Directed evolution of the dioxygenase complex for the synthesis of furanone flavor compounds

Article abstract of DOI:10.1016/j.tet.2003.10.105

Herein we report a new preparation of 4-hydroxy-2,5-dimethyl-2,3- dihydrofuran-3-one, the flavor compound strawberry furanone, based on a 'green' approach with a minimum number of steps. The first step is an enzymatic dioxygenation of p-xylene to form cyclohexadiene-cis-diol, followed by ring opening via ozonolysis, and ring closure to form the furanone. In efforts to improve the efficiency of the enzymatic step, a directed evolution approach was taken to increase the substrate specificity and selectivity of the toluene dioxygenase enzyme system.

Full text of DOI:10.1016/j.tet.2003.10.105

Tetrahedron 60 (2004) 729–734
Directed evolution of the dioxygenase complex for the synthesis of
furanone flavor compoundsq
†
c
,
a
b
Lisa M. Newman,a Henry Garcia, Tomas Hudlicky and Sergey A. Selifonov
,
*
aCodexis, Inc., 515 Galveston Drive, Redwood City, CA 94063, USA
bDepartment of Chemistry, University of Florida, Gainesville, FL 32611-7200, USA
cMaxygen, Inc., 515 Galveston Dr., Redwood City, CA 94063, USA
Received 11 July 2003; revised 1 September 2003; accepted 17 October 2003
Abstract—Herein we report a new preparation of 4-hydroxy-2,5-dimethyl-2,3-dihydrofuran-3-one, the flavor compound strawberry
furanone, based on a ‘green’ approach with a minimum number of steps. The first step is an enzymatic dioxygenation of p-xylene to form
cyclohexadiene-cis-diol, followed by ring opening via ozonolysis, and ring closure to form the furanone. In efforts to improve the efficiency
of the enzymatic step, a directed evolution approach was taken to increase the substrate specificity and selectivity of the toluene dioxygenase
enzyme system.
q 2003 Elsevier Ltd. All rights reserved.
1. Introduction
example through oxidations, cycloadditions, ring opening
and ring closure reactions, to generate precursors for
synthetic applications. Economic application of microbial
dioxygenases for the production of substituted cyclo-
hexadiene-cis-diols has been hampered by limited volu-
metric productivity.
The flavor compound strawberry furanone (1)1,2 is a high
value fine chemical commercially available as Furaneolw
from Firmenich.3 It is found naturally in pineapples,
strawberries, mangoes, as well as in roasted or cooked
foods such as malt, sesame seeds, roasted beef, and soy
sauce, where it is formed via the Maillard reaction.4 Despite
this natural occurrence of 1 in fruits and cooked foods, most
of the commercially available flavor compound is a non-
natural, synthetic product manufactured from relatively
expensive 2,5-hex-3-ynediol.5 Herein, we report a new
preparation of 1 based on a chemoenzymatic approach with
a minimum number of steps from inexpensive p-xylene
using bacterial dioxygenase enzymes for enabling the
strategy and the execution of the synthesis (Fig. 1). In
addition, we explore the directed evolution of these bacterial
dioxygenase enzymes for increased efficiency and potential
for commercial applications of the synthesis of 1 and similar
flavor compounds.
Directed evolution is a technology that allows for the
diversification of enzymes into biocatalyst platforms by
Microorganisms contain aryl dioxygenases that in a single
step convert aromatic hydrocarbons to corresponding diene-
cis-diols. Substituted cyclohexadiene-cis-diols are interest-
ing synthons for the manufacture of a range of fine
chemicals.6 The various functionalities and stereospecific
centers of these molecules can be exploited chemically, for
Figure 1. A novel route to strawberry furanone was envisioned in which the
p-xylene derived cis-1,2-dihydroxy-3,6-dimethyl-cyclohexa-3,5-diene is
converted in a short reaction sequence to the desired product. B. The
strategy and execution of the synthesis. Conditions (a) E. coli JM109
(pTrctodNK1); (b) O3, MeOH; then aq. Na2S2O3, NaHCO3 (84–92%); (c)
nBuOAc, H2O, phosphate buffer, 95 8C, 5 h (39–40%).
q Supplementary data associated with this article can be found at doi:10.
1016/J.tet.2003.10.105
*
Corresponding author; e-mail address: lisa.newman@codexis.com
Present address: Aromagen Corp., Plymouth, MN, 55446, USA; e-mail
address: selifonov@attbi.com.
