Deciphering the Late Biosynthetic Steps of Antimalarial Compound FR-900098

Article abstract of DOI:10.1016/j.chembiol.2009.12.009

FR-900098 is a potent chemotherapeutic agent for the treatment of malaria. Here we report the heterologous production of this compound in Escherichia coli by reconstructing the entire biosynthetic pathway using a three-plasmid system. Based on this system, whole-cell feeding assays in combination with in vitro enzymatic activity assays reveal an unusual functional role of nucleotide conjugation and lead to the complete elucidation of the previously unassigned late biosynthetic steps. These studies also suggest a biosynthetic route to a second phosphonate antibiotic, FR-33289. A thorough understanding of the FR-900098 biosynthetic pathway now opens possibilities for metabolic engineering in E. coli to increase production of the antimalarial antibiotic and combinatorial biosynthesis to generate novel derivatives of FR-900098.

Full text of DOI:10.1016/j.chembiol.2009.12.009

Chemistry & Biology
Article
Deciphering the Late Biosynthetic Steps
of Antimalarial Compound FR-900098
Tyler W. Johannes,1,6,7 Matthew A. DeSieno,1,2,6 Benjamin M. Griffin,2,3 Paul M. Thomas,2,4 Neil L. Kelleher,2,4
William W. Metcalf,2,3 and Huimin Zhao1,2,4,5,
*
1Department of Chemical and Biomolecular Engineering
2Institute for Genomic Biology
3Department of Microbiology
4Department of Chemistry
5Departments of Biochemistry and Bioengineering, Center for Biophysics and Computational Biology
University of Illinois at Urbana-Champaign, Urbana, IL 61801, USA
6These authors contributed equally to this work
7Present address: Department of Chemical Engineering, The University of Tulsa, 800 S. Tucker Drive, Tulsa, OK 74104, USA
*Correspondence: zhao5@illinois.edu
DOI 10.1016/j.chembiol.2009.12.009
SUMMARY
Gram-positive bacteria, plants, and the parasite causing the
most virulent form of malaria, Plasmodium falciparum (Boucher
FR-900098 is a potent chemotherapeutic agent for and Doolittle, 2000; Lange et al., 2000). Mammals produce iso-
prenoids via the mevalonate pathway; therefore, FR-900098
and fosmidomycin should not be detrimental to humans (Beytia
and Porter, 1976). Both compounds are currently undergoing
the treatment of malaria. Here we report the heterolo-
gous production of this compound in Escherichia coli
by reconstructing the entire biosynthetic pathway
using a three-plasmid system. Based on this system,
whole-cell feeding assays in combination with in vitro
enzymatic activity assays reveal an unusual func-
tional role of nucleotide conjugation and lead to the
complete elucidation of the previously unassigned
clinical trials and have shown great promise as malaria treat-
ments (Borrmann et al., 2004, 2006). Earlier work revealed that
FR-900098 is twice more effective than fosmidomycin against
various strains of P. falciparum in vitro and the closely related
P. vinckei in mice (Jomaa et al., 1999). Progress has been
made in the development of chemical synthetic routes for fosmi-
late biosynthetic steps. These studies also suggest
a biosynthetic route to a second phosphonate antibi-
otic, FR-33289. A thorough understanding of the
FR-900098 biosynthetic pathway now opens possi-
bilities for metabolic engineering in E. coli to increase
production of the antimalarial antibiotic and combi-
natorial biosynthesis to generate novel derivatives
of FR-900098.
domycin (Hemmi et al., 1982) and FR-900098 (Fokin et al., 2007;
Hemmi et al., 1982), but we chose to pursue a biological route for
a more cost-effective way to produce these antimalarial
compounds.
Heterologous production of FR-900098 in Streptomyces livid-
ans and a proposed biosynthetic pathway were reported by Eliot
et al. (2008). The first step involved in FR-900098 biosynthesis,
a phosphoenolpyruvate (PEP) mutase reaction catalyzed by
FrbD, is common in almost all other known phosphonate
biosyntheses. Currently, only three other complete phosphonate
biosynthetic pathways have been identified (2-aminoethyl-
phosphonate [Barry et al., 1988], fosfomycin [Hidaka et al.,
INTRODUCTION
Malaria remains one of the most common life-threatening infec- 1995; Woodyer et al., 2006], and bialaphos [Blodgett et al.,
tions in developing countries, with up to 500 million infections 2005; Schwartz et al., 2004]), all of which have a common
and over 1 million deaths per year (Snow et al., 2005). Current second step. In all of these cases, an irreversible thiamine pyro-
antimalarial treatments include chloroquine and sulfadoxin pyri- phosphate-dependent decarboxylation of phosphonopyruvate
methamine, but the rapid emergence of antibiotic-resistant (PnPy) to form phosphonoacetaldehyde is required to provide
strains is rendering these current therapies futile, creating an a sufficient thermodynamic driving force to overcome the
urgent need for new effective drugs (Ridley, 2002; Wiesner unfavorable equilibrium created by PEP mutase. However, the
et al., 2003). FR-900098 and fosmidomycin represent a new FR-900098 biosynthetic cluster does not contain a PnPy decar-
class of antimalarial compounds. These compounds were first boxylase. FrbC, a homocitrate synthase homolog, catalyzes the
isolated from Streptomyces rubellomurinus and Streptomyces formation of 2-phosphonomethylmalate, thus creating a novel
lavendulae, respectively (Okuhara et al., 1980a). Studies have route for phosphonate biosynthesis (Eliot et al., 2008).
