Genome mining in streptomyces clavuligerus: Expression and biochemical characterization of two new cryptic sesquiterpene synthases

Article abstract of DOI:10.1016/j.chembiol.2010.11.008

Two presumptive terpene synthases of unknown biochemical function encoded by the sscg-02150 and sscg-03688 genes of Streptomyces clavuligerus ATCC 27074 were individually expressed in Escherichia coli as N-terminal-His ₆-tag proteins, using codon-optimized synthetic genes. Incubation of recombinant SSCG-02150 with farnesyl diphosphate (1, FPP) gave (-)-δ-cadinene (2) while recombinant SSCG-03688 converted FPP to (+)-T-muurolol (3). Individual incubations of (-)-δ-cadinene synthase with [1,1-²H₂]FPP (1a), (1S)-[1-²H]-FPP (1b), and (1R)-[1-²H]-FPP (1c) and NMR analysis of the resulting samples of deuterated (-)-δ-cadinene supported a cyclization mechanism involving the intermediacy of nerolidyl diphosphate (4) leading to a helminthogermacradienyl cation 5. Following a 1,3-hydride shift of the original H-1_si of FPP, cyclization and deprotonation will give (-)-δ-cadinene. Similar incubations with recombinant SSCG-03688 supported an analogous mechanism for the formation of (+)-T-muurolol (3), also involving a 1,3-hydride shift of the original H-1_si of FPP.

Full text of DOI:10.1016/j.chembiol.2010.11.008

Chemistry & Biology
Brief Communication
Genome Mining in Streptomyces clavuligerus:
Expression and Biochemical Characterization
of Two New Cryptic Sesquiterpene Synthases
Yunfeng Hu,1 Wayne K.W. Chou,1 Russell Hopson,1 and David E. Cane1,
*
1Department of Chemistry, Brown University, Box H, Providence, RI 02912-9108, USA
*Correspondence: david_cane@brown.edu
DOI 10.1016/j.chembiol.2010.11.008
SUMMARY
together harbor several dozen presumed terpene synthases,
the vast majority of which are of unknown biochemical function.
Two presumptive terpene synthases of unknown The two recently independently determined genomic sequences
of S. clavuligerus ATCC 27604, the producer of the widely used
biochemical function encoded by the sscg_02150
b-lactamase inhibitor clavulanic acid, indicate the presence of as
and sscg_03688 genes of Streptomyces clavuligerus
many as 20 presumptive terpene synthase genes, most of which
are located on a giant, 1.8 Mb linear plasmid, pSCL4 (Fischbach
et al., 2008; Medema et al., 2010). Of these 20 ORFs, the
ATCC 27074 were individually expressed in Escheri-
chia coli as N-terminal-His6-tag proteins, using
codon-optimized synthetic genes. Incubation of
recombinant SSCG_02150 with farnesyl diphosphate
(1, FPP) gave (-)-d-cadinene (2) while recombinant
SSCG_03688 converted FPP to (+)-T-muurolol (3).
Individual incubations of (-)-d-cadinene synthase
with [1,1-2H2]FPP (1a), (1S)-[1-2H]-FPP (1b), and
biochemical function of only sclav_p0159 can be predicted
with any certainty, based on its 62%–64% identity and 74%–
75% similarity (E value 0.0) over 724 amino acids (aa) to the
known S. coelicolor and S. avermitilis geosmin synthases (Jiang
et al., 2006; Cane et al., 2006).
Of the remaining S. clavuligerus terpene synthase genes, two
(1R)-[1-2H]-FPP (1c) and NMR analysis of the result- in particular attracted our attention. Each is part of an apparent
3-4 gene operon, suggesting the encoded terpene synthases
catalyze the first, committed step in the biosynthesis of oxidized
and methylated sesquiterpene metabolites of as yet unknown
ing samples of deuterated (-)-d-cadinene supported
a cyclization mechanism involving the intermediacy
of nerolidyl diphosphate (4) leading to a helmintho-
germacradienyl cation 5. Following a 1,3-hydride
shift of the original H-1si of FPP, cyclization and
deprotonation will give (-)-d-cadinene. Similar incu-
bations with recombinant SSCG_03688 supported
an analogous mechanism for the formation of (+)-T-
structure. Thus, sscg_02150 (sclav_p0328; UniProt ID B5GS26)
encodes a predicted 327 aa protein that is flanked downstream
by a unidirectionally transcribed putative cytochrome P450 and
upstream by a predicted methyl transferase and a diver-
gently transcribed oxidoreductase (Figure 1A). Similarly, the
sscg_03688 gene (sclav_p0068, UniProt ID B5GW45) encodes
muurolol (3), also involving a 1,3-hydride shift of the
original H-1si of FPP.
