β-Lactone formation during product release from a nonribosomal peptide synthetase

Article abstract of DOI:10.1038/nchembio.2374

Nonribosomal peptide synthetases (NRPSs) are multidomain modular biosynthetic assembly lines that polymerize amino acids into a myriad of biologically active nonribosomal peptides (NRPs). NRPS thioesterase (TE) domains employ diverse release strategies for off-loading thioester-tethered polymeric peptides from termination modules typically via hydrolysis, aminolysis, or cyclization to provide mature antibiotics as carboxylic acids/esters, amides, and lactams/lactones, respectively. Here we report the enzyme-catalyzed formation of a highly strained β-lactone ring during TE-mediated cyclization of a β-hydroxythioester to release the antibiotic obafluorin (Obi) from an NRPS assembly line. The Obi NRPS (ObiF) contains a type I TE domain with a rare catalytic cysteine residue that plays a direct role in β-lactone ring formation. We present a detailed genetic and biochemical characterization of the entire Obi biosynthetic gene cluster in plant-associated Pseudomonas fluorescens ATCC 39502 that establishes a general strategy for β-lactone biogenesis.

Full text of DOI:10.1038/nchembio.2374

article
puBLIsHed OnLIne: 15 MaY 2017 | dOI: 10.1038/nCHeMBIO.2374
b-Lactone formation during product release from
a nonribosomal peptide synthetase
Jason e schaffer1,2, Margaret r reck1,2, neha K prasad1 & timothy a Wencewicz1*
Nonribosomal peptide synthetases (NRPSs) are multidomain modular biosynthetic assembly lines that polymerize amino
acids into a myriad of biologically active nonribosomal peptides (NRPs). NRPS thioesterase (TE) domains employ diverse
release strategies for off-loading thioester-tethered polymeric peptides from termination modules typically via hydrolysis,
aminolysis, or cyclization to provide mature antibiotics as carboxylic acids/esters, amides, and lactams/lactones, respec-
tively. Here we report the enzyme-catalyzed formation of a highly strained b-lactone ring during TE-mediated cyclization
of a b-hydroxythioester to release the antibiotic obafluorin (Obi) from an NRPS assembly line. The Obi NRPS (ObiF) contains
a type I TE domain with a rare catalytic cysteine residue that plays a direct role in b-lactone ring formation. We present a
detailed genetic and biochemical characterization of the entire Obi biosynthetic gene cluster in plant-associated Pseudomonas
fluorescens ATCC 39502 that establishes a general strategy for b-lactone biogenesis.
trained β-lactone rings are found in diverse classes of natural is proposed to catalyze the ATP-dependent cyclization of syn and
products, including polyketides (PKs), nonribosomal peptides anti β-hydroxy acids to the corresponding cis and trans β-lactones,
S
(NRPs), amino acids, terpenoids, and hybrid molecules1. Little which undergo decarboxylation to give the cis and trans double
is known about the biosynthetic origins of β-lactones despite the bonds found in olefinic hydrocarbons. The biosynthetic gene
widetherapeuticvalueofnaturallyoccurringβ-lactonesasinhibitors clusters for the natural product β-lactones lipstatin13, ebelactone7,
of enzymes in the serine hydrolase superfamily2. Tetrahydrolipstatin salinosporamide14, and oxazolomycin15 are known, and their bio-
is a hybrid PK–NRP lipase inhibitor, known commercially as synthetic pathways have been proposed. However, the mechanisms
Orlistat, that is approved by the US Food and Drug Administration for β-lactone ring formation in these systems remain unclear, and
for the treatment of obesity3. Salinosporamide A, also known as no enzyme domains have been experimentally linked to β-lactone
marizomib and NPI-0052, is a hybrid PK–NRP proteasome inhibi- cyclization. Interestingly, the terminating PKS or NRPS module
tor that is under clinical investigation for use as a treatment of mul- for each β-lactone antibiotic lacks an embedded TE domain16. TE
tiple myeloma and other advanced malignancies4. Other natural domains can participate in acyl transfers17, epimerization of ste-
β-lactones show promise as antimicrobial5, anticancer4, antiviral6, reogenic centers18, proofreading19, and product release via hydro-
and anti-obesity3 agents (Supplementary Results, Supplementary lysis or macrocyclization20. The β-lactone rings of lactacystin and
Fig. 1). The genetic and biochemical basis for β-lactone ring forma- ebelactone can form nonenzymatically at pH 7 from precursor
tion is an unsolved problem in natural product biosynthesis7.
β-hydroxythioesters, which challenges the evolutionary need for
There is a precedent for enzymatic formation of strained four- enzyme catalysis of β-lactone ring formation7,21. We hypothesized
membered rings in the closely related β-lactam family of antibi- that enzyme catalysis could still be at play for these β-lactones, a
otics8. Three chemically distinct biosynthetic pathways leading to factor that might have been overlooked as a result of the relative sta-
β-lactams have been reported. Penicillin and cephalosporin bicyclic bility of the β-lactones. Thus, we turned our search for a ‘β-lactone
β-lactam scaffolds form via oxidative cyclization of an NRP tripep- synthase’ to scenarios in which the β-lactone is rapidly hydrolyzed
tide precursor by isopenicillin N synthase9. The β-lactam rings in in aqueous environments, where enzyme catalysis would be needed
clavulanic acid and carbapenems arise via ATP-dependent cycliza- to increase the rate of β-lactone ring formation to maintain titer
tion of β-amino acid precursors catalyzed by the β-lactam synthetase levels that are beneficial to the producing microbe.
enzyme family10. The nocardicin family of monocyclic β-lactam
We discovered that enzyme catalysis is required for strained ring
antibiotics are derived from an NRPS assembly line comprised of formation during biosynthesis of the NRPS β-lactone antibiotic
five catalytic modules that each covalently tether the evolving sub- obafluorin (Obi, 1), which rapidly hydrolyzes to the correspond-
strate in the form of a thioester on a peptidyl carrier protein, also ing β-hydroxycarboxylic acid (Obi-COOH, 2) at neutral pH22,23. We
known as the thiolation (T) domain. The condensation (C) domain used full genome sequencing and comparative sequence analysis
of module 5 has a rare HHHXXDG motif that is important for dehy- to identify a putative Obi biosynthetic gene cluster in the known
dration of a T4-serine thioester to a T4-dehydroalanine thioester Obi producer P. fluorescens ATCC 39502. We biochemically char-
electrophile that undergoes a Michael addition–nucleophilic acyl acterized five recombinant enzymes, ObiD, ObiF, ObiG, ObiH, and
substitution reaction cascade with the nucleophilic α-amino group ObiL, that convert chorismate-derived 2,3-dihydroxybenzoic acid
of a downstream T5-L-(p-hydroxyphenyl)glycine thioester to pro- (2,3-DHB, 3) and p-NH2-phenylpyruvic acid (PAPPA, 4) to Obi by
duce the nocardicin β-lactam warhead11. We hypothesized that way of the biosynthetic intermediates p-NO2-phenylpyruvic acid
similar biosynthetic strategies could conceivably be employed to (PNPPA, 5), p-NO2-phenylacetaldehyde (PNPAA, 6), and (R)-β-
assemble β-lactone rings.
hydroxy-p-nitro-L-homophenylalanine (β-OH-p-NO2-homoPhe, 7).
The β-lactone synthetase enzyme OleC produces β-lactone We found that the type I TE domain of ObiF catalyzed the cyclization
intermediates during olefinic hydrocarbon biosynthesis12. OleC of an active site–tethered Cys-β-hydroxythioester during product
1department of chemistry, Washington university in St. louis, St. louis, Missouri, uSA. 2these authors contributed equally to this work.
*e-mail: wencewicz@wustl.edu
nature CHeMICaL BIOLOGY | AdvAnce online publicAtion | www.nature.com/naturechemicalbiology
1

NATuRE CHEmICAl bIOlOgy dOI: 10.1038/nCHeMBIO.2374
article
NO2
release from the NRPS assembly line. Our findings establish a
genetic signature and biochemical mechanism for β-lactone bio-
synthesis from NRPS assembly lines that will guide future β-lactone
natural product discovery efforts through genome mining, com-
putational structure prediction, chemoenzymatic synthesis, and
a
H
N
H2O
HO
H
N
HO
OH
OH
O
O
NO2
O
OH
O
O
OH
Obi (1)
Obi-COOH (2)
directed biosynthesis24,25
.
b
RESulTS
Identification of the obafluorin biosynthetic gene cluster
Obi is a cis-monocyclic β-lactone antibiotic produced by plant-asso-
ciatedstrainsofP.fluorescens22,23.Underphysiologicalconditions,Obi
spontaneouslyhydrolyzestothecorrespondingβ-hydroxycarboxylic
acid Obi-COOH (Fig. 1a)26. The unique structure of Obi contains a
2,3-DHB unit coupled through an amide linkage to an α-amino-β-
lactone ring of the unusual nonproteinogenic amino acid β-OH-p-
NO2-homoPhe. Stable-isotope feeding studies suggest that 2,3-DHB
and β-OH-p-NO2-homoPhe are advanced intermediates in the
biosynthesis of Obi27,28. We sequenced the genome of the known
Obi producer P. fluorescens ATCC 39502 to search for a putative
Obi biosynthetic gene cluster. We identified a ~20-kilobase-pair
gene cluster with 14 candidate coding sequences (CDSs), includ-
ing one encoding an AurF di-iron nonheme arylamine oxygenase
homolog (obiL) predicted to install the aryl nitro functional group
TAr
C
A
T
TE AAr
0 kb
5 kb
10 kb
15 kb
20 kb
2,3-DHB biosynthesis
NRPS
Regulation
β-OH-p-NO2-homoPhe biosynthesis
Figure 1 | biosynthetic gene cluster for obafluorin b-lactone from
P. fluorescens ATCC 39502. (a) obafluorin β-lactone (obi) undergoes
nonenzymatic hydrolysis to form a β-hydroxycarboxylic acid (obi-cooH).
