Ambiguity of NRPS Structure Predictions: Four Bidentate Chelating Groups in the Siderophore Pacifibactin

Article abstract of DOI:10.1021/acs.jnatprod.8b01073

Identified through a bioinformatics approach, a nonribosomal peptide synthetase gene cluster in Alcanivorax pacificus encodes the biosynthesis of the new siderophore pacifibactin. The structure of pacifibactin differs markedly from the bioinformatic prediction and contains four bidentate metal chelation sites, atypical for siderophores. Genome mining and structural characterization of pacifibactin is reported herein, as well as characterization of pacifibactin variants accessible due to a lack of adenylation domain fidelity during biosynthesis. A spectrophotometric titration of pacifibactin with Fe(III) and ¹³C NMR spectroscopy of the Ga(III)-pacifibactin complex establish 1:1 metal:pacifibactin coordination and reveal which of the bidentate binding groups are coordinated to the metal. The photoreaction of Fe(III)-pacifibactin, resulting from Fe(III) coordination of the β-hydroxyaspartic acid ligands, is reported.

Full text of DOI:10.1021/acs.jnatprod.8b01073

Article
pubs.acs.org/jnp
Cite This: J. Nat. Prod. XXXX, XXX, XXX−XXX
Ambiguity of NRPS Structure Predictions: Four Bidentate Chelating
Groups in the Siderophore Pacifibactin
Clifford D. Hardy and Alison Butler*
Department of Chemistry & Biochemistry, University of California Santa Barbara, Santa Barbara, California 93106-9510, United
States
S
* Supporting Information
ABSTRACT: Identified through a bioinformatics approach, a
nonribosomal peptide synthetase gene cluster in Alcanivorax
pacificus encodes the biosynthesis of the new siderophore
pacifibactin. The structure of pacifibactin differs markedly
from the bioinformatic prediction and contains four bidentate
metal chelation sites, atypical for siderophores. Genome
mining and structural characterization of pacifibactin is
reported herein, as well as characterization of pacifibactin
variants accessible due to a lack of adenylation domain fidelity
during biosynthesis. A spectrophotometric titration of
pacifibactin with Fe(III) and 13C NMR spectroscopy of the
Ga(III)-pacifibactin complex establish 1:1 metal:pacifibactin coordination and reveal which of the bidentate binding groups are
coordinated to the metal. The photoreaction of Fe(III)-pacifibactin, resulting from Fe(III) coordination of the β-
hydroxyaspartic acid ligands, is reported.
utomated genome mining tools enable high-throughput
scanning of bacterial genomes for gene clusters encoding
are synthesized by NRPSs, and genome mining has enabled the
prediction and discovery of many new siderophore struc-
tures.9−13 NRPS-directed siderophore biosynthesis often
employs extensive tailoring of both the amino acid substrates
and the assembled product to yield Fe(III)-chelating functional
groups. These tailoring reactions may be carried out by
standalone proteins not accounted for in the commonly
utilized NRPS analysis tools.
The genomes of many species within the obligate hydro-
carbon-degrading microbial genus Alcanivorax have been
sequenced, yet siderophore biosynthesis has only been
identified in Alcanivorax borkumensis.14 This abundance of
sequenced Alcanivorax genomes enables a genome mining
approach for further characterization of siderophore produc-
tion in the genus. To this end, fully sequenced genomes of
Alcanivorax species not known to produce siderophores were
screened for potential NRPS gene clusters encoding side-
rophore biosynthesis. Of the analyzed genomes, only
Alcanivorax pacificus contains a candidate NRPS gene cluster
for siderophore production.
A
biosynthetic machinery.1,2 Natural products produced by
nonribosomal peptide synthetases (NRPSs) are particularly
amenable to discovery through bioinformatics approaches.
NRPSs are multimodular proteins that work in an assembly
line fashion to synthesize a peptidic natural product.3−6
A
typical NRPS consists of at least one adenylation domain, a
thiolation domain, and a condensation domain. This triad of
catalytic domains governs selection and activation of a
substrate (generally an amino acid), anchors the activated
substrate to the NRPS assembly, and incorporates the amino
acid into the synthesized product through peptide bond
formation. Domains responsible for tailoring reactions may
also be present that carry out tailoring reactions such as
epimerization and cyclization.4 The organization of an NRPS
into distinct domains with predictable functions and amino
acid substrates is a key feature driving software such as
antiSMASH and PRISM, which utilize sequence analysis to
identify NRPS-encoding gene clusters and predict their
functionality.7,8
Iron plays a central role in many enzymatic processes and
protein functions across all organisms and, thus, is a required
nutrient for virtually all life. The requirement for iron
complicates bacterial growth, as readily bioavailable iron is
limited by the low solubility of Fe(III) in aerobic environ-
ments. In response to low iron stress, many bacteria synthesize
siderophores, small-molecule Fe(III) chelators produced by a
bacterium to solubilize Fe(III) in the environment. Fe(III)-
siderophore complexes are recognized by an outer membrane
receptor protein and taken up by the cell. Many siderophores
We report herein that A. pacificus contains a biosynthetic
gene cluster that encodes the synthesis of the previously
unknown siderophore pacifibactin. Structural characterization
of pacifibactin reveals limitations of current automated genome
mining approaches, highlighting several tailoring steps as yet
undetectable through existing software. Structural variants of
pacifibactin resulting from substrate selection promiscuity by
Received: December 19, 2018
© XXXX American Chemical Society and
American Society of Pharmacognosy
A
DOI: 10.1021/acs.jnatprod.8b01073
J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
Figure 1. Pacifibactin biosynthetic gene cluster, annotated with NRPS and PKS domains as identified by antiSMASH and PRISM. Adenylation
domains (lavender boxes) are labeled with their substrate specificity prediction. The adenylation domain of PfbG is predicted to incorporate Ser;
however structural characterization of pacifibactin establishes the presence of Ala. Te, thioesterase domain; C, condensation domain; T, thiolation
domain; KS, ketosynthase domain; Mal, malonic acid (substrate specificity); KR, ketoreductase domain; E, epimerization domain.
