Subscriber access provided by UNIV OF CALIFORNIA SAN DIEGO LIBRARIES Article Enzymatic Menthol Production: One-pot Approach Using Engineered Escherichia coli Helen Toogood, Aisling Ni Cheallaigh, Shirley Tait, David J Mansell, Adrian Jervis, Antonios Lygidakis, Luke D Humphreys, Eriko Takano, John M Gardiner, and Nigel S. Scrutton ACS Synth. Biol., Just Accepted Manuscript • DOI: 10.1021/acssynbio.5b00092 • Publication Date (Web): 27 May 2015 Downloaded from http://pubs.acs.org on June 2, 2015

Just Accepted

“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

ACS Synthetic Biology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties. Page 1 of 30 ACS Synthetic Biology

1 2 Enzymatic Menthol Production: One-pot Approach Using 3 4 5 Engineered Escherichia coli 6 7 8

9 a b a b a 10 Helen S. Toogood , Aisling Ní Cheallaigh , Shirley Tait , David J Mansell , Adrian Jervis , Antonios 11 12 Lygidakis a, Luke Humphreys c, Eriko Takano a, John M. Gardiner b, Nigel S. Scrutton a* 13 14 SYNBIOCHEM, Manchester Institute of Biotechnology, aFaculty of Life Sciences and bSchool of 15 16 Chemistry, University of Manchester, Manchester M1 7DN, UK. 17 18 cGlaxoSmithKline, Medicines Research Centre, Gunnel's Wood Road, Stevenage, Herts SG1 2NY, UK. 19 20 21 22 Corresponding authors: Nigel S. Scrutton and Helen S. Toogood: Manchester Institute of 23 24 Biotechnology, Faculty of Life Sciences, University of Manchester, Manchester M1 7DN, UK. 25 26 Phone: +44 161 3065152; Fax: +44 161 3068918; E-mail addresses: [email protected] 27 28 29 and [email protected] 30 31 Present address: David J. Mansell: Glythera Ltd. Herschel Annex, King’s Road, Newcastle-upon- 32 33 Tyne, NE1 7RU 34 35 36 E-mail addresses: [email protected]; [email protected]; 37 38 [email protected]; [email protected]; 39 40 [email protected]; [email protected]; 41 42 43 [email protected]; [email protected] 44 45 46 47 Abbreviations: 48 49 50 NtDBR: double bond reductase from Nicotiana tabacum ; MMR = (-)-menthone:(-)menthol 51 reductase from Mentha piperita ; MNMR = (-)-menthone:(-)neomenthol reductase from M. 52 piperita ; GDH = glucose dehydrogenase; DTT = dithiothreitol; SD = Shine-Dalgarno sequence; IPTG 53 = isopropyl-β-D-1-thiogalactopyranoside; TB = Terrific broth; TBAIM = Terrific broth auto induction 54 medium. Expression gene clusters: DM = NtDBR and MMR; DN = NtDBR and MNMR; DMN = 55 56 NtDBR, MMR and MNMR. 57 58 59 60 1 ACS Paragon Plus Environment ACS Synthetic Biology Page 2 of 30

1 2 ABSTRACT 3 4 5 Menthol isomers are high-value monoterpenoid commodity chemicals, produced naturally 6 7 by mint plants, Mentha spp. Alternative clean biosynthetic routes to these compounds are 8 9 10 commercially attractive. Optimization strategies for biocatalytic terpenoid production are mainly 11 12 focused on metabolic engineering of the biosynthesis pathway within an expression host. We 13 14 circumvent this bottleneck by combining pathway assembly techniques with classical biocatalysis 15 16 17 methods to engineer and optimize cell-free one-pot biotransformation systems and apply this 18 19 strategy to the mint biosynthesis pathway. Our approach allows optimization of each pathway 20 21 enzyme and avoidance of monoterpenoid toxicity issues to the host cell. We have developed a 22 23 24 one-pot (bio)synthesis of (1 R,2 S,5R)-(-)-menthol and (1 S,2 S,5 R)-(+)-neomenthol from pulegone, 25 26 using recombinant Escherichia coli extracts containing the biosynthetic genes for an ‘ene’- 27 28 29 reductase (NtDBR from Nicotiana tabacum ) and two menthone dehydrogenases (MMR and 30 31 MNMR from M. piperita ). Our modular engineering strategy allowed each step to be optimized to 32 33 improve the final production level. Moderate to highly pure menthol (79.1 %) and neomenthol 34 35 36 (89.9 %) were obtained when E. coli strains co-expressed NtDBR with only MMR or MNMR, 37 38 respectively. This one-pot biocatalytic method allows easier optimization of each enzymatic step 39 40 and easier modular combination of reactions to ultimately generate libraries of pure compounds 41 42 43 for use in high throughput screening. It will, therefore, be a valuable addition to the arsenal of 44 45 biocatalysis strategies, especially when applied for (semi)-toxic chemical compounds. 46 47 48 49 50 Keywords: 51 52 53 Menthol production; one-pot biosynthesis; recombinant biosynthetic pathways; Escherichia coli . 54 55 56 57 58 59 60 2 ACS Paragon Plus Environment Page 3 of 30 ACS Synthetic Biology

1 2 INTRODUCTION 3 4 5 Limonene biosynthesis MEP 6 OPP glyceraldehyde-3-phosphate Dimethylallyl diphosphate pathway + 7 OPP pyruvate GPPS 8 IDI 9 Geranyl OH diphosphate Mevalonate 'Mint' biosynthesis acetyl-CoA 10 OPP Monoterpenoid pathway (+)-Neomenthol Isopentyl diphosphate MNMR 11 Precursor Eukaryotes only

12 LimS MMR PP Peppermint pathway 13 (-)-Menthone O OH

NADP+ 14 For L3H/CPR IPDH IPR (-)-Menthol IPGI PGR both 15 NADPH NADPH NADP+ OH NAD+ NADH O NADPH NADP+ O O NADPH NADP+ + H O MNMR 16 + O2 2 LIMONENE (-)- trans-Isopiperitenol (-)-Isopiperitenone (+)- cis-Isopulegone (+)-Pulegone O OH 17 MMR NADPH + O2 NADPH + O2 (+)-Isomenthone (+)-Isomenthol 18 MFS- + NADP + H2O NAD + Spearmint pathway CPR 19 L6H/CPR P + H2O 20 HO CDH O Methanofuran OH

AD+ ADH biosynthesis 21 N N O (+)-Neoisomenthol 22 Menthol (-)- trans-Carveol (-)-Carvone (+)-Menthofuran Isomers 23

24 1 25 Scheme 1. Monoterpenoid biosynthesis pathways in different Mentha species. IDI = Isopentenyl-diphosphate Delta- 26 isomerase; GPPS = geranyl diphosphate synthase; LimS = (-)-limonene synthase; L3H = (-)-limonene-3-hydroxylase; 27 CPR = cytochrome P450 reductase; IPDH = (-)-trans -isopiperitenol dehydrogenase; IPR = (-)-isopiperitenone reductase; 28 IPGI = (+)-cis -isopulegone isomerase; PGR = (+)-pulegone reductase; MMR = (-)-menthone:(-)menthol reductase; 29 MNMR = menthone:(+)-neomenthol reductase; MFS = (+)-menthofuran synthase; L6H = (-)-limonene-6-hydroxylase; 30 CDH = (-)-trans -carveol dehydrogenase. 31 32 33 34 Limonene enantiomers are the most abundant monocyclic monoterpenoids in nature; (-)- 35 36 limonene is found in herbs such as Mentha spp. (e.g., peppermint and spearmint), while the (+)- 37 38 enantiomer is the major oil constituent of orange and lemon peel.2 Natural limonene derivatives 39 40 41 are known to be important precursors in the production of several pharmaceutical and commodity 42 43 chemicals, such as fragrances, perfumes and flavors. For example, essential oils of mint contain a 44 45 variety of limonene derivatives (Scheme 1), such as menthol isomers, pulegone and methanofuran 46 47 48 in peppermint ( Mentha piperita ) and carveol/carvone in spearmint ( Mentha spicata ). Menthol 49 50 isomers, (1 R,2 S,5 R)-(-)-menthol, (1 R,2 S,5 S)-(+)-isomenthol, (1 S,2 S,5 R)-(+)-neomenthol and 51 52 53 (1 R,2 R,5 R)-(+)-neoisomenthol, and carvone are used as additives in oral hygiene products, and 54 55 flavors in food and beverages. Carveol is found in cosmetics, while pulegone is commonly found as 56 57 a flavor in perfumery and aromatherapy products. Carvone and carveol are also known to have 58 59 60 3 ACS Paragon Plus Environment ACS Synthetic Biology Page 4 of 30

1 2 anticancer properties, while menthol has antibacterial activity against Staphylococcus aureus and 3 4 Escherichia coli .3 5 6 7 There is a high demand for limonene and derivatives (e.g., menthol oil ca 3,000 t/$373–401 8 9 US million pa), which are traditionally obtained from natural sources (Mentha spp.) due to the 10 11 flavor/fragrance industries demanding so-called ‘natural’ sources that are compatible with use in 12 13 14 food products. However, the production of natural menthol relies heavily on ca 0.29 million 15 16 hectares of arable land, and requires expensive steam distillation and filtration processes.4 Given 17 18 the volatility of mint crop prices due to unpredictable harvest yields and the environmental 19 20 21 footprint of intensive mint cultivation, alternative clean (bio)synthetic routes to these compounds 22 23 are commercially attractive.5 24 25 An alternative ‘natural’ route to highly pure complex organic compounds utilizes 26 27 28 microorganisms as biological factories. They are built from existing or de novo biosynthetic 29 30 pathways, incorporated into rapidly growing, cost effective and even food compatible 31 32 microorganisms grown on non-petroleum based renewable feedstock. 6 For example, a precursor 33 34 35 of Taxol (paclitaxel), a potent anticancer drug found naturally in Taxus brevifolia (Pacific yew tree), 36 37 has been successfully produced in E. coli .6 Several reports describe the production of limonene 38 39 40 and other terpenoids in E. coli , Saccharomyces cerevisiae and cyanobacteria by incorporating 41 42 genes encoding plant terpene synthases and ones that up-regulate isoprene precursor 43 44 production. 5a, 7 The semisynthetic industrial-scale production (ca 35 tonnes per annum) of 45 46 47 artemisinin (major active ingredient in modern malarial treatments) by Sanofi has shown 48 49 commercial success, in comparison to traditional extractions/purifications from natural sources 50 51 (sweet wormwood). 8 In all of these examples, the optimisation of terpene yields was conducted in 52 53 54 vivo , i.e. in the production cell. One-pot biosynthesis of complex chemical compounds are not 55 56 common, and may be difficult to achieve due to technical challenges in successfully completing a 57 58 production process.9 59 60 4 ACS Paragon Plus Environment Page 5 of 30 ACS Synthetic Biology

