Total synthesis and structure revision of petrobactin

Article abstract of DOI:10.1016/S0040-4020(03)00103-0

The total synthesis and the revised structural assignment of petrobactin, a siderophore isolated from the marine bacterium Marinobacter hydrocarbonoclasticus, is reported. The key step in the synthesis involved condensation of N¹-(2,3-dibenzoyl

Full text of DOI:10.1016/S0040-4020(03)00103-0

Tetrahedron 59 (2003) 2007–2014
Total synthesis and structure revision of petrobactin
†
,
a
Raymond J. Bergeron,a Guangfei Huang, Richard E. Smith,a Neelam Bharti,a
,
*
James S. McManisa and Alison Butlerb
aDepartment of Medicinal Chemistry, University of Florida, Gainesville, Florida 32610 USA
bDepartment of Chemistry and Biochemistry, University of California, Santa Barbara, California 93106 USA
Received 15 November 2002; revised 10 January 2003; accepted 10 January 2003
Abstract—The total synthesis and the revised structural assignment of petrobactin, a siderophore isolated from the marine bacterium
Marinobacter hydrocarbonoclasticus, is reported. The key step in the synthesis involved condensation of N 1-(2,3-dibenzoyloxybenzoyl)-
N 4-benzylspermidine with 1,3-di-(p-nitrophenyl)-2-tert-butyl citrate. Proton NMR spectra of the synthesized product compared with those
reported for the natural product revealed that the compound did not contain 2,3-dihydroxybenzoyl moieties as published; instead, the splitting
pattern suggested 3,4-dihydroxybenzoyl fragments. The 3,4-dihydroxybenzoyl analogue was accessed via a similar route; the proton and
carbon-13 NMR spectra of this compound were consistent with those reported for natural petrobactin. q 2003 Elsevier Science Ltd. All
rights reserved.
1. Introduction
compliance is a problem.15–17 Both synthetic ligands18 and
natural products19 have been targeted in the search for more
efficient and orally deliverable therapeutic agents. The
current paper focuses on the total synthesis of a natural
product ligand, petrobactin, which is a candidate in these
investigations.
Iron serves as a prosthetic for many different redox systems,
including enzymes1 and cytochromes,2 which are essential
for life itself. Although iron comprises about 5% of the
earth’s crust, it is not easily accessible to living systems. In
the biosphere, iron primarily exists as Fe(III), which forms
highly insoluble ferric hydroxide polymers under physio-
logical conditions, Ksp ¼ 2 £ 10239; corresponding to a
solution concentration of free Fe(III) of approximately
1£10219 M in this pH range.3,4 Because of this insolubility,
most life forms have developed specific ligands to sequester
and manage the metal. The bacteria assemble relatively low-
molecular-weight, virtually ferric iron-specific, chelators,
siderophores, which they secrete into the surroundings.5–8
These ligands sequester Fe(III) and render it utilizable by
the microorganism.
Petrobactin is a siderophore that was isolated from an oil-
degrading marine bacterium, Marinobacter hydrocarbono-
clasticus.20,21 It is a hexacoordinate ligand and forms a 1:1
complex with Fe(III). Although not confirmed by its
synthesis, the initially reported structure of petrobactin
(Fig. 1) suggested that the compound was predicated on a
citrate bis-spermidine backbone, the 2,3-dihydroxybenzoyl
In humans, a number of disease states have been identified
in which iron transport and storage systems are over-
whelmed with excess metal; significant damage to tissues
and organs often occurs.9,10 Left untreated, these diseases
can be fatal; the management of these iron overload
syndromes requires chelation therapy. Interestingly, the
treatment of choice for iron overload is a hydroxamate
siderophore, desferrioxamine,11–14 but because of the poor
efficiency of the drug and the required time-consuming (10
to 12 h/day, 5 or 6 days/week) infusion of the drug, patient
Keywords: siderophore; Marinobacter hydrocarbonoclasticus; synthesis;
citrate; spermidine.
*
Corresponding author. Tel.: þ1-352-846-1956; fax: þ1-352-392-8406;
e-mail: bergeron@mc.cop.ufl.edu
Present address: Alchem Laboratories Corporation, 13305 Rachael Blvd.,
Alachua, Florida 32615, USA.
†
Figure 1. Retrosynthetic schematic of the original structure of petrobactin.
0040–4020/03/$ - see front matter q 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0040-4020(03)00103-0

2008
R. J. Bergeron et al. / Tetrahedron 59 (2003) 2007–2014
Scheme 1. Total synthesis of petrobactin: generation of the 2,3- and 3,4-dihydroxybenzoyl compounds.
moieties providing four of the requisite six donor groups for
Fe(III). This latter functional group is found in a number of
well-characterized iron chelators, most notably L-parabac-
tin, which is isolated from Paracoccus denitrificans.22
Interestingly, petrobactin participates in a photolytic ligand-
to-metal charge-transfer reaction, leading to decarboxyla-
tion and likely reduction of Fe(III) to Fe(II).23 This occurs in
solution when exposed to sunlight. Such Fe(III)-mediated
photolytic decarboxylations of a-hydroxy carboxylic acids
are well-known.24,25 Although this may play a role in how
the ligand processes the metal, further investigation is
needed. The availability of either radiolabeled ligand or
greater quantities of the compound would help address some
of these issues. Of course, facile access to relatively large
amounts of the chelator is requisite if any iron clearance
studies are planned in animals.
that petrobactin utilizes a 3,4-dihydroxybenzoyl fragment,
rather than a 2,3-dihydroxybenzoyl donor. No hexadentate
iron chelator from a natural source containing the 3,4-
dihydroxybenzamide moiety has been reported previously,
in spite of the fact that 3,4-dihydroxybenzoic acid binds
iron(III) more tightly than its 2,3-isomer at physiological
pH.26
2. Results and discussion
2.1. Synthetic design considerations
From a retrosynthetic standpoint (Fig. 1), the original
structure of petrobactin23 is defined by a central citrate
moiety, two spermidine fragments, and two terminal 2,3-
dihydroxybenzoyl groups. From the perspective of synthetic
design, the asymmetry of spermidine, that is, its 3-
aminopropyl and 4-aminobutyl limbs, suggests a selectively
protected spermidine fragment.27 Substitution at the term-
inal carboxyls of citrate warrants a central citric acid
fragment protected at the central carboxyl.
