Synthesis of selenium analogues of the naturally occurring glycosidase inhibitor salacinol and their evaluation as glycosidase inhibitors

Article abstract of DOI:10.1021/ja020299g

The syntheses of two selenium analogues (10 and 11) of the naturally occurring sulfonium ion, salacinol (3), are described. Salacinol is one of the active principles in the aqueous extracts of Salacia reticulata that are traditionally used in Sri Lanka and India for the treatment of diabetes. The synthetic strategy relies on the nucleophilic attack of a 2,3,5-tri-O-benzyl-1,4-anhydro-4-seleno-D-arabinitol at the least hindered carbon of benzyl- or benzylidene-protected D- or L-erythritol-1,3-cyclic sulfate. The use of 1,1,1,3,3,3-hexafluoro-2-propanol as a solvent in the coupling reaction proves to be beneficial. Enzyme inhibition assays indicate that 10 is a better inhibitor (K_i = 0.72 mM) of glucoamylase than 3, which has a K_i value of 1.7 mM. In contrast, 11 showed no significant inhibition of glucoamylase. Compounds 10 and 11 showed no significant inhibition of barley-α-amylase or porcine pancreatic-α-amylase.

Full text of DOI:10.1021/ja020299g

Published on Web 06/18/2002
Synthesis of Selenium Analogues of the Naturally Occurring
Glycosidase Inhibitor Salacinol and Their Evaluation as
Glycosidase Inhibitors†
Blair D. Johnston,‡ Ahmad Ghavami,‡ Morten T. Jensen,§ Birte Svensson,§ and
B. Mario Pinto*,‡
Contribution from the Departments of Chemistry, Simon Fraser UniVersity,
Burnaby, B.C., Canada V5A 1S6, and Carlsberg Laboratory, Gamle Carlsberg Vej 10,
DK-2500, Valby, Copenhagen, Denmark
Received February 26, 2002
Abstract: The syntheses of two selenium analogues (10 and 11) of the naturally occurring sulfonium ion,
salacinol (3), are described. Salacinol is one of the active principles in the aqueous extracts of Salacia
reticulata that are traditionally used in Sri Lanka and India for the treatment of diabetes. The synthetic
strategy relies on the nucleophilic attack of a 2,3,5-tri-O-benzyl-1,4-anhydro-4-seleno-D-arabinitol at the
least hindered carbon of benzyl- or benzylidene-protected D- or L-erythritol-1,3-cyclic sulfate. The use of
1,1,1,3,3,3-hexafluoro-2-propanol as a solvent in the coupling reaction proves to be beneficial. Enzyme
inhibition assays indicate that 10 is a better inhibitor (Ki ) 0.72 mM) of glucoamylase than 3, which has a
Ki value of 1.7 mM. In contrast, 11 showed no significant inhibition of glucoamylase. Compounds 10 and
11 showed no significant inhibition of barley-R-amylase or porcine pancreatic-R-amylase.
Chart 1
Introduction
Glycomimetics in which specific oxygen atoms of carbohy-
drates have been replaced by different heteroatoms have been
intensely investigated in the search for novel glycosidase
inhibitors.1 Our efforts in recent years have focused on a novel
class of glycosidase inhibitors containing sulfonium ions as
putative mimics of the oxacarbenium ion intermediates in
glycosidase hydrolysis reactions.2,3 Thus, we have described the
synthesis and conformational analysis of a sulfonium ion
analogue (1) of the naturally occurring glycosidase inhibitor of
the indolizidine alkaloid class, castanospermine (2) (Chart 1).2
Yoshikawa et al.4 also described the isolation of a naturally
occurring glycosidase inhibitor containing a zwitterionic sul-
fonium-sulfate structure, namely salacinol (3), from the plant
Salacia reticulata, which prompted us3 and others5 to synthesize
3 and its stereoisomers, 4 and 5 (Chart 1), and provide
conclusive proof of structure of the natural product. We have
also reported the syntheses of the corresponding nitrogen
congeners 6 and 7 (Chart 2) as potential glycosidase inhibitors,6
and this has been followed by a similar report from another
group.7 We have also reported the syntheses of 1,4-anhydro-
D-xylitol heteroanalogues 8 and 9 (Chart 2) and their evaluation
as glycosidase inhibitors.8 Most recently, Yoshikawa et al.9
reported the inhibitory activity of 3 against several R-glucosi-
dases and also the inhibitory effects of extracts of Salacia
† Presented at the 20th International Symposium on Carbohydrates,
Hamburg, Germany, September 2000; Abstract B-102.
* To whom correspondence should be addressed. Tel.: (604) 291-4327.
Fax: (604) 291-5424. E-mail: bpinto@sfu.ca.
‡ Simon Fraser University.
§ Carlsberg Laboratory.
(1) (a) Sears, P.; Wong C.-H. Angew. Chem., Int. Ed. 1999, 38, 2300-2324.
(b) Yuasa, H.; Hashimoto, H. ReV. Heteroatom Chem. 1999, 19, 35-65.
(c) Driguez, H. Top. Curr. Chem. 1997, 187, 85-116.
(2) Svansson, L.; Johnston, B. D.; Gu, J.-H.; Patrick, B.; Pinto, B. M. J. Am.
Chem. Soc. 2000, 122, 10769-10775.
(6) Ghavami, A.; Johnston, B. D.; Jensen, M. T.; Svensson, B.; Pinto, B. M.
J. Am. Chem. Soc. 2001, 123, 6268-6271.
(3) Ghavami, A.; Johnston, B. D.; Pinto, B. M. J. Org. Chem. 2001, 66, 2312-
2317.
(7) Muraoka, O.; Ying, S.; Yoshikai, K.; Matsuura, Y.; Yamada, E.; Minematsu,
T.; Tanabe, G.; Matsuda, H. Yoshikawa, M. Chem. Pharm. Bull. 2001,
49, 1503-1505.
(4) Yoshikawa, M.; Murakami, T.; Shimada, H.; Matsuda, H.; Yamahara, J.
Tanabe, G.; Muraoka, O. Tetrahedron Lett. 1997, 38, 8367-8370.
Yoshikawa, M.; Murakami, T.; Yashiro, K.; Matsuda, H. Chem. Pharm.
