2
184
M. Nóbile et al. / Process Biochemistry 47 (2012) 2182–2188
1
1
−1
2
.5. Adenosine synthesis by immobilized C. koseri cells
C. koseri whole cells (8 g of immobilized biocatalyst, 1.4 × 10 cells g ), 3 mM 2-
fluoroadenine and 6 mM AraU in case of fludarabine and 10 mM adenine and 20 mM
of AraU for vidarabine, in 30 mM potassium phosphate buffer pH 7.
The stirred-tank reactor consisted of a spinner glass reactor (Celstir, Wheaton,
Millville, USA) equipped with radial flow impellers. The stirring at 200 rpm was
produced by the magnetic rotation of the impellers using a hot plate (Heidolph,
UK).
The standard reaction mixture for adenosine synthesis containing: 30 mM ade-
nine, 30 mM uridine and 30 mM potassium phosphate buffer pH 7 (final volume
3
allowed to proceed for 3 h (63% conversion). Aliquots taken at different reaction
times were centrifuged at 10,000 × g for 30 s and the supernatants were analyzed
by HPLC, using as operating conditions: 5 min water/methanol (90:10, v/v), 1.5 min
gradient to water/methanol (80:20, v/v) and 5 min water/methanol (80:20, v/v),
◦
mL) and 1 g biocatalyst, was stirred at 200 rpm and 60 C. The reactions were
The packed-bed reactor consisted of a cylindrical glass of 10 cm length and
1.5 cm width. The reaction mixture was introduced from the bottom using a peri-
−
1
−1
1
mL min flow rate and detector set at ꢀ = 254 nm (Rt: uracil = 2.6; uridine = 3.1;
staltic pump operating at 1.6 mL h
.
adenine = 6.7; adenosine = 9.1).
The bubble column reactor consisted of a 10 cm length and 2.6 cm width-glass
column. To maintain the necessary fluidization of the bed, and external pump sup-
−
1
2.6. Reuse capacity
plying an air flow of 0.15 L min was coupled to the column. Evaporation of the
reaction mixture was minimized (less than 1% in volume) attaching a condenser
◦
To assess the reuse potential of the C. koseri entrapped cells, repeated batch
on top of the reactor. In all cases the temperature was maintained at 60 C using a
transglycosylation reactions for the synthesis of adenosine were carried out under
the experimental conditions described above. After each cycle, the beads were fil-
tered from the reaction media, washed with phosphate buffer and transferred into
a fresh reaction mixture.
silicone flexible heating tape rolled around the reactor. Once the optimal reaction
times were achieved, the mixtures were centrifuged and the supernatants analyzed
by HPLC.
The synthesis of vidarabine in a final volume of 150 mL was performed using a
5
00 mL spinner glass reactor (Celstir, Wheaton, Millville,US) equipped with radial
−
1
2.7. Analytical methods
impellers. The reaction mixture comprised 0.35 g mL agarose beads containing
1
1
−1
C. koseri whole cells (1.4 × 10 cells g ), 10 mM adenine and 20 mM of AraU, in
◦
HPLC analyses were performed in a modular Gilson instrument (321Pump, 156
30 mM potassium phosphate buffer pH 7. The reaction was kept at 60 C and 200 rpm
UV/VIS detector and 234 Autoinjector Series) (Middleton, WI, USA) with an Alltech
Apollo RP18 column (150 × 4.6 mm, 5 ) (Deerfield, IL, USA), at room temperature
and eluted with water/acetonitrile or water/methanol mixtures using authentic
materials as reference standards, when available.
MS analysis was carried out using a Thermo-Finnigan LCQ Advantage Max spec-
trometer (San Jose, CA, USA) by direct injection, in positive mode, after solid phase
extraction (SPE). SPE was performed using a C18 silica gel cartridge (Phenomenex,
Torrance, CA, USA) and eluted with 1% methanol/formic acid (70:30, v/v).
and was allowed to proceed during 26 h. After purification following the protocol
previously detailed, 175.2 mg of vidarabine (62.5% yield) were recovered from the
reaction mixture.
The synthesis of adenosine as model reaction was also carried out in con-
tinuous operating packed-bed reactors (100 and 200 mL volume glass columns;
0.25 cm × 20 cm and 40 cm, respectively) (Supplementary information Figs. S1 and
◦
S2). The reactors were pre-heated at 60 C and this temperature was kept dur-
ing the whole process. The synthesis of adenosine was performed using 72 g
(
100 mL reactor) or 141.5 g (200 mL reactor) of biocatalyst and the reaction mix-
ture described above was circulated from a reservoir through a peristaltic pump.
2
.8. Synthesis of fludarabine
−
1
The determined optimal flow and residence time were 1.6 mL min /124 min and
−
1
The reaction mixture comprising: C. koseri (wet cell paste or 1 g agarose beads)
3.4 mL min /110 min for 100 and 200 mL reactors respectively, affording a produc-
tive capacity of 0.45 and 0.96 mg/h of adenosine (61.7 ± 0.1% yield during 27 h for
100 mL reactor and 61.8 ± 0.1% yield during 24 h for 200 mL reactor).
