An improved method for culturing Streptomyces sahachiroi: Biosynthetic origin of the enol fragment of azinomycin B

Article abstract of DOI:10.1016/j.bioorg.2007.08.002

Azinomycin B is an environmental DNA crosslinking agent produced by the soil microorganism Streptomyces sahachiroi. While the agent displays potent cytotoxic activities against leukemic cell lines and animal mouse models, the lack of a consistent supply of the natural product has hampered detailed biological investigations on the compound, including its mode of action and biosynthesis. We report here a significant methodological improvement in the culturing of the bacterium, which allows reliable and steady production of the natural product in good yields. The key experimental step involves the culturing of the strain on dehydrated plates, followed by the generation of a two-stage starter culture and subsequent fermentation of the strain under nutrient-starved conditions. We illustrate use of this culture system by investigating the formation of the enol fragment of the molecule in isotopic labeling experiments with threonine and several advanced precursors (β-ketoamino acid 3, β-hydroxyamino aldehyde 4, and β-ketoaminoaldehyde 5). The results unequivocally show that threonine is the most advanced precursor accepted by the NRPS (non-ribosomal peptidyl synthetase) machinery for final processing and construction of the enol moiety of the natural product.

Full text of DOI:10.1016/j.bioorg.2007.08.002

Available online at www.sciencedirect.com
Bioorganic Chemistry 36 (2008) 4–15
www.elsevier.com/locate/bioorg
An improved method for culturing Streptomyces
sahachiroi: Biosynthetic origin of the enol fragment of azinomycin B
1
1
,2
*
Gilbert T. Kelly , Vasudha Sharma , Coran M.H. Watanabe
Department of Chemistry, Texas A&M University, P.O. Box 30012, College Station, TX 77843, USA
Received 20 July 2007
Available online 27 September 2007
Abstract
Azinomycin B is an environmental DNA crosslinking agent produced by the soil microorganism Streptomyces sahachiroi. While the
agent displays potent cytotoxic activities against leukemic cell lines and animal mouse models, the lack of a consistent supply of the nat-
ural product has hampered detailed biological investigations on the compound, including its mode of action and biosynthesis. We report
here a significant methodological improvement in the culturing of the bacterium, which allows reliable and steady production of the nat-
ural product in good yields. The key experimental step involves the culturing of the strain on dehydrated plates, followed by the gener-
ation of a two-stage starter culture and subsequent fermentation of the strain under nutrient-starved conditions. We illustrate use of this
culture system by investigating the formation of the enol fragment of the molecule in isotopic labeling experiments with threonine and
several advanced precursors (b-ketoamino acid 3, b-hydroxyamino aldehyde 4, and b-ketoaminoaldehyde 5). The results unequivocally
show that threonine is the most advanced precursor accepted by the NRPS (non-ribosomal peptidyl synthetase) machinery for final pro-
cessing and construction of the enol moiety of the natural product.
ꢀ 2007 Elsevier Inc. All rights reserved.
Keywords: Azinomycin; Biosynthesis; Streptomyces; PKS/NRPS
1. Introduction
16 lg/kg/d and an ILS of 161% at 32 lg/kg/d for Ehrlich
carcinoma comparable to the clinical drug, mitomycin C
(204% at 1 mg/kg/d, the intraperitoneal dosage of mitomy-
cin C was ꢀ62-fold that of azinomycin B) [3]. In contrast,
azinomycin A gave a comparably lower ILS (76% at 16 lg/
kg/d), and a much narrower therapeutic index suggesting
some degree of selectivity between these two agents.
In vitro experiments reveal the inherent ability of azino-
mycin B to bind within the major groove of DNA. The
electrophilic C-10 and C-21 carbons contained within the
aziridino[1,2a]pyrrolidine (1-azabicyclo[3.1.0]hexane) and
epoxide fragments impart the natural product to form
interstrand crosslinks with the N7 positions of suitably dis-
posed purine bases of DNA [4]. Recent studies with DNA
microarrays and fluorescence imaging with the natural
product provide the first demonstration of DNA damage
caused by azinomycin B in whole cells, correlating
in vitro DNA cross-linking observed with the metabolite
with an in vivo cellular response [5].
The azinomycins (A and B, Scheme 1) comprise a family
of aziridine-containing natural products produced by
Streptomyces sp. that possess potent anti-tumor activity
[1]. Both compounds exhibit cytotoxicity in the nanomolar
range against the leukemic cell line L5178 [2]. Likewise,
in vivo studies against P388 leukemic mice indicate that azi-
nomycin B exhibits an ILS (increased lifespan) of 193% at
*
Corresponding author. Fax: +1 979 458 8095.
E-mail address: watanabe@mail.chem.tamu.edu (C.M.H. Watanabe).
These authors contributed equally to the experimental portion of this
1
manuscript.
2
This manuscript is dedicated in memory of my academic grandfather,
Distinguished Professor Ian Scott, whose pioneering efforts in natural
product biosynthesis and passion for science will continue to inspire many
of us.
0045-2068/$ - see front matter ꢀ 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.bioorg.2007.08.002

G.T. Kelly et al. / Bioorganic Chemistry 36 (2008) 4–15
5
OH
O
O
O
O
N
O
H
N
O
H
N
O
O
O
O
O
N
H
O
N
H
O
O
O
O
O
O
N
HO
HO
Azinomycin B (1b)
Azinomycin A (1a)
Scheme 1. Structures of azinomycin A and B.
1
The unusual architecture of the azinomycins coupled
with its potent anti-tumor activity has captured the atten-
tion of both the biosynthetic and synthetic communities.
The total synthesis of azinomycin A was achieved in 2001
by Coleman and co-workers and a number of synthetic
analogues have also been reported [6]. Notwithstanding
these gains, even after 50 years beyond its initial isolation
[7] and more than two decades of structure correction [1],
there is no total synthesis yet reported for azinomycin B
which is a major barrier to in-depth studies involving bio-
logical activity of the natural product. In addition, biosyn-
thetic investigations on the compound have lagged
considerably. While some gains have been made to estab-
lish the polyketide origin of the naphthoate moiety [8]
and a cell-free system developed to support synthesis of azi-
nomycin B in vitro [9], progress in this area has been largely
impeded by difficulties with the culture method and secur-
ing a consistent source of the natural product. Following
literature protocols [1,8], we found production of the natu-
ral product by Streptomyces sahachiroi to be highly erratic
(production would be observed once then not seen for 3-
weeks or more). As a great majority of biosynthetic studies
hinge upon having reliable production of the natural prod-
uct including isotopic labeling studies and gene disruption
experiments, this necessitated the development of a new
culture method. We detail here, experiments performed
to achieve these optimized growth conditions and utilize
these refined culture conditions to investigate the biosyn-
thetic route to the enol fragment of azinomycin B.
ian Inova 500 or Varian Inova 300. H NMR chemical
shifts are reported as d values in ppm relative to CDCl3
(7.26 ppm) and coupling constants (J) are reported in Hertz
(Hz). Infrared spectra were recorded on a Bruker Tensor 27
spectrometer. Unless otherwise indicated, deuterochloro-
form (CDCl3) served as an internal standard (77.0 ppm)
for all 13C spectra. Mass spectra (ESI) were obtained at
the Laboratory for Biological Mass Spectrometry at the
Department of Chemistry, Texas A&M University, with
API QStar Pulsar, MDS Sciex (Toronto, ON, Canada)
Quadrupole-TOF hybrid spectrometer. Gas chromatogra-
phy/low resolution mass spectra were recorded on a Trace
DSQ GCMS spectrometer, ThermoElectron Corporation
(Austin, TX, USA). APCI was recorded on a Thermofinn-
igan LC-Q DECA mass spectrometer. Fermentations were
run on Fermentation Design Inc. Model # MS21 (Allen-
town, PA, USA). The total capacity of the fermentation
system is 15 L.
2.2. Organism
Streptomyces sahachiroi (NRRL 2485) was obtained
from the American Type Culture Collection (ATCC).
2.3. Culture conditions
Spore stocks: Streptomyces sahachiroi spores from dehy-
drated glucose, yeast extract, and maltose extract (GYM)
plates (per liter of medium: glucose monohydrate, 4 g yeast
extract, 4 g malt extract, 10 g CaCO3, 2 g and tap water
adjusted to pH 6.8 with 1 M NaOH prior to sterilization)
were streaked onto large MS (Mannitol Soya flour, per liter
medium: Mannitol, 20 g Soya flour, 20 g and deionized
water) plates and allowed to incubate at 30 ꢁC for 15 days.
At this time the grey spores were removed with sterile water
and agitation. The spores were then filtered through sterile
cotton, washed three times with sterile water, centrifuged at
3000 rpm, re-suspended in a minimal amount of 10% glyc-
erol solution, flash frozen, and stored at ꢁ80 ꢁC.
2. Materials and methods
2.1. Instrumentation and general methods
Reactions were carried out in flame-dried glassware
under nitrogen or an argon atmosphere, unless otherwise
noted. Commercial solvents and reagents were used as
received from Sigma–Aldrich. All isotopically labeled
material was purchased from Cambridge Isotope Labora-
tories, Inc. Anhydrous solvents were dried over neutral alu-
mina (MBraun system). All reactions were magnetically
stirred and monitored by thin layer chromatography
(TLC), utilizing glass-backed silica gel plates from Anal-
tech (#47011). Yields refer to chromatographically and
spectroscopically pure compounds unless otherwise stated.
(i) Solid plates. Solid media formulations (See Support-
ing Information A) were evaluated by streaking a
loop full of S. sahachiroi spore stock onto prepared
plates. The plates were grown at 30 ꢁC in a Fisher Sci-
entific Isotemp incubator for 5–7 days. Once grown,
1/4 of the plate was inoculated into Erlenmeyer flasks
containing 100 mL PS5 medium and shaken at
250 rpm at 30 ꢁC. After 72 h of growth, the cultures
˚
Flash column chromatography was performed using 60 A
Silica Gel (Silacycle, 230–400 mesh) as a stationary pha-
se.1H and 13C NMR spectra were recorded on either a Var-

