Synthesis of 1-azagalactofagomine, a potent galactosidase inhibitor

Article abstract of DOI:10.1039/b007973l

1-Azagalactofagomine [(+)-(3R,4S,5R)-4,5-Dihydroxy-3-(hydroxymethyl)hexahydropyridazine, 2] was synthesised from achiral starting materials in a chemoenzymic synthesis. The racemic Diels-Alder adduct (2-hydroxymethyl-8-methyl-1,6,8-triazabicyclo[4.3.0]non-3-ene-7,9-dione, 5) from addition of pentadienol to 4-methyl-1,2,4-triazoline3,5-dione was resolved using lipase R-catalysed acetylation. The acetate [(S)-2-acetoxymethyl-8-methyl-1,6,8-triazabicyclo[4.3.0]non-3-ene-7,9-dione, 6] was saponified and treated with MCPBA to give a majority of the syn epoxide [(2R,3S,4R)-3,4-epoxy-2-hydroxymethyl-8-methyl-1,6,8-triazabicyclo[4.3.0]nonane- 7,9-dione, 7]. This isomer was subjected to epoxide opening with HI followed by a Woodward reaction-like displacement of the iodide with water and peracetylation to give an all-syn triacetate [(2R,3S,4R)-3,4-diacetoxy-2-acetoxymethyl-1,6,8-triazabicyclo[4.3.0]nonane-7,9-d ione, 11]. Finally deacetylation and hydrazinolysis gave 2. The pK_a of 2 was determined to be 5.7. 1-Azagalactofagomine was found to be a potent competitive galactosidase inhibitor. The inhibition constants, K_i, were 40, 300 and 7800 nM versus β-galactosidase from Aspergillus oryzae, Eschinchia coli and Saccharomyces fragilis, respectively, and 280 nM vs. α-galactosidase from green coffee beans.

Full text of DOI:10.1039/b007973l

View Article Online / Journal Homepage / Table of Contents for this issue
Synthesis of 1-azagalactofagomine, a potent galactosidase
inhibitor
Henrik Helligsø Jensen and Mikael Bols*
1
Department of Chemistry, University of Aarhus, Aarhus C, DK-8000, Denmark
Received (in Cambridge, UK) 3rd October 2000, Accepted 18th December 2000
First published as an Advance Article on the web 29th January 2001
1-Azagalactofagomine [(ϩ)-(3R,4S,5R)-4,5-Dihydroxy-3-(hydroxymethyl)hexahydropyridazine, 2] was synthesised
from achiral starting materials in a chemoenzymic synthesis. The racemic Diels–Alder adduct (2-hydroxymethyl-8-
methyl-1,6,8-triazabicyclo[4.3.0]non-3-ene-7,9-dione, 5) from addition of pentadienol to 4-methyl-1,2,4-triazoline-
3,5-dione was resolved using lipase R-catalysed acetylation. The acetate [(S)-2-acetoxymethyl-8-methyl-1,6,8-triaza-
bicyclo[4.3.0]non-3-ene-7,9-dione, 6] was saponified and treated with MCPBA to give a majority of the syn epoxide
[(2R,3S,4R)-3,4-epoxy-2-hydroxymethyl-8-methyl-1,6,8-triazabicyclo[4.3.0]nonane-7,9-dione, 7]. This isomer was
subjected to epoxide opening with HI followed by a Woodward reaction-like displacement of the iodide with water
and peracetylation to give an all-syn triacetate [(2R,3S,4R)-3,4-diacetoxy-2-acetoxymethyl-1,6,8-triazabicyclo[4.3.0]-
nonane-7,9-dione, 11]. Finally deacetylation and hydrazinolysis gave 2. The pKa of 2 was determined to be 5.7.
1-Azagalactofagomine was found to be a potent competitive galactosidase inhibitor. The inhibition constants, Ki,
were 40, 300 and 7800 nM versus β-galactosidase from Aspergillus oryzae, Eschinchia coli and Saccharomyces fragilis,
respectively, and 280 nM vs. α-galactosidase from green coffee beans.
