Expanding the σ⁵⁴-dependent transcription process with orthogonal designs

Abstract

The significance of orthogonal gene expression lies in its ability to ensure consistent and predictable operation of genetic pathways in synthetic biology. In bacteria, σ factors are responsible for promoter recognition, where the recognition pattern of σ⁵⁴ is distinct from that of σ⁷⁰. Moreover, σ⁵⁴-dependent promoters require bacterial enhancer-binding proteins (bEBPs) for transcription initiation, which are stringently regulated and strongly activated. Thus, σ⁵⁴ appears to be a promising candidate for orthogonal designs. In this study, through knowledge-based screening and rewiring of the RpoN box in σ⁵⁴, together with its partnered promoter, we identified three sets of orthogonal expression systems based on σ⁵⁴-R456H, R456Y, and R456L, with different promoter preferences and ideal mutual orthogonality toward each other and the native σ⁵⁴. The orthogonality is transferable, as specific transcription via σ⁵⁴-R456H was demonstrated in three non-model bacteria. When combined with different bEBPs, the system can be employed to control orthogonal downstream output in response to environmental or chemical signals. The orthogonal σ⁵⁴ factors proved to be capable of orthogonalizing complex biological pathways and genetic circuits. Therefore, the orthogonal transcription system will contribute to the expansion of synthetic biology toolkits, thereby providing reliable and diversified gene expression in a wide range of hosts.

Graphical Abstract

Graphical Abstract.

Introduction

Synthetic biology has emerged as a promising approach to program genetic codes of life to endow cells with new functions. By rationally designing and assembling basic biological building blocks, a cell can be engineered to perform sophisticated tasks following a pre-determined design, such as the production of chemicals [1–3], biocomputing and information storage [4, 5], and intelligent drug development [6, 7]. In synthetic biology practices, reliable transcription is required for smooth and robust running of genetic circuits, where the transcriptional machinery from the chassis acts as the fundamental driving force to power the performance of system outputs. However, the complexity of intrinsic regulation at the transcription level has posed barriers to the fine-tuning of functional modules, as undesired expression patterns can be frequently observed, sacrificing the predictability of synthetic biological systems.

As a solution to this challenge, the concept of orthogonality has been proposed, which stands for the decoupling of foreign genetic circuits from the host's regulatory networks. A widely applied example is T7 RNA polymerase (T7 RNAP), derived from the T7 bacteriophage, which specifically recognizes its cognate promoters to achieve orthogonal transcription [8]. While T7 RNAP has proven to be suitable for various tasks, its cellular toxicity and the limited availability of promoters/terminators cannot be overlooked, especially in the stable expression of complex genetic circuits in the long term [9]. We thus wondered, whether the native transcription machinery of bacteria could be harnessed for orthogonal rewiring, thereby creating a novel set of endogenous but orthogonal transcription rules with lower toxicity and higher adjustability.

In bacteria, transcription is carried out by RNAP, which consists of a core component and a dissociable σ factor. The core RNAP (E) has five subunits: β, β′, two α, and ω. When the core RNAP is combined with a certain σ factor, the Eσ complex turns into a functional holoenzyme, forming a functional transcription machinery. In the holoenzyme, the σ factor is responsible for the recognition of promoters, conferring RNAP with the capability to initiate transcription at specific sites in DNA. Although the sequence and structure of core RNAP show high conservation in the bacterial kingdom, σ factors vary significantly among different species [10]. Based on their homologous partners in Escherichia coli, σ factors can be broadly classified into two main groups: the major σ⁷⁰ class, which includes the housekeeping σ factor along with several structurally related alternative σ factors, and the σ⁵⁴ class, which includes only one member, designated σ⁵⁴ [11].

Transcription by the σ⁷⁰ class is initiated by the recognition of consensus promoter sequences at the −10 and −35 elements. This process occurs when σ⁷⁰ is incorporated into RNAP, exposing its DNA-binding determinants to guide the formation of a closed promoter complex (RP_C). Shortly after a spontaneous conformational change, RP_C isomerizes to an open promoter complex (RP_O), and mRNA synthesis begins [12]. Another type of σ factor, σ⁵⁴, seems to be more enigmatic, showing complex regulation patterns compared with σ⁷⁰. When σ⁵⁴ in RNAP recognizes its adaptive promoter, a stable RP_C is formed in the promoter region. In contrast to the σ⁷⁰-dependent process, RP_C cannot spontaneously turn into RP_O because of the autoinhibition of the N-terminal part of σ⁵⁴. Thus, the holoenzyme remains inactive and is trapped at the promoter region until a bacterial enhancer-binding protein (bEBP) is recruited to unlock the complex [13]. The bEBPs are usually oligomers with ATPase activity, which can hydrolyze ATP to remodel the architecture of RP_C to form RP_O [14]. This mechanism is somewhat analogous to eukaryotic transcription initiation carried out by RNAP II, where the participation of activator proteins is also essential [15]. This “eukaryotic-like” regulation mechanism gives us space to design novel tools to control gene expression at various steps, notably those expecting low basal leakage or high fold change upon induction [16]. Early research explored the unique advantages of σ⁵⁴-dependent transcription in the construction of logic circuits (e.g. AND gates and NAND gates). The generated biological process showed predictable computational performance in response to bEBP inputs as electronic equivalents [17]. Modular and tunable genetic amplifiers have also been designed based on adjustable functional bEBP inputs, allowing the customization of the induction fold [18]. These σ⁵⁴-based transcriptional tools facilitate the construction of sophisticated genetic circuits and the achievement of refined regulatory control, highlighting the remarkable engineering flexibility of this mechanism.

