Skip to main content
Download PDF

Dynamic feedback regulation for efficient membrane protein production using a small RNA-based genetic circuit in Escherichia coli

Microbial Cell Factories volume 21, Article number: 260 (2022) Cite this article

Abstract

Background

Membrane proteins (MPs) are an important class of molecules with a wide array of cellular functions and are part of many metabolic pathways. Despite their great potential—as therapeutic drug targets or in microbial cell factory optimization—many challenges remain for efficient and functional expression in a host such as Escherichia coli.

Results

A dynamically regulated small RNA-based circuit was developed to counter membrane stress caused by overexpression of different MPs. The best performing small RNAs were able to enhance the maximum specific growth rate with 123%. On culture level, the total MP production was increased two-to three-fold compared to a system without dynamic control. This strategy not only improved cell growth and production of the studied MPs, it also suggested the potential use for countering metabolic burden in general.

Conclusions

A dynamically regulated feedback circuit was developed that can sense metabolic stress caused by, in casu, the overexpression of an MP and responds to it by balancing the metabolic state of the cell and more specifically by downregulating the expression of the MP of interest. This negative feedback mechanism was established by implementing and optimizing simple-to-use genetic control elements based on post-transcriptional regulation: small non-coding RNAs. In addition to membrane-related stress when the MP accumulated in the cytoplasm as aggregates, the sRNA-based feedback control system was still effective for improving cell growth but resulted in a decreased total protein production. This result suggests promiscuity of the MP sensor for more than solely membrane stress.

Background

Industrial biotechnology has become key in mitigating today’s main economic and environmental issues through the sustainable and efficient production of enzymes, pharmaceuticals, chemicals, fuels and foods [1, 2]. To this end, employing microorganisms as microbial cell factories (MCFs) shows great promise. To develop an industrial biotechnology process that is scalable and economically viable, extensive host optimization is crucial. These hosts have naturally evolved to be in complete balance with their variable environment which translates into a vast collection of genes, pathways and complex regulatory mechanisms to tightly control both gene expression and enzyme activity [3, 4]. For this reason, maintenance and expression of heterologous pathways in these wild-type (WT) organisms results in low yields or productivities due to the unnatural load on the host cells, e.g., consumption of cellular resources or metabolites or production of toxic intermediates [5,6,7]. Through metabolic engineering and synthetic biology, both complex native and heterologous biosynthetic pathways can be altered, combined and even expanded to create MCFs which are efficient, productive and robust [8,9,10,11].

Despite these achievements, synthetic biology tools are still lagging behind for the efficient expression of membrane proteins (MPs). MPs take up 50% of the cell envelope, where they fulfil central enzymatic reactions, many of which entailing interesting industrial applications. First, MPs play key roles for the design of therapeutic drugs, e.g., membrane-integrated β-glycosyltransferases are implicated in the transfection mechanism of several pathogens. High levels of MP overexpression are required for structural studies and hence development of novel therapeutic targets. More specifically, MPs are involved in several diseases, e.g., cystic fibrosis, cancer and Alzheimer’s disease. Concerning the COVID-19 pandemic and from an engineering perspective, efficient production of biosensing MPs could be a great asset for nanotechnology-enabled solutions to fight COVID-19 and maybe future pandemics [12]. However, MP expression and optimization is also becoming increasingly important for MCF optimization, e.g., to influence substrate uptake or product transport [13, 14]. Secondly, MPs are crucial pathway enzymes for bacterial fermentation of some interesting pharmaceutical or cosmetic products, e.g., hyaluronic acid [15] or flavonoids [16], for which functional MP expression often remains a daunting task.

MPs greatly rely on the cellular membranes for correct folding, structural integrity and activity [17, 18]. Elevated levels of MPs can cause membrane disruption or aggregation in inclusion bodies which makes the isolation of sufficient quantities complicated. Next, heterologous MP expression results in a heavy load on the membrane transport machinery and hence problems with incorporation in the membrane. In addition, titration of protein synthesis precursors to the expression of heterologous MPs lowers MCF viability. In general, MP production often results in only a small amount of well-folded and functional protein, toxicity to the host and little membrane-incorporated protein per cell [17, 18]. One of the most commonly used expression systems for the production of MPs remains Escherichia coli (E. coli) due to wide knowledge in physiology, fast growth, effective genetic tools and inexpensive culture costs. Several systems that can contribute to resolving part of these MP expression problems are available, e.g., the E. coli BL21 (DE3) strain and derivatives together with T7 promoter-based plasmids leading to an optimal level of mRNA of the target gene [19,20,21], use of a truncated RNase E to stabilise toxicity caused by MP expression [22] or the E. coli K12-derived SuptoxD or SuptoxR strains [23] which co-express potent effectors for suppressing membrane protein-induced toxicity. However, these systems deliver only a static solution for optimizing MP expression. Current strains do not sense the stress caused by MP expression. As a result, overall membrane protein expression is lowered without tackling the problem: MP stress.

In contrast, dynamic solutions, using genetic circuits, allow for a more efficient, robust and controlled strategy with minimal side-effects on the native metabolism. Genetic circuits enable cells to dynamically respond and perform logic functions upon a certain input signal by influencing the expression of one or multiple enzyme(s) on a transcriptional, translational or post-translational level (Fig. 1a). Although many accomplishments have been established using protein-based, transcriptional regulation [24,25,26,27], an attractive alternative are small RNA-based (sRNA) genetic circuits. sRNAs are non-coding RNA-molecules of 50 to 500 nucleotides, highly structured and play a role in the regulatory network of the cell [28]. Most of the sRNAs are directly transcribed from DNA (trans-acting). Others are localized in the 3’ untranslated region (UTR) and are obtained by mRNA processing (cis-acting). By base-pairing with their targets, sRNAs can repress or stimulate translation and transcription elongation and control the stability of transcripts. The use of sRNAs offers many advantages. As sRNAs are not translated into a protein, this results in a faster production, less consumption of cellular resources and lower energy requirement for production. The fast production in combination with a fast degradation rate allows efficient signal propagation, causing an almost linear response curve which is useful for tuning feedback circuits [29, 30]. Additionally, RNA regulators are highly composable, meaning that multiple regulators can be readily combined. Moreover, RNA regulators have relatively simple structures which can, in many cases, be predicted and designed by computational tools [30,31,32].

Fig. 1
figure 1

a General dynamic regulation of membrane stress. 1. The cells produce an IMP of interest upon addition of the inducer. 2. Accumulation of IMPs causes IM stress signals. 3. A synthetic genetic circuit consisting of a controller and actuator senses and acts upon the IM stress. 4. IM stress diminishes which allows further IMP production. b Dynamic regulation of IM stress using a sRNA-based genetic circuit. 1. The cells produce an IMP of interest upon addition of the inducer. 2. Accumulation of IMPs activates the Cpx pathway. 3. The phosphorylated CpxR protein induces the circuit which results in the production of the designed sRNA (cpxQmut). 4. The sRNA represses translation of the IMP. 5. IM stress diminishes and the regulatory systems are downregulated. The amount of sRNA in the cell decreases which allows further IMP production. The cells can now re-enter this cycle. IM inner-membrane, IMP inner-membrane protein, OM outer-membrane, GOI gene of interest, RBS ribosome binding site

Full size image

To develop a dynamic biological circuit which is fast, composable and enhances the yield of well-folded and functional MP expression in E. coli, we sourced and exploited natural regulatory networks and more specifically the envelope stress response regulation mechanisms in E. coli for biological DNA parts which can detect membrane stress (i.e., sensor) and respond immediately (i.e., actuator). Several native E. coli regulatory systems exist to prevent mistargeting of MPs and associated cell lysis [18, 33,34,35]. The CpxAR regulatory pathway, further referred to as the Cpx pathway, is a two-component system in E. coli that senses and acts on a variety of signals coming from the inner membrane (IM) [36,37,38,39,40]. This compelling system was therefore exploited to create a control loop to dynamically control MP expression in view of increasing the yield of well-folded and functional MPs (Fig. 1b). Therefore, the DNA parts of this CpxAR regulatory system (i.e., promoter and sRNA) were evaluated and adapted to build a dynamically regulated negative feedback mechanism. The usefulness was evaluated for the expression of different MPs in E. coli.

Results

A membrane stress sensor: evaluation of promotor PcpxP

The Cpx system consists of the membrane-anchored sensor kinase CpxA, the response regulator CpxR and the periplasmic protein CpxP (Fig. 2). Under IM stress conditions, CpxR is phosphorylated by CpxA and activates the transcription of several genes including the cpxP gene. The transcribed cpxP mRNA is translated to make the CpxP protein and is further processed by RNase E at the 3’UTR to produce cpxQ sRNA. Whereas CpxP acts in the membrane to degrade misfolded IM proteins (IMPs) and regulate CpxA activity at the post-translational level, cpxQ together with the Hfq protein post-transcriptionally reduce the synthesis of IMPs in the cytoplasm [36].

Fig. 2
figure 2

Native regulatory CpxAR two-component system countering IM stress in Escherichia coli (E. coli). Triggered by misfolded IMPs and IM damage, structural changes in the CpxA sensor domain cause the histidine kinase domain located in the cytoplasm to autophosphorylate. CpxA subsequently phosphorylates CpxR, the response regulator, activating it for transcriptional regulation. One of the activated genes encodes CpxP, which inhibits CpxA activation and functions in a negative-feedback loop. In parallel, the cpxP mRNA is further processed by RNase E to form the small RNA cpxQ that in turn post-transcriptionally reduces the synthesis of IMPs in the cytoplasm. IM inner membrane, OM outer membrane, IMP inner membrane proteins

Full size image

The ability to sense membrane stress could provide a first insight into the potential damage caused by interference with the membrane integrity upon (heterologous) MP expression. Furthermore, membrane stress would be an ideal input trigger to dynamically regulate MP expression. Upon IM stress, phosphorylated CpxR (CpxR-P) binds to DNA to regulate transcription of about 70 genes [37, 41]. The activity of the cpxP promoter (PcpxP) was chosen and evaluated as a biosensor for IM-related stress using the red fluorescent reporter protein mKate2 [42] (Fig. 3a). The PcpxP promoter is recognized by the sigma 70 factor, generally known as the major or primary sigma factor, which is essential for general transcription in exponentially growing cells. In addition, three CpxR-P binding sites were identified of which one is located up to five nucleotides directly downstream of the transcription start site (TSS) [43]. Therefore, the promoter—including these five nucleotides—used for this indirect analysis and further experiments will be referred to as PcpxP(+5).

