REPORTS

24. S. A. Zimov et al., Clim. Change 33, 111120 (1996). 45. E. S. Kasischke et al., Can. J. Res. 40, 13131324 L.R.W. was supported by NASA award NNX11AF36G. Early
25. See section SM7 in the supplementary materials. (2010). observations at BRW were funded by U.S. Navy/Office of Naval
26. FRA, Global Forest Resources Assessment 2010 (Food 46. S. A. Zimov et al., Science 284, 19731976 (1999). Research contract N00014-67-A-0103-0007. Online access to
and Agriculture Organization of the United Nations, 47. L. R. Welp, J. T. Randerson, H. P. Liu, J. Geophys. Res. all observational data is summarized in section SM10 of the
Rome, 2010). 111, G03007 (2006). supplementary materials. NCAR is supported by the NSF. Any
27. G. C. Hurtt et al., Clim. Change 109, 117161 (2011). 48. S. J. Goetz, A. G. Bunn, G. J. Fiske, R. A. Houghton, opinions, findings, and conclusions or recommendations
28. Y. Pan et al., Science 333, 988993 (2011). Proc. Natl. Acad. Sci. U.S.A. 102, 1352113525 (2005). expressed in this material are those of the authors and do not
29. O. L. Phillips et al., Science 323, 13441347 (2009). 49. D. Verbyla, Glob. Ecol. Biogeogr. 17, 547555 (2008). necessarily reflect the views of NOAA, NSF, DOE or NASA. We
30. K. R. Gurney, W. J. Eckels, Tellus B Chem. Phys. Meterol. 50. P. E. Thornton, J.-F. Lamarque, N. A. Rosenbloom, thank the HIPPO science team and the crew and support staff
63, 328339 (2011). N. M. Mahowald, Global Biogeochem. Cycles 21, at the NCAR Research Aviation Facility. We acknowledge the
31. K. E. Taylor, R. J. Stouffer, G. A. Meehl, Bull. Am. GB4018 (2007). World Climate Research Programmes Working Group on
Meteorol. Soc. 93, 485498 (2012). 51. P. P. Tans, I. Y. Fung, T. Takahashi, Science 247, Coupled Modelling, which is responsible for CMIP, and we
32. See section SM9 in the supplementary materials. 14311438 (1990). thank the climate modeling groups for producing and making
33. A. D. McGuire et al., Global Biogeochem. Cycles 15, 52. K. R. Gurney et al., Nature 415, 626630 (2002). available their model output. Support of the CMIP data
183206 (2001). 53. B. B. Stephens et al., Science 316, 17321735 (2007). sets is provided by the Office of Science, U.S. Department
34. K. Schaefer et al., J. Geophys. Res. 117, G03010 (2012). 54. C. D. Keeling, S. C. Piper, T. P. Whorf, R. F. Keeling, of Energy. C. Roedenbeck provided assistance with the TM3
35. L. Rustad et al., Oecologia 126, 543562 (2001). Tellus B Chem. Phys. Meterol. 63, 122 (2011). model. P.K.P. is partially supported by the Ministry of
36. R. A. Houghton, J. Geophys. Res. 92, 4223 55. We specify the uncertainty in CO2 amplitude with 95% Education, Culture, Sports, Science and Technology Green
confidence intervals estimated using a jackknife Network of Excellence program. G.W.S. acknowledges support
(1987).
procedure (fig. S3). from the NSF Graduate Research Fellowship Program and
37. G. H. Kohlmaier et al., Tellus B Chem. Phys. Meterol. 41,
56. Only two models simulations extended to 2011 the Environmental Protection Agencys Science to Achieve
487510 (1989).
(HadGEM2-ES and NorESM-1). Other models output was Results program.
38. R. J. Norby et al., Proc. Natl. Acad. Sci. U.S.A. 102, available only through 2005, so NEP for 2009 to 2011 is
1805218056 (2005). given by the mean over 2001 to 2005. In these models,
39. P. D. Jones et al., J. Geophys. Res. 117, D05127 (2012). the change in NEP amplitude up to 2001 to 2005 may be Supplementary Materials
40. J. Hansen, R. Ruedy, M. Sato, K. Lo, Rev. Geophys. 48, www.sciencemag.org/cgi/content/full/science.1239207/DC1
10% smaller than the change up to 2009 to 2011.

Downloaded from http://science.sciencemag.org/ on October 19, 2017

RG4004 (2010). Materials and Methods
41. G. R. Walther et al., Nature 416, 389395 (2002). Figs. S1 to S9
Acknowledgments: The Scripps CO2 Program is supported by
42. S. C. Elmendorf et al., Nat. Clim. Change 2, 453457 Tables S1 to S7
DOE grant DE-SC0005090. HIPPO was supported by NSF grants
(2012). ATM-0628575, ATM-0628519, ATM-0628388, ATM-0628452, References (5781)
43. K. E. N. Tape, M. Sturm, C. Racine, Glob. Change Biol. and ATM-1036399, and by the National Center for Atmospheric 16 April 2013; accepted 17 July 2013
12, 686702 (2006). Research (NCAR). Recovery and updating of early aircraft, Published online 8 August 2013;
44. A. J. Soja et al., Global Planet. Change 56, 274296 (2007). MLO, and BRW data was supported by NSF grant ATM-1036399. 10.1126/science.1239207

