Insulated transcriptional elements enable precise design of genetic circuits

Rational engineering of biological systems is often complicated by the complex but unwanted interactions between cellular components at multiple levels. Here we address this issue at the level of prokaryotic transcription by insulating minimal promoters and operators to prevent their interaction and enable the biophysical modeling of synthetic transcription without free parameters. This approach allows genetic circuit design with extraordinary precision and diversity, and consequently simplifies the design-build-test-learn cycle of circuit engineering to a mix-and-match workflow. As a demonstration, combinatorial promoters encoding NOT-gate functions were designed from scratch with mean errors of <1.5-fold and a success rate of >96% using our insulated transcription elements. Furthermore, four-node transcriptional networks with incoherent feed-forward loops that execute stripe-forming functions were obtained without any trial-and-error work. This insulation-based engineering strategy improves the resolution of genetic circuit technology and provides a simple approach for designing genetic circuits for systems and synthetic biology.


Supplementary Figure 1. Segment-by-segment saturation mutagenesis of P ECF11 on a medium-copy-number backbone.
Each dot represents a randomly selected mutant containing mutations in the corresponding sequence segment. The promoter activity was measured using superfolder GFP as the reporter and quantified as the arithmetic mean of flow cytometry fluorescence data. CV, coefficient of variation; FV, relative (-fold) variation determined by the ratio of maximal promoter activity to the minimal observed value. Data represent the averages of at least three replicate experiments conducted on different days.

Supplementary Figure 2. Genetic Refinement of Additional ECF and T7-Family Promoters.
(a) Effect of random mutagenesis on the promoter activity of P T3 . (b) Effect of random mutagenesis on the promoter activity of P MmP1 . (c) Effect of random mutagenesis on the promoter activity of P gh-1 . (d) Effect of random mutagenesis on the promoter activity of P ECF16 . (e) Effect of random mutagenesis on the promoter activity of P ECF20 . Each dot represents a randomly selected mutant containing mutations in the corresponding sequence segment. The promoter activity was measured using superfolder GFP as the reporter and quantified as the arithmetic mean of flow cytometry fluorescence data. CV, coefficient of variation; FV, relative (-fold) variation determined by the ratio of maximal promoter activity to the minimal observed value. Data represent the averages of at least three replicate experiments conducted on different days.

Supplementary Figure 3. Effects of activator expressions on the growth rate of E.coli. (a) Effect of T7 RNAP and its
sfGFP reporter. (b) Effect of σ ECF11 and sfGFP reporter. The relative growth rate was obtained by extracting the "cell event number per ten seconds" from the flow cytometry data of each measurement. Data represent the means ± SD from at least three replicate experiments conducted on different days.

Supplementary Figure 4. Experimental Measurements and Parameter
Fitting of Response Functions for cI434 acting on P T3 (a) and P gh-1 (b). Solid lines represent the data of parameter fitting using a non-equilibrium correction term. Open circuits represent experimental measurements. δ R values for each response function obtained from model fitting are given.
The promoter activity was calculated as the arithmetic mean of flow cytometry fluorescence data using sfGFP as the reporter. Data represent the means ± SD from at least three replicate experiments conducted on different days.

Supplementary Figure 5. Raw Data and Model Fitting for T7-RNAP Promoter Cores and 32 Repressor-Operator
Pairs. (a) Transcriptional activity of T7-RNAP and ECF11 promoter cores. Promoter cores described in Fig. 4 are highlighted in red and those used for the fitting of repression-dependent parameters are highlighted in blue. The corresponding sequences of the promoter cores are listed in Supplementary Table 3. (b) Experimental measurements and parameter fitting of response functions for 32 repressor-operator pairs. The response functions for the four operators targeted by cI434 are shown in Fig. 3. Open circles and solid lines represent experimental data and model fitting results, respectively. Data represent the means ± SD from at least three replicate experiments conducted on different days. Figure 6. Experimentally measured and predicted response functions of 107 Combinational Promoters Selected for Genetic Implementation. Boxed in red are the combinational promoter designs using cymR as the repressor which were selected as a group. Others were selected randomly. Mean relative (-fold) errors for each response function and the corresponding promoter design are given. The promoter activity was measured using superfolder GFP as the reporter and quantified as the arithmetic mean of flow cytometry fluorescence data. Data represent the means ± SD from at least three replicate experiments conducted on different days. Figure 7. Parameterization of P TAC Promoter Variants. (a) Response functions of P TAC promoter variants using IPTG concentrations as the input. Open circles represent experimental data. The solid curves represent the fitting results according to the biophysical model in (c). Data were obtained by measuring the flow cytometric fluorescence of cells harboring pRG-sfGFP under the control of wild-type and mutant P TAC promoters. Error bars represent the standard deviations from at least three biological replicates. (b) Sequence alignment of P TAC promoter variants. Mutations are marked in red. (c) The biophysical model used to fit the response functions of P TAC promoter variants. K I and n I represent the dissociation constants for IPTG and LacI, respectively; α and β denote the maximal and basal promoter activity.

