Geosmin biosynthesis. Streptomyces coelicolor germacradienol/germacrene D synthase converts farnesyl diphosphate to geosmin

Article abstract of DOI:10.1021/ja062669x

Geosmin is responsible for the characteristic odor of moist soil. Incubation of recombinant germacradienol synthase, encoded by the SCO6073 (SC9B1.20) gene of the Gram-positive soil bacterium Streptomyces coelicolor, with farnesyl diphosphate (2, FPP) in the presence of Mg²⁺ gave a mixture of (4S,7R)-germacra-1(10)E,5E-diene-11-ol (3) (74%), (-)-(7S)-germacrene D (4) (10%), geosmin (1) (13%), and a hydrocarbon, tentatively assigned the structure of octalin 5 (3%). Individual incubations of recombinant germacradienol synthase with [1,1-²H₂]FPP (2a), (1R)-[1-²H]-FPP (2b), and (1S)-[1-²H]-FPP (2c), as well as with FPP (2) in D₂O, and GC-MS analysis of the resulting deuterated products supported a mechanism of geosmin formation involving proton-initiated cyclization and retro-Prins fragmentation of the initially formed germacradienol to give intermediate 5, followed by protonation of 5, 1,2-hydride shift, and capture of water. Copyright

Full text of DOI:10.1021/ja062669x

Published on Web 06/01/2006
Geosmin Biosynthesis. Streptomyces coelicolor Germacradienol/Germacrene
D Synthase Converts Farnesyl Diphosphate to Geosmin
Jiaoyang Jiang, Xiaofei He, and David E. Cane*
Department of Chemistry, Box H, Brown UniVersity, ProVidence, Rhode Island 02912-9108
Received April 17, 2006; E-mail: david_cane@brown.edu
Scheme 1
The characteristic odor of moist soil is due to the presence of
two terpene derivatives, methyl isoborneol and geosmin (1).1,2
Produced in soil primarily by streptomycetessubiquitous Gram-
positive, filamentous, saprophytic bacteriasgeosmin has also been
isolated from a variety of cyanobacteria, fungi, and liverworts.3
Geosmin has an exceptionally low detection threshold of the order
of 10-100 ppt. Besides its pleasant, characteristic earthy aroma,
geosmin is also associated with an undesirable musty odor or off
flavor in drinking water, as well as in wine, fish, and other food
stuffs, making its detection and elimination important in the
management of water and food quality.4
Scheme 2. Mechanism of Cyclization of FPP (2) to
Germacradienol (3), Germacrene D (4), Hydrocarbon 5, and
Geosmin (1)
The structure of geosmin (1) was first established as trans-
1,10-dimethyl-trans-9-decalol by Gerber1c who detected the volatile
oil in 17 different species of Streptomyces and a blue green alga
following its initial isolation from S. griseus.1 Shortly thereafter,
Bentley provided evidence that the C12 metabolite geosmin was
likely a degraded sesquiterpene, based on the apparent incorporation
of both [1-14C]- and [2-14C]acetate into geosmin by strains of S.
