Metabolic Engineering 48 (2018) 121–128

Contents lists available at ScienceDirect

Metabolic Engineering
journal homepage: www.elsevier.com/locate/meteng

Synthetic auxotrophs for stable and tunable maintenance of plasmid copy T

number
⁎
Chae Won Kanga,1, Hyun Gyu Lima,1, Jina Yangb, Myung Hyun Noha, Sang Woo Seob, ,
⁎⁎
Gyoo Yeol Junga,c,
a
Department of Chemical Engineering, Pohang University of Science and Technology, 77 Cheongam-Ro, Nam-Gu, Pohang, Gyeongbuk 37673, Republic of Korea
b
School of Chemical and Biological Engineering, Institute of Chemical Process, Seoul National University, 1 Gwanak-Ro, Gwanak-Gu, Seoul 08826, Republic of Korea
c
School of Interdisciplinary Bioscience and Bioengineering, Pohang University of Science and Technology, 77 Cheongam-Ro, Nam-Gu, Pohang, Gyeongbuk 37673,
Republic of Korea

A R T I C LE I N FO A B S T R A C T

Keywords: Although plasmid-based expression systems have advantages in multi-copy expression of genes, heterogeneity of
Synthetic auxotrophy plasmid copy number (PCN) in individual cells is inevitable even with the addition of antibiotics. Here, we
Plasmid copy number developed a synthetic auxotrophic system for stable and tunable maintenance of the PCN in Escherichia coli
Expression control without addition of antibiotics. This auxotroph expresses infA, one of the essential genes encoding a translation
Antibiotic-free
initiation factor, on a plasmid instead of on the chromosome. With this system, the gene expression was stably
Stable expression
Plasmid stability
maintained for 40 generations with minimized cell-to-cell variation under antibiotic-free conditions. Moreover,
varying the expression level of infA enabled us to rationally tune the PCN by more than 5.6-fold. This antibiotic-
free PCN control system significantly improved the production of itaconic acid and lycopene compared to the
conventional system based on antibiotics (2-fold). Collectively, the developed strategy could be a platform for
the production of value-added products in antibiotic-free cultivation.

1. Introduction that the cell-to-cell variations can induce stochastic phenotypes which
greatly lower the productivity of bioprocesses (Xiao et al., 2016).
Plasmids serve as DNA shuttles that deliver genes to host cells To maintain plasmids, antibiotics are usually added to minimize
(Taylor et al., 2004). Easy manipulation and high transformation effi- subpopulations that have none or a reduced number of plasmids rising
ciency facilitate a wide range of applications in reconstruction of me- from segregational instability (Lenski and Bouma, 1987; Popov et al.,
tabolic pathways (Jung et al., 2016; Noh et al., 2017a), complex genetic 2014). However, concerns about the use of antibiotics have been raised
circuits (Brophy and Voigt, 2014; Min et al., 2017), and production of due to expensive price (Vandermeulen et al., 2011), contamination of
recombinant proteins (Chen, 2012). Notably, plasmids are known to be the final product (Peubez et al., 2010), and potential emergence of
the only way to achieve hundreds of gene doses. In plasmid-related multi-drug resistant organisms (Davison, 1999). More importantly,
systems, plasmid copy number (PCN), which is mainly governed by antibiotics may not be suitable for stable and tight maintenance of the
replication origin, is an important characteristic. Varying the PCN en- plasmid for long-term cultivation due to its degradation (Jung et al.,
ables controllable expression of target genes or pathways; such varia- 1988). Therefore, the development of efficient and predictable biolo-
tion has been utilized as a tool for metabolic engineering purpose in gical systems with improved performance requires that PCNs are tun-
balancing pathways to achieve efficient production of biochemicals able and tightly controllable even without the addition of antibiotics.
(Ajikumar et al., 2010; Lim et al., 2016; Xu et al., 2013). However, Previously, antibiotic-free plasmid systems have been developed to
there are few available replication origins with constrained ranges of enable the stable maintenance of plasmids (Hanak and Cranenburgh,
PCNs; this can be a limitation to the amenable design of biological 2001; Mignon et al., 2015; Peubez et al., 2010; Sodoyer et al., 2012).
systems. Furthermore, PCN easily fluctuates among cells because of its Examples of such systems include essential gene complementation
intrinsic instability (Xiao et al., 2016; Zenobi, 2013). It has been known systems (essential genes are expressed by plasmids rather than

⁎
Corresponding author.
⁎⁎
Corresponding author at: Gyoo Yeol Jung Mailing address: Department of Chemical Engineering, Pohang University of Science and Technology, 77 Cheongam-Ro, Nam-Gu, Pohang,
Gyeongbuk 37673, Republic of Korea.
E-mail addresses: swseo@snu.ac.kr (S.W. Seo), gyjung@postech.ac.kr (G.Y. Jung).
1
These authors contributed equally to this work.

https://doi.org/10.1016/j.ymben.2018.05.020
Received 8 November 2017; Received in revised form 28 May 2018; Accepted 31 May 2018
Available online 02 June 2018
1096-7176/ © 2018 International Metabolic Engineering Society. Published by Elsevier Inc. All rights reserved.
C.W. Kang et al. Metabolic Engineering 48 (2018) 121–128