Introduction
Results and discussion
Aza sugars and imino sugars are subject to intense current
interest.1–3 Some time ago it was found that a subtle change in
the classical imino sugar inhibitor of nojirimycin type, by
moving the nitrogen to the pseudo-anomeric position (the
position that corresponds to the anomeric position in a mono-
saccharide), gave a very potent class of glycosidase inhibitors
the so-called 1-aza sugars.4,5 A member of this class of com-
pounds is 1-azafagomine 1, a hydrazine, that inhibits both
α- and β-glucosidase strongly (Fig. 1).6,7 The reason for the
biological activity of 1 is perhaps that, in protonated form, it
mimics the transition states of α-glucoside (A) and β-glucoside
cleavage (B). In order to get more evidence for this theory it was
the subject of this study to investigate how general this idea was
and whether it could be extended to other glycosidases. Ideally
one would wish to investigate the bioactivity of the lyxo isomer
2 as it has been shown that 3 is a very potent β-galactosidase
inhibitor, while being weak against α-galactosidase.8 Nojirimy-
cin analogue 4, on the other hand, has the opposite inhibitory
profile (Fig. 1).9 Thus if 1 was mimicking both transition states
one would expect 2 to be potent against either type of galacto-
sidase. In this paper we report the synthesis of 2 and report that
it is indeed a potent competitive inhibitor of both α- and
β-galactosidase.
Optically pure 1 can be synthesised both from chiral pool7 or
achiral starting materials.10 However, extension of these
methods to the synthesis of 2 was not trivial. In order to obtain
2 by the chiral pool synthesis it would require that expensive
-ribose be used as starting material. The chemoenzymic
synthesis would require that the double bond in 5 be cis-
dihydroxylated syn to the α-hydroxymethyl group (Scheme 1).
Scheme 1 Enzymic resolution of alkene 5.
However OsO4-catalysed dihydroxylation was known to give
exclusively the anti product.6
As a consequence we searched for a procedure that would
lead to cis-dihydroxylation syn to the α-hydroxymethyl group.11
One of the few available methods was that of Woodward and
Brutcher, in which an alkene is treated with I2 and AgOAc in
wet acetic acid.12 This results in formation of a iodo acetate C,
with iodine at the less hindered face, which subsequently forms
an acetoxonium ion D (Fig. 2). Spontaneous hydrolysis of this
ion by the water present gave a partially acetylated compound
Fig.
1
Structure of azafagomine (1), azagalactofagomine (2),
isogalactofagomine (3) and galactostatin (4).
DOI: 10.1039/b007973l
J. Chem. Soc., Perkin Trans. 1, 2001, 905–909
This journal is © The Royal Society of Chemistry 2001
905

View Article Online
Table 1 Results from various reaction conditions employed in the
transformation of iodides 9 and 10 to triacetates 11 and 12 (Scheme 3)
%
Water
Ratio
11/12
Entry
Reagent
Solvent
Time/d
1
2
3
4
5
AgOAc
AcOH
AcOH
AcOH
AcOH
6
6
6
50
6
2
2
2
1
4
70 : 30
70 : 30
91 : 9
87 : 13
96 : 4
Fig. 2 Some of the intermediates from the Woodward reaction: C
shows one of the iodides expected to be formed by addition of AgOAc
and I2 to an alkene, D shows the subsequent aceteoxonium ion and E
one of the monoacetates formed from hydrolysis of D.
AgOSO2CF3
AgO2CCF3
AgO2CCF3
AgO2CCF3
CF3CO2H
E, which by deacetylation gave the syn-diol at the more
hindered face.12
isomers 9 and 10 was determined from a COSY spectrum.
The major isomer 10 has a low-field (δ 5.17) quartet
from a proton (H-4) that couples with the two protons on
C-5 and H-3. Therefore this proton must be next to a acetoxy
group.
Now treatment of the mixture of iodo acetates 9 and 10 with
AgOAc in acetic acid containing 6% of water gave a mixture of
partially acetylated compounds that were immediately acetyl-
ated with acetic anhydride and Et3N to give a 7 : 3 mixture of
the triacetates 11 and 12 (Scheme 3, Table 1). Not only was the
When the Woodward and Brutcher protocol was attempted
on substrate 5 no significant reaction was observed. Presumably
the alkene is too unreactive, as shown by the previous observ-
ation that epoxidation of 5 can only occur with the most
reactive reagents or under forcing conditions.6 However the
idea was conceived that if the intermediate iodo acetate C could
be obtained by a different route the acetoxonium chemistry
could be relied upon to give the desired product E. An obvious
source for iodo acetate C was the epoxide 7, a previously
unwanted by-product from the epoxidation of 5.10 It was
expected, based on previous work,6 that epoxidation of 5 with
m-chloroperbenzoic acid (MCPBA) would give 7 as the major
product.
