Cloning and characterization of the biosynthetic gene cluster of 16-membered macrolide antibiotic FD-891: Involvement of a dual functional cytochrome P450 monooxygenase catalyzing epoxidation and hydroxylation

Article abstract of DOI:10.1002/cbic.201000214

FD-891 is a 16-membered cytotoxic antibiotic macrolide that is especially active against human leukemia such as HL-60 and Jurkat cells. We identified the FD-891 biosynthetic (gfs) gene cluster from the producer Streptomyces graminofaciens A-8890 by using

Full text of DOI:10.1002/cbic.201000214

DOI: 10.1002/cbic.201000214
Cloning and Characterization of the Biosynthetic Gene
Cluster of 16-Membered Macrolide Antibiotic FD-891:
Involvement of a Dual Functional Cytochrome P450
Monooxygenase Catalyzing Epoxidation and
Hydroxylation
Fumitaka Kudo,[b] Atsushi Motegi,[b] Kazutoshi Mizoue,[c] and Tadashi Eguchi*[a]
FD-891 is a 16-membered cytotoxic antibiotic macrolide that is
especially active against human leukemia such as HL-60 and
Jurkat cells. We identified the FD-891 biosynthetic (gfs) gene
cluster from the producer Streptomyces graminofaciens A-8890
by using typical modular type I polyketide synthase (PKS)
genes as probes. The gfs gene cluster contained five typical
modular type I PKS genes (gfsA, B, C, D, and E), a cytochro-
me P450 gene (gfsF), a methyltransferase gene (gfsG), and a
regulator gene (gfsR). The gene organization of PKSs agreed
well with the basic polyketide skeleton of FD-891 including the
oxidation states and a-alkyl substituent determined by the
substrate specificities of the acyltransferase (AT) domains. To
clarify the involvement of the gfs genes in the FD-891 biosyn-
thesis, the P450 gfsF gene was inactivated; this resulted in the
loss of FD-891 production. Instead, the gfsF gene-disrupted
mutant accumulated a novel FD-891 analogue 25-O-methyl-
FD-892, which lacked the epoxide and the hydroxyl group of
FD-891. Furthermore, the recombinant GfsF enzyme coex-
pressed with putidaredoxin and putidaredoxin reductase con-
verted 25-O-methyl-FD-892 into FD-891. In the course of the
GfsF reaction, 10-deoxy-FD-891 was isolated as an enzymatic
reaction intermediate, which was also converted into FD-891
by GfsF. Therefore, it was clearly found that the cytochro-
me P450 GfsF catalyzes epoxidation and hydroxylation in a
stepwise manner in the FD-891 biosynthesis. These results
clearly confirmed that the identified gfs genes are responsible
for the biosynthesis of FD-891 in S. graminofaciens.
Introduction
FD-891 is a 16-membered macrolide antibiotic isolated from
the culture broth of Streptomyces graminofaciens A-8890 that
is known to induce morphological change of human promye-
locytic leukemia (HL-60) cells.[1] FD-891 also showed strong cy-
tocidal activities against mice leukemia cells P388 and L1210,
but not antibacterial activities against Gram-positive and
Gram-negative bacteria. Among various human cancer cell
lines, four leukemia cell lines, HL-60, Jurkat, THP-1, and U-937
were highly sensitive to FD-891.[2] Further biochemical studies
of FD-891 in Jurkat cells suggested that FD-891 strongly indu-
ces apoptosis of Jurkat cells through mitochondrial release of
cytochrome c as well as caspase-9 processing triggered by cas-
pase-8 activation. This apoptosis-signaling pathway caused by
FD-891 was thus distinct from the other caspase-dependent
apoptosis by chemotherapic drugs such as doxorubicin, etopo-
side, and vincristine. Cell cycle arrest at G2/M phase by FD-891
was also suggested to contribute to the FD-891-induced apop-
tosis.[3] FD-891 also prevents cytotoxic T lymphocyte (CTL)-
mediated killing pathway as an immunosuppressive activity.[4]
However, in contrast to a structurally related concanamycin A,
which blocks the perforin-dependent killing pathway in CTL-
mediated cytotoxicity, FD-891 did not affect the vacuolar
acidification and only slightly decreased perforin activity. Fur-
ther biochemical studies illustrated that FD-891 inhibits CTL-
[a] Prof. Dr. T. Eguchi
Department of Chemistry and Materials Science, Tokyo Institute
of Technology 2-12-1 O-okayama, Meguro-ku, Tokyo 152-8551 (Japan)
Fax: (+81)3-5734-2631
E-mail: eguchi@cms.titech.ac.jp
[b] Dr. F. Kudo, A. Motegi
Department of Chemistry, Tokyo Institute of Technology
2-12-1 O-okayama, Meguro-ku, Tokyo 152-8551 (Japan)
[c] Dr. K. Mizoue
Taisho Pharmaceutical Co. Ltd.
1-403, Yoshino-cho, Kita-ku, Saitama, Saitama 331-9530 (Japan)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cbic.201000214.
1574
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 1574 – 1582

Biosynthesis of FD-891
target cell conjugate formation with the T-cell receptor (TCL)
down regulation. However, the molecular target(s) of FD-891
still remains unclear.
Results
Identification of the gene cluster
Structurally related FD-892, which lacks the epoxide at C8-
C9, the hydroxyl group at C10, and the 25-O-methyl group of
FD-891, was also isolated from the same microorganism and
showed the same induction of morphological change of HL-60
as FD-891.[1] FD-892 is thus presumed to be a biosynthetic in-
termediate of FD-891. However, FD-892 showed very weak cy-
tocidal activities against HL-60, P388, and L-1210, probably due
to the lack of an epoxide ring compared with FD-891. There-
fore, the molecular targets of FD-891 and FD-892 appear to be
partly different to induce distinct signaling pathways in HL-60.