6
G.T. Kelly et al. / Bioorganic Chemistry 36 (2008) 4–15
were extracted with CH2Cl2, concentrated and ana-
3000 rpm and the supernatant discarded. The resulting res-
idue was dissolved in dichloromethane (600 lL) to which
hexanes (2 mL) was added. The heterogeneous mixture
was centrifuged at 3000 rpm and the supernatant retained.
To the solution was added hexanes (4 mL) and the suspen-
sion centrifuged at 3000 rpm to give azinomycin B as a
solid.
If full purification is not achieved, azinomycin B can be
further purified by flash column chromatography (95:5
CH2Cl2:methanol). By TLC azinomycin exhibits an Rf of
0.23. A short column should be utilized to minimize overall
contact with the silica gel and degradation by hydrolysis.
The process can be repeated if necessary. The compound
can be safely stored at ꢁ80 ꢁC under anhydrous diethyl
ether. The azinomycin B isolated by us matched the
NMR spectrum provided by Yokoi et al. [1b].
lyzed by TLC and MS. First and second stage cul-
tures were prepared by streaking an aliquot of the
S. sahachiroi spore stock suspension onto the surface
of the GYM plates and storing them at room temper-
ature to dryness.
(ii) First stage culture. Streptomyces sahachiroi (inocu-
lated from dehydrated plates) was grown on GYM
plates for 5–7 days. A 1 cm2 piece of the GYM plate
was used to inoculate 100 mL of PS5 medium in a
250 mL Erlenmeyer flask. The culture was incubated
at 30 ꢁC for 24 h at 250 rpm.
(iii) Second stage culture. The second stage culture was
prepared by inoculating 2 L Erlenmeyer baffled flasks
(Fernbach containing 600 mL of PS5 medium) with
25 mL of the first stage culture. The culture was incu-
bated at 30 ꢁC for 24 h at 250 rpm.
(iv) Fermentation.The fermenter containing 10 L of the
reduced PS5 (Pharmamedia/Starch plus additives,
per liter of medium: Pharmamedia (yellow cotton
seed flour), 5 g soluble starch, 5 g glucose, 2 g casein
hydrolysate, 2 g NH4SO4, 0.5 g lysine, 0.5 g orni-
thine, 0.5 g glycine, 0.5 g and deionized water
adjusted to pH 7.0 prior to sterilization) medium
(reduced by 75%) was autoclaved at 121 ꢁC for
20 min. Following inoculation (with two 600 mL sec-
ond stage cultures) the fermenter was agitated at
ꢀ300 rpm and aerated with sterile filtered air (8 L/
min) for 72 h.
2.5.1. Characterization of azinomycin B
Pale-white amorphous powder (1:9 CH2Cl2:hexane) IR
(neat) mmax 3338.4(br), 2957.1, 2925.3, 2872.8, 1725.92(br),
1619.3, 1601.7, 1511.2, 1417.6 cmꢁ11 H NMR (300 MHz,
CDCl3) d 12.40 (1H, br), 12.32 (1H, s), 8.54 (1H, dd,
J = 3.6, 7.0 Hz,), 8.20 (1H, br), 7.94 (1H, d, J = 2.9 Hz),
7.46 (1H, d, J = 2.9 Hz), 7.32 (1H, s), 7.32 (1H, s), 7.32
(1H, s), 5.50 (1H, d, J = 4.0 Hz), 5.12 (1H, s), 4.64 (1H, dd,
J = 4.0, 4.8 Hz), 3.96 (3H, s), 3.96 (1H,br), 3.36 (1H, m),
2.98 (1H, d, J = 4.3 Hz), 2.80 (1H, d, J = 4.3 Hz), 2.70
(1H, s), 2.66 (3H, s), 2.30 (1H, s), 2.24 (1H, s), 2.18 (1H, s),
1.52 (1H, s) 13C NMR (75 MHz, CDCl3) d 191.5, 173.0,
165.7, 164.0, 162.0, 156.0, 153.0, 150.8, 134.5, 133.3, 128.1,
127.9, 127.0, 125.4, 123.9, 122.3, 119.3, 118.6, 108.5, 84.4,
77.4*, 77.1*, 56.2, 55.7, 53.9, 46.4, 36.7, 24.5, 21.0, 20.3,
17.2. APCI-MS (LRMS) 624.2. Found: 624.2 (*obscured
by CDCl3 solvent peak).
2.4. General feeding conditions
The labeled material was weighed in equal portions and
solubilized in autoclaved distilled water. The first aliquot
was administered after 24 h, followed by addition of the
second provided 24 h later. The culture was harvested
72 h post-induction (inoculation of the second stage culture
into the fermenter). Note. Compound 4 and 5 were admin-
istered in D2O immediately after synthesis. Table 3 (sup-
porting information) provides details on the amounts of
each compound fed.
2.6. Synthesis of isotopically labeled compounds (3), (4)
and (5)
2.6.1. Threonine methyl ester hydrochloride (6)
Thionyl chloride (398.5 mg, 3.4 mmol) was added drop-
wise to cold anhydrous methanol (15 mL) at 0 ꢁC. The
solution was allowed to stir at 0 ꢁC for 10 min. Threonine,
2 (0.4 g, 3.35 mmol) was added and the reaction mixture
allowed to reflux for 1 h. The reaction was allowed to cool
and the solvent evaporated in vacuo. A 2 N solution of
anhydrous HCl in methanol was generated and the reac-
tion was again refluxed for 1 h. The solution was allowed
to cool and the solvent evaporated in vacuo on a rotary
evaporator. A foamy solid was obtained in quantitative
yield. IR (NaCl, thin film) mmax 3375.3, 3236.2,
2960.6(br), 1744.1, 1599.0, 1516.1, 1240.9 cmꢁ11 H NMR
(CD3OD, 500 MHz) d 1.35 (d, 3H, J = 6.5 Hz), 3.87 (s,
3H), 3.97 (d, 1H, J = 4.5 Hz), 4.29 (dq, 1H, J1 = 6.5 Hz,
J2 = 4.0 Hz), 13C NMR and DEPT(CD3OD, 500 MHz) d
20.5, 53.7, 59.7, 66.3,169.6 HRMS (ESI+): m/z calcd for
C5H12NO3Cl (M+H), 134.0818. Found: 134.0812. ESI-
(LRMS): m/z 34.96, 36.99.
2.5. Isolation and purification of azinomycin B
Following fermentation, the cultures were centrifuged at
7000 rpm at 4 ꢁC. The cell pellets were discarded and the
medium extracted with an equal volume of methylene chlo-
ride (1X). The organic layer was collected, dried over anhy-
drous magnesium sulfate, and concentrated in vacuo. The
resulting crude extract was stored under diethyl ether at
ꢁ80 ꢁC. The solid was dissolved in a minimal amount of
dichloromethane and precipitated with the addition of hex-
ane to give a ratio of 1:29 CH2Cl2/hexane. The resulting
suspension was centrifuged at 1500 rpm and the superna-
tant discarded. Diethyl ether (2 mL) was added to the pel-
let, which was subsequently agitated, centrifuged at