Thus ( )-510 was subjected to enzymic resolution by
acetylation with lipase R and vinyl acetate to give acetate 6 and
alcohol (ϩ)-5 (Scheme 1). This was essentially performed as
previously described10 except that only one enzyme was used.
Acetate 6 and alcohol (ϩ)-5 were obtained in 88% and 92% ee,
respectively. After deacetylation of 6 with NaOMe–MeOH and
recrystallisation the optically pure alcohol (Ϫ)-5 was obtained.
Now treatment of this alkene with MCPBA at 80 ЊC in
(ClCH2)2 for 18 h, necessary to promote reaction, gave a mix-
ture of the epoxides 7 and 8 in the ratio 2 : 1 (Scheme 2). These
Scheme 3 Opening of epoxide 7 with HI and subsequent solvolysis.
stereoselectivity in this reaction disappointing, but it was also
remarkable that the arabino isomer 12 was formed in relatively
large amounts. Formation of an arabino isomer can occur by
substitution, with retention of configuration, of iodide 10
(Scheme 4). Alternatively 12 could be imagined to be formed
from intermediate 13 by SN2 substitution with acetate. This is,
however, unlikely because if nucleophilic attack of 12 by acetate
were occurring, a xylo isomer should also be formed, and this
was not observed. The desired lyxo isomer 11, on the other
hand, could be formed both through the planned hydrolysis of
acetoxonium ion 13 and by direct substitution of 9 and 10 with
inversion of configuration (Scheme 4). The large amounts of
12 formed showed that nucleophilic substitution of the iodide
10 was a major reaction pathway. In order to suppress this
undesired pathway a series of different reaction conditions
were investigated (Table 1). We attempted to replace AgOAc
with AgO2CCF3 and AgOSO2CF3, which have less nucleophilic
counter-ions. As seen from Table 1 the least nucleophilic
reagent AgOSO2CF3 gave unchanged stereoselectivity, while
AgO2CCF3 actually improved the selectivity considerably
Scheme 2 Deacetylation and epoxidation of acetate 6.
two isomers were separated by chromatography and isolated
pure in 42% and 23% yield, respectively.
The epoxide 7 was now subjected to treatment with 57% aq.
HI in AcOH followed by in situ acetylation by addition of
Ac2O, which gave a mixture of the acetylated iodides 9 and 10
in the ratio 1 : 3 and a yield of 75%. The preferential opening in
the 3 position of epoxide 7 in this reaction is in contrast to the
opening of the corresponding anti epoxide 8, which is hydro-
lysed very selectively with attack at the 4 position.10 While
the selective hydrolysis of 8 undoubtedly is caused by steric
hindrance from the 2-hydroxymethyl group, steric reasons
cannot play a role in the preferential formation of iodide 10.
One explanation could be that geometrical constraints in
the bicyclic system favor 3R,4R-diaxial opening regardless of
the stereochemistry of the epoxide. The regiochemistry of the
906
J. Chem. Soc., Perkin Trans. 1, 2001, 905–909

View Article Online
towards 11. It was possible that, in these reactions, acetate ions
from the acetic acid medium were still acting as nucleophiles,
leading to formation of 12 by substitution with retention. It
was therefore decided to investigate the medium as well. First
the water content was investigated. Woodward and Brutcher
used as little as 1 mol equiv. of water in their procedure;12 how-
ever, decreasing the water content in this case decreased the
reaction rate too much to be practical. Increasing the water
content to as much as 50% increased the reaction rate, and gave
virtually unchanged stereoselectivity (Table 1). Finally, acetic
acid was substituted with trifluoroacetic acid (TFA). As seen
from Table 1, when AgO2CCF3 in TFA containing 6% water
was used the side reaction was suppressed giving a stereo-
selectivity of 96 : 4 and a 79% yield (Scheme 3). These experi-
ments show that a large fraction of 12 is formed by attack of
acetate ions, and removing acetate improves stereoselectivity
tremendously. However, since a small amount of 12 is obtained
even when no acetate is present it appears that water substitutes
the iodide with retention to some extent.
