Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Oct 24.
Published in final edited form as: Biotechnol Bioeng. 2020 Jun 10;117(9):2715–2727. doi: 10.1002/bit.27440

Scalable, two-stage, autoinduction of recombinant protein expression in E. coli utilizing phosphate depletion

Romel Menacho-Melgar 1, Zhixia Ye 1, Eirik A Moreb 1, Tian Yang 1, John P Efromson 1, John S Decker 1, Ruixin Wang 1, Michael D Lynch 1
PMCID: PMC9589519  NIHMSID: NIHMS1614236  PMID: 32441815

Abstract

We report the scalable production of recombinant proteins in Escherichia coli, reliant on tightly controlled autoinduction, triggered by phosphate depletion in the stationary phase. The method, reliant on engineered strains and plasmids, enables improved protein expression across scales. Expression levels using this approach have reached as high as 55% of the total cellular protein. The initial use of the method in instrumented fed-batch fermentations enables cell densities of ~30 gCDW/L and protein titers up to 8.1 ± 0.7 g/L (~270 mg/gCDW). The process has also been adapted to an optimized autoinduction media, enabling routine batch production at culture volumes of 20 μl (384-well plates), 100 μl (96-well plates), 20 ml, and 100 ml. In batch cultures, cell densities routinely reach ~5–7 gCDW/L, offering protein titers above 2 g/L. The methodology has been validated with a set of diverse heterologous proteins and is of general use for the facile optimization of routine protein expression from high throughput screens to fed-batch fermentation.

Keywords: autoinduction, phosphate depletion, protein expression, stationary phase

1 |. INTRODUCTION

Heterologous protein expression is a standard workflow common in numerous fields of biology and Escherichia coli is the workhorse microbe for routine protein production in academia and industry. E. coli-based processes are used for the production of over 30% of protein-based drugs that are in the market today, (Sanchez-Garcia et al., 2016) and pET-based expression in E. coli strain BL21(DE3) and its derivatives are a mainstay of heterologous expression in many labs.

Standard protocols rely on easily prepared media (Luria broth [LB] and or terrific broth) but require culture monitoring to optimize induction in exponential phase. (Glazyrina et al., 2010; Neidhardt, Ingraham, & Schaechter, 1990) Autoinduction protocols removing the need for manual additions have been developed, most notably by Studier (2005) and (2014), and require the use of multiple carbon substrates, such as glucose and lactose. After glucose depletion the consumption of lactose induces heterologous expression. Significant recent work has been done in developing new protocols enabling autoinduction systems focused on using novel auto inducing promoters that respond to a variety of signals from cell density to oxygen limitation (Anilionyte, Liang, Ma, Yang, & Zhou, 2018; Baez, Majdalani, & Shiloach, 2014; Ben et al., 2016; Briand et al., 2016; Nocadello & Swennen, 2012). Despite simplifying expression protocols, many of these approaches still result in relatively low biomass and protein levels and have not been validated in multiple culture systems including instrumented bioreactors. Scale up of protein expression to higher cell density fermentations remains a nontrivial task. The use of BL21 and its derivatives can be further complicated by heterogenous induction, resulting from lactose-based inducers (Labhsetwar, Cole, Roberts, Price, & Luthey-Schulten, 2013; Novick & Weiner, 1957), as well as the accumulation of acetic acid in fermentations with excess carbon source, which can have toxic effects on both cell growth and protein expression (Eiteman & Altman, 2006; Shiloach & Fass, 2005).

There remains a need for autoinducible protein expression methods with tightly controlled expression, minimal overflow metabolism, and a high level of protein expression. Ideally new methods will be adaptable to numerous workflows and culture volumes, from high throughput screening approaches in microtiter plates and be readily scalable to higher cell densities in larger instrumented bioreactors, in commercially relevant media.

We report the development of a facile protocol for the routine high level expression of proteins. The method relies on a promoter that is induced by phosphate depletion, where protein expression is induced at the entry into stationary phase. While the expression of heterologous proteins during stationary phase may seem counter-intuitive and at odds with maximal production, stationary phase cells can maintain significant metabolic activity and produce high levels of protein (Burg et al., 2016; Chubukov & Sauer, 2014; Lynch, 2014, 2016). Specifically, phosphate depletion has been used routinely for heterologous protein expression (Lübke, Boidol, & Petri, 1995; Song, Jiang, Wang, & Zhang, 2017). In addition, it has been shown that phosphate depletion can be used to amplify the expression of heterologous proteins using the pET-based T7 promoters in E. coli (Huber, Roth, Rahmen, & Buchs, 2011). Phosphate-dependent promoters are used in an engineered strain of E. coli with minimal acetate production, and near optimal growth rates and yields, offering tightly controlled expression. These strains and plasmids can be used in minimal media in instrumented bioreactors as well as with an optimized autoinduction broth (AB) enabling high level batch expression, in cultures as small as 20 μl in 384-well plates, to 100 ml in larger shake flasks.

2 |. MATERIALS AND METHODS

2.1 |. Reagents and media

Unless otherwise stated, all materials and reagents were purchased from Sigma-Aldrich (St. Louis, MO). LB, Lennox formulation was used for routine strain and plasmid propagation and construction and is referred to as LB below. All media formulations including stock solutions and working antibiotic concentrations are described in Supporting Information Materials Section 4.

2.2 |. Strains and strain construction

E. coli strains BL21(DE3) (Cat #C2527) and BL21(DE3) pLysS (Cat #C3010) were obtained from New England Biolabs (NEB), Ipswich, MA. Strain BW25113 was obtained from the Yale E. coli Genetic Stock Center (https://cgsc.biology.yale.edu/; Baba et al., 2006; Datsenko & Wanner, 2000). Strain BWapldf was a gift from George Chen (Tsinghua University; Jian et al., 2010). Chromosomal modifications were made using standard recombineering methodologies (Sharan, Lynn, Kuznetsov, & Court, 2009) through scarless tet-sacB selection and counterselection, strictly following the protocols of Li, Thomason, Sawitzke, Costantino, and Court (2013). The recombineering plasmid pSIM5 and the tet-sacB selection/counterselection marker cassette were kind gifts from Donald Court (NCI, https://redrecombineering.ncifcrf.gov/court-lab.html). The ompT protease gene was deleted using standard recombineering methods by selection for an apramycin selectable marker obtained from the pMDIA plasmid (Yang et al., 2014). pMDIA was a gift from Sheng Yang (Addgene plasmid #51655; http://n2t.net/addgene:51655; refer to the Supporting Information Materials Section 3, for additional details of strain construction). Chromosomal modifications were confirmed by polymerase chain reaction (PCR) amplification and sequencing (Genewiz, NC) using paired oligonucleotides, either flanking the entire region.

2.3 |. Plasmids

pETM6 and pETM6-mCherry were a gift from Mattheos Koffas (Addgene plasmids #49795 and #66534). pLysS was obtained from NEB. Plasmids made in this study were constructed using G-blocks™ and/or PCR products and assembled using NEBuilder® HiFi DNA Assembly Master Mix following manufacturer’s protocol (NEB). PCRs were performed with Q5 DNA Polymerase (NEB). pSMART-HC-Kan (Lucigen, WI), pTWIST-Chlor-Medium Copy, pTWIST-Kan-High Copy (Twist Biosciences, San Francisco, CA), and pCDF (derived from pCDF-1b; EMD Millipore, Burlington, MA) were used as a backbone vectors in these studies. Sequences of all oligos and synthetic DNA are given in Supporting Information Materials Section 3, Tables S3.1 and S3.2. All plasmid sequences were confirmed by DNA sequencing (Genewiz). Sequences and maps are available with Addgene (refer to Table 1 for Addgene numbers; refer to Supporting Information Materials Section).

