A practical synthesis of kifunensine analogues as inhibitors of endoplasmic reticulum α-mannosidase I

Article abstract of DOI:10.1021/jo0516382

A practical synthesis of the potent class I α-mannosidase inhibitor kifunensine (1) beginning from the inexpensive and readily available starting material L-ascorbic acid (15) is described. The protected amino-alcohol ((2R,3R,4R,5R)-5-amino-2,3:4,6-diisop

Full text of DOI:10.1021/jo0516382

A Practical Synthesis of Kifunensine Analogues as Inhibitors of
Endoplasmic Reticulum r-Mannosidase I
Kirk W. Hering, Khanita Karaveg,†,‡ Kelley W. Moremen,†,‡ and William H. Pearson*,§
Department of Chemistry, University of Michigan, Ann Arbor, Michigan 48109-1055
wpearson@berryassoc.com
Received August 4, 2005
A practical synthesis of the potent class I R-mannosidase inhibitor kifunensine (1) beginning from
the inexpensive and readily available starting material L-ascorbic acid (15) is described. The
protected amino-alcohol ((2R,3R,4R,5R)-5-amino-2,3:4,6-diisopropylidenedioxyhexanol, 11) served
as a key intermediate from which several N-1 substituted kifunensine analogues (including
N-methyl, N-cyclohexyl, and N-bis(hydroxymethyl)methyl) and 2-desoxakifunensine analogues
(including N-H and N-methyl) were prepared and screened for inhibition of human endoplasmic
reticulum R-mannosidase I (ER Man I) and mouse Golgi R-mannosidase IA (Golgi Man IA). In
addition, several pseudodisaccharide kifunensine analogues in which a mannose residue was
tethered to N-1 of kifunensine via a two-, three-, or four-carbon linker and an affinity-bound
kifunensine analogue were also prepared and evaluated for biological activity. While the synthesized
N-1 kifunesine analogues were found to be less potent inhibitors of Class I R-mannosidases than
kifuensine itself, the bis(hydroxymethyl)methylkifunensine analogue 6 was shown to selectively
inhibit ER Man I over Golgi Man IA.
Introduction
respectively.4b,5 In addition to aiding in the synthesis and
maturation of N-linked glycoproteins,6 ER Man I also
plays a key role in the quality control of glycoprotein
folding within the endoplasmic reticulum (ER), a process
The class I R-mannosidase inhibitor kifunensine (1) is
a potent, naturally occurring glycosidase inhibitor of
Class I (CAZy glycosylhydrolase family 471)2 Kifunensine
has been reported to inhibit both human endoplasmic
reticulum R-1,2-mannosidase I (ER Man I) and members
of the Golgi subfamily of the Class I mannosidases,3 Golgi
R-mannosidase IA, IB, and IC (Golgi Man IA/IB/IC),4 in
addition to Class I mannosidases from plant origin (mung
bean R-1,2-mannosidase I (Golgi Man I)) with Ki values
of 130 nm, 23 nm, and an IC50 value of 20-50 nm,
(2) For reviews of the biological activity of kifunensine and other
glycosidase inhibitors, see: (a) Elbein, A. D. FASEB J. 1991, 5, 3055-
3063. (b) Winchester, B.; Fleet, G. W. J. Glycobiology 1992, 2, 199-
210. (c) Elbein, A. D.; Molyneux, R. J. In Iminosugars as Glycosidase
Inhibitors; Stutz, A. E., Ed.; Wiley-VCH: Weinheim, Germany, 1999;
pp 216-251. (d) Asano, N.; Hashimoto, H. In Glycoscience: Chemistry
and Chemical Biology; Fraser-Reid, B. O., Tatsuta, K., Thiem, J., Eds.;
Springer: Berlin; New York, 2001; Vol. 3, pp 2541-2594. (e) Asano,
N.; Nash, R. J.; Molyneux, R. J.; Fleet, G. W. J. Tetrahedron Asymmetry
2000, 11, 1645-1680.
(3) Moremen, K. W. In Oligosaccharides in Chemistry and Biology:
A Comprehensive Handbook, Vol. II. Biology of Saccharides, Part 1.
Biosynthesis of Glycoconjugates; Ernst, B., Hart, G., Sinay, P., Eds.;
Wiley and Sons: New York, 2000; pp 81-117.
(4) (a) Lal, A.; Pang, P.; Kalelkar, S.; Romero, P. A.; Herscovics, A.;
Moremen, K. W. Glycobiology 1998, 8, 981-995. (b) Tremblay, L. O.;
Herscovics, A. J. Biol. Chem. 2000, 275, 31655-31660.
† Complex Carbohydrate Research Center, University of Georgia,
Athens, GA 30602.
‡ Department of Biochemistry and Molecular Biology, University of
Georgia, Athens, GA 30602.
§ Current address: Berry & Associates, Inc., 2434 Bishop Circle
East, Dexter, MI 48130.
(1) (a) Henrissat, B.; Bairoch, A. Biochem. J. 1996, 316, 695-696.
(b) Henrissat, B.; Davies, G. Curr. Opin. Struct. Biol. 1997, 7, 637-
644. (c) Henrissat, B. Biochem. Soc. Trans. 1998, 26, 153-156. (d)
Bourne, Y.; Henrissat, B. Curr. Opin. Struct. Biol. 2001, 11, 593-600.
(e) Davies, G. J.; Henrissat, B. Biochem. Soc. Trans. 2002, 30, 291-
297.
(5) (a) Elbein, A. D.; Tropea, J. E.; Mitchell, M.; Kaushal, G. P. J.
Biol. Chem. 1990, 265, 15599-15605. (b) Vallee, F.; Karaveg, K.;
Herscovics, A.; Moremen, K. W.; Howell, P. L. J. Biol. Chem. 2000,
275, 41287-41298. (c) Karaveg, K.; Moremen, K. W. J. Biol. Chem.
2005, 280, 29837-29848.
10.1021/jo0516382 CCC: $30.25 © 2005 American Chemical Society
Published on Web 11/01/2005
9892
J. Org. Chem. 2005, 70, 9892-9904

Synthesis of Kifunensine Analogues
R1-antitrypsin Z variant.13 In contrast, overexpression of
ER Man I was found to accelerate the disposal of
misfolded glycoproteins.14 Surprisingly, acceleration of
the “disposal clock” by ER Man I overexpression also
triggered a significant disposal of wild-type glycoproteins.
Thus a model was proposed in which glycan trimming
to Man8GlcNAc2 moieties in the context of incompletely
folded or misfolded polypeptide structures acts as the
rate-determining step in generating the signal for gly-
coprotein disposal.14a In this model, ER Man I occupies
a unique intersection of both glycoprotein biosynthetic
and catabolic pathways and defines the binary decision
for the fate of nascent glycoproteins toward folding or
disposal. These results suggest that inhibition of ER Man
I may serve as a possible approach for the treatment of
genetic diseases that occur as a direct result of the
processes of ERAD.5c,14a,15 As a result of its potent
inhibition of ER Man I, kifunensine presents an excellent
means for the study and regulation of this important
enzyme and its role in ERAD. However, kifunensine is
also a potent inhibitor of Golgi Man I, an enzyme that is
necessary for the maturation of N-linked glycoproteins
into hybrid and complex type glycoproteins.4,5a,6 Thus, it
would be desirable to generate kifunensine analogues
capable of selectively inhibiting ER Man I over Golgi Man
I in order to effectively study the mechanism of ER Man
I and to aid in the development of novel therapeutic
agents for the treatment of genetic diseases related to
ERAD. A review of the literature provides several
examples in which pseudodisaccharide glycomimetics are
used in place of the corresponding monosaccharide
moieties to mimic an oligosaccharide in an effort to attain
greater potency and/or more selectivity in recognition by
a targeted enzyme.16 Accordingly, we believed that a
kifunensine pseudodisaccharide could be developed that
would have increased specificity for ER Man I over other
class I R-mannosidases. We report herein the synthesis
and preliminary biological evaluation of kifunensine, the
N-1 substituted 2-desoxakifunensine analogues 2 and 3,
and the N-1 substituted kifunensine analogues 4-11,
which include the tethered disaccharide mimetics 7-9
and the affinity bound kifunensine analogue 10 (Figure
1).
FIGURE 1. Structures of kifunensine and analogues.
that is commonly referred to as endoplasmic reticulum
associated degradation (ERAD).7 A considerable number
of genetic diseases are linked to the processes of ERAD.8
Recently, several studies have demonstrated that inhibi-
tion of ER Man I with kifunensine (1, Figure 1) or
deoxymannojirimycin9 leads to blockage of the degrada-
tion of mutant glycoproteins including the T cell receptor
subunit CD3-δ,10 tyrosinase,11 R2-plasmin inhibitor,12 and
(6) For reviews of N-linked gylocoprotein biosynthesis, see: (a)
Hubbard, S. C.; Ivatt, R. J. Annu. Rev. Biochem. 1981, 50, 555-583.
(b) Elbein, A. D. CRC Crit. Rev. Biochem. 1984, 16, 21-29. (c) Kornfeld,
R.; Kornfeld, S. Annu. Rev. Biochem. 1985, 54, 631-664. (d) Herscovics,
A. Biochim. Biophys. Acta 1999, 1473, 96-107. (e) Herscovics, A.
Biochim. Biophys. Acta 1999, 1426, 275-285.
(7) For reviews of quality control within the endoplasmic reticulum,
see: (a) Ellgaard, L.; Molinari, M.; Helenius, A. Science 1999, 286,
1882-1888. (b) Herscovics, A. In Comprehensive Natural Products
Chemistry; Pinto, B. M., Ed.; Elsevier: Amsterdam, 1999; Vol. 3, pp
13-35. (c) Parodi, A. J. Biochem. J. 2000, 348, 1-13. (d) Roth, J. Chem.
Rev. 2002, 102, 285-303. (e) Trombetta, E. S. Glycobiology 2003, 13,
77R-91R. (f) Spiro, R. G. Cell. Mol. Life Sci. 2004, 61, 1025-1041.
(8) For reviews of diseases related to ERAD, see: (a) Thomas, P. I.;
Qu, B.-H.; Pedersen, P. L. Trends. Biol. Sci. 1995, 20, 456-459. (b)
Choudhury, P.; Liu, Y.; Sifers, R. N. News Physiol. Sci. 1997, 12, 162-
165. (c) Brodsky, J. L.; McCracken, A. A. Semin. Cell Dev. Biol. 1999,
10, 507-513.
(9) For isolation of deoxymannojirimycin, see: (a) Legler, G.; Julich,
E. Carbohydr. Res. 1984, 128. For inhibition of Class I mannosidases
by deoxymannojirimycin, see: (b) Fuhrmann, U.; Bause, E.; Legler,
G.; Ploegh, H. Nature 1984, 307, 755-758. (c) Reference 5b.
(10) Yang, M.; Omura, S.; Bonifacino, J. S.; Weissman, A. M. J. Exp.
Med. 1998, 187, 835-846.
(11) Wang, Y.; Androlewicz, M. J. Biochem. Biophys. Res. Commun.
2000, 271, 22-27.
(12) Chung, D. H.; Ohashi, K.; Watanabe, M.; Miyasaka, N.;
Hirosawa, S. J. Biol. Chem. 2000, 275, 4981-4987.
Our studies began with an improved synthesis of
kifunensine itself. It was our wish to develop a convenient
and inexpensive preparation of kifunensine from which
intermediates could be used to generate kifunensine
analogues that may be useful for the study of class I
R-mannosidases. Kifunensine (1) was originally isolated
by Iwami et al. from the actinomycete Kitasatosporia
kifunensine no. 9482 and was designated FR900494 in
(13) (a) Liu, Y.; Choudhury, P.; Cabral, C. M.; Sifers, R. N. J. Biol.
Chem. 1999, 274, 5861-5867. (b) Marcus, N. Y.; Perlmutter, D. H. J.
Biol. Chem. 2000, 275, 1987-1992.
(14) (a) Wu, Y.; Swulius, M. T.; Moremen, K. W.; Sifers, R. N. Proc.
Natl. Acad. Sci. U.S.A. 2003, 100, 8229-8234. (b) Hosokawa, N.;
Tremblay, L. O.; You, Z.; Herscovics, A.; Wada, I.; Nagata, K. J. Biol.
Chem. 2003, 278, 26287-26294.
(15) (a) Tempel, W.; Karaveg, K.; Liu, Z. J.; Rose, J.; Wang, B. C.;
Moremen, K. W. J. Biol. Chem. 2004, 279, 29774-29786. (b) Karaveg,
K.; Siriwardena, A.; Tempel, W.; Liu, Z. J.; Glushka, J.; Wang, B. C.;
Moremen, K. W. J. Biol. Chem. 2005, 280, 16197-16207.
(16) For reviews of oligosaccharide mimetics, see: (a) Sinay, P. Pure
Appl. Chem. 1997, 69, 459-463. (b) Sears, P.; Wong, C.-H. Angew.
Chem., Int. Ed. 1999, 38, 2300-2324. (c) Vogel, P. Chimia 2001, 55,
359-365.
J. Org. Chem, Vol. 70, No. 24, 2005 9893

