Different Biosynthetic Pathways to Fosfomycin in Pseudomonas
syringae and Streptomyces Species
Seung Young Kim,aKou-San Ju,aWilliam W. Metcalf,aBradley S. Evans,aTomohisa Kuzuyama,band Wilfred A. van der Donka,c
Institute of Genomic Biology, University of Illinois at Urbana—Champaign, Urbana, Illinois, USAa; Biotechnology Research Center, University of Tokyo, Tokyo, Japanb; and
Howard Hughes Medical Institute and Department of Chemistry, University of Illinois at Urbana—Champaign, Urbana, Illinois, USAc
mycin production in Pseudomonas syringae PB-5123. Surprisingly, the biosynthetic pathway in this organism is very different
from that in Streptomyces fradiae and Streptomyces wedmorensis. The pathways share the first and last steps, involving conver-
tively, but the enzymes converting PnPy to 2-HPP are different. The genome of P. syringae PB-5123 lacks a gene encoding the
PnPydecarboxylasefoundinthe Streptomyces strains.Instead,itcontainsagenecodingforacitratesynthase-likeenzyme,Psf2,
ologous production of fosfomycin in Pseudomonas aeruginosa from a fosmid encoding the fosfomycin biosynthetic cluster from
P. syringae PB-5123 confirmed that the gene cluster is functional. Therefore, two different pathways have evolved to produce
pound blocks peptidoglycan biosynthesis by inhibiting UDP-
GlcNAc enolpyruvyl transferase (MurA) (29, 38). In the clinic,
times gastrointestinal infections (1, 2, 15, 43, 45, 56). Previous
synthetic pathway shown in Fig. 1 in Streptomyces fradiae and
Streptomyces wedmorensis (37, 50, 59, 60). Like in nearly all phos-
phonate biosynthetic pathways (39), the first committed step in
fosfomycin biosynthesis in these streptomycetes is the conversion
of phosphoenolpyruvate (PEP) to phosphonopyruvate (PnPy)
PnPy to phosphonoacetaldehyde (PnAA) by a thiamine-depen-
dent decarboxylase (41). Subsequent reduction to 2-hydroxyeth-
ylphosphonate (2-HEP) (51, 59), methylation to generate 2-hy-
droxypropylphosphonate (2-HPP) (20, 34, 59, 60), and epoxide
formation (19, 25, 37, 49, 62) complete the biosynthesis of fosfo-
mycin (Fig. 1). All enzymatic steps in this pathway have been
putatively catalyzed by the radical S-adenosylmethionine (SAM)
methyltransferase Fom3 (58, 59, 61). Attempts to achieve in vitro
activity with purified Fom3 enzymes from streptomycetes have
the reaction that Fom3 is proposed to catalyze, experimental ver-
ification of its proposed role would be valuable.
Fosfomycin and various analogs are also produced by several
nas fluorescens PK-52, and Pseudomonas syringae PB-5123 (30,
from these Gram-negative bacteria, the fosfomycin biosynthetic
gene cluster from Pseudomonas syringae PB-5123 was identified.
ortholog, nor does it contain a gene encoding a PnPy decarboxyl-
osfomycin is an FDA-approved antibiotic containing epoxide
and phosphonate functional groups (Fig. 1) (8). The com-
sis in P. syringae are the same as those reported for previously
investigated pathways in streptomycetes but that an entirely dif-
ferent set of enzymes is used to convert PnPy to 2-HPP.
MATERIALS AND METHODS
Materials. Chemical reagents used in this study were purchased from
Sigma-Aldrich (St. Louis, MO) or Thermo Fisher Scientific (Pittsburgh,
PA) and were used without further purification. Medium components
were purchased from Thermo Fisher Scientific or VWR (West Chester,
PA). Bacterial strains, plasmids, and the sequences of PCR primers are
Tryptone-yeast extract medium (TY; 1% tryptone, 0.5% yeast extract)
and minimal-salts broth (MSB) (55) containing 1% (vol/vol) Balch’s vi-
was solidified with 1.8% (wt/vol) Noble agar, and LB and TY were solid-
ified with 1.6% (wt/vol) Bacto agar. Antibiotics were added at the follow-
ing concentrations for plasmid selection and maintenance; kanamycin
(Kan), 50 ?g ml?1; apramycin (Apr), 25 ?g ml?1; chloramphenicol
(Cm), 15 ?g ml?1; and gentamicin, 5 ?g ml?1.
DNA isolation and manipulation. All cloning was performed by es-
tablished methods (46). Endonucleases and T4 DNA ligase were pur-
chased from Invitrogen (Carlsbad, CA) and New England BioLabs
(Ipswich, MA). Shrimp alkaline phosphatase was purchased from Roche
Received 2 January 2012 Returned for modification 30 January 2012
Accepted 9 May 2012
Published ahead of print 21 May 2012
Address correspondence to Wilfred A. van der Donk, email@example.com.
This paper is dedicated to the memory of Professor Haruo Seto, honoring his
pioneering contributions to the natural-product field.
Supplemental material for this article may be found at http://aac.asm.org/.
Copyright © 2012, American Society for Microbiology. All Rights Reserved.
