Non-heme iron oxygenases generate natural structural diversity in carbapenem antibiotics

Article abstract of DOI:10.1021/ja907320n

(Figure Presented) Carbapenems are a clinically important antibiotic family. More than 50 naturally occurring carbapenam/ems are known and are distinguished primarily by their C-2/C-6 side chains where many are only differentiated by the oxidation states of these substituents. With a limited palette of variations the carbapenem family comprises a natural combinatorial library, and C-2/C-6 oxidation is associated with increased efficacy. We demonstrate that ThnG and ThnQ encoded by the thienamycin gene cluster in Streptomyces cattleya oxidize the C-2 and C-6 moieties of carbapenems, respectively. ThnQ stereospecifically hydroxylates PS-5 (5) giving N-acetyl thienamycin (2). ThnG catalyzes sequential desaturation and sulfoxidation of PS-5 (5), giving PS-7 (7) and its sulfoxide (9). The enzymes are relatively substrate selective but are proposed to give rise to the oxidative diversity of carbapenems produced by S. cattleya, and orthologues likely function similarly in allied streptomyces. Elucidating the roles of ThnG and ThnQ will focus further investigations of carbapenem antibiotic biosynthesis.

Full text of DOI:10.1021/ja907320n

Published on Web 12/17/2009
Non-Heme Iron Oxygenases Generate Natural Structural Diversity in
Carbapenem Antibiotics
Micah J. Bodner, Ryan M. Phelan, Michael F. Freeman, Rongfeng Li, and Craig A. Townsend*
Department of Chemistry, Johns Hopkins UniVersity, 3400 North Charles Street, Baltimore, Maryland 21218
Received August 29, 2009; E-mail: ctownsend@jhu.edu
Carbapenem antibiotics are of clinical importance because of
their high potency, broad spectrum of antimicrobial activity, and
resistance to most ꢀ-lactamases.1 Thienamycin (1) (Figure 1), the
most potent natural member of this family, co-occurs in Strepto-
myces cattleya with four carbapenems that are distinguished by their
C-2/C-6 substituents.2 There are more than 50 known carbapenam/
em metabolites, many of which are differentiated only by the
oxidation state of their C-2/C-6 substituents. The C-6 ethyl side
chain of 1 is derived by C1-donations from methionine3,4 and can
be methyl, ethyl, or isopropyl, which can be saturated, unsaturated,
hydroxylated, or sulfated. Recent work has established that
coenzyme A is successively truncated by three enzymes encoded
by the thienamycin gene cluster to give the C-2 cysteamine moiety.5
This side chain can be pantetheine, but is generally cysteamine,
which can be N-acetylated or N-propionylated, desaturated and
further oxidized to the sulfoxide, or cleaved and oxidized in a
stepwise fashion to the sulfonic acid. These carbapenem metabolites
comprise a natural combinatorial library whose structural modifica-
tions temper the high intrinsic hydrolytic instability of the carbap-
enem nucleus, as well as affect the antimicrobial spectrum and
ꢀ-lactamase resistance of each family member.1 Some of the higher
oxidation state carbapenems have either enhanced antibiotic activity
or increased ꢀ-lactamase resistance, and so, in the broad context
of carbapenem biosynthesis, the origin of this oxidative diversity
is of particular interest.
have been postulated to catalyze steps in thienamycin biosynthesis
analogous to the coupled C-5 epimerization and C-2/C-3 desatu-
ration of (2S,5S)-carbapenam (11) to (5R)-carbapenem-3-carboxy-
late (13) catalyzed by CarC.11 To discern their roles in thienamycin
biosynthesis, ThnG and ThnQ were analyzed for carbapenem-
oxidizing activity, as well as for the ability to catalyze C-5
epimerization and coupled or uncoupled C-2/C-3 desaturation of