In the current study, we focused on the recognition process between σ⁵⁴ and its promoters, with the aim of identifying key participants with the potential to perform orthogonal rewiring. By altering the native protein–DNA interaction preferences, we successfully expanded the σ⁵⁴-dependent expression system from one to four. The newly established σ⁵⁴-dependent orthogonal expression systems exhibited excellent mutual orthogonality and strict internal compatibility with the host. This σ⁵⁴-dependent orthogonality is transferable, as demonstrated in Klebsiella oxytoca, Pseudomonas fluorescens, and Sinorhizobium meliloti. Additionally, we found that the bEBP-dependent activation mechanism was preserved in the orthogonal system and could be employed to control orthogonal downstream outputs in response to environmental or chemical signals. Orthogonality can also be extended to other σ⁵⁴-type promoters by modifying the −24 region, and varying strengths can be achieved by adjusting the spacer region of the orthogonal promoter, which favors the application of orthogonal σ⁵⁴ factors in managing complex biological pathways and layered logic gates. In summary, the reliance on bEBP for activation and the availability of multiple orthogonal σ⁵⁴ factors facilitate the harmonious integration of expression precision with flexible tunability, thus broadening the application scenarios of our system.

Materials and methods

Strains and media

Escherichia coli JM109 was used for routine molecular cloning and preliminary screening processes. Klebsiella oxytoca, P. protegens Pf-5, and S. meliloti Sm1021 were used as chassis for evaluating the performance of the orthogonal transcription system. LB medium (10 g/l tryptone, 5 g/l yeast extract, and 10 g/l NaCl) was used to culture E. coli, K. oxytoca, and P. protegens. TY medium (5 g/l tryptone, 3 g/l yeast extract, and 0.795 g/l CaCl₂·2H₂O) was used to culture S. meliloti Sm1021. KPM minimal medium [10.4 g/l Na₂HPO₄, 3.4 g/l KH₂PO₄, 26 mg/l CaCl₂·2H₂O, 30 mg/l MgSO₄, 0.3 mg/l MnSO₄, 36 mg/l ferric citrate, 10 mg/l para-aminobenzoic acid, 5 mg/l biotin, 1 mg/l vitamin B1, 0.05% casamino acids, and 0.8% (w/v) glucose], supplied with 20 mM ammonium sulfate (high nitrogen level, KPM-HN) or 0.1% glutamate (low nitrogen level, KPM-LN) was used for the RcNifA signal response and nitrogenase activity assays. KPM minimal medium was also used to test the function of the sucrose utilization pathway, with glucose substituted with sucrose. Antibiotics were added at the following final concentrations: 75 μg/ml ampicillin, 25 μg/ml kanamycin, 25 μg/ml chloramphenicol, and 10 μg/ml (E. coli) or 30 μg/ml (S. meliloti) gentamycin sulfate (Gm).

Construction of the rpoN knockout strain

The E. coli ΔrpoN was constructed using the λ-red homologous recombination method [19]. The temperature-sensitive plasmid pKD46 carrying red recombinase was transformed into E. coli JM109. After inducing the expression of red recombinase with arabinose, the bacteria were electroporated with a linear DNA fragment consisting of a Gm-resistant gene flanked by 60 bp homologous arms, near the rpoN gene. Subsequently, the cells were spread on Gm plates and cultured overnight at 37°C. The E. coli ΔrpoN strain was screened by PCR and further confirmed by DNA sequencing. The strain was passaged continuously at 42°C for 3 days to remove the temperature-sensitive plasmid pKD46.

Molecular cloning and library construction.

The plasmids used in this study are listed in Supplementary Dataset S1 and were verified by sequencing before use in further experiments. The corresponding expression cassettes were assembled using the Golden Gate assembly. The cscA, cscB, and cscK genes, which encode sucrose utilization pathways, were chemically synthesized by GenScript. Other genes were directly amplified from the genome of the corresponding bacteria, with the BpiI and BsaI restriction sites removed using overlap extension PCR. The constitutive promoter P_bla2 was used to express E. coli-derived σ⁵⁴/σ⁵⁴ mutants. KoNifA and RcNifA were expressed under the P_tet promoter. The gfp and rfp genes were used as reporters to characterize the outputs of different orthogonal systems. Inverse PCR was used to introduce random mutation libraries for rpoN R456/R457 sites and −24 elements of the promoters, with plasmids carrying the corresponding sequences used as templates. To validate the orthogonality in P. protegens and S. meliloti, a pBBR-derived broad-host-range plasmid was used as the final carrier vector [20]. The rpoN genes were expressed from their native promoters, and the corresponding mutations were introduced using inverse PCR. To fit the codon usage of different hosts, nifA from Azotobacter vinelandii expressed from P_tet was used in P. protegens. Similarly, nifA from the S. meliloti genome, which is driven by the P_cat promoter, was used in S. meliloti.

Transformation of P. protegens and S. meliloti

Electroporation was used to transform plasmids into P. protegens Pf-5, as previously described [21]. Overnight cultures (1.5 ml) were collected by centrifugation and washed twice with 300 mM sucrose. The resulting bacterial cells were resuspended in 75 μl of 300 mM sucrose and mixed with 100 ng of plasmid DNA. Subsequently, the mixture was transferred to a 0.1 cm gap sterile electroporation cuvette, and electroporation was performed using a Bio-Rad Gene Pulser X-cell apparatus (Bio-Rad Laboratories, USA). Immediately after electroporation, 1 ml of LB medium was added. After further incubation for 60 min at 30°C, the cells were plated on selective LB plates. Biparental mating was used to deliver plasmids into S. meliloti. The recipient, S. meliloti Sm1021, was grown in TY medium for 40 h at 30°C. The donor E. coli ST18 strain [22] carrying the corresponding plasmids was cultured overnight in LB medium supplemented with 64 μg/ml 5-ALA (5-aminolevulinic acid hydrochloride) at 37°C. Equal amounts of both the donor and recipient were mixed and mated for 12 h on a TY plate containing 5-ALA. The mating mixture was plated on a TY plate containing 30 μg/ml Gm but without 5-ALA when appropriately diluted.