†
0040–4020/$ - see front matter q 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tet.2003.10.105

730
L. M. Newman et al. / Tetrahedron 60 (2004) 729–734
recombination of homologous gene sequences that encode
similar enzymes.7 These platforms can be screened for
desired biocatalyst characteristics such as substrate speci-
ficity, specific activity, enantioselectivity, regioselectivity,
substrate tolerance (to allow conversions at economically
relevant substrate concentrations), product tolerance (to
prevent inhibition of conversion as the conversion
approaches completion), and half-life time of the enzyme.
The technology is applicable to single genes, encoding
homomeric enzymes, multiple genes, encoding heteromeric
enzymes or enzyme pathways, and even whole genomes.8
in this program will enable novel chemical processes at
commercial scale. One such new application is the synthesis
of 3(2H)-furanone flavor compounds. Here we describe the
application of dioxygenase to a novel route for strawberry
furanone synthesis starting from p-xylene.
2. Results and discussions
2.1. Directed evolution of aryl dioxygenase
Multigene DNA shuffling was applied to the complete
toluene dioxygenase and tetrachloro-benzene dioxygenase
operons to improve the biotransformation of a range of
aromatics. The four gene operons encoding toluene and
tetrachlorobenzene dioxygenase,20 todC1C2BA and
tecA1A2A3A4, were shuffled (Fig. 2A). The shuffled library
was pre-screened according to the tier 1 assay (see Section
4) in E. coli DH5a for active clones using the indole
oxidation assay21 and was 70% active. Restriction enzyme
digests of 24 clones from the library indicated that 23 of the
randomly picked clones were chimeras of the parental genes
(results not shown).
Aryl dioxygenases are multi-subunit, heteromeric, iron
containing enzyme complexes that catalyze the incorpor-
ation of both atoms of molecular oxygen into aromatic
hydrocarbons.9 Toluene and biphenyl dioxygenase oxidize
their substrates toluene and biphenyl to cis-1,2-dihydroxy-
3-methyl-cyclohexa-3,5-diene and cis-1,2-dihydroxy-3-
phenyl-cyclohexa-3,5-diene, respectively.9,10 Directed
evolution of toluene dioxygenase has resulted in improved
substrate specificity,11 while biphenyl dioxygenase has been
evolved for remediation purposes by extending its substrate
range.12,13
Aryl dioxygenases are encoded by four genes that constitute
a gene cluster or operon. The four genes encode each of the
two subunits of the oxygenase component, ferredoxin and
ferredoxin reductase14 (Fig. 2A). In all reported aryl
dioxygenase enzyme improvement studies to date, only
the large subunit of the oxygenase was subjected to directed
evolution.11–16 While this subunit is generally regarded as
the main contributor to substrate specificity,17 – 19 all
components of the enzyme complex can in principle
represent the rate-limiting step in substrate oxidation. In
this report, we present the first directed evolution studies on
a multi enzyme complex in which all components were
subjected to recombination of genetic diversity. It is
expected that the improved biocatalysts that are discovered
The diene-cis-diol resulting from p-xylene oxidation is of
interest as a starting material for a novel synthetic route to
strawberry furanone as described below. Active clones from
the first tier assay were therefore, evaluated for the ability to
oxidize p-xylene to cis-1,2-dihydroxy-3,6-dimethyl-3,5-
cyclohexadiene. A high throughput screen was developed
in which p-xylene toxicity to the bacterial host was
circumvented by supplying the substrate in a slow-release
system (Fig. 3). The amount of diol produced was compared
to the amount of product produced by the parental clones. In
total, 1044 different dioxygenases were screened and nine
exhibited 2 to 3.5-fold higher activity when compared to the
best parent.
The best nine clones were retested in multiples of eight
(Table 1). The greatest improvement in activity was that of
clone C10, which had a 4.4-fold higher activity. It is known
that the details of a screening procedure may have an effect
Figure 2. A. Genetic maps of the toluene (todC1C2BA) and tetrachlor-
obenzene dioxygenase (tecA1A2A3A4) gene clusters. The operons are
approximately 3.6 kb in length and encode the oxygenase large subunit (LS,
encoded by todC1 and tecA1, respectively), the oxygenase small subunit
(SS, todC2 and tecA2), a ferredoxin (Fd, todB and tecA3) and reductase (Rd,
todA and tecA4). Homology between the genes and encoded enzymes is
given in percentages at the nucleotide sequence level and amino acid
sequence level (in parentheses) above the gene arrows. B. Composition of
the chimeric dioxygenase clusters for nine improved dioxygenase clones.
