Constructing Biocatalytic Cascades:In Vitro Andin Vivo Approaches to De Novo Multi-Enzyme
Constructing Biocatalytic Cascades:In Vitro andIn Vivo Approaches to De Novo Multi-Enzyme Pathways
Scott P. France,‡ Lorna J. Hepworth,‡ Nicholas J. Turner* and Sabine L. Flitsch*.
School of Chemistry, Manchester Institute of Biotechnology, The University of Manchester, 131 Princess Street, Manchester M1 7DN, United Kingdom.
‡ These authors contributed equally to the manuscript.
ABSTRACT
The combination of sequentialbiocatalyticreactions,via non-natural synthetic cascades,is a rapidly developing field andleads to the generation of complex valuable chemicals from simple precursors. As the toolbox of available biocatalysts continues to expand, so do the options for biocatalytic retrosynthesis of a target molecule, leading to alternative routes employing enzymatic transformations. The implementation of such cascade reactions requires careful consideration, particularly with respect to whether the pathway is constructed in vitro or in vivo. In this Perspective, we describe the relative merits of in vitro, in vivo or hybrid approaches to building biocatalytic cascades and showcase recent developments in the area. We also highlight the factors that influence the design and implementation of purely enzymatic or chemo-enzymatic, one-pot, multi-step pathways.
KEYWORDS: biocatalysis, enzymatic cascades, in vitrobiotransformations, in vivobiotransformations, biocatalytic retrosynthesis, whole-cell biocatalysis, enzymes, cofactors.
INTRODUCTION
The repertoire of organic reactions that can be mediated by biocatalysts is rapidly increasing, driven by improved methods for enzyme discovery and screening together with faster and cheaper gene synthesis.1–9The outcome is an ever-expanding biocatalytic toolbox consisting of natural enzymes, engineered or evolved enzymes10–16 and artificial enzymes17–19which are capable of catalyzing an increasing range of chemical transformations. The availability of a broader range of biocatalysts also means that it is increasingly possible to build entirely de novo enzymatic pathways, both in vitro and in vivo, conceived and designed through a biocatalytic retrosynthetic approach (Figure 1).20–22
Figure 1.Design-Implement-Analyze Cycle of Multi-Enzyme Cascade Development.
The term ‘enzymatic cascades’ has been used to describe (chemo)enzymatic processes consisting of two or more steps for the production of valuable chemical compounds.23–35These cascades are distinct from natural biosynthetic pathways (not discussed in this Perspective) that have evolved over time to ensure often exquisite flow of simple metabolites to complex natural products. However, an analysis of natural biosynthetic pathways reveals some of the hallmarks and important features of efficient enzymatic cascades:
(1) The overall thermodynamic parameters of the cascade are favorable (ΔGcascade < 0).
(2)Selectivity: Enzymes catalyze reactions with high reaction specificity and functional group orthogonality in order to avoid unwanted cross-reactivity between different substrates.
(3) The overall kinetic reaction parameters are controlled by enzyme activities to ensure reaction flux.
In this Perspective we review recently published examples of synthetically useful non-natural enzyme cascades to illustratethe design principles involved and also how these important features of cascade processes are addressed in practice. We discuss these cascades in terms of their operation as either cell-free systems (in vitro) or in whole-cells (in vivo), and focus on concurrent cascade reactions rather than sequential processes. The selection of examples is not intended to be comprehensive, but rather didactic in nature, and hence we direct the reader to several recent excellent in-depth reviews.36–43 Although not covered in this Perspective, considerable effort in recent years has been made regarding the engineering of native biosynthetic pathways, and a variety of valuable natural products, such as vanillin44 and artemisinin,45 have been successfully obtained in considerable yields through a synthetic biology approach. This area has been extensively reviewed elsewhere in many informative articles.46–52
IN VITRO VS IN VIVO
Enzyme cascades can be broadly classified as either (i) in vitro, (ii)in vivo or (iii)hybrid as shown in Figure 2. The selection amongst these three options for a particular application typically depends upon a number of factors, includingavailability of gene sequences and heterologous enzymes, cofactor requirements, substrate and product uptake and release, and the metabolic stability of substrates and product.
Figure 2.Comparison of (i)in vitro, (ii)in vivo and (iii) hybrid cascade reactions. Product D is generated from starting material A using highly selective biocatalysts cat i-iii.
