Flow-Synthesis of Nucleosides Catalyzed by an Immobilized Purine Nucleoside Phosphorylase from Aeromonas hydrophila: Integrated Systems of Reaction Control and Product Purification

Article abstract of DOI:10.1002/adsc.201500133

A purine nucleoside phosphorylase from Aeromonas hydrophyla (AhPNP) was covalently immobilized in a pre-packed stainless steel column containing aminopropylsilica particles via Schiff base chemistry upon glutaraldehyde activation. The resulting AhPNP-IMER (Immobilized Enzyme Reactor, immobilization yield ≈50%) was coupled on-line through a 6-way switching valve to an HPLC apparatus containing an analytical or a semi-preparative chromatographic column. The synthesis of five 6-modified purine ribonucleosides was carried out by continuously pumping the reaction mixture through the AhPNP-IMER until the highest conversion was reached, and then directing the reaction mixture to chromatographic separation. The conditions of the AhPNP-catalyzed transglycosylations (2:1 ratio sugar donor:base acceptor; 10 mM phosphate buffer; pH 7.5; temperature 37 °C, flow rate 0.5 mL min^-1) were optimized by a fractional factorial experimental design. Coupling the bioconversion step with the product purification in such an integrated platform resulted in a fast and efficient synthetic process (yield=52-89%; <10 mg) where sample handling was minimized. To date, AhPNP-IMER has retained completely its activity upon 50 reactions in 10 months.

Full text of DOI:10.1002/adsc.201500133

FULL PAPERS
DOI: 10.1002/adsc.201500133
Flow-Synthesis of Nucleosides Catalyzed by an Immobilized
Purine Nucleoside Phosphorylase from Aeromonas hydrophila:
Integrated Systems of Reaction Control and Product Purification
Enrica Calleri,a,* Giulia Cattaneo,a Marco Rabuffetti,b Immacolata Serra,a,e
Teodora Bavaro,a Gabriella Massolini,a Giovanna Speranza,b,c,d
and Daniela Ubialia,c,*
a
University of Pavia, Department of Drug Sciences, Viale T. Taramelli 12, I-27100 Pavia, Italy
Fax: (+39)-0382-422975; phone: (+39)-0382-987383 or (+39)-0382-987889; e-mail: enrica.calleri@unipv.it or
daniela.ubiali@unipv.it
University of Milan, Department of Chemistry, Via C. Golgi 19, I-20133 Milano, Italy
Italian Biocatalysis Center, Viale T. Taramelli 12, I-27100 Pavia, Italy
ISTM-CNR, Via C. Golgi 19, I-20133, Milano, Italy
b
c
d
e
Present address: Department of Food, Environmental and Nutritional Sciences – DeFENS, University of Milan, via L.
Mangiagalli 25, I-20133 Milano, Italy
Received: February 17, 2015; Revised: May 8, 2015; Published online: August 3, 2015
Supporting information for this article is available on the WWW under http://dx.doi.org/10.1002/adsc.201500133.
Abstract: A purine nucleoside phosphorylase from AhPNP-catalyzed transglycosylations (2:1 ratio sugar
Aeromonas hydrophyla (AhPNP) was covalently im- donor:base acceptor; 10 mM phosphate buffer;
mobilized in a pre-packed stainless steel column con- pH 7.5; temperature 378C, flow rate 0.5 mLminÀ1)
taining aminopropylsilica particles via Schiff base were optimized by a fractional factorial experimental
chemistry upon glutaraldehyde activation. The re- design. Coupling the bioconversion step with the
sulting AhPNP-IMER (Immobilized Enzyme Reac- product purification in such an integrated platform
tor, immobilization yield ꢀ50%) was coupled on-line resulted in a fast and efficient synthetic process
through a 6-way switching valve to an HPLC appara- (yield=52–89%; <10 mg) where sample handling
tus containing an analytical or a semi-preparative was minimized. To date, AhPNP-IMER has retained
chromatographic column. The synthesis of five 6- completely its activity upon 50 reactions in 10
modified purine ribonucleosides was carried out by months.
continuously pumping the reaction mixture through
the AhPNP-IMER until the highest conversion was Keywords: Aeromonas hydrophila; biotransforma-
reached, and then directing the reaction mixture to tions; flow-synthesis; nucleosides; purine nucleoside
chromatographic separation. The conditions of the phosphorylase; transglycosylation
Introduction
on synthetic chemistry as an alternative to batch proc-
essing. A key advantage of such a technology is the
In recent years, the use of biocatalyzed reactions in possibility to scale-up a reaction without scaling-up
the production of fine chemicals has become increas- the equipment by exploiting the continuous flow of
ingly important. This mainly derives from the high the reagents through the system. Flow reactor tech-
chemo-, regio-, and stereoselectivity of biocatalysts nology allows for a very accurate control of the reac-
under mild reaction conditions (temperature, pH, tion parameters through automation, thus increasing
pressure) which turn into less reaction steps. More- reproducibility, safety, efficiency and, not less impor-
over, the reduced use of toxic and hazardous re- tant, reducing the R&D time.[3,4] In particular, this
agents/solvents and the cut down of polluting waste latter factor is a pivotal driver in the drug discovery
associated with biotransformations usually afford en- process where a large number of compounds (as po-
vironmentally friendly and safer processes.[1,2]
tential drug candidates for primary assays) is expected
On the other hand, recent developments in “contin- to be delivered in a short time. Many continuous flow
uous flow” synthesis are also significantly impacting processes have already been established in synthetic/
2520
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2015, 357, 2520 – 2528

FULL PAPERS
Flow-Synthesis of Nucleosides Catalyzed
Scheme 1. PNP-catalyzed reactions: phosphorolysis (a); transglycosylation (b). Legend: Ino=inosine (1); Hpx=hypoxan-
thine (2); PNP=purine nucleoside phosphorylase; Pi =inorganic phosphate; Ade=adenine (3); Ado=adenosine (9).