an adenylation domain are also described. Pacifibactin is
unique among siderophores in that it contains two hydroxamic
acid and two β-hydroxyaspartic acid functional groups. Four
potential bidentate Fe(III) binding groups are rarely observed
in siderophore structures and were not predicted from the
genome mining analysis. The coordination chemistry of
Fe(III)-pacifibactin is reported, as is the photoreactivity of
the Fe(III)-siderophore complex due to coordination by β-
hydroxyaspartic acid ligands.15
923.41 [M + H]+ and m/z 462.20 [M + 2H]2+. The putative
siderophore, named herein as pacifibactin, was purified by
semipreparative RP-HPLC. HR-ESIMS of purified pacifibactin
detects a protonated molecule of m/z 923.4081 (Figure S1).
MS/MS peptide b/y fragmentation reveals the constituent
amino acids of pacifibactin (Figure 2, Figure S2), with an
RESULTS AND DISCUSSION
■
Analysis of the pfb Gene Cluster for Siderophore
Biosynthesis. The genomes of eight fully sequenced
Alcanivorax species available through NCBI were analyzed
using the bioinformatics software tools antiSMASH and
PRISM,7,8 which identify and annotate putative nonribosomal
peptide synthetase genes. The genome of A. pacificus W11-5T,
a species isolated from a bacterial consortium found within
seafloor sediment in the Pacific Ocean,16 contains a putative
siderophore biosynthetic gene cluster centered around three
NRPS-encoding genes and one polyketide synthase (PKS)-
encoding gene, identified by both antiSMASH and PRISM
(Figure 1). Putative genes involved in siderophore transport,
Fe(III)-siderophore reduction, and amino acid tailoring are
also present within the cluster (full annotation, Table S1). A
siderophore structure comprising L-Ser, malonic acid (PKS), L-
Asp, D-Arg, D-Asp, L-Ser, L-OH-Orn, and D-OH-Orn is
predicted from the adenylation domain specificity and the
location of epimerization domains within the NPRS/PKS
assembly line (Figure 1). The presence of genes predicted to
encode a TonB-dependent receptor protein and siderophore-
iron reductase (Table S1), and the incorporation of aspartic
acid and ornithine, which are commonly functionalized as
Fe(III) binding groups, further supported siderophore
production. Additionally, the genome of A. pacificus contains
no other NRPS-encoding gene clusters, nor does it contain any
biosynthetic gene clusters indicating currently known NRPS-
independent siderophore biosynthesis.
Figure 2. Structure of pacifibactin, with b/y peptide fragment masses
(fragments with m/z in red were not observed). Observed internal
fragments are shown above the structure.
amino acid sequence from the carboxylate terminus of cyclized
N5-OH-Orn, N5-acetyl-N5-OH-Orn, Ser, β-OH-Asp, Arg, and
β-OH-Asp. The two β-OH-Asp residues, N5-acetyl-N5-OH-
Orn residue, and cyclized N5-OH-Orn residue yield in total
four bidentate metal binding sites. The presence of four sites
stands in contrast to almost all known siderophores, which
generally contain no more than three bidentate metal binding
sites.
MS/MS fragmentation suggested a mass for the N-terminal
amino acid that did not match any amino acid previously
found in peptidic siderophores. Analysis of the NRPS domains
within the gene cluster suggested that Ser and malonic acid
(presumably incorporated by the PKS enzyme PfbH) should
be incorporated at the N-terminus of pacifibactin. Mixed
NRPS-PKS siderophore biosyntheses involve a decarboxylative
Claisen condensation between the carboxylate functional
groups of an NRPS-recruited amino acid and a PKS-recruited
substrate,12,18 and indeed the mass of the N-terminal
pacifibactin amino acid determined by MS/MS fractionation
is consistent with a decarboxylative Claisen condensation of
Isolation and Structural Characterization of Pacifi-
bactin. To induce siderophore production, A. pacificus was
grown in an iron-deficient artificial seawater medium. A.
pacificus grows readily under iron-starvation conditions, and
aliquots of the culture tested positive in the liquid chrome
azurol S (CAS) assay for strong Fe(III)-binding ligands.17
UPLC-ESIMS analysis of the supernatant extract revealed a
candidate compound with protonated molecule masses of m/z
B
DOI: 10.1021/acs.jnatprod.8b01073
J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
Ser and malonate followed by stepwise reduction of the β-keto
the PKS-encoding gene identifies a ketoreductase domain that
could carry out this reduction to a hydroxyl functionality
(Figure 1).
group to an alkyl moiety.
1
Complete assignment of each H and 13C chemical shift of
pacifibactin was accomplished through COSY, HSQC, and
HMBC NMR techniques (Table 1, Figure 3). The NMR-
Table 1. NMR Spectroscopic Data (500 MHz, (CD3)2SO)
for Pacifibactin (20 mg)
δC, type
δH (J in Hz)
COSY
HMBC
1, 3, 6
1
2
3
4
164.5, C
49.6, CH
20.1, CH2
27.5, CH2
51.2, CH2
4.30, m
1.85, m; 1.59, m
1.88, m
3.48, m
8.12, d (8.0)
3, N1
2, 4
3, 5
4
2
2, 3, 5
3, 4
2, 6
5
Figure 3. Structure of pacifibactin.