IN VITROCASCADES
In vitro cascades based on enzymatic or chemo-enzymatic reactions have been developed for a number of years, with an early example reported byFessner and co-workers usingaldolases for the synthesis of branched-chain saccharides.53These cascade reactions use purified enzymes, cell lysates, cell-free extracts or freeze-dried whole-cells, and offer advantages such as the relative ease of fine-tuning the amounts of each (bio)catalyst present in the system to maximize the overall flux and yield of products.The application of purified enzymes avoids any complications that arise from the complex metabolic pathways operating in living whole-cells, although the use of crude enzyme preparations does not completely eliminate these potential side reactions. Additionally, enzyme purification is an expensive and time-consuming process, especially for scale-up and potential industrial application.Likewise, if biocatalysts are cofactor dependent, expensive exogenous cofactors need to be added in stoichiometric amounts or alternatively the cascadeneeds to be supplemented with a recycling system.
Scheme 1.In Vitro Hydrogen-Borrowing Cascade for the Production of Chiral Amines from Alcohols.
A solution to the issue of cofactor requirements was recently reported for the production of chiral amines from the corresponding alcohols via an intermediary ketone compound (Scheme 1).54In this example an alcohol dehydrogenase (ADH) was employed to oxidize the alcohol starting material 1to the ketone 2.The NADH generated during the first step was used in the subsequent step in which an amine dehydrogenase (AmDH) catalyzed a reductive amination of the ketone with ammonia. Overall, the cascade is redox neutral and the NAD(H) cofactor was recycled by a ‘hydrogen-borrowing’ process55,56 in which the hydrogen abstracted in the first step was reinstalled in the second. Ultimately, an in vitro approach using purified enzymes was required due to competing reactions catalyzed by endogenous proteins present when a crude cell preparation of the ADH was used. These side reactions sequestered and oxidized the NADH cofactor required for the AmDHstep and thus disrupted the hydrogen-borrowing nature of the cascade. High conversions were achieved (85%, >99% ee) andpreparative-scale reactions (100 - 126 mg of starting material1)were demonstrated to afford moderate to excellentisolated yields (30 to 91%) and high ee values (82 to >99%).
Scheme 2.In Vitro Cascade for the Production of para-Vinylphenols8 Starting from Phenols 5 and Pyruvate 4.
Kroutiland co-workers developed a highly selective biocatalytic route to perform the formal para-vinylation of phenols via a three enzyme in vitro cascade (Scheme 2).57 The important initial C-C bond forming step wascatalyzed by a tyrosine phenol lyase (TPL) that coupled the phenol substrate5 with pyruvate 4and ammonia to generate an L-tyrosine intermediate6. This compound was then deaminated by a tyrosine ammonia lyase (TAL) to afford a coumaric acid derivate7 that was finally decarboxylated by a ferulic acid decarboxylase (FAD). The ammonia by-product generated as part of the TAL transformation could be reused in the first TPL-catalyzed step and all reactions were run in a one-pot concurrent fashion. A wild-type or engineered TPL was used as a cell-free extract and the TAL and FAD were recombinantly expressed separately in E. coli and employed as freeze-dried cell preparations. A range of 2-, 3- and 2,3-substituted phenols could be successfully vinylated in high conversion (>99%) and the synthetic utility of the cascade was demonstrated by high yielding preparative-scale examples(65 to 83%, on approximately 50 mg of starting material 5).
Scheme 3.In Vitro Cascade for the Production of (1R,2R)-Norpseudoephedrine (1R,2R)-11 and (1R,2S)-Norephedrine (1R,2S)-11 from Benzaldehyde 9 and Pyruvate 4 with an Optional Recycling Mode for Pyruvate.
Rother et al. have devised a synthetic route to (1R,2R)-norpseudoephedrine(1R,2R)-11, or (1R,2S)-norephedrine(1R,2S)-11 by means of a two-step biocatalytic cascade (Scheme 3).58 A thiamine diphosphate (ThDP)-dependent acetohydroxyacid synthase I (AHAS-I) was employed in the first step to decarboxylate pyruvate4 and perform a ligation to benzaldehyde9, affording the intermediate (R)-phenylacetylcarbinol10 with high stereoselectivity (>98%). Subsequently an (R)- or (S)-selective transaminase converted10 into the final products (1R,2R)-11 and (1R,2S)-11, respectively, using alanine as a co-substrate. Interestingly, the cascade could be operated in a ‘recycling’ mode in which the pyruvate by-product generated by the transaminase step could then re-enter the cascade, either directly as a substrate in the first step or via the reversible formation of an acetolactate intermediate 12. One problem identified when operating the cascade as a concurrent process was the fact that benzaldehyde 9was also a substrate for the transaminase biocatalyst. Indeed, using an (S)-selective transaminase gave only 2% product (1R,2S)-11 with the major product being undesired benzylamine. However, a commercially available (R)-selective transaminase was identified that possessed initial rate activities for benzaldehyde9 that were approximately ten times lower than the AHAS-I step. This enabled greater conversion to 10 by the AHAS-I before 9could beintercepted by the transaminase, and allowed the cascade to be run in a concurrent fashion to afford product (1R,2R)-11 with 85% conversion.In comparison, when run in the concurrent ‘recycling’ mode, only 70% conversions could be achieved. To overcome the problems associated with the cross-reactivity between 9 and the transaminases, a sequential process was developed in which the transaminase was added after complete conversion in the first step. This enabled higher conversions to be achieved compared to the concurrent process: (1R,2S)-11 was obtained with 80% conversion and (1R,2R)-11 with >96% conversion with high de and ee (>98%). The temporal separation of the two biocatalytic cascade steps in the sequential process meant that the ‘recycling’ mode was not possible directly; however, more 9 could be added after the complete sequential cascade had been run to achieve further conversion.