medicinal chemistry,[5,6] whereas the use of flow bio-
Accumulated data about the PNP from Aeromonas
processes is still in its infancy. It is foreseen that cou- hydrophila (AhPNP)[14,17,20] have clearly shown that
pling the advantages of biocatalysis and continuous this enzyme can be successfully used in the synthesis
flow-synthesis will make it plausible for this novel of a wide range of nucleoside analogues, that are
technology to grow rapidly, as proved by some recent either routinely used as drugs (i.e., arabinosylade-
interesting applications.[7,8]
nine),[17,20] or can have promising pharmacological ac-
One of the areas where biocatalysis has clearly tivities, such as some 6-substituted purine ribonucleo-
shown its potential is nucleoside chemistry. Enzymatic sides.[14,21]
synthesis of nucleosides entails several advantages
Since many years some of the authors have been in-
over “conventional” chemical routes, such as smooth- volved in the development of analytical bioreactors as
er synthetic schemes where protection/deprotection tools for the study of enzymatic reactions.[22–24] Re-
steps can be omitted through more selective reactions. cently, a fused silica open tubular capillary system
Nucleoside phosphorylases (NPs, EC 2.4.2), which based on AhPNP has been developed to assay the
participate in the metabolism of nucleotides salvage/ substrate specificity of this enzyme by phosphorolysis
recycling,[9,10] are among the most studied enzymes as (see Scheme 1a) towards some modified nucleosides.
biocatalysts for the synthesis of nucleosides.[11–13] Par-
To this aim, through a bidimensional chromato-
ticularly, purine nucleoside phosphorylases (PNPs, graphic system, the reaction mixture was transferred
EC 2.4.2.1) catalyze the removal of the glycosyl on-line from the bioreactor to the analytical column
moiety from purine nucleosides (Scheme 1a, phos- for the separation and quantification of substrate (nu-
phorolysis) and can transfer it to another purine base, cleoside) and product (nucleobase). The AhPNP-
thus forming a new nucleoside (Scheme 1b, transgly- based bioreactor was shown to afford the same activi-
cosylation). The whole process can be therefore con- ty ranking order as determined with the soluble
veniently exploited to synthesize purine nucleoside enzyme towards a set of modified nucleosides.
analogues which are modified on the sugar or/and on AhPNP was immobilized on the inner surface of the
the nucleobase, in one step. PNPs from different bio- silica capillary previously functionalized with 3-ami-
logical sources have been used as biocatalysts for this nopropyltriethoxysilane and glutaraldehyde. The re-
aim, either as soluble or immobilized enzymes, either sulting imino groups were chemically reduced to co-
[25]
alone[14,15] or coupled with a second enzyme,[16–19] typi- valent, stable C N bonds.
cally an NP with specificity for pyrimidine nucleo-
À
As a step forward of this latter study, we have
sides, i.e., a pyrimidine nucleoside phosphorylase hereby developed a packed bioreactor based on
(PyNP, EC 2.4.2.2) or thymidine phosphorylase (TP, AhPNP for the flow-synthesis of a set of 6-substituted
EC 2.4.2.4).
purine ribonucleosides (10–14, Scheme 1b) with the
Adv. Synth. Catal. 2015, 357, 2520 – 2528
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2521

FULL PAPERS
Enrica Calleri et al.
goal to overcome one of the major bottlenecks in bio-
The activity of the AhPNP-IMER was assayed by
catalysis that is downstream of pure products. The bi- submitting the bioreactor to a chromatographic activi-
oreactor was indeed interfaced on-line through ty assay (phosphorolysis of inosine to hypoxanthine,
a switching valve to an analytical/semipreparative see Scheme 1a). This assay was periodically carried
column for the separation/isolation of the target nu- out also to test the stability of the bioreactor over
cleosides from the reaction mixture. Thus, coupling time. To date, the AhPNP-IMER has retained the ini-
bioconversion and separation steps made the whole tial catalytic activity over 50 reactions in 10 months.
process smoother and free from any further down-
stream work-up and purification.