Bull. 1998, 46, 1339-1340.
(8) Ghavami, A.; Johnston, B. D.; Maddess, M. D.; Chinapoo, S. M.; Jensen,
M. T.; Svensson, B.; Pinto, B. M. Can. J. Chem., in press.
(9) Yoshikawa, M.; Morikawa, T.; Matsuda, H.; Tanabe, G.; Muraoka, O.
Bioorg. Med. Chem. 2002, 10, 1547-1554.
(5) Yuasa, H.; Takada, J.; Hashimoto, H. Tetrahedron Lett. 2000, 41, 6615-
6618.
9
10.1021/ja020299g CCC: $22.00 © 2002 American Chemical Society
J. AM. CHEM. SOC. 2002, 124, 8245-8250
8245

A R T I C L E S
Johnston et al.
Chart 2
Scheme 1
Scheme 2
reticulata on serum blood glucose levels in maltose- and
sucrose-loaded rats. Those authors also assigned the “absolute
stereostructure” of 3 (Chart 1) to a zwitterionic compound
comprised of a 1,4-anhydro-4-thio-D-arabinitol moiety and a 1′-
deoxy-D-erythrosyl-3′-sulfate unit.9 However, we point out here
that carbohydrate nomenclature dictates that the latter unit be
designated a 1-deoxy-L-erythrosyl-3′-sulfate unit, as also proven
by our earlier work on the synthesis of the different stereoiso-
mers of salacinol.3
As part of our continuing interest in evaluating the effect of
the heteroatom substitution in the sugar ring on glycosidase
inhibitory activity, we report in the present work the synthesis
of novel analogues of 3 and its diastereomer 5, in which the
sulfur atom has been replaced by the heavier cognate atom,
selenium, to give 10 and 11, respectively (Chart 3). We report
also their evaluation as glycosidase inhibitors of porcine
pancreatic R-amylase (PPA), barley R-amylase (AMY1), and
glucoamylase G2.
Scheme 3
Chart 3
produced the dimesylate 13,10b which was reacted with freshly
prepared Na2Se to yield 14 in 80% yield (Scheme 2).
The 2,4-O-benzylidene-L- and D-erythritol-1,3-cyclic sulfates
15 and 16 were synthesized from L- and D-glucose, respectively,
using our methods reported previously (Scheme 3).3
The selenonium salt 19 was synthesized by alkylation of the
protected seleno-arabinitol 14 with the cyclic sulfate 16 (1.2
equiv) in dry acetone containing K2CO3, in excellent yield (86%)
(Scheme 4), but NMR spectroscopy showed the presence of
two isomers in a ratio of 7:1.
We have also investigated an alternative route to 19 which
avoided the use of expensive L-glucose as a starting material.
Yuasa et al.5 have reported the preparation of the cyclic sulfates
17 and 18 from D-glucose (Scheme 3) and investigated their
utility in the synthesis of salacinol. They concluded that the
more reactive isopropylidene derivative 18 was the reagent of
choice under their conditions and that 17 decomposed at
temperatures of 60-70 °C in DMF. We also prepared 17 from
D-glucose by a similar route (Scheme 3) and have confirmed
that this derivative is a much less reactive alkylating agent than
the corresponding benzylidene compound 15 that we had
employed previously.3 Thus, attempted reactions with 17 for
the preparation of precursors to salacinol, under a variety of
Results and Discussion
Retrosynthetic analysis indicated that salacinol 3 or its
analogues A could be obtained by alkylation of anhydro-alditol
derivatives at the ring heteroatom (Scheme 1).3 This strategy
was chosen in order to provide flexibility for the synthesis of
analogues having different heteroatoms (X) in the ring. We
envisaged that the opening of appropriately protected cyclic
sulfates with selenoether nucleophiles (X ) Se) would likely
proceed as well, or better, with the corresponding thioether
derivatives.
The required seleno-anhydroarabinitol, 2,3,5-tri-O-benzyl-1,4-
anhydro-4-seleno-D-arabinitol (14), was prepared from L-xylose
as shown in Scheme 2. Thus, 2,3,5-tri-O-benzyl-L-xylitol (12)
was synthesized in four steps starting from L-xylose, as described
by Satoh et al.10a for the synthesis of its enantiomer. Mesylation
(10) (a) Satoh, H.; Yoshimura, Y.; Mirua, S.; Machida, H.; Matsuda, A. Bioorg.
Med. Chem. Lett. 1998, 8, 989-992. (b) Fleet, G. W.; Smith, P. W.
Tetrahedron 1986, 42, 5685-5692.
9
8246 J. AM. CHEM. SOC. VOL. 124, NO. 28, 2002

Se Analogues of the Glycosidase Inhibitor Salacinol
A R T I C L E S
Scheme 4
hexafluoro alcohol (pKa ) 9.3). Since cyclic sulfates are often
considered to be synthetic equivalents of epoxides, we tested
HFIP as a reaction medium for selenonium salt formation. The
reaction of the selenoether 14 with the cyclic sulfate 17 in HFIP
was much faster and proceeded essentially to completion in less
than 1 day at 80 °C (Scheme 4). Two more polar products were
formed and were separated by chromatography. The major
product was obtained in 78% yield and proved to be compound
20, isolated as a 3:1 mixture of diasteromers at the stereogenic
selenium center. The major diastereomer in the mixture was
identical to the selenonium salt, trans-20, obtained from the
reaction in acetone. The minor cis-20 diastereomer showed the
expected H-1′/H-5 correlation in the NOESY spectrum. The
other minor product, 21 (4%), was obtained isomerically pure
and was assigned to be the product resulting from attack of the
selenoether at the secondary carbon (C-3′) of the cyclic sulfate.
This mode of attack at the more hindered center was not
observed in the corresponding reaction of the thioether3 and
presumably arises because of the longer C-Se versus C-S
bond, thereby leading to fewer steric interactions during Se-C
bond formation, and also because of the affinity of the softer
Se atom for the softer secondary carbon center. Alternatively,
the acidic nature of HFIP could have resulted in an SN1-type
reaction. The 1H NMR spectrum of 21, compared to that of 20,
showed downfield shifts for the resonances of H-1′ and an
upfield shift of the H-3′ resonance. Similar differences in the
chemical shifts for C-1′ and C-3′ in the 13C NMR spectrum
provided confirmation of the interchange of selenonium and
sulfate centers. Thus, although HFIP resulted in a faster reaction
and a higher overall yield, the selectivity for the desired trans-
20 isomer had decreased.