(b) Annotated biosynthetic gene cluster of obi identified from
whole-genome sequencing of P. fluorescens Atcc 39502. Arrows represent
coding sequences. A, acid adenylation; AAr, aryl acid adenylation;
c, condensation; t, acid thiolation; tAr, aryl acid thiolation; te, thioesterase;
nRpS, nonribosomal peptide synthetase.
via oxidation of the precursor aniline (Fig. 1b and Supplementary Conversion of PAPPA to b-OH-p-NO2-homoPhe
Table 1). The Obi gene cluster encodes putative enzymes required Three enzymes (ObiL, ObiG, and ObiH) are required to convert
for the biosynthesis of 2,3-DHB (obiA, obiB, obiC, and obiE) and PAPPA to β-OH-p-NO2-homoPhe (Fig. 2a). We found that ObiL,
the nonproteinogenic amino acid β-OH-p-NO2-homoPhe (obiG, a putative nonheme di-iron oxygenase, catalyzed the six-electron
obiH, obiI, obiJ, obiK, and obiL). CDSs encoding an aryl acyl car- oxidation of the arylamine PAPPA to the aryl nitro PNPPA. Related
rier protein (obiD) and an NRPS module (obiF) with condensation examples of arylamine oxidases are found in the biosynthetic path-
(C), acid adenylation (A), aryl acid adenylation (AAr), thiolation ways of aureothin (AurF; GenBank EJZ60582.1) and chloram-
(T), and thioesterase (TE) domains make up the remaining biosyn- phenicol (CmlI; PDB 5HYH). ObiL showed 24% and 18% sequence
thetic genes. There are no obvious antibiotic resistance genes in the homology to AurF and CmlI, respectively (Supplementary Table 3;
cluster. The genes obiM and obiN encode proteins that have strong Supplementary Fig. 6a)33,34. Similarly to ObiL, AurF acts early in
homology to the acylhomoserine lactone (AHL) synthase LuxI and the biosynthetic pathway of aureothin to oxidize p-aminobenzoic
the AHL-binding transcriptional regulator LuxR, respectively. acid (PABA) to p-nitrobenzoic acid (PNBA). CmlI catalyzes the
The LuxI and LuxR pair is a common quorum-sensing system oxidation of the penultimate aniline intermediate to the aryl nitro
found in many pseudomonads. A basic local alignment search as the final step of chloramphenicol biosynthesis34. Treatment of
tool (BLAST) search of the NCBI database revealed homologous recombinant ObiL with PAPPA under reaction conditions similar
Obi clusters in other pseudomonads including Pseudomonas sp. to those used to functionally characterize AurF resulted in an
37_R_15 and 34_E_7, the environmental chitin-degrading bacte- immediate color change of the reaction mixture from clear and col-
rium Chitiniphilus shinanonensis SAY3, and the plant-growth-pro- orless to sky blue (Supplementary Fig. 7)33. Subsequent treatment
moting rhizobacterium Burkholderia diffusa RF8-non_BP2, which of the ObiL–PAPPA reaction mixture with recombinant ObiG and
is also an opportunistic human pathogen (Supplementary Table 2; ObiH in the presence of thiamine diphosphate (ThDP), pyridoxal
Supplementary Figs. 2 and 3).
phosphate (PLP), and L-Thr produced β-OH-p-NO2-homoPhe,
which was detected by LC–MS (Fig. 2b). The results of the tri-
ple-enzyme ObiL–ObiG–ObiH reaction confirmed that all three
2,3-DHb and PAPPA are precursors to Obi
Chorismate is the primary carbon source of the Obi biosynthetic recombinant enzymes were capable of converting PAPPA directly
pathway (Supplementary Fig. 4a)27,28. The gene obiB encodes an to β-OH-p-NO2-homoPhe, but the timing of each reaction required
extra copy of the first biosynthetic enzyme from the shikimate further clarification.
pathway, 3-deoxy-D-arabinoheptulosonate-7-phosphate (DAHP)
We discovered that ObiG, a ThDP-dependent phenylpyruvate
synthase, which converts phosphoenolpyruvate (PEP) and eryth- decarboxylase,catalyzedtheconversionofPNPPAtoPNPAA(Fig.2a;
rose-4-phosphate to chorismate29. Chorismate is converted to 2,3- Supplementary Figs. 6b and 8). A colorimetric Purpald assay sug-
DHB by isochorismate synthase (ObiA), isochorismatase (ObiE), gested that PNPPA is decarboxylated to the aldehyde PNPAA in
and a dehydrogenase (ObiC), which are analogous to EntC, EntB, the presence of ThDP and ObiG (Supplementary Fig. 9). A dou-
and EntA, respectively, involved in biosynthesis of the siderophore ble-enzyme reaction with ObiG and ObiH in the presence of ThDP,
enterobactin30. Chorismate is also converted to PAPPA by the PLP, and L-Thr revealed that PNPPA is directly converted to β-OH-
4-amino-4-deoxychorismate (4-ADC) synthase (ObiJ and ObiK) p-NO2-homoPhe, which was detected by LC–MS (Fig. 2c) and
and 4-ADC mutase/dehydratase (ObiI), a process analogous to that 1H-NMR (Fig. 2d). PAPPA was not accepted as a substrate by ObiG
of the CmlB, CmlC, and CmlD and PapA, PapB, and PapC enzymes and ObiH, suggesting that ObiL oxidized PAPPA before catalysis
that lead to PAPPA in the chloramphenicol31 and pristinamycin32 by ObiG and ObiH (Fig. 2c,d and Supplementary Fig. 10).
biosynthetic pathways, respectively (Supplementary Fig. 4b). We
BLAST searches suggested that ObiH was a PLP-dependent thre-
undertook in vitro reconstitution of the remaining biosynthetic onine aldolase35. Similar aldolases are involved in the biosynthesis
enzymes (ObiD, ObiF, ObiG, ObiH, and ObiL) to establish the of natural products containing β-hydroxy-α-amino acids, including
timing and action of each step in the conversion of 2,3-DHB and the peptidyl nucleoside antibiotics A090289 A, caprazamycin A, and
PAPPA to Obi (Supplementary Fig. 5).
muramycin A1 (ref. 36), the 4-methyl-oxazoline antibiotics JBIR-34
2
nature CHeMICaL BIOLOGY | AdvAnce online publicAtion | www.nature.com/naturechemicalbiology

NATuRE CHEmICAl bIOlOgy dOI: 10.1038/nCHeMBIO.2374
article
a
CH3CHO
L-Thr
CO2
O
O
H2N
ObiH
PLP
ObiG
ObiL
O
Ha
HO
HO
HO
Fe(II)SO4, PMS,
NADH, O2
MES buꢀer
pH 5.5
ThDP, MgCl2
MES buꢀer
pH 7
O
O
O
OH
NO2
MES buꢀer
pH 7
NH2
NO2
NO2
β-OH-p-NO2-homoPhe (7)
PAPPA (4)
PNPPA (5)
PNPAA (6)
b
c
d
ObiG,H substrate comparison
1H-NMR signal of Ha
ObiL,G,H assay
ObiG,H pyruvic acid screen
5
17.5
15
4
3
2
1
δ = 4.43 p.p.m.
δ = 4.40 p.p.m.
12.5
10
Chemical shift
100
7.5
5
+ PNPPA + ObiG – ObiH
+ PNPPA – ObiG + ObiH
+ PNPPA – ObiG – ObiH
50
0
+ PAPPA – ObiL + ObiG + ObiH
+ PAPPA + ObiL + ObiG + ObiH
2.5
0
+ PNPPA + ObiG + ObiH
+ PAPPA + ObiG + ObiH
0
2
4
6
8
PAPPA
PNPPA
2
4
6
Time (min)
Time (min)
Substrate
Figure 2 | Enzymatic conversion of PAPPA to b-OH-p-NO2-homoPhe using recombinant oxidase Obil, decarboxylase Obig, and aldolase ObiH.
(a) Scheme depicting the conversion of pAppA to β-oH-p-no2-homophe by tandem actions of obil, obiG, and obiH. (b) extracted ion chromatograms
(eics) (β-oH-p-no2-homophe [M+H]+ ion m/z = 241) for the triple-enzyme reaction of pAppA with obil, obiG, and obiH. pMS, phenazine methosulfate.
(c) eics for the double-enzyme reaction of pAppA and pnppA with obiG and obiH. eics were generated by lc–MS in positive ion mode using single
ion monitoring (SiM) for the predicted m/z values for the pAppA (m/z = 211) and pnppA (m/z = 241) product [M+H]+ ions. data represents a single
measurement from two independent experiments. (d) the box shows a zoomed region of the 1H-nMR spectra for the obiG and obiH double-enzyme
reactions with l-thr and pAppA or pnppA after 3 h of equilibration in which Ha on c3 of β-oH-p-no2-homophe appears as a doublet of doublet of
doublets. the bar graph depicts the relative percent of l-thr remaining for each reaction, based on integration of the l-thr methyl signal in the 1H-nMR
spectra. error bars represent s.d. for three independent trials (P < 0.05).
and BE-32030 (ref. 37), and the α-methyl-L-Ser antibiotics amicetin We examined whether recombinant ObiF and ObiD, together,
and bamicetin38. We showed that ObiH catalyzed a retro-aldol reac- couldconvertβ-OH-p-NO2-homoPheand2,3-DHBtoObi(Fig. 3a).
tion on L-Thr, yielding acetaldehyde and a PLP-stabilized glycine Prior to running in vitro enzyme reactions, we primed recom-
enolate that underwent a crossed aldol reaction with PNPAA to binant ObiD and ObiF with 4-phosphopantetheine transferase
yield (2S,3R)-β-OH-p-NO2-homoPhe, setting the final stereochem- (Sfp) and coenzyme A to install the phosphopantetheine post-
istry found in the Obi β-lactone scaffold (Fig. 2a; Supplementary translational modification on the T domains. Treatment of β-OH-
Figs. 6c and 11). Double-enzyme reactions with decarboxylase ObiG p-NO2-homoPhe and 2,3-DHB with holo-ObiF and holo-ObiD,
and aldolase ObiH monitored by LC–MS established that L-Thr is together, in the presence of ATP and Mg2+ gave clean conversion to
greatly preferred to D-Thr, L-allo-Thr, D-allo-Thr, L-Ser, D-Ser, and the ring-closed Obi β-lactone and the corresponding ring-opened
Gly as the amino acid substrate (Supplementary Fig. 10). The ste- hydrolysis product Obi-COOH, presumably originating from the
reochemistry of β-OH-p-NO2-homoPhe matches that of L-Thr, sug- nonenzymatic hydrolysis of Obi (Fig. 3b; Supplementary Fig. 15;
gesting a preference of this substrate for giving the required enolate Supplementary Table 4).
stereochemistry for the asymmetric aldol reaction (Supplementary
Unlike the lipstatin13, ebelactone7, salinosporamide14, and oxazo-
Fig. 11). The ObiH PLP cofactor formed a glycine enolate quinon- lomycin15 PKS and NRPS assembly lines, ObiF has a TE domain in
oid intermediate upon exposure to L-Thr, as observed by optical the termination module. Primary sequence analysis of the ObiF TE
absorption spectroscopy (λ = 495 nm), which was stable through- domain revealed that the expected catalytic serine of the conserved
out the entire protein expression and purification process, including GXSXG motif is replaced by a cysteine at residue 1141; this mutation
dialysis (Supplementary Figs. 12 and 13). Kinetic analysis of the has been found previously in type II TEs but is rarely reported for
ObiH reaction by 1H-NMR suggested strong product feedback inhi- type I TEs16,39,40. We prepared phosphopantetheinylated C1141S and
bition of ObiH (Fig. 2d and Supplementary Fig. 10). We hypoth- C1141A ObiF variants to test the importance of C1141 in β-lactone
esized that downstream utilization of the nonproteinogenic amino formation. The C1141S mutation converted ObiF from a ‘β-lactone
acid β-OH-p-NO2-homoPhe by the ATP-consuming NRPS ObiF synthase’ to a classic hydrolase that produced only Obi-COOH
would drive the ObiH equilibrium toward product.