1 2 We have demonstrated, as ‘proof of principle’, the successful one-pot biosynthesis of 3 4 menthol isomers from pulegone 10 by incorporating part of the peppermint biosynthetic pathway 5 6 7 into E. coli . This hybrid approach integrates both pathway engineering and classical biocatalysis 8 9 strategies, where the best performing enzymes (according to factors such as catalytic rate, 10 11 enantiospecificity, soluble expression) were identified and assembled into a functional biocatalytic 12 13 14 cascade. Further optimization for production was conducted by host strain selection, varying the 15 16 pathway enzyme expression strength and increasing the coenzyme (NADPH) availability to 17 18 increase in vitro production. Our approach combines enzyme selection and validation, followed by 19 20 21 seamless PCR-based operon assembly and one-pot biotransformations using E. coli extracts. We 22 23 demonstrate the effectiveness of this approach in generating moderately to highly pure menthol 24 25 isomers from recombinant E. coli cell extracts. This alternative one-pot biocatalytic method, which 26 27 28 avoids the bottlenecks of host optimization and in vivo metabolic engineering, has the potential to 29 30 be applied as a general strategy for optimizing (semi-)toxic chemical compounds production, 31 32 based on its ease of optimization of each enzymatic step and the possibility of using the resulting 33 34 35 engineered in vitro pathways in high-throughput screening. 36 37 38 39 40 RESULTS AND DISCUSSION 41 42 43 Product synthesis. A synthetic route was designed to give access to all menthone and 44 45 menthol isomers required for biotransformations, starting from menthone (Scheme 2). This was 46 47 motivated by the fact that isomenthone and neoisomenthol are not commercially available, while 48 49 50 the remaining menthols can be derived from the purification of essential oil mixtures. Further 51 52 details of the synthesis, purification and identification of the isomers is described in the supporting 53 54 information. Our chemical synthetic approach has provided routes to obtain diastereomerically 55 56 57 pure isomers of menthone and menthol for the first time, overcoming existing limitations in the 58 59 commercial supply of some these compounds. 60 5 ACS Paragon Plus Environment ACS Synthetic Biology Page 6 of 30

1 2 3 4 5 NaBH 4, THF, MeOH 6 OH O OH 7

8 menthone menthol neomenthol 9 10% NaOH 10 in MeOH 11 1. DBAD, p- nitrobenoic acid, 12 NaBH 4, THF, PPh 3, THF 13 MeOH 2. LiOH, H2O, THF 14 O OH OH

15 isomethone neoisomenthol isomenthol 16 17 DBAD: Di- tert -butyl azodicarboxylate 18 19 Scheme 2 . Chemical synthesis of isomenthone and all four menthols from menthone 20 21 22 23 Enzyme production and validation. Our cloning strategies enabled the rapid generation and 24 25 manipulation of multi-gene expression constructs from individual ‘components’, including the 26 27 28 presence of repeated sequences of relatively high GC content. However, to design the best 29 30 production pathway, it is necessary to characterize the individual enzyme activity, and to use the 31 32 most active enzymes for the desired reaction. Initially, the three recombinant enzymes (NtDBR, 33 34 35 MMR and MNMR) were purified, and preliminary biocatalytic ability of each was assessed to test 36 37 their functionality when produced in a bacterial system. The three genes encoding these plant 38 39 40 enzymes were first codon-optimized for E. coli and separately cloned into pET21b to generate C- 41 42 His 6-tagged recombinant enzymes. After production and purification, each enzyme was tested for 43 44 activity under steady state conditions, via indirect NADPH oxidation monitoring. The ene- 45 46 -1 47 reductase activity of NtDBR-His 6 with (5 R)-pulegone was relatively low (<0.1 s ), however this is 48 11 49 known to be a poor kinetic substrate for this enzyme. Ketoreductase activity of MMR-His 6 (1.63 50 51 -1 -1 -1 -1 s and 2.00 s ) and His 6-MNMR-His 6 (0.96 s and 0.20 s ) were determined with both (2 S,5 R)- 52 53 54 menthone and (2 R,5 R)-isomenthone, respectively. This shows that MMR has a similar rate with 55 56 both isomers, while MNMR has a preference for menthone over isomenthone. Further 57 58 characterization studies are currently being performed, and will be the focus of a later report. 59 60 6 ACS Paragon Plus Environment Page 7 of 30 ACS Synthetic Biology

1 2 Initial in vitro biotransformation reactions were performed with each enzyme, and an 3 4 equimolar combination of all three (2 M), in the presence of an externally added cofactor 5 6 + 7 recycling system (GDH/glucose/NADP ), to identify the yields and ratio of each monoterpenoid 8 9 product formed (SI Table 2). For NtDBR, the reduction of pulegone yielded near equivalent 10 11 12 amounts of menthone and isomenthone, similar to the non His 6-tagged recombinant enzyme. 12 13 11 14 This differs from previous studies with the His 6-tagged enzyme where the ratio was nearly 1:2. 15 16 Prior studies suggested that variations can be obtained under different reaction conditions. 12 17 18 Interestingly, under biotransformation conditions (24 h; cofactor recycling system) MMR appears 19 20 21 to generate more products from menthone than isomenthone. This differs from the kinetic 22 23 studies, where the rates slightly favored the reaction with isomenthone. Product ratios agree with 24 25 prior studies 10 where MMR has a preference for menthol over neomenthol from menthone, and 26 27 28 neoisomenthol over isomenthol with isomenthone whilst the opposite is true for MNMR. This 29

30 shows these His 6-tagged-enzymes have comparable activity to the native biocatalysts in M. 31 32 33 piperita . 34 35 Small amounts of the non-standard products (e.g., neomenthol from ‘isomenthone’) with 36 37 both enzymes were obtained due to the presence of approximately 5% of the opposite substrate 38 39 40 enantiomer contaminating the reactions. Reactions with a mixture of equivalent amounts of each 41 42 enzyme show all the menthone generated by NtDBR has been consumed to form menthol and 43 44 neomenthol, while activity utilizing isomenthone was poor. This shows the recombinant enzymes 45 46 47 were all active in vitro and display previously determined activities. In a stoichiometric mixture of 48 49 all three enzymes, MMR activity is biased over that of MNMR as expected from the steady-state 50 51 kinetics of each enzyme. 52 53 54 Operon construction. As each recombinant enzyme shows the required activities, a synthetic 55 56 operon was designed (Figure 1) to enable the co-expression of each gene within E. coli under the 57 58 control of one promoter (T7 lac ). The initial construct encoding NtDBR, MMR and MNMR (pDMN) 59 60 7 ACS Paragon Plus Environment ACS Synthetic Biology Page 8 of 30

1 2 was designed to include a Shine-Dalgarno (SD) sequence between successive genes, and maintain 3 4 the His 6-tags (N-His 6 for MNMR). Our initial attempts to construct this vector using standard PCR 5 6 7 and cloning methods were complicated due to the presence of a His 6-tag next to the SD sequence 8 9 with a repeating unit of 43 bp and a high G/C content (T m 73 °C). To overcome this issue, we 10 11 designed a cloning protocol (Figure 1) based on the In-Fusion recombination system, avoiding the 12 13 14 use of PCR primers annealing to the His 6-tag when the repeating unit was present. This sequential 15 16 approach was based on the following rules: i) the vector containing the first gene is opened by 17 18 reverse PCR, maintaining the C-His 6-tag and incorporating the SD sequence at the 3’ end; ii ) each 19 20 21 inserted gene contains a 5’ SD sequence, and all except the last gene are amplified without their 22 23 His 6-tag at the 3’ end; iii ) the third and successive gene(s) contain the missing His 6-tag of the 24 25 previously inserted gene at the 5’ end before the SD sequence; iv ) the 3’ overlaps of the inserted 26 27 28 genes anneal to the terminator region of the vector and v) each PCR amplified product contains a 29 30 15 bp overlap annealing to the PCR product it will be recombined with. 31 32 This general protocol can be used for assembly of pathways to enable multiple genes (up 33 34 35 to 5) to be added in one reaction in the correct order, providing the 15 bp overlaps are specific for 36

37 the next gene, and not the His 6-tag/SD region (5’ region of all except the first gene insert contains 38 39 40 the His 6-tag, SD region and 15 bp overlap with the previous gene). Only the last gene to be 41 42 inserted contains a 3’ overlap annealing to the terminator region. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 8 ACS Paragon Plus Environment Page 9 of 30 ACS Synthetic Biology

1 2 Linearised NtDBR-pET21b (1) T pET21b DBR His GGAGGACAGCTAAAT 3 6 4 (2) + (3) + GGAGGACAGCTAAAT MMR His GATCCGGCTGCTAAC GGAGGACAGCTAAAT His MNMR GATCCGGCTGCTAAC 5 6 6 6 In-Fusion cloning In-Fusion cloning 7 DM DN 8 DBR His6 SD MMR His6 T DBR His6 SD His6 MNMR T 9 + 10 (4) GGAGGACAGCTAAAT MMR GATCCGGCTGCTAAC 11 50 50 DBR MNMR In-Fusion cloning DM DN DMMn 12 13 37 37 pET21b DBR His6 SD MMR T 14 MMR All PCR linearisa on 15 25 Inset A 25 Inset B

