Modular plasmid design for autonomous multi-protein expression in Escherichia coli

Abstract

Background

Molecular and synthetic biology tools enable the design of new-to-nature biological systems, including genetically engineered microorganisms, recombinant proteins, and novel metabolic pathways. These tools simplify the development of more efficient, manageable, and tailored solutions for specific applications, biocatalysts, or biosensors that are devoid of undesirable characteristics. The key aspect of preparing these biological systems is the availability of appropriate strategies for designing novel genetic circuits. However, there remains a pressing need to explore independent and controllable systems for the co-expression of multiple genes.

Results

In this study, we present the characterisation of a set of bacterial plasmids dedicated to recombinant expression in broadly used Escherichia coli. The set includes plasmids with four different, most commonly used bacterial expression cassettes - RhaS/RhaBAD, LacI/Trc, AraC/AraBAD, and XylS/Pm, which can be used alone or freely combined in up to three-gene monocistronic expression systems using Golden Standard Molecular Cloning kit assembly. The independent induction of each of the designed cassettes enables the autonomous expression of up to three recombinant proteins from one plasmid. The expression of a triple-enzyme cascade consisting of sucrose synthase, UDP-rhamnose synthase and flavonol-7-O-rhamnosyltransferase, confirmed that the designed system can be applied for the complex biocatalysts production.

Conclusions

Presented herein strategy for the multigene expression is a valuable addition to the current landscape of different co-expression approaches. The thorough characterisation of each expression cassette indicated their strengths and potential limitations, which will be useful for subsequent investigations in the field. The defined cross-talks brought a better understanding of the metabolic mechanisms that may affect the heterologous expression in the bacterial hosts.

Background

The design of artificial genetic circuits forms the foundation of the synthetic biology. Discovered over the last few decades and still carefully investigated, natural mechanism of gene expression or genome editing are extensively harnessed to build artificial biological systems tailored for purposes, such as recombinant proteins and enzymatic cascades production, or even entire metabolic pathways assembly. The deconstruction and redesign of such mechanisms enable a better understanding of the biochemical principles and further result in more optimised processes. Among all model organisms for synthetic biology the most exploited, due to their relative simplicity and ease of manipulation is Escherichia coli [1]. It is broadly used as a model organisms for metabolic engineering studies [2] and are a well-characterised platforms for the production of recombinant proteins [3, 4].

Various strategies for the heterologous expression in bacterial hosts are developed, including plasmid-based [5, 6] and genome integration [7, 8] approaches. In case of synthetic plasmids design different cloning strategies, artificial vectors and genetic parts were developed. Starting from simple restriction enzyme digestion and ligation protocol, through BioBricks assembly [9], Gibson assembly [10], and Golden Gate assembly [11], to the most recent such as CIDAR [12], or Golden Standard Modular Cloning (GS MoClo) [13] assemblies. They offer different levels of complexity of final genetic circuits, including multi-genetic complexes. However, there is still an urgent need for investigations focusing on the independent and steerable co-expression of several genes in one host. Often developing a metabolic pathways or enzymatic cascades, it is crucial to deliver an optimised proportions between involved proteins [14, 15]. Apart from utilisation of couples of plasmids with various copy numbers [16] or ribosome binding sites (RBSs) of different strengths [17], the most precise and presumably autonomous control of the expression levels offer regulator/promoter inducible systems.

In this study we present a set of new plasmids, dedicated for the recombinant expression in E. coli. Plasmids design was based on the recently developed GS MoClo assembly based on Standard European Vector Architecture (SEVA) [18]. The aim was to prepare ready-to-use expression cassettes with different induction systems that requires only the chosen protein tag, RBS, and coding sequence genetic parts for assembly. Furthermore, the single-gene cassettes were assembled into a combinatorial multigene monocistronic configuration, dedicated for independent co-expression of up to three proteins from single plasmid.

Methods

Materials

Microbiological media components were purchased from Sigma-Aldrich (St Louis, USA) and SERVA Electrophoresis GmbH (Heidelberg, Germany). Antibiotics were bought from Cayman Chemical Company (Ellsworth, USA). Chemical inducers rhamnose (≥ 99% purity), arabinose (99% purity), m-toluic acid (99% purity) were purchased from Sigma-Aldrich (St Louis, USA), and isopropyl-1-thio-β-D-galactopyranoside (IPTG, ≥ 98% purity) was purchased from SERVA Electrophoresis GmbH (Heidelberg, Germany). Plasmid Miniprep Kit, Taq PCR Kit, T4 DNA ligase, and restriction enzymes (BsaI-HFv2, BbsI-HF) were bought from New England Biolabs Inc. (NEB, Ipswich, USA). All genes and expression cassettes were codon-optimised and synthesised by Invitrogen GeneArt Gene Synthesis (Thermo Fisher Scientific, Waltham, MA, USA). Primers utilised for PCR and sequencing were acquired from Sigma-Aldrich (St Louis, USA). Chemically competent E. coli cells – NEB 5-alpha, NEB 10-beta, and BL21 (DE3) were purchased from NEB (Ipswich, USA). All primers, genes, plasmids and E. coli strains utilised within this study are listed in Table S1. Naringenin (1) (98% purity) was acquired from Sigma-Aldrich (St Louis, USA). The UPLC-grade solvents (acetonitrile (≥ 99.9% purity)), and methanol (99.9% purity)) were bought from Merck KGaA (Darmstadt, Germany).

Plasmid design and cloning

Plasmid design was based on the SEVA [18] and adapted to GS MoClo assembly principles [13]. The simplified design of the expression cassettes and framework of the modular cloning procedure is presented in Fig. 1. Expression cassettes were designed in silico using SnapGene software (version 5.1.7) in three variants, each flanked with different fusion site pairs denoted as 12, 23 and 34 (which include BbsI restriction sites as shown in Fig. 1A), corresponding to the 1st level vectors of GS MoClo assembly [13].

Fig. 1

A The framework of the expression cassettes design and modular cloning assemblies. B Description of the GS fusion sites [13] employed in the assemblies. C Detailed structure of expression cassettes. Abbreviation: TF – transcriptional factor; RS – restriction site; RBS – ribosome binding site; CDS – coding sequence, T - terminator

Each cassette comprises an appropriate transcriptional factor (TF), promoter, cargo region (suitable for introduction of RBS, N- or C-tag and CDS according to GS MoClo principles [13] flanked with BD fusion sites (which include BsaI restriction sites)) and T7 terminator (Fig. 1). Subsequently, the designed expression cassettes were cloned into pSEVA23g19[g1], pSEVA23g19[g2], and pSEVA23g19[g3] vectors [13], based on their corresponding fusion sites. Prepared 1st level backbone plasmids (pRhaBAD_12, pRhaBAD_23, pRhaBAD_34, pTrc_12, pTrc_23, pTrc_34, pAraBAD_12, pAraBAD_23, pAraBAD_34, pPm_12, pPm_23, pPm_34) were ready for modular cloning of expression plasmids according to GS MoClo 1st and 2nd level assemblies [13]. All used and constructed plasmids are listed in Table S1.