August 2012 Volume 56 Number 8Antimicrobial Agents and Chemotherapyp. 4175–4183 aac.asm.org
Diagnostics GmbH (Mannheim, Germany). Oligonucleotides were ob-
tained from Integrated DNA Technologies (Coralville, IA). Plasmids and
ase (Novagen, EMD Chemicals Inc., Gibbstown, NJ). Plasmids used for
functionalization of Pseudomonas aeruginosa were purified with a Fer-
mentas GeneJet miniprep kit (Glen Burnie, MD), and DNA fragments
were purified with a Fermentas GeneJet gel extraction kit. An UltraClean
microbial DNA isolation kit (Mo Bio, Carlsbad, CA) was used for the
purification of genomic DNA for genome sequencing. DNA sequencing
was carried out at the Roy J. Carver Biotechnology Center at the Univer-
sity of Illinois, Urbana—Champaign, on an ABI 3730XL capillary se-
E. coli S17-1 ?-pir was used to introduce plasmids into Pseudomonas
for mobilization were cross-streaked with Pseudomonas strains and incu-
bated at 30°C for 16 h. Cells were then resuspended in 10 ml of MSB,
homogenized by vortexing, and plated onto MSB plates containing 10
purified by repeated single-colony isolations.
syringae PB-5123 was sequenced on a Roche 454 GS-FLX system at the
Champaign. The sequence reads were assembled using the Newbler pro-
gram (454 Life Sciences).
Construction of a library of genomic DNA of P. syringae PB-5123.
homogenized, and genomic DNA was isolated using a DNeasy blood and
by FIGE. The fosmid vector pJK050 was prepared by sequential NheI
tion. The genomic DNA fraction was ligated with digested pJK050 over-
night at 16°C, followed by ethanol precipitation and packaging into
lambda phage using a MaxPlax packaging extract according to the man-
ufacturer’s instructions (Epicentre, Madison, WI). E. coli WM4489 cells
were transfected with the packaged library and plated on LB–12-?g/ml
Cm agar plates. WM4489 is an E. coli DH10B derivative engineered to
provide copy-number control for pJK050 fosmids through regulation of
the trfA33 gene (encoding plasmid replication protein) by a rhamnose-
inducible promoter (14). Individual colonies (?1,300) were picked into
standard 96-well plates, with each well containing a single 3-mm glass
bead (to aid mixing) and 200 ?l of LB–12 ?g/ml Cm. The samples were
grown overnight with shaking at 37°C.
Library screening. Cultures (10 ?l) from 48 clones were pooled,
boiled in water (100 ?l), and used as the template for PCR mixtures
containing 500 nmol of primers FomScreenF and FomScreenR (Table 1)
and KOD Hot Start DNA polymerase in 1? PCR premix G, with melting
at 95°C for 30 s, annealing at 60°C for 30 s, and extension at 72°C for 60 s
for 30 cycles in total. Pools that yielded DNA bands corresponding to the
psf1 gene fragment were then individually screened by PCR to identify
DNA was isolated from single positive clones grown overnight in 5 ml
LB–12 ?g/ml Cm–0.2% rhamnose using a Qiagen miniprep kit. The pu-
rified fosmids were then individually recombined in vitro with similarly
purified pAE4 using a BP clonase kit according to the manufacturer’s
instructions (14). The reaction mixtures were used to transform E. coli
WM4489, and successfully recombined plasmids were identified by the
screened by PCR for the presence of several genes in the putative cluster
(psf1, psf2, psf3, psf5, psf6, and psf7) using the primers listed in Table 1.
One fosmid (46-2-11) containing a 22.5-kb insert was positive for the
presence of the psf1-3 and psf5-7 genes and was sequenced. E. coli trans-
formants containing 46-2-11 were used as donors for conjugal transfer of
the fosmid to P. aeruginosa PAO1-LAC after retrofitting and strain con-
struction (see below).
created to extend the range of target strains amenable for ?C31 integra-
?C31 attB site. Plasposon pTnMod-RCm (12) was first tailored to confer
ment containing the gentamicin acetyltransferase gene (aacC1) from
SacI-digested pTnMod-OGm was purified and ligated to the ?3.5-kb
fragment of similarly digested pTnMod-RCm to produce pKSJ248. The
orientation of aacC1 was verified by restriction digestion with BsrGI and
SpeI. A DNA fragment containing the ?C31 attB site was amplified from
Streptomyces coelicolor A3(2) by PCR using primers PhiC31attBF-2 (with
restriction sites XbaI, SpeI, and NotI; Table 1) and PhiC31attBR (with
restriction sites HindIII, PacI, and MfeI; Table 1). The 291-bp product
was purified, digested with XbaI and HindIII, and ligated with similarly
digested pUC18 to produce pKSJ256. After verification of the desired
construct by sequencing, the ?C31 attB site was excised from pKSJ256
using SpeI and PacI, and the resulting fragment was ligated with similarly
digested pKSJ248 to produce pKSJ259.
Heterologous expression of the fosfomycin biosynthetic gene clus-
ter. P. aeruginosa PAO1-LAC was functionalized with the ?C31 attB site
by introducing pKSJ259 through conjugative transfer from E. coli S17-1
?-pir. A gentamicin-resistant derivative containing the ?C31 attB site
(verified by PCR) was designated KSJ554. Fosmids 46-2-11 and pJK50-
?-pir. Single gentamicin-, chloramphenicol-, and apramycin-resistant
colonies were selected and designated strains KSJ570 (harboring the fos-
mid without the insert) and KSJ629 (harboring fosmid 46-2-11). The
presence of the intact fosfomycin gene cluster in KSJ629 was verified by
diagnostic PCRs using the primers in Table 1.
