Effect of flexible linker length on the activity of fusion protein 4-coumaroyl-CoA ligase::stilbene synthase

Article abstract of DOI:10.1039/c6mb00563b

In order to elucidate the effect of flexible linker length on the catalytic efficiency of fusion proteins, two short flexible peptide linkers of various lengths were fused between Arabidopsis thaliana 4-coumaroyl-CoA ligase (4CL) and Polygonum cuspidatum stilbene synthase (STS) to generate fusion proteins 4CL-(GSG)_n-STS (n ≤ 5) and 4CL-(GGGGS)_n-STS (n ≤ 4). The fusion proteins were expressed in both Escherichia coli and Saccharomyces cerevisiae, and their bioactivities were tested in vitro and in vivo using purified proteins and engineered strains, respectively. The catalytic efficiency of the fusions decreased gradually with the increase of GSG or GGGGS repeats. In both engineered S. cerevisiae and E. coli in vivo experiments, the capacity of resveratrol production decreased gradually with increasing linker length. In silico analysis showed that the prediction of homology models of fusion proteins was consistent with the in vitro and in vivo results.

Full text of DOI:10.1039/c6mb00563b

Molecular
BioSystems
View Article Online
View Journal
PAPER
Effect of flexible linker length on the activity of
fusion protein 4-coumaroyl-CoA ligase::stilbene
synthase†
Cite this:DOI: 10.1039/c6mb00563b
Huili Guo,‡a Yadong Yang,‡a Feiyan Xue,‡a Hong Zhang,a Tiran Huang,a
Wenbin Liu,a Huan Liu,a Fenqiang Zhang,a Mingfeng Yang,a Chunmei Liu,a
Heshu Lu,*a Yansheng Zhang*b and Lanqing Ma*ac
In order to elucidate the effect of flexible linker length on the catalytic efficiency of fusion proteins, two
short flexible peptide linkers of various lengths were fused between Arabidopsis thaliana 4-coumaroyl-
CoA ligase (4CL) and Polygonum cuspidatum stilbene synthase (STS) to generate fusion proteins
4CL–(GSG)n–STS (n r 5) and 4CL–(GGGGS)n–STS (n r 4). The fusion proteins were expressed in both
Escherichia coli and Saccharomyces cerevisiae, and their bioactivities were tested in vitro and in vivo
using purified proteins and engineered strains, respectively. The catalytic efficiency of the fusions
decreased gradually with the increase of GSG or GGGGS repeats. In both engineered S. cerevisiae and
E. coli in vivo experiments, the capacity of resveratrol production decreased gradually with increasing
linker length. In silico analysis showed that the prediction of homology models of fusion proteins was
consistent with the in vitro and in vivo results.
Received 3rd August 2016,
Accepted 24th January 2017
DOI: 10.1039/c6mb00563b
rsc.li/molecular-biosystems
However, only the chimera with the linker (GGGGS)4 was desired
when penicillin amidase and thermophilic chaperonin were fused.6
Introduction
Linking genes for the expression of unnatural fusion proteins is Compared to the G5 linker, the insertion of the G8 linker
a powerful strategy for generating proteins with improved greatly improved the in vivo function of epitope-tagged
functions and bioactivities.1–5 The linker regions between the Est2p.17 (Gly–Ser–Gly)n and (Gly–Gly–Gly–Gly–Ser)n are the most
functional domains can influence the production and properties commonly used flexible linkers in the construction of fusion
of fusion proteins, and can be modified to improve structural proteins.8,18–20 By adjusting the copy number ‘‘n’’, the length of
folding and stability,5–8 enhance expression,9,10 and elevate the linker can be optimized to achieve appropriate separation
biological activity.11–13 The correct design of the linker peptide of the functional domains, or to maintain necessary inter-domain
can be crucial for fusion protein function.14,15 Generally, linker interaction.
length, composition, hydrophobicity, sensitivity to proteases,
Resveratrol is a polyphenolic compound from the stilbene
and secondary structure are worthy of careful consideration family. It shows a wide range of beneficial properties such
when designing linkers,14 wherein the flexible linker length as antitumor, antithrombotic, antidiabetic, anti-inflammatory
has a direct impact on the function of the fusion protein. For and antiaging properties.21,22 Resveratrol is synthesized by the
constructing scFv, the short linker (GGGGS)1 was demonstrated phenylpropanoid pathway, in which 4-coumaric acid is converted
to be the best.16 For fusing b-glucanase (Glu) and xylanase (Xyl), into resveratrol under the concerted activities of 4-coumaroyl-CoA
(GGGGS)2 was the best among the tested linkers (GGGGS)n (n r 3).8 ligase (4CL) and stilbene synthase (STS). After the successful
elucidation of the resveratrol biosynthesis pathway, engineered
microbial organisms for the production of resveratrol have been
a Key Laboratory of Urban Agriculture (North) of Ministry of Agriculture,
reported in many cases.23–28 Among these efforts, the construction
Beijing University of Agriculture, Beijing 102206, China.
of unnatural fusion proteins to engineer resveratrol biosynthesis
was shown to be the most powerful strategy.18 Fusion of Arabidopsis
thaliana 4CL and Vitis vinifera STS via a simple GSG tripeptide to
give At4CL–GSG–VvSTS resulted in a 15-fold improvement in
resveratrol production compared with separate expression of 4CL
and STS.18 Further crystallographic analysis of this fusion protein
suggested that the improved production likely resulted from the
E-mail: lqma@bua.edu.cn; Tel: +86-10-80797305
b CAS Key Laboratory of Plant Germplasm Enhancement and Specialty Agriculture,
Wuhan Botanical Garden, Chinese Academy of Sciences, Wuhan 430074, Hubei,
China
c Beijing Collaborative Innovation Center for Eco-Environmental Improvement with
Forestry and Fruit Trees, Beijing 102206, China
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6mb00563b
‡ Huili Guo, Yadong Yang and Feiyan Xue contributed equally to this work.
This journal is ©The Royal Society of Chemistry 2017
Mol. BioSyst.