Scheme 4.In Vitro ‘Triangular’ Cascade for the Formation of Benzylisoquinoline Alkaloids 15 from Phenylethylamines13.
Recent work by Hailes, Ward and co-workers demonstrated the construction of a ‘triangular’ cascade for the formation of benzylisoquinoline alkaloids (Scheme 4).59 The reaction sequence was initiated by the addition of amine 13, which generatedthe aldehyde 14 by means of a transaminase (ω-TA) step with pyruvate as the amineacceptor. Subsequently, a norcoclaurine synthase (NCS) catalyzed the asymmetric Pictet-Spengler reaction between 13 and 14 to give the benzylisoquinoline15. Control of the number of equivalents of pyruvate in the transaminase step enabled the correct stoichiometry of 13 and 14 in the system and problems often associated with transaminase equilibria were avoided by the subsequent removal of equal quantities of the amine and aldehyde by the NCS. It was important to optimize the reaction parameters to ensure the rates of the two biocatalytic steps were sufficiently matched to circumvent aldehyde 14accumulation that would lead to a competing, non-enzyme catalyzed, Pictet-Spengler reaction and formation of racemic product. The cascade was operated with a purified NCS and a transaminase employed as a lysate. Optimization of the reaction parameters resulted in a system capable of converting 20 mM starting material in 87% conversion and in 99% ee. Tetrahydroprotoberberine alkaloids could also be accessed by incorporating an additional step after the biocatalytic cascade in which formaldehyde was added after 3 h reaction time. This initiated a non-enzymatic Pictet-Spengler condensation to afford the alkaloid 16 as the major regioisomer. The cascade was also utilized for the preparative-scale synthesis (using 94.5 mg of 13) of 15 and 16 in 62% and 42% isolated yield, respectively, and both with an enantiomeric excess of >95%.
Scheme 5. Complementary In Vitro Cascades for the Production of EnantiopureD- or L-Arylalanines18 from Cinnamic Acids 17.
A cascade route to both enantiomers of phenylalanine derivatives has been developed by the Turner group (Scheme 5).60 Phenylalanine ammonia lyases (PALs) have previously been used for the amination of cinnamic acid derivatives 17using free ammonia to generate L-arylalaninesL-18, although not always with perfect enantioselectivity. To access the more challenging D-enantiomers, an amination/racemization cascade system was constructed, in which the first step was a PAL-mediated amination of the cinnamic acid employing an engineered variant to generate the arylalanine with imperfect enantioselectivity. The L-enantiomer was then oxidized by an L-amino acid deaminase (LAAD) to an imine 19which was then reduced by the non-selective chemical reducing agent ammonia borane. In this way, all of the L-enantiomer was converted to the D-enantiomer which was not a substrate for the LAAD, and therefore accumulated in the system. This approach enabled the synthesis of a range of D-arylalanines in good conversion and excellent ee(62 to 80% conv., 98 to >99% ee). Some α-ketoacid by-product 20was observed as a result of the interception and hydrolysis of the imine before it could be reduced by ammonia borane. Furthermore, the simple substitution of the engineered PAL variant with the wild-type PAL enzyme, and of the LAAD with a D-amino acid oxidase (DAAO), generated a new cascade to upgrade the ee of nitro- and cyano-L-arylalanines (>99% ee), known to be produced in lower enantiopurity by PAL alone (0 to >99% ee).
Scheme 6.In Vitro Cascade for the Production of Tetrasaccharide24.