G.T. Kelly et al. / Bioorganic Chemistry 36 (2008) 4–15
7
2.6.2. Boc-protected threonine methyl ester (7)
,CHCl3) IR (NaCl, thin film)mmax 3384.2, 2931.2, 2851.3,
1758.9, 1720.4, 1708.6, 1483.6, 1365.1, 1157.9 cmꢁ1 1H
NMR (CDCl3, 300 MHz) d 1.42 (s, 9H), 2.35 (s, 3H), 3.78
(s,3H), 5.02 (d, 1H, J = 7.2 Hz), 5.74 (br, 1H). 13C NMR
(CDCl3, 300 MHz) d 7.8, 28.1, 53.2, 64.2, 80.8, 154.9,
166.9, 198.8 HRMS (ESI+): m/z calcd for C10H17NO5
(M+H), 232.1185. Found: 232.1178.
Method A. A flame dried 25 mL three-necked round-
bottomed flsk, equipped with a magnetic stirring bar, ther-
mometer, reflux condenser (protected from moisture by a
calcium chloride-filled drying tube), and a pressure-equaliz-
ing dropping funnel connected to a N2-line was charged
with a solution of di-tert-butyl dicarbonate (0.34 g,
1.56 mmol) in THF (2.5 mL). A suspension of methyl thr-
eoninate hydrochloride 6 (0.27 g, 1.59 mmol) in THF
(5 mL) and triethylamine (0.34 g, 3.4 mmol) was main-
tained at 0 ꢁC and allowed to stir for 5 min. The solution
of di-tert-butyl dicarbonate was added dropwise over a per-
iod of 1 h at 0 ꢁC. After 10 min of additional stirring, the
ice-water bath was removed and the suspension was stirred
overnight (14 h) at room temperature, then warmed at
50 ꢁC for a further 3 h. The solvent was removed under
reduced pressure and the residue was partitioned between
diethyl ether (20 mL) and saturated aqueous bicarbonate
solution (25 mL). The aqueous phase was extracted with
ether 3· 15 mL. The combined organic phases were dried
with anhydrous Na2SO4 and concentrated under reduced
pressure to give 0.33 g (88% yield) of N-Boc-L-threonine
methyl ester, 7 as a thick colorless oil that was used without
further purification.
2.6.4. 2-Amino 3-oxobutanoic acid TFA salt (3)
A solution of tert-butyl 1-(methoxycarbonyl)-2-oxopro-
pylcarbamate, 8 (0.13 g, 0.56 mmol) in 5 mL TFA:H2O (v/
v 1:1) was refluxed for 8 h and allowed to stir for an addi-
tional 5 h. The solvent was evaporated in vacuo to give the
desired trifluoroacetate salt as a light yellow sticky solid, 3.
[a]D ꢁ3.24 (c 0.4, MeOH) IR (NaCl, thin film)mmax 3422.7,
3014.1, 2857.2, 1732.2, 1693.4, 1675.9, 1199.0 cmꢁ1 1H
NMR (D2O, 500 MHz) d 2.06(s, 3H), 3.89 (s, 1H) 13C
NMR (D2O + 1 drop CD3OD, 500 MHz) d 26.6, 47.7,
116.4 (q, JCꢁF = 1160 Hz), 162.8(q, JCꢁF = 141 Hz), 171.6,
204.03 HRMS (ESI+): m/z calcd for 13C4H7NO3((M+H)-
H2O), 104.0. Found: 103.99.
2.6.5. (4S,5R)-3-tert-butyl 4-methyl 2,2,5-
trimethyloxazolidine-3,4-dicarboxylate (9)
Method B. Threonine methyl ester hydrochloride salt, 6
(3.0 g, 17.7 mmol) and NaHCO3 (4.6 g, 53.2 mmol) were
dissolved in a 1:1 (v/v) mixture of water and methanol
(36 mL). (Boc)2O (5.8 g, 26.6 mmol) was added drop-wise
and the solution was stirred at room temperature for
20 h. The reaction mixture was concentrated under vac-
uum. The reaction mixture was then acidified with aqueous
citric acid (1 M) to pH 4.5. The reaction mixture was
extracted with ethyl acetate (30· 4 mL). The organic layers
were combined, dried with MgSO4, filtered and concen-
trated in vacuo to obtain the product 7 in quantitative yield
and was used without further purification.
IR (NaCl, thin film) mmax 3432.1, 2982.4, 2853.2, 1753.4,
1719.9, 1510.4, 1365.2, 1172.4 cmꢁ1 1H NMR (CDCl3,
500 MHz) and gCOSY d 1.21 (d, 3H, J = 7 Hz), 1.45 (s,
9H), 2.23 (br, 1H), 3.74 (s, 3H), 4.22 (d, 1H, J = 9 Hz),
4.26 (d, 1H, J = 6 Hz), 5.41 (d, 1H, J = 7 Hz). 13C NMR
(CDCl3, 500 MHz) and gHMQC d 19.8, 28.2, 52.5, 58.6,
68.1, 80.1, 156.1, 172.0 HRMS (ESI+): m/z calcd for
C10H19NO5 (M+Li), 240.1423. Found: 240.1415.
2-Methoxypropene (0.31 g, 4.29 mmol) and camphor sul-
phonic acid (6.7 mg, 0.02 mmol) was added to a solution of
N-Boc-L-threonine methyl ester, 7 (0.1 g, 0.43 mmol) in ace-
tone (2 mL). The resulting orange solution was stirred at
room temperature for 3.5 h (TLC analysis indicated comple-
tion of reaction). The reaction mixture was quenched with
10 lL of triethylamine and the solvent removed under
reduced pressure. The residual brown syrup was partitioned
between diethyl ether (20 mL) and saturated aqueous
sodium bicarbonate solution (30 mL). The aqueous layer
was extracted with diethyl ether (2· 30 mL) and the com-
bined organic phases were dried with anhydrous sodium sul-
fate and concentrated under reduced pressure to give 105 mg
(90%) of oxazolidine methyl ester, 9 as a yellow oil in both
rotameric forms (3:1). IR (NaCl, thin film) mmax 2984.5,
2928.3, 2857.2, 1755.9, 1720.4, 1376.9, 1362.2, 1258.6 cmꢁ1
1H NMR and gCOSY (CDCl3, 300 MHz) d 1.31, 1.32 (s,
9H), 1.41 (s, 3H), 1.49–1.57 (m, 6H), 3.69 (s, 3H), 3.83,
3.91 (d, 2H, J = 7.5 Hz) 4.05–4.09 (m, 1H) 13C NMR
(CDCl3, 300 MHz) and DEPT d 18.7, 18.5, 23.9,24.7, 26.4,
27.7, 28.1, 28.2, 52.1, 52.3, 65.9, 66.1, 73.4,73.7, 80.2, 80.7,
94.4, 95.0, 151.2, 152.3, 171.0, 171.5 HRMS (ESI+): m/z
calcd for C13H23NO5, (M+H), 274.1654. Found: 274.1645.
2.6.3. tert-Butyl 1-(methoxycarbonyl)-2-
oxopropylcarbamate (Boc-protected b-keto-threonine
methyl ester) (8)
A solution of Boc-protected threonine methyl ester, 7
(0.1 g, 0.43 mmol) and Dess–Martin periodinane (0.22 g,
0.52 mmol) in dichloromethane (2 mL), was stirred for 1 h
at room temperature. The reaction was then quenched with
15 mL of 1:1 v/v solution of NaHCO3 and 10% Na2S2O3
and extracted with ethyl acetate (3· 30 mL). The organic
phase was collected, dried over Na2SO4 and concentrated
in vacuo to afford the product (8) as a low melting white solid
in 91% yield (92 mg). Mp = 61 ꢁC [a]D = -3.8(c = 2.5
2.6.6. (4R,5R)-tert-butyl 4-(hydroxymethyl)-2,2,5-
trimethyloxazolidine-3-carboxylate (10)
A 25-mL, two-necked, round-bottomed flask was
equipped with a magnetic stirring bar, reflux condenser bear-
ing a drying tube and a pressure-equalizing dropping funnel
fitted with a rubber septum. The flask was charged with
10 mL of tetrahydrofuran and 20.9 mg (0.55 mmol) of lith-
ium aluminum hydride. While the suspension in the flask
was stirred, a solution of the oxazolidine ester, 9 (100 mg,