Expanding the Fluorine Chemistry of powerful strategy for the design of synthetic phar-
maceuticals. Estimates indicate that 20 to 30% of
drugs, including many of the top sellers, contain
Living Systems Using Engineered at least one fluorine atom (57). Recent innova-
tions have expanded the scope of synthetic CF
Polyketide Synthase Pathways bond-forming methodologies, but the unusual ele-
mental properties of fluorine that serve as the basis
for its success also continue to restrict the range of
Mark C. Walker,1* Benjamin W. Thuronyi,2* Louise K. Charkoudian,3 Brian Lowry,4 molecular structures that can be accessed (811).
Chaitan Khosla,3,4,5 Michelle C. Y. Chang1,2 As such, the invention of alternative routes for the
site-selective introduction of fluorine into struc-
Organofluorines represent a rapidly expanding proportion of molecules that are used in turally diverse molecules, particularly under mild
pharmaceuticals, diagnostics, agrochemicals, and materials. Despite the prevalence of fluorine in conditions, remains an outstanding challenge.
synthetic compounds, the known biological scope is limited to a single pathway that produces In comparison to synthetic small molecules,
fluoroacetate. Here, we demonstrate that this pathway can be exploited as a source of fluorinated fluorine has limited distribution in naturally oc-
building blocks for introduction of fluorine into natural-product scaffolds. Specifically, we curring organic compounds; the only organofluo-
have constructed pathways involving two polyketide synthase systems, and we show that rine natural products characterized to date consist
fluoroacetate can be used to incorporate fluorine into the polyketide backbone in vitro. We further of a small set of simple molecules associated with
show that fluorine can be inserted site-selectively and introduced into polyketide products the fluoroacetate pathway of Streptomyces cattleya,
in vivo. These results highlight the prospects for the production of complex fluorinated natural a soil bacterium that has the ability to catalyze the
products using synthetic biology. formation of CF bonds from aqueous fluoride
(Fig. 1A) (12, 13). Although these compounds
he catalytic diversity of biological systems tion of pharmaceuticals, fuels, and materials lack the intricacy typically expected of secondary

T provides enormous potential for applica-

tion of living cells to the scalable produc-
(14). However, the scope of innovation of living
organisms is typically limited to functions that
confer a direct advantage for cell growth, thereby
metabolites, they represent a potentially rich source
of modular organofluorine building blocks for
the production of complex fluorinated natural pro-
1
maximizing biomass as the end product rather ducts. In this regard, the backbones of several large
Department of Molecular and Cell Biology, University of California,
Berkeley, Berkeley, CA 947201460, USA. 2Department of Chem-
than a distinct molecule or reaction of interest. In classes of medicinally relevant natural products
istry, University of California, Berkeley, Berkeley, CA 947201460, contrast, synthetic biology approaches allow us including polyketides, isoprenoids, steroids, alka-
USA. 3Department of Chemistry, Stanford University, Stanford, CA to disconnect some of these biochemical trans- loids, eicosanoids, leukotrienes, and othersare
94305, USA. 4Department of Chemical Engineering, Stanford formations from cell survival and reconnect them biosynthesized directly from the assembly and tai-
University, Stanford, CA 94305, USA. 5Department of Biochem- in different ways for the targeted synthesis of al- loring of simple acetate units (Fig. 1A). Introduc-
istry, Stanford University, Stanford, CA 94305, USA.
ternative classes of compounds. One particularly tion of the fluoroacetate monomer in place of
*These authors contributed equally to this work.
Present address: Department of Chemistry, Haverford College,
interesting area of opportunity is the development acetate would allow us to incorporate fluorine
Haverford, PA 19041, USA. of methods to introduce fluorine into complex into the backbone of these targets and create new
Corresponding author. E-mail: [email protected] small-molecule scaffolds, which has become a molecular function by combining the medicinal

www.sciencemag.org SCIENCE VOL 341 6 SEPTEMBER 2013 1089

 REPORTS
chemistry advantages of fluorine with the struc- to be accommodating with regard to their starter that the irreversible alkylation of active-site nucleo-
tural complexity and bioactivity of natural products. units (23), the encoding of extender units has philes could also create problems (27). Thus, the
For example, the introduction of fluorine via been found to be quite selective, and many cellu- development of a system to incorporate fluorinated
synthetic or semisynthetic routes has enabled lar acyl-CoAs are excluded from the backbone extender units could dramatically increase the range
the improvement of the clinical properties of sev- (22). However, progress in engineering extender- of complex structures that can be accessed but
eral natural products, but this remains challeng- unit incorporation has been made by domain en- must also address the challenges in activating the
ing to achieve (1417). Although previous studies gineering (2325) or incorporation via a domain fluoroacetate monomer for the downstream CC
have shown that distal fluorine substituents can that encodes a rare extender unit (17, 26). Al- bond-forming chemistry involved in chain-extension
be accommodated in natural-product biosyn- though fluoroacetate serves as a starter unit in reactions.
thetic pathways (18, 19), access to fluoromalonyl nature to produce highly toxic w-fluoro fatty acids Chain elongation in polyketides and related
coenzyme A (CoA), a fluorinated analog of one (Fig. 1A) (13), fluorine has never been observed fatty acidbased natural products relies on a separate
of natures most powerful carbon nucleophiles, as to date within the backbone, implying that chain- pool of extender units formed by carboxylation
an extender unit would enable a general meth- extension reactions with the fluorinated acyl-CoA of acyl-CoAs at the a position. These malonyl-
od for direct incorporation of fluorine into any do not occur in these systems. The apparent in- CoA derivatives are then used as masked enolates
polyketide structure. ability of living systems to use fluoroacetate for for CC bond formation after decarboxylation.
Many acetate-based natural products, poly- the biosynthesis of complex small molecules prob- The fluorinated extender, fluoromalonyl-CoA, can
ketides in particular, are generated through the ably results in part from the extreme properties be made through two routes: either (i) a two-step
iterative condensation of activated thioesters, re- of fluorine that affect biological and chemical activation of the biogenic fluoroacetate or (ii) a
sulting in reactive b-keto units that condense fur- synthesis. For example, the pKa (where Ka is the direct ligation of CoA to fluoromalonate (Fig. 2).
ther to produce a wide range of structures (20, 21) acid dissociation constant) of the a proton, elec- We reasoned that the acetate kinase (AckA)
(Fig. 1B). The structural diversity of polyketides trophilicity of the carbonyl group, and the stability phosphotransacetylase (Pta) pair would be effec-

