Synthesis and accumulation of aromatic aldehydes in an engineered strain of escherichia coli

Article abstract of DOI:10.1021/ja506664a

Aromatic aldehydes are useful in numerous applications, especially as flavors, fragrances, and pharmaceutical precursors. However, microbial synthesis of aldehydes is hindered by rapid, endogenous, and redundant conversion of aldehydes to their corresponding alcohols. We report the construction of an Escherichia coli K-12 MG1655 strain with reduced aromatic aldehyde reduction (RARE) that serves as a platform for aromatic aldehyde biosynthesis. Six genes with reported activity on the model substrate benzaldehyde were rationally targeted for deletion: three genes that encode aldo-keto reductases and three genes that encode alcohol dehydrogenases. Upon expression of a recombinant carboxylic acid reductase in the RARE strain and addition of benzoate during growth, benzaldehyde remained in the culture after 24 h, with less than 12% conversion of benzaldehyde to benzyl alcohol. Although individual overexpression results demonstrated that all six genes could contribute to benzaldehyde reduction in vivo, additional experiments featuring subset deletion strains revealed that two of the gene deletions were dispensable under the conditions tested. The engineered strain was next investigated for the production of vanillin from vanillate and succeeded in preventing formation of the byproduct vanillyl alcohol. A pathway for the biosynthesis of vanillin directly from glucose was introduced and resulted in a 55-fold improvement in vanillin titer when using the RARE strain versus the wild-type strain. Finally, synthesis of the chiral pharmaceutical intermediate l-phenylacetylcarbinol (l-PAC) was demonstrated from benzaldehyde and glucose upon expression of a recombinant mutant pyruvate decarboxylase in the RARE strain. Beyond allowing accumulation of aromatic aldehydes as end products in E. coli, the RARE strain expands the classes of chemicals that can be produced microbially via aldehyde intermediates.

Full text of DOI:10.1021/ja506664a

Article
pubs.acs.org/JACS
Synthesis and Accumulation of Aromatic Aldehydes in an
Engineered Strain of Escherichia coli
Aditya M. Kunjapur,†,‡ Yekaterina Tarasova, and Kristala L. J. Prather*,†,‡,§
‡,§
†
‡
§
Department of Chemical Engineering, Synthetic Biology Engineering Research Center (SynBERC), Microbiology Graduate
Program, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, United States
*
S Supporting Information
ABSTRACT: Aromatic aldehydes are useful in numerous
applications, especially as flavors, fragrances, and pharmaceut-
ical precursors. However, microbial synthesis of aldehydes is
hindered by rapid, endogenous, and redundant conversion of
aldehydes to their corresponding alcohols. We report the
construction of an Escherichia coli K-12 MG1655 strain with
reduced aromatic aldehyde reduction (RARE) that serves as a
platform for aromatic aldehyde biosynthesis. Six genes with
reported activity on the model substrate benzaldehyde were
rationally targeted for deletion: three genes that encode aldo-
keto reductases and three genes that encode alcohol dehydrogenases. Upon expression of a recombinant carboxylic acid
reductase in the RARE strain and addition of benzoate during growth, benzaldehyde remained in the culture after 24 h, with less
than 12% conversion of benzaldehyde to benzyl alcohol. Although individual overexpression results demonstrated that all six
genes could contribute to benzaldehyde reduction in vivo, additional experiments featuring subset deletion strains revealed that
two of the gene deletions were dispensable under the conditions tested. The engineered strain was next investigated for the
production of vanillin from vanillate and succeeded in preventing formation of the byproduct vanillyl alcohol. A pathway for the
biosynthesis of vanillin directly from glucose was introduced and resulted in a 55-fold improvement in vanillin titer when using
the RARE strain versus the wild-type strain. Finally, synthesis of the chiral pharmaceutical intermediate L-phenylacetylcarbinol (L-
PAC) was demonstrated from benzaldehyde and glucose upon expression of a recombinant mutant pyruvate decarboxylase in the
RARE strain. Beyond allowing accumulation of aromatic aldehydes as end products in E. coli, the RARE strain expands the classes
of chemicals that can be produced microbially via aldehyde intermediates.
INTRODUCTION
The primary barrier to overproduction of aromatic aldehydes
in engineered microorganisms is the rapid conversion of
desired aldehydes into undesired alcohols by numerous
■
Aromatic aldehydes are used widely in industrial applications,
such as in flavors and fragrances.
primary ingredient in vanilla extract,
which is used to create almond and cherry flavors,
examples of aromatic aldehydes and are the two most widely
used flavoring agents in food products. Biotechnological
production of vanillin is of special interest given that less than
1
−5
Vanillin, which is the
17−21
6
−8
endogenous enzymes.
Previous work aimed at enabling
and benzaldehyde,
biotechnological production of vanillin and L-PAC illustrates
how long this obstacle has been recognized. Sixteen years have
passed since a pathway from glucose to an immediate precursor
9
−11
are
2
,5
22
of vanillin was constructed in E. coli. In the same report, an
argument was made that commercial relevance depended on
engineering the ability to synthesize and accumulate vanillin in
1
% of the 16 000 tons of vanillin sold annually originates from
1
2
vanilla beans. Due to the limited supply and high price of
extract from the vanilla bean, most of the market consists of
vanillin that is chemically synthesized from either lignin or
22
a single microbe. However, this feat has not been reported to
date, prompting some researchers to search for organisms with
intrinsically lower reductase activity on aromatic aldehydes,
1
2
18
petroleum. Consumer preference for flavors in which no
such as Schizosaccharomyces pombe. Similarly, L-PAC bio-
transformations performed by the addition of exogenous
benzaldehyde to yeast cultures suffer from low yields, largely
due to benzyl alcohol byproduct formation. The use of purified
enzymes to achieve higher yields has been explored but is likely
4
chemicals are used (i.e., natural flavors ) has led to a price
differential of $1,200-$4,000/kg for natural vanillin compared
to $15/kg for artificial vanillin. Beyond their uses in flavoring,
these and other aromatic aldehydes serve as intermediates for
12
1
2,13
14
the production of industrial chemicals and pharmaceuticals.
to increase production costs. Alternatively, others have
For example, benzaldehyde is used to produce L-phenyl-
acetylcarbinol (L-PAC, also commonly referred to in the
literature as (R)-PAC), which is a precursor to the
decreased benzyl alcohol formation by adding benzaldehyde
Received: February 28, 2014
Published: July 30, 2014
14−16
pharmaceutical ephedrine.
©
2014 American Chemical Society
11644
dx.doi.org/10.1021/ja506664a | J. Am. Chem. Soc. 2014, 136, 11644−11654

Journal of the American Chemical Society
Article
to resting E. coli cells that express a recombinant pyruvate
decarboxylase (PDC) from Zymomonas mobilis.
amplified from genomic DNA (ATCC 39676). The gene encoding the
15
mutant PDC_E473Q enzyme was generated using Polymerase
3
6
Incomplete Primer Extension (PIPE) cloning. All codon-optimized
and mutant gene sequences are included in SI text. E. coli AKR and
ADH gene targets were amplified from MG1655(DE3) genomic DNA
using PCR and cloned into the Duet vector system (Novagen) using
PIPE cloning. Unless otherwise specified, all other genes were cloned
into the Duet vector system (Novagen) using restriction digest-based
cloning (see SI text for more details). Restriction enzymes and T4
DNA ligase were purchased from New England Biolabs. Propagated
constructs were purified using a QIAprep Miniprep Kit (Qiagen), and
agarose gel fragments were purified using a Zymoclean Gel DNA
Recovery Kit (Zymo Research). All constructs were confirmed to be
correct by nucleotide sequencing (Genewiz).