TABLE 1.

Plasmids and strain used in this study

Plasmid Insert Promoter ori Res Addgene Source
pSMART-HC-Kan None None colE1 Kan NA Lucigen
pLysS T7 lysozyme NA p15a Cm NA NEB
pHCKan-yibDp-GFPuv GFPuv yibDp colE1 Kan 127078 This study
pHCKan-yibDp-mCherry 6xhis-mCherry yibDp colE1 Kan 127058 This study
pETM6 None T7 (pET) colE1 Amp 49795 Jones et al. (2015)
pETM6-mCherry mCherry T7 (pET) colE1 Amp 66534
pCDF None None cloDF13 Sm 89596 This study
pTCmc-yibDp-SBS-mCherry SBS-mCherry yibDp p15a Cm 134598 This study
pTKhc-yibDp-GFP-β20-cp6 GFP-β20-cp6 yibD colE1 Kan 127060 This study
pTKhc-yibDp-GFP-cp6 GFP-cp6 yibD colE1 Kan 134938 This study
pCDF-yibDp-matB matB yibDp cloDF13 Sm 134597 This study
pSMART-Ala1 AlaDh(D196A/L197R) yibDp colE1 Kan 65814 This study
pCDF-yibDp-mdlC-his mdlC-6xhis yibDp cloDF13 Sm 134590 This study
pHCKan-yibDp-GST GST-6xHis yibD colE1 Kan 134592 This study
pHCKan-yibDp-Nef Nef yibD colE1 Kan 134593 This study
pTKhc-yibDp-cimA3.7 cimA3.7 yibDp colE1 Kan 134595 This study
pHCKan-yibDp-CBD-hGLY hGLYAT2 yibD colE1 Kan 134596 This study
Strains used in this study
Strain Genotype Source
BL21(DE3) F– ompT gal dcm Lon hsdSB(rB–mB–) λ(DE3 [lacI lacUV5-T7p07 ind1 sam7 nin5]) [malB+]K-12(λS) NEB
BWapldf F-, λ-, Δ(araD-araB)567, lacZ4787(del)(::rrnB-3), rph-1, Δ(rhaD-rhaB)568, hsdR51, ΔackA-pta, ΔpoxB,
ΔpflB, ΔldhA, ΔadhE 		Jian et al. (2010)
DLF_R002 BWapldf, ΔiclR, ΔarcA This study
DLF_R003 DLF_R002, ΔompT::apmR This study
Open in a new tab

Abbreviations: Amp, ampicillin; Cm, chloramphenicol; Kan, kanamycin; NEB, New England BioLabs; Res, resistance marker; Sm, spectinomycin.

2.4 |. BioLector™ experiments

Growth and fluorescence measurements were obtained in a BioLector (m2p-labs, Baesweiler, Germany) using a high mass transfer Flower-Plate (Cat #MTP-48-B; m2p-labs, BioLector settings were as follows: RFP gain = 40, GFP gain = 20, biomass gain = 20, shaking speed 1,300 rpm, temperature 37°C, humidity 85%. Single colonies of each strain were inoculated into 5 ml LB with appropriate antibiotics and cultured at 37°C, 150 rpm overnight. Overnight culture optical density (OD) was measured and normalized to OD600 nm = 25. Subsequently, 8 μl of normalized overnight culture was inoculated into 792 μl of the appropriate medium with appropriate antibiotics and transferred into wells of the FlowerPlate. Every strain was analyzed in at least triplicate.

2.5 |. Microtiter plate-based growth and expression

Ninety-six and three hundred eighty-four-well plates were obtained from Genesee Scientific, San Diego, CA (Cat #25-104) and VWR, Suwanee, GA (Cat #10814-224). Three microliters of strain glycerol stocks were used to inoculate 150 μl LB overnight cultures with appropriate antibiotics. Plates were covered with sandwich covers (Model #CR1596, 96-well plates and Model #CR1384, 384-well plates) obtained from EnzyScreen, Haarlem, The Netherlands). These covers ensure minimal evaporative loss during incubation. Microtiter plates were cultured at 37°C, 300 rpm for 16 hr, shaker orbit is 50 mm. This combination of orbit and minimal shaking speed is required to obtain needed mass transfer coefficient and enable adequate culture oxygenation. After 16 hr of growth, a 1% volume of the overnight culture was inoculated into the desired volume of autoinduction media plus the appropriate antibiotics. Plates were again covered with sandwich covers and grown at 37°C, 300 rpm for 24 hr at which point samples were harvested for analysis.

2.6 |. Autoinduction media development

The autoinduction media was developed using design of experiments (DoE) definitive screening designs and JMP software (SAS, Cary, NC). Media were prepared as detailed in Supporting Information Materials Section 4; 2X trace metal mix, 2.5 g/L yeast extract, and 2.5 g/L casamino acid were used as the starting center point, 4X and 1/4X of the center point values were used as the upper and lower concentration ranges. Definitive screening design was performed in five iterations. Center point, upper and lower concentration ranges for future iterations were determined based on DoE results from the previous iteration. A total of 148.5 μl of media was distributed to triplicate 96-well plates and each well was inoculated with 1.5 μl of overnight LB culture. The plates were covered with EnzyScreen covers and shaken at 300 rpm at 37°C. After 24 hr, OD and fluorescence were measured.

2.7 |. Shake flask growth and expression

Glycerol stocks were used to inoculate overnight cultures in 5 ml of LB media, with appropriate antibiotics. After 16 hr of growth, a 1% volume of overnight culture was inoculated into autoinduction media plus the appropriate antibiotics. Flasks cultures were grown at 37°C, 150 rpm in baffled 250 ml Erlenmeyer flasks for 24 hr at which point samples were harvested for analysis.

2.8 |. Fermentation seeds

Single colony from transformation plate was inoculated into 5 ml LB with appropriate antibiotics and cultured at 37°C, 150 rpm for 16 hr. Two hundred microlitres of the LB culture was inoculated into 20 ml SM10 + seed media with appropriate antibiotics in 250 ml shaker flasks. The culture was incubated at 37°C with a shaking speed of 150 rpm for 16 hr, at which time OD600 nm is usually between 6 and 10. The culture was harvested by centrifugation at 4,000 rpm for 15 min, the supernatant was discarded and the cell culture was normalized to OD600 nm = 10 using FGM10 media. Seed vials were prepared by adding 1.5 ml of 50% glycerol to 6.5 ml of normalized OD600 nm = 10 culture in cryovials, and stored at −60°C.

2.9 |. One liter fermentations

An Infors-HT Multifors (Laurel, MD) parallel bioreactor system, as described in Supporting Information Materials Section 5, was used to perform 1 L fermentations. Tanks were filled with 800 ml of either FGM10 or FGM30 medium, which have enough phosphate to target a final E. coli biomass concentration of ~10 or 30 gCDW/L, respectively. Antibiotics were added as appropriate. Phosphate, glucose, thiamine, and antibiotics were added after cooling the tank vessel containing the rest of the media components. Frozen seed vials containing 8 ml of seed culture were used to inoculate the tanks. Bioreactors were controlled at 37°C and pH 6.8 using 14.5 M ammonium hydroxide and 1 M hydrochloric acid as titrants. Oxygen control and the glucose feeding strategies are detailed in Supporting Information Materials Section 5.