N1
6
7
8
9
2
171.0, C
Comparison of the structure of pacifibactin to the
bioinformatic prediction reveals several surprises. The
adenylation domain of PfbG is predicted by both antiSMASH
and PRISM to incorporate Ser, yet Ala appears to be
incorporated into pacifibactin at the N-terminus instead. The
ensemble algorithm SANDPUMA, which applies several
different adenylation domain specificity predictors and has
outperformed any individual method in accuracy,19 also
predicts Ser incorporation by PfbG. The incorporation of Ala
highlights the need for continuous refinement of adenylation
domain predictor tools as more experimental data are
generated.
Perhaps even more intriguing in the characterized structure
is the presence of four bidentate binding groups in pacifibactin,
as siderophores generally have three bidentate binding groups
to satisfy hexadentate coordination to Fe(III). The bio-
informatic analysis predicts incorporation of two Asp and
two OH-Orn by the NRPS assembly line; however neither Asp
nor OH-Orn acts as a bidentate metal chelator without further
tailoring. While these residues were expected to provide the
metal chelation sites typical of siderophores, unmodified Asp
and OH-Orn residues have been identified in other side-
rophore structures;20−23 thus not all four residues were
expected to be modified as bidentate metal chelators. An
expanded manual analysis of the pfb gene cluster helps explain
this unexpected result. The amino acid sequence of PfbN
exhibits homology to N-acetyl transferases implicated in the
acetylation of Lys and Orn in other siderophores.24,25 N5-
acetyl-N5-OH-Orn is known to be synthesized prior to
adenylation;25 thus further development of predictor tools is
needed to distinguish this substrate from OH-Orn. The
mechanism of N5-OH-Orn cyclization is as yet unknown;
however in consideration of the mechanism of NRPS peptide
bond formation cyclization must occur after incorporation of
OH-Orn, making prediction through bioinformatics tools
difficult.
52.9, CH
29.1, CH2
23.0, CH2
46.7, CH2
170.3, C
4.24, m
1.52, m
1.57, m
3.46, m
8, 9, N2
7, 9
8, 10
9
6, 8
7, 9, 10
8, 10
8, 11
10
11
12
N2
13
14
15
N3
16
17
18
19
N4
20
21
22
23
24
N5
25
N6
26
27
28
29
N7
30
31
20.4, CH3
1.97, s
8.22, d (7.5)
11
7, 13
14, 15
13, 15, 16
13, 14
14, 16
7
170.0, C
54.6, CH
61.9, CH2
4.38, m
3.48, m; 3.65, m
7.62, d (7.5)
15, N3
14
14
168.9, C
55.5, CH
70.3, CH
173.0, C
4.76, t (2.6)
4.56, d (2.5)
18, N4
17
16, 18, 19, 20
16, 17, 19
17, 18
8.18, d (8.8)
17
20
171.4, C
51.9, CH
29.6, CH2
24.5, CH2
40.3, CH2
4.47, m
22, N6
21
24
23, N5
24
20, 22
1.74, m; 1.55, m
1.44, m; 1.48, m
3.07, m
20, 21, 23
21, 22, 24
22, 23
7.47, t (5.6)
24, 25
156.7, C
7.87, d (8.0)
21
21, 26
168.8, C
55.5, CH
70.1, CH
172.8, C
4.74, t (2.6)
4.54, d (2.5)
28, N7
27
26, 28, 29, 30
26, 27, 29
8.03, d (9.0)
27
32
27, 30
170.2, C
38.9, CH2
2.32, dd (14.27, 5.49)
2.43, dd (14.33, 8.35)
4.04, m
3.19, m
1.09, d
30, 32, 33
32
33
34
67.6, CH
50.0, CH
11.9, CH3
31, 33
32, 34
33
31, 33, 34
34
32, 33
Both Asp residues in pacifibactin are hydroxylated at the β-
carbon to form β-OH-Asp. Accordingly, PfbF exhibits
homology to the Fe(II)/α-ketoglutarate-dependent Asp β-
hydroxylase SyrP of Pseudomonas syringae.26 Additionally, a
domain exhibiting homology with SyrP is found in the PKS
protein PfbH. Notably, neither antiSMASH nor PRISM picked
up this domain as an Asp β-hydroxylase. Putative Asp β-
hydroxylating domains have been observed in other side-
rophore gene clusters.9,12,23,27 In the biosynthetic gene clusters
of serobactin and cupriachelin, both Asp β-hydroxylating
NRPS domains and discrete Asp β-hydroxylating enzymes have
supported structure of pacifibactin predicts an exact mass of
m/z 923.4065 [C34H59N12O18]+, which is within 2 ppm of the
measured mass of m/z 923.4081 (Figure S1). NMR analysis
unambiguously assigns the N-terminal amino acid as a γ-amino
acid methylated at the γ-carbon and hydroxylated at the β-
carbon. This unusual amino acid is consistent with a Claisen
condensation between Ala (presumably incorporated by PfbG)
and malonate (presumably incorporated by PfbH) followed by
reduction of the β-keto group to a hydroxy group, instead of
the predicted condensation of Ser and malonate. Analysis of
C
DOI: 10.1021/acs.jnatprod.8b01073
J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
Figure 4. Proposed biosynthesis of pacifibactin, based on bioinformatics analysis of the putative biosynthetic gene cluster.
been identified.9,27 Each of these siderophores, like paci-
fibactin, contain two β-OH-Asp residues. β-Hydroxylation of
Asp occurs after tethering to the assembly line; thus
adenylation domain analysis alone is insufficient to predict
incorporation of β-OH-Asp residues in an NRPS product.
However, development of future bioinformatics tools that pair
adenylation domain analysis with identification of these
putative β-hydroxylases could successfully predict incorpo-
ration of β-OH-Asp.
mass increase localized to the N-terminal amino acid (Figure 5,
Figure S12).
