Deoxyiminoalditols from aldonic acids VI. Preparation of the four stereoisomeric 4-amino-3-hydroxypyrrolidines from bromodeoxytetronic acids. Discovery of a new α-mannosidase inhibitor

Article abstract of DOI:10.1055/s-1999-3680

A convenient four step synthesis of 4-amino-3-hydroxypyrrolidines is presented. From the readily available D- and L-tetronic acids the four possible stereoisomeric 4-amino-3-hydroxypyrrolidines 14 (3R,4R), 15 (3R,4S), ent-14 (3S,4S) and ent-15 (3S,4R) could be obtained as crystalline compounds, avoiding any chromatographic purification. The key step in the reactions was the regioselective formation of either the 2,4-diamino-2,4-di-deoxy-D- threono-D-1,4-lactam (5) or the 3,4-diamino-3,4-dideoxy-L-erythrono-1,4- lactam (10) by treatment of the methyl 4-bromo-4-deoxy-2,3-cis- or -2,3- trans-anhydrotetronates (3 or 9), respectively, with liquid ammonia. Thus, opposite regioselectivity for the opening of the c/s-configurated epoxide 3 (4:1, C-2/c-3) and the trans-configurated epoxide 9 (3:7, C-2/c-3) by ammonia was observed. Preliminary testing as glycosidase inhibitors of the 4-amino- 3-hydroxypyrrolidines formed by reduction of the lactams showed an inhibition of α-mannosidase (K₁ 40 μM) by isomer 14, (3R,4R)-4-amino-3- hydroxypyrrolidine.

Full text of DOI:10.1055/s-1999-3680

1
78
PAPER
Deoxyiminoalditols from Aldonic Acids VI.
Preparation of the Four Stereoisomeric 4-Amino-3-hydroxypyrrolidines from
Bromodeoxytetronic Acids. Discovery of a New α-Mannosidase Inhibitor
1
Gerrit Limberg, Inge Lundt,* John Zavilla
Department of Organic Chemistry, Technical University of Denmark, Building 201, DK-2800 Lyngby, Denmark
Fax: + 45(45)933968; E-mail: okil@pop.dtu.dk
Received 6 February 1998; revised 19 June 1998
sequently attacked by ammonia to yield a primary amine
Abstract: A convenient four step synthesis of 4-amino-3-hydroxy-
which in turn intramolecularly opened the 2,3-epoxide to
pyrrolidines is presented. From the readily available D- and L-
give either the five-membered pyrrolidines or the six-
membered piperidines.
tetronic acids the four possible stereoisomeric 4-amino-3-hydroxy-
pyrrolidines 14 (3R,4R), 15 (3R,4S), ent-14 (3S,4S) and ent-15
3S,4R) could be obtained as crystalline compounds, avoiding any
(
We recently published the reaction of 5-bromo-5-deoxy-
pentonolactones with ammonia to give the corresponding
chromatographic purification. The key step in the reactions was the
regioselective formation of either the 2,4-diamino-2,4-di-deoxy-D-
threono-1,4-lactam (5) or the 3,4-diamino-3,4-dideoxy-L-erythro-
no-1,4-lactam (10) by treatment of the methyl 4-bromo-4-deoxy-
9
δ-lactams. This reaction could also be shown to proceed
via an initially formed 4,5-epoxide. Attack at the primary
position by ammonia gave the 5-amino derivative of the
2
,3-cis- or -2,3-trans-anhydrotetronates (3 or 9), respectively, with
liquid ammonia. Thus, opposite regioselectivity for the opening of aldonic amide, which subsequently formed the δ-lactam.
the cis-configurated epoxide 3 (4:1, C-2/C-3) and the trans-config-
Reduction of the lactam using boraneÐdimethyl sulfide
urated epoxide 9 (3:7, C-2/C-3) by ammonia was observed. Prelim-
inary testing as glycosidase inhibitors of the 4-amino-3-
hydroxypyrrolidines formed by reduction of the lactams showed an
9
complex gave the desired trihydroxypiperidines.
As an extension of this methodology we have now inves-
tigated the reactions of 2,4-dibromo-2,4-dideoxytetronic
acid methyl esters with ammonia which might lead to
aminohydroxy-substituted pyrrolidines. The required 2,4-
inhibition of α-mannosidase (K 40 µM) by isomer 14, (3R,4R)-4-
i
amino-3-hydroxypyrrolidine.
