Journal of Industrial Microbiology and Biotechnology, 2022, 49, kuac018 https://doi.org/10.1093/jimb/kuac018 Advance access publication date: 6 June 2022 Fermentation, Cell Culture and Bioengineering – Original Paper Process intensification at the expression system level for the production of 1-phosphate aldolase in antibiotic-free E. coli fed-batch cultures 1 ,2 Martina Pasini , Alfred Fernández-Castané3 ,4 , Gloria Caminal2 ,5 , Tim W Overton6 ,7 , Pau Ferrer2 1 Aston institute of Photonic technologies (AiPT), Aston University, Birmingham, B4 7ET, UK Downloaded from https://academic.oup.com/jimb/article/49/4/kuac018/6601392 by guest on 17 October 2022 2 Department of Chemical, Biological, and Environmental Engineering, Escolad’Enginyeria, Universitat Autònoma de Barcelona, Bellaterra (Cerdanyola del Vallès) 08193, Catalonia, Spain. 3 Aston Institute of Materials Research, Aston University, Birmingham, B4 7ET, UK 4 Energy and Bioproducts Research Institute, Aston University, Birmingham, B4 7ET, UK 5 Institute of Advanced Chemical of Catalonia, IQAC−CSIC, 08034, Barcelona, Spain 6 School of Chemical Engineering, College of Engineering and Physical Sciences, University of Birmingham, Edgbaston, Birmingham, B15 2TT, UK 7 Institute for Microbiology and Infection, University of Birmingham, Edgbaston, Birmingham, B15 2TT, UK Correspondence should be addressed to: Martina Pasini. Phone: +44 7480295916. E-mail: m.pasini@aston.ac.uk Abstract: To successfully design expression systems for industrial biotechnology and biopharmaceutical applications; plasmid stability, efficient synthesis of the desired product and the use of selection markers acceptable to regulatory bodies are of utmost importance. In this work we demonstrate the application of a set of IPTG-inducible protein expression systems – harboring different features namely, antibiotic vs auxotrophy marker; two-plasmids vs single plasmid expression system; expression levels of the repres- sor protein (LacI) and the auxotrophic marker (glyA) – in high-cell density cultures to evaluate their suitability in bioprocess conditions that resemble industrial settings. Results revealed that the first generation of engineered strain showed a 50% reduction in the pro- duction of the model recombinant protein fuculose-1-phosphate aldolase (FucA) compared to the reference system from QIAGEN. The over-transcription of glyA was found to be a major factor responsible for the metabolic burden. The second- and third-generation of expression systems presented an increase in FucA production and advantageous features. In particular, the third-generation ex- pression system is antibiotic-free, autotrophy-selection based and single-plasmid and, is capable to produce FucA at similar levels compared to the original commercial expression system. These new tools open new avenues for high-yield and robust expression of recombinant proteins in E. coli. Keywords: High-cell-density fed-batch cultures, Recombinant protein production, Antibiotic-free expression system, Escherichia coli, Bioprocess optimization Introduction vast majority of expression systems for lab-based investigations include antibiotic resistance markers that enable screening of Escherichia coli is the most widely used host for the development of positive clones and help in plasmid retention. However, the use of bacterial expression systems for overexpression of recombinant expression systems with antibiotic markers is considered unac- proteins and remains the first choice for laboratory investigations ceptable by regulatory authorities in relevant areas of industrial and industrial production of recombinant proteins within the biotechnology (e.g., medical, therapeutic, and agricultural appli- biopharmaceutical sector (Rosano & Ceccarelli, 2014; Castiñeiras cations; Glenting & Wessels, 2005) and follows current trends in et al., 2018). The main advantages of E. coli are rapid growth rate, improvements in antibiotic use and stewardship to reduce antimi- capability to reach high cell densities, growth on inexpensive sub- crobial resistance (Mignon et al., 2015). strates, well-characterized genetics, and the availability of excel- In order to produce high levels of protein, it is often useful to lent tools for genetic manipulation (Brown, 1995). clone the gene of interest downstream to a well-characterized, A wide range of expression systems with different features strong, and tightly regulated promoter. The E. coli lac promoter have been developed for recombinant protein production (RPP) in is arguably the most extensively studied (Germán et al., 2019). E. coli, comprising plasmids for constitutive or inducible expres- Within the realm of E. coli expression, the T5 promoter system sion, different mechanisms of induction (e.g., temperature and (and its derivatives) available from QIAGEN is one of the most IPTG-inducible systems), the use of strains with different geno- popular choices for protein production (Fernández-Castané et al., types (Waegeman & Soetaert, 2011), and additional features to aid 2012a; Calleja et al., 2014). This IPTG-induced promoter has a high production (e.g., use of tuneable strains or co-expression of chap- transcription rate that can only be tightly regulated in the pres- erones; Marschall et al., 2017). Moreover, expression systems may ence of high levels of the Lac repressor protein (LacI), which is incorporate fusion tags to increase protein solubility or enable pu- encoded by the lacI gene (Xu & Matthews, 2009). The commercial rification using affinity chromatography (Costa et al., 2014). The pQE- vectors commercialized by QIAGEN utilize two lac operator Received: April 28, 2022. Accepted: May 4, 2022. © The Author(s) 2022. Published by Oxford University Press on behalf of Society of Industrial Microbiology and Biotechnology. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. 2 | Journal of Industrial Microbiology and Biotechnology, 2022, Vol. 49, No. 4 (lacO) regions in order to guarantee strong repression under non- production levels of the protein of interest in shake flasks ex- induced conditions (Rosano & Ceccarelli, 2014). periments (Pasini et al., 2016). Finally, the third generation strain It is widely recognized that RPP consumes cellular energy and (‘‘AmpR– ’’) was generated by deleting the ampicillin resistance metabolites (Rahmen et al., 2015), whereby high level expres- gene (bla) from the pQE-FucA plasmid, resulting in solely aux- sion of heterologous proteins has a direct impact on the central otrophic selection. metabolism of cells and causes the activation of stress responses In this work, we demonstrate that the third generation of (Weber et al., 2002; Neubauer et al., 2003). This physiological re- antibiotic-free pQE-derived expression systems developed re- sponse is known as the metabolic burden caused by RPP draining cently in our laboratory can be successfully used in industri- the host cell’s resources, either in the form of energy such as ATP ally relevant bioprocessing strategies (i.e., in high-cell-density or GTP, or nutrients such as amino acids and nucleotides that are cultures) for the expression of FucA, an enzyme of interest for required to express and synthesize the protein of interest and its biocatalytic pharmaceutical applications (Koeller & Wong, 2000; associated cellular machinery (Glick, 1995). Typically, cells over- Clapés et al., 2010) . Growth and production of FucA were com- come the metabolic burden by triggering stress-response mecha- pared in the reference and three generations of auxotrophic sys- nisms that adapt and readjust the metabolism in order to restore tems, and cell physiology was measured using FCM. Our