Contents lists available at ScienceDirect

Metabolic Engineering

journal homepage: www.elsevier.com/locate/ymben

Original Research Article

Development of a Bacillus subtilis cell-free transcription-translation systemfor prototyping regulatory elements

Richard Kelwicka,b,1, Alexander J. Webba,b,1, James T. MacDonalda,b, Paul S. Freemonta,b,⁎

a Centre for Synthetic Biology and Innovation, Imperial College London, London SW7 2AZ, UKb Section of Structural Biology, Department of Medicine, Imperial College London, London SW7 2AZ, UK

A R T I C L E I N F O

Keywords:Synthetic biologyBacillus subtilisCell-free transcription-translationRegulatory element prototypingPromoter libraryLuciferase assay

A B S T R A C T

Cell-free transcription-translation systems were originally applied towards in vitro protein production. Morerecently, synthetic biology is enabling these systems to be used within a systematic design context forprototyping DNA regulatory elements, genetic logic circuits and biosynthetic pathways. The Gram-positive soilbacterium, Bacillus subtilis, is an established model organism of industrial importance. To this end, wedeveloped several B. subtilis-based cell-free systems. Our improved B. subtilis WB800N-based system wascapable of producing 0.8 µM GFP, which gave a ~72x fold-improvement when compared with a B. subtilis 168cell-free system. Our improved system was applied towards the prototyping of a B. subtilis promoter library inwhich we engineered several promoters, derived from the wild-type Pgrac (σA) promoter, that display a rangeof comparable in vitro and in vivo transcriptional activities. Additionally, we demonstrate the cell-freecharacterisation of an inducible expression system, and the activity of a model enzyme - renilla luciferase.

1. Introduction

Cell-free systems that are based on cellular extracts were originallydeveloped as experimental systems to understand fundamental aspectsof molecular biology, cellular biochemistry and for in vitro proteinproduction (Hodgman and Jewett, 2012; Nirenberg, 2004; Sullivanet al., 2016; Zubay, 1973). Synthetic biology approaches are enablingthe re-purposing of cell-free systems as coupled in vitro transcription–translation characterisation platforms for the prototyping of DNAbased parts, devices and systems (Kelwick et al., 2014). Cell-freetranscription-translation systems have been employed to rapidly pro-totype DNA regulatory elements (Chappell et al., 2013), logic systems(Niederholtmeyer et al., 2015; Shin and Noireaux, 2012; Sun et al.,2014; Takahashi et al., 2015) and medical biosensor devices (Pardeeet al., 2014) with workflows that can be completed within severalhours. In contrast, typical in vivo approaches may take several days.Another distinguishing advantage of cell-free systems is that they canbe coupled with model-guided design strategies to create ‘biomolecu-lar-breadboards’ that enable the robust cell-free characterisation ofbioparts that can then be implemented as final designs in vivo (Siegal-Gaskins et al., 2014). These developments are also enabling cell-free

protein synthesis driven metabolic engineering approaches for thebiochemical characterisation of novel enzymes and prototyping ofbiosynthetic pathways (Karim and Jewett, 2016).

Several cell-free systems have been developed, of which, the mostwell established systems use cellular extracts from Escherichia coli(Garamella et al., 2016), Wheat Germ (Ogawa et al., 2016), Yeast (Ganand Jewett, 2014) or HeLa cells (Gagoski et al., 2016). Additionally,more specialist cell-free systems including the PURE express system,which uses purified cellular machinery rather than cellular extracts,have also been established (Shimizu et al., 2005). Whilst these cell-freesystems have been continually improved through developments in themethods used for their preparation (Shrestha et al., 2012) andoptimisation of energy buffers (Caschera and Noireaux, 2015a,2014), there have been fewer reports of Bacillus subtilis cell-freesystems. Yet, the development of robust B. subtilis cell-free systemscould have applicability to a broad array of microbiology, syntheticbiology and industrial biotechnology applications. Applications for B.subtilis are diverse and include the production of industrial orpharmaceutical proteins, and more recently for use as whole-cellbiosensors (Harwood, 1992; Pohl et al., 2013; Webb et al., 2016;Westers et al., 2004). Cell-free systems could be applied to support

http://dx.doi.org/10.1016/j.ymben.2016.09.008Received 6 May 2016; Received in revised form 8 September 2016; Accepted 29 September 2016

⁎ Correspondence to: Section of Structural Biology, Department of Medicine, Sir Alexander Fleming Building, South Kensington Campus, Exhibition Road, London SW7 2AZ, UK.

1 Joint First AuthorsE-mail address: p.freemont@imperial.ac.uk (P.S. Freemont).

Abbreviations: ATP, adenosine triphosphate; CTP, cytidine triphosphate; GTP, guanosine triphosphate; UTP, uridine triphosphate; GFPmut3b, green fluorescent protein mut3bvariant; Mg, magnesium; K, potassium; tRNA, transfer ribonucleic acid; 3-PGA, 3-phosphoglycerate; NAD, nicotinamide; CoA, coenzyme A; cAMP, adenosine 3′,5′-cyclicmonophosphate; DTT, dithiothreitol

Metabolic Engineering xx (xxxx) xxxx–xxxx

1096-7176/ © 2016 The Authors. Published by Elsevier Inc. on behalf of International Metabolic Engineering Society. This is an open access article under the CC BY license(http://creativecommons.org/licenses/by/4.0/).

Available online xxxx

Please cite this article as: Kelwick, R., Metabolic Engineering (2016), http://dx.doi.org/10.1016/j.ymben.2016.09.008

developments across these applications, particularly, where the func-tionality of the engineered system relates to aspects of the biochem-istry, metabolism and/or regulatory processes of B. subtilis as well aspotentially other Gram-positive bacteria. For instance, the cell-freeprototyping of B. subtilis regulatory elements (e.g. promoter libraries)may provide synergistic benefits when coupled with in vivo studies,such that multiple rounds of cell-free characterisation workflows mayresult in more rapid iterations of the design cycle towards the final invivo design (Chappell et al., 2013; Karim and Jewett, 2016; Tuza et al.,2013).

However, the initially reported B. subtilis cell-free systems weretypically encumbered by the requirement to use exogenous mRNA,protease inhibitors, DNAse treatments or less efficient energy systems(Legault-Demare and Chambliss, 1974; Leventhal and Chambliss,1979; Nes and Eklund, 1983; Okamoto et al., 1985; Zaghloul andDoi, 1987) which is perhaps why, despite their potential, these systemshave been largely neglected. In the present study, we report on thedevelopment and improvement of a B. subtilis cell-free system, using astandardised workflow that has no such limitations. We demonstratethe utility of B. subtilis cell-free transcription-translation systems as auseful tool for genetic regulatory element prototyping through thecharacterisation of an engineered promoter library that enables a rangeof comparable in vitro and in vivo transcriptional activities.Additionally, as a step towards additional applications for B. subtiliscell-free systems, we characterise an inducible expression system (aprecursor to genetic circuit prototyping) and, characterise the activityof the Renilla (sea pansy) luciferase (a model enzyme).

2. Materials and methods

2.1. Bacterial strains, plasmids and growth conditions

Bacterial strains used in this study are listed in Table S1. E. colistrains were grown in Luria-Bertani (LB) medium at 37 °C whilst B.subtilis strains were grown in 2× YTP medium (31 g/L 2× YT, 40 mMpotassium phosphate dibasic, 22 mM potassium phosphate monoba-sic) at 30 °C. When applicable, the medium was supplemented with thefollowing antibiotics: E. coli cultures - ampicillin (Amp) 100 μg/ml;chloramphenicol (Cam) 50 μg/ml; kanamycin (Kan) 35 μg/ml; B.subtilis WB800N cultures – chloramphenicol 5 μg/ml; kanamycin10 μg/ml. Kanamycin is used to select for the neomycin (Neo)resistance gene in B. subtilis WB800N.

2.2. Strain and plasmid construction

Oligonucleotide primers for plasmid construction and sequencingare listed in Table S2.

2.2.1. GFPmut3b expression vectorThe GFPmut3b expression vector pHT01-gfpmut3b was con-

structed as follows. The insert gfpmut3b was amplified from plasmidpAJW26 (BBa_K316008) using primer pair AJW289/AJW290, theresultant PCR product was purified, digested with enzymes BamHI andXbaI and ligated with the vector pHT01, which had been digested withthe same enzymes, resulting in the construction of plasmid pHT01-gfpmut3b (pAJW107). To remove LacI control, lacI was deleted fromplasmids pHT01 and pHT01-gfpmut3b as follows: inverted PCRreactions using primer pair AJW320/AJW321 and plasmids pAJW9and pAJW107 as templates were undertaken, the DNA products werepurified, phosphorylated, self-ligated and transformed into E. coliNEB10-beta, resulting in the plasmids pHT01-ΔlacI (pAJW118) andpHT01-ΔlacI-gfpmut3b (pWK-WT). To generate a pHT01-ΔlacI-gfpmut3b construct lacking the −35 and −10 boxes and the regionbetween the two boxes, plasmid pWK-WT was used as the template inan inverted PCR reaction with primers WK5/WK6. The resultant DNAproduct was purified, phosphorylated, self-ligated and transformed

into E. coli NEB10-beta, resulting in the plasmid pHT01-ΔlacI-Δbox-gfpmut3b (pWK-Δbox).