Chemicals. The following compounds were purchased from
Sigma: sodium benzoate, benzaldehyde, benzyl alcohol, vanillic acid,
vanillin, isovanillin, vanillyl alcohol, 3,4-dihydroxybenzoic acid, 3,4-
dihydroxybenzaldehyde, and dimethyl sulfoxide (DMSO). 3,4-
Dihydroxybenzyl alcohol was purchased from TCI America. L-
Phenylacetylcarbinol was purchased from Santa Cruz Biotechnology.
Isopropyl β-D-1-thiogalactopyranoside (IPTG) was purchased from
Denville Scientific. Ampicillin sodium salt, chloramphenicol, and
kanamycin sulfate were purchased from Affymetrix.
Several factors motivate the selection of E. coli for
identification and deletion of genes that encode aromatic
aldehyde reductases. The superior tools and know-how
established with this organism enable rapid evaluation of gene
targets. Strategies to increase flux from central metabolism to
the biosynthesis of aromatics in E. coli are well-docu-
2
3−26
mented
and thus expected to facilitate swift improvements
in aromatic aldehyde production if their accumulation were
realizable. Furthermore, E. coli K-12 strains have been used to
produce food additives designated as GRAS (Generally
27
Recognized As Safe), such as chymosin. Knowledge of the
E. coli genes responsible for aromatic aldehyde reductase
activity can also help inform efforts to engineer other microbial
hosts.
We began this endeavor by exploring serial deletions of
different combinations of genes that were reported to act on
benzaldehyde, which is structurally the simplest aromatic
aldehyde. Our search initially focused on aldo-keto reductases
(AKRs), which form a superfamily of enzymes that have broad
Culture Conditions. With the exception of growth rate experi-
ments, all experiments were performed in 50 mL Pyrex VISTA screw-
cap culture tubes (Sigma), which contained 5 mL of culture in order to
maintain aerobic conditions and limit evaporation of volatile
metabolites. Experimental cultures were initiated using 1% (v/v)
inoculum volumes of overnight culture that were transferred into
either LB medium or M9 minimal medium containing 1.2% glucose,
incubated at 30 °C, and agitated at 250 rpm. Overnight cultures were
grown in 3 mL of the same medium in 14 mL round-bottom tubes
substrate specificity and convert aldehydes and ketones to
28
alcohols in the presence of NADPH. In the few previously
reported attempts at constructing microbial pathways to
produce aldehydes, only genes encoding alcohol dehydro-
1
8,19
genases (ADHs) were targeted for deletion.
Unlike AKRs,
ADHs typically use NADH as their cofactor, and under
anaerobic conditions they perform the important function of
recycling cofactors. We hypothesized that rational deletion of
AKRs with activity on benzaldehyde, in addition to the deletion
of select ADHs, would be a promising route to engineering the
accumulation of aromatic aldehydes in E. coli.
We report the construction of an E. coli MG1655 strain with
reduced aromatic aldehyde reduction (RARE) that can serve as
a platform for the synthesis of aromatic aldehydes with minimal
or no conversion to their corresponding alcohols. We then use
the RARE strain to enable the synthesis of vanillin and L-PAC,
demonstrating the utility of this particular engineered strain and
the general approach of rationally combining gene knockouts to
overcome a highly redundant endogenous activity.
(Corning). In general, experiments were performed in triplicates, and
results are presented as averages with error bars representing one
standard deviation.
For all substrate-feeding experiments excluding glucose as a
substrate, filter-sterilized and pH-neutralized substrates were added
to the cultures upon induction with IPTG. During 5 mM benzoate-
feeding experiments, cultures were induced with 1 mM IPTG between
optical densities (OD600) of 0.7−1.0. During 5 mM vanillate-feeding
experiments, cultures were induced within an OD600 range of 1.0−1.3.
Culture medium was supplemented with 50 mg/L ampicillin and 17
mg/L chloramphenicol to provide selective pressure for plasmid
maintenance.
For experiments testing the production of vanillin directly from
glucose, 1.2% (w/v) glucose was added prior to inoculation and
cultures were induced with 0.5 mM IPTG between an OD600 of 0.8
and 1.1. Screw-caps remained tightly closed until sampling at final time
points, but mass balances in the liquid phase did not always close,
indicating some evaporation of volatile products. Either LB medium or
M9 minimal medium was used. Culture medium was supplemented
with 50 mg/L ampicillin and 17 mg/L chloramphenicol to provide
selective pressure for plasmid maintenance.
For experiments testing the synthesis of L-phenylacetylcarbinol,
1.2% (w/v) glucose was added prior to inoculation and cultures were
induced with 0.5 mM IPTG between an OD600 of 0.8 and 1.1. At
induction, a filter-sterilized solution of 50 mM benzaldehyde in 10%
DMSO was added to the cultures, resulting in an initial concentration
of 5 mM benzaldehyde and 1% DMSO. For these experiments, culture
medium was supplemented with either only 50 mg/L kanamycin or 25
mg/L kanamycin and 17 mg/L chloramphenicol depending on
whether strains were expressing Car.
Growth rates for wild-type and RARE strains were determined from
50 mL shake flask cultures. Cultures were initiated with 1% inoculum
volumes of overnight culture that were transferred into either LB
medium containing 1.2% glucose or M9 minimal medium containing
1.2% glucose. Overnight cultures were grown in the same media. In
both cases, cultures were incubated at 30 °C and agitated at 250 rpm.
The OD600 was measured regularly during exponential growth using a
DU800 UV/vis spectrophotometer (Beckman Coulter).
MATERIALS AND METHODS
Strains and Plasmids. E. coli strains and plasmids used in this
study are listed in Table S1 (Supporting Information (SI)). Molecular
biology techniques were performed according to standard practices
unless otherwise stated. Molecular cloning and vector propagation
were performed in DH5α. All targeted genes were deleted from E. coli
K-12 MG1655(DE3). The genes dkgB, yeaE, yahK, yjgB, endA, and
■
2
9
3
0
recA were deleted using donor strains from the Keio collection and
31
P1 transduction. P1 bacteriophage was obtained from ATCC
25404-B1). The operon encoding yqhC-dkgA was deleted using the
(
32
λ Red system. To generate homology, three pairs of oligonucleotides
were used as PCR primers. These and other oligonucleotides are
shown in Table S2. Oligonucleotides were purchased from Sigma. Q5
High Fidelity DNA Polymerase (New England Biolabs) was used for
DNA amplification. In all cases of gene deletions, pCP20 was used to
3
2
cure the kanamycin resistance cassette.
3
3
The car gene from Nocardia iowensis, the sfp gene from Bacillus
subtilis, and the Hs-S-COMT gene from Homo sapiens were
synthesized and codon-optimized for expression in E. coli (GenScript).
The asbF gene from Bacillus thuringiensis was synthesized as a DNA
string codon-optimized for expression in E. coli (GeneArt, Invitrogen).
