Characterization of an experimental miniature bioreactor for cellular perturbation studies

AUC-PR is a More Informative Metric for Assessing the Biological Relevance of In Silico Cellular Perturbation Prediction Models

In silico perturbation models, computational methods which can predict cellular responses to perturbations, present an opportunity to reduce the need for costly and time-intensive in vitro experiments. Many recently proposed models predict high-dimensional cellular responses, such as gene or protein expression to perturbations such as gene knockout or drugs. However, evaluating in silico performance has largely relied on metrics such asR 2 , which assess overall prediction accuracy but fail to capture biologically significant outcomes like the identification of differentially expressed genes. In this study, we present a novel evaluation framework that introduces the AUC-PR metric to assess the precision and recall of DE gene predictions. By applying this framework to both single-cell and pseudo-bulked datasets, we systematically benchmark simple and advanced computational models. Our results highlight a significant discrepancy betweenR 2 and AUC-PR, with models achieving highR 2 values but struggling to identify Differentially expressed genes accurately, as reflected in their low AUC-PR values. This finding underscores the limitations of traditional evaluation metrics and the importance of biologically relevant assessments. Our framework provides a more comprehensive understanding of model capabilities, advancing the application of computational approaches in cellular perturbation research.

Metabolic control analysis enables rational improvement of E. coli L-tryptophan producers but methylglyoxal formation limits glycerol-based production

BACKGROUND

Although efficient L-tryptophan production using engineered Escherichia coli is established from glucose, the use of alternative carbon sources is still very limited. Through the application of glycerol as an alternate, a more sustainable substrate (by-product of biodiesel preparation), the well-studied intracellular glycolytic pathways are rerouted, resulting in the activity of different intracellular control sites and regulations, which are not fully understood in detail. Metabolic analysis was applied to well-known engineered E. coli cells with 10 genetic modifications. Cells were withdrawn from a fed-batch production process with glycerol as a carbon source, followed by metabolic control analysis (MCA). This resulted in the identification of several additional enzymes controlling the carbon flux to L-tryptophan.

RESULTS

These controlling enzyme activities were addressed stepwise by the targeted overexpression of 4 additional enzymes (trpC, trpB, serB, aroB). Their efficacy regarding L-tryptophan productivity was evaluated under consistent fed-batch cultivation conditions. Although process comparability was impeded by process variances related to a temporal, unpredictable break-off in L-tryptophan production, process improvements of up to 28% with respect to the L-tryptophan produced were observed using the new producer strains. The intracellular effects of these targeted genetic modifications were revealed by metabolic analysis in combination with MCA and expression analysis. Furthermore, it was discovered that the E. coli cells produced the highly toxic metabolite methylglyoxal (MGO) during the fed-batch process. A closer look at the MGO production and detoxification on the metabolome, fluxome, and transcriptome level of the engineered E. coli indicated that the highly toxic metabolite plays a critical role in the production of aromatic amino acids with glycerol as a carbon source.

CONCLUSIONS

A detailed process analysis of a new L-tryptophan producer strain revealed that several of the 4 targeted genetic modifications of the E. coli L-tryptophan producer strain proved to be effective, and, for others, new engineering approaches could be derived from the results. As a starting point for further strain and process optimization, the up-regulation of MGO detoxifying enzymes and a lowering of the feeding rate during the last third of the cultivation seems reasonable.

Metabolic control analysis of L-tryptophan producing Escherichia coli applying targeted perturbation with shikimate

L-tryptophan production from glycerol with Escherichia coli was analysed by perturbation studies and metabolic control analysis. The insertion of a non-natural shikimate transporter into the genome of an Escherichia coli L-tryptophan production strain enabled targeted perturbation within the product pathway with shikimate during parallelised short-term perturbation experiments with cells withdrawn from a 15 L fed-batch production process. Expression of the shikimate/H⁺-symporter gene (shiA) from Corynebacterium glutamicum did not alter process performance within the estimation error. Metabolic analyses and subsequent extensive data evaluation were performed based on the data of the parallel analysis reactors and the production process. Extracellular rates and intracellular metabolite concentrations displayed evident deflections in cell metabolism and particularly in chorismate biosynthesis due to the perturbations with shikimate. Intracellular flux distributions were estimated using a thermodynamics-based flux analysis method, which integrates thermodynamic constraints and intracellular metabolite concentrations to restrain the solution space. Feasible flux distributions, Gibbs reaction energies and concentration ranges were computed simultaneously for the genome-wide metabolic model, with minimum bias in relation to the direction of metabolic reactions. Metabolic control analysis was applied to estimate elasticities and flux control coefficients, predicting controlling sites for L-tryptophan biosynthesis. The addition of shikimate led to enhanced deviations in chorismate biosynthesis, revealing a so far not observed control of 3-dehydroquinate synthase on L-tryptophan formation. The relative expression of the identified target genes was analysed with RT-qPCR. Transcriptome analysis revealed disparities in gene expression and the localisation of target genes to further improve the microbial L-tryptophan producer by metabolic engineering.

