FULL PAPER
DOI: 10.1002/ejoc.200700333
Synthetic Routes to Three Novel Scaffolds for Potential Glycosidase Inhibitors
Michael Rommel,[a] Alexander Ernst,[b][‡] and Ulrich Koert*[a]
Keywords: Synthesis / Ketalisation / Hydroxypyridine / Cyclopentane / Glycosidase inhibitor
Efficient syntheses of three novel scaffolds for potential β-
glycosidase inhibitors were developed: The first consists of a
2,7-dioxabicyclo[2.2.1]heptane derivative, which was pre-
pared by an intramolecular ketalisation. The second scaffold
consists of a hydroxylated cyclopentylamine, which could be
synthesised stereoselectively from 2-azabicyclo[2.2.1]hept-5-
en-3-one. The third scaffold, a 4,5-dihydroxynicotinic acid,
Introduction
The oligosaccharides of the glycocalix are involved in
many disease-relevant cellular molecular-recognition events.
Glycosidases, which interfere and control the cellular oligo-
saccharide processing are therefore an important class of
targets for pharmaceutical research. For that reason, gly-
cosidase inhibitors present an important substance class for
drug development.[1] Currently, glycosidase inhibitors are
established for the treatment of diabetes[2] and influenza.[3]
Furthermore, their function as antiviral agents is also useful
for the development of potential applications against hepa-
titis,[4] HIV[5] and cancer.[6] Representative examples for β-
glycosidase inhibitors of natural-product and non-natural-
product origin are shown in Figure 1. Isofagomine (1) is an
isomer of the natural product fagomine,[7] while siastatin B
(2), a natural product from Clostridium perfringens, inhibits
sialidase.[8] The isoquinuclidine 3, a mimic of the β-d-man-
nopyranoside 1,4B conformer,[9] as well as the iminosugar
4,[10] are the results of inhibitor design and synthetic efforts.
The search for efficient and selective glycosidase inhibi-
tors challenges transition-state-analogue design[11] and or-
ganic synthesis. Based on the mechanism for a retaining β-
glycosidase, we had proposed a 2,7-dioxabicyclo[2.2.1]hep-
tane derivative as potential inhibitor.[12] Here, we report in
full detail the synthetic route to this novel scaffold, as well
as another two inhibitor frameworks: 4,5-dihydroxynico-
tinic acid derivatives and hydroxylated cyclopentylamines.
In addition, selected compounds were tested as inhibitors
for a number of glycosidases and the results are reported.
[a] Fachbereich Chemie, Philipps-Universität Marburg,
35032 Marburg, Germany
Fax: +49-6421-2825677
E-mail: koert@chemie.uni-marburg.de
[b] Schering AG, Medicinal Chemistry IV,
Müllerstrasse 178, 13342 Berlin, Germany
[‡] Current address: Polyphor AG,
Gewerbestrasse 14, 4123 Allschwil (BL), Switzerland
© 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Eur. J. Org. Chem. 2007, 4408–44304408
was accessible through a sequence of substituent directed
ortho-lithiations. Selected compounds were tested as inhibi-
tors for a number of glycosidases. Three nicotinic acid deriva-
tives were found to be selective β-glucosidase inhibitors.
(© Wiley-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim,
Germany, 2007)
Figure 1. Naturally occurring and synthetic β-glycosidase inhibi-
tors.
Results and Discussion
Design of the Scaffolds
For retaining β-glycosidases the cleavage of the glycosidic
bond proceeds through a boat-like conformer in which the
scissile C–O bond is oriented antiperiplanar to the doubly
occupied non-bonding orbital of the endocyclic oxygen
atom.[13,14] The involvement of a boat-type conformer in
the reaction pathway of retaining β-glycosidases is shown
schematically in Figure 2 for a β-glucuronidase. Guided by
the pioneering work of Vasella and co-workers[9,13d,15] to
use locked boat conformations as potential inhibitors we
devised a novel bicyclic ketal: the dioxabicyclo[2.2.1]hep-
tane derivative 5 should imitate the substrate in a conforma-
tion related to the transition state. The amino group in 5 is
intended to mimic the positive charge of the oxonium ion.
