- Research article
- Open Access
Layering genetic circuits to build a single cell, bacterial half adder
© Wong et al. 2015
- Received: 28 April 2015
- Accepted: 3 June 2015
- Published: 16 June 2015
Gene regulation in biological systems is impacted by the cellular and genetic context-dependent effects of the biological parts which comprise the circuit. Here, we have sought to elucidate the limitations of engineering biology from an architectural point of view, with the aim of compiling a set of engineering solutions for overcoming failure modes during the development of complex, synthetic genetic circuits.
Using a synthetic biology approach that is supported by computational modelling and rigorous characterisation, AND, OR and NOT biological logic gates were layered in both parallel and serial arrangements to generate a repertoire of Boolean operations that include NIMPLY, XOR, half adder and half subtractor logics in a single cell. Subsequent evaluation of these near-digital biological systems revealed critical design pitfalls that triggered genetic context-dependent effects, including 5′ UTR interferences and uncontrolled switch-on behaviour of the supercoiled σ54 promoter. In particular, the presence of seven consecutive hairpins immediately downstream of the promoter transcription start site severely impeded gene expression.
As synthetic biology moves forward with greater focus on scaling the complexity of engineered genetic circuits, studies which thoroughly evaluate failure modes and engineering solutions will serve as important references for future design and development of synthetic biological systems. This work describes a representative case study for the debugging of genetic context-dependent effects through principles elucidated herein, thereby providing a rational design framework to integrate multiple genetic circuits in a single prokaryotic cell.
- Logic Gate
- Genetic Circuit
- Half Adder
- High Copy Plasmid
- pBAD Promoter
Gene regulation in biological systems behaves like a molecular computer whereby the gene’s output can be modelled as on-off states of Boolean (digital) logic [1–3]. However, programming gene regulation is far from trivial and requires considerable time and effort during functional testing and tuning of the synthetic genetic circuits under development. Apart from the scarcity of reliable and well-characterised biological parts, digital performance in biological systems is further impacted by the cellular and genetic context-dependent effects of the biological parts which comprise the circuit [4–6]. Recent studies have shown that genetic crosstalk between the engineered circuits and endogenous networks of the host cell can lead to cellular context-dependent effects [7, 8]. For this reason, molecular parts and devices that are orthogonal to the cell native machineries with roles in either genetic transcription or protein translation have been created to enable predictable engineering of genetic circuits [9–13]. Demonstrations of layered genetic circuits in a single cell, such as the execution of a 4-input AND gate in bacteria  and biological half adders and half subtractors in mammalian cells  have revealed that orthogonal logic gates can be interlinked to perform digital operations of higher complexity and diversified outputs. While the capability to program cells with memory and decision-making functions [15–19] presents many opportunities in biotechnological applications, a lack of formal understanding associated with genetic context-dependent effects has limited progress in engineering biology. In this respect, two studies have shown that the 5′ untranslated region (5′-UTR) of mRNA can affect the temporal control of multigene operons or inverter-based genetic circuits, and RNA processing using clustered regularly interspaced short palindromic repeats (CRISPRs) or ribozymes can serve as effective genetic insulators to buffer such context-dependent effects [5, 20]. In this paper, we have sought to elucidate the limitations of engineering biology from an architectural point of view, with the aim of creating a set of engineering solutions for overcoming failure modes during the development of complex, synthetic genetic circuits.
Design of biological half adder
In this study we were interested in developing biological half adders in prokaryotic systems — particularly in microbes which exhibit much faster cell division and shorter cycle time — so that they can be broadly applied in different biotechnological applications. In contrast to the mammalian cell-based half adder, which is developed mainly for therapeutic and biosensing applications, a prokaryotic half adder can be used to enhance molecular process control and decision making, for example, in drug and biofuel production, biosensing, bioremediation  and probiotic engineering for the treatment of metabolic disorders , cancer  and infectious diseases [24, 25]. In digital processing, half adders form the key building blocks for shift registers, binary counters and serial parallel data converters. Likewise in biological systems, a combination of half adders can be connected in various arrangements to regulate gene expression with diverse, digital-like performance. In doing so, biological systems can be made to interface with novel biomolecular devices, allowing the repurposing of cellular phenotype, as well as providing new platforms to probe and elucidate biological functions [26–28].
On induction with arabinose and rhamnose, the transcription factors AraC and RhaS, both of which are constitutively expressed in a single transcript by promoter pCON, associate with their corresponding inducers to activate expression of the enhancer-binding proteins HrpS and HrpR. This results in the activation of the AND logic and the concurrent synthesis of GFP reporter and lambda repressor (λCl) by the pHrpL promoter. Consequently, genetic events of the OR gate, which run in parallel with the HrpRS AND gate, are then turned off due to obstructive repression by λCl molecules. In all, the half adder demonstrates both AND (SUM Output) and XOR (CARRY Output) logic operations; the latter operation is a processed outcome achieved by sequential and parallel layering of AND, OR and NOT logic (Fig. 1b). By comparison, induction with either inducer singly will trigger only genetic operation of the OR gate, resulting in the synthesis of RFP reporter, but not GFP and λCl molecules (Fig. 1c). Finally, we also demonstrate the development of a single cell prokaryotic, half subtractor via slight modifications to the half adder circuit.