Carbohydrates are complex targets for chemical synthesis and biocatalysis has become a very attractive strategy to assemble both natural and unnatural glycosides. A recent example is the biomimetic route to O-mannosylglycans which are important biomarkers on cell surfaces (Scheme 6).61 The synthetic strategy employed purified glycosyltransferases to sequentially attach sugar monomer units to the glycan chain. A mannosyl unit was first attached to a peptide chain of interest using solid phase peptide synthesis techniques to form glycopeptide21. Sequentially, N-acetyl-glucosamine (GlcNAc), galactose (Gal) and sialic acid moieties were then attached, catalyzed by human protein O-mannoseβ-1,2-N-acetyl-glucosaminyltransferase 1 (POMGnT1), bovine β-1,4-galactosyltransferase (β-1,4-GalT) and Trypanozomacruzi trans-sialidase (TcTS) in 85% (22), 87% (23) and 47% (24) yield after HPLC, respectively. The TcTS enzyme used the sialoprotein fetuin as the sugar donor for this step and required careful reaction monitoring to avoid hydrolysis of the newly created sialosidic linkage, making this step difficult to drive to completion. Due to the very high selectivity of all three biocatalysts in this synthetic route, a one-pot cascade was implemented which significantly shortened the amount of time needed to generate the desired tetrasaccharide, and removed all intermediate purification processes. Using this cascade approach, the tri- 23and tetrasaccharides24 were formed as a 1:1 mixture with only trace amounts of mono- 21and disaccharide22 observed.
Scheme 7.In Vitro Isomerization/Reduction Cascade for the Production of Chiral Saturated Alcohols 27 from Racemic Allylic Alcohols 25.
The simultaneous application of a chemocatalyst and a biocatalyst has been thoroughly studied in the context of dynamic kinetic resolutions. Pioneering work by Bäckvall employed a chemocatalyst to equilibrate two enantiomers, followed by a successive enzymatic reaction catalyzed by a lipase to achieve enantiopure products.62–64A recent development in the field ofchemoenzymatic reactions was a concurrent cascade reported by González-Sabín, García-Álvarezand co-workers for the formal asymmetric reduction of allylic alcohols (Scheme 7).65 Their approach was to combine a bis(allyl)-ruthenium (IV) complex to catalyze the redox isomerization of the allylic alcohol25 to the saturated ketone 26,followed by the biocatalytic reduction accomplished by a ketoreductase (KRED). A range of commercially available NADPH-dependent KREDs from Codexis were shown to afford high conversions and enantioselectivities on a panel of ketones with isopropanol used for cofactor recycling. First, isomerization was run to completion, before the appropriate KRED and NADPH were added to the reaction. This process was able to achieve very high conversions (92 to 97%), isolated yields (85 to 90%) and ee values (98 to >99%). Aconcurrent cascade was then developed with both chemo- and biocatalysts present from the start. To achieve this, a compromise in temperature was employed and a higher [Ru]catloading was required to ensure turnover of the first step. This approach led to good conversion (85 to 94%) and isolated yield of 27(60 to 86% yield, on 26 – 33 mg of starting material 25). The limiting factors under the concurrent mode were the rate of the ruthenium-catalyzed isomerization and the biocatalyst stability i.e. the enzyme lost activity before complete isomerization had occurred. However, this was overcome in a sequential process as the ketone intermediate produced in the first step could be converted at high rate during the most active period in the lifetime of the biocatalyst.
Scheme 8.In Vitro Cascades Involving Artificial Transfer Hydrogenases (ATHases) for the Production of Chiral Amines via(a) Deracemization, and (b) Imine Formation from L-30, Followed by Deracemization.
The co-location of a transition metal catalyst and a biocatalyst in the same pot often results in the mutual inactivation or attenuation of the activity of both and hence the implementation of such systems is not a trivial undertaking.66–70One approach to overcome inactivation is to tether the transitionmetal catalyst within a protein scaffold, creating an artificial metalloenzyme which can then be applied alongside other biocatalysts without inhibitory or deactivating effects.71–73 This concept was demonstrated by Ward, Turner, Hollmann and co-workers,by coupling an artificial transfer hydrogenase (ATHase) with various enzymes to create in vitro cascades that were not possible when employing the free transition metal catalyst.74 The ATHase was constructed by incorporating a biotinylated iridium d6-pianostool complex within a streptavidin isoform and could be regenerated in situ by sodium formate as a source of hydride. The artificial enzyme was then employed in one-pot with a purified monoamine oxidase variant (MAO-N-9) for the deracemization of cyclic amines (Scheme 8a). A variant streptavidin isoform (S112T) was identified that gave some (R)-selectivity for the reduction of cyclic imine 28. The variant was then used in the cascade in which (S)-29was selectively oxidized by the MAO-N-9, leaving (R)-29, and the imine 28generated was then reduced by the ATHase to (R)-29. Catalase (CAT) was added to eliminate the build-up of hydrogen peroxide that was speculated to deactivate the iridium catalytic center. A different cascade was also built for the synthesis of L-pipecolic acidL-32 based on the same deracemization principle, however the cyclic imine 31 was generated in an initial step from L-lysineL-30, catalyzed by an L-amino acid oxidase (LAAO) (Scheme 8b). In these cascades, the artificial metalloenzyme approach demonstrates how the encapsulating streptavidin protein effectively retains and shields the active iridium core from deactivating effects and provides a local chiral environment for the chemical transformations that occur there.