Dual phenazine gene clusters enable diversification during biosynthesis

Article abstract of DOI:10.1038/s41589-019-0246-1

Biosynthetic gene clusters (BGCs) bridging genotype and phenotype continuously evolve through gene mutations and recombinations to generate chemical diversity. Phenazine BGCs are widespread in bacteria, and the biosynthetic mechanisms of the formation of the phenazine structural core have been illuminated in the last decade. However, little is known about the complex phenazine core-modification machinery. Here, we report the diversity-oriented modifications of the phenazine core through two distinct BGCs in the entomopathogenic bacterium Xenorhabdus szentirmaii, which lives in symbiosis with nematodes. A previously unidentified aldehyde intermediate, which can be modified by multiple enzymatic and non-enzymatic reactions, is a common intermediate bridging the pathways encoded by these BGCs. Evaluation of the antibiotic activity of the resulting phenazine derivatives suggests a highly effective strategy to convert Gram-positive specific phenazines into broad-spectrum antibiotics, which might help the bacteria–nematode complex to maintain its special environmental niche.

Full text of DOI:10.1038/s41589-019-0246-1

Articles
https://doi.org/10.1038/s41589-019-0246-1
Dual phenazine gene clusters enable
diversification during biosynthesis
Yi-MingShiꢀ ꢀ1, AlexanderO.Brachmann1,3, Margaretha A.Westphalen1, NickNeubacher1,
NicholasJ.Tobias1 and HelgeB.Bodeꢀ ꢀ1,2*
Biosynthetic gene clusters (BGCs) bridging genotype and phenotype continuously evolve through gene mutations and recom-
binations to generate chemical diversity. Phenazine BGCs are widespread in bacteria, and the biosynthetic mechanisms of the
formation of the phenazine structural core have been illuminated in the last decade. However, little is known about the complex
phenazine core-modification machinery. Here, we report the diversity-oriented modifications of the phenazine core through
two distinct BGCs in the entomopathogenic bacterium Xenorhabdus szentirmaii, which lives in symbiosis with nematodes. A
previously unidentified aldehyde intermediate, which can be modified by multiple enzymatic and non-enzymatic reactions,
is a common intermediate bridging the pathways encoded by these BGCs. Evaluation of the antibiotic activity of the resulting
phenazine derivatives suggests a highly effective strategy to convert Gram-positive specific phenazines into broad-spectrum
antibiotics, which might help the bacteria–nematode complex to maintain its special environmental niche.
enes encoding biosynthetic pathways for natural products Gram-negative bacteria (for example, Pseudomonas) are able to
are frequently clustered in bacterial and fungal genomes. A carry out simple modifications, such as hydroxylation of the aro-
G
reason for that might be genome evolution driven by hori- matic ring, decarboxylation, N-methylation, and N-oxidation13.
zontal gene transfer in the producer1, thus allowing for simplified More complex core modifications abound in Streptomyces and
regulation via transcriptional units or co-localization2,3. Since the Kitasatospora, exemplified by C-/N-alkylation with isoprenyl
evolutionary speed of BGCs is closely linked to the rapid replication or terpene moieties, esterification with polyketides, fatty acids,
of their microbial hosts and the natural tendency toward horizontal amino acids, or monosaccharides, and radical dimerization13,14
.
transmission among prokaryotic and/or eukaryotic species4,5, BGCs Phenazines have various bioactivities, acting as virulence factors in
are considered the most diverse and expeditious evolving genetic Pseudomonas lung infections, promoting anaerobic survival, serv-
elements6. As a critical bridge between genotype and phenotype ing as intercellular signaling molecules in the quorum-sensing net-
(that is, the produced natural products), BGCs constantly evolve work, and having antimicrobial and cytotoxic activities13,17,18
.
through mutation within individual genes, duplication followed by
Although over 50 naturally occurring phenazine hybrids and
mutation, and alteration to the gene repertoire to fulfill chemical dimers have been discovered from a wide range of bacteria13,14, the
diversity and ultimately produce a functional molecule that justi- complex core-modification machinery has not been well explored
fies the BGC’s existence, thereby adapting potent forces of natural so far, except for two examples involving polyketide biochemis-
selection6,7. Mutations of active sites in domains or genes give rise try in saphenamycin19 and methyl-rhamnosylation in endophen-
to changes in substrate specificity as in modular assembly lines8 or azines20. Here, we present the structures and biosyntheses of diverse
functional capability, exemplified by (mero)terpenoid cyclases9 and phenazine derivatives from the entomopathogenic bacterium
oxygenases10 that diversify potential skeletal arrangements. Changes Xenorhabdus szentirmaii that lives in mutualistic symbiosis with
to the gene repertoire reflect nature’s capability to achieve innova- Steinernema rarum nematodes21. After synthesis of 1 through the
tion through a combinatorial manner of recruiting, mixing, and proteins encoded by a typical phz operon, 1 is diversified by the
matching genes, by which BGCs with recombined architecture tend enzymes from two discrete BGCs, including two monooxygenases,
to construct innovative patterns11 and hybrid scaffolds12.
a reductase, a methyltransferase, and a non-ribosomal peptide syn-
Phenazines are widespread in Gram-negative (for example, thetase (NRPS)-like module. The resulting antibiotic activity of the
Pseudomonas, Burkholderia, and Xanthomonas) and Gram-positive products suggests a protective function for insect cadavers in the
(for example, Streptomyces, Brevibacterium, and Kitasatospora) ecological niche of the bacteria–nematode–insect interaction.
bacteria, as well as in the archaeal genus Methanosarcina13,14. A
conserved operon, phzABCDEFG (Fig. 1), encodes enzymes that Results
catalyze dimerization and subsequent oxidation of chorismic acid- Identification of the phenazine gene clusters. From an antiS-
derived 6-amino-5-oxocyclohex-2-en-1-carboxylic acid to afford MASH22 analysis of 28 Xenorhabdus and Photorhabdus genome
a basic tricyclic planar scaffold, phenazine-1,6-dicarboxylic acid sequences23, we identified four strains encoding phenazine BGC(s),
(PDC, 1) or phenazine-1-carboxylic acid (PCA, 2)15,16. These phen- with X. szentirmaii as the only strain encoding two phenazine BGCs
azine structural cores can be further modified at the electron-rich (xpz; Fig. 1 and Supplementary Table 1). The first xpz BGC contains
aromatic rings, the nitrogen(s) with a lone pair in the reduced form a typical phz operon (xpzABCDETF; Fig. 1) which is responsible for
of phenazine, and the carboxylic acid, generating various derivatives. the biosynthesis of phenazine structural cores 1 and 2 (Fig. 2) in
1Molekulare Biotechnologie, Fachbereich Biowissenschaften, Goethe Universität Frankfurt, Frankfurt am Main, Germany. 2Buchmann Institute for
Molecular Life Sciences, Goethe Universität Frankfurt, Frankfurt am Main, Germany. 3Present address: Institute of Microbiology, Eidgenössische
Technische Hochschule Zurich, Zurich, Switzerland. *e-mail: h.bode@bio.uni-frankfurt.de
NATURe CHeMiCAl BiOlOgY | www.nature.com/naturechemicalbiology