Downloaded from http://science.sciencemag.org/ on October 19, 2017

is especially notable given that the majority of of the acyl-CoA and its corresponding carbanion tive at fluoroacetate activation, as mutations in
polyketides draw on only two monomers, acetate are all highly affected by fluorine substitution. Fur- this gene locus have been shown to lead to fluoro-
and propionate, as the extender units that form thermore, the fluoroacetyl group bears a clear acetate resistance in Escherichia coli (28). The
their carbon skeletons (3, 20, 22). Although similarity to the fluoromethylketone motif used enzymes from E. coli were overexpressed and char-
polyketide synthases (PKSs) have been observed for the design of covalent inhibitors, suggesting acterized biochemically, confirming that AckA

A
Met
OH F F Adenine S+ Adenine
_ O H3C
F COO O O O
F _
_
H F
NH3+ OPO32
HO OH HO OH HO OH

Fluorothreonine 5-Fluorodeoxyadenosine S-Adenosylmethionine

O
F _
O
Fluoroacetate

-Fluorofatty acids _
COO
O
O
AcO O O
F _ AcO O OH
O MeO OH OH Ph NH O OH
n O OH OH
H Eicosanoids/Leukotrienes
N Ph O
+ F
_ SCoA O OH
OOC OH O O _ HO H O
_ _ O COO BzO
OOC COO O
AcO

F Fluoroacetyl-CoA Polyketides Isoprenoids HO

Fluorocitrate > 10,000 > 50,000 Steroids

B O HO
O O
O O O O
O or
_ O
O SR
OH SR O O
O OH
Acetate Chain elongation
Cyclization and aromatization Lactonization

Fig. 1. Synthetic biology of fluorine. (A) The fluoroacetate pathway from acetate backbones (left to right, red box). Red dots represent po-
and its metabolites represent the known scope of biological fluorine sitions that could, in principle, be fluorinated by incorporation of a fluoroacetate
chemistry, starting with fluoride and S-adenosylmethionine, to produce monomer without altering the carbon skeleton, including locations where
fluoroacetate and fluorothreonine as the end products (right to left, gray fluorine would replace a methyl group derived from propionate or where
box). This scope could be greatly expanded by engineering downstream downstream tailoring steps have occurred on the final structure. (B) As-
pathways to use fluoroacetate as a building block for site-selective in- sembly of acetate units in the biosynthesis of polyketide natural products. R,
troduction of fluorine into large families of natural products constructed CoA or PKS.

1090 6 SEPTEMBER 2013 VOL 341 SCIENCE www.sciencemag.org

 REPORTS
and Pta serve as an effective activation system to fluoromalonyl-CoA is still produced at reason- and fluoromalonyl-CoA extender with only a five-
rapidly produce both acetyl- and fluoroacetyl- able efficiency (Fig. 2B and figs. S3 and S4). fold defect in catalytic efficiency (kcat/KM, where
CoA in nearly quantitative yield (figs. S1 and S2). Both of these systems also provide in situ re- kcat is the turnover rate and Km is the Michaelis
Analysis of the kinetic parameters for these en- generation capacity that can amplify product constant) derived from a drop in kcat with the flu-
zymes with respect to fluorinated substrates in- yields from polyketide synthases, and we found orinated substrate (Fig. 3). This lower turnover
dicated that neither appears to be affected by the that either system increased polyketide produc- rate observed with the fluorinated substrate is pos-
fluorine substituent beyond inductive effects that tion by tetrahydroxynaphthalene synthase (31) sibly related to the reduced reactivity of the enolate
alter the nucleophilicity of the carboxylic acid compared with simple addition of malonyl-CoA species, which would be stabilized by the fluorine
(AckA) or electrophilicity of the carbonyl (Pta) (fig. S5). substituent. However, the overall yield was com-
(29). Next, we purified the individual AccABCD We next turned our attention to the use of the parable for both fluorinated and nonfluorinated
subunits that make up the acetyl-CoA carboxyl- fluoromalonyl-CoA monomer for downstream substrates, which shows that a decarboxylative
ase (ACCase) from E. coli and added these en- chain-elongation reactions. To start, we exam- Claisen condensation with fluoromalonyl-CoA can
zymes to the AckA-Pta system to carry out the ined the behavior of a simple polyketide syn- take place at a similar extent of conversion com-
carboxylation of fluoroacetate in a one-pot reaction thase system with regard to one cycle of chain pared to malonyl-CoA. Furthermore, these exper-
to generate the fluoromalonyl-CoA extender unit extension and ketoreduction, which is a key iments also show that the 2-fluoro-3-keto motif
(Fig. 2A and fig. S1). Under these conditions, the functionality of larger multimodular systems produced with the fluoromalonyl-CoA extender
ligation of CoA by AckA-Pta to produce the for controlling downstream cyclization and rear- can be accepted by ketoreductases, as PhaB is ca-
acyl-CoA is rapid, and production of the carbox- rangements within the polyketide backbone (Fig. pable of efficiently reducing the acetofluoroacetyl-
ylated product is limited by the ACCase. Although 3A) (3, 20). We constructed a synthetic gene en- CoA substrate (fig. S7). The 1H and 19F nuclear
the rate of conversion is 4.5 times slower for fluo- coding NphT7 (32), which appears to be a free- magnetic resonance (NMR) spectra of the reduced
roacetate as compared with acetate, the overall standing ketosynthase that is related at the structural product indicate that both diastereomers are