8
G.T. Kelly et al. / Bioorganic Chemistry 36 (2008) 4–15
0.37 mmol) in tetrahydrofuran (1 mL) was added dropwise
over 20 min. The dropping funnel was washed with two
1-mL portions of tetrahydrofuran and the suspension stirred
for an additional 20 min, when TLC analysis showed the
complete formation of the alcohol. The reaction mixture
was cooled with an ice-water bath while 1 mL of a 10% aque-
ous KOH solution was added drop-wise over 10 min. Cau-
tion! The reaction is exothermic. The reaction was stirred
for another hour at room temperature, and filtered through
a Celite pad (1 · 2.5 cm) that was subsequently rinsed with
diethyl ether (3· 10 mL). The combined organic filtrates
were washed with 25 mL of aqueous phosphate buffer (pH
7.0), and the aqueous layer extracted with diethyl ether (3·
30 mL). The combined organic phases were dried with anhy-
drous sodium sulfate, filtered and concentrated in vacuo to
give 85.1 mg (ꢀ95%) of the desired product, 10 as a pale yel-
low oil which was used without further purification. IR
(NaCl, thin film) mmax 3437.5, 2978.6, 2928.3, 2875.0,
1696.7, 1670.1, 1456.9, 1406.6, 1255.6 cmꢁ1 1H NMR and
gCOSY (CDCl3 + 1 drop D2O), 300 MHz) d 1.32 (d, 3H,
J = 6 Hz), 1.44 (s, 3H), 1.47 (s, 3H,), 1.55 (s, 3H), 3.49 (m,
1H), 3.62 (d, 2H, J = 3 Hz) 3.71 (m, 1H) 13C NMR (CDCl3,
300 MHz) and HMQC d 18.1, 25.9, 27.8, 28.3, 64.7, 67.2,
71.9, 81.3, 94.1, 154.2 HRMS (ESI+): m/z calcd for
C12H23NO4, (M+H), 246.1705. Found: 246.1699.
grease-like material separated instantaneously. The slurry
was stirred for 20 min to yield the product as a trifluoroace-
tate salt, 4 in the D2O layer. Note: all attempts to concen-
trate the product led to decomposition of the material. This
adduct was used for feeding studies immediately. Stability
studies indicated the presence of ꢀ70% of the compound
after 48 h in D2O at room temperature. IR (NaCl, thin film)
mmax 3393.1, 1675.9 cmꢁ1 1H NMR and gCOSY(D2O,
500 MHz) d 1.01 (d, 3H, J = 6.5 Hz), 2.76–2.80 (m,1H),
3.76–3.84 (m, 1H), 4.08 (d,1H, J = 5 Hz). 13C NMR
(D2O, 500 MHz) d 19.2, 61.1, 63.9, 86.6, 116.0 (q,
JCꢁF = 1153 Hz), 162.2 (q, JCꢁF = 139.5 Hz) HRMS
(ESI+) m/z calcd for C4H11NO3(M+H), 121.0739. Found:
122.0819 C4H9NO2(M+H) 104.0712. Found: 104.0715.
2.6.9. Methyl (2S,3R)-2-(tert-Butoxycarbonylamino)-3-O-
(tert-butyldimethylsilyl) butanoate,(12)
A mixture of Boc-protected methyl ester 7 (1.0 g,
4.29 mmol), triethylamine (0.99 mL, 6.86 mmol), 4-N,
N-(dimethylamino)pyridine (52.3 mg, 0.42 mmol) in
anhydrous DMF (16 mL) was cooled at 0 ꢁC. tert-Butyldim-
ethylsilyl chloride (0.84 g, 5.57 mmol) was added and the
mixture was stirred for 1.5 h. The ice-bath was removed
and the reaction mixture was stirred at room temperature
for 36 h. The reaction was quenched with methanol
(1 mL), stirred for an additional 30 min, diluted with Et2O
(100 mL), and washed with saturated aqueous NH4Cl (3·
25 mL). The organic phase was dried (NaSO4), filtered and
concentrated in vacuo to give the corresponding silyl deriva-
tive, 12 (1.35 g, 91.2%). [a]D = ꢁ28.6 (c 0.73, MeOH) IR
(NaCl, thin film) mmax 2916.3, 2860.0, 1757.8, 1713.3,
1380.2, 1162.2 cmꢁ1 1H NMR and gCOSY (CDCl3,
500 MHz) d 0.01 (s, 3H), 0.06 (s, 3H), 0.87 (s, 9H), 1.21 (d,
3H, J = 6.5 Hz), 1.48 (s, 9H), 3.74 (s, 3H), 4.24 (dd, 1H,
J = 2, 8 Hz), 4.44 (m, 1H), 5.20 (1H, J = 8 Hz) 13C NMR
(CDCl3, 500 MHz) d ꢁ5.2, ꢁ4.3, 17.9, 20.7, 25.5, 28.3,
52.1, 59.5, 68.8, 79.7, 156.2, 171.7 HRMS (ESI+): m/z calcd
for (M+H) C16H33NO5Si, 348.2206. Found: 348.2208.
2.6.7. (4S,5R)-tert-Butyl 4-formyl-2,2,5-
trimethyloxazolidine-3-carboxylate (11)
Oxazolidine alcohol 10 (85 mg, 0.35 mmol) was dissolved
in ethyl acetate (2.5 mL, 0.14 M final concentration), and
IBX (307.2 mg, 1.1 mmol) was added. The resulting suspen-
sion was refluxed in an oil bath set to 80 ꢁC with vigorous
stirring. After 2 h (TLC monitoring), the reaction was cooled
to room temperature and filtered through a medium glass
frit. The filter cake was washed with ethyl acetate (3·
2 mL), and the combined filtrates washed with pre-chilled
NaHCO3 (1·). The organic layer was collected, dried over
Na2SO4, filtered and concentrated in vacuo to yield 82 mg
(98% yield) of the desired product 11 as a mixture of rotom-
ers (2:1). Note: the aldehydes were immediatelyused for depro-
tection. Prolonged storage in an organic solvent led to
decomposition of the product. IR (NaCl, thin film) mmax
2963.8, 2860.2, 1717.4, 1684.9, 1374.0, 1264.5 cmꢁ1 1H
NMR and gCOSY (CDCl3), 300 MHz) d 1.34 (d, 3H,
J = 6 Hz), 1.41, 1.49 (s, 9H,), 1.55–1.64 (m, 6H), 4.02–4.08,
4.19–4.26 (m, 1H), 3.68, 3.80 (dd, 1H, J = 2.4, 2.7 Hz),
9.37, 9.46 (d, 1H, J = 1.8 Hz) 13C NMR (CDCl3,
300 MHz) and gHMQC d 17.6, 17.7, 25.0, 25.8, 26.2, 27.3,
28.1, 28.2, 69.8, 70.0, 70.9, 71.0, 81.4, 81.5, 94.1, 94.9,
150.9, 152.5, 197.5 HRMS (ESI+): m/z calcd for
C12H21NO4 (M+H), 244.1549. Found: 244.1546.
2.6.10. (2R,3R)-2-(tert-Butoxycarbonylamino)-3-O-(tert-
butyldimethylsilyl)-1,3-butandiol, (13)
A well-stirred suspension of LiAlH4 (0.59 g, 15.5 mmol)
in anhydrous THF (24 mL) was maintained at ꢁ50 ꢁC.
Boc-protected silyl derivative, 12 (1.35 g, 3.88 mmol) in
anhydrous THF (2 mL) was added over a period of
20 min. The mixture was stirred at ꢁ50 ꢁC for an additional
30 min, diluted with 1 M phosphate buffer at pH 7 (3 mL)
and EtOAc (30 mL), warmed to room temperature, and fil-
tered through a pad of Celite. The organic layer was collected
and concentrated. The residue was eluted from a column of
silica gel with 4:1 cyclohexane–AcOEt to give product 13
(0.99 g, 80%) as a thick colorless oil. [a]D = ꢁ6.9 (c 0.8,
CHCl3) lit. [a]D = ꢁ7.5 (c 0.8, CHCl3) IR (NaCl, thin film)m-
2.6.8. (R)-2-Aminobutane-1,1,3-triol trifluroacetate salt (4)
A solution of aldehyde 11 (0.14 g, 0.58 mmol) was stir-
red in deuterated acetone (15% in D2O) at room tempera-
ture. Trifluoroacetic acid (N2 flushed, Aldrich, 0.21 g,
1.84 mmol) was added drop-wise to this solution. A brown
3446.3, 2929.1, 2850.6, 1702.7, 1504.2, 1169.2 cmꢁ1 1H
max
NMR and gCOSY (CDCl3, 500 MHz) d 0.09 (s, 3H, Si–
CH3), 0.1 (s, 3H, Si–CH3), 0.90 (s, 9H, Si–C(CH3)3), 1.19
(d, 3H, J = 6 Hz), 1.47 (s, 9H), 3.52–3.70 (m, 3H), 4.06–