The pfb gene cluster contains three epimerization domains
within the NRPS assembly line (Figure 4). Given the
placement of the epimerization domains, incorporation of
(starting from the N-terminus) L-Ala, L-β-OH-Asp, D-Arg, D-β-
OH-Asp, L-Ser, L-N5-acetyl-N5-OH-Orn, and D-N5-OH-Orn
is expected. Hydrolysis of pacifibactin with HCl or HI (for
reductive hydrolysis to yield unfunctionalized ornithine) and
derivatization of the resultant hydrolysates with Marfey’s
reagent (1-fluoro-2,4-dinitrophenyl-5-L-alanine amide, FDAA)
identifies L-threo-β-OH-Asp, D-Arg, D-threo-β-OH-Asp, L-Ser, L-
Orn, and D-Orn in the hydrolysate through co-injections with
the corresponding derivatized amino acid standards (Figures
S9, S10). The identified amino acids are consistent with the
bioinformatic prediction. Co-injections with D,L-threo-β-OH-
Asp confirmed the presence of the threo diastereomers in
pacifibactin. Moreover, co-injections with a D,L-erythro-β-OH-
Asp establish that pacifibactin does not incorporate the erythro
diastereomers (Figure S9). The incorporation of L-Ala was not
confirmed, as the PKS-governed reaction forms a carbon−
carbon bond between Ala and malonate that is not broken
under the hydrolytic conditions.
Variations in Biosynthetic Substrate Incorporation.
The unexpected incorporation of Ala into pacifibactin at the N-
terminal position, despite a consensus bioinformatic prediction
of Ser incorporation, calls into question the substrate
specificity of the corresponding adenylation domain. To
probe whether L-Ser could be incorporated into pacifibactin
in place of L-Ala, A. pacificus was cultured in media
supplemented with L-Ser. ESIMS analysis of the resultant
culture extracts identified a coeluting compound of m/z 470.2
(z = 2), 16 amu higher than the mass of pacifibactin. This mass
increase matches the expected mass increase of Ser
incorporation over Ala incorporation (Figure S11). The
fragmentation pattern observed by ESIMS/MS confirms that
this mass corresponds to a variant of pacifibactin, with the
Figure 5. MS/MS b/y fragmentation patterns for pacifibactin Ser
(470.2 m/z [M + 2H]2+) and Gly (455.2 m/z [M + 2H]2+) variants
(fragments with m/z in red were not observed). Observed internal
fragments are shown above the structure.
To further test substrate variation at the pacifibactin N-
terminus, A. pacificus was cultured with supplements of Gly, L-
Thr, and L-Val. While no incorporation of L-Thr or L-Val was
observed, A. pacificus produced a compound of m/z 455.2 (z =
2) when supplemented with Gly, 14 amu lower than the mass
of pacifibactin (Figure S11). The difference is again localized
D
DOI: 10.1021/acs.jnatprod.8b01073
J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
to the N-terminal amino acid, based on MS/MS fragmentation
(Figure 5, Figure S13). The interchange between incorpo-
ration of L-Ala, L-Ser, and Gly in pacifibactin biosynthesis is
reminiscent of the moanachelin siderophores produced by
Vibrio sp. NT1, which incorporate either Ala or Gly as the
third amino acid residue.28 The N-terminus of pacifibactin is
presumably biosynthesized by PfbG based on bioinformatics
analysis; thus further investigation into structural differences
between the adenylation domain of PfbG and known Ser-
incorporating domains could deepen the understanding of
adenylation domain selectivity.
Coordination Chemistry of Pacifibactin. The structure
of pacifibactin contains four bidentate metal binding sites, an
unusual feature for siderophores only observed among some
desferrioxamines and malleobactin D.20,29 The presence of four
bidentate ligands in the structure of pacifibactin calls into
question which groups coordinate Fe(III). Titrating a buffered
solution of apo pacifibactin with Fe(III) under neutral pH
conditions indicates that despite the extra binding group,
pacifibactin coordinates Fe(III) in a 1:1 ratio, as observed for
hexadentate siderophores (Figure 6). The UV−visible
residue for Fe(III) coordination.9 This resemblance suggests
that pacifibactin coordinates one Fe(III) through both β-OH-
Asp residues and either the cyclized N5-OH-Orn or the N5-
acetyl-N5-OH-Orn residue.
To identify the binding groups involved in Fe(III)
coordination in pacifibactin, the Ga(III)-pacifibactin complex
was prepared as an NMR-compatible mimic to Fe(III)-
pacifibactin. After Ga(III) complexation, 13C chemical shifts
in both β-OH-Asp residues and the N5-acetyl-N5-OH-Orn
residue showed significant changes relative to apo pacifibactin,
while no significant differences were observed in the 13C
chemical shifts of the cyclized N5-hydroxyornithine residue
(Table 2). The observed chemical shifts indicate that Ga(III)
likely coordinates to pacifibactin through both β-OH-Asp
residues and N5-acetyl-N5-OH-Orn, but not the cyclized N5-
OH-Orn residue (Figure 7).
Figure 7. Proposed coordination mode of Ga(III)-pacifibactin.
Ga(III) is bound by both β-OH-Asp residues and the N5-acetyl-
N5-OH-Orn residue, while the cyclized N5-OH-Orn residue remains
an open coordination site.
