Transglycosylation by A. hydrophila 397
Synthesis of ribavirin
Results
Substrate specificity of A. hydrophila strains
The reaction mixture comprised wet cell paste of
A. hydrophila CECT4226 containing 5 ϫ 1010 cells
mL−1, 30 mM 1,2,4-triazole-3-carboxamide (TCA,
13.4 mg), 30 mM uridine (29.3 mg) and 20 mM
potassium phosphate buffer pH 7 (4 mL) which was
stirred at 200 rpm and 60°C. After 26 h, the mixture
was centrifuged and the ribavirin conversion (77%)
in the supernatant was determined by HPLC using
as eluent (1) 4 min water/acetonitrile (99.4:0.6, v/v),
(2) 1 min gradient to water/acetonitrile (95:5, v/v) and
(3) 4 min water/acetonitrile (95:5, v/v) and setting the
detector at l ϭ 235 nm; Rt (min): uracil ϭ 2.5; riba-
virin ϭ 3.5; uridine ϭ 4;TCA ϭ 7. Finally, ribavirin
was obtained in 65% yield (14.2 mg, 99% purity
by HPLC) after purification from the supernatant
using first a variable volume column (10 ϫ 200 mm,
Kontes Flex-Column,Vineland, NJ), containing C18
silica gel (10 g, Phenomenex, Torrance, CA) eluting
successively with five volumes of H2O, acetonitrile 5%
and 10%.The fractions containingTCA and ribavirin
were further purified by normal flash chromatography
with methanol:chloroform 20:80 v/v.
Several natural and unnatural pyrimidine nucleosides
and purine bases were used as starting materials
for transglycosylation biocatalyzed by two previously
selected A.hydrophila strains (CECT 4226 and 4221)
(Tables I–III).All biotransformations were performed
using the standard methodology and the products
were characterized by HPLC when commercial
reference materials were available or by MS in the
case of non-commercial products (2’-deoxyribavirin
M/Zϩ1 ϭ 231.2; 2-fluoradenine-2’-deoxyriboside
M/Zϩ1 ϭ 270.1; 2-amino-6-chloropurine-2’-deoxy-
riboside M/Zϩ1 ϭ 302.2; 2-fluoradenine riboside
M/Zϩ1 ϭ 286.1; 2-amino-6-methoxypurine riboside
M/Zϩ1 ϭ 298.5; 6-mercaptopurine riboside M/Zϩ1
ϭ 285.2; 8-azaguanosine M/Zϩ1 ϭ 258.5; 6-iodopu-
rine riboside M/Zϩ1 ϭ 369.4; 6-methoxypurine
riboside M/Zϩ1 ϭ 286.3).
The base specificity in the synthesis of ribosides
was studied initially with A. hydrophila CECT 4226
using uridine and different purine bases as sub-
strates (Table I).The presence of substituents in the
purine ring was necessary for biocatalyst activity
since no transglycosylation was observed starting
from purine alone (Table I, entry 14). Unexpect-
edly, among the 6-substituted purines (Table I,
entries 1–6) adenine was not a substrate and the
best result was obtained starting with 6-iodopurine.
Incorporation of an amino group at the C-2 of the
purine ring (Table I, entries 7–10) dramatically
increased the yields, except when guanine was the
substrate, although the latter may be attributed
to the low solubility of this base (Medici et al.
2008). The presence of a 2-fluorine (Table I, entry
11) also improved the yield and the reaction time.
Synthesis of didanosine
The reaction mixture comprised wet cell paste of
A. hydrophila CECT4221 containing 3 ϫ 1010 cells
mL−1, 10 mM hypoxanthine (6.8 mg), 30 mM dide-
oxyuridine (31.8 mg) and 30 mM pH 7 potassium
phosphate buffer (5 mL) which was stirred at 200
rpm and 45°C. After 16 h, the mixture was centri-
fuged and DDI conversion (62%) was determined
by HPLC using water/acetonitrile (92.5:7.5 v/v) as
eluent and setting the detector at λ ϭ 248 nm;
Rt(min): uracil ϭ 1.8; dideoxyuridine ϭ 2.2; hypox-
anthine ϭ 4.8; didanosine ϭ 6.3. Finally, DDI was
obtained in 58% yield (7.1 mg, 99% purity by
HPLC) after purification from the supernatant using
a variable volume column (10 ϫ 200 mm, Kontes
Flex-Column), containing C18 silica gel (10 g, Phe-
nomenex) eluting successively with five volumes of
H2O, acetonitrile 2%, 10% and 50%.
Since benzimidazole,
a
1,3-deazapurine ring
(Bentancor et al. 2004) (Table I, entry 15) afforded
good conversions in the transglycosylation reac-
tions, other purine analogues with different nitrogen
contents were assessed.Thus,TCA, a triazole deriv-
ative (Table I, entry 16) was also a good substrate
whereas 7-deaza-adenine and 8-azaguanine (Table I,
entries 12 and 13) were poorly accepted.The trans-
glycosylation was not successful using bases con-
taining indol, imidazole or pyridine rings (Table I,
entries 17–19).
Table II shows the pyrimidine nucleoside accep-
tance by A. hydrophila CECT 4226. Nucleosides
containing different sugar residues and uracil or
thymine as bases were assessed for the synthesis of
benzimidazole nucleosides. This base was chosen
since it had already been demonstrated that benz-
imidazole displays good affinity for PNP but, in
contrast, the corresponding nucleosides are poor
A. hydrophila immobilization
The pellet, prepared as above described, was mixed
with previously sterilized agarose (3 mL, 2% (w/v)).
Then, the mixture was slowly added to stirred sun-
flower oil (10 mL) at 25°C for 5 min. The resulting
gel beads (mean diameter 3.5 mm) were cooled,
filtered, washed with hexane and then with physio-
logical solution, to obtain solvent-free beads. They
were used directly as the biocatalyst (catalyst load:
2.9 ϫ 1010 cells g–1).