Hydrogenolysis of the benzyl protecting groups of 20 (trans:
cis ) 3:1) yielded 10 as a mixture of diastereomers. Precipitation
from MeOH yielded amorphous selenosalacinol (trans-10) of
>90% diastereomeric purity at the stereogenic selenium center.
We have chosen to name the selenium analogue of salacinol
1
(trans-10) “blintol”. The H and 13C NMR spectra of blintol
(trans-10) were very similar to those of salacinol (3) except for
slight upfield shifts for resonances corresponding to the nuclei
in close proximity to the selenium center and the magnitudes
of the coupling constants. As previously noted for salacinol
itself,3 these spectra were sensitive to concentration effects due
to solvation of molecular aggregates of charged zwitterionic
species.
different reaction conditions, were unsuccessful. However, the
more nucleophilic selenium derivative 14 did lead, under our
standard conditions, to a low yield of the selenonium salt 20,
which was obtained as a single diastereomer (Scheme 4). The
reaction proceeded very slowly at 85 °C in acetone and was
terminated before complete consumption of the selenoether
because decomposition products were formed after extended
periods. The NOESY spectrum of 20 showed a clear correlation
between H-4 and H-1′, thus indicating the presence of the isomer
with a trans relationship between C-5 and C-1′.
As part of a study of the influence of different solvents in
such reactions, we investigated the effects of the unusual solvent
1,1,1,3,3,3-hexafluoro-2-propanol (HFIP). This solvent has been
shown to promote the nucleophilic ring-opening of epoxides
by amines.11 The beneficial effects were attributed to activation
of epoxides through hydrogen bonding with the relatively acidic
Since we had previously shown that a salacinol analogue
having the side chain derived from D-erythritol instead of
L-erythritol showed some glycosidase inhibitory activity,6 the
corresponding selenium analogue 22 was prepared starting from
the seleno-arabinitol 14 and the cyclic sulfate 15. Reaction of
these partners in acetone under our standard conditions yielded
the selenonium salt 22 (78%) as a mixture of two diasteromers
(5:1) at the stereogenic selenium center. The major component
of the mixture was trans-22, having C-1′ and C-5 in a trans
relationship. Deprotection of 22 was problematic. Essentially
no hydrogenolysis occurred under 1 atm of H2, despite several
additions of more Pd/C catalyst. Eventually the starting material
was isolated and repurified to remove catalyst-poisoning im-
purites. Deprotection by hydrogenolysis was then achieved,
albeit in very low yield (19%), and the selenonium salt 11 was
obtained as a crystalline solid (trans:cis, 8:1). We attribute the
(11) Das, U.; Crousse, B.; Kesavam, V.; Bonnet-Delpon, D.; Be´gue´, J.-P. J.
Org. Chem. 2000, 65, 6749-6751.
9
J. AM. CHEM. SOC. VOL. 124, NO. 28, 2002 8247

A R T I C L E S
Johnston et al.
2,4-Di-O-benzyl-L-erythritol-1,3-O-cyclic Sulfate (17). A solution
of 1,3-di-O-benzyl-D-erythritol17 (10.0 g, 33.1 mmol) in CH2Cl2 and
Et3N (15 mL, 108 mmol) was cooled and stirred in an ice bath. Thionyl
chloride (2.6 mL, 36 mmol) in CH2Cl2 (20 mL) was added dropwise
over 0.5 h. After an additional 5 min, the mixture was poured onto
crushed ice (∼100 g), and the aqueous phase was separated and
extracted with additional CH2Cl2. The combined extracts were washed
with cold water and dried over MgSO4. The solvent was removed to
give a mixture of the two diastereomeric cyclic sulfites as a pale brown
oil. The mixture was passed through a short silica gel column with
hexanes:EtOAc 3:1, without attempting to separate the isomers, and
the mixture of cyclic sulfites (a pale yellow oil, 10.5 g) was immediately
dissolved in 1:1 CH3CN:CCl4 (200 mL). Sodium periodate (10.6 g,
49.6 mmol) and RuCl3‚xH2O (150 mg) were added, and the mixture
was stirred rapidly while H2O (100 mL) was added. After 75 min,
analysis by TLC showed formation of a single, slightly more polar
product. The mixture was poured into a separatory funnel, and the lower
organic phase was separated. The dark red-brown aqueous phase was
extracted with additional CCl4, and the combined extracts were filtered
and concentrated to a dark syrup that was dissolved in EtOAc (200
mL) and filtered to remove black material. The colorless filtrate was
washed with saturated aqueous NaHCO3 and saturated aqueous NaCl
and dried over MgSO4. Solvent removal gave an oil which was purified
by flash chromatography (hexanes:EtOAc, 3:1) to yield 17 as a colorless
syrup (9.66 g, 80%). Storage at -20 °C resulted in slow crystallization.
low yield to loss of the selenonium salt intermediates by
adsorption on the large amounts of Pd/C used. Recrystallization
of the diasteromeric mixture from MeOH gave pure trans-11.
Enzyme Inhibition Assays. Compounds trans-10 and trans-
11 were tested for their inhibition of three glycosidase enzymes,
namely glucoamylase G2,12,13 porcine pancreatic R-amylase
(PPA), and barley R-amylase (AMY1).14 The effects were
compared to those of salacinol (3). Glucoamylase G2 was
weakly inhibited by 3 (Ki ) 1.7 mM), whereas trans-10 was a
better inhibitor of this enzyme, with a Ki value of 0.72 mM. In
contrast, trans-11 showed no significant inhibition of glucoamy-
lase. Salacinol inhibited AMY1 and PPA, with Ki values of 15
( 1 and 10 ( 2 µM, respectively. Surprisingly, trans-10 and
trans-11 showed no significant inhibition of either AMY1 or
PPA. It would appear, then, that analogues trans-10, trans-11,
and 3 show discrimination for certain glycosidase enzymes and
could be promising candidates for selective inhibition of a wider
panel of enzymes that includes human small intestinal maltase-
glucoamylase15 and human pancreatic R-amylase.16 Such studies
to map the enzyme selectivity profiles of these compounds are
planned.