Fig. 3
figure 3

a Principle of the indirect stress sensor using positive and negative control to trigger IM stress. b Relative mKate2 fluorescence of an equal number (OD600 = 0.3, mid-exponential phase) of Escherichia coli (E. coli) DE3 cells (See Additional file 1: Table S2 for more details) producing either NlpE (IMP of E. coli, positive control) or sfGFP (cytoplasmatic superfolder green fluorescent protein, negative control) for several IPTG-concentrations. The fluorescence of NlpE and sfGFP producing DE3 cells at 0 mM IPTG is arbitrarily set to one and fluorescence of other IPTG-concentrations was compared to this value. IM = inner-membrane, IMP = inner-membrane protein, mKate2 = red fluorescent protein

Full size image

Overexpression of the lipoprotein NlpE served as positive control as it is known membrane protein and proven trigger of the Cpx pathway [36, 39, 40, 43, 44]. As negative control, expression of the cytoplasmic fluorescent protein sfGFP [45] was chosen. Both proteins were controlled by the T7 expression system, in order to induce protein synthesis at a specific moment in time. Expression of NlpE lead to a sevenfold increase in mKate2 fluorescence for an IPTG-concentration of 0.05 mM which indicated the ability of the PcpxP(+5) promoter to detect membrane stress (Fig. 3b). In contrast, the negative control, cytoplasmic sfGFP expression, did not result in a significant mKate2 increase (p = 0.316, Fig. 3b and Additional file 1: Table S1 and Fig. S8). In addition, the ideal IPTG-concentration was found to be ranging from 0 to 0.2 mM due to the high levels of metabolic burden and cell lysis above 0.2 mM (Additional file 1: Fig. S1).

The actuator: cpxQ sRNA engineering

As shown in Fig. 2, the Cpx pathway consists of a trans-encoded sRNA, cpxQ, which recognizes designated binding sites in target mRNAs of proteins with predominant periplasmic or IM localization [46, 47]. The region of complementarity between the sRNA and the mRNA of the protein, i.e., the seed region, determines the efficiency of silencing. In E. coli, one of these target mRNAs is nhaB, which translates into a sodium-proton antiporter, for which the cpxQ seed region is fully known [36]. This seed region can bind both to part of the 5’UTR and coding sequence of nhaB (See Additional file 1: Fig. S2 for cpxQ seed regions to nhaB mRNA).

As sRNAs are known to facilitate rapid, reversible, dynamic and efficient signal propagation, the ability to alter these seed regions to bind new and specific targets could be a powerful tool for the purpose of dynamic pathway regulation. In a first step, the known seed region of the native cpxQ sequence (Fig. 4a) was designed to specifically recognize a 5’UTR of interest, i.e., the 5’UTR(T7), which is typically used in protein expression plasmids [48, 49]. The better the complementarity, the better the sRNA binds to this translation initiation region (TIR) and the more stable the sRNA-mRNA complex is, which is reflected by the Gibbs free energy of the sRNA-mRNA complex (ΔG1). Next to the sRNA-mRNA complex, the sRNA stability (reflected by the Gibbs free energy of the secondary structure of the unbound sRNA (ΔG2)) is an important parameter to avoid rapid degradation and successful targeting of the mRNA as the sRNA has to remain able to interact with the RNA-binding protein Hfq [50,51,52,53,54,55,56]. The RNA secondary structure and interaction with 5’UTR(T7) were constructed using RNAcofold [57]. By adding mismatches to the seed region of the cpxQ sRNA (Additional file 1: Table S3), the ΔG1 of the sRNA-mRNA complex was varied to in silico create an engineered cpxQ sRNA (cpxQmut1) with improved binding to the 5’UTR(T7) (Fig. 4a). In addition, the secondary structure of cpxQmut1 closely resembles the native cpxQ structure to ensure sRNA stability due to optimal interaction with Hfq.

Fig. 4
figure 4

a Interaction between part of the native and rationally designed cpxQmut1 (marked green) and the 5’UTR(T7) (marked red). Gibbs free energy (ΔG) of the interaction is depicted. The secondary RNA structure of both the native and rationally designed cpxQmut1 with ΔG-value are also illustrated. b Designs of control plasmid without engineered cpxQ sequence and plasmid with cpxQmut1 sequence controlled by the inducible promoter PBAD. c Bar plots representing the output sfGFP/OD600 for the control plasmid (blue) and the plasmid expressing the sRNA cpxQmut1 (grey) and this for different L-arabinose concentrations. sfGFP/mKate2 values for an equal number (OD600 = 0.4, mid-exponential phase) of Escherichia coli DH10B cells are plotted. All experiments were carried out in replica triplicates (biological variation) and the error bars represent one standard deviation from the mean value. * = p-value < 0.05, *** = p-value < 0.001 obtained by conducting a two-sample t-test between strains expressing the control plasmid and strain expressing the sRNA cpxQmut1 and this for each L-arabinose concentration, araC regulator of the PBAD promoter, mKate2 red fluorescent protein, OD600 optical density measured at 600 nm, pBR322 medium-copy origin of replication (~ 15–20 copies per cell), RBS ribosome binding site, sfGFP superfolder green fluorescent protein, 5’UTR(T7) 5’ untranslated region of T7 expression system, sRNA = small RNA, TIR translation initiation region, ss secondary structure

Full size image

As proof of concept that the engineered sRNA cpxQmut1 can specifically in vivo recognize the 5’UTR(T7), regulating the expression of the fluorescent protein sfGFP [45], a plasmid was introduced in E. coli containing a sfGFP under control of the constitutive promoter P14 [58] and 5’UTR(T7) and the designed cpxQmut1 controlled by the PBAD promoter [59] (Fig. 4b). Constitutive expression of the red fluorescent protein mKate2 [42] was used as a reference to account for extrinsic factors influencing gene expression (such as cell growth). A plasmid without the designed sRNA sequence served as a control (Fig. 4b). It is important to note that for both strains—control and cpxQmut1—the native cpxQ sRNA is still present as being part of the native CpxAR-two component system in E. coli. No knockout strains were considered in this set-up. Fluorescence and OD600 were measured over time for different L-arabinose concentrations. A significant decrease in fluorescence per cell was observed indicating the ability of the engineered sRNA cpxQmut1 to specifically bind to 5’UTR(T7) and prohibit further translation (Fig. 4c and Additional file 1: Fig. S1 and Table S4).

Evaluation of a dynamically regulated sRNA-based genetic circuit

The promoter PcpxP(+5) and the rationally designed sRNA cpxQmut1 were combined in one genetic circuit to achieve membrane stress-driven dynamic regulation: sensing stress resulting from MP overexpression, and subsequently regulating the synthesis of these proteins to reduce their toxicity. To this end, four different designs were constructed and compared in order to aid the cells combat membrane-related stress signals and hence increase MP production (Fig. 5). The four designs were evaluated for the efficient production of a membrane-targeted sfGFP [45]. The use of fluorescence as a quantification of membrane integration has been proven in the past, even for analysis of highly complex transmembrane (TM) proteins [60,61,62]. Here, the fluorescent protein sfGFP was targeted to the membrane using the E. coli transmembrane domain (TMD) of the IM protein SohB [63, 64]. The SohB TMD was selected from a list of different bacterial and human TMDs and signal peptides which were tested for their capacity to guide sfGFP to the E. coli membrane (Additional file 1: Fig. S4). Of note is that both the signal peptide of PelB, commonly used to target proteins to the periplasm of E. coli [49], and the TMD of 17aCYP, commonly used to target proteins like cytochrome P450 enzymes to the IM of E. coli [16], yield cytoplasmatic sfGFP instead of cell membrane targeted sfGFP.

Fig. 5
figure 5

Comparison of four different sRNA-based (inner-membrane) stress circuit designs on growth and protein production, in relation to the original SohB(TMD)-sfGFP expressing Escherichia coli cells as a reference. a Total sfGFP results for overexpression of membrane targeted sfGFP (inducer concentrations of 0.1 and 0.2 mM IPTG). The fluorescence of SohB(TMD)-sfGFP producing Escherichia coli (E. coli) MG1656 DE3 cells (Additional file 1: Table S2) without stress circuit was arbitrarily set to one and fluorescence of the same strains with respectively D1-4 was compared to this value (Timepoint: 10 h, early stationary phase). b Bar plots representing the output sfGFP/OD600 for overexpression of membrane targeted sfGFP (inducer concentrations of 0.1 and 0.2 mM IPTG). sfGFP/OD600 values for an equal number (OD600 = 0.2) of E. coli MG1656 DE3 cells were plotted. c Specific growth rate (µmax) for strains (E. coli MG1656 DE3) expressing Ref and D1-4, respectively (inducer concentrations of 0.1 and 0.2 mM IPTG). Error bars represent standard deviations based on 3 biological replications. For statistical comparison, a one-way ANOVA was performed. * = p-value < 0.05, *** = p-value < 0.001, IPTG isopropyl β-D-1-thiogalactopyranoside, TMD transmembrane domain, UTR untranslated region, SohB inner-membrane protein from E. coli, sfGFP superfolder green fluorescent protein

Full size image

For the first three designs (D1-D3), a dual plasmid system was created comprising a low copy plasmid (pSC101 ori), bearing the cpxQmut1 sRNA expression operon, and a medium copy plasmid (pBR322 ori), expressing SohB(TMD)-sfGFP under control of the promoter PT7 and the 5’UTR(T7) [48, 49], on which the cpxQmut1 sRNA can bind, hindering translation. Design D1 is based on the natural processing of the cpxP mRNA in the CpxAR two-component system in E. coli: upon activation of the stress-sensitive promoter PcpxP, cpxP mRNA is transcribed and further processed by the RNase E to produce cpxQ sRNA [36, 52]. For this, in design D1 the 3’ end of the native cpxP mRNA was altered to match the sequence of cpxQmut1. Design D2 circumvents RNase E cleavage and immediately transcribes cpxQmut1 sRNA upon promoter PcpxP activation. Finally, in design D3 the first five nucleotides of the cpxP mRNA were fused with the cpxQmut1 sRNA. The preservation of these first five nucleotides after the transcription start site in the cpxP operon was found to be important for CpxR-P binding to the third CpxR-P binding site in the PcpxP promoter [43]. All dual plasmid constructs were introduced into E. coli MG1656 DE3 cells (Additional file 1: Table S2). The effect of the three designs (D1-D3) on cell growth and fluorescence was compared to the reference strain that expresses SohB(TMD)-sfGFP in absence of cpxQmut1 sRNA expression (Ref), for two different IPTG-concentrations (Fig. 5a–c). Although no significant differences were noted for maximum specific growth rate, total protein production or protein production per cell, an average increase in final OD600 was observed for an inducer concentration of 0.1 mM (Additional file 1: Fig. S5a and Table S5).