16 (5) T pET21b DBR His6 SD MMR 17 +

18 (6) TGTCTGTATCTCGAG His6 GGAGGACAGCTAAAT His6 MNMR GATCCGGCTGCTAAC 19 20 In-Fusion cloning 21 DMN 22 DBR His SD MMR His SD His MNMR T 23 6 6 6 24 25 26 Figure 1. Method of generating multi-gene expression constructs with a single promoter using In-Fusion cloning. 27 Sequences are colour coded to show the overlap with the next gene. Large overlaps (e.g. MNMR 5’ region) were 28 generated using 2-3 overlapping forward PCR primers within one PCR reaction. Red = Shine-Dalgarno sequence (SD); 29 Orange = T7 terminator region of pET21b (T). The numbers in parentheses refer to the PCR steps, and correlate with 30 31 the oligos in SI Table 5. Inset A shows the SDS-PAGE analysis of the three purified enzymes (20-30 pmol each). Inset B 32 is a Western blot (anti-His 6) of soluble protein extracts from whole cells expressing the three multi-gene constructs. 33 34 The new constructed pathway (DMN; Figure 1) was expressed in E. coli (strain 1; 35 36 BL21(DE3)pLysS; Figure 1 inset A), and expression of the tagged proteins was analyzed by Western 37 38 39 blot (Figure 1 Inset B). Unfortunately, the three enzymes are similar in size (Figure 1 inset A), so in 40 41 most cases only two of the three proteins could be identified. Whole cell soluble protein extracts 42 43 44 of E. coli strain 1 containing the DMN construct underwent biotransformations with the substrates 45 46 pulegone, menthone and isomenthone to determine the activity of each enzyme. The results (SI 47 48 Table 3; strain 1) showed the DMN extract had activity of all three enzymes, and no product 49 50 51 formation was observed in the control reactions (cell extracts with empty pET21b). In the 52 53 presence of pulegone, both NtDBR (78 M menthone and 3.5 M isomenthone) and MNMR (22 54 55 M neomenthol) activities were detected but no MMR activity. Reactions with menthone showed 56 57 58 59 60 9 ACS Paragon Plus Environment ACS Synthetic Biology Page 10 of 30

1 2 primarily MNMR activity (212 M neomenthol) but also that of MMR (21 M menthol). However, 3 4 product yields were poor (20-220 M) and further optimization was clearly required. 5 6 7 A) B) 8 100" 100"

9 O 10 80" 80" 11 O 12 60" 60"

13 %"Yield" 40" %"Yield" 40" 14

15 20" 20" 16 17 0" 0" 18 4" 6" 7" 8" 9" 10" 4" 6" 7" 8" 9" 10" Strain" 19 Strain" 20 C) D) 21 22 80" 100" 23 O 80" kDa 60" O 50 24 37 25 25 60" 26 40" 50

%"Yield" 37 27 %"Yield" 40" 25 28 29 20" 20" 30 0" 0" 31 0" 10" 50" 100" 500" 1000"TBAIM" 32 4" 6" 7" 8" 9" 10" [IPTG]"(M)" 33 Strain" 34 35 Figure 2. Ratio of products formed during biotransformations of DMN in 6 strains with substrates A) pulegone, B) 36 menthone and C) isomenthone. D) Products formed during biotransformations of DMN cell extracts in strain 4 at 37 different IPTG concentrations and TBAIM media. Control reactions (strain 1 with empty pET21b) yielded no products. 38 Inset: SDS-PAGE and Western blot analysis of the cell extracts from the biotransformations. Reactions (2 mL) were 39 performed in buffer (50 mM Tris pH 7.0) containing monoterpenoid (1 mM), cell extracts (0.5 mL), NADP + (10 μM), 40 glucose (15 mM) and GDH (10 U). The reactions were agitated at 30 °C for 24 h at 130 rpm. Product yields were 41 determined by GC analysis using a DB-WAX column. Data point colors: menthone = grey; isomenthone = red; menthol 42 = green; neoisomenthol = purple; neomenthol = blue; isomenthol = orange. Strain identification is found in the 43 Methods section. 44 45 46 47 Comparative cell extract biotransformations. The DMN construct (pDMN) was 48 49 transformed into a further eleven E. coli strains to screen for improved expression of each gene. 50 51 Reactions with pulegone, menthone and isomenthone were performed with each extract as 52 53 54 before, but in the presence of an externally added cofactor recycling system for the production of 55 56 NADPH required by each enzyme. To our surprise, the results showed a wide variation in the 57 58 product yields and ratios (Figure 2 and SI Table 3), suggesting that the E. coli strain strongly 59 60 10 ACS Paragon Plus Environment Page 11 of 30 ACS Synthetic Biology

1 2 impacts on the expression levels of the individual genes independently. For example, strains 3 4 NiCO 2(DE3) (4) and Tuner(DE3) (7) showed high activity of each enzyme with pulegone, and 5 6 7 generated at least 75% yield of neomenthol with menthone (Figure 2a-b). In contrast, many other 8 9 extracts produced very little product, especially with isomenthone as the substrate (Figure 2c and 10 11 SI Table 3). Interestingly, strains 8-10 generated proportionally more menthol over neomenthol, 12 13 14 suggesting higher MMR activity over MNMR. These proportions of menthol isomers differ from 15 16 essential oils of M. piperita ; menthol yields are typically around 50 %, while neomenthol is a 17 18 relatively minor component (< 3 %). 10 The low yields of isomenthol from isomenthone is not 19 20 21 surprising, as MMR is known to have a 10-fold higher specificity for menthone over 22 23 isomenthone. 10 No activity was seen with the control reactions, although some minor unidentified 24 25 by-products were observed (results not shown). 26 27 28 Given that NtDBR generates nearly equal amounts of menthone and isomenthone (54:46), it 29 30 was surprising that reactions of strain 4 with pulegone gave a total yield of menthone and 31 32 subsequent reduction products of 84 % (840 M). Additionally, reactions with isomenthone of this 33 34 35 strain showed at least 20 % yield of products from menthone reduction (menthol and 36 37 neomenthol), in spite of there being a maximum of 5 % contaminating menthone in the reaction. 38 39 40 These findings are not observed in reactions with the purified enzymes, suggesting the E. coli 41 42 extracts contain epimerase activity, where isomenthone is isomerized to menthol. This was 43 44 confirmed by control reactions ( E. coli extracts without recombinant enzymes) in the presence of 45 46 47 menthone or isomenthone, showing a change from the 95:5 ratio to 60:40 (results not shown). 48 49 Given that reactions of many strains with menthone generated little or no isomenthol or 50 51 neoisomenthol, this suggests that the equilibrium strongly favors the direction of menthone 52 53 54 formation. Work is currently underway to identify and characterize the E. coli epimerase(s). 55 56 To further optimize the production level of menthol, the best three strains were identified (4, 57 58 6-7) based on product yields, and the balance of all three activities (Figure 2d and SI Figure 2). 59 60 11 ACS Paragon Plus Environment ACS Synthetic Biology Page 12 of 30

1 2 These strains underwent studies to determine the optimal IPTG concentration for protein 3 4 expression and thereby high activity. Strain 4 showed high NtDBR activity (~75 % yield) in 5 6 7 uninduced cells, suggesting high levels of leaky expression of this gene. In contrast, strain 6 8 9 (Rosetta2(DE3)pLysS) gave almost no products in the absence of IPTG, consistent with its pLysS 10 11 phenotype (tight regulation of the T7 lac promoter). Western blots of each extract (Figure 2d inset) 12 13 14 showed a correlation between expression levels and product yields, with the optimal IPTG 15 16 concentration as low as 50 M. 17 18 Construct optimization. The product ratios from DMN biotransformations suggested that 19 20 21 modifications of the multi-gene construct combined with further expression optimization and 22 23 native E. coli epimerization activity may lead to near single product formation. Therefore, two new 24 25 operons were designed and engineered, namely NtDBR-MMR (DM) and NtDBR-MNMR (DN; Figure 26 27 28 1). These were designed to enrich the products with menthol and neomenthol, respectively. 29 30 Comparative biotransformations were performed from the three multi-gene constructs in strain 4, 31 32 33 using pulegone as the substrate (Table 1). The results clearly showed an enrichment of the desired 34 35 product (menthol or neomenthol) from DM and DN, respectively, compared to DMN. However, 36 37 there were still significant levels of menthone/isomenthone and the original substrate at the end 38 39 40 of the reaction, suggesting further optimization is needed to increase MMR/MNMR activity. 41 42 Further studies were performed to see if the production of additional NADPH increases 43 44 enzyme activity. This was tested by the addition of either glucose, to supply the native E. coli 45 46 47 cofactor recycling systems, or a full externally provided cofactor-recycling system 48 49 (glucose/NADP +/GDH). Reactions in the presence of an externally added cofactor recycling system 50 51 showed higher product yields in all cases (0.5-3-fold; Table 1). This suggests the incorporation of 52 53 54 cofactor recycling genes to the multi-gene constructs may increase strain productivity. This was 55 56 previously reported in an E. coli strain expressing a glucose dehydrogenase gene from Bacillus 57 58 megaterium , which showed an increase in chiral alcohol production. 13 59 60 12 ACS Paragon Plus Environment Page 13 of 30 ACS Synthetic Biology