Experimental Section
A sample was recrystallized from Et2O/hexanes. Mp: 63-65 °C. [R]20
:
D
General. Optical rotations were measured at 23 °C. 1H and 13C NMR
spectra were recorded at 400.13 and 100.6 MHz, respectively, unless
otherwise stated. All assignments were confirmed with the aid of two-
-9.4° (c 1.1, CHCl3). 1H NMR (CDCl3): δ 7.40-7.20 (m, 10H, Ph),
4.76 (ddd, 1H, J2,3 ) 9.4, J3,4a ) 3.3, J3,4b ) 2.1 Hz, H-3), 4.64 and
4.53 (2d, each 1H, JA,B ) 11.9 Hz, CH2Ph), 4.59 and 4.50 (2d, each
1H, JA,B ) 11.6 Hz, CH2Ph), 4.44 (dd, 1H, J1ax,1eq ) 11.0, J1ax,2 ) 10.1
Hz, H-1ax), 4.33 (dd, 1H, J1eq,2 ) 5.2 Hz, H-1eq), 4.17 (ddd, 1H, H-2),
3.88 (dd, 1H, J4a,4b ) 12.0 Hz, H-4a), 3.76 (dd, 1H, H-4b). 13C NMR
(CDCl3): δ 137.19 and 136.53, (2 × Cipso, 2 × Ph), 128.75, 128.65,
128.56, 128.14, 128.07 and 127.90 (each 2C, Cortho, Cmeta, Cpara, 2 ×
Ph), 84.99 (C-3), 73.69 and 73.64 (2 × CH2Ph), 71.59 (C-2), 66.92
and 66.50 (C-1, C-4). MALDI MS: m/e 387.2 (M+ + Na), 403.1 (M+
+ K). Anal. Calcd for C18H20O6S: C, 59.33; H, 5.53. Found: C, 59.38;
H, 5.52.
1
1
1
dimensional H, H (COSYDFTP) or H, 13C (INVBTP) experiments
using standard Bruker pulse programs. Column chromatography was
performed with Merck silica gel 60 (230-400 mesh). MALDI-TOF
mass spectra were measured on a PerSeptive Biosystems Voyager-DE
spectrometer, using 2,5-dihydroxybenzoic acid as a matrix.
1,4-Anhydro-2,3,5-tri-O-benzyl-4-seleno-D-arabinitol (14). Sele-
nium metal (1.1 g, 14 mmol) was added to liquid NH3 (60 mL) in a
-50 °C bath, and small pieces of sodium (0.71 g) were added until a
blue color appeared. A small portion of selenium (20 mg) was added
to remove the blue color. NH3 was removed by warming on a water
bath, and DMF was added and removed under high vacuum to remove
the rest of the NH3. A solution of the dimesylate (13) (7.4 g, 12.7 mmol)
in DMF (100 mL) was added, and the mixture was stirred under N2 in
a 70 °C bath for 3 h. The mixture was cooled, and the solvent was
removed under high vacuum. The product was partitioned between CH2-
Cl2 (150 mL) and water (50 mL), and the organic solution was washed
with water (50 mL) and brine (50 mL) and dried (MgSO4). The solvent
was removed, and the product was purified by flash chromatography
(hexanes:EtOAc, 3:1) to give 14 as a pale yellow oil (4.74 g, 80%).
[R]D: +22° (c 1.3, CHCl3). 1H NMR (CDCl3): δ 7.48-7.22 (15H, m,
Ar), 4.67, 4.61 (2H, 2d, JA,B ) 11.8 Hz, CH2Ph), 4.56, 4.48 (2H, 2d,
JA,B ) 12.1 Hz, CH2Ph), 4.53, 4.50 (2H, 2d, CH2Ph), 4.22 (1H, ddd,
2,3,5-Tri-O-benzyl-1,4-dideoxy-1,4-[[(2S,3S)-2,4-benzylidenedioxy-
3-(sulfoxy)butyl]episelenoniumylidene]-D-arabinitol Inner Salt (19).
The selenoether 14 (0.20 g, 0.43 mmol) was reacted with the cyclic
sulfate 15 (0.14 g, 1.2 equiv) in acetone (0.7 mL) by the procedure
used previously for the corresponding sulfide.3 Column chromatography
(CHCl3:MeOH, 15:1) of the crude product gave an amorphous solid
(0.27 g, 86%). NMR showed the presence of two isomers (7:1) at the
stereogenic selenium center which were separable on analytical HPLC
(acetonitrile/H2O).
1
Data for the major diastereomer trans-19 follow. H NMR (CD2-
Cl2): δ 7.50-7.10 (20H, m, Ar), 5.55 (1H, s, CHPh), 4.58 (1H, m,
H-2), 4.56-4.45 (5H, m, H-3′, 3 CH2Ph, H-4′eq), 4.38 (1H, dd, J1′a,2′
) 2.2 Hz, H-1′a), 4.38-4.34 (2H, m, H-3, CH2Ph), 4.34 and 4.26 (2H,
2d, JA,B ) 12.1 Hz, CH2Ph), 4.25 (1H, m, H-2′), 4.14-4.08 (2H, m,
H-1a, H-4), 3.97 (1H, dd, J1′a,1′b ) 12.1, J1′b, 2′ ) 3.3 Hz, H-1′b), 3.80
(1H, dd, J4′ax,4′eq ) J3′,4′ax ) 10.1 Hz, H-4′ax), 3.65-3.52 (3H, m, H-1b,
H-5a, H-5b). 13C NMR (CD2Cl2): δ 137.28-126.68 (24CAr), 102.10
(CHPh), 84.55 (C-3), 83.36 (C-2), 77.18 (C-2′), 73.70, 72.81, 72.36 (3
× CH2Ph), 69.57 (C-4′), 67.76 (C-3′), 67.02 (C-5), 66.30 (C-4), 48.77
(C-1′), 46.43 (C-1). Anal. Calcd for C37H40O9SSe: C, 59.99; H, 5.45.
