Beyond Protein Engineering: its Applications in Synthetic Biology

After decades of development, protein engineering has become sufficiently important to industry that its focus has shifted from technology innovation to application exploration. While protein engineering continues to play an important role in established industries like detergent enzymes, it also promises to provide new solutions in emerging industries such as biopharmaceuticals [evolved biologics] and renewable energy [engineered enzymes for biofuel production].


Introduction
After decades of development, protein engineering has become sufficiently important to industry that its focus has shifted from technology innovation to application exploration. While protein engineering continues to play an important role in established industries like detergent enzymes, it also promises to provide new solutions in emerging industries such as biopharmaceuticals [evolved biologics] and renewable energy [engineered enzymes for biofuel production].
In contrast to protein engineering, synthetic biology is still in its infancy, even though the term, "synthetic biology", was coined over thirty years ago [1]. Synthetic biology has great potential for enabling sustainable economic growth. Indeed, synthetic pathways can produce many chemicals, including biofuels and drugs, from renewable resources. Currently, synthetic pathways are assembled from biological components culled from nature, and may not be optimal since those biological components were evolved by nature to benefit their native species' fitness and were not customized for performance in synthetic pathways. Biological components from different species [or even different kingdoms] may not function optimally when simply put together in a pathway. Therefore, to meet industrial standards, synthetic pathways usually require optimization to produce chemicals economically, and this is where protein engineering can play significant roles. In fact, protein engineering has already been applied to optimize a number of synthetic biological components, from pathway enzymes to regulatory elements. It has also been used to balance pathway redox equivalents and control the expression of pathway genes. This editorial highlights applications of protein engineering to synthetic biology in the areas of improving enzyme activity, changing substrate or product specificity, modifying cofactor usage for redox balance, tuning protein expression, and creating scaffold multi-enzyme complexes for metabolite flux control. Due to limited space, only one to two examples are given for each application to demonstrate the concept.