Characterisation of input devices
The choice of input signals presents the first possible complication in terms of parts modularity. For this reason, genetic circuits of higher complexity with multiple inputs often utilise promoter systems which are activated by inducers of vastly dissimilar chemical nature, namely IPTG, tetracycline, arabinose, 3OC12HSL and C4HSL. Previous studies have shown that a subset of quorum sensing promoters can be activated by homoserine lactone inducers of similar carbon chain length [29, 30]. Likewise, the wild-type pBAD promoter is affected by lactose analogues, requiring further mutagenesis to avoid crosstalk inhibition . Instances of cross-phosphorylation have also been observed in two component signal transduction systems between otherwise distinct pathways . Thus, it is important for inducible input devices to be carefully characterised for their steady state transfer function and pairing compatibility before further assembly into higher ordered logic devices.
While previous studies with pRHAB promoter involved genetic circuits that include both RhaR and RhaS transcription factors [33–35], in this paper we demonstrate that the rhamnose inducible promoter pRHAB requires only RhaS for full activation and displays tight regulation even when RhaS is overexpressed. Additional file 1: Figures S2C and S3C show the steady state transfer functions of input device A, pBAD (Additional file 1: Figure S2A) and input device B, pRHAB (Additional file 1: Figure S3A) expressing RFP under strong ribosome binding sites (RBSs) by their corresponding inducers, respectively.
To examine the possibility of genetic cross-communication, we constructed genetic circuits that couple GFP production to pBAD activation and RFP production to pRHAB activation. The results show that varying the concentration of arabinose did not activate pRHAB promoter activity (Additional file 1: Figure S4A). A similar trend was observed in pBAD promoter with rhamnose (Additional file 1: Figure S4B). Interestingly, the simultaneous introduction of both sugars modified the transfer function of each promoter slightly, which may be a result of differential cell growth, sugar import rate or antagonistic effect of one sugar to another. This effect, however, is insignificant as definite ON and OFF switch behaviours are apparent — thereby confirming the pairing compatibility of pBAD and pRHAB promoters.
Design and characterisation of AND logic gate
Designs of highly modularised, prokaryotic AND logic devices have hitherto involved the use of multiple plasmids [10, 12, 16, 36, 37]. In this work, we assembled the AND logic gate in a single plasmid. This procedure has enabled us to localise the AND logic gate in a single vector, and facilitated the downstream troubleshooting and tuning of layered genetic circuits.
To assess the effect of plasmid copy number on the performance of the AND gate, modules were constructed which generate the HrpRS transcription activators (pBAD-HrpS-pRHAB-HrpR). This produces a GFP output (pHrpL-GFP) into separate low and high copy plasmids (co-transforming the plasmids into E. coli cells).The relative GFP output of each system was measured (Fig. 2d). The results show that the AND gate system with the GFP-producing module in the high copy plasmid and the HrpRS transcription activators in the low copy plasmid produced a >4-fold greater GFP output than AND gate systems with GFP-producing module in low copy plasmid and HrpRS (as compared to transcription activators in either low or high copy plasmids). The result indicates that a higher concentration of HrpRS transcription activators, above the saturation limit of the pHrpL promoter, do not produce a greater GFP output. It is likely that the transcriptional output of the HrpRS AND gate is limited by the strength of the weak pHrpL promoter. Hence, the conclusion is that when pHrpL-GFP module was expressed in high copy plasmids, the intracellular availability of pHrpL promoters was increased — resulting in the amplification of GFP output.
Design and characterisation of OR logic gate
Genetic context effect of σ54-dependent pHrpL promoter
To enable sufficient expression of the λCl repressor by an AND gate system, the gene encoding for λCl repressor was assembled downstream of a σ54-dependent pHrpL promoter on a high copy plasmid. Fortuitously, we discovered that a pHrpL promoter located downstream of another pHrpL expression cassette can be turned on even in the absence of its cognate HrpRS transcription factors (Additional file 1: Figure S5C). The converse is not true for an upstream pHrpL promoter (Additional file 1: Figure S5B). Negative controls with just the GFP reporter or RBS-λCl gene upstream of the pHrpL-GFP module confirmed that the pHrpL promoter alone is not leaky and that a cryptic promoter is absent in the λCl gene (Additional file 1: Figure S5A and S5D). To buffer against this genetic context-dependent effect of the pHrpL promoters, pHrpL-GFP and pHrpL-λCl modules were assembled on separate plasmids. This successfully prevented the genetic interference of both pHrpL expression modules with each other (Additional file 1: Figure S5E and S5F). Additional file 1: Figure S5G shows a quantitative assessment of pHrpL promoter activation due to the presence of another upstream pHrpL promoter and the use of plasmids as genetic insulators.