1 2 Table 1. Comparative yields of each of the six products from pulegone biotransformations catalyzed by cell extracts of 3 DM, DN and DMN in E. coli strain NiCO 2(DE3). 4 Product Yield ( M) 5 DM DN DMN a b a b a b 6 Product Glucose Cofact. Glucose Cofact. Glucose Cofact. 7 Menthone 121 + 2 144 + 3 107 + 2 213 + 5 108 + 1 192 + 3 8 Isomenthone 82 + 1 ND 73 + 2 143 + 3 84 + 1 144 + 2 c 9 Menthol 109 + 1 339 + 2 ND ND 37 + 2 132 + 2 c 10 Neoisomenthol ND ND ND ND ND ND Neomenthol d ND ND 132 + 4 247 + 1 103 + 1 149 + 4 11 d 12 Isomenthol ND ND ND 28 + 1 ND ND 13 Total % yield 31 2 (352) 48 3 (820) 31 2 (543) 63 2 (766) 33 2 (404) 61 8 (705) 14 Reactions (2 mL) were performed in buffer (50 mM Tris pH 7.0) containing pulegone (1 mM), cell extracts (0.5 mL), glucose (15 mM) +/- cofactor recycling system (10 μM NADP + and 10 U GDH). The reactions were agitated at 30 °C for 15 a 16 24 h at 130 rpm. Product yields were determined by GC analysis using a DB-WAX column. Reactions containing glucose and NADP+, but relying on native E. coli cofactor recycling enzymes; bReactions containing a complete 17 c d 18 externally added cofactor recycling system; Product from a reaction with menthone; Product from a reaction with 19 isomenthone. Data in parentheses are % conversion data; ND = none detected. 20 21 22 Biotransformation optimization. To further maximize product yields, biotransformations 23 24 with constructs DM and DN were performed with different levels of cell extracts. Control reactions 25 26 27 were performed where the DN/DM extracts and cofactor-recycling system was incubated in the 28 29 absence of monoterpenoids. There was only a moderate improvement in product yields, with 30 31 32 most effect seen in the increasing NtDBR activity in DN (Table 2A). In some cases, high 33 34 concentrations of cell extract in biotransformations inhibited product production (DN reactions), 35 36 and side product formation was significant. The most abundant side product (up to 470 M) 37 38 39 detected during reaction optimization studies (Table 2A) was identified as the non-terpenoid ester 40 41 ethyl propanoate. Therefore, this product has likely resulted from metabolic processes 42 43 independent of the introduced pathways. Increasing the cell extract quantity generated problems 44 45 46 with product extractions and clean ups due to viscosity, so further enzyme loading optimization 47 48 studies will concentrate on increasing protein expression levels. 49 50 51 Reactions also became significantly cloudy after a few hours of incubation, suggesting 52 53 protein precipitation and/or extract degradation. Therefore, parallel reactions were performed 54 55 each construct using pulegone as the substrate, and samples were analyzed at time points 1, 2, 6 56 57 58 and 24 h to determine the optimal reaction time (Table 2B). The results clearly show that long 59 60 13 ACS Paragon Plus Environment ACS Synthetic Biology Page 14 of 30

1 2 incubations with cell extracts lead to product loss in most cases (10-20 %). The exception was 3 4 menthone formation, where yields increased with time. Menthone gain and isomenthone loss 5 6 7 over time may simply be due to epimerization activity. 8 9 Table 2. Optimization of biotransformations from cell extracts of DM and DN in E. coli strain NiCO 2(DE3). 10 Product Yield (M) a b a b 11 Construct Menthone Isomenthone Menthol Neoisomenthol Neomenthol Isomenthol 12 A) Effect of the enzyme loading on product yields over 24 hours 13 DM extract volume (mL) 14 0.6 160 + 1 106 + 1 307 + 1 ND ND ND 15 0.8 165 + 4 106 + 2 359 + 10 ND ND ND 16 1.0 235 + 17 149 + 9 489 + 112 trace trace ND 17 DN extract volume (mL) 18 0.6 207 + 8 131 + 4 ND ND 265 + 57 ND 19 0.8 171 + 12 107 + 7 ND ND 317 + 7 ND 20 1.0 160 + 6 99 + 3 ND ND 371 + 12 ND 21 22 B) Effect of reaction time on product yields 23 DM reaction time (h) 24 1 37 + 2 229 + 7 359 + 10 ND ND ND 25 2 63 + 2 203 + 6 354 + 9 ND ND ND 26 6 114 + 3 153 + 4 351 + 6 ND ND ND 27 24 117 + 10 90 + 6 270 + 16 ND ND ND 28 DN reaction time (h) 29 1 25 + 1 205 + 2 ND ND 370 + 30 ND 30 2 125 + 4 163 + 7 ND ND 254 + 16 25 + 2 6 170 + 25 134 + 17 trace ND 189 + 27 trace 31 24 167 + 26 118 + 18 ND ND 180 + 28 trace 32 Reactions (2 mL) were performed in buffer (50 mM Tris pH 7.0) containing pulegone (1 mM), cell extracts (0.5 mL), 33 NADP + (10 μM), glucose (15 mM) and GDH (10 U). The reactions were agitated at 30 °C for 1-24 h at 130 rpm. Product 34 yields were determined by GC analysis using a DB-WAX column. aProduct from a reaction with menthone; bProduct 35 from a reaction with isomenthone. Total % yields were within 1-18% of the % conversion in each case. ND = none 36 detected; trace = <20 M product detected in some cultures. 37 38 39 40 The loss of menthol isomers over time suggests either product breakdown or utilization via 41 42 other E. coli pathways (side products of 4-43% yield). Interestingly, the reaction between DN and 43 44 45 pulegone yielded a small quantity (10 M) of menthol from menthone (MMR activity). However 46 47 studies have shown that MMR and MNMR do not have absolute stereochemistry, and can 48 49 produce both enantiomers from menthone and isomenthone. 10 50 51 52 To further investigate the catalytic abilities of MMR and MNMR, biotransformations with 53 54 DM and DN were performed with menthone and isomenthone at high levels of cell extract (1 mL) 55 56 for 1 hour. Complete conversion of menthone with DM was obtained, yielding highly pure 57 58 59 menthol (79.1 %). However reactions with isomenthone gave lower product yields (43.1 %) with a 60 14 ACS Paragon Plus Environment Page 15 of 30 ACS Synthetic Biology

1 2 near equal ratio of menthol and neoisomenthol (16.3:19.7), presumably due to epimerization 3 4 activity on isomenthone. Similarly, reactions of DN with menthone (735 M) produced almost 5 6 7 entirely neomenthol (89.9 % purity), while isomenthone reactions showed lower product yields 8 9 (440 M) with a ratio of mostly neomenthol and isomenthol (28:11). Therefore, there is potential 10 11 to generate highly pure menthol or neomenthol by eliminating MMR or MNMR from the operon, 12 13 14 respectively. 15 16 Table 3 . Menthol production biotransformations from optimization studies with DM cell extracts in E. coli strain 17 NiCO 2(DE3). 18 Condition Menthol Yield Menthol Purity Condition Menthol Yield Menthol Purity 19 (mg/g protein) (% total product) (mg/g protein) (% total product) 20 A) Medium B) IPTG concentration ( M) 21 Minimal 1.40 + 0.09(ND) 18 + 1(ND) 0 ND ND 22 LB 0.92 + 0.25(1.51) 31 + 8(54) 10 0.42 + 0.27 30 + 19 23 2YT 1.33 + 0.28(1.60) 37 + 8(56) 50 0.46 + 0.22 28 + 13 24 TB 1.39 + 0.34(0.51) 45 + 11(62) 100 0.91 + 0.26 42 + 12 25 SB 1.31 + 0.29(0.93) 56 + 13(59) 500 0.32 + 0.13 14 + 6 26 TBAIM 1.61 + 0.39 56 + 14 1000 0.17 + 0.01 7 + 1 27 28 C) Induction temperature (°C) D) Cell growth at induction (OD 600 nm) 29 15 0.91 + 0.04 58 + 3 Initial (induction) 2.02 + 0.40 57 + 11 30 20 1.32 + 0.14 57 + 6 0.3 1.93 + 0.54 56 + 16 31 25 0.82 + 0.11 42 + 6 0.7 1.28 + 0.18 55 + 8 32 30 0.60 + 0.15 19 + 5 1.1 1.44 + 0.40 57 + 16 33 37 ND ND 2.1 1.64 + 0.24 59 + 9 34 35 E) Induction length (h) 36 3 1.11 + 0.11 76 + 7 37 5 1.27 + 0.11 57 + 5 38 16 1.80 + 0.19 53 + 6 39 Culture growth conditions and extract production protocols for each optimization trial are described in the Methods section. Reactions (1 mL) were performed in buffer (50 mM Tris pH 7.0) containing pulegone (1 mM), cell extracts 40 + 41 (0.25 mL), NADP (10 μM), glucose (15 mM) and GDH (10 U). The reactions were agitated at 30 °C for 2 h at 130 rpm. 42 Product yields were determined by GC analysis using a DB-WAX column. Menthol yields are expressed as the mg 43 produced per g of protein in the cell extracts. Purity is the % menthol produced compared to the total product yields. 44 Minimal = minimal medium; LB = lysogeny broth; 2YT = 2YT broth; TB = Terrific broth; SB = Super broth; TBAIM = 45 Terrific broth auto induction medium; ND = none detected. Menthol yields in parentheses represents the activity 46 detected in uninduced culture extracts. Full data showing other products produced are shown in SI Table 4. 47 48 49 Given the large consumer and industrial market for menthol, further expression 50 51 optimization studies were performed with the DM construct (Table 3). The efficiency of menthol 52 53 54 production (mg/g extract protein) and purity (% of total product yield) was assessed (Table 3) 55 56 based on the total menthol produced, cell mass generated and relative proportions of DBR vs 57 58 59 MMR production/activity (SI Table 4). The effect of the medium composition showed that on 60 15 ACS Paragon Plus Environment ACS Synthetic Biology Page 16 of 30

1 2 average richer media gave higher yields and purity of menthol (Table 3A). Interestingly, uninduced 3 4 cultures in less rich media generated more menthol than induced cells, which is likely related to 5 6 7 the observation of a significant reduction in the levels of insoluble protein production (results not 8 9 shown). Auto induction media also gave slightly higher menthol yields than IPTG induced cells. 10 11 Using rich medium (TB), the optimal IPTG concentration for menthol production over 12 13 14 menthone/isomenthone was 100 M, with a post induction temperature of 20°C (Table 3B-C; SI 15 16 Table 4B-C). Slightly higher yields of menthol were produced when induction occurred within the 17 18 first couple of hours of culture growth (Table 3D), although the menthol purity was essentially the 19 20 21 same at any induction time (55-59 %). Interestingly, the highest purity of menthol (76 %) was 22 23 obtained at the shortest induction length (3 hours at 20°C; Table 3E), however the overall menthol 24 25 yield was reduced due to lower cell density. 26 27 28 Side reactions. An important aspect of designing an optimal pathway is the elimination of any 29 30 unwanted side reactions. An interesting observation in this context is the apparent correlation 31 32 33 between neomenthol loss and menthone gain (Table 2). This suggests the presence of an oxidase 34 35 acting on neomenthol, converting it back to menthone. We performed reactions where 36 37 neomenthol was incubated with DN and control E. coli extracts for 24 hours, to check if E. coli 38 39 40 contained a contaminating neomenthol oxidase. Significant menthone (261 M) and isomenthone 41 42 (173 M) were produced in the DN reactions, but not in the control ones, suggesting the oxidase 43 44 activity is due solely to the reversibility of the ketoreduction reaction by MNMR (Scheme 3a). This 45 46 47 reaction is likely to proceed via a proton abstraction from the hydroxyl group by a nearby basic 48 49 residue and a simultaneous hydride transfer to NADP+ resulting in oxidation at the 3-position. Prior 50 51 52 studies of menthone reductases showed that ketoreduction of menthone by MNMR, but not 53 10 54 MMR, was reversible, with a relatively high Km for neomenthol (1 mM). 55 56 Menthone and isomenthone isomerization within E. coli extracts (Figure 2) is likely to 57 58 59 proceed via a classical glutamate racemase-type mechanism (Scheme 3b). Firstly, a base extraction 60 16 ACS Paragon Plus Environment Page 17 of 30 ACS Synthetic Biology