2.2.2. Promoter library constructionTo construct the promoter library of clones with changes to the −35

and −10 boxes, inverted PCR was undertaken using pWK-WT as thetemplate and primer pair WK1/WK2. The resultant PCR product waspurified, phoshporylated, self-ligated, transformed into E. coli NEB10-beta and the colonies cultured on plates incubated at either 30 °C or37 °C. This resulted in the production of the pWK(n) plasmid promotervariants. To create targeted changes to the −10 box, inverted PCR wasundertaken using pWK-WT and pWK5 as the templates and primerpair WK7/WK8. The products were purified, phosphorylated, self-ligated and transformed into E. coli NEB10-beta, resulting in theplasmids pWK403 and pWK501 respectively. Promoter library clonestested in this study are listed in Table S6.

2.2.3. GFPmut3b purification vectorPrimer pair RK003 and RK004 were designed to PCR amplify

gfpmut3b along with the addition of restriction sites BamHI andHindIII from plasmid pRK1. The subsequent PCR product wasdesigned so that it could be digested with BamHI and HindIII andligated into pre-digested vector pPROEX HTb to form pRK2 – a vectorin which N-terminally His-tagged GFPmut3b protein production couldbe induced.

2.2.4. Renilla luciferase vectorPrimer pair RK005/RK006 were designed to PCR amplify the

renilla luciferase gene along with the addition of restriction sitesBamHI and XbaI from plasmid pRK5. The subsequent PCR productwas designed so that it could be digested with BamHI and XbaI andligated into pre-digested vector pWK-WT to form pRK6 – a vector inwhich Renilla Luciferase enzyme could be constitutively expressed.

The DNA of all inserts/constructs were verified by the sequencingservice provided by Eurofins Genomics GmbH (Ebersberg, Germany).Primers AJW10 and AJW11 were used to sequence pSB1C3 basedconstructs and primers AJW77, AJW78, AJW322 and AJW376 wereused to sequence either pHT01 or pHT01-ΔlacI based constructs.Primer WK3 was used to sequence the gfpmut3b constructs whilstprimers RK001 and RK002 were used to sequence pPROEX HTb His-gfpmut3b.

2.3. Cell-free extract preparation

To prepare cell-free extracts, B. subtilis 168 cells were revived fromglycerol stocks onto LB plates whilst B. subtilis WB800N cells wererevived from glycerol stocks onto LB plates supplemented withkanamycin (Kan;10 µg/ml). Once streaked, plates were incubated for48 h at 30 °C. Individual colonies were inoculated into 5 ml 2× YTPmedium and incubated for 10 h with shaking (180 rpm) at 30 °C. Theresultant cultures were diluted (1:500) into flasks containing 50 ml 2×YTP medium and incubated for 10 h with shaking (180 rpm) at 30 °C.Resultant cultures were either harvested for cell lysis or, for largerscales of production, they were diluted (1:500) into flasks containing500 ml 2× YTP medium and incubated for 10 h with shaking (180 rpm)at 30 °C. To harvest cells, 500 ml cultures were centrifuged at 3,220gfor 15 min. Cell pellets were re-suspended into 20 ml S30-A buffer(14 mM Magnesium (Mg) glutamate, 60 mM Potassium (K) glutamate,50 mM Tris, 2 mM DTT, pH 7.7) and transferred into a pre-weighed50 ml Falcon tube. Each 50 ml Falcon tube was centrifuged (2,000g,10 min, 4 °C), pellets washed with 20 ml S30-A buffer and subse-quently re-centrifuged (2,000g, 10 min, 4 °C) to form the final cellpellets in preparation for cell lysis. To determine the weight of the cellpellet, the weight of the 50 ml falcon tube was subtracted from thecombined weight of the 50 ml tube and cell pellet. Pellets were stored at−80 °C for no more than 48 h, prior to cell lysis.

R. Kelwick et al. Metabolic Engineering xx (xxxx) xxxx–xxxx

2

To lyse the cells, pellets were defrosted on ice and re-suspendedinto 1 ml S30-A buffer per gram of cell pellet and aliquoted as 1 mlsamples in 1.5 ml microtubes. Samples were sonicated on ice (3×40 swith 1-min cooling interval; output frequency: 20 kHz; amplitude:50%) and then centrifuged (12,000g at 4 °C for 10 min). The super-natants were removed, aliquoted at 500 μl into 2 ml screw cap tubesand incubated with shaking (180 rpm) at 37 °C for either 0, 30 or80 min. Post pre-incubation, samples were stored on ice and thencentrifuged (12,000g at 4 °C for 10 min). Supernatants were removedand were either aliquoted into 1.5 ml tubes that were stored on ice orinto dialysis cassettes (GeBAflex-Maxi Dialysis Tubes −8 kDa MWCO,Generon) for dialysis into S30-B buffer (14 mM Mg-glutamate, 60 mMK-glutamate, ~5 mM Tris, 1 mM DTT; pH 8.2) with stirring at 4 °C for3 h. Post-dialysis samples were centrifuged (12,000g at 4 °C for10 min), the extract supernatants from all conditions were aliquotedinto 1.5 ml tubes, flash frozen in liquid nitrogen and stored at −80 °Cfor use in cell-free reactions. The protein concentration of cell extractswas measured using a Bradford Assay (Biorad, CA, USA).

2.4. Cell-free transcription-translation reactions

Cell-free reactions were 10 μl in total and consisted of three partsmixed together in the indicated ratios: cell extract (33% or 50% v/v),energy buffer (42% v/v) and DNA (25% or 8% v/v). The final reactionconditions were: 4–12 mM Mg-glutamate, 40–160 mM K-glutamate,1.5 mM each amino acid (except leucine −1.25 mM leucine) [RTSAmino Acid Sampler, 5-Prime, DE], 50 mM HEPES, 1.5 mM ATP andGTP, 0.9 mM CTP and UTP, 0–0.2 mg/ml E. coli tRNA, 0.26 mM CoA,0.33 mM NAD, 0.75 mM cAMP, 0.068 mM folinic acid, 1 mM spermi-dine, 2% (w/v) PEG-8000, 30 mM 3-PGA and 0–12 nM plasmid DNA.10 μl cell-free reactions were aliquoted into individual wells of 384-wellplates (Griener bio-one, NC, USA) and measured using a Clariostarplate reader (BMG, UK) with the following settings: excitation 483 nmand emission 530–30 nm. Plates were sealed, shaken prior to eachreading cycle (500 rpm) and the plate reader was set to incubate thecell-free reactions at 30 °C.

2.5. In vitro cell-free transcription-translation promoter librarycharacterisation

Cell-free transcription-translation reactions were 10 μl in total andconsisted of three parts mixed together in the indicated ratios: cellextract (33% v/v), optimised energy buffer (42% v/v) and DNA (25% v/v). The final reaction conditions were: 8 mM Mg-glutamate,160 mM K-glutamate, 1.5 mM each amino acid (except leucine−1.25 mM leucine), 50 mM HEPES, 1.5 mM ATP and GTP, 0.9 mMCTP and UTP, 0.2 mg/ml E. coli tRNA, 0.26 mM CoA, 0.33 mM NAD,0.75 mM cAMP, 0.068 mM folinic acid, 1 mM spermidine, 2% PEG-8000, 30 mM 3-PGA and 10 nM plasmid DNA. 10 μl cell-free reactionswere aliquoted into individual wells of 384-well plates (Griener bio-one), measured using a Clariostar plate reader (BMG) with thefollowing settings; Excitation 483 nm and Emission 530–30 nm.Plates were sealed, shaken prior to each reading cycle (500 rpm) andthe plate reader was set to incubate the cell-free reactions at 30 °C. Therelative strength of the promoters was calculated from the rate offluorescence increase during a phase of increasing GFPmut3b expres-sion (20–80 min). The background fluorescence of cell-free reactionsusing the control plasmid (pHT01-ΔlacI), were subtracted and thesedata were normalised to the relative strength of prWK-WT which wasdenoted a relative strength of 1.

2.6. In vivo promoter library characterisation

Promoter library constructs selected for in vivo characterisation,along with the pHT01-ΔlacI empty vector control, were transformedinto B. subtilis WB800N using the two-step transformation procedure

as described previously (Cutting and Vander Horn, 1990) and trans-formants were selected on LB agar containing the appropriate anti-biotics. This resulted in strains WB800N pHT01-ΔlacI (AJW25),WB800N pWK-WT (WK1), WB800N pWK-Δbox (WK2), WB800NpWK1 (WK3), WB800N pWK28 (WK4), WB800N pWK76 (WK5),WB800N pWK104 (WK6), WB800N pWK105 (WK7), WB800NpWK118 (WK8), WB800N pWK120 (WK9), WB800N pWK301(WK10), WB800N pWK319 (WK11), WB800N pWK603 (WK12) andWB800N pWK609 (WK13).

2.6.1. Plate reader characterisationPromoter library strains were revived from glycerol stocks onto LB

plates supplemented with the appropriate antibiotics and incubated for48 h at 30 °C. Individual colonies were inoculated into 5 ml 2x YTPmedium with appropriate antibiotics and incubated overnight withshaking (180 rpm) at 30 °C. The overnight cultures were diluted to anOD600 nm of 0.05 in fresh 2x YTP with appropriate antibiotics and100 μl aliquots loaded onto a 96-well black plate with clear flat bottoms(Greiner Bio-one, At; Cat #655076). Absorbance (600 nm) and fluor-escence (Excitation 483 nm, Emission 530–30 nm) was measuredevery ten minutes at 30 °C, with shaking at 700 rpm between eachmeasurement in a BMG Clariostar plate reader (BMG, UK). Each strainwas analysed using 3 independent cultures, with each culture beingtested in triplicate. The relative strength of the promoters wascalculated as the rate of fluorescence (GFPmut3b) per cell growth(OD600 nm) increase during a set time period (240–300 min). Thebackground fluorescence of 2x YTP cell growth media and cellstransformed with the negative control plasmid (pHT01-ΔlacI) weresubtracted and these data were normalised to the relative strength ofpWK-WT which was denoted an RPU of 1.