Unfortunately, the FD-891 producer S. graminofaciens A-8890
does not accumulate FD-892 anymore under the laboratory
culture conditions, and the biochemical studies of the FD-892
are suspended at the moment.[5]
The FD-891 biosynthetic gene cluster was presumed to contain
typical modular type I PKS genes in Actinomycetes. Thus in
order to identify typical b-ketoacylsynthases (KS) genes, which
are highly conserved in many known type I PKSs for bacterial
macrolides such as erythromycin and rapamycin, PCR by using
the previously reported primers KSMA-F and KSMB-R[12] was
carried out by using genomic DNA of S. graminofaciens A-8890
as a template. The expected size of the DNA fragment was suc-
cessfully amplified and subsequently cloned by a standard ge-
netic operation. As we anticipated, the DNA sequences of the
PCR products showed significant similarity to the known type I
KS genes (data not shown). The cloned KS genes were then
used as a DNA probe to identify continuous PKS gene cluster
over 65 kbp, which was estimated from the number of the ex-
tension units of FD-891 by assuming the average size of a
gene for one module of type I PKS is 5 kbp. A pOJ446-based
cosmid library with the randomly Sau3AI digested genome
DNA of S. graminofaciens A-8890 was constructed and further
screened by hybridization with the DIG-labeled KS DNAs. As a
result, one of the KS gene probes gave a gene cluster contain-
ing total 69 kbp of five modular type I PKS genes (gfsA, B, C, D,
and E), a cytochrome P450 monooxygenase gene (gfsF), a
methyltransferase gene (gfsG), and a NarL family two-compo-
nent response regulator gene (gfsR),[13] which could be activat-
ed by a certain sensor protein in the microorganism (Figure 1
and Table 1). There is no open-reading frame (ORF) over 1 kbp
outside of the gene cluster; this suggests that this gene cluster
is one operon for the biosynthesis of this particular polyketide.
We then named this gene cluster as gfs cluster from S. grami-
nofaciens.
Although two groups established the total synthesis of FD-
891 and its analogue,[6,7] the bioengineering approach is an al-
ternative and attractive way to provide such structurally relat-
ed FD-891 analogues, including FD-892, for biochemical study
because the fermentation of the producer microorganisms is
well established. FD-891 is presumed to be constructed by typ-
ical modular type I polyketide synthase (PKS) followed by ep-
oxidation, hydroxylation, and methylation. Although genetic
manipulation of PKS genes is indeed an attractive methodolo-
gy to create novel polyketide molecules,[8,9] the genes for the
tailoring enzymes can be more easily inactivated to create bio-
synthetic intermediates that were not accumulated in the wild-
type producer strains.[10] Furthermore, such enzymes can be
utilized for the in vitro transformation to create diverse struc-
tures of analogous compounds. Glycosyltransferase is one of
the most applicable enzymes for the purpose, and thus glyco-
randomization methodology are widely accepted to create bio-
logically relevant glycosides.[11] If we can control the order of
the modification reactions in the FD-891 biosynthesis, seven
FD-891 analogues, including FD-892, can be logically created
through engineering of biosynthetic genes/enzymes. We thus
tried to clone the FD-891 biosynthetic genes and to produce
novel FD-891 analogues with biosynthetic technology.
PKS genes: gfsA, B, C, D, and E
Five PKS genes, gfsA, B, C, D, and E, were identified in the gfs
cluster spanning 68.5 kbp. The domain structures of each PKS
(GfsA, GfsB, GfsC, GfsD, and GfsE) were predicted by compari-
son with known modular type I PKSs through Pfam analysis[14]
as shown in Figure 2. As a result, the modular structure of the
PKSs (GfsA, 5 modules; GfsB, 3 modules; GfsC, 1 module; GfsD
2 modules; GfsE, 2 modules) was clearly determined. By com-
Figure 1. The 74.3 kb region of the Streptomyces graminofaciens A-8890 chromosome containing the FD-891 biosynthetic gene cluster, showing open reading
frames (ORFs) and cosmid boundaries.
ChemBioChem 2010, 11, 1574 – 1582
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
1575

T. Eguchi et al.
of FD-891, DH10 might not accept the elongated polyketide
intermediate having an S configuration of the a-substituent for
dehydration, and thus the skipped substrate would be pro-
cessed by the next module, module 11. The catalytic region of
KR10 domain has two unique amino acid residues compared
with the other B1 type KR domains (KR3, KR4, KR5, KR6, KR7,
KR8, KR9, KR11, and KR12) indicating a B2-type KR domain
(Supporting Information).[21] The KR1 domain belongs to the
A1-type KRs that produces 2R-methyl and 3S-hydroxyl configu-
ration and KR2 domain belongs to A2-type KRs that produces
the 2S-methyl and 3S-hydroxyl configuration. The stereochem-
istry of the enoyl reduction catalyzed by the ER5 domain was
also predicted to be the 2S configuration because the unique
tyrosine residue exists.[23] Overall, the domain structure of Gfs
PKSs agreed well with the chemical structure of FD-891, includ-
ing its stereochemistry. Therefore, the identified gfs gene clus-
ter was strongly supported to be responsible to the FD-891
biosynthesis.
Table 1. Deduced functions of ORFs in the FD-891 biosynthetic gene
cluster.
Gene Size[a] Homologue (biosynthetic product, Predicted function in
identity [%]/similarity [%])
FD-891 biosynthesis
gfsA 7480 polyketide synthase PimS1 (pimari- PKS loading, module 1–
cin, 54/66)
4
gfsB 5571 polyketide synthase NysJ (nystatin, PKS modules 5–7
51/62)
gfsC 2228 polyketide synthase Orf17 (ECO-
PKS module 8
02301, 63/74)
gfsD 3639 polyketide synthase PteA2 (filipin, PKS modules 9–10
53/65)
gfsD 3799 polyketide synthase Orf17 (ECO-
PKS modules 11–12
P450 monooxygenase
methyltransferase
02301, 51/63)
gfsE
gfsF
gfsG
414 P450 monooxygenase Mflv2418
(57/70)
278 Methyltransferase AtM (AT2433,
57/70)
243 NarL family two-component re-
sponse regulator
transcriptional regula-
tor
[a] Numbers are in amino acids.
Tailoring enzymatic genes: gfsF and gfsG
parison with the domain structures of the PKSs comprising of
b-ketosynthase (KS), acyltransferase (AT), b-ketoreductase (KR),
dehydratase (DH), enoylreductase (ER) and acyl carrier protein
(ACP), and the chemical structure of FD-892, the catalytic order
of the modular PKSs has been predicted as shown in Figure 2.
The probable substrate specificity of AT domains for malonyl
CoA (MT) or methylmalonyl CoA (MMT),[15,16] the existence of
KSQ domain[17] for lording of the starter unit in the front of the
first module in GfsA, and a termination domain thioesterase
(TE) in the tail of GfsE also agreed well with the chemical struc-
ture of FD-892. Docking domain analysis of PKSs based on the
indicative amino acid residues suggested specific interaction of
GfsA_C terminus (H1d)/GfsB_N terminus (T1d), GfsB_C terminus
(H1b)/GfsC_N terminus (T1b) and GfsD_C terminus (H1c)/GfsE_
N terminus (T1c).[18,19] GfsC_C terminus (H1d-like) and GfsD_
N terminus (T1b-like) showed a different type of combination
for the interaction, which should be distinguished from the
other PKS docking domains (see the Supporting Information).
