Engineering an E. coli Near-Infrared Light SensorNicholas T. Ong,† Evan J. Olson,† and Jeffrey J. Tabor*,†,‡

†Department of Bioengineering, ‡Department of Biosciences, Rice University, 6100 Main Street, Houston, Texas 77005, UnitedStates

*S Supporting Information

ABSTRACT: Optogenetics is a technology wherein researcherscombine light and genetically engineered photoreceptors to controlbiological processes with unrivaled precision. Near-infrared (NIR)wavelengths (>700 nm) are desirable optogenetic inputs due to theirlow phototoxicity and spectral isolation from most photoproteins.The bacteriophytochrome photoreceptor 1 (BphP1), found in severalpurple photosynthetic bacteria, senses NIR light and activatestranscription of photosystem promoters by binding to and inhibitingthe transcriptional repressor PpsR2. Here, we examine the responseof a library of output promoters to increasing levels ofRhodopseudomonas palustris PpsR2 expression, and we identify thatof Bradyrhizobium sp. BTAi1 crtE as the most strongly repressed in Escherichia coli. Next, we optimize Rps. palustris bphP1 andppsR2 expression in a strain engineered to produce the required chromophore biliverdin IXα in order to demonstrate NIR-activated transcription. Unlike a previously engineered bacterial NIR photoreceptor, our system does not require production of asecond messenger, and it exhibits rapid response dynamics. It is also the most red-shifted bacterial optogenetic tool yet reportedby approximately 50 nm. Accordingly, our BphP1−PpsR2 system has numerous applications in bacterial optogenetics.

KEYWORDS: synthetic biology, optogenetics, near-infrared light, bacteriophytochrome

In previous work, we and others have engineered bacterialphotoreceptors that regulate transcription in response to

ultraviolet (maximal wavelength (λmax) = 382−405 nm),1 blue(λmax = 450 nm),2−4 green (λmax = 535 nm),5−8 red (λmax = 650nm),6,9 and short wavelength near-infrared (NIR; λmax = 712nm)10 light. Several of these photoreceptors have been used tospatially pattern gene expression across macroscopic5,9,11 andmicroscopic2 cell populations, engineer a bacterial edgedetector,12 induce permanent genetic memory via recombinaseexpression,13 program tailor-made expression dynamics ofone1,14 and two independent15 proteins, place proteinexpression under in silico feedback control,16 and reveal novelinput/output dynamics of a widely used synthetic geneticcircuit.14

Bacteriophytochromes (BphPs) are the most red-shiftedbiological photoreceptors known. BphPs contain an N-terminalphotosensory core domain (PCD) and C-terminal effectordomain that regulates the biological response. The PCD iscomposed of a Per/Arnt/Sim (PAS), cGMP phosphodiester-ase/adenyl cyclase/FhlA (GAF), phytochrome (PHY) motifwith the chromophore biliverdin IXα (BV) covalently boundnear the N-terminus.17 Canonical BphPs adopt an inactiveground (i.e., dark-adapted) conformation with maximalabsorption between 690 and 710 nm.18,19 This ground stateis referred to as Pr, for phytochrome red absorbing state, byliterature convention. Upon exposure to red or shortwavelength NIR light, these BphPs switch to the biologicallyactive Pfr (for phytochrome far-red absorbing state, byconvention) conformation, with maximal absorption between

750 and 760 nm. Canonical BphPs revert from Pfr to Pr inmilliseconds after exposure to NIR light20 or in minutes tohours by thermal reversion in the dark.17,21

A subfamily of BphPs, referred to as “bathy” variants, exhibitan inverted photocycle, switching between the inactive Pfrground and active Pr states, respectively.

22 The dark reversionprocess is also inverted: bathy BphPs revert from Pr to Pfr on atime scale of minutes to hours.23,24

Ryu and Gomelsky recently utilized a canonical BphP toengineer the most red-shifted bacterial optogenetic toolreported to date.10 In particular, these authors fused the PCDof Rhodobacter sphaeroides BphG1 (Pr λmax = 712 nm; Pfr λmax =756 nm) to the diguanylate cyclase effector domain ofSynechocystis PCC6803 Slr1143 in Escherichia coli engineeredto produce BV. When activated by red or short wavelengthNIR, the BphG1−Slr1143 chimera (named BphS) synthesizescyclic di-GMP (c-di-GMP), a bacterial second messenger.MrkH, a Klebsiella pneumoniae transcription factor, binds this c-di-GMP, resulting in transcription from the PmrkA outputpromoter with a reported 40-fold dynamic range (ratio ofreporter gene levels in the activated versus inactive state).Despite its utility, BphS−MrkH has several limitations. First,

heterologous production of c-di-GMP has been shown to alterthe expression of approximately 200 E. coli genes, includingthose involved in the central processes of sugar metabolism andcell-wall modification.25 c-di-GMP also controls biofilm

Received: August 14, 2017Published: November 1, 2017

Research Article

pubs.acs.org/synthbio

© 2017 American Chemical Society 240 DOI: 10.1021/acssynbio.7b00289ACS Synth. Biol. 2018, 7, 240−248

Figure 1. Engineering PpsR2-repressible E. coli promoters. (a) Annotated sequences of 10 candidate promoters. (b) Device schematic for the PpsR2induction and promoter activity measurement plasmid system. (c) Promoter transfer functions. Dashed lines indicate the maximal cellularfluorescence levels. Data points represent the total fluorescence levels of cells carrying the promoter−reporter plasmid and pNO15. Data are fromexperiments on three separate days. Error bars represent the standard deviation (SD) of the mean. Solid lines represent the deactivating form of the

Hill function, = − ·+f I b( ) a I

I K

n

n n1/2, fit to each data set, where I is the concentration of IPTG (mM), b and a are the fit parameters for the initial and

range of sfGFP (MEFL) output values, respectively, K1/2 represents the concentration of IPTG (mM) needed to reach half-maximal repression ofthe promoters, and n is the Hill coefficient. Fit parameter values are listed in Table S2. E. coli cell autofluorescence is denoted by the gray area.