Hering et al.
that report.17 Subsequently, Kayakiri et al. reported the
structure and stereochemistry of kifunensine.18 The
structural complexity of the fused heterocyclic ring of
kifunensine has limited the synthetic approaches to only
two to date.19 The first synthesis, reported by Kayakiri
et al. in the early 1990s, was efficient and high yielding,
furnishing kifunensine in eight steps with a 32.5% overall
yield beginning from D-mannosamine.19a,b More recently,
Hudlicky and co-workers reported a formal synthesis of
kifunensine beginning from chlorobenzene, which utilizes
the same key intermediate and key synthetic step as was
first developed by Kayakiri.19c,d Of the two approaches,
Kayakiri’s synthesis presents the most desirable prepa-
ration of kifunensine known to date because of its short
synthetic sequence and high overall yield. However, the
relatively high cost and low availability of D-man-
nosamine as a starting material remains an unattractive
aspect of Kayakiri’s synthesis. As a result of our group’s
long-standing interest in novel preparations of azasugars
and azasugar analogues,20 we chose to undertake a
practical synthesis of kifunensine that could also be used
to generate novel N-1 substituted kifunensine analogues
for use in the study of class I R-mannosidases. We report
herein a practical synthesis of kifunensine and several
N-1 substituted kifunensine analogues beginning from
the inexpensive and readily available L-ascorbic acid. Our
preparation is relatively short and efficient, utilizing
simple, highly reproducible steps, and has allowed for the
synthesis of multiple gram quantities of 11, a precursor
first utilized in Hulicky’s formal synthesis of kifunensine
(Scheme 1).19c,d This intermediate has been readily
transformed into kifunensine and has served as an ideal
branching point for the synthesis and preliminary bio-
logical evaluation of a variety of novel N-1 substituted
kifunensine analogues.
SCHEME 1. Retrosynthetic Analysis of
Kifunensine and Analogues
SCHEME 2. Synthesis of Azido-lactone 13a
A retrosynthetic analysis of our strategy toward kifu-
nensine and analogues is outlined in Scheme 1. In a
formal synthesis, Hudlicky has previously obtained a
fully protected kifunensine precursor from the amino-
alcohol 11.19c,d We envisioned that 11 could also serve as
a key intermediate in the synthesis of a variety of N-1
substituted kifunensine analogues. Amino-alcohol 11
could be generated in several steps from the triol 12,
which could be readily obtained from reductive ring
a Conditions: (a) H2, 10% Pd/C, H2O, 50 °C, 7 days (75%); (b)
CuSO4, H2SO4, acetone (76%); (c) (1) AcOH, H2O, (2) TBSCl, Im,
DMF (87%, 2 steps); (d) HN3, Ph3P, DEAD, THF, rt (74%).
(17) Iwami, M.; Nakayama, O.; Terano, H.; Kohsaka, M.; Aoki, H.;
Imanaka, H. J. Antibiot. 1987, 40, 612-622.
(18) (a) Kayakiri, H.; Takase, S.; Shibata, T.; Okamoto, M.; Terano,
H.; Hashimoto, M. J. Org. Chem. 1989, 54, 4015-4016. (b) Kayakiri,
H.; Takase, S.; Shibata, T.; Hashimoto, M.; Tada, T.; Koda, S. Chem.
Pharm. Bull. 1989, 39, 1378-1381.
opening and deprotection of the known TBS-protected
azido-lactone 13.21 Azide 13 has been previously synthe-
sized from L-gulonolactone (14) by Fleet,21 and 14 can be
readily obtained by palladium-catalyzed hydrogenation
of L-ascorbic acid (15) using the method of Andrews.22
(19) (a) Kayakiri, H.; Kasahara, C.; Oku, T.; Hashimoto, M. Tetra-
hedron Lett. 1990, 31, 225-226. (b) Kayakiri, H.; Kasahara, C.;
Nakamura, K.; Oku, T.; Hashimoto, M. Chem. Pharm. Bull. 1991, 39,
1392-1396. (c) Rouden, J.; Hudlicky, T. J. Chem. Soc., Perkin Trans.
1 1993, 1095-1097. (d) Hudlicky, T.; Rouden, J.; Luna, H.; Allen, S.
J. Am. Chem. Soc. 1994, 116, 5099-5107.
(20) Swainsonine and swainsonine analogues: (a) Pearson, W. H.;
Chung, L. K. Tetrahedron Lett. 1990, 31, 7571-7574. (b) Pearson, W.
H.; Hembre, E. J. J. Org. Chem. 1996, 61, 5537-5545. (c) Pearson, W.
H.; Hembre, E. J. J. Org. Chem. 1996, 61, 7217-7221. (d) Hembre, E.
J.; Pearson, W. H. Tetrahedron 1997, 53, 11021-11032. (e) Pearson,
W. H.; Hines, J. V. J. Org. Chem. 2000, 65, 5785-5793. (f) Pearson,
W. H.; Hembre, E. J. Tetrahedron Lett. 2001, 42, 8273-8276. (g)
Pearson, W. H.; Guo, L.; Jewell, T. M. Tetrahedron Lett. 2002, 43,
2175-2178. (h) Pearson, W. H.; Ren, Y.; Powers, J. D. Heterocycles
2002, 58, 421-430. Australine and alexine analogues: (i) Pearson, W.
H.; Hines, J. V. Tetrahedron Lett. 1991, 32, 5513-5516. (j) Pearson,
W. H.; Hembre, E. J. J. Org. Chem. 1996, 61, 5546-5556. Other novel
glycosidases: (k) Pearson, W. H.; Hembre, E. J. Tetrahedron Lett. 1993,
34, 8221-8224.
Results and Discussion
1. Practical Synthesis of Kifunensine. The synthe-
sis of kifunensine (1) began with preparation of 13 using
previously reported methods (Scheme 2). Hydrogenation
of L-ascorbic acid (15) to L-gulonolactone (14) was ac-
complished in 75% yield by the method of Andrews.22, 23
For the next step, L-gulonolactone (14) was treated
with copper sulfate, sulfuric acid, and acetone using the
(21) Fleet, G. W. J.; Ramsden, N. G.; Witty, D. R. Tetrahedron 1989,
45, 319-326.
(22) Andrews, G. C.; Crawford, T. C.; Bacon, B. E. J. Org. Chem.
1981, 46, 2976-2977.
9894 J. Org. Chem., Vol. 70, No. 24, 2005