G.T. Kelly et al. / Bioorganic Chemistry 36 (2008) 4–15
9
4.11 (m, 1H), 4.91 (1H, J = 8 Hz) 13C NMR (CDCl3,
500 MHz) d ꢁ5.3, ꢁ4.3, 17.8, 20.8, 26.0 , 28.3, 57.3, 64.2,
67.7,79.7, 156.8 HRMS (ESI+): m/z calcd for (M+H)
C15H33NO4Si, 320.2257. Found: 320.2260.
30 mL). The organic layer was collected, dried over Na2SO4,
filtered and concentrated in vacuo to give crude product 16 as
a thick yellow oil and a mixture of rotamers as a thick yellow
oil (0.17 mg, 90%). IR (NaCl, thin film) mmax 2917.2, 2847.7,
1732.6, 1697.8, 1366.3 cmꢁ1 1H NMR(CDCl3, 500 MHz) d
1.43, 1.51(s, 9H),1.48, 1.54 (s, 3H), 1.66, 1.72 (s,3H), 2.20,
2.22 (s, 3H), 3.93, 3.98 (dd, 1H, J = 3.0, 6.5 Hz), 4.10–4.18
(m,1H), 4.28, 4.42 (dd, 1H, J = 3.0, 5.5 Hz) 13C NMR
(CDCl3, 500 MHz) d 24.7, 25.6, 25.4, 25.9, 26.3, 26.5, 28.3,
28.2, 65.1, 65.5, 65.5, 65.6, 81.1, 80.7, 94.5, 95.2, 151.4,
152.4, 206.5, 207 HRMS (ESI+): m/z calcd for (M+H)
C12H21NO4, 244.1549. Found: 244.1537.
2.6.11. (4R)-4-[(R)-1-O-(tert-butyldimethylsilyl)ethyl]-
2,2-dimethyl-N-(tert-butoxycarbonyl)-1,3-oxazolidine, (14)
A solution of Boc-protected silyl alcohol, 13 (0.85 g,
2.66 mmol) and 2-methoxypropene (1.73 g, 23.98 mmol) in
acetone (10 mL) was maintained at 0 ꢁC. 10-Camphorsulf-
onic acid (61.79 mg, 0.26 mmol) was added and the mixture
stirred for 1 h at 0 ꢁC, followed by 30 min at room tempera-
ture. The reaction was quenched with Et3N (0.2 mL, color
changes from dark red to yellow) and the solvent was
removed under reduced pressure. The residual brown syrup
was partitioned between diethyl ether (50 mL) and saturated
aqueous sodium bicarbonate solution (4· 30 mL). The
organic layer was collected and washed once with brine.
The organic phase was dried (Na2SO4) and concentrated to
give the desired product 14 as a mixture of rotamers (2:1)
(0.9 g, 95%). IR (NaCl, thin film) mmax 2929.3, 2856.7,
1705.9, 1471.3, 1381.9, 1256.2 cmꢁ1 1H NMR and gCOSY
(CDCl3, 500 MHz) d 0.07 (m, 6H), 0.88 (s, 9H), 1.09 (d,
3H, J = 6.5 Hz), 1.27, 1.34 (s, 3H), 1.47, 1.49 (s, 9H), 1.56,
1.62 (s, 3H), 3.78–3.82, 3.92–3.96 (m, 1H), 3.87–3.92, 4.14–
4.19 (m, 1H) 4.22–4.28, 4.31–4.37 (m, 1H) 13C NMR
(CDCl3, 500 MHz) d ꢁ4.8, ꢁ4.7, 17.8, 22.6, 22.7, 25.7,
28.1, 28.3, 28.6, 28.6, 29.7, 60.6, 61.2, 62.8, 63.0, 66.2, 66.9,
79.7, 79.9, 94.6, 93.9, 152.3, 152.7 HRMS (ESI+): m/z calcd
for (M+H) C18H37NO4Si, 360.2570. Found: 360.2580.
2.6.14. tert-Butyl (S)-1-hydroxy-3-oxobutan-2-ylcarbamate
(17)
A solution of ketone 16 (0.17 g, 0.71 mmol) in CH2Cl2
(1 mL) was mixed with 0.5 M TFA in anhydrous CH2Cl2
(15 mL) and stirred at room temperature for 15 min. The
reaction was quenched with cold NaHCO3 (30 mL) and
extracted in ether (3· 30 mL). The organic layer was col-
lected, dried over Na2SO4, filtered and concentrated in vacuo
to give the crude product as a thick yellow oil. The crude
material was then purified by column chromatography by
eluting with ethyl acetate:hexane (1:3) Rf = 0.25 to yield
acetyl glycinol 17 (0.12 g, 85%). IR (NaCl, thin film) mmax
3418.4, 2919.2, 2851.1, 1711.9, 1687.7, 1366.7 cmꢁ1 1H
NMR(CDCl3, 500 MHz) d 1.46 (s, 9H), 2.29 (s, 3H), 3.96
(ddd, J = 4, 8, 25 Hz), 4.32 (m, 1H), 5.68 (br,1H) 13C
NMR (CDCl3, 500 MHz) d 27.3, 28.2, 62.1,62.9, 80.2,
155.9, 205.6 HRMS (ESI+): m/z calcd for (M+H)
C9H17NO4, 204.1236. Found: 204.1239.
2.6.12. (4R)-4-[(R)-1-Hydroxyethyl]-2,2-dimethyl-N-
(tert-butoxycarbonyl)-1,3-oxazolidine (15)
2.6.15. 3-Amino-4-hydroxy but-3-en-2-one trifluoroacetate
salt (5)
A solution of the silyl ether derivative (0.9 g, 2.82 mmol)
in anhydrous THF (10 mL) was treated with n-
Bu4N+FꢁÆ3H2O (3.38 mL (1 M in THF), 3.39 mmol) at
room temperature for 8 h and concentrated. The residue
was dissolved in CH2Cl2 (20 mL), washed with H2O (2·
20 mL), dried (Na2SO4), and concentrated to give the crude
product as a dark yellow oil. The residue was eluted from a
plug of silica gel with 4:1 hexane–AcOEt (containing 0.3%
of Et3N) to afford the product (15) as a white solid (0.55 g,
88%). IR (NaCl, thin film) mmax 3443.4, 2972.7, 2827.3,
1696.7, 1379.9 cmꢁ1 1H NMR(CDCl3, 500 MHz) d 1.14
(d, 3H, J = 6.5 Hz), 1.46 (s, 3H),1.47 (s, 9H), 1.55 (s,
3H), 3.78–3.98, 4.11–4.19 (2m, 4H) 13C NMR (CDCl3,
500 MHz) d 24.4, 25.9, 28.2, 28.3, 28.4, 29.8, 63.2, 64.7,
70.1, 81.3, 94.0,152.3 HRMS (ESI+): m/z calcd for
(M+H) C12H23NO4, 246.1705. Found: 246.1697.
A suspension of ketone 17 (0.08 g, 0.4 mmol) in dichloro-
ethane (2 mL) was refluxed gently with IBX (0.7 g, 1.2 mmol,
45%) for 3 h. The suspension was allowed to cool to room
temperature and further cooled at ꢁ20 ꢁC for 20 min. The
resulting white precipitate was filtered. The filtrate was
mixed with aqueous TFA (50 mg in 1 mL water) and allowed
to stir for 25 min at room temperature. The aqueous layer
was collected and characterized. Note: the keto-amino alde-
hyde was found to be unstable to isolation conditions. Any
attempts to concentrate this material led to extensive decom-
position of the product. This precursor was used for feeding
studies immediately. Stability studies indicated the presence
of ꢀ35% of the compound after 48 h in D2O at room temper-
ature. 1H NMR (D2O, 500 MHz) d 1.69(s,3H), 1.83 (s,
3H), 4.69 (s, 1H), 5.06 (s, 1H). HRMS (ESI+): m/z calcd
for (M+H) C4H7NO2, 102.0555. Found: 102.0551.
2.6.13. (4R)-4-Acetyl-2,2-dimethyl-N-(tert-
butoxycarbonyl)-1,3-oxazolidine, (16)
3. Results and discussion
A mixture of alcohol 15 (0.19 g, 0.79 mmol), and Dess–
Martin reagent (0.37 g, 0.88 mmol) in anhydrous CH2Cl2
(10 mL) was stirred at room temperature for 2 h away from
light. The reaction was quenched with a 1:1 solution of cold
Na2S2O3 and NaHCO3(v/v) and extracted in ether (3·
3.1. Development of the culture system
As the production of secondary metabolites from Strep-
tomycetes frequently coincides or precedes the formation of

10
G.T. Kelly et al. / Bioorganic Chemistry 36 (2008) 4–15
aerial hyphae on solid media [10], we initially examined a
variety of media conditions including LB, ISP+ [11],
NB+, YPD [12], TO [13], GYM [4], PS5 [4], MS [14],
YEME [14], and R2YE [14]. Several solid media conditions
were evaluated by streaking a loop full of S. sahachiroi
spore stock upon prepared plates. The growth was evalu-
ated at regular intervals. All plates showed growth of
Streptomyces, however, only GYM, MS, PS5 and PS5+
plates exhibited sporulation (see Supporting information
A, Table 1). As azinomycin B production was undetectable
in organic extracts of the spores, the plates were inoculated
into Erlenmeyer flasks containing 100 mL of PS5 medium.
After a 72 h post-inoculation period at 30 ꢁC, the cultures
were extracted with dichloromethane, concentrated, and
analyzed by TLC and mass spectrometry. Results revealed
that while the MS plates showed traces of azinomycin B
production, the GYM plates produced the most robust
and productive spores at 5–7 days growth when inoculated
from spores stored on dehydrated GYM plates (see Sup-
porting information A, Table 1).
lated with two 24 h first stage cultures (100 mL, generated
from GYM plates), agitated at 250 rpm, and aerated at
6 L/min with sterile filtered air for 72 h. The experiment
resulted in considerable foaming leading to loss of culture
and low production of the azinomycins. We, therefore,
decreased the volume of medium to 10 L and repeated
the experiment inoculating with two 24 h second stage cul-
tures (600 mL each) that were generated by inoculating 2 L
baffled flasks with 25 mL of a 24 h first stage culture. As
azinomycin production was minimal, nutrient starvation
was explored (see Table 1, entry 4). The component com-
position of the PS5 medium (10 L) was reduced by 50%
and inoculated with two 600 mL second stage cultures.
After 72 h, the cells were harvested and ꢀ30 mg of azino-
mycin B was obtained from 10 L. To further investigate
the stress imposed by medium deprivation, the medium
composition was reduced by 75%, inoculated with two sec-
ond stage cultures (600 mL), and harvested after 72 h. This
sequence yielded approximately 40 mg of azinomycin B
from 10 L of culture. Further reduction of the PS5 medium
composition resulted in lower levels of azinomycin produc-
tion (Fig. 1c) [15]. By increasing the two second stage cul-
tures to 1 L and the aeration rate to 8 L/min, we can now
reproducibly obtain about 60 mg of azinomycin from 10 L.
Interestingly, the pH of the cell culture steadily increased to
ꢀ8.0 over a period of 4 days. Fig. 1b and d show the pH
profile and amount of azinomycin B produced as functions
of time. Optimal production of the natural product was
observed between 64–72 h (Fig. 1d) corresponding to pH
7.4–7.8 (Fig. 1b), probably due to instability of the mole-
cule under strongly acidic or alkaline conditions.
Proceeding to the evaluation of liquid culture condi-
tions, we examined a variety of media formulations
(YPD, R2YE, PS5, PS5+, LB, TO, GYM, MS, YEME,
NB+, and ISP+) [11–14] (see Supporting Information A
for media formulations and abbreviations) to support azi-
nomycin production by S. sahachiroi in shake flasks and by
fermentation (as summarized in Table 1 and Supporting
information A, Table 2). Since natural product production
is generally confined to stationary phase of the growth
curve, in shake culture (Erlenmeyer flasks or baffled flasks)
we evaluated each medium composition by generating a
first stage culture (from GYM plates), which was used to
inoculate and produce the culture in large scale (in 2 L baf-
fled flasks). The cells were harvested at a fixed time of 72 h.
The cultures were centrifuged and extracted with dichloro-
methane. Under these growth conditions, azinomycin B
production was observed in YPD, R2YE, PS5, and
PS5+, but not in LB, TO, GYM, MS, YEME, NB+,
Manual adjustment of the pH throughout the culture
period with bicarbonate did not have a dramatic effect on
production levels, 40–60 mg of azinomycin was obtained.
3.2. Whole cell feeding studies: biosynthetic route to the enol
fragment
1
and ISP+ as determined by TLC, mass spectrometry, H
With a reliable culture system in hand, our aim was to
employ these new growth conditions to investigate the for-
mation of the enol fragment of azinomycin B. Initially, to
examine our feeding conditions, we fed [1-13C] acetate
and [methyl-13C]methionine to cell suspensions of S. sah-
achiroi. The isotopically labeled compounds were provided
to the cultures in two separate aliquots (in equal portions),
the first after 24 h of incubation and the second 24 h later.
As expected, whole cell feeding of [1-13C] acetate gave rise
to an alternate labeling pattern within the naphthoate frag-
ment owing to its PKS origin and confirming the results
reported by Lowden [8] (Fig. 2a and 2b, blue C20-C8a0,
7.9–11.4% incorporation). Furthermore, both carbons C1
and C4 revealed moderate incorporation (Fig. 2a and b,
blue, C1, 2.8% C4, 2.9%) supporting the hypothesis that
scrambling of label could occur to give threonine via oxa-
loacetate in the Kreb’s cycle [16]. The results further sug-
gest that C-14 (Fig. 2a and b, blue, 8.2%) of the molecule
is derived from acetate and does not arise from rearrange-
NMR and 13C NMR spectroscopy. Azinomycin B produc-
tion, however, remained erratic, with the final pH of the
culture varying from pH 4.25 to as high as pH 8.75.
As the highest levels of azinomycin B production typi-
cally correlated with an ending pH value of 8, we attempted
to adjust the pH throughout the culturing period (using
PS5 medium) by the addition of base (NaOH) or use of
phosphate buffer but improvement in either case was not
observed (see Supporting information Table 2 entry 9
and 10). Nutrient limitation was also investigated as well
as varying the culture period from 1–6 days. In all cases,
azinomycin production was low and/or inconsistent. As
modifying liquid shake culture protocols did not give much
improvement, we decided to explore usage of a large scale
fermentation system (Fig. 1). Such an approach would
afford more control over several factors not easily con-
trolled with shake flasks such as degree of aeration and
foaming. Initially, the fermenter (15 L capacity) was inocu-