ACS Synthetic Biology Research Article

DOI: 10.1021/acssynbio.7b00289ACS Synth. Biol. 2018, 7, 240−248

241

formation,26 cell division,27 the production of extracellular curliprotein fibers,28 polynucleotide phosphorylase activity,29 andflagellar rotation bias.30 Second, bacterial genomes encodemany poorly characterized c-di-GMP synthases and phospho-diesterases (e.g., E. coli encodes 12 and 13, respectively) thatcould cross-activate or cross-repress the PmrkA output promoterin unwanted or unknown ways.31 Third, BphS−MrkH requires1.5 h after exposure to activating light to produce an observablegene expression response and 6 h to reach half-maximalactivation (t1/2). Most model bacterial strains double in 0.3−1 hunder laboratory conditions, and many regulatory processesoccur on these time scales as well.32 Thus, these slow dynamicsmake BphS−MrkH a poor choice for characterizing thedynamics of many pathways.33 Finally, BphS is likely to bespectrally incompatible with red light responsive photo-receptors as well as red-shifted fluorescent proteins. Forexample, Ryu and Gomelsky activated BphS−MrkH with 660nm light emitting diodes (LEDs),10 which strongly repress thewidely used E. coli green light sensor CcaSR and red lightsensor Cph8−OmpR (λmax = 650 nm).15 Additionally, imaging

of the commonly used infrared fluorescent protein IFP1.4(excitation maximum = 684 nm)34 would likely cross-activateBphS−MrkH.Rhodopseudomonas palustris CGA009 BphP1 is a bathy BphP

with a more red-shifted ground state absorption spectrum (λmax= 760 nm) than that of any previously reported bacterialoptogenetic tool.35 BphP1 possesses a C-terminal HOS (2-helixoutput sensor) effector domain that is inactive in the Pfr state,but it binds and inactivates the transcriptional repressor PpsR2in the Pr conformation.35 PpsR2 binds to TGT-N12-ACAoperator sites that often overlap the −35/−10 regions of σ70-type promoters that drive the expression of genes involved inphotosynthesis.36,37

To engineer BphP1−PpsR2 to function as an optogenetictool, we first developed a strong E. coli promoter that is highlyrepressed by PpsR2 expression. Next, we optimized BphP1 andPpsR2 expression and BV production to achieve the desiredlight response. Finally, we performed kinetic and spectralexperiments to demonstrate that our BphP1−PpsR2 systemresponds to changes in light input in minutes and exhibits the

Figure 2. Engineering BphP1−PpsR2 to function as an E. coli photoreceptor. (a) Device schematic for the chemically inducible BphP1−PpsR2system. (b) Dependence of system response on BphP1 and PpsR2 expression. (c) Device schematic for the hardwired BphP1−PpsR2 system. (d)Response of hardwired system and negative controls to NIR, R, and dark. WT: wild-type. bphP1ΔHOS: deletion of BphP1 HOS domain.ppsR2ΔDBD: deletion of PpsR2 DNA binding domain. Δho1: deletion of BV production plasmid. Bar heights represent the mean of experimentsperformed on three separate days. Error bars represent the SD of the mean. Reported sfGFP values are corrected for cell autofluorescence.Statistically significant differences are annotated as such: *, p < 0.05; **, p < 0.01 (two-way ANOVA followed by Tukey’s multiple comparison test).(e) Representative flow cytometry histograms of the data in panel D.

ACS Synthetic Biology Research Article

DOI: 10.1021/acssynbio.7b00289ACS Synth. Biol. 2018, 7, 240−248

242

most red-shifted action spectrum of any bacterial optogenetictool.

■ RESULTS AND DISCUSSIONEngineering PpsR2-Repressible E. coli Promoters. To

engineer a BphP1−PpsR2 output promoter in E. coli, weconsidered the intergenic regions upstream of bchC, crtD, andcrtE. We chose these promoters because they have beenvalidated to be directly bound by PpsR2 in DNA footprintingexperiments.38 We examined sequences from both Rps. palustrisand the closely related Bradyrhizobium BTAi1, which has ahomologous BphP1−PpsR2 pair.39 Using the BPROM webt-ool,40 we identified a single putative E. coli promoter with onePpsR2 operator overlapping the −35 site and a secondoverlapping the −10 site in all six intergenic regions (FigureS1). BTAi1 bchC contains a third operator upstream of the−35, and Rps. palustris crtD and crtE each possess oneadditional semiconserved operator (TGA-N12-ACA) near thetranscriptional start site (+1). Each +1 site is separated from thestart codon of the downstream ORF by 30−50 bp 5′untranslated regions (UTRs) (Figure S1).Next, we redesigned each of these promoter regions to

function reliably in E. coli. First, to reduce the possibility ofunwanted regulation by alternative pathways, we truncated allDNA greater than 10 bp upstream of the −35 or most distalPpsR2 operator (Figure S1). Additionally, we deleted all DNAdownstream of each terminal promoter element (i.e., +1, orthird operator site) to eliminate any possible regulatory impactfrom the 5′ UTRs. We named the resulting minimizedpromoters PRp_bchC, PRp_crtD, PRp_crtE, PBr_bchC, PBr_crtD, andPBr_crtE to indicate the organism and gene from which theyare derived. Because the third operator sites of PBr_bchC, PRp_crtD,and PRp_crtE do not overlap predicted E. coli −35/−10 elementsand may thus contribute less to transcriptional repression in ourexperimental system, we also designed versions of each whereinthese operators are deleted (PBr_bchC2, PRp_crtD2, and PRp_crtE2)(Figure S1). Finally, we designed a synthetic PpsR2-repressibleE. coli promoter by replacing the cI operator sites in the strongbacteriophage λ promoter PL

41 with PpsR2 operator sites. Thispromoter, which we named PL_ppsR2, is based on the design ofthe strongly repressible engineered tetracycline repressor(TetR)-repressed promoter PLTetO‑1

42 (Figure S2).To characterize the resulting library (Figure 1a), we next

constructed a series of plasmids (pNO1−10) wherein weencoded superfolder green fluorescent protein (sfgfp)43 down-stream of each candidate promoter along with a computation-ally designed synthetic E. coli ribosome binding site (sRBS)44,45

(Figures 1b and S3). Then, we transformed each plasmid intoE. coli and measured the resulting sfGFP expression levels byflow cytometry (Methods). Though PRp_crtD and PRp_crtD2 arevery weak, the remaining eight promoters are clearly functional,driving total cellular fluorescence levels between 523 ± 17 and6900 ± 230 (PBr_crtE) molecules of equivalent fluorescein(MEFL) compared to the E. coli autofluorescence level of 165± 10 MEFL (Figure 1c). PBr_bchC2 and PRp_crtE2 are weaker thantheir parental promoters, demonstrating that the regioncontaining the third operator enhances the overall transcriptionrate (Figure 1c).Next, we characterized the extent to which each promoter is

repressed by PpsR2 expression. To this end, we constructedplasmid pNO15, wherein Rps. palustris ppsR2 is transcribedunder inducible control of isopropyl β-D-1-thiogalactopyrano-side (IPTG) (Figure S3). Then, we cotransformed pNO15

pairwise with each member of the pNO1−10 set, grew theresulting strains in the presence of variable levels of IPTG, andmeasured sfGFP levels as before. As expected, transcriptionfrom each of the eight functional promoters decreases withPpsR2 induction (Figure 1c). PBr_crtE (encoded on pNO4)again exhibits the greatest response, with a 76 ± 20-folddynamic range (Figure 1c). Accordingly, we selected PBr_crtE asthe BphP1−PpsR2 output promoter for subsequent experi-ments.