Heterologous production of fosfomycin detected by GC-MS. Vola-
tile, silylated derivatives of fosfomycin were prepared by the addition of
200 ?l N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA)–1% trimeth-
ylchlorosilane (TMCS) (Aldrich) (54) to a 2-ml amber glass sample vial
and then addition of 100 ?l sample (either fosfomycin standard, crude
spent medium of P. syringae PB-5123, crude spent medium of P. aerugi-
methanol. Derivatization was performed for 40 min at 70°C. Reaction
mixtures were then analyzed on an Agilent 6890N gas chromatograph-
the University of Illinois Urbana—Champaign. Sample introduction was
confirmed by experimental data with purified enzymes are shown with solid
arrows, whereas putative transformations are indicated with dashed arrows.
Kim et al.
aac.asm.orgAntimicrobial Agents and Chemotherapy
via split injection onto an HP-5 (5% phenyl methylpolysiloxane) column
(30 m, 0.25-mm inner diameter, 0.25-?m film thickness). The injector
temperature was 250°C. The initial column temperature was 40°C, and
the temperature was held at this temperature for 5 min after injection
at 230°C for the remainder of the 27-min program. Under these condi-
tions, derivatized fosfomycin was detected at 14 min. Peaks at this reten-
210 with a diagnostic fragmentation pattern.
NMR spectroscopy. All nuclear magnetic resonance (NMR) experi-
ments were performed at the Varian Oxford Center for Excellence in
NMR spectroscopy at the University of Illinois, Urbana—Champaign.
The presence of phosphonates was determined using1H-decoupled31P
20% D2O as a lock solvent. The31P NMR spectra were externally refer-
enced to an 85% phosphoric acid standard (0 ppm). Spectra were ac-
quired at room temperature on a Varian Unity Inova-600 spectrometer
equipped with a 5-mm Varian 600DB AutoX probe with ProTune acces-
sory for detection at the31P frequency.
Preparation of recombinant Psf1 and Psf2. The psf1 and psf2 genes
TABLE 1 Bacterial strains, plasmids, and sequences of oligonucleotide primers for PCR experiments used in this studya
Strain, plasmid, or
primer nameRelevant characteristics or sequenceReference or source
Cloning host; contains ?-pir thi
Host for plasmid mobilization; contains ?-pir thi
P. syringae PB-5123Fosfomycin producer53
PAO1 derivative; lacIq?delta(lacZ)M15?tetA?tetR?
Derivative of PAO1-LAC, contains ?C31 attB Gmr
Derivative of KSJ554, pJK50-AE4 integrated in PhiC31 attB GmrCmrAprr; contains 46-2-11
Derivative of KSJ554, fosmid 46-2-11 integrated in ?C31, attB GmrCmrAprr
pAE4Plasmid for retrofitting pJK050 derivatives for conjugal transfer and integration; ? attP, ?C31 int, ?C31
Plasmid for constructing fosmid libraries; ? attB Clmr
Cloning vector, Ampr
Expression vector, Kanr
Plasposon, pMB1 ori Gmr
Plasposon, R6K ori Cmr
Plasposon (pTnMod-RGm), R6K ori Gmr
pUC18 containing ?C31 attB Ampr
pKSJ248 containing ?C31 attB Gmr
Derivative of pJK050, recombined with pAE4, contains fosfomycin gene cluster from PB-5123; ClmrAprr
Derivative of pJK050, recombined with pAE4; AprrClmr
pET26b containing psf1, Kanr
pET26b containing psf2, Kanr
aAmpr, ampicillin resistant; Aprr, apramycin resistant; Cmr, chloramphenicol resistant; Kanr, kanamycin resistant. Restriction sites are underlined.
bItalics, bold, and underlining indicate the restriction sites for XbaI, SpeI, and NotI, respectively.
cItalics, bold, and underlining indicate the restriction sites for HindIII, PacI, MfeI, respectively.
Two Biosynthetic Pathways to Fosfomycin
August 2012 Volume 56 Number 8 aac.asm.org 4177
is underlined) and the reverse primer Psf2RP (Table 1; the BamHI site is
underlined), respectively. The PCR products were then cloned into a
pET26b vector to produce pPSF1-his and pPSF2-his, respectively (Table
1). The expression plasmids containing psf1 and psf2 were used to trans-
form E. coli BL21(DE3) cells. These strains were then grown in LB me-
sion was induced by the addition of isopropyl-?-D-thiogalactoside
(IPTG) to 0.1 mM. Cultures shaken at 20°C overnight were harvested by
centrifugation and stored at ?80°C.
For purification, cell pellets were thawed and resuspended in 45 ml of
lysis buffer (50 mM Tris-HCl, 300 mM NaCl, 10% glycerol, 20 mM imi-
dazole, pH 7.5). Lysozyme was added to a concentration of 1 mg/ml, and
the resulting suspension was incubated on ice for 30 min. Cells were dis-
rupted by two passes through a French press (20,000 lb/in2), and debris
was removed by centrifugation. The resulting supernatant was slowly ag-
itated with 5 ml (bed volume of resin) of Ni-nitrilotriacetic acid resin
(Qiagen, Valencia, CA) prewashed with lysis buffer at 4°C for 3 h. The
suspension was loaded onto a column, and the flowthrough fraction was
collected. The resin was washed with lysis buffer containing 20 mM imi-
was eluted with a buffer containing 250 mM imidazole. The desired frac-
tions, as detected by SDS-PAGE, were pooled and concentrated using an
protein sample was loaded onto a PD-10 desalting column (GE Health-
7.5, as per the column manufacturer’s instructions.