Transformation and clone selection

Plasmids were transformed into the chemically competent E. coli cells – NEB 5-alpha or NEB 10-beta for plasmid maintenance, or BL21 (DE3) for expression tests and protein production. Positive clones were identified through blue-white screening on selective media supplemented with X-Gal (30 µg/mL) (except pLacI/Trc plasmids, which are not compatible with blue-white screening, due to the presence of lac operator next to the lacZα promoter) and the appropriate antibiotic (ampicillin (100 mg/L), kanamycin (30 mg/L), or gentamicin (20 mg/L)), as well as through colony PCR using Taq DNA polymerase. The plasmid sequences were confirmed through Sanger sequencing conducted by Macrogen Europe BV (Amsterdam, The Netherlands).

Fluorescent microplate assays

In vivo fluorescence measurements were performed at 37 °C in the 96-well microplate (ScreenStar black 96-well microplate, clean F-bottom with lid; Greiner Bio-One, Kremsmünster, Austria) in Synergy H1 microplate reader (BioTek Instruments, Vermont, USA) using 24 h continuous assay with 45 min intervals for shaking (15 s, 282 rpm), absorbance (OD₆₀₀) and fluorescence (Electra1: excitation 402 nm, emission 454 nm, GFP: excitation 485 nm, emission 513 nm; mCherry: excitation 579 nm, emission 616 nm) measurements. The working volume in each well was 200 µL. In each plate, the outer wells (rows A and H, columns 1 and 12) were filled with distilled water to minimize evaporation from the inner wells, which contained investigated cultures. The growth was carried out in M9 minimal medium with improved buffering capacity (2xM9) [19] to account for the pH dependence of GFP fluorescence [20]. The medium was supplemented with kanamycin (30 mg/L) or gentamicin (20 mg/L), 1% (v/v) inoculum of overnight preculture, 0.8% (w/w) of glucose or 0.8% (w/w) of glycerol, and the inducers (rhamnose, arabinose, m-toluic acid, or IPTG). The uninduced cultures were used as a control for expression of fluorescent proteins (FPs). The cultures of BL21 (DE3) strain carrying plasmid backbone were used as a growth and background fluorescence control, and their values were subtracted from the sample values. The growth medium was used as a control for background absorbance/fluorescence. If not stated differently, all variants were cultivated in 6 independent replicates. Fluorescence values were normalised against the OD₆₀₀ values measured directly in 200 µL culture volumes, without adjustment to a 1 cm pathlength.

Enzyme production and purification

The recombinant enzyme production (from single-gene and triple-gene plasmids) was carried out overnight in 75 mL Dynamite medium [21] (single-gene plasmids) or in M9 medium (triple-gene plasmids), supplemented with kanamycin (30 mg/L) or gentamicin (20 mg/L) in a 250 mL baffled flask at 25 °C with 120 rpm agitation using an incubator shaker (New Brunswick Innova 44, Eppendorf, Hamburg, Germany). A 1% (v/v) of overnight preculture was used as an inoculum. Expression was induced when the OD₆₀₀ reached 0.6–0.8. Expression from single-gene plasmids was induced using 10 mM rhamnose, and from triple-gene plasmids was induced using10 mM rhamnose or combination of 10 mM rhamnose, 0.5 mM IPTG, and 0.0025 mM m-toluic acid. The overnight culture was centrifuged (4,000 x g, 30 min, 4 °C; Centrifuge 5810R, Eppendorf, Hamburg, Germany) and the cell pellet was resuspended in 10 mL Binding Buffer (20 mM sodium phosphate, 500 mM NaCl, 20 mM imidazole, pH 7.5), digested by lysozyme (0.3 mg/mL) for 1 h, and disrupted by sonication (6 min program with 20 s pulse and 20 s pause, 80% amplitude) in an ice bath using Vibra-Cell Ultrasonic Liquid Processor VCX 130 (Sonics & Materials, Inc., Newtown, USA). The cell debris was harvested by centrifugation (14,000 x g, 30 min, 4 °C; Centrifuge 5430R, Eppendorf, Hamburg, Germany). The cell lysate was either directly utilised for reactions, stored at − 80 °C, or subjected to the purification protocol. The total protein concentration of the cell lysate was determined using the Bicinchoninic Acid (BCA) Protein Assay Kit (Sigma-Aldrich, St Louis, USA), following the manufacturer’s protocol, with BSA as a reference for the calibration curve. For purification, cell lysate was filtrated through a 0.45 μm filter, applied to a IMAC column (His-TrapTM, GE Healthcare) and proceeded according to the manufacturer’s protocol. The His_6x-tagged enzyme was eluted in Elution Buffer (20 mM sodium phosphate, 500 mM NaCl, 500 mM imidazole, pH 7.5). Finally, selected fractions were combined, and the buffer exchange was carried on centrifugal filters Amicon Ultra 30 kDa (AtUGT89C1) or 50 kDa (GmSuSy and VvRHM_NRS) cut-off (Merck Milipore, Massachusetts, USA) until ~ 800-1000x dilution of imidazole to a Storage Buffer (50 mM phosphate buffer, 150 mM NaCl, pH 7.5). The purified enzyme was directly used for reactions or stored at − 80 °C. Enzyme concentration was measured according to protein absorbance at 280 nm (GmSuSy (M_w = 92240 M, ɛ = 104210 M^− 1 cm^− 1), VvRHM-NRS (M_w = 76000 M, ɛ = 81750 M^− 1 cm^− 1), AtUGT89C1 (M_w = 48120 M, ɛ = 53065 M^− 1 cm^− 1)). The expression and purification of the enzyme were evaluated by 10% SDS-PAGE analysis [22], with proteins stained by Coomassie Brilliant Blue (Cepham Life Sciences, Inc). PageRuler Plus Prestained Protein Ladder (Thermo Fisher Scientific, Waltham, MA, USA) was used as a molecular size marker.