In addition, the straightforward PKS gene organization, gfsA, B,
C, D, and E seemed to be reasonable to form the appropriate
quaternary structural organization of PKSs to construct this
particular polyketide molecule.
The deduced gfsG gene product showed moderate similarity
to various S-adenosyl methionine (SAM)-dependent methyl-
transferase, such as AtM in the indolocarbazoles antibiotic
AT2433 biosynthesis,[24] RebM in the rebeccamycin biosynthe-
sis,[25] MitM in the mitomycin C biosynthesis,[26] and AveD in
the avermectin biosynthesis.[27] Thus, GfsG seemed to catalyze
a methylation of the hydroxyl group at C-25 of FD-892 or the
other related biosynthetic intermediates.
The deduced gfsF gene product showed high similarity to
various cytochrome P450 in Actinomycetes and belongs to the
CYP105 family.[28] Cytochrome P450 contains a heme–iron com-
plex in the active site and catalyzes various oxidation reactions,
including hydroxylation and epoxidation. TylHI[29] and NysN[30]
belong to the CYP105 family and reportedly catalyze the oxi-
dations of polyketide skeletons in each biosynthetic pathway.
Thus, GfsF was also presumed to catalyze the oxidations in the
FD-891 biosynthesis, which should contain epoxidation at C8-
C9 and hydroxylation at C10. However, only GfsF was encoded
in the gfs gene cluster as a potential oxygenase; this suggests
that it might catalyze both epoxidation and hydroxylation. Al-
ternatively, GfsF could catalyze either of the oxidations, and an-
other enzyme outside of the gfs gene cluster might be respon-
sible for the remaining oxidation. To investigate these hypo-
thetical functions of GfsF, the gfsF gene inactivation study and
enzymatic characterization of the GfsF protein were carried
out.
One exception was found in module 10, because an unnec-
essary DH domain exists judging from the chemical structure
of FD-892. The functionality at the corresponding C7 position
should remain a hydroxyl group. However, the DH10 is quite
similar to the other active DH domains in Gfs PKSs and does
not show any indication of being an inactive DH domain. The
stereochemistry of a-substituent determined by the preceding
KR domain is known to be critical for the following dehydra-
tion reaction catalyzed by the DH domain.[20] Stereospecificity
of the KR domains in the front of the active DH domains have
been speculated by comparison with those of the other KR do-
mains indicating that 2R-methyl and 3R-hydroxyl functionality
could be formed.[21,22] Because KR10 should afford 2S-methyl
and 3R-hydroxyl functionality judging from the stereochemistry
Inactivation of the gfsF gene
According to a conventional method,[31] we inactivated the
gfsF gene in S. graminofaciens by the insertion of apramycin-re-
sistant aac(3)IV gene derived from pOJ446 (see the Experimen-
tal Section for details). The resultant DgfsF strain of S. gramino-
faciens did not produce FD-891; this confirms that this gene is
involved in the biosynthesis of FD-891 (Figure 3). Instead, this
mutant accumulated a FD-891-analogous compound as the
1576
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 1574 – 1582

Biosynthesis of FD-891
Figure 2. FD-891 modular polyketide synthase (PKS) encoded by gfsA–E, showing loading module and modules 1–12 and predicted ACP-bound polyketide in-
termediates. KS: b-ketosynthase, MT: malonyltransferase, MMT: methylmalonyltransferase, KR: b-ketoreductase, DH: dehydratase, ER: enoylreductase, ACP:
acyl carrier protein and TE: thioesterase. The DH10 domain of module 10 indicated as an asterisk should not work in the biosynthesis of FD-891.
major product, judging from the HPLC-photodiode array (PDA)
analysis. From the precipitate of 1.2 L culture of DgfsF strain,
280 mg of the novel compound was isolated by silica gel chro-
matography (2ꢁ), and its molecular formula was established
to be C33H54O6 by positive HR-FAB-MS indicating the loss of
two oxygen atoms from FD-891.
10.1 Hz, 1H), 5.35 (dd, J=8.0, 15.4 Hz, 1H), and 5.30 ppm (m,
1H)]. The methyl group at 3.34 ppm appeared to be methoxy
group. The 13C NMR and DEPT spectra exhibited 33 carbon res-
onances, consisting of three quaternary carbons, including an
ester carbonyl carbon (dC =169.6 ppm), and two olefinic car-
bons (dC =123.1 and 132.4 ppm), eight methyls including me-
thoxy carbon (dC =56.0 ppm), five methylenes, and seventeen
methines, including five oxymethines (dC =82.6, 78.8, 78.1,
76.4, and 72.9 ppm), and eight olefinic carbons (dC =144.27,
144.34, 133.2, 133.1, 132.7 130.6, 129.5, and 125.7 ppm). These
NMR spectral data of this compound are quite similar to those
of FD-892 except for the existence of methoxy group. Further
1
The H NMR spectrum showed three singlet methyl groups
[dH =1.95, 2.03, and 3.34 ppm], five doublet methyl groups
[dH =0.78 (d, J=7.0 Hz, 3H), 0.92 (d, J=7.1 Hz, 3H), 0.93 (d, J=
6.8 Hz, 3H), 1.15 (d, J=6.7 Hz, 3H), and 1.18 ppm (d, J=6.3 Hz,
3H)], and eight olefinic protons [dH =7.20 (m, 1H), 5.77 (ddd,
J=6.8, 8,4, 15.4 Hz, 1H), 5.53 (m, 2H), 5.47 (m, 1H), 5.40 (d, J=
ChemBioChem 2010, 11, 1574 – 1582
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
1577

T. Eguchi et al.
molecule, or one of these reactions. Thus, we attempted to ex-
press GfsF in E. coli and have investigated its enzymatic reac-
tion. The vast majority of the P450 enzymes require two-elec-
tron reduction by redox partner proteins such as NADPH and
NADH.[32,33] However, because no such protein was encoded at
the adjacent region of the gfs gene cluster in chromosome of
S. graminofaciens, a set of known heterologous putidaredoxin
and putidaredoxin reductase derived from Pseudomonas putida
were coexpressed with the gfsF gene to reconstitute the ex-
pected oxidation reactions.[34] This coexpression system report-
edly worked well with heterologous P450 enzymes.[35] The GfsF
expressing E. coli coexpressed with putidaredoxin and putidar-
edoxin reductase was reacted with 25-O-methyl-FD-892 in the
appropriate reaction buffer, and the reaction products were
analyzed by HPLC (Figure 5). As a result, the GfsF reaction af-
Figure 3. HPLC analysis of crude extracts from the culture of S. graminofa-
ciens. A) The ethyl acetate extract of precipitate from the wild-type S. grami-
nofaciens culture, and B) the ethyl acetate extract of precipitate from DgfsF
S. graminofaciens culture.
NMR spectroscopic analysis, including COSY, HMQC, and
HMBC, indicated that the chemical structure of the novel FD-
891 analogue from the gfsF gene-disrupted mutant was deter-
mined as 25-O-methyl-FD-892 (Figure 4A).