Found: C, 59.91; H, 5.44.
H-2), 4.07 (1H, dd, J2,3 ) J3,4 ) 4.6 Hz, H-3), 3.85 (1H, dd, J5a,5b
)
9.2, J4,5a ) 7.6 Hz, H-5a), 3.77 (1H, ddd, J4,5b ) 6.9 Hz, H-4), 3.53
(1H, dd, H-5b), 3.11 (1H, dd, J1a,1b ) 10.4, J1a,2 ) 5.1 Hz, H-1a), 2.96
(1H, dd, J1b,2 ) 5.3 Hz H-1b). 13C NMR (CDCl3): δ 138.24, 138.21,
138.06 (3Cipso), 128.40-127.60 (15C, Ph), 85.93 (C-2), 85.63 (C-3),
72.96 (2C, C-5, CH2Ph), 72.14, 71.50 (2 × CH2Ph), 42.59 (C-4), 23.96
(C-1). MALDI MS: m/e 491.2 (M+ + Na). Anal. Calcd for
C26H28O3S: C, 74.25; H, 6.71. Found: C, 74.18; H, 6.53.
(12) Svensson, B.; Pedersen, T.; Svendsen, I.; Sakai, T.; Ottesen, M. Carlsberg
Res. Commun. 1982, 47, 55-69.
2,3,5-Tri-O-benzyl-1,4-dideoxy-1,4-[[(2S,3S)-2,4-di(benzyloxy)-3-
(sulfoxy)butyl]episelenoniumylidene]-D-arabinitol Inner Salt (20).
(a) By Reaction in (CH3)2CO. The selenoether 14 (117 mg, 0.250
mmol), cyclic sulfate 17 (84 mg, 0.23 mmol), and K2CO3 (80 mg, 0.58
(13) Stoffer, B.; Frandsen, T. P.; Busk, P. K.; Schneider, P.; Svendsen, I.;
Svensson, B. Biochem J. 1993, 292, 197-202.
(14) Juge, N.; Andersen, J. S.; Tull, D.; Roepstorff, P.; Svensson, B. Protein
Expression Purif. 1996, 8, 204-214.
(15) Nichols, B. L.; Eldering, J.; Avery, S.; Hahn, D.; Quaroni, A.; Sterchi, E.
J. Biol. Chem. 1998, 273, 3076-3081.
(16) Braun, C.; Brayer, G. D.; Withers, S. G. J. Biol. Chem. 1995, 270, 26778-
26781.
(17) Jiricek, R.; Lehmann, J.; Rob, B.; Scheuring, M. Carbohydr. Res. 1993,
250, 31-43.
9
8248 J. AM. CHEM. SOC. VOL. 124, NO. 28, 2002

Se Analogues of the Glycosidase Inhibitor Salacinol
A R T I C L E S
1
mmol) were added to anhydrous acetone (3.0 mL), and the mixture
was stirred in a sealed tube with heating at 70 °C for 20 h and then at
85 °C for 48 h. Periodic analysis by TLC (CHCl3:MeOH, 10:1) showed
that the reaction was proceeding very slowly and that substantial
fractions of the cyclic sulfate 17 and the selenoether 14 remained
unreacted. Slow decomposition of the cyclic sulfate to produce polar
impurities was noted at the higher temperature. At the end of 68 h, the
extent of reaction to yield the desired product was estimated to be
<30%, but decomposition products were becoming significant, and thus,
the reaction was terminated at this point. The mixture was cooled and
filtered through Celite with the aid of CH2Cl2. The solvents were
removed, and the residue was purified by column chromatography
(gradient of CHCl3 to CHCl3:MeOH, 10:1). The selenonium salt 20
(39 mg, 20%) was obtained as a colorless syrup. Compound 20 was
isomerically pure and was assigned to be the isomer with a trans
relationship between C-5 and C-1′ by analysis of the NOESY spectrum.
Data for 21 follow. [R]20D: -68° (c 2.2, CHCl3). H NMR (600
MHz, CDCl3): δ 7.38-7.00 (25H, m, Ph), 4.76 (1H, dt, J2′,3′ ) 7.0
Hz, H-3′), 4.71 and 4.42 (2H, 2d, JA,B ) 10.9 Hz, CH2Ph), 4.68 and
4.51 (2H, 2d, JA,B ) 12.4 Hz, CH2Ph), 4.50 and 4.36 (2H, 2d, JA,B
)
11.6 Hz, CH2Ph), 4.50 (1H, ddd, H-4), 4.38 (1H, dd, J1′a,1′b ) 12.6,
J1′a,2′ ) 5.1 Hz, H-1′a), 4.36-4.32 (2H, m, H-2, H-3), 4.32 and 4.14
(2H, 2d, JA,B ) 11.7 Hz, CH2Ph), 4.29 (1H, dd, J1′b,2′ ) 3.1 Hz, H-1′b),
4.28 and 4.21 (2H, 2d, JA,B ) 11.8 Hz, CH2Ph), 4.13 (1H, ddd, H-2′),
4.10 (1H, dd, J4′a,4′b ) 11.8, J3′,4′a ) 3.8 Hz, H-4′a), 3.84 (1H, dd, J3′,4′b
) 3.8 Hz, H-4′b), 3.48 (1H, dd, J5a,5b ) 10.0, J4,5a ) 9.1 Hz, H-5a),
3.44 (1H, dd, J1a,1b ) 12.0 Hz, H-1a), 3.42 (1H, dd, J4,5b ) 9.9 Hz,
H-5b), 3.14 (1H, dd, J1b,2 ) 2.2 Hz, H-1b). 13C NMR (100 MHz,
CDCl3): δ 137.15, 136.63, 136.57, 136.34 and 136.11 (5 × Cipso, Ph),
128.89-127.50 (25C, Ph), 83.43 (C-3), 81.12 (C-2), 75.35 (C-2′), 73.55
(C-1′), 73.06, 72.40, 71.85 and 71.65 (4 × CH2Ph), 66.31 (C-5), 65.90
(C-4′), 63.79 (CH2Ph), 62.65 (C-4), 61.73 (C-3′), 39.77 (C-1). MALDI
MS: m/e 833.8 (M+ + H), 753.8 (M+ + H - SO3). Anal. Calcd for
C44H48O9SSe: C, 63.53; H, 5.82. Found: C, 63.79; H, 5.83.