Fe(III) complexes of α-hydroxycarboxylate siderophores,
including β-OH-Asp siderophores, are photoreac-
tive.12,15,27,30−34 To probe photoreactivity, Fe(III)-pacifibactin
was photolyzed with a 450 W mercury-arc lamp and monitored
by UV−visible spectrophotometry. Through 8 h of continuous
photolysis, clear shifts in the UV−visible spectrum are
observable (Figure 8). Upon photolysis, the Fe(III)-hydrox-
amate absorbance band around 400 nm nearly doubles in
intensity and red shifts slightly, while two near-isosbestic
points are present, suggesting photolysis of Fe(III)-pacifibactin
initially yields one Fe(III)-coordinating photoproduct. Con-
tinued photolysis after 8 h leads to a loss of isosbestic points
and eventual elimination of the Fe(III)-α-hydroxycarboxylate
charge transfer band around 300 nm (Figure S21). After
photolysis, UPLC-ESIMS/MS analysis of the reaction mixture
identifies a photoproduct resulting from photooxidative
cleavage of the peptide backbone at the C-16 to C-19 β-
Figure 6. UV−visible absorption spectrum of 0.1 mM Fe(III)-
pacifibactin in 100 mM MOPS buffer pH 7.1. Inset: Spectrophoto-
metric titration of apo pacifibactin with Fe(III). Apo pacifibactin (0.1
mM, in 100 mM MOPS pH 7.1) was titrated with Fe(III) stock (2.14
mM in 40 mM HNO3). A break point in absorbance is observed at 0.9
equiv of Fe(III), suggesting a 1:1 coordination mode.
absorption spectrum of Fe(III)-pacifibactin features a peak at
305 nm characteristic of Fe(III)-α-hydroxycarboxylate ligand-
to-metal charge transfer (LMCT) and a broad shoulder at 400
nm indicative of Fe(III)-hydroxamate LMCT (Figure 6). The
spectrum resembles that of Fe(III)-serobactin, a siderophore
with two β-OH-Asp residues and one cyclized N5-OH-Orn
a
Table 2. 13C Chemical Shifts of Ga(III)-Pacifibactin (in D2O) Compared to Apo Pacifibactin
CO (ppm)
Cα (ppm)
Cβ (ppm)
Cγ (ppm)
Cδ (ppm)
C acetyl (ppm)
CyOHOrn
AcOHOrn
Apo
Ga(III)
Apo
Ga(III)
Apo
Ga(III)
Apo
166.41
166.18
173.49
173.78
170.37
175.94
171.56
173.16
50.32
49.97
53.62
54.28
55.76
61.23
55.66
59.09
26.56
26.83
27.91
26.11
70.27
73.18
70.07
74.10
19.93
19.83
22.36
22.99
174.21
182.65
174.09
180.05
51.57
51.65
47.20
49.87
173.84
164.61
βOHAsp
(C#16−19)
βOHAsp
(C#26−29)
Ga(III)
a
Bolded resonances indicate a significant change in the 13C chemical shift after Ga(III) coordination.
E
DOI: 10.1021/acs.jnatprod.8b01073
J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
clear that automated tools such as antiSMASH remain limited
in their ability to generate accurate structure predictions of
NRPS-synthesized siderophores, yet also details a path to
improving these predictions. While antiSMASH by design
takes a conservative approach to the chemistry prediction of
NRPS and PKS products,7 we propose that the identification
of tailoring enzymes and domains associated with siderophore
production could be incorporated into automated genome
mining tools. The accurate prediction of OH-Orn tailoring
enzymes and Asp β-hydroxylases within NRPS gene clusters
could not only improve the accuracy of structure predictions
but also aid in distinguishing NRPS clusters as siderophore
producers.
Figure 8. UV−visible absorbance spectra of 0.1 mM Fe(III)-
pacifibactin in 100 mM MOPS buffer pH 7.1 subjected to 8 h of
continuous photolysis with a 450 W UV mercury-arc lamp, collected
from 220 to 700 nm. Scans are taken at 2 h time points, and lighter
gray represents increased time. Arrows indicate increases and
decreases in absorbance over time. Inset: Proposed structure of
photoproduct detected by UPLC-ESIMS/MS (390.2 m/z, z = 1),
with MS/MS b/y fragmentation pattern. The observed internal
fragment is shown by the orange bar above the structure.
EXPERIMENTAL SECTION
■
General Experimental Procedures. Optical rotation was
measured on a Rudolph Autopol III polarimeter with a 50 mm
microcell (1.2 mL). UV−visible absorbance was measured on an
Agilent Cary 300 UV Vis spectrophotometer using 3 mL quartz
cuvettes. NMR spectroscopy was carried out on 500 MHz (1H, 13C)
and 600 MHz (COSY, HSQC, HMBC) Varian Unity Inova
spectrometers. Chemical shifts were referenced through residual
solvent peaks [1H (DMSO-d6) 2.50 ppm, 13C (DMSO-d6) 39.51
ppm] or an external reference for samples dissolved in D2O [1H, 13C
(TMS) 0.0 ppm]. Mass spectrometry analysis was carried out on a
Waters Xevo G2-XS QTof with positive mode electrospray ionization
coupled to an ACQUITY UPLC H-Class system with a Waters BEH
C18 column. Culture extracts were analyzed with a linear gradient of
0% to 30% CH3CN (0.1% formic acid) in ddH2O (0.1% formic acid)
over 10 min. For MS/MS analysis, a collision energy profile of 20, 25,
and 30 kEV was employed.
Genome Mining and Gene Cluster Annotation. The genome
of Alcanivorax pacificus W11-535 was accessed through NCBI and
analyzed with the NRPS cluster-predicting software PRISM and
antiSMASH.7,8 Genes within the pacifibactin cluster and their
corresponding amino acid sequences were analyzed using BLAST
and the PFAM database to predict function of proteins encoded by
the cluster.