2.6.2. Flow cytometry characterisationPromoter library strains were revived from glycerol stocks onto LB

plates supplemented with the appropriate antibiotics and incubated for48 h at 30 °C. Three individual colonies were selected for each strain,separately inoculated into 5 ml 2x YTP medium with appropriateantibiotics and incubated overnight with shaking (180 rpm) at 30 °C.Overnight cultures were diluted to an OD600 nm of 1.0. 1 ml of dilutedcell cultures were centrifuged (12,470g) and washed twice with 1 mlPhosphate Buffered Saline (1X PBS). Finally, cell pellets were re-suspended into 1 ml PBS, then diluted (1:1000) into PBS before beingloaded onto an Attune NxT (ThermoFisher Scientific, MA, USA) flowcytometer. The fluorescence (Geometric mean BL1-A; Ex. 488 nm, Em.530/30) of at least 30,000 cells per sample were measured and thesedata were analysed using FlowJo (vX 10.1r5) software. The backgroundfluorescence (BL1-A) of cells transformed with the negative controlplasmid (pHT01-ΔlacI) were subtracted and these data were normal-ised to the relative strength of pWK-WT which was denoted a relativestrength of 1.

2.7. GFPmut3b expression and purification

A culture of glycerol stocked E. coli containing plasmid pROEXHTb His-gfpmut3b (pRK2) was used to inoculate 5 ml LB supplemen-ted with 100 µg/ml ampicillin. The culture was grown with shaking(180 rpm) for 16 h at 37 °C. Subsequently, the culture was diluted(1:500) into 500 ml LB supplemented with 100 µg/ml ampicillin, and1 mM IPTG and grown with shaking (180 rpm) for 24 h at 37 °C. Thecell pellet was harvested through centrifugation at 3,220g for 15 minand stored at −80 °C. Pellets were defrosted on ice and re-suspendedinto 1 ml re-suspension buffer (50 mM Na2HPO4, 100 mM NaCltitrated to pH 8 with HCl/NaOH) per gram of cells. Re-suspendedcells were sonicated (3×40 s with 1-min cooling interval; outputfrequency: 20 kHz, Amplitude: 50%) and centrifuged (2000g, 10 min,4 °C) to produce a clarified extract. His–GFPmut3b was purified fromthe clarified extract using a Ni-NTA column with wash buffer (50 mM

R. Kelwick et al. Metabolic Engineering xx (xxxx) xxxx–xxxx

3

Na2HPO4, 100 mM NaCl, 25 mM Imidazole, titrated to pH 8 with HCl/NaOH) and elution buffer (50 mM Na2HPO4, 100 mM NaCl, 500 mMImidazole, titrated to pH 8 with HCl/NaOH). The eluted His–GFPmut3b purified fraction was dialysed using dialysis tubing(MWCO 12–14,000 kDa) suspended in 1 L dialysis buffer (20 mMHEPES, 100 mM NaCl, dH2O pH 8 with KOH) overnight with stirring,in the dark, at 4 °C. The protein concentration of the purified His–GFPmut3b was determined using a Nanodrop ND-1000 spectrophot-ometer (Thermo Scientific). GFP purification samples were analysedvia sodium dodecyl sulphate (SDS)-polyacrylamide gel electrophoresis(PAGE) using 4–12% Bis-Tris gels (NuPAGE Novex, Lifetech), followedby either coomassie blue staining or western blot analysis with HRP-conjugated GFP-specific polyclonal antibody (1:4,000 dilution;#A10260, Thermo Fisher Scientific Ltd, UK). Western blots weredeveloped by enhanced chemiluminence (ECL).

2.8. GFPmut3b calibration curve

GFPmut3b calibration curve reactions (10 μl) were setup similarlyto typical cell-free reactions, but without DNA, and consisted of threeparts mixed together in the indicated ratios: cell extract (33% v/v),buffer (42% v/v) and purified His-GFPmut3b (25% v/v). TheGFPmut3b calibration solutions contained: 8 mM Mg-glutamate,160 mM K-glutamate, 1.5 mM each amino acid (except leucine−1.25 mM leucine), 50 mM HEPES, 1.5 mM ATP and GTP, 0.9 mMCTP and UTP, 0–0.2 mg/ml E. coli tRNA, 0.26 mM CoA, 0.33 mMNAD, 0.75 mM cAMP, 0.068 mM folinic acid, 1 mM spermidine, 2%PEG-8000, 30 mM 3-PGA and 0–1.09 µM His–GFPmut3b. The cali-bration curve reactions were aliquoted into 384-well plates (GreinerBio-One, At) and measured using a Clariostar plate reader (BMG, UK)with the following settings: Excitation 483 nm, Emission 530–30 nm.

2.9. Luciferase assay

Cell-free transcription-translation reactions were 10 μl in total andconsisted of three parts mixed together in the indicated ratios: cellextract (33% v/v), optimised energy buffer (42% v/v) and DNA (25% v/v). Cell-free reactions were setup to include either 10 nM (finalconcentration) of pHT01-ΔlacI or pHT01-ΔlacI-Renilla plasmid con-structs. Cell-free reactions were incubated for 3 h at 30 °C. Post-incubation, cell-free reactions were transferred into white assay plates(Greiner Bio-One 96-well half-area) and assayed, according to manu-facturer's guidelines, for detection of luciferase activity using acommercially available Renilla luciferase assay kit (Promega, WI,USA; Cat# E2810). Bioluminescence was measured using a Clariostarplate reader (BMG, UK) with the following settings: bioluminescenceemission was measured at 480–80 nm and the plate reader was set toincubate the cell-free reactions at 30 °C.

3. Results and discussion

3.1. Cell-free workflow

The primary focus of this study was to develop a robust B. subtiliscell-free transcription-translation system that could be applied towardsapplications that are of interest to synthetic biologists and metabolicengineers such as cell-free protein synthesis, or regulatory elementprototyping. To this end, our initial aim was to implement a standar-dised workflow of extract preparation and reaction optimisations thatcould be tailored towards the minimisation of cell extract batchvariation and improvement of cell-free protein production yields. Theworkflow was developed for B. subtilis, though it incorporates aspectsof several different established cell-free transcription-translation pro-tocols (Shin and Noireaux, 2012; Shrestha et al., 2012; Sun et al.,2013) and consists of three phases – harvest cells, extract preparationand cell-free reaction optimisation (Fig. 1)..

During the cell growth phase B. subtilis strains were revived fromglycerol stocks onto LB plates and incubated for 48 h at 30 °C. Fromthese, individually selected colonies were used to inoculate 5 mlcultures (2× YTP media) that were grown for 10 h with shaking at30 °C. These cultures were then diluted (1:500) into 50 ml 2× YTPmedia and grown for 10 h with shaking at 30 °C, after which the cellswere harvested during a late log phase of cell growth (Fig. S1a).Alternatively, if larger batches were required then cultures could bediluted once more (1:500) into 500 ml 2× YTP media and grown for afinal 10 h with shaking at 30 °C. For practical purposes, cells wereharvested at the 50 ml growth stage since this produced enough cellextract for downstream characterisation experiments. These growthconditions were sufficient such that the final harvest OD600 nm of bothB. subtilis strains −168 (3.133 ± 0.033) and WB800N (2.722 ± 0.072)were consistent between batches of the same strain (Fig. S1b). Cellpellets were harvested through centrifugation, washed and stored at−80 °C until the extract preparation phase.

During the extract preparation phase cell pellets were defrostedslowly on ice, re-suspended in 1 ml S30-A buffer per gram of cell pelletand each millilitre of cell and buffer mixture aliquoted into separate1.5 ml microtubes. These samples were sonicated and then centrifugedto produce a clarified cellular extract. In order to produce cellularextracts that have optimal cell-free transcription-translation activity,clarified cellular extracts were aliquoted into distinct downstreamprocessing groups. Cellular extracts were pre-incubated at 37 °C withshaking (180 rpm) for either 0, 30 or 80 min. Similarly, to reports of E.

Fig. 1. Cell-free workflow. A graphical depiction of the workflow used to develop andcharacterise B. subtilis cell-free systems.

R. Kelwick et al. Metabolic Engineering xx (xxxx) xxxx–xxxx

4

coli cell-free extract preparation methods, the pre-incubation tempera-ture influences the performance of cellular extracts in downstream cell-free transcription-translation reactions (Sushmita et al., 2015). Weobserved that B. subtilis WB800N extracts prepared with a pre-incubation temperature of 30 °C displayed a reduction in cell-freetranscription-translation activity in comparison to cell extracts pre-pared with a pre-incubation temperature of 37 °C (Fig. S2). Therefore,we typically pre-incubated cell extracts at 37 °C.

Upon completion of the pre-incubation step the cell extracts wereprocessed either with or without dialysis treatment at 4 °C, withstirring, for 3 h. This resulted in the generation of six differentlyprocessed cellular extracts from each cell batch. The total proteincontent of these cellular extracts was analysed using Bradford assays.The total protein concentrations of B. subtilis 168 cellular extractsvaried between cell batches and between the different extract proces-sing methods, particularly for those extracts that were pre-incubated(Fig. S1c). In contrast, the total protein concentrations of B. subtilisWB800N cellular extracts were largely consistent across batches andprocessing methods (Fig. S1c). Processed cellular extracts were flashfrozen in liquid nitrogen and stored at −80 °C until used.