Validation of the extracellular signal responses of KoNifA and RcNifA

To validate the temperature response of KoNifA, single colonies of each recombinant E. coli strain were picked and cultured in liquid LB with continuous shaking at 30, 37, or 42°C for 16 h. Green fluorescence signals were detected using a microplate reader (BioTek Synergy H1). For the chemical response assay for RcNifA, bacterial cells carrying the corresponding constructs were pre-cultured overnight in KPM-HN medium. The bacteria were collected and resuspended in KPM supplemented with ammonium sulfate (high nitrogen level) or glutamate (low nitrogen level) to an OD₆₀₀ of ∼0.1. Oxygen levels were controlled using static (low oxygen levels) or oscillating (high oxygen levels) cultures. The output signals were monitored every 20 min for 24 h using a microplate reader. The maximum ratio of the fluorescence signal to OD₆₀₀ was selected to represent the output for further analyses.

Nitrogenase activity assay

The acetylene reduction assay (ARA), modified from a previous method, was used to measure nitrogenase activity [23]. Briefly, recombinant E. coli JM109 strains were initially grown overnight in KPM-HN medium and then diluted into 2 ml of KPM-LN medium in 25 ml sealed tubes to a final OD₆₀₀ of ∼0.4. Subsequently, 2 ml of C₂H₂ was injected and statically incubated at 30°C for 12–16 h. The gas phase was analyzed using a Shimadzu GC-2014 gas chromatograph. When nitrogenase activity was assayed in combination with the sucrose utilization pathway, glucose in KPM-LN medium was replaced with sucrose. The data presented are mean values based on at least three replicate cultures.

Results

Semi-rational screening for orthogonal σ⁵⁴ factors in E. coli

σ⁵⁴ proteins in bacteria can be divided into three regions based on sequence, structure, and function [24] (Fig. 1A). The three-helical bundle lying at the C-terminus of region III, designated as the RpoN box [25], is responsible for the specific interaction with the −24 element in the promoter [26]. Therefore, we sought to determine whether modifying the sequences in both the RpoN box and −24 element would create new recognition pairs while maintaining the biological activity of both. To achieve this, we focused on the interaction forces between the RpoN box and the −24 element based on the structure of σ⁵⁴ from Aquifex aeolicus (Fig. 1B) [27]. Our analysis revealed that seven residues in the RpoN box are likely to be involved in promoter binding. Specifically, the σ⁵⁴-R455 and σ⁵⁴-R456 residues may play crucial roles in the specific recognition of bases via hydrogen bonds (Fig. 1C). Therefore, we selected these two residues and their associated bases for orthogonal rewiring.

Figure 1.

Mechanisms of σ⁵⁴-dependent transcription initiation. (A) σ⁵⁴ domain composition in E. coli. Region I helps maintain the closed initiation complex by binding to bEBPs and directing the formation of a fork junction at the −12 element in DNA. Region II consists of repeated acidic residues that may function as internal linkers. Region III contains a core binding domain (CBD) and interacts with both RNAP and its promoter DNA. The C-terminal RpoN box domain is indicated in yellow. (B) Schematic representation of the transcription initiation complex of σ⁵⁴-dependent promoters. The RNAP holoenzyme containing σ⁵⁴ interacts with its target promoter with the aid of integration host factor (IHF) and bEBP to form an open complex. (C) Cryo-electron microscopy (EM) structure indicating the interactions between the RpoN box and promoter. As deduced from sequence homology, the RpoN box in E. coli, ranging from A456 to E463, is purple and lies in the major groove of the promoter DNA. PDB ID: 5NSR. (D) Inferred protein–DNA interactions between the RpoN box and promoter. Colored lines denote interaction forces: van der Waals (≤ 4.5 Å, purple), ironic bonds (≤ 4.5 Å, red), and hydrogen bonds (≤ 3.5 Å, green). Rectangles denote nucleotides or amino acids at specific sites.

A screening system based on two libraries was developed to obtain orthogonal pairs (Fig. 2A). The input library used the degenerate codon NNS (N = A/G/C/T, S = G/C) to generate random mutations in σ⁵⁴-R455 and σ⁵⁴-R456 residues. The reporter library selected the nifH promoter (P_KonifH) from K. oxytoca, which is activated by NifA, as the initial parental promoter [28, 29]. The six bases at −24 of P_KonifH were randomized by 6 × N (N = A/G/C/T) to generate a P_KonifH variant pool, with their activities reported by green fluorescent protein (GFP). Additionally, red fluorescent protein (RFP) expressed from wild-type P_KonifH was designed as a negative selection to eliminate the cross-activation of the σ⁵⁴ mutant to the original −24 element (Fig. 2A). When the two libraries were co-transformed into the E. coli ΔrpoN strain, colonies with strong GFP signals and negligible RFP signals may represent new orthogonal pairs. Through screening, the σ⁵⁴-R456H mutant, together with its partner promoter OH-P_KonifH (with CTGGTA changed to ATGAGA at −24), best fitted the output rules for the new orthogonal pair (Fig. 2B). To confirm orthogonality, the transcription initiation efficiencies of σ⁵⁴ and σ⁵⁴-R456H towards P_KonifH and OH-P_KonifH were measured. The results showed that σ⁵⁴-R456H displayed strong and specific activation of OH-P_KonifH (∼118%), but only limited initiation efficiency towards P_KonifH (∼3.8%). In contrast, wild-type σ⁵⁴ showed high activation efficiency towards P_KonifH (∼100.0%), and only minimal cross-activation was observed for OH-P_KonifH (∼1.4%) (Fig. 2C).