The structure of the naturally occurring β-glucuronidase
inhibitor siastatin B (2) served as a lead structure for the
development of further two types of potential inhibitors
(Figure 3). The change of the piperidine ring to a cyclopen-
tylamine (dashed line in 2 of Figure 3) results in the cyclo-
Three Novel Scaffolds for Potential Glycosidase Inhibitors FULL PAPER
Figure 2. Section of the reaction pathway (A) for a retaining β-
glucuronidase.
pentane 7. If one assumes the imine 6 (or its iminium ion)
as the biologically active form of siastatin B, then the 4,5-
dihydroxynicotinic acid 8 might be another interesting can-
didate for glycosidase inhibition. The last scaffold evades
on purpose stereochemical issues but offers the advantages
of a rigid structure with a possible positive charge at the
pyridine nitrogen atom (after N-protonation or N-alky-
lation).
Figure 3. Siastatin B (2), a naturally occurring glycosidase inhibi-
tor, as a lead structure for the inhibitor candidates 7 and 8.
Synthesis of the Dioxabicyclo[2.2.1]heptane Scaffold
The synthetic plan for the bicyclic amino acid 5 relied on
an intramolecular ketalisation of the open-chain dihydroxy
ketone 9 as the key step (Scheme 1). Compound 9 should
be available from d-galactose (10) by a C1 homologation,
e.g., by a Wittig reaction.
Scheme 1. Retrosynthesis of bicyclic galacturonic acid derivative 5.
According to this plan β-d-galactose pentaacetate[16] (11)
was chosen as starting material (Scheme 2). By using allylic
alcohol and boron trifluoride–diethyl ether[17] compound
11 was transformed into the corresponding anomeric allyl
ether. After methanolysis of the remaining acetates with cat.
Eur. J. Org. Chem. 2007, 4408–4430 © 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.eurjoc.org 4409
NaOMe in MeOH,[18] the 4,6-diol was selectively protected
as benzylidene acetal[19] to yield compound 12. The ben-
zylation of the two hydroxy groups in 2- and 3-position gave
the fully protected galactose building block 13. The subse-
quent selective cleavage of the allyl ether (13� 14) could be
achieved by its base-mediated (KOtBu in DMSO, 100 °C)
rearrangement into the corresponding enol ether and the
cleavage of the latter with HgCl2/HgO in acetone/water.[20]
Scheme 2. a) CH2CHCH2OH, BF3·OEt2, CH2Cl2, 0 °C � room
temp., 24 h; b) NaOMe, MeOH, room temp., 18 h; c) PhCH-
(OMe)2, CSA, MeCN, room temp., 1.5 h, 63% over three steps; d)
NaH, BnBr, THF, room temp., 24 h, 93%; e) i. KOtBu, DMSO,
100 °C, 20 min, ii. yellow HgO, HgCl2, acetone/H2O, 9:1, room
temp., 15 h, 83%; f) [Ph3PMe]+ Br–, nBuLi, THF, –30 °C � room
temp., 24 h, 90%; g) HS(CH2)3SH, CSA, MeOH, room temp., 15 h,
97%; h) TBDPS-Cl, Et3N, DMAP, CH2Cl2, 0 °C � room temp.,
3 d, 83%; i) 1,1-dimethoxycyclopentane, CSA, MeCN, room temp.,
25 min, 93%.