from the in vitro reaction of the holo-ObiF C1141S mutant and
holo-ObiD with β-OH-p-NO2-homoPhe and 2,3-DHB (Fig. 3b,c;
Supplementary Fig. 15; Supplementary Table 5). The C1141A
ObiF is a b-lactone-forming NRPS
The NRPS ObiF is a single-module, 209-kDa protein with C–A–T– mutation brought the catalytic activity of ObiF to a near halt, with
TE–AAr catalytic domains (Supplementary Figs. 6e,f and 14), and ObiD only trace amounts of Obi-COOH detected by LC–MS (Fig. 3b,c;
is a separate aryl acyl carrier protein (TAr) (Supplementary Fig. 6d). Supplementary Fig. 15; Supplementary Table 6). Presumably, the
nature CHeMICaL BIOLOGY | AdvAnce online publicAtion | www.nature.com/naturechemicalbiology
3

NATuRE CHEmICAl bIOlOgy dOI: 10.1038/nCHeMBIO.2374
article
a
Nonenzymatic hydrolysis
NO2
ObiD
ObiF
NH2
HO
H
H
N
N
OH
HO
+
+
HO
OH
HO
ATP, Mg2+
75 mM Tris
pH 7.5
OH
O
O
O
OH
OH
O
OH
O
NO2
NO2
O
O
OH
Obi (1)
m/z [M+H]+ = 359
Obi-COOH (2)
β-OH-p-NO2-homoPhe (7)
2,3-DHB (3)
m/z [M+H]+ = 377
b
c
d
60 min
ObiF C1141A
ObiF C1141S
ObiF WT
ObiF C1141A
ObiF C1141S
ObiF WT
30
25
20
15
10
5
12
10
8
32P exchange assay
30 min
20,000
15,000
10,000
5,000
0
60 min
10 min
6
30 min
4
10 min
2
0
0
9
10
9
10 9
10
9
9
9
Retention time (min)
Retention time (min)
Figure 3 | Enzymatic conversion of b-OH-p-NO2-homoPhe and 2,3-DHb to Obi b-lactone using recombinant NRPS assembly line ObiF and ObiD.
(a) Scheme depicting the enzymatic conversion of β-oH-p-no2-homophe and 2,3-dHb to obi (m/z = 359 for [M+H]+) and obi-cooH (m/z = 377 for
[M+H]+). (b) eics (m/z = 359) for reactions of 2,3-dHb, β-oH-p-no2-homophe, and holo-obid with holo-obiF wild type (Wt; black), holo-obiF c1141S
mutant (red), and holo-obiF c1141A mutant (blue). (c) eics (m/z = 377) for reactions of 2,3-dHb, β-oH-p-no2-homophe, and holo-obid with holo-obiF
Wt (black), holo-obiF c1141S mutant (red), and holo-obiF c1141A mutant (blue). (d) Atp–[32p]ppi exchange assay to detect carboxylic acid adenylation
by obiF. the graph shows counts per minute (c.p.m.) for each carboxylate substrate. Additional amino acid substrates (Arg, Asn, Asp, cys, Glu, Gln, Gly,
His, ile, leu, lys, Met, orn, pro, trp, val) are not included in the graph because of the negligible c.p.m. observed for them. error bars represent s.d. for three
independent trials (P < 0.05).
penultimate 2,3-DHB-β-OH-p-NO2-homoPhe thioester stalled Analysis of the ObiF primary sequence with NRPSPredictor2
on the T domain, as acyl transfer to the TE domain of the holo- software predicted that the embedded A domain would adenylate
ObiF C1141A mutant was blocked. The trace Obi-COOH that L-Thr and that the terminal AAr domain would adenylate 2,3-DHB
was detected by LC–MS is predicted to form via slow hydrolysis (Supplementary Table 8)41. The specificity-conferring sequence42
of the T-domain thioester. Prolonged exposure (24 h) of the holo- of the embedded ObiF A domain (D/A/W/G/C/G/L/I) showed sim-
ObiF C1141A mutant and holo-ObiD to 2,3-DHB, β-OH-p-NO2- ilarity to the consensus signature sequences for A domains that acti-
homoPhe, and ATP resulted in steady production of Obi-COOH, vate L-Thr (D/F/W/N/I/G/M/V) and L-Phe (D/A/W/T/I/A/A/V),
supporting the theory that the NRPS module was functional but with some key differences that help account for accommodation
required slow, nonenzymatic hydrolysis of the 2,3-DHB-β-OH-p- of the sterically larger β-OH-p-NO2-homoPhe substrate. In com-
NO2-homoPhe T-domain thioester for turnover. We also expressed, parison to L-Phe A domains, Thr at position 4 is changed to a Gly,
purified,andfunctionallycharacterizedrecombinantObiFandObiD extending the active site pocket to accommodate the chain extended
homologs from C. shinanonensis SAY3 (Supplementary Fig. 16; homoPhe. In comparison to L-Thr A domains, Phe at position 2 is
Supplementary Table 2). Reaction of holo-ObiF and holo-ObiD changed to Ala, and Asn at position 4 is changed to Gly, which leaves
from C. shinanonensis with β-OH-p-NO2-homoPhe and 2,3-DHB room to accommodate the benzyl group. Both L-Phe and L-Thr A
gave enzyme-, time-, and ATP-dependent conversion to Obi and domains have an Ile at position 5, which is changed to Cys in ObiF.
Obi-COOH (Supplementary Fig. 17; Supplementary Table 7). The The Cys might interact favorably with the electron-deficient nitrop-
ability of recombinant holo-ObiF and holo-ObiD from C. shinano- henyl group (Supplementary Fig. 18).
nensis to catalyze the in vitro conversion of β-OH-p-NO2-homoPhe
and 2,3-DHB to Obi with similar efficiency to the P. fluorescens b-lactone formation requires Cys1141
enzymes suggests that the NRPS module would be functional when We propose a direct role for the ObiF TE domain in the formation
expressed and phosphopantetheinylated in C. shinanonensis.
of the Obi β-lactone ring as the final biosynthetic step, releasing the
We employed an ATP–[32P]pyrophosphate (PPi) exchange mature cyclized product from the NRPS assembly line. The ObiF
assay to test for reversible acyl-adenylate intermediate forma- 2,3-DHB–β-OH-p-NO2-homoPhe T thioester must be shuttled
tion and showed that the ObiF A domains are highly selective for from the T domain and loaded to the TE domain via energy-neu-
the activation of β-OH-p-NO2-homoPhe and 2,3-DHB (Fig. 3d). tral transthioesterification, where it resides as the C1141 thioester
Proteinogenic amino acids, including the β-hydroxy (L-Thr and and undergoes C3-OH-to-C1 cyclization to release Obi β-lactone.
L-Ser) and aromatic (L-Phe) amino acids, produced signals compa- The weaker C–S bond of the thioester, compared to the C–O bond
rable to those of control reactions lacking an amino acid substrate. of an oxo-ester, is likely required to make strained β-lactone ring
4
nature CHeMICaL BIOLOGY | AdvAnce online publicAtion | www.nature.com/naturechemicalbiology

NATuRE CHEmICAl bIOlOgy dOI: 10.1038/nCHeMBIO.2374
article
NO2
a
3.2 h (Fig. 4c). The rate of Obi-SNAC hydrolysis was comparable to
that of the ObiF C1141A mutant (Fig. 3c; Supplementary Fig. 15),
which further supports the role of the ObiF TE domain as a catalyst
for β-lactone ring formation.
N-Acetylcysteamine
H
H
O
H
H
N
N
(SNAC)
HO
HO
OH
OH
O
OH
O
NO2
O
S
O
H
N
Obi (1)
Obi-SNAC (8)
O
A model for Obi biosynthesis
NO2
MES buꢀer
pH 7
MES buꢀer
pH 7
Our in vitro characterization of ObiD, ObiF, ObiG, ObiH, and ObiL;
computational prediction of ObiA, ObiB, ObiC, ObiE, ObiI, ObiJ,
and ObiK function; and previously reported isotope labeled pre-
cursor studies27,28 enabled us to propose a complete model for Obi
biosynthesis in P. fluorescens (Fig. 5). Carbon flux to Obi biosynthe-
sis starts from the primary metabolites erythrose-4-phosphate and
PEP, which are converted to chorismate by the upregulated DAHP
synthase ObiB and the endogenous shikimate pathway. ObiA, ObiC,
and ObiE convert chorismate to 2,3-DHB, and ObiI, ObiJ, and ObiK
convert it to PAPPA. Aryl amine oxidase ObiL converts PAPPA to
PNPPA, andThDP-dependentdecarboxylaseObiGconvertsPNPPA
to PNPAA. Threonine aldolase ObiH establishes equilibrium
between PNPAA, L-Thr, acetaldehyde, and β-OH-p-NO2-homoPhe.
The ATP-consuming NRPS ObiF activates β-OH-p-NO2-homoPhe
as the acyl adenylate with the embedded A domain and covalently
loads the amino acid as a phosphopantetheinyl thioester on the
embedded T domain. 2,3-DHB is similarly activated as the acyl ade-
nylate by the C-terminal AAr domain and loaded on the stand-alone
TAr domain ObiD as a phosphopantetheinyl thioester. The C domain
is predicted to catalyze amide bond formation between the α-amino
group of the ObiF β-OH-p-NO2-homoPhe-T-thioester and the car-
bonyl of the ObiD 2,3-DHB-TAr-thioester, releasing free TAr (ObiD)
and forming the 2,3-DHB-β-OH-p-NO2-homoPhe-T-thioester.
Transthioesterification to the active site C1141 of the TE domain
leads to cyclization of the β-hydroxythioester to the corresponding
β-lactone, releasing Obi and turning over the NRPS assembly line
for another round of catalysis (Supplementary Fig. 20).
H
N
HO
OH
OH
O
O
OH
Obi-COOH (2)
b
c
Fmoc-Ala
Obi-COOH
400
300
200
100
0
200
175
150
125
100
75
Unknown
Obi
Obi-COOH
Fmoc-Ala
intermediate
Obi-SNAC
5.5 h
5.0 h
4.5 h
4.0 h
3.5 h
3.0 h
2.5 h
2.0 h
1.5 h
std
std
std
6 h
2 h
50
25
1.0 h
0.5 h
0 h
0.5 h
0 h
0
14
16
18
20
22
15
16
17
18
19 20 21
Retention time (min)
22
Retention time (min)
Figure 4 | ObiF catalysis is required for b-lactone ring formation.