Chromatographic System Set-Up (Analytical Scale)
Results and Discussion
The chromatographic system previously reported[25]
was modified as depicted in Figure 1 (panel a). This
system can take on two different configurations de-
pending on the 6-way switching valve (V) position (a
Bioreactor Preparation
AhPNP was covalently immobilized on aminopropyl- and b). When V is in a position, the reaction mixture
silica particles (5 mm, 100 Š) pre-packed in a stainless is continuously recirculated through the bioreactor
steel column (50”4.6 mm), by exploiting the same thus allowing for the biotransformation to occur. On-
binding chemistry previously used for the immobiliza- line reaction monitoring can be easily performed by
tion of AhPNP on the inner surface of a silica capilla- automatically switching V to b position at fixed times
ry:[25] Lys residues of the enzyme reacted with the al- allowing an assessment of the time necessary to reach
dehyde groups provided by the support, upon its func- the equilibrium. This value is crucial in the view of
tionalization with glutaraldehyde (GA). In the case of the application of the system for preparative purposes
the silica capillary, amino groups reacting with GA as the purification is performed once the highest con-
were the result of the functionalization of the silica version has been achieved. By switching V in b posi-
surface of the capillary with 3-aminopropyltriethoxy- tion, a 5-mL aliquot of the reaction mixture is diverted
silane; in the case of the on-column immobilization, to the analytical column where the separation of sub-
the stationary phase was constituted of aminopropyl- strates (nucleoside donor, i.e., inosine, purine acceptor
silica particles and, indeed, the support was ready to 3–8, see Scheme 1) and products (i.e., hypoxanthine 2
directly react with GA. In both cases, the resulting coming from the phosphorolysis of inosine 1, synthe-
À
imino groups were reduced to C N bonds by sodium sized nucleosides 9–14) occurs.
cyanoborohydride.
The higher surface area of the packed silica bed in
comparison with the capillary surface[25] allowed Design of Experiments (DoE)
a higher protein loading to be achieved, resulting in
the immobilization of approximately 25 mg of enzyme To characterize the AhPNP-IMER, a two-level, five-
(immobilization yield ꢀ50%), as roughly estimated factor fractional factorial design was used.
by the Bradford assay.[26]
The synthesis of Ado (9) by transglycosylation
Considering that the amount of immobilized (Scheme 1b) was used as the model reaction. Each re-
enzyme on the inner surface of the capillary was action mixture (exp. no. 1–11, Table 1) was pumped
about 0.03 mg (immobilization yield ꢀ10%),[25] a dra- into the bioreactor using the chromatographic analyti-
matic improvement was achieved in the case of the cal settings (Figure 1, panel a). The considered factors
packed bioreactor.
were: molar ratio between sugar donor (Ino, 1) and
However, it is worth pointing out that immobiliza- the nucleobase (Ade, 3) (x1), phosphate concentra-
tion of AhPNP in the capillary was driven by the goal tion (x2), temperature (x3), flow rate (x4) and reac-
of setting up a miniaturized bioreactor to be routinely tion pH (x5). The levels of factors x1–x5 were select-
used instead of batch assays for a faster substrate ed on the basis of theoretical considerations and/or
screening; therefore, the use of a low amount of preliminary experiments.
enzyme (with the advantage of reusing it) and a re-
The percentage of phosphorolysis (y1), the percent-
duced sample size were desirable. Conversely, a bio- age of transglycosylation (y2) and the reaction time
reactor designed for synthetic applications required (y3) were the responses (Table 1). The mathematical
a higher efficiency. Immobilization of AhPNP on model fitted was: y=b0+b1”1+b2”2+b3”3+b4”
a packed column provided, as expected, an Immobi- 4+b5”5. Interactions were not assessable with
lized Enzyme Reactor (AhPNP-IMER) that answered a 2(nÀ2) design in agreement with the literature.[27] The
the need for a highly active bioreactor and the suita- coefficients for the regression of the model were eval-
ble scale for preparative purposes.
uated using MATLAB 5.2 (The Math Works Inc.,
Natick MA, USA).
2522
asc.wiley-vch.de
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2015, 357, 2520 – 2528

FULL PAPERS
Flow-Synthesis of Nucleosides Catalyzed
Figure 1. Panel a): analytical apparatus setting; panel b): semi-preparative apparatus setting.
Table 1. Experimental conditions (x1–x5) and responses (y1–y3) for the synthesis of Ado (9) by transglycosylation.[a]
x1
x2
x3
x4
x5 y1
y2
y3
Exp.
No.
Ino:Ade
[mM]
Phosphate
[mM]
Temperature Flow rate
[8C]
pH Phosphorolysis Transglycosylation Reaction time
(mLminÀ1)
[%]
[% ]
[min]
1
2
3
4
5
6
7
8
10:5
5:5
10:5
5:5
10:5
5:5
10:5
5:5
100
100
10
37
37
37
37
25
25
25
25
31
31
31
0.5
0.1
0.1
0.5
0.5
0.1
0.1
0.5
0.3
0.3
0.3
8
7
8
7
7
8
7
8
59.35
73.81
49.70
68.45
56.83
74.70
48.47
65.74
68.36
51.81
76.32
60.73
70.47
51.92
77.88
63.84
66.75
65.21
66.41
10
50
50
10
10
50
50
10
17
17
17
10
100
100
10
10
55
9
10
11
7.5:5
7.5:5
7.5:5
7.5 61.69
7.5 62.75
7.5 62.41
55
55
[a]
x1=Ino:Ade (5:5–10:5 mM); x2=phosphate (10–100 mM); x3=temperature (25–378C); x4=flow rate (0.1–
0.5 mLminÀ1); x5=reaction pH (7–8); y1=phosphorolysis (%); y2=transglycosylation (%); y3=reaction time (min).
Reaction volume=5 mL. Responses y1 and y2 were calculated as follows: y1 (%)=[Hpx area/(Hpx area+Ino area)]”
100; y2 (%)=[Ado area/(Ado area+Ade area)]”100. Response y3 (min) was determined as the time required to reach
the reaction equilibrium.