Metabolic control analysis of L-tryptophan production with Escherichia coli based on data from short-term perturbation experiments

E. coli strain NT1259 /pF112aroFBL_kan was able to produce 14.3 g L^-1 L-tryptophan within 68 h in a fed-batch process from glycerol on a 15 L scale. To gain detailed insight into metabolism of this E. coli strain in the fed-batch process, a sample of L-tryptophan producing cells was withdrawn after 47 h, was separated rapidly and then resuspended in four parallel stirred-tank bioreactors with fresh media. Four different carbon sources (glucose, glycerol, succinate, pyruvate) were supplied individually with varying feeding rates within 19 min and the metabolic reactions of the cells in the four parallel reactors were analyzed by quantification of extracellular and intracellular substrate, product and metabolite concentrations. Data analysis allowed the estimation of intracellular carbon fluxes and of thermodynamic limitations concerning intracellular concentrations and reaction energies. Carbon fluxes and intracellular metabolite concentrations enabled the estimation of elasticities and flux control coefficients by applying metabolic control analysis making use of a metabolic model considering 48 enzymatic reactions and 56 metabolites. As the flux control coefficients describe connections between enzyme activities and metabolic fluxes, they reveal genetic targets for strain improvement. Metabolic control analysis of the recombinant E. coli cells withdrawn from the fed-batch production process clearly indicated that (i) the supply of two precursors for L-tryptophan biosynthesis, L-serine and phosphoribosyl-pyrophosphate, as well as (ii) the formation of aromatic byproducts and (iii) the enzymatic steps of igps and trps2 within the L-tryptophan biosynthesis pathway have major impact on fed-batch production of L-tryptophan from glycerol and should be the targets for further strain improvements.

Intracellular product recycling in high succinic acid producing yeast at low pH

BACKGROUND

The metabolic engineering of Saccharomyces cerevisiae for the production of succinic acid has progressed dramatically, and a series of high-producing hosts are available. At low cultivation pH and high titers, the product transport can become bidirectional, i.e. the acid is reentering the cell and is again exported or even catabolized. Here, a quantitative approach for the identification of product recycling fluxes is developed.

RESULTS

The metabolic flux distributions at two time-points of the fermentation process were analyzed. ¹³C labeled succinic acid was added to the extracellular space and intracellular enrichments were measured and subsequently used for the estimation of metabolic fluxes. The labeling was introduced by a labeling switch experiment, leading to an immediate labeling of about 85% of the acid while keeping the total acid concentration constant. Within 100 s significant labeling enrichment of the TCA cycle intermediates fumarate, iso-citrate and α-ketoglutarate was observed, while no labeling was detected for malate and citrate. These findings suggest that succinic acid is rapidly exchanged over the cellular membrane and enters the oxidative TCA cycle. Remarkably, in the oxidative direction malate ¹³C enrichment was not detected, indicating that there is no flux going through this metabolite pool. Using flux modeling and thermodynamic assumptions on compartmentation it was concluded that malate must be predominantly cytosolic while fumarate and iso-citrate were more dominant in the mitochondria.

CONCLUSIONS

Adding labeled product without changing the extracellular environment allowed to quantify intracellular metabolic fluxes under high producing conditions and identify product degradation cycles. In the specific case of succinic acid production, compartmentation was found to play a major role, i.e. the presence of metabolic activity in two different cellular compartments lead to intracellular product degradation reducing the yield. We also observed that the flux from glucose to succinic acid branches at two points in metabolism: (1) At the level of pyruvate, and (2) at cytosolic malate which was not expected.

Perturbation Experiments: Approaches for Metabolic Pathway Analysis in Bioreactors

In the last decades, targeted metabolic engineering of microbial cells has become one of the major tools in bioprocess design and optimization. For successful application, a detailed knowledge is necessary about the relevant metabolic pathways and their regulation inside the cells. Since in vitro experiments cannot display process conditions and behavior properly, process data about the cells' metabolic state have to be collected in vivo. For this purpose, special techniques and methods are necessary. Therefore, most techniques enabling in vivo characterization of metabolic pathways rely on perturbation experiments, which can be divided into dynamic and steady-state approaches. To avoid any process disturbance, approaches which enable perturbation of cell metabolism in parallel to the continuing production process are reasonable. Furthermore, the fast dynamics of microbial production processes amplifies the need of parallelized data generation. These points motivate the development of a parallelized approach for multiple metabolic perturbation experiments outside the operating production reactor. An appropriate approach for in vivo characterization of metabolic pathways is presented and applied exemplarily to a microbial L-phenylalanine production process on a 15 L-scale.