chromosomes) (Hägg et al., 2004), toxin-antitoxin systems (cells must F1, O-mCherry-F2, and O-mCherry-B primers with pZ-mCherry (Seo
harbor plasmids in order to detoxify toxins expressed by chromosome) et al., 2013a, 2013b) as a template into SacI site of pETDuet. Then, an
(Szpirer and Milinkovitch, 2005), operator-repressor titration systems efp expression cassette was amplified using O-efp-F1, O-efp-F2 and O-
(operator-carrying plasmids competitively titrate the action of re- efp-B primers and inserted into XhoI site of pET-mCherry plasmid
pressors that regulate chromosomal genes) (Cranenburgh et al., 2001), (Table S1). pFRT2161 plasmid was constructed via TA cloning of am-
and RNA-based selection marker system (RNA I derived from the ColE1 plified DNA fragment using FRT2161_F and FRT2161_B primers with
origin of replication inhibits the translation of repressor on chromo- pFRT72variant plasmid as a template.
somal essential gene) (Mairhofer et al., 2010). Among them, essential It should be noted that the synthetic constitutive promoter and
gene complementation has been generally favored for several reasons: terminator (BBa_B1002 and BBa_B1005) sequences were obtained from
i) the presence of many putative markers in the host genome, ii) no the Registry of Standard Biological Parts (http://parts.igem.org/). In
toxicity to cells, and iii) no off-target effects. In particular, several es- addition, the 5′-UTR sequences were designed by UTR Designer (http://
sential genes involved in key metabolites (Chen et al., 2012; Vidal et al., sbi.postech.ac.kr/rbs) (Seo et al., 2013a, 2013b). Especially, desired
2008), redox cofactors (Dong et al., 2010), and cell wall synthesis expression levels of essential genes were set to 1000 for low expression;
(Degryse, 1991) were employed in a few host organisms. Although 100,000 was input for mid expression of reporter genes.
these previous studies demonstrated great potential in stable plasmid
maintenance without the addition of antibiotics, there still exist diffi- 2.3. Construction of auxotrophic strains for the STAPL system
culties in the tight control of PCN due to potential cross-feeding,
especially in the case of metabolite auxotrophy (Hägg et al., 2004). For To develop a model STAPL system, a synthetic expression cassette
example, cells do not maintain plasmid can exist as a minor population for infA from E. coli W3110 was inserted into a rare restriction enzyme
due to the secreted metabolites from the plasmid-harboring cells. Fur- site (Bsu36I) of target plasmids. The designed plasmid was introduced
thermore, these systems have not been systematically characterized at to E. coli W3110; thereafter, the chromosomal infA was deleted for
the single-cell level yet. auxotrophic maintenance and the strain was labeled STG1. For chro-
In this study, we developed synthetic auxotrophs with controllable mosomal deletion of infA, pKD46 was additionally introduced to the
PCNs in Escherichia coli via expression of infA (encoding translation host strain for Red recombination (Datsenko and Wanner, 2000). Then,
initiation factor IF-1) using predictable genetic parts. We named the a kanamycin resistance gene fragment amplified from pFRT72variant
developed strategy the Stable and TunAble PLasmid (STAPL) system. plasmid by polymerase chain reaction (PCR) using the primers D-infA-F
With the STAPL system, we demonstrated that plasmid could be stably and D-infA-B was introduced to the cell. The disruption of chromosomal
maintained for more than 40 generations even without the addition of infA was confirmed by colony PCR and sequencing. The inserted gene
antibiotics. More importantly, by controlling the expression level of for kanamycin resistance was removed by introduction of pCP20
infA, the PCN was tunable up to 5.6-fold with minimized cell-to-cell (Datsenko and Wanner, 2000). When efp was employed as an auxo-
variations. We also illustrated the adaptability of the STAPL system by trophic marker, D-efp-F and D-efp-B primers and pFRT2161 plasmid
demonstrating significant increases in production of itaconic acid (2- were used for chromosomal deletion.
fold) and lycopene (2-fold) compared to that of conventional cultiva-
tion. The results indicate that the STAPL system is a highly promising 2.4. Plasmid stability assay
strategy for improving the economic feasibility of bioprocessing.
To test the STAPL system, the STG0 – STG5 strains were cultured in
2. Material and methods 15-mL test tubes with 3 mL of modified M9 medium (47.8 mM
Na2HPO4, 22.0 mM KH2PO4, 18.7 mM NH4Cl, 8.6 mM NaCl, 2 mM
2.1. Bacterial strains, plasmids, and reagents MgSO4, and 0.1 mM CaCl2) additionally containing 4 g/L glucose and
1 g/L casamino acids. To initiate the cultures, a single colony from a
The bacterial strains and plasmids are listed in Table S1. Mach-T1R streaked LB agar plate was inoculated into a test tube containing 3 mL
was utilized as a cloning host, and W3110 was used for the STAPL medium. The saturated seed culture was refreshed by re-inoculation at
system. The primers were synthesized by Cosmogenetech (Seoul, Korea) optical density (OD600) 0.05. This refreshed seed was used as a main
and are listed in Table S2. Genomic DNA was isolated using the Ex- culture. Serial sub-cultures were conducted every 5 generations (when
gene™ Cell SV kit (GeneAll, Seoul, Korea). Plasmid DNA was isolated OD600 reached to 1.6 from 0.05) until the 40th generation. The OD600
using the Exprep™ Plasmid SV kit (GeneAll). DNA fragments amplified was monitored using a UV-1700 spectrophotometer (Shimadzu, Kyoto,
by PCR were purified using the Expin™ Gel SV kit (GeneAll). For typical Japan). All cell cultures were conducted with three biological replicates
DNA work, Q5 polymerase and restriction enzymes (New England at 37 °C and continuous shaking (250 rpm). It should be noted that
Biolabs, Ipswich, MA, USA) were used. Reagents for cell culture were Streptomycin (50 µg/mL) and Ampicillin (50 μg/mL) were added to the
purchased from BD Bioscience (Sparks, MD, USA). All other chemicals agar plates and liquid culture for control.
were supplied by Sigma (St. Louis, MO, USA) unless otherwise in- For measurement of the fluorescence of eGFP at OD600 1, a VICTOR3
dicated. 1420 multilabel plate reader (PerkinElmer, Wellesley, MA, USA) was
used with a 486-nm excitation filter and a 535-nm emission filter. A
2.2. Cloning of plasmids Hidex Sense microplate reader (Hidex, Turku, Finland) was used with a
575-nm excitation filter and a 610-nm emission filter for measurement
To construct pCDF-eGFP, egfp was synthesized by GeneArt Strings™ of the fluorescence of mCherry. The fluorescence at the single-cell level
DNA fragments service (Thermo Fisher, Waltham, MA, USA) and am- was measured by an S3e cell sorter (Bio-Rad, Hercules, CA, USA).
plified using O-eGFP-F1, O-eGFP-F2, and O-eGFP-B primers. The pur-
ified DNA fragments were digested using SacI and inserted into 2.5. Quantification of PCN
pCDFDuet. Synthetic infA expression cassettes were prepared by am-
plification using O-infA-F1, O-infA-B, and O-infA-F2 primer variants PCN, defined as the number of plasmids per chromosome, was de-
with the genomic DNA of W3110 strain as a template. These cassettes termined by a previously reported method (Skulj et al., 2008). Cultured
were inserted into the Bsu36I cloning site of pCDF-eGFP, pCDF-CAD samples at OD600 1 (corresponding to 106 cells/μL) or diluted samples
(Noh et al., 2017b), and pCDF_idi_ispA_crtEBI (Jung et al., 2016) plas- (OD600 1) were denatured by heating at 95 °C for 10 min and used as a
mids to yield STAPL plasmids (Table S1). pET-mCherry plasmid was template (105 cells/μL, additional 1/10 dilution) to reduce possible
prepared by insertion of the DNA fragment amplified using O-mCherry- experimental errors during the DNA isolation procedure (Anindyajati