G.T. Kelly et al. / Bioorganic Chemistry 36 (2008) 4–15
11
Fig. 1. (a) Sequence of fermentation system: culture of S. sahachiroi (b) pH profile as a time course. (c) Effect of nutrient deprivation on production of
azinomycin B. (d) Time course production of azinomycin B with 75% nutrient deprivation (second stage culture, 600 mL baffled flask). Crude yield of
azinomycins and purified azinomycin B are shown as a function of time.
Table 1
Examination of culture conditions for azinomycin production in Streptomyces sahachiroi
Entry
Culture conditions
Result
1
2
3
Solid media, culture flasks, 72 h
Solid media, first stage culture, baffled flasks, 72 h
Solid media, first stage culture, fermenter (300 rpm aerated at 6 L/min),
72 h
Little or no production
Little production, erratic
Little production, erratic
4
5
6
7
Solid media, first stage culture, second stage culture (600 mL baffled flask),
fermenter (250 rpm aerated at 6 L/min), 72 h
Traces of azinomycin B observed
Solid media, first stage culture, second stage culture (600 mL baffled flask),
50% nutrient deprivation (300 rpm aerated at 8 L/min), 72 h
Solid media, first stage culture, second stage culture (600 mL baffled flask),
75% nutrient deprivation (300 rpm aerated at 8 L/min), 72 h
Solid media, first stage culture, second stage culture (1 L baffled flask), 75%
nutrient deprivation (300 rpm aerated at 8 L/min), 72 h
ꢀ15 mg azinomycin B from 10 L of
culture
ꢀ30–40 mg azinomycin B from 10 L of
culture
ꢀ60 mg azinomycin B from 10 L of
culture
ment or Baeyer–Villiger oxidation of a more advanced pre-
cursor. Likewise, feeding of [methyl-13C]methionine to cul-
tures of S. sahachiroi gave an unambiguous clear
enhancement of signal (33.8% incorporation, Fig. 2b) at
the methoxy carbon of the molecule (Fig. 2a, green). The
finding not only supports the involvement of the co-factor
SAM (S-adenosylmethionine) in the biosynthesis of the
methoxy group of the naphthoate, but also excludes the
involvement of SAM in the formation of the aziridinopyrr-
olidine ring system (specifically, the electrophilic C-10 car-
bon does not arise from SAM).
Having established a proper feeding regimen with our
new culture system, our focus shifted to utilizing our new
fermentation strategy to probe the biosynthesis of the late
stages of the pathway, in the construction of the enol frag-
ment of azinomycin B. The structure of the enol fragment
suggests that it might originate from (L)-threonine the b-
alcohol of the amino acid would necessitate oxidation
and the terminal carboxylate reduction (Scheme 2). Our
earlier results from cell-free extract studies [9] and acetate
labeling experiments support such an argument. Neither
approach has shown, however, that the amino acid is

12
G.T. Kelly et al. / Bioorganic Chemistry 36 (2008) 4–15
Fig. 2. (a) Summary of feeding studies: [methyl-13C]methionine incorporates at –OMe [1-13C]sodium acetate incorporates at C1, C4, C6, C14, C10CO–,
C20, C40, C50, C70, C8a0 [U–13C]threonine incorporates C1–C4. (b) % incorporation = [(A ꢁ B)/B] · 1.07 where A, intensity of labeled carbon; B, intensity
of unlabeled carbon; 1.07, is the natural abundance of 13C; n.d., not detectable by 13C NMR. (c) Representative comparison of C1 of azinomycin B via
feeding of [U–13C]-labeled threonine, 3, 4 and 5 to that of the negative control.
O
O
OH
NH2
3
OH
O
O
OH
O
O
OH
H
H
NH2
NH2
NH2
5
OH OH
2
OH
O
H
H
NH2
NH2
4
Scheme 2. Proposed biosynthetic routes to the enol fragment of azinomycin B.
site-specifically incorporated into the enol fragment of the
in the oxidation of the alcohol and reduction of the carbox-
ylate of 2, we synthesized b-ketoamino acid 3, b-hydroxya-
mino aldehyde 4, and b-ketoaminoaldehyde 5 in labeled
form. Synthesis of such precursors is important to establish
whether the enol fragment or its respective intermediates
are pre-formed prior to loading onto the NRPS or gener-
ated at a later stage in the biosynthesis.
natural product. We, therefore, examined uptake of
[U–13C]- L-threonine (2, Scheme 2). The experiment
resulted in labeling of all four carbons (C1–C4) of the enol
fragment (Fig. 2a, red). This ‘tail-to-tail’ incorporation
clearly suggested that threonine was site-specifically incor-
porated. Close inspection of the 13C NMR indicated intact
incorporation (C1–C4, 7.9–11.7% incorporation) of the
[U–13C]-labeled threonine. Isotopic labeling was evident
at C1 as a doublet of a doublet (J1 = 55.8, J2 = 180.0 Hz)
flanking the natural abundance peak at 24.1 ppm. Simi-
3.3. Synthesis
The synthetic routes were designed to allow easy and
efficient preparation in high yields, selective deprotection
of O- and N-functional groups and minimal chromato-
graphic separations to afford synthesis of the molecules in
a cost effective manner for the implementation of stable
isotopes. The availability of synthetic routes to a-amino-
larly, C4 gave
a
doublet of doublet (J1 = 38.7,
J2 = 324.6 Hz) flanking the natural abundance peak at
149.5 ppm. Multiplets were observed for carbons C2 and
C3 probably owing to extensive 13C–13C coupling. To
address the order of events, the timing of the biosynthesis