The current study focuses on synthetic approaches to
petrobactin and related compounds. Comparison of the
proton NMR of a 2,3-dihydroxybenzamide model com-
pound with the published assignment23 suggested that the
bidentate donor fragment was not, in fact, a 2,3-dihydroxy-
benzoyl moiety. Closer review of the spectra of the isolated
material revealed that the distinct splitting in the aromatic
region characteristic of a 2,3-dihydroxybenzoyl fragment
was absent. Interestingly, the splitting pattern more nearly
approximated that found in 3,4-dihydroxybenzoyl-substi-
tuted systems. Total synthesis of both the 2,3- and 3,4-
dihydroxybenzoyl compounds unequivocally demonstrates
2.2. Synthesis of ‘petrobactin’, original structure
The key step in the synthesis of what was initially reported
as petrobactin (8, Scheme 1) involved the condensation of
the citrate triester 1,3-di-(p-nitrophenyl)-2-tert-butyl
citrate28 with N 1-(2,3-dibenzyloxybenzoyl)-N 4-benzyl-

R. J. Bergeron et al. / Tetrahedron 59 (2003) 2007–2014
Table 1. 13C and 1H resonances for the synthetic 2,3-dihydroxy compound (8)
2009
Atom
Found DMSO 13
C
Found DMSO 1H mult
(J in Hz)
Found DMSO:D2O ca 4:1 1H mult
(J in Hz)
C1 (C29)
C2 (C30)
C3 (C31)
C4 (C32)
C5 (C33)
C6 (C34)
C7 (C28)
C8 (C27)a
C9 (C26)a
C10 (C25)a
C11 (C24)a
C12 (C23)a
C13 (C22)a
C14 (C21)a
C15 (C20)
C16 (C19)
–
115.06
146.21
149.40
118.83
117.98
117.22
169.90
36.16
26.01
44.74
46.47
25.78
22.96
37.67
169.52
43.30
–
73.44
174.99
na
na
na
na
na
na
na
na
na
na
na
na
na
6.92 dd (1, 7.8)
6.71 dd (7.8, 8.1)
7.18 dd (1, 8.1)
na
3.26 t (6.6)
1.83 m
2.86 m
2.86 m
1.39 m
1.52 m
3.01 t (6.8)
na
2.46 d (14.6)
2.56 d (14.6)
na
na
ex
ex
ex
no
ex
ex
no
6.92 dd (1.3, 7.8)
6.69 dd (7.8, 8.1)
7.26 dd (1.3, 8.1)
na
3.36 m
1.86 m
2.92 m
2.92 m
1.42 m
1.55 m
3.04 m
na
2.50 d (14.5)
2.58 d (14.5)
na
C17
C18
na
N1 (N6)
N2 (N5)
N3 (N4)
OHb
8.86 t (5.6)
8.33 br m
7.97 t (5.6)
no
9.20 s
5.52 br s
no
OHc
OHd
OHe
br¼broad, d¼doublet, dd¼doubled doublet, ex¼exchanged, m¼multiplet, mult¼multiplicity, na¼not applicable, no¼not observed, nr¼not reported,
s¼singlet, t¼triplet.
a
1H chemical shifts are the center of the observed multiplet.
OH of C2 and C30, not observed within spectral width of 22 to 11 ppm.
OH of C3 and C31, tentative assignment.
OH of C17.
Carboxylic OH of C18.
b
c
d
e
spermidine (6). The assembly of the spermidine reagent (6)
began with cyanoethylation of N 1-benzyl-N 4-(tert-butoxy-
carbonyl)-1,4-diaminobutane (1)29 to afford nitrile 2 in
quantitative yield. Exposure of 2 to hydrogen over Raney
nickel reduced the nitrile to provide N 4-benzyl-N 8-(tert-
butoxycarbonyl)spermidine (3) in 96% yield. This amine
was acylated at the N1-position with 2,3-dibenzyloxy-
benzoyl chloride (4),30,31 producing the fully protected ‘side
fragments’ N 1-(2,3-dibenzyloxybenzoyl)-N 4-benzyl-N 8-
(tert-butoxycarbonyl)spermidine (5) in 74% yield. Removal
of the tert-butoxycarbonyl group of 5 with trifluoroacetic
acid (TFA) generated the synthon N 1-(2,3-dibenzyloxy-
benzoyl)-N 4-benzylspermidine (6) in 73% yield.
2.3. Comparison of the proton NMR spectrum of the
synthetic product with that of the reported molecule;
synthesis of the 3,4-dihydroxy compound
Unfortunately, the 500-MHz 1H NMR spectrum of the 2,3-
dihydroxy synthetic product (8) differed significantly from
that reported for the siderophore.23 The observed aromatic
proton resonances and their associated coupling constants
for the synthetic product were 6.69 ppm (7.8 and 8.1 Hz),
6.92 ppm (1.3 and 7.8 Hz) and 7.26 ppm (1.3 and 8.1 Hz).
The two larger coupling constants for the 6.69 ppm
resonance were consistent with two ortho couplings,34
confirming assignment of this resonance to position 5 of the
aromatic rings. The two remaining resonances displayed
both a small meta and a large ortho coupling, indicating that
these protons flank the 5 position. Comparison of the
observed shifts with those predicted for a model compound,
N-methyl-2,3-dihydroxybenzamide,‡ permitted assignment
of the 7.26 ppm resonance to position 6 and the 6.92 ppm
resonance to position 4. With the aromatic protons assigned,
the aromatic 13C resonances were easily assignable from the
Coupling of the amine (6, 2 equiv.) with the citrate triester
was accomplished in acetonitrile and N,N-diisopropylethyl-
amine (DIEA), providing the protected skeleton of the target
ligand (7) in 71% yield. The tert-butyl ester 7 was cleaved to
the corresponding acid, 1,3-bis[N 8-(N 1-2,3-dibenzyloxy-
benzoyl-N 4-benzyl)spermidinyl] citrate, with TFA in
CH2Cl2. Finally, the six benzyl protecting groups were
removed by hydrogenolysis over PdCl2 to provide the
putative siderophore (8) in 17% calculated yield for the two
steps. The synthetic sequence proceeded efficiently except
for the unmasking of 7 to chelator 8. A major byproduct,
which was separated by HPLC, resulted from cyclization of
8 to the five-membered imide on the basis of NMR analysis.
