Machine learning modeling of the effects of media formulated with various yeast extracts on heterologous protein production in Escherichia coli

Unlocking the genome of the non-sourdough Kazachstania humilis MAW1: insights into inhibitory factors and phenotypic properties

BACKGROUND

Ascomycetous budding yeasts are ubiquitous environmental microorganisms important in food production and medicine. Due to recent intensive genomic research, the taxonomy of yeast is becoming more organized based on the identification of monophyletic taxa. This includes genera important to humans, such as Kazachstania. Until now, Kazachstania humilis (previously Candida humilis) was regarded as a sourdough-specific yeast. In addition, any antibacterial activity has not been associated with this species.

RESULTS

Previously, we isolated a yeast strain that impaired bio-hydrogen production in a dark fermentation bioreactor and inhibited the growth of Gram-positive and Gram-negative bacteria. Here, using next generation sequencing technologies, we sequenced the genome of this strain named K. humilis MAW1. This is the first genome of a K. humilis isolate not originating from a fermented food. We used novel phylogenetic approach employing the 18 S-ITS-D1-D2 region to show the placement of the K. humilis MAW1 among other members of the Kazachstania genus. This strain was examined by global phenotypic profiling, including carbon sources utilized and the influence of stress conditions on growth. Using the well-recognized bacterial model Escherichia coli AB1157, we show that K. humilis MAW1 cultivated in an acidic medium inhibits bacterial growth by the disturbance of cell division, manifested by filament formation. To gain a greater understanding of the inhibitory effect of K. humilis MAW1, we selected 23 yeast proteins with recognized toxic activity against bacteria and used them for Blast searches of the K. humilis MAW1 genome assembly. The resulting panel of genes present in the K. humilis MAW1 genome included those encoding the 1,3-β-glucan glycosidase and the 1,3-β-glucan synthesis inhibitor that might disturb the bacterial cell envelope structures.

CONCLUSIONS

We characterized a non-sourdough-derived strain of K. humilis, including its genome sequence and physiological aspects. The MAW1, together with other K. humilis strains, shows the new organization of the mating-type locus. The revealed here pH-dependent ability to inhibit bacterial growth has not been previously recognized in this species. Our study contributes to the building of genome sequence-based classification systems; better understanding of K.humilis as a cell factory in fermentation processes and exploring bacteria-yeast interactions in microbial communities.

Optimization of medium components for protein production by Escherichia coli with a high-throughput pipeline that uses a deep neural network

To optimize rapidly the medium for green fluorescent protein expression by Escherichia coli with an introduced plasmid, pRSET/emGFP, a single-cycle optimization pipeline was applied. The pipeline included a deep neural network (DNN) and mathematical optimization algorithms with simultaneous optimization of 18 medium components. To evaluate the DNN data sampling method, two methods, orthogonal array (OA) and Latin hypercube sampling (LHS), were used to design 64 initial media for each sampling method. The OA- and LHS-based data sampling resulted in green fluorescent protein fluorescence intensities of 0.088 × 103-1.85 × 104 and 3.30 × 103-1.50 × 104, respectively. Fifty DNN models were built using the OA and LHS datasets. Hold-out validation was performed using 15 % test of OA and LHS data. Mean square errors of the DNN models were 0.015-0.64, indicating the estimation accuracies were sufficient. However, the sensitivities of components in the DNN models varied and were grouped into six major classes by the index of k-means clustering. A representative model was selected for each class. Mathematical optimization algorithms using Bayesian optimization and genetic algorithm were applied to the representative models, and representative optimized medium (OM) compositions were selected by k-means clustering from the proposed OMs. A total of 54 OMs were obtained from the OA and LHS datasets. In the validating cultivation, the best OMs of OA and LHS were 2.12-fold and 2.13-fold higher, respectively, than those of the learning data.

Metabolome analysis of metabolic burden in Escherichia coli caused by overexpression of green fluorescent protein and delta-rhodopsin

Overexpression of proteins by introducing a DNA vector is among the most important tools for the metabolic engineering of microorganisms such as Escherichia coli. Protein overexpression imposes a burden on metabolism because metabolic pathways must supply building blocks for protein and DNA synthesis. Different E. coli strains have distinct metabolic capacities. In this study, two proteins were overexpressed in four E. coli strains (MG1655(DE3), W3110(DE3), BL21star(DE3), and Rosetta(DE3)), and their effects on metabolic burden were investigated. Metabolomic analysis showed that E. coli strains overexpressing green fluorescent protein had decreased levels of several metabolites, with a positive correlation between the number of reduced metabolites and green fluorescent protein expression levels. Moreover, nucleic acid-related metabolites decreased, indicating a metabolic burden in the E. coli strains, and the growth rate and protein expression levels were improved by supplementation with the five nucleosides. In contrast, two strains overexpressing delta rhodopsin, a microbial membrane rhodopsin from Haloterrigena turkmenica, led to a metabolic burden and decrease in the amino acids Ala, Val, Leu, Ile, Thr, Phe, Asp, and Trp, which are the most frequent amino acids in the delta rhodopsin protein sequence. The metabolic burden caused by protein overexpression was influenced by the metabolic capacity of the host strains and the sequences of the overexpressed proteins. Detailed characterization of the effects of protein expression on the metabolic state of engineered cells using metabolomics will provide insights into improving the production of target compounds.