View Article Online
Paper
Molecular BioSystems
closer proximity of the 4CL and STS active sites.29 However, same restriction sites, under the control of T7 and Gal1
whether GSG is the optimal linker for fusing 4CL and STS promoters, to yield pET-30a-4CL and pESC-TRP-4CL, respec-
remains unknown, since the effects of adjusting the linker type, tively. The STS gene was also cloned into pET-30a(+) and
length, composition and flexibility on bioactivity have not been pESC-TRP at the above sites to generate the pET-30a-STS and
fully investigated.
pESC-TRP-STS vectors, respectively.
To answer these questions, two short flexible peptide linkers
To construct nine 4CL–linker–STS fusion genes, the stop
of various lengths were fused between Arabidopsis thaliana codon of the 4CL gene was removed, and linkers of different
4-coumaroyl-CoA ligase (4CL) and Polygonum cuspidatum stilbene lengths, (Gly–Ser–Gly)n (n = 1, 2, 3, 4, or 5), or (Gly–Gly–Gly–Gly–Ser)n
synthase (STS) to generate fusion proteins 4CL–(GSG)n–STS (n r 5) (n = 1, 2, 3, or 4), were introduced between the open reading frames
and 4CL–(GGGGS)n–STS (n r 4). The catalytic efficiency of the of the 4CL and STS genes via two rounds of PCR;32 a linker-free
fusion proteins with various linker lengths was tested in vitro and fusion gene was also constructed. The first round of the PCR
in vivo using purified enzymes and engineered strains, respectively. introduced the linkers (GSG)n (n r 5) and (GGGGS)n (n r 4) into
Subsequently, protein structure-based homology modeling was the 4CL gene using 10 pairs of primers. Simultaneously, the
used to analyze the results.
complementary linkers (GSG)n (n r 5) and (GGGGS)n (n r 4) were
introduced into the STS gene using 10 other pairs of primers. Each
PCR product was purified and served as a template in the second
round of PCR. Next, the PCR products of the 4CL and STS genes
were joined by overlapping extension PCR33 using the primers
linker-up and linker-dw. The PCR program included a 4 min period
at 94 1C, 35 cycles at 94 1C (30 s), 62 1C (30 s) and 72 1C (45 s), and a
final 10 min extension at 72 1C. The gel purified PCR products were
ligated into a pMD18-T vector. The 10 fusion genes were confirmed
by digestion checks and sequencing. Finally, the 10 fusion genes
were ligated to pET-30a(+) and pESC-TRP between BamHI and XhoI
sites, yielding pET-30a-4CL–STS, pET-30a-4CL–(GSG)n–STS (n r 5),
pET-30a-4CL–(GGGGS)n–STS (n r 4), pESC-TRP–4CL–STS, pESC-
TRP-4CL–(GSG)n–STS (n r 5), and pESC-TRP–4CL–(GGGGS)n–STS
(n r 4), respectively.
Materials and methods
Chemicals
Authentic compounds, including 4-coumaric acid, ATP, coenzyme
A (CoA), malonyl-CoA and resveratrol, were purchased from
Sigma-Aldrich (St. Louis, MO, USA). Anti-actin antibody produced
in rabbits was also purchased from Sigma-Aldrich. All solvents
for high-performance liquid chromatography (HPLC) and liquid
chromatography-coupled mass spectrometry (LC-MS) were acquired
from Fisher Scientific (Waltham, MA, USA). Other enzymes were
obtained from TaKaRa (Shiga, Japan) unless specified otherwise.
Strains and culture conditions
Expression and purification of fusion proteins
Escherichia coli Top10 (Invitrogen, Carlsbad, CA, USA) was used
for routine cloning and plasmid propagation. E. coli Rosetta
(Novagen, Darmstadt, Germany) was used for protein expression
and resveratrol biosynthesis. The E. coli strains were grown at
37 1C on Luria–Bertani (LB) agar plates or in liquid LB medium.
Enzymes were expressed in E. coli Rosetta at 25 1C. M9 medium
was used for resveratrol synthesis in E. coli Rosetta. Saccharomyces
cerevisiae WAT11 was also used as the host for the production of
resveratrol. S. cerevisiae cells were cultured in Yeast Peptone
Dextrose (YPD) medium at 30 1C. S. cerevisiae cells transformed
with integrative plasmids were selected on synthetic complete (SC)
dropout medium.
Each expression construct, pET-30a-4CL–(GSG)n–STS (n r 5),
pET-30a-4CL–(GGGGS)n–STS (n r 4), or pET-30a-4CL–STS, was
transformed into E. coli Rosetta grown in LB medium containing
kanamycin and chloramphenicol, and grown at 37 1C until
reaching an optical density of 600 nm (OD600) of 0.6–0.8.
Protein expression was induced by adding 1 mM isopropyl
thio-b-D-galactopyranoside (IPTG), and the culture was grown
at 25 1C for 6 h. The cells were harvested by centrifugation
(8000 Â g; 10 min; 4 1C). The pellets were re-suspended in lysis
buffer (20 mM Tris, pH 7.5, 300 mM NaCl, 20 mM imidazole,
1 mg mlÀ1 leupeptin, 2 mg mlÀ1 aprotinin, and 100 mg mlÀ1
lysozyme). The cells were disrupted by sonication (30 Â 3 s;
4 1C). The homogenate was centrifuged at 10 000 Â g for 10 min
Gene fusion and plasmid construction
The 4CL gene (GenBank Accession number AY376729)30 from at 4 1C. The supernatant was passed through a Ni–NTA His-Bindt
Arabidopsis thaliana was amplified from pCRT7/CT-TOP-4CL Resin (Novagen, Madison, WI, USA) column containing Ni2+ as an
(preserved in our laboratory) using the primers 4CL-up and affinity ligand. After loading of the protein sample, the column
4CL-dw. The STS gene (GenBank Accession number ACC76753)31 was washed (20 mM Tris, pH 7.5, 300 mM NaCl, and 20 mM
from Polygonum cuspidatum was cloned from pET-30b-STS imidazole) to remove unbound contaminant proteins. His-tagged
(preserved in our laboratory) using a pair of primers STS-up proteins were eluted (20 mM Tris, pH 7.5, 300 mM NaCl, and
and STS-dw. The primers used in this study are summarized in 300 mM imidazole). For long-term storage, proteins were
the ESI,† Table S1. The two PCR products were gel purified and purified using a PD-10 column (Amersham Pharmacia Biotech,
ligated into a pMD18-T cloning vector. The genes were con- Uppsala, Sweden), and the samples were stored at À80 1C. The
firmed by digestion checks and sequencing. The 4CL gene was efficiency of purification was monitored by sodium dodecyl
excised from pMD18-T-4CL via BamHI/XhoI and ligated to the sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Proteins
prokaryotic expression vector pET-30a(+) and the S. cerevisiae were quantified by the method of Bradford using bovine serum
expression vector pESC-TRP (Stratagene, La Jolla, CA, USA) at the albumin as the standard.