Processed extracts were assessed in terms of their cell-free tran-scription-translation reaction activities using a previously described(Sun et al., 2013) standard energy buffer (Table S3), and 10 nM of

plasmid DNA. In this study, we chose to use the B. subtilis plasmidpHT01 as the backbone for our constructs in cell-free reactions since ithas been previously validated as a stable expression vector for theproduction of recombinant proteins (Nguyen et al., 2007). Thefluorescent reporter GFPmut3b was cloned into this plasmid to createthe construct pHT01-gfpmut3b, such that the expression of gfpmut3bwould be under the control of the lacI repressible Pgrac (σA)promoter. Addition of Isopropyl β-D-1-thiogalactopyranoside (IPTG)for inhibition of LacI repression enables the inducible expression ofGFPmut3b (Fig. 2a; Fig. 5b). However, when pHT01-gfpmut3b wastested in B. subtilis cell-free reactions, GFPmut3b production occurredregardless of IPTG induction (Fig. S3a). As such, the pHT01-gfpmut3bplasmid effectively resulted in the constitutive expression of gfpmut3b.Since the LacI repressor proteins are not present in the B. subtilisWB800N cell extract and are instead constitutively expressed from thelacI gene that is encoded into pHT01, it is likely that there areinsufficient LacI repressor proteins during the early stages of the cell-free reaction. The unnecessary repressor lacI gene was removed, usingPCR, from plasmids pHT01 and pHT01-gfpmut3b to create a negativecontrol plasmid (pHT01-ΔlacI) and a constitutive gfpmut3b expressionplasmid (pWK-WT) (Fig. 2a; Fig. S3b; Fig. S4) that were subsequentlyused to test and compare cell-free reaction activities..

During the final phase of the workflow the most productive cellular

Fig. 2. Characterisation of a Bacillus subtilis 168 cell-free transcription-translation system: (a) Schematic of the constructs used to characterise B. subtilis cell free systems. Circuitswere visualised using Pigeon (Bhatia and Densmore, 2013). (b) Endpoints (5 h) of cell-free reactions using cell-extract batches prepared as indicated - including 0, 30 or 80 min pre-incubation at 37 °C, followed either with or without dialysis treatment. The background fluorescence of cell-free reactions using the negative control plasmid (pHT01-ΔlacI), weresubtracted. (c) Example time-course cell-free reactions using cell extracts that were prepared as indicated - including 0, 30 or 80 min pre-incubation at 37 °C, followed either with orwithout dialysis treatment. Cell-free reactions contained either the negative control plasmid – pHT01-ΔlacI (indicated black) or pWK-WT (indicated green). Error bars denote standarderror of the mean.

R. Kelwick et al. Metabolic Engineering xx (xxxx) xxxx–xxxx

5

extracts were improved through additional cell-free reaction optimisa-tion steps. Emphasis was placed on changing the concentrations ofmagnesium glutamate and potassium glutamate in the energy buffersince these have previously been shown to have a significant influenceon cell-free reaction activity (Cai et al., 2015; Sun et al., 2013). Uponcompletion of the workflow the extract preparation method and energybuffer composition that resulted in the greatest yield of GFPmut3bproduction could then be used for all subsequent batches. We initiallyused the workflow to characterise a B. subtilis 168 cell-free system.

3.2. Characterisation of a Bacillus subtilis 168 cell-free system

B. subtilis 168 is an established and highly characterised strainwhose genomic heritage spans several decades to the extent that itsorigins can be traced back to some of the earliest isolated legacy strains(Burkholder and Giles, 1947; Zeigler et al., 2008). B. subtilis 168 is adomesticated strain and as such, it is relatively easy to culture andgenetically engineer (Guan et al., 2016; Zeigler et al., 2008). Thesecharacteristics have made B. subtilis 168 a suitable choice for a broadarray of industrial biotechnology applications and more recently as asuitable host for synthetic biology (Harwood et al., 2013). Theuniversal utility of B. subtilis 168 suggests that the development of aB. subtilis 168 cell-free transcription-translation system would be auseful platform for synthetic biology and metabolic engineeringapplications.

In order to develop a B. subtilis 168 cell-free system, threeindependently generated batches of B. subtilis 168 were cultured,harvested and the resultant cell extracts processed using the cell-freeworkflow described in Fig. 1. These extracts were combined with thestandard energy buffer and 10 nM (final concentration) of either thenegative control plasmid (pHT01-ΔlacI) or the constitutive GFPmut3bexpression plasmid (pWK-WT) to form cell-free reactions for testing.Replicate reactions were aliquoted into a 384 well plate and measuredin parallel. Cell-free production of GFPmut3b was measured every tenminutes for five hours at 30 °C using a Clariostar plate reader, withshaking before each measurement cycle to support oxygenation andmixing of the cell-free reactions. However, cell-free reactions using B.subtilis 168 cell extracts showed relatively little transcription and/ortranslation activity (Fig. 2b; Fig. 2c). Indeed, end-point (5 h) analysis ofGFPmut3b production across all cell-extract batches that were pre-pared as indicated - including 0, 30 or 80 min pre-incubation (37 °C),followed either with, or without dialysis treatment, produced low andunreliable yields of GFPmut3b (Fig. 2b). GFPmut3b yields, calculatedusing a GFPmut3b calibration curve (Fig. S5c), ranged from effectively0–0.011 µM ± 0.003 (pre-incubation for 0 min, without dialysis). Acomplete analysis of these data is shown in Table S4. Whilst alternativeextract processing methods or energy buffers may improve cell-freeactivity, it is possible that despite the advantages of B. subtilis 168 invivo, the strain is not intrinsically suitable for use in cell-freetranscription-translation reactions. This could be due to the presenceof endogenously expressed proteases and a resultant degradation oftranslated proteins. Indeed, previously reported B. subtilis cell-freesystems were typically encumbered by a requirement to includeprotease inhibitors (Table 1). Rather than to optimise B. subtilis 168cell-free activity through the addition of a cocktail of proteaseinhibitors, we decided to circumvent these limitations through thedevelopment of a B. subtilis cell-free system that uses cellular extractsfrom a protease deficient strain - B. subtilis WB800N (Nguyen et al.,2011).

3.3. The development of a B. subtilis WB800N cell-free system

B. subtilis WB800N (MoBiTech, GmbH) is a commercially acces-sible strain that has been developed for the production and secretion ofheterologous proteins (Nguyen et al., 2011). B. subtilis WB800N wasengineered to be neomycin resistant and deficient for the expression of T

able

1Com

parison

ofB.su

btilis

cell-freetran

scription

-translationsystem

s.

This

study

Legau

lt-D

emarean

dCham

bliss(197

4)

Leven

thal

andCham

bliss

(197

9)Nes

andEklund(198

3)

Oka

motoet

al.(198

5)

Zag

hloulan

dDoi

(198

7)

Strain

(s)

WB80

0N16

8T+

–ATCC66

331A

292

168

Reaction

type

Cou

pledtran

scription

-tran

slation

mRNA-directed

Cou

pledtran

scription

-tran

slation

mRNA-directed

Cou

pledtran

scription

-translation

Cou

pledtran

scription

-translation

Energ

yso

urc

e(s)

3-PGA,glutamate

Phosphoe

nolpyruva

te,pyruva

tekinase

–Phosphoe

nolpyruva

te,pyruva

tekinase

Phosphoe

nolpyruva

te,pyruva

tekinase

Phosphoe

nolpyruva

te,pyruva

tekinase

Tre

atm

ent

Not

requ

ired

Proteaseinhibitors

DNAse

treatm

entof

ribo

somes

2-mercaptoethan

olExo

genou

sribo

somes,protease

inhibitors

Exo

genou

sribo

somes,protease

inhibitors

Reaction

yield

~0.8µM

GFPmut3b

Incorporationof

826pmoles

[14C]leu

cine

orphen

ylalan

ine

Incorporationof

300pmol

methionine

48nmol/l

polyp

hen

ylalan

ine

Incorporationof

197pmoles

[14C]-

leucine

Incorporationof

20cp

m/p

mol

ofmethionine

Reaction

dura

tion

150min

30min

30–60

min

15min

30min

15–60

min–

–den

otes

that

theinform

ationwas

not

accessible

tous.

R. Kelwick et al. Metabolic Engineering xx (xxxx) xxxx–xxxx

6

several proteases (nprE aprE epr bpr mpr::ble nprB::bsr ΔvprwprA::hyg cm::neo; NeoR) (Fig. 3a). Similarly, to the generation ofB. subtilis 168 extracts, three independently generated batches of B.subtilis WB800N cells were cultured, harvested and the resultant

extracts were processed using the cell-free workflow described inFig. 1. Cell-free reactions using B. subtilis WB800N cell-extracts weremost active during the first 0–150 min and were generally more active,in terms of GFPmut3b production, than B. subtilis 168 cell-extracts

Fig. 3. Development of a Bacillus subtilis WB800N cell-free transcription-translation system. (a) B. subtilis WB800N is an engineered strain in which the indicated proteases (X) havebeen knocked-out. Adapted from (Westers et al., 2004). (b) Endpoints (5 h) of cell-free reactions using cell extract batches prepared as indicated - including 0, 30 or 80 min pre-incubation at 37 °C, followed either with or without dialysis treatment. The background fluorescence of cell-free reactions using the negative control plasmid (pHT01-ΔlacI), weresubtracted. (c) Representative time-courses of cell-free reactions using cell-extracts that were prepared as indicated - including 0, 30 or 80 min pre-incubation at 37 °C, followed eitherwith or without dialysis treatment. Cell-free reactions contained either the negative control plasmid – pHT01-ΔlacI (indicated black) or pWK-WT (indicated green). (d) Optimisation ofcell-free buffer components: magnesium glutamate and potassium glutamate. These data are representative of endpoint (5 h) analysis of cell-free reactions from three independentlyprepared extracts. The background fluorescence of cell-free reactions using the negative control plasmid (pHT01-ΔlacI), were subtracted. (e) Endpoints (5 h) of optimised cell-freereactions that include a range of different plasmid DNA (pWK-WT) concentrations. The background fluorescence of cell-free reactions using the negative control plasmid (pHT01-ΔlacI),were subtracted. Error bars denote standard error of the mean.