Cascade assays

Reactions were conducted in triplicates in a 50 mM HEPES buffer (50 mM KCl, 10 mM MgCl₂, pH 7.5) with 0.5 mM uridine diphosphate (UDP), 1 mM nicotinamide adenine dinucleotide (NAD⁺), 500 mM sucrose, 0.05 mM naringenin dissolved in 1% (v/v) DMSO, and cell lysate (~ 2 mg/mL) or purified enzymes (Table S11) as the source of catalyst. Reactions were stopped after 4 h (cell lysates) or after 1 h (purified enzymes) by the addition of ethyl acetate and vortexing. The reaction mixtures were centrifuged, and extracts were diluted 8-fold with methanol prior to analysis. The reaction progress was evaluated by the UPLC-DAD assay performed on Dionex Ultimate 3000 UHPLC + instrument (Thermo Fisher Scientific, Waltham, MA, USA) The UPLC system was equipped with a DGP-3600 A dual pump liquid control compartment, a TCC-3200 thermostated column module, a WPS-3000 autosampler, a diode array detector (DAD), and an analytical C-18 Acclaim RSLC PolarAdvantage II (2.2 μm, 2.1 × 100 mm, Thermo Fisher Scientific, Waltham, MA, USA) column thermostated at 40 °C with two mobile phases: (A) 0.1% formic acid solution in water and (B) 0.1% formic acid solution in acetonitrile, and following gradient elution program − 0–3 min: 15–98% B; 3–4.2 min: 98% B; 4.2–4.4 min: 98 − 15% B; 4.4–6 min: 15% B at 0.7 mL/min flow rate. System control and data acquisition were managed in Chromeleon 6.80 software (Dionex, Sunnyvale, USA). Detection was carried out at 280 nm. Conversion was calculated as the percentage of naringenin converted. The identification of naringenin and naringenin 7-O-rhamnoside was based on authentic standard compounds, determined by their UPLC retention times and UV-Vis spectra.

Statistical analysis

The means, standard deviations of the mean were calculated from triplicate or sextuplicate experiments. Data were analysed for statistical significance with Statistica software (version 13.3) by unpaired t-test, one-way analysis of variance (one-way ANOVA) with Tukey’s post hoc test. Equality of the variance was verified using Levene’s test. If needed, a Box-Cox transformation was performed to unify the variance of the data.

Accession numbers

Synthetic genes used in the study were deposited in NCBI Genbank database under accession numbers listed in the Table S1.

Results

In this study, we present our approach to extend GS MoClo kit and create a plasmid-based platform for simultaneous and independent expression of up to three heterologous genes in E. coli under different control. Investigation includes a detailed characterisation of synthetic plasmids. Moreover, as an example of its application, a one-step production of a triple-enzyme cascade for flavonoid rhamnosylation was performed.

Characterisation of single-gene expression plasmids

Designed expression cassettes were based on four bacterial regulator/promoter systems – RhaS/RhaBAD [23], LacI/Trc [24], AraC/AraBAD [25], and XylS/Pm [26], responsive to small molecules – rhamnose, IPTG (or other allolactose analogues), arabinose, or aromatic hydrocarbons, respectively. The organisation of the functional parts of each cassette, including their order and direction of transcription, is depicted in Fig. 1C, as they contain some modifications compared to their native versions. RhaS/RhaBAD cassette includes rhaS gene under synthetic and weak J23103 promoter from Anderson Promoter Collection and T500 terminator put parallelly to RhaBAD promoter. LacI/Trc cassette contains lacI gene under mutant LacI^q promoter and rmBT1T terminator localised parallelly to Trc promoter as well. In contrary, AraC/AraBAD and XylS/Pm cassettes have the TF genes placed divergently to the CDS region. AraC/AraBAD cassette has araC gene placed under the Pc promoter (AraC promoter) and Trp reverse terminator. XylS/Pm cassette includes xylS gene under medium strong synthetic J23114 promoter and T500 terminator. Each cassette includes a T7 terminator at the end of the transcriptional unit for the expression of the gene of interest.

Firstly, the background expression levels and dynamic range were investigated, to define basic characteristics of designed plasmids. The analysis was based on the expression of mCherry, as an easily detectable fluorescence protein from pRhaBAD_12_mCherry, pTrc_12_mCherry, pAraBAD_12_mCherry and pPm_12_mCherry (Table S1) as model plasmids. All tested plasmids showed a dose-dependent response to the inducers within the tested concentration ranges (Fig. 2). One-way ANOVA of average normalised fluorescence (RFU/OD₆₀₀) after 18 h of cultivation against inducer dosage revealed statistically significant differences in induction strength (Fig. 2B, D, F, H) and confirmed the capability to control gene expression by titrating the inducer from each cassette.

Fig. 2

The background expression and dynamic range of A RhaS/RhaBAD, C LacI/Trc, E AraC/AraBAD, G XylS/Pm single-gene cassettes monitored through the fluorescence intensity of mCherry over a 24 h cultivation. The normalised fluorescence of mCherry at 18 h of cultivation was plotted against inducer concentration for B RhaS/RhaBAD, D LacI/Trc, F AraC/AraBAD, H XylS/Pm cassette induction and analysed by one-way ANOVA and Levene’s tests, which results are shown in Table S2. Specific differences between neighbouring groups were tested using Tukey’s post hoc analysis. Error bars represent the standard deviations of 6 individual replicates. Fluorescence values were normalised against the OD₆₀₀. Plain E. coli BL21 (DE3) strain was used as a control. Abbreviation: ns – not significant

In cultures where mCherry expression was controlled by LacI/Trc and XylS/Pm systems, the fluorescence was detected at already first measurement in the test (~ 2000 and ~ 800 RFU/OD₆₀₀ respectively). Possibly, the LB medium used for precultures was contaminated with compounds that induced heterologous expression (e.g., lactose, aromatic hydrocarbons). This is suggested by the fact that the fluorescence diminished over time in the control samples or samples with low inducers concentration when they were cultivated in the 2xM9 medium. The OD₆₀₀ and unnormalised fluorescence values from the test are included in the Supplementary material (Fig. S1). High rhamnose concentrations turn out to inhibit the growth of E. coli BL21 (DE3). The higher the rhamnose concentration, the more postponed the exponential phase was, but only the two highest rhamnose concentrations (25 and 50 mM) caused a significant decrease in the final cell density (Fig. S2). No other noticeable impact on E. coli growth was detected, even though post hoc analyses showed several other statistically significant differences of OD₆₀₀ between the control and induced cultures (Fig S2) – it more results from low variances within the tested groups. The effect of inducers on bacterial growth was also checked on plain E. coli BL21 (DE3) with selected inducers’ concentrations (Fig. S3). One-way ANOVA of average OD₆₀₀ after 24 h of cultivation demonstrated significant differences between tested groups, and Tukey’s post hoc analysis revealed that 1 mM IPTG had the strongest negative impact on final OD₆₀₀ (Fig. S3). 10 mM rhamnose and 1 mM m-toluic acid caused moderate decrease in the final cell density (Fig. S3).