Mol. BioSyst.
This journal is © The Royal Society of Chemistry 2017

View Article Online
Molecular BioSystems
Paper
Enzyme assays and steady-state kinetic analysis
tested concentrations of 4-coumaric acid. The OD600 value was
monitored and compared with that of the control culture
without adding 4-coumaric acid.
To determine the optimal conditions of the enzyme assays, all
of the fusion proteins were assayed at a pH gradient of 5.0–9.0,
temperatures ranging from 35 to 65 1C, and incubation times
ranging from 5 to 60 min. The standard assays (300 ml) containing
150 mM 4-coumaric acid, 280 mM malonyl-CoA, 250 mM ATP,
120 mM CoA, 0.1 M potassium phosphate (pH 7.0), and 1.0 mg of
fusion protein were incubated at 45 1C for 30 min, extracted
twice with 600 ml of ethyl acetate, and centrifuged at 10 000 Â g
for 10 min. Acetic acid (5% final concentration) was added
before the extraction step to detect any side product. After
drying under vacuum, the residue was dissolved in 50 ml of
50% (v/v) methanol. The enzymatic products were analyzed by
HPLC, and confirmed by LC-MS.
Kinetic constants were determined using five concentrations
of the substrate covering the Km range from 0.2 to 6.0, and the
concentration of the second substrate was the saturation
point under the optimal conditions. The assay was performed
three times with 1.0 mg of purified enzyme in a final volume of
300 ml of 0.1 M potassium phosphate buffer (pH 7.0), and
incubation was at 45 1C for 30 min. The kinetic parameters
were calculated for the formation of the major product of the
enzymatic reactions at the optimum pH. Lineweaver–Burk plots
of the data were used to derive the apparent Km and Kcat values
(average of triplicates Æ SD).
Biosynthesis and extraction of resveratrol in Escherichia coli
As in S. cerevisiae, twelve constructs (pET-30a-4CL–STS, pET-30a-
4CL–(GSG)n–STS (n r 5), pET-30a-4CL–(GGGGS)n–STS (n r 4),
pET-30a-4CL and pET-30a-STS) were transformed into the E. coli
Rosetta strain, independently or in combination. Each recombi-
nant E. coli Rosetta strain was pre-cultured overnight at 37 1C in
LB medium. Then, the overnight culture was diluted into fresh
LB supplemented with kanamycin and chloramphenicol. The
cells were incubated at 37 1C until the OD600 value reached 0.6, at
which point they were induced with 1 mM IPTG and grown for
another 6 h at 25 1C for protein production. Next, the cells were
harvested and re-suspended in M9 medium containing kanamycin
and chloramphenicol, 0.075 mM 4-coumaric acid, and 1 mM
IPTG. Incubation continued at 25 1C for 48 h prior to HPLC
analysis of the heterologous products.
Previous work has shown that less than 5% of the products
were found in the cell pellets;36 therefore, only the culture
medium was extracted. For extraction, 500 ml of the culture was
centrifuged to pellet the cells. The medium was extracted three
times with an equal volume (500 ml) of ethyl acetate. The
extracts were dried under vacuum, and the dried residue was
re-suspended in 100 ml of 50% (v/v) methanol for HPLC and
LC-MS analyses.
Biosynthesis and extraction of resveratrol in Saccharomyces
For optimum substrate concentration and incubation time
assays, each recombinant E. coli Rosetta strain was incubated
in M9 medium containing 0.025, 0.05, 0.075, and 0.1 mM
4-coumaric acid. The cultures were collected at 24, 48, 72,
and 96 h, and analyzed as above. HPLC was used to detect
the accumulation of resveratrol. E. coli Rosetta strains contain-
ing recombinant plasmids were grown in the presence of the
above tested concentrations of 4-coumaric acid. The OD600
value was monitored and compared with that of the control
culture without adding 4-coumaric acid.
cerevisiae
Ten constructs, pESC-TRP-4CL–STS, pESC-TRP-4CL–(GSG)n–STS
(n r 5), and pESC-TRP-4CL–(GGGGS)n–STS (n r 4), were trans-
formed using the lithium acetate method34 into the S. cerevisiae
WAT11 strain, independently. Two constructs, pESC-TRP-4CL and
pESC-TRP-STS, were co-transformed into the S. cerevisiae WAT11
strain as controls. In vivo S. cerevisiae assays were conducted as
previously described, with minor modifications.35 Each recombi-
nant S. cerevisiae WAT11 strain was grown at 30 1C for 24 h in
dropout SC liquid medium without tryptophan (FunGenome
Company, Beijing, China) containing 2% (w/v) glucose. The
cultures were collected by centrifugation and washed three times
Western blotting
in sterile de-ionized water. Cells were re-suspended in induction To test whether the fusion proteins degraded in the S. cerevisiae
medium containing 2% (w/v) galactose to an OD600 of 0.8. Then, expression systems, anti-STS Western blotting was performed.
4-coumaric acid was added to a final concentration of 0.075 mM Equivalent amounts of total protein isolated 24 h post incuba-
as the substrate. After incubation for 24 h, aliquots (500 ml) of the tion were separated by SDS-PAGE and transferred to polyviny-
cultures were extracted three times with an equal volume of ethyl lidene fluoride (PVDF) membranes with a current of 50 mA for
acetate without separating the cells, because the culture broth 2 h. The membranes were blocked with 5% skimmed milk in
contained about 10-fold more resveratrol than the cell extract.24 tris-buffered saline (TBS, 0.02 M Tris-HCl, 0.5 M NaCl, pH 7.5)
After drying under vacuum, the residue was re-dissolved in 100 ml for 1 h at 37 1C, the pooled rabbit antiserum against the
50% (v/v) of methanol for HPLC and LC-MS analyses.
prepared STS diluted to 1 : 1000 in TBS was added and incu-
For the time course and substrate concentration assays, bated for 1 h at 37 1C. The membranes were washed three times
each recombinant S. cerevisiae WAT11 strain was grown in for 10 min with 0.05% Tween-20 in TBS, followed by 1 h of
SC-Trp liquid medium containing 0.05, 0.075, 0.1, and 0.125 incubation with horseradish peroxidase (HRP)-conjugated goat
mM 4-coumaric acid. The cultures were collected at 6, 12, 24, anti-rabbit IgG (Sigma, St. Louis, MO) at 37 1C. After washing
and 36 h, and analyzed as above. HPLC was used to determine three times with TBST, the immunoblotting images were devel-
the resveratrol yield. S. cerevisiae WAT11 strains containing oped on the membrane using 6 mg methoxy-naphthol dissolved
recombinant plasmids were grown in the presence of the above in 2 ml of methanol and 10 ml of TBS with 6 ml of 30% H2O2.
