Synthetic Methanol and Formate Assimilation Via Modular Engineering and Selection Strategies

Nico J. Claassens, Hai He and Arren Bar-Even*

Max Planck Institute of Molecular Plant Physiology, Potsdam-Golm, Germany. *Correspondence: [email protected] htps://doi.org/10.21775/cimb.033.237

Abstract Introduction One-carbon (C1) feedstocks can provide a One-carbon (C1) compounds could prove to be vital link between cheap and sustainable abiotic a crucial link between the abiotic and the biotic resources and microbial bioproduction. Soluble C1 worlds. Tese feedstocks can be obtained from substrates – methanol and formate – could prove low-cost and abundant sources, such as syngas and to be more suitable than gaseous feedstocks as they natural gas (Dürre and Eikmanns, 2015; Clomburg avoid mass transfer barriers. However, microorgan- et al., 2017), and can be produced directly from isms that naturally assimilate methanol and formate CO2 using energy sources such as sunlight and are limited by a narrow product spectrum and a renewable electricity (Kumar et al., 2012; Martín et restricted genetic toolbox. Engineering biotech- al., 2015; Claassens et al., 2018; Jouny et al., 2018). nological organisms to assimilate these soluble C1 Multiple microorganisms can be cultivated on C1 substrates has therefore become an atractive goal. compounds as sole carbon and energy sources, thus Here, we discuss the use of a step-wise, modular opening new avenues for sustainable bioproduc- engineering approach for the implementation of tion. C1 assimilation pathways. In this strategy, pathways However, the use of microorganisms that can are divided into metabolic modules, the activities naturally grow on C1 substrates is limited by mul- of which are selected for in dedicated gene-deletion tiple factors, including a narrow product spectrum, strains whose growth directly depends on module low yields, titres, and productivities, a restricted activity. Tis provides an easy way to identify and genetic toolbox for engineering, and gaps in our resolve metabolic barriers hampering pathway per- understanding of their cellular physiology and formance. Optimization of gene expression levels metabolism (Whitaker et al., 2015; Clomburg et al., and adaptive laboratory evolution can be used to 2017). To overcome these difculties, recent meta- establish the desired activity if direct selection fails. bolic engineering eforts are aiming to introduce We exemplify this approach using several pathways, C1 assimilation pathways into model biotechno- focusing especially on the ribulose monophosphate logical microorganisms that are easier to engineer cycle for methanol assimilation and the reduc- and that can be beter optimized for industrially tive glycine pathway for formate assimilation. We relevant conditions. Tese eforts use either natural argue that such modular engineering and selection pathways that are known to sustain high yields, or, strategies will prove essential for rewiring microbial more boldly, synthetic pathways with low ATP cost metabolism towards new growth phenotypes and that could theoretically support increased yields sustainable bioproduction. (Bar-Even et al., 2013; Siegel et al., 2015; Bar-Even,

Curr. Issues Mol. Biol. (2019) Vol. 33 caister.com/cimb 238 | Claassens et al.

2016;). Some of these synthetic pathways can Modularity and selection be established by combining naturally existing as metabolic engineering enzymes, while others include novel enzyme activi- strategies ties that can be realized by protein engineering Engineering synthetic C1 metabolism requires (Erb et al., 2017). In fact, engineered enzymes have the overexpression of pathway enzymes, especially already been demonstrated in vitro to support for- those that are missing in the host or that are natively mate assimilation (Siegel et al., 2015) and carbon expressed at insufcient levels. However, simple fxation (Schwander et al., 2016). overexpression is unlikely to be sufcient for realiz- In this review, we discuss metabolic engineering the activity of the entire pathway. Tis is mainly ing studies aiming to introduce pathways for the because of the overlap between the introduced assimilation of the soluble C1 compounds metha- pathway and the host central metabolism, result- nol and formate, the utilization of which bypasses ing in disrupted fuxes through both systems. To the challenges associated with mass transfer of beter identify and resolve problematic metabolic gaseous C1 substrates, such as methane and carbon interactions, it is helpful to divide the synthetic monoxide (Henstra et al., 2007; Fei et al., 2014). pathway into smaller metabolic modules, i.e. sub- We specifcally focus on modular and selection- pathways consisting of several reactions (Fig. 14.1). based engineering strategies in which the activity Te in vivo implementation of these modules can of pathway segments is coupled to cellular growth. be considerably easier than the full pathway and We show that this step-wise approach is vital for the provide vital information on the metabolic context realization of synthetic C1 assimilation. that enables or constrains the newly introduced

1 2 3 4 Divide synthetic pathway Multiple expression levels for into several metabolic each enzyme are tested, modules, each corresponds for example by varying to a discrete metabolic goal strength of promoters, RBS, and plasmid origins of replication

Express each module in a low high dedicated auxotrophic selection selection strain, the growth of which Increase selection for module

strictly depends upon Fraction of cells activity in different strains, the module activity, enabling growth of each requires a

direct selection for activity Protein level (log scale) different level of module activity If direct selection fails, or Feeding with 13C-labeled formate results in sub-optimal growth, (or another carbon source) and + long-term cultivation under monitoring the labeling pattern in selective conditions proteinogenic amino acids, we (chemostat or turbidostat) can confirm module activity or = evolves desired growth identify competing routes that properties should be deleted Sequence evolved strains, Integrate modules into a full O O feed introduce mutations to a H pathway within a strain carrying naïve strain to identify the 2 the beneficial mutations previously HC C H C contribution of each mutation, OH HO C OH identified, and select for growth via identify minimal set of the pathway, upon expression of the effluent mutations to enable activity NH2 (potentially evolved) pathway enzymes 5 6 7 8

Figure 14.1 A schematic representation of the modular engineering and selection approach outlined in this paper.

