Engineering orthogonal dual transcription factors for multi-input synthetic promoters

Abstract

Synthetic biology has seen an explosive growth in the capability of engineering artificial gene circuits from transcription factors (TFs), particularly in bacteria. However, most artificial networks still employ the same core set of TFs (for example LacI, TetR and cI). The TFs mostly function via repression and it is difficult to integrate multiple inputs in promoter logic. Here we present to our knowledge the first set of dual activator-repressor switches for orthogonal logic gates, based on bacteriophage λ cI variants and multi-input promoter architectures. Our toolkit contains 12 TFs, flexibly operating as activators, repressors, dual activator-repressors or dual repressor-repressors, on up to 270 synthetic promoters. To engineer non cross-reacting cI variants, we design a new M13 phagemid-based system for the directed evolution of biomolecules. Because cI is used in so many synthetic biology projects, the new set of variants will easily slot into the existing projects of other groups, greatly expanding current engineering capacities.

Figure 1. Developing a set of orthogonal dual transcription factors for synthetic logic gates.

Flow chart of the selection process and characterization of orthogonal transcription factor-promoter pairs. For the selection of cI variants, a new M13 phagemid-based system was developed (Fig. 2). The selected TFs were characterized, and checked for orthogonality, by a GFP and mCherry dual reporter assay employing bidirectional promoters that integrate both activation and repression activities.

Figure 2. Developing a phagemid-based system for the selection of combinatorial libraries.

(a) Scheme of the phage-assisted selection system. E. coli cells containing the modified M13 helper phage (HP; contains all phage genes except genes III and VI) and an accessory plasmid (AP; containing a conditional gene VI expression circuit, dependent on variant cI activity) are infected with selection phages encoding a combinatorial cI library member on phagemids (PM; contain the variable cI genes and gene III). After infection, a protein with desired characteristics leads to an upregulated gene VI expression and therefore increased phage production. In this way, a protein with desired properties can be selected after several rounds of reinfection. (b) Testing for phage infection resistance from gene III or gene VI expression under the constitutive λ P_RM promoter. TG1 cells expressing no phage genes are used as a benchmark for maximum infection potential. A t-test was performed to test significance (***=P value <0.001; n.s.= not significant: P value >0.05). (c) Enrichment assays of λ cI_opt from mixed phage populations, diluted 1:10³ or 1:10⁶ with excess of RFP-expressing phagemid. Enrichment of cI_opt was analysed by calculating the ratio of white (cI_opt) to red (RFP) colonies on agar plates. In all experiments, phage encoding cI_opt were fully enriched after the selection process. Error bars are 1 s.d.

Figure 3. Design and characterization of engineered promoters with no activation by wild-type (WT) cI.

(a) Synthetic promoters were designed by making symmetric variants of the consensus sequence (CS) based on the six natural λ operators. Each operator contains a consensus (1–8) and a non-consensus half-site (8–1), which were modified for each synthetic promoter. The engineered promoters were named after the position of the base substitution in the consensus half-site. As an example, the promoter 5C6A contains two base pair mutations at positions 5 and 6 (overlined). The −35 and −10 regions are underlined for each bidirectional promoter. (b) Scheme of the λ cI binding assay with GFP and mCherry as reporter proteins. Binding of WT cI to the WT promoter (P_R/P_RM) results in activation of GFP and simultaneous repression of mCherry. (c) Experimental results of the binding assay. GFP and mCherry were normalized to OD₆₀₀ and the ratio was calculated. The lack of binding of λ cI and cI_opt against synthetic promoters was confirmed by the reporter assay. Error bars are 1 s.d.

Figure 4. Selection and characterization of cI library members.

(a) Sequencing results of selected cI variants against synthetic promoters while counterselecting against wild-type binding. Wild-type amino acids are highlighted in blue, randomized positions are underlined and evolved amino acids that were not part of the combinatorial library are annotated with an asterisk. (b) The selected cI variants were re-engineered by rational design (35 S, 38 D, 39 K, highlighted in blue) in order to obtain activators with varying strengths. (c) Fold-activation of engineered λ P_M promoters by selected cI variants. Stronger activation domain mutants are denoted by a ‘_P' (for example cI_5C6A,P). (d) Repression of engineered λ P promoters (0.0=100% repression). GFP and mCherry expression was normalized to OD₆₀₀ and data were obtained from four replicates. Activation and repression were normalized to the basal expression of each promoter in the absence of a TF.

Figure 5. Construction and characterization of gene circuits.

(a) Design of a 2-input gene network. Two sensors act on an integrating circuit with two cI variants operating on a bidirectional promoter and two reporter genes. (b) Experimental data for the 2-input system illustrating the concentration-dependent response of GFP and mCherry. (c) Design of a 3-input network. This network consists of three sensors, an integrated circuit with three cI variants operating on two unidirectional promoters, and two reporter genes. (d) Complex logic function for the 3-input system showing the concentration-dependent correlation between inducers (Ara, IPTG, 3OC6-HSL) and output signals. The maximum GFP output is achieved by IPTG only (cI_5C6A activates GFP expression) and the minimum GFP output by arabinose only (cI represses GFP expression). In an analogous manner, the maximum mCherry output is obtained by 3OC6-HSL (cI_5G6G,P activates mCherry expression) and the minimum mCherry output is with Ara only (cI represses mCherry expression). As expected, combinations of the inducers resulted in intermediate GFP and mCherry expression levels, in an inducer-dependent manner. The three inducer concentrations used for the on state (1) were 0.1% Ara, 0.01 mM IPTG and 1.0 μM 3OC6-HSL. All data represent the average of four replicates and error bars correspond to 1 s.d. between the measurements.