This journal is ©The Royal Society of Chemistry 2017
Mol. BioSyst.

View Article Online
Paper
Molecular BioSystems
The membranes were scanned and evaluated by using the
computer software ImageMaster (Pharmacia Biotech).
Resveratrol analysis
HPLC samples were analyzed using a Waters Alliance 1525
HPLC system on a Sunfiret reverse-phase C18 column (5 mm,
150 mm  4.6 mm), which was connected to a Waters 2998
photodiode array detector, and controlled by a computer running
Empower software (Waters, Milford, MA, USA). The eluents were
methanol (A) and water (B) at a flow rate of 0.8 ml minÀ1. The
following gradients were used: 10–70% A for 30 min, and 70% A
for 10 min. The detection wavelength was 306 nm. A standard
solution of resveratrol was used for quantification.
For the on-line HPLC-MS analysis, liquid chromatography
was performed on a Shimadzu LC-10ADvp HPLC system, which
was coupled to a Shimadzu LC-MS-2010A single quadrupole
mass spectrometer with an electrospray ionization (ESI) inter-
face (Shimadzu, Kyoto, Japan). Data acquisition and processing
were performed using Shimadzu LC-MS solution software for
LC-ESIMS-2010. HPLC separations were conducted under the
conditions described above. The optimized MS conditions were
as follows: all spectra were obtained in negative mode over an
m/z range of 150–1000; sheath gas: N2, 40 ml minÀ1; auxiliary
gas flow, 3 ml minÀ1; curved desolvation line temperature,
300 1C; and probe voltage, 35 V.
Fig. 1 SDS-PAGE analysis of purified fusion proteins expressed in E. coli.
(A) Purification of 4CL–STS and 4CL–(GSG)n–STS (n r 5). M: molecular
weight standard, Lane 1: 4CL–STS (103.5 kDa), Lane 2: 4CL–(GSG)1–STS
(103.7 kDa), Lane 3: 4CL–(GSG)2–STS (103.9 kDa), Lane 4: 4CL–(GSG)3–
STS (104.1 kDa), Lane 5: 4CL–(GSG)4–STS (104.3 kDa), Lane 6: 4CL–
(GSG)5–STS (104.5 kDa). (B) Purification of 4CL–(GGGGS)n–STS (n r 4).
M: molecular weight standard, Lane 1: 4CL–(GGGGS)1–STS (103.8 kDa),
Lane 2: 4CL–(GGGGS)2–STS (104.1 kDa), Lane 3: 4CL–(GGGGS)3–STS
(104.4 kDa), Lane 4: 4CL–(GGGGS)4–STS (104.7 kDa).
in vitro kinetic assays. The effect of pH and temperature on
the fusion protein in vitro enzymatic reaction was determined
using 4-coumaric acid and malonyl-CoA as substrates. All the
fusion proteins shared an optimal value of pH 7.0 and tem-
perature 45 1C (data not shown).
To compare the catalytic properties of 4CL–(GSG)n–STS (n r 5)
and 4CL–STS, in vitro kinetic assays under the optimal conditions
were performed (Table 1). 4CL–STS exhibited much lower catalytic
activity, and the activity of the 4CL–(GSG)n–STS (n r 5) series
decreased with increasing linker length. The specificity constant
(Kcat/Km) values of 4CL–(GSG)1–STS were 9.2-fold and 6.7-fold
higher than 4CL–(GSG)5–STS with 4-coumaric acid and malonyl-
CoA substrates, respectively.
The 4CL–(GGGGS)n–STS (n r 4) series was also tested
in vitro (Table 1), and the (GGGGS)1 linker proved best for
both 4-coumaric acid and malonyl-CoA substrates, followed by
(GGGGS)2, (GGGGS)3 and (GGGGS)4. The activity of the fusion
proteins thus decreased with increasing linker length for both
the GSG and GGGGS series.
Homology model building
The homology-based structures for the fusion protein of Arabidopsis
thaliana 4CL and Polygonum cuspidatum STS, At4CL (22–482)::PcSTS
(1–387), were modeled using SWISS-MODEL (http://www.swissmo
del.expasy.org/) with the template of the coordinates from the
crystal structure of At4CL::VvSTS (PDB: 3TSY) fusing Arabidopsis
thaliana 4CL and Vitis vinifera STS.29 The amino acid sequence
identity between the PcSTS and the templates was as high as
86.86%. The position of the missing C-terminal amino acids of
At4CL (483–561) and the GSG linker region were determined in
COOT on the basis of the SSM superposing the poplar 4CL (PDB: Selecting appropriate substrate concentration
3A9V) structure to the At4CL in the fusion protein, which showed
It was reported that 4-coumaric acid could inhibit cell growth
root-mean-square deviations (rmsd) of 1.12 Å2 for 422 aligned
N-terminal Ca atoms of the poplar 4CL.
at high concentrations.37 To find an appropriate substrate
Table
1
Kinetic parameters of the 4CL–(GSG)n–STS (n
r
5) and
4CL–(GGGGS)n–STS (n r 4)a fusion proteins
Results
Expression and purification of fusion proteins
4-Coumaric acid
Malonyl-CoA
Km
5.35
The fusion genes constructed by linking the 4CL and STS with
nine of the peptide linkers, 4CL–STS, 4CL–(GSG)n–STS (n r 5),
and 4CL–(GGGGS)n–STS (n r 4), were heterologously expressed
in E. coli Rosetta with an additional hexahistidine tag at the C
terminus. The His-tagged fusion proteins were purified by Ni
affinity chromatography. The purified enzyme gave a single band
with a molecular mass of ca. 104 kDa on SDS-PAGE (Fig. 1).