third gen- Downloaded from https://academic.oup.com/jimb/article/49/4/kuac018/6601392 by guest on 17 October 2022 functionality and viability. However, such cellular responses often eration system is advantageous as it does not require antibiotics negatively affect growth parameters such as growth rate, biomass and comprises a single plasmid. yield, and cellular viability, as well as recombinant protein pro- ductivity (Carneiro et al., 2012). Plasmid-based expression systems are not always stable, par- Materials and Methods ticularly in cultivation strategies where cells are grown for many Strains and Plasmids generations, that is, in high-cell-density or continuous cultures The K-12 derived E. coli M15 (QIAGEN) and M15glyA strains were (Pierce & Gutteridge, 1985). Hence, vector stability is of utmost im- used for recombinant FucA expression. These strains were stored portance. Expression systems using two or more plasmids may at –80°C in cryo-stock aliquots prepared from exponential phase present high instability, thus enabling their use only as ‘‘tran- grown cultures in Luria-Bertani (LB) medium. The list of E. coli sient’’ expression systems (Al-Allaf et al., 2013). During cell growth strains, plasmids, and the expression systems used in this study in high-cell-density cultures, overexpression of recombinant pro- and their abbreviated names are summarized in Table 1. teins often results in (i) cell aggregation and (ii) bacteria becoming viable but non-culturable, a phenotype which means the loss of the ability to form colonies on agar plates (Jeong & Lee, 2003; Wyre Culture Media & Overton, 2014a; Fernández-Castané et al., 2017). Flow cytometry All chemicals and reagents were purchased from Sigma-Aldrich (FCM) has been extensively used as a rapid and high-throughput unless otherwise indicated. Luria-Bertani medium, containing analytical technique to determine the viability and physiology of 10 g L–1 peptone, 5 g L–1 yeast extract, and 10 g L–1 NaCl, was used cells and to detect heterogeneities in cultures (Hewitt et al., 1999; for pre-cultures. Wyre and Overton, 2014b). Defined medium (DM) used for shake flasks cultures and biore- Previous studies carried out in our laboratory developed aux- actors contained per liter: 5 g glucose; 2.97 g K2 HPO4 ; 0.60 g otrophic selection systems based on glycine auxotrophy derived KH2 PO4 ; 0.46 g NaCl; 0.75 g (NH4 )2 SO4 ; 0.11 g MgSO4 ·7H2 O; 0.006 g from the commercial two-plasmid pQE-based expression system FeCl3 ; 0.025 g thiamine; 1.44 g CaCl2 ·2H2 O; and 0.7 mL of trace el- (QIAGEN), comprising an expression plasmid (pQE), where the T5 ements solution (TES). TES contained per liter: 0.04 g AlCl3 ·6H2 O; promoter regulates expression of the recombinant gene of inter- 1.74 g ZnSO4 ·7H2 O; 0.16 g CoCl2 ·6H2 O; 2.18 g CuSO4 ·5H2 O; est, and pREP4, which expresses lacI that allows tight regulation 0.01 g H3 BO3 ; 1.42 g MnCl2 ·6H2 O; 0.01 g NiCl2 ·6H2 O; and 0.23 g of the T5 promoter on the pQE plasmid (Gronenborn, 1976). The E. Na2 MoO4·5H2 O. coli glyA gene encoding serine hydroxymethyl transferase (SHMT), Stock solutions of kanamycin and chloramphenicol were dis- which catalyzes the reversible interconversion of L-threonine and solved in dH2 O at a concentration of 100 mg mL–1 and 30 mg mL–1 , glycine and is essential for cell growth in the absence of glycine respectively and stored at –20 °C. Ampicillin was dissolved in ab- (Plamann & Stauffer, 1983), was used as a selective marker. In solute EtOH and prepared at a concentration of 100 mg L–1 and the first generation auxotrophic system (Vidal et al., 2008), glyA stored at –20 °C. IPTG stock was prepared at 100 mM, and stored was cloned into the pQE-FucA vector expressing fucA, encod- at –20 °C. ing fuculose-1-phosphate aldolase (FucA). The resultant plasmid, Vitamins, antibiotics, TES, FeCl3 , MgSO4 ·4H2 O, CaCl2 ·2H2 O, and pQEαβFucA, was transformed into the glycine auxotroph E. coli inducer were sterilized by filtration (0.2 μm syringe filter made M15glyA[pREP4] yielding M15glyA[pREP4] pQEαβFucA. Expres- from a blend of cellulose esters, Sartorius). Glucose and saline sion studies in shake flask cultures revealed that this strain pre- solutions were separately sterilized by autoclaving at 121°C for sented a slight decrease in recombinant protein yield compared 30 min. to the commercial reference strain (Pasini et al., 2016). Feeding media for the fed-batch phase contained per litre: In the second generation of the expression system (named 490 g glucose; 9.56 g MgSO4 ·7H2 O; 0.49 g FeCl3 ; 0.33 g thiamine; ‘‘Puzzle’’), we optimized expression levels of lacI and glyA us- 0.10 g CaCl2 ·2H2 O; and 63 mL of TES. Phosphates (P) were not in- ing the weak, constitutive J23110 synthetic promoter (Registry for cluded in the feeding solution in order to avoid co-precipitation Standard Biological Parts, http://parts.igem.com) and cloned both with magnesium salts. Instead, a 5-mL pulse of a concentrated as cassettes into the pQE-FucA vector, which was transformed phosphates solution containing 500 g L−1 K2 HPO4 , and 100 g L−1 into E. coli M15glyA, removing the need for the pREP4 plasmid KH2 PO4 was added every 30-OD600 step increase based on a cal- and thus alleviating the metabolic burden. This resulted in im- culated biomass from phosphates yield (YX/P ) of 18 g DCW g P–1 proved plasmid maintenance while substantially increasing the (Vidal et al., 2003). Pasini et al. | 3 Table 1. Bacterial strains and plasmids used in this work and abbreviation for the four E. coli expression systems used in this study Name Characteristics Reference or source Strains M15 K12 derived QIAGEN (Zamenhof & Villarejo, 1972; Villarejo & Zabin 1974) M15glyA Deletion of the glyA locus of the Vidal et al. (2008) chromosome Plasmids pQE-FucA pQE-40 derived (QIAGEN) with fucA Pasini et al. (2016) gene cloned pREP4 LacIq kanR QIAGEN pQEαβFucA pQE-FucA derived with fragment, Pasini et al. (2016) containing the glyA gene, under Downloaded from https://academic.oup.com/jimb/article/49/4/kuac018/6601392 by guest on 17 October 2022 the P3 promoter transcriptional control pQE-FucA_puzzle pQE-FucA derived with insertion of Pasini et al. (2016) the cassette (J23110-lacI-glyA) for lacI and glyA genes transcription pQE-FucA_puzzle_AmpR– pQE-FucA_puzzle (J23110) derived Pasini et al. (2016) without the bla gene Expression system name Abbreviation Generation Number of plasmids M15[pREP4] pQE-FucA M15[pREP4] Reference system 2 M15glyA[pREP4] pQEαβFucA M15glyA[pREP4] First generation 2 system M15glyA pQE-FucA_puzzle (J23110) Puzzle Second 1 generation system M15glyA pQE-FucA_puzzle (J23110)_AmpR– AmpR– Third generation 1 system Cultivation Conditions The pH was set at 7 and maintained by adding a 15% (v/v) NH4OH Cryo-stocks stored at –80°C, were used to inoculate 50 mL Fal- solution. con tubes with 15 mL of LB medium supplemented with the For experiments performed in the Fermac 310/60 bioreactor, corresponding antibiotic when necessary. Growth was performed the volume of inoculum was 200 mL and the DM volume was overnight at 37 °C at 200 rpm. 1300 mL, making up a total of 1500 mL batch volume in a 5-L jar A total of 3 mL of overnight pre-inoculum were transferred into equipped with 4 baffles and an agitator with 2 six-bladed Rushton a 500 mL conical flask containing 100 mL of DM, following the turbines. Aeration was achieved by sparging air from below the same growing conditions as pre-inoculum cultures for approxi- lower impeller at a rate of 3 L hr–1 (2 vvm). pO2 was measured on- mately 4.5 hr until an OD600 of 1.1–1.2 was achieved. line using a D150 Oxyprobe (Broadley James) and was maintained Bioreactor experiments were carried