Curr. Issues Mol. Biol. (2019) Vol. 33 caister.com/cimb Strategies for Synthetic C1 Assimilation | 239 activities. To probe the implementation of meta- and potentially also of related host enzymes, e.g. bolic modules, it is useful to couple their activity deletion or down-regulation of enzymes that with the growth of the host. divert metabolic intermediates from the pathway. Coupling module activity with growth usu- In addition, diferent enzyme variants or codon ally requires modifying the metabolic network of optimization of the relevant genes can sup- the host by performing strategic gene deletions. port increased expression and activity. Another Tese are made to generate a strain auxotrophic method, which does not rely on genetic tools, is for certain essential metabolic intermediates – for the addition of small molecules that specifcally example, an amino acid – which can be exclusively inhibit interfering enzymes, as demonstrated synthesized via the synthetic module. As a result, for the glycolytic glyceraldehyde 3-phosphate cellular growth becomes dependent on the activ- dehydrogenase in the engineering of methanol ity of the module. A range of selection strains can assimilation in Escherichia coli (Woolston et al., be designed with increasing selection pressure for 2018a). pathway activity: a ‘minimal’ selection is sustained If these approaches fail to establish module if the module provides a single required metabo- activity, adaptive laboratory evolution (ALE) can lite, higher selection pressure is obtained when be performed to increase module functionality module activity is responsible for the biosynthesis and establish module-dependent growth (Portnoy of multiple building blocks, and very high selection et al., 2011). For this process, the overexpressed pressure is imposed when the biosynthesis of all or genes should preferably be integrated into the most biomass is dependent on the module. genome rather than carried on a plasmid as to Te design of modules and selection strains increase the chance of benefcial mutations to be can be assisted by computational tools based on fxed in the population. Diferent types of ALE Flux Balance Analysis, for example OptKnock can be applied; a prominent approach being con- or FlexFlux (Burgard et al., 2003; Marmiesse et tinuous cultivation on a selective medium, with al., 2015; Meyer et al., 2018). Yet, in most cases, limiting amounts of the compounds for which the manual design based on biochemical and metabolic cells are auxotrophic. Tis method was applied knowledge sufces. Specifcally, when dividing a for the successful engineering of the CO2-fxing pathway into metabolic modules, several factors Calvin cycle in E. coli (Antonovsky et al., 2016), should be taken into consideration (Wenk et al., where the concentration of the limiting sub- 2018): (1) the module should be linked to a clear strate xylose was gradually decreased as growth growth phenotype within an appropriate selec- improved. Another approach involves swapping tion strain, resulting in growth versus no growth between permissive and selection media, where readout; (2) the number of enzymes in a module the former contains the auxotrophy-related com- should be limited, to enable easy expression and pounds and the later does not (Marlière et al., optimization, and to allow straightforward inter- 2011). Te relative dosages of the diferent media pretation of growth phenotypes; (3) modules are coupled to the growth of the population, should together cover the whole pathway and could such that increased cell density leads to addi- overlap with one another, such that enzymes occur- tion of selection medium, and a decrease in cell ring in multiple modules can be tested in diferent density results in more permissive medium. Such metabolic contexts; (4) ideally, modules should cultivation regime adapts the population towards be easy to combine, that is, dedicated selection using the selection medium until the permissive strains – whose growth is dependent on the activity medium is no longer required (Marlière et al., of several consecutive modules – should be easy to 2011; Bouzon et al., 2017; Döring et al., 2018). construct. Following successful evolution of module activity, When direct selection for module activity fails the evolved genomes can be sequenced to identify or results in poor growth, further optimization the accumulated mutations. Te specifc efect of is required (Fig. 14.1). Tis can be achieved by diferent mutations can be interpreted and further modulating the expression levels of the pathway studied by reintroducing them into a non-evolved enzymes (Zelcbuch et al., 2013; Wenk et al., 2018), selection strain.

Curr. Issues Mol. Biol. (2019) Vol. 33 caister.com/cimb 240 | Claassens et al.

Synthetic methanol assimilation tetrahydrofolate (THF) system and glycine pro- Methanol can be produced from diverse fossil and duction solely from threonine cleavage (Yishai et renewable sources and has been proposed as a al., 2017). promising feedstock for industrial applications and As the RuMP cycle (Fig. 14.2a) atracts the microbial growth (Olah et al., 2009; Olah, 2013; most research atention so far, we will focus on Zhang et al., 2017). Hence, biological conversion atempts for its engineering in biotechnological of methanol to products has received considerable hosts. Non-methylotrophic hosts typically lack atention (Schrader et al., 2009; Whitaker et al., only three enzymes of the pathway: methanol 2015; Pfeifenschneider et al., 2017; Bennet et al., dehydrogenase, hexulose-6-phosphate synthase 2018a). As genetic toolboxes for the engineering and 6-phospho-3-hexulosiomerase. Several stud- of most promising natural methylotrophs (e.g. ies have overexpressed these enzymes in E. coli Bacillus methanolicus) are still underdeveloped, (Müller et al., 2015; Price et al., 2016; Whitaker engineering model biotechnological microbes for et al., 2017; Bennet et al., 2018b; Gonzalez et al., growth on methanol has become an atractive 2018) and Corynebacterium glutamicum (Leßmeier target. et al., 2015; Withof et al., 2015) and have dem- Tree native pathways are known to support onstrated methanol assimilation, albeit at low growth on methanol: the ribulose monophosphate rates and assimilation efciencies. Tese eforts, (RuMP) cycle, the xylulose monophosphate cycle, however, did not apply a selection strategy, that is, and the serine pathway (Kato et al., 2006; Chistoser- cellular growth was not dependent on methanol dova et al., 2009). Among these, the engineering assimilation. of the RuMP cycle has received most atention, as Te RuMP cycle can be divided into three main this route supports the highest yield (Bar-Even et modules: methanol oxidation to formaldehyde, al., 2013). Te heterologous establishment of the formaldehyde assimilation into central metabolism, serine pathway has not yet been reported and only and regeneration of the acceptor metabolite ribulose a single study aimed at engineering the xylulose 5-phosphate (Ru5P) (Fig. 14.2b). In the assimila- monophosphate pathway in Saccharomyces cerevi- tion module, formaldehyde is condensed with siae (Dai et al., 2017). Ru5P to generate F6P, which can be metabolized A synthetic methanol assimilation pathway has to all biomass building blocks. Te assimilation and also been proposed, where three formaldehyde regeneration modules can be supported by several molecules are condensed to dihydroxyacetone by a alternative metabolic structures, the most efcient rationally engineered formolase enzyme. Tis path- one uses glycolysis and the non-oxidative pentose way was recently introduced in E. coli, but did not phosphate pathway (Quayle and Ferenci, 1978; lead to substantial methanol assimilation, probably Zhang et al., 2017). due to the poor kinetics of the formolase enzyme Several recent studies have atempted direct (Wang et al., 2017). Another proposed synthetic selection for the activities of the methanol oxida- methanol condensation cycle, which was dem- tion module and the formaldehyde assimilation onstrated in vitro, consists of the RuMP cycle module in E. coli (Chen et al., 2018; Meyer et al., combined with non-oxidative glycolysis (Bogorad 2018) and C. glutamicum (Tuyishime et al., 2018). et al., 2014). In this pathway, fructose-6-phosphate Tese studies generated a selection strain in which (F6P) is cleaved by phosphoketolases to produce ribose 5-phosphate isomerase is deleted (ΔrpiAB). acetyl-CoA, bypassing pyruvate decarboxylation Tis knockout blocked growth on xylose or gluco- and preventing carbon loss. Finally, an alternative, nate (the later with the additional deletion of edd, synthetic structure of the serine pathway was pro- encoding phosphogluconate dehydratase). Te posed, in which serine is deaminated to pyruvate, methanol oxidation and formaldehyde assimila- and glycine is recycled via threonine biosynthesis tion modules were expected to rescue growth on and degradation, further generating acetyl-CoA as xylose or gluconate by enabling their assimilation a biomass precursor (Bar-Even, 2016). A recent via a ‘RuMP shunt’, a linear route converting Ru5P study describes the successful selection for key to F6P via condensation with formaldehyde (Fig. metabolic modules of this synthetic route, that 14.2c and d). Yet, none of these studies was able is, formate assimilation into serine via the to demonstrate the required activity upon direct