To further optimize this sRNA-based membrane stress circuit design, we investigated the response to an important genetic parameter for sRNAs: the impact of the sRNA-mRNA ratio [65, 66], as an excess of sRNA is often required to saturate binding of the target mRNA. Additionally, tight spatial correlation of the sRNA and its target could be important to increase effective binding due to a decrease in search space of the sRNA for its target [29, 67]. This local concentration burst of sRNA will likely avoid the short-circuiting of other Hfq-mediated regulatory networks as the Hfq chaperone is a limiting factor in the bacterial cell [51]. For this reason, design D3 of the sRNA-based (IM) stress circuit was transferred to the reference plasmid (SohB(TMD)-sfGFP). Application of the stress circuit by the single plasmid design (D4) significantly reduced protein production per cell, indicating the functionality of the circuit (Fig. 5b). Additionally, there was a significant improvement in cell growth and a sixfold increase in total protein production (one-way ANOVA, p = 0.018 and p = 0.001, respectively) in contrast to the double plasmid designs (D1-3) (Fig. 5a, c).

Smart sRNA library for increased membrane protein production

To rationally design a sRNA to specifically bind a target, two important parameters need to be considered: sRNA stability and sRNA-mRNA interaction. Therefore, a library of nine different variants of the native cpxQ was created to find the optimal cpxQ with the best balance between a stable secondary structure (ΔG2) and interaction with the 5’UTR(T7) (ΔG1). To this end, the library is called ‘smart’ as several combinations of both parameters were made, e.g., strong sRNA-mRNA interaction but low sRNA stability or stable sRNA secondary structure but intermediate sRNA-mRNA interaction (Table 1). In this way, the smart library covers a relevant range of both parameters. The sRNAs cpxQmut1, cpxQmut2 and cpxQmut3 were designed to have a secondary structure which closely resembled to the native cpxQ but with higher complementarity to the 5’UTR(T7). Next, for cpxQmut4 the region of complementarity increased, improving the ΔG1 value, although ΔG2 was close to the native cpxQ but with different folding characteristics. Finally, sRNAs cpxQmut5-9 have an expanded region of complementarity to 5’UTR(T7), resulting in decreased ΔG1s, up to − 40.07 kcal/mol [68]. However, for cpxQmut7-8 this resulted in a more unstable secondary structure (Additional file 1: Fig. S6). This smart sRNA library was evaluated for the efficient production of a membrane-targeted sfGFP (SohB(TMD)-sfGFP). The vectors expressing the native cpxQ or the respective cpxQmut variant are similar to design D4 (Fig. 5, one plasmid design) for which cpxQmut1 was replaced by one of the other nine cpxQ(mut) sequences.

Table 1 Gibbs free energies (ΔG) of the sRNA-mRNA complex and secondary structures of the unbound sRNA
Full size table

Again, fluorescent intensity was used as an indication for the amount of protein targeted to the membrane. A strain producing cytoplasmic sfGFP served as reference for fluorescence production without membrane stress. Both OD600, as a measure of cell growth, and green fluorescence, as a measure of protein production, were continuously followed over time (Fig. 6a and Additional file 1: Fig. S7a). Targeting sfGFP to the membrane, without the sRNA-based circuit, significantly reduced cell growth and fitness (Fig. 6a–c and Additional file 1: Table S5). Cytoplasmic sfGFP expression decreased cell growth as well, probably due to metabolic burden issues (µmax of 0.37 ± 0.04 h−1 for cytoplasmic sfGFP expression compared to 0.54 ± 0.02 h−1 for E. coli MG1656 DE3 cells without plasmid). Overall, addition of the stress circuit positively affects cell growth and total protein production (Fig. 6a and Additional file 1: Fig. S7). Overexpression of the native cpxQ sRNA, as a result of the applied membrane stress, did not significantly improve cell growth or protein production, highlighting the orthogonality of the other rationally designed sRNAs for 5’UTR(T7). Rationally designed sRNAs cpxQmut1, 3, 5, 6 and 9 resulted in a significant increase in µmax compared to the µmax of the strain without stress circuit (SohB(TMD)-sfGFP). The same could be concluded from sfGFP intensity measurements (Additional file 1: Fig. S7b). Overall, rationally designed sRNAs cpxQmut1 and 3 performed significantly better compared to the other rationally designed cpxQ sRNAs (ΔG2 of − 17.50 and − 18.50 kcal/mol, respectively). More specifically, the decreased protein production per cell (Fig. 6b) resulted in an improved growth rate and subsequently increased total protein production (Fig. 6d). The balance between sRNA stability and complementarity to 5’UTR(T7) resulted in significantly increased maximum specific growth rates (p = 0.009 and p = 0.001, respectively, Additional file 1: Table S5), total protein production (p = 0.001 and p = 0.001, respectively, Additional file 1: Table S5) and final OD600 values (Fig. 6a). Therefore, these two sRNAs were chosen in further evaluation of the stress circuit for the overexpression of MPs.

Fig. 6
figure 6

Effect of the smart library of sRNA-based stress circuits on cell growth and protein production of a membrane targeted sfGFP (SohB(TMD)-sfGFP). The strain which expressed construct SohB(TMD)-sfGFP without stress circuit, was used as reference (R). The strain with native cpxQ served as negative control (N). Expression of sfGFP in the cytoplasm was also included (Cy). a OD600 measured in time. b sfGFP/OD600-values of the different strains (inducer concentration 0.2 mM IPTG). sfGFP/OD600 values for an equal number (OD600 = 0.25) of E. coli MG1656 DE3 cells were plotted. This ratio was corrected for background fluorescence and optical density of the medium and the cell culture (Escherichia coli MG1656 DE3). c Maximum specific growth rate (µmax) of the different strains. The experiment was carried out in triplicate (biological variation) and the error bars represent one standard deviation from the mean value. For panel b and c, significant differences between strains expressing construct SohB(TMD)-sfGFP and all constructs with native or mutated cpxQ were calculated using one-way ANOVA. d Scatter plot representing fold change of sfGFP in function of the ΔG2 absolute value for the secondary sRNA structure (Timepoint: 10 h, stationary phase). Complete plasmid details can be found in Additional file 1: Table S6. The colormaps represents the ΔG1 absolute value. * = p-value < 0.05, ** = p-value < 0.01, *** = p-value < 0.001, ΔG1 Gibbs free energy for sRNA-mRNA interaction, ΔG2 Gibbs free energy for sRNA stability, sfGFP superfolder green fluorescent protein, SohB(TMD) transmembrane domain of inner-membrane protein SohB from E. coli, TIR translation initiation region

Full size image

Dynamic regulation for the overproduction of membrane proteins

Next, the optimized stress circuits (design D4 with cpxQmut1/mut3) were applied for the overexpression of functional MPs. For this purpose, two bacterial MPs were chosen: GarP and YidC (Additional file 1: Table S7). GarP is a TM transporter, probably of galactarate and D-glucarate. No crystal structure for this protein is yet obtained but homology modelling predicts 12 TMDs with a cytoplasmic loop between TMD6 and 7. Both the N- and C-termini are located in the cytoplasm [69]. The second MP, YidC, is a MP insertase. This protein is inserted with six TMDs in the E. coli IM and contains a large periplasmic loop between TMD1 and 2. YidC plays an important role in the correct folding of MPs, both dependent and independent of the Sec-translocon [70]. Localization of both proteins was checked with the Zeiss LSM 780 confocal scanning light microscope with Airyscan technology. Both proteins were fused to sfGFP to allow visualization. In contrast to what was suggested in literature, GarP overexpression resulted in the aggregation of GarP in the cytoplasm. YidC was predominantly localized in the membrane with some aggregation at the poles (Fig. 7a). In both cases, elongated cells were observed and were able to significantly activate the membrane stress sensor (Additional file 1: Fig. S8 and Table S8). In analogy with E. coli overexpressing membrane-targeted sfGFP, overexpression of GarP and YidC severely hampered cell growth and hence total protein production (Additional file 1: Fig. S9). Co-expression of both MPs with the optimized stress circuit (design D4 with cpxQmut1/mut3) was evaluated. Both OD600 and green fluorescence intensity were measured in time. The TIR (5’UTR(T7)) of the two MPs was the same as for the membrane-targeted sfGFP.