1 2 of an acidic proton alpha to the carbonyl group results in an enolate formation. This acts as a 3 4 nucleophile, and abstracts a proton from an acidic residue. The proton could potentially be 5 6 7 attacked from either face of neomenthol, resulting in reformation of the initial substrate or 8 9 formation the isomerized product (Scheme 3b). 10 11 12 13 14 A) 15 16 NADPH + H + NADP O OH 17 18 19 20 ketone B) enolate reformation 21 formation 22 23 O O O O O 24 H H 25 B A 26 B = base 27 AH = protonated acid 28 29 Scheme 3. Possible mechanisms of the enzymatic a) neomenthol oxidation to menthone by MNMR and b) 30 menthone/isomenthone epimerization. The epimerization reaction is based on the current mechanism of glutamate 31 racemase. 14 32 33 34 35 CONCLUSIONS 36 37 38 We have successfully engineered the later stages of the mint production pathway, 39 40 employing a careful quantitative characterization of all individual building blocks (e.g. selection of 41 42 active and stereo/enantiomerically suitable biocatalysts) and developing simple unidirectional 43 44 45 gene expression systems for codon-optimized genes. In addition, we optimized the production 46 47 further by assessing the expression of each part within an assembled pathway by incorporating 48 49 protein identification tags. Other optimization strategies employed were the selection of the best 50 51 52 production host strain, characterizing the influence of the promoter strength on the concerted 53 54 enzyme expression and reaction in a pathway, optimizing the cofactor availability and 55 56 understanding and decreasing of side reactions (Figure 3). These optimization steps enabled 50% 57 58 59 improvement in production of menthol and in another case increased purity to 76% from 60 17 ACS Paragon Plus Environment ACS Synthetic Biology Page 18 of 30

1 2 pulegone, thus greatly improving the production from 0% in the first construct (DMN in strain 1). 3 4 The optimization steps described in this manuscript are required to achieve the maximum 5 6 7 production levels of any compound. By sidestepping the time-consuming host optimization and in 8 9 vivo biosynthesis for most of the optimization pipeline, we could substantially speed up the 10 11 process. 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 Figure 3. Optimization scheme and achieved menthol production and/or purity using cell extracts of recombinant E. 36 coli expressing plant biosynthetic genes (DMN and DM) in a one-pot biosynthesis starting from pulegone. CoF = 37 external cofactor recycling system incorporated in the reaction; Bl21(DE3)pLysS/NiCO 2(DE3) = E. coli strains; Init = 38 induction at inoculation. Menthol purity (%) reflects the proportion of menthol vs other monoterpenoid products. 39 40 41 42 Our studies have demonstrated the successful integration of chemical synthesis of pure 43 44 precursor(s) with biochemical analysis of each enzymatic steps to select the best combination of 45 46 enzymes, and powerful improved pathway assembly methods for easy engineering. In addition, 47 48 49 we have demonstrated moderate final production of the end compound(s); production enabling 50 51 easy analysis for further high throughput screening, by deploying one-pot biosynthesis of menthol. 52 53 Compared to the use of in vivo cells for production, this approach has particularly important 54 55 56 benefits when producing (semi-)toxic chemical compounds. This approach is also useful when 57 58 aiming for the easy optimization of individual enzyme steps or their combinations for the 59 60 18 ACS Paragon Plus Environment Page 19 of 30 ACS Synthetic Biology

1 2 production of chemically diverse compound libraries. These benefits are highly relevant when 3 4 optimizing the production of high-value chemicals in a high throughput manner. 5 6 7 8 9 METHODS 10 11 12 General reagents and procedures. All chemicals and solvents were purchased from 13 14 commercial suppliers, except where specified, and were of analytical grade or better. Media 15 16 17 components were obtained from Formedium (Norfolk, UK). Gene sequencing and oligonucleotide 18 19 synthesis were performed by Eurofins MWG (Ebersberg, German). Chemical syntheses were 20 21 monitored by thin layer chromatography using Merck aluminum foil coated TLC plates carrying 22 23 24 silica gel 60 F254 (0.2mm thickness). Ultraviolet light, cerium molybdate (12 g of ammonium 25 26 molybdate, 0.5 g of ceric ammonium molybdate, and 15 mL of concentrated sulfuric acid) and/or 27 28 29 phosphomolybdic acid (10 g of phosphomolybdic acid in 100 mL of absolute ethanol) were used to 30 31 detect compounds. Purification of compounds was carried out using column chromatography 32 33 (Fluka Analytical high-purity grade silica gel, 60 Å pore size, 220-440 mesh particle size, 35 -75 34 35 36 micron particle size). NMR spectra were recorded on a 400 MHz spectrometer and referenced to 37 38 the solvent. All monoterpenoid compounds were dissolved as stock solutions in absolute ethanol, 39 40 with a final concentration of 2 % (v/v) in the reactions. The 1H and 13 C NMR spectra for the 41 42 43 synthesized compounds can be found in the supporting information document (SI Figure 1). 44 45 Synthesis of isomenthone. Menthone (1.5g, 9.7 mmol) was dissolved in methanol (15 mL) 46 47 and 10% aqueous solution of NaOH (1.5 mL, 10% w/v) was added. The solution was stirred at 48 49 50 room temperature for 3 h followed by solvent removal. The compound(s) were dissolved in ethyl 51 52 acetate and washed with water and brine. The organic layer was dried over magnesium sulfate, 53 54 filtered and reduced. Isomenthone (451 mg, 30%) was separated from menthone (959 mg, 64%) 55 56 1 57 by column chromatography using a gradient elution (hexane/diethyl ether, 98/2 to 90/10, v/v). H 58 59 NMR (400 MHz; CDCl 3) δ 2.30 (ddt, 1H, H-2a, J = 13.1, 4.5, 1.2 Hz), 2.11 (dd, 1H, H-2b, J = 13.2, 10.1 60 19 ACS Paragon Plus Environment ACS Synthetic Biology Page 20 of 30

1 2 Hz), 2.06-1.91 (m, 4H, H-1, H-4, H-5a, H-8), 1.76-1.66 (m, 2H, H-5b, H-6a), 1.52-1.43 (m, 1H, H-6b), 3 4 0.99 (d, 3H, H-7a-c, J = 6.6 Hz), 0.93 (d, 3H, H-9a-c/ H-10a-c, J = 6.5 Hz), 0.84 (d, 3H, H-9a-c/ H-10a- 5 6 13 7 c, J = 6.6 Hz) C NMR (101 MHz; CDCl 3) δ 57.3 (C4), 48.1 (C2), 34.5 (C1), 29.5 (C6), 27.05 (C8), 27.0 8 9 (C5), 21.6 (C7), 21.0, 20.0 (C9, C10). 10 11 Synthesis of menthol and neomenthol. Sodium borohydride (90 mg, 3.36 mmol) was added 12 13 14 portion wise to a stirred solution of menthone (320 mg, 2.077 mmol) in tetrahydrofuran/methanol 15 16 (10 mL, 9/1, v/v) at 0 °C. The solution was returned to room temperature and stirred for 30 min. 17 18 The reaction was quenched at 0 °C by slow addition of water (4 mL), followed by addition of 1M 19 20 21 HCl (4 mL). The menthol derivatives were extracted with dichloromethane and the organic layer 22 23 was washed with water and brine. The organic layer was then dried over magnesium sulphate, 24 25 filtered and reduced. GC analysis was performed on a sample of the crude, which gave the product 26 27 28 ratio of menthol/neomenthol/neoisomenthol as 60/30/4. Menthol (165 mg, 51%), neomenthol 29 30 (88 mg, 27%) and neoisomenthol (trace) were then separated and isolated by column 31 32 chromatography using a gradient elution system (hexane/diethyl ether, 49/1 to 9/1 , v/v). 33 34 1 35 Menthol: H NMR (CDCl 3, 400 MHz) δ 3.39 (td, 1H, H-3, J = 10.5, 4.3 Hz), 2.16 (dtd, 1H, H-8, J = 36 37 14.0, 7.0, 2.8 Hz), 1.97-1.92 (m, 1H, H-2a), 1.67-1.56 (m, 3H, OH, H-5a, H-6a), 1.44-1.36 (m, 1H, H- 38 39 40 1), 1.09 (ddt, 1H, H-4, J = 12.4, 9.8, 2.8 Hz), 1.04-0.81 (m, 3H, H-2b, H-5b, H-6b), 0.91 – 0.90 (d, 3H, 41 13 42 CH 3, J = 6.27 Hz), 0.90 - 0.88 (d, 3H, CH 3, J = 5.83 Hz), 0.79 (d, 3H, CH 3, J = 7.0 Hz,) C NMR (101 43 44 MHz; CDCl 3) δ 71.6 (C3), 50.2 (C4), 45.1 (C2), 34.6 (C6), 31.7 (C1), 25.9 (C8), 23.2 (C5), 22.3 (C7), 45 46 1 47 21.1 (C10), 16.2 (C9). Neomenthol: H NMR (CDCl 3, 400 MHz) δ 4.10 – 4.09 (d, 1H, H-3, J = 2.85 Hz), 48 49 1.85 – 1.80 (dq, 1H, H-2a, J = 13.72, 3.02 Hz), 1.74 – 1.63 (m, 3H, H-1, H-5a, H-6a), 1.56 – 1.47 (ddt, 50 51 1H, H-8, J = 13.33, 9.34, 6.67 Hz), 1.37 (s, 1H, OH), 1.28 – 1.21 (m, 1H, H-5b), 1.11 – 1.10 (m, 1H, H- 52 53 54 2b), 0.96 – 0.94 (d, 3H, H-7a-c, J = 6.67), 0.92 – 0.90 (d, 3H, H-9a-c/10a-c, J = 6.65), 0.87 – 0.85 (d, 55 13 56 3H, H-9a-c/10a-c, J = 6.33), 1.00 – 0.84 (m, 2H, H-4, H-6b) C NMR (101 MHz; CDCl 3) δ 67.8 (C3), 57 58 59 60 20 ACS Paragon Plus Environment Page 21 of 30 ACS Synthetic Biology