2.10 |. Organic acid quantification

Two orthogonal methods were used to quantify organic acids including lactate, acetate, succinate, fumarate, pyruvate, malate, and others. The first method was a reverse phase ultra performance liquid chromatography method. Chromatographic separation was performed using a Restek Ultra AQ C18 column (150 mm × 2.1 i.d., 3 μm; Cat #9178362; Restek Corporation, Bellefonte, PA) at 30°C. Twenty millimolar phosphoric acid was used as the eluent. The isocratic elution rate was at 0.8 ml/min, run time was 1.25 min. Sample injection volume was 10 μl. Absorbance was monitored at 210 nm. The second method relied on ion exchange chromatography and refractive index detection. A Phenomenex Rezex™ ROA-Organic Acid H+ (8%) (30 × 4.6 mm; Cat #00A-0138-E0; Phenomenex, Torrance, CA) was used for a 30 min isocratic separation using a mobile phase of 5 mM H2SO4, at a flow rate of 0.5 ml/min. Again sample injections were 10 μl. Organic acid elution times were as follows: pyruvate 13.3 min, citrate 10.9 min, lactate 17.5 min, and acetate 20.3 min.

2.11 |. Glucose quantification

Two methods were used to quantify glucose. The first was identical to the organic acid method, utilizing the Resex column, discussed above, wherein glucose eluted at 12.5 min. The second method also relied on ion exchange and refractive index detection. Chromatographic separation was performed using a Bio-Rad Fast Acid Analysis HPLC Column (100 × 7.8 mm, 9 μm particle size; Cat #1250100; Bio-Rad Laboratories Inc., Hercules, CA) at 65°C. Five millimolar sulfuric acid was used as the eluent at a flow rate of 0.48 ml/min. Sample injection volume was 10 μl.

2.12 |. Determination of strain dry weight

Culture samples (5 ml, n = 3) were taken and washed twice with deionized water via centrifugation and resuspension. After wash steps the OD of the samples were determined at 600 nm. Subsequently, samples were filtered over preweighed nitrocellulose filters (pore size, 0.45 μm). Filters were washed extensively with demineralized water and dried in a microwave oven until weights were stable to determine correlation of OD600 nm and gram dry cell weight (gDCW), which was 0.35.

2.13 |. Phosphate quantification

Phosphate concentrations were determined using the BioMOL Green colorimetric assay from Enzo Life Sciences (Farmingdale, NY) according to the manufacturer’s instructions.

2.14 |. Biomass and fluorescence measurements

Optical densities and fluorescent were measured using a Tecan Infinite 200 plate reader, using 200 μl in black-walled 96-well plates (Reference Number: 655087; Greiner Bio-One). OD was read at 600 nm (Part Number: 3019445; filter from Omega Optical). GFP fluorescence was measured using excitation at 412 nm (Part Number: 3024970; Omega Optical) and emission at 530 nm (Part Number: 3032166; Omega Optical) using a gain of 60. Readings were adjusted for blanks, path length, and dilutions as appropriate.

2.15 |. Sodium dodecyl sulfate–polyacrylamide gel electrophoresis and protein quantification

The OD600 nm of cultures was measured and then cells were harvested by centrifugation at 4,000 rpm for 15 min and resuspended in 50 μl of phosphate-buffered saline with protease inhibitors (Product Number A32965; Thermo Fisher Scientific, MA) and 5 mM ethylenediaminetetraacetic acid. Twenty-five microliters of the resuspended cells were mixed with 25 μl of 2X Laemmli sample buffer (Bio-Rad Laboratories Inc.), boiled for 5 min at 95°C and then centrifuged at 14,000 rpm for 10 min. For each sample, 20 μg of total protein was loaded into a 4–15% gradient Mini-Protean TGX precast protein gel (Bio-Rad Laboratories Inc.) and run at 140 V. The gels were stained using Coomassie Brilliant Blue R-250 and imaged using a UVP PhotoDoc-It™ Imaging System (Analytik Jena, CA.) Expression levels were quantified using ImageJ (NIH, MD). To correlate GFPuv fluorescence with grams of GFPuv, samples were taken wherein both (a) fluorescence was measured as described above and (b) expression level was calculated as described above. Total cellular protein was estimated at 500 mg/gDCW or 50% of DCW (Long & Antoniewicz, 2014). In these comparisons, 3.24e9 relative fluorescent units corresponded to 1 g of GFPuv. This correlation was also used to calculate GFPuv titers across all experiments.

2.16 |. Solubility evaluation and scoring

To quantify the soluble fractions for the group of test proteins, samples were sonicated using 10/30 s on/off cycles for 20 min. After clearing the lysates, the insoluble fraction was resuspended to load in a protein gel. The expression level of the heterologous proteins in the soluble and insoluble fractions were quantified using Image J, except for mCherry and SBS-mCherry where fluorescence of mCherry was used as described above, but with excitation/emission at 585/612 nm (with a gain of 60). Solubility scores were calculated according to Berrow et al. (2006), where the level of soluble expression was evaluated using a standardized regime as follows: 0, no detectable expression or degraded protein; 1, expression predicted to give <0.5 mg/L at scale: 2 (0.5–5.0 mg/L), 3 (>5 mg/L), wherein scale is at least a 1 L culture.

2.17 |. Cytometry

Strains were grown in 5 ml of LB at 37°C, 150 rpm for 16 hr. A 1% volume of of either BL21(DE3) pLys or DLF_R002 overnight culture were used to inoculate 20 ml of LB or autoinduction broth (AB) media in 250 ml baffled Erlenmeyer flasks incubated at 37°C, 150 rpm. BL21(DE3) cultures were induced at OD600 nm ~0.3 with 1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG). Samples for BL21(DE3) were collected 20 hr after induction. DLF_R002 samples were collected 24 hr after inoculation. Samples were diluted 1,000-fold with sterile distilled water before analysis in a Thermo Attune NXT flow cytometer (Thermo Fisher Scientific), at a 12.5 μl/min flow rate. Fluorescence measurements were taken from the 620/15 band pass filter with excitation at 561 nm. DLF_R002 bearing an empty vector control (pSMART-HC-Kan) was used to determine the appropriate gating to exclude small particles from being counted as events. A forward scatter height of 10,000 and a side scatter of 2,500 were used for gating for all samples. Data were analyzed using FlowJo v10.6.1 (BD, NJ).

3 |. RESULTS

3.1 |. Initial characterization of phosphate induction with the yibDp gene promoter

To move from IPTG-based induction to autoinduction via phosphate depletion, we leveraged a previously reported phoB-regulated promoter, a modified promoter of the E. coli yibD (waaH) gene, referred to herein as yibDp, and constructed a plasmid enabling the induction of mCherry upon phosphate depletion (pHCKan-yibDp-mCherry; Table 1, refer to Supporting Information Materials Section 2 Table S2.1 for the promoter sequence (Lynch, Gill, & Lipscomb, 2018; Lynch et al., 2015). The modified yibD promoter is a 5′ truncation of the native yibD promoter wherein PhoB activation sites are retained and the NsrR repressor site is deleted. A recent additional study from our group confirmed the performance of this promoter by comparing expression levels and the robustness of this modified yibD promoter with 16 additional PhoB-regulated E. coli promoters, including the commonly used phoA promoter (Moreb et al., 2020). We initially evaluated the expression of this construct in BL21(DE3), BL21(DE3) with pLysS and a well characterized E. coli K12 derivative: BW25113 (Grenier, Matteau, Baby, & Rodrigue, 2014). The accessory plasmid (pLysS) expressing T7 lysozyme, is routinely used to reduce leaky induction in pET-based systems (Studier, 1991). Unexpectedly, significant basal expression was observed in BL21(DE3; Figure S1). In contrast, no significant basal expression was observed in BW25113. More work is needed to understand the mechanisms underlying the basal expression in BL21(DE3).