This instability is inherent in citrate-derived amides such as
the siderophores rhizoferrin32 and acinetoferrin.33
1H–13C HMQC spectrum. Table 1 lists observed 1H and 13
C
resonances and their assignments for the synthetic 2,3-
dihydroxy product (8).
Chemical shifts and coupling constants as listed for the
‡
ChemNMR Pro ver. 1.0 as implemented in ChemDraw Pro ver. 4.5,
CambridgeSoft Corp, Cambridge, MA

2010
R. J. Bergeron et al. / Tetrahedron 59 (2003) 2007–2014
Table 2. Comparison of reported petrobactin23 and synthetic petrobactin 13C and 1H resonances
Atom
Literature
DMSO 13C
Synthetic found
DMSO 13
Literature
DMSO 1H
(J in Hz)
Synthetic found
DMSO 1H mult
(J in Hz)
Synthetic found
DMSO:D2O ca. 4:1 1H mult
(J in Hz)
C
C1 (C29)
C2 (C30)
C3 (C31)
C4 (C32)a
C5 (C33)a
C6 (C34)
C7 (C28)
C8 (C27)b
C9 (C26)b
C10 (C25)b
C11 (C24)b
C12 (C23)b
C13 (C22)b
C14 (C21)b
C15 (C20)
C16 (C19)
–
125.37
148.47
144.86
114.81
118.95
115.07
166.65
36.15
26.23
44.77
46.45
26.03
22.98
37.69
169.54
43.32
–
73.47
175.02
na
na
na
na
na
na
na
125.36
na
na
na
na
na
7.22 d (2.1)
na
na
6.76 d (8.3)
7.15 dd (2.1, 8.3)
na
3.26 t (6.6)
1.78 m
2.86 m
2.86 m
1.40 m
1.52 m
3.01 t (6.8)
na
2.47 d (14.3)
2.56 d (14.3)
na
na
ex
ex
ex
ex
ex
ex
no
115.04
144.84
148.45
114.79
118.93
166.65
36.13
26.21
44.76
46.44
26.00
22.95
37.67
169.53
43.30
–
73.45
174.98
na
na
na
na
na
na
na
7.28 d (1.9)
na
na
6.76 d (8.2)
7.18 dd (1.9, 8.2)
na
3.28 m
1.80 m
2.90 m
2.90 m
1.42 m
1.55 m
3.04 m
na
6.76 (10)
7.18 (2, 10)
7.28 (2)
na
3.27
1.80
2.90
2.90
1.43
1.55
3.04
na
2.50
2.58
na
2.50 d (14.5)
2.59 d (14.5)
na
C17
C18
na
na
N1 (N6)
N2 (N5)b
N3 (N4)
OHc
8.31
8.41
8.01
nr
nr
nr
8.30 t (5.7)
8.37 m
7.98 t (5.6)
9.10 s
9.53 s
5.53 br s
no
OHd
OHe
OHf
nr
br¼broad, d¼doublet, dd¼doubled doublet, ex¼exchanged, m¼multiplet, mult¼multiplicity, na¼not applicable, no¼not observed, nr¼not reported,
s¼singlet, t¼triplet.
a
Literature values reflect correction to the apparently transposed 1H–1H J values listed in Table 1 of Ref. 23.
b
1H chemical shifts are the center of the observed multiplet.
OH of C3 and C31.
c
d
OH of C4 and C32.
OH of C17.
Carboxylic OH of C18.
e
f
aromatic region of the natural product, however, were
6.76 ppm (2 Hz), 7.18 ppm (2 and 10 Hz), and 7.28 (10 Hz)
and were assigned to positions 4, 5 and 6, respectively.23
But the observation of one large and one small coupling
constant suggests two protons ortho to one another (at 7.18
and 7.28 ppm) with an isolated proton (6.76 ppm) meta to
the 7.18 ppm resonance and para to the 7.28 ppm resonance.
Chemical shift predictions§ of the model compound N-
methyl-2,4-dihydroxybenzamide, together with the reported
coupling data, suggested a 2,4-dihydroxy arrangement. This
result seemed anomalous—as nonadjacent hydroxyl posi-
tioning would be unlikely for an efficient iron chelator.
Upon review of a plot trace of the 1H 500 MHz spectrum of
the natural product, it was apparent that the coupling
constants for the 6.76 and 7.28 ppm resonances had been
transposed in the initial report.23 When the chemical shifts
were matched to the correct coupling constants and shift-
structure correlations were reconsidered, 3,4-dihydroxy
substitution provided the best fit.
similar to those obtained for the 2,3-dihydroxy compound,
except for the final step, in which direct HPLC purification
of the crude product afforded 13 in much greater yield
(61%).
As shown in Table 2, both the 1H (500 MHz) and 13C
(125 MHz) NMR spectra of the synthetic 3,4-dihydroxy-
benzoyl compound were identical to those of the naturally
occurring siderophore. The major impurity present in the
crude synthetic petrobactin was also isolated during final
HPLC purification. Its retention time was longer than that of
synthetic petrobactin, suggesting a less polar structure;
proton NMR and HRMS analyses were consistent with the
imide structure 14 (Fig. 2). The proton NMR spectrum
clearly showed a loss of symmetry as evidenced by
To definitively settle the question, the assembly of the 3,4-
dihydroxy petrobactin analogue (13) was initiated. Simply
substituting 3,4-dibenzyloxybenzoyl chloride (9)30,35 for
2,3-dibenzyloxybenzoyl chloride in the synthetic route
(Scheme 1) provided the target compound. Yields were
§
ChemNMR Pro ver. 1.0 as implemented in ChemDraw Pro ver. 4.5,
CambridgeSoft Corp, Cambridge, MA
Figure 2. Structure of the imide (14) formed upon cyclization of 13.