out using two types of above a set point of 50% by increasing agitation to a maximum of stirred tank bioreactors: Braun Biotech Int. Biostat® B (Germany) 1000 rpm from a minimum of 200 rpm. Off-gas passed through a and Electrolab Fermac 310/60 (Tewkesbury, UK). For both biore- condenser, an autoclavable 0.22-μm filter (Sartorius, Goettingen, actors, the temperature was maintained at 37°C and the pH was Germany), and a HEPA filter (Millipore, Darmstadt, Germany). pH kept at 7.00 ± 0.05 by adding 15% (v/v) NH4 OH solution (NH4 + was measured using an F-695 FermProbe (Broadley James). The from base addition served also as a nitrogen source). Dissolved temperature was maintained at 37 °C by a heating jacket and a oxygen level (pO2 ) was kept above 60% saturation. After glucose cold finger. The pH was set at 7 and maintained by adding a 15% was depleted during the batch phase, feeding was started with the (v/v) NH4OH solution. addition of the feed solution. Bioreactor off gasses were automatically collected for Induction of cultures was carried out by a single-pulse of 70 μM subsequent compositional analysis using a PrimaDB gas- IPTG as previously described by (Fernández-Castané et al., 2012b). chromatograph mass spectrometer (Thermo) and compared During the induction period, 1.5 mL samples were removed from to atmospheric air in order to calculate the O2 uptake rate, CO2 the bioreactor every 10–30 min intervals. One sample was re- evolution rate, and subsequent estimation of the respiratory moved prior to induction for the measurement of basal FucA quotient (RQ). These data were logged automatically by Gas- expression. Works v1.0 (Thermo). The increase of pO2 , O2 , and pH values, For experiments performed in the Biostat® B bioreactor, a vol- together with a decrease in CO2 values, indicate the complete ume of 80 mL of the inoculum culture was added to 720 mL of DM consumption of the 10 g L–1 of initial glucose and the end of the in a 2-L vessel (Vidal et al., 2005; Calleja et al., 2014). pO2 set point batch phase (Supplementary data, Fig. S2-1), (Sooan at al., 2009). was fixed at 60% saturation, and pO2 was controlled by adapting The feeding strategy was performed to maintain the specific the stirring speed between 350 and 1120 rpm and by supplying growth rate (μfix ) constant through the fed-batch phase, using air (enriched with pure oxygen when necessary) at a flow rate of a pre-defined exponential feeding profile (Pinsach et al., 2008). 1.5 vvm. A microburette was used for the discrete addition of feed. Feeding was manually and periodically interrupted and restarted 4 | Journal of Industrial Microbiology and Biotechnology, 2022, Vol. 49, No. 4 in order to avoid glucose accumulation during the induction concentration was quantified using Kodak Digital Science® 1D phase. 3.0.2 densitometry software. Average values of triplicates experiments were plotted with er- Analytical Methods ror bars. The error indicates the confidence interval of 90%. Cell concentration was determined by optical density measure- ments at 600 nm (OD600 ) using a spectrophotometer (Uvicon 941 Flow Cytometry Plus, Kontrol, and Evolution 300 UV–vis, Thermo Scientific, for the Bacterial samples taken directly from the bioreactor were resus- Biostat® B and the Fermac 310/60 bioreactors, respectively). OD600 pended in phosphate-buffered saline to a final concentration of values were correlated to biomass concentration expressed as 105 –106 cells mL–1 and then analyzed directly or after staining Dry Cell Weight (DCW), being 1 OD600 equivalent to 0.3 gDCW L–1 with various fluorescent dyes using a BD AccuriTM C6 flow cy- (Pinsach et al., 2008). tometer (Becton, Dickinson and Company, Oxford, UK). Samples Glucose and acetate concentrations in the fermentation broth were stained with bis (1,3-dibutybarbituric acid) trimethine ox- were analyzed. One milliliter of culture medium was separated onol (0.05 μg BOX mL–1 ) to determine membrane potential; pro- pidium iodide (4 μg PI mL–1 ) to determine membrane integrity; Downloaded from https://academic.oup.com/jimb/article/49/4/kuac018/6601392 by guest on 17 October 2022 from biomass by centrifugation at 14 000 rpm for 6 min and filtered (0.45 μm membrane filter of cellulose esters, Millipore) or Congo red (CR) (40 μg CR mL–1 ) to determine amyloid content prior to analysis. Glucose concentration was determined en- and thus the presence of inclusion bodies (IBs). During FCM on zymatically using an YSI 2070 biochemical analyzer (Yellow fluorescently labeled cells, samples were excited using a 488 nm Spring Systems). Acetic acid was analyzed by HPLC (Hewlett solid-state laser, and fluorescence was detected using two differ- Packard 1050) equipped with an ICSep COREGEL 87H3 ICE-99-9861 ent filters: a 533/30 BP (filter FL1-A) for BOX; and a 670 LP filter (Transgenomic) column and an IR detector (HP 1047), using 6 M (FL3-A) for PI and CR. Particulate noise was eliminated using a H2 SO4 (pH 2.0) as the mobile phase at a flow rate of 0.3 mL min–1 . FSC-H threshold. A total of 20 000 data points were collected at a The column was kept at 40°C. maximum rate of 2500 events s–1 . Data were analyzed using BD The maximum specific growth rate (μmax ) of the strains was CFlow software. estimated from microbial growth curves. Equation 1 shows the re- lationship between the cell concentration (X), maximum specific growth rate (μmax ), and time (t). Log-linearized Equation 2 yields a Results Comparison of Cell Growth, Substrate linear relationship, where the μmax is represented by the slope of Consumption, and Acetate Formation the linear portion in the plot of the natural log of cell concentra- tion versus time. The performance of the three generations of auxotrophic expres- sion systems (Table 1) was compared to that of the initial QIA- GEN pQE/pREP4 system in fed-batch fermentations for the pro- Xt = X0 · eμmax ·t (1) duction of FucA. These strains had previously only been compared in shake flask cultures (Pasini et al., 2016). In this study, triplicate lnXt = lnX0 + μmax · t, (2) fed-batch fermentations were performed using defined medium and glucose as a carbon source in BioStat bioreactors to determine where X0 and Xt are the OD600 or cell concentration at the start the robustness of each expression system in terms of growth yield, and during the exponential phase of growth. plasmid stability, and RPP. The biomass yield, YX/S was calculated using the following The time-profiles of the biomass, glucose consumption, FucA equation: (DCWmax − DCW0 ) mass, specific activity, and acetate production of the four expres- YX/S = . (3) sion systems evaluated in this work are represented in Fig. 1. Key (Glc0 − Glcf ) fermentation parameters are listed in Table 2. DCWmax and DCW0 (g L−1 ) are the maximum and the initial The comparison of the maximum specific growth rate (μmax ) in biomass values, respectively. Glc0 and Glcf (g L−1 ) are initial and the batch phase between the reference commercial M15[pREP4] final value of glucose concentration, respectively. E. coli strain and the three M15glyA-derived generations shows that the latter present slightly lower μmax , decreasing from FucA Quantification 0.49 ± 0.01 hr–1 , of the reference strain to 0.44 ± 0.01 hr–1 , Samples from fermentation broth were withdrawn, adjusted to a 0.45 ± 0.01 hr–1 , and 0.41 ± 0.01 hr–1 of the first, second, and third final OD600 of 4, centrifuged, and subsequently processed as de- generations, respectively (Table 2). Furthermore, in both the ref- scribed elsewhere (Durany et al., 2005; Vidal et al., 2008). Briefly, erence and the first generation strains, the concentration of the pellets were resuspended in 100 mM Tris HCl (pH 7.5). Cell suspen- biomass progressively increased until 2.5 hr after induction and sions were placed in ice and sonicated [VibracellTM model VC50 then stopped around 4 hr after induction, reaching a final value (Sonics & Materials)], over four pulses of 15 s each at 50 W with of 40 g DCW L–1 . While for the third generation strain the fermen- 2 min intervals in ice between each pulse. Cellular debris was then tation ended with a final biomass concentration of 53.5 ± 0.7 g removed by centrifugation, and the cleared supernatant was col- DCW L–1 , with a concentration of dissolved oxygen dropped be- lected for FucA analysis. One unit of FucA activity is defined as low 10%, after 26.3 hr (Fig. 1). the amount of enzyme required to convert 1 μmol of fuculose-1- Substrate uptake rates (qS ) along the induction phase for both phosphate in DHAP and L-lactaldehyde for minute at 25 °C and reference and the first generation strains were calculated, being pH 7.5 (Vidal et al., 2008). 