Fusion protein
Kcat
Km
Kcat/Km Kcat
Kcat/Km
4CL–STS
0.001 0.21
78
1.07 17 287
0.007
1.39
0.87
1.09
0.89
0.57
1.55
0.88
0.40
0.78
21
4CL–(GSG)1–STS
4CL–(GSG)2–STS
4CL–(GSG)3–STS
4CL–(GSG)4–STS
4CL–(GSG)5–STS
1.11
0.37
0.5
0.49
0.30
19.88 1164
1.21
2.34
3.22
2.66
0.92
2.23
3.17
2.72
5096
3567
2536
1880
8107
3354
2087
1692
44.29
73.46
81.87
54.73
53.49
69.27
40.36
87.20
326
247
182
174
482
212
164
149
4CL–(GGGGS)1–STS 0.45
4CL–(GGGGS)2–STS 0.45
4CL–(GGGGS)3–STS 0.40
4CL–(GGGGS)4–STS 0.28
In vitro kinetic assays of fusion proteins
a
Units for Kcat, Km and Kcat/Km are minÀ1, mM, and MÀ1
s
À1, respec-
The ability of the 10 fusion proteins mentioned above to
catalyze the biosynthesis of resveratrol was assessed using tively. Mean values (n = 3) are shown, the errors are less than 10%.
Mol. BioSyst.
This journal is © The Royal Society of Chemistry 2017

View Article Online
Molecular BioSystems
Paper
concentration, engineered S. cerevisiae WAT11 containing
recombinant plasmid pESC-TRP-4CL–(GSG)1–STS was grown
in different 4-coumaric acid concentrations. As shown in
Fig. 2, inhibition was observed at more than 0.075 mM of
supplemented 4-coumaric acid. Accordingly, a concentration of
0.075 mM 4-coumaric acid was chosen for the feeding assays,
and feeding did not inhibit growth relative to un-fed cultures.
Resveratrol levels reached a maximum value at 24 h after
incubation in engineered S. cerevisiae. Engineered S. cerevisiae
strains expressing the other fusion proteins showed similar
results (data not shown).
In this way, we optimized the substrate concentration and
incubation time for the engineered E. coli strains. As shown in
Fig. 3, a concentration of 0.075 mM 4-coumaric acid and an
incubation time of 48 h were subsequently chosen for the
resveratrol biotransformation experiments, and feeding did
not inhibit growth compared with un-fed cultures. Engineered
E. coli strains expressing the other fusion proteins showed
similar results (data not shown).
Fig. 3 Effects of incubation time and 4-coumaric acid concentration on
the yield of resveratrol from an engineered E. coli strain expressing the
4CL–(GSG)1–STS fusion protein. The error bar represents the standard
deviation of three biological replicates.
in S. cerevisiae decreased gradually with an increasing number
of GSG linker repeats. A similar linker effect was apparent in
the in vivo E. coli assays (Fig. 4) in which (GSG)1 was the best
linker; the expression of 4CL–(GSG)1–STS produced a resvera-
trol yield of 32.5 mg mlÀ1, which was 2.1-fold higher than
expression of 4CL–(GSG)5–STS, and 2.0-fold higher than that
previously reported in an E. coli host.25
In vivo expression of 4CL–(GGGGS)1–STS in S. cerevisiae
WAT11 gave a resveratrol yield of 8.7 mg mlÀ1, which was
2.6-fold higher than that achieved by the expression of 4CL–
(GGGGS)4–STS (Fig. 5). In vivo expression of 4CL–(GGGGS)1–STS
in E. coli Rosetta increased resveratrol production by 110.5-fold
compared with the expression of 4CL–STS, and 1.7-fold com-
pared with the expression of 4CL–(GGGGS)4–STS (Fig. 5).
Resveratrol production decreased with an increasing number
of GSG or GGGGS linker repeats in both S. cerevisiae and E. coli,
consistent with the in vitro results.
In vivo assays of fusion proteins
In vivo production of resveratrol in the engineered S. cerevisiae
and E. coli strains expressing 4CL–(GSG)n–STS (n r 5) and
4CL–(GGGGS)n–STS (n r 4) was investigated, respectively.
S. cerevisiae WAT11 and E. coli Rosetta strains expressing 4CL
or STS alone were used as controls, and resveratrol was not
detected (ESI,† Fig. S1). In contrast, a peak with the same
retention time and UV spectrum as the authentic resveratrol
standard was detected in both the transgenic S. cerevisiae and
E. coli strains when 4-coumaric acid was added to the culture.
The identity of the resveratrol product was confirmed by LC-MS
using the resveratrol standard (ESI,† Fig. S1).
In S. cerevisiae, the expression of 4CL–STS resulted in the
lowest resveratrol production (0.09 mg mlÀ1), which was 10.4-fold
lower than that obtained by separate expression of 4CL and STS
(Fig. 4). The highest resveratrol production (12.2 mg mlÀ1) achieved
in S. cerevisiae was produced by expressing 4CL–(GSG)1–STS, which
was 2.3-fold and 8100-fold higher than that reported previously by
Zhang et al.18 and Becker et al.,23 respectively. Resveratrol production
Fig. 2 Effects of incubation time and 4-coumaric acid concentration on
the yield of resveratrol from an engineered S. cerevisiae strain expressing
the 4CL–(GSG)1–STS fusion protein. The error bar represents the standard
deviation of three biological replicates.
Fig. 4 Production of resveratrol by engineered S. cerevisiae and E. coli
strains expressing 4CL–STS, 4CL–(GSG)n–STS (n r 5) or 4CL and STS after
24 h and 48 h of incubation, respectively. Data are the means and standard
deviations of three biological repeats.
This journal is ©The Royal Society of Chemistry 2017
Mol. BioSyst.