antibioticus.5
We recently reported expression in Escherichia coli of a 2181-
bp gene from S. coelicolor A3(2) (SCO6073, SC9B1.20) to give a
726 amino acid protein with significant similarity in both the
N-terminal and C-terminal halves to the well-characterized ses-
quiterpene synthase, pentalenene synthase.6 The full-length recom-
binant protein was shown to catalyze the Mg2+-dependent conver-
sion of farnesyl diphosphate (FPP, 2) to a 85:15 mixture of (4S,7R)-
germacra-1(10)E,5E-diene-11-ol (3) and the sesquiterpene hydro-
carbon (-)-(7S)-germacrene D (4), with a kcat of 6 × 10-3 s-1 and
a Km for FPP of 60 nM (Scheme 1).6-8 Both products result from
partitioning of a common intermediate, with formation of germa-
cradienol involving loss of the H-1si proton of FPP (2), H-1re being
retained at C-6 of 3, while formation of germacrene D (4) involves
a competing 1,3-hydride shift of H-1si to the isopropyl side chain
(Scheme 2).7 Expression of the N-terminal (366 amino acid) domain
of the SCO6073 protein gave a fully functional germacradienol
synthase with steady-state catalytic parameters similar to those of
the full-length protein.6 The expressed C-terminal domain of
SC06073 had no detectable FPP cyclase activity.6
Independently, Chater and co-workers reported that deletion of
the entire SCO6073 gene from S. coelicolor A3(2) results in the
complete loss of geosmin production.9 In-frame deletion of the
N-terminal domain of SCO6073 also abolished geosmin formation,
while deletion of only the C-terminal domain had no apparent effect
on geosmin biosynthesis. The combined biochemical and molecular
genetic evidence thus established conclusively that the SCO6073
gene encodes a germacradienol synthase and that this enzyme
catalyzes an essential step in the biosynthesis of geosmin. We
originally proposed that formation of geosmin (1) from germacra-
dienol (3) would involve a multistep biochemical sequence requiring
both oxidative and reductive transformations catalyzed by several
hypothetical enzymes.6 Multistep redox pathways have also been
suggested by other investigators.3 We now report that germacra-
dienol/germacrene D synthase itself generates geosmin (1), thereby
establishing that the biosynthesis of geosmin from farnesyl diphos-
phate is catalyzed by a single enzyme, without intervention of any
additional enzymes and without any requirement for redox cofactors.
GC-MS analysis of the products resulting from incubation of
FPP with 2.3 µM recombinant S. coelicolor germacradienol synthase
(4 h, 30 °C) revealed the presence of a minor (2.5%) but
reproducible peak, retention time, 9.79 min, m/z 182, identical in
GC retention time and mass spectrum with (-)-geosmin (1),
obtained from the volatile extract of S. coelicolor, as well as with
synthetic (()-geosmin.10,11 An additional 1% component of the
reaction mixture, retention time, 8.51 min, m/z 164, was tentatively
assigned the octalin structure 5.12 Interestingly, increasing the
concentration of cyclase from 2.3 to 9.2 µM increased not only
the total yield of geosmin (1) and hydrocarbon 5 but also the relative
proportion of both 1 and 5 (13 and 3.5%, respectively), with a
concomitant decrease in germacradienol (3) from 86 to 74% of the
total product mixture. By contrast, the relative proportion of
germacrene D (4) remained constant at 10-11%, independent of
protein concentration. As already established for generation of
germacradienol and germacrene D,6,7 the enzymatic conversion of
FPP to geosmin (1) showed an absolute requirement for a divalent
cation, Mg2+ being strongly preferred. Notably, Fe2+ or Cu2+, alone
or in binary combination with Mg2+, each suppressed the formation
of geosmin by a factor of 3 or more, thus ruling out any requirement
for either redox-active metal ion. No other organic or inorganic
cofactor is required for the formation of geosmin from FPP.
9
8128
J. AM. CHEM. SOC. 2006, 128, 8128-8129
10.1021/ja062669x CCC: $33.50 © 2006 American Chemical Society

C O M M U N I C A T I O N S
incubations, was a mixture of 6% d2-, 36% d1-, and 58% d0-species.
The deuterium isotope effect also resulted in a perturbation of the
normal product ratios, with an enhancement of the proportion of
germacrene D (4d) (24%) and hydrocarbon 5d (8%), and a decrease
in the proportions of germacradienol (3d) and geosmin (1d) (65
and 5%, respectively.)
Figure 1. MS fragmentation of geosmin (1).
The above-described results provide strong experimental evidence
for the previously inferred intermediacy of germacradienol (3) in
geosmin biosynthesis. The demonstration that a single enzyme is
both necessary and sufficient to convert farnesyl diphosphate to
geosmin solves a long-standing biosynthetic mystery, while expand-
ing our knowledge of the already impressive synthetic virtuosity
of terpene synthases. Experiments to establish further details of
this remarkable biochemical reaction are in progress.