Fig. 7
figure 7

Effect of the sRNA-based stress circuit on protein production of functional MPs expression. a Confocal scanning light microscopy images with intensity cross-section profiles for both the red FM4-64 membrane dye (purple colour) and the green fluorescent protein signal (GFP, green colour). Yellow bars represent the position used to make a cross-section profile. b Specific growth rate (µmax) for strains (Escherichia coli (E. coli) MG1656 DE3, Additional file 1: Table S2) expressing SohB(TMD)-sfGFP, YidC-sfGFP and GarP-sfGFP. c Total protein production results for overexpression of proteins, SohB(TMD)-sfGFP, YidC-sfGFP and GarP-sfGFP, localized in the membrane. The fluorescence of SohB(TMD)-sfGFP, YidC-sfGFP and GarP-sfGFP producing E. coli MG1656 DE3 cells without stress circuit was arbitrarily set to one and fluorescence of the same strains with respectively stress circuits 1 and 3 were compared to this value. d sfGFP/OD600-values for overexpression of proteins, SohB(TMD)-sfGFP, YidC-sfGFP and GarP-sfGFP with and without stress circuit. sfGFP/OD600 results for an equal number (OD600 = 0.25 for GarP-sfGFP and SohB(TMD)-sfGFP, OD600 = 0.18 for YidC-sfGFP) of E. coli MG1656 DE3 cells were plotted. This ratio was corrected for background fluorescence and optical density of the medium and the cell culture (Escherichia coli MG1656 DE3). All experiments were carried out in triplicates (biological variation) and the error bars represent the standard deviation from the mean value. Statistical difference between strains expressing the MP without and with stress circuit 1 or 3, was calculated using a one-way ANOVA. * = p-value < 0.05, ** = p-value < 0.01, *** = p-value < 0.001, IPTG (0.2 mM) was added to induce protein expression. YidC = membrane protein insertase in E. coli, GarP = TM transporter in E. coli, MP = membrane protein, sfGFP = superfolder green fluorescent protein, SohB(TMD) = transmembrane domain of inner-membrane protein SohB from E. coli, Stress circuit 1 (mut1) = strain expressing MP and sRNA cpxQmut1, Stress circuit 3 (mut3) = strain expressing MP and sRNA cpxQmut3

Full size image

For the overexpression of all proteins, growth curves regained the sigmoid shape and a higher final OD600 was obtained (Additional file 1: Fig. S9). These results were further reflected in the µmax-values: application of the stress circuit with sRNAs cpxQmut1 and 3 significantly increased µmax (Fig. 7b and Additional file 1: Table S5) and decreased protein production per cell (Fig. 7d). Additionally, adding the stress circuit to the overexpression of YidC resulted in a significant increase in total production (Fig. 7 c) and Additional file 1: Table S5). Similar to the results obtained with the smart sRNA library (Fig. 6), cpxQmut3 sRNA performed better than cpxQmut1 (Fig. 7b, c). Although, a notable difference was observed depending on the source of stress. Direct GarP visualization (Fig. 7a) indicated cytoplasmic localization in aggregates. Overexpression of the latter protein in combination with the stress circuit, resulted in a significant decrease in protein productivity (Fig. 7c and Additional file 1: Table S5). Finally, the influence of both circuits on protein localization or membrane insertion was visualized with the Airyscan technology. In contrast to the improved cell growth and productivity results, no effect on localization or cell physiology was observed. Cells retained an elongated form and in the case of GarP overexpression, aggregation still occurred (Additional file 1: Fig. S10). To demonstrate the specificity of cpxQmut1 and 3, a non-specific cpxQmut (ΔG1 = − 3.94 kcal/mol for binding with the 5’UTR of interest) was designed. The non-specific cpxQmut sRNA (Additional file 1: Fig. S11a) could not significantly increase µmax (Additional file 1: Fig. S11b) and did not significantly decrease protein production per cell (Additional file 1: Fig. S11c).

Discussion

In this study, a dynamically regulated sRNA-based circuit was developed to counter stress caused by MP production. Due to the many advantages connected to the use of sRNAs in metabolic engineering, we designed and built a dynamic, membrane stress-driven, negative feedback circuit that combined the use of sRNAs with the membrane stress-sensitive promoter PcpxP(+5) [29, 68, 71, 72]. The sRNA cpxQ was designed to specifically bind to a 5’UTR of interest (5’UTR(T7)) which controlled expression of the fluorescent protein sfGFP. A significant decrease in fluorescence per cell was observed, suggesting the ability of our engineered sRNA to specifically bind to the 5’UTR(T7) and prohibit further translation. Three different sRNA-based circuit designs were assessed for their ability to improve cell growth and/or total membrane-targeted sfGFP production. All three tested dual plasmid designs (D1-D3) had an average beneficial effect on cell growth but not on membrane-targeted sfGFP production for 0.1 mM IPTG induction. At a higher level of induction (0.2 mM IPTG), no positive effect on cell growth or protein production was observed for D1-D3. Multiple hypotheses could clarify these observations. First, D1 needs an additional step to fully activate the circuit: cleavage of cpxP mRNA by RNase E to form cpxQmut1 sRNA. This extra step possibly delayed the speed at which the circuit could react to membrane-related stress. It seemed unlikely that RNase E did not recognize the correct cleavage site as this was not altered in the sRNA design process [52]. Additionally, in D1 cpxP mRNA was translated to CpxP which could cause an allocation of resources without any added benefit for the cells. Moreover, CpxP is known for its repressing activity of the Cpx pathway and hence decreases the amount of CpxR-P [73,74,75,76]. Designs D2 and D3 skip this RNase E cleavage step and directly produce the required sRNA. For D2, an incomplete interaction between CpxR-P and the cpxP promoter could have occurred. The extra 5-nucleotide extension between the PcpxP promoter and the sRNA sequence in D3 should allow for an optimal CpxR-P—PcpxP interaction however no significant improvement in cell growth and total protein production was observed [43, 77, 78]. For this reason, a final important parameter was investigated: the impact of the sRNA-mRNA ratio. Tight spatial correlation of the sRNA and its target could be important to increase effective binding due to a decrease in search space of the sRNA for its target and/or by preventing Hfq short-circuiting [79]. Therefore, a single plasmid design (D4) in contrast to the dual plasmid design (D3) of the sRNA-based (membrane) stress circuit was evaluated (Fig. 5). The single plasmid design—in contrast to the dual plasmid design—of the sRNA-based (membrane) stress circuit successfully improved cell growth and membrane-targeted sfGFP production (Fig. 5a–c).

Next, a smart library of different cpxQ sRNAs was designed and evaluated, using D4, for the optimization of membrane-targeted sfGFP production. These cpxQ sRNA variants differed in binding strength to the 5’UTR(T7) (region of complementarity) and secondary structure. The application of the stress circuits resulted in (partly) restored growth rate, cell fitness and increased total protein production (Fig. 6 and Additional file 1: Fig. S7). As protein production per cell did decrease, this indicated the ability of the circuit to dynamically regulate overexpression of the membrane-targeted protein sfGFP (Fig. 6b). However, we want to stress the promiscuity and difficulty for interpreting protein production per cell. More specifically, overexpression of the selected membrane proteins resulted in severe growth rate retardation and final OD600-values and therefore complicating a completely correct comparison of protein production per cell at a pre-set moment on the growth curve, i.e., exponential or stationary phase. After reaching the stationary phase, productivity stopped improving. The PcpxP promoter has been reported to be activated in response to entry into stationary phase and hence, negative feedback is activated and protein production halted [40]. Altogether, similar trends were observed for most of the designed sRNAs: improved cell growth and total protein production. As expected, the non-optimized, native cpxQ sRNA exerted no beneficial effect on both growth and total protein production due to the inability to target our 5’UTR(T7). However, increased expression of the native cpxQ could still influence other E. coli proteins but not to an extent that stress levels were relieved. The best performing cpxQmut sRNAs (cpxQmut1 and 3) had better complementarity to the 5’UTR(T7) and a secondary structure closely resembling the native cpxQ structure (Table 1 and Additional file 1: Fig. S6). In both sRNAs the Rho-independent terminator region, required for interaction with Hfq, is unchanged and available for association with the RNA-chaperone Hfq. The sRNAs cpxQmut5, 6 and 9 also significantly increased cell growth and overall protein yield (p = 0.048, p = 0.046 and p = 0.002, respectively, Additional file 1: Table S5). Even though these three sRNAs have better complementarity to the 5’UTR(T7), e.g., ΔG1 = − 33.90 kcal/mol for cpxQmut9 compared to ΔG1 = − 12.12 kcal/mol for cpxQmut3, the influence on growth and total protein production was less pronounced than for cpxQmut1 and 3 (Fig. 6c and Additional file 1: Fig. S7b). This indicated that the stability of the sRNA plays an important role in silencing. The sRNAs cpxQmut7 and 8 had a long region of complementarity to the 5’UTR(T7), with a ΔG1 of approximately − 40 kcal/mol. Nevertheless, the negative ΔG1-value did not compensate for the low stability of the secondary structure, with ΔG2-values of − 6.40 kcal/mol and − 6.20 kcal/mol, respectively. If the sRNA is immediately degraded, it cannot form a stable complex with the mRNA making the ΔG1-value of no relevance. A balance is needed between the sRNA secondary structure and complementarity to the 5’UTR(T7) in order to significantly improve production of the membrane-targeted protein sfGFP (Fig. 6d).

Finally, the sRNA-based stress circuit was applied for the overexpression of functional MPs. Two E. coli MPs were chosen: GarP and YidC. Microscopic visualization using the Airyscan technology, showed aggregation and membrane localization of GarP and YidC overexpression, respectively. Since GarP is a proven TM protein in E. coli, these aggregates were caused by the saturation of insertion machineries rather than the lack of recognition as a MP. YidC is mainly localized in the membrane with more proteins situated near the poles [80]. According to Urbanus et al. [81], the accumulation at the poles is localized in the membrane and not in the cytoplasm, as is the case with inclusion bodies. Possibly, GarP aggregated more quickly than YidC due to the number of TMDs, namely 12, compared to YidC, which contains six TMDs. The higher hydrophobicity of the protein is likely to increase the chance of hydrophobic interactions, which lead to the formation of aggregates. Without addition of the stress circuit, overexpression of both proteins severely hampered cell growth (Fig. 7b). Addition of the dynamically regulated stress circuit significantly increased cell growth in both cases and decreased protein production per cell (Fig. 7d). However, total protein production was not increased for GarP, even though cell growth was improved (Fig. 7b, c). The decreased GarP protein production suggested that the circuit responded to more than solely membrane stress. Many of the triggers of the Cpx-pathway are still unknown but CpxA-independent activation of CpxR has already been studied. There is a growing body of evidence that cytoplasmic signals such as the presence of certain salts, metals, the metabolic acetate pathway, decreased levels of cyclic AMP, among others, can trigger CpxR phosphorylation independent of CpxA. More specifically, it has been proposed that acetyl phosphate (acetyl-P), the intermediate of the phosphotransacetylase (Pta)-acetatekinase (AckA) pathway, can transfer its phosphoryl group to CpxR [40, 82]. This pathway is a result from the overflow metabolism which occurs in fast growing bacterial cells. At high growth rates, the E. coli cell switches from aerobic respiration to anaerobic fermentation. Despite the availability of oxygen, fermentation allows faster ATP production per unit membrane area and results in a lower proteome cost of energy biogenesis [83, 84]. The Pta-AckA pathway will not only generate ATP but also recycles CoA facilitating the glycolytic flux. In addition, other non-envelope associated stimuli of the Cpx pathway exist which could be correlated to heterologous protein overexpression. In the case of GarP, protein aggregate formation—before and after application of the stress circuit—was observed. GarP contains 12 TMDs probably causing hydrophobic interactions between un- and misfolded GarP proteins which resulted in the formation of protein aggregates in the crowded intracellular environment. The field of protein aggregation and their impact on reproduction and fitness remains a grey zone although recent work suggests that protein aggregation can have an effect on gene expression and cellular resource allocation [85]. One of these cell-survival strategies employs heat-shock proteins (HSPs), e.g., the HSP70 system, which also mediates refolding of unfolded or aggregated proteins by redirecting cellular resources towards several chaperones and proteases, e.g., the DnaKJE or GroESL chaperone systems [86,87,88]. A crucial component in all these responses is the heat-shock response σ-factor (σ32 or RpoH). RpoH is not only implicated in the regulation of several chaperone systems but also in the control of genes that are upregulated upon metabolic burden, e.g., IbpAB [89]. Several of these processes are connected to the CpxAR mechanism. For example, it is known that CpxR-P activates RpoH [77] or that the Cpx system can react to several stresses, e.g., aminoglycoside-caused stress, and this by activating several HSPs such as IbpAB [90]. In conclusion, PcpxP(+5) is not only directly activated by membrane stress but can also indirectly be triggered by other imbalances and stress responses, such as several heat shock responses, caused by the overexpression of membrane proteins. For this reason, the stress source could be an important factor in the level of activation of PcpxP(+5). More in-depth experiments, e.g., RNAseq or microarray analysis, are needed to study the sRNA levels in these cells and understand the reason for productivity decrease.