Figure 2.

Screening of orthogonal σ⁵⁴. (A) Screening system for obtaining orthogonal σ⁵⁴ and its adaptive promoters. Mixed RpoN box variants were randomly co-transformed with mixed reporter plasmids, the latter containing constitutively expressed NifA, an expression cassette for RFP under the wild-type P_KonifH, and GFP under the P_KonifHmix. (B) Screened orthogonal pair of the σ⁵⁴ mutant (O-σ⁵⁴) and its corresponding promoter (O-P_KonifH). Orthogonality testing was performed on the screened orthogonal pairs. (C) σ⁵⁴ and σ⁵⁴-R456H were co-transformed with P_KonifH-gfp and O-P_KonifH-gfp to test their recognition abilities and specificities. The transcription initiation strength produced by σ⁵⁴ and P_KonifH was assigned as 100%. (D) Reporter system for testing the activation effect of native σ⁵⁴-dependent promoters. The −24, −12, and spacer sequences were replaced with those from native σ⁵⁴-dependent promoters to form a series of hybrid promoters. (E) Transcription initiation strengths of σ⁵⁴ and σ⁵⁴-R456H were tested using native σ⁵⁴-dependent promoters in E. coli. The transcription initiation strength produced by σ⁵⁴ and native promoters was assigned as 100%. All data are presented as the mean of at least two biological replicates.

Next, we tested whether σ⁵⁴-R456H is orthogonal to native σ⁵⁴-dependent promoters in E. coli to reflect the orthogonality towards the host transcriptional machinery. However, the functionality of σ⁵⁴-dependent promoters largely relies on individualized bEBPs, which are difficult to test under the same conditions because of the temporal heterogeneity of their expression and functionality. To tackle this task, we cloned the core sequences from positions −24 to −12 in the promoters and inserted them into a unified promoter backbone derived from P_KonifH, lacking positions −24 to −12 (Fig. 2D). These hybrid promoters can mimic the binding of σ⁵⁴ or σ⁵⁴-R456H in vivo and can be activated simultaneously by NifA. We selected 16 typical σ⁵⁴-dependent promoters from E. coli and tested their responses to σ⁵⁴ and σ⁵⁴-R456H. The results demonstrated that all hybrid promoters could be efficiently activated by σ⁵⁴, whereas most of them were poorly activated by σ⁵⁴-R456H, with relative ratios generally falling below 15%, except for the nac promoter, which was at 37% (Fig. 2E).

In the present screening procedure, we acquired the σ⁵⁴-R456H mutant, which exhibited favorable orthogonality with both the original parental and intrinsic promoters of E. coli. However, because of the vast array of possibilities generated by the combination of two libraries, a relatively low output efficiency could be a main limiting factor. Thus, we decided to proceed with an additional round of screening using specific σ⁵⁴-R456 mutants.

Screening novel mutually orthogonal σ⁵⁴ factors based on σ⁵⁴-R456

To enhance efficiency and reduce the number of permutations during the screening process, we performed a saturated mutation on σ⁵⁴-R456. This was done to eliminate any mutants that were still capable of recognizing the initial promoter P_KonifH (Fig. 3A). Subsequently, we selected seven mutants (σ⁵⁴-R456Y, -R456H, -R456M, -R456L, -R456W, -R456D, and -R456F) that almost completely lost their ability to activate wild-type P_KonifH for further screening of their paired orthogonal promoters. These σ⁵⁴-R456 mutants were separately co-transformed with the reporter library carrying a randomized −24 region of P_KonifH. We observed many colonies with varied GFP signals and negligible RFP signals for each σ⁵⁴-R456 mutant, except for R456D, for which no GFP expression was observed on the plates. 12 colonies were randomly sequenced to confirm the −24 element sequences for each σ⁵⁴-R456 mutant (Supplementary Table S1). Sequence logos showed that, unlike wild-type σ⁵⁴, which prefers GG at loci −23 and −22, σ⁵⁴-R456Y, -R456M, -R456W, and -R456F mutants preferred GC. In contrast, σ⁵⁴-R456H and σ⁵⁴-R456L preferred GA and GT, respectively (Fig. 3B; Supplementary Fig. S1). Because σ⁵⁴-R456Y, -R456M, -R456W, and -R456F mutants shared a similar preference for the paired promoter, there may be cross-recognition among them. Therefore, we chose σ⁵⁴-R456Y, σ⁵⁴-R456L, and σ⁵⁴-R456H, which exhibited distinct promoter preferences, for orthogonal verification. Our results demonstrated that these three σ⁵⁴ mutants displayed highly orthogonal behavior towards each other's partner promoter (Fig. 3C).

Figure 3.

Different promoter recognition preferences were achieved by mutating R456. (A) Recognition effect of the R456 variants towards P_KonifH-gfp. The transcription initiation strength produced by σ⁵⁴ and P_KonifH was assigned as 100%. (B) Sequence logo maps indicating promoter recognition patterns of σ⁵⁴, σ⁵⁴-R456H, σ⁵⁴-R456Y, and σ⁵⁴-R456L. (C) Cross-recognition testing between σ⁵⁴, σ⁵⁴-R456H, σ⁵⁴-R456L, and σ⁵⁴-R456Y with their partner orthogonal promoters. (D) Fluorescence of wild-type E. coli cells carrying the GFP reporter (OY-P_KonifH), mCherry reporter (OL-P_KonifH), and tagBFP (blue fluorescent protein) reporter (OH-P_KonifH) when σ⁵⁴ variants were introduced. (E) Genetic circuits for dual orthogonal σ⁵⁴ inputs where the expression of σ⁵⁴-R456Y was fixed, and the expression of σ⁵⁴-R456L was coupled with the addition of anhydrotetracycline (aTc). The ribosome-binding site in the NifA expression cassette can be tuned to provide high, medium, or low NifA input. (F and G) GFP and mCherry fluorescence signals were output according to the aTc concentration. Error bars represent the mean ± SD of at least three biological replicates.