Engineering BphP1−PpsR2 To Function as an E. coliPhotoreceptor. The architecture of BphP1−PpsR2 is similarto that of bacterial two-component histidine kinase signaltransduction systems (TCSs).46 We have previously shown thatTCS performance is sensitive to the expression levels of boththe sensor kinase (analogous to BphP1) and response regulator(analogous to PpsR2).1,6 To optimize BphP1−PpsR2 functionin E. coli, we therefore added a cassette wherein Rps. palustrisbphP1 is transcribed under control of an anhydrotetracyline(aTc)-inducible promoter to pNO15, resulting in pNO17(Figures 2a and S4). To engineer E. coli to produce BV, we alsoengineered plasmid pNO41, from which a heme oxygenasegene is constitutively expressed (Figures 2a and S4). Weoptimized the expression levels of a heme oxygenase gene onpNO41 to maximize BV production (Figure S5).To assay the function of this engineered system, we grew E.

coli carrying pNO4, pNO17, and pNO41 in variousconcentrations of aTc and IPTG while exposing them to either760 nm (NIR) or 660 nm (red) light (Methods). Indeed, weobserved NIR activation under a wide range of BphP1 andPpsR2 expression levels, with a maximum activation of 2.35 ±0.28-fold (Figure 2b).While convenient for preliminary system validation, the

requirement for IPTG and aTc limits the compatibility ofoptogenetic tools with many other engineered genetic systemsthat utilize these inducers or their target repressor proteins.6

Therefore, we next replaced the two inducible promoters with aconstitutive version driving a synthetic bphP1−ppsR2 operon(Figures 2c and S4). To reoptimize BphP1 and PpsR2expression levels, we designed eight sRBSs with variablepredicted strengths for each gene (Table S1). We thenconstructed 64 plasmids (pNO131−194, Figure S4) to testthe 8 × 8 matrix of sRBS strength combinations for these twogenes (Table S1). Then, we cotransformed each plasmid withpNO4 and pNO41 and characterized the NIR light response asbefore. While more than half of the designs do not appreciablyrespond to light, approximately one-third exhibit between 1.1-and 2-fold NIR activation (Figure S6), and five yield a ≥2-foldresponse (Figure S6). The best performing system is encodedon pNO183 (Figure S4) and has a dynamic range of 2.47 ±0.17-fold, similar to the best aTc and IPTG inductionconditions.To be certain that we had achieved the maximum dynamic

range, we also tested the other seven PpsR2-repressiblepromoters with pNO183 and pNO41. However, these allyield smaller dynamic ranges than PBr_crtE (Figure S7).Next, we performed a series of control experiments to

validate that the light response is due to BphP1−PpsR2signaling rather than an alternative pathway. As expected,removal of the BphP1 HOS domain results in constitutively lowsfGFP levels (Figure 2d). Similarly, deletion of the PpsR2 DBDresults in constitutively high sfGFP (Figure 2d). Furthermore,the lack of BV abolishes the light response (Figure 2d). Darktreatment also results in intermediate sfGFP output relative to

ACS Synthetic Biology Research Article

DOI: 10.1021/acssynbio.7b00289ACS Synth. Biol. 2018, 7, 240−248

243

NIR and R illumination (Figure 2d,e). These data areconsistent with in vitro results demonstrating that dark-treatedholo-BphP1 first adopts the Pr state and then gradually convertsto Pfr,

47 which should generate an equilibrium between thesetwo states in actively growing dark-treated cells. Takentogether, these results clearly demonstrate that we haverecapitulated the function of the BphP1−PpsR2 system in E.coli.Transfer Functions. The ability to use variable light

intensities or wavelength combinations to tune the outputtranscription rate enables researchers to program geneexpression levels and dynamics with high quantitative precision,an important feature of optogenetic tools.14,48 Thus, wecharacterized the BphP1−PpsR2 transfer function, or steady-state response to increasing NIR intensity, in the presence ofdifferent intensities of deactivating red light. In the absence ofred light, sfGFP output increases between 0 and 72 μmol/m2·sNIR light. The response is well-fit by a Hill function with anintensity resulting in 50% maximal response (K1/2) equal to8.63 ± 0.54 μmol/m2·s and a Hill coefficient (n) equal to 1.209± 0.091 (Figure 3). This Hill coefficient reveals that the system

exhibits a relatively linear response over a wide range of lightintensities, a desirable performance feature for tuning thetranscription rate. As expected, the presence of red lightreduces the NIR response (Figure 3).By repeating the transfer function measurement in the

presence of variable red light intensities (Figure S8), weobserved that approximately 1.2 additional NIR photons arerequired to overcome the inhibitory effect of each red photon(Figure S9). This result suggests the system has similarsensitivity to NIR and red light, a feature that should make iteasy to modulate transcriptional output under diverseexperimental conditions.Input/Output Dynamics. Next, we characterized the

step−response dynamics of our engineered BphP1−PpsR2system. Specifically, we preconditioned cells in saturating redlight and then instantaneously switched to saturating NIR (stepON), performed the converse experiment (step OFF), andswitched from saturating NIR to dark (step DARK). Then, weused our previously developed kinetic protocol based on theinhibition of transcription and maturation of sfGFP (Methods)to monitor system output over time (Figure 4).

