Introduction
Erythromycin is an important broad-spectrum antibiotic of macrolides
(Oliynyk et al., 2007) and Saccharopolyspora erythraea is the
major industrial erythromycin-producing strain (Butler, 2008). With the
accumulation of knowledge on erythromycin biosynthesis genes and
biochemical pathways, rational strain design has become a powerful
strategy to improve the titer and productivity of erythromycin and the
proportion of active component (Chen et al., 2008; Li et al., 2013; Wu
et al., 2011; Liu et al., 2017). The biosynthesis of erythromycin is
mainly divided into two parts. One propionyl-CoA and six
methylmalonyl-CoA are firstly condensed to 6-deoxyerythromycinB (6-dEB),
then 6-dEB is post-modified to erythromycin A (Mironov et al., 2004).
Wang et al. (2007) integrated the exogenous S-adenosylmethionine (SAM)
synthase gene into the S. erythraea genome to promote the
post-modification of 6-deoxyerythronolideb, which ultimately increased
the yield of erythromycin A by 132%. Chen and Liu et al. (2008)
succeeded in reducing the contents of the impurity components of
erythromycin B and C in the fermentation broth by regulating the copy
number of the genes (eryG and eryK ) encoding the
erythromycin post-modification enzymes.
Propanol has been widely used as a precursor of propionyl-CoA for
erythromycin synthesis (Chen et al. 2013). And it has been well known
that the addition of propionic acid or n-propanol significantly
stimulates the synthesis of erythromycin. Bojanowsa et al. (1973)
firstly found that propionate kinase activity was positively correlated
with the rate of erythromycin synthesis. Recently, Hong Ming et al.
(2016) pointed out that the change of intracellular propionyl-CoA
concentration presented a direct positive correlation with the
erythromycin productivity. On the other hand, Chen et al. (2013)
demonstrated that the addition of n-propanol caused an obvious drop of
dissolved oxygen, indicating that n-propanol might be utilized as a
carbon source apart from glucose, however, it is considered to be not
cost-effective for industrial fermentation process as the relatively
high price of commercial n-propanol.
13C isotope labeling experiment is a quantitative
method enabling to directly study the effects of environmental or
genetic changes on cell metabolism, thereby providing scientific
guidance for further metabolic engineering and fermentation process
regulation (Becker and Wittmann, 2012). Hong Ming et al. (2017) carried
out isotope labeling experiments with [U-13C]
glucose and naturally-labelled proline in S. erythraea and showed
that proline was catabolized into TCA cycle to provide energy and NADPH.
In addition, proline could also increase the energy load of the cells,
thereby reducing the cellular metabolism burden.
In this study, on the basis of carbon atom transition map of propanol
catabolism derived from KEGG database, two possible propanol catabolism
pathways in S. erythraea were identified by labeling experiments
using 13C-labeled sodium propionate as the substrate.
Subsequently, a novel strain was constructed by knocking-out the genes
involved in the unexpected uneconomic propanol catabolism pathway. And
the effects of gene modification was investigated by integration of
isotopic dilution mass spectrometry (IDMS) metabolomics and isotopically
non-steady 13C metabolic flux analysis
(13C-MFA). This study demonstrated how13C labeling experiments can be applied to identify
unknown metabolic pathways and to rationally explore the potential
targets of gene engineering of S. erythraea , thus achieving a
”design-build-validate-guidance” metabolic engineering cycle (Fig. 1).
Materials and methods
Strains and culture conditions
S. erythraea E3 was used as the parent strain. Other strains and
plasmids used were listed in the supplementary materials.
Spores were inoculated to 500 mL shake flasks at 220 rpm and cultured at
34 ℃ for 48 h. Before inoculating the culture solution to a 5 L tank for
fermentation culture, the shake flask culture solution was firstly
washed, and then cultured and sampled according to the previous reported
method (Hong et al., 2017). 5 μg/mL thiostrepton and 10 μg/mL apramycin
were added to the medium. The E. coli DH5α was used in cloning
experiments, which was grown at 37 ℃ in the Luria-Bertani liquid medium
(Hopwood et al., 2000).