3.2 |. Host strain engineering

With BL21(DE3) demonstrating baseline heterogeneous leaky expression with the yibDp promoter, and in light of other routine issues encountered in using BL21 and its derivatives, such as accumulation of acetic acid, we turned to engineering a BW25113 derivative for optimal growth and minimal byproduct formation. We began with a previously reported derivative, strain BWapldf, with deletions in genes leading to common mixed acid fermentation products, such as lactic and acetic acid (Jian et al., 2010). BWapldf has deletions in the following genes: ackA-pta, pflB, adhE, ldhA, and poxB reducing the rates of production of acetate, formate, lactate, and ethanol from overflow metabolites. Deletions of the two global regulators iclR and arcA were next incorporated into this strain. These mutations have been shown to improve biomass yield and reduce overflow metabolism in K12 derivatives. Together these mutations increase flux through the citric acid cycle and glyoxylate bypass and reduce overflow metabolism by increasing the rate of oxidation of excess carbon to carbon dioxide and increasing ATP supply (Waegeman et al., 2011; Waegeman, Maertens, Beauprez, De Mey, & Soetaert, 2012).

These strains, as well as a BL21(DE3) pLysS control were initially evaluated in controlled fed-batch fermentations, using a defined minimal media (FGM10 media, refer to Section 2) wherein phosphate concentrations limit biomass levels. Growth rates, biomass and by-products, including acetic acid, were measured. Results are given in Figure 1 as well as in Supporting Information Table S1 and Figure S2. In these studies organic acid byproducts, other than acetic acid were not observed. As expected, BL21(DE3) produced acetic acid during growth (Figure 1e; Shiloach, Kaufman, Guillard, & Fass, 1996) Interestingly, strain BWapldf, despite having numerous deletions had a significantly decreased biomass yield and increased acetic acid production compared to BW25113 (Figure 1f,g). The deletion of the two global regulators, arcA and iclR, (strain DLF_R002) recovered biomass yield and virtually eliminated acetic acid production in this host (Figure 1h; refer to Supporting Information Materials and Figure S2 for maximal growth rates. It should be noted that all of these fed-batch fermentations were run with excess residual glucose (refer to Figure 1a–d). Glucose limitation can be expected to reduce overflow metabolites including acetic acid but often requires optimization of feed rates (Eiteman & Altman, 2006; Shiloach et al., 1996) Specifically, while acetate production is lower in BL21 compared to many E. coli strains, glucose limitation is still needed for effective reduction in acetate acid production (Shiloach et al., 1996). To our knowledge DLF_R002 is somewhat unique in that no measurable overflow metabolites were observed even with excess residual sugar, simplifying process development.

FIGURE 1.

FIGURE 1

Open in a new tab

Growth and byproduct formation of Escherichia coli strains in minimal media fermentations. Biomass levels and residual glucose concentration (blue) as a function of time for (a) BL21(DE3)pLys, (b) BW25113, (c) BWapldf, and (d) DLF_R002, respectively. (e) Distribution of glucose utilized during growth in minimal medium fermentations for BW25113 (white), BL21(DE3) pLys (light gray), BWapldf (dark gray), and DLF_R002 (black). Results are averages of duplicate fermentations. CO2 was explicitly measured via off-gas analysis for strain BW25113, BWapldf, and DLF_R002. In the case of BL21(DE3) pLys, CO2 is included in unknown products required to account for glucose consumption. OD, optical density

Using strain DLF_R002 we next turned to evaluate protein expression in bioreactors using FGM10 media. As mentioned biomass levels supported by FGM10 media are limited by phosphate, and phosphate depletion occurs when biomass levels reach an OD of ~30–35 or ~10 gCDW/L. In this case we constructed an additional plasmid with GFPuv driven by the yibDp promoter (pHCKan-yibDp-GFPuv; Table 1; Crameri, Whitehorn, Tate, & Stemmer, 1996; results are given in Figure 2a). Biomass levels reached ~10 gCDW/L producing final GFPuv titers of ~2.7 g/L or 270 mg/gCDW. With the success in low cell density we turned to develop a process with higher cell density utilizing FGM30 media with enough phosphate to support three times the biomass (~30 gCDW/L). Results are given in Figure 2b. In addition we evaluated expression in strain DLF_R003 a derivative of DLF_R002 with a deletion in the outer membrane ompT protease (Figure 2c), which has proteolytic activity even under denaturing conditions, creating issues with purification of recombinant proteins (Gill, DeLisa, Shiloach, Holoman, & Bentley, 2000; White, Chen, Kenyon, & Babbitt, 1995). Specific expression levels were maintained at higher biomass levels resulting in a threefold improvement in GFPuv titer reaching ~8.1 g/L.

FIGURE 2.

FIGURE 2

Open in a new tab

Autoinduction of GPFuv expression in 1 L bioreactors with (a) FGM10 minimal media and host strain DLF_R002 (triplicates, solid line, open symbols) and DLF_R003 (DLF_R002, ΔompT; single run, dotted line, filled squares) bearing plasmid pHCKan-yibDp-GFPuv. (b) FGM30 minimal media and host strain DLF_R003 (triplicates) bearing plasmid pHCKan-yibDp-GFPuv. Optical density (black lines) and GFPuv were measured over time. Shaded area is standard error of triplicate growth profiles. X’s, triangles, circles, and squares are normalized GFPuv fluorescence units, each symbol corresponding to a single fermentation. Green line is the best fit of the expression profiles. OD, optical density

3.3 |. Development of phosphate-limited media for autoinduction

We next turned to the optimization of media formulations for more routine autoinduction via phosphate depletion. Importantly, the fermentations discussed above (Figure 1) were performed with defined minimal media, which while lower in cost in larger scale production, can lead to significant lags when cells transition from a richer cloning and propagation media such as LB. To overcome this, seed cultures are often used to adapt the cells to a more minimal media (as they were in this case, refer to Section 2) before inoculation of bioreactors. For routine lab scale protein expression, media adaptation is not desirable, and rather protocols enabling direct inoculation of production flasks from overnight LB cultures is preferred. As a result, we developed batch AB with more complex nutrient sources including yeast extract and casamino acids. Media formulations were developed using standard DoE methodology and evaluated in 96-well plates. These experiments were performed using strain DLF_R002 bearing plasmid pHCKan-yibDp-GFPuv, described above. Briefly, overnight LB cultures were used to inoculate various media in 96-well plates. Biomass and GFPuv levels were measured after 24 hr. Importantly, no phosphate was added to these media, as adequate batch phosphate is supplied in the complex nutrient sources (yeast extract and casamino acids); results are given in Figure 3. Models built based on these results did not predict significant improvements in expression over the best performing experimentally tested formulations Figure 3a. The media formulation producing the most GFPuv (as measured by relative fluorescence), was renamed AB and used in subsequent studies. To evaluate the time course of growth, phosphate depletion and autoinduction in AB, DLF_R002 pHCKan-yibDp-GFPuv, was grown in AB in the BioLector™ Microreactor; results are given in Figure 3b.

FIGURE 3.