Curr. Issues Mol. Biol. (2019) Vol. 33 caister.com/cimb Strategies for Synthetic C1 Assimilation | 241

(a) (b)

formaldehyde production

formaldehyde ribulose-5P assimilation regeneration

xylose

(e) (f) 3% of biomass 14% of biomass carbon from C1 carbon from C1

xylose xylose

acetate

(g) (h) 16% of biomass 14% of biomass carbon from C1 carbon from C1

xylose xylose

Figure 14.2 Engineering the RuMP shunt in E. coli. (a) Metabolic structure of a variant of the RuMP cycle. (b) Subdivision of the RuMP cycle into metabolic modules; (c, d) Selection schemes for methanol-dependent growth of E. coli via the RuMP shunt as described by Meyer et al. (2018) (c) and Chen et al. (2018) (d). (e–h) Selection schemes for growth via the RuMP shunt at diferent selection strengths as described in He et al. (2018). Pie charts indicate the minimum fraction of carbons in biomass that are derived from C1 (formaldehyde or methanol) in the diferent selection schemes, as calculated using (Neidhardt et al., 1990). Gene deletions are shown in red; overexpressed enzymes in purple; enzymes that mutated during ALE in brown; metabolites that are dependent on the RuMP shunt in green; and substrates that have to be co-fed in blue. Enzyme abbreviations: aceA, isocitrate lyase; aceB, malate synthase B; edd, phosphogluconate dehydratase; fbp; fructose-1,6-bisphosphatase; frmRAB, glutathione-dependent formaldehyde detoxifcation system; glcB, malate synthase G; glpX, fructose-1,6-bisphosphatase 2; HPS, 3-hexulose-6-phosphate synthase; maldh, malate dehydrogenase; MDH, methanol dehydrogenase; PHI, 6-phospho-3-hexuloisomerase; rpiAB, ribose 5-phosphate isomerase; SOX, sarcosine oxidase; tktAB, transketolase A and B; zwf, glucose 6-phosphate dehydrogenase. Metabolite abbreviations: A3H6P, arabino 3-hexulose 6-phosphate; DHAP, dihydroxyacetone phosphate; E4P, erythrose 4-phosphate; F6P, fructose 6-phosphate; FBP, fructose bisphosphate; GAP, glyceraldehyde 3-phosphate; 6PGC, 6-phospho-gluconate; 6PGL, 6-phospho-glucono-lactone; R5P, ribose 5-phosphate; Ru5P, ribulose 5-phosphate; S7P, sedoheptulose 7-phosphate; Xu5P, xylulose 5-phosphate.

Curr. Issues Mol. Biol. (2019) Vol. 33 caister.com/cimb 242 | Claassens et al.

selection, and ALE was necessary to establish we obtained with the ΔfmRB ΔrpiAB Δzwf strain methanol-assimilation-dependent growth. was considerably poorer than that observed using Te strict requirement for ALE to achieve RuMP the ΔfmRB ΔtktAB Δzwf strain. Tis serves as shunt-dependent growth can be atributed to the a clear demonstration of the importance of the unfavorability of methanol oxidation. Supporting metabolic context for establishing activity of a this notion, all of these studies found mutations newly introduced pathway. In this specifc case, linked to a decreased NADH/NAD ratio, e.g. by the poor growth associated with the ΔfmRB interrupting or inhibiting the TCA cycle (Chen ΔrpiAB Δzwf strain probably stems from costly et al., 2018; Meyer et al., 2018; Tuyishime et al., metabolism of F6P ‘back’ to xylulose-5-phosphate 2018). Tis likely relates to the fact that methanol and Ru5P, and potentially from the inhibition of oxidation is thermodynamically limited by a high essential transketolase reactions due to the accu- NADH/NAD ratio. Using deuterated methanol mulation of the coproduct xylulose-5-phosphate

(CD3OD), it was shown that methanol dehydroge- (He et al., 2018). nase also kinetically limits the activity of the RuMP Te major challenge of establishing a fully func- shunt in E. coli (Woolston et al., 2018a). Methanol tional RuMP cycle lies in the regeneration module. oxidation could be improved by identifying or To make things even more difcult, the establish- engineering kinetically superior variants, which ment of autocatalytic cycles, such as the RuMP could be directly screened using a formaldehyde cycle, requires the kinetic parameters of enzymes biosensor (Woolston et al., 2018b) or selected for to be carefully balanced as to avoid draining the in appropriate selection strains. pathway intermediates (Barenholz et al., 2017). A To overcome the barriers associated with recent study suggested an interesting way to tackle methanol oxidation, we decided to separate the this challenge, by overexpressing the irreversible formaldehyde production module from that of SBPase to force fux towards the regeneration of formaldehyde assimilation, such that we could Ru5P (Woolston et al., 2018a). Another approach test the later in more detail. Towards this aim, we is to overexpress the enzymes of the pentose phos- replaced methanol dehydrogenase with the kineti- phate pathway from an organism that supports cally efcient and thermodynamically favourable more efcient regeneration of Ru5P (Bennet et sarcosine oxidase that metabolizes sarcosine to for- al., 2018b). Still, as was previously shown with maldehyde and glycine (He et al., 2018). We further ALE of E. coli to achieve a functional Calvin Cycle, constructed several selection strains, imposing dif- down-regulation of branching enzymes, e.g. ribose- ferent levels of selective pressure on formaldehyde phosphate diphosphokinase, might be necessary to assimilation via the RuMP shunt. Tese included establish a sustainable cyclic fux (Antonovsky et al., (i) ΔfmRB ΔrpiAB Δfp ΔglpX Δzwf strain in 2017; Herz et al., 2017). which the synthetic shunt provides only essential sugar phosphates while succinate serves as the main growth substrate (Fig. 14.2e); (ii) ΔfmRB ΔrpiAB Synthetic formate assimilation ΔaceBA ΔglcB Δzwf strain in which the RuMP shunt Formate is a promising microbial feedstock that can

provides almost all cellular building blocks while be efciently produced from CO2 by electrochemi- acetate oxidation provides reducing power and cal and photochemical processes (Kopljar et al., energy (Fig. 14.2f); (iii) ΔfmRB ΔrpiAB Δzwf 2016; Zhou et al., 2016; Yang et al., 2017). Formate strain, where the shunt is responsible for satisfying can also be obtained from partial oxidation of bio-

all cellular carbon and energy needs (Fig. 14.2g); mass or natural gas and by hydrogenation of CO2 and (iv) ΔfmRB ΔtktAB Δzwf strain in which (Shen et al., 2015; Wang et al., 2015). almost all cellular carbon and energy needs are sup- Microbial growth on formate as a carbon and ported by the RuMP shunt (Fig. 14.2h). energy source is reported for diverse groups of We were able to directly select for growth of microorganisms. Acetogenic and methanogenic all of these strains via the RuMP shunt without microbes can grow on formate using the reductive the need for ALE (He et al., 2018), most probably acetyl-CoA pathway (Kerby and Zeikus, 1987), because our selection scheme bypasses the chal- representing the most efcient way to convert this lenge of methanol oxidation. However, the growth feedstock into a product (Bar-Even et al., 2013).