a predicted 418 aa terpene synthase protein flanked immediately
upstream and downstream by two unidirectionally transcribed
cytochrome P450s and a downstream methyl transferase
(Figure 1B). The cytochrome P450s and methyl transferases
are presumed to carry out an unspecified series of oxidations
and methylations of the individual sesquiterpene products
INTRODUCTION
Terpenoids constitute the largest and most varied group of initially produced by the parent terpene synthase from each
natural products. To date, more than 50,000 terpenoid operon. Both terpene synthases display the two highly
compounds with a wide range of chemical structures have conserved Mg2+-binding motifs characteristic of essentially all
been reported, most of which have been isolated from both known and presumed terpene synthases, the aspartate-rich
terrestial and marine plants, as well as numerous fungi and a rela- 84DDRID motif and the downstream ‘‘NSE/DTE’’ triad
tively small but increasing number of bacteria. The C15 sesquiter- 223NDLMTVDKE in SSCG_02150 and the corresponding
pene hydrocarbons and alcohols, which display the greatest 83DDEYCD and 225NDILSHHKE motifs in SSCG_03688. Similar
range of structural diversity among the terpenes, include more to all other terpene synthases, these conserved motifs are
than 300 identified carbon skeletons, each of which is derived each separated by 139 and 142 amino acids, respectively. On
by cyclization of the universal acyclic precursor farnesyl diphos- the other hand, neither protein shows any significant match in
phate (1, FPP). Extensive further metabolic transformations pairwise sequence alignments either to each other or to any
generate the enormous range of sesquiterpenoid metabolites.
terpene synthase of known biochemical function, and at best
Streptomyces are already noteworthy for the production of an low-level (<33%) identity to predicted proteins generically anno-
enormous variety of natural products, many of which play impor- tated only as ‘‘terpene synthase’’ or erroneously as ‘‘pentalenene
tant roles in human and animal medicine and in agriculture. More synthase.’’ We have now determined the biochemical function of
than 20 partial or complete Streptomyces genome sequences both enzymes, demonstrating that the SSCG_02150 protein
32 Chemistry & Biology 18, 32–37, January 28, 2011 ª2011 Elsevier Ltd All rights reserved

Chemistry & Biology
Sesquiterpene Biosynthesis in S. clavuligerus
Figure 1. S. clavuligerus Terpene Synthase Gene
Clusters
(A) SSCG_02150 biosynthetic gene cluster.
(B) SSCG_03688 biosynthetic gene cluster.
See also Figures S1 and S2.
catalyzes the cyclization of FPP to (-)-d-cadinene (2) while lactopyranoside (IPTG) and supplemented with lithium aceto-
the SSCG_03688 protein is shown to be a (+)-T-muurolol acetate and then incubated at 18ꢀC for 48 hr. Extraction of the
(3) synthase. We have also investigated the mechanism of both harvested cells with acetone followed by SiO2 column chromato-
enzyme-catalyzed FPP cyclizations and determined the stereo- graphic purification of the derived pentane-soluble products
chemistry of the 1,3-hydride transfers involved in each reaction. gave 10 mg/l of a sesquiterpene hydrocarbon, m/z 204, identical
Although both (+)-d-cadinene and (-)-T-muurolol are widely to that generated in vitro by SSCG_02150-catalyzed cyclization
occurring plant metabolites, neither (-)-2 nor (+)-3, nor any of FPP. In like manner, cotransformation of E. coli JM109(DE3)
derived sesquiterpenoid metabolites have previously been with pAC-Mev/Scidi/AacI and pET-03688 and pentane extrac-
reported as constituents of S. clavuligerus or any other tion of IPTG-induced, lithium acetoacetate-supplemented
Streptomycete.
cultures of E. coli JM109(DE3)/pAC-Mev/Scidi/AacI/pET-03688
gave, after chromatographic purification of the concentrated
pentane extract, 20 mg/l of a sesquiterpene alcohol, m/z 222,
identical to the product previously obtained by SSCG_03688-
catalyzed cyclization of FPP.