R. J. Bergeron et al. / Tetrahedron 59 (2003) 2007–2014
2011
replacement of the two citrate doublets of synthetic
petrobactin (2.50 and 2.59 ppm) with four one-hydrogen
doublets (2.59, 2.70, 2.83 and 2.87 ppm).33 Additionally,
the high resolution mass spectrum indicated a parent (plus
hydrogen) ion of 701 m/z, consistent with the loss of a
molecule of water from synthetic petrobactin in forming the
cyclic imide 14. The kinetics of the petrobactin cyclization
are currently under investigation. A better understanding of
this reaction will help increase the overall efficiency of the
synthetic route.
C19H29N3O2: C, 68.85; H, 8.82; N, 12.68. Found: C,
68.76; H, 8.73; N, 12.76.
4.2.2. N 4-Benzyl-N 8-(tert-butoxycarbonyl)spermidine
(3). Raney Ni (4.45 g, 50% slurry in water) was added to
a solution of 2 (6.63 g, 20.0 mmol) in EtOH (25 mL) and 1N
NaOH (190 mL) in 95% EtOH in a Parr bottle; the slurry
was shaken overnight under a hydrogen atmosphere
maintained at 2–3 atm. The mixture was filtered, solids
were washed with 95% EtOH (50 mL), and the combined
filtrate was concentrated under reduced pressure. Flash
chromatography of the residue, eluting with 3% concen-
trated NH4OH in MeOH, gave 3 (6.42 g, 96%) as a colorless
3. Conclusion
1
oil. H NMR d 1.44 (9H, s, (CH3)3C), 1.45–1.60 (8H, m,
NH–CH2–CH2–CH2–CH2 and CH2–CH2–NH2), 2.38–
2.48 (4H, m, N–CH2 and CH2–NH2), 2.70 (2H, t,
J¼6.81 Hz, N–CH2–CH2–CH2–NH2), 3.05–3.15 (2H,
m, NH–CH2), 3.53 (2H, s, CH2–Ar), 4.68 (1H, br s, CO–
NH), 7.15–7.25 (5H, m, Ar). HRMS: found (MþH)þ,
336.2648. C19H34N3O2 requires (MþH) 336.2651. Analysis
calcd for C19H33N3O2·0.5H2O: C, 66.24; H, 9.95; N, 12.20.
Found: C, 66.22; H, 9.93; N, 12.20.
The structure of petrobactin has been revised and confirmed
by its synthesis and that of its 2,3-dihydroxy analogue.
Thus, the first preparation of a naturally occurring
hexacoordinate chelator containing 3,4-dihydroxybenza-
mides has been accomplished. The method described here
offers efficient access to petrobactin as well as analogues
differing in the arrangement of the aromatic hydroxyl
groups. Additionally, the spermidine reagent 6 provides a
useful synthon for producing the purported photoinduced
oxidative breakdown product, 1,5-bis[N 8-(N 1-2,3-dihy-
droxybenzoyl)spermidinyl]-3-oxoglutarate23 and its 3,4-
dihydroxy isomer.
4.2.3. N 1-[2,3-Bis(benzyloxy)benzoyl]-N 4-benzyl-N 8-
(tert-butoxycarbonyl)spermidine (5). A solution of oxalyl
chloride (7.28 g, 57.3 mmol) in toluene (25 mL) was added
dropwise to an ice-bath cooled slurry of 2,3-bis(benzyloxy)-
benzoic acid30,31 (4.85 g, 14.5 mmol) and dimethylforma-
mide (DMF) (0.2 mL) in toluene (50 mL); the mixture was
stirred for 1 h. The ice bath was removed; stirring was
continued at rt for 1 h. Volatiles were removed under high
vacuum, and 4 was dissolved in CH2Cl2 (50 mL) and added
dropwise to an ice-bath cooled solution of 3 (3.24 g,
9.66 mmol) and DIEA (13.75 g, 106 mmol) in CH2Cl2
(125 mL). The mixture was allowed to warm to rt and was
stirred overnight. The solution was washed with water
(2£150 mL) and brine (2£150 mL), dried over Na2SO4, and
filtered; solvents were removed under reduced pressure.
Flash chromatography of the residue, eluting with 1:1.5
hexane/EtOAc, gave 5 (4.66 g, 74%) as a pale yellow oil. 1H
NMR d 1.36–1.41 (4H, m, NH–CH2–CH2–CH2–CH2),
1.43 (9H, s, (CH3)3C), 1.45–1.58 (2H, m, N–CH2–CH2–
CH2–NH), 2.28–2.37 (4H, m, CH2–N–CH2), 2.94–3.06
(2H, m, OCO–NH–CH2), 3.24–3.34 (2H, m, Ar–CO–
NH–CH2), 3.48 (2H, s, N–CH2–Ar), 4.68 (1H, br m,
OCO–NH), 5.02 (2H, s, O–CH2–Ar), 5.08 (2H, s, O–
CH2–Ar), 7.10–7.50 (17H, m, Ar), 7.66–7.73 (1H, m, Ar),
7.92 (1H, br t, J¼5.28 Hz, Ar–CO–NH). HRMS: found
(MþH)þ, 652.3735. C40H50N3O5 requires (MþH)
652.3750. Analysis calcd for C40H49N3O5: C, 73.70; H,
7.58; N, 6.45. Found: C, 73.44; H, 7.69; N, 6.45.
4. Experimental
4.1. General details
All reactions were performed under a nitrogen atmosphere
at ambient pressure unless otherwise indicated. Acryloni-
trile and trifluoroacetic acid (TFA) were freshly distilled
prior to use. Percent NH4OH solutions are expressed as
volume solutions of 30% (by weight) NH4OH. Except
where stated, 1H NMR spectra were obtained at 300 MHz in
CDCl3 and 13C NMR at 75 MHz in CDCl3; chemical shifts
(d) are given in ppm downfield from tetramethylsilane
(TMS) as standard.