NATurE CHEmiCAl BioloGy
Articles
phz
M
G1
A1 B1
C1
D1
E1
F1
S
1st
xpz
T
A
B
C
PBAD
D
E
V
N
G
H
U
F
PBAD
PBAD
xpz
2nd
A2
Q
R
I
J
K
L
M
O
P
S
ehp
R
N
A
B
C
D
E
F
G
H
I
J
K
L
M
O
0 kb
5 kb
10 kb
15 kb
Phenazine core biosynthesis
NRPS-like
Iodinin biosynthesis
Pyocyanin biosynthesis
Griseoluteic acid biosynthesis
Resistance
Others
Fig. 1 | Comparison of the phenazine BgCs in Pseudomonas aeruginosa PAO1 (phz), X. szentirmaii (xpz), and Pantoea agglomerans eh1087 (ehp). Similar
genes are connected with dashed lines. The second xpz BGC is silent under laboratory conditions. A red arrow shows the position where the putative
natural promoter is exchanged by the arabinose-inducible promoter PBAD. kb, kilobase.
Pseudomonas strains18. The phenazine core biosynthetic genes are we detected a larger molecule with an m/z of 368 [M+H]+. Since
flanked by three genes encoding monooxygenases (XpzUGH), of XpzPQS is homologous to EhpMNO, which is responsible for incor-
which XpzU is homologous to an antibiotic biosynthesis monooxy- porating a d-alanine to griseoluteic acid in Pantoea agglomerans27,
genase, XpzG belongs to a putative type I Baeyer–Villiger monooxy- we expected the product to be a phenazine–peptide hybrid, but
genase harboring a fingerprint motif sequence FXGXXXHXXWP24, failed to isolate it due to its instability and the low production level
and XpzH is homologous to an NAD(P)/FAD-dependent oxidore- in E. coli.
ductase (Fig. 1).
The second xpz BGC is located 1.05 megabase pairs (Mb) away Oxidatively modified phenazines in the wild-type strain. To ver-
from the first and encodes a set of putative griseoluteic acid syn- ify if phenazines expressed in E. coli are the native products of the
thases (XpzJKLMNO) and an NRPS-like module with free-standing clusters, we compared the phenazine production in the native host
adenylation (A, XpzP), thiolation (T, XpzQ), and FabH family25 con- with that in E. coli. The wild-type strain shows brownish colonies on
densation (XpzS) proteins (Fig. 1). The remaining second xpz BGC LB media agar plates (Supplementary Fig. 2). HPLC–UV–MS analy-
is made up of xpzR that encodes a glyoxalase I/bleomycin resistance sis of methanol/acetone extracts (Fig. 3a) from cultures grown in
protein homolog26 and two phz homologs, xpzA2 and xpzI. The M63 minimal medium31 revealed four peaks (1, 2, 5, and 6) identi-
second xpz BGC, encoding enzymes XpzJKLMNOPQRS, is highly cal to E. coli harboring xpzABCDETF and xpzGH as well as two new
similar to the ehp cluster from Pantoea agglomerans Eh1087 (ref. 27), peaks from the native host identified to be 6-hydroxyphenazine-
a producer of 1, 2, griseoluteic acid, and d-alanylgriseoluteic acid. 1-carboxylic acid (3) and 1,6-dihydroxyphenazine (4) by HRMS
A previous report assigned the ehp genes to four functional groups (Fig. 2 and Supplementary Table 5). Given that the differences
and proposed conversions from 1 to d-alanylgriseoluteic acid27.
among compounds 1–6 were observed at C-1/C-6 and N-5/N-10,
To verify the putative function of two xpz BGCs, we heterolo- we postulated that two monooxygenases (XpzGH) were sufficient
gously expressed the phenazine core biosynthetic genes xpzABC- to fulfill the oxidation of multiple symmetric carbons and nitrogens.
DETF in combination with postulated core-modification xpz genes To determine the roles of xpzGH in forming 3–6, we constructed
inEscherichia coli BL21(DE3)(SupplementaryFig. 1). Co-expression a ∆xpzGH deletion strain that abolished the brownish phenotype
of the core genes xpzABCDETF with the monooxygenases encoding (Supplementary Fig. 2) and the production of compounds 3–6
genes xpzUGH or xpzGH yielded 1 and 2, together with two oxi- (Supplementary Fig. 3). Complementation with xpzG alone restored
datively modified products, 1,6-dihydroxyphenazine-5-oxide (5) the production of 3 and 4, while complementation with xpzGH
and the magenta pigment iodinin (6; Fig. 2)28,29, as determined by additionally restored the brownish colonies and production of 5
high-resolution MS (HRMS) (Supplementary Table 5). As omission and 6, suggesting that these compounds contribute to the brownish
of XpzU had no effect on the product profile, these results suggest phenotype in the wild-type colonies.
that only xpzGH are required for the oxidations. E. coli encoding the
For a more detailed analysis of XpzG and XpzH, we reconstituted
core genes and xpzRA2IJKLMNO produced demethoxygriseoluteic the biosynthetic pathways by in vivo experiments (Supplementary
acid (8; Fig. 2) and griseoluteic acid (9; Supplementary Fig. 1), both Fig. 4). E. coli harboring xpzG or xpzGH converted 1 to 4, or 1 to
of which were first reported from Streptomyces griseoluteus30. With 5 and 6, respectively, while no conversion of 1 was observed when
an additional NRPS-like module XpzPQS (Supplementary Fig. 1), incubated with E. coli expressing only xpzH. These observations
NATURe CHeMiCAl BiOlOgY | www.nature.com/naturechemicalbiology

NATurE CHEmiCAl BioloGy
Articles
Chorismic acid
XpzG
XpzH
XpzABCDETF
In vitro
O
OH
OH
COOH
COOH
COOH
OH
N
N
N
N
N
N
N
N
XpzG
XpzH
+
+
+
In vivo
In vitro
In vivo
In vitro
N
N
N
N
2
1
3
4
OH
OH
O
COOH
XpzN
OH
OH
O
5
6
Spontaneous
COOH
COOH
COOH
XpzL?
N
N
N
N
N
Non-enzymatic
In vitro
N
12
21
7
COOMe
CHO
16 R = OH
17 Δ1',12
1'
-H2O
R
O
Spontaneous
XpzN
Non-enzymatic
In vitro
COOH
COOH
N
COOH
COOH
O
N
N
N
N
XpzLO
XpzPQS
In vivo
N
N
N
18
22
8
COOMe
R
OH
O
10 i-Pr
11 i-Bu
14* s-Bu
H2N
O
O
XpzLO
XpzLO
R
COOH
O
COOH
O
COOH
O
COOH
O
N
N
N
N
N
XpzPQS
?
In vivo
N
N
N
20
9
19
R
O
OH
OH
O
12 i-Pr
13 i-Bu
15* s-Bu
OH
H2N
O
O
R
Fig. 2 | Structures and biosynthetic pathway of phenazines from X. szentirmaii. The asterisks indicate compounds resulting from the feeding experiment.
suggest that XpzG appears to catalyze Baeyer–Villiger oxidations with the expression of the first. Among the structural xpz genes, the
of 1 at C-1/C-6, followed by spontaneous decarboxylations to gen- most highly expressed are xpzFGH, which suggests a rapid conver-
erate 4 that is a putative substrate of XpzH for subsequent N-5/N-10 sion of 1 into 6 as is also evident from the brownish pigmentation
oxidations to afford the final product 6. We subsequently confirmed phenotype of the wild-type strain.
this conversion in vitro using recombinant XpzG and XpzH pro-
To mimic the natural growth condition of the bacteria, which
teins (Fig. 4a), both of which contain a conserved flavin-binding might induce a change in expression of BGCs and result in produc-
domain GXGXXG32, non-covalently bound to a flavin mononucleo- tion of additional phenazines, we injected X. szentirmaii into lar-
tide cofactor (Supplementary Fig. 5). Incubation of XpzG with 1 and vae of the model insect Galleria mellonella and carried out a time
NADPH (or NADH) as a cofactor showed the formation of 4 via 3 course analysis of phenazine production. Besides the products 1
(Supplementary Fig. 6a,b). Subsequent treatment of 4 with XpzH in and 3–6, which arise from expression of the first xpz BGC (Fig. 3a
the presence of NADPH yielded 5 and 6 (Supplementary Fig. 6c). and Supplementary Fig. 7), the second BGC was also activated
Unlike XpzG accepting both NADPH and NADH, XpzH has a pref- (Fig. 3a,c) leading to the production of 8 and the major derivative 9,
erence for NADPH. When NADH was used, XpzH converted 4 only together with two compounds possessing larger masses (m/z 384
to 5 (Supplementary Fig. 6c).
[M+H]+ and 398 [M+H]+) at day four post-infection. The two
larger compounds with characteristic phenazine UV absorptions30
Activation of the silent second xpz gene cluster. No phenazine were postulated to be phenazine–peptide hybrids. Inspired by the in
derivatives derived from the second xpz BGC were detected in vivo activation of the silent cluster in infected insects, we attempted
the native host (Fig. 3a), suggesting this BGC to be silent under to mimic the natural growth environment in vitro by using LB
laboratory conditions. To obtain more information about the tran- medium with added insect homogenate and Galleria instant broth
scription of both xpz BGCs, we performed RNA sequencing of the (Fig. 3a) that is made of dead insects33. Although both insect media
wild-type strain under normal growth condition in LB medium. were shown to induce the expression of the silent cluster, only low
The results show that both BGCs have different expression levels in titers of 8 and 9 were produced and were unsuitable for scale-up to
the exponential growth phase (Fig. 3b and Supplementary Table 6). use for isolation and structural elucidation of the unknown phen-
The second xpz BGC is expressed at much lower levels compared azine derivatives.
NATURe CHeMiCAl BiOlOgY | www.nature.com/naturechemicalbiology