Synthesis of Kifunensine Analogues
conditions reported by Sano24 to furnish the desired bis-
acetonide 16 in 76% yield.25 Deprotection of the side chain
acetonide of 16 followed by selective protection of the
resulting primary alcohol provided silyl-ether 17 in
excellent yield.21 A variety of methods commonly used
in the formation of azides from alcohols with inversion
of stereochemistry were explored for the conversion of
17 to 13.26 The best conditions were determined to be
treatment of 17 with hydrazoic acid, triphenylphosphine,
and DEAD at room temperature, which provided the
azido-lactone 13 in 74% yield. This reaction has been
performed on a 15 g scale with no decrease in yield.
Once a high yielding route to azide 13 had been
developed, we next explored approaches to Hudlicky’s
intermediate 11 (Scheme 3).19c,d Sodium borohydride
reduction of lactone 13 using the conditions reported by
Fleet for the selective reduction of a similar azido-
lactone27 furnished diol 18 with the azide intact in
excellent yield. Unfortunately, attempts to obtain 11
directly via the triol 12 were unsuccessful due to the
instability of 12.28 To circumvent this problem, the
primary alcohol of 18 was selectively protected as the
pivalate ester 19 in 88% yield by treatment with pivaloyl
chloride and pyridine. Treatment with TBAF deprotected
the silyl ether, and the resulting diol was readily pro-
tected to give the acetonide 20.29 Lithium aluminum
hydride effected reduction of both the azide and ester
moieties of 20, furnishing key intermediate 11 in 98%
yield. This amino-alcohol could be crystallized by tritu-
ration in ether and proved to be stable at room temper-
ature for at least 1 year.
SCHEME 3. Completion of Kifunensine Synthesisa
a Conditions: (a) NaBH4, EtOH (87%); (b) TBAF, THF, 0 °C
(89%); (c) PivCl, pyr, CH2Cl2 (88%); (d) (1) TBAF, THF, -20 °C,
(2) 2-methoxypropene, PTSA, CH2Cl2 (91%, 2 steps); (e) LiAlH4,
Et2O (98%); (f) (1) oxamic acid, EDC, HOBT, DMF, rt, (2) mp-
carbonate resin (93%); (g) (1) CrO3‚pyr, (2) NH3/MeOH (76%, 2
steps); (h) (1) Dess-Martin periodinane, CH2Cl2, (2) NH3/MeOH
(89%, 2 steps); (i) 75% TFA/H2O (80%).
Finally, to complete the synthesis of kifunensine, EDC
coupling of amino-alcohol 11 with oxamic acid in the
(23) (a) We found that in our hands the reduction of L-ascorbic acid
(15) to L-gulonolactone (14) could be accomplished in no greater than
75% yield with the reaction taking an average of 7 days at 50 °C under
75 psi hydrogen. The original report by Andrews indicates that this
transformation proceeded in 99% yield over 24 h at 50 °C under 50
psi hydrogen (see ref 22). However, in more recent reports, other groups
have also noted difficulty reproducing this result (see ref 23b,c). (b)
Vekemans, J. A. J. M.; Boerekamp, J.; Godefroi, E. F. Recl. Trav. Chim.
Pays-Bas 1985, 104, 266-272. (c) Czarnocki, Z.; Mieczkowski, J. B.;
Ziolkowski, M. Tetrahedron: Asymmetry 1996, 7, 2711-2720.
(24) Sano, H.; Sugai, S. Tetrahedron: Asymmetry 1995, 6, 1143-
1150.
(25) We found that treatment of 14 using 2,2-DMP and catalytic
PTSA as reported by Fleet (see ref 21) afforded a 1:1 mixture of the
desired product 16 and the product resulting from ring opening of the
lactone by MeOH generated in the reaction. We were unable to avoid
formation of this mixture, thus an alternate method of generating 16
was pursued.
(26) (a) Fleet reported that conversion of the secondary alcohol of
17 to the triflate, followed by displacement with sodium azide furnished
the azido-lactone 13 in 76% over two steps (see ref 21). However, we
were never able to obtain greater than 30% yield using this two step
sequence. Therefore, a variety of conditions were screened for the
conversion of 17 to 13. For the two-step sequence, the use of tetra-n-
butylammonium azide in place of sodium azide afforded no increase
in yield. Thus, a variety of Mitsunobu conditions were explored for
the direct conversion of 17 to 13 in one step using diphenylphosphoryl
azide (see ref 26b), zinc azide/bispyridine complex (see ref 26c), and
hydrazoic acid (see ref 26d). (b) Lal, B.; Pramanik, B. N.; Manhas, M.
S.; Bose, A. K. Tetrahedron Lett. 1977, 18, 1977-1980. (c) Viaud, M.
C.; Rollin, P. Synthesis 1990, 130-132. (d) Wolff, H. In Organic
Reactions; Adams, R., Ed.; Wiley: New York, 1946; Vol. 3, pp 307-
336.
presence of HOBT afforded Kayakiri’s key intermediate
21.19b Intermediate 21 was then subjected to Kayakiri’s
oxidation/double cyclization protocol19b followed by ac-
etonide deprotection to furnish kifunensine (1) in 12.2%
yield over 14 steps from ascorbic acid. However, we found
that the Collins oxidation conditions originally used by
Kayakiri et al. for the oxidation/double cyclization of 21
to 22 were not conducive toward analogue synthesis (vida
infra). Therefore, alternative oxidation conditions were
explored for this substrate. Kayakiri et al. report that
the aldehyde resulting from oxidation of 21 is unstable
to both silica gel and aqueous workup, and thus only
oxidations with a nonaqueous workup could be utilized.19b
After screening a variety of oxidants with 21 it was found
that the oxidation/double cyclization of 21 was best
achieved by treatment first with Dess-Martin periodi-
nane and then with excess ammonia in methanol to give
protected kifunensine 22 in 89% yield over two steps
(Scheme 3).30 Thus, a total synthesis of kifunensine was
completed by our optimized route in 14.2% yield over 14
steps from ascorbic acid.
While our preparation is slightly longer than the
previously reported approaches to kifunensine,19 this
synthesis requires only nine purification steps, of which
(27) Fairbanks, A. J.; Fleet, G. W. J.; Jones, A. H.; Bruce, I.; Al
Daher, S.; di Bello, I. C.; Winchester, B. Tetrahedron 1991, 47, 131-
138.
(28) Treatment of TBS-ether 18 with TBAF afforded a highly
unstable triol (12) in which the acetonide was easily scrambled at room
temperature.
(29) Wolfrom, M. L.; Diwadkar, A. B.; Gelas, J.; Horton, D. Carbo-
hydr. Res. 1974, 35, 87-96.
(30) We observed no formation of the protected kifunensine 8a-
epimer of 22 upon oxidation/double cyclization of 21, as previously
reported by Kayakiri et al. (see ref 19b). However, this was likely due
to the small scale at which the reactions were performed.
J. Org. Chem, Vol. 70, No. 24, 2005 9895

Hering et al.
could also be useful for mechanistic studies with ER Man
I to determine the amino acid residues that serve as the
catalytic acid and base within the enzyme active site.31
We also planned to synthesize the 2-desoxakifunen-
esine analogue 25 as another potential ER Man I inhibi-
tor. Compounds such as 25 lacking the C-2 carbonyl
should be more flexible than the corresponding kifun-
ensine analogues, allowing the desoxa compounds to
better mimic the O-glycosidic bond in the natural sub-
strate 23 yet remain resistant to hydrolysis.
Retrosynthetically, the kifunensine analogues of the
general structure 30 could be generated from an oxida-
tion/double cyclization reaction32 between Kayakiri’s
intermediate 21 and a suitable amine (Scheme 4). We
envisioned that 2-desoxakifunensine analogues of the
general structure 28 could be accessed by oxidation/
double cyclization of 29 with the desired amine. Amide
29 could in turn be easily generated by acylation of 11,
a key intermediate in our synthesis of kifunensine.
Initially, we planned to prepare cyclohexylkifunensine
analogues 5 and 26 and bis(hydroxymethyl)methylkifu-
nensine analogues 6 and 27 in order to test the feasibility
of this method for the generation of the more complicated
pseudodisaccharides 24 and 25 (Figure 2).
At the outset of the synthesis, little was known about
the position of binding of kifunensine within the ER Man
I active site. Thus the N-1 substituted kifunensine
analogues shown in Figure 2 were considered to be
excellent choices to serve as suitable mimics for the
mannobiose disaccharide 23. However, during the course
of this project, Moremen and colleagues published crystal
structures of human ER Man I with kifunensine and
deoxymannojirimycin bound within the enzyme active
site.5b These reports revealed that the N-H bond at N-1
of kifunensine was pointed toward the active site wall
and that the two carbonyls at C-2 and C-3 were solvent
exposed. This information implied that the mannobiose
disaccharide is more likely bound within the ER Man I
active site in a conformation such as that shown in 23a
(Figure 3), a conclusion that was further bolstered in a
recent computational study33 and confirmed in a recent
crystal structure.15b Given that this conformation is
unattainable by the analogues shown in Scheme 4, we
proposed to synthesize kifunensine analogues in which
the mannose moiety is tethered to the kifunensine moiety
by a linker arm. A review of the literature provides
several examples of azasugars tethered to sugar moieties
via a variety of linkers to serve as oligosaccharide
mimics.34 Thus, the tethered pseudodisaccharide kifun-
FIGURE 2. Proposed kifunensine analogues.
only five were chromatographic separations. In addition,
this synthesis allowed for the generation of gram quanti-
ties of the key intermediate 11. This intermediate was
readily transformed into kifunensine using reported
methods and provided an excellent branching point for
the synthesis of the novel kifunensine analogues pre-
sented in the next section.
2. Synthesis of Novel N-1 Substituted Kifun-
ensine Analogues. In addition to developing a new
route to kifunensine, we were also interested in the
preparation of novel N-1 substituted kifunensine ana-
logues that could be used as biochemical tools for the
study of class I R-mannosidases, as well as serve as
possible leads in the development of inhibitors that are
specific for ER Man I over other class I R-mannosidases.
Kifunensine pseudodisaccharide 24 was chosen as our
initial synthetic target to mimic mannobiose disaccharide
23, an ER Man I substrate (Figure 2). We believed that
pseudodisaccharide 24 should retain the inhibitory prop-
erties of kifunensine in the glycone binding site, allow
for additional inhibitor-enzyme interactions within the
aglycone binding site, and remain resistant to hydrolysis.
It was thought that a compound such as 24 could
capitalize upon added interactions within the aglycone
binding site allowing it to serve as a better inhibitor of
class I R-mannosidases than kifunensine. In addition, 24
SCHEME 4. Retrosynthesis of Proposed Kifunensine Analogues
9896 J. Org. Chem., Vol. 70, No. 24, 2005

Synthesis of Kifunensine Analogues
SCHEME 5. Synthesis of Amine Starting
Materialsa
a Conditions: (a) LiAlH4, Et2O; (b) EDC, CH2Cl2; (c) H2, 10%
Pd/C, EtOAc.
FIGURE 3. Structure of proposed tethered kifunensine
proven extremely difficult to purify, despite the cloning
and expression work accomplished by Moremen.37 Al-
though deoxymannojirimycin-bound affinity resins have
already been successfully used for the purification of pig
liver R-mannosidase I,38 it was thought that a kifun-
ensine-bound affinity resin could provide a more valuable
tool for the purification of class I R-mannosidases because
kifunensine is approximately 100-fold more potent of an
inhibitor of these enzymes than deoxymannojirimycin.5b
The development of an affinity-bound kifunensine chro-
matography matrix could simplify the methods required
to purify class I R-mannosidases and aid in the discovery
of other undetermined class I R-mannosidases located in
the golgi and ER. We planned to synthesize 10, with the
tether arm linked to N-1 of kifunensine, which corre-
sponds to the position of the glycosyl linkage. This
linkage should not interfere with the binding of kifun-
ensine to the enzyme active site as the D-manno config-
ured hydroxyl groups critical for binding would remain
unhindered.
analogues.
ensine analogues 7, 8, and 9 were chosen as suitable
targets.35 These could be synthesized by the same route
described in Scheme 4 using the amino-tethered, pro-
tected mannosamines 33, 34, and 35 (Figure 3). We
envisioned that the variation in tether length would allow
for different levels of flexibility in binding for each
tethered monosaccharide moiety such that one or more
of the analogues 7-9 could bind ER Man I in a potent
and selective manner.
Finally, the synthesis of an affinity-bound kifunensine
analogue such as 10 (Figure 2) was planned for use in
the purification of class I R-mannosidases. Many of the
glycosidases from the N-linked glycoprotein processing
pathway are membrane-bound proteins, which are gen-
erally more difficult to purify than their soluble coun-
terparts.36 Mammalian class I R-mannosidases have
(31) (a) Recent crystal structures of ER Man I obtained in the
presence of Man9GlcNAc2 (see 31b), kifunensine, or deoxymannojiri-
mycin (see ref 5b) did not conclusively determine the identity of the
catalytic acid and catalytic base required for the hydrolysis of the
necessary mannose residue from the oligosaccharide moiety. However,
during the course of this project Moremen and colleagues were able to
identify the catalytic acid and catalytic base upon determination of a
cocrystal structure of ER Man I with a nonhydrolyzable thiodisaccha-
ride pseudosubstrate (see ref 15b). (b) Vallee, F.; F., L.; Yip, P.; Sleno,
B.; Herscovics, A.; Howell, P. L. EMBO J. 2000, 19, 581-588.
(32) For details on the oxidation/double cyclization reaction sequence
towards the synthesis of kifunensine, see ref 19b.
(33) Mulakala, C.; Reilly, P. J. Proteins 2002, 49, 125-134.
(34) For selected examples, see: (a) Qiao, L.; Murray, B. W.;
Shimazaki, M.; Schultz, J.; Wong, C.-H. J. Am. Chem. Soc. 1996, 118,
7653-7662. (b) Johns, B. A.; Johnson, C. R. Tetrahedron Lett. 1998,
39, 749-752.
(35) Examination of molecular models indicated that a tether length
of four atoms between kifunensine and the sugar ring would be optimal
for mimicking the likely conformation of a disaccharide bound in the
ER Man I active site.
(36) (a) Kaushal, G. P.; Elbein, A. D. Methods Enzymol. 1989, 179,
452-475. (b) Tulsiani, D. R. P.; Touster, O. Methods Enzymol. 1989,
179, 446-451. (c) Kaushal, G. P.; Elbein, A. D. Methods Enzymol. 1994,
230, 316-329.
Synthesis of Amine Starting Materials. Before
undertaking the synthesis of the kifunensine analogues
proposed above, the prerequisite amine starting materi-
als were prepared as shown in Scheme 5. First, treatment
of the known azide 3639 with lithium aluminum hydride
afforded the protected 2-mannosamine 32 in excellent
yield. Next, EDC coupling of 32 with 2-azidoacetic acid
(37),40 3-azidopropionic acid (38),41 and 4-azidobutyric
acid (39)42 furnished the tethered azides 40, 41, and 42,
respectively. Palladium-catalyzed reduction of the azide
(37) Gonzalez, D. S.; Karaveg, K.; Vandersall-Nairn, A. S.; Lal, A.;
Moremen, K. W. J. Biol. Chem. 1999, 274, 21375-21386.
(38) Schweden, J.; Bause, E. Biochem. J. 1989, 264, 347-355.
(39) Sugawara, T.; Igarashi, K. Carbohydr. Res. 1988, 172, 195-
207.
(40) Lundquist, J. T.; Pelletier, J. C. Org. Lett. 2001, 3, 781-783.
(41) Lakanen, J. R.; Coward, J. K.; Pegg, A. E. J. Med. Chem. 1992,
35, 724-734.
(42) Khoukhi, N.; Vaultier, M.; Carrie, R. Tetrahedron 1987, 43,
1811-1822.
J. Org. Chem, Vol. 70, No. 24, 2005 9897