Downloaded from http://science.sciencemag.org/ on October 19, 2017

extent of reaction is similar for both congeners, level to the ketosynthase domain of more com- produced in this reaction (fig. S7), which may re-
which suggests that covalent inactivation of the plex polyketide synthases (fig. S6), and we iso- sult from lack of stereochemical preference of
ACCase by fluoroacetyl-CoA is not important if lated the heterologously expressed enzyme for NphT7 with respect to the fluorine substituent or
it occurs. In addition to the route from fluoro- biochemical characterization (fig. S1). With the from racemization of the product before reduc-
acetate, we also tested a malonyl-CoA synthetase use of a coupled assay with an R-hydroxyl forming tion by PhaB. Although PhaB does not appear
(MatB) (30) for coupling CoA directly to fluoro- acetoacetyl-CoA reductase (PhaB), we found that to show diastereoselectivity with respect to the
malonate. Although MatB exhibits a 103-fold NphT7 is competent to catalyze the formation of fluorine group, the polyketide synthase keto-
selectivity for malonate over fluoromalonate, acetofluoroacetyl-CoA, using an acetyl-CoA starter reductases are known to be selective with regard
to their native a substituent and could potentially
A carry out the stereochemical resolution of the
O AckA_Pta O ACCase O O fluorine modification upon reduction (33).
_ _ With this information in hand, we sought to
R ATP, CoA R _
O SCoA ATP, HCO3 O SCoA extend our biosynthetic method for fluorine intro-
R duction to more complex polyketide synthase sys-
tems, which use the chain-elongation reaction for
R=H R= F the biosynthesis of many bioactive and clinically
400 400
Fluoromalonyl-CoA (M)

important natural products, such as erythromy-

Malonyl-CoA (M)

cin and rapamycin (3, 20). Of the multimodular

300 300
polyketide systems, 6-deoxyerythronolide B syn-
thase (DEBS) is probably the best understood and
200 200
is responsible for production of the erythromycin
precursor (34). We therefore focused our studies
100 100 on the sixth module of DEBS, including the ter-
minal thioesterase (DEBSMod6+TE) (35). Using a
0 0 diketide substrate [natural diketide N-acetyl cyste-
0 10 20 30 40 50 60 70 0 20 40 60 80 100
Time (min) Time (min) amine thioester (NDK-SNAC)], DEBSMod6+TE
can catalyze a single round of chain elongation
with its native methylmalonyl-CoA extender unit
B O O MatB O O and then cyclize the tethered product to form a
_ _ _ methyltriketide lactone (TKL) (Fig. 4A, R =
O O ATP, CoA O SCoA CH3; Fig. 4B, 1; and fig. S8) (36). We found that
R R DEBSMod6+TE is also able to accept the fluori-
R= kcat (s-1) KM (M) kcat/KM (M-1s-1) nated monomer in chain-extension catalysis to form
the 2-fluoro-2-desmethyltriketide lactone (F-TKL)
Malonate H 11.8 0.3 71 5 (1.7 0.1) x 105 and incorporate fluorine into the polyketide back-
Methylmalonate CH3 1.49 0.02 38 2 (3.9 0.2) x 104 bone (Fig. 4B, 2 to 4; and fig. S9). The identity of
Fluoromalonate F 0.29 0.01 550 40 (5.3 0.4) x 102
the F-TKL was established by comparison to an
authentic synthetic standard using reverse-phase
Fig. 2. Enzymatic production of activated extender units for CC bond-formation reactions. (A) liquid chromatographymass spectrometry (LC-
Formation of malonyl-CoA (left) and fluoromalonyl-CoA (right) from 500 mM CoA and either acetate or MS) monitored by electrospray ionization (ESI)
fluoroacetate, respectively. Values are reported as the mean T SD (n = 3 replicates). (B) Kinetic parameters and further confirmed by characterization of the
for malonate activation. Kinetic parameters are reported as mean T SE (n = 3), as determined from isolated compound by high-resolution MS, gas
nonlinear curve-fitting. Error in the kcat /KM parameter was obtained from propagation of error from the chromatographymass spectrometry, and 19F NMR
individual kinetic terms. spectroscopy (figs. S10 to S13). Although the

www.sciencemag.org SCIENCE VOL 341 6 SEPTEMBER 2013 1091

 REPORTS
2S keto tautomer is generated in 94% diaster- A
eomeric excess (figs. S12 and S13), this ratio O O O NphT7 O O PhaB OH O
appears to be set by the compounds stereoelec- +_ _
tronic factors rather than the stereochemical pref- CO2 NADPH
SCoA O SCoA SCoA SCoA
erence of DEBSMod6+TE, as the F-TKL is fully R R R
enolized in aqueous solution. The F-TKL can also
be produced directly from fluoroacetate using the
AckA-Pta/ACCase activation system in either a B R=H R= F
multistage (Fig. 4B, 5 and 6) or single-pot re- 5 0.8
action (Fig. 4B, 7 and 8) with DEBSMod6+TE in 4

Initial rate (s-1)

Initial rate (s-1)

a similar yield to the MatB reaction, which allows 0.6
us to connect fluorinated polyketide production 3
directly to the biosynthetically available fluori- 0.4
nated building block (Fig. 1A and scheme S1). 2
In contrast to the chain-extension reaction 1 0.2
catalyzed by NphT7, DEBSMod6+TE does not in-
corporate fluorinated extender units into the 0 0.0
0 40 80 120 160 0 40 80 120 160 200
triketide lactone product as efficiently as its na-
tive methylmalonyl-CoA extender. Preliminary Malonyl-CoA (M) Fluoromalonyl-CoA (M)
studies indicate that the reduced efficiency of R= kcat (s-1) KM (M) kcat/KM (M-1s-1)
DEBSMod6+TE with the fluorinated extender is