The aroG* gene was kindly provided by Professor Jay D. Keasling at
the University of California, Berkeley (USA). The gene encoding the
pyruvate decarboxylase (PDC) enzyme from Zymomonas mobilis was
34
18
3
5
1
1645
dx.doi.org/10.1021/ja506664a | J. Am. Chem. Soc. 2014, 136, 11644−11654

Journal of the American Chemical Society
Article
Table 1. E. coli Strains Featuring Different Combinations of Gene Deletions (“X” indicates deletion)
Gene
Enzyme Family
MG endA- recA-
Subset 1
Subset 2
Subset 3
Subset 4
RARE endA- recA-
dkgB
yeaE
dkgA
yqhC
yqhD
yahK
yjgB
AKR
X
X
X
X
X
X
X
X
X
X
X
X
X
X
AKR
AKR
X
X
X
X
X
X
X
X
activator
ADH
ADH
ADH
X
X
Metabolite Analysis. Culture samples were pelleted by
centrifugation, and aqueous supernatant was collected for HPLC
analysis using either an Agilent 1100 series or 1200 series instrument
equipped with a diode array detector. Wavelengths of 223, 242, and
by qPCR were performed in triplicate. 2 μL of cDNA from each RNA
isolation were added to Brilliant II SYBR Green High ROX QPCR
Mix (Agilent Technologies), and 0.5 μM of appropriate primers was
added to a final volume of 25 μL per well. Amplification was
performed according to the following program: an initial step of 95 °C
for 10 min, followed by 40 cycles of 95 °C for 30 s and 60 °C for 1
min. The number of cycles to reach the threshold (CT value) was
measured for each primer pair in triplicate samples of each cDNA. The
192 nm were used to detect benzoic acid, benzaldehyde, and benzyl
alcohol, respectively. The benzoate family of analytes was separated
using an Aminex HPX-87H anion-exchange column (Bio-Rad
Laboratories), with a mobile phase consisting of 70% 5 mM H SO4
2
and 30% acetonitrile. All three compounds eluted within 35 min at a
flow rate of 0.4 mL/min. The column temperature was maintained at
“
Auto C ” option in the 7300 System SDS RQ Study Software was
T
used to determine the threshold values. Sequences of primers used for
analysis of dkgA, dkgB, yeaE, yqhD, yahK, and yjgB are listed in Table
S2. Primer sequences were designed to have melting temperatures
ranging from 56 to 60 °C and to generate roughly 100 base pair
amplicons. The specificity of primers was verified using gel
electrophoresis using gDNA from MG1655(DE3) ΔendA ΔrecA as a
template. No-template and no-RT controls confirmed that primer
3
0 °C. All chemicals reported in figures were quantified using
calibration of standards on the HPLC instrument and linear
interpolation.
Compounds used in vanillin experiments were separated using a
Zorbax Eclipse XDB-C18 column (Agilent) and detected using a
wavelength of 280 nm. A gradient method used the following solvents:
(A) 50% acetonitrile + 0.1% trifluoroacetic acid (TFA); (B) water +
dimer formation was absent or negligible (i.e., C values greater than
T
0.1% TFA. The gradient began with 5% Solvent A and 95% Solvent B.
3
3). Plasmids containing gene deletion targets were diluted to
The setting at 20 min was 60% Solvent A and 40% Solvent B. The
program restored the original ratio at 22 min and ended at 25 min.
The flow rate was 1.0 mL/min, and all compounds of interest eluted
within 15 min. The column temperature was maintained at 30 °C.
Phenylacetylcarbinol was detected using a Zorbax Eclipse XDB-C18
column (Agilent) and detected using a wavelength of 210 nm. A
gradient method used the following solvents: (A) 100% acetonitrile +
−4
−8
concentrations ranging from roughly 10 to 10 ng/μL and analyzed
in triplicate as standards during each respective run. Using linear
standard curves, CT values were used to quantify absolute
concentrations of reverse transcribed mRNA corresponding to each
gene of interest, with and without benzaldehyde treatment. All PCR
efficiencies were similar.
0
.1% trifluoroacetic acid (TFA); (B) water + 0.1% TFA. The gradient
began with 5% Solvent A and 95% Solvent B. The setting at 20 min
was 60% Solvent A and 40% Solvent B. The program restored the
original ratio at 22 min and ended at 25 min. The column temperature
was maintained at 30 °C. The flow rate was 1.0 mL/min, and the
retention time of phenylacetylcarbinol was 8.3 min. The peak area
corresponding to a coeluting and static background peak in LB
medium was subtracted in order to quantify concentrations of
phenylacetylcarbinol produced. Although the enantiomeric excess
RESULTS
■
A combination of rationally targeted gene deletions
enables benzaldehyde accumulation in E. coli. Deletions
3
7
of AKR genes were guided by literature reported activities of
gene products on benzaldehyde (Table S3). Protein BLAST
was used to organize E. coli AKRs based on sequence similarity
to DkgA, the AKR with the greatest reported activity on
benzaldehyde (Table S4). The gene encoding DkgA is located
downstream of yqhD in the same operon, and transcription of
both genes is activated by the product of yqhC, which is located
(
ee) was not determined, it has been shown previously using chiral
HPLC and near-UV circular dichroism spectroscopy that both wild-
type PDC and PDC_E473Q catalyze the formation of the R
16
stereoisomer in 98−99% ee.
38
immediately upstream of the operon. Given that YqhD is a
broad substrate ADH that is also reported to act on
Quantitative Reverse Transcription PCR (qRT-PCR). For
isolation of RNA and generation of cDNA in biological duplicate,
two cultures of MG1655(DE3) ΔendA ΔrecA were grown overnight in
3
9−41
benzaldehyde,
the entire operon was deleted. Benzalde-
3
mL of LB medium contained in 14 mL round-bottom tubes. Each
hyde reductase activity was unaffected by the deletion of this
operon and by the subsequent deletion of the two genes
encoding AKRs with the greatest protein sequence similarity to
DkgA (dkgB and yeaE). The activity of two cinnamyl alcohol
dehydrogenases, YahK and YjgB, on benzaldehyde was
overnight culture was used to inoculate two cultures in 4.5 mL of LB
medium (1% v/v inoculum) contained in 50 mL PYREX VISTA tubes.
Once cultures reached an OD600 of 0.7−0.9, benzaldehyde was added
to two out of the four cultures to a final concentration of 5 mM. After
1
h of further incubation, 0.5 mL of cells was harvested for RNA
4
2
reported recently. A few months prior to that report,
overexpression of the yahK and yjgB genes had also been
reported to improve the production of aromatic alcohols in E.
isolation. RNA protect bacterial reagent (Qiagen) was added to cells
prior to centrifugation and lysis. Total RNA was isolated using the
illustra RNAspin Mini Isolation Kit (GE healthcare) with an on-
column DNaseI treatment according to protocol. Turbo DNA-free
reagents (Ambion) were used to further treat isolated RNA for
removal of genomic DNA. Next, QuantiTect Reverse Transcriptase
43
coli. After deleting these two additional genes, we observed a
significant change in the ability of E. coli to accumulate
benzaldehyde. The final engineered strain (RARE ΔendA
ΔrecA) contains nine gene deletions: dkgB, yeaE, yqhC, yqhD,
dkgA, yahK, yjgB, endA, and recA. The genes endA and recA,
which encode an endonuclease and a recombinase, were
deleted to increase plasmid stability. The deletion of yqhC
(
Qiagen) was used to generate cDNA from 500 ng of total RNA for
each isolation. Concentrations of RNA and DNA were measured using
a NanoDrop 2000 (Thermo Scientific).