Hering et al.
SCHEME 6. Synthesis of 2-Desoxakifunensine
Analoguesa
SCHEME 7. Synthesis of Kifunensine Analogues
a Conditions: (a) ClCOCH2Cl, NaOAc, acetone, H2O (83%); (b)
(1) (COCl)2, DMSO, Et3N, CH2Cl2, -78 °C, (2) NH3 (100 equiv),
MeOH, rt (79%); (c) (1) (COCl)2, DMSO, Et3N, CH2Cl2, -78 °C,
(2) MeNH2 (100 equiv), MeOH, rt (68%, 2 steps); (d) (1) TPAP,
NMO, 4Å m.s., CH2Cl2, rt, (2) cyclohexylamine (5 equiv), NaOMe,
4Å m.s., MeOH, reflux (24%, 2 steps).
moieties of 40, 41, and 42 then furnished the desired
tethered mannosamines 33, 34, and 35, respectively, also
in excellent yields.
sequence. As mentioned previously, it was found that the
Dess-Martin oxidation afforded the best yields of product
after the two-step sequence. Using this protocol, protected
N-methylkifunensine 30a was prepared in excellent yield
(entry 1). For the model system, N-cyclohexylkifunensine
derivative 30b was obtained in 54% yield using the
conditions shown in entry 2. Thus, the model substrate
30b was successfully prepared in an acceptable yield
using only 2 equiv of amine in the oxidation/double
cyclization reaction. The bis(hydroxymethyl)methyl-
kifunensine derivative 30c was also obtained using the
known protected aminodiol 3144 with these conditions,
albeit in fairly low yield (entry 3).
Given the success obtained with the model system, a
synthesis of aza-N-disaccharide 30d was attempted.
However, oxidation of 21 followed by treatment of the
resulting aldehyde with protected mannosamine 32 using
the conditions developed for the model system resulted
in no product formation (entry 4). Several conditions were
investigated in an attempt to drive the double cyclization
to completion without success.45 We speculate that
protected mannosamine 32 is likely too sterically encum-
bered to undergo the double cyclization reaction with the
aldehyde that results from oxidation of 21.
Next, the synthesis of the tethered kifunensine ana-
logues 30e, 30f, and 30g was attempted. Oxidation of
21 followed by treatment with the appropriate tethered
amino sugar 33, 34, or 35, using the conditions developed
for the cyclohexylamine model system, resulted in forma-
tion of the desired protected kifunensine pseudodi-
saccharides 30e, 30f, and 30g, respectively (entries 5-7).
The reported yields for the synthesis of 30e-g are
optimized, indicating that these compounds are sterically
and perhaps electronically poor substrates for use in the
double cyclization reaction.
Synthesis of N-1 Substituted, 2-Desoxakifun-
ensine Analogues. The first series of analogues to be
investigated were the 2-desoxakifunensine analogues of
the general structure 28 shown in Scheme 6. Thus,
treatment of 11 with chloroacetyl chloride in the presence
of sodium acetate furnished the desired oxidation/double
cyclization precursor 29 in good yield. A variety of
conditions were attempted for the oxidation of 29 to the
corresponding aldehyde, and it was found that the Swern
and tetra-n-propylammonium perruthenate (TPAP) oxi-
dations with nonaqueous workups provided the best
yields of the desired aldehyde. This aldehyde was found
to be unstable and was thus used immediately in the
double cyclization step without purification. Treatment
of the aldehyde with a large excess of ammonia or
methylamine in methanol provided good yields of the
2-desoxakifunensine analogues 28a and 28b. The cyclo-
hexylamine derivative 28c proved much more difficult
to prepare. Although many conditions were screened,43
the best results obtained were by treating the crude
aldehyde with 5 equiv of cyclohexylamine and sodium
methoxide in refluxing methanol. However, a 24% yield
of model substrate 28c was not acceptable for further
pursuit of other inhibitors related to this class of ana-
logues.
Synthesis of N-1 Substituted Kifunensine Ana-
logues. Next, reaction conditions were explored for the
synthesis of kifunensine analogues of the general struc-
ture 30 (Scheme 7). Early on, it was discovered that
subjecting 21 to a Collin’s oxidation followed by double
cyclization with 2-5 equiv of amine did not furnish any
of the desired products. This was likely due to poisoning
of the double cyclization reaction by residual chromium
salts from the oxidation step. Given that a synthetically
complex, sugar-like amine was ultimately intended for
use in the oxidation/double cyclization reaction, alterna-
tive oxidation conditions were explored that would allow
for the use of 1-2 equiv of amine to accomplish this
Once the protected kifunensine and 2-desoxakifun-
ensine analogues had been obtained, methods were
explored for deprotection of these compounds to the
(44) Harada, H.; Morie, T.; Hirokawa, Y.; Kato, S. Chem. Pharm.
Bull. 1996, 44, 2205-2212.
(45) Conditions attempted included heating the imine in tert-butyl
alcohol or toluene and treatment of the imine with pyridinium
p-toluenesulfonate (PPTS), BF3‚OEt2 or triethylamine in the presence
methylene chloride. However, no product formation was observed with
any of these reaction conditions.
(43) A variety of other conditions were attempted to generate 28c
including treatment with 100 equiv of cyclohexylamine in a variety of
solvents at ambient and elevated temperatures, heating in refluxing
toluene, and heating in a variety of solvents under acidic conditions
(PTSA, Sc(OTf)3). However, none of these conditions resulted in greater
than 10% yields of the desired product.
9898 J. Org. Chem., Vol. 70, No. 24, 2005

Synthesis of Kifunensine Analogues
TABLE 1. Conditions Attempted for the Synthesis of
SCHEME 9. Deprotection of Tethered
Kifunensine Analoguesa
30a-g
reagents
(equiv amine)
temp time
yield
entry
solvent (°C) (h) product (%)a
1
2
MeNH2 (100)
CyNH2 (2), Et3N,
ui4Å m.s.
MeOH 23
CH2Cl2 60b
16
16
30a
30b
72
54
3
4
5
6
7
31 (2), Et3N, 4Å m.s. CH2Cl2 60b
32 (2), Et3N, 4Å m.s. CH2Cl2 60b
33 (2), Et3N, 4Å m.s. MeOH 65
34 (2), Et3N, 4Å m.s. MeOH 65
35 (2), Et3N, 4Å m.s. CH2Cl2 60b
16
16
16
16
16
30c
30d
30e
30f
22
0
30
34
35
30g
a Combined yield over two steps. b Reaction performed in a
sealed tube.
SCHEME 8. Deprotection of Kifunensine and
2-Desoxakifunensine Analoguesa
a Conditions: (a) 45 psi H2, 10% Pd/C, MeOH; (b) PTSA, MeOH,
ethylene glycol, reflux; (c) Ac2O, pyr; (d) 20% Pd(OH)2/C, cyclo-
hexene, EtOH, reflux; (e) NaOMe, MeOH; (f) Dowex 50Wx8-100
(pyridinium form).
ene in refluxing ethanol were necessary to remove the
benzyl ether and the benzylidene acetal groups of 30f and
30g. In addition, these conditions resulted in partial
acetonide removal for both 30f and 30g. Because of the
partial removal of the benzylidene and isopropylidene
acetals of 30e, 30f, and 30g during the debenzylation
step, the crude products were taken on directly for acetal
removal.
Several different conditions were attempted for the
deprotection of the remaining acetonides, often resulting
in a mixture of desired and undesired products.46 Ulti-
mately, it was found that treatment with PTSA and
excess ethylene glycol in refluxing methanol using a
modification of the conditions reported by Nicolaou47
formed the desired crude products. However, these
compounds were not amenable to purification by HPLC
or chromatography. Therefore, they were peracetylated
to furnish moderate yields of 43, 44, and 45 over three
steps. Removal of the acetates was accomplished by
treatment with sodium methoxide in methanol, and the
a Conditions: (a) 1 N HCl, THF; (b) Dowex 1x8-200 (OH form).
desired azasugar analogues for enzymatic testing. Thus,
treatment of kifunensine derivative 30a with 1 N HCl
in THF afforded an excellent yield of N-methylkifuensine
4, previously prepared by Kayakiri (Scheme 8).19a,b Next,
treatment of 30b and 30c with the same conditions
resulted in formation of 5 and 6, respectively, also in
excellent yields. The 2-desoxakifunensine derivatives 28a
and 28b were similarly deprotected and then passed over
a short column of Dowex 1x8-200 resin (OH- form) to
obtain the free base analogues 2 and 3.
The tethered pseudodisaccharide kifunensine ana-
logues 30e-g proved extremely difficult to deprotect
because of the sensitive functionalities present within
these compounds (Scheme 9). Different conditions were
required for removal of the benzyl groups and benz-
ylidene acetals of 30e than for 30f and 30g. Thus,
treatment of 30e with 10% Pd/C under 45 psi H2 resulted
in complete removal of the benzyl ether and partial
removal of the benzylidene acetal. Transfer hydrogena-
tion conditions using Pearlman’s catalyst and cyclohex-
(46) Treatment with the 1 N HCl/THF conditions used for depro-
tection of the other kifunensine analogues caused hydrolysis of the
methyl glycosidic bond of the attached mannosamine. Alternatively,
treatment with refluxing 80% acetic acid in water gave incomplete
acetonide removal even after several days. Treatment with 75%
trifluoroacetic acid in water, conditions employed by Kayakiri et al.
for diacetonide removal in the original kifunensine synthesis (see ref
19b), resulted in hydrolysis of the amide tether, as did treatment with
2% HCl in methanol.
(47) Nicolaou, K. C.; Mitchell, H. J.; Fylaktakidou, K. C.; Rodriguez,
R. M.; Suzuki, H. Chem. Eur. J. 2000, 6, 3116-3148.
J. Org. Chem, Vol. 70, No. 24, 2005 9899