FIGURE 3

Open in a new tab

Media development using design of experiment (DoE) methodology. 212 Media formulations were evaluated for autoinduction based on phosphate depletion, each comprising different “levels” of casamino acids, yeast extract, trace metals (TM mix), calcium sulfate (CaSO4), magnesium sulfate (MgSO4), iron(II) sulfate (FeSO4), ammonium sulfate ((NH4)2SO4), and citric acid. (a) Upper panel: GFP (green bars) and OD600 nm (gray bars) rank ordered plot for all media formulations. Standard deviations are from triplicate experiments. Lower panel: Nutrient concentration levels for all media (refer to Supporting Information Materials Section 4, for specific concentrations for each level). Strain DLF_R002 with plasmid pHCKan-yibDp-GFPuv was used for all experiments. (b) GFP fluorescence (green line), phosphate levels (black circles) and OD600 nm (black line) for strain DLF_R002 with plasmid pHCKan-yibDp-GFPuv in media #36 (autoinduction broth) media. Standard deviations (shaded regions) are from triplicate experiments. OD, optical density

3.4 |. Comparison with current approaches

With the successful development of an optimal AB, we turned to a head to head comparison of this approach with the traditional protocols based in LB media as well as the lactose-based autoinduction system as developed by Studier (2005, 2014). Due to the availability of a pET-mCherry plasmid (Table 1) mCherry was used as the reporter for this comparison. Specifically, induction of mCherry in BL21(DE3) with pLysS and pETM6-mCherry, using either IPTG-based induction in LB media, or lactose autoinduction media was compared to strain DLF_R002 with plasmid pHCKan-yibDp-mCherry in AB. To monitor not only endpoint expression but the dynamics of growth and autoinduction, these studies were performed in the BioLector™. Results are shown in Figure 4. As expected, using E. coli BL21(DE3) and pET-based expression, lactose-based autoinduction media enabled higher cell densities and higher expression levels of mCherry than induction with IPTG (Figure 4a). Phosphate-based autoinduction using strain DLF_R002 enabled a further 40% increase in final mCherry levels at 24 hr over BL21(DE3) (Figure 4b; refer to Figure S3 for a comparison of BL21(DE3) vs. BL21(DE3) plus pLys). Cytometry was used to further characterize these two expression systems (Figure 4c). Phosphate-based autoinduction not only had more homogeneous induction but also more expression per cell. In addition, one of the major potential reported advantages of BL21(DE3) and related strains is reduction in Lon protease activity (Ratelade et al., 2009). To investigate the impact of Lon activity in these strains, a previously reported fluorescent Lon substrate was used to monitor the impact of this protease. Specifically, a circular permutation variant of GFP with a Lon degradation tag (GFP-β20-cp6) was used (Wohlever, Nager, Baker, & Sauer, 2013; results are given in Figure 4d). The expression level of the Lon substrate was significantly reduced compared to a non-Lon substrate for both strains, but at least with this specific reporter, no significant difference in cell specific Lon activity was observed between BL21(DE3) and DLF_R002.

FIGURE 4.

FIGURE 4

Open in a new tab

Head to head comparison of autoinduction via phosphate depletion with pET-based expression in BL21(DE3). (a) pET-based mCherry expression in BL21(DE3) with pLysS. mCherry (red lines) and biomass levels (OD600 nm, black lines) over time. Solid lines: lactose-based autoinduction. Dashed lines: IPTG induction in LB media. (b) yibDp-based mCherry expression in DLF_R002 in AB media mCherry (red lines) and biomass levels (OD600 nm, black lines). (c) Cytometry of induced populations (gray: empty vector control, red: pET-mCherry in BL21(DE3) + pLysS, green: yibDp-mCherry in DLF_R002). (d) Expression of the Lon substrate (GFP-β20-cp6) in BL21(DE3) and DLF_R002. Normalized fluorescence is relative fluorescence normalized to OD. Black line: GFP control (non-Lon substrate) in DLF_R002. Red line: BL21(DE3) expressing GFP-β20-cp6. Green line: DLF_R002 expressing GFP-β20-cp6. Shaded areas are standard deviations of at least three replicates. AB, autoinduction broth; IPTG, isopropyl β-D-1-thiogalactopyranoside; LB, Luria broth; OD, optical density

3.5 |. Optimization of high throughput expression protocols

The DoE results discussed above in Figure 2, were generated in 96-well plates using high shaking speeds in combination with the Duetz system, which utilizes a series of specialty plate covers to minimize evaporative volume loss, while enabling adequate aeration (Duetz, Kuhner, & Lohser, 2006; Duetz & Witholt, 2001). As rapid growth and expression are not only a function of media, but culture aeration, we sought to evaluate the optimal aeration conditions for microtiter plate-based expression (96- and 384-well plates). In addition to orbital shaking speed and orbit diameter, culture volume (impacting the surface area to volume ratio) can have a significant impact on oxygen transfer and in this case protein expression as shown in Figure 5. In standard 96-well plates, volumes <100 μl gave optimal expression. As 384-well plates have a very small area, the surface tension at the culture meniscus can limit mixing. As a result, small amounts of surfactant (commercial antifoam) were added to improve aeration in 384-well plates; Figure S4. In 384-well cultures volumes >20 μl gave optimal expression with AB media. As can be seen in Figure 5, expression levels, using AB in 384-well plates, did not reach levels observed in the 96-well plates or other culture systems. We hypothesized this was due to remaining mass transfer limitations. We tested this hypothesis by evaluating an autoinduction media identified in the DoE results (AB-C7) that yielded reduced biomass and expression levels, but as a result would have a lower maximal aeration requirement. With lowered biomass levels, and aeration demands, expression levels in 384-well plates reached that of other culture systems using this media (Figures S5 and S6). Although total protein levels are higher in AB media, the use of AB-C7 media may be preferred when using 384-well plates to minimize oxygen limitations.

FIGURE 5.

FIGURE 5

Open in a new tab

Optimization of autoinduction in batch cultures at various scales. Impact of various fill volumes on expression in AB. Varying fill volumes in 384- and 96-well plates as well as 250 ml baffled Erlenmeyer and 2.8 L Fernbach flasks. When using 384-well plates, 0.05% polypropylene glycol (2,000 MW) was added to the media. DLF_R002 with plasmid pHCKan-yibDp-GFPuv was used for all experiments

3.6 |. Development of shake flask protocols

For any expression protocol to be widely applicable, it cannot rely on controlled bioreactors and/or specialty plate systems, but be accessible to the average laboratory. Toward this goal, we turned to the optimization of the protocol in shake flask cultures. As mentioned above, one primary difference between bioreactor experiments and shake flask cultivation is oxygen transfer. While instrumented bioreactors and microreactors such as the BioLector™ can easily meet these mass transfer targets, standard shake flask have reported oxygen transfer rates anywhere from 20 mmol·L−1·hr−1 (for unbaffled flasks) to 120 mmol·L−1·hr−1 for baffled glassware (Running & Bansal, 2016). A key potential consequence of shake flasks is oxygen limitation and reduced growth rates and expression. As a consequence we sought to evaluate the optimal culture conditions to achieve maximal expression in shake flask cultures with a focus on baffled 250 ml Erlenmeyer flasks and 2.8 L Fernbach flasks. As seen in Figure 5, again culture volume plays a key role in optimal protein expression, with 20 ml or lower being optimal in baffled 250 ml Erlenmeyer flasks and 100 ml or lower being optimal in 2.8 L Fernbach flasks. These results were obtained in shakers where an adhesive mat is used to hold flasks and shaking speeds are limited to 150 rpm. Using clamps, higher shaking speeds may enable optimal expression using larger shake flask fill volumes.