Quantitative PCR was performed using an ABI 7300 Real Time
PCR System Instrument (Applied Biosystems). All samples analyzed
1
1646
dx.doi.org/10.1021/ja506664a | J. Am. Chem. Soc. 2014, 136, 11644−11654

Journal of the American Chemical Society
Article
Figure 1. A combination of rational gene deletions enables benzaldehyde accumulation in E. coli. (A) Scheme depicting intracellular formation of
benzaldehyde from benzoate and endogenous conversion to the byproduct benzyl alcohol. (B) Conversion of 5 mM benzoate after 24 h in strains
transformed with pETDuet-1 and pACYC-car-sfp. (C) Conversion of 5 mM benzoate after 24 h in RARE strains transformed with pACYC-car-sfp
and a pET plasmid harboring the gene indicated below the x axis.
forms a convenient but nonessential control. The remaining six
genes constitute the rationally targeted set.
experiments, pH-neutralized acid substrates were added to the
medium to obtain greater solubility and mimic product
formation in an engineered pathway. The corresponding
aldehydes were then generated intracellularly by the action of
Car on the acid substrate.
To compare the ability of the RARE strain to accumulate
aromatic aldehydes, four additional strains containing comple-
mentary subsets of the total set of gene deletions were
constructed (Table 1). In short, the “Subset 1” strain contains
the yqhC-yqhD-dkgA operon intact, the “Subset 2” strain
contains all gene targets that were not in the operon, the
Accumulation of benzaldehyde was investigated in the six
different strains (Figure 1A). The wild-type strain and three out
of the four strains containing subsets of gene knockouts
converted all of the supplied 5 mM benzoate into benzyl
alcohol within 24 h, with no accumulation of benzaldehyde. On
the other hand, the RARE strain accumulated 3.3 ± 0.1 mM
benzaldehyde and displayed less than 12% conversion of
benzaldehyde to benzyl alcohol (Figure 1B). Similarly, the
“Subset 4” strain also enabled the accumulation of benzalde-
hyde. The “Subset 4” strain contains the dkgB and yeaE genes
intact in the genome, indicating that the deletions of dkgB and
yeaE were not necessary for benzaldehyde accumulation under
these conditions.
“Subset 3” strain contains only the targeted ADHs that were
not in the operon, and the “Subset 4” strain contains only the
targeted AKRs that were not in the operon. All strains were also
built to express a recombinant and activated carboxylic acid
reductase (Car). Car, from N. iowensis, has broad substrate
specificity and was previously used to catalyze the formation of
benzaldehyde and vanillin from their corresponding acids in
33
vivo in E. coli and in vitro. Expression of Car was paired with
expression of Sfp from B. subtilis. Sfp is a phosphopantetheinyl
3
4,44
transferase that has been shown to activate Car.
In these
1
1647
dx.doi.org/10.1021/ja506664a | J. Am. Chem. Soc. 2014, 136, 11644−11654

Journal of the American Chemical Society
Article
Figure 2. In vanillate-feeding experiments, the RARE strain eliminates conversion of vanillin to vanillyl alcohol. (A) Scheme depicting intracellular
formation of vanillin from vanillate and endogenous conversion to the byproduct vanillyl alcohol. (B) Conversion of 5 mM vanillate after 48 h in
strains transformed with pETDuet-1 and pACYC-car-sfp. (C) Conversion of 5 mM vanillate after 24 h in RARE strains transformed with pACYC-
car-sfp and a pET plasmid harboring the gene indicated below the x axis.
Deleted genes were overexpressed individually alongside car
in the RARE strain to determine whether each gene could
contribute to benzaldehyde reductase activity in vivo (Figure
significance of gene deletions. qRT-PCR results provide further
support for our hypothesis of low levels of baseline expression
of dkgB and yeaE compared to the other targeted genes (Figure
S1). Furthermore, expression of these two genes is not
significantly different in the presence or absence of
benzaldehyde, whereas expression of all other targeted genes
increases upon benzaldehyde addition.
1C). Benzoate was supplied and formation of aldehyde and
alcohol products was monitored as before. Individual over-
expression of each of the six target genes prevented detectable
accumulation of benzaldehyde. Conversely, benzaldehyde
accumulated when the control gene yqhC was overexpressed.
For some strains, lower conversion rates were observed in this
experiment relative to the previous experiment, which may have
been due to lower expression of car in the presence of native
gene overexpression. Interestingly, individual overexpression of
dkgB and yeaE also restored the inability to accumulate
benzaldehyde, even though the deletion strain experiment
demonstrated that these deletions were not required under
these conditions. The results from these two experiments
suggest that native expression of dkgB and yeaE may be minimal
under these conditions and that, in general, overexpression
experiments alone may mislead efforts to determine the
In vanillate-feeding experiments, the RARE strain
eliminates conversion of vanillin into vanillyl alcohol.
We next investigated the utility of the engineered strain by
attempting to produce and accumulate other aromatic
aldehydes. The RARE strain and subset deletion strains were
fed vanillate to assess the effect of the same set of gene
deletions on the undesired conversion of vanillin to vanillyl
alcohol (Figure 2A). Our hypothesis was that a strain that
displays minimal reductase activity on benzaldehyde would also
display low reductase activity on vanillin. Although the probable
physiological substrates of AKRs are smaller aldehydes such as
37
methylglyoxal, benzaldehyde and vanillin are structurally
1
1648
dx.doi.org/10.1021/ja506664a | J. Am. Chem. Soc. 2014, 136, 11644−11654

Journal of the American Chemical Society
Article
Figure 3. The RARE strain enables production of vanillin from glucose in E. coli. (A) Scheme depicting complete pathway from glucose to vanillin
with overexpressed E. coli aroG* indicated in bold typeface [PYR = pyruvate, F6P = fructose 6-phosphate, G3P = glyceraldehyde 3-phosphate, X5P =
xylulose 5-phosphate, PEP = phosphoenolpyruvate, E4P = erythrose 4-phosphate, DAHP = 3-deoxy-D-arabinoheptulosonate 7-phosphate, DHQ = 3-
dehydroquinate, DHS = 3-dehydroshikimate]. (B) Scheme illustrating heterologous portion of pathway with two possible undesired alcohol
byproducts. (C) Concentration profiles of the six heterologous metabolites of interest (including vanillin) when wild-type and RARE hosts are
transformed with pET-OMT-asbF and pACYC-car-sfp-aroG* plasmids and grown in LB + 1.2% glucose for 48 h. (D) Concentration profiles of the
six heterologous metabolites of interest (including vanillin) produced from glucose as a sole carbon source after 60 h in the same strains grown in
M9 + 1.2% glucose.
similar, with vanillin differing only by the presence of additional
hydroxyl and methoxy groups distant from the aldehyde group.
Cultures were supplied with vanillate and compared 48 h
after induction due to the slower kinetics of Car and
endogenous enzymes on vanillate and vanillin relative to
and that overexpression results can mislead gene deletion
efforts. These results also validate that the RARE strain can be
used for the bioconversion of multiple acid substrates into their
corresponding aldehydes.
The RARE strain enables production of vanillin from
glucose in E. coli. To investigate whether the RARE strain
could enable the production of an aromatic aldehyde directly
from glucose, a non-optimized pathway from glucose to vanillin
was assembled in the RARE and wild-type strains. Previous
33
benzoate and benzaldehyde. After 48 h, the RARE and
Subset 4” strains were the only strains that resulted in no
“
detectable formation of vanillyl alcohol (Figure 2B). As before,
in order to investigate whether each gene could contribute to
vanillin reductase activity in vivo, individual overexpression of
deleted genes in the RARE strain was examined for the
presence of the alcohol after 24 h. All gene products were active
on vanillin, but surprisingly the overexpression of targeted
AKRs (DkgB, YeaE, and DkgA) resulted in significantly more
vanillyl alcohol production compared to the overexpression of
targeted ADHs (YahK, YjgB, and YqhD) (Figure 2C). As with
benzaldehyde, these results suggest that dkgB and yeaE
expression may remain minimal in the presence of vanillin
22
reports established a route from glucose to vanillate in E. coli,
and a route featuring the same metabolites but different
18
enzymes was also assembled in yeast. In both cases, the native
branch-point metabolite was 3-dehydroshikimate, which is part
of the aromatic amino acid biosynthesis pathway (Figures 3A
45
and S2). A feedback-resistant form of E. coli aroG (aroG*),
which encodes a 3-deoxy-D-arabinoheptulosonate 7-phosphate
(DAHP) synthase, was included to ensure that flux enters this
endogenous pathway.