122
C.W. Kang et al. Metabolic Engineering 48 (2018) 121–128

et al., 2016; Skulj et al., 2008). The Q-pCDF-F/Q-pCDF-B, and Q-polQ- overexpressed based on the STAPL system to convert the isoprenoid
F/Q-polQ-B primer pairs were employed to specifically amplify the precursors (isopentenyl pyrophosphate and dimethylallyl pyropho-
plasmid and chromosome, respectively. For quantitative PCR (qPCR), a sphate) into lycopene. The STL0 – 5 strains were cultivated as described
StepOnePlus Real-time PCR system (Applied Biosystems, Foster City, in a previous study (Jung et al., 2016). Briefly, the overnight cultures in
CA, USA) and TOPreal™ qPCR 2 × PreMIX (SYBR Green with high Luria-Bertani (LB) broth were refreshed by inoculation into 2 × M9
ROX) (Enzynomics, Daejeon, Korea) were used for amplification and medium. When OD600 reached 0.9–1.2, the refreshed seed was in-
signal detection. Standard curves were plotted with known copies of oculated at OD600 0.05 in 300-mL flask containing 20 mL medium. The
plasmid and chromosome using the following equation (Whelan et al., cells were cultured aerobically at 37 °C with continuous shaking
2003). (250 rpm). After 24 h, the cells were harvested to quantify produced
lycopene. The culture experiments were conducted in triplicate.
6.02×1023 (copy/mol) × DNA amount(g)
DNA(copy) = For lycopene extraction, the collected cell pellet was washed using
DNA length(dp) × 660 (g/mol/dp) phosphate buffer (pH 7.0) and lysed by addition of 1 mL acetone. The
After quantifying the copies of plasmid and chromosome respec- tubes were incubated at 55 °C for 15 min and intermittently vortexed to
tively, the number of plasmid copies was divided by the number of facilitate extraction. Subsequently, the cell debris was removed by
chromosome copies to obtain the PCN. centrifuge and only supernatants were obtained. For quantification,
absorbance at 475 nm was measured. A standard curve was prepared
based on commercial lycopene. The concentration of lycopene was
2.6. Quantification of relative promoter strengths
normalized to dry cell weight (DCW); one unit of OD600 correlates to
0.252 g DCW/L (Jung et al., 2016).
The relative promoter strengths of STG1–5 variants were de-
termined by the amount of infA transcript divided by the measured
3. Results
PCN. Total RNA from refreshed cells was extracted using the Ribospin™
total RNA purification kit (GeneAll) when OD600 reached 1. Residual
3.1. Stable plasmid maintenance with minimized cell-to-cell variations with
DNA was removed by Riboclear™ (GeneAll). Complementary DNA was
the STAPL system
synthesized using SuperScript III Reverse Transcriptase (Invitrogen,
Carlsbad, CA, USA). The relative amount of infA transcript was quan-
Without any selection pressure, cells begin to lose their plasmids
tified using the comparative CT method (Lee et al., 2006; Livak and
after several rounds of division (Fig. 1a). When antibiotics are added,
Schmittgen, 2001) with hcaT and cysG as reference genes (Brynildsen
cells should harbor plasmids to obtain the resistance to the antibiotic.
et al., 2011). The Q-infA-F and Q-infA-B primers were used to amplify
However, the plasmids can be unevenly distributed due to degradation
infA transcript; the Q-hcaT-F/Q-hcaT-B and Q-cysG-F/Q-cysG-B primer
or the shortage of selection pressure (Nicole and Bentley, 2015; Wright,
pairs were utilized for amplification of hacT and cysG transcripts. The
2005). We expected that the STAPL system, based on auxotrophic
experiments were conducted by following the MIQE guidelines (Sinton
complementation, would enable stable maintenance of the plasmids
et al., 2009) with biological triplicate (Table S3). All nucleic acid
without the addition of antibiotics, because the essential gene should be
concentrations (Table S4) were measured by a UV-1700 spectro-
expressed by the plasmid instead of the chromosome (Fig. 1b). More-
photometer (Shimadzu, Kyoto, Japan).
over, choosing the proper essential gene directly related to cell viability
can tightly control the PCN depending on its expression level and thus
2.7. Itaconic acid production and quantification can minimize cell-to-cell variations during antibiotic-free cultivation.
To develop a model STAPL system (Fig. 1b), we employed a com-
For itaconic acid production, the developed STI0 – STI5 strains were monly used plasmid (pCDFDuet, CloDF13 replicon). Its PCN is known to
grown in M9 medium supplemented with 2 g/L yeast extract, 10 mL/L be 20 − 40 copies/cell measured by gel analysis (Held et al., 2003),
ATCC trace mineral solution, and 20 g/L glucose. Overnight cultured however it was assessed to be 300 copies/cell in this study (Table S5).
cells were re-inoculated into the fresh medium at OD600 0.05. When the We also selected infA, which is involved in the translation process, for
OD of the refreshed seed reached 1.0, 25-mL cultures were started with the STAPL system, because its expression is vital to cell growth and does
OD600 0.05 in a 300-mL flask under aerobic conditions (30 °C, 200 rpm) not cause a cross-feeding effect (Hägg et al., 2004). For infA expression,
for itaconic acid production. When the cells were in the middle of the we constructed an expression cassette with rationally designable ge-
exponential phase (OD600 0.6), 0.01 mM IPTG (isopropyl β-D-1-thio- netic parts consisting of a mid-strength constitutive promoter (J23116,
galactopyranoside) was added to induce gene expression under the Ptac Table S6) from a promoter library (Registry of standard biological
promoter. The pH was adjusted to approximately 7.0 using 10 M NaOH parts), a low-strength 5′-untranslated region (computationally designed
solution every 6 h. by UTR Designer (Seo et al., 2013a, 2013b), see Section 2.2), and a
The concentration of metabolites was determined with an UltiMate synthetic terminator (BBa_B1002 from Registry of standard biological
3000 analytical HPLC system (Dionex, Sunnyvale, CA, USA) equipped parts). The amount of expression was designed to be low because strong
with an Aminex HPX-87H column (Bio-Rad Laboratories, Richmond, plasmid-based expression could potentially induce cellular burden
CA, USA). As a mobile phase, 5 mM sulfuric acid was used at 0.6 mL/ (Vidal et al., 2008), which inhibits cell growth. In addition, we ex-
min flow rate and 14 °C. Glucose and acetate signals were monitored pressed a gene encoding green fluorescent protein (eGFP) under a
through a Shodex RI-101 refractive index detector (Shodex, constitutive promoter from the same plasmid to observe the PCN in situ
Klokkerfaldet, Denmark) and itaconic acid signals were detected with a by measuring fluorescence. The designed plasmid was introduced to E.
UV–vis diode array detector (at 210 nm). All experiments in shake coli W3110; thereafter, the chromosomal infA was deleted for auxo-
flasks were performed with in triplicate. trophic maintenance and the strain was labeled STG1.
We then evaluated the long-term stability of the developed STAPL
2.8. Lycopene production and quantification system compared to that of the conventional case. Serial sub-cultures
were conducted every 5 generations (when OD600 reached to 1.6 from
For lycopene production, 3 heterologous genes (crtE, crtB, and crtI, 0.05) until the 40th generation. The fluorescence of the STG0 strain
encoding geranylgeranyl pyrophosphate synthase, phytoene synthase, (W3110 harboring pCDF-eGFP, Table S1), rapidly decreased and
and phytoene desaturase, respectively) from Lamprocystis purpurea and reached 13.7% of initial fluorescence at the 40th generation when
2 native genes (ispA and idi, encoding farnesyl diphosphate synthase streptomycin (a selection pressure to maintain pCDFDuet) was not
and isopentenyl-diphosphate delta-isomerase, respectively) were added, indicating significant loss of plasmids (Fig. 2a). When