Subsequently, the expression strength of the designed plasmids was compared to that of the pSEVA23g19[g1] plasmid (which has similar backbone parts) where mCherry was placed under the control of strong constitutive J23100 promoter (Fig. S4A, B, and G). The comparison showed that expression from the LacI/Trc cassette is up to 4 times stronger than constitutive expression, while RhaS/RhaBAD is up to 1.8 times stronger, and XylS/Pm is up to 1.5 times stronger (Fig. 3).

Fig. 3

Relative expression from RhaS/RhaBAD, LacI/Trc, AraC/AraBAD and XylS/Pm cassettes in comparison to constitutive expression of mCherry at 6, 12, 18 and 24 h of the cultivation. Inducers concentrations were: 10 mM rhamnose, 0.5 mM IPTG, 10 mM arabinose, and 0.0025 mM m-toluic acid. The results of unpaired t-test between constitutive and induced mCherry expression are showed in the Table S3. Error bars represent the standard deviations of 6 individual replicates. Fluorescence values were normalised against the OD₆₀₀ and compared to the normalised fluorescence values of constitutively expressed mCherry. Plain E. coli BL21 (DE3) strain was used as a control

On the other hand, expression from AraC/AraBAD turned out to be weaker by up to 30% in comparison to plasmid with constitutive J23100 promoter (Fig. 3).

Next, we evaluated possible cross-reactivity, including nonspecific activation and inhibition of expression caused by the inducers.The tests showed no significant nonspecific activation between tested expression cassettes and their inducers (Fig. 4A) but revealed a strong inhibition of the AraC/AraBAD system by IPTG (Fig. 4B).

Fig. 4

Cross-reactivity heat map of nonspecific A activation and B inhibition of mCherry expression from investigated single-gene plasmids by different inducers. In the nonspecific activation test, nonspecific inducers were added to uninduced cultures, whereas in the nonspecific inhibition test, nonspecific inducers were added to induced cultures. Inducers concentrations were: 10 mM rhamnose, 1 mM IPTG, 10 mM arabinose, and 0.0025 mM m-toluic acid. Fluorescence values were normalised against the OD₆₀₀

Some minor synergistic coactions between expression cassettes and nonspecific inducers were also detected. Addition of rhamnose or m-toluic acid in the presence of IPTG increased expression from LacI/Trc cassette (Fig. 4B). We also observed a positive effect of IPTG on the RhaS/RhaBAD cassette, which was already described in our previously published paper [27]. Rhamnose also slightly boosted expression from the XylS/Pm cassette (Fig. 4B). All one-way ANOVA and Tukey’s post hoc analysis results can be found in the Supplementary material (Table S4 and S5). Due to the IPTG inhibition of expression from the AraC/AraBAD system, we decided to verify if lactose will affect it as well. Analysis showed that lactose cannot non-specifically activate or inhibit any of the tested expression cassettes, including AraC/AraBAD (Fig. S5). Only a slight synergistic effect with rhamnose for RhaS/RhaBAD induction was noted (Fig. S5B). Furthermore, the dynamic range of the lactose induction was verified. Compared to IPTG, it shows much weaker induction strength but retains titrability (Fig. S5D-F). Also, no significant influence on E. coli growth was detected (Fig. S5C).

The last step in single-gene plasmid characterisation was an investigation of the influence of carbon source on the induction efficiency. Two sources of carbon were investigated - glucose and glycerol. The test revealed that LacI/Trc, AraC/AraBAD and XylS/Pm systems are significantly more effective when glycerol serves as a carbon source (Fig. 5).

Fig. 5

Influence of the carbon source on the induction. The differences in fluorescence intensity of mCherry (bar-plot) and OD₆₀₀ (cross-plot) at 18 h of cultivation between cultures supplemented with 0.8% glucose or 0.8% glycerol were analysed by unpaired t-test (Table S6 and Table S7). Strains were cultured at 37 °C for 24 h in 2xM9 minimal medium. Inducers concentrations were: 10 mM rhamnose, 1 mM IPTG, 10 mM arabinose, and 0.0025 mM m-toluic acid. Error bars represent the standard deviations of 6 individual replicates. Fluorescence values were normalised against the OD₆₀₀. Plain E. coli BL21 (DE3) strain was used as a control

Expression from RhaS/RhaBAD plasmid showed inverse action (Fig. 5) with higher normalized mCherry fluorescence observed in cultures grown on glucose compared to glycerol. Nevertheless, in all samples supplemented in glycerol final cell density was lower than in cultures with glucose in medium. This effect was stronger in strains induced by rhamnose and m-toluic acid.

Characterisation of triple-gene expression plasmids

This part of the investigation focused on verifying if the simultaneous expression of three genes from three differently induced promoters placed on one plasmid is feasible. Furthermore, we analysed the range of the independent control of each expression system when organised into such a monocistronic configuration (Fig. 1A) using different configurations and concentrations of the inducers. Moreover, it was verified if the order of single-gene cassettes in this final triple-gene expression system influences on the expression from individual promotor. To characterise the simultaneous expression of three heterologous genes in one bacterial strain, three fluorescent proteins (FPs) with no significant spectra overlap and non-toxic to the host – mCherry, GFPmut3* and Electra1 – were selected as reporter proteins. Three combinations of expression cassettes were chosen for testing – RhaS/RhaBAD, AraC/AraBAD with LacI/Trc (pTriple_colors_A and pTriple_color_D-H plasmids), RhaS/RhaBAD, XylS/Pm with LacI/Trc (pTriple_colors_B plasmid), and XylS/Pm, AraC/AraBAD with LacI/Trc (pTriple_colors_C plasmid). The order of the cassettes in each triple-gene expression system is depicted in Fig. 6D. A description of which fluorescent protein corresponds to each expression cassette in the plasmids is provided in Table S1. To standardise the comparison of expression levels between three different fluorescent proteins, their fluorescence was normalised against their constitutive expression from identical plasmids (Fig S4), similar to the approach used for single-gene plasmids (Fig. 3). As previously, the fluorescence of each protein was monitored over 24 h cultivation alongside OD₆₀₀ of the culture (Fig. S6).