Downloaded from http://science.sciencemag.org/ on October 19, 2017

not due to covalent inactivation of the enzyme Malonyl-CoA H 5.1 0.1 21 2 (2.4 0.2) x 105
(fig. S14), but rather to the more complex bio- Fluoromalonyl-CoA F 0.81 0.03 15 2 (5.4 0.7) x 104
chemistry of polyketide synthases with regard to
monomer selection (37). Extender unit hydrol- Fig. 3. A chain-extension and ketoreduction cycle with a fluorinated extender using a simple
ysis, which occurs even for the native substrate polyketide synthase, NphT7. (A) Reactions catalyzed by NphT7 and PhaB. NADPH, reduced form of
(table S2), appears to limit fluoromalonyl-CoA nicotinamide adenine dinucleotide phosphate. (B) Steady-state kinetic parameters for NphT7-catalyzed
incorporation based on the observations that CC bond formation measured using a coupled assay with PhaB. Data points are reported as the mean T SD
MatB and adenosine triphosphate (ATP) are (n = 3). Kinetic parameters are reported as mean T SE (n = 3), as determined from nonlinear curve-fitting.
needed for fluoromalonyl-CoA regeneration and Error in the kcat /KM parameter was obtained from propagation of error from the individual kinetic terms.
that fluoromalonate remains the major organo-
fluorine species, even in their presence (fig. S15).
The fluoromalonyl-CoA extender is, however, in- hance F-TKL formation by the AT-null mutant (43) but almost no methylmalonyl-CoA (44, 45).
corporated at higher efficiency by DEBSMod6+TE (Fig. 4C). Using this approach, we began to explore We carried out preliminary 19F-NMR studies of
than malonyl-CoA (R = H), which is reported to the possibility of site-selective fluorine incorpo- cells expressing MatB, NphT7, and PhaB and fed
be naturally excluded by DEBS (38). In fact, ration with a mini-PKS model system, consist- with nontoxic levels of fluoromalonate. Analysis
DEBSMod6+TE produces at least 10 times more ing of DEBSMod2 and DEBSMod3+TE, that was of the media and cell extracts indicated that flux
F-TKL than H-TKL in a direct competition ex- designed to carry out two chain-extension re- through fluoromalonyl-CoA could reach 100 mM
periment with equimolar amounts (1 mM) of actions from the NDK-SNAC substrate (42). Using to 1 mM, which is sufficient for use by PKSs in
fluoromalonyl-CoA and malonyl-CoA (table S3). the appropriate AT-null constructs, we were able live cells (table S4). Next, we tested the ability of
To address the issue of site- or regioselective to observe exclusive production of either regio- DEBSMod6+TE to catalyze chain elongation in
fluorine incorporation, we turned our attention to isomer of the fluoro-methyl tetraketide lactone cell lysates prepared from E. coli BAP1 coex-
exploiting the greater reactivity of the fluorinated (tetraKL). The identity of the 2-fluoro-4-methyl pressing DEBSMod6+TE and MatB. Under these
extender unit toward acylation reactions. In this tetraKL and 2-methyl-4-fluoro tetraKL was es- conditions, F-TKL is produced with no observ-
regard, we hypothesized that it would be possible tablished by both high-resolution ESI-MS and able H-TKL upon addition of only NDK-SNAC,
for a fluorinated substrate to selectively acylate LC-MS on the basis of their different retention fluoromalonate, CoA, ATP, and the ATP regen-
either the acyltransferase (AT) or acyl carrier times, as well as their mass fragmentation pat- eration system (fig. S18A). Negative controls with
protein domains of individual DEBS modules in terns, which are consistent with the incorpora- either no DEBSMod6+TE/MatB expressed or no
the presence of a catalytically compromised or tion of fluorine at the expected sites (Fig. 4D NDK-SNAC substrate show no production of
inactive AT domain, an approach that has been and fig. S17). These studies also indicate that F-TKL (fig. S18A). These results demonstrate
shown to facilitate malonyl incorporation by further chain extension after fluorine insertion that the intracellular level of expression of the
DEBS (39). Experiments with DEBSMod6+TE can be achieved and that fluorinated intermedi- DEBSMod6+TE and MatB enzymes is sufficient
showed that not only does F-TKL yield increase as ates could potentially be tolerated in downstream for the incorporation of the fluorinated extender
expected, but fluorine selectivity also improves reactions. This observation is consistent with pre- unit. They also further imply that fluorine could
upon introduction of a key S2107A (Ser2107 vious work that has shown that intermediates be introduced into the polyketide backbone in-
Ala2107) mutation, reversing the selectivity of the with non-native substituents, including fluorine, side living cells, which are capable of generat-
wild-type module (Fig. 4C). When the NDK- can be extended and tailored to the final structure ing ATP through normal metabolic processes.
SNAC substrate is used with its native module, (3, 1720, 23) and gives promise that larger flu- Therefore, we cultured E. coli BAP1 coexpressing
DEBSMod2, in conjunction with the analogous orinated polyketide targets may be accessible DEBSMod6+TE and MatB and harvested the
S2652A mutation, extension with fluoromalonyl- through this approach. cells after induction. These cells were then fed
CoA to form F-TKL reaches 30% efficiency The observed selectivity for fluoromalonyl- over with the fluoromalonate precursor, which resulted
compared with methylmalonyl-CoA (fig. S16). Fur- malonyl-CoA extender units suggests that polyketide in the production of F-TKL upon addition of
thermore, we found that the stand-alone trans-AT chain-extension reactions with fluoromalonyl-CoA NDK-SNAC (Fig. 4B, 9; and fig. S18B). The
from the disorazole polyketide synthase (40, 41) could possibly be catalyzed in vivo in E. coli, which identity of the F-TKL under these conditions
accepts fluoromalonyl-CoA and can further en- contains a sizable malonyl-CoA pool (~35 mM) was established by LC-MS, co-injection with an

1092 6 SEPTEMBER 2013 VOL 341 SCIENCE www.sciencemag.org

 REPORTS

A
O
C KS AT KR* ACP
Wild-type
TE KS
X AT KR* ACP
ATnull
TE KS
X
AT KR* ACP TE
ATnull + trans-AT
+ AT

DEBSMod6 TE 4000
OH O R 3000
+ CO2 2000 TKL (R = CH3)
1000
SNAC O O O O 80 F-TKL (R = F )

TKL or F-TKL (M)

70
_
O SCoA 60
R = CH3 TKL
R F F-TKL 50
H H-TKL
40
30
20
B 10
0
Mod3 Mod6 Mod3 Mod6 Mod3 Mod6

D DEBSMod2 DEBSMod3+TE O
E. coli cells (9)
Co-injection (8) OH
KS AT KR ACP KS AT ACP TE
One-pot, fluoroacetate (7) O O

Co-injection (6) Me/Me (1)

O
Telescoped, fluoroacetate (5)
F
Co-injection (4)
MatB, fluoromalonate (3)
KS AT KR ACP KS
X
AT ACP TE
OH