Intracellular sample preparation
The samples in the labeling experiment were obtained by a fast sampling
device independently designed by our research group (Fig. 2). The rapid
sampling device has the characteristics of simplicity, reliability,
strong applicability, which can quickly draw a sample within 1 second
under minimum hardware conditions, and introduce multiple sets of
sampling pipes of different pipe diameters to improve the compatibility
of different fermentation processes (broth with low or high viscosity)
(Wang et al., 2018). Methods for extracting different intracellular
metabolites were performed according to the previous report (Hong et
al., 2017). Briefly, phosphorylated sugar and coenzyme A substances were
extracted by liquid nitrogen grinding method, while boiling ethanol
method was suitable for organic acid extraction.
Analytical methods
Cell concentration and erythromycin titer
5 mL fermentation broth was centrifuged at 4000 rpm for 5min. After
removing the supernatant, the precipitate was dried at 80 ℃ for 48 h,
and dry cell weight was used to represent the cell concentration (Zhang
et al., 2014). Oxygen uptake rate (OUR) and carbon dioxide evolution
rate (CER) was calculated online as reported (Liang et al., 2011; Parekh
et al., 2000). The erythromycin titer and components were measured by
high performance liquid chromatography as previously described by Zou et
al. (2009).
[1-13C] sodium propionate labeling
experiment
The labeling experiments were carried out at 34°C in a 250 mL mini tank
(Fig. 3). Due to the absence of labeled n-propanol for sale on the
market and considering the fact that n-propanol firstly converts into
propionic acid under the presence of alcohol dehydrogenase,13C labeled sodium propionate instead of n-propanol
was used as labled substrate. The temperature of the fermentation broth
was controlled at 34°C using the heating coil of the tank. By
controlling the stirring rate and the aeration, the dissolved oxygen of
the fermentation broth was maintained over 30%. The maximum agitation
was 500 rpm and the maximum aeration was 300 mL/min.
After inoculating for 48 h in a 5 L bioreactor, 150 mL of the
fermentation broth in the 5 L bioreactor was pumped into a 250 mL mini
bioreactor using a sterile tube. At the same time, 0.2 g/L
[1-13C] sodium propionate (Cambridge Isotope
Laboratories, 98% isotopic purity) (the final concentration in the
fermentation broth) was added into the 250 mL mini tank. Then,
continuous sampling was carried out from the mini tank at the designed
time point.
Metabolic transformation of S. erythraea strain
Plasmid pOJ260 (Bierman et al., 1992), which was used for generating the
knockout of sucC in strain S. erythraea E3, has
thiostrepton-resistance gene (tsr ) and apramycin-resistance gene
(acc ). Plasmid pOJ260 can be amplified in E. coli , but it
cannot be amplified in S. erythraea . The plasmid
pOJ260-ΔsucC as a gene knockout cassette was generated as
following steps (Fig. S1). The 1.021-kb fragment of sucC -H was
made by PCR amplification of the S. erythraea genomic DNA using
the primers A1, 5’-CCC AAG CTT GGG ATG AGG CCA AGA CGA A-3’and A2,
5’-GAA GAT CTT CGC CCT GGA CGA TGA CCT TG -3’. The sucC-H fragment and
the plasmid pOJ260-tsr were digested with restriction enzymesHin dIII and Bgl II simultaneously, and the products were
ligated. The plasmid pOJ260-ΔsucC was constructed and propagated
in E. coli DH5α. The DNA sequence of the gene knockout cassette
was verified by sequencing (Takara). Finally, it was introduced fromE. coli to S. erythraea E3 using the standard
conjugational transfer methods.
The mutant S. erythraea E3-ΔsucC (E3-ΔsucC ) was
constructed by the addition of sucC partial ORF via the sucC-H
fragment from the cassette (Fig. S1). A single crossover plasmid
insertion was performed with pOJ260-ΔsucC by homologous
recombination (Matsushima et al., 1994) into the genome of original
strain E3 with screening by thiostrepton resistance and apramycin
resistance, to form the gene-knockout strain E3-ΔsucC .