Next, a lactol Wittig reaction[21] (14 � 15) was per-
formed in 90% yield by adding 14 to [Ph3PMe]+Br–/nBuLi
at –30 °C.[22] The subsequent cleavage of the benzylidene
acetal 15 to the triol 16 with CSA in MeOH proceeded
quantitatively only upon addition of 1,3-propanedithiol to
trap the emerging benzaldehyde. The primary hydroxy func-
tion in 16 was protected selectively using TBDPS-Cl and
TEA/DMAP to give the silyl ether 17. The following choice
of the protecting group for the remaining 1,2-diol was cru-
cial for the final ketalisation. Initial attempts with an iso-
propylidene ketal suffered from problems in this key step.
Thus, the more labile cyclopentylidene ketal[23] was chosen
and compound 18 was obtained.
The final reaction sequence dealt with the introduction
of the amino functionality and the construction of the bicy-
clic ketal (Scheme 3). First, the epoxidation of the alkene
18 led to the epoxide 19. The opening of the epoxide with
sodium azide at the less substituted position resulted in the
azido alcohol 20. A Swern oxidation converted the alcohol
20 into the ketone 21. The following bicyclisation required
the selective deprotection of the cyclopentylidene ketal
without cleavage of the primary TBDPS ether. This was
M. Rommel, A. Ernst, U. KoertFULL PAPER
possible under anhydrous reaction conditions at 20 °C in a
mixture of CH2Cl2 and TFA containing powdered molecu-
lar sieves. After chromatography the desired dioxabicy-
clo[2.2.1]heptane derivative 22 was obtained in 89% yield.
The primary TBDPS ether was now cleaved with TBAF
to afford the primary alcohol 23. A subsequent one-step
oxidation using diacetoxyiodobenzene and cat. TEMPO[24]
led to the carboxylic acid 24. The final deprotection of the
benzyl ethers and the simultaneous reduction of the azide
was accomplished by hydrogenation with Pd(OH)2/C. The
crude product was dissolved in MeOH and precipitated on
addition of ethyl acetate to yield the target compound 5 as
a colourless powder. The structure of 5 was confirmed by
X-ray crystal structure analysis.[12]
Scheme 3. a) mCPBA, CH2Cl2, 0 °C � room temp., 60 h, 87%; b)
NaN3, NH4Cl, EtOH, 78 °C, 45 h, 87% (98% based on conver-
sion); c) (COCl)2, DMSO, Et3N, CH2Cl2, –60 °C � room temp.,
1.5 h, 83%; d) TFA/CH2Cl2 (1:1), MS (4 Å), room temp., 30 min,
89 %; e) TBAF, THF, room temp., 1 h, 93 %; f) PhI(OAc)2,
TEMPO, wet CH2Cl2, room temp., 90 min, 87%; g) H2, Pd(OH)2/
C, EtOAc/MeOH, 2:1, room temp., 90 min, quant.
With an efficient route to the new scaffold in hand, a
number of derivatives was prepared (Scheme 4). From 23
the amino alcohol 25 was obtained by reduction of the az-
ide and simultaneous cleavage of the benzyl ethers. The azi-
docarboxylic acid 24 was the starting point for monomeric
(26, 29) and dimeric derivatives (27, 28). Esterification of
24 with ethyleneglycol and subsequent azide reduction and
benzyl ether cleavage yielded the monoester 26 and the dies-
ter 27. Amide formation with ethylenediamine followed by
hydrogenolysis led to the diamide 28, while amide forma-
tion with Z-protected ethylenediamine and subsequent hy-
drogenolysis provided the monoamide 29.
The synthetic routes described above allow an efficient
elaboration of the novel dioxabicyclo[2.2.1]heptane scaffold
in enantiopure form. The variability in the end game of the
synthesis opens the possibility to a wide range of variations
and derivative formation.