Compounds 11 and 12 could not be separated. Therefore 11,
containing 4% 12, was deacetylated with NaOMe and the
product subjected to hydrazinolysis with aq. NH2NH2 at
100 ЊC. This gave a product that could be purified to give the
target compound 2 stereochemically pure in 56% yield over the
two steps (Scheme 3).
The basicity of 2 was determined by measuring the pKa
of the compound. The titration curve for titration of the hydro-
chloride of 2 with NaOH is shown in Fig. 3. From the curve the
pKa was found to be 5.7. As this value was in total contrast to
the pKa-value of the arabino-isomer 1, which was previously
found to be 3.9,6 it was decided to measure the pKa of 1 again.
A titration curve for the titration of the hydrochloride of 1 is
shown in Fig. 4. From the curve the pKa of 1 was found to be
5.3, which means that our previous determination of pKa was
incorrect. To confirm this, NMR spectra of 1 were measured at
different pH-values. They showed that 1 was 71% protonated
at pH 4.8 and 15% protonated at pH 5.5. Therefore 1 should
be approximately 50% protonated at pH 5.3, and this thus
confirmed the new pKa-value.
Azagalactofagomine 2 was tested for inhibition of a series
of glycosidases. Not surprisingly, 2 was a poor inhibitor of
α-glucosidase, but a strong inhibitor of β-glucosidase and a
series of galactosidases. It is noteworthy that compound 2 is a
slightly weaker inhibitor of β-galactosidase and β-glucosidase
than is isogalactofagomine 3, but more potent than galacto-
deoxynojirimycin 4. In contrast, 2 is a stronger inhibitor of
α-galactosidase than is 3, but weaker than 4 (Table 2). These
observations are similar to what has been found for the corre-
sponding arabino/gluco-isomers of 2, 3 and 4,6 and as such sup-
port the transition-state theory outlined in the Introduction and
Fig. 1. One may speculate as to why 2 is a significantly weaker
inhibitor of α-galactosidase than is 4 and the explanation is
Fig. 3 Titration curve for 2. Titration performed with 0.099 M NaOH
on 2 in water, with 1.13 M HCl added. pKa determined to 5.7.
Fig. 4 Titration curve for 1. Titration performed with 0.099 M NaOH
on 6.5 mg 1 in 5 ml of water, with 0.2 ml of 1.13 M HCl added. pKa
determined to be 5.3.
Scheme 4 Possible reaction routes in the transformation of iodides 9
and 10 to triacetates 11 and 12.
Table 2 Ki-values in µM for inhibition of glycosidases by 2, 3 and 4 at pH 6.8 and 25 ЊC
Enzyme
α-Glucosidase (baker’s yeast)
β-Glucosidase (almonds)
β-Galactosidase (Asperigillus Oryzae)
β-Galactosidase (E. coli)
β-Galactosidase (Saccharomyces fragilis)
α-Galactosidase (green coffee beans)
570
0.13
0.04
0.30
7.8
>2000a
0.097b
540e
12.5e
0.004c
0.2d
0.33
81
0.28
50
0.0016e
a An IC50-value taken from ref. 5. b Taken from ref. 13. c Taken from ref. 5. d Measured on racemic inhibitor. e Taken from ref. 9.
J. Chem. Soc., Perkin Trans. 1, 2001, 905–909
907

View Article Online
probably that 2 lacks a (carbohydrate) C-2 hydroxy group. It is
more potent than 3 against this enzyme because it has a second
ring nitrogen atom. The reason why 2 is more potent against the
β-glycosidases than is 4 is presumably due to the presence and
lack of an anomeric N-atom, respectively. The fact that 2 is
slightly weaker than 3 towards the β-glycosidases is probably
explained by the lower basicity of 2.
In this paper we have reported synthesis of the potent galacto-
sidase inhibitor 2 for the first time. A new modified version of
Woodward’s reaction was used in this synthesis to introduce
the lyxo stereochemistry. This method should be useful for
de novo synthesis of other galactose mimics.
Epoxide 7 (340 mg, 1.60 mmol) was dissolved in AcOH (4.5
mL) and aq. HI (57%; 0.42 mL, 3.2 mmol) was added at room
temperature. After 5 h all starting material had disappeared
(TLC control, AcOEt), Ac2O was added (5 mL), and the mix-
ture was stirred at room temperature overnight. To quench the
reaction water (10 mL) was carefully added and allowed to
react for 1 h. The reaction mixture was then extracted with
AcOEt (3 × 30 mL) and the combined organic phases were
washed with saturated aq. of NaHCO3 and saturated aq.