R. Kelwick et al. Metabolic Engineering xx (xxxx) xxxx–xxxx

7

(Fig. 3b; Fig. 3c). Based upon an analysis of end-point (5 h) GFPmut3bcell-free reaction yields, WB800N cell-extracts (Fig. 3b) producedhigher and more consistent yields of GFPmut3b than B. subtilis 168cell-extracts (Fig. 2b). WB800N cell-free GFPmut3b production yields,calculated using a GFPmut3b calibration curve (Fig. S5c), ranged from0.041 µM ± 0.008 (pre-incubation for 80 min, without dialysis) to0.116 µM ± 0.009 (pre-incubation at 37 °C for 80 min, followed bydialysis) (Table S4)..

In continuation of the cell-free workflow (Fig. 1) additional experi-ments were undertaken to further improve B. subtilis WB800N cell-free transcription-translation reactions through changes in the compo-sition of the energy buffer. The concentrations of magnesium glutamate(Mg-glutamate) and potassium glutamate (K-glutamate) in the stan-dard energy buffer were altered since these have been previously shownto have a significant influence on cell-free reaction activity. Previousreports have demonstrated that glutamate is utilised in the TCA cycle ofcellular extracts and is therefore able to serve as an energy substratethat supports ATP generation and protein production (Jewett et al.,2008). The processing conditions that included 80 min pre-incubationat 37 °C and dialysis produced the most active cellular extracts andtherefore, these conditions were used for the generation of threeadditional B. subtilis WB800N extract batches. These processed cellextracts were combined with 10 nM (final concentration) of eitherpHT01-ΔlacI (control) or pWK-WT plasmid DNA and energy buffersthat were similar in composition to the standard energy buffer, exceptthat they contained different combinations of K-glutamate(40−160 mM) and Mg-glutamate (4 mM−12 mM) concentrations. Incomparison to the standard energy buffer, which enabled the cell-freeproduction of 0.281 µM ± 0.019 GFPmut3b, an optimised energybuffer that included 160 mM K-glutamate and 8 mM Mg-glutamateresulted in an improved GFPmut3b production yield of 0.500 µM ±0.052 (Fig. 3d; Table S3; Table S5).

An additional batch of B. subtilis WB800N cell extract wasprocessed using our improved protocol. This new batch of cell extractwas used to setup several cell-free reactions that when tested with arange of different pWK-WT plasmid DNA concentrations produced upto 0.848 µM ± 0.047 of GFPmut3b (Fig. 3e). In these experiments,there was generally a linear relationship between DNA concentration(pWK-WT) and cell-free GFPmut3b production however, when thefinal DNA concentration was increased above 8 nM this linearity didnot continue. Essentially, further increases in DNA concentrationabove 8 nM did not result in significant increases in GFPmut3bproduction. It is possible that relatively high concentrations of plasmidDNA pose a maximal usage of available transcription (e.g. RNApolymerases) and/or translation (e.g. ribosomes) machinery in the cellextract that is limiting any further improvements in cell-free reactionperformance (Ceroni et al., 2015). As a preliminary test, we exploredwhether an increase in the proportion of cellular extract from 33% (v/v) to 50% (v/v) of the total cell-free reaction volume would improvecell-free transcription-translation activity. As expected, increasing thevolume of cellular extract (to 50% v/v) and as a likely consequence,increasing the availability of transcription and/or translation machin-ery, resulted in a greater level of cell-free reaction activity (Fig. S6). Inseparate experiments, we observed that the addition of E. coli tRNAsimproved the activity of B. subtilis WB800N cell-free transcription-translation reactions but were not essential (Fig. S7). Thus, whilst wecannot discount the possibility that further improvements in B. subtilisWB800N cell-free activity might be hampered by the availability of corecellular machinery (e.g. active ribosomes) in the cell-free extract, thesedata suggest that at least some of the translational resources (e.g.tRNAs) may not be a significant limiting factor.

Future studies may also need to consider additional factors thatmight be impacting B. subtilis WB800N cell-free reaction activity. Forinstance, there are reports that changes in pH as a result of the build-up of glycolytic lactate and acetate production, as well as the produc-tion of inorganic phosphates during cell-free reactions can also be

deleterious to cell-free activity (Caschera and Noireaux, 2015b, 2014).These limiting factors may be mitigated through changes in the energysource that is used. For example, Maltose is a novel energy source thatenables the recycling of inhibitory inorganic phosphates in E. coli cell-free systems (Caschera and Noireaux, 2014). Therefore, in futurestudies, it may be possible to further improve B. subtilis cell-freeactivity through an investigation of these potentially inhibitory factorsand to assess whether changes in the energy buffer composition mightmitigate any limiting effects.

3.4. Generation of an engineered promoter library

In order to improve the ability of synthetic biologists and metabolicengineers to precisely fine tune gene expression there is an increasinginterest towards expanding the availability of characterised geneticregulatory elements (e.g. promoters) (Guiziou et al., 2016; Moore et al.,2016; Patron et al., 2015). As demonstration of the applicability of B.subtilis cell-free transcription-translation systems for regulatory ele-ment prototyping we applied our improved B. subtilis WB800N cell-free system towards the characterisation of an engineered B. subtilispromoter library that enables a range of transcriptional activities invitro (cell-free) and in vivo. The engineered promoter library wascreated using degenerate oligonucleotides that were designed to bind tothe −35 and −10 boxes of the Pgrac (σA) promoter in plasmid pWK-WT, and introduce, through an inverted PCR, changes into thepromoter sequence (Fig. 4a). This strategy should result in enoughrandom changes to both boxes (−10 and −35) to give a theoreticallibrary size of up to 412 (1,6777,216) different engineered promoters.Several library clones were screened via sequencing and those withsequence changes that differed from the reference wildtype Pgracpromoter, which we renamed as prWK-WT, were retained for furtherstudies. The promoter library sequences obtained are shown in Fig. S8.The plasmid - pWK-Δbox was also engineered as an additional negativecontrol in which primers were designed to remove, via PCR fromplasmid pWK-WT, both the −35 and −10 boxes of the Pgrac (σA)promoter, as well as the sequence between them. Thus, plasmid pWK-Δbox did not promote gfpmut3b expression (Fig. 4b; Fig. S9; Fig. S10b;Fig. S11; Table S7; Table S8)..

An initial screen of cell-free GFPmut3b production from 58different engineered promoter constructs was carried out using theimproved B. subtilis WB800N cell-free system. Our initial screeningidentified several promoters that displayed a range of differenttranscriptional strengths. Engineered promoters that resulted in thehighest levels of gfpmut3b expression were selected for further analysis(Fig. S9). Of these, clones: pWK1, pWK28, pWK76, pWK104, pWK105,pWK118, pWK120, pWK301, pWK319, pWK603, and pWK609, werecharacterised for both in vitro (cell-free transcription-translation) andin vivo gfpmut3b expression.

Similarly to previous reports (Chappell et al., 2013; Kelly et al.,2009) we assessed the relative strength of these engineered promotersusing cell-free transcription-translation (in vitro), plate reader (invivo) and flow cytometry (in vivo) assays (Fig. 4b). These assays aredescribed in more detail in the materials and methods section. Briefly,for cell-free transcription-translation characterisation of the promoterlibrary the relative strength of the promoters was calculated from therate of fluorescence increase during a phase of increasing GFPmut3bexpression (20–80 min) and then these rate change data were normal-ised (including removal of the background signal) to the relativestrength of prWK-WT. Whilst, for in vivo analysis, promoter librarystrains (B. subtilis WB800N) were cultured in 96-well plates andassayed using a Clariostar plate reader (Fig. S10). The relative strengthof the promoters was calculated as the rate of fluorescence (GFPmut3b)increase per cell growth (OD600 nm) during a set time period (240–300 min). and then these rate change data were normalised (includingremoval of the background signal) to the relative strength of prWK-WT.Finally, as an additional assessment, the promoter library strains (B.

R. Kelwick et al. Metabolic Engineering xx (xxxx) xxxx–xxxx

8

subtilis WB800N) were cultured overnight, washed in PBS and loadedinto an Attune NxT flow cytometer for analysis of GFPmut3b fluores-cence (Geometric mean BL1-A) (Table S8.). These data were subse-quently normalised (including removal of the background signal) to therelative strength of prWK-WT.

We observed significant differences in activity between promotersthat were largely similar in sequence. For example, the sequences ofprWK104 and prWK1 differ by only two base changes (−35 box: T > C;−10 box: G > A) yet, on relative terms, prWK104 is roughly four timesstronger than prWK1. It is likely, that these base changes areinfluencing the biophysical interaction (e.g. binding affinity) betweenthe promoter region on the plasmid DNA and the transcriptionalmachinery (e.g. RNA polymerase and sigma factor) (Guiziou et al.,2016). Interestingly, the general pattern of relative promoter strengthsin the engineered library were similar both in vitro (cell-free) and invivo (Fig. 4b; Table S7). Comparability between the relative activity ofgenetic regulatory elements tested in cell-free transcription-translationreactions and in vivo has been previously demonstrated in E. coli(Chappell et al., 2013; Sun et al., 2014). Comparability between cell-free and in vivo characterised DNA regulatory elements is importantwithin a systematic design context – through which, it is envisionedthat cell-free workflows may rapidly provide the characterisation data

required to rationally select a smaller number of designs for finaltesting in vivo. Therefore, whilst further investigations are needed toassess the in vivo comparability of in vitro characterised B. subtilisgenetic regulatory elements, characterisation of our engineered pro-moter library suggests that B. subtilis cell-free systems are a viableplatform for regulatory element prototyping.