Bacterial Growth and Siderophore Isolation. Alcanivorax
pacificus W11-5T, obtained from Dr. Zongze Shao (Marine Culture
Collection of China, Third Institute of State Oceanic Administration,
P. R. China), was cultured on Difco 2216 Marine medium agar plates
amended with sodium pyruvate. Single colonies were inoculated in a
liquid artificial seawater medium ASW+Py (10 g CAS amino acids
L−1, 1 g NH4Cl L−1, 1 g glycerol phosphate L−1, 12.35 g MgSO4 L−1,
1.45 g CaCl2 L−1, 16.55 g NaCl L−1, 0.75 g KCl L−1, 5 g sodium
pyruvate L−1 in ddH2O, amended with 10 mL of 1.0 M HEPES L−1, 2
mL of 1.0 M NaHCO3 L−1, and 6 mL of glycerol L−1) or in a liquid
single carbon source medium (24.6 g NaCl L−1, 0.67 g KCl L−1, 1.36
g CaCl2 L−1, 6.29 g MgSO4 L−1, 4.66 g MgCl2 L−1, 0.18 g NaHCO3
L−1, 10.0 g sodium pyruvate L−1, 2.0 g NH4Cl L−1, 0.2 g Na2HPO4
L−1, 10 mM of Ser/Gly) for amino acid amendment, with microbial
growth monitored by OD600. Cultures of 2 L and 500 mL (for amino
acid amendment) volumes were grown. Cultures were harvested in
the late log phase of growth by centrifugation (SLA-3000 rotor,
ThermoScientific) at 6000 rpm for 30 min at 4 °C. Culture
supernatants were decanted and shaken with 100 g of XAD-2
polystyrene resin for 3 h at 4 °C to adsorb organics. The resin was
filtered from the supernatant, washed with 250 mL of 90/10%
ddH2O/MeOH, and eluted with 300 mL of 10/90% ddH2O/MeOH.
The eluent was concentrated under vacuum to 40 mL and stored at 4
°C for analysis. Eluent was further purified by semipreparative HPLC
on a YMC 20 × 250 mm C18-AQ column, with a linear gradient of
10% MeOH in ddH2O (+0.1% trifluoroacetic acid) to 30% MeOH in
ddH2O (+0.1% trifluoroacetic acid) over 40 min, yielding pure
product (30.02 mg from 2 L of culture).
OH-Asp residue (Figure 8), consistent with photoproducts of
other β-OH-Asp siderophores.12,15,27 Further investigations are
focused on elucidating the mechanism of pacifibactin photo-
oxidation, including identification of intermediates in the
reaction.
CONCLUSIONS
■
Characterization of pacifibactin revealed a highly unusual
siderophore with marked structural differences from the
predictions inferred through initial bioinformatic analysis.
Two β-OH-Asp functional groups combined with two Orn-
derived hydroxamate functional groups yield four bidentate
metal binding sites, a very uncommon feature in siderophores.
UV−visible spectrophotometry of Fe(III)-pacifibactin and 13C
NMR spectroscopy of Ga(III)-pacifibactin are consistent with
metal coordination by both β-OH-Asp residues and the N5-
acetyl-N5-OH-Orn residue, thus leaving a free cyclized N5-
OH-Orn residue. Owing to the coordination of Fe(III) by β-
OH-Asp, UV irradiation of Fe(III)-pacifibactin triggers a
photooxidative breakdown of the ligand, as evinced through
shifts in UV−visible absorbance.
The structure of pacifibactin highlights several limitations to
existing NRPS cluster analysis programs. While programs such
as antiSMASH identify and annotate NRPS domains within a
gene cluster, both noncanonical NRPS domains and tailoring
enzymes acting externally from the NRPS proteins elude
identification. Initial bioinformatic analysis predicted incorpo-
ration of two Asp residues into pacifibactin; however a more
exhaustive manual analysis of the gene cluster identified an
NRPS domain and a standalone enzyme that are each likely
responsible for aspartic acid β-hydroxylation. Similarly,
antiSMASH predicted incorporation of two OH-Orn residues,
while manual analysis identified an N-acetyl transferase likely
responsible for N5-acetylation of N5-OH-Orn. This extended
manual analysis of the gene cluster allowed a refinement of the
initial structure prediction of the corresponding natural
product. Structural characterization of pacifibactin then
confirmed the accuracy of the refinements, as pacifibactin
indeed incorporates two β-OH-Asp residues and one N5-
acetyl-N5-OH-Orn. The characterization of pacifibactin makes
F
DOI: 10.1021/acs.jnatprod.8b01073
J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
1
Pacifibactin: white solid; [α]18 −63 (c 0.100, MeOH); H and
After photolysis, the reaction mixture was analyzed by UPLC-ESIMS/
MS to detect and characterize any photoproducts.
D
13C NMR data, Table 1; HRESIMS m/z 923.4081 [M + H]+ (calcd
for C34H59N12O18, 923.4065).
ASSOCIATED CONTENT
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.jnat-
prod.8b01073.
Amino Acid Analysis of Pacifibactin with Marfey’s Reagent.
Purified apo pacifibactin (2 mg) was dissolved in 6 M HCl, sealed in
an ampule under argon, and heated at 80 °C for 8 h to hydrolyze the
siderophore. The hydrolysis mixture was evaporated to dryness to
remove HCl and redissolved in ddH2O. After two additional cycles of
evaporation and dissolution in ddH2O, the hydrolysis mixture was
derivatized with 1-fluoro-2−4-dinitrophenyl-5-L-alanine amide (Mar-
fey’s reagent) using standard procedures.36 The hydrolysis procedure
was also performed as described using 45% HI in place of 6 M HCl to
reduce any N5-acetyl-N5-hydroxyornithine and N5-hydroxyornithine
to ornithine to aid in analysis. Derivatized hydrolysis products were
separated by HPLC on a YMC 4.6 × 250 mm C18-AQ column with a
gradient from 15% CH3CN in ddH2O (0.05% trifluoroacetic acid) to
50% CH3CN in ddH2O (0.05% trifluoroacetic acid) over 60 min.