Na2SO3 (each 10 mL). After the solution had been dried over
anhydrous MgSO4, filtered and evaporated, the residue under-
went column chromatography (first AcOEt–pentane 1 : 2, then
AcOEt–pentane 1 : 1) which resulted in 516 mg (75%) of iodides
9 and 10 (Rf 0.38 in AcOEt–pentane 1 : 1) in the ratio 1 : 3.
The combination of iodides appeared as a colorless solid.
HRMS(ES) Calc. for C12H16N3O6I ϩ Na: m/z 447.9983. Found:
m/z, 447.9984; δH(CHCl3). 9: 5.12 (m, 1H, H-3), 4.17–4.67 (m,
5H, H-2Јa, H-2Јb, H-4, H-5a), 3.57 (dd, J 11.4, 13.6 Hz, 1H,
H-5b), 3.00 (s, 3H, NCH3), 2.13 and 1.95 (each s, 3H, CH3). 10:
5.17 (q, 1H, J 2.4 Hz, H-4), 4.54 (dt, J2,3 1.6, J2,2Ј 7.0 Hz, 1H,
H-2), 4.43 (br s, 1H, H-3), 4.33 (d, 2H, H2-2Ј), 3.95 (br dd, 1H,
H-5a), 3.86 (dd, J5a,5b 13.2 Hz, 1H, H-5b), 3.04 (s, 3H, NCH3),
2.05 and 1.98 (each s, 3H, CH3).
Experimental
General
1
13C NMR and H NMR spectra were recorded on a Varian
Gemini 2000 (200 MHz) instrument. D2O was used as solvent
with DHO (1H NMR: δH 4.79) and acetone (1H NMR: δH 2.05;
13C NMR: δC 29.8) as reference. With CHCl3 as solvent, SiMe4
(TMS) and CHCl3 (13C NMR: δC 76.93) were used as references.
Mass spectra were obtained on a Micromass LCT instrument.
Concentrations were performed on a rotary evaporator at a
temperature below 40 ЊC.
The 3,4-elimination product was also isolated in a yield of
24 mg (6%, Rf 0.2 in AcOEt–pentane 1 : 1).
(R)-2-Hydroxymethyl-8-methyl-1,6,8-triazabicyclo[4.3.0]non-3-
ene-7,9-dione (؉)-5 and (S)-2-hydroxymethyl-8-methyl-1,6,8-
triazabicyclo[4.3.0]non-3-ene-7,9-dione (؊)-5
(2R,3S,4R)-3,4-Diacetoxy-2-acetoxymethyl-1,6,8-triazabicyclo-
[4.3.0]nonane-7,9-dione 11
A mixture of iodides 9 and 10 (180 mg, 0.42 mmol) was dis-
solved in wet TFA (4 mL containing 6% water). To the solution
was added AgO2CCF3 (186 mg, 0.85 mmol). The reaction vessel
was sealed and heated in the dark to 90 ЊC for 4 days. The
reaction mixture was then cooled and NaCl (50 mg, 0.85 mmol)
was added. Filtration and washing with CH3OH removed the
precipitate. The solvent was then removed under reduced pres-
sure and the residue underwent acetylation by stirring it in
CHCl3 (3 mL) with Ac2O (1 mL) and Et3N (1 mL) for 5 h at
room temperature. After the reaction was complete, excess of
Ac2O was destroyed by slowly adding water (5 mL) and stirring
of the mixture for 30 min. The two phases were then separated
and the aqueous phase was extracted with CHCl3 (4 × 5 mL).