Data for trans-20 follow. [R]D: -15° (c 1.0, CHCl3). 1H NMR
(CDCl3): δ 7.38-7.05 (25H, m, Ph), 4.65 and 4.46 (2H, 2d, JA,B
)
12.0 Hz, CH2Ph), 4.64 (1H, dt, J2′,3′ ) 6.9 Hz, H-3′), 4.63 and 4.47
(2H, 2d, JA,B ) 11.6 Hz, CH2Ph), 4.48 and 4.33 (2H, 2d, JA,B ) 11.8
Hz, CH2Ph), 4.46 (1H, dd, H-2), 4.43 and 4.38 (2H, 2d, JA,B ) 11.2
Hz, CH2Ph), 4.38 and 4.28 (2H, 2d, JA,B ) 12.0 Hz, CH2Ph), 4.28
(1H, ddd, H-2′), 4.24 (1H, br d, J2,3 ) 1.6 Hz, H-3), 4.22 (1H, br d,
H-1a), 4.17 (1H, dd, J1′a,1′b ) 12.1, J1′a,2′ ) 1.1 Hz, H-1′a), 4.04 (1H,
dd, J4′a,4′b ) 11.1, J3′,4′a ) 2.9 Hz, H-4′a), 3.87 (1H, dd, J3′,4′b ) 2.5
Hz, H-4′b), 4.50 (1H, br t, H-4), 3.67 (1H, dd, J1′b,2′ ) 4.3 Hz, H-1′b),
3.60 (1H, dd, J1a,1b ) 12.8, J1b,2 ) 3.0 Hz, H-1b), 3.57 (1H, dd, J5a,5b
2,3,5-Tri-O-benzyl-1,4-dideoxy-1,4-[[(2R,3R)-2,4-benzylidenedioxy-
3-(sulfoxy)butyl]episelenoniumylidene]-D-arabinitol Inner Salt (22).
The selenoether 14 (760 mg, 1.63 mmol), cyclic sulfate 163 (467 mg,
1.72 mmol), and K2CO3 (102 mg, 0.74 mmol) were added to anhydrous
acetone (3.0 mL), and the mixture was stirred in a sealed tube with
heating at 80 °C for 13 h. Analysis by TLC (CHCl3:MeOH, 10:1)
showed that the cyclic sulfate 16 had been consumed but that there
was a substantial amount of the selenoether 14 remaining. Another
portion of the cyclic sulfate (180 mg, 0.66 mmol) was therefore added,
and the reaction was continued at 80 °C for a further 12 h. After cooling
to room temperature, the mixture was diluted with CH2Cl2 and
processed and purified as described for 19. Compound 22 was obtained
as a colorless foam (0.939 g, 78%). Analysis by NMR showed that the
product was a mixture of two isomers (∼5:1) at the stereogenic selenium
center. The major component of the mixture was assigned to be the
diastereomer with a trans relationship between C-5 and C-1′ on the
basis of analysis of the NOESY spectrum.
) 10.1, J4,5a ) 7.1 Hz, H-5a), 3.54 (1H, dd, J4,5b ) 9.0 Hz, H-5b). 13
NMR (CDCl3): δ 137.07, 137.00, 136.87, 136.15 and 135.88 (5 ×
ipso, Ph), 128.81-127.60 (25C, Ph), 83.83 (C-3), 81.16 (C-2), 74.99
C
C
(C-3′), 73.79 and 73.40 (2 × CH2Ph), 75.18 (C-2′), 72.85, 72.01, and
71.59 (3 × CH2Ph), 69.14 (C-4′), 67.13 (C-5), 64.83 (C-4), 50.08 (C-
1′), 46.34 (C-1). MALDI MS: m/e 833.8 (M+ + H), 753.7 (M+ + H
- SO3). Anal. Calcd for C44H48O9SSe: C, 63.53; H, 5.82. Found: C,
63.39; H, 5.86.
1
(b) By Reaction in (CF3)2CHOH. The selenoether 14 (633 mg,
1.35 mmol), cyclic sulfate 17 (531 mg, 1.46 mmol), and K2CO3 (73
mg, 0.53 mmol) were added to 1,1,1,3,3,3-hexafluoro-2-propanol (2.0
mL), and the mixture was stirred in a Caries tube while being warmed
slowly. The K2CO3 dissolved with evolution of gas between 60 and
80 °C. The tube was cooled to room temperature, opened to relieve
pressure, and then resealed and kept at 80 °C for 22 h. Analysis by
TLC (CHCl3:MeOH, 10:1) showed virtually complete consumption of
the cyclic sulfate 17, although some of the selenoether 14 remained
unreacted. Two products of increased polarity had been formed (major,
rf 0.40; minor, rf 0.35). The mixture was cooled and filtered through
Celite with the aid of CH2Cl2. The solvents were removed, and the
residue was purified by column chromatography (two successive silica
gel columns: first with a gradient of CHCl3 to CHCl3:MeOH, 10:1, to
separate the starting materials, and then with EtOAc:MeOH, 25:1, to
separate the two products). The selenonium salts 20 (827 mg, 78%)
and 21 (45 mg, 4%) were obtained as syrupy oils. Compound 20 proved
to be a 3:1 mixture of isomers at the stereogenic selenium center. The
major component (trans-20) was identical to the compound obtained
from the reaction in acetone (that is, the trans C-5, C-1′ isomer), while
the minor component was assigned to be the corresponding cis-20
isomer. Partial 13C NMR data for the cis isomer were obtained from a
spectrum of the mixture, and assignments were made by analogy to
those of the trans isomer. Compound 21 was isomerically pure and
was assigned to be the isomer with a trans relationship between C-5
and C-3′ by analysis of the NOESY spectrum.