Curr. Issues Mol. Biol. (2019) Vol. 33 caister.com/cimb Strategies for Synthetic C1 Assimilation | 243

However, this anaerobic pathway can support only et al., 2017). Its limited overlap with central metab- a narrow product spectrum (Bertsch and Müller, olism and its very high ATP-efciency further make 2015). Aerobic growth on formate is sustained by the reductive glycine pathway especially promising the Calvin cycle and the serine pathway; however, to support aerobic formate assimilation. the high ATP costs of these pathways reduce the To facilitate the implementation of the reductive potential biomass and product yields (Bar-Even glycine pathway in E. coli, we divided it into four et al., 2013). To overcome this problem, several modules (Fig. 14.3a): (i) a C1 activation module in synthetic pathways have been designed to support which formate is condensed with the THF system high yields under aerobic conditions (Bar-Even et and reduced to 5,10-methylene-THF; (ii) a glycine al., 2013; Siegel et al., 2015; Bar-Even, 2016). biosynthesis module that condenses the C1 moiety

In the synthetic formolase pathway, formate from 5,10-methylene-THF with CO2 and ammo- is frst reduced to formaldehyde by promiscuous nia to generate the C2 metabolite glycine; (iii) a activity of two enzymes: acetyl-CoA synthase serine biosynthesis module that condenses glycine and acetylaldehyde dehydrogenase (Siegel et al., with another C1 moiety to give the C3 metabolite 2015). Ten, three formaldehyde molecules are serine; and (iv) a serine assimilation module that condensed into dihydroxyacetone by the formo- deaminates this amino acid to produce pyruvate as lase enzyme described before. Te activity of the a biomass precursor. pathway was demonstrated in vitro, but its in vivo We constructed several selection strains to functionality was very poor due to the low activities demonstrate the activities of the pathway modules of formaldehyde dehydrogenase and the formolase in E. coli (Yishai et al., 2017, 2018). First, we con- enzyme (Siegel et al., 2015). To boost the activi- structed a C1-auxotroph strain (ΔglyA ΔgcvTHP) ties of these limiting enzymes, a modular selection and showed that overexpression of formate THF strategy can be used. For example, a module ligase (FTL) enabled formate to serve as sole responsible for formate reduction to formaldehyde source of all C1-dependent building blocks, includ- can be tested in a formaldehyde-dependent strain ing purines, thymidine, and methionine (Fig. – for example, the RuMP shunt-dependent strains 14.3b). Tis confrmed the efcient activity of the described above – and ALE can be used to increase C1-activation module. Next, we selected for the this activity. Similarly, activity of the formolase combined activity of the C1-activation module and enzyme might be initially tested and optimized in the serine biosynthesis module. Towards this aim, a strain which produces formaldehyde via sarco- we constructed a strain deleted in 3-phosphoglyc- sine oxidation and assimilate dihydroxyacetone to erate dehydrogenase (ΔserA, the frst enzyme of provide only a fraction of cellular building blocks. serine biosynthesis) and the glycine cleavage system For example, by deleting phosphoglycerate mutase, (ΔgcvTHP). Only upon overexpression of FTL central metabolism can be divided into upper and and the native bifunctional 5,10-methenyl-THF lower segments, where succinate provides carbon cyclohydrolase/5,10-methylene-THF dehydroge- and energy for lower metabolism and dihydroxy- nase (FolD), glycine and formate could serve as acetone phosphorylation provides carbon only C1 and serine sources to support cell growth (Fig. for upper metabolism (Zelcbuch et al., 2015). Tis 14.3c). Tis demonstrates that testing a module – would impose moderate selection for the activity in this case the C1 activation module – in diferent of the formolase enzyme and would thus be a more metabolic contexts is important to uncover hidden reasonable initial target. botlenecks: while the endogenous activity of FolD A more feasible approach to establish for- sufced for the frst selection strain, the higher mate assimilation would be to focus on existing activity required for the growth of the second selec- enzymatic activities that could be combined tion strain necessitated dedicated overexpression of to realize a new pathway. Tis is exactly the the enzyme (Yishai et al., 2017). case of the synthetic reductive glycine pathway In a follow-up study, we selected for the activi- (Fig. 14.3a) (Bar-Even et al., 2013). All the reactions ties of the C1-assimilation module and the glycine of this pathway are catalysed by known and ubiqui- biosynthesis module (Yishai et al., 2018). We tous enzymes. It is even possible that the pathway constructed a strain auxotrophic to C1 and glycine operates endogenously in some microbes (Figueroa (ΔglyA ΔltaE Δkbl ΔaceA). We could establish

Curr. Issues Mol. Biol. (2019) Vol. 33 caister.com/cimb 244 | Claassens et al.

(b) (c) 3% of biomass 4% of biomass formate carbon from C1 formate carbon from C1 (a) THF FTL THF FTL 10-formyl-THF 10-formyl-THF FolD formate 5,10-methylene-THF ATP,THF 5,10-methylene-THF

ADP,Pi CO2 NH3 CO2 NH3 C1-activation 10-formyl-THF ΔgcvTHP ΔgcvTHP THF THF H O 2 glycine glycine 5,10-methenyl-THF ΔglyA THF GlyA THF NAD(P)H NAD+ serine serine 5,10-methylene-THF dihydro-LP ΔserA THF NH3 NAD+ GLYCOLYSIS/ GLYCOLYSIS/ aminomethyl- TCA CYCLE TCA CYCLE serine dihydro-LP NADH CO glycine biosynthesis 2 biosynthesis LP glucose glucose glycine (d) (e) THF 8% of biomass 11% of biomass formate serine formate carbon from C1 carbon from C1 H O THF THF serine 2 FTL FTL NH assimilation 3 10-formyl-THF 10-formyl-THF pyruvate MtdA + Fch MtdA + Fch 5,10-methylene-THF 5,10-methylene-THF

CO2 NH3 CO2 NH3 gcvTHP Δkbl gcvTHP Δkbl THF ΔltaE THF ΔltaE glycine threonine glycine threonine