0.37 ± 0.04 and 0.50 ± 0.13 g Glc g–1 DCW hr–1 , respectively. As Total protein content was determined by means of the Bradford a consequence, the first generation strain accumulated higher method using a Coomassie® Protein Assay Reagent Kit (Thermo amounts of acetate throughout, reaching a final concentration of Scientific). To quantify the amount of FucA relative to total intra- 0.54 ± 0.03 g L–1 . Both strains started to accumulate glucose and cellular soluble proteins, NuPAGE® 12% Bis–Tris gels were used acetate in the medium after 2 hr of induction, remaining below according to the manufacturer’s instructions (Invitrogen). Protein inhibitory levels (32 g L–1 and 2.4 g L–1 , respectively) (Fig. 1A and B) Pasini et al. | 5 Table 2. Calculated maximum specific growth rate and biomass from substrate yield of the E. coli strains used in this work along the batch phase Escherichia coli strains μmax (hr–1 ) Yx/s (g g–1 ) M15[pREP4] (reference strain) 0.49 ± 0.02 0.43 ± 0.04 M15glyA[pREP4] (first generation strain) 0.44 ± 0.01 0.36 ± 0.06 Puzzle (second generation strain) 0.45 ± 0.01 0.37 ± 0.09 AmpR– (third generation strain) 0.41 ± 0.01 0.35 ± 0.02 Note. Determination of FucA expression levels. Recently, it has been demonstrated that this ‘‘metabolic burden’’ is not caused by energy limitation; on the contrary, this is due to Downloaded from https://academic.oup.com/jimb/article/49/4/kuac018/6601392 by guest on 17 October 2022 restrictions in anabolic functions (Weber et al., 2021). This energy excess due to RPP causes a reduction of catabolic carbon process- ing, hence, (i) affecting negatively the growth rate; (ii) enhancing acetate formation, and (iii) reducing the carbon substrate uptake, evidenced by the accumulation of glucose (Weber et al., 2002, 2021). The feeding strategy of our fed-batch cultivations was pre- set to maintain constant values of μ and YX/S , and these were not modified during the induction phase, that is, the substrate feeding program assumed these parameters were constant even though they changed as a result of the metabolic imbalance gen- erated by RPP, leading to an accumulation of glucose into the medium. When this state was reached, glucose feeding was man- ually stopped to avoid its further accumulation and re-established once the residual glucose was depleted. Remarkably, the second and third generation strains (each with a single plasmid), unlike the previous two-plasmid systems, did not generate acetate or accumulate glucose during the induction phase (Fig. 1C and D). Thus, the tuning of lacI-glyA expression by the weaker constitutive J23110 promoter leads to a reduction in the transcription levels of these genes, resulting in a reduction of the metabolic load imposed to the cells. In order to evaluate the levels of FucA production in fed- batch cultures of the different strains, specific protein produc- tion (mg g–1 DCW) and activity (AU g–1 DCW) were measured as shown in Figs. 1 and 2, respectively. Overall, FucA activity and FucA production decreased more than 35% when comparing the first generation to the reference strain, which reached a final pro- duction of 240 ± 16 mg FucA g–1 DCW, with an activity of 1214 ± 47 AU g–1 DCW. These values were reduced to 165 ± 13 mg FucA g–1 Fig. 1. Time profiles of fed-batch cultures of FucA production strains DCW and 806 ± 12 AU g–1 DCW, respectively, in the first genera- performed under the following conditions: [IPTG], 70 μM; Xind , 20 g · L–1 ; tion strain (Fig. 2). These effects may be caused by the increase in μfix , 0.22 hr–1 . (A) Reference strain, M15[pREP4] pQE-FucA; (B) first the metabolic burden due to the maintenance of the pQEαβFucA generation, M15glyA[pREP4]; (C) second generation, Puzzle; (D) third expression vector in the first generation auxotrophic strain. The generation, AmpR– . Key to symbols: ( ) biomass DCW (g L–1 ); ( ) glucose concentration (g L–1 ); ( ) FucA activity (AU g–1 DCW); ( ) FucA glyA gene, even though encoded by the low-copy number plas- specific mass (mgFucA –1 DCW); and ( ) acetate concentration (g L–1 ). mid (ColE1 replication origin) under the control of the P3 from the Batch, fed-batch, and induction phases are indicated. Triplicate gen blaTEM-1A (Lartigue et al., 2002), leads to substantially higher fermentations were performed and mean ± SD are presented. The amounts of SHMT protein products which accumulate in the cy- arrows indicate the stop of the feeding. toplasm compared to the reference strain, which contains a single genomic copy of glyA (Pasini et al., 2016). (Kazan et al., 1995; Phue et al., 2005). It is widely known that glu- The maximal FucA specific productivity, both in terms of cose acts as a catabolite repressor, thus repressing the genes in- activity and mass, was highest for the reference strain, being volved in metabolizing other sugars and, in IPTG-inducible sys- 62.04 ± 0.26 AU g–1 DCW hr–1 and 14.95 ± 0.39 mg g–1 DCW hr–1 , tems, to repress the expression of recombinant proteins (Amma et respectively (Table 3). Clearly, the first generation strain showed al., 2018). The accumulation of glucose and acetate during the in- reduced specific productivity of 44.02 ± 0.68 AU g–1 DCW hr–1 duction phase occurred at the same time as a decrease in the FucA and 7.61 ± 0.17 mg g–1 DCW hr–1 . A slight improvement was ob- activity and production in the first-generation system (Fig. 1B). tained in the third generation strain, obtaining a final value of This accumulation is probably due to the high energy demand 48.51 ± 0.31 AU g–1 DCW hr–1 and 8.95 ± 0.39 mg g–1 DCW hr–1 . imposed to the cell for the synthesis of the recombinant protein. Nevertheless, this strain showed volumetric productivity in terms 6 | Journal of Industrial Microbiology and Biotechnology, 2022, Vol. 49, No. 4 the second generation, and 1322 ± 19 AU g–1 DCW with a spe- cific mass of 245 ± 13 mg FucA g–1 DCW for the third generation strain (Fig. 2). Comparing these values with those obtained with the reference strain, both the specific activity and the amount of the recombinant protein are comparable. Remarkably, the glyA strain generations based on a single- plasmid expression system allow for an improvement of FucA pro- duction levels compared to the two-plasmid system. In particu- lar, the down-regulation of the glyA gene is a key factor that is likely to contribute to improve the regulation of fucA expression, thus leading to a higher FucA specific activity. These conclusions are in accordance with the results obtained with the third gener- ation strain. When comparing the first and the third generations, FucA specific activity increased more than 60% (from 806 ± 12 to Downloaded from