The primers B1, 5’-AAC TCT TCG CCT CCC ACG GCG T-3’ , B2, 5’-TCA CCA GCT
CCG CGA AGT CGC T-3’, C1, 5’-CTG ACC GGC AAT CAC CAA CGC AA-3’ and C2,
5’-ACG GAG CCG AAG ACG GTG AGC T-3’ flanking, the inserted gene knockout
cassette were used for the identification of correct mutants by PCR.
Isotopically nonstationary labeling experiments
Triplicate cultures of E3 and E3-ΔsucC were inoculated to an
initial 0.78 gDCW/L in 3 L optimized synthetic medium in the 5 L
fermenter. The labeling experiment was initiated at 50h, when the cells
reached a cell density of approximately 5~6 gDCW/L. A
140-mL sample corresponding to t=0 (unlabeled) was withdrawn from the 5
L fermenter using a pump and sterilized latex tube to the three parallel
250 mL fermenters. 6 M NaOH filter bottle was used to prevent unlabeled
CO2 in the air entering the system. A 30 mL aliquot of
media with different ratios of labeling glucose (Cambridge Isotope
Laboratories, 99% isotopic purity) was quickly introduced to the
fermenter to achieve a final trace concentration. Following this pulse,
a series of 2-mL samples were withdrawn and rapidly quenched at time
points 10, 20, 30, 40, 50, 60, 80, 100, 130, 150, 180 s, respectively.
Each sample was quenched in a 50-mL centrifuge tube that contained 30-mL
liquid nitrogen. In particular, one needed to be filtered into the
liquid nitrogen after the supernatant was filtered for removal of
extracellular organic acids, since the concentration of extracellular
organic acids are several orders of magnitude higher than the
intracellular organic acids (Mashego et al., 2007). The cell pellet was
extracted using a previously reported extraction method (Hong et al.,
2017). Dried extracts were stored at -80℃ prior to detection.
LC–MS/MS measurement of metabolite abundance information and
pool size
The major fragment patterns of metabolites and its 13C
derivative was adapted from Seifar et al. (2013) and was performed on a
Thermal Ultimate 3000 UPLC system coupled to a Thermal TSQ QUANTUM ULTRA
mass spectrum system at the State Key Laboratory of Bioengineering, East
China University of Science and Technology, Shanghai China. An ACQUITY
UPLC BEH C18 column (1.7 μm, 2.1*150 mm) was used with
gradient elution. The eluents used were 5% acetonitrile aqueous
solution with 5mM dibutylammonium acetate (A) and 84% acetonitrile
aqueous solution with 5mM dibutylammonium acetate (B). All the chemicals
were HPLC or LC-MS grade and purchased from Sigma Aldrich, USA. An
injection volume of 2 μL was used, and the column flow rate and
temperature were kept constant at 5 μL/min and 25℃, respectively. In
order to determine the intracellular metabolites precisely,13C-labeled internal standards were added to the
samples after quenching (Wu et al., 2005). The gradient profile of
different metabolites was adapted from Hong et al. (2017). The
acquisition of concentration and isotope data was performed using
negative electrospray ionization in the selected reaction monitoring
(SRM) mode. The final parameters used for isotomer measurements are
listed in Supplementary materials. All data acquisition and evaluation
were performed on the Xcalibur software (Thermal Scientific) supplied
with the instrument.
Metabolic network model and 13C-metabolic flux
analysis
The central carbon metabolism network of S. erythraea consists of
EMP pathway, TCA cycle, PP pathway, ED pathway, glyoxylate shunt, amino
acid synthesis reaction, one-carbon unit metabolism, amphibolic
reactions, acetic acid synthesis, exchange reaction (transport reaction)
and bacterial synthesis reaction with a total of 68 reactions. The
metabolic network is a simplification of the metabolic model of S.
erythraea in the KEGG database. The central carbon metabolic network ofS. erythraea can be found in the supplementary materials (Table
S3).
