Synthesis of the liposidomycin diazepanone nucleoside

Article abstract of DOI:10.1021/ol0102904

(formula presented) The synthesis of the liposidomycin degradation product 4 from D-glucose establishes its stereochemistry as 5′S,6′S and, by incorporation of the earlier diazepanone relative stereochemical assignment, establishes the absolute stereochem

Full text of DOI:10.1021/ol0102904

ORGANIC
LETTERS
2002
Vol. 4, No. 4
603-606
Synthesis of the Liposidomycin
Diazepanone Nucleoside
,†
Spencer Knapp,* Gregori J. Morriello,† and George A. Doss‡
Department of Chemistry & Chemical Biology,
Rutgers The State UniVersity of New Jersey, Piscataway, New Jersey 08854-8087,
and Merck & Co., P.O. Box 2000, Rahway, New Jersey 07065-0900
knapp@rutchem.rutgers.edu
Received December 10, 2001
ABSTRACT
The synthesis of the liposidomycin degradation product 4 from D-glucose establishes its stereochemistry as 5′S,6′S and, by incorporation of
the earlier diazepanone relative stereochemical assignment, establishes the absolute stereochemistry of the liposidomycins 1 and 2 as
5′S,6′S,2′′′S,3′′′S.
The liposidomycins,1-4 including B, 1, and C, 2, constitute
a family of complex nucleoside antibiotics5 that inhibit
phospho-N-acetylmuramylpentapeptide transferase (translo-
case I),6,7 which catalyzes the first membrane step in the
peptidoglycan lipid cycle.8 High in vitro specificity and
potency (ID50 0.03 µg/mL) were observed for 2, reflecting
its structural resemblance to uridine diphosphate N-acetyl-
muramylpentapeptide, undecaprenyl phosphate, their transi-
tion state, and/or their product, Lipid I. Extensive NMR and
mass spectral studies by Isono, Ubukata, McCloskey, and
their co-workers led to the liposidomycin structural assign-
ments illustrated, but these stopped short of specifying the
relative and absolute stereochemistry at C-5′, C-6′, C-2′′′,
and C-3′′′. Model studies on 3,6,7-trisubstituted-1,4-dimeth-
yldiazepan-2-ones9 showed close NMR correlation of a
6′S,2′′′S,3′′′S isomer (or its enantiomer) with 1/2 and their
degradation products, leaving only the absolute stereochem-
istry at C-5′ and C-6′ to be determined to completely assign
the diazepanone nucleoside portion of 1/2. Fortunately, the
degradation studies3 had produced a nucleoside triol acid (see
3/4) that features C-5′ and C-6′ as the only unassigned
stereogenic centers, and thus an exact match of this com-
pound with synthetic material of known structure would
provide the missing stereochemical information.
† Rutgers The State University of New Jersey.
‡ Merck & Co.
(1) Isono, K.; Uramoto, M.; Kusakabe, H.; Kimura, K.; Izaki, K.; Nelson,
C. C.; McCloskey, J. A. J. Antibiot. 1985, 38, 1617-1621.
(2) Ubukata, M.; Isono, K.; Kimura, K.; Nelson, C. C.; McCloskey, J.
A. J. Am. Chem. Soc. 1988, 110, 4416-4417.
(3) Ubukata, M.; Kimura, K.; Isono, K.; Nelson, C. C.; Gregson, J. M.;
McCloskey, J. A. J. Org. Chem. 1992, 57, 6392-6403.
(4) Esumi, Y.; Suzuki, Y.; Kimura, K.; Yoshihama, M.; Ichikawa, T.;
Uramoto, M. J. Antibiot. 1999, 52, 281-287.
(5) Knapp, S. Chem. ReV. 1995, 95, 1859-1876.
(6) Kimura, K.; Miyata, N.; Kawanishi, G.; Kamio, Y.; Izaki, K.; Isono,
K. Agric. Biol. Chem. 1989, 53, 1811-1815.
(7) Brandish, P. E.; Kimura, K.; Inukai, M.; Southgate, R.; Lonsdale, J.
T.; Bugg, T. D. H. Antimicrob. Agents Chemother. 1996, 40, 1640-1644.
(8) Van Heijenoort, J. In Bacterial Cell Wall; Ghuysen, J.-M., Hakenbeck,
R., Eds.; Elsevier: Amsterdam, 1994; Vol. 27, pp 39-54.
Our analysis of the liposidomycin degradation product
structures compatible with the vicinal proton coupling
10.1021/ol0102904 CCC: $22.00 © 2002 American Chemical Society
Published on Web 01/26/2002