Firstly, the E. coli strains carrying pTriple_colors_A, pTriple_colors_B or pTriple_colors_C plasmids served as a model to verify whether all three expression system sets allow independent expression of selected proteins through the addition of different combinations of inducers. According to the results, the aforementioned control is feasible, but we observed some cross-interactions, that did not appear at the single-gene plasmids characterisations. In general, the addition of a single inducer (rhamnose, arabinose, IPTG or m-toluic acid) resulted in the expression of the FP from the corresponding cassette (Fig. 6A-C). Although, in cultures where rhamnose was supplemented, the relative expression of the other two proteins also appeared. On the other hand, induction with arabinose, IPTG or m-toluic acid turned out to be quite specific (Fig. 6A-C). It also needs to be mentioned that the addition of the rhamnose diminished the growth of E. coli strains with the triple-gene plasmids more, than with previously described single-gene plasmids (Fig. S6A, E). Supplementation of all three inducers resulted in the expression of all three FPs in the range, accurate for each promoter (Fig. 6A-C). Additionally, the pairs of the inducers were supplemented to broaden the understanding of possible cross-actions. Once again, in the presence of rhamnose, relative expression of all three FPs was detected, while induction with other compounds resulted in the expression of corresponding FPs (Fig. 6A-C). When exposed to rhamnose the relative expression from LacI/Trc, turned out to be much higher than in samples, where only IPTG was added (Fig. 6A, B). A similar, but weaker effect was observed for the relative expression from AraC/AraBAD and XylS/Pm as well. On the other hand, the addition of the IPTG caused a decrease in the relative expression from AraC/AraBAD or XylS/Pm and an increase in the relative expression from RhaS/RhaBAD (Fig. 6,). The least nonspecific cross-actions between inducers and expression system were observed for the expression from pTriple_colors_C plasmid (Fig. 6C). One-way ANOVA and Tukey’s post hoc analyses of these interactions are presented in the Table S8-S10.

Fig. 6

A-C Relative co-expression of mCherry, GFPmut3*, and Electra1 from from A pTriple_colors_A B pTriple_colors_B C pTriple_colors_C induced by various combinations of the inducers. Inducers concentrations: 10 mM rhamnose, 0.5 mM IPTG, 10 mM arabinose, or 0.0025 mM m-toluic acid. D Order of the cassettes in each plasmid. Error bars represent the standard deviations of 6 individual replicates. Fluorescence values were normalised against the OD₆₀₀ and the constitutive fluorescence of the corresponding fluorescent protein. Abbreviation: A - pTriple_colors_A plasmid; B - pTriple_colors_B plasmid, C - pTriple_colors_C etc.; none – without induction; all – induced by three inducers; rha – rhamnose; ara – arabinose

Subsequently, it was tested whether the sensitivity to different inducer concentrations observed for the single-gene plasmids was retained in the triple-gene plasmid and whether lowering the inducer concentrations could reduce unwanted cross-activities. All expression systems maintained a dose-dependent induction pattern (Fig. 7). However, the LacI/Trc system demonstrated the highest susceptibility to rhamnose concentration — the higher the rhamnose concentration applied, the higher the relative expression from the LacI/Trc system was detected (Fig. 7A-B).

Fig. 7

Relative co-expression of mCherry, GFPmut3*, and Electra1 from A pTriple_colors_A B pTriple_colors_B C pTriple_colors_C induced by various concentrations of the inducers. The inducer concentrations are indicated in mM. Error bars represent the standard deviations of 6 individual replicates. Fluorescence values were normalised against the OD₆₀₀ and the constitutive fluorescence of the corresponding fluorescent protein. Abbreviation: A - pTriple_colors_A plasmid; B - pTriple_colors_B plasmid, C - pTriple_colors_C

Moreover, unexpected inhibition of LacI/Trc system was detected in strain carrying pTriple_colors_C plasmid when induced with 0.5 mM IPTG, 0.0005 mM m-toluic acid and 10 mM arabinose (Fig. 7C).

Next step focused on the verification if order of the cassettes in the triple-gene monocistronic system affects the expression of the genes of interest. Only set of RhaS/RhaBAD, LacI/Trc with AraC/AraBAD cassette was included in this part of the investigation. Tested orders are depicted in Fig. 6D.

Results showed, that each organisation of the plasmid displayed similar expression of the FPs (Fig. S7), but some slight cross-interactions can be pointed out. When the RhaS/RhaBAD cassette was placed on 2nd or 3rd position, in samples without rhamnose but only with IPTG or arabinose, mCherry relative expression was detected (Fig. S7A-E). This effect is stronger when RhaS/RhaBAD cassette is placed behind LacI/Trc cassette. No other clear interactions was detected. Overall, the expression levels of all three fluorescent proteins were comparable across the tested plasmid configurations.

One-step production of flavonoid rhamnosylation cascade

As an example of practical application of designed plasmids, a one-step production of a triple-enzyme flavonoid rhamnosylation cascade (Fig. 8) was performed. This in vitro cascade includes: sucrose synthase from Glycine max (GmSuSy) [28], chimeric UDP-rhamnose synthase (VvRHM_NRS/ER) [29], and flavonol 7-O-rhamnosyltransferase from Arabidopsis thaliana (AtUGT89C1) [30]. Cooperation of this three enzymes results in two-step recycling system of UDP-rhamnose from UDP and sucrose with UDP-glucose as intermediate, and fructose as an side product (Fig. 8). Chimeric three-functional VvRHM_NRS/ER holds UDP-glucose 4,6-dehydratase, UDP-4-keto-6-deoxy-glucose 3,5-epimerase, and UDP-4-keto-rhamnose 4-keto-reductase activities and is composed of N- and C-domains from different origins – Vitis vinifera and Arabidopsis thaliana, respectively. Thanks to that it has an inner recycling system of its co-factors NAD⁺ and NADH (native rhamnose synthases utilises both NAD⁺ and NADP⁺ cofactors), and makes this cascade independent of the addition of not only co-substrates but also co-factors.

Fig. 8

Scheme of triple-enzyme cascade for flavonoid rhamnosylation. Abbreviation: UDP – uridine diphosphate; NAD - nicotinamide adenine dinucleotide; SuSy – sucrose synthase; RHM – UDP- rhamnose synthase; RT – rhamnosyltransferase

Firstly, the optimal working ratio among the three enzymes was determined to facilitate the selection of an expression cassette with suitable expression strength for each enzyme. Reactions were set up with purified enzymes (Fig. S8A-C) in three rounds, each time varying the concentration of one enzyme while keeping the amounts of the other two constant (Table S11). As the concentration of AtUGT89C1 increased from 0.002 to 0.2 mg/mL, the conversion percentage increased, peaking at 82.8% at 0.2 mg/mL. Beyond this concentration, the conversion of naringenin decreased (Table S11). Reactions with VvRHM_NRS/ER concentrations below 0.04 mg/mL showed no cascade activity, while concentrations in the range of 0.2-1 mg/mL resulted in similar conversion rates between 96% and 100% (Table S11). GmSuSy already at relatively low concentration (0.02 mg/mL) resulted in satisfying conversion of 75.6%, peaking at 98.6% at 1 mg/mL. However, reactions with varying GmSuSy concentrations lasted only 30 min. to capture conversions below 100%, unlike the other reactions which lasted 1 h. Considering all results, VvRHM_NRS/ER was recognised as a limiting step in the cascade, hence the strongest designed expression cassette (LacI/Trc) was chosen for its production. For AtUGT89C1 and GmSuSy expression RhaS/RhaBAD and XylS/Pm systems were selected, respectively.