Derivatized hydrolysis products were co-injected with standards of
Marfey’s derivatized amino acids to determine the constituent amino
acids of pacifibactin: D,L-threo-β-OH-Asp (Sigma-Aldrich), L-Ser
(Alfa-Aesar), D-Arg (Alfa-Aesar), D-Orn (Sigma-Aldrich), L-Orn
(Sigma-Aldrich). D,L-erythro-β-OH-Asp was synthesized through
treatment of 2,3-trans-expoxysuccinic acid (50 mg) with 375 μL of
concentrated aqueous NH4OH (28%).33 The reaction was sealed in a
glass ampule and heated for 20 h at 50 °C. The crude mixture was
dried, then dissolved in 1.5 mL of ddH2O. The product was then
derivatized with Marfey’s reagent, and the formation of derivatized
D,L-erythro-β-OH-Asp as the dominant product was confirmed by
UPLC-ESIMS, noting the mass of the derivatized amino acid and the
difference in retention time in comparison to the D,L-threo-β-OH-Asp-
derivatized standard.
■
S
Gene cluster annotation, ESIMS/ESIMS/MS spectra,
NMR spectra, UV−visible spectra from titration and
photolysis, HPLC chromatograms from amino acid
analysis (PDF)
AUTHOR INFORMATION
Corresponding Author
*E-mail: butler@chem.ucsb.edu.
■
ORCID
Alison Butler: 0000-0002-3525-7864
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We are grateful for support from NSF CHE-1710761. C.D.H.
thanks the UCSB Mellichamp Academic Initiative in
Sustainability for a fellowship in support of his research. This
work made use of the MRL Shared Experimental Facilities
supported by the MRSEC Program of the NSF, DMR
1121053. We thank R. Behrens (UPLC-MS) and H. Zhou
(NMR) for technical help.
Fe(III) Titration of Pacifibactin. A 2.12 mM stock solution of
Fe(III) was prepared by diluting a 1 mg/mL Fe(NO3)3 atomic
absorption standard solution with ddH2O and standardized
spectrophotometrically with 1,10-phenanthroline using established
procedures.37 A stock solution of apo pacifibactin was prepared by
dissolving freeze-dried siderophore in ddH2O. To standardize the
pacifibactin stock solution, a 400 μL aliquot of apo pacifibactin stock
solution was lyophilized then dissolved with 2.77 mg of dried maleic
REFERENCES
■
(1) Jensen, P. R. Trends Microbiol. 2016, 24 (12), 968−977.
(2) Ziemert, N.; Alanjary, M.; Weber, T. Nat. Prod. Rep. 2016, 33
(8), 988−1005.
(3) Koglin, A.; Walsh, C. T. Nat. Prod. Rep. 2009, 26 (8), 987−1000.
(4) Sattely, E. S.; Fischbach, M. A.; Walsh, C. T. Nat. Prod. Rep.
2008, 25 (4), 757−793.
1
acid in 700 μL of 99% (CD3)2SO. H NMR peak integrations of
pacifibactin and the maleic acid internal standard were then taken,
establishing a stock concentration of 3.5 mM. A solution of apo
pacifibactin (2 mL, 0.1 mM in ddH2O buffered with 100 mM MOPS
pH 7.1) was prepared in a 3 mL quartz cuvette and titrated with the
standardized Fe(III) stock solution (2.14 mM in 40 mM HNO3).
After each aliquot of Fe(III) was added, the solution was allowed to
equilibrate for 24 h, a period determined by monitoring changes in
the UV−vis absorption spectrum of the solution after Fe(III)
addition. After equilibration, the UV−vis absorbance spectrum of
the solution was measured.
Preparation of Ga(III)-Pacifibactin. A stock solution of Ga(III)
was prepared by dissolving a gallium metal ingot in boiling 30%
HNO3 for 24 h. The resultant Ga(NO3)3 stock solution was
standardized by colorimetric titration with EDTA (pyrocatechol
violet indicator).38 Ga(III)-bound pacifibactin was prepared by
adding Ga(III) stock solution to a solution of apo pacifibactin at a
4:3 molar ratio in ddH2O (pH adjusted to 7.0 with NaOH addition)
and equilibrating for 48 h. The solution was then purified by solid-
phase extraction using a Waters C18 SepPak to remove any excess
Ga(III) and salts.
Photolysis of Fe(III)-Pacifibactin. Fe(III)-bound pacifibactin
was prepared by adding Fe(III) stock solution to a solution of apo
pacifibactin at a 1:1 ratio for a final concentration of 0.1 mM Fe(III)-
pacifibactin in 100 mM MOPS pH 7.1 buffer (2 mL) in a 3 mL quartz
cuvette. The solution was equilibrated for 24 h before photolysis.
Solutions were photolyzed using a 450 W mercury arc lamp (Ace
Glass, 40−48% output in UV) as the light source, with cuvettes placed
on a stand approximately 15 cm away from the lamp. UV−vis
absorbance spectra of the solutions were measured at timed intervals.
(5) Crosa, J. H.; Walsh, C. T. Microbiol. Mol. Biol. Rev. 2002, 66 (2),
223−249.
(6) Samel, S. A.; Marahiel, M. A.; Essen, L.-O. Mol. BioSyst. 2008, 4
(5), 387−393.
(7) Blin, K.; Wolf, T.; Chevrette, M. G.; Lu, X.; Schwalen, C. J.;
Kautsar, S. A.; Suarez Duran, H. G.; de los Santos, E. L. C.; Kim, H.
U.; Nave, M.; Dickschat, J. S.; Mitchell, D. A.; Shelest, E.; Breitling,
R.; Takano, E.; Lee, S. Y.; Weber, T.; Medema, M. H. Nucleic Acids
Res. 2017, 45 (W1), W36−W41.