Hering et al.
TABLE 2. Inhibition (IC50 Values) of Human ER Man I
and Mouse Golgi ManIA by Kifunensine Analogues
SCHEME 10. Synthesis of Affinity-Bound
Kifunensine Analoguea
IC50 (mM)
compd
human Er Man I
mouse Golgi Man IA
1
<0.03
0.03
0.9
0.08
2.6
0.2
>5.0
2.1
2.2
>5.0
5.0
<0.03
0.15
1.0
0.17
>5.0
2.1
2
3
4
5
6
7
>5.0
>5.0
1.8
8
9
10a
49
ndb
ndb
a Slurries of 0.5%, 5% and 50% were used for the three different
concentrations assayed. b Not determined.
Although none of our compounds inhibit ER Man I as
well as kifunensine (1), these data indicates that the N-1
substituted kifunensine and 2-desoxakifunensine deriva-
tives do indeed inhibit both human ER Man I and mouse
Golgi Man IA. The N-H 2-desoxakifunensine analogue 2
and the N-methyl kifunensine analogue 4 are the best
inhibitors, likely due to the small steric presence around
N-1 of the kifunensine substructure. Unfortunately, the
2-desoxakifunensine analogues 2 and 3 are relatively
poorer inhibitors of ER Man I as compared to kifun-
ensine. This may indicate that the free amine is interact-
ing unfavorably within the glycone binding site or that
replacement of the C-2 carbonyl with a methylene results
in loss of the structural rigidity necessary to hold the
fused pyranose ring hydroxyls in the desired 1C4 confor-
mation necessary for activity.5b The N-cyclohexyl kifun-
ensine and N-bis(hydroxymethyl)methyl kifunensine ana-
logues 5 and 6, respectively, demonstrate significant
inhibition of human ER Man I, with the latter compound
having a ∼10-fold greater selectivity for inhibition of ER
man I over Golgi Man IA. This was unexpected given the
anticipated steric interactions with active site residues
for the N-1 substituents of the analogues for both of the
Class I mannosidases and the similarities in positions
of the active site residues in the +1 and -1 subsites for
both enzymes.5b,49 It appears that the tethered pseudo-
disaccharide analogues 8 and 9 may indeed be able to
bind the attached mannosamine residue in the aglycone
binding site as demonstrated by IC50 values of ∼2 mM.
However, 7, which contains the shortest tether, does not
inhibit human ER Man I to any considerable degree.
Similarly, both the Affi-Gel linked kifunensine inhibitor
10 and the tethered amide 49 were poor inhibitors of
human ER Man I even at the highest concentrations
tested. Thus, the affinity bound kifunensine analogue 10
would likely not be useful for the purification of class I
R-mannosidases. Given the fact that the tethered pseudo-
disaccharide kifunensine analogues 8 and 9 exhibited
reasonable inhibition of ER Man I, it is interesting that
the N-hexylamido kifunensine analogue 49 exhibited
such poor inhibition (IC50 value of ∼5 mM). This may
indicate that the amide functionality exhibits an unfa-
vorable interaction within the aglycone binding site, in
which case a switch to either an ether or ester linkage
a Conditions: (a) (1) Dess-Martin [O], CH2Cl2, 4Å m.s., (2)
H2N(CH2)6NHBoc, CH2Cl2, Et3N, 4Å m.s., 60 °C, sealed tube (66%,
2 steps); (b) 1 N HCl, THF (98%); (c) Ac2O, Et3N, DMF (69%); (d)
NH3, MeOH (64%); (e) Affigel-10, MeOH, Et3N, 4 °C (100%).
basic salts were removed by passage over a short column
of Dowex 50Wx8-100 resin (pyridinium form) to furnish
the desired products 7, 8, and 9 in excellent yields.
Synthesis of an Affinity-Bound N-1 Linked Kifu-
nensine Analogue. Next the synthesis of an affinity-
bound kifunensine analogue was undertaken. Dess-
Martin oxidation of oxamido-alcohol 21 followed by
treatment of the resulting aldehyde with 2 equiv of
commercially available H2N(CH2)6NHBoc‚HCl, using the
conditions developed previously in our model system,
yielded 66% of the Boc-protected kifunensine analogue
46. Global deprotection was accomplished by treatment
with 1 N HCl in THF to give 47 as the hydrochloride
salt. Compound 47 was quantitatively appended to Affi-
Gel 10 resin to furnish the desired affinity-bound kifu-
nensine analogue 10.48 In the event that the Affi-Gel-
bound analogue 10 did not bind to ER Man I, the
N-acetamido derivative 49 was also prepared to serve as
a control for enzyme screening. The N-acetamido ana-
logue 49 was prepared by peracetylation of 47 to generate
48, followed by removal of the acetate protecting groups
by treatment with ammonia.
R-Mannosidase I Inhibitory Activities of Kifun-
ensine Analogues. All of the synthetic kifunensine and
2-desoxakifunensine analogues were then screened for
inhibitory activity against human ER Man I and mouse
Golgi Man IA by determining IC50 values for the respec-
tive compounds (Table 2).
(48) Coupling procedures may be found in the booklet Activated
Immunoaffinity Supports and Bulletin 1085, available from Bio-Rad
Laboratories, Inc., Life Sciences Group, 2000 Alfred Nobel Drive,
Hercules, CA 94547.
(49) Tempel, W.; Karaveg, K.; Lui, Z.-J.; Rose, J.; Wang, B.-C.;
Moremen, K. W. J. Biol. Chem. 2004, 279, 29774-29786.
9900 J. Org. Chem., Vol. 70, No. 24, 2005