Key words: 4-amino-3-hydroxypyrrolidines, tetrono-1,4-lactams,
α-mannosidase inhibitor
dibromo-2,4-dideoxytetronic acid methyl esters
2
and 8
are readily available by treatment of either potassium
1
0
10
D-erythronate (1) or calcium D-threonate (7) with hy-
drogen bromide in glacial acetic acid followed by esterifi-
1
1
cation with methanol (Scheme 1). Since the reaction of
the brominated methyl esters 2 or 8 with ammonia did not
give homogeneous products, we prepared the 2,3-ep-
oxides which we assumed to be the primary intermediates.
We have previously shown that epoxylactones can be ob-
tained by treatment of the bromodeoxyaldonolactones
with either potassium fluoride or potassium carbonate un-
Both naturally occurring and synthetic polyhydroxylated
piperidines and pyrrolidines have been shown to exhibit
interesting biological activities as glycosidase inhibitors.2
Glycosidases are key enzymes in the biosynthesis and
processing of glycoproteins and the catabolism of glyco-
conjugates. Carbohydrate pyranoses and furanoses in
which the ring oxygen has been replaced by a nitrogen are
inert to metabolism, but can still interact with glycosidas-
es and other carbohydrate recognizing proteins. They in-
hibit glycosidases by mimicking the oxocarbenium-ion-
like transition state of the enzymatic hydrolysis of the
glycopyranoside substrates. Thus, substances that are able
to inhibit processing glycosidases of the biosynthetic
pathway of glycoproteins have become important as po-
1
2
der nonaqueous conditions. Thus, when the methyl thre-
onate 2 was treated with potassium carbonate the 2,3-cis-
epoxide 3 was formed exclusively. Similarly, the 2,3-
trans-epoxide 9 was formed selectively from the 2,4-di-
bromo-2,4-dideoxy-D-erythronate 8. Thus, only second-
ary epoxides were formed under these conditions.
Reaction of either 3 or 9 with aqueous ammonia resulted
in complex mixtures. Monitoring the reaction by 13
C
3
4
tential antiviral and antitumor agents and those that in-
NMR revealed that the present O- and N-nucleophiles
were competing in opening the epoxide at either C-2 or C-
2
b,5
hibit intestinal disaccharidases, as antidiabetic agents.
The synthetic approach to this class of compounds is
based on our ongoing work on the application of mono-
and dibrominated aldonolactones as readily accessible
3
and in replacing the primary bromine at C-4. This finally
led to a mixture of open-chain amides and γ-lactams con-
taining either one or two hydroxy and/or amino groups.
6
starting materials. The key step in our concept is the in-
To avoid competition between the nucleophiles we then
carried out the reaction in liquid ammonia. Optimization
of reaction time and temperature gave complete conver-
sions to amino hydroxy γ-lactams only, as outlined in
Scheme 1. The cis-epoxide 3 showed full conversion after
two hours at 60 °C to give amino hydroxy lactams. The
troduction of the amino function by the reaction of the
brominated compounds with either aqueous or liquid am-
7
monia. In the case of the dibromohexono- and
8
dibromoheptonolactones or their corresponding alditols
it could be shown that the reactions proceed via the initial
formation of a diepoxide. The primary epoxide was sub-
Synthesis 1999, No. 1, 178Ð183 ISSN 0039-7881 © Thieme Stuttgart á New York

PAPER
Deoxyiminoalditols from Aldonic Acids VI
179
amino-substitution pattern of 5 while 6 was shown to be a
-amino-substituted lactam.
3
Treatment of the trans-configurated epoxide 9 with liquid
ammonia required a reaction time of 18 hours at 90 °C to
achieve full conversion. In this case the main product was
the 3-amino-substituted lactam 10 which likewise could
be isolated by crystallization in 38% yield from the crude
product mixture. Separation of the mother liquor by chro-
matography yielded a further amount of 10 (22%), a
mixed fraction containing the two C-2-epimeric amino-
lactams 5 and 4 in a 2:1 ratio, together with a fraction
(
20%) of almost pure 4 containing only traces of 5. The
latter fraction could be crystallized to give 4 (13%). The
ratio of the products formed could thereby finally be cal-
culated to be 10/4/5 ≈ 70:24:6. Due to the prolonged reac-
tion time the primary formed cis-2-aminolactam 4 had
epimerized to 5 to a larger extent (5/4 ≈ 1:4) than dis-
cussed above.