ΔglyA THF THF

serine serine

ΔserA GLYCOLYSIS/ GLYCOLYSIS/ TCA CYCLE TCA CYCLE

glucose glucose

Figure 14.3 Engineering the reductive glycine pathway in E. coli. (a) Metabolic scheme of the reductive glycine pathway and its subdivision into modules; (b–e) Selection schemes for the activity of diferent modules of the reductive glycine pathway, as demonstrated in E. coli (Yishai et al., 2017, 2018). Pie charts indicate the minimum

fraction of carbons in biomass that are derived from formate and CO2 in the diferent selection schemes, as calculated using (Neidhardt et al., 1990). Gene deletions are shown in red; overexpressed foreign enzymes

in purple; overexpressed native enzymes in grey; metabolites that are produced from formate and CO2 in green; and substrates that have to be co-fed in blue. Abbreviations: Fch, 5,10-methenyl-THF cyclohydrolase; FolD, bifunctional 5,10-methenyl-THF cyclohydrolase/5,10-methylene-THF dehydrogenase; FTL, formate- THF ligase; gcvTHP(L), glycine cleavage system subunits T, H and P (and lipoamide dehydrogenase subunit); glyA, serine hydroxymethyltransferase; ltaE, threonine aldolase; kbl, 2-amino-3-ketobutyrate CoA ligase; LP, lipoamide-protein; MtdA, 5,10-methylene-THF dehydrogenase; serA, 3-phosphoglycerate dehydrogenase; THF, tetrahydrofolate.

formate conversion to C1-activated-THF and a C1-glycine-serine auxotrophic strain (ΔserA glycine upon overexpression of the enzymes ΔltaE Δkbl ΔaceA) (Fig. 14.3e). Overexpression of the native glycine cleavage system as well as of enzymes mentioned above resulted in C1, gly- FTL, 5,10-methenyl-THF cyclohydrolase, and cine and serine production solely from formate

5,10-methylene-THF dehydrogenase from Methy- and CO2. Overall, ≈ 10% of carbons in biomass lobacterium extorquens (Fig. 14.3d). Te later two were provided by the pathway and the fast growth enzymes were necessary to replace the native FolD, obtained (doubling rate of ≈ 1.7 hours) indicates whose activity was too low to support the required high activity of all pathway components. fux even when overexpressed, probably since it is Another recent study also focused on modular inhibited by the key intermediate 10-formyl-THF engineering of the reductive glycine pathway in E. (Yishai et al., 2018). Next, to select for the activity coli (Tashiro et al., 2018). Similar to the approach of three of the pathway modules, we constructed described above, this study demonstrated that

Curr. Issues Mol. Biol. (2019) Vol. 33 caister.com/cimb Strategies for Synthetic C1 Assimilation | 245 the overexpression of three foreign genes from Conclusions Clostridium ljungdahlii can support C1 and serine Recent eforts using modular and selection-based biosynthesis from formate and glycine (Fig. 14.3c). engineering approach have enabled C1-routes to However, for demonstration of glycine and serine support the biosynthesis of a substantial fraction biosynthesis via the reductive activity of the gly- of the host’s biomass. Tese studies demonstrate cine cleavage system, leaky auxotrophic strains a general approach of iterative design, build, test were employed in which only serA or glyA were and learn cycles. First, pathway modules and selec- deleted, while threonine cleavage to glycine (via tion strains are designed and constructed. Testing Kbl or LtaE) was lef untouched. Consequently, as module activity using these selection strains pro- is supported by the labelling data, in these strains, vides insights into the metabolic constraints that the majority of glycine and serine were most prob- limit fux, from which beter designs and improved ably derived from threonine cleavage rather than activity can emerge. Selection strains are therefore formate assimilation. an essential tool for these cycles, as they provide In a parallel efort to the rational engineering simple performance readout that facilitates optimi- approach, we demonstrated the establishment of zation of module and pathway activity. the three modules of the reductive glycine pathway As mentioned above for the RuMP cycle, a major using ALE (Döring et al., 2018). Tis work used challenge for establishing C1 pathways relates to the same tight selection strategies as above but the regulation of fux that is diverted away from the introduced only a single foreign enzyme: Clostrid- pathway towards other biosynthesis routes. Tis is ial FTL that was integrated into the genome. Te especially true for autocatalytic cycles in which the cells were cultivated continuously and provided product is also an intermediate of the pathway. To with two types of alternating media: a selection realize stable activity, proper balancing is required medium (which does not contain the auxotrophy- between the rates of pathway enzymes and those relieving compounds) that was added when the of the branching enzymes. As previously demon- culture turbidity surpassed a threshold, and a strated, for autocatalytic cycles, this likely requires permissive medium (containing the compounds ensuring low afnities of the branching enzymes for which the strains are auxotrophic) that was towards the pathway metabolites (Barenholz et al., supplied upon decrease of the culture turbidity 2017). Indeed, the use of ALE for the establishment below this threshold. Tis procedure was used to of a functional Calvin cycle in E. coli resulted in select for growth on the stressing medium and, lower activity of branching reactions (Antonovsky using sequential rounds of evolution, established et al., 2016). Fortunately, branching reactions are metabolism of formate to C1-activated-THF, gly- easy to identify such that a rational design approach cine, and, fnally, serine. Genome sequencing of might be able to achieve the same goals as ALE, these strains revealed several mutations, including e.g. genetic or protein engineering aiming at lower in the coding region of the key limiting enzyme expression levels or afnities of branching enzymes FolD. Several genes were duplicated in the genome, could directly establish the desired growth pheno- including FTL and FolD, probably to increase type. their expression levels. Tis study confrms that Most eforts to establish synthetic C1 metabo- the same metabolic goals can be achieved either lism have so far focused on the RuMP cycle, the via a rational engineering approach or via long reductive glycine pathway, and the formolase path- term evolution. way. Yet, given the diversity of metabolic solutions To complete the establishment of the reductive for C1 assimilation, more pathways are likely to glycine pathway, the next challenge is to integrate be designed and tested, and the modular selection the frst three modules – converting formate to approach described here would be vital for their serine – with the last one, that is, serine assimila- implementation. Currently, E. coli serves as the tion into central metabolism. Tis would require prime host for engineering synthetic C1 assimila- an order of magnitude increase in the fux via the tion, mainly due to the highly developed genetic pathway and would most probably require ALE toolbox for its engineering and our extensive to adapt the cellular physiology for the novel knowledge of its physiology and metabolism. How- growth mode. ever, other microbial hosts might prove to be more

Curr. Issues Mol. Biol. (2019) Vol. 33 caister.com/cimb 246 | Claassens et al.