NATurE CHEmiCAl BioloGy
Articles
a
b
5
4
3
2
1
0
2
`
3
5
1
4
6
(i)
9
12
13
8
(ii)
xpz genes
c
9
100
80
60
40
20
0
8
9
12
13
(iii)
(iv)
9
8
1
2
3
4
5
5.25
5.75
6.25
6.75
7.25
Time (min)
Time (days)
d
2
8+12
10
17
WT strain
extracts
P
xpzI mutant
BADextracts
9+13+16
3
1
11
+
20
6
5
(v)
(vi)
18
19
21
22
6.5
4
5.0
5.5
6.0
7.0
7.5
8.0 Time (min)
Fig. 3 | Activation of the silent xpz biosynthetic gene cluster. a, HPLC–MS analysis of X. szentirmaii wild-type strain with respect to phenazine production
in different media. Extracted ion chromatograms (EICs) of phenazines produced by the wild-type strain in M63 medium (i); injected to G. mellonella
larvae at day four post-infection (ii); in LB medium with G. mellonella homogenate (iii); and in Galleria instant broth, in which the EIC intensities of 8 and
9 were magnified tenfold (iv). b, Transcriptional analysis [log10(read counts)] of xpz genes (see Supplementary Table 6 for actual values). c, Time course
analysis for phenazine production from the second xpz BGC in G. mellonella larvae infection assays. The production of individual phenazines at each
time point is relative (in percentage, %) to its total production during a five-day time course (the total production sums to 100%). d, HPLC–UV analysis
(detection wavelength 350–800ꢀnm) of the wild-type strain and the constructed mutant in which a PBAD promoter was introduced in front of xpzI and was
induced by arabinose. Methanol/acetone extracts with magenta pigment produced by the wild-type strain (v) and with brown pigment produced by the
induced PBAD xpzI mutant (vi). In a and d, representative data from three independent experiments are shown. c, Data represent meanꢀ ꢀs.d. from three
independent experiments.
We then applied a promoter-exchange strategy to replace the Compound 12 is pelagiomicin B and 10, 11, and 13 represent new
natural promoter region of xpzR, xpzA2, or xpzI with an arabinose- phenazine derivatives that we designate as phenaszentines A–C.
inducible PBAD promoter34 (Fig. 1). The PBAD xpzI mutant was the Additionally, two new hybrids, phenaszentines D (14) and E (15)
most efficient (Supplementary Fig. 8), yielding a brownish culture (Fig. 2), were produced when the culture was fed with l-isoleucine
and additional peaks showing typical phenazine UV absorptions (Supplementary Fig. 8). To support the putative role of the NRPS-
with three UV bands at 205–220, 250–290, and 350–370nm30 in like module (XpzPQS) in installing the amino acid to the struc-
the HPLC profile as compared with those of the wild-type strain tural core, we constructed a PBAD xpzI mutant carrying a ∆xpzPQS
(Fig. 3d). In addition to 8 and 9, four phenazine–peptide hybrids deletion that abolished the production of 10–13 (Supplementary
(10–13; Fig. 2) with 8 or 9 as a structural core were found, as identi- Fig. 12).
fied from the characteristic losses for valine and leucine/isoleucine
in their MS/MS fragmentation patterns. The structures of 10–13 Phenazine–polyketides via non-enzymatic condensation. On
were identified by stable isotope labeling (Supplementary Figs. 8 closer examination of the HPLC–UV profiles of PBAD xpzI and PBAD
and 9) and in vivo conversion experiments (Supplementary Fig. 10), xpzI ∆xpzPQS mutants, in addition to 1–13, we noticed some other
chemical synthesis (Supplementary Note), and NMR methods peaks with phenazine UV absorptions present in both of the mutants
(Supplementary Fig. 11 and Supplementary Note), revealing that 10 but absent in the wild-type strain (Fig. 3d and Supplementary
and 12 carry an l-valine while 11 and 13 possess an l-leucine. Fig. 12). Due to their low production titers, a 12-liter fermentation
NATURe CHeMiCAl BiOlOgY | www.nature.com/naturechemicalbiology

NATurE CHEmiCAl BioloGy
Articles
O2
CO2
O2
H2O
a
XpzG
XpzH
NADPH or NADH
Tris buffer
O2
CO2
O2
CO2
pH 8.5, RT
O
N
COOH
OH
OH
OH
COOH
N
N
N
N
XpzG
XpzH
+
+
NADPH or NADH
Tris buffer
pH 8.5, RT
NADPH
Tris buffer
pH 8.5, RT
N
N
N
N
N
1
3
4
OH
OH
OH
OH
COOH
O
O
5
6
b
CO2
MeOH
COOH
COOH
COOH
O
O
N
N
N
N
N
Non-enzymatic
+
OMe
Tris buffer or diisopropylamine
pH 9.0, 30 °C
N
7
O
HO
16
17
O
O
c
CO2
MeOH
COOH
COOH
O
O
N
N
N
Non-enzymatic
+
OMe
Tris buffer or diisopropylamine
pH 9.0, 30 °C
N
17
18
O
O
Fig. 4 | in vitro characterization of iodinin and phenazine–polyketide conversions. a, Enzymatic conversion of PDC (1) to iodinin (6) using the
recombinant monooxygenases XpzG and XpzH. RT, room temperature. b, Non-enzymatic polyketide chain extension from the active aldehyde
intermediate (7) via an aldol condensation under basic conditions. c, Non-enzymatic six-membered ring annulation via a Michael addition and an aldol
condensation under basic conditions.
of PBAD xpzI ∆xpzPQS mutant was performed, allowing the isolation of 7, we incubated the culture with an aldehyde capture reagent,
of seven additional yellow or orange phenazine derivatives (16–22; O-(2,3,4,5,6-pentafluorobenzyl)hydroxylamine (PFBHA)38. Two
Fig. 2) including the known phencomycin (21)35 and six new com- new aldoxime ethers (29 and 30), as determined by their HRMS
pounds that we named phenaszenketides A–D (16–19) and phen- data (Supplementary Table 5), were found in all PBAD xpzI mutants
cocins A (20) and B (22). The structures of all compounds were that are supposed to generate the aldehyde intermediate, except for
confirmed by HRMS (Supplementary Table 5) and 1D and 2D NMR the PBAD xpzI ∆xpzN mutant (Supplementary Fig. 13a). Additionally,
data (Supplementary Note and Supplementary Fig. 11).
a PBAD xpzI mutant fed with PFBHA gave rise to lower production
Compounds 16–19 have a PDC core that is modified by titers of 8–20 compared with the control culture (Supplementary
polyketide moieties, such as β-hydroxybutanone (16), butenone (17 Fig. 13b), supporting 7 as a central intermediate in the divergent
and 19), or β-methylcyclohexenone (18), through a carbon–car- pathways.
bon connection. However, in contrast to the saphenamycin BGC in
To investigate the non-enzymatic formation of 16, we initially
Streptomyces antibioticus Tü 2706, which contains a polyketide syn- attempted to obtain 7 by 2-iodoxybenzoic acid oxidation of the
thase involved in the one-carbon extension of PDC19, no polyketide synthesized 8 (Supplementary Note). However, after purification
synthase candidate genes are present in the xpz BGCs. Given the we observed the rapid conversion of 7 to 1, as well as spontaneous
reactivity of aldehyde intermediates for the spontaneous formation oxidation of 8 to 7 and 1, in solution at room temperature (Fig. 2
of C–C and C–N bonds, as exemplified by the jadomycin aglycone36 and Supplementary Fig. 14). Due to this reactivity of 7, the mix-
and discoipyrrole core37, we speculated that 6-formylphenazine- ture was incubated with methyl acetoacetate in Tris buffer at pH
1-carboxylic acid (7) might act as an intermediate to react with an 9.0 to simulate the basic environment of the culture medium. As
acetoacetyl-CoA via a non-enzymatic aldol condensation.
expected, under these conditions we observed two products with
To verify 7 as a phenazine–polyketide hybrid precursor, we identical retention time and MS/MS fragmentation patterns as 16
individually deleted the reductase-encoding genes xpzK and xpzN and 17 (Fig. 4b and Supplementary Fig. 15)
and determined whether they are involved in converting 1 to its
The β-methylcyclohexenone moiety of 18 could be annulated via
alcohol 8 via intermediate 7. While the obtained PBAD xpzI ∆xpzK a Michael addition and an aldol condensation of 17 with a second
and PBAD xpzI mutants shared an identical metabolite profile and molecule of acetoacetyl-CoA, followed by dehydration and decar-
pigmentation phenotypes, 8–20 were completely absent in the PBAD boxylation (Supplementary Fig. 16a). A similar process of ring
xpzI ∆xpzN mutant (Supplementary Fig. 12), pointing to the essen- annulationhasbeendescribedforcyclohexanediones39,aclassofquo-
tial role of xpzN in governing the two-step reduction of 1 for the rum-sensing signaling molecules produced in Photorhabdus lumine-
downstream biosynthetic pathway (Fig. 2). To validate the existence scens TT01 (Supplementary Fig. 16b). However, while the unusual
NATURe CHeMiCAl BiOlOgY | www.nature.com/naturechemicalbiology