1 2 48.0 (C4), 42.7 (C2), 35.2 (C6), 29.3 (C8), 25.9 (C1), 24.3 (C5), 22.4 (C7), 21.3 (C9/C10), 20.8 3 4 (C9/C10). 5 6 7 Synthesis of neoisomenthol. Sodium borohydride (45 mg, 1.168 mmol) was added portion 8 9 wise to a stirred solution of menthone (150 mg, 0.974 mmol) in tetrahydrofuran/methanol (5 mL, 10 11 9/1, v/v) at 0 °C. The solution was returned to room temperature and stirred for 30 mins. The 12 13 14 reaction was quenched at 0 °C by slow addition of water (2 mL) followed by addition of 1M HCl (2 15 16 mL). The menthol derivatives were extracted with dichloromethane and the organic layer was 17 18 washed with water and brine. The organic layer was then dried over magnesium sulphate, filtered 19 20 21 and reduced. GC analysis was performed on a sample of the crude, which gave the product ratio of 22 23 neoisomenthol/menthol/neomenthol/isomenthol as 93/4/1.5/0.6. Neoisomenthol (123 mg, 82%) 24 25 was separated from the other isomers by column chromatography (hexane/diethyl ether, 99/1 to 26

27 1 28 95/5, v/v). H NMR (CDCl 3, 400 MHz) δ 4.02 - 3.99 (dt, 1H, H-3, J = 6.26, 3.22 Hz), 1.78 – 1.70 (m, 29 30 1H, H-1), 1.68 – 1.52 (m, 4H, H-8, H-2a, H-2b, H-5a), 1.46 – 1.36 (m, 4H, H-5b. H-6a, H-6b, OH), 31 32 1.15 -1.083 (m, 1H, H04), 1.067 -1.049 (d, 3H, CH , J = 7.10 Hz), 0.99 -0.98 (d, 3H, CH , J = 6.63 Hz), 33 3 3 34 13 35 0.92 -0.90 (d, 3H, CH 3, J = 6.67 Hz) C NMR (CDCl 3, 101 MHz) δ 70.8 (C3), 47.5 (C4), 39.1 (C2), 31.1 36

37 (C6), 28.4 (C1), 27.6 (C8), 21.9 (C5), 21.6 (3 x CH3). 38 39 40 Synthesis of isomenthol. Neomenthol (80 mg, 0.519 mmol) was dissolved in anhydrous 41 42 tetrahydrofuran (5 mL). To this solution, triphenylphosphine (163 mg, 0.622 mmol) and p- 43 44 nitrobenzoic acid (104 mg, 0.622 mmol) was added. Upon dissolution of these reagents, di-tert- 45 46 47 butylazodicarboxylate (143 mg, 0.622 mmol) was added portion wise over 30 mins. The reaction 48 49 was stirred at room temperature overnight. The solvent was reduced and the diethyl ether was 50 51 added to the crude, causing triphenylphosphine oxide to precipitate. This was filtered off, the 52 53 54 solvent was removed and the process repeated till no more triphenylphosphine oxide 55 56 precipitated. The crude was then purified by column chromatography (hexane/diethyl ether, 98/2, 57 58 v/v) and the p-nitrobenzoate derivative (107 mg) was isolated in 68 % yield. 1H-NMR (400 MHz; 59 60 21 ACS Paragon Plus Environment ACS Synthetic Biology Page 22 of 30

1 2 CDCl 3) δ 8.29-8.20 (m, 4H, ArC H), 5.36 – 5.32 (td, 1H, H-3, J = 6.2, 3.3 Hz), 1.98 -1.32 (qd, 1H, H-1, J 3 4 = 7.4, 3.9 Hz), 1.83-1.73 (m, 2H, H-2a, H-8), 1.69 (dt, 1H, H-5a, J = 8.9, 4.3 Hz), 1.63-1.48 (m, 4H, H- 5 6 7 2b, H-4, H-5b, H-6a), 1.28 (dtd, 1H, H-6b, J = 12.6, 8.6, 3.7 Hz), 0.98 (dd, 6H, H-9a-c, H-10a-c, J = 8 13 9 8.0, 6.9 Hz), 0.90 (d, 3H, H-7a-c, J = 6.7 Hz) C NMR (101 MHz; CDCl 3) δ 164.1 (C=O), 150.5 10 11 (Ar CNO 2), 136.5 (Ar C), 130.7 (Ar CH), 123.6 (Ar CH), 74.0 (C3), 45.7 (C4), 35.7 (C2), 29.9 (C6), 27.8 12 13 14 (C1), 26.5 (C8), 21.4 (C5), 21.01, 20.81, 19.4 (C7, C9, C10). 15 16 The p-nitrobenzoate derivative was dissolved in a tetrahydrofuran/water mixture (2 mL, 4/1, 17 18 v/v) and lithium hydroxide monohydrate (47 mg) was added. The solution was placed in a sealed 19 20 21 sample vial and heated to 40°C for 4h. Water was then added, and the compound was extracted 22 23 with diethyl ether. The organic layer was dried over magnesium sulphate, filtered and reduced. 24 25 Column chromatography (hexane/diethyl ether, 98/2, v/v) gave isomenthol (46 mg) in an 85 % 26

27 1 28 yield. H NMR (CDCl 3, 400 MHz) δ 3.83 – 3.78 (td, 1H, H-3, J = 7.89, 3.79), 2.024 (m, 2H, H-1, H-8), 29 30 1.65 – 1.27 (m, 6H, H-2a, H-2b, H-5a, H-5b, H-6a, H-6b), 1.419 (s, 1H, OH), 1.19 – 1.12 (m, 1H, H-4), 31 32 0.953 – 9.35 (d, 6H, H-9a-c, H-10a-c, J = 7.0), 0.88 – 0.87 (d, 3H, H-7a-c, J = 6.83 Hz) 13 C NMR (101 33 34 35 MHz; CDCl 3) δ 68.0 (C3), 49.7 (C4), 40.1 (C2), 30.5 (C6), 27.6, 26.1 (C8, C1), 21.0, 19.9 (C9/ C10), 36 37 19.5 (C5), 18.1 (C7). 38 39 40 Gene synthesis and modifications. The double bond reductase from Nicotiana tabacum 41 11 42 (NtDBR-His 6 in pET21b; Uniprot: Q9SLN8) was prepared as described previously. The protein 43 44 sequences for the following enzymes were obtained from UniProt (http://www.uniprot. org): i) (-)- 45 46 47 menthone:(-)menthol reductase from M. piperita (MMR; UniProt: Q5CAF4) and ii ) (-)- 48 49 menthone:(+)neomenthol reductase from M. piperita (MNMR; UniProt: Q06ZW2). The respective 50 51 gene sequences were designed and synthesised by GenScript (USA), incorporating codon 52 53 54 optimisation techniques of rare codon removal for optimal expression in E. coli . The genes were 55 56 sub cloned individually into pET21b (Novagen) via Nde I/ Xho I restriction sites, without a stop 57 58 codon, to incorporate a C-terminal His -tag to allow expression monitoring by Western blotting. 59 6 60 22 ACS Paragon Plus Environment Page 23 of 30 ACS Synthetic Biology

1 2 Due to poor expression of the MNMR-His 6 construct, the gene was sub cloned into pET15b, via 3 4 Nde I/ Xho I restriction sites, to generate a N- and C-terminally His 6-tagged protein. Each construct 5 6 7 was transformed into the E. coli strain BL21(DE3)pLysS (Stratagene) for soluble protein over- 8 9 expression according to the manufacturer’s protocol. 10 11 Protein production and purification. A general protocol for the production and purification 12 13 14 of each individual His 6-tagged protein was used, based on the NtDBR method described 15 16 previously. 11 Cultures of E. coli BL21(DE3)pLysS containing expression vectors were grown in (12 x 17 18 1 L) Terrific broth (TB; tryptone 12 gL -1 and yeast extract 24 gL -1; pH 7.0), supplemented with 19 20 -1 -1 21 glycerol (0.4 %), ampicillin (100 mgmL ; 15 mgmL kanamycin for MNMR in pET15b) and 22 -1 23 chloramphenicol (34 mgmL ). Cultures were incubated at 37°C until OD 600nm reached 0.5, 24 25 followed by a 16 hour induction with isopropyl-β-D-1-thiogalactopyranoside (IPTG; 10 M) at 26 27 28 25 °C. Cells were harvested by centrifugation at 5000 g for 10 min at 4°C. Cell pellets were 29 30 resuspended in lysis buffer 1 (50 mM Tris pH 8.0 containing the EDTA-free complete protease 31 32 -1 -1 33 inhibitor cocktail, 1 mM MgCl 2, 0.1 mg mL DNase I, 0.1 mg mL lysozyme and 10% glycerol) and 34 35 stirred for 20 min at 4 °C. Cells were disrupted by sonication (Sonics Vibra Cell) followed by extract 36 37 clarification by centrifugation for 60 min at 26600g. The clarified supernatants were passaged 38 39 2+ 11 40 twice through Ni Sepharose, as described previously. Subsequent gel filtration was required for 41 42 MMR and MNMR on a HiLoad 16/600 Superdex 200pg column (GE Healthcare) column pre- 43 44 equilibrated in buffer A (50 mM Tris pH 8.0 containing 1 mM β-mercaptoethanol, 10% sorbitol, 45 46 47 10% glycerol). An isocratic elution of the protein from the column was carried out in the same 48 49 buffer. Purified enzymes were dialysed into cryobuffer (10 mM Tris pH 7.0 containing 10 % 50 51 52 glycerol), and flash frozen in liquid nitrogen for storage at -80°C. Protein concentration was 53 15 54 determined using the Bradford and extinction coefficient methods. In the case of MMR and 55 56 MNMR, 2-mercaptoethanol (1 mM) was included in all buffers. 57 58 59 60 23 ACS Paragon Plus Environment ACS Synthetic Biology Page 24 of 30