https://academic.oup.com/jimb/article/49/4/kuac018/6601392 by guest on 17 October 2022 1322 ± 19 AU g–1 DCW) and FucA specific mass increased c.a. 50% (from 165 ± 13 to 245 ± 13 mgFucA g–1 DCW) (Fig. 2). SHMT Production Levels A potential drawback of using SHMT as an auxotrophic marker is accumulation to high concentrations, leading to metabolic bur- den, and resultant decreases in biomass and RPP (Vidal et al., 2008; Pasini et al., 2016). The second and third generation strains aimed to reduce SHMT levels by regulating glyA expression from the weak, constitutive, J23110 promoter (Pasini et al., 2016). Figure 3 depicts the concentration of the SHMT protein both pre-induction and during the induction phases in the high-cell- density fed-batch cultures carried out in Fig. 1. It can be seen how the SHMT levels are around 130 mg SHMT g–1 DCW in the first generation strain, which corresponds to a c.a. 4.5-fold increase compared to the reference system (20 mg SHMT g–1 DCW). The constitutive high SHMT production can be clearly observed in Supplementary data, Fig. S1A, which as an example presents the SDS-PAGE analysis of the samples taken during the induction phase of the first generation fed-batch culture. Hence, it is clear that the glyA gene encoded on a high-copy number plasmid leads to substantially higher amounts of SHMT accumulated in the Fig. 2. (A) Maximum enzyme activity (AU g–1 DCW) and (B) maximum specific mass (mgFucA g–1 DCW) for the strains presented in this study. cytoplasm compared to the reference strain containing a single The dashed line indicates the reference value corresponding to the genomic copy of glyA. reference (M15[pREP4]) strain. However, by tuning glyA expression levels, SHMT production decreased in the single-plasmid constructs (second and third generations), resulting in an overall reduced metabolic burden of enzymatic activity comparable to the reference strain, despite (Pasini et al., 2016). As it can be seen in Fig. 3 and Supplementary still being 22% lower in terms of enzyme mass. data, Fig. S1B, even though there is constitutive overexpression of Unlike the first generation, previous results in shake flasks the SHMT protein in both pre-induction and induction phases, for showed that FucA production improved in the second and third the second and third generation strains, these values were almost generation expression strains, both of which have a single plas- 50% reduced compared to the first generation strains. Thus, these mid, compared to the reference (Pasini et al., 2016). The results data confirm that the reduction of SHMT production is observed obtained in shake flask were consistent with the results pre- in both bioreactors and shake flasks. As an explanatory exam- sented in this study in bioreactor cultures where the maximum ple, 85 ± 0.5 and 72 ± 13 mg SHMT g DCW–1 were produced af- FucA specific production and specific activity increased up to ter 4 hr of induction for the second and third generation strains, 202 ± 18 mg–1 FucA g–1 DCW and 1176 ± 19 AU g–1 DCW for respectively. In contrast, 132 ± 4 mg SHMT g DCW–1 was produced Table 3. Specific productivity and volumetric productivity for both fuca activity and mg for the four strains presented along this work Specific productivity Volumetric productivity Escherichia coli strains AU FucA g–1 DCW hr–1 mgFucA g–1 DCW hr–1 AU FucA L–1 hr–1 mgFucA L–1 hr–1 M15[pREP4] (reference strain) 62.04 ± 0.26 14.95 ± 0.39 2536 ± 44 611 ± 24 M15glyA[pREP4] (first generation strain) 44.02 ± 0.68 7.61 ± 0.17 1652 ± 13 241 ± 80 Puzzle (second generation strain) 45.46 ± 0.26 7.86 ± 0.18 2328 ± 67 407 ± 19 AmpR– (third generation strain) 48.51 ± 0.31 8.95 ± 0.39 2597 ± 56 479 ± 26 Pasini et al. | 7 Downloaded from https://academic.oup.com/jimb/article/49/4/kuac018/6601392 by guest on 17 October 2022 Fig. 3. SHMT production (mg g DCW–1 ) for the four strains presented along this study during pre-induction (pre-Ind) and induction phase of high-cell-density fed-batch cultures run in triplicate. for the reference strain (Fig. 3). These findings are equivalent to a of experiments carried out in Biostat® B and the Fermac 310/60 1.6-fold reduction in the accumulation of the SHMT protein in the bioreactors were similar. cytoplasm as a result of replacing the strong P3 promoter with the Propidium iodide (PI), which only stains cells with compro- weaker J23110 constitutive promoter (Pasini et al., 2016). There- mised membranes, and bis-oxanol (BOX), which only stains cells fore, this enabled a reduction of the metabolic burden imposed with a collapsed membrane potential, were used to determine vi- to the cells. These results are in accordance with the observations ability (Fig. 4). Cells were classified as ‘‘healthy’’ with a membrane by Mairhofer et al., (2013), who demonstrated that the folding ma- potential (PI– /BOX– ), ‘‘damaged’’, being intact but with no mem- chinery is severely overstrained in plasmid-based expression sys- brane potential (PI– /BOX+ ), or ‘‘dead’’, having a non-intact mem- tems compared with plasmid-free cells (i.e., with a single copy of brane and no membrane potential (PI+ /BOX+ ). the gene of interest integrated into the genome) and this is due to As can be seen in Fig. 4A, from the pre-induction sample to the different expression vector dosage. end of the fermentation, the reference strain presented a signif- As stated earlier, similarities can be observed among all the icant percentage of unhealthy and dead cells even under unin- strains when comparing the fed-batch culture results presented duced conditions. The high proportion of damaged or dead cells here with the previous experiments performed in shake flask present before the induction stage is probably due to the pres- cultures (Pasini et al., 2016): (i) the reference strain presents a ence of the antibiotic in the medium, which is used to maintain slightly higher maximum specific growth rate compared to the plasmid stability (Feizollahzadeh et al., 2017). The percentage of new set of engineered strains; (ii) the SHMT production increased damaged cells remained relatively constant through growth (3– significantly when moving from the reference to the to the first 5%). The percentage of dead cells reached the maximum value generation strain; (iii) tuning glyA levels is a key factor to allow (18%) after 15 min of induction, which indicates the stress caused for improved fucA expression regulation, leading to a higher FucA to the cells by the induction of FucA expression. specific activity in the second and third generations strains; (iv) For the two-plasmid based expression systems (the reference the second and third generation strains presented a reduction in and the first generation strains), it can be seen that at the end of the amount of acetate production and glucose accumulation. the fed-batch cultures, the proportion of damaged and dead cells increased (Fig. 4A and B). It is well known that the production of a foreign protein causes additional stress for the host strain and Flow Cytometry Study of Physiological leads to a decrease in the overall cell fitness (Hoffmann & Rinas, Responses Caused by FucA Expression 2004). The dead population at the end of the fermentation corre- To gain new insights into the physiology of the strains upon sponded to 16% and 8% for the reference and the first generation FucA overexpression, FCM analysis was performed on the high- strains, respectively. This reflects the stress caused to the cells due cell-density fed-batch culture experiments performed in Fermac to the induction. The state of the cells of the reference strain is 310/60 bioreactors. The time profiles of the biomass, pH, pO2 , more unhealthy than the first generation strain, probably due to CO2 , O2 , and RQ of the four expression systems evaluated in a higher FucA overexpression and also due to the presence of the these bioreactors are represented as an example in Supplemen- antibiotic. In addition to PI&BOX staining, we also stained cells tary data, Fig. S2. Importantly, pO2 time profile between the set with CR, which stains amyloids such as IBs (Carrió et al., 2005). 