shown that these two compounds bind in the active site and
The subsequent steps parallel the tricarboxylic cycle, resulting
effectively inhibit 1-deoxy-D-xylulose 5-phosphate reductoiso- in the formation of 2-oxo-4-phosphonobutyrate, an analog of
merase (DXR), the first step in the nonmevalonate pathway for a-ketoglutarate. The final steps were proposed based only on
isoprenoid biosynthesis (MacSweeney et al., 2005). Genomic sequence homology of the remaining genes, without any defini-
analyses indicate that the genes encoding the nonmevalonate tive assignment of order or function. A complete understanding
pathway are present in a wide variety of Gram-negative and of the FR-900098 pathway could provide significant insight into
Chemistry & Biology 17, 57–64, January 29, 2010 ª2010 Elsevier Ltd All rights reserved 57

Chemistry & Biology
FR-900098 Biosynthesis
the biosynthesis of all phosphonates. More importantly, heterol-
ogous expression inside a more genetically tractable host would
act as a platform toward creating novel and potentially more
potent derivatives of FR-900098 and increasing production of
the antimalarial compound to industrially desirable levels. Here
we report the heterologous production of FR-900098 in E. coli
and the deciphering of the late steps in FR-900098 biosynthesis,
verified through whole-cell feeding experiments and in vitro
assays with purified enzymes.
RESULTS AND DISCUSSION
Heterologous Production of FR-900098 in E. coli
To achieve heterologous production of FR-900098 in E. coli,
each of the genes in the biosynthetic gene cluster were first
checked for soluble expression inside the new host. Each of
the 11 genes was cloned, expressed, and then analyzed by
SDS-PAGE and western blot. All of the proteins had high expres-
sion and showed clear overexpression bands in E. coli, except
for FrbE and DxrB, which demonstrated low levels of soluble
expression detected only by western blot analysis (data not
shown). To reconstitute the FR-900098 biosynthetic pathway
in E. coli, each of the 11 genes in the biosynthetic cluster was
cloned into a set of three compatible Duet expression vectors
(pETDuet, pRSFDuet, and pACYCDuet). Each individual gene
is under the transcriptional control of a strong T7 promoter
and expression is inducible by the addition of isopropyl-b-D-
thiogalactopyranoside (IPTG). The constructed plasmids were
transformed into the heterologous host E. coli to create the
production strain. The heterologous production of FR-900098
was confirmed using liquid chromatography/mass spectrometry
(LC-MS) analysis as described in Eliot et al. (2008).
Figure 1. Heterologous Production of FR-900098 in E. coli
(A) Scheme for the three-plasmid system in an FR-900098-producing strain.
Each gene is under transcriptional control by strong T7 promoters indicated
by arrows.
(B) Time course production of FR-900098 in E. coli. Error bars represent stan-
dard deviations.
Deletion analysis was utilized to determine which genes were phosphonobutyrate (2APn), a glutamate analog which is effi-
required for heterologous production of FR-900098. Different ciently taken up by E. coli and also shown to crossfeed into
strains of E. coli were created which lacked a single gene from S. lividans deletion mutants (data not shown). Culture superna-
the pathway and each was tested for its ability to produce tants were analyzed for possible phosphonate intermediates
the compound. The strain lacking dxrB grew very slowly using a coupled LC-MS approach. Cells expressing FrbH
upon protein induction and, although the strain eventually were found to accumulate CMP-50-3-aminopropylphosphonate
grew to saturation, it produced significantly lower amounts of (CMP-50-3APn), and cells expressing FrbH, FrbF, and FrbG
FR-900098. As mentioned earlier, the dxrB gene encodes had detectable levels of CMP-50-N-acetyl-3-aminopropyl-
a homolog of 1-deoxy-D-xylulose-5-phosphate reductoisomer- phosphonate (CMP-50-Ac3APn), CMP-50-FR-900098, and FR-
ase, the known target of the antibiotic. This provided evidence 900098. Thus, these initial whole-cell feeding studies suggested
for the necessary inclusion of the dxrB gene in the E. coli that cytidine-50-monophosphate (CMP) -conjugated phospho-
FR-900098-producing strains as a means of resistance from nate intermediates are present during the late steps of
increasing concentrations of the compound. The genes frbI FR-900098 biosynthesis. These studies also showed that FrbG
and frbJ were not essential for FR-900098 production in is an essential component in the late steps of FR-900098 biosyn-
E. coli; the overexpression of these two proteins more than likely thesis, as definitive assignment of the function of the enzyme
results in the unnecessary consumption of cellular resources, was never made (Eliot et al., 2008). It is also interesting to note
which results in lower FR-900098 production. The final produc- that 2-oxo-4-phosphonobutyrate was detected in samples
tion strain (pETDuet-frbABCD, pRSFDuet-frbHFG, pACYCDuet- containing a negative control (empty vector), providing evidence
frbE-dxrB) did not include frbI or frbJ, resulting in the highest that a ubiquitous cellular transaminase is able to carry out the
production of FR-900098, 6.3 mg/L (Figure 1). To our knowledge, transamination of 2OPn to form 2APn.
this is the first known heterologous production of a phosphonate
antibiotic in E. coli.