View Article Online
Paper
Molecular BioSystems
Homology model analysis of the fusion protein
Homology modeling using SWISS-MODEL with the coordinates
from the crystal structure of At4CL::VvSTS (PDB: 3TSY)29 as a
template was used to evaluate the modeled fusion proteins
At4CL::PcSTS containing variations of GSG and GGGGS. In the
4CL–(GSG)1–STS model (Fig. 7A), two STS monomers form a
homodimer, which leads to a Y-shaped structure as described
previously.29 Amino acids from both the N- and C-terminal
sides, as well as the GSG linker residues themselves, are
involved in two conserved helices in the 4CL and STS struc-
tures, suggesting that the relative positioning of the two fused
molecules is largely dependent on the conformation of the
linker. In the 4CL–(GSG)1–STS model, the GSG linker region is
extended and the 4CL and STS domains lie in close proximity
with each other at a distance of B5 Å, not close enough to form
strong polar interactions such as hydrogen bonds or salt
bridges (Fig. 7B). However, the absence of GSG in 4CL–STS
would cause a steric clash between the two domains due to the
distance between them being decreased by B10 Å, which might
lead to loss of the enzymatic function of one or both enzyme
domains. There is sufficient room between the last 4CL residue
Leu561 and the first STS residue of the (GSG)n (n r 5) or
(GGGGS)n (n r 4) linker (Fig. 7C and D). The alteration to an
extended linker conformation may result in an increased
physical distance between the two active sites (Fig. 7E). We
speculated that the increased physical distance between active
sites with increasing linker length may reduce catalytic effi-
ciency, which was consistent with the in vitro and in vivo results.
Fig. 5 Production of resveratrol by engineered S. cerevisiae and E. coli
strains expressing 4CL–STS, 4CL–(GGGGS)n–STS (n r 4) or 4CL and STS
after 24 h and 48 h of incubation, respectively. Data are the means and
standard deviations of three independent determinations.
Anti-STS Western blotting was carried out in the engineered
S. cerevisiae to check the integrity of the fusion proteins. As
shown in Fig. 6, for the nine fusion proteins, there was no
detectable free STS present in the samples, indicating that the
nine fusion proteins had high integrity in engineered S. Cerevisiae
in vivo.
For engineered E. coli, the nine total proteins from the cell
lysate after 48 h incubation were separated by SDS-PAGE (ESI,†
Fig. S2). The amount of total proteins loaded was the same for
each lane. There was no obvious correlation between the band
intensity and the catalytic efficiency of the fusion proteins with
different linker lengths. Thus, the possibility that increased
linker length can induce enhanced proteolysis of fusion proteins
in engineered E. coli in vivo can be excluded.
Discussion
Fusion proteins have a wide range of applications in metabolic
engineering,18,38,39 biological research,6,40–43 synthetic biology,44,45
and as biopharmaceuticals.46–49 Direct fusion of function domains
without a linker may give rise to many undesirable outcomes such
as misfolding of the fusion protein,50 impaired bioactivity,51,52 low
yield in protein production,10 or even termination of bioactivity.53
The linker is believed to be a deciding factor for successful fusion
construction.
In this work, the catalytic efficiency of the 4CL–STS fusion
can be greatly improved by inserting optimized peptide linkers
between the moieties. The short linker (GSG)1 was proved to
be best for fusing 4CL and STS among the tested peptides, it
was excellent to link the 4CL and STS at a reasonable distance
for beneficial interaction and thus for enhanced bifunctional
activities. The catalytic efficiencies of the fusion proteins
decreased gradually with the increasing repeat times of either
the GSG or GGGGS linker, which was also observed between the
two types of linkers. We speculate that the tendency in the
result might be related to the proximity effect. This proximity
effect might increase the reaction rate by facilitating transfer of
the product of the first reaction to the active site of the second
enzyme, thus preventing loss of the intermediate by either
diffusion or competition with other pathways. In addition, it
is likely that unprotected and flexible longer linker regions are
Fig. 6 Western blot analysis of the integrity of the fusion proteins using
anti-STS antibodies. Lane 1: 4CL–(GSG)1–STS, Lane 2: 4CL–(GSG)2–STS,
Lane 3: 4CL–(GSG)3–STS, Lane 4: 4CL–(GSG)4–STS, Lane 5: 4CL–(GSG)5–
STS, Lane 6: 4CL–(GGGGS)1–STS, Lane 7: 4CL–(GGGGS)2–STS, Lane 8:
4CL–(GGGGS)3–STS, Lane 9: 4CL–(GGGGS)4–STS. Data shown are from
one representative experiment.
Mol. BioSyst.
This journal is © The Royal Society of Chemistry 2017

View Article Online
Molecular BioSystems
Paper
Fig. 7 Structure-based homology models of the 4CL–linker–STS fusion protein. (A) Overall structure of the 4CL–(GSG)1–STS model. Monomer 1 and 2 are
colored red and green, respectively. Catalytic sites are shown in stick representation and highlighted with black boxes. (B) Detailed structure of the linker
region of 4CL–(GSG)1–STS. The amino acids between adjacent 4CL and STS domains are shown in stick representation with distances displayed. (C) The
conformation of the (GSG)n (n r 5) linkers are shown in red, blue, yellow, pink and cyan, respectively, for linkers with three, six, nine, twelve or fifteen amino
acids. (D) The conformation of the (GGGGS)n (n r 4) linkers are shown in red, violet, orange and marine, respectively, for linkers with five, ten, fifteen or
twenty amino acids. (E) Superimposition of 4CL–(GSG)1–STS and the structure with a fully extended (GSG)5 linker conformation. Monomers are colored red
and green for 4CL–(GSG)1–STS and grey for 4CL–(GSG)5–STS. Catalytic sites are shown in stick representation and highlighted with black boxes.
often susceptible to proteolytic cleavage during recombinant results excluded the possibility that increased linker length
protein production, especially in eukaryotic expression systems, could induce enhanced proteolysis on the fusion proteins
which have more complex protease systems. Western blotting in vivo. Therefore, the reduced catalytic efficiency of the fusion
This journal is ©The Royal Society of Chemistry 2017
Mol. BioSyst.