Design and characterisation of NOT and NIMPLY logic gates
As part of the development of XOR logic operations of the half adder, repressor binding sites are required downstream of the OR gate promoters. To examine the minimal number of λCl repressor binding sites required for effective repression, a single λCl operator site and dual λCl operator sites of perfect dyad symmetry were fused downstream of the pBAD promoter, before the RFP gene . The repressibility of both circuits was tested by generating λCl repressors from an HrpRS AND gate in a separate plasmid. Negligible repression was observed when only one λCl repressor operator site was present. In the presence of two operator sites of perfect dyad symmetry, RFP expression from the pBAD promoter was greatly attenuated — even when the λCl repressor was not synthesised. We postulate that the observed reduction of RFP expression might be caused by the presence of secondary hairpin structures immediately downstream of the TSS acting as pseudo transcription terminators or locking the RBS in conformations that prevented translation initiation (Additional file 1: Figure S6A).
In order to examine this further, random mutagenesis on the natural sequence of the λCl repressor operator sites was performed with screening for mutants with significant difference in RFP expression levels, in the absence and presence of the λCl repressor. Accordingly, an evolved candidate (Cl2B) with 4 mutations in the inverted sequence of the λCl repressor binding (Additional file 1: Figure S6B) was obtained. Sequence comparison with the original λCl repressor binding sites (Cl2A) with the evolved candidate revealed that the directed evolution process had eliminated the effect of secondary hairpin structures from 7 to 3. Next, the efficiency of λCl-mediated transcription termination in the context of a genetic NIMPLY gate was studied. This was achieved by placing repressor binding sites directly downstream of tandem pBAD-pRHAB promoters and generating λCl repressors from a separate pBAD expression cassette.
Design and characterisation of XOR logic gate
Accordingly, only design IV was able to achieve well-balanced outputs which accurately described XOR logic operations (Fig. 5c). While design I demonstrated the strong suppression of RFP output in the presence of both inputs (arabinose and rhamnose), when characterised as a NIMPLY gate (as described earlier), the same design failed to function in the context of a XOR gate in which a weaker pHrpL promoter was used to drive the synthesis λCl repressors instead of the strong pBAD promoter. Interestingly, the results imply that when employing transcription repressors as molecular blockers to mRNA elongation, a higher concentration of λCl molecules is needed to completely suppress transcription as λCl binding sites are engineered further away from the transcription start site. This observation may be an effect of RNAP gaining momentum as it runs down template DNA to perform transcription, inadvertently enabling RNAP to continue its course of action as a result of the inadequacy of ‘molecular brakes’.
While designs II and III, which were developed with λCl binding sites downstream of both pBAD and pRHAB promoters, exhibited a slight semblance of XOR logic operations, the presence of multiple, repeated sequences of λCl binding sites in the transcript generated from the pBAD promoter greatly reduced the RFP output from input A. Using untagged RFP gene in design III led to a slight increase in overall RFP output but did not alleviate the signal balancing issue. The result implies that the 5′UTR structural effect is more dominant than RFP half-life in determining the success of the layered XOR gate. In order to apply the XOR gate in the implementation of the half adder, design IV was characterised for its steady state profile by titrating with varying concentrations of arabinose and rhamnose as shown in Fig. 5d. It is noteworthy that the XOR gate developed in this work possesses higher single cell computational capability compared to that achieved by Tamsir and colleagues using a network of inter-communicating cells , hence circumventing problems associated with cell-cell communication.
Design and characterisation of single cell half adder and half subtractor
Failure modes and engineering solutions for the design and building of layered genetic circuits in a single (bacterial) cell
Genetic crosstalk: Input switch devices crosstalk with one another.
▪ Check pairwise compatibility by placing GFP and RFP under the regulation of each input switch device
▪ Perform mutagenesis on promoter or DNA-binding protein to identify orthogonal pairs.
Stoichiometric mismatch: Amount of AND gate’s transcription activators are disproportionately matched, resulting in ‘leaky’ AND gate.
▪ Characterise the expression profile of input genetic switches with different RBSs and input the resultant transfer function equations into a steady state AND gate computational model. Match AND gate sub-modules to obtain stoichiometric balance using this forward engineering approach.
DNA supercoiling: σ54 AND gate promoter is turned on by the DNA supercoil effects of upstream σ54 promoter.
▪ Insulate σ54 promoters using different plasmid vectors.
Stoichiometric mismatch: Outputs from input device I and II are disproportionately matched, resulting in skewed OR gate.
▪ Characterise the expression profile of input genetic switches with different RBSs and input the resultant transfer function equations into a steady state OR gate computational model. Match OR gate sub-modules to obtain stoichiometric balance using this forward engineering approach.
Transcription interference: Tandem promoter OR gate design fails due to downstream DNA sequence acting as a repressor to upstream promoter.
▪ Characterise different permutation of tandem promoter OR gate to identify the optimal genetic architecture.
▪ Separate OR gate promoters into distinct expression cassettes.
Layering OR-NOT into NIMPLY gate
Insufficient repression: Placing single repressor binding site downstream of inducible promoter cannot fully repress gene expression.
▪ Increase repression efficiency by introducing additional repressor binding sites to the NOT gate. Note that the introduction of extra repressor binding sites may also lead to extensive 5′UTR effects.