(8) Skinnider, M. A.; Dejong, C. A.; Rees, P. N.; Johnston, C. W.; Li,
H.; Webster, A. L. H.; Wyatt, M. A.; Magarvey, N. A. Nucleic Acids
Res. 2015, 43 (20), 9645−9662.
(9) Rosconi, F.; Davyt, D.; Martínez, V.; Martínez, M.; Abin-
Carriquiry, J. A.; Zane, H.; Butler, A.; de Souza, E. M.; Fabiano, E.
Environ. Microbiol. 2013, 15 (3), 916−927.
(10) Bosello, M.; Zeyadi, M.; Kraas, F. I.; Linne, U.; Xie, X.;
Marahiel, M. A. J. Nat. Prod. 2013, 76 (12), 2282−2290.
(11) Zane, H. K.; Naka, H.; Rosconi, F.; Sandy, M.; Haygood, M.
G.; Butler, A. J. Am. Chem. Soc. 2014, 136 (15), 5615−5618.
(12) Kurth, C.; Schieferdecker, S.; Athanasopoulou, K.; Seccareccia,
I.; Nett, M. J. Nat. Prod. 2016, 79 (4), 865−872.
(13) Hermenau, R.; Ishida, K.; Gama, S.; Hoffmann, B.; Pfeifer-
Leeg, M.; Plass, W.; Mohr, J. F.; Wichard, T.; Saluz, H.-P.; Hertweck,
C. Nat. Chem. Biol. 2018, 14 (9), 841−843.
(14) Kem, M. P.; Zane, H. K.; Springer, S. D.; Gauglitz, J. M.; Butler,
A. Metallomics 2014, 6 (6), 1150−1155.
(15) Barbeau, K.; Rue, E. L.; Bruland, K. W.; Butler, A. Nature 2001,
413, 409−413.
G
DOI: 10.1021/acs.jnatprod.8b01073
J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products
Article
(16) Lai, Q.; Wang, L.; Liu, Y.; Fu, Y.; Zhong, H.; Wang, B.; Chen,
L.; Wang, J.; Sun, F.; Shao, Z. Int. J. Syst. Evol. Microbiol. 2011, 61 (6),
1370−1374.
(17) Schwyn, B.; Neilands, J. B. Anal. Biochem. 1987, 160 (1), 47−
56.
(18) Quadri, L. E. N.; Sello, J.; Keating, T. A.; Weinreb, P. H.;
Walsh, C. T. Chem. Biol. 1998, 5 (11), 631−645.
(19) Chevrette, M. G.; Aicheler, F.; Kohlbacher, O.; Currie, C. R.;
Medema, M. H. Bioinformatics 2017, 33 (20), 3202−3210.
(20) Franke, J.; Ishida, K.; Hertweck, C. Chem. - Eur. J. 2015, 21
(22), 8010−8014.
(21) Franke, J.; Ishida, K.; Ishida-Ito, M.; Hertweck, C. Angew.
Chem., Int. Ed. 2013, 52 (32), 8271−8275.
(22) Johnston, C. W.; Skinnider, M. A.; Wyatt, M. A.; Li, X.; Ranieri,
M. R. M.; Yang, L.; Zechel, D. L.; Ma, B.; Magarvey, N. A. Nat.
Commun. 2015, 6, 8421.
(23) Kreutzer, M. F.; Nett, M. Org. Biomol. Chem. 2012, 10 (47),
9338−9343.
(24) Heydel, P.; Plattner, H.; Diekmann, H. FEMS Microbiol. Lett.
1987, 40 (2−3), 305−309.
(25) Robbel, L.; Helmetag, V.; Knappe, T. A.; Marahiel, M. A.
Biochemistry 2011, 50 (27), 6073−6080.
(26) Singh, G. M.; Fortin, P. D.; Koglin, A.; Walsh, C. T.
Biochemistry 2008, 47 (43), 11310−11320.
(27) Kreutzer, M. F.; Kage, H.; Nett, M. J. Am. Chem. Soc. 2012, 134
(11), 5415−5422.
(28) Gauglitz, J. M.; Butler, A. JBIC, J. Biol. Inorg. Chem. 2013, 18
(5), 489−497.
(29) Feistner, G. J.; Stahl, D. C.; Gabrik, A. H. Org. Mass Spectrom.
1993, 28 (3), 163−175.
(30) Kupper, F. C.; Carrano, C. J.; Kuhn, J.-U.; Butler, A. Inorg.
̈
Chem. 2006, 45 (15), 6028−6033.
(31) Amin, S. A.; Green, D. H.; Kupper, F. C.; Carrano, C. J. Inorg.
̈
Chem. 2009, 48 (23), 11451−11458.
(32) Butler, A.; Theisen, R. M. Coord. Chem. Rev. 2010, 254 (3),
288−296.
(33) Robertson, A. W.; McCarville, N. G.; MacIntyre, L. W.; Correa,
H.; Haltli, B.; Marchbank, D. H.; Kerr, R. G. J. Nat. Prod. 2018, 41,
858.
(34) Hardy, C. D.; Butler, A. JBIC, J. Biol. Inorg. Chem. 2018, 23,
957.
(35) Lai, Q.; Shao, Z. J. Bacteriol. 2012, 194 (24), 6936.
(36) Marfey, P. Carlsberg Res. Commun. 1984, 49 (6), 591.
(37) Fortune, W. B.; Mellon, M. G. Ind. Eng. Chem., Anal. Ed. 1938,
10, 60−64.
(38) Loomis, L. D.; Raymond, K. N. J. Coord. Chem. 1991, 23 (1−
4), 361−387.
H
DOI: 10.1021/acs.jnatprod.8b01073
J. Nat. Prod. XXXX, XXX, XXX−XXX