Phosphonomethylmalate formation by Psf2. The phosphonometh-
spectroscopy and Fourier transform mass spectrometry (FTMS). The as-
say mixture (1 ml) contained 50 mM HEPES-K?(pH 7.2), 5 mM MgCl2,
Psf2. The reaction mixture was incubated at 30°C for 1 h. Then, the pro-
teins were removed using a Microcon YM-10 filter unit, and the solution
was lyophilized. Samples were reconstituted in 90% acetonitrile contain-
linear trap quadrupole-FTMS with an attached Agilent 1200 high-pres-
arated on a Zic pHILIC column (2.1 by 150 mm; SeQuant, Umeå, Swe-
den) using 90% acetonitrile containing 10 mM ammonium bicarbonate
(solvent B) and 10 mM ammonium bicarbonate (solvent A). The mobile
phase was held at initial conditions (100% B) for 2 min, followed by a
linear gradient down to 40% B over 15 min, a 3-min linear gradient back
acquired in negative mode, scanning m/z 100 to 700 at a resolution of
the Qualbrowser application of Xcalibur software (Thermo Fisher Scien-
tific, San Jose, CA).
fosmid 46-2-11 has been deposited in GenBank under accession number
Fosfomycin gene cluster in P. syringae PB-5123. To identify the
genes encoding fosfomycin biosynthesis, genomic DNA was pre-
sequence reads (estimated coverage, 95-fold) yielded 124 contigs
totaling 5,828,174 bp. A gene (psf1; Fig. 2) that encodes a protein
erates the POC bond in the first step of nearly all phosphonate
natural-product biosynthetic pathways by converting PEP to
PnPy (Fig. 1) (39), was identified. The genome of P. syringae PB-
5123 contains only one gene (psf1) homologous to pepM. Sur-
rounding the psf1 gene are several open reading frames (ORFs)
with homology to genes present in the phosphonate biosynthetic
pathways of FR-900098 (14) and phosphinothricin (5, 48) (Fig. 2
and Table 2). As expected from previous studies on the epoxidase
responsible for the last step of fosfomycin biosynthesis in both
streptomycetes (37) and pseudomonads (36, 40) (Fig. 1), a gene
encoding the epoxidase was present (here designated psf4). An
ortholog of the fomA gene (psf7), which has previously been
shown to encode a protein that confers self-resistance by phos-
phorylating fosfomycin (16, 31, 35, 42), was also present. How-
ever, a gene encoding a FomB ortholog, another protein involved
in self-resistance in fosfomycin-producing streptomycetes by
converting the compound to its bisphosphorylated analog (31,
35), was not found in the PB-5123 genome. More surprisingly,
vert PnPy to 2-HPP in Streptomyces (Fig. 1) were also absent.
ogy with the citrate synthase-like proteins FrbC and Pms (see Fig.
S1 in the supplemental material) that are encoded in the biosyn-
thetic gene clusters of the phosphonate natural products FR-
900098 and phosphinothricin, respectively (5, 14, 24, 48). These
the carbonyl group of phosphonopyruvate and phosphinopyru-
vate, respectively (14, 24). Thus, it appears that the unfavorable
forward by decarboxylation in PB-5123, as in S. fradiae and S.
group to the carbonyl moiety of PnPy, similar to the pathways of
FR-900098 (14) and phosphinothricin (24, 52).
teins were heterologously expressed in E. coli as N-terminally
His6-tagged proteins and purified by immobilized metal affinity
chromatography (IMAC). Incubation of PEP with both proteins
indeed produced phosphonomethylmalate (Pmm), as confirmed
dard prepared using the homologous enzymes from the FR-
900098 pathway (Fig. 3) (14). Additional peaks in the phosphate
When either Psf1 or Psf2 was omitted from reaction mixtures,
PEP was not converted to new products, consistent with the ob-
servation that the equilibrium between PEP and PnPy favors PEP
by more than 500-fold (7).
FIG 2 (Top) Fosfomycin biosynthetic gene cluster in S. fradiae; (Bottom)
DNA fragment of the genome from P. syringae PB-5123 that includes the
biosynthetic gene cluster for fosfomycin production. Red, the biosynthetic
genes that are present in S. fradiae, S. wedmorensis, and P. syringae; blue, the
gae; green, a gene encoding a citrate synthase-like protein similar to enzymes
genes in the bottom panel, see Table 2.
Kim et al.
aac.asm.orgAntimicrobial Agents and Chemotherapy
Heterologous production of fosfomycin in P. aeruginosa.
We next set out to demonstrate that the 12-kb genomic DNA
fragment shown in Fig. 2 indeed encodes a functional fosfomycin
gene cluster via heterologous expression in P. aeruginosa. We first
constructed a large-insert fosmid library from PB-5123 genomic
DNA. The fosmids were individually recombined in vitro with
selected on LB agar plates supplemented with Cm and Apr. The
library was then screened by PCR with primers specific to psf1.
Nine positive clones were obtained, and these were further
screened by PCR for the presence of psf2, psf3, psf5, psf6, and psf7
using the primers in Table 1. One fosmid (46-2-11) contained an
insert of 22.5 kb with all six genes. Sequencing and annotation of
the insert revealed the 12-kb DNA fragment shown in Fig. 2. This
clone was then chromosomally integrated into the ?C31 attB site
of fosfomycin by this gene cluster.
P. aeruginosa strains KSJ570 (negative control; pJK050-AE4
integrated), KSJ629 (46-2-11 integrated), and P. syringae PB-
5123 (positive control) were analyzed for fosfomycin produc-
tion by31P NMR spectroscopy, GC-MS, and liquid chromatog-
raphy (LC)-FTMS. As reported previously, the native producer
generates only very small amounts of fosfomycin (53) (see Fig.