Bifunctional Activity of FrbH
The role of frbH was investigated after overexpression and puri-
fication of a histidine-tagged FrbH fusion from E. coli. Based on
Whole-Cell Feeding Studies
To identify possible intermediates in the late steps of FR-900098 sequence homology to other known PLP-dependent enzymes,
biosynthesis, a series of whole-cell feeding experiments was the protein was originally believed to catalyze the decarboxyl-
conducted. Resting cells of E. coli were fed with 2-amino-4- ation of 2APn to yield 3APn (Eliot et al., 2008). However, repeated
58 Chemistry & Biology 17, 57–64, January 29, 2010 ª2010 Elsevier Ltd All rights reserved

Chemistry & Biology
FR-900098 Biosynthesis
Figure 2. FrbH Reaction
(A) Proposed order of reaction where CMP attachment occurs before decarboxylation.
(B) HPLC analysis of the FrbH reaction product.
(C) MS/MS data of CMP-50-3APn.
(D) HPLC analysis of FrbH-K466A mutant, which should remove decarboxylase activity.
(E) HPLC analysis of FrbH-K38A mutant, which should remove nucleotide transferase activity.
attempts to demonstrate this in vitro reaction failed. A BLAST characteristic fragment ion of m/z 322, indicating the CMP
search of full-length FrbH revealed two distinct domains: a nucle- moiety on both of these compounds. These results suggest
otide transferase domain, followed by a PLP-dependent amino- FrbH condenses 2APn and CTP to form CMP-50-2APn and
transferase/decarboxylase domain. Only the second domain then decarboxylates CMP-50-2APn to yield CMP-50-3APn.
was originally believed to be active, but based on the whole-
Single-point mutations were designed to eliminate the activity
cell feeding studies with 2APn, we began to investigate whether of each of the individual domains of FrbH to verify the proposed
FrbH is bifunctional (Figure 2A). An in vitro reaction mixture con- order of reaction. PLP-dependent enzymes are known to bind to
taining both 2APn and cytidine-50-triphosphate (CTP) showed the cofactor with a conserved lysine residue. A sequence align-
nearly complete conversion of CTP to several compounds, ment with CobD from Salmonella typhimurium revealed the
including CMP, cytidine-50-diphosphate (CDP), CMP-50-2APn, PLP-binding motif and the lysine at position 466 (Brushaber
and predominantly CMP-50-3APn (Figure 2B). LC-MS of the et al., 1998). A K466A mutant was constructed, expressed,
products CMP-50-2APn and CMP-50-3APn revealed the masses and purified which should be unable to bind the PLP and thus
m/z 487 and 443, respectively (Figure 2C). Furthermore, frag- perform the decarboxylation. A second mutant was designed
mentation of these two products in negative mode showed the to disrupt nucleotide transferase activity of FrbH. A sequence
Chemistry & Biology 17, 57–64, January 29, 2010 ª2010 Elsevier Ltd All rights reserved 59

Chemistry & Biology
FR-900098 Biosynthesis
alignment with IspD (an enzyme in the nonmevalonate pathway Removal of the CMP moiety from CMP-50-FR-900098 is neces-
for isoprenoid biosynthesis) from E. coli revealed a common sary in order to produce the final compound FR-900098. Nucleo-
lysine residue (K38 in FrbH) that eliminated enzymatic activity tide hydrolases are known to catalyze the hydrolytic breakdown
(Richard et al., 2004). Under the same in vitro reaction conditions of a wide variety of nucleotide derivatives (McLennan, 2006).
as the wild-type FrbH, the K466A mutant produced only CMP-50- FrbI has high homology to this family of enzymes and could likely
2APn (Figure 2D). Although unable to totally abolish nucleotide catalyze the hydrolysis of CMP-50-FR-900098 to produce
transferase activity, the K38A mutant had a mostly unreacted FR-900098 and CMP. In order to perform in vitro enzymatic
CTP peak with a small peak corresponding to CMP-50-3APn studies, the enzyme FrbI was overexpressed in E. coli and puri-
(Figure 2E). These results are conclusive evidence for FrbH cata- fied to homogeneity. An in vitro reaction involving FrbI was tested
lyzing the CMP attachment to 2APn followed by decarboxyl- and found to produce a significant amount of FR-900098,
ation, yielding CMP-50-3APn (Figure 2A).
whereas only a very small amount of FR-900098 was present
in the absence of FrbI, likely due to the instability of CMP-50-
FR-900098 (Figures 3C and 3D). Thus, it is clear that FrbI cata-
lyzes the hydrolysis of CMP-50-FR-900098 to form FR-900098
N-Acetyltransferase Activity of FrbF and N-Hydroxylase
Activity of FrbG
To complete the biosynthesis of FR-900098, the amino group of and CMP. Interestingly, FrbI also catalyzed the hydrolysis of all
CMP-50-3APn requires both hydroxylation and acetylation. The the other CMP-containing intermediates within the pathway
order and identity of the enzymes catalyzing these two steps and even showed a small amount of activity toward CTP. This
were never definitively assigned in the original pathway (Eliot is not surprising given that nucleotide hydrolases typically have
et al., 2008). FrbF has strong homology to known N-acetyltrans- broad substrate specificity (Bessman et al., 1996). This promis-
ferases, and it was originally believed that FrbJ may be involved cuous behavior is evident from the gene deletion analysis previ-
in the hydroxylation step. However, based on the results of the ously mentioned. The gene frbI is not required for FR-900098
whole-cell feeding assays and deletion studies described above, biosynthesis in E. coli as a ubiquitous nucleotide hydrolase is
FrbG, a FAD/NADPH-dependent monooxygenase, was now able to cleave the CMP, yielding the final compound FR-900098.