All 13C-metabolic flux analysis was calculated with
INCA software (Young, 2014), which was based on the EMU algorithm
framework (Young et al., 2008). Metabolic fluxes were estimated by
minimizing the variance-weighted sum of squared residuals (SSR) between
the experimentally measured and model predicted external rates and mass
isotopomer distributions using non-linear least-squares regression
(Antoniewicz et al., 2006). All measured mass isotopomers were provided
in Supplemental materials. Flux estimation was repeated more than 10
times starting with random initial values for some key fluxes to find a
global answer. The exact 95% confidence interval for all estimated
fluxes was calculated by evaluating the sensitivity of the minimized SSR
to flux changes. Standard deviations of estimated fluxes were determined
as follows:
SD= [(fluxupper bound 95%)-(fluxlower
bound 95%)]/4 (1)
To determine the goodness-of-fit, 13C-MFA fitting
results were subjected to a χ2-statistical test. In
short, assuming the model is correct and the data has no serious
measurement errors, the minimized SSR is a random variable with a
χ2 distribution (Antoniewicz et al., 2006). Acceptable
range of SSR values is between χ2α/
2(n-p ) and χ21-α/
2(n-p ), where α is a certain selected threshold value, for
example, 0.05 for 95% confidence interval.
Results
Identification of n-propanol catabolism pathways
According to KEGG database, two pathways were speculated to catabolize
the propanol into tricarboxylic acid (TCA) cycle (Fig. 4). Propanol was
firstly converted to propionyl-CoA, then to methylmalonyl-CoA and
succinyl-CoA sequentially in pathway I, or to malonyl-CoA and acetyl-CoA
sequentially in pathway II. In terms of the previous report (Mironov et
al., 2004), malonyl-CoA decarboxylase in S. erythraea catalyzes
the conversion of malonyl-CoA to acetyl-CoA. However, it existed many
gaps in the remaining metabolic pathways, and the complete metabolic
pathway cannot be actually presented.
The labeling experiments using 13C-labeled sodium
propionate were conducted to identify the activities of these two
possible pathways on the basis of carbon atom transition map derived
from KEGG database (Fig. 5). The temporal significant increases of
labelling enrichments (M1, M2 and M3) in propionyl-CoA (Fig. 6A) and
succinyl-CoA (Fig. 6B) indicated the obvious activity of pathway I. The
increase of M2 and M3 in malonyl-CoA (Fig. 6C) hinted the activities of
the reactions from propionyl-CoA to malonyl-CoA in pathway II, but the
almost unchanged labelling enrichments in acetyl-CoA (Fig. 6D) showed
the very low or even no activity of the reaction which converts
malonyl-CoA to acetyl-CoA. In order to increase the industrial
conversion of n-propanol to erythromycin, blocking pathway I to reduce
the catabolism of n-propanol should be a feasible strategy.
Construction of engineered strain E3-ΔsucC and comparison
of cell growth and erythromycin synthesis
As the fact that propanol can be catalyzed through the pathway I and TCA
cycle (Fig. 5), it was assumed that the block of the connection from
n-propanol metabolism to TCA cycle could improve the conversion yield of
n-propanol to erythromycin. A novel strain E3-ΔsucC was
constructed by knocking-out the sucC gene (succinyl-CoA
synthetase), which catalyzes succinyl-CoA to succinate. Theoretically,
more carbon flux can be pushed towards erythromycin production insucC -knockout strain due to succinyl-CoA and unable to be
converted to succinate (Fig. 7). Moreover, as the conversion of
α-ketoglutarate to succinyl-CoA was irreversible, this strategy can not
only decrease the catabolism of n-propanol, but also can promote the
precursors derived from glucose to erythromycin synthesis.
Initially, it was hypothesized that the partial defect of TCA cycle in
E3-ΔsucC might result in the decrease of cell growth, however,
the influence on DCW of E3-ΔsucC was not significantly different
with E3 (Fig. 8A), indicating the low metabolic burden due to initial
medium composition and increased glyoxylate cycle activity to compensate
for short circuit in TCA cycle. As expected, it was observed that the
erythromycin concentration of E3-ΔsucC (786.6 mg/L) was almost
two-fold of E3 (392.94 mg/L) in 5 L fermenter (Fig. 8A), and similar
trends were observed in flasks (Fig. S3), which was in supplementary
materials.