1
1649
dx.doi.org/10.1021/ja506664a | J. Am. Chem. Soc. 2014, 136, 11644−11654

Journal of the American Chemical Society
Article
Figure 4. The RARE strain enables the synthesis of L-phenylacetylcarbinol (L-PAC). (A) Scheme depicting the synthesis of L-PAC from the
condensation of exogenously supplied benzaldehyde and metabolized pyruvate, catalyzed by either PDC or PDC_E473Q. (B) Concentration
profiles of PAC, benzaldehyde, and benzaldehyde oxidation/reduction products 24 h after addition of 5 mM benzaldehyde to RARE and wild-type
host strains. (C) Time course of benzaldehyde reduction using the wild-type host transformed with pRSF/PDC_E473Q and pACYC/Car/Sfp.
The heterologous pathway constructed for our experiments
(0.014 ± 0.001 mM). Conversely, the dominant products made
by the RARE strain are vanillin and its precursor aldehyde,
protocatechualdehyde. No detectable protocatechuic alcohol
formed, revealing that the RARE host strain is capable of
accumulating at least three different aromatic aldehydes
(benzaldehyde, vanillin, and protocatechualdehyde). Although
some vanillyl alcohol was detected, it represents less than 14%
conversion of vanillin formed. Using the RARE strain, the
average vanillin titer was 0.78 ± 0.02 mM (119 ± 3 mg/L),
representing more than a 55-fold increase in production over
the wild-type strain.
Finally, the same strains were cultivated in M9 minimal
medium + 1.2% glucose instead of LB medium + 1.2% glucose
to determine whether vanillin could be produced from glucose
as a sole carbon source. Concentrations of the six metabolites
of interest demonstrate that the RARE strain enables the
accumulation of vanillin in minimal medium (Figure 3D).
Under these conditions, vanillin is the dominant product of the
six metabolites of interest even after 60 h. Compared to results
obtained from growth in LB medium, the ratios of proto-
catechuate to protocatechualdehyde and vanillate to vanillin
sharply increase in M9 minimal medium. Although the buildup
of protocatechualdehyde in the cultures grown in LB indicated
35
consists of three genes: asbF from B. thuringiensis, Hs-S-
1
8,46
COMT from H. sapiens,
and car (Figure 3B). Together with
aroG* and sfp, a total of five genes were overexpressed. The
asbF gene encodes a 3-dehydroshikimate dehydrogenase, which
efficiently converts 3-dehydroshikimate into protocatechuate.
The Hs-S-COMT gene encodes a soluble O-methyltransferase
(
OMT) that has activity on catechols and related compounds.
Depending on the relative enzyme kinetics and availability of
cofactors, protocatechuate can either be converted into
protocatechualdehyde by Car or be converted into vanillate
by OMT. The final step in the pathway is either the conversion
of protocatechualdehyde to vanillin by OMT or the conversion
of vanillate to vanillin by Car. Because this pathway can lead to
the production of two possible alcohol byproducts (proto-
catechuic alcohol and vanillyl alcohol), we expected vanillin
titers to be greater when using the RARE strain host rather than
the wild-type host.
Figure 3C displays the concentrations of the six metabolites
of interest produced as a result of the vanillin pathway 48 h
after induction in wild-type and RARE strains. As expected, the
dominant products generated using the wild-type host are the
two alcohol byproducts, with minimal formation of vanillin
1
1650
dx.doi.org/10.1021/ja506664a | J. Am. Chem. Soc. 2014, 136, 11644−11654

Journal of the American Chemical Society
Article
that the O-methyltransferase was limiting, the greater pool size
of vanillate relative to protocatechualdehyde suggests that this
enzyme was no longer limiting in M9. Given the documented
complete set of targeted genes and increase the industrial
relevance of the engineered strain.
18
lack of specificity of the O-methyltransferase, the byproduct
isovanillin was also produced in these experiments (Figure S3).
The RARE strain enables the synthesis of L-PAC in E.
coli. We next sought to demonstrate the utility of the RARE
strain as a platform for the biocatalysis of products derived from
aromatic aldehyde intermediates. Given the reactivity of
aldehyde functional groups, there are numerous enzymatic
chemistries that may be enabled by the accumulation of
aromatic aldehydes. We were particularly interested in
carboligations because these reactions are known to yield
chiral products. We hypothesized that use of the RARE host
strain could enhance the synthesis of L-PAC. Wild-type and
RARE strains were transformed to express a recombinant
mutant PDC (PDC_E473Q) with improved kinetic properties
over the wild-type enzyme from Z. mobilis. Specifically,
PDC_E473Q displays inverted partitioning between aldehyde
release and carboligation compared to the wild-type PDC, with
DISCUSSION
■
AKRs are found in organisms ranging from vertebrates to
4
7
archaebacteria and are believed to be responsible for
catalyzing the conversion of methylglyoxal and related reactive
37
metabolites into less toxic compounds. Microbial AKRs
2
8
belong to 10 families: AKR2, AKR3, AKR5, and AKR8−14.
The preference of AKRs for NADPH over NADH suggests that
AKRs may be physiologically relevant under aerobic conditions.
A total of nine open reading frames that encode AKRs in E. coli
48
have been identified using sequence similarity searches.
Unlike many other E. coli AKRs, DkgA has been characterized
and a crystal structure has been resolved. Activity assays
performed in vitro constitute the majority of published data on
E. coli AKRs and demonstrate that most of these gene products
49
37
act on overlapping sets of substrates. Perhaps because of the
known redundancy of AKR activity in E. coli, no one has
previously reported the intentional deletion of AKRs for the
purpose of building up aldehyde pools. To our knowledge,
there are no previous publications even describing the
simultaneous deletion of two or more AKRs in E. coli. In our
case, although we initially focused on AKRs as rational targets
for deletion, we found that two out of the three targeted AKRs
were not contributing to benzaldehyde or vanillin reduction
under the conditions tested. Nevertheless, we also found that all
AKRs were active on these substrates when overexpressed, and
in the case of vanillin, they were significantly more active than
targeted ADHs. From our experiments, it is still unclear what
conditions, if any, may lead to significant expression of dkgB
and yeaE, but there was no reduction in growth rate or other
disadvantage incurred by their deletion.
16
an up to 100-fold preference for carboligation. In our case,
the PDC_E473Q enzyme was expected to catalyze the
condensation of benzaldehyde supplied exogenously and
pyruvate resulting from metabolism of glucose (Figure 4A).
Expression of only PDC_E473Q resulted in the synthesis of
2
.59 ± 0.04 mM PAC in the RARE strain 24 h after
benzaldehyde addition, along with less than 4% reduction of
benzaldehyde to benzyl alcohol (Figure 4B). We also observed
roughly 8% oxidation of benzaldehyde to benzoate. To prevent
net oxidation of benzaldehyde, we expressed both Car and
PDC_E473Q. In this case, no benzoate was detected, and
similar levels of PAC and benzyl alcohol were produced in the
RARE strain. Surprisingly, wild-type strains transformed with
the same constructs resulted in no synthesis of PAC (Figure
Previously, yields of some aldehydes have been increased by
deleting ADH genes responsible for reductase activity in model
organisms such as Saccharomyces cerevisiae and E. coli. ADHs are
4B). This result suggested that the time scale of benzaldehyde
reduction maybe much shorter than 24 h. To investigate this
further, a time course study was performed to monitor the
conversion of benzaldehyde to benzyl alcohol using the wild-
type strain expressing Car and PDC_E473Q (Figure 4C).