thought to catalyze the hydroxylation step instead of FrbJ. The
whole-cell feeding studies also suggest the order of the two Monooxygenase Activity of FrbJ
steps as acetylation followed by hydroxylation, because CMP- FrbJ is homologous to a-ketoglutarate-dependent monooxyge-
50-Ac3APn was seen and not N-hydroxy-3-aminopropylphosph- nases capable of functionalizing an inactivated C-H bond.
onate (CMP-50-H3APn). Coupled in vitro reactions with FrbF and Although not required for FR-900098 biosynthesis, we believed
FrbG also supported this order of reaction, as there was always the enzyme may be involved in the synthesis of another phos-
a CMP-50-Ac3APn peak present but little to no CMP-50-H3APn phonate antibiotic, FR-33289. Production of this compound
monitored (Figures 3A and 3B). To determine the order of the was observed in both the native S. rubellomurinus and heterolo-
acetylation and hydroxylation steps, each of the potential gous S. lividans 4G7 host (data not shown). FR-33289 has similar
substrates was purified and kinetic parameters for FrbF and antibacterial properties compared to FR-900098 and could also
FrbG were obtained.
be another effective antimalarial compound (Okuhara et al.,
Although CMP-50-Ac3APn was believed to be an intermediate 1980b). An in vitro reaction mixture containing active FrbJ was
in the pathway instead of CMP-50-H3APn, no activity was able to convert FR-900098 to FR-33289 (Figures 3E and 3F).
observed for FrbG toward this substrate. FrbF was able to Expression of frbJ along with the remaining genes in the pathway
accept either of the two potential substrates, meaning CMP-50- could provide a means for production of FR-33289.
Ac3APn is actually a dead-end in the FR-900098 biosynthetic
In summary, we have heterologously produced FR-900098 in
pathway rather than a true intermediate (Table 1). Kinetic E. coli and demonstrated that the late steps of FR-900098
analysis of these two enzymes indicated that FrbF and FrbG biosynthesis follow a series of novel intermediates (Figure 4).
compete for the same intermediate, CMP-50-3APn, as they Of special note, the nucleotide attachment to the intermediates
both prefer this compound compared to the alternative (CMP- in FR-900098 biosynthesis seems to play an unprecedented
H3APn for FrbF and CMP-50-Ac3APn for FrbG). However, FrbG role. It is traditionally believed that attachment of the nucleotide
has 2.3-fold lower KM for CMP-50-3APn in comparison to FrbF, serves only as an activator for otherwise challenging chemistry in
so the enzyme can effectively direct the flux away from the natural product biosyntheses containing nucleotide-conjugated
dead end. The reason for observing CMP-50-Ac3APn rather intermediates. For example, in the biosynthesis of phosphino-
than CMP-50-H3APn throughout the feeding experiments is thricin tripeptide, PhpF catalyzes the condensation of CTP
due to the buildup of the dead-end intermediate CMP-50- with phosphonoformate to form CMP-50-phosphonoformate
Ac3APn, whereas the true intermediate, CMP-50-H3APn, is pre- (CMP-50-PF). This reaction was required so that the activated
vented from accumulating because it is further processed by intermediate would be primed for transfer of the phosphonofor-
FrbF. These results demonstrate that the flux of the pathway mate group to the appropriate acceptor molecule (Blodgett
must go through N-hydroxylation by FrbG followed by N-acety- et al., 2007). As part of the nonmevalonate pathway, IspD has
lation by FrbF and subsequently the more electrophilic CMP-50- been shown to catalyze the formation of 4-diphosphocytidyl-2-
H3APn intermediate.
C-methyl-D-erythritol (CDP-ME) from 2-C-methyl-D-erythritol
4-phosphate and CTP, which facilitates the subsequent forma-
tion of 2-C-methyl-D-erythritol 2,4-cyclodiphosphate (Rohdich
Nucleotide Hydrolase Activity of FrbI
The last step in the biosynthesis of FR-900098 was evident based et al., 1999). However, in FR-900098 biosynthesis, nucleotide
on the presence of the intermediate CMP-50-FR-900098. attachment clearly does not activate the bond for subsequent
60 Chemistry & Biology 17, 57–64, January 29, 2010 ª2010 Elsevier Ltd All rights reserved

Chemistry & Biology
FR-900098 Biosynthesis
Figure 3. FrbG, FrbF, FrbI, and FrbJ Reactions
(A) HPLC analysis of the FrbG and FrbF combined reaction product.
(B) MS/MS data of CMP-50-FR-900098.
(C) HPLC analysis of the FrbI reaction product.
(D) MS/MS data of FR-900098.
(E) HPLC analysis of the FrbJ reaction product.