We fit each data set to a simplified version of a previousordinary differential equation model we developed to captureTCS photoreceptor response dynamics14 (Figure S10). ThisBphP1−PpsR2 model depicts a pure delay after a change inlight input during which time the output promoter activity isconstant (τ, min) followed by a change in promoter activitythat obeys single exponential kinetics. For step ON, τ fits to 5.3± 3.0 min and the time required to reach half-maximalactivation (t1/2) fits to 26.9 (−5.4/+6.0) min (Figure 4;Methods). For step OFF, we estimate τ to be 8.6 ± 2.3 min andt1/2 as 38.2 (−4.3/+4.6) min (Figure 4). Finally, for stepDARK, τ fits to 24.9 ± 2.9 min and t1/2 fits to 63.5 (−6.5/+7.4)min (Figure 4). These response times are similar to the timescales of bacterial gene expression and to the dynamics of ourprevious green, red, and UV sensors.1,14 The slower responsefor the step DARK is also consistent with in vitro measurementsof a 50 min half-life for BphP1 dark reversion.35

Action Spectra. The action spectrum(a) of an optogenetictool depicts the relationship between input wavelength(s),intensity(ies), and system output and is crucial for selectingoptimal light conditions, especially when multiplexing withother photoreceptors.15 Therefore, we characterized theforward (i.e., Pfr to Pr) and reverse (i.e., Pr to Pfr) actionspectra of our engineered BphP1−PpsR2 system. For theforward experiment, we exposed bacteria to 0−20.0 μmol/m2·slight from each of 23 different LEDs with emission wavelengthsranging from 369 to 954 nm.15 As expected, wavelengthsbetween 740 and 782 nm induce the strongest responses,whereas those between 636 and 677 nm maximally deactivatethe system (Figures 5a and S11). Interestingly, the 782 nmresponse is quantitatively similar to that of 740 and 755 nm(Figure S12), demonstrating robust activation by this highlyred-shifted wavelength. Other wavelengths have little effect,though there is some activation in the 388−451 nm region atthe higher intensities (Figure 5a). BphP1, like otherphytochrome-family photoreceptors, has a small absorptionpeak in the UV-blue region,49 also known as the Soret band,which may be responsible for its activation at those wave-lengths. However, we are able to repress this apparent Soretband activation of BphP1−PpsR2 by applying deactivating redlight (Figure S13).For the reverse action spectrum, we applied sufficient NIR to

activate the system to an intermediate level and then performedthe same spectral experiment (Figure 5b). As expected, thesystem is maximally deactivated by 636−677 nm light (Figures

Figure 3. BphP1−PpsR2 steady-state transfer functions. Markers anderror bars represent the means and SD, respectively, of n = 3experiments conducted on separate days. Reported sfGFP values arecorrected for cell autofluorescence. Lines represent the activating form

of the Hill function, = + ·+f I b( ) a I

I K

n

n n1/2, fit to each data set. Fit

parameters for each data set can be found in Table S3.

Figure 4. Step−response dynamics. Response to step changes in lightfrom R to NIR (dark red), NIR to R (red), and NIR to dark (black).Markers and error bars represent the means and SD, respectively, of n= 3 experiments conducted on separate days. Reported sfGFP valuesare corrected for cell autofluorescence. Lines represent model fits(Figure S10). Fit parameters are listed in Table S4.

ACS Synthetic Biology Research Article

DOI: 10.1021/acssynbio.7b00289ACS Synth. Biol. 2018, 7, 240−248

244

5b and S11). Overall, both action spectrum measurements arehighly consistent with in vitro BphP1 absorption and actionspectra.35,49

■ CONCLUSIONS

Our engineered BphP1−PpsR2 system is the most red-shiftedbacterial optogenetic tool yet reported. Due to its unique actionspectrum and rapid response dynamics, BphP1−PpsR2 shouldbe able to be combined with existing UV,1 blue,2−4 and greensensors6 to achieve independent control of the expressiondynamics of multiple genes in the same cell. Such multiplexedoptogenetic programming of gene expression dynamicspromises to be a powerful new method for revealing newaspects of how information flows through gene networks.15

Additionally, BphP1−PpsR2 should be able to be controlledwhile imaging most commonly used fluorescent proteinreporters,50 including infrared fluorescent proteins,51 a benefitthat will facilitate single cell time lapse experiments.The major limitation to our system is its relatively low (∼2.5-

fold) dynamic range. However, we previously engineered an E.coli green light sensor with only 2-fold dynamic range5 that wassubsequently used to reveal novel input/output dynamics of asynthetic transcriptional inverter circuit,14 modulate growthrate via programmed expression of a methionine biosyntheticgene,52 and demonstrate that synthetic genetic circuits canadaptively filter noise.53 Thus, even without further optimiza-tion, we believe that BphP1−PpsR2 can be used for a range ofoptogenetic applications.Nonetheless, the utility of BphP1−PpsR2 should be able to

be improved by increasing its dynamic range in future work.Because the PBr_crtE output promoter has a relatively largedynamic range of nearly 80-fold (Figure 1c), we hypothesizethat this small response is due to intrinsically small changes inthe BphP1 Pr/Pfr ratio in response to NIR versus red light. Onthe other hand, BphP1−PpsR2 has recently been combinedwith TetR DNA binding and VP16 transcriptional activationdomains along with a minimal CMV output promoter

containing seven tetO operator sites to achieve ∼40-fold NIRactivation in mammalian cells.49 Thus, we believe that thedynamic range of our system could be improved by addingbacterial signal-amplifying genetic modules. Available E. colitranscriptional inverters,2,54 RNA-based transcription regula-tors,55,56 viral protease systems,57 and recombinases thatactivate strong output promoters58 have been shown to amplifytranscriptional responses by 1−3 orders of magnitude.Alternatively, directed evolution could be performed onBphP1 to increase its photoswitchability.59,60

Other NIR photoreceptors could also be explored. Lagariasand co-workers have recently discovered cyanobacteriochromesensor histidine kinases (CBCR SKs) that bind phycocyano-bilin (PCB) instead of BV, yet absorb strongly in the NIRspectrum up to λmax = 740 nm in the ground state.61 The majorchallenge is that these photoreceptors lack known partnerresponse regulators and/or output promoters that are neededto control gene expression. However, domain swapping9 couldpotentially be used to replace the PCD of our related greenlight-sensing CBCR SK CcaS6 with that from one of these NIRsensors. Ideally, the >100-fold dynamic range and rapidresponse dynamics of our CcaS-based system would bepreserved, resulting in a new NIR sensor that retains thebenefits of BphP1−PpsR2 while finding additional utility due tothe larger response.