The highest phosphorolysis percentage (y1) was ob- donor and the nucleobase (1:1, 5 mM) and at the
tained at the lowest assayed ratio between the sugar highest tested concentration of the second substrate
Adv. Synth. Catal. 2015, 357, 2520 – 2528
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2523

FULL PAPERS
Enrica Calleri et al.
Figure 2. Results of DoE. Panel a): coefficient plot of phosphorolysis percentage (y1); panel b): coefficient plot of transgly-
cosylation percentage (y2).
(100 mM phosphate buffer) (exp no. 6, Table 1). This reactor, thus making redundant the mathematical
is coherent with the pattern of the phosphorolysis re- elaboration of response y3 (i.e., reaction time).
action where a higher concentration of phosphate The dependence of phosphorolysis and transglyco-
shifts the equilibrium towards the product formation sylation on factors x1–x5 is resumed in the coefficient
(a-d-ribose-1-phosphate).
plots depicted in Figure 2.
On the contrary, the highest transglycosylation per-
centage (y2) was achieved at the highest assayed ratio
between the sugar donor and the nucleobase (2:1, Flow-Synthesis of 6-Substituted Ribonucleosides (10–
10:5 mM) and at the lowest tested concentration of 14)
the second substrate (10 mM phosphate buffer) (exp
no. 7, Table 1). This was hardly surprising as a greater Based on the DoE results, the reaction conditions se-
amount of sugar donor and a lower amount of phos- lected for the flow-synthesis of 6-substituted ribonu-
phate (the latter acting as a substrate in the phosphor- cleosides (10–14) were a 2:1 ratio between the sugar
olysis and as a product in the transglycosylation) to- donor and the nucleobase (10:5 mM) and 10 mM
gether shift the equilibrium to the nucleoside (prod- phosphate buffer.
uct) formation.
The less significant experimental parameters (tem-
The resultant model for the responses y1 and y2 re- perature, pH and flow rate) were set to such values
flects the importance of the molar ratio between the that substrate solubility could be enhanced (378C)
nucleoside (sugar donor) and the nucleobase as well and reaction times reduced, thus using the highest
as of the second substrate concentration (phosphate), flow rate (0.5 mLminÀ1). According to previous stud-
whereas temperature, flow rate and pH of the ies,[17] the pH value was set at 7.5 to preserve the long
medium were found not to affect the reaction equilib- term stability of the enzyme and, indeed, of the bio-
rium.
In each reaction set of the model, the reaction equi-
reactor.
One of the main advantages of running biocat-
librium was reached as soon as the entire volume of alyzed reactions is the possibility to work in aqueous
the reaction mixture was pumped through the bio- medium. However, the poor water solubility of some
Table 2. Flow-synthesis of nucleosides 9–14 by transglycosylation.[a]
Nucleoside (product)
Purine base
[mM]
Volume
[mL]
DMSO
[%]
C
Time
Yield
[mg]
[%][b]
[min][c]
Adenosine (Ado, 9)
3 [5]
4 [5]
5 [5]
6 [2.5]
5
5
5
10
5
0
0
5
7.5
5
78
89
52
77
81
10
10
30
20
10
n.i.[d]
6.6
3.8
5.9
6.1
2-Amino-6-methoxypurine-9-riboside, 10
6-Chloropurine-9-riboside, 11
6-(n-Butylamino)purine-9-riboside, 12
6-(2-Hydroxyethylamino)purine-9-riboside, 7 [5]
13
6-(Cyclohexylamino)purine-9-riboside, 14
8 [2.5]
10
5
87
20
7.5
[a]
Experimental conditions: molar ratio Ino/purine base=2:1; solvent: 10 mm phosphate buffer, pH 7.5; T=378C; flow
rate=0.5 mLminÀ1.
[b]
[c]
[d]
Conversion (%)=[product area/(product area+substrate area)]”100.
Time required to reach the equilibrium.
Not isolated
2524
asc.wiley-vch.de
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2015, 357, 2520 – 2528

FULL PAPERS
Flow-Synthesis of Nucleosides Catalyzed
Figure 3. Sequence of chromatograms of the synthesis of 6-chloropurine-9-riboside (11). Analytical instrumental set: 0–
6.00 min (10 mM ammonium acetate buffer, pH 5.5-MeOH 90:10; step gradient at 6.01–15.00 10 mM ammonium acetate
buffer, pH 5.5-MeOH 85:15. X axis reports the reaction time (0 min corresponds to the beginning of the transglycosylation
reaction, when the reaction mixture is pumped through the bioreactor). Arrows indicate the start of each chromatographic
run acquired after switching the 6-way valve for reaction monitoring. Calculated retention factors (k) are given in the Exper-
imental section.
purines often requires the use of a cosolvent, typically all advantage of the flow approach in terms of yield,
DMSO[25] or glycerol.[14] Both these solvents were reaction time and purification. However, this previous
found not to negatively affect the enzyme stability up study, that will be reported elsewhere, was aimed ex-
to 10 or 20% v/v, respectively, as previously report- clusively at a preliminary assessment with regard to
ed.[14,25] In this case, we opted for DMSO owing to its recognition by AhPNP of 6-modified purines (4–8) as
higher dissolving capacities. Under the experimental substrates in the view to use this information for de-
conditions described above, the synthesis of five nu- veloping the preparative process hereby described.
cleoside analogues (10–14, Scheme 1b) was performed
Among the assayed purines, 2-amino-6-methoxy-
on an analytical scale by using the chromatographic purine (4), the nucleobase counterpart of 2-amino-6-
system depicted in Figure 1 (panel a). Each reaction methoxypurine ribonucleoside (10), represents an in-
mixture contained inosine (1) as the sugar donor and teresting substrate as it is contained in Nelarabine, an
the appropriate nucleobase (4–8) in phosphate buffer arabinonucleoside having a recognized pharmacologi-
(in some cases in the presence of DMSO, see cal activity as an antitumor agent. On the other hand,
Table 2). The reaction mixture was then continuously 6-N- and 6-O-substituted purine ribonucleosides have
pumped through the bioreactor and monitored until been reported to display cytotoxic/cytostatic effects.[21]
the equilibrium was reached. The time required to
reach the reaction equilibrium ranged from 10 to
60 min depending on the nucleobase.