Supplementary
respectively; K lac is the constant for LacI binding to lacO. (d) Parameter database of P TAC promoters. The output of a P TAC promoter carried on p15A-AmpR (pRG plasmid) is the input of transcriptional repression and the output of a P TAC promoter on the chromosome is the input of transcriptional activation.  3-cI434 . NE+, with non-equilibrium correction term; NE-, without non-equilibrium term. MFE, mean relative (-fold) error. A secondary increase of GFP expression attributed to the non-equilibrium term is indicated. The output, GFP fluorescence, was measured using flow cytometry and quantified as the arithmetic means of measurements. Data represent the means ± SD from triplicate experiments conducted on different days. Figure 10. Experimental and predicted response functions of re-designed IFFL networks. The wild-type T7 promoter of the highest-fitness IFFL network was replaced by five new promoter cores. (a) Schematic representation of network link assignment, design space and computation task. (b) Experimental and predicted response functions of IFFL networks using P T7M1 , P T7M3 , P T7M13 , P T7M14 and P T7M16 as the promoter cores instead of P T7wt . The dashed line indicates the peak position of the highest-fitness IFFL circuit in Fig. 6f. Data represent the means ± SD from three replicate experiments conducted on different days. Figure 11. Emergent Transcriptional Activity in Two Failed Combinatorial Promoter Designs. (a) Spontaneous (emergent) transcriptional activity of promoter cores, operators and their combinations in the absence of T7 RNAP. (b) Experimental and predicted response functions for two failed promoter designs before and after integrating the contribution of emergent promoter activity. MEF was calculated as described in Fig. 4. The promoter activity was calculated as the arithmetic mean of flow cytometry fluorescence data using sfGFP as the reporter. (c) Spontaneous Transcriptional Activity of all 83 Combinational Promoters in the Absence of T7 RNAP. The two combinational promoters shown in (b) are highlighted in green. The promoter activity was measured using superfolder GFP as the reporter and quantified as the arithmetic mean of flow cytometry fluorescence data. Data represent the means ± SD from at least three replicate experiments conducted on different days. Figure 12. Plasmid Architecture of pPT and pRG. The vast majority of the plasmids used in this study were derived from two basic vectors: pPT and pRG. We named the derived plasmids according to the schemes "pPT-XYZ" and "pRG-XYZ", whereby XYZ denotes the sequence of interest by which the lacZα fragment was replaced via Golden Gate Assembly. The crucial parts of the plasmid backbones are labeled with Roman numerals and identical parts are labeled with the same number. The corresponding sequences are summarized in Supplementary Table 4.   Tables 5-7). atgcctccacaccgctcgtcacatcctgAGTTCATGAAACGTGAACTatgctatgCTCATCGACTCACT ATAGGGGAGTTCATGAAACGTGAACTtgccggtcgatc a Operators are indicated by plain capital letters; promoter cores are indicated by bold capital letters. The -35 and -10 regions of putative σ 70 -dependent promoters are underlined. pPT-promoteroperator pRG-repressor P C -RNAP IPTG gradient IFFL pPT-promoteroperator pRG-repressor P TAC -RNAP IPTG gradient a These plasmid specifications were also used for the measurement of promoter core libraries. b In this study, P Sal was used as a constitutive promoter due to its high-level basal transcriptional activity. c P c : constitutive promoter. gaattcttgacggctagctcagtcctaggtacagtgctagcagctgtcaccggatgtgctttccggtc tgatgagtccgtgaggacgaaacagcctctacaaataattttgtttaa P C -T3 RNAP c gaattcttgacggctagctcagtcctaggtatagtgctagcagctgtcaccggatgtgctttccggtc tgatgagtccgtgaggacgaaacagcctctacaaataattttgtttaa a The sequences of RNAP or ECF σ immediately downstream of these sequences are omitted for clarity. The EcoRI sites are shown in bold. The underlined sequences indicate the -35 and -10 regions of the promoters. b The CDS of NahR is shown in capital letters. c The sequence of RiboJ is included. pRG-σ ECF20 a These sequences correspond to the sequences in shown in lower-case letters in Supplementary Table 1.