3.5. Characterisation of a B. subtilis WB800N-pHT01 cell extract forinducible gene expression

Genetic circuit engineering lies at the foundations of syntheticbiology (Collins et al., 2000; Elowitz and Leibler, 2000) and there is agrowing interest in utilising cell-free transcription-translation systemsto iterate increasingly more complex genetic circuit designs(Niederholtmeyer et al., 2015; Noireaux et al., 2003; Zhang et al.,2016). Therefore, it is likely that as the availability of characterised B.subtilis regulatory elements increases there will be an increase in thepotential for using B. subtilis cell-free transcription-translation systemsto support the development of B. subtilis-based genetic circuits (Jeonget al., 2015). Consequently, as an initial step towards the emergence ofgenetic circuit prototyping in B. subtilis cell-free systems we charac-terised an inducible expression system. The protease deficient strain -B. subtilis WB800N was previously transformed with plasmid pHT01to create the engineered strain - WB800N-pHT01 (AJW5) whichconstitutively expresses the lacI gene (Fig. 5a) (Webb et al., 2016).Three independent batches of B. subtilis WB800N-pHT01 cell extractswere generated using the same method that improved the cell-freeactivity of B. subtilis WB800N cell extracts (pre-incubation at 37 °C for80 min, followed by dialysis). As expected, the presence of LacIrepressor protein within B. subtilis WB800N-pHT01 cell extracts wassufficient to inhibit (turn “OFF”) GFPmut3b expression from pHT01-based plasmids in cell-free transcription-translation reactions (Fig. 5b;Fig. 5c; Fig. S12). Likewise, addition of 0.0625–1 mM IPTG forinhibition of LacI repression resulted in a robust induction (turn“ON”) of GFPmut3b expression from pHT01-gfpmut3b plasmid –

producing a fluorescent signal that was 117–147 fold above back-ground (Fig. 5b; Fig. 5c). Whilst, there was a modest induction increasebetween the lowest (0.0625 mM) and highest (0.5–1 mM) concentra-tions of IPTG that were tested, it may be possible to achieve a largerdynamic range of induction levels if a broader range of IPTGconcentrations are used. Nevertheless, these inducible expression datasuggest that it may be possible, in future studies, to use these B. subtiliscell-free systems to characterise simple genetic circuits..

3.6. Characterisation of Renilla luciferase activity in B. subtilisWB800N cell-free reactions

The luciferase gene from Renilla reniformis (sea pansy) encodes adecarboxylating enzyme that is capable of catalyzing the oxidation ofCoelenterazine to Coelenteramide. As a consequence, cabon dioxide(CO2) and bioluminescence (480 nm) are produced (Lorenz et al.,1991; Matthews et al., 1977). The bioluminescence capabilities ofRenilla luciferase have been extensively characterised and repurposedfor use in an array of contexts and biological assays (Hampf andGossen, 2006). We carried out cell-free protein expression andcharacterisation of Renilla luciferase activity in B. subtilis WB800Ncell-free extracts. Cell-free reactions including either 10 nM (finalconcentration) of pHT01-ΔlacI or pHT01-ΔlacI-renilla plasmid DNAwere incubated at 30 °C for three hours prior to the addition of theluciferase assay buffer and substrate. Bioluminescence (480–80 nm)was detected in cell-free reactions that expressed Renilla luciferasefrom plasmid pHT01-ΔlacI-renilla and as expected control reactions(with plasmid pHT01-ΔlacI) produced negligible bioluminescence(Fig. 6b, Fig. 6c). These experiments were carried out in order todemonstrate the potential applicability of B. subtilis cell-free transcrip-tion-translation reactions for cell-free protein production and in situ

Fig. 4. Characterisation of an engineered Bacillus subtilis promoter library. (a)Schematic displaying the sequences of engineered promoters that were derived fromPCR reactions involving degenerate primers that target the −10 and −35 boxes within thewildtype Pgrac (σA) promoter (prWK-WT) (b) For in vitro cell-free characterisation, therelative strength of the promoters was calculated from the rate of fluorescence increaseduring a phase of increasing GFPmut3b expression (20–80 min). The backgroundfluorescence of cell-free reactions using the negative control plasmid (pHT01-ΔlacI),were subtracted and these data were normalised to the relative strength of prWK-WTwhich was denoted a relative strength of 1. For in vivo plate reader characterisation, therelative strength of the promoters was calculated as the rate of fluorescence (GFPmut3b)per cell growth (OD600 nm) increase during a set time period (240–300 min). Thebackground fluorescence of 2x YTP cell growth media and cells transformed with thenegative control plasmid (pHT01-ΔlacI) were subtracted and these data were normalisedto the relative strength of pWK-WT which was denoted a relative strength of 1. For invivo flow cytometry characterisation, the fluorescence (Geometric mean BL1-A; Ex.488 nm, Em. 530/30) was measured using an Attune NxT flow cytometer and wereanalysed using FlowJo (vX 10.1r5) software. The background fluorescence of cellstransformed with the negative control plasmid (pHT01-ΔlacI) were subtracted and thesedata were normalised to the relative strength of pWK-WT which was denoted a relativestrength of 1. Error bars denote standard error of the mean.

R. Kelwick et al. Metabolic Engineering xx (xxxx) xxxx–xxxx

9

characterisation of enzyme performance. tableIt is likely that as cell-free protein synthesis driven metabolic engineering approaches devel-op the parallel characterisation of enzymes, that are encoded withinDNA libraries, may enable the rapid optimisation of biosyntheticpathways (Dudley et al., 2015; Karim and Jewett, 2016)..

4. Conclusions

B. subtilis is an established model organism of broad importance tomicrobiology, synthetic biology and industrial biotechnology.Therefore, the development of B. subtilis cell-free transcription-trans-lation systems are desirable since, much like existing cell-free tran-scription-translation systems, they could be applied to several applica-tions. Yet, despite their potential B. subtilis cell-free transcription-translation systems have been largely neglected. Previously reported B.subtilis cell-free transcription-translation systems were typically en-cumbered by a range of different factors that significantly hindered theaccessibility of the methods used to generate them and their capabil-ities (e.g. poor reaction dynamics and low cell-free protein productionyields) (Table 1).

Herein, we report on the development of several B. subtilis cell-freesystems that unlike previously reported methods do not require thepreparation or use of exogenous mRNAs, ribosomes, DNAse orprotease inhibitors. Consequently, we report that our method is muchmore accessible and easier to carry out in a shorter time frame -typically several batches of cell extract can be generated in just a fewdays. Additionally, in contrast to previous reports, we describe the useand improvement of a relatively more efficient energy regenerationsystem, based on 3-phosphoglycerate (3-PGA) and optimised concen-trations of magnesium and potassium glutamate, that is now typicallyused in E. coli cell-free transcription-translation reactions. In combi-nation these improvements have enabled the development of animproved B. subtilis WB800N system that is capable of robust cell-free transcription-translation reactions that last for several hours andcan produce up to 0.8 µM GFPmut3b, which is a ~72× fold-improve-ment when compared with a B. subtilis 168 cell-free transcription-translation system (0.011 µM GFPmut3b). However, additional im-provements are needed to increase B. subtilis cell-free activity tocomparable levels of recently published E. coli cell-free transcription-translation systems (production of over 40 µM of reporter protein)(Caschera and Noireaux, 2015b, 2014). Conversely, an examination ofdevelopments in E. coli or other cell-free transcription systems mayprovide insights into how further improvements in B. subtilis cell-freeactivity may be achieved. In particular, an examination of alternativeenergy sources (e.g. maltose, maltodextrin or hexametaphosphate), co-factor regeneration systems and an analysis of the availability of thecellular machinery (e.g. ribosomes) in cell extract batches may be themost significant priorities for future studies.

Whilst, further improvements in B. subtilis cell-free activity aredesirable we demonstrate the applicability of our reported B. subtilisWB800N cell-free transcription-translation system towards regulatoryelement prototyping. Essentially, we engineered several promoters,derived from the wild-type Pgrac (σA) promoter, that display a rangeof comparable in vitro and in vivo transcriptional activities. Thus, weanticipate that these cell-free systems will be useful to applications

Fig. 5. Characterisation of a Bacillus subtilis WB800N-pHT01 cell-free transcription-translation system for inducible gene expression. (a) Schematic depicts the generation ofB. subtilis WB800N-pHT01 cell extracts for inducible gene expression (b) PlasmidpHT01-gfpmut3b enables IPTG inducible gfpmut3b expression within B. subtilisWB800N-pHT01-based cell-free transcription-translation reactions. (c) Endpoint (5 h)characterisation of IPTG induced gene expression from plasmid pHT01-gfpmut3b(10 nM final concentration) in B. subtilis WB800N-pHT01 cell-free transcription-translation reactions. The background fluorescence of cell-free reactions using thenegative control plasmid (pHT01), and the indicated range of IPTG were subtracted.Error bars denote standard error of the mean. Student t-test, *P < 0.05.