www.eurjoc.org © 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Eur. J. Org. Chem. 2007, 4408–44304410
Scheme 4. a) H2, Pd(OH)2/C, EtOAc/MeOH, 2:1, room temp., 3 h,
99%; b) HOCH2CH2OH, EDC·HCl, Et3N, DMAP, CH2Cl2, 0 °C
� room temp., 24 h, monomer: 22% (32% based on conversion),
dimer: 12% (17% based on conversion); c) H2, Pd(OH)2/C, EtOAc/
MeOH (2:1), room temp., 3–18 h, quant. monomer and 92% dimer;
d) H2NCH2CH2NH2, HBTU, HOBt, iPr2NEt, CH2Cl2, room
temp., 2 h, 40% (48% based on conversion); e) H2, Pd(OH)2/C,
EtOAc/MeOH, 2:1, room temp., 18 h, 96%; f) H2NCH2CH2NHZ,
HATU, HOAt, iPr2NEt, CH2Cl2, room temp., 2 h, 78%; g) H2,
Pd(OH)2/C, EtOAc/MeOH, 2:1, room temp., 3.5 h, 95%.
Synthesis of the Cyclopentane Scaffold
The synthetic strategy for the siastatin structure derived
dihydroxycyclopentane amino acid 7 is shown in Scheme 5.
A cis-dihydroxylation of an olefin precursor and the epi-
merisation of the carboxylic substituent leads to the cis-
substituted precursor 30, which should be accessible from
the commercially available enantiopure lactam 31. This ap-
proach should also lead to stereoanalogs of compound 7.
Scheme 5. Retrosynthesis of the dihydroxycyclopentane amino acid
7.
The opening of the bicyclic lactam 31 in HCl/MeOH[25]
gave the corresponding γ-amino methyl ester, whose amino
group was directly Z-protected[26] to 32 (Scheme 6). The cis
configuration of 32 was confirmed by the NOESY spec-
trum, and an epimerisation under the esterification condi-
tions could be excluded.[27] The dihydroxylation of 32 with
K2OsO4/NMO provided two diastereomeric cis-diols in a
1:1 ratio, which were in our hands not separable directly by
silica gel chromatography. In contrast, the corresponding
acetonides 33 and 34, which were obtained from the diol
mixture, were easily separated by flash column chromatog-
raphy.
Three Novel Scaffolds for Potential Glycosidase Inhibitors FULL PAPER
Scheme 6. a) SOCl2, MeOH, 0 °C, 2 h; b) benzyl chloroformate,
NaHCO3, 1,4-dioxane/H2O (4:3), room temp., 2 h, 93% over two
steps; c) K2OsO4, NMO, acetone/H2O (9:1), room temp., 40 h,
87%; d) 2,2-dimethoxypropane, CSA, MeCN, room temp., 30 min,
42% 33 and 49% 34.
NaOMe in MeOH was a suitable base for the epimer-
isation 33� 35. Under these conditions the thermodynam-
ically controlled deprotonation avoided β-elimination of the
alkoxy substituent. Both epimeric esters 33 and 35 could be
separated by chromatography and converted into the de-
sired dihydroxy-γ-amino acids by microwave-assisted aceto-
nide cleavage (33 � 36, 35 � 38), ester hydrolysis and hy-
drogenolytic cleavage of the carbamate (36 � 37, 38 � 7,
Scheme 7).
Scheme 7. a) NaOMe, MeOH, room temp., 100 min, 40% 35 and
45% 33; b) AcOH/H2O (4:1), 100 °C (microwave), 10 min; c)
LiOH·H2O, THF/H2O (3:1), room temp., 18 h, 79% over two steps;
d) AcOH/H2O (4:1), 100 °C (microwave), 10 min; e) LiOH·H2O,
THF/H2O (3:1), room temp., 1 h, 95% over two steps; f) H2,
Pd(OH)2/C, MeOH, room temp., 90 min, 98%; g) H2, Pd(OH)2/C,
MeOH, room temp., 1 h, 90%.
The diastereomeric acetonide ester 34 was epimerised to
39 (Scheme 8). Cleavage of the acetonide (34 � 40, 39 �
42), and subsequent carbamate cleavage afforded the corre-
sponding dihydroxy-γ-amino esters 41 and 43.