4.2. Synthesis of ‘petrobactin’, original structure (8)
4.2.1. N-Benzyl-N-(2-cyanoethyl)-N0-(tert-butoxycarbo-
nyl)-1,4-butanediamine (2). Acrylonitrile (4.03 g,
76 mmol) was added to a solution of 129 (6.75 g,
24.2 mmol) in MeOH (50 mL), and the mixture was stirred
overnight at 50–538C. The resulting solution was cooled to
room temperature. Removal of volatiles under reduced
1
pressure gave 2 (8.02 g, quantitative) as a colorless oil. H
NMR d 1.44 (9H, s, (CH3)3C), 1.48–1.56 (4H, m, CH2–
CH2–CH2–CH2), 2.41 (2H, t, J¼6.81 Hz, CH2–CH2–CN),
2.48–2.56 (2H, m, N–CH2–CH2), 2.78 (2H, t, J¼6.81 Hz,
CH2–CH2–CN), 3.03–3.22 (2H, m, CO–NH–CH2), 3.60
(2H, s, CH2–Ar), 4.60 (1H, br m, NH), 7.25–7.38 (5H, m,
Ar); 13C NMR d 16.3 (CH2–CN), 24.4 (NH–CH2–CH2–
CH2), 27.6 (NH–CH2–CH2), 28.4 ((CH)3C), 40.2 (NH–
CH2), 49.2 (CH2–CH2–CN), 53.3 (NH–CH2–CH2–CH2–
CH2), 58.4 (CH2–Ar), 79.0 ((CH)3C), 118.9 (CN), 127.3
(ArCH), 128.3 (ArCH), 128.6 (ArCH), 138.7 (ArC), 156.0
(CO). HRMS: found (MþH)þ, 332.2338. C19H30N3O2
requires (MþH) 332.2338. Analysis calcd for
4.2.4. N 1-[2,3-Bis(benzyloxy)benzoyl]-N 4-benzylspermi-
dine (6). A solution of TFA (10 mL) in CH2Cl2 (20 mL) was
added dropwise to an ice-bath cooled solution of 5 in
CH2Cl2 (20 mL) and water (0.05 mL). The mixture was
stirred for 5 min, then concentrated under reduced pressure
at 3–58C. The yellow residue was dissolved in CH2Cl2
(250 mL) and washed with satd. Na2CO3 (2£200 mL),
water (2£100 mL), and brine (2£100 mL), dried over
Na2SO4, and filtered; solvents were removed under reduced
pressure. 1H NMR analysis of a 5-mg sample indicated 36%
conversion to 6. The remaining residue was retreated as

2012
R. J. Bergeron et al. / Tetrahedron 59 (2003) 2007–2014
above, stirring for 15 min after TFA addition. Flash
chromatography of the residue remaining after the second
workup, eluting with 3% concentrated NH4OH in MeOH,
removed under reduced pressure (3£) to facilitate removal
of excess HOAc. The residue was dissolved in MeOH
(25 mL), Sephadex LH-20 (1.35 g) was added, and the
mixture was allowed to stand overnight at 0 to 258C. The
solvents were removed under reduced pressure; the residue
was loaded onto a wet-packed Sephadex LH-20 (27 g,
2.5£15 cm) column, eluting with 1:8:13 water/EtOH/
toluene at 0.25 mL/min. The iron active fractions were
combined and concentrated. An aliquot of approximately a
third of the resulting solution was passed through an anion
exchange column (TFA, 14 mL bed); the iron active
fractions were combined and lyophilized to give crude 8
(0.023 g, 25%). Final purification of an analytical sample by
HPLC on a C4 column (solvent: 94.4% water, 5.5% MeCN,
0.1% TFA, isocratic elution, detection by uv absorption at
280 nm, Rt¼11.0–12.5 min), followed by lyophilization,
gave 8 as a white amorphous solid, calculated yield 17%.
1
gave 6 (2.10 g, 73%) as a colorless oil. H NMR d 1.26–
1.51 (8H, m, NH2–CH2–CH2–CH2 and CH2–CH2–NH–
CO), 2.26–2.39 (4H, m, NH2–CH2–CH2–CH2–CH2–N),
2.59 (2H, t, J¼6.59 Hz, N–CH2–CH2–CH2–NH), 3.26–
3.35 (2H, m, CO–NH–CH2), 3.45 (2H, s, N–CH2–Ar),
5.02 (2H, s, O–CH2–Ar), 5.16 (2H, s, O–CH2–Ar), 7.10–
7.50 (17H, m, Ar), 7.64–7.73 (1H, m, Ar), 7.92 (1H, br t,
J¼5.28 Hz, Ar–CO–NH). HRMS: found (MþH)þ,
552.3216. C35H42N3O3 requires (MþH) 552.3226. Analysis
calcd for C35H41N3O3·1H2O: C, 73.78; H, 7.61; N, 7.38.
Found: C, 73.68; H, 7.37; N, 7.24.
4.2.5. tert-Butyl 4-[[4-[[3-[[2,3-Bis(benzyloxy)benzoyl]-
amino]propyl](phenylmethyl)amino]butyl]amino]-2-[2-
[[4-[[3-[[2,3-bis(benzyloxy)benzoyl]amino]propyl](phe-
nylmethyl)amino]butyl]amino]-2-oxoethyl]-2-hydroxy-
4-oxo-butanoate (7). A solution of 6 (1.07 g, 1.94 mmol) in
CH3CN (20 mL) was added dropwise to a solution of 1,3-
1
Detailed H (500 MHz) and 13C (125 MHz) NMR analysis
is given in Table 1. HRMS: found (MþH)þ, 719.3606.
C34H51N6O11 requires (MþH) 719.3616.
bis(p-nitrophenyl)-2-(tert-butyl)
citrate28
(0.40 g,
4.3. Synthesis of petrobactin (13)
0.82 mmol) in CH3CN (40 mL) with ice-bath cooling. A
solution of DIEA (0.25 g, 1.97 mmol) in CH3CN (8 mL)
was added dropwise; the mixture was stirred for 2 h and
then stored overnight at 0 to 258C. The solvents were
removed under reduced pressure, and the residue was
dissolved in CHCl3 (300 mL), washed with satd. NaHCO3
(6£150 mL) and brine (150 mL), dried over Na2SO4, and
filtered; solvents were removed under reduced pressure.