Enzyme Engineering to Improve Enzyme Performance in a Pathway
Creation of a synthetic pathway often involves heterologous expression of multiple enzymes from diverse species. The properties of one or more wild type enzymes, such as activity or specificity, may not meet industrial needs. The problem of low enzyme activity may be addressed by simply increasing the expression level, but this approach may increase the cellular protein burden and may reduce the cellular capacity to produce other necessary pathway enzymes. Changes to other enzymatic properties like substrate specificity require alteration of the enzymes themselves. Protein engineering is an efficient way to change an enzyme's kinetic parameters to improve its in vivo performance, and thus enhance pathway productivity.
One approach to increase pathway productivity is to improve the specific activity of a critical or rate-limiting enzyme through protein engineering. There are many successful examples of this. To increase the production of polyhydroxyalkanoate [PHA] in E. coli, Amara et al. [2] engineered a PHA synthase from Aeromonas punctata using the mutator strain E. coli XL1-Red. They screened about 200,000 mutants on medium containing Nile red stain and identified five variants with enhanced fluorescence. Four of them exhibited both elevated in vitro and in vivo PHA synthase activity, with as much as 5-fold improvement over the wild type. This improvement in enzymatic activity resulted in over 126% PHA accumulation in cells, and all five mutants increased PHA weight-average molar mass yield [2].  [3] systematically elevated metabolic flux toward IPP and DMAPP but failed to increase LP titers. However, subsequent combinatorial mutagenesis [site directed mutagenesis and site-saturation mutagenesis based on homology modeling and phylogenetic analysis] of GGPPS and LPS followed by library screening identified improved variants of both enzymes. Combined with the improvements in precursor flux, LP productivity was improved by 2,600-fold [3].
In certain scenarios, an enzyme employed in a pathway need catalyze the conversion of a non-native substrate. Promiscuous enzymes are suitable for this purpose, but low substrate specificity may reduce desired product yield due to inefficient substrate conversion and byproduct formation. Altering and/or improving substrate specificity can lead to higher yields of desired pathway products. To reduce the formation of byproducts, Zhang et al. increased the substrate specificity of 2-ketoisovalerate decarboxylase from Lactococcus lactis on a desired substrate, 2-keto-4-methylhexanoate, by modifying its substratebinding sites based on a homology model. The authors also altered the substrate specificity of another pathway enzyme, 2-isopropylmalate synthase [LeuA] from E. coli. Based on its crystal structure, the authors introduced mutations that enlarged the substrate-binding pocket to accommodate the larger unnatural substrate, 2-keto-3methylvalerate. This further increased the pathway's productivity of nonnatural alcohols up to C 8 [4]. In another attempt to produce nonnatural chemicals using a synthetic pathway, Zhang et al. [5] changed the substrate preference of E. coli glutamate dehydrogenase [GDH] from 2-ketoglutarate to 2-ketobutyrate to produce L-homoalanine, a valuable drug precursor. Together with other pathway optimization steps, the authors managed to produce 5.4 g/L of L-homoalanine from a feedstock of 30 g/L glucose [5]. The work of Zhang et al. [5] has demonstrated that synthetic pathways can be utilized to produce many non-natural chemicals. Moreover, these efforts have shown that the productivity of synthetic pathways can be dramatically enhanced through the alteration of pathway enzyme specificity.
Some enzymes produce multiple products from the same substrate. If such enzymes are to be employed in synthetic pathways, their product specificity may require narrowing to improve the productivity of the desired pathway product. For example, γ-humulene synthase [HUM] from Abies grandis can produce as many as 52 different sesquiterpenes from a sole substrate, farnesyl diphosphate [FPP]. To alter the product specificity of HUM, Yoshikuni et al. [6] first identified 19 amino acid residues as targets for mutagenesis based on a homology model. After performing saturation mutagenesis on these residues, they found that only four of them had a significant effect on catalysis. The authors then developed an algorithm to systematically recombine the mutations found to alter product specificity. This resulted in the creation of seven HUM variants, each of which specifically produces a different product [6]. The same team later changed the product specificity of the [+]-δ-cadinene synthase from Gossypium arboretum to create an entirely novel sesquiterpene synthase that produces germacrene D-4-ol from FPP, in order to meet the requirement of a synthetic pathway [7].

Engineering cofactor utilization of enzymes in metabolic pathways
Redox [reduction-oxidation] reactions are chemical reactions involving changes in the oxidation state of molecules. In a typical pathway, redox is reflected by the status of pyridine nucleotides, NAD[H] and NADP [H]. Redox balance means that the production and consumption of reduced cofactors [NADPH and/or NADH] are approximately equal. Unbalanced redox can damage cells, waste energy and/or carbon, and even lead to metabolic arrest [8]. Fortunately, redoximbalanced pathways can sometimes be re-balanced through cofactor engineering. Two approaches to cofactor engineering are (1) removal of excess cofactors via enzymatic conversion [e.g., by expression of a cofactor oxidase or transhydrogenase], and, (2) switching the cofactorutilization preference [specificity] of pathway enzymes.
The concentration of NADH in a cell is generally about an order of magnitude higher than that of NADPH. [For example, the reported value for NAD[H] is 1.3 μM and that for NADP[H] is 0.39 μM for exponentially-growing Escherichia coli [9]. In addition, NADH is more stable than NADPH, and is also roughly an order of magnitude less expensive. Therefore, many efforts to balance redox aim for pathway utilization of NADH.
To optimize the vitamin C biosynthetic pathway, Banta et al. [9] switched the cofactor preference of Corynebacterium 2,5-Diketo-D-gluconic acid [2,5-DKG] reductase from NADPH to NADH. The authors first subjected the binding site of the 2'-phosphate group of NADPH to site-directed mutagenesis [3]. Several beneficial mutations were indentified and were combined with other substitutions suggested by the sequences of wild type NADH-dependent aldo-keto reductases. The resultant quadruple mutant, F22Y/K232G/R238H/A272G, has NADH-dependent activity over 2 orders of magnitude greater than that of wild type 2,5-DKG reductase [3,5].
The status of the reduced forms of NAD[H] and NADP[H] may greatly affect product yield or the formation of byproducts [8]. To test this idea, Heux et al. [8] expressed an NADH oxidase from Lactococcus lactis in S. cerevisiae strain V5 to reduce the intracellular NADH pool [8]. Under controlled microaerobic conditions, the intracellular NADH concentration was reduced five-fold and the NADH/NAD + ratio was decreased six-fold. Several metabolic fluxes were found to be redistributed in response to the imposed NADH consumption, presumably to maintain redox balance. The production levels of ethanol, glycerol, succinate and hydroxyglutarate were significantly decreased, whereas the levels of many oxidized metabolites such as acetaldehyde, acetate and acetoin were increased [8]. At a glucose concentration of 10% [w/v], the biomass yield and glucose consumption were both lower than in the control strain lacking heterologous NADH oxidase [8]. This work demonstrated that when cellular redox balance is changed, cells may be able to make their own adjustments, and this may affect pathway productivity.
Another way to balance redox is to remove excess cofactor by converting it into a more useful form [e.g. NADH to NADPH]. Pyridine nucleotide transhydrogenase catalyzes the reversible transfer of a hydride ion between NADH and NADP + . By co-expressing NADP +dependent alcohol dehydrogenase and NAD + -dependent formate dehydrogenase, Weckbecker and Hummel managed to improve the production of chiral alcohols in E. coli [10]. This strategy has limited applicability, however. For example, the conversion of NADH to NADPH requires a proton motive force and one proton per hydride. This consumes energy and can substantially decrease a pathway's productivity [11].