Fig. 6. Characterisation of Renilla luciferase in B. subtilis WB800N cell-free reactions.(a) Schematic depicts the cell-free expression and enzymatic activity of Renilla luciferasewithin B. subtilis WB800N cell-free transcription-translation reactions. Cell-free reac-tions including either 10 nM (final concentration) of pHT01-ΔlacI or pHT01-ΔlacI-renilla were incubated at 30 °C for three hours prior to the addition of the luciferaseassay buffer and substrate. (b) Time course analysis of Renilla luciferase activity withinCell-fee reactions. Luminesence (480–80 nm) was measured, post-2-second delay, for10 s using a Clariostar plate reader. (c) Integrated analysis was calculated as thesummation of luminescence, throughout the 10 s measurement cycle, for each sample.Error bars denote standard error of the mean.

R. Kelwick et al. Metabolic Engineering xx (xxxx) xxxx–xxxx

10

such as the engineering of in vivo metabolic pathways or geneticcircuits that require characterised regulatory elements (e.g. promoters)in order to rationally fine tune gene expression. Additionally, as a steptowards future applications for B. subtilis cell-free systems, wedescribed the characterisation of an inducible expression system (aprecursor to genetic circuit prototyping) and we also characterised theactivity of Renilla luciferase (a model enzyme). More broadly, weenvision that these improved B. subtilis cell-free transcription-transla-tion systems will spur a renewal in efforts to continue the developmentof these systems for the benefit of an array of synthetic biology andmetabolic engineering applications.

Acknowledgements

We wish to acknowledge the support of the Engineering andPhysical Science Research Council (EPSRC) – [EP/K034359/1; EP/J02175X/1] and that of our colleagues in the Centre for SyntheticBiology and Innovation (CSynBI) at Imperial College. We would alsolike to thank colleagues from The Flowers Consortium, particularlyProfessor Colin Harwood, for their support and advice. We also thankProfessor Angelika Gründling for the kind gift of Bacillus subtilis 168.

Appendix A. Supporting information

Supplementary data associated with this article can be found in theonline version at doi:10.1016/j.ymben.2016.09.008.

References

Bhatia, S., Densmore, D., 2013. Pigeon: A Design Visualizer for Synthetic Biology. ACSSynth. Biol. 2, 348–350. http://dx.doi.org/10.1021/sb400024s.

Burkholder, P.R., Giles, N.H., 1947. Induced biochemical mutations in Bacillus subtilis.Am. J. Bot. 34, 345–348.

Cai, Q., Hanson, J. a, Steiner, A.R., Tran, C., Masikat, M.R., Chen, R., Zawada, J.F., Sato,A.K., Hallam, T.J., Yin, G., 2015. A simplified and robust protocol forimmunoglobulin expression in E scherichia coli cell-free protein synthesis systems.Biotechnol. Prog. 31, 823–831. http://dx.doi.org/10.1002/btpr.2082.

Caschera, F., Noireaux, V., 2015a. Preparation of amino acid mixtures for cell-freeexpression systems. Biotechniques. http://dx.doi.org/10.2144/000114249.

Caschera, F., Noireaux, V., 2015b. A cost-effective polyphosphate-based metabolism fuelsan all E. coli cell-free expression system. Metab. Eng. 27, 29–37. http://dx.doi.org/10.1016/j.ymben.2014.10.007.

Caschera, F., Noireaux, V., 2014. Synthesis of 2.3 mg/ml of protein with an allEscherichia coli cell-free transcription-translation system. Biochimie 99, 162–168.http://dx.doi.org/10.1016/j.biochi.2013.11.025.

Ceroni, F., Algar, R., Stan, G.-B., Ellis, T., 2015. Quantifying cellular capacity identifiesgene expression designs with reduced burden. Nat. Methods 12, 415–418. http://dx.doi.org/10.1038/nmeth.3339.

Chappell, J., Jensen, K., Freemont, P.S., 2013. Validation of an entirely in vitro approachfor rapid prototyping of DNA regulatory elements for synthetic biology. Nucleic AcidsRes 41, 1–11. http://dx.doi.org/10.1093/nar/gkt052.

Collins, J.J., Gardner, T.S., Cantor, C.R., 2000. Construction of a genetic toggle switch inEscherichia coli. Nature 403, 339–342. http://dx.doi.org/10.1038/35002131.

Cutting, S., Vander Horn, P.B., 1990. Genetic analysis. In: Harwood, C.R., Cutting, S.(Eds.), Molecular Biological Methods for Bacillus. John Wiley and Sons, Chichester,United Kingdom, 27–74. http://dx.doi.org/10.1016/S0232-4393(11)80231-3.

Dudley, Q.M., Karim, A.S., Jewett, M.C., 2015. Cell-free metabolic engineering:Biomanufacturing beyond the cell. Biotechnol. J. http://dx.doi.org/10.1002/biot.201400330.

Elowitz, M.B., Leibler, S., 2000. A synthetic oscillatory network of transcriptionalregulators. Nature 403, 335–338. http://dx.doi.org/10.1038/35002125.

Gagoski, D., Polinkovsky, M.E., Mureev, S., Kunert, A., Johnston, W., Gambin, Y.,Alexandrov, K., 2016. Performance benchmarking of four cell-free proteinexpression systems. Biotechnol. Bioeng. 113, 292–300. http://dx.doi.org/10.1002/bit.25814.

Gan, R., Jewett, M.C., 2014. A combined cell-free transcription-translation system fromSaccharomyces cerevisiae for rapid and robust protein synthe. Biotechnol. J. 9,641–651. http://dx.doi.org/10.1002/biot.201300545.

Garamella, J., Marshall, R., Rustad, M., Noireaux, V., 2016. The all E. coli TX-TL toolbox2.0: A platform for cell-free synthetic biology. ACS Synth. Biol. 5, 344–355. http://dx.doi.org/10.1021/acssynbio.5b00296.

Guan, C., Cui, W., Cheng, J., Zhou, L., Liu, Z., Zhou, Z., 2016. Development of an efficientautoinducible expression system by promoter engineering in Bacillus subtilis.Microb. Cell Fact. 15, 66. http://dx.doi.org/10.1186/s12934-016-0464-0.

Guiziou, S., Sauveplane, V., Chang, H.-J., Cler, E., C., Declerck, N., Jules, M., Bonnet, J.,2016. A part toolbox to tune genetic expression in Bacillus subtilis. Nucleic AcidsRes, 1–14. http://dx.doi.org/10.1093/nar/gkw624.

Hampf, M., Gossen, M., 2006. A protocol for combined Photinus and Renilla luciferasequantification compatible with protein assays. Anal. Biochem. 356, 94–99. http://dx.doi.org/10.1016/j.ab.2006.04.046.

Harwood, C.R., 1992. Bacillus subtilis and its relatives: molecular biological andindustrial workhorses. Trends Biotechnol 10, 247–256.

Harwood, C.R., Pohl, S., Smith, W., Wipat, A., 2013. Bacillus subtilis. Model gram-positive synthetic biology chassis. Methods Microbiol 40, 87–117. http://dx.doi.org/10.1016/B978-0-12-417029-2.00004-2.

Hodgman, C.E., Jewett, M.C., 2012. Cell-free synthetic biology: Thinking outside the cell.Metab. Eng. 14, 261–269. http://dx.doi.org/10.1016/j.ymben.2011.09.002.

Jeong, D.-E., Park, S.-H., Pan, J.-G., Kim, E.-J., Choi, S.-K., 2015. Genome engineeringusing a synthetic gene circuit in Bacillus subtilis. Nucleic Acids Res 43. http://dx.doi.org/10.1093/nar/gku1380.

Jewett, M.C., Calhoun, K.A., Voloshin, A., Wuu, J.J., Swartz, J.R., 2008. An integratedcell-free metabolic platform for protein production and synthetic biology. Mol. Syst.Biol. 4. http://dx.doi.org/10.1038/msb.2008.57.

Karim, A.S., Jewett, M.C., 2016. A cell-free framework for rapid biosynthetic pathwayprototyping and enzyme discovery. Metab. Eng. 36, 116–126. http://dx.doi.org/10.1016/j.ymben.2016.03.002.

Kelly, J.R., Rubin, A.J., Davis, J.H., Ajo-Franklin, C.M., Cumbers, J., Czar, M.J., de Mora,K., Glieberman, A.L., Monie, D.D., Endy, D., 2009. Measuring the activity of BioBrickpromoters using an in vivo reference standard. J. Biol. Eng. 3, 4. http://dx.doi.org/10.1186/1754-1611-3-4.

Kelwick, R., MacDonald, J.T., Webb, A.J., Freemont, P., 2014. Developments in the toolsand methodologies of synthetic biology. Front. Bioeng. Biotechnol. 2. http://dx.doi.org/10.3389/fbioe.2014.00060.

Legault-Demare, L., Chambliss, G.H., 1974. Natural messenger ribonucleic acid-directedcell-free protein-synthesizing system of Bacillus subtilis. J. Bacteriol. 120,1300–1307.

Leventhal, J.M., Chambliss, G.H., 1979. DNA-directed cell-free protein-synthesizingsystem of Bacillus subtilis. Biochim. Biophys. Acta 564, 162–171.

Lorenz, W.W., McCann, R.O., Longiaru, M., Cormier, M.J., 1991. Isolation andexpression of a cDNA encoding Renilla reniformis luciferase. Proc. Natl. Acad. Sci. U.S. A 88, 4438–4442. http://dx.doi.org/10.1073/pnas.88.10.4438.