123
C.W. Kang et al. Metabolic Engineering 48 (2018) 121–128

a Conventional plasmid maintenance system

Without antibiotics With antibiotics

Plasmid copy number

High
Plasmid loss Broad PCN

b Stable and TunAble PLasmid (STAPL) system

Chromosome
infA infA

infA infA neo

infA pCDFDuet

infA GOI

infA CloDF13
Promoter strength of infA

High PCN Mid PCN Low PCN

Fig. 1. An overview of the STAPL system. a, Conventional plasmid maintenance system. For plasmid maintenance, antibiotics are added to prevent loss of plasmids.
However, even with antibiotics, population heterogeneity can occur because of the difficulty in maintaining tight control. b, A schematic representation of the STAPL
system. In the STAPL system, chromosomal infA is deleted and expressed on a plasmid via a synthetic expression cassette. With the STAPL system, plasmids can be
stably maintained in cells with tunable copy numbers by controlled auxotrophic selection. Moreover, the STAPL system also minimizes cell-to-cell variations; these
properties can significantly improve the economic feasibility of industrial fermentation. Green circle: cell harboring plasmid; intensity of green color corresponds to
the PCN.

streptomycin was added, the fluorescence level decreased to 65% at the maintenance. However, the STAPL system showed a homogenous po-
40th generation; this result indicates that addition of antibiotics is ef- pulation (CV = 80.2) without any minor populations (Fig. 2b, STG1).
fective but cannot guarantee the stable maintenance of plasmids. On These results clearly indicate that the STAPL system with synthetic
the other hand, the fluorescence of the STG1 strain (W3110 ΔinfA strain expression control of infA is highly effective in tightly maintaining the
harboring pSTAPL_eGFPv1, Table S1) with the STAPL system was sur- plasmids and minimizing the cell-to-cell variations.
prisingly consistent over the entire 40th generation; 97.2% of fluores-
cence was maintained without streptomycin. This result shows that 3.2. Modulation of the PCN with varied infA expression in the STAPL
plasmids are well maintained by the STAPL system without antibiotics. system
Plasmid maintenance system without tight control can inherently
induce cell-to-cell variations (Popov et al., 2014; Xiao et al., 2016); we PCN control is often obtained by random mutagenesis of the
measured the fluorescence at the single-cell level using flow cytometry plasmid origin or its control region (Tao et al., 2005); however, rational
for detailed population analysis in each system. We observed severe loss control of the PCN is virtually impossible with this strategy. Alter-
of plasmids during serial sub-cultures without addition of streptomycin, natively, another strategy, one of varying the concentration of anti-
resulting in heterogeneous populations (coefficient of variance (CV) biotics, was applied in Saccharomyces cerevisiae (Lian et al., 2016). Al-
= 327), where 76.4% of the total cell fraction was plasmid-free at the though this strategy demonstrated that PCN could be tuned depending
40th generation (Fig. 2b, STG0 w/o Sm). Due to this heterogeneity, the on the level of antibiotic concentration, this is impractical because of
overall specific fluorescence was gradually reduced. With the addition the problems arising from the use of antibiotics as well as high cost.
of streptomycin, there also existed a minor population consisting of In the STAPL system, we hypothesized that changing transcription
28.5% (at the 20th generation) and 45.1% (at the 40th generation) of levels of infA would affect PCN (Fig. 1b). When the transcription level
total cells showing lower fluorescence (Fig. 2b, STG0 w/ Sm). This of infA is lowered, the cells with high PCN should dominate by com-
emergence of minor population can be explained by “downward se- pensating for the reduced expression of a gene essential to growth. On
lective pressure” mechanism (Bentley and Quiroga, 1993; Nicole and the other hand, low-PCN cells are favored when the transcription level
Bentley, 2015). The cells should maintain the plasmid but favor low- of infA is high because of the low metabolic cost as described above
ering their PCN due to the cost of metabolic burden during plasmid (Nicole and Bentley, 2015). To test this concept, we varied the strength

124
C.W. Kang et al. Metabolic Engineering 48 (2018) 121–128

Fig. 2. Stable plasmid maintenance of the STAPL system. a, Long-term plasmid stability of the STAPL system. The y-axis and x-axis represent relative plasmid
maintenance ratio and generation, respectively. The ratio was calculated by dividing the measured fluorescence by initial fluorescence (at 0 generation) of each
strain. Sm: streptomycin; Light gray circle: without streptomycin; dark gray rectangle: with streptomycin; red triangle: the STAPL system. Error bars represent the
standard deviations of biological triplicates (n = 3). b, Fluorescence at the single-cell level analyzed by flow cytometry during the 40th generation. A total of 100,000
cells were analyzed in each case. The y-axis and x-axis represent cell counts and fluorescence signals, respectively. The cells in the range of 0 – 101 were regarded as
plasmid-free cells.

of promoters for infA expression and investigated the effect on PCN; values of STG0 strain at the 40th generation, insisting that STAPL is
additional 4 different infA variants (STG2–5 strains, Table S1) were effective to minimize cell-to-cell variation.
constructed using other constitutive promoters (J23112, J23109, In addition to the fluorescence, PCN and infA expression were also
J23118, and J23100) which cover the full range of promoter strength in highly correlated, as we hypothesized (Fig. 3b and Table S5); the PCN
the library (Table S6). could be varied by more than 5.6-fold (from 46 to 261 copies per cell)
After similar long-term cultivation, we investigated the correlation and maintained without changing the plasmid origin. We measured
between infA promoter strength and fluorescence of STG1–5 strains. For growth rates in the STAPL system to determine the possible inhibitory
this analysis, the strengths of infA promoter were re-measured to con- effect of changing infA transcription levels, but we were not able to
firm the reported strength (Table S6). Although the strength of the observe significant metabolic burden in terms of growth rates (Fig. S3);
J23116 promoter was much lower than reported, our observations of all variants showed similar growth rates. Collectively, changing the
the others were generally consistent with their reported strengths. transcription levels of infA enabled successful modulation of PCNs by
Under this re-evaluation, the fluorescence at the 40th generation was more than 5.6-fold in a controllable manner.
inversely proportional to infA transcription level (Fig. 3a). Moreover,
this trend was consistently maintained during the entire cultivation
period (Fig. S1). Single-cell level analysis was also conducted for each 3.3. Applications of the STAPL system for the production of biochemicals
variant (Fig. S2). Although a minor population was observed for STG4
strain, most strains showed homogeneous populations during entire For the production of biochemicals, related metabolic genes are
cultivation period. All CV values of STAPL strains were lower than the usually overexpressed in order to drive carbon flux toward product
synthesis by using strong promoters and ribosome binding sites (Seo

Fig. 3. Tunable plasmid copy number (PCN) of

the STAPL system. a, Relative specific fluores-
cence of STG0 – 5 strains at the 40th genera-
tion. The values were normalized to that of the
STG5 strain. b, The correlation between the
PCNs of the STG0 – 5 strains (red) and relative
promoter strength (dark gray). The values were
normalized to the lowest value in each experi-
ment. a and b, Error bars represent the standard
deviations of biological triplicates (n = 3).