Flow-Synthesis and On-Line Purification of 6-
As an example, in Figure 3, the sequence of four Substituted Ribonucleosides (10-14)
chromatograms concerning the synthesis of 6-chloro-
purine-9-riboside (11) at 10, 30, 60 and 90 min is re- Starting from the optimized transglycosylation condi-
ported, where the progressive increase of the peak tions on an analytical scale, we moved to a semi-prep-
area of the nucleoside product can be observed. It is arative system with the aim to synthesize and isolate
worth noting that the highest conversion was achieved ribonucleosides 10–14 in high purity (Scheme 1 and
after only 30 min. With the exception of product 11, Table 2). The on-line synthesis and purification of
a conversion percentage higher than 75% was ob- compounds 10–14 were performed by the instrumen-
tained in only 10-20 min.
tal set-up depicted in Figure 1 (panel B b).
The semi-preparative set-up requires
When compounds 9 and 11–14 were synthesized by
a 6-way
in-batch AhPNP-catalyzed transglycosylation, under switching valve (V) and a 3-way junction (Y). When
non-optimized conditions, a 20–47% conversion was V and Y are in position a and c, respectively, systems
achieved after 6 h (data not reported). Although 1 and 2 are independent and, while the reaction circu-
a direct comparison with the same biotransformations lates through the bioreactor, the semi-preparative
carried out in-batch is not feasible due to the different column is conditioning. When Y moves to position d,
experimental conditions used (i.e., donor nucleoside, the 1.4 mL loop, located in V, is loaded with the reac-
enzyme amount, etc.), these results highlight the over- tion mixture. Separation and purification occur when
Adv. Synth. Catal. 2015, 357, 2520 – 2528
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2525

FULL PAPERS
Enrica Calleri et al.
V switches to b and, in the meantime, Y returns to c . Experimental Section
The semi-preparative chromatograms of nucleosides
10–14 are reported in Figures S1–S5 (Supporting In- General
formation).
All chemicals were of analytical or reagent grade and were
used without further purification. The substrates inosine
(Ino, 1), 6-chloropurine (5), Bradford reagent and the mate-
rials for enzyme immobilization (hydrochloric acid, glutaral-
dehyde, phosphate buffer, sodium cyanoborohydride, mono-
ethanolamine), ammonium acetate buffer and DMSO were
purchased from Sigma Aldrich (Milano, Italy). Adenine
(Ade, 3), hypoxanthine (Hpx, 2) and adenosine (Ado, 9),
were purchased from Alfa Aesar (Novachimica, Cinisello
Balsamo, Italy). Water was deionized by a Direct-QTM water
purification system (Millipore S.p.A., Vimodrone, Milano,
Italy). Stainless steel column (50”4.6 mm) was from Alltech
(Sedriano, Milano, Italy). Aminopropylsilica (5 mm, 100 Š)
was from Macherey–Nagel GmbH & Co. KG (Düren, Ger-
many).
2-Amino-6-methoxypurine (4) was synthesized following
the general procedure reported by Sakakibara et al. in 94%
yield.[28] 6-(n-Butylamino) purine (6), 6-(2-hydroxyethylami-
no) purine (7) and 6-(cyclohexylamino) purine (8) were pre-
pared as previously reported in 80–86% yield.[29]
PNP from A. hydrophila was prepared by Gnosis S.p.A.
(Desio, MB, Italy).[20] Protein concentration was
21.5 mgmLÀ1 and the specific activity toward Ino was
39 IUmgÀ1. pH measurements were performed with a pHme-
ter 718 Stat Titrino from Metrohm (Herisau, Switzerland).
The quantification of immobilized enzyme was carried out
by Bradford assay[26] using a UV 1601 spectrophotometer
(UVProbe 1.0 Shimadzu, Kyoto, Japan).
As the volume of the reaction mixture was 5 or
10 mL (Table 2), consecutive on-line sample transfer
from the reaction mixture to the semi-preparative
column was performed until mixture consumption.
All the aliquots containing the product eluted from
the semi-preparative column were collected, lyophi-
lized and structurally characterized by 1H NMR,
13C NMR and UV spectroscopy (Figures S6–S20, Sup-
porting Information). Yields corresponded to conver-
sions values, being in the range 52–89% (3.8–7.5 mg,
Table 2).
Conclusions
In this work the flow chemistry concept has been
combined to biocatalysis in order to carry out the on-
line synthesis and purification of modified nucleosides
at a mg scale.
The here proposed “flow-based” approach for the
preparation of 6-substituted purine ribonucleosides
makes the process free from any laborious step of
downstream work-up and purification, also minimiz-
ing sample handling. The target compounds were syn-
thesized in moderate to high yield and purity.