Sequence derived from todC1C2BA is in dark grey, sequence derived from
tecA1A2A3A4 is in light grey, point mutations due to the PCR process are
indicated as white lines.
Figure 3. High throughput screen for p-xylene oxidation. A 96-well system
was developed in which toxicity of high p-xylene concentrations was
prevented by supplying the substrate from a slow-release formulation. Cell
suspensions were added on top of the semi-solid substrate and efficient
mixing was achieved by shaking with a 1 mm steel ball. Evaporation of
substrate was prevented by sealing the plates with an adhesive aluminum
foil. Head space over the reaction mixture was sufficient to provide oxygen
for the bioconversion. Upon termination of the reaction after 1 h of
incubation, the presence of diene-cis-diol was measured
spectrophotometrically.

L. M. Newman et al. / Tetrahedron 60 (2004) 729–734
Table 1. Apparent activitya of diol formation in parental and family shuffled clones
731
DH5a clones 2£YT
LS5218 clones minimal media
Average activity
sdb
Fold improvement
1.0
Average activity
sd
Fold improvement
1.00
tod
tec
B9
C9
C10
E2
F1
F5
H3
H6
G12
0.830
0.663
1.59
3.06
3.67
2.18
3.03
2.85
2.99
2.38
2.63
0.134
0.070
0.259
0.970
0.894
0.402
0.451
0.404
0.902
0.481
0.438
1.64
1.24
1.10
1.45
4.12
2.46
3.40
2.16
3.59
3.51
2.88
0.292
0.550
0.242
0.178
0.683
0.761
0.743
0.389
1.13
1.9
3.7
4.4
2.6
3.7
3.4
3.6
2.9
3.2
(0.7)
(0.9)
2.5
1.5
2.1
1.3
2.2
2.1
1.8
1.10
0.706
a
Apparent activity¼mmol/min mL whole cells. Activities are normalized to optical density of cell suspensions.
b
Standard deviation.
on the outcome of such screen. To test the influence of the
large dioxygenase subunit in clone F1 consisted primarily of
tod sequence where the tod operon was shown to have less
activity on toluene than the tec operon. This could indicate
that for oxidation of toluene, the large subunit is not
necessarily the limiting factor. On this substrate, activity
does not seem to be limited by the large subunit of the
complex.
host cell background and the growth medium on the activity
of the shuffled dioxygenases, plasmid purified from the nine
clones and the parental clones were transformed into E. coli
LS5218. The transformants were tested for activity on
p-xylene after growth in minimal media with fructose as
carbon source. Improved activity was reconfirmed for seven
of the nine shuffled clones, although the extent of such
improvements were reduced to 1.3- to 2.5-fold. This
indicates that either the host strain or the growth conditions
or both have a direct impact on overall dioxygenase activity.
The exact cause of this effect remains to be determined.
The availability of a wide range of highly active
dioxygenase enzymes may enable the commercial appli-
cation of these catalysts. One novel application would be in
the manufacture of 3(2H)-furanone flavor compounds.
The same nine clones were examined for improved activity
on a range of substituted benzene derivatives (Fig. 4).
Depending on the substrate, among the nine clones
improvements ranged from 2.7-fold (on m-xylene) to 16-
fold on cumene. In addition to p-xylene, clone C10
exhibited the highest activity on m-xylene, 1,2,3-trimethyl-
benzene and naphthalene, while clone F1 showed the
highest activity on cumene and toluene. DNA sequence
analysis of these nine clones indicated that significant
diversification had occurred (Fig. 2B). Clone C10 consisted
primarily of the tec sequence, with significant amounts of
tod sequence in the reductase gene only. Surprisingly, the
2.2. Dioxygenase enabled route to strawberry furanone
A novel route to strawberry furanone was envisioned in
which the p-xylene derived diene-cis-diol is converted in a
short reaction sequence to the desired product (Fig. 1A). Our
chemoenzymatic approach as well as the execution of the
synthesis are outlined in Figure 1B.