Figure 5. HPLC analysis of GfsF reaction from 25-O-methyl-FD-892 coupled
with putidaredoxin and putidaredoxin reductase.
forded FD-891 accompanied by a probable reaction intermedi-
ate. This result clearly verified that GfsF consecutively catalyzes
both epoxidation and hydroxylation as a dual functional P450
monooxygenase in the biosynthesis of FD-891. The novel FD-
891 analogue in the GfsF reaction was also isolated, and its
molecular formula was established to be C33H54O7 by positive
HR-FAB-MS; this indicates either an epoxidized or hydroxylated
compound.
1
The H NMR spectrum showed three singlet methyl groups
[dH =2.02, 2.08, and 3.34 ppm], five doublet methyl groups
[dH =0.79 (d, J=7.0 Hz, 3H), 0.90 (d, J=6.5 Hz, 3H), 0.92 (d, J=
7.1 Hz, 3H), 1.14 (d, J=7.2 Hz, 3H), and 1.18 ppm (d, J=6.3 Hz,
3H)], and six olefinic protons [dH =5.60 (m, 5H) and 7.32 ppm
1
(m, 1H)]. In the H NMR spectrum, two olefinic protons of 25-
O-methyl-FD-892 disappeared and instead two protons of oxy-
methines appeared at d=2.91 and 3.13 ppm indicating that
one double bond was epoxidized by GfsF. The 13C NMR spec-
trum exhibited 33 carbon resonances consisting of an ester
carbonyl carbon (dC =169.0 ppm), eight olefinic carbons (dC =
144.2, 142.1, 135.5, 132.9, 130.1, 129.5, 125.7, and 124.4 ppm),
eight oxygenated carbons (82.7, 78.4, 76.5, 73.1, 71.3, 57.2,
56.0, and 53.7 ppm), nine of which were methylenes and me-
thines (39.65, 39.63, 36.8, 36.0, 35.8, 34.6, 34.2, and 29.2 ppm),
and seven methyls (dC =16.6, 16.4, 16.2, 15.5, 12.6, 11.6, and
Figure 4. A) 25-O-methyl-FD-892 from DgfsF S. graminofaciens culture, and
B) 10-deoxy-FD-891 in the GfsF reaction. Arrows indicate the HMBC correla-
1
tions and bold lines indicate the spin couplings detected in the H,1H COSY
spectrum.
GfsF reaction with 25-O-methyl-FD-892
The 25-O-methyl-FD-892 that accumulated in the DgfsF strain
was presumed to be a substrate of GfsF P450 monooxygenase,
although it was still unclear whether GfsF catalyzes both the
epoxidation at C8ÀC9 and the hydroxylation at C10 of this
1578
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 1574 – 1582

Biosynthesis of FD-891
4.9 ppm); this suggests that one double bond was converted
to one epoxide functionality at d=57.2 and 53.7 ppm. Further
NMR spectroscopic analysis clearly indicated that the novel FD-
891 analogue in the GfsF reaction is 10-deoxy-FD-891 as
shown in Figure 4B.
The recognition mechanism of the dual functional P450
monooxygenase GfsF is an intriguing issue in the biosynthesis
of FD-891. GfsF first catalyzes the regio- and stereospecific ep-
oxidation of 25-O-methyl-FD-892 from the Si-face of C8–C9
double bond. Then, the pro-R hydrogen atom of 10-deoxy-FD-
891 is abstracted by GfsF to form a probable substrate hydox-
yl–iron–heme complex, which leads to FD-891. Initially, we
speculated that the C10 position would be hydroxylated at
first, because an allylic radical species is a likely intermediate.
However, this is not the case. The electron-rich olefinic bond
reacts with the ferryl-oxo intermediate (called compound I) in
the active site of GfsF at first leading to the epoxide formation.
The conformation of the first reaction product 10-deoxy-FD-
891 is somehow modulated and the pro-R hydrogen atom
would be stereospecifically abstracted by the ferryl-oxo inter-
mediate.[38] Therefore, overall, modulation of substrate confor-
mations appears to be critical for the stereospecific reactions
catalyzed by GfsF. Thus, the C10 carbon of 25-O-methyl-FD-892
should not be reacted with the ferryl-oxo intermediate in GfsF,
even though it can be closely positioned. Detailed study for
the recognition mechanism of this unique dual functional P450
monooxygenase GfsF is an intriguing issue in enzymatic
chemistry and should shed a light on the future enzymatic ap-
plication as a biocatalyst.[39,40]
Then, the isolated 10-deoxy-FD-891 was reacted again with
the GfsF reaction system, and this substrate was efficiently
converted to FD-891. These results clearly indicated that GfsF
first catalyzes the epoxidation at C8-C9 and then the hydroxyl-
ation at C10 to convert 25-O-methyl-FD-892 into FD-891 in a
stepwise manner. The stereochemistry of 10-deoxy-FD-891 and
25-O-methyl-FD-892 should be as same as those of FD-891.
Discussion
We identified the FD-891 biosynthetic gene cluster from the
producer strain S. graminofaciens A-8890 by screening with a
KS probe, which was turned out to be the KS7 domain in
module 7 (Figure 1 and Table 1). Among the PCR-amplified KS
genes with genomic DNA of S. graminofaciens as the template,
the KS3 domain was also found to be amplified by the PCR.
The same PCR also afforded a probable concanamycin PKS
gene (77% identity to ConE),[36] which was anticipated to be
amplified because this strain reportedly produces concanamy-
cin.[37] Three other KS genes were also amplified, indicating
that this strain has the potential to produce unidentified poly-
ketides.