(F) MS/MS data of FR-33289.
chemical reactions. Conjugation to CMP likely functions as a enzymatic activity. This hypothesis was substantiated by our
recognition mechanism to help distinguish among other metab- recent determination of the cocrystal structure of FrbG with the
olites within the cell and, more importantly, as a requirement for cofactor NADPH (data not shown). Examination of this structure
Chemistry & Biology 17, 57–64, January 29, 2010 ª2010 Elsevier Ltd All rights reserved 61

Chemistry & Biology
FR-900098 Biosynthesis
to play an unprecedented role in natural product biosyn-
thesis. Instead of activating a chemical bond, the nucleotide
conjugation seems to serve as a substrate recognition
mechanism. In addition, a biosynthetic route to a second
phosphonic acid antibiotic, FR-33289, was discovered,
which is embedded in the biosynthetic route to FR-900098,
and its production is regulated by a single enzyme, FrbJ.
This compound could also be effective in the treatment of
malaria. These studies now open possibilities for metabolic
engineering in E. coli to increase production of the antima-
larial antibiotic and combinatorial biosynthesis to generate
novel derivatives of FR-900098 with more potent antimalarial
activity.
Table 1. Kinetic Parameters for FrbF and FrbG Reactions with
Each of the Potential Substrates
kcat (minꢂ1
(Mean SD) (Mean SD) (mMꢂ1 minꢂ1
)
KM (mM)
kcat/KM
Enzyme Substrate
)
FrbF
CMP-50-3APn
3.46 0.16 35.2 3.2
98.0
39.0
CMP-50-H3APn 2.04 0.43 52.3 3.0
FrbG
CMP-50-3APn
5.92 0.38 15.1 1.5
391
ND
CMP-50-Ac3APn
ND, no detectable activity; SD, standard deviation.
reveals that there is no room for the binding of the substrate in
the active site. Thus, we proposed a novel reaction mechanism
for the N-hydroxylation catalyzed by FrbG. Briefly, the substrate
and NADPH cofactor compete for the same binding pocket in the
enzyme, and the nucleotide helps dock the substrate in the
correct position. The remaining enzymes in the pathway likely
evolved to accept substrates conjugated to CMP rather than
those proposed in the original pathway.
EXPERIMENTAL PROCEDURES
Heterologous Production of FR-900098 in E. coli
An overnight culture of the production strain (pETDuet-frbABCD, pRSFDuet-
frbHFG, pACYCDuet-frbE-dxrB in E. coli BL21[DE3] cells) was grown
at 37ꢀC in Luria-Bertani (LB) medium broth using the following antibiotic
concentrations: 50 mg/ml ampicillin, 25 mg/ml kanamycin, and 12.5 mg/ml
chloramphenicol. The cultures were diluted 1:100 in fresh medium and grown
at 37ꢀC to an OD600 of ꢁ0.7. At that point, the expression of the recombinant
genes was induced with 0.3 mM IPTG and the culture was incubated at 30ꢀC
with shaking (250 rpm). Samples (1 ml) were then removed from the cultures at
specific time points and stored at ꢂ80ꢀC. Frozen samples were thawed at
room temperature and centrifuged at maximum speed for 2 min. Supernatants
from the samples were then analyzed by LC-MS to determine the concentra-
tion of FR-900098.
SIGNIFICANCE
We have successfully produced a potent antimalarial agent,
FR-900098, in a recombinant E. coli strain and elucidated
the biosynthetic mechanisms of the late steps involved
in FR-900098 biosynthesis. Notably, a nucleotide CMP is
conjugated to several of the intermediates, which seems
Figure 4. Revised FR-900098 Pathway
Each of the final steps in the FR-900098 biosynthetic pathway has been demonstrated in vitro using purified enzymes. Intermediates in the FR-900098
biosynthetic pathway are as follows: I, phosphoenolpyruvate; II, phosphonopyruvate; III, 2-phosphonomethylmalate; IV, 3-phosphonomethylmalate; V,
2-oxo-4-phosphonobutyrate; VI, 2-amino-4-phosphonobutyrate; VII, CMP-50-3-aminopropylphosphonate; VIII, CMP-50-N-hydroxy-3-aminopropylphospho-
nate; IX, CMP-50-FR-900098.
62 Chemistry & Biology 17, 57–64, January 29, 2010 ª2010 Elsevier Ltd All rights reserved

Chemistry & Biology
FR-900098 Biosynthesis
Whole-Cell Feeding Studies
FrbJ Activity
E. coli BL21(DE3) cells expressing empty vector (control) and FrbH and cells
overexpressing FrbH, FrbF, FrbG, and DxrB were cultured overnight at 37ꢀC
in Terrific Broth medium with appropriate antibiotics. The cultures were diluted
1:100 in fresh medium and grown at 37ꢀC to an OD600 of ꢁ0.7. Protein expres-
sion was induced by the addition of 0.3 mM IPTG final concentration and
cultures were grown at 25ꢀC for 24 hr. Cultures were then centrifuged at
3200 xg for 15 min and the supernatant was discarded. The cell pellet was re-
suspended in 1 ml of 10 mM Tris-HCl buffer (pH 7.0). Resuspended cells were
mixed with 20 mM final concentration ( )-2-amino-4-phosphonobutyrate
(2APn) and incubated at 30ꢀC for 24 hr. Samples were lysed by freeze/thaw
and centrifuged at maximum speed in a microcentrifuge for 10 min. The super-
natant (1 ml) was mixed with 9 ml of methanol, vortexed, and centrifuged at
3200 xg for 15 min. The pellet was discarded and the supernatant was rotary
evaporated down to a volume of 1 ml. Samples were placed in a ꢂ80ꢀC freezer
until analyzed by LC-MS using a 7 Tesla LTQ-FT mass spectrometer (Thermo
Finnigan, San Jose, CA, USA).