Scheme 3. Cyclization of FPP (2) to 1, 3, 4, and 5 in D2O
(predicted labeling)
Acknowledgment. GC-MS analysis of the volatile extract of
S. coelicolor was carried out by Xin Lin. We thank Dr. Tun-Li
Shen for assistance with the mass spectrometric analysis, and Dr.
Pascal Gosselin for a gift of synthetic (()-geosmin. This research
was supported by NIH Grant GM30301 to D.E.C.
To explore the mechanism of this unexpected transformation,
we carried out incubations of recombinant germacradienol synthase
with individual deuterated samples of FPP and analyzed the
resulting mixtures of 1, 3, 4, and 5 by GC-MS.7,11 Cyclization of
[1,1-2H2]FPP (2a)13 or (1R)-[1-2H]FPP (2b)13 each gave [6-2H]-
germacradienol (3a and 3b) as previously reported,7 as well as both
geosmin-d1 (1a and 1b; m/z 183) and octalin-d1 (5a and 5b; m/z
165) (Scheme 2). The deuterium atom in both 1a and 1b, predicted
to be at C-6, could be localized in ring B, on the basis of the
observed shift from m/z 112 to 113 of the characteristic ring B MS
fragment ion (Figure 1).3 The corresponding ring A-derived
fragment ion from either 1a or 1b, m/z 126, was devoid of
deuterium. As expected, germacradienol (3c), geosmin (1c), and
hydrocarbon 5c derived from (1S)-[1-2H]FPP (2c)13 were all
unlabeled, consistent with the previously established loss of H-1si
of FPP in the formation of germacradienol. The retention and
distribution of label in the samples of germacrene D (4a, 4b, 4c)
obtained from each incubation were all as previously reported.7
Although the formation of geosmin (1) and hydrocarbon 5 from
FPP, catalyzed by a single terpene synthase, has no biochemical
precedent, the observed labeling results can be fully rationalized
by the mechanism illustrated in Scheme 2. Proton-initiated cycliza-
tion of germacradienol followed by a retro-Prins-type fragmentation
of the resulting eudesmanoid cation will lead to release of the
2-propanol side chain, presumably as acetone, and generation of
the trisnorsesquiterpene 5. Protonation of 5, followed by a 1,2-
hydride shift and capture of the bridgehead cation by water, will
generate geosmin (1).14 The observation that the proportion of
geosmin is enhanced at increased concentrations of protein suggests
that a substantial proportion of the initially generated germacra-
dienol dissociates from the enzyme before being rebound and further
converted to 5 and geosmin.
To test further the proposed cyclization mechanism, we carried
out a series of incubations with FPP and recombinant germacra-
dienol synthase in D2O (∼95 atom 2H) using selected ion monitor-
ing (SIM) to quantitate the relevant M, M + 1, M + 2, and M +
3 peaks corresponding to each incubation product (Scheme 3).11
The resulting germacradienol (3d) consisted of 22% d1-species and
77% d0, apparently due to a pronounced isotope effect on the first
protonation step. Geosmin (1d) obtained from each incubation
consisted of 13% d3-, 58% d2-, 26% d1, and 3% d0-species,
consistent with the required three protonation steps. Notably, only
one of the three deuterium atoms is located in ring B of geosmin
(3d), as evidenced by the observation of the predicted d1-fragment
at m/z 113.14 Hydrocarbon 5d, resulting from the same D2O
Supporting Information Available: Experimental methods, incu-
bation conditions, and GC-MS data. This material is available free of
charge via the Internet at http://pubs.acs.org.