Although cell fitness did improve, by means of specific growth rate, microscopy results still indicated elongated cells. This shape plasticity is considered a survival strategy to cope with stressful conditions, e.g., DNA damage, antibiotic treatment, or host immune systems [91]. For this reason, we concluded that the dynamic stress circuit temporarily relieves the cell from additional stress which allows the cell to divide more frequently. Consequently, the designed IM stress circuit aims to decrease protein production per cell and hence stress levels which ultimately will allow the cells to grow and divide. As a result, protein numbers per cell decrease but on a culture level total protein production increases. Similar systems exist such as the E. coli BL21(DE3) strain or transcriptional feedback regulation leading to an optimal level of mRNA of the target MP [14, 21]. However, to our knowledge the designed membrane stress circuit is the first reported system using the cpxQ sRNA to dynamically regulate this stress in cell factories. Up to now, it was not possible to completely remove the stressful conditions connected to heterologous overexpression. More fundamental research is needed to understand the interaction of the Cpx pathway with other metabolic pathways in the cell. It is important to note that fundamental in-depth experiments such as RT-qPCR or sRNA-mRNA binding assays can be performed to further study off-target effects and interaction with the target mRNA. Moreover, to conclude general applicability of the stress circuit, expression of additional MPs, e.g., pro- and eukaryotic, are needed.

Finally, the choice of sRNA design can be an important tool to obtain a desirable output. The smart library of sRNAs pointed towards the importance of the sRNA design to obtain a significant effect on cell growth and productivity [89]. A stress source mainly originating from the membrane, was effective in both increasing cell fitness and total protein production and this in contrast to a cytoplasmic stress source which decreased productivity. As discussed, this could be correlated with the responsiveness of the stress-sensitive promoter to multiple triggers and hence a higher level of activation. The promiscuous effect of the promoter combined with the smart library offers the opportunity to design a sRNA which forms a less stable sRNA-mRNA complex which could finally result in higher productivities for the overexpression of almost all kinds of proteins. More specifically, we advise the implementation of the cpxQmut3 variant for optimal and increased membrane protein production. This variant outperforms other cpxQmut variants because of its optimal balance between sRNA-mRNA interaction (ΔG1-value of − 12.12) and sRNA stability (ΔG2-value of − 18.80). However, due to the promiscuity of the stress-sensitive promoter, other variants could be considered if one would opt for dynamic gene regulation of non-membrane proteins.

Conclusions

A dynamically regulated feedback circuit was developed that can sense membrane-related toxicity signals and responds to it by balancing the metabolic state of the cell and more specifically by downregulating the expression of the MP of interest. This negative feedback mechanism was established using a simple-to-use genetic control element based on post-transcriptional regulation: small non-coding RNAs. The seed region of the native sRNA cpxQ was altered and optimized to bind to the 5’UTR upstream of our gene of interest. This strategy significantly increased the maximum specific growth rate and enhanced total protein production two- to three-fold for E. coli strains expressing membrane-targeted proteins. In contrast, if the protein accumulated in the cytoplasm as aggregates, the sRNA-based feedback control was still effective for improving cell growth but resulted in a decreased total protein production. This result suggests promiscuity of the sensor (PcpxP(+5)) for more than solely membrane stress.

Material and methods

Chemical, oligonucleotides and molecular biology

All the products were purchased from Sigma-Aldrich (Diegem, Belgium) unless stated otherwise. Agarose and ethidium bromide were purchased from Thermo Fisher Scientific (Erembodegem, Bel- gium). Standard molecular biology procedures were conducted as described by Sambrook et al. [92]. A variety of polymerases were used for the different types of PCR reactions: PrimeSTAR HS DNA polymerase (Takara, Westburg, Leusden, The Netherlands) and Q5 polymerase (New England Biolabs, County Road, Ipswich, MA, USA) were used for short DNA fragments (< 4 kb). PrimeSTAR GXL polymerase (Takara Bio, France) was used for long DNA fragments (> 4 kb). All DNA fragments were purified using the innuPREP PCR-pure Kit (Analytik Jena AG, Germany). All plasmids were isolated from bacterial cultures using the innuPREP Plasmid Mini Kit (Analytik Jena AG, Germany). Oligonucleotides were purchased from Integrated DNA Technologies (Leuven, Belgium). Sequencing services were conducted by Macrogen (Amsterdam, The Netherlands).

Bacterial strains

E. coli One Shot Top10 Electrocomp™ (Invitrogen, Carlsbad, California, USA) were used for the construction and maintenance of all plasmids. E. coli DH10B was used to allow inducible expression with L-arabinose. For membrane protein expression, E. coli DE3 K12 MG1656ΔendAΔrecA was used to allow inducible expression of the PT7 promoter using isopropyl β-D-1-thiogalactopyranoside (IPTG) as inducer. Expression vectors were introduced into their host cells by electroporation [92]. Bacterial strains that were used in this study were listed in Additional file 1: Table S2.

Media and culture conditions

The culture medium lysogeny broth (LB) was used for cloning purposes. Transformed cultures were plated out on lysogenic broth agar (LBA) and grown overnight at 30 °C. If required, LB and LBA medium were supplemented with appropriate antibiotics (100 μg/mL ampicillin or 50 μg/ml kanamycin). For microscopy and the indirect analysis method, strains were grown in minimal medium glucose (MMGlc). If using DH10B cells, 0.4% casamino acids were added. If required, the culture medium was supplemented with appropriate antibiotics and inducers (0–1% L-arabinose or 0–0.2 mM IPTG). For fluorescence experiments with the smart library of sRNAs, strains were grown in MOPS EZ Rich Defined medium (Teknova) with 100 μg/mL ampicillin and 2% glucose as carbon source. For all fluorescent measurements, independent of medium (EZ Rich or MMGlc) used, precultures were grown overnight on a Compact Digital Microplate Shaker (Thermo Fisher Scientific) at 30 °C and 800 rpm, in 96-well flat bottomed microtiter plates (Greiner Bio-One), in 150 μL LB per well with appropriate antibiotics, enclosed by a Breathe-Easy sticker. Precultures were used for 1/150 dilution in 150 μL of defined medium (EZ Rich or MMGlc) containing a specific concentration of the inducer, and were grown for 24 h at 30 °C and 800 rpm unless stated otherwise.

Plasmid construction

All plasmids for indirect inner-membrane (IM) stress analysis, microscopy (TMD-sfGFP fusion proteins), design evaluation (D1-D4), the smart sRNA library and overproduction of MPs were constructed using a combination of circular polymerase extension cloning (CPEC) and single strand assembly (SSA) [93]. These plasmids will be referred to as production plasmids. Characteristics of the corresponding plasmid backbones are listed in Additional file 1: Table S9. The genes yidC, garP, nlpE, sohB(TMD) and yhcB(TMD) were PCR amplified using purified E. coli MG1655 genome (Genbank accession number U00096) as template. All proteins, used in this study, are listed in Additional file 1: Table S7. The rationally designed sRNA sequences and stress-sensitive promoter (PcpxP(+5)) were ordered as single stranded oligonucleotides from Integrated DNA Technologies. Part of the gene cpxP and PcpxP promoter were PCR amplified using purified E. coli MG1655 genome as template. The cpxP 3’ end, containing cpxQmut1, was added using CPEC. The corresponding plasmids are listed in Additional file 1: Table S6. More detailed plasmid maps are given in Additional file 1: Fig. S12. All DNA sequences (coding, promoter and 5’UTR) are listed in Additional file 1: Table S10.

Plasmids for small RNA engineering, were constructed using cloning plasmids (pCPs) and pBR322-based expression plasmids (pEX-BR322) as described in Coussement et al. [94]. To this end, two sets of pCPs, i.e., pCP(1–2) and pCP(2–3), were generated using CPEC. In short, each rationally designed sRNA sequence (cpxQmut1-3) with inducible PBAD promoter – and AraC for regulation—was cloned in a pCP(1–2) containing two consecutive BsaI cut sites (Golden Gate sites, or GG sites). Besides, a pCP(1–2) plasmid was subcloned with a truncated, non-functional, part of sfgfp. This truncated sfgfp (sfGFP(trunc)) is based on the sfgfp gene devoid of promoter, RBS and start codon [95]. The pCP(2–3) plasmid was subcloned with the PT7 promoter, RBS(T7) and sfgfp gene.

These pCPs were assembled into an expression plasmid (pEX) with a specific origin of replication (here pBR322) using Golden Gate assembly [96]. To use mKate2 expression for normalization of fluorescence data, the proB promoter (PproB) and mKate2 sequence were prior to the Golden Gate assembly subcloned on pEX-BR322 using CPEC. More detailed plasmid maps are given in Additional file 1: Fig. S12. All DNA sequences (coding, promoter and 5’UTR) are listed in Additional file 1: Table S10.