In addition, a tri-fluorescence reporting system was constructed to test the mutual orthogonality among these σ⁵⁴ mutants. The activation of paired promoters for σ⁵⁴-R456Y, σ⁵⁴-R456L, and σ⁵⁴-R456H was reported by GFP (for OY-P_KonifH), mCherry (for OL-P_KonifH), and tagBFP (blue fluorescent protein) (for OH-P_KonifH), respectively (Fig. 3D). The output fluorescence intensity of each σ⁵⁴ factor was set to 100%. When the orthogonal σ⁵⁴ factors were double or triple inputted, the activation status of the promoters was as anticipated, indicating the excellent performance of the orthogonal σ⁵⁴ factors, even when they were present in the same bacterial cell (Fig. 3D). However, the fluorescence signal decreased for each factor when multiple σ⁵⁴ factors were introduced (Fig. 3D), which may be due to competition for the same pool of host core RNAP or shared NifA. To quantitatively describe the allocation of shared transcriptional resources between the dual orthogonal σ⁵⁴ inputs, we performed a titration experiment with a specific input. By employing a constitutive promoter to maintain the expression of σ⁵⁴-R456Y at a constant level and an inducible promoter to regulate the expression of σ⁵⁴-R456L according to varying concentrations of anhydrotetracycline (aTc), we assessed the dynamic adjustment of shared resources. The co-transformation experiment with adaptive reporters (Fig. 3E) revealed an increase in mCherry expression and a corresponding decrease in GFP expression as the aTc concentration increased (Fig. 3F). This observation was particularly pronounced under lower NifA input, highlighting the competition for shared resources within the orthogonal transcription system (Fig. 3G).

Amino acid substitutions may alter the stability of a protein; therefore, it is necessary to test whether the orthogonal σ⁵⁴ factors carrying R456H, R456Y, or R456L mutations are stable under different expression levels. We transformed them into E. coli ΔrpoN and conducted western blot experiments to estimate their expression under low and high induction conditions (Supplementary Fig. S2A). The results showed that although the expression levels for σ⁵⁴-R456Y and σ⁵⁴-R456L experienced minor declines, σ⁵⁴-R456H showed comparable stability to that of wild-type σ⁵⁴ (Supplementary Fig. S2B).

Cellular toxicity is also an important aspect for evaluating the reliability of orthogonal transcription systems. Thus, we characterized the fitness costs of introducing orthogonal σ⁵⁴ factors. Escherichia coli strains carrying the plasmids coding for isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible σ⁵⁴-R456H, -R456Y, or -R456L were grown in liquid LB medium or on LB plates under different induction concentrations, and their growth curves or colony sizes were quantified. Compared with bacteria carrying the empty vector, we did not observe significant growth defects in the lag and logarithmic phases when the orthogonal σ⁵⁴ factors were expressed at varied levels (Supplementary Fig. S3A–F). Meanwhile, there were no significant changes in colony size, regardless of whether the bacteria carried the orthogonal σ⁵⁴ factors or the empty vector (Supplementary Fig. S4). Collectively, these results indicate that the orthogonal transcription systems pose little threat of cellular toxicity and do not appreciably impair bacterial growth, highlighting their suitability for biotechnological applications requiring stable genetic circuit performance.

Expanding the orthogonal transcription system to other chassis

Based on the high degree of sequence homology, σ⁵⁴-R456 is conserved across various bacterial species (Supplementary Fig. S5A, B). This observation led us to hypothesize that the proposed orthogonality would also be effective in other bacteria. To test this hypothesis, we selected K. oxytoca, P. protegens Pf-5, and S. meliloti Sm1021 for further investigation (Fig. 4A). Our results showed that the OH-P_KonifH promoter could be activated by an additional copy of σ⁵⁴ with the arginine residue corresponding to E. coli σ⁵⁴-R456 replaced by a histidine residue in all three bacteria. In contrast, no output signals were detected when an additional copy of wild-type σ⁵⁴ or an empty vector was introduced into the bacterial cells (Fig. 4B–D). These outcomes are highly encouraging as they indicate that the system possesses broad adaptability, thus greatly expanding its potential range of applications.

Figure 4.

Functional testing of orthogonal promoters in non-model chassis. (A) Workflow indicating orthogonal rewiring of native σ⁵⁴-dependent transcription in K. oxytoca M5a1 (B), P. protegens Pf-5 (C), and S. meliloti Sm1021 (D). NifA was constitutively expressed, whereas host σ⁵⁴ (overexpression) or σ⁵⁴-R456H (or the identical position in E. coli) was introduced into the chassis. GFP or RFP controlled by OH-P_KonifH was used as a reporter. Error bars represent the mean ± SD of at least three biological replicates.

Screening orthogonal promoters with varied initiation strength

Exploiting the diverse nature of promoter types can broaden the utility of our orthogonalization strategy, as naturally occurring σ⁵⁴-dependent promoters may represent viable candidates for engineering orthogonal derivatives by redesigning their −24 element. Therefore, we aimed to extend this orthogonal relationship to other promoters in the study. In addition to P_KonifH, five σ⁵⁴-dependent promoters (P_KonifJ, P_KonifE, P_KonifU, P_KonifB, and P_KonifF) are present in the nif gene cluster of K. oxytoca, which can be activated by NifA. Consequently, these promoters were selected to screen for orthogonal promoter pairs compatible with σ⁵⁴-R456H, utilizing strategies similar to those used for P_KonifH. Through this screening process, we obtained orthogonal promoters with varied strengths for P_KonifH, P_KonifJ, P_KonifU, and P_KonifB, while only one paired promoter for P_KonifE and no paired promoter for P_KonifF was obtained (Fig. 5; Supplementary Table S2).

