Synthesis of glycocinnasperimicin D

Article abstract of DOI:10.1002/chem.200901745

The first total synthesis of amino sugar antibiotic glycocinnasperimicin D (1) has been achieved by a convergent, three-component coupling strategy. The key steps involve the Heck-Mizoroki reaction by using the iodophenyl glycoside 50 and acryl amide 32 to furnish the right core structure of 1, and the construction of the urea glycoside employing the reaction of glycosyl isocyanate 8 with amino sugar 9. Glycosyl isocyanate 8 was prepared by the oxidation of isonitrile 10, which displayed excellent reactivity in the coupling event. Synthetic roadblocks, encountered during this synthetic effort, have led to the development of the α-selective, Lewis acid catalyzed phenyl glycosylation process with 2-amino-hexopyranose and a procedure for acetonide deprotection without affecting the silyl ethers.

Full text of DOI:10.1002/chem.200901745

DOI: 10.1002/chem.200901745
Synthesis of Glycocinnasperimicin D
Taihei Nishiyama,[b] Yoshifumi Kusumoto,[a] Ken Okumura,[a] Kanako Hara,[a]
Shohei Kusaba,[a] Keiko Hirata,[b] Yukihiro Kamiya,[b] Minoru Isobe,[b] Keiji Nakano,[a]
Hiyoshizo Kotsuki,[a] and Yoshiyasu Ichikawa*[a]
Abstract: The first total synthesis of
amino sugar antibiotic glycocinnasperi-
micin D (1) has been achieved by a
convergent, three-component coupling
strategy. The key steps involve the
Heck–Mizoroki reaction by using the
iodophenyl glycoside 50 and acryl
amide 32 to furnish the right core
structure of 1, and the construction of
the urea glycoside employing the reac-
tion of glycosyl isocyanate
8
with
tivity in the coupling event. Synthetic
roadblocks, encountered during this
synthetic effort, have led to the devel-
opment of the a-selective, Lewis acid
catalyzed phenyl glycosylation process
with 2-amino-hexopyranose and a pro-
cedure for acetonide deprotection
without affecting the silyl ethers.
amino sugar 9. Glycosyl isocyanate 8
was prepared by the oxidation of isoni-
trile 10, which displayed excellent reac-
Keywords: amino sugars · antibiot-
ics · glycosylation · Heck reaction ·
isonitriles · sigmatropic rearrange-
ment
Introduction
In 1985, Umezawa and co-workers at the Institute of Micro-
bial Chemistry (Tokyo) in Japan reported the isolation and
characterization of glycocinnasperimicin D (1), an antibiotic
produced by the Nocardia sp. strain MG615-7F6 isolated
from a soil sample collected in Fuchu City.[1] Analysis of its
spectroscopic properties demonstrated that this substance
belongs to the glycocinnaspermidine family of amino sugar
antibiotics that includes LL-BM123b (2)[2] and coumamidine
g1 (3),[3] all of which were isolated and identified by re-
searchers at the Lederle (New York) and Abbott (Illinois)
Laboratories in the United States of America.
The structurally elaborate, nitrogen-rich and highly func-
tionalized glycocinnasperimicin D possesses two unusual
carbohydrates, including a 2-ureido-pentose in its left por-
[a] Y. Kusumoto, K. Okumura, K. Hara, S. Kusaba, Dr. K. Nakano,
Prof. Dr. H. Kotsuki, Prof. Dr. Y. Ichikawa
Faculty of Science, Kochi University, Akebono-cho
Kochi 780-8520 (Japan)
Fax : (+81)88-844-8359
E-mail: ichikawa@kochi-u.ac.jp
[b] T. Nishiyama, K. Hirata, Y. Kamiya, Prof. Dr. M. Isobe
School of Bioagricultural Sciences, Nagoya University
Chikusa, Nagoya 464-8601 (Japan)
tion and a 2-guanidino-4-ureido-6-deoxy-a-d-glucopyranose
with p-cinnamoylspermidine aglycon on the right side. In ad-
dition, these two amino sugars are joined by a unique glyco-
syl urea linkage. The stereochemistry of the urea glycoside
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.200901745.
600
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 600 – 610

FULL PAPER
on the left-hand amino sugar was determined to be a b-link-
age, assignment of which is based on the observation that
the anomeric proton appearing at d=4.84 ppm in the
1H NMR spectrum of 1 (408C, D2O) has a J value of 9.0 Hz.
Although glycocinnasperimicin D is of special interest,
owing to its broad-spectrum activity against gram-negative
organisms and its undetermined mechanism of action, this
antibiotic is no longer available from natural sources as a
result of a mutation of the Nocardia strain.[4] The structural
complexity, potent biological activity, and lack of availability
of glycocinnasperimicin D have combined to stimulate the
interest of synthetic chemists.[5] We now report a full ac-
count of our studies and observations leading to the first
synthesis of glycocinnasperimicin D.[6,7]
Results and Discussion
Synthetic plan: The major synthetic challenge posed by gly-
cocinnasperimicin D is related to the difficulty associated
with selective installation of the glycosyl urea linkage when
a variety of nitrogen-containing functional groups are pres-
ent. To surmount this problem, we have developed a new
approach for the preparation of urea glycosides
(Scheme 1).[8] The key features of this approach include
Scheme 2. Retrosynthetic plan of glycocinnasperimicin D (1). Boc: tert-
butyloxycarbonyl; Bz: benzoyl; Troc: 2,2,2-trichloroethoxycarbonyl.
alkene moiety in allyl amine 11 would deliver the pyranose
ring of 10. The allyl amine 11 was to be constructed from
the chiral, selectively protected butantetraol derivative 16,
prepared from d-tartaric acid (17). Based on the results of
our previous studies exploring stereoselective methods for
allyl amine synthesis,[9] we anticipated that enantioselective
addition of diethylzinc (Et2Zn) to aldehyde 15 would pro-
vide the allyl alcohol 14, which in turn would be converted
to allyl isocyanate 12 by [3.3] sigmatropic rearrangement of
the allyl cyanate 13. Based on this idea, we launched a syn-
thetic venture toward glycocinnasperimicin D.
Scheme 1. A new approach to the synthesis of urea glycosides.