▪ Attenuate expression ‘leakiness’ by using weaker RBS for the NOT gate
Translation interference: Placing repressor binding sites downstream of inducible promoter creates extensive 5′UTR structural effects.
▪ Perform mutagenesis to relieve RNA hairpin structures at selected sites.
▪ Use RNA processing tools to remove undesired 5′UTR sequences.
Layering AND-OR-NOT into XOR gate
Insufficient repression: Insufficient transcription repressors are generated by upstream genetic circuit to stop transcription elongation, level mismatch.
▪ Reduce repressors required in NOT gate by designing repressor binding sites such that they are immediately downstream of transcription start site.
▪ Increase production of repressor in the AND gate by expressing transcription repressors in high copy plasmid.
Translation interference: Placing repressor binding sites downstream of OR gate tandem promoter creates extensive 5′UTR structural effects.
▪ Separate OR gate promoters into distinct expression cassettes.
▪ Use RNA processing tools to remove undesired 5′UTR sequences.
Overall, the presence of secondary structures in the 5′-UTR of mRNA affects genetic expression most. We discovered that the presence of seven consecutive hairpins immediately downstream of the promoter transcription start site would severely impede gene expression. Although an OR gate design made up of tandem promoters can be subjected to the undesirable effects of the 5′-UTR secondary structure, we showed that the effect is not pronounced in the digital performance of the OR logic when the promoters and DNA operator sites involved are of markedly different DNA sequences. The OR gate design that comprises a separate gene expression cassette also reliably demonstrates digital operation. However, the involvement of larger DNA modules and repetitive use of transcription terminators that are rich in secondary hairpin structures may impede system assembly in terms of construction efficiency and accuracy. Where identical DNA sequences are incorporated in a single mRNA transcript, as shown in designs II and III of the XOR gate, the effect of 5′-UTR secondary structure preventing gene expression is significantly more pronounced. Thus, it is proposed that XOR gate logic in layered genetic circuits should be designed with two discrete expression cassettes instead of employing a tandem promoter circuit design. It would also be interesting to test if RNA processing tools can be employed in multiplex mode to insulate the myriad of biological devices from RNA genetic context-dependent effects in layered genetic circuits concurrently.
Perhaps of particular interest, we discovered that σ54 promoters can exhibit genetic context-dependent effects if two σ54 promoters are placed close to each other. Previously, σ54-dependent NtrC-binding promoters have been reported. These promoters permit transcription in vitro in the absence of enhancer-binding proteins and ATP under conditions that promote DNA melting. These include DNA supercoiling, temperature rise and lower ionic strength, or the implemention of characteristic point mutations on the σ54 protein [46, 47]. In this paper, we show that an upstream σ54 pHrpL promoter could also activate a downstream pHrpL promoter in vivo if the two promoters are in close proximity — possibly as a result of plasmid DNA supercoiling. This undesired switched-on activity can be avoided by designing pHrpL expression modules in different plasmids, that is, using plasmids as genetic buffers to insulate such genetic context-dependent effects.
While recombinases have been intelligently crafted into Boolean logic gates with DNA-encoded memory functions, it is important to note that biosensors connected in AND, OR and XOR operations with recombinase-based logic gates may not be able to distinguish inputs from different environments and provide the desired response. For example, a probiotic that is genetically programmed in AND logic to sense two inputs such as hypoxia and low pH may be activated for hypoxia and low pH signals in two different locations, as compared to sensing both signals in situ. The same may be applicable for other logic operations with recombinase-based logic gates. Thus, layered genetic circuits that are capable of sensing and providing location-sensitive Boolean logic operations are still useful in programming cellular behaviour. Of particular interest is a combination of layered genetic circuits with the synthesis of recombinases as an intermediary output. This may provide a novel and better platform for programmable cellular behaviour in terms of both accuracy and memory.
With a few notable exceptions [19, 48, 49], most studies of synthetic biological systems are centred on the development of rational engineering approaches, reporting successful and advantageous aspects of the engineered systems, with a lesser focus on reporting failure modes and compiling the engineering solutions applied to troubleshoot system failures. As synthetic biology moves forward with greater focus on scaling the complexity of engineered genetic circuits, studies which thoroughly evaluate failure modes and engineering solutions will serve as important references for future design and development of synthetic biological systems.
Strains, plasmids and growth conditions
E. coli strain Top10 (Invitrogen) was used for all the cloning and characterisation experiments. The genes and oligonucleotides used in this study were synthesised by either Geneart (Life Technologies, Grand Island, NY) or Sigma Aldrich (St. Louis, MO). All the enzymes used in this study, including OneTaq and Phusion polymerase, T4 ligase, EcoRI, XbaI, SpeI, PstI and DpnI, were obtained from New England Biolabs. Chloramphenicol (35 μg ml−1) and ampicillin (100 μg ml−1) were added to culture media for experiments involving pSB1C3 and pSB4A5 plasmid vectors, where appropriate. In all the characterisation experiments, cells were inoculated from freshly transformed plates and grown in 2 ml LB (Miller, BD Bioscience, San Jose, CA) with appropriate antibiotic in 50-ml Falcon tubes overnight at 37 °C with 225 rpm shaking unless otherwise stated. Overnight cultures were then diluted to OD600 about 0.002 in 5 ml LB antibiotic and further grown to a final OD of 0.5 ± 0.05 under the same culture conditions (37 °C and 225 rpm shaking). Harvested cells were kept on ice until induction. All inducers used in this study were purchased from Sigma Aldrich with final concentration ranging from 0 to 28 mM.
