Metabolic Flux Analysis in Mammalian Cell Culture: Uncovering Dynamic Metabolism

In mammalian cell culture system, Metabolic Flux Analysis (MFA) is of great value because of its unique dynamic analytical ability. MFA can systematically reveal the metabolic response mechanism of cells to nutrient supply (such as glucose and glutamine), environmental stress (hypoxia and oxidative stress) and growth factor stimulation. By analyzing the distribution of metabolic flux, researchers can accurately locate the dynamic regulatory nodes of energy metabolism (such as glycolysis, oxidative phosphorylation), biomacromolecule synthesis (lipid, protein) and disease-related pathways (such as tumor Warburg effect, immune cell metabolic reprogramming), which provides a key basis for disease mechanism analysis, cell therapy process optimization and bioreactor design.

Metabolic flux describes the dynamics of cellular metabolism (Lagziel S et al., 2019).

Key technical framework of MFA in mammalian cells

Isotope labeling strategy design

• Selection of labeling substrates: According to the characteristics of cell metabolism, [U-13C] glucose, [13C5] glutamine or mixed labeling substrates (such as the combination of [1,2-13C] glucose and [3-13C] lactic acid) are usually used to cover the central carbon metabolism and branching pathways.
• Dynamic labeling experiment: The isotope transfer dynamics in TCA cycle, pentose phosphate pathway and other key networks are tracked by time series sampling (such as 0/6/12/24 hours), and the labeling efficiency is improved by combining the pulse-chase strategy.

Multidimensional quantitative analysis of metabolites

Technical platform:

• High-resolution mass spectrometry (HRMS): Q-Exactive Orbitrap or TOF-MS is used to realize the full-spectrum scanning of metabolites, and the detection limit can reach pmol level, which supports the discrimination of isotopic isomers (such as γ -phosphate labeled isomers of ATP).
• Multidimensional NMR techniques: such as 1H-13C HSQC spectrum, can analyze the structure and marker sites of metabolites, which is suitable for dynamic monitoring of extracellular metabolites.

Data standardization

• introduce internal standard (such as D4- succinic acid) to correct instrument drift, and match fragment ion spectrum through metabolite database (HMDB, METLIN) to ensure quantitative accuracy.

Metabolic network modeling and flux calculation

Network topology construction

• Based on genome-scale metabolic model (such as Recon3D), cell-specific metabolic network is constructed by integrating transcriptome/proteome data, with emphasis on reversible reactions and branching pathways (such as PPP and glycolysis shunting).

Calculation algorithm optimization

• Software tools: INCA (based on EMU framework) is used to estimate the unsteady flux, and Monte Carlo sensitivity analysis is carried out with 13CFLUX2; COBRA toolbox can be used for constraint optimization of large-scale networks.
• Machine learning integration: neural network algorithm (such as DeepFlux) is introduced to process high-dimensional isotope labeling data to improve the flux resolution of complex metabolic networks.

Metabolic network used for metabolic flux analysis (Niklas J et al., 2010).

Technical challenges and coping strategies of MFA in mammalian cells

1. Complexity and modeling problems of metabolic networks.

The metabolic network of mammalian cells has the characteristics of high redundancy and dynamic regulation, which are as follows:

• Pathway crossing and feedback regulation: such as carbon shunting of glycolysis and pentose phosphate pathway (PPP), coordinated regulation of TCA cycle and lipid synthesis, and dynamic regulation of metabolic flux by mTOR signaling pathway, all increase the complexity of network topology.

• Subcellular compartmentalization effect: it is difficult to quantitatively characterize the inter-compartment metabolite transport in mitochondria, cytoplasm and peroxisome (such as NADH/NAD+ shuttle system), which leads to uncertainty of flux distribution.
• Cell heterogeneity interference: the difference of cell cycle and the existence of phenotypic differentiation of subsets in the culture system (such as the metabolic preference of tumor stem cells and differentiated cells) may mask the true flux distribution.

Coping strategies:

• The cell type-specific metabolic model was constructed, and the network topology was optimized by integrating single cell transcriptome data.
• Compartment metabolic model (such as MITOsym platform) and isotope labeling were used to track the exchange of metabolites between compartments.