Although the feeding rates of n-propanol were basically the same in the
fermentation processes of two strains, the residual n-propanol
concentration of E3-∆sucC was obviously lower than that of E3
(Fig. 8B), demonstrating that the n-propanol consumption of
E3-∆sucC was higher. The decline of residual n-propanol
concentration after 84 h might imply the insufficient supply of
n-propanol. The metabolic activity of the engineered strain might be
stronger compared with the starting strain. After entering the
stationary period, the engineered strain showed a stable OUR level
compared with the OUR of the starting strain, which presented a gradual
decline in the late phase, moreover, the CER level of the engineered
stain in the whole fermentation process were higher than that of the
starting strain (Fig. 8C).
After knocking out of sucC gene, S. erythraea exhibited
obvious changes in some key physiological parameters (Table S3). The
specific glucose consumption rate and propanol consumption rate were
increased from 0.049 to 0.067 mmol/gDCW/h and 0.058 to 0.083
mmol/gDCW/h, respectively. In contrast, the specific erythromycin
production rate (qerythromycin) and the conversion yield
of n-propanol to erythromycin (Yerythromycin/propanol)
were significantly increased by 52% and 81%, respectively. In terms of
cell respiratory metabolism, including the specific carbon dioxide
evolution rate and specific oxygen uptake rate, the engineered strain
had no significant changes compared to the starting strain.
Intracellular metabolites pool size and isotopically
non-stationary metabolic flux analysis
Targeted metabolomics analysis was applied to examine the intracellular
metabolite levels in each strain since intracellular reaction rates are
correlated with intracellular metabolite pool sizes and activities of
correspondent enzymes (Jazmin et al., 2017). Fig. 9 showed the pool
sizes of key metabolites in glycolysis (EMP), TCA cycle and pentose
phosphate (PP) pathway during the stationary phase. Knocking outsucC gene led to an almost 2-fold increase in α-ketoglutaric pool
size compared to E3, indirectly identifying the inactivation of
succinyl-CoA synthetase. This result may be caused by the feedback
inhibition of α-oxoglutarate (AKG) on citrate synthase (Wright et al.,
1967). The intracellular phosphorylated compounds, such as glucose
6-phosphate (G6P) and fructose 1,6-bisphosphate (FBP)presented downward
trends. Notably, the concentration of FBP in E3-ΔsucC was
relatively lower than that in E3 (Fig. 9), indicating a low metabolic
flux of the EMP pathway (Bennett et al., 2009). Interestingly, the
intracellular metabolites in PP pathway showed opposite trends. The
ribulose-5-phosphate (Ru5P), R5P (ribose-5-phosphate) and E4P (erythrose
4-phosphate) in E3-ΔsucC were in upward trends during the
stationary period, which reflected a high activity of PP pathway.
The synthesis of erythromycin required one propionyl-CoA and six
methylmalonyl-CoA (Chan et al., 2009). As two important precursors of
erythromycin, the rise of the intracellular pools of propionyl-CoA and
methylmalonyl-CoA was consistent with the increase of erythromycin
synthesis rate. Comparing to the coenzyme A trend in E3-ΔsucC ,
the intracellular pools of both propionyl-CoA and methylmalonyl-CoA
increased, implying that the significant decrease in the concentration
of intracellular precursor coenzyme A was a key factor affecting
erythromycin synthesis.
INST-MFA was applied to characterize the metabolic phenotypes of E3 and
E3-ΔsucC . The measured MIDs, the substrate consumption rates, the
cell growth rate and the erythromycin productivity were used to
construct comprehensive flux networks using a modified version of the
reaction network developed by Hong et al. (2017). The flux network of
E3-ΔsucC and E3 (Fig. 10) were statistically acceptable based on
a chi-square test of the SSR (SSR = 471.9 was within the expected range
[380.0, 495.7] for E3 and SSR = 193.9 was also within the expected
range of [184.4, 267.3] for E3-ΔsucC ). The full list of
optimal parameter estimates including net fluxes and exchange fluxes for
both E3 and E3-ΔsucC could be found in the Supplementary
Materials.