1) generation of a highly reactive glycosyl isocyanate 5 by
the oxidation of glycosyl isonitrile 4 under mild conditions
(catalytic iodine, pyridine N-oxide, acetonitrile, room tem-
perature) and 2) without isolation, reaction of 5 with amino
sugar 6 to furnish the urea glycoside 7 in excellent yield.
Armed with this method, the convergent tactic for the syn-
thesis of glycocinnasperimicin D and the left-hand amino
sugar, outlined a retrosynthetic style in Scheme 2, was
evolved.
In the retrosynthetic route, disconnection of the urea gly-
cosyl linkage in the target 1 generates glycosyl isocyanate 8
and the highly functionalized right-hand amino sugar 9. Iso-
cyanate 8 would arise by oxidation of the isonitrile group in
left-hand amino sugar 10, and oxidative cleavage of the
Studies directed at the preparation of the left-hand amino
sugar 10: The synthesis of the left-hand amino sugar started
with the chiral C-4 synthon 16, which was prepared from d-
tartaric acid (17) in three steps (Scheme 3) by using known
procedures.[10] The robust TBDPS group was selected for
protection of the primary hydroxyl, since we believed it
would survive the conditions used for all of the reactions in
the sequence. Swern oxidation of 16 in THF followed by
Horner–Wadsworth–Emmons olefination in a one-pot oper-
ation afforded ester 19 in 70% yield.[11] DIBAL reduction of
19 gave the allyl alcohol 20, which was subjected to IBX[12]
Chem. Eur. J. 2010, 16, 600 – 610
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
601

Y. Ichikawa et al.
Table 1. Stereoselective construction of a new allyl carbinol stereocenter
by using the enantioselective addition of diethylzinc.
Entry
Scale[a]
[mg]
Catalyst
([mol%])
Yield
[%]
Selectivity[b]
(dr, 14/21)
N
A
B
C
D
E
F
100
600
200
5000
500
200
300
(S)-DPMPM (5)
(S)-DPMPM (5)
(S)-DPMPM (10)
(S)-DPMPM (10)
(ꢀ)-MIB (8)
75–80
62
82
79
90
93:7
94:6
94:6
94:6
96:4
7:93
5:95
(R)-DPMPM (10)
(+)-MIB (8)
77
90
G
1
[a] Scale given by mg substrate. [b] Determined by H NMR spectroscop-
ic analysis of the products after acetylation.
Scheme 3. Preparation of the unsaturated aldehyde 15. DIBAL: diisobu-
tylaluminun hydride; IBX: 1-hydroxy-1,2-benziodoxol-3(1H)one 1-oxide;
NCS: N-chlorosuccinimide; TBDPS: tert-butyldiphenylsilyl; TEMPO:
2,2,6,6-tetramethylpiperidine N-oxyl.
found that a complete reversal of stereospecificity took
place without affecting the yield (77% yield with 7:93 dr in
entry F, and 90% with 5:95 selectivity in entry G). Since the
yields and selectivities in each pair of processes, (entries A/
F) by using DPMPM and (entries E/G) employing MIB, are
comparable, reagent control appears to be dominant and no
matched/mismatched issues exist. In summary, enantioselec-
tive addition of Et2Zn to the a,b-unsaturated aldehydes can
be performed in a highly selective manner at remote posi-
tions in which the effect of chiral centers in the substrates
are negligible.
or TEMPO oxidation[13] to produce the a,b-unsaturated al-
dehyde 15 in good yields, respectively.
Enantioselective addition of Et2Zn to aldehyde 15: Enantio-
selective addition of Et2Zn to the a,b-unsaturated aldehyde
15 was examined in the presence of the catalyst (S)-diphen-
yl(1-methylpyrrolidin-2-yl)methanol (DPMPM), reported by
Soai [Eq. (1)].[14]
The next task in this synthetic sequence involves the prep-
aration of allyl amine 23 through exploitation of an allyl cy-
anate-to-isocyanate rearrangement (Scheme 4).[17] For this
purpose, the allyl alcohol 14 was transformed to the corre-
sponding carbamate 22 by treatment with trichloroacetyl
isocyanate followed by hydrolysis with potassium carbonate
in aqueous methanol. Dehydration of carbamate 22 with tri-
phenylphosphine, carbon tetrabromide, and triethylamine at
The reaction with a 5 mol% loading of (S)-DPMPM and
100 mg of 15 led to the generation of a mixture of allyl alco-
hols in 75–80% yields and with a 93:7 dr (favoring 14; deter-
1
mined by H NMR spectroscopic analysis of the correspond-
ing acetates, Table 1, entry A; dr=diastereromeric ratio).
Disappointingly, the yield of this process decreased consider-
ably (62%, entry B) when 600 mg of the substrate was em-
ployed. An increase of the catalyst loading to 10 mol% in a
reaction carried out on a 200 mg scale enabled attainment
of the small-scale yield (82%, entry C). More importantly,
the use of 10 mol% of the catalyst realized an efficient reac-
tion on a 5 g scale (79% yield with a 94:6 dr, entry D).
The catalyst, 3-exo-morpholinoisoborneol (MIB) devel-
oped by Nugent, was also probed.[15] Although the prepara-
tion of MIB is somewhat more laborious than that of
DPMPA, better results were obtained when 8 mol% loading
of (ꢀ)-MIB was employed for a 500 mg scale reaction (90%
yield, 96:4 dr, entry E).[16] To explore the scope and limita-
tion of the enantioselective addition of Et2Zn for stereose-
lective introduction of an allyl carbinol stereocenter, we car-
ried out reactions with (R)-DPMPM and (+)-MIB, and
Scheme 4. Synthesis of allyl amine 23 by using an allyl cyanate-to-isocya-
nate rearrangement.
602
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 600 – 610

Synthesis of Glycocinnasperimicin D
FULL PAPER
ꢀ408C generated the allyl cyanate 13, which spontaneously
underwent [3.3] sigmatropic rearrangement to afford allyl
isocyanate 12.[18] Due to the pronounced lability of isocya-
nates toward hydrolysis, transformation of 12 was carried
out without using an aqueous workup; the reaction mixture
was treated with a large excess of 2,2,2-trichloroethanol
(12 equiv). Workup and purification then furnihsed the
Troc-carbamate 23 in 76% overall yield from allyl alcohol
14.
