Optimal Protein Allocation Controls the Inhibition of GltA and AcnB in Neisseria gonorrhoeae

Neisseria gonorrhea (Ngo) is a major concern for global public health due to its severe implications for reproductive health. Understanding its metabolic phenotype is crucial for comprehending its pathogenicity. Despite Ngo’s ability to encode TCA cycle proteins, GltA and AcnB, their activities are notably restricted. To investigate this phenomenon, we used the iNgo_557 metabolic model and incorporated a constraint on total cellular protein content. Our results indicate that low cellular protein content severely limits GltA and AcnB activity, leading to a shift towards acetate overflow for ATP production, which is more efficient in terms of protein usage. Surprisingly, increasing cellular protein content alleviates this restriction on GltA and AcnB and delays the onset of acetate overflow, highlighting protein allocation as a critical determinant in understanding Ngo’s metabolic phenotype. These findings underscore the significance of Ngo’s metabolic adaptation in light of optimal protein allocation, providing a blueprint to understand Ngo’s metabolic landscape.


Introduction
Neisseria gonorrhoeae (Ngo), the causative agent of gonorrhea, is one of the most prevalent sexually transmitted infections globally, affecting approximately 87 million people each year, including 1.6 million within the United States (Kreisel et al. 2021).This pathogen's persistence and increasing resistance to antibiotics pose a substantial public health challenge (Chinemerem Nwobodo et al. 2022).Ngo's ability to persist in the human host and to evade the immune system emphasizes the need to understand its unique metabolic phenotype.
Despite having the ability to encode a complete tricarboxylic acid (TCA) cycle, previous studies revealed Ngo's puzzling phenotype of restricting crucial TCA cycle enzymes, such as GltA and AcnB (Potter and Criss 2023a).Additionally, it does not have the ability to encode phosphofructokinase (PFK).To probe deeper into metabolic anomaly in the TCA cycle, the application of systems biology, particularly, genome-scale metabolic models (GEMs), provides a promising avenue, since it has been proven to answer unusual bacterial phenomena (Chowdhury et al. 2023).Previous works has shown that integrating protein allocation constraints within GEMs can significantly enhance our understanding of bacterial metabolism, reflecting more accurately the in vivo conditions that might influence metabolic functionality (O'Brien et al. 2013;Chen and Nielsen 2019;Chowdhury, Alsiyabi and Saha 2022).
Thus, in this study, we applied a total protein capacity constraint to the recently published iNgo_557 GEM to test the hypothesis that limitations in protein availability affect the activity of TCA cycle enzymes -GltA and AcnB.Our findings indicate that limited protein resources are crucial in suppressing the function of GltA and AcnB enzymes.Notably, by adjusting protein levels within the model, we observed that increased protein availability could relax these restrictions, allowing higher flux through GltA and AcnB and reducing the organism's reliance on less efficient metabolic pathways such as acetate overflow.These results not only elucidate the significant impact of protein allocation on metabolic pathways but also provide valuable insights into potential metabolic vulnerabilities of Ngo, thus enhancing our understanding on how Ngo manages to sustain itself under the nutrient-limited and immune-rich environment of the human host.

Initial Model Setup, Data Integration, and Reaction Bounds Modification
The study utilized the previously published Ngo genome-scale metabolic model, iNgo_557 (Potter et al. 2023).Initial preparations involved adding pathway information and making the model irreversible (https://github.com/ssbio/Gc_protein_constrained).Following that filtering the model to identify reactions with gene-protein-reaction (GPR) associations.For these reactions, enzyme kinetics (  ) were sourced predominantly using DLKcat (Li et al. 2022a), which provides estimated enzyme turnover rates based on homology.Molecular weights (MW) for these enzymes were also calculated using amino acid sequences derived from the KEGG database.Additionally, to accurately reflect the in vivo environment, reaction bounds were adjusted according to minimal defined medium (MDM) (Potter et al. 2023), ensuring that the simulated nutrient availabilities aligned with the physiological conditions under which N. gonorrhoeae thrives.

Handling Data Gaps with Monte Carlo Simulations and Model Selection
Due to the unavailability of   values for 373 reactions and MW data for 12 reactions, Monte Carlo simulations were conducted to stochastically generate 100 ensemble models (Figure 1A and 1B).These simulations predicted missing   and MW values, filling the gaps necessary for a comprehensive metabolic analysis (Figure 1C and 1D).Each of the 100 simulations provided a unique set of   and MW values for the missing data, which were then integrated into the iNgo_557 model to create 100 ensemble GEMs.For each of these 100 modified GEMs, the cellular ) was calculated.Finally, the model with the lowest overall Φ value was selected for further analysis, as it represented the most protein-efficient configuration under protein-limited conditions.