As shown in Scheme 9, all attempts to hydrolyse the ester
after cleavage of the acetonide failed for compounds 34 (�
44) and 39 (� 45). The cis-configured Z-protected amino
alcohol led to an undesired oxazolidinone formation in
these cases.
With compounds 33 and 35 the hydrogenolytic cleavage
of the Z group after the acetonide cleavage was possible
and resulted in the formation of the dihydroxy-γ-amino es-
ters 46 and 47 (Scheme 10).
Eur. J. Org. Chem. 2007, 4408–4430 © 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.eurjoc.org 4411
Scheme 8. a) NaOMe, MeOH, room temp., 3 h, 37% 39 and 58%
34; b) AcOH/H2O (1:1), 100 °C (microwave), 30 min, 90%; c)
AcOH/H2O (4:1), 100 °C (microwave), 10 min, 80%; d) H2,
Pd(OH)2/C, MeOH, room temp., 1 h, 74%; e) H2, Pd(OH)2/C,
MeOH, room temp., 1 h, 90%.
Scheme 9. Acetonide cleavage and subsequent saponification did
not give the corresponding carboxylic acids 44 and 45.
Scheme 10. a) i: AcOH/H2O (4:1), 100 °C (microwave), 10 min; ii:
H2, Pd(OH)2/C, MeOH, room temp., 90 min.
To summarise the synthesis of the cyclopentane scaffold,
the choice of the enantiopure lactam 31, combined with a
stereodivergent approach, gave access to a series of dihy-
droxycyclopentane γ-amino acids (7, 37) and esters (41, 43,
46, 47).
Synthesis of the 4,5-Dihydroxynicotinic Acid Scaffold
With no synthetic access to the 4,5-dihydroxynicotinic
acid (8) known, we focused on the preparation of substi-
tuted heteroaromatics using directed ortho-metallation.[28]
The retrosynthetic considerations of 4,5-dihydroxynicotinic
acid (8) led to a route (Scheme 11) where the carboxylic
group would be introduced at last by a carbonylation reac-
tion. A properly O-protected 4-chloropyridin-3-ol 48 might
be a precursor for this step. The latter should be available
by 4-directed ortho-metallation from pyridin-3-ol (49). Vari-
ous derivatives of compound 8 should be accessible accord-
ing to this approach.
M. Rommel, A. Ernst, U. KoertFULL PAPER
Scheme 11. Retrosynthesis of hydroxypyridine 8.
The synthesis of 8 commenced with the MOM protection
of pyridin-3-ol (49). Compared to the literature pro-
cedure,[29] a better yield of 50 was achieved using DMF as
solvent. The ortho-lithiation of 50 took place at –78 °C in
Et2O and after treatment with hexachloroethane, the chlo-
ropyridine 51 was obtained in 87% yield. In order to avoid
a directing effect of the MOM group towards the 2-position
in the next lithiation step, it was exchanged for a benzyl
ether (51 � 52 � 53), which has the additional advantage
of neutral (hydrogenolytic) cleavage conditions at the end
of the synthesis. A selective lithiation of 51 at the 5-position
with LDA at –78 °C was achieved. It should be noticed that
with pyridines a halogen substituent can have a stronger
metal-directing effect than an alkoxy substituent[28a]
(Scheme 12).
Scheme 12. a) NaH, MOM-Cl, DMF, 0 °C, 90 min, 76%; b) tBuLi,
C2Cl6, Et2O, –78 °C � room temp., 2 h, 87%; c) TFA, CH2Cl2,
0 °C� room temp., 14 h, 94%; d) NaH, 15-crown-5, BnBr, DMF,
0 °C, 80 min, 55%; e) LDA, I2, THF, hexanes, –78 °C, 75 min, 78%
(83% based on conversion); f) CO, Et3N, (rac-BINAP)PdCl2,
MeOH, 85 °C, 5 h, 95%; g) AcCl, MeOH, 85 °C, 16 h, 80%; h) H2,
Pd(OH)2/C, TFE, room temp., 90 min, 98%; i) LiOH·H2O, H2O,
75 °C, 7 h, 92%.
