 and 9 0.73 for cellular energy and biomass production, respectively. The biomass simulations included 10 an aggregate amino acid substrate pool, which was not considered for cellular energy 11 simulations, and as anticipated, the aggregate amino acid pool greatly reduced the requirement 12 for enzymatic steps by negating de novo amino acid synthesis reactions (Fig. 7a, supplemental 13 material S14, S15). 14 The maximization of rate criterion was also tested with the additional substrates. The 15 analysis assumed optimal product yields on substrate were proportional to the optimal product 16 rates. The maximum rate criterion did not predict the experimental preference for organic acids 17 over glucose. In fact, the predicted preferences had negative correlations with the experimental 18 data (Fig. 7b). Additional optimizations and aggregate substrate simulations were considered 19 (supplemental material S16-S18). None outperformed the presented approach in terms of 20 accuracy and simplicity. 21

Discussion 22
P. aeruginosa preferentially consumes nonfermentable, lower energy substrates, such as 23 acetate over glucose. This ecological strategy has enabled its broad, global distribution in the 24 environment and medical niches including chronic, diabetic ulcers. The hierarchy of substrate 25 preferences for Pa 215 was: amino acids such as aspartate, followed by citrate, succinate, 1 lactate, acetate, and finally glucose. These preferences were also observed with P. aeruginosa 2 PAOI grown on CSP GLAS medium (supplemental material S10). Pa 215 maintained, 3 constitutively, core TCA cycle enzymes and regulated the abundance of the proteins required 4 for specific substrates as needed to convert the substrates into central metabolism 5 intermediates. Analysis using an in silico metabolic model determined the rCCR phenotype was 6 consistent with a multidimensional, resource utilization strategy where substrate preference was 7 based on minimizing the proteome investment required to completely oxidize the metabolite. 8 The rCCR phenotypes were not consistent with the commonly applied, systems biology 9 criterion, which maximizes growth rate [45][46][47].  in consortia based on complementary rCCR and cCCR metabolisms could lead to emergent 25 properties such as enhanced biomass productivity based on enhanced resource acquisition and 1 better metabolic return on investment of scarce nutrients, ultimately leading to greater virulence. 2 Mitigating these consortia, through rational countermeasures, will require quantitative 3 knowledge of the metabolic organization which forms the bases of all virulence mechanisms. 4

Bacterial strain and cultivation 6
All experiments used P. aeruginosa str. 215, a clinical isolate obtained from a chronic wound at 7 the Southwest Regional Wound Care Center in Lubbock, Texas, USA [107]. Frozen stocks of P. 8 aeruginosa 215 were prepared by growing cultures in 10 mL of 1/10 strength tryptone soy broth 9 (TSB) at 37°C with shaking (150 rpm), adding 3 mL of 20% glycerol, mixing well, and making 10 aliquots of 1 mL volumes then stored until use at -80°C. 11 Frozen stocks were plated on tryptic soy agar (TSA) at 37°C for 12 h, five colonies were picked 12 to inoculate 10 mL of Clostridium, Staphylococcus, Pseudomonas (CSP) medium in culture 13 tubes. CSP is a chemically defined medium developed to support the growth of P. aeruginosa, 14 Staphylococcus aureus, and Clostridium perfringens as monocultures or consortia [37]. CSP 15 consists of 1.7 g/L Yeast Nitrogen Base without Amino Acids and Ammonium Sulfate (BD 16 Difco™), 0.7 g/L sodium citrate, 0.1 g/L EDTA tetrasodium salt, 100 mL MEM Non-Essential AA 17 Solution (Thermo Fisher Scientific/Life Technologies), 50 mL MEM AA Solution (Thermo 18 Fisher Scientific/Life Technologies), 4.74 g/L KH2PO4, 8.208 g/L Na2HPO4, 0.147 g/L glutamine, 19 2.00 ug/L B12, 2.80 mg/L FeSO4·7H2O, 1.20E-5 g/L CoCl2·6H2O, and 0.02 g/L of each the 20 following nucleosides: adenine, uracil, cytosine, guanine. For CSP supplemented with one or 21 more organic acids, 22 mM of each organic acid specified was added: CSP G, CSP GL, CSP 22 GA, CSP GLA, and CSP GLAS were supplemented with glucose, glucose and lactate, glucose 23 and acetate, glucose and lactate and acetate, and glucose and lactate and acetate and 1 succinate, respectively. 2 A total of three culturing tubes containing 10 mL of CSP were each inoculated with about five 3 colonies from the TSA plates, incubated at 37°C with shaking at 150 rpm (tubes were placed at 4 a 45⁰ angle in the shaker to increase mixing) and grown until the cultures reach an OD600 of 0.5. 5 1 mL of each culture was then added to 49 mL of fresh CSP medium in 250 mL baffled flasks 6 giving a culture volume to flask volume ratio of 1:5 and an OD600 reading of 0.010. The baffled 7 flasks were capped with gas permeable foam lids and incubated at 37°C with shaking at 150 8 rpm. Sampling occurred about every hour during the first 12 h and less frequently afterwards. 9