10
−1
5
× 10 cells mL , 3 mM 2-fluoradenine, 6 mM AraU and 30 mM potassium phos-
◦
phate buffer pH 7 (3 mL), was stirred at 200 rpm and 60 C. After 14 h, the mixture
was centrifuged and 58% of fludarabine was determined in the supernatant by
HPLC, using as eluent: (1) 3 min water/methanol (85:15, v/v), (2) 2 min gradi-
ent to water/methanol (70:30, v/v) and (3) 4 min water/methanol (70:30, v/v),
3. Results and discussion
−
1
0
.9 mL min
flow rate and setting the detector at ꢀ = 260 nm (Rt: uracil = 2.4;
AraU = 3.1; 2-fluoradenine = 6.6; fludarabine = 7.7). Finally, the fludarabine was puri-
fied from the supernatant using a variable volume column (10 mm × 200 mm, Kontes
Flex-Column, Vineland, NJ, USA), containing C18 silica gel (Phenomenex, Torrance,
CA, USA) eluting successively with 5 volumes of each: H2O, 5% and 10% acetonitrile.
Fludarabine was obtained in 48% yield (99% purity by HPLC).
3
.1. Biocatalyst selection
Microorganisms supply a large diversity of biocatalysts with
different substrate specificities. In order to make use of this enzy-
matic diversity for the synthesis of bioactive purine arabinosides,
a screening process capable of identifying the strains with the
proper transglycosylation activity has been carried out based on
the methodology previously developed in the group [29]. Starting
from AraU and the corresponding purine bases, almost 100 strains
were screened to afford vidarabine, fludarabine and DAPA. Since it
is known that AraU is substrate of UP [30], all the transglycosyla-
tions were carried out at 60 C. This temperature was used to avoid
deamination and to improve NP activities. The most representative
results are shown in Table 1.
The only microorganism that efficiently catalyzed the synthesis
of the three arabinonucleosides was C. koseri (CECT 856). Although
C. koseri is frequently used as microbial sensor in bioremedia-
tion [31,32], its transglycosylation activity had not been previously
reported.
2.9. Synthesis of vidarabine
The reaction mixture comprising: C. koseri (wet cell paste or 1 g agarose beads)
1
0
−1
5
× 10 cells mL , 10 mM adenine, 20 mM AraU and 20 mM potassium phosphate
◦
buffer pH 7 (3 mL), was stirred at 200 rpm and 60 C. After 26 h, the mixture was cen-
trifuged and 71% of vidarabine was determined in the supernatant by HPLC, using
−1
as eluent water/acetonitrile (95.2:4.8, v/v), 1 mL min flow rate and setting the
detector at ꢀ = 254 nm (Rt: uracil = 2.1; AraU = 2.7; adenine = 3.6; vidarabine = 4.9).
Finally, vidarabine was purified from the supernatant using a variable volume col-
umn (10 mm × 200 mm, Kontes Flex-Column, Vineland, NJ, USA), containing C18
silica gel (Phenomenex, Torrance, CA, USA) eluting successively with 5 volumes of
each: H2O, 5% and 10% acetonitrile. Vidarabine was obtained in 62% yield (99% purity
by HPLC).
◦
2.10. Synthesis of 2-6-diaminopurinarabinoside
1
0
−1
,
The reaction mixture comprising: C. koseri (wet cell paste) 1.1 × 10 cells mL
1
0 mM 2,6-diamopurine, 30 mM AraU and 30 mM potassium phosphate buffer
pH 7 (3 mL), was stirred at 200 rpm and 60 C. After 24 h, the mixture was cen-
trifuged and 77% of DAPA was determined in the supernatant by HPLC, using as
eluent water/methanol (90:10, v/v), 1 mL min flow rate and setting the detec-
tor at ꢀ = 254 nm (Rt: uracil = 2.4; AraU = 3.4; 2,6-diaminopurine = 4.7; DAPA = 6.3).
Finally, DAPA was purified from the supernatant using first a variable volume col-
umn (10 mm × 200 mm, Kontes Flex-Column, Vineland, NJ, USA), containing C18
silica gel (Phenomenex, Torrance, CA, USA) eluting successively with 5 volumes of
each: H2O, 5% and 10% acetonitrile. DAPA was obtained in 66% yield (99% purity by
HPLC).
◦
−1
3.2. 3 mL scale reaction
3
.2.1. Biocatalyst load
The influence of the biocatalyst load on vidarabine transgly-
9
cosylation reaction was investigated varying from 3.6 × 10 to
10
−1
5 × 10 cells mL . The results (Table 2, entries 4–8) showed that
10
−1
1
× 10 C. koseri cells mL was sufficient to reach a good trans-
2.11. Scaled reactions (25–200 mL)
glycosylation yield but, as expected, shorter time was observed
10
−1
were
increasing the biomass amount. Then, 5 × 10 cells mL
The 25 mL scale syntheses of fludarabine and vidarabine were performed in
used to prepare vidarabine and fludarabine in 26 and 14 h respec-
tively. This reaction profile was not conserved in the case of
glass reactors operated in batch mode (stirred-tank, packed-bed or bubble-column
−1
reactors). The reaction mixture comprised 0.35 g mL agarose beads containing