2. Limitation of isotope labeling efficiency

The dynamic range and accuracy of isotope labeling directly affect the flux resolution ability. The main problems include:

• Metabolic state dependence of cells: the rapid consumption of labeling substrates by highly proliferating cells (such as CHO cells) may lead to unstable labeling, while the isotope enrichment efficiency of quiescent cells (such as primary hepatocytes) is low.
• Microenvironment influence: The local metabolic heterogeneity caused by oxygen/nutrient gradient in three-dimensional culture system or organ-like body may cause spatial deviation of isotope marker distribution.
• Isotope dilution effect: the dilution of labeled signals by endogenous metabolites (such as amino acids released by autophagy) needs to be compensated by the metabolite pool size correction model.

Coping strategies:

• Development of mixed isotope labeling strategy (such as using [U-13C] glucose and [2H] leucine at the same time), combined with multi-element tracing to improve labeling coverage.
• Unsteady MFA(INST-MFA) algorithm is applied to dynamically fit the marked time series data.

3. Refined management requirements of experimental design

Improper selection of experimental parameters may lead to the decrease of data signal-to-noise ratio;

• Sampling time window: it is necessary to balance the establishment of isotope steady state (usually 5-7 cell doubling time) and the stability of cell state (such as avoiding stress reaction induced by nutrient depletion).
• The bottleneck of metabolite detection: the synchronous extraction and quantitative technology of polar metabolites (such as ATP and NADPH) and hydrophobic metabolites (such as lipids) is still immature, so the strategy of fractional extraction combined with LC-MS/GC-MS should be adopted.
• Challenge of data standardization: Batch effect (such as medium composition fluctuation) needs to be controlled by introducing isotope internal standard (such as 13C6- glucose -6- phosphate) and experimental repeated design (n≥3).

Coping strategies:

• Optimization of sampling time point design based on metabolite turnover prediction tools (such as Isodyn).
• Establish a standardized metabolite extraction process (such as MeRy-BM scheme) to ensure the comparability of cross-experimental data.

4. The bottleneck of computing resources and algorithm efficiency

Computational challenges brought by mammalian genome-scale metabolic models (such as Recon3D containing more than 3,000 reactions);

• High-dimensional nonlinear optimization: the flux space search of large-scale metabolic network is easy to fall into local optimal solution, so it is necessary to use global optimization algorithm (such as genetic algorithm) or parallel computing architecture to accelerate the solution.
• Multiomics data integration: the introduction of transcriptome/proteome constraints (such as GECKO method) can improve the model accuracy, but significantly increase the computational complexity.
• Uncertainty quantification demand: Monte Carlo simulation of flux confidence interval needs tens of thousands of iterations, which requires extremely high computational resources.

Coping strategies:

• Adopt cloud computing platform (such as AWS metabolic analysis suite) to realize distributed computing.
• Develop dimension reduction algorithms (such as flux compression technology based on principal component analysis) to reduce the degree of freedom of the model.
• Integrate machine learning frameworks (such as deep reinforcement learning) to optimize the flux search path.

Application of MFA in mammalian cell culture

Biological production and biotechnology

In the context of biological production of mammalian cells, MFA can be used to optimize the production of therapeutic protein, vaccines and other biological products. By describing metabolic pathways involved in energy production, amino acid synthesis and glycosylation, researchers can identify metabolic bottlenecks that limit production.

• PPP is an important part of cell metabolism, involving the production of nucleotides, amino acid precursors and reducing equivalents (such as NADPH), which are essential for cell biosynthesis, antioxidant stress and other physiological functions. It is found that, besides the classic oxidative and non-oxidative branches, there may be some undisclosed reactions in PPP, especially in some cancer cells or rapidly proliferating cells, and these hidden reactions may be related to transketolase-like protein 1(TKTL1). TKTL1 is considered as a possible catalyst for a new reaction in PPP, which involves the cleavage of xylulose 5- phosphate (X5P) into glyceraldehyde 3- phosphate (GAP) and acetyl coenzyme A(AcCoA). 13C-MFA analysis was used to prove whether TKTL1 participates in PPP in mammalian cells, and novel isotope labeling technology and GC-MS method were used to improve the flux resolution. CHO cell line was used in the experiment, and the labeling of metabolites in the cell was measured by GC-MS, and the mass isotope distribution of 3PG and PEP was obtained. The results show that the ratio of (M1+2*M2)/M3 measured in the experiment of [1-13C] glucose+[4,5,6-13C] glucose is low, suggesting that a large amount of glucose is metabolized by oxPPP, while other tracer combinations can effectively show the activity of TKTL1. Most of glucose is metabolized through glycolysis and oxidative phosphorylation, among which 55% glucose is metabolized through oxPPP, and part of it forms R5P, and then it is converted into X5P to activate TKTL1 reaction. The estimated flux of TKTL1 is 33 4%, which indicates that this pathway is very important for glucose metabolism. In addition, the addition of F6P tag data also improves the estimation accuracy of exchange flux in PPP. The study also revealed that oxPPP not only produced NADPH, but also promoted TKTL1 reaction in CHO cells on stationary phase, which may be used for anti-oxidative stress. Future work will further explore whether cytoplasmic acetyl-CoA produced by TKTL1 reaction is used for fatty acid synthesis, which may provide new insights into the relationship between glucose metabolism and lipogenesis (Ahn WS et al., 2016).