All genes from E. coli (pBAD, pRHAB, araC and rhaS) were cloned from genomic DNA of strain MG1655 (ATCC 700926). hrpS, hrpR and pHrpL were cloned from an earlier study , while pCON (Bba_J23101), double terminator (Bba_B0015), GFPmut3b (Bba_E0040), RFP (Bba_E1010) and λCl (Bba_C0051) were cloned from the Biobrick registry. PCR was performed using Phusion DNA polymerase in a dual cycle PCR programme at annealing temperatures of 53 °C and 60 °C for the first 7 and subsequent 20 cycles, respectively. Biological parts were spliced by overlap extension PCR and ligated to vectors pSB4A5 (low copy, pSC101 replication origin) and pSB1C3 (high copy, pMB1 replication origin) using XbaI and PstI restriction sites. Composite systems with two or more biological modules were sequentially assembled as previously described .
Parts mutation of λCl repressor binding sites
To obtain sequence variants of λCl repressor binding sites, PCR was performed on a pHrpL-λCl-pBAD-Cl2A template with mutagenesis primers 5′-ttcgaattcgcggccgcttctagaggccggattat and 5′-gctactagtatatNNNNNNNNccggtgatatatggagaaacagta (restriction sites underlined) using Phusion DNA polymerase. The resultant amplificons (about 1.4 kb) were then ligated upstream to the pSB1C3 vector containing an RFP reporter and transformed to competent cells carrying HrpRS AND gate modules in pSB4A5. Single colonies of uniform size were inoculated into a 96-well microplate loaded with 200 ul LBAC (LB with chloramphenicol and ampicillin) and grown in a microplate incubator set at 37 °C with 750 rpm shaking for 6 hours. Accordingly, cultures in each well were triplicated and diluted 10× into 200 ul LBAC with 3.5 mM arabinose and 28 mM rhamnose, 3.5 mM arabinose, and no arabinose under the same growth conditions. Evolved mutants were identified by observable differences in RFP expression and inhibition after 6 hours of induction using a Fluostar OPTIMA microplate reader (BMG Labtechnologies). Validation and characterisation of isolated candidate parts was independently performed in 175 ul LBAC in 1.5-ml microcentrifuge tubes after 4 hours induction at 37 °C and 1,000 rpm shaking under four different logic conditions.
Modelling of AND, OR and NIMPLY logic gates
To enable model-driven design synthetic biological systems, we examine the effect of ribosome binding sites (RBSs) on the steady state transfer function of input switch devices. By analysing reference data  that had previously characterised the input–output relationship of genetic switches in the form of Eq. 1, we observed that parameters that are most sensitive to changes in RBS are parameters A and B. Hence, by knowing the relative output of switch devices with weaker RBSs by either prediction from reliable software or by single experimental measurement of a device’s output at input maximal, the parameters A and B can be scaled proportionally to obtain a priori parameters that accurately predict the transfer function of other devices with weaker RBSs (Additional file 1: Figure S1A). We validated our approach with previously published data sets (Additional file 1: Table S1) and showed that the transfer function of input devices pLuxR (Additional file 1: Figure S1A) and pBAD (Additional file 1: Figure S1B) with different RBSs can be reliably estimated without additional experimentation. MATLAB modelling scripts are available in Additional file 2.
The transfer functions of input switch devices used in this work with strong RBSs were empirically fitted into the Hill-like equation (Eq. 1), while those of input switch devices with weak RBSs were predicted using the validated method as discussed above. Additional file 1: Table S3 shows the empirical transfer function parameters of the various input switch devices. AND and OR gate profiles were then modelled and predicted using these parameters and equations as shown in Additional file 2: Eqs. 7 and 10. Additional file 1: Figures S11, S12 and S13 show the predicted normalised output of the HrpRS AND gate and various OR gate combinations. The NIMPLY gate was empirically modelled using Additional file 2: Eq. 2 and parameters from Additional file 1: Table S2. Additional file 1: Figure S10 shows the predicted output of the NIMPLY gate.
Characterisation and orthogonality testing of input switch devices
Equation 1 models reporter output (Y) as a function of input concentration of inducer ([X]). The four parameters (A, B, C, n) were estimated to obtain the best fit curve by performing a nonlinear curve fitting using the experimental results. This curve fitting was performed using the nonlinear least square fitting functions in the MATLAB Curve Fitting Toolbox (The Mathworks, Natick, MA, USA).