NATurE CHEmiCAl BioloGy
Articles
a
Robinson annulation. Indeed, the spontaneous occurrence of 17
annulation under basic conditions was demonstrated by in vitro
assays (Fig. 4c and Supplementary Fig. 17), and the identification of
three intermediates by HRMS (Supplementary Table 5) supported
the proposed reaction mechanism (Supplementary Fig. 17d). We
also searched for other non-enzymatic products, such as Schiff bases
formed from 7 and metabolites containing primary amino group
(for example, amino acids or peptides), but no additional products
were identified on the basis of the characteristic UV absorption of
phenazines and MS/MS data. This suggests that although the non-
enzymatic formation of 16–18 was observed in in vitro assays,
their exclusive production in vivo might imply the involvement of
enzymes that are not identified yet.
The methoxy group derivative 19 might be formed from 17,
analogous to the conversion of 8 to 9 catalyzed by both XpzO
(homologous to naphthocyclinone hydroxylase) and XpzL (homol-
ogous to SAM-dependent methyltransferase). Compound 20,
with a glycerol moiety, appears to be an artifact since it was only
observed in glycerol-containing medium (Supplementary Fig. 18).
Compounds 21 and 22 represent O-methylation and/or methoxyl-
ation derivatives of the PDC core, which could also be catalyzed by
XpzL and/or XpzO.
Q
xpz
P
S
A
T
KS
COOH
R
COOH
R
N
N
N
N
S
H2N
O
R'
X
X
H2N
R'
O
Val i-Pr
Leu i-Bu
Ile s-Bu
R'
X
R
H
OMe
H
X
R
R'
8
9
OH
OH
10
11
12
13
14
15
24
O
O
O
O
O
O
NH
H
H
i-Pr
i-Bu
OMe i-Pr
OMe i-Bu
23 NH2
H
H
s-Bu
s-Bu
H
H
i-Pr
i-Bu/s-Bu
25/26 NH
b
0.7
0.6
0.5
XpzPQS as an ester- and amide-forming NRPS-like module. The
free-standing A (XpzP), T (XpzQ), and FabH family condensation
(XpzS) proteins constitute a minimal NRPS elongation module.
The transformation of 8 and 9 to 10–13 by an E. coli strain har-
boring xpzPQS implied that XpzS is involved in attaching an amino
acid to the structural core through an ester bond (Fig. 5a and
Supplementary Fig. 19). Inspired by the ester-bond formation, we
examined whether XpzS also retains amide bond-forming capa-
bility. The incubation of E. coli expressing xpzPQS with synthetic
6-(aminomethyl)phenazine-1-carboxylic acid (23; Supplementary
Note) as a nucleophilic substrate yielded three new phenazine–pep-
tide hybrids (24–26) consisting of valine, leucine, and isoleucine
bound to 23 (Fig. 5a and Supplementary Fig. 20). These findings
motivated us to pursue further acceptor substrate flexibility of XpzS.
However, the use of carboxylic acid methyl ester substrates (37 and
40; Supplementary Fig. 21a) failed to generate any peptide hybrid,
pointing to the negatively charged carboxylic group being an essen-
tial interaction site (Supplementary Fig. 21b) that might interact
with a residue possessing a positively charged side chain in the
donor site of XpzS. Simplified substrates containing one or two ben-
zene rings (Supplementary Fig. 21a) were also not modified by an
amino acid. This suggests that the rigid, flat tricyclic aromatic ring
could contribute to a π-stacking interaction with residues having
an aromatic moiety (Supplementary Fig. 21b), with a binding mode
analogous to that of griseoluteic acid interacting with the phenazine
resistance protein EhpR26.
0.4
0.3
0.2
0.1
0.0
–0.1
–0.2
–0.3
5
10
15
20
25
30
0
Time (h)
xpzPQS + 8 (induced)
xpzPQS + 8 (non-induced)
xpzPQS + 23 (induced)
xpzPQS + 9 (induced)
xpzPQS + 9 (non-induced)
xpzPQS
xpzPQS + 23 (non-induced)
c
R'
COOH
O
N
N
H2N
27
28
O
R'
O
H2N
Fig. 5 | Characterization of XpzPQS as free-standing A, T, and FabH family
condensation (KS) enzymes catalyzing phenazine–peptide antibiotics
formation via ester and amide bonds. a, In vivo characterization of the
condensation between amino acids and demethoxygriseoluteic acid (8),
griseoluteic acid (9), and 6-(aminomethyl)phenazine-1-carboxylic acid
(23) fed to E. coli BL21(DE3) expressing xpzPQS (Supplementary Figs. 19
and 20). b, Growth curves [log10(OD)] of E. coli BL21(DE3) harboring
pCOLADuet-1 xpzPQS with added 8, 9, or 23 in LB media with and without
isopropyl β-d-1-thiogalactopyranoside (IPTG) induction. Data represent
meanꢀ ꢀstandard deviation from three independent experiments.
c, Chemically synthesized phenazine–peptides (27 and 28). A, adenylation;
T, thiolation; KS, ketoacylsynthase.
Examples of FabH family enzymes (also known as 3-oxoacyl-
ACP synthases III belonging to the ketoacylsynthase family)
catalyzing ester-bond formation have been described in the bio-
synthesis of cervimycins41, xenocyloins42, and nonactin43. XpzS is
a case of a FabH family enzyme accepting carrier protein-tethered
amino acids as substrates and catalyzing both ester- and amide-
bond formation. Phylogenetic analysis indicates that XpzS is
homologous to 3-oxoacyl-ACP synthases III and falls within a
clade of proteins that have not been biochemically characterized
(Supplementary Fig. 22). Moreover, a multiple sequence align-
ment (Supplementary Fig. 23) indicated a low sequence identity
(13.9%) of XpzS with FabH. In particular, a closer inspection of
the conserved catalytic triads (Cys-His-Asn) revealed that XpzS
lacks a second histidine residue that is responsible for decarbox-
ylation of a malonyl-ACP to generate a carbanion for subsequent
ketosynthase, StlD, catalyzes the annulation in Photorhabdus39,40
,
no StlD homolog exists in the X. szentirmaii genome, and Claisen condensation44. Although XpzS harbors identical func-
thus we envisaged 17 more likely to undergo a non-enzymatic tions to SgcC5, which is the only example of a C domain in NRPS
NATURe CHeMiCAl BiOlOgY | www.nature.com/naturechemicalbiology

NATurE CHEmiCAl BioloGy
Articles
activity, we modified the structural core of 9 by l-alanine (27; as a
control) and β-alanine (28; Fig. 5c). The β-alanine residue pos-
sesses a positively charged terminal amino group that is a pharma-
cophore of antimicrobial cationic peptides51 able to potently bind
to the negatively charged phosphate-containing elements in bac-
terial cells through electrostatic interactions. Replacement of the
native amino acid residue by an l-alanine as in 27 showed simi-
lar inhibition to 12 and 13, except for the loss of activity against
Enterococcus faecalis, while a β-alanine residue as in 28 resulted
in an increased inhibition of Micrococcus luteus, but narrowed the
antibiotic spectrum (Table 1).
Table 1 | Antibiotic activity of phenazines against selected
gram-positive and -negative bacteria
Compound Bs
Ml Sa
1.5
efs eh efm Ab ec
Kp Se
3
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
6
1.5 1.5
–
–
–
10
11
12
13
27
28
1.3
–