MycG is a recently characterized dual-functional P450 mono-
oxygenase in the mycinamicin biosynthesis in Micromonospora
griseorubida.[41] MycG catalyzes stereospecific hydroxylation at
C14 of mycinamicin IV (M-IV) and epoxidation at C12–C13 of
M-V (Scheme 2). The order of oxidations catalyzed by MycG is
opposite to that of GfsF in FD-891 biosynthesis. Thus, in the
MycG reaction, conformation of the first hydroxylated product
M-V seems to be modulated to react with MycG again, result-
ing to the epoxidation at C12–C13 for completion of the M-II
biosynthesis. The apparently opposite reaction scheme ob-
served in the GfsF and MycG reactions is an intriguing issues
how the enzymes distinguish the structure of substrates. MycG
reportedly catalyzes the epoxidation at C12-C13 of M-IV to
give M-I, which is not converted to M-II by MycG (Scheme 2,
bottom). Therefore, only a subtle difference of conformation of
substrates and/or functional groups appears to be critical for
recognition by this unique enzyme. In addition, two desmety-
lated mycinosyl analogues M-III and M-VI cannot be accepted
by MycG, suggesting that the dimethyl functionality in the my-
cinose moiety of M-IV is critical for the recognition of MycG.
The modular structures of GfsA, GfsB, GfsC, GfsD, and GfsE
clearly indicated the proposed polyketide assembly line to con-
struct FD-892 as shown in Figure 2. Also, the gfsF gene inacti-
vation experiment and the GfsF enzymatic activity revealed the
order of post-PKS tailoring steps as shown in Scheme 1. Thus,
the hydroxyl group at C25 of FD-892 should be methylated by
a methyltransfearse GfsG to give 25-O-methyl-FD-892, which is
first epoxidized and then hydroxylated to afford FD-891 in a
stepwise manner by a P450 monooxygenase GfsF. In the pro-
posed biosynthetic pathway, it is still unclear how GfsF distin-
guishes between FD-892 and 25-O-methyl-FD-892, because the
oxidized moiety (C8, C9, and C10) is quite far from C25, which
is the only difference between these two molecules. To verify
the substrate specificity of GfsF, the gfsG gene-disrupted
mutant is currently under construction, assuming that the
mutant would accumulate FD-892 as discussed later.
Scheme 1. Proposed biosynthetic pathway for FD-891.
ChemBioChem 2010, 11, 1574 – 1582
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
1579

T. Eguchi et al.
Scheme 2. A dual functional P450 monooxygenase MycG involved in mycinamicin biosynthesis.
Accordingly, in the case of GfsF reaction, the methyl functional-
ity on the hydroxyl group at C25 of 25-O-methyl-FD-892 might
be important for the recognition of GfsF. Thus, we are now
constructing the methyltransferase gfsG gene disrupted
mutant to produce FD-892, which is not accumulated in wild
type and also express the GfsG to create various FD-891 ana-
logues by combination of genetic engineering and tailoring
enzymatic transformations.
ing to a standard method.[31] The FD-891 production medium con-
tained potato starch 3%, soya flake 1.5%, yeast extract 0.2%, corn
steep liquor 0.5%, NaCl 0.3%, 0.05% MgSO4·7H2O, 0.0005%
CoCl2·6H2O, and 0.3% CaCO3 (pH 7.1). The seed cultures of disrupt-
ed mutants (1 mL) were inoculated to the medium (100 mL) in a
500 mL baffled flask, and further cultivation was continued for five
days at 288C, 200 rpm.
Escherichia coli DH5a was routinely used for plasmid preparation.
E. coli XL-1 Blue MRF’ was used for cosmid manipulations. E. coli
ET12567 was used for preparation of nonmethylated plasmids for
transformation of S. graminofaciens. E. coli BL21(DE3) was used for
the gfsF gene expression. Luria–Bertani (LB) medium supplemented
with standard amounts of antibiotic as required (50 mgmLÀ1 ampi-
cillin for LITMUS28 and pT7-Blue) was used for the culture of
E. coli. Genetic manipulation in E. coli was carried out according to
a standard manner.[42]
Conclusions
We have produced two novel FD-891 analogues, 25-O-methyl-
FD-892 and 10-deoxy-FD-891 through biosynthetic technology.
These FD-891 analogues show unique bioactivity against
cancer cell lines and thus chemical biological studies of these
compounds are now in progress. Although the reaction order
of the oxidation catalyzed by P450 GfsF restricts the number
of novel FD-891 analogues, GfsF itself can be engineered to
change its specificity. For this purpose, a detailed enzymatic re-
action analysis of these biosynthetic enzymes would be neces-
sary. Studies of the relatively simple macrolide antibiotic FD-
891 biosynthetic pathway could provide an important solution
to other problems in the biosynthetic technology.
Preparation of DNA probe to identify the gfs gene cluster: Ge-
nomic DNA of S. graminofaciens A-8890 was used as a template for
the PCR amplification of typical type I PKS KS genes with KSMA 5’-
TSGCSATGGACCCSCAGCAG-3’
and
KSMB
5’-
CCSGTSCCGTGSGCCTCSAC-3’ as primers (S: a mixture of C and
G).[12] PCR conditions were 948C, 5 min for denaturing, 30 cycles of
948C, 30 s, 508C, 30 s, 728C, 40 s for extension of DNA in 2ꢁ GC
buffer I (5 mL), dNTP (1.6 mL; 2.5 mm each), 50 mm KSMA (0.2 mL ),
50 mm KSMB (0.2 ), distilled H2O (0.8 mL), LA-Taq DNA polymerase
(TaKaRa; 0.1 mL), and template DNA (2.1 mL; 2.4 ngmLÀ1 of chromo-
somal DNA). The amplified PCR products were subcloned with the
T vector of pT7-blue (TaKaRa) to obtain plasmids containing a par-
tial sequence of type I PKS KS genes. The DNA sequences of the
cloned plasmids were analyzed by using a DNA sequencer LONG
READER 4200 (Li-Cor) and SequiTherm EXCEL II DNA Sequencing
Kit-LC (for 66 cm gels, Epicentre Biotechnologies) with M13 Fwd
(À29)/IRDye 800 and M13 Reverse (À29)/IRDye 800 (Li-Cor) accord-
ing to the manufacturer’s protocol.