O O
F /Me (3)
F-TKL standard (2) Me/F (2)
Me/ F (2)
MatB, methylmalonate (1) O

Downloaded from http://science.sciencemag.org/ on October 19, 2017

4 6 8 10
Time (min)
12 14 KS
X AT KR ACP KS AT ACP TE OH
F

2 4 6 8 10
Me/Me (1)

O O
Time (min)
F /Me (3)

Fig. 4. Production of fluorinated polyketides in vitro and in vivo. Values are reported as the mean T SD (n = 3). KR* denotes that the KR domain
(A) Reaction catalyzed by DEBSMod6+TE using the NDK-SNAC substrate with of Mod3 is inactive. (D) LC-MS traces showing regioselective tetraketide lactone
various extender units (NDK-SNAC, (2S,3R)-2-methyl-3-hydroxypentanoyl-N- formation using the DEBS mini-PKS consisting of DEBSMod2 and DEBSMod3+TE
acetylcysteamine thioester). (B) Chain extension by DEBSMod6+TE to form (Me/Me, 2-methyl-4-methyl-tetraketide lactone, m/z = 227; Me/F, 2-fluoro-4-
triketide lactones monitored by LC-MS [TKL, mass/charge ratio (m/z) = 169; methyl-tetraketide lactone, m/z = 231; F/Me, 2-methyl-4-fluoro-tetraketide lactone,
F-TKL, m/z = 173]. CoA, ATP, and ATP regeneration system are included in all m/z = 231). Me/Me was produced using DEBSMod2/DEBSMod3+TE and methyl-
in vitro reactions. Data are normalized with respect to the TKL peak. (C) Selectivity malonate (1). Me/F was produced using DEBSMod2/DEBSMod3AT0+TE, DszsAT,
of DEBSMod6+TE and DEBSMod3+TE for the methylmalonyl-CoA versus methylmalonyl-CoA, and fluoromalonate (2). F/Me was produced using
fluoromalonyl-CoA extender unit, as monitored by TKL (m/z = 169) and F-TKL DEBSMod2AT0/DEBSMod3+TE, methylmalonyl-CoA, and fluoromalonate (3).
(m/z = 173) formation. Conditions include wild-type modules, AT0 modules, and Data are normalized with respect to the Me/Me peak. All reactions contained
AT0 modules in conjunction with the trans-AT from the disorazole PKS (DszsAT). MatB and the ATP regeneration system.

authentic standard, as well as high-resolution MS. 6. D. OHagan, Chem. Soc. Rev. 37, 308319 (2008). 25. I. Koryakina, J. B. McArthur, M. M. Draelos, G. J. Williams,
Moreover, F-TKL can also be produced directly 7. T. Furuya, A. S. Kamlet, T. Ritter, Nature 473, 470477 Org. Biomol. Chem. 11, 44494458 (2013).
(2011). 26. A. S. Eustquio, D. OHagan, B. S. Moore, J. Nat. Prod.
in cell culture with the simple addition of a mix- 8. N. D. Ball, M. S. Sanford, J. Am. Chem. Soc. 131, 73, 378382 (2010).
ture of both substrates to the media after induc- 37963797 (2009). 27. J. C. Powers, J. L. Asgian, . D. Ekici, K. E. James,
tion of DEBSMod6+TE and MatB (fig. S18C). 9. D. A. Watson et al., Science 325, 16611664 (2009). Chem. Rev. 102, 46394750 (2002).
Taken together, these studies show that the nat- 10. V. Rauniyar, A. D. Lackner, G. L. Hamilton, F. D. Toste, 28. T. D. K. Brown, M. C. Jones-Mortimer, H. L. Kornberg,
Science 334, 16811684 (2011). J. Gen. Microbiol. 102, 327336 (1977).
ural selectivity of the polyketide synthase allows 11. E. Lee et al., Science 334, 639642 (2011). 29. M. C. Walker, M. Wen, A. M. Weeks, M. C. Y. Chang, ACS
for the site-selective introduction of fluorine 12. C. Dong et al., Nature 427, 561565 (2004). Chem. Biol. 7, 15761585 (2012).
over hydrogen into the polyketide backbone in- 13. D. OHagan, J. Fluor. Chem. 127, 14791483 30. A. J. Hughes, A. Keatinge-Clay, Chem. Biol. 18, 165176
side living cells. (2006). (2011).
14. A. Rivkin, K. Biswas, T.-C. Chou, S. J. Danishefsky, 31. M. Izumikawa et al., J. Ind. Microbiol. Biotechnol. 30,
Using engineered pathways to link simple bio- Org. Lett. 4, 40814084 (2002). 510515 (2003).
genic organofluorine building blocks into more 15. J.-P. Bgu, D. Bonnet-Delpon, J. Fluor. Chem. 127, 32. E. Okamura, T. Tomita, R. Sawa, M. Nishiyama,
complex fluorinated small-molecule targets, we 9921012 (2006). T. Kuzuyama, Proc. Natl. Acad. Sci. U.S.A. 107,
have demonstrated that we can expand the fluo- 16. B. Llano-Sotelo et al., Antimicrob. Agents Chemother. 54, 1126511270 (2010).
49614970 (2010). 33. A. P. Siskos et al., Chem. Biol. 12, 11451153 (2005).
rine chemistry of living systems. Because of the
17. S. Mo et al., J. Am. Chem. Soc. 133, 976985 (2011). 34. C. Khosla, Y. Tang, A. Y. Chen, N. A. Schnarr, D. E. Cane,
modular nature of the biosynthetic pathways used 18. W. Runguphan, J. J. Maresh, S. E. OConnor, Proc. Natl. Annu. Rev. Biochem. 76, 195221 (2007).
to produce polyketides and related acetate-derived Acad. Sci. U.S.A. 106, 1367313678 (2009). 35. R. S. Gokhale, S. Y. Tsuji, D. E. Cane, C. Khosla, Science
natural products, these findings open the door to 19. R. J. M. Goss et al., ChemBioChem 11, 698702 284, 482485 (1999).
general strategies for exploring the fluorine syn- (2010). 36. N. Wu, F. Kudo, D. E. Cane, C. Khosla, J. Am. Chem. Soc.
20. J. Staunton, K. J. Weissman, Nat. Prod. Rep. 18, 122, 48474852 (2000).
thetic biology of complex natural products. 380416 (2001). 37. S. A. Bonnett et al., Chem. Biol. 18, 10751081 (2011).
21. R. Croteau, T. M. Kutchan, N. G. Lewis, in Biochemistry 38. G. F. Liou, J. Lau, D. E. Cane, C. Khosla, Biochemistry 42,
References and Notes and Molecular Biology of Plants, R. B. Buchanan, 200207 (2003).
1. D. K. Ro et al., Nature 440, 940943 (2006). W. Gruissem, R. Jones, Eds. (American Society of Plant 39. P. Kumar, A. T. Koppisch, D. E. Cane, C. Khosla,
2. S. Atsumi, T. Hanai, J. C. Liao, Nature 451, 8689 (2008). Biologists, Rockville, MD, 2000), pp. 12501318. J. Am. Chem. Soc. 125, 1430714312 (2003).
3. D. E. Cane, C. T. Walsh, C. Khosla, Science 282, 6368 22. Y. A. Chan, A. M. Podevels, B. M. Kevany, M. G. Thomas, 40. F. T. Wong, A. Y. Chen, D. E. Cane, C. Khosla,
(1998). Nat. Prod. Rep. 26, 90114 (2009). Biochemistry 49, 95102 (2010).
4. A. M. Weeks, M. C. Y. Chang, Biochemistry 50, 54045418 23. R. McDaniel et al., Proc. Natl. Acad. Sci. U.S.A. 96, 41. F. T. Wong, X. Jin, I. I. Mathews, D. E. Cane, C. Khosla,
(2011). 18461851 (1999). Biochemistry 50, 65396548 (2011).
5. K. Mller, C. Faeh, F. Diederich, Science 317, 18811886 24. U. Sundermann et al., ACS Chem. Biol. 8, 443450 42. S. Y. Tsuji, D. E. Cane, C. Khosla, Biochemistry 40,
(2007). (2013). 23262331 (2001).