View Article Online
Paper
Molecular BioSystems
proteins is most likely due to the proximity effect rather than the National Science and Technology Program of China during the
cleavability of the linker. 25th-Year Plan period (No. 2012AA02A704), Foundation of
To further explain this trend, we constructed homology models Beijing Municipal Education Committee (No. KM201310020015
of two fusion protein series, 4CL–(GSG)n–STS (n r 5) and 4CL– and KM201410020001), Beijing Municipal Education Commission
(GGGGS)n–STS (n r 4) using coordinates from the crystal structure (CEFF-PXM2016_014207_000038), Beijing outstanding talent
of At4CL–GSG–VvSTS (PDB: 3TSY).29 The linker regions in the two project (No. 2013D05021000003) and Beijing Municipal Natural
fusion protein series are extended and no strong polar interactions Science Foundation (No. 2164059).
such as hydrogen bonds or salt bridges exist between the 4CL and
STS domains. Physical distance between active sites may be the only
factor affecting fusion protein 4CL–linker–STS activity. A longer
linker would increase the physical distance between both enzyme
Notes and references
´ ´
1 F. Amaro, A. P. Turkewitz, A. Martın-Gonzalez and J. C.
active sites which may lead to inadequate interaction between the
active centers and reduce catalytic efficiency. Whereas, in fusions of
Glu and Xyl, (GGGGS)2 was best for linking the Glu and Xyl among
the tested peptides (GGGGS)n (n r 3).8 This may be due to the
presence of polar interactions such as hydrogen bonds or salt
bridges between the Glu and Xyl domains. In addition to physical
distance, polar interactions may also affect the Glu–linker–Xyl
activity. The structure-based homology modeling analysis provided
an aid to speculate and explain the effects observed in the in vitro
and in vivo assays.
The analysis of homology-based 3D structures of protein is a
powerful method for understanding the structure–activity rela-
tionship of proteins.54–56 An increasing number of proteins
have been engineered for metabolic and pathway engineering
by structure-guided rational design.57–60 The homology model-
ing could be an aid to the preliminary analysis of the structure
of fusion proteins, which is fine to get a rough idea of a protein.
Structural prediction can facilitate the structure-based rational
design of fusion proteins. For sound prediction and analysis,
other modeling software such as YASARA, AMBER software or
MODELLER should be used. This modeling software would be
conducive to the rational design of fusion protein linkers.
´
Gutierrez, BioMetals, 2014, 27, 195–205.
2 X. Chen, H. Liu, T. Zhang, Y. Liu, X. Xie, Z. Wang and X. Xu,
PLoS One, 2014, 9, e95448.
3 D. M. Czajkowsky, J. Hu, Z. Shao and R. J. Pleass, EMBO Mol.
Med., 2012, 4, 1015–1028.
4 M. R. Bell, M. J. Engleka, A. Malik and J. E. Strickler, Protein
Sci., 2013, 22, 1466–1477.
5 S. T. Reddy, X. Ge, A. E. Miklos, R. A. Hughes, S. H. Kang,
K. H. Hoi, C. Chrysostomou, S. P. Hunicke-Smith, B. L.
Iverson, P. W. Tucker, A. D. Ellington and G. Georgiou,
Nat. Biotechnol., 2010, 28, 965–969.
6 L. M. Bergeron, L. Gomez, T. A. Whitehead and D. S. Clark,
Biotechnol. Bioeng., 2009, 102, 1316–1322.
7 S. Werner, S. Marillonnet, G. Hause, V. Klimyuk and
Y. Gleba, Proc. Natl. Acad. Sci. U. S. A., 2006, 103,
17678–17683.
8 P. Lu and M. G. Feng, Appl. Microbiol. Biotechnol., 2008, 79,
579–587.
9 X. Chen, Y. Bai, J. L. Zaro and W. C. Shen, Biotechniques,
2010, 49, 513–518.
10 N. Amet, H. F. Lee and W. C. Shen, Pharm. Res., 2009, 26,
523–528.
11 N. Amet, W. Wang and W. C. Shen, J. Controlled Release,
2010, 141, 177–182.
12 H. L. Zhao, C. Xue, J. L. Du, M. Ren, S. Xia and Z. M. Liu,
Mol. Pharmaceutics, 2012, 9, 664–670.
13 T. Weimer, W. Wormsbacher, U. Kronthaler, W. Lang,
U. Liebing and S. Schulte, Thromb. Haemostasis, 2008, 99,
659–667.
14 X. Chen, J. L. Zaro and W. C. Shen, Adv. Drug Delivery Rev.,
2013, 65, 1357–1369.
15 K. Yu, C. Liu, B. G. Kim and D. Y. Lee, Biotechnol. Adv., 2015,
33, 155–164.
16 D. Shan, O. W. Press, T. T. Tsu, M. S. Hayden and J. A.
Ledbetter, J. Immunol., 1999, 162, 6589–6595.
17 M. Sabourin, C. T. Tuzon, T. S. Fisher and V. A. Zakian,
Yeast, 2007, 24, 39–45.
18 Y. Zhang, S. Z. Li, J. Li;, X. Pan, R. E. Cahoon, J. G. Jaworski,
X. Wang, J. M. Jez, F. Chen and O. Yu, J. Am. Chem. Soc.,
2006, 128, 13030–13031.
Conclusions
In this work, different 4CL–linker–STS fusion proteins with two
types of amino acids in various length sizes between the 4CL and
STS were evaluated for their catalytic efficiency by either purified
protein-based in vitro assays or in vivo metabolic engineering in
S. cerevisiae and E. coli cells. Our results experimentally confirmed
that flexible linker length has a direct effect on the catalytic efficiency
of fusion proteins. The performance of the 4CL and STS fusion
protein can be greatly improved by inserting optimized flexible
linkers between the moieties. The catalytic efficiencies of the fusion
proteins decreased gradually with the increasing of flexible linker
length. GSG was proved to be a better linker than GGGGS for fusing
4CL and STS enzymes. The structures of the fusion proteins were
preliminarily speculated using a structure-based homology model,
which is helpful for designing linkers for unnatural fusion proteins.
19 R. Trinh, B. Gurbaxani, S. L. Morrison and M. Seyfzadeh,
Mol. Immunol., 2004, 40, 717–722.
Acknowledgements
This work was supported by the National Natural Science 20 J. S. Klein, S. Jiang, R. P. Galimidi, J. R. Keeffe and
Foundation of China (No. 31300620, 31370674 and 21606020),
P. J. Bjorkman, Protein Eng., Des. Sel., 2014, 27, 325–330.
Mol. BioSyst.
This journal is © The Royal Society of Chemistry 2017

View Article Online
Molecular BioSystems
Paper
21 P. Jeandet, B. Delaunois, A. Aziz, D. Donnez, Y. Vasserot, 41 K. Y. Choi, E. O. Jung, D. H. Jung, B. P. Pandey, N. Lee,
S. Cordelier and E. Courot, J. Biomed. Biotechnol., 2012,
2012, 579089.