Biotransformation of p-xylene by whole cells of E. coli
strains JM109 (pTrctodNK1) or LS5218 (pTrctodNK1)
expressing toluene dioxygenase produced 2. Cultures were
grown in minimal media with fructose as the carbon source
and p-xylene was fed as a vapor introduced into the air
stream. Fed-batch cultures of 1.0 and 5.0 L volume yielded
20 and 88 g of 2, respectively, and which was purified from
cell-free culture broth by extraction. Synthesis of 2 up to
0.2 g/L had been reported for Pseudomonas mutants that
were unable to convert p-xylene beyond the diol.10
The meso diol was subjected to ozonolysis in MeOH at
278 8C. When the solution turned a light blue color, it was
quenched with aqueous thiosulfate/bicarbonate, and the
mixture was extracted with ethylacetate. A semisolid was
obtained in ,90% yield. NMR analysis of this material
indicated the desired molecule as a 1:1 mixture of the
diketone 3 and its diastereomeric methoxyketal–hemiketal
4.22 As similar mixtures were obtained when this sequence
was performed on the acetonide derived from 2, protection
of the diol was not required. The crude ozonolysis products
were dissolved in n-butylacetate and added to phosphate
buffer. The mixture was heated at 95 8C for 5 h, then
allowed to stand for 12 h. Furanone 1 was isolated by
extraction with ethylacetate or by steam distillation. Its
Figure 4. Substrate specificity of nine clones that were improved for
oxidation of p-xylene. Clones were examined individually for activity on
the indicated substrates. Relative activity is measured as the UV absorbance
of the supernatant of cultures of each clone at the lmax for each product
diene-cis-diol. Note that the absolute activity varies per substrate. The clone
with highest activity, with respect to the best parent, is indicated by the
black bar with the fold improvement noted.

732
L. M. Newman et al. / Tetrahedron 60 (2004) 729–734
physical and spectral properties were identical to those of a
commercial sample. Consistent yields of ca. 40% were
attained with n-butylacetate, even with varying cosolvents/
buffers and temperature regimes. The entire sequence can be
successfully performed on a 5 g scale without purification of
intermediates. Further optimization at the process level will
provide for a potential commercial process for this
important flavor additive.
initiated to maintain the dissolved oxygen at 30%. After the
culture reached an OD600 15, the temperature was reduced
to 34 8C and 200 mM ferrous ammonium sulfate was added.
The expression of toluene dioxygenase was induced by the
addition of 250 mM of IPTG. The culture was grown for
another 2 h, after which p-xylene was fed as vapor to the
culture through the air stream at a flow rate of 3 L/min. The
formation of cis-1,2-dihydroxy-3,6-dimethyl-hexa-3,5-
diene, was monitored by measuring the absorbance of
cell-free supernatant at a wavelength of 280 nm and
estimating the concentration using an extinction coefficient
3. Conclusions
of 1¼6500 mol21 cm21 10
.
Improved dioxygenase biocatalysts were generated through
multigene DNA shuffling of the entire dioxygenase gene
cluster. Because such improvements were generated by
recombining different fragments of the dioxygenase oper-
ons, we expect that further improvement of these catalysts is
possible. Such improvements are desirable to further lower
the cost of these diene-cis-diols. It is our estimate that with a
product titer in the range of 60–80 g/L in the biotransform-
ation, the novel route to strawberry furanone will be
competitive with the current process. Analogous routes to
other commercial flavor compounds such as 4-hydroxy-5-
methyl-3(2H)-furanone and 5-ethyl-2-methyl-4-hydroxy-
3(2H)-furanone can be developed based on toluene and
p-ethyl-toluene, respectively and in fact, a range of new
flavor compounds may be available through dioxygenase-
mediated supply of diene-cis-diols (Fig. 5). As directed
evolution of enzymes is applicable to provide efficient
catalysts for any desired substrate, such applications are
coming closer to reality.
After the formation of the product was complete, the culture
was harvested and the cells removed by centrifugation at
5,000 rpm, 20 min, at 4 8C. NaCl (280 g) was dissolved in the
supernatant and the product phase extracted three times with
approximately 0.6 L of ethyl acetate. The organic layers were
combined and the solvent evaporated under reduced pressure
to yield 20 g of nearly colorless crystals of the desired
compound. The extracted samples were homogenous by TLC
analysis after development in ethyl acetate giving an UV
absorbance at 254 nm and staining positive with iodine vapor.
The biotransformation procedure was scaled to a 5.0 L vessel
(BioFlo 3000, New Brunswick, Edison, NJ) using the same
medium and process conditions as described above. The 5.0 L
of cell culture yielded approximately 88 g or 17.6 g/L of the
desired product.