It is noteworthy that the regioselectivity for the epoxide
opening is different for the two epoxides. For the cis-con-
figurated epoxide 3 opening at C-2 is favored to give a
4
.5:1 ratio of 2-amino and 3-amino products, while for the
trans-epoxide 9 this ratio is changed to 3:7 in favor for
1
3
opening at C-3. In earlier studies we have shown that the
reaction of 2-bromo-2-deoxy-D-threono-1,4-lactone with
liquid ammonia for 18 hours at 90 °C resulted in predom-
inant formation of the 3-amino-3-deoxy-D-threonic acid
amide via a 2,3-cis-epoxide. Regarding the preference for
opening at C-3 as the ÒusualÓ regioselectivity to be ex-
pected for the reaction of open chain 2,3-epoxy amides
Scheme 1
14
with N-nucleophiles the ÒunusualÓ regioselective out-
come of the reaction of cis-epoxide 3 calls for further ex-
planation.
main product was the 2-amino-substituted 1,4-lactam 5,
which could be obtained pure in 35% yield by direct crys- As we reported earlier13 treatment of 2,5-dibromo-2,5-
tallization from the crude mixture. Investigation of the dideoxy-D-xylono-1,4-lactone with liquid ammonia led to
13
mother liquor by C NMR spectroscopy showed the pres- predominant formation of the 2,5-diamino-2,5-dideoxy-
ence of about equal amounts of the lactam 5 and the 3- D-xylono-1,5-lactam. In this case monitoring of the reac-
1
3
aminolactam 6, besides a small amount of the C-2 epimer- tion in aqueous ammonia by C NMR spectroscopy re-
ic 2-aminolactam 4. Purification of this mixture by col- vealed the presence of an intermediately formed 2,3-
umn chromatography yielded a further amount of pure 5 anhydrolyxono-1,5-lactam, a 2,3-cis-epoxide, which in
(19%) and 6 (8%) besides fractions consisting of 4 with turn was opened regioselectively at C-2 by ammonia to
traces of 5, and a mixture of equal amounts of 5 and 6. The yield almost exclusively the 2-amino-2-deoxy-D-xylono-
ratio of the amino hydroxy lactams formed could thereby 1,5-lactam. The presence of a similar intermediate in the
finally be calculated to be 5/6/4 ≈ 78:18:4. Prolonged re- case of the four carbon compounds could explain the re-
action times or higher temperatures resulted in an increas- sults since the formation of a cyclic lactam is only possi-
ing amount of 4 and correspondingly decreasing yields of ble for a cis-configurated epoxide. This might be opened
5
, indicating that 4 arises from epimerization at C-2 of the predominantly at C-2 in contrast to the acyclic carboxam-
initially formed lactam 5.
ide having a trans-configurated epoxide.
An independent experiment was carried out in which pure In order to further investigate this hypothesis we worked
was treated with liquid ammonia at 60 °C for three days out a synthesis for a 2,3-epoxy-1,4-lactam, which might
5
either in the presence or absence of ammonium bromide. be an intermediate in the reactions. Thus, selective re-
If the ammonium salt was added a 1:1 mixture of 4 and 5 placement of the bromine by azide was obtained simply
was obtained, whereas the lactam 5 remained unchanged by treatment of 3 or 9 with sodium azide in DMF for 18
without the presence of the salt. This proves the structure hours at room temperature. This facile and almost quanti-
of 5 to be the 2,4-diamino-2,4-dideoxy-D-threono-1,4-lact- tative conversion gave the epoxy azides 11 and 13
am and 4 to be the corresponding D-erythro-isomer. Fur- (Scheme 2). Selective reduction of the terminal azide
thermore, CH-correlated NMR spectra confirmed the 2- could be achieved using tributyltin hydride and AIBN,15
leaving the 2,3-epoxide untouched during the radical re-
Synthesis 1999, No. 1, 178–183 ISSN 0039-7881 © Thieme Stuttgart · New York

1
80
G. Limberg et al.
PAPER
duction. The cis-configurated epoxy-azide 11 yielded the lowing reduction with boraneÐdimethyl sulfide complex.
,3-epoxy-1,4-lactam 12 as the only detectable product. Subsequent hydrolysis with aqueous hydrochloric acid
2
1
3
The intermediately formed primary amine must have re- gave a crude product which by C NMR spectroscopy
acted spontaneously with the methyl ester to give the lac- showed quantitative conversion to the corresponding 4-
tam which could be isolated by crystallization in 56% amino-3-hydroxypyrrolidine dihydrochlorides 14, 15,
yield based on 3. Reaction of 12 with liquid ammonia for ent-14 and ent-15 (Scheme 3), which by recrystallization
2
hours at 60¡C gave the 2-aminolactam 5 as the only yielded analytically pure compounds in around 70%
product. No evidence of any 3-amino or epimerized 2- yields.
amino compounds could be detected in the C NMR
spectrum.
1
3
The initially observed product ratio for the reaction of 3
with liquid ammonia can now be rationalized as a compe-
tition between two different reaction pathways. If the pri-
mary bromine is replaced first, the intermediate formation
of epoxylactam 12 leads exclusively to production of the
2
-amino compound 5, whereas epoxide opening prior to
bromine substitution results in a mixture of 2- and 3-ami-
no-compounds 5 and 6.