suitable. Specifcally, some microbes can produce Acta 1827, 1039–1047. htps://doi.org/10.1016/j. bbabio.2012.10.013 specifc chemicals beter than E. coli and others Barenholz, U., Davidi, D., Reznik, E., Bar-On, Y., Antonovsky, display higher tolerance towards substrates or N., Noor, E., and Milo, R. (2017). Design principles of products. For example Corynebacterium glutamicum autocatalytic cycles constrain enzyme kinetics and force is arguably the best host to produce amino acids low substrate saturation at fux branch points. Elife 6, e20667. htps://doi.org/10.7554/eLife.20667 (Wendisch et al., 2006), and Saccharomyces cerevi- Bennet, R.K., Steinberg, L.M., Chen, W., and Papoutsakis, siae can tolerate high concentrations of the formate E.T. (2018a). Engineering the bioconversion of methane feedstock and alcohol products (Overkamp et and methanol to fuels and chemicals in native and al., 2002; Mohd Azhar et al., 2017). Also, some synthetic methylotrophs. Curr. Opin. Biotechnol. 50, 81–93. metabolic modules can directly integrate with C1 Bennet, R.K., Gonzalez, J.E., Whitaker, W.B., Antoniewicz, metabolism, making microbes that naturally har- M.R., and Papoutsakis, E.T. (2018b). Expression bour these beter hosts. A primary example is the of heterologous non-oxidative pentose phosphate metal-dependent formate dehydrogenase, which pathway from Bacillus methanolicus and phosphoglucose isomerase deletion improves methanol assimilation and can support highly efcient utilization of formate as metabolite production by a synthetic Escherichia coli cellular energy source but is difcult to introduce methylotroph. Metab. Eng. 45, 75–85. into a new host (Maia et al., 2017). Bertsch, J., and Müller, V. (2015). Bioenergetic constraints In the next few years, we will likely witness the for conversion of syngas to biofuels in acetogenic bacteria. Biotechnol. Biofuels 8, 210. htps://doi. completion of synthetic C1 metabolism in model org/10.1186/s13068-015-0393-x microbes. While we cannot predict which pathway Bogorad, I.W., Chen, C.T., Teisen, M.K., Wu, T.Y., will prove to be the most benefcial in the long term, Schlenz, A.R., Lam, A.T., and Liao, J.C. (2014). it is clear that eforts such as those described in this Building carbon-carbon bonds using a biocatalytic methanol condensation cycle. Proc. Natl. Acad. Sci. review are bringing efcient bioproduction from C U.S.A. 111, 15928–15933. htps://doi.org/10.1073/ feedstocks closer. pnas.1413470111 Bouzon, M., Perret, A., Loreau, O., Delmas, V., Perchat, Acknowledgements N., Weissenbach, J., Taran, F., and Marlière, P. (2017). A Synthetic Alternative to Canonical One-Carbon Te authors thank Charles Coton for his feedback Metabolism. ACS Synth. Biol. 6, 1520–1533. htps:// and critical reading of this manuscript. Tis work doi.org/10.1021/acssynbio.7b00029 was funded by the Max Planck Society and European Burgard, A.P., Pharkya, P., and Maranas, C.D. (2003). Commission Horizon 2020 grant eForFuel. Nico Optknock: a bilevel programming framework for identifying gene knockout strategies for microbial strain Claassens is supported by Te Netherlands Organi- optimization. Biotechnol. Bioeng. 84, 647–657. htps:// sation for Scientifc Research (NWO) through a doi.org/10.1002/bit.10803 Rubicon Grant (Project 019.163LW.035). Hai He Chen, C.T., Chen, F.Y., Bogorad, I.W., Wu, T.Y., Zhang, R., is funded by the China Scholarship Council. Lee, A.S., and Liao, J.C. (2018). Synthetic methanol auxotrophy of Escherichia coli for methanol-dependent growth and production. Metab. Eng. 49, 257–266. References Chistoserdova, L., Kalyuzhnaya, M.G., and Lidstrom, Antonovsky, N., Gleizer, S., Noor, E., Zohar, Y., Herz, E., M.E. (2009). Te expanding world of methylotrophic Barenholz, U., Zelcbuch, L., Amram, S., Wides, A., metabolism. Annu. Rev. Microbiol. 63, 477–499. htps:// Tepper, N., et al. (2016). Sugar synthesis from CO2 doi.org/10.1146/annurev.micro.091208.073600 in Escherichia coli. Cell 166, 115–125. htps://doi. Claassens, N.J., Sánchez-Andrea, I., Sousa, D.Z., and Bar- org/10.1016/j.cell.2016.05.064 Even, A. (2018). Towards sustainable feedstocks: A Antonovsky, N., Gleizer, S., and Milo, R. (2017). Engineering guide to electron donors for microbial carbon fxation. carbon fxation in E. coli: from heterologous RuBisCO Curr. Opin. Biotechnol. 50, 195–205. expression to the Calvin-Benson-Bassham cycle. Curr. Clomburg, J.M., Crumbley, A.M., and Gonzalez, R. (2017). Opin. Biotechnol. 47, 83–91. Industrial biomanufacturing: Te future of chemical Bar-Even, A. (2016). Formate assimilation: Te metabolic production. Science 355, aag0804. architecture of natural and synthetic pathways. Dai, Z., Gu, H., Zhang, S., Xin, F., Zhang, W., Dong, W., Ma, Biochemistry 55, 3851–3863. htps://doi.org/10.1021/ J., Jia, H., and Jiang, M. (2017). Metabolic construction acs.biochem.6b00495 strategies for direct methanol utilization in Saccharomyces Bar-Even, A., Noor, E., Flamholz, A., and Milo, R. cerevisiae. Bioresour. Technol. 245, 1407–1412. htps:// (2013). Design and analysis of metabolic pathways doi.org/10.1016/j.biortech.2017.05.100 supporting formatotrophic growth for electricity- Döring, V., Darii, E., Yishai, O., Bar-Even, A., and Bouzon, dependent cultivation of microbes. Biochim. Biophys. M. (2018). Implementation of a reductive route of one- carbon assimilation in Escherichia coli through directed

Curr. Issues Mol. Biol. (2019) Vol. 33 caister.com/cimb Strategies for Synthetic C1 Assimilation | 247