RESULTS AND DISCUSSION
Expression and Biochemical Characterization
of Recombinant SSCG_02150 and SSCG_03688
Synthetic genes for S. clavuligerus sscg_02150 and sscg_03688,
optimized for heterologous expression in Escherichia coli, were
ligated into the pET-28a(+) expression vector and the derived
plasmids, pET-02150 and pET-03688, were each transformed
into E. coli BL21(DE3). The resultant recombinant N-terminal
His6-tagged proteins, after purification to homogeneity by
Ni-NTA chromatography (see Figure S1 available online), were
each incubated with 60 mM FPP in the presence of 15 mM
Mg2+ for 16 hr at 30ꢀC. GC-MS analysis of the pentane extracts
of the SSCG_02150 incubation mixture indicated the formation
of a single (>95%) sesquiterpene hydrocarbon, C15H24, m/z
204, while recombinant SSCG_03688 yielded a sesquiterpene
alcohol, C15H26O, m/z 222 as the predominant (>95%) product
(Figure S2).
Identification of (-)-d-Cadinene (2) and (+)-T-Muurolol (3)
The structures of the sesquiterpene hydrocarbon produced by
SSCG_02150 and the sesquiterpene alcohol generated by
SSCG_03688 were deduced as (-)-d-cadinene (2) and (+)-T-
muurolol (3), respectively, by a combination of 1D and 2D NMR
1
(Figure S3), as well as GC-MS analysis. The H NMR spectrum
of 2 displayed one olefinic proton (d 5.43 [s]) while one C-H
(
13C 124.7 ppm) and three quaternary olefinic carbon signals
(134.2, 129.9, 124.5 ppm) were apparent in the 13C NMR spec-
trum, indicating the presence of two double bonds and therefore
two rings (Table S1). Of the four methyl signals, two belonged to
an isopropyl group (d 0.77 [d] and 0.94 [d]; 13C 15.7 and 21.8
ppm) and two showed typical resonances for allylic methyl
groups (d 1.63 [bs] and 1.65 [bs]; 13C 18.5 and 23.6 ppm). The
remaining connectivity was readily deduced from the HSQC
and HMBC spectra, and the deduced d-cadinene structure
was confirmed by comparison with the published 1H and 13C
NMR spectra of (+)-d-cadinene (Davis et al., 1996b). The abso-
lute configuration of the enzymatically generated (-)-d-cadinene
(2) was assigned by direct GC-MS comparison with authentic
(+)-d-cadinene ((+)-2), a major component of commercial cade
oil (Davis et al., 1996b). While the two samples showed identical
retention times (Retention Index [RI] 1520) and mass spectra
when analyzed by capillary GC-MS, they were clearly distin-
guished by chiral capillary GC-MS.
Steady-State Kinetic Parameters
Kinetic parameters were measured for each synthase by
carrying out 30 min incubations over a range of concentrations
of [1-3H]-FPP from 0.25 to 7.5 mM giving a kcat 1.14 0.03 3
10À3 s-1 and Km 1.4 0.1 mM for recombinant SSCG_02150,
while recombinant SSCG_03688 had a kcat 1.63
10À3 s-1 and Km 2.7 0.3 mM.
0.09 3
Production in E. coli of Sesquiterpenes Generated
by S. clavuligerus SSCG_02150 and SSCG_03688
In order to obtain sufficient quantities of the sesquiterpene prod-
1H and 13C NMR analysis of the purified sesquiterpene alcohol
ucts of each of the two S. clavuligerus terpene synthases, we produced by E. coli JM109(DE3)/pAC-Mev/Scidi/AacI/pET-
utilized a metabolically engineered E. coli host designed for the 03688 showed the presence of a single trisubsituted double
efficient in vivo synthesis of terpenoid metabolites (Harada bond as evidenced by the proton signal at d 5.54 (d) correlated
et al., 2009). The two plasmids, pAC-Mev/Scidi/AacI and pET- with the 13C resonance at 124.8 ppm and a second olefinic
02150 were used to cotransform E. coli JM109(DE3). Cultures carbon at 133.5 ppm attached to an allylic methyl (d 1.63 [bs],
of the resultant transformant, E. coli JM109(DE3)/pAC-Mev/ 13C 23.6 ppm), indicating the presence of a bicyclic structure.
Scidi/AacI/pET-02150, were induced with isopropyl-b-D-thioga- A pair of geminal methyl groups indicated the presence of an
Chemistry & Biology 18, 32–37, January 28, 2011 ª2011 Elsevier Ltd All rights reserved 33

Chemistry & Biology
Sesquiterpene Biosynthesis in S. clavuligerus
Figure 2. Mechanism and Stereochemistry of the Enzymatic Cyclization of Chirally Deuterated FPP
SSCG_02150-catalyzed cyclization of FPP to (-)-d-cadinene (2) and SSCG_03688-catalyzed cyclization of FPP to (+)-T-muurolol (3). See also Figures S3, S6, and
S7 and Table S1.