(a) Scheme depicting the synthesis of obi-N-acetylcysteamine (SnAc) and
the hydrolysis of obi and obi-SnAc to obi-cooH. treatment of obi with
neat SnAc results in quantitative conversion to obi-SnAc. Hydrolysis of
obi (b) and obi-SnAc (c) at pH 7 in MeS buffer was monitored by Hplc
with detection by optical absorbance spectroscopy at 270 nm using an
Fmoc-l-Ala internal standard. peak identities were confirmed using lc–MS
and comparison to authentic standards. Std, pure analytical standard. the
traces represent single measurements from two independent experiments.
DISCuSSION
formation thermodynamically favorable43. The higher ground state There are some general features from the ObiF TE domain that might
energy of the Cys thioester, compared to that of the Ser oxo-ester, be genetic and biochemical signatures for β-lactone biosynthesis.
is also predicted to increase the rate of alcohol addition to the car- A GXCXG motif is likely required for β-lactone ring formation. The
bonyl group, which is likely the rate-determining step. The thermo- active site Cys provides a favorable kinetic and thermodynamic
dynamic requirements for β-lactone formation are not met by the scenario for strained-ring formation, proceeding through a more
oxo-ester intermediate created by the ObiF C1141S mutant, result- reactive thioester compared to the traditional oxo-ester intermedi-
ing in the accumulation of only the hydrolysis product Obi-COOH ate. The advantages of activated TE domain thioesters as substrates
(Fig. 3c). The ObiF C1141A mutant also failed to generate detect- for NRPS product release might extend to other strained-ring sys-
able concentrations of Obi β-lactone, which supports the theory tems including β-lactams. Recently, the biosynthetic gene cluster for
that TE catalysis is required in β-lactone ring formation.
the monobactam sulfazecin was reported to encode an NRPS with
TofurtherprobetheroleoftheTEdomaininβ-lactoneformation, a terminal type I TE domain that contains an active site Cys resi-
wepreparedanN-acetylcysteamine(SNAC)thioester(Obi-SNAC,8) due44. GXCXG motifs are rare among type I TE domains16. Analysis
and compared its rate of hydrolysis with that of Obi β-lactone. of the TE protein family (PF00975) revealed that 264 out of 3,863
Dissolving purified Obi from P. fluorescens culture supernatant in members (6.8%) contained a GXCXG motif, whereas the remain-
neat SNAC thiol resulted in quantitative conversion to Obi-SNAC ing 93.2% contained a GXSXG motif. Analysis of the top 20,000
thioester (Fig. 4a; Supplementary Fig. 19). We monitored the hits from standard protein BLAST search of the full ObiF sequence
hydrolysis of Obi (Fig. 4b) and Obi-SNAC (Fig. 4c) in MES buffer against nonredundant protein sequences in the NCBI database
at pH 7 by HPLC with detection by optical absorbance spectroscopy revealed that 13,944 sequences possessed a GXSXG motif, 1,240
at 270 nm. We observed rapid hydrolysis of Obi in MES buffer at pH sequences contained a GXCXG motif, and 194 sequences contained
7 with a half-life <30 min and immediate formation of two products a GHCXG motif. Homologous ObiF TE domains from P. fluore-
(retention times of 15.2 min and 17.2 min) with optical absorbance scens, C. shinanonensis, and B. diffusa all contain a GHCAG motif
spectra that are consistent with the presence of an aryl nitro group and show >50% total sequence homology (Supplementary Fig. 16;
(Fig. 4b). Comparison to an analytical standard of Obi-COOH Supplementary Table 9). Comparison to a diverse sampling of
purified from P. fluorescens culture supernatant confirmed the peak PKS and NRPS TE domains, including pyochelin and AB3403 TE
at retention time 17.2 min to be Obi-COOH. The peak at retention domains with GXCXG motifs, showed that the ObiF TE domain
time 15.2 min is an uncharacterized intermediate that converted is sequence unique (Supplementary Fig. 21; Supplementary
directly to Obi-COOH. Previous studies of Obi hydrolysis using Table 10)39,40. Homology modeling predicted overall structural sim-
1H-NMR in D2O/CD3CN (1:4) showed that an oxazoline interme- ilarities to canonical type I α/β hydrolases, with the conserved Asp
diate forms and hydrolyzes to give O-(2,3-dihydroxybenzoyl)-β- residue of the catalytic triad moved from β-strand 6 to β-strand 7
OH-p-NO2-homoPhe26. We observed that Obi-SNAC hydrolyzed (Supplementary Fig. 22)16. Future mechanistic and structural stud-
directly to Obi-COOH in MES buffer at pH 7 with a half-life of ies will help elucidate the catalytic roles of the His–Cys–Asp triad
nature CHeMICaL BIOLOGY | AdvAnce online publicAtion | www.nature.com/naturechemicalbiology
5

NATuRE CHEmICAl bIOlOgy dOI: 10.1038/nCHeMBIO.2374
article
CH3CHO
L-Thr
ObiH
PLP
CO2
H
H
H
N
NH2
HO
O
O
HO
O
O
OH
NO2
O
OH OH
O
PNPPA (5)
NO2
NO2
NO2
ObiG
ThDP
Obi (1)
β-OH-p-NO2-homoPhe (7)
PNPAA (6)
TE
A
ObiL
AAr
C
T
Fe(II), O2
SH
ATP
II.
SH
HO
V.
AMP, PPi
O
PAPPA (4)
NH2
TE
TE
A
A
AAr
AAr
C
C
T
T
SH
ObiI
ObiJ
ObiK
ObiF/D
catalytic cycle
HS
OH
O
S
S
O
O
HO
OH
OH
N
H
H2N
7-T-thioester
2-TE-thioester
NO2
NO2
Chorismate
TAr
IV.
III.
O
S
3-TAr-thioester
TE
SH
A
Endogenous
shikimate
pathway
AAr
OH
C
T
ObiB
OH
TE
A
O
S
OH
O
AAr
C
T
HO
OH
SH
PEP,
erythrose-4-P
N
H
SH
i
TAr
SH
NO2
2-T-thioester
AMP, PPi
I.
Endogenous
shikimate
pathway
ATP
ObiB
HO
O
ObiA
ObiC
ObiE
OH
OH
Chorismate
2,3-DHB (3)
Figure 5 | model for Obi biosynthesis and the catalytic cycle of NRPS assembly line ObiF. obib directs carbon flux from the primary metabolic pool to
chorismate via the endogenous shikimate pathway. obiA, obic, and obie convert chorismate to 2,3-dHb. obii, obiJ, and obiK convert chorismate to
pAppA. obil converts pAppA to pnppA. obiG converts pnppA to pnpAA. obiH establishes equilibrium between l-thr, pnpAA, acetaldehyde, and β-oH-
p-no2-homophe. obiF and obid are the nRpS assembly line components that convert β-oH-p-no2-homophe and 2,3-dHb to obi at the cost of two
Atp molecules. the nRpS catalytic cycle starts (i) with AAr catalyzed activation and loading of 2,3-dHb to the tAr domain as the phosphopantetheinyl
thioester. Similarly, the obiF embedded A domain activates and loads β-oH-p-no2-homophe to the obiF t domain (ii). the c domain catalyzes amide bond
formation giving the 2-t-thioester (iii). transthioesterification to active site cys1141 of the te domain (iv) leads to cyclization releasing obi β-lactone (v).
and will clarify whether substrate preorganization plays a role in of EntF45. The ObiF MbtH-like domain might play a similar role to
formation of the strained four-membered β-lactone ring10,20
.
YbdZ. Recent structural and biophysical studies of NRPS modules
Inclusion of the 2,3-DHB-activating AAr domain as part of the with C–A–T–TE domains have provided insight into the motions
ObiF NRPS module might be optional. In the P. fluorescens and and dynamics of individual domains during coordinated catalysis
C. shinanonensis ObiF NRPSs the AAr domain is included in the by the enzyme assembly lines40,46. The TE domain is highly dynamic
CDS as the C-terminal catalytic domain. In the B. diffusa NRPS, the and is distal from the T domain until invoked for final product
AAr domain is predicted to be a stand-alone enzyme that activates release. In the case of ObiF, this would leave AAr free to catalyze the
2,3-DHB as the acyl adenylate for in trans transfer to the TAr carrier formation of the 2,3-DHB acyl adenylate and keep the TAr carrier
protein in analogy to the enterobactin NRPS machinery (EntF, EntE, protein loaded as a 2,3-DHB phosphopantetheinyl thioester. In the
and EntB) (Supplementary Fig. 23). An MbtH domain is encoded case of enterobactin biosynthesis, the AAr and TAr domains, EntE and
in the ObiF primary sequence between the TE and AAr domains. The EntB, respectively, are separate from the main NRPS module, EntF,
MbtH-like protein YbdZ is encoded in the enterobactin biosynthetic which activates and loads Ser as a T-domain phosphopantetheinyl
gene cluster and has been shown to enhance the adenylation activity thioester45. The full-length EntF is required for trimerization of Ser,
6
nature CHeMICaL BIOLOGY | AdvAnce online publicAtion | www.nature.com/naturechemicalbiology

NATuRE CHEmICAl bIOlOgy dOI: 10.1038/nCHeMBIO.2374
article
3. Pemble, C.W. IV, Johnson, L.C., Kridel, S.J. & Lowther, W.T. Crystal structure
of the thioesterase domain of human fatty acid synthase inhibited by Orlistat.
Nat. Struct. Mol. Biol. 14, 704–709 (2007).
amide bond formation with 2,3-DHB, and cyclization to the mac-
rolactone47. Thus, only AAr and TAr domains have been shown to be
active in trans, supporting the existence of ObiF NRPS modules with
and without the AAr domain as part of the main NRPS module.
ItisconceivablethatothercatalyticdomainswithactivesiteCysres-
idues might catalyze cyclization of β-hydroxythioesters to β-lactones.
For example, an uncharacterized ketosynthase (KS) domain resides
at the C terminus of the ebelactone PKS termination module7. KS
domains undergo transthioesterification reactions with T domain
thioesters to form intermediate KS-Cys thioesters on a conserved
Cys–His–His triad48. KS domains typically catalyze thio-Claisen
condensations with malonyl T-domain thioesters, but might also be
capable of β-lactone ring forming condensations. The termination
module of the hybrid PKS–NRPS assembly line for oxazolomycin fea-
tures a C-terminal C domain of unknown function15. Condensation
domains typically catalyze amide bond formation between two T
domain thioesters, but there are documented cases of C domains cata-
lyzing cyclization during product release such as in cyclosporin A and
enniatin biosynthesis20. The oxazolomycin terminal C domain might
catalyze the intramolecular condensation of a T domain thioester to
the spirocyclic β-lactone. The salinosporamide biosynthetic gene clus-
ter encodes for several uncharacterized enzymes, including a stand-
alone KS domain, SalC, that might participate in bicyclic β-lactone
formation4,14. We predict that more β-lactone-forming “logic gates”
4. Gulder, T.A. & Moore, B.S. Salinosporamide natural products: Potent 20S
proteasome inhibitors as promising cancer chemotherapeutics. Angew. Chem.