In vivo fluorescence measurements

Strains, as three biological replicates (n = 3 for each IPTG or L-arabinose concentration), were prepared as described above. They were grown for 24 h at 30 °C and fluorescence and OD600 were continuously measured using a Tecan M200 infinite PRO with i-control software version 1.1. For measuring mKate2 fluorescence, excitation and emission wavelengths of 588 and 633 nm were used (gain 100), respectively. For measuring sfGFP fluorescence excitation and emission wavelengths of 488 and 510 nm were used (gain 70), respectively. Optical density was measured at a wavelength of 600 nm.

Confocal light scanning microscopy

For microscopy analysis, Zeiss LSM 780 confocal scanning light microscopy, with Airyscan technology, was used. This service was provided by the Bio-imaging Facility at VIB Ghent. Microscopy was done using μ-slide 8 well coverslips (ibidi) coated with a 0.01% poly-L-lysine (molecular weight of 70–150 kDa) solution. Cell culture of interest was grown in 5 mL LB medium for 18 h and diluted 10 times in phosphate-buffered saline (PBS). Subsequently, 500 μL of this medium was dyed with 1 μL of the red FM4-64 membrane dye (1 mg/mL stock) (Thermo Fisher Scientific). The dyed cell culture was applied on the μ-slide 8 well coverslip and incubated for 30 min at room temperature. Next, the coverslip well was pipetted until dry and 2–3 droplets of mounting medium (1% n-propyl-gallate in glycerol) were added. The image processing package Fiji v1.8.0172, a distribution of ImageJ, was used to compose intensity profiles.

Smart sRNA library construction

The RNA secondary structure and associated ΔG2-value was calculated using RNAfold. The interaction of the designed sRNA with 5’UTR(T7) and associated ΔG1-value was calculated using RNAcofold [57].

Data and statistical analysis

Data were analysed using pandas (www.pandas.pydata.org) unless stated otherwise. Maximal growth rates were determined by plotting \(\mathrm{ln}\frac{{({OD}_{600})}_{t}}{{({OD}_{600})}_{{t}_{0}}}\) values as a function of time and fitting Richards growth model. Pairwise comparisons between different strains were done by a two-sided t-test using the scipy.stats package in Python. One-way ANOVA was performed using the scipy.stats package in Python. Linear regression was performed using the statsmodels.anova package in Python. In all cases, a significance level of 0.05 was applied.

For fluorescence measurements, defined medium (MM or MOPS EZ Rich medium) without cell culture was used to correct for background fluorescence and optical density of the medium (FPmed and ODmed, respectively) for each imposed inducer concentration. To account for background fluorescence and optical density (FPBlank and ODBlank, respectively) of the cell culture, E. coli TOP10, DE3 or DH10B cells without plasmid, were used. Thus, the corrected fluorescence, normalized for optical density, was calculated as follows:

$$\left( {\frac{FP}{{OD}}} \right)_{cor} = \frac{{FP - FP_{med} }}{{OD - OD_{med} }} - \frac{{FP_{Blank} - FP_{med} }}{{OD_{Blank} - OD_{med} }}$$

Availability of data and materials

All data generated or analyzed during this study are included in this published article [and its Additional file 1].

References

  1. Pontrelli S, Chiu TY, Lan EI, Chen FYH, Chang P, Liao JC. Escherichia coli as a host for metabolic engineering. Metab Eng. 2018;50:16–46.

    Article  CAS  Google Scholar 

  2. Keasling JD. Manufacturing molecules through metabolic engineering. Science. 2010;330:1355–8.

    Article  CAS  Google Scholar 

  3. Mu Q, Zhang S, Mao X, Tao Y, Yu B. Highly efficient production of L-homoserine in Escherichia coli by engineering a redox balance route. Metab Eng. 2021;67:321–9.

    Article  CAS  Google Scholar 

  4. Zhu M, Dai X. Growth suppression by altered (p)ppGpp levels results from non-optimal resource allocation in Escherichia coli. Nucleic Acids Res. 2019;47:4684–93.

    Article  CAS  Google Scholar 

  5. Pasini M, Fernández-Castané A, Jaramillo A, de Mas C, Caminal G, Ferrer P. Using promoter libraries to reduce metabolic burden due to plasmid-encoded proteins in recombinant Escherichia coli. N Biotechnol. 2016;33:78–90.

    Article  CAS  Google Scholar 

  6. Tyo KEJ, Ajikumar PK, Stephanopoulos G. Stabilized gene duplication enables long-term selection-free heterologous pathway expression. Nat Biotechnol. 2009;27:760–5.

    Article  CAS  Google Scholar 

  7. Alice G, Roberto DB, Francesca C. Experimental tools to reduce the burden of bacterial synthetic biology. Curr Opin Syst Biol. 2021. https://doi.org/10.1016/j.coisb.2021.100393.

    Article  Google Scholar 

  8. Kim SG, Noh MH, Lim HG, Jang S, Jang S, Koffas MAG, et al. Molecular parts and genetic circuits for metabolic engineering of microorganisms. FEMS Microbiol Lett. 2018;365:1–10.

    Article  Google Scholar 

  9. Brophy JAN, Voigt CA. Principles of genetic circuit design. Nat Methods. 2014;11:508–20.

    Article  CAS  Google Scholar 

  10. Shen X, Wang J, Li C, Yuan Q, Yan Y. Dynamic gene expression engineering as a tool in pathway engineering. Curr Opin Biotechnol. 2019;59:122–9.

    Article  CAS  Google Scholar 

  11. Biggs BW, De Paepe B, Santos CNS, De Mey M, Kumaran AP. Multivariate modular metabolic engineering for pathway and strain optimization. Curr Opin Biotechnol. 2014;29:156–62.

    Article  CAS  Google Scholar 

  12. Jones GW, Monopoli M, Campagnolo L, Pietroiusti A. No small matter: a perspective on nanotechnology-enabled solutions to fight COVID-19. Nanomedicine. 2020. https://doi.org/10.2217/nnm-2020-0286.

    Article  Google Scholar 

  13. Chubukov V, Mukhopadhyay A, Petzold CJ, Keasling JD, Martín HG. Synthetic and systems biology for microbial production of commodity chemicals. Syst Biol Appl. 2016;2(1):1.

    Google Scholar 

  14. Boyarskiy S, Davis López S, Kong N, Tullman-Ercek D. Transcriptional feedback regulation of efflux protein expression for increased tolerance to and production of n-butanol. Metab Eng. 2016;33:130–7.

    Article  CAS  Google Scholar 

  15. Gunasekaran V, Gowdhaman D, Ponnusami V. Role of membrane proteins in bacterial synthesis of hyaluronic acid and their potential in industrial production. Int J Biol Macromol. 2020;164:1916–26.

    Article  CAS  Google Scholar 

  16. Zelasko S, Palaria A, Das A. Optimizations to achieve high-level expression of cytochrome P450 proteins using Escherichia coli expression systems. Protein Expr Purif. 2013;92:77–87.

    Article  CAS  Google Scholar 

  17. Wagner S, Bader ML, Drew D, De Gier J. Rationalizing membrane protein overexpression. Trends Biotechnol. 2006. https://doi.org/10.1016/j.tibtech.2006.06.008.

    Article  Google Scholar 

  18. Mathieu K, Javed W, Vallet S, Lesterlin C, Candusso MP, Ding F, et al. Functionality of membrane proteins overexpressed and purified from E. coli is highly dependent upon the strain. Sci Rep. 2019;9:1–15.

    Article  CAS  Google Scholar 

  19. Miroux B, Walker JE. Overproduction of proteins in Escherichia coli: mutant hosts that allow synthesis of some membrane proteins and globular proteins at high levels. J Mol Biol. 1996;19:71–89.

    Google Scholar 

  20. Wagner S, Klepsch MM, Schlegel S, Appel A, Draheim R, Tarry M, et al. Tuning Escherichia coli for membrane protein overexpression. PNAS. 2008;105:14371–6.

    Article  CAS  Google Scholar 

  21. Schlegel S, Löfblom J, Lee C, Hjelm A, Klepsch M, Strous M, et al. Optimizing membrane protein overexpression in the Escherichia coli strain Lemo21(DE3). J Mol Biol. 2012;423:648–59.

    Article  CAS  Google Scholar 

  22. Heyde SAH, Nørholm MHH. Tailoring the evolution of BL21(DE3) uncovers a key role for RNA stability in gene expression toxicity. Commun Biol. 2021;21:1–9.

    Google Scholar 

  23. Gialama D, Kostelidou K, Michou M, Delivoria DC, Kolisis FN, Skretas G. Development of Escherichia coli strains that withstand membrane protein-induced toxicity and achieve high-level recombinant membrane protein production. ACS Synth Biol. 2016;6:284–300.

    Article  Google Scholar 

  24. De Paepe B, Maertens J, Vanholme B, De Mey M. Modularization and response curve engineering of a naringenin-responsive transcriptional biosensor. ACS Synth Biol. 2018;7:1303–14.

    Article  Google Scholar 

  25. Hussein R, Lim HN. Direct comparison of small RNA and transcription factor signaling. Nucleic Acids Res. 2012;40:7269–79.

    Article  CAS  Google Scholar 

  26. De Paepe B, Maertens J, Vanholme B, De Mey M. Chimeric LysR-type transcriptional biosensors for customizing ligand specificity profiles toward flavonoids. ACS Synth Biol. 2019;8:318–31.

    Article  Google Scholar 

  27. Bervoets I, Van Brempt M, Van Nerom K, Van Hove B, Maertens J, De Mey M, et al. A sigma factor toolbox for orthogonal gene expression in Escherichia coli. Nucleic Acids Res. 2018;46:2133–44.

    Article  CAS  Google Scholar 

  28. Dutta T, Srivastava S. Small RNA-mediated regulation in bacteria: a growing palette of diverse mechanisms. Gene. 2018;656:60–72.

    Article  CAS  Google Scholar 

  29. Kelly CL, Harris AWK, Steel H, Hancock EJ, Heap JT, Papachristodoulou A. Synthetic negative feedback circuits using engineered small RNAs. Nucleic Acids Res. 2018;46:9875–89.

    Article  CAS  Google Scholar 

  30. Peters G, Coussement P, Maertens J, Lammertyn J, De Mey M. Putting RNA to work: Translating RNA fundamentals into biotechnological engineering practice. Biotechnol Adv. 2015;33:1829–44.