3.7 |. Utility with a diverse group of recombinant proteins

All results discussed to this point relied on easily quantified reporter proteins (GFPuv and mCherry), which are easily expressed to high levels in most expression hosts. To evaluate the broader applicability of the approach, the expression of a group of other diverse proteins was evaluated in several vector backbone contexts in the phosphate autoinduction protocol. These included: a borneol diphosphate synthase, a terpene synthase with a C-terminal mCherry tag (Wise, Savage, Katahira, & Croteau, 1998), a mutant alanine dehydrogenase (Lerchner, Jarasch, & Skerra, 2016), a malonyl-CoA synthetase (An & Kim, 1998), a benzoylformate decarboxylase (Tsou et al., 1990), glutathione S-transferase (Oakley, 2011), HIV-1 Nef protein (Pereira & daSilva, 2016), a mutant citramalate synthase (Atsumi & Liao, 2008), and a human glycine acyltransferase with an N-terminal chitin binding tag (Waluk, Schultz, & Hunt, 2010; refer to Table 1 for construct details). As can be seen in Figure 6, expression levels ranged from ~10% of total protein for a large terpene synthase to 55% in the case of alanine dehydrogenase, achieving maximal protein concentrations of 275 mg/gCDW in the best case. In most cases these proteins were found predominantly in the soluble fraction with only one protein (hGLYAT2) being exclusively in the insoluble fraction (Figure S7 and Table S1.1). In fact, when given a solubility score, as described previously (Berrow et al., 2006), all proteins (except hGLYAT2) received the highest score of 3.

FIGURE 6.

FIGURE 6

Open in a new tab

Autoinduction in AB in 96-well plates for a diverse set of recombinant proteins including: GFPuv, mCherry, AlaDh* (a mutant alanine dehydrogenase), Nef (HIV-1 Nef protein), hGLYAT2 (human glycine acyltransferase-2 an N-terminal chitin binding tag), cimA3.7 (a mutant citramalate synthase), GST, mdlC (benzoylformate decarboxylase), matB (malonyl-CoA synthetase), and SBS (bornyl-diphosphate synthase with a C-terminal mCherry tag). Percent of total expression is given for three replicates. Refer to Figure S7 for an example SDS-PAGE result. AB, autoinduction broth; SDS-PAGE, sodium dodecyl sulfate–polyacrylamide gel electrophoresis

4 |. DISCUSSION

Two-stage expression induced upon phosphate depletion enables a facile and versatile approach to routine high level recombinant protein production from high throughput screens to instrumented bioreactors. Numerous previous studies have utilized phosphate starvation to induce protein expression in E. coli as well as many other hosts, including Bacillus, Streptomyces, and Pichia (Ahn et al., 2009; Kerovuo, von Weymarn, Povelainen, Auer, & Miasnikov, 2000; Lübke et al., 1995; Sola-Landa, Rodríguez-García, Franco-Domínguez, & Martín, 2005; Song et al., 2017; Trung et al., 2019). Some of these systems enable high levels of expression and in some cases autoinduction protocols have been developed (Kerovuo et al., 2000; Lübke et al., 1995; Trung et al., 2019). In many cases, protein titers are reported in units of enzyme making direct comparisons across studies difficult (Ahn et al., 2009; Kerovuo et al., 2000; Lübke et al., 1995; Sola-Landa et al., 2005; Song et al., 2017; Trung et al., 2019). In the common laboratory workhorse, E. coli, the phoA (alkaline phosphatase) promoter has primarily been leveraged for expression upon phosphate starvation, (Lübke et al., 1995; Song et al., 2017) where expression levels primarily vary as a function of the target protein. Recently, building on the results and methodologies in this study, our group has reported the evaluation of several E. coli promoters induced upon phosphate starvation. In these studies, with an identical GFP reporter, the modified yibD promoter enabled increased expression levels when compared to the phoA promoter (Moreb et al., 2020). Of course, if another promoter was chosen as a starting point, a similar optimization methodology may potentially yield very different results. That is to say that optimal media and process conditions are not expected to be universal and may likely be promoter dependent. Nonetheless the general approach taken here may be adapted for any autoinduction process.

In the case of GFPuv, engineered strains along with the modified yibD promoter enable protein titers approaching 2 g/L in batch microtiter plates and shake flasks. These titers correspond to protein yields of 20 μg of protein per well in 384-well plates, 170 μg per well in 96-well plates, and 40 and 180 mg of protein in 250 ml Erlenmeyer and Fernbach flasks, respectively. Importantly, current results also support homogenous expression using phosphate depletion. Expression levels will of course vary as a function of the protein and expression construct, but initial testing with additional proteins supports expression levels from 10% to 55% of total cellular protein. At the high end is ~275 mg/gCDW of recombinant protein and represents significant improvements in heterologous protein expression in E. coli, although we anticipate that expression levels using this approach, as with any expression system, will vary as a function of the target protein (Chew, Tan, Boo, & Tey, 2012; Striedner et al., 2010). More work is needed to better understand the mechanisms of unexpectedly high expression levels observed in this system. Initial adaptation to instrumented bioreactors, enabled GFP titers as high as 2.7 g/L, 270 mg/gCDW and 55% expression, with 10 gCDW/L. Process intensification to increase biomass levels to ~30 gCDW/L, while maintaining specific expression levels, resulting in a approximately threefold improvement in GFP titers reaching 8.1 g/L. Further optimization of bioreactor protocols may enable much higher cell density cultures. If truly high cell density fermentations (from 50 to 100 gCDW/L of biomass) can be developed with equivalent expression levels, protein titers in the range of 15–30 g/L or higher in some cases can be expected.

There are, however, several challenges with the existing protocol. First, proteins of interest must be cloned into a plasmid with the yibDp promoter. Adaptation of the system for use with existing pET-based plasmids would also be of utility for proteins that are already cloned into these standard vectors. Second, preparation of AB media is more complicated than making routine LB media. Finally, this system is not designed to address challenges with difficult to express proteins common to other standard heterologous expression systems.

Despite these potential limitations, the development of strains, plasmids and protocols for autoinduction based on phosphate depletion not only enables improved expression, with impressive protein titers, but also a scalable methodology. A single host and plasmid can be used in high throughput screening of initial expression constructs or mutant variants all the way through to instrumented bioreactors. These results support the biosynthetic potential of phosphate depleted stationary phase cultures of E. coli. Decoupling growth from production also has the potential to enable future studies focused on key remaining limitations in protein biosynthesis in this well characterized host.

Supplementary Material

supinfo

ACKNOWLEDGMENTS

The authors would like to acknowledge the following support: the NSF EAGER: #1445726 DARPA# HR0011-14-C-0075, ONR YIP #N00014-16-1-2558 and DOE EERE grant #EE0007563, as well as the North Carolina Biotechnology Center 2018-BIG-6503. John S. Decker was supported in part by the NIH Biotechnology Training Grant (T32GM008555). The authors would also like to thank M. Munson for help with cytometry experiments.

Funding information

Defense Advanced Research Projects Agency, Grant/Award Number: DARPA# HR0011-14-C-0075; National Institutes of Health, Grant/Award Number: T32GM008555; Office of Naval Research, Grant/Award Number: YIP #N00014-16-1-2558; Department of Energy EERE, Grant/Award Number: #EE0007563; North Carolina Biotechnology Center, Grant/Award Number: 2018-BIG-6503; National Science Foundation, Grant/Award Number: EAGER: #1445726

Footnotes

CONFLICT OF INTERESTS

Michael D. Lynch and Zhixia Ye have a financial interest in DMC Biotechnologies Inc. Romel Menacho-Melgar, Zhixia Ye, and Michael D. Lynch have filed patent applications on strains and methods discussed in this manuscript. Other authors declare that there are no conflict of interests.

SUPPORTING INFORMATION

Additional supporting information may be found online in the Supporting Information section.