–
0.9
2.7 2.5 2.5 0.8 2.4 2.3 0.8 0.8 1.0 0.8
2.7 2.5 2.5 1.4 2.5 2.5 1.0 0.9 1.3 1.0
2.5 2.6 2.2
2.0 3.2 2.0
–
2.3 2.3 1.3 1.3
2.2 2.3
1.1 0.9
1.4 1.1
Discussion
In contrast to precise target-oriented biosynthesis in primary meta-
bolic pathways, secondary metabolic pathways in microorgan-
isms and plants are often more diversity oriented52. Such a strategy
enables efficient synthesis of compound libraries deriving from a
simple, common starter. By doing so, natural product producers
might improve the hit rate of gaining biologically important mol-
ecules that are beneficial to the producer itself. X. szentirmaii is
shown here to be a proteobacterium with BGCs encoding enzymes
–
–
–
Numerical values indicate the diameters of inhibition zones (cm) observed in disc diffusion
assays. Data represent nꢀ=ꢀ1 experiment; see Supplementary Fig. 27 for pictures of the assay.
Bs, Bacillus subtilis; Ml, Micrococcus luteus; Sa, Staphylococcus aureus; Efs, Enterococcus faecalis;
Eh, Enterococcus hirae; Efm, Enterococcus faecium; Ab, Acinetobacter baumannii; Ec, Escherichia coli;
Kp, Klebsiella pneumoniae; and Se, Salmonella enterica subsp. enterica.
systems capable of forming both ester and amide bonds45, it is for both oxidative and complex core-phenazine modifications.
unsurprising that the sequence of XpzS is distinct from that of Importantly, the previously unidentified highly reactive aldehyde 7
other C domains (Supplementary Fig. 24).
serves as a critical branching point that not only bridges two xpz
BGCs for downstream enzymatic modifications to construct the
Antibiotic activity of phenazines. When xpzPQS were expressed phenazine–peptides, but also enables additional non-enzymatic
in E. coli fed with 8 or 9 for in vivo conversion, reduced growth was reactions leading to a new set of phenazine–polyketides. The newly
observed (Fig. 5b) concurrent with detection of phenazines 10–13 identified phenazine–peptide and polyketide hybrids, as well as
(Supplementary Fig. 19). This coincides with the previous descrip- derivatives with variable patterns from X. szentirmaii, expand the
tions of phenazine–peptide hybrids as antibiotics against Gram- phenazine repertoire and showcase the abilities of an under-appre-
negative bacteria46. Earlier reports suggest that a 40-fold lower ciated producer. More importantly, modifications to the planar aro-
antibiotic activity of 9 compared with its peptide hybrid is due to matic phenazine structural core introduce a moiety (for example,
a poorer uptake of 9 (ref. 47). With the synthetic 8 in hand, we per- polyketides or amino acids) bearing stereochemistry and carbo-
formed uptake and conversion studies and observed that 8 is indeed genic sp3 content, which could enhance bioactive potency and selec-
taken up by E. coli (Supplementary Fig. 25a), without a growth inhi- tivity53 as in the case of 12 and 13, although the functions of some of
bition effect (Fig. 5b). E. coli harboring xpzPQS converts 8 into 10 these hybrids are unknown.
and 11 in the cell, which are subsequently exported (Supplementary
In the context of bacteria–nematode–insect interactions, the role
Fig. 25b). Additionally, when E. coli was incubated with phenaszen- of phenazine-producing bacteria remains mostly unknown. We pos-
tine A (10), we observed that the amount of extracellular 8 increases tulate that once inside insects, X. szentirmaii produces and secretes
while the amount of 10 decreases (Supplementary Fig. 25c). This a broad-spectrum phenazine antibiotic cocktail that is made of
suggests that, along with suppressing the growth of E. coli (Fig. 5b), cytotoxic 6 and various antibiotics with both specific activity against
the active form 10 might be transformed into the inactive metabo- Gram-positive bacteria (for example, 3, 6, and 9)13,50 and with
lite 8. Therefore, we conclude that an amino acid moiety is indis- broad-spectrum activity (for example, 10–13). This would protect
pensable for inhibiting E. coli. This is not because of a change in the insect cadaver from competing bacteria in the soil and therefore
physical properties, but presumably because the primary amine help maintain the lifecycle of its nematode host within the cadaver.
(pKa≈9.6) of a valine/leucine residue being positively charged Furthermore, phenazines might have additive or synergistic effects
under normal physiological conditions (pH≈7.0) might guide in the defensive system, together with fabclavines that have been
binding to negatively charged phosphate groups in the DNA, result- previously described as broad-spectrum antibiotics from X. szen-
ing in DNA-damaging activity47. A similar mechanism of action has tirmaii54. The observation of phenazine–peptide formation implies
been identified in bleomycins whose C-terminal cation provides that the approach of producing simply modified hybrids could be a
major DNA binding affinity48. Further exploration of efflux pumps highly effective strategy for the producer to convert Gram-positive-
conferring phenazine resistance in Gram-negative bacteria revealed specific phenazines like 9 (ref. 50) into broad-spectrum antibiotics
an essential role of the AcrAB–TolC complex49 on phenazine export (Table 1). Recently, four BGCs with diverse architectures that poten-
(Supplementary Fig. 26).
tially encode the biosynthesis of phenazine–peptides have been
We subsequently screened most of the isolated compounds identified from various actinobacteria and gammaproteobacteria55,
for antibiotic activity. Compounds 3 and 6 produced by the first suggesting such a strategy to be more widespread than we initially
xpz BGC showed specific activity against Gram-positive bacteria thought.
(Table 1). The peptide hybrids 10 and 11 from the second xpz BGC
were active against Bacillus subtilis (Table 1) and E. coli (Fig. 5b). Online content
Notably, 12 and 13, both derived from 9 that was reported to be Any methods, additional references, Nature Research reporting
active against Bacillus and Staphylococcus strains13,50, exhibited summaries, source data, statements of data availability and asso-
potent and broad-spectrum antibiotic activity (Table 1). Compared ciated accession codes are available at https://doi.org/10.1038/
with 10 and 11, a methoxy substituent at C-9, as in 12 and 13, s41589-019-0246-1.
also substantially increased the potency and expanded their spec-
trum. Motivated by the preliminary structure–activity relation-
ship in which the amino acid residue is essential for antibiotic
Received: 24 June 2018; Accepted: 13 February 2019;
Published: xx xx xxxx
NATURe CHeMiCAl BiOlOgY | www.nature.com/naturechemicalbiology