Preliminary experiments of quenching the pyridinylli-
thium intermediate from 53 with CO2 (dry ice) gave only
47% yield of the desired carboxylic acid. A higher yield was
achieved with the two-step procedure iodination/carbon-
ylation (53� 54� 55). The use of CO in MeOH and (rac-
BINAP)PdCl2 as catalyst[30] gave the methyl ester 55 in 95%
yield. The conversion of the 4-chloropyridine 55 to the 4-
www.eurjoc.org © 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Eur. J. Org. Chem. 2007, 4408–44304412
hydroxypyridine 56 was best accomplished with HCl/
MeOH. A mechanistic rationale of this reaction consists of
a nucleophilic aromatic substitution of chloride by meth-
oxide and a subsequent SN2 attack of chloride to yield the
hydroxypyridine and methyl chloride which is converted
into dimethyl ether.[31] An X-ray crystal structure analysis
confirmed the constitution of 56. The following deben-
zylation suffered from the low solubility of 56 in most com-
mon solvents. Trifluoroethanol was the only suitable sol-
vent, which allowed a clean hydrogenolysis leading to 57.
Finally, the methyl ester was hydrolysed and the desired di-
hydroxynicotinic acid (8) was obtained.
The integration of the dihydroxynicotinic acid into a di-
saccharide mimic could be achieved by a glycosidic linkage
to a sugar residue (Scheme 13). d-Glucose was chosen as
the sugar moiety. The 4-hydroxypyridine 56 was allowed to
react with 2,3,4,6-tetra-O-acetyl-α-d-glucopyranosyl bro-
mide under modified Koenigs–Knorr conditions[32] to yield
the β-glucopyranoside 58 as the exclusive product. The O-
glycosylation (no N-glycosylation was observed) was con-
firmed at a later stage of the deprotected compound 61: in
the HMBC spectrum the anomeric proton 1�-H showed a
3JCH coupling to C-4 of the pyridine. The deprotection pro-
ceeded in the following three steps. After methanolytic
cleavage of the acetates in 58, the tetraol 59 was obtained.
The hydrogenolysis of the benzyl ether in 59 led to the 3-
hydroxypyridine 60. At last, the methyl ester could be hy-
drolysed to the corresponding carboxylic acid, which was
stable as its lithium carboxylate 61.
Scheme 13. a) 2,3,4,6-Tetra-O-acetyl-α-d-glucopyranosyl bromide,
AgOTf, 2,6-lutidine, MS (4 Å), CH2Cl2, room temp., 20 h, 78%; b)
NaOMe, MeOH, room temp., 15 h, 94%; c) H2, Pd black, EtOAc/
MeOH (4:1), room temp., 150 min, quant.; d) LiOH·H2O, THF/
H2O (3:1), room temp., 18 h, 93%.
For comparison, a dihydroxypyridine derivative lacking
the carboxylic acid of the nicotinic acid was synthesised
(Scheme 14). The 4-chloropyridine 51 was converted into
the 3-methoxypyridine 62, which after MOM cleavage af-
forded the 3-hydroxy-4-methoxypyridine (63).
Scheme 14. a) NaOMe, MeOH, 80 °C, 32 h, 24%; b) TFA, CH2Cl2,
0 °C, 5 h, 87%.
Three Novel Scaffolds for Potential Glycosidase Inhibitors FULL PAPER
To explore variations in the 4-position of the pyridine, a
series of compounds with a 4-methoxy group was prepared
(Scheme 15). Starting with the 4-chloronicotinic acid deriv-
ative 55, a nucleophilic substitution by methoxide gave
compound 64. Hydrogenolytic cleavage of the benzyl ether
resulted in 65. The ester group in 64 was reduced to the