Int. Edn Engl. 49, 9346–9367 (2010).
5. De Pascale, G., Nazi, I., Harrison, P.H. & Wright, G.D. β-Lactone natural
products and derivatives inactivate homoserine transacetylase, a target for
antimicrobial agents. J. Antibiot. (Tokyo) 64, 483–487 (2011).
6. Lall, M.S., Ramtohul, Y.K., James, M.N.G. & Vederas, J.C. Serine and
threonine β-lactones: a new class of hepatitis A virus 3C cysteine proteinase
inhibitors. J. Org. Chem. 67, 1536–1547 (2002).
7. Wyatt, M.A. et al. Biosynthesis of ebelactone A: isotopic tracer, advanced
precursor and genetic studies reveal a thioesterase-independent cyclization to
give a polyketide β-lactone. J. Antibiot. (Tokyo) 66, 421–430 (2013).
8. Hamed, R.B. et al. e enzymes of β-lactam biosynthesis. Nat. Prod. Rep. 30,
21–107 (2013).
9. Roach, P.L. et al. Structure of isopenicillin N synthase complexed with
substrate and the mechanism of penicillin formation. Nature 387,
827–830 (1997).
10. Bachmann, B.O., Li, R. & Townsend, C.A. β-Lactam synthetase: a new
biosynthetic enzyme. Proc. Natl. Acad. Sci. USA 95, 9082–9086 (1998).
11. Gaudelli, N.M., Long, D.H. & Townsend, C.A. β-Lactam formation by a
non-ribosomal peptide synthetase during antibiotic biosynthesis. Nature 520,
383–387 (2015).
12. Christenson, J.K. et al. β-Lactone synthetase found in the olefin biosynthesis
pathway. Biochemistry 56, 348–351 (2017).
13. Bai, T. et al. Operon for biosynthesis of lipstatin, the β-lactone inhibitor of
human pancreatic lipase. Appl. Environ. Microbiol. 80, 7473–7483 (2014).
will be discovered as part of NRPS–PKS modular assembly lines and 14. Eustáquio, A.S. et al. Biosynthesis of the salinosporamide A polyketide
possibly as autonomous catalysts10,16,49
.
synthase substrate chloroethylmalonyl-coenzyme A from S-adenosyl-L-
methionine. Proc. Natl. Acad. Sci. USA 106, 12295–12300 (2009).
15. Zhao, C. et al. Oxazolomycin biosynthesis in Streptomyces albus JA3453
featuring an “acyltransferase-less” type I polyketide synthase that incorporates
two distinct extender units. J. Biol. Chem. 285, 20097–20108 (2010).
16. Horsman, M.E., Hari, T.P. & Boddy, C.N. Polyketide synthase and
nonribosomal peptide synthetase thioesterase selectivity: logic gate or a
victim of fate? Nat. Prod. Rep. 33, 183–202 (2016).
Obi represents a novel structural class of β-lactone antibiotics
from nature’s chemical inventory. The Obi biosynthetic machin-
ery characterized in this study offers a versatile platform for the
chemoenzymatic synthesis of β-lactones from β-hydroxy-α-amino
acid precursors. Here, the new amino acid β-OH-p-NO2-homoPhe
was produced by a threonine aldolase, possibly representing a gen-
eral strategy for producing β-hydroxy-α-amino acid precursors to
peptide β-lactones. Fe(II)/α-KG-dependent oxygenases and non-
heme di-iron oxygenases are also known to produce β-hydroxy-
α-amino acids via direct C–H bond activation chemistry on free
amino acids50,51. Hydroxylation of amino acid side chains can also
take place while loaded as thioesters on NRPS T domains, making A
domain substrate prediction insufficient for the potential incorpo-
ration of β-hydroxy-α-amino acids into NRP scaffolds52. Our work
shows that β-hydroxy-α-amino acids can be directly incorporated
into peptide scaffolds on an NRPS assembly line, leading to peptides
capped at the C terminus with a β-lactone warhead. Our genetic and
biochemical characterization of the Obi biosynthetic pathway in
P. fluorescens will guide future genome mining efforts and expands
the known chemistry of NRPS and PKS product release mecha-
nisms to include strained 4-membered ring formation10,20,24. Our
chemoenzymatic approach to C-terminal β-lactone peptides has the
potential for applications in precursor-directed feeding studies and
NRPS engineering to produce targeted peptide β-lactone inhibitors
of proteases for the treatment of diseases associated with microbial
and viral infections5,6,25, cancer4, and obesity3.
17. Jiang, Y., Morley, K.L., Schrag, J.D. & Kazlauskas, R.J. Different active-site
loop orientation in serine hydrolases versus acyltransferases. ChemBioChem
12, 768–776 (2011).
18. Gaudelli, N.M. & Townsend, C.A. Epimerization and substrate gating
by a TE domain in β-lactam antibiotic biosynthesis. Nat. Chem. Biol. 10,
251–258 (2014).
19. Jensen, K. et al. Polyketide proofreading by an acyltransferase-like enzyme.
Chem. Biol. 19, 329–339 (2012).
20. Kopp, F. & Marahiel, M.A. Macrocyclization strategies in polyketide
and nonribosomal peptide biosynthesis. Nat. Prod. Rep. 24, 735–749
(2007).
21. Dick, L.R. et al. Mechanistic studies on the inactivation of the proteasome by
lactacystin: a central role for clasto-lactacystin β-lactone. J. Biol. Chem. 271,
7273–7276 (1996).
22. Wells, J.S., Trejo, W.H., Principe, P.A. & Sykes, R.B. Obafluorin, a novel
β-lactone produced by Pseudomonas fluorescens. Taxonomy, fermentation and
biological properties. J. Antibiot. (Tokyo) 37, 802–803 (1984).
23. Tymiak, A.A., Culver, C.A., Malley, M.F. & Gougoutas, J.Z. Structure of
obafluorin: an antibacterial β-lactone from Pseudomonas fluorescens.
J. Org. Chem. 50, 5491–5495 (1985).
24. Dejong, C.A. et al. Polyketide and nonribosomal peptide retro-biosynthesis
and global gene cluster matching. Nat. Chem. Biol. 12, 1007–1014 (2016).
25. Hamed, R.B. et al. Crotonase catalysis enables flexible production
of functionalized prolines and carbapenams. J. Am. Chem. Soc. 134,
471–479 (2012).
26. Pu, Y., Lowe, C., Sailer, M. & Vederas, J.C. Synthesis, stability, and
antimicrobial activity of (+)-obafluorin and related β-lactone antibiotics.
J. Org. Chem. 59, 3642–3655 (1994).
received 1 June 2016; accepted 7 March 2017;
published online 15 May 2017
27. Herbert, R.B. & Knaggs, A.R. Biosynthesis of the antibiotic obafluorin from
D-[U-13C]glucose and p-aminophenylalanine in Pseudomonas fluorescens.
J. Chem. Soc. Perkin Trans. I 1992, 103–107 (1992).
mETHODS
28. Herbert, R.B. & Knaggs, A.R. Biosynthesis of the antibiotic obafluorin from
p-aminophenylalanine and glycine (glyoxylate). J. Chem. Soc. Perkin Trans. I
1992, 109–113 (1992).
Methods, including statements of data availability and any associated
accession codes and references, are available in the online version
of the paper.
29. Bentley, R. e shikimate pathway—a metabolic tree with many branches.
Crit. Rev. Biochem. Mol. Biol. 25, 307–384 (1990).
references
30. Walsh, C.T., Liu, J., Rusnak, F. & Sakaitani, M. Molecular studies on enzymes
in chorismate metabolism and the enterobactin biosynthetic pathway.
Chem. Rev. 90, 1105–1129 (1990).
1. Lowe, C. & Vederas, J.C. Naturally occuring β-lactones: occurence, synthesis
and properties. A review. Org. Prep. Proced. Int. 27, 305–346 (1995).
2. Bachovchin, D.A. et al. Superfamily-wide portrait of serine hydrolase
inhibition achieved by library-versus-library screening. Proc. Natl. Acad.
Sci. USA 107, 20941–20946 (2010).
31. Fernández-Martínez, L.T. et al. New insights into chloramphenicol
biosynthesis in Streptomyces venezuelae ATCC 10712. Antimicrob. Agents
Chemother. 58, 7441–7450 (2014).
nature CHeMICaL BIOLOGY | AdvAnce online publicAtion | www.nature.com/naturechemicalbiology
7

NATuRE CHEmICAl bIOlOgy dOI: 10.1038/nCHeMBIO.2374
article
32. Blanc, V. et al. Identification and analysis of genes from Streptomyces
pristinaespiralis encoding enzymes involved in the biosynthesis of the
4-dimethylamino-L-phenylalanine precursor of pristinamycin I. Mol.
Microbiol. 23, 191–202 (1997).
33. Choi, Y.S., Zhang, H., Brunzelle, J.S., Nair, S.K. & Zhao, H. In vitro
reconstitution and crystal structure of p-aminobenzoate N-oxygenase
(AurF) involved in aureothin biosynthesis. Proc. Natl. Acad. Sci. USA 105,
6858–6863 (2008).
34. Makris, T.M. et al. An unusual peroxo intermediate of the arylamine
oxygenase of the chloramphenicol biosynthetic pathway. J. Am. Chem. Soc.
137, 1608–1617 (2015).
35. Contestabile, R. et al. l-reonine aldolase, serine hydroxymethyltransferase
and fungal alanine racemase. A subgroup of strictly related enzymes
specialized for different functions. Eur. J. Biochem. 268, 6508–6525 (2001).
36. Barnard-Britson, S. et al. Amalgamation of nucleosides and amino acids in
antibiotic biosynthesis: discovery of an L-threonine:uridine-5′-aldehyde
transaldolase. J. Am. Chem. Soc. 134, 18514–18517 (2012).
37. Muliandi, A. et al. Biosynthesis of the 4-methyloxazoline-containing
nonribosomal peptides, JBIR-34 and -35, in Streptomyces sp. Sp080513GE-23.
Chem. Biol. 21, 923–934 (2014).
38. Zhang, G. et al. Characterization of the amicetin biosynthesis gene cluster
from Streptomyces vinaceusdrappus NRRL 2363 implicates two alternative
strategies for amide bond formation. Appl. Environ. Microbiol. 78,
2393–2401 (2012).
39. Reimmann, C., Serino, L., Beyeler, M. & Haas, D. Dihydroaeruginoic
acid synthetase and pyochelin synthetase, products of the pchEF genes,
are induced by extracellular pyochelin in Pseudomonas aeruginosa.
Microbiology 144, 3135–3148 (1998).
40. Drake, E.J. et al. Structures of two distinct conformations of holo-non-
ribosomal peptide synthetases. Nature 529, 235–238 (2016).