■ METHODS

Plasmids, Strains, and Media. All plasmids wereconstructed using Golden Gate62 or Gibson63 assembly. Thegenes encoding PpsR2 and BphP1 (rpa1536 and rpa1537,respectively) were amplified from Rps. palustris CGA009genomic DNA (ATCC), and the frameshift mutation inrpa1537 was fixed via insertion of a guanine residue at position430.64 pEO100c14 and pSR346 were used as template vectorsfor building plasmids pNO1−10 and pNO41, respectively. Tobuild pNO15, we replaced sfgf p downstream of the Ptacpromoter65 on pSR11-21 with ppsR2. We replaced mCherry

Figure 5. BphP1−PpsR2 action spectra. (A) Forward action spectrum. LED centroid wavelength is shown. (B) Reverse action spectrum. Cells weretreated with 7.82 μmol/m2·s 755 nm light to activate the system to an intermediate level and simultaneously exposed to the LED centroidwavelengths shown. All values represent the mean of n = 3 experiments conducted on separate days. The 369 nm LED is incapable of reaching thebrightest intensities (gray squares). Absolute sfGFP responses are shown in Figure S11. LED details are listed in Table S5.

ACS Synthetic Biology Research Article

DOI: 10.1021/acssynbio.7b00289ACS Synth. Biol. 2018, 7, 240−248

245

downstream of the PLTetO‑1 promoter42 on pNO15 with thecorrected bphP1 gene to build pNO17. All synthetic ribosomebinding sites (sRBSs) were designed using DeNovoDNAsoftware.44,45 To build the ppsR2−bphP1 bicistronic operonplasmids pNO131−194, we computationally designed a set ofeight sRBSs with predicted strengths varying over ∼57-fold(Table S1) each for ppsR2 and bphP1 and then assembled the64 possible combinations of sRBS strengths for the 2 genesusing pSR58.66 as a template vector. The apFAB30 constitutivepromoter66 controls operon expression on pNO131−194.Cloning was carried out using NEB 10-β cells (New EnglandBiolabs, catalog no. C3019H). Primers were ordered fromIntegrated DNA Technologies, Inc., IA, and DNA sequencingservices from Genewiz or Lone Star Laboratories were used.All experiments were performed in E. coli BW29655

(BW28357 Δ(envZ-ompR)520(::FRT)).67 LB media wasused to culture transformed bacteria at 37 °C, 250 rpm, withappropriate antibiotics added (50 μg/mL ampicillin, 34 μg/mLchloramphenicol, and/or 100 μg/mL spectinomycin). Allexperiments were conducted using M9 media (1× M9 salts,0.4% w/v glucose, 0.2% w/v casamino acids, 2 mM MgSO4,100 μM CaCl2) supplemented with appropriate antibioticslisted earlier. LB cultures of transformed bacteria were grown toexponential phase (OD600 < 0.3), diluted 1:1000 in M9 media,and allowed to grow to exponential phase. By keeping the cellsgrowing in exponential phase instead of reaching stationaryphase, we minimized any potential physiological changes to thecells. The OD600 of each M9 culture was recorded prior to theaddition of filter-sterilized glycerol to a final concentration of30% (v/v); cultures were then transferred into PCR tubes in100 μL aliquots and frozen at −80 °C. All experimentsdescribed here were started from such individual aliquots ofculture frozen at exponential phase. In addition to beingconvenient, we have previously determined that this protocolreduces day-to-day variability in the outputs of engineeredgenetic systems.1,15,68−70

Optical Hardware. All optogenetic experiments wereperformed in six 24-well Light Plate Apparatus (LPA) devices70

mounted in a shaking incubator operating at 37 °C, 250 rpm,allowing for precise optical control with up to two LED inputsper culture. LEDs were calibrated using a spectrophotometer(StellarNet UVN-SR-25 LT16) following our previousmethod.15 Replicates of each experimental data point wererandomized to different well positions where possible tominimize systematic errors from well-specific effects such asuneven LED calibration. Unless otherwise stated, NIR and Rillumination used were set at 72 μmol/m2·s for 760 nm and 40μmol/m2·s for 660 nm.Experimental Protocols. All experiments were performed

according to our previous protocol,15 which we summarizehere. Frozen aliquots of engineered cells were thawed at thebench and inoculated at low densities (OD600 = 2 × 10−5 to 1 ×10−4) into fresh media. Inoculation densities were picked onthe basis that cell cultures would reach steady state and remainin exponential phase (OD600 < 0.3) by the end of theexperiment, thus ensuring the cells undergo minimalphysiological changes during the experiment. Chemicalinducers anhydrotetracycline (aTc) (Clontech cat. no.631310) and isopropyl β-D-1-thiogalactopyranoside (IPTG)(VWR cat. no. 14213-263) were added as needed. 500 μL ofeach inoculated M9 culture was then added into a well of aclear-bottomed black-walled 24-well plate (ArcticWhite AWLS-303008) that was then sealed with adhesive foil (VWR 60941-

126). Culture plates were mounted onto preconfigured LPAdevices running the selected light exposure program.70 Fordynamics experiments, cells were preconditioned for 4 h underselected light conditions and then switched to the final lightconditions using a previously described staggered-startprotocol14 over the next 4 h. The staggered-start protocolallows ease of capturing light dynamics data with precisecontrol of time points while eliminating the need for repeatedsamplings of a continuous culture. Assays were run for 5 h(PpsR2-promoter screen) or 8 h (optogenetic experiments). Atthe end of each experiment, the plates were removed from theincubator and placed in an ice−water slurry for 15 min toinhibit further cell growth. 100 μL cell sample aliquots weretransferred from each well into prechilled flow cytometry tubesholding 1 mL PBS + 500 μg/mL rifampicin, and these wereplaced in a 37 °C water bath for 1 h to allow for maturation oftranslated sfGFP while inhibiting new sfGFP transcription.They were placed back in ice−water for 15 min and thenmeasured for sfGFP fluorescence using flow cytometry.

Flow Cytometry and Data Analysis. Single cellfluorescence was measured using a BD FACScan flowcytometer as previously described.14 Calibration bead samples(Spherotech RCP-30-5A) in PBS were measured with eachday’s experiments. Following data acquisition, raw flowcytometry files were processed using FlowCal.69 Cellpopulations were gated by density, and the arbitraryfluorescence units were converted by FlowCal into stand-ardized MEFL values based on the calibration bead samples.The individual arithmetic means of all gated events werecalculated and reported as the population mean of each sample.

Model Fitting. All Hill function (deactivating form:

= − ·+f I b( ) a I

I K

n

n n1/2; activating form: = + ·

+f I b( ) a II K

n

n n1/2) and

dynamics model (Figure S10) fits were performed using anonlinear least-squares curve-fitting algorithm in Prism v7.0afor Mac OSX (GraphPad Software, Inc., La Jolla, CA).Reported t1/2 values were calculated by addition of therespective fit values of the delay parameter τ and half-life ofthe exponential parameter k. Because the four functionparameters including k and τ (Figure S10) were fit locally toeach step−response curve, k and τ fit values have symmetricuncertainties (Table S4). However, t1/2 fit values inheritasymmetrical uncertainties from logarithmic calculation of khalf-life uncertainties. Model function parameters were fitlocally to each data set of n = 3 replicates per data point unlessotherwise stated.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acssynbio.7b00289.