8 | Journal of Industrial Microbiology and Biotechnology, 2022, Vol. 49, No. 4 Downloaded from https://academic.oup.com/jimb/article/49/4/kuac018/6601392 by guest on 17 October 2022 Fig. 4. BOX/PI co-staining, indicating the percentage of ( ) live (PI– /BOX– ), ( ) damaged (PI– /BOX+ ), and ( ) dead (PI+ /BOX+ ) cells for the four expression systems presented along this study during pre-induction (pre-Ind) and induction phase (from 5 to 7 hr) of duplicate Fig. 5. Time profiles of the percentage of Congo red-stained cells as high-cell-density fed-batch cultures using the Fermac 310/60 bioreactor. determined by FCM. (A) The reference strain; (B) the first generation; (C) (A) The reference strain; (B) the first generation; (C) the second the second generation; and (D) the third generation. Average values and generation; and (D) the third generation. error bars are presented from two independent experiments. Only at the very end of the fermentation a slight increase of Figure 5A and B shows that, for the reference and first generation both damaged and dead cells appears, for the Puzzle and the strains, there were around 1% of cells containing IBs; for the first AmpR– strains, indicating an initial loss of viability. This is in ac- generation strain, this increased after 3 hr of induction. cordance with previous works, where the number of cells identi- FCM data reveal that for the second generation strain, Puzzle, fied as either being stressed or dead increased at the end of the more than 95% of cells were healthy throughout the fermentation fermentation, due to stress condition (Lewis et al., 2004). This in- (Fig. 4C). Similar observations were obtained for the third genera- crease is probably due to the non-optimal conditions presented tion strain (AmpR– ) (Fig. 4D), with even fewer unhealthy cells (96% at the end of the fermentation, in particular the low pO2 lev- healthy throughout). Almost no change in PI or BOX staining was els due to the high oxygen demands of the culture (data not observed through the induction phase for the second and third shown). This stress condition can also be observed in the FCM generation strains, indicating that physiology was not negatively analysis of cells stained with CR (Fig. 5C and D), showing an in- impacted by RPP. crease in the proportion of cells containing IBs (up to 2% and 3% Pasini et al. | 9 of cells, respectively). This result can be explained by an increase antibiotic-free, which makes it suitable for biopharmaceutical of the stress condition of the protein folding machinery resulting and agritech applications. in an increase of aggregated protein. This can be reflected also We envisage that our antibiotic-free expression system has the in the specific enzyme activity values of the strains presented in potential to become an attractive platform for the production of this study, being: 5.06 ± 0.01 UA mg–1 for the reference strain; a wide range of proteins as well as for other applications, where 4.88 ± 0.01 UA mg–1 for the first generation; 5.82 ± 0.01 UA mg–1 the use of antibiotics is restricted, such as vaccine delivery (Saubi for the second generation; and 5.40 ± 0.01 UA mg–1 for the third et al., 2014). generation. These results indicate that the first generation strain presents the lower FucA enzyme activity/protein concentration ratio indicating a lower folded enzyme activity, in accordance with Acknowledgements the CR results that indicate the higher amount of IBs. It is interest- M. P. acknowledges the Universitat Autònoma de Barcelona ing to note that the second generation strain showed the higher for the pre-doctoral fellowship and EMBO for a Short Term specific enzyme activity. Fellowship. Bacterial exposure to environmental stress (including glu- Downloaded from https://academic.oup.com/jimb/article/49/4/kuac018/6601392 by guest on 17 October 2022 cose/oxygen availability) often triggers protein misfolding and the formation of aggregates as IBs (Kwiatkowska et al., 2008; Supplementary Material Schramm et al., 2020). These results are consistent with those Supplementary material is available online at JIMB (www.academic. presented by Castellanos-Mendoza et al., where IB formation oup.com/jimb). (as detected using CR) increased at uncontrolled pH conditions (Castellanos-Mendoza et al., 2014). The presence of IBs is generally not desirable in the recombi- Author Contributions nant production of enzymes because a further refolding step dur- M.P.: performed all experiments, acquisition, and analysis of all ing purification is often required (Vallejo & Rinas, 2004). In future the data, as well as drafting the manuscript. A.F.C., G.C., and P.F.: studies, a trade-off between maximizing protein production and contributed to the overall conceptual design of the study and correct folding of the enzyme should be taken into consideration data interpretation, as well as in the drafting and revision of if further downstream processing—harvesting, purification, and the manuscript. T.W.O.: contributed to the conceptual design of application of FucA—is intended (Mukhopadhyay, 1997; Kim at el., the study and manuscript editing. All authors read and approved 2020). the manuscript. Funding Conclusion This work was supported by the Spanish MICINN, project number We have demonstrated that by tuning the expression levels of the CTQ2011-28398-CO2-01, the research group 2009SGR281, and by repressor (lacI) and the auxotrophic gene (glyA) using synthetic bi- the Bioprocess Engineering and Applied Biocatalisys Group, De- ology tools and methods, the strains developed in our lab can be partment of Chemical, Biological and Environmental Engineering used for efficient and stable expression of an industrially relevant of the Universitat Autònoma de Barcelona, Cerdanyola del Valles enzyme, namely, FucA, in cultivation conditions that resemble in- (Spain). dustrial production settings. FucA production (mg FucA g DCW–1 ) in the optimized third generation strain was c.a. 5%, 50%, and 10% higher comparing with the reference, first generation, and second Conflict of Interest generation strains, respectively. FucA activity (AU g–1 DCW) was in- All authors concur with the submission and agree with its publi- creased through the different stepwise improvements performed cation. The authors declare that they have no conflict of interest. along this work. The best performing engineered strain reached The authors confirm that this work is original and has not been more than 60% higher values in terms of activity compared to the published elsewhere, nor is it currently under consideration for first generation of M15glyA strain. publication elsewhere. The manuscript does not contain experi- Tuning of glyA transcriptional levels using the weak constitu- ments using animals or human studies. tive promoter J23110 resulted in the reduction of more than 50% of SHMT expression levels in the single plasmid expression system strains, which led to a reduction of the energy demand toward References SHMT production, thus redirecting nutrients/energy for recombi- Al-Allaf, F. A., Tolmachov, O. E., Zambetti, L. P., Tchetchelnitski, V., & nant FucA production. Mehmet, H. (2013). Remarkable stabilityu of an instability-prone To conclude, the optimized third generation strain presented lentiviral vector plasmid in Escherichia coli Stbl3. 