References
(1) (a) Gerber, N. N. CRC Crit. ReV. Microbiol. 1979, 7, 191-214. (b) Gerber,
N. N.; Lechevalier, H. A. Appl. Microbiol. 1965, 13, 935-938. (c) Gerber,
N. N. Tetrahedron Lett. 1968, 2971-2974.
(2) Buttery, R. G.; Garibaldi, J. A. J. Agric. Food Chem. 1976, 24, 1246-
1247.
(3) For leading references, see: (a) Spiteller, D.; Jux, A.; Piel, J.; Boland,
W. Phytochemistry 2002, 61, 827-834. (b) Dickschat, J. S.; Bode, H.
B.; Mahmud, T.; Muller, R.; Schulz, S. J. Org. Chem. 2005, 70, 5174-
5182.
(4) (a) Darriet, P.; Pons, M.; Lamy, S.; Dubourdieu, D. J. Agric. Food. Chem.
2000, 48, 4835-4838. (b) Heil, T. P.; Lindsay, R. C. J. EnViron. Sci.
Health B 1988, 23, 489-512.
(5) Bentley, R.; Meganathan, R. FEBS Lett. 1981, 125, 220-222.
(6) Cane, D. E.; Watt, R. M. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 1547-
1551.
(7) He, X.; Cane, D. E. J. Am. Chem. Soc. 2004, 126, 2678-2679.
(8) Germacradienol (3) was first isolated from S. citreus as a cometabolite of
geosmin (1), accompanied by several germacrenoid and eudesmanoid
metabolites, including germacrene D (4), bicyclogermacrene, and dihy-
droagarofuran. (a) Gansser, D.; Pollak, F. C.; Berger, R. G. J. Nat. Prod.
1995, 58, 1790-1793. (b) Pollak, F. C.; Berger, R. G. Appl. EnViron.
Microbiol. 1996, 62, 1295-1299. 3, 4, and 1 frequently co-occur in
Streptomyces.
(9) Gust, B.; Challis, G. L.; Fowler, K.; Kieser, T.; Chater, K. F. Proc. Natl.
Acad. Sci. U.S.A. 2003, 100, 1541-1546.
(10) Gosselin, P.; Joulain, D.; Laurin, P.; Rouessac, F. Tetrahedron Lett. 1989,
30, 2775-2778.
(11) Earlier incubations (cf. ref 7) had been carried out using 0.5-1.0 µM
protein. GC-MS conditions: Hewlett-Packard Series 2 GC-MSD, 70
eV; 30 m × 0.25 mm HP5MS capillary column and temperature program
of 50-280 °C, 20 °C/min. Retention time, germacrene D (4), 10.24 min;
germacradienol (3), 11.20 min. See Supporting Information for experi-
mental details.
(12) Although hydrocarbon 5 has not previously been reported, the mass
spectrum is very similar, but not identical, to that of the isomeric 4,8a-
dimethyl-1,2,3,4,6,7,8,8a-octahydronaphthalene, m/z 164 (MassFinder 2.3,
retention index 1233, scan #479; D. H. Hochmuth, http://www.massfind-
er.com).
(13) Cane, D. E.; Oliver, J. S.; Harrison, P. H. M.; Abell, C.; Hubbard, B. R.;
Kane, C. T.; Lattman, R. J. Am. Chem. Soc. 1990, 112, 4513-4524.
(14) The deuterium labeling results summarized in Scheme 2 are consistent
with the reported conversion of [5,5-2H2]-1-deoxy-D-xylulose to [2H5]-
geosmin by cultures of Streptomyces sp JP95 (ref 3a), as well as the
incorporation of [4,4,6,6,6-2H5]mevalonate into [2H9]geosmin by both
Myxococcus xanthus and Stigmatella aurantiaca (ref 3b). The formation
of germacradienol (3) as well as the incorporation of three deuterium atoms
from water into geosmin (1d) rules out, however, the mechanism of
geosmin biosynthesis proposed in ref 3a.
JA062669X
9
J. AM. CHEM. SOC. VOL. 128, NO. 25, 2006 8129