5B and F). P. aeruginosa containing the fosmid with the puta-
tive fosfomycin gene cluster also produced small amounts of
fosfomycin, detected by31P NMR spectroscopy, GC-MS, and
material, respectively). The negative control did not show any
fosfomycin. Thus, the genes shown in Fig. 2 indeed confer the
ability to produce fosfomycin. To further corroborate this con-
clusion, deletion of psf1 from PB-5123 abolished fosfomycin
FIG 4 NMR detection of fosfomycin produced by P. aeruginosa KSJ629.31P
NMR spectra of partially purified spent medium of P. aeruginosa KSJ629 (A)
and the same sample supplemented with fosfomycin authentic standard (B).
at 11.9 ppm. The identity of the other phosphonate products is not known.
Production of multiple phosphonates upon heterologous expression of phos-
phonate biosynthetic gene clusters is a common observation (6, 14) and may
be a consequence of either buildup of biosynthetic intermediates or detoxifi-
cation by the heterologous host.
TABLE 2 Summary of open reading frames in the fosfomycin biosynthetic gene cluster of P. syringae PB-5123a
Pseudomonas putida F1 5,10-methylene-tetrahydrofolate dehydrogenase/cyclohydrolase (YP_001268782)
Chromobacterium violaceum ATCC 12472 MFS family transporter (NP_900838)
Pseudovibrio sp. strain JE062 hypothetical protein (ZP_05084540)
Beggiatoa sp. strain PS phosphoenolpyruvate phosphomutase (ZP_01998559)
Planctomyces maris DSM 8797 hypothetical protein (ZP_01854986)
Burkholderia oklahomensis C6786 hypothetical protein (ZP_02365922)
Pseudomonas syringae epoxidase (BAA94418)
Pseudomonas syringae phosphotransferase (CAA83855)
Pseudomonas syringae unknown (CAA83856)
Pectobacterium carotovorum subsp. carotovorum PC1 6-phosphogluconate dehydrogenase (YP_003016070)
Burkholderia oklahomensis EO147 trans-homoaconitate synthase (ZP_02358889)
Photorhabdus asymbiotica subsp. asymbiotica ATCC 43949 non-heme Fe-binding protein (YP_003042097)
Pseudomonas putida GB-1 hypothetical protein (YP_001668139)
aHomologous proteins and their putative functions are listed along with the sequence identity/similarity.
bThe designations in parentheses are the GenBank accession numbers.
FIG 3 NMR and FTMS detection of the product of incubation of PEP and
Ac-CoA with Psf1 and Psf2. (A)31P NMR spectrum of 2-phosphomethyl-
malate (Pmm) formation from Ac-CoA and PEP in the presence of Psf1 and
summed mass spectrum is shown on the right.
Two Biosynthetic Pathways to Fosfomycin
August 2012 Volume 56 Number 8aac.asm.org 4179
production (see the discussion and Fig. S2 in the supplemental
Attempted in vitro reconstitution of Psf3, Psf5, Psf6, and
Psf7. Having established that the gene cluster is responsible for
fosfomycin biosynthesis, we attempted to determine the biosyn-
converts 2-HPP to fosfomycin in the last step of the pathway, the
sequences of the proteins encoded by the additional ORFs in the
gene cluster do not provide immediate clues for the manner in
homology with an NAD-dependent 6-phosphogluconate dehy-
drogenase, whereas Psf5 has sequence homology with nonheme
iron-dependent proteins (Table 2). We expressed these two pro-
teins as well as another putative dehydrogenase (Psf6) in E. coli as
N-terminally His6-tagged proteins, purified all three proteins by
IMAC, and evaluated their activity on the last known intermedi-
ates in the pathway, Pmm in the forward direction and 2-HPP in
pathway, require specialized cofactors, or act on intermediates
that are conjugated to nucleotides. The last possibility is sup-
ported by the presence of psf7, which encodes a protein with ho-
one of its biosynthetic intermediates. Indeed, a nucleotide analog
of fosfomycin, fosfadecin, has been isolated from Pseudomonas
viridiflava PK-5 (30), supporting the potential conjugation. Con-
jugation of nucleotides to pathway intermediates is also found in
the biosynthesis of FR-900098 and phosphinothricin. In both
cases, a nucleotidyltransferase transfers a CMP moiety from CTP
to one of the phosphonate oxygen atoms of a biosynthetic inter-
mediate about halfway into the overall pathway (4, 27). We het-
date have not been able to detect any activity using a series of
nucleotides and potential phosphonate substrates (fosfomycin,
dicate that Psf10, Psf11, Psf12, Psf13, and Psf15 do not have clear
homology with proteins of known function. These gene products
may carry out steps in the biosynthetic pathway between Pmm
and 2-HPP and generate intermediates that are the actual sub-
strates for Psf3/Psf5/Psf6/Psf7.
cin biosynthesis in several streptomycetes. Surprisingly, no gene
encoding such an ortholog was found in a draft genome of P.