Promoter design, plasmid maps, functional character-ization (PDF)

■ AUTHOR INFORMATIONCorresponding Author*E-mail: jeff.tabor@rice.edu. Phone: +1 (713) 348-8316.ORCIDNicholas T. Ong: 0000-0002-6338-668XAuthor ContributionsN.T.O., E.J.O., and J.J.T conceived of the study and designedthe experiments. N.T.O. performed the experiments and

ACS Synthetic Biology Research Article

DOI: 10.1021/acssynbio.7b00289ACS Synth. Biol. 2018, 7, 240−248

246

analyzed the data. N.T.O. and J.J.T wrote the manuscript. Allauthors discussed the project and commented on themanuscript.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTS

We thank Karl Gerhardt and other Tabor lab members forassistance in configuring and calibrating optogenetic hardware.We also thank the Joel Moake lab for generously sharing theirflow cytometer. This work was supported by the Office ofNaval Research (MURI N000141310074) and a NationalScience Scholarship provided by the Agency for Science,Technology and Research (A*STAR), Singapore. Plasmids areavailable from Addgene.

■ ABBREVIATIONS

NIR, near-infrared; BphPs, bacteriophytochrome photorecep-tors; PCD, photosensory core domain; BV, biliverdin IXα;MEFL, molecules of equivalent fluorescein; TCS, two-component system; CBCR, cyanobacteriochrome; SK, sensorkinase

■ REFERENCES(1) Ramakrishnan, P., and Tabor, J. J. (2016) RepurposingSynechocystis PCC6803 UirS-UirR as a UV-Violet/Green Photo-reversible Transcriptional Regulatory Tool in E. coli. ACS Synth. Biol.5, 733−740.(2) Ohlendorf, R., Vidavski, R. R., Eldar, A., Moffat, K., and Moglich,A. (2012) From dusk till dawn: one-plasmid systems for light-regulated gene expression. J. Mol. Biol. 416, 534−542.(3) Chen, X., Liu, R., Ma, Z., Xu, X., Zhang, H., Xu, J., Ouyang, Q.,and Yang, Y. (2016) An extraordinary stringent and sensitive light-switchable gene expression system for bacterial cells. Cell Res. 26, 854−857.(4) Jayaraman, P., Devarajan, K., Chua, T. K., Zhang, H. Z.,Gunawan, E., and Poh, C. L. (2016) Blue light-mediated transcrip-tional activation and repression of gene expression in bacteria. NucleicAcids Res. 44, 6994−7005.(5) Tabor, J. J., Levskaya, A., and Voigt, C. A. (2011) Multichromaticcontrol of gene expression in Escherichia coli. J. Mol. Biol. 405, 315−324.(6) Schmidl, S. R., Sheth, R. U., Wu, A., and Tabor, J. J. (2014)Refactoring and optimization of light-switchable Escherichia coli two-component systems. ACS Synth. Biol. 3, 820−831.(7) Abe, K., Miyake, K., Nakamura, M., Kojima, K., Ferri, S.,Ikebukuro, K., and Sode, K. (2014) Engineering of a green-lightinducible gene expression system in Synechocystis sp. PCC6803.Microb. Biotechnol. 7, 177−183.(8) Sugie, Y., Hori, M., Oka, S., Ohtsuka, H., and Aiba, H. (2016)Reconstruction of a chromatic response system in Escherichia coli. J.Gen. Appl. Microbiol. 62, 140−143.(9) Levskaya, A., Chevalier, A. A., Tabor, J. J., Simpson, Z. B., Lavery,L. A., Levy, M., Davidson, E. A., Scouras, A., Ellington, A. D., Marcotte,E. M., and Voigt, C. A. (2005) Synthetic biology: engineeringEscherichia coli to see light. Nature 438, 441−442.(10) Ryu, M. H., and Gomelsky, M. (2014) Near-infrared lightresponsive synthetic c-di-GMP module for optogenetic applications.ACS Synth. Biol. 3, 802−810.(11) Fernandez-Rodriguez, J., Moser, F., Song, M., and Voigt, C. A.(2017) Engineering RGB color vision into Escherichia coli. Nat. Chem.Biol. 13, 706−708.(12) Tabor, J. J., Salis, H. M., Simpson, Z. B., Chevalier, A. A.,Levskaya, A., Marcotte, E. M., Voigt, C. A., and Ellington, A. D. (2009)A synthetic genetic edge detection program. Cell 137, 1272−1281.