Flux Balance Analysis and Comparative Metabolic Flux Profiling
Next, we employed a sequential optimization strategy within the confines of Flux Balance Analysis (FBA).Initially, the model was tasked with maximizing biomass production.Once the optimal biomass growth rate was established, we fixed this growth rate as a constraint, to maximize acetate production under different glucose uptake while generating fluxes for all the reactions in the model.Moreover, the ATP synthase (ATPS) reaction was not connected to the TCA cycle through the succinate dehydrogenase (SDH) which was resolved by adding an additional constraint.This configuration was employed for the two distinct protein contents, Φ = 0.46   (lower protein content) and Φ = 0.60   (higher protein content) to assess the flux distribution in different pathways.Following is the mathematical formulation of this optimization problem: = 0.65 ×   (4) Here,  and  are the sets of metabolites and reactions in the model, respectively.  is the stoichiometric coefficient of metabolite  in reaction ,  ′ is the set of reactions for which GPR is available, Φ is total cellular protein content, and   is the flux value of reaction .Parameters   and   denote the minimum and maximum allowable fluxes for reaction , respectively.
Subsequent analysis involved combining the flux distributions for each metabolic pathway.This comparative analysis between the two protein conditions highlighted significant metabolic shifts in revealed critical insights into the metabolic adjustments of Ngo, especially concerning the TCA cycle and acetate overflow pathways.

Computational Tools
Simulations were conducted using COBRApy, a Python library designed for computational modeling of biological networks, with Gurobi solver for the linear programming tasks.The entire process from data integration to simulation was automated using custom Python scripts, enhancing reproducibility and computational efficiency.

Limited Protein Availability and Its Impact on TCA Cycle Restriction
In this work, we presented a protein-constrained genome-scale metabolic (pcGEM) model of Ngo to test the hypothesis that optimal proteome allocation influences its unique metabolic phenotype.
This includes suppression of the first three steps of the TCA cycle, despite its ability to encode the necessary proteins (Potter and Criss 2023b).To reconstruct the pcGEM, we selected a recently published GEM of Ngo, iNgo_557 (D. et al. 2023).We incorporated the pathway information for all the reactions (Dataset S1) and subsequently added unit-flux protein cost data for reactions that has GPR in the model (Chen and Nielsen 2019).The protein cost unit-flux is the ratio of molecular weight (MW) and enzyme turnover rate (  ) of the enzyme that catalyzes a reaction.Due to the scarcity of experimental   of Ngo, we used  (Li et al. 2022b) to predict it (Dataset S2).
For MW calculations, we used amino acid sequences (Dataset S3).Further details can be found in the Materials and Methods section.
As GltA and AcnB of the TCA cycle are severely restricted and TCA cycle is associated with energy generation, we first determined the ATP producing modules of Ngo.ATP can be produced

Increased Protein Availability Send more Flux through GltA and AcnB
To further test the hypothesis that proteome allocation plays the decisive role in restricting the first three steps of the TCA cycle, we increased the Φ of Ngo to 0.60

𝑔𝑔 𝑔𝑔𝑔𝑔𝑔𝑔𝑀𝑀
. If the hypothesis is accurate, three aspects will be noticed in the Ngo metabolism.First, the onset of acetate secretion will be delayed, as more protein is available for investment in the TCA cycle.Second, the bulk of the ATP will be generated from the TCA cycle and the contribution of substrate-level phosphorylation will remain similar.Third, the first three steps of the TCA cycle will carry much higher fluxes compared to the low Φ.

Figure 1 .
Figure 1.Generation of missing enzyme turn-over rate (  ) and molecular weight (MW) for missing enzymes.A) Distribution of   for known enzymes.B) Distribution of MW for known enzymes.C) Generated   for 373 missing enzymes for 100 ensembles run using Monte Carlo simulation.D) Generated MW for 12 missing enzymes for 100 ensembles run using Monte Carlo simulation.
in three different modules: substrate-level phosphorylation of glucose, acetate overflow metabolism, and TCA cycle.Reactions associated with each module that are carrying flux can be found in Dataset S4.Next, we require the cellular protein content for Ngo.Since this data is also unavailable, we minimized the total protein content, such that the predicted growth rate matches closely to experimental observation, 0.73 ℎ −1(D.et al. 2023).From the optimization problem, the minimum protein content (Φ) found was 0.46   .This value is in par with the protein content of E. coli, which is 0.55   (B., M. and E. 2021) Additionally, we connected ATP synthase (ATPS) with the TCA cycle with an additional constraint, ATPS can carry 65% of the flux of succinate dehydrogenase (SDH).