All three cassettes were cloned into one final monocistronic system in following CDS order: GmSuSy, VvRHM_NRS/ER and AtUGT89C1 resulting in pTriple_cascade plasmid (Table S1). For comparison of enzyme production efficiency, an alternative monocistronic plasmid with all genes under the control of RhaBAD promoters was constructed – pRhaS_cascade (Table S1). Small scale His-Trap purification of crude protein extracts from strains holding pTriple_cascade and pRhaS_cascade plasmids showed three distinct bands corresponding to three co-expressed enzymes (Fig. 9A). In addition, a test of independent induction of each gene was performed. In samples where only one inducer was added, the major band corresponded to the induced enzyme. However, in each sample all three bands appeared due to the previously described crosstalk between the expression systems (Fig. 9A) – to exclude the interference of the medium composition in the induction process, cascade production was carried out in minimal M9 medium.

Fig. 9

Production of flavonoid rhamnosylation cascade A SDS-PAGE gel showing co-expressed GmSuSy, VvRHM_NRS/ER and AtUGT89C1 purified on His SpinTrap columns from cell lysate of differently induced. E. coli harbouring pTriple_cascade (Lane 3–6) or pRhaS_cascade (Lane 2) plasmid. Lane 1 – PageRuler Plus Prestained Protein Ladder; Lane 3 – rhamnose induction; Lane 4 – IPTG induction; Lane 5 - m-toluic acid induction; Lane 6 – induction with all three inducers. Proteins mass – GmSuSy – 92 kDa, VvRHM_NRS/ER – 76 kDa, AtUGT89C1 –48 kDa. B Conversions of 0.05 mM naringenin to naringenin 7-O-rhamnoside conducted using corresponding from SDS-PAGE gel cell lysates

Results of conducted reactions showed that cascade is active in cell lysates. The best conversion of naringenin was reached in reaction with cell lysate from fully induced strain with pTriple-cascade plasmid.

Discussion

This study presents a detailed characterisation of new expression plasmids with four different bacterial expression systems built on and compatible with GS MoClo assembly [13], dedicated to recombinant protein production in E. coli. Their architecture enables the design of a monocistronic expression cassette containing up to three genes of interest, each regulated by individually inducible promoters, however addition of more expression cassettes is possible to due to high versatility and modularity of GS MoClo, provided the plasmid size limits are not exceeded. Application of SEVA and GS MoClo assembly also enables easy switch in antibiotic resistance or origin of replication. Most importantly, a such genetic circuit can be constructed combinatorial within only two rounds of modular cloning, using a minimal set of genetic parts and result in a broad range of variants for screening studies. Selected bacterial regulator/promoter systems are governed by both positive and negative regulatory mechanisms, and mainly originate from bacterial carbohydrate metabolism. They are also functional in various proteobacteria strains. RhaS/RhaBAD and LacI/Trc cassettes were already described in our previously published paper [27]. They were used to construct a double expression system for the production of a two-enzyme cascade. However, this investigation presents a more comprehensive characterisation.

RhaS/RhaBAD cassette was based on the E. coli rhamnose operon, which is responsible for rhamnose transport and metabolism. RhaBAD promoter characterises low basal transcriptional activity and slow response, which is beneficial for expressing toxic proteins [31]. It is naturally regulated by two activators, RhaR and RhaS. In the presence of rhamnose constitutively expressed RhaR binds to RhaRS promoter and initiates expression of RhaR and RhaS TFs. The latter protein, linked with rhamnose, binds to the RhaBAD promoter and activates the expression of downstream genes [32]. Kelly et al. reconfigured this two-step regulation and proved that only terminal TF (RhaS) is needed to control expression from the RhaBAD promoter [33], and this simplified configuration was implemented herein. RhaS gene was placed under constitutive J23103 promoter to ensure the constant presence of this TF [34]. Similarly to other investigations on this expression system [33,34,35], our plasmids with RhaS/RhaBAD cassette has low basal expression, fine expression strength and satisfying titrability. Recently, it was verified that the well-known dose-dependent regulation of the induction strength from the RhaS/RhaBAD-related systems comes from rhamnose consumption, not from real tunability [36]. Authors were able to engineer the dose-dependent regulation by deleting gene of rhamnose-proton symporter (rhaT) and hampering rhamnose digestion. Thanks to that, the sensitivity of the system increased. Nevertheless, in both configurations expression titration is possible, regardless of the mechanism. Moreover, the RhaS/RhaBAD system does not cross-talk with other inducers, apart from interaction with IPTG, which was already described in our previous paper [27]. What is interesting, the strain harbouring pRhaBAD_12_mCherry plasmid in the presence of rhamnose had postponed the exponential phase of the growth – the higher the rhamnose concentration, the more growth was delayed. Such strong response was not detected for plain E. coli BL21 (DE3), which suggests that growth reduction must be related mainly to the heterologous expression from the recombinant plasmid. This is evident when analysing OD₆₀₀ and fluorescence curves over 24 h cultivation (Fig. S1). As soon as the mCherry fluorescence stops increasing, the culture enters the exponential phase. Surely, at this time point, rhamnose is completely metabolised. To our best knowledge, any similar observations have been reported in the literature, making it difficult to explain the underlying mechanism. However, it is possible that constitutive, unregulated expression of RhaS, may be the cause. RhaS belongs to the vast AraC TFs family [31], and it is possible, that when overproduced it may non-specifically bind to other AraC-like binding sites and downregulate other metabolic pathways, crucial for E. coli growth. This can be proven by the fact, that in native rhamnose operon, RhaS downregulates its own expression [34]. Here this mechanism is disabled, along with the catabolite repression effect [32], by replacing the RhaRS promoter with the J23103 promoter. Moreover, bacteria incorporates rhamnose into the cell surface polysaccharides, that are essential for interactions between cells [37]. Thus, variations in rhamnose concentrations in the cytosol, related to the external addition of rhamnose, may also affect their synthesis and condition of the cells. Nonetheless, to prove such a mechanism a detailed transcriptomic/proteomic analysis is needed, which is beyond the scope of this investigation. Additionally, it would be worth checking how the deletion of the genes responsible for rhamnose anabolism and catabolism, would affect expression from RhaS/RhaBAD system, as well as E. coli growth.