Figure 5.

Screening of adaptive promoters for σ⁵⁴-R456H. The sequence compositions of the −24 elements in the promoters and their strengths are shown. (A) Functionality and orthogonality testing of screened adaptive promoters based on P_KonifH for σ⁵⁴-R456H. The corresponding transcription initiation strengths activated by σ⁵⁴ (blue) and σ⁵⁴-R456H (red) are shown as bar graphs. The transcription initiation strength produced by σ⁵⁴ and P_KonifH was assigned as 100%. Error bars represent the mean ± SD of at least two biological replicates. (B) Functionality and orthogonality testing of screened adaptive promoters based on P_KonifU for σ⁵⁴-R456H. The corresponding transcription initiation strengths activated by σ⁵⁴ (blue) and σ⁵⁴-R456H (red) are shown as bar graphs. The transcription initiation strength produced by σ⁵⁴-R456H and P_KonifU was assigned as 100%. Error bars represent the mean ± SD of at least two biological replicates.

The specificity of σ⁵⁴ factor–promoter recognition is contingent upon the direct binding of the RpoN box with −24 and −12 elements, while the spacer sequence between the conserved elements can contribute to promoter initiation strength by influencing the process of transcription complex remodeling [30]. We hypothesized that fixing the sequence of the −24 and −12 elements in pre-screened orthogonal promoters and adjusting the spacer sequences between them would generate a series of promoters with varying initiation strengths while maintaining orthogonality. By doing so, we could potentially acquire a range of orthogonal promoters adapted to σ⁵⁴-R456H, covering a wide range of expression levels. We then randomized the spacer sequences of OH-P_KonifH, P_O1-KonifE, P_O3-KonifU, and P_O1-KonifB to evaluate the functionality and orthogonality of the derived promoters. As expected, promoter variants with varying strengths for each promoter were obtained (Fig. 6; Supplementary Table S3), indicating that altering spacer sequences can be a universal strategy for expanding our orthogonal promoter library.

Figure 6.

Adjusting promoter activity by altering spacer sequences. The sequence compositions of the spacer regions in the promoters and their strengths are shown. (A) Functionality and orthogonality testing of spacer-randomized OH-P_KonifH promoters. The corresponding transcription initiation strengths activated by σ⁵⁴ (blue) and σ⁵⁴-R456H (red) are shown as bar graphs. The transcription initiation strength produced by σ⁵⁴ and OH-P_KonifH was assigned as 100%. Error bars represent the mean ± SD of at least three biological replicates. (B) Functionality and orthogonality testing of spacer-randomized P_O3-KonifH promoters. The corresponding transcription initiation strengths activated by σ⁵⁴ (blue) and σ⁵⁴-R456H (red) are shown as bar graphs. The transcription initiation strength produced by σ⁵⁴-R456H and P_O3-KonifH was assigned as 100%. Error bars represent the mean ± SD of at least two biological replicates.

Engineering the orthogonal transcription system to sense extracellular signals

The orthogonal transcription system is based on σ⁵⁴-dependent transcription mechanisms, which involve bEBPs as essential participants. This characteristic provides an extra dimension for the convergent control of gene expression. Naturally occurring bEBPs can sense environmental or chemical signals to regulate their activity as transcriptional activators [24, 31]. This inspired us to utilize them to engineer a more flexible orthogonal transcription system whose function depends on the culture conditions. Thus, transcription behavior can be abstracted as an AND logic gate, with dual input modules responsible for signal processing (bEBP) and orthogonality (orthogonal σ⁵⁴) (Fig. 7A). To achieve signal-dependent orthogonal transcription, we selected KoNifA from K. oxytoca and RcNifA from Rhodobacter capsulatus for demonstration (Fig. 7B).

Figure 7.

Control of the orthogonal transcription system in wild-type E. coli by signal processing of bEBP. (A) Boolean logic for transcription behavior of the orthogonal system. bEBP acts as a signal processing module through protein-level activity control, such as environmental and chemical signals. (B) Regulatory effects of environmental signals (temperature) and chemical signals (oxygen and nitrogen sources) on KoNifA and RcNifA, thereby testing the behavior of downstream transcription. (C) Functionality testing of the transcription behavior that responds to temperature when KoNifA acts as a signal-processing module. (D) Functionality testing of the transcription behavior that responds to the nitrogen source and oxygen when RcNifA acts as a signal-processing module. Error bars represent the mean ± SD of at least four biological replicates.

KoNifA, a collective bEBP for six separate nif operons, can respond to temperature in its native chassis [32]. It exhibited superior activity at temperatures below 30°C, but its performance was poor at ≥ 37°C. By incorporating KoNifA into the orthogonal σ⁵⁴-R456H-based transcription system (Fig. 7C), we observed a clear temperature-dependent transcription of downstream gfp, with optimal expression at 30°C (100%) and significantly lower fluorescence at 37°C (∼20%) or 42°C (∼1%). Notably, in the absence of σ⁵⁴-R456H, no fluorescence was observed regardless of the temperature. These distinct expression patterns indicate the precise role of bEBP in processing environmental signals that regulate downstream orthogonal transcription behavior. Next, we explored the possibility that chemical signals could serve as effective inducers of orthogonal gene transcription. RcNifA responds to both oxygen and nitrogen levels and exhibits maximum activity when both signals are present at low levels [33]. Given its unique chemical signal response feature, we utilized RcNifA as a bEBP and combined it with σ⁵⁴-R456H to create an orthogonal chemical signal-dependent transcription system. Output signals for bacteria carrying this system were detected under different combinations of oxygen and nitrogen levels (Fig. 7D). The results showed that the reporter GFP could be efficiently expressed only when both the oxygen level and nitrogen source were completely depleted. High oxygen levels with low nitrogen sources (∼20%) or high nitrogen sources with low oxygen levels (∼1%) led to the inhibition of downstream orthogonal output. Moreover, the system failed to function in the absence of the orthogonal σ⁵⁴-R456H input.