Cell line development and engineering

MFA can be used to design and engineer mammalian cell lines with higher productivity and stronger robustness. By understanding the metabolic requirements of specific recombinant protein production process, cell lines can be engineered to make better use of nutrients, reduce by-products and maintain cell health in long-term culture. These engineered cell lines can lead to a more efficient and lower-cost biological treatment process.

• HEK293 cell line has high transfection ability, which makes it the preferred host for recombinant protein expression. ECD-Her1 is a complex protein, which is usually used to develop anticancer vaccines. Researchers pay attention to how to increase the yield of ECD-Her1 by optimizing cell culture conditions and improving cell metabolic activity. In order to achieve this goal, studies show that adding sodium acetate (NaAc) can enhance the glycolysis and aerobic metabolic flux of cells, thus enhancing the synthesis of protein. Under the condition of different concentrations of NaAc, the production rate of heterologous protein has nothing to do with the exponential period and stable period of cell growth. Effect of NaAc on cell density: In exponential period, cell density is not affected by sodium acetate, but with the increase of NaAc concentration, the nutritional depth (that is, the nutritional level required for cell growth) gradually decreases. Protein (ECD-Her1) was produced at the beginning of exponential phase, and the productivity of ECD-Her1 was significantly improved in the presence of NaAc. Experiments show that cells consume glucose and secrete lactic acid in exponential phase, and when glucose is exhausted, cells begin to consume lactic acid, which is more obvious in the presence of NaAc. Through MFA, it was found that HEK293 cells showed a typical Warburg effect, that is, even in the presence of oxygen, cells met their metabolic needs by increasing glucose intake and lactic acid production. The addition of NaAc further strengthened this effect. The existence of NaAc increases the metabolic flux related to TCA cycle, which indicates that NaAc may promote the production of ECD-Her1. The extracellular flux analyzer (Seahorse XFe96) was used to monitor the metabolic activity of cells. The results showed that in the presence of NaAc, the glycolysis and respiratory flux of cells increased, indicating that the energy state of cells improved. Although the exact mechanism of NaAc improving ECD-Her1 production cannot be clearly determined, the research shows that NaAc may promote the synthesis of protein by acting as an energy source and participating in histone acetylation. Based on the experimental results, it is suggested that NaAc concentration of about 8 mM should be used in batch culture of HEK293 cells to maximize the yield of ECD-Her1 in protein (Pérez-Fernández BA et al., 2024).

Disease research and cancer metabolism

In disease research, especially in cancer metabolism, MFA provides an in-depth understanding of how cells adjust their metabolism to cope with the increased energy and synthesis demand of rapid cell division. Researchers can describe metabolic reprogramming in cancer, which may provide new therapeutic targets for metabolic intervention.