Experimental Section
Bacterial strains and growth conditions: Streptomyces graminofa-
ciens A-8890 was used as the FD-891 producer strain and source of
the FD-891 biosynthetic (gfs) genes.[1] S. graminofaciens A-8890 was
maintained on ISP2 (0.4% yeast extract, 1% malt extract, 0.4% glu-
cose, pH 7.3) agar medium (2.0% agar, 288C, 3 d) and the spores
were used for a seed culture in ISP2 liquid medium (288C,
200 rpm, 4–6 d). Genomic DNA was extracted from the cultured
cells in 100 mL of ISP2 liquid medium (288C, 200 rpm, 4 d) accord-
Identification of the FD-891 gene (gfs) cluster: The chromosomal
DNA of S. graminofaciens was partially digested with Sau3AI and
1580
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 1574 – 1582

Biosynthesis of FD-891
the restriction enzyme was denatured with the phenol-chloroform
extraction. After ethanol precipitation, the digested DNA fragments
of the chromosomal DNA were treated with calf intestine alkaline
phosphatase (CIAP) at 508C and CIAP were denatured with the
phenol–CHCl3extraction. A cosmid vector pOJ446 was separately
digested with HpaI, and after the treatment of CIAP, followed by di-
gestion with BamHI. After the phenol–CHCl3 extraction and the
EtOH precipitation, the resulting vector DNA was dissolved with
the TE buffer. The digested pOJ446 and the partially digested chro-
mosomal DNA were ligated by DNA Ligation kit ver. 2 (TaKaRa) at
48C overnight. After EtOH precipitation, the ligated DNA was dis-
solved in the TE buffer. The resulting ligation mixture was pack-
aged into l phage followed by phage transfection to E. coli XL1
Blue MRF’ by using with Gigapack III XL Packaging Extract (Strata-
gene) according to the manufacturer’s protocol. The host strain
E. coli XL1 Blue MRF’ was cultured in LB (3 mL) containing 10 mm
MgSO4 and 0.2% maltose by OD600 0.5–1. The culture was centri-
fuged, and the resulting wet cells were suspended in a 10 mm
MgSO4 up to OD600 =1 for transfection.
with SphI and BamHI, Klenow fragment, and bacterial alkaline
phosphatase (BAP) in advance. The resultant plasmid pW-D2N3-Bs
was digested with PstI and further treated with the Klenow frag-
ment and BAP. An apramycin-resistance gene aac(3)IV cassette was
inserted into the blunt-ended site, to obtain plasmid pWHM3-df.
An EcoRI-PstI digested fragment containing the aac(3)IV gene de-
rived from pOJ446 was subcloned into the corresponding site of
pBluescript II KS+. From the plasmid, EcoRV-SmaI fragment was re-
covered and used as the aac(3)IV cassette.
Non-methylated plasmid pWHM3-df was recovered from E. coli
ET12567 transformant and used for transformation of S. graminofa-
ciens. Preparation of protoplasts of S. graminofaciens and transfor-
mation were performed according to
a standard protoplast
method.[31] The transformants of S. graminofaciens harboring
pWHM3-df were inoculated on R5 agar medium containing
40 mgmLÀ1 apramycin and 16 mgmLÀ1 thiostrepton at 288C. Appro-
priate transformants were then cultured in ISP2 liquid medium
(3 mL) containing 5 mgmLÀ1 of thiostrepton at 288C for six days.
The seed cultures were spread on R5 agar medium containing
50 mgmLÀ1 of apramycin. After 10 days’ culture at 288C, several
colonies were inoculated in 3 mL of SK No.2 liquid medium con-
taining 12.5 mgmLÀ1 of apramycin at 288C for 5 d. The cultures
were again spread on R5 agar medium containing 50 mgmLÀ1 of
apramycin and further cultured for 4 d at 288C; 148 colonies were
screened on R5 agar medium containing 50 mgmLÀ1 of apramycin
or 25 mgmLÀ1 of thiostrepton or no antibiotics. As a result, 32 thio-
strepton-sensitive and apramycin-resistant strains were obtained.
The BamHI digested genomic DNAs of the strains were used for
Southern hybridization with a DIG-labeled probe to confirm the in-
sertion of aac(3)IV cassette in the expected site by double cross-
over homologous recombination. The probe DNA was derived
from 865 bp of ApaI-digested DNA fragment in the gfsF gene. The
desired gfsF gene mutants of S. graminofaciens were stored as
spore suspension at À308C.
A cosmid library of 2960 clones in E. coli grown on 50 mgmLÀ1 ge-
neticin-containing LB medium was screened by the hybridization
with a Dig-labeled DNA probe, which was made by the DIG DNA
Labeling Kit (Roche). Among several KS genes obtained by using
the above-mentioned PCR, KS5 (which was turned out to be a par-
tial KS gene in module 7 of gfsB) was used as a DNA probe to iden-
tify type I PKS gene containing clusters. Hybridization was carried
out with the DIG Nucleic Acid Detection Kit and NBT/BCIP solution
(Roche) according to the manufacturer’s protocol.
Thirty-eight positive clones were cultured and the corresponding
cosmids were extracted by using a standard protocol. The cosmids
were further screened by using the above-mentioned PCR and
Southern hybridization with the same probe to obtain target
cosmid cgra05, which contained the target partial KS gene. Over-
lapping cosmids, cgra09, cgra01, cgraU04, and cgraD02 spanning
an entire PKS gene cluster were subsequently identified by chro-
mosomal walking.
Isolation of 25-O-methyl-FD-892 from the gfsF gene mutant:
Culture broth (1.2 L) of DgfsF was centrifuged to separate superna-
tant and precipitate at 7000 rpm for 20 min. The precipitate was
extracted with acetone. After removal of acetone by rotary evapo-
ration, the aqueous residue was extracted with EtOAc. The com-
bined organic layer was dried on anhyd Na2SO4, and the solvent
was removed by rotary evaporation. The extract (yellow syrup) was
purified twice by silica gel chromatography (silica gel 60N, 63–
210 mm; Kanto Chemical, Tokyo, Japan; first: 1.5ꢁ10 cm gel with
hexane/acetone (1:1), second 1.5ꢁ20 cm gel with hexane/EtOAc
(1:4)) to obtain 282 mg of 25-O-methyl-FD-892. 1H and 13C NMR
spectra were taken by DRX-500 (Bruker) or ECX500 (JEOL) spec-
trometer, and FAB-MS spectra were measured by JMS-700 (JEOL).
The assignments of 1H and 13C NMR spectroscopy signals are
shown in the Supporting Information. HR-FAB-MS (positive, glycer-
ol): m/z: calcd for C33H55O6: 547.3999 [M+H]+, found: 547.3979.