Considering the direct protein-level sensing capability of NifA, this system may have higher sensitivity and lower latency. It can quickly capture changes in signals, thereby adjusting the orthogonal downstream outputs within minutes. Using the KoNifA-based orthogonal transcription system as a demonstration, bacteria carrying the plasmid were pre-cultured at 30°C to achieve maximum gfp mRNA expression. Upon rapid temperature elevation to 37°C, samples were collected at 5 min intervals. We observed a rapid decrease in gfp levels, which dropped to 82% within 5 min and to 65% within 15 min. Meanwhile, KoNifA mRNA levels scarcely changed, indicating the direct protein-level regulation of bEBP activity for orthogonal output (Supplementary Fig. S6).

These findings suggest that by altering the bEBP component, we can endow the orthogonal transcription system with the ability to sense chemical or environmental signals, thereby providing an additional dimension for controlling the system output. Many naturally occurring or artificial bEBPs have been shown to be regulated by different signals, providing us with various options. For instance, NorR can be regulated by nitric oxide [34], XylR by aromatic compounds [35], and BmoR by alcohols [36]. This characteristic could be particularly useful for dynamically controlling system output in genetic circuit engineering with minute-scale response kinetics.

Functional reconstruction of genetic circuits with an orthogonal expression system

Considering the working mechanism of the orthogonal transcription system, it can be effectively utilized to assemble orthogonal genetic circuits that are specifically designed to function in response to the corresponding orthogonal σ⁵⁴. To achieve this, we utilized an orthogonal expression system to reconstruct the sucrose utilization pathway. This pathway comprises three genes, cscBAK, and endows bacteria with the capacity to use sucrose as the sole carbon source (Fig. 8A) [37]. The cscABK genes were arranged into a single operon and placed under the regulation of partnered promoters for σ⁵⁴-R456H (Fig. 8B). We then co-transformed the orthogonal cscABK pathway with σ⁵⁴-R456H in E. coli cells. The results showed that efficient growth in the minimal medium with sucrose as the sole carbon source was only observed in the presence of σ⁵⁴-R456H (Fig. 8C).

Figure 8.

Heterologous pathway reconstruction and functional demonstration in wild-type E. coli using orthogonal σ⁵⁴ factors. (A) Schematic diagram showing the functions of the three enzymes involved in sucrose utilization. (B) Orthogonal csc gene pathway directed by orthogonal OH-P_KonifH. (C) Growth curves of wild-type E. coli cells carrying orthogonal csc genes, with (red) or without (brown) σ⁵⁴-R456H input. An empty vector without csc genes (gray) was used as a negative control. Error bars represent the mean ± SD of at least four biological replicates. (D) Construction of the nitrogenase biosynthesis pathway using orthogonal promoters for σ⁵⁴-R456H. The characters in the coding sequences are shown in the nif-omitted form. (E) Nitrogenase activity was tested using the acetylene reduction assay, where the orthogonal nif cluster was transformed with or without σ⁵⁴-R456H or carrying an empty vector as a blank control. (F) Demonstration of NAND gate performance based on the combination of orthogonal σ⁵⁴-R456H and LacI repressors. Error bars represent the mean ± SD of at least three biological replicates.

Next, to demonstrate the potential application of the orthogonal expression system in engineering complex genetic pathways, we reconstructed the nitrogen fixation pathway from K. oxytoca in E. coli. The nif gene cluster of K. oxytoca comprises 18 structural genes arranged into six operons: nifHDKTY, nifENX, nifBQ, nifUSVWZM, nifJ, and nifF. The promoters of the nifHDKTY, nifENX, nifBQ, nifUSVWZM, and nifJ operons were replaced with their corresponding orthogonal variants of σ⁵⁴-R456H (Fig. 8D). Additionally, P_O1-KonifH was employed to drive the expression of the nifF operon, as no orthogonal promoter was obtained in the previous screening process. Subsequently, we assembled six orthogonal operons with a constitutively expressed nifA cassette to generate the O-nif system. After co-transformation with σ⁵⁴-R456H, ∼98.67 ± 7.50 nmol/h/OD₆₀₀ acetylene reduction activity was obtained (Fig. 8E). In contrast, virtually no nitrogenase activity was observed in the absence of σ⁵⁴-R456H because the orthogonal promoters could not be activated by the native σ⁵⁴ present in the bacterial genome.