REFERENCES

  1. Ahn J, Hong J, Park M, Lee H, Lee E, Kim C, … Lee H (2009). Phosphate-responsive promoter of a Pichia pastoris sodium phosphate symporter. Applied and Environmental Microbiology, 75(11), 3528–3534. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. An JH, & Kim YS (1998). A gene cluster encoding malonyl-CoA decarboxylase (MatA), malonyl-CoA synthetase (MatB) and a putative dicarboxylate carrier protein (MatC) in Rhizobium trifolii—Cloning, sequencing, and expression of the enzymes in Escherichia coli. European Journal of Biochemistry, 257(2), 395–402. [DOI] [PubMed] [Google Scholar]
  3. Anilionyte O, Liang H, Ma X, Yang L, & Zhou K (2018). Short, auto-inducible promoters for well-controlled protein expression in Escherichia coli. Applied Microbiology and Biotechnology, 102(16), 7007–7015. [DOI] [PubMed] [Google Scholar]
  4. Atsumi S, & Liao JC (2008). Directed evolution of Methanococcus jannaschii citramalate synthase for biosynthesis of 1-propanol and 1-butanol by Escherichia coli. Applied and Environmental Microbiology, 74(24), 7802–7808. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Baba T, Ara T, Hasegawa M, Takai Y, Okumura Y, Baba M, … Mori H (2006). Construction of Escherichia coli K-12 in-frame, single-gene knockout mutants: The Keio collection. Molecular Systems Biology, 2(1), 2006.0008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Baez A, Majdalani N, & Shiloach J (2014). Production of recombinant protein by a novel oxygen-induced system in Escherichia coli. Microbial Cell Factories, 13(1), 50. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Ben R, Jiying F, Jian’an S, Tu TN, Jing S, Jingsong Z, … Yaling S (2016). An auto-inducible expression system based on the RhlI-RhlR quorum-sensing regulon for recombinant protein production in E. coli. Biotechnology and Bioprocess Engineering, 21(1), 160–168. [Google Scholar]
  8. Berrow NS, Büssow K, Coutard B, Diprose J, Ekberg M, Folkers GE, … Busso D (2006). Recombinant protein expression and solubility screening in Escherichia coli: A comparative study. Acta Crystallographica. Section D Biological Crystallography, 62(Pt 10), 1218–1226. [DOI] [PubMed] [Google Scholar]
  9. Briand L, Marcion G, Kriznik A, Heydel JM, Artur Y, Garrido C, … Neiers F (2016). A self-inducible heterologous protein expression system in Escherichia coli. Scientific Reports, 6, 33037. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Burg JM, Cooper CB, Ye ZX, Reed BR, Moreb EA, & Lynch MD (2016). Large-scale bioprocess competitiveness: The potential of dynamic metabolic control in two-stage fermentations. Current Opinion in Chemical Engineering, 14, 121–136. [Google Scholar]
  11. Chew FN, Tan WS, Boo HC, & Tey BT (2012). Statistical optimization of green fluorescent protein production from Escherichia coli BL21(DE3). Preparative Biochemistry & Biotechnology, 42(6), 535–550. [DOI] [PubMed] [Google Scholar]
  12. Chubukov V, & Sauer U (2014). Environmental dependence of stationary-phase metabolism in Bacillus subtilis and Escherichia coli. Applied and Environmental Microbiology, 80(9), 2901–2909. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Crameri A, Whitehorn EA, Tate E, & Stemmer WP (1996). Improved green fluorescent protein by molecular evolution using DNA shuffling. Nature Biotechnology, 14(3), 315–319. [DOI] [PubMed] [Google Scholar]
  14. Datsenko KA, & Wanner BL (2000). One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proceedings of the National Academy of Sciences of the United States of America, 97(12), 6640–6645. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Duetz WA, Kuhner M, & Lohser R (2006). Microbial and cell growth in microtiter plates. Genetic Engineering News, 26(11), 44–46. [Google Scholar]
  16. Duetz WA, & Witholt B (2001). Effectiveness of orbital shaking for the aeration of suspended bacterial cultures in square-deepwell microtiter plates. Biochemical Engineering Journal, 7(2), 113–115. [DOI] [PubMed] [Google Scholar]
  17. Eiteman MA, & Altman E (2006). Overcoming acetate in Escherichia coli recombinant protein fermentations. Trends in Biotechnology, 24(11), 530–536. [DOI] [PubMed] [Google Scholar]
  18. Gill RT, DeLisa MP, Shiloach M, Holoman TR, & Bentley WE (2000). OmpT expression and activity increase in response to recombinant chloramphenicol acetyltransferase overexpression and heat shock in E. coli. Journal of Molecular Microbiology and Biotechnology, 2(3), 283–289. [PubMed] [Google Scholar]
  19. Glazyrina J, Materne E-M, Dreher T, Storm D, Junne S, Adams T, … Neubauer P (2010). High cell density cultivation and recombinant protein production with Escherichia coli in a rocking-motion-type bioreactor. Microbial Cell Factories, 9, 42. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Grenier F, Matteau D, Baby V, & Rodrigue S (2014). Complete genome sequence of Escherichia coli BW25113. Genome Announcements, 2(5), e01038-14. 10.1128/genomeA.01038-14 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Huber R, Roth S, Rahmen N, & Buchs J (2011). Utilizing high-throughput experimentation to enhance specific productivity of an E. coli T7 expression system by phosphate limitation. BMC Biotechnology, 11, 22. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Jian J, Zhang S-Q, Shi Z-Y, Wang W, Chen G-Q, & Wu Q (2010). Production of polyhydroxyalkanoates by Escherichia coli mutants with defected mixed acid fermentation pathways. Applied Microbiology and Biotechnology, 87(6), 2247–2256. [DOI] [PubMed] [Google Scholar]
  23. Jones JA, Vernacchio VR, Lachance DM, Lebovich M, Fu L, Shirke AN, … Koffas MA (2015). ePathOptimize: A combinatorial approach for transcriptional balancing of metabolic pathways. Scientific Reports, 5, 11301. 10.1038/srep11301 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Kerovuo J, von Weymarn N, Povelainen M, Auer S, & Miasnikov A (2000). A new efficient expression system for Bacillus and its application to production of recombinant phytase. Biotechnology Letters, 22(16), 1311–1317. [Google Scholar]
  25. Labhsetwar P, Cole JA, Roberts E, Price ND, & Luthey-Schulten ZA (2013). Heterogeneity in protein expression induces metabolic variability in a modeled Escherichia coli population. Proceedings of the National Academy of Sciences of the United States of America, 110(34), 14006–14011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Lerchner A, Jarasch A, & Skerra A (2016). Engineering of alanine dehydrogenase from Bacillus subtilis for novel cofactor specificity. Biotechnology and Applied Biochemistry, 63(5), 616–624. [DOI] [PubMed] [Google Scholar]
  27. Li X-T, Thomason LC, Sawitzke JA, Costantino N, & Court DL (2013). Positive and negative selection using the tetA-sacB cassette: Recombineering and P1 transduction in Escherichia coli. Nucleic Acids Research, 41(22), e204. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Long CP, & Antoniewicz MR (2014). Quantifying biomass composition by gas chromatography/mass spectrometry. Analytical Chemistry, 86(19), 9423–9427. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Lübke C, Boidol W, & Petri T (1995). Analysis and optimization of recombinant protein production in Escherichia coli using the inducible pho A promoter of the E. coli alkaline phosphatase. Enzyme and Microbial Technology, 17, 923–928. 10.1016/0141-0229(94)00130-j [DOI] [Google Scholar]