Deprotection of the isopropylidene ketal moiety in 23
proved to be unexpectedly difficult (Table 2). Application of
a number of protocols resulted in hydrolysis of both the ace-
tonide and TBDPS groups. Further experimentation found
that Kimꢁs method was appropriate for this task
(entry A).[19] Treatment of 23 with ferric chloride hexahy-
Scheme 5. Transformation of allyl amine 11 into the pyranose 25.
DMAP: 4-dimethylaminopyridine.
predominantly (71%, for three steps). In practice, this reac-
tion sequence could be performed conveniently and effi-
ciently on a 30 mg scale, but on larger scales the yields drop-
ped to less than 50% and the process was accompanied by a
significant amount of formation of a byproduct, the struc-
ture of which was tentatively assigned as the pentose 26,
presumably stemming from 1,2-migration of the acetyl
group.
Table 2. New procedures for unmasking the acetonide group in 23 with-
out affecting the TBDPS silyl ether moiety.
As a result of the problem encountered in this sequence,
an alternative approach was investigated (Scheme 6). We
envisaged that byproduct formation could be avoided by
carrying out ozonolysis prior to silyl deprotection. It was
Entry
Lewis acid
Solvent
Yield [%]
70 (17)[a]
84
86
89
A
B
C
D
FeCl3·6H2O
BCl3·Me2S
BCl3·Me2S
CHCl3
EtSH
HSCH2CH2SH
HSCH2CH2SH
ZnACHTUNGTRENNUNG(OTf)2
[a] Value in parentheses indicates the recovery of starting material.
drate (FeCl3·6H2O) in chloroform followed by immediate
acetylation of the resultant diol, to prevent silyl-group mi-
gration, afforded the requisite acetate 11 in 70% yield
(based on the consumed starting material) along with recov-
ered 23. Further efforts to improve this deprotection step
led to the discovery of unusual reaction conditions involving
Lewis acid catalyzed thiolysis with thiols as the solvent.[20]
For example, combinations of trichloroborane–dimethyl sul-
fide complex with either ethanethiol (entry B) or 1,2-ethane-
dithiol (entry C), and zinc triflate with 1,2-ethanedithiol
(entry D) were found to promote smooth deprotection reac-
tions. As a result, facile reactions leading to complete con-
sumption of 23 on a 500 mg scale were observed to take
place in these cases, affording 11 in good yields (84–89%
after acetylation and purification). Since excess thiol can be
removed during aqueous workup, this procedure appears to
be an excellent method for acetonide deprotection, with the
exception of the disagreeable odors.
Scheme 6. Synthesis of the left-hand amino sugar 10. TMS: trimethylsilyl.
Procedures to transform the allyl amine 11 into the left-
hand amino sugar were investigated next (Scheme 5). Al-
though removal of the TBDPS group in 11 with tetrabutyl-
ammonium fluoride (nBu4NF) buffered with AcOH[21] was
complicated by acetyl-group migration, treatment of 11 with
HF/pyridine in THF[22] cleanly provided a fragile intermedi-
ate alcohol, which was immediately subjected to ozonolysis
(CH2Cl2, ꢀ788C). Acetylation gave the acetyl a-glycoside 25
speculated that rapid hemiacetal formation (28!29) would
block the primary hydroxyl in 28 and prevent both acetyl
1,2-migration and epimerization of the a-amino aldehyde.
Indeed, this change led to a reliable transformation of 11
into 25 on a 1 gram scale in 77% overall yield.
Treatment of the acetyl glycoside 25 with trimethylsilyl
azide in the presence of tin(VI) chloride furnished the b-gly-
Chem. Eur. J. 2010, 16, 600 – 610
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
603

Y. Ichikawa et al.
cosyl azide 30 exclusively (86%).[23] The isonitrile group was
introduced at the anomeric position by using a four-step se-
quence involving the Staudinger reaction of glycosyl azide
30 with triphenylphosphine followed by hydrolysis of the re-
sultant iminophosphorane to give the glycosylamine, which
was immediately treated with acetic formic anhydride to
give the formamide 31 (72% yield). Dehydration of 31 with
triphosgene generated the desired isonitrile, albeit in modest
and variable yields (40–60%). However, implementation of
the more reliable, modified Appelꢁs procedure (PPh3, CBr4,
Et3N, CH2Cl2), furnished the left-hand amino sugar 10 in
85% yield.[24]
the phenyl b-glycopyranosides into the a-isomers (anomeri-
zation) have been described,[28] the yields of the reported
procedures are relatively low (23–32%). In contrast, b-selec-
tive syntheses of phenyl glucosides[29] and galactosides[30] are
well documented. Prompted by these earlier reports, pre-
liminary investigations began with the Lewis acid catalyzed
glycosylation of phenol with penta-O-acetyl-d-galactopyra-
noside 33 to obtain the b-anomer 35 (Scheme 8). During
Synthetic plan for the construction of the right-hand amino
sugar 9: The strategy designed for preparation of the right-
hand amino sugar 9 relies on the Heck–Mizoroki reaction,
which would couple the iodophenyl-6-deoxy-2-amino-a-d-
glycopyranoside A and acrylamide 32 subunits (Scheme 7).
This convergent tactic was selected to alleviate concerns
about potential problems that might be encountered in the
glycosylation reaction of the poorly nucleophilic phenol
with a highly functionalized pyranose.[25] In studies, de-
scribed in detail elsewhere,[26] we demonstrated that the
Heck–Mizoroki reaction of iodophenyl glycopyranoside
with acrylamide 32 proceeded smoothly to introduce the
cinnamoyl structure and sperimidine moiety on the iodo-
phenyl glycopyranoside in one step. The remaining task in
the execution of this plan is the multigram synthesis of
phenyl 2-amino-a-d-glycopyranoside B and its transforma-
tion into subunit A by deoxygenation of the C-6 hydroxy
group and iodination of the phenyl ring. Owing to the fact
that our preliminary experiments showed that catalytic hy-
drogenolysis of 6-iodo-glycopyranose is the most simple and
high-yielding method for multigram synthesis of 6-deoxy-
glycopyranoside, we did not chose a route involving glycosy-
lation of iodophenol followed by deoxygenation of the C-6
hydroxyl. An iodophenyl glycopyranoside would not be
compatible with the conditions employed in the deoxygena-
tion step.[27]
Scheme 8. The unexpected a-selective galactosylation of phenol.