125
C.W. Kang et al. Metabolic Engineering 48 (2018) 121–128

a ...
b 0.7
c 1000
Glucose G3P PYR AcCoA
0.6

Itaconic acid titer (g/L)

CIT 800
OAA
0.5
Chromosome
CCA
600
SUC 0.4

PCN
infA cad
0.3 400
neo Itaconic acid
0.2
200
0.1
Plasmid for itaconic acid production
0.0 0
PJ23100

I0

I4
I0

I1
I2
I3

I5
I0

I0
I1
I2
I3
I4
I5

ST

ST
ST

ST
ST
ST

ST
ST
ST

ST
ST
ST
ST
ST
cad Sm - + - - - - -
Sm - + - - - - -
Promoter library
CloDF13 infA promoter strength infA promoter strength
infA

d Glucose G3P ... PYR AcCoA

e f
30

Specific lycopene (mg/g DCW)

800
...

CIT 25
OAA
HMBPP
CCA
idi 20 600
IPP SUC
DMAPP

PCN
Chromosome
ispA 15
infA 400
crtEBI
10
Lycopene
neo
200
5
Plasmid for lycopene production
PJ23100 0 0

ispA L0 L0 L1 L2 L3 L4 L5 L0 L0 L1 L2 L3 L4 L5
idi crtE crtB crtI ST ST ST ST ST ST ST ST ST ST ST ST ST ST
Sm - + - - - - - Sm - + - - - - -
Promoter library CloDF13
infA infA promoter strength infA promoter strength
Fig. 4. Application of the STAPL system in itaconic acid and lycopene production. a, Metabolic pathway for itaconic acid production. Itaconic acid is produced by cis-
aconitate decarboxylase (encoded by cad) from TCA cycle. Abbreviations: G3P, glyceraldehyde 3-phosphate; PYR, pyruvate; CIT, citrate; CCA, cis-aconitate; SUC,
succinate; OAA, oxaloacetate; infA, gene encoding translation initiation factor-I; cad, gene encoding cis-aconitate decarboxylase. b, Itaconic acid production of the
STI0 – 5 strains after 36 h. The y-axis represents itaconic acid production (g/L). c, The PCN of the STI0–5 strains measured after 36-h cultivation. d, Metabolic
pathway for lycopene production. Lycopene is produced by serial enzymatic reactions via the methyl-D-erythritol 4-phosphate (MEP) pathway with condensation of
two intermediates, DMAPP and IPP. Abbreviations: HMBPP, 4-hydroxy-3-methyl-but-2-enyl pyrophosphate; DMAPP, dimethylallyl pyrophosphate; IPP, isopentenyl-
diphosphate; idi, gene encoding isopentenyl-diphosphate delta-isomerase; ispA, gene encoding farnesyl diphosphate (FPP) synthase; crtE, gene encoding ger-
anylgeranyl pyrophosphate (GGPP) synthase; crtB, gene encoding phytoene synthase; crtI, gene encoding phytoene desaturase. e, The specific lycopene production of
the STL0 – 5 strains after 24 h. The y-axis represents the specific lycopene production (mg/g DCW). f, The PCN of the STL0 – 5 strains measured at OD600 0.8. b, c, e,
and f, Error bars represent the standard deviations of biological triplicate experiments (n = 3).

et al., 2013a, 2013b). Additionally, copy number of plasmid, a struc- cad from Aspergillus terreus (Harder et al., 2016) to produce itaconic
tural basis for the expression cassettes, is also considered to be im- acid in E. coli, and we cultivated the developed strains (STI0–5). The
portant; plasmids with different copy numbers are utilized for the op- STI0 strain with plasmid maintenance by addition of streptomycin
timization of metabolic activity of cells (Ajikumar et al., 2010). Thus, produced 0.27 g/L of itaconic acid during 36-h fermentation (Fig. 4b).
modulation of PCN can be an effective strategy for maximizing bio- Without addition of streptomycin, a smaller amount (0.04 g/L) of ita-
chemical production. Furthermore, a previously proposed concept for conic acid was produced, indicating reduced biosynthesis due to a loss
population quality control (Xiao et al., 2016) suggested that reduction of plasmids during cultivation. On the other hand, the STI4 strain
of heterogeneity arising from many factors (Zenobi, 2013) could en- showed significantly higher itaconic acid production (0.56 g/L) which
hance the production of biochemicals. As the STAPL system enables is a 2-fold higher than that of the STI0 strain. In this case, too much
modulation of PCN as well as quality control of populations by mini- expression of cad was not beneficial in the production of itaconic acid,
mizing cell-to-cell variations, we tested a biochemical production because the enzyme mediates conversion of a TCA intermediate; a low
pathway using the STAPL system to investigate the potential increase in PCN (Fig. 4c) could enhance itaconic acid production. In this case, PCN
cellular performance. at 36 h was ranged from 184 to 769 copies per cell (more than 4.2-fold,
To this end, we first applied the STAPL system to production of Table S5); the titer (0.56 g/L) and productivity (0.02 g/L/h) of the
itaconic acid, a promising platform chemical (Okabe et al., 2009). Ita- STAPL system were superior to those of the systems (0.47 g/L and
conic acid can be biologically synthesized through decarboxylation of 0.01 g/L/h) with expression of single cad in a previous study (Jeon
cis-aconitic acid by cis-aconitic acid decarboxylase (encoded by cad) et al., 2016). Ultimately, these results indicated that the STAPL system
(Fig. 4a). Thus, we engineered a plasmid to overexpress heterologous successfully maintained the plasmids and could be applied to find