Characterisation of AND, OR, NIMPLY, XOR, half adder and half subtractor
To characterise the steady state profile of AND and OR logic devices, reinoculated cultures at OD600 about 0.5 were transferred to black, flat-bottom 96-well plates (Greiner Bio-One, catalogue number 655090) in aliquots of 150 μl for induction with varying concentrations of rhamnose and arabinose ranging from 8.38E-07 M to 2.74E-02 M. The plates were sealed with gas-permeable foils and incubated at 37 °C with 750 rpm shaking for 3 hours. Fluorescence and optical density data were collected using a Fluostar Optima microplate reader (BMG Labtech) and zeroed with blank LB media.
To characterise the steady state profile of NIMPLY, XOR, half adder and half subtractor logic devices, the above procedures were repeated with slight modifications to reduce evaporation losses in constructs with weaker RBS-RFP modules. Briefly, reinoculated cultures were dispensed in 175 μl aliquots into 1.5-ml capped tubes and induced with varying concentrations of rhamnose and arabinose, as described above. The aliquots were incubated on a thermomixer platform (Eppendorf) set at 37 °C with 1,000 rpm shaking for 4 hours. 150-μl aliquots from each tube were transferred to black, flat-bottom 96-well plates and assayed for fluorescence and optical density with Synergy HT or H1m microplate readers (Biotek Instruments Inc.). To assess the digital performance of all logic devices, cell cultures were separately induced with water, 28 mM rhamnose and/or 7 mM arabinose in four different logic conditions. The induced cultures were incubated in the respective conditions as described above and assayed for fluorescence. All results were normalised with OD600-estimated cell density and provided in arbitrary units.
Fluorescence imaging of AND and OR gates
For the acquisition of fluorescent images in AND and OR logic devices, reinoculated cultures were transferred to 50-ml tubes in aliquots of 5 ml and separately induced with water, 28 mM rhamnose and/or 7 mM arabinose in four different logic conditions overnight. After 15 hours, cell pellets were harvested and transferred to 1.5-ml tubes for fluorescent imaging with suitable filters. Images were acquired with a high mega-pixel mobile phone camera.
Reinoculated cultures were dispensed in 175-μl aliquots into 1.5-ml capped tubes and separately induced with water, 28 mM rhamnose and/or 7 mM arabinose in four different logic conditions. The aliquots were grown on a thermomixer platform (Eppendorf, Germany) set at 37 °C with 1,000 rpm shaking for 4 hours. Before assay, 5 μl culture from each sample were diluted 200x in 0.22 μm filtered DI water (pH 7). All expression data were collected using a BD LSRFortessa X-20 flow cytometer (BD Biosciences, San Jose, CA) with a 488 nm argon excitation laser, and 530 nm ± 30 (FITC) and 610 nm ± 20 (PE-CF594) emission filters. The data were gated using both forward (550 v, threshold 1,500 v) and side scatter (310 v) with the neutral density filter removed. At least 10,000 events were recorded per sample. FITC and PE-CF594 channels were set at 466 v and 852 v, respectively. Data analysis was performed with FlowJo (TreeStar Inc., Ashland, OR).
We are grateful to P. Freemont, K. Jensen, C. Hirst and F. Jonas from Imperial College London for suggestions, M.W. Chang from the University of Singapore for experimental support and the use of the flow cytometer for the work, and B.J. Wang for the donation of plasmids containing hrpS, hrpR and pHrpL promoter. We would also like to thank the anonymous reviewers for their comments on this paper.
AW is funded by the NTU-IC joint Ph.D. scholarship programme. RIK would like to acknowledge the financial support of the UK Engineering and Physical Sciences Research Council for the project, and CLP would like to acknowledge the financial support of the Ministry of Education of Singapore (AcRF ARC43/13) for the project.
- Mayo AE, Setty Y, Shavit S, Zaslaver A, Alon U. Plasticity of the cis-regulatory input function of a gene. PLoS Biol. 2006;4:e45.PubMed CentralPubMedView ArticleGoogle Scholar
- Peter IS, Faure E, Davidson EH. Predictive computation of genomic logic processing functions in embryonic development. Proc Natl Acad Sci U S A. 2012;109:16434–42.PubMed CentralPubMedView ArticleGoogle Scholar
- Thomas R. Boolean formalization of genetic control circuits. J Theor Biol. 1973;42:563–85.PubMedView ArticleGoogle Scholar