Sequencing and analysis: Because cosmid cgra09 (3663~41679)
and cgra01 (29211~72473) seemed to contain the majority of a
gene cluster (74321 bp), each of the two cosmids was randomly
sequenced by a shotgun sequence method on double-stranded
DNA templates with more than tenfold coverage and a minimum
of three times each portion of the DNA sequence (Shimadzu Bio-
tech, Kyoto, Japan). Further, several DNA fragments from cgraU04
(U04KpnI04, U04EcoRI05, U04BssHII03, and U04NcoI03) and
cgraD02 (D02PstI04, D02BamHI09, D02BamHI10, and D02KpnI06)
were subcloned into the plasmid vector LITMUS28 and sequenced
on double-stranded DNA templates by using a LONG READER 4200
(Li-Cor). As a result, a 74321 bp of DNA sequence containing a
type I PKS gene cluster was determined. ORFs were determined by
FramePlot
analysis
(http://www.nih.go.jp/~jun/cgi-bin/frame-
plot.pl)[43] and BLAST homology search by using the NCBI BLAST
server. The domain structure of PKS was analyzed by Pfam pro-
gram[14] by using the Sanger institute server (http://pfam.sanger.a-
c.uk/). The determined DNA sequence data of the gene cluster in
S. graminofaciens A-8890 has been deposited to the DDBJ databas-
es under accession number AB469193.
Expression of GfsF: The gfsF gene was amplified by PCR with the
primers 891P450-N: 5’-AACATATGACCGACACGACACTC-3’ and
891P450-C: 5’-GAA CTA GTG GCA GGG CGG GGC-3’. PCR conditions
were 30 cycles of 988C, 10 s, 508C, 5 s, 728C, 90 s for extension of
DNA in 5ꢁ PrimeSTAR buffer (2 mL), 0.8 mL of dNTP (2.5 mm each),
DMSO (0.5 mL), 0.1 mm 891P450-N (0.1 mL), 0.1 mm 891P450-C
(0.1 mL), PrimeSTAR polymerase (TaKaRa; 0.1 mL ), template DNA
(0.1 mL; gfsF gene containing plasmid derived from cgraD02,
2.2 mgmLÀ1), and H2O (6.3 mL). The amplified PCR product was sub-
cloned into the LITMUS28. After digestion of LT28 with EcoRV, we
dephosphorylated the vector with BAP. After confirmation of the
DNA sequence, an appropriate plasmid was digested with NdeI
Inactivation of the gfsF gene: NcoI digested fragment (7.5 kbp)
from cgraD02 was subcloned into LITMUS28. From the plasmid,
3.4 kb of the BssHII-digested fragment was recovered and treated
with Klenow fragment. The resultant blunt-ended DNA was ligated
with E. coli--Streptomyces shuttle vector pWHM3, which was treated
ChemBioChem 2010, 11, 1574 – 1582
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
1581

T. Eguchi et al.
[10] C. Olano, C. Mendez, J. A. Salas, Nat. Prod. Rep. 2010, 27, 571–616.
and SpeI, and the resulting DNA fragment was inserted into the
corresponding site of pCYP-camAB[44] to obtain pGfsF-camAB. The
pGfsF-camAB was introduced into E. coli BL21(DE3) by a standard
chemical transformation. The E. coli harboring pGfsF-camAB was
precultured in LB medium (3 mL) with 100 mgmLÀ1 of ampicillin at
378C overnight and then 1 mL of the culture was used for the
main culture in M9 mix medium (6.78 gLÀ1 Na2HPO4, 3 gLÀ1
KH2PO4, 0.5 gLÀ1 NaCl, 1 gLÀ1 NH4Cl, 1% casamino acid, 0.4% glu-
cose, 0.1 mm CaCl2, 1 mm MgCl2 were autoclaved and then mixed
with 0.1 mm FeSO4 20 mgLÀ1 thymine, 80 mgLÀ1 5-aminolevulinic
acid) supplemented with 100 mgmLÀ1 ampicillin. The main culture
was carried out at 378C by OD600 =0.7, and a final 0.1 mm isopro-
pyl b-d-thiogalactoside was then added for induction of overex-
pression. The culture was continued at 228C overnight and the
cells were harvested by centrifugation (2300g for 10 min). The wet
cells (2.3 g from 100 mL culture) were suspended in CV-2 mm DTT
buffer [25 mL; 50 mm NaH2PO4, 1 mm EDTA, 10% glycerol, 1 mm
glucose (pH 7.3) were autoclaved, and then mixed with 2 mm DTT]
and were stored at À808C.
GfsF reaction with 25-O-Methyl-FD-892: 25-O-Methyl-FD-892
(10 mg) was dissolved in DMSO (1 mL) and added to the above-
mentioned GfsF-containing reactant (100 mL) in a baffled flask.
After 24 h incubation at 288C, 200 rpm, the reaction product was
extracted with EtOAc (3ꢁ100 mL). The combined organic layers
were dried on anhyd Na2SO4, and the solvent was removed by
rotary evaporator. The crude extract was purified by a standard
silica gel chromatography (1.0ꢁ20 cm gel with hexane/acetone
(1:1)) and a preparative TLC with hexane/EtOAc (1:4; Rf =0.3) to
obtain 2.6 mg of 10-deoxy-FD-891. The assignments of 1H and
13C NMR signals are shown in the Supporting Information. HR-FAB-
MS (positive, glycerol): m/z: calcd for C33H55O7: 563.3948 [M+H]+,
found: 563.3942.
[11] C. J. Thibodeaux, C. E. Melancon, III, H. W. Liu, Angew. Chem. 2008, 120,
9960–10007; Angew. Chem. Int. Ed. 2008, 47, 9814–9859.
[12] M. Otsuka, K. Ichinose, I. Fujii, Y. Ebizuka, Antimicrob. Agents Chemother.
2004, 48, 3468–3476.
[13] A. E. Maris, M. R. Sawaya, M. Kaczor-Grzeskowiak, M. R. Jarvis, S. M. Bear-
son, M. L. Kopka, I. Schroder, R. P. Gunsalus, R. E. Dickerson, Nat. Struct.
Biol. 2002, 9, 771–778.
[14] R. D. Finn, J. Mistry, J. Tate, P. Coggill, A. Heger, J. E. Pollington, O. L.
Gavin, P. Gunasekaran, G. Ceric, K. Forslund, L. Holm, E. L. Sonnhammer,
S. R. Eddy, A. Bateman, Nucleic Acids Res. 2010, 38, D211–222.
[15] G. Yadav, R. S. Gokhale, D. Mohanty, J. Mol. Biol. 2003, 328, 335–363.