4.2. Bacterial strains and plasmid construction
P. putida F1 (ATCC 700007) was obtained from the
American Type Culture Collection. The toluene dioxygen-
ase operon todC1C2BA (GenBank accession number
J04996) were PCR amplified using the forward primer
CCATG GCTTGAAAAGTGAGAAGAC and the reverse
primer TCTAGAGCGCCACCGCC TGTCACC. pSTE7
containing the tetra-chlorobenzoate dioxygenase genes
from Burkholderia sp. PS12, was a kind gift from Dr
Deitmar H. Pieper, Division of Microbiology, GBF-
National Research Centre for Biotechnology, Braunsch-
weig, Germany. The tec operon tecA1A2A3A4 (GenBank
accession number U78099) was PCR amplified from
plasmid pSTE7 using the forward primer CCATGGCAT
GAAACG TGAGGAGAC and reverse primer TCTA
GAGCG CCTCCGCCCGTCACC. The PCR product was
digested with NcoI and XbaI, gel purified, ligated into
similarly digested expression vector pTrc99a (Amersham
Pharmacia Biotech, Piscataway, NJ), and transformed into
E. coli DH5a (Invitrogen Corp., Carlsbad, CA). Recombinant
strains were maintained on LB agar plates or liquid media
containing 100 mg/mL ampicillin. Strains were stored at
280 8C as 20% glycerol stocks. DNA sequences of the
corresponding pTrctodNK1 and pTrctecNK1 inserts were
verified by sequencing both DNA strands on an ABI 377
sequencer (Perkin–Elmer, Foster City, CA). PCR reagents
were obtained from Qiagen (Valencia, CA) and restriction
enzymes from New England Biolabs (Beverly, MA).
4. Experimental
4.1. Biotransformation; purification of the
cyclohexadiene-cis-diol; directed evolution
For biotransformation studies at the 1 and 5 L scale, clones
containing pTrctodNK1 were preferred over pTrctecNK1.
E. coli LS5218 (pTrctodNK1) was grown in a 2 L fermentor
(New Brunswick BioFlo 2000, Edison, NJ), on E223 with R2
trace elements,24 0.2% yeast extract, 5 mM MgSO4 and
0.5% fructose at 37 8C. Ampicillin was added to a final
concentration of 100 mg/mL. Aeration was at 1.8 L/min and
pH was maintained at 7.0 with KOH. The culture was grown
to an OD600 of 3.5, at which time a 50% fructose, 6%
ammonium chloride and 2% magnesium sulfate feed was
Figure 5. Analogous routes to other flavor compounds such as 4-hydroxy-
5-methyl-3(2H)-furanone (5) and 5-ethyl-2-methyl-4-hydroxy-3(2H)-fur-
anone (6) can be developed based on toluene and p-ethyl-toluene,
respectively and in fact, a range of new flavor compounds may be available
through dioxygenase-mediated supply of diene-cis-diols.
4.3. Shuffling
The two multigene operons from parent plasmids
pTrctodNK1 and pTrctecNK1, were shuffled as previously

L. M. Newman et al. / Tetrahedron 60 (2004) 729–734
733
described.25 The PCR product was digested with NcoI and
XbaI, gel purified, ligated into similarly digested expression
vector pTrc99a and transformed into E. coli DH5a. PCR
product of insert DNA from randomly selected clones was
digested with restriction enzyme HincII, and the restriction
digest pattern compared to that of similarly digested
parental PCR product to confirm that the shuffling reaction
produced chimeric sequences. DNA sequences of improved
shuffled clones were determined by sequencing both DNA
strands on an ABI 377 sequencer (Perkin–Elmer, Foster
City, CA). The sequences were compared to the parental
gene sequences to confirm the formation of chimeric
sequences. DNA sequences were analyzed using
Sequencher version 4.0.5 (Gene Codes Corp., Ann Arbor,
MI) and Vector NTI 6 (InforMax Inc., Bethesda, MD).
(3000 rpm, 10 min, 10 8C) after which 200 mL of the
supernatants were transferred to a Corning UV plate (VWR,
So. Plainfield, NJ) and the absorbance of the cell-free
supernatants was read at 280 and 320 nm on a SpectraMax
190 plate reader (Molecular Devices, Sunnyvale, CA).
4.6. Analytical methods
TLC analysis of cyclohexadiene-cis-diols was performed
on Merck silica gel 60 F254 plates (2.5£7.5 cm, Aldrich,
Milwaukee, WI). Aliquots of p-xylene transformation assays
were extracted three times with an equal volume of ethyl
acetate, the organic layers combined and dried under nitrogen.
The extracted samples were dissolved in 20 mL ethyl acetate
and developed in ethyl acetate vapor giving an UV absorbance
at 254 nm and staining positive with iodine vapor. The
samples were compared to cyclohexadiene-cis-diol standard
prepared from the wild type toluene dioxygenase clone.