Synthesis of Kifunensine Analogues
The reaction was quenched with the addition of an excess of
ammonium chloride with effervescence, filtered and concen-
trated to yield an oily white solid. Chromatography through a
plug of silica (30% EtOAc/hexanes) afforded 8.25 g (89%) of
the title compound as a clear oil. Data for 18: Rf ) 0.03 (30%
may improve the inhibitory properties of these kifun-
ensine analogues.
In summary, a practical synthesis of kifunensine has
been accomplished using L-ascorbic acid as an inexpen-
sive and readily available starting material. The syn-
thetic route allowed for the synthesis of a number of novel
N-1 substituted kifunensine and 2-desoxakifunensine
derivatives including an affinity-bound kifunensine ana-
logue. Preliminary biological assays of the reported
compounds with human ER Man I and mouse Golgi Man
IA yielded varying results, with none of the compounds
exhibiting inhibition comparable to kifunensine alone.
However, the improved specificity for the N-bis(hy-
droxymethyl)methyl kifunensine analogue 6 and the
reasonable affinity for the tethered kifunensine pseudo-
disaccharides 8 and 9 show promise that future N-1
substituted kifunensine analogues may be developed that
are capable of possessing selectivity in the binding of
class I R-mannosidases. Such compounds could enhance
our understanding of the role that ER Man I plays in
the degradation of misfolded proteins within the ER and
may be useful for the treatment of genetic diseases
related to ERAD.
EtOAc/hexanes); [R]23 ) -43.8° (c 1.58, CH2Cl2); 1H NMR
D
(CDCl3, 400 MHz) δ 4.41 (dd, J ) 1.4, 7.3 Hz, 1H), 4.30 (dt, J
) 4.4, 7.3 Hz, 1H), 4.09 (dd, J ) 3.1, 10.6 Hz, 1H), 3.90-3.85
(m, 2H), 3,80 (dd, J ) 4.4, 12.4 Hz, 1H), 3.62 (dd, J ) 1.4, 9.1
Hz, 1H) 3.48 (ddd, J ) 3.1, 6.3, 9.1 Hz, 1H), 1.53 (s, 3H), 1.41
(s, 3H), 0.92 (s, 9H), 0.11 (app d, J ) 2.2 Hz, 6H); 13C NMR
(CDCl3, 100 MHz) δ 108.4, 75.5, 68.3, 64.2, 64.0, 60.9, 26.7,
25.7, 24.7, 18.2, -5.6, -5.7; IR (neat) 3414 (br m), 2100 (s)
cm-1; MS (EI) m/z (rel int) 346 (M - CH3, 4), 328 (9), 246 (24),
131 (40), 116 (83), 73 (100), 59 (94); HRMS (DCI w/ NH3) calcd
for C15H32N3O5Si (M + H) 362.2111, found 362.2113.
(2R,3R,4R,5R)-5-Amino-2,3:4,6-diisopropylidenedioxy-
hexanol (11). By an alternate method of preparation for the
known title compound,19c,19d lithium aluminum hydride (1.94
g, 51.2 mmol) was added to a solution of azido-ester 20 (6.34
g, 17.1 mmol) in Et2O (400 mL) at room temperature and
stirred 1 h. The reaction was quenched with successive
addition of water (1.95 mL), 10% NaOH (1.95 mL), and water
(3.90 mL), the ether was poured off, and the solid was
extracted with EtOAc. The combined organic phases were
dried (Na2SO4) and concentrated to give an oily yellow solid,
which was triterated in hot ether and filtered to yield 3.99 g
(89%) of the title compound as a pale yellow solid. The crude
material was sufficiently pure for characterization. Data for
Experimental Section
11: mp ) 95.5-96.5 °C; Rf ) 0.21 (10% MeOH/CH2Cl2); [R]23
For general experimental details, see Supporting Informa-
tion.
D
) -38.7° (c ) 1.00, CH2Cl2); 1H NMR (CDCl3, 400 MHz) δ 4.50
(dd, J ) 2.3, 6.7 Hz, 1H), 4.29 (dt, J ) 4.8, 6.7 Hz, 1H), 3.89
(dd, J ) 5.3, 11.5 Hz, 1H), 3.80, (dd, J ) 4.8, 12.0 Hz, 1H),
3.76 (dd, J ) 4.8, 12.0 Hz, 1H), 3.53 (dd, J ) 2.3, 9.3 Hz, 1H),
3,50 (dd, J ) 9.3, 11.5 Hz, 1H), 3.11 (td, J ) 5.3, 9.3 Hz, 1H),
1.85 (br s, 3H), 1.53 (s, 3H), 1.48 (s, 3H), 1.43 (s, 3H), 1.39 (s,
3H); 13C NMR (CDCl3, 100 MHz) δ 108.7, 98.9, 77.7, 74.7, 73.9,
66.2, 61.3, 45.9, 28.4, 26.6, 25.7, 19.3; IR (neat) 3355 (br m),
3287 (m), 1606 (w) cm-1; MS (EI) m/z (rel int) 246 (M - CH3,
10), 72 (18), 59 (22), 43 (100); HRMS (ES) calcd for C17H29N3-
NaO6 (M + Na) 284.1474, found 284.1463.
(3S,4S,5R)-5-[(1′R)-1′-Azido-2′-tert-butyldimethylsila-
nyloxyethyl]-3,4-isopropylidenedioxytetrahydrofuran-
2-one (13). As an alternate method of preparation, the known
title compound21 was synthesized by Mitsunobu reaction using
freshly prepared hydrazoic acid.26d Caution: Hydrazoic acid
should be handled with extreme care. All work involving
hydrazoic acid solutions were carried out in an efficient fume
hood. The solution was transferred by cannula and any excess
hydrazoic acid was quenched by the addition of 10% NaOH.
Diethylazodicarboxylate (3.61 mL, 3.99 g, 22.9 mmol) and
hydrazoic acid (98.0 mL of a 0.86 M solution in toluene, 84.3
mmol) were added to a solution of triphenylphosphine (22.11
g, 84.28 mmol) in THF (280 mL) at room temperature and
stirred for 1 min. Then a solution of the known alcohol 1721
(14.01 g, 42.14 mmol) in THF (140 mL) was cannulated into
the reaction mixture and stirred for 20 h. The reaction was
quenched with the addition of methanol and concentrated to
yield a light orange oil. The residue was triterated in 30%
EtOAc/hexanes, filtered and concentrated to give a yellow oil.
Chromatography (10% EtOAc/hexanes) afforded 9.20 g (61%)
of the title compound as a white crystalline solid. We discov-
ered that the literature data for 13 was misreported.21 Thus,
full characterization was obtained for 13. Data for 13: mp )
84-85 °C; Rf ) 0.66 (30% EtOAc/hexanes); [R]23D ) -11.0° (c
(2R,3R,4R,5R)-2,3:4,6-Diisopropylidenedioxy-5-oxa-
moylaminohexanol (21). By an alternate method of prepa-
ration for this known compound,19b 1-(3-dimethylaminopropyl)-
3-ethylcarbodiimide hydrochloride (333 mg, 1.74 mmol) was
added to a solution of oxamic acid (104 mg, 1.16 mmol),
1-hydroxybenzotriazole hydrate (236 mg, 1.74 mmol), and
amino-alcohol 11 (304 mg, 1.16 mmol) in DMF (5.8 mL) and
stirred for 1 h at room temperature. Then macroporous
triethylammonium methylpolystyrene carbonate resin (mp-
carbonate, Argonaut Technologies) (1.83 g, 5.23 mmol, of 2.85
meq/g) was added in one portion and the reaction was stirred
for an additional 3 h. The resin was removed by vacuum
filtration and washed with CH2Cl2, and the filtrate was
concentrated to yield a viscous, yellow oil. Chromatography
(10:1:0.1 EtOAc/hexanes/Et3N) afforded 358 mg (93%) of the
1
1.10, CHCl3); H NMR (CDCl3, 400 MHz) δ 4.90 (dd, J ) 3.3,
5.1 Hz, 1H), 4.85 (d, J ) 5.1 Hz, 1H), 4.41 (dd, J ) 3.3, 9.9 Hz,
1H), 4.07 (dd, J ) 2.4, 10.8 Hz, 1H), 3,87 (dd, J ) 5.7, 10.8
Hz, 1H), 3.76 (ddd, J ) 2.4, 5.5, 9.9 Hz, 1H) 1.49 (s, 3H), 1.44
(s, 3H), 0.92 (s, 9H), 0.11 (app d, J ) 1.4 Hz, 6H); 13C NMR
(CDCl3, 100 MHz) δ 173.2, 114.3, 75.9, 75.8, 75.4, 63.1, 60.4,
26.8, 25.9, 25.7, 18.2, -5.6, -5.7; IR (neat) 2105 (s), 1796 (s)
cm-1; HRMS (ES) calcd for C15H27N3NaO5Si (M + Na) 380.1618,
found 380.1616. Anal. Calcd for C15H27N3O5Si: C, 50.40; H,
7.61; N, 11.75. Found: C, 50.31; H, 7.60; N, 11.86.
title compound as an amorphous white solid. 1H NMR and 13
C
NMR spectral data for 21 matched published reports.19b
(3aS,3bS,6bR,10aR,10bS)-2,2,9,9-Tetramethyl-1,3,8,10-
tetraoxacyclohexa-[e]cyclopenta[g]-5,6-dione-4-azain-
dolizidine (22). By an alternate method of preparation for
this known compound,19b Dess-Martin periodinane (31 mg,
0.072 mmol) was added to a solution of alcohol 21 (16 mg, 0.048
mmol) in CH2Cl2 (1.0 mL) at room temperature and stirred
for 4 h. The reaction was then diluted with ether (4 mL),
filtered through Celite and concentrated to yield the interme-
diate aldehyde as a clear oil. The crude intermediate aldehyde
was dissolved in 7.0 N NH3-MeOH (0.70 mL, 4.9 mmol NH3)
and stirred at room temperature for 24 h. The reaction was
then concentrated to yield a yellow oil. Chromatography (75%
EtOAc/hexanes) afforded 13 mg (89%) of the title compound
(2R,3R,4R,5R)-5-Azido-6-tert-butyldimethylsilanyloxy-
2,3-isopropylidene-dioxy-1,4-hexanediol (18). The title
compound was prepared by the method of Fairbanks for the
reduction of azide-containing γ-lactones to the corresponding
diol.27 Sodium borohydride (1.95 g, 51.5 mmol) was added to
a solution of azide-containing γ-lactone 13 (9.20 g, 25.8 mmol)
in ethanol (400 mL) at room temperature and stirred for 4 h.
J. Org. Chem, Vol. 70, No. 24, 2005 9901