Lewis acid screening studies, a reaction mixture containing
pentaacetyl galactopyranoside 33, phenol, and tin(IV) chlo-
ride in CH2Cl2 was accidentally left at room temperature for
several days and then subjected to workup. Surprisingly, the
predominant product was found to be the a-galactopyrano-
side 34. The significance of this fortuitous experimental
result led us to carry out this process under controlled con-
ditions, and we found that the glycosylation reaction initially
led to a mixture of anomers, which, with a prolonged reac-
tion time, underwent a gradual decrease in the amount of
the b-isomer and a concomitant increase in the a-anomer.
Contrary to our initial plan, phenyl a-d-galactopyranoside
34 was obtained in a modest yield (47%, unoptimized) after
recrystallization.[31]
These unexpected results led us to explore the reaction of
the 2-azido-d-galactose derivative 36[32] with phenol in the
presence of tin(VI) chloride (Scheme 9).[33] This reaction
proceeded smoothly in CH2Cl2 at room temperature over-
night to produce a 4:1 mixture (1H NMR spectroscopic anal-
ysis of the crude product mixture) of phenyl galactopyrano-
sides favoring the a-anomer 37 in 86% yield. Removal of
Synthesis of the phenyl 2-amino-a-d-glycopyranoside B: In
studies performed to find deoxygenation and iodination re-
action conditions, we initially planned to prepare phenyl
acetyl galactopyranoside as model compounds. Although
synthesis of aryl a-glycopyranosides and isomerization of
Scheme 7. Plan for synthesis of the right-hand amino sugar 9.
604
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 600 – 610

Synthesis of Glycocinnasperimicin D
FULL PAPER
more robust methoxycarbonyl-protected derivative 41 was
prepared and then employed in the Lewis acid catalyzed
glycosylation to afford the phenyl a-glycopyranoside 43 with
a/b=10:1 selectivity. Treatment of 43 with triethylamine in
methanol produced the crystalline triol 44, which was easily
purified by recrystallization (80% yield, three steps from
41).
Further efforts established a four-step synthesis of 44
from commercially available and inexpensive d-glucosamine
hydrochloride (45) without the need for chromatographic
purifications (Scheme 11). By using the procedure reported
Scheme 9. a-Selective glycosylation of phenol with 2-azido-d-galactose
derivative 36.
acetyl groups in 37 afforded the highly crystalline triol 38,
and the b-isomer was conveniently removed by recrystalliza-
tion. As a result, multigram quantities of the a-anomer 38, a
synthetic equivalent to phenyl 2-amino-a-d-glycopyranoside
B, were obtained in 48% yield.
At first, we had excluded a plan to synthesize phenyl a-
glycopyranosides by starting from N-alkoxycarbonyl deriva-
tives of glucosamine, because literature precedent clearly in-
dicated that the neighboring effect of the N-alkoxycarbonyl
group results in formation of 1,2-trans-glycosylation prod-
ucts.[34] However, the results uncovered in studying Lewis
acid catalyzed a-selective glycosylation reactions with
phenol guided us to reconsider this plan (Scheme 10). Ac-
Scheme 11. Synthesis of phenyl 2-amino-a-d-glycopyranoside 44 from d-
glucosamine hydrochloride.
by Schmidt,[36] 45 was transformed into a mixture of N-me-
thoxycarbonyl-protected O-acetyl glucosamine 41. Glycosy-
lation of phenol with 41, followed by removal of the acetyl
groups in the product 43 and recrystallization led to produc-
tion of the triol 44 in 71% yield (four steps).
Synthesis of the right-hand amino sugar 9: A synthetic route
for the preparation of subunit A, by starting from galactose
derivative 38, was explored (Scheme 12). Tosylation of the
primary C-6 hydroxyl in 38 was accomplished by using the
conventional method employing TsCl in pyridine. A subse-
quent displacement reaction of the resulting tosylate 46 with
sodium iodide and tetrabutylammonium iodide (nBu4NI) in
refluxing DME provided the iodide 47 in 73% yield (two
steps). After selective protection of the equatorial C-3 hy-
droxy group in 47 as a benzoate (benzoyl chloride, pyri-
dine), reductive hydrogenolysis of the iodide 48 was carried
out with concomitant reduction of the azide group (H2, 5%
Pd/C, Et3N). It should be noted that while the use of AcOEt
or MeOH as solvents gave poor results, the reduction pro-
ceeded smoothly in EtOH. Subsequent protection of the re-
sulting amine as a Troc-carbamate (TrocCl, pyridine) gave
rise to the 6-deoxy-pyranose 49 in 81% yield (two steps). Fi-
nally, iodination of 49 with CAN and nBu4NI in acetonitrile
Scheme 10. 1,2-cis-Glycosylation of phenol with a N-alkoxycarbonyl de-
rivative of glucosamine.
cordingly, the N-trichloroethoxy carbonyl derivative 40 was
prepared from tetraacetyl glucosamine 39[35] by a conven-
tional method (TrocCl, iPr2NEt, CH2Cl2) and then subjected
to the glycosylation reaction conditions. We were pleased to
find that this process was also attended by anomerization of
the initially formed b-isomer into the a-anomer; as a result,
it furnished the phenyl a-d-glycopyranoside 42 predomi-
nantly (1H NMR spectroscopic analysis). Unfortunately, re-
moval of the acetyl groups in 42 was complicated by loss of
the Troc protecting group. To circumvent this problem, the
at 708C furnished the iodophenyl galactoside 50 (86%),[37]
synthetic equivalent of subunit A.
a
A second route for the synthesis of subunit A was devel-
oped (Scheme 13). The initial step in this sequence, which
Chem. Eur. J. 2010, 16, 600 – 610
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
605

Y. Ichikawa et al.
standard conditions (TsCl, pyridine), was sluggish and re-
sulted in low conversion and poor yield, the diamine-cata-
lyzed tosylation method developed by Tanabe was investi-
gated.[38] Although original protocols by using toluene or
acetonitrile as the solvent were problematic due to the poor
solubility of the triol 44, employment of pyridine as a sol-
vent led to the smooth reaction of 44 with TsCl in the pres-
ence of one equivalent of tetramethyl-1,3-diaminopropane
to afford the tosylate 51 in 77% yield. As a consequence of
the lower reactivity of C-4 hydroxy groups in pynanosides
that have the gluco configuration,[39] selective protection of
the C-3 hydroxyl in the diol 51 could be carried out by treat-
ment with pivaloyl chloride in pyridine to afford 52 in 75%
yield.