FrbJ activity was assayed by HPLC and LC-MS. The assay mixture (100 ml)
contained 1 mM FR-900098, 1 mM a-ketoglutarate, 1 mM FeSO4, 4 mM ascor-
bic acid, and His6-FrbJ in 50 mM HEPES-Na+ buffer (pH 7.25). The reaction
was initiated with the addition of 25 mg of purified FrbJ and incubated at
25ꢀC for 4 hr. Reaction mixtures were transferred into Microcon centrifugal
filter devices (MWCO 10,000) and centrifuged for 15 min at maximum speed.
The reactions were then analyzed by LC-MS.
LC-MS Analysis of CMP-Containing Compounds
Products of the enzymatic reactions were analyzed by an Agilent 1100 series
LC/MSD XCT plus-ion-trap mass spectrometer (Agilent, Palo Alto, CA, USA) at
the Roy J. Carver Metabolomics Center at the University of Illinois (Urbana, IL,
USA). The liquid chromatography was performed using an Agilent 1100 series
HPLC equipped with an online degasser, binary pump, and autosampler.
Analytes were resolved on a 150 3 4.6 mm Agilent C18 reverse-phase HPLC
column. HPLC parameters were as follows: 25ꢀC; solvent A: 15 mM ammo-
nium formate; solvent B: 20 mM ammonium acetate; gradient: 25% B at
0 min, then to 100% B in 10 min, and finally maintain 100% B for 3 min; flow
rate: 0.3 ml/min; detection by UV absorbance at 254 nm. The mass spectrom-
etry was performed using an Agilent XCT ion-trap MSD mass spectrometer.
Mass spectra were acquired in ultrascan mode using electrospray ionization
(ESI) with negative polarity. The MS system was operated using a drying
temperature of 350ꢀC, a nebulizer pressure of 35 psi, a drying gas flow of
8.5 L/min, and a capillary voltage of 4500 V.
Cloning, Expression, and Purification
The cloning of genes frbH, frbF, frbG, frbI, and frbJ from fosmid 4G7 was
similar to that in Eliot et al. (2008). Briefly, the genes were PCR amplified
with NdeI and HindIII sites and cloned into corresponding sites of pET28a,
designed to create an N-terminal histidine tag and transformed BL21(DE3)
cells. The cells were grown in LB medium supplemented with kanamycin
(50 mg/ml) at 37ꢀC to OD600 ꢁ0.7 and then induced at 25ꢀC by the addition
of IPTG to a final concentration of 0.3 mM. During the purification of FrbH,
PLP was added to the media for a final concentration of 50 mg/ml. After
6–8 hr, the cells were harvested by centrifugation and resuspended in
20 mM Tris-HCl (pH 7.65), 15% glycerol, and 0.5 M NaCl at 4ꢀC. The
sample was freeze-thawed and then lysed further by sonication. After cellular
debris was removed by centrifugation, the N-His6-tagged proteins were
purified using immobilized metal-affinity chromatography with TALON Super-
flow Co2+ resin coupled to fast-performance liquid chromatography. The
purified protein was washed three times with 50 mM HEPES-Na+ buffer
and concentrated a final time. Concentrated protein was stored in 20% glyc-
erol at ꢂ80ꢀC.
LC-MS Analysis of FR-900098 and FR-33289
LC-MS was performed using an Agilent 1100 series LC/MSD XCT plus-ion-
trap mass spectrometer. The liquid chromatography was performed using
an Agilent 1100 series HPLC equipped with an online degasser, binary
pump, and autosampler. The mobile phase contained 0.1% formic acid in
ddH2O (solvent A) and 100% acetonitrile (solvent B). Analytes were resolved
on a 100 3 4.6 mm Synergi 4u Fusion-RP 80A analytical column (Phenomenex,
Torrance, CA, USA) at a flow rate of 200 ml/min with a gradient of 0%–40%
solvent B for 10 min, a gradient of 40%–100% solvent B for 3 min, and
100% solvent B for 5 min. The mass spectrometry was performed using an
Agilent XCT ion-trap MSD mass spectrometer. Mass spectra were acquired
in ultrascan mode using ESI with positive polarity. The system was operated
using a drying temperature of 350ꢀC, a nebulizer pressure of 35 psi, a drying
gas flow of 8.5 L/min, and a capillary voltage of 4500 V. For determination of
titers, extracted ion chromatograms were generated from suitable ions of
each analyte, and the integrated areas corresponding to each compound
were compared to standards. Fosmidomycin was used as an internal standard
to measure FR-900098 concentrations.
FrbH Activity
Wild-type and mutant FrbH activity was assayed by HPLC and LC-MS. The
assay mixture (100 ml) contained 2 mM ( )-2-amino-4-phosphonobutyric
acid, 1 mM CTP, 10 mM MgCl2, and His6-FrbH in 50 mM HEPES-Na+ buffer
at pH 7.25. The reaction was initiated with the addition of 25 mg of purified
wild-type or mutant FrbH and incubated at 30ꢀC for 2 hr. Reaction mixtures
were transferred into Microcon centrifugal filter devices (MWCO 10,000) and
centrifuged for 15 min at maximum speed. The reactions were then analyzed
by LC-MS.