Figure 1. C-5′/C-6′ Newman projections of two proposed struc-
tures, 3 and 4, for the liposidomycin nucleoside triol acid, showing
selected NOEs and vicinal coupling constants observed for the
degradation product and the related C-3′′′ hydrate, a nucleoside
1
tetrol acid. H NMR data are taken from refs 3 and 10.
constants3 and NOE’s10 reported by Ubukata and co-workers
is shown in Figure 1. Given the uridine C-4′ S stereochem-
istry, an average H-4′/H-5′ dihedral angle close to 80°
(dictated by the small J), and the proximity of H-3′ and H-5′
(from NOE measurements), one may logically assign C-5′
the S configuration. However, the stereochemistry at C-6′
cannot, in our view, be determined from the data available.
Both structures 3 (with C-6′ as R) and 4 (with C-6′ as S)
would seem to conform to the requirements set by an H-5′/
H-6′ dihedral angle close to 180° and an NOE for H-5′ and
NCH3. We therefore undertook the synthesis of 3 and 4 by
unambiguous routes.
Figure 2.
analysis, indicating that the ring-opening reaction is highly
site selective.
The azido acid 8, which matches the stereochemistry of 3
at C-5′ and C-6′, was coupled to the sarcosine-acrolein
adducts9 9 under the action of the peptide coupling reagent
EEDQ in CH2Cl2.14 A mixture of all four stereoisomers of 9
was used to ensure the formation of at least one stereoisomer
of the dipeptide 10, and in fact, only two isomers of 10 (2:
1, 28%), of the four possible, were isolated. The efficiency
of this coupling could undoubtedly be improved by selecting
the best-matching amine partner 9, but for the purpose at
hand the mixture 10 proved to be a workable intermediate.
Cyclization of 10 to the desired diazepanone 11 was
accomplished under conditions related to previous reductive
amination sequences,9 but with some surprising and helpful
features. Exposure of 10 to ozone converted it to a stable
azido ozonide, which was separately combined with triph-
enylphosphine in THF solution. The presumed imine inter-
mediate was reduced in THF solution by treatment with
excess sodium triacetoxyborohydride.15 A single cyclized
(and, under the basic conditions, eliminated) product (11,
28%) was formed. Astonishingly, the C-4 amine arrived
N-methylated in 11 (-CH3 in box), undoubtedly as the result
of its reaction at some point with the formaldehyde liberated
by reduction of the ozonide derived from 10 (dCH2 in box).
The synthesis of 3 (see Figure 2) began with commercially
available 1,2:5,6-di-O-isopropylidene-R-D-allofuranose, 5.
Protection of O-3 as its TBS ether11 was followed by
selective hydrolysis of the 5,6-O-isopropylidene with aqueous
HOAc and then cleavage of the resulting 5,6-diol with NaIO4
in aqueous THF to afford the C-5 carboxaldehyde. Direct
treatment with methyl (triphenylphosphoranylidene)acetate
in CH2Cl2 solution provided the trans unsaturated ester 6 in
66% overall yield from 5. Reduction of 6 with DIBAL in
THF gave the allylic alcohol (67% + 10% recovered 6), and
then Sharpless epoxidation12 mediated by (-)-diethyl D-
tartrate delivered the 5S,6R epoxy alcohol 7. The primary
hydroxyl of 7 was oxidized with RuCl3 and NaIO4,13 and
then the crude 5S,6S epoxy acid was treated with NaN3 in
refluxing methanol to give the 5S,6R azido acid 8 (84% for
two steps). No other isomer of 8 was detected by NMR
(9) Knapp, S.; Morriello, G. J.; Nandan, S. R.; Emge, T. J.; Doss, G.
A.; Mosley, R. T.; Chen, L. J. Org. Chem. 2001, 66, 5822-5831, and
references therein.
(10) Spada, M. R.; Ubukata, M.; Isono, K. Heterocycles 1992, 34, 1147-
1167.
(11) Ramasamy, K. S.; Bandaru, R.; Averett, D. Synth. Commun. 1999,
29, 2881-2894.
(12) Katsuki, T.; Martin, V. S. Org. React. 1996, 48, 1-299.
(13) Carlsen, P. H. J.; Katsuki, T.; Martin, V. S.; Sharpless, K. B. J.
Org. Chem. 1981, 46, 3936-3938.
(14) Belleau, B.; Malek, G. J. Am. Chem. Soc. 1968, 90, 1651-1652.
Lotz, B. T.; Miller, M. J. J. Org. Chem. 1993, 58, 618-625.
(15) Abdel-Magid, A. F.; Maryanoff, C. A.; Carson, K. G. Tetrahedron
Lett. 1990, 39, 5595-5598.
604
Org. Lett., Vol. 4, No. 4, 2002