carbapenams/ems.
The envisioned experiments required carbapenam/ems varying
in stereochemical configuration at C-6 as well as C-2/C-6 oxidation
state/substitution pattern to serve as substrates and reference
standards. Two methods were employed to establish the C-6
substituent and C-5/C-6 configuration by synthesizing precursor
azetidinones. The first method provided the trans (3S,4R)-config-
uration by alkylating the enolate of an azetidinone derived from
L-aspartic acid.12 The second method employed a catalytic asym-
metric azetidinone-forming reaction that produced either enantiomer
of the cis 3,4-disubstituted azetidinones with independent control
of the carbapenem C-8 stereocenter.13 These compounds could be
used as precursors of cis or trans carbapenems. Azetidinones were
converted to carbapenems by the Merck method, which allowed
various C-2 groups to be introduced.14,15 The intermediate 2-oxo-
carbapenams in this route can be reduced and directed to the
preparation of carbapenams bearing thioether substituents at C-2
(Figure 1c).5 Carbapenam thioethers 14-17, (2S,5S)-carbapenam
(11), (2S,5R)-carbapenam (12),16 and 5-epi-PS-5 (6) were synthe-
sized to test ThnG and ThnQ for coupled or uncoupled carbapenam
ring epimerization and desaturation. PS-5 (5) was synthesized to
test for side chain oxidation activity, because the acetylated
cysteaminyl side chain is more stable than the unacetylated
deshydroxy thienamycin (10). PS-7 (7), PS-7 sulfoxide (9), PS-5
sulfoxide (8), N-acetyl thienamycin (2), and the diastereomeric
mixture (8S,R)-N-acetyl thienamycin (4) were synthesized as
additional substrates and reference standards.
Thienamycin biosynthetic cluster genes thnG and thnQ were
cloned from genomic DNA and inserted into pET29b each bearing
a C-terminal His6-tag. The recombinant proteins were overproduced
in E. coli Rosetta2(DE3) and purified by Ni-NTA affinity chro-
matography. In Vitro reactions in MOPS, pH 7.0, containing
Fe(NH4)2(SO4)2, R-KG, ascorbate, the subject carbapenam/em, and
either ThnG or ThnQ were incubated and analyzed by HPLC for
the formation of new product(s).17,18 Clear outcomes were observed
for both ThnG- and ThnQ-catalyzed reactions with PS-5 (5) (Figure
2 and Supporting Information). The products were immediately
Figure 1. (a) Representative carbapenems. (b) The CarC-catalyzed reaction.
(c) Carbapenam thioethers synthesized in this study.
The variability of the carbapenem side chain oxidation state as
well as the discovery of a mutant strain6 of Streptomyces cattleya
that produced deshydroxy thienamycin (10) instead of thienamycin
(1) led us to believe that the thienamycin gene cluster7 encoded
one or more enzyme(s) capable of oxidizing the C-2/C-6 moieties
of carbapenems. Protein sequence analysis of ThnG and ThnQ
indicated that each enzyme contained the Hx(D/E)xnH motif
characteristic of nonheme Fe(II)/R-ketoglutarate (R-KG)-dependent
dioxygenases, and these seemed promising candidates.8 However,
ThnG and ThnQ are in the same family as CarC encoded by the
(5R)-carbapenem-3-carboxylate (13) gene cluster in Pectobacterium
carotoVorum.9,10 Despite low homology to CarC, ThnG and ThnQ
identifiable as carbapenems by their unique chromophores (λmax
)
290-320 nm). ThnQ produced a single new product more polar
than 5, while ThnG produced two products, one with a shorter
retention time and one with a longer retention time than that of 5.
ESI mass spectrometric analysis of the new product (m/z )
313.05) in the ThnQ-catalyzed reaction established that a single
oxygen had been incorporated. Its identity was determined by
9
12 J. AM. CHEM. SOC. 2010, 132, 12–13
10.1021/ja907320n  2010 American Chemical Society