By coupling the bioconversion step with the prod-
uct purification in an integrated platform, this work
represents an innovation in the biocatalyzed synthesis
of nucleoside analogues. As most of the cost of manu-
facturing of a biotech process is attributed to the
downstream processing, this advancement answers the
need to circumvent this constraint typically associated
to in-batch biotransformations. Moreover, reuse and
stability of enzymes are pre-requisites to their appli-
cation in bioreactors and in this regard the described
AhPNP-IMER is characterized by an outstanding sta-
bility profile.
Taking into account the wide specificity of AhPNP
towards nucleosides, this system configuration is
a powerful tool to easily access, in a short time, nu-
cleoside libraries for preliminary biological testing.
However, we aim at further scaling up this prototype
1H and 13C NMR spectra were recorded on a Bruker
AVANCE 400 or a Bruker AVANCE 500 spectrometer
(Bruker, Karlsruhe, Germany) both interfaced with a work-
station running a Windows operating system equipped with
an XWIN-NMR or TOPSPIN software package at 300 K.
Chemical shifts (d) are given in parts per million and are
referenced to the solvent signals [dH =2.50 and dC =
39.52 ppm, dH =3.31 and dC =49.00 ppm from tetramethylsi-
lane (TMS) for DMSO-d6 and CD3OD, respectively].
1
13C NMR signals were assigned on the basis of H-13C corre-
lation experiments (heteronuclear multiple quantum corre-
lation spectroscopy, HSQC).
UV-Vis spectra were recorded on a Jasco V-630 spectro-
photometer (Jasco, Easton, Maryland, USA), over the range
210–400 nm.
Chromatographic Systems
Enzyme immobilization and flow biotransformations were
to increase the productivity (i.e., recycling the un- performed by two HPLC modular systems (namely System
1 and System 2 in Figure 1) that were interfaced by a 6-way
switching valve (V) and used in two different settings
(Figure 1, panel a: analytical set-up; panel b: semi-prepara-
tive set-up). System 1 consisted of an Agilent HP-1100 (Palo
Alto, CA, USA) pump (pump 1), delivering the reaction
mixture. Pump 1 was directly connected to the bioreactor
(AhPNP-IMER) and to a 6-way switching valve (V) con-
taining the loop (5 mL or 1.4 mL for analytical or semi-prep-
arative purposes, respectively). In the semi-preparative set-
ting a 3-way junction (Y) was inserted between the AhPNP-
IMER and V. System 2 consisted of an Agilent HP-1100
reacted substrates, etc.). In this frame, besides techni-
cal modifications of the chromatographic set-up, we
are currently exploring on-column co-immobilization
of AhPNP with a uridine phosphorylase in order to
perform bi-enzymatic transglycosylations.
2526
asc.wiley-vch.de
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2015, 357, 2520 – 2528

FULL PAPERS
Flow-Synthesis of Nucleosides Catalyzed
(Palo Alto, CA, USA) pump (pump 2) connected through V
to the chromatographic column (Phenomenex Gemini RP-
18, 250”4.6 mm, 5 mm, 110 Š equipped with a Phenomenex
RP-18 precolumn, or Lichrocart Lichrosphere RP-18, 300”
10 mm, 10 mm, 100 Š, for analytical and semi-preparative
purposes, respectively) and to the UV detector set at
260 nm. The systems were controlled by an HPLC ChemSta-
tion (Revision A.04.01).
order to assess the time required to reach the highest con-
version. The percentage of transglycosylation was calculated
on the basis of the depletion of the substrate nucleobase (3--
8) by monitoring the formation of the synthesized nucleo-
side (9--14): conversion (%)=[product area/(product area+
substrate area)]”100.
A third reaction was performed and purified on-line by
the semi-preparative instrument set (Figure 1, panel b) in
order to obtain the synthesized nucleoside by collecting the
corresponding peak (Supporting Information, Figures S1–
S5). The collected solutions were lyophilized and character-
Column Packing and Enzyme Immobilization
1
ized by H NMR, 13C NMR and UV spectroscopy (Support-
AhPNP was immobilized by Schiff base chemistry modifying
a previously reported protocol.[30] Briefly, an aminopropylsil-
ica stationary phase, previously packed in a stainless steel
column (50”4.6 mm), was inserted into an HPLC system
and conditioned for 20 min with the mobile phase (50 mM
phosphate buffer, pH 7.5, 0.5 mLminÀ1). The column was
then activated with a 10% (v/v) glutaraldehyde solution in
phosphate buffer (50 mM, pH 7.5) for 5 h at 0.3 mLminÀ1.
The stationary phase was then washed with phosphate
buffer (100 mM, pH 7.5) for 10 min at 0.3 mLminÀ1 to
remove the glutaraldehyde excess.
ing Information, Figures S6–S20); yield: 52–89% (see
Table 2). Experimental conditions: [3–5, 7]=5 mM and [6,
8]=2.5 mM; molar ratio Ino (1)/nucleobase (3--8)=2:1; sol-
vent: 10 mM phosphate buffer, pH 7.5 (cosolvent DMSO 5–
7.5% v/v as reported in Table 2); T=378C; flow rate=
0.5 mLminÀ1.