NATurE CHEmiCAl BioloGy
Articles
29. Turner, J. M. & Messenger, A. J. in Advances in Microbial Physiology Vol. 27
(eds A. H. Rose & D. W. Tempest) 211–275 (Academic Press, 1986).
30. Yagishita, K. Production of phenazine compounds by Streptomyces
griseoluteus. J. Antibiot. A 13, 83–96 (1960).
References
1. Lawrence, J. Selfsh operons: the evolutionary impact of gene clustering in
prokaryotes and eukaryotes. Curr. Opin. Genet. Dev. 9, 642–648 (1999).
2. Mao, D., Bushin, L. B., Moon, K., Wu, Y. & Seyedsayamdost, M. R. Discovery
of scmR as a global regulator of secondary metabolism and virulence
in Burkholderia thailandensis E264. Proc. Natl Acad. Sci. USA 114,
E2920–E2928 (2017).
31. Ghoul, M., Bernard, T. & Cormier, M. Evidence that Escherichia coli
accumulates glycine betaine from marine sediments. Appl. Environ. Microbiol.
56, 551–554 (1990).
32. Dym, O. & Eisenberg, D. Sequence-structure analysis of FAD-containing
proteins. Protein Sci. 10, 1712–1728 (2001).
3. Wang, Z. & Cirino, P. C. New and improved tools and methods for enhanced
biosynthesis of natural products in microorganisms. Curr. Opin. Biotechnol.
42, 159–168 (2016).
4. Koonin, E. V., Makarova, K. S. & Aravind, L. Horizontal gene transfer in
prokaryotes: quantifcation and classifcation. Annu. Rev. Microbiol. 55,
709–742 (2001).
5. Keeling, P. J. & Palmer, J. D. Horizontal gene transfer in eukaryotic evolution.
Nat. Rev. Genet. 9, 605–618 (2008).
6. Fischbach, M. A., Walsh, C. T. & Clardy, J. Te evolution of gene collectives:
how natural selection drives chemical innovation. Proc. Natl Acad. Sci. USA
105, 4601–4608 (2008).
7. Medema, M. H., Cimermancic, P., Sali, A., Takano, E. & Fischbach, M. A.
A systematic computational analysis of biosynthetic gene cluster evolution:
lessons for engineering biosynthesis. PLoS Comput. Biol. 10, e1004016 (2014).
8. Cai, X. et al. Entomopathogenic bacteria use multiple mechanisms for
bioactive peptide library design. Nat. Chem. 9, 379–386 (2017).
9. Yoshikuni, Y., Ferrin, T. E. & Keasling, J. D. Designed divergent evolution of
enzyme function. Nature 440, 1078–1082 (2006).
10. Nakashima, Y. et al. Structure function and engineering of multifunctional
non-heme iron dependent oxygenases in fungal meroterpenoid biosynthesis.
Nat. Commun. 9, 104 (2018).
11. Pan, G. et al. Discovery of the leinamycin family of natural products by
mining actinobacterial genomes. Proc. Natl Acad. Sci. USA 114,
E11131–E11140 (2017).
12. Wolpert, M., Gust, B., Kammerer, B. & Heide, L. Efects of deletions of
mbtH-like genes on clorobiocin biosynthesis in Streptomyces coelicolor.
Microbiology 153, 1413–1423 (2007).
13. Laursen, J. B. & Nielsen, J. Phenazine natural products: biosynthesis, synthetic
analogues, and biological activity. Chem. Rev. 104, 1663–1685 (2004).
14. Guttenberger, N., Blankenfeldt, W. & Breinbauer, R. Recent developments in
the isolation, biological function, biosynthesis, and synthesis of phenazine
natural products. Bioorg. Med. Chem. 25, 6149–6166 (2017).
15. Blankenfeldt, W. & Parsons, J. F. Te structural biology of phenazine
biosynthesis. Curr. Opin. Struct. Biol. 29, 26–33 (2014).
16. Mentel, M. et al. Of two make one: the biosynthesis of phenazines.
ChemBioChem 10, 2295–2304 (2009).
17. Price-Whelan, A., Dietrich, L. E. & Newman, D. K. Rethinking ‘secondary’
metabolism: physiological roles for phenazine antibiotics. Nat. Chem. Biol. 2,
71–78 (2006).
18. Mavrodi, D. V., Blankenfeldt, W. & Tomashow, L. S. Phenazine compounds
in fuorescent Pseudomonas spp. biosynthesis and regulation. Annu. Rev.
Phytopathol. 44, 417–445 (2006).
19. Rui, Z. et al. Insights into a divergent phenazine biosynthetic pathway
governed by a plasmid-born esmeraldin gene cluster. Chem. Biol. 19,
1116–1125 (2012).
20. Wu, C. et al. Leucanicidin and endophenasides result from methyl-
rhamnosylation by the same tailoring enzymes in Kitasatospora sp. MBT66.
ACS Chem. Biol. 11, 478–490 (2016).
21. Lengyel, K. et al. Description of four novel species of Xenorhabdus, family
Enterobacteriaceae: Xenorhabdus budapestensis sp. nov., Xenorhabdus ehlersii
sp. nov., Xenorhabdus innexi sp. nov., and Xenorhabdus szentirmaii sp. nov.
Syst. Appl. Microbiol. 28, 115–122 (2005).
22. Weber, T. et al. antiSMASH 3.0—a comprehensive resource for the
genome mining of biosynthetic gene clusters. Nucleic Acids Res. 43,
W237–W243 (2015).
23. Tobias, N. J. et al. Natural product diversity associated with the
nematode symbionts Photorhabdus and Xenorhabdus. Nat. Microbiol. 2,
1676–1685 (2017).
24. Leisch, H., Morley, K. & Lau, P. C. K. Baeyer−Villiger monooxygenases: more
than just green chemistry. Chem. Rev. 111, 4165–4222 (2011).
25. White, S. W., Zheng, J., Zhang, Y.-M. & Rock, C. O. Te structural
biology of type II fatty acid biosynthesis. Annu. Rev. Biochem. 74,
791–831 (2005).
26. Yu, S. et al. Atomic resolution structure of EhpR: phenazine resistance in
Enterobacter agglomerans Eh1087 follows principles of bleomycin/mitomycin
C resistance in other bacteria. BMC Struct. Biol. 11, 33 (2011).
27. Giddens, S. R., Feng, Y. J. & Mahanty, H. K. Characterization of a novel
phenazine antibiotic gene cluster in Erwinia herbicola Eh1087. Mol. Microbiol.
45, 769–783 (2002).
28. Myhren, L. E. et al. Iodinin (1,6-dihydroxyphenazine 5,10-dioxide) from
Streptosporangium sp. induces apoptosis selectively in myeloid leukemia cell
lines and patient cells. Mar. Drugs 11, 332–349 (2013).
33. Bode, E. et al. Biosynthesis and function of simple amides in Xenorhabdus
doucetiae. Environ. Microbiol. 19, 4564–4575 (2017).
34. Bode, E. et al. Simple “on-demand” production of bioactive natural products.
ChemBioChem 16, 1115–1119 (2015).
35. Chatterjee, S. et al. Phencomycin, a new antibiotic from a Streptomyces
species HIL Y-9031725. J. Antibiot. 48, 1353–1354 (1995).
36. Rix, U. et al. Te dynamic structure of jadomycin B and the amino
acid incorporation step of its biosynthesis. J. Am. Chem. Soc. 126,
4496–4497 (2004).
37. Colosimo, D. A. & MacMillan, J. B. Detailed mechanistic study of the
non-enzymatic formation of the discoipyrrole family of natural products.
J. Am. Chem. Soc. 138, 2383–2388 (2016).
38. Vidoudez, C. & Pohnert, G. Growth phase-specifc release of polyunsaturated
aldehydes by the diatom Skeletonema marinoi. J. Plankton Res. 30,
1305–1313 (2008).
39. Brameyer, S., Kresovic, D., Bode, H. B. & Heermann, R. Dialkylresorcinols
as bacterial signaling molecules. Proc. Natl Acad. Sci. USA 112,
572–577 (2015).
40. Mori, T. et al. Structural insight into the enzymatic formation of bacterial
stilbene. Cell Chem. Biol. 23, 1468–1479 (2016).
41. Bretschneider, T. et al. A ketosynthase homolog uses malonyl
units to form esters in cervimycin biosynthesis. Nat. Chem. Biol. 8,
154–161 (2012).
42. Proschak, A. et al. Biosynthesis of the insecticidal xenocyloins in
Xenorhabdus bovienii. ChemBioChem 15, 369–372 (2014).
43. Kwon, H.-J. et al. C-O bond formation by polyketide synthases. Science 297,
1327–1330 (2002).
44. Qiu, X. et al. Crystal structure of β-ketoacyl-acyl carrier protein synthase III:
a key condensing enzyme in bacterial fatty acid biosynthesis. J. Biol. Chem.
274, 36465–36471 (1999).
45. Lin, S., Van Lanen, S. G. & Shen, B. A free-standing condensation enzyme
catalyzing ester bond formation in C-1027 biosynthesis. Proc. Natl Acad. Sci.
USA 106, 4183–4188 (2009).
46. Imamura, N. et al. New anticancer antibiotics pelagiomicins,
produced by a new marine bacterium Pelagiobacter variabilis. J. Antibiot. 50,
8–12 (1997).
47. Singh, M. P. et al. Biological and mechanistic activities of
phenazine antibiotics produced by culture LL-14I352. J. Antibiot. 50,
785–787 (1997).
48. Burger, R. M. Cleavage of nucleic acids by bleomycin. Chem. Rev. 98,
1153–1169 (1998).
49. Nikaido, H. & Takatsuka, Y. Mechanisms of RND multidrug efux pumps.
Biochim. Biophys. Acta 1794, 769–781 (2009).
50. Wang, Y., Luo, Q., Zhang, X. & Wang, W. Isolation and purifcation of a
modifed phenazine, griseoluteic acid, produced by Streptomyces griseoluteus
P510. Res. Microbiol. 162, 311–319 (2011).
51. Moon, S. H. et al. Novel linear lipopeptide paenipeptins with potential for
eradicating bioflms and sensitizing gram-negative bacteria to rifampicin and
clarithromycin. J. Med. Chem. 60, 9630–9640 (2017).
52. Fischbach, M. A. & Clardy, J. One pathway, many products. Nat. Chem. Biol.
3, 353–355 (2007).
53. Clemons, P. A. et al. Quantifying structure and performance diversity for sets
of small molecules comprising small-molecule screening collections. Proc.
108, 6817–6822 (2011).
Natl Acad. Sci. USA
54. Shi, Y.-M. & Bode, H. B. Chemical language and warfare of bacterial natural
products in bacteria-nematode-insect interactions. Nat. Prod. Rep. 35,
309–335 (2018).
55. Hadjithomas, M. et al. IMG-ABC: a knowledge base to fuel discovery
of biosynthetic gene clusters and novel secondary metabolites. mBio 6,
e00932 (2015).
Acknowledgements
The authors are grateful to T. A. Wichelhaus from Universitätsklinikum Frankfurt for
the help with antibiotic susceptibility testing and the following colleagues from Goethe
Universität Frankfurt: K. A. J. Bozhüyük for constructive discussion, K. M. Pos for
providing the E. coli efflux pump mutant strains, and Y. Kopp and S. Mauer for the initial
analysis of the phenazine biosynthesis in X. szentirmaii. This work was supported by the
LOEWE program of the state of Hesse (LOEWE Schwerpunkt MegaSyn and LOEWE
Zentrum TBG). Y.-M.S. is supported by a Postdoctoral Research Fellowship from the
Alexander von Humboldt Foundation.
NATURe CHeMiCAl BiOlOgY | www.nature.com/naturechemicalbiology