Within just 2 h of supplying 5 mM benzaldehyde, all of the
benzaldehyde was either reduced to benzyl alcohol (∼4 mM)
or lost to the headspace (∼1 mM). This result explains why the
wild-type E. coli host strain failed to produce any PAC under
these conditions. The RARE strain enables the synthesis of
PAC in E. coli by extending the duration of benzaldehyde
availability more than 10-fold, to the relevant time scale of
PDC_E473Q kinetics.
5
0−52
generally classified under the short-chain
chain
or medium-
53,54
dehydrogenase/reductase families and, like AKRs, are
known to have broad and redundant substrate specificity.
Specifically, the ADH6 gene in S. cerevisiae was deleted in
another study in order to produce 45 mg/L of vanillin from
18
glucose. However, other genes known to convert aldehydes to
alcohols in S. cerevisiae remained intact in that study and no
further deletions have been reported to our knowledge.
Another report described an effort to improve the production
19
of isobutyraldehyde, which is a bulk chemical feedstock. In
that study, eight genes were deleted (yqhD, adhP, eutG, yiaY,
yjgB, betA, fucO, eutE). Unlike in our study, only five of these
targeted genes (yqhD, adhP, eutG, yiaY, yjgB) were found to be
capable of reducing isobutyraldehyde to isobutanol when
individually overexpressed. Overall, the combination of all
eight deletions resulted in an improvement in the ratio of
isobutyraldehyde to isobutanol produced from 0.14 galdehyde/L/
The deletion of all targeted genes to form the RARE
strain has no effect on growth rate. To test whether the
full set of gene deletions affects cell growth, the RARE and
wild-type strains were grown in LB medium + 1.2% glucose and
in M9 minimal medium + 1.2% glucose. In the absence of any
particular stress, no effect on specific growth rate was observed
under either condition (Figure S4). The average specific growth
OD600 and 1.5 g
/L/OD600 to 1.5 galdehyde/L/OD600 and 0.4
However, the deletion of the yqhD gene
alcohol
1
9
galcohol/L/OD600
.
−1
−1
rates were μ = 1.07 h and μ = 0.37 h in LB and M9,
respectively. Furthermore, final OD600 measurements were
taken in the presence of aldehydes synthesized from experi-
ments featuring the RARE and wild-type strains (Figure S5).
These measurements demonstrate that the presence of vanillin
at concentrations explored in this study does not significantly
affect the growth rate of the RARE strain relative to the wild-
type strain. These results reveal the nonessential nature of the
alone led to a 1:1 ratio of aldehyde to alcohol, indicating a
diminishing marginal return on the deletions of the four other
genes that may contribute to isobutryaldehyde reduction.
Neither of the aforementioned studies described an attempt to
delete genes encoding AKRs nor did they culture their strains
under aerobic conditions.
By rationally evaluating three AKR deletions in combination
with three ADH deletions, we constructed an E. coli host strain
1
1651
dx.doi.org/10.1021/ja506664a | J. Am. Chem. Soc. 2014, 136, 11644−11654

Journal of the American Chemical Society
Article
that displayed a significant step-change in the ability to
accumulate aromatic aldehydes relative to several engineered
strains containing subsets of these deletions. The brute-force
method of rationally targeting and combining several knockouts
to overcome a redundant endogenous activity should become
more accessible given significant advances in genome engineer-
ing and an increasing rate of functional gene annotation.
Techniques that enable prompt construction of rational
combinations of gene knockouts, such as Multiplex Automated
stration of L-PAC synthesis illustrates that aldehydes can be
biologically converted into new chemical classes if they can
avoid the fate of rapid reduction. This previously unattainable
option may be more attractive than harvesting resting cells or
using purified enzymes for many biocatalytic processes.
Furthermore, it opens up the potential for synthesis of end
products directly from glucose as a sole carbon source. Besides
pyruvate decarboxylase, other carboligases have been used in
vitro to produce enantiopure products using benzaldehyde as an
55
6
5
Genome Engineering (MAGE), or gene expression knock-
intermediate, such as benzoins
and 2-hydroxy-
56
66
downs, such as RNA or CRISPR interference, can be
harnessed in future studies like this.
propriphenones. However, as just one enzyme class relying
on a single aldehyde substrate, carboligations that utilize
benzaldehyde offer a small view into the range of microbial
products potentially attainable from aromatic aldehyde
intermediates. For example, aromatic aldehydes can also serve
as efficient amine acceptors, and there are transaminases
reported to catalyze the formation of aromatic primary
However, selection of the correct target genes to eliminate a
highly redundant activity is not straightforward. Kinetic data
from purified enzyme assays may be misleading given that
enzyme activity and gene expression have no correlation but
together affect endogenous activity. Similarly, we have shown
that gene products displaying undesired activity in vivo when
overexpressed may not need to be deleted in order to eliminate
the undesired endogenous activity. A further complication is
that not all of the necessary target genes may be fully
characterized. Current genome modeling approaches for
67
amines.
CONCLUSION
■
Aromatic aldehydes are valuable chemicals, either as end
products or as intermediates. Microbial synthesis of aromatic
aldehydes is ordinarily hindered by rapid, endogenous, and
redundant conversion to alcohol byproducts. We have
constructed an E. coli K-12 MG1655 strain that can accumulate
a variety of industrially relevant aromatic aldehydes, including
benzaldehyde, protocatechualdehyde, and vanillin. This result
was achieved by rationally targeting six genes for deletion,
spanning the aldo-keto reductase and alcohol dehydrogenase
superfamilies. This particular combination of deletions resulted
in a step-change in aromatic aldehyde accumulation, whereas
several strains containing subsets of these deletions did not
deviate significantly from the wild-type strain. In spite of all of
these deletions, the growth rate of the engineered strain was
unperturbed. The engineered strain was used to enable the
synthesis and accumulation of vanillin and L-phenylacetyl-
carbinol. Finally, the use of this engineered strain may enable
microbial synthesis of a variety of other compounds enzymati-
cally derived from aromatic aldehyde intermediates, expanding
the boundaries of metabolic engineering and biocatalysis.
57
determining knockouts such as OptStrain would not predict
these deletions. In fact, OptStrain has already been used in an
attempt to optimize vanillin production in E. coli and did not
57
include any of the aldehyde reductases targeted in our study.
The alternative of transcriptional profiling is limited by the
inability to account for redundant or constitutively expressed
genes, as well as by off-target effects and secondary responses.
5
8,59
Finally, combinatorial approaches to gene knockouts
have
led to the generation of numerous strain improvements but are
unlikely to surmount sufficiently redundant activities, especially
in the absence of an effective selection strategy.