Hering et al.
1
as a white solid. H NMR and 13C NMR spectral data for 22
hexahydroimidazo[1,2-a]pyridine-2,3-dione (43). A solution
of benzyl ether 30e (49 mg, 0.068 mmol) in MeOH (1.4 mL)
was transferred to a Parr bottle and 10% palladium on carbon
(10 mg) was added to the solution. The Parr bottle was
assembled onto a Parr Shaker, evacuated via an aspirator, and
purged first with nitrogen (2×) and then hydrogen (2×).
Hydrogenation was carried out under 45 psi of hydrogen at
room temperature with vigorous shaking. After 48 h, analysis
of the reaction by LCMS indicated complete removal of the
benzyl ether so the hydrogen was evacuated, the mixture
filtered through Celite and concentrated to obtain a clear oil.
The residue was dissolved in dry MeOH (1.4 mL) and ethylene
glycol (28 µL, 32 mg, 0.51 mmol) and p-toluenesulfonic acid
(13 mg, 0.068 mmol) were added. The reaction was stirred at
reflux for 2 h, at which point analysis by LCMS indicated that
all of the acetonides had been removed. Therefore, the reaction
was cooled to room temperature, quenched with pyridine (30
µL) and concentrated to give a pale yellow oil. The oil was
dissolved in pyridine (0.7 mL) and acetic anhydride (0.7 mL)
and stirred at room temperature for 16 h. The reaction was
then concentrated to yield a pale yellow oil. Normal phase
HPLC (95% EtOAc/hexanes, 10 mL/min, tR ) 20.4 min)
afforded 31 mg (59%) of the title compound as a clear oil. Data
matched published reports.19b
(2R,3R,4R,5R)-N-(2,3:4,6-Diisopropylidenedioxyhexanol-
5-yl)-2′-chloro-acetamide (29). Chloroacetyl chloride (249
µL, 4.0 M solution in acetone, 0.998 mmol) was added dropwise
with stirring to a solution of amino-alcohol 11 (261 mg, 0.998
mmol) and sodium acetate (164 mg, 2.00 mmol) in a mixture
of 1.4 mL of acetone and 0.7 mL of water at 0 °C. The reaction
was allowed to slowly come to room temperature and stirred
for 22 h, at which point it was shown to be incomplete by TLC.
The reaction was cooled to 0 °C, chloroacetyl chloride (125 µL,
4.0 M solution in acetone, 0.499 mmol) was added dropwise,
and the resulting mixture was then allowed to come to room
temperature. After stirring for 1 h the reaction was concen-
trated, and the residue was diluted with water and extracted
with CH2Cl2. The combined organic phases were washed with
brine, dried (MgSO4) and concentrated to yield 280 mg (83%)
of the title compound as an amorphous white solid. The crude
material was sufficiently pure for characterization. Data for
29: Rf ) 0.61 (10% MeOH/CH2Cl2); [R]23 ) -58.7° (c 1.03,
D
CH2Cl2); 1H NMR (benzene-d6, 400 MHz) δ 6.26 (d, J ) 8.4
Hz, 1H), 4.33-4.24 (m, 1H), 4.22 (d, J ) 6.6 Hz, 1H), 4.14-
4.09 (m, 1H), 3.89 (d, J ) 9.5 Hz, 1H), 3.81-3.69, (m, 3H),
3.56 (dd, J ) 8.1, 11.7 Hz, 1H), 3.45 (s, 2H), 1.60 (s, 3H), 1.33
(s, 6H),1.28 (s, 3H); 13C NMR (benzene-d6, 100 MHz) δ 166.0,
109.5, 99.8, 78.3, 76.1, 70.3, 62.4, 62.0, 47.4, 42.9, 28.1, 27.4,
26.3, 20.6; IR (neat) 3286 (m), 1665 (s) cm-1; MS (EI) m/z (rel
int) 322 (M - CH3, 15), 264 (27), 148 (58), 119 (100); HRMS
(ES) calcd for C14ClH24NaNO6 (M + Na) 360.1190, found
360.1191.
for 43: Rf ) 0.39 (100% EtOAc); [R]23 ) 13.1° (c 0.52, CH2-
D
1
Cl2); H NMR (CDCl3, 400 MHz) δ 6.70 (d, J ) 9.1 Hz, 1H),
5.53 (t, J ) 3.1 Hz, 1H), 5.42 (d, J ) 9.5 Hz, 1H), 5.31 (dd, J
) 4.4, 9.5 Hz, 1H), 5.11-5.01 (m, 3H), 4.77 (dd, J ) 5.5, 9.6
Hz, 1H), 4.68 (d, J ) 1.4 Hz, 1H), 4.60-4.53 (m, 2H), 4.33 (ABq,
J ) 16.5, ∆ν ) 65.3 Hz, 2H), 4.33-4.22 (m, 2H), 4.06 (dd, J )
2.5, 12.1 Hz, 1H), 3.98 (ddd, J ) 2.5, 5.9, 9.2 Hz, 1H), 3.41 (s,
3H), 2.17 (s, 3H), 2.13 (s, 3H), 2.12 (s, 3H), 2.08 (s, 3H), 2.07
(s, 3H), 2.04 (s, 3H), 2.02 (s, 3H); 13C NMR (CDCl3, 100 MHz)
δ 170.8, 169.8, 169.8, 169.0, 168.8, 168.5, 166.6, 158.3, 156.6,
99.6, 72.0, 69.2, 68.0, 67.2, 67.0, 66.3, 64.1, 62.7, 60.4, 55.3,
52.4, 50.6, 45.8, 20.7 (2 carbons), 20.6 (4 carbons), 20.6; IR
(3aS,3bS,6bR,10aR,10bS)-2,2,9,9-Tetramethyl-1,3,8,10-
tetraoxacyclohexa-[e]cyclopenta[g]-6-one-4-azaindolizi-
dine (28a). Dimethyl sulfoxide (11 µL, 12 mg, 0.16 mmol) was
added with stirring to a solution of oxalyl chloride (39 µL, 2.0
M solution in CH2Cl2, 0.078 mmol) in CH2Cl2 (1.0 mL) at -78
°C. After 30 min, a solution of alcohol 29 (24 mg, 0.071 mmol)
in CH2Cl2 (0.4 mL) was added dropwise and the reaction
stirred at -78 °C for an additional 30 min. Triethylamine (50
µL, 36 mg, 0.36 mmol) was then added at -78 °C, and the
reaction was stirred for a further 20 min, warmed to room
temperature and concentrated to yield an intermediate alde-
hyde/aminal as an oily, pale yellow solid. The crude material
was unstable to chromatography, although the presence of a
1:2 mixture of the aldehyde and aminal was verified by
resonances of δ 9.64 (d, J ) 1.6 Hz, 0.33H) and δ 5.52 (br s,
0.67H), respectively, in the 1H NMR spectrum (CDCl3, 400
MHz). The crude residue was treated with 7.0 N NH3-MeOH
(1.0 mL, 7.0 mmol NH3) and stirred for 20 h at room
temperature. The reaction was concentrated to yield an oily,
brown solid. Chromatography (2% MeOH/CH2Cl2) afforded 17
mg (79%) of the title compound as a white solid. Data for 28a:
mp ) 190-192 °C; Rf ) 0.18 (100% EtOAc); [R]23D ) -71.4° (c
) 0.74, CH2Cl2); 1H NMR (CDCl3, 400 MHz) δ 4.64 (d, J ) 8.6
Hz, 1H), 4.53 (dd, J ) 5.1, 10.8 Hz, 1H), 4.26 (t, J ) 8.6 Hz,
1H), 4.06 (dd, J ) 8.6, 12.3 Hz, 1H), 4.04 (t, J ) 8.6 Hz, 1H),
3.71 (t, J ) 10.8 Hz, 1H), 3.54 (ABq, J ) 16.5, ∆ν ) 87.1 Hz,
2H), 3.46-3.39 (m, 1H), 1.55 (s, 3H), 1.53 (s, 3H), 1.48 (s, 3H),
1.36 (s, 3H); 13C NMR (CDCl3, 100 MHz) δ 175.2, 110.7, 99.8,
77.5, 75.9, 72.9, 71.1, 62.2, 48.1, 47.9, 29.1, 26.5, 23.9, 19.0;
IR (neat) 3338 (w), 1695 (s) cm-1; MS (EI) m/z (rel int) 298
(M+, 14), 283 (24), 240 (39), 110 (100), 85 (37); HRMS (ES)
calcd for C15H26N2NaO6 (M + Na + MeOH) 353.1689, found
353.1697. Stereochemical assignment for 28a was determined
on the basis of COSY analysis and comparison of the observed
J-coupling between H8 (4.04 ppm, t, 8.6 Hz) and H8a (4.64 ppm,
d, 8.6 Hz) to those reported in the literature for the closely
related structures of kifunensine diacetonide (J-coupling
between H8 and H8a of 8.0 Hz) and 8a-epi kifunensine di-
acetonide (J-coupling between H8 and H8a of 3.0 Hz).18,19b
(neat) 3334 (br w), 1751 (vs), 1697 (m), 1223 (vs), 1050 (s) cm-1
;
MS (EI) m/z (rel int) 728 (M - OMe, <1), 640 (4), 566 (19),
524 (67), 242 (17), 43 (100); HRMS (ES) calcd for C31H41N3-
NaO19 (M + Na) 782.2232, found 782.2195.
(5R,6R,7S,8S,8aS)-N-(6,7,8-Triacetoxy-5-acetoxymethyl-
1-{[1′S,2′S,3′R,4′S,5′R]-3′,4′-diacetoxy-5′-acetoxymethyl-
1′-methoxytetrahydropyran-2′-ylpropion-amide-3′′-yl})-
hexahydroimidazo[1,2-a]pyridine-2,3-dione (44). 20%
palladium hydroxide on carbon (13 mg) was added to a solution
of benzyl ether 30f (50 mg, 0.068 mmol) and cyclohexene (0.35
mL) in EtOH (1.4 mL) and refluxed with stirring for 4 h. At
this point, analysis of the reaction by LCMS indicated complete
removal of the benzyl ether so it was cooled to room temper-
ature, filtered through Celite and concentrated to obtain a
white oil. The residue was dissolved in dry MeOH (1.4 mL),
and ethylene glycol (19 µL, 21 mg, 0.34 mmol) and p-
toluenesulfonic acid (13 mg, 0.068 mmol) were added. The
reaction was stirred at reflux for 2 h at which point analysis
by LCMS indicated that all of the acetonides had been
removed. Therefore, the reaction was cooled to room temper-
ature, quenched with pyridine (20 µL), and concentrated to
give a clear oil. The oil was dissolved in pyridine (0.7 mL) and
acetic anhydride (0.7 mL) and stirred at room temperature
for 16 h. The reaction was then concentrated to yield a clear
oil. Normal phase HPLC (95% EtOAc/hexanes, 10 mL/min, tR
) 29.2 min) afforded 19 mg (36%) of the title compound as a
clear oil. Data for 44: Rf ) 0.35 (100% EtOAc); [R]23D ) 19.7°
(c 0.94, CH2Cl2); 1H NMR (CDCl3, 500 MHz) δ 6.34 (d, J ) 8.8
Hz, 1H), 5.48 (t, J ) 3.1 Hz, 1H), 5.34-5.28 (m, 2H), 5.13-
5.06 (m, 2H), 5.00 (dd, J ) 3.1, 9.2 Hz, 1H), 4.75 (dd, J ) 5.4,
9.6 Hz, 1H), 4.68 (d, J ) 1.6 Hz, 1H), 4.62 (dd, J ) 9.5, 11.5
Hz, 1H), 4.58 (ddd, J ) 1.6, 4.6, 9.0 Hz, 1H), 4.28-4.20 (m,
2H), 4.08 (dd, J ) 2.4, 12.2 Hz, 1H), 3.98-3.90 (m, 1H), 3.76
(ddd, J ) 7.0, 8.5, 14.1 Hz, 1H), 3.40 (s, 3H), 2.22 (s, 3H), 2.13
(s, 3H), 2.12 (s, 3H), 2.11 (s, 3H), 2.05 (s, 3H), 2.04 (s, 3H),
1.98 (s, 3H); 13C NMR (CDCl3, 100 MHz) δ 170.7, 170.7, 169.8,
169.8, 169.7, 169.1, 168.9, 168.7, 158.1, 157.3, 99.7, 72.3, 69.2,
(5R,6R,7S,8S,8aS)-N-(6,7,8-Triacetoxy-5-acetoxymethyl-
1-{[1′S,2′S,3′R,4′S,5′R]-3′,4′-diacetoxy-5′-acetoxymethyl-
1′-methoxytetrahydropyran-2′-ylacetamide-2′′-yl})-
9902 J. Org. Chem., Vol. 70, No. 24, 2005