Flash chromatography of the residue, eluting with 1:2:15
MeOH/acetone/CHCl3, gave 7 (0.77 g, 71%) as a yellow oil.
1H NMR d 1.43 (9H, s, (CH3)3C), 1.43–1.58 (12H, m,
CH2–CH2–CH2–CH2 and CH2–CH2–CH2), 2.28–2.36
(8H, m, CH2–N–CH2), 2.49 (2H, d, J¼14.9 Hz, CHH–
CO), 2.65 (2H, d, J¼14.9 Hz, CHH–CO), 3.03–3.18 (4H,
m, CH2–CO–NH–CH2), 3.25–3.34 (4H, m, Ar–CO–
NH–CH2), 3.43 (4H, s, N–CH2–Ar), 5.02 (4H, s, O–CH2–
Ar), 5.15 (4H, s, O–CH2–Ar), 5.55 (1H, br s, OH), 6.83
(2H, t, J¼5.4 Hz, CH2–CO–NH), 7.10–7.50 (34H, m, Ar),
7.63–7.71 (2H, m, Ar), 7.94 (2H, br t, J¼5.5 Hz, Ar–CO–
NH). Analysis calcd for C80H94N6O11: C, 73.03; H, 7.20; N,
6.39. Found: C, 72.91; H, 7.31; N, 6.38.
4.3.1. N 1-[3,4-Bis(benzyloxy)benzoyl]-N 4-benzyl-N 8-
(tert-butoxycarbonyl)spermidine (10). A solution of
oxalyl chloride (15.04 g, 120 mmol) in CH2Cl2 (50 mL)
was added dropwise to an ice-bath cooled slurry of 3,4-
bis(benzyloxy)benzoic acid30,35 (10.02 g, 30 mmol) and
DMF (0.4 mL) in toluene (50 mL); the mixture was stirred
for 1 h. The ice-bath was removed and stirring continued at
rt for 1 h. Volatiles were removed under high vacuum; 9 was
dissolved in CH2Cl2 (100 mL) and added dropwise to an
ice-bath cooled solution of 3 (6.70 g, 20 mmol) and DIEA
(28.54 g, 220 mmol) in CH2Cl2 (250 mL). The mixture was
allowed to warm to rt and was stirred overnight. The
solution was washed with water (2£250 mL) and brine
(2£250 mL), dried over Na2SO4, and filtered; solvents were
removed under reduced pressure. Flash chromatgraphy of
the residue, eluting with EtOAc, gave 10 (9.32 g, 72%) as a
pale yellow solid, mp 978C. 1H NMR d 1.42 (9H, s,
(CH3)3C), 1.58–1.73 (6H, m, N–CH2–CH2–CH2–CH2
and N–CH2–CH2–CH2–N), 2.46 (2H, t, J¼6.9 Hz,
NH–CH2–CH2–CH2–CH2–N), 2.53 (2H, t, J¼6.0 Hz,
N–CH2–CH2–CH2–NH), 3.02–3.09 (2H, m, OCO–NH–
CH2), 3.42–3.48 (2H, m, Ar–CO–NH–CH2), 3.52 (2H, s,
N–CH2–Ar), 4.58 (1H, br m, OCONH), 5.14 (2H, s,
O–CH2–Ar), 5.21 (2H, s, O–CH2–Ar), 6.83 (1H, d,
J¼8.4 Hz, Ar), 7.01 (1H, d, J¼8.4 Hz, Ar), 7.13–7.48
(16H, m, Ar). Analysis calcd for C40H49N3O5: C, 73.70; H,
7.58; N, 6.45. Found: C, 73.59; H, 7.67; N, 6.54.
4.2.6. 4-[[4-[[3-[(2,3-Dihydroxybenzoyl)amino]propyl]-
amino]butyl]amino]-2-[2-[[4-[[3-[(2,3-dihydroxyben-
zoyl)amino]propyl]amino]butyl]amino]-2-oxoethyl]-2-
hydroxy-4-oxo-butanoic acid, bis(trifluoroacetate) salt
(reported petrobactin, 8). TFA (15 mL) was added
dropwise to an ice-bath cooled solution of 7 (0.385 g,
0.293 mmol) in CH2Cl2 (15 mL) and water (0.15 mL), and
the mixture was stirred for 1 h. The ice-bath was removed
and stirring continued for 1 h. The mixture was concentrated
under reduced pressure, dissolved in CH2Cl2 (25 mL) and
water (0.15 mL) and reconcentrated (3£). The concentrate
was used directly in the next reaction and was dissolved in a
solution of glacial HOAc (15 mL) and water (0.75 mL),
PdCl2 (163 mg) was added, and the mixture was stirred
under H2 at ambient pressure for 7 h. The mixture was
filtered, solids were washed with a solution of glacial HOAc
(15 mL) and water (0.75 mL); the combined filtrate and
washings were concentrated under reduced pressure.
Toluene (15 mL) and water (0.25 mL) were added and
4.3.2. N 1-[3,4-Bis(benzyloxy)benzoyl]-N 4-benzylspermi-
dine (11). A solution of 10 (2.5 g, 3.8 mmol) in CH3OH
(41 mL) and concentrated HCl (16 mL) was stirred at rt for
12 h. The mixture was concentrated to dryness. Water
(100 mL) was added to the residue, the pH was adjusted to 9
with satd. Na2CO3, and the aqueous layer was extracted
with EtOAc (3£100 mL). The combined organic extracts
were washed with H2O (2£100 mL), and the solvent was
removed in vacuo. Silica gel flash chromatography of the
residue, eluting with 3% concentrated NH4OH in MeOH,
gave 11 (1.79 g, 84%) as a white solid, mp 688C. 1H NMR d
1.35–1.70 (8H, m, NH2–CH2–CH2–CH2 and CH2–CH2–

R. J. Bergeron et al. / Tetrahedron 59 (2003) 2007–2014
2013
NH–CO), 2.46 (2H, t, J¼7.8 Hz, NH2–CH2), 2.55 (2H, t,
J¼6.0 Hz, NH2–CH2–CH2–CH2–CH2–N), 2.62 (2H, t,
J¼6.6 Hz, N–CH2–CH2–CH2–NH), 3.44–3.49 (2H, m,
CO–NH–CH2), 3.54 (2H, s, N–CH2–Ar), 5.13 (2H, s, O–
CH2–Ar), 5.21 (2H, s, O–CH2–Ar), 6.84 (1H, d, J¼8.4 Hz,
Ar), 7.02 (1H, d, J¼8.4 Hz, Ar), 7.18–7.48 (16H, m, Ar).