Culture sampling 10
Samples were drawn from each flask for OD600, pH, amino acid, and carbon metabolite 11 measurements. An aliquot of 1.5 mL of culture was collected at each sampling, cells were 12 separated from the supernatant using centrifugation at 7000 rpm for 10 min (Eppendorf 5415D 13 microcentrifuge). Supernatants were then filtered using 0.22 µm syringe filters prior to being 14 stored at -20°C. 15 At each sampling, a volume of culture was collected for OD600 measurement. This volume was 16 discarded after measurement. OD600 readings were blanked with fresh CSP and samples were 17 diluted, if necessary, to keep OD600 measurements ≤ 0.30. 18

Organic acid and sugar analyses 19
HPLC analysis of select carbon metabolites including glucose and organic acids was performed 20 with an Agilent 1200 series HPLC equipped with a variable wavelength detector (VWD) and a 21 refractive index detector (RID) and an Aminex HPX-87H ion exclusion column, 300 mm x 7.8 22 mm. A mobile phase of 5 mM H2SO4 was run at a flow rate of 0.6 mL/min for 25 min/injection. A 23 volume of 200 µL of sample was added to an HPLC vial with 200 µL of an internal standard of 1 24 g/L fucose dissolved in 10 mM H2SO4. Samples were then loaded into an autosampler, and 1 each was injected twice for a total of two technical replicates for each of the three biological 2 replicates for each time point sampled. 3

Amino acid analysis 4
HPLC analysis of amino acids was performed with an Agilent 1100 series equipped with a diode 5 array detector (DAD) and a ZORBAX Eclipse XDB-C18 column, 4.6 mm ID x 250 mm (5 µm) 80 6 Å. This setup was used with the Agilent protocol for HPLC analysis of amino acids [108]. 7

Cell dry weight measurement 8
A correlation curve between OD600 and grams of cell dry weight (g CDW) per liter was 9 constructed. 5 mL aliquots of P. aeruginosa culture harvested in mid-exponential growth phase 10 diluted to a range of densities were dried at 80⁰C for 24 h in aluminum drying pans and 11 Toolbox (https://opencobra.github.io/cobratoolbox/stable/cite.html) in MATLAB using the Gurobi 10 optimization program (http://www.gurobi.com) (supplemental material S11, S19). Carbon and O2 11 limitations were modeled by setting the carbon (5 mmol/g/h) or O2 (20 mmol/g/h) uptake rates, 12 respectively, for each of the examined carbon sources and maximizing the production of 13 biomass or cellular energy (i.e. quantified as the number of ATP bonds hydrolyzed). Enzyme The algorithms can be found in supplemental material S19. The authors declare no competing interests.   The experimental substrate preference was: 1. amino acids, 2. citrate, 2. succinate, 3. lactate, 4. 7 acetate, and 5. glucose. a. Two-dimensional, substrate preference predictions for cellular 8 energy production and biomass production. Simulations used the minimal proteome 9 investment proxy and refined, core proteome theory. See text for details. b. Predicted substrate 10 preference based on the maximization of rate criterion for cellular energy production and 11 biomass production. In silico product yield on substrate was assumed proportional to rate.  Cultures had two exponential growth phases with different specific growth rates ( 1 ,  2 ). X = dried biomass concentration (g/L). Initial culture pH was 7.0. C:N = ratio of total moles of carbon to total moles of nitrogen in the medium. See text for more details. *The specific growth rate during the first exponential growth phase was not significantly different between condition G and the four other conditions (p-values all > 0.05 using a paired two tailed t-test). †Carbon limited medium. All specific growth rates were calculated from three biological replicates during exponential growth.   The experimental substrate preference was: 1. amino acids, 2. citrate, 2. succinate, 3. lactate, 4. 4 acetate, and 5. glucose. a. Two-dimensional, substrate preference predictions for cellular 5 energy production and biomass production. Simulations used the minimal proteome 6 investment proxy and refined, core proteome theory. See text for details. b. Predicted substrate 7 preference based on the maximization of rate criterion for cellular energy production and 8 biomass production. In silico product yield on substrate was assumed proportional to rate. 9 10 11