Deoxygenation of the C-6 hydroxy group was then accom-
plished by using reaction conditions similar to those shown
in Scheme 12 to afford the 6-deoxypyranose 53 in 65% yield
(two steps). Inversion of the C-4 hydroxyl stereochemistry
in 53 was accomplished by using an intramolecular displace-
ment reaction of the triflate with the carbonyl oxygen atom
in the pivalate group at C-3, which involves the dioxoleni-
um-ion intermediate 54.[40] Following this strategy, the tri-
flate was prepared by using the standard method (Tf2O, pyr-
idine, CH2Cl2). Unfortunately, the previously described sol-
volytic conditions, which involve addition of water to the re-
action mixture followed by heating, was complicated by hy-
drolysis of the phenyl glycoside. An exploration of this
process revealed that treatment of triflate with sodium ace-
tate in a mixture of THF and H2O (4:1) at room tempera-
ture effectively promoted formation of the pivaloyl-group
migration product 55 (82%) along with a minor amount of
the C-3 pivalate 56 (ca. 3%). The small J3,4 value of 2.0 Hz
Scheme 12. Synthesis of subunit A from d-galactose derivative 38. CAN:
ceric ammonium nitrate; DME: dimethoxyethane; TsCl: p-toluenesulfon-
yl chloride.
1
observed in the H NMR spectrum of 55 is consistent with
the assignment of axial stereochemistry of the C-4 hydroxyl.
After protection of the hydroxy group in 55 as an acetate,
iodination of the phenyl group as described before
(Scheme 12) provided the iodophenyl glycopyranoside 57 in
90% yield. It should be noted that the yield of this iodina-
tion reaction was considerably lower (46%) when the C-4
hydroxyl unprotected substrate was used. Removal of the
acetyl, pivaloyl, and N-methoxycarbonyl groups in 57 with
barium hydroxide in a mixture of ethanol and H2O (4:1) at
708C gave the amino triol, which was then subjected to
Schotten–Baumann reaction conditions (TrocCl, aq.
NaHCO3/CH2Cl2) to furnish the Troc-carbamate 58 in 60%
yield. Finally, selective protection of the equatorial C-3 hy-
droxyl in 58 as a benzoate (benzoyl chloride, pyridine) then
delivered subunit A (50, 96%), which is identical to the
product synthesized earlier in Scheme 12.
Synthesis of the right-hand amino sugar 9: With sufficient
quantities of iodophenyl galactoside 50 in hand, we em-
barked on a study of the Heck–Mizoroki reaction of 50 with
acryl amide 32 (Scheme 14). The preparation of 32 began
with conjugate addition of 1,4-diaminobutane to acryloni-
trile (60%).[41] Protection of the amine moieties in 60 as
Boc-carbamates followed by reduction of the nitrile 61 with
Scheme 13. A second approach to the synthesis of subunit A (50) from d-
glucosamine derivative 44. Tf: trifluoromethanesulfonyl.
involves selective tosylation of the primary hydroxyl in 44,
was unexpectedly difficult. Since this reaction, when using
606
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 600 – 610

Synthesis of Glycocinnasperimicin D
FULL PAPER
Scheme 14. Synthesis of the right-hand amino sugar 9.
lithium aluminum hydride (LiAlH4) gave rise to the protect-
ed spermidine 62, which was then treated with acryl chloride
under Schotten–Baumann conditions to afford the acryl
amide 32 in 63% yield (three steps). As foreshadowed by
the successful model experiments,[26] the Heck–Mizoroki re-
action enabled the uneventful construction of the right core
structure of glycocinnasperimicin D. Thus, joining iodophen-
yl glycoside 50 and acryl amide 32 in the presence of palla-
Completion of the total synthesis of glycocinnasperimicin D:
With the fully elaborated amino sugars 9 and 10 in hand, we
were well positioned to attempt the critical coupling reac-
tion of the two intermediates (Scheme 15). In the event, oxi-
dation of the isonitrile 10 with pyridine N-oxide in the pres-
ence of a catalytic amount of iodine and MS3A in acetoni-
trile generated the isocyanate 8, which was subsequently
treated with amino sugar 9. The high reactivity associated
with the glycosyl isocyanate 8 was manifested in this crucial
coupling reaction; as a result, the stereochemically congest-
ed 4-amino-pyranose 9 reacted smoothly with isocyanate 8
to furnish the urea glycoside 66 in good yield (85%).
dium catalyst (PdACHTUNGTRENNUNG(OAc)2 (0.19 equiv), PCAHTUGNTRNEN(UGN oTol)3 (0.33 equiv),
Et3N, CH3CN, 708C) led to the production of the cinnamoyl
galactoside 63 in 73% yield. The amine, formed by depro-
tection of the Troc-carbamate in 63 (Zn, AcOH, THF), was
treated with N,N-di-(tert-butoxycarbonyl)-(S)-methyl iso-
thiourea and mercuric chloride
in DMF to provide bis(Boc-
protected) guanidine 64 in 72%
yield (two steps).[42] Finally, the
C-4 amino group was intro-
duced by using a SN2 displace-
ment reaction of the triflate,
prepared from 64 by the stan-
dard method (Tf2O, pyridine,
CH2Cl2) with sodium azide in
DMF to furnish 65 in 73%
yield (two steps). A number of
methods, involving the use of
PPh3, SnCl2, nBu3SnH, or H2/
Pd/C, were applied for the re-
duction of the azide 65, but
none led to formation of the
desired
product.