    Article  CAS  Google Scholar 

  31. Zadeh JN, Steenberg CD, Bios JS, Wolfe BR, Pierce MB, Khan AR, et al. NUPACK: analysis and design of nucleic acid systems. J Comput Chem. 2011;32:170–3.

    Article  CAS  Google Scholar 

  32. Zuker M. Mfold web server for nucleic acid folding and hybridization prediction. Nucleic Acids Res. 2003;31:3406–15.

    Article  CAS  Google Scholar 

  33. Grabowicz M, Koren D, Silhavy TJ. The cpxQ sRNA negatively regulates Skp to prevent mistargeting of β-barrel outer membrane proteins into the cytoplasmic membrane. Am Soc Microbiol. 2016;7:1–8.

    Google Scholar 

  34. Mitchell AM, Silhavy TJ. Envelope stress responses: balancing damage repair and toxicity. Nat Rev Microbiol. 2019;17:417–28.

    Article  CAS  Google Scholar 

  35. Hews CL, Cho T, Rowley G, Raivio TL. Maintaining integrity under stress: Envelope stress response regulation of pathogenesis in Gram-negative bacteria. Front Cell Infect Microbiol. 2019;9:1–25.

    Article  Google Scholar 

  36. Chao Y, Vogel J. A 3’ UTR-derived small RNA provides the regulatory noncoding arm of the inner membrane stress response. Mol Cell. 2016;61:352–63.

    Article  CAS  Google Scholar 

  37. Vogt SL, Raivio TL. Just scratching the surface: an expanding view of the Cpx envelope stress response. FEMS Microbiol Lett. 2012;326:2–11.

    Article  CAS  Google Scholar 

  38. Delhaye A, Collet J. Fine-tuning of the Cpx envelope stress response is required for cell wall homeostasis in Escherichia coli. Am Soc Microbiol. 2016;7:1–10.

    Google Scholar 

  39. Delhaye A, Laloux G, Collet J. The lipoprotein NlpE is a Cpx sensor that serves as a sentinel for protein sorting and folding defects in the Escherichia coli envelope. J Bacteriol. 2019. https://doi.org/10.1128/JB.00611-18.

    Article  Google Scholar 

  40. Wolfe AJ, Parikh N, Lima BP, Zemaitaitis B. Signal integration by the two-component signal transduction response regulator CpxR. J Bacteriol. 2008;190:2314–22.

    Article  CAS  Google Scholar 

  41. Keseler IM, Mackie A, Santos-Zavaleta A, Billington R, Bonavides-Martínez C, Caspi R, et al. The EcoCyc database: Reflecting new knowledge about Escherichia coli K-12. Nucleic Acids Res. 2017;45:D543–50.

    Article  CAS  Google Scholar 

  42. Shcherbo D, Merzlyak EM, Chepurnykh TV, Fradkov AF, Ermakova GV, Solovieva E, a, et al. Bright far-red fluorescent protein for whole-body imaging. Nat Methods. 2007;4:741–6.

    Article  CAS  Google Scholar 

  43. Yamamoto K, Ishihama A. Characterization of copper-inducible promoters regulated by CpxA/CpxR in Escherichia coli. Biosci Biotechnol Biochem. 2006;70:1688–95.

    Article  CAS  Google Scholar 

  44. Shimizu T, Ichimura K, Noda M. The surface sensor NlpE of enterohemorrhagic Escherichia coli contributes to regulation of the Type III secretion system and flagella by the Cpx response to adhesion. Am Soc Microbiol. 2016;84:537–49.

    CAS  Google Scholar 

  45. Frenzel E, Legebeke J, Van Stralen A, Van Kranenburg R, Kuipers OP. In vivo selection of sfGFP variants with improved and reliable functionality in industrially important thermophilic bacteria. Biotechnol Biofuels. 2018;11:8.

    Article  Google Scholar 

  46. Sharma CM, Darfeuille F, Plantinga TH, Vogel J. A small RNA regulates multiple ABC transporter mRNAs by targeting C/A-rich elements inside and upstream of ribosome-binding sites. Genes Dev. 2007;21:2804–17.

    Article  CAS  Google Scholar 

  47. Waters LS, Storz G. Regulatory RNAs in bacteria. Cell. 2009;136:615–28.

    Article  CAS  Google Scholar 

  48. Tabor S, Richardson CC. A bacteriophage T7 RNA polymerase/promoter system for controlled exclusive expression of genes. Proc National Academy Sci. 1985;82:1074–8.

    Article  CAS  Google Scholar 

  49. Rosenberg AH, Lade BN, Dao-shan C, Lin SW, Dunn JJ, Studier FW. Vectors for selective expression of cloned DNAs by T7 RNA polymerase. Gene. 1987;56:125–35.

    Article  CAS  Google Scholar 

  50. Soper T, Mandin P, Majdalani N, Gottesman S, Woodson SA. Positive regulation by small RNAs and the role of Hfq. PNAS. 2010;107:2–7.

    Article  Google Scholar 

  51. Vogel J, Luisi BF. Hfq and its constellation of RNA. Nat Rev Microbiol. 2011;9:578–89.

    Article  CAS  Google Scholar 

  52. Chao Y, Li L, Girodat D, Wieden H, Luisi BF, Vogel J. In vivo cleavage map illuminates the central role of RNase E in coding and non-coding RNA pathways. Mol Cell. 2017;65:39–51.

    Article  CAS  Google Scholar 

  53. Moller T, Franch T, Hojrup P, Keene DR, Bächinger HP, Brennan RG, et al. Hfq: a bacterial Sm-like protein that mediates RNA-RNA interaction. Mol Cell. 2002;9:23–30.

    Article  CAS  Google Scholar 

  54. Zhang A, Altuvia S, Tiwari A, Argaman L, Hengge-Aronis R, Storz G. The OxyS regulatory RNA represses rpoS translation and binds the Hfq (HF-I) protein. EMBO J. 1998;17:6061–8.

    Article  CAS  Google Scholar 

  55. Lorenz R, Bernhart SH, Siederdissen CH, Tafer H, Flamm C, Stadler PF, et al. ViennaRNA package 2.0. Algorithms Mol Biol. 2011;6:1–14.

    Article  Google Scholar 

  56. Kawamoto H, Morita T, Shimizu A, Inada T, Aiba H. Implication of membrane localization of target mRNA in the action of a small RNA: mechanism of post-transcriptional regulation of glucose transporter in Escherichia coli. Genes Dev. 2005;19:328–38.

    Article  CAS  Google Scholar 

  57. Gruber AR, Lorenz R, Bernhart SH, Neuböck R, Hofacker IL. The Vienna RNA websuite. Nucleic Acids Res. 2008;36:W70–4.

    Article  CAS  Google Scholar 

  58. De Mey M, Maertens J, Lequeux GJ, Soetaert WK, Vandamme EJ. Construction and model-based analysis of a promoter library for E. coli: an indispensable tool for metabolic engineering. BMC Biotechnol. 2007;7:1–14.

    Google Scholar 

  59. Guzman LM, Belin D, Carson MJ, Beckwith J. Tight regulation, modulation, and high-level expression by vectors containing the arabinose PBAD promoter. J Bacteriol. 1995;177:4121–30.

    Article  CAS  Google Scholar 

  60. Bird LE, Rada H, Verma A, Gasper R, Birch J, Jennions M, et al. Green fluorescent protein-based expression screening of membrane proteins in Escherichia coli. J Vis Exp. 2015. https://doi.org/10.3791/52357.

    Article  Google Scholar 

  61. Lo CA, Kays I, Emran F, Lin TJ, Cvetkovska V, Chen BE. Quantification of protein levels in single living cells. Cell Rep The Authors. 2015;13:2634–44.

    Article  CAS  Google Scholar 

  62. Soboleski MR, Oaks J, Halford WP. Green fluorescent protein is a quantitative reporter of gene expression in individual eukaryotic cells. FASEB J. 2005;19:1–20.

    Article  Google Scholar 

  63. Baird L, Lipinska B, Raina S, Georgopoulos C. Identification of the Escherichia coli sohB gene, a multicopy suppressor of the HtrA (DegP) null phenotype. J Bacteriol. 1991;173:5763–70.

    Article  CAS  Google Scholar 

  64. Stenberg F, Chovanec P, Maslen SL, Robinson CV, Ilag LL, Von Heijne G, et al. Protein complexes of the Escherichia coli cell envelope. J Biol Chem. 2005;280:34409–19.

    Article  CAS  Google Scholar 

  65. Lenz DH, Mok KC, Lilley BN, Kulkarni RV, Wingreen NS, Bassler BL. The small RNA chaperone Hfq and multiple small RNAs control quorum sensing in Vibrio harveyi and Vibrio cholerae. Cell. 2004;118:69–82.

    Article  CAS  Google Scholar 

  66. Levine E, Zhang Z, Kuhlman T, Hwa T. Quantitative characteristics of gene regulation by small RNA. PLoS Biol. 2007;5:1998–2010.

    Article  CAS  Google Scholar 

  67. Ghodasara A, Voigt CA. Balancing gene expression without library construction via a reusable sRNA pool. Nucleic Acids Res. 2017;45:8116–27.

    Article  CAS  Google Scholar 

  68. Na D, Yoo SM, Chung H, Park H, Park JH, Lee SY. Metabolic engineering of Escherichia coli using synthetic small regulatory RNAs. Nat Biotechnol. 2013;31:170–4.

    Article  CAS  Google Scholar 

  69. Saier MH, Beatty JT, Goffeau A, Harley KT, Heijne WHM, Huang SC, et al. The major facilitator superfamily. J Mol Microbiol Biotechnol. 1999;1:257–79.

    CAS  Google Scholar 

  70. Tsukazaki T. Structural basis of the Sec translocon and YidC revealed through X-ray crystallography. Protein J. 2019;38:249–61.

    Article  CAS  Google Scholar 

  71. Kang Z, Wang X, Li Y, Wang Q, Qi Q. Small RNA RyhB as a potential tool used for metabolic engineering in Escherichia coli. Biotechnol Lett. 2012;34:527–31.