isopropyl group (d 0.82 [d] and 0.87 [d]; 13C 21.6 and 15.4 ppm), Mechanism and Stereochemistry of the Cyclization
while a fourth methyl group (d 1.18 [s], 13C 29.3 ppm) was of FPP to (-)-d-Cadinene (2) and (+)-T-Muurolol (3)
attached to the quaternary carbinol carbon (13C 72.4 ppm). To elucidate the mechanism and stereochemistry of the
The remaining connectivity and relative stereochemistry was enzymatic formation of formation of both (-)-d-cadinene (2) and
deduced from the 2D NMR analysis and the T-muurolol structure (+)-T-muurolol (3), we carried out preparative-scale incubations
was confirmed by the precise match to the reported NMR of each protein with individual samples of [1,1-2H2]-FPP (1a),
spectra of (-)-T-muurolol in both CDCl3 and C6D6 (Ding et al., (1S)-[1-2H]-FPP (1b), and (1R)-[1-2H]-FPP (1c) (Figure 2). The
2009) and (+)-T-muurolol in CDCl3 (Nagashima et al., 1994), resultant deuterated samples of (-)-d-cadinene (2a-2c) (Figure 3)
which all differed in several significant details from those and (+)-T-muurolol (3a-3c) (Figure 4), were analyzed by GC-MS
reported for a-cadinol, amorphenol, and related diastereomers and NMR (Figures S4 and S5). (-)-d-Cadinene (2a) derived from
of 3. The (+)-T-muurolol had an RI value of 1639 (cf. RI 1633 [1,1-2H2]-FPP (1a) retained both deuterium atoms, as evidenced
for (-)-3, MassFinder 4.0,). The absolute configuration of enzy- by the parent peak at m/z 206 [M+] and the base peak at m/z
matically-generated 3 was assigned from the [a]D +85.79ꢀ, cor- 162 [M–44] due to loss of the deuterated isopropyl side chain
responding to that reported for (+)-T-muurolol (Nagashima (Figure S4). The mass spectrum of [11-2H]-2b derived from
et al., 1994).
(1S)-[1-2H]-FPP (1b) exhibited a parent peak of m/z 205 and
34 Chemistry & Biology 18, 32–37, January 28, 2011 ª2011 Elsevier Ltd All rights reserved

Chemistry & Biology
Sesquiterpene Biosynthesis in S. clavuligerus
Figure 3. 1H NMR Analysis of (-)-d-Cadinene (2)
Generated by SSCG_02150-Catalyzed Cyclization
of Chirally Deuterated FPP
(A) [11-2H]-(-)-d-cadinene (2b) from incubation with (1S)-
[1-2H]-FPP (1b).
(B) [5-2H]-(-)-d-cadinene (2c) from incubation with (1R)-
[1-2H]-FPP (1c).
See Figure S4.
undetermined absolute configuration have
been detected in the volatile organic products
of the North Sea Streptomyces strain GWS-
BW-H5 in which epicubenol (8) of unspecified
configuration is the major metabolite (Dickschat
et al., 2005). Recently (-)-T-muurolol ([-]-3) has
been isolated from the marine Streptomyces
sp. M491, along with several oxidized T-muuro-
lol derivatives, including 3-oxo-T-muurolol and
3-oxo-15-hydroxy-T-muurolol (Ding et al.,
2009). The only previously documented occur-
rence of (+)-T-muurolol (3) is its isolation from
a base peak m/z 161 [M–44] indicating the presence of the the Belgian liverwort Scapania undulata along with the closely
deuterium in the isopropyl side chain, while [5-2H]-2c derived related (+)-epicubenol ([+]-8) (Nagashima and Asakawa, 2001;
from (1R)-[1-2H]-FPP (1c) had the complementary pattern with Nagashima et al., 1994). The latter sesquiterpene alcohol has
a parent peak of m/z 205 and a base peak m/z 162 [M–43]. previously been isolated from both Streptomyces sp. LL-B7
The 1H NMR spectrum of [11-2H]-2b lacked the multiplet at (Gerber, 1971) and the liverwort Heteroscyphus planus (Nabeta
d 2.03 corresponding to the H-11 isopropyl methine proton, while et al., 1994, 1995). The discovery that S. clavuligerus
the two neighboring isopropyl methyl protons each now ap- SSCG_02150 is
a (-)-d-cadinene synthase and that the
peared as singlets (Figure 3A). In the complementary experi- SSCG_03688 protein is a (+)-T-muurolol synthase is the first
ment, [5-2H]-2c lacked the bridgehead H-5 allylic proton reso- documented evidence for the biosynthesis of the specific enan-
nance at d 2.50 (Figure 3B). Consistent with these results, the tiomeric forms of each of these sesquiterpenes in Streptomyces.