4.4. Screening. First tier assay
Shuffled dioxygenase libraries were initially screened for
activity in vivo using the ability of the dioxygenase to
transform endogenous indole to indigo.21 Colonies contain-
ing active dioxygenase protein turn blue from the pro-
duction of indigo. The shuffled libraries were transformed
into E. coli DH5a and plated onto LB agar containing
100 mg/mL ampicillin. Plates were incubated for 20 h at
37 8C and then moved to room temperature for 24 h. The
plates were checked for blue colored colonies indicating the
presence of active dioxygenase. The blue color of the
colonies varied from light to dark among the active shuffled
clones. These active clones were picked to 96 well plates
containing LB media, 100 mg/mL ampicillin, 0.2% glucose,
and 20% glycerol. These master plates were incubated 20 h
at 37 8C and stored at ,20 8C and used to inoculate cultures
for substrate activity assays.
4.7. Oxidation by ozonolysis of an unprotected diol–
diene to form a diol–dione
The cis-1,2-dihydroxy-3,6-dimethylhexa-3,5-diene 2 was
not purified prior to ozonolysis. All reagents were purchased
from Fischer Scientific (Pittsburgh, PA) and used without
further purification. Sodium sesquicarbonate solution A was
prepared by dissolving 84 g of NaHCO3 and 53 g of
Na2CO3 in 1 L of water. The final pH of the solution was
about 10–11. 5.07 g of 1,2-dihydroxy-3,6-dimethylhexa-
3,5-diene 2 (vacuum dried) was dissolved in 50 mL of
MeOH and ozonized at 278 8C for 3 h (until blue color
persisted) (65 kV, 3 lpm). The resulting solution was
transferred via insulated canula to a solution of 21 mL of
1 M Na2S2O3 and 13 mL of 1.5 M sesquicarbonate solution
A and stirred at 0 8C. The transfer took 5 min. The resulting
solution quickly became peroxide negative. Stirring was
continued for 30 min at 0 8C and for another 30 min at room
temperature. Methanol was distilled off under reduced
pressure at 40 8C, and the resulting solution (with white
precipitate) was extracted 4 times with 125 mL ethyl
acetate. Combined extracts were dried over anhydrous
sodium sulfate and evaporated to give colorless and
practically odorless oil, which after vacuum drying partially
crystallized on standing (4.92 g). This material was the
desired product as a 1:1 mixture of the free 2,5-diketo-3,4-
dihydroxyhexane 3 and it’s methoxyketal–hemiketal 4,
with combined yields in 84–92% range.
4.5. Second tier assay
Active shuffled clones were screened for the conversion of
p-xylene to cis-1,2-dihydroxy-3,6-dimethylcyclohexa-3,5-
diene in 96-well-plate format. Clones were grown in 200 mL
of 2£YT media containing 100 mg/mL ampicillin and 0.2%
glucose. Plate wells were inoculated with 3 mL/well of
inoculum from the master plate and incubated for 20 h at
37 8C (250 rpm and 85% relative humidity (RH)). The 20-h
cultures were subcultured into 2.2-mL-deep well plates
containing 500 mL/well of 2£YT media containing
100 mg/mL ampicillin and 1 mM IPTG. These plates were
incubated under the same conditions for 6 h. The plates were
centrifuged using an Allegra 6KR centrifuge (Beckman
Instruments, Fullerton, CA) (3000 rpm, 10 min, 10 8C) to
pellet cells, washed once in 40 mM phosphate buffer, pH
7.0, and resuspended in 0.5 mL of 40 mM potassium
phosphate buffer, pH 7.0.
4.8. Cyclization of an unprotected diol–dione
The following condition was used after modification of the
procedure published in US Patent 5,149,840,26 which
describes an optimized cyclization of rhamnose and other 6-
deoxyhexoses to 1. Crude ozonolysis product, 800 mg, was
added to buffer containing: 340 mg of NaH2PO4£1H2O,
80 mg NaHCO3, 0.260 mL of H2O, 0.100 mL of 40% NaOH.
The semi-solid mixture was added to n-butylacetate or
ethylacetate and the two-phase system was, after degassing
with argon, heated in a closed vial (see Table 2). The solvent
was separated from the dark-reddish-brown aqueous phase,
the aqueous layer was extracted with 3£5 mL EtOAc, and
the organic layer and extracts were combined, dried
over anhydrous MgSO4 and the solvent evaporated. All
The washed cell suspensions (300 mL) were transferred to a
96-deep well assay plate that contained 250 mL of solid
agarose/10% p-xylene emulsion on the bottom of each well.