Analysis calcd for C35H41N3O3·0.5H2O: C, 74.97; H, 7.55;
N, 7.49. Found: C, 74.90; H, 7.39; N, 7.41.
NH–CO–CH2 and N(CO)2–CH2–CH2–CH2), 1.79–1.87
(4H, m, N–CH2–CH2–CH2–N), 2.59 (1H, d, J¼15.3 Hz,
NC(O)–CHH–C(OH)), 2.70 (1H, d, J¼15.3 Hz, NC(O)–
CHH–C(OH)), 2.83 (1H, d, J¼15.4 Hz, C(OH)–C(O)–N–
C(O)–CHH), 2.87 (1H, d, J¼15.4 Hz, C(OH)–C(O)–N–
C(O)–CHH), 2.88–2.94 (8H, m, CH2–NH–CH2), 2.96–
3.07 (2H, m, CH2–CH2–CH2–CH2–NH–CO–CH2), 3.28
(4H, m, Ar–CO–NH–CH2), 3.39–3.43 (2H, m, N(CO)2–
CH2), 6.82 (2H, d, J¼8.2 Hz, C(C)–CH–C(OH)–C(OH)–
CH–CH), 7.20 (2H, dd, J¼2.3, 8.2 Hz, C(C)–CH–
C(OH)–C(OH)–CH–CH), 7.27 (2H, d, J¼2.3 Hz, C(C)–
CH–C(OH)–C(OH)–CH–CH). HRMS: found (MþH)þ,
701.3546. C34H49N6O10 requires (MþH) 701.3510.
4.3.3. tert-Butyl 4-[[4-[[3-[[3,4-Bis(benzyloxy)benzoyl]-
amino]propyl](phenylmethyl)amino]butyl]amino]-2-[2-
[[4-[[3-[[3,4-bis(benzyloxy)benzoyl]amino]propyl]-
(phenylmethyl)amino]butyl]amino]-2-oxoethyl]-2-
hydroxy-4-oxo-butanoate (12). A solution of 11 (1.59 g,
2.88 mmol) in CH2Cl2 (25 mL) was added dropwise to a
solution of 1,3-bis(p-nitrophenyl)-2-(tert-butyl) citrate28
(0.59 g, 1.2 mmol) in CH2Cl2 (50 mL) with ice-bath cool-
ing. A solution of NEt3 (0.156 g, 1.42 mmol) in CH2Cl2
(10 mL) was added dropwise; the mixture was stirred for 2 h
then stored overnight at 0 to 258C. Workup according to the
method described for 7 and flash chromatography of the
residue, eluting with 2:10 MeOH/CHCl3, afforded 12
Acknowledgements
This work was supported by the US National Institutes of
Health Grant No. R01-DK49108. We thank James R. Rocca
(Advanced Magnetic Resonance Imaging and Spectroscopy
Facility, McKnight Brain Institute, University of Florida)
for the HMQC spectra. We also appreciate the technical
assistance of J. R. Timothy Vinson and the aid of Dr Eileen
Eiler-McManis for the organization and editing of this
manuscript.
1
(1.15 g, 72%) as a yellow solid, mp 658C. H NMR d 1.42
(9H, s, (CH3)3C), 1.66–1.72 (12H, m, CH2–CH2–CH2–
CH2 and CH2–CH2–CH2), 2.43–2.63 (12H, m, CH2–N–
CH2, CHH–CO, CHH–CO), 3.13–3.15 (4H, m, CH2–
CO–NH–CH2), 3.42–3.43 (4H, m, Ar–CO–NH–CH2),
3.52 (4H, s, N–CH2–Ar), 5.13 (4H, s, O–CH2–Ar), 5.19
(4H, s, O–CH2–Ar), 6.62 (1H, br s, OH), 6.84 (2H, d,
J¼8.4 Hz, Ar), 7.08 (2H, d, J¼8.4 Hz, Ar), 7.16–7.48
(32H, m, Ar). Analysis calcd for C80H94N6O11: C, 73.03; H,
7.20; N, 6.39. Found: C, 72.65; H, 7.23; N, 6.45.
References
1. Sahlin, M.; Petersson, L.; Graslund, A.; Ehrenberg, A.;
Sjoberg, B.-M.; Thelander, L. Biochemistry 1987, 26,
5541–5548.
4.3.4. 4-[[4-[[3-[(3,4-Dihydroxybenzoyl)amino]propyl]-
amino]butyl]amino]-2-[2-[[4-[[3-[(3,4-dihydroxyben-
zoyl)amino]propyl]amino]butyl]amino]-2-oxoethyl]-2-
hydroxy-4-oxo-butanoic acid, bis(trifluoroacetate) salt
(petrobactin, 13). TFA (15 mL) was added dropwise to an
ice-bath cooled solution of 12 (0.385 g, 0.293 mmol) in
CH2Cl2 (15 mL) and water (0.15 mL); the mixture was
stirred for 1 h. The ice-bath was removed and stirring
continued for 1 h. The mixture was concentrated under
reduced pressure, dissolved in CH2Cl2 (25 mL) and water
(0.15 mL) and reconcentrated (3£); this concentrate was
used directly in the next reaction. The residue was dissolved
in a solution of glacial HOAc (15 mL) and water (0.75 mL),
PdCl2 (163 mg) was added, and the mixture was stirred
under H2 at ambient pressure overnight. Filtration and
workup were as described for compound 8. Purification of
the crude product was directly accomplished by HPLC on a
C4 column in the manner described for compound 8
(Rt¼11.0–12.5 min). Lyophilization of the pooled products
afforded 13 as a solid in 61% calculated yield. Detailed 1H
(500 MHz) and 13C (125 MHz) NMR analysis is given in
Table 2. HRMS: found (MþH)þ, 719.3606. C34H51N6O11
requires (MþH) 719.3615.