HSQC spectrum of [5,11-2H2]-2a lacked both the H-5/C-5 and Since both of these S. clavuligerus sesquiterpene synthases are
H-11/C-11 cross-peaks that were present in the spectrum of located within apparent gene clusters that also harbor genes for
unlabeled 2 (data not shown).
1-2 additional cytochrome P450s and a methyl transferase, it is
The SSCG_03688-catalyzed formation of (+)-T-muurolol (3) unlikely that either the parent sesquiterpene hydrocarbon 2 or
showed the same stereochemistry for the 1,3-hydride shift as the sesquiterpene alcohol 3 accumulates in S. clavuligerus.
that determined for (-)-d-cadinene synthase (Figure 2). Thus
[5,11-2H2]-(+)-T-muurolol (3a) derived from [1,1-2H2]-FPP (1a)
Mechanism and Stereochemistry of the Cyclization
retained both deuterium atoms, as evidenced by the parent
peak at m/z 224 [M+], the [M – H2O] peak at m/z 206, and
Reactions
Theenzyme-catalyzed formation of both(-)-d-cadinene and(+)-T-
a peak at m/z 162 [M – 64] due to loss of the deuterated isopropyl
muurolol mediated by the S. clavuligerus synthases encoded by
side chain plus water (Figure S5). Similarly, the mass spectrum of
[11-2H]-3b ([M+] m/z 223) derived from (1S)-[1-2H]-FPP (1b)
displayed a peak at m/z 161 resulting from loss of the deuterated
thesscg_02150andsscg_03688genesbothinvolvea1,3-hydride
shift of the original H-1si of farnesyl diphosphate, presumably
involving a common (Z,E)-helminthogermacradienyl cation inter-
isopropyl side chain plus water, while [5-2H]-3c ([M+] m/z 223)
mediate 5 (Figure 2). Since the bridgehead stereochemistry at
derived from (1R)-[1-2H]-FPP (1c) retained the deuterium in the
C-10 of the derived bicyclic cation 6 is uncertain in the case of
(-)-d-cadinene biosynthesis, while (+)-T-muurolol must be formed
m/z 162 fragment. The 1H NMR spectrum of [5,11-2H2]-3a lacked
the signals for both H-5 (d 2.25) and H-11 (d 1.96), while the
from the corresponding cis-fused muurolyl cation 7, it is unclear at
which point the two enzymatic cyclization pathways diverge. We
corresponding spectrum of [11-2H]-3b lacked only the signal
for H-11, with the isopropyl methyl signals appearing as singlets
have previously established the stereochemistry of the analogous
1,3-hydride shift in the cyclization of FPP to (+)-epicubenol (8)
catalyzed by the Streptomyces sp. LL-B7 (+)-epicubenol syn-
at d 0.80 and 0.86 (Figure 4).
Occurrence of d-Cadinene and T-Muurolol
thase (Cane and Tandon, 1995) (Figure S6). The formation of
Although both (+)-d-cadinene and (-)-T-muurolol, the enantio- (+)-8 has been shown to involve the intermediacy of (3R)-nerolidyl
morphs of the sesquiterpene cyclization products produced by diphosphate (4, NPP) that is generated by an initial syn allylic rear-
the two S. clavuligerus terpene synthases, are common plant rangement of FPP. The cisoid conformer of (3R)-NPP then
metabolites, the isolation of either enantiomer of 2 and 3 from undergoes cyclization to the helminthogermacradienyl cation 5.
bacterial cultures is very rare. Trace amounts of d-cadinene of A similar 1,3-hydride shift has also been implicated in the
Chemistry & Biology 18, 32–37, January 28, 2011 ª2011 Elsevier Ltd All rights reserved 35

Chemistry & Biology
Sesquiterpene Biosynthesis in S. clavuligerus
Figure 4. 1H NMR Analysis of (+)-T-Muurolol (3)
Generated by SSCG_03688-Catalyzed Cyclization
of Chirally Deuterated FPP
(A) [5,11-2H2]-(+)-T-muurolol (3a) from incubation with
[1,1-2H2]-FPP (1a).
(B) [11-2H]-(+)-T-muurolol (3b) from incubation with (1S)-
[1-2H]-FPP (1b).