The metabolic flux distribution indicated that most of the NADPH was
from PP pathway, which provides reducing power for erythromycin
synthesis. It could be speculated that the overexpression of NADPH
synthase or increasing the metabolic flux of PP pathway would further
improve the erythromycin yield (Chen et al., 2017). We also observed
little or no flux through the reaction of ‘SucCoA→Suc’, which was in
agreement with the expectation of genetic manipulation in
E3-ΔsucC . Additionally, there was minimal flux through the EMP
pathway, which was consistent with intracellular phosphorylated sugar
concentration. In particular, the fluxes of reaction ‘F6P→FBP’ of
E3-ΔsucC was almost zero (Fig. 10), indicating that the glucose
was seldom metabolized through the EMP pathway. This result might be
caused by the inhibition of phosphofructokinase (PFK) (Borodina et al.,
2008; Hong et al., 2017).
The flux analysis highlighted the activity of glyoxylate shunt in
E3-ΔsucC (0.23 vs 0.00 mmol/gDCW/h). Glyoxylate by-pass can
replenish and repair defective TCA cycles. However, there was minimal
flux through the oxidative arm of the TCA cycle (from citrate to
α-ketoglutarate), which reduced the level of oxidative phosphorylation
and limited the energy supply. The significant decrease in succinyl-CoA
synthetase flux was necessary to supply substrate for erythromycin
production. This finding verified a major alternative route for succinic
acid formation and demonstrated that the strains constructed for
metabolic engineering exhibited a relatively significant advantage in
the erythromycin synthesis.
Discussion
This study demonstrated that the knowledge from pathway identification
could be used to guide rational metabolic engineering to improve strain
performance. The labeling experiments using 1-13C
sodium propionate identified the activity of n-propanol catabolism
pathway in S. erythraea based on the carbon atom transitions of
n-propanol and the measured enrichments of intracellular coenzyme A
metabolites. The availability of carbon atom transitions is critical for
revealing the activity of alternative metabolic pathways (Fan Jing et
al., 2014; Li Bo et al., 2014; Andrew R. Mullen et al., 2012). To the
best of our knowledge, the carbon atom transitions of n-propanol
catabolism pathways in S. erythraea was constructed and used to
demonstrate the presence of oxidative catabolism in the utilization of
propanol in S. erythraea for the first time.
According to the fact that n-propanol was not only acting as a precursor
for erythromycin synthesis but also could be catabolized into TCA circle
through pathway I, the genes involved in this pathway reasonably became
the target for gene manipulations in order to enhance erythromycin
synthesis. The significant increase of erythromycin production in strain
E3-ΔsucC (Fig. 8) was expectable. Although the beneficial effects
on erythromycin synthesis by deleting genes involved in n-propanol
catabolism have been reported in our previous study (Chen et al., 2016),
where the knock-out of mutB gene encoding the beta subunit of
methylmalonyl-CoA mutase led to an obvious increase of erythromycin
synthesis, the results of knocking-out sucC gene was reported for
the first time. In this study, atomic migration and labeling experiments
rationally confirmed the n-propanol metabolic pathway and corresponding
transformation target, and combining the intracellular metabolomics with
metabolic flux analysis elucidate the mechanism and guide the future
industrial production, which had universal significance and application
for the antibiotic strain construction.
Metabolic modification of specific targets for industrial production
strains is one of the most important ways to improve antibiotic
production (Young et al., 2011; Jazmin et al., 2017). Deletion ofSACE_Lrp (an efficient regulator for transporting and
catabolizing branched-chain amino acids (BCAAs)) of industrial strain WB
combined with overexpression of its target SACE_5387-5386enhanced erythromycin production by 41% (Liu et al., 2017). In this
study, deletion of SACE_6669 (sucC ) of industrial strain
E3 significantly increased erythromycin production. It is of great
interest to improve the potential synergies of erythromycin production
through different synergistic mechanisms. Therefore, further study will
continue to knock out gene encoding succinyl-CoA mutase and succinyl-CoA
synthetase to block the flow of n-propanol to the TCA cycle and
over-express the polyketide synthase gene to introduce the precursor
into erythromycin synthesis, which is expected to obtain high-efficient
industrial recombinant producer. Importantly, our work also reveals a
potential for broader applications for metabolic reconstruction of
industrial strains.