However,
treatment of 65 with propanedi-
thiol and triethylamine in
MeOH[43] successfully yielded
the desired left-hand amino
sugar 9 in 73% yield.
Scheme 15. Completion of the total synthesis of glycocinnasperimicin D.
Chem. Eur. J. 2010, 16, 600 – 610
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
607

Y. Ichikawa et al.
Installation of the urea group on the left-hand amino
sugar unit in 66 was then carried out by using three opera-
tions involving 1) deprotection of the Troc group in 66 (Zn,
AcOH, THF), 2) treatment of the resulting amine with tri-
chloroacetyl isocyanate (CCl3CONCO, CH2Cl2), and
3) methanolysis (Et3N, MeOH) of the trichloroacetyl urea
67 giving rise to the fully protected glycocinnasperimicin D
68 in a 77% overall yield.
addition, ESI Mass spectrometric analysis of the crude prod-
uct mixture revealed the presence of a peak at 521, which is
not observed in the spectrum of the natural product. This
result indicates that partial hydrolysis of urea glycoside oc-
curred to form product 70, arising from the right-hand
amino sugar.[45] Finally, reverse-phase HPLC analysis of the
crude product mixture showed the presence of four bands,
comprised by a major peak associated with glycocinnasperi-
micin D and three minor peaks attributed to the cis-isomer
69, an unidentified substance, and the hydrolyzed products
corresponding to 70 and its cis-alkene isomer.
All that remained to complete the total synthesis was to
remove the seven protecting groups present in 68. Observa-
tions made in studies of a model compound indicated that
removal of Boc protection followed by hydrolysis of acetyl
and benzoyl groups would be an optimal deprotection pro-
tocol.[44] Accordingly, the Boc group was removed by expo-
sure of 68 to TFA in CH2Cl2. Subsequent cleavage of the
acetyl and benzoyl groups was accomplished by treatment
with aqueous ammonia in methanol at 858C in a sealed tube
overnight. An extended reaction time and an elevated tem-
perature were needed for this process to remove the obsti-
nate benzoyl group. Lyophilization of the reaction mixture
1
furnished a residue that was analyzed by H NMR spectros-
copy. While we were excited to find resonances correspond-
ing to glycocinnasperimicin D in the crude mixture, several
riddle signals were noted in the region of d=5–8 ppm (indi-
cated by arrows in Figure 1). To our surprise, these weak
1
resonaces are also observed in the reported H NMR spec-
trum of the natural product.[2]
Although a detailed understanding of the facility of the
cis/trans isomerization of the cinnamoyl group that took
place during or/after deprotection is not currently in
hand,[46] this obstacle could be circumvented by implementa-
tion of a simple experimental technique. The deprotection
and purification steps were carried out in glass apparatus
wrapped with aluminum foil. After an exhaustive and labor-
intensive task of purification, which led to loss of a consider-
able amount of our precious compound, we were able to iso-
late pure compound 1 as its TFA salt. Careful treatment of
this TFA salt in H2O with Amberlite IRA 410 ion-exchange
resin (Clꢀ form) followed by filtration and lyophilization af-
forded the glycocinnasperimicin D (1) as its hydrochloride
salt (19% yield).[47] The spectroscopic data (1H NMR in
D2O at 408C and 13C NMR spectra), TLC behavior, and an-
timicrobial activity of the synthetic material were in good
agreement with those of naturally occurring glycocinnasperi-
micin D.
Figure 1. 1H NMR spectra of synthetic and natural glycocinnasperimi-
cin D.
Conclusions
The total synthesis of glycocinnasperimicin D has been ac-
complished with a highly convergent, three-component cou-
pling strategy (Scheme 16). The key features in the sequence
involve a Heck–Mizoroki reaction by using the iodophenyl
glycoside 50 and acryl amide 32, and the construction of the
urea glycoside employing glycosyl isocyanate 8 and amino
sugar 9. Oxidation of isonitrile 10 generated the reactive
glycosyl isocyanate 8, which displayed excellent reactivity in
its coupling event. Several synthetic roadblocks, encoun-
Extensive analysis of the crude product mixture unraveled
the following key clues; the puzzling small peaks in the
1H NMR spectra were finally worked out by analysis of the
coupling constants of the weak alkene proton resonances,
which have a J value of 12.5 Hz. This observation suggested
that the photochemically induced olefin isomerization of the
trans-cinnamoyl moiety took place during and/or after re-
moval of the protecting groups to from the cis-isomer 69. In
608
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 600 – 610

Synthesis of Glycocinnasperimicin D
FULL PAPER
1985, 1166–1170; b) K. Dobashi, PhD Thesis, Kyushu University
(Japan), 1990.
[2] a) G. A. Ellestad, D. B. Cosulich, R. W. Broschard, J. H. Martin,
M. P. Kunstmann, G. O. Morton, J. E. Lancaster, W. Fulmor, F. M.
Lovell, J. Am. Chem. Soc. 1978, 100, 2515–2524; b) H. D. Tresner,
J. H. Korshalla, A. A. Fantini, J. D. Korshalla, J. P. Kirby, J. J. Good-
man, R. A. Kele, A. J. Shay, D. B. Borders, J. Antibiot. 1978, 31,
394–397; c) J. H. Martin, M. P. Kunstmann, F. Barbatschi, M. Hertz,
G. A. Ellestad, M. Dann, G. S. Reiden, A. C. Dornbushi, N. A.
Kuck, J. Antibiot. 1978, 31, 398–404; d) S. H. L. Chiu, R. Fiala, R.
Kennett, L. Wozniak, M. W. Bullock, J. Antibiot. 1984, 1000–1006;
e) S. H. L. Chiu, R. Fiala, M. W. Bullock, J. Antibiot. 1984, 37, 1079–
1081; f) H.-R. Tsou, R. R. Fiala, P. C. Mowery, M. W. Bullock, J. An-
tibiot. 1984, 1382–1387. Biological activity; g) M. Greenstein, J. L.
Speth, W. M. Maise, Antimicrob. Agents Chemother. 1981, 20, 425-
432; h) M. S. Osburne, W. M. Maiese, M. Greenstein, Antimicrob.
Agents Chemother. 1990, 34, 1450- 1452.