  30. Lynch MD (2014). US Patent 8,652,816. Boulder, CO: OPX Biotechnologies Inc.
  31. Lynch MD (2016). Into new territory: Improved microbial synthesis through engineering of the essential metabolic network. Current Opinion in Biotechnology, 38, 106–111. [DOI] [PubMed] [Google Scholar]
  32. Lynch MD, Gill RT, & Lipscomb TEW (2018). US Patent 10,100,342. Wayzata, MN: Cargill, Incorporated.
  33. Lynch MD, Trahan AD, Rodriguez DL, Ye Z, Cooper CB, & Bozdag A (2015). United Kingdom Patent GB 2,528,177. [Google Scholar]
  34. Moreb EA, Ye Z, Efromson JP, Hennigan JN, Menacho-Melgar R, & Lynch MD (2020). Robustness testing and scalability of phosphate regulated promoters useful for two-stage autoinduction in E. coli. bioRxiv, 10.1021/acssynbio.0c00182 [DOI] [PubMed] [Google Scholar]
  35. Neidhardt FC, Ingraham JL, & Schaechter M (1990). Physiology of the bacterial cell: A molecular approach. Sunderland, MA: Sinauer Associates. [Google Scholar]
  36. Nocadello S, & Swennen EF (2012). The new pLAI (lux regulon based auto-inducible) expression system for recombinant protein production in Escherichia coli. Microbial Cell Factories, 11, 3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Novick A, & Weiner M (1957). Enzyme induction as an all-or-none phenomenon. Proceedings of the National Academy of Sciences of the United States of America, 43(7), 553–566. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Oakley A (2011). Glutathione transferases: A structural perspective. Drug Metabolism Reviews, 43(2), 138–151. [DOI] [PubMed] [Google Scholar]
  39. Pereira EA, & daSilva LLP (2016). HIV-1 Nef: Taking control of protein trafficking. Traffic, 17(9), 976–996. [DOI] [PubMed] [Google Scholar]
  40. Ratelade J, Miot M-C, Johnson E, Betton J-M, Mazodier P, & Benaroudj N (2009). Production of recombinant proteins in the lon-deficient BL21(DE3) strain of Escherichia coli in the absence of the DnaK chaperone. Applied and Environmental Microbiology, 75, 3803–3807. 10.1128/aem.00255-09 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Running JA, & Bansal K (2016). Oxygen transfer rates in shaken culture vessels from fernbach flasks to microtiter plates. Biotechnology and Bioengineering, 113(8), 1729–1735. [DOI] [PubMed] [Google Scholar]
  42. Sanchez-Garcia L, Martín L, Mangues R, Ferrer-Miralles N, Vázquez E, & Villaverde A (2016). Recombinant pharmaceuticals from microbial cells: A 2015 update. Microbial Cell Factories, 15, 33. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Sharan SK, Lynn CT, Kuznetsov SG, & Court DL (2009). Recombineering: A homologous recombination-based method of genetic engineering. Nature Protocols, 4(2), 206–223. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Shiloach J, & Fass R (2005). Growing E. coli to high cell density—A historical perspective on method development. Biotechnology Advances, 23(5), 345–357. [DOI] [PubMed] [Google Scholar]
  45. Shiloach J, Kaufman J, Guillard AS, & Fass R (1996). Effect of glucose supply strategy on acetate accumulation, growth, and recombinant protein production by Escherichia coli BL21 (λDE3) and Escherichia coli JM109. Biotechnology and Bioengineering, 49(4), 421–428. [DOI] [PubMed] [Google Scholar]
  46. Sola-Landa A, Rodríguez-García A, Franco-Domínguez E, & Martín JF (2005). Binding of PhoP to promoters of phosphate-regulated genes in streptomyces coelicolor: Identification of PHO boxes. Molecular Microbiology, 56(5), 1373–1385. [DOI] [PubMed] [Google Scholar]
  47. Song H, Jiang J, Wang X, & Zhang J (2017). High purity recombinant human growth hormone (rhGH) expression in Escherichia coli under phoA promoter. Bioengineered, 8, 147–153. 10.1080/21655979.2016.1212137 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Striedner G, Pfaffenzeller I, Markus L, Nemecek S, Grabherr R, & Bayer K (2010). Plasmid-free T7-based Escherichia coli expression systems. Biotechnology and Bioengineering, 105(4), 786–794. [DOI] [PubMed] [Google Scholar]
  49. Studier FW (1991). Use of bacteriophage T7 lysozyme to improve an inducible T7 expression system. Journal of Molecular Biology, 219(1), 37–44. [DOI] [PubMed] [Google Scholar]
  50. Studier FW (2005). Protein production by auto-induction in high density shaking cultures. Protein Expression and Purification, 41(1), 207–234. [DOI] [PubMed] [Google Scholar]
  51. Studier FW (2014). Stable expression clones and auto-induction for protein production in E. coli. Methods in Molecular Biology, 1091, 17–32. [DOI] [PubMed] [Google Scholar]
  52. Trung NT, Hung NM, Thuan NH, Canh NX, Schweder T, & Jürgen B (2019). An auto-inducible phosphate-controlled expression system of Bacillus licheniformis. BMC Biotechnology, 19(1), 3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Tsou AY, Ransom SC, Gerlt JA, Buechter DD, Babbitt PC, & Kenyon GL (1990). Mandelate pathway of Pseudomonas putida: Sequence relationships involving mandelate racemase, (S)-mandelate dehydrogenase, and benzoylformate decarboxylase and expression of benzoylformate decarboxylase in Escherichia coli. Biochemistry, 29, 9856–9862. 10.1021/bi00494a015 [DOI] [PubMed] [Google Scholar]
  54. Waegeman H, Beauprez J, Moens H, Maertens J, Mey MD, Foulquié-Moreno MR, … Soetaert W (2011). Effect of iclR and arcA knockouts on biomass formation and metabolic fluxes in Escherichia coli K12 and its implications on understanding the metabolism of Escherichia coli BL21 (DE3). BMC Microbiology, 11, 70. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Waegeman H, Maertens J, Beauprez J, De Mey M, & Soetaert W (2012). Effect of iclR and arcA deletions on physiology and metabolic fluxes in Escherichia coli BL21 (DE3). Biotechnology Letters, 34(2), 329–337. [DOI] [PubMed] [Google Scholar]
  56. Waluk DP, Schultz N, & Hunt MC (2010). Identification of glycine N-acyltransferase-like 2 (GLYATL2) as a transferase that produces N-acyl glycines in humans. FASEB Journal, 24(8), 2795–2803. [DOI] [PubMed] [Google Scholar]
  57. White CB, Chen Q, Kenyon GL, & Babbitt PC (1995). A novel activity of OmpT. proteolysis under extreme denaturing conditions. The Journal of Biological Chemistry, 270(22), 12990–12994. [DOI] [PubMed] [Google Scholar]
  58. Wise ML, Savage TJ, Katahira E, & Croteau R (1998). Monoterpene synthases from common sage (Salvia officinalis). cDNA isolation, characterization, and functional expression of (+)-sabinene synthase, 1,8-cineole synthase, and (+)-bornyl diphosphate synthase. The Journal of Biological Chemistry, 273(24), 14891–14899. [DOI] [PubMed] [Google Scholar]
  59. Wohlever ML, Nager AR, Baker TA, & Sauer RT (2013). Engineering fluorescent protein substrates for the AAA+ lon protease. Protein Engineering, Design & Selection: PEDS, 26(4), 299–305. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Yang J, Sun B, Huang H, Jiang Y, Diao L, Chen B, … Yang S (2014). High-efficiency scarless genetic modification in Escherichia coli by using lambda red recombination and I-SceI cleavage. Applied and Environmental Microbiology, 80(13), 3826–3834. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

supinfo
NIHMS1614236-supplement-supinfo.docx (1.6MB, docx)

RESOURCES