G.T. Kelly et al. / Bioorganic Chemistry 36 (2008) 4–15
13
compounds (8, Scheme 3), oxazolidine-aldehyde (11,
Scheme 3), and amino alcohol (17, Scheme 4) is of addi-
tional importance as they can serve as models for stereo-
chemical studies [17] as well as serve as building blocks in
the synthesis of amino-sugars [18], aza-sugars [19],
sphingosines [20], and unnatural amino acids/derivatives
[21].
The syntheses of all three precursors 3, 4 and 5 began
from commercially available (L)-threonine. Threonine, 2
was converted into its methyl ester hydrochloride using
SOCl2 in MeOH under reflux conditions. The methyl ester
6 was then protected with Boc-anhydride in THF in the
presence of triethylamine to provide the Boc-protected
methyl ester of threonine 7 in ꢀ88% yield. This Boc-pro-
tected methyl ester 7 served as the common substrate for
all precursors (Scheme 3).
pletely as an oxazolidine 9 which when subjected to reduc-
tion with LiAlH4 afforded the protected alcohol 10 in
excellent yields. The alcohol was further oxidized to the
threonine analog of Garner’s aldehyde 11 in 98% yield.
The protected amino-aldehyde was not stable over long
periods of time and was immediately subjected to deprotec-
tion by treatment with TFA/D2O. This TFA salt of the
amino aldehyde hydrate system generated was unstable
to isolation and was characterized and directly utilized in
feeding experiments.
3.3.3. b-Ketoaminoaldehyde 5
We next examined a synthetic route for the preparation
of b-ketoaminoaldehyde 5. Since the b-ketoenamine system
is synthetically equivalent to the corresponding ketoalde-
hyde, we expected it to be directly formed via oxidation
of the corresponding 1,3 diol system. Treatment with
TPAP/NMO [23], PCC [24], PDC [25], Dess–Martin [26],
and IBX(2-iodoxy benzoic acid) [27], however, resulted in
a complex mixture of products unstable to column chroma-
tography. Alternatively, access to the Boc-protected
aminoaldehyde using reduction of Boc-protected esters or
Weinreb amides [28] led to overreduction to the corre-
sponding alcohol or complicated and tedious column chro-
matographic separations, not suitable for implementation
of stable isotopes. Based on synthesis and stability studies
of 3, we envisaged N-Boc-protected acetyl glycinol 17
(Scheme 4) as a convenient moiety to undergo oxidation
and subsequent deprotection. Thus, chiral acetyl oxazoli-
dine 16 was synthesized from the common precursor tert-
butoxycarbonyl-protected methyl L-threoninate 7 using a
3.3.1. b-Keto amino acid 3
Upon oxidation with Dess–Martin periodinane, Boc-
protected threonine methyl ester was converted into the
corresponding Boc-protected keto-ester 8 in 91%, which
when hydrolyzed with TFA/H2O under reflux afforded
the desired compound 3 in an overall 72% yield (Scheme 3).
3.3.2. Hydroxy-aldehyde 4
The hydroxy-aldehyde 4 (Scheme 3) was synthesized
over six steps by modification of the synthesis of Garner’s
aldehyde [22]. The N-Boc 2,2-dimethyl oxazolidine ring
constituted a convenient masked modified amino acid that
was expected to tolerate synthetic elaboration. Thus, Boc-
protected amino acid methyl ester, 7 was protected com-
HO
COOH
NH2
HO
COOMe
NH2. HCl
O
COOH
HO
COOMe
NHBoc
O
COOMe
NHBoc
i
ii
iii
iv
NH
2.TFA
2
8
6
7
3
HO
CH2 OH
NBoc
COOMe
CHO
CHOH
OH
vv
iv
ii
viii
7
O
NBoc
O
O
NBoc
.TFA
NH2
HO
NH2 .TFA
HO
9
10
11
4
Scheme 3. Reagents: (i) MeOH, HCl (dry) reflux, 2 h; (ii) Boc2O, Et3N, THF, 0ꢁC–r.t. 14 h, then 50 ꢁC 3 h; (iii) Dess–Martin periodinane, CH2Cl2, r.t.,
1 h; (iv) TFA (aq.), reflux; (v) 2- methoxypropene, acetone, r.t., 3 h; (vi) LiAlH4, THF, 400; (vii) IBX, EtOAc, reflux, 3 h; (viii) TFA, water, 250.
OTBDMS
OH
OTBDMS
COOMe
OTBDMS
CH2OH
i
ii
iii
iv
7
O
O
BocN
BocN
NHBoc
12
NHBoc
13
14
15
O
O
O
v
vii
vi
O
OH
NH2.TFA
OH
NHBoc
17
BocN
16
5
Scheme 4. Reagents: (i) TBDMSCl, DMAP, Et3N, dry DMF, 0 ꢁC–r.t., 36 h; (ii) LiAlH4, THF, ꢁ50 ꢁC, 300; (iii) 2-methoxypropene, acetone, r.t., 2 h; (iv)
TBAF, THF, 8 h; (v) Dess–Martin periodinane, CH2Cl2, r.t., 1 h; (vi) 0.5 M TFA, CH2Cl2, r.t., 150; (vi) IBX, DCE, 3 h, cool, filter then TFA, H2O, 150.

14
G.T. Kelly et al. / Bioorganic Chemistry 36 (2008) 4–15
modified procedure employed by Dondoni et al. [29]. This
ester was first transformed into the alcohol 13 by silylation
(91%) of the secondary hydroxyl group with tert-butyldim-
ethylsilyl chloride (TBDMSCl) and subsequent reduction
of the ester group with LiAlH4. Our earlier efforts and
studies by others [29] revealed that the selective step-wise
protection of the two hydroxyl groups was important. Ace-
tonation with 2-methoxypropene (95%) and desilylation
(n-Bu4N+Fꢁ) converted 13 into the (R,R)-hydroxyethyl
oxazolidine 15 (88%) in rotameric forms. The secondary
alcohol was then subjected to oxidation with Dess–Martin
periodinane to yield 16. Use of PCC for oxidation gave a
complex mixture of products that remained impure even
after column chromatography. The oxazolidine ring was
cleaved with very dilute TFA (boc group still intact) to
yield the desired acetyl Boc-protected glycinol 17. Alterna-
tively, the acetyl Boc-protected glycinol was also synthe-
sized starting from N-Boc-protected serine, which was
converted into its Weinreb amide. Further acetonation
with 2-methoxypropene, followed by treatment with organ-
olithium (MeLi) in presence of CH3MgBr as the sacrificial
base provided acetylation of the a-C [30]. However, overall
yields from this methodology and purification were not
found to be cost-effective in our hands. This primary alco-
hol was now ready for oxidation and subsequent final
deprotection. Heterogenous IBX oxidation [27] provided
a mild method to form the aldehyde, however, all efforts
to isolate the keto-amino aldehyde resulted in rapid decom-
position of the product. Hence, the reaction was simply fil-
tered and treated with TFA/H2O for 25 min at room
temperature to afford final deprotection, producing the
desired keto-amino aldehyde 5.
Each of the threonine derivatives (compounds 3, 4, and
5) were synthesized in universally labeled form and fed
individually to whole cell suspension cultures as detailed
previously. Interestingly, none of these amino acid precur-
sors gave any site-specific incorporation above background
(Fig. 2c, supporting information c). As a control for cellu-
lar uptake, since the majority of the amino acid derivatives
were synthesized as their respective TFA salts, we fed
[methyl-13C] methionine as its TFA salt and monitored
its incorporation ([methyl-13C]methionine labels specifi-
cally the methoxy group of azinomycin B, Fig. 2a, b and
supporting information b). Intact incorporation was
observed as reported previously, confirming that the TFA
salts of amino acids are capable of penetrating the cell
membrane of the producer strain, S. sahachiroi. As noted
previously [31], both of the enol systems (4 and 5) were
found to be relatively unstable as compared to the corre-
sponding acetylglycine 3. Fig. 3 illustrates the stability
analysis of the compounds. After 48 h at room temperature
under aqueous conditions, only 70% of the hydroxyl-amino
aldehyde 4 and 35% of the keto-amino aldehyde 5 were
detected (Supporting Information B, Table 4).
120
110
100
90
80
70
60
50
40
30
20
10
0
3
4
5
0
10
20
30
Time
40
50
60
70
Fig. 3. Stability curve as a function of time for 3, 4 and 5.
been precursors in the biosynthesis, intact incorporation
into the molecule should have been detectable. The gate-
keeper in the biosynthesis of the enol fragment is presumed
to be that of the NRPS adenylation domain, where in this
case, based upon data reported here, threonine appears to
be the most advanced precursor to be accepted by the mod-
ule. One possibility is that the adenylation domain itself
catalyzes (all or in part) the oxidation of the alcohol and
the reduction of the carboxylate. In myxochelin biosynthe-
sis, reductase domains have been found to replace the TE
(thioesterase) domain in the gene cluster resulting in release
of the natural product with concomitant production of an
aldehyde moiety [32]. Moreover, adenylation domains have
also been found to harbor N- and C-methyltransferases,
decarboxylases, as well as a-ketoreductase domains [33].
Alternatively, the required modifications might be post-
NRPS processes. The localization of the azinomycin bio-
synthetic gene cluster will certainly shed light on these
issues.
4. Conclusion
The azinomycins represent a structurally unusual class
of DNA crosslinking agents produced by Nature. While
the in vivo actions of azinomycin B show significant prom-
ise, the clinical potential of the agent remains largely unex-
plored due to dearth of synthetic procedures for the
molecule as well as unreliable culture methods for the pro-
ducing organism. We report here a reliable fermentation
procedure for S. sahachiroi, the azinomycin producer,
where up to 60 mg of purified compound can be obtained
from 10 L of culture. We successfully employed this system
to investigate the biosynthetic origin of azinomycin B, prin-
cipally in the late stages of the pathway, the identification
of the source of the enol fragment of the molecule. Threo-
nine and advanced precursors (3, 4 and 5) that were synthe-
sized in isotopically labeled form were fed to suspension
cultures of the Streptomyces strain. The results indicated
that threonine is the most advanced precursor accepted
by the NRPS module for further processing. Elucidation
of the azinomycin biosynthetic gene cluster is currently in
progress, which will pave the way for detailed characteriza-
Despite the relative instability of some of these amino
acid derivatives, given our feeding regimen (two separate
and equal aliquots fed 24 h apart), had these compounds