Engineering of genetic regulatory elements to tune protein expression
Unlike native cellular pathways [e.g., many secondary metabolic pathways], which are usually tightly-regulated, synthetic pathways may not be as well-tuned to respond to dynamic internal and external environments. Up-or down-regulation of the expression of some pathway genes is generally needed to achieve balanced expression of the various pathway enzymes. Genes are regulated at several levels: transcription, post-transcription, translation, and post-translation. In response to changes in the concentration of effector molecules [e.g., activators or inhibitors], transcriptional regulatory proteins can regulate mRNA expression and allosteric regulatory domains in enzymes can modulate enzymatic activity [12]. Regulatory proteins can thus control protein expression or metabolic flux. Protein engineering concepts can also be applied to engineer non-protein regulatory elements, like promoters or other DNA or RNA regulatory regions to affect transcription and/or translation and thus control protein expression.
Pathway productivity can be increased by reducing the inhibition of pathway enzymes by products, intermediates, or other inhibitory small molecules. Such inhibition can be modulated through an allosteric regulatory domain. Deng et al. created error-prone PCR libraries of glucosamine synthase [GlmS] and screened for mutants that were less inhibited by the intermediate, glucosamine-6-phosphate. They discovered a GlmS variant that increased glucosamine production by 15-fold [13]. Baez-Viveros et al. [14] engineered chorismate mutaseprephenate dehydratase [CM-PDT] for resisting feedback inhibition and managed to enhance the yield of L-phenylalanine by 1.6-fold using the evolved CM-PDT in concert with other optimization strategies [14].
DNA regulatory regions, like promoters, operators, and enhancers, affect genetic transcription and consequently, protein expression. Many synthetic promoters have been engineered to fine-tune pathway enzyme expression [15]. A library of mutated promoters is usually Eng  ISSN: EEG, an open access journal cloned upstream of a reporter gene [e.g., a fluorescent protein]. After screening and identification of desirable promoter variants, the reporter genes are replaced with pathway genes. Alper et al. [1] created an errorprone PCR library of P L _ λ promoter, which was cloned into a reporter plasmid upstream of a green fluorescenct protein [GFP] gene [1]. After screening in E. coli, nearly 200 promoter variants with a wide range of strengths were generated. The authors applied the same concept to engineer the TEF1 promoter with yellow fluorescent protein [YFP] as the reporter gene in S. cerevisiae. The resulting mutated promoters also covered a wide range of strengths, and can in principle serve as a "catalog" of choices to fine-tune the expression of a multitude of pathway genes [1].
RNA molecules are well-known to form secondary structural elements. These folds can have a great impact on gene expression by modulating mRNA stability or controlling the rate of translation initiation. Pfleger et al. [16] generated libraries of tunable intergenic regions [TIGRs] with various mRNA secondary structures and RNase cleavage sites to coordinate the expression of multiple genes in a heterologous mevalonate biosynthetic pathway. By this approach, they improved mevalonate production by 7-fold [16]. Smolke et al. used specially-designed RNA sequences to control gene expression both temporally and spatially [17], and exploited this technology to increase secondary metabolite biosynthesis in yeast [18].
Many other genetic regulatory elements, like ribosome binding sites [19], the ribosome itself [20], and operon structures [21], have also been engineered to alter pathway gene expression and affect pathway productivity.