The combined organic phases were washed first with saturated
aq. NaHCO3 (10 mL), then brine (10 mL), and dried over
anhydrous MgSO4. The solvent was removed and the remaining
oil underwent flash chromatography using AcOEt–pentane
1 : 1 as eluent (Rf 0.26) to afford 120 mg (79%) of the desired
product as a colorless oil containing 4% of the 3-epimer
(NMR analysis); δH (CDCl3) 5.29 (q, J 3.4 Hz, 1H, H-4), 5.17
(dd, J3,4, 3.4, J2,3 5.2 Hz, 1H, H-3), 4.67 (dd, J2,2Јa 9.0, J2Јa,2Јb
11.6, 1H, H-2Јa), 4.42–4.56 (m, 1H, H-2), 4.32 (dd, J2,2Јb 3.4 Hz,
1H, H-2Јb), 3.92 (dd, J4,5a 12.8 Hz, 1H, H-5a), 3.39 (dd, 1H,
H-5b), 3.00 (s, 3H, NCH3), 2.09, 2.05 and 1.96 (each s, 3H,
CH3); δC(CDCl3) 170.9, 169.6 and 169.3 [C(O)CH3], 155.0 and
153.2 [NC(O)N], 67.2, 65.4 (C-3, C-4), 60.0 (C-2Ј), 54.4 (C-2),
46.9 (C-5), 25.5 (NCH3), 21.1, 20.9, 20.8 (C(O)CH3).
HRMS(ES) Calc. for C14H18O8N3 ϩ Na: 380.1070. Found:
380.1072.
A mixture of ( )-5 (3.040 g, 15.4 mmol) and lipase R (Penicil-
lium Roqueforti, 4.0 g) was stirred in vinyl acetate (200 mL) at
room temperature. The reaction was monitored by taking
samples for NMR analysis. ≈50% Conversion was reached after
66 h, when the solution was filtered and the enzyme cake was
washed thoroughly with AcOEt. The solution was concentrated
in vacuo and the residue underwent flash chromatography (first
CHCl3–AcOEt 1 : 1, then AcOEt) which gave (ϩ)-5 (1.314 g,
43%, 92% ee) as a colorless solid, and then acetate 6 which
appeared as a colorless oil. The ester was deacetylated by dis-
solving the compound in 70 mL of methanol containing a
catalytical amount of NaOCH3. After total conversion (TLC
control, AcOEt) a piece of solid CO2 was added to neutralise
the solution, which was evaporated to dryness and filtered
through silica gel (AcOEt) to obtain the desired alcohol (Ϫ)-5
as a colorless solid (1.256 g, 41%, 88% ee). Recrystallisation
of each enantiomer from AcOEt–hexane gave enantiopure
alcohols (ϩ)-5 and (Ϫ)-5. The NMR spectra of isomers 5
were identical with those previously reported.10 The ees were
determined by HPLC (Daicel AD, hexane–PriOH 80 : 20,
flow rate 1.0 mL minϪ1, UV detection at 210 nm), tr = 11.1
min [(Ϫ) enantiomer], tr = 16.8 min [(ϩ)-enantiomer].
(؊)-(2R,3S,4R)- and (؊)-(2R,3R,4S)-3,4-Epoxy-2-hydroxy-
methyl-8-methyl-1,6,8-triazabicyclo[4.3.0]nonane-7,9-dione
7 and 8
Alkenol (Ϫ)-5 (774 mg, 3.9 mmol) was dissolved in 1,2-
dichloroethane (10 mL) and MCPBA (2 g) was added. The
solution was stirred for 18 h at 80 ЊC after which the solvent was
removed under reduced pressure. The residue was put directly
on a column of silica gel and eluted (first CHCl3, then AcOEt),
which consequently gave epoxides 7 and 8 in an isolated yield
of 347 mg (42%) and 190 mg (23%), respectively. Rf (7) 0.22
in AcOEt, Rf (8) 0.13 in AcOEt. The NMR spectra of 7 and
8 were identical with those previously reported.10
General procedure for iodide substitution
A 3 : 1 mixture of iodides 9 and 10 (20 mg) was dissolved in
AcOH or TFA containing water (0.45 mL). 2 Mole equivalents
of the silver salt (AgOAc, AgOTf or AgO2CCF3) was added.
The reaction mixture was heated to 90 ЊC in darkness in a
sealed flask for 24–96 hours (TLC control, AcOEt–pentane
1 : 1). The reaction mixture was allowed to cool to room tem-
perature and 2 mole equivalents of NaCl was added. Filtration
and washing with CH3OH removed the precipitate. The solvent
was removed under reduced pressure and the residue underwent
(2R,3S,4S)-3-Acetoxy-2-acetoxymethyl-4-iodo-8-methyl-1,6,8-
triazabicyclo[4.3.0]nonane-7,9-dione and (2R,3R,4R)-4-acetoxy-
2-acetoxymethyl-3-iodo-8-methyl-1,6,8-triazabicyclo[4.3.0]-
nonane-7,9-dione 9 and 10
908
J. Chem. Soc., Perkin Trans. 1, 2001, 905–909

View Article Online
acetylation by being stirred it in CHCl3 (0.3 mL) with Ac2O
(0.1 mL) and Et3N (0.1 mL) for 5 h at room temperature.