[3] R. H. Chen, D. N. Whittern, A. M. Buko, J. B. McAlpine, J. Antibiot.
1989, 533–537.
[4] Private communication from Dr. H. Naganawa and Dr. Y. Takaha-
shi.
[5] For synthetic studies of glycocinnamoylspermidine antibiotics, see:
a) K. Araki, K. Miyazawa, J. Yoshimura, Tetrahedron Lett. 1982, 23,
1705–1708; b) H. Hashimoto, K. Araki, H. Hashimoto, J. Yoshi-
mura, Carbohydr. Res. 1982, 99, 59–69; c) K. Araki, H. Hashimoto,
J. Yoshimura, Carbohydr. Res. 1982, 109, 143–160; d) K. Araki, M.
Kawai, K. Miyazawa, H. Hashimoto, J. Yoshimura, Bull. Chem. Soc.
Jpn. 1986, 59, 3137–3143.
[6] This work has been communicated in a preliminary form: T. Nish-
iyama, M. Isobe, Y. Ichikawa, Angew. Chem. 2005, 117, 4446–4449;
Angew. Chem. Int. Ed. 2005, 44, 4372–4375.
[7] Taken in part from: T. Nishiyama, PhD Thesis, Nagoya University
(Japan), 2005.
[8] a) Y. Ichikawa, T. Nishiyama, M. Isobe, Synlett 2000, 1253–1256;
b) Y. Ichikawa, T. Nishiyama, M. Isobe, J. Org. Chem. 2001, 66,
4200–4205.
[9] a) Y. Ichikawa, T. Ito, T. Nishiyama, M. Isobe, Synlett 2003, 1034–
1036; b) Y. Ichikawa, T. Ito, M. Isobe, Chem. Eur. J. 2005, 11, 1949–
1957.
Scheme 16. Summary of the three-component coupling route to glycocin-
nasperimicin D.
[10] For the preparation of 18: E. A. Mash, K. A. Nelson, S. V. Deusen,
S. B. Hemperly, Org. Synth. Coll. Vol. 8 1993, 155–161; for the syn-
thesis of 16: K. Ucihda, K. Kato, H. Akita, Synthesis 1999, 1678–
1686; see also: H. Iida, N. Yamazaki, C. Kibayashi, J. Org. Chem.
1987, 52, 3337–3342.
tered in this synthetic route, have led to the development of
new methodologies including the a-selective, Lewis acid cat-
alyzed phenyl glycosylation process and an interesting pro-
cedure for acetonide deprotection without affecting the silyl
ethers.
[11] R. E. Ireland, W. Norbeck, J. Org. Chem. 1985, 50, 2198–2200.
[12] M. Frigerio, M. Santagostino, S. Sputore, G. Palmisano, J. Org.
Chem. 1995, 60, 7272–7276.
[13] J. Einhorn, C. Einhorn, F. Ratajczak, J.-L. Pierre, J. Org. Chem.
1996, 61, 7452–7454.
[14] K. Soai, A. Ookawa, T. Kaba, K. Ogawa, J. Am. Chem. Soc. 1987,
109, 7111–7115.
[15] W. A. Nugent, Chem. Commun. 1999, 1369–1370.
[16] Recently, a simple three-step synthesis of MIB has appeared: Y. K.
Chen, S.-J. Jeon, P. J. Walsh, W. A. Nugent, Org. Synth. 2005, 82, 87–
92.
Acknowledgements
[17] For representative examples of the allyl cyanate-to-isocyanate rear-
rangement, see: a) C. Christophersen, A. Holm, Acta Chem. Scand.
1970, 24, 1512–1526; b) L. E. Overman, M. Kakimoto, J. Org.
Chem. 1978, 43, 4564–4567; c) K. Banert, Angew. Chem. 1992, 104,
865–867; Angew. Chem. Int. Ed. Engl. 1992, 31, 866–868.
[18] a) Y. Ichikawa, Synlett 1991, 238–240; b) Y. Ichikawa, Synlett 2007,
2927–2936.
We thank Dr. Y. Takahashi (Microbial Chemistry Research Center) for
antimicrobial assays, ESI and 1H NMR spectroscopic measurement of
both synthetic and natural products, cellulose TLC analysis, and valuable
discussions and suggestions at the final stages of the synthetic studies. We
are grateful for financial support in the form of a Grant-in-Aid for Scien-
tific Research (17310129) and Scientific Research on Priority Areas
(18032055 and 18037053) from MEXT (Japan). This work was supported
in part by a Special Research Grant for Green Science from Kochi Uni-
versity and the Kochi University Presidentꢁs Discretionary Grant.
[19] K. S. Kim, Y. H. Song, B. H. Lee, C. S. Hahn, J. Org. Chem. 1986,
51, 404–407.
[20] Deprotection of either benzylidene acetal or acetonide in the pres-
ence of silyl protecting groups with ethanethiol and zinc triflate in
CH2Cl2 has been reported: a) K. C. Nicolaou, M. E. Bunnage, D. G.
McGarry, S. Shi, P. K. Somers, P. A. Wallace, X-J. Chu, K. A. Agrios,
J. L. Gunzner, Z. Yang, Chem. Eur. J. 1999, 5, 599–617; b) K. Fuji-
[1] a) K. Dobashi, K. Nagaoka, Y. Watanabe, M. Nishida, M. Hamada,
H. Naganawa, T. Takita, T. Takeuchi, H. Umezawa, J. Antibiot.
Chem. Eur. J. 2010, 16, 600 – 610
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
609

Y. Ichikawa et al.
wara, S. Yoshimoto, A. Takizawa, S, Souma, H. Mishima, A.
Murai, H. Kawai, T. Suzuki, Tetrahedron Lett. 2005, 46, 6819–6822;
in our case, use of CH2Cl2 as a solvent resulted in slow and incom-
plete deprotection accompanied by recovery of the starting materi-
al.
[21] A. B. Smith III, G. R. Ott, J. Am. Chem. Soc. 1996, 118, 13095–
13096.
[22] D. A. Evans, S. W. Kaldor, T. K. Jones, J. Clardy, T. J. Stout, J. Am.
Chem. Soc. 1990, 112, 7001–7031.