evolution. ACS Synth. Biol. 7, 2029–2036. htps://doi. Leßmeier, L., Pfeifenschneider, J., Carnicer, M., Heux, S., org/10.1021/acssynbio.8b00167 Portais, J.C., and Wendisch, V.F. (2015). Production Dürre, P., and Eikmanns, B.J. (2015). C1-carbon sources of carbon-13-labeled cadaverine by engineered for chemical and fuel production by microbial gas Corynebacterium glutamicum using carbon-13-labeled fermentation. Curr. Opin. Biotechnol. 35, 63–72. methanol as co-substrate. Appl. Microbiol. Biotechnol. htps://doi.org/10.1016/j.copbio.2015.03.008 99, 10163–10176. htps://doi.org/10.1007/s00253- Erb, T.J., Jones, P.R., and Bar-Even, A. (2017). Synthetic 015-6906-5 metabolism: metabolic engineering meets enzyme Maia, L.B., Moura, I., and Moura, J.J.G. (2017). design. Curr. Opin. Chem. Biol. 37, 56–62. Molybdenum and tungsten-containing formate Fei, Q., Guarnieri, M.T., Tao, L., Laurens, L.M., Dowe, N., dehydrogenases: Aiming to inspire a catalyst for carbon and Pienkos, P.T. (2014). Bioconversion of natural gas dioxide utilization. Inorganica Chim. Acta 455, 350– to liquid fuel: opportunities and challenges. Biotechnol. 363. htps://doi.org/10.1016/j.ica.2016.07.010 Adv. 32, 596–614. htps://doi.org/10.1016/j. Marlière, P., Patrouix, J., Döring, V., Herdewijn, P., Tricot, biotechadv.2014.03.011 S., Cruveiller, S., Bouzon, M., and Mutzel, R. (2011). Figueroa, I.A., Barnum, T.P., Somasekhar, P.Y., Carlström, Chemical evolution of a bacterium’s genome. Angew. C.I., Engelbrektson, A.L., and Coates, J.D. (2018). Chem. Int. Ed. Engl. 50, 7109–7114. htps://doi. Metagenomics-guided analysis of microbial org/10.1002/anie.201100535 chemolithoautotrophic phosphite oxidation yields Marmiesse, L., Peyraud, R., and Cotret, L. (2015). evidence of a seventh natural CO2 fxation pathway. FlexFlux: combining metabolic fux and regulatory Proc. Natl. Acad. Sci. U.S.A. 115, E92–E101. htps:// network analyses. BMC Syst. Biol. 9, 93. htps://doi. doi.org/10.1073/pnas.1715549114 org/10.1186/s12918-015-0238-z Gonzalez, J.E., Bennet, R.K., Papoutsakis, E.T., and Martín, A.J., Larrazábal, G.O., and Pérez-Ramírez, J. (2015). Antoniewicz, M.R. (2018). Methanol assimilation in Towards sustainable fuels and chemicals through the Escherichia coli is improved by co-utilization of threonine electrochemical reduction of CO2: lessons from water and deletion of leucine-responsive regulatory protein. electrolysis. Green Chem. 17, 5114–5130. htps://doi. Metab. Eng. 45, 67–74. org/10.1039/C5GC01893E He, H., Edlich-Muth, C., Lindner, S.N., and Bar-Even, A. Meyer, F., Keller, P., Hartl, J., Gröninger, O.G., Kiefer, P., (2018). Ribulose monophosphate shunt provides nearly and Vorholt, J.A. (2018). Methanol-essential growth all biomass and energy required for growth of E. coli. ACS of Escherichia coli. Nat. Commun. 9, 1508. htps://doi. Synth. Biol. 7, 1601–1611. htps://doi.org/10.1021/ org/10.1038/s41467-018-03937-y acssynbio.8b00093 Mohd Azhar, S.H., Abdulla, R., Jambo, S.A., Marbawi, H., Henstra, A.M., Sipma, J., Rinzema, A., and Stams, A.J. Gansau, J.A., Mohd Faik, A.A., and Rodrigues, K.F. (2007). Microbiology of synthesis gas fermentation (2017). Yeasts in sustainable bioethanol production: A for biofuel production. Curr. Opin. Biotechnol. 18, review. Biochem. Biophys. Rep. 10, 52–61. htps://doi. 200–206. org/10.1016/j.bbrep.2017.03.003 Herz, E., Antonovsky, N., Bar-On, Y., Davidi, D., Gleizer, S., Müller, J.E.N., Meyer, F., Litsanov, B., Kiefer, P., Pothof, Prywes, N., Noda-Garcia, L., Lyn Frisch, K., Zohar, Y., E., Heux, S., Quax, W.J., Wendisch, V.F., Brautaset, T., Wernick, D.G., et al. (2017). Te genetic basis for the Portais, J.C., et al. (2015). Engineering Escherichia coli adaptation of E. coli to sugar synthesis from CO2. Nat. for methanol conversion. Metab. Eng. 28, 190–201. Commun. 8, 1705. htps://doi.org/10.1038/s41467- Neidhardt, F.C., Ingraham, J.L., and Schaechter, M. 017-01835-3 (1990). Building blocks needed to produce 1g of E. Jouny, M., Luc, W.W., and Jiao, F. (2018). A general techno- coli protoplasm. In Physiology of the Bacterial Cell: economic analysis of CO2 electrolysis systems. Ind. Eng. A Molecular Approach, Neidhardt, F.C., Ingraham, Chem. Res. 57, 2165–2177. htps://doi.org/10.1021/ J.L., and Schaechter, M., eds. (Sinaver Associates, acs.iecr.7b03514 Sunderland, MA), pp. 134–143. Kato, N., Yurimoto, H., and Tauer, R.K. (2006). Te Olah, G.A. (2013). Towards oil independence through physiological role of the ribulose monophosphate renewable methanol chemistry. Angew. Chem. Int. pathway in bacteria and archaea. Biosci. Biotechnol. Ed. Engl. 52, 104–107. htps://doi.org/10.1002/ Biochem. 70, 10–21. anie.201204995 Kerby, R., and Zeikus, J.G. (1987). Anaerobic catabolism Olah, G.A., Goeppert, A., and Prakash, G.K.S. (2009). of formate to acetate and CO2 by Butyribacterium Beyond Oil and Gas: Te Methanol Economy, Second methylotrophicum. J. Bacteriol. 169, 2063–2068. Edition. (Wiley VCH, Weinheim), pp. 1–334 htps:// Kopljar, D., Wagner, N., and Klemm, E. (2016). Transferring doi.org/10.1002/9783527627806. electrochemical CO2 reduction from semi-batch Overkamp, K.M., Köter, P., van der Hoek, R., into continuous operation mode using gas difusion Schoondermark-Stolk, S., Lutik, M.A., van Dijken, J.P., electrodes. Chem. Eng. Technol. 39, 2042–2050. and Pronk, J.T. (2002). Functional analysis of structural htps://doi.org/10.1002/ceat.201600198 genes for NAD(+)-dependent formate dehydrogenase Kumar, B., Llorente, M., Froehlich, J., Dang, T., Sathrum, in Saccharomyces cerevisiae. Yeast 19, 509–520. A., and Kubiak, C.P. (2012). Photochemical and Pfeifenschneider, J., Brautaset, T., and Wendisch, photoelectrochemical reduction of CO2. Annu. Rev. V.F. (2017). Methanol as carbon substrate in the Phys. Chem. 63, 541–569. htps://doi.org/10.1146/ bio-economy: Metabolic engineering of aerobic annurev-physchem-032511-143759 methylotrophic bacteria for production of value-added

Curr. Issues Mol. Biol. (2019) Vol. 33 caister.com/cimb 248 | Claassens et al.