41. Röttig, M. et al. NRPSpredictor2—a web server for predicting NRPS
adenylation domain specificity. Nucleic Acids Res. 39, W362–W367 (2011).
42. Challis, G.L., Ravel, J. & Townsend, C.A. Predictive, structure-based model of
amino acid recognition by nonribosomal peptide synthetase adenylation
domains. Chem. Biol. 7, 211–224 (2000).
48. Smith, S. & Tsai, S.C. e type I fatty acid and polyketide synthases: a tale of
two megasynthases. Nat. Prod. Rep. 24, 1041–1072 (2007).
49. Kohli, R.M., Takagi, J. & Walsh, C.T. e thioesterase domain from
a nonribosomal peptide synthetase as a cyclization catalyst for
integrin binding peptides. Proc. Natl. Acad. Sci. USA 99, 1247–1252
(2002).
50. Makris, T.M., Chakrabarti, M., Münck, E. & Lipscomb, J.D. A family
of di-iron monooxygenases catalyzing amino acid β-hydroxylation in
antibiotic biosynthesis. Proc. Natl. Acad. Sci. USA 107, 15 391–15396
(2010).
51. Jiang, W. et al. EcdGHK are three tailoring iron oxygenases for amino acid
building blocks of the echinocandin scaffold. J. Am. Chem. Soc. 135,
4457–4466 (2013).
52. Haslinger, K. et al. e structure of a transient complex of a nonribosomal
peptide synthetase and a cytochrome P450 monooxygenase. Angew. Chem.
Int. Edn Engl. 53, 8518–8522 (2014).
acknowledgments
This work was supported with start-up funds provided by Washington University
in St. Louis and the Research Corporation for Science Advancement through
a Cottrell Scholar award to T.A.W. We thank C.T. Walsh (Stanford ChEM-H) for
productive scientific discussions. We thank Cofactor Genomics (St. Louis, MO) for
bioinformatics consultation and genome sequence analysis. We thank S. Alvarez and
B. Evans at the Proteomics & Mass Spectrometry Facility at the Donald Danforth Plant
Science Center (St. Louis, MO) for the acquisition of high-resolution MS spectra (NSF
Grant No. DBI-0922879). We thank J.-S. Taylor (WUSTL Department of Chemistry) for
assistance with the [32P]PPi exchange assay. We thank J. Kao (WUSTL Department of
Chemistry) for assistance with NMR experiments.
author Contributions
T.A.W., J.E.S., and M.R.R. wrote the paper and prepared the supplementary
information. T.A.W. oversaw all of the experiments. J.E.S. and M.R.R. cloned and
purified ObiG, ObiH, and ObiL. J.E.S. functionally characterized ObiG, ObiH, and
ObiL. M.R.R. cloned, purified, and functionally characterized ObiD and ObiF. J.E.S.
and M.R.R. purified and characterized all compounds. T.A.W. and N.K.P. isolated
Pseudomonas fluorescens gDNA, analyzed sequencing data, and annotated the
Obi biosynthetic gene cluster. N.K.P. performed protein homology modeling and
helped with preparation of the supplementary information.
43. McGrath, N.A. & Raines, R.T. Chemoselectivity in chemical biology: acyl
transfer reactions with sulfur and selenium. Acc. Chem. Res. 44, 752–761 (2011).
44. Li, R., Oliver, R.A. & Townsend, C.A. Identification and characterization of
the sulfazecin monobactam biosynthetic gene cluster. Cell Chem. Biol. 24,
24–34 (2017).
45. Miller, B.R., Drake, E.J., Shi, C., Aldrich, C.C. & Gulick, A.M. Structures of a
nonribosomal peptide synthetase module bound to MbtH-like proteins
support a highly dynamic domain architecture. J. Biol. Chem. 291,
22559–22571 (2016).
Competing financial interests
The authors declare no competing financial interests.
46. Reimer, J.M., Aloise, M.N., Harrison, P.M. & Schmeing, T.M. Synthetic cycle
of the initiation module of a formylating nonribosomal peptide synthetase.
Nature 529, 239–242 (2016).
additional information
Any supplementary information, chemical compound information and source data are
available in the online version of the paper. Reprints and permissions information is
available online at http://www.nature.com/reprints/index.html. Publisher’s note: Springer
Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations. Correspondence and requests for materials should be
addressed to T.A.W.
47. Ehmann, D.E., Shaw-Reid, C.A., Losey, H.C. & Walsh, C.T. e EntF and
EntE adenylation domains of Escherichia coli enterobactin synthetase:
sequestration and selectivity in acyl-AMP transfers to thiolation domain
cosubstrates. Proc. Natl. Acad. Sci. USA 97, 2509–2514 (2000).
8
nature CHeMICaL BIOLOGY | AdvAnce online publicAtion | www.nature.com/naturechemicalbiology

ONlINE mETHODS
agitation at 37 °C. A 200 μl aliquot of this culture was used to inoculate 500 ml
of terrific broth (12 g/L tryptone, 24 g/L yeast extract, 5 g/L glycerol, 17 mM
Strains, materials, and instrumentation. P. fluorescens ATCC 39502 was
purchased from ATCC, and glycerol stocks were made from cells grown in KH2PO4, and 72 mM K2HPO4) containing 50 μg/ml kanamycin. The culture
Bennett’s media and stored at −80 °C. DNA primers were purchased from was grown at 37 °C with agitation until OD600 reached approximately 0.4.
Integrated DNA Technologies. Herculase II DNA polymerase kit was pur- The culture was cooled in an ice bath for 20 min, then 500 μl of a sterile 0.5 M
chased from Agilent. T4 DNA ligase and NheI were purchased from New IPTG solution was added. Culture was then incubated with agitation for 18 h
England BioLabs. T4 Polynucleotide Kinase and FastDigest (FD) restriction (at 15 °C for ObiG/ObiF/CS-ObiF, and at 20 °C for ObiD/CS-ObiD/ObiH/ObiL).
enzymes NdeI, HindII, and DpnI were purchased from Thermo Scientific. From this point on, all protein purification steps were performed at 4 °C. Cells
TOP10 Escherichia coli cells were purchased from Invitrogen and BL21-
Gold(DE3) E. coli cells were purchased from Agilent. PCR was performed
on a Bio-Rad MyCycler thermal cycler. Nickel-nitrilotriacetic acid (Ni-NTA)
were harvested by centrifugation at 5,000 r.p.m. for 20 min. Supernatant was
discarded, and cell pellets were each suspended in 40 ml cold lysis buffer
(50 mM K2HPO4 pH 8.0, 500 mM NaCl, 5 mM β-mercaptoethanol, 20 mM
agarose was purchased from Invitrogen. Any kD SDS–PAGE gels were pur- imidazole, and 10% glycerol). Cell suspensions were transferred to 50 ml
chased from Bio-Rad. Plasmid DNA sequencing was performed by Genewiz. Falcon tubes and flash frozen in liquid nitrogen. Frozen cells were thawed
Buffers, salts, and chemical reagents were purchased from Sigma-Aldrich, and gently rocked for 30 min before being mechanically lysed using an Avestin
unless otherwise stated. Fully deuterated MES buffer and D2O were purchased
from Cambridge Isotope Laboratories. PAPPA was purchased from Tractus
Chemistry. PNPPA was purchased from Frinton Laboratories, Inc. PPA and
phenazine methosulfate were purchased from Acros Organics. 32P- radiolabeled
sodium pyrophosphate was purchased from PerkinElmer. Samples for HPLC
EmulsiFlex-C5 cell disruptor. Cell lysate was centrifuged at 45,000 r.p.m. for
35 min and supernatant was incubated with pre-washed Ni-NTA resin for
30 min. Resin was washed twice with 40 ml lysis buffer and then eluted five
times with 10 ml elution buffer (50 mM K2HPO4 pH 8.0, 500 mM NaCl,
5 mM β-mercaptoethanol, 300 mM imidazole, and 10% glycerol). Fractions
and LC–MS were prepared in 0.45 μm PTFE MiniUniPrep vials from Agilent. containing the majority of protein, as judged by SDS–PAGE with Coomassie
All pH measurements were recorded using an Orion Star A111 pH meter and blue visualization (Supplementary Fig. 5), were combined in 10,000 MWCO
a PerpHecT ROSS micro combination pH electrode from Thermo Scientific. SnakeSkin dialysis tubing from Thermo Scientific and soaked overnight in
Analytical HPLC was performed using Beckman Coulter SYSTEM GOLD 126
solvent module, 508 autosampler, and 168 detector with a Phenomenex Luna
1.8 L phosphate buffer (50 mM K2HPO4 pH 8.0, 150 mM NaCl, 1 mM DTT).
Dialyzed protein solution was concentrated using an appropriately sized spin
5 μm C18(2) 100A column, 250 × 4.6 mm with guard column. All preparatory filter (EMD Millipore Amicon Ultra 15 ml Centrifugal Filters). Concentrated
HPLC was performed using a Beckman Coulter SYSTEM GOLD 127P solvent protein solutions were flash frozen in liquid nitrogen and stored at –80 °C.
module and 168 detector with a Phenomenex Luna 10u C18(2) 100A column,
250 × 21.20 mm, 10 μm with guard column. Analytical and prep HPLC were Mutagenesis of WT-ObiF to ObiF C1141S and ObiF C1141A. Mutagenesis
performed with mobile phases of 0.1% TFA in (A) water and (B) acetonitrile, primers (Supplementary Tables 11 and 12) were phosphorylated using T4
and data were processed using 32 Karat software, version 7.0. Low-resolution
LC–MS was performed on an Agilent 6130 single quadrupole LC–MS with
Polynucleotide Kinase according to product instructions. Primers were taken
directly from the phosphorylation mixture for use in PCR amplification. PCR
G1313 autosampler, G1315 diode array detector, and 1200 series solvent mod- was carried out using a Herculase II Fusion DNA Polymerase kit. After ampli-
ule. A Phenomenex Gemini C18 column, 50 × 2 mm, 5 μm with guard column fication, 1 μl FastDigest DpnI was added to PCR reaction mixture to digest
was used for LC–MS separations. LC–MS mobile phases were 0.1% formic template plasmid with wild-type gene. The reaction was incubated at 37 °C for
acid in (A) water and (B) acetonitrile, and data were processed using G2710 60 min, and plasmids were then purified using gel electrophoresis. Purified
ChemStation software. NMR was performed on Varian Unity Plus-300 MHz,
Varian Unity Inova-500 MHz, and Varian Unity-600 MHz (with cold probe)
plasmids were ligated overnight at 16 °C using T4 DNA ligase, then trans-
formed into electrocompetent E. coli TOP10 cells. Plasmids were sequenced to
instruments. Optical absorption spectroscopy was performed on a Cary 50 fit confirm the desired mutations.
with an autosampler and water Peltier thermostat system using 1 cm quartz
cuvettes. High-resolution MS spectra were collected using an LTW-Velos Pro
Orbitrap at the Donald Danforth Plant Science Center, St. Louis, MO.