1 2 Protein detection and identification. Purity was assessed by SDS-PAGE, using 10-12% Mini- 3 4 PROTEAN® TGX Stain-Free™ gels and Precision Plus protein unstained markers (BioRad) according 5 6 7 to the manufacturers instructions. Identification of His 6-tagged proteins was performed by 8 9 Western blots using the Trans-Blot® Turbo™ Transfer system (PVDF membranes; BioRad) and the 10 11 Western Breeze Chemiluminescent Immunodetection kit (alkaline phosphatase; Life Technologies) 12 13 14 with mouse (His tag monoclonal antibody) and alkaline phosphatase-containing (Anti-C-My) 15 16 primary and secondary antibodies, respectively. 17 18 Enzyme kinetics. The concentration of nicotinamide coenzymes (Melford) was determined by 19 20 -1 -1 21 the extinction coefficient method ( ε340 = 6220 M cm ). Steady-state kinetic analyses were 22 23 performed on a Cary UV-50 Bio UV/Vis scanning spectrophotometer using a quartz cuvette (1 mL; 24 25 Hellma) with a 1 cm path length. Standard reactions (1 mL) were performed in buffer (50 mM Tris 26 27 28 pH 7.0) containing NADPH (50-100 M for MMR/MNMR and NtDBR, respectively), monoterpenoid 29 30 (1 mM) and enzyme (30 nM to 2 M). Reactions were followed by continuously monitoring 31 32 33 NADPH oxidation at 340 nm for 1 min at 25 °C. The standard monoterpenoid substrates used were 34 35 pulegone for NtDBR (alkene reduction) 11 and menthone/isomenthone for both MMR and MNMR 36 37 (ketoreduction). 16 Ketoreductase reactions were performed in an alternative buffer (12.5 mM tri- 38 39 40 sodium citrate, 12.5 mM KH 2PO 4, 12.5 mM K 2HPO 4 and 12.5 mM CHES) containing dithiothreitol (1 41 42 mM; DTT). All steady state reactions were performed in at least duplicate. 43 44 Operon construction. Co-expressing multi-gene constructs were prepared in pET21b using 45 46 47 the recombination-based In-Fusion® HD Cloning Kit (Takara/Clontech) with sequential gene 48 49 addition according to the manufacturers protocols. These protocols (Figure 2) are summarised by 50 51 52 the following steps: i) vector linearisation (containing gene 1) and new gene amplification by PCR. 53 54 A 13 bp Shine-Dalgarno sequence (SD; GGAGGACAGCTAA) was incorporated between the stop 55 56 and start codons of successive genes to allow expression from one promoter (T7 lac ; Figure 2); ii ) 57 58 59 template removal by cloning enhancer (Takara/Clontech) or Dpn I (New England Biolabs; NEB) 60 24 ACS Paragon Plus Environment Page 25 of 30 ACS Synthetic Biology

1 2 restriction digest; iii ) gel extraction and purification of PCR products; iv ) In-Fusion cloning reaction 3 4 between the vector and new gene; v) transformation into an E. coli cloning strain (Stellar; 5 6 7 Takara/Clontech); vi ) plasmid preparation and sequencing; vii ) repeating the above steps to 8 9 incorporate any additional genes. Each PCR product contained a 15 bp overlap (Figure 2) between 10 11 the vector and insert pairs to facilitate recombination. To generate the long overhangs, 12 13 14 overlapping sets of PCR primers were used, with the outermost oligo at a considerably higher 15 16 concentration than the innermost one (ratio 5:1). The PCR primers used are found in SI Table 5. In 17 18 most cases the PCR reactions, template removal and gel purification of DNA were performed using 19 20 21 the kit protocols and enzymes (CloneAmp PCR premix; cloning enhancer). However, in some cases 22 23 alternative enzymes were used, such as Q5 DNA polymerase and Dpn I restriction enzyme (NEB). 24 25 The following three constructs were generated: i) NtDBR-His –SD–MMR-His –SD–His -MNMR 26 6 6 6 27 28 (DMN); ii ) NtDBR-His 6–SD–MMR-His 6 (DM) and iii ) NtDBR-His 6–SD–His 6-MNMR (DN). Constructs 29 30 were transformed into the E. coli expression strain BL21(DE3)pLysS, according to the 31 32 manufacturers’ protocols, to check the expression levels of the individual genes. Cell extracts of 33 34 35 each culture were obtained (Section 2.9.2), and checked for recombinant protein expression by 36 37 SDS PAGE, Western blotting and biotransformations. 38 39 40 Optimisation of multi-gene expression. The multi-gene construct DMN was transformed into 41 42 a further eleven E. coli expression strains according to the manufacturers’ protocols (Table 5). 43 44 Small-scale cultures of each strain (1 L) were produced as described in Table 5. A generalised 45 46 47 Terrific broth autoinduction medium (TBAIM; Formedium) protocol was chosen, except for strains 48 49 with the pLysS phenotype and Arctic Express (DE3) cells. However, autoinduction protocols were 50 51 used for the control cells (Bl21(DE3)pLysS containing pET21b), as no recombinant gene expression 52 53 54 is needed. Further optimisation was performed by culture growth under the conditions used to 55 56 generate the individual enzymes, except the IPTG concentration was varied (10 M to 1 mM). Cell 57 58 59 60 25 ACS Paragon Plus Environment ACS Synthetic Biology Page 26 of 30

1 2 extracts of each culture were obtained (Section 2.9.2), and checked for recombinant protein 3 4 expression by SDS PAGE, Western blotting and biotransformations. 5 6 Table 5. E. coli strains and growth conditions with pET21b constructs. 7 Name Strain Media Antibiotics Growth conditions a Source 8 b 9 Control Bl21(DE3)pLysS TBAIM Amp + Chl Autoinduction Novagen 10 Strain 1 Bl21(DE3)pLysS TB + 0.4% glycerol Amp + Chl Standard Novagen 11 Strain 2 Bl21-Codon Plus(DE3)RIPL TB + 0.4% glycerol Amp + Chl Standard Agilent 12 Strain 3 Arctic Express(DE3) TB + 0.4% glycerol Amp + Gent Low temperature Agilent 13 Strain 4 NiCO 2 21(DE3) TBAIM Amp Autoinduction NEB 14 Strain 5 Rosetta(DE3)pLysS TB + 0.4% glycerol Amp + Chl Standard Novagen 15 Strain 6 Rosetta2(DE3)pLysS TB + 0.4% glycerol Amp + Chl Standard Novagen 16 17 Strain 7 Tuner(DE3) TBAIM Amp Autoinduction Novagen 18 Strain 8 Bl21(DE3) TBAIM Amp Autoinduction Novagen 19 Strain 9 Rosetta2(DE3) TBAIM Amp + Chl Autoinduction Novagen 20 Strain 10 Origami2(DE3) TBAIM Amp + Strep Autoinduction Novagen 21 Strain 11 BLR(DE3) TBAIM Amp Autoinduction Novagen 22 Strain 12 HMS174(DE3) TBAIM Amp + Rifam Autoinduction Novagen 23 aStandard expression conditions: Incubate at 37 °C until OD =0.5, induce with 0.4 mM IPTG and incubate at 25 °C 24 600 overnight at 200 rpm shaking. Autoinduction conditions: Incubate at 25 C for 24 h at 200 rpm shaking. Low 25 ° 26 temperature = Incubate at 37 °C for 3 hours, then 10 minutes at 12 °C. Induce with 1 mM IPTG and incubate at 12 °C overnight at 200 rpm shaking. bContains empty pET21b. Amp = 100 gmL -1 ampicillin; Chl = 34 gmL -1 27 -1 -1 -1 28 chloramphe nicol; Gent = 20 gmL gentamycin; Strep = 50 gmL streptomycin; Rifam = 200 gmL rifampicin. 29 30 31 Biotransformation reactions. All biotransformation reactions were performed in at least 32 33 duplicates, and the results are averages of the data. Biotransformations with DBR were performed 34 35 11 36 as described previously. Ketoreductase reactions (1.0 mL) with purified enzymes were 37 38 performed in biotrans buffer (50 mM Tris pH 7.0) containing the monoterpenoid (5 mM), NADP + 39 40 41 (10 μM), glucose (15 mM), glucose dehydrogenase (GDH from Pseudomonas sp.; 10 U) and 42 43 enzyme (2 M). Reactions were shaken at 30 °C for 24 h at 130 rpm, and terminated by extraction 44 45 with ethyl acetate (0.9 mL) containing an internal standard (1% sec-butyl-benzene in ethyl 46 47 48 acetate). The extracts were dried using anhydrous magnesium sulphate, and analysed by GC. 49 50 Quantitative analysis was carried out by a comparison of product peak areas to standards of 51 52 known concentrations. Products were identified by comparison with authentic standards. 53 54 55 Cell pellets of multi-gene constructs (from 1 L culture; Table 5) were combined with equal 56 57 volumes (15 mL) of lysis buffer 2 (lysis buffer 1 at pH 7.0 containing 1 mM 2-mercaptoethanol), 58 59 60 26 ACS Paragon Plus Environment Page 27 of 30 ACS Synthetic Biology