An AhPNP solution (25 mL, 2 mgmLÀ1) in phosphate
buffer (100 mM, pH 7.5), containing approximately 50 mg of
protein, was recirculated through the column overnight at
0.3 mLminÀ1 (T=258C). The Schiff bases were then re-
duced by pumping a 0.1M NaBH3CN solution in phosphate
buffer (100 mm, pH 7.5) for 1.5 h at 0.3 mLminÀ1. The
column was washed with phosphate buffer for 10 min
(100 mm, pH 7.5), and a 0.1M monoethanolamine solution
in the same buffer to quench the unreacted groups (2 h),
and again with phosphate buffer (100 mm, pH 7.5). Approxi-
mately 25 mg of AhPNP (53% yield) were immobilized, as
estimated by Bradford method from the difference between
the UV absorbance at the beginning and at the end of the
immobilization procedure.[26] Whenever not in use, the
AhPNP-IMER was stored at 48C in phosphate buffer
(100 mM, pH 7.5).
Chromatographic Methods
Analytical Scale
Depending on the pair nucleoside/nucleobase, different ana-
lytical methods were developed keeping a constant flow-
rate (1.0 mLminÀ1) and l (260 nm). For each compound the
retention factor k [(tRÀt0)/t0], where tR is the retention time
of the analyte and t0 is the void volume, is listed below.
Synthesis of 9 (Ado): H2O:MeOH, 90:10. k 1 (Ino)=1.75;
k 2 (Hpx)=0.93; k 3 (Ade)=2.38; k 9 (Ado)=6.18.
Synthesis of 10: 0–6.00 min 10 mM ammonium acetate
buffer pH 5.5: MeOH, 90:10; 6.01–15.00 min 10 mM ammo-
nium acetate buffer pH 5.5: MeOH, 80:20. k 1 (Ino)=2.07;
k 2 (Hpx)=1.64; k 4=4.63; k 10=10.24.
Synthesis of 11 and 13: 0–6.00 min 10 mm ammonium ace-
tate buffer pH 5.5:MeOH, 90:10; 6.01–15.00 min 10 mM am-
monium acetate buffer pH 5.5:MeOH, 85:15. k 1 (Ino)=
2.07; k 2 (Hpx)=1.64; k 5=5.72; k 11=6.04; k 7=4.18; k
13=5.15.
Synthesis of 12 and 14: 0–6.00 min 10 mM ammonium
acetate buffer pH 5.5: MeOH, 90:10; 6.01–15.00 min 10 mM
ammonium acetate buffer pH 5.5: MeOH, 60:40. k 1 (Ino)=
2.07; k 2 (Hpx)=1.64; k 6=7.06; k 12=6.87; k 8=5.34; k
14=5.14.
Activity Assay
The activity of AhPNP-IMER was assessed by measuring
the phosphorolysis of Ino (1, see Scheme 1a). A solution of
1 (40 mM) in 0.1M phosphate buffer pH 7.5 at 378C
(volume: 5 mL) was pumped through the bioreactor. After
the solution was circulated once through the AhPNP-IMER,
the reaction mixture was collected and analyzed off-line (di-
lution 1:10; eluent: H2O:MeOH, 90:10; column: Phenomen-
ex Gemini RP-18, 250”4.6 mm, 5 mm, 110 Š, equipped with
a
RP-18 Phenomenex precolumn; l: 260 nm; flow:
1 mLminÀ1; tR 1: 8.1 min, tR 2: 4.8 min). The percentage of
phosphorolysis was calculated on the basis of the depletion
of nucleoside (Ino, 1) by monitoring the formation of the
nucleobase (Hpx, 2): conversion (%)=[product area/(prod-
uct area+substrate area)]”100.
Semi-Preparative Scale
Depending on the pair nucleoside/nucleobase, different
semi-preparative methods were developed keeping a con-
stant flow-rate (4.0 mLminÀ1) and l (260 nm).
Synthesis of 10, 11 and 13: 10 mM ammonium acetate
buffer pH 5.5:MeOH, 90:10 (Supporting Information,
Figures S1, S2, S4).
Synthesis of 12 and 14: 10 mM ammonium acetate buffer
pH 5.5:MeOH, 50:50 (Supporting Information, Figures S3,
S5).
Transglycosylation Reaction
Nucleosides 9--14 were synthesized by transglycosylation
(Scheme 1b) using Ino (1) as the ribose donor and the corre-
sponding purine (3--8). Each transglycosylation reaction was
performed (in duplicate) through the AhPNP-IMER by
using the analytical instrument set (Figure 1, panel a) in
Adv. Synth. Catal. 2015, 357, 2520 – 2528
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2527

FULL PAPERS
Enrica Calleri et al.
[16] D. Ubiali, R. Rocchietti, F. Scaramozzino, M. Terreni,
A. M. Albertini, R. Fernµndez-Lafuente, J. M. Guisµn,
M. Pregnolato, Adv. Synth. Catal. 2004, 346, 1361–1366.
Stability Assay
Stability of AhPNP-IMER was assessed weekly by using the
standard activity assay reported above. For each assay, the
activity of the bioreactor was expressed as percentage of re-
sidual activity with respect to the initial activity of the bio-
reactor.
ˇ
[17] I. Serra, D. Ubiali, J. Piskur, S. Christoffersen, E. S.
Lewkowicz, A. M. Iribarren, A. M. Albertini, M. Terre-
ni, ChemPlusChem 2013, 78, 157–165.
[18] K. Szeker, X. Zhou, T. Schwab, A. Casanueva, D.