fosfomycin, albeit in very small quantities, as had also been ob-
served previously (53). Furthermore, a fosmid containing the
open reading frames shown in Fig. 2 (bottom) was able to confer
heterologous production of fosfomycin upon P. aeruginosa
PAO1-LAC. Collectively, these results establish two distinct bio-
P. syringae. As demonstrated by the activity of the purified en-
zymes Psf1 and Fom4 in this and previous studies (36, 40), the
ments show that Fom1 and Fom4 from S. fradiae have 32% and
FIG 5 GC-MS analysis of derivatized fosfomycin. (A to D) Extracted ion chromatograms monitoring for ions with m/z 210 from concentrated extracts of the
heterologous producer P. aeruginosa KSJ629 (Me, methyl; TMSO, trimethylsilyloxy) (A), concentrated extracts of the native producer P. syringae PB-5123 (B),
authentic fosfomycin (C), and concentrated extracts of P. aeruginosa KSJ570 (negative control) (D). The arrows in panels A to D indicate the presence of an ion
with m/z of 210. In panel D, a peak with an m/z of 210 is observed, but it elutes slightly earlier (14.0 versus 14.1 min) and the mass spectrum and fragmentation
pattern do not correspond to those of the standard. (E) Mass spectrum of the GC peak at 14.1 min produced by P. aeruginosa KSJ629; (F) mass spectrum of the
at 14.0 min produced by P. aeruginosa KSJ570. All samples were silylated as described in Materials and Methods.
Kim et al.
aac.asm.orgAntimicrobial Agents and Chemotherapy
35% identity to Psf1 and Psf4, respectively (see Fig. S3 in the sup-
plemental material). These numbers demonstrate that these pro-
result of convergent evolution.
Whereas the first and last steps are the same, the genes for the
decarboxylase Fom2 and methyltransferase Fom3, used by the
thus far, are absent in P. syringae. Instead, a longer pathway ap-
pears to be used by P. syringae to convert PnPy to 2-HPP. The
details of the pathway are still unclear but likely involve the addi-
tion of an acetyl group to the carbonyl of PnPy to generate Pmm.
A chemically feasible path to fosfomycin from Pmm can be pro-
posed (Fig. 6), taking into account the sequence homology of
some of the genes found in the cluster. Hydroxylation of Pmm
(possibly by Psf5) would provide intermediate 1. Subsequent de-
carboxylation of the C-3 carboxyl group with concomitant elim-
ination of the hydroxyl group at C-1, possibly assisted by prior
phosphorylation by Psf7, would provide intermediate 2. This re-
action sequence would be analogous to phosphorylation prior to
or sulfation prior to decarboxylation in curacin A biosynthesis
(18). Alternatively, concomitant decarboxylation and dehydra-
tion without phosphorylation could take place as observed for
ger driving force as a consequence of aromatization of the sub-
strate. None of the proteins encoded by the ORFs of unknown
function in P. syringae have sequences that are homologous with
known decarboxylases. However, chemically favorable decarbox-
ylation reactions such as those involving substrates with a ?-car-
bonyl functionality or with a good ?-leaving group are catalyzed
by a wide variety of enzymes. Examples are the above-mentioned
decarboxylation during curacin biosynthesis catalyzed by a thio-
decarboxylation/elimination reactions, compound 2 would tau-
tomerize to compound 3, and this ?-keto acid could undergo
another decarboxylation to produce 2-oxopropionyl phospho-
Subsequent reduction, possibly by Psf3, would generate 2-HPP,
which is the substrate for Psf4-catalyzed formation of fosfomycin
Fig. 6 may be conjugated to phosphate or nucleotide groups.
is correct, the pathways in both Streptomyces strains as well as in
Pseudomonas add additional carbon atoms to ultimately arrive at
is performed by a radical SAM-dependent methyltransferase
(Fom3), whereas a citrate synthase-type reaction is used in pseu-
domonads (Psf2). Of note, the two pathways utilize the two pre-
viously identified strategies of driving the unfavorable equilib-
rium of PEP and PnPy forward (39), decarboxylation and
addition of Ac-CoA. These two distinct pathways to fosfomycin
natural-product biosynthesis. Similar evolutionarily distinct
pathways toward the same chemical structures have been de-
scribed in primary metabolism (e.g., menaquinone, thiamine,
lysine, and isopentenyl diphosphate) (10, 13, 28, 32), and a few
cases have been described in secondary metabolism (e.g.,
We thank Shionogi Co. Ltd. for providing P. syringae PB-5123, Jonathan
Dennis (University of Alberta) for providing plasposons pTnModR-Cm
and pTnMod-OGm, and Alfred Pühler (University of Bielefeld) for pro-
viding plasmid pK19mobSacB.
This work was supported by the U.S. National Institutes of Health
(GM077596 to W.A.V.D.D. and W.W.M.).
with solid arrows, whereas putative transformations are indicated with dashed arrows.
Two Biosynthetic Pathways to Fosfomycin
August 2012 Volume 56 Number 8 aac.asm.org 4181
1. Bailey RR. 1990. Brief overview of single-dose therapy for uncomplicated
urinary tract infections. Chemotherapy 36(Suppl 1):27–30.
2. Bailey RR. 1993. Management of lower urinary tract infections. Drugs
3. Bertani G. 1951. Studies on lysogenesis. I. The mode of phage liberation
by lysogenic Escherichia coli. J. Bacteriol. 62:293–300.
4. Blodgett JA, et al. 2007. Unusual transformations in the biosynthesis of
the antibiotic phosphinothricin tripeptide. Nat. Chem. Biol. 3:480–485.
5. Blodgett JA, Zhang JK, Metcalf WW. 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.
6. Borisova SA, Circello BT, Zhang JK, van der Donk WA, Metcalf WW.
2010. Biosynthesis of rhizocticins, antifungal phosphonate oligopeptides
produced by Bacillus subtilis ATCC6633. Chem. Biol. 17:28–37.