chemicals. Biofuels Bioprod. Biorefning 11, 719–731. based on the biology of aerobic methanol utilization. htps://doi.org/10.1002/bbb. Curr. Opin. Biotechnol. 33, 165–175. htps://doi. Portnoy, V.A., Bezdan, D., and Zengler, K. (2011). org/10.1016/j.copbio.2015.01.007 Adaptive laboratory evolution – harnessing the power Whitaker, W.B., Jones, J.A., Bennet, R.K., Gonzalez, J.E., of biology for metabolic engineering. Curr. Opin. Vernacchio, V.R., Collins, S.M., Palmer, M.A., Schmidt, Biotechnol. 22, 590–594. htps://doi.org/10.1016/j. S., Antoniewicz, M.R., Kofas, M.A., et al. (2017). copbio.2011.03.007 Engineering the biological conversion of methanol to Price, J.V., Chen, L., Whitaker, W.B., Papoutsakis, E., and specialty chemicals in Escherichia coli. Metab. Eng. 39, Chen, W. (2016). Scafoldless engineered enzyme 49–59. assembly for enhanced methanol utilization. Proc. Withof, S., Schmitz, K., Niedenführ, S., Nöh, K., Noack, Natl. Acad. Sci. U.S.A. 113, 12691–12696. htps://doi. S., Bot, M., and Marienhagen, J. (2015). Metabolic org/10.1073/pnas.1601797113 engineering of Corynebacterium glutamicum for Quayle, J.R., and Ferenci, T. (1978). Evolutionary aspects of methanol metabolism. Appl. Environ. Microbiol. 81, autotrophy. Microbiol. Rev. 42, 251–273. 2215–2225. htps://doi.org/10.1128/AEM.03110-14 Schrader, J., Schilling, M., Holtmann, D., Sell, D., Filho, Woolston, B.M., King, J.R., Reiter, M., Van Hove, B., and M.V., Marx, A., and Vorholt, J.A. (2009). Methanol- Stephanopoulos, G. (2018a). Improving formaldehyde based industrial biotechnology: current status and consumption drives methanol assimilation in future perspectives of methylotrophic bacteria. Trends engineered E. coli. Nat. Commun. 9, 2387. htps://doi. Biotechnol. 27, 107–115. htps://doi.org/10.1016/j. org/10.1038/s41467-018-04795-4 tibtech.2008.10.009 Woolston, B.M., Roth, T., Kohale, I., Liu, D.R., and Schwander, T., Schada von Borzyskowski, L., Burgener, S., Stephanopoulos, G. (2018b). Development of a Cortina, N.S., and Erb, T.J. (2016). A synthetic pathway formaldehyde biosensor with application to synthetic for the fxation of carbon dioxide in vitro. Science 354, methylotrophy. Biotechnol. Bioeng. 115, 206–215. 900–904. htps://doi.org/10.1002/bit.26455 Shen, Y., Jarboe, L., Brown, R., and Wen, Z. (2015). A Yang, H., Kaczur, J.J., Sajjad, S.D., and Masel, R.I. (2017). thermochemical-biochemical hybrid processing CO2 Conversion to formic acid in a three compartment of lignocellulosic biomass for producing fuels and cell with SustainionTM membranes. ECS Trans. 77, chemicals. Biotechnol. Adv. 33, 1799–1813. htps://doi. 1425–1431. htps://doi.org/10.1149/07711.1425ecst. org/10.1016/j.biotechadv.2015.10.006 Yishai, O., Goldbach, L., Tenenboim, H., Lindner, S.N., Siegel, J.B., Smith, A.L., Poust, S., Wargacki, A.J., Bar-Even, and Bar-Even, A. (2017). Engineered assimilation of A., Louw, C., Shen, B.W., Eiben, C.B., Tran, H.M., Noor, exogenous and endogenous formate in Escherichia E., et al. (2015). Computational protein design enables a coli. ACS Synth. Biol. 6, 1722–1731. htps://doi. novel one-carbon assimilation pathway. Proc. Natl. Acad. org/10.1021/acssynbio.7b00086 Sci. U.S.A. 112, 3704–3709. htps://doi.org/10.1073/ Yishai, O., Bouzon, M., Döring, V., and Bar-Even, A. (2018). pnas.1500545112 In vivo assimilation of one-carbon via a synthetic Tashiro, Y., Hirano, S., Matson, M.M., Atsumi, S., and reductive glycine pathway in Escherichia coli. ACS Kondo, A. (2018). Electrical-biological hybrid system Synth. Biol. 7, 2023–2028. htps://doi.org/10.1021/ for CO2 reduction. Metab. Eng. 47, 211–218. acssynbio.8b00131 Tuyishime, P., Wang, Y., Fan, L., Zhang, Q., Li, Q., Zheng, P., Zelcbuch, L., Antonovsky, N., Bar-Even, A., Levin-Karp, A., Sun, J., and Ma, Y. (2018). Engineering Corynebacterium Barenholz, U., Dayagi, M., Liebermeister, W., Flamholz, glutamicum for methanol-dependent growth and A., Noor, E., Amram, S., et al. (2013). Spanning high- glutamate production. Metab. Eng. 49, 220–231. dimensional expression space using ribosome-binding Wang, W.H., Himeda, Y., Muckerman, J.T., Manbeck, G.F., site combinatorics. Nucleic Acids Res. 41, e98. htps:// and Fujita, E. (2015). CO2 hydrogenation to formate and doi.org/10.1093/nar/gkt151 methanol as an alternative to photo- and electrochemical Zelcbuch, L., Razo-Mejia, M., Herz, E., Yahav, S., Antonovsky, CO2 reduction. Chem. Rev. 115, 12936–12973. htps:// N., Kroytoro, H., Milo, R., and Bar-Even, A. (2015). An doi.org/10.1021/acs.chemrev.5b00197 in vivo metabolic approach for deciphering the product Wang, X., Wang, Y., Liu, J., Li, Q., Zhang, Z., Zheng, P., Lu, specifcity of glycerate kinase proves that both E. coli’s F., and Sun, J. (2017). Biological conversion of methanol glycerate kinases generate 2-phosphoglycerate. PLOS by evolved Escherichia coli carrying a linear methanol ONE 10, e0122957. htps://doi.org/10.1371/journal. assimilation pathway. Bioresour. Bioprocess. 4, 4–9. pone.0122957 htps://doi.org/10.1186/s40643-017-0172-6. Zhang, W., Zhang, T., Wu, S., Wu, M., Xin, F., Dong, W., Wendisch, V.F., Bot, M., and Eikmanns, B.J. (2006). Ma, J., Zhang, M., and Jiang, M. (2017). Guidance Metabolic engineering of Escherichia coli and for engineering of synthetic methylotrophy based on Corynebacterium glutamicum for biotechnological methanol metabolism in methylotrophy. RSC Adv. 7, production of organic acids and amino acids. Curr. 4083–4091. htps://doi.org/10.1039/C6R27038G Opin. Microbiol. 9, 268–274. Zhou, X., Liu, R., Sun, K., Chen, Y., Verlage, E., Francis, S.A., Wenk, S., Yishai, O., Lindner, S.N., and Bar-Even, A. Lewis, N.S., and Xiang, C. (2016). Solar-driven reduction (2018). An engineering approach for rewiring microbial of 1 atm of CO2 to formate at 10% energy-conversion metabolism. Meth. Enzymol. 608, 329–367. efciency by use of a TiO2-protected III-V tandem Whitaker, W.B., Sandoval, N.R., Bennet, R.K., Fast, A.G., photoanode in conjunction with a bipolar membrane and Papoutsakis, E.T. (2015). Synthetic methylotrophy: and a Pd/C cathode. ACS Energy Let. 1, 764–770. engineering the production of biofuels and chemicals htps://doi.org/10.1021/acsenergylet.6b00317

Curr. Issues Mol. Biol. (2019) Vol. 33 caister.com/cimb