1 2 and clarified extracts were produced as above. Standard biotransformation reactions were 3 4 performed with modifications (2 mL; 25 mL reaction vials; 25 °C) with cell extracts (0.5 mL per 2 mL 5 6 7 reaction, equivalent to around 30 mL of culture) in the presence or absence of an externally added 8 9 cofactor recycling system. The substrates tested were pulegone, menthone and isomenthone (1 10 11 mM), and all possible products were processed as described above. Control reactions were 12 13 14 performed with cell extracts of an E. coli construct containing an empty pET21b vector. To test the 15 16 effect of enzyme loading on product yields, the cell extract volume was varied (0.6-1.0 mL/2 mL 17 18 reaction). To minimize substrate/product decomposition and/or utilization by other E. coli 19 20 21 pathways, reactions (1 mL) were performed and analyzed at different times (1, 2, 6 and 24 h). 22 23 Latter optimization studies with DM cultures were performed as described above, except a 24 25 smaller culture volume was used (50 mL). The different medium compositions trialed (ForMedium 26 27 28 preparations unless stated) were minimal medium (M9 salts containing 10 M FeSO 4, 1 mM 29 30 MgSO 4, 100 M CaCl 2 and 0.4 % glycerol), LB, 2YT, TB, SB (35 g/L tryptone, 20 g/L yeast extract and 31 32 33 5 g/L NaCl) and TB auto induction medium. Induction parameters varied were IPTG concentration 34 35 (0 – 1 mM), induction time (OD 600 nm = 0.01-2.1), post induction temperature (15 - 37 °C) and 36 37 length (3-16 hours). Cell-free extracts and biotransformations (1.0 mL) were performed as 38 39 40 described above. 41 42 Analytical procedures. Reaction extracts (1 L) were analysed by gas chromatography on an 43 44 45 Agilent Technologies 7890A GC system equipped with an FID detector and a 7693 autosampler. A 46 47 DB-WAX column (30 m; 0.32 mm; 0.25 μm film thickness; JW Scientific) was used to separate the 48 49 seven substrates and product isomers/enantiomers: pulegone (co-elutes with isomenthol), 50 51 52 menthone, isomenthone, menthol, neomenthol and neoisomenthol. In this method the injector 53 54 temperature was at 220°C with a split ratio of 20:1 (1 L injection). The carrier gas was helium 55 56 with a flow rate of 1 mLmin -1 and a pressure of 5.1 psi. The program began at 40°C with a hold for 57 58 59 1 min followed by an increase of temperature to 210°C at a rate of 15°C/minute, with a hold at 60 27 ACS Paragon Plus Environment ACS Synthetic Biology Page 28 of 30

1 2 210 °C for 3 min. The FID detector was maintained at a temperature of 250°C with a flow of 3 4 hydrogen at 30mL/min. To quantify isomenthol yields, the program was modified (SI Figure 2) to 5 6 7 allow the separation of pulegone and isomenthol (increase of temperature to 210°C at a rate of 8 9 10°C/minute, with a hold at 210 °C for 1 min). Unknown products were identified by gas 10 11 chromatography in combination with mass spectrometry on an Agilent Technologies 7890A GC 12 13 14 with an Agilent Technologies 5975C inert XL-EI/CI MSD with triple axis detector and a Zebron ZB- 15 16 Semi Volatiles column (15m x 0.25mm x 0.25um film thickness, Phenomenex). In this method the 17 18 injector temperature was at 220°C with a split ratio of 10:1 (1 L injection). The carrier gas was 19 20 -1 21 helium with a flow rate of 1 mLmin and a pressure of 5.1 psi. The program began at 40°C with a 22 23 hold for 3 min followed by an increase of temperature to 210°C at a rate of 10°C/minute, with a 24 25 hold at 210 °C for 3 min. The mass spectra fragmentation patterns were entered into the 26 27 28 NIST/EPA/NIH 11 (mass spectral library for identification of a potential match. 29 30 31 32 33 34 ACKNOWLEDGEMENTS 35 36 We would like to thank Rainer Breitling for his critical reading of the manuscript. This work was 37 38 39 funded by the UK Biotechnology and Biological Sciences Research Council (BBSRC BB/J015512/1; 40 41 BB/M017702/1) and GlaxoSmithKline. NSS is a Royal Society Wolfson Merit Award holder and an 42 43 Engineering and Physical Sciences Research Council (EPSRC; EP/J020192/1) Established Career 44 45 46 Fellow. 47 48 49 50 51 Supporting Information Available : Additional methodology and results can be found in the 52 53 Supporting Information document. This material is available free of charge via the Internet at 54 55 http://pubs.acs.org. 56 57 58 59 60 28 ACS Paragon Plus Environment Page 29 of 30 ACS Synthetic Biology

1 2 REFERENCES 3 4 5 1. Turner, G. W., and Croteau, R. (2004) Organization of monoterpene biosynthesis in 6 Mentha . Immunocytochemical localizations of geranyl diphosphate synthase, limonene-6- 7 hydroxylase, isopiperitenol dehydrogenase, and pulegone reductase. Plant Physiology 136 (4), 8 4215-4227. 9 10 2. Duetz, W. A., Bouwmeester, H., van Beilen, J. B., and Witholt, B. (2003) Biotransformation 11 of limonene by bacteria, fungi, yeasts, and plants. Applied Microbiol. Biotechnol. 61 (4), 269-277. 12 3. Ajikumar, P. K., Tyo, K., Carlsen, S., Mucha, O., Phon, T. H., and Stephanopoulos, G. (2008) 13 Terpenoids: Opportunities for Biosynthesis of Natural Product Drugs Using Engineered 14 Microorganisms. Molecular Pharmaceutics 5 (2), 167-190. 15 16 4. (a) Lange, B. M., Mahmoud, S. S., Wildung, M. R., Turner, G. W., Davis, E. M., Lange, I., 17 Baker, R. C., Boydston, R. A., and Croteau, R. B. (2011) Improving peppermint essential oil yield 18 and composition by metabolic engineering. Proc. Natl. Acad. Sci. USA 108 (41), 16944-16949; (b) 19 Lawrence, B. M., Mint: The genus Mentha . Hardman, R., Ed. CRC Press: 2007; pp. 1-547. (accessed 20 Jul 10). 21 5. (a) Alonso-Gutierrez, J., Chan, R., Batth, T. S., Adams, P. D., Keasling, J. D., Petzold, C. J., and 22 23 Lee, T. S. (2013) Metabolic engineering of Escherichia coli for limonene and perillyl alcohol 24 production. Metab. Engin. 19 , 33-41; (b) Bicas, J. L., Dionísio, A. P., and Pastore, G. u. M. (2009) 25 Bio-oxidation of terpenes: An approach for the flavor industry. Chem. Rev. 109 (9), 4518-4531; (c) 26 Bohlmann, J., Steele, C. L., and Croteau, R. (1997) Monoterpene synthases from Grand Fir ( Abies 27 grandis ) cDNA isolation, characterization, and functional expression of myrcene synthase, (-)-(4 S)- 28 29 limonene synthase, and (-)-(1 S,5 S)-pinene synthase. J. Biol. Chem. 272 (35), 21784-21792. 30 6. Ajikumar, P. K., Xiao, W. H., Tyo, K. E. J., Wang, Y., Simeon, F., Leonard, E., Mucha, O., Phon, 31 T. H., Pfeifer, B., and Stephanopoulos, G. (2010) Isoprenoid pathway optimization for taxol 32 precursor overproduction in Escherichia coli . Science 330 (6000), 70-74. 33 7. (a) Carter, O. A., Peters, R. J., and Croteau, R. (2003) Monoterpene biosynthesis pathway 34 construction in Escherichia coli . Phytochemistry 64 (2), 425-433; (b) Farmer, W. R., and Liao, J. C. 35 36 (2001) Precursor balancing for metabolic engineering of lycopene production in Escherichia coli . 37 Biotechnol. Progr. 17 (1), 57-61; (c) Jackson, B. E., Hart-Wells, E. A., and Matsuda, S. P. T. (2003) 38 Metabolic engineering to produce sesquiterpenes in yeast. Org. Lett. 5 (10), 1629-1632; (d) Kiyota, 39 H., Okuda, Y., Ito, M., Hirai, M. Y., and Ikeuchi, M. (2014) Engineering of cyanobacteria for the 40 photosynthetic production of limonene from CO2. J. Biotechnol. 185 , 1-7; (e) Martin, V. J. J., Pitera, 41 42 D. J., Withers, S. T., Newman, J. D., and Keasling, J. D. (2003) Engineering a mevalonate pathway in 43 Escherichia coli for production of terpenoids. Nat. Biotechnol. 21 (7), 796-802. 44 8. Chang, M. C. Y., and Keasling, J. D. (2006) Production of isoprenoid pharmaceuticals by 45 engineered microbes. Nat. Chem. Biol. 2 (12), 674-681. 46 9. Denard, C. A., Hartwig, J. F., and Zhao, H. (2013) Multistep one-pot reactions combining 47 biocatalysts and chemical catalysts for asymmetric synthesis. ACS Catalysis 3 (12), 2856-2864. 48 49 10. Davis, E. M., Ringer, K. L., McConkey, M. E., and Croteau, R. (2005) Monoterpene 50 metabolism. Cloning, expression, and characterization of menthone reductases from peppermint. 51 Plant Physiol. 137 (3), 873-881. 52 11. Mansell, D. J., Toogood, H. S., Waller, J., Hughes, J. M. X., Levy, C. W., Gardiner, J. M., and 53 Scrutton, N. S. (2013) Biocatalytic asymmetric alkene reduction: crystal structure and 54 characterization of a double bond reductase from Nicotiana tabacum . ACS Catalysis 3 , 370-379. 55 56 12. Hirata, T., Matsushima, A., Sato, Y., Iwasaki, T., Nomura, H., Watanabe, T., Toyoda, S., and 57 Izumi, S. (2009) Stereospecific hydrogenation of the C=C double bond of enones by Escherichia 58 coli overexpressing an enone reductase of Nicotiana tabacum . J. Mol. Catal. B: Enzym 59 (1-3), 59 158-162. 60 29 ACS Paragon Plus Environment ACS Synthetic Biology Page 30 of 30

1 2 13. Kataoka, M., Rohani, L. P. S., Wada, M., Kita, K., Yanase, H., Urabe, I., and Shimizu, S. 3 (1998) Escherichia coli transformant expressing the glucose dehydrogenase gene from Bacillus 4 megaterium as a cofactor regenerator in a chiral alcohol production system. Biosci. Biotechnol. 5 Biochem. 62 (1), 167-169. 6 7 14. Yoshimura, T., and Esak, N. (2003) Amino acid racemases: functions and mechanisms. J. 8 Biosci. Bioengin. 96 (2), 103-109. 9 15. Peterson, G. L. (1983) The determination of total protein. Meth. Enzymol. 91 , 95-119. 10 16. Croteau, R. B., Davis, E. M., Ringer, K. L., and Wildung, M. R. (2005) (−)-Menthol 11 biosynthesis and molecular genetics. Naturwissenschaften 92 (12), 562-577. 12 13 14 15 16 17 18 GRAPHICAL ABSTRACT 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 30 ACS Paragon Plus Environment