• TME consists of cancer cells, stromal cells (such as fibroblasts, immune cells, endothelial cells and pericytes) and extracellular matrix. The interaction between stromal cells and cancer cells can promote tumor growth, metastasis and resist immune response and treatment. Cancer-associated fibroblasts (CAF) are one of the main components of TME, which help tumor cells to acquire therapeutic resistance by secreting soluble factors. Exosomes (30-100nm in diameter) are small vesicles secreted by cells, which can carry biomolecules such as protein and RNA (including miRNA), mediate cell-to-cell communication and support tumor progression. In order to study the effect of exosomes on tumor cell metabolism, the transport flux of metabolites from CAF to cancer cells was predicted by Exo-MFA. The results show that exosomes can effectively restore the proliferation of PDAC cells in the medium lacking glutamine and phenylalanine, which proves the importance of exosomes in tumor cell metabolism. The effect of exocrine internalization on cell metabolism is instantaneous, and it is mainly limited by the internalization rate of exocrine and the release rate of goods. Exosomes can be quickly internalized by cancer cells and supplement necessary metabolic intermediates, thus helping cancer cells overcome the proliferation obstacles caused by nutritional deficiency. Metabolites after exocrine internalization quickly participate in the central carbon metabolism of cancer cells, and significantly improve the proliferation ability of cells by supplementing central carbon intermediates and amino acid precursors. In addition, the study also found that exosomes have a time-sensitive effect on glycolytic pathway and TCA cycle. Glycolysis-related amino acids and lactic acid increased steadily from 6 to 24 hours, while TCA cycle-related amino acids and metabolites showed periodic changes, which indicated that exosomes had dynamic regulation on cancer cell metabolism. In the medium lacking glutamine and phenylalanine, although the activity of PDAC cells decreased significantly at the initial stage, the activity of PDAC cells was significantly restored by supplementing exosomes. It is found that cancer cells quickly absorb exosomes within 24 hours, leading to the consumption of metabolites in exosomes. In glycolytic pathway, the supply of lactic acid can significantly regulate metabolic flux and affect the metabolism of cancer cells. In addition, exosomes can also provide intermediate products of TCA cycle and glutamine to support the energy production and growth of cancer cells (Achreja A et al., 2017).

Understanding cell stress response

Mammalian cells are often exposed to various stresses during culture, including nutrient deficiency, hypoxia and oxidative stress. MFA can monitor the changes of metabolic flux under different stress conditions. For example, under hypoxic conditions, cells may shift their metabolic activities to anaerobic pathways, such as glycolysis, while reducing oxidative phosphorylation.

• Glucose metabolism in endothelial cells plays an important role in many diseases, especially in atherosclerosis, cancer and Alzheimer's disease. Metabolic enzyme inhibitors (such as aldose reductase inhibitor Fidarestat, sex hormone precursor dehydroepiandrosterone DHEA and glutamine analogue Azaserine) have effects on glucose metabolism in endothelial cells. The 13C- MFA model was constructed to improve the understanding of endothelial cell metabolism. The treatment of Fidarestat significantly reduced the 13C enrichment of metabolites of glycolysis and TCA cycle, amino acids (such as serine, glycine, aspartic acid and alanine), UDP- glucose and UDP-GlcNAc, the 13C enrichment of glycolytic metabolites pyruvate and lactic acid, and the 13C enrichment of most metabolites in TCA cycle, especially serine and glycine. The distribution of isotopic isomers of UDP- glucose has changed. Although the total 13C enrichment has not changed significantly, the enrichment level of some isomers has changed. After Fidarestat treatment, the metabolic flow of glycolysis remained almost unchanged, but the flux of non-oxidizing PPP decreased. Cells increased TCA flux by increasing the intake of glutamine and malic acid. The treatment of DHEA has mixed effects on glycolytic metabolites, but in general, the 13C enrichment of glycolytic and TCA metabolites has decreased significantly, which has a wide impact on TCA metabolites. The 13C enrichment of all measured metabolites has decreased significantly, the PPP flux in cells has changed, the isotopic isomer distribution of R5P has also been adjusted, the 13C enrichment of nucleotides UMP and AMP has decreased, and the [m+6] isomer of UDP- glucose has increased. The effects of these inhibitors on metabolic pathways were further clarified by 13C MFA model. For example, fidarestat may regulate metabolism through SIRT1 or Nrf-2 pathway by enhancing glutamine uptake and TCA activity. DHEA-treated cells showed the mechanism of restoring NADPH by increasing malic enzyme activity. Azaserine further affects cell metabolism by affecting HBP pathway, possibly by reducing the labeling of UDP-GlcNAc (Moiz B et al., 2021).
• Hematopoietic stem cells (HSC) are powerful tissue stem cells located at the top of hematopoietic system. HSC has a strong resistance to stress (such as oxidative stress, aging, etc.) and has the characteristics of static cell cycle. During the division of HSC, metabolism will be reprogrammed, which will activate fatty acid β -oxidation (FAO) and purine metabolism, while during rest or stress, HSC mainly relies on glycolysis as the energy source. HSC will undergo metabolic reprogramming under stress, but it is still unclear whether this change occurs uniformly in deep static HSC. The purpose of this study is to find out the key metabolic enzymes, which may regulate the metabolic system of HSC under pressure load, and then affect its function. In this paper, the metabolic mechanism of glycolysis activation of HSCs after 5-FU treatment and OXPHOS inhibition was discussed by 13C-MFA. After 5-FU treatment, the number of quiescent (Ki67-) HSC cells decreased and the proportion of proliferating (Ki67+) cells increased. ATP concentration decreased briefly on the sixth day, indicating the change of metabolic demand. In HSC treated with 5-FU, the concentration of metabolites of glycolysis and pentose phosphate pathway increased significantly, and the level of metabolites of glycolysis was twice that of PBS group, and it was more dependent on anaerobic glycolysis. Seahorse analysis showed that glycolysis could be activated flexibly under the inhibition of OXPHOS, and glucose uptake was positively correlated with cell proliferation. PFKFB3 is a key metabolic enzyme, which mainly maintains ATP production in HSC by activating glycolysis. In the process of proliferation, PFKFB3 accelerates glycolysis to make up for the deficiency of OXPHOS metabolic pathway. In the mouse model treated with 5-FU, PFKFB3 activated ATP production, although AMPK inhibition had little effect on ATP level. HSC relies on glycolysis to maintain ATP level when OXPHOS is damaged, showing strong metabolic plasticity. The inhibition of PFKFB3 makes the ATP level in HSC decrease rapidly, indicating its important role in metabolic regulation. Under stress conditions, HSC mainly relies on glycolysis and PPP (pentose phosphate pathway) to produce ATP, especially when the cell cycle progresses and proliferates. Although under steady-state conditions, HSC mainly relies on mitochondrial respiration (OXPHOS) to maintain ATP level, under acute stress conditions, HSC chooses to accelerate glycolysis to make up for the lack of ATP when OXPHOS is inhibited. In addition, HSC does rely on PFKFB3 to support proliferation in the process of stress hematopoiesis, especially in the initial stress stage after bone marrow transplantation (Watanuki S et al., 2024).