Discussion

The genetic machinery of the host powers the operation of foreign pathways, governing the performance of expected functions from a fundamental basis. Although the aims and tasks differ among individualized applications, stable and robust running of the introduced parts is a universal demand for synthetic biology, in which reliable transcription behavior plays a key role. In this study, we propose and demonstrate a novel strategy for constructing an orthogonal transcription system. By simply introducing a single amino acid mutation in the promoter-binding region of σ⁵⁴, a dramatic change in protein–DNA interaction was achieved, showing totally different promoter recognition preferences. This phenomenon is consistent with the basic knowledge of biochemistry, as hydrogen bonds serve as the main contributor to specific interactions, and the process is strictly coordinated from a co-evolutionary perspective. The library versus library in vivo screening strategy has proven successful in other studies concerning the alteration of protein–protein [38] or protein–DNA/RNA [39, 40] interaction interfaces. In other words, exploring sequence space is likely to yield matched pairs with novel specificities compared with the naturally occurring parental sequences, some of which may exhibit ideal orthogonality with high application value in synthetic biology. In short, our work enriched successful cases of screening strategies by making a first step towards mining the potential of σ factor–promoter interaction for orthogonal pairs, standing as a demonstration to inspire further research on orthogonalizing other regulatory parts, including transcription factors, G protein-coupled receptors, and proteases.

The orthogonal transcription system has several advantages. First, owing to the strict dependency on a certain bEBP, the system can only work upon the dual input of both orthogonal σ⁵⁴ and bEBP, thus forming a natural AND logic gate. This allows for the conditional expression and convergent regulation of downstream pathways and provides a simultaneous adjustment platform for guiding the allocation of cellular resources among various modules. In addition, by combining it with a NOT gate (e.g. a repressor or inhibitor), the system can be expanded to form other common logic gates, such as the NAND gate (Fig. 8E). Second, the compositions of spacer sequences and −24 element sequences have tremendous possibilities, covering a wide range of promoter initiation strengths. The candidate promoter pool provides abundant choices for constructing complex genetic circuits involving the expression of multiple genes, and facilitates the adjustment of expression levels while maintaining orthogonality. Third, the system is highly modularized. For example, the bEBP component can be freely replaced to achieve signal-dependent downstream transcription, providing an extra dimension for controlling cellular resource influx for orthogonal transcription behavior.

Chromatin immunoprecipitation (ChIP)-seq experiments revealed that σ⁵⁴ could also occupy non-promoter regions in E. coli; however, its biological functions remain unclear [41]. Engineering orthogonal σ⁵⁴ factors with altered DNA recognition specificities may inadvertently introduce unintended chromosomal-binding profiles, potentially triggering unforeseen biological consequences. Although no apparent growth defects were observed in our study, transcriptomic (RNA-seq) and chromatin occupancy (ChIP-seq) profiling would be instrumental in delineating what exactly happened in the cells.

We rewired promoter recognition preferences by changing a single amino acid residue in σ⁵⁴, which is 100% conserved among the bacterial kingdom. No amino acid other than R has been discovered at this site, indicating that during the course of evolution, there has never been any other possibility for σ⁵⁴–promoter recognition. However, our experiments showed that other residues at R456 could also maintain perfect promoter-specific recognition effects, along with overall ideal stability and host compatibility. Why has transcription, a fundamental biological behavior, not evolved to exhibit greater diversity? According to the fitness landscape theory, in the early stages when the interaction between σ⁵⁴ and its promoter is initially formed, there may be multiple interaction candidates with lower energy, corresponding to various potential recognition preferences [42, 43]. As evolution continued, this interaction relationship finally landed on the possibility of R456 (σ⁵⁴) and TGGCA (−24 region). After the formation of strains carrying this interaction, the relationship was reinforced to decode beneficial genes from other bacteria during frequent horizontal gene transfer events among bacteria [44, 45]. Although other possibilities were successfully explored in our attempts, the composition of naturally occurring RpoN boxes may work better in balancing protein stability, functionality, and metabolic toxicity, leading to a higher chance of preservation during evolution.

Previous studies have reported other orthogonal transcription strategies based on σ factors, mainly achieved by heterologous members from other hosts [46, 47]. The principle lies in the evolutionarily distant relationships that result in vastly different promoter recognition patterns. Heterologous σ factors from Bacillus subtilis have been shown to work well in building complex orthogonal biosynthetic pathways in E. coli [48]. Moreover, extracytoplasmic function (ECF) σ factors are candidates for achieving orthogonal and predictable transcription [49]. These explorations represent a classic research paradigm that utilizes naturally occurring, but evolutionarily barriered transcriptional toolkits to present orthogonality in bacteria. In contrast, our strategy focuses on the native host-derived σ⁵⁴, which directly alters its protein–DNA recognition preference to achieve orthogonality. This process caused minimal alterations, wherein only individual amino acids and bases were modified, thereby maintaining excellent host fitness levels. Furthermore, the σ⁵⁴-based orthogonal transcription system is rarely affected by anti-σ factors, which typically act on ECF σ factors, and its performance can be flexibly controlled at the bEBP level. Nevertheless, there is an intrinsic limitation for σ-dependent orthogonal systems, as competition among all functional σ factors in a cell cannot be entirely eliminated [50]. This limitation may be mitigated in our system by controlling the expression of σ factors to the extent that it does not affect host growth (Supplementary Fig. S3), while efficient activation of the partnered promoter could be realized by elevating the protein levels of bEBPs. Although it is not feasible to construct fully isolated orthogonal genetic systems that function within a cell, the concept of orthogonality remains an appropriate description for our system. Utilizing this system, we can reduce the dependence of artificial genetic circuits on host resources and acquire a platform for separate and convergent transcriptional controls. In conclusion, we believe that the orthogonal transcription system represents a valuable addition to the orthogonal toolkit of synthetic biology and holds promise as a tool for building more controllable and predictable genetic circuits.

Supplementary data

Supplementary data is available at NAR online.

Conflict of interest

None declared.

Funding

The National Key R&D Program of China [2022YFF1000400]; the National Science Foundation of China [32122006 and 32070052]; Biological Breeding-National Science and Technology Major Project [2024ZD04079]; and Yazhouwan National Laboratory [2310WC01].

Data availability

All data are available in the main text and in the Supplementary data.