(13) Farzadfard, F., and Lu, T. K. (2014) Genomically encodedanalog memory with precise in vivo DNA writing in living cellpopulations. Science 346, 1256272.(14) Olson, E. J., Hartsough, L. A., Landry, B. P., Shroff, R., andTabor, J. J. (2014) Characterizing bacterial gene circuit dynamics withoptically programmed gene expression signals. Nat. Methods 11, 449−455.(15) Olson, E. J., Tzouanas, C. N., and Tabor, J. J. (2017) Aphotoconversion model for full spectral programming and multi-plexing of optogenetic systems. Mol. Syst. Biol. 13, 926.(16) Milias-Argeitis, A., Rullan, M., Aoki, S. K., Buchmann, P., andKhammash, M. (2016) Automated optogenetic feedback control forprecise and robust regulation of gene expression and cell growth. Nat.Commun. 7, 12546.(17) Lamparter, T., Michael, N., Mittmann, F., and Esteban, B.(2002) Phytochrome from Agrobacterium tumefaciens has unusualspectral properties and reveals an N-terminal chromophore attachmentsite. Proc. Natl. Acad. Sci. U. S. A. 99, 11628−11633.(18) Bhoo, S. H., Davis, S. J., Walker, J., Karniol, B., and Vierstra, R.D. (2001) Bacteriophytochromes are photochromic histidine kinasesusing a biliverdin chromophore. Nature 414, 776−779.(19) Lamparter, T., Carrascal, M., Michael, N., Martinez, E.,Rottwinkel, G., and Abian, J. (2004) The biliverdin chromophorebinds covalently to a conserved cysteine residue in the N-terminus ofAgrobacterium phytochrome Agp1. Biochemistry 43, 3659−3669.(20) Borucki, B., von Stetten, D., Seibeck, S., Lamparter, T., Michael,N., Mroginski, M. A., Otto, H., Murgida, D. H., Heyn, M. P., andHildebrandt, P. (2005) Light-induced proton release of phytochromeis coupled to the transient deprotonation of the tetrapyrrolechromophore. J. Biol. Chem. 280, 34358−34364.(21) Burgie, E. S., Wang, T., Bussell, A. N., Walker, J. M., Li, H., andVierstra, R. D. (2014) Crystallographic and electron microscopicanalyses of a bacterial phytochrome reveal local and globalrearrangements during photoconversion. J. Biol. Chem. 289, 24573−24587.(22) Karniol, B., and Vierstra, R. D. (2003) The pair ofbacteriophytochromes from Agrobacterium tumefaciens are histidinekinases with opposing photobiological properties. Proc. Natl. Acad. Sci.U. S. A. 100, 2807−2812.(23) Rottwinkel, G., Oberpichler, I., and Lamparter, T. (2010) Bathyphytochromes in rhizobial soil bacteria. J. Bacteriol. 192, 5124−5133.(24) Yang, X., Kuk, J., and Moffat, K. (2008) Crystal structure ofPseudomonas aeruginosa bacteriophytochrome: Photoconversion andsignal transduction. Proc. Natl. Acad. Sci. U. S. A. 105, 14715−14720.(25) Mendez-Ortiz, M. M., Hyodo, M., Hayakawa, Y., andMembrillo-Hernandez, J. (2006) Genome-wide transcriptional profileof Escherichia coli in response to high levels of the second messenger3′,5′-cyclic diguanylic acid. J. Biol. Chem. 281, 8090−8099.(26) Boehm, A., Steiner, S., Zaehringer, F., Casanova, A., Hamburger,F., Ritz, D., Keck, W., Ackermann, M., Schirmer, T., and Jenal, U.(2009) Second messenger signalling governs Escherichia coli biofilminduction upon ribosomal stress. Mol. Microbiol. 72, 1500−1516.(27) Kim, H. K., and Harshey, R. M. (2016) A Diguanylate CyclaseActs as a Cell Division Inhibitor in a Two-Step Response to Reductiveand Envelope Stresses. mBio 7, e00822-16.(28) Weber, H., Pesavento, C., Possling, A., Tischendorf, G., andHengge, R. (2006) Cyclic-di-GMP-mediated signalling within thesigma(S) network of Escherichia coli. Mol. Microbiol. 62, 1014−1034.(29) Tuckerman, J. R., Gonzalez, G., and Gilles-Gonzalez, M. A.(2011) Cyclic di-GMP Activation of Polynucleotide PhosphorylaseSignal-Dependent RNA Processing. J. Mol. Biol. 407, 633−639.(30) Fang, X., and Gomelsky, M. (2010) A post-translational, c-di-GMP-dependent mechanism regulating flagellar motility. Mol. Micro-biol. 76, 1295−1305.(31) Povolotsky, T. L., and Hengge, R. (2016) Genome-BasedComparison of Cyclic Di-GMP Signaling in Pathogenic andCommensal Escherichia coli Strains. J. Bacteriol. 198, 111−126.

ACS Synthetic Biology Research Article

DOI: 10.1021/acssynbio.7b00289ACS Synth. Biol. 2018, 7, 240−248

247

(32) Olson, E. J., and Tabor, J. J. (2012) Post-translational toolsexpand the scope of synthetic biology. Curr. Opin. Chem. Biol. 16,300−306.(33) Castillo-Hair, S. M., Igoshin, O. A., and Tabor, J. J. (2015) Howto train your microbe: methods for dynamically characterizing genenetworks. Curr. Opin. Microbiol. 24, 113−123.(34) Shu, X., Royant, A., Lin, M. Z., Aguilera, T. A., Lev-Ram, V.,Steinbach, P. A., and Tsien, R. Y. (2009) Mammalian expression ofinfrared fluorescent proteins engineered from a bacterial phytochrome.Science 324, 804−807.(35) Bellini, D., and Papiz, M. Z. (2012) Structure of abacteriophytochrome and light-stimulated protomer swapping with agene repressor. Structure 20, 1436−1446.(36) Gomelsky, M., and Kaplan, S. (1995) Genetic-Evidence ThatPpsr from Rhodobacter-Sphaeroides-2.4.1 Functions as a Repressor ofPuc and Bchf Expression. J. Bacteriol. 177, 1634−1637.(37) Ponnampalam, S. N., and Bauer, C. E. (1997) DNA bindingcharacteristics of CrtJ - A redox-responding repressor of bacterio-chlorophyll, carotenoid, and light harvesting-II gene expression inRhodobacter capsulatus. J. Biol. Chem. 272, 18391−18396.(38) Jaubert, M., Zappa, S., Fardoux, J., Adriano, J. M., Hannibal, L.,Elsen, S., Lavergne, J., Vermeglio, A., Giraud, E., and Pignol, D. (2004)Light and redox control of photosynthesis gene expression inBradyrhizobium: dual roles of two PpsR. J. Biol. Chem. 279, 44407−44416.(39) Giraud, E., and Vermeglio, A. (2008) Bacteriophytochromes inanoxygenic photosynthetic bacteria. Photosynth. Res. 97, 141−153.(40) Solovyev, V., and Salamov, A. (2011) Automatic Annotation ofMicrobial Genomes and Metagenomic Sequences, in Metagenomicsand Its Applications in Agriculture, Biomedicine and EnvironmentalStudies (Li, R. W., Ed.), pp 61−78, Nova Science Publishers.(41) Kincade, J. M., and deHaseth, P. L. (1991) Bacteriophagelambda promoters pL and pR: sequence determinants of in vivoactivity and of sensitivity to the DNA gyrase inhibitor, coumermycin.Gene 97, 7−12.(42) Lutz, R., and Bujard, H. (1997) Independent and tightregulation of transcriptional units in Escherichia coli via the LacR/O,the TetR/O and AraC/I1-I2 regulatory elements. Nucleic Acids Res. 25,1203−1210.(43) Pedelacq, J. D., Cabantous, S., Tran, T., Terwilliger, T. C., andWaldo, G. S. (2006) Engineering and characterization of a superfoldergreen fluorescent protein. Nat. Biotechnol. 24, 79−88.(44) Salis, H. M., Mirsky, E. A., and Voigt, C. A. (2009) Automateddesign of synthetic ribosome binding sites to control proteinexpression. Nat. Biotechnol. 27, 946−950.(45) Borujeni, A. E., Channarasappa, A. S., and Salis, H. M. (2014)Translation rate is controlled by coupled trade-offs between siteaccessibility, selective RNA unfolding and sliding at upstream standbysites. Nucleic Acids Res. 42, 2646−2659.(46) Gao, R., and Stock, A. M. (2009) Biological Insights fromStructures of Two-Component Proteins. Annu. Rev. Microbiol. 63,133−154.(47) Evans, K., Fordham-Skelton, A. P., Mistry, H., Reynolds, C. D.,Lawless, A. M., and Papiz, M. Z. (2005) A bacteriophytochromeregulates the synthesis of LH4 complexes in Rhodopseudomonaspalustris. Photosynth. Res. 85, 169−180.(48) Olson, E. J., and Tabor, J. J. (2014) Optogenetic character-ization methods overcome key challenges in synthetic and systemsbiology. Nat. Chem. Biol. 10, 502−511.(49) Kaberniuk, A. A., Shemetov, A. A., and Verkhusha, V. V. (2016)A bacterial phytochrome-based optogenetic system controllable withnear-infrared light. Nat. Methods 13, 591−597.(50) Shaner, N. C., Steinbach, P. A., and Tsien, R. Y. (2005) A guideto choosing fluorescent proteins. Nat. Methods 2, 905−909.(51) Shcherbakova, D. M., Baloban, M., and Verkhusha, V. V. (2015)Near-infrared fluorescent proteins engineered from bacterial phyto-chromes. Curr. Opin. Chem. Biol. 27, 52−63.