Artificial intelligence-assisted analysis reveals amino acid effects and interactions on Limosilactobacillus fermentum growth

To understand the growth of lactic acid bacteria (LAB), Limosilactobacillus fermentum, in response to medium compositions, a deep neural network (DNN) was designed using amino acids (AAs) as explanatory variables and LAB growth as the objective variable. Sixty-four different patterns of free AAs were set using an orthogonal array. The best DNN model had high accuracy with low mean square errors and predicted that Asp would affect LAB growth. Bayesian optimization (BO) using this model recommended an optimal growth media comprising maximum amounts of Asn, Asp, Lys, Thr, and Tyr and minimum amounts of Gln, Pro, and Ser. Furthermore, this proposed media was empirically validated to promote LAB growth. The absence of Gln, Ser, and Pro indicates that the different growth trends among the DNN-BO-optimized media were likely caused by the interactions among the AAs and the other components.

A review of algorithmic approaches for cell culture media optimization

Cell culture media composition and culture conditions play a crucial role in product yield, quality and cost of production. Culture media optimization is the technique of improving media composition and culture conditions to achieve desired product outcomes. To achieve this, there have been many algorithmic methods proposed and used for culture media optimization in the literature. To help readers evaluate and decide on a method that best suits their specific application, we carried out a systematic review of the different methods from an algorithmic perspective that classifies, explains and compares the available methods. We also examine the trends and new developments in the area. This review provides recommendations to researchers regarding the suitable media optimization algorithm for their applications and we hope to also promote the development of new cell culture media optimization methods that are better suited to existing and upcoming challenges in this biotechnology field, which will be essential for more efficient production of various cell culture products.

High throughput optimization of medium composition for Escherichia coli protein expression using deep learning and Bayesian optimization

To improve synthetic media for protein expression in Escherichia coli, a strategy using deep neural networks (DNN) and Bayesian optimization was performed in this study. To obtain training data for a deep learning algorithm, E. coli harvesting a plasmid pRSET/emGFP, which introduces the green fluorescence protein (GFP), was cultivated in 81 media designed using a Latin square in deepwell-scale cultivation. The media were composed of 31 components with three levels. The resultant GFP fluorescence intensities were evaluated using a fluorescence spectrometer, and the intensities were in the range 2.69-7.99 × 10³. A deep neural network model was used to estimate the GFP fluorescence intensities from the culture media compositions, and accuracy was evaluated using cross-validation with 15% test data. Bayesian optimization using the best DNN model was used to calculate 20 representative compositions optimized for GFP expression. According to the validating cultivation, the simulated GFP expression levels included large errors between the estimated and experimental data. The DNN model was retrained using data from the validating cultivation, and secondary estimations were performed. The secondary estimations fit the corresponding experimental data well, and the best GFP fluorescence intensity was 1.4-fold larger than the best of the initial test composition.

Combined multivariate statistical and flux balance analyses uncover media bottlenecks to the growth and productivity of CHO cell cultures

Chinese hamster ovary (CHO) cells are widely used for producing recombinant proteins. To enhance their productivity and product quality, media reformulation has been a key strategy, albeit with several technical challenges, due to the myriad of complex molecular mechanisms underlying media effects on culture performance. Thus, it is imperative to characterize metabolic bottlenecks under various media conditions systematically. To do so, we combined partial least square regression (PLS-R) with the flux balance analysis of a genome-scale metabolic model to elucidate the physiological states and metabolic behaviors of human alpha-1 antitrypsin producing CHO-DG44 cells grown in one commercial and another two in-house media under development. At the onset, PLS-R was used to identify metabolite exchanges that were correlated to specific growth and productivity. Then, by comparing metabolic states described by resultant flux distributions under two of the media conditions, we found sub-optimal level of four nutrients and two metabolic wastes, which plausibly hindered cellular growth and productivity; mechanistically, lactate and ammonia recycling were modulated by glutamine and asparagine metabolisms in the media conditions, and also by hitherto unsuspected folate and choline supplements. Our work demonstrated how multivariate statistical analysis can be synergistically combined with metabolic modelling to uncover the mechanistic elements underlying differing media performance. It thus paved the way for the systematic identification of nutrient targets for medium reformulation to enhance recombinant protein production in CHO cells. This article is protected by copyright. All rights reserved.