    Article  CAS  Google Scholar 

  72. Yang D, Yoo SM, Gu C, Ryu JY, Lee JE, Lee SY. Expanded synthetic small regulatory RNA expression platforms for rapid and multiplex gene expression knockdown. Metab Eng. 2019;54:180–90.

    Article  CAS  Google Scholar 

  73. Raivio TL, Silhavy TJ. Transduction of envelope stress in Escherichia coli by the Cpx two-component system. J Bacteriol. 1997;179:7724–33.

    Article  CAS  Google Scholar 

  74. Raivio TL, Popkin DL, Silhavy TJ. The Cpx envelope stress response is controlled by amplification and feedback inhibition. J Bacteriol. 1999;181:5263–72.

    Article  CAS  Google Scholar 

  75. Fleischer R, Heermann R, Jung K, Hunke S, Mu D. Purification, reconstitution, and characterization of the CpxRAP envelope stress system of Escherichia coli. J Biol Chem. 2007;282:8583–93.

    Article  CAS  Google Scholar 

  76. DiGiuseppe PA, Silhavy TJ. Signal detection and target gene induction by the CpxRA two-component system. J Bacteriol. 2003;185:2432–40.

    Article  CAS  Google Scholar 

  77. De Wulf P, McGuire AM, Liu X, Lin ECC. Genome-wide profiling of promoter recognition by the two-component response regulator CpxR-P in Escherichia coli. J Biol Chem. 2002;277:26652–61.

    Article  Google Scholar 

  78. Dorel C, Lejeune P, Rodrigue A. The Cpx system of Escherichia coli, a strategic signaling pathway for confronting adverse conditions and for settling biofilm communities? Res Microbiol. 2006;157:306–14.

    Article  CAS  Google Scholar 

  79. Hussein R, Lim HN. Disruption of small RNA signaling caused by competition for Hfq. Proc Natl Acad Sci USA. 2011;108:1110–5.

    Article  CAS  Google Scholar 

  80. Samuelson JC, Chen M, Jiang F, Möller I, Wiedmann M, Kuhn A, et al. YidC mediates membrane protein insertion in bacteria. Nature. 2000;406:637–41.

    Article  CAS  Google Scholar 

  81. Urbanus ML, Fröderberg L, Drew D, Björk P, De Gier JWL, Brunner J, et al. Targeting, insertion, and localization of Escherichia coli YidC. J Biol Chem. 2002;277:12718–23.

    Article  CAS  Google Scholar 

  82. Raivio TL. Everything old is new again: an update on current research on the Cpx envelope stress response. Biochim Biophys Acta Mol Cell Res. 2014;1843:1529–41.

    Article  CAS  Google Scholar 

  83. Szenk M, Dill KA, de Graff AMR. Why do fast-growing bacteria enter overflow metabolism? Testing the membrane real estate hypothesis. Cell Syst. 2017;5:95–104.

    Article  CAS  Google Scholar 

  84. Basan M, Hui S, Okano H, Zhang Z, Shen Y, Williamson JR, et al. Overflow metabolism in Escherichia coli results from efficient proteome allocation. Nature. 2015;528:99–104.

    Article  CAS  Google Scholar 

  85. Lim B, Miyazaki R, Neher S, Siegele DA, Ito K, Walter P, et al. Heat shock transcription factor σ32 co-opts the signal recognition particle to regulate protein homeostasis in E. coli. PLoS Biol. 2013. https://doi.org/10.1371/journal.pbio.1001735.

    Article  Google Scholar 

  86. Kumar M, Sourjik V. Physical map and dynamics of the chaperone network in Escherichia coli. Mol Microbiol. 2012;84:736–47.

    Article  CAS  Google Scholar 

  87. Noor R. Mechanism to control the cell lysis and the cell survival strategy in stationary phase under heat stress. Springerplus. 2015. https://doi.org/10.1186/s40064-015-1415-7.

    Article  Google Scholar 

  88. Baig F, Fernando LP, Salazar MA, Powell RR, Bruce FB, Harcum SW. Dynamic transcriptional response of Escherichia coli to inclusion body formation. Biotechnol Bioeng. 2014;23:1–7.

    Google Scholar 

  89. Ceroni F, Boo A, Furini S, Gorochowski TE, Borkowski O, Ladak YN, et al. Burden-driven feedback control of gene expression. Nat Methods. 2018;15(5):387.

    Article  CAS  Google Scholar 

  90. Ćudić E, Surmann K, Panasia G, Hammer E, Hunke S. The role of the two-component systems Cpx and Arc in protein alterations upon gentamicin treatment in Escherichia coli. BMC Microbiol. 2017;17:1–17.

    Article  Google Scholar 

  91. Wehrens M, Ershov D, Rozendaal R, Walker N, Schultz D, Kishony R, et al. Size laws and division ring dynamics in filamentous Escherichia coli cells. Curr Biol. 2018;28:972-979.e5.

    Article  CAS  Google Scholar 

  92. Sambrook JF, Russell DW. Molecular cloning: A laboratory manual. 3rd ed. New York: Cold Spring Harbor Laboratory Press; 2001.

    Google Scholar 

  93. Coussement P, Maertens J, Beauprez J, Van Bellegem W, De Mey M. One step DNA assembly for combinatorial metabolic engineering. Metab Eng. 2014;23:70–7.

    Article  CAS  Google Scholar 

  94. Coussement P, Bauwens D, Maertens J, De Mey M. Direct combinatorial pathway optimization. ACS Synth Biol. 2017;6:224–32.

    Article  CAS  Google Scholar 

  95. Pédelacq JD, Cabantous S, Tran T, Terwilliger TC, Waldo GS. Engineering and characterization of a superfolder green fluorescent protein. Nat Biotechnol. 2006;24:79–88.

    Article  Google Scholar 

  96. Engler C, Kandzia R, Marillonnet S. A one pot, one step, precision cloning method with high throughput capability. PLoS ONE. 2008. https://doi.org/10.1371/journal.pone.0003647.

    Article  Google Scholar 

Download references

Acknowledgements

Confocal scanning light microscopy, with Airyscan technology was provided by the Bio-imaging Facility at VIB Ghent.

Funding

C. Guidi, L. De Wannemaeker, W. Demeester and J. De Baets hold a PhD grant (1S16017N, 1S01018N, 1SC6820N and 1S29521N, respectively) from the Research Foundation—Flanders (FWO). This research was also supported by the BOF-IOP project “MLSB” (BOF16/ IOP/040) of the Bijzonder Onderzoeksfonds and the FWO research project “SynSysBio4COS” (G0B8118) of the Research Foundation—Flanders (FWO).

Author information

Authors and Affiliations

  1. Centre for Synthetic Biology, Ghent University, 9000, Ghent, Belgium

    Chiara Guidi, Lien De Wannemaeker, Jasmine De Baets, Wouter Demeester, Jo Maertens, Brecht De Paepe & Marjan De Mey

Authors
  1. Chiara Guidi
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Lien De Wannemaeker
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Jasmine De Baets
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Wouter Demeester
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Jo Maertens
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Brecht De Paepe
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Marjan De Mey
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

CG, LDeW, WD and JDeB designed and performed the experiments. CG analysed the data and wrote the manuscript. JM, BDeP and MDeM were involved in the conception and design of this work and in drafting, discussing, correcting and revising the manuscript. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Marjan De Mey.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no conflict of interest.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1

: Figure S1. Growth curves corresponding to indirect stress detection tool. Figure S2. cpxQ binding sites for nhaB mRNA. Figure S3. Bar plots representing the output sfGFP/OD600, mKate2, sfGFP and µmax for the control plasmid (blue) and the plasmid expressing the sRNA cpxQmut1 (grey) and this for different L-arabinose concentrations. Figure S4. Confocal scanning light microscopy images with intensity cross-section profiles for both the red FM4-64 membrane dye (purple colour) and the green fluorescent protein signal of TMD-sfGFP fusion proteins. Figure S5. Evaluation of the three dual plasmid designs (D1-3) on growth and protein production of the membrane targeted sfGFP. Figure S6. Library of nine different cpxQ mutants in their secondary structure with the associated ΔG2 value. Figure S7. Effect of the smart library of sRNA-based stress circuits on fluorescence intensity. Figure S8. Relative mKate2/OD600 fluorescence of Escherichia coli (E. coli) DE3 cells producing either sfGFP (cytoplasmatic superfolder green fluorescent protein, negative control), NlpE (IMP of E. coli, positive control), SohB(TMD)-sfGFP, YidC-sfGFP and GarP-sfGFP for several IPTG-concentrations (Timepoint: 10h, stationary phase). Figure S9. Effect of the sRNA-based circuit, respectively with cpxQmut1 (Stress Circuit 1) and cpxQmut3 (Stress Circuit 3), on cell growth and fluorescence intensity. Figure S10. Confocal scanning light microscopy images with intensity cross-section profiles for both the red FM4-64 membrane dye (purple colour) and the green fluorescent protein signal of MPs YidC and GarP (sfGFP, green colour). Figure S11. Effect of the non-specific and specific sRNA-based stress circuit on protein production of functional MPs expression. Figure S12. Detailed plasmid maps of the vectors created in this work. Table S1. List of fold changes for NlpE and sfGFP expression (at OD600 = 0.3) measured for indirect IM stress analysis. Table S2. Bacterial Escherichia coli strains that were used in this study. Table S3. List of DNA sequences used in this study to design the smart small RNA library. Table S4. Two sample t-tests performed in this study. Table S5. Statistical one-way ANOVA performed in this study. Table S6. Plasmids that were used and constructed throughout this study. Table S7. List of proteins used in this study for membrane localisation and membrane targeting. Table S8. List of fold changes for NlpE, sfGFP, SohB(TMD)-sfGFP, YidC-sfGFP and GarP-sfGFP expression (at t = 20h) measured for indirect IM stress analysis. Table S9. Overview of the different plasmid backbones used in this study and their assigned function. Table S10. List of DNA sequences used in this study (coding sequences, promoter, 5’UTRs and terminator sequences).

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. 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 in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Guidi, C., De Wannemaeker, L., De Baets, J. et al. Dynamic feedback regulation for efficient membrane protein production using a small RNA-based genetic circuit in Escherichia coli. Microb Cell Fact 21, 260 (2022). https://doi.org/10.1186/s12934-022-01983-2

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12934-022-01983-2

Keyword

Microbial Cell Factories

ISSN: 1475-2859

Contact us