- Khalil AS, Lu TK, Bashor CJ, Ramirez CL, Pyenson NC, Joung JK, et al. A synthetic biology framework for programming eukaryotic transcription functions. Cell. 2012;150:647–58.PubMed CentralPubMedView ArticleGoogle Scholar
- Lou C, Stanton B, Chen Y-J, Munsky B, Voigt CA. Ribozyme-based insulator parts buffer synthetic circuits from genetic context. Nat Biotechnol. 2012;30:1137–42.PubMed CentralPubMedView ArticleGoogle Scholar
- Mutalik VK, Guimaraes JC, Cambray G, Lam C, Christoffersen MJ, Mai Q-A, et al. Precise and reliable gene expression via standard transcription and translation initiation elements. Nat Methods. 2013;10:354–60.PubMedView ArticleGoogle Scholar
- Cardinale S, Joachimiak Marcin P, Arkin Adam P. Effects of genetic variation on the E. coli host-circuit interface. Cell Reports. 2013;4:231–7.PubMedView ArticleGoogle Scholar
- Cardinale S, Arkin AP. Contextualizing context for synthetic biology – identifying causes of failure of synthetic biological systems. Biotechnol J. 2012;7:856–66.PubMed CentralPubMedView ArticleGoogle Scholar
- Rackham O, Chin JW. A network of orthogonal ribosome-mRNA pairs. Nat Chem Biol. 2005;1:159–66.PubMedView ArticleGoogle Scholar
- Moon TS, Lou C, Tamsir A, Stanton BC, Voigt CA. Genetic programs constructed from layered logic gates in single cells. Nature. 2012;491:249–53.PubMed CentralPubMedView ArticleGoogle Scholar
- Lucks JB, Qi L, Mutalik VK, Wang D, Arkin AP. Versatile RNA-sensing transcriptional regulators for engineering genetic networks. Proc Natl Acad Sci U S A. 2011;108:8617–22.PubMed CentralPubMedView ArticleGoogle Scholar
- Wang B, Kitney RI, Joly N, Buck M. Engineering modular and orthogonal genetic logic gates for robust digital-like synthetic biology. Nat Commun. 2011;2:508.PubMed CentralPubMedView ArticleGoogle Scholar
- An W, Chin JW. Synthesis of orthogonal transcription-translation networks. Proc Natl Acad Sci U S A. 2009;106:8477–82.PubMed CentralPubMedView ArticleGoogle Scholar
- Auslander S, Auslander D, Muller M, Wieland M, Fussenegger M. Programmable single-cell mammalian biocomputers. Nature. 2012;487:123–7.PubMedGoogle Scholar
- Siuti P, Yazbek J, Lu TK. Synthetic circuits integrating logic and memory in living cells. Nat Biotechnol. 2013;31:448–52.PubMedView ArticleGoogle Scholar
- Bonnet J, Yin P, Ortiz ME, Subsoontorn P, Endy D. Amplifying genetic logic gates. Science. 2013;340:599–603.PubMedView ArticleGoogle Scholar
- Gardner TS, Cantor CR, Collins JJ. Construction of a genetic toggle switch in Escherichia coli. Nature. 2000;403:339–42.PubMedView ArticleGoogle Scholar
- Friedland A, Lu T, Wang X, Shi D, Church G, Collins J. Synthetic gene networks that count. Science. 2009;324:1199–202.PubMed CentralPubMedView ArticleGoogle Scholar
- Bonnet J, Subsoontorn P, Endy D. Rewritable digital data storage in live cells via engineered control of recombination directionality. Proc Natl Acad Sci U S A. 2012;109:8884–9.PubMed CentralPubMedView ArticleGoogle Scholar
- Qi L, Haurwitz RE, Shao W, Doudna JA, Arkin AP. RNA processing enables predictable programming of gene expression. Nat Biotechnol. 2012;30:1002–6.PubMedView ArticleGoogle Scholar
- Nikel PI, Martinez-Garcia E, de Lorenzo V. Biotechnological domestication of pseudomonads using synthetic biology. Nat Rev Microbiol. 2014;12:368–79.PubMedView ArticleGoogle Scholar
- Chen ZY, Guo LL, Zhang YQ, Walzem RL, Pendergast JS, Printz RL, et al. Incorporation of therapeutically modified bacteria into gut microbiota inhibits obesity. J Clin Invest. 2014;124:3391–406.PubMed CentralPubMedView ArticleGoogle Scholar
- Xiang S, Fruehauf J, Li CJ. Short hairpin RNA-expressing bacteria elicit RNA interference in mammals. Nat Biotechnol. 2006;24:697–702.PubMedView ArticleGoogle Scholar
- Duan F, March JC. Engineered bacterial communication prevents Vibrio cholerae virulence in an infant mouse model. Proc Natl Acad Sci U S A. 2010;107:11260–4.PubMed CentralPubMedView ArticleGoogle Scholar
- Saeidi N, Wong CK, Lo TM, Nguyen HX, Ling H, Leong SSJ, et al. Engineering microbes to sense and eradicate Pseudomonas aeruginosa, a human pathogen. Mol Syst Biol. 2011;7:521.PubMed CentralPubMedView ArticleGoogle Scholar
- Callura J, Dwyer D, Isaacs F, Cantor C, Collins J. Tracking, tuning, and terminating microbial physiology using synthetic riboregulators. Proc Natl Acad Sci U S A. 2010;107:15898–903.PubMed CentralPubMedView ArticleGoogle Scholar
- Callura JM, Cantor CR, Collins JJ. Genetic switchboard for synthetic biology applications. Proc Natl Acad Sci U S A. 2012;109:5850–5.PubMed CentralPubMedView ArticleGoogle Scholar