2. Ortiz de Montellano, P. R. Cytochrome P450 – Structure,
Metabolism, and Biochemistry; Plenum: New York, 1986.
3. Raymond, K. N.; Carrano, C. J. Acc. Chem. Res. 1979, 12,
183–190.
4. Tufano, T. P.; Raymond, K. N. J. Am. Chem. Soc. 1981, 103,
6617–6624.
5. Neilands, J. B. Bacteriol. Rev. 1957, 21, 101–111.
6. Neilands, J. B. J. Biol. Chem. 1995, 270, 26723–26726.
7. Byers, B. R.; Arceneaux, J. E. Met. Ions Biol. Syst. 1998, 35,
37–66.
8. Leong, S. A.; Winkelmann, G. Met. Ions Biol. Syst. 1998, 35,
147–186.
9. Hershko, C.; Weatherall, D. J. Crit. Rev. Clin. Lab. Sci. 1988,
26, 303–345.
10. Lieu, P. T.; Heiskala, M.; Peterson, P. A.; Yang, Y. Mol. Asp.
Med. 2001, 22, 1–87.
11. Bickel, H.; Hall, G. E.; Keller-Schierlein, W.; Prelog, V.;
Vischer, E.; Wettstein, A. Helv. Chim. Acta 1960, 43,
2129–2138.
12. Aksoy, M.; Seyithanoglu, B. Y.; Bozbora, A. In Hypertension
and Iron Chelation in Thalassemia; Aksoy, A., Birdwood,
G. F. B., Eds.; Huber: Berne, Switzerland, 1984; p 11.
13. Olivieri, N. F.; Brittenham, G. M. Blood 1997, 89, 739–761.
14. Giardina, P. J.; Grady, R. W. Semin. Hematol. 2001, 38,
360–366.
The imide 14 (Fig. 2) was also isolated by HPLC during the
petrobactin preparative runs (retention time 17–19 min).
Lyophilization gave 14 as a white amorphous solid,
calculated yield 15%: 1H NMR (500 MHz, DMSO/D2O
4:1) d 1.39–1.47 (2H, m, CH2–CH2–CH2–CH2–NH–
CO–CH2), 1.52–1.62 (6H, m, CH2–CH2–CH2–CH2–
15. Lee, P.; Mohammed, N.; Marshall, L.; Abeysinghe, R. D.;
Hider, R. C.; Porter, J. B.; Singh, S. Drug Metab. Dispos.
1993, 21, 640–644.
16. Pippard, M. J.; Callender, S. T.; Finch, C. A. Blood 1982, 60,
288–294.

2014
R. J. Bergeron et al. / Tetrahedron 59 (2003) 2007–2014
17. Pippard, M. J. Baillieres Clin. Haematol. 1989, 2, 323–343.
18. Peter, H. H.; Bergeron, R. J.; Streiff, R. R.; Wiegand, J. In The
Development of Iron Chelators for Clinical Use; Bergeron,
R. J., Brittenham, G. M., Eds.; CRC: Boca Raton, 1994; pp
373–394.
27. Bergeron, R. J.; Garlich, J. R.; Stolowich, N. J. J. Org. Chem.
1984, 49, 2997–3001.
28. Ghosh, A.; Miller, M. J. J. Org. Chem. 1993, 58, 7652–7659.
29. Muller, D.; Zeltser, I.; Bitan, G.; Gilon, C. J. Org. Chem. 1997,
62, 411–416.
19. Bergeron, R. J.; Streiff, R. R.; King, W.; Daniels, Jr. R. D.;
Wiegand, J. Blood 1993, 82, 2552–2557.
30. Blagbrough, I. S.; Maya, E. Tetrahedron Lett. 1995, 36,
9393–9396.
20. Gauthier, M. J.; Lafay, B.; Christen, R.; Fernandez, L.;
Acquaviva, M.; Bonin, P.; Bertrand, J.-C. Int. J. Syst.
Bacteriol. 1992, 42, 568–576.
31. Rastetter, W. H.; Erickson, T. J.; Venuti, M. C. J. Org. Chem.
1981, 46, 3579–3590.
32. Bergeron, R. J.; Xin, M. G.; Smith, R. E.; Wollenweber, M.;
McManis, J. S.; Ludin, C.; Abboud, K. Tetrahedron 1997, 53,
427–434.
21. Barbeau, K.; Rue, E. L.; Bruland, K. W.; Butler, A. Nature
2001, 413, 409–413.
22. Peterson, T.; Neilands, J. B. Tetrahedron Lett. 1979,
4805–4808.
33. Wang, Q. X.; Phanstiel, O., IV. J. Org. Chem. 1998, 63,
1491–1495.
23. Barbeau, K.; Zhang, G.; Live, D. H.; Butler, A. J. Am. Chem.
Soc. 2002, 124, 378–379.
34. Silverstein, R. M.; Bassler, G. C.; Morrill, T. Spectrometric
Identification of Organic Compounds; 5th ed. Wiley: New
York, 1991.
24. Balzani, V.; Carassiti, V. Photochemistry of Coordination
Compounds; Academic: New York, 1970; Chapter 10.
25. Faust, B. C.; Zepp, R. G. Environ. Sci. Technol. 1993, 27,
2517–2522.
35. Deng, B.-L.; Lepoivre, J. A.; Lemiere, G.; Dommisse, R.;
Claeys, M.; Boers, F.; De Groot, A. Liebigs Ann. Chem. 1997,
2169–2175.
26. Aplincourt, M.; Bee, A.; Gerard, C.; Hugel, R.; Njomgang, R.;
Prudhomme, J.-C. J. Chem. Res., Synop. 1987, 398–399.