3 Biotech, 3(1), 61– a reduction of the metabolic burden and an improvement of the 70. https://doi.org/10.1007/s13205- 012- 0070- 8. strain fitness and overall performance of the expression. The ex- Ammar, E. M., Wang, X., & Rao, C. V. (2018). Regulation of metabolism pression levels were similar to the commercial strain, but our in Escherichia coli during growth on mixtures of the non-glucose novel expression system presented advantageous features: (i) it sugars: arabinose, lactose, and xylose. Scientific Reports, 8(1), 609. is a single-plasmid tightly regulated expression system; (ii) it pre- https://doi.org/10.1038/s41598- 017- 18704- 0. sented significant reduced metabolic burden to the cells, resulted Brown, T. A. (1995) Gene cloning: an introduction (4th edition), Wiley- in no acetate production nor glucose accumulation and in healthy Blackwell. cells with intact and polarized cytoplasmatic membrane all along Calleja, D., Fernández-Castañé, A., Pasini, M., de Mas, C., & López- the fermentation; and (iii) no antibiotic is needed. Santín, J. (2014). Quantitative modeling of inducer transport in Remarkably, all this is achieved by means of (i) a system with fed-batch cultures of Escherichia coli. Biochemical Engineering Journal, a reduced number of genetic components and, importantly, (ii) 91, 210–219. https://doi.org/10.1016/j.bej.2014.08.017. 10 | Journal of Industrial Microbiology and Biotechnology, 2022, Vol. 49, No. 4 Carneiro, S., Ferreira, E. C., & Rocha, I. (2013). Metabolic responses Glenting, J. & Wessels, S. (2005). Ensuring safety of DNA vac- to recombinant bioprocesses in Escherichia coli. Journal of Biotech- cines. Microbial Cell Factories, 4(1), 26. https://doi.org/10.1186/ nology, 164(3), 396–408. https://doi.org/10.1016/j.jbiotec.2012. 1475- 2859- 4- 26. 08.026. Gronenborn, B. (1976). Overproduction of phage lambda repressor Carrió, M., Gonzalez-Montalbán, N., Vera, A., Villaverde, A., & Ven- under control of the lac promotor of Escherichia coli. Molecular tura, S. (2005). Amyloid-like properties of bacterial inclusion bod- and General Genetics MGG, 148(3), 243–250. https://doi.org/10.1007/ ies. Journal of Molecular Biology, 347(5), 1025–1037. https://doi.org/ BF00332898. 10.1016/j.jmb.2005.02.030. Hewitt, C. J., Nebe-von Caron, G., Nienow, A. W., & Mcfarlane, C. M. Castellanos-Mendoza, A., Castro-Acosta, R. M., Olvera, A., Zavala, (1999). The use of multi-parameter flow cytometry to compare G., Menodza-Vera, M., Garcia-Hernandez, E., Alagon, A., Trujillo- the physiological response of Escherichia coli w3110 to glucose lim- Roldan, M. A., & Valdez-Cruz, N. A. (2014). Influence of pH control itation during batch, fed-batch and continuous culture cultiva- in the formaiton of inclusion bodies during production of recom- tions. Journal of Biotechnology, 75(2-3), 251–264. https://doi.org/10. binant sphyingomyelinase-D in Escherichia coli. Microbial Cell Facto- 1016/s0168-1656(99)00168-6. ries, 13(1), 137. https://doi.org/10.1186/s12934- 014- 0137- 9. Hoffmann, F. & Rinas, U. (2004). Stress induced by recombinant pro- Downloaded from https://academic.oup.com/jimb/article/49/4/kuac018/6601392 by guest on 17 October 2022 Castiñeiras, T. S., Williams, S. G., Hitchcock, A. G., & Smith, D. C. (2018). tein production in Escherichia coli. Advances Biochemical Engineering E. coli strain engineering for the production of advanced bio- Biotecnology, 89, 73–92. https://doi.org/10.1007/b93994. pharmaceutical products. FEMS Microbiology Letters, 365(15), 162. Jeong, K. J. & Lee, S. Y. (2003). Enhanced production of recombinant https://doi.org/10.1093/femsle/fny162. proteins in Escherichia coli by filamentation suppression. Applied Clapés, P., Fessner, W. D., Sprenger, G. A., & Samland, A. K. (2010). Re- and Environmental Microbiology, 69(2), 1295–1298. https://doi.org/ cent progress in stereoselective synthesis with alsolases. Current 10.1128/AEM.69.2.1295-1298.2003. Opinion in Chemical Biology, 14(2), 154–167. https://doi.org/10.1016/ Kim, J., Darlington, A., Salvador, M., Utrilla, J., & Jimenez, J. I. (2020). j.cbpa.2009.11.029. Trade-offs between gene exession, growth and phenotypic diver- Costa, S., Almeida, A., Castro, A., & Domingues, L. (2014). Fusion sity in microbial populations. Current Opinion in Biotechnology, 62, tags for protein solubility, purification, and immunogenicity in Es- 29–37. https://doi.org/10.1016/j.copbio.2019.08.004. cherichia coli: the novel Fh8 system. Frontiers in Microbiology, 5(5), Kazan, D., Çamurdan, A., & Hortaçsu, A. (1995). The effect of glucose 63. https://doi.org/10.3389/fmicb.2014.00063. concentration on the growth rate and some intracellular compo- Durany, O., de Mas, C., & López-Santıń, J. (2005). Fed-batch produc- nents of a recombinant E. coli culture. Process Biochemistry, 30(3), tion of recombinant fuculose-1-phosphate aldolase in E. coli. Pro- 269–273. https://doi.org/10.1016/0032-9592(95)85008-2. cess Biochemistry, 40(2). 707–716. https://doi.org/10.1016/j.procbio. Koeller, K. M. & Wong, C. H. (2000). Caomplex carbohydrate synthe- 2004.01.058. sis tools for glycobiologists: enzyme-based approach and pro- Feizollahzadeh, S., Kouhpayeh, S., Rahimmansh, I., Khanahmad, H., grammable one-pot strategies. Glycobiology, 10(11), 1157–1169. Sabzehei, F., Ganjalikhani-hakemi, M., Andalib, A., Hejazi, Z., & https://doi.org/10.1093/glycob/10.11.1157. Razaei, A. (2017). The increase in preotein and palmsid yields Kwiatkowska, J., Matuszewska, E., Kuczyńska-Wiśnik, D., & of E. coli with optmized concentration of Ampicillin as selec- Laskowska, E. (2008). Aggregation of Escherichia coli pro- tion marker. Iranian Journal of Biotechnology, 15(2), 128–134. https:// teins during stationary phase depends on glucose and oxy- doi.org/10.15171/ijb.1467. gen availability. Research in Microbiology, 159(9-10), 651–657. Fernández-Castané, A., Caminal, G., & López-Santín, J. (2012). Direct https://doi.org/10.1016/j.resmic.2008.09.008. measurement of IPTG enable analysis of the induction of E. coli in Lartigue, M. F., Leflon-Guibout, W., Poirel L, L., Nordmann, P., & high cell density cultures. Microbial Cell Factories, 11( 1), 58. https:// Nicolas-Chanoine, M. H. (2002). HPromoters P3, Pa/Pb, P4, and P5 doi.org/10.1186/1475- 2859- 11- 58. upstream from bla(TEM) genes and their relationship to beta- Fernández-Castané, A., Li, H., Thomas, O. R. T., & Overton, T. W. (2017). lactam resistance. Antimicrobial Agents and Chemotherapy, 46(12), Flow cytometry as a rapid analytical tool to determine physio- 4035–7. https://doi.org/10.1128/AAC.46.12.4035-4037. logical responses to changin O2 and iron concentration by Mag- Lewis, G., Taylor, I. W, Nienow, A. W., & Hewitt, C. J. (2005). The appli- netospirillum gryphiswaldense strain MSR-1. Scientific Reports, 7(1), cation of multi-parameter flow cytometry to the study of recom- 13118. https://doi.org/10.1038/s41598- 017- 13414- z. binant Escherichia coli batch fermentation processes. Microbial Cell Fernández-Castané, A., Vine, C. E., Caminal, G., & López-Santín, J. Factories, 4(1), 311–322. https://doi.org/10.1186/1475- 2859- 4- 1. (2012). Evidencing the role of lactose permease in IPTG uptake Liu, M., Feng, X., Ding, Y., Zhao, G., Liu, H., & Xian, M. (2015). Metabolic by Escherichia coli in fed-batch high cell density cultures. Journal engineering of Escherichia coli to improve recombinant protein of Biotechnology, 157(3). 391–398. https://doi.org/10.1016/j.jbiotec. production. Applied Microbiology and Biotechnology, 99(24), 10367– 2011.12.007. 