ObiG Purpald reactions. A pre-incubation solution was made with 10 μM
ThDP, 10 μM MgCl2, 25 mM MES buffer pH 7.5, and 10 μM ObiG from a
freshly thawed frozen stock. The enzyme pre-incubation solution incubated
Sequencing of P. fluorescens ATCC 39502 Genomic DNA. The P. fluorescens at room temperature for 5 min. In a separate 1.5 ml Eppendorf tube, the pyru-
ATCC 39502 gDNA was isolated using a Qiagen DNeasy Blood & Tissue kit vic acid substrate (PAPPA, PNPPA, PHPPA, or PPA) was diluted to 1 mM in
following the provided instructions. The gDNA was sequenced by Ambry 25 mM MES pH 7.5 buffer. The substrate mix was transferred in one aliquot
Genetics (Aliso Viejo, CA) using Illumnia MiSeq. Contigs containing ORFs were into the pre-incubation solution. The final concentrations of reagents were
assembled, annotated, and deposited in GenBank (Accession #s KX134682– 10 μM ObiG, 5μM ThDP, 5μM MgCl2, 1 mM pyruvic acid substrate, and 25 mM
KX134695) with assistance from Cofactor Genomics (St. Louis, MO).
MES buffer pH 7.5. Aliquots of the enzyme solution were added to an equal
volume of a 20 mg/ml solution of Purpald in 1 M aqueous NaOH at vari-
ous time points. The resulting solution was mixed and then left open to air
to develop the purple color (Supplementary Fig. 9). All reactions and time
Cloning, expression, and purification of ObiG, ObiH, ObiL, ObiD, ObiF,
ObiF C1141S, ObiF C1141A, CS-ObiD and CS-ObiF. Gene sequences were
amplified from either P. fluorescens ATCC 39502 gDNA or C. shinanonensis points were single trials.
DSM 23277 (CS) gDNA (isolated using Qiagen DNeasy Blood & tissue kit
following provided instructions) using forward and reverse primers designed
for each gene (Supplementary Tables 11–13) and a Herculase II Fusion
DNA Polymerase kit. The resulting PCR fragments were purified by gel elec- stock, 50 μM PLP, and 25 mM MES buffer was prepared and analyzed by opti-
Detection of ObiH quinonoid by optical absorption spectroscopy. In a 1.5 ml
Eppendorf tube, a master mix of 50 μM ObiH from a freshly thawed frozen
trophoresis. PCR products and pET28a vector were separately digested with cal absorption spectroscopy scanning from 200 to 600 nm. Three solutions at
either FD-NdeI and FD-HindIII or NheI and FD-HindIII. Cut DNA was puri-
fied using a QIAquick Gel Extraction Kit. Ligation reactions were carried out
overnight at 16 °C using T4 DNA Ligase. Ligation mixtures were purified and
a final volume of 500 μl were analyzed as single trials: (1) L-Thr was added to
the master mix a 1 mM final concentration; (2) PAA was added to the master
mix at 1 mM final concentration; (3) L-Thr and PAA were added to the master
transformed into electrocompetent E. coli TOP10 cells. Clones were selected mix at 1 mM final concentration each. Samples were then analyzed by optical
for on LB agar containing 50 μg/ml kanamycin. Sequencing of selected colo- absorption spectroscopy scanning from 200–600 nm in order to observe the
nies revealed clones containing the desired constructs. Plasmids were purified
and used to transform electrocompetent E. coli BL21 (DE3) cells.
quinonoid peak, which absorbs at ~495 nm (Supplementary Fig. 12).
For protein expression, a 5 ml culture of E. coli BL21 harboring the appropri- Combined ObiG and ObiH double enzyme reactions. Three solutions were
ate plasmid was grown overnight in LB containing 50 μg/ml kanamycin with prepared using individual Eppendorf tubes. First, an ObiG solution containing
nature CHeMICaL BIOLOGY
doi:10.1038/nchembio.2374

10 μM ThDP, 10 μM MgCl2, and 10 μM ObiG was adjusted to 333 μl final vol- C. shinanonensis homologs) was carried out in separate 500 μl solutions con-
ume using 25 mM MES buffer pH 7.5 and allowed to pre-incubate for 5 min. taining 180 μM CoASH, 5 mM DTT, 10 mM MgCl2, 400 nM Sfp, 25 μM apo
Second, apyruvicacidsolution, 1mMfinalconcentration, wasadjustedto333μl enzyme, and 75 mM Tris–HCl at pH 7.5 (ref. 54). Reactions were left at room
with 25 mM MES buffer pH 7.5. Third, an ObiH solution containing 10 μM PLP, temperature for 2 h, and then used directly as the source of holo-ObiD and
1 mM amino acid (L-Thr, D-Thr, L-Ser, D-Ser, Gly, L-alloThr, or D-alloThr), and
10 μM ObiH was adjusted to 333 μl final volume. After the pre-incubation time
holo-ObiF. Reactions were performed in duplicate at 500 μl total volume.
Full reactions contained 5 mM ATP, 5 mM DTT, 1 mM 2,3-DHB, 1 mM β-OH-
was complete, the three solutions were combined in a single Eppendorf tube p-NO2-homoPhe, 1 μM holo-ObiF (ObiF, CS-ObiF, ObiF-C1141A, or ObiF-
and allowed to rest for 3 h at room temperature before quenching with MeCN C1141S), 1 μM holo-ObiD, and 75 mM Tris–HCl at pH 7.5. For reactions with
to crash the enzyme. The mixtures were centrifuged and the supernatants were CS-ObiF, the holo-ObiD homolog from C. shinanonensis (CS-ObiD) was used
analyzed by LC–MS using a gradient of 0% B held for 5 min then 0% B to 95%
as the aryl acyl adenylating enzyme. Control reactions were also prepared by
B over 10 min using single ion monitoring for the expected ions in positive ion replacing various components with an equivalent volume of 75 mM Tris–HCl
mode (retention times and observed product ions: β-OH-homoPhe, 1.7 min, buffer. Three control experiments were performed for the ObiF, ObiF-C1141S,
[M+H]+ = 196; β-OH-p-OH-homoPhe, 0.8 min, [M+H]+ = 212; β-OH-p-NO2- ObiF-C1141A, CS-ObiF reactions: (−) ObiD; (−) 2,3-DHB/(−) β-OH-p-NO2-
homoPhe, 2.9 min, [M+H]+ = 241; β-OH-p-NH2-homoPhe, not observed; homoPhe; and (−) ATP. Two additional controls were prepared as either (−)
Fig. 2c and Supplementary Fig. 10). Product masses were confirmed by high- ObiF or (−) ObiF/(−) ObiD (Supplementary Tables 4–7). Reactions were per-
resolution LC–MS analysis. For NMR analysis the same samples were prepared
formed at room temperature. At 10, 30, and 60 min time points, a 50 μl aliq-
using fully deuterated MES buffer. After equilibration for 3 h, the reactions uot from each reaction was quenched with 50 μl acidic MeCN (acidified with
were quenched with TFA to a pH ~2. The samples were then flash frozen in HCl) to give a final pH of approximately 3.5. Mixtures were centrifuged for
liquid nitrogen, lyophilized to dryness, resuspended in 750 μl of D2O, and cen- 2 min at 13,000 r.p.m. to remove precipitated enzyme, and the supernatants
trifuged to pellet any insoluble particulate. The soluble supernatants were then
transferred to NMR tubes and analyzed (Fig. 2d; Supplementary Fig. 10). All
reactions were performed in triplicate.
were analyzed by LC–MS using a gradient of 5% solvent B to 95% solvent B
over 20 min. Ion counts for Obi and Obi-COOH were extracted from total ion
chromatograms in positive ion mode (retention times and observed ions: Obi,
9.5min,[M+H]+ =359;Obi-COOH,9.0min,[M+H]+ =377)(Fig.3b,c).Reaction
Purification of β-OH-p-NO2-homoPhe from the ObiG/ObiH reaction. The mixtures were also analyzed by HPLC using a gradient of 5% B to 95% B over
ObiG/ObiH double-enzyme reaction described above was scaled up with 20 min, 95% B to 100% B over 3 min, and 100% B to 5% B over 2 min at a flow
respect to the substrates to produce enough material for NMR analysis. The
concentrations of reaction components were 10 μM ObiG, 10 μM ThDP, Fig. 15). Product ions and retention times were confirmed using purified
rate of 1 ml/min with optical absorbance detection at 270 nm (Supplementary
10 μM MgCl2, 10 μM ObiH, 10 μM PLP, 5 mM L-Thr, 5 mM PNPPA, and 25 mM
pH 7.5 MES buffer, with a final reaction volume of 1 ml. After 3 h the reaction
was quenched with TFA to a pH ~2, centrifuged at 13,000 r.p.m. for 2 min, flash
frozen in liquid nitrogen, and lyophilized to dryness. The resulting solid was
dissolved in 1:10 MeCN:H2O, filtered, and purified by preparatory HPLC using
standards of Obi and Obi-COOH isolated from P. fluorescens ATCC 39502 fer-
mentations (Supplementary Fig. 19).
[32P]PPi exchange assay. Reaction mixtures (650 μl) contained 2 μM ObiF,
5 mM acid substrate, 1 mM ATP, 1 mM MgCl2, 40 mM KCl, 1 mM DTT,
a solvent gradient of 0% B held for 5 min then 0% B to 100% B over 10 min 5 mM Na[32P]PPi (3.3 × 105 c.p.m./ml), and 50 mM Tris–HCl (pH 8)55. Mixtures
(β-OH-p-NO2-homoPhe retention time = 14.3 min). The desired TFA salt were incubated at room temperature for 30 min, then three 200 μl aliquots
1
of β-OH-p-NO2-homoPhe product was isolated as a white powder. H NMR were removed and each were quenched with 500 μl of a charcoal suspension
(500 MHz, D2O) δ(p.p.m) = 8.24 (d, J = 8.4 Hz, 2 H), 7.55 (d, J = 9.0 Hz, 2 H), 4.49 (100 mM NaPPi, 350 mM HClO4, and 16 g/L powdered charcoal). The mixtures
– 4.43 (ddd, J = 3.6, 4.8, 10.2 Hz, 1 H), 3.98 (d, J = 4.8 Hz, 1 H), 3.20 (dd, J = 3.6,
were shaken, then centrifuged at 13,000 r.p.m. for 3 min. Charcoal pellets were
14.4 Hz, 1 H), 2.99 (dd, J = 10.2, 13.8 Hz, 1 H); 13C NMR (126 MHz, D2O) washed with 750 μl of wash solution (100 mM NaPPi, 350 mM HClO4) and
δ (p.p.m.) = 160.6, 136.0, 134.9, 119.7, 113.2, 59.1, 47.5, 28.9; HRMS (ESI) cal- centrifuged again for 3 min. Washing step was repeated once more, followed
culated for C10H11N2O5: 241.0819 [(M+H)+], found peak when extracting ions by suspension of charcoal pellets in 1.5 ml EcoLite(+) scintillation fluid from
for 241.0795–241.0843. See Supplementary Note for tabulated NMR data and
MP Biomedicals. Charcoal-bound radioactivity was measured on a Beckman
Coulter LS 6500 scintillation counter. Graph of results indicates measured
c.p.m. (Fig. 3d). All proteinogenic amino acids were L-stereoisomers.
copies of spectra.