ACKNOWLEDGMENTS
We thank Lucas Li at the Metabolomics Center at the University of Illinois for
help in LC-MS analysis. This work was supported by grants from the National
Institute of General Medical Sciences GM077596 (H.Z., W.W.M., and N.L.K.)
and GM59334 (W.W.M.). M.A.D. acknowledges support from the U.S. National
Institutes of Health under Ruth L. Kirschstein national research award 5 T32
GM070421 from the National Institute of General Medical Sciences.
FrbG and FrbF Activity
The activity of FrbG and FrbF was demonstrated in a coupled assay with FrbH.
The assay mixture (100 ml) contained 2 mM ( )-2-amino-4-phosphonobutyric
acid, 1 mM CTP, 10 mM MgCl2, 1 mM AcCoA, 1 mM NADPH, His6-FrbH,
His6-FrbF, and His6-FrbG in 50 mM HEPES-Na+ buffer (pH 7.25). The reaction
was initiated with the addition of 25 mg of purified FrbF and 35 mg of purified
FrbG. Reaction mixtures were transferred into Microcon centrifugal filter
devices (MWCO 10,000) and centrifuged for 15 min at maximum speed. The
reactions were then analyzed by LC-MS.
Received: November 5, 2009
Revised: December 1, 2009
Accepted: December 9, 2009
Published: January 28, 2010
FrbI Activity
REFERENCES
The activity of FrbI was demonstrated in a coupled assay with FrbH, FrbF,
and FrbG. The assay mixture (100 ml) contained 2 mM ( )-2-amino-4-phospho-
nobutyric acid, 1 mM CTP, 10 mM MgCl2, 1 mM AcCoA, 1 mM NADPH,
His6-FrbH, His6-FrbF, His6-FrbG, and His6-FrbI in 50 mM HEPES-Na+ buffer
(pH 7.25). The reaction was initiated with the addition of 25 mg of purified
FrbI and incubated at 30ꢀC for 2 hr. Reaction mixtures were transferred
into Microcon centrifugal filter devices (MWCO 10,000) and centrifuged for
15 min at maximum speed. The reactions were then analyzed by LC-MS.
Barry, R.J., Bowman, E., McQueney, M., and Dunaway-Mariano, D. (1988).
Elucidation of the 2-aminoethylphosphonate biosynthetic pathway in Tetrahy-
mena pyriformis. Biochem. Biophys. Res. Commun. 153, 177–182.
Bessman, M.J., Frick, D.N., and O’Handley, S.F. (1996). The MutT proteins or
‘‘Nudix’’ hydrolases, a family of versatile, widely distributed, ‘‘housecleaning’’
enzymes. J. Biol. Chem. 271, 25059–25062.
Chemistry & Biology 17, 57–64, January 29, 2010 ª2010 Elsevier Ltd All rights reserved 63

Chemistry & Biology
FR-900098 Biosynthesis
Beytia, E.D., and Porter, J.W. (1976). Biochemistry of polyisoprenoid biosyn-
(1999). Inhibitors of the nonmevalonate pathway of isoprenoid biosynthesis
thesis. Annu. Rev. Biochem. 45, 113–142.
as antimalarial drugs. Science 285, 1573–1576.
Lange, B.M., Rujan, T., Martin, W., and Croteau, R. (2000). Isoprenoid biosyn-
thesis: the evolution of two ancient and distinct pathways across genomes.
Proc. Natl. Acad. Sci. USA 97, 13172–13177.
Blodgett, J.A., Zhang, J.K., and Metcalf, W.W. (2005). Molecular cloning,
sequence analysis, and heterologous expression of the phosphinothricin
tripeptide biosynthetic gene cluster from Streptomyces viridochromogenes
DSM 40736. Antimicrob. Agents Chemother. 49, 230–240.
MacSweeney, A., Lange, R., Fernandes, R.P., Schulz, H., Dale, G.E.,
Douangamath, A., Proteau, P.J., and Oefner, C. (2005). The crystal structure
of E. coli 1-deoxy-D-xylulose-5-phosphate reductoisomerase in a ternary
complex with the antimalarial compound fosmidomycin and NADPH reveals
a tight-binding closed enzyme conformation. J. Mol. Biol. 345, 115–127.
Blodgett, J.A., Thomas, P.M., Li, G., Velasquez, J.E., van der Donk, W.A.,
Kelleher, N.L., and Metcalf, W.W. (2007). Unusual transformations in the
biosynthesis of the antibiotic phosphinothricin tripeptide. Nat. Chem. Biol. 3,
480–485.
McLennan, A.G. (2006). The Nudix hydrolase superfamily. Cell. Mol. Life Sci.
Borrmann, S., Adegnika, A.A., Matsiegui, P.B., Issifou, S., Schindler, A.,
Mawili-Mboumba, D.P., Baranek, T., Wiesner, J., Jomaa, H., and Kremsner,
P.G. (2004). Fosmidomycin-clindamycin for Plasmodium falciparum infections
in African children. J. Infect. Dis. 189, 901–908.
63, 123–143.
Okuhara, M., Kuroda, Y., Goto, T., Okamoto, M., Terano, H., Kohsaka, M.,
Aoki, H., and Imanaka, H. (1980a). Studies on new phosphonic acid antibiotics.