The metabolic flux networks of E3 and E3-ΔsucC were determined
using chemically-defined medium by INST-MFA. INST-MFA is ideally suited
to systems that label slowly due to the presence of large intermediate
pools or pathway bottlenecks (Cheah and Young, 2018). The observation of
very little flux of reaction converting succinyl-CoA to succinate in
strain E3-ΔsucC was consistent with the expectation of knocking
out sucC gene. Previous studies have inferred that the
erythromycin synthesis was increased by 126% by blocking the flux to
TCA cycle (Reeves et al., 2006). Therefore, the low flux throughsucC may serve as a key route in order to pull more carbon to
erythromycin synthesis. In spite of the apparent importance ofsucC flux, the E3-ΔsucC strain showed a rapid drop of cell
density in the stationary phase, which seemed to be ascribed to the
depletion of intracellular succinate (Fig. 9), suggesting that
decreasing the flux from succinyl-CoA toward succinate should be
balanced by increasing the flux of glyoxylate bypass simultaneously.
One potential concern is that we did not attempt to optimize the culture
conditions of fermentation systematically, since this was not a goal of
our study. We performed all experiments under the same standard culture
conditions described in the Materials and Methods section, and when the
engineered strain was in the stationary phase, the cell mass decreased
rapidly, leading to a shortened final fermentation duration, which
restricted the further synthesis of erythromycin. According to the
literatures, we noted that nitrogen source in the erythromycin
production phase played an important role that stimulating the synthesis
of ppGpp and initiating transcription and translation of antibiotic
synthesis genes (Chakraburtty and Bibb, 1997). In previous study,
Enshasy et al. (2008) used molasses as a carbon source, corn syrup and
ammonium sulphate as nitrogen sources, and erythromycin production
increased by 33% compared to the original formulation. Therefore, we
performed an additional experiment to investigate the effects of
ammonium sulphate supplementation on the erythromycin of E3-ΔsucCstrain (Fig. S4). Importantly, the total content of erythromycin A was
significantly improved than E3-ΔsucC in ammonium sulphate-free
media. This result suggests that the addition of the ammonium sulphate
can enhance cell growth of mid-late of phase of fermentation. The
mechanism of the enhancement of erythromycin A by the addition of
ammonium sulfate will be further explored. Finally, it is important to
note that this article is a general DBVG method that can be extended to
the other strains for its genetic modification and process optimization.
Conclusions
Overall, this study demonstrated the metabolic fate of propanol in the
erythromycin production process via 13C labeling
experiment. At first, the metabolic engineering target was identified,
and subsequently an engineered erythromycin high producer was
constructed. Finally, the effectiveness of engineering strain
modification was analyzed through combining INST-MFA with targeted
metabolomics. [1-13C] sodium propionate labeling
experiment revealed an active pathway in S. erythraea E3 and
suggested that enzymes involved in propanol catabolism could be
potential targets for debottlenecking flux towards TCA cycle. The target
was deleted in the parental S. erythraea E3 strain, and
erythromycin production were significantly enhanced. It was the first
report, to the best of our knowledge, that the gene encoding
succinyl-CoA synthetase has been involved in microbial research, and
using unsteady isotope labeling experiment for metabolic flux analysis
of secondary metabolism of S. erythraea . However, further work
shall quantify the medium composition and optimization of culture
process in order to realize the industrialization of engineered strain.
In addition, future studies are required to examine additional
engineering targets to increase precursor conversion rate. In short, our
work on account of 13C flux analysis was proved useful
to identify rational targets for strain improvement that provide basis
and valuable information for further enhancement in erythromycin
production.