CombinedObiG,ObiH,andObiLtripleenzymereaction. Asolutionof100μM
ObiL from a freshly thawed frozen stock, 3 mM phenazine methosulfate (PMS), Isolation of Obi and Obi-COOH. Obi and the hydrolysis product Obi-COOH
1 mM PAPPA, 100 μM iron(II) sulfate, and 25 mM NADH was prepared in
were isolated from Pseudomonas fluorescens ATCC 39502 fermentations as
25 mM MES buffer at pH 5.5 at a final volume of 500 μl (refs. 33,53). NADH was previously described22,23. Bennet’s Agar slants (1 g/L yeast extract, 1 g/L beef
added last to initiate the reaction and the mixture was incubated at room tem- extract, 2 g/L NZ amine, 10 g/L glucose, 15 g/L agar) were inoculated with
perature for 1 h. Upon addition of NADH, the color of the reaction changed from streaks of P. fluorescens ATCC 39502 from a frozen glycerol stock. The slants
light yellow to sky blue (Supplementary Fig. 7). In a separate Eppendorf tube, were incubated at 25 °C for 48 h. 5 ml of sterile saline was added to the top of
a 500 μl solution was prepared containing 10 μM ObiG, 10 μM ThDP, 10 μM the slant and shaken gently. Inoculated saline (1 mL) was transferred to 100
MgCl2, 10 μM ObiH, 10 μM PLP, 1 mM L-Thr, and 25 mM MES buffer at pH
7. The pH of the ObiL solution changed from 5.5 to 7.0 over the course of the 0.1 g/L FeSO4·7H2O, 200 ml soil filtrate extract from campus of Washington
reaction. The control assay reached an end pH of 6.0 and was adjusted to a University in St. Louis, 800 ml tap water, autoclaved). This starter culture was
ml of sterile media (5 g/L yeast extract, 5 g/L glucose, 0.1 g/L MgSO4·7H2O,
pH of 7 using 1 M NaOH before addition of the ObiG/ObiH solution. The incubated with shaking at 225 r.p.m. at 25 °C for 24 h. 5 ml of starter culture
ObiG/ObiH solution was added directly to the ObiL solution, allowed to react was transferred to each 3 L baffled flask containing 500 ml of the same media,
for 2 h at room temperature, and quenched with MeCN to crash the enzymes.
The mixtures were centrifuged and the supernatants were analyzed by LC–MS
using a gradient of 0% B held for 5 min then 0% B to 95% B over 10 min. Ion
and the cultures were incubated with shaking at 225 r.p.m. at 25 °C for 17 h.
The cultures were pooled and centrifuged at 5,000 r.p.m. at 4 °C for 25 min to
pellet cells. The supernatant was pooled and acidified to pH 3 with 1M aque-
counts for expected products were extracted from total ion chromatograms in ous HCl. The pooled supernatant was saturated with EtOAc then extracted
positive ion mode (Fig. 2b; retention times and observed product ions: β-OH- with three 100 ml volumes of EtOAc. Extractions were pooled and dried using
p-NO2-homoPhe, 1.7 min, [M+H]+ = 241). The β-OH-p-NO2-homoPhe ion rotary evaporation to yield a gray-brown solid. The solid was dissolved in 10 ml
was confirmed by HRMS. Reaction was performed as two independent trials.
acetonitrile, filtered, and concentrated by rotary evaporation under reduced
pressure to yield 221 mg of clear, brown oil. The oil was dissolved in 10 ml
CombinedObiFandObiDdoubleenzymereaction.Phosphopantetheinylation of acetonitrile, filtered, and purified with Prep HPLC with a gradient of 5%
of apo-ObiD and apo-ObiF (wild type, C1141S and C1141A mutants, and
solvent B to 95% solvent B over 20 min. The two largest peaks were isolated
nature CHeMICaL BIOLOGY
doi:10.1038/nchembio.2374

which upon NMR analysis proved to be Obi (retention time = 23.0 min) and
(d, J = 8.7 Hz, 1 H), 4.59 (t, J = 6.7 Hz, 1 H), 3.05 – 2.94 (m, 4 H), 1.89 (s, 3 H);
Obi-COOH (retention time = 20.5 min), both isolated as brown oils. Obi:1H 13C NMR (151 MHz, methanol-d4) δ (p.p.m.) = 201.7, 170.3, 148.6, 148.2, 147.3,
NMR (500 MHz, methanol-d4) δ (p.p.m.) = 8.12 (d, J = 9.0 Hz, 2 H), 7.49
(d, J = 8.4 Hz, 2 H), 7.28 (d, J = 9.6 Hz, 1 H), 6.99 (d, J = 6.6 Hz, 1 H), 6.78
(t, J = 1.0 Hz, 1 H), 5.83 (d, J = 6.0 Hz, 1 H), 5.08 (td, J = 5.7, 9.0 Hz, 1 H), 3.40
(dd, J = 9.0, 15.0 Hz, 1 H), 3.21 (dd, J = 5.1, 14.7 Hz, 1 H); 13C NMR (126 MHz,
methanol-d4) δ (p.p.m.) = 171.8, 170.2, 150.5, 148.7, 147.8, 145.8, 131.6, 130.5,
147.2, 131.7, 124.4, 120.8, 120.2, 119.9, 72.7, 64.2, 41.6, 39.8, 29.4, 22.5; LC-MS
(ESI) calculated for C21H24N3O8S: 478.1 [(M+H)+], found: 478.1 (retention time
9.1 min). See Supplementary Note for tabulated NMR data and copies of spec-
tra. See Supplementary Figure 19 for LC–MS analysis of purified Obi-SNAC.
129.9, 128.1, 124.9, 120.8, 120.4, 119.5, 116.4, 79.2, 60.5, 36.8; HRMS (ESI) cal- Hydrolysis of Obi and Obi-SNAC to Obi-COOH. Solutions of purified Obi
culated for C17H15N2O7: 359.0874 [(M+H)+], found peak when extracting ions (1 mM) and Obi-SNAC (1 mM) were prepared in MES buffer at pH ~7. Aliquots
for 359.0858–359.0930. See Supplementary Note for tabulated NMR data and
copies of spectra. Obi-COOH:1H NMR (500 MHz, methanol-d4) δ (p.p.m.) =
8.15 (d, J = 8.4 Hz, 2 H), 7.50 (d, J = 8.4 Hz, 2 H), 7.37 (d, J = 9.6 Hz, 1 H),
6.98 (d, J = 9.0 Hz, 1 H), 6.78 (t, J = 8.1 Hz, 1 H), 4.75 (d, J = 2.4 Hz, 1 H), 4.54
(100μl)weretakenatvarioustimepointsanalyzedbyHPLC(gradientof5%Bto
95% B over 20 min, 95% B to 100% B over 3 min, and 100% B to 5% B over 2 min
at a flow rate of 1 ml/min) with detection by optical absorbance spectroscopy at
270 nm (Fig. 4b,c). HPLC peak identities were confirmed by LC–MS and reten-
(ddd, J = 2.1, 5.6, 8.0 Hz, 1 H), 3.03 (t, J = 1.0 Hz, 1 H), 2.96 (t, J = 1.0 Hz, 1 tion times were normalized using an Fmoc-Ala internal standard. Experiments
H); 13C NMR (126 MHz, methanol-d4) δ (p.p.m.) = 173.8, 170.7, 149.3, 148.3, were performed in duplicate.
147.8, 147.4, 131.8, 124.7, 124.6, 124.5, 120.2, 120.1, 120.0, 117.8, 73.3, 63.4,
57.7, 41.8; HRMS (ESI) calculated for C17H15N2O8: 375.0834 [(M+H)+], found Data availability. P. fluorescens ATCC 39502 sequencing data for the Obi bio-
peak when extracting ions for 375.0796–375.0872. See Supplementary Note synthetic gene cluster is deposited in GenBank (Accession numbers KX134682,
for tabulated NMR data and copies of spectra. See Supplementary Figure 19
for LC–MS analysis of P. fluorescens culture supernatant, purified Obi, and
purified Obi-COOH.
KX134683, KX134684, KX134685, KX134686, KX134687, KX134688,
KX134689, KX134690, KX134691, KX134692, KX134693, KX134694,
KX134695). All other data supporting the findings of this study are contained
in the published article (and its supplementary information files) or are avail-
Synthesis of Obi-SNAC. 1 mg purified Obi β-lactone was dissolved in 5 mg able from the corresponding author upon request.
SNAC. The solution was left at room temperature overnight, and then diluted
with 0.5 ml MeCN and purified by reverse phase preparatory HPLC (gradient
of 5% B to 95% B over 20 min at flow rate of 10 ml/min). Product-containing
53. Chanco, E., Choi, Y.S., Sun, N., Vu, M. & Zhao, H. Characterization of the
fraction (retention time 21.2 min) was concentrated by rotary evaporation, dis-
solved in 0.5 ml DMSO, and transferred to a 1.5 ml Eppendorf tube. Solution
was lyophilized overnight and dissolved in methanol-d4 for NMR analysis.
Obi-SNAC forms a 2:1 mixture of rotamers in methanol-d4. The NMR data for
the major rotamer is given below. 1H NMR (600 MHz, methanol-d4) δ (p.p.m.)
= 9.08 (d, J = 8.7 Hz, 1 H), 8.13 (d, J = 8.7 Hz, 1 H), 7.48 (d, J = 7.9 Hz, 1 H),
7.41 (d, J = 7.9 Hz, 1 H), 6.99 (d, J = 7.9 Hz, 1 H), 6.80 (t, J = 7.9 Hz, 1 H), 4.75
N-oxygenase AurF from Streptomyces thioletus. Bioorg. Med. Chem. 22,
5569–5577 (2014).
54. Quadri, L.E. et al. Characterization of Sfp, a Bacillus subtilis
phosphopantetheinyl transferase for peptidyl carrier protein domains in
peptide synthetases. Biochemistry 37, 1585–1595 (1998).
55. Hollenhorst, M.A., Clardy, J. & Walsh, C.T. e ATP-dependent amide ligases
DdaG and DdaF assemble the fumaramoyl-dipeptide scaffold of the
dapdiamide antibiotics. Biochemistry 48, 10467–10472 (2009).
nature CHeMICaL BIOLOGY
doi:10.1038/nchembio.2374