126
C.W. Kang et al. Metabolic Engineering 48 (2018) 121–128

optimal expression of target genes and maximize biochemical produc- that the STAPL system prevents the loss of plasmids, as the marker gene
tion. is essential to cell survival. In addition, the lower expression of infA
We further applied the STAPL system to lycopene production, which guaranteed tight PCN control with minimized cell-to-cell variations.
requires expression of a larger number of genes for its synthesis. We This can be adapted for quality control of PCN in cells for improved
overexpressed 3 heterologous genes (crtE, crtB, and crtI, encoding ger- biochemical production (Xiao et al., 2016).
anylgeranyl pyrophosphate synthase, phytoene synthase, and phytoene Multiple plasmids could also be maintained through the STAPL
desaturase, respectively) from Lamprocystis purpurea and 2 native genes system. In addition to infA gene, other essential genes directly related to
(ispA and idi, encoding farnesyl diphosphate synthase and isopentenyl- cell viability (e.g., rpsL encoding ribosomal protein, fusA encoding
diphosphate delta-isomerase, respectively) based on the STAPL system elongation factor G) can be potential candidates for markers. As an
to convert the isoprenoid precursors (isopentenyl pyrophosphate and example, we constructed pSTAPL-mCherry based on pETduet plasmid
dimethylallyl pyrophosphate) into lycopene (Fig. 4d) (Jung et al., (pMB1 origin, medium copy) by using efp encoding Elongation Factor P
2016). After 24-h cultivation, we observed that specific lycopene pro- as an auxotrophic marker gene. When we expressed efp under J23116
duction was highly correlated to the promoter strength of infA (Fig. 4e). promoter in STG1 strain, we observed higher maintenance ratios for
The STL1 strain with the lowest expression of infA exhibited 2-fold pSTAPL-eGFP (Fig. S4a) and pSTAPL-mCherry (Fig. S4b) plasmids than
enhanced lycopene production (27.1 mg/g DCW) compared to that of those of STGC0 strain. Interestingly, mCherry signal (indicating PCN of
the STL0 (13.6 mg/g DCW) strain, indicating increased carbon flux pSTAPL-mCherry) was continuously increased until the 40th generation
toward lycopene synthesis. PCNs were ranged from 38 to 684 copies per while eGFP signal (indicating PCN of pSTAPL-eGFP) was maintained at
cell (18-fold, Fig. 4f, and Table S5). Contrary to itaconic acid produc- a similar level of initial PCN. The reason is currently unclear; probably,
tion, increased expression of biosynthetic genes via high PCN was more it requires further optimization for controlled maintenance. Multiple
effective for lycopene synthesis. The overall activity of the pathway was plasmid maintenance will be helpful in expressing larger pathways, and
easily tunable by changing the expression level of single gene, infA. The their modulation of PCNs will allow efficient balancing of multigene
production of lycopene during 24 h by the STL1 strain (27.1 mg/g pathways in antibiotic-free conditions. Furthermore, this modulation
DCW) was much improved when compared to 6.20 mg/g DCW reported can be linked to genetic circuits for autonomous control and would
in a previous study (Jung et al., 2016) with the same plasmid. enable the expression of target plasmids at specific timings in bio-
Collectively, our results clearly indicated that addition of antibiotics chemical production (Gupta et al., 2017).
is not required to maintain the plasmids expressing many genes via the
STAPL system; the STAPL system with its PCN control can effectively Author contributions
drive carbon flux toward biochemical production as well as provide
population quality control. The production can be further synergisti- C.W.K. and H.G.L. designed the project and performed experiments
cally improved when the system is combined with the genome en- together with J.Y. and M.H.N. C.W.K., H.G.L., S.W.S., and G.Y.J. con-
gineering and/or optimization process conducted in previous studies ducted data analysis and interpretation and wrote the manuscript.
(Jung et al., 2016; Jeon et al., 2016; Rad et al., 2012). S.W.S. and G.Y.J. supervised the project. All authors read and approved
the final manuscript.
4. Discussion
Competing Interests
For the production of high-value-added chemicals, stable and tun-
able expression of the gene(s) in hosts is essential for high production. The authors declare that they have no competing interests.
The plasmid is definitely an efficient tool for expression of desired
genes; however, its stability is vulnerable when antibiotics are not Acknowledgments
added. As an alternative strategy, recombinant DNA can be integrated
into the host chromosome. However, chromosome-based expression is This research was supported by grants from C1 Gas Refinery
significantly lower than plasmid-based expression due to lowered copy Program [NRF-2018M3D3A1A01055754], Global Research Laboratory
number (from 10–500 to 1). Although a recent approach enabled the Program [NRF-2016K1A1A2912829], and Fostering Core Leaders of
increase of copy number to up to 40 consecutive copies in the genome the Future Basic Science Program/Global Ph.D. Fellowship Program
(Tyo et al., 2009), difficulties, such as disruption of the gene repair [NRF-2017H1A2A1044539] funded by the Ministry of Science and ICT.
system and fluctuation of gene expression depending on integration
locus on the genome, (Sousa et al., 1997) limit the ease of this en- Appendix A. Supplementary material
gineering. With the STAPL system, plasmids can be stably maintained
without significant alterations to the native system. Supplementary data associated with this article can be found in the
Furthermore, the STAPL system also enabled copy number control online version at http://dx.doi.org/10.1016/j.ymben.2018.05.020.
via quantitative expression control of an auxotrophic marker gene.
With the STAPL system, we demonstrated that a wide range of PCNs References
(40 − 260 copies per cell for gfp expression, 184 − 769 copies per cell
for cad expression, 30 − 680 copies per cell for crtEBI expression, Table Adams, C.W., Hatfield, G.W., 1984. Effects of promoter strengths and growth conditions
S5) can be obtainable without changing the plasmid origin. Although on copy number of transcription-fusion vectors. J. Biol. Chem. 259, 7399–7403.
Ajikumar, P.K., Xiao, W.-H., Tyo, K.E.J., Wang, Y., Simeon, F., Leonard, E., Mucha, O.,
the actual PCNs and fold-changes varied probably due to the susceptible Phon, T.H., Pfeifer, B., Stephanopoulos, G., 2010. Isoprenoid pathway optimization
nature of plasmid to expressed genes, growth rate, culture condition, for taxol precursor overproduction in Escherichia coli. Science 330, 70–74. http://dx.
medium composition (Adams and Hatfield, 1984; Corchero and doi.org/10.1126/science.1191652.
Anindyajati, Anita Artarini, A., Riani, C., Retnoningrum, D.S., 2016. Plasmid copy
Villaverde, 1998; Lian et al., 2016; Prather, 2010; Stueber and Bujard, number determination by quantitative polymerase chain reaction. Sci. Pharm. 84,
1982), the trends of PCN with changed infA expression were highly 89–101. http://dx.doi.org/10.3797/scipharm.ISP.2015.02.
consistent. More detailed expression control via synthetic biology-based Bentley, W.E., Quiroga, O.E., 1993. Investigation of subpopulation heterogeneity and
plasmid stability in recombinant Escherichia coli via a simple segregated model.
genetic tools (Seo et al., 2013a, 2013b) will allow more programmable Biotechnol. Bioeng. 42, 222–234. http://dx.doi.org/10.1002/bit.260420210.
and diversified PCNs. Brophy, J.A.N., Voigt, C.A., 2014. Principles of genetic circuit design. Nat. Methods 11,
The STAPL system is also effective in minimizing the heterogeneity 508–520. http://dx.doi.org/10.1038/nmeth.2926.
Brynildsen, M.P., Liao, J.C., Lim, Q.’En, Zou, R., Stephanopoulos, G., Too, H.-P.,
of populations, which has been a drawback in plasmid-based expression
Brynildsen, M., Liao, J., Baez-Viveros, J., Flores, N., et al., 2011. Novel reference
(Magdanova and Golyasnaya, 2013; Xiao et al., 2016). We observed