Both features (the in situ elimination and the N-methylation)
advanced the cyclization product along the route toward 3,
and neither over-reduction nor dehydration interfered to any
great extent.
Acetonide 11 was converted to its anomeric acetate 12
(72%, 4:1 mixture of C-1 isomers) by careful treatment with
acetic anhydride and methanesulfonic acid (other acids, such
as sulfuric or trifluoroacetic, were ineffective). Under these
conditions, the hydroxyls at C-2 and C-5 were also acety-
lated, and the O-TBS group at C-3 was cleaved and the
hydroxyl there acetylated as well. The resulting tetra-O-
acetate 12 superficially resembles a wide variety of other
anomeric acetates previously used for nucleoside (even
“complex” nucleoside5) synthesis.16 However, application of
the standard conditions [bis(trimethylsilyl)uracil, TMS-OTf,
refluxing acetonitrile, or 1,2-dichloroethane] to 12 gave no
nucleoside product whatsoever. The problem was determined
to be the nucleophilic N-4 (in box), which is positioned just
six atoms away from the anomeric center. Lewis basic atoms
have previously been observed to interfere with O- and
N-glycosylations,17 and the usual solution is to use an excess
of the glycosyl donor to temporarily block the more Lewis
basic site. As this is not an option here, the N-4 amino was
instead pretreated with each of several different acids,
including SnCl4, methanesulfonic acid, BF3, and HCl. Only
as the hydrochloride salt did 12 provide any of the corre-
sponding uracil nucleoside 13 (one isomer, 18%). The
structure of 13 follows from the nearly first order 1H NMR
spectrum and its spectroscopic resemblance to similar
compounds.3,9 Deprotection with 0.5 N LiOH in ethanol
solution and then purification by reverse-phase HPLC
afforded the nucleoside triol acid 3 (64%).
For the synthesis of the 6′S diastereomer 4 (Figure 3), the
C-5 carboxaldehyde derived from 5 as above was chain
extended with the Still-Gennari reagent10,18 to provide the
cis unsaturated ester 14 (60%). The ester was reduced with
DIBAL as before (55%), and the resulting primary alcohol
was subjected to Sharpless epoxidation. Although some of
the desired 5S,6S epoxy acid 15 was obtained in the presence
of (+)-diethyl L-tartrate, the reaction was slow and did not
proceed to completion. A more satisfactory solution was to
epoxidize with m-chloroperoxybenzoic acid (CH2Cl2, 15 h),
which led to the same diastereomer (67%). Oxidation as
before gave the 5S,6R epoxy acid 16 (92%); however, the
reaction of 16 with sodium azide was nonselective, giving
rise to a 3:2 mixture of azido acids (78%). That the major
product is the desired 5S,6S isomer 17 was established by
subsequent transformations. The mixture of azido acids was
coupled with amine 9 as before, leading to a mixture of two
peptide isomers 18 (48%, theoretical 60%) with the same
configuration at C-5′ and C-6′ (matching the major azido
acid isomer) but differing in the C-2′′′ and C-3′′′ (liposido-
mycin numbering) configurations. The minor azido acid
evidently did not couple. Cyclization of 18 to the diaz-
epanone 19 was carried out as follows. A solution in CH2-
Figure 3.
Cl2 was treated with ozone, concentrated, and then catalyti-
cally hydrogenated (10% Pd-C) in ethanol solution in the
presence of HCl. The intermediate amino ozonide (m/z 685)
was reduced with sodium triacetoxyborohydride as for 11
to give a 3:1 mixture of two diazepanone isomers (19,
8-21%) as the only cyclization products isolated. Neither
elimination nor N-methylation occurred in this case, and the
diazepanone ring stereochemistries could be assigned as
2′′′S,3′′′S and 2′′′R,3′′′R, respectively, by comparison to
closely related model compounds (isopropyl replaces C-1′-
C-5′ as the diazepanone substituent) with very similar proton
coupling patterns.9 Conversely, the match of the model
systems with the more elaborate diazepanones 19 justifies
somewhat further the earlier use of the models to assign
relative stereochemistry. The C-2′′′ and C-3′′′ stereocenters
are of no consequence here, however, as both isomers were
readily converted to the same unsaturated diazepanone by
treatment with 1,8-diazabicyclo[5.4.0]undec-7-ene (toluene,
1 h).19 Separate N-methylation (formalin, sodium triacetoxy-
borohydride, CH3CN, HOAc, 60 °C, 12 h) gave 20 (83%
for two steps), the diazepanone epimeric at C-6′ with 11.
The reductive N-alkylation that provides 20 requires more
forcing conditions than that for 11 owing to the formation
of a stable intermediate oxazolidine (m/z 513) involving a
methylene-bridged C-5′/C-6′ erythro amino alcohol. The
situation where the C-6′ amino and C-5′ hydroxy are in close
proximity (small dihedral angle) is seen in the Newman
projection of 4 (Figure 1); O-cyclization of an (N-meth-
(16) Vorbru¨ggen, H.; Ruh-Polenz, C. Org. React. 2000, 55, 1-630.
(17) Knapp, S.; Gore, V. K. J. Org. Chem. 1996, 61, 6744-6747, and
references therein.
(19) Knapp, S.; Nandan, S.; Resnick, L. Tetrahedron Lett. 1992, 33,
5485-5486.
(18) Still, W.; Gennari, C. Tetrahedron Lett. 1983, 24, 4405-4408.
Org. Lett., Vol. 4, No. 4, 2002
605