The solidified agarose/substrate emulsion served as the source
of volatile substrate for the biotransformation reaction. The
deep well plates were sealed with Biomek Seal and Sample
foil seals (Beckman Instruments, Fullerton, CA) and shaken at
room temperature for 1 h. The cell suspensions were then
transferred to a 96 shallow well plate, the cells pelleted

734
L. M. Newman et al. / Tetrahedron 60 (2004) 729–734
Table 2. Summary of two-phase cyclization reactions for 4-hydroxy-3[2H]-furanone 1 synthesis
Mass of starting material
Solvent
(volume)
Buffer, amounts of NaH2PO4·H2O and NaHCO3
Temp.
(8C)
Time
(h)
Weight
recovered
Ratio of 1:3
(by NMR)
800 mg
1.3 g
1 g
n-BuOAc (3 mL)
EtOAc (3 mL)
n-BuOAc (5 mL)
n-BuOAc (10 mL)
340 mgþ80 mg
510 mgþ120 mg
425 mgþ100 mg
170 mgþ40 mg
95, then rt
Reflux ,80
95, then rt
110
4–16
22
8–15
22
280 mg
630 mg
390 mg
1.3 g
4:1
1:1
3.5:1
3:1
2.6 g
Progress; Teranishi, R., Wick, E. L., Hornstein, I., Eds.;
Kluwer Academic Plenum: New York, 1999; p 305.
5. Re, L.; Maurer, B.; Ohloff, G. Helv. Chim. Acta 1973, 56,
1882–1894.
experiments were repeated at least 3 times. The experiments
were reproducible up to 2–6 g. The material was analyzed by
TLC, 13C NMR, and 1H NMR with 4-hydroxy-2,5-dimethyl-
2,3-dihydrofuran-3-one1 ascomparativestandard.22 Deutero-
chloroform was used as the solvent and D2O exchange was
also performed. The final 4-hydroxy-2,5-dimethyl-2,3-
dihydrofuran-3-one 1 product is very volatile and oxidizes
readily in the presence of air, as evidenced by appearance of
additional spots on subsequent TLC analysis.
6. (a) Hudlicky, T.; Gonzalez, D.; Gibson, D. T. Aldrichim. Acta
1999, 32, 35–62. (b) Boyd, D. R.; Sheldrake, G. N. Nat. Prod.
Rep. 1998, 309–324. (c) Brown, S. M.; Hudlicky, T. Organic
Synthesis: Theory and Applications, Hudlicky, T., Ed.; JAI:
Greenwich, CT, 1993; Vol. 2, p 113. (d) Hudlicky, T.; Reed,
J. W. In Advances in Asymmetric Synthesis; Hassner, A., Ed.;
JAI: Greenwich, CT, 1995; p 271.
4.8.1. 4-Hydroxy-3,5-dimethyl-3(2H)-furanone-1. 1H NMR
(400 MHz, CDCl3): 7.6 (1H), 4.5 (dq, J¼7, 1 Hz, 1H), 2.26 (d,
J¼1 Hz, 3H), 1.44 (d, J¼7 Hz, 3H). 13C NMR (100 MHz,
CDCl3): 198.84, 174.89, 133.95, 80.02, 16.27, 13.37.
7. Powell, K. A.; Ramer, S. W.; DelCardayre, S. B.; Stemmer,
W. P. C.; Tobin, M. B.; Longchamp, P. F.; Huisman, G.
Angew. Chem. 2001, 40, 3948–3959.
8. Huisman, G. W.; Gray, D. Curr. Opin. Biotechnol. 2002, 13,
352–358.
4.8.2. cis-1,2-Dihydroxy-3,6-dimethyl-hexa-3,5-diene-2.
1H NMR (400 MHz, CDCl3): 5.63 (2H), 4.06 (2H), 2.55
(2H), 1.88 (6H). 13C NMR (100 MHz, CDCl3): 135.14,
120.25, 72.05, 19.48. Compound 2 is unstable and
dehydrates readily to 2,5-dimethylphenol.7 This degradation
9. Boyd, D. R.; Sharma, N. D.; Allen, C. C. R. Curr. Opin.
Biotechnol. 2001, 12, 564–573.
10. Gibson, D. T.; Mahadevan, V.; Davey, J. F. J. Bacteriol. 1974,
119, 930–936.