7. Bowman E, McQueney M, Barry RJ, Dunaway-Mariano D. 1988.
Catalysis and thermodynamics of the phosphoenolpyruvate/
phosphopyruvate rearrangement. Entry into the phosphonate class of
8. Christensen BG, et al. 1969. Phosphonomycin: structure and synthesis.
9. Cotton RG, Gibson F. 1965. The biosynthesis of phenylalanine and ty-
rosine; enzymes converting chorismic acid into prephenic acid and their
relationships to prephenate dehydratase and prephenate dehydrogenase.
Biochim. Biophys. Acta 100:76–88.
10. Dairi T, Kuzuyama T, Nishiyama M, Fujii I. 2011. Convergent strategies
in biosynthesis. Nat. Prod. Rep. 28:1054–1086.
11. de Lorenzo V, Cases I, Herrero M, Timmis KN. 1993. Early and late
responses of TOL promoters to pathway inducers: identification of post-
exponential promoters in Pseudomonas putida with lacZ-tet bicistronic
reporters. J. Bacteriol. 175:6902–6907.
12. Dennis JJ, Zylstra GJ. 1998. Plasposons: modular self-cloning minitrans-
poson derivatives for rapid genetic analysis of gram-negative bacterial
genomes. Appl. Environ. Microbiol. 64:2710–2715.
13. Elbein AD, Pan YT, Pastuszak I, Carroll D. 2003. New insights on
trehalose: a multifunctional molecule. Glycobiology 13:17R–27R. doi:
14. Eliot AC, et al. 2008. Cloning, expression, and biochemical characteriza-
timalarial compound FR900098. Chem. Biol. 15:765–770.
15. Falagas ME, Giannopoulou KP, Kokolakis GN, Rafailidis PI. 2008.
Infect. Dis. 46:1069–1077.
16. Garcia P, Arca P, Evaristo Suarez J. 1995. Product of fosC, a gene from
Pseudomonas syringae, mediates fosfomycin resistance by using ATP as
cosubstrate. Antimicrob. Agents Chemother. 39:1569–1573.
17. Gerhardt P, Murray RGE, Wood WA, Krieg NR (ed). 1994. Methods for
general and molecular bacteriology. American Society for Microbiology,
18. Gu L, et al. 2009. Polyketide decarboxylative chain termination preceded
19. Hammerschmidt F. 1991. Biosynthesis of natural products with a P-C
bond. Part 8. On the origin of the oxirane oxygen atom of fosfomycin in
Streptomyces fradiae. J. Chem. Soc. Perkin Trans. 1:1993–1996.
20. Hammerschmidt F. 1994. Incorporation of L-[methyl-2H3]methionine
and 2-[hydroxy-18O]hydroxyethylphosphonic acid into fosfomycin in
21. Harwood CR, Cutting SM. 1990. Molecular biological methods for Ba-
cillus. John Wiley & Sons, West Sussex, England.
22. Hidaka T, et al. 1995. Cloning and nucleotide sequence of fosfomycin
biosynthetic genes of Streptomyces wedmorensis. Mol. Gen. Genet. 249:
23. Hidaka T, Iwakura H, Imai S, Seto H. 1992. Studies on the biosynthesis
tivity in a fosfomycin high-producing strain of Streptomyces wedmorensis
and characterization of its blocked mutant NP-7. J. Antibiot. 45:1008–
24. Hidaka T, Shimotohno KW, Morishita T, Seto H. 1999. Studies on the
biosynthesis of bialaphos (SF-1293). 18. 2-Phosphinomethylmalic acid
synthase: a descendant of (R)-citrate synthase? J. Antibiot. 52:925–931.
25. Higgins LJ, Yan F, Liu P, Liu HW, Drennan CL. 2005. Structural insight
into antibiotic fosfomycin biosynthesis by a mononuclear iron enzyme.
26. Iyengar R, Cardemil E, Frey PA. 1986. Mevalonate-5-diphosphate de-
mevalonate 5-diphosphate. Biochemistry 25:4693–4698.
27. Johannes TW, et al. 2010. Deciphering the late biosynthetic steps of
antimalarial compound FR-900098. Chem. Biol. 17:57–64.
28. Jurgenson CT, Begley TP, Ealick SE. 2009. The structural and biochem-
ical foundations of thiamin biosynthesis. Annu. Rev. Biochem. 78:569–
29. Kahan FM, Kahan JS, Cassidy PJ, Kropp H. 1974. The mechanism of
action of fosfomycin (phosphonomycin). Ann. N. Y. Acad. Sci. 235:364–
30. Katayama N, Tsubotani S, Nozaki Y, Harada S, Ono H. 1990. Fosfade-
cin and fosfocytocin, new nucleotide antibiotics produced by bacteria. J.
31. Kobayashi S, Kuzuyama T, Seto H. 2000. Characterization of the fomA
and fomB gene products from Streptomyces wedmorensis, which confer
32. Krishnamoorthy K, Begley TP. 2011. Protein thiocarboxylate-dependent
33. Kuzuyama T, Hidaka T, Imai S, Seto H. 1993. Studies on the biosyn-
thesis of fosfomycin. V. Cloning of genes for fosfomycin biosynthesis. J.
34. Kuzuyama T, Hidaka T, Kamigiri K, Imai S, Seto H. 1992. Studies on the
of fosfomycin. J. Antibiot. 45:1812–1814.
35. Kuzuyama T, Kobayashi S, O’Hara K, Hidaka T, Seto H. 1996. Fosfo-
fomycin derivatives formed by gene products of FomA and FomB from a
fosfomycin producing organism Streptomyces wedmorensis. J. Antibiot.