H. Yun, H. Y. Park and B. G. Kim, Biotechnol. Bioeng., 2012,
109, 630–636.
22 Y. Z. Mei, R. X. Liu, D. P. Wang, X. Wang and C. C. Dai, 42 A. Sokolowska-Wedzina, A. Borek, J. Chudzian, P. Jakimowicz,
Biotechnol. Lett., 2015, 37, 9–18.
23 J. V. Becker, G. O. Armstrong, M. J. van der Merwe, M. G.
M. Zakrzewska and J. Otlewski, Protein Expression Purif., 2014,
99, 50–57.
Lambrechts, M. A. Vivier and I. S. Pretorius, FEMS Yeast Res., 43 C. E. Hagemeyer, C. von Zur Muhlen, D. von Elverfeldt and
2003, 4, 79–85. K. Peter, Thromb. Haemostasis, 2009, 101(6), 1012–1019.
24 S. Y. Shin, N. S. Han, Y. C. Park, M. D. Kim and J. H. Seo, 44 R. Gru¨nberg and L. Serrano, Nucleic Acids Res., 2010, 38(8),
Enzyme Microb. Technol., 2011, 48, 48–53. 2663–2675.
25 J. Beekwilder, R. Wolswinkel, H. Jonker, R. Hall, C. H. de Vos 45 K. Y. Choi, E. Jung, H. Yun, Y. H. Yang and B. G. Kim,
and A. Bovy, Appl. Environ. Microbiol., 2006, 72, 5670–5672. Appl. Microbiol. Biotechnol., 2014, 98, 8191–8200.
26 Y. Wang, C. Halls, J. Zhang, M. Matsuno, Y. Zhang and 46 A. Beck and J. M. Reichert, mAbs, 2011, 3, 415–416.
O. Yu, Metab. Eng., 2011, 13, 455–463.
27 M. Li, K. R. Kildegaard, Y. Chen, A. Rodriguez, I. Borodina
and J. Nielsen, Metab. Eng., 2015, 32, 1–11.
47 X. Chen, J. L. Zaro and W. C. Shen, Expert Opin. Drug Metab.
Toxicol., 2012, 8, 581–595.
48 B. Leader, Q. J. Baca and D. E. Golan, Nat. Rev. Drug Discovery,
2008, 7, 21–39.
28 Y. Wang and O. Yu, J. Biotechnol., 2012, 157, 258–260.
29 Y. Wang, H. Yi, M. Wang, O. Yu and J. M. Jez, J. Am. Chem. 49 G. Molineux, Ann. N. Y. Acad. Sci., 2011, 1222, 55–63.
Soc., 2011, 133, 20684–20687.
30 D. Lee, M. Ellard, L. A. Wanner, K. R. Davis and C. J.
Douglas, Plant Mol. Biol., 1995, 28, 871–884.
31 Y. Guo, H. Guo, X. Li, L. Huang, B. Zhang, X. Pang, B. Liu, L. Ma
and Y. Wang, Plant Biotechnology Reports, 2013, 7, 371–381.
50 H. L. Zhao, X. Q. Yao, C. Xue, Y. Wang, X. H. Xiong and
Z. M. Liu, Protein Expression Purif., 2008, 61, 73–77.
51 Y. Bai, D. K. Ann and W. C. Shen, Proc. Natl. Acad. Sci. U. S. A.,
2005, 102, 7292–7296.
52 Y. Bai and W. C. Shen, Pharm. Res., 2006, 23, 2116–2121.
32 M. Brodelius, A. Lundgren, P. Mercke and P. E. Brodelius, 53 M. Ikebe, T. Kambara, W. F. Stafford, M. Sata, E. Katayama
Eur. J. Biochem., 2002, 269, 3570–3577. and R. Ikebe, J. Biol. Chem., 1998, 273, 17702–17707.
33 R. M. Horton, H. D. Hun, S. N. Ho, J. K. Pullen and L. R. 54 T. M. Cheng, T. L. Blundell and J. Fernandez-Recio,
Pease, Gene, 1989, 77, 61–68. BMC Bioinf., 2008, 9, 441.
34 R. D. Gietz and R. A. Woods, Methods Enzymol., 2002, 350, 87–96. 55 A. M. Wollacott, A. Zanghellini, P. Murphy and D. Baker,
35 L. Ralston, S. Subramanian, M. Matsuno and O. Yu, Plant
Physiol., 2005, 137, 1375–1388.
36 K. T. Watts, P. C. Lee and C. Schmidt-Dannert, ChemBioChem,
2004, 5, 500–507.
37 K. T. Watts, P. C. Lee and C. Schmidt-Dannert, BMC Biotechnol.,
2006, 6, 22.
38 M. Rizk, G. Antranikian and S. Elleuche, Biochem. Biophys.
Res. Commun., 2012, 428, 1–5.
Protein Sci., 2007, 16, 165–175.
56 S. Lorenz, P. Deng, O. Hantschel, G. Superti-Furga and
J. Kuriyan, Biochem. J., 2015, 468, 283–291.
57 Y. Yoshikuni, J. A. Dietrich, F. F. Nowroozi, P. C. Babbitt and
J. D. Keasling, Chem. Biol., 2008, 15, 607–618.
58 D. T. Eriksen, J. Lian and H. Zhao, J. Struct. Biol., 2014, 185,
234–242.
59 Y. S. Cheng, C. C. Chen, J. W. Huang, T. P. Ko, Z. Huang and
R. T. Guo, Appl. Microbiol. Biotechnol., 2015, 99, 9503–9510.
39 C. Formighieri and A. Melis, Metab. Eng., 2015, 32, 116–124.
´
´
40 L. Iturrate, I. Sanchez-Moreno, I. Oroz-Guinea, J. Perez-Gil 60 L. Han, P. Liu, J. Sun, Y. Wu, Y. Zhang, W. Chen, J. Lin,
´
and E. Garcıa-Junceda, Chemistry, 2010, 16, 4018–4030.
Q. Wang and Y. Ma, Sci. Rep., 2015, 5, 13435.
This journal is ©The Royal Society of Chemistry 2017
Mol. BioSyst.