NATurE CHEmiCAl BioloGy
Articles
Author contributions
Additional information
Y.-M.S. and H.B.B. conceived the project and wrote the paper. All experiments were
performed by Y.-M.S., except initial analysis and heterologous expression of the gene
clusters performed by A.O.B. RNA sequencing was performed by M.W. and N.J.T. and
G. mellonella injection was performed by N.N. Y.-M.S., N.J.T., and H.B.B. discussed the
results and commented on the manuscript.
Supplementary information is available for this paper at https://doi.org/10.1038/
s41589-019-0246-1.
Reprints and permissions information is available at www.nature.com/reprints.
Correspondence and requests for materials should be addressed to H.B.B.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
Competing interests
© The Author(s), under exclusive licence to Springer Nature America, Inc. 2019
The authors declare no competing interests.
NATURe CHeMiCAl BiOlOgY | www.nature.com/naturechemicalbiology

NATurE CHEmiCAl BioloGy
Articles
with l-valine-d8 and l-leucine-d10 at a final concentration of 1mM. For inverse
feeding experiments with l-isoleucine, the cell pellets of the 100μL overnight
culture were washed once with ISOGRO-13C medium (100μL) and resuspended
in 13C medium (100μL). The feeding culture in 13C medium (5mL) was inoculated
with a washed overnight culture (50μL) and additional l-isoleucine was added at a
final concentration of 1mM.
Methods
General experimental procedures. All chemicals were purchased from Sigma-
Aldrich Corporation. Isotope-labeled chemicals were purchased from Cambridge
Isotope Laboratories, Inc. PCR amplifcations were carried out on thermocyclers
(SensoQuest). DNA polymerases (Taq, Phire, and Phusion) and restriction
enzymes were purchased from New England Biolabs or Termo Fisher Scientifc.
DNA primers were purchased from Eurofns MWG Operon. Protein purifcation
was performed on the ÄKTA purifer system (GE Healthcare) with a His/Strep
Trap HP 5mL column (GE Healthcare). HPLC–UV–MS analysis was conducted
on an UltiMate 3000 system (Termo Fisher) coupled to an AmaZonX mass
spectrometer (Bruker) with an ACQUITY UPLC BEH C18 column (130Å,
2.1mm×100mm, 1.7μm particle size, Waters) at a fow of 0.6mLmin−1 (5–95%
acetonitrile/water with 0.1% formic acid, v/v, 22min, UV detection wavelength
350–800nm). HPLC–UV–HRMS analysis was performed on an UltiMate 3000
system (Termo Fisher) coupled to an Impact II qTof mass spectrometer (Bruker)
with an ACQUITY UPLC BEH C18 column (130Å, 2.1mm×100mm, 1.7 μm
particle size, Waters) at a fow of 0.4mLmin−1 (5–95% acetonitrile/water with 0.1%
formic acid, v/v, 16min, UV detection wavelength 350–800nm). Semipreparative
HPLC was performed on an Agilent 1260 system coupled to a DAD detector
with a C-18 ZORBAX Eclipse XDB column (9.4mm×250mm, 5 μm). NMR
experiments were acquired on a Bruker AVANCE 500MHz spectrometer equipped
with a 5mm cryoprobe.
Isolation and purification of phenazines and NMR spectroscopy. For the
isolation and purification of phenazines, XAD-16 resins from a 12liter culture
of cells of PBAD xpzI ∆xpzPQS mutant induced by l-arabinose were collected
after 72h of incubation at 30°C with shaking at 120r.p.m., and were washed
with water and extracted with methanol/acetone (1:1, v/v, 2×12liters) to yield a
crude extract (9g) after evaporation. The extract was dissolved in methanol and
was chromatographed on a Sephadex LH-20 (GE Healthcare) column eluting
with methanol to furnish seven fractions. The fractions were further purified by
semipreparative HPLC using an acetonitrile/water gradient (0.1% formic acid)
0–35min, 5–95% to afford 1 (3.8mg), 3 (5.9mg), 4 (1.8mg), 8 (120mg), 9 (78mg),
16 (3.8mg), 17 (65mg), 18 (8.2mg), 19 (0.8mg), 20 (0.5mg), 21 (0.7mg), and
22 (0.9mg). 1H and 13C NMR, 1H–13C heteronuclear single quantum coherence
(HSQC), 1H-13C heteronuclear multiple bond correlation (HMBC), and 1H-1H
correlation spectroscopy (COSY) were measured. Chemical shifts (δ) were
reported in parts per million (ppm) and referenced to the solvent signals. Data are
reported as follows: chemical shift, multiplicity (br, broad, s, singlet, d, doublet,
t, triplet, dd, doublet of doublet, m, multiplet), and coupling constants in hertz (Hz).
Strain and culture conditions. Xenorhabdus szentirmaii DSM 16338 wild-type
strain, and the mutants thereof, and E. coli were cultivated on LB agar plates
at 30°C overnight and were subsequently inoculated into liquid LB culture
and cultivated at 30°C with shaking at 200r.p.m. Phenazine production in X.
szentirmaii and in E. coli were performed in M63 and LB medium, respectively. The
overnight LB culture was transferred into the corresponding medium (1:100, v/v)
with 2% (v/v) of Amberlite XAD-16 resins, 0.1% of l-arabinose or 1mM isopropyl
β-d-1-thiogalactopyranoside (IPTG) as the inducer, and selective antibiotics such
as ampicillin (Am, 100µgmL−1), kanamycin (Km, 50µgmL−1), chloramphenicol
(Cm, 34µgmL−1), or spectinomycin (Sm, 50µgmL−1) for phenazine production.
In vivo enzymatic assays of XpzG, XpzGH, and XpzPQS, growth curves
measurement, and uptake and export experiments. An overnight culture of
BL21_YS31 (carrying xpzG) was transferred into 5mL LB medium. When the
culture grew to an OD600 of 0.5–0.7, 0.1mM PDC (1) and 1mM IPTG were fed
with shaking at 200r.p.m. at 30°C. A non-induced (without IPTG) negative control
was also conducted. After 24h, the culture was added 2% (v/v) of XAD-16 resins
with shaking for 1h. After removal of the culture, the resins were extracted with
5mL methanol/acetone (1:1, v/v) for HPLC–UV–MS analysis. The same was done
for BL21_YS35 (carrying xpzGH) with PDC (1) and BL21_YS53 (carrying xpzPQS)
with demethoxygriseoluteic acid (8), griseoluteic acid (9), and 6-(aminomethyl)
phenazine-1-carboxylic acid (23).
For growth curves measurement and uptake and export experiments, cell
densities (OD600) of the induced and non-induced E. coli cultures incubated with
the substrates above were measured over a 27-hour time course. A 24-hour time
point was chosen for metabolite analyses in all experiments after bacteria had
reached stationary growth phase. An E. coli culture without any treatment was
conducted as a negative control. Logarithms of OD600 from each strain were plotted
against the corresponding time point.
Culture extraction and HPLC–UV–MS analysis. The XAD-16 resins were
collected after 72h and extracted with methanol/acetone (1:1, v/v). The crude
extracts were diluted 1:5 (v/v) with methanol. Crude extracts (5μL) were injected
and analyzed by HPLC–UV–MS. Unless otherwise specified, HPLC–UV–MS
chromatograms in the figures are shown on the same scale.
Construction of insertion mutants. The first 500–800base pairs (bp) of the
target gene (xpzR, xpzA2, or xpzI) were amplified with a corresponding primer
pair listed in Supplementary Table 4. Resulting fragments were cloned using
Hot Fusion56 into pCEP_KmR backbone that was amplified by pCEP_Fw and
pCEP_Rv. After transformation of the constructed plasmid into E. coli S17-1 λ pir,
clones were verified by PCR with primers YS-pCEP-Ve-Fw and YS-pDS132-Ve-
Rv. X. szentirmaii (recipient) was mated with E. coli S17-1 λ pir (donor) carrying
respective constructed plasmids. Both strains were grown in LB medium to an
optical density OD600 of 0.6 to 0.7, and the cells were washed once with fresh LB
medium. Subsequently, the donor and recipient strains were mixed on an LB agar
plate in ratios of 1:3 and 3:1, and incubated at 37°C for 3h followed by incubation
at 30°C for 21h. After this time, the bacterial cell layer was collected with an
inoculating loop and resuspended in 2mL fresh LB medium. Resuspended culture
(200μL) was spread out on an LB agar plate with ampicillin and kanamycin and
incubated at 30°C for 2 days. Individual insertion clones were cultivated and
analyzed by HPLC–UV–MS, and the genotype of all mutants was verified for by
using plasmid- and genome-specific primers.
In vitro enzymatic assays of XpzG and XpzH. For heterologous expression and
purification of the recombinant XpzG, overnight culture of BL21_YS31 (carrying
xpzG) was transferred into 1liter fresh LB medium with additional 50mL 20×NPS
(66g (NH4)2SO4, 136g KH2PO4, 142g Na2HPO4, and 900mL ddH2O, sterile
filtrated), 20mL 50×5052 (250g glycerol, 25g glucose,100g α-lactose, and 730mL
ddH2O, sterile filtrated), 1mL of 1M MgSO4 for auto induction58, and 5mL 10%
arabinose for induction of recombinant protein expression, followed by incubation
at 37°C with shaking at 200r.p.m. When the cultures grew to an OD600 of 0.5–0.7,
they were kept growing at 16°C with shaking at 200r.p.m. for 72h. After this time,
the cells were collected by centrifugation (10,000r.p.m., 15min, 4°C) and the
pellets were frozen at −20°C. The defrosted cell pellets were resuspended in 50mL
His-Tag binding buffer [20mM HEPES, 500mM NaCl, 20mM imidazole, pH 8.5]
supplemented with one protease inhibitor tablet (Roche), 0.5mgmL−1 lysozyme
(10UmL−1), and 3μL Benzonase Nuklease (25 UμL−1), incubated at 4°C for 1h,
and additionally lysed by sonication. Cell debris was removed from the lysate
by centrifugation (20,000r.p.m., 30min, 4°C) and the lysate was loaded onto a
HisTrap HP 5mL column and purified in the purifier system with His-Tag elution
buffer (20mM HEPES, 500mM NaCl, 500mM imidazole, pH 8.5). The collected
fractions, including the protein of interest, were pooled and concentrated with
Centritrep units, MWCO=30kDa (Merck Millipore) and the buffer was changed
into assay buffer (50mM Tris–HCl, pH 8.5) with PD-10 desalting columns
(GE Healthcare). The protein concentration was measured using the NanoDrop
ND 1000 (Thermo Fisher Scientific). The expression and purification of XpzH
was performed in BL21_YS33 (carrying xzpH) in the same procedure except for
using Strep-Tag binding (100mM Tris–HCl, 150mM NaCl, pH 8.5) and elution
(100mM Tris–HCl, 150mM NaCl, 2.5mM desthiobiotin, pH 8.5) buffers and
Strep Trap HP 5mL column.
Construction of deletion mutants. Deletions of the genes (xpzGH, xpzK, xpzN,
or xpzPQS) were created by amplifying upstream and downstream fragments of
the respective genes. The amplified fragments were cloned into a pCKcipB vector
using Hot Fusion56. Transformation of E. coli S17-1 λ pir with the resulting plasmid
and conjugation of the plasmid in X. szentirmaii strain, as well as the generation
of double crossover mutants via counterselection, was done as described before57.
Individual deletion clones were confirmed via PCR with the verification primers
listed in Supplementary Table 4.
Heterologous expression. All plasmids carrying target genes for heterologous
expression were constructed via Hot Fusion56. The phenazine core biosynthetic
genes (xpzABCDETF) were amplified and cloned into pCOLADuet-1. The
oxidases encoded genes in the first xpz BGC (xpzGH, xpzU, and xpzUGH) and
the griseoluteic acid biosynthetic genes (xpzRA2IJKLMNO and xpzA2IJKLMNO)
were separately cloned in pACYCDuet-1. The NPRS-like module (xpzPQS)
was constructed in pCDFDuet-1 and pCOLADuet-1. E. coli BL21(DE3) was
transformed with plasmids carrying different genes for co-expression as shown in
Supplementary Fig. 1.
The enzymatic reactions of XpzG with PDC (1) and 6-hydroxyphenazine-1-
carboxylic acid (3), and XpzH with 1,6-dihydroxyphenazine (4), were performed in
reaction mixtures containing 50mM Tris–HCl (pH 8.5), 0.5mM substrate (pure 1,
3, or 4) in DMSO, 4mM NADPH (or NADH for XpzG), 20μM FAD, and 10μM
purified enzyme (XpzG or XpzH) in a final volume of 100μL. After incubation at
room temperature for 2h, the reaction was quenched by adding 100μL methanol/
acetone (1:1, v/v) and mixed by vortex. After centrifugation, the supernatant was
analyzed by HPLC–UV–MS.
Isotope labeling experiments for structural elucidation of phenazine–peptides
by MS. The overnight culture was transferred into M63 medium additionally fed
NATURe CHeMiCAl BiOlOgY | www.nature.com/naturechemicalbiology