ylidene)iminium intermediate to an oxazolidine can more
readily occur from a stable conformation when C-6′ has the
S (rather than R) configuration.
liposidomycin degradation product under the same conditions
of measurement confirms their identity down to the peak
shapes. Therefore, the liposidomycins possess the 5′S,6′S
absolute stereochemistry, and given the relative stereochem-
ical determinations made previously,9 C-2′′′ (S) and C-3′′′
(S) of 1 and 2 may also be assigned with confidence. At
least seven research groups have published papers describing
synthetic approaches to liposidomycin components;9 clari-
fication of the synthetic targets should sustain these efforts.
Conversion of 20 to the tetraacetate 21 (1:1 anomeric
mixture, 52%) was carried out as for 12, and then 21 was
pretreated with HCl and subjected to N-glycosylation as
before to give the nucleoside 22 (15-20%). Deprotection
of 22 with ethanolic 0.5 N LiOH gave after HPLC purifica-
tion two nucleoside products: the desired nucleoside triol
acid 4 (24%) and an isomeric cyclic ether 23 (60%), each
as the trifluoroacetate salt. NMR analysis of 23 revealed
reasonable upfield resonances for C-2′′′ and C′3′′′, and the
appearance of an H-2′′′/H-3′′′ coupling (J ) 4.8 Hz). H-6′
is a singlet, reflecting the profound change in the H-5′/H-6′
dihedral angle as the result of the O-5′/C-3′′′ cyclization.
Comparison of the 1H NMR spectra of the nucleoside triol
acids 3 and 4 with the published spectra3 of the liposidomycin
degradation product in D2O solution reveals a close match
with 4 but not 3 (see tabulated data in Supporting Informa-
tion). Furthermore, direct comparison of the spectrum of 4
at 600 MHz with that of an authentic sample of the
Acknowledgment. We thank Merck & Co. and SynChem
Research Inc. for financial support and Prof. Makoto Ubukata
(Toyama Prefectural University) for an authentic sample of
the liposidomycin diazepanone nucleoside.
Supporting Information Available: Experimental details
and spectral characterization for all new compounds and 1H
NMR comparisons for 3 and 4. This material is available
free of charge via the Internet at http://pubs.acs.org.
OL0102904
606
Org. Lett., Vol. 4, No. 4, 2002