Enzyme scaffold engineering to control metabolite flux
Pathway intermediates can be toxic to host cells, consumed by endogenous cellular activities, or reduced through secretion [22]. The issues could be mitigated by spatially arranging pathway enzymes so the substrate moves efficiently from one step to the next. The increased local concentration of pathway intermediates, reduction of opportunities for diversion of intermediates to other cellular enzymes, and decreased chance of secretion are all potential advantages afforded by this approach. Scaffold proteins are known in nature that spatially organize specific proteins in the cell, and are very important for coordinating certain cellular activities [23]. Recently, researchers have exploited such scaffolds to co-localize synthetic pathway enzymes into complexes with optimal enzyme stoichiometries.
Inspired by natural substrate-channeling systems, Dueber et al. [24] engineered protein scaffolds using protein-protein interaction domains and ligands from metazoan cells [mouse SH3 and PDZ domains, and rat GBD domain] for pathway enzymes in an effort to improve pathway productivity [24]. Through varying x, y, and z, the pathway was optimized and the product titer was increased by 77-fold [24]. Because the pathway enzymes are co-localized in the same scaffold and enzyme stoichiometry can be independently adjusted, the optimal balance of enzyme activities within the complex can be achieved, improving pathway productivity. Using a similar scaffold design, Moon et al. [25] increased glucaric acid titer by 5-fold compared to the non-scaffolded control [25].

Conclusion and Perspectives
In the past several decades, particularly the past two, many robust and efficient protein engineering technologies have been developed. These technologies are now starting to be employed in a very exciting and nascent field -synthetic biology, as reviewed in this editorial. Some concepts from protein engineering have also been borrowed by synthetic biologists. For example, rational design and directed evolution, two general strategies for protein engineering [28], have been deployed in synthetic biology [26 27]. More techniques of protein engineering are expected to be used in synthetic biology in the future.
One area of protein engineering not mentioned in this editorial involves unnatural amino acids [29]. There are a few examples of the application of unnatural amino acids in synthetic biology [30]. Since unnatural amino acids can be used to functionalize proteins and immobilize pathway enzymes in vivo, one potential application is to channel substrates for synthetic pathways to improve pathway productivity, in a similar fashion to the synthetic scaffold structure approach described above [24].
With the rapid development of "-omics" technologies, an overwhelming amount of biological data have become available [31]. In addition, computational power has been dramatically expanding and molecular modeling has become more accurate. All these developments make rational design more attractive, rapid, and powerful. As a result, more rational and semi-rational designs are likely to be seen in protein engineering and its applications to synthetic biology going forward.
While we appreciate the importance of protein engineering to synthetic biology, we should not underestimate the role of synthetic biology in further advancing protein engineering. For example, synthetic biology can be used to develop more efficient highthroughput screening methods to capitalize on protein engineering techniques [32].
It is clear that the marriage of protein engineering and synthetic biology promises to be a fruitful one. Trained to think on a molecular level, protein engineers and synthetic biologists routinely "think big" while working on the small [e.g. microbial cells]. And though these ideas are all implemented first in microscopic single cells, the biggest of them will surely do nothing less than change the world.