Identification of informative metabolic responses using a minibioreactor: a small step change in the glucose supply rate creates a large metabolic response in Saccharomyces cerevisiae

In this study, a previously developed mini-bioreactor, the Biocurve, was used to identify an informative stimulus-response experiment. The identified stimulus-response experiment was a modest 50% shift-up in glucose uptake rate (qGLC) that unexpectedly resulted in a disproportionate transient metabolic response. The 50% shift-up in qGLC in the Biocurve resulted in a near tripling of the online measured oxygen uptake (qO2) and carbon dioxide production (qCO2) rates, suggesting a considerable mobilization of glycogen and trehalose. The 50% shift-up in qGLC was subsequently studied in detail in a conventional bioreactor (4 l working volume), which confirmed the results obtained with the Biocurve. Especially relevant is the observation that the 50% increase in glucose uptake rate led to a three-fold increase in glycolytic flux, due to mobilization of storage materials. This explains the unexpected ethanol and acetate secretion after the shift-up, in spite of the fact that after the shift-up the qGLC was far less than the critical value. Moreover, these results show that the correct in vivo fluxes in glucose pulse experiments cannot be obtained from the uptake and secretion rates only. Instead, the storage fluxes must also be accurately quantified. Finally, we speculate on the possible role that the transient increase in dissolved CO2 immediately after the 50% shift-up in qGLC could have played a part in triggering glycogen and trehalose mobilization.

An industrial perspective on bioreactor scale-down: what we can learn from combined large-scale bioprocess and model fluid studies

For industrial bioreactor design, operation, control and optimization, the scale-down approach is often advocated to efficiently generate data on a small scale, and effectively apply suggested improvements to the industrial scale. In all cases it is important to ensure that the scale-down conditions are representative of the real large-scale bioprocess. Progress is hampered by limited detailed and local information from large-scale bioprocesses. Complementary to real fermentation studies, physical aspects of model fluids such as air-water in large bioreactors provide useful information with limited effort and cost. Still, in industrial practice, investments of time, capital and resources often prohibit systematic work, although, in the end, savings obtained in this way are trivial compared to the expenses that result from real process disturbances, batch failures, and non-flyers with loss of business opportunity. Here we try to highlight what can be learned from real large-scale bioprocess in combination with model fluid studies, and to provide suitable computation tools to overcome data restrictions. Focus is on a specific well-documented case for a 30-m(3) bioreactor. Areas for further research from an industrial perspective are also indicated.

A noninvasive technique for the measurement of the energetic state of free-suspension mammalian cells

A perfusion small-scale bioreactor allowing on-line monitoring of the cell energetic state was developed for free-suspension mammalian cells. The bioreactor was designed to perform in vivo nuclear magnetic resonance (NMR) spectroscopy, which is a noninvasive and nondestructive method that permits the monitoring of intracellular nutrient concentrations, metabolic precursors and intermediates, as well as metabolites and energy shuttles, such as ATP, ADP, and NADPH. The bioreactor was made of a 10-mm NMR tube following a fluidized bed design. Perfusion flow rate allowing for adequate oxygen supply was found to be above 0.79 mL min(-1) for high-density cell suspensions (10(8) cells). Chinese hamster ovary (CHO) cells were studied here as model system. Hydrodynamic studies using coloration/decoloration and residence time distribution measurements were realized to perfect bioreactor design as well as to determine operating conditions bestowing adequate homogeneous mixing and cell retention in the NMR reading zone. In vivo (31)P NMR was performed and demonstrated the small-scale bioreactor platform ability to monitor the cell physiological behavior for 30-min experiments.

CO2 production as an indicator of biofilm metabolism

Biofilms are important in aquatic nutrient cycling and microbial proliferation. In these structures, nutrients like carbon are channeled into the production of extracellular polymeric substances or cell division; both are vital for microbial survival and propagation. The aim of this study was to assess carbon channeling into cellular or noncellular fractions in biofilms. Growing in tubular reactors, biofilms of our model strain Pseudomonas sp. strain CT07 produced cells to the planktonic phase from the early stages of biofilm development, reaching pseudo steady state with a consistent yield of approximately 10(7) cells.cm(-2).h(-1) within 72 h. Total direct counts and image analysis showed that most of the converted carbon occurred in the noncellular fraction, with the released and sessile cells accounting for <10% and <2% of inflowing carbon, respectively. A CO(2) evolution measurement system (CEMS) that monitored CO(2) in the gas phase was developed to perform a complete carbon balance across the biofilm. The measurement system was able to determine whole-biofilm CO(2) production rates in real time and showed that gaseous CO(2) production accounted for 25% of inflowing carbon. In addition, the CEMS made it possible to measure biofilm response to changing environmental conditions; changes in temperature or inflowing carbon concentration were followed by a rapid response in biofilm metabolism and the establishment of new steady-state conditions.