NATurE CHEmiCAl BioloGy
Articles
In vitro non-enzymatic assays of phenazine–polyketide formation. The in
vitro reactions were performed in 50mM Tris–HCl pH 7.0 and 9.0 that was used
to simulate the basic environment of the culture medium (pH ∼9). The reaction
mixtures contain 0.5mM substrate (a mixture containing 7 or pure 17) in DMSO
and 0.5mM methyl acetoacetate in a final volume of 100μL. After incubation at
30°C for 24h, the reaction was quenched by adding 100μL methanol and mixed
thoroughly. After centrifugation, the supernatant was analyzed by HPLC–UV–MS.
to early exponential phase (OD600 ≈ 1) and collected. Total RNA was extracted
exactly as described61. RNA quality control, rRNA depletion, and RNA sequencing
were performed in duplicate by IMGM Laboratories GmbH using an Illumina
HiSeq2500. The sequence data were mapped to the genome of X. szentirmaii DSM
16338 and analyzed as described61.
Phenazine production in infected Galleria. Precultures of X. szentirmaii were
grown in LB medium and inoculated into fresh cultures at an OD600 of 0.1. Cells
were grown to exponential phase (OD600 ≈1) and then diluted to an OD600 of
0.00025. Diluted X. szentirmaii wild-type culture (5µl) was injected into the left last
pro-leg of the larvae (LB medium as a negative control). G. mellonella larvae were
kept at 4°C for 10min before injection. After infection the larvae were incubated at
25°C. Dead Galleria larvae were frozen at −20°C, then at −80°C, and freeze-dried
for 1day. Freeze-dried larvae were ground and extracted with n-heptane/methanol
(v/v, 1:1). The methanol phase was subjected to HPLC–MS–UV analysis.
Phylogenetic analysis. The amino acid sequences of XpzS along with 29 homologs
from the BLAST search of XpzS to the National Center for Biotechnology
Information database and 59 condensation domains/enzymes from the five known
categories from Xenorhabdus and Photorhabdus were collected to construct
two phylogenetic trees. Sequences were aligned using ClustalW. The trees were
generated using the Geneious tree builder incorporated into Geneious (v.6.1.8)
utilizing the Jukes–Cantor distance model, and were visualized by iTOL59.
Disk diffusion antibiotic susceptibility testing. The antibiotic activities of
compounds 1, 3, 4, 6, 8–13, 16–23, 27, and 28 were evaluated by using the
following indicator organisms: Bacillus subtilis American Type Culture Collection
(ATCC) 6633, Escherichia coli ATCC 25922, Pseudomonas aeruginosa ATCC
27853, Staphylococcus aureus ATCC 29213, Enterococcus faecalis ATCC 29212,
Salmonella enterica subsp. enterica ATCC 43971, Enterococcus faecium ATCC
19434, Acinetobacter baumannii ATCC 19606, Stenotrophomonas maltophilia
ATCC 13637, Enterococcus hirae ATCC 10541, Klebsiella pneumoniae ATCC 10031,
Micrococcus luteus ATCC 9341, and Proteus vulgaris. Bacteria were suspended
in sterile saline to the density of a McFarland 0.5 standard, corresponding to
1–2×108 colony forming units per ml and the inoculum was spread on Mueller
Hinton agar (Oxoid) by a sterile cotton swab according to EUCAST guidelines60.
Disks (Macherey–Nagel) with a diameter of 6mm were loaded with 10µL (10mM)
of phenazine, and placed on the agar. After incubation at 36°C for 16–20h, the
zones of inhibition were determined.
Reporting Summary. Further information on research design is available in
the Nature Research Reporting Summary linked to this article.
Data availability.
All data generated or analyzed in this study are available within the article and
its Supplementary Information files.
References
56. Fu, C., Donovan, W. P., Shikapwashya-Hasser, O., Ye, X. & Cole, R. H. Hot
Fusion: an efcient method to clone multiple DNA fragments as well as
inverted repeats without ligase. PLoS ONE 9, e115318 (2015).
57. Brachmann, A. O. et al. A type II polyketide synthase is responsible for
anthraquinone biosynthesis in Photorhabdus luminescens. ChemBioChem 8,
1721–1728 (2007).
58. Studier, F. W. Protein production by auto-induction in high-density shaking
cultures. Protein Expr. Purif. 41, 207–234 (2005).
59. Letunic, I. & Bork, P. Interactive tree of life (iTOL)v3: an online tool for the
display and annotation of phylogenetic and other trees. Nucleic Acids Res. 44,
W242–W245 (2016).
60. Matuschek, E., Brown, D. F. J. & Kahlmeter, G. Development of the EUCAST
disk difusion antimicrobial susceptibility testing method and its
implementation in routine microbiology laboratories. Clin. Microbiol. Infect.
20, O255–O266 (2014).
61. Heinrich, A. K., Glaeser, A., Tobias, N. J., Heermann, R. & Bode, H. B.
Heterogeneous regulation of bacterial natural product biosynthesis via a novel
transcription factor. Heliyon 2, e00197 (2016).
Preparation of insect homogenate. Larvae of Galleria mellonella were snap frozen
in liquid nitrogen for 2min. The frozen larvae were transferred into a 15mL tube
containing 10mL precooled Tris–HCl (20mM, pH 7.5) and Teen A Lysing Matrix
(MP Biomedicals). The larvae were then homogenized twice for 40s (4ms−1) using
the P Fast Prep-24 Tissue and Cell Homogenizer (MP Biomedicals). To get rid of
cell debris, we centrifuged the homogenate for 10min at 10,000r.p.m. at 4°C. The
homogenate was transferred into a round-bottomed flask, frozen at −80°C, and
freeze-dried for 14–16h.
RNA sequencing. Precultures of X. szentirmaii DSM 16338 were grown in LB
medium and inoculated into fresh cultures at an OD600 of 0.3. Cells were grown
NATURe CHeMiCAl BiOlOgY | www.nature.com/naturechemicalbiology

Corresponding author(s):
Helge B. Bode
Reporting Summary
Nature Research wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency
in reporting. For further information on Nature Research policies, see Authors & Referees and the Editorial Policy Checklist.
Statistical parameters
When statistical analyses are reported, confirm that the following items are present in the relevant location (e.g. figure legend, table legend, main
text, or Methods section).
n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement
An indication of whether measurements were taken from distinct samples or whether the same sample was measured repeatedly
The statistical test(s) used AND whether they are one- or two-sided
Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested
A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons
A full description of the statistics including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND
variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals)
For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted
Give P values as exact values whenever suitable.
For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings
For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes
Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated
Clearly defined error bars
State explicitly what error bars represent (e.g. SD, SE, CI)
Our web collection on statistics for biologists may be useful.
Software and code
Policy information about availability of computer code
Bruker Compass DataAnalysis 4.3 was used for chromatography and mass spectrometry. Bruker TopSpin 4.0 was used for NMR data.
Geneious 6.1.8 was used for sequence data.
Data collection
Recent releases of standard mass spectrometry data analysis in GNPS were used to identify known compounds. Anti-smash 3.0.5 and
NCBI protein BLAST were used for gene annotation. ClustalW integrated in Geneious 6.1.8 and iTOL v4 were used for sequence
alignments and phylogeny. Data was analyzed and statistically evaluated using OriginPro 2017 SR1 b9.4.2.380 and Excel of Microsoft
office 365. RNA reads were mapped using bowtie2, counted using nesoni and Degust was used to analyze read counts and their
statistical significance.
Data analysis
For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors/reviewers
upon request. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information.
1

Data
Policy information about availability of data
All manuscripts must include a data availability statement. This statement should provide the following information, where applicable:
- Accession codes, unique identifiers, or web links for publicly available datasets
- A list of figures that have associated raw data
- A description of any restrictions on data availability
The authors declare that the data supporting the findings of this study are available within the paper and its supplementary information files.
Field-specific reporting
Please select the best fit for your research. If you are not sure, read the appropriate sections before making your selection.
Life sciences
Behavioural & social sciences
Ecological, evolutionary & environmental sciences
For a reference copy of the document with all sections, see nature.com/authors/policies/ReportingSummary-flat.pdf
Life sciences study design
All studies must disclose on these points even when the disclosure is negative.
Sample sizes were determined according to the field standards. In general, all experiments were run at least three independent times with
representative data shown. Three to five replicates is a standard sample size for measuring production titer of compounds and bacterial
growth curve.
Sample size
No data were excluded from analyses.
Data exclusions
Replication
All determination were repeated in at least triplicates. All replication attempts were successful. Disc diffusion antibiotic susceptibility testing
was only performed once.
Randomization is irrelevant since our study does not involve group allocation.
Blinding is irrelevant since our study does not involve group allocation.
Randomization
Blinding
Reporting for specific materials, systems and methods
Materials & experimental systems
n/a Involved in the study
Unique biological materials
Antibodies
Methods
n/a Involved in the study
ChIP-seq
Flow cytometry
Eukaryotic cell lines
MRI-based neuroimaging
Palaeontology
Animals and other organisms
Human research participants
2