www.sciencemag.org SCIENCE VOL 341 6 SEPTEMBER 2013 1093

 REPORTS
43. B. D. Bennett et al., Nat. Chem. Biol. 5, 593599 (2009). The College of Chemistry NMR Facility at the University of Supplementary Materials
44. T. Haller, T. Buckel, J. Rtey, J. A. Gerlt, Biochemistry 39, California, Berkeley (UC Berkeley) is supported in part by the www.sciencemag.org/cgi/content/full/341/6150/1089/DC1
46224629 (2000). NIH (grants 1S10RR023679-01 and S10 RR16634-01). M.C.W. Materials and Methods
45. B. A. Pfeifer, S. J. Admiraal, H. Gramajo, D. E. Cane, and B.W.T acknowledge the support of a NIH National Figs. S1 to S18
C. Khosla, Science 291, 17901792 (2001). Research Service Award Training Grant (1 T32 GMO66698) and Tables S1 to S4
the Gerald K. Branch Predoctoral Fellowship and UC Cancer Scheme S1
Acknowledgments: We thank B. Bond-Watts and I. Aanei for Research Council Committee Predoctoral Fellowship (to B.W.T). References (4660)
assembly and cloning of the synthetic nphT7 gene, X. Yu for L.K.C. acknowledges support from the National Cancer Institute
providing plasmids for the E. coli ACCase, C. Harvey for (grant F32 CA137994). This work was funded by generous
synthetic chemistry assistance, F. T. Wong for helpful support from UC Berkeley and the NIH (grant 1 DP2 24 June 2013; accepted 5 August 2013
discussions, and A. Lingel (Novartis) for QCI-F CryoProbe use. OD008696 to M.C.Y.C. and grant R01 GM087934 to C.K.). 10.1126/science.1242345

imperfect agreement with theory in the latter

Observation of Partial Wave studies. Here, we report a joint experimental and
theoretical study on O2H2 inelastic collisions for
Resonances in Low-Energy O2 (N = 1, j = 0 N = 1, j = 1) rotational ex-
citation, a transition which violates the semi-
O2H2 Inelastic Collisions classical propensity rules for rotational energy
transfer involving a molecule in a 3S state (10). The
results demonstrate the purely quantum mechan-
Simon Chefdeville,1,2 Yulia Kalugina,3,4 Sebastiaan Y. T. van de Meerakker,5 ical nature of the process and offer a complete
Christian Naulin,1,2 Franois Lique,3* Michel Costes1,2* characterization of fully resolved partial wave res-

Downloaded from http://science.sciencemag.org/ on October 19, 2017

onances with close agreement between theory
Partial wave resonances predicted to occur in bimolecular collision processes have proven challenging and experiment.
to observe experimentally. Here, we report crossed-beam experiments and quantum-scattering We performed our experiments with a crossed-
calculations on inelastic collisions between ground-state O2 and H2 molecules that provide state-to-state beam apparatus, which allowed us to tune the
cross sections for rotational excitation of O2 (rotational state N = 1, j = 0) to O2 (N = 1, j = 1) in the collision energy (the relative translational energy
vicinity of the thermodynamic threshold at 3.96 centimeter1. The close agreement between experimental of colliding partners with reduced mass m and
and theoretical results confirms the classically forbidden character of this collision-induced transition, relative velocity vr) by varying the beam inter-
which occurs exclusively in a purely quantum mechanical regime via shape and Feshbach resonances section angle, c, between 12.5 and 90: ET =
arising from partial waves with total angular momentum ( J) = 2 to 4. mvr2 = m(vO22 + vH22 2vO2vH2cosc) (9, 11).
H2 and O2 beams with low and matched ve-
he accurate description of collisions be- molecular collision experiments remains highly locities and high velocity resolution and quantum