- Temme K, Hill R, Segall-Shapiro TH, Moser F, Voigt CA. Modular control of multiple pathways using engineered orthogonal T7 polymerases. Nucleic Acids Res. 2012;40:8773–81.PubMed CentralPubMedView ArticleGoogle Scholar
- Collins CH, Leadbetter JR, Arnold FH. Dual selection enhances the signaling specificity of a variant of the quorum-sensing transcriptional activator LuxR. Nat Biotechnol. 2006;24:708–12.PubMedView ArticleGoogle Scholar
- Canton B, Labno A, Endy D. Refinement and standardization of synthetic biological parts and devices. Nat Biotechnol. 2008;26:787–93.PubMedView ArticleGoogle Scholar
- Lee SK, Chou HH, Pfleger BF, Newman JD, Yoshikuni Y, Keasling JD. Directed evolution of AraC for improved compatibility of arabinose- and lactose-inducible promoters. Appl Environ Microb. 2007;73:5711–5.View ArticleGoogle Scholar
- Laub MT, Goulian M. Specificity in two-component signal transduction pathways. Annu Rev Genet. 2007;41:121–45.PubMedView ArticleGoogle Scholar
- Yu BJ, Kang KH, Lee JH, Sung BH, Kim MS, Kim SC. Rapid and efficient construction of markerless deletions in the Escherichia coli genome. Nucleic Acids Res. 2008;36:e84.PubMed CentralPubMedView ArticleGoogle Scholar
- Bloodworth RAM, Gislason AS, Cardona ST. Burkholderia cenocepacia conditional growth mutant library created by random promoter replacement of essential genes. Microbiology Open. 2013;2:243–58.PubMed CentralPubMedView ArticleGoogle Scholar
- Cardona ST, Mueller CL, Valvano MA. Identification of essential operons with a rhamnose-inducible promoter in Burkholderia cenocepacia. Appl Environ Microb. 2006;72:2547–55.View ArticleGoogle Scholar
- Shis DL, Bennett MR. Library of synthetic transcriptional AND gates built with split T7 RNA polymerase mutants. Proc Natl Acad Sci U S A. 2013;110:5028–33.PubMed CentralPubMedView ArticleGoogle Scholar
- Anderson JC, Voigt CA, Arkin AP. Environmental signal integration by a modular AND gate. Mol Syst Biol. 2007;3:133.PubMed CentralPubMedView ArticleGoogle Scholar
- Tamsir A, Tabor JJ, Voigt CA. Robust multicellular computing using genetically encoded NOR gates and chemical ‘wires’. Nature. 2011;469:212–5.PubMed CentralPubMedView ArticleGoogle Scholar
- Glick BR. Metabolic load and heterologous gene expression. Biotechnol Adv. 1995;13:247–61.PubMedView ArticleGoogle Scholar
- Benson N, Sugiono P, Youderian P. DNA sequence determinants of lambda repressor binding in vivo. Genetics. 1988;118:21–9.PubMed CentralPubMedGoogle Scholar
- Andersen JB, Sternberg C, Poulsen LK, Bjørn SP, Givskov M, Molin S. New unstable variants of green fluorescent protein for studies of transient gene expression in bacteria. Appl Environ Microb. 1998;64:2240–6.Google Scholar
- Daniel R, Rubens JR, Sarpeshkar R, Lu TK. Synthetic analog computation in living cells. Nature. 2013;497:619–23.PubMedView ArticleGoogle Scholar
- Kitney R, Freemont P. Synthetic biology - the state of play. FEBS Lett. 2012;586:2029–36.PubMedView ArticleGoogle Scholar
- Tabor J, Salis H, Simpson Z, Chevalier A, Levskaya A, Marcotte E, et al. A synthetic genetic edge detection program. Cell. 2009;137:1272–81.PubMed CentralPubMedView ArticleGoogle Scholar
- Zhang H, Lin M, Shi H, Ji W, Huang L, Zhang X, et al. Programming a Pavlovian-like conditioning circuit in Escherichia coli. Nat Commun. 2014;5:3102.PubMedGoogle Scholar
- Wang JT, Syed A, Gralla JD. Multiple pathways to bypass the enhancer requirement of sigma 54 RNA polymerase: roles for DNA and protein determinants. Proc Natl Acad Sci U S A. 1997;94:9538–43.PubMed CentralPubMedView ArticleGoogle Scholar
- Wang JT, Syed A, Hsieh M, Gralla JD. Converting Escherichia coli RNA polymerase into an enhancer-responsive enzyme: role of an NH2-terminal leucine patch in sigma 54. Science. 1995;270:992–4.PubMedView ArticleGoogle Scholar
- Brophy JAN, Voigt CA. Principles of genetic circuit design. Nat Methods. 2014;11:508–20.PubMed CentralPubMedView ArticleGoogle Scholar
- Slusarczyk AL, Lin A, Weiss R. Foundations for the design and implementation of synthetic genetic circuits. Nat Rev Genet. 2012;13:406–20.PubMedView ArticleGoogle Scholar
- Rhodius VA, Segall-Shapiro TH, Sharon BD, Ghodasara A, Orlova E, Tabakh H, et al. Design of orthogonal genetic switches based on a crosstalk map of rs, anti-rs, and promoters. Mol Syst Biol. 2013;9:702.PubMed CentralPubMedView ArticleGoogle Scholar
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.