Data for the major diastereomer (trans-22) follow. H NMR (CD2-
Cl2): δ 7.50-7.10 (25H, m, Ph), 5.54 (1H, s, CHPh), 4.58 and 4.50
(2H, 2d, JA,B ) 12.0 Hz, CH2Ph), 4.55 (1H, dd, J4′ax,4′eq ) 10.6, J3′,4′eq
) 5.1 Hz, H-4′eq), 4.50 (1H, dd, H-2), 4.45 (1H, br d, J2,3 ) 2.6 Hz,
H-3), 4.44 (1H, ddd, H-3′), 4.40 and 4.35 (2H, 2d, JA,B ) 11.7 Hz,
CH2Ph), 4.35 and 4.23 (2H, 2d, JA,B ) 11.9 Hz, CH2Ph), 4.34 (1H,
ddd, H-2′), 4.33 (1H, br t, H-4), 4.16 (2H, br d, J1′,2′ ) 4.9 Hz, H-1′a,
H-1′b), 3.91 (1H, dd, J1a,2 ) 1.5, J1a,1b ) 12.1 Hz, H-1a), 3.78 (1H, dd,
J3′,4′ax ) 9.8 Hz, H-4′ax), 3.67-3.59 (2H, m, H-5a, H-5b), 3.56 (1H,
dd, J1b,2 ) 3.4 Hz, H-1b). 13C NMR (CD2Cl2): δ 137.66, 137.31, 136.72
and 136.49 (4 × Cipso, Ph), 129.73-126.66 (25C, Ph), 102.04 (CHPh),
84.27 (C-2), 83.04 (C-3), 77.04 (C-2′), 73.60, 72.51 and 72.14 (3 ×
CH2Ph), 69.77 (C-4′), 68.82 (C-3′), 67.05 (C-5), 64.81 (C-4), 48.19
(C-1′), 46.35 (C-1). MALDI MS: m/e 741.6 (M+ + H), 661.5 (M+
H - SO3). Anal. Calcd for C37H40O9SSe: C, 60.08; H, 5.45. Found:
+
C, 59.91; H, 5.45.
1,4-Dideoxy-1,4-[[(2S,3S)-2,4-dihydroxy-3-(sulfoxy)butyl]epi-
selenoniumylidene]-D-arabinitol Inner Salt (10). To a solution of
selenonium salt 20 (744 mg, 0.894 mmol, 3:1 mixture of isomers) in
HOAc (10 mL) was added 10% Pd/C catalyst (200 mg), and the mixture
was stirred under an atmosphere of H2 for 16 h. More Pd/C (200 mg)
was added, and the hydrogenolysis was continued for an additional 24
h. The mixture was filtered through Celite with MeOH (80 mL) and
concentrated to give a syrup. Purification by column chromatography
(EtOAc:MeOH:H2O, 6:3:1) gave 10 as a colorless gum (225 mg, 66%).
Analysis by 1H NMR indicated a mixture of isomers (5:1). The product
was dissolved in a minimum amount of warm MeOH and cooled slowly
to deposit an amorphous solid (112 mg). This proved to be >90% pure
10 which was assigned, by analysis of the NOESY spectrum, to be the
major trans-10 isomer corresponding to the configuration of salacinol.
Data for the cis-20 isomer follow. 13C NMR (CDCl3): 84.31 (C-3),
82.78 (C-2), 75.42 (C-3′), 73.84 and 73.52, (2 × CH2Ph), 73.18 (C-
2′), 72.86 (CH2Ph), 71.72 (2 × CH2Ph), 68.78(C-4′), 65.46 (C-5), 58.33
(C-4), 42.71 (C-1′), 40.16 (C-1).
9
J. AM. CHEM. SOC. VOL. 124, NO. 28, 2002 8249

A R T I C L E S
Johnston et al.
Data for the diastereomer trans-10 follow. [R]20D: +20° (c 0.5, D2O).
1H NMR (600 MHz, D2O): δ 4.84 (1H, ddd, H-2), 4.53 (1H, dd, J2,3
) 3.5 Hz, H-3), 4.43 (1H, ddd, J2′,3′ ) 7.0 Hz, H-2′), 4.37 (1H, ddd,
H-3′), 4.22 (1H, ddd, J3,4 ) 3.2 Hz, H-4), 4.12 (1H, dd, J4,5a ) 5.1,
J5a,5b ) 12.5 Hz, H-5a), 4.04 (1H, dd, J1′a,2′ ) 3.7, J1′a,1′b ) 12.4 Hz,
H-1′a), 3.98 (1H, dd, J4,5b ) 8.9 Hz, H-5b), 3.97 (1H, dd, J3′,4′a ) 3.6,
J4′a,4′b ) 12.8 Hz, H-4′a), 3.90 (1H, dd, J1′b,2′ ) 7.5 Hz, H-1′b), 3.88
(1H, dd, J3′,4′b ) 3.4 Hz, H-4′b), 3.86 (1H, dd, J1a,2 ) 3.2 Hz, H-1a),
3.83 (1H, dd, J1b,2 ) 4.0, J1a,1b ) 12.2 Hz, H-1b). 13C NMR (100 MHz,
D2O): δ 83.10 (C-2′), 80.98 (C-3), 80.27 (C-2), 72.58 (C-4), 68.32
(C-2′), 62.34 (C-4′), 61.93 (C-5), 50.29 (C-1′), 47.75 (C-1). MALDI
MS: m/e 383.2 (M+ + H), 303.2 (M+ + H - SO3). Anal. Calcd for
C6H18O9SSe: C, 28.35; H, 4.76. Found: C, 28.12; H, 4.83.