After the reaction had gone to completion, excess of Ac2O
was destroyed by addition of water (0.5 mL) and stirring of the
mixture for 30 min. All the solvent was removed by evaporating
several times with toluene. The product was sufficiently pure for
NMR analysis from which the ratio between the two epimers
was determined by integration.
Acknowledgements
This paper is dedicated to the memory of Göran Magnusson.
He was a great friend and an outstanding scientist. We
thank Ms Anne Bülow, Mr Helmer Søhoel and Mr
Xifu Liang for some of the enzyme kinetic data and the
Danish National Research Council (SNF) for financial
support.
(؉)-(3R,4S,5R)-4,5-Dihydroxy-3-(hydroxymethyl)hexahydro-
pyridazine 2
References
1 T. D. Heightman and A. T. Vasella, Angew. Chem., Int. Ed., 1999,
38, 750.
Triacetate 11 (120 mg, 0.34 mmol) was dissolved in methanol
(3 mL) containing a catalytic amount of NaOCH3. The
deacetylation was complete in 30 min, and the solvent was
removed. To the residue was added hydrazine hydrate (4 mL)
and the mixture was refluxed for 18 h. The solvent was then
removed under reduced pressure and the remaining oil under-
went ion exchange (Amberlite IR-120, Hϩ). The product was
released from the resin with 2.5% NH4OH. Concentration
followed by chromatography (ethanol–25% NH4OH 20 : 1, Rf
0.18) gave 28 mg (56% over two steps) of 2 (without any trace of
the 4-epimer) as a colorless solid, [α]D22 ϩ11.9 × 10Ϫ1 deg cm2 gϪ1
(H2O); δH(D2O) 3.96 (br s, 1H, H-4), 3.69–3.79 (m, 1H, H-5),
3.59 (d, J3,3Ј 6.6 Hz, 2H, H2-3Ј), 2.89 (dt, J3,4 1.6 Hz, 1H, H-3),
2.85 (dd, J5,6eq 3.2 Hz, 1H, H-6eq), 2.75 (dd, J5,6ax 10.8, JH6eq,H6ax
12.8 Hz, 1H, H-6ax); δC(D2O) 68.3, 67.1 (C-4, C-5), 61.5, 60.9
(C-3, C-3Ј), 46.9 (C-6); HRMS(ES) Calc. for C5H12N2O3 ϩ Na:
m/z, 171.0746. Found: m/z, 171.0740.
2 D. L. Zechel and S. G. Withers, Acc. Chem. Res., 2000, 33, 11.
3 A. E. Stütz, Iminosugars as Glycosidase Inhibitors: Nojirimycin and
Beyond, Wiley–VCH, Weinheim, 1999.
4 M. Bols, Acc. Chem. Res., 1998, 31, 1.
5 Y. Ichikawa, Y. Igarashi, M. Ichikawa and Y. Suhura, J. Am. Chem.
Soc., 1998, 120, 5854.
6 M. Bols, R. Hazell and I. Thomsen, Chem. Eur. J., 1997, 3,
940.
7 B. V. Ernholt, I. B. Thomsen, A. Lohse, K. B. Jensen, R. G. Hazell,
I. Plesner, X. Liang, A. Jacobsen and M. Bols, Chem. Eur. J., 2000,
6, 278.
8 Y. Ichikawa and Y. Igarashi, Tetrahedron Lett., 1995, 36, 4585.
9 G. Legler, Adv. Carbohydr. Chem. Biochem., 1990, 48, 319.
10 X. Liang and M. Bols, J. Org. Chem., 1999, 64, 8485.
11 R. P. Clausen and M. Bols, J. Org. Chem., 2000, 65, 2797.
12 R. B. Woodward and F. V. Brutcher, Jr., J. Am. Chem. Soc., 1958, 80,
209.
13 A. Bülow, I. Plesner and M. Bols, J. Am. Chem. Soc., 2000, 122,
8567.
J. Chem. Soc., Perkin Trans. 1, 2001, 905–909
909