Trc promoter, which is a hybrid of Trp and LacUV5 promoters, belongs to IPTG-inducible promoters, negatively regulated by a well-known LacI transcriptional factor [24]. In the absence of allolactose or its analogue, such as IPTG, LacI binds to the promoter region and disables expression. As soon as the inducer appears in the cell, the LacI-inducer complex dissociates from the promoter side and enables the transcription of downstream genes. Trc promoter is renowned for its high expression strength, which can lead to the accumulation of up to 30% heterologous protein in the total cell protein content [6]. As predicted, the LacI/Trc cassette exhibited the highest expression strength among designed plasmids, as well as the highest level of basal expression. Moreover, a satisfying titrability at low IPTG concentrations was observed. Interestingly, in the presence of m-toluic acid and rhamnose, the LacI/Trc cassette exhibited elevated expression. As the interactions with rhamnose could be explained by the carbohydrate origin, co-action with m-toluic acid is more surprising. Once again, a transcriptomic/proteomic analysis could shed some light on this. Without it, the underlying reasons are unknown. Nevertheless, both compounds were unable to non-specifically induce Trc promoter in the absence of IPTG, so the cross-talk is not problematic in terms of combining both systems into one expression plasmid. Regarding the cross-reactivity between the IPTG and the other cassettes, a strong inhibition of the AraC/AraBAD cassette was observed. This effect has already been described in the literature [38], but the inhibition detected in this study turned out to be much stronger. According to Lee et al. IPTG can bind to the AraC protein and impair the induction of the AraBAD promoter [38]. However, this is supposed to appear only at low arabinose concentrations (≤ 1 mM). In our investigation, minimal expression even with 10 mM arabinose and 1 mM IPTG was observed. In contrary to the RhaS/RhaBAD cassette, the AraC promoter wasn’t changed, so the results should correspond to other investigations, and the reason for that strong inhibition remains unclear. To omit that issue, induction with lactose and its influence on other expression systems was verified. Lactose does not inhibit expression from AraC/AraBAD and any other expression system, but due to the lactose metabolism results in expression levels that are nearly three times lower compared to induction with IPTG [39]. Thus, substituting IPTG with lactose when using both expression systems in one strain could be a solution, but it would come at the price of reduced productivity in the LacI/Trc cassette. However, analysing expression from triple-gene plasmids, where both LacI/Trc and AraC/AraBAD system were present, and induced with both arabinose and IPTG, a much weaker inhibition effect (or none) was observed. Since IPTG naturally should have a higher affinity for LacI than AraC, constitutively expressed LacI prevents significant inhibition of the AraBAD promoter, allowing for the parallel use of both expression systems in a single strain. Nonetheless, to eliminate this cross-talk a variant of AraC, with increased insensitivity to IPTG [38], could be applied in the AraC/AraBAD cassette design.

AraC/AraBAD architecture was based on the E. coli arabinose operon, responsible for arabinose metabolism and transport. In the absence of arabinose, AraC dimer binds to the AraBAD promoter region and precludes transcription. When arabinose attaches to the AraC, changes its configuration, and enables the expression of downstream genes [40]. AraC also downregulates its expression, by binding to the Pc promoter, which overlaps with the AraBAD promoter region [41]. Due to this close relationship between both promoters, we decided not to change their native configuration. AraC/AraBAD expression system turned out to have the lowest expression level, with minimal basal expression and satisfying titrability. Arabinose does not activate other expression systems, and aside from IPTG inhibition, the AraC/AraBAD cassette does not interact with other inducers. This makes it suitable for incorporation into higher genetic circuits alongside other expression systems. The weak performance of the AraC/AraBAD system results not only from lower promoter strength but also from E. coli carbohydrate metabolism. Utilised in this study the E. coli strains can consume arabinose, and its concentration depletes over time. Genes responsible for arabinose digestion could be deleted to increase the productivity. However, low expression strength can be beneficial for the production of toxic proteins, hence developing that type of expression system is also reasonable. What is more, the AraBAD promoter is positively controlled by the cyclic AMP and catabolite activator protein (CAP) due to the well-known catabolite repression mechanism [42]. Therefore, we decided to investigate how different carbon sources affect the induction of the tested expression systems, especially since not only the AraC/AraBAD can be affected [34, 43, 44]. We chose to evaluate glucose and glycerol – a biodiesel by-product, preferred feedstock for large-scale cultivation. All cassettes, except for RhaS/RhaBAD, achieved higher expressions of mCherry growing on glycerol than glucose. The presence of glucose as the most preferred carbon source causes catabolite repression, and restricts expression from promoters, which regulates the metabolism of other, less favourable carbohydrates [45]. In the RhaS/RhaBAD system, the RhaS promoter was exchanged for the J23103 promoter, successfully decoupling induction from catabolite repression. The highest expression increase, nearly threefold, was detected for AraC/AraBAD cassette. LacI/Trc and XylS/Pm systems gained around 50 and 90% of expression level, respectively. However, it needs to be highlighted that cultures with glycerol as a sole carbon source grow much slower, achieving three to four times lower OD₆₀₀ than cultures with glucose. Therefore, despite eliminating catabolite repression and the lower cost, using glycerol as a culture feed requires extended cultivation time or larger culture volumes to achieve the same biomass yield. An interesting approach could be also the design of a mixed feed system [46] or auto-induction medium, compatible with each carbohydrate based expression system.

XylS/Pm system originates from Pseudomonas putida TOL plasmid pWW0 and controls the expression of aromatic hydrocarbons degradation pathway [47]. XylS belongs to AraC/XylS transcriptional factors family and positively regulates transcription from the Pm promoter [48]. In the native configuration of this expression system, xylS gene is under the control of two overlapping promoters - inducible Ps1 and weak constitutive Ps2 [49], which here are replaced by one constitutive J23114 promoter. XylS TF in the presence of its inducer (aromatic hydrocarbon) dimerizes and binds to the Pm promoter region, inducing transcription [50]. However, XylS dimerization can also occur at high intracellular XylS concentrations [51]. What is important, only protonated inducer molecules can passively diffuse into the cells [52]. Thus, monitoring the pH of the culture is vital while utilising this expression system. Results showed that the designed XylS/Pm system has moderate expression strength and nice titrability up to 0.0025 mM m-toluic acid concentration. The high basal expression can result from impurities in the growth medium, that non-specifically induced expression, but also from the constitutive expression of XylS from medium-strong J23114. This can lead to auto-induction by too high XylS concentration in the cell. XylS/Pm and its inducer do not cross-talk with other expression systems and their inducers, which makes it suitable for inclusion in more complex genetic circuits.