C O M M U N I C A T I O N S
respectively, of carbapenem substrates. No evidence was found that
they catalyze coupled or uncoupled C-5 epimerization and/or C-2/
C-3 desaturation in the carbapenam/ems tested. On these bases it
appears that the latter two reactions rely on other proteins encoded
by the thienamycin gene cluster that are distinct from the non-
heme iron oxygenases like CarC, key to the biosynthesis of (5R)-
carbpenem-3-carboxylate (13).10,11,21 Oxidative modifications of
the C-2 and C-6 side chains of carbapenems are major determinants
of their antimicrobial spectrum and ꢀ-lactamase resistance. These
activities strongly suggest that the known carbapenems produced
by S. cattleya arise from ThnG and ThnQ oxidation of a common
biosynthetic precursor and that much of the structural diversity
exemplified by this class of antibiotics likely derives from ortho-
logues present in other carbapenem producers. Knowledge of these
oxidative relationships will more sharply refine further biosynthetic
investigations of this antibiotic family.
Figure 2. HPLC analysis of ThnG and ThnQ reactions with PS-5 (5). (a)
PS-7 sulfoxide diasteriomers standard (9). (b) PS-7 standard (7). (c) ThnG-
catalyzed reaction with 5. (d) N-Acetyl thienamycin standard (2). (e) ThnQ-
catalyzed reaction with 5.
Acknowledgment. We thank K. A. Moshos for carbapenams
and Nitrocefin and the NIH (AI014937) for financial support.
coinjection with N-acetyl thienamycin (2) and (8S,R)-N-acetyl
thienamycin (4) (Figures 2d,e, S3). This HPLC comparison
demonstrated that ThnQ stereospecifically hydroxylated PS-5 (5)
to produce 2. The chromatographically distinct (8S)-N-acetyl
thienamycin diastereomer was not detected (Figure S4). ESI-MS
analysis of the products of the ThnG-catalyzed reaction with 5
indicated that the late eluting product (m/z ) 295.05) had an
additional degree of unsaturation relative to 5, and the early eluting
product (m/z ) 311.15) was both desaturated and oxidized. The
carbapenems were identified as PS-7 (7) and PS-7 sulfoxide 9 by
HPLC comparison to synthetic standards (Figures 2a-c, S5). ThnG
was also able convert 7 to its sulfoxide but unable to catalyze
desaturation when given PS-5 sulfoxide 8, indicating that desatu-
ration precedes sulfoxidation (Figure S6).
The oxidative relationships among carbapenems were further
demonstrated by conversion of PS-7 (7) and N-acetyl thienamycin
(2) into the S. cattleya metabolite N-acetyl dehydrothienamycin (3).
Upon reaction with 2, ThnG produced a less polar product and
ESI-MS indicated it contained an additional degree of unsaturation.
ThnQ produced a more polar product on reaction with 7. These
new products coeluted under HPLC, had identical masses by ESI-
MS, and were assigned the same structure, N-acetyl dehydrothie-
namycin (3). Notably no sulfoxide product was observed, consistent
with the metabolite profile in S. cattleya and suggesting that ThnQ
reaction precedes that of ThnG.
Supporting Information Available: Experimental procedures,
HPLC comparisons, and compound characterizations. This material is
available free of charge via the Internet at http://pubs.acs.org.
References
(1) Bonfiglio, G.; Russo, G.; Nicoletti, G. Expert Opin. InVest. Drugs 2002,
11 (4), 529–544.
(2) Wilson, K. E.; Kempf, A. J.; Liesch, J. M.; Arison, B. H. J. Antibiot. 1983,
36 (9), 1109–1117.
(3) Houck, D. R.; Kobayashi, K.; Williamson, J. M.; Floss, H. G. J. Am. Chem.
Soc. 1986, 108 (17), 5365–5366.
(4) Williamson, J. M.; Inamine, E.; Wilson, K. E.; Douglas, A. W.; Liesch,
J. M.; Albersschonberg, G. J. Biol. Chem. 1985, 260 (8), 4637–4647.
(5) Freeman, M. F.; Moshos, K. A.; Bodner, M. J.; Li, R. F.; Townsend, C. A.
Proc. Natl. Acad. Sci. U.S.A. 2008, 105 (32), 11128–11133.
(6) Rosi, D.; Drozd, M. L.; Kuhrt, M. F.; Terminiello, L.; Came, P. E.; Daum,
S. J. J. Antibiot. 1981, 34 (3), 341–343.
(7) Nunez, L. E.; Mendez, C.; Brana, A. F.; Blanco, G.; Salas, J. A. Chem.
Biol. 2003, 10 (4), 301–311.
(8) Hausinger, R. P. Crit. ReV. Biochem. Mol. 2004, 39 (1), 21–68.
(9) McGowan, S. J.; Sebaihia, M.; Porter, L. E.; Stewart, G. S. A. B.; Williams,
P.; Bycroft, B. W.; Salmond, G. P. C. Mol. Microbiol. 1996, 22 (3), 415–
426.
(10) Li, R. F.; Stapon, A.; Blanchfield, J. T.; Townsend, C. A. J. Am. Chem.
Soc. 2000, 122 (38), 9296–9297.
(11) Hamed, R. B.; Batchelar, E. T.; Mecinovic, J.; Claridge, T. D. W.; Schofield,
C. J. ChemBioChem 2009, 10 (2), 246–250.
(12) Reider, P. J.; Grabowski, E. J. J. Tetrahedron Lett. 1982, 23 (22), 2293–
2296.
(13) Bodner, M. J.; Phelan, R. M.; Townsend, C. A. Org. Lett. 2009, 11 (16),
3606–3609.
(14) Salzmann, T. N.; Ratcliffe, R. W.; Christensen, B. G.; Bouffard, F. A. J. Am.
Chem. Soc. 1980, 102 (19), 6161–6163.
HPLC analyses of other carbapenam/ems (6, 11, 12, 14-17)
tested with ThnG and ThnQ did not show appearance of a
carbapenem chromophore or other new products. Additionally, in
Vitro reactions with carbapenam/ems (6, 11, 12, 14-17) employing
cell-free extracts harboring ThnG or ThnQ were analyzed with the
supersensitive E. coli SC 12155 and the Nitrocefin ꢀ-lactamase
induction assay using Bacillus licheniformis ATCC 14580.19,20 By
none of these three sensitive measures could antibiotic production
be detected.
(15) Ueda, Y.; Roberge, G.; Vinet, V. Can. J. Chem. 1984, 62 (12), 2936–
2940.
(16) Stapon, A.; Li, R. F.; Townsend, C. A. J. Am. Chem. Soc. 2003, 125 (51),
15746–15747.
(17) Salowe, S. P.; Marsh, E. N.; Townsend, C. A. Biochemistry 1990, 29 (27),
6499–6508.
(18) Decristoforo, G. Anal. Chim. Acta 1984, 163 (SEP), 25–33.
(19) Sykes, R. B.; Wells, J. S. J. Antibiot. 1985, 38 (1), 119–121.
(20) Gerratana, B.; Stapon, A.; Townsend, C. A. Biochemistry 2003, 42 (25),
7836–7847.
(21) Stapon, A.; Li, R. F.; Townsend, C. A. J. Am. Chem. Soc. 2003, 125 (28),
8486–8493.
These experiments demonstrate that the Fe(II)/R-KG-dependent
oxygenases ThnG and ThnQ oxidize the C-2 and C-6 side chains,
JA907320N
9
J. AM. CHEM. SOC. VOL. 132, NO. 1, 2010 13