using an enzyme concentration of 7.0 × 10-8 M and five inhibitor
concentrations in the range from 1 µm to 5 mM. The effects of the
inhibition on rates of substrate hydrolysis were compared for the
different compounds. The glucose released was analyzed in aliquots
removed at appropriate time intervals using a glucose oxidase assay
adapted to microtiter plate reading and using a total reaction volume
for the enzyme reaction mixtures of 150 or 300 µL.18 The Ki values
were calculated assuming competitive inhibition from 1/V ) (1/Vmax
)
+ [(Km)/(Vmax[S]Ki)][I], where V is the rate measured in the presence
or absence of inhibitor, [I] and [S] the concentrations of inhibitor and
substrate, Km ) 1.6 mM, and kcat ) 11.3 s-1, using ENZFITTER.19
Porcine pancreatic R-amylase (PPA) and bovine serum albumin
(BSA) were purchased from Sigma. Amylose EX-1 (DP17; average
degree of polymerization 17) was purchased from Hayashibara Chemi-
cal Laboratories (Okayama, Japan). Recombinant barley R-amylase
isozyme 1 (AMY1) was produced and purified as described.14 An
aliquot of the porcine pancreatic R-amylase (PPA) crystalline suspension
(in ammonium sulfate) was dialyzed extensively against the assay buffer
without BSA. The enzyme concentration was determined by aid of
amino acid analysis as determined using an LKB model Alpha Plus
amino acid analyzer. The inhibition of AMY1 (3 × 10-9 M) and PPA
(9 × 10-9 M) activity toward DP17 amylose was measured at 37 °C
in 20 mM sodium acetate, pH 5.5, 5 mM CaCl2, 0.005% BSA (for
AMY1), and 20 mM sodium phosphate, pH 6.9, 10 mM NaCl, 0.1
mM CaCl2, 0.005% BSA (for PPA). Six different final inhibitor
concentrations were used in the range from 1 µM to 5 mM. The inhibitor
was preincubated with enzyme for 5 min at 37 °C before addition of
substrate. Initial rates were determined by measuring reducing sugar
by the copper-bicinchoninate method as described.14,20 The Ki values
were calculated assuming competitive inhibition, as described above
for the case of glucoamylase, and a Km of 0.57 mg/mL and kcat of 165
s-1 for AMY1 and 1 mg/mL and 1200 s-1, respectively, for PPA, as
determined in the substrate concentration range 0.03-10 mg/mL using
ENZFITTER.19 For the Ki determinations, [S] ) 0.7 mg/mL amylose
DP 17 for the AMY1 binding and [S] ) 2.5 mg/mL amylose DP 17
for the PPA binding.
1,4-Dideoxy-1,4-[[(2R,3R)-2,4-dihydroxy-3-(sulfoxy)butyl]epi-
selenoniumylidene]-D-arabinitol Inner Salt (11). Hydrogenolysis
of the selenonium salt 22 (0.906 g, 1.22 mmol, trans:cis ) 5:1) by the
same procedure reported above for 19 was extremely sluggish. After a
total of 1.2 g of Pd/C had been added in three portions over 3 days of
stirring under an atmosphere of H2, little reaction had occurred. The
reaction was stopped and the catalyst removed by filtration through
Celite with methanol. The solvents were removed, and the residue,
consisting mostly of unreacted starting material 22, was repurified by
column chromatography. The fractions containing 22 were combined
and concentrated to a syrup. Hydrogenolysis of this material in acetic
acid (6 mL) with Pd/C (0.4 g) was now relatively rapid, and TLC
indicated complete reaction after 24 h. Processing and purification as
described for 10 gave 11 as a colorless foam (88 mg, 19%). This low
yield was attributed to losses of the selenonium salt by adsorption on
1
the large amounts of Pd/C that had been used. Analysis by H NMR
indicated that 11 was a mixture of isomers (8:1) at the selenium center.
The major component of the mixture was assigned to be the diastere-
omer with a trans relationship between C-5 and C-1′ on the basis of
observation of a strong H-1′/H-4 correlation in the NOESY spectrum.
The pure trans-11 was obtained by crystallization from MeOH.
Data for the major isomer trans-11 follow. Mp: 137-140 °C. [R]20
-33° (c 0.3, H2O). H NMR (600 MHz, D2O): δ 4.83 (1H, ddd,
:
D
1
H-2), 4.54 (1H, dd, J2,3 ) 3.6 Hz, H-3), 4.43 (1H, td, J1′a,2′ ) J2′,3′
6.9, J1′b,2′ ) 5.5 Hz, H-2′), 4.36 (1H, ddd, H-3′), 4.28 (1H, ddd, J3,4
)
)
Acknowledgment. We thank Professor Barry B. Snider for
a fruitful discussion and for the suggestion to investigate HFIP
as a reaction solvent. We are grateful to Sidsel Ehlers for
technical assistance with the glucoamylase assays. We also thank
the Natural Sciences and Engineering Research Council of
Canada for financial support and the Michael Smith Foundation
for Health Research for a trainee fellowship (to A.G.).
3.2 Hz, H-4), 4.10 (1H, dd, J4,5a ) 5.3, J5a,5b ) 12.7 Hz, H-5a), 4.00
(1H, dd, J4,5b ) 8.0 Hz, H-5b), 3.98 (1H, dd, J3′,4′a ) 3.1 Hz, H-4′a),
3.96 (2H, m, H-1′a, H-1′b), 3.87 (1H, dd, J3′,4′b ) 3.4, J4′a,4′b ) 12.8
Hz, H-4′b), 3.82 (2H, d, J1a,2 ) J1b,2 ) 4.0 Hz, H-1a, H-1b). 13C NMR
(D2O): δ 83.33 (C-3′), 81.15 (C-3), 80.34 (C-2), 75.54 (C-4), 68.67
(C-2′), 62.36 (C-4′), 61.86 (C-5), 49.96 (C-1′), 47.30 (C-1). MALDI
MS: m/e 383.0 (M+ + H), 303.0 (M+ + H - SO3). Anal. Calcd for
C6H18O9SSe: C, 28.35; H, 4.76. Found: C, 28.30; H, 5.01.
JA020299G
Enzyme Inhibition Assays. The glucoamylase G2 form from
Aspergillus niger was purified from a commercial enzyme (Novo
Nordisk, Bagsvaerd, Denmark) as described.12,13 The initial rates of
glucoamylase G2-catalyzed hydrolysis of maltose were tested with 1
mM maltose as substrate in 0.1 M sodium acetate pH 4.5 at 45 °C
(18) Frandsen, T. P.; Dupont, C.; Lehmbeck, J.; Stoffer, B.; Sierks, M. R.;
Honzatko, R. B.; Svensson, B. Biochemistry 1994, 33, 13808-13816.
(19) Leatherbarrow, R. J. Enzfitter, a nonlinear regression data analysis program
for IBM PC; Elsevier Science Publishers BV: Amsterdam, The Netherlands,
1987.
(20) Fox, J. D.; Robyt, J. F. Anal. Biochem. 1991, 195, 93-96.
9
8250 J. AM. CHEM. SOC. VOL. 124, NO. 28, 2002