(52) Davidson, E. A., Basu, A. S., and Bayer, T. S. (2013)Programming microbes using pulse width modulation of opticalsignals. J. Mol. Biol. 425, 4161−4166.(53) Zechner, C., Seelig, G., Rullan, M., and Khammash, M. (2016)Molecular circuits for dynamic noise filtering. Proc. Natl. Acad. Sci. U.S. A. 113, 4729−4734.(54) Karig, D., and Weiss, R. (2005) Signal-amplifying genetic circuitenables in vivo observation of weak promoter activation in the Rhlquorum sensing system. Biotechnol. Bioeng. 89, 709−718.(55) Westbrook, A. M., and Lucks, J. B. (2017) Achieving largedynamic range control of gene expression with a compact RNAtranscription-translation regulator. Nucleic Acids Res. 45, 5614−5624.(56) Liu, C. C., Qi, L., Lucks, J. B., Segall-Shapiro, T. H., Wang, D.,Mutalik, V. K., and Arkin, A. P. (2012) An adaptor from translationalto transcriptional control enables predictable assembly of complexregulation. Nat. Methods 9, 1088−1094.(57) Fernandez-Rodriguez, J., and Voigt, C. A. (2016) Post-translational control of genetic circuits using Potyvirus proteases.Nucleic Acids Res. 44, 6493−6502.(58) Yang, L., Nielsen, A. A. K., Fernandez-Rodriguez, J., McClune,C. J., Laub, M. T., Lu, T. K., and Voigt, C. A. (2014) Permanentgenetic memory with > 1-byte capacity. Nat. Methods 11, 1261−1266.(59) Esvelt, K. M., Carlson, J. C., and Liu, D. R. (2011) A system forthe continuous directed evolution of biomolecules. Nature 472, 499−503.(60) Ellefson, J. W., Meyer, A. J., Hughes, R. A., Cannon, J. R.,Brodbelt, J. S., and Ellington, A. D. (2013) Directed evolution ofgenetic parts and circuits by compartmentalized partnered replication.Nat. Biotechnol. 32, 97−101.(61) Rockwell, N. C., Martin, S. S., and Lagarias, J. C. (2016)Identification of Cyanobacteriochromes Detecting Far-Red Light.Biochemistry 55, 3907−3919.(62) Engler, C., Gruetzner, R., Kandzia, R., and Marillonnet, S.(2009) Golden Gate Shuffling: A One-Pot DNA Shuffling MethodBased on Type IIs Restriction Enzymes. PLoS One 4, e5553.(63) Gibson, D. G., Young, L., Chuang, R. Y., Venter, J. C.,Hutchison, C. A., and Smith, H. O. (2009) Enzymatic assembly ofDNA molecules up to several hundred kilobases. Nat. Methods 6, 343−345.(64) Braatsch, S., Johnson, J. A., Noll, K., and Beatty, J. T. (2007)The O2-responsive repressor PpsR2 but not PpsR1 transduces a lightsignal sensed by the BphP1 phytochrome in Rhodopseudomonaspalustris CGA009. FEMS Microbiol. Lett. 272, 60−64.(65) Deboer, H. A., Comstock, L. J., and Vasser, M. (1983) The TacPromoter - a Functional Hybrid Derived from the Trp and LacPromoters. Proc. Natl. Acad. Sci. U. S. A. 80, 21−25.(66) Mutalik, V. K., Guimaraes, J. C., Cambray, G., Lam, C.,Christoffersen, M. J., Mai, Q. A., Tran, A. B., Paull, M., Keasling, J. D.,Arkin, A. P., and Endy, D. (2013) Precise and reliable gene expressionvia standard transcription and translation initiation elements. Nat.Methods 10, 354−360.(67) Zhou, L., Lei, X. H., Bochner, B. R., and Wanner, B. L. (2003)Phenotype microarray analysis of Escherichia coli K-12 mutants withdeletions of all two-component systems. J. Bacteriol. 185, 4956−4972.(68) Daeffler, K. N., Galley, J. D., Sheth, R. U., Ortiz-Velez, L. C.,Bibb, C. O., Shroyer, N. F., Britton, R. A., and Tabor, J. J. (2017)Engineering bacterial thiosulfate and tetrathionate sensors fordetecting gut inflammation. Mol. Syst. Biol. 13, 923.(69) Castillo-Hair, S. M., Sexton, J. T., Landry, B. P., Olson, E. J.,Igoshin, O. A., and Tabor, J. J. (2016) FlowCal: A User-Friendly,Open Source Software Tool for Automatically Converting FlowCytometry Data from Arbitrary to Calibrated Units. ACS Synth. Biol. 5,774−780.(70) Gerhardt, K. P., Olson, E. J., Castillo-Hair, S. M., Hartsough, L.A., Landry, B. P., Ekness, F., Yokoo, R., Gomez, E. J., Ramakrishnan,P., Suh, J., Savage, D. F., and Tabor, J. J. (2016) An open-hardwareplatform for optogenetics and photobiology. Sci. Rep. 6, 35363.

ACS Synthetic Biology Research Article

DOI: 10.1021/acssynbio.7b00289ACS Synth. Biol. 2018, 7, 240−248

248