Matthews, J.C., Hori, K., Cormier, M.J., 1977. Purification and properties of Renillareniformis luciferase. Biochemistry 16, 85–91. http://dx.doi.org/10.1021/bi00620a014.

Moore, S.J., Lai, H.-E., Kelwick, R.J.R., Chee, S.M., Bell, D.J., Polizzi, K.M., Freemont,P.S., 2016. EcoFlex: A Multifunctional MoClo kit for E. coli synthetic biology. ACSSynth. Biol. acssynbio. http://dx.doi.org/10.1021/acssynbio.6b00031.

Nes, I.F., Eklund, T., 1983. The effect of parabens on DNA, RNA and protein synthesis inEscherichia coli and Bacillus subtilis. J. Appl. Bacteriol. 54, 237–242.

Nguyen, H., Phan, T., Schumann, W., 2011. Analysis and application of Bacillus subtilissortases to anchor recombinant proteins on the cell wall. AMB Express 1, 22. http://dx.doi.org/10.1186/2191-0855-1-22.

Nguyen, H.D., Phan, T.T.P., Schumann, W., 2007. Expression vectors for the rapidpurification of recombinant proteins in Bacillus subtilis. Curr. Microbiol. 55, 89–93.http://dx.doi.org/10.1007/s00284-006-0419-5.

Niederholtmeyer, H., Sun, Z.Z., Hori, Y., Yeung, E., Verpoorte, A., Murray, R.M., Maerkl,S.J., 2015. Rapid cell-free forward engineering of novel genetic ring oscillators. Elife4, 1–5. http://dx.doi.org/10.7554/eLife.09771.

Nirenberg, M., 2004. Historical review: Deciphering the genetic code – a personalaccount. Trends Biochem. Sci. 29, 46–54. http://dx.doi.org/10.1016/j.tibs.2003.11.009.

Noireaux, V., Bar-Ziv, R., Libchaber, A., 2003. Principles of cell-free genetic circuitassembly. Proc. Natl. Acad. Sci. 100, 12672–12677. http://dx.doi.org/10.1073/pnas.2135496100.

Ogawa, A., Namba, Y., Gakumasawa, M., 2016. Rational optimization of ambersuppressor tRNAs toward efficient incorporation of a non-natural amino acid intoprotein in a eukaryotic wheat germ extract. Org. Biomol. Chem. 14, 2671–2678.http://dx.doi.org/10.1039/c5ob02533h.

Okamoto, M., Fukui, S., Kobayashi, Y., 1985. In vitro expression of plasmid pUB110DNA with Bacillus subtilis cell-free extracts. Agric. Biol. Chem. 49, 1077–1082.http://dx.doi.org/10.1080/00021369.1985.10866864.

Pardee, K., Green, A.A., Ferrante, T., Cameron, D.E., DaleyKeyser, A., Yin, P., Collins,J.J., 2014. Paper-based synthetic gene networks. Cell 159, 940–954. http://dx.doi.org/10.1016/j.cell.2014.10.004.

Patron, N.J., Orzaez, D., Marillonnet, S., Warzecha, H., Matthewman, C., Youles, M.,Raitskin, O., Leveau, A., Farré, G., Rogers, C., Smith, A., Hibberd, J., Webb, A.A.R.,Locke, J., Schornack, S., Ajioka, J., Baulcombe, D.C., Zipfel, C., Kamoun, S., Jones,J.D.G., Kuhn, H., Robatzek, S., Van Esse, H.P., Sanders, D., Oldroyd, G., Martin, C.,Field, R., O’Connor, S., Fox, S., Wulff, B., Miller, B., Breakspear, A., Radhakrishnan,G., Delaux, P.-M., Loqué, D., Granell, A., Tissier, A., Shih, P., Brutnell, T.P., Quick,W.P., Rischer, H., Fraser, P.D., Aharoni, A., Raines, C., South, P.F., Ané, J.-M.,Hamberger, B.R., Langdale, J., Stougaard, J., Bouwmeester, H., Udvardi, M.,Murray, J.A.H., Ntoukakis, V., Schäfer, P., Denby, K., Edwards, K.J., Osbourn, A.,Haseloff, J., 2015. Standards for plant synthetic biology: a common syntax forexchange of DNA parts. New Phytol 208, 13–19. http://dx.doi.org/10.1111/nph.13532.

Pohl, S., Bhavsar, G., Hulme, J., Bloor, A.E., Misirli, G., Leckenby, M.W., Radford, D.S.,Smith, W., Wipat, A., Williamson, E.D., Harwood, C.R., Cranenburgh, R.M., 2013.Proteomic analysis of Bacillus subtilis strains engineered for improved production ofheterologous proteins. Proteomics 13, 3298–3308. http://dx.doi.org/10.1002/pmic.201300183.

Shimizu, Y., Kanamori, T., Ueda, T., 2005. Protein synthesis by pure translation systems.Methods 36, 299–304. http://dx.doi.org/10.1016/j.ymeth.2005.04.006.

R. Kelwick et al. Metabolic Engineering xx (xxxx) xxxx–xxxx

11

Shin, J., Noireaux, V., 2012. An E. coli cell-free expression toolbox: Application tosynthetic gene circuits and artificial cells. ACS Synth. Biol. 1, 29–41. http://dx.doi.org/10.1021/sb200016s.

Shrestha, P., Holland, T.M., Bundy, B.C., 2012. Streamlined extract preparation forEscherichia coli-based cell-free protein synthesis by sonication or bead vortexmixing. Biotechniques 53, 163–174. http://dx.doi.org/10.2144/0000113924.

Siegal-Gaskins, D., Tuza, Z.A., Kim, J., Noireaux, V., Murray, R.M., 2014. Gene circuitperformance characterization and resource usage in a cell-free “breadboard.”. ACSSynth. Biol. 3, 416–425. http://dx.doi.org/10.1021/sb400203p.

Sullivan, C.J., Pendleton, E.D., Sasmor, H.H., Hicks, W.L., Farnum, J.B., Muto, M.,Amendt, E.M., Schoborg, J.A., Martin, R.W., Clark, L.G., Anderson, M.J.,Choudhury, A., Fior, R., Lo, Y.-H., Griffey, R.H., Chappell, S.A., Jewett, M.C., Mauro,V.P., Dresios, J., 2016. A cell-free expression and purification process for rapidproduction of protein biologics. Biotechnol. J 11, 238–248. http://dx.doi.org/10.1002/biot.201500214.

Sun, Z.Z., Hayes, C.A., Shin, J., Caschera, F., Murray, R.M., Noireaux, V., 2013. Protocolsfor implementing an Escherichia coli based TX-TL cell-free expression system forsynthetic biology. J. Vis. Exp., e50762. http://dx.doi.org/10.3791/50762.

Sun, Z.Z., Yeung, E., Hayes, C.A., Noireaux, V., Murray, R.M., 2014. Linear DNA forRapid Prototyping of Synthetic Biological Circuits in an Escherichia coli Based TX-TL Cell-Free System. ACS Synth. Biol. 3, 387–397. http://dx.doi.org/10.1021/sb400131a.

Sushmita, M.R., Sunil, B., Evan, G., Henry, H., 2015. Method for enhancing recombinantprotein production by cell-free protein expression system US9040253 (B2).

Takahashi, M.K., Hayes, C. a, Chappell, J., Sun, Z.Z., Murray, R.M., Noireaux, V., Lucks,

J.B., 2015. Characterizing and prototyping genetic networks with cell-freetranscription–translation reactions. Methods 86, 60–72. http://dx.doi.org/10.1016/j.ymeth.2015.05.020.

Tuza, Z.a., Singhal, V., Jongmin Kim, Murray, R.M., 2013. An in silico modeling toolboxfor rapid prototyping of circuits in a biomolecular “breadboard” system. in: I52ndIEEE Conference on Decision and Control. IEEE, pp. 1404–1410. doi: ⟨10.1109/CDC.2013.6760079⟩

Webb, A.J., Kelwick, R., Doenhoff, M.J., Kylilis, N., MacDonald, J.T., Wen, K.Y.,McKeown, C., Baldwin, G., Ellis, T., Jensen, K., Freemont, P.S., 2016. A protease-based biosensor for the detection of schistosome cercariae. Sci. Rep. 6, 24725.http://dx.doi.org/10.1038/srep24725.

Westers, L., Westers, H., Quax, W.J., 2004. Bacillus subtilis as cell factory forpharmaceutical proteins: a biotechnological approach to optimize the host organism.Biochim. Biophys. Acta - Mol. Cell Res 1694, 299–310. http://dx.doi.org/10.1016/j.bbamcr.2004.02.011.

Zaghloul, T.I., Doi, R.H., 1987. In vitro expression of a Tn9-derived chloramphenicolacetyltransferase gene fusion by using a Bacillus subtilis system. J. Bacteriol. 169,1212–1216.

Zeigler, D.R., Pragai, Z., Rodriguez, S., Chevreux, B., Muffler, A., Albert, T., Bai, R., Wyss,M., Perkins, J.B., 2008. The origins of 168, W23, and other Bacillus subtilis legacystrains. J. Bacteriol. 190, 6983–6995. http://dx.doi.org/10.1128/JB.00722-08.

Zhang, C., Tsoi, R., You, L., 2016. Addressing biological uncertainties in engineering genecircuits. Integr. Biol. 8, 456–464. http://dx.doi.org/10.1039/C5IB00275C.

Zubay, G., 1973. In vitro synthesis of protein in microbial systems. Annu. Rev. Genet. 7,267–287. http://dx.doi.org/10.1146/annurev.ge.07.120173.001411.

R. Kelwick et al. Metabolic Engineering xx (xxxx) xxxx–xxxx

12