With the defined characteristics of single-gene expression cassettes established, we constructed triple-gene expression cassettes, with all genes arranged in a monocistronic configuration, and tested their potential for multiple gene expression. Several plasmid-based strategies for protein co-expression were developed, but few of them enable independent control of the expression of each gene of interest or rely on handy modular cloning assembly, which are crucial in the process development. Existing commercial backbone vectors, such as pETDuet (Merck KGaA, Darmstadt, Germany), most often include one type of promoter, preferably the T7 promoter [53]. Apart from using varying promoters, the application of RBSs with different strengths can differentiate expression levels to some extent, but it is not fully steerable [54]. Another strategy for protein co-expression is the utilisation of several plasmids with different copy numbers, to vary the expression level of the recombinant proteins [16]. However, this technique requires additional environmental pressure and increases complexity and the costs of the process. Promoter induction remains the most reliable method for achieving dose-dependent and predictable expression regulation. Nonetheless, co-expression strategies based on this approach are still relatively rare. A noteworthy example is the study by Meyer et al. [8], where engineered regulatory elements (transcriptional factors and transporters) from 12 biosensors, were integrated into the E. coli genome, enabling the independent control of the recombinant expression from 12 different promoters. As an example they cloned and balanced five-enzyme lycopene biosynthesis pathway, which is an notable achievement. Our approach focused on the development of a single-plasmid strategy, where all necessary regulatory sequences were integrated into one recombinant plasmid. This aimed firstly to minimise resources required to modular cloning elements, and secondly to evaluate the capacity of such multigene plasmids. The work by Meyer et al. demonstrated that their strains are genetically stable and that their method minimise the size of the required plasmid. While we acknowledge that cloning 12 genes of interest using our approach would be infeasible due to plasmid size limitations, we did not observe a significant drop in cloning efficiency with our triple-gene plasmids encoding fluorescent proteins. We did observe a reduced success rate during the cloning of the pTriple_cascade plasmid. Notably, even the single-gene plasmid carrying VvRHM-NRS/ER proved challenging to obtain, likely due to the potential toxicity of this protein. Therefore, the issues encountered with the pTriple_cascade plasmid may be attributed to the specific gene rather than the cloning strategy itself proving that it can be useful for at least three-enzyme pathway expression. The combination of three out of four investigated expression systems, into one monocistronic genetic circuit, succeeded in simultaneous and independent expression of three genes of interest. Moreover, each expression cassette maintained titrability observed in single-gene plasmids. Interestingly, the observed cross-talks in single-gene plasmids changed, as there was no longer AraC/AraBAD inhibition by IPTG, as described above. Moreover, rhamnose seemed to boost up (or even induce) the expression of other cassettes when combined in the triple-gene plasmid, especially LacI/Trc. The reason for this could be related to the reduced growth rate of the E. coli strain in the presence of rhamnose, even stronger when holding triple- than single-gene plasmid. The high induction rate and leakiness of the Trc promoter, combined with slow growth, resulted in a higher amount of recombinant protein per cell, captured in the relative expression unit. The high variance in the results reflecting relative expression from LacI/Trc may also results from the choice of the reporter protein, Electra1. Its detection wavelength is close to the native fluorescence of E. coli cultures, leading to a high fluorescence background. This also makes Electra1 fluorescence more sensitive to changes in the optical density of the cultures caused by rhamnose addition, compared to mCherry and GFPmut3*. This should be kept in mind when analysing the results. Nevertheless, we chosen the selected set of fluorescent proteins on the basis of minimal overlap with one another, which otherwise would cause more problematic interferences. Additionally, the transcriptional read-through was observed, when LacI/Trc was followed by RhaS/RhaBAD, which led to some leakage of mCherry expression. The observed cross-talks between expression systems did not prevent the control of each gene’s expression. Yet, some redesign of the RhaS/RhaBAD cassette or deletion of the native rhamnose metabolic pathway seems crucial for a complete understanding of the detected interactions.

The practical test of the developed expression system – producing the triple-enzyme flavonoid rhamnosylation cascade – was also successful. All three enzymes were detected in the cell lysate. However, when cultured in the rich Dynamite medium, VvRHM_NRS/ER, driven by the strong LacI/Trc expression system, was predominantly expressed as inclusion bodies (data not shown), which negatively affected the host cells. Cultures induced by IPTG exhibited only half the growth compared to the other cultures, resulting in much weaker expression of all recombinant proteins. This issue was resolved by switching to M9 minimal medium, which reduced nonspecific induction caused by medium components and possibly improved balance in recombinant expression due to a lower growth rate. Finally, the best cascade activity and presumably the best expressed enzyme proportions was attained from fully induced strain carrying pTriple_cascade plasmid. Despite, the optimisation of the one-step production of the cascade, based on the enzymes’ activities, brought forward acceptable results, the potential toxicity of the recombinant proteins and their tendency to form inclusion bodies have to be carefully considered while choosing the expression strength. It is especially important in the aspect of multigene co-expression, which may even more unpredictably affect the vitality of the host.

Conclusions

In summary, this study presents a plasmid-based strategy for the independent and steerable expression of up to three recombinant proteins in one bacterial host, using different expression systems. Furthermore, it delivers a comprehensive overview of the practical aspects of using the developed plasmids. The investigation showed that each designed expression cassette has different expression strength – ranging from low (AraC/AraBAD, XylS/Pm), and medium (RhaS/RhaBAD), to strong (LacI/Trc) – and that expression from all of them can be controlled in dose-dependent manner by the inducer concentration. The examination of the nonspecific cross-talks between the cassettes and inducers unveiled that IPTG can effectively inhibit expression from the AraC/AraBAD system, but this influence fades in the higher level plasmids in the presence of the LacI/Trc system. Moreover, it was observed that strains holding plasmids with the RhaS/RhaBAD system have postponed exponential phase when supplemented with rhamnose. Most probably, it is caused by the nonspecific inhibition of the host’s metabolic pathways by the constitutively expressed RhaS transcriptional factor in complex with rhamnose. Further, the influence of the carbon source on the investigated expression cassettes was characterised. The LacI/Trc, AraC/AraBAD, XylS/Pm systems exhibited response to the catabolic repression, while RhaS/RhaBAD was successfully decoupled from this mechanism. The simultaneous co-expression of three fluorescent proteins, using different combinations of the characterised expression cassettes, confirmed they ability to independently function on the same plasmid. Finally, the attempt to the production of the rhamnosylation cascade demonstrated that also expression of functional proteins can be performed with developed strategy, and can be used for complex biocatalysts preparation. However, it have to be underlined, that the production optimisation cannot be based only on the enzymes activity, but also on their potential toxicity and inclusion bodies formation, to ensure the planned enzyme ratio in the final batch. Outlined in this study details about the cross-reactivities, dynamic ranges, and boundaries of recombinant proteins co-expression, should be valuable for other investigations in the field.