G.T. Kelly et al. / Bioorganic Chemistry 36 (2008) 4–15
15
[9] C. Liu, G.T. Kelly, C.M.H. Watanabe, Organic Lett. 8 (2006) 1065–
1068.
[10] M.J. Bibb, Curr. Opin. Microbiol. 8 (2005) 208–215.
[11] E.B. Shirling, D. Gottlieb, Int. J. Syst. Bacteriol. 16 (1966) 313–340.
tion and functional analysis of individual biosynthetic steps
of the pathway.
Acknowledgments
[12] J. Sambrook, D.W. Russell, Molecular Cloning
a Laboratory
Manual, third ed., Cold Spring Harbor Laboratory Press, Cold
Spring Harbor, New York, 2001, pp. A2.12.
[13] L.-H. Malmberg, W.-S. Hu, D.H. Sherman, J. Bacteriol. 175 (1993)
6916–6924.
The authors gratefully acknowledge financial support
from the Elsa Pardee Foundation, Welch Foundation
(A-1587), Texas A&M University, American Cancer Soci-
ety, and the NIH sponsored CBI Program (GM008523)
for providing partial research support for G.T. Kelly. We
thank the NMR facility at Texas A&M University and
Dr. D. Holmes at Michigan State University for helpful
discussions.
[14] T. Kieser, M.J. Bibb, M.J. Buttner, K.F. Chater, D.A. Hopwood,
Media, buffers and suppliers, in: Practical Streptomyces Genetics, The
John Innes Foundation, Norwich, United Kingdom, 2000, pp. 408–
418.
[15] By increasing the nutrient deprivation to 90%, we isolated only 17 mg
of pure azinomycin B out of 65 mg of crude material extracted.
[16] C.K. Mathews, K.E. van Holde, K.G. Ahern, Biochemistry, third ed.,
Addison Wesley Longman, San Francisco, 1999.
[17] (a) R.S. Coleman, A.J. Carpenter, Tetrahedron Lett. 33 (1992) 1697–
1700;
Appendix A. Supplementary data
(b) T. Fujisawa, M. Nagai, Y. Koike, M. Shimizu, J. Org. Chem. 59
(1994) 5865–5867;
(c) F. D’Aniello, A. Mann, M. Taddei, C.-G. Wermuth, Tetrahedron
Lett. 35 (1994) 7775–7778.
Details on the culturing of Streptomyces sahachiroi,
media formulations,1H and 13C spectra of azinomycin B,
details on feeding studies with [1-13C]sodium acetate,
[methyl-13C]methionine, [U–13C]threonine, [U–13C]-3,
[U–13C]-4, [U–13C]-5, comparison of feeding of all threo-
[18] (a) D. Kahne, D. Yang, M.D. Lee, Tetrahedron Lett. 31 (1990) 21–22;
(b) W.R. Roush, J.A. Hunt, J. Org. Chem. 60 (1995) 798–806.
[19] A. Dondoni, P. Merino, D. Perrone, Tetrahedron 49 (1993) 2939–
2956.
[20] (a) S. Nimkar, D. Menaldino, A.H. Merrill, D. Liotta, Tetrahedron
Lett. 29 (1988) 3037–3040;
1
nine derivatives, H and 13C spectra for compounds 3-17
is available via the Internet. Supplementary data associated
with this article can be found, in the online version, at
doi:10.1016/j.bioorg.2007.08.002.
(b) P. Garner, J.M. Park, E. Malecki, J. Org. Chem. 53 (1988) 4395–
4398;
(c) A. Dondoni, G. Fantin, M. Fogagnolo, P. Pedrini, J. Org. Chem.
55 (1990) 1439–1446;
References
(d) K. Soai, K.J. Takahashi, Chem. Soc., Perkin Trans. I (1994) 1257–
1258;
(e) F. Ruan, S. Yamamura, S.-I. Hakomori, Y. Igarashi, Tetrahedron
Lett. 36 (1995) 6615–6618.
[1] (a) K. Nagaoka, M. Matsumoto, J. Oono, K. Yokoi, S. Ishizeki, T.
Nakashima, J. Antibiot. 39 (1986) 1527–1532;
(b) K. Yokoi, K. Nagaoka, T. Nakashima, Chem. Pharm. Bull. 34
(1986) 4554–4561.
[2] S. Ishizeki, M. Ohtsuka, K. Irinoda, K. Kukita, K. Nagaoka, T.
Nakashima, J. Antibiot. 40 (1987) 60–65.
[3] (a) Y. Yamada, T. Kubota, F. Asanuma, T. Yamada, T. Suzuki, K.
Ishibiki, E. Kawamura, Surg. today 23 (1993) 193–199;
(b) S.K. Carter, Cancer Chemother. Rep. Supp. 1 (Pt. 3) (1968) 99–
114.
[4] (a) J.W. Lown, K.C. Majumdar, Can. J. Biochem. 55 (1977) 630–635;
(b) R.S. Coleman, R.J. Perez, C.H. Burk, A. Navarro, J. Am. Chem.
Soc. 124 (2002) 13008–13017;
[21] (a) P. Garner, J.M. Park, J. Org. Chem. 53 (1988) 2979–2984;
(b) N. Sakai, Y. Ohfune, Tetrahedron Lett. 31 (1990) 4151–4154;
(c) I. Jako, P. Uiber, A. Mann, C.-G. Wermuth, T. Boulanger, B.
Norberg, G. Evrard, F. Durant, J. Org. Chem. 56 (1991) 5729–5733;
(d) A. Golebiowski, J. Kozak, J. Jurczak, J. Org. Chem. 56 (1991)
7344–7347.
[22] A. Dondoni, D. Perrone, Organic Syn. 77 (2000) 64–70.
[23] W.P. Griffith, S.V. Ley, G.P. Whitcombe, A.D. White, J. Chem. Soc.,
Chem. Commun. (1987) 1625–1627.
[24] E.J. Corey, J.W. Suggs, Tetrahedron Lett. 31 (1975) 2647–2650.
[25] E.J. Corey, G. Schmidt, Tetrahedron Lett. 20 (1979) 399–402.
[26] D.B. Dess, J.C. Martin, J. Org. Chem. 48 (1983) 4155–4156.
[27] J.D. More, N.S. Finney, Organic Lett. 4 (2002) 3001–3003.
[28] S. Nahn, S. Weinreb, Tetrahedron Lett. 22 (1981) 3815–3818.
[29] A. Dondoni, A. Marra, A. Massi, J. Org. Chem. 64 (1999) 933–944.
[30] (a) R.C. So, R. Ndonye, D.P. Izmirian, S.K. Richardson, R.L.
Guerrera, A.R. Howell, J. Org. Chem. 69 (2004) 3233–3235;
(b) G. Ageno, L. Banfi, G. Cascio, G. Guanti, E. Manghisi, R. Riva,
V. Roccaa, Tetrahedron 51 (1995) 8121–8134.
[31] (a) M.T. Reetz, Angew. Chem. Int. J. Engl. 30 (1991) 1531–1546;
(b) A.G. Myers, D.W. Kung, B. Zhong, J. Am. Chem. Soc. 122 (2000)
3236–3237.
[32] N. Gaitatzis, B. Kunze, R. Muller, Proc. Natl. Acad. Sci. USA 98
(2001) 11136–11141.
(c) R.C. LePla, C.A.S. Landreau, M. Shipman, G.D.D. Jones, Org.
Biomol. Chem. 3 (2005) 1174–1175;
(d) R.W. Armstrong, M.E. Salvati, M. Nguyen, J. Am. Chem. Soc. 14
(1992) 3144–3145;
(e) H. Zang, K.S. Gates, Biochemistry 39 (2000) 14968–14975;
(f) T. Fujiwara, I. Saito, H. Sugiyama, Tetrahedron Lett. 40 (1999)
315–318.
[5] G.T. Kelly, C. Liu, R. Smith III, R.S. Coleman, C.M.H. Watanabe,
Chem. Biol. 13 (2006) 1–8.
[6] R.S. Coleman, J. Li, A. Navarro, Angew. Chem. Int. Ed. Engl. 40
(2001) 1736–1739.
[7] F. Hata, Y. Koga, K. Sano, A. Kanamori, R. Matsumae, T.
Sugawara, T. Hoshi, S. Shimi, S. Ito, S. Tomizawa, J. Antibiot.
(Tokyo) 7 (1954) 107–112.
[8] (a) C. Corre, P.A.S. Lowden, Chem. Commun. (2004) 990–991;
(b) C. Corre, C.A.S. Landreau, M. Shipman, P.A.S. Lowden, Chem.
Commun. (2004) 2600–2601.
[33] E. Conti, T. Stachelhaus, M.A. Marahiel, P. Brick, EMBO J. 14
(1997) 4174–4183.