Although the creation of an E. coli strain that has minimal
conversion of vanillin to vanillyl alcohol has been previously
2
2,33,60
sought after,
some groups have reported alternative
strategies to use E. coli for vanillin production. At least one
report documents the use of resting E. coli cells for the
bioconversion of ferulic acid into vanillin with minimal vanillyl
61
alcohol formation under these conditions. Under nongrowing
conditions, it is likely that the AKR and ADH genes identified
in this study are not expressed to a high degree. However, the
use of resting E. coli cells is not a viable strategy for producing
vanillin from glucose, which is an order of magnitude more
affordable as a starting substrate than ferulic acid ($0.3/kg
ASSOCIATED CONTENT
■
*
S
Supporting Information
1
8
versus $5/kg). Additionally, microbial cells in a nongrowing
state lose desired activity more rapidly than cells in a growing
state, frequently resulting in lower overall productivity. Another
group has published the use of growing E. coli cells for the
bioconversion of ferulic acid into vanillin, asserting that E. coli
has no degradation pathway from vanillin to vanillyl
Five figures, four tables, and text showing qRT-PCR results
(
(
Figure S1), chemical structures of select pathway metabolites
Figure S2), isovanillin titers (Figure S3), growth curves and
growth rates of wild-type and RARE strains (Figure S4), final
OD600 measurements (Figure S5), strains and plasmids (Table
S1), oligonucleotides (Table S2), reported enzyme activities on
benzaldehyde (Table S3), results of protein sequence align-
ments (Table S4), and codon-optimized or mutant gene
sequences (SI text). This material is available free of charge via
the Internet at http://pubs.acs.org.
6
2−64
alcohol.
It may be that no attempt to detect vanillyl
alcohol was made in these particular studies. As illustrated
throughout this report, even when performing simple
bioconversions using vanillate as a substrate, we observe
significant vanillyl alcohol formation. Whether for the purpose
of bioconversions or utilization of glucose as a sole carbon
source, the engineered strain reported in this study is expected
to be a superior E. coli host strain for producing aromatic
aldehydes.
AUTHOR INFORMATION
■
Corresponding Author
kljp@mit.edu
The ability of the RARE strain to accumulate aromatic
aldehydes has broad implications for the biosynthesis of
products derived from aldehyde intermediates. The demon-
Notes
The authors declare no competing financial interest.
1
1652
dx.doi.org/10.1021/ja506664a | J. Am. Chem. Soc. 2014, 136, 11644−11654

Journal of the American Chemical Society
Article
(
31) Thomason, L. C.; Costantino, N.; Court, D. L. In E. coli Genome
ACKNOWLEDGMENTS
■
Manipulation by P1 Transduction; Current Protocols in Molecular
Biology; John Wiley & Sons, Inc.: 2001.
The authors would like to acknowledge Micah J. Sheppard for
engaging in helpful discussions throughout the project. This
work was supported by the Synthetic Biology Engineering
Research Center (SynBERC, Grant No. EEC-0540879).
A.M.K. is a recipient of a National Science Foundation
Graduate Research Fellowship.
(32) Datsenko, K. A.; Wanner, B. L. Proc. Natl. Acad. Sci. U.S.A. 2000,
97, 6640.
(33) He, A.; Li, T.; Daniels, L.; Fotheringham, I.; Rosazza, J. P. N.
Appl. Environ. Microbiol. 2004, 70, 1874.
(
34) Nakano, M. M.; Corbell, N.; Besson, J.; Zuber, P. Mol. Gen.
Genet. 1992, 232, 313.
35) Fox, D. T.; Hotta, K.; Kim, C.; Koppisch, A. T. Biochemistry
(
REFERENCES
■
2008, 47, 12251.
(
1) Hagedorn, S.; Kaphammer, B. Annu. Rev. Microbiol. 1994, 48,
(36) Klock, H. E.; Lesley, S. A. In The Polymerase Incomplete Primer
Extension (PIPE) Method Applied to High-Throughput Cloning and Site-
Directed Mutagenesis; Doyle, S. A., Ed.; High Throughput Protein
Expression and Purification; Humana Press: 2009; p 91.
(37) Ko, J.; Kim, I.; Yoo, S.; Min, B.; Kim, K.; Park, C. J. Bacteriol.
2005, 187, 5782.
(38) Turner, P. C.; Miller, E. N.; Jarboe, L. R.; Baggett, C. L.;
Shanmugam, K. T.; Ingram, L. O. J. Ind. Microbiol. Biotechnol. 2011, 38,
431.
773.
(
(
2) Krings, U.; Berger, R. G. Appl. Microbiol. Biotechnol. 1998, 49:1, 1.
3) Fahlbusch, K.; Hammerschmidt, F.; Panten, J.; Pickenhagen, W.;
Schatkowski, D.; Bauer, K.; Garbe, D.; Surburg, H. In Flavors and
Fragrances; Ullmann’s Encyclopedia of Industrial Chemistry; Wiley-VCH
Verlag GmbH & Co. KGaA: 2000.
4) Burdock, G. A. Fenaroli’s Handbook of Flavor Ingredients; CRC
Press: Boca Raton, FL, 2005.
5) Feron, G.; Wache, Y. 1.16 Microbial Biotechnology of Food
(
(
(39) Sulzenbacher, G.; Alvarez, K.; van den Heuvel, R. H. H.;
Versluis, C.; Spinelli, S.; Campanacci, V.; Valencia, C.; Cambillau, C.;
Eklund, H.; Tegoni, M. J. Mol. Biol. 2004, 342, 489.
Flavor Production. In Food Biotechnology, 2nd ed.; Pometto, A., Shetty,
K., Paliyath, G., Levin, R., Eds.; CRC Press: Boca Raton, FL, 2006; p
407.
(40) Per
́
ez, J. M.; Arenas, F. A.; Pradenas, G. A.; Sandoval, J. M.;
(
(
6) Li, T.; Rosazza, J. P. N. Appl. Environ. Microbiol. 2000, 66, 684.
7) Priefert, H.; Rabenhorst, J.; Steinbuchel, A. Appl. Microbiol.
Vasquez, C. C. J. Biol. Chem. 2008, 283, 7346.
́
̈
(41) Miller, E. N.; Jarboe, L. R.; Yomano, L. P.; York, S. W.;
Shanmugam, K. T.; Ingram, L. O. Appl. Environ. Microbiol. 2009, 75,
4315.
(42) Pick, A.; Ruhmann, B.; Schmid, J.; Sieber, V. Appl. Microbiol.
̈
Biotechnol. 2013, 97:13, 5815.
Biotechnol. 2001, 56:3−4, 296.
8) Kaur, B.; Chakraborty, D. Appl. Biochem. Biotechnol. 2013, 169,
353.
9) Duff, S. J. B.; Murray, W. D. Biotechnol. Bioeng. 1989, 34, 153.
10) Nierop Groot, M. N.; de Bont, J. A. M. Appl. Environ. Microbiol.
998, 64, 3009.
11) Jain, A. N.; Khan, T. R.; Daugulis, A. J. Biotechnol. Lett. 2010,
2:11, 1649.
12) Walton, N. J.; Mayer, M. J.; Narbad, A. Phytochemistry 2003, 63,
05.
(
1
(
(
1
(
3
(
5
(
(
1
(43) Koma, D.; Yamanaka, H.; Moriyoshi, K.; Ohmoto, T.; Sakai, K.
Appl. Environ. Microbiol. 2012, 78, 6203.
(44) Venkitasubramanian, P.; Daniels, L.; Rosazza, J. P. N. J. Biol.
Chem. 2007, 282, 478.
(45) Juminaga, D.; Baidoo, E. E. K.; Redding-Johanson, A. M.; Batth,
T. S.; Burd, H.; Mukhopadhyay, A.; Petzold, C. J.; Keasling, J. D. Appl.
Environ. Microbiol. 2012, 78, 89.
(46) Zhang, J.; Klinman, J. P. J. Am. Chem. Soc. 2011, 133, 17134.