10377. https://doi.org/10.1007/s00253- 015- 6955- 9. Garcia-Junceda, E., Gwo-Jenn, S., Takeshi, S., & Wong, C. H. (1995). Mairhofer, J., Scharl, T., Marisch, K., Cserjan-Puschmann, M., & Stried- A new strategy for the cloning, overexpression and one step ner, G. (2013). Comparative transcription profiling and in-depth purification of Tthree DHAP-dependent aldolases: rhamnulose- characterization of plasmid-based and plasmid-free Escherichia 1-phosphate aldolase, fuculose-1-phosphate aldolase and coli expression systems under production conditions. Applied and tagatose-1,6-diphosphate aldolase. Bioorganic & Medicinal Chem- Environmental Microbiology, 79(12), 3802–3812. https://doi.org/10. istry, 3(7), 945–953. https://doi.org/10.1016/0968-0896(95)00077-t. 1128/AEM.00365-13. Germán, L. R., Morales, E. S., & Ceccarelli, E. A. (2019). New tools for Marschall, L., Segmeister, P., & Herwig, C. (2017). Tunable recom- recombinant protein production in Escherichia coli: a 5-year up- biant protein expression in E. coli promoter systems and genetic date. Protein Science, 28(8). 1412–1422. https://doi.org/10.1002/pro. constraints. Applied Microbiology and Biotechnology, 101(2), 501–512. 3668. https://doi.org/10.1007/s00253- 016- 8045- z. Glick, B. R. (1995). Metabolic load and heterologous gene expres- Mignon, C., Sodoyer, R., & Werle, B. (2015). Antibiotic-free selection in sion. Biotechnology Advances, 13(2), 247–261. https://doi.org/10. biotherapeutics: now and forever. Pathogens, 4(2), 157–181. https:// 1016/0734-9750(95)00004-A. doi.org/10.3390/pathogens4020157. Pasini et al. | 11 Mukhopadhyay, A. (1997). Inclusion bodies and purification of cherichia coli W3110 growth. Journal of Microbiology and Biotechnology, proteins in biologically active forms. Advances in Biochemi- 19(10), 1127–1134. https://doi.org/10.4014/jmb.0902.0097. cal Engineering/Biotechnology, 56. 61–109. https://doi.org/10.1007/ Vallejo, L. F. & Rinas, U. (2004). Strategies for the recovery of ac- BFb0103030. tive proteins through refolding of bacterial inclusio body pro- Neubauer, P., Lin, H. Y., & Mathiszik, B. (2003). Metabolic load of re- teins. Microbial Cell Factories, 3(1), 11. https://doi.org/10.1186/ combinant protein production: inhibition of cellular capacities 1475- 2859- 3- 11. for glucose uptake and respiration after induction of a heterolo- Vidal, L., Durany, O., Suau, T., Ferrer, P., & Caminal, G. (2003). gous gene in Escherichia coli. Biotechnology and Bioengineering, 83(1), High-level production of recombinant His-tagged rhamnulose 1- 53–64. https://doi.org/10.1002/bit.10645. phosphate aldolase in Escherichia coli. Journal of Chemical Technology Pasini, M., Fernández-Castané, A., Jaramillo, A., De Mas, C., Cam- & Biotechnology, 78(11), 1171–1179. inal, G., & Ferrer, P. (2016). Using promoter libraries to reduce Vidal, L., Ferrer, P., Alvaro, G., Benaiges, M. D., & Caminal, G. (2005). In- metabolic burden due to plasmid-encoded proteins in recombi- fluence of induction and operation mode on recombinant rham- nant Escherichia coli. New Biotechnology, 33(1), 78–90. https://doi. nulose 1-phosphate aldolase production by Escherichia coli using org/10.1016/j.nbt.2015.08.003. the T5 promoter. Journal of Biotechnology, 118(1), 75–87. Downloaded from https://academic.oup.com/jimb/article/49/4/kuac018/6601392 by guest on 17 October 2022 Phue, J.-N. & Shiloach, J. (2005). Impact of dissolved oxygen concen- Vidal, L., Pinsach, J., Striedner, G., & Ferrer, P. (2008). Development of tration on aceate accumulation and physiology of E. coli BL21, an antibiotic-free plasmid selection system based on glycine aux- evaluating transciption levels of key genes at different dissolved otrophy for recombinant protein overproduction in Escherichia coli. oxygen conditions. Metabolic Engineering, 7(5-6), 353–363. https: Journal of Biotechnology, 134(1-2), 127–136. https://doi.org/10.1016/ //doi.org/10.1016/j.ymben.2005.06.003. j.jbiotec.2008.01.011. Pierce, J. & Gutteridge, S. (1985). Large-scale preparation of ribu- Villarejo, M. R. & Zabin, I. (1974). β -galactosidase from termination losebisphosphate carboxylase from a recombinant system in Es- and deletion mutant strains. Journal of Bacteriology, 120(1). 466– cherichia coli characterized by extreme plasmid instability. Appl 474. https://doi.org/10.1128/jb.120.1.466-474.1974. Env Microbiol, 49(5). 1094–1100. https://doi.org/10.1128/aem.49.5. Waegeman, H. & Soetaert, W. (2011). Increasing recombinant protein 1094-1100. production in Escherichia coli through metabolic and genetic en- Pinsach, J., de Mas, C., & López-Santín, J. (2008). Induction strate- gineering. Journal of Industrial Microbiology & Biotechnology, 38(12), gies in fed-batch cultures for recombinant protein production in 1891–1910. https://doi.org/10.1007/s10295- 011- 1034- 4. Escherichia coli: application to rhamnulose 1-phosphate aldolase. Weber, J., Hoffmann, F., & Rinas, U. (2002). Metabolic adaptation Biochemical Engineering Journal, 41(2), 181–187. https://doi.org/10. of Escherichia coli during temperature-induced recombinant pro- 1016/j.bej.2008.04.013. tein production: 2. Redirection of metabolic fluxes. Biotechnol- Plamann, M. D. & Stauffer, G. V. (1983). Characterisation of the ogy and Bioengineering, 80(3), 320–330. https://doi.org/10.1002/bit. Escherichia coli gene for serine hydroxymethyltransferase. Gene, 10380. 22(1), 9–18. https://doi.org/10.1016/0378-1119(83)90059-8. Weber, J., Li., Z., & Rinas, U. (2021). Recombinant protein production Rahmen, N., Fulton, A., Ihling, N., Magni, M., Jaeger, K., & Büchs, provoked accumulation of ATP, fructose-1,6-bisphosphate and J. (2015). Exchange of single amino acids at different posi- pyruvate in E. coli K12 strain TG1. Microbial Cell Factories, 20(1), 169. tions of a recombinant protein affects metabolic burden in Es- https://doi.org/10.1186/s12934- 021- 01661- 9. cherichia coli. Microbial Cell Factories, 14(1). https://doi.org/10.1186/ Wyre, C. & Overton, T. W. (2014a). Flow cytometric analysis of E. s12934- 015- 0191- y. coli on agar plates: implications for recombinant protein pro- Rosano, G. L. & Ceccarelli, E. A. (2014). Recombinant protein expres- duction. Biotechnology Letters, 36(7), 1485–1494. https://doi.org/10. sion in Escherichia coli: advances and challenges. Frontiers in Micro- 1007/s10529- 014- 1511- 8. biology, 5. 172. https://doi.org/10.3389/fmicb.2014.00172. Wyre, C. & Overton, T. W. (2014b). Use of a stress-minimisation Saubi, N., Gea-Mallorquí, E., Ferrer, P., Hurtado, C., Sánchez-Úbeda, paradigm in high cell density fed-batch Escherichia coli fermen- S., Eto, Y., Gatell, J. M., Hanke, T., & Joseph, J. (2014). Engineering tations to optimise recombinant protein production. Journal of new mycobacterial vaccine design for HIV-TB pediatric vaccine Industrial Microbiology and Biotechnology, 41(9), 1391–1404. https:// vectored by lysine auxotroph of BCG. Molecular Therapy - Methods & doi.org/10.1007/s10295- 014- 1489- 1. Clinical Development, 1. 14017. https://doi.org/10.1038/mtm.2014. Xu, J. & Matthews, K. S. (2009). Flexibility in the inducer binding region 17. is crucial for allostery in the Escherichia coli lactose repressor. Bio- Schramm, F. D, Schroeder, K., & Jonas, K. (2020). Protein aggregation in chemistry, 48(22), 4988–4998. https://doi.org/10.1021/bi9002343. bacteria. FEMS Microbiology Reviews, 44(1). 54–72. https://doi.org/ Zamenhof, P. J. & Villarejo, M. (1972). Construction and properties 10.1093/femsre/fuz026. of Escherichia coli strains exhibiting α-complementation of β- Sooan, S., Chang, D., & Pan, J. G. (2009). Acetate consuption activity galactosidase fragments in vivo. Journal of Bacteriology, 110(1), 171. directly determines the level of acetate accumulation during Es- https://doi.org/10.1128/jb.110.1.171-178.1972.