Synthesis of Kifunensine Analogues
68.0, 67.3, 67.1, 66.0, 63.7, 62.6, 60.3, 55.2, 52.4, 50.4, 39.7,
33.8, 20.7 (2 carbons), 20.7 (2 carbons), 20.6 (3 carbons); IR
dione Chloride (47). A solution of 46 (44 mg, 0.085 mmol)
in THF (0.85 mL) was treated with 1 N HCl (0.85 mL) and
stirred at room temperature for 4 h. The reaction was then
concentrated to yield 31 mg (98%) of the title compound as a
yellow oil. The crude material was sufficiently pure for
(neat) 3346 (w), 1750 (s), 1679 (m), 1223 (s), 1052 (m) cm-1
;
MS (EI) m/z (rel int) 742 (M - OMe, <1), 654 (1), 538 (5), 455
(6), 113 (29), 55 (39), 43 (100); HRMS (ES) calcd for C32H43N3-
NaO19 (M + Na) 796.2388, found 796.2394.
characterization. Data for 47: Rf ) 0.00 (10% MeOH/CH2Cl2);
1
[R]23 ) 22.3° (c 0.92, MeOH); H NMR (CD3OD, 500 MHz) δ
D
(5R,6R,7S,8S,8aS)-N-(6,7,8-Triacetoxy-5-acetoxymethyl-
1-{[1′S,2′S,3′R,4′S,5′R]-3′,4′-diacetoxy-5′-acetoxymethyl-
1′-methoxytetrahydropyran-2′-ylbutyr-amide-4′′-yl})-
hexahydroimidazo[1,2-a]pyridine-2,3-dione (45). 20%
palladium hydroxide on carbon (15 mg) was added to a solution
of benzyl ether 30g (30 mg, 0.040 mmol) and cyclohexene (0.15
mL) in EtOH (0.60 mL) and refluxed with stirring for 2 h. At
this point, analysis of the reaction by LCMS indicated complete
removal of the benzyl ether so it was cooled to room temper-
ature, filtered through Celite and concentrated to obtain a
clear oil. The residue was dissolved in dry MeOH (0.8 mL),
and ethylene glycol (11 µL, 12 mg, 0.20 mmol) and p-
toluenesulfonic acid (8 mg, 0.04 mmol) were added. The
reaction was stirred at reflux for 9 h at which point analysis
by LCMS indicated that all of the acetonides had been
removed. Therefore, the reaction was cooled to room temper-
ature, quenched with pyridine (20 µL) and concentrated to give
a pale yellow oil. The oil was dissolved in pyridine (0.4 mL)
and acetic anhydride (0.4 mL) and stirred at room temperature
for 16 h. The reaction was then concentrated to yield a pale
yellow oil. Normal phase HPLC (100% EtOAc, 10 mL/min, tR
) 38.0 min) afforded 12 mg (39%) of the title compound as a
clear oil. Data for 45: Rf ) 0.22 (100% EtOAc); [R]23D ) 13.2°
(c 0.60, CH2Cl2); 1H NMR (CDCl3, 500 MHz) δ 6.45 (d, J ) 9.0
Hz, 1H), 5.49 (t, J ) 3.4 Hz, 1H), 5.30 (dd, J ) 4.6, 10.2 Hz,
1H), 5.26 (d, J ) 9.2 Hz, 1H), 5.10 (t, J ) 10.2 Hz, 1H), 5.07
(d, J ) 3.6 Hz, 1H), 4.95 (dd, J ) 3.2, 9.3 Hz, 1H), 4.79-4.72
(m, 2H), 4.70 (d, J ) 1.0 Hz, 1H), 4.56 (ddd, J ) 1.4, 4.5, 9.0
Hz, 1H), 4.26-4.19 (m, 2H), 4.08-4.01 (m, 2H), 3.96 (ddd, J
) 2.4, 5.9, 10.0 Hz, 1H), 3.41 (s, 3H), 3.34 (dt, 4.8, 14.2 Hz,
1H), 2.36-2.17 (m, 3H), 2.22 (s, 3H), 2.11 (app d, J ) 3.9 Hz,
6H), 2.10 (app d, J ) 3.1 Hz, 6H), 2.03 (s, 3H), 2.01-1.93 (m,
1H), 1.96 (s, 3H); 13C NMR (CDCl3, 100 MHz) δ 171.9, 171.8,
170.6, 169.8, 169.7, 168.9, 168.8, 168.6, 158.2, 157.1, 99.9, 72.4,
69.3, 67.9, 67.2, 66.2, 62.7, 62.4, 60.3, 55.2, 52.7, 50.2, 42.1,
33.3, 29.7, 23.5, 20.7 (3 carbons), 20.6 (3 carbons), 20.6; IR
7.78 (s, 1H), 4.99 (d, J ) 8.8 Hz, 1H), 4.40 (dd, J ) 4.6, 8.8
Hz, 2H), 4.02 (d, J ) 3.4 Hz, 1H), 3.98 (dd, J ) 9.0, 11.7 Hz,
1H), 3.92 (t, J ) 3.1 Hz, 1H), 3.83-3.70 (m, 3H), 3.67 (dd, J )
3.1, 9.0 Hz, 1H), 2.97-2.90 (m, 2H), 1.86-1.64 (m, 4H), 1.50-
1.35 (m, 4H); 13C NMR (CD3OD, 100 MHz) δ 160.9, 160.5, 73.6,
73.6, 70.5, 67.8, 61.8, 60.2, 44.1, 40.8, 28.5, 28.5, 27.3, 26.9;
IR (MeOH) 1730 (s) cm-1; MS (ES) m/z (rel int) 332 (M + H,
100); HRMS (ES) calcd for C14H26N3O6 (M + H) 332.1822,
found 332.1819.
(5R,6R,7S,8S,8aS)-1-(6′-Acetylaminohexyl)-6,7,8-trihy-
droxy-5-hydroxymethyl-hexahydroimidazo[1,2-a]pyri-
dine-2,3-dione (49). Compound 48 (24 mg, 0.044 mmol) was
treated with 2 N NH3-MeOH (0.90 mL, 1.8 mmol NH3) and
stirred at 40 °C for 1 h. The reaction was then concentrated
to yield a clear oil. Chromatography (20% MeOH/CH2Cl2)
afforded 11 mg (64%) of the title compound as a clear oil. Data
for 49: Rf ) 0.07 (20% MeOH/CH2Cl2); 1H NMR (CD3OD, 400
MHz) δ 4.97 (d, J ) 9.2 Hz, 1H), 4.40 (dd, J ) 5.0, 8.6 Hz,
1H), 4.01 (dd, J ) 1.1, 3.6 Hz, 1H), 3.97 (dd, J ) 9.0, 11.9 Hz,
1H), 3.91 (t, J ) 3.1 Hz, 1H), 3.86-3.68 (m, 1H), 3.66 (dd, J )
3.3, 9.2 Hz, 1H), 3.15 (t, J ) 7.0 Hz, 2H), 1.91 (s, 3H), 1.84-
1.64 (m, 2H), 1.55-1.45 (m, 2H), 1.43-1.32 (m, 4H); 13C NMR
(CD3OD, 100 MHz) δ 173.2, 160.5, 160.3, 73.6, 73.5, 73.4, 67.6,
67.6, 61.7, 60.0, 44.1, 40.3, 30.2, 28.6, 27.4, 22.5; IR (neat) 3296
(br m), 1732 (s) cm-1; MS (EI) m/z (rel int) 373 (M+, 2), 355
(9), 337 (40), 306 (100); HRMS (ES) calcd for C16H27N3NaO7
(M + Na) 396.1747, found 396.1746.
Anhydrous Coupling of (5R,6R,7S,8S,8aS)-1-(6′-Am
moniohexyl)-6,7,8-trihydroxy-5-hydroxymethyl-hexa-
hydroimidazo[1,2-a]pyridine-2,3-dione Chloride (47) to
Affi-Gel 10 (10). Compound 10 was prepared using the
anhydrous coupling protocol of Affi-Gel 10 (Bio-Rad) described
in the manufacturer’s manual.48 A solution of the hydrochloride
salt 47 (51 mg, 0.14 mmol) and triethylamine (0.5 mL) in
MeOH (5 mL) was added to a slurry of Affi-Gel 10 resin (8.30
mL settled resin volume, 0.125 mmol, prewashed with 6 bed
volumes of cold iPrOH and 6 bed volumes of cold MeOH) in
MeOH (20 mL) and agitated on a shaker at 4 °C for 16 h. The
gel was filtered and washed with cold MeOH (6 bed volumes)
and cold water (6 bed volumes), and the combined filtrates
were concentrated. The amount of 47 in the combined filtrates
was determined to be approximately 0.013 mmol by 1H NMR
using dimethyl acetemide as an internal standard, correspond-
ing to a loading of 0.015 mmol 47 per mL of gel. The gel was
added to a solution of ethanolamine (6.25 mL, 0.1 M solution
in water, 0.625 mmol) in cold water (25 mL) and agitated on
a shaker at 4 °C for 4 h. The gel was then filtered and washed
with cold water (6 bed volumes) and cold 0.2% aqueous NaN3
(3 bed volumes) and stored at 0 °C in a solution of 0.2%
aqueous NaN3 (inhibits bacterial growth).
Assay of R1,2-Mannosidase Inhibition. The inhibitory
activity of each compound was assayed at three concentrations
(0.05, 0.5, and 5 mM) using recombinant human ER mannosi-
dase I5b or recombinant mouse Golgi Man IA.5b ER Man I
assays were performed using 20 µg of enzyme and Man9-
GlcNAc2-PA (20 µM) as substrate in a final volume of 20 µL
containing 20 mM MES/NaOH, pH 7.0, 150 mM NaCl, and 5
mM CaCl2. The enzyme reactions were allowed to proceed for
15 min at 37 °C and were stopped by addition of 20 µL of 1.25
M Tris-HCl pH 7.6. The human ER Man I product, Man8-
GlcNAc2-PA, was resolved from Man9GlcNAc2-PA by HPLC on
a Hypersil APS-2 NH2 column.50 Man9GlcNAc2-PA was pre-
(neat) 3358 (w), 1750 (s), 1680 (m), 1224 (s), 1051 (s) cm-1
;
MS (EI) m/z (rel int) 788 (M + H, 1), 756 (4), 527 (18), 469
(22), 242 (14), 84 (15), 43 (100); HRMS (ES) calcd for C33H45N3-
NaO19 (M + Na) 810.2545, found 810.2548.
(5R,6R,7S,8S,8aS)-N-(6,7,8-Trihydroxy-5-hydroxy-
methyl-1-{[1′S,2′S,3′R,4′S,5′R]-3′,4′-dihydroxy-5′-hydroxy-
methyl-1′-methoxytetrahydropyran-2′-ylacet-amide-
2′′-yl})-hexahydroimidazo[1,2-a]pyridine-2,3-dione (7).
Sodium methoxide powder (15 mg, 0.34 mmol) was added to
a solution of compound 43 (26 mg, 0.034 mmol) in MeOH (0.7
mL) and stirred at room temperature for 30 min. The reaction
was then passed over a short column of Dowex 50Wx8-100
resin (pyridinium form), the product was eluted with MeOH
and concentrated to yield 13 mg (84%) of the title compound
as a clear oil. Data for 7: Rf ) 0.01 (30% MeOH/CH2Cl2); [R]23
D
1
) 31.4° (c 0.67, MeOH); H NMR (CD3OD, 400 MHz) δ 8.18
(d, J ) 9.2 Hz, 1H), 5.03 (d, J ) 9.1 Hz, 1H), 4.63 (d, J ) 1.1
Hz, 1H), 4.56 (ABq, J ) 16.7, ∆ν ) 18.8 Hz, 2H), 4.43 (dd, J
) 5.0, 8.7 Hz, 1H), 4.35 (dd, J ) 1.1, 4.8 Hz, 1H), 4.03 (d, J )
2.9 Hz, 1H), 3.98 (dd, J ) 8.7, 11.9 Hz, 1H), 3.93-3.89 (m,
2H), 3.82 (d, J ) 3.3 Hz, 2H), 4.76-3.70 (m, 2H), 3.57 (t, J )
9.7 Hz, 1H), 3.52 (dt, J ) 3.3, 9.7 Hz, 1H), 3.37 (s, 3H); 13C
NMR (CD3OD, 100 MHz) δ 170.0, 161.1, 159.8, 101.6, 74.0,
73.4, 73.3, 70.8, 70.4, 68.4, 68.1, 62.3, 61.7, 60.2, 55.3, 54.4,
46.2; IR (MeOH) 1739 (vs), 1673 (m), 1063 (s) cm-1; MS (EI)
m/z (rel int) 429 (M - 2 H2O, <1), 339 (5), 137 (31), 59 (100);
HRMS (ES) calcd for C17H27N3NaO12 (M + Na) 488.1492, found
488.1482.
(5R,6R,7S,8S,8aS)-1-(6′-Ammoniohexyl)-6,7,8-trihydroxy-
5-hydroxymethyl-hexahydroimidazo[1,2-a]pyridine-2,3-
(50) Lal, A.; Pang, P.; Kalelkar, S.; Romero, P. A.; Herscovics, A.;
Moremen, K. W. Glycobiology 1998, 8, 981-995.
J. Org. Chem, Vol. 70, No. 24, 2005 9903

Hering et al.
pared by reductive pyridylamination of Man9GlcNAc2 isolated
fromcrudesoybeanagglutininextractedasdescribedpreviously.15b
Golgi Man IA assays were performed using 10 µg of enzyme
in a reaction containing 20 mM potassium phosphate, pH 6.0,
150 mM NaCl, 5 mM CaCl2, and 5 mM Man-R-1,2-Man-R-O-
CH3 in a total volume of 25 µL. Enzyme reactions were allowed
to proceed for 30 min at 37 °C and terminated by the addition
of 25 µL of 1.25 M Tris-HCl, pH 7.6. The amount of mannose
released was quantified using a glucose oxidase and peroxidase
reagent as previously described.50 The recombinant enzymes
were expressed and purified as previously described.5b For
determination of compound inhibition, enzyme activity was
plotted as a percentage of residual activity (compared to no
added compound) versus compound concentration and the IC50
values were estimated from the plots.
Acknowledgment. We thank Pharmacia Labora-
tories for a graduate fellowship for K.H. (2001-2002)
for funding this research. This work was supported by
National Institutes of Health Research Grants GM47533
and RR05351 (to K.W.M.).
Supporting Information Available: Experimental pro-
cedures for compounds 2-6, 8, 9, 12, 19, 20, 28b, 28c, 30a-
c, 30e-g, 32-35, 40-42, 46, and 48 and 1H spectra for all
new compounds. This material is available free of charge via
the Internet at http://pubs.acs.org.
JO0516382
9904 J. Org. Chem., Vol. 70, No. 24, 2005