References

1. Ahn WS, Crown SB, Antoniewicz MR. "Evidence for transketolase-like TKTL1 flux in CHO cells based on parallel labeling experiments and (13)C-metabolic flux analysis." Metab Eng. 2016 ;37:72-78. doi: 10.1016/j.ymben.2016.05.005
2. Pérez-Fernández BA, Calzadilla L, Enrico Bena C, Del Giudice M, Bosia C, Boggiano T, Mulet R. "Sodium acetate increases the productivity of HEK293 cells expressing the ECD-Her1 protein in batch cultures: experimental results and metabolic flux analysis." Front Bioeng Biotechnol. 2024 ;12:1335898. doi: 10.3389/fbioe.2024.1335898
3. Achreja A, Zhao H, Yang L, Yun TH, Marini J, Nagrath D. "Exo-MFA - A 13C metabolic flux analysis framework to dissect tumor microenvironment-secreted exosome contributions towards cancer cell metabolism." Metab Eng. 2017 ;43(Pt B):156-172. doi: 10.1016/j.ymben.2017.01.001
4. Moiz B, Garcia J, Basehore S, Sun A, Li A, Padmanabhan S, Albus K, Jang C, Sriram G, Clyne AM. "13C Metabolic Flux Analysis Indicates Endothelial Cells Attenuate Metabolic Perturbations by Modulating TCA Activity." Metabolites. 2021 ;11(4):226. doi: 10.3390/metabo11040226
5. Watanuki S, Kobayashi H, Sugiura Y, Yamamoto M, Karigane D, Shiroshita K, Sorimachi Y, Fujita S, Morikawa T, Koide S, Oshima M, Nishiyama A, Murakami K, Haraguchi M, Tamaki S, Yamamoto T, Yabushita T, Tanaka Y, Nagamatsu G, Honda H, Okamoto S, Goda N, Tamura T, Nakamura-Ishizu A, Suematsu M, Iwama A, Suda T, Takubo K. "Context-dependent modification of PFKFB3 in hematopoietic stem cells promotes anaerobic glycolysis and ensures stress hematopoiesis." Elife. 2024 ;12:RP87674. doi: 10.7554/eLife.87674