See Figure S5.
very different Mg2+-binding motifs. The S. clavu-
ligerus (-)-d-cadinene synthase also has insignif-
icant similarity either at the amino acid level or in
the identity of presumptive active site residues to
the well-studied G. arboreum (+)-d-cadinene
synthase (UniProt ID Q39761) (Benedict et al.,
2001; Davis et al., 1996a). Although the cotton
enzyme has
a typical aspartate-rich motif
308DDTYD, it lacks the canonical NSE/NTE
Mg2+-binding triad, displaying instead a second
aspartate-rich motif, 451DDVAE. The S. clavuli-
gerus (-)-d-cadinene synthase is also distinct in
formation of the enantiomeric (+)-d-cadinene by Gossypium ar- both amino acid sequence and Mg2+-binding motifs from the
boreum (cotton) cadinene synthase (Benedict et al., 2001). Inter- recently described d-cadinene synthase of the basidiomycete
estingly, intermolecular competition experiments using the cotton Coprinus cinereus (Cop4, UniProt ID A8NU13) which generates
cyclase demonstrated that (3R)-NPP (4) is an intermediate in the d-cadinene of unspecified absolute configuration (Agger et al.,
formation of (+)-d-cadinene, the enantiomer of the product gener- 2009). S. clavuligerus SSCG_03688, the first reported T-muurolol
ated by the S. clavuligerus SSCG_02150 enzyme. Assuming that synthase, also has insignificant amino acid sequence similarity to
this (3R)-NPP would be generated from (E,E)-FPP by the usual syn the g-muurolene synthase of C. cinereus (Cop3, UniProt ID
allylic rearrangement, it can be inferredthat the 1,3-hydrideshift in A8NE23). In spite of the substantial differences in both overall
the biosynthesis of (+)-d-cadinene also involves the H-1si (Hb) of amino acid sequence and specific Mg2+-binding motifs, both
FPP. In fact, as Arigoni has elegantly demonstrated in a pene- SSCG_02150 and SSCG_03688 are predicted by protein
trating analysis of FPP cyclization reactions involving 1,3-hydride homology modeling using the Modweb server (http://modbase.
shifts that generate products with cis double bonds and isopropyl compbio.ucsf.edu/modbase-cgi/index.cgi) (Pieper et al., 2009)
side chains, there is a strict correlation between the identity of the to have the common all-a-helical class I terpenoid synthase fold
˚
enantiotopic H-1 proton of FPP that undergoes the 1,3-hydride that has been established by the 2.4 A crystal structure of the
shift and the relative cis- or trans-configuration of the bridgehead cotton (+)-d-cadinene synthase (Gennadios et al., 2009) and the
proton and the isopropyl side chain in the ultimately formed prototype bacterial sesquiterpene synthase pentalenene syn-
sesquiterpene product (Arigoni, 1975). On the other hand, the thase (Lesburg et al., 1997), as well as all other terpene synthases
absolute configuration of the enzyme-bound NPP intermediate analyzed to date (Christianson, 2006).
cannot be deduced solely from the absolute or relative configura-
Since terpene synthase sequences cluster in multiple
tion of the eventually formed cadinene, muurolene, or amorphene sequence alignments predominantly according to phylogenetic
product. We have provisionally invoked the intermediacy of (3R)- origin rather than biochemical reaction type (Bohlmann et al.,
NPP (4) in the SSCG_02150-catalyzed formation of (-)-d-cadinene 1998), bioinformatic analysis of deduced amino acid sequence
(2) and SSCG_03688-catalyzed formation of (+)-T-muurolol alone is incapable of assigning the specific biochemical function
(3) (Figure 2), based on analogy to the demonstrated role of of any newly recognized terpene synthase gene. Direct biochem-
(3R)-NPP (4) in the closely related reaction catalyzed by ical characterization of terpene synthases is therefore essential
(+)-epicubenol (8) synthase from Streptomyces LL-B7 (Cane for assignment of the actual terpene cyclization product and
and Tandon, 1995) (Figure S6). An alternative cyclization pathway even the native biological substrate (FPP or geranyl diphosphate;
for (-)-d-cadinene synthase involving (3S)-NPP that would give the Bohlmann and Gershenzon, 2009). Of the 20 or so sequences in
same observed labeling results is illustrated in Figure S7. These the S. clavuligerus 27604 genome that are currently annotated as
two theoretical possibilities will have to be distinguished ‘‘terpene synthase,’’ only two of these ORFs are now of experi-
experimentally.