127
C.W. Kang et al. Metabolic Engineering 48 (2018) 121–128

genes for quantifying transcriptional responses of Escherichia coli to protein over- microorganisms: recent advances in synthetic dynamic regulators. J. Ind. Microbiol.
expression by quantitative PCR. Mol. Biol. 12, 1–9. http://dx.doi.org/10.1038/msb. Biotechnol. 44, 89–98. http://dx.doi.org/10.1007/s10295-016-1867-y.
2009.34. Nicole, B., Bentley, W., 2015. Systems and technology involving bacteria. In: Bronzino,
Chen, R., 2012. Bacterial expression systems for recombinant protein production: E. coli J.D., Peterson, D.R. (Eds.), Molecular, cellular, and tissue engineering. CRC Press,
and beyond. Biotechnol. Adv. 30, 1102–1107. http://dx.doi.org/10.1016/j. Boca Raton, pp. 4–1-4-2.
biotechadv.2011.09.013. Noh, M.H., Lim, H.G., Park, S., Seo, S.W., Jung, G.Y., 2017a. Precise flux redistribution to
Chen, Y., Partow, S., Scalcinati, G., Siewers, V., Nielsen, J., 2012. Enhancing the copy glyoxylate cycle for 5-aminolevulinic acid production in Escherichia coli. Metab. Eng.
number of episomal plasmids in Saccharomyces cerevisiae for improved protein pro- 43, 1–8. http://dx.doi.org/10.1016/j.ymben.2017.07.006.
duction. FEMS Yeast Res. 12, 598–607. http://dx.doi.org/10.1111/j.1567-1364. Noh, M.H., Lim, H.G., Woo, S.H., Song, J., Jung, G.Y., 2017b. Production of itaconic acid
2012.00809.x. from acetate by engineering acid-tolerant Escherichia coli W. Biotechnol. Bioeng.
Corchero, J.L., Villaverde, A., 1998. Plasmid maintenance in Escherichia coli recombinant 729–738. http://dx.doi.org/10.1002/bit.26508.
cultures is dramatically, steadily, and specifically influenced by features of the en- Okabe, M., Lies, D., Kanamasa, S., Park, E.Y., 2009. Biotechnological production of ita-
coded proteins. Biotechnol. Bioeng. 58, 625–632. http://dx.doi.org/10.1002/(SICI) conic acid and its biosynthesis in Aspergillus terreus. Appl. Microbiol. Biotechnol. 84,
1097-0290(19980620)58:6<625::AID-BIT8>3.0.CO;2-K. 597–606. http://dx.doi.org/10.1007/S00253-009-2132-3.
Cranenburgh, R.M., Hanak, J.A., Williams, S.G., Sherratt, D.J., 2001. Escherichia coli Peubez, I., Chaudet, N., Mignon, C., Hild, G., Husson, S., Courtois, V., De Luca, K., Speck,
strains that allow antibiotic-free plasmid selection and maintenance by repressor ti- D., Sodoyer, R., 2010. Antibiotic-free selection in E. coli: new considerations for
tration. Nucleic Acids Res 29, E26. http://dx.doi.org/10.1093/nar/29.5.e26. optimal design and improved production. Microb. Cell Fact. 9, 65. http://dx.doi.org/
Datsenko, K.A., Wanner, B.L., 2000. One-step inactivation of chromosomal genes in 10.1186/1475-2859-9-65.
Escherichia coli K-12 using PCR products. Proc. Natl. Acad. Sci. USA 97, 6640–6645. Popov, M., Petrov, S., Kirilov, K., Nacheva, G., Ivanov, I., 2014. Segregational instability
http://dx.doi.org/10.1073/pnas.120163297. in E. coli of expression plasmids carrying human Interferon gamma gene and its 3′-
Davison, J., 1999. Genetic exchange between bacteria in the environment. Plasmid 42, end truncated variants. Biotechnol. Biotechnol. Equip. 23, 840–843. http://dx.doi.
73–91. http://dx.doi.org/10.1006/plas.1999.1421/rS0147-619X(99)91421. org/10.1080/13102818.2009.10818553.
Degryse, E., 1991. Stability of a host-vector system based on complementation of an es- Prather, K.L.J., 2010. Low-copy number plasmids as artificial chromosome. In: Smolke, C.
sential gene in Escherichia coli. J. Biotechnol. 18, 29–39. http://dx.doi.org/10.1016/ D.(Eds.), The metabolic pathway engineering handbook. Tools and applications. CRC
0168-1656(91)90233-L. Press, Boca Raton, pp. 6–3.
Dong, W.R., Xiang, L.X., Shao, J.Z., 2010. Novel antibiotic-free plasmid selection system Rad, S.A., Zahiri, H.S., Noghabi, K.A., Rajaei, S., Heidari, R., Mojallali, L., 2012. Type 2
based on complementation of host auxotrophy in the NAD de novo synthesis IDI performs better than type 1 for improving lycopene production in metabolically
pathway. Appl. Environ. Microbiol. 76, 2295–2303. http://dx.doi.org/10.1128/AEM. engineered E. coli strains. World J. Microbiol. Biotechnol. 28, 313–321. http://dx.doi.
02462-09. org/10.1007/s11274-011-0821-4.
Gupta, A., Reizman, I.M.B., Reisch, C.R., Prather, K.L.J., 2017. Dynamic regulation of Seo, S.W., Yang, J., Min, B.E., Jang, S., Lim, J.H., Lim, H.G., Kim, S.C., Kim, S.Y., Jeong,
metabolic flux in engineered bacteria using a pathway-independent quorum-sensing J.H., Jung, G.Y., 2013a. Synthetic biology: tools to design microbes for the produc-
circuit. Nat. Biotechnol. 35, 273–279. http://dx.doi.org/10.1038/nbt.3796. tion of chemicals and fuels. Biotechnol. Adv. 31, 811–817. http://dx.doi.org/10.
Hägg, P., De Pohl, J.W., Abdulkarim, F., Isaksson, L.A., 2004. A host/plasmid system that 1016/j.biotechadv.2013.03.012.
is not dependent on antibiotics and antibiotic resistance genes for stable plasmid Seo, S.W., Yang, J.S., Kim, I., Yang, J., Min, B.E., Kim, S., Jung, G.Y., 2013b. Predictive
maintenance in Escherichia coli. J. Biotechnol. 111, 17–30. http://dx.doi.org/10. design of mRNA translation initiation region to control prokaryotic translation effi-
1016/j.jbiotec.2004.03.010. ciency. Metab. Eng. 15, 67–74. http://dx.doi.org/10.1016/j.ymben.2012.10.006.
Hanak, J.A.J., Cranenburgh, R.M., 2001. Antibiotic-free plasmid selection and main- Sinton, L.W., Finlay, R.K., Lynch, P.A., 2009. The MIQE guidelines: minimum information
tenance in bacteria. In: Merten, O.-W., Mattanovich, D., Lang, C., Larsson, G., for publication of quantitative real-time PCR experiments. Clin. Chem. 55, 1–12.
Neubauer, P., Porro, D., Postma, P., Teixeirade Mattos, J., Cole, J.A. (Eds.), http://dx.doi.org/10.1373/clinchem.2008.112797.
Recombinant protein production with prokaryotic and eukaryotic cells. A Skulj, M., Okrslar, V., Jalen, S., Jevsevar, S., Slanc, P., Strukelj, B., Menart, V., 2008.
Comparative View on Host Physiology. E-publishing Inc., New York, pp. 111–124. Improved determination of plasmid copy number using quantitative real-time PCR