Cowan, I. A. Mikhailopulo, P. Neubauer, J. Mol. Catal.
B: Enzym. 2012, 84, 27–34.
[19] X. Zhou, K. Szeker, B. Janocha, T. Bçhme, D. Al-
brecht, I. A. Mikhailopulo, P. Neubauer, FEBS J. 2013,
280, 1475–1490.
References
[1] A. Wells, H.-P. Meyer, ChemCatChem 2014, 6, 918–920.
[2] P. Hoyos, V. Pace, M. J. Hernaiz, A. R. Alcantara, Curr.
Green Chem. 2014, 1, 155–181.
[3] R. Wohlgemuth, New Biotechnol. 2009, 25, 204–213.
[4] L. Malet-Sanz, S. Flavien, J. Med. Chem. 2012, 55,
4062–4098.
[5] D. M. Roberge, B. Zimmermann, F. Rainone, M.
Gottsponer, M. Eyholzer, N. Kockmann, Org. Process
Res. Dev. 2008, 12, 905–910.
[20] I. Serra, S. Daly, A. R. Alcàntara, D. Bianchi, M. Terre-
ni, D. Ubiali, RSC Adv. 2015, 5, 23569–23577.
ˇ
[21] a) M. Hocek, P. Silhµr, I. Shih, E. Mabery, R. Mack-
man, Bioorg. Med. Chem. Lett. 2006, 16, 5290–5293;
ˇ
b) P. Silhµr, M. Hocek, R. Pohl, I. Votruba, I. Shih, E.
Mabery, R. Mackman, Bioorg. Med. Chem. 2008, 16,
2329–2366.
[22] G. Massolini, E. Calleri, A. Lavecchia, F. Loiodice, D.
Lubda, C. Temporini, G. Fracchiolla, P. Tortorella, E.
Novellino, G. Caccialanza, Anal. Chem. 2003, 75, 535–
542.
[23] E. Calleri, C. Temporini, S. Furlanetto, F. Loiodice, G.
Fracchiolla, G. Massolini, J. Pharm. Biomed. Anal.
2003, 32, 715–724.
[24] C. Temporini, E. Perani, E. Calleri, L. Dolcini, D.
Lubda, G. Caccialanza, G. Massolini, Anal. Chem.
2007, 79, 355–363.
[25] E. Calleri, D. Ubiali, I. Serra, C. Temporini, G. Catta-
neo, G. Speranza, C. F. Morelli, G. Massolini, J. Chro-
matogr. B: Analyt. Technol. Biomed. Life Sci. 2014,
968, 79–86.
[6] M. D. Hopkin, I. R. Baxendale, S. V. Ley, Chem.
Commun. 2010, 46, 2450–2452.
[7] L. Tamborini, D. Romano, A. Pinto, M. Contente,
M. C. Iannuzzi, P. Conti, F. Molinari, Tetrahedron Lett.
2013, 54, 6090–6093.
[8] J. Lawrence, B. OꢁSullivan, G. J. Lye, R. Wohlgemuth,
N. Szita, J. Mol. Catal. B: Enzym. 2013, 95, 111–117.
[9] A. Bzowska, E. Kulikowska, D. Shugar, Pharmacol.
Ther. 2000, 88, 349–425.
ˇ
[10] A. Konrad, J. Piskur, D. A. Liberles, Gene 2012, 510,
154–161.
[11] L. A. Condezo, J. Fernµndez-Lucas, C. A. García-
Burgos, A. R. Alcµntara, J. V. Sinisterra, in: Biocataly-
sis in the Pharmaceutical and Biotechnology Industries,
(Ed.: R. N. Patel), Taylor & Francis, 2010, chap. 14,
pp 401–423.
[12] I. A. Mikhailopulo, Curr. Org. Chem. 2007, 11, 317–
335.
[13] I. A. Mikhailopulo, A. I. Miroshnikov, Mendeleev
Commun. 2011, 21, 57–68.
[26] M. M. Bradford, Anal. Biochem. 1976, 72, 248–254.
[27] M. C. Montgomery, Design and Analysis of Experi-
ments, 7th edn., John Wiley & Sons Inc., Hoboken,
2009.
[28] N. Sakakibara, R. Konishi, I. Tsukamoto, T. Maruyama,
Heterocycles 2012, 85, 1105–1116.
[29] U. Prad›re, V. Roy, A. Montagu, O. Sari, M. Hamada,
J. Balzarini, R. Snoeck, G. Andrei, L. A. Agrofoglio,
Eur. J. Med. Chem. 2012, 57, 126–133.
[14] D. Ubiali, C. D. Serra, I. Serra, C. F. Morelli, M. Terre-
ni, A. M. Albertini, P. Manitto, G. Speranza, Adv.
Synth. Catal. 2012, 354, 96–104.
[30] E. Calleri, G. Fracchiolla, R. Montanari, G. Pochetti,
A. Lavecchia, F. Loiodice, A. Laghezza, L. Piemontese,
G. Massolini, C. Temporini, J. Chromatogr. A 2012,
1232, 84–92.
[15] I. V. Fateev, K. V. Antonov, I. D. Konstantinova, T. I.
Muravyova, F. Seela, R. S. Esipov, A. I. Miroshnikov,
I. A. Mikhailopulo, Beilstein J. Org. Chem. 2014, 10,
1657–1669.
2528
asc.wiley-vch.de
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2015, 357, 2520 – 2528