In spite of the fact that the S. clavuligerus (-)-d-cadinene (SSCG_02150) and (+)-T-muurolol synthase (SSCG_03688).
synthase (SSCG_02150) and (+)-T-muurulol synthase
mentally defined biochemical function: (-)-d-cadinene synthase
(SSCG_03688) are derived from the same organism and catalyze SIGNIFICANCE
cyclization reactions that are closely related both mechanistically
and stereochemically, the two proteins are remarkably different, Genome mining has provided a powerful and versatile new
showing <25% identity in pairwise alignment and possessing paradigm for the discovery of new biochemical
36 Chemistry & Biology 18, 32–37, January 28, 2011 ª2011 Elsevier Ltd All rights reserved

Chemistry & Biology
Sesquiterpene Biosynthesis in S. clavuligerus
Cane, D.E., He, X., Kobayashi, S., Omura, S., and Ikeda, H. (2006). Geosmin
biosynthesis in Streptomyces avermitilis. Molecular cloning, expression, and
mechanistic study of the germacradienol/geosmin synthase. J. Antibiot.
(Tokyo) 59, 471–479.
transformations and natural product pathways (Corre and
Challis, 2009; Walsh and Fischbach, 2010). Although generic
terpene synthases from both bacterial and eukaryotic
genomes can be readily recognized by the presence of the
characteristic pair of conserved Mg2+-binding motifs, sepa-
rated in all cases by 140 5 aa, a major challenge in the
annotation of presumptive terpene synthases is the lack of
an obvious correlation between detailed protein sequence
and any specific terpene cyclization pathway. Direct
biochemical characterization is therefore indispensable for
the identification of both the actual acyclic terpene
substrate and the derived cyclization product. The
sscg_02150 and sscg_03688 gene products of S. clavulige-
rus have now been established to be (-)-d-cadinene syn-
thase and (+)-T-muurolol synthase, respectively. This repre-
sents the first isolation and characterization of either
sesquiterpene synthase from any biological source and the
first documented evidence for the occurrence of the specific
enantiomers of either sesquiterpene in a bacterial host. The
role of the two sesquiterpenes in S. clavuligerus metabolism
and the structure of the ultimately derived terpenoid metab-
olites produced by the corresponding biosynthetic gene
clusters remain unknown, as does the function of the
impressive number of remaining uncharacterized terpene
synthases that make up the S. clavuligerus terpenome.
Christianson, D.W. (2006). Structural biology and chemistry of the terpenoid
cyclases. Chem. Rev. 106, 3412–3442.
Corre, C., and Challis, G.L. (2009). New natural product biosynthetic chemistry
discovered by genome mining. Nat. Prod. Rep. 26, 977–986.
Davis, E.M., Tsuji, J., Davis, G.D., Pierce, M.L., and Essenberg, M. (1996a).
Purification of (+)-d-cadinene synthase,
a sesquiterpene cyclase from
bacteria-inoculated cotton foliar tissue. Phytochemistry 41, 1047–1055.
Davis, G.D., Essenberg, M., Berlin, K.D., Faure, R., and Gaydou, E.M. (1996b).
Complete 1H and 13C NMR spectral assignment of d-cadinene, a bicyclic
sesquiterpene hydrocarbon. Magn. Reson. Chem. 34, 156–161.
Dickschat, J.S., Martens, T., Brinkhoff, T., Simon, M., and Schulz, S. (2005).
Volatiles released by a Streptomyces species isolated from the North Sea.
Chem. Biodivers. 2, 837–865.
Ding, L., Pfoh, R., Ruhl, S., Qin, S., and Laatsch, H. (2009). T-muurolol sesqui-
terpenes from the marine Streptomyces sp. M491 and revision of the configu-
ration of previously reported amorphanes. J. Nat. Prod. 72, 99–101.
Fischbach, M., Ward, D., Young, S., Jaffe, D., Gnerre, S., Berlin, A., Heiman,
D., Hepburn, T., Sykes, S., Alvarado, L., et al. (2008). Annotation of
Streptomyces clavuligerus ATCC 27064; whole genome shotgun sequencing
project, GenBank NCBI Nucleotide Accession Nr. ABJH00000000.
Gennadios, H.A., Gonzalez, V., Di Costanzo, L., Li, A., Yu, F., Miller, D.J.,
Allemann, R.K., and Christianson, D.W. (2009). Crystal structure of (+)-d-cadi-
nene synthase from Gossypium arboreum and evolutionary divergence of
metal binding motifs for catalysis. Biochemistry 48, 6175–6183.
SUPPLEMENTAL INFORMATION
Gerber, N.N. (1971). Sesquiterpenoids from actinomycetes: Cadin-4-ene-1-ol.
Supplemental Information includes Supplemental Experimental Procedures,
seven figures, and one table and can be found with this article online at
doi:10.1016/j.chembiol.2010.11.008.
Phytochemistry 10, 185–189.
Harada, H., Yu, F., Okamoto, S., Kuzuyama, T., Utsumi, R., and Misawa, N.