http://dx.doi.org/10.1007/978-94-015-9749-4_9. for monitoring fermentation processes. Microb. Cell Fact. 7, 6. http://dx.doi.org/10.
Harder, B.-J., Bettenbrock, K., Klamt, S., 2016. Model-based metabolic engineering en- 1186/1475-2859-7-6.
ables high yield itaconic acid production by Escherichia coli. Metab. Eng. 38, 29–37. Sodoyer, R., Courtois, V., Peubez, I., Mignon, C., 2012. Antibiotic-free selection for bio-
http://dx.doi.org/10.1016/j.ymben.2016.05.008. production: moving towards a new "Gold standard". In: Pana, M. (Ed.), Antibiotic
Held, D., Yaeger, K., Novy, R., 2003. New coexpression vectors for expanded compat- resistant bacteria-A continuous challenge in the new millennium. CC BY, Croatia, pp.
ibilities in E. coli. Innovations 18, 4–6. 531–548.
Jeon, H.G., Cheong, D.E., Han, Y., Song, J.J., Choi, J.H., 2016. Itaconic acid production Sousa, C., Lorenzo, V., De, Cebollat, A., 1997. Modulation of gene expression through
from glycerol using Escherichia coli harboring a random synonymous codon-sub- chromosomal positioning in Escherichia coli. Microbiology 143, 2071–2078.
stituted 5′-coding region variant of the cadA gene. Biotechnol. Bioeng. 113, Stueber, D., Bujard, H., 1982. Transcription from efficient promoters can interfere with
1504–1510. http://dx.doi.org/10.1002/bit.25914. plasmid replication and diminish expression of plasmid specified genes. EMBO J. 1,
Jung, G., Denèfle, P., Becquart, J., Mayaux, J.-F., 1988. High-cell density fermentation 1399–1404.
studies of recombinant Escherichia coli strains expressing human interleukin-1 beta. Szpirer, C.Y., Milinkovitch, M.C., 2005. Separate-component-stabilization system for
Ann. Inst. Pasteur Microbiol. 139, 129–146. http://dx.doi.org/10.1016/0769- protein and DNA production without the use of antibiotics. Biotechniques 38,
2609(88)90100-7. 775–781. http://dx.doi.org/10.2144/05385RR02.
Jung, J., Lim, J.H., Kim, S.Y., Im, D.K., Seok, J.Y., Lee, S.J.V., Oh, M.K., Jung, G.Y., 2016. Tao, L., Jackson, R.E., Cheng, Q., 2005. Directed evolution of copy number of a broad
Precise precursor rebalancing for isoprenoids production by fine control of gapA host range plasmid for metabolic engineering. Metab. Eng. 7, 10–17. http://dx.doi.
expression in Escherichia coli. Metab. Eng. 38, 401–408. http://dx.doi.org/10.1016/j. org/10.1016/j.ymben.2004.05.006.
ymben.2016.10.003. Taylor, D., Gibreel, A., Tracz, D., Lawley, T., 2004. Plasmid biology. Plasmid Biol. 35,
Lee, C., Kim, J., Shin, S.G., Hwang, S., 2006. Absolute and relative QPCR quantification of 473–492. http://dx.doi.org/10.1128/9781555817732.
plasmid copy number in Escherichia coli. J. Biotechnol. 123, 273–280. http://dx.doi. Tyo, K.E.J., Ajikumar, P.K., Stephanopoulos, G., 2009. Stabilized gene duplication en-
org/10.1016/j.jbiotec.2005.11.014. ables long-term selection-free heterologous pathway expression. Nat. Biotechnol. 27,
Lenski, R.E., Bouma, J.E., 1987. Effects of segregation and selection on instability of 760–765. http://dx.doi.org/10.1038/nbt.1555.
plasmid pACYC184 in Escherichia coli B. J. Bacteriol. 169, 5314–5316. Vandermeulen, G., Marie, C., Scherman, D., Préat, V., 2011. New generation of plasmid
Lian, J., Jin, R., Zhao, H., 2016. Construction of plasmids with tunable copy numbers in backbones devoid of antibiotic resistance marker for gene therapy trials. Mol. Ther.
Saccharomyces cerevisiae and their applications in pathway optimization and multi- 19, 1942–1949. http://dx.doi.org/10.1038/mt.2011.182.
plex genome integration. Biotechnol. Bioeng. 113, 2462–2473. http://dx.doi.org/10. Vidal, L., Pinsach, J., Striedner, G., Caminal, G., Ferrer, P., 2008. Development of an
1002/bit.26004. antibiotic-free plasmid selection system based on glycine auxotrophy for recombinant
Lim, H.G., Noh, M.H., Jeong, J.H., Park, S., Jung, G.Y., 2016. Optimum rebalancing of the protein overproduction in Escherichia coli. J. Biotechnol. 134, 127–136. http://dx.doi.
3-hydroxypropionic acid production pathway from glycerol in Escherichia coli. ACS org/10.1016/j.jbiotec.2008.01.011.
Synth. Biol. 5, 1247–1255. http://dx.doi.org/10.1021/acssynbio.5b00303. Whelan, J.A., Russell, N.B., Whelan, M.A., 2003. A method for the absolute quantification
Livak, K.J., Schmittgen, T.D., 2001. Analysis of relative gene expression data using real- of cDNA using real-time PCR. J. Immunol. Methods 278, 261–269. http://dx.doi.org/
time quantitative PCR and the 2-△△CT method. Methods 25, 402–408. http://dx.doi. 10.1016/S0022-1759(03)00223-0.
org/10.1006/meth.2001.1262. Wright, G.D., 2005. Bacterial resistance to antibiotics: enzymatic degradation and mod-
Magdanova, L.A., Golyasnaya, N.V., 2013. Heterogeneity as an adaptive trait of microbial ification. Adv. Drug Deliv. Rev. 57, 1451–1470. http://dx.doi.org/10.1016/j.addr.
populations. Microbiology 82, 1–10. http://dx.doi.org/10.1134/ 2005.04.002.
S0026261713010074. Xiao, Y., Bowen, C.H., Liu, D., Zhang, F., 2016. Exploiting non-genetic, cell-to-cell var-
Mairhofer, J., Cserjan-Puschmann, M., Striedner, G., Nöbauer, K., Razzazi-Fazeli, E., iation for enhanced biosynthesis. Nat. Chem. Biol. 12, 339–344. http://dx.doi.org/
Grabherr, R., 2010. Marker-free plasmids for gene therapeutic applications-Lack of 10.1038/nchembio.2046.
antibiotic resistance gene substantially improves the manufacturing process. J. Xu, P., Gu, Q., Wang, W., Wong, L., Bower, A.G.W., Collins, C.H., Koffas, M.A.G., 2013.
Biotechnol. 146, 130–137. http://dx.doi.org/10.1016/j.jbiotec.2010.01.025. Modular optimization of multi-gene pathways for fatty acids production in E. coli.
Mignon, C., Sodoyer, R., Werle, B., 2015. Antibiotic-free selection in biotherapeutics: now Nat. Commun. 4, 1409. http://dx.doi.org/10.1038/ncomms2425.
and forever. Pathogens 4, 157–181. http://dx.doi.org/10.3390/pathogens4020157. Zenobi, R., 2013. Single-cell metabolomics: analytical and biological perspectives.
Min, B.E., Hwang, H.G., Lim, H.G., Jung, G.Y., 2017. Optimization of industrial Science 342, 1243259. http://dx.doi.org/10.1126/science.1243259.

128