The Need for Integrated Approaches in Metabolic Engineering

Abstract

This review highlights state-of-the-art procedures for heterologous small-molecule biosynthesis, the associated bottlenecks, and new strategies that have the potential to accelerate future accomplishments in metabolic engineering. We emphasize that a combination of different approaches over multiple time and size scales must be considered for successful pathway engineering in a heterologous host. We have classified these optimization procedures based on the "system" that is being manipulated: transcriptome, translatome, proteome, or reactome. By bridging multiple disciplines, including molecular biology, biochemistry, biophysics, and computational sciences, we can create an integral framework for the discovery and implementation of novel biosynthetic production routes.

Figure 1.

Integrated approaches in metabolic engineering. To achieve high metabolite flux through a biosynthetic pathway, it is essential to consider bottlenecks on the transcriptome, translatome, protein/proteome, and reactome level. Integrated metabolic engineering approaches need to occur both on the molecular-scale and system-scale level.

Figure 2.

Predictable and reliable gene expression. (A) To precisely predict translation initiation rates, the ideal transcript would be depleted in secondary structures close to the ribosome-binding site (RBS). However, mRNA secondary structures are crucial to overall transcript stability and translation speed control. (B) Secondary structures present in most nascent mRNA molecules complicate accurate computational predictions of expression strength. Depending on the location of the hairpin, secondary structure formation can slow down translation initiation rates or completely prevent translation initiation in case of an occluded RBS. (C) The bicistronic gene design developed by Mutalik and coworkers (Davis et al. 2011; Mutalik et al. 2013) allows for higher predictability, because a constant short open reading frame is placed upstream of the target expression cassette that allows for read-through despite the presence of downstream hairpins.

Figure 3.

Soluble and active protein biosynthesis. (A) Proteins expressed in their native host naturally assume a soluble and catalytically active tertiary structure, whereas protein synthesis in nonnative hosts leads to misfolded or aggregated proteins. (B) Relative synonymous codon usage plots (RSCUs) for a representative protein in native and nonnative hosts correlated with structural motifs within the protein.

Figure 4.

Proteome-level optimization. (A) Optimal versus nonoptimal enzyme-level activities in engineered designs. Nonoptimal factors that influence activities may include allosteric regulation, protein posttranslational modifications, incorrect protein–protein interactions, and many others. (B) Graphical representation of directed evolution of enzyme that directly influences its stability and reactivity. (Adapted from Mollwitz et al. 2012.) (C) An in silico workflow to establish rational engineering through novel complementary computational methods, such as machine learning, genetic algorithms, and molecular modeling.

Figure 5.

Reactome-level optimization. (A) Nonoptimal metabolite flux limited by low turnover of one pathway enzyme. (B) Optimal flux through the pathway can be achieved by engineering the bottleneck enzyme leading to either higher enzyme levels or increased catalytic activity.