(2009). Efficient synthesis of functional isoprenoids from acetoacetate through
metabolic pathway-engineered Escherichia coli. Appl. Microbiol. Biotechnol.
81, 915–925.
ACKNOWLEDGMENTS
We thank Dr. Norihiko Misawa of Kirin Holdings Co. Ltd. Japan for providing
plasmid pAC-Mev/Scidi/AacI. This work was supported by National Institutes
of Health Grant GM30301 to D.E.C. We thank Dr. Tun-Li Shen for assistance
with mass spectrometry.
Jiang, J., He, X., and Cane, D.E. (2006). Geosmin biosynthesis. Streptomyces
coelicolor germacradienol/germacrene D synthase converts farnesyl diphos-
phate to geosmin. J. Am. Chem. Soc. 128, 8128–8129.
Lesburg, C.A., Zhai, G., Cane, D.E., and Christianson, D.W. (1997). Crystal
structure of pentalenene synthase: mechanistic insights on terpenoid cycliza-
tion reactions in biology. Science 277, 1820–1824.
Received: September 14, 2010
Revised: October 17, 2010
Accepted: November 9, 2010
Published: January 27, 2011
Medema, M.H., Trefzer, A., Kovalchuk, A., van den Berg, M., Muller, U., Heijne,
W., Wu, L., Alam, M.T., Ronning, C.M., Nierman, W.C., et al. (2010). The
sequence of a 1.8-Mb bacterial linear plasmid reveals a rich evolutionary reser-
voir of secondary metabolic pathways. Genome Biol. Evol. 2, 212–224.
REFERENCES
Nabeta, K., Mototani, Y., Tazaki, H., and Okuyama, H. (1994). Biosynthesis of
sesquiterpenes of cadinane type in cultured cells of Heteroscyphus planus.
Phytochemistry 35, 915–920.
Agger, S., Lopez-Gallego, F., and Schmidt-Dannert, C. (2009). Diversity of
sesquiterpene synthases in the basidiomycete Coprinus cinereus. Mol.
Microbiol. 72, 1181–1195.
Nabeta, K., Kigure, K., Fujita, M., Nagoya, T., Ishikawa, T., Okuyama, H., and
Takasawa, T. (1995). Biosynthesis of (+)-cubenene and (+)-epicubenol by cell-
free extracts of cultured cells of Heteroscyphus planus and cyclization of [2H]
farnesyl diphosphates. J. Chem. Soc., Perkin Trans. 1, 1935–1939.
Arigoni, D. (1975). Stereochemical aspects of sesquiterpene biosynthesis.
Pure Appl. Chem. 41, 219–245.
Benedict, C.R., Lu, J.L., Pettigrew, D.W., Liu, J., Stipanovic, R.D., and
Williams, H.J. (2001). The cyclization of farnesyl diphosphate and nerolidyl
diphosphate by a purified recombinant d-cadinene synthase. Plant Physiol.
125, 1754–1765.
Nagashima, F., and Asakawa, Y. (2001). Sesqui- and diterpenoids from two
Japanese and three European liverworts. Phytochemistry 56, 347–352.
Nagashima, F., Suda, K., and Asakawa, Y. (1994). Cadinane-type sesquiterpe-
Bohlmann, J., and Gershenzon, J. (2009). Old substrates for new enzymes of
noids from the liverwort Scapania undulata. Phytochemistry 37, 1323–1325.
terpenoid biosynthesis. Proc. Natl. Acad. Sci. USA 106, 10402–10403.
Pieper, U., Eswar, N., Webb, B.M., Eramian, D., Kelly, L., Barkan, D.T., Carter,
H., Mankoo, P., Karchin, R., Marti-Renom, M.A., et al. (2009). MODBASE,
a database of annotated comparative protein structure models and associated
resources. Nucleic Acids Res. 37, D347–D354.
Bohlmann, J., Meyer-Gauen, G., and Croteau, R. (1998). Plant terpenoid syn-
thases: molecular biology and phylogenetic analysis. Proc. Natl. Acad. Sci.
USA 95, 4126–4133.
Cane, D.E., and Tandon, M. (1995). Epicubenol synthase and the stereochem-
istry of the enzymic cyclization of farnesyl and nerolidyl diphosphate. J. Am.
Chem. Soc. 117, 5602–5603.
Walsh, C.T., and Fischbach, M.A. (2010). Natural products version 2.0: con-
necting genes to molecules. J. Am. Chem. Soc. 132, 2469–2493.
Chemistry & Biology 18, 32–37, January 28, 2011 ª2011 Elsevier Ltd All rights reserved 37