T tween individual molecules remains a

long-standing goal in physical chemistry.
In a classical picture, the outcome of a collision
challenging. In fact, partial waves can only be
distinguished when giving rise to scattering reso-
nances that manifest as resolved peaks in dif-
state purity were obtained by using cryogenically
cooled Even-Lavie fast-pulsed valves (12) [see table
S1 (13)]. Neat beams of para-H2 and normal-H2
is determined by the interaction potential and ini- ferential or integral cross sections (DCSs and ICSs) were used and were characterized in the crossing
tial conditions such as relative velocity and im- as measured in crossed-beam experiments. Such region by 3 + 1 resonance-enhanced multiphoton
pact parameter, that is, the distance of closest very rare findings then provide a unique insight ionization (REMPI) time-of-flight mass spectrom-
approach between the molecules in the absence in the most critical part of the potential energy etry using three-photon (C1Pu, v = 0 X1Sg+,
of any interaction. Surely, molecules are quantum surface (PES), the transition state region (2). How- v = 0) R branch transitions near 99,150 cm1 (14).
objects in nature, and the collisions must be de- ever, in most cases, the collision energy spread Populations >95% for j = 0 and <5% for j = 1 were
scribed in terms of interfering quantum-scattering and the participation of many overlapping par- deduced when using freshly prepared samples
states, or partial waves, instead. Each partial wave tial waves corresponding to the energetically al- of paraH2 and 25% j = 0, 75% j = 1 when using
is the quantum mechanical analog of a classical lowed values of J tend to average the individual normalH2. The O2 beam (0.3% O2 in He) was
trajectory with a given impact parameter and is resonance signatures and still render them elu- probed by using 2 + 1 (3S0, v = 2 X3Sg, v = 0)
characterized by a definite value of total angular sive to experimental observation. and (3S1, v = 2 X3Sg, v = 0) REMPI
momentum, J, which is conserved throughout the Partial wave resonances were first observed transitions around 88,900 cm1 (15). The relative
collision (1). for a few elastic collisions that involve simple populations of the three fine-structure components
The contribution of individual partial waves atoms (35) and then for the F + HD HF + D of ground rotational state N = 1 were estimated to
to the scattering cross sections constitutes the system. In this textbook three-atom reaction, the be >85% in j = 0, <15% in j = 2 at E1,2 = 2.08 cm1,
ultimate information that can be obtained about resonances could be identified by comparison and <0.5% in j = 1 at E1,1 = 3.96 cm1, whereas
a collision event. Whereas molecular scattering of experimentally determined cross sections with those of first excited rotational state N = 3 were
processes are nowadays described successfully the outcome of high-level electronic structure and negligible (13). Cross-section measurements at-
by quantum mechanical methods throughout, the quantum-scattering calculations. A clear oscil- tributed to N = 1, j = 0 N = 1, j = 1 rotational
direct observation of individual partial waves in latory structure assigned to J = 12 to 14 partial energy transfer were performed by probing the
1
waves appears in the collision energy dependence collision-induced population in the N = 1, j =
Universit de Bordeaux, Institut des Sciences Molculaires,
of the state and angle-resolved DCS (6), where- 1 level as a function of the crossing angle. The
33400 Talence, France. 2CNRS, UMR 5255, 33400 Talence,
France. 3Laboratoire Ondes et Milieux Complexes, UMR 6294 as a steplike feature characterizes the resonance initial residual population in N = 1, j = 1 level was
CNRSUniversit du Havre, 25 rue Philippe Lebon, BP 540, behavior of the total ICS (7). More recently, reso- offset by shot-to-shot background subtraction when
76058 Le Havre, France. 4Department of Optics and Spectros- nance structures have been revealed in total ICSs triggering the probe laser and the O2 beam at 10 Hz
copy, Tomsk State University, 36 Lenin av., Tomsk 634050, for Penning ionization reactions of metastable with the H2 beam at 5 Hz. The integral cross sec-
Russia. 5Radboud University Nijmegen, Institute for Mole-
cules and Materials, Heijendaalseweg 135, 6525 AJ Nijmegen, He with H2 or Ar in the cold regime (8) and in tions for O2para-H2 collisions between ET =
Netherlands. state-to-state ICSs for low-energy inelastic scat- 3.7 cm1 and 20 cm1 are displayed in Figs. 1A and
*Corresponding author. E-mail: [email protected] tering of CO with H2 (9). Yet, the identification 2. Those for O2normal-H2 collisions are shown
(M.C.); [email protected] (F.L.) of the contributing partial waves suffered from in Fig. 2. The excitation function displays well-

1094 6 SEPTEMBER 2013 VOL 341 SCIENCE www.sciencemag.org

Expanding the Fluorine Chemistry of Living Systems Using Engineered Polyketide Synthase
Pathways
Mark C. Walker, Benjamin W. Thuronyi, Louise K. Charkoudian, Brian Lowry, Chaitan Khosla and Michelle C. Y. Chang

Science 341 (6150), 1089-1094.

DOI: 10.1126/science.1242345

Stitching in Fluoroacetate
Polyketide synthase enzymes stitch together an impressively diverse series of organic compounds from simple
acetate and propionate building blocks. Walker et al. (p. 1089) now show that these biochemical pathways can be
engineered to incorporate fluoroacetatea primary product of the only known native enzymatic fluorination routeinto

Downloaded from http://science.sciencemag.org/ on October 19, 2017

tri- and tetraketides. In Escherichia coli cells, this process shows potential as a versatile means of inserting fluorine
substituents into a range of complex molecules for use in pharmaceutical and agrochemical research.

ARTICLE TOOLS http://science.sciencemag.org/content/341/6150/1089

SUPPLEMENTARY http://science.sciencemag.org/content/suppl/2013/09/04/341.6150.1089.DC1
MATERIALS

RELATED http://science.sciencemag.org/content/sci/341/6150/1052.1.full
CONTENT

REFERENCES This article cites 59 articles, 12 of which you can access for free
http://science.sciencemag.org/content/341/6150/1089#BIBL

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Use of this article is subject to the Terms of Service

Science (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement of
Science, 1200 New York Avenue NW, Washington, DC 20005. 2017 The Authors, some rights reserved; exclusive
licensee American Association for the Advancement of Science. No claim to original U.S. Government Works. The title
Science is a registered trademark of AAAS.