[16] L. Tang, Y. J. Yoon, C. Y. Choi, C. R. Hutchinson, Gene 1998, 216, 255–
265.
[17] C. Bisang, P. F. Long, J. Cortes, J. Westcott, J. Crosby, A. L. Matharu, R. J.
Cox, T. J. Simpson, J. Staunton, P. F. Leadlay, Nature 1999, 401, 502–505.
[18] M. Thattai, Y. Burak, B. I. Shraiman, PLoS Comput. Biol. 2007, 3, 1827–
1835.
[19] K. J. Weissman, ChemBioChem 2006, 7, 485–494.
[20] A. Keatinge-Clay, J. Mol. Biol. 2008, 384, 941–953.
[21] A. T. Keatinge-Clay, Chem. Biol. 2007, 14, 898–908.
[22] C. R. Valenzano, R. J. Lawson, A. Y. Chen, C. Khosla, D. E. Cane, J. Am.
Chem. Soc. 2009, 131, 18501–18511.
[23] D. H. Kwan, Y. Sun, F. Schulz, H. Hong, B. Popovic, J. C. Sim-Stark, S. F.
Haydock, P. F. Leadlay, Chem. Biol. 2008, 15, 1231–1240.
[24] Q. Gao, C. Zhang, S. Blanchard, J. S. Thorson, Chem. Biol. 2006, 13, 733–
743.
[25] H. Onaka, S. Taniguchi, Y. Igarashi, T. Furumai, Biosci. Biotechnol. Bio-
chem. 2003, 67, 127–138.
[26] Y. Mao, M. Varoglu, D. H. Sherman, Chem. Biol. 1999, 6, 251–263.
[27] H. Ikeda, T. Nonomiya, M. Usami, T. Ohta, S. Omura, Proc. Natl. Acad. Sci.
USA 1999, 96, 9509–9514.
[28] D. C. Lamb, H. Ikeda, D. R. Nelson, J. Ishikawa, T. Skaug, C. Jackson, S.
Omura, M. R. Waterman, S. L. Kelly, Biochem. Biophys. Res. Commun.
2003, 307, 610–619.
[29] R. Fouces, E. Mellado, B. Diez, J. L. Barredo, Microbiology 1999, 145,
855–868.
[30] T. Brautaset, O. N. Sekurova, H. Sletta, T. E. Ellingsen, A. R. Strøm, S. Valla,
S. B. Zotchev, Chem. Biol. 2000, 7, 395–403.
[31] T. Kieser, M. J. Bibb, M. J. Buttner, K. F. Chater, D. A. Hopwood, Practical
Streptomyces Genetics, The John Innes Foundation, Norwich, 2000.
[32] A. W. Munro, H. M. Girvan, K. J. McLean, Nat. Prod. Rep. 2007, 24, 585–
609.
[33] I. G. Denisov, T. M. Makris, S. G. Sligar, I. Schlichting, Chem. Rev. 2005,
105, 2253–2277.
[34] J. A. Peterson, M. C. Lorence, B. Amarneh, J. Biol. Chem. 1990, 265,
6066–6073.
Acknowledgements
This work was supported in part by the Uehara Memorial Foun-
dation (T.E.) and Naito Foundation (T.E. and F.K.). The putidare-
doxin and putidaredoxin reductase coexpression system was
kindly gifted by Dr. Arisawa at Mercian Corporation.
[35] Y. Fujii, S. Hirosue, T. Fujii, N. Matsumoto, H. Agematu, A. Arisawa, Biosci.
Biotechnol. Biochem. 2006, 70, 2299–2302.
[36] S. F. Haydock, A. N. Appleyard, T. Mironenko, J. Lester, N. Scott, P. F. Lea-
dlay, Microbiology 2005, 151, 3161–3169.
Keywords: biosynthesis
macrolides · polyketides
· cytochromes · gene clusters ·
[1] M. Seki-Asano, T. Okazaki, M. Yamagishi, N. Sakai, K. Hanada, K. Mizoue,
J. Antibiot. 1994, 47, 1226–1233.
[37] M. Seki-Asano, Y. Tsuchida, K. Hanada, K. Mizoue, J. Antibiot. 1994, 47,
1234–1241.
[2] S. Inaba, T. Eguchi, A. Motegi, K. Mizoue, T. Usui, K. Nagai, T. Kataoka, J.
Antibiot. 2009, 62, 507–512.
[3] J. Garcia-Fortanet, J. Murga, M. Carda, J. A. Marco, R. Matesanz, J. F.
Diaz, I. Barasoain, Chem. Eur. J. 2007, 13, 5060–5074.
[4] T. Kataoka, A. Yamada, M. Bando, T. Honma, K. Mizoue, K. Nagai, Immu-
nology 2000, 100, 170–177.
[5] T. Eguchi, K. Yamamoto, K. Mizoue, K. Kakinuma, J. Antibiot. 2004, 57,
156–157.
[6] J. Garcia-Fortanet, J. Murga, M. Carda, J. A. Marco, Org. Lett. 2006, 8,
2695–2698.
[38] J. A. Krauser, F. P. Guengerich, J. Biol. Chem. 2005, 280, 19496–19506.
[39] V. B. Urlacher, S. Eiben, Trends Biotechnol. 2006, 24, 324–330.
[40] R. Bernhardt, J. Biotechnol. 2006, 124, 128–145.
[41] Y. Anzai, S. Li, M. R. Chaulagain, K. Kinoshita, F. Kato, J. Montgomery,
D. H. Sherman, Chem. Biol. 2008, 15, 950–959.
[42] J. Sambrook, E. F. Fritsch, T. Maniatis, Molecular Cloning. A Laboratory
Manual, 2nd ed., Cold Spring Harbor Laboratory Press, New York, 1989.
[43] J. Ishikawa, K. Hotta, FEMS Microbiol. Lett. 1999, 174, 251–253.
[44] H. Agematu, N. Matsumoto, Y. Fujii, H. Kabumoto, S. Doi, K. Machida, J.
Ishikawa, A. Arisawa, Biosci. Biotechnol. Biochem. 2006, 70, 307–311.
[7] M. T. Crimmins, F. Caussanel, J. Am. Chem. Soc. 2006, 128, 3128–3129.
[8] C. D. Reeves, E. Rodriguez, Methods Enzymol. 2009, 459, 295–318.
[9] S. Peiru, H. Gramajo, H. G. Menzella, Methods Enzymol. 2009, 459, 319–
337.
Received: April 7, 2010
Published online on June 29, 2010
1582
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 1574 – 1582