I. FR-900098, isolation and characterization. J. Antibiot. (Tokyo) 33, 13–17.
Borrmann, S., Lundgren, I., Oyakhirome, S., Impouma, B., Matsiegui, P.B.,
Adegnika, A.A., Issifou, S., Kun, J.F., Hutchinson, D., Wiesner, J., et al.
(2006). Fosmidomycin plus clindamycin for treatment of pediatric patients
aged 1 to 14 years with Plasmodium falciparum malaria. Antimicrob. Agents
Chemother. 50, 2713–2718.
Okuhara, M., Kuroda, Y., Goto, T., Okamoto, M., Terano, H., Kohsaka, M.,
Aoki, H., and Imanaka, H. (1980b). Studies on new phosphonic acid
antibiotics. III. Isolation and characterization of FR-31564, FR-32863 and
FR-33289. J. Antibiot. (Tokyo) 33, 24–28.
Boucher, Y., and Doolittle, W.F. (2000). The role of lateral gene transfer in the
Richard, S.B., Lillo, A.M., Tetzlaff, C.N., Bowman, M.E., Noel, J.P., and Cane,
D.E. (2004). Kinetic analysis of Escherichia coli 2-C-methyl-D-erythritol-4-
phosphate cytidyltransferase, wild type and mutants, reveals roles of active
site amino acids. Biochemistry 43, 12189–12197.
evolution of isoprenoid biosynthesis pathways. Mol. Microbiol. 37, 703–716.
Brushaber, K.R., O’Toole, G.A., and Escalante-Semerena, J.C. (1998). CobD,
a novel enzyme with L-threonine-O-3-phosphate decarboxylase activity, is
responsible for the synthesis of (R)-1-amino-2-propanol O-2-phosphate,
a proposed new intermediate in cobalamin biosynthesis in Salmonella typhi-
murium LT2. J. Biol. Chem. 273, 2684–2691.
Ridley, R.G. (2002). Medical need, scientific opportunity and the drive for anti-
malarial drugs. Nature 415, 686–693.
Rohdich, F., Wungsintaweekul, J., Fellermeier, M., Sagner, S., Herz, S., Kis, K.,
Eisenreich, W., Bacher, A., and Zenk, M.H. (1999). Cytidine 50-triphosphate-
dependent biosynthesis of isoprenoids: YgbP protein of Escherichia coli
catalyzes the formation of 4-diphosphocytidyl-2-C-methylerythritol. Proc.
Natl. Acad. Sci. USA 96, 11758–11763.
Eliot, A.C., Griffin, B.M., Thomas, P.M., Johannes, T.W., Kelleher, N.L., Zhao,
H., and Metcalf, W.W. (2008). Cloning, expression, and biochemical character-
ization of Streptomyces rubellomurinus genes required for biosynthesis of
antimalarial compound FR900098. Chem. Biol. 15, 765–770.
Fokin, A.A., Yurchenko, A.G., Rodionov, V.N., Gunchenko, P.A., Yurchenko,
R.I., Reichenberg, A., Wiesner, J., Hintz, M., Jomaa, H., and Schreiner, P.R.
(2007). Synthesis of the antimalarial drug FR900098 utilizing the nitroso-ene
reaction. Org. Lett. 9, 4379–4382.
Schwartz, D., Berger, S., Heinzelmann, E., Muschko, K., Welzel, K., and
Wohlleben, W. (2004). Biosynthetic gene cluster of the herbicide phosphino-
thricin tripeptide from Streptomyces viridochromogenes Tu494. Appl. Environ.
Microbiol. 70, 7093–7102.
Hemmi, K., Takeno, H., Hashimoto, M., and Kamiya, T. (1982). Studies on
phosphonic acid antibiotics. IV. Synthesis and antibacterial activity of analogs
of 3-(N-acetyl-N-hydroxyamino)-propylphosphonic acid (FR-900098). Chem.
Pharm. Bull. (Tokyo) 30, 111–118.
Snow, R.W., Guerra, C.A., Noor, A.M., Myint, H.Y., and Hay, S.I. (2005). The
global distribution of clinical episodes of Plasmodium falciparum malaria.
Nature 434, 214–217.
Wiesner, J., Ortmann, R., Jomaa, H., and Schlitzer, M. (2003). New antimalarial
Hidaka, T., Goda, M., Kuzuyama, T., Takei, N., Hidaka, M., and Seto, H. (1995).
Cloning and nucleotide sequence of fosfomycin biosynthetic genes of Strep-
tomyces wedmorensis. Mol. Gen. Genet. 249, 274–280.
drugs. Angew. Chem. Int. Ed. Engl. 42, 5274–5293.
Woodyer, R.D., Shao, Z., Thomas, P.M., Kelleher, N.L., Blodgett, J.A., Metcalf,
W.W., van der Donk, W.A., and Zhao, H. (2006). Heterologous production of
fosfomycin and identification of the minimal biosynthetic gene cluster.
Chem. Biol. 13, 1171–1182.
Jomaa, H., Wiesner, J., Sanderbrand, S., Altincicek, B., Weidemeyer, C.,
Hintz, M., Turbachova, I., Eberl, M., Zeidler, J., Lichtenthaler, H.K., et al.
64 Chemistry & Biology 17, 57–64, January 29, 2010 ª2010 Elsevier Ltd All rights reserved