1
11. Sakamoto, T.; Joern, J. M.; Arisawa, A.; Arnold, F. H. Appl.
Environ. Microbiol. 2001, 67, 3882–3887.
product was detected in the spectra of compound 2: H
NMR (400 MHz, CDCl3): 6.97 (d, J¼7.6 Hz, 1H), 6.63 (d,
12. Kumamaru, T.; Suenaga, H.; Mitsuoka, M.; Watanabe, T.;
Furukawa, K. Nat. Biotechnol. 1998, 16, 663–666.
13. Bruehlmann, F.; Chen, W. Biotechnol. Bioengng 1999, 63,
544–551.
J¼7.6 Hz, 1H), 6.59 (1H), 2.25 (3H), 2.19 (3H).
13C NMR (100 MHz, CDCl3): 163.8, 136.8, 130.66, 121.04,
120.68, 115.62, 20.88, 15.3.
14. Zylstra, G. J.; Gibson, D. T. Genet. Engng 1991, 13, 183–203.
15. Misawa, N.; Shindo, K.; Takahashi, H.; Suenaga, H.; Iguchi,
K.; Okazaki, H.; Harayama, S.; Furukawa, K. Tetrahedron
2002, 58, 9605–9612.
4.8.3. 2,5-Diketo-3,4-dihydroxyhexane-3/methoxyketal–
hemiketal-4 mixture: 3. 1H NMR (400 MHz, CDCl3): 4.38
(2H), 2.31 (6H). 13C NMR (100 MHz, CDCl3): 207.87,
78.53, 16.31. 4: 1H NMR (400 MHz, CDCl3): 3.83 (d,
J¼3.5 Hz, 1H), 3.69 (d, J¼3.5 Hz, 1H), 3.36 (3H), 1.4 (m,
6H). 13C NMR (100 MHz, CDCl3): 105.56, 101.8, 70.89,
66.88, 48.88, 27.03, 21.46.
16. Suenaga, H.; Mitsuoka, M.; Ura, Y.; Watanabe, T.; Furukawa,
K. J. Bacteriol. 2001, 183, 5441–5444.
17. Parales, J. V.; Parales, R. E.; Resnick, S. M.; Gibson, D. T.
J. Bacteriol. 1998, 180, 1194–1199.
18. Barriault, D.; Simard, C.; Chatel, H.; Sylvestre, M. Can.
J. Microbiol. 2001, 47, 1025–1032.
19. Beil, S.; Mason, J. R.; Timmis, K. N.; Pieper, D. H.
J. Bacteriol. 1998, 180, 5520–5528.
Acknowledgements
20. Beil, S.; Happe, B.; Timmis, K. N.; Pieper, D. H. Eur.
J. Biochem. 1997, 247, 190–199.
The project was supported by Maxygen, Inc. Partial support
(T. Hudlicky) was also received by the National Science
Foundation (CHE-9910412). The authors thank Dr. Chris
Davis and Dr. Gjalt Huisman for critical review and
assistance in the preparation of the manuscript.
21. Ensley, B. D.; Ratzkin, B. J.; Osslund, T. D.; Simon, M. J.;
Wackett, L. P.; Gibson, D. T. Science 1983, 222, 167–169.
22. Briggs, M. A.; Haines, A. H.; Jones, H. F. J. Chem. Soc.,
Perkin Trans. 1: Org. Bio-Org. Chem. 1985, 4, 795–798.
23. Lageveen, R. G.; Huisman, G. W.; Preusting, H.; Ketelaar, P.;
Eggink, G.; Witholt, B. Appl. Environ. Microbiol. 1988, 54,
2924–2932.
References and notes
24. Riesenberg, D.; Menzel, K.; Schulz, V.; Schumann, K.; Veith,
G.; Zuber, G.; Knorre, W. A. Appl. Microbiol. Biotechnol.
1990, 34, 77–82.
1. Slaughter, J. C. Biol. Rev. 1999, 7, 259–276.
2. Pickenhagen, W. In Flavor Chemistry: Thirty Years of
Progress; Teranishi, R., Wick, E. L., Hornstein, I., Eds.;
Kluwer Academic/Plenum: New York, 1999; p 75.
25. Cramer, A.; Raillard, S. A.; Bermudez, E.; Stemmer, W. P.
Nature 1998, 391, 288–291.
3. Furaneolw is a trademark of Firmenich, Geneva.
26. Decnop, C.; van Dort, J. M.; de Hey, J. T. US Patent 5,149,
840, 1992.
4. Tressel, R.; Rewicki, D. In Flavor Chemistry: Thirty Years of