[23] H. Paulsen, Z. Gyorgydeak, M. Friedman, Chem. Ber. 1974, 107,
1568–1578.
[24] Y. Ichikawa, J. Chem. Soc. Perkin Trans. 1 1992, 2135–2139.
[25] K. Oyama, T. Kondo, Synlett 1999, 1627–1629, and references there-
in; see also reference [6c].
[26] Model studies by using iodophenyl glycoside I and acryamide 32
provided II in 75% yield, see: T. Nishiyama, Y. Ichikawa, M. Isobe,
Synlett 2004, 89–92.
[34] a) S. Kusumoto, H. Yoshimura, M. Imoto, T. Shimamoto, T. Shiba,
Tetrahedron Lett. 1985, 26, 909–912; b) P. Boullanger, J. Banoub, G.
Descotes, Can. J. Chem. 1987, 65, 1343–1348.
[35] Tetra-O-acetyl-glucosamine 39 was prepared from glucosamine hy-
drochloride 45 in three steps, see reference [27].
[36] W. Dullenkopf, J. C. Castro-Palomino, L. Manzoni, R. R. Schmidt,
Carbohydr. Res. 1996, 296, 135–147.
[37] T. Sugiyama, Bull. Chem. Soc. Jpn. 1981, 54, 2847–2848.
[38] Y. Yoshida, K. Shimonishi, Y. Sakakura, S. Okada, N. Aso, Y.
Tanabe, Synthesis 1999, 1633–1636.
[39] a) J. M. Williams, A. C. Richardson, Tetrahedron 1967, 23, 1369–
1378; b) O. Varela, D. Cicero, R. M. de Lederkremer, J. Org. Chem.
1989, 54, 1884–1890; c) C. Gallo-Rodriguez, O. Varela, R. M. de Le-
derkremer, J. Org. Chem. 1996, 61, 1886–1889.
[40] a) R. W. Binkley, M. R. Sivik, J. Org. Chem. 1986, 51, 2619–2621;
b) R. W. Binkley, J. Org. Chem. 1991, 56, 3892–3896; c) V. R.
Bouvet, R. N. Ben, J. Org. Chem. 2006, 71, 3619–3622.
[41] a) M. Israel, J. S. Rosenfield, E. J. Modest, J. Med. Chem. 1964, 7,
710–716; b) M. Humora, J. Quick, J. Org. Chem. 1979, 44, 1166–
1168.
[42] Z. X. Guo, A. N. Cammidge, D. Horwell, Synth. Commun. 2000, 30,
2933–2943.
[43] H. Bayley, D. N. Stadring, J. R. Knowles, Tetrahedron Lett. 1978, 19,
3633–3634.
[27] For a recent example of C-6 deoxygenation by using catalytic hydo-
genolysis of 6-idodo-glycopyranose, see: A. Medgyes, E. Farkas, A.
Liptak, V. Pozsgay, Tetrahedron 1997, 53, 4159–4178.
[28] a) W. E. Trevelyan, Carbohydr. Res. 1966, 2, 418–420; b) J. L. Bose,
T. R. Ingle, Chem. Ind. 1967, 1451; c) T. D. Audichya, T. R. Ingle,
J. L. Bose, Indian J. Chem. 1971, 9, 315–317; d) T. D. Audichya,
T. R. Ingle, J. L. Bose, Indian J. Chem. 1973, 11, 704–705; e) K.
Honma, K. Nakazima, T. Uematsu, A. Hamada, Chem. Pharm. Bull.
1976, 24, 394–399.
[29] a) R. U. Lemieux, W. P. Shyluk, Can. J. Chem. 1953, 31, 528–535;
b) E. Smits, J. B. F. N. Engberts, R. M. Kellogg, H. A. van Doren, J.
Chem. Soc. Perkin Trans. 1 1996, 2873–2877.
[44] The model compound shown below was employed to examine the
deprotection conditions.
[30] C. Murakata, T. Ogawa, Carbohydr. Res. 1992, 235, 95–114.
[31] During the preparation of this manuscript, an extensive literature
search uncovered the following reference: T. R. Ingle, J. L. Bose,
Carbohydr. Res. 1970, 12, 459–462. In this report, Bose described
the stannic chloride catalyzed glycosylation reaction of phenol with
penta-O-acetyl-b-d-galactopyranose. In this case, a- and b-anomers
were isolated in 40 and 12% yields, respectively. In reference [28b],
Bose reported the reaction of b-d-galactopyranose pentaacetate
with phenol in the presence of stannic chloride to afford a mixture
of phenyl a- and b-galactopyranoside in a 6:4 ratio.
[32] 2-Azido-d-galactose derivative 36 was prepared from d-galactal by
using a two-step route involving azido nitration of d-galactal fol-
lowed by acetolysis by using the method reported by Lemieux, see:
R. U. Lemieux, R. M. Ratcliffe, Can. J. Chem. 1979, 57, 1244–1251.
[33] Initial efforts to prepare phenyl a-d-galactopyranoside 37 focused
on glycosylation of phenol with 2-azido galactose derivatives I, be-
cause an azide group is known to show a small neighboring effect in
the Konig–Knorr and the imidate glycosylation reactions. Unfortu-
nately, this approach resulted in only moderate selectivity (a/b=
7:3), which is not amenable to a large-scale synthesis. For the pio-
neering work on glycosylation by using 2-azide glycopyranose, see:
H. Paulsen, Angew. Chem. 1982, 94, 184–201; Angew. Chem. Int.
Ed. Engl. 1982, 21, 155–224.
[45] In our exploration to find deprotection conditions with the model
compound shown in reference [44], hydrolysis of urea glycoside was
never observed under similar conditions to those used in Scheme 15.
[46] After completion of the synthetic efforts, we noticed the photolysis
of a mixture of LL-BM123g1 and g2, see: J. J. Hlavka, P. Bitha, J.
Boothe, T. Fields, J. Antibiot. 1978, 31, 477–479.
[47] This yield is calculated on the basis of the trihydrochloride of 1. The
previous value (23%) in reference [6] was estimated from the mono-
hydrochloride.
Received: June 24, 2009
Published online: November 20, 2009
610
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 600 – 610