(47) Jez, J. M.; Penning, T. M. Chem. Biol. Interact. 2001, 130−132,
499.
13) Pohl, M.; Lingen, B.; Muller, M. Chem.Eur. J. 2002, 8, 5288.
14) Tripathi, C. M.; Agarwal, S. C.; Basu, S. K. J. Ferment. Bioeng.
997, 84, 487.
(
15) Yun, H.; Kim, B. Biotechnol. Bioprocess Eng. 2008, 13, 372.
(
16) Meyer, D.; Walter, L.; Kolter, G.; Pohl, M.; Muller, M.;
(48) Lapthorn, A. J.; Zhu, X.; Ellis, E. M. Chem. Biol. Interact. 2013,
Tittmann, K. J. Am. Chem. Soc. 2011, 133, 3609.
202, 168.
(
(
17) Li, T.; Rosazza, J. P. N. Appl. Environ. Microbiol. 2000, 66, 684.
18) Hansen, E. H.; Møller, B. L.; Kock, G. R.; Bunner, C. M.;
(49) Jeudy, S.; Monchois, V.; Maza, C.; Claverie, J.; Abergel, C.
Proteins: Struct., Funct., Bioinf. 2006, 62, 302.
(50) Joernvall, H.; Persson, B.; Krook, M.; Atrian, S.; Gonzalez-
Duarte, R.; Jeffery, J.; Ghosh, D. Biochemistry 1995, 34, 6003.
(51) Oppermann, U.; Filling, C.; Hult, M.; Shafqat, N.; Wu, X.;
̈
Kristensen, C.; Jensen, O. R.; Okkels, F. T.; Olsen, C. E.; Motawia, M.
S.; Hansen, J. Appl. Environ. Microbiol. 2009, 75, 2765.
(
(
19) Rodriguez, G.; Atsumi, S. Microb. Cell Fact. 2012, 11, 90.
20) Akhtar, M. K.; Turner, N. J.; Jones, P. R. Proc. Natl. Acad. Sci.
̈
Lindh, M.; Shafqat, J.; Nordling, E.; Kallberg, Y.; Persson, B.; Jornvall,
U.S.A. 2013, 110, 87.
21) Howard, T. P.; Middelhaufe, S.; Moore, K.; Edner, C.; Kolak, D.
H. Chem. Biol. Interact. 2003, 143−144, 247.
(
(52) Kavanagh, K. L.; Jornvall, H.; Persson, B.; Oppermann, U. Cell.
Mol. Life Sci. 2008, 65, 3895.
M.; Taylor, G. N.; Parker, D. A.; Lee, R.; Smirnoff, N.; Aves, S. J.;
Love, J. Proc. Natl. Acad. Sci. U.S.A. 2013, 110, 7636.
(53) Persson, B.; Zigler, J. S.; Jo
15.
(54) Nordling, E.; Jo
4267.
̈
rnvall, H. Eur. J. Biochem. 1994, 226,
(22) Li, K.; Frost, J. W. J. Am. Chem. Soc. 1998, 120, 10545.
(23) Draths, K. M.; Frost, J. W. J. Am. Chem. Soc. 1991, 113, 9361.
(24) Draths, K. M.; Pompliano, D. L.; Conley, D. L.; Frost, J. W.;
̈
rnvall, H.; Persson, B. Eur. J. Biochem. 2002, 269,
Berry, A.; Disbrow, G. L.; Staversky, R. J.; Lievense, J. C. J. Am. Chem.
Soc. 1992, 114, 3956.
(55) Wang, H. H.; Isaacs, F. J.; Carr, P. A.; Sun, Z. Z.; Xu, G.; Forest,
C. R.; Church, G. M. Nature 2009, 460, 894.
(56) Qi, L.; Larson, M.; Gilbert, L.; Doudna, J.; Weissman, J.; Arkin,
A.; Lim, W. Cell 2013, 152, 1173.
(
25) Snell, K. D.; Draths, K. M.; Frost, J. W. J. Am. Chem. Soc. 1996,
118, 5605.
(
26) Berry, A. Trends Biotechnol. 1996, 14, 250.
(57) Pharkya, P.; Burgard, A. P.; Maranas, C. D. Genome Res. 2004,
(
27) Food and Drug Administration, Department of Health and
14, 2367.
Human Services 21 CFR 184.1685. 2012.
(58) Alper, H.; Miyaoku, K.; Stephanopoulos, G. Nat. Biotechnol.
2005, 23, 612.
(59) Santos, C. N. S.; Xiao, W.; Stephanopoulos, G. Proc. Natl. Acad.
(
(
28) Ellis, E. M. FEMS Microbiol. Lett. 2002, 216, 123.
29) Sambrook, J.; Russell, D. W. Molecular cloning: a laboratory
manual; Cold Spring Harbor Laboratory Press: Cold Spring Harbor,
Sci. U.S.A. 2012, 109, 13538.
NY, 2001.
(60) Venkitasubramanian, P.; Daniels, L.; Das, S.; Lamm, A. S.;
Rosazza, J. P. N. Enzyme Microb. Technol. 2008, 42, 130.
(61) Barghini, P.; Di Gioia, D.; Fava, F.; Ruzzi, M. Microb. Cell Fact.
2007, 6, 13.
(
30) Baba, T.; Ara, T.; Hasegawa, M.; Takai, Y.; Okumura, Y.; Baba,
M.; Datsenko, K. A.; Tomita, M.; Wanner, B. L.; Mori, H. Mol. Syst.
Biol. 2006, 2, 2006−0008.
1
1653
dx.doi.org/10.1021/ja506664a | J. Am. Chem. Soc. 2014, 136, 11644−11654

Journal of the American Chemical Society
Article
(
62) Yoon, S.; Li, C.; Kim, J.; Lee, S.; Yoon, J.; Choi, M.; Seo, W.;
Yang, J.; Kim, J.; Kim, S. Biotechnol. Lett. 2005, 27, 1829.
63) Yoon, S.; Lee, E.; Das, A.; Lee, S.; Li, C.; Ryu, H.; Choi, M.; Seo,
W.; Kim, S. Biotechnol. Prog. 2007, 23, 1143.
64) Lee, E.; Yoon, S.; Das, A.; Lee, S.; Li, C.; Kim, J.; Choi, M.; Oh,
D.; Kim, S. Biotechnol. Bioeng. 2009, 102, 200.
65) Dunkelmann, P.; Kolter-Jung, D.; Nitsche, A.; Demir, A. S.;
(
(
(
Siegert, P.; Lingen, B.; Baumann, M.; Pohl, M.; Muller, M. J. Am.
Chem. Soc. 2002, 124, 12084.
(
66) Brandt, G. S.; Kneen, M. M.; Petsko, G. A.; Ringe, D.; McLeish,
M. J. J. Am. Chem. Soc. 2010, 132, 438.
67) Fuchs, M.; Tauber, K.; Sattler, J.; Lechner, H.; Pfeffer, J.;
Kroutil, W.; Faber, K. RSC Adv. 2012, 2, 6262.
68) Tseng, H.; Martin, C. H.; Nielsen, D. R.; Prather, K. L. J. Appl.
Environ. Microbiol. 2009, 75, 3137.
69) Tseng, H.; Harwell, C. L.; Martin, C. H.; Prather, K. L. J. Microb.
Cell Fact. 2010, 9, 96.
70) Salis, H. M.; Mirsky, E. A.; Voigt, C. A. Nat. Biotechnol. 2009, 27,
46.
(
(
(
(
9
1
1654
dx.doi.org/10.1021/ja506664a | J. Am. Chem. Soc. 2014, 136, 11644−11654