In silico analysis of design of experiment methods for metabolic pathway optimization.

Microbial cell factories allow the production of chemicals presenting an alternative to traditional fossil fuel-dependent production. However, finding the optimal expression of production pathway genes is crucial for the development of efficient production strains. Unlike sequential experimentation, combinatorial optimization captures the relationships between pathway genes and production, albeit at the cost of conducting multiple experiments. Fractional factorial designs followed by linear modeling and statistical analysis reduce the experimental workload while maximizing the information gained during experimentation. Although tools to perform and analyze these designs are available, guidelines for selecting appropriate factorial designs for pathway optimization are missing. In this study, we leverage a kinetic model of a seven-genes pathway to simulate the performance of a full factorial strain library. We compare this approach to resolution V, IV, III, and Plackett Burman (PB) designs. Additionally, we evaluate the performance of these designs as training sets for a random forest algorithm aimed at identifying best-producing strains. Evaluating the robustness of these designs to noise and missing data, traits inherent to biological datasets, we find that while resolution V designs capture most information present in full factorial data, they necessitate the construction of a large number of strains. On the other hand, resolution III and PB designs fall short in identifying optimal strains and miss relevant information. Besides, given the small number of experiments required for the optimization of a pathway with seven genes, linear models outperform random forest. Consequently, we propose the use of resolution IV designs followed by linear modeling in Design-Build-Test-Learn (DBTL) cycles targeting the screening of multiple factors. These designs enable the identification of optimal strains and provide valuable guidance for subsequent optimization cycles.

Machine Learning-Guided Optimization of p-Coumaric Acid Production in Yeast.

Industrial biotechnology uses Design-Build-Test-Learn (DBTL) cycles to accelerate the development of microbial cell factories, required for the transition to a biobased economy. To use them effectively, appropriate connections between the phases of the cycle are crucial. Using p-coumaric acid (pCA) production in Saccharomyces cerevisiae as a case study, we propose the use of one-pot library generation, random screening, targeted sequencing, and machine learning (ML) as links during DBTL cycles. We showed that the robustness and flexibility of the ML models strongly enable pathway optimization and propose feature importance and Shapley additive explanation values as a guide to expand the design space of original libraries. This approach allowed a 68% increased production of pCA within two DBTL cycles, leading to a 0.52 g/L titer and a 0.03 g/g yield on glucose.

Combinatorial optimization of pathway, process and media for the production of p-coumaric acid by Saccharomyces cerevisiae.

Microbial cell factories are instrumental in transitioning towards a sustainable bio-based economy, offering alternatives to conventional chemical processes. However, fulfilling their potential requires simultaneous screening for optimal media composition, process and genetic factors, acknowledging the complex interplay between the organism's genotype and its environment. This study employs statistical design of experiments to systematically explore these relationships and optimize the production of p-coumaric acid (pCA) in Saccharomyces cerevisiae. Two rounds of fractional factorial designs were used to identify factors with a significant effect on pCA production, which resulted in a 168-fold variation in pCA titre. Moreover, a significant interaction between the culture temperature and expression of ARO4 highlighted the importance of simultaneous process and strain optimization. The presented approach leverages the strengths of experimental design and statistical analysis and could be systematically applied during strain and bioprocess design efforts to unlock the full potential of microbial cell factories.

Enzyme-constrained models predict the dynamics of Saccharomyces cerevisiae growth in continuous, batch and fed-batch bioreactors.

Genome-scale, constraint-based models (GEM) and their derivatives are commonly used to model and gain insights into microbial metabolism. Often, however, their accuracy and predictive power are limited and enable only approximate designs. To improve their usefulness for strain and bioprocess design, we studied here their capacity to accurately predict metabolic changes in response to operational conditions in a bioreactor, as well as intracellular, active reactions. We used flux balance analysis (FBA) and dynamic FBA (dFBA) to predict growth dynamics of the model organism Saccharomyces cerevisiae under different industrially relevant conditions. We compared simulations with the latest developed GEM for this organism (Yeast8) and its enzyme-constrained version (ecYeast8) herein described with experimental data and found that ecYeast8 outperforms Yeast8 in all the simulations. EcYeast8 was able to predict well-known traits of yeast metabolism including the onset of the Crabtree effect, the order of substrate consumption during mixed carbon cultivation and production of a target metabolite. We showed how the combination of ecGEM and dFBA links reactor operation and genetic modifications to flux predictions, enabling the prediction of yields and productivities of different strains and (dynamic) production processes. Additionally, we present flux sampling as a tool to analyse flux predictions of ecGEM, of major importance for strain design applications. We showed that constraining protein availability substantially improves accuracy of the description of the metabolic state of the cell under dynamic conditions. This therefore enables more realistic and faithful designs of industrially relevant cell-based processes and, thus, the usefulness of such models.

Inhibition of Xanthine Oxidoreductase Enhances the Potential of Tyrosine Kinase Inhibitors against Chronic Myeloid Leukemia.

Chronic myeloid leukemia (CML) is characterized by the expression of the oncogenic kinase BCR-ABL. Although tyrosine kinase inhibitors (TKIs) against BCR-ABL represent the standard therapeutic option for CML, resistances to TKIs can be a serious problem. Thus, the search for novel therapeutic approaches is still needed. CML cells show an increased ROS production, which is required for maintaining the BCR-ABL signaling cascade active. In line with that, reducing ROS levels could be an interesting therapeutic strategy for the clinical management of resistant CML. To analyze the therapeutic potential of xanthine oxidoreductase (XOR) in CML, we tested the effect of XOR inhibitor allopurinol. Here, we show for the first time the therapeutic potential of allopurinol against BCR-ABL-positive CML cells. Allopurinol reduces the proliferation and clonogenic ability of the CML model cell lines K562 and KCL22. More importantly, the combination of allopurinol with imatinib or nilotinib reduced cell proliferation in a synergistic manner. Moreover, the co-treatment arms hampered cell clonogenic capacity and induced cell death more strongly than each single-agent arm. The reduction of intracellular ROS levels and the attenuation of the BCR-ABL signaling cascade may explain these effects. Finally, the self-renewal potential of primary bone marrow cells from CML patients was also severely reduced especially by the combination of allopurinol with TKIs. In summary, here we show that XOR inhibition is an interesting therapeutic option for CML, which can enhance the effectiveness of the TKIs currently used in clinics.