Reduction of the azido group of the 4-azido-2,3-trans-ep-
oxide 13 was carried out as described above, but all at-
tempts to isolate or trap the expected open-chain amino
epoxide failed, probably due to formation of polymers.
Scheme 3
Thus, all four possible diastereoisomers of 4-amino-3-hy-
droxypyrrolidines are easily available in overall yields of
1
5Ð20% from the lactones or salts of either D- or L-eryth-
ronic or D- or L-threonic acid in a four step synthesis
avoiding any chromatographic purification. Furthermore,
a procedure for the preparation of 2,4-diamino-2,4-
dideoxy-D-threono-1,4-lactam (5) without any contami-
nation of regio- or stereoisomeric byproducts, was elabo-
rated (Scheme 2).
Preliminary testing as glycosidase inhibitors of the four
pyrrolidines (3R,4R)-4-amino-3-hydroxypyrrolidine (14),
the enantiomer ent-14, the (3R,4S)-4-amino-3-hydroxy-
pyrrolidine (15) and the enantiomer ent-15 towards α-glu-
cosidase from bakers yeast and β-glucosidase from E. coli
showed no activity.
Scheme 2
Most interestingly, however, compound 14 showed inhi-
In analogy to the synthesis of the amino-lactams 5 and 10 bition of α-mannosidase from Jack beans. The K was de-
i
from the epoxides 3 and 9, their L-enantiomers ent-5 and termined to be 40 µM which is slightly better than the
ent-10 could be prepared by simply starting with potassium reported K of 68 µM for 1-deoxymannonoijrimycin
i
L-erythronate1 or calcium L-threonate. Reaction with against the same enzyme and in the same order of
0,11
16
17
hydrogen bromide in acetic acid, followed by methyl ester magnitude as reported for the inhibition of some swainso-
1
8
formation and selective conversion to the 2,3-epoxides, nine epimers (K range: 2Ð120 µM against mammalian
i
yielded ent-3 and ent-9, respectively, which on treatment α-mannosidases). Swainsonine itself, of course, shows a
1
8
with liquid ammonia gave the corresponding 2-amino- substantially better K of 0.07 µM.
i
(
ent-5) and 3-aminolactam (ent-10), as the main products.
The effect of substituting a hydroxy group by an amino
Likewise, they could be crystallized straight from the prod-
uct mixture in 37% and 33% yield, respectively.
group in α-mannosidase inhibitors is difficult to predict.
19
20
By replacing the OH-6 or OH-5 of the strong α-man-
9
18
The lactams 5, 10, ent-5 and ent-10 were silylated to pro- nosidase inhibitor 1,4-imino-D-mannitol with an amino
vide sufficient solubility in dioxane necessary for the fol- group gave inactive compounds. In contrast, the exchange
Synthesis 1999, No. 1, 178–183 ISSN 0039-7881 © Thieme Stuttgart · New York

PAPER
Deoxyiminoalditols from Aldonic Acids VI
181
of OH-2 in 1-deoxymannonoijrimycin with an amino
group, increased the inhibitory properties (to K 20 µM) of
i
2
1
this compound towards α-mannosidase.
The good and selective inhibition of α-mannosidase by
the 4-amino-3-hydroxypyrrolidine 14, with a four-carbon
framework and having only two chiral centers, is still re-
markable.
Table 1 13C NMR Chemical Shifts (δ)a
Compound C-1
C-2
C-3
C-4
C-5
OMe
52.5
b
55.6*
55.3
58.6
75.8
56.3*
70.7
57.9*
54.6
55.5
50.2
51.9
53.2*
68.7
73.8
55.4
53.8*
49.7
52.8*
53.5
52.7
71.2
67.3
3
4
5
6
9
0
1
2
3
4
5
167.3
26.7
47.8
46.0
45.0
29.8
46.3
48.4
45.2
50.0
54.8
50.7
c
c
c
b
c
b
c
b
c
c
178.8
177.5
177.0
169.4
177.2
167.6
175.7
168.2
52.5
51.0
50.2
1
1
1
1
1
1
46.1
44.5
a
b
c
Signals marked* may be interconverted.
Solvent CDCl3.
Solvent D O.
2
Mps are uncorrected. Optical rotations were determined on a Per-
kinÐElmer 241 polarimeter. NMR spectra were recorded on Bruker
AC-250 and AM-500 instruments. Chemical shifts (δ) are given in
ppm and coupling constants (J) in Hz. For NMR spectra in D O,
2
1
3
MeCN was used as internal reference (CH δ = 0.8 for C and δ =
3
1
1
(
.95 for H). For spectra in CDCl (CD at δ = 76.9) and CD OD
3
3
1
13
CD at δ = 49.0) were used as internal reference. H and C NMR
3
1
1
1
13
signals were assigned by H H COSY and H C correlated spec-
tra. All evaporations were carried out at or below 40¡C in vacuo.