36. Kuzuyama T, Seki T, Kobayashi S, Hidaka T, Seto H. 1999. Cloning and
expression in Escherichia coli of 2-hydroxypropylphosphonic acid epoxi-
5123. Biosci. Biotechnol. Biochem. 63:2222–2224.
37. Liu P, et al. 2001. Protein purification and function assignment of the
38. Marquardt JL, et al. 1994. Kinetics, stoichiometry, and identification of
the reactive thiolate in the inactivation of UDP-GlcNAc enolpyruvoyl
transferase by the antibiotic fosfomycin. Biochemistry 33:10646–10651.
39. Metcalf WW, van der Donk WA. 2009. Biosynthesis of phosphonic and
phosphinic acid natural products. Annu. Rev. Biochem. 78:65–94.
40. Munos JW, et al. 2008. Purification and characterization of the epoxidase
catalyzing the formation of fosfomycin from Pseudomonas syringae. Bio-
41. Nakashita H, Kozuka K, Hidaka T, Hara O, Seto H. 2000. Identification
Streptomyces hygroscopicus. Biochim. Biophys. Acta 1490:159–162.
42. Pakhomova S, Bartlett SG, Augustus A, Kuzuyama T, Newcomer ME.
2008. Crystal structure of fosfomycin resistance kinase FomA from Strep-
tomyces wedmorensis. J. Biol. Chem. 283:28518–28526.
43. Patel SS, Balfour JA, Bryson HM. 1997. Fosfomycin tromethamine. A
review of its antibacterial activity, pharmacokinetic properties and thera-
peutic efficacy as a single-dose oral treatment for acute uncomplicated
lower urinary tract infections. Drugs 53:637–656.
44. Pollack SJ, Freeman S, Pompliano DL, Knowles JR. 1992. Cloning, over-
expression and mechanistic studies of carboxyphosphonoenolpyruvate mu-
45. Reeves DS. 1992. Treatment of bacteriuria in pregnancy with single dose
fosfomycin trometamol: a review. Infection 20(Suppl 4):S313–S316.
46. Sambrook J, Fritsch EF, Maniatis T. 1989. Molecular cloning: a labora-
tory manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring
47. Schmit JC, Zalkin H. 1969. Chorismate mutase-prephenate dehydratase.
Partial purification and properties of the enzyme from Salmonella typhi-
murium. Biochemistry 8:174–181.
Kim et al.
aac.asm.orgAntimicrobial Agents and Chemotherapy
48. Schwartz D, et al. 2004. Biosynthetic gene cluster of the herbicide phos-
phinothricin tripeptide from Streptomyces viridochromogenes Tu494.
Appl. Environ. Microbiol. 70:7093–7102.
49. Seto H, et al. 1991. Studies on the biosynthesis of fosfomycin. 2. Conver-
sion of 2-hydroxypropyl-phosphonic acid to fosfomycin by blocked mu-
tants of Streptomyces wedmorensis. J. Antibiot. 44:1286–1288.
50. Seto H, Kuzuyama T. 1999. Bioactive natural products with carbon-
phosphorus bonds and their biosynthesis. Nat. Prod. Rep. 16:589–596.
51. Shao Z, et al. 2008. Biosynthesis of 2-hydroxyethylphosphonate, an un-
expected intermediate common to multiple phosphonate biosynthetic
pathways. J. Biol. Chem. 283:23161–23168.
52. Shimotohno KW, Seto H, Otake N, Imai S, Murakami T. 1988. Studies
on the biosynthesis of bialaphos (SF-1293). 8. Purification and character-
ization of 2-phosphinomethylmalic acid synthase from Streptomyces hy-
groscopicus SF-1293. J. Antibiot. 41:1057–1065.
53. Shoji J, et al. 1986. Production of fosfomycin (phosphonomycin) by
Pseudomonas syringae. J. Antibiot. 39:1011–1012.
54. Stalling DL, Gehrke CW, Zumwalt RW. 1968. A new silylation reagent
for amino acids bis(trimethylsilyl)trifluoroacetamide (BSTFA). Biochem.
Biophys. Res. Commun. 31:616–622.
55. Stanier RY, Palleroni NJ, Doudoroff M. 1966. The aerobic pseu-
domonads: a taxonomic study. J. Gen. Microbiol. 43:159–271.
56. Stein GE. 1998. Single-dose treatment of acute cystitis with fosfomycin
tromethamine. Ann. Pharmacother. 32:215–219.
57. Studier FW, Rosenberg AH, Dunn JJ, Dubendorff JW. 1990. Use of T7
RNA polymerase to direct expression of cloned genes. Methods Enzymol.
of a bioorganic laboratory. J. Org. Chem. 71:9561–9571.
59. Woodyer RD, Li G, Zhao H, van der Donk WA. 2007. New insight into
the mechanism of methyl transfer during the biosynthesis of fosfomycin.
Chem. Commun. (Camb.), p359–361.
60. Woodyer RD, et al. 2006. Heterologous production of fosfomycin and
identification of the minimal biosynthetic gene cluster. Chem. Biol. 13:
61. Zhang Q, van der Donk WA, Liu W. 2012. Radical-mediated enzymatic
methylation: a tale of two SAMs. Acc. Chem. Res. 45:555–564.
62. Zhao ZB, et al. 2002. Mechanistic studies of HPP epoxidase: configura-
tion of the substrate governs its enzymatic fate. Angew. Chem. Int. Ed.
Two Biosynthetic Pathways to Fosfomycin
August 2012 Volume 56 Number 8aac.asm.org 4183