Microanalysis were performed at the Analytical Department of the
Research Institute for Biochemistry and Pharmacy, Prague, The
Czech Republic. Column chromatography was performed on silica
gel (40Ð63 µm, Merck) using the flash technique. TLC spots were
visualized by dipping in a solution of 2% (w/v) p-anisaldehyde in
9
5% (v/v) EtOH and 5% (v/v) concd H SO . Compounds 2 and 8 as
2 4
well as their L-enantiomers were synthesized according to
1
0,11
10,11
literature
starting with either potassium D- or L- erythronate,
D-erythronic acid 1,4-lactone (Fluka) or calcium D- or L-threonate,
10,11,16
the latter also obtainable from Fluka.
Methyl 2,3-Anhydro-4-bromo-4-deoxy-D-erythronate (3)
1
1
Crystalline 2 (8 g, 29 mmol) was treated with anhyd K CO3
2
(20.0 g, 145 mmol) in anhyd acetone (100 mL) for 4 h at r.t. MgSO4
and activated charcoal were added and stirring was continued for
another 15 min. The solids were filtered off and the solvent evapo-
rated to give 3 (5.5 g, 98%) as a slightly colored oil. The compound
was used in the following reactions without further purification.
Methyl 2,3-Anhydro-4-bromo-4-deoxy-L-erythronate (L-3)
The title compound was prepared from L-2 analogously to the prep-
aration of 3. Syrupy L-3 was used without further purification.
2
,4-Diamino-2,4-dideoxy-D-threono-1,4-lactam (5) and 3,4-Di-
amino-3,4-dideoxy-L-threono-1,4-lactam (6)
Crude cis-epoxide 3 (5.28 g, 27 mmol) was placed in an autoclave
(100-mL volume) equipped with a magnetic stirring bar. Liquid
NH (~50 mL) was added while stirring and the autoclave was
3
Synthesis 1999, No. 1, 178–183 ISSN 0039-7881 © Thieme Stuttgart · New York

1
82
G. Limberg et al.
PAPER
closed. The mixture was allowed to warm up to r.t., after which the
autoclave was placed in an oil bath and left at 60¡C for 2 h. After
cooling the autoclave in dry ice/iPrOH, it was carefully opened and
was added and solids were removed by filtration. The solvents were
evaporated and the residual oil was redissolved in EtOAc (50 mL).
The organic phase was extracted twice with H O and brine, dried
2
the NH was evaporated using a stream of air while stirring. The
(MgSO ) and concentrated to leave 11 (1.78 g, 89%) as an oil. The
crude compound was shown to be >95% pure by C NMR and was
used without any further purification in the following synthesis.
3
4
13
crude product obtained was dissolved in H O (~100 mL) and treated
2
Ð
with ion exchange resin (Amberlite IRA 420, OH , ~100 mL) for 30
min. The resin was filtered off and washed extensively with water
and H O/MeOH (1:1). The combined solutions were concentrated.
2
4
-Amino-2,3-anhydro-4-deoxy-D-erythrono-1,4-lactam (12)
Three co-evaporations with EtOH yielded the crude mixtures of lac-
1
3
Crude 11 (3.0 g, 19 mmol) was dissolved in anhyd toluene (25 mL),
kept under argon and heated to 80¡C. A solution of Bu SnH
tams as seen by C NMR. Crystallization from MeOH yielded the
main product 5 (1.1 g, 35%). The mother liquor from the crystalli-
zation was concentrated and the residue subjected to column chro-
3
(10.2 mL, 38.2 mmol) and AIBN (100 mg, 0.6 mmol) in anhyd tol-
uene (25 mL) was added dropwise during 90 min. The reaction was
kept at 80¡C for another 60 min before the solvents were evaporat-
ed. The residue was dissolved in MeCN (30 mL) which was extract-
ed five times with pentane. The MeCN was evaporated and the
residue was dissolved in EtOH (50 mL) and treated with activated
charcoal. Filtration and partial concentration to ca. 10 mL caused
crystallization of the pure product 12. Further concentration of the
mother liquor yielded another amount of 12 bringing the total yield
matography (EtOH/25% aq NH 9:1). This yielded a further amount
3
of 5 (597 mg, 19%) as well as 6 (260 mg, 8%) together with two
mixed fractions one of which contained 4 (134 mg, 4%) with traces
of 5, and another consisting of a ~1:1 mixture of 5 and of 6 (510 mg,
1
6%).This gave pure crystalline 5 (54%) and 6 (8%). Subsequent
recrystallizations from MeOH/H O yielded analytical samples of 5
2
and 6.
: mp 162Ð165¡C (dec.); [α]2 Ð92.0 (c = 1.00, H O).
0
to 56% (1.05 g). Further recrystallization from EtOH furnished an
5
D
2
20
analytical sample: mp 106Ð107¡C;[α]D +36.6 (c = 0.56, H O).
2
Anal. calcd for C H N O (116.1): C, 41.37; H, 6.94; N, 24.12.
4
8
2
2
Found C, 41.32; H, 6.97; N, 23.95.
Anal. calcd for C H NO (99.1): C, 48.49; H, 5.09; N, 14.14. Found
4 5 2
C, 48.59; H, 5.17; N, 14.16.
: mp 182Ð184¡C (dec.); [α]2 +88.4 (c = 1.01, H O).
0
6
D
2
Anal. calcd for C H N O (116.1): C, 41.37; H, 6.94; N, 24.12.
Found C, 41.61; H, 6.96; N, 23.84.
Methyl 2,3-Anhydro-4-azido-4-deoxy-D-threonate (13)
Crystalline trans-epoxide 9 (1 g, 5.1 mmol) was dissolved in DMF
4
8
2
2
(
10 mL) and NaN (0.5 g, 7.65 mmol) was added. Reaction and
3
Methyl 2,3-Anhydro-4-bromo-4-deoxy-D-threonate (9)
Syrupy 8 (20.0 g, 72 mmol) was treated with anhyd K CO3
workup were carried out as described for 11 to yield 13 (680 mg,
4.33 mmol, 85%) as an oil. The crude compound proved to be
>95% pure as shown by 13C NMR.
1
1
2
(50.0 g, 362 mmol) in anhyd acetone (200 mL) for 4 h at r.t.. Work-
up as described for 3 yielded a syrup, which crystallized from
EtOAc/pentane to give 9 (11.5 g, 81%); mp 50Ð51¡C. An analytical
sample was obtained by recrystallization from the same solvent; mp
(
3R,4R)-4-Amino-3-hydroxypyrrolidine Dihydrochloride (14)
Lactam 5 (500 mg, 4.35 mmol) was dissolved in anhyd MeCN
20 mL) and (Me Si) NH (2.89 mL, 3.2 equiv, 13.9 mmol) and
2
0
5
3Ð54¡C; [α] +7.5 (c = 1.1, CHCl3).
D
(
3
2
Me SiCl (0.1 mL, 0.7 mmol) were added. The solution was heated to
reflux for 1 h. After cooling to r.t. the precipitate was filtered off,
3
Methyl 2,3-Anhydro-4-bromo-4-deoxy-L-threonate (L-9)
The title compound was prepared from L-8 analogously to the prep-
washed with CHCl and the solvents were evaporated at 30¡C. The
2
0
3
aration of 9; mp 51Ð52¡C; [α] Ð7.2 (c = 1.0, CHCl3).
D
residue was dissolved in anhyd dioxane (20 mL) and kept under ar-
13C NMR data were identical with those for 9.
gon. An excess of 10 M BH •SMe (2.2 mL, 22 mmol) was added.
3
2
The mixture was refluxed for 5 h and reacted for further 18 h at r.t.
followed by addition of 1 M aq HCl (20 mL). Refluxing for 2 h fol-
lowed by concentration and co-evaporation with MeOH (~20 mL)
containing concd HCl (ca. 4 drops) was performed 4 times to remove
the boronic acid as its volatile trimethyl ester. This gave the pure pyr-
rolidine dihydrochloride 14 (541 mg, 71%) after the first recrystalli-
Anal. calcd for C H BrO (195.0): C, 30.80; H, 3.62; Br, 40.97.
Found C, 30.88; H, 3.61; Br, 40.32.
5
7
3
3
,4-Diamino-3,4-dideoxy-L-erythrono-1,4-lactam (10) and 2,4-
Diamino-2,4-dideoxy-D-erythrono-1,4-lactam (4)
Crystalline trans-epoxide 9 (5.0 g, 25.6 mmol) was reacted with
zation from MeOH/H O. Further recrystallizations from the same
2
NH for 18 h at 90¡C (as described above for the preparation of 5)
20
3
solvents yielded an analytical sample; mp 256Ð259¡C (dec.); [α]D
to give 10 (1.12 g, 38%) after crystallization from MeOH. Chroma-
tography of the mother liquor yielded as the first fraction a ~2:1
mixture of 5 and 4 (167 mg, 6%) followed by 4 (595 mg, 20%) con-
taining traces of 5. Recrystallization from MeOH yielded pure 4
Ð14.6 (c = 1.08, H O).
2
Anal. calcd for C H Cl N O (175.1): C, 27.44; H, 6.91; Cl, 40.50;
4
12
2
2
2
N, 16.00. Found C, 27.81; H, 6.92; Cl, 39.79; N, 16.21.
(380 mg, 12.8%). The third fraction gave another amount of pure 10
(
651 mg, 22%), increasing the yield to 60%. Subsequent recrystal-
(
3R,4S)-4-Amino-3-hydroxypyrrolidine Dihydrochloride (15)
lizations from MeOH/H O yielded analytical samples of 10 and 4.
2
Lactam 10 (500 mg, 4.35 mmol) was reduced to the pyrrolidine fol-
lowing the procedure described above to yield 15 (551 mg, 72%) af-
0: mp 177Ð178¡C (dec.); [α]2 +5.2 (c = 1.08, H O).
0
1
D
2
Anal. calcd for C H N O (116.1): C, 41.37; H, 6.94; N, 24.12.
ter the first recrystallization from MeOH/H O. Further
4
8
2
2
2
Found C, 41.23; H, 6.94; N, 23.98.
recrystallizations from this solvent mixture yielded an analytical
sample; mp 245Ð248¡C (dec.); [α]2
0
Ð23.9 (c = 1.12, H
O).
: mp 167Ð169¡C (dec.); [α]2 Ð 22.7 (c = 1.01, H O).
0
D
2
4
D
2
Anal. calcd for C H N O (116.1): C, 41.37; H, 6.94; N, 24.12.
Anal. calcd for C H Cl N O (175.1): C, 27.44; H, 6.91; Cl, 40.50;
4 12 2 2
4
8
2
2
Found C, 41.27; H, 6.92; N, 24.19.
N, 16.00. Found C, 27.38; H, 6.81; Cl, 39.46; N, 16.06.
Methyl 2,3-Anhydro-4-azido-4-deoxy-D-erythronate (11)
2,4-Diamino-2,4-dideoxy-L-threono-1,4-lactam (ent-5)
Crude cis-epoxide 3 (2.48 g, 12.7 mmol) was dissolved in DMF
Crude methyl 2,3-anhydro-4-bromo-4-deoxy-L-erythronate (L-3)
(5.4 g, 27.6 mmol) was reacted as described for 3 to yield ent-5
(1.17 g, 8.35 mmol, 37%) after crystallization from MeOH. Further
(
20 mL) and NaN (1.24 g, 19 mmol) was added. The suspension
3
was stirred for 18 h with protection from light. EtOAc (100 mL)
Synthesis 1999, No. 1, 178–183 ISSN 0039-7881 © Thieme Stuttgart · New York

PAPER
Deoxyiminoalditols from Aldonic Acids VI
183
recrystallizations from MeOH/H O yielded an analytical sample; References
2
2
0
mp 161Ð162¡C (dec.); [α] +92.9 (c = 1.08, H O). The NMR spec-
D
2
(
1) New address for Gerrit Limberg: Danisco Biotechnology,
tra were identical with those of the enantiomer 5.
Langebrogade 1, P.O. Box 17, DK-1001 Copenhagen, Den-
mark.
Anal. calcd for C H N O (116.1): C, 41.37; H, 6.94; N, 24.12.
4
8
2
2
Found C, 41.41; H, 6.96; N, 23.85.
(2) (a) Elbein, A.D. Annu. Rev. Biochem. 1987, 56, 497.
(
(
3
b) Winchester, B.; Fleet, G. W. J. Glycobiology 1992, 2, 199.
c) Legler, G. Adv. Carbohydr. Chem. Biochem. 1990, 48,
19.
(
1
3S,4S)-4-Amino-3-hydroxypyrrolidine Dihydrochloride (ent-
4)
Lactam ent-5 (500 mg, 4.35 mmol) was reduced to the pyrrolidine
as described for the lactam 5 to yield ent-14 (528 mg, 70%) after
crystallization from MeOH/H O. Further recrystallizations from
(
3) Stephens, E. B.; Monck, E.; Reppas, K.; Butfiloski, E. J. J. Vi-
rol. 1991, 65, 1114.
2
(4) Olden, K.; Breton, P.; Grzegorzewski, K.; Yasuda, Y.; Gause,
B. L.; Ordelipe, O. A.; Newton, S. A.; White, S. L. Pharma-
col. Ter. 1991, 50, 285.
5) Robinson, K. M.; Begovic, M. E.; Reinhart, B. L.; Heineke, E.
W.; Ducep, J.-B.; Kastner, P. R.; Marshall, F. N.; Danzin, C.
Diabetes 1991, 40, 825.
6) Lundt, I. Top. Curr. Chem. 1997, 187, 117.
7) Lundt, I.; Madsen, R. Synthesis 1993, 714.
Lundt, I.; Madsen, R. Synthesis 1993, 720.
Lundt, I.; Madsen, R.; Al Daher, S.; Winchester, B. Tetrahe-
dron 1994, 50, 7513.
8) Lundt, I.; Madsen, R. Synthesis 1995, 787.
(9) Godskesen, M.; Lundt, I.; Madsen, R.; Winchester, B.
J. Bioorg. Med. Chem. 1996, 4, 1857.
10) Humphlett, W. J. Carbohydr. Res. 1967, 4, 157.
(11) Bock, K.; Lundt, I.; Pedersen, C. Acta Chem. Scand. 1983,
B37, 341.
this solvent mixture yielded an analytical sample; mp 239Ð245¡C
2
0
(dec.); [α] +14.0 (c = 1.0, H O). The NMR spectra were identical
D
2
with those of its enantiomer 14.
(
Anal. calcd for C H Cl N O (175.1): C, 27.44; H, 6.91; Cl, 40.50;
N, 16.00. Found C, 27.49; H, 6.91; Cl, 39.58; N, 16.30.
4
12
2
2
(
(
3
,4-Diamino-3,4-dideoxy-D-erythrono-1,4-lactam (ent-10)
Crystalline methyl 2,3-anhydro-4-bromo-4-deoxy-L-threonate (L-
) (5.0 g, 25.6 mmol) was treated with ammonia as described for its
9
enantiomer 9 to yield ent-10 (960 mg, 33%) after crystallization
from MeOH. Further recrystallizations from MeOH/H O yielded an
analytical sample; mp 174Ð176¡C (dec.); [α] Ð4.8 (c = 1.0, H O).
The NMR spectra were identical with those of the enantiomer 10.
(
2
2
0
D
2
(
Anal. calcd for C H N O (116.1): C, 41.37; H, 6.94; N, 24.12.
Found 41.32; H, 7.00; N, 23.91.
4
8
2
2
(
12) Lundt, I.; Pedersen, C. Synthesis 1992, 669.
(
1
3S,4R)-4-Amino-3-hydroxypyrrolidine Dihydrochloride (ent-
5)
(13) Bols, M.; Lundt, I. Acta Chem. Scand. 1991, 45, 280.
(14) Behrens, C. H.; Sharpless, K. B. J. Am. Chem. Soc. 1985, 50,
5696.
(15) Wassermann, H. H.; Brunner, R. K.; Buynak, J. D.; Carter, C.
G.; Oku, T.; Robinson, R. P. J. Am. Chem. Soc. 1985, 107,
519.
(16) Isbell, H. S.; Frush, H. L. Carbohydr. Res. 1979, 72, 301.
(17) Wong, C.-H.; Provencher, L.; Porco, J. A.; Jung, S.-H.; Wang,
Y.-F.; Chen, L.; Wang, R.; Steensma, D. H. J. Org. Chem.
Lactam ent-10 (500 mg, 4.35 mmol) was reduced to the pyrrolidine
as described for 10 to yield ent-15 (483 mg, 64%) after the first re-
crystallization from MeOH/H O. Further recrystallization from this
2
solvent mixture yielded an analytical sample; mp 245Ð250¡C
2
0
(dec.); [α] +24.4 (c = 1.0, H O). The NMR spectra were identical
D
2
with those of 15.
Anal. calcd for C H Cl N O (175.1): C, 27.44; H, 6.91; Cl, 40.50;
N, 16.00. Found C, 27.23; H, 6.95; Cl, 39.82; N, 15.71.
4
12
2
2
2
1
995, 60, 1492.
(18) Cenci di Bello, I.; Fleet, G.; Namgoong, S. K.; Tadano, K.-I.;
Winchester, B. Biochem. J. 1989, 259, 855.
Acknowledgement
(19) Farr, R. A.; Holland, A. K.; Huber, E. W.; Peet, N.; Weint-
raub, P. M. Tetrahedron 1994, 50, 1033.
Financial support of this project by the Danish Technical Research
Council is gratefully acknowledged.
(20) Kiess, F.-M.; Poggendorf, P.; Picasso, S.; JŠger, V. J. Chem.
Soc., Chem. Commun. 1998, 119.
(21) Merrer, Y. L.; Poitout, L.; Depezay, J.-C.; Dosbaa, I.; Geoff-
roy, S.; Foglietti, M.-J. Bioorg. Med. Chem. 1997, 